{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.981973610853001,
  "eval_steps": 1346,
  "global_step": 21524,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00018583906337112061,
      "grad_norm": 0.27501940727233887,
      "learning_rate": 1e-05,
      "loss": 1.4066,
      "step": 1
    },
    {
      "epoch": 0.00018583906337112061,
      "eval_loss": 1.6832497119903564,
      "eval_runtime": 22.6047,
      "eval_samples_per_second": 48.309,
      "eval_steps_per_second": 24.154,
      "step": 1
    },
    {
      "epoch": 0.00037167812674224123,
      "grad_norm": 0.6742346286773682,
      "learning_rate": 2e-05,
      "loss": 1.4238,
      "step": 2
    },
    {
      "epoch": 0.0005575171901133618,
      "grad_norm": 0.29993563890457153,
      "learning_rate": 3e-05,
      "loss": 1.2451,
      "step": 3
    },
    {
      "epoch": 0.0007433562534844825,
      "grad_norm": 0.28028905391693115,
      "learning_rate": 4e-05,
      "loss": 1.1953,
      "step": 4
    },
    {
      "epoch": 0.000929195316855603,
      "grad_norm": 0.2644942104816437,
      "learning_rate": 5e-05,
      "loss": 1.3122,
      "step": 5
    },
    {
      "epoch": 0.0011150343802267236,
      "grad_norm": 0.30770039558410645,
      "learning_rate": 6e-05,
      "loss": 1.5342,
      "step": 6
    },
    {
      "epoch": 0.0013008734435978444,
      "grad_norm": 0.2737680971622467,
      "learning_rate": 7e-05,
      "loss": 1.1543,
      "step": 7
    },
    {
      "epoch": 0.001486712506968965,
      "grad_norm": 0.3635600805282593,
      "learning_rate": 8e-05,
      "loss": 1.4659,
      "step": 8
    },
    {
      "epoch": 0.0016725515703400855,
      "grad_norm": 0.2415798455476761,
      "learning_rate": 9e-05,
      "loss": 1.0784,
      "step": 9
    },
    {
      "epoch": 0.001858390633711206,
      "grad_norm": 0.4179717004299164,
      "learning_rate": 0.0001,
      "loss": 1.4486,
      "step": 10
    },
    {
      "epoch": 0.0020442296970823266,
      "grad_norm": 1.9943130016326904,
      "learning_rate": 0.00011000000000000002,
      "loss": 4.5419,
      "step": 11
    },
    {
      "epoch": 0.002230068760453447,
      "grad_norm": 0.3632005453109741,
      "learning_rate": 0.00012,
      "loss": 1.4089,
      "step": 12
    },
    {
      "epoch": 0.0024159078238245677,
      "grad_norm": 0.30995360016822815,
      "learning_rate": 0.00013000000000000002,
      "loss": 1.047,
      "step": 13
    },
    {
      "epoch": 0.0026017468871956887,
      "grad_norm": 0.40917372703552246,
      "learning_rate": 0.00014,
      "loss": 1.239,
      "step": 14
    },
    {
      "epoch": 0.0027875859505668093,
      "grad_norm": 0.43013089895248413,
      "learning_rate": 0.00015000000000000001,
      "loss": 1.4529,
      "step": 15
    },
    {
      "epoch": 0.00297342501393793,
      "grad_norm": 0.4031105935573578,
      "learning_rate": 0.00016,
      "loss": 1.2565,
      "step": 16
    },
    {
      "epoch": 0.0031592640773090504,
      "grad_norm": 2.158144474029541,
      "learning_rate": 0.00017,
      "loss": 4.981,
      "step": 17
    },
    {
      "epoch": 0.003345103140680171,
      "grad_norm": 0.4983663558959961,
      "learning_rate": 0.00018,
      "loss": 1.2107,
      "step": 18
    },
    {
      "epoch": 0.0035309422040512915,
      "grad_norm": 0.7070329189300537,
      "learning_rate": 0.00019,
      "loss": 1.1058,
      "step": 19
    },
    {
      "epoch": 0.003716781267422412,
      "grad_norm": 0.36021122336387634,
      "learning_rate": 0.0002,
      "loss": 1.0866,
      "step": 20
    },
    {
      "epoch": 0.0039026203307935326,
      "grad_norm": 0.9074283242225647,
      "learning_rate": 0.00019999999572339194,
      "loss": 1.0985,
      "step": 21
    },
    {
      "epoch": 0.004088459394164653,
      "grad_norm": 0.4314013719558716,
      "learning_rate": 0.00019999998289356812,
      "loss": 0.9917,
      "step": 22
    },
    {
      "epoch": 0.004274298457535774,
      "grad_norm": 0.4853290319442749,
      "learning_rate": 0.0001999999615105296,
      "loss": 1.1093,
      "step": 23
    },
    {
      "epoch": 0.004460137520906894,
      "grad_norm": 0.4557550847530365,
      "learning_rate": 0.0001999999315742783,
      "loss": 1.3921,
      "step": 24
    },
    {
      "epoch": 0.004645976584278015,
      "grad_norm": 0.4223005771636963,
      "learning_rate": 0.00019999989308481668,
      "loss": 1.1424,
      "step": 25
    },
    {
      "epoch": 0.004831815647649135,
      "grad_norm": 0.6201490759849548,
      "learning_rate": 0.00019999984604214812,
      "loss": 1.4515,
      "step": 26
    },
    {
      "epoch": 0.005017654711020256,
      "grad_norm": 0.5055346488952637,
      "learning_rate": 0.00019999979044627657,
      "loss": 0.9974,
      "step": 27
    },
    {
      "epoch": 0.005203493774391377,
      "grad_norm": 0.4199519753456116,
      "learning_rate": 0.0001999997262972068,
      "loss": 1.3593,
      "step": 28
    },
    {
      "epoch": 0.0053893328377624975,
      "grad_norm": 0.42003971338272095,
      "learning_rate": 0.00019999965359494435,
      "loss": 1.1522,
      "step": 29
    },
    {
      "epoch": 0.0055751719011336185,
      "grad_norm": 0.4767492413520813,
      "learning_rate": 0.00019999957233949534,
      "loss": 1.1272,
      "step": 30
    },
    {
      "epoch": 0.005761010964504739,
      "grad_norm": 0.5115839242935181,
      "learning_rate": 0.00019999948253086685,
      "loss": 1.1056,
      "step": 31
    },
    {
      "epoch": 0.00594685002787586,
      "grad_norm": 0.48334285616874695,
      "learning_rate": 0.00019999938416906645,
      "loss": 1.0248,
      "step": 32
    },
    {
      "epoch": 0.00613268909124698,
      "grad_norm": 0.3692367672920227,
      "learning_rate": 0.00019999927725410262,
      "loss": 1.0031,
      "step": 33
    },
    {
      "epoch": 0.006318528154618101,
      "grad_norm": 0.38361677527427673,
      "learning_rate": 0.00019999916178598445,
      "loss": 1.221,
      "step": 34
    },
    {
      "epoch": 0.006504367217989221,
      "grad_norm": 0.34514594078063965,
      "learning_rate": 0.00019999903776472187,
      "loss": 1.1274,
      "step": 35
    },
    {
      "epoch": 0.006690206281360342,
      "grad_norm": 0.4714662730693817,
      "learning_rate": 0.00019999890519032545,
      "loss": 1.0562,
      "step": 36
    },
    {
      "epoch": 0.006876045344731463,
      "grad_norm": 0.47009018063545227,
      "learning_rate": 0.00019999876406280655,
      "loss": 1.2228,
      "step": 37
    },
    {
      "epoch": 0.007061884408102583,
      "grad_norm": 0.4776786267757416,
      "learning_rate": 0.00019999861438217722,
      "loss": 1.2817,
      "step": 38
    },
    {
      "epoch": 0.007247723471473704,
      "grad_norm": 0.42615750432014465,
      "learning_rate": 0.0001999984561484503,
      "loss": 1.2395,
      "step": 39
    },
    {
      "epoch": 0.007433562534844824,
      "grad_norm": 0.39289531111717224,
      "learning_rate": 0.00019999828936163928,
      "loss": 1.245,
      "step": 40
    },
    {
      "epoch": 0.007619401598215945,
      "grad_norm": 0.4392164349555969,
      "learning_rate": 0.00019999811402175848,
      "loss": 1.1604,
      "step": 41
    },
    {
      "epoch": 0.007805240661587065,
      "grad_norm": 0.7305529713630676,
      "learning_rate": 0.00019999793012882283,
      "loss": 1.2129,
      "step": 42
    },
    {
      "epoch": 0.007991079724958185,
      "grad_norm": 0.4402279257774353,
      "learning_rate": 0.0001999977376828481,
      "loss": 0.9151,
      "step": 43
    },
    {
      "epoch": 0.008176918788329306,
      "grad_norm": 0.5128088593482971,
      "learning_rate": 0.00019999753668385072,
      "loss": 1.3266,
      "step": 44
    },
    {
      "epoch": 0.008362757851700427,
      "grad_norm": 0.40549737215042114,
      "learning_rate": 0.00019999732713184797,
      "loss": 1.2442,
      "step": 45
    },
    {
      "epoch": 0.008548596915071548,
      "grad_norm": 0.5504630208015442,
      "learning_rate": 0.00019999710902685765,
      "loss": 1.36,
      "step": 46
    },
    {
      "epoch": 0.00873443597844267,
      "grad_norm": 0.4083457589149475,
      "learning_rate": 0.00019999688236889846,
      "loss": 1.1267,
      "step": 47
    },
    {
      "epoch": 0.008920275041813789,
      "grad_norm": 0.467245489358902,
      "learning_rate": 0.00019999664715798985,
      "loss": 1.0897,
      "step": 48
    },
    {
      "epoch": 0.00910611410518491,
      "grad_norm": 0.42215853929519653,
      "learning_rate": 0.00019999640339415188,
      "loss": 1.0611,
      "step": 49
    },
    {
      "epoch": 0.00929195316855603,
      "grad_norm": 0.48484039306640625,
      "learning_rate": 0.00019999615107740536,
      "loss": 1.015,
      "step": 50
    },
    {
      "epoch": 0.009477792231927152,
      "grad_norm": 0.4362781047821045,
      "learning_rate": 0.00019999589020777196,
      "loss": 1.0966,
      "step": 51
    },
    {
      "epoch": 0.00966363129529827,
      "grad_norm": 0.42648428678512573,
      "learning_rate": 0.00019999562078527395,
      "loss": 1.0121,
      "step": 52
    },
    {
      "epoch": 0.009849470358669392,
      "grad_norm": 0.4830460548400879,
      "learning_rate": 0.00019999534280993437,
      "loss": 1.0878,
      "step": 53
    },
    {
      "epoch": 0.010035309422040513,
      "grad_norm": 0.4462554156780243,
      "learning_rate": 0.000199995056281777,
      "loss": 1.324,
      "step": 54
    },
    {
      "epoch": 0.010221148485411634,
      "grad_norm": 0.3730710744857788,
      "learning_rate": 0.00019999476120082635,
      "loss": 0.9527,
      "step": 55
    },
    {
      "epoch": 0.010406987548782755,
      "grad_norm": 0.5048691034317017,
      "learning_rate": 0.00019999445756710765,
      "loss": 1.28,
      "step": 56
    },
    {
      "epoch": 0.010592826612153874,
      "grad_norm": 0.40299585461616516,
      "learning_rate": 0.00019999414538064686,
      "loss": 1.1119,
      "step": 57
    },
    {
      "epoch": 0.010778665675524995,
      "grad_norm": 0.4745088219642639,
      "learning_rate": 0.0001999938246414707,
      "loss": 1.029,
      "step": 58
    },
    {
      "epoch": 0.010964504738896116,
      "grad_norm": 0.4021121859550476,
      "learning_rate": 0.00019999349534960666,
      "loss": 0.9435,
      "step": 59
    },
    {
      "epoch": 0.011150343802267237,
      "grad_norm": 0.4799672067165375,
      "learning_rate": 0.0001999931575050828,
      "loss": 1.3012,
      "step": 60
    },
    {
      "epoch": 0.011336182865638356,
      "grad_norm": 0.4614189565181732,
      "learning_rate": 0.00019999281110792807,
      "loss": 1.1238,
      "step": 61
    },
    {
      "epoch": 0.011522021929009477,
      "grad_norm": 0.36898913979530334,
      "learning_rate": 0.00019999245615817213,
      "loss": 1.1332,
      "step": 62
    },
    {
      "epoch": 0.011707860992380598,
      "grad_norm": 0.4333168864250183,
      "learning_rate": 0.00019999209265584526,
      "loss": 1.1219,
      "step": 63
    },
    {
      "epoch": 0.01189370005575172,
      "grad_norm": 0.4542793333530426,
      "learning_rate": 0.0001999917206009786,
      "loss": 1.1224,
      "step": 64
    },
    {
      "epoch": 0.01207953911912284,
      "grad_norm": 0.41229432821273804,
      "learning_rate": 0.00019999133999360397,
      "loss": 1.2116,
      "step": 65
    },
    {
      "epoch": 0.01226537818249396,
      "grad_norm": 0.5130184888839722,
      "learning_rate": 0.00019999095083375395,
      "loss": 1.0225,
      "step": 66
    },
    {
      "epoch": 0.01245121724586508,
      "grad_norm": 0.45014601945877075,
      "learning_rate": 0.00019999055312146176,
      "loss": 1.2452,
      "step": 67
    },
    {
      "epoch": 0.012637056309236202,
      "grad_norm": 0.5515113472938538,
      "learning_rate": 0.00019999014685676144,
      "loss": 1.2329,
      "step": 68
    },
    {
      "epoch": 0.012822895372607323,
      "grad_norm": 0.43442612886428833,
      "learning_rate": 0.0001999897320396878,
      "loss": 0.951,
      "step": 69
    },
    {
      "epoch": 0.013008734435978442,
      "grad_norm": 0.46794602274894714,
      "learning_rate": 0.00019998930867027624,
      "loss": 1.155,
      "step": 70
    },
    {
      "epoch": 0.013194573499349563,
      "grad_norm": 0.45471006631851196,
      "learning_rate": 0.000199988876748563,
      "loss": 0.9123,
      "step": 71
    },
    {
      "epoch": 0.013380412562720684,
      "grad_norm": 0.3680677115917206,
      "learning_rate": 0.00019998843627458505,
      "loss": 1.1987,
      "step": 72
    },
    {
      "epoch": 0.013566251626091805,
      "grad_norm": 0.5155819654464722,
      "learning_rate": 0.00019998798724838005,
      "loss": 1.1141,
      "step": 73
    },
    {
      "epoch": 0.013752090689462926,
      "grad_norm": 0.40624114871025085,
      "learning_rate": 0.0001999875296699864,
      "loss": 1.2618,
      "step": 74
    },
    {
      "epoch": 0.013937929752834045,
      "grad_norm": 0.437930166721344,
      "learning_rate": 0.00019998706353944318,
      "loss": 1.0078,
      "step": 75
    },
    {
      "epoch": 0.014123768816205166,
      "grad_norm": 0.710334837436676,
      "learning_rate": 0.00019998658885679035,
      "loss": 1.4127,
      "step": 76
    },
    {
      "epoch": 0.014309607879576287,
      "grad_norm": 0.5053734183311462,
      "learning_rate": 0.00019998610562206847,
      "loss": 1.0414,
      "step": 77
    },
    {
      "epoch": 0.014495446942947408,
      "grad_norm": 0.5894109606742859,
      "learning_rate": 0.00019998561383531888,
      "loss": 1.3005,
      "step": 78
    },
    {
      "epoch": 0.014681286006318527,
      "grad_norm": 0.39885780215263367,
      "learning_rate": 0.00019998511349658368,
      "loss": 1.0932,
      "step": 79
    },
    {
      "epoch": 0.014867125069689648,
      "grad_norm": 0.4869062006473541,
      "learning_rate": 0.00019998460460590554,
      "loss": 1.1018,
      "step": 80
    },
    {
      "epoch": 0.01505296413306077,
      "grad_norm": 0.3895723521709442,
      "learning_rate": 0.00019998408716332815,
      "loss": 1.1326,
      "step": 81
    },
    {
      "epoch": 0.01523880319643189,
      "grad_norm": 0.49565768241882324,
      "learning_rate": 0.00019998356116889568,
      "loss": 1.2525,
      "step": 82
    },
    {
      "epoch": 0.015424642259803011,
      "grad_norm": 0.46484237909317017,
      "learning_rate": 0.00019998302662265312,
      "loss": 1.1684,
      "step": 83
    },
    {
      "epoch": 0.01561048132317413,
      "grad_norm": 0.4285367727279663,
      "learning_rate": 0.00019998248352464618,
      "loss": 1.2158,
      "step": 84
    },
    {
      "epoch": 0.015796320386545253,
      "grad_norm": 0.4140075743198395,
      "learning_rate": 0.00019998193187492137,
      "loss": 1.1595,
      "step": 85
    },
    {
      "epoch": 0.01598215944991637,
      "grad_norm": 0.4073302447795868,
      "learning_rate": 0.0001999813716735258,
      "loss": 0.9106,
      "step": 86
    },
    {
      "epoch": 0.016167998513287492,
      "grad_norm": 0.47974786162376404,
      "learning_rate": 0.00019998080292050744,
      "loss": 1.0553,
      "step": 87
    },
    {
      "epoch": 0.016353837576658613,
      "grad_norm": 0.47133541107177734,
      "learning_rate": 0.00019998022561591494,
      "loss": 1.1256,
      "step": 88
    },
    {
      "epoch": 0.016539676640029734,
      "grad_norm": 0.45809057354927063,
      "learning_rate": 0.00019997963975979763,
      "loss": 1.0548,
      "step": 89
    },
    {
      "epoch": 0.016725515703400855,
      "grad_norm": 0.358278751373291,
      "learning_rate": 0.00019997904535220564,
      "loss": 0.8885,
      "step": 90
    },
    {
      "epoch": 0.016911354766771976,
      "grad_norm": 0.4621461033821106,
      "learning_rate": 0.00019997844239318983,
      "loss": 0.9386,
      "step": 91
    },
    {
      "epoch": 0.017097193830143097,
      "grad_norm": 0.5164687037467957,
      "learning_rate": 0.00019997783088280177,
      "loss": 1.1284,
      "step": 92
    },
    {
      "epoch": 0.017283032893514218,
      "grad_norm": 0.4301346242427826,
      "learning_rate": 0.00019997721082109376,
      "loss": 1.1444,
      "step": 93
    },
    {
      "epoch": 0.01746887195688534,
      "grad_norm": 0.5308482646942139,
      "learning_rate": 0.0001999765822081188,
      "loss": 1.1924,
      "step": 94
    },
    {
      "epoch": 0.017654711020256456,
      "grad_norm": 0.45287665724754333,
      "learning_rate": 0.0001999759450439307,
      "loss": 1.1199,
      "step": 95
    },
    {
      "epoch": 0.017840550083627577,
      "grad_norm": 0.42649218440055847,
      "learning_rate": 0.00019997529932858394,
      "loss": 1.3226,
      "step": 96
    },
    {
      "epoch": 0.018026389146998698,
      "grad_norm": 0.4206051826477051,
      "learning_rate": 0.00019997464506213376,
      "loss": 0.9306,
      "step": 97
    },
    {
      "epoch": 0.01821222821036982,
      "grad_norm": 0.4551459848880768,
      "learning_rate": 0.00019997398224463612,
      "loss": 1.1994,
      "step": 98
    },
    {
      "epoch": 0.01839806727374094,
      "grad_norm": 0.44016626477241516,
      "learning_rate": 0.0001999733108761477,
      "loss": 1.0889,
      "step": 99
    },
    {
      "epoch": 0.01858390633711206,
      "grad_norm": 0.405900776386261,
      "learning_rate": 0.00019997263095672594,
      "loss": 0.8563,
      "step": 100
    },
    {
      "epoch": 0.018769745400483182,
      "grad_norm": 0.4155542254447937,
      "learning_rate": 0.00019997194248642898,
      "loss": 1.1767,
      "step": 101
    },
    {
      "epoch": 0.018955584463854303,
      "grad_norm": 0.4083574116230011,
      "learning_rate": 0.0001999712454653157,
      "loss": 1.1509,
      "step": 102
    },
    {
      "epoch": 0.019141423527225424,
      "grad_norm": 2.3378918170928955,
      "learning_rate": 0.00019997053989344573,
      "loss": 4.8463,
      "step": 103
    },
    {
      "epoch": 0.01932726259059654,
      "grad_norm": 0.421448290348053,
      "learning_rate": 0.00019996982577087945,
      "loss": 1.1293,
      "step": 104
    },
    {
      "epoch": 0.019513101653967663,
      "grad_norm": 0.40957626700401306,
      "learning_rate": 0.00019996910309767784,
      "loss": 1.2171,
      "step": 105
    },
    {
      "epoch": 0.019698940717338784,
      "grad_norm": 0.4505980312824249,
      "learning_rate": 0.00019996837187390284,
      "loss": 0.9156,
      "step": 106
    },
    {
      "epoch": 0.019884779780709905,
      "grad_norm": 0.39445918798446655,
      "learning_rate": 0.0001999676320996169,
      "loss": 1.1852,
      "step": 107
    },
    {
      "epoch": 0.020070618844081026,
      "grad_norm": 0.41280612349510193,
      "learning_rate": 0.00019996688377488334,
      "loss": 1.0868,
      "step": 108
    },
    {
      "epoch": 0.020256457907452147,
      "grad_norm": 0.3600105047225952,
      "learning_rate": 0.00019996612689976618,
      "loss": 1.0886,
      "step": 109
    },
    {
      "epoch": 0.020442296970823268,
      "grad_norm": 0.44562339782714844,
      "learning_rate": 0.00019996536147433012,
      "loss": 1.1261,
      "step": 110
    },
    {
      "epoch": 0.02062813603419439,
      "grad_norm": 0.5070035457611084,
      "learning_rate": 0.00019996458749864062,
      "loss": 1.3632,
      "step": 111
    },
    {
      "epoch": 0.02081397509756551,
      "grad_norm": 0.4161199927330017,
      "learning_rate": 0.0001999638049727639,
      "loss": 0.9583,
      "step": 112
    },
    {
      "epoch": 0.020999814160936627,
      "grad_norm": 0.32975199818611145,
      "learning_rate": 0.00019996301389676687,
      "loss": 0.907,
      "step": 113
    },
    {
      "epoch": 0.021185653224307748,
      "grad_norm": 0.4182499945163727,
      "learning_rate": 0.00019996221427071725,
      "loss": 1.3569,
      "step": 114
    },
    {
      "epoch": 0.02137149228767887,
      "grad_norm": 0.5010649561882019,
      "learning_rate": 0.00019996140609468343,
      "loss": 1.0246,
      "step": 115
    },
    {
      "epoch": 0.02155733135104999,
      "grad_norm": 0.40202227234840393,
      "learning_rate": 0.00019996058936873443,
      "loss": 0.924,
      "step": 116
    },
    {
      "epoch": 0.02174317041442111,
      "grad_norm": 0.4302928149700165,
      "learning_rate": 0.00019995976409294022,
      "loss": 1.2835,
      "step": 117
    },
    {
      "epoch": 0.021929009477792232,
      "grad_norm": 0.4092571437358856,
      "learning_rate": 0.00019995893026737133,
      "loss": 1.2428,
      "step": 118
    },
    {
      "epoch": 0.022114848541163353,
      "grad_norm": 0.5921717882156372,
      "learning_rate": 0.0001999580878920991,
      "loss": 1.2466,
      "step": 119
    },
    {
      "epoch": 0.022300687604534474,
      "grad_norm": 0.4174133837223053,
      "learning_rate": 0.00019995723696719555,
      "loss": 1.0471,
      "step": 120
    },
    {
      "epoch": 0.022486526667905595,
      "grad_norm": 0.40855398774147034,
      "learning_rate": 0.00019995637749273352,
      "loss": 0.9747,
      "step": 121
    },
    {
      "epoch": 0.022672365731276713,
      "grad_norm": 0.43846818804740906,
      "learning_rate": 0.00019995550946878647,
      "loss": 1.3309,
      "step": 122
    },
    {
      "epoch": 0.022858204794647834,
      "grad_norm": 0.38275861740112305,
      "learning_rate": 0.00019995463289542868,
      "loss": 1.1033,
      "step": 123
    },
    {
      "epoch": 0.023044043858018955,
      "grad_norm": 0.4742163419723511,
      "learning_rate": 0.0001999537477727351,
      "loss": 1.145,
      "step": 124
    },
    {
      "epoch": 0.023229882921390076,
      "grad_norm": 0.4117315113544464,
      "learning_rate": 0.00019995285410078143,
      "loss": 1.021,
      "step": 125
    },
    {
      "epoch": 0.023415721984761197,
      "grad_norm": 0.4802894592285156,
      "learning_rate": 0.00019995195187964414,
      "loss": 1.2366,
      "step": 126
    },
    {
      "epoch": 0.023601561048132318,
      "grad_norm": 0.4284798502922058,
      "learning_rate": 0.00019995104110940036,
      "loss": 1.0542,
      "step": 127
    },
    {
      "epoch": 0.02378740011150344,
      "grad_norm": 0.48981085419654846,
      "learning_rate": 0.00019995012179012807,
      "loss": 1.2871,
      "step": 128
    },
    {
      "epoch": 0.02397323917487456,
      "grad_norm": 0.4514794647693634,
      "learning_rate": 0.00019994919392190576,
      "loss": 1.3027,
      "step": 129
    },
    {
      "epoch": 0.02415907823824568,
      "grad_norm": 0.4130529463291168,
      "learning_rate": 0.00019994825750481291,
      "loss": 1.1686,
      "step": 130
    },
    {
      "epoch": 0.024344917301616798,
      "grad_norm": 0.5316373109817505,
      "learning_rate": 0.0001999473125389296,
      "loss": 1.3746,
      "step": 131
    },
    {
      "epoch": 0.02453075636498792,
      "grad_norm": 0.49608317017555237,
      "learning_rate": 0.00019994635902433663,
      "loss": 1.2536,
      "step": 132
    },
    {
      "epoch": 0.02471659542835904,
      "grad_norm": 0.38635924458503723,
      "learning_rate": 0.00019994539696111555,
      "loss": 1.2128,
      "step": 133
    },
    {
      "epoch": 0.02490243449173016,
      "grad_norm": 0.37796691060066223,
      "learning_rate": 0.00019994442634934867,
      "loss": 1.2026,
      "step": 134
    },
    {
      "epoch": 0.025088273555101282,
      "grad_norm": 0.4202409088611603,
      "learning_rate": 0.00019994344718911897,
      "loss": 1.2832,
      "step": 135
    },
    {
      "epoch": 0.025274112618472403,
      "grad_norm": 0.4410860538482666,
      "learning_rate": 0.00019994245948051024,
      "loss": 1.1883,
      "step": 136
    },
    {
      "epoch": 0.025459951681843524,
      "grad_norm": 1.8298826217651367,
      "learning_rate": 0.00019994146322360694,
      "loss": 4.4886,
      "step": 137
    },
    {
      "epoch": 0.025645790745214645,
      "grad_norm": 0.4328976273536682,
      "learning_rate": 0.00019994045841849429,
      "loss": 1.1451,
      "step": 138
    },
    {
      "epoch": 0.025831629808585766,
      "grad_norm": 0.3915058374404907,
      "learning_rate": 0.00019993944506525825,
      "loss": 0.9214,
      "step": 139
    },
    {
      "epoch": 0.026017468871956884,
      "grad_norm": 0.44726213812828064,
      "learning_rate": 0.00019993842316398544,
      "loss": 1.1946,
      "step": 140
    },
    {
      "epoch": 0.026203307935328005,
      "grad_norm": 0.47718754410743713,
      "learning_rate": 0.00019993739271476332,
      "loss": 1.1276,
      "step": 141
    },
    {
      "epoch": 0.026389146998699126,
      "grad_norm": 0.3767591118812561,
      "learning_rate": 0.00019993635371768002,
      "loss": 0.9029,
      "step": 142
    },
    {
      "epoch": 0.026574986062070247,
      "grad_norm": 0.3988798260688782,
      "learning_rate": 0.00019993530617282436,
      "loss": 1.0303,
      "step": 143
    },
    {
      "epoch": 0.026760825125441368,
      "grad_norm": 0.3820064067840576,
      "learning_rate": 0.00019993425008028602,
      "loss": 1.116,
      "step": 144
    },
    {
      "epoch": 0.02694666418881249,
      "grad_norm": 0.4157227873802185,
      "learning_rate": 0.00019993318544015527,
      "loss": 1.0564,
      "step": 145
    },
    {
      "epoch": 0.02713250325218361,
      "grad_norm": 0.4232051372528076,
      "learning_rate": 0.00019993211225252316,
      "loss": 1.2417,
      "step": 146
    },
    {
      "epoch": 0.02731834231555473,
      "grad_norm": 0.47376081347465515,
      "learning_rate": 0.00019993103051748153,
      "loss": 1.0504,
      "step": 147
    },
    {
      "epoch": 0.02750418137892585,
      "grad_norm": 0.4919112026691437,
      "learning_rate": 0.00019992994023512292,
      "loss": 1.3752,
      "step": 148
    },
    {
      "epoch": 0.02769002044229697,
      "grad_norm": 0.5118665099143982,
      "learning_rate": 0.00019992884140554047,
      "loss": 1.0661,
      "step": 149
    },
    {
      "epoch": 0.02787585950566809,
      "grad_norm": 0.44284725189208984,
      "learning_rate": 0.00019992773402882828,
      "loss": 1.233,
      "step": 150
    },
    {
      "epoch": 0.02806169856903921,
      "grad_norm": 0.4117078483104706,
      "learning_rate": 0.000199926618105081,
      "loss": 1.169,
      "step": 151
    },
    {
      "epoch": 0.028247537632410332,
      "grad_norm": 0.549756646156311,
      "learning_rate": 0.00019992549363439412,
      "loss": 1.2191,
      "step": 152
    },
    {
      "epoch": 0.028433376695781453,
      "grad_norm": 0.39329949021339417,
      "learning_rate": 0.00019992436061686382,
      "loss": 1.2276,
      "step": 153
    },
    {
      "epoch": 0.028619215759152574,
      "grad_norm": 0.4098329246044159,
      "learning_rate": 0.00019992321905258699,
      "loss": 1.1029,
      "step": 154
    },
    {
      "epoch": 0.028805054822523695,
      "grad_norm": 0.5076633095741272,
      "learning_rate": 0.00019992206894166125,
      "loss": 1.1196,
      "step": 155
    },
    {
      "epoch": 0.028990893885894816,
      "grad_norm": 0.41998520493507385,
      "learning_rate": 0.000199920910284185,
      "loss": 1.1913,
      "step": 156
    },
    {
      "epoch": 0.029176732949265937,
      "grad_norm": 0.4319917857646942,
      "learning_rate": 0.00019991974308025733,
      "loss": 1.1782,
      "step": 157
    },
    {
      "epoch": 0.029362572012637055,
      "grad_norm": 0.6997054815292358,
      "learning_rate": 0.00019991856732997807,
      "loss": 1.0852,
      "step": 158
    },
    {
      "epoch": 0.029548411076008176,
      "grad_norm": 0.48971933126449585,
      "learning_rate": 0.00019991738303344782,
      "loss": 1.1819,
      "step": 159
    },
    {
      "epoch": 0.029734250139379297,
      "grad_norm": 0.5128749012947083,
      "learning_rate": 0.00019991619019076783,
      "loss": 1.3655,
      "step": 160
    },
    {
      "epoch": 0.029920089202750418,
      "grad_norm": 0.41675087809562683,
      "learning_rate": 0.00019991498880204018,
      "loss": 1.0505,
      "step": 161
    },
    {
      "epoch": 0.03010592826612154,
      "grad_norm": 0.48276373744010925,
      "learning_rate": 0.00019991377886736754,
      "loss": 1.2697,
      "step": 162
    },
    {
      "epoch": 0.03029176732949266,
      "grad_norm": 0.4491844177246094,
      "learning_rate": 0.00019991256038685352,
      "loss": 1.2506,
      "step": 163
    },
    {
      "epoch": 0.03047760639286378,
      "grad_norm": 0.4582407772541046,
      "learning_rate": 0.0001999113333606022,
      "loss": 1.2433,
      "step": 164
    },
    {
      "epoch": 0.0306634454562349,
      "grad_norm": 0.4408119320869446,
      "learning_rate": 0.00019991009778871864,
      "loss": 1.1231,
      "step": 165
    },
    {
      "epoch": 0.030849284519606023,
      "grad_norm": 0.4083813726902008,
      "learning_rate": 0.00019990885367130847,
      "loss": 1.0707,
      "step": 166
    },
    {
      "epoch": 0.031035123582977144,
      "grad_norm": 0.5064361095428467,
      "learning_rate": 0.00019990760100847813,
      "loss": 1.0206,
      "step": 167
    },
    {
      "epoch": 0.03122096264634826,
      "grad_norm": 0.41701388359069824,
      "learning_rate": 0.00019990633980033473,
      "loss": 1.2338,
      "step": 168
    },
    {
      "epoch": 0.031406801709719386,
      "grad_norm": 0.457020103931427,
      "learning_rate": 0.00019990507004698615,
      "loss": 1.2124,
      "step": 169
    },
    {
      "epoch": 0.031592640773090507,
      "grad_norm": 0.45959389209747314,
      "learning_rate": 0.00019990379174854102,
      "loss": 1.1267,
      "step": 170
    },
    {
      "epoch": 0.03177847983646163,
      "grad_norm": 0.45844098925590515,
      "learning_rate": 0.00019990250490510867,
      "loss": 1.1652,
      "step": 171
    },
    {
      "epoch": 0.03196431889983274,
      "grad_norm": 0.5090590119361877,
      "learning_rate": 0.00019990120951679914,
      "loss": 1.0736,
      "step": 172
    },
    {
      "epoch": 0.03215015796320386,
      "grad_norm": 0.4741423726081848,
      "learning_rate": 0.00019989990558372326,
      "loss": 1.1206,
      "step": 173
    },
    {
      "epoch": 0.032335997026574984,
      "grad_norm": 0.4410255253314972,
      "learning_rate": 0.00019989859310599254,
      "loss": 1.2636,
      "step": 174
    },
    {
      "epoch": 0.032521836089946105,
      "grad_norm": 0.4406038522720337,
      "learning_rate": 0.00019989727208371926,
      "loss": 0.987,
      "step": 175
    },
    {
      "epoch": 0.032707675153317226,
      "grad_norm": 0.4072116017341614,
      "learning_rate": 0.00019989594251701635,
      "loss": 1.1572,
      "step": 176
    },
    {
      "epoch": 0.032893514216688347,
      "grad_norm": 0.4296364486217499,
      "learning_rate": 0.00019989460440599758,
      "loss": 1.046,
      "step": 177
    },
    {
      "epoch": 0.03307935328005947,
      "grad_norm": 0.4294045865535736,
      "learning_rate": 0.00019989325775077743,
      "loss": 0.9476,
      "step": 178
    },
    {
      "epoch": 0.03326519234343059,
      "grad_norm": 0.5257508158683777,
      "learning_rate": 0.00019989190255147103,
      "loss": 1.0242,
      "step": 179
    },
    {
      "epoch": 0.03345103140680171,
      "grad_norm": 0.36463358998298645,
      "learning_rate": 0.00019989053880819428,
      "loss": 0.7214,
      "step": 180
    },
    {
      "epoch": 0.03363687047017283,
      "grad_norm": 0.49668338894844055,
      "learning_rate": 0.00019988916652106392,
      "loss": 1.2282,
      "step": 181
    },
    {
      "epoch": 0.03382270953354395,
      "grad_norm": 0.47722092270851135,
      "learning_rate": 0.0001998877856901972,
      "loss": 1.2502,
      "step": 182
    },
    {
      "epoch": 0.03400854859691507,
      "grad_norm": 0.40731075406074524,
      "learning_rate": 0.00019988639631571232,
      "loss": 1.063,
      "step": 183
    },
    {
      "epoch": 0.034194387660286193,
      "grad_norm": 0.45999324321746826,
      "learning_rate": 0.00019988499839772804,
      "loss": 0.9978,
      "step": 184
    },
    {
      "epoch": 0.034380226723657314,
      "grad_norm": 0.4087058901786804,
      "learning_rate": 0.00019988359193636398,
      "loss": 1.231,
      "step": 185
    },
    {
      "epoch": 0.034566065787028435,
      "grad_norm": 0.3663274645805359,
      "learning_rate": 0.00019988217693174044,
      "loss": 0.8827,
      "step": 186
    },
    {
      "epoch": 0.034751904850399556,
      "grad_norm": 0.6143786907196045,
      "learning_rate": 0.00019988075338397843,
      "loss": 1.1063,
      "step": 187
    },
    {
      "epoch": 0.03493774391377068,
      "grad_norm": 0.36634212732315063,
      "learning_rate": 0.0001998793212931997,
      "loss": 1.1026,
      "step": 188
    },
    {
      "epoch": 0.0351235829771418,
      "grad_norm": 0.48718002438545227,
      "learning_rate": 0.00019987788065952677,
      "loss": 1.1274,
      "step": 189
    },
    {
      "epoch": 0.03530942204051291,
      "grad_norm": 0.4710601270198822,
      "learning_rate": 0.00019987643148308286,
      "loss": 1.2466,
      "step": 190
    },
    {
      "epoch": 0.035495261103884033,
      "grad_norm": 0.4751073718070984,
      "learning_rate": 0.00019987497376399186,
      "loss": 1.1662,
      "step": 191
    },
    {
      "epoch": 0.035681100167255154,
      "grad_norm": 0.35801300406455994,
      "learning_rate": 0.00019987350750237852,
      "loss": 0.926,
      "step": 192
    },
    {
      "epoch": 0.035866939230626275,
      "grad_norm": 0.40735355019569397,
      "learning_rate": 0.00019987203269836824,
      "loss": 1.2028,
      "step": 193
    },
    {
      "epoch": 0.036052778293997396,
      "grad_norm": 0.3875338137149811,
      "learning_rate": 0.00019987054935208712,
      "loss": 1.0547,
      "step": 194
    },
    {
      "epoch": 0.03623861735736852,
      "grad_norm": 0.3859501779079437,
      "learning_rate": 0.0001998690574636621,
      "loss": 0.9658,
      "step": 195
    },
    {
      "epoch": 0.03642445642073964,
      "grad_norm": 0.47865334153175354,
      "learning_rate": 0.0001998675570332207,
      "loss": 1.1934,
      "step": 196
    },
    {
      "epoch": 0.03661029548411076,
      "grad_norm": 0.5451048016548157,
      "learning_rate": 0.00019986604806089133,
      "loss": 1.1744,
      "step": 197
    },
    {
      "epoch": 0.03679613454748188,
      "grad_norm": 0.40316689014434814,
      "learning_rate": 0.00019986453054680306,
      "loss": 1.0095,
      "step": 198
    },
    {
      "epoch": 0.036981973610853,
      "grad_norm": 0.3943663537502289,
      "learning_rate": 0.00019986300449108564,
      "loss": 1.1106,
      "step": 199
    },
    {
      "epoch": 0.03716781267422412,
      "grad_norm": 0.4198141396045685,
      "learning_rate": 0.00019986146989386957,
      "loss": 1.041,
      "step": 200
    },
    {
      "epoch": 0.03735365173759524,
      "grad_norm": 0.3879221975803375,
      "learning_rate": 0.0001998599267552862,
      "loss": 0.948,
      "step": 201
    },
    {
      "epoch": 0.037539490800966364,
      "grad_norm": 0.39751577377319336,
      "learning_rate": 0.00019985837507546745,
      "loss": 1.0038,
      "step": 202
    },
    {
      "epoch": 0.037725329864337485,
      "grad_norm": 0.37878310680389404,
      "learning_rate": 0.00019985681485454605,
      "loss": 1.1636,
      "step": 203
    },
    {
      "epoch": 0.037911168927708606,
      "grad_norm": 0.48006248474121094,
      "learning_rate": 0.00019985524609265548,
      "loss": 1.2288,
      "step": 204
    },
    {
      "epoch": 0.03809700799107973,
      "grad_norm": 0.40840205550193787,
      "learning_rate": 0.0001998536687899299,
      "loss": 1.1478,
      "step": 205
    },
    {
      "epoch": 0.03828284705445085,
      "grad_norm": 0.500511884689331,
      "learning_rate": 0.00019985208294650422,
      "loss": 1.0763,
      "step": 206
    },
    {
      "epoch": 0.03846868611782197,
      "grad_norm": 0.45608195662498474,
      "learning_rate": 0.00019985048856251405,
      "loss": 1.2623,
      "step": 207
    },
    {
      "epoch": 0.03865452518119308,
      "grad_norm": 0.4747142195701599,
      "learning_rate": 0.0001998488856380958,
      "loss": 1.2437,
      "step": 208
    },
    {
      "epoch": 0.038840364244564204,
      "grad_norm": 0.5317223072052002,
      "learning_rate": 0.00019984727417338656,
      "loss": 1.114,
      "step": 209
    },
    {
      "epoch": 0.039026203307935325,
      "grad_norm": 0.4821191728115082,
      "learning_rate": 0.00019984565416852415,
      "loss": 1.2248,
      "step": 210
    },
    {
      "epoch": 0.039212042371306446,
      "grad_norm": 0.42245104908943176,
      "learning_rate": 0.00019984402562364716,
      "loss": 1.2469,
      "step": 211
    },
    {
      "epoch": 0.03939788143467757,
      "grad_norm": 1.796623945236206,
      "learning_rate": 0.00019984238853889484,
      "loss": 4.2855,
      "step": 212
    },
    {
      "epoch": 0.03958372049804869,
      "grad_norm": 0.44016581773757935,
      "learning_rate": 0.00019984074291440728,
      "loss": 1.066,
      "step": 213
    },
    {
      "epoch": 0.03976955956141981,
      "grad_norm": 0.4538607597351074,
      "learning_rate": 0.00019983908875032515,
      "loss": 1.3343,
      "step": 214
    },
    {
      "epoch": 0.03995539862479093,
      "grad_norm": 0.4110296368598938,
      "learning_rate": 0.00019983742604679002,
      "loss": 1.2131,
      "step": 215
    },
    {
      "epoch": 0.04014123768816205,
      "grad_norm": 0.606137216091156,
      "learning_rate": 0.00019983575480394402,
      "loss": 0.9551,
      "step": 216
    },
    {
      "epoch": 0.04032707675153317,
      "grad_norm": 0.4060904085636139,
      "learning_rate": 0.00019983407502193015,
      "loss": 1.0849,
      "step": 217
    },
    {
      "epoch": 0.04051291581490429,
      "grad_norm": 0.41577842831611633,
      "learning_rate": 0.00019983238670089214,
      "loss": 1.1291,
      "step": 218
    },
    {
      "epoch": 0.040698754878275414,
      "grad_norm": 0.43058526515960693,
      "learning_rate": 0.00019983068984097426,
      "loss": 1.2347,
      "step": 219
    },
    {
      "epoch": 0.040884593941646535,
      "grad_norm": 0.3922870457172394,
      "learning_rate": 0.00019982898444232173,
      "loss": 1.0996,
      "step": 220
    },
    {
      "epoch": 0.041070433005017656,
      "grad_norm": 0.49691110849380493,
      "learning_rate": 0.00019982727050508038,
      "loss": 1.1778,
      "step": 221
    },
    {
      "epoch": 0.04125627206838878,
      "grad_norm": 0.45880937576293945,
      "learning_rate": 0.00019982554802939687,
      "loss": 1.1532,
      "step": 222
    },
    {
      "epoch": 0.0414421111317599,
      "grad_norm": 0.49630632996559143,
      "learning_rate": 0.00019982381701541847,
      "loss": 1.1742,
      "step": 223
    },
    {
      "epoch": 0.04162795019513102,
      "grad_norm": 0.3757645785808563,
      "learning_rate": 0.00019982207746329324,
      "loss": 0.8219,
      "step": 224
    },
    {
      "epoch": 0.04181378925850214,
      "grad_norm": 0.423665851354599,
      "learning_rate": 0.00019982032937316998,
      "loss": 0.9377,
      "step": 225
    },
    {
      "epoch": 0.041999628321873254,
      "grad_norm": 0.5444109439849854,
      "learning_rate": 0.0001998185727451982,
      "loss": 1.02,
      "step": 226
    },
    {
      "epoch": 0.042185467385244375,
      "grad_norm": 0.4525771141052246,
      "learning_rate": 0.00019981680757952822,
      "loss": 1.2185,
      "step": 227
    },
    {
      "epoch": 0.042371306448615496,
      "grad_norm": 0.4341532289981842,
      "learning_rate": 0.0001998150338763109,
      "loss": 1.0239,
      "step": 228
    },
    {
      "epoch": 0.04255714551198662,
      "grad_norm": 0.4187489151954651,
      "learning_rate": 0.00019981325163569803,
      "loss": 0.9313,
      "step": 229
    },
    {
      "epoch": 0.04274298457535774,
      "grad_norm": 0.38820531964302063,
      "learning_rate": 0.000199811460857842,
      "loss": 0.9375,
      "step": 230
    },
    {
      "epoch": 0.04292882363872886,
      "grad_norm": 0.36113470792770386,
      "learning_rate": 0.00019980966154289602,
      "loss": 1.0695,
      "step": 231
    },
    {
      "epoch": 0.04311466270209998,
      "grad_norm": 4.252074241638184,
      "learning_rate": 0.00019980785369101396,
      "loss": 4.6367,
      "step": 232
    },
    {
      "epoch": 0.0433005017654711,
      "grad_norm": 0.43251553177833557,
      "learning_rate": 0.00019980603730235044,
      "loss": 1.0693,
      "step": 233
    },
    {
      "epoch": 0.04348634082884222,
      "grad_norm": 0.47350192070007324,
      "learning_rate": 0.00019980421237706088,
      "loss": 1.2545,
      "step": 234
    },
    {
      "epoch": 0.04367217989221334,
      "grad_norm": 0.3853001296520233,
      "learning_rate": 0.00019980237891530133,
      "loss": 0.8562,
      "step": 235
    },
    {
      "epoch": 0.043858018955584464,
      "grad_norm": 0.46240493655204773,
      "learning_rate": 0.00019980053691722856,
      "loss": 1.2941,
      "step": 236
    },
    {
      "epoch": 0.044043858018955585,
      "grad_norm": 0.4637151062488556,
      "learning_rate": 0.00019979868638300021,
      "loss": 1.223,
      "step": 237
    },
    {
      "epoch": 0.044229697082326706,
      "grad_norm": 0.40852031111717224,
      "learning_rate": 0.00019979682731277452,
      "loss": 0.8145,
      "step": 238
    },
    {
      "epoch": 0.04441553614569783,
      "grad_norm": 0.4328407943248749,
      "learning_rate": 0.0001997949597067105,
      "loss": 1.1434,
      "step": 239
    },
    {
      "epoch": 0.04460137520906895,
      "grad_norm": 0.42404770851135254,
      "learning_rate": 0.00019979308356496786,
      "loss": 0.9842,
      "step": 240
    },
    {
      "epoch": 0.04478721427244007,
      "grad_norm": 0.40816017985343933,
      "learning_rate": 0.00019979119888770714,
      "loss": 1.0351,
      "step": 241
    },
    {
      "epoch": 0.04497305333581119,
      "grad_norm": 0.44897130131721497,
      "learning_rate": 0.00019978930567508948,
      "loss": 1.0483,
      "step": 242
    },
    {
      "epoch": 0.04515889239918231,
      "grad_norm": 0.40088385343551636,
      "learning_rate": 0.00019978740392727687,
      "loss": 0.9936,
      "step": 243
    },
    {
      "epoch": 0.045344731462553425,
      "grad_norm": 0.4536771774291992,
      "learning_rate": 0.00019978549364443188,
      "loss": 1.2624,
      "step": 244
    },
    {
      "epoch": 0.045530570525924546,
      "grad_norm": 0.4574795067310333,
      "learning_rate": 0.00019978357482671797,
      "loss": 1.1316,
      "step": 245
    },
    {
      "epoch": 0.04571640958929567,
      "grad_norm": 0.41690966486930847,
      "learning_rate": 0.00019978164747429925,
      "loss": 1.1572,
      "step": 246
    },
    {
      "epoch": 0.04590224865266679,
      "grad_norm": 0.4959973692893982,
      "learning_rate": 0.0001997797115873406,
      "loss": 1.2667,
      "step": 247
    },
    {
      "epoch": 0.04608808771603791,
      "grad_norm": 0.4050540626049042,
      "learning_rate": 0.0001997777671660075,
      "loss": 1.2102,
      "step": 248
    },
    {
      "epoch": 0.04627392677940903,
      "grad_norm": 0.3968220055103302,
      "learning_rate": 0.00019977581421046637,
      "loss": 1.3496,
      "step": 249
    },
    {
      "epoch": 0.04645976584278015,
      "grad_norm": 0.38303324580192566,
      "learning_rate": 0.0001997738527208842,
      "loss": 1.0241,
      "step": 250
    },
    {
      "epoch": 0.04664560490615127,
      "grad_norm": 0.36732155084609985,
      "learning_rate": 0.00019977188269742878,
      "loss": 0.9691,
      "step": 251
    },
    {
      "epoch": 0.04683144396952239,
      "grad_norm": 0.44628089666366577,
      "learning_rate": 0.00019976990414026857,
      "loss": 1.0435,
      "step": 252
    },
    {
      "epoch": 0.047017283032893514,
      "grad_norm": 0.42364564538002014,
      "learning_rate": 0.00019976791704957284,
      "loss": 1.1622,
      "step": 253
    },
    {
      "epoch": 0.047203122096264635,
      "grad_norm": 0.4465533196926117,
      "learning_rate": 0.00019976592142551152,
      "loss": 1.0775,
      "step": 254
    },
    {
      "epoch": 0.047388961159635756,
      "grad_norm": 0.41187724471092224,
      "learning_rate": 0.00019976391726825535,
      "loss": 0.907,
      "step": 255
    },
    {
      "epoch": 0.04757480022300688,
      "grad_norm": 0.41359013319015503,
      "learning_rate": 0.0001997619045779757,
      "loss": 1.2421,
      "step": 256
    },
    {
      "epoch": 0.047760639286378,
      "grad_norm": 0.5085680484771729,
      "learning_rate": 0.0001997598833548447,
      "loss": 1.2285,
      "step": 257
    },
    {
      "epoch": 0.04794647834974912,
      "grad_norm": 0.5501344799995422,
      "learning_rate": 0.00019975785359903535,
      "loss": 1.2846,
      "step": 258
    },
    {
      "epoch": 0.04813231741312024,
      "grad_norm": 0.3470396101474762,
      "learning_rate": 0.00019975581531072112,
      "loss": 0.9717,
      "step": 259
    },
    {
      "epoch": 0.04831815647649136,
      "grad_norm": 0.41229280829429626,
      "learning_rate": 0.00019975376849007642,
      "loss": 1.0211,
      "step": 260
    },
    {
      "epoch": 0.04850399553986248,
      "grad_norm": 0.6339378356933594,
      "learning_rate": 0.0001997517131372763,
      "loss": 1.2711,
      "step": 261
    },
    {
      "epoch": 0.048689834603233596,
      "grad_norm": 0.4885227084159851,
      "learning_rate": 0.00019974964925249656,
      "loss": 1.1049,
      "step": 262
    },
    {
      "epoch": 0.04887567366660472,
      "grad_norm": 0.5992847681045532,
      "learning_rate": 0.00019974757683591373,
      "loss": 1.1126,
      "step": 263
    },
    {
      "epoch": 0.04906151272997584,
      "grad_norm": 0.4222111105918884,
      "learning_rate": 0.0001997454958877051,
      "loss": 1.2963,
      "step": 264
    },
    {
      "epoch": 0.04924735179334696,
      "grad_norm": 0.4678804576396942,
      "learning_rate": 0.0001997434064080486,
      "loss": 1.0687,
      "step": 265
    },
    {
      "epoch": 0.04943319085671808,
      "grad_norm": 0.6195970177650452,
      "learning_rate": 0.000199741308397123,
      "loss": 1.0591,
      "step": 266
    },
    {
      "epoch": 0.0496190299200892,
      "grad_norm": 2.341949224472046,
      "learning_rate": 0.00019973920185510768,
      "loss": 4.5625,
      "step": 267
    },
    {
      "epoch": 0.04980486898346032,
      "grad_norm": 0.42770177125930786,
      "learning_rate": 0.0001997370867821829,
      "loss": 0.8979,
      "step": 268
    },
    {
      "epoch": 0.04999070804683144,
      "grad_norm": 0.4548147916793823,
      "learning_rate": 0.00019973496317852952,
      "loss": 1.1994,
      "step": 269
    },
    {
      "epoch": 0.050176547110202564,
      "grad_norm": 0.521209180355072,
      "learning_rate": 0.00019973283104432918,
      "loss": 1.4071,
      "step": 270
    },
    {
      "epoch": 0.050362386173573685,
      "grad_norm": 0.49684473872184753,
      "learning_rate": 0.00019973069037976428,
      "loss": 1.2518,
      "step": 271
    },
    {
      "epoch": 0.050548225236944806,
      "grad_norm": 0.41964155435562134,
      "learning_rate": 0.00019972854118501782,
      "loss": 1.2989,
      "step": 272
    },
    {
      "epoch": 0.05073406430031593,
      "grad_norm": 0.41438028216362,
      "learning_rate": 0.0001997263834602737,
      "loss": 1.154,
      "step": 273
    },
    {
      "epoch": 0.05091990336368705,
      "grad_norm": 0.42565253376960754,
      "learning_rate": 0.0001997242172057165,
      "loss": 1.0732,
      "step": 274
    },
    {
      "epoch": 0.05110574242705817,
      "grad_norm": 0.43003037571907043,
      "learning_rate": 0.00019972204242153148,
      "loss": 1.1901,
      "step": 275
    },
    {
      "epoch": 0.05129158149042929,
      "grad_norm": 0.38172245025634766,
      "learning_rate": 0.00019971985910790464,
      "loss": 1.0215,
      "step": 276
    },
    {
      "epoch": 0.05147742055380041,
      "grad_norm": 0.5161172151565552,
      "learning_rate": 0.00019971766726502267,
      "loss": 0.788,
      "step": 277
    },
    {
      "epoch": 0.05166325961717153,
      "grad_norm": 0.4897576570510864,
      "learning_rate": 0.00019971546689307316,
      "loss": 1.2448,
      "step": 278
    },
    {
      "epoch": 0.05184909868054265,
      "grad_norm": 0.5034477114677429,
      "learning_rate": 0.00019971325799224423,
      "loss": 1.2172,
      "step": 279
    },
    {
      "epoch": 0.05203493774391377,
      "grad_norm": 0.3733575642108917,
      "learning_rate": 0.00019971104056272485,
      "loss": 1.0159,
      "step": 280
    },
    {
      "epoch": 0.05222077680728489,
      "grad_norm": 0.5659055709838867,
      "learning_rate": 0.00019970881460470467,
      "loss": 1.0378,
      "step": 281
    },
    {
      "epoch": 0.05240661587065601,
      "grad_norm": 0.44712385535240173,
      "learning_rate": 0.00019970658011837404,
      "loss": 1.0341,
      "step": 282
    },
    {
      "epoch": 0.05259245493402713,
      "grad_norm": 0.40552452206611633,
      "learning_rate": 0.00019970433710392416,
      "loss": 1.309,
      "step": 283
    },
    {
      "epoch": 0.05277829399739825,
      "grad_norm": 2.332465410232544,
      "learning_rate": 0.00019970208556154684,
      "loss": 4.0356,
      "step": 284
    },
    {
      "epoch": 0.05296413306076937,
      "grad_norm": 0.5010231137275696,
      "learning_rate": 0.00019969982549143464,
      "loss": 1.1609,
      "step": 285
    },
    {
      "epoch": 0.05314997212414049,
      "grad_norm": 0.4107954502105713,
      "learning_rate": 0.00019969755689378085,
      "loss": 1.1437,
      "step": 286
    },
    {
      "epoch": 0.053335811187511614,
      "grad_norm": 0.40947437286376953,
      "learning_rate": 0.0001996952797687796,
      "loss": 1.319,
      "step": 287
    },
    {
      "epoch": 0.053521650250882735,
      "grad_norm": 0.41476190090179443,
      "learning_rate": 0.0001996929941166256,
      "loss": 1.0567,
      "step": 288
    },
    {
      "epoch": 0.053707489314253856,
      "grad_norm": 0.5421552062034607,
      "learning_rate": 0.00019969069993751431,
      "loss": 1.2611,
      "step": 289
    },
    {
      "epoch": 0.05389332837762498,
      "grad_norm": 0.3736020624637604,
      "learning_rate": 0.000199688397231642,
      "loss": 1.1123,
      "step": 290
    },
    {
      "epoch": 0.0540791674409961,
      "grad_norm": 0.4207514226436615,
      "learning_rate": 0.00019968608599920564,
      "loss": 1.1748,
      "step": 291
    },
    {
      "epoch": 0.05426500650436722,
      "grad_norm": 0.5403153896331787,
      "learning_rate": 0.0001996837662404029,
      "loss": 1.2441,
      "step": 292
    },
    {
      "epoch": 0.05445084556773834,
      "grad_norm": 0.3881579041481018,
      "learning_rate": 0.00019968143795543215,
      "loss": 1.1158,
      "step": 293
    },
    {
      "epoch": 0.05463668463110946,
      "grad_norm": 0.47338834404945374,
      "learning_rate": 0.0001996791011444926,
      "loss": 1.2112,
      "step": 294
    },
    {
      "epoch": 0.05482252369448058,
      "grad_norm": 0.46377336978912354,
      "learning_rate": 0.00019967675580778412,
      "loss": 1.0821,
      "step": 295
    },
    {
      "epoch": 0.0550083627578517,
      "grad_norm": 0.37327149510383606,
      "learning_rate": 0.00019967440194550723,
      "loss": 1.138,
      "step": 296
    },
    {
      "epoch": 0.055194201821222824,
      "grad_norm": 0.45194652676582336,
      "learning_rate": 0.00019967203955786335,
      "loss": 1.0792,
      "step": 297
    },
    {
      "epoch": 0.05538004088459394,
      "grad_norm": 0.442440927028656,
      "learning_rate": 0.0001996696686450545,
      "loss": 1.1935,
      "step": 298
    },
    {
      "epoch": 0.05556587994796506,
      "grad_norm": 0.5025745630264282,
      "learning_rate": 0.0001996672892072835,
      "loss": 1.1351,
      "step": 299
    },
    {
      "epoch": 0.05575171901133618,
      "grad_norm": 0.47533366084098816,
      "learning_rate": 0.0001996649012447538,
      "loss": 1.3472,
      "step": 300
    },
    {
      "epoch": 0.0559375580747073,
      "grad_norm": 0.45607975125312805,
      "learning_rate": 0.00019966250475766972,
      "loss": 1.1532,
      "step": 301
    },
    {
      "epoch": 0.05612339713807842,
      "grad_norm": 0.4451979994773865,
      "learning_rate": 0.0001996600997462362,
      "loss": 0.8268,
      "step": 302
    },
    {
      "epoch": 0.05630923620144954,
      "grad_norm": 0.4001930356025696,
      "learning_rate": 0.00019965768621065896,
      "loss": 0.9162,
      "step": 303
    },
    {
      "epoch": 0.056495075264820664,
      "grad_norm": 0.415159672498703,
      "learning_rate": 0.00019965526415114445,
      "loss": 1.1613,
      "step": 304
    },
    {
      "epoch": 0.056680914328191785,
      "grad_norm": 0.4185488224029541,
      "learning_rate": 0.0001996528335678998,
      "loss": 1.1483,
      "step": 305
    },
    {
      "epoch": 0.056866753391562906,
      "grad_norm": 1.838326334953308,
      "learning_rate": 0.0001996503944611329,
      "loss": 4.0785,
      "step": 306
    },
    {
      "epoch": 0.05705259245493403,
      "grad_norm": 0.42079976201057434,
      "learning_rate": 0.0001996479468310524,
      "loss": 0.9382,
      "step": 307
    },
    {
      "epoch": 0.05723843151830515,
      "grad_norm": 0.3876133859157562,
      "learning_rate": 0.00019964549067786762,
      "loss": 0.9098,
      "step": 308
    },
    {
      "epoch": 0.05742427058167627,
      "grad_norm": 0.4491420388221741,
      "learning_rate": 0.0001996430260017887,
      "loss": 1.104,
      "step": 309
    },
    {
      "epoch": 0.05761010964504739,
      "grad_norm": 0.3816368281841278,
      "learning_rate": 0.00019964055280302637,
      "loss": 0.9202,
      "step": 310
    },
    {
      "epoch": 0.05779594870841851,
      "grad_norm": 0.3864871859550476,
      "learning_rate": 0.00019963807108179223,
      "loss": 0.8936,
      "step": 311
    },
    {
      "epoch": 0.05798178777178963,
      "grad_norm": 0.4503324329853058,
      "learning_rate": 0.00019963558083829853,
      "loss": 1.051,
      "step": 312
    },
    {
      "epoch": 0.05816762683516075,
      "grad_norm": 0.4208626449108124,
      "learning_rate": 0.00019963308207275823,
      "loss": 1.2242,
      "step": 313
    },
    {
      "epoch": 0.058353465898531874,
      "grad_norm": 0.5481116771697998,
      "learning_rate": 0.00019963057478538513,
      "loss": 1.5282,
      "step": 314
    },
    {
      "epoch": 0.058539304961902995,
      "grad_norm": 0.46902161836624146,
      "learning_rate": 0.0001996280589763936,
      "loss": 1.2003,
      "step": 315
    },
    {
      "epoch": 0.05872514402527411,
      "grad_norm": 0.37822434306144714,
      "learning_rate": 0.00019962553464599887,
      "loss": 0.7875,
      "step": 316
    },
    {
      "epoch": 0.05891098308864523,
      "grad_norm": 0.4753560423851013,
      "learning_rate": 0.00019962300179441687,
      "loss": 1.2323,
      "step": 317
    },
    {
      "epoch": 0.05909682215201635,
      "grad_norm": 0.5222552418708801,
      "learning_rate": 0.00019962046042186417,
      "loss": 1.3808,
      "step": 318
    },
    {
      "epoch": 0.05928266121538747,
      "grad_norm": 0.37479904294013977,
      "learning_rate": 0.0001996179105285582,
      "loss": 1.1761,
      "step": 319
    },
    {
      "epoch": 0.05946850027875859,
      "grad_norm": 0.5287699103355408,
      "learning_rate": 0.00019961535211471704,
      "loss": 0.9187,
      "step": 320
    },
    {
      "epoch": 0.059654339342129714,
      "grad_norm": 0.5424025058746338,
      "learning_rate": 0.0001996127851805595,
      "loss": 1.3986,
      "step": 321
    },
    {
      "epoch": 0.059840178405500835,
      "grad_norm": 0.3869524896144867,
      "learning_rate": 0.0001996102097263052,
      "loss": 0.9046,
      "step": 322
    },
    {
      "epoch": 0.060026017468871956,
      "grad_norm": 0.42366644740104675,
      "learning_rate": 0.00019960762575217434,
      "loss": 1.064,
      "step": 323
    },
    {
      "epoch": 0.06021185653224308,
      "grad_norm": 0.4955057203769684,
      "learning_rate": 0.00019960503325838798,
      "loss": 1.0469,
      "step": 324
    },
    {
      "epoch": 0.0603976955956142,
      "grad_norm": 0.49584606289863586,
      "learning_rate": 0.00019960243224516783,
      "loss": 1.1859,
      "step": 325
    },
    {
      "epoch": 0.06058353465898532,
      "grad_norm": 0.3556949496269226,
      "learning_rate": 0.0001995998227127364,
      "loss": 1.1647,
      "step": 326
    },
    {
      "epoch": 0.06076937372235644,
      "grad_norm": 0.41173893213272095,
      "learning_rate": 0.00019959720466131687,
      "loss": 1.145,
      "step": 327
    },
    {
      "epoch": 0.06095521278572756,
      "grad_norm": 0.44232895970344543,
      "learning_rate": 0.00019959457809113316,
      "loss": 0.9857,
      "step": 328
    },
    {
      "epoch": 0.06114105184909868,
      "grad_norm": 0.43240708112716675,
      "learning_rate": 0.00019959194300240993,
      "loss": 1.1169,
      "step": 329
    },
    {
      "epoch": 0.0613268909124698,
      "grad_norm": 0.5672433972358704,
      "learning_rate": 0.00019958929939537258,
      "loss": 1.1568,
      "step": 330
    },
    {
      "epoch": 0.061512729975840924,
      "grad_norm": 0.30468809604644775,
      "learning_rate": 0.00019958664727024722,
      "loss": 0.8156,
      "step": 331
    },
    {
      "epoch": 0.061698569039212045,
      "grad_norm": 2.1194663047790527,
      "learning_rate": 0.0001995839866272607,
      "loss": 3.7851,
      "step": 332
    },
    {
      "epoch": 0.061884408102583166,
      "grad_norm": 0.4063419699668884,
      "learning_rate": 0.0001995813174666405,
      "loss": 1.0279,
      "step": 333
    },
    {
      "epoch": 0.06207024716595429,
      "grad_norm": 0.4712132513523102,
      "learning_rate": 0.00019957863978861505,
      "loss": 1.1407,
      "step": 334
    },
    {
      "epoch": 0.0622560862293254,
      "grad_norm": 0.44568800926208496,
      "learning_rate": 0.0001995759535934133,
      "loss": 1.1316,
      "step": 335
    },
    {
      "epoch": 0.06244192529269652,
      "grad_norm": 0.4557202160358429,
      "learning_rate": 0.00019957325888126507,
      "loss": 1.0661,
      "step": 336
    },
    {
      "epoch": 0.06262776435606765,
      "grad_norm": 0.38041767477989197,
      "learning_rate": 0.00019957055565240076,
      "loss": 0.9131,
      "step": 337
    },
    {
      "epoch": 0.06281360341943877,
      "grad_norm": 0.4428039491176605,
      "learning_rate": 0.00019956784390705163,
      "loss": 1.2109,
      "step": 338
    },
    {
      "epoch": 0.06299944248280989,
      "grad_norm": 0.4144112467765808,
      "learning_rate": 0.0001995651236454496,
      "loss": 1.0889,
      "step": 339
    },
    {
      "epoch": 0.06318528154618101,
      "grad_norm": 0.3976646363735199,
      "learning_rate": 0.0001995623948678274,
      "loss": 1.0306,
      "step": 340
    },
    {
      "epoch": 0.06337112060955213,
      "grad_norm": 0.47308364510536194,
      "learning_rate": 0.00019955965757441834,
      "loss": 1.2456,
      "step": 341
    },
    {
      "epoch": 0.06355695967292325,
      "grad_norm": 0.3834746778011322,
      "learning_rate": 0.00019955691176545663,
      "loss": 1.0008,
      "step": 342
    },
    {
      "epoch": 0.06374279873629438,
      "grad_norm": 0.46271997690200806,
      "learning_rate": 0.00019955415744117707,
      "loss": 1.1123,
      "step": 343
    },
    {
      "epoch": 0.06392863779966548,
      "grad_norm": 0.4702741503715515,
      "learning_rate": 0.00019955139460181527,
      "loss": 1.2039,
      "step": 344
    },
    {
      "epoch": 0.0641144768630366,
      "grad_norm": 0.4005366861820221,
      "learning_rate": 0.00019954862324760752,
      "loss": 0.955,
      "step": 345
    },
    {
      "epoch": 0.06430031592640773,
      "grad_norm": 0.44210031628608704,
      "learning_rate": 0.00019954584337879085,
      "loss": 1.2393,
      "step": 346
    },
    {
      "epoch": 0.06448615498977885,
      "grad_norm": 0.4394225478172302,
      "learning_rate": 0.00019954305499560307,
      "loss": 1.2445,
      "step": 347
    },
    {
      "epoch": 0.06467199405314997,
      "grad_norm": 0.6319501399993896,
      "learning_rate": 0.00019954025809828266,
      "loss": 1.2729,
      "step": 348
    },
    {
      "epoch": 0.06485783311652109,
      "grad_norm": 0.49698418378829956,
      "learning_rate": 0.00019953745268706883,
      "loss": 1.3173,
      "step": 349
    },
    {
      "epoch": 0.06504367217989221,
      "grad_norm": 2.311923027038574,
      "learning_rate": 0.00019953463876220154,
      "loss": 3.9918,
      "step": 350
    },
    {
      "epoch": 0.06522951124326333,
      "grad_norm": 0.4096338450908661,
      "learning_rate": 0.0001995318163239215,
      "loss": 1.0602,
      "step": 351
    },
    {
      "epoch": 0.06541535030663445,
      "grad_norm": 0.5735236406326294,
      "learning_rate": 0.00019952898537247004,
      "loss": 1.1047,
      "step": 352
    },
    {
      "epoch": 0.06560118937000557,
      "grad_norm": 0.43088850378990173,
      "learning_rate": 0.0001995261459080894,
      "loss": 1.1304,
      "step": 353
    },
    {
      "epoch": 0.06578702843337669,
      "grad_norm": 0.39601054787635803,
      "learning_rate": 0.0001995232979310224,
      "loss": 1.0685,
      "step": 354
    },
    {
      "epoch": 0.06597286749674781,
      "grad_norm": 0.5507276654243469,
      "learning_rate": 0.0001995204414415126,
      "loss": 1.1807,
      "step": 355
    },
    {
      "epoch": 0.06615870656011894,
      "grad_norm": 0.4136437773704529,
      "learning_rate": 0.00019951757643980433,
      "loss": 1.038,
      "step": 356
    },
    {
      "epoch": 0.06634454562349006,
      "grad_norm": 0.4400745928287506,
      "learning_rate": 0.00019951470292614272,
      "loss": 1.1044,
      "step": 357
    },
    {
      "epoch": 0.06653038468686118,
      "grad_norm": 0.3629709482192993,
      "learning_rate": 0.00019951182090077344,
      "loss": 1.0114,
      "step": 358
    },
    {
      "epoch": 0.0667162237502323,
      "grad_norm": 0.6199255585670471,
      "learning_rate": 0.00019950893036394307,
      "loss": 1.3218,
      "step": 359
    },
    {
      "epoch": 0.06690206281360342,
      "grad_norm": 0.4702179729938507,
      "learning_rate": 0.00019950603131589883,
      "loss": 1.1172,
      "step": 360
    },
    {
      "epoch": 0.06708790187697454,
      "grad_norm": 0.42372721433639526,
      "learning_rate": 0.00019950312375688865,
      "loss": 1.1621,
      "step": 361
    },
    {
      "epoch": 0.06727374094034566,
      "grad_norm": 0.4708528518676758,
      "learning_rate": 0.00019950020768716128,
      "loss": 1.222,
      "step": 362
    },
    {
      "epoch": 0.06745958000371678,
      "grad_norm": 0.4498105049133301,
      "learning_rate": 0.00019949728310696604,
      "loss": 1.1895,
      "step": 363
    },
    {
      "epoch": 0.0676454190670879,
      "grad_norm": 0.4580400586128235,
      "learning_rate": 0.00019949435001655313,
      "loss": 1.1215,
      "step": 364
    },
    {
      "epoch": 0.06783125813045902,
      "grad_norm": 0.4212903678417206,
      "learning_rate": 0.00019949140841617348,
      "loss": 1.3536,
      "step": 365
    },
    {
      "epoch": 0.06801709719383014,
      "grad_norm": 0.4488227069377899,
      "learning_rate": 0.00019948845830607863,
      "loss": 1.1161,
      "step": 366
    },
    {
      "epoch": 0.06820293625720127,
      "grad_norm": 0.46963340044021606,
      "learning_rate": 0.00019948549968652087,
      "loss": 0.859,
      "step": 367
    },
    {
      "epoch": 0.06838877532057239,
      "grad_norm": 0.489118367433548,
      "learning_rate": 0.00019948253255775336,
      "loss": 1.221,
      "step": 368
    },
    {
      "epoch": 0.06857461438394351,
      "grad_norm": 0.4626411199569702,
      "learning_rate": 0.0001994795569200298,
      "loss": 1.0484,
      "step": 369
    },
    {
      "epoch": 0.06876045344731463,
      "grad_norm": 0.47765642404556274,
      "learning_rate": 0.00019947657277360478,
      "loss": 1.1136,
      "step": 370
    },
    {
      "epoch": 0.06894629251068575,
      "grad_norm": 0.45885488390922546,
      "learning_rate": 0.00019947358011873344,
      "loss": 1.2611,
      "step": 371
    },
    {
      "epoch": 0.06913213157405687,
      "grad_norm": 0.6871397495269775,
      "learning_rate": 0.0001994705789556718,
      "loss": 1.3397,
      "step": 372
    },
    {
      "epoch": 0.06931797063742799,
      "grad_norm": 0.3958967626094818,
      "learning_rate": 0.00019946756928467658,
      "loss": 1.0389,
      "step": 373
    },
    {
      "epoch": 0.06950380970079911,
      "grad_norm": 0.47687655687332153,
      "learning_rate": 0.0001994645511060052,
      "loss": 1.1862,
      "step": 374
    },
    {
      "epoch": 0.06968964876417023,
      "grad_norm": 0.45492956042289734,
      "learning_rate": 0.00019946152441991576,
      "loss": 1.1136,
      "step": 375
    },
    {
      "epoch": 0.06987548782754135,
      "grad_norm": 0.5021520256996155,
      "learning_rate": 0.00019945848922666717,
      "loss": 1.1335,
      "step": 376
    },
    {
      "epoch": 0.07006132689091248,
      "grad_norm": 0.5270656943321228,
      "learning_rate": 0.00019945544552651905,
      "loss": 0.829,
      "step": 377
    },
    {
      "epoch": 0.0702471659542836,
      "grad_norm": 0.39411312341690063,
      "learning_rate": 0.00019945239331973174,
      "loss": 1.0567,
      "step": 378
    },
    {
      "epoch": 0.07043300501765472,
      "grad_norm": 0.4584178328514099,
      "learning_rate": 0.00019944933260656624,
      "loss": 1.158,
      "step": 379
    },
    {
      "epoch": 0.07061884408102582,
      "grad_norm": 0.4279657304286957,
      "learning_rate": 0.00019944626338728444,
      "loss": 1.0037,
      "step": 380
    },
    {
      "epoch": 0.07080468314439695,
      "grad_norm": 0.554434061050415,
      "learning_rate": 0.00019944318566214876,
      "loss": 1.2635,
      "step": 381
    },
    {
      "epoch": 0.07099052220776807,
      "grad_norm": 0.4390695095062256,
      "learning_rate": 0.0001994400994314225,
      "loss": 0.8383,
      "step": 382
    },
    {
      "epoch": 0.07117636127113919,
      "grad_norm": 0.49963220953941345,
      "learning_rate": 0.00019943700469536962,
      "loss": 1.2483,
      "step": 383
    },
    {
      "epoch": 0.07136220033451031,
      "grad_norm": 0.5585063099861145,
      "learning_rate": 0.0001994339014542548,
      "loss": 1.3737,
      "step": 384
    },
    {
      "epoch": 0.07154803939788143,
      "grad_norm": 0.4825543165206909,
      "learning_rate": 0.00019943078970834348,
      "loss": 0.9761,
      "step": 385
    },
    {
      "epoch": 0.07173387846125255,
      "grad_norm": 0.3667401969432831,
      "learning_rate": 0.00019942766945790186,
      "loss": 0.9599,
      "step": 386
    },
    {
      "epoch": 0.07191971752462367,
      "grad_norm": 0.5182320475578308,
      "learning_rate": 0.00019942454070319673,
      "loss": 1.1975,
      "step": 387
    },
    {
      "epoch": 0.07210555658799479,
      "grad_norm": 0.4698030352592468,
      "learning_rate": 0.00019942140344449577,
      "loss": 1.2344,
      "step": 388
    },
    {
      "epoch": 0.07229139565136591,
      "grad_norm": 0.42861101031303406,
      "learning_rate": 0.0001994182576820673,
      "loss": 1.2487,
      "step": 389
    },
    {
      "epoch": 0.07247723471473703,
      "grad_norm": 0.5619314908981323,
      "learning_rate": 0.00019941510341618036,
      "loss": 1.3243,
      "step": 390
    },
    {
      "epoch": 0.07266307377810816,
      "grad_norm": 0.4856956899166107,
      "learning_rate": 0.0001994119406471048,
      "loss": 1.1531,
      "step": 391
    },
    {
      "epoch": 0.07284891284147928,
      "grad_norm": 0.3697066903114319,
      "learning_rate": 0.00019940876937511108,
      "loss": 1.1304,
      "step": 392
    },
    {
      "epoch": 0.0730347519048504,
      "grad_norm": 0.417986124753952,
      "learning_rate": 0.00019940558960047046,
      "loss": 1.0578,
      "step": 393
    },
    {
      "epoch": 0.07322059096822152,
      "grad_norm": 0.4160363972187042,
      "learning_rate": 0.0001994024013234549,
      "loss": 1.3094,
      "step": 394
    },
    {
      "epoch": 0.07340643003159264,
      "grad_norm": 0.4469776153564453,
      "learning_rate": 0.00019939920454433715,
      "loss": 1.0983,
      "step": 395
    },
    {
      "epoch": 0.07359226909496376,
      "grad_norm": 0.5031948685646057,
      "learning_rate": 0.0001993959992633906,
      "loss": 1.536,
      "step": 396
    },
    {
      "epoch": 0.07377810815833488,
      "grad_norm": 0.45184874534606934,
      "learning_rate": 0.00019939278548088937,
      "loss": 0.8456,
      "step": 397
    },
    {
      "epoch": 0.073963947221706,
      "grad_norm": 0.45181095600128174,
      "learning_rate": 0.00019938956319710843,
      "loss": 1.2099,
      "step": 398
    },
    {
      "epoch": 0.07414978628507712,
      "grad_norm": 0.49429285526275635,
      "learning_rate": 0.00019938633241232335,
      "loss": 1.1858,
      "step": 399
    },
    {
      "epoch": 0.07433562534844824,
      "grad_norm": 0.41390714049339294,
      "learning_rate": 0.0001993830931268104,
      "loss": 0.9845,
      "step": 400
    },
    {
      "epoch": 0.07452146441181937,
      "grad_norm": 0.42566490173339844,
      "learning_rate": 0.00019937984534084674,
      "loss": 1.0536,
      "step": 401
    },
    {
      "epoch": 0.07470730347519049,
      "grad_norm": 0.4550681412220001,
      "learning_rate": 0.00019937658905471012,
      "loss": 1.1484,
      "step": 402
    },
    {
      "epoch": 0.07489314253856161,
      "grad_norm": 0.4505470097064972,
      "learning_rate": 0.00019937332426867905,
      "loss": 1.3151,
      "step": 403
    },
    {
      "epoch": 0.07507898160193273,
      "grad_norm": 0.3663678765296936,
      "learning_rate": 0.00019937005098303277,
      "loss": 1.1038,
      "step": 404
    },
    {
      "epoch": 0.07526482066530385,
      "grad_norm": 0.4451432228088379,
      "learning_rate": 0.00019936676919805126,
      "loss": 1.0466,
      "step": 405
    },
    {
      "epoch": 0.07545065972867497,
      "grad_norm": 0.4404486417770386,
      "learning_rate": 0.00019936347891401522,
      "loss": 1.2363,
      "step": 406
    },
    {
      "epoch": 0.07563649879204609,
      "grad_norm": 0.42253008484840393,
      "learning_rate": 0.0001993601801312061,
      "loss": 0.8975,
      "step": 407
    },
    {
      "epoch": 0.07582233785541721,
      "grad_norm": 0.4789395034313202,
      "learning_rate": 0.000199356872849906,
      "loss": 1.0806,
      "step": 408
    },
    {
      "epoch": 0.07600817691878833,
      "grad_norm": 0.4530208706855774,
      "learning_rate": 0.00019935355707039785,
      "loss": 1.0968,
      "step": 409
    },
    {
      "epoch": 0.07619401598215945,
      "grad_norm": 0.47005361318588257,
      "learning_rate": 0.00019935023279296524,
      "loss": 1.2328,
      "step": 410
    },
    {
      "epoch": 0.07637985504553058,
      "grad_norm": 0.4853919446468353,
      "learning_rate": 0.00019934690001789247,
      "loss": 1.148,
      "step": 411
    },
    {
      "epoch": 0.0765656941089017,
      "grad_norm": 0.5219566226005554,
      "learning_rate": 0.00019934355874546466,
      "loss": 1.1759,
      "step": 412
    },
    {
      "epoch": 0.07675153317227282,
      "grad_norm": 0.3366680145263672,
      "learning_rate": 0.0001993402089759675,
      "loss": 0.8707,
      "step": 413
    },
    {
      "epoch": 0.07693737223564394,
      "grad_norm": 0.6554648876190186,
      "learning_rate": 0.0001993368507096876,
      "loss": 1.3711,
      "step": 414
    },
    {
      "epoch": 0.07712321129901506,
      "grad_norm": 0.44505739212036133,
      "learning_rate": 0.0001993334839469122,
      "loss": 1.0383,
      "step": 415
    },
    {
      "epoch": 0.07730905036238617,
      "grad_norm": 0.44871413707733154,
      "learning_rate": 0.00019933010868792916,
      "loss": 1.1774,
      "step": 416
    },
    {
      "epoch": 0.07749488942575729,
      "grad_norm": 0.39043372869491577,
      "learning_rate": 0.00019932672493302729,
      "loss": 1.1545,
      "step": 417
    },
    {
      "epoch": 0.07768072848912841,
      "grad_norm": 0.5223948955535889,
      "learning_rate": 0.00019932333268249595,
      "loss": 1.3809,
      "step": 418
    },
    {
      "epoch": 0.07786656755249953,
      "grad_norm": 0.4341309368610382,
      "learning_rate": 0.00019931993193662528,
      "loss": 0.9368,
      "step": 419
    },
    {
      "epoch": 0.07805240661587065,
      "grad_norm": 0.5227263569831848,
      "learning_rate": 0.0001993165226957062,
      "loss": 1.1342,
      "step": 420
    },
    {
      "epoch": 0.07823824567924177,
      "grad_norm": 0.40590155124664307,
      "learning_rate": 0.00019931310496003029,
      "loss": 1.3317,
      "step": 421
    },
    {
      "epoch": 0.07842408474261289,
      "grad_norm": 0.4488816559314728,
      "learning_rate": 0.00019930967872988984,
      "loss": 0.9919,
      "step": 422
    },
    {
      "epoch": 0.07860992380598401,
      "grad_norm": 0.3798006772994995,
      "learning_rate": 0.00019930624400557796,
      "loss": 0.7698,
      "step": 423
    },
    {
      "epoch": 0.07879576286935513,
      "grad_norm": 0.41681668162345886,
      "learning_rate": 0.00019930280078738836,
      "loss": 1.096,
      "step": 424
    },
    {
      "epoch": 0.07898160193272626,
      "grad_norm": 0.40335264801979065,
      "learning_rate": 0.00019929934907561564,
      "loss": 1.113,
      "step": 425
    },
    {
      "epoch": 0.07916744099609738,
      "grad_norm": 0.4952360689640045,
      "learning_rate": 0.00019929588887055495,
      "loss": 1.2477,
      "step": 426
    },
    {
      "epoch": 0.0793532800594685,
      "grad_norm": 0.44737347960472107,
      "learning_rate": 0.00019929242017250226,
      "loss": 1.0625,
      "step": 427
    },
    {
      "epoch": 0.07953911912283962,
      "grad_norm": 0.4532301723957062,
      "learning_rate": 0.00019928894298175428,
      "loss": 1.1246,
      "step": 428
    },
    {
      "epoch": 0.07972495818621074,
      "grad_norm": 0.4013027250766754,
      "learning_rate": 0.00019928545729860846,
      "loss": 1.0591,
      "step": 429
    },
    {
      "epoch": 0.07991079724958186,
      "grad_norm": 0.4169690012931824,
      "learning_rate": 0.00019928196312336285,
      "loss": 1.2644,
      "step": 430
    },
    {
      "epoch": 0.08009663631295298,
      "grad_norm": 0.549383282661438,
      "learning_rate": 0.00019927846045631638,
      "loss": 1.093,
      "step": 431
    },
    {
      "epoch": 0.0802824753763241,
      "grad_norm": 1.4828532934188843,
      "learning_rate": 0.0001992749492977686,
      "loss": 3.8312,
      "step": 432
    },
    {
      "epoch": 0.08046831443969522,
      "grad_norm": 0.414670467376709,
      "learning_rate": 0.00019927142964801985,
      "loss": 1.0559,
      "step": 433
    },
    {
      "epoch": 0.08065415350306634,
      "grad_norm": 0.46947577595710754,
      "learning_rate": 0.00019926790150737118,
      "loss": 0.6598,
      "step": 434
    },
    {
      "epoch": 0.08083999256643747,
      "grad_norm": 0.37407925724983215,
      "learning_rate": 0.00019926436487612435,
      "loss": 1.0784,
      "step": 435
    },
    {
      "epoch": 0.08102583162980859,
      "grad_norm": 0.4735274910926819,
      "learning_rate": 0.00019926081975458183,
      "loss": 1.1369,
      "step": 436
    },
    {
      "epoch": 0.08121167069317971,
      "grad_norm": 0.3918091952800751,
      "learning_rate": 0.00019925726614304688,
      "loss": 1.0932,
      "step": 437
    },
    {
      "epoch": 0.08139750975655083,
      "grad_norm": 0.4890015125274658,
      "learning_rate": 0.00019925370404182341,
      "loss": 1.4067,
      "step": 438
    },
    {
      "epoch": 0.08158334881992195,
      "grad_norm": 0.5494944453239441,
      "learning_rate": 0.00019925013345121616,
      "loss": 1.0676,
      "step": 439
    },
    {
      "epoch": 0.08176918788329307,
      "grad_norm": 0.4678218364715576,
      "learning_rate": 0.00019924655437153046,
      "loss": 0.9595,
      "step": 440
    },
    {
      "epoch": 0.08195502694666419,
      "grad_norm": 0.42315244674682617,
      "learning_rate": 0.0001992429668030725,
      "loss": 1.1847,
      "step": 441
    },
    {
      "epoch": 0.08214086601003531,
      "grad_norm": 0.4802471399307251,
      "learning_rate": 0.00019923937074614907,
      "loss": 0.9514,
      "step": 442
    },
    {
      "epoch": 0.08232670507340643,
      "grad_norm": 0.38350144028663635,
      "learning_rate": 0.00019923576620106776,
      "loss": 0.9091,
      "step": 443
    },
    {
      "epoch": 0.08251254413677755,
      "grad_norm": 0.42246341705322266,
      "learning_rate": 0.0001992321531681369,
      "loss": 1.2286,
      "step": 444
    },
    {
      "epoch": 0.08269838320014868,
      "grad_norm": 0.48033586144447327,
      "learning_rate": 0.0001992285316476655,
      "loss": 1.2414,
      "step": 445
    },
    {
      "epoch": 0.0828842222635198,
      "grad_norm": 0.42830902338027954,
      "learning_rate": 0.00019922490163996336,
      "loss": 0.8,
      "step": 446
    },
    {
      "epoch": 0.08307006132689092,
      "grad_norm": 0.4135034382343292,
      "learning_rate": 0.0001992212631453409,
      "loss": 1.0903,
      "step": 447
    },
    {
      "epoch": 0.08325590039026204,
      "grad_norm": 0.4331395924091339,
      "learning_rate": 0.00019921761616410939,
      "loss": 1.1565,
      "step": 448
    },
    {
      "epoch": 0.08344173945363316,
      "grad_norm": 0.530741274356842,
      "learning_rate": 0.00019921396069658072,
      "loss": 1.2175,
      "step": 449
    },
    {
      "epoch": 0.08362757851700428,
      "grad_norm": 0.5240445137023926,
      "learning_rate": 0.00019921029674306755,
      "loss": 1.0252,
      "step": 450
    },
    {
      "epoch": 0.0838134175803754,
      "grad_norm": 0.47946205735206604,
      "learning_rate": 0.00019920662430388328,
      "loss": 0.9948,
      "step": 451
    },
    {
      "epoch": 0.08399925664374651,
      "grad_norm": 0.4071696698665619,
      "learning_rate": 0.00019920294337934203,
      "loss": 1.0508,
      "step": 452
    },
    {
      "epoch": 0.08418509570711763,
      "grad_norm": 0.528746485710144,
      "learning_rate": 0.0001991992539697586,
      "loss": 1.0941,
      "step": 453
    },
    {
      "epoch": 0.08437093477048875,
      "grad_norm": 0.44651922583580017,
      "learning_rate": 0.0001991955560754486,
      "loss": 1.0259,
      "step": 454
    },
    {
      "epoch": 0.08455677383385987,
      "grad_norm": 0.5037238001823425,
      "learning_rate": 0.0001991918496967283,
      "loss": 1.2394,
      "step": 455
    },
    {
      "epoch": 0.08474261289723099,
      "grad_norm": 0.4429932236671448,
      "learning_rate": 0.00019918813483391472,
      "loss": 1.0982,
      "step": 456
    },
    {
      "epoch": 0.08492845196060211,
      "grad_norm": 0.5680214166641235,
      "learning_rate": 0.0001991844114873256,
      "loss": 1.1919,
      "step": 457
    },
    {
      "epoch": 0.08511429102397323,
      "grad_norm": 0.43642061948776245,
      "learning_rate": 0.00019918067965727938,
      "loss": 1.1807,
      "step": 458
    },
    {
      "epoch": 0.08530013008734436,
      "grad_norm": 0.4607391655445099,
      "learning_rate": 0.00019917693934409528,
      "loss": 1.1641,
      "step": 459
    },
    {
      "epoch": 0.08548596915071548,
      "grad_norm": 0.48521819710731506,
      "learning_rate": 0.00019917319054809317,
      "loss": 0.9252,
      "step": 460
    },
    {
      "epoch": 0.0856718082140866,
      "grad_norm": 0.5308165550231934,
      "learning_rate": 0.00019916943326959378,
      "loss": 1.3597,
      "step": 461
    },
    {
      "epoch": 0.08585764727745772,
      "grad_norm": 0.5356461405754089,
      "learning_rate": 0.00019916566750891843,
      "loss": 1.2329,
      "step": 462
    },
    {
      "epoch": 0.08604348634082884,
      "grad_norm": 0.45560821890830994,
      "learning_rate": 0.0001991618932663892,
      "loss": 1.027,
      "step": 463
    },
    {
      "epoch": 0.08622932540419996,
      "grad_norm": 0.44379356503486633,
      "learning_rate": 0.0001991581105423289,
      "loss": 1.2826,
      "step": 464
    },
    {
      "epoch": 0.08641516446757108,
      "grad_norm": 0.5654692649841309,
      "learning_rate": 0.00019915431933706113,
      "loss": 1.0153,
      "step": 465
    },
    {
      "epoch": 0.0866010035309422,
      "grad_norm": 0.4437659978866577,
      "learning_rate": 0.00019915051965091013,
      "loss": 1.0302,
      "step": 466
    },
    {
      "epoch": 0.08678684259431332,
      "grad_norm": 0.41596749424934387,
      "learning_rate": 0.00019914671148420085,
      "loss": 1.2477,
      "step": 467
    },
    {
      "epoch": 0.08697268165768444,
      "grad_norm": 0.40698888897895813,
      "learning_rate": 0.00019914289483725909,
      "loss": 1.1617,
      "step": 468
    },
    {
      "epoch": 0.08715852072105557,
      "grad_norm": 0.5427649021148682,
      "learning_rate": 0.00019913906971041122,
      "loss": 0.9785,
      "step": 469
    },
    {
      "epoch": 0.08734435978442669,
      "grad_norm": 0.4189436137676239,
      "learning_rate": 0.0001991352361039845,
      "loss": 1.025,
      "step": 470
    },
    {
      "epoch": 0.08753019884779781,
      "grad_norm": 0.5531293153762817,
      "learning_rate": 0.00019913139401830674,
      "loss": 1.4385,
      "step": 471
    },
    {
      "epoch": 0.08771603791116893,
      "grad_norm": 0.46829280257225037,
      "learning_rate": 0.00019912754345370663,
      "loss": 1.1963,
      "step": 472
    },
    {
      "epoch": 0.08790187697454005,
      "grad_norm": 0.5418177247047424,
      "learning_rate": 0.00019912368441051345,
      "loss": 1.412,
      "step": 473
    },
    {
      "epoch": 0.08808771603791117,
      "grad_norm": 0.4771590828895569,
      "learning_rate": 0.00019911981688905734,
      "loss": 1.1606,
      "step": 474
    },
    {
      "epoch": 0.08827355510128229,
      "grad_norm": 0.4091399013996124,
      "learning_rate": 0.00019911594088966905,
      "loss": 1.1281,
      "step": 475
    },
    {
      "epoch": 0.08845939416465341,
      "grad_norm": 0.47988924384117126,
      "learning_rate": 0.00019911205641268012,
      "loss": 0.8196,
      "step": 476
    },
    {
      "epoch": 0.08864523322802453,
      "grad_norm": 0.5251743793487549,
      "learning_rate": 0.00019910816345842277,
      "loss": 0.9623,
      "step": 477
    },
    {
      "epoch": 0.08883107229139565,
      "grad_norm": 0.44282639026641846,
      "learning_rate": 0.00019910426202723004,
      "loss": 1.1464,
      "step": 478
    },
    {
      "epoch": 0.08901691135476678,
      "grad_norm": 0.6290245056152344,
      "learning_rate": 0.00019910035211943557,
      "loss": 1.3983,
      "step": 479
    },
    {
      "epoch": 0.0892027504181379,
      "grad_norm": 0.4757908880710602,
      "learning_rate": 0.0001990964337353738,
      "loss": 1.2815,
      "step": 480
    },
    {
      "epoch": 0.08938858948150902,
      "grad_norm": 0.43329253792762756,
      "learning_rate": 0.00019909250687537986,
      "loss": 1.28,
      "step": 481
    },
    {
      "epoch": 0.08957442854488014,
      "grad_norm": 0.4964252710342407,
      "learning_rate": 0.00019908857153978966,
      "loss": 1.2799,
      "step": 482
    },
    {
      "epoch": 0.08976026760825126,
      "grad_norm": 0.4180728793144226,
      "learning_rate": 0.00019908462772893977,
      "loss": 1.1443,
      "step": 483
    },
    {
      "epoch": 0.08994610667162238,
      "grad_norm": 0.45169374346733093,
      "learning_rate": 0.00019908067544316751,
      "loss": 0.8931,
      "step": 484
    },
    {
      "epoch": 0.0901319457349935,
      "grad_norm": 0.5057590007781982,
      "learning_rate": 0.00019907671468281097,
      "loss": 1.218,
      "step": 485
    },
    {
      "epoch": 0.09031778479836462,
      "grad_norm": 0.40992191433906555,
      "learning_rate": 0.00019907274544820887,
      "loss": 1.162,
      "step": 486
    },
    {
      "epoch": 0.09050362386173574,
      "grad_norm": 0.5198535323143005,
      "learning_rate": 0.00019906876773970072,
      "loss": 1.0092,
      "step": 487
    },
    {
      "epoch": 0.09068946292510685,
      "grad_norm": 0.44954633712768555,
      "learning_rate": 0.00019906478155762677,
      "loss": 1.2348,
      "step": 488
    },
    {
      "epoch": 0.09087530198847797,
      "grad_norm": 0.4218129813671112,
      "learning_rate": 0.00019906078690232792,
      "loss": 1.057,
      "step": 489
    },
    {
      "epoch": 0.09106114105184909,
      "grad_norm": 0.5415510535240173,
      "learning_rate": 0.0001990567837741459,
      "loss": 0.9229,
      "step": 490
    },
    {
      "epoch": 0.09124698011522021,
      "grad_norm": 0.41418707370758057,
      "learning_rate": 0.00019905277217342304,
      "loss": 1.3003,
      "step": 491
    },
    {
      "epoch": 0.09143281917859133,
      "grad_norm": 0.47480088472366333,
      "learning_rate": 0.00019904875210050252,
      "loss": 1.0083,
      "step": 492
    },
    {
      "epoch": 0.09161865824196246,
      "grad_norm": 0.4657514989376068,
      "learning_rate": 0.00019904472355572814,
      "loss": 1.1367,
      "step": 493
    },
    {
      "epoch": 0.09180449730533358,
      "grad_norm": 0.6550506949424744,
      "learning_rate": 0.0001990406865394445,
      "loss": 1.364,
      "step": 494
    },
    {
      "epoch": 0.0919903363687047,
      "grad_norm": 0.44624167680740356,
      "learning_rate": 0.0001990366410519969,
      "loss": 1.0305,
      "step": 495
    },
    {
      "epoch": 0.09217617543207582,
      "grad_norm": 0.5374132990837097,
      "learning_rate": 0.00019903258709373132,
      "loss": 1.3057,
      "step": 496
    },
    {
      "epoch": 0.09236201449544694,
      "grad_norm": 0.5603333711624146,
      "learning_rate": 0.00019902852466499452,
      "loss": 1.2736,
      "step": 497
    },
    {
      "epoch": 0.09254785355881806,
      "grad_norm": 0.43850839138031006,
      "learning_rate": 0.000199024453766134,
      "loss": 1.207,
      "step": 498
    },
    {
      "epoch": 0.09273369262218918,
      "grad_norm": 0.47750213742256165,
      "learning_rate": 0.00019902037439749792,
      "loss": 1.158,
      "step": 499
    },
    {
      "epoch": 0.0929195316855603,
      "grad_norm": 0.46765515208244324,
      "learning_rate": 0.00019901628655943522,
      "loss": 1.0762,
      "step": 500
    },
    {
      "epoch": 0.09310537074893142,
      "grad_norm": 0.46739229559898376,
      "learning_rate": 0.00019901219025229554,
      "loss": 1.177,
      "step": 501
    },
    {
      "epoch": 0.09329120981230254,
      "grad_norm": 0.5190978050231934,
      "learning_rate": 0.0001990080854764292,
      "loss": 1.0188,
      "step": 502
    },
    {
      "epoch": 0.09347704887567367,
      "grad_norm": 0.49740681052207947,
      "learning_rate": 0.00019900397223218734,
      "loss": 1.2849,
      "step": 503
    },
    {
      "epoch": 0.09366288793904479,
      "grad_norm": 0.4239771068096161,
      "learning_rate": 0.00019899985051992177,
      "loss": 1.2079,
      "step": 504
    },
    {
      "epoch": 0.09384872700241591,
      "grad_norm": 0.5220719575881958,
      "learning_rate": 0.000198995720339985,
      "loss": 1.1537,
      "step": 505
    },
    {
      "epoch": 0.09403456606578703,
      "grad_norm": 0.4205484986305237,
      "learning_rate": 0.0001989915816927303,
      "loss": 1.057,
      "step": 506
    },
    {
      "epoch": 0.09422040512915815,
      "grad_norm": 0.4098726511001587,
      "learning_rate": 0.0001989874345785117,
      "loss": 1.0374,
      "step": 507
    },
    {
      "epoch": 0.09440624419252927,
      "grad_norm": 0.4050571024417877,
      "learning_rate": 0.00019898327899768386,
      "loss": 0.9851,
      "step": 508
    },
    {
      "epoch": 0.09459208325590039,
      "grad_norm": 0.560594916343689,
      "learning_rate": 0.00019897911495060224,
      "loss": 1.032,
      "step": 509
    },
    {
      "epoch": 0.09477792231927151,
      "grad_norm": 0.5087330937385559,
      "learning_rate": 0.00019897494243762301,
      "loss": 1.0799,
      "step": 510
    },
    {
      "epoch": 0.09496376138264263,
      "grad_norm": 0.40081334114074707,
      "learning_rate": 0.00019897076145910302,
      "loss": 1.1713,
      "step": 511
    },
    {
      "epoch": 0.09514960044601375,
      "grad_norm": 0.4222041964530945,
      "learning_rate": 0.0001989665720153999,
      "loss": 1.0581,
      "step": 512
    },
    {
      "epoch": 0.09533543950938488,
      "grad_norm": 0.47098875045776367,
      "learning_rate": 0.00019896237410687198,
      "loss": 1.1315,
      "step": 513
    },
    {
      "epoch": 0.095521278572756,
      "grad_norm": 0.41529735922813416,
      "learning_rate": 0.00019895816773387833,
      "loss": 1.1249,
      "step": 514
    },
    {
      "epoch": 0.09570711763612712,
      "grad_norm": 0.5826699733734131,
      "learning_rate": 0.0001989539528967787,
      "loss": 1.3797,
      "step": 515
    },
    {
      "epoch": 0.09589295669949824,
      "grad_norm": 0.4980536997318268,
      "learning_rate": 0.00019894972959593363,
      "loss": 1.0622,
      "step": 516
    },
    {
      "epoch": 0.09607879576286936,
      "grad_norm": 0.4493926763534546,
      "learning_rate": 0.00019894549783170434,
      "loss": 1.0394,
      "step": 517
    },
    {
      "epoch": 0.09626463482624048,
      "grad_norm": 0.4388639032840729,
      "learning_rate": 0.00019894125760445275,
      "loss": 1.0333,
      "step": 518
    },
    {
      "epoch": 0.0964504738896116,
      "grad_norm": 0.5359880924224854,
      "learning_rate": 0.00019893700891454157,
      "loss": 1.4027,
      "step": 519
    },
    {
      "epoch": 0.09663631295298272,
      "grad_norm": 0.459553062915802,
      "learning_rate": 0.00019893275176233416,
      "loss": 1.2656,
      "step": 520
    },
    {
      "epoch": 0.09682215201635384,
      "grad_norm": 0.5185438990592957,
      "learning_rate": 0.0001989284861481947,
      "loss": 0.9231,
      "step": 521
    },
    {
      "epoch": 0.09700799107972496,
      "grad_norm": 0.3912256062030792,
      "learning_rate": 0.000198924212072488,
      "loss": 1.1295,
      "step": 522
    },
    {
      "epoch": 0.09719383014309609,
      "grad_norm": 0.41744834184646606,
      "learning_rate": 0.0001989199295355796,
      "loss": 1.0672,
      "step": 523
    },
    {
      "epoch": 0.09737966920646719,
      "grad_norm": 0.4016173779964447,
      "learning_rate": 0.0001989156385378359,
      "loss": 1.0234,
      "step": 524
    },
    {
      "epoch": 0.09756550826983831,
      "grad_norm": 0.422979474067688,
      "learning_rate": 0.0001989113390796238,
      "loss": 1.0978,
      "step": 525
    },
    {
      "epoch": 0.09775134733320943,
      "grad_norm": 0.529803991317749,
      "learning_rate": 0.00019890703116131112,
      "loss": 1.3601,
      "step": 526
    },
    {
      "epoch": 0.09793718639658056,
      "grad_norm": 0.3872774839401245,
      "learning_rate": 0.00019890271478326628,
      "loss": 1.0105,
      "step": 527
    },
    {
      "epoch": 0.09812302545995168,
      "grad_norm": 0.5113834738731384,
      "learning_rate": 0.0001988983899458585,
      "loss": 1.3075,
      "step": 528
    },
    {
      "epoch": 0.0983088645233228,
      "grad_norm": 0.4002736508846283,
      "learning_rate": 0.00019889405664945766,
      "loss": 1.294,
      "step": 529
    },
    {
      "epoch": 0.09849470358669392,
      "grad_norm": 0.49016720056533813,
      "learning_rate": 0.00019888971489443445,
      "loss": 0.9481,
      "step": 530
    },
    {
      "epoch": 0.09868054265006504,
      "grad_norm": 0.6148933172225952,
      "learning_rate": 0.00019888536468116018,
      "loss": 1.0486,
      "step": 531
    },
    {
      "epoch": 0.09886638171343616,
      "grad_norm": 0.4160175621509552,
      "learning_rate": 0.00019888100601000695,
      "loss": 0.9821,
      "step": 532
    },
    {
      "epoch": 0.09905222077680728,
      "grad_norm": 0.4974881410598755,
      "learning_rate": 0.00019887663888134755,
      "loss": 1.0666,
      "step": 533
    },
    {
      "epoch": 0.0992380598401784,
      "grad_norm": 0.5117452144622803,
      "learning_rate": 0.00019887226329555553,
      "loss": 1.2383,
      "step": 534
    },
    {
      "epoch": 0.09942389890354952,
      "grad_norm": 0.4444761276245117,
      "learning_rate": 0.00019886787925300515,
      "loss": 1.1308,
      "step": 535
    },
    {
      "epoch": 0.09960973796692064,
      "grad_norm": 0.5025230646133423,
      "learning_rate": 0.00019886348675407137,
      "loss": 0.9898,
      "step": 536
    },
    {
      "epoch": 0.09979557703029177,
      "grad_norm": 0.4477807283401489,
      "learning_rate": 0.00019885908579912994,
      "loss": 1.1329,
      "step": 537
    },
    {
      "epoch": 0.09998141609366289,
      "grad_norm": 0.5180255174636841,
      "learning_rate": 0.00019885467638855716,
      "loss": 1.2191,
      "step": 538
    },
    {
      "epoch": 0.10016725515703401,
      "grad_norm": 0.48357728123664856,
      "learning_rate": 0.0001988502585227303,
      "loss": 0.9723,
      "step": 539
    },
    {
      "epoch": 0.10035309422040513,
      "grad_norm": 0.5262537598609924,
      "learning_rate": 0.0001988458322020272,
      "loss": 1.2641,
      "step": 540
    },
    {
      "epoch": 0.10053893328377625,
      "grad_norm": 0.39484769105911255,
      "learning_rate": 0.0001988413974268264,
      "loss": 1.0507,
      "step": 541
    },
    {
      "epoch": 0.10072477234714737,
      "grad_norm": 0.4654625654220581,
      "learning_rate": 0.00019883695419750728,
      "loss": 1.0045,
      "step": 542
    },
    {
      "epoch": 0.10091061141051849,
      "grad_norm": 0.4511539041996002,
      "learning_rate": 0.00019883250251444983,
      "loss": 1.0578,
      "step": 543
    },
    {
      "epoch": 0.10109645047388961,
      "grad_norm": 0.48877015709877014,
      "learning_rate": 0.00019882804237803488,
      "loss": 0.892,
      "step": 544
    },
    {
      "epoch": 0.10128228953726073,
      "grad_norm": 0.4010559916496277,
      "learning_rate": 0.0001988235737886438,
      "loss": 1.0008,
      "step": 545
    },
    {
      "epoch": 0.10146812860063185,
      "grad_norm": 0.4214247763156891,
      "learning_rate": 0.00019881909674665892,
      "loss": 1.2221,
      "step": 546
    },
    {
      "epoch": 0.10165396766400298,
      "grad_norm": 0.47292011976242065,
      "learning_rate": 0.0001988146112524631,
      "loss": 1.1664,
      "step": 547
    },
    {
      "epoch": 0.1018398067273741,
      "grad_norm": 0.6377806663513184,
      "learning_rate": 0.00019881011730644004,
      "loss": 1.1985,
      "step": 548
    },
    {
      "epoch": 0.10202564579074522,
      "grad_norm": 0.46788981556892395,
      "learning_rate": 0.00019880561490897406,
      "loss": 1.236,
      "step": 549
    },
    {
      "epoch": 0.10221148485411634,
      "grad_norm": 0.42801347374916077,
      "learning_rate": 0.0001988011040604503,
      "loss": 1.183,
      "step": 550
    },
    {
      "epoch": 0.10239732391748746,
      "grad_norm": 0.4006738066673279,
      "learning_rate": 0.00019879658476125458,
      "loss": 1.1887,
      "step": 551
    },
    {
      "epoch": 0.10258316298085858,
      "grad_norm": 0.45762765407562256,
      "learning_rate": 0.00019879205701177344,
      "loss": 1.2906,
      "step": 552
    },
    {
      "epoch": 0.1027690020442297,
      "grad_norm": 0.4175111949443817,
      "learning_rate": 0.0001987875208123941,
      "loss": 1.1363,
      "step": 553
    },
    {
      "epoch": 0.10295484110760082,
      "grad_norm": 0.4322377145290375,
      "learning_rate": 0.00019878297616350465,
      "loss": 1.0614,
      "step": 554
    },
    {
      "epoch": 0.10314068017097194,
      "grad_norm": 0.5138376355171204,
      "learning_rate": 0.00019877842306549376,
      "loss": 1.0583,
      "step": 555
    },
    {
      "epoch": 0.10332651923434306,
      "grad_norm": 0.4600732624530792,
      "learning_rate": 0.0001987738615187508,
      "loss": 1.0225,
      "step": 556
    },
    {
      "epoch": 0.10351235829771419,
      "grad_norm": 0.549647867679596,
      "learning_rate": 0.00019876929152366603,
      "loss": 1.2555,
      "step": 557
    },
    {
      "epoch": 0.1036981973610853,
      "grad_norm": 0.47968605160713196,
      "learning_rate": 0.00019876471308063028,
      "loss": 1.1949,
      "step": 558
    },
    {
      "epoch": 0.10388403642445643,
      "grad_norm": 0.4664756655693054,
      "learning_rate": 0.00019876012619003515,
      "loss": 1.2603,
      "step": 559
    },
    {
      "epoch": 0.10406987548782753,
      "grad_norm": 0.45363089442253113,
      "learning_rate": 0.00019875553085227295,
      "loss": 1.2025,
      "step": 560
    },
    {
      "epoch": 0.10425571455119866,
      "grad_norm": 0.4678725302219391,
      "learning_rate": 0.00019875092706773682,
      "loss": 1.1634,
      "step": 561
    },
    {
      "epoch": 0.10444155361456978,
      "grad_norm": 0.4928404986858368,
      "learning_rate": 0.00019874631483682046,
      "loss": 1.2043,
      "step": 562
    },
    {
      "epoch": 0.1046273926779409,
      "grad_norm": 0.39862075448036194,
      "learning_rate": 0.00019874169415991832,
      "loss": 1.1535,
      "step": 563
    },
    {
      "epoch": 0.10481323174131202,
      "grad_norm": 0.4637607932090759,
      "learning_rate": 0.0001987370650374257,
      "loss": 1.1165,
      "step": 564
    },
    {
      "epoch": 0.10499907080468314,
      "grad_norm": 0.4636867046356201,
      "learning_rate": 0.0001987324274697385,
      "loss": 1.3045,
      "step": 565
    },
    {
      "epoch": 0.10518490986805426,
      "grad_norm": 0.48317477107048035,
      "learning_rate": 0.00019872778145725342,
      "loss": 1.2164,
      "step": 566
    },
    {
      "epoch": 0.10537074893142538,
      "grad_norm": 0.5793139338493347,
      "learning_rate": 0.00019872312700036778,
      "loss": 0.9508,
      "step": 567
    },
    {
      "epoch": 0.1055565879947965,
      "grad_norm": 0.49341946840286255,
      "learning_rate": 0.00019871846409947973,
      "loss": 1.2695,
      "step": 568
    },
    {
      "epoch": 0.10574242705816762,
      "grad_norm": 0.4591580033302307,
      "learning_rate": 0.0001987137927549881,
      "loss": 1.2874,
      "step": 569
    },
    {
      "epoch": 0.10592826612153874,
      "grad_norm": 0.4602017402648926,
      "learning_rate": 0.00019870911296729243,
      "loss": 1.3729,
      "step": 570
    },
    {
      "epoch": 0.10611410518490987,
      "grad_norm": 0.34223201870918274,
      "learning_rate": 0.00019870442473679298,
      "loss": 0.6184,
      "step": 571
    },
    {
      "epoch": 0.10629994424828099,
      "grad_norm": 0.49571168422698975,
      "learning_rate": 0.00019869972806389074,
      "loss": 1.0509,
      "step": 572
    },
    {
      "epoch": 0.10648578331165211,
      "grad_norm": 0.4025332033634186,
      "learning_rate": 0.00019869502294898744,
      "loss": 1.0083,
      "step": 573
    },
    {
      "epoch": 0.10667162237502323,
      "grad_norm": 0.5795102119445801,
      "learning_rate": 0.00019869030939248555,
      "loss": 1.2375,
      "step": 574
    },
    {
      "epoch": 0.10685746143839435,
      "grad_norm": 0.49539127945899963,
      "learning_rate": 0.00019868558739478815,
      "loss": 1.1855,
      "step": 575
    },
    {
      "epoch": 0.10704330050176547,
      "grad_norm": 0.44701632857322693,
      "learning_rate": 0.00019868085695629923,
      "loss": 1.0034,
      "step": 576
    },
    {
      "epoch": 0.10722913956513659,
      "grad_norm": 0.4629371762275696,
      "learning_rate": 0.0001986761180774233,
      "loss": 1.0642,
      "step": 577
    },
    {
      "epoch": 0.10741497862850771,
      "grad_norm": 0.44119590520858765,
      "learning_rate": 0.00019867137075856573,
      "loss": 0.9856,
      "step": 578
    },
    {
      "epoch": 0.10760081769187883,
      "grad_norm": 0.5623904466629028,
      "learning_rate": 0.00019866661500013256,
      "loss": 1.2922,
      "step": 579
    },
    {
      "epoch": 0.10778665675524995,
      "grad_norm": 0.5019487142562866,
      "learning_rate": 0.00019866185080253057,
      "loss": 1.2963,
      "step": 580
    },
    {
      "epoch": 0.10797249581862108,
      "grad_norm": 0.46231526136398315,
      "learning_rate": 0.00019865707816616723,
      "loss": 1.1691,
      "step": 581
    },
    {
      "epoch": 0.1081583348819922,
      "grad_norm": 0.3911663591861725,
      "learning_rate": 0.0001986522970914508,
      "loss": 1.0292,
      "step": 582
    },
    {
      "epoch": 0.10834417394536332,
      "grad_norm": 0.48484066128730774,
      "learning_rate": 0.00019864750757879015,
      "loss": 1.3642,
      "step": 583
    },
    {
      "epoch": 0.10853001300873444,
      "grad_norm": 0.44198179244995117,
      "learning_rate": 0.000198642709628595,
      "loss": 0.8633,
      "step": 584
    },
    {
      "epoch": 0.10871585207210556,
      "grad_norm": 0.5480027198791504,
      "learning_rate": 0.00019863790324127568,
      "loss": 1.2279,
      "step": 585
    },
    {
      "epoch": 0.10890169113547668,
      "grad_norm": 0.5162123441696167,
      "learning_rate": 0.00019863308841724334,
      "loss": 1.3003,
      "step": 586
    },
    {
      "epoch": 0.1090875301988478,
      "grad_norm": 0.4140394628047943,
      "learning_rate": 0.00019862826515690973,
      "loss": 0.8132,
      "step": 587
    },
    {
      "epoch": 0.10927336926221892,
      "grad_norm": 0.4862360954284668,
      "learning_rate": 0.00019862343346068747,
      "loss": 0.9631,
      "step": 588
    },
    {
      "epoch": 0.10945920832559004,
      "grad_norm": 0.4251142144203186,
      "learning_rate": 0.00019861859332898977,
      "loss": 0.9675,
      "step": 589
    },
    {
      "epoch": 0.10964504738896116,
      "grad_norm": 0.44398531317710876,
      "learning_rate": 0.00019861374476223064,
      "loss": 1.1292,
      "step": 590
    },
    {
      "epoch": 0.10983088645233229,
      "grad_norm": 0.42411577701568604,
      "learning_rate": 0.0001986088877608248,
      "loss": 1.0539,
      "step": 591
    },
    {
      "epoch": 0.1100167255157034,
      "grad_norm": 0.3475162088871002,
      "learning_rate": 0.00019860402232518766,
      "loss": 0.5955,
      "step": 592
    },
    {
      "epoch": 0.11020256457907453,
      "grad_norm": 0.43855538964271545,
      "learning_rate": 0.00019859914845573543,
      "loss": 1.1307,
      "step": 593
    },
    {
      "epoch": 0.11038840364244565,
      "grad_norm": 0.5022986531257629,
      "learning_rate": 0.00019859426615288488,
      "loss": 0.9466,
      "step": 594
    },
    {
      "epoch": 0.11057424270581677,
      "grad_norm": 0.44647181034088135,
      "learning_rate": 0.00019858937541705364,
      "loss": 1.0578,
      "step": 595
    },
    {
      "epoch": 0.11076008176918788,
      "grad_norm": 0.41280534863471985,
      "learning_rate": 0.00019858447624866008,
      "loss": 0.9337,
      "step": 596
    },
    {
      "epoch": 0.110945920832559,
      "grad_norm": 0.4094390273094177,
      "learning_rate": 0.00019857956864812316,
      "loss": 1.1393,
      "step": 597
    },
    {
      "epoch": 0.11113175989593012,
      "grad_norm": 0.4634653329849243,
      "learning_rate": 0.00019857465261586272,
      "loss": 1.1014,
      "step": 598
    },
    {
      "epoch": 0.11131759895930124,
      "grad_norm": 0.46472883224487305,
      "learning_rate": 0.00019856972815229917,
      "loss": 0.8368,
      "step": 599
    },
    {
      "epoch": 0.11150343802267236,
      "grad_norm": 0.4838216304779053,
      "learning_rate": 0.00019856479525785372,
      "loss": 1.0311,
      "step": 600
    },
    {
      "epoch": 0.11168927708604348,
      "grad_norm": 0.5469513535499573,
      "learning_rate": 0.00019855985393294834,
      "loss": 1.2409,
      "step": 601
    },
    {
      "epoch": 0.1118751161494146,
      "grad_norm": 0.5099655985832214,
      "learning_rate": 0.00019855490417800562,
      "loss": 1.2431,
      "step": 602
    },
    {
      "epoch": 0.11206095521278572,
      "grad_norm": 0.6649165749549866,
      "learning_rate": 0.00019854994599344893,
      "loss": 1.3869,
      "step": 603
    },
    {
      "epoch": 0.11224679427615684,
      "grad_norm": 0.46297532320022583,
      "learning_rate": 0.00019854497937970238,
      "loss": 1.2108,
      "step": 604
    },
    {
      "epoch": 0.11243263333952797,
      "grad_norm": 0.46323227882385254,
      "learning_rate": 0.00019854000433719074,
      "loss": 1.0765,
      "step": 605
    },
    {
      "epoch": 0.11261847240289909,
      "grad_norm": 0.44460347294807434,
      "learning_rate": 0.00019853502086633956,
      "loss": 1.1857,
      "step": 606
    },
    {
      "epoch": 0.11280431146627021,
      "grad_norm": 0.47829505801200867,
      "learning_rate": 0.00019853002896757512,
      "loss": 1.2685,
      "step": 607
    },
    {
      "epoch": 0.11299015052964133,
      "grad_norm": 0.5406739711761475,
      "learning_rate": 0.00019852502864132433,
      "loss": 1.1095,
      "step": 608
    },
    {
      "epoch": 0.11317598959301245,
      "grad_norm": 0.46139219403266907,
      "learning_rate": 0.00019852001988801486,
      "loss": 1.0732,
      "step": 609
    },
    {
      "epoch": 0.11336182865638357,
      "grad_norm": 0.47796913981437683,
      "learning_rate": 0.00019851500270807518,
      "loss": 1.184,
      "step": 610
    },
    {
      "epoch": 0.11354766771975469,
      "grad_norm": 0.4437905251979828,
      "learning_rate": 0.00019850997710193444,
      "loss": 1.0998,
      "step": 611
    },
    {
      "epoch": 0.11373350678312581,
      "grad_norm": 0.48870012164115906,
      "learning_rate": 0.00019850494307002242,
      "loss": 1.133,
      "step": 612
    },
    {
      "epoch": 0.11391934584649693,
      "grad_norm": 0.4321851134300232,
      "learning_rate": 0.0001984999006127697,
      "loss": 1.125,
      "step": 613
    },
    {
      "epoch": 0.11410518490986805,
      "grad_norm": 0.5969387292861938,
      "learning_rate": 0.00019849484973060764,
      "loss": 1.3145,
      "step": 614
    },
    {
      "epoch": 0.11429102397323918,
      "grad_norm": 0.5733935832977295,
      "learning_rate": 0.00019848979042396816,
      "loss": 1.2648,
      "step": 615
    },
    {
      "epoch": 0.1144768630366103,
      "grad_norm": 0.39343923330307007,
      "learning_rate": 0.0001984847226932841,
      "loss": 1.0274,
      "step": 616
    },
    {
      "epoch": 0.11466270209998142,
      "grad_norm": 0.46703800559043884,
      "learning_rate": 0.0001984796465389888,
      "loss": 1.4018,
      "step": 617
    },
    {
      "epoch": 0.11484854116335254,
      "grad_norm": 0.4704144299030304,
      "learning_rate": 0.00019847456196151652,
      "loss": 1.087,
      "step": 618
    },
    {
      "epoch": 0.11503438022672366,
      "grad_norm": 0.5104250311851501,
      "learning_rate": 0.0001984694689613021,
      "loss": 1.0647,
      "step": 619
    },
    {
      "epoch": 0.11522021929009478,
      "grad_norm": 0.47510913014411926,
      "learning_rate": 0.0001984643675387812,
      "loss": 0.9258,
      "step": 620
    },
    {
      "epoch": 0.1154060583534659,
      "grad_norm": 0.5313277840614319,
      "learning_rate": 0.00019845925769439014,
      "loss": 1.3115,
      "step": 621
    },
    {
      "epoch": 0.11559189741683702,
      "grad_norm": 0.42490604519844055,
      "learning_rate": 0.00019845413942856595,
      "loss": 0.9596,
      "step": 622
    },
    {
      "epoch": 0.11577773648020814,
      "grad_norm": 0.40668928623199463,
      "learning_rate": 0.00019844901274174644,
      "loss": 0.9975,
      "step": 623
    },
    {
      "epoch": 0.11596357554357926,
      "grad_norm": 0.44957441091537476,
      "learning_rate": 0.0001984438776343701,
      "loss": 0.9959,
      "step": 624
    },
    {
      "epoch": 0.11614941460695039,
      "grad_norm": 0.427727609872818,
      "learning_rate": 0.0001984387341068761,
      "loss": 1.0412,
      "step": 625
    },
    {
      "epoch": 0.1163352536703215,
      "grad_norm": 0.4397459328174591,
      "learning_rate": 0.00019843358215970445,
      "loss": 1.2047,
      "step": 626
    },
    {
      "epoch": 0.11652109273369263,
      "grad_norm": 0.522113561630249,
      "learning_rate": 0.00019842842179329577,
      "loss": 0.9631,
      "step": 627
    },
    {
      "epoch": 0.11670693179706375,
      "grad_norm": 0.4980854094028473,
      "learning_rate": 0.00019842325300809143,
      "loss": 1.1708,
      "step": 628
    },
    {
      "epoch": 0.11689277086043487,
      "grad_norm": 0.4923211634159088,
      "learning_rate": 0.00019841807580453359,
      "loss": 1.0419,
      "step": 629
    },
    {
      "epoch": 0.11707860992380599,
      "grad_norm": 0.38797327876091003,
      "learning_rate": 0.00019841289018306498,
      "loss": 0.9371,
      "step": 630
    },
    {
      "epoch": 0.11726444898717711,
      "grad_norm": 0.37108826637268066,
      "learning_rate": 0.00019840769614412916,
      "loss": 0.7634,
      "step": 631
    },
    {
      "epoch": 0.11745028805054822,
      "grad_norm": 0.44877228140830994,
      "learning_rate": 0.0001984024936881704,
      "loss": 0.8744,
      "step": 632
    },
    {
      "epoch": 0.11763612711391934,
      "grad_norm": 0.5255552530288696,
      "learning_rate": 0.0001983972828156337,
      "loss": 1.2011,
      "step": 633
    },
    {
      "epoch": 0.11782196617729046,
      "grad_norm": 0.46886783838272095,
      "learning_rate": 0.00019839206352696476,
      "loss": 1.2743,
      "step": 634
    },
    {
      "epoch": 0.11800780524066158,
      "grad_norm": 0.6389307379722595,
      "learning_rate": 0.00019838683582260993,
      "loss": 1.2846,
      "step": 635
    },
    {
      "epoch": 0.1181936443040327,
      "grad_norm": 0.39743301272392273,
      "learning_rate": 0.0001983815997030164,
      "loss": 0.865,
      "step": 636
    },
    {
      "epoch": 0.11837948336740382,
      "grad_norm": 0.5184935331344604,
      "learning_rate": 0.00019837635516863205,
      "loss": 1.0444,
      "step": 637
    },
    {
      "epoch": 0.11856532243077494,
      "grad_norm": 0.4453527629375458,
      "learning_rate": 0.0001983711022199054,
      "loss": 0.9113,
      "step": 638
    },
    {
      "epoch": 0.11875116149414607,
      "grad_norm": NaN,
      "learning_rate": 0.0001983711022199054,
      "loss": 5.2668,
      "step": 639
    },
    {
      "epoch": 0.11893700055751719,
      "grad_norm": 0.5004279017448425,
      "learning_rate": 0.00019836584085728575,
      "loss": 0.9127,
      "step": 640
    },
    {
      "epoch": 0.11912283962088831,
      "grad_norm": 0.5233067274093628,
      "learning_rate": 0.00019836057108122316,
      "loss": 1.1963,
      "step": 641
    },
    {
      "epoch": 0.11930867868425943,
      "grad_norm": 0.5132344961166382,
      "learning_rate": 0.00019835529289216833,
      "loss": 1.169,
      "step": 642
    },
    {
      "epoch": 0.11949451774763055,
      "grad_norm": 0.3598637282848358,
      "learning_rate": 0.00019835000629057271,
      "loss": 0.5946,
      "step": 643
    },
    {
      "epoch": 0.11968035681100167,
      "grad_norm": 0.5927154421806335,
      "learning_rate": 0.0001983447112768885,
      "loss": 1.1492,
      "step": 644
    },
    {
      "epoch": 0.11986619587437279,
      "grad_norm": 0.4621802568435669,
      "learning_rate": 0.00019833940785156858,
      "loss": 1.1075,
      "step": 645
    },
    {
      "epoch": 0.12005203493774391,
      "grad_norm": 0.5019433498382568,
      "learning_rate": 0.00019833409601506657,
      "loss": 1.1041,
      "step": 646
    },
    {
      "epoch": 0.12023787400111503,
      "grad_norm": 0.46227413415908813,
      "learning_rate": 0.00019832877576783683,
      "loss": 1.2795,
      "step": 647
    },
    {
      "epoch": 0.12042371306448615,
      "grad_norm": 0.6081055998802185,
      "learning_rate": 0.00019832344711033436,
      "loss": 1.297,
      "step": 648
    },
    {
      "epoch": 0.12060955212785728,
      "grad_norm": 0.457120418548584,
      "learning_rate": 0.00019831811004301494,
      "loss": 1.0885,
      "step": 649
    },
    {
      "epoch": 0.1207953911912284,
      "grad_norm": 0.33518850803375244,
      "learning_rate": 0.0001983127645663351,
      "loss": 0.8171,
      "step": 650
    },
    {
      "epoch": 0.12098123025459952,
      "grad_norm": 0.55155348777771,
      "learning_rate": 0.00019830741068075202,
      "loss": 1.4775,
      "step": 651
    },
    {
      "epoch": 0.12116706931797064,
      "grad_norm": 0.44404229521751404,
      "learning_rate": 0.0001983020483867236,
      "loss": 1.2144,
      "step": 652
    },
    {
      "epoch": 0.12135290838134176,
      "grad_norm": 0.4665181338787079,
      "learning_rate": 0.00019829667768470858,
      "loss": 1.1258,
      "step": 653
    },
    {
      "epoch": 0.12153874744471288,
      "grad_norm": 0.4867573380470276,
      "learning_rate": 0.0001982912985751662,
      "loss": 1.1113,
      "step": 654
    },
    {
      "epoch": 0.121724586508084,
      "grad_norm": 0.4458369016647339,
      "learning_rate": 0.00019828591105855667,
      "loss": 1.1641,
      "step": 655
    },
    {
      "epoch": 0.12191042557145512,
      "grad_norm": 0.3621409833431244,
      "learning_rate": 0.00019828051513534075,
      "loss": 0.8111,
      "step": 656
    },
    {
      "epoch": 0.12209626463482624,
      "grad_norm": 0.4944764971733093,
      "learning_rate": 0.00019827511080597993,
      "loss": 1.1098,
      "step": 657
    },
    {
      "epoch": 0.12228210369819736,
      "grad_norm": 0.4163343608379364,
      "learning_rate": 0.00019826969807093647,
      "loss": 1.0126,
      "step": 658
    },
    {
      "epoch": 0.12246794276156849,
      "grad_norm": 0.43734899163246155,
      "learning_rate": 0.00019826427693067338,
      "loss": 1.0961,
      "step": 659
    },
    {
      "epoch": 0.1226537818249396,
      "grad_norm": 0.5198599100112915,
      "learning_rate": 0.00019825884738565428,
      "loss": 1.0699,
      "step": 660
    },
    {
      "epoch": 0.12283962088831073,
      "grad_norm": 0.548116147518158,
      "learning_rate": 0.0001982534094363436,
      "loss": 1.2081,
      "step": 661
    },
    {
      "epoch": 0.12302545995168185,
      "grad_norm": 0.526623010635376,
      "learning_rate": 0.0001982479630832065,
      "loss": 1.2044,
      "step": 662
    },
    {
      "epoch": 0.12321129901505297,
      "grad_norm": 0.4783700108528137,
      "learning_rate": 0.00019824250832670872,
      "loss": 1.3622,
      "step": 663
    },
    {
      "epoch": 0.12339713807842409,
      "grad_norm": 0.5150986313819885,
      "learning_rate": 0.00019823704516731691,
      "loss": 0.9491,
      "step": 664
    },
    {
      "epoch": 0.12358297714179521,
      "grad_norm": 0.3757990896701813,
      "learning_rate": 0.0001982315736054983,
      "loss": 0.8711,
      "step": 665
    },
    {
      "epoch": 0.12376881620516633,
      "grad_norm": 0.599030077457428,
      "learning_rate": 0.0001982260936417209,
      "loss": 0.9816,
      "step": 666
    },
    {
      "epoch": 0.12395465526853745,
      "grad_norm": 0.5053818821907043,
      "learning_rate": 0.0001982206052764534,
      "loss": 0.8597,
      "step": 667
    },
    {
      "epoch": 0.12414049433190857,
      "grad_norm": 0.46872928738594055,
      "learning_rate": 0.00019821510851016525,
      "loss": 1.1802,
      "step": 668
    },
    {
      "epoch": 0.12432633339527968,
      "grad_norm": 0.4599286913871765,
      "learning_rate": 0.00019820960334332658,
      "loss": 1.0911,
      "step": 669
    },
    {
      "epoch": 0.1245121724586508,
      "grad_norm": 0.564627468585968,
      "learning_rate": 0.00019820408977640831,
      "loss": 1.0628,
      "step": 670
    },
    {
      "epoch": 0.12469801152202192,
      "grad_norm": 0.5068032741546631,
      "learning_rate": 0.00019819856780988197,
      "loss": 1.3099,
      "step": 671
    },
    {
      "epoch": 0.12488385058539304,
      "grad_norm": 0.5990276336669922,
      "learning_rate": 0.00019819303744421992,
      "loss": 1.1733,
      "step": 672
    },
    {
      "epoch": 0.12506968964876417,
      "grad_norm": 0.4374839663505554,
      "learning_rate": 0.00019818749867989515,
      "loss": 1.1439,
      "step": 673
    },
    {
      "epoch": 0.1252555287121353,
      "grad_norm": 0.511102557182312,
      "learning_rate": 0.00019818195151738137,
      "loss": 1.1636,
      "step": 674
    },
    {
      "epoch": 0.1254413677755064,
      "grad_norm": 0.5559630990028381,
      "learning_rate": 0.00019817639595715313,
      "loss": 1.1273,
      "step": 675
    },
    {
      "epoch": 0.12562720683887754,
      "grad_norm": 0.4792366027832031,
      "learning_rate": 0.00019817083199968552,
      "loss": 1.1224,
      "step": 676
    },
    {
      "epoch": 0.12581304590224865,
      "grad_norm": 0.46028703451156616,
      "learning_rate": 0.00019816525964545448,
      "loss": 0.9323,
      "step": 677
    },
    {
      "epoch": 0.12599888496561978,
      "grad_norm": 0.5629606246948242,
      "learning_rate": 0.0001981596788949366,
      "loss": 1.2004,
      "step": 678
    },
    {
      "epoch": 0.1261847240289909,
      "grad_norm": 0.5521604418754578,
      "learning_rate": 0.00019815408974860926,
      "loss": 1.0562,
      "step": 679
    },
    {
      "epoch": 0.12637056309236203,
      "grad_norm": 0.41555026173591614,
      "learning_rate": 0.00019814849220695048,
      "loss": 0.9565,
      "step": 680
    },
    {
      "epoch": 0.12655640215573313,
      "grad_norm": 0.5460541248321533,
      "learning_rate": 0.00019814288627043904,
      "loss": 1.113,
      "step": 681
    },
    {
      "epoch": 0.12674224121910427,
      "grad_norm": 0.4217720031738281,
      "learning_rate": 0.0001981372719395544,
      "loss": 1.1451,
      "step": 682
    },
    {
      "epoch": 0.12692808028247538,
      "grad_norm": 0.4192173182964325,
      "learning_rate": 0.00019813164921477682,
      "loss": 1.1911,
      "step": 683
    },
    {
      "epoch": 0.1271139193458465,
      "grad_norm": 0.5131157636642456,
      "learning_rate": 0.00019812601809658718,
      "loss": 1.2259,
      "step": 684
    },
    {
      "epoch": 0.12729975840921762,
      "grad_norm": 0.42514634132385254,
      "learning_rate": 0.00019812037858546713,
      "loss": 1.2219,
      "step": 685
    },
    {
      "epoch": 0.12748559747258875,
      "grad_norm": 0.4878459572792053,
      "learning_rate": 0.00019811473068189904,
      "loss": 1.2935,
      "step": 686
    },
    {
      "epoch": 0.12767143653595986,
      "grad_norm": 0.42948704957962036,
      "learning_rate": 0.00019810907438636596,
      "loss": 1.0748,
      "step": 687
    },
    {
      "epoch": 0.12785727559933097,
      "grad_norm": 0.44495806097984314,
      "learning_rate": 0.00019810340969935174,
      "loss": 1.2191,
      "step": 688
    },
    {
      "epoch": 0.1280431146627021,
      "grad_norm": 0.35490602254867554,
      "learning_rate": 0.00019809773662134083,
      "loss": 0.7925,
      "step": 689
    },
    {
      "epoch": 0.1282289537260732,
      "grad_norm": 0.5111147165298462,
      "learning_rate": 0.0001980920551528185,
      "loss": 1.352,
      "step": 690
    },
    {
      "epoch": 0.12841479278944434,
      "grad_norm": 0.5153658390045166,
      "learning_rate": 0.0001980863652942707,
      "loss": 1.2032,
      "step": 691
    },
    {
      "epoch": 0.12860063185281545,
      "grad_norm": 0.47706758975982666,
      "learning_rate": 0.00019808066704618408,
      "loss": 0.8932,
      "step": 692
    },
    {
      "epoch": 0.12878647091618659,
      "grad_norm": 0.5452196598052979,
      "learning_rate": 0.000198074960409046,
      "loss": 1.1013,
      "step": 693
    },
    {
      "epoch": 0.1289723099795577,
      "grad_norm": 0.43593546748161316,
      "learning_rate": 0.0001980692453833446,
      "loss": 0.9039,
      "step": 694
    },
    {
      "epoch": 0.12915814904292883,
      "grad_norm": 0.4466397166252136,
      "learning_rate": 0.0001980635219695687,
      "loss": 1.206,
      "step": 695
    },
    {
      "epoch": 0.12934398810629993,
      "grad_norm": 0.4870753884315491,
      "learning_rate": 0.00019805779016820782,
      "loss": 1.0871,
      "step": 696
    },
    {
      "epoch": 0.12952982716967107,
      "grad_norm": 0.43623775243759155,
      "learning_rate": 0.0001980520499797522,
      "loss": 1.1557,
      "step": 697
    },
    {
      "epoch": 0.12971566623304218,
      "grad_norm": 0.443454384803772,
      "learning_rate": 0.00019804630140469285,
      "loss": 1.2333,
      "step": 698
    },
    {
      "epoch": 0.1299015052964133,
      "grad_norm": 0.5000869631767273,
      "learning_rate": 0.00019804054444352143,
      "loss": 1.0621,
      "step": 699
    },
    {
      "epoch": 0.13008734435978442,
      "grad_norm": 0.5300038456916809,
      "learning_rate": 0.00019803477909673034,
      "loss": 1.2468,
      "step": 700
    },
    {
      "epoch": 0.13027318342315555,
      "grad_norm": 0.45927757024765015,
      "learning_rate": 0.00019802900536481272,
      "loss": 1.1653,
      "step": 701
    },
    {
      "epoch": 0.13045902248652666,
      "grad_norm": 0.5365282893180847,
      "learning_rate": 0.0001980232232482624,
      "loss": 1.1903,
      "step": 702
    },
    {
      "epoch": 0.1306448615498978,
      "grad_norm": 0.5043763518333435,
      "learning_rate": 0.00019801743274757395,
      "loss": 0.9898,
      "step": 703
    },
    {
      "epoch": 0.1308307006132689,
      "grad_norm": 0.4623168706893921,
      "learning_rate": 0.00019801163386324263,
      "loss": 1.0392,
      "step": 704
    },
    {
      "epoch": 0.13101653967664004,
      "grad_norm": 0.6235971450805664,
      "learning_rate": 0.0001980058265957644,
      "loss": 1.1318,
      "step": 705
    },
    {
      "epoch": 0.13120237874001114,
      "grad_norm": 0.47475048899650574,
      "learning_rate": 0.00019800001094563604,
      "loss": 0.9881,
      "step": 706
    },
    {
      "epoch": 0.13138821780338228,
      "grad_norm": 0.579794704914093,
      "learning_rate": 0.00019799418691335494,
      "loss": 1.0279,
      "step": 707
    },
    {
      "epoch": 0.13157405686675339,
      "grad_norm": 0.48420289158821106,
      "learning_rate": 0.00019798835449941923,
      "loss": 1.1624,
      "step": 708
    },
    {
      "epoch": 0.13175989593012452,
      "grad_norm": 0.49966883659362793,
      "learning_rate": 0.00019798251370432778,
      "loss": 1.2414,
      "step": 709
    },
    {
      "epoch": 0.13194573499349563,
      "grad_norm": 0.5556090474128723,
      "learning_rate": 0.00019797666452858016,
      "loss": 1.1312,
      "step": 710
    },
    {
      "epoch": 0.13213157405686676,
      "grad_norm": 0.474415123462677,
      "learning_rate": 0.00019797080697267667,
      "loss": 1.0425,
      "step": 711
    },
    {
      "epoch": 0.13231741312023787,
      "grad_norm": 0.4529750347137451,
      "learning_rate": 0.00019796494103711834,
      "loss": 1.1246,
      "step": 712
    },
    {
      "epoch": 0.132503252183609,
      "grad_norm": 0.49492666125297546,
      "learning_rate": 0.00019795906672240686,
      "loss": 1.2322,
      "step": 713
    },
    {
      "epoch": 0.1326890912469801,
      "grad_norm": 0.6376146674156189,
      "learning_rate": 0.00019795318402904467,
      "loss": 1.0772,
      "step": 714
    },
    {
      "epoch": 0.13287493031035125,
      "grad_norm": 0.4275801479816437,
      "learning_rate": 0.00019794729295753496,
      "loss": 1.2186,
      "step": 715
    },
    {
      "epoch": 0.13306076937372235,
      "grad_norm": 0.4138648211956024,
      "learning_rate": 0.00019794139350838157,
      "loss": 1.0326,
      "step": 716
    },
    {
      "epoch": 0.1332466084370935,
      "grad_norm": 0.4423656761646271,
      "learning_rate": 0.0001979354856820891,
      "loss": 1.2348,
      "step": 717
    },
    {
      "epoch": 0.1334324475004646,
      "grad_norm": 0.4263252019882202,
      "learning_rate": 0.00019792956947916292,
      "loss": 1.2048,
      "step": 718
    },
    {
      "epoch": 0.13361828656383573,
      "grad_norm": 0.4504089951515198,
      "learning_rate": 0.000197923644900109,
      "loss": 0.9089,
      "step": 719
    },
    {
      "epoch": 0.13380412562720684,
      "grad_norm": 0.5284350514411926,
      "learning_rate": 0.00019791771194543407,
      "loss": 1.0991,
      "step": 720
    },
    {
      "epoch": 0.13398996469057797,
      "grad_norm": 0.5844881534576416,
      "learning_rate": 0.00019791177061564562,
      "loss": 1.2395,
      "step": 721
    },
    {
      "epoch": 0.13417580375394908,
      "grad_norm": 0.5521482825279236,
      "learning_rate": 0.00019790582091125183,
      "loss": 1.1193,
      "step": 722
    },
    {
      "epoch": 0.1343616428173202,
      "grad_norm": 0.6519663333892822,
      "learning_rate": 0.00019789986283276156,
      "loss": 1.0444,
      "step": 723
    },
    {
      "epoch": 0.13454748188069132,
      "grad_norm": 0.5339438915252686,
      "learning_rate": 0.0001978938963806844,
      "loss": 1.235,
      "step": 724
    },
    {
      "epoch": 0.13473332094406243,
      "grad_norm": 0.5313912630081177,
      "learning_rate": 0.00019788792155553075,
      "loss": 1.1505,
      "step": 725
    },
    {
      "epoch": 0.13491916000743356,
      "grad_norm": 0.49416816234588623,
      "learning_rate": 0.0001978819383578116,
      "loss": 0.8922,
      "step": 726
    },
    {
      "epoch": 0.13510499907080467,
      "grad_norm": 0.4863441586494446,
      "learning_rate": 0.0001978759467880387,
      "loss": 1.0582,
      "step": 727
    },
    {
      "epoch": 0.1352908381341758,
      "grad_norm": 0.501708984375,
      "learning_rate": 0.00019786994684672452,
      "loss": 1.205,
      "step": 728
    },
    {
      "epoch": 0.1354766771975469,
      "grad_norm": 0.5261609554290771,
      "learning_rate": 0.0001978639385343823,
      "loss": 1.1186,
      "step": 729
    },
    {
      "epoch": 0.13566251626091805,
      "grad_norm": 0.4654827415943146,
      "learning_rate": 0.00019785792185152587,
      "loss": 1.1819,
      "step": 730
    },
    {
      "epoch": 0.13584835532428916,
      "grad_norm": 0.42379230260849,
      "learning_rate": 0.0001978518967986699,
      "loss": 1.1737,
      "step": 731
    },
    {
      "epoch": 0.1360341943876603,
      "grad_norm": 0.5935065746307373,
      "learning_rate": 0.00019784586337632973,
      "loss": 1.1976,
      "step": 732
    },
    {
      "epoch": 0.1362200334510314,
      "grad_norm": 0.4229237139225006,
      "learning_rate": 0.00019783982158502137,
      "loss": 1.149,
      "step": 733
    },
    {
      "epoch": 0.13640587251440253,
      "grad_norm": 0.4713400900363922,
      "learning_rate": 0.0001978337714252616,
      "loss": 1.2643,
      "step": 734
    },
    {
      "epoch": 0.13659171157777364,
      "grad_norm": 0.49922773241996765,
      "learning_rate": 0.00019782771289756794,
      "loss": 1.2761,
      "step": 735
    },
    {
      "epoch": 0.13677755064114477,
      "grad_norm": 0.4869403541088104,
      "learning_rate": 0.00019782164600245855,
      "loss": 0.9267,
      "step": 736
    },
    {
      "epoch": 0.13696338970451588,
      "grad_norm": 0.40278908610343933,
      "learning_rate": 0.00019781557074045235,
      "loss": 1.0307,
      "step": 737
    },
    {
      "epoch": 0.13714922876788702,
      "grad_norm": 0.4315333962440491,
      "learning_rate": 0.00019780948711206899,
      "loss": 1.1248,
      "step": 738
    },
    {
      "epoch": 0.13733506783125812,
      "grad_norm": 0.5250226259231567,
      "learning_rate": 0.0001978033951178288,
      "loss": 1.412,
      "step": 739
    },
    {
      "epoch": 0.13752090689462926,
      "grad_norm": 0.45913055539131165,
      "learning_rate": 0.00019779729475825284,
      "loss": 1.361,
      "step": 740
    },
    {
      "epoch": 0.13770674595800036,
      "grad_norm": 0.49358081817626953,
      "learning_rate": 0.0001977911860338629,
      "loss": 1.1859,
      "step": 741
    },
    {
      "epoch": 0.1378925850213715,
      "grad_norm": 0.4311683177947998,
      "learning_rate": 0.00019778506894518145,
      "loss": 1.3164,
      "step": 742
    },
    {
      "epoch": 0.1380784240847426,
      "grad_norm": 0.47601014375686646,
      "learning_rate": 0.0001977789434927317,
      "loss": 0.8034,
      "step": 743
    },
    {
      "epoch": 0.13826426314811374,
      "grad_norm": 0.5088006258010864,
      "learning_rate": 0.0001977728096770376,
      "loss": 1.1709,
      "step": 744
    },
    {
      "epoch": 0.13845010221148485,
      "grad_norm": 0.5519676804542542,
      "learning_rate": 0.0001977666674986238,
      "loss": 1.2942,
      "step": 745
    },
    {
      "epoch": 0.13863594127485598,
      "grad_norm": 0.440140038728714,
      "learning_rate": 0.0001977605169580156,
      "loss": 0.8967,
      "step": 746
    },
    {
      "epoch": 0.1388217803382271,
      "grad_norm": 0.3971608877182007,
      "learning_rate": 0.0001977543580557391,
      "loss": 0.708,
      "step": 747
    },
    {
      "epoch": 0.13900761940159823,
      "grad_norm": 0.42031824588775635,
      "learning_rate": 0.0001977481907923211,
      "loss": 0.92,
      "step": 748
    },
    {
      "epoch": 0.13919345846496933,
      "grad_norm": 0.49634236097335815,
      "learning_rate": 0.00019774201516828906,
      "loss": 0.9919,
      "step": 749
    },
    {
      "epoch": 0.13937929752834047,
      "grad_norm": 0.45243629813194275,
      "learning_rate": 0.0001977358311841712,
      "loss": 1.0909,
      "step": 750
    },
    {
      "epoch": 0.13956513659171157,
      "grad_norm": 0.4697221517562866,
      "learning_rate": 0.00019772963884049648,
      "loss": 1.0697,
      "step": 751
    },
    {
      "epoch": 0.1397509756550827,
      "grad_norm": 0.5083733797073364,
      "learning_rate": 0.00019772343813779454,
      "loss": 0.9514,
      "step": 752
    },
    {
      "epoch": 0.13993681471845382,
      "grad_norm": 0.43233105540275574,
      "learning_rate": 0.00019771722907659572,
      "loss": 1.0183,
      "step": 753
    },
    {
      "epoch": 0.14012265378182495,
      "grad_norm": 0.6939888596534729,
      "learning_rate": 0.0001977110116574311,
      "loss": 1.2529,
      "step": 754
    },
    {
      "epoch": 0.14030849284519606,
      "grad_norm": 0.4073755145072937,
      "learning_rate": 0.00019770478588083249,
      "loss": 0.9524,
      "step": 755
    },
    {
      "epoch": 0.1404943319085672,
      "grad_norm": 0.43617358803749084,
      "learning_rate": 0.00019769855174733234,
      "loss": 1.1203,
      "step": 756
    },
    {
      "epoch": 0.1406801709719383,
      "grad_norm": 0.3306081295013428,
      "learning_rate": 0.00019769230925746392,
      "loss": 0.438,
      "step": 757
    },
    {
      "epoch": 0.14086601003530944,
      "grad_norm": 0.4973095953464508,
      "learning_rate": 0.00019768605841176116,
      "loss": 1.0405,
      "step": 758
    },
    {
      "epoch": 0.14105184909868054,
      "grad_norm": 0.4343370497226715,
      "learning_rate": 0.00019767979921075866,
      "loss": 1.2194,
      "step": 759
    },
    {
      "epoch": 0.14123768816205165,
      "grad_norm": 0.5016842484474182,
      "learning_rate": 0.00019767353165499186,
      "loss": 1.2143,
      "step": 760
    },
    {
      "epoch": 0.14142352722542278,
      "grad_norm": 0.563551127910614,
      "learning_rate": 0.00019766725574499677,
      "loss": 1.4206,
      "step": 761
    },
    {
      "epoch": 0.1416093662887939,
      "grad_norm": 0.33113259077072144,
      "learning_rate": 0.00019766097148131025,
      "loss": 0.872,
      "step": 762
    },
    {
      "epoch": 0.14179520535216503,
      "grad_norm": 0.4594592750072479,
      "learning_rate": 0.0001976546788644697,
      "loss": 0.9665,
      "step": 763
    },
    {
      "epoch": 0.14198104441553613,
      "grad_norm": 0.37534263730049133,
      "learning_rate": 0.00019764837789501346,
      "loss": 0.8677,
      "step": 764
    },
    {
      "epoch": 0.14216688347890727,
      "grad_norm": 0.45439133048057556,
      "learning_rate": 0.0001976420685734804,
      "loss": 1.1933,
      "step": 765
    },
    {
      "epoch": 0.14235272254227838,
      "grad_norm": 0.5969803929328918,
      "learning_rate": 0.00019763575090041017,
      "loss": 1.356,
      "step": 766
    },
    {
      "epoch": 0.1425385616056495,
      "grad_norm": 0.49011388421058655,
      "learning_rate": 0.00019762942487634315,
      "loss": 1.1676,
      "step": 767
    },
    {
      "epoch": 0.14272440066902062,
      "grad_norm": 0.4911096692085266,
      "learning_rate": 0.00019762309050182044,
      "loss": 0.9579,
      "step": 768
    },
    {
      "epoch": 0.14291023973239175,
      "grad_norm": 0.5238345861434937,
      "learning_rate": 0.00019761674777738378,
      "loss": 1.2124,
      "step": 769
    },
    {
      "epoch": 0.14309607879576286,
      "grad_norm": 0.5594388842582703,
      "learning_rate": 0.00019761039670357572,
      "loss": 1.2953,
      "step": 770
    },
    {
      "epoch": 0.143281917859134,
      "grad_norm": 0.43093204498291016,
      "learning_rate": 0.00019760403728093947,
      "loss": 1.1205,
      "step": 771
    },
    {
      "epoch": 0.1434677569225051,
      "grad_norm": 0.537513256072998,
      "learning_rate": 0.00019759766951001897,
      "loss": 1.0663,
      "step": 772
    },
    {
      "epoch": 0.14365359598587624,
      "grad_norm": 0.39102646708488464,
      "learning_rate": 0.00019759129339135881,
      "loss": 1.0509,
      "step": 773
    },
    {
      "epoch": 0.14383943504924734,
      "grad_norm": 0.6008369326591492,
      "learning_rate": 0.00019758490892550447,
      "loss": 1.2021,
      "step": 774
    },
    {
      "epoch": 0.14402527411261848,
      "grad_norm": 0.4833708703517914,
      "learning_rate": 0.0001975785161130019,
      "loss": 1.1288,
      "step": 775
    },
    {
      "epoch": 0.14421111317598959,
      "grad_norm": 0.5020431280136108,
      "learning_rate": 0.000197572114954398,
      "loss": 1.2715,
      "step": 776
    },
    {
      "epoch": 0.14439695223936072,
      "grad_norm": 0.45390018820762634,
      "learning_rate": 0.0001975657054502402,
      "loss": 0.9566,
      "step": 777
    },
    {
      "epoch": 0.14458279130273183,
      "grad_norm": 0.4128166437149048,
      "learning_rate": 0.00019755928760107675,
      "loss": 1.0689,
      "step": 778
    },
    {
      "epoch": 0.14476863036610296,
      "grad_norm": 0.4302813410758972,
      "learning_rate": 0.0001975528614074566,
      "loss": 1.1927,
      "step": 779
    },
    {
      "epoch": 0.14495446942947407,
      "grad_norm": 0.45130646228790283,
      "learning_rate": 0.00019754642686992935,
      "loss": 1.1991,
      "step": 780
    },
    {
      "epoch": 0.1451403084928452,
      "grad_norm": 0.4731914699077606,
      "learning_rate": 0.00019753998398904538,
      "loss": 1.175,
      "step": 781
    },
    {
      "epoch": 0.1453261475562163,
      "grad_norm": 0.6347445249557495,
      "learning_rate": 0.00019753353276535577,
      "loss": 1.1229,
      "step": 782
    },
    {
      "epoch": 0.14551198661958745,
      "grad_norm": 0.495601624250412,
      "learning_rate": 0.00019752707319941232,
      "loss": 0.9881,
      "step": 783
    },
    {
      "epoch": 0.14569782568295855,
      "grad_norm": 0.5375710129737854,
      "learning_rate": 0.00019752060529176752,
      "loss": 1.1683,
      "step": 784
    },
    {
      "epoch": 0.1458836647463297,
      "grad_norm": 0.454044371843338,
      "learning_rate": 0.00019751412904297455,
      "loss": 0.8788,
      "step": 785
    },
    {
      "epoch": 0.1460695038097008,
      "grad_norm": 0.4273867905139923,
      "learning_rate": 0.0001975076444535874,
      "loss": 1.1718,
      "step": 786
    },
    {
      "epoch": 0.14625534287307193,
      "grad_norm": 0.4461952745914459,
      "learning_rate": 0.00019750115152416067,
      "loss": 1.179,
      "step": 787
    },
    {
      "epoch": 0.14644118193644304,
      "grad_norm": 0.5559700727462769,
      "learning_rate": 0.0001974946502552497,
      "loss": 1.3204,
      "step": 788
    },
    {
      "epoch": 0.14662702099981417,
      "grad_norm": NaN,
      "learning_rate": 0.0001974946502552497,
      "loss": 5.4482,
      "step": 789
    },
    {
      "epoch": 0.14681286006318528,
      "grad_norm": 0.5265190601348877,
      "learning_rate": 0.00019748814064741058,
      "loss": 1.0285,
      "step": 790
    },
    {
      "epoch": 0.14699869912655641,
      "grad_norm": 0.524591326713562,
      "learning_rate": 0.00019748162270120012,
      "loss": 1.0945,
      "step": 791
    },
    {
      "epoch": 0.14718453818992752,
      "grad_norm": 0.4927036762237549,
      "learning_rate": 0.0001974750964171758,
      "loss": 1.171,
      "step": 792
    },
    {
      "epoch": 0.14737037725329866,
      "grad_norm": 0.5644413232803345,
      "learning_rate": 0.00019746856179589573,
      "loss": 1.2689,
      "step": 793
    },
    {
      "epoch": 0.14755621631666976,
      "grad_norm": 0.5673896670341492,
      "learning_rate": 0.000197462018837919,
      "loss": 1.3814,
      "step": 794
    },
    {
      "epoch": 0.14774205538004087,
      "grad_norm": 0.461016446352005,
      "learning_rate": 0.00019745546754380513,
      "loss": 1.3054,
      "step": 795
    },
    {
      "epoch": 0.147927894443412,
      "grad_norm": 0.49385157227516174,
      "learning_rate": 0.00019744890791411447,
      "loss": 1.0074,
      "step": 796
    },
    {
      "epoch": 0.1481137335067831,
      "grad_norm": 0.43251967430114746,
      "learning_rate": 0.0001974423399494081,
      "loss": 0.9248,
      "step": 797
    },
    {
      "epoch": 0.14829957257015425,
      "grad_norm": 0.5019454956054688,
      "learning_rate": 0.00019743576365024784,
      "loss": 1.089,
      "step": 798
    },
    {
      "epoch": 0.14848541163352535,
      "grad_norm": 0.49163103103637695,
      "learning_rate": 0.00019742917901719608,
      "loss": 1.0207,
      "step": 799
    },
    {
      "epoch": 0.1486712506968965,
      "grad_norm": 0.5061166882514954,
      "learning_rate": 0.00019742258605081608,
      "loss": 1.0134,
      "step": 800
    },
    {
      "epoch": 0.1488570897602676,
      "grad_norm": 0.5079135298728943,
      "learning_rate": 0.00019741598475167175,
      "loss": 1.2507,
      "step": 801
    },
    {
      "epoch": 0.14904292882363873,
      "grad_norm": 0.4781422019004822,
      "learning_rate": 0.00019740937512032768,
      "loss": 1.1281,
      "step": 802
    },
    {
      "epoch": 0.14922876788700984,
      "grad_norm": 0.45758572220802307,
      "learning_rate": 0.00019740275715734925,
      "loss": 1.044,
      "step": 803
    },
    {
      "epoch": 0.14941460695038097,
      "grad_norm": 0.4596036970615387,
      "learning_rate": 0.00019739613086330247,
      "loss": 0.9013,
      "step": 804
    },
    {
      "epoch": 0.14960044601375208,
      "grad_norm": 0.4866902232170105,
      "learning_rate": 0.00019738949623875413,
      "loss": 0.9927,
      "step": 805
    },
    {
      "epoch": 0.14978628507712322,
      "grad_norm": 0.41937676072120667,
      "learning_rate": 0.00019738285328427165,
      "loss": 0.9932,
      "step": 806
    },
    {
      "epoch": 0.14997212414049432,
      "grad_norm": 0.4618746340274811,
      "learning_rate": 0.00019737620200042328,
      "loss": 0.9386,
      "step": 807
    },
    {
      "epoch": 0.15015796320386546,
      "grad_norm": 0.44216638803482056,
      "learning_rate": 0.00019736954238777792,
      "loss": 0.9385,
      "step": 808
    },
    {
      "epoch": 0.15034380226723656,
      "grad_norm": 0.49641454219818115,
      "learning_rate": 0.00019736287444690513,
      "loss": 1.0367,
      "step": 809
    },
    {
      "epoch": 0.1505296413306077,
      "grad_norm": 0.5560820698738098,
      "learning_rate": 0.00019735619817837528,
      "loss": 1.3052,
      "step": 810
    },
    {
      "epoch": 0.1507154803939788,
      "grad_norm": 0.6007441282272339,
      "learning_rate": 0.00019734951358275934,
      "loss": 1.3654,
      "step": 811
    },
    {
      "epoch": 0.15090131945734994,
      "grad_norm": 0.47275739908218384,
      "learning_rate": 0.00019734282066062914,
      "loss": 1.3,
      "step": 812
    },
    {
      "epoch": 0.15108715852072105,
      "grad_norm": 0.43406975269317627,
      "learning_rate": 0.00019733611941255712,
      "loss": 1.2628,
      "step": 813
    },
    {
      "epoch": 0.15127299758409218,
      "grad_norm": 0.5152772665023804,
      "learning_rate": 0.00019732940983911641,
      "loss": 1.2358,
      "step": 814
    },
    {
      "epoch": 0.1514588366474633,
      "grad_norm": 0.4792425036430359,
      "learning_rate": 0.00019732269194088091,
      "loss": 1.1774,
      "step": 815
    },
    {
      "epoch": 0.15164467571083443,
      "grad_norm": 0.46853065490722656,
      "learning_rate": 0.00019731596571842525,
      "loss": 1.3217,
      "step": 816
    },
    {
      "epoch": 0.15183051477420553,
      "grad_norm": 0.394144207239151,
      "learning_rate": 0.00019730923117232472,
      "loss": 0.9721,
      "step": 817
    },
    {
      "epoch": 0.15201635383757667,
      "grad_norm": 0.5005031228065491,
      "learning_rate": 0.0001973024883031553,
      "loss": 1.0645,
      "step": 818
    },
    {
      "epoch": 0.15220219290094777,
      "grad_norm": 0.5808000564575195,
      "learning_rate": 0.0001972957371114938,
      "loss": 0.9737,
      "step": 819
    },
    {
      "epoch": 0.1523880319643189,
      "grad_norm": 0.5996263027191162,
      "learning_rate": 0.0001972889775979176,
      "loss": 1.011,
      "step": 820
    },
    {
      "epoch": 0.15257387102769002,
      "grad_norm": NaN,
      "learning_rate": 0.0001972889775979176,
      "loss": 5.2029,
      "step": 821
    },
    {
      "epoch": 0.15275971009106115,
      "grad_norm": 0.5036181211471558,
      "learning_rate": 0.00019728220976300486,
      "loss": 1.1093,
      "step": 822
    },
    {
      "epoch": 0.15294554915443226,
      "grad_norm": 0.5951871275901794,
      "learning_rate": 0.00019727543360733448,
      "loss": 1.23,
      "step": 823
    },
    {
      "epoch": 0.1531313882178034,
      "grad_norm": 0.5592852830886841,
      "learning_rate": 0.00019726864913148604,
      "loss": 1.2979,
      "step": 824
    },
    {
      "epoch": 0.1533172272811745,
      "grad_norm": 0.4100774824619293,
      "learning_rate": 0.00019726185633603976,
      "loss": 1.0589,
      "step": 825
    },
    {
      "epoch": 0.15350306634454564,
      "grad_norm": 0.5126859545707703,
      "learning_rate": 0.00019725505522157676,
      "loss": 0.9879,
      "step": 826
    },
    {
      "epoch": 0.15368890540791674,
      "grad_norm": 0.49354124069213867,
      "learning_rate": 0.00019724824578867865,
      "loss": 1.0719,
      "step": 827
    },
    {
      "epoch": 0.15387474447128788,
      "grad_norm": 0.39361080527305603,
      "learning_rate": 0.00019724142803792793,
      "loss": 1.0554,
      "step": 828
    },
    {
      "epoch": 0.15406058353465898,
      "grad_norm": 0.48714399337768555,
      "learning_rate": 0.0001972346019699077,
      "loss": 1.1068,
      "step": 829
    },
    {
      "epoch": 0.15424642259803012,
      "grad_norm": 0.6365624666213989,
      "learning_rate": 0.0001972277675852018,
      "loss": 0.7661,
      "step": 830
    },
    {
      "epoch": 0.15443226166140123,
      "grad_norm": 0.4671623706817627,
      "learning_rate": 0.00019722092488439478,
      "loss": 1.1978,
      "step": 831
    },
    {
      "epoch": 0.15461810072477233,
      "grad_norm": 0.4496230483055115,
      "learning_rate": 0.00019721407386807195,
      "loss": 0.978,
      "step": 832
    },
    {
      "epoch": 0.15480393978814347,
      "grad_norm": 0.568945050239563,
      "learning_rate": 0.0001972072145368193,
      "loss": 1.3419,
      "step": 833
    },
    {
      "epoch": 0.15498977885151458,
      "grad_norm": 0.425636351108551,
      "learning_rate": 0.00019720034689122345,
      "loss": 1.1048,
      "step": 834
    },
    {
      "epoch": 0.1551756179148857,
      "grad_norm": 0.6280021071434021,
      "learning_rate": 0.00019719347093187192,
      "loss": 1.289,
      "step": 835
    },
    {
      "epoch": 0.15536145697825682,
      "grad_norm": 0.49712350964546204,
      "learning_rate": 0.0001971865866593527,
      "loss": 1.1054,
      "step": 836
    },
    {
      "epoch": 0.15554729604162795,
      "grad_norm": 0.5043768882751465,
      "learning_rate": 0.00019717969407425468,
      "loss": 0.9071,
      "step": 837
    },
    {
      "epoch": 0.15573313510499906,
      "grad_norm": 0.5186277627944946,
      "learning_rate": 0.0001971727931771674,
      "loss": 1.2653,
      "step": 838
    },
    {
      "epoch": 0.1559189741683702,
      "grad_norm": 0.4321137070655823,
      "learning_rate": 0.0001971658839686811,
      "loss": 1.1057,
      "step": 839
    },
    {
      "epoch": 0.1561048132317413,
      "grad_norm": 64.90776824951172,
      "learning_rate": 0.00019715896644938676,
      "loss": 6.1631,
      "step": 840
    },
    {
      "epoch": 0.15629065229511244,
      "grad_norm": 0.4284045696258545,
      "learning_rate": 0.00019715204061987603,
      "loss": 0.9935,
      "step": 841
    },
    {
      "epoch": 0.15647649135848354,
      "grad_norm": 0.529630184173584,
      "learning_rate": 0.00019714510648074127,
      "loss": 0.8112,
      "step": 842
    },
    {
      "epoch": 0.15666233042185468,
      "grad_norm": 0.5717606544494629,
      "learning_rate": 0.0001971381640325756,
      "loss": 1.2969,
      "step": 843
    },
    {
      "epoch": 0.15684816948522579,
      "grad_norm": 0.4687287211418152,
      "learning_rate": 0.00019713121327597283,
      "loss": 0.9925,
      "step": 844
    },
    {
      "epoch": 0.15703400854859692,
      "grad_norm": 0.5366423726081848,
      "learning_rate": 0.00019712425421152747,
      "loss": 1.419,
      "step": 845
    },
    {
      "epoch": 0.15721984761196803,
      "grad_norm": 0.6493159532546997,
      "learning_rate": 0.0001971172868398347,
      "loss": 1.3085,
      "step": 846
    },
    {
      "epoch": 0.15740568667533916,
      "grad_norm": 0.5041580200195312,
      "learning_rate": 0.00019711031116149052,
      "loss": 1.0164,
      "step": 847
    },
    {
      "epoch": 0.15759152573871027,
      "grad_norm": 0.46504315733909607,
      "learning_rate": 0.00019710332717709153,
      "loss": 1.2274,
      "step": 848
    },
    {
      "epoch": 0.1577773648020814,
      "grad_norm": 0.5412839651107788,
      "learning_rate": 0.0001970963348872351,
      "loss": 0.9564,
      "step": 849
    },
    {
      "epoch": 0.1579632038654525,
      "grad_norm": 0.5719959735870361,
      "learning_rate": 0.00019708933429251927,
      "loss": 1.0234,
      "step": 850
    },
    {
      "epoch": 0.15814904292882365,
      "grad_norm": 0.42526480555534363,
      "learning_rate": 0.00019708232539354287,
      "loss": 1.0373,
      "step": 851
    },
    {
      "epoch": 0.15833488199219475,
      "grad_norm": 0.5555219650268555,
      "learning_rate": 0.00019707530819090537,
      "loss": 1.1596,
      "step": 852
    },
    {
      "epoch": 0.1585207210555659,
      "grad_norm": 0.4331360161304474,
      "learning_rate": 0.00019706828268520693,
      "loss": 1.0561,
      "step": 853
    },
    {
      "epoch": 0.158706560118937,
      "grad_norm": 0.44076040387153625,
      "learning_rate": 0.00019706124887704848,
      "loss": 0.9639,
      "step": 854
    },
    {
      "epoch": 0.15889239918230813,
      "grad_norm": 0.6320684552192688,
      "learning_rate": 0.0001970542067670316,
      "loss": 0.9481,
      "step": 855
    },
    {
      "epoch": 0.15907823824567924,
      "grad_norm": 0.43688324093818665,
      "learning_rate": 0.00019704715635575872,
      "loss": 0.9687,
      "step": 856
    },
    {
      "epoch": 0.15926407730905037,
      "grad_norm": 0.37447649240493774,
      "learning_rate": 0.00019704009764383277,
      "loss": 0.9623,
      "step": 857
    },
    {
      "epoch": 0.15944991637242148,
      "grad_norm": 0.5964031219482422,
      "learning_rate": 0.00019703303063185756,
      "loss": 0.918,
      "step": 858
    },
    {
      "epoch": 0.15963575543579261,
      "grad_norm": 0.502717137336731,
      "learning_rate": 0.00019702595532043754,
      "loss": 1.0252,
      "step": 859
    },
    {
      "epoch": 0.15982159449916372,
      "grad_norm": 0.4272380471229553,
      "learning_rate": 0.0001970188717101778,
      "loss": 0.9805,
      "step": 860
    },
    {
      "epoch": 0.16000743356253486,
      "grad_norm": 0.453313946723938,
      "learning_rate": 0.00019701177980168433,
      "loss": 1.033,
      "step": 861
    },
    {
      "epoch": 0.16019327262590596,
      "grad_norm": 0.5758691430091858,
      "learning_rate": 0.00019700467959556363,
      "loss": 1.182,
      "step": 862
    },
    {
      "epoch": 0.1603791116892771,
      "grad_norm": 0.48800668120384216,
      "learning_rate": 0.00019699757109242305,
      "loss": 0.9889,
      "step": 863
    },
    {
      "epoch": 0.1605649507526482,
      "grad_norm": 0.594584047794342,
      "learning_rate": 0.00019699045429287057,
      "loss": 1.3348,
      "step": 864
    },
    {
      "epoch": 0.16075078981601934,
      "grad_norm": 0.47926321625709534,
      "learning_rate": 0.0001969833291975149,
      "loss": 1.0722,
      "step": 865
    },
    {
      "epoch": 0.16093662887939045,
      "grad_norm": 0.4536885619163513,
      "learning_rate": 0.0001969761958069655,
      "loss": 1.0953,
      "step": 866
    },
    {
      "epoch": 0.16112246794276155,
      "grad_norm": 0.5602397322654724,
      "learning_rate": 0.00019696905412183247,
      "loss": 1.0182,
      "step": 867
    },
    {
      "epoch": 0.1613083070061327,
      "grad_norm": 0.5677642822265625,
      "learning_rate": 0.00019696190414272664,
      "loss": 1.201,
      "step": 868
    },
    {
      "epoch": 0.1614941460695038,
      "grad_norm": 0.5723028779029846,
      "learning_rate": 0.00019695474587025962,
      "loss": 1.4633,
      "step": 869
    },
    {
      "epoch": 0.16167998513287493,
      "grad_norm": 0.4471625089645386,
      "learning_rate": 0.00019694757930504363,
      "loss": 1.0097,
      "step": 870
    },
    {
      "epoch": 0.16186582419624604,
      "grad_norm": 0.4626026153564453,
      "learning_rate": 0.00019694040444769164,
      "loss": 0.94,
      "step": 871
    },
    {
      "epoch": 0.16205166325961717,
      "grad_norm": 0.4059935212135315,
      "learning_rate": 0.00019693322129881734,
      "loss": 0.8778,
      "step": 872
    },
    {
      "epoch": 0.16223750232298828,
      "grad_norm": 0.4863290786743164,
      "learning_rate": 0.00019692602985903513,
      "loss": 1.2015,
      "step": 873
    },
    {
      "epoch": 0.16242334138635942,
      "grad_norm": 0.49636152386665344,
      "learning_rate": 0.00019691883012896012,
      "loss": 1.1677,
      "step": 874
    },
    {
      "epoch": 0.16260918044973052,
      "grad_norm": 0.46879225969314575,
      "learning_rate": 0.00019691162210920806,
      "loss": 1.1934,
      "step": 875
    },
    {
      "epoch": 0.16279501951310166,
      "grad_norm": 0.45269060134887695,
      "learning_rate": 0.00019690440580039554,
      "loss": 0.8255,
      "step": 876
    },
    {
      "epoch": 0.16298085857647276,
      "grad_norm": 0.49863696098327637,
      "learning_rate": 0.00019689718120313973,
      "loss": 1.0585,
      "step": 877
    },
    {
      "epoch": 0.1631666976398439,
      "grad_norm": 0.5355581641197205,
      "learning_rate": 0.0001968899483180586,
      "loss": 1.2448,
      "step": 878
    },
    {
      "epoch": 0.163352536703215,
      "grad_norm": 0.5691109895706177,
      "learning_rate": 0.0001968827071457708,
      "loss": 1.2223,
      "step": 879
    },
    {
      "epoch": 0.16353837576658614,
      "grad_norm": 0.5250231623649597,
      "learning_rate": 0.00019687545768689563,
      "loss": 1.344,
      "step": 880
    },
    {
      "epoch": 0.16372421482995725,
      "grad_norm": 0.47329142689704895,
      "learning_rate": 0.00019686819994205323,
      "loss": 1.115,
      "step": 881
    },
    {
      "epoch": 0.16391005389332838,
      "grad_norm": 0.5414997935295105,
      "learning_rate": 0.00019686093391186427,
      "loss": 1.107,
      "step": 882
    },
    {
      "epoch": 0.1640958929566995,
      "grad_norm": 0.48563218116760254,
      "learning_rate": 0.00019685365959695034,
      "loss": 1.1476,
      "step": 883
    },
    {
      "epoch": 0.16428173202007063,
      "grad_norm": 0.5817862749099731,
      "learning_rate": 0.00019684637699793358,
      "loss": 1.2211,
      "step": 884
    },
    {
      "epoch": 0.16446757108344173,
      "grad_norm": 0.5920995473861694,
      "learning_rate": 0.00019683908611543686,
      "loss": 1.1086,
      "step": 885
    },
    {
      "epoch": 0.16465341014681287,
      "grad_norm": 0.49546167254447937,
      "learning_rate": 0.0001968317869500838,
      "loss": 1.0569,
      "step": 886
    },
    {
      "epoch": 0.16483924921018397,
      "grad_norm": 0.5465673804283142,
      "learning_rate": 0.00019682447950249875,
      "loss": 1.1475,
      "step": 887
    },
    {
      "epoch": 0.1650250882735551,
      "grad_norm": 0.6050189137458801,
      "learning_rate": 0.00019681716377330667,
      "loss": 1.2038,
      "step": 888
    },
    {
      "epoch": 0.16521092733692622,
      "grad_norm": 0.4171687364578247,
      "learning_rate": 0.00019680983976313337,
      "loss": 1.0506,
      "step": 889
    },
    {
      "epoch": 0.16539676640029735,
      "grad_norm": 0.4742315113544464,
      "learning_rate": 0.00019680250747260522,
      "loss": 1.133,
      "step": 890
    },
    {
      "epoch": 0.16558260546366846,
      "grad_norm": 0.4795233905315399,
      "learning_rate": 0.00019679516690234938,
      "loss": 1.1432,
      "step": 891
    },
    {
      "epoch": 0.1657684445270396,
      "grad_norm": 0.5627532601356506,
      "learning_rate": 0.00019678781805299372,
      "loss": 0.88,
      "step": 892
    },
    {
      "epoch": 0.1659542835904107,
      "grad_norm": 0.4793561100959778,
      "learning_rate": 0.00019678046092516676,
      "loss": 1.1651,
      "step": 893
    },
    {
      "epoch": 0.16614012265378184,
      "grad_norm": 0.42398396134376526,
      "learning_rate": 0.00019677309551949785,
      "loss": 1.0936,
      "step": 894
    },
    {
      "epoch": 0.16632596171715294,
      "grad_norm": 0.4386990964412689,
      "learning_rate": 0.00019676572183661693,
      "loss": 0.9839,
      "step": 895
    },
    {
      "epoch": 0.16651180078052408,
      "grad_norm": 0.41191449761390686,
      "learning_rate": 0.00019675833987715463,
      "loss": 0.9892,
      "step": 896
    },
    {
      "epoch": 0.16669763984389518,
      "grad_norm": 0.5318143367767334,
      "learning_rate": 0.00019675094964174245,
      "loss": 0.8916,
      "step": 897
    },
    {
      "epoch": 0.16688347890726632,
      "grad_norm": 0.43514615297317505,
      "learning_rate": 0.00019674355113101242,
      "loss": 1.1414,
      "step": 898
    },
    {
      "epoch": 0.16706931797063743,
      "grad_norm": 0.5335797071456909,
      "learning_rate": 0.00019673614434559739,
      "loss": 1.0711,
      "step": 899
    },
    {
      "epoch": 0.16725515703400856,
      "grad_norm": 0.6108746528625488,
      "learning_rate": 0.00019672872928613085,
      "loss": 1.0356,
      "step": 900
    },
    {
      "epoch": 0.16744099609737967,
      "grad_norm": 0.5601165890693665,
      "learning_rate": 0.00019672130595324702,
      "loss": 1.2728,
      "step": 901
    },
    {
      "epoch": 0.1676268351607508,
      "grad_norm": 0.6302157640457153,
      "learning_rate": 0.00019671387434758085,
      "loss": 1.3459,
      "step": 902
    },
    {
      "epoch": 0.1678126742241219,
      "grad_norm": 0.4896230697631836,
      "learning_rate": 0.000196706434469768,
      "loss": 1.0278,
      "step": 903
    },
    {
      "epoch": 0.16799851328749302,
      "grad_norm": 0.5244489908218384,
      "learning_rate": 0.00019669898632044478,
      "loss": 0.9506,
      "step": 904
    },
    {
      "epoch": 0.16818435235086415,
      "grad_norm": 0.49130362272262573,
      "learning_rate": 0.0001966915299002483,
      "loss": 1.079,
      "step": 905
    },
    {
      "epoch": 0.16837019141423526,
      "grad_norm": 0.4698750972747803,
      "learning_rate": 0.00019668406520981626,
      "loss": 1.1445,
      "step": 906
    },
    {
      "epoch": 0.1685560304776064,
      "grad_norm": 0.5378125309944153,
      "learning_rate": 0.00019667659224978718,
      "loss": 1.0339,
      "step": 907
    },
    {
      "epoch": 0.1687418695409775,
      "grad_norm": 0.5768793225288391,
      "learning_rate": 0.0001966691110208002,
      "loss": 1.3645,
      "step": 908
    },
    {
      "epoch": 0.16892770860434864,
      "grad_norm": 0.49996206164360046,
      "learning_rate": 0.00019666162152349523,
      "loss": 1.2858,
      "step": 909
    },
    {
      "epoch": 0.16911354766771974,
      "grad_norm": 0.4342251420021057,
      "learning_rate": 0.00019665412375851287,
      "loss": 0.8591,
      "step": 910
    },
    {
      "epoch": 0.16929938673109088,
      "grad_norm": 0.4541759192943573,
      "learning_rate": 0.0001966466177264944,
      "loss": 1.2329,
      "step": 911
    },
    {
      "epoch": 0.16948522579446199,
      "grad_norm": 0.48625192046165466,
      "learning_rate": 0.00019663910342808183,
      "loss": 1.0584,
      "step": 912
    },
    {
      "epoch": 0.16967106485783312,
      "grad_norm": 0.4969070553779602,
      "learning_rate": 0.00019663158086391787,
      "loss": 1.1282,
      "step": 913
    },
    {
      "epoch": 0.16985690392120423,
      "grad_norm": 0.4897630214691162,
      "learning_rate": 0.00019662405003464598,
      "loss": 0.8879,
      "step": 914
    },
    {
      "epoch": 0.17004274298457536,
      "grad_norm": 0.5059701800346375,
      "learning_rate": 0.00019661651094091023,
      "loss": 1.141,
      "step": 915
    },
    {
      "epoch": 0.17022858204794647,
      "grad_norm": 0.578790545463562,
      "learning_rate": 0.0001966089635833555,
      "loss": 1.2052,
      "step": 916
    },
    {
      "epoch": 0.1704144211113176,
      "grad_norm": 0.4383024275302887,
      "learning_rate": 0.00019660140796262733,
      "loss": 0.8611,
      "step": 917
    },
    {
      "epoch": 0.1706002601746887,
      "grad_norm": 0.5056631565093994,
      "learning_rate": 0.00019659384407937192,
      "loss": 1.1212,
      "step": 918
    },
    {
      "epoch": 0.17078609923805985,
      "grad_norm": 0.48702114820480347,
      "learning_rate": 0.00019658627193423626,
      "loss": 1.1543,
      "step": 919
    },
    {
      "epoch": 0.17097193830143095,
      "grad_norm": 0.47001418471336365,
      "learning_rate": 0.00019657869152786802,
      "loss": 1.1391,
      "step": 920
    },
    {
      "epoch": 0.1711577773648021,
      "grad_norm": 0.4468410313129425,
      "learning_rate": 0.00019657110286091557,
      "loss": 0.8197,
      "step": 921
    },
    {
      "epoch": 0.1713436164281732,
      "grad_norm": 0.413083553314209,
      "learning_rate": 0.00019656350593402796,
      "loss": 1.0223,
      "step": 922
    },
    {
      "epoch": 0.17152945549154433,
      "grad_norm": 0.533121645450592,
      "learning_rate": 0.00019655590074785502,
      "loss": 0.9284,
      "step": 923
    },
    {
      "epoch": 0.17171529455491544,
      "grad_norm": 0.45572182536125183,
      "learning_rate": 0.0001965482873030472,
      "loss": 1.1174,
      "step": 924
    },
    {
      "epoch": 0.17190113361828657,
      "grad_norm": 0.5130087733268738,
      "learning_rate": 0.00019654066560025567,
      "loss": 1.1895,
      "step": 925
    },
    {
      "epoch": 0.17208697268165768,
      "grad_norm": 0.5442116856575012,
      "learning_rate": 0.00019653303564013235,
      "loss": 1.1295,
      "step": 926
    },
    {
      "epoch": 0.17227281174502881,
      "grad_norm": 0.4951963424682617,
      "learning_rate": 0.00019652539742332986,
      "loss": 1.2979,
      "step": 927
    },
    {
      "epoch": 0.17245865080839992,
      "grad_norm": 0.6164078712463379,
      "learning_rate": 0.00019651775095050155,
      "loss": 1.2801,
      "step": 928
    },
    {
      "epoch": 0.17264448987177106,
      "grad_norm": 0.45666176080703735,
      "learning_rate": 0.00019651009622230134,
      "loss": 1.2134,
      "step": 929
    },
    {
      "epoch": 0.17283032893514216,
      "grad_norm": 0.5629132390022278,
      "learning_rate": 0.00019650243323938408,
      "loss": 1.4352,
      "step": 930
    },
    {
      "epoch": 0.1730161679985133,
      "grad_norm": 0.5350096225738525,
      "learning_rate": 0.0001964947620024051,
      "loss": 1.2231,
      "step": 931
    },
    {
      "epoch": 0.1732020070618844,
      "grad_norm": 0.5338471531867981,
      "learning_rate": 0.00019648708251202054,
      "loss": 1.2145,
      "step": 932
    },
    {
      "epoch": 0.17338784612525554,
      "grad_norm": 0.479168176651001,
      "learning_rate": 0.00019647939476888734,
      "loss": 0.9549,
      "step": 933
    },
    {
      "epoch": 0.17357368518862665,
      "grad_norm": 0.548399031162262,
      "learning_rate": 0.00019647169877366297,
      "loss": 0.9933,
      "step": 934
    },
    {
      "epoch": 0.17375952425199778,
      "grad_norm": 0.4960792362689972,
      "learning_rate": 0.00019646399452700566,
      "loss": 1.2743,
      "step": 935
    },
    {
      "epoch": 0.1739453633153689,
      "grad_norm": 0.4801279902458191,
      "learning_rate": 0.00019645628202957444,
      "loss": 1.1701,
      "step": 936
    },
    {
      "epoch": 0.17413120237874002,
      "grad_norm": 0.47602880001068115,
      "learning_rate": 0.00019644856128202897,
      "loss": 1.0865,
      "step": 937
    },
    {
      "epoch": 0.17431704144211113,
      "grad_norm": 0.5058044195175171,
      "learning_rate": 0.00019644083228502956,
      "loss": 1.2168,
      "step": 938
    },
    {
      "epoch": 0.17450288050548224,
      "grad_norm": 0.46569764614105225,
      "learning_rate": 0.00019643309503923737,
      "loss": 1.126,
      "step": 939
    },
    {
      "epoch": 0.17468871956885337,
      "grad_norm": 0.5684424042701721,
      "learning_rate": 0.0001964253495453141,
      "loss": 0.9557,
      "step": 940
    },
    {
      "epoch": 0.17487455863222448,
      "grad_norm": 0.5311669111251831,
      "learning_rate": 0.0001964175958039223,
      "loss": 0.9657,
      "step": 941
    },
    {
      "epoch": 0.17506039769559562,
      "grad_norm": 2.8396852016448975,
      "learning_rate": 0.00019640983381572516,
      "loss": 3.9658,
      "step": 942
    },
    {
      "epoch": 0.17524623675896672,
      "grad_norm": 0.55197674036026,
      "learning_rate": 0.00019640206358138656,
      "loss": 1.1804,
      "step": 943
    },
    {
      "epoch": 0.17543207582233786,
      "grad_norm": 0.48992422223091125,
      "learning_rate": 0.00019639428510157109,
      "loss": 1.1826,
      "step": 944
    },
    {
      "epoch": 0.17561791488570896,
      "grad_norm": 0.508070170879364,
      "learning_rate": 0.0001963864983769441,
      "loss": 1.0286,
      "step": 945
    },
    {
      "epoch": 0.1758037539490801,
      "grad_norm": 0.5306305289268494,
      "learning_rate": 0.00019637870340817156,
      "loss": 1.2903,
      "step": 946
    },
    {
      "epoch": 0.1759895930124512,
      "grad_norm": 0.5408816933631897,
      "learning_rate": 0.00019637090019592024,
      "loss": 1.1369,
      "step": 947
    },
    {
      "epoch": 0.17617543207582234,
      "grad_norm": 0.45540884137153625,
      "learning_rate": 0.00019636308874085753,
      "loss": 1.0051,
      "step": 948
    },
    {
      "epoch": 0.17636127113919345,
      "grad_norm": 0.4946479797363281,
      "learning_rate": 0.0001963552690436516,
      "loss": 1.306,
      "step": 949
    },
    {
      "epoch": 0.17654711020256458,
      "grad_norm": 0.5659436583518982,
      "learning_rate": 0.00019634744110497122,
      "loss": 1.0725,
      "step": 950
    },
    {
      "epoch": 0.1767329492659357,
      "grad_norm": 0.4694545269012451,
      "learning_rate": 0.00019633960492548597,
      "loss": 1.3804,
      "step": 951
    },
    {
      "epoch": 0.17691878832930683,
      "grad_norm": 0.5113719701766968,
      "learning_rate": 0.0001963317605058661,
      "loss": 1.0854,
      "step": 952
    },
    {
      "epoch": 0.17710462739267793,
      "grad_norm": 0.579812228679657,
      "learning_rate": 0.00019632390784678257,
      "loss": 1.1792,
      "step": 953
    },
    {
      "epoch": 0.17729046645604907,
      "grad_norm": 0.556024968624115,
      "learning_rate": 0.000196316046948907,
      "loss": 1.1594,
      "step": 954
    },
    {
      "epoch": 0.17747630551942017,
      "grad_norm": 0.5733163952827454,
      "learning_rate": 0.00019630817781291177,
      "loss": 1.2513,
      "step": 955
    },
    {
      "epoch": 0.1776621445827913,
      "grad_norm": 0.47930943965911865,
      "learning_rate": 0.00019630030043946993,
      "loss": 1.2074,
      "step": 956
    },
    {
      "epoch": 0.17784798364616242,
      "grad_norm": 0.5812712907791138,
      "learning_rate": 0.00019629241482925532,
      "loss": 0.9204,
      "step": 957
    },
    {
      "epoch": 0.17803382270953355,
      "grad_norm": 0.6513651013374329,
      "learning_rate": 0.0001962845209829423,
      "loss": 1.1605,
      "step": 958
    },
    {
      "epoch": 0.17821966177290466,
      "grad_norm": 0.5434703826904297,
      "learning_rate": 0.0001962766189012061,
      "loss": 1.3342,
      "step": 959
    },
    {
      "epoch": 0.1784055008362758,
      "grad_norm": 0.49916622042655945,
      "learning_rate": 0.00019626870858472262,
      "loss": 0.9762,
      "step": 960
    },
    {
      "epoch": 0.1785913398996469,
      "grad_norm": 0.48941847681999207,
      "learning_rate": 0.00019626079003416845,
      "loss": 0.6947,
      "step": 961
    },
    {
      "epoch": 0.17877717896301804,
      "grad_norm": 0.6125507950782776,
      "learning_rate": 0.00019625286325022082,
      "loss": 1.0488,
      "step": 962
    },
    {
      "epoch": 0.17896301802638914,
      "grad_norm": 0.5615450143814087,
      "learning_rate": 0.0001962449282335578,
      "loss": 1.269,
      "step": 963
    },
    {
      "epoch": 0.17914885708976028,
      "grad_norm": 0.5081421732902527,
      "learning_rate": 0.00019623698498485802,
      "loss": 0.9589,
      "step": 964
    },
    {
      "epoch": 0.17933469615313138,
      "grad_norm": 0.5921791195869446,
      "learning_rate": 0.00019622903350480098,
      "loss": 0.9497,
      "step": 965
    },
    {
      "epoch": 0.17952053521650252,
      "grad_norm": 0.5271924734115601,
      "learning_rate": 0.00019622107379406667,
      "loss": 1.248,
      "step": 966
    },
    {
      "epoch": 0.17970637427987363,
      "grad_norm": 0.5222536325454712,
      "learning_rate": 0.00019621310585333598,
      "loss": 1.1401,
      "step": 967
    },
    {
      "epoch": 0.17989221334324476,
      "grad_norm": 0.46469056606292725,
      "learning_rate": 0.0001962051296832904,
      "loss": 0.9384,
      "step": 968
    },
    {
      "epoch": 0.18007805240661587,
      "grad_norm": 0.495740681886673,
      "learning_rate": 0.00019619714528461212,
      "loss": 1.178,
      "step": 969
    },
    {
      "epoch": 0.180263891469987,
      "grad_norm": 0.4830992817878723,
      "learning_rate": 0.00019618915265798414,
      "loss": 1.3077,
      "step": 970
    },
    {
      "epoch": 0.1804497305333581,
      "grad_norm": 0.5313519835472107,
      "learning_rate": 0.00019618115180409002,
      "loss": 1.2829,
      "step": 971
    },
    {
      "epoch": 0.18063556959672925,
      "grad_norm": 0.45389869809150696,
      "learning_rate": 0.0001961731427236141,
      "loss": 1.064,
      "step": 972
    },
    {
      "epoch": 0.18082140866010035,
      "grad_norm": 0.4235461354255676,
      "learning_rate": 0.00019616512541724147,
      "loss": 0.8205,
      "step": 973
    },
    {
      "epoch": 0.1810072477234715,
      "grad_norm": 0.5951849222183228,
      "learning_rate": 0.0001961570998856578,
      "loss": 1.2295,
      "step": 974
    },
    {
      "epoch": 0.1811930867868426,
      "grad_norm": 0.4912169277667999,
      "learning_rate": 0.00019614906612954952,
      "loss": 1.1792,
      "step": 975
    },
    {
      "epoch": 0.1813789258502137,
      "grad_norm": 0.6600517630577087,
      "learning_rate": 0.00019614102414960386,
      "loss": 1.1413,
      "step": 976
    },
    {
      "epoch": 0.18156476491358484,
      "grad_norm": 0.48177197575569153,
      "learning_rate": 0.00019613297394650858,
      "loss": 1.2551,
      "step": 977
    },
    {
      "epoch": 0.18175060397695594,
      "grad_norm": 0.4557987153530121,
      "learning_rate": 0.00019612491552095228,
      "loss": 1.1687,
      "step": 978
    },
    {
      "epoch": 0.18193644304032708,
      "grad_norm": 0.47915035486221313,
      "learning_rate": 0.0001961168488736242,
      "loss": 1.0173,
      "step": 979
    },
    {
      "epoch": 0.18212228210369819,
      "grad_norm": 0.7031456828117371,
      "learning_rate": 0.00019610877400521432,
      "loss": 1.1247,
      "step": 980
    },
    {
      "epoch": 0.18230812116706932,
      "grad_norm": 0.4912133812904358,
      "learning_rate": 0.00019610069091641324,
      "loss": 0.8791,
      "step": 981
    },
    {
      "epoch": 0.18249396023044043,
      "grad_norm": 0.5140329003334045,
      "learning_rate": 0.00019609259960791243,
      "loss": 0.8195,
      "step": 982
    },
    {
      "epoch": 0.18267979929381156,
      "grad_norm": 0.4770338237285614,
      "learning_rate": 0.00019608450008040383,
      "loss": 1.2822,
      "step": 983
    },
    {
      "epoch": 0.18286563835718267,
      "grad_norm": 0.4585348665714264,
      "learning_rate": 0.0001960763923345803,
      "loss": 1.0453,
      "step": 984
    },
    {
      "epoch": 0.1830514774205538,
      "grad_norm": 0.5334509611129761,
      "learning_rate": 0.00019606827637113527,
      "loss": 1.1493,
      "step": 985
    },
    {
      "epoch": 0.1832373164839249,
      "grad_norm": 0.5813632607460022,
      "learning_rate": 0.0001960601521907629,
      "loss": 1.2191,
      "step": 986
    },
    {
      "epoch": 0.18342315554729605,
      "grad_norm": 0.5605377554893494,
      "learning_rate": 0.00019605201979415817,
      "loss": 0.9412,
      "step": 987
    },
    {
      "epoch": 0.18360899461066715,
      "grad_norm": 0.5268937945365906,
      "learning_rate": 0.00019604387918201657,
      "loss": 1.0694,
      "step": 988
    },
    {
      "epoch": 0.1837948336740383,
      "grad_norm": 0.5224151611328125,
      "learning_rate": 0.00019603573035503437,
      "loss": 1.016,
      "step": 989
    },
    {
      "epoch": 0.1839806727374094,
      "grad_norm": 0.45612266659736633,
      "learning_rate": 0.00019602757331390865,
      "loss": 1.1016,
      "step": 990
    },
    {
      "epoch": 0.18416651180078053,
      "grad_norm": 0.49727365374565125,
      "learning_rate": 0.000196019408059337,
      "loss": 0.9785,
      "step": 991
    },
    {
      "epoch": 0.18435235086415164,
      "grad_norm": 0.5259670615196228,
      "learning_rate": 0.00019601123459201788,
      "loss": 1.124,
      "step": 992
    },
    {
      "epoch": 0.18453818992752277,
      "grad_norm": 0.4743773937225342,
      "learning_rate": 0.00019600305291265032,
      "loss": 1.1237,
      "step": 993
    },
    {
      "epoch": 0.18472402899089388,
      "grad_norm": 0.5588434934616089,
      "learning_rate": 0.00019599486302193417,
      "loss": 1.2736,
      "step": 994
    },
    {
      "epoch": 0.18490986805426501,
      "grad_norm": 0.5143772959709167,
      "learning_rate": 0.00019598666492056993,
      "loss": 1.0648,
      "step": 995
    },
    {
      "epoch": 0.18509570711763612,
      "grad_norm": 0.5187076330184937,
      "learning_rate": 0.00019597845860925878,
      "loss": 0.8345,
      "step": 996
    },
    {
      "epoch": 0.18528154618100726,
      "grad_norm": 0.5211664438247681,
      "learning_rate": 0.0001959702440887026,
      "loss": 1.0942,
      "step": 997
    },
    {
      "epoch": 0.18546738524437836,
      "grad_norm": 0.4638168513774872,
      "learning_rate": 0.00019596202135960407,
      "loss": 1.1404,
      "step": 998
    },
    {
      "epoch": 0.1856532243077495,
      "grad_norm": 0.4244816303253174,
      "learning_rate": 0.0001959537904226664,
      "loss": 1.1829,
      "step": 999
    },
    {
      "epoch": 0.1858390633711206,
      "grad_norm": 0.4680684208869934,
      "learning_rate": 0.0001959455512785937,
      "loss": 1.1177,
      "step": 1000
    },
    {
      "epoch": 0.18602490243449174,
      "grad_norm": 0.5658469200134277,
      "learning_rate": 0.0001959373039280906,
      "loss": 1.4114,
      "step": 1001
    },
    {
      "epoch": 0.18621074149786285,
      "grad_norm": 0.5405275225639343,
      "learning_rate": 0.00019592904837186257,
      "loss": 1.0142,
      "step": 1002
    },
    {
      "epoch": 0.18639658056123398,
      "grad_norm": 0.5917633771896362,
      "learning_rate": 0.0001959207846106157,
      "loss": 1.3173,
      "step": 1003
    },
    {
      "epoch": 0.1865824196246051,
      "grad_norm": 0.5246212482452393,
      "learning_rate": 0.00019591251264505682,
      "loss": 1.0017,
      "step": 1004
    },
    {
      "epoch": 0.18676825868797622,
      "grad_norm": 0.5147131085395813,
      "learning_rate": 0.0001959042324758934,
      "loss": 1.2113,
      "step": 1005
    },
    {
      "epoch": 0.18695409775134733,
      "grad_norm": 0.4775300920009613,
      "learning_rate": 0.0001958959441038337,
      "loss": 1.072,
      "step": 1006
    },
    {
      "epoch": 0.18713993681471847,
      "grad_norm": 0.4175073504447937,
      "learning_rate": 0.00019588764752958668,
      "loss": 1.0136,
      "step": 1007
    },
    {
      "epoch": 0.18732577587808957,
      "grad_norm": 0.5542589426040649,
      "learning_rate": 0.00019587934275386193,
      "loss": 1.3874,
      "step": 1008
    },
    {
      "epoch": 0.1875116149414607,
      "grad_norm": 0.44940391182899475,
      "learning_rate": 0.00019587102977736972,
      "loss": 1.0096,
      "step": 1009
    },
    {
      "epoch": 0.18769745400483182,
      "grad_norm": 0.6043052077293396,
      "learning_rate": 0.00019586270860082118,
      "loss": 1.2021,
      "step": 1010
    },
    {
      "epoch": 0.18788329306820295,
      "grad_norm": 0.5352486371994019,
      "learning_rate": 0.000195854379224928,
      "loss": 1.1452,
      "step": 1011
    },
    {
      "epoch": 0.18806913213157406,
      "grad_norm": 0.4326801896095276,
      "learning_rate": 0.00019584604165040254,
      "loss": 0.8873,
      "step": 1012
    },
    {
      "epoch": 0.18825497119494516,
      "grad_norm": 0.5577175617218018,
      "learning_rate": 0.00019583769587795802,
      "loss": 1.2128,
      "step": 1013
    },
    {
      "epoch": 0.1884408102583163,
      "grad_norm": 0.4985814392566681,
      "learning_rate": 0.0001958293419083082,
      "loss": 1.3347,
      "step": 1014
    },
    {
      "epoch": 0.1886266493216874,
      "grad_norm": 0.5640441179275513,
      "learning_rate": 0.00019582097974216772,
      "loss": 1.0509,
      "step": 1015
    },
    {
      "epoch": 0.18881248838505854,
      "grad_norm": 0.48311781883239746,
      "learning_rate": 0.0001958126093802517,
      "loss": 1.1147,
      "step": 1016
    },
    {
      "epoch": 0.18899832744842965,
      "grad_norm": 0.5454368591308594,
      "learning_rate": 0.00019580423082327613,
      "loss": 1.1961,
      "step": 1017
    },
    {
      "epoch": 0.18918416651180078,
      "grad_norm": 0.49283385276794434,
      "learning_rate": 0.00019579584407195764,
      "loss": 1.1866,
      "step": 1018
    },
    {
      "epoch": 0.1893700055751719,
      "grad_norm": 0.6507675647735596,
      "learning_rate": 0.00019578744912701358,
      "loss": 1.2556,
      "step": 1019
    },
    {
      "epoch": 0.18955584463854303,
      "grad_norm": 0.5064658522605896,
      "learning_rate": 0.00019577904598916192,
      "loss": 1.0026,
      "step": 1020
    },
    {
      "epoch": 0.18974168370191413,
      "grad_norm": 0.76408451795578,
      "learning_rate": 0.00019577063465912149,
      "loss": 1.2837,
      "step": 1021
    },
    {
      "epoch": 0.18992752276528527,
      "grad_norm": 0.6197009086608887,
      "learning_rate": 0.00019576221513761168,
      "loss": 1.3769,
      "step": 1022
    },
    {
      "epoch": 0.19011336182865637,
      "grad_norm": 0.5020315051078796,
      "learning_rate": 0.00019575378742535266,
      "loss": 1.1076,
      "step": 1023
    },
    {
      "epoch": 0.1902992008920275,
      "grad_norm": 0.541795015335083,
      "learning_rate": 0.00019574535152306523,
      "loss": 1.1054,
      "step": 1024
    },
    {
      "epoch": 0.19048503995539862,
      "grad_norm": 0.49482569098472595,
      "learning_rate": 0.00019573690743147096,
      "loss": 1.2124,
      "step": 1025
    },
    {
      "epoch": 0.19067087901876975,
      "grad_norm": 0.5648143887519836,
      "learning_rate": 0.00019572845515129208,
      "loss": 1.2169,
      "step": 1026
    },
    {
      "epoch": 0.19085671808214086,
      "grad_norm": 0.4914950728416443,
      "learning_rate": 0.00019571999468325152,
      "loss": 1.0648,
      "step": 1027
    },
    {
      "epoch": 0.191042557145512,
      "grad_norm": 0.5238738656044006,
      "learning_rate": 0.00019571152602807296,
      "loss": 1.0846,
      "step": 1028
    },
    {
      "epoch": 0.1912283962088831,
      "grad_norm": 0.4733150005340576,
      "learning_rate": 0.00019570304918648071,
      "loss": 1.0583,
      "step": 1029
    },
    {
      "epoch": 0.19141423527225424,
      "grad_norm": 0.5372300148010254,
      "learning_rate": 0.00019569456415919985,
      "loss": 1.1714,
      "step": 1030
    },
    {
      "epoch": 0.19160007433562534,
      "grad_norm": 0.5832946300506592,
      "learning_rate": 0.00019568607094695603,
      "loss": 1.3292,
      "step": 1031
    },
    {
      "epoch": 0.19178591339899648,
      "grad_norm": 0.5379717350006104,
      "learning_rate": 0.00019567756955047582,
      "loss": 1.2471,
      "step": 1032
    },
    {
      "epoch": 0.19197175246236758,
      "grad_norm": 0.5493462681770325,
      "learning_rate": 0.00019566905997048627,
      "loss": 1.0719,
      "step": 1033
    },
    {
      "epoch": 0.19215759152573872,
      "grad_norm": 0.4469130337238312,
      "learning_rate": 0.00019566054220771526,
      "loss": 1.2158,
      "step": 1034
    },
    {
      "epoch": 0.19234343058910983,
      "grad_norm": 0.4941439628601074,
      "learning_rate": 0.0001956520162628913,
      "loss": 1.0047,
      "step": 1035
    },
    {
      "epoch": 0.19252926965248096,
      "grad_norm": 0.5384143590927124,
      "learning_rate": 0.0001956434821367437,
      "loss": 1.1704,
      "step": 1036
    },
    {
      "epoch": 0.19271510871585207,
      "grad_norm": 0.4670869708061218,
      "learning_rate": 0.00019563493983000232,
      "loss": 0.961,
      "step": 1037
    },
    {
      "epoch": 0.1929009477792232,
      "grad_norm": 0.43627220392227173,
      "learning_rate": 0.00019562638934339787,
      "loss": 1.0979,
      "step": 1038
    },
    {
      "epoch": 0.1930867868425943,
      "grad_norm": 0.5042278170585632,
      "learning_rate": 0.00019561783067766167,
      "loss": 0.9918,
      "step": 1039
    },
    {
      "epoch": 0.19327262590596545,
      "grad_norm": 0.46678847074508667,
      "learning_rate": 0.0001956092638335257,
      "loss": 1.0757,
      "step": 1040
    },
    {
      "epoch": 0.19345846496933655,
      "grad_norm": 0.9079703092575073,
      "learning_rate": 0.0001956006888117228,
      "loss": 0.994,
      "step": 1041
    },
    {
      "epoch": 0.1936443040327077,
      "grad_norm": 0.4500991404056549,
      "learning_rate": 0.00019559210561298637,
      "loss": 1.1286,
      "step": 1042
    },
    {
      "epoch": 0.1938301430960788,
      "grad_norm": 0.49819493293762207,
      "learning_rate": 0.00019558351423805054,
      "loss": 1.0163,
      "step": 1043
    },
    {
      "epoch": 0.19401598215944993,
      "grad_norm": 0.507444441318512,
      "learning_rate": 0.00019557491468765017,
      "loss": 1.027,
      "step": 1044
    },
    {
      "epoch": 0.19420182122282104,
      "grad_norm": 0.5294500589370728,
      "learning_rate": 0.00019556630696252074,
      "loss": 0.9572,
      "step": 1045
    },
    {
      "epoch": 0.19438766028619217,
      "grad_norm": 0.5559682250022888,
      "learning_rate": 0.00019555769106339857,
      "loss": 1.1363,
      "step": 1046
    },
    {
      "epoch": 0.19457349934956328,
      "grad_norm": 0.5073032379150391,
      "learning_rate": 0.00019554906699102053,
      "loss": 1.1022,
      "step": 1047
    },
    {
      "epoch": 0.19475933841293439,
      "grad_norm": 0.4607917368412018,
      "learning_rate": 0.0001955404347461243,
      "loss": 1.0064,
      "step": 1048
    },
    {
      "epoch": 0.19494517747630552,
      "grad_norm": 0.4846749007701874,
      "learning_rate": 0.00019553179432944817,
      "loss": 1.1871,
      "step": 1049
    },
    {
      "epoch": 0.19513101653967663,
      "grad_norm": 0.7353876829147339,
      "learning_rate": 0.0001955231457417312,
      "loss": 1.168,
      "step": 1050
    },
    {
      "epoch": 0.19531685560304776,
      "grad_norm": 0.5777313709259033,
      "learning_rate": 0.00019551448898371315,
      "loss": 1.0699,
      "step": 1051
    },
    {
      "epoch": 0.19550269466641887,
      "grad_norm": 0.5727697014808655,
      "learning_rate": 0.0001955058240561344,
      "loss": 1.0917,
      "step": 1052
    },
    {
      "epoch": 0.19568853372979,
      "grad_norm": 0.4747198522090912,
      "learning_rate": 0.00019549715095973613,
      "loss": 1.1084,
      "step": 1053
    },
    {
      "epoch": 0.1958743727931611,
      "grad_norm": 0.46111252903938293,
      "learning_rate": 0.00019548846969526013,
      "loss": 0.9441,
      "step": 1054
    },
    {
      "epoch": 0.19606021185653225,
      "grad_norm": 0.5396764874458313,
      "learning_rate": 0.00019547978026344893,
      "loss": 1.2142,
      "step": 1055
    },
    {
      "epoch": 0.19624605091990335,
      "grad_norm": 0.6169037222862244,
      "learning_rate": 0.00019547108266504577,
      "loss": 1.2052,
      "step": 1056
    },
    {
      "epoch": 0.1964318899832745,
      "grad_norm": 0.5420907139778137,
      "learning_rate": 0.00019546237690079457,
      "loss": 1.1335,
      "step": 1057
    },
    {
      "epoch": 0.1966177290466456,
      "grad_norm": 0.5372158288955688,
      "learning_rate": 0.00019545366297143995,
      "loss": 0.8468,
      "step": 1058
    },
    {
      "epoch": 0.19680356811001673,
      "grad_norm": 0.5079149603843689,
      "learning_rate": 0.00019544494087772723,
      "loss": 1.1215,
      "step": 1059
    },
    {
      "epoch": 0.19698940717338784,
      "grad_norm": 0.49826011061668396,
      "learning_rate": 0.00019543621062040244,
      "loss": 0.9791,
      "step": 1060
    },
    {
      "epoch": 0.19717524623675897,
      "grad_norm": 0.49958309531211853,
      "learning_rate": 0.00019542747220021228,
      "loss": 1.0473,
      "step": 1061
    },
    {
      "epoch": 0.19736108530013008,
      "grad_norm": 0.5966783165931702,
      "learning_rate": 0.0001954187256179042,
      "loss": 1.355,
      "step": 1062
    },
    {
      "epoch": 0.19754692436350121,
      "grad_norm": 0.5301747918128967,
      "learning_rate": 0.00019540997087422627,
      "loss": 1.1467,
      "step": 1063
    },
    {
      "epoch": 0.19773276342687232,
      "grad_norm": 0.4955202341079712,
      "learning_rate": 0.00019540120796992733,
      "loss": 1.2297,
      "step": 1064
    },
    {
      "epoch": 0.19791860249024346,
      "grad_norm": 0.4707731008529663,
      "learning_rate": 0.00019539243690575687,
      "loss": 0.9205,
      "step": 1065
    },
    {
      "epoch": 0.19810444155361456,
      "grad_norm": 0.6613312363624573,
      "learning_rate": 0.00019538365768246517,
      "loss": 0.9754,
      "step": 1066
    },
    {
      "epoch": 0.1982902806169857,
      "grad_norm": 0.47768664360046387,
      "learning_rate": 0.00019537487030080303,
      "loss": 1.0455,
      "step": 1067
    },
    {
      "epoch": 0.1984761196803568,
      "grad_norm": 0.6217352151870728,
      "learning_rate": 0.00019536607476152207,
      "loss": 1.4062,
      "step": 1068
    },
    {
      "epoch": 0.19866195874372794,
      "grad_norm": 0.6218780279159546,
      "learning_rate": 0.00019535727106537463,
      "loss": 1.2606,
      "step": 1069
    },
    {
      "epoch": 0.19884779780709905,
      "grad_norm": 0.6484935283660889,
      "learning_rate": 0.00019534845921311374,
      "loss": 1.319,
      "step": 1070
    },
    {
      "epoch": 0.19903363687047018,
      "grad_norm": 0.493118554353714,
      "learning_rate": 0.00019533963920549306,
      "loss": 1.218,
      "step": 1071
    },
    {
      "epoch": 0.1992194759338413,
      "grad_norm": 0.6731880903244019,
      "learning_rate": 0.00019533081104326696,
      "loss": 1.1587,
      "step": 1072
    },
    {
      "epoch": 0.19940531499721242,
      "grad_norm": 0.7866933941841125,
      "learning_rate": 0.00019532197472719057,
      "loss": 1.5491,
      "step": 1073
    },
    {
      "epoch": 0.19959115406058353,
      "grad_norm": 0.6097940802574158,
      "learning_rate": 0.00019531313025801964,
      "loss": 1.1939,
      "step": 1074
    },
    {
      "epoch": 0.19977699312395467,
      "grad_norm": 0.5222741365432739,
      "learning_rate": 0.00019530427763651072,
      "loss": 1.2106,
      "step": 1075
    },
    {
      "epoch": 0.19996283218732577,
      "grad_norm": 0.5517898797988892,
      "learning_rate": 0.00019529541686342093,
      "loss": 0.8951,
      "step": 1076
    },
    {
      "epoch": 0.2001486712506969,
      "grad_norm": 0.5374531745910645,
      "learning_rate": 0.00019528654793950817,
      "loss": 1.0006,
      "step": 1077
    },
    {
      "epoch": 0.20033451031406802,
      "grad_norm": 0.572433590888977,
      "learning_rate": 0.000195277670865531,
      "loss": 1.1501,
      "step": 1078
    },
    {
      "epoch": 0.20052034937743915,
      "grad_norm": 0.4874914586544037,
      "learning_rate": 0.00019526878564224878,
      "loss": 0.9626,
      "step": 1079
    },
    {
      "epoch": 0.20070618844081026,
      "grad_norm": 0.520273745059967,
      "learning_rate": 0.00019525989227042138,
      "loss": 0.9302,
      "step": 1080
    },
    {
      "epoch": 0.2008920275041814,
      "grad_norm": 0.45634725689888,
      "learning_rate": 0.00019525099075080952,
      "loss": 1.1159,
      "step": 1081
    },
    {
      "epoch": 0.2010778665675525,
      "grad_norm": 0.49319830536842346,
      "learning_rate": 0.00019524208108417454,
      "loss": 0.8992,
      "step": 1082
    },
    {
      "epoch": 0.20126370563092363,
      "grad_norm": 0.4766017496585846,
      "learning_rate": 0.00019523316327127853,
      "loss": 1.2904,
      "step": 1083
    },
    {
      "epoch": 0.20144954469429474,
      "grad_norm": 0.5087282061576843,
      "learning_rate": 0.00019522423731288427,
      "loss": 1.1372,
      "step": 1084
    },
    {
      "epoch": 0.20163538375766585,
      "grad_norm": 0.591387927532196,
      "learning_rate": 0.00019521530320975518,
      "loss": 1.5339,
      "step": 1085
    },
    {
      "epoch": 0.20182122282103698,
      "grad_norm": 0.6128246784210205,
      "learning_rate": 0.00019520636096265537,
      "loss": 1.1951,
      "step": 1086
    },
    {
      "epoch": 0.2020070618844081,
      "grad_norm": 0.46142295002937317,
      "learning_rate": 0.00019519741057234978,
      "loss": 1.2229,
      "step": 1087
    },
    {
      "epoch": 0.20219290094777922,
      "grad_norm": 0.537652313709259,
      "learning_rate": 0.00019518845203960388,
      "loss": 1.1164,
      "step": 1088
    },
    {
      "epoch": 0.20237874001115033,
      "grad_norm": 0.6652979850769043,
      "learning_rate": 0.000195179485365184,
      "loss": 1.2889,
      "step": 1089
    },
    {
      "epoch": 0.20256457907452147,
      "grad_norm": 0.6123540997505188,
      "learning_rate": 0.000195170510549857,
      "loss": 1.1483,
      "step": 1090
    },
    {
      "epoch": 0.20275041813789257,
      "grad_norm": 0.46437254548072815,
      "learning_rate": 0.00019516152759439055,
      "loss": 1.0709,
      "step": 1091
    },
    {
      "epoch": 0.2029362572012637,
      "grad_norm": 0.5575506091117859,
      "learning_rate": 0.00019515253649955296,
      "loss": 1.1623,
      "step": 1092
    },
    {
      "epoch": 0.20312209626463482,
      "grad_norm": 0.4591430127620697,
      "learning_rate": 0.00019514353726611327,
      "loss": 1.0653,
      "step": 1093
    },
    {
      "epoch": 0.20330793532800595,
      "grad_norm": 0.5363798141479492,
      "learning_rate": 0.0001951345298948412,
      "loss": 1.1025,
      "step": 1094
    },
    {
      "epoch": 0.20349377439137706,
      "grad_norm": 0.5441762208938599,
      "learning_rate": 0.0001951255143865072,
      "loss": 1.3256,
      "step": 1095
    },
    {
      "epoch": 0.2036796134547482,
      "grad_norm": 0.5097872614860535,
      "learning_rate": 0.00019511649074188233,
      "loss": 1.1426,
      "step": 1096
    },
    {
      "epoch": 0.2038654525181193,
      "grad_norm": 0.4579317271709442,
      "learning_rate": 0.00019510745896173846,
      "loss": 0.9945,
      "step": 1097
    },
    {
      "epoch": 0.20405129158149043,
      "grad_norm": 0.49984660744667053,
      "learning_rate": 0.00019509841904684808,
      "loss": 0.9922,
      "step": 1098
    },
    {
      "epoch": 0.20423713064486154,
      "grad_norm": 0.5130319595336914,
      "learning_rate": 0.00019508937099798436,
      "loss": 1.0339,
      "step": 1099
    },
    {
      "epoch": 0.20442296970823268,
      "grad_norm": 0.6417672038078308,
      "learning_rate": 0.00019508031481592123,
      "loss": 0.9417,
      "step": 1100
    },
    {
      "epoch": 0.20460880877160378,
      "grad_norm": 0.4974082410335541,
      "learning_rate": 0.00019507125050143324,
      "loss": 1.0728,
      "step": 1101
    },
    {
      "epoch": 0.20479464783497492,
      "grad_norm": 0.5567024350166321,
      "learning_rate": 0.0001950621780552958,
      "loss": 1.2673,
      "step": 1102
    },
    {
      "epoch": 0.20498048689834603,
      "grad_norm": 0.5510198473930359,
      "learning_rate": 0.00019505309747828474,
      "loss": 1.1641,
      "step": 1103
    },
    {
      "epoch": 0.20516632596171716,
      "grad_norm": 0.5027584433555603,
      "learning_rate": 0.00019504400877117686,
      "loss": 0.9991,
      "step": 1104
    },
    {
      "epoch": 0.20535216502508827,
      "grad_norm": 0.5015250444412231,
      "learning_rate": 0.00019503491193474947,
      "loss": 1.1148,
      "step": 1105
    },
    {
      "epoch": 0.2055380040884594,
      "grad_norm": 0.5679439306259155,
      "learning_rate": 0.00019502580696978067,
      "loss": 0.9935,
      "step": 1106
    },
    {
      "epoch": 0.2057238431518305,
      "grad_norm": 0.4760668873786926,
      "learning_rate": 0.00019501669387704925,
      "loss": 0.9816,
      "step": 1107
    },
    {
      "epoch": 0.20590968221520164,
      "grad_norm": 0.5136030912399292,
      "learning_rate": 0.0001950075726573346,
      "loss": 1.0082,
      "step": 1108
    },
    {
      "epoch": 0.20609552127857275,
      "grad_norm": 0.5538644790649414,
      "learning_rate": 0.00019499844331141695,
      "loss": 1.1304,
      "step": 1109
    },
    {
      "epoch": 0.2062813603419439,
      "grad_norm": 0.5204370021820068,
      "learning_rate": 0.0001949893058400771,
      "loss": 1.2725,
      "step": 1110
    },
    {
      "epoch": 0.206467199405315,
      "grad_norm": 0.7252811789512634,
      "learning_rate": 0.00019498016024409664,
      "loss": 1.1573,
      "step": 1111
    },
    {
      "epoch": 0.20665303846868613,
      "grad_norm": 0.5194401144981384,
      "learning_rate": 0.0001949710065242578,
      "loss": 1.2174,
      "step": 1112
    },
    {
      "epoch": 0.20683887753205724,
      "grad_norm": 0.5207048654556274,
      "learning_rate": 0.0001949618446813435,
      "loss": 1.2053,
      "step": 1113
    },
    {
      "epoch": 0.20702471659542837,
      "grad_norm": 0.5159600973129272,
      "learning_rate": 0.0001949526747161374,
      "loss": 1.1801,
      "step": 1114
    },
    {
      "epoch": 0.20721055565879948,
      "grad_norm": 0.5342578291893005,
      "learning_rate": 0.0001949434966294238,
      "loss": 1.0367,
      "step": 1115
    },
    {
      "epoch": 0.2073963947221706,
      "grad_norm": 0.5836411714553833,
      "learning_rate": 0.00019493431042198774,
      "loss": 1.1718,
      "step": 1116
    },
    {
      "epoch": 0.20758223378554172,
      "grad_norm": 0.6905984878540039,
      "learning_rate": 0.00019492511609461491,
      "loss": 1.1931,
      "step": 1117
    },
    {
      "epoch": 0.20776807284891285,
      "grad_norm": 0.5296867489814758,
      "learning_rate": 0.0001949159136480918,
      "loss": 1.1577,
      "step": 1118
    },
    {
      "epoch": 0.20795391191228396,
      "grad_norm": 0.5768712162971497,
      "learning_rate": 0.0001949067030832054,
      "loss": 1.1849,
      "step": 1119
    },
    {
      "epoch": 0.20813975097565507,
      "grad_norm": 0.59405118227005,
      "learning_rate": 0.00019489748440074358,
      "loss": 1.0622,
      "step": 1120
    },
    {
      "epoch": 0.2083255900390262,
      "grad_norm": 0.4969901740550995,
      "learning_rate": 0.00019488825760149482,
      "loss": 1.0519,
      "step": 1121
    },
    {
      "epoch": 0.2085114291023973,
      "grad_norm": 0.4819211959838867,
      "learning_rate": 0.0001948790226862483,
      "loss": 0.7888,
      "step": 1122
    },
    {
      "epoch": 0.20869726816576845,
      "grad_norm": 0.5228898525238037,
      "learning_rate": 0.00019486977965579389,
      "loss": 0.9494,
      "step": 1123
    },
    {
      "epoch": 0.20888310722913955,
      "grad_norm": 0.4955415725708008,
      "learning_rate": 0.00019486052851092222,
      "loss": 0.9514,
      "step": 1124
    },
    {
      "epoch": 0.2090689462925107,
      "grad_norm": 0.5162959098815918,
      "learning_rate": 0.0001948512692524245,
      "loss": 1.1837,
      "step": 1125
    },
    {
      "epoch": 0.2092547853558818,
      "grad_norm": 0.4941028952598572,
      "learning_rate": 0.00019484200188109275,
      "loss": 1.1001,
      "step": 1126
    },
    {
      "epoch": 0.20944062441925293,
      "grad_norm": 0.5514204502105713,
      "learning_rate": 0.0001948327263977196,
      "loss": 1.3927,
      "step": 1127
    },
    {
      "epoch": 0.20962646348262404,
      "grad_norm": 0.5498420596122742,
      "learning_rate": 0.00019482344280309836,
      "loss": 1.3795,
      "step": 1128
    },
    {
      "epoch": 0.20981230254599517,
      "grad_norm": 0.5233075022697449,
      "learning_rate": 0.00019481415109802314,
      "loss": 1.2404,
      "step": 1129
    },
    {
      "epoch": 0.20999814160936628,
      "grad_norm": 0.5122292637825012,
      "learning_rate": 0.00019480485128328868,
      "loss": 1.1559,
      "step": 1130
    },
    {
      "epoch": 0.2101839806727374,
      "grad_norm": 0.5212186574935913,
      "learning_rate": 0.00019479554335969038,
      "loss": 1.1201,
      "step": 1131
    },
    {
      "epoch": 0.21036981973610852,
      "grad_norm": 0.5237042903900146,
      "learning_rate": 0.0001947862273280243,
      "loss": 0.995,
      "step": 1132
    },
    {
      "epoch": 0.21055565879947966,
      "grad_norm": 0.5185428857803345,
      "learning_rate": 0.00019477690318908742,
      "loss": 0.9331,
      "step": 1133
    },
    {
      "epoch": 0.21074149786285076,
      "grad_norm": 0.4496161937713623,
      "learning_rate": 0.00019476757094367714,
      "loss": 1.0886,
      "step": 1134
    },
    {
      "epoch": 0.2109273369262219,
      "grad_norm": 0.5477585792541504,
      "learning_rate": 0.00019475823059259174,
      "loss": 1.0802,
      "step": 1135
    },
    {
      "epoch": 0.211113175989593,
      "grad_norm": 0.6424799561500549,
      "learning_rate": 0.00019474888213663003,
      "loss": 1.0755,
      "step": 1136
    },
    {
      "epoch": 0.21129901505296414,
      "grad_norm": 0.47722670435905457,
      "learning_rate": 0.00019473952557659167,
      "loss": 1.1139,
      "step": 1137
    },
    {
      "epoch": 0.21148485411633525,
      "grad_norm": 0.5191829800605774,
      "learning_rate": 0.00019473016091327693,
      "loss": 0.978,
      "step": 1138
    },
    {
      "epoch": 0.21167069317970638,
      "grad_norm": 0.4725750684738159,
      "learning_rate": 0.0001947207881474868,
      "loss": 0.9026,
      "step": 1139
    },
    {
      "epoch": 0.2118565322430775,
      "grad_norm": 0.5386378169059753,
      "learning_rate": 0.0001947114072800229,
      "loss": 1.3181,
      "step": 1140
    },
    {
      "epoch": 0.21204237130644862,
      "grad_norm": 0.5343483686447144,
      "learning_rate": 0.00019470201831168766,
      "loss": 1.1293,
      "step": 1141
    },
    {
      "epoch": 0.21222821036981973,
      "grad_norm": 0.6510689854621887,
      "learning_rate": 0.0001946926212432841,
      "loss": 1.031,
      "step": 1142
    },
    {
      "epoch": 0.21241404943319087,
      "grad_norm": 0.5814812779426575,
      "learning_rate": 0.00019468321607561602,
      "loss": 1.0944,
      "step": 1143
    },
    {
      "epoch": 0.21259988849656197,
      "grad_norm": 0.4645356237888336,
      "learning_rate": 0.00019467380280948783,
      "loss": 1.0804,
      "step": 1144
    },
    {
      "epoch": 0.2127857275599331,
      "grad_norm": 2.570280075073242,
      "learning_rate": 0.00019466438144570466,
      "loss": 3.3341,
      "step": 1145
    },
    {
      "epoch": 0.21297156662330421,
      "grad_norm": 0.5189799070358276,
      "learning_rate": 0.00019465495198507234,
      "loss": 1.2346,
      "step": 1146
    },
    {
      "epoch": 0.21315740568667535,
      "grad_norm": 0.5512979626655579,
      "learning_rate": 0.0001946455144283974,
      "loss": 1.179,
      "step": 1147
    },
    {
      "epoch": 0.21334324475004646,
      "grad_norm": 0.555311381816864,
      "learning_rate": 0.00019463606877648704,
      "loss": 0.9932,
      "step": 1148
    },
    {
      "epoch": 0.2135290838134176,
      "grad_norm": 0.5090897679328918,
      "learning_rate": 0.0001946266150301492,
      "loss": 1.049,
      "step": 1149
    },
    {
      "epoch": 0.2137149228767887,
      "grad_norm": 0.537129819393158,
      "learning_rate": 0.00019461715319019248,
      "loss": 1.2181,
      "step": 1150
    },
    {
      "epoch": 0.21390076194015983,
      "grad_norm": 0.567992091178894,
      "learning_rate": 0.00019460768325742613,
      "loss": 1.0256,
      "step": 1151
    },
    {
      "epoch": 0.21408660100353094,
      "grad_norm": 0.5266989469528198,
      "learning_rate": 0.00019459820523266017,
      "loss": 1.0451,
      "step": 1152
    },
    {
      "epoch": 0.21427244006690208,
      "grad_norm": 0.5035572052001953,
      "learning_rate": 0.00019458871911670524,
      "loss": 1.2143,
      "step": 1153
    },
    {
      "epoch": 0.21445827913027318,
      "grad_norm": 0.4713079035282135,
      "learning_rate": 0.00019457922491037275,
      "loss": 1.093,
      "step": 1154
    },
    {
      "epoch": 0.21464411819364432,
      "grad_norm": 0.508577823638916,
      "learning_rate": 0.00019456972261447475,
      "loss": 1.3012,
      "step": 1155
    },
    {
      "epoch": 0.21482995725701542,
      "grad_norm": 0.9455997347831726,
      "learning_rate": 0.00019456021222982397,
      "loss": 1.1844,
      "step": 1156
    },
    {
      "epoch": 0.21501579632038653,
      "grad_norm": 0.5246649980545044,
      "learning_rate": 0.00019455069375723386,
      "loss": 1.1327,
      "step": 1157
    },
    {
      "epoch": 0.21520163538375767,
      "grad_norm": 0.5420750379562378,
      "learning_rate": 0.00019454116719751856,
      "loss": 1.1217,
      "step": 1158
    },
    {
      "epoch": 0.21538747444712877,
      "grad_norm": 0.49900150299072266,
      "learning_rate": 0.00019453163255149289,
      "loss": 0.8934,
      "step": 1159
    },
    {
      "epoch": 0.2155733135104999,
      "grad_norm": 0.5530839562416077,
      "learning_rate": 0.00019452208981997243,
      "loss": 1.0734,
      "step": 1160
    },
    {
      "epoch": 0.21575915257387102,
      "grad_norm": 0.5028192400932312,
      "learning_rate": 0.0001945125390037733,
      "loss": 1.0318,
      "step": 1161
    },
    {
      "epoch": 0.21594499163724215,
      "grad_norm": 0.4835618734359741,
      "learning_rate": 0.00019450298010371242,
      "loss": 1.1212,
      "step": 1162
    },
    {
      "epoch": 0.21613083070061326,
      "grad_norm": 0.7778669595718384,
      "learning_rate": 0.00019449341312060745,
      "loss": 1.5447,
      "step": 1163
    },
    {
      "epoch": 0.2163166697639844,
      "grad_norm": 0.5164691209793091,
      "learning_rate": 0.0001944838380552766,
      "loss": 1.261,
      "step": 1164
    },
    {
      "epoch": 0.2165025088273555,
      "grad_norm": 0.5152435302734375,
      "learning_rate": 0.00019447425490853887,
      "loss": 1.0973,
      "step": 1165
    },
    {
      "epoch": 0.21668834789072663,
      "grad_norm": 0.48500069975852966,
      "learning_rate": 0.00019446466368121398,
      "loss": 1.0991,
      "step": 1166
    },
    {
      "epoch": 0.21687418695409774,
      "grad_norm": 0.7115551829338074,
      "learning_rate": 0.00019445506437412217,
      "loss": 1.1862,
      "step": 1167
    },
    {
      "epoch": 0.21706002601746888,
      "grad_norm": 0.469462126493454,
      "learning_rate": 0.0001944454569880846,
      "loss": 1.0227,
      "step": 1168
    },
    {
      "epoch": 0.21724586508083998,
      "grad_norm": 0.46283483505249023,
      "learning_rate": 0.00019443584152392296,
      "loss": 1.0887,
      "step": 1169
    },
    {
      "epoch": 0.21743170414421112,
      "grad_norm": 0.594821572303772,
      "learning_rate": 0.0001944262179824597,
      "loss": 1.1053,
      "step": 1170
    },
    {
      "epoch": 0.21761754320758223,
      "grad_norm": 0.5451986193656921,
      "learning_rate": 0.00019441658636451794,
      "loss": 1.041,
      "step": 1171
    },
    {
      "epoch": 0.21780338227095336,
      "grad_norm": 3.8261632919311523,
      "learning_rate": 0.00019440694667092147,
      "loss": 4.0756,
      "step": 1172
    },
    {
      "epoch": 0.21798922133432447,
      "grad_norm": 0.5103700160980225,
      "learning_rate": 0.00019439729890249484,
      "loss": 1.1334,
      "step": 1173
    },
    {
      "epoch": 0.2181750603976956,
      "grad_norm": 0.5776328444480896,
      "learning_rate": 0.00019438764306006317,
      "loss": 1.2469,
      "step": 1174
    },
    {
      "epoch": 0.2183608994610667,
      "grad_norm": 0.5896931886672974,
      "learning_rate": 0.00019437797914445242,
      "loss": 0.9341,
      "step": 1175
    },
    {
      "epoch": 0.21854673852443784,
      "grad_norm": 0.5199896097183228,
      "learning_rate": 0.00019436830715648912,
      "loss": 1.1746,
      "step": 1176
    },
    {
      "epoch": 0.21873257758780895,
      "grad_norm": 0.5308344960212708,
      "learning_rate": 0.00019435862709700057,
      "loss": 1.1905,
      "step": 1177
    },
    {
      "epoch": 0.2189184166511801,
      "grad_norm": 0.6593711376190186,
      "learning_rate": 0.0001943489389668147,
      "loss": 1.3087,
      "step": 1178
    },
    {
      "epoch": 0.2191042557145512,
      "grad_norm": 0.4791925847530365,
      "learning_rate": 0.00019433924276676017,
      "loss": 1.2127,
      "step": 1179
    },
    {
      "epoch": 0.21929009477792233,
      "grad_norm": 0.48847538232803345,
      "learning_rate": 0.0001943295384976663,
      "loss": 0.7622,
      "step": 1180
    },
    {
      "epoch": 0.21947593384129344,
      "grad_norm": 0.614321768283844,
      "learning_rate": 0.00019431982616036314,
      "loss": 1.1416,
      "step": 1181
    },
    {
      "epoch": 0.21966177290466457,
      "grad_norm": 0.43737366795539856,
      "learning_rate": 0.0001943101057556814,
      "loss": 0.8295,
      "step": 1182
    },
    {
      "epoch": 0.21984761196803568,
      "grad_norm": 0.5130372047424316,
      "learning_rate": 0.00019430037728445247,
      "loss": 1.1073,
      "step": 1183
    },
    {
      "epoch": 0.2200334510314068,
      "grad_norm": 0.5877685546875,
      "learning_rate": 0.00019429064074750848,
      "loss": 1.2899,
      "step": 1184
    },
    {
      "epoch": 0.22021929009477792,
      "grad_norm": 0.543999969959259,
      "learning_rate": 0.00019428089614568217,
      "loss": 1.4243,
      "step": 1185
    },
    {
      "epoch": 0.22040512915814905,
      "grad_norm": 0.4639683961868286,
      "learning_rate": 0.00019427114347980707,
      "loss": 1.1822,
      "step": 1186
    },
    {
      "epoch": 0.22059096822152016,
      "grad_norm": 0.5469757914543152,
      "learning_rate": 0.0001942613827507173,
      "loss": 1.2563,
      "step": 1187
    },
    {
      "epoch": 0.2207768072848913,
      "grad_norm": 0.5121818780899048,
      "learning_rate": 0.00019425161395924775,
      "loss": 1.1868,
      "step": 1188
    },
    {
      "epoch": 0.2209626463482624,
      "grad_norm": 0.4748496115207672,
      "learning_rate": 0.00019424183710623397,
      "loss": 0.9397,
      "step": 1189
    },
    {
      "epoch": 0.22114848541163354,
      "grad_norm": 0.5774352550506592,
      "learning_rate": 0.00019423205219251216,
      "loss": 1.304,
      "step": 1190
    },
    {
      "epoch": 0.22133432447500465,
      "grad_norm": 0.4741893410682678,
      "learning_rate": 0.00019422225921891925,
      "loss": 0.905,
      "step": 1191
    },
    {
      "epoch": 0.22152016353837575,
      "grad_norm": 0.500920832157135,
      "learning_rate": 0.0001942124581862929,
      "loss": 1.043,
      "step": 1192
    },
    {
      "epoch": 0.2217060026017469,
      "grad_norm": 0.4882694184780121,
      "learning_rate": 0.00019420264909547136,
      "loss": 0.9522,
      "step": 1193
    },
    {
      "epoch": 0.221891841665118,
      "grad_norm": 0.4808083176612854,
      "learning_rate": 0.00019419283194729365,
      "loss": 1.125,
      "step": 1194
    },
    {
      "epoch": 0.22207768072848913,
      "grad_norm": 0.5128915905952454,
      "learning_rate": 0.00019418300674259946,
      "loss": 0.9932,
      "step": 1195
    },
    {
      "epoch": 0.22226351979186024,
      "grad_norm": 0.5226528644561768,
      "learning_rate": 0.0001941731734822291,
      "loss": 1.1762,
      "step": 1196
    },
    {
      "epoch": 0.22244935885523137,
      "grad_norm": 0.504802405834198,
      "learning_rate": 0.00019416333216702375,
      "loss": 1.0157,
      "step": 1197
    },
    {
      "epoch": 0.22263519791860248,
      "grad_norm": 0.5046318173408508,
      "learning_rate": 0.00019415348279782504,
      "loss": 0.9394,
      "step": 1198
    },
    {
      "epoch": 0.2228210369819736,
      "grad_norm": 0.524804949760437,
      "learning_rate": 0.00019414362537547545,
      "loss": 1.0875,
      "step": 1199
    },
    {
      "epoch": 0.22300687604534472,
      "grad_norm": 0.45535653829574585,
      "learning_rate": 0.0001941337599008181,
      "loss": 1.1935,
      "step": 1200
    },
    {
      "epoch": 0.22319271510871586,
      "grad_norm": 0.5665996670722961,
      "learning_rate": 0.00019412388637469685,
      "loss": 1.2213,
      "step": 1201
    },
    {
      "epoch": 0.22337855417208696,
      "grad_norm": 0.6267208456993103,
      "learning_rate": 0.00019411400479795617,
      "loss": 0.9289,
      "step": 1202
    },
    {
      "epoch": 0.2235643932354581,
      "grad_norm": 0.5627426505088806,
      "learning_rate": 0.00019410411517144124,
      "loss": 1.1286,
      "step": 1203
    },
    {
      "epoch": 0.2237502322988292,
      "grad_norm": 0.5556461215019226,
      "learning_rate": 0.00019409421749599795,
      "loss": 1.0649,
      "step": 1204
    },
    {
      "epoch": 0.22393607136220034,
      "grad_norm": 0.45988309383392334,
      "learning_rate": 0.00019408431177247288,
      "loss": 0.9952,
      "step": 1205
    },
    {
      "epoch": 0.22412191042557145,
      "grad_norm": 0.6533599495887756,
      "learning_rate": 0.00019407439800171326,
      "loss": 1.2076,
      "step": 1206
    },
    {
      "epoch": 0.22430774948894258,
      "grad_norm": 0.49994662404060364,
      "learning_rate": 0.0001940644761845671,
      "loss": 0.7867,
      "step": 1207
    },
    {
      "epoch": 0.2244935885523137,
      "grad_norm": 0.46270355582237244,
      "learning_rate": 0.00019405454632188295,
      "loss": 0.9882,
      "step": 1208
    },
    {
      "epoch": 0.22467942761568482,
      "grad_norm": 0.4713519215583801,
      "learning_rate": 0.0001940446084145102,
      "loss": 1.1609,
      "step": 1209
    },
    {
      "epoch": 0.22486526667905593,
      "grad_norm": 0.5355565547943115,
      "learning_rate": 0.0001940346624632988,
      "loss": 0.8165,
      "step": 1210
    },
    {
      "epoch": 0.22505110574242707,
      "grad_norm": 0.6384341716766357,
      "learning_rate": 0.00019402470846909952,
      "loss": 1.2009,
      "step": 1211
    },
    {
      "epoch": 0.22523694480579817,
      "grad_norm": 0.7215434908866882,
      "learning_rate": 0.0001940147464327637,
      "loss": 1.2914,
      "step": 1212
    },
    {
      "epoch": 0.2254227838691693,
      "grad_norm": 0.5999711751937866,
      "learning_rate": 0.00019400477635514346,
      "loss": 1.1607,
      "step": 1213
    },
    {
      "epoch": 0.22560862293254041,
      "grad_norm": 0.5356886386871338,
      "learning_rate": 0.00019399479823709146,
      "loss": 1.2823,
      "step": 1214
    },
    {
      "epoch": 0.22579446199591155,
      "grad_norm": 0.555874764919281,
      "learning_rate": 0.00019398481207946126,
      "loss": 1.3053,
      "step": 1215
    },
    {
      "epoch": 0.22598030105928266,
      "grad_norm": 0.5327717661857605,
      "learning_rate": 0.00019397481788310694,
      "loss": 1.0022,
      "step": 1216
    },
    {
      "epoch": 0.2261661401226538,
      "grad_norm": 0.530863344669342,
      "learning_rate": 0.00019396481564888337,
      "loss": 1.1806,
      "step": 1217
    },
    {
      "epoch": 0.2263519791860249,
      "grad_norm": 0.5350611805915833,
      "learning_rate": 0.00019395480537764598,
      "loss": 1.1462,
      "step": 1218
    },
    {
      "epoch": 0.22653781824939603,
      "grad_norm": 0.41025102138519287,
      "learning_rate": 0.00019394478707025105,
      "loss": 0.7575,
      "step": 1219
    },
    {
      "epoch": 0.22672365731276714,
      "grad_norm": 0.5089498162269592,
      "learning_rate": 0.00019393476072755546,
      "loss": 1.0725,
      "step": 1220
    },
    {
      "epoch": 0.22690949637613828,
      "grad_norm": 0.5464990735054016,
      "learning_rate": 0.00019392472635041674,
      "loss": 0.9933,
      "step": 1221
    },
    {
      "epoch": 0.22709533543950938,
      "grad_norm": 0.5634016990661621,
      "learning_rate": 0.00019391468393969316,
      "loss": 0.9743,
      "step": 1222
    },
    {
      "epoch": 0.22728117450288052,
      "grad_norm": 10.35197639465332,
      "learning_rate": 0.00019390463349624372,
      "loss": 3.9092,
      "step": 1223
    },
    {
      "epoch": 0.22746701356625162,
      "grad_norm": 0.462494820356369,
      "learning_rate": 0.000193894575020928,
      "loss": 0.7876,
      "step": 1224
    },
    {
      "epoch": 0.22765285262962276,
      "grad_norm": 0.465165913105011,
      "learning_rate": 0.0001938845085146064,
      "loss": 0.9355,
      "step": 1225
    },
    {
      "epoch": 0.22783869169299387,
      "grad_norm": 1.9577351808547974,
      "learning_rate": 0.0001938744339781398,
      "loss": 3.4486,
      "step": 1226
    },
    {
      "epoch": 0.228024530756365,
      "grad_norm": 0.6147559285163879,
      "learning_rate": 0.00019386435141239004,
      "loss": 1.0793,
      "step": 1227
    },
    {
      "epoch": 0.2282103698197361,
      "grad_norm": 0.5095495581626892,
      "learning_rate": 0.00019385426081821938,
      "loss": 0.9592,
      "step": 1228
    },
    {
      "epoch": 0.22839620888310722,
      "grad_norm": 0.5296030640602112,
      "learning_rate": 0.00019384416219649098,
      "loss": 1.3726,
      "step": 1229
    },
    {
      "epoch": 0.22858204794647835,
      "grad_norm": 0.7352867722511292,
      "learning_rate": 0.00019383405554806855,
      "loss": 1.1072,
      "step": 1230
    },
    {
      "epoch": 0.22876788700984946,
      "grad_norm": 0.40808743238449097,
      "learning_rate": 0.00019382394087381655,
      "loss": 0.701,
      "step": 1231
    },
    {
      "epoch": 0.2289537260732206,
      "grad_norm": 0.5722739696502686,
      "learning_rate": 0.0001938138181746001,
      "loss": 1.2682,
      "step": 1232
    },
    {
      "epoch": 0.2291395651365917,
      "grad_norm": 0.6272040009498596,
      "learning_rate": 0.00019380368745128504,
      "loss": 0.7458,
      "step": 1233
    },
    {
      "epoch": 0.22932540419996283,
      "grad_norm": 0.5764666199684143,
      "learning_rate": 0.00019379354870473786,
      "loss": 0.9817,
      "step": 1234
    },
    {
      "epoch": 0.22951124326333394,
      "grad_norm": 0.682600736618042,
      "learning_rate": 0.00019378340193582568,
      "loss": 1.1695,
      "step": 1235
    },
    {
      "epoch": 0.22969708232670508,
      "grad_norm": 0.5942116379737854,
      "learning_rate": 0.00019377324714541652,
      "loss": 1.0587,
      "step": 1236
    },
    {
      "epoch": 0.22988292139007618,
      "grad_norm": 0.5158277153968811,
      "learning_rate": 0.00019376308433437882,
      "loss": 1.0289,
      "step": 1237
    },
    {
      "epoch": 0.23006876045344732,
      "grad_norm": 0.6414231657981873,
      "learning_rate": 0.00019375291350358186,
      "loss": 1.3772,
      "step": 1238
    },
    {
      "epoch": 0.23025459951681843,
      "grad_norm": 0.578842282295227,
      "learning_rate": 0.00019374273465389564,
      "loss": 1.1186,
      "step": 1239
    },
    {
      "epoch": 0.23044043858018956,
      "grad_norm": 0.6373087763786316,
      "learning_rate": 0.00019373254778619066,
      "loss": 0.9283,
      "step": 1240
    },
    {
      "epoch": 0.23062627764356067,
      "grad_norm": 0.5056608319282532,
      "learning_rate": 0.00019372235290133832,
      "loss": 0.9324,
      "step": 1241
    },
    {
      "epoch": 0.2308121167069318,
      "grad_norm": 0.5139119625091553,
      "learning_rate": 0.00019371215000021054,
      "loss": 1.1976,
      "step": 1242
    },
    {
      "epoch": 0.2309979557703029,
      "grad_norm": 0.4746711552143097,
      "learning_rate": 0.00019370193908368008,
      "loss": 0.8965,
      "step": 1243
    },
    {
      "epoch": 0.23118379483367404,
      "grad_norm": 0.580740213394165,
      "learning_rate": 0.00019369172015262022,
      "loss": 0.9536,
      "step": 1244
    },
    {
      "epoch": 0.23136963389704515,
      "grad_norm": 0.48911336064338684,
      "learning_rate": 0.00019368149320790506,
      "loss": 1.0331,
      "step": 1245
    },
    {
      "epoch": 0.2315554729604163,
      "grad_norm": 0.5302042365074158,
      "learning_rate": 0.00019367125825040928,
      "loss": 1.0308,
      "step": 1246
    },
    {
      "epoch": 0.2317413120237874,
      "grad_norm": 0.5719622373580933,
      "learning_rate": 0.00019366101528100834,
      "loss": 1.3409,
      "step": 1247
    },
    {
      "epoch": 0.23192715108715853,
      "grad_norm": 0.5955301523208618,
      "learning_rate": 0.00019365076430057836,
      "loss": 1.167,
      "step": 1248
    },
    {
      "epoch": 0.23211299015052964,
      "grad_norm": 0.5893456339836121,
      "learning_rate": 0.0001936405053099961,
      "loss": 1.211,
      "step": 1249
    },
    {
      "epoch": 0.23229882921390077,
      "grad_norm": 0.5408098697662354,
      "learning_rate": 0.000193630238310139,
      "loss": 0.9324,
      "step": 1250
    },
    {
      "epoch": 0.23248466827727188,
      "grad_norm": 0.5345112085342407,
      "learning_rate": 0.0001936199633018853,
      "loss": 1.1313,
      "step": 1251
    },
    {
      "epoch": 0.232670507340643,
      "grad_norm": 0.6104554533958435,
      "learning_rate": 0.0001936096802861138,
      "loss": 1.1836,
      "step": 1252
    },
    {
      "epoch": 0.23285634640401412,
      "grad_norm": 0.4254988729953766,
      "learning_rate": 0.000193599389263704,
      "loss": 0.8247,
      "step": 1253
    },
    {
      "epoch": 0.23304218546738525,
      "grad_norm": 0.5082131624221802,
      "learning_rate": 0.00019358909023553614,
      "loss": 1.1895,
      "step": 1254
    },
    {
      "epoch": 0.23322802453075636,
      "grad_norm": 0.5012628436088562,
      "learning_rate": 0.00019357878320249114,
      "loss": 1.1176,
      "step": 1255
    },
    {
      "epoch": 0.2334138635941275,
      "grad_norm": 0.5261410474777222,
      "learning_rate": 0.00019356846816545055,
      "loss": 1.0318,
      "step": 1256
    },
    {
      "epoch": 0.2335997026574986,
      "grad_norm": 0.49757251143455505,
      "learning_rate": 0.00019355814512529666,
      "loss": 1.0032,
      "step": 1257
    },
    {
      "epoch": 0.23378554172086974,
      "grad_norm": 0.5406329035758972,
      "learning_rate": 0.00019354781408291243,
      "loss": 1.0992,
      "step": 1258
    },
    {
      "epoch": 0.23397138078424085,
      "grad_norm": 0.5017756819725037,
      "learning_rate": 0.00019353747503918144,
      "loss": 1.1414,
      "step": 1259
    },
    {
      "epoch": 0.23415721984761198,
      "grad_norm": 0.5275353193283081,
      "learning_rate": 0.00019352712799498804,
      "loss": 1.0786,
      "step": 1260
    },
    {
      "epoch": 0.2343430589109831,
      "grad_norm": 0.5688086152076721,
      "learning_rate": 0.0001935167729512173,
      "loss": 1.2523,
      "step": 1261
    },
    {
      "epoch": 0.23452889797435422,
      "grad_norm": 0.5454958081245422,
      "learning_rate": 0.0001935064099087548,
      "loss": 0.9357,
      "step": 1262
    },
    {
      "epoch": 0.23471473703772533,
      "grad_norm": 0.5572336316108704,
      "learning_rate": 0.000193496038868487,
      "loss": 1.2626,
      "step": 1263
    },
    {
      "epoch": 0.23490057610109644,
      "grad_norm": 0.6043594479560852,
      "learning_rate": 0.0001934856598313009,
      "loss": 1.1696,
      "step": 1264
    },
    {
      "epoch": 0.23508641516446757,
      "grad_norm": 0.502384603023529,
      "learning_rate": 0.00019347527279808428,
      "loss": 1.0992,
      "step": 1265
    },
    {
      "epoch": 0.23527225422783868,
      "grad_norm": 0.5702905654907227,
      "learning_rate": 0.00019346487776972555,
      "loss": 1.0492,
      "step": 1266
    },
    {
      "epoch": 0.2354580932912098,
      "grad_norm": 0.6055151224136353,
      "learning_rate": 0.00019345447474711382,
      "loss": 1.0244,
      "step": 1267
    },
    {
      "epoch": 0.23564393235458092,
      "grad_norm": 0.5263853669166565,
      "learning_rate": 0.00019344406373113886,
      "loss": 1.1904,
      "step": 1268
    },
    {
      "epoch": 0.23582977141795206,
      "grad_norm": 0.5104175806045532,
      "learning_rate": 0.0001934336447226912,
      "loss": 0.9947,
      "step": 1269
    },
    {
      "epoch": 0.23601561048132316,
      "grad_norm": 0.6638094186782837,
      "learning_rate": 0.00019342321772266197,
      "loss": 1.1783,
      "step": 1270
    },
    {
      "epoch": 0.2362014495446943,
      "grad_norm": 0.4550234079360962,
      "learning_rate": 0.00019341278273194302,
      "loss": 1.0144,
      "step": 1271
    },
    {
      "epoch": 0.2363872886080654,
      "grad_norm": 0.5424718260765076,
      "learning_rate": 0.00019340233975142685,
      "loss": 1.1628,
      "step": 1272
    },
    {
      "epoch": 0.23657312767143654,
      "grad_norm": 0.512442409992218,
      "learning_rate": 0.00019339188878200666,
      "loss": 1.0532,
      "step": 1273
    },
    {
      "epoch": 0.23675896673480765,
      "grad_norm": 0.5735439658164978,
      "learning_rate": 0.00019338142982457644,
      "loss": 0.8838,
      "step": 1274
    },
    {
      "epoch": 0.23694480579817878,
      "grad_norm": 0.5866655707359314,
      "learning_rate": 0.00019337096288003067,
      "loss": 1.0592,
      "step": 1275
    },
    {
      "epoch": 0.2371306448615499,
      "grad_norm": 0.5844191908836365,
      "learning_rate": 0.00019336048794926466,
      "loss": 1.0053,
      "step": 1276
    },
    {
      "epoch": 0.23731648392492102,
      "grad_norm": 0.4920348525047302,
      "learning_rate": 0.00019335000503317433,
      "loss": 0.943,
      "step": 1277
    },
    {
      "epoch": 0.23750232298829213,
      "grad_norm": 0.47772717475891113,
      "learning_rate": 0.0001933395141326563,
      "loss": 1.2133,
      "step": 1278
    },
    {
      "epoch": 0.23768816205166327,
      "grad_norm": 0.5314902067184448,
      "learning_rate": 0.0001933290152486079,
      "loss": 1.1368,
      "step": 1279
    },
    {
      "epoch": 0.23787400111503437,
      "grad_norm": 0.5484104752540588,
      "learning_rate": 0.00019331850838192713,
      "loss": 1.0915,
      "step": 1280
    },
    {
      "epoch": 0.2380598401784055,
      "grad_norm": 0.5783165097236633,
      "learning_rate": 0.00019330799353351266,
      "loss": 1.0065,
      "step": 1281
    },
    {
      "epoch": 0.23824567924177661,
      "grad_norm": 0.4743618965148926,
      "learning_rate": 0.0001932974707042638,
      "loss": 0.8908,
      "step": 1282
    },
    {
      "epoch": 0.23843151830514775,
      "grad_norm": 0.5681834816932678,
      "learning_rate": 0.00019328693989508063,
      "loss": 1.1216,
      "step": 1283
    },
    {
      "epoch": 0.23861735736851886,
      "grad_norm": 0.5138354301452637,
      "learning_rate": 0.0001932764011068639,
      "loss": 0.8675,
      "step": 1284
    },
    {
      "epoch": 0.23880319643189,
      "grad_norm": 0.6256782412528992,
      "learning_rate": 0.00019326585434051497,
      "loss": 1.0162,
      "step": 1285
    },
    {
      "epoch": 0.2389890354952611,
      "grad_norm": 0.5428333878517151,
      "learning_rate": 0.00019325529959693597,
      "loss": 1.1342,
      "step": 1286
    },
    {
      "epoch": 0.23917487455863223,
      "grad_norm": 0.5870194435119629,
      "learning_rate": 0.0001932447368770296,
      "loss": 1.2211,
      "step": 1287
    },
    {
      "epoch": 0.23936071362200334,
      "grad_norm": 0.6258403658866882,
      "learning_rate": 0.0001932341661816994,
      "loss": 1.1359,
      "step": 1288
    },
    {
      "epoch": 0.23954655268537448,
      "grad_norm": 0.5419567227363586,
      "learning_rate": 0.00019322358751184944,
      "loss": 1.038,
      "step": 1289
    },
    {
      "epoch": 0.23973239174874558,
      "grad_norm": 5.651763439178467,
      "learning_rate": 0.00019321300086838457,
      "loss": 3.7309,
      "step": 1290
    },
    {
      "epoch": 0.23991823081211672,
      "grad_norm": 0.5438091158866882,
      "learning_rate": 0.00019320240625221024,
      "loss": 1.2439,
      "step": 1291
    },
    {
      "epoch": 0.24010406987548782,
      "grad_norm": 0.5790950059890747,
      "learning_rate": 0.0001931918036642327,
      "loss": 0.7855,
      "step": 1292
    },
    {
      "epoch": 0.24028990893885896,
      "grad_norm": 0.6340652108192444,
      "learning_rate": 0.00019318119310535875,
      "loss": 1.3861,
      "step": 1293
    },
    {
      "epoch": 0.24047574800223007,
      "grad_norm": 0.5647812485694885,
      "learning_rate": 0.000193170574576496,
      "loss": 1.2188,
      "step": 1294
    },
    {
      "epoch": 0.2406615870656012,
      "grad_norm": 0.5352049469947815,
      "learning_rate": 0.0001931599480785526,
      "loss": 1.1037,
      "step": 1295
    },
    {
      "epoch": 0.2408474261289723,
      "grad_norm": 0.6570692658424377,
      "learning_rate": 0.00019314931361243752,
      "loss": 1.0727,
      "step": 1296
    },
    {
      "epoch": 0.24103326519234344,
      "grad_norm": 0.5156503319740295,
      "learning_rate": 0.00019313867117906033,
      "loss": 1.1296,
      "step": 1297
    },
    {
      "epoch": 0.24121910425571455,
      "grad_norm": 0.4969639182090759,
      "learning_rate": 0.0001931280207793313,
      "loss": 1.0714,
      "step": 1298
    },
    {
      "epoch": 0.24140494331908569,
      "grad_norm": 0.49877431988716125,
      "learning_rate": 0.00019311736241416135,
      "loss": 0.9473,
      "step": 1299
    },
    {
      "epoch": 0.2415907823824568,
      "grad_norm": 0.5098116993904114,
      "learning_rate": 0.00019310669608446214,
      "loss": 1.0151,
      "step": 1300
    },
    {
      "epoch": 0.2417766214458279,
      "grad_norm": 0.47907474637031555,
      "learning_rate": 0.00019309602179114598,
      "loss": 1.0961,
      "step": 1301
    },
    {
      "epoch": 0.24196246050919903,
      "grad_norm": 0.4947602450847626,
      "learning_rate": 0.0001930853395351259,
      "loss": 1.1161,
      "step": 1302
    },
    {
      "epoch": 0.24214829957257014,
      "grad_norm": 0.5896280407905579,
      "learning_rate": 0.00019307464931731555,
      "loss": 1.1756,
      "step": 1303
    },
    {
      "epoch": 0.24233413863594128,
      "grad_norm": 0.5432973504066467,
      "learning_rate": 0.00019306395113862922,
      "loss": 1.8116,
      "step": 1304
    },
    {
      "epoch": 0.24251997769931238,
      "grad_norm": 0.6156009435653687,
      "learning_rate": 0.00019305324499998208,
      "loss": 1.2343,
      "step": 1305
    },
    {
      "epoch": 0.24270581676268352,
      "grad_norm": 0.47537344694137573,
      "learning_rate": 0.00019304253090228975,
      "loss": 1.155,
      "step": 1306
    },
    {
      "epoch": 0.24289165582605463,
      "grad_norm": 0.4657548666000366,
      "learning_rate": 0.00019303180884646867,
      "loss": 0.8811,
      "step": 1307
    },
    {
      "epoch": 0.24307749488942576,
      "grad_norm": 0.6590142846107483,
      "learning_rate": 0.0001930210788334359,
      "loss": 1.2152,
      "step": 1308
    },
    {
      "epoch": 0.24326333395279687,
      "grad_norm": 0.6140265464782715,
      "learning_rate": 0.00019301034086410923,
      "loss": 1.2579,
      "step": 1309
    },
    {
      "epoch": 0.243449173016168,
      "grad_norm": 0.5011003613471985,
      "learning_rate": 0.00019299959493940708,
      "loss": 0.9862,
      "step": 1310
    },
    {
      "epoch": 0.2436350120795391,
      "grad_norm": 0.4542607069015503,
      "learning_rate": 0.00019298884106024855,
      "loss": 0.8064,
      "step": 1311
    },
    {
      "epoch": 0.24382085114291024,
      "grad_norm": 0.5873056054115295,
      "learning_rate": 0.0001929780792275535,
      "loss": 1.1433,
      "step": 1312
    },
    {
      "epoch": 0.24400669020628135,
      "grad_norm": 0.4630652666091919,
      "learning_rate": 0.00019296730944224236,
      "loss": 1.0874,
      "step": 1313
    },
    {
      "epoch": 0.2441925292696525,
      "grad_norm": 0.7520668506622314,
      "learning_rate": 0.00019295653170523633,
      "loss": 1.2873,
      "step": 1314
    },
    {
      "epoch": 0.2443783683330236,
      "grad_norm": 0.5418159365653992,
      "learning_rate": 0.00019294574601745723,
      "loss": 1.0383,
      "step": 1315
    },
    {
      "epoch": 0.24456420739639473,
      "grad_norm": 0.5760306715965271,
      "learning_rate": 0.00019293495237982758,
      "loss": 1.1608,
      "step": 1316
    },
    {
      "epoch": 0.24475004645976584,
      "grad_norm": 0.4945579469203949,
      "learning_rate": 0.0001929241507932706,
      "loss": 1.0581,
      "step": 1317
    },
    {
      "epoch": 0.24493588552313697,
      "grad_norm": 0.5649460554122925,
      "learning_rate": 0.00019291334125871022,
      "loss": 0.9331,
      "step": 1318
    },
    {
      "epoch": 0.24512172458650808,
      "grad_norm": 0.5532613396644592,
      "learning_rate": 0.00019290252377707087,
      "loss": 1.0461,
      "step": 1319
    },
    {
      "epoch": 0.2453075636498792,
      "grad_norm": 0.4638751745223999,
      "learning_rate": 0.00019289169834927792,
      "loss": 0.9448,
      "step": 1320
    },
    {
      "epoch": 0.24549340271325032,
      "grad_norm": 0.46956831216812134,
      "learning_rate": 0.00019288086497625724,
      "loss": 1.0126,
      "step": 1321
    },
    {
      "epoch": 0.24567924177662145,
      "grad_norm": 0.6272051930427551,
      "learning_rate": 0.0001928700236589354,
      "loss": 1.3591,
      "step": 1322
    },
    {
      "epoch": 0.24586508083999256,
      "grad_norm": 0.5483406782150269,
      "learning_rate": 0.00019285917439823975,
      "loss": 1.212,
      "step": 1323
    },
    {
      "epoch": 0.2460509199033637,
      "grad_norm": 0.5835994482040405,
      "learning_rate": 0.00019284831719509822,
      "loss": 0.9838,
      "step": 1324
    },
    {
      "epoch": 0.2462367589667348,
      "grad_norm": 0.5645042657852173,
      "learning_rate": 0.00019283745205043946,
      "loss": 1.2848,
      "step": 1325
    },
    {
      "epoch": 0.24642259803010594,
      "grad_norm": 0.5618262887001038,
      "learning_rate": 0.00019282657896519276,
      "loss": 1.2191,
      "step": 1326
    },
    {
      "epoch": 0.24660843709347705,
      "grad_norm": 0.4727591872215271,
      "learning_rate": 0.00019281569794028813,
      "loss": 1.032,
      "step": 1327
    },
    {
      "epoch": 0.24679427615684818,
      "grad_norm": 1.797974705696106,
      "learning_rate": 0.00019280480897665626,
      "loss": 3.749,
      "step": 1328
    },
    {
      "epoch": 0.2469801152202193,
      "grad_norm": 0.5174382925033569,
      "learning_rate": 0.0001927939120752285,
      "loss": 0.9723,
      "step": 1329
    },
    {
      "epoch": 0.24716595428359042,
      "grad_norm": 0.5607216358184814,
      "learning_rate": 0.0001927830072369369,
      "loss": 1.2211,
      "step": 1330
    },
    {
      "epoch": 0.24735179334696153,
      "grad_norm": 0.5592674612998962,
      "learning_rate": 0.00019277209446271414,
      "loss": 0.9497,
      "step": 1331
    },
    {
      "epoch": 0.24753763241033266,
      "grad_norm": 0.592069685459137,
      "learning_rate": 0.00019276117375349364,
      "loss": 1.1111,
      "step": 1332
    },
    {
      "epoch": 0.24772347147370377,
      "grad_norm": 0.5631759166717529,
      "learning_rate": 0.0001927502451102095,
      "loss": 1.2154,
      "step": 1333
    },
    {
      "epoch": 0.2479093105370749,
      "grad_norm": 0.5535110235214233,
      "learning_rate": 0.00019273930853379637,
      "loss": 1.1496,
      "step": 1334
    },
    {
      "epoch": 0.248095149600446,
      "grad_norm": 0.5004221200942993,
      "learning_rate": 0.0001927283640251898,
      "loss": 1.2262,
      "step": 1335
    },
    {
      "epoch": 0.24828098866381715,
      "grad_norm": 0.526226282119751,
      "learning_rate": 0.00019271741158532582,
      "loss": 0.9087,
      "step": 1336
    },
    {
      "epoch": 0.24846682772718826,
      "grad_norm": 0.504088819026947,
      "learning_rate": 0.00019270645121514124,
      "loss": 1.0982,
      "step": 1337
    },
    {
      "epoch": 0.24865266679055936,
      "grad_norm": 0.650536060333252,
      "learning_rate": 0.00019269548291557352,
      "loss": 1.2494,
      "step": 1338
    },
    {
      "epoch": 0.2488385058539305,
      "grad_norm": 0.588284969329834,
      "learning_rate": 0.0001926845066875608,
      "loss": 1.3854,
      "step": 1339
    },
    {
      "epoch": 0.2490243449173016,
      "grad_norm": 0.609093427658081,
      "learning_rate": 0.00019267352253204192,
      "loss": 1.0135,
      "step": 1340
    },
    {
      "epoch": 0.24921018398067274,
      "grad_norm": 0.5409224629402161,
      "learning_rate": 0.00019266253044995636,
      "loss": 0.918,
      "step": 1341
    },
    {
      "epoch": 0.24939602304404385,
      "grad_norm": 0.6197240352630615,
      "learning_rate": 0.0001926515304422443,
      "loss": 1.2752,
      "step": 1342
    },
    {
      "epoch": 0.24958186210741498,
      "grad_norm": 0.6102698445320129,
      "learning_rate": 0.0001926405225098466,
      "loss": 0.8095,
      "step": 1343
    },
    {
      "epoch": 0.2497677011707861,
      "grad_norm": 0.5725678205490112,
      "learning_rate": 0.00019262950665370476,
      "loss": 1.0331,
      "step": 1344
    },
    {
      "epoch": 0.24995354023415722,
      "grad_norm": 0.686580240726471,
      "learning_rate": 0.00019261848287476104,
      "loss": 1.2452,
      "step": 1345
    },
    {
      "epoch": 0.25013937929752833,
      "grad_norm": 0.6485561728477478,
      "learning_rate": 0.00019260745117395832,
      "loss": 0.9583,
      "step": 1346
    },
    {
      "epoch": 0.25013937929752833,
      "eval_loss": 1.1052125692367554,
      "eval_runtime": 22.8091,
      "eval_samples_per_second": 47.876,
      "eval_steps_per_second": 23.938,
      "step": 1346
    },
    {
      "epoch": 0.25032521836089944,
      "grad_norm": 0.5164663791656494,
      "learning_rate": 0.0001925964115522401,
      "loss": 0.6874,
      "step": 1347
    },
    {
      "epoch": 0.2505110574242706,
      "grad_norm": 0.4719383716583252,
      "learning_rate": 0.00019258536401055073,
      "loss": 0.9657,
      "step": 1348
    },
    {
      "epoch": 0.2506968964876417,
      "grad_norm": 0.6003237366676331,
      "learning_rate": 0.00019257430854983506,
      "loss": 0.9867,
      "step": 1349
    },
    {
      "epoch": 0.2508827355510128,
      "grad_norm": 0.5269563794136047,
      "learning_rate": 0.00019256324517103867,
      "loss": 0.9536,
      "step": 1350
    },
    {
      "epoch": 0.2510685746143839,
      "grad_norm": 0.502997636795044,
      "learning_rate": 0.00019255217387510788,
      "loss": 1.0991,
      "step": 1351
    },
    {
      "epoch": 0.2512544136777551,
      "grad_norm": 0.5721033811569214,
      "learning_rate": 0.00019254109466298964,
      "loss": 1.1603,
      "step": 1352
    },
    {
      "epoch": 0.2514402527411262,
      "grad_norm": 0.5710723996162415,
      "learning_rate": 0.00019253000753563154,
      "loss": 1.1508,
      "step": 1353
    },
    {
      "epoch": 0.2516260918044973,
      "grad_norm": 0.5256808996200562,
      "learning_rate": 0.00019251891249398196,
      "loss": 0.7742,
      "step": 1354
    },
    {
      "epoch": 0.2518119308678684,
      "grad_norm": 0.6614579558372498,
      "learning_rate": 0.0001925078095389898,
      "loss": 1.297,
      "step": 1355
    },
    {
      "epoch": 0.25199776993123957,
      "grad_norm": 0.4824221432209015,
      "learning_rate": 0.00019249669867160476,
      "loss": 0.9859,
      "step": 1356
    },
    {
      "epoch": 0.2521836089946107,
      "grad_norm": 0.5019868612289429,
      "learning_rate": 0.00019248557989277716,
      "loss": 1.0879,
      "step": 1357
    },
    {
      "epoch": 0.2523694480579818,
      "grad_norm": 0.6684537529945374,
      "learning_rate": 0.00019247445320345804,
      "loss": 1.1135,
      "step": 1358
    },
    {
      "epoch": 0.2525552871213529,
      "grad_norm": 0.5733745098114014,
      "learning_rate": 0.00019246331860459906,
      "loss": 1.1445,
      "step": 1359
    },
    {
      "epoch": 0.25274112618472405,
      "grad_norm": 0.5823042392730713,
      "learning_rate": 0.0001924521760971526,
      "loss": 1.1241,
      "step": 1360
    },
    {
      "epoch": 0.25292696524809516,
      "grad_norm": 0.5691298246383667,
      "learning_rate": 0.0001924410256820717,
      "loss": 1.2497,
      "step": 1361
    },
    {
      "epoch": 0.25311280431146627,
      "grad_norm": 0.630459189414978,
      "learning_rate": 0.0001924298673603101,
      "loss": 1.2617,
      "step": 1362
    },
    {
      "epoch": 0.2532986433748374,
      "grad_norm": 0.4968092739582062,
      "learning_rate": 0.00019241870113282216,
      "loss": 1.0298,
      "step": 1363
    },
    {
      "epoch": 0.25348448243820854,
      "grad_norm": 0.5145949721336365,
      "learning_rate": 0.00019240752700056297,
      "loss": 1.0263,
      "step": 1364
    },
    {
      "epoch": 0.25367032150157964,
      "grad_norm": 0.5401347875595093,
      "learning_rate": 0.0001923963449644883,
      "loss": 1.1958,
      "step": 1365
    },
    {
      "epoch": 0.25385616056495075,
      "grad_norm": 0.4620814323425293,
      "learning_rate": 0.0001923851550255545,
      "loss": 0.9539,
      "step": 1366
    },
    {
      "epoch": 0.25404199962832186,
      "grad_norm": 0.7093937993049622,
      "learning_rate": 0.00019237395718471877,
      "loss": 1.0059,
      "step": 1367
    },
    {
      "epoch": 0.254227838691693,
      "grad_norm": 0.4676429331302643,
      "learning_rate": 0.00019236275144293884,
      "loss": 1.0978,
      "step": 1368
    },
    {
      "epoch": 0.2544136777550641,
      "grad_norm": 0.6647943258285522,
      "learning_rate": 0.00019235153780117314,
      "loss": 1.0992,
      "step": 1369
    },
    {
      "epoch": 0.25459951681843523,
      "grad_norm": 0.5866450667381287,
      "learning_rate": 0.00019234031626038083,
      "loss": 1.3438,
      "step": 1370
    },
    {
      "epoch": 0.25478535588180634,
      "grad_norm": 0.5575841665267944,
      "learning_rate": 0.00019232908682152163,
      "loss": 1.1333,
      "step": 1371
    },
    {
      "epoch": 0.2549711949451775,
      "grad_norm": 0.6370420455932617,
      "learning_rate": 0.00019231784948555615,
      "loss": 1.0891,
      "step": 1372
    },
    {
      "epoch": 0.2551570340085486,
      "grad_norm": 0.49752578139305115,
      "learning_rate": 0.00019230660425344546,
      "loss": 0.9855,
      "step": 1373
    },
    {
      "epoch": 0.2553428730719197,
      "grad_norm": 0.517709493637085,
      "learning_rate": 0.00019229535112615142,
      "loss": 1.1083,
      "step": 1374
    },
    {
      "epoch": 0.2555287121352908,
      "grad_norm": 0.5422150492668152,
      "learning_rate": 0.0001922840901046365,
      "loss": 1.2918,
      "step": 1375
    },
    {
      "epoch": 0.25571455119866193,
      "grad_norm": 0.5656682848930359,
      "learning_rate": 0.00019227282118986394,
      "loss": 1.0533,
      "step": 1376
    },
    {
      "epoch": 0.2559003902620331,
      "grad_norm": 0.4613041579723358,
      "learning_rate": 0.00019226154438279752,
      "loss": 1.0798,
      "step": 1377
    },
    {
      "epoch": 0.2560862293254042,
      "grad_norm": 0.4737628102302551,
      "learning_rate": 0.00019225025968440181,
      "loss": 1.0431,
      "step": 1378
    },
    {
      "epoch": 0.2562720683887753,
      "grad_norm": 0.5070370435714722,
      "learning_rate": 0.00019223896709564202,
      "loss": 1.1703,
      "step": 1379
    },
    {
      "epoch": 0.2564579074521464,
      "grad_norm": 0.49695461988449097,
      "learning_rate": 0.00019222766661748398,
      "loss": 1.2816,
      "step": 1380
    },
    {
      "epoch": 0.2566437465155176,
      "grad_norm": 0.5315166115760803,
      "learning_rate": 0.00019221635825089433,
      "loss": 1.0998,
      "step": 1381
    },
    {
      "epoch": 0.2568295855788887,
      "grad_norm": 0.5896740555763245,
      "learning_rate": 0.00019220504199684023,
      "loss": 0.9867,
      "step": 1382
    },
    {
      "epoch": 0.2570154246422598,
      "grad_norm": 0.6068691611289978,
      "learning_rate": 0.00019219371785628964,
      "loss": 1.2522,
      "step": 1383
    },
    {
      "epoch": 0.2572012637056309,
      "grad_norm": 0.375446617603302,
      "learning_rate": 0.00019218238583021107,
      "loss": 0.7037,
      "step": 1384
    },
    {
      "epoch": 0.25738710276900206,
      "grad_norm": 0.5455111861228943,
      "learning_rate": 0.00019217104591957382,
      "loss": 0.9547,
      "step": 1385
    },
    {
      "epoch": 0.25757294183237317,
      "grad_norm": 0.5593006014823914,
      "learning_rate": 0.00019215969812534783,
      "loss": 1.2173,
      "step": 1386
    },
    {
      "epoch": 0.2577587808957443,
      "grad_norm": 0.5944668054580688,
      "learning_rate": 0.00019214834244850364,
      "loss": 1.2137,
      "step": 1387
    },
    {
      "epoch": 0.2579446199591154,
      "grad_norm": 0.6287859082221985,
      "learning_rate": 0.00019213697889001256,
      "loss": 1.3013,
      "step": 1388
    },
    {
      "epoch": 0.25813045902248655,
      "grad_norm": 0.4995611906051636,
      "learning_rate": 0.00019212560745084656,
      "loss": 1.0576,
      "step": 1389
    },
    {
      "epoch": 0.25831629808585765,
      "grad_norm": 0.494229793548584,
      "learning_rate": 0.00019211422813197824,
      "loss": 0.9894,
      "step": 1390
    },
    {
      "epoch": 0.25850213714922876,
      "grad_norm": 0.4791043698787689,
      "learning_rate": 0.0001921028409343809,
      "loss": 0.9988,
      "step": 1391
    },
    {
      "epoch": 0.25868797621259987,
      "grad_norm": 0.5609716773033142,
      "learning_rate": 0.00019209144585902852,
      "loss": 1.0351,
      "step": 1392
    },
    {
      "epoch": 0.25887381527597103,
      "grad_norm": 0.619724452495575,
      "learning_rate": 0.00019208004290689572,
      "loss": 1.1413,
      "step": 1393
    },
    {
      "epoch": 0.25905965433934214,
      "grad_norm": 0.5314267873764038,
      "learning_rate": 0.00019206863207895785,
      "loss": 1.5014,
      "step": 1394
    },
    {
      "epoch": 0.25924549340271325,
      "grad_norm": 0.49732571840286255,
      "learning_rate": 0.0001920572133761909,
      "loss": 1.0477,
      "step": 1395
    },
    {
      "epoch": 0.25943133246608435,
      "grad_norm": 0.45689892768859863,
      "learning_rate": 0.00019204578679957152,
      "loss": 1.0746,
      "step": 1396
    },
    {
      "epoch": 0.2596171715294555,
      "grad_norm": 0.6244208216667175,
      "learning_rate": 0.00019203435235007706,
      "loss": 1.3679,
      "step": 1397
    },
    {
      "epoch": 0.2598030105928266,
      "grad_norm": 0.5010833144187927,
      "learning_rate": 0.00019202291002868554,
      "loss": 1.0572,
      "step": 1398
    },
    {
      "epoch": 0.25998884965619773,
      "grad_norm": 0.6845056414604187,
      "learning_rate": 0.00019201145983637562,
      "loss": 1.0037,
      "step": 1399
    },
    {
      "epoch": 0.26017468871956884,
      "grad_norm": 0.5309386849403381,
      "learning_rate": 0.0001920000017741267,
      "loss": 1.0226,
      "step": 1400
    },
    {
      "epoch": 0.26036052778294,
      "grad_norm": 0.6004118323326111,
      "learning_rate": 0.00019198853584291876,
      "loss": 1.1238,
      "step": 1401
    },
    {
      "epoch": 0.2605463668463111,
      "grad_norm": 0.4703200161457062,
      "learning_rate": 0.0001919770620437326,
      "loss": 1.0052,
      "step": 1402
    },
    {
      "epoch": 0.2607322059096822,
      "grad_norm": 0.5243449807167053,
      "learning_rate": 0.00019196558037754946,
      "loss": 1.3672,
      "step": 1403
    },
    {
      "epoch": 0.2609180449730533,
      "grad_norm": 0.5253371596336365,
      "learning_rate": 0.00019195409084535146,
      "loss": 1.0317,
      "step": 1404
    },
    {
      "epoch": 0.2611038840364245,
      "grad_norm": 2.241456985473633,
      "learning_rate": 0.0001919425934481214,
      "loss": 3.0421,
      "step": 1405
    },
    {
      "epoch": 0.2612897230997956,
      "grad_norm": 0.5534107089042664,
      "learning_rate": 0.00019193108818684257,
      "loss": 1.0637,
      "step": 1406
    },
    {
      "epoch": 0.2614755621631667,
      "grad_norm": 0.5403827428817749,
      "learning_rate": 0.0001919195750624991,
      "loss": 1.1521,
      "step": 1407
    },
    {
      "epoch": 0.2616614012265378,
      "grad_norm": 0.6491701602935791,
      "learning_rate": 0.0001919080540760757,
      "loss": 1.2381,
      "step": 1408
    },
    {
      "epoch": 0.2618472402899089,
      "grad_norm": 0.5003953576087952,
      "learning_rate": 0.00019189652522855777,
      "loss": 1.1388,
      "step": 1409
    },
    {
      "epoch": 0.2620330793532801,
      "grad_norm": 0.5262916088104248,
      "learning_rate": 0.00019188498852093145,
      "loss": 1.1532,
      "step": 1410
    },
    {
      "epoch": 0.2622189184166512,
      "grad_norm": 0.4679466784000397,
      "learning_rate": 0.00019187344395418346,
      "loss": 0.9516,
      "step": 1411
    },
    {
      "epoch": 0.2624047574800223,
      "grad_norm": 0.5046184062957764,
      "learning_rate": 0.00019186189152930124,
      "loss": 1.2808,
      "step": 1412
    },
    {
      "epoch": 0.2625905965433934,
      "grad_norm": 0.595723032951355,
      "learning_rate": 0.00019185033124727288,
      "loss": 1.2959,
      "step": 1413
    },
    {
      "epoch": 0.26277643560676456,
      "grad_norm": 0.5154600739479065,
      "learning_rate": 0.0001918387631090872,
      "loss": 0.9881,
      "step": 1414
    },
    {
      "epoch": 0.26296227467013567,
      "grad_norm": 0.5345718264579773,
      "learning_rate": 0.00019182718711573365,
      "loss": 1.1117,
      "step": 1415
    },
    {
      "epoch": 0.26314811373350677,
      "grad_norm": 0.4770624041557312,
      "learning_rate": 0.00019181560326820226,
      "loss": 0.9044,
      "step": 1416
    },
    {
      "epoch": 0.2633339527968779,
      "grad_norm": 0.5534773468971252,
      "learning_rate": 0.00019180401156748396,
      "loss": 1.0823,
      "step": 1417
    },
    {
      "epoch": 0.26351979186024904,
      "grad_norm": 0.5540618300437927,
      "learning_rate": 0.00019179241201457007,
      "loss": 1.3371,
      "step": 1418
    },
    {
      "epoch": 0.26370563092362015,
      "grad_norm": 0.5049322247505188,
      "learning_rate": 0.00019178080461045282,
      "loss": 1.2151,
      "step": 1419
    },
    {
      "epoch": 0.26389146998699126,
      "grad_norm": 0.605796754360199,
      "learning_rate": 0.00019176918935612496,
      "loss": 1.0641,
      "step": 1420
    },
    {
      "epoch": 0.26407730905036236,
      "grad_norm": 0.4726361036300659,
      "learning_rate": 0.00019175756625258004,
      "loss": 0.9796,
      "step": 1421
    },
    {
      "epoch": 0.2642631481137335,
      "grad_norm": 0.6339431405067444,
      "learning_rate": 0.00019174593530081212,
      "loss": 1.3727,
      "step": 1422
    },
    {
      "epoch": 0.26444898717710463,
      "grad_norm": 0.6950168013572693,
      "learning_rate": 0.00019173429650181606,
      "loss": 0.8703,
      "step": 1423
    },
    {
      "epoch": 0.26463482624047574,
      "grad_norm": 0.5827997922897339,
      "learning_rate": 0.0001917226498565874,
      "loss": 1.2306,
      "step": 1424
    },
    {
      "epoch": 0.26482066530384685,
      "grad_norm": 0.5235970616340637,
      "learning_rate": 0.00019171099536612225,
      "loss": 1.25,
      "step": 1425
    },
    {
      "epoch": 0.265006504367218,
      "grad_norm": 0.5464771389961243,
      "learning_rate": 0.00019169933303141742,
      "loss": 0.8001,
      "step": 1426
    },
    {
      "epoch": 0.2651923434305891,
      "grad_norm": 0.5554638504981995,
      "learning_rate": 0.00019168766285347047,
      "loss": 1.3444,
      "step": 1427
    },
    {
      "epoch": 0.2653781824939602,
      "grad_norm": 0.5135937333106995,
      "learning_rate": 0.00019167598483327955,
      "loss": 1.0415,
      "step": 1428
    },
    {
      "epoch": 0.26556402155733133,
      "grad_norm": 0.5569391846656799,
      "learning_rate": 0.00019166429897184354,
      "loss": 1.0694,
      "step": 1429
    },
    {
      "epoch": 0.2657498606207025,
      "grad_norm": 0.5736503005027771,
      "learning_rate": 0.0001916526052701619,
      "loss": 0.9726,
      "step": 1430
    },
    {
      "epoch": 0.2659356996840736,
      "grad_norm": 0.47747477889060974,
      "learning_rate": 0.0001916409037292348,
      "loss": 1.0458,
      "step": 1431
    },
    {
      "epoch": 0.2661215387474447,
      "grad_norm": 0.6593626141548157,
      "learning_rate": 0.00019162919435006321,
      "loss": 1.0182,
      "step": 1432
    },
    {
      "epoch": 0.2663073778108158,
      "grad_norm": 0.5842790603637695,
      "learning_rate": 0.00019161747713364858,
      "loss": 1.1817,
      "step": 1433
    },
    {
      "epoch": 0.266493216874187,
      "grad_norm": 0.5009592175483704,
      "learning_rate": 0.00019160575208099307,
      "loss": 1.112,
      "step": 1434
    },
    {
      "epoch": 0.2666790559375581,
      "grad_norm": 0.6723164916038513,
      "learning_rate": 0.00019159401919309965,
      "loss": 1.1049,
      "step": 1435
    },
    {
      "epoch": 0.2668648950009292,
      "grad_norm": 0.530133068561554,
      "learning_rate": 0.00019158227847097177,
      "loss": 1.2624,
      "step": 1436
    },
    {
      "epoch": 0.2670507340643003,
      "grad_norm": 0.6356481313705444,
      "learning_rate": 0.0001915705299156137,
      "loss": 1.3338,
      "step": 1437
    },
    {
      "epoch": 0.26723657312767146,
      "grad_norm": 0.5509076714515686,
      "learning_rate": 0.0001915587735280303,
      "loss": 1.1009,
      "step": 1438
    },
    {
      "epoch": 0.26742241219104257,
      "grad_norm": 0.6118791699409485,
      "learning_rate": 0.0001915470093092271,
      "loss": 1.2114,
      "step": 1439
    },
    {
      "epoch": 0.2676082512544137,
      "grad_norm": 0.5365028977394104,
      "learning_rate": 0.00019153523726021034,
      "loss": 1.0852,
      "step": 1440
    },
    {
      "epoch": 0.2677940903177848,
      "grad_norm": 0.4970124661922455,
      "learning_rate": 0.00019152345738198692,
      "loss": 1.2104,
      "step": 1441
    },
    {
      "epoch": 0.26797992938115595,
      "grad_norm": 0.7996759414672852,
      "learning_rate": 0.00019151166967556438,
      "loss": 1.3305,
      "step": 1442
    },
    {
      "epoch": 0.26816576844452705,
      "grad_norm": 0.4470215141773224,
      "learning_rate": 0.0001914998741419509,
      "loss": 1.0159,
      "step": 1443
    },
    {
      "epoch": 0.26835160750789816,
      "grad_norm": 0.4929320216178894,
      "learning_rate": 0.0001914880707821555,
      "loss": 0.8658,
      "step": 1444
    },
    {
      "epoch": 0.26853744657126927,
      "grad_norm": 0.5478307008743286,
      "learning_rate": 0.00019147625959718765,
      "loss": 1.3146,
      "step": 1445
    },
    {
      "epoch": 0.2687232856346404,
      "grad_norm": 0.6132805943489075,
      "learning_rate": 0.0001914644405880576,
      "loss": 1.1372,
      "step": 1446
    },
    {
      "epoch": 0.26890912469801154,
      "grad_norm": 0.7843652963638306,
      "learning_rate": 0.00019145261375577626,
      "loss": 1.1238,
      "step": 1447
    },
    {
      "epoch": 0.26909496376138264,
      "grad_norm": 0.4491705298423767,
      "learning_rate": 0.0001914407791013552,
      "loss": 0.91,
      "step": 1448
    },
    {
      "epoch": 0.26928080282475375,
      "grad_norm": 0.46373412013053894,
      "learning_rate": 0.00019142893662580667,
      "loss": 1.0154,
      "step": 1449
    },
    {
      "epoch": 0.26946664188812486,
      "grad_norm": 0.5411079525947571,
      "learning_rate": 0.00019141708633014362,
      "loss": 1.2359,
      "step": 1450
    },
    {
      "epoch": 0.269652480951496,
      "grad_norm": 0.5943930149078369,
      "learning_rate": 0.00019140522821537958,
      "loss": 1.208,
      "step": 1451
    },
    {
      "epoch": 0.26983832001486713,
      "grad_norm": 0.5719057321548462,
      "learning_rate": 0.00019139336228252883,
      "loss": 1.1051,
      "step": 1452
    },
    {
      "epoch": 0.27002415907823824,
      "grad_norm": 0.5495271682739258,
      "learning_rate": 0.00019138148853260626,
      "loss": 1.114,
      "step": 1453
    },
    {
      "epoch": 0.27020999814160934,
      "grad_norm": 0.47945669293403625,
      "learning_rate": 0.0001913696069666275,
      "loss": 1.1378,
      "step": 1454
    },
    {
      "epoch": 0.2703958372049805,
      "grad_norm": 0.5663251876831055,
      "learning_rate": 0.00019135771758560876,
      "loss": 1.0401,
      "step": 1455
    },
    {
      "epoch": 0.2705816762683516,
      "grad_norm": 0.5720750093460083,
      "learning_rate": 0.000191345820390567,
      "loss": 0.988,
      "step": 1456
    },
    {
      "epoch": 0.2707675153317227,
      "grad_norm": 0.5430013537406921,
      "learning_rate": 0.00019133391538251978,
      "loss": 1.1438,
      "step": 1457
    },
    {
      "epoch": 0.2709533543950938,
      "grad_norm": 0.5141093730926514,
      "learning_rate": 0.0001913220025624854,
      "loss": 0.8791,
      "step": 1458
    },
    {
      "epoch": 0.271139193458465,
      "grad_norm": 0.5556092858314514,
      "learning_rate": 0.00019131008193148274,
      "loss": 1.318,
      "step": 1459
    },
    {
      "epoch": 0.2713250325218361,
      "grad_norm": 0.5598298907279968,
      "learning_rate": 0.00019129815349053144,
      "loss": 1.0279,
      "step": 1460
    },
    {
      "epoch": 0.2715108715852072,
      "grad_norm": 0.5438913106918335,
      "learning_rate": 0.00019128621724065175,
      "loss": 1.0816,
      "step": 1461
    },
    {
      "epoch": 0.2716967106485783,
      "grad_norm": 0.4408967196941376,
      "learning_rate": 0.0001912742731828646,
      "loss": 0.9494,
      "step": 1462
    },
    {
      "epoch": 0.2718825497119495,
      "grad_norm": 0.5961295962333679,
      "learning_rate": 0.0001912623213181916,
      "loss": 1.2759,
      "step": 1463
    },
    {
      "epoch": 0.2720683887753206,
      "grad_norm": 0.577762246131897,
      "learning_rate": 0.000191250361647655,
      "loss": 1.0705,
      "step": 1464
    },
    {
      "epoch": 0.2722542278386917,
      "grad_norm": 0.5204886794090271,
      "learning_rate": 0.00019123839417227779,
      "loss": 1.1823,
      "step": 1465
    },
    {
      "epoch": 0.2724400669020628,
      "grad_norm": 0.5674437880516052,
      "learning_rate": 0.0001912264188930835,
      "loss": 0.9889,
      "step": 1466
    },
    {
      "epoch": 0.27262590596543396,
      "grad_norm": 0.8880316615104675,
      "learning_rate": 0.00019121443581109643,
      "loss": 1.1098,
      "step": 1467
    },
    {
      "epoch": 0.27281174502880506,
      "grad_norm": 0.5054218769073486,
      "learning_rate": 0.00019120244492734154,
      "loss": 1.1621,
      "step": 1468
    },
    {
      "epoch": 0.27299758409217617,
      "grad_norm": 0.5926326513290405,
      "learning_rate": 0.00019119044624284442,
      "loss": 1.1382,
      "step": 1469
    },
    {
      "epoch": 0.2731834231555473,
      "grad_norm": 2.5959131717681885,
      "learning_rate": 0.00019117843975863136,
      "loss": 3.5187,
      "step": 1470
    },
    {
      "epoch": 0.27336926221891844,
      "grad_norm": 0.5273279547691345,
      "learning_rate": 0.00019116642547572924,
      "loss": 1.1472,
      "step": 1471
    },
    {
      "epoch": 0.27355510128228955,
      "grad_norm": 0.6036831736564636,
      "learning_rate": 0.00019115440339516576,
      "loss": 1.1127,
      "step": 1472
    },
    {
      "epoch": 0.27374094034566065,
      "grad_norm": 0.5351670384407043,
      "learning_rate": 0.0001911423735179691,
      "loss": 1.1082,
      "step": 1473
    },
    {
      "epoch": 0.27392677940903176,
      "grad_norm": 0.7399309873580933,
      "learning_rate": 0.0001911303358451683,
      "loss": 1.1346,
      "step": 1474
    },
    {
      "epoch": 0.2741126184724029,
      "grad_norm": 0.6517348885536194,
      "learning_rate": 0.00019111829037779288,
      "loss": 1.0512,
      "step": 1475
    },
    {
      "epoch": 0.27429845753577403,
      "grad_norm": 0.5458797812461853,
      "learning_rate": 0.00019110623711687317,
      "loss": 1.0631,
      "step": 1476
    },
    {
      "epoch": 0.27448429659914514,
      "grad_norm": 0.5033796429634094,
      "learning_rate": 0.00019109417606344008,
      "loss": 0.9589,
      "step": 1477
    },
    {
      "epoch": 0.27467013566251625,
      "grad_norm": 0.6234574317932129,
      "learning_rate": 0.00019108210721852523,
      "loss": 1.2062,
      "step": 1478
    },
    {
      "epoch": 0.2748559747258874,
      "grad_norm": 0.5345573425292969,
      "learning_rate": 0.00019107003058316092,
      "loss": 1.0058,
      "step": 1479
    },
    {
      "epoch": 0.2750418137892585,
      "grad_norm": 0.589379072189331,
      "learning_rate": 0.00019105794615838003,
      "loss": 1.1664,
      "step": 1480
    },
    {
      "epoch": 0.2752276528526296,
      "grad_norm": 0.5771492123603821,
      "learning_rate": 0.00019104585394521623,
      "loss": 1.1288,
      "step": 1481
    },
    {
      "epoch": 0.27541349191600073,
      "grad_norm": 0.5232322812080383,
      "learning_rate": 0.00019103375394470375,
      "loss": 1.1587,
      "step": 1482
    },
    {
      "epoch": 0.27559933097937184,
      "grad_norm": 0.5217462778091431,
      "learning_rate": 0.0001910216461578776,
      "loss": 1.068,
      "step": 1483
    },
    {
      "epoch": 0.275785170042743,
      "grad_norm": 0.5919577479362488,
      "learning_rate": 0.00019100953058577328,
      "loss": 0.9965,
      "step": 1484
    },
    {
      "epoch": 0.2759710091061141,
      "grad_norm": 0.5517082810401917,
      "learning_rate": 0.0001909974072294271,
      "loss": 1.3083,
      "step": 1485
    },
    {
      "epoch": 0.2761568481694852,
      "grad_norm": 0.7410999536514282,
      "learning_rate": 0.00019098527608987602,
      "loss": 1.1973,
      "step": 1486
    },
    {
      "epoch": 0.2763426872328563,
      "grad_norm": 0.5045968890190125,
      "learning_rate": 0.00019097313716815763,
      "loss": 1.1717,
      "step": 1487
    },
    {
      "epoch": 0.2765285262962275,
      "grad_norm": 0.4956260323524475,
      "learning_rate": 0.0001909609904653102,
      "loss": 1.0734,
      "step": 1488
    },
    {
      "epoch": 0.2767143653595986,
      "grad_norm": 0.486935555934906,
      "learning_rate": 0.00019094883598237266,
      "loss": 1.0529,
      "step": 1489
    },
    {
      "epoch": 0.2769002044229697,
      "grad_norm": 0.48205116391181946,
      "learning_rate": 0.00019093667372038463,
      "loss": 1.0731,
      "step": 1490
    },
    {
      "epoch": 0.2770860434863408,
      "grad_norm": 0.4888480305671692,
      "learning_rate": 0.00019092450368038636,
      "loss": 1.0925,
      "step": 1491
    },
    {
      "epoch": 0.27727188254971197,
      "grad_norm": 0.5380078554153442,
      "learning_rate": 0.00019091232586341875,
      "loss": 0.9717,
      "step": 1492
    },
    {
      "epoch": 0.2774577216130831,
      "grad_norm": 0.4658101797103882,
      "learning_rate": 0.00019090014027052342,
      "loss": 0.7607,
      "step": 1493
    },
    {
      "epoch": 0.2776435606764542,
      "grad_norm": 0.5170148015022278,
      "learning_rate": 0.00019088794690274265,
      "loss": 0.9027,
      "step": 1494
    },
    {
      "epoch": 0.2778293997398253,
      "grad_norm": 0.5996540784835815,
      "learning_rate": 0.00019087574576111935,
      "loss": 1.3506,
      "step": 1495
    },
    {
      "epoch": 0.27801523880319645,
      "grad_norm": 0.4204680621623993,
      "learning_rate": 0.0001908635368466971,
      "loss": 0.6852,
      "step": 1496
    },
    {
      "epoch": 0.27820107786656756,
      "grad_norm": 0.49904823303222656,
      "learning_rate": 0.00019085132016052016,
      "loss": 0.9465,
      "step": 1497
    },
    {
      "epoch": 0.27838691692993867,
      "grad_norm": 0.4688251316547394,
      "learning_rate": 0.00019083909570363346,
      "loss": 1.0922,
      "step": 1498
    },
    {
      "epoch": 0.2785727559933098,
      "grad_norm": 0.5986531972885132,
      "learning_rate": 0.00019082686347708254,
      "loss": 1.2231,
      "step": 1499
    },
    {
      "epoch": 0.27875859505668094,
      "grad_norm": 0.4802838861942291,
      "learning_rate": 0.00019081462348191373,
      "loss": 1.057,
      "step": 1500
    },
    {
      "epoch": 0.27894443412005204,
      "grad_norm": 0.6052405834197998,
      "learning_rate": 0.00019080237571917388,
      "loss": 1.046,
      "step": 1501
    },
    {
      "epoch": 0.27913027318342315,
      "grad_norm": 0.5468049645423889,
      "learning_rate": 0.00019079012018991058,
      "loss": 1.0525,
      "step": 1502
    },
    {
      "epoch": 0.27931611224679426,
      "grad_norm": 0.4576660692691803,
      "learning_rate": 0.00019077785689517208,
      "loss": 0.9774,
      "step": 1503
    },
    {
      "epoch": 0.2795019513101654,
      "grad_norm": 0.5179756879806519,
      "learning_rate": 0.00019076558583600728,
      "loss": 1.1046,
      "step": 1504
    },
    {
      "epoch": 0.2796877903735365,
      "grad_norm": 0.7189592123031616,
      "learning_rate": 0.00019075330701346578,
      "loss": 1.2653,
      "step": 1505
    },
    {
      "epoch": 0.27987362943690763,
      "grad_norm": 0.48832884430885315,
      "learning_rate": 0.00019074102042859775,
      "loss": 1.1849,
      "step": 1506
    },
    {
      "epoch": 0.28005946850027874,
      "grad_norm": 0.5543190836906433,
      "learning_rate": 0.00019072872608245412,
      "loss": 0.8673,
      "step": 1507
    },
    {
      "epoch": 0.2802453075636499,
      "grad_norm": 0.5351570844650269,
      "learning_rate": 0.00019071642397608648,
      "loss": 1.0057,
      "step": 1508
    },
    {
      "epoch": 0.280431146627021,
      "grad_norm": 0.4867394268512726,
      "learning_rate": 0.00019070411411054706,
      "loss": 1.0946,
      "step": 1509
    },
    {
      "epoch": 0.2806169856903921,
      "grad_norm": 0.47239750623703003,
      "learning_rate": 0.00019069179648688867,
      "loss": 1.0842,
      "step": 1510
    },
    {
      "epoch": 0.2808028247537632,
      "grad_norm": 0.5905438661575317,
      "learning_rate": 0.00019067947110616492,
      "loss": 0.9838,
      "step": 1511
    },
    {
      "epoch": 0.2809886638171344,
      "grad_norm": 0.464809387922287,
      "learning_rate": 0.00019066713796943007,
      "loss": 0.9241,
      "step": 1512
    },
    {
      "epoch": 0.2811745028805055,
      "grad_norm": 0.548126220703125,
      "learning_rate": 0.0001906547970777389,
      "loss": 0.9682,
      "step": 1513
    },
    {
      "epoch": 0.2813603419438766,
      "grad_norm": 0.5304195880889893,
      "learning_rate": 0.00019064244843214706,
      "loss": 1.1021,
      "step": 1514
    },
    {
      "epoch": 0.2815461810072477,
      "grad_norm": 0.5824160575866699,
      "learning_rate": 0.00019063009203371066,
      "loss": 1.1086,
      "step": 1515
    },
    {
      "epoch": 0.28173202007061887,
      "grad_norm": 0.5206829905509949,
      "learning_rate": 0.0001906177278834866,
      "loss": 1.1925,
      "step": 1516
    },
    {
      "epoch": 0.28191785913399,
      "grad_norm": 0.6318472623825073,
      "learning_rate": 0.00019060535598253243,
      "loss": 0.9323,
      "step": 1517
    },
    {
      "epoch": 0.2821036981973611,
      "grad_norm": 0.5612223744392395,
      "learning_rate": 0.0001905929763319064,
      "loss": 1.1283,
      "step": 1518
    },
    {
      "epoch": 0.2822895372607322,
      "grad_norm": 0.4812374711036682,
      "learning_rate": 0.00019058058893266724,
      "loss": 0.9585,
      "step": 1519
    },
    {
      "epoch": 0.2824753763241033,
      "grad_norm": 0.5273768305778503,
      "learning_rate": 0.00019056819378587454,
      "loss": 1.3129,
      "step": 1520
    },
    {
      "epoch": 0.28266121538747446,
      "grad_norm": 0.5103570818901062,
      "learning_rate": 0.00019055579089258853,
      "loss": 1.1644,
      "step": 1521
    },
    {
      "epoch": 0.28284705445084557,
      "grad_norm": 0.5185193419456482,
      "learning_rate": 0.00019054338025386994,
      "loss": 1.0037,
      "step": 1522
    },
    {
      "epoch": 0.2830328935142167,
      "grad_norm": 0.5031711459159851,
      "learning_rate": 0.0001905309618707804,
      "loss": 0.8256,
      "step": 1523
    },
    {
      "epoch": 0.2832187325775878,
      "grad_norm": 0.5190253257751465,
      "learning_rate": 0.00019051853574438203,
      "loss": 0.9677,
      "step": 1524
    },
    {
      "epoch": 0.28340457164095895,
      "grad_norm": 0.5510692000389099,
      "learning_rate": 0.00019050610187573762,
      "loss": 0.9814,
      "step": 1525
    },
    {
      "epoch": 0.28359041070433005,
      "grad_norm": 0.5186879634857178,
      "learning_rate": 0.00019049366026591073,
      "loss": 1.1876,
      "step": 1526
    },
    {
      "epoch": 0.28377624976770116,
      "grad_norm": 0.5891383290290833,
      "learning_rate": 0.00019048121091596551,
      "loss": 1.0911,
      "step": 1527
    },
    {
      "epoch": 0.28396208883107227,
      "grad_norm": 0.5759264230728149,
      "learning_rate": 0.00019046875382696676,
      "loss": 0.9242,
      "step": 1528
    },
    {
      "epoch": 0.28414792789444343,
      "grad_norm": 0.6390319466590881,
      "learning_rate": 0.00019045628899997996,
      "loss": 1.1155,
      "step": 1529
    },
    {
      "epoch": 0.28433376695781454,
      "grad_norm": 0.5420219898223877,
      "learning_rate": 0.00019044381643607125,
      "loss": 1.2446,
      "step": 1530
    },
    {
      "epoch": 0.28451960602118564,
      "grad_norm": 0.6645781993865967,
      "learning_rate": 0.00019043133613630745,
      "loss": 1.1495,
      "step": 1531
    },
    {
      "epoch": 0.28470544508455675,
      "grad_norm": 0.6481037139892578,
      "learning_rate": 0.00019041884810175604,
      "loss": 0.9835,
      "step": 1532
    },
    {
      "epoch": 0.2848912841479279,
      "grad_norm": 0.7210860252380371,
      "learning_rate": 0.0001904063523334851,
      "loss": 1.0869,
      "step": 1533
    },
    {
      "epoch": 0.285077123211299,
      "grad_norm": 0.5078470706939697,
      "learning_rate": 0.00019039384883256347,
      "loss": 1.1299,
      "step": 1534
    },
    {
      "epoch": 0.28526296227467013,
      "grad_norm": 0.5624436736106873,
      "learning_rate": 0.00019038133760006058,
      "loss": 1.2529,
      "step": 1535
    },
    {
      "epoch": 0.28544880133804124,
      "grad_norm": 0.49437078833580017,
      "learning_rate": 0.00019036881863704656,
      "loss": 1.0552,
      "step": 1536
    },
    {
      "epoch": 0.2856346404014124,
      "grad_norm": 0.53767329454422,
      "learning_rate": 0.00019035629194459215,
      "loss": 1.0651,
      "step": 1537
    },
    {
      "epoch": 0.2858204794647835,
      "grad_norm": 0.5064103603363037,
      "learning_rate": 0.0001903437575237688,
      "loss": 0.9809,
      "step": 1538
    },
    {
      "epoch": 0.2860063185281546,
      "grad_norm": 0.5985023975372314,
      "learning_rate": 0.00019033121537564862,
      "loss": 0.9943,
      "step": 1539
    },
    {
      "epoch": 0.2861921575915257,
      "grad_norm": 0.6447902917861938,
      "learning_rate": 0.00019031866550130438,
      "loss": 1.1544,
      "step": 1540
    },
    {
      "epoch": 0.2863779966548969,
      "grad_norm": 0.4907586872577667,
      "learning_rate": 0.00019030610790180943,
      "loss": 1.0605,
      "step": 1541
    },
    {
      "epoch": 0.286563835718268,
      "grad_norm": 0.5801265239715576,
      "learning_rate": 0.00019029354257823792,
      "loss": 1.2279,
      "step": 1542
    },
    {
      "epoch": 0.2867496747816391,
      "grad_norm": 0.5001637935638428,
      "learning_rate": 0.00019028096953166456,
      "loss": 0.9343,
      "step": 1543
    },
    {
      "epoch": 0.2869355138450102,
      "grad_norm": 0.5724277496337891,
      "learning_rate": 0.00019026838876316476,
      "loss": 1.0683,
      "step": 1544
    },
    {
      "epoch": 0.28712135290838137,
      "grad_norm": 0.5190825462341309,
      "learning_rate": 0.00019025580027381454,
      "loss": 1.2547,
      "step": 1545
    },
    {
      "epoch": 0.2873071919717525,
      "grad_norm": 0.6240419149398804,
      "learning_rate": 0.0001902432040646907,
      "loss": 0.9924,
      "step": 1546
    },
    {
      "epoch": 0.2874930310351236,
      "grad_norm": 0.48464152216911316,
      "learning_rate": 0.00019023060013687056,
      "loss": 0.9693,
      "step": 1547
    },
    {
      "epoch": 0.2876788700984947,
      "grad_norm": 0.503472089767456,
      "learning_rate": 0.00019021798849143217,
      "loss": 0.997,
      "step": 1548
    },
    {
      "epoch": 0.28786470916186585,
      "grad_norm": 0.580359697341919,
      "learning_rate": 0.00019020536912945423,
      "loss": 1.1117,
      "step": 1549
    },
    {
      "epoch": 0.28805054822523696,
      "grad_norm": 0.622320294380188,
      "learning_rate": 0.0001901927420520161,
      "loss": 1.2558,
      "step": 1550
    },
    {
      "epoch": 0.28823638728860806,
      "grad_norm": 0.5121637582778931,
      "learning_rate": 0.00019018010726019785,
      "loss": 1.0791,
      "step": 1551
    },
    {
      "epoch": 0.28842222635197917,
      "grad_norm": 0.567800760269165,
      "learning_rate": 0.00019016746475508008,
      "loss": 1.0396,
      "step": 1552
    },
    {
      "epoch": 0.2886080654153503,
      "grad_norm": 0.5097231268882751,
      "learning_rate": 0.00019015481453774422,
      "loss": 1.012,
      "step": 1553
    },
    {
      "epoch": 0.28879390447872144,
      "grad_norm": 0.556196928024292,
      "learning_rate": 0.00019014215660927218,
      "loss": 1.2133,
      "step": 1554
    },
    {
      "epoch": 0.28897974354209255,
      "grad_norm": 0.5323203206062317,
      "learning_rate": 0.00019012949097074668,
      "loss": 0.9587,
      "step": 1555
    },
    {
      "epoch": 0.28916558260546366,
      "grad_norm": 0.5991493463516235,
      "learning_rate": 0.00019011681762325103,
      "loss": 1.049,
      "step": 1556
    },
    {
      "epoch": 0.28935142166883476,
      "grad_norm": 0.4759868085384369,
      "learning_rate": 0.00019010413656786918,
      "loss": 1.1119,
      "step": 1557
    },
    {
      "epoch": 0.2895372607322059,
      "grad_norm": 0.5300564765930176,
      "learning_rate": 0.0001900914478056858,
      "loss": 1.2332,
      "step": 1558
    },
    {
      "epoch": 0.28972309979557703,
      "grad_norm": 0.7315372824668884,
      "learning_rate": 0.0001900787513377862,
      "loss": 0.8838,
      "step": 1559
    },
    {
      "epoch": 0.28990893885894814,
      "grad_norm": 0.5923600196838379,
      "learning_rate": 0.00019006604716525626,
      "loss": 0.9924,
      "step": 1560
    },
    {
      "epoch": 0.29009477792231925,
      "grad_norm": 0.5271015763282776,
      "learning_rate": 0.00019005333528918268,
      "loss": 1.1689,
      "step": 1561
    },
    {
      "epoch": 0.2902806169856904,
      "grad_norm": 0.5299558043479919,
      "learning_rate": 0.00019004061571065267,
      "loss": 1.2,
      "step": 1562
    },
    {
      "epoch": 0.2904664560490615,
      "grad_norm": 0.5884658694267273,
      "learning_rate": 0.00019002788843075426,
      "loss": 1.0522,
      "step": 1563
    },
    {
      "epoch": 0.2906522951124326,
      "grad_norm": 0.40285414457321167,
      "learning_rate": 0.0001900151534505759,
      "loss": 0.8925,
      "step": 1564
    },
    {
      "epoch": 0.29083813417580373,
      "grad_norm": 0.46703270077705383,
      "learning_rate": 0.00019000241077120697,
      "loss": 1.0306,
      "step": 1565
    },
    {
      "epoch": 0.2910239732391749,
      "grad_norm": 0.49443286657333374,
      "learning_rate": 0.0001899896603937373,
      "loss": 1.2487,
      "step": 1566
    },
    {
      "epoch": 0.291209812302546,
      "grad_norm": 0.5805255174636841,
      "learning_rate": 0.00018997690231925752,
      "loss": 1.1078,
      "step": 1567
    },
    {
      "epoch": 0.2913956513659171,
      "grad_norm": 0.5543115735054016,
      "learning_rate": 0.0001899641365488588,
      "loss": 1.0998,
      "step": 1568
    },
    {
      "epoch": 0.2915814904292882,
      "grad_norm": 0.5571877956390381,
      "learning_rate": 0.000189951363083633,
      "loss": 1.1004,
      "step": 1569
    },
    {
      "epoch": 0.2917673294926594,
      "grad_norm": 0.4987258315086365,
      "learning_rate": 0.00018993858192467277,
      "loss": 1.1713,
      "step": 1570
    },
    {
      "epoch": 0.2919531685560305,
      "grad_norm": 0.6147323250770569,
      "learning_rate": 0.00018992579307307119,
      "loss": 1.1392,
      "step": 1571
    },
    {
      "epoch": 0.2921390076194016,
      "grad_norm": 0.5259654521942139,
      "learning_rate": 0.00018991299652992217,
      "loss": 0.8903,
      "step": 1572
    },
    {
      "epoch": 0.2923248466827727,
      "grad_norm": 0.6485708355903625,
      "learning_rate": 0.00018990019229632031,
      "loss": 1.045,
      "step": 1573
    },
    {
      "epoch": 0.29251068574614386,
      "grad_norm": 0.49851977825164795,
      "learning_rate": 0.00018988738037336062,
      "loss": 0.7989,
      "step": 1574
    },
    {
      "epoch": 0.29269652480951497,
      "grad_norm": 0.6040017008781433,
      "learning_rate": 0.00018987456076213904,
      "loss": 1.0092,
      "step": 1575
    },
    {
      "epoch": 0.2928823638728861,
      "grad_norm": 0.5967439413070679,
      "learning_rate": 0.00018986173346375205,
      "loss": 1.0741,
      "step": 1576
    },
    {
      "epoch": 0.2930682029362572,
      "grad_norm": 0.6299907565116882,
      "learning_rate": 0.00018984889847929677,
      "loss": 1.2439,
      "step": 1577
    },
    {
      "epoch": 0.29325404199962835,
      "grad_norm": 0.635490894317627,
      "learning_rate": 0.00018983605580987102,
      "loss": 0.9893,
      "step": 1578
    },
    {
      "epoch": 0.29343988106299945,
      "grad_norm": 0.501427173614502,
      "learning_rate": 0.00018982320545657324,
      "loss": 0.9781,
      "step": 1579
    },
    {
      "epoch": 0.29362572012637056,
      "grad_norm": 0.6587309837341309,
      "learning_rate": 0.00018981034742050258,
      "loss": 1.2243,
      "step": 1580
    },
    {
      "epoch": 0.29381155918974167,
      "grad_norm": 0.524802565574646,
      "learning_rate": 0.0001897974817027588,
      "loss": 1.2386,
      "step": 1581
    },
    {
      "epoch": 0.29399739825311283,
      "grad_norm": 0.4880768656730652,
      "learning_rate": 0.0001897846083044423,
      "loss": 1.1358,
      "step": 1582
    },
    {
      "epoch": 0.29418323731648394,
      "grad_norm": 0.46930983662605286,
      "learning_rate": 0.00018977172722665423,
      "loss": 1.0273,
      "step": 1583
    },
    {
      "epoch": 0.29436907637985504,
      "grad_norm": 0.5368134379386902,
      "learning_rate": 0.0001897588384704963,
      "loss": 1.1655,
      "step": 1584
    },
    {
      "epoch": 0.29455491544322615,
      "grad_norm": 0.7028413414955139,
      "learning_rate": 0.00018974594203707094,
      "loss": 1.0176,
      "step": 1585
    },
    {
      "epoch": 0.2947407545065973,
      "grad_norm": 0.5124669075012207,
      "learning_rate": 0.00018973303792748115,
      "loss": 1.1427,
      "step": 1586
    },
    {
      "epoch": 0.2949265935699684,
      "grad_norm": 0.6192018985748291,
      "learning_rate": 0.0001897201261428307,
      "loss": 1.0188,
      "step": 1587
    },
    {
      "epoch": 0.2951124326333395,
      "grad_norm": 0.5622596740722656,
      "learning_rate": 0.000189707206684224,
      "loss": 1.0261,
      "step": 1588
    },
    {
      "epoch": 0.29529827169671063,
      "grad_norm": 0.5319574475288391,
      "learning_rate": 0.000189694279552766,
      "loss": 1.0369,
      "step": 1589
    },
    {
      "epoch": 0.29548411076008174,
      "grad_norm": 0.5429655313491821,
      "learning_rate": 0.0001896813447495624,
      "loss": 0.822,
      "step": 1590
    },
    {
      "epoch": 0.2956699498234529,
      "grad_norm": 0.49856165051460266,
      "learning_rate": 0.00018966840227571953,
      "loss": 0.9698,
      "step": 1591
    },
    {
      "epoch": 0.295855788886824,
      "grad_norm": 0.5880312323570251,
      "learning_rate": 0.00018965545213234445,
      "loss": 1.2094,
      "step": 1592
    },
    {
      "epoch": 0.2960416279501951,
      "grad_norm": 0.6625062823295593,
      "learning_rate": 0.0001896424943205448,
      "loss": 1.2754,
      "step": 1593
    },
    {
      "epoch": 0.2962274670135662,
      "grad_norm": 0.542192816734314,
      "learning_rate": 0.00018962952884142884,
      "loss": 1.0705,
      "step": 1594
    },
    {
      "epoch": 0.2964133060769374,
      "grad_norm": 0.5409523844718933,
      "learning_rate": 0.00018961655569610557,
      "loss": 1.0441,
      "step": 1595
    },
    {
      "epoch": 0.2965991451403085,
      "grad_norm": 0.5490151047706604,
      "learning_rate": 0.0001896035748856846,
      "loss": 1.0385,
      "step": 1596
    },
    {
      "epoch": 0.2967849842036796,
      "grad_norm": 0.5456315279006958,
      "learning_rate": 0.00018959058641127623,
      "loss": 1.0773,
      "step": 1597
    },
    {
      "epoch": 0.2969708232670507,
      "grad_norm": 0.5822267532348633,
      "learning_rate": 0.00018957759027399135,
      "loss": 1.0326,
      "step": 1598
    },
    {
      "epoch": 0.2971566623304219,
      "grad_norm": 0.6549869775772095,
      "learning_rate": 0.00018956458647494158,
      "loss": 1.0455,
      "step": 1599
    },
    {
      "epoch": 0.297342501393793,
      "grad_norm": 0.6730489730834961,
      "learning_rate": 0.00018955157501523913,
      "loss": 0.9874,
      "step": 1600
    },
    {
      "epoch": 0.2975283404571641,
      "grad_norm": 0.5480064153671265,
      "learning_rate": 0.00018953855589599696,
      "loss": 0.9857,
      "step": 1601
    },
    {
      "epoch": 0.2977141795205352,
      "grad_norm": 0.5919156670570374,
      "learning_rate": 0.00018952552911832856,
      "loss": 1.2629,
      "step": 1602
    },
    {
      "epoch": 0.29790001858390636,
      "grad_norm": 0.5088599920272827,
      "learning_rate": 0.00018951249468334818,
      "loss": 1.073,
      "step": 1603
    },
    {
      "epoch": 0.29808585764727746,
      "grad_norm": 0.6555619835853577,
      "learning_rate": 0.00018949945259217067,
      "loss": 1.3914,
      "step": 1604
    },
    {
      "epoch": 0.29827169671064857,
      "grad_norm": 0.5444729924201965,
      "learning_rate": 0.00018948640284591152,
      "loss": 1.1564,
      "step": 1605
    },
    {
      "epoch": 0.2984575357740197,
      "grad_norm": 0.5014877915382385,
      "learning_rate": 0.00018947334544568696,
      "loss": 0.9872,
      "step": 1606
    },
    {
      "epoch": 0.29864337483739084,
      "grad_norm": 0.5414091348648071,
      "learning_rate": 0.0001894602803926138,
      "loss": 0.8565,
      "step": 1607
    },
    {
      "epoch": 0.29882921390076195,
      "grad_norm": 0.5945007801055908,
      "learning_rate": 0.00018944720768780945,
      "loss": 1.238,
      "step": 1608
    },
    {
      "epoch": 0.29901505296413305,
      "grad_norm": 0.6132047772407532,
      "learning_rate": 0.00018943412733239215,
      "loss": 0.9816,
      "step": 1609
    },
    {
      "epoch": 0.29920089202750416,
      "grad_norm": 0.5965632796287537,
      "learning_rate": 0.00018942103932748065,
      "loss": 1.2321,
      "step": 1610
    },
    {
      "epoch": 0.2993867310908753,
      "grad_norm": 0.5612810254096985,
      "learning_rate": 0.00018940794367419437,
      "loss": 1.1615,
      "step": 1611
    },
    {
      "epoch": 0.29957257015424643,
      "grad_norm": 0.5566758513450623,
      "learning_rate": 0.00018939484037365346,
      "loss": 1.099,
      "step": 1612
    },
    {
      "epoch": 0.29975840921761754,
      "grad_norm": 0.4823967218399048,
      "learning_rate": 0.0001893817294269786,
      "loss": 1.065,
      "step": 1613
    },
    {
      "epoch": 0.29994424828098865,
      "grad_norm": 0.505115270614624,
      "learning_rate": 0.00018936861083529126,
      "loss": 1.0358,
      "step": 1614
    },
    {
      "epoch": 0.3001300873443598,
      "grad_norm": 0.4831530749797821,
      "learning_rate": 0.0001893554845997135,
      "loss": 0.9381,
      "step": 1615
    },
    {
      "epoch": 0.3003159264077309,
      "grad_norm": 0.5076389312744141,
      "learning_rate": 0.00018934235072136802,
      "loss": 0.9582,
      "step": 1616
    },
    {
      "epoch": 0.300501765471102,
      "grad_norm": 0.5244806408882141,
      "learning_rate": 0.0001893292092013782,
      "loss": 1.0831,
      "step": 1617
    },
    {
      "epoch": 0.30068760453447313,
      "grad_norm": 0.4544525146484375,
      "learning_rate": 0.000189316060040868,
      "loss": 1.2185,
      "step": 1618
    },
    {
      "epoch": 0.3008734435978443,
      "grad_norm": 0.553446888923645,
      "learning_rate": 0.0001893029032409622,
      "loss": 0.9928,
      "step": 1619
    },
    {
      "epoch": 0.3010592826612154,
      "grad_norm": 0.5081769227981567,
      "learning_rate": 0.00018928973880278605,
      "loss": 1.0391,
      "step": 1620
    },
    {
      "epoch": 0.3012451217245865,
      "grad_norm": 0.5252265930175781,
      "learning_rate": 0.00018927656672746554,
      "loss": 1.0321,
      "step": 1621
    },
    {
      "epoch": 0.3014309607879576,
      "grad_norm": 0.5269213914871216,
      "learning_rate": 0.00018926338701612738,
      "loss": 1.2644,
      "step": 1622
    },
    {
      "epoch": 0.3016167998513288,
      "grad_norm": 0.5293154120445251,
      "learning_rate": 0.00018925019966989877,
      "loss": 1.0577,
      "step": 1623
    },
    {
      "epoch": 0.3018026389146999,
      "grad_norm": 0.593284010887146,
      "learning_rate": 0.00018923700468990767,
      "loss": 1.1221,
      "step": 1624
    },
    {
      "epoch": 0.301988477978071,
      "grad_norm": 0.5898813605308533,
      "learning_rate": 0.00018922380207728274,
      "loss": 1.0924,
      "step": 1625
    },
    {
      "epoch": 0.3021743170414421,
      "grad_norm": 0.5219821333885193,
      "learning_rate": 0.00018921059183315316,
      "loss": 1.0935,
      "step": 1626
    },
    {
      "epoch": 0.3023601561048132,
      "grad_norm": 0.5993270874023438,
      "learning_rate": 0.00018919737395864882,
      "loss": 1.0921,
      "step": 1627
    },
    {
      "epoch": 0.30254599516818437,
      "grad_norm": 0.4896244704723358,
      "learning_rate": 0.00018918414845490035,
      "loss": 0.9751,
      "step": 1628
    },
    {
      "epoch": 0.3027318342315555,
      "grad_norm": 0.5135372281074524,
      "learning_rate": 0.0001891709153230389,
      "loss": 1.0925,
      "step": 1629
    },
    {
      "epoch": 0.3029176732949266,
      "grad_norm": 0.6543980240821838,
      "learning_rate": 0.0001891576745641963,
      "loss": 1.2144,
      "step": 1630
    },
    {
      "epoch": 0.3031035123582977,
      "grad_norm": 0.5897946953773499,
      "learning_rate": 0.00018914442617950513,
      "loss": 1.1757,
      "step": 1631
    },
    {
      "epoch": 0.30328935142166885,
      "grad_norm": 0.5730814933776855,
      "learning_rate": 0.00018913117017009854,
      "loss": 0.8796,
      "step": 1632
    },
    {
      "epoch": 0.30347519048503996,
      "grad_norm": 0.5452990531921387,
      "learning_rate": 0.0001891179065371103,
      "loss": 1.0472,
      "step": 1633
    },
    {
      "epoch": 0.30366102954841107,
      "grad_norm": 0.5573520064353943,
      "learning_rate": 0.00018910463528167492,
      "loss": 1.2547,
      "step": 1634
    },
    {
      "epoch": 0.3038468686117822,
      "grad_norm": 0.4705549478530884,
      "learning_rate": 0.0001890913564049275,
      "loss": 0.9675,
      "step": 1635
    },
    {
      "epoch": 0.30403270767515334,
      "grad_norm": 0.525519073009491,
      "learning_rate": 0.00018907806990800383,
      "loss": 0.9774,
      "step": 1636
    },
    {
      "epoch": 0.30421854673852444,
      "grad_norm": 0.5609502792358398,
      "learning_rate": 0.0001890647757920403,
      "loss": 0.9778,
      "step": 1637
    },
    {
      "epoch": 0.30440438580189555,
      "grad_norm": 0.5060504674911499,
      "learning_rate": 0.00018905147405817399,
      "loss": 1.1342,
      "step": 1638
    },
    {
      "epoch": 0.30459022486526666,
      "grad_norm": 0.7549655437469482,
      "learning_rate": 0.00018903816470754265,
      "loss": 0.8456,
      "step": 1639
    },
    {
      "epoch": 0.3047760639286378,
      "grad_norm": 0.5245809555053711,
      "learning_rate": 0.00018902484774128463,
      "loss": 1.2306,
      "step": 1640
    },
    {
      "epoch": 0.3049619029920089,
      "grad_norm": 0.5347371101379395,
      "learning_rate": 0.000189011523160539,
      "loss": 0.883,
      "step": 1641
    },
    {
      "epoch": 0.30514774205538003,
      "grad_norm": 0.5013769865036011,
      "learning_rate": 0.0001889981909664454,
      "loss": 1.1944,
      "step": 1642
    },
    {
      "epoch": 0.30533358111875114,
      "grad_norm": 0.5568968653678894,
      "learning_rate": 0.0001889848511601442,
      "loss": 1.2865,
      "step": 1643
    },
    {
      "epoch": 0.3055194201821223,
      "grad_norm": 0.5005005598068237,
      "learning_rate": 0.00018897150374277635,
      "loss": 1.0849,
      "step": 1644
    },
    {
      "epoch": 0.3057052592454934,
      "grad_norm": 0.5866604447364807,
      "learning_rate": 0.00018895814871548348,
      "loss": 1.2455,
      "step": 1645
    },
    {
      "epoch": 0.3058910983088645,
      "grad_norm": 0.5890522003173828,
      "learning_rate": 0.0001889447860794079,
      "loss": 0.9689,
      "step": 1646
    },
    {
      "epoch": 0.3060769373722356,
      "grad_norm": 0.5062227845191956,
      "learning_rate": 0.00018893141583569252,
      "loss": 1.1549,
      "step": 1647
    },
    {
      "epoch": 0.3062627764356068,
      "grad_norm": 0.5845811367034912,
      "learning_rate": 0.0001889180379854809,
      "loss": 1.1489,
      "step": 1648
    },
    {
      "epoch": 0.3064486154989779,
      "grad_norm": 0.641802191734314,
      "learning_rate": 0.00018890465252991738,
      "loss": 1.0747,
      "step": 1649
    },
    {
      "epoch": 0.306634454562349,
      "grad_norm": 0.5519110560417175,
      "learning_rate": 0.00018889125947014676,
      "loss": 1.1263,
      "step": 1650
    },
    {
      "epoch": 0.3068202936257201,
      "grad_norm": 0.594745934009552,
      "learning_rate": 0.0001888778588073146,
      "loss": 1.0389,
      "step": 1651
    },
    {
      "epoch": 0.30700613268909127,
      "grad_norm": 0.44811272621154785,
      "learning_rate": 0.00018886445054256705,
      "loss": 0.8878,
      "step": 1652
    },
    {
      "epoch": 0.3071919717524624,
      "grad_norm": 0.6732388734817505,
      "learning_rate": 0.00018885103467705102,
      "loss": 1.1783,
      "step": 1653
    },
    {
      "epoch": 0.3073778108158335,
      "grad_norm": 0.6434346437454224,
      "learning_rate": 0.00018883761121191395,
      "loss": 1.2321,
      "step": 1654
    },
    {
      "epoch": 0.3075636498792046,
      "grad_norm": 0.5560142993927002,
      "learning_rate": 0.000188824180148304,
      "loss": 1.0874,
      "step": 1655
    },
    {
      "epoch": 0.30774948894257576,
      "grad_norm": 0.6674293279647827,
      "learning_rate": 0.00018881074148736993,
      "loss": 1.165,
      "step": 1656
    },
    {
      "epoch": 0.30793532800594686,
      "grad_norm": 0.5190140604972839,
      "learning_rate": 0.00018879729523026123,
      "loss": 0.9781,
      "step": 1657
    },
    {
      "epoch": 0.30812116706931797,
      "grad_norm": 0.5726615786552429,
      "learning_rate": 0.00018878384137812793,
      "loss": 1.1098,
      "step": 1658
    },
    {
      "epoch": 0.3083070061326891,
      "grad_norm": 0.6239640712738037,
      "learning_rate": 0.00018877037993212075,
      "loss": 1.0968,
      "step": 1659
    },
    {
      "epoch": 0.30849284519606024,
      "grad_norm": 0.5197991132736206,
      "learning_rate": 0.00018875691089339117,
      "loss": 0.9939,
      "step": 1660
    },
    {
      "epoch": 0.30867868425943135,
      "grad_norm": 0.5677801966667175,
      "learning_rate": 0.00018874343426309118,
      "loss": 1.4178,
      "step": 1661
    },
    {
      "epoch": 0.30886452332280245,
      "grad_norm": 0.504206120967865,
      "learning_rate": 0.0001887299500423734,
      "loss": 1.0849,
      "step": 1662
    },
    {
      "epoch": 0.30905036238617356,
      "grad_norm": 0.7125438451766968,
      "learning_rate": 0.00018871645823239128,
      "loss": 0.922,
      "step": 1663
    },
    {
      "epoch": 0.30923620144954467,
      "grad_norm": 0.6472058892250061,
      "learning_rate": 0.00018870295883429873,
      "loss": 1.3155,
      "step": 1664
    },
    {
      "epoch": 0.30942204051291583,
      "grad_norm": 0.5385541915893555,
      "learning_rate": 0.0001886894518492504,
      "loss": 0.9653,
      "step": 1665
    },
    {
      "epoch": 0.30960787957628694,
      "grad_norm": 0.5295284986495972,
      "learning_rate": 0.00018867593727840153,
      "loss": 0.9995,
      "step": 1666
    },
    {
      "epoch": 0.30979371863965804,
      "grad_norm": 0.5476770997047424,
      "learning_rate": 0.00018866241512290814,
      "loss": 1.2066,
      "step": 1667
    },
    {
      "epoch": 0.30997955770302915,
      "grad_norm": 0.5373952984809875,
      "learning_rate": 0.00018864888538392676,
      "loss": 1.0628,
      "step": 1668
    },
    {
      "epoch": 0.3101653967664003,
      "grad_norm": 0.5620145797729492,
      "learning_rate": 0.00018863534806261458,
      "loss": 1.1074,
      "step": 1669
    },
    {
      "epoch": 0.3103512358297714,
      "grad_norm": 0.49720069766044617,
      "learning_rate": 0.00018862180316012953,
      "loss": 1.0306,
      "step": 1670
    },
    {
      "epoch": 0.31053707489314253,
      "grad_norm": 0.5404595136642456,
      "learning_rate": 0.00018860825067763015,
      "loss": 1.0686,
      "step": 1671
    },
    {
      "epoch": 0.31072291395651364,
      "grad_norm": 0.4632123112678528,
      "learning_rate": 0.00018859469061627556,
      "loss": 0.9225,
      "step": 1672
    },
    {
      "epoch": 0.3109087530198848,
      "grad_norm": 0.6441605687141418,
      "learning_rate": 0.0001885811229772256,
      "loss": 1.3782,
      "step": 1673
    },
    {
      "epoch": 0.3110945920832559,
      "grad_norm": 0.5591496229171753,
      "learning_rate": 0.0001885675477616407,
      "loss": 0.9357,
      "step": 1674
    },
    {
      "epoch": 0.311280431146627,
      "grad_norm": 0.5491276979446411,
      "learning_rate": 0.0001885539649706821,
      "loss": 0.9397,
      "step": 1675
    },
    {
      "epoch": 0.3114662702099981,
      "grad_norm": 0.5409485697746277,
      "learning_rate": 0.00018854037460551144,
      "loss": 0.9025,
      "step": 1676
    },
    {
      "epoch": 0.3116521092733693,
      "grad_norm": 0.5407100319862366,
      "learning_rate": 0.00018852677666729117,
      "loss": 1.1127,
      "step": 1677
    },
    {
      "epoch": 0.3118379483367404,
      "grad_norm": 0.6072363257408142,
      "learning_rate": 0.0001885131711571844,
      "loss": 1.1601,
      "step": 1678
    },
    {
      "epoch": 0.3120237874001115,
      "grad_norm": 0.4891136586666107,
      "learning_rate": 0.00018849955807635475,
      "loss": 1.0007,
      "step": 1679
    },
    {
      "epoch": 0.3122096264634826,
      "grad_norm": 0.7494361400604248,
      "learning_rate": 0.00018848593742596664,
      "loss": 1.3113,
      "step": 1680
    },
    {
      "epoch": 0.31239546552685377,
      "grad_norm": 0.6173309087753296,
      "learning_rate": 0.00018847230920718508,
      "loss": 0.9148,
      "step": 1681
    },
    {
      "epoch": 0.3125813045902249,
      "grad_norm": 0.6358309388160706,
      "learning_rate": 0.0001884586734211757,
      "loss": 1.104,
      "step": 1682
    },
    {
      "epoch": 0.312767143653596,
      "grad_norm": 0.47288426756858826,
      "learning_rate": 0.0001884450300691048,
      "loss": 0.8665,
      "step": 1683
    },
    {
      "epoch": 0.3129529827169671,
      "grad_norm": 0.643383264541626,
      "learning_rate": 0.00018843137915213932,
      "loss": 1.2428,
      "step": 1684
    },
    {
      "epoch": 0.31313882178033825,
      "grad_norm": 0.42575201392173767,
      "learning_rate": 0.0001884177206714468,
      "loss": 0.865,
      "step": 1685
    },
    {
      "epoch": 0.31332466084370936,
      "grad_norm": 0.662293553352356,
      "learning_rate": 0.00018840405462819558,
      "loss": 1.2715,
      "step": 1686
    },
    {
      "epoch": 0.31351049990708046,
      "grad_norm": 0.6092833280563354,
      "learning_rate": 0.0001883903810235545,
      "loss": 1.3039,
      "step": 1687
    },
    {
      "epoch": 0.31369633897045157,
      "grad_norm": 0.5167418718338013,
      "learning_rate": 0.0001883766998586931,
      "loss": 1.0148,
      "step": 1688
    },
    {
      "epoch": 0.31388217803382273,
      "grad_norm": 0.580491840839386,
      "learning_rate": 0.00018836301113478156,
      "loss": 1.1673,
      "step": 1689
    },
    {
      "epoch": 0.31406801709719384,
      "grad_norm": 0.6332157850265503,
      "learning_rate": 0.0001883493148529907,
      "loss": 1.3328,
      "step": 1690
    },
    {
      "epoch": 0.31425385616056495,
      "grad_norm": 0.4925262928009033,
      "learning_rate": 0.000188335611014492,
      "loss": 1.0126,
      "step": 1691
    },
    {
      "epoch": 0.31443969522393606,
      "grad_norm": 0.4704776704311371,
      "learning_rate": 0.00018832189962045753,
      "loss": 0.923,
      "step": 1692
    },
    {
      "epoch": 0.3146255342873072,
      "grad_norm": 0.6361016631126404,
      "learning_rate": 0.00018830818067206012,
      "loss": 0.9724,
      "step": 1693
    },
    {
      "epoch": 0.3148113733506783,
      "grad_norm": 0.5294798016548157,
      "learning_rate": 0.00018829445417047316,
      "loss": 1.1158,
      "step": 1694
    },
    {
      "epoch": 0.31499721241404943,
      "grad_norm": 0.4817502498626709,
      "learning_rate": 0.00018828072011687074,
      "loss": 1.0515,
      "step": 1695
    },
    {
      "epoch": 0.31518305147742054,
      "grad_norm": 0.5335044264793396,
      "learning_rate": 0.00018826697851242747,
      "loss": 1.0616,
      "step": 1696
    },
    {
      "epoch": 0.3153688905407917,
      "grad_norm": 0.4775748550891876,
      "learning_rate": 0.00018825322935831878,
      "loss": 0.881,
      "step": 1697
    },
    {
      "epoch": 0.3155547296041628,
      "grad_norm": 0.5682370066642761,
      "learning_rate": 0.00018823947265572065,
      "loss": 0.9613,
      "step": 1698
    },
    {
      "epoch": 0.3157405686675339,
      "grad_norm": 0.5231326818466187,
      "learning_rate": 0.00018822570840580972,
      "loss": 0.9672,
      "step": 1699
    },
    {
      "epoch": 0.315926407730905,
      "grad_norm": 0.6463721394538879,
      "learning_rate": 0.00018821193660976325,
      "loss": 1.0961,
      "step": 1700
    },
    {
      "epoch": 0.31611224679427613,
      "grad_norm": 0.5318092703819275,
      "learning_rate": 0.00018819815726875918,
      "loss": 1.017,
      "step": 1701
    },
    {
      "epoch": 0.3162980858576473,
      "grad_norm": 0.5542148947715759,
      "learning_rate": 0.0001881843703839761,
      "loss": 0.9725,
      "step": 1702
    },
    {
      "epoch": 0.3164839249210184,
      "grad_norm": 0.606061577796936,
      "learning_rate": 0.00018817057595659327,
      "loss": 1.1379,
      "step": 1703
    },
    {
      "epoch": 0.3166697639843895,
      "grad_norm": 0.5355798006057739,
      "learning_rate": 0.00018815677398779048,
      "loss": 0.9458,
      "step": 1704
    },
    {
      "epoch": 0.3168556030477606,
      "grad_norm": 0.534319281578064,
      "learning_rate": 0.00018814296447874828,
      "loss": 0.8033,
      "step": 1705
    },
    {
      "epoch": 0.3170414421111318,
      "grad_norm": 0.5069194436073303,
      "learning_rate": 0.00018812914743064784,
      "loss": 1.087,
      "step": 1706
    },
    {
      "epoch": 0.3172272811745029,
      "grad_norm": 0.5822552442550659,
      "learning_rate": 0.00018811532284467096,
      "loss": 1.1817,
      "step": 1707
    },
    {
      "epoch": 0.317413120237874,
      "grad_norm": 0.46280261874198914,
      "learning_rate": 0.00018810149072200004,
      "loss": 0.9893,
      "step": 1708
    },
    {
      "epoch": 0.3175989593012451,
      "grad_norm": 0.6027824282646179,
      "learning_rate": 0.00018808765106381823,
      "loss": 1.0487,
      "step": 1709
    },
    {
      "epoch": 0.31778479836461626,
      "grad_norm": 0.5475772023200989,
      "learning_rate": 0.00018807380387130923,
      "loss": 1.216,
      "step": 1710
    },
    {
      "epoch": 0.31797063742798737,
      "grad_norm": 0.5941911935806274,
      "learning_rate": 0.00018805994914565744,
      "loss": 1.4315,
      "step": 1711
    },
    {
      "epoch": 0.3181564764913585,
      "grad_norm": 0.6065940856933594,
      "learning_rate": 0.0001880460868880479,
      "loss": 1.0772,
      "step": 1712
    },
    {
      "epoch": 0.3183423155547296,
      "grad_norm": 0.6566287875175476,
      "learning_rate": 0.00018803221709966624,
      "loss": 1.2141,
      "step": 1713
    },
    {
      "epoch": 0.31852815461810075,
      "grad_norm": 0.5035297870635986,
      "learning_rate": 0.00018801833978169876,
      "loss": 1.3613,
      "step": 1714
    },
    {
      "epoch": 0.31871399368147185,
      "grad_norm": 0.5885191559791565,
      "learning_rate": 0.00018800445493533246,
      "loss": 1.1476,
      "step": 1715
    },
    {
      "epoch": 0.31889983274484296,
      "grad_norm": 0.6840378046035767,
      "learning_rate": 0.00018799056256175493,
      "loss": 1.1244,
      "step": 1716
    },
    {
      "epoch": 0.31908567180821407,
      "grad_norm": 0.5499553084373474,
      "learning_rate": 0.00018797666266215441,
      "loss": 1.0127,
      "step": 1717
    },
    {
      "epoch": 0.31927151087158523,
      "grad_norm": 0.5777943730354309,
      "learning_rate": 0.0001879627552377198,
      "loss": 1.15,
      "step": 1718
    },
    {
      "epoch": 0.31945734993495634,
      "grad_norm": 0.5416708588600159,
      "learning_rate": 0.00018794884028964064,
      "loss": 1.0631,
      "step": 1719
    },
    {
      "epoch": 0.31964318899832744,
      "grad_norm": 0.6631436944007874,
      "learning_rate": 0.00018793491781910705,
      "loss": 1.1563,
      "step": 1720
    },
    {
      "epoch": 0.31982902806169855,
      "grad_norm": 0.658031702041626,
      "learning_rate": 0.00018792098782730992,
      "loss": 1.1838,
      "step": 1721
    },
    {
      "epoch": 0.3200148671250697,
      "grad_norm": 0.5618680715560913,
      "learning_rate": 0.00018790705031544063,
      "loss": 0.9808,
      "step": 1722
    },
    {
      "epoch": 0.3202007061884408,
      "grad_norm": 0.5322409868240356,
      "learning_rate": 0.00018789310528469138,
      "loss": 0.8012,
      "step": 1723
    },
    {
      "epoch": 0.3203865452518119,
      "grad_norm": 0.5504143238067627,
      "learning_rate": 0.00018787915273625486,
      "loss": 1.1341,
      "step": 1724
    },
    {
      "epoch": 0.32057238431518303,
      "grad_norm": 0.5948871374130249,
      "learning_rate": 0.00018786519267132446,
      "loss": 0.9778,
      "step": 1725
    },
    {
      "epoch": 0.3207582233785542,
      "grad_norm": 0.5112269520759583,
      "learning_rate": 0.00018785122509109426,
      "loss": 1.0508,
      "step": 1726
    },
    {
      "epoch": 0.3209440624419253,
      "grad_norm": 0.4630531072616577,
      "learning_rate": 0.0001878372499967589,
      "loss": 0.9361,
      "step": 1727
    },
    {
      "epoch": 0.3211299015052964,
      "grad_norm": 0.6956179141998291,
      "learning_rate": 0.00018782326738951368,
      "loss": 1.3476,
      "step": 1728
    },
    {
      "epoch": 0.3213157405686675,
      "grad_norm": 0.6133003234863281,
      "learning_rate": 0.0001878092772705546,
      "loss": 1.2831,
      "step": 1729
    },
    {
      "epoch": 0.3215015796320387,
      "grad_norm": 0.5532038807868958,
      "learning_rate": 0.00018779527964107824,
      "loss": 0.9402,
      "step": 1730
    },
    {
      "epoch": 0.3216874186954098,
      "grad_norm": 0.5056849122047424,
      "learning_rate": 0.0001877812745022819,
      "loss": 0.8827,
      "step": 1731
    },
    {
      "epoch": 0.3218732577587809,
      "grad_norm": 0.5501548051834106,
      "learning_rate": 0.0001877672618553634,
      "loss": 1.0141,
      "step": 1732
    },
    {
      "epoch": 0.322059096822152,
      "grad_norm": 0.5648378729820251,
      "learning_rate": 0.00018775324170152132,
      "loss": 0.812,
      "step": 1733
    },
    {
      "epoch": 0.3222449358855231,
      "grad_norm": 0.6281271576881409,
      "learning_rate": 0.00018773921404195481,
      "loss": 1.3273,
      "step": 1734
    },
    {
      "epoch": 0.32243077494889427,
      "grad_norm": 0.6104331016540527,
      "learning_rate": 0.0001877251788778637,
      "loss": 1.0296,
      "step": 1735
    },
    {
      "epoch": 0.3226166140122654,
      "grad_norm": 0.6585345268249512,
      "learning_rate": 0.00018771113621044843,
      "loss": 1.1113,
      "step": 1736
    },
    {
      "epoch": 0.3228024530756365,
      "grad_norm": 0.5706427693367004,
      "learning_rate": 0.00018769708604091012,
      "loss": 1.0446,
      "step": 1737
    },
    {
      "epoch": 0.3229882921390076,
      "grad_norm": 0.6643130779266357,
      "learning_rate": 0.00018768302837045048,
      "loss": 0.883,
      "step": 1738
    },
    {
      "epoch": 0.32317413120237876,
      "grad_norm": 0.4990970492362976,
      "learning_rate": 0.00018766896320027195,
      "loss": 0.9353,
      "step": 1739
    },
    {
      "epoch": 0.32335997026574986,
      "grad_norm": 0.48843303322792053,
      "learning_rate": 0.0001876548905315775,
      "loss": 0.9064,
      "step": 1740
    },
    {
      "epoch": 0.32354580932912097,
      "grad_norm": 0.6881991028785706,
      "learning_rate": 0.00018764081036557083,
      "loss": 1.3455,
      "step": 1741
    },
    {
      "epoch": 0.3237316483924921,
      "grad_norm": 0.5047183036804199,
      "learning_rate": 0.00018762672270345623,
      "loss": 0.874,
      "step": 1742
    },
    {
      "epoch": 0.32391748745586324,
      "grad_norm": 0.5843253135681152,
      "learning_rate": 0.00018761262754643862,
      "loss": 1.3657,
      "step": 1743
    },
    {
      "epoch": 0.32410332651923435,
      "grad_norm": 0.743747889995575,
      "learning_rate": 0.00018759852489572368,
      "loss": 1.2149,
      "step": 1744
    },
    {
      "epoch": 0.32428916558260545,
      "grad_norm": 0.5197893381118774,
      "learning_rate": 0.00018758441475251754,
      "loss": 1.038,
      "step": 1745
    },
    {
      "epoch": 0.32447500464597656,
      "grad_norm": 0.5472385287284851,
      "learning_rate": 0.0001875702971180271,
      "loss": 0.9697,
      "step": 1746
    },
    {
      "epoch": 0.3246608437093477,
      "grad_norm": 0.5840916037559509,
      "learning_rate": 0.00018755617199345988,
      "loss": 1.0975,
      "step": 1747
    },
    {
      "epoch": 0.32484668277271883,
      "grad_norm": 0.5640180706977844,
      "learning_rate": 0.00018754203938002408,
      "loss": 0.9969,
      "step": 1748
    },
    {
      "epoch": 0.32503252183608994,
      "grad_norm": 0.5140122771263123,
      "learning_rate": 0.00018752789927892845,
      "loss": 1.0121,
      "step": 1749
    },
    {
      "epoch": 0.32521836089946105,
      "grad_norm": 0.595289409160614,
      "learning_rate": 0.0001875137516913824,
      "loss": 1.228,
      "step": 1750
    },
    {
      "epoch": 0.3254041999628322,
      "grad_norm": 0.519131600856781,
      "learning_rate": 0.000187499596618596,
      "loss": 1.2358,
      "step": 1751
    },
    {
      "epoch": 0.3255900390262033,
      "grad_norm": 0.617352306842804,
      "learning_rate": 0.00018748543406178002,
      "loss": 0.8571,
      "step": 1752
    },
    {
      "epoch": 0.3257758780895744,
      "grad_norm": 0.5419536828994751,
      "learning_rate": 0.00018747126402214576,
      "loss": 0.9787,
      "step": 1753
    },
    {
      "epoch": 0.32596171715294553,
      "grad_norm": 0.5365861654281616,
      "learning_rate": 0.00018745708650090527,
      "loss": 1.0241,
      "step": 1754
    },
    {
      "epoch": 0.3261475562163167,
      "grad_norm": 0.5210004448890686,
      "learning_rate": 0.00018744290149927113,
      "loss": 1.1246,
      "step": 1755
    },
    {
      "epoch": 0.3263333952796878,
      "grad_norm": 0.6406294107437134,
      "learning_rate": 0.00018742870901845664,
      "loss": 1.1412,
      "step": 1756
    },
    {
      "epoch": 0.3265192343430589,
      "grad_norm": 0.6498095393180847,
      "learning_rate": 0.0001874145090596757,
      "loss": 1.1282,
      "step": 1757
    },
    {
      "epoch": 0.32670507340643,
      "grad_norm": 0.5301967859268188,
      "learning_rate": 0.00018740030162414287,
      "loss": 1.0865,
      "step": 1758
    },
    {
      "epoch": 0.3268909124698012,
      "grad_norm": 0.6150500774383545,
      "learning_rate": 0.00018738608671307338,
      "loss": 1.1464,
      "step": 1759
    },
    {
      "epoch": 0.3270767515331723,
      "grad_norm": 0.5350655317306519,
      "learning_rate": 0.000187371864327683,
      "loss": 0.9322,
      "step": 1760
    },
    {
      "epoch": 0.3272625905965434,
      "grad_norm": 0.5245082378387451,
      "learning_rate": 0.00018735763446918826,
      "loss": 1.1232,
      "step": 1761
    },
    {
      "epoch": 0.3274484296599145,
      "grad_norm": 0.5193549394607544,
      "learning_rate": 0.0001873433971388062,
      "loss": 0.927,
      "step": 1762
    },
    {
      "epoch": 0.32763426872328566,
      "grad_norm": 0.5565290451049805,
      "learning_rate": 0.0001873291523377546,
      "loss": 1.1293,
      "step": 1763
    },
    {
      "epoch": 0.32782010778665677,
      "grad_norm": 0.5595194101333618,
      "learning_rate": 0.0001873149000672519,
      "loss": 1.1835,
      "step": 1764
    },
    {
      "epoch": 0.3280059468500279,
      "grad_norm": 0.5816195011138916,
      "learning_rate": 0.00018730064032851707,
      "loss": 1.0722,
      "step": 1765
    },
    {
      "epoch": 0.328191785913399,
      "grad_norm": 0.5451340675354004,
      "learning_rate": 0.00018728637312276976,
      "loss": 1.021,
      "step": 1766
    },
    {
      "epoch": 0.32837762497677014,
      "grad_norm": 0.3959139883518219,
      "learning_rate": 0.00018727209845123034,
      "loss": 0.7137,
      "step": 1767
    },
    {
      "epoch": 0.32856346404014125,
      "grad_norm": 0.6754137873649597,
      "learning_rate": 0.00018725781631511975,
      "loss": 1.3023,
      "step": 1768
    },
    {
      "epoch": 0.32874930310351236,
      "grad_norm": 0.6047104597091675,
      "learning_rate": 0.00018724352671565947,
      "loss": 1.1471,
      "step": 1769
    },
    {
      "epoch": 0.32893514216688347,
      "grad_norm": 0.60149085521698,
      "learning_rate": 0.00018722922965407182,
      "loss": 1.2296,
      "step": 1770
    },
    {
      "epoch": 0.3291209812302546,
      "grad_norm": 0.5902987718582153,
      "learning_rate": 0.00018721492513157966,
      "loss": 0.9665,
      "step": 1771
    },
    {
      "epoch": 0.32930682029362573,
      "grad_norm": 0.6506375074386597,
      "learning_rate": 0.00018720061314940643,
      "loss": 0.846,
      "step": 1772
    },
    {
      "epoch": 0.32949265935699684,
      "grad_norm": 0.5661017894744873,
      "learning_rate": 0.00018718629370877626,
      "loss": 1.0438,
      "step": 1773
    },
    {
      "epoch": 0.32967849842036795,
      "grad_norm": 0.6380808353424072,
      "learning_rate": 0.00018717196681091402,
      "loss": 0.9613,
      "step": 1774
    },
    {
      "epoch": 0.32986433748373906,
      "grad_norm": 0.6345396637916565,
      "learning_rate": 0.00018715763245704504,
      "loss": 1.1511,
      "step": 1775
    },
    {
      "epoch": 0.3300501765471102,
      "grad_norm": 0.6821529865264893,
      "learning_rate": 0.00018714329064839532,
      "loss": 0.9459,
      "step": 1776
    },
    {
      "epoch": 0.3302360156104813,
      "grad_norm": 0.4994625449180603,
      "learning_rate": 0.00018712894138619168,
      "loss": 0.9316,
      "step": 1777
    },
    {
      "epoch": 0.33042185467385243,
      "grad_norm": 0.579918384552002,
      "learning_rate": 0.00018711458467166137,
      "loss": 0.7732,
      "step": 1778
    },
    {
      "epoch": 0.33060769373722354,
      "grad_norm": 0.602846622467041,
      "learning_rate": 0.00018710022050603232,
      "loss": 1.2335,
      "step": 1779
    },
    {
      "epoch": 0.3307935328005947,
      "grad_norm": 0.5414668917655945,
      "learning_rate": 0.00018708584889053317,
      "loss": 0.8378,
      "step": 1780
    },
    {
      "epoch": 0.3309793718639658,
      "grad_norm": 0.5418664813041687,
      "learning_rate": 0.00018707146982639317,
      "loss": 1.2916,
      "step": 1781
    },
    {
      "epoch": 0.3311652109273369,
      "grad_norm": 0.5478229522705078,
      "learning_rate": 0.00018705708331484213,
      "loss": 1.0765,
      "step": 1782
    },
    {
      "epoch": 0.331351049990708,
      "grad_norm": 0.6811211705207825,
      "learning_rate": 0.00018704268935711062,
      "loss": 1.1639,
      "step": 1783
    },
    {
      "epoch": 0.3315368890540792,
      "grad_norm": 0.676049530506134,
      "learning_rate": 0.0001870282879544298,
      "loss": 1.1275,
      "step": 1784
    },
    {
      "epoch": 0.3317227281174503,
      "grad_norm": 0.4629632234573364,
      "learning_rate": 0.00018701387910803136,
      "loss": 0.7645,
      "step": 1785
    },
    {
      "epoch": 0.3319085671808214,
      "grad_norm": 0.5445986390113831,
      "learning_rate": 0.0001869994628191478,
      "loss": 1.0727,
      "step": 1786
    },
    {
      "epoch": 0.3320944062441925,
      "grad_norm": 0.5960368514060974,
      "learning_rate": 0.00018698503908901216,
      "loss": 0.9489,
      "step": 1787
    },
    {
      "epoch": 0.33228024530756367,
      "grad_norm": 0.536932647228241,
      "learning_rate": 0.00018697060791885814,
      "loss": 0.9583,
      "step": 1788
    },
    {
      "epoch": 0.3324660843709348,
      "grad_norm": 0.7085352540016174,
      "learning_rate": 0.00018695616930992004,
      "loss": 1.3772,
      "step": 1789
    },
    {
      "epoch": 0.3326519234343059,
      "grad_norm": 0.5107738971710205,
      "learning_rate": 0.00018694172326343285,
      "loss": 1.1724,
      "step": 1790
    },
    {
      "epoch": 0.332837762497677,
      "grad_norm": 0.6691588759422302,
      "learning_rate": 0.00018692726978063214,
      "loss": 1.1536,
      "step": 1791
    },
    {
      "epoch": 0.33302360156104815,
      "grad_norm": 0.5077880620956421,
      "learning_rate": 0.0001869128088627542,
      "loss": 0.8661,
      "step": 1792
    },
    {
      "epoch": 0.33320944062441926,
      "grad_norm": 0.655665934085846,
      "learning_rate": 0.00018689834051103588,
      "loss": 1.1782,
      "step": 1793
    },
    {
      "epoch": 0.33339527968779037,
      "grad_norm": 0.5770731568336487,
      "learning_rate": 0.00018688386472671465,
      "loss": 0.942,
      "step": 1794
    },
    {
      "epoch": 0.3335811187511615,
      "grad_norm": 0.531545877456665,
      "learning_rate": 0.00018686938151102868,
      "loss": 1.1079,
      "step": 1795
    },
    {
      "epoch": 0.33376695781453264,
      "grad_norm": 0.59685879945755,
      "learning_rate": 0.0001868548908652168,
      "loss": 0.9696,
      "step": 1796
    },
    {
      "epoch": 0.33395279687790375,
      "grad_norm": 0.597497284412384,
      "learning_rate": 0.00018684039279051837,
      "loss": 0.987,
      "step": 1797
    },
    {
      "epoch": 0.33413863594127485,
      "grad_norm": 0.5863851308822632,
      "learning_rate": 0.00018682588728817346,
      "loss": 0.9386,
      "step": 1798
    },
    {
      "epoch": 0.33432447500464596,
      "grad_norm": 0.557304859161377,
      "learning_rate": 0.00018681137435942277,
      "loss": 0.9928,
      "step": 1799
    },
    {
      "epoch": 0.3345103140680171,
      "grad_norm": 0.49487656354904175,
      "learning_rate": 0.00018679685400550757,
      "loss": 1.1482,
      "step": 1800
    },
    {
      "epoch": 0.33469615313138823,
      "grad_norm": 0.5904025435447693,
      "learning_rate": 0.0001867823262276699,
      "loss": 0.878,
      "step": 1801
    },
    {
      "epoch": 0.33488199219475934,
      "grad_norm": 0.49566707015037537,
      "learning_rate": 0.00018676779102715226,
      "loss": 1.2314,
      "step": 1802
    },
    {
      "epoch": 0.33506783125813044,
      "grad_norm": 0.5034186244010925,
      "learning_rate": 0.00018675324840519797,
      "loss": 0.8979,
      "step": 1803
    },
    {
      "epoch": 0.3352536703215016,
      "grad_norm": 0.5835298299789429,
      "learning_rate": 0.0001867386983630508,
      "loss": 1.2786,
      "step": 1804
    },
    {
      "epoch": 0.3354395093848727,
      "grad_norm": 0.602591872215271,
      "learning_rate": 0.00018672414090195534,
      "loss": 1.2119,
      "step": 1805
    },
    {
      "epoch": 0.3356253484482438,
      "grad_norm": 0.5360019207000732,
      "learning_rate": 0.00018670957602315663,
      "loss": 1.2997,
      "step": 1806
    },
    {
      "epoch": 0.33581118751161493,
      "grad_norm": 0.7245522141456604,
      "learning_rate": 0.0001866950037279005,
      "loss": 1.1139,
      "step": 1807
    },
    {
      "epoch": 0.33599702657498604,
      "grad_norm": 0.5866603851318359,
      "learning_rate": 0.00018668042401743333,
      "loss": 1.4327,
      "step": 1808
    },
    {
      "epoch": 0.3361828656383572,
      "grad_norm": 0.5497808456420898,
      "learning_rate": 0.00018666583689300214,
      "loss": 1.2068,
      "step": 1809
    },
    {
      "epoch": 0.3363687047017283,
      "grad_norm": 0.49823257327079773,
      "learning_rate": 0.00018665124235585462,
      "loss": 0.9279,
      "step": 1810
    },
    {
      "epoch": 0.3365545437650994,
      "grad_norm": 0.5769509673118591,
      "learning_rate": 0.00018663664040723905,
      "loss": 1.0357,
      "step": 1811
    },
    {
      "epoch": 0.3367403828284705,
      "grad_norm": 0.3853382468223572,
      "learning_rate": 0.0001866220310484044,
      "loss": 0.6868,
      "step": 1812
    },
    {
      "epoch": 0.3369262218918417,
      "grad_norm": 0.5640473961830139,
      "learning_rate": 0.0001866074142806002,
      "loss": 1.27,
      "step": 1813
    },
    {
      "epoch": 0.3371120609552128,
      "grad_norm": 0.5786116719245911,
      "learning_rate": 0.0001865927901050767,
      "loss": 1.2678,
      "step": 1814
    },
    {
      "epoch": 0.3372979000185839,
      "grad_norm": 0.5291271209716797,
      "learning_rate": 0.0001865781585230847,
      "loss": 0.9834,
      "step": 1815
    },
    {
      "epoch": 0.337483739081955,
      "grad_norm": 0.6278622150421143,
      "learning_rate": 0.00018656351953587566,
      "loss": 1.0949,
      "step": 1816
    },
    {
      "epoch": 0.33766957814532617,
      "grad_norm": 0.5260774493217468,
      "learning_rate": 0.00018654887314470173,
      "loss": 1.1183,
      "step": 1817
    },
    {
      "epoch": 0.3378554172086973,
      "grad_norm": 0.5557003021240234,
      "learning_rate": 0.00018653421935081561,
      "loss": 0.7845,
      "step": 1818
    },
    {
      "epoch": 0.3380412562720684,
      "grad_norm": 0.6177703142166138,
      "learning_rate": 0.0001865195581554707,
      "loss": 1.0623,
      "step": 1819
    },
    {
      "epoch": 0.3382270953354395,
      "grad_norm": 0.6486766338348389,
      "learning_rate": 0.000186504889559921,
      "loss": 1.2507,
      "step": 1820
    },
    {
      "epoch": 0.33841293439881065,
      "grad_norm": 0.5549860596656799,
      "learning_rate": 0.00018649021356542108,
      "loss": 1.0893,
      "step": 1821
    },
    {
      "epoch": 0.33859877346218176,
      "grad_norm": 0.48546648025512695,
      "learning_rate": 0.0001864755301732263,
      "loss": 0.704,
      "step": 1822
    },
    {
      "epoch": 0.33878461252555286,
      "grad_norm": 2.3476552963256836,
      "learning_rate": 0.0001864608393845925,
      "loss": 2.9748,
      "step": 1823
    },
    {
      "epoch": 0.33897045158892397,
      "grad_norm": 0.5599921941757202,
      "learning_rate": 0.00018644614120077625,
      "loss": 1.0073,
      "step": 1824
    },
    {
      "epoch": 0.33915629065229513,
      "grad_norm": 0.5544316172599792,
      "learning_rate": 0.00018643143562303475,
      "loss": 0.9475,
      "step": 1825
    },
    {
      "epoch": 0.33934212971566624,
      "grad_norm": 0.5891472697257996,
      "learning_rate": 0.0001864167226526257,
      "loss": 1.0265,
      "step": 1826
    },
    {
      "epoch": 0.33952796877903735,
      "grad_norm": 0.6864466667175293,
      "learning_rate": 0.00018640200229080763,
      "loss": 1.0996,
      "step": 1827
    },
    {
      "epoch": 0.33971380784240846,
      "grad_norm": 0.5325470566749573,
      "learning_rate": 0.00018638727453883957,
      "loss": 1.1052,
      "step": 1828
    },
    {
      "epoch": 0.3398996469057796,
      "grad_norm": 0.5583032369613647,
      "learning_rate": 0.0001863725393979812,
      "loss": 1.1012,
      "step": 1829
    },
    {
      "epoch": 0.3400854859691507,
      "grad_norm": 0.6103065013885498,
      "learning_rate": 0.00018635779686949284,
      "loss": 1.0018,
      "step": 1830
    },
    {
      "epoch": 0.34027132503252183,
      "grad_norm": 0.7236659526824951,
      "learning_rate": 0.0001863430469546355,
      "loss": 1.2731,
      "step": 1831
    },
    {
      "epoch": 0.34045716409589294,
      "grad_norm": 0.5036203861236572,
      "learning_rate": 0.0001863282896546707,
      "loss": 0.9995,
      "step": 1832
    },
    {
      "epoch": 0.3406430031592641,
      "grad_norm": 0.585512638092041,
      "learning_rate": 0.00018631352497086074,
      "loss": 1.1056,
      "step": 1833
    },
    {
      "epoch": 0.3408288422226352,
      "grad_norm": 0.6121904253959656,
      "learning_rate": 0.00018629875290446845,
      "loss": 1.114,
      "step": 1834
    },
    {
      "epoch": 0.3410146812860063,
      "grad_norm": 0.5517188310623169,
      "learning_rate": 0.00018628397345675727,
      "loss": 1.3077,
      "step": 1835
    },
    {
      "epoch": 0.3412005203493774,
      "grad_norm": 0.5181845426559448,
      "learning_rate": 0.0001862691866289914,
      "loss": 1.193,
      "step": 1836
    },
    {
      "epoch": 0.3413863594127486,
      "grad_norm": 0.6143492460250854,
      "learning_rate": 0.00018625439242243548,
      "loss": 0.8677,
      "step": 1837
    },
    {
      "epoch": 0.3415721984761197,
      "grad_norm": 0.7705954313278198,
      "learning_rate": 0.00018623959083835504,
      "loss": 0.9832,
      "step": 1838
    },
    {
      "epoch": 0.3417580375394908,
      "grad_norm": 0.5456838011741638,
      "learning_rate": 0.00018622478187801594,
      "loss": 1.258,
      "step": 1839
    },
    {
      "epoch": 0.3419438766028619,
      "grad_norm": 0.4819016456604004,
      "learning_rate": 0.00018620996554268493,
      "loss": 0.9829,
      "step": 1840
    },
    {
      "epoch": 0.34212971566623307,
      "grad_norm": 0.5512806177139282,
      "learning_rate": 0.00018619514183362923,
      "loss": 1.0721,
      "step": 1841
    },
    {
      "epoch": 0.3423155547296042,
      "grad_norm": 0.6261305809020996,
      "learning_rate": 0.00018618031075211676,
      "loss": 0.9484,
      "step": 1842
    },
    {
      "epoch": 0.3425013937929753,
      "grad_norm": 0.5673807263374329,
      "learning_rate": 0.00018616547229941604,
      "loss": 1.1987,
      "step": 1843
    },
    {
      "epoch": 0.3426872328563464,
      "grad_norm": 0.4913477599620819,
      "learning_rate": 0.00018615062647679627,
      "loss": 0.8978,
      "step": 1844
    },
    {
      "epoch": 0.3428730719197175,
      "grad_norm": 0.566608190536499,
      "learning_rate": 0.0001861357732855272,
      "loss": 0.9981,
      "step": 1845
    },
    {
      "epoch": 0.34305891098308866,
      "grad_norm": 0.6863612532615662,
      "learning_rate": 0.00018612091272687925,
      "loss": 1.07,
      "step": 1846
    },
    {
      "epoch": 0.34324475004645977,
      "grad_norm": 0.5646424293518066,
      "learning_rate": 0.00018610604480212356,
      "loss": 1.1925,
      "step": 1847
    },
    {
      "epoch": 0.3434305891098309,
      "grad_norm": 0.6170990467071533,
      "learning_rate": 0.00018609116951253171,
      "loss": 1.0215,
      "step": 1848
    },
    {
      "epoch": 0.343616428173202,
      "grad_norm": 0.6480187177658081,
      "learning_rate": 0.0001860762868593761,
      "loss": 1.1675,
      "step": 1849
    },
    {
      "epoch": 0.34380226723657314,
      "grad_norm": 0.5635117292404175,
      "learning_rate": 0.00018606139684392963,
      "loss": 1.1116,
      "step": 1850
    },
    {
      "epoch": 0.34398810629994425,
      "grad_norm": 0.7464002966880798,
      "learning_rate": 0.00018604649946746585,
      "loss": 1.1706,
      "step": 1851
    },
    {
      "epoch": 0.34417394536331536,
      "grad_norm": 0.5671253204345703,
      "learning_rate": 0.00018603159473125905,
      "loss": 1.0751,
      "step": 1852
    },
    {
      "epoch": 0.34435978442668647,
      "grad_norm": 0.5876680612564087,
      "learning_rate": 0.00018601668263658398,
      "loss": 1.3353,
      "step": 1853
    },
    {
      "epoch": 0.34454562349005763,
      "grad_norm": 0.5108693838119507,
      "learning_rate": 0.00018600176318471617,
      "loss": 1.1154,
      "step": 1854
    },
    {
      "epoch": 0.34473146255342874,
      "grad_norm": 0.5718084573745728,
      "learning_rate": 0.00018598683637693165,
      "loss": 1.111,
      "step": 1855
    },
    {
      "epoch": 0.34491730161679984,
      "grad_norm": 0.599010705947876,
      "learning_rate": 0.0001859719022145072,
      "loss": 0.9496,
      "step": 1856
    },
    {
      "epoch": 0.34510314068017095,
      "grad_norm": 0.5230289697647095,
      "learning_rate": 0.00018595696069872013,
      "loss": 1.0775,
      "step": 1857
    },
    {
      "epoch": 0.3452889797435421,
      "grad_norm": 0.5170990824699402,
      "learning_rate": 0.00018594201183084842,
      "loss": 1.092,
      "step": 1858
    },
    {
      "epoch": 0.3454748188069132,
      "grad_norm": 0.57718425989151,
      "learning_rate": 0.00018592705561217072,
      "loss": 0.899,
      "step": 1859
    },
    {
      "epoch": 0.3456606578702843,
      "grad_norm": 0.5379887223243713,
      "learning_rate": 0.00018591209204396622,
      "loss": 1.1148,
      "step": 1860
    },
    {
      "epoch": 0.34584649693365543,
      "grad_norm": 0.604630172252655,
      "learning_rate": 0.0001858971211275148,
      "loss": 1.1184,
      "step": 1861
    },
    {
      "epoch": 0.3460323359970266,
      "grad_norm": 0.6093925833702087,
      "learning_rate": 0.00018588214286409697,
      "loss": 0.8449,
      "step": 1862
    },
    {
      "epoch": 0.3462181750603977,
      "grad_norm": 0.6406834721565247,
      "learning_rate": 0.00018586715725499387,
      "loss": 1.1169,
      "step": 1863
    },
    {
      "epoch": 0.3464040141237688,
      "grad_norm": 0.49252834916114807,
      "learning_rate": 0.00018585216430148723,
      "loss": 0.7095,
      "step": 1864
    },
    {
      "epoch": 0.3465898531871399,
      "grad_norm": 0.6328451037406921,
      "learning_rate": 0.00018583716400485937,
      "loss": 1.2131,
      "step": 1865
    },
    {
      "epoch": 0.3467756922505111,
      "grad_norm": 0.6539022326469421,
      "learning_rate": 0.0001858221563663934,
      "loss": 0.9528,
      "step": 1866
    },
    {
      "epoch": 0.3469615313138822,
      "grad_norm": 0.5577411651611328,
      "learning_rate": 0.00018580714138737288,
      "loss": 1.2018,
      "step": 1867
    },
    {
      "epoch": 0.3471473703772533,
      "grad_norm": 0.5997073650360107,
      "learning_rate": 0.00018579211906908215,
      "loss": 1.1088,
      "step": 1868
    },
    {
      "epoch": 0.3473332094406244,
      "grad_norm": 0.5570213198661804,
      "learning_rate": 0.000185777089412806,
      "loss": 0.9635,
      "step": 1869
    },
    {
      "epoch": 0.34751904850399556,
      "grad_norm": 0.5627360343933105,
      "learning_rate": 0.00018576205241983005,
      "loss": 1.1156,
      "step": 1870
    },
    {
      "epoch": 0.34770488756736667,
      "grad_norm": 0.6400758028030396,
      "learning_rate": 0.00018574700809144035,
      "loss": 1.3113,
      "step": 1871
    },
    {
      "epoch": 0.3478907266307378,
      "grad_norm": 0.6254924535751343,
      "learning_rate": 0.00018573195642892376,
      "loss": 0.9771,
      "step": 1872
    },
    {
      "epoch": 0.3480765656941089,
      "grad_norm": 0.5834730267524719,
      "learning_rate": 0.00018571689743356764,
      "loss": 0.985,
      "step": 1873
    },
    {
      "epoch": 0.34826240475748005,
      "grad_norm": 0.6734195351600647,
      "learning_rate": 0.00018570183110666003,
      "loss": 1.2304,
      "step": 1874
    },
    {
      "epoch": 0.34844824382085116,
      "grad_norm": 0.6961857080459595,
      "learning_rate": 0.0001856867574494896,
      "loss": 1.1581,
      "step": 1875
    },
    {
      "epoch": 0.34863408288422226,
      "grad_norm": 0.6982626914978027,
      "learning_rate": 0.0001856716764633456,
      "loss": 0.9181,
      "step": 1876
    },
    {
      "epoch": 0.34881992194759337,
      "grad_norm": 0.7757996916770935,
      "learning_rate": 0.00018565658814951788,
      "loss": 1.0588,
      "step": 1877
    },
    {
      "epoch": 0.3490057610109645,
      "grad_norm": 0.5235817432403564,
      "learning_rate": 0.00018564149250929713,
      "loss": 0.9972,
      "step": 1878
    },
    {
      "epoch": 0.34919160007433564,
      "grad_norm": 0.5383930206298828,
      "learning_rate": 0.0001856263895439744,
      "loss": 0.9844,
      "step": 1879
    },
    {
      "epoch": 0.34937743913770675,
      "grad_norm": 0.5330451130867004,
      "learning_rate": 0.0001856112792548415,
      "loss": 0.7695,
      "step": 1880
    },
    {
      "epoch": 0.34956327820107785,
      "grad_norm": 0.5530948042869568,
      "learning_rate": 0.0001855961616431909,
      "loss": 1.0495,
      "step": 1881
    },
    {
      "epoch": 0.34974911726444896,
      "grad_norm": 0.5738743543624878,
      "learning_rate": 0.00018558103671031554,
      "loss": 0.9853,
      "step": 1882
    },
    {
      "epoch": 0.3499349563278201,
      "grad_norm": 0.5287640690803528,
      "learning_rate": 0.0001855659044575092,
      "loss": 1.1136,
      "step": 1883
    },
    {
      "epoch": 0.35012079539119123,
      "grad_norm": 0.5047253966331482,
      "learning_rate": 0.00018555076488606606,
      "loss": 0.9412,
      "step": 1884
    },
    {
      "epoch": 0.35030663445456234,
      "grad_norm": 0.5331687927246094,
      "learning_rate": 0.00018553561799728113,
      "loss": 1.174,
      "step": 1885
    },
    {
      "epoch": 0.35049247351793344,
      "grad_norm": 0.6587241888046265,
      "learning_rate": 0.00018552046379244992,
      "loss": 1.0395,
      "step": 1886
    },
    {
      "epoch": 0.3506783125813046,
      "grad_norm": 0.541508674621582,
      "learning_rate": 0.0001855053022728686,
      "loss": 1.1467,
      "step": 1887
    },
    {
      "epoch": 0.3508641516446757,
      "grad_norm": 0.6289540529251099,
      "learning_rate": 0.000185490133439834,
      "loss": 1.0423,
      "step": 1888
    },
    {
      "epoch": 0.3510499907080468,
      "grad_norm": 0.5184619426727295,
      "learning_rate": 0.00018547495729464352,
      "loss": 1.0425,
      "step": 1889
    },
    {
      "epoch": 0.35123582977141793,
      "grad_norm": 0.541866660118103,
      "learning_rate": 0.0001854597738385952,
      "loss": 1.2454,
      "step": 1890
    },
    {
      "epoch": 0.3514216688347891,
      "grad_norm": 0.46014341711997986,
      "learning_rate": 0.0001854445830729877,
      "loss": 1.0373,
      "step": 1891
    },
    {
      "epoch": 0.3516075078981602,
      "grad_norm": 0.5548123717308044,
      "learning_rate": 0.00018542938499912034,
      "loss": 1.0437,
      "step": 1892
    },
    {
      "epoch": 0.3517933469615313,
      "grad_norm": 0.7150872945785522,
      "learning_rate": 0.0001854141796182931,
      "loss": 1.3672,
      "step": 1893
    },
    {
      "epoch": 0.3519791860249024,
      "grad_norm": 0.6109015345573425,
      "learning_rate": 0.0001853989669318064,
      "loss": 1.3214,
      "step": 1894
    },
    {
      "epoch": 0.3521650250882736,
      "grad_norm": 0.606502115726471,
      "learning_rate": 0.0001853837469409615,
      "loss": 1.0622,
      "step": 1895
    },
    {
      "epoch": 0.3523508641516447,
      "grad_norm": 0.6415699124336243,
      "learning_rate": 0.0001853685196470602,
      "loss": 1.1894,
      "step": 1896
    },
    {
      "epoch": 0.3525367032150158,
      "grad_norm": 0.5605230331420898,
      "learning_rate": 0.00018535328505140495,
      "loss": 0.9746,
      "step": 1897
    },
    {
      "epoch": 0.3527225422783869,
      "grad_norm": 0.5477644205093384,
      "learning_rate": 0.00018533804315529873,
      "loss": 0.8564,
      "step": 1898
    },
    {
      "epoch": 0.35290838134175806,
      "grad_norm": 0.5427902340888977,
      "learning_rate": 0.00018532279396004522,
      "loss": 1.1638,
      "step": 1899
    },
    {
      "epoch": 0.35309422040512917,
      "grad_norm": 0.5802868604660034,
      "learning_rate": 0.00018530753746694872,
      "loss": 1.2356,
      "step": 1900
    },
    {
      "epoch": 0.3532800594685003,
      "grad_norm": 0.5529875755310059,
      "learning_rate": 0.0001852922736773142,
      "loss": 1.0866,
      "step": 1901
    },
    {
      "epoch": 0.3534658985318714,
      "grad_norm": 0.5103433728218079,
      "learning_rate": 0.00018527700259244717,
      "loss": 0.9164,
      "step": 1902
    },
    {
      "epoch": 0.35365173759524254,
      "grad_norm": 0.5571877360343933,
      "learning_rate": 0.0001852617242136538,
      "loss": 1.2392,
      "step": 1903
    },
    {
      "epoch": 0.35383757665861365,
      "grad_norm": 0.5150656700134277,
      "learning_rate": 0.00018524643854224087,
      "loss": 1.0823,
      "step": 1904
    },
    {
      "epoch": 0.35402341572198476,
      "grad_norm": 0.49546682834625244,
      "learning_rate": 0.00018523114557951578,
      "loss": 1.077,
      "step": 1905
    },
    {
      "epoch": 0.35420925478535586,
      "grad_norm": 0.5201963782310486,
      "learning_rate": 0.00018521584532678667,
      "loss": 1.1797,
      "step": 1906
    },
    {
      "epoch": 0.354395093848727,
      "grad_norm": 0.604653537273407,
      "learning_rate": 0.00018520053778536212,
      "loss": 1.1176,
      "step": 1907
    },
    {
      "epoch": 0.35458093291209813,
      "grad_norm": 0.44055116176605225,
      "learning_rate": 0.0001851852229565514,
      "loss": 0.7782,
      "step": 1908
    },
    {
      "epoch": 0.35476677197546924,
      "grad_norm": 0.5750049352645874,
      "learning_rate": 0.00018516990084166442,
      "loss": 1.1154,
      "step": 1909
    },
    {
      "epoch": 0.35495261103884035,
      "grad_norm": 0.5687276124954224,
      "learning_rate": 0.0001851545714420118,
      "loss": 1.0966,
      "step": 1910
    },
    {
      "epoch": 0.3551384501022115,
      "grad_norm": 0.5168128609657288,
      "learning_rate": 0.00018513923475890463,
      "loss": 1.1899,
      "step": 1911
    },
    {
      "epoch": 0.3553242891655826,
      "grad_norm": 0.4693703055381775,
      "learning_rate": 0.0001851238907936547,
      "loss": 0.8872,
      "step": 1912
    },
    {
      "epoch": 0.3555101282289537,
      "grad_norm": 0.5975525379180908,
      "learning_rate": 0.00018510853954757441,
      "loss": 1.1444,
      "step": 1913
    },
    {
      "epoch": 0.35569596729232483,
      "grad_norm": 0.5854206681251526,
      "learning_rate": 0.00018509318102197678,
      "loss": 1.2371,
      "step": 1914
    },
    {
      "epoch": 0.35588180635569594,
      "grad_norm": 0.5721655488014221,
      "learning_rate": 0.00018507781521817548,
      "loss": 1.1206,
      "step": 1915
    },
    {
      "epoch": 0.3560676454190671,
      "grad_norm": 0.659113883972168,
      "learning_rate": 0.00018506244213748477,
      "loss": 1.269,
      "step": 1916
    },
    {
      "epoch": 0.3562534844824382,
      "grad_norm": 0.5849035382270813,
      "learning_rate": 0.00018504706178121952,
      "loss": 1.1486,
      "step": 1917
    },
    {
      "epoch": 0.3564393235458093,
      "grad_norm": 0.6287510991096497,
      "learning_rate": 0.00018503167415069528,
      "loss": 1.1698,
      "step": 1918
    },
    {
      "epoch": 0.3566251626091804,
      "grad_norm": 0.5501243472099304,
      "learning_rate": 0.00018501627924722818,
      "loss": 0.9855,
      "step": 1919
    },
    {
      "epoch": 0.3568110016725516,
      "grad_norm": 0.5094524025917053,
      "learning_rate": 0.00018500087707213497,
      "loss": 1.1109,
      "step": 1920
    },
    {
      "epoch": 0.3569968407359227,
      "grad_norm": 0.53468918800354,
      "learning_rate": 0.00018498546762673302,
      "loss": 0.6887,
      "step": 1921
    },
    {
      "epoch": 0.3571826797992938,
      "grad_norm": 0.5579971671104431,
      "learning_rate": 0.00018497005091234036,
      "loss": 1.0723,
      "step": 1922
    },
    {
      "epoch": 0.3573685188626649,
      "grad_norm": 0.5719289183616638,
      "learning_rate": 0.00018495462693027557,
      "loss": 1.1419,
      "step": 1923
    },
    {
      "epoch": 0.35755435792603607,
      "grad_norm": 0.6961529850959778,
      "learning_rate": 0.00018493919568185796,
      "loss": 1.1488,
      "step": 1924
    },
    {
      "epoch": 0.3577401969894072,
      "grad_norm": 0.5444719195365906,
      "learning_rate": 0.00018492375716840735,
      "loss": 1.1814,
      "step": 1925
    },
    {
      "epoch": 0.3579260360527783,
      "grad_norm": 0.5759077072143555,
      "learning_rate": 0.00018490831139124424,
      "loss": 1.1933,
      "step": 1926
    },
    {
      "epoch": 0.3581118751161494,
      "grad_norm": 0.6321165561676025,
      "learning_rate": 0.00018489285835168978,
      "loss": 1.1398,
      "step": 1927
    },
    {
      "epoch": 0.35829771417952055,
      "grad_norm": 0.6945244669914246,
      "learning_rate": 0.00018487739805106562,
      "loss": 1.1798,
      "step": 1928
    },
    {
      "epoch": 0.35848355324289166,
      "grad_norm": 0.5583364367485046,
      "learning_rate": 0.0001848619304906942,
      "loss": 0.9665,
      "step": 1929
    },
    {
      "epoch": 0.35866939230626277,
      "grad_norm": 0.511746883392334,
      "learning_rate": 0.00018484645567189845,
      "loss": 1.1767,
      "step": 1930
    },
    {
      "epoch": 0.3588552313696339,
      "grad_norm": 0.5670949816703796,
      "learning_rate": 0.00018483097359600195,
      "loss": 0.9885,
      "step": 1931
    },
    {
      "epoch": 0.35904107043300504,
      "grad_norm": 0.5351964235305786,
      "learning_rate": 0.00018481548426432895,
      "loss": 1.0039,
      "step": 1932
    },
    {
      "epoch": 0.35922690949637615,
      "grad_norm": 0.5773396492004395,
      "learning_rate": 0.00018479998767820425,
      "loss": 1.0051,
      "step": 1933
    },
    {
      "epoch": 0.35941274855974725,
      "grad_norm": 0.5246624946594238,
      "learning_rate": 0.00018478448383895332,
      "loss": 0.88,
      "step": 1934
    },
    {
      "epoch": 0.35959858762311836,
      "grad_norm": 0.5461812615394592,
      "learning_rate": 0.00018476897274790228,
      "loss": 0.9274,
      "step": 1935
    },
    {
      "epoch": 0.3597844266864895,
      "grad_norm": 0.5961292386054993,
      "learning_rate": 0.00018475345440637777,
      "loss": 1.0634,
      "step": 1936
    },
    {
      "epoch": 0.35997026574986063,
      "grad_norm": 0.6102057695388794,
      "learning_rate": 0.00018473792881570713,
      "loss": 0.9261,
      "step": 1937
    },
    {
      "epoch": 0.36015610481323174,
      "grad_norm": 0.549595057964325,
      "learning_rate": 0.0001847223959772183,
      "loss": 0.9047,
      "step": 1938
    },
    {
      "epoch": 0.36034194387660284,
      "grad_norm": 0.6503207683563232,
      "learning_rate": 0.00018470685589223984,
      "loss": 1.164,
      "step": 1939
    },
    {
      "epoch": 0.360527782939974,
      "grad_norm": 0.5422487258911133,
      "learning_rate": 0.00018469130856210093,
      "loss": 1.0569,
      "step": 1940
    },
    {
      "epoch": 0.3607136220033451,
      "grad_norm": 0.6560564637184143,
      "learning_rate": 0.00018467575398813132,
      "loss": 1.0021,
      "step": 1941
    },
    {
      "epoch": 0.3608994610667162,
      "grad_norm": 0.5428600907325745,
      "learning_rate": 0.0001846601921716615,
      "loss": 1.1048,
      "step": 1942
    },
    {
      "epoch": 0.3610853001300873,
      "grad_norm": 0.5088539123535156,
      "learning_rate": 0.00018464462311402243,
      "loss": 0.9902,
      "step": 1943
    },
    {
      "epoch": 0.3612711391934585,
      "grad_norm": 0.5354966521263123,
      "learning_rate": 0.00018462904681654582,
      "loss": 1.1174,
      "step": 1944
    },
    {
      "epoch": 0.3614569782568296,
      "grad_norm": 0.581659734249115,
      "learning_rate": 0.00018461346328056392,
      "loss": 1.0791,
      "step": 1945
    },
    {
      "epoch": 0.3616428173202007,
      "grad_norm": 0.6271023750305176,
      "learning_rate": 0.00018459787250740966,
      "loss": 1.2299,
      "step": 1946
    },
    {
      "epoch": 0.3618286563835718,
      "grad_norm": 0.5690140724182129,
      "learning_rate": 0.0001845822744984165,
      "loss": 1.1757,
      "step": 1947
    },
    {
      "epoch": 0.362014495446943,
      "grad_norm": 0.6378099322319031,
      "learning_rate": 0.00018456666925491862,
      "loss": 0.9182,
      "step": 1948
    },
    {
      "epoch": 0.3622003345103141,
      "grad_norm": 0.5846841335296631,
      "learning_rate": 0.00018455105677825072,
      "loss": 0.7915,
      "step": 1949
    },
    {
      "epoch": 0.3623861735736852,
      "grad_norm": 0.5810773372650146,
      "learning_rate": 0.0001845354370697482,
      "loss": 1.0371,
      "step": 1950
    },
    {
      "epoch": 0.3625720126370563,
      "grad_norm": 0.5026060342788696,
      "learning_rate": 0.00018451981013074703,
      "loss": 1.2072,
      "step": 1951
    },
    {
      "epoch": 0.3627578517004274,
      "grad_norm": 0.5775339007377625,
      "learning_rate": 0.00018450417596258388,
      "loss": 1.2526,
      "step": 1952
    },
    {
      "epoch": 0.36294369076379857,
      "grad_norm": 0.5179687142372131,
      "learning_rate": 0.00018448853456659586,
      "loss": 1.0594,
      "step": 1953
    },
    {
      "epoch": 0.3631295298271697,
      "grad_norm": 0.5170058608055115,
      "learning_rate": 0.00018447288594412094,
      "loss": 1.0907,
      "step": 1954
    },
    {
      "epoch": 0.3633153688905408,
      "grad_norm": 0.5153071284294128,
      "learning_rate": 0.00018445723009649748,
      "loss": 1.19,
      "step": 1955
    },
    {
      "epoch": 0.3635012079539119,
      "grad_norm": 0.6060937643051147,
      "learning_rate": 0.00018444156702506457,
      "loss": 0.9486,
      "step": 1956
    },
    {
      "epoch": 0.36368704701728305,
      "grad_norm": 0.5591087341308594,
      "learning_rate": 0.00018442589673116195,
      "loss": 0.8211,
      "step": 1957
    },
    {
      "epoch": 0.36387288608065416,
      "grad_norm": 0.4953506588935852,
      "learning_rate": 0.0001844102192161299,
      "loss": 0.7849,
      "step": 1958
    },
    {
      "epoch": 0.36405872514402526,
      "grad_norm": 0.6325752139091492,
      "learning_rate": 0.0001843945344813094,
      "loss": 1.2812,
      "step": 1959
    },
    {
      "epoch": 0.36424456420739637,
      "grad_norm": 0.5928200483322144,
      "learning_rate": 0.00018437884252804195,
      "loss": 1.2031,
      "step": 1960
    },
    {
      "epoch": 0.36443040327076753,
      "grad_norm": 0.6781278848648071,
      "learning_rate": 0.0001843631433576697,
      "loss": 1.2147,
      "step": 1961
    },
    {
      "epoch": 0.36461624233413864,
      "grad_norm": 0.6580811142921448,
      "learning_rate": 0.00018434743697153552,
      "loss": 1.1414,
      "step": 1962
    },
    {
      "epoch": 0.36480208139750975,
      "grad_norm": 0.5909239053726196,
      "learning_rate": 0.00018433172337098272,
      "loss": 1.1057,
      "step": 1963
    },
    {
      "epoch": 0.36498792046088085,
      "grad_norm": 0.6667724847793579,
      "learning_rate": 0.00018431600255735538,
      "loss": 1.0801,
      "step": 1964
    },
    {
      "epoch": 0.365173759524252,
      "grad_norm": 0.6188902854919434,
      "learning_rate": 0.00018430027453199808,
      "loss": 1.0105,
      "step": 1965
    },
    {
      "epoch": 0.3653595985876231,
      "grad_norm": 0.5609699487686157,
      "learning_rate": 0.00018428453929625614,
      "loss": 1.0674,
      "step": 1966
    },
    {
      "epoch": 0.36554543765099423,
      "grad_norm": 0.6792697906494141,
      "learning_rate": 0.00018426879685147537,
      "loss": 1.193,
      "step": 1967
    },
    {
      "epoch": 0.36573127671436534,
      "grad_norm": 0.6148858666419983,
      "learning_rate": 0.00018425304719900227,
      "loss": 1.1392,
      "step": 1968
    },
    {
      "epoch": 0.3659171157777365,
      "grad_norm": 0.4912833869457245,
      "learning_rate": 0.00018423729034018396,
      "loss": 0.8163,
      "step": 1969
    },
    {
      "epoch": 0.3661029548411076,
      "grad_norm": 0.6004575490951538,
      "learning_rate": 0.00018422152627636814,
      "loss": 1.1595,
      "step": 1970
    },
    {
      "epoch": 0.3662887939044787,
      "grad_norm": 0.6387303471565247,
      "learning_rate": 0.00018420575500890314,
      "loss": 1.2459,
      "step": 1971
    },
    {
      "epoch": 0.3664746329678498,
      "grad_norm": 0.5127397179603577,
      "learning_rate": 0.00018418997653913793,
      "loss": 0.9239,
      "step": 1972
    },
    {
      "epoch": 0.366660472031221,
      "grad_norm": 0.5673999786376953,
      "learning_rate": 0.00018417419086842206,
      "loss": 1.0092,
      "step": 1973
    },
    {
      "epoch": 0.3668463110945921,
      "grad_norm": 0.8210087418556213,
      "learning_rate": 0.00018415839799810577,
      "loss": 1.0895,
      "step": 1974
    },
    {
      "epoch": 0.3670321501579632,
      "grad_norm": 0.5619094967842102,
      "learning_rate": 0.00018414259792953977,
      "loss": 0.9131,
      "step": 1975
    },
    {
      "epoch": 0.3672179892213343,
      "grad_norm": 0.5745676755905151,
      "learning_rate": 0.0001841267906640755,
      "loss": 0.7174,
      "step": 1976
    },
    {
      "epoch": 0.36740382828470547,
      "grad_norm": 0.6106050610542297,
      "learning_rate": 0.00018411097620306502,
      "loss": 0.991,
      "step": 1977
    },
    {
      "epoch": 0.3675896673480766,
      "grad_norm": 0.5693312883377075,
      "learning_rate": 0.00018409515454786095,
      "loss": 0.9161,
      "step": 1978
    },
    {
      "epoch": 0.3677755064114477,
      "grad_norm": 0.623209536075592,
      "learning_rate": 0.00018407932569981656,
      "loss": 1.0186,
      "step": 1979
    },
    {
      "epoch": 0.3679613454748188,
      "grad_norm": 0.5974239110946655,
      "learning_rate": 0.00018406348966028575,
      "loss": 1.1224,
      "step": 1980
    },
    {
      "epoch": 0.36814718453818995,
      "grad_norm": 0.5626535415649414,
      "learning_rate": 0.00018404764643062296,
      "loss": 1.0819,
      "step": 1981
    },
    {
      "epoch": 0.36833302360156106,
      "grad_norm": 0.5896697044372559,
      "learning_rate": 0.0001840317960121833,
      "loss": 1.1705,
      "step": 1982
    },
    {
      "epoch": 0.36851886266493217,
      "grad_norm": 0.5947336554527283,
      "learning_rate": 0.00018401593840632253,
      "loss": 1.2353,
      "step": 1983
    },
    {
      "epoch": 0.3687047017283033,
      "grad_norm": 0.5700284242630005,
      "learning_rate": 0.00018400007361439697,
      "loss": 0.9938,
      "step": 1984
    },
    {
      "epoch": 0.36889054079167444,
      "grad_norm": 0.513639509677887,
      "learning_rate": 0.00018398420163776358,
      "loss": 1.1357,
      "step": 1985
    },
    {
      "epoch": 0.36907637985504554,
      "grad_norm": 0.6426970958709717,
      "learning_rate": 0.0001839683224777799,
      "loss": 1.1692,
      "step": 1986
    },
    {
      "epoch": 0.36926221891841665,
      "grad_norm": 0.5685890913009644,
      "learning_rate": 0.0001839524361358041,
      "loss": 1.1505,
      "step": 1987
    },
    {
      "epoch": 0.36944805798178776,
      "grad_norm": 0.5381999015808105,
      "learning_rate": 0.000183936542613195,
      "loss": 1.0664,
      "step": 1988
    },
    {
      "epoch": 0.36963389704515887,
      "grad_norm": 0.5211588740348816,
      "learning_rate": 0.00018392064191131202,
      "loss": 1.1045,
      "step": 1989
    },
    {
      "epoch": 0.36981973610853003,
      "grad_norm": 0.605248749256134,
      "learning_rate": 0.00018390473403151513,
      "loss": 1.0103,
      "step": 1990
    },
    {
      "epoch": 0.37000557517190114,
      "grad_norm": 0.6501337885856628,
      "learning_rate": 0.000183888818975165,
      "loss": 1.1475,
      "step": 1991
    },
    {
      "epoch": 0.37019141423527224,
      "grad_norm": 0.5290714502334595,
      "learning_rate": 0.00018387289674362292,
      "loss": 0.9015,
      "step": 1992
    },
    {
      "epoch": 0.37037725329864335,
      "grad_norm": 0.5104148983955383,
      "learning_rate": 0.00018385696733825066,
      "loss": 0.9868,
      "step": 1993
    },
    {
      "epoch": 0.3705630923620145,
      "grad_norm": 0.6122913956642151,
      "learning_rate": 0.00018384103076041078,
      "loss": 1.0051,
      "step": 1994
    },
    {
      "epoch": 0.3707489314253856,
      "grad_norm": 6.534750461578369,
      "learning_rate": 0.0001838250870114663,
      "loss": 3.3172,
      "step": 1995
    },
    {
      "epoch": 0.3709347704887567,
      "grad_norm": 0.4851309657096863,
      "learning_rate": 0.00018380913609278098,
      "loss": 0.9435,
      "step": 1996
    },
    {
      "epoch": 0.37112060955212783,
      "grad_norm": 0.6162280440330505,
      "learning_rate": 0.0001837931780057191,
      "loss": 1.0495,
      "step": 1997
    },
    {
      "epoch": 0.371306448615499,
      "grad_norm": 0.5363066792488098,
      "learning_rate": 0.00018377721275164558,
      "loss": 1.0672,
      "step": 1998
    },
    {
      "epoch": 0.3714922876788701,
      "grad_norm": 0.5488103032112122,
      "learning_rate": 0.00018376124033192602,
      "loss": 0.985,
      "step": 1999
    },
    {
      "epoch": 0.3716781267422412,
      "grad_norm": 0.5667101740837097,
      "learning_rate": 0.00018374526074792656,
      "loss": 1.0454,
      "step": 2000
    },
    {
      "epoch": 0.3718639658056123,
      "grad_norm": 0.6776452660560608,
      "learning_rate": 0.00018372927400101395,
      "loss": 1.1813,
      "step": 2001
    },
    {
      "epoch": 0.3720498048689835,
      "grad_norm": 0.6331282258033752,
      "learning_rate": 0.00018371328009255552,
      "loss": 1.2781,
      "step": 2002
    },
    {
      "epoch": 0.3722356439323546,
      "grad_norm": 0.5698820352554321,
      "learning_rate": 0.00018369727902391937,
      "loss": 1.2935,
      "step": 2003
    },
    {
      "epoch": 0.3724214829957257,
      "grad_norm": 0.6152218580245972,
      "learning_rate": 0.00018368127079647405,
      "loss": 1.1618,
      "step": 2004
    },
    {
      "epoch": 0.3726073220590968,
      "grad_norm": 0.6485865116119385,
      "learning_rate": 0.00018366525541158875,
      "loss": 1.2801,
      "step": 2005
    },
    {
      "epoch": 0.37279316112246796,
      "grad_norm": 0.5456370711326599,
      "learning_rate": 0.00018364923287063334,
      "loss": 1.0006,
      "step": 2006
    },
    {
      "epoch": 0.37297900018583907,
      "grad_norm": 0.5099388957023621,
      "learning_rate": 0.00018363320317497825,
      "loss": 0.9357,
      "step": 2007
    },
    {
      "epoch": 0.3731648392492102,
      "grad_norm": 0.5599960088729858,
      "learning_rate": 0.00018361716632599455,
      "loss": 1.1451,
      "step": 2008
    },
    {
      "epoch": 0.3733506783125813,
      "grad_norm": 0.6054771542549133,
      "learning_rate": 0.00018360112232505387,
      "loss": 1.1288,
      "step": 2009
    },
    {
      "epoch": 0.37353651737595245,
      "grad_norm": 0.7153000235557556,
      "learning_rate": 0.00018358507117352852,
      "loss": 1.2604,
      "step": 2010
    },
    {
      "epoch": 0.37372235643932356,
      "grad_norm": 0.6568436026573181,
      "learning_rate": 0.0001835690128727914,
      "loss": 1.1132,
      "step": 2011
    },
    {
      "epoch": 0.37390819550269466,
      "grad_norm": 0.6042101979255676,
      "learning_rate": 0.00018355294742421598,
      "loss": 1.3149,
      "step": 2012
    },
    {
      "epoch": 0.37409403456606577,
      "grad_norm": 0.6615718603134155,
      "learning_rate": 0.0001835368748291764,
      "loss": 1.3576,
      "step": 2013
    },
    {
      "epoch": 0.37427987362943693,
      "grad_norm": 0.5358205437660217,
      "learning_rate": 0.00018352079508904734,
      "loss": 1.0024,
      "step": 2014
    },
    {
      "epoch": 0.37446571269280804,
      "grad_norm": 0.44465547800064087,
      "learning_rate": 0.00018350470820520417,
      "loss": 0.8894,
      "step": 2015
    },
    {
      "epoch": 0.37465155175617915,
      "grad_norm": 0.4735083281993866,
      "learning_rate": 0.00018348861417902287,
      "loss": 0.8917,
      "step": 2016
    },
    {
      "epoch": 0.37483739081955025,
      "grad_norm": 0.6190890073776245,
      "learning_rate": 0.00018347251301187992,
      "loss": 1.2724,
      "step": 2017
    },
    {
      "epoch": 0.3750232298829214,
      "grad_norm": 0.5531015396118164,
      "learning_rate": 0.0001834564047051525,
      "loss": 1.1366,
      "step": 2018
    },
    {
      "epoch": 0.3752090689462925,
      "grad_norm": 0.6023117899894714,
      "learning_rate": 0.00018344028926021844,
      "loss": 0.9135,
      "step": 2019
    },
    {
      "epoch": 0.37539490800966363,
      "grad_norm": 0.5503703355789185,
      "learning_rate": 0.0001834241666784561,
      "loss": 1.0205,
      "step": 2020
    },
    {
      "epoch": 0.37558074707303474,
      "grad_norm": 0.554999828338623,
      "learning_rate": 0.00018340803696124448,
      "loss": 1.1605,
      "step": 2021
    },
    {
      "epoch": 0.3757665861364059,
      "grad_norm": 0.5042999386787415,
      "learning_rate": 0.00018339190010996321,
      "loss": 1.1332,
      "step": 2022
    },
    {
      "epoch": 0.375952425199777,
      "grad_norm": 0.64976567029953,
      "learning_rate": 0.00018337575612599246,
      "loss": 1.2517,
      "step": 2023
    },
    {
      "epoch": 0.3761382642631481,
      "grad_norm": 0.7032577991485596,
      "learning_rate": 0.00018335960501071312,
      "loss": 1.26,
      "step": 2024
    },
    {
      "epoch": 0.3763241033265192,
      "grad_norm": 0.5371274948120117,
      "learning_rate": 0.0001833434467655066,
      "loss": 1.0916,
      "step": 2025
    },
    {
      "epoch": 0.37650994238989033,
      "grad_norm": 0.6726012229919434,
      "learning_rate": 0.00018332728139175492,
      "loss": 1.1531,
      "step": 2026
    },
    {
      "epoch": 0.3766957814532615,
      "grad_norm": 0.5239533185958862,
      "learning_rate": 0.00018331110889084078,
      "loss": 0.9916,
      "step": 2027
    },
    {
      "epoch": 0.3768816205166326,
      "grad_norm": 0.542515754699707,
      "learning_rate": 0.00018329492926414745,
      "loss": 1.2967,
      "step": 2028
    },
    {
      "epoch": 0.3770674595800037,
      "grad_norm": 0.542534351348877,
      "learning_rate": 0.00018327874251305879,
      "loss": 1.0778,
      "step": 2029
    },
    {
      "epoch": 0.3772532986433748,
      "grad_norm": 0.7218822836875916,
      "learning_rate": 0.00018326254863895928,
      "loss": 1.3548,
      "step": 2030
    },
    {
      "epoch": 0.377439137706746,
      "grad_norm": 0.5446364283561707,
      "learning_rate": 0.00018324634764323402,
      "loss": 1.0741,
      "step": 2031
    },
    {
      "epoch": 0.3776249767701171,
      "grad_norm": 0.5216005444526672,
      "learning_rate": 0.00018323013952726875,
      "loss": 0.9263,
      "step": 2032
    },
    {
      "epoch": 0.3778108158334882,
      "grad_norm": 0.5345918536186218,
      "learning_rate": 0.00018321392429244976,
      "loss": 1.1902,
      "step": 2033
    },
    {
      "epoch": 0.3779966548968593,
      "grad_norm": 0.5378860831260681,
      "learning_rate": 0.00018319770194016397,
      "loss": 1.1874,
      "step": 2034
    },
    {
      "epoch": 0.37818249396023046,
      "grad_norm": 2.5068178176879883,
      "learning_rate": 0.00018318147247179894,
      "loss": 3.1407,
      "step": 2035
    },
    {
      "epoch": 0.37836833302360157,
      "grad_norm": 0.5298924446105957,
      "learning_rate": 0.00018316523588874278,
      "loss": 1.1361,
      "step": 2036
    },
    {
      "epoch": 0.3785541720869727,
      "grad_norm": 0.6050952672958374,
      "learning_rate": 0.00018314899219238422,
      "loss": 0.869,
      "step": 2037
    },
    {
      "epoch": 0.3787400111503438,
      "grad_norm": 0.5876106023788452,
      "learning_rate": 0.00018313274138411268,
      "loss": 1.1049,
      "step": 2038
    },
    {
      "epoch": 0.37892585021371494,
      "grad_norm": 0.501067578792572,
      "learning_rate": 0.0001831164834653181,
      "loss": 1.049,
      "step": 2039
    },
    {
      "epoch": 0.37911168927708605,
      "grad_norm": 0.5056796073913574,
      "learning_rate": 0.00018310021843739104,
      "loss": 0.866,
      "step": 2040
    },
    {
      "epoch": 0.37929752834045716,
      "grad_norm": 0.574049174785614,
      "learning_rate": 0.0001830839463017227,
      "loss": 0.8973,
      "step": 2041
    },
    {
      "epoch": 0.37948336740382826,
      "grad_norm": 0.5803636312484741,
      "learning_rate": 0.00018306766705970486,
      "loss": 1.3646,
      "step": 2042
    },
    {
      "epoch": 0.3796692064671994,
      "grad_norm": 0.5793257355690002,
      "learning_rate": 0.00018305138071272992,
      "loss": 1.0705,
      "step": 2043
    },
    {
      "epoch": 0.37985504553057053,
      "grad_norm": 0.4978286921977997,
      "learning_rate": 0.0001830350872621909,
      "loss": 1.0592,
      "step": 2044
    },
    {
      "epoch": 0.38004088459394164,
      "grad_norm": 0.5665377378463745,
      "learning_rate": 0.00018301878670948137,
      "loss": 0.8908,
      "step": 2045
    },
    {
      "epoch": 0.38022672365731275,
      "grad_norm": 0.5234101414680481,
      "learning_rate": 0.00018300247905599564,
      "loss": 1.0787,
      "step": 2046
    },
    {
      "epoch": 0.3804125627206839,
      "grad_norm": 0.5455564856529236,
      "learning_rate": 0.00018298616430312845,
      "loss": 1.0485,
      "step": 2047
    },
    {
      "epoch": 0.380598401784055,
      "grad_norm": 0.6050750613212585,
      "learning_rate": 0.00018296984245227526,
      "loss": 1.1934,
      "step": 2048
    },
    {
      "epoch": 0.3807842408474261,
      "grad_norm": 0.6124750375747681,
      "learning_rate": 0.00018295351350483216,
      "loss": 1.2551,
      "step": 2049
    },
    {
      "epoch": 0.38097007991079723,
      "grad_norm": 0.6014724373817444,
      "learning_rate": 0.00018293717746219575,
      "loss": 1.0813,
      "step": 2050
    },
    {
      "epoch": 0.3811559189741684,
      "grad_norm": 0.5594990253448486,
      "learning_rate": 0.0001829208343257633,
      "loss": 1.0199,
      "step": 2051
    },
    {
      "epoch": 0.3813417580375395,
      "grad_norm": 0.5794000029563904,
      "learning_rate": 0.00018290448409693266,
      "loss": 1.0601,
      "step": 2052
    },
    {
      "epoch": 0.3815275971009106,
      "grad_norm": 0.6697391867637634,
      "learning_rate": 0.00018288812677710233,
      "loss": 1.341,
      "step": 2053
    },
    {
      "epoch": 0.3817134361642817,
      "grad_norm": 0.6012246608734131,
      "learning_rate": 0.00018287176236767136,
      "loss": 1.2732,
      "step": 2054
    },
    {
      "epoch": 0.3818992752276529,
      "grad_norm": 0.5211855173110962,
      "learning_rate": 0.00018285539087003946,
      "loss": 1.0806,
      "step": 2055
    },
    {
      "epoch": 0.382085114291024,
      "grad_norm": 0.6030214428901672,
      "learning_rate": 0.0001828390122856069,
      "loss": 1.1769,
      "step": 2056
    },
    {
      "epoch": 0.3822709533543951,
      "grad_norm": 0.6156930923461914,
      "learning_rate": 0.00018282262661577456,
      "loss": 1.272,
      "step": 2057
    },
    {
      "epoch": 0.3824567924177662,
      "grad_norm": 0.5623907446861267,
      "learning_rate": 0.000182806233861944,
      "loss": 1.0137,
      "step": 2058
    },
    {
      "epoch": 0.3826426314811373,
      "grad_norm": 0.5875430703163147,
      "learning_rate": 0.00018278983402551728,
      "loss": 1.4113,
      "step": 2059
    },
    {
      "epoch": 0.38282847054450847,
      "grad_norm": 0.5854385495185852,
      "learning_rate": 0.0001827734271078971,
      "loss": 1.416,
      "step": 2060
    },
    {
      "epoch": 0.3830143096078796,
      "grad_norm": 0.6129041314125061,
      "learning_rate": 0.00018275701311048682,
      "loss": 1.571,
      "step": 2061
    },
    {
      "epoch": 0.3832001486712507,
      "grad_norm": 0.5709010362625122,
      "learning_rate": 0.00018274059203469034,
      "loss": 1.3855,
      "step": 2062
    },
    {
      "epoch": 0.3833859877346218,
      "grad_norm": 0.5533404350280762,
      "learning_rate": 0.0001827241638819122,
      "loss": 1.0585,
      "step": 2063
    },
    {
      "epoch": 0.38357182679799295,
      "grad_norm": 0.45654740929603577,
      "learning_rate": 0.00018270772865355755,
      "loss": 0.9902,
      "step": 2064
    },
    {
      "epoch": 0.38375766586136406,
      "grad_norm": 0.5475004315376282,
      "learning_rate": 0.00018269128635103207,
      "loss": 1.1629,
      "step": 2065
    },
    {
      "epoch": 0.38394350492473517,
      "grad_norm": 0.6893115639686584,
      "learning_rate": 0.00018267483697574218,
      "loss": 1.1441,
      "step": 2066
    },
    {
      "epoch": 0.3841293439881063,
      "grad_norm": 0.4979449212551117,
      "learning_rate": 0.00018265838052909482,
      "loss": 0.9994,
      "step": 2067
    },
    {
      "epoch": 0.38431518305147744,
      "grad_norm": 0.6472446322441101,
      "learning_rate": 0.0001826419170124975,
      "loss": 1.1945,
      "step": 2068
    },
    {
      "epoch": 0.38450102211484855,
      "grad_norm": 0.5857232213020325,
      "learning_rate": 0.0001826254464273584,
      "loss": 1.2508,
      "step": 2069
    },
    {
      "epoch": 0.38468686117821965,
      "grad_norm": 0.6424605250358582,
      "learning_rate": 0.0001826089687750863,
      "loss": 0.9354,
      "step": 2070
    },
    {
      "epoch": 0.38487270024159076,
      "grad_norm": 0.5561656951904297,
      "learning_rate": 0.00018259248405709056,
      "loss": 1.1487,
      "step": 2071
    },
    {
      "epoch": 0.3850585393049619,
      "grad_norm": 0.6111963391304016,
      "learning_rate": 0.00018257599227478112,
      "loss": 1.0517,
      "step": 2072
    },
    {
      "epoch": 0.38524437836833303,
      "grad_norm": 0.6646915674209595,
      "learning_rate": 0.00018255949342956863,
      "loss": 0.7623,
      "step": 2073
    },
    {
      "epoch": 0.38543021743170414,
      "grad_norm": 0.4870254397392273,
      "learning_rate": 0.00018254298752286424,
      "loss": 0.842,
      "step": 2074
    },
    {
      "epoch": 0.38561605649507524,
      "grad_norm": 0.47323906421661377,
      "learning_rate": 0.0001825264745560797,
      "loss": 1.0857,
      "step": 2075
    },
    {
      "epoch": 0.3858018955584464,
      "grad_norm": 0.6508793234825134,
      "learning_rate": 0.00018250995453062742,
      "loss": 1.1488,
      "step": 2076
    },
    {
      "epoch": 0.3859877346218175,
      "grad_norm": 0.7815990447998047,
      "learning_rate": 0.00018249342744792045,
      "loss": 1.1072,
      "step": 2077
    },
    {
      "epoch": 0.3861735736851886,
      "grad_norm": 0.5861072540283203,
      "learning_rate": 0.0001824768933093723,
      "loss": 0.9545,
      "step": 2078
    },
    {
      "epoch": 0.3863594127485597,
      "grad_norm": 0.5428565144538879,
      "learning_rate": 0.00018246035211639722,
      "loss": 1.0341,
      "step": 2079
    },
    {
      "epoch": 0.3865452518119309,
      "grad_norm": 0.6659338474273682,
      "learning_rate": 0.00018244380387040998,
      "loss": 1.1575,
      "step": 2080
    },
    {
      "epoch": 0.386731090875302,
      "grad_norm": 0.6334238052368164,
      "learning_rate": 0.00018242724857282604,
      "loss": 1.108,
      "step": 2081
    },
    {
      "epoch": 0.3869169299386731,
      "grad_norm": 0.6278782486915588,
      "learning_rate": 0.00018241068622506138,
      "loss": 1.1232,
      "step": 2082
    },
    {
      "epoch": 0.3871027690020442,
      "grad_norm": 0.6690854430198669,
      "learning_rate": 0.0001823941168285326,
      "loss": 1.1608,
      "step": 2083
    },
    {
      "epoch": 0.3872886080654154,
      "grad_norm": 0.6730903387069702,
      "learning_rate": 0.00018237754038465692,
      "loss": 1.2147,
      "step": 2084
    },
    {
      "epoch": 0.3874744471287865,
      "grad_norm": 0.5820775628089905,
      "learning_rate": 0.0001823609568948522,
      "loss": 0.8705,
      "step": 2085
    },
    {
      "epoch": 0.3876602861921576,
      "grad_norm": 0.630463182926178,
      "learning_rate": 0.00018234436636053683,
      "loss": 0.9977,
      "step": 2086
    },
    {
      "epoch": 0.3878461252555287,
      "grad_norm": 0.5832675099372864,
      "learning_rate": 0.0001823277687831298,
      "loss": 1.0188,
      "step": 2087
    },
    {
      "epoch": 0.38803196431889986,
      "grad_norm": 0.6170270442962646,
      "learning_rate": 0.0001823111641640508,
      "loss": 1.2257,
      "step": 2088
    },
    {
      "epoch": 0.38821780338227097,
      "grad_norm": 0.5397138595581055,
      "learning_rate": 0.00018229455250472003,
      "loss": 1.0039,
      "step": 2089
    },
    {
      "epoch": 0.38840364244564207,
      "grad_norm": 0.6032971739768982,
      "learning_rate": 0.00018227793380655834,
      "loss": 1.1537,
      "step": 2090
    },
    {
      "epoch": 0.3885894815090132,
      "grad_norm": 0.725053608417511,
      "learning_rate": 0.0001822613080709871,
      "loss": 1.1994,
      "step": 2091
    },
    {
      "epoch": 0.38877532057238434,
      "grad_norm": 0.5189559459686279,
      "learning_rate": 0.0001822446752994284,
      "loss": 0.9984,
      "step": 2092
    },
    {
      "epoch": 0.38896115963575545,
      "grad_norm": 0.5911458134651184,
      "learning_rate": 0.0001822280354933049,
      "loss": 1.1677,
      "step": 2093
    },
    {
      "epoch": 0.38914699869912656,
      "grad_norm": 0.5729151368141174,
      "learning_rate": 0.0001822113886540398,
      "loss": 1.0192,
      "step": 2094
    },
    {
      "epoch": 0.38933283776249766,
      "grad_norm": 0.5103919506072998,
      "learning_rate": 0.0001821947347830569,
      "loss": 0.9939,
      "step": 2095
    },
    {
      "epoch": 0.38951867682586877,
      "grad_norm": 0.6922236084938049,
      "learning_rate": 0.00018217807388178076,
      "loss": 0.8513,
      "step": 2096
    },
    {
      "epoch": 0.38970451588923993,
      "grad_norm": 0.863508939743042,
      "learning_rate": 0.00018216140595163629,
      "loss": 0.7811,
      "step": 2097
    },
    {
      "epoch": 0.38989035495261104,
      "grad_norm": 0.5084664225578308,
      "learning_rate": 0.00018214473099404922,
      "loss": 0.9129,
      "step": 2098
    },
    {
      "epoch": 0.39007619401598215,
      "grad_norm": 2.070678949356079,
      "learning_rate": 0.00018212804901044578,
      "loss": 2.9204,
      "step": 2099
    },
    {
      "epoch": 0.39026203307935325,
      "grad_norm": 0.4668201506137848,
      "learning_rate": 0.00018211136000225277,
      "loss": 0.7666,
      "step": 2100
    },
    {
      "epoch": 0.3904478721427244,
      "grad_norm": 0.5377861261367798,
      "learning_rate": 0.0001820946639708977,
      "loss": 1.0295,
      "step": 2101
    },
    {
      "epoch": 0.3906337112060955,
      "grad_norm": 0.6034281849861145,
      "learning_rate": 0.0001820779609178086,
      "loss": 1.0837,
      "step": 2102
    },
    {
      "epoch": 0.39081955026946663,
      "grad_norm": 0.5661458373069763,
      "learning_rate": 0.0001820612508444141,
      "loss": 0.9965,
      "step": 2103
    },
    {
      "epoch": 0.39100538933283774,
      "grad_norm": 2.1413791179656982,
      "learning_rate": 0.00018204453375214344,
      "loss": 3.3321,
      "step": 2104
    },
    {
      "epoch": 0.3911912283962089,
      "grad_norm": 0.5809323191642761,
      "learning_rate": 0.0001820278096424265,
      "loss": 0.9947,
      "step": 2105
    },
    {
      "epoch": 0.39137706745958,
      "grad_norm": 0.47406235337257385,
      "learning_rate": 0.00018201107851669373,
      "loss": 0.9195,
      "step": 2106
    },
    {
      "epoch": 0.3915629065229511,
      "grad_norm": 0.686505913734436,
      "learning_rate": 0.00018199434037637614,
      "loss": 1.0664,
      "step": 2107
    },
    {
      "epoch": 0.3917487455863222,
      "grad_norm": 0.5684332251548767,
      "learning_rate": 0.00018197759522290544,
      "loss": 1.0698,
      "step": 2108
    },
    {
      "epoch": 0.3919345846496934,
      "grad_norm": 0.6020855903625488,
      "learning_rate": 0.0001819608430577138,
      "loss": 1.0924,
      "step": 2109
    },
    {
      "epoch": 0.3921204237130645,
      "grad_norm": 0.6147074103355408,
      "learning_rate": 0.00018194408388223416,
      "loss": 1.1372,
      "step": 2110
    },
    {
      "epoch": 0.3923062627764356,
      "grad_norm": 0.662574291229248,
      "learning_rate": 0.00018192731769789988,
      "loss": 1.2889,
      "step": 2111
    },
    {
      "epoch": 0.3924921018398067,
      "grad_norm": 0.6683513522148132,
      "learning_rate": 0.00018191054450614503,
      "loss": 1.0685,
      "step": 2112
    },
    {
      "epoch": 0.39267794090317787,
      "grad_norm": 0.6317805051803589,
      "learning_rate": 0.00018189376430840436,
      "loss": 1.1599,
      "step": 2113
    },
    {
      "epoch": 0.392863779966549,
      "grad_norm": 0.5959762930870056,
      "learning_rate": 0.00018187697710611298,
      "loss": 1.1292,
      "step": 2114
    },
    {
      "epoch": 0.3930496190299201,
      "grad_norm": 0.5756658315658569,
      "learning_rate": 0.00018186018290070676,
      "loss": 1.1887,
      "step": 2115
    },
    {
      "epoch": 0.3932354580932912,
      "grad_norm": 0.5529783368110657,
      "learning_rate": 0.0001818433816936222,
      "loss": 1.0081,
      "step": 2116
    },
    {
      "epoch": 0.39342129715666235,
      "grad_norm": 0.6137245297431946,
      "learning_rate": 0.00018182657348629632,
      "loss": 0.8836,
      "step": 2117
    },
    {
      "epoch": 0.39360713622003346,
      "grad_norm": 0.5938009023666382,
      "learning_rate": 0.00018180975828016677,
      "loss": 1.2131,
      "step": 2118
    },
    {
      "epoch": 0.39379297528340457,
      "grad_norm": 0.6098518967628479,
      "learning_rate": 0.00018179293607667178,
      "loss": 0.9022,
      "step": 2119
    },
    {
      "epoch": 0.3939788143467757,
      "grad_norm": 0.5324256420135498,
      "learning_rate": 0.00018177610687725016,
      "loss": 1.1064,
      "step": 2120
    },
    {
      "epoch": 0.39416465341014684,
      "grad_norm": 0.5962520837783813,
      "learning_rate": 0.00018175927068334142,
      "loss": 1.1095,
      "step": 2121
    },
    {
      "epoch": 0.39435049247351794,
      "grad_norm": 0.6412689685821533,
      "learning_rate": 0.00018174242749638552,
      "loss": 1.3036,
      "step": 2122
    },
    {
      "epoch": 0.39453633153688905,
      "grad_norm": 0.537754476070404,
      "learning_rate": 0.00018172557731782314,
      "loss": 1.0712,
      "step": 2123
    },
    {
      "epoch": 0.39472217060026016,
      "grad_norm": 0.6352578401565552,
      "learning_rate": 0.0001817087201490955,
      "loss": 1.19,
      "step": 2124
    },
    {
      "epoch": 0.3949080096636313,
      "grad_norm": 0.5872593522071838,
      "learning_rate": 0.00018169185599164442,
      "loss": 1.1057,
      "step": 2125
    },
    {
      "epoch": 0.39509384872700243,
      "grad_norm": 0.5860781669616699,
      "learning_rate": 0.00018167498484691238,
      "loss": 1.0346,
      "step": 2126
    },
    {
      "epoch": 0.39527968779037354,
      "grad_norm": 0.5678336024284363,
      "learning_rate": 0.00018165810671634232,
      "loss": 0.9507,
      "step": 2127
    },
    {
      "epoch": 0.39546552685374464,
      "grad_norm": 0.4740240275859833,
      "learning_rate": 0.00018164122160137792,
      "loss": 1.143,
      "step": 2128
    },
    {
      "epoch": 0.3956513659171158,
      "grad_norm": 0.6389288902282715,
      "learning_rate": 0.00018162432950346338,
      "loss": 1.1391,
      "step": 2129
    },
    {
      "epoch": 0.3958372049804869,
      "grad_norm": 0.6023207306861877,
      "learning_rate": 0.00018160743042404352,
      "loss": 0.9583,
      "step": 2130
    },
    {
      "epoch": 0.396023044043858,
      "grad_norm": 0.5941089987754822,
      "learning_rate": 0.00018159052436456373,
      "loss": 1.2258,
      "step": 2131
    },
    {
      "epoch": 0.3962088831072291,
      "grad_norm": 0.5679216384887695,
      "learning_rate": 0.0001815736113264701,
      "loss": 0.9775,
      "step": 2132
    },
    {
      "epoch": 0.39639472217060023,
      "grad_norm": 0.5501899719238281,
      "learning_rate": 0.00018155669131120919,
      "loss": 1.2804,
      "step": 2133
    },
    {
      "epoch": 0.3965805612339714,
      "grad_norm": 0.5905486941337585,
      "learning_rate": 0.00018153976432022816,
      "loss": 0.9472,
      "step": 2134
    },
    {
      "epoch": 0.3967664002973425,
      "grad_norm": 0.7558655142784119,
      "learning_rate": 0.00018152283035497492,
      "loss": 1.3491,
      "step": 2135
    },
    {
      "epoch": 0.3969522393607136,
      "grad_norm": 0.6427619457244873,
      "learning_rate": 0.00018150588941689776,
      "loss": 1.1921,
      "step": 2136
    },
    {
      "epoch": 0.3971380784240847,
      "grad_norm": 0.6187041997909546,
      "learning_rate": 0.00018148894150744573,
      "loss": 1.1544,
      "step": 2137
    },
    {
      "epoch": 0.3973239174874559,
      "grad_norm": 0.5489941239356995,
      "learning_rate": 0.00018147198662806844,
      "loss": 1.1024,
      "step": 2138
    },
    {
      "epoch": 0.397509756550827,
      "grad_norm": 0.5993608832359314,
      "learning_rate": 0.000181455024780216,
      "loss": 0.9939,
      "step": 2139
    },
    {
      "epoch": 0.3976955956141981,
      "grad_norm": 0.5785315036773682,
      "learning_rate": 0.00018143805596533926,
      "loss": 1.1785,
      "step": 2140
    },
    {
      "epoch": 0.3978814346775692,
      "grad_norm": 0.5734258890151978,
      "learning_rate": 0.0001814210801848896,
      "loss": 1.1233,
      "step": 2141
    },
    {
      "epoch": 0.39806727374094036,
      "grad_norm": 0.6431028842926025,
      "learning_rate": 0.00018140409744031897,
      "loss": 1.0709,
      "step": 2142
    },
    {
      "epoch": 0.39825311280431147,
      "grad_norm": 0.553656816482544,
      "learning_rate": 0.00018138710773307995,
      "loss": 1.0857,
      "step": 2143
    },
    {
      "epoch": 0.3984389518676826,
      "grad_norm": 0.639657199382782,
      "learning_rate": 0.0001813701110646257,
      "loss": 1.2033,
      "step": 2144
    },
    {
      "epoch": 0.3986247909310537,
      "grad_norm": 0.5474963784217834,
      "learning_rate": 0.00018135310743641,
      "loss": 1.1524,
      "step": 2145
    },
    {
      "epoch": 0.39881062999442485,
      "grad_norm": 0.5809098482131958,
      "learning_rate": 0.00018133609684988717,
      "loss": 1.3077,
      "step": 2146
    },
    {
      "epoch": 0.39899646905779595,
      "grad_norm": 0.9639946222305298,
      "learning_rate": 0.00018131907930651222,
      "loss": 1.1468,
      "step": 2147
    },
    {
      "epoch": 0.39918230812116706,
      "grad_norm": 0.676054835319519,
      "learning_rate": 0.00018130205480774061,
      "loss": 1.0494,
      "step": 2148
    },
    {
      "epoch": 0.39936814718453817,
      "grad_norm": 0.5698658227920532,
      "learning_rate": 0.0001812850233550286,
      "loss": 1.0787,
      "step": 2149
    },
    {
      "epoch": 0.39955398624790933,
      "grad_norm": 0.5421280264854431,
      "learning_rate": 0.0001812679849498328,
      "loss": 0.9483,
      "step": 2150
    },
    {
      "epoch": 0.39973982531128044,
      "grad_norm": 0.6609428524971008,
      "learning_rate": 0.00018125093959361067,
      "loss": 1.1642,
      "step": 2151
    },
    {
      "epoch": 0.39992566437465155,
      "grad_norm": 0.5440996289253235,
      "learning_rate": 0.00018123388728782,
      "loss": 1.2694,
      "step": 2152
    },
    {
      "epoch": 0.40011150343802265,
      "grad_norm": 0.5741338133811951,
      "learning_rate": 0.00018121682803391942,
      "loss": 1.236,
      "step": 2153
    },
    {
      "epoch": 0.4002973425013938,
      "grad_norm": 0.6801538467407227,
      "learning_rate": 0.000181199761833368,
      "loss": 0.6522,
      "step": 2154
    },
    {
      "epoch": 0.4004831815647649,
      "grad_norm": 0.6567083597183228,
      "learning_rate": 0.00018118268868762546,
      "loss": 1.3321,
      "step": 2155
    },
    {
      "epoch": 0.40066902062813603,
      "grad_norm": 0.6132495403289795,
      "learning_rate": 0.00018116560859815208,
      "loss": 0.9719,
      "step": 2156
    },
    {
      "epoch": 0.40085485969150714,
      "grad_norm": 0.582120954990387,
      "learning_rate": 0.00018114852156640878,
      "loss": 0.8756,
      "step": 2157
    },
    {
      "epoch": 0.4010406987548783,
      "grad_norm": 0.715114414691925,
      "learning_rate": 0.00018113142759385706,
      "loss": 1.1517,
      "step": 2158
    },
    {
      "epoch": 0.4012265378182494,
      "grad_norm": 0.5721078515052795,
      "learning_rate": 0.000181114326681959,
      "loss": 0.8977,
      "step": 2159
    },
    {
      "epoch": 0.4014123768816205,
      "grad_norm": 0.5104873776435852,
      "learning_rate": 0.00018109721883217725,
      "loss": 1.0237,
      "step": 2160
    },
    {
      "epoch": 0.4015982159449916,
      "grad_norm": 0.5614782571792603,
      "learning_rate": 0.0001810801040459751,
      "loss": 0.8917,
      "step": 2161
    },
    {
      "epoch": 0.4017840550083628,
      "grad_norm": 0.5537241101264954,
      "learning_rate": 0.0001810629823248164,
      "loss": 1.2027,
      "step": 2162
    },
    {
      "epoch": 0.4019698940717339,
      "grad_norm": 0.7075382471084595,
      "learning_rate": 0.00018104585367016563,
      "loss": 1.0896,
      "step": 2163
    },
    {
      "epoch": 0.402155733135105,
      "grad_norm": 0.503906786441803,
      "learning_rate": 0.00018102871808348784,
      "loss": 0.7954,
      "step": 2164
    },
    {
      "epoch": 0.4023415721984761,
      "grad_norm": 0.5689190626144409,
      "learning_rate": 0.00018101157556624865,
      "loss": 1.2011,
      "step": 2165
    },
    {
      "epoch": 0.40252741126184727,
      "grad_norm": 0.5936292409896851,
      "learning_rate": 0.00018099442611991436,
      "loss": 1.1824,
      "step": 2166
    },
    {
      "epoch": 0.4027132503252184,
      "grad_norm": 0.5619094371795654,
      "learning_rate": 0.0001809772697459517,
      "loss": 1.0462,
      "step": 2167
    },
    {
      "epoch": 0.4028990893885895,
      "grad_norm": 0.4826868772506714,
      "learning_rate": 0.00018096010644582817,
      "loss": 1.0125,
      "step": 2168
    },
    {
      "epoch": 0.4030849284519606,
      "grad_norm": 0.633266806602478,
      "learning_rate": 0.00018094293622101173,
      "loss": 1.1802,
      "step": 2169
    },
    {
      "epoch": 0.4032707675153317,
      "grad_norm": 0.6388394832611084,
      "learning_rate": 0.00018092575907297106,
      "loss": 1.0138,
      "step": 2170
    },
    {
      "epoch": 0.40345660657870286,
      "grad_norm": 0.5150460600852966,
      "learning_rate": 0.00018090857500317525,
      "loss": 0.9774,
      "step": 2171
    },
    {
      "epoch": 0.40364244564207397,
      "grad_norm": 0.6779277920722961,
      "learning_rate": 0.00018089138401309422,
      "loss": 1.0853,
      "step": 2172
    },
    {
      "epoch": 0.4038282847054451,
      "grad_norm": 0.5205432176589966,
      "learning_rate": 0.00018087418610419827,
      "loss": 0.7876,
      "step": 2173
    },
    {
      "epoch": 0.4040141237688162,
      "grad_norm": 0.4968721568584442,
      "learning_rate": 0.00018085698127795835,
      "loss": 0.7495,
      "step": 2174
    },
    {
      "epoch": 0.40419996283218734,
      "grad_norm": 0.5260478258132935,
      "learning_rate": 0.0001808397695358461,
      "loss": 0.8344,
      "step": 2175
    },
    {
      "epoch": 0.40438580189555845,
      "grad_norm": 0.6795833706855774,
      "learning_rate": 0.00018082255087933363,
      "loss": 1.0586,
      "step": 2176
    },
    {
      "epoch": 0.40457164095892956,
      "grad_norm": 0.5038895010948181,
      "learning_rate": 0.00018080532530989372,
      "loss": 0.8544,
      "step": 2177
    },
    {
      "epoch": 0.40475748002230066,
      "grad_norm": 0.6297782063484192,
      "learning_rate": 0.00018078809282899967,
      "loss": 1.3135,
      "step": 2178
    },
    {
      "epoch": 0.4049433190856718,
      "grad_norm": 0.5927119255065918,
      "learning_rate": 0.00018077085343812545,
      "loss": 1.0277,
      "step": 2179
    },
    {
      "epoch": 0.40512915814904293,
      "grad_norm": 0.5536563992500305,
      "learning_rate": 0.00018075360713874558,
      "loss": 1.0211,
      "step": 2180
    },
    {
      "epoch": 0.40531499721241404,
      "grad_norm": 0.5813071131706238,
      "learning_rate": 0.00018073635393233513,
      "loss": 0.9992,
      "step": 2181
    },
    {
      "epoch": 0.40550083627578515,
      "grad_norm": 0.5180925130844116,
      "learning_rate": 0.00018071909382036985,
      "loss": 1.1269,
      "step": 2182
    },
    {
      "epoch": 0.4056866753391563,
      "grad_norm": 0.5994383692741394,
      "learning_rate": 0.000180701826804326,
      "loss": 1.0403,
      "step": 2183
    },
    {
      "epoch": 0.4058725144025274,
      "grad_norm": 0.6202577948570251,
      "learning_rate": 0.0001806845528856805,
      "loss": 0.9314,
      "step": 2184
    },
    {
      "epoch": 0.4060583534658985,
      "grad_norm": 0.6377980709075928,
      "learning_rate": 0.00018066727206591084,
      "loss": 0.8948,
      "step": 2185
    },
    {
      "epoch": 0.40624419252926963,
      "grad_norm": 0.5450910925865173,
      "learning_rate": 0.000180649984346495,
      "loss": 1.2289,
      "step": 2186
    },
    {
      "epoch": 0.4064300315926408,
      "grad_norm": 0.6690120697021484,
      "learning_rate": 0.00018063268972891172,
      "loss": 1.1064,
      "step": 2187
    },
    {
      "epoch": 0.4066158706560119,
      "grad_norm": 0.9592549800872803,
      "learning_rate": 0.00018061538821464022,
      "loss": 1.4348,
      "step": 2188
    },
    {
      "epoch": 0.406801709719383,
      "grad_norm": 0.5875871181488037,
      "learning_rate": 0.0001805980798051603,
      "loss": 1.0296,
      "step": 2189
    },
    {
      "epoch": 0.4069875487827541,
      "grad_norm": 0.5648946166038513,
      "learning_rate": 0.00018058076450195248,
      "loss": 1.1743,
      "step": 2190
    },
    {
      "epoch": 0.4071733878461253,
      "grad_norm": 0.5465866923332214,
      "learning_rate": 0.0001805634423064977,
      "loss": 1.2059,
      "step": 2191
    },
    {
      "epoch": 0.4073592269094964,
      "grad_norm": 0.6858041882514954,
      "learning_rate": 0.00018054611322027752,
      "loss": 1.1359,
      "step": 2192
    },
    {
      "epoch": 0.4075450659728675,
      "grad_norm": 0.5024155974388123,
      "learning_rate": 0.00018052877724477424,
      "loss": 0.9183,
      "step": 2193
    },
    {
      "epoch": 0.4077309050362386,
      "grad_norm": 1.0227012634277344,
      "learning_rate": 0.00018051143438147058,
      "loss": 1.3601,
      "step": 2194
    },
    {
      "epoch": 0.40791674409960976,
      "grad_norm": 0.6042284369468689,
      "learning_rate": 0.00018049408463184996,
      "loss": 1.1513,
      "step": 2195
    },
    {
      "epoch": 0.40810258316298087,
      "grad_norm": 0.6936082243919373,
      "learning_rate": 0.00018047672799739628,
      "loss": 1.0927,
      "step": 2196
    },
    {
      "epoch": 0.408288422226352,
      "grad_norm": 0.6657347083091736,
      "learning_rate": 0.00018045936447959413,
      "loss": 0.9907,
      "step": 2197
    },
    {
      "epoch": 0.4084742612897231,
      "grad_norm": 0.5620197057723999,
      "learning_rate": 0.00018044199407992863,
      "loss": 1.1356,
      "step": 2198
    },
    {
      "epoch": 0.40866010035309425,
      "grad_norm": 0.5103034973144531,
      "learning_rate": 0.00018042461679988555,
      "loss": 0.9557,
      "step": 2199
    },
    {
      "epoch": 0.40884593941646535,
      "grad_norm": 0.5765652656555176,
      "learning_rate": 0.00018040723264095112,
      "loss": 1.215,
      "step": 2200
    },
    {
      "epoch": 0.40903177847983646,
      "grad_norm": 0.568516194820404,
      "learning_rate": 0.00018038984160461237,
      "loss": 0.9904,
      "step": 2201
    },
    {
      "epoch": 0.40921761754320757,
      "grad_norm": 0.6059340834617615,
      "learning_rate": 0.00018037244369235666,
      "loss": 1.1956,
      "step": 2202
    },
    {
      "epoch": 0.4094034566065787,
      "grad_norm": 0.5740835070610046,
      "learning_rate": 0.00018035503890567216,
      "loss": 1.0662,
      "step": 2203
    },
    {
      "epoch": 0.40958929566994984,
      "grad_norm": 8.388848304748535,
      "learning_rate": 0.00018033762724604753,
      "loss": 3.5506,
      "step": 2204
    },
    {
      "epoch": 0.40977513473332094,
      "grad_norm": 0.6180132031440735,
      "learning_rate": 0.00018032020871497196,
      "loss": 0.9254,
      "step": 2205
    },
    {
      "epoch": 0.40996097379669205,
      "grad_norm": 0.7836258411407471,
      "learning_rate": 0.00018030278331393539,
      "loss": 1.1685,
      "step": 2206
    },
    {
      "epoch": 0.41014681286006316,
      "grad_norm": 0.6342977285385132,
      "learning_rate": 0.0001802853510444282,
      "loss": 1.1987,
      "step": 2207
    },
    {
      "epoch": 0.4103326519234343,
      "grad_norm": 0.4958232045173645,
      "learning_rate": 0.0001802679119079414,
      "loss": 0.7183,
      "step": 2208
    },
    {
      "epoch": 0.41051849098680543,
      "grad_norm": 0.48592817783355713,
      "learning_rate": 0.0001802504659059666,
      "loss": 0.8992,
      "step": 2209
    },
    {
      "epoch": 0.41070433005017654,
      "grad_norm": 0.5504721999168396,
      "learning_rate": 0.00018023301303999602,
      "loss": 0.8599,
      "step": 2210
    },
    {
      "epoch": 0.41089016911354764,
      "grad_norm": 0.5736424326896667,
      "learning_rate": 0.00018021555331152245,
      "loss": 1.0319,
      "step": 2211
    },
    {
      "epoch": 0.4110760081769188,
      "grad_norm": 0.751387357711792,
      "learning_rate": 0.00018019808672203923,
      "loss": 1.0879,
      "step": 2212
    },
    {
      "epoch": 0.4112618472402899,
      "grad_norm": 0.5974605083465576,
      "learning_rate": 0.0001801806132730403,
      "loss": 1.209,
      "step": 2213
    },
    {
      "epoch": 0.411447686303661,
      "grad_norm": 0.516734778881073,
      "learning_rate": 0.00018016313296602022,
      "loss": 1.0002,
      "step": 2214
    },
    {
      "epoch": 0.4116335253670321,
      "grad_norm": 0.5020787715911865,
      "learning_rate": 0.00018014564580247415,
      "loss": 0.9484,
      "step": 2215
    },
    {
      "epoch": 0.4118193644304033,
      "grad_norm": 0.5782486796379089,
      "learning_rate": 0.00018012815178389775,
      "loss": 1.1124,
      "step": 2216
    },
    {
      "epoch": 0.4120052034937744,
      "grad_norm": 0.6170997619628906,
      "learning_rate": 0.00018011065091178737,
      "loss": 1.0163,
      "step": 2217
    },
    {
      "epoch": 0.4121910425571455,
      "grad_norm": 0.5772647261619568,
      "learning_rate": 0.00018009314318763984,
      "loss": 1.1784,
      "step": 2218
    },
    {
      "epoch": 0.4123768816205166,
      "grad_norm": 0.5786311626434326,
      "learning_rate": 0.00018007562861295273,
      "loss": 1.0668,
      "step": 2219
    },
    {
      "epoch": 0.4125627206838878,
      "grad_norm": 0.602414071559906,
      "learning_rate": 0.000180058107189224,
      "loss": 1.2294,
      "step": 2220
    },
    {
      "epoch": 0.4127485597472589,
      "grad_norm": 0.5060234069824219,
      "learning_rate": 0.00018004057891795236,
      "loss": 0.9069,
      "step": 2221
    },
    {
      "epoch": 0.41293439881063,
      "grad_norm": 0.7384883761405945,
      "learning_rate": 0.000180023043800637,
      "loss": 1.0488,
      "step": 2222
    },
    {
      "epoch": 0.4131202378740011,
      "grad_norm": 0.5549197196960449,
      "learning_rate": 0.0001800055018387777,
      "loss": 1.0974,
      "step": 2223
    },
    {
      "epoch": 0.41330607693737226,
      "grad_norm": 0.6316516995429993,
      "learning_rate": 0.00017998795303387496,
      "loss": 1.2475,
      "step": 2224
    },
    {
      "epoch": 0.41349191600074336,
      "grad_norm": 0.6400758028030396,
      "learning_rate": 0.00017997039738742974,
      "loss": 1.186,
      "step": 2225
    },
    {
      "epoch": 0.41367775506411447,
      "grad_norm": 0.637119710445404,
      "learning_rate": 0.00017995283490094356,
      "loss": 1.2223,
      "step": 2226
    },
    {
      "epoch": 0.4138635941274856,
      "grad_norm": 0.5269392728805542,
      "learning_rate": 0.00017993526557591864,
      "loss": 1.0976,
      "step": 2227
    },
    {
      "epoch": 0.41404943319085674,
      "grad_norm": 0.6631988883018494,
      "learning_rate": 0.00017991768941385766,
      "loss": 1.0029,
      "step": 2228
    },
    {
      "epoch": 0.41423527225422785,
      "grad_norm": 0.6144427061080933,
      "learning_rate": 0.000179900106416264,
      "loss": 0.903,
      "step": 2229
    },
    {
      "epoch": 0.41442111131759896,
      "grad_norm": 2.708913564682007,
      "learning_rate": 0.00017988251658464154,
      "loss": 2.9746,
      "step": 2230
    },
    {
      "epoch": 0.41460695038097006,
      "grad_norm": 0.5651253461837769,
      "learning_rate": 0.0001798649199204948,
      "loss": 0.9768,
      "step": 2231
    },
    {
      "epoch": 0.4147927894443412,
      "grad_norm": 0.5956056118011475,
      "learning_rate": 0.00017984731642532882,
      "loss": 0.9158,
      "step": 2232
    },
    {
      "epoch": 0.41497862850771233,
      "grad_norm": 0.6976494193077087,
      "learning_rate": 0.00017982970610064933,
      "loss": 1.1362,
      "step": 2233
    },
    {
      "epoch": 0.41516446757108344,
      "grad_norm": 0.6097948551177979,
      "learning_rate": 0.00017981208894796253,
      "loss": 0.9367,
      "step": 2234
    },
    {
      "epoch": 0.41535030663445455,
      "grad_norm": 0.6046789288520813,
      "learning_rate": 0.00017979446496877528,
      "loss": 1.3803,
      "step": 2235
    },
    {
      "epoch": 0.4155361456978257,
      "grad_norm": 0.5067242383956909,
      "learning_rate": 0.00017977683416459494,
      "loss": 0.7473,
      "step": 2236
    },
    {
      "epoch": 0.4157219847611968,
      "grad_norm": 0.5810707807540894,
      "learning_rate": 0.0001797591965369296,
      "loss": 1.0738,
      "step": 2237
    },
    {
      "epoch": 0.4159078238245679,
      "grad_norm": 0.4900633990764618,
      "learning_rate": 0.00017974155208728778,
      "loss": 1.023,
      "step": 2238
    },
    {
      "epoch": 0.41609366288793903,
      "grad_norm": 2.034982204437256,
      "learning_rate": 0.00017972390081717865,
      "loss": 2.4168,
      "step": 2239
    },
    {
      "epoch": 0.41627950195131014,
      "grad_norm": 0.5499010682106018,
      "learning_rate": 0.000179706242728112,
      "loss": 1.1621,
      "step": 2240
    },
    {
      "epoch": 0.4164653410146813,
      "grad_norm": 0.6297591328620911,
      "learning_rate": 0.0001796885778215981,
      "loss": 1.1059,
      "step": 2241
    },
    {
      "epoch": 0.4166511800780524,
      "grad_norm": 0.601097822189331,
      "learning_rate": 0.00017967090609914796,
      "loss": 1.0177,
      "step": 2242
    },
    {
      "epoch": 0.4168370191414235,
      "grad_norm": 0.6372743844985962,
      "learning_rate": 0.000179653227562273,
      "loss": 1.1184,
      "step": 2243
    },
    {
      "epoch": 0.4170228582047946,
      "grad_norm": 0.6114941239356995,
      "learning_rate": 0.00017963554221248534,
      "loss": 1.0708,
      "step": 2244
    },
    {
      "epoch": 0.4172086972681658,
      "grad_norm": 0.5649812817573547,
      "learning_rate": 0.00017961785005129766,
      "loss": 0.8418,
      "step": 2245
    },
    {
      "epoch": 0.4173945363315369,
      "grad_norm": 0.632774829864502,
      "learning_rate": 0.00017960015108022318,
      "loss": 1.0592,
      "step": 2246
    },
    {
      "epoch": 0.417580375394908,
      "grad_norm": 0.6083325743675232,
      "learning_rate": 0.00017958244530077573,
      "loss": 1.0943,
      "step": 2247
    },
    {
      "epoch": 0.4177662144582791,
      "grad_norm": 0.6060488224029541,
      "learning_rate": 0.00017956473271446972,
      "loss": 1.0025,
      "step": 2248
    },
    {
      "epoch": 0.41795205352165027,
      "grad_norm": 0.6035066843032837,
      "learning_rate": 0.00017954701332282018,
      "loss": 1.2486,
      "step": 2249
    },
    {
      "epoch": 0.4181378925850214,
      "grad_norm": 0.536759614944458,
      "learning_rate": 0.00017952928712734268,
      "loss": 0.9172,
      "step": 2250
    },
    {
      "epoch": 0.4183237316483925,
      "grad_norm": 0.6202712059020996,
      "learning_rate": 0.00017951155412955331,
      "loss": 1.1223,
      "step": 2251
    },
    {
      "epoch": 0.4185095707117636,
      "grad_norm": 0.5908938646316528,
      "learning_rate": 0.00017949381433096892,
      "loss": 1.2079,
      "step": 2252
    },
    {
      "epoch": 0.41869540977513475,
      "grad_norm": 0.6541932225227356,
      "learning_rate": 0.00017947606773310678,
      "loss": 1.211,
      "step": 2253
    },
    {
      "epoch": 0.41888124883850586,
      "grad_norm": 0.5047724843025208,
      "learning_rate": 0.00017945831433748477,
      "loss": 0.8668,
      "step": 2254
    },
    {
      "epoch": 0.41906708790187697,
      "grad_norm": 0.616878867149353,
      "learning_rate": 0.0001794405541456214,
      "loss": 1.1663,
      "step": 2255
    },
    {
      "epoch": 0.4192529269652481,
      "grad_norm": 0.5981701016426086,
      "learning_rate": 0.00017942278715903575,
      "loss": 1.1325,
      "step": 2256
    },
    {
      "epoch": 0.41943876602861924,
      "grad_norm": 0.5608638525009155,
      "learning_rate": 0.0001794050133792475,
      "loss": 0.9859,
      "step": 2257
    },
    {
      "epoch": 0.41962460509199034,
      "grad_norm": 0.7004113793373108,
      "learning_rate": 0.00017938723280777678,
      "loss": 1.0358,
      "step": 2258
    },
    {
      "epoch": 0.41981044415536145,
      "grad_norm": 0.5976504683494568,
      "learning_rate": 0.00017936944544614449,
      "loss": 1.0715,
      "step": 2259
    },
    {
      "epoch": 0.41999628321873256,
      "grad_norm": 0.49607938528060913,
      "learning_rate": 0.00017935165129587198,
      "loss": 1.0598,
      "step": 2260
    },
    {
      "epoch": 0.4201821222821037,
      "grad_norm": 0.6711708903312683,
      "learning_rate": 0.00017933385035848125,
      "loss": 1.3064,
      "step": 2261
    },
    {
      "epoch": 0.4203679613454748,
      "grad_norm": 0.6147506237030029,
      "learning_rate": 0.00017931604263549483,
      "loss": 1.1359,
      "step": 2262
    },
    {
      "epoch": 0.42055380040884593,
      "grad_norm": 0.5291450619697571,
      "learning_rate": 0.00017929822812843584,
      "loss": 1.0914,
      "step": 2263
    },
    {
      "epoch": 0.42073963947221704,
      "grad_norm": 0.5078874826431274,
      "learning_rate": 0.00017928040683882804,
      "loss": 1.0915,
      "step": 2264
    },
    {
      "epoch": 0.4209254785355882,
      "grad_norm": 0.5273508429527283,
      "learning_rate": 0.00017926257876819566,
      "loss": 1.1543,
      "step": 2265
    },
    {
      "epoch": 0.4211113175989593,
      "grad_norm": 0.6677791476249695,
      "learning_rate": 0.0001792447439180636,
      "loss": 0.9763,
      "step": 2266
    },
    {
      "epoch": 0.4212971566623304,
      "grad_norm": 0.5593759417533875,
      "learning_rate": 0.00017922690228995737,
      "loss": 1.1768,
      "step": 2267
    },
    {
      "epoch": 0.4214829957257015,
      "grad_norm": 0.6982876062393188,
      "learning_rate": 0.00017920905388540295,
      "loss": 1.1277,
      "step": 2268
    },
    {
      "epoch": 0.4216688347890727,
      "grad_norm": 0.5563878417015076,
      "learning_rate": 0.00017919119870592694,
      "loss": 0.964,
      "step": 2269
    },
    {
      "epoch": 0.4218546738524438,
      "grad_norm": 0.7023888230323792,
      "learning_rate": 0.00017917333675305654,
      "loss": 1.2069,
      "step": 2270
    },
    {
      "epoch": 0.4220405129158149,
      "grad_norm": 0.8888300061225891,
      "learning_rate": 0.00017915546802831957,
      "loss": 1.1435,
      "step": 2271
    },
    {
      "epoch": 0.422226351979186,
      "grad_norm": 0.6172018647193909,
      "learning_rate": 0.00017913759253324432,
      "loss": 1.2328,
      "step": 2272
    },
    {
      "epoch": 0.4224121910425572,
      "grad_norm": 0.6272783875465393,
      "learning_rate": 0.00017911971026935973,
      "loss": 1.0586,
      "step": 2273
    },
    {
      "epoch": 0.4225980301059283,
      "grad_norm": 0.6125234961509705,
      "learning_rate": 0.00017910182123819532,
      "loss": 1.0063,
      "step": 2274
    },
    {
      "epoch": 0.4227838691692994,
      "grad_norm": 0.6082481741905212,
      "learning_rate": 0.0001790839254412812,
      "loss": 1.1442,
      "step": 2275
    },
    {
      "epoch": 0.4229697082326705,
      "grad_norm": 0.6373506188392639,
      "learning_rate": 0.000179066022880148,
      "loss": 1.0645,
      "step": 2276
    },
    {
      "epoch": 0.4231555472960416,
      "grad_norm": 0.5189694166183472,
      "learning_rate": 0.00017904811355632698,
      "loss": 1.2191,
      "step": 2277
    },
    {
      "epoch": 0.42334138635941276,
      "grad_norm": 0.5406830906867981,
      "learning_rate": 0.00017903019747134998,
      "loss": 1.0508,
      "step": 2278
    },
    {
      "epoch": 0.42352722542278387,
      "grad_norm": 0.5542166829109192,
      "learning_rate": 0.00017901227462674933,
      "loss": 1.0213,
      "step": 2279
    },
    {
      "epoch": 0.423713064486155,
      "grad_norm": 0.6017323136329651,
      "learning_rate": 0.0001789943450240581,
      "loss": 1.0446,
      "step": 2280
    },
    {
      "epoch": 0.4238989035495261,
      "grad_norm": 0.46722128987312317,
      "learning_rate": 0.00017897640866480983,
      "loss": 0.731,
      "step": 2281
    },
    {
      "epoch": 0.42408474261289725,
      "grad_norm": 0.6106616854667664,
      "learning_rate": 0.0001789584655505386,
      "loss": 1.2458,
      "step": 2282
    },
    {
      "epoch": 0.42427058167626835,
      "grad_norm": 0.7112687230110168,
      "learning_rate": 0.00017894051568277918,
      "loss": 1.3797,
      "step": 2283
    },
    {
      "epoch": 0.42445642073963946,
      "grad_norm": 0.5603358149528503,
      "learning_rate": 0.00017892255906306684,
      "loss": 1.1287,
      "step": 2284
    },
    {
      "epoch": 0.42464225980301057,
      "grad_norm": 0.5825716853141785,
      "learning_rate": 0.00017890459569293746,
      "loss": 1.2238,
      "step": 2285
    },
    {
      "epoch": 0.42482809886638173,
      "grad_norm": 0.6225450038909912,
      "learning_rate": 0.00017888662557392744,
      "loss": 1.1294,
      "step": 2286
    },
    {
      "epoch": 0.42501393792975284,
      "grad_norm": 0.5144187808036804,
      "learning_rate": 0.00017886864870757387,
      "loss": 1.0482,
      "step": 2287
    },
    {
      "epoch": 0.42519977699312395,
      "grad_norm": 0.6005070209503174,
      "learning_rate": 0.0001788506650954143,
      "loss": 0.9388,
      "step": 2288
    },
    {
      "epoch": 0.42538561605649505,
      "grad_norm": 0.7564705014228821,
      "learning_rate": 0.00017883267473898693,
      "loss": 1.1563,
      "step": 2289
    },
    {
      "epoch": 0.4255714551198662,
      "grad_norm": 0.600664496421814,
      "learning_rate": 0.00017881467763983052,
      "loss": 1.2571,
      "step": 2290
    },
    {
      "epoch": 0.4257572941832373,
      "grad_norm": 0.5442487001419067,
      "learning_rate": 0.0001787966737994844,
      "loss": 1.1045,
      "step": 2291
    },
    {
      "epoch": 0.42594313324660843,
      "grad_norm": 0.3654262125492096,
      "learning_rate": 0.00017877866321948843,
      "loss": 0.6109,
      "step": 2292
    },
    {
      "epoch": 0.42612897230997954,
      "grad_norm": 0.605783224105835,
      "learning_rate": 0.00017876064590138316,
      "loss": 1.3888,
      "step": 2293
    },
    {
      "epoch": 0.4263148113733507,
      "grad_norm": 0.5427509546279907,
      "learning_rate": 0.00017874262184670961,
      "loss": 1.1819,
      "step": 2294
    },
    {
      "epoch": 0.4265006504367218,
      "grad_norm": 0.5091795921325684,
      "learning_rate": 0.00017872459105700944,
      "loss": 1.0144,
      "step": 2295
    },
    {
      "epoch": 0.4266864895000929,
      "grad_norm": 0.7157513499259949,
      "learning_rate": 0.00017870655353382486,
      "loss": 1.02,
      "step": 2296
    },
    {
      "epoch": 0.426872328563464,
      "grad_norm": 0.6378082633018494,
      "learning_rate": 0.00017868850927869862,
      "loss": 1.0082,
      "step": 2297
    },
    {
      "epoch": 0.4270581676268352,
      "grad_norm": 0.5989370942115784,
      "learning_rate": 0.00017867045829317416,
      "loss": 1.0008,
      "step": 2298
    },
    {
      "epoch": 0.4272440066902063,
      "grad_norm": 0.5860329866409302,
      "learning_rate": 0.00017865240057879532,
      "loss": 1.1185,
      "step": 2299
    },
    {
      "epoch": 0.4274298457535774,
      "grad_norm": 0.6344106793403625,
      "learning_rate": 0.0001786343361371067,
      "loss": 1.0703,
      "step": 2300
    },
    {
      "epoch": 0.4276156848169485,
      "grad_norm": 0.6763262152671814,
      "learning_rate": 0.00017861626496965333,
      "loss": 1.2203,
      "step": 2301
    },
    {
      "epoch": 0.42780152388031967,
      "grad_norm": 0.6049442887306213,
      "learning_rate": 0.00017859818707798093,
      "loss": 1.0438,
      "step": 2302
    },
    {
      "epoch": 0.4279873629436908,
      "grad_norm": 0.5617985725402832,
      "learning_rate": 0.0001785801024636357,
      "loss": 1.1025,
      "step": 2303
    },
    {
      "epoch": 0.4281732020070619,
      "grad_norm": 0.7062197923660278,
      "learning_rate": 0.00017856201112816447,
      "loss": 1.3451,
      "step": 2304
    },
    {
      "epoch": 0.428359041070433,
      "grad_norm": 0.6399662494659424,
      "learning_rate": 0.00017854391307311468,
      "loss": 0.9795,
      "step": 2305
    },
    {
      "epoch": 0.42854488013380415,
      "grad_norm": 0.5584330558776855,
      "learning_rate": 0.0001785258083000342,
      "loss": 1.1063,
      "step": 2306
    },
    {
      "epoch": 0.42873071919717526,
      "grad_norm": 0.525244414806366,
      "learning_rate": 0.0001785076968104716,
      "loss": 1.1579,
      "step": 2307
    },
    {
      "epoch": 0.42891655826054637,
      "grad_norm": 0.6120824813842773,
      "learning_rate": 0.00017848957860597606,
      "loss": 1.0426,
      "step": 2308
    },
    {
      "epoch": 0.4291023973239175,
      "grad_norm": 0.5395716428756714,
      "learning_rate": 0.00017847145368809724,
      "loss": 0.7092,
      "step": 2309
    },
    {
      "epoch": 0.42928823638728864,
      "grad_norm": 0.520346999168396,
      "learning_rate": 0.00017845332205838533,
      "loss": 0.9699,
      "step": 2310
    },
    {
      "epoch": 0.42947407545065974,
      "grad_norm": 0.5615527629852295,
      "learning_rate": 0.00017843518371839126,
      "loss": 1.228,
      "step": 2311
    },
    {
      "epoch": 0.42965991451403085,
      "grad_norm": 0.6460365056991577,
      "learning_rate": 0.00017841703866966642,
      "loss": 1.1718,
      "step": 2312
    },
    {
      "epoch": 0.42984575357740196,
      "grad_norm": 0.5339592099189758,
      "learning_rate": 0.00017839888691376275,
      "loss": 0.8472,
      "step": 2313
    },
    {
      "epoch": 0.43003159264077306,
      "grad_norm": 0.5476655960083008,
      "learning_rate": 0.00017838072845223284,
      "loss": 1.0667,
      "step": 2314
    },
    {
      "epoch": 0.4302174317041442,
      "grad_norm": 0.5243486166000366,
      "learning_rate": 0.00017836256328662985,
      "loss": 1.0041,
      "step": 2315
    },
    {
      "epoch": 0.43040327076751533,
      "grad_norm": 0.7483263611793518,
      "learning_rate": 0.0001783443914185074,
      "loss": 0.9907,
      "step": 2316
    },
    {
      "epoch": 0.43058910983088644,
      "grad_norm": 7.6981706619262695,
      "learning_rate": 0.0001783262128494199,
      "loss": 3.1886,
      "step": 2317
    },
    {
      "epoch": 0.43077494889425755,
      "grad_norm": 0.5558903813362122,
      "learning_rate": 0.0001783080275809221,
      "loss": 1.128,
      "step": 2318
    },
    {
      "epoch": 0.4309607879576287,
      "grad_norm": 0.5539029240608215,
      "learning_rate": 0.00017828983561456941,
      "loss": 1.0576,
      "step": 2319
    },
    {
      "epoch": 0.4311466270209998,
      "grad_norm": 0.6026525497436523,
      "learning_rate": 0.00017827163695191794,
      "loss": 1.1955,
      "step": 2320
    },
    {
      "epoch": 0.4313324660843709,
      "grad_norm": 0.4793394207954407,
      "learning_rate": 0.00017825343159452416,
      "loss": 0.9578,
      "step": 2321
    },
    {
      "epoch": 0.43151830514774203,
      "grad_norm": 0.6857821345329285,
      "learning_rate": 0.00017823521954394526,
      "loss": 1.3858,
      "step": 2322
    },
    {
      "epoch": 0.4317041442111132,
      "grad_norm": 0.6874255537986755,
      "learning_rate": 0.00017821700080173896,
      "loss": 0.9156,
      "step": 2323
    },
    {
      "epoch": 0.4318899832744843,
      "grad_norm": 0.5313177704811096,
      "learning_rate": 0.0001781987753694635,
      "loss": 1.0231,
      "step": 2324
    },
    {
      "epoch": 0.4320758223378554,
      "grad_norm": 0.7569800615310669,
      "learning_rate": 0.0001781805432486778,
      "loss": 1.2913,
      "step": 2325
    },
    {
      "epoch": 0.4322616614012265,
      "grad_norm": 0.5684303641319275,
      "learning_rate": 0.00017816230444094125,
      "loss": 1.2554,
      "step": 2326
    },
    {
      "epoch": 0.4324475004645977,
      "grad_norm": 0.6719896793365479,
      "learning_rate": 0.00017814405894781387,
      "loss": 1.3118,
      "step": 2327
    },
    {
      "epoch": 0.4326333395279688,
      "grad_norm": 0.5125758051872253,
      "learning_rate": 0.0001781258067708562,
      "loss": 1.0935,
      "step": 2328
    },
    {
      "epoch": 0.4328191785913399,
      "grad_norm": 0.6126227974891663,
      "learning_rate": 0.0001781075479116295,
      "loss": 1.2605,
      "step": 2329
    },
    {
      "epoch": 0.433005017654711,
      "grad_norm": 0.44674035906791687,
      "learning_rate": 0.00017808928237169537,
      "loss": 0.7494,
      "step": 2330
    },
    {
      "epoch": 0.43319085671808216,
      "grad_norm": 0.5729489326477051,
      "learning_rate": 0.00017807101015261616,
      "loss": 0.8303,
      "step": 2331
    },
    {
      "epoch": 0.43337669578145327,
      "grad_norm": 0.6711477041244507,
      "learning_rate": 0.0001780527312559547,
      "loss": 1.0835,
      "step": 2332
    },
    {
      "epoch": 0.4335625348448244,
      "grad_norm": 0.5800749659538269,
      "learning_rate": 0.00017803444568327445,
      "loss": 1.2225,
      "step": 2333
    },
    {
      "epoch": 0.4337483739081955,
      "grad_norm": 0.5604079365730286,
      "learning_rate": 0.00017801615343613942,
      "loss": 1.0564,
      "step": 2334
    },
    {
      "epoch": 0.43393421297156665,
      "grad_norm": 0.5705206990242004,
      "learning_rate": 0.00017799785451611414,
      "loss": 1.1846,
      "step": 2335
    },
    {
      "epoch": 0.43412005203493775,
      "grad_norm": 2.446877956390381,
      "learning_rate": 0.00017797954892476384,
      "loss": 3.1014,
      "step": 2336
    },
    {
      "epoch": 0.43430589109830886,
      "grad_norm": 0.5746934413909912,
      "learning_rate": 0.00017796123666365414,
      "loss": 1.061,
      "step": 2337
    },
    {
      "epoch": 0.43449173016167997,
      "grad_norm": 0.5936819314956665,
      "learning_rate": 0.0001779429177343514,
      "loss": 1.1547,
      "step": 2338
    },
    {
      "epoch": 0.43467756922505113,
      "grad_norm": 0.5753687620162964,
      "learning_rate": 0.0001779245921384224,
      "loss": 0.9682,
      "step": 2339
    },
    {
      "epoch": 0.43486340828842224,
      "grad_norm": 0.5654261112213135,
      "learning_rate": 0.00017790625987743468,
      "loss": 1.1469,
      "step": 2340
    },
    {
      "epoch": 0.43504924735179334,
      "grad_norm": 0.5429582595825195,
      "learning_rate": 0.00017788792095295615,
      "loss": 1.069,
      "step": 2341
    },
    {
      "epoch": 0.43523508641516445,
      "grad_norm": 0.5560867786407471,
      "learning_rate": 0.00017786957536655539,
      "loss": 0.9442,
      "step": 2342
    },
    {
      "epoch": 0.4354209254785356,
      "grad_norm": 0.5235719680786133,
      "learning_rate": 0.00017785122311980155,
      "loss": 1.1919,
      "step": 2343
    },
    {
      "epoch": 0.4356067645419067,
      "grad_norm": 0.5630365610122681,
      "learning_rate": 0.00017783286421426436,
      "loss": 1.2617,
      "step": 2344
    },
    {
      "epoch": 0.43579260360527783,
      "grad_norm": 0.5522767901420593,
      "learning_rate": 0.00017781449865151406,
      "loss": 1.0568,
      "step": 2345
    },
    {
      "epoch": 0.43597844266864894,
      "grad_norm": 0.8357774615287781,
      "learning_rate": 0.00017779612643312152,
      "loss": 1.1691,
      "step": 2346
    },
    {
      "epoch": 0.4361642817320201,
      "grad_norm": 0.6235195994377136,
      "learning_rate": 0.00017777774756065813,
      "loss": 0.9429,
      "step": 2347
    },
    {
      "epoch": 0.4363501207953912,
      "grad_norm": 0.5671910047531128,
      "learning_rate": 0.0001777593620356959,
      "loss": 1.0226,
      "step": 2348
    },
    {
      "epoch": 0.4365359598587623,
      "grad_norm": 0.5591475963592529,
      "learning_rate": 0.00017774096985980736,
      "loss": 0.9408,
      "step": 2349
    },
    {
      "epoch": 0.4367217989221334,
      "grad_norm": 0.5957860350608826,
      "learning_rate": 0.00017772257103456567,
      "loss": 1.162,
      "step": 2350
    },
    {
      "epoch": 0.4369076379855045,
      "grad_norm": 0.6221433281898499,
      "learning_rate": 0.00017770416556154453,
      "loss": 1.2008,
      "step": 2351
    },
    {
      "epoch": 0.4370934770488757,
      "grad_norm": 0.5632703304290771,
      "learning_rate": 0.0001776857534423181,
      "loss": 1.2612,
      "step": 2352
    },
    {
      "epoch": 0.4372793161122468,
      "grad_norm": 0.5558828711509705,
      "learning_rate": 0.00017766733467846133,
      "loss": 1.0839,
      "step": 2353
    },
    {
      "epoch": 0.4374651551756179,
      "grad_norm": 0.532751739025116,
      "learning_rate": 0.00017764890927154952,
      "loss": 1.129,
      "step": 2354
    },
    {
      "epoch": 0.437650994238989,
      "grad_norm": 0.5031901597976685,
      "learning_rate": 0.00017763047722315871,
      "loss": 1.1353,
      "step": 2355
    },
    {
      "epoch": 0.4378368333023602,
      "grad_norm": 0.6349141597747803,
      "learning_rate": 0.00017761203853486538,
      "loss": 1.152,
      "step": 2356
    },
    {
      "epoch": 0.4380226723657313,
      "grad_norm": 0.6550739407539368,
      "learning_rate": 0.0001775935932082467,
      "loss": 1.0943,
      "step": 2357
    },
    {
      "epoch": 0.4382085114291024,
      "grad_norm": 0.5785501599311829,
      "learning_rate": 0.00017757514124488023,
      "loss": 1.1742,
      "step": 2358
    },
    {
      "epoch": 0.4383943504924735,
      "grad_norm": 0.5431879758834839,
      "learning_rate": 0.0001775566826463443,
      "loss": 1.0897,
      "step": 2359
    },
    {
      "epoch": 0.43858018955584466,
      "grad_norm": 0.5559978485107422,
      "learning_rate": 0.00017753821741421769,
      "loss": 1.1531,
      "step": 2360
    },
    {
      "epoch": 0.43876602861921576,
      "grad_norm": 0.5944271683692932,
      "learning_rate": 0.00017751974555007974,
      "loss": 1.0854,
      "step": 2361
    },
    {
      "epoch": 0.43895186768258687,
      "grad_norm": 0.545141875743866,
      "learning_rate": 0.0001775012670555104,
      "loss": 0.986,
      "step": 2362
    },
    {
      "epoch": 0.439137706745958,
      "grad_norm": 0.6592658758163452,
      "learning_rate": 0.0001774827819320902,
      "loss": 0.9982,
      "step": 2363
    },
    {
      "epoch": 0.43932354580932914,
      "grad_norm": 0.6145473122596741,
      "learning_rate": 0.00017746429018140024,
      "loss": 1.0719,
      "step": 2364
    },
    {
      "epoch": 0.43950938487270025,
      "grad_norm": 0.5475202798843384,
      "learning_rate": 0.00017744579180502205,
      "loss": 1.1697,
      "step": 2365
    },
    {
      "epoch": 0.43969522393607136,
      "grad_norm": 0.5210398435592651,
      "learning_rate": 0.00017742728680453796,
      "loss": 1.084,
      "step": 2366
    },
    {
      "epoch": 0.43988106299944246,
      "grad_norm": 0.6031140685081482,
      "learning_rate": 0.00017740877518153064,
      "loss": 1.298,
      "step": 2367
    },
    {
      "epoch": 0.4400669020628136,
      "grad_norm": 0.588117241859436,
      "learning_rate": 0.00017739025693758352,
      "loss": 0.9898,
      "step": 2368
    },
    {
      "epoch": 0.44025274112618473,
      "grad_norm": 0.5167676210403442,
      "learning_rate": 0.00017737173207428043,
      "loss": 0.836,
      "step": 2369
    },
    {
      "epoch": 0.44043858018955584,
      "grad_norm": 0.5224547386169434,
      "learning_rate": 0.00017735320059320589,
      "loss": 1.0271,
      "step": 2370
    },
    {
      "epoch": 0.44062441925292695,
      "grad_norm": 0.5996924638748169,
      "learning_rate": 0.00017733466249594492,
      "loss": 1.2124,
      "step": 2371
    },
    {
      "epoch": 0.4408102583162981,
      "grad_norm": 0.6365364789962769,
      "learning_rate": 0.0001773161177840831,
      "loss": 1.3201,
      "step": 2372
    },
    {
      "epoch": 0.4409960973796692,
      "grad_norm": 0.5757020115852356,
      "learning_rate": 0.00017729756645920664,
      "loss": 1.0181,
      "step": 2373
    },
    {
      "epoch": 0.4411819364430403,
      "grad_norm": 0.5585982203483582,
      "learning_rate": 0.00017727900852290227,
      "loss": 1.1259,
      "step": 2374
    },
    {
      "epoch": 0.44136777550641143,
      "grad_norm": 0.7039685845375061,
      "learning_rate": 0.0001772604439767573,
      "loss": 1.4213,
      "step": 2375
    },
    {
      "epoch": 0.4415536145697826,
      "grad_norm": 0.5538169145584106,
      "learning_rate": 0.00017724187282235955,
      "loss": 0.9053,
      "step": 2376
    },
    {
      "epoch": 0.4417394536331537,
      "grad_norm": 0.5275604724884033,
      "learning_rate": 0.00017722329506129745,
      "loss": 0.9567,
      "step": 2377
    },
    {
      "epoch": 0.4419252926965248,
      "grad_norm": 0.5697493553161621,
      "learning_rate": 0.00017720471069516006,
      "loss": 1.0802,
      "step": 2378
    },
    {
      "epoch": 0.4421111317598959,
      "grad_norm": 0.5514183044433594,
      "learning_rate": 0.00017718611972553688,
      "loss": 0.9211,
      "step": 2379
    },
    {
      "epoch": 0.4422969708232671,
      "grad_norm": 0.5721895098686218,
      "learning_rate": 0.00017716752215401807,
      "loss": 1.1159,
      "step": 2380
    },
    {
      "epoch": 0.4424828098866382,
      "grad_norm": 0.5930714011192322,
      "learning_rate": 0.0001771489179821943,
      "loss": 1.1339,
      "step": 2381
    },
    {
      "epoch": 0.4426686489500093,
      "grad_norm": 0.5223318934440613,
      "learning_rate": 0.00017713030721165684,
      "loss": 0.889,
      "step": 2382
    },
    {
      "epoch": 0.4428544880133804,
      "grad_norm": 0.5683040022850037,
      "learning_rate": 0.0001771116898439975,
      "loss": 1.2511,
      "step": 2383
    },
    {
      "epoch": 0.4430403270767515,
      "grad_norm": 0.5765540599822998,
      "learning_rate": 0.00017709306588080867,
      "loss": 0.9842,
      "step": 2384
    },
    {
      "epoch": 0.44322616614012267,
      "grad_norm": 0.568500280380249,
      "learning_rate": 0.00017707443532368332,
      "loss": 1.1352,
      "step": 2385
    },
    {
      "epoch": 0.4434120052034938,
      "grad_norm": 0.5285328030586243,
      "learning_rate": 0.00017705579817421492,
      "loss": 1.0351,
      "step": 2386
    },
    {
      "epoch": 0.4435978442668649,
      "grad_norm": 0.6016953587532043,
      "learning_rate": 0.00017703715443399752,
      "loss": 1.2857,
      "step": 2387
    },
    {
      "epoch": 0.443783683330236,
      "grad_norm": 0.6040376424789429,
      "learning_rate": 0.00017701850410462586,
      "loss": 1.1826,
      "step": 2388
    },
    {
      "epoch": 0.44396952239360715,
      "grad_norm": 0.600718080997467,
      "learning_rate": 0.00017699984718769505,
      "loss": 1.1051,
      "step": 2389
    },
    {
      "epoch": 0.44415536145697826,
      "grad_norm": 0.6064468026161194,
      "learning_rate": 0.0001769811836848009,
      "loss": 1.1616,
      "step": 2390
    },
    {
      "epoch": 0.44434120052034937,
      "grad_norm": 0.6028340458869934,
      "learning_rate": 0.00017696251359753976,
      "loss": 1.1355,
      "step": 2391
    },
    {
      "epoch": 0.4445270395837205,
      "grad_norm": 0.61343914270401,
      "learning_rate": 0.00017694383692750843,
      "loss": 1.2456,
      "step": 2392
    },
    {
      "epoch": 0.44471287864709164,
      "grad_norm": 0.6040683388710022,
      "learning_rate": 0.0001769251536763045,
      "loss": 1.1576,
      "step": 2393
    },
    {
      "epoch": 0.44489871771046274,
      "grad_norm": 0.5966310501098633,
      "learning_rate": 0.00017690646384552588,
      "loss": 1.4009,
      "step": 2394
    },
    {
      "epoch": 0.44508455677383385,
      "grad_norm": 0.5714512467384338,
      "learning_rate": 0.0001768877674367712,
      "loss": 1.0652,
      "step": 2395
    },
    {
      "epoch": 0.44527039583720496,
      "grad_norm": 0.6570279002189636,
      "learning_rate": 0.0001768690644516396,
      "loss": 1.2492,
      "step": 2396
    },
    {
      "epoch": 0.4454562349005761,
      "grad_norm": 0.5377715229988098,
      "learning_rate": 0.00017685035489173076,
      "loss": 1.0869,
      "step": 2397
    },
    {
      "epoch": 0.4456420739639472,
      "grad_norm": 0.6347850561141968,
      "learning_rate": 0.000176831638758645,
      "loss": 1.1475,
      "step": 2398
    },
    {
      "epoch": 0.44582791302731833,
      "grad_norm": 0.5425981283187866,
      "learning_rate": 0.00017681291605398312,
      "loss": 1.0795,
      "step": 2399
    },
    {
      "epoch": 0.44601375209068944,
      "grad_norm": 0.6422131061553955,
      "learning_rate": 0.0001767941867793465,
      "loss": 1.2814,
      "step": 2400
    },
    {
      "epoch": 0.4461995911540606,
      "grad_norm": 0.6414985060691833,
      "learning_rate": 0.00017677545093633713,
      "loss": 1.0609,
      "step": 2401
    },
    {
      "epoch": 0.4463854302174317,
      "grad_norm": 0.5891706943511963,
      "learning_rate": 0.0001767567085265575,
      "loss": 1.3514,
      "step": 2402
    },
    {
      "epoch": 0.4465712692808028,
      "grad_norm": 0.5836945176124573,
      "learning_rate": 0.00017673795955161067,
      "loss": 1.0508,
      "step": 2403
    },
    {
      "epoch": 0.4467571083441739,
      "grad_norm": 0.5158883333206177,
      "learning_rate": 0.00017671920401310032,
      "loss": 1.0526,
      "step": 2404
    },
    {
      "epoch": 0.4469429474075451,
      "grad_norm": 0.603083610534668,
      "learning_rate": 0.00017670044191263063,
      "loss": 1.0876,
      "step": 2405
    },
    {
      "epoch": 0.4471287864709162,
      "grad_norm": 0.5200565457344055,
      "learning_rate": 0.0001766816732518064,
      "loss": 0.8484,
      "step": 2406
    },
    {
      "epoch": 0.4473146255342873,
      "grad_norm": 0.6235826015472412,
      "learning_rate": 0.0001766628980322329,
      "loss": 1.17,
      "step": 2407
    },
    {
      "epoch": 0.4475004645976584,
      "grad_norm": 0.6319602727890015,
      "learning_rate": 0.00017664411625551606,
      "loss": 0.9634,
      "step": 2408
    },
    {
      "epoch": 0.44768630366102957,
      "grad_norm": 0.4631809592247009,
      "learning_rate": 0.0001766253279232623,
      "loss": 0.8415,
      "step": 2409
    },
    {
      "epoch": 0.4478721427244007,
      "grad_norm": 0.5904662013053894,
      "learning_rate": 0.00017660653303707864,
      "loss": 0.7638,
      "step": 2410
    },
    {
      "epoch": 0.4480579817877718,
      "grad_norm": 0.5749669075012207,
      "learning_rate": 0.00017658773159857263,
      "loss": 0.9454,
      "step": 2411
    },
    {
      "epoch": 0.4482438208511429,
      "grad_norm": 0.6295116543769836,
      "learning_rate": 0.00017656892360935243,
      "loss": 1.256,
      "step": 2412
    },
    {
      "epoch": 0.44842965991451406,
      "grad_norm": 0.5321747660636902,
      "learning_rate": 0.00017655010907102666,
      "loss": 1.1843,
      "step": 2413
    },
    {
      "epoch": 0.44861549897788516,
      "grad_norm": 0.5966020822525024,
      "learning_rate": 0.00017653128798520467,
      "loss": 1.1913,
      "step": 2414
    },
    {
      "epoch": 0.44880133804125627,
      "grad_norm": 0.615644633769989,
      "learning_rate": 0.0001765124603534962,
      "loss": 1.0032,
      "step": 2415
    },
    {
      "epoch": 0.4489871771046274,
      "grad_norm": 0.5704545378684998,
      "learning_rate": 0.0001764936261775116,
      "loss": 0.9007,
      "step": 2416
    },
    {
      "epoch": 0.44917301616799854,
      "grad_norm": 0.5010755658149719,
      "learning_rate": 0.00017647478545886186,
      "loss": 1.116,
      "step": 2417
    },
    {
      "epoch": 0.44935885523136965,
      "grad_norm": 0.6472548246383667,
      "learning_rate": 0.0001764559381991584,
      "loss": 1.0135,
      "step": 2418
    },
    {
      "epoch": 0.44954469429474075,
      "grad_norm": 0.49504557251930237,
      "learning_rate": 0.00017643708440001334,
      "loss": 1.0561,
      "step": 2419
    },
    {
      "epoch": 0.44973053335811186,
      "grad_norm": 0.724803626537323,
      "learning_rate": 0.00017641822406303922,
      "loss": 0.989,
      "step": 2420
    },
    {
      "epoch": 0.44991637242148297,
      "grad_norm": 0.5256615877151489,
      "learning_rate": 0.00017639935718984925,
      "loss": 1.2047,
      "step": 2421
    },
    {
      "epoch": 0.45010221148485413,
      "grad_norm": 0.6002311706542969,
      "learning_rate": 0.00017638048378205713,
      "loss": 1.1932,
      "step": 2422
    },
    {
      "epoch": 0.45028805054822524,
      "grad_norm": 0.6114380955696106,
      "learning_rate": 0.00017636160384127715,
      "loss": 1.1895,
      "step": 2423
    },
    {
      "epoch": 0.45047388961159635,
      "grad_norm": 0.6021878123283386,
      "learning_rate": 0.0001763427173691242,
      "loss": 1.2897,
      "step": 2424
    },
    {
      "epoch": 0.45065972867496745,
      "grad_norm": 0.6550626754760742,
      "learning_rate": 0.00017632382436721358,
      "loss": 0.9298,
      "step": 2425
    },
    {
      "epoch": 0.4508455677383386,
      "grad_norm": 0.6041695475578308,
      "learning_rate": 0.00017630492483716132,
      "loss": 1.1527,
      "step": 2426
    },
    {
      "epoch": 0.4510314068017097,
      "grad_norm": 0.6602818369865417,
      "learning_rate": 0.00017628601878058394,
      "loss": 1.1716,
      "step": 2427
    },
    {
      "epoch": 0.45121724586508083,
      "grad_norm": 0.5621387958526611,
      "learning_rate": 0.00017626710619909848,
      "loss": 1.1978,
      "step": 2428
    },
    {
      "epoch": 0.45140308492845194,
      "grad_norm": 0.5593321919441223,
      "learning_rate": 0.0001762481870943226,
      "loss": 1.0507,
      "step": 2429
    },
    {
      "epoch": 0.4515889239918231,
      "grad_norm": 0.5860761404037476,
      "learning_rate": 0.00017622926146787447,
      "loss": 1.3819,
      "step": 2430
    },
    {
      "epoch": 0.4517747630551942,
      "grad_norm": 0.5836736559867859,
      "learning_rate": 0.00017621032932137287,
      "loss": 0.9615,
      "step": 2431
    },
    {
      "epoch": 0.4519606021185653,
      "grad_norm": 0.5087476372718811,
      "learning_rate": 0.00017619139065643707,
      "loss": 1.0476,
      "step": 2432
    },
    {
      "epoch": 0.4521464411819364,
      "grad_norm": 0.5831040740013123,
      "learning_rate": 0.00017617244547468697,
      "loss": 0.8807,
      "step": 2433
    },
    {
      "epoch": 0.4523322802453076,
      "grad_norm": 0.5577731132507324,
      "learning_rate": 0.00017615349377774297,
      "loss": 1.1035,
      "step": 2434
    },
    {
      "epoch": 0.4525181193086787,
      "grad_norm": 0.6123347878456116,
      "learning_rate": 0.00017613453556722608,
      "loss": 1.0745,
      "step": 2435
    },
    {
      "epoch": 0.4527039583720498,
      "grad_norm": 0.5789678692817688,
      "learning_rate": 0.0001761155708447578,
      "loss": 1.1167,
      "step": 2436
    },
    {
      "epoch": 0.4528897974354209,
      "grad_norm": 0.59694504737854,
      "learning_rate": 0.00017609659961196022,
      "loss": 1.0494,
      "step": 2437
    },
    {
      "epoch": 0.45307563649879207,
      "grad_norm": 0.709152340888977,
      "learning_rate": 0.00017607762187045605,
      "loss": 0.9394,
      "step": 2438
    },
    {
      "epoch": 0.4532614755621632,
      "grad_norm": 0.4953658878803253,
      "learning_rate": 0.00017605863762186843,
      "loss": 0.9409,
      "step": 2439
    },
    {
      "epoch": 0.4534473146255343,
      "grad_norm": 0.5981800556182861,
      "learning_rate": 0.00017603964686782115,
      "loss": 1.3633,
      "step": 2440
    },
    {
      "epoch": 0.4536331536889054,
      "grad_norm": 0.604300320148468,
      "learning_rate": 0.00017602064960993853,
      "loss": 0.8273,
      "step": 2441
    },
    {
      "epoch": 0.45381899275227655,
      "grad_norm": 0.5925197601318359,
      "learning_rate": 0.00017600164584984546,
      "loss": 1.0395,
      "step": 2442
    },
    {
      "epoch": 0.45400483181564766,
      "grad_norm": 0.5409677028656006,
      "learning_rate": 0.00017598263558916735,
      "loss": 0.9716,
      "step": 2443
    },
    {
      "epoch": 0.45419067087901877,
      "grad_norm": 0.6586957573890686,
      "learning_rate": 0.0001759636188295302,
      "loss": 0.854,
      "step": 2444
    },
    {
      "epoch": 0.4543765099423899,
      "grad_norm": 0.6706559062004089,
      "learning_rate": 0.00017594459557256054,
      "loss": 1.3125,
      "step": 2445
    },
    {
      "epoch": 0.45456234900576103,
      "grad_norm": 0.6373844742774963,
      "learning_rate": 0.00017592556581988547,
      "loss": 1.1909,
      "step": 2446
    },
    {
      "epoch": 0.45474818806913214,
      "grad_norm": 0.5949018597602844,
      "learning_rate": 0.0001759065295731327,
      "loss": 1.0114,
      "step": 2447
    },
    {
      "epoch": 0.45493402713250325,
      "grad_norm": 0.6161789298057556,
      "learning_rate": 0.00017588748683393038,
      "loss": 1.2209,
      "step": 2448
    },
    {
      "epoch": 0.45511986619587436,
      "grad_norm": 0.6025707721710205,
      "learning_rate": 0.0001758684376039073,
      "loss": 1.1802,
      "step": 2449
    },
    {
      "epoch": 0.4553057052592455,
      "grad_norm": 2.4854700565338135,
      "learning_rate": 0.0001758493818846928,
      "loss": 2.2678,
      "step": 2450
    },
    {
      "epoch": 0.4554915443226166,
      "grad_norm": 0.6085222363471985,
      "learning_rate": 0.0001758303196779167,
      "loss": 1.1835,
      "step": 2451
    },
    {
      "epoch": 0.45567738338598773,
      "grad_norm": 0.6468694806098938,
      "learning_rate": 0.00017581125098520945,
      "loss": 1.1857,
      "step": 2452
    },
    {
      "epoch": 0.45586322244935884,
      "grad_norm": 0.6589061617851257,
      "learning_rate": 0.0001757921758082021,
      "loss": 0.9729,
      "step": 2453
    },
    {
      "epoch": 0.45604906151273,
      "grad_norm": 0.7410516142845154,
      "learning_rate": 0.0001757730941485261,
      "loss": 0.9728,
      "step": 2454
    },
    {
      "epoch": 0.4562349005761011,
      "grad_norm": 0.5850430130958557,
      "learning_rate": 0.00017575400600781363,
      "loss": 0.9746,
      "step": 2455
    },
    {
      "epoch": 0.4564207396394722,
      "grad_norm": 0.4909527599811554,
      "learning_rate": 0.00017573491138769728,
      "loss": 0.9898,
      "step": 2456
    },
    {
      "epoch": 0.4566065787028433,
      "grad_norm": 2.1568737030029297,
      "learning_rate": 0.0001757158102898103,
      "loss": 1.937,
      "step": 2457
    },
    {
      "epoch": 0.45679241776621443,
      "grad_norm": 0.5108540654182434,
      "learning_rate": 0.0001756967027157864,
      "loss": 0.7381,
      "step": 2458
    },
    {
      "epoch": 0.4569782568295856,
      "grad_norm": 0.5407970547676086,
      "learning_rate": 0.0001756775886672599,
      "loss": 0.8545,
      "step": 2459
    },
    {
      "epoch": 0.4571640958929567,
      "grad_norm": 0.6552363634109497,
      "learning_rate": 0.00017565846814586572,
      "loss": 1.2524,
      "step": 2460
    },
    {
      "epoch": 0.4573499349563278,
      "grad_norm": 0.5322883129119873,
      "learning_rate": 0.0001756393411532392,
      "loss": 1.1172,
      "step": 2461
    },
    {
      "epoch": 0.4575357740196989,
      "grad_norm": 0.7922711968421936,
      "learning_rate": 0.00017562020769101638,
      "loss": 1.1572,
      "step": 2462
    },
    {
      "epoch": 0.4577216130830701,
      "grad_norm": 0.5557255744934082,
      "learning_rate": 0.00017560106776083376,
      "loss": 1.0438,
      "step": 2463
    },
    {
      "epoch": 0.4579074521464412,
      "grad_norm": 0.5572389960289001,
      "learning_rate": 0.00017558192136432843,
      "loss": 1.0503,
      "step": 2464
    },
    {
      "epoch": 0.4580932912098123,
      "grad_norm": 0.5063178539276123,
      "learning_rate": 0.00017556276850313798,
      "loss": 1.2155,
      "step": 2465
    },
    {
      "epoch": 0.4582791302731834,
      "grad_norm": 0.6309200525283813,
      "learning_rate": 0.00017554360917890064,
      "loss": 1.0558,
      "step": 2466
    },
    {
      "epoch": 0.45846496933655456,
      "grad_norm": 0.6432199478149414,
      "learning_rate": 0.00017552444339325514,
      "loss": 1.1363,
      "step": 2467
    },
    {
      "epoch": 0.45865080839992567,
      "grad_norm": 0.583766758441925,
      "learning_rate": 0.00017550527114784073,
      "loss": 1.1827,
      "step": 2468
    },
    {
      "epoch": 0.4588366474632968,
      "grad_norm": 0.5922523736953735,
      "learning_rate": 0.00017548609244429734,
      "loss": 1.1427,
      "step": 2469
    },
    {
      "epoch": 0.4590224865266679,
      "grad_norm": 0.6233006715774536,
      "learning_rate": 0.0001754669072842653,
      "loss": 1.1008,
      "step": 2470
    },
    {
      "epoch": 0.45920832559003905,
      "grad_norm": 0.5502414703369141,
      "learning_rate": 0.00017544771566938559,
      "loss": 1.199,
      "step": 2471
    },
    {
      "epoch": 0.45939416465341015,
      "grad_norm": 0.7010074257850647,
      "learning_rate": 0.00017542851760129965,
      "loss": 1.1342,
      "step": 2472
    },
    {
      "epoch": 0.45958000371678126,
      "grad_norm": 0.5409691333770752,
      "learning_rate": 0.0001754093130816496,
      "loss": 0.8252,
      "step": 2473
    },
    {
      "epoch": 0.45976584278015237,
      "grad_norm": 0.6597109436988831,
      "learning_rate": 0.00017539010211207806,
      "loss": 1.0491,
      "step": 2474
    },
    {
      "epoch": 0.45995168184352353,
      "grad_norm": 0.5203375816345215,
      "learning_rate": 0.00017537088469422808,
      "loss": 1.0829,
      "step": 2475
    },
    {
      "epoch": 0.46013752090689464,
      "grad_norm": 0.6410231590270996,
      "learning_rate": 0.00017535166082974351,
      "loss": 1.0531,
      "step": 2476
    },
    {
      "epoch": 0.46032335997026574,
      "grad_norm": 0.6432588696479797,
      "learning_rate": 0.00017533243052026846,
      "loss": 1.0289,
      "step": 2477
    },
    {
      "epoch": 0.46050919903363685,
      "grad_norm": 0.6830189228057861,
      "learning_rate": 0.00017531319376744783,
      "loss": 0.9012,
      "step": 2478
    },
    {
      "epoch": 0.460695038097008,
      "grad_norm": 0.6746105551719666,
      "learning_rate": 0.000175293950572927,
      "loss": 1.2264,
      "step": 2479
    },
    {
      "epoch": 0.4608808771603791,
      "grad_norm": 0.6015942692756653,
      "learning_rate": 0.00017527470093835182,
      "loss": 1.0476,
      "step": 2480
    },
    {
      "epoch": 0.46106671622375023,
      "grad_norm": 0.6453880667686462,
      "learning_rate": 0.00017525544486536877,
      "loss": 1.0024,
      "step": 2481
    },
    {
      "epoch": 0.46125255528712134,
      "grad_norm": 0.5611870884895325,
      "learning_rate": 0.0001752361823556249,
      "loss": 1.1538,
      "step": 2482
    },
    {
      "epoch": 0.4614383943504925,
      "grad_norm": 0.5102313756942749,
      "learning_rate": 0.00017521691341076774,
      "loss": 0.8601,
      "step": 2483
    },
    {
      "epoch": 0.4616242334138636,
      "grad_norm": 0.5948174595832825,
      "learning_rate": 0.00017519763803244538,
      "loss": 1.2159,
      "step": 2484
    },
    {
      "epoch": 0.4618100724772347,
      "grad_norm": 0.523780345916748,
      "learning_rate": 0.00017517835622230654,
      "loss": 0.8715,
      "step": 2485
    },
    {
      "epoch": 0.4619959115406058,
      "grad_norm": 0.5988032221794128,
      "learning_rate": 0.0001751590679820004,
      "loss": 1.208,
      "step": 2486
    },
    {
      "epoch": 0.462181750603977,
      "grad_norm": 0.54287189245224,
      "learning_rate": 0.00017513977331317678,
      "loss": 1.1722,
      "step": 2487
    },
    {
      "epoch": 0.4623675896673481,
      "grad_norm": 0.5695911645889282,
      "learning_rate": 0.0001751204722174859,
      "loss": 1.0592,
      "step": 2488
    },
    {
      "epoch": 0.4625534287307192,
      "grad_norm": 0.606904149055481,
      "learning_rate": 0.00017510116469657868,
      "loss": 1.1955,
      "step": 2489
    },
    {
      "epoch": 0.4627392677940903,
      "grad_norm": 0.5442674160003662,
      "learning_rate": 0.0001750818507521065,
      "loss": 0.9511,
      "step": 2490
    },
    {
      "epoch": 0.46292510685746147,
      "grad_norm": 0.6489145755767822,
      "learning_rate": 0.0001750625303857214,
      "loss": 1.2446,
      "step": 2491
    },
    {
      "epoch": 0.4631109459208326,
      "grad_norm": 0.5138381123542786,
      "learning_rate": 0.00017504320359907586,
      "loss": 0.8772,
      "step": 2492
    },
    {
      "epoch": 0.4632967849842037,
      "grad_norm": 0.5864812135696411,
      "learning_rate": 0.0001750238703938229,
      "loss": 1.0736,
      "step": 2493
    },
    {
      "epoch": 0.4634826240475748,
      "grad_norm": 0.5524820685386658,
      "learning_rate": 0.00017500453077161617,
      "loss": 1.1108,
      "step": 2494
    },
    {
      "epoch": 0.4636684631109459,
      "grad_norm": 0.6688693761825562,
      "learning_rate": 0.0001749851847341098,
      "loss": 0.9646,
      "step": 2495
    },
    {
      "epoch": 0.46385430217431706,
      "grad_norm": 0.5650538206100464,
      "learning_rate": 0.0001749658322829585,
      "loss": 1.0911,
      "step": 2496
    },
    {
      "epoch": 0.46404014123768816,
      "grad_norm": 0.6584517955780029,
      "learning_rate": 0.00017494647341981755,
      "loss": 1.1071,
      "step": 2497
    },
    {
      "epoch": 0.46422598030105927,
      "grad_norm": 0.62405925989151,
      "learning_rate": 0.00017492710814634275,
      "loss": 1.0248,
      "step": 2498
    },
    {
      "epoch": 0.4644118193644304,
      "grad_norm": 0.5927742719650269,
      "learning_rate": 0.00017490773646419046,
      "loss": 0.956,
      "step": 2499
    },
    {
      "epoch": 0.46459765842780154,
      "grad_norm": 0.5736260414123535,
      "learning_rate": 0.0001748883583750176,
      "loss": 1.1814,
      "step": 2500
    },
    {
      "epoch": 0.46478349749117265,
      "grad_norm": 0.6146401166915894,
      "learning_rate": 0.00017486897388048156,
      "loss": 0.9654,
      "step": 2501
    },
    {
      "epoch": 0.46496933655454376,
      "grad_norm": 0.5440464019775391,
      "learning_rate": 0.00017484958298224037,
      "loss": 0.9954,
      "step": 2502
    },
    {
      "epoch": 0.46515517561791486,
      "grad_norm": 0.6333348751068115,
      "learning_rate": 0.00017483018568195257,
      "loss": 1.3325,
      "step": 2503
    },
    {
      "epoch": 0.465341014681286,
      "grad_norm": 0.669081449508667,
      "learning_rate": 0.0001748107819812773,
      "loss": 1.1671,
      "step": 2504
    },
    {
      "epoch": 0.46552685374465713,
      "grad_norm": 0.5631642937660217,
      "learning_rate": 0.0001747913718818741,
      "loss": 1.1348,
      "step": 2505
    },
    {
      "epoch": 0.46571269280802824,
      "grad_norm": 0.565836489200592,
      "learning_rate": 0.00017477195538540325,
      "loss": 1.156,
      "step": 2506
    },
    {
      "epoch": 0.46589853187139935,
      "grad_norm": 0.4931422472000122,
      "learning_rate": 0.00017475253249352548,
      "loss": 0.9889,
      "step": 2507
    },
    {
      "epoch": 0.4660843709347705,
      "grad_norm": 0.6236067414283752,
      "learning_rate": 0.000174733103207902,
      "loss": 1.1636,
      "step": 2508
    },
    {
      "epoch": 0.4662702099981416,
      "grad_norm": 0.5879521369934082,
      "learning_rate": 0.0001747136675301947,
      "loss": 1.419,
      "step": 2509
    },
    {
      "epoch": 0.4664560490615127,
      "grad_norm": 0.5449656844139099,
      "learning_rate": 0.00017469422546206596,
      "loss": 1.059,
      "step": 2510
    },
    {
      "epoch": 0.46664188812488383,
      "grad_norm": 0.5169174671173096,
      "learning_rate": 0.00017467477700517864,
      "loss": 1.0173,
      "step": 2511
    },
    {
      "epoch": 0.466827727188255,
      "grad_norm": 0.5268780589103699,
      "learning_rate": 0.00017465532216119625,
      "loss": 1.1683,
      "step": 2512
    },
    {
      "epoch": 0.4670135662516261,
      "grad_norm": 0.5800816416740417,
      "learning_rate": 0.00017463586093178287,
      "loss": 0.8585,
      "step": 2513
    },
    {
      "epoch": 0.4671994053149972,
      "grad_norm": 0.5580684542655945,
      "learning_rate": 0.00017461639331860294,
      "loss": 1.0921,
      "step": 2514
    },
    {
      "epoch": 0.4673852443783683,
      "grad_norm": 0.6631280183792114,
      "learning_rate": 0.00017459691932332163,
      "loss": 0.923,
      "step": 2515
    },
    {
      "epoch": 0.4675710834417395,
      "grad_norm": 0.5217365026473999,
      "learning_rate": 0.00017457743894760456,
      "loss": 1.1524,
      "step": 2516
    },
    {
      "epoch": 0.4677569225051106,
      "grad_norm": 0.6715142726898193,
      "learning_rate": 0.000174557952193118,
      "loss": 1.0847,
      "step": 2517
    },
    {
      "epoch": 0.4679427615684817,
      "grad_norm": 0.6288979053497314,
      "learning_rate": 0.0001745384590615286,
      "loss": 1.2511,
      "step": 2518
    },
    {
      "epoch": 0.4681286006318528,
      "grad_norm": 0.6113332509994507,
      "learning_rate": 0.00017451895955450375,
      "loss": 1.1654,
      "step": 2519
    },
    {
      "epoch": 0.46831443969522396,
      "grad_norm": 0.5559727549552917,
      "learning_rate": 0.0001744994536737112,
      "loss": 1.2103,
      "step": 2520
    },
    {
      "epoch": 0.46850027875859507,
      "grad_norm": 0.6704648733139038,
      "learning_rate": 0.00017447994142081935,
      "loss": 1.2595,
      "step": 2521
    },
    {
      "epoch": 0.4686861178219662,
      "grad_norm": 0.9100911021232605,
      "learning_rate": 0.00017446042279749716,
      "loss": 1.1599,
      "step": 2522
    },
    {
      "epoch": 0.4688719568853373,
      "grad_norm": 0.5992748141288757,
      "learning_rate": 0.0001744408978054141,
      "loss": 0.8977,
      "step": 2523
    },
    {
      "epoch": 0.46905779594870844,
      "grad_norm": 0.576352596282959,
      "learning_rate": 0.00017442136644624015,
      "loss": 1.0218,
      "step": 2524
    },
    {
      "epoch": 0.46924363501207955,
      "grad_norm": 0.73007732629776,
      "learning_rate": 0.00017440182872164586,
      "loss": 1.2671,
      "step": 2525
    },
    {
      "epoch": 0.46942947407545066,
      "grad_norm": 0.5133440494537354,
      "learning_rate": 0.00017438228463330236,
      "loss": 0.8163,
      "step": 2526
    },
    {
      "epoch": 0.46961531313882177,
      "grad_norm": 0.636014461517334,
      "learning_rate": 0.0001743627341828813,
      "loss": 1.1354,
      "step": 2527
    },
    {
      "epoch": 0.4698011522021929,
      "grad_norm": 0.5063263177871704,
      "learning_rate": 0.00017434317737205488,
      "loss": 0.7853,
      "step": 2528
    },
    {
      "epoch": 0.46998699126556404,
      "grad_norm": 0.5486884713172913,
      "learning_rate": 0.00017432361420249582,
      "loss": 0.9424,
      "step": 2529
    },
    {
      "epoch": 0.47017283032893514,
      "grad_norm": 0.5266200304031372,
      "learning_rate": 0.00017430404467587738,
      "loss": 1.146,
      "step": 2530
    },
    {
      "epoch": 0.47035866939230625,
      "grad_norm": 0.626629114151001,
      "learning_rate": 0.0001742844687938734,
      "loss": 1.0168,
      "step": 2531
    },
    {
      "epoch": 0.47054450845567736,
      "grad_norm": 0.6614012122154236,
      "learning_rate": 0.0001742648865581583,
      "loss": 1.0367,
      "step": 2532
    },
    {
      "epoch": 0.4707303475190485,
      "grad_norm": 0.685212254524231,
      "learning_rate": 0.00017424529797040693,
      "loss": 1.3456,
      "step": 2533
    },
    {
      "epoch": 0.4709161865824196,
      "grad_norm": 0.636879563331604,
      "learning_rate": 0.00017422570303229473,
      "loss": 1.2354,
      "step": 2534
    },
    {
      "epoch": 0.47110202564579073,
      "grad_norm": 0.4891170263290405,
      "learning_rate": 0.00017420610174549773,
      "loss": 1.1914,
      "step": 2535
    },
    {
      "epoch": 0.47128786470916184,
      "grad_norm": 0.5971581935882568,
      "learning_rate": 0.0001741864941116925,
      "loss": 1.1496,
      "step": 2536
    },
    {
      "epoch": 0.471473703772533,
      "grad_norm": 0.6186808347702026,
      "learning_rate": 0.00017416688013255607,
      "loss": 1.0804,
      "step": 2537
    },
    {
      "epoch": 0.4716595428359041,
      "grad_norm": 0.5329042673110962,
      "learning_rate": 0.00017414725980976608,
      "loss": 1.1438,
      "step": 2538
    },
    {
      "epoch": 0.4718453818992752,
      "grad_norm": 0.590497612953186,
      "learning_rate": 0.00017412763314500072,
      "loss": 1.1196,
      "step": 2539
    },
    {
      "epoch": 0.4720312209626463,
      "grad_norm": 0.6951502561569214,
      "learning_rate": 0.00017410800013993867,
      "loss": 1.061,
      "step": 2540
    },
    {
      "epoch": 0.4722170600260175,
      "grad_norm": 0.5251403450965881,
      "learning_rate": 0.00017408836079625922,
      "loss": 0.7278,
      "step": 2541
    },
    {
      "epoch": 0.4724028990893886,
      "grad_norm": 0.7289116978645325,
      "learning_rate": 0.00017406871511564215,
      "loss": 1.0795,
      "step": 2542
    },
    {
      "epoch": 0.4725887381527597,
      "grad_norm": 0.5959666967391968,
      "learning_rate": 0.00017404906309976777,
      "loss": 1.1282,
      "step": 2543
    },
    {
      "epoch": 0.4727745772161308,
      "grad_norm": 0.6236711144447327,
      "learning_rate": 0.000174029404750317,
      "loss": 1.2404,
      "step": 2544
    },
    {
      "epoch": 0.47296041627950197,
      "grad_norm": 0.5691270232200623,
      "learning_rate": 0.00017400974006897123,
      "loss": 1.2028,
      "step": 2545
    },
    {
      "epoch": 0.4731462553428731,
      "grad_norm": 0.5911795496940613,
      "learning_rate": 0.0001739900690574124,
      "loss": 0.9487,
      "step": 2546
    },
    {
      "epoch": 0.4733320944062442,
      "grad_norm": 0.6401910185813904,
      "learning_rate": 0.00017397039171732312,
      "loss": 1.2378,
      "step": 2547
    },
    {
      "epoch": 0.4735179334696153,
      "grad_norm": 0.4885672330856323,
      "learning_rate": 0.00017395070805038633,
      "loss": 1.0692,
      "step": 2548
    },
    {
      "epoch": 0.47370377253298646,
      "grad_norm": 0.6205880641937256,
      "learning_rate": 0.00017393101805828563,
      "loss": 0.9946,
      "step": 2549
    },
    {
      "epoch": 0.47388961159635756,
      "grad_norm": 0.5263411402702332,
      "learning_rate": 0.00017391132174270521,
      "loss": 1.0809,
      "step": 2550
    },
    {
      "epoch": 0.47407545065972867,
      "grad_norm": 0.7040189504623413,
      "learning_rate": 0.00017389161910532968,
      "loss": 1.2357,
      "step": 2551
    },
    {
      "epoch": 0.4742612897230998,
      "grad_norm": 0.561153769493103,
      "learning_rate": 0.00017387191014784427,
      "loss": 0.8271,
      "step": 2552
    },
    {
      "epoch": 0.47444712878647094,
      "grad_norm": 0.611548900604248,
      "learning_rate": 0.00017385219487193473,
      "loss": 1.1188,
      "step": 2553
    },
    {
      "epoch": 0.47463296784984205,
      "grad_norm": 0.6051244735717773,
      "learning_rate": 0.00017383247327928735,
      "loss": 0.9883,
      "step": 2554
    },
    {
      "epoch": 0.47481880691321315,
      "grad_norm": 0.6681859493255615,
      "learning_rate": 0.00017381274537158896,
      "loss": 1.1764,
      "step": 2555
    },
    {
      "epoch": 0.47500464597658426,
      "grad_norm": 0.5694065093994141,
      "learning_rate": 0.00017379301115052692,
      "loss": 1.1938,
      "step": 2556
    },
    {
      "epoch": 0.4751904850399554,
      "grad_norm": 0.5876331329345703,
      "learning_rate": 0.00017377327061778914,
      "loss": 1.0628,
      "step": 2557
    },
    {
      "epoch": 0.47537632410332653,
      "grad_norm": 0.6016728281974792,
      "learning_rate": 0.00017375352377506408,
      "loss": 1.1583,
      "step": 2558
    },
    {
      "epoch": 0.47556216316669764,
      "grad_norm": 0.4448058307170868,
      "learning_rate": 0.00017373377062404076,
      "loss": 0.8662,
      "step": 2559
    },
    {
      "epoch": 0.47574800223006874,
      "grad_norm": 0.6803490519523621,
      "learning_rate": 0.00017371401116640867,
      "loss": 1.0937,
      "step": 2560
    },
    {
      "epoch": 0.4759338412934399,
      "grad_norm": 0.6000282168388367,
      "learning_rate": 0.00017369424540385783,
      "loss": 1.1161,
      "step": 2561
    },
    {
      "epoch": 0.476119680356811,
      "grad_norm": 0.5419662594795227,
      "learning_rate": 0.00017367447333807898,
      "loss": 1.0714,
      "step": 2562
    },
    {
      "epoch": 0.4763055194201821,
      "grad_norm": 0.543039083480835,
      "learning_rate": 0.00017365469497076318,
      "loss": 0.918,
      "step": 2563
    },
    {
      "epoch": 0.47649135848355323,
      "grad_norm": 0.46435242891311646,
      "learning_rate": 0.0001736349103036021,
      "loss": 0.9592,
      "step": 2564
    },
    {
      "epoch": 0.47667719754692434,
      "grad_norm": 0.49158206582069397,
      "learning_rate": 0.00017361511933828801,
      "loss": 1.0331,
      "step": 2565
    },
    {
      "epoch": 0.4768630366102955,
      "grad_norm": 0.5796807408332825,
      "learning_rate": 0.00017359532207651365,
      "loss": 1.1738,
      "step": 2566
    },
    {
      "epoch": 0.4770488756736666,
      "grad_norm": 0.576512336730957,
      "learning_rate": 0.00017357551851997232,
      "loss": 1.0411,
      "step": 2567
    },
    {
      "epoch": 0.4772347147370377,
      "grad_norm": 0.6543241739273071,
      "learning_rate": 0.0001735557086703579,
      "loss": 1.1288,
      "step": 2568
    },
    {
      "epoch": 0.4774205538004088,
      "grad_norm": 0.6577771306037903,
      "learning_rate": 0.00017353589252936476,
      "loss": 1.1625,
      "step": 2569
    },
    {
      "epoch": 0.47760639286378,
      "grad_norm": 0.5853607654571533,
      "learning_rate": 0.00017351607009868778,
      "loss": 0.9686,
      "step": 2570
    },
    {
      "epoch": 0.4777922319271511,
      "grad_norm": 0.6094310283660889,
      "learning_rate": 0.00017349624138002242,
      "loss": 1.0382,
      "step": 2571
    },
    {
      "epoch": 0.4779780709905222,
      "grad_norm": 0.5882353782653809,
      "learning_rate": 0.0001734764063750647,
      "loss": 1.0502,
      "step": 2572
    },
    {
      "epoch": 0.4781639100538933,
      "grad_norm": 0.6566441655158997,
      "learning_rate": 0.00017345656508551115,
      "loss": 1.3288,
      "step": 2573
    },
    {
      "epoch": 0.47834974911726447,
      "grad_norm": 0.4997236430644989,
      "learning_rate": 0.0001734367175130588,
      "loss": 1.0965,
      "step": 2574
    },
    {
      "epoch": 0.4785355881806356,
      "grad_norm": 0.5745366811752319,
      "learning_rate": 0.0001734168636594053,
      "loss": 0.9885,
      "step": 2575
    },
    {
      "epoch": 0.4787214272440067,
      "grad_norm": 0.6105988621711731,
      "learning_rate": 0.00017339700352624879,
      "loss": 1.1263,
      "step": 2576
    },
    {
      "epoch": 0.4789072663073778,
      "grad_norm": 0.5690290331840515,
      "learning_rate": 0.00017337713711528796,
      "loss": 1.2127,
      "step": 2577
    },
    {
      "epoch": 0.47909310537074895,
      "grad_norm": 0.6462541818618774,
      "learning_rate": 0.00017335726442822195,
      "loss": 1.216,
      "step": 2578
    },
    {
      "epoch": 0.47927894443412006,
      "grad_norm": 0.640511155128479,
      "learning_rate": 0.00017333738546675058,
      "loss": 0.8523,
      "step": 2579
    },
    {
      "epoch": 0.47946478349749116,
      "grad_norm": 0.5977886319160461,
      "learning_rate": 0.00017331750023257415,
      "loss": 1.0885,
      "step": 2580
    },
    {
      "epoch": 0.47965062256086227,
      "grad_norm": 0.6122512221336365,
      "learning_rate": 0.00017329760872739343,
      "loss": 0.9032,
      "step": 2581
    },
    {
      "epoch": 0.47983646162423343,
      "grad_norm": 0.6344446539878845,
      "learning_rate": 0.00017327771095290983,
      "loss": 1.1595,
      "step": 2582
    },
    {
      "epoch": 0.48002230068760454,
      "grad_norm": 0.7006237506866455,
      "learning_rate": 0.00017325780691082528,
      "loss": 1.2291,
      "step": 2583
    },
    {
      "epoch": 0.48020813975097565,
      "grad_norm": 0.66008061170578,
      "learning_rate": 0.00017323789660284212,
      "loss": 1.1691,
      "step": 2584
    },
    {
      "epoch": 0.48039397881434676,
      "grad_norm": 0.6128259301185608,
      "learning_rate": 0.0001732179800306634,
      "loss": 0.9029,
      "step": 2585
    },
    {
      "epoch": 0.4805798178777179,
      "grad_norm": 0.6071939468383789,
      "learning_rate": 0.0001731980571959926,
      "loss": 1.1245,
      "step": 2586
    },
    {
      "epoch": 0.480765656941089,
      "grad_norm": 0.5949906706809998,
      "learning_rate": 0.00017317812810053378,
      "loss": 0.8654,
      "step": 2587
    },
    {
      "epoch": 0.48095149600446013,
      "grad_norm": 0.6016258001327515,
      "learning_rate": 0.0001731581927459915,
      "loss": 0.8913,
      "step": 2588
    },
    {
      "epoch": 0.48113733506783124,
      "grad_norm": 0.6196433305740356,
      "learning_rate": 0.00017313825113407085,
      "loss": 1.1088,
      "step": 2589
    },
    {
      "epoch": 0.4813231741312024,
      "grad_norm": 0.7779860496520996,
      "learning_rate": 0.00017311830326647751,
      "loss": 0.9617,
      "step": 2590
    },
    {
      "epoch": 0.4815090131945735,
      "grad_norm": 0.5172240138053894,
      "learning_rate": 0.0001730983491449177,
      "loss": 0.9737,
      "step": 2591
    },
    {
      "epoch": 0.4816948522579446,
      "grad_norm": 0.6355507969856262,
      "learning_rate": 0.0001730783887710981,
      "loss": 1.0966,
      "step": 2592
    },
    {
      "epoch": 0.4818806913213157,
      "grad_norm": 0.5902290940284729,
      "learning_rate": 0.0001730584221467259,
      "loss": 1.1274,
      "step": 2593
    },
    {
      "epoch": 0.4820665303846869,
      "grad_norm": 0.6762295365333557,
      "learning_rate": 0.00017303844927350903,
      "loss": 1.3456,
      "step": 2594
    },
    {
      "epoch": 0.482252369448058,
      "grad_norm": 0.6468362808227539,
      "learning_rate": 0.0001730184701531557,
      "loss": 1.2443,
      "step": 2595
    },
    {
      "epoch": 0.4824382085114291,
      "grad_norm": 0.5669111609458923,
      "learning_rate": 0.00017299848478737482,
      "loss": 1.044,
      "step": 2596
    },
    {
      "epoch": 0.4826240475748002,
      "grad_norm": 0.6939884424209595,
      "learning_rate": 0.00017297849317787574,
      "loss": 1.2908,
      "step": 2597
    },
    {
      "epoch": 0.48280988663817137,
      "grad_norm": 0.6661247611045837,
      "learning_rate": 0.00017295849532636846,
      "loss": 1.1268,
      "step": 2598
    },
    {
      "epoch": 0.4829957257015425,
      "grad_norm": 0.6555077433586121,
      "learning_rate": 0.00017293849123456334,
      "loss": 1.1174,
      "step": 2599
    },
    {
      "epoch": 0.4831815647649136,
      "grad_norm": 0.5667238831520081,
      "learning_rate": 0.00017291848090417145,
      "loss": 1.0486,
      "step": 2600
    },
    {
      "epoch": 0.4833674038282847,
      "grad_norm": 0.6136776804924011,
      "learning_rate": 0.00017289846433690426,
      "loss": 1.1264,
      "step": 2601
    },
    {
      "epoch": 0.4835532428916558,
      "grad_norm": 0.5599073171615601,
      "learning_rate": 0.0001728784415344739,
      "loss": 1.1205,
      "step": 2602
    },
    {
      "epoch": 0.48373908195502696,
      "grad_norm": 0.583046555519104,
      "learning_rate": 0.0001728584124985929,
      "loss": 1.1435,
      "step": 2603
    },
    {
      "epoch": 0.48392492101839807,
      "grad_norm": 0.5243253707885742,
      "learning_rate": 0.00017283837723097443,
      "loss": 1.0822,
      "step": 2604
    },
    {
      "epoch": 0.4841107600817692,
      "grad_norm": 0.6579024195671082,
      "learning_rate": 0.00017281833573333212,
      "loss": 1.1433,
      "step": 2605
    },
    {
      "epoch": 0.4842965991451403,
      "grad_norm": 0.593559205532074,
      "learning_rate": 0.00017279828800738017,
      "loss": 0.987,
      "step": 2606
    },
    {
      "epoch": 0.48448243820851145,
      "grad_norm": 0.5837587714195251,
      "learning_rate": 0.0001727782340548333,
      "loss": 1.3241,
      "step": 2607
    },
    {
      "epoch": 0.48466827727188255,
      "grad_norm": 0.5321952104568481,
      "learning_rate": 0.0001727581738774068,
      "loss": 0.935,
      "step": 2608
    },
    {
      "epoch": 0.48485411633525366,
      "grad_norm": 0.6163881421089172,
      "learning_rate": 0.00017273810747681645,
      "loss": 1.1309,
      "step": 2609
    },
    {
      "epoch": 0.48503995539862477,
      "grad_norm": 0.6759121417999268,
      "learning_rate": 0.00017271803485477852,
      "loss": 1.1814,
      "step": 2610
    },
    {
      "epoch": 0.48522579446199593,
      "grad_norm": 0.5714459419250488,
      "learning_rate": 0.00017269795601300994,
      "loss": 1.231,
      "step": 2611
    },
    {
      "epoch": 0.48541163352536704,
      "grad_norm": 0.5658890604972839,
      "learning_rate": 0.000172677870953228,
      "loss": 1.3298,
      "step": 2612
    },
    {
      "epoch": 0.48559747258873814,
      "grad_norm": 0.6630015969276428,
      "learning_rate": 0.00017265777967715072,
      "loss": 1.175,
      "step": 2613
    },
    {
      "epoch": 0.48578331165210925,
      "grad_norm": 0.6266465187072754,
      "learning_rate": 0.00017263768218649654,
      "loss": 1.2725,
      "step": 2614
    },
    {
      "epoch": 0.4859691507154804,
      "grad_norm": 0.5513420701026917,
      "learning_rate": 0.00017261757848298438,
      "loss": 0.9279,
      "step": 2615
    },
    {
      "epoch": 0.4861549897788515,
      "grad_norm": 3.998467445373535,
      "learning_rate": 0.00017259746856833382,
      "loss": 3.0076,
      "step": 2616
    },
    {
      "epoch": 0.48634082884222263,
      "grad_norm": 0.5792141556739807,
      "learning_rate": 0.0001725773524442648,
      "loss": 1.0691,
      "step": 2617
    },
    {
      "epoch": 0.48652666790559373,
      "grad_norm": 0.5263779163360596,
      "learning_rate": 0.00017255723011249803,
      "loss": 1.0758,
      "step": 2618
    },
    {
      "epoch": 0.4867125069689649,
      "grad_norm": 0.5941771268844604,
      "learning_rate": 0.0001725371015747545,
      "loss": 1.0486,
      "step": 2619
    },
    {
      "epoch": 0.486898346032336,
      "grad_norm": 0.6138995885848999,
      "learning_rate": 0.00017251696683275594,
      "loss": 1.1304,
      "step": 2620
    },
    {
      "epoch": 0.4870841850957071,
      "grad_norm": 0.5196933746337891,
      "learning_rate": 0.00017249682588822446,
      "loss": 0.9873,
      "step": 2621
    },
    {
      "epoch": 0.4872700241590782,
      "grad_norm": 0.8865917921066284,
      "learning_rate": 0.00017247667874288277,
      "loss": 1.0958,
      "step": 2622
    },
    {
      "epoch": 0.4874558632224494,
      "grad_norm": 0.6708472371101379,
      "learning_rate": 0.0001724565253984541,
      "loss": 0.9633,
      "step": 2623
    },
    {
      "epoch": 0.4876417022858205,
      "grad_norm": 0.5149091482162476,
      "learning_rate": 0.00017243636585666225,
      "loss": 0.8691,
      "step": 2624
    },
    {
      "epoch": 0.4878275413491916,
      "grad_norm": 0.7160931825637817,
      "learning_rate": 0.00017241620011923146,
      "loss": 1.2219,
      "step": 2625
    },
    {
      "epoch": 0.4880133804125627,
      "grad_norm": 0.6450536251068115,
      "learning_rate": 0.00017239602818788652,
      "loss": 1.1747,
      "step": 2626
    },
    {
      "epoch": 0.48819921947593387,
      "grad_norm": 0.5064762830734253,
      "learning_rate": 0.00017237585006435287,
      "loss": 0.7801,
      "step": 2627
    },
    {
      "epoch": 0.488385058539305,
      "grad_norm": 0.5990808010101318,
      "learning_rate": 0.0001723556657503563,
      "loss": 1.1578,
      "step": 2628
    },
    {
      "epoch": 0.4885708976026761,
      "grad_norm": 0.665460467338562,
      "learning_rate": 0.00017233547524762331,
      "loss": 1.0568,
      "step": 2629
    },
    {
      "epoch": 0.4887567366660472,
      "grad_norm": 0.5016782283782959,
      "learning_rate": 0.00017231527855788073,
      "loss": 1.0718,
      "step": 2630
    },
    {
      "epoch": 0.48894257572941835,
      "grad_norm": 0.5329027771949768,
      "learning_rate": 0.0001722950756828561,
      "loss": 0.9948,
      "step": 2631
    },
    {
      "epoch": 0.48912841479278946,
      "grad_norm": 0.639634370803833,
      "learning_rate": 0.0001722748666242774,
      "loss": 1.02,
      "step": 2632
    },
    {
      "epoch": 0.48931425385616056,
      "grad_norm": 0.6515493988990784,
      "learning_rate": 0.00017225465138387316,
      "loss": 1.0507,
      "step": 2633
    },
    {
      "epoch": 0.48950009291953167,
      "grad_norm": 0.6322322487831116,
      "learning_rate": 0.0001722344299633724,
      "loss": 1.0638,
      "step": 2634
    },
    {
      "epoch": 0.48968593198290283,
      "grad_norm": 0.6251794099807739,
      "learning_rate": 0.00017221420236450475,
      "loss": 1.1392,
      "step": 2635
    },
    {
      "epoch": 0.48987177104627394,
      "grad_norm": 0.6864203214645386,
      "learning_rate": 0.0001721939685890003,
      "loss": 1.17,
      "step": 2636
    },
    {
      "epoch": 0.49005761010964505,
      "grad_norm": 0.5467199683189392,
      "learning_rate": 0.00017217372863858964,
      "loss": 1.1392,
      "step": 2637
    },
    {
      "epoch": 0.49024344917301615,
      "grad_norm": 0.5941484570503235,
      "learning_rate": 0.00017215348251500403,
      "loss": 1.0009,
      "step": 2638
    },
    {
      "epoch": 0.49042928823638726,
      "grad_norm": 0.660819411277771,
      "learning_rate": 0.0001721332302199751,
      "loss": 0.9249,
      "step": 2639
    },
    {
      "epoch": 0.4906151272997584,
      "grad_norm": 0.5856802463531494,
      "learning_rate": 0.0001721129717552351,
      "loss": 1.0534,
      "step": 2640
    },
    {
      "epoch": 0.49080096636312953,
      "grad_norm": 0.6267640590667725,
      "learning_rate": 0.00017209270712251673,
      "loss": 1.1636,
      "step": 2641
    },
    {
      "epoch": 0.49098680542650064,
      "grad_norm": 0.6789271235466003,
      "learning_rate": 0.00017207243632355335,
      "loss": 1.3032,
      "step": 2642
    },
    {
      "epoch": 0.49117264448987175,
      "grad_norm": 0.5530579090118408,
      "learning_rate": 0.0001720521593600787,
      "loss": 1.3485,
      "step": 2643
    },
    {
      "epoch": 0.4913584835532429,
      "grad_norm": 0.6183351874351501,
      "learning_rate": 0.00017203187623382715,
      "loss": 0.9769,
      "step": 2644
    },
    {
      "epoch": 0.491544322616614,
      "grad_norm": 0.5712160468101501,
      "learning_rate": 0.00017201158694653352,
      "loss": 1.1888,
      "step": 2645
    },
    {
      "epoch": 0.4917301616799851,
      "grad_norm": 0.6171578764915466,
      "learning_rate": 0.00017199129149993327,
      "loss": 1.0783,
      "step": 2646
    },
    {
      "epoch": 0.49191600074335623,
      "grad_norm": 0.5475873351097107,
      "learning_rate": 0.00017197098989576222,
      "loss": 1.0462,
      "step": 2647
    },
    {
      "epoch": 0.4921018398067274,
      "grad_norm": 0.6114869117736816,
      "learning_rate": 0.00017195068213575688,
      "loss": 1.1484,
      "step": 2648
    },
    {
      "epoch": 0.4922876788700985,
      "grad_norm": 0.5490365028381348,
      "learning_rate": 0.00017193036822165417,
      "loss": 1.1458,
      "step": 2649
    },
    {
      "epoch": 0.4924735179334696,
      "grad_norm": 0.5615825653076172,
      "learning_rate": 0.0001719100481551916,
      "loss": 0.9985,
      "step": 2650
    },
    {
      "epoch": 0.4926593569968407,
      "grad_norm": 0.6177067160606384,
      "learning_rate": 0.00017188972193810722,
      "loss": 0.9851,
      "step": 2651
    },
    {
      "epoch": 0.4928451960602119,
      "grad_norm": 0.550390362739563,
      "learning_rate": 0.00017186938957213953,
      "loss": 0.8936,
      "step": 2652
    },
    {
      "epoch": 0.493031035123583,
      "grad_norm": 0.5912401676177979,
      "learning_rate": 0.0001718490510590276,
      "loss": 1.1753,
      "step": 2653
    },
    {
      "epoch": 0.4932168741869541,
      "grad_norm": 0.7773668169975281,
      "learning_rate": 0.00017182870640051107,
      "loss": 1.1211,
      "step": 2654
    },
    {
      "epoch": 0.4934027132503252,
      "grad_norm": 0.6435762643814087,
      "learning_rate": 0.00017180835559833004,
      "loss": 1.2553,
      "step": 2655
    },
    {
      "epoch": 0.49358855231369636,
      "grad_norm": 0.570472240447998,
      "learning_rate": 0.00017178799865422514,
      "loss": 1.1645,
      "step": 2656
    },
    {
      "epoch": 0.49377439137706747,
      "grad_norm": 0.6846404075622559,
      "learning_rate": 0.0001717676355699376,
      "loss": 1.1736,
      "step": 2657
    },
    {
      "epoch": 0.4939602304404386,
      "grad_norm": 0.6840716004371643,
      "learning_rate": 0.00017174726634720905,
      "loss": 1.2924,
      "step": 2658
    },
    {
      "epoch": 0.4941460695038097,
      "grad_norm": 0.6342020034790039,
      "learning_rate": 0.00017172689098778177,
      "loss": 1.0777,
      "step": 2659
    },
    {
      "epoch": 0.49433190856718084,
      "grad_norm": 0.5968679785728455,
      "learning_rate": 0.00017170650949339847,
      "loss": 0.7009,
      "step": 2660
    },
    {
      "epoch": 0.49451774763055195,
      "grad_norm": 0.6246665716171265,
      "learning_rate": 0.00017168612186580238,
      "loss": 1.0159,
      "step": 2661
    },
    {
      "epoch": 0.49470358669392306,
      "grad_norm": 0.5966857671737671,
      "learning_rate": 0.00017166572810673744,
      "loss": 1.1489,
      "step": 2662
    },
    {
      "epoch": 0.49488942575729417,
      "grad_norm": 0.5603086352348328,
      "learning_rate": 0.0001716453282179478,
      "loss": 1.1315,
      "step": 2663
    },
    {
      "epoch": 0.49507526482066533,
      "grad_norm": 0.5681596994400024,
      "learning_rate": 0.00017162492220117845,
      "loss": 1.2733,
      "step": 2664
    },
    {
      "epoch": 0.49526110388403644,
      "grad_norm": 0.6216878294944763,
      "learning_rate": 0.00017160451005817468,
      "loss": 1.0994,
      "step": 2665
    },
    {
      "epoch": 0.49544694294740754,
      "grad_norm": 0.6957216262817383,
      "learning_rate": 0.00017158409179068244,
      "loss": 0.8731,
      "step": 2666
    },
    {
      "epoch": 0.49563278201077865,
      "grad_norm": 0.7972751259803772,
      "learning_rate": 0.00017156366740044807,
      "loss": 1.0435,
      "step": 2667
    },
    {
      "epoch": 0.4958186210741498,
      "grad_norm": 0.5390364527702332,
      "learning_rate": 0.0001715432368892186,
      "loss": 1.1918,
      "step": 2668
    },
    {
      "epoch": 0.4960044601375209,
      "grad_norm": 0.5905940532684326,
      "learning_rate": 0.00017152280025874142,
      "loss": 1.1615,
      "step": 2669
    },
    {
      "epoch": 0.496190299200892,
      "grad_norm": 0.5240758657455444,
      "learning_rate": 0.00017150235751076456,
      "loss": 1.1469,
      "step": 2670
    },
    {
      "epoch": 0.49637613826426313,
      "grad_norm": 0.630538284778595,
      "learning_rate": 0.00017148190864703654,
      "loss": 0.9101,
      "step": 2671
    },
    {
      "epoch": 0.4965619773276343,
      "grad_norm": 0.5258409976959229,
      "learning_rate": 0.00017146145366930634,
      "loss": 1.2836,
      "step": 2672
    },
    {
      "epoch": 0.4967478163910054,
      "grad_norm": 0.5378925800323486,
      "learning_rate": 0.00017144099257932364,
      "loss": 0.9881,
      "step": 2673
    },
    {
      "epoch": 0.4969336554543765,
      "grad_norm": 0.6253618597984314,
      "learning_rate": 0.00017142052537883834,
      "loss": 0.6877,
      "step": 2674
    },
    {
      "epoch": 0.4971194945177476,
      "grad_norm": 0.5238915681838989,
      "learning_rate": 0.00017140005206960123,
      "loss": 0.9998,
      "step": 2675
    },
    {
      "epoch": 0.4973053335811187,
      "grad_norm": 0.6569894552230835,
      "learning_rate": 0.0001713795726533633,
      "loss": 1.1629,
      "step": 2676
    },
    {
      "epoch": 0.4974911726444899,
      "grad_norm": 0.6058579683303833,
      "learning_rate": 0.00017135908713187628,
      "loss": 1.218,
      "step": 2677
    },
    {
      "epoch": 0.497677011707861,
      "grad_norm": 0.6192750930786133,
      "learning_rate": 0.0001713385955068923,
      "loss": 1.1703,
      "step": 2678
    },
    {
      "epoch": 0.4978628507712321,
      "grad_norm": 0.5908500552177429,
      "learning_rate": 0.00017131809778016404,
      "loss": 1.2753,
      "step": 2679
    },
    {
      "epoch": 0.4980486898346032,
      "grad_norm": 0.6565137505531311,
      "learning_rate": 0.00017129759395344475,
      "loss": 1.0112,
      "step": 2680
    },
    {
      "epoch": 0.49823452889797437,
      "grad_norm": 0.621427059173584,
      "learning_rate": 0.00017127708402848817,
      "loss": 1.0148,
      "step": 2681
    },
    {
      "epoch": 0.4984203679613455,
      "grad_norm": 0.5895846486091614,
      "learning_rate": 0.00017125656800704852,
      "loss": 1.039,
      "step": 2682
    },
    {
      "epoch": 0.4986062070247166,
      "grad_norm": 0.5865383148193359,
      "learning_rate": 0.00017123604589088063,
      "loss": 1.0531,
      "step": 2683
    },
    {
      "epoch": 0.4987920460880877,
      "grad_norm": 0.6337679624557495,
      "learning_rate": 0.00017121551768173977,
      "loss": 1.2365,
      "step": 2684
    },
    {
      "epoch": 0.49897788515145886,
      "grad_norm": 0.6088318824768066,
      "learning_rate": 0.00017119498338138172,
      "loss": 1.3512,
      "step": 2685
    },
    {
      "epoch": 0.49916372421482996,
      "grad_norm": 0.6825537085533142,
      "learning_rate": 0.00017117444299156293,
      "loss": 1.2751,
      "step": 2686
    },
    {
      "epoch": 0.49934956327820107,
      "grad_norm": 0.5798403024673462,
      "learning_rate": 0.00017115389651404018,
      "loss": 0.8005,
      "step": 2687
    },
    {
      "epoch": 0.4995354023415722,
      "grad_norm": 0.5042392611503601,
      "learning_rate": 0.00017113334395057087,
      "loss": 1.1635,
      "step": 2688
    },
    {
      "epoch": 0.49972124140494334,
      "grad_norm": 0.5946384072303772,
      "learning_rate": 0.0001711127853029129,
      "loss": 1.1139,
      "step": 2689
    },
    {
      "epoch": 0.49990708046831445,
      "grad_norm": 0.5286500453948975,
      "learning_rate": 0.00017109222057282476,
      "loss": 0.7885,
      "step": 2690
    },
    {
      "epoch": 0.5000929195316856,
      "grad_norm": 0.6756643056869507,
      "learning_rate": 0.0001710716497620653,
      "loss": 1.3519,
      "step": 2691
    },
    {
      "epoch": 0.5002787585950567,
      "grad_norm": 0.6427111625671387,
      "learning_rate": 0.00017105107287239402,
      "loss": 1.0801,
      "step": 2692
    },
    {
      "epoch": 0.5002787585950567,
      "eval_loss": 1.073055386543274,
      "eval_runtime": 23.0111,
      "eval_samples_per_second": 47.455,
      "eval_steps_per_second": 23.728,
      "step": 2692
    },
    {
      "epoch": 0.5004645976584278,
      "grad_norm": 0.6261104345321655,
      "learning_rate": 0.00017103048990557094,
      "loss": 0.956,
      "step": 2693
    },
    {
      "epoch": 0.5006504367217989,
      "grad_norm": 0.5339560508728027,
      "learning_rate": 0.00017100990086335653,
      "loss": 1.0753,
      "step": 2694
    },
    {
      "epoch": 0.5008362757851701,
      "grad_norm": 0.6238815188407898,
      "learning_rate": 0.00017098930574751183,
      "loss": 1.1003,
      "step": 2695
    },
    {
      "epoch": 0.5010221148485412,
      "grad_norm": 0.6724877953529358,
      "learning_rate": 0.00017096870455979838,
      "loss": 1.2206,
      "step": 2696
    },
    {
      "epoch": 0.5012079539119123,
      "grad_norm": 0.49802348017692566,
      "learning_rate": 0.00017094809730197825,
      "loss": 0.9445,
      "step": 2697
    },
    {
      "epoch": 0.5013937929752834,
      "grad_norm": 0.6113379597663879,
      "learning_rate": 0.000170927483975814,
      "loss": 1.1136,
      "step": 2698
    },
    {
      "epoch": 0.5015796320386545,
      "grad_norm": 0.49268612265586853,
      "learning_rate": 0.0001709068645830688,
      "loss": 1.0709,
      "step": 2699
    },
    {
      "epoch": 0.5017654711020256,
      "grad_norm": 0.7463451623916626,
      "learning_rate": 0.00017088623912550616,
      "loss": 1.2494,
      "step": 2700
    },
    {
      "epoch": 0.5019513101653967,
      "grad_norm": 3.880950450897217,
      "learning_rate": 0.0001708656076048903,
      "loss": 2.6297,
      "step": 2701
    },
    {
      "epoch": 0.5021371492287678,
      "grad_norm": 0.679601788520813,
      "learning_rate": 0.00017084497002298592,
      "loss": 1.0374,
      "step": 2702
    },
    {
      "epoch": 0.5023229882921391,
      "grad_norm": 0.5280315279960632,
      "learning_rate": 0.00017082432638155807,
      "loss": 1.0968,
      "step": 2703
    },
    {
      "epoch": 0.5025088273555102,
      "grad_norm": 0.676991879940033,
      "learning_rate": 0.00017080367668237254,
      "loss": 1.1099,
      "step": 2704
    },
    {
      "epoch": 0.5026946664188813,
      "grad_norm": 0.6209761500358582,
      "learning_rate": 0.0001707830209271955,
      "loss": 1.1345,
      "step": 2705
    },
    {
      "epoch": 0.5028805054822524,
      "grad_norm": 0.5609131455421448,
      "learning_rate": 0.00017076235911779372,
      "loss": 0.9599,
      "step": 2706
    },
    {
      "epoch": 0.5030663445456235,
      "grad_norm": 0.5717629790306091,
      "learning_rate": 0.00017074169125593443,
      "loss": 1.2308,
      "step": 2707
    },
    {
      "epoch": 0.5032521836089946,
      "grad_norm": 0.5387877225875854,
      "learning_rate": 0.00017072101734338536,
      "loss": 1.0986,
      "step": 2708
    },
    {
      "epoch": 0.5034380226723657,
      "grad_norm": 0.6486318111419678,
      "learning_rate": 0.00017070033738191483,
      "loss": 1.1697,
      "step": 2709
    },
    {
      "epoch": 0.5036238617357368,
      "grad_norm": 0.5692694783210754,
      "learning_rate": 0.0001706796513732917,
      "loss": 1.0235,
      "step": 2710
    },
    {
      "epoch": 0.5038097007991079,
      "grad_norm": 0.7457370162010193,
      "learning_rate": 0.00017065895931928516,
      "loss": 0.9989,
      "step": 2711
    },
    {
      "epoch": 0.5039955398624791,
      "grad_norm": 0.49672433733940125,
      "learning_rate": 0.00017063826122166518,
      "loss": 0.8731,
      "step": 2712
    },
    {
      "epoch": 0.5041813789258502,
      "grad_norm": 0.6872747540473938,
      "learning_rate": 0.000170617557082202,
      "loss": 1.2155,
      "step": 2713
    },
    {
      "epoch": 0.5043672179892214,
      "grad_norm": 0.6377733945846558,
      "learning_rate": 0.00017059684690266656,
      "loss": 0.9905,
      "step": 2714
    },
    {
      "epoch": 0.5045530570525925,
      "grad_norm": 0.6393523216247559,
      "learning_rate": 0.0001705761306848302,
      "loss": 1.2727,
      "step": 2715
    },
    {
      "epoch": 0.5047388961159636,
      "grad_norm": 0.5900831818580627,
      "learning_rate": 0.00017055540843046485,
      "loss": 1.0815,
      "step": 2716
    },
    {
      "epoch": 0.5049247351793347,
      "grad_norm": 0.5988309383392334,
      "learning_rate": 0.00017053468014134295,
      "loss": 1.099,
      "step": 2717
    },
    {
      "epoch": 0.5051105742427058,
      "grad_norm": 0.6535857915878296,
      "learning_rate": 0.0001705139458192374,
      "loss": 1.0273,
      "step": 2718
    },
    {
      "epoch": 0.5052964133060769,
      "grad_norm": 0.5833021998405457,
      "learning_rate": 0.00017049320546592167,
      "loss": 0.9237,
      "step": 2719
    },
    {
      "epoch": 0.5054822523694481,
      "grad_norm": 0.6732074618339539,
      "learning_rate": 0.00017047245908316972,
      "loss": 0.962,
      "step": 2720
    },
    {
      "epoch": 0.5056680914328192,
      "grad_norm": 0.5957641005516052,
      "learning_rate": 0.000170451706672756,
      "loss": 1.1259,
      "step": 2721
    },
    {
      "epoch": 0.5058539304961903,
      "grad_norm": 0.640815794467926,
      "learning_rate": 0.0001704309482364556,
      "loss": 1.0514,
      "step": 2722
    },
    {
      "epoch": 0.5060397695595614,
      "grad_norm": 0.5656802654266357,
      "learning_rate": 0.00017041018377604393,
      "loss": 1.0022,
      "step": 2723
    },
    {
      "epoch": 0.5062256086229325,
      "grad_norm": 0.6468055844306946,
      "learning_rate": 0.00017038941329329709,
      "loss": 1.1897,
      "step": 2724
    },
    {
      "epoch": 0.5064114476863036,
      "grad_norm": 0.6397135853767395,
      "learning_rate": 0.00017036863678999157,
      "loss": 0.6603,
      "step": 2725
    },
    {
      "epoch": 0.5065972867496747,
      "grad_norm": 0.5529398918151855,
      "learning_rate": 0.0001703478542679045,
      "loss": 0.9799,
      "step": 2726
    },
    {
      "epoch": 0.5067831258130459,
      "grad_norm": 0.7279483079910278,
      "learning_rate": 0.0001703270657288134,
      "loss": 1.4637,
      "step": 2727
    },
    {
      "epoch": 0.5069689648764171,
      "grad_norm": 0.5699883699417114,
      "learning_rate": 0.00017030627117449636,
      "loss": 1.0844,
      "step": 2728
    },
    {
      "epoch": 0.5071548039397882,
      "grad_norm": 0.663766086101532,
      "learning_rate": 0.000170285470606732,
      "loss": 1.1391,
      "step": 2729
    },
    {
      "epoch": 0.5073406430031593,
      "grad_norm": 0.5575130581855774,
      "learning_rate": 0.00017026466402729942,
      "loss": 1.1047,
      "step": 2730
    },
    {
      "epoch": 0.5075264820665304,
      "grad_norm": 0.6974256634712219,
      "learning_rate": 0.00017024385143797827,
      "loss": 0.9739,
      "step": 2731
    },
    {
      "epoch": 0.5077123211299015,
      "grad_norm": 0.5451050996780396,
      "learning_rate": 0.0001702230328405487,
      "loss": 0.9685,
      "step": 2732
    },
    {
      "epoch": 0.5078981601932726,
      "grad_norm": 0.5307935476303101,
      "learning_rate": 0.00017020220823679133,
      "loss": 1.31,
      "step": 2733
    },
    {
      "epoch": 0.5080839992566437,
      "grad_norm": 0.5251340866088867,
      "learning_rate": 0.00017018137762848742,
      "loss": 1.0396,
      "step": 2734
    },
    {
      "epoch": 0.5082698383200148,
      "grad_norm": 0.5450385212898254,
      "learning_rate": 0.00017016054101741857,
      "loss": 1.0322,
      "step": 2735
    },
    {
      "epoch": 0.508455677383386,
      "grad_norm": 0.661077618598938,
      "learning_rate": 0.000170139698405367,
      "loss": 1.2001,
      "step": 2736
    },
    {
      "epoch": 0.5086415164467571,
      "grad_norm": 0.5541077852249146,
      "learning_rate": 0.00017011884979411542,
      "loss": 0.9658,
      "step": 2737
    },
    {
      "epoch": 0.5088273555101283,
      "grad_norm": 0.610466718673706,
      "learning_rate": 0.0001700979951854471,
      "loss": 1.3137,
      "step": 2738
    },
    {
      "epoch": 0.5090131945734994,
      "grad_norm": 0.6601592898368835,
      "learning_rate": 0.00017007713458114574,
      "loss": 1.1539,
      "step": 2739
    },
    {
      "epoch": 0.5091990336368705,
      "grad_norm": 0.6103355288505554,
      "learning_rate": 0.0001700562679829956,
      "loss": 0.6126,
      "step": 2740
    },
    {
      "epoch": 0.5093848727002416,
      "grad_norm": 0.6720930933952332,
      "learning_rate": 0.00017003539539278144,
      "loss": 1.3313,
      "step": 2741
    },
    {
      "epoch": 0.5095707117636127,
      "grad_norm": 0.5944845080375671,
      "learning_rate": 0.00017001451681228856,
      "loss": 0.7785,
      "step": 2742
    },
    {
      "epoch": 0.5097565508269838,
      "grad_norm": 0.5247288942337036,
      "learning_rate": 0.00016999363224330273,
      "loss": 1.1125,
      "step": 2743
    },
    {
      "epoch": 0.509942389890355,
      "grad_norm": 0.6319442987442017,
      "learning_rate": 0.00016997274168761026,
      "loss": 1.3289,
      "step": 2744
    },
    {
      "epoch": 0.5101282289537261,
      "grad_norm": 0.7262590527534485,
      "learning_rate": 0.00016995184514699796,
      "loss": 1.2919,
      "step": 2745
    },
    {
      "epoch": 0.5103140680170972,
      "grad_norm": 0.5297901630401611,
      "learning_rate": 0.00016993094262325316,
      "loss": 0.9359,
      "step": 2746
    },
    {
      "epoch": 0.5104999070804683,
      "grad_norm": 0.6656481623649597,
      "learning_rate": 0.0001699100341181637,
      "loss": 1.1935,
      "step": 2747
    },
    {
      "epoch": 0.5106857461438394,
      "grad_norm": 0.6488802433013916,
      "learning_rate": 0.00016988911963351794,
      "loss": 1.0575,
      "step": 2748
    },
    {
      "epoch": 0.5108715852072105,
      "grad_norm": 0.6008021235466003,
      "learning_rate": 0.0001698681991711047,
      "loss": 1.1663,
      "step": 2749
    },
    {
      "epoch": 0.5110574242705817,
      "grad_norm": 0.7873336672782898,
      "learning_rate": 0.00016984727273271338,
      "loss": 1.2606,
      "step": 2750
    },
    {
      "epoch": 0.5112432633339528,
      "grad_norm": 0.612949788570404,
      "learning_rate": 0.0001698263403201339,
      "loss": 1.0826,
      "step": 2751
    },
    {
      "epoch": 0.5114291023973239,
      "grad_norm": 0.5440847873687744,
      "learning_rate": 0.00016980540193515658,
      "loss": 0.9785,
      "step": 2752
    },
    {
      "epoch": 0.5116149414606951,
      "grad_norm": 0.6771647930145264,
      "learning_rate": 0.00016978445757957238,
      "loss": 0.9802,
      "step": 2753
    },
    {
      "epoch": 0.5118007805240662,
      "grad_norm": 0.5409290194511414,
      "learning_rate": 0.0001697635072551727,
      "loss": 1.0132,
      "step": 2754
    },
    {
      "epoch": 0.5119866195874373,
      "grad_norm": 0.5121529698371887,
      "learning_rate": 0.00016974255096374944,
      "loss": 1.1907,
      "step": 2755
    },
    {
      "epoch": 0.5121724586508084,
      "grad_norm": 0.9652408957481384,
      "learning_rate": 0.0001697215887070951,
      "loss": 1.1317,
      "step": 2756
    },
    {
      "epoch": 0.5123582977141795,
      "grad_norm": 0.5094935894012451,
      "learning_rate": 0.00016970062048700258,
      "loss": 0.9606,
      "step": 2757
    },
    {
      "epoch": 0.5125441367775506,
      "grad_norm": 0.6005422472953796,
      "learning_rate": 0.00016967964630526533,
      "loss": 0.9366,
      "step": 2758
    },
    {
      "epoch": 0.5127299758409217,
      "grad_norm": 0.647412121295929,
      "learning_rate": 0.00016965866616367733,
      "loss": 1.3646,
      "step": 2759
    },
    {
      "epoch": 0.5129158149042928,
      "grad_norm": 0.6332395672798157,
      "learning_rate": 0.0001696376800640331,
      "loss": 0.9569,
      "step": 2760
    },
    {
      "epoch": 0.513101653967664,
      "grad_norm": 0.5023225545883179,
      "learning_rate": 0.00016961668800812753,
      "loss": 0.8663,
      "step": 2761
    },
    {
      "epoch": 0.5132874930310352,
      "grad_norm": 0.554136335849762,
      "learning_rate": 0.0001695956899977562,
      "loss": 1.206,
      "step": 2762
    },
    {
      "epoch": 0.5134733320944063,
      "grad_norm": 0.5565813183784485,
      "learning_rate": 0.00016957468603471512,
      "loss": 1.0772,
      "step": 2763
    },
    {
      "epoch": 0.5136591711577774,
      "grad_norm": 3.6855132579803467,
      "learning_rate": 0.00016955367612080075,
      "loss": 2.6807,
      "step": 2764
    },
    {
      "epoch": 0.5138450102211485,
      "grad_norm": 0.8236385583877563,
      "learning_rate": 0.0001695326602578101,
      "loss": 1.0427,
      "step": 2765
    },
    {
      "epoch": 0.5140308492845196,
      "grad_norm": 0.5622274279594421,
      "learning_rate": 0.00016951163844754078,
      "loss": 0.9239,
      "step": 2766
    },
    {
      "epoch": 0.5142166883478907,
      "grad_norm": 0.580856442451477,
      "learning_rate": 0.0001694906106917908,
      "loss": 0.8981,
      "step": 2767
    },
    {
      "epoch": 0.5144025274112618,
      "grad_norm": 0.577639102935791,
      "learning_rate": 0.00016946957699235865,
      "loss": 1.1712,
      "step": 2768
    },
    {
      "epoch": 0.514588366474633,
      "grad_norm": 0.7837035655975342,
      "learning_rate": 0.00016944853735104348,
      "loss": 1.3395,
      "step": 2769
    },
    {
      "epoch": 0.5147742055380041,
      "grad_norm": 0.4840414524078369,
      "learning_rate": 0.0001694274917696448,
      "loss": 0.9164,
      "step": 2770
    },
    {
      "epoch": 0.5149600446013752,
      "grad_norm": 0.6124632954597473,
      "learning_rate": 0.00016940644024996272,
      "loss": 0.8041,
      "step": 2771
    },
    {
      "epoch": 0.5151458836647463,
      "grad_norm": 0.5656507015228271,
      "learning_rate": 0.00016938538279379772,
      "loss": 1.2659,
      "step": 2772
    },
    {
      "epoch": 0.5153317227281174,
      "grad_norm": 0.545505702495575,
      "learning_rate": 0.00016936431940295106,
      "loss": 1.0758,
      "step": 2773
    },
    {
      "epoch": 0.5155175617914886,
      "grad_norm": 0.5941258668899536,
      "learning_rate": 0.00016934325007922417,
      "loss": 0.9539,
      "step": 2774
    },
    {
      "epoch": 0.5157034008548597,
      "grad_norm": 0.5610512495040894,
      "learning_rate": 0.00016932217482441928,
      "loss": 1.2581,
      "step": 2775
    },
    {
      "epoch": 0.5158892399182308,
      "grad_norm": 0.655250072479248,
      "learning_rate": 0.00016930109364033893,
      "loss": 1.1122,
      "step": 2776
    },
    {
      "epoch": 0.516075078981602,
      "grad_norm": 0.5420289635658264,
      "learning_rate": 0.00016928000652878628,
      "loss": 0.9652,
      "step": 2777
    },
    {
      "epoch": 0.5162609180449731,
      "grad_norm": 0.5724037289619446,
      "learning_rate": 0.00016925891349156493,
      "loss": 1.0721,
      "step": 2778
    },
    {
      "epoch": 0.5164467571083442,
      "grad_norm": 0.4908211827278137,
      "learning_rate": 0.000169237814530479,
      "loss": 1.1539,
      "step": 2779
    },
    {
      "epoch": 0.5166325961717153,
      "grad_norm": 0.690380334854126,
      "learning_rate": 0.0001692167096473332,
      "loss": 1.3287,
      "step": 2780
    },
    {
      "epoch": 0.5168184352350864,
      "grad_norm": 0.5678565502166748,
      "learning_rate": 0.00016919559884393258,
      "loss": 1.4168,
      "step": 2781
    },
    {
      "epoch": 0.5170042742984575,
      "grad_norm": 5.411264419555664,
      "learning_rate": 0.00016917448212208283,
      "loss": 3.2767,
      "step": 2782
    },
    {
      "epoch": 0.5171901133618286,
      "grad_norm": 0.5628544688224792,
      "learning_rate": 0.00016915335948359018,
      "loss": 1.141,
      "step": 2783
    },
    {
      "epoch": 0.5173759524251997,
      "grad_norm": 0.4003027081489563,
      "learning_rate": 0.0001691322309302612,
      "loss": 0.6452,
      "step": 2784
    },
    {
      "epoch": 0.5175617914885708,
      "grad_norm": 0.5444847345352173,
      "learning_rate": 0.00016911109646390306,
      "loss": 1.0142,
      "step": 2785
    },
    {
      "epoch": 0.5177476305519421,
      "grad_norm": 0.498196005821228,
      "learning_rate": 0.0001690899560863235,
      "loss": 0.918,
      "step": 2786
    },
    {
      "epoch": 0.5179334696153132,
      "grad_norm": 0.5275631546974182,
      "learning_rate": 0.00016906880979933068,
      "loss": 1.3385,
      "step": 2787
    },
    {
      "epoch": 0.5181193086786843,
      "grad_norm": 0.5868602991104126,
      "learning_rate": 0.00016904765760473328,
      "loss": 1.2404,
      "step": 2788
    },
    {
      "epoch": 0.5183051477420554,
      "grad_norm": 0.5333208441734314,
      "learning_rate": 0.0001690264995043405,
      "loss": 0.8179,
      "step": 2789
    },
    {
      "epoch": 0.5184909868054265,
      "grad_norm": 0.6185091137886047,
      "learning_rate": 0.00016900533549996203,
      "loss": 1.0924,
      "step": 2790
    },
    {
      "epoch": 0.5186768258687976,
      "grad_norm": 0.517379641532898,
      "learning_rate": 0.00016898416559340805,
      "loss": 0.8873,
      "step": 2791
    },
    {
      "epoch": 0.5188626649321687,
      "grad_norm": 0.6053199768066406,
      "learning_rate": 0.0001689629897864893,
      "loss": 0.9401,
      "step": 2792
    },
    {
      "epoch": 0.5190485039955398,
      "grad_norm": 0.6556801199913025,
      "learning_rate": 0.00016894180808101698,
      "loss": 1.0591,
      "step": 2793
    },
    {
      "epoch": 0.519234343058911,
      "grad_norm": 0.60248863697052,
      "learning_rate": 0.00016892062047880283,
      "loss": 1.1708,
      "step": 2794
    },
    {
      "epoch": 0.5194201821222821,
      "grad_norm": 0.5984275937080383,
      "learning_rate": 0.00016889942698165907,
      "loss": 0.9006,
      "step": 2795
    },
    {
      "epoch": 0.5196060211856532,
      "grad_norm": 0.5322675704956055,
      "learning_rate": 0.00016887822759139836,
      "loss": 1.0042,
      "step": 2796
    },
    {
      "epoch": 0.5197918602490244,
      "grad_norm": 0.5319322943687439,
      "learning_rate": 0.00016885702230983403,
      "loss": 1.0598,
      "step": 2797
    },
    {
      "epoch": 0.5199776993123955,
      "grad_norm": 0.6360406875610352,
      "learning_rate": 0.00016883581113877973,
      "loss": 0.9902,
      "step": 2798
    },
    {
      "epoch": 0.5201635383757666,
      "grad_norm": 0.5423364043235779,
      "learning_rate": 0.0001688145940800497,
      "loss": 1.0555,
      "step": 2799
    },
    {
      "epoch": 0.5203493774391377,
      "grad_norm": 0.5447130799293518,
      "learning_rate": 0.00016879337113545877,
      "loss": 1.0019,
      "step": 2800
    },
    {
      "epoch": 0.5205352165025088,
      "grad_norm": 0.6677650213241577,
      "learning_rate": 0.0001687721423068221,
      "loss": 1.3007,
      "step": 2801
    },
    {
      "epoch": 0.52072105556588,
      "grad_norm": 0.6144031882286072,
      "learning_rate": 0.00016875090759595546,
      "loss": 1.2085,
      "step": 2802
    },
    {
      "epoch": 0.5209068946292511,
      "grad_norm": 0.5660108327865601,
      "learning_rate": 0.0001687296670046751,
      "loss": 1.1709,
      "step": 2803
    },
    {
      "epoch": 0.5210927336926222,
      "grad_norm": 0.49554237723350525,
      "learning_rate": 0.0001687084205347978,
      "loss": 1.0253,
      "step": 2804
    },
    {
      "epoch": 0.5212785727559933,
      "grad_norm": 0.7389078140258789,
      "learning_rate": 0.0001686871681881408,
      "loss": 1.0017,
      "step": 2805
    },
    {
      "epoch": 0.5214644118193644,
      "grad_norm": 0.598188042640686,
      "learning_rate": 0.0001686659099665218,
      "loss": 1.024,
      "step": 2806
    },
    {
      "epoch": 0.5216502508827355,
      "grad_norm": 0.5329128503799438,
      "learning_rate": 0.00016864464587175912,
      "loss": 1.3472,
      "step": 2807
    },
    {
      "epoch": 0.5218360899461066,
      "grad_norm": 0.6566879749298096,
      "learning_rate": 0.00016862337590567157,
      "loss": 1.0788,
      "step": 2808
    },
    {
      "epoch": 0.5220219290094777,
      "grad_norm": 0.6406189799308777,
      "learning_rate": 0.00016860210007007832,
      "loss": 1.0195,
      "step": 2809
    },
    {
      "epoch": 0.522207768072849,
      "grad_norm": 0.612294614315033,
      "learning_rate": 0.00016858081836679923,
      "loss": 1.1664,
      "step": 2810
    },
    {
      "epoch": 0.5223936071362201,
      "grad_norm": 0.5791414976119995,
      "learning_rate": 0.00016855953079765448,
      "loss": 1.0084,
      "step": 2811
    },
    {
      "epoch": 0.5225794461995912,
      "grad_norm": 0.552058219909668,
      "learning_rate": 0.00016853823736446487,
      "loss": 1.3248,
      "step": 2812
    },
    {
      "epoch": 0.5227652852629623,
      "grad_norm": 0.6082121729850769,
      "learning_rate": 0.0001685169380690517,
      "loss": 0.9751,
      "step": 2813
    },
    {
      "epoch": 0.5229511243263334,
      "grad_norm": 0.5541410446166992,
      "learning_rate": 0.00016849563291323675,
      "loss": 1.2256,
      "step": 2814
    },
    {
      "epoch": 0.5231369633897045,
      "grad_norm": 0.5580676794052124,
      "learning_rate": 0.0001684743218988423,
      "loss": 1.1095,
      "step": 2815
    },
    {
      "epoch": 0.5233228024530756,
      "grad_norm": 0.5919464230537415,
      "learning_rate": 0.00016845300502769107,
      "loss": 1.0043,
      "step": 2816
    },
    {
      "epoch": 0.5235086415164467,
      "grad_norm": 0.5584767460823059,
      "learning_rate": 0.00016843168230160638,
      "loss": 1.1576,
      "step": 2817
    },
    {
      "epoch": 0.5236944805798178,
      "grad_norm": 0.5251010060310364,
      "learning_rate": 0.000168410353722412,
      "loss": 1.1172,
      "step": 2818
    },
    {
      "epoch": 0.523880319643189,
      "grad_norm": 0.5411672592163086,
      "learning_rate": 0.00016838901929193222,
      "loss": 1.0392,
      "step": 2819
    },
    {
      "epoch": 0.5240661587065601,
      "grad_norm": 0.5992206931114197,
      "learning_rate": 0.0001683676790119918,
      "loss": 1.056,
      "step": 2820
    },
    {
      "epoch": 0.5242519977699313,
      "grad_norm": 0.7021682858467102,
      "learning_rate": 0.00016834633288441607,
      "loss": 1.5045,
      "step": 2821
    },
    {
      "epoch": 0.5244378368333024,
      "grad_norm": 0.6393280625343323,
      "learning_rate": 0.00016832498091103073,
      "loss": 1.1608,
      "step": 2822
    },
    {
      "epoch": 0.5246236758966735,
      "grad_norm": 0.6091069579124451,
      "learning_rate": 0.0001683036230936621,
      "loss": 1.278,
      "step": 2823
    },
    {
      "epoch": 0.5248095149600446,
      "grad_norm": 0.6061083078384399,
      "learning_rate": 0.00016828225943413697,
      "loss": 1.0032,
      "step": 2824
    },
    {
      "epoch": 0.5249953540234157,
      "grad_norm": 0.5480764508247375,
      "learning_rate": 0.00016826088993428264,
      "loss": 1.1557,
      "step": 2825
    },
    {
      "epoch": 0.5251811930867868,
      "grad_norm": 0.5672341585159302,
      "learning_rate": 0.00016823951459592684,
      "loss": 1.1974,
      "step": 2826
    },
    {
      "epoch": 0.525367032150158,
      "grad_norm": 0.6575129628181458,
      "learning_rate": 0.00016821813342089787,
      "loss": 1.1193,
      "step": 2827
    },
    {
      "epoch": 0.5255528712135291,
      "grad_norm": 0.5681663751602173,
      "learning_rate": 0.00016819674641102453,
      "loss": 0.9461,
      "step": 2828
    },
    {
      "epoch": 0.5257387102769002,
      "grad_norm": 0.5635049343109131,
      "learning_rate": 0.00016817535356813604,
      "loss": 1.0586,
      "step": 2829
    },
    {
      "epoch": 0.5259245493402713,
      "grad_norm": 0.6108316779136658,
      "learning_rate": 0.00016815395489406227,
      "loss": 1.2576,
      "step": 2830
    },
    {
      "epoch": 0.5261103884036424,
      "grad_norm": 0.6750257611274719,
      "learning_rate": 0.0001681325503906334,
      "loss": 1.1878,
      "step": 2831
    },
    {
      "epoch": 0.5262962274670135,
      "grad_norm": 0.6346907019615173,
      "learning_rate": 0.00016811114005968025,
      "loss": 1.0485,
      "step": 2832
    },
    {
      "epoch": 0.5264820665303847,
      "grad_norm": 0.6290209889411926,
      "learning_rate": 0.00016808972390303407,
      "loss": 1.2745,
      "step": 2833
    },
    {
      "epoch": 0.5266679055937558,
      "grad_norm": 0.6105024218559265,
      "learning_rate": 0.00016806830192252664,
      "loss": 1.0585,
      "step": 2834
    },
    {
      "epoch": 0.526853744657127,
      "grad_norm": 0.5764310359954834,
      "learning_rate": 0.00016804687411999024,
      "loss": 1.0018,
      "step": 2835
    },
    {
      "epoch": 0.5270395837204981,
      "grad_norm": 0.6331760287284851,
      "learning_rate": 0.00016802544049725763,
      "loss": 0.9499,
      "step": 2836
    },
    {
      "epoch": 0.5272254227838692,
      "grad_norm": 0.5340588092803955,
      "learning_rate": 0.0001680040010561621,
      "loss": 0.9208,
      "step": 2837
    },
    {
      "epoch": 0.5274112618472403,
      "grad_norm": 0.6449952721595764,
      "learning_rate": 0.00016798255579853735,
      "loss": 1.0791,
      "step": 2838
    },
    {
      "epoch": 0.5275971009106114,
      "grad_norm": 0.5729663372039795,
      "learning_rate": 0.00016796110472621766,
      "loss": 1.1602,
      "step": 2839
    },
    {
      "epoch": 0.5277829399739825,
      "grad_norm": 0.6265697479248047,
      "learning_rate": 0.00016793964784103782,
      "loss": 0.9851,
      "step": 2840
    },
    {
      "epoch": 0.5279687790373536,
      "grad_norm": 0.5786687731742859,
      "learning_rate": 0.00016791818514483306,
      "loss": 1.2305,
      "step": 2841
    },
    {
      "epoch": 0.5281546181007247,
      "grad_norm": 0.5565997362136841,
      "learning_rate": 0.00016789671663943912,
      "loss": 0.9734,
      "step": 2842
    },
    {
      "epoch": 0.528340457164096,
      "grad_norm": 0.5070616602897644,
      "learning_rate": 0.00016787524232669232,
      "loss": 0.9516,
      "step": 2843
    },
    {
      "epoch": 0.528526296227467,
      "grad_norm": 0.6512296199798584,
      "learning_rate": 0.0001678537622084293,
      "loss": 1.2005,
      "step": 2844
    },
    {
      "epoch": 0.5287121352908382,
      "grad_norm": 0.6731481552124023,
      "learning_rate": 0.00016783227628648734,
      "loss": 1.1391,
      "step": 2845
    },
    {
      "epoch": 0.5288979743542093,
      "grad_norm": 0.5603753328323364,
      "learning_rate": 0.0001678107845627042,
      "loss": 1.1172,
      "step": 2846
    },
    {
      "epoch": 0.5290838134175804,
      "grad_norm": 0.6080132126808167,
      "learning_rate": 0.0001677892870389181,
      "loss": 1.0396,
      "step": 2847
    },
    {
      "epoch": 0.5292696524809515,
      "grad_norm": 0.5874385237693787,
      "learning_rate": 0.00016776778371696774,
      "loss": 1.2197,
      "step": 2848
    },
    {
      "epoch": 0.5294554915443226,
      "grad_norm": 0.615352988243103,
      "learning_rate": 0.00016774627459869242,
      "loss": 1.2805,
      "step": 2849
    },
    {
      "epoch": 0.5296413306076937,
      "grad_norm": 0.6553687453269958,
      "learning_rate": 0.00016772475968593176,
      "loss": 1.0638,
      "step": 2850
    },
    {
      "epoch": 0.5298271696710649,
      "grad_norm": 0.6943587064743042,
      "learning_rate": 0.00016770323898052607,
      "loss": 1.0408,
      "step": 2851
    },
    {
      "epoch": 0.530013008734436,
      "grad_norm": 3.6750288009643555,
      "learning_rate": 0.00016768171248431602,
      "loss": 2.4315,
      "step": 2852
    },
    {
      "epoch": 0.5301988477978071,
      "grad_norm": 0.6327362060546875,
      "learning_rate": 0.00016766018019914283,
      "loss": 1.2868,
      "step": 2853
    },
    {
      "epoch": 0.5303846868611782,
      "grad_norm": 0.5573206543922424,
      "learning_rate": 0.00016763864212684818,
      "loss": 1.0675,
      "step": 2854
    },
    {
      "epoch": 0.5305705259245493,
      "grad_norm": 0.5509533286094666,
      "learning_rate": 0.0001676170982692743,
      "loss": 1.0416,
      "step": 2855
    },
    {
      "epoch": 0.5307563649879204,
      "grad_norm": 0.676953136920929,
      "learning_rate": 0.00016759554862826384,
      "loss": 1.4656,
      "step": 2856
    },
    {
      "epoch": 0.5309422040512916,
      "grad_norm": 0.6586757898330688,
      "learning_rate": 0.00016757399320566,
      "loss": 1.4603,
      "step": 2857
    },
    {
      "epoch": 0.5311280431146627,
      "grad_norm": 0.6420878171920776,
      "learning_rate": 0.00016755243200330648,
      "loss": 1.0874,
      "step": 2858
    },
    {
      "epoch": 0.5313138821780338,
      "grad_norm": 0.5888837575912476,
      "learning_rate": 0.00016753086502304749,
      "loss": 1.055,
      "step": 2859
    },
    {
      "epoch": 0.531499721241405,
      "grad_norm": 0.5788093209266663,
      "learning_rate": 0.00016750929226672762,
      "loss": 1.0413,
      "step": 2860
    },
    {
      "epoch": 0.5316855603047761,
      "grad_norm": 0.5567480325698853,
      "learning_rate": 0.0001674877137361921,
      "loss": 1.0751,
      "step": 2861
    },
    {
      "epoch": 0.5318713993681472,
      "grad_norm": 0.8277568817138672,
      "learning_rate": 0.00016746612943328656,
      "loss": 1.0339,
      "step": 2862
    },
    {
      "epoch": 0.5320572384315183,
      "grad_norm": 0.44506174325942993,
      "learning_rate": 0.00016744453935985714,
      "loss": 0.963,
      "step": 2863
    },
    {
      "epoch": 0.5322430774948894,
      "grad_norm": 0.62749844789505,
      "learning_rate": 0.00016742294351775053,
      "loss": 1.2259,
      "step": 2864
    },
    {
      "epoch": 0.5324289165582605,
      "grad_norm": 0.5781571865081787,
      "learning_rate": 0.00016740134190881383,
      "loss": 1.0872,
      "step": 2865
    },
    {
      "epoch": 0.5326147556216316,
      "grad_norm": 0.5317525863647461,
      "learning_rate": 0.00016737973453489468,
      "loss": 0.9075,
      "step": 2866
    },
    {
      "epoch": 0.5328005946850027,
      "grad_norm": 0.588940441608429,
      "learning_rate": 0.0001673581213978412,
      "loss": 1.0165,
      "step": 2867
    },
    {
      "epoch": 0.532986433748374,
      "grad_norm": 0.57525634765625,
      "learning_rate": 0.00016733650249950202,
      "loss": 1.0519,
      "step": 2868
    },
    {
      "epoch": 0.5331722728117451,
      "grad_norm": 0.6532301902770996,
      "learning_rate": 0.00016731487784172626,
      "loss": 1.1657,
      "step": 2869
    },
    {
      "epoch": 0.5333581118751162,
      "grad_norm": 0.5266667008399963,
      "learning_rate": 0.0001672932474263635,
      "loss": 1.1751,
      "step": 2870
    },
    {
      "epoch": 0.5335439509384873,
      "grad_norm": 0.6372961401939392,
      "learning_rate": 0.00016727161125526388,
      "loss": 0.9298,
      "step": 2871
    },
    {
      "epoch": 0.5337297900018584,
      "grad_norm": 0.5796231031417847,
      "learning_rate": 0.0001672499693302779,
      "loss": 1.0757,
      "step": 2872
    },
    {
      "epoch": 0.5339156290652295,
      "grad_norm": 0.6466894149780273,
      "learning_rate": 0.00016722832165325672,
      "loss": 1.1214,
      "step": 2873
    },
    {
      "epoch": 0.5341014681286006,
      "grad_norm": 0.5859202146530151,
      "learning_rate": 0.00016720666822605192,
      "loss": 1.1376,
      "step": 2874
    },
    {
      "epoch": 0.5342873071919717,
      "grad_norm": 1.9871423244476318,
      "learning_rate": 0.0001671850090505155,
      "loss": 1.9804,
      "step": 2875
    },
    {
      "epoch": 0.5344731462553429,
      "grad_norm": 0.5480605959892273,
      "learning_rate": 0.00016716334412850007,
      "loss": 0.9871,
      "step": 2876
    },
    {
      "epoch": 0.534658985318714,
      "grad_norm": 0.6779521107673645,
      "learning_rate": 0.00016714167346185862,
      "loss": 1.1147,
      "step": 2877
    },
    {
      "epoch": 0.5348448243820851,
      "grad_norm": 0.5643601417541504,
      "learning_rate": 0.00016711999705244475,
      "loss": 1.1982,
      "step": 2878
    },
    {
      "epoch": 0.5350306634454562,
      "grad_norm": 0.662219762802124,
      "learning_rate": 0.0001670983149021125,
      "loss": 0.9183,
      "step": 2879
    },
    {
      "epoch": 0.5352165025088274,
      "grad_norm": 0.6409224271774292,
      "learning_rate": 0.00016707662701271632,
      "loss": 1.1694,
      "step": 2880
    },
    {
      "epoch": 0.5354023415721985,
      "grad_norm": 0.5614042282104492,
      "learning_rate": 0.00016705493338611124,
      "loss": 0.9345,
      "step": 2881
    },
    {
      "epoch": 0.5355881806355696,
      "grad_norm": 0.6522765159606934,
      "learning_rate": 0.00016703323402415286,
      "loss": 0.7476,
      "step": 2882
    },
    {
      "epoch": 0.5357740196989407,
      "grad_norm": 0.6119543313980103,
      "learning_rate": 0.00016701152892869701,
      "loss": 0.9991,
      "step": 2883
    },
    {
      "epoch": 0.5359598587623119,
      "grad_norm": 0.611212968826294,
      "learning_rate": 0.0001669898181016003,
      "loss": 0.9248,
      "step": 2884
    },
    {
      "epoch": 0.536145697825683,
      "grad_norm": 0.5680248141288757,
      "learning_rate": 0.00016696810154471966,
      "loss": 0.9241,
      "step": 2885
    },
    {
      "epoch": 0.5363315368890541,
      "grad_norm": 0.7113536596298218,
      "learning_rate": 0.00016694637925991255,
      "loss": 1.2047,
      "step": 2886
    },
    {
      "epoch": 0.5365173759524252,
      "grad_norm": 0.5832658410072327,
      "learning_rate": 0.00016692465124903695,
      "loss": 1.1214,
      "step": 2887
    },
    {
      "epoch": 0.5367032150157963,
      "grad_norm": 0.6345791816711426,
      "learning_rate": 0.00016690291751395129,
      "loss": 1.2211,
      "step": 2888
    },
    {
      "epoch": 0.5368890540791674,
      "grad_norm": 0.7019723057746887,
      "learning_rate": 0.00016688117805651446,
      "loss": 1.2782,
      "step": 2889
    },
    {
      "epoch": 0.5370748931425385,
      "grad_norm": 0.5996314883232117,
      "learning_rate": 0.00016685943287858595,
      "loss": 1.1402,
      "step": 2890
    },
    {
      "epoch": 0.5372607322059096,
      "grad_norm": 0.6830651164054871,
      "learning_rate": 0.00016683768198202562,
      "loss": 1.2012,
      "step": 2891
    },
    {
      "epoch": 0.5374465712692807,
      "grad_norm": 0.5295338034629822,
      "learning_rate": 0.00016681592536869392,
      "loss": 1.2169,
      "step": 2892
    },
    {
      "epoch": 0.537632410332652,
      "grad_norm": 0.676923394203186,
      "learning_rate": 0.0001667941630404517,
      "loss": 1.3105,
      "step": 2893
    },
    {
      "epoch": 0.5378182493960231,
      "grad_norm": 0.6796035170555115,
      "learning_rate": 0.00016677239499916037,
      "loss": 1.2213,
      "step": 2894
    },
    {
      "epoch": 0.5380040884593942,
      "grad_norm": 0.6050610542297363,
      "learning_rate": 0.0001667506212466818,
      "loss": 0.9687,
      "step": 2895
    },
    {
      "epoch": 0.5381899275227653,
      "grad_norm": 0.5061158537864685,
      "learning_rate": 0.0001667288417848783,
      "loss": 0.9394,
      "step": 2896
    },
    {
      "epoch": 0.5383757665861364,
      "grad_norm": 0.7453888058662415,
      "learning_rate": 0.0001667070566156127,
      "loss": 1.2559,
      "step": 2897
    },
    {
      "epoch": 0.5385616056495075,
      "grad_norm": 0.5495951771736145,
      "learning_rate": 0.00016668526574074846,
      "loss": 1.0272,
      "step": 2898
    },
    {
      "epoch": 0.5387474447128786,
      "grad_norm": 0.5371149182319641,
      "learning_rate": 0.00016666346916214926,
      "loss": 1.1627,
      "step": 2899
    },
    {
      "epoch": 0.5389332837762497,
      "grad_norm": 0.6507683396339417,
      "learning_rate": 0.00016664166688167947,
      "loss": 1.024,
      "step": 2900
    },
    {
      "epoch": 0.5391191228396209,
      "grad_norm": 0.6115374565124512,
      "learning_rate": 0.0001666198589012039,
      "loss": 1.0213,
      "step": 2901
    },
    {
      "epoch": 0.539304961902992,
      "grad_norm": 0.6327989101409912,
      "learning_rate": 0.0001665980452225878,
      "loss": 1.1811,
      "step": 2902
    },
    {
      "epoch": 0.5394908009663631,
      "grad_norm": 0.6254445910453796,
      "learning_rate": 0.00016657622584769697,
      "loss": 1.0541,
      "step": 2903
    },
    {
      "epoch": 0.5396766400297343,
      "grad_norm": 0.6406470537185669,
      "learning_rate": 0.0001665544007783976,
      "loss": 0.9014,
      "step": 2904
    },
    {
      "epoch": 0.5398624790931054,
      "grad_norm": 0.5992531776428223,
      "learning_rate": 0.00016653257001655652,
      "loss": 1.2726,
      "step": 2905
    },
    {
      "epoch": 0.5400483181564765,
      "grad_norm": 0.7481580972671509,
      "learning_rate": 0.00016651073356404098,
      "loss": 1.0581,
      "step": 2906
    },
    {
      "epoch": 0.5402341572198476,
      "grad_norm": 0.5833855271339417,
      "learning_rate": 0.0001664888914227186,
      "loss": 1.216,
      "step": 2907
    },
    {
      "epoch": 0.5404199962832187,
      "grad_norm": 0.5238146185874939,
      "learning_rate": 0.00016646704359445764,
      "loss": 1.0835,
      "step": 2908
    },
    {
      "epoch": 0.5406058353465899,
      "grad_norm": 0.5387495160102844,
      "learning_rate": 0.0001664451900811268,
      "loss": 0.9602,
      "step": 2909
    },
    {
      "epoch": 0.540791674409961,
      "grad_norm": 0.6715323328971863,
      "learning_rate": 0.00016642333088459524,
      "loss": 0.8797,
      "step": 2910
    },
    {
      "epoch": 0.5409775134733321,
      "grad_norm": 0.6089659333229065,
      "learning_rate": 0.0001664014660067326,
      "loss": 0.9706,
      "step": 2911
    },
    {
      "epoch": 0.5411633525367032,
      "grad_norm": 0.5779519081115723,
      "learning_rate": 0.0001663795954494091,
      "loss": 1.1606,
      "step": 2912
    },
    {
      "epoch": 0.5413491916000743,
      "grad_norm": 0.6019100546836853,
      "learning_rate": 0.0001663577192144953,
      "loss": 1.1748,
      "step": 2913
    },
    {
      "epoch": 0.5415350306634454,
      "grad_norm": 0.6508989334106445,
      "learning_rate": 0.0001663358373038624,
      "loss": 1.1944,
      "step": 2914
    },
    {
      "epoch": 0.5417208697268165,
      "grad_norm": 0.630411684513092,
      "learning_rate": 0.00016631394971938193,
      "loss": 1.1656,
      "step": 2915
    },
    {
      "epoch": 0.5419067087901877,
      "grad_norm": 0.7168655395507812,
      "learning_rate": 0.00016629205646292604,
      "loss": 1.2304,
      "step": 2916
    },
    {
      "epoch": 0.5420925478535589,
      "grad_norm": 0.6478018760681152,
      "learning_rate": 0.00016627015753636727,
      "loss": 1.1461,
      "step": 2917
    },
    {
      "epoch": 0.54227838691693,
      "grad_norm": 0.7113860845565796,
      "learning_rate": 0.0001662482529415787,
      "loss": 1.3356,
      "step": 2918
    },
    {
      "epoch": 0.5424642259803011,
      "grad_norm": 0.6111336350440979,
      "learning_rate": 0.0001662263426804339,
      "loss": 1.1173,
      "step": 2919
    },
    {
      "epoch": 0.5426500650436722,
      "grad_norm": 0.6446161866188049,
      "learning_rate": 0.00016620442675480685,
      "loss": 1.0263,
      "step": 2920
    },
    {
      "epoch": 0.5428359041070433,
      "grad_norm": 0.7231898307800293,
      "learning_rate": 0.0001661825051665721,
      "loss": 1.3269,
      "step": 2921
    },
    {
      "epoch": 0.5430217431704144,
      "grad_norm": 4.894625663757324,
      "learning_rate": 0.0001661605779176046,
      "loss": 2.7246,
      "step": 2922
    },
    {
      "epoch": 0.5432075822337855,
      "grad_norm": 0.5375463962554932,
      "learning_rate": 0.00016613864500977996,
      "loss": 1.192,
      "step": 2923
    },
    {
      "epoch": 0.5433934212971566,
      "grad_norm": 0.5629100203514099,
      "learning_rate": 0.000166116706444974,
      "loss": 0.7375,
      "step": 2924
    },
    {
      "epoch": 0.5435792603605278,
      "grad_norm": 0.5796539783477783,
      "learning_rate": 0.00016609476222506327,
      "loss": 0.9775,
      "step": 2925
    },
    {
      "epoch": 0.543765099423899,
      "grad_norm": 0.5823481678962708,
      "learning_rate": 0.00016607281235192468,
      "loss": 1.0259,
      "step": 2926
    },
    {
      "epoch": 0.54395093848727,
      "grad_norm": 0.7050068378448486,
      "learning_rate": 0.00016605085682743566,
      "loss": 1.2885,
      "step": 2927
    },
    {
      "epoch": 0.5441367775506412,
      "grad_norm": 0.5624426007270813,
      "learning_rate": 0.00016602889565347408,
      "loss": 1.0661,
      "step": 2928
    },
    {
      "epoch": 0.5443226166140123,
      "grad_norm": 0.6064251065254211,
      "learning_rate": 0.00016600692883191834,
      "loss": 1.0146,
      "step": 2929
    },
    {
      "epoch": 0.5445084556773834,
      "grad_norm": 0.5572468638420105,
      "learning_rate": 0.00016598495636464735,
      "loss": 0.8867,
      "step": 2930
    },
    {
      "epoch": 0.5446942947407545,
      "grad_norm": 0.7270289063453674,
      "learning_rate": 0.0001659629782535404,
      "loss": 1.1612,
      "step": 2931
    },
    {
      "epoch": 0.5448801338041256,
      "grad_norm": 0.5419548749923706,
      "learning_rate": 0.00016594099450047736,
      "loss": 0.9358,
      "step": 2932
    },
    {
      "epoch": 0.5450659728674967,
      "grad_norm": 0.6731230616569519,
      "learning_rate": 0.00016591900510733854,
      "loss": 0.9637,
      "step": 2933
    },
    {
      "epoch": 0.5452518119308679,
      "grad_norm": 0.6420273184776306,
      "learning_rate": 0.00016589701007600476,
      "loss": 1.2075,
      "step": 2934
    },
    {
      "epoch": 0.545437650994239,
      "grad_norm": 0.48401394486427307,
      "learning_rate": 0.00016587500940835727,
      "loss": 0.8759,
      "step": 2935
    },
    {
      "epoch": 0.5456234900576101,
      "grad_norm": 0.6045053601264954,
      "learning_rate": 0.00016585300310627786,
      "loss": 1.1209,
      "step": 2936
    },
    {
      "epoch": 0.5458093291209812,
      "grad_norm": 0.5479567050933838,
      "learning_rate": 0.00016583099117164877,
      "loss": 1.0256,
      "step": 2937
    },
    {
      "epoch": 0.5459951681843523,
      "grad_norm": 0.5547533631324768,
      "learning_rate": 0.00016580897360635272,
      "loss": 0.9392,
      "step": 2938
    },
    {
      "epoch": 0.5461810072477234,
      "grad_norm": 0.7662529349327087,
      "learning_rate": 0.00016578695041227293,
      "loss": 1.0937,
      "step": 2939
    },
    {
      "epoch": 0.5463668463110946,
      "grad_norm": 0.5871464014053345,
      "learning_rate": 0.0001657649215912931,
      "loss": 0.9771,
      "step": 2940
    },
    {
      "epoch": 0.5465526853744657,
      "grad_norm": 0.5434662699699402,
      "learning_rate": 0.00016574288714529735,
      "loss": 0.9684,
      "step": 2941
    },
    {
      "epoch": 0.5467385244378369,
      "grad_norm": 0.577323853969574,
      "learning_rate": 0.00016572084707617042,
      "loss": 1.1616,
      "step": 2942
    },
    {
      "epoch": 0.546924363501208,
      "grad_norm": 0.7086620330810547,
      "learning_rate": 0.00016569880138579736,
      "loss": 1.2126,
      "step": 2943
    },
    {
      "epoch": 0.5471102025645791,
      "grad_norm": 0.6588655710220337,
      "learning_rate": 0.00016567675007606383,
      "loss": 1.1606,
      "step": 2944
    },
    {
      "epoch": 0.5472960416279502,
      "grad_norm": 0.6107214093208313,
      "learning_rate": 0.00016565469314885594,
      "loss": 0.9897,
      "step": 2945
    },
    {
      "epoch": 0.5474818806913213,
      "grad_norm": 0.6048929691314697,
      "learning_rate": 0.0001656326306060602,
      "loss": 0.946,
      "step": 2946
    },
    {
      "epoch": 0.5476677197546924,
      "grad_norm": 0.6101149916648865,
      "learning_rate": 0.00016561056244956377,
      "loss": 1.0612,
      "step": 2947
    },
    {
      "epoch": 0.5478535588180635,
      "grad_norm": 0.6024320125579834,
      "learning_rate": 0.00016558848868125411,
      "loss": 1.0689,
      "step": 2948
    },
    {
      "epoch": 0.5480393978814346,
      "grad_norm": 0.7112870216369629,
      "learning_rate": 0.00016556640930301926,
      "loss": 0.9647,
      "step": 2949
    },
    {
      "epoch": 0.5482252369448058,
      "grad_norm": 0.6747824549674988,
      "learning_rate": 0.0001655443243167477,
      "loss": 1.0305,
      "step": 2950
    },
    {
      "epoch": 0.548411076008177,
      "grad_norm": 0.5641652941703796,
      "learning_rate": 0.00016552223372432844,
      "loss": 1.1324,
      "step": 2951
    },
    {
      "epoch": 0.5485969150715481,
      "grad_norm": 0.6192693114280701,
      "learning_rate": 0.0001655001375276509,
      "loss": 1.1574,
      "step": 2952
    },
    {
      "epoch": 0.5487827541349192,
      "grad_norm": 0.607811689376831,
      "learning_rate": 0.00016547803572860504,
      "loss": 1.0986,
      "step": 2953
    },
    {
      "epoch": 0.5489685931982903,
      "grad_norm": 0.627953827381134,
      "learning_rate": 0.00016545592832908126,
      "loss": 0.9785,
      "step": 2954
    },
    {
      "epoch": 0.5491544322616614,
      "grad_norm": 0.5804354548454285,
      "learning_rate": 0.00016543381533097048,
      "loss": 1.2391,
      "step": 2955
    },
    {
      "epoch": 0.5493402713250325,
      "grad_norm": 0.7040599584579468,
      "learning_rate": 0.00016541169673616406,
      "loss": 0.8315,
      "step": 2956
    },
    {
      "epoch": 0.5495261103884036,
      "grad_norm": 0.9636831879615784,
      "learning_rate": 0.00016538957254655384,
      "loss": 1.0614,
      "step": 2957
    },
    {
      "epoch": 0.5497119494517748,
      "grad_norm": 0.679150402545929,
      "learning_rate": 0.00016536744276403212,
      "loss": 1.0342,
      "step": 2958
    },
    {
      "epoch": 0.5498977885151459,
      "grad_norm": 0.5816363096237183,
      "learning_rate": 0.00016534530739049176,
      "loss": 1.2106,
      "step": 2959
    },
    {
      "epoch": 0.550083627578517,
      "grad_norm": 0.8184868693351746,
      "learning_rate": 0.00016532316642782605,
      "loss": 1.291,
      "step": 2960
    },
    {
      "epoch": 0.5502694666418881,
      "grad_norm": 0.565895676612854,
      "learning_rate": 0.0001653010198779287,
      "loss": 1.245,
      "step": 2961
    },
    {
      "epoch": 0.5504553057052592,
      "grad_norm": 0.6959781050682068,
      "learning_rate": 0.000165278867742694,
      "loss": 1.3576,
      "step": 2962
    },
    {
      "epoch": 0.5506411447686304,
      "grad_norm": 0.7933211922645569,
      "learning_rate": 0.00016525671002401668,
      "loss": 1.2011,
      "step": 2963
    },
    {
      "epoch": 0.5508269838320015,
      "grad_norm": 0.5321523547172546,
      "learning_rate": 0.00016523454672379191,
      "loss": 1.0696,
      "step": 2964
    },
    {
      "epoch": 0.5510128228953726,
      "grad_norm": 0.5358680486679077,
      "learning_rate": 0.00016521237784391534,
      "loss": 1.07,
      "step": 2965
    },
    {
      "epoch": 0.5511986619587437,
      "grad_norm": 0.5598793029785156,
      "learning_rate": 0.00016519020338628317,
      "loss": 1.2435,
      "step": 2966
    },
    {
      "epoch": 0.5513845010221149,
      "grad_norm": 0.5243983268737793,
      "learning_rate": 0.000165168023352792,
      "loss": 1.1711,
      "step": 2967
    },
    {
      "epoch": 0.551570340085486,
      "grad_norm": 0.5723912715911865,
      "learning_rate": 0.00016514583774533893,
      "loss": 1.0636,
      "step": 2968
    },
    {
      "epoch": 0.5517561791488571,
      "grad_norm": 2.9442546367645264,
      "learning_rate": 0.0001651236465658216,
      "loss": 2.5894,
      "step": 2969
    },
    {
      "epoch": 0.5519420182122282,
      "grad_norm": 0.7351272702217102,
      "learning_rate": 0.000165101449816138,
      "loss": 0.9057,
      "step": 2970
    },
    {
      "epoch": 0.5521278572755993,
      "grad_norm": 0.60108882188797,
      "learning_rate": 0.0001650792474981867,
      "loss": 0.9787,
      "step": 2971
    },
    {
      "epoch": 0.5523136963389704,
      "grad_norm": 0.6729848980903625,
      "learning_rate": 0.0001650570396138667,
      "loss": 1.1786,
      "step": 2972
    },
    {
      "epoch": 0.5524995354023415,
      "grad_norm": 0.6495280265808105,
      "learning_rate": 0.0001650348261650775,
      "loss": 1.1397,
      "step": 2973
    },
    {
      "epoch": 0.5526853744657126,
      "grad_norm": 0.7092133164405823,
      "learning_rate": 0.00016501260715371906,
      "loss": 1.346,
      "step": 2974
    },
    {
      "epoch": 0.5528712135290839,
      "grad_norm": 0.5815352201461792,
      "learning_rate": 0.0001649903825816918,
      "loss": 1.1502,
      "step": 2975
    },
    {
      "epoch": 0.553057052592455,
      "grad_norm": 0.49748045206069946,
      "learning_rate": 0.00016496815245089672,
      "loss": 1.0467,
      "step": 2976
    },
    {
      "epoch": 0.5532428916558261,
      "grad_norm": 0.6422550082206726,
      "learning_rate": 0.0001649459167632351,
      "loss": 1.1578,
      "step": 2977
    },
    {
      "epoch": 0.5534287307191972,
      "grad_norm": 0.6007086634635925,
      "learning_rate": 0.00016492367552060886,
      "loss": 1.0099,
      "step": 2978
    },
    {
      "epoch": 0.5536145697825683,
      "grad_norm": 0.5576412081718445,
      "learning_rate": 0.00016490142872492032,
      "loss": 1.0978,
      "step": 2979
    },
    {
      "epoch": 0.5538004088459394,
      "grad_norm": 0.8643796443939209,
      "learning_rate": 0.00016487917637807232,
      "loss": 1.209,
      "step": 2980
    },
    {
      "epoch": 0.5539862479093105,
      "grad_norm": 0.5304890871047974,
      "learning_rate": 0.00016485691848196812,
      "loss": 1.0879,
      "step": 2981
    },
    {
      "epoch": 0.5541720869726816,
      "grad_norm": 0.6655973196029663,
      "learning_rate": 0.00016483465503851154,
      "loss": 1.0499,
      "step": 2982
    },
    {
      "epoch": 0.5543579260360528,
      "grad_norm": 0.6618862748146057,
      "learning_rate": 0.00016481238604960677,
      "loss": 1.0499,
      "step": 2983
    },
    {
      "epoch": 0.5545437650994239,
      "grad_norm": 0.6254931092262268,
      "learning_rate": 0.00016479011151715857,
      "loss": 1.2534,
      "step": 2984
    },
    {
      "epoch": 0.554729604162795,
      "grad_norm": 0.5860945582389832,
      "learning_rate": 0.0001647678314430721,
      "loss": 1.1679,
      "step": 2985
    },
    {
      "epoch": 0.5549154432261661,
      "grad_norm": 0.6401891112327576,
      "learning_rate": 0.000164745545829253,
      "loss": 1.1576,
      "step": 2986
    },
    {
      "epoch": 0.5551012822895373,
      "grad_norm": 0.5567334890365601,
      "learning_rate": 0.00016472325467760743,
      "loss": 1.0597,
      "step": 2987
    },
    {
      "epoch": 0.5552871213529084,
      "grad_norm": 0.5591250061988831,
      "learning_rate": 0.000164700957990042,
      "loss": 0.9265,
      "step": 2988
    },
    {
      "epoch": 0.5554729604162795,
      "grad_norm": 0.6554935574531555,
      "learning_rate": 0.0001646786557684638,
      "loss": 0.8308,
      "step": 2989
    },
    {
      "epoch": 0.5556587994796506,
      "grad_norm": 0.7079557180404663,
      "learning_rate": 0.00016465634801478044,
      "loss": 1.1058,
      "step": 2990
    },
    {
      "epoch": 0.5558446385430218,
      "grad_norm": 0.613620936870575,
      "learning_rate": 0.0001646340347308998,
      "loss": 1.1594,
      "step": 2991
    },
    {
      "epoch": 0.5560304776063929,
      "grad_norm": 0.5244887471199036,
      "learning_rate": 0.00016461171591873055,
      "loss": 0.7287,
      "step": 2992
    },
    {
      "epoch": 0.556216316669764,
      "grad_norm": 0.832699179649353,
      "learning_rate": 0.00016458939158018156,
      "loss": 1.0544,
      "step": 2993
    },
    {
      "epoch": 0.5564021557331351,
      "grad_norm": 0.546475350856781,
      "learning_rate": 0.00016456706171716232,
      "loss": 0.9461,
      "step": 2994
    },
    {
      "epoch": 0.5565879947965062,
      "grad_norm": 0.6768338084220886,
      "learning_rate": 0.00016454472633158274,
      "loss": 1.0346,
      "step": 2995
    },
    {
      "epoch": 0.5567738338598773,
      "grad_norm": 0.7314834594726562,
      "learning_rate": 0.0001645223854253532,
      "loss": 1.1819,
      "step": 2996
    },
    {
      "epoch": 0.5569596729232484,
      "grad_norm": 0.600952684879303,
      "learning_rate": 0.00016450003900038461,
      "loss": 1.2066,
      "step": 2997
    },
    {
      "epoch": 0.5571455119866195,
      "grad_norm": 0.4913482367992401,
      "learning_rate": 0.00016447768705858828,
      "loss": 0.8089,
      "step": 2998
    },
    {
      "epoch": 0.5573313510499907,
      "grad_norm": 0.6177974343299866,
      "learning_rate": 0.00016445532960187602,
      "loss": 1.179,
      "step": 2999
    },
    {
      "epoch": 0.5575171901133619,
      "grad_norm": 0.5990849137306213,
      "learning_rate": 0.0001644329666321601,
      "loss": 1.0434,
      "step": 3000
    },
    {
      "epoch": 0.557703029176733,
      "grad_norm": 0.685162365436554,
      "learning_rate": 0.0001644105981513533,
      "loss": 0.5839,
      "step": 3001
    },
    {
      "epoch": 0.5578888682401041,
      "grad_norm": 0.5808719992637634,
      "learning_rate": 0.00016438822416136881,
      "loss": 1.2187,
      "step": 3002
    },
    {
      "epoch": 0.5580747073034752,
      "grad_norm": 0.5311502814292908,
      "learning_rate": 0.00016436584466412036,
      "loss": 1.1093,
      "step": 3003
    },
    {
      "epoch": 0.5582605463668463,
      "grad_norm": 0.525296688079834,
      "learning_rate": 0.0001643434596615221,
      "loss": 1.0029,
      "step": 3004
    },
    {
      "epoch": 0.5584463854302174,
      "grad_norm": 0.7411049604415894,
      "learning_rate": 0.0001643210691554887,
      "loss": 1.2051,
      "step": 3005
    },
    {
      "epoch": 0.5586322244935885,
      "grad_norm": 0.6270830631256104,
      "learning_rate": 0.00016429867314793518,
      "loss": 0.9569,
      "step": 3006
    },
    {
      "epoch": 0.5588180635569596,
      "grad_norm": 0.6153439283370972,
      "learning_rate": 0.00016427627164077724,
      "loss": 1.1596,
      "step": 3007
    },
    {
      "epoch": 0.5590039026203308,
      "grad_norm": 0.5672575831413269,
      "learning_rate": 0.00016425386463593083,
      "loss": 0.969,
      "step": 3008
    },
    {
      "epoch": 0.559189741683702,
      "grad_norm": 0.673835813999176,
      "learning_rate": 0.00016423145213531251,
      "loss": 1.428,
      "step": 3009
    },
    {
      "epoch": 0.559375580747073,
      "grad_norm": 0.5811061859130859,
      "learning_rate": 0.00016420903414083926,
      "loss": 1.0928,
      "step": 3010
    },
    {
      "epoch": 0.5595614198104442,
      "grad_norm": 0.5861353874206543,
      "learning_rate": 0.00016418661065442858,
      "loss": 1.1264,
      "step": 3011
    },
    {
      "epoch": 0.5597472588738153,
      "grad_norm": 0.6847012042999268,
      "learning_rate": 0.0001641641816779983,
      "loss": 1.3918,
      "step": 3012
    },
    {
      "epoch": 0.5599330979371864,
      "grad_norm": 0.6058685779571533,
      "learning_rate": 0.00016414174721346696,
      "loss": 1.066,
      "step": 3013
    },
    {
      "epoch": 0.5601189370005575,
      "grad_norm": 0.6190561056137085,
      "learning_rate": 0.00016411930726275333,
      "loss": 1.1328,
      "step": 3014
    },
    {
      "epoch": 0.5603047760639286,
      "grad_norm": 0.6666110754013062,
      "learning_rate": 0.00016409686182777673,
      "loss": 1.258,
      "step": 3015
    },
    {
      "epoch": 0.5604906151272998,
      "grad_norm": 0.59378981590271,
      "learning_rate": 0.00016407441091045706,
      "loss": 1.2295,
      "step": 3016
    },
    {
      "epoch": 0.5606764541906709,
      "grad_norm": 0.6505338549613953,
      "learning_rate": 0.0001640519545127145,
      "loss": 1.1902,
      "step": 3017
    },
    {
      "epoch": 0.560862293254042,
      "grad_norm": 0.5768793225288391,
      "learning_rate": 0.00016402949263646984,
      "loss": 0.9078,
      "step": 3018
    },
    {
      "epoch": 0.5610481323174131,
      "grad_norm": 0.5398329496383667,
      "learning_rate": 0.00016400702528364432,
      "loss": 1.4141,
      "step": 3019
    },
    {
      "epoch": 0.5612339713807842,
      "grad_norm": 0.6005418300628662,
      "learning_rate": 0.0001639845524561596,
      "loss": 1.0095,
      "step": 3020
    },
    {
      "epoch": 0.5614198104441553,
      "grad_norm": 0.5389949679374695,
      "learning_rate": 0.00016396207415593777,
      "loss": 0.9768,
      "step": 3021
    },
    {
      "epoch": 0.5616056495075264,
      "grad_norm": 0.7075584530830383,
      "learning_rate": 0.0001639395903849015,
      "loss": 0.999,
      "step": 3022
    },
    {
      "epoch": 0.5617914885708976,
      "grad_norm": 0.5680679082870483,
      "learning_rate": 0.0001639171011449739,
      "loss": 1.2847,
      "step": 3023
    },
    {
      "epoch": 0.5619773276342688,
      "grad_norm": 0.8404898643493652,
      "learning_rate": 0.00016389460643807845,
      "loss": 1.3546,
      "step": 3024
    },
    {
      "epoch": 0.5621631666976399,
      "grad_norm": 0.5268566012382507,
      "learning_rate": 0.0001638721062661393,
      "loss": 1.1228,
      "step": 3025
    },
    {
      "epoch": 0.562349005761011,
      "grad_norm": 0.591393232345581,
      "learning_rate": 0.0001638496006310808,
      "loss": 0.9728,
      "step": 3026
    },
    {
      "epoch": 0.5625348448243821,
      "grad_norm": 0.5140171051025391,
      "learning_rate": 0.00016382708953482795,
      "loss": 1.0738,
      "step": 3027
    },
    {
      "epoch": 0.5627206838877532,
      "grad_norm": 0.5632948279380798,
      "learning_rate": 0.00016380457297930623,
      "loss": 1.1832,
      "step": 3028
    },
    {
      "epoch": 0.5629065229511243,
      "grad_norm": 0.6482004523277283,
      "learning_rate": 0.00016378205096644145,
      "loss": 1.2979,
      "step": 3029
    },
    {
      "epoch": 0.5630923620144954,
      "grad_norm": 0.6070230007171631,
      "learning_rate": 0.00016375952349816,
      "loss": 1.1065,
      "step": 3030
    },
    {
      "epoch": 0.5632782010778665,
      "grad_norm": 0.5631871223449707,
      "learning_rate": 0.0001637369905763887,
      "loss": 1.1694,
      "step": 3031
    },
    {
      "epoch": 0.5634640401412377,
      "grad_norm": 0.6535444259643555,
      "learning_rate": 0.00016371445220305487,
      "loss": 1.1037,
      "step": 3032
    },
    {
      "epoch": 0.5636498792046089,
      "grad_norm": 0.5825177431106567,
      "learning_rate": 0.0001636919083800862,
      "loss": 0.8396,
      "step": 3033
    },
    {
      "epoch": 0.56383571826798,
      "grad_norm": 0.49941158294677734,
      "learning_rate": 0.00016366935910941098,
      "loss": 0.8458,
      "step": 3034
    },
    {
      "epoch": 0.5640215573313511,
      "grad_norm": 0.5689782500267029,
      "learning_rate": 0.00016364680439295785,
      "loss": 1.1708,
      "step": 3035
    },
    {
      "epoch": 0.5642073963947222,
      "grad_norm": 8.309523582458496,
      "learning_rate": 0.00016362424423265598,
      "loss": 5.9499,
      "step": 3036
    },
    {
      "epoch": 0.5643932354580933,
      "grad_norm": 0.7303856015205383,
      "learning_rate": 0.000163601678630435,
      "loss": 0.9228,
      "step": 3037
    },
    {
      "epoch": 0.5645790745214644,
      "grad_norm": 0.6042539477348328,
      "learning_rate": 0.00016357910758822502,
      "loss": 1.0577,
      "step": 3038
    },
    {
      "epoch": 0.5647649135848355,
      "grad_norm": 0.7997288703918457,
      "learning_rate": 0.0001635565311079565,
      "loss": 1.2253,
      "step": 3039
    },
    {
      "epoch": 0.5649507526482066,
      "grad_norm": 0.554591715335846,
      "learning_rate": 0.00016353394919156052,
      "loss": 1.1309,
      "step": 3040
    },
    {
      "epoch": 0.5651365917115778,
      "grad_norm": 0.6072766184806824,
      "learning_rate": 0.0001635113618409686,
      "loss": 1.0735,
      "step": 3041
    },
    {
      "epoch": 0.5653224307749489,
      "grad_norm": 0.693913996219635,
      "learning_rate": 0.00016348876905811257,
      "loss": 0.9651,
      "step": 3042
    },
    {
      "epoch": 0.56550826983832,
      "grad_norm": 0.5827034115791321,
      "learning_rate": 0.00016346617084492493,
      "loss": 0.9317,
      "step": 3043
    },
    {
      "epoch": 0.5656941089016911,
      "grad_norm": 0.6475703716278076,
      "learning_rate": 0.00016344356720333854,
      "loss": 1.1251,
      "step": 3044
    },
    {
      "epoch": 0.5658799479650622,
      "grad_norm": 0.5898675322532654,
      "learning_rate": 0.00016342095813528672,
      "loss": 1.0938,
      "step": 3045
    },
    {
      "epoch": 0.5660657870284334,
      "grad_norm": 0.5563342571258545,
      "learning_rate": 0.00016339834364270325,
      "loss": 1.0096,
      "step": 3046
    },
    {
      "epoch": 0.5662516260918045,
      "grad_norm": 0.6287338733673096,
      "learning_rate": 0.0001633757237275225,
      "loss": 1.2837,
      "step": 3047
    },
    {
      "epoch": 0.5664374651551756,
      "grad_norm": 0.71528160572052,
      "learning_rate": 0.00016335309839167906,
      "loss": 1.1279,
      "step": 3048
    },
    {
      "epoch": 0.5666233042185468,
      "grad_norm": 0.6881284117698669,
      "learning_rate": 0.00016333046763710823,
      "loss": 1.1145,
      "step": 3049
    },
    {
      "epoch": 0.5668091432819179,
      "grad_norm": 0.5819098353385925,
      "learning_rate": 0.00016330783146574558,
      "loss": 1.2411,
      "step": 3050
    },
    {
      "epoch": 0.566994982345289,
      "grad_norm": 0.5303094983100891,
      "learning_rate": 0.0001632851898795273,
      "loss": 1.1144,
      "step": 3051
    },
    {
      "epoch": 0.5671808214086601,
      "grad_norm": 0.7490119934082031,
      "learning_rate": 0.00016326254288038998,
      "loss": 1.1791,
      "step": 3052
    },
    {
      "epoch": 0.5673666604720312,
      "grad_norm": 0.5890512466430664,
      "learning_rate": 0.00016323989047027063,
      "loss": 0.752,
      "step": 3053
    },
    {
      "epoch": 0.5675524995354023,
      "grad_norm": 0.5626150965690613,
      "learning_rate": 0.00016321723265110675,
      "loss": 1.1475,
      "step": 3054
    },
    {
      "epoch": 0.5677383385987734,
      "grad_norm": 0.6347929239273071,
      "learning_rate": 0.00016319456942483633,
      "loss": 1.4247,
      "step": 3055
    },
    {
      "epoch": 0.5679241776621445,
      "grad_norm": 0.6369238495826721,
      "learning_rate": 0.00016317190079339784,
      "loss": 1.1994,
      "step": 3056
    },
    {
      "epoch": 0.5681100167255158,
      "grad_norm": 0.5677321553230286,
      "learning_rate": 0.0001631492267587301,
      "loss": 1.1179,
      "step": 3057
    },
    {
      "epoch": 0.5682958557888869,
      "grad_norm": 0.6081541776657104,
      "learning_rate": 0.00016312654732277254,
      "loss": 1.1653,
      "step": 3058
    },
    {
      "epoch": 0.568481694852258,
      "grad_norm": 0.5795355439186096,
      "learning_rate": 0.0001631038624874649,
      "loss": 0.839,
      "step": 3059
    },
    {
      "epoch": 0.5686675339156291,
      "grad_norm": 0.7947715520858765,
      "learning_rate": 0.00016308117225474758,
      "loss": 1.2332,
      "step": 3060
    },
    {
      "epoch": 0.5688533729790002,
      "grad_norm": 0.5681910514831543,
      "learning_rate": 0.00016305847662656127,
      "loss": 1.0693,
      "step": 3061
    },
    {
      "epoch": 0.5690392120423713,
      "grad_norm": 0.6374632716178894,
      "learning_rate": 0.00016303577560484713,
      "loss": 1.0293,
      "step": 3062
    },
    {
      "epoch": 0.5692250511057424,
      "grad_norm": 0.5624377131462097,
      "learning_rate": 0.00016301306919154685,
      "loss": 1.0867,
      "step": 3063
    },
    {
      "epoch": 0.5694108901691135,
      "grad_norm": 0.5568638443946838,
      "learning_rate": 0.0001629903573886026,
      "loss": 1.2122,
      "step": 3064
    },
    {
      "epoch": 0.5695967292324847,
      "grad_norm": 0.7387270927429199,
      "learning_rate": 0.00016296764019795693,
      "loss": 1.1513,
      "step": 3065
    },
    {
      "epoch": 0.5697825682958558,
      "grad_norm": 0.6199186444282532,
      "learning_rate": 0.0001629449176215529,
      "loss": 1.0889,
      "step": 3066
    },
    {
      "epoch": 0.5699684073592269,
      "grad_norm": 0.5959807634353638,
      "learning_rate": 0.00016292218966133405,
      "loss": 0.9602,
      "step": 3067
    },
    {
      "epoch": 0.570154246422598,
      "grad_norm": 0.555349588394165,
      "learning_rate": 0.0001628994563192443,
      "loss": 1.3097,
      "step": 3068
    },
    {
      "epoch": 0.5703400854859692,
      "grad_norm": 3.8160369396209717,
      "learning_rate": 0.00016287671759722814,
      "loss": 2.6136,
      "step": 3069
    },
    {
      "epoch": 0.5705259245493403,
      "grad_norm": 0.6192765831947327,
      "learning_rate": 0.00016285397349723039,
      "loss": 1.293,
      "step": 3070
    },
    {
      "epoch": 0.5707117636127114,
      "grad_norm": 0.5929650068283081,
      "learning_rate": 0.00016283122402119645,
      "loss": 0.9145,
      "step": 3071
    },
    {
      "epoch": 0.5708976026760825,
      "grad_norm": 0.6493182182312012,
      "learning_rate": 0.00016280846917107215,
      "loss": 1.1481,
      "step": 3072
    },
    {
      "epoch": 0.5710834417394536,
      "grad_norm": 0.5538601875305176,
      "learning_rate": 0.00016278570894880372,
      "loss": 1.0587,
      "step": 3073
    },
    {
      "epoch": 0.5712692808028248,
      "grad_norm": 0.5534785985946655,
      "learning_rate": 0.0001627629433563379,
      "loss": 0.985,
      "step": 3074
    },
    {
      "epoch": 0.5714551198661959,
      "grad_norm": 0.6992794275283813,
      "learning_rate": 0.0001627401723956219,
      "loss": 1.1699,
      "step": 3075
    },
    {
      "epoch": 0.571640958929567,
      "grad_norm": 0.545943558216095,
      "learning_rate": 0.0001627173960686033,
      "loss": 1.0472,
      "step": 3076
    },
    {
      "epoch": 0.5718267979929381,
      "grad_norm": 0.5662815570831299,
      "learning_rate": 0.00016269461437723034,
      "loss": 0.7866,
      "step": 3077
    },
    {
      "epoch": 0.5720126370563092,
      "grad_norm": 0.6630524396896362,
      "learning_rate": 0.00016267182732345145,
      "loss": 1.1837,
      "step": 3078
    },
    {
      "epoch": 0.5721984761196803,
      "grad_norm": 0.527268648147583,
      "learning_rate": 0.00016264903490921574,
      "loss": 1.0633,
      "step": 3079
    },
    {
      "epoch": 0.5723843151830514,
      "grad_norm": 0.6216735243797302,
      "learning_rate": 0.0001626262371364727,
      "loss": 1.0545,
      "step": 3080
    },
    {
      "epoch": 0.5725701542464225,
      "grad_norm": 0.5457433462142944,
      "learning_rate": 0.00016260343400717223,
      "loss": 1.121,
      "step": 3081
    },
    {
      "epoch": 0.5727559933097938,
      "grad_norm": 0.6372676491737366,
      "learning_rate": 0.00016258062552326473,
      "loss": 1.1908,
      "step": 3082
    },
    {
      "epoch": 0.5729418323731649,
      "grad_norm": 0.5404830574989319,
      "learning_rate": 0.00016255781168670108,
      "loss": 0.9754,
      "step": 3083
    },
    {
      "epoch": 0.573127671436536,
      "grad_norm": 0.6268422603607178,
      "learning_rate": 0.00016253499249943261,
      "loss": 1.1317,
      "step": 3084
    },
    {
      "epoch": 0.5733135104999071,
      "grad_norm": 0.5977557301521301,
      "learning_rate": 0.00016251216796341105,
      "loss": 1.3657,
      "step": 3085
    },
    {
      "epoch": 0.5734993495632782,
      "grad_norm": 0.5620158314704895,
      "learning_rate": 0.00016248933808058865,
      "loss": 1.1449,
      "step": 3086
    },
    {
      "epoch": 0.5736851886266493,
      "grad_norm": 0.6281202435493469,
      "learning_rate": 0.00016246650285291813,
      "loss": 1.044,
      "step": 3087
    },
    {
      "epoch": 0.5738710276900204,
      "grad_norm": 0.5784666538238525,
      "learning_rate": 0.00016244366228235262,
      "loss": 1.0241,
      "step": 3088
    },
    {
      "epoch": 0.5740568667533915,
      "grad_norm": 0.5703860521316528,
      "learning_rate": 0.0001624208163708457,
      "loss": 1.0912,
      "step": 3089
    },
    {
      "epoch": 0.5742427058167627,
      "grad_norm": 0.6101926565170288,
      "learning_rate": 0.00016239796512035145,
      "loss": 1.2968,
      "step": 3090
    },
    {
      "epoch": 0.5744285448801338,
      "grad_norm": 0.5332390666007996,
      "learning_rate": 0.0001623751085328244,
      "loss": 0.9362,
      "step": 3091
    },
    {
      "epoch": 0.574614383943505,
      "grad_norm": 2.209155321121216,
      "learning_rate": 0.00016235224661021946,
      "loss": 2.555,
      "step": 3092
    },
    {
      "epoch": 0.574800223006876,
      "grad_norm": 0.5431200265884399,
      "learning_rate": 0.00016232937935449214,
      "loss": 0.9771,
      "step": 3093
    },
    {
      "epoch": 0.5749860620702472,
      "grad_norm": 0.7131924629211426,
      "learning_rate": 0.0001623065067675983,
      "loss": 0.9736,
      "step": 3094
    },
    {
      "epoch": 0.5751719011336183,
      "grad_norm": 0.5774680376052856,
      "learning_rate": 0.00016228362885149425,
      "loss": 0.8636,
      "step": 3095
    },
    {
      "epoch": 0.5753577401969894,
      "grad_norm": 0.5699325203895569,
      "learning_rate": 0.00016226074560813682,
      "loss": 1.2025,
      "step": 3096
    },
    {
      "epoch": 0.5755435792603605,
      "grad_norm": 0.5468953847885132,
      "learning_rate": 0.00016223785703948327,
      "loss": 1.0436,
      "step": 3097
    },
    {
      "epoch": 0.5757294183237317,
      "grad_norm": 0.5742730498313904,
      "learning_rate": 0.0001622149631474913,
      "loss": 1.25,
      "step": 3098
    },
    {
      "epoch": 0.5759152573871028,
      "grad_norm": 0.8017017841339111,
      "learning_rate": 0.00016219206393411903,
      "loss": 1.2828,
      "step": 3099
    },
    {
      "epoch": 0.5761010964504739,
      "grad_norm": 0.6201735138893127,
      "learning_rate": 0.00016216915940132513,
      "loss": 1.2306,
      "step": 3100
    },
    {
      "epoch": 0.576286935513845,
      "grad_norm": 0.5407164096832275,
      "learning_rate": 0.00016214624955106867,
      "loss": 1.1117,
      "step": 3101
    },
    {
      "epoch": 0.5764727745772161,
      "grad_norm": 0.6462402939796448,
      "learning_rate": 0.00016212333438530918,
      "loss": 1.3912,
      "step": 3102
    },
    {
      "epoch": 0.5766586136405872,
      "grad_norm": 0.6825852990150452,
      "learning_rate": 0.00016210041390600662,
      "loss": 1.0587,
      "step": 3103
    },
    {
      "epoch": 0.5768444527039583,
      "grad_norm": 0.5548142790794373,
      "learning_rate": 0.00016207748811512146,
      "loss": 0.8915,
      "step": 3104
    },
    {
      "epoch": 0.5770302917673295,
      "grad_norm": 0.599079966545105,
      "learning_rate": 0.00016205455701461454,
      "loss": 0.8589,
      "step": 3105
    },
    {
      "epoch": 0.5772161308307006,
      "grad_norm": 0.5806437134742737,
      "learning_rate": 0.00016203162060644726,
      "loss": 1.1802,
      "step": 3106
    },
    {
      "epoch": 0.5774019698940718,
      "grad_norm": 0.7707300186157227,
      "learning_rate": 0.00016200867889258142,
      "loss": 1.0741,
      "step": 3107
    },
    {
      "epoch": 0.5775878089574429,
      "grad_norm": 0.5511063933372498,
      "learning_rate": 0.0001619857318749792,
      "loss": 1.0173,
      "step": 3108
    },
    {
      "epoch": 0.577773648020814,
      "grad_norm": 0.5827605128288269,
      "learning_rate": 0.00016196277955560345,
      "loss": 1.0117,
      "step": 3109
    },
    {
      "epoch": 0.5779594870841851,
      "grad_norm": 0.5755023956298828,
      "learning_rate": 0.00016193982193641717,
      "loss": 1.0535,
      "step": 3110
    },
    {
      "epoch": 0.5781453261475562,
      "grad_norm": 0.5576872229576111,
      "learning_rate": 0.00016191685901938407,
      "loss": 0.9355,
      "step": 3111
    },
    {
      "epoch": 0.5783311652109273,
      "grad_norm": 0.6232908964157104,
      "learning_rate": 0.00016189389080646823,
      "loss": 1.0984,
      "step": 3112
    },
    {
      "epoch": 0.5785170042742984,
      "grad_norm": 0.558789074420929,
      "learning_rate": 0.00016187091729963409,
      "loss": 1.1206,
      "step": 3113
    },
    {
      "epoch": 0.5787028433376695,
      "grad_norm": 0.5674159526824951,
      "learning_rate": 0.0001618479385008467,
      "loss": 1.0263,
      "step": 3114
    },
    {
      "epoch": 0.5788886824010407,
      "grad_norm": 0.6567408442497253,
      "learning_rate": 0.00016182495441207148,
      "loss": 1.2807,
      "step": 3115
    },
    {
      "epoch": 0.5790745214644119,
      "grad_norm": 0.6544413566589355,
      "learning_rate": 0.00016180196503527426,
      "loss": 1.0089,
      "step": 3116
    },
    {
      "epoch": 0.579260360527783,
      "grad_norm": 0.6177968382835388,
      "learning_rate": 0.0001617789703724214,
      "loss": 1.2623,
      "step": 3117
    },
    {
      "epoch": 0.5794461995911541,
      "grad_norm": 0.6059399843215942,
      "learning_rate": 0.0001617559704254797,
      "loss": 1.1049,
      "step": 3118
    },
    {
      "epoch": 0.5796320386545252,
      "grad_norm": 0.5528789162635803,
      "learning_rate": 0.00016173296519641635,
      "loss": 1.039,
      "step": 3119
    },
    {
      "epoch": 0.5798178777178963,
      "grad_norm": 0.5349646210670471,
      "learning_rate": 0.00016170995468719904,
      "loss": 0.9782,
      "step": 3120
    },
    {
      "epoch": 0.5800037167812674,
      "grad_norm": 0.5830336809158325,
      "learning_rate": 0.00016168693889979601,
      "loss": 1.3013,
      "step": 3121
    },
    {
      "epoch": 0.5801895558446385,
      "grad_norm": 0.5634336471557617,
      "learning_rate": 0.00016166391783617573,
      "loss": 1.022,
      "step": 3122
    },
    {
      "epoch": 0.5803753949080097,
      "grad_norm": 0.57351154088974,
      "learning_rate": 0.00016164089149830728,
      "loss": 1.0899,
      "step": 3123
    },
    {
      "epoch": 0.5805612339713808,
      "grad_norm": 0.5886953473091125,
      "learning_rate": 0.00016161785988816018,
      "loss": 1.2963,
      "step": 3124
    },
    {
      "epoch": 0.5807470730347519,
      "grad_norm": 0.6066156029701233,
      "learning_rate": 0.00016159482300770432,
      "loss": 1.1767,
      "step": 3125
    },
    {
      "epoch": 0.580932912098123,
      "grad_norm": 0.6183861494064331,
      "learning_rate": 0.00016157178085891014,
      "loss": 1.1532,
      "step": 3126
    },
    {
      "epoch": 0.5811187511614941,
      "grad_norm": 0.5010476112365723,
      "learning_rate": 0.00016154873344374846,
      "loss": 1.0039,
      "step": 3127
    },
    {
      "epoch": 0.5813045902248652,
      "grad_norm": 0.5397800803184509,
      "learning_rate": 0.00016152568076419058,
      "loss": 0.9084,
      "step": 3128
    },
    {
      "epoch": 0.5814904292882364,
      "grad_norm": 0.5681976675987244,
      "learning_rate": 0.00016150262282220828,
      "loss": 0.9105,
      "step": 3129
    },
    {
      "epoch": 0.5816762683516075,
      "grad_norm": 0.686458170413971,
      "learning_rate": 0.0001614795596197737,
      "loss": 1.3427,
      "step": 3130
    },
    {
      "epoch": 0.5818621074149787,
      "grad_norm": 0.6114237904548645,
      "learning_rate": 0.00016145649115885952,
      "loss": 1.0371,
      "step": 3131
    },
    {
      "epoch": 0.5820479464783498,
      "grad_norm": 0.6143605709075928,
      "learning_rate": 0.00016143341744143884,
      "loss": 0.8177,
      "step": 3132
    },
    {
      "epoch": 0.5822337855417209,
      "grad_norm": 0.569019079208374,
      "learning_rate": 0.00016141033846948517,
      "loss": 0.9501,
      "step": 3133
    },
    {
      "epoch": 0.582419624605092,
      "grad_norm": 0.5694284439086914,
      "learning_rate": 0.00016138725424497252,
      "loss": 1.0626,
      "step": 3134
    },
    {
      "epoch": 0.5826054636684631,
      "grad_norm": 0.5643896460533142,
      "learning_rate": 0.00016136416476987534,
      "loss": 1.111,
      "step": 3135
    },
    {
      "epoch": 0.5827913027318342,
      "grad_norm": 0.7149565815925598,
      "learning_rate": 0.00016134107004616853,
      "loss": 1.3123,
      "step": 3136
    },
    {
      "epoch": 0.5829771417952053,
      "grad_norm": 0.6421211361885071,
      "learning_rate": 0.00016131797007582743,
      "loss": 0.9872,
      "step": 3137
    },
    {
      "epoch": 0.5831629808585764,
      "grad_norm": 0.5691509246826172,
      "learning_rate": 0.00016129486486082782,
      "loss": 1.2416,
      "step": 3138
    },
    {
      "epoch": 0.5833488199219476,
      "grad_norm": 0.6846590042114258,
      "learning_rate": 0.00016127175440314596,
      "loss": 1.0807,
      "step": 3139
    },
    {
      "epoch": 0.5835346589853188,
      "grad_norm": 0.6126776933670044,
      "learning_rate": 0.00016124863870475847,
      "loss": 1.0882,
      "step": 3140
    },
    {
      "epoch": 0.5837204980486899,
      "grad_norm": 0.8353591561317444,
      "learning_rate": 0.0001612255177676426,
      "loss": 1.3001,
      "step": 3141
    },
    {
      "epoch": 0.583906337112061,
      "grad_norm": 0.6794141530990601,
      "learning_rate": 0.00016120239159377582,
      "loss": 1.1654,
      "step": 3142
    },
    {
      "epoch": 0.5840921761754321,
      "grad_norm": 0.6769196391105652,
      "learning_rate": 0.00016117926018513622,
      "loss": 1.2275,
      "step": 3143
    },
    {
      "epoch": 0.5842780152388032,
      "grad_norm": 0.6092231273651123,
      "learning_rate": 0.0001611561235437023,
      "loss": 0.9933,
      "step": 3144
    },
    {
      "epoch": 0.5844638543021743,
      "grad_norm": 0.5706498622894287,
      "learning_rate": 0.00016113298167145292,
      "loss": 1.0099,
      "step": 3145
    },
    {
      "epoch": 0.5846496933655454,
      "grad_norm": 0.5918547511100769,
      "learning_rate": 0.00016110983457036753,
      "loss": 1.1331,
      "step": 3146
    },
    {
      "epoch": 0.5848355324289165,
      "grad_norm": 0.6145870089530945,
      "learning_rate": 0.00016108668224242587,
      "loss": 0.9591,
      "step": 3147
    },
    {
      "epoch": 0.5850213714922877,
      "grad_norm": 0.5824666023254395,
      "learning_rate": 0.00016106352468960825,
      "loss": 1.2682,
      "step": 3148
    },
    {
      "epoch": 0.5852072105556588,
      "grad_norm": 0.6346530318260193,
      "learning_rate": 0.0001610403619138954,
      "loss": 1.0673,
      "step": 3149
    },
    {
      "epoch": 0.5853930496190299,
      "grad_norm": 0.5495458841323853,
      "learning_rate": 0.00016101719391726853,
      "loss": 1.0197,
      "step": 3150
    },
    {
      "epoch": 0.585578888682401,
      "grad_norm": 0.6226973533630371,
      "learning_rate": 0.00016099402070170912,
      "loss": 1.1676,
      "step": 3151
    },
    {
      "epoch": 0.5857647277457722,
      "grad_norm": 0.687059760093689,
      "learning_rate": 0.00016097084226919932,
      "loss": 0.8824,
      "step": 3152
    },
    {
      "epoch": 0.5859505668091433,
      "grad_norm": 0.5375981330871582,
      "learning_rate": 0.0001609476586217216,
      "loss": 1.0266,
      "step": 3153
    },
    {
      "epoch": 0.5861364058725144,
      "grad_norm": 0.5993423461914062,
      "learning_rate": 0.00016092446976125892,
      "loss": 0.9489,
      "step": 3154
    },
    {
      "epoch": 0.5863222449358855,
      "grad_norm": 0.670810878276825,
      "learning_rate": 0.00016090127568979464,
      "loss": 1.406,
      "step": 3155
    },
    {
      "epoch": 0.5865080839992567,
      "grad_norm": 0.5641027688980103,
      "learning_rate": 0.00016087807640931265,
      "loss": 1.0512,
      "step": 3156
    },
    {
      "epoch": 0.5866939230626278,
      "grad_norm": 0.6967179775238037,
      "learning_rate": 0.0001608548719217972,
      "loss": 1.0222,
      "step": 3157
    },
    {
      "epoch": 0.5868797621259989,
      "grad_norm": 0.9438024163246155,
      "learning_rate": 0.00016083166222923304,
      "loss": 1.0211,
      "step": 3158
    },
    {
      "epoch": 0.58706560118937,
      "grad_norm": 0.595064640045166,
      "learning_rate": 0.00016080844733360535,
      "loss": 0.9823,
      "step": 3159
    },
    {
      "epoch": 0.5872514402527411,
      "grad_norm": 0.5441033244132996,
      "learning_rate": 0.0001607852272368997,
      "loss": 1.1615,
      "step": 3160
    },
    {
      "epoch": 0.5874372793161122,
      "grad_norm": 0.5815873146057129,
      "learning_rate": 0.0001607620019411022,
      "loss": 0.9964,
      "step": 3161
    },
    {
      "epoch": 0.5876231183794833,
      "grad_norm": 0.5449917912483215,
      "learning_rate": 0.00016073877144819933,
      "loss": 1.1755,
      "step": 3162
    },
    {
      "epoch": 0.5878089574428544,
      "grad_norm": 0.6668300628662109,
      "learning_rate": 0.0001607155357601781,
      "loss": 1.2527,
      "step": 3163
    },
    {
      "epoch": 0.5879947965062257,
      "grad_norm": 0.5341411828994751,
      "learning_rate": 0.00016069229487902586,
      "loss": 1.0948,
      "step": 3164
    },
    {
      "epoch": 0.5881806355695968,
      "grad_norm": 0.6133508086204529,
      "learning_rate": 0.00016066904880673047,
      "loss": 1.0737,
      "step": 3165
    },
    {
      "epoch": 0.5883664746329679,
      "grad_norm": 0.5782324075698853,
      "learning_rate": 0.00016064579754528023,
      "loss": 0.9634,
      "step": 3166
    },
    {
      "epoch": 0.588552313696339,
      "grad_norm": 0.6144787073135376,
      "learning_rate": 0.0001606225410966638,
      "loss": 0.9183,
      "step": 3167
    },
    {
      "epoch": 0.5887381527597101,
      "grad_norm": 0.6035318970680237,
      "learning_rate": 0.00016059927946287045,
      "loss": 1.1995,
      "step": 3168
    },
    {
      "epoch": 0.5889239918230812,
      "grad_norm": 0.822420597076416,
      "learning_rate": 0.00016057601264588974,
      "loss": 1.1864,
      "step": 3169
    },
    {
      "epoch": 0.5891098308864523,
      "grad_norm": 0.5931529402732849,
      "learning_rate": 0.00016055274064771176,
      "loss": 0.8851,
      "step": 3170
    },
    {
      "epoch": 0.5892956699498234,
      "grad_norm": 0.6320171356201172,
      "learning_rate": 0.00016052946347032697,
      "loss": 0.8221,
      "step": 3171
    },
    {
      "epoch": 0.5894815090131946,
      "grad_norm": 0.5823320746421814,
      "learning_rate": 0.0001605061811157264,
      "loss": 1.2303,
      "step": 3172
    },
    {
      "epoch": 0.5896673480765657,
      "grad_norm": 0.6609289050102234,
      "learning_rate": 0.00016048289358590133,
      "loss": 0.8762,
      "step": 3173
    },
    {
      "epoch": 0.5898531871399368,
      "grad_norm": 0.5508584976196289,
      "learning_rate": 0.0001604596008828437,
      "loss": 1.3369,
      "step": 3174
    },
    {
      "epoch": 0.590039026203308,
      "grad_norm": 0.5863260626792908,
      "learning_rate": 0.0001604363030085457,
      "loss": 0.9728,
      "step": 3175
    },
    {
      "epoch": 0.590224865266679,
      "grad_norm": 0.5487418174743652,
      "learning_rate": 0.0001604129999650001,
      "loss": 1.2077,
      "step": 3176
    },
    {
      "epoch": 0.5904107043300502,
      "grad_norm": 0.6902508735656738,
      "learning_rate": 0.00016038969175420005,
      "loss": 1.0947,
      "step": 3177
    },
    {
      "epoch": 0.5905965433934213,
      "grad_norm": 0.5682591199874878,
      "learning_rate": 0.00016036637837813915,
      "loss": 1.2425,
      "step": 3178
    },
    {
      "epoch": 0.5907823824567924,
      "grad_norm": 0.7730516791343689,
      "learning_rate": 0.00016034305983881142,
      "loss": 1.1954,
      "step": 3179
    },
    {
      "epoch": 0.5909682215201635,
      "grad_norm": 0.6822097897529602,
      "learning_rate": 0.0001603197361382114,
      "loss": 1.175,
      "step": 3180
    },
    {
      "epoch": 0.5911540605835347,
      "grad_norm": 0.5383156538009644,
      "learning_rate": 0.0001602964072783339,
      "loss": 0.9459,
      "step": 3181
    },
    {
      "epoch": 0.5913398996469058,
      "grad_norm": 0.6112632155418396,
      "learning_rate": 0.00016027307326117445,
      "loss": 1.1796,
      "step": 3182
    },
    {
      "epoch": 0.5915257387102769,
      "grad_norm": 0.592560350894928,
      "learning_rate": 0.00016024973408872872,
      "loss": 0.9495,
      "step": 3183
    },
    {
      "epoch": 0.591711577773648,
      "grad_norm": 0.6019397377967834,
      "learning_rate": 0.00016022638976299303,
      "loss": 1.0865,
      "step": 3184
    },
    {
      "epoch": 0.5918974168370191,
      "grad_norm": 0.6122562885284424,
      "learning_rate": 0.00016020304028596408,
      "loss": 1.373,
      "step": 3185
    },
    {
      "epoch": 0.5920832559003902,
      "grad_norm": 0.5218804478645325,
      "learning_rate": 0.00016017968565963897,
      "loss": 0.78,
      "step": 3186
    },
    {
      "epoch": 0.5922690949637613,
      "grad_norm": 0.6451453566551208,
      "learning_rate": 0.00016015632588601528,
      "loss": 1.087,
      "step": 3187
    },
    {
      "epoch": 0.5924549340271325,
      "grad_norm": 0.5636645555496216,
      "learning_rate": 0.000160132960967091,
      "loss": 1.0118,
      "step": 3188
    },
    {
      "epoch": 0.5926407730905037,
      "grad_norm": 0.5245563983917236,
      "learning_rate": 0.00016010959090486464,
      "loss": 1.0635,
      "step": 3189
    },
    {
      "epoch": 0.5928266121538748,
      "grad_norm": 0.5599002838134766,
      "learning_rate": 0.000160086215701335,
      "loss": 1.0792,
      "step": 3190
    },
    {
      "epoch": 0.5930124512172459,
      "grad_norm": 0.5101682543754578,
      "learning_rate": 0.00016006283535850152,
      "loss": 0.9534,
      "step": 3191
    },
    {
      "epoch": 0.593198290280617,
      "grad_norm": 0.6640803813934326,
      "learning_rate": 0.00016003944987836392,
      "loss": 1.0931,
      "step": 3192
    },
    {
      "epoch": 0.5933841293439881,
      "grad_norm": 0.5303892493247986,
      "learning_rate": 0.0001600160592629224,
      "loss": 0.9207,
      "step": 3193
    },
    {
      "epoch": 0.5935699684073592,
      "grad_norm": 0.6699718236923218,
      "learning_rate": 0.0001599926635141776,
      "loss": 1.1941,
      "step": 3194
    },
    {
      "epoch": 0.5937558074707303,
      "grad_norm": 0.5338372588157654,
      "learning_rate": 0.00015996926263413066,
      "loss": 1.188,
      "step": 3195
    },
    {
      "epoch": 0.5939416465341014,
      "grad_norm": 0.5052550435066223,
      "learning_rate": 0.00015994585662478306,
      "loss": 0.9666,
      "step": 3196
    },
    {
      "epoch": 0.5941274855974726,
      "grad_norm": 0.6414618492126465,
      "learning_rate": 0.00015992244548813678,
      "loss": 1.191,
      "step": 3197
    },
    {
      "epoch": 0.5943133246608437,
      "grad_norm": 0.5840882658958435,
      "learning_rate": 0.00015989902922619421,
      "loss": 1.0446,
      "step": 3198
    },
    {
      "epoch": 0.5944991637242149,
      "grad_norm": 0.5422523617744446,
      "learning_rate": 0.00015987560784095823,
      "loss": 1.0008,
      "step": 3199
    },
    {
      "epoch": 0.594685002787586,
      "grad_norm": 0.619379460811615,
      "learning_rate": 0.0001598521813344321,
      "loss": 1.2946,
      "step": 3200
    },
    {
      "epoch": 0.5948708418509571,
      "grad_norm": 0.5059217810630798,
      "learning_rate": 0.00015982874970861955,
      "loss": 0.9146,
      "step": 3201
    },
    {
      "epoch": 0.5950566809143282,
      "grad_norm": 0.6319354176521301,
      "learning_rate": 0.00015980531296552476,
      "loss": 1.1363,
      "step": 3202
    },
    {
      "epoch": 0.5952425199776993,
      "grad_norm": 0.6196834444999695,
      "learning_rate": 0.00015978187110715224,
      "loss": 1.1737,
      "step": 3203
    },
    {
      "epoch": 0.5954283590410704,
      "grad_norm": 0.5323222279548645,
      "learning_rate": 0.0001597584241355071,
      "loss": 1.0392,
      "step": 3204
    },
    {
      "epoch": 0.5956141981044416,
      "grad_norm": 0.6693430542945862,
      "learning_rate": 0.00015973497205259477,
      "loss": 1.1555,
      "step": 3205
    },
    {
      "epoch": 0.5958000371678127,
      "grad_norm": 0.5676935911178589,
      "learning_rate": 0.0001597115148604212,
      "loss": 1.31,
      "step": 3206
    },
    {
      "epoch": 0.5959858762311838,
      "grad_norm": 0.5511360764503479,
      "learning_rate": 0.00015968805256099273,
      "loss": 1.1491,
      "step": 3207
    },
    {
      "epoch": 0.5961717152945549,
      "grad_norm": 0.5314706563949585,
      "learning_rate": 0.0001596645851563161,
      "loss": 1.0161,
      "step": 3208
    },
    {
      "epoch": 0.596357554357926,
      "grad_norm": 0.5497950911521912,
      "learning_rate": 0.00015964111264839853,
      "loss": 1.1909,
      "step": 3209
    },
    {
      "epoch": 0.5965433934212971,
      "grad_norm": 0.564395546913147,
      "learning_rate": 0.00015961763503924772,
      "loss": 1.0012,
      "step": 3210
    },
    {
      "epoch": 0.5967292324846682,
      "grad_norm": 0.5982910394668579,
      "learning_rate": 0.0001595941523308717,
      "loss": 1.0582,
      "step": 3211
    },
    {
      "epoch": 0.5969150715480394,
      "grad_norm": 0.5818560123443604,
      "learning_rate": 0.00015957066452527906,
      "loss": 0.9864,
      "step": 3212
    },
    {
      "epoch": 0.5971009106114106,
      "grad_norm": 0.4957444369792938,
      "learning_rate": 0.00015954717162447874,
      "loss": 0.9415,
      "step": 3213
    },
    {
      "epoch": 0.5972867496747817,
      "grad_norm": 0.6664856672286987,
      "learning_rate": 0.00015952367363048016,
      "loss": 1.2168,
      "step": 3214
    },
    {
      "epoch": 0.5974725887381528,
      "grad_norm": 0.5591182708740234,
      "learning_rate": 0.00015950017054529306,
      "loss": 1.3039,
      "step": 3215
    },
    {
      "epoch": 0.5976584278015239,
      "grad_norm": 0.6910569667816162,
      "learning_rate": 0.00015947666237092783,
      "loss": 1.228,
      "step": 3216
    },
    {
      "epoch": 0.597844266864895,
      "grad_norm": 0.533061683177948,
      "learning_rate": 0.00015945314910939507,
      "loss": 1.1549,
      "step": 3217
    },
    {
      "epoch": 0.5980301059282661,
      "grad_norm": 0.6971861720085144,
      "learning_rate": 0.00015942963076270603,
      "loss": 1.0395,
      "step": 3218
    },
    {
      "epoch": 0.5982159449916372,
      "grad_norm": 0.648344874382019,
      "learning_rate": 0.00015940610733287218,
      "loss": 1.4975,
      "step": 3219
    },
    {
      "epoch": 0.5984017840550083,
      "grad_norm": 0.5853761434555054,
      "learning_rate": 0.0001593825788219056,
      "loss": 1.0498,
      "step": 3220
    },
    {
      "epoch": 0.5985876231183794,
      "grad_norm": 0.5414695143699646,
      "learning_rate": 0.0001593590452318187,
      "loss": 1.1884,
      "step": 3221
    },
    {
      "epoch": 0.5987734621817506,
      "grad_norm": 0.6624699234962463,
      "learning_rate": 0.00015933550656462437,
      "loss": 1.0484,
      "step": 3222
    },
    {
      "epoch": 0.5989593012451218,
      "grad_norm": 0.6307032704353333,
      "learning_rate": 0.00015931196282233594,
      "loss": 1.0314,
      "step": 3223
    },
    {
      "epoch": 0.5991451403084929,
      "grad_norm": 0.5564342141151428,
      "learning_rate": 0.00015928841400696713,
      "loss": 0.9162,
      "step": 3224
    },
    {
      "epoch": 0.599330979371864,
      "grad_norm": 0.6190139055252075,
      "learning_rate": 0.0001592648601205321,
      "loss": 0.9379,
      "step": 3225
    },
    {
      "epoch": 0.5995168184352351,
      "grad_norm": 0.5595014691352844,
      "learning_rate": 0.00015924130116504553,
      "loss": 1.1747,
      "step": 3226
    },
    {
      "epoch": 0.5997026574986062,
      "grad_norm": 0.6404126882553101,
      "learning_rate": 0.00015921773714252244,
      "loss": 1.0451,
      "step": 3227
    },
    {
      "epoch": 0.5998884965619773,
      "grad_norm": 0.5678627490997314,
      "learning_rate": 0.00015919416805497825,
      "loss": 0.9317,
      "step": 3228
    },
    {
      "epoch": 0.6000743356253484,
      "grad_norm": 0.5714607238769531,
      "learning_rate": 0.000159170593904429,
      "loss": 1.0092,
      "step": 3229
    },
    {
      "epoch": 0.6002601746887196,
      "grad_norm": 0.5801486372947693,
      "learning_rate": 0.00015914701469289095,
      "loss": 1.2109,
      "step": 3230
    },
    {
      "epoch": 0.6004460137520907,
      "grad_norm": 0.5355969071388245,
      "learning_rate": 0.00015912343042238087,
      "loss": 1.0206,
      "step": 3231
    },
    {
      "epoch": 0.6006318528154618,
      "grad_norm": 0.5958425402641296,
      "learning_rate": 0.00015909984109491605,
      "loss": 0.9693,
      "step": 3232
    },
    {
      "epoch": 0.6008176918788329,
      "grad_norm": 0.644477367401123,
      "learning_rate": 0.00015907624671251407,
      "loss": 1.1284,
      "step": 3233
    },
    {
      "epoch": 0.601003530942204,
      "grad_norm": 0.7528777718544006,
      "learning_rate": 0.000159052647277193,
      "loss": 1.0145,
      "step": 3234
    },
    {
      "epoch": 0.6011893700055752,
      "grad_norm": 0.6112080812454224,
      "learning_rate": 0.0001590290427909714,
      "loss": 0.9781,
      "step": 3235
    },
    {
      "epoch": 0.6013752090689463,
      "grad_norm": 0.6840320229530334,
      "learning_rate": 0.0001590054332558682,
      "loss": 1.2317,
      "step": 3236
    },
    {
      "epoch": 0.6015610481323174,
      "grad_norm": 0.6607563495635986,
      "learning_rate": 0.00015898181867390277,
      "loss": 1.0902,
      "step": 3237
    },
    {
      "epoch": 0.6017468871956886,
      "grad_norm": 0.7982823848724365,
      "learning_rate": 0.0001589581990470949,
      "loss": 1.4083,
      "step": 3238
    },
    {
      "epoch": 0.6019327262590597,
      "grad_norm": 0.5766059160232544,
      "learning_rate": 0.00015893457437746484,
      "loss": 1.1384,
      "step": 3239
    },
    {
      "epoch": 0.6021185653224308,
      "grad_norm": 0.6066285967826843,
      "learning_rate": 0.00015891094466703325,
      "loss": 1.0282,
      "step": 3240
    },
    {
      "epoch": 0.6023044043858019,
      "grad_norm": 0.6731379628181458,
      "learning_rate": 0.00015888730991782125,
      "loss": 1.0858,
      "step": 3241
    },
    {
      "epoch": 0.602490243449173,
      "grad_norm": 0.5809962749481201,
      "learning_rate": 0.00015886367013185035,
      "loss": 1.2507,
      "step": 3242
    },
    {
      "epoch": 0.6026760825125441,
      "grad_norm": 0.5508452653884888,
      "learning_rate": 0.00015884002531114255,
      "loss": 0.8706,
      "step": 3243
    },
    {
      "epoch": 0.6028619215759152,
      "grad_norm": 0.5738691091537476,
      "learning_rate": 0.0001588163754577202,
      "loss": 0.9869,
      "step": 3244
    },
    {
      "epoch": 0.6030477606392863,
      "grad_norm": 0.6129263639450073,
      "learning_rate": 0.00015879272057360612,
      "loss": 1.0251,
      "step": 3245
    },
    {
      "epoch": 0.6032335997026576,
      "grad_norm": 0.5459702014923096,
      "learning_rate": 0.0001587690606608236,
      "loss": 1.304,
      "step": 3246
    },
    {
      "epoch": 0.6034194387660287,
      "grad_norm": 0.6796538233757019,
      "learning_rate": 0.0001587453957213963,
      "loss": 0.9258,
      "step": 3247
    },
    {
      "epoch": 0.6036052778293998,
      "grad_norm": 0.6355002522468567,
      "learning_rate": 0.00015872172575734833,
      "loss": 0.7917,
      "step": 3248
    },
    {
      "epoch": 0.6037911168927709,
      "grad_norm": 0.782307505607605,
      "learning_rate": 0.00015869805077070427,
      "loss": 1.2612,
      "step": 3249
    },
    {
      "epoch": 0.603976955956142,
      "grad_norm": 0.48565757274627686,
      "learning_rate": 0.00015867437076348906,
      "loss": 0.878,
      "step": 3250
    },
    {
      "epoch": 0.6041627950195131,
      "grad_norm": 0.6093220114707947,
      "learning_rate": 0.00015865068573772808,
      "loss": 1.12,
      "step": 3251
    },
    {
      "epoch": 0.6043486340828842,
      "grad_norm": 0.5323510766029358,
      "learning_rate": 0.0001586269956954472,
      "loss": 0.9756,
      "step": 3252
    },
    {
      "epoch": 0.6045344731462553,
      "grad_norm": 0.6668819785118103,
      "learning_rate": 0.00015860330063867267,
      "loss": 0.8406,
      "step": 3253
    },
    {
      "epoch": 0.6047203122096264,
      "grad_norm": 0.6621644496917725,
      "learning_rate": 0.0001585796005694312,
      "loss": 1.0424,
      "step": 3254
    },
    {
      "epoch": 0.6049061512729976,
      "grad_norm": 0.5795639157295227,
      "learning_rate": 0.00015855589548974982,
      "loss": 1.0071,
      "step": 3255
    },
    {
      "epoch": 0.6050919903363687,
      "grad_norm": 0.6052048802375793,
      "learning_rate": 0.00015853218540165623,
      "loss": 0.9134,
      "step": 3256
    },
    {
      "epoch": 0.6052778293997398,
      "grad_norm": 0.6029704809188843,
      "learning_rate": 0.0001585084703071783,
      "loss": 1.1407,
      "step": 3257
    },
    {
      "epoch": 0.605463668463111,
      "grad_norm": 0.5548236966133118,
      "learning_rate": 0.0001584847502083444,
      "loss": 1.0075,
      "step": 3258
    },
    {
      "epoch": 0.6056495075264821,
      "grad_norm": 0.679028332233429,
      "learning_rate": 0.0001584610251071834,
      "loss": 1.3603,
      "step": 3259
    },
    {
      "epoch": 0.6058353465898532,
      "grad_norm": 0.7591668963432312,
      "learning_rate": 0.00015843729500572464,
      "loss": 1.1176,
      "step": 3260
    },
    {
      "epoch": 0.6060211856532243,
      "grad_norm": 0.5672114491462708,
      "learning_rate": 0.00015841355990599772,
      "loss": 1.2445,
      "step": 3261
    },
    {
      "epoch": 0.6062070247165954,
      "grad_norm": 0.5262810587882996,
      "learning_rate": 0.00015838981981003273,
      "loss": 1.2205,
      "step": 3262
    },
    {
      "epoch": 0.6063928637799666,
      "grad_norm": 0.5443841814994812,
      "learning_rate": 0.00015836607471986033,
      "loss": 0.7721,
      "step": 3263
    },
    {
      "epoch": 0.6065787028433377,
      "grad_norm": 0.568134605884552,
      "learning_rate": 0.00015834232463751135,
      "loss": 1.2441,
      "step": 3264
    },
    {
      "epoch": 0.6067645419067088,
      "grad_norm": 0.5961521863937378,
      "learning_rate": 0.00015831856956501726,
      "loss": 0.9117,
      "step": 3265
    },
    {
      "epoch": 0.6069503809700799,
      "grad_norm": 0.6283470988273621,
      "learning_rate": 0.0001582948095044099,
      "loss": 0.9078,
      "step": 3266
    },
    {
      "epoch": 0.607136220033451,
      "grad_norm": 0.6529102921485901,
      "learning_rate": 0.00015827104445772144,
      "loss": 1.2285,
      "step": 3267
    },
    {
      "epoch": 0.6073220590968221,
      "grad_norm": 0.55194491147995,
      "learning_rate": 0.00015824727442698467,
      "loss": 0.6904,
      "step": 3268
    },
    {
      "epoch": 0.6075078981601932,
      "grad_norm": 0.6459614634513855,
      "learning_rate": 0.0001582234994142326,
      "loss": 1.027,
      "step": 3269
    },
    {
      "epoch": 0.6076937372235643,
      "grad_norm": 0.491305410861969,
      "learning_rate": 0.00015819971942149875,
      "loss": 0.7555,
      "step": 3270
    },
    {
      "epoch": 0.6078795762869356,
      "grad_norm": 0.5508794188499451,
      "learning_rate": 0.00015817593445081716,
      "loss": 1.0445,
      "step": 3271
    },
    {
      "epoch": 0.6080654153503067,
      "grad_norm": 0.4993383288383484,
      "learning_rate": 0.00015815214450422217,
      "loss": 0.9549,
      "step": 3272
    },
    {
      "epoch": 0.6082512544136778,
      "grad_norm": 0.6566349864006042,
      "learning_rate": 0.00015812834958374853,
      "loss": 1.1194,
      "step": 3273
    },
    {
      "epoch": 0.6084370934770489,
      "grad_norm": 0.6416013836860657,
      "learning_rate": 0.00015810454969143156,
      "loss": 1.0529,
      "step": 3274
    },
    {
      "epoch": 0.60862293254042,
      "grad_norm": 0.6909444332122803,
      "learning_rate": 0.00015808074482930685,
      "loss": 0.8588,
      "step": 3275
    },
    {
      "epoch": 0.6088087716037911,
      "grad_norm": 0.6137370467185974,
      "learning_rate": 0.00015805693499941056,
      "loss": 1.1165,
      "step": 3276
    },
    {
      "epoch": 0.6089946106671622,
      "grad_norm": 0.6334401369094849,
      "learning_rate": 0.00015803312020377908,
      "loss": 1.3178,
      "step": 3277
    },
    {
      "epoch": 0.6091804497305333,
      "grad_norm": 0.5722105503082275,
      "learning_rate": 0.00015800930044444943,
      "loss": 1.1094,
      "step": 3278
    },
    {
      "epoch": 0.6093662887939045,
      "grad_norm": 0.651226282119751,
      "learning_rate": 0.00015798547572345892,
      "loss": 0.8808,
      "step": 3279
    },
    {
      "epoch": 0.6095521278572756,
      "grad_norm": 0.6072734594345093,
      "learning_rate": 0.00015796164604284533,
      "loss": 1.0669,
      "step": 3280
    },
    {
      "epoch": 0.6097379669206467,
      "grad_norm": 0.5889689326286316,
      "learning_rate": 0.00015793781140464695,
      "loss": 1.2274,
      "step": 3281
    },
    {
      "epoch": 0.6099238059840179,
      "grad_norm": 0.573512852191925,
      "learning_rate": 0.0001579139718109023,
      "loss": 1.1719,
      "step": 3282
    },
    {
      "epoch": 0.610109645047389,
      "grad_norm": 0.55435711145401,
      "learning_rate": 0.00015789012726365046,
      "loss": 1.1348,
      "step": 3283
    },
    {
      "epoch": 0.6102954841107601,
      "grad_norm": 0.7388445734977722,
      "learning_rate": 0.00015786627776493095,
      "loss": 1.3228,
      "step": 3284
    },
    {
      "epoch": 0.6104813231741312,
      "grad_norm": 0.6711801886558533,
      "learning_rate": 0.0001578424233167836,
      "loss": 0.9831,
      "step": 3285
    },
    {
      "epoch": 0.6106671622375023,
      "grad_norm": 0.5652539730072021,
      "learning_rate": 0.00015781856392124878,
      "loss": 1.3369,
      "step": 3286
    },
    {
      "epoch": 0.6108530013008734,
      "grad_norm": 0.6947058439254761,
      "learning_rate": 0.00015779469958036724,
      "loss": 1.3382,
      "step": 3287
    },
    {
      "epoch": 0.6110388403642446,
      "grad_norm": 0.5150282979011536,
      "learning_rate": 0.00015777083029618015,
      "loss": 1.0123,
      "step": 3288
    },
    {
      "epoch": 0.6112246794276157,
      "grad_norm": 0.603293776512146,
      "learning_rate": 0.00015774695607072902,
      "loss": 1.0484,
      "step": 3289
    },
    {
      "epoch": 0.6114105184909868,
      "grad_norm": 0.5617212057113647,
      "learning_rate": 0.00015772307690605598,
      "loss": 1.0026,
      "step": 3290
    },
    {
      "epoch": 0.6115963575543579,
      "grad_norm": 0.5847578644752502,
      "learning_rate": 0.00015769919280420345,
      "loss": 0.9991,
      "step": 3291
    },
    {
      "epoch": 0.611782196617729,
      "grad_norm": 0.4738200604915619,
      "learning_rate": 0.0001576753037672142,
      "loss": 0.8151,
      "step": 3292
    },
    {
      "epoch": 0.6119680356811001,
      "grad_norm": 0.6319113373756409,
      "learning_rate": 0.00015765140979713158,
      "loss": 1.1612,
      "step": 3293
    },
    {
      "epoch": 0.6121538747444712,
      "grad_norm": 2.3821051120758057,
      "learning_rate": 0.00015762751089599925,
      "loss": 2.1318,
      "step": 3294
    },
    {
      "epoch": 0.6123397138078424,
      "grad_norm": 0.6564528942108154,
      "learning_rate": 0.00015760360706586142,
      "loss": 1.1375,
      "step": 3295
    },
    {
      "epoch": 0.6125255528712136,
      "grad_norm": 0.6328415870666504,
      "learning_rate": 0.00015757969830876254,
      "loss": 1.149,
      "step": 3296
    },
    {
      "epoch": 0.6127113919345847,
      "grad_norm": 0.6952791810035706,
      "learning_rate": 0.00015755578462674764,
      "loss": 1.1789,
      "step": 3297
    },
    {
      "epoch": 0.6128972309979558,
      "grad_norm": 0.6253474354743958,
      "learning_rate": 0.00015753186602186209,
      "loss": 1.1331,
      "step": 3298
    },
    {
      "epoch": 0.6130830700613269,
      "grad_norm": 0.6182775497436523,
      "learning_rate": 0.00015750794249615166,
      "loss": 0.9761,
      "step": 3299
    },
    {
      "epoch": 0.613268909124698,
      "grad_norm": 0.3781263530254364,
      "learning_rate": 0.00015748401405166264,
      "loss": 0.5632,
      "step": 3300
    },
    {
      "epoch": 0.6134547481880691,
      "grad_norm": 0.6519707441329956,
      "learning_rate": 0.00015746008069044166,
      "loss": 1.0675,
      "step": 3301
    },
    {
      "epoch": 0.6136405872514402,
      "grad_norm": 0.5394522547721863,
      "learning_rate": 0.0001574361424145358,
      "loss": 0.8303,
      "step": 3302
    },
    {
      "epoch": 0.6138264263148113,
      "grad_norm": 0.6013384461402893,
      "learning_rate": 0.00015741219922599253,
      "loss": 1.0751,
      "step": 3303
    },
    {
      "epoch": 0.6140122653781825,
      "grad_norm": 2.4101686477661133,
      "learning_rate": 0.00015738825112685976,
      "loss": 2.4912,
      "step": 3304
    },
    {
      "epoch": 0.6141981044415536,
      "grad_norm": 0.5980067253112793,
      "learning_rate": 0.0001573642981191859,
      "loss": 1.1398,
      "step": 3305
    },
    {
      "epoch": 0.6143839435049248,
      "grad_norm": 0.6038574576377869,
      "learning_rate": 0.00015734034020501957,
      "loss": 1.2101,
      "step": 3306
    },
    {
      "epoch": 0.6145697825682959,
      "grad_norm": 0.5979435443878174,
      "learning_rate": 0.00015731637738641005,
      "loss": 1.2162,
      "step": 3307
    },
    {
      "epoch": 0.614755621631667,
      "grad_norm": 0.5230640769004822,
      "learning_rate": 0.0001572924096654069,
      "loss": 1.1457,
      "step": 3308
    },
    {
      "epoch": 0.6149414606950381,
      "grad_norm": 0.5864366888999939,
      "learning_rate": 0.0001572684370440601,
      "loss": 0.9144,
      "step": 3309
    },
    {
      "epoch": 0.6151272997584092,
      "grad_norm": 0.5331762433052063,
      "learning_rate": 0.00015724445952442013,
      "loss": 1.0474,
      "step": 3310
    },
    {
      "epoch": 0.6153131388217803,
      "grad_norm": 0.582480788230896,
      "learning_rate": 0.0001572204771085378,
      "loss": 1.2336,
      "step": 3311
    },
    {
      "epoch": 0.6154989778851515,
      "grad_norm": 0.638606607913971,
      "learning_rate": 0.0001571964897984644,
      "loss": 1.2695,
      "step": 3312
    },
    {
      "epoch": 0.6156848169485226,
      "grad_norm": 0.6707527041435242,
      "learning_rate": 0.00015717249759625163,
      "loss": 1.1273,
      "step": 3313
    },
    {
      "epoch": 0.6158706560118937,
      "grad_norm": 0.6105638742446899,
      "learning_rate": 0.00015714850050395154,
      "loss": 1.0156,
      "step": 3314
    },
    {
      "epoch": 0.6160564950752648,
      "grad_norm": 0.5666206479072571,
      "learning_rate": 0.0001571244985236167,
      "loss": 1.011,
      "step": 3315
    },
    {
      "epoch": 0.6162423341386359,
      "grad_norm": 0.5360713005065918,
      "learning_rate": 0.00015710049165730002,
      "loss": 0.822,
      "step": 3316
    },
    {
      "epoch": 0.616428173202007,
      "grad_norm": 0.5776330828666687,
      "learning_rate": 0.00015707647990705493,
      "loss": 1.1726,
      "step": 3317
    },
    {
      "epoch": 0.6166140122653782,
      "grad_norm": 0.5878480672836304,
      "learning_rate": 0.0001570524632749351,
      "loss": 1.0647,
      "step": 3318
    },
    {
      "epoch": 0.6167998513287493,
      "grad_norm": 0.7072561383247375,
      "learning_rate": 0.0001570284417629948,
      "loss": 1.3886,
      "step": 3319
    },
    {
      "epoch": 0.6169856903921205,
      "grad_norm": 0.6697565913200378,
      "learning_rate": 0.00015700441537328864,
      "loss": 0.8962,
      "step": 3320
    },
    {
      "epoch": 0.6171715294554916,
      "grad_norm": 0.6354570388793945,
      "learning_rate": 0.00015698038410787163,
      "loss": 1.0034,
      "step": 3321
    },
    {
      "epoch": 0.6173573685188627,
      "grad_norm": 0.6120606660842896,
      "learning_rate": 0.00015695634796879915,
      "loss": 1.1362,
      "step": 3322
    },
    {
      "epoch": 0.6175432075822338,
      "grad_norm": 0.7120777368545532,
      "learning_rate": 0.00015693230695812722,
      "loss": 1.0888,
      "step": 3323
    },
    {
      "epoch": 0.6177290466456049,
      "grad_norm": 0.5528488159179688,
      "learning_rate": 0.00015690826107791198,
      "loss": 1.0157,
      "step": 3324
    },
    {
      "epoch": 0.617914885708976,
      "grad_norm": 0.6340495347976685,
      "learning_rate": 0.00015688421033021018,
      "loss": 0.9895,
      "step": 3325
    },
    {
      "epoch": 0.6181007247723471,
      "grad_norm": 0.5119288563728333,
      "learning_rate": 0.00015686015471707892,
      "loss": 1.185,
      "step": 3326
    },
    {
      "epoch": 0.6182865638357182,
      "grad_norm": 0.6284135580062866,
      "learning_rate": 0.00015683609424057578,
      "loss": 0.9463,
      "step": 3327
    },
    {
      "epoch": 0.6184724028990893,
      "grad_norm": 0.7082629203796387,
      "learning_rate": 0.00015681202890275865,
      "loss": 1.1147,
      "step": 3328
    },
    {
      "epoch": 0.6186582419624606,
      "grad_norm": 0.5838453769683838,
      "learning_rate": 0.00015678795870568583,
      "loss": 1.0441,
      "step": 3329
    },
    {
      "epoch": 0.6188440810258317,
      "grad_norm": 0.7504107356071472,
      "learning_rate": 0.00015676388365141623,
      "loss": 1.102,
      "step": 3330
    },
    {
      "epoch": 0.6190299200892028,
      "grad_norm": 0.5921317338943481,
      "learning_rate": 0.00015673980374200896,
      "loss": 0.9912,
      "step": 3331
    },
    {
      "epoch": 0.6192157591525739,
      "grad_norm": 0.6119334697723389,
      "learning_rate": 0.00015671571897952368,
      "loss": 1.0653,
      "step": 3332
    },
    {
      "epoch": 0.619401598215945,
      "grad_norm": 0.5734863877296448,
      "learning_rate": 0.00015669162936602033,
      "loss": 0.9782,
      "step": 3333
    },
    {
      "epoch": 0.6195874372793161,
      "grad_norm": 0.6472792029380798,
      "learning_rate": 0.00015666753490355944,
      "loss": 1.1381,
      "step": 3334
    },
    {
      "epoch": 0.6197732763426872,
      "grad_norm": 0.6566014289855957,
      "learning_rate": 0.00015664343559420178,
      "loss": 1.1101,
      "step": 3335
    },
    {
      "epoch": 0.6199591154060583,
      "grad_norm": 0.5882189273834229,
      "learning_rate": 0.00015661933144000862,
      "loss": 1.2242,
      "step": 3336
    },
    {
      "epoch": 0.6201449544694295,
      "grad_norm": 0.6419895887374878,
      "learning_rate": 0.00015659522244304173,
      "loss": 1.0805,
      "step": 3337
    },
    {
      "epoch": 0.6203307935328006,
      "grad_norm": 0.6279326677322388,
      "learning_rate": 0.00015657110860536314,
      "loss": 1.0568,
      "step": 3338
    },
    {
      "epoch": 0.6205166325961717,
      "grad_norm": 0.7833722829818726,
      "learning_rate": 0.00015654698992903534,
      "loss": 1.213,
      "step": 3339
    },
    {
      "epoch": 0.6207024716595428,
      "grad_norm": 0.6131447553634644,
      "learning_rate": 0.00015652286641612125,
      "loss": 1.1272,
      "step": 3340
    },
    {
      "epoch": 0.620888310722914,
      "grad_norm": 0.5262309908866882,
      "learning_rate": 0.0001564987380686843,
      "loss": 0.9924,
      "step": 3341
    },
    {
      "epoch": 0.6210741497862851,
      "grad_norm": 0.6729386448860168,
      "learning_rate": 0.00015647460488878813,
      "loss": 1.2097,
      "step": 3342
    },
    {
      "epoch": 0.6212599888496562,
      "grad_norm": 0.5907789468765259,
      "learning_rate": 0.00015645046687849697,
      "loss": 1.1616,
      "step": 3343
    },
    {
      "epoch": 0.6214458279130273,
      "grad_norm": 0.7607451677322388,
      "learning_rate": 0.00015642632403987535,
      "loss": 1.1074,
      "step": 3344
    },
    {
      "epoch": 0.6216316669763985,
      "grad_norm": 0.6834059357643127,
      "learning_rate": 0.00015640217637498827,
      "loss": 1.0589,
      "step": 3345
    },
    {
      "epoch": 0.6218175060397696,
      "grad_norm": 0.6914005279541016,
      "learning_rate": 0.00015637802388590122,
      "loss": 1.167,
      "step": 3346
    },
    {
      "epoch": 0.6220033451031407,
      "grad_norm": 0.48968812823295593,
      "learning_rate": 0.00015635386657467985,
      "loss": 1.0032,
      "step": 3347
    },
    {
      "epoch": 0.6221891841665118,
      "grad_norm": 0.5632472634315491,
      "learning_rate": 0.00015632970444339053,
      "loss": 1.0351,
      "step": 3348
    },
    {
      "epoch": 0.6223750232298829,
      "grad_norm": 0.5951436758041382,
      "learning_rate": 0.00015630553749409982,
      "loss": 1.2193,
      "step": 3349
    },
    {
      "epoch": 0.622560862293254,
      "grad_norm": 0.5487954616546631,
      "learning_rate": 0.0001562813657288748,
      "loss": 0.9555,
      "step": 3350
    },
    {
      "epoch": 0.6227467013566251,
      "grad_norm": 0.6174929141998291,
      "learning_rate": 0.00015625718914978292,
      "loss": 1.1938,
      "step": 3351
    },
    {
      "epoch": 0.6229325404199962,
      "grad_norm": 0.6154797673225403,
      "learning_rate": 0.0001562330077588921,
      "loss": 1.067,
      "step": 3352
    },
    {
      "epoch": 0.6231183794833675,
      "grad_norm": 0.6975616812705994,
      "learning_rate": 0.00015620882155827054,
      "loss": 1.145,
      "step": 3353
    },
    {
      "epoch": 0.6233042185467386,
      "grad_norm": 0.6247215867042542,
      "learning_rate": 0.00015618463054998703,
      "loss": 0.9011,
      "step": 3354
    },
    {
      "epoch": 0.6234900576101097,
      "grad_norm": 2.737778425216675,
      "learning_rate": 0.0001561604347361106,
      "loss": 2.8023,
      "step": 3355
    },
    {
      "epoch": 0.6236758966734808,
      "grad_norm": 0.6489484906196594,
      "learning_rate": 0.00015613623411871085,
      "loss": 1.1984,
      "step": 3356
    },
    {
      "epoch": 0.6238617357368519,
      "grad_norm": 0.7646564841270447,
      "learning_rate": 0.00015611202869985765,
      "loss": 0.9968,
      "step": 3357
    },
    {
      "epoch": 0.624047574800223,
      "grad_norm": 0.6500785946846008,
      "learning_rate": 0.00015608781848162136,
      "loss": 0.9897,
      "step": 3358
    },
    {
      "epoch": 0.6242334138635941,
      "grad_norm": 0.5417594909667969,
      "learning_rate": 0.00015606360346607273,
      "loss": 1.1919,
      "step": 3359
    },
    {
      "epoch": 0.6244192529269652,
      "grad_norm": 0.68771892786026,
      "learning_rate": 0.00015603938365528293,
      "loss": 1.2118,
      "step": 3360
    },
    {
      "epoch": 0.6246050919903363,
      "grad_norm": 0.5929567217826843,
      "learning_rate": 0.00015601515905132353,
      "loss": 1.078,
      "step": 3361
    },
    {
      "epoch": 0.6247909310537075,
      "grad_norm": 0.7144112586975098,
      "learning_rate": 0.0001559909296562665,
      "loss": 1.2353,
      "step": 3362
    },
    {
      "epoch": 0.6249767701170786,
      "grad_norm": 0.6896948218345642,
      "learning_rate": 0.00015596669547218427,
      "loss": 1.0764,
      "step": 3363
    },
    {
      "epoch": 0.6251626091804497,
      "grad_norm": 0.6907390356063843,
      "learning_rate": 0.00015594245650114958,
      "loss": 1.3201,
      "step": 3364
    },
    {
      "epoch": 0.6253484482438209,
      "grad_norm": 0.5990051031112671,
      "learning_rate": 0.00015591821274523571,
      "loss": 1.1825,
      "step": 3365
    },
    {
      "epoch": 0.625534287307192,
      "grad_norm": 0.6666750907897949,
      "learning_rate": 0.00015589396420651624,
      "loss": 1.1073,
      "step": 3366
    },
    {
      "epoch": 0.6257201263705631,
      "grad_norm": 0.6805914044380188,
      "learning_rate": 0.00015586971088706522,
      "loss": 1.3364,
      "step": 3367
    },
    {
      "epoch": 0.6259059654339342,
      "grad_norm": 0.6045242547988892,
      "learning_rate": 0.00015584545278895707,
      "loss": 1.094,
      "step": 3368
    },
    {
      "epoch": 0.6260918044973053,
      "grad_norm": 0.5706300139427185,
      "learning_rate": 0.00015582118991426663,
      "loss": 0.9863,
      "step": 3369
    },
    {
      "epoch": 0.6262776435606765,
      "grad_norm": 0.6760212779045105,
      "learning_rate": 0.00015579692226506918,
      "loss": 1.2693,
      "step": 3370
    },
    {
      "epoch": 0.6264634826240476,
      "grad_norm": 0.5949139595031738,
      "learning_rate": 0.00015577264984344038,
      "loss": 1.1793,
      "step": 3371
    },
    {
      "epoch": 0.6266493216874187,
      "grad_norm": 0.5809699296951294,
      "learning_rate": 0.00015574837265145628,
      "loss": 0.9453,
      "step": 3372
    },
    {
      "epoch": 0.6268351607507898,
      "grad_norm": 0.6867097020149231,
      "learning_rate": 0.00015572409069119337,
      "loss": 1.0772,
      "step": 3373
    },
    {
      "epoch": 0.6270209998141609,
      "grad_norm": 0.7444250583648682,
      "learning_rate": 0.0001556998039647286,
      "loss": 1.0622,
      "step": 3374
    },
    {
      "epoch": 0.627206838877532,
      "grad_norm": 0.5732178092002869,
      "learning_rate": 0.0001556755124741392,
      "loss": 0.9385,
      "step": 3375
    },
    {
      "epoch": 0.6273926779409031,
      "grad_norm": 0.6235290169715881,
      "learning_rate": 0.00015565121622150286,
      "loss": 0.8565,
      "step": 3376
    },
    {
      "epoch": 0.6275785170042743,
      "grad_norm": 0.6951614618301392,
      "learning_rate": 0.00015562691520889775,
      "loss": 1.1083,
      "step": 3377
    },
    {
      "epoch": 0.6277643560676455,
      "grad_norm": 0.5099678039550781,
      "learning_rate": 0.0001556026094384023,
      "loss": 1.1066,
      "step": 3378
    },
    {
      "epoch": 0.6279501951310166,
      "grad_norm": 0.575692892074585,
      "learning_rate": 0.00015557829891209554,
      "loss": 0.9044,
      "step": 3379
    },
    {
      "epoch": 0.6281360341943877,
      "grad_norm": 0.6911464929580688,
      "learning_rate": 0.00015555398363205675,
      "loss": 1.015,
      "step": 3380
    },
    {
      "epoch": 0.6283218732577588,
      "grad_norm": 0.7588498592376709,
      "learning_rate": 0.00015552966360036567,
      "loss": 1.2013,
      "step": 3381
    },
    {
      "epoch": 0.6285077123211299,
      "grad_norm": 0.6335672736167908,
      "learning_rate": 0.00015550533881910242,
      "loss": 0.927,
      "step": 3382
    },
    {
      "epoch": 0.628693551384501,
      "grad_norm": 0.6448066830635071,
      "learning_rate": 0.00015548100929034764,
      "loss": 0.9687,
      "step": 3383
    },
    {
      "epoch": 0.6288793904478721,
      "grad_norm": 0.7308876514434814,
      "learning_rate": 0.00015545667501618216,
      "loss": 0.9884,
      "step": 3384
    },
    {
      "epoch": 0.6290652295112432,
      "grad_norm": 0.5620437860488892,
      "learning_rate": 0.00015543233599868742,
      "loss": 1.04,
      "step": 3385
    },
    {
      "epoch": 0.6292510685746144,
      "grad_norm": 0.7370177507400513,
      "learning_rate": 0.0001554079922399452,
      "loss": 1.0136,
      "step": 3386
    },
    {
      "epoch": 0.6294369076379855,
      "grad_norm": 0.5509874224662781,
      "learning_rate": 0.00015538364374203764,
      "loss": 0.9867,
      "step": 3387
    },
    {
      "epoch": 0.6296227467013567,
      "grad_norm": 0.7003196477890015,
      "learning_rate": 0.00015535929050704733,
      "loss": 1.1006,
      "step": 3388
    },
    {
      "epoch": 0.6298085857647278,
      "grad_norm": NaN,
      "learning_rate": 0.00015535929050704733,
      "loss": 4.1587,
      "step": 3389
    },
    {
      "epoch": 0.6299944248280989,
      "grad_norm": 0.5861777067184448,
      "learning_rate": 0.00015533493253705726,
      "loss": 0.9348,
      "step": 3390
    },
    {
      "epoch": 0.63018026389147,
      "grad_norm": 0.6192556619644165,
      "learning_rate": 0.00015531056983415081,
      "loss": 1.0898,
      "step": 3391
    },
    {
      "epoch": 0.6303661029548411,
      "grad_norm": 0.6822803616523743,
      "learning_rate": 0.00015528620240041178,
      "loss": 1.2059,
      "step": 3392
    },
    {
      "epoch": 0.6305519420182122,
      "grad_norm": 0.5751301646232605,
      "learning_rate": 0.00015526183023792438,
      "loss": 1.2787,
      "step": 3393
    },
    {
      "epoch": 0.6307377810815834,
      "grad_norm": 0.6449471712112427,
      "learning_rate": 0.00015523745334877317,
      "loss": 1.1165,
      "step": 3394
    },
    {
      "epoch": 0.6309236201449545,
      "grad_norm": 0.612648069858551,
      "learning_rate": 0.00015521307173504325,
      "loss": 0.7657,
      "step": 3395
    },
    {
      "epoch": 0.6311094592083256,
      "grad_norm": 0.5513817667961121,
      "learning_rate": 0.00015518868539881995,
      "loss": 0.9294,
      "step": 3396
    },
    {
      "epoch": 0.6312952982716967,
      "grad_norm": 0.6446556448936462,
      "learning_rate": 0.00015516429434218908,
      "loss": 1.0127,
      "step": 3397
    },
    {
      "epoch": 0.6314811373350678,
      "grad_norm": 0.4763330817222595,
      "learning_rate": 0.00015513989856723691,
      "loss": 0.916,
      "step": 3398
    },
    {
      "epoch": 0.6316669763984389,
      "grad_norm": 14.481205940246582,
      "learning_rate": 0.00015511549807605004,
      "loss": 3.4295,
      "step": 3399
    },
    {
      "epoch": 0.63185281546181,
      "grad_norm": 0.5876346826553345,
      "learning_rate": 0.00015509109287071548,
      "loss": 0.9924,
      "step": 3400
    },
    {
      "epoch": 0.6320386545251812,
      "grad_norm": 0.6718258857727051,
      "learning_rate": 0.00015506668295332072,
      "loss": 1.103,
      "step": 3401
    },
    {
      "epoch": 0.6322244935885523,
      "grad_norm": 0.6530651450157166,
      "learning_rate": 0.00015504226832595354,
      "loss": 1.2541,
      "step": 3402
    },
    {
      "epoch": 0.6324103326519235,
      "grad_norm": 0.5370796918869019,
      "learning_rate": 0.00015501784899070217,
      "loss": 1.2188,
      "step": 3403
    },
    {
      "epoch": 0.6325961717152946,
      "grad_norm": 0.635278582572937,
      "learning_rate": 0.00015499342494965527,
      "loss": 1.2426,
      "step": 3404
    },
    {
      "epoch": 0.6327820107786657,
      "grad_norm": 0.49484050273895264,
      "learning_rate": 0.0001549689962049019,
      "loss": 0.8188,
      "step": 3405
    },
    {
      "epoch": 0.6329678498420368,
      "grad_norm": 0.5422400832176208,
      "learning_rate": 0.0001549445627585315,
      "loss": 1.0764,
      "step": 3406
    },
    {
      "epoch": 0.6331536889054079,
      "grad_norm": 0.637448251247406,
      "learning_rate": 0.00015492012461263384,
      "loss": 0.7457,
      "step": 3407
    },
    {
      "epoch": 0.633339527968779,
      "grad_norm": 0.5702096223831177,
      "learning_rate": 0.00015489568176929922,
      "loss": 1.1017,
      "step": 3408
    },
    {
      "epoch": 0.6335253670321501,
      "grad_norm": 0.5875257849693298,
      "learning_rate": 0.00015487123423061833,
      "loss": 0.9366,
      "step": 3409
    },
    {
      "epoch": 0.6337112060955212,
      "grad_norm": 0.6866803765296936,
      "learning_rate": 0.00015484678199868219,
      "loss": 1.1363,
      "step": 3410
    },
    {
      "epoch": 0.6338970451588924,
      "grad_norm": 0.5591610074043274,
      "learning_rate": 0.00015482232507558222,
      "loss": 1.244,
      "step": 3411
    },
    {
      "epoch": 0.6340828842222636,
      "grad_norm": 0.6071485280990601,
      "learning_rate": 0.0001547978634634103,
      "loss": 1.0588,
      "step": 3412
    },
    {
      "epoch": 0.6342687232856347,
      "grad_norm": 0.6361015439033508,
      "learning_rate": 0.0001547733971642587,
      "loss": 0.9244,
      "step": 3413
    },
    {
      "epoch": 0.6344545623490058,
      "grad_norm": 0.560198187828064,
      "learning_rate": 0.00015474892618022005,
      "loss": 1.1407,
      "step": 3414
    },
    {
      "epoch": 0.6346404014123769,
      "grad_norm": 0.6156582832336426,
      "learning_rate": 0.0001547244505133874,
      "loss": 1.0394,
      "step": 3415
    },
    {
      "epoch": 0.634826240475748,
      "grad_norm": 0.6680965423583984,
      "learning_rate": 0.00015469997016585426,
      "loss": 1.0968,
      "step": 3416
    },
    {
      "epoch": 0.6350120795391191,
      "grad_norm": 0.54937744140625,
      "learning_rate": 0.00015467548513971443,
      "loss": 0.7901,
      "step": 3417
    },
    {
      "epoch": 0.6351979186024902,
      "grad_norm": 0.6295645833015442,
      "learning_rate": 0.0001546509954370622,
      "loss": 1.1601,
      "step": 3418
    },
    {
      "epoch": 0.6353837576658614,
      "grad_norm": 0.6593343615531921,
      "learning_rate": 0.00015462650105999218,
      "loss": 1.0454,
      "step": 3419
    },
    {
      "epoch": 0.6355695967292325,
      "grad_norm": 0.6861504316329956,
      "learning_rate": 0.0001546020020105995,
      "loss": 1.2475,
      "step": 3420
    },
    {
      "epoch": 0.6357554357926036,
      "grad_norm": 0.5935508608818054,
      "learning_rate": 0.00015457749829097956,
      "loss": 1.1566,
      "step": 3421
    },
    {
      "epoch": 0.6359412748559747,
      "grad_norm": 0.6246775984764099,
      "learning_rate": 0.00015455298990322823,
      "loss": 0.9687,
      "step": 3422
    },
    {
      "epoch": 0.6361271139193458,
      "grad_norm": 0.6714119911193848,
      "learning_rate": 0.00015452847684944176,
      "loss": 0.9973,
      "step": 3423
    },
    {
      "epoch": 0.636312952982717,
      "grad_norm": 0.6952091455459595,
      "learning_rate": 0.00015450395913171685,
      "loss": 1.3226,
      "step": 3424
    },
    {
      "epoch": 0.6364987920460881,
      "grad_norm": 0.7437610030174255,
      "learning_rate": 0.0001544794367521505,
      "loss": 1.1179,
      "step": 3425
    },
    {
      "epoch": 0.6366846311094592,
      "grad_norm": 0.5943450331687927,
      "learning_rate": 0.00015445490971284018,
      "loss": 1.0157,
      "step": 3426
    },
    {
      "epoch": 0.6368704701728304,
      "grad_norm": 0.5073950886726379,
      "learning_rate": 0.0001544303780158837,
      "loss": 0.9862,
      "step": 3427
    },
    {
      "epoch": 0.6370563092362015,
      "grad_norm": 0.6207197904586792,
      "learning_rate": 0.00015440584166337938,
      "loss": 1.0514,
      "step": 3428
    },
    {
      "epoch": 0.6372421482995726,
      "grad_norm": 0.6031699776649475,
      "learning_rate": 0.00015438130065742586,
      "loss": 1.1663,
      "step": 3429
    },
    {
      "epoch": 0.6374279873629437,
      "grad_norm": 0.6535128355026245,
      "learning_rate": 0.00015435675500012212,
      "loss": 1.2384,
      "step": 3430
    },
    {
      "epoch": 0.6376138264263148,
      "grad_norm": 0.5577135682106018,
      "learning_rate": 0.00015433220469356765,
      "loss": 0.9835,
      "step": 3431
    },
    {
      "epoch": 0.6377996654896859,
      "grad_norm": 0.5418036580085754,
      "learning_rate": 0.00015430764973986235,
      "loss": 0.8347,
      "step": 3432
    },
    {
      "epoch": 0.637985504553057,
      "grad_norm": 0.7411396503448486,
      "learning_rate": 0.00015428309014110632,
      "loss": 1.1138,
      "step": 3433
    },
    {
      "epoch": 0.6381713436164281,
      "grad_norm": 0.6470693349838257,
      "learning_rate": 0.00015425852589940025,
      "loss": 1.1362,
      "step": 3434
    },
    {
      "epoch": 0.6383571826797992,
      "grad_norm": 0.6360995769500732,
      "learning_rate": 0.00015423395701684526,
      "loss": 1.0167,
      "step": 3435
    },
    {
      "epoch": 0.6385430217431705,
      "grad_norm": 0.5424529910087585,
      "learning_rate": 0.00015420938349554268,
      "loss": 0.8663,
      "step": 3436
    },
    {
      "epoch": 0.6387288608065416,
      "grad_norm": 0.598699688911438,
      "learning_rate": 0.00015418480533759435,
      "loss": 1.1082,
      "step": 3437
    },
    {
      "epoch": 0.6389146998699127,
      "grad_norm": 0.5573627948760986,
      "learning_rate": 0.00015416022254510257,
      "loss": 1.006,
      "step": 3438
    },
    {
      "epoch": 0.6391005389332838,
      "grad_norm": 0.5559731721878052,
      "learning_rate": 0.00015413563512016985,
      "loss": 1.0956,
      "step": 3439
    },
    {
      "epoch": 0.6392863779966549,
      "grad_norm": 0.6015757322311401,
      "learning_rate": 0.00015411104306489926,
      "loss": 1.191,
      "step": 3440
    },
    {
      "epoch": 0.639472217060026,
      "grad_norm": 0.6004772782325745,
      "learning_rate": 0.0001540864463813942,
      "loss": 1.2299,
      "step": 3441
    },
    {
      "epoch": 0.6396580561233971,
      "grad_norm": 0.6631706953048706,
      "learning_rate": 0.00015406184507175852,
      "loss": 1.2987,
      "step": 3442
    },
    {
      "epoch": 0.6398438951867682,
      "grad_norm": 0.6589309573173523,
      "learning_rate": 0.00015403723913809639,
      "loss": 1.1748,
      "step": 3443
    },
    {
      "epoch": 0.6400297342501394,
      "grad_norm": 0.574073851108551,
      "learning_rate": 0.00015401262858251235,
      "loss": 1.1021,
      "step": 3444
    },
    {
      "epoch": 0.6402155733135105,
      "grad_norm": 0.7661458849906921,
      "learning_rate": 0.0001539880134071115,
      "loss": 1.2319,
      "step": 3445
    },
    {
      "epoch": 0.6404014123768816,
      "grad_norm": 0.6638999581336975,
      "learning_rate": 0.00015396339361399916,
      "loss": 0.9982,
      "step": 3446
    },
    {
      "epoch": 0.6405872514402527,
      "grad_norm": 0.5810231566429138,
      "learning_rate": 0.00015393876920528115,
      "loss": 1.1258,
      "step": 3447
    },
    {
      "epoch": 0.6407730905036239,
      "grad_norm": 0.5583173036575317,
      "learning_rate": 0.00015391414018306362,
      "loss": 1.1781,
      "step": 3448
    },
    {
      "epoch": 0.640958929566995,
      "grad_norm": 0.6540656685829163,
      "learning_rate": 0.00015388950654945314,
      "loss": 1.1244,
      "step": 3449
    },
    {
      "epoch": 0.6411447686303661,
      "grad_norm": 0.5573239922523499,
      "learning_rate": 0.0001538648683065567,
      "loss": 1.1899,
      "step": 3450
    },
    {
      "epoch": 0.6413306076937372,
      "grad_norm": 0.6116712093353271,
      "learning_rate": 0.0001538402254564817,
      "loss": 1.1629,
      "step": 3451
    },
    {
      "epoch": 0.6415164467571084,
      "grad_norm": 0.5380821228027344,
      "learning_rate": 0.0001538155780013358,
      "loss": 1.199,
      "step": 3452
    },
    {
      "epoch": 0.6417022858204795,
      "grad_norm": 0.4550449550151825,
      "learning_rate": 0.00015379092594322723,
      "loss": 1.0023,
      "step": 3453
    },
    {
      "epoch": 0.6418881248838506,
      "grad_norm": 0.5812940001487732,
      "learning_rate": 0.00015376626928426452,
      "loss": 1.008,
      "step": 3454
    },
    {
      "epoch": 0.6420739639472217,
      "grad_norm": 0.5325769782066345,
      "learning_rate": 0.00015374160802655658,
      "loss": 0.868,
      "step": 3455
    },
    {
      "epoch": 0.6422598030105928,
      "grad_norm": 0.650091826915741,
      "learning_rate": 0.00015371694217221273,
      "loss": 1.03,
      "step": 3456
    },
    {
      "epoch": 0.6424456420739639,
      "grad_norm": 0.7132396101951599,
      "learning_rate": 0.00015369227172334272,
      "loss": 1.4007,
      "step": 3457
    },
    {
      "epoch": 0.642631481137335,
      "grad_norm": 0.5265616774559021,
      "learning_rate": 0.0001536675966820567,
      "loss": 1.074,
      "step": 3458
    },
    {
      "epoch": 0.6428173202007061,
      "grad_norm": 0.6761186718940735,
      "learning_rate": 0.00015364291705046514,
      "loss": 1.0798,
      "step": 3459
    },
    {
      "epoch": 0.6430031592640774,
      "grad_norm": 0.5889752507209778,
      "learning_rate": 0.00015361823283067892,
      "loss": 1.1018,
      "step": 3460
    },
    {
      "epoch": 0.6431889983274485,
      "grad_norm": 0.6676234006881714,
      "learning_rate": 0.00015359354402480938,
      "loss": 1.1207,
      "step": 3461
    },
    {
      "epoch": 0.6433748373908196,
      "grad_norm": 0.6662126779556274,
      "learning_rate": 0.00015356885063496818,
      "loss": 0.8809,
      "step": 3462
    },
    {
      "epoch": 0.6435606764541907,
      "grad_norm": 0.6595667600631714,
      "learning_rate": 0.00015354415266326742,
      "loss": 0.9899,
      "step": 3463
    },
    {
      "epoch": 0.6437465155175618,
      "grad_norm": 0.7342430949211121,
      "learning_rate": 0.00015351945011181955,
      "loss": 0.725,
      "step": 3464
    },
    {
      "epoch": 0.6439323545809329,
      "grad_norm": 0.5512349009513855,
      "learning_rate": 0.00015349474298273746,
      "loss": 1.1645,
      "step": 3465
    },
    {
      "epoch": 0.644118193644304,
      "grad_norm": 0.6810212135314941,
      "learning_rate": 0.00015347003127813436,
      "loss": 1.5384,
      "step": 3466
    },
    {
      "epoch": 0.6443040327076751,
      "grad_norm": 0.38854822516441345,
      "learning_rate": 0.0001534453150001239,
      "loss": 0.498,
      "step": 3467
    },
    {
      "epoch": 0.6444898717710462,
      "grad_norm": 0.6276911497116089,
      "learning_rate": 0.0001534205941508202,
      "loss": 1.0133,
      "step": 3468
    },
    {
      "epoch": 0.6446757108344174,
      "grad_norm": 0.7050503492355347,
      "learning_rate": 0.00015339586873233758,
      "loss": 1.2785,
      "step": 3469
    },
    {
      "epoch": 0.6448615498977885,
      "grad_norm": 0.6047936677932739,
      "learning_rate": 0.00015337113874679093,
      "loss": 1.0657,
      "step": 3470
    },
    {
      "epoch": 0.6450473889611597,
      "grad_norm": 0.7037943005561829,
      "learning_rate": 0.0001533464041962954,
      "loss": 1.1488,
      "step": 3471
    },
    {
      "epoch": 0.6452332280245308,
      "grad_norm": 0.5448530912399292,
      "learning_rate": 0.00015332166508296664,
      "loss": 0.981,
      "step": 3472
    },
    {
      "epoch": 0.6454190670879019,
      "grad_norm": 0.5085611343383789,
      "learning_rate": 0.00015329692140892063,
      "loss": 0.933,
      "step": 3473
    },
    {
      "epoch": 0.645604906151273,
      "grad_norm": 0.637241780757904,
      "learning_rate": 0.00015327217317627373,
      "loss": 1.2294,
      "step": 3474
    },
    {
      "epoch": 0.6457907452146441,
      "grad_norm": 0.5922691822052002,
      "learning_rate": 0.00015324742038714276,
      "loss": 1.1787,
      "step": 3475
    },
    {
      "epoch": 0.6459765842780152,
      "grad_norm": 0.5249952673912048,
      "learning_rate": 0.00015322266304364476,
      "loss": 1.0432,
      "step": 3476
    },
    {
      "epoch": 0.6461624233413864,
      "grad_norm": 0.591828465461731,
      "learning_rate": 0.00015319790114789742,
      "loss": 0.8827,
      "step": 3477
    },
    {
      "epoch": 0.6463482624047575,
      "grad_norm": 0.646541953086853,
      "learning_rate": 0.00015317313470201863,
      "loss": 1.3075,
      "step": 3478
    },
    {
      "epoch": 0.6465341014681286,
      "grad_norm": 0.660753607749939,
      "learning_rate": 0.00015314836370812666,
      "loss": 1.0812,
      "step": 3479
    },
    {
      "epoch": 0.6467199405314997,
      "grad_norm": 0.5507911443710327,
      "learning_rate": 0.0001531235881683403,
      "loss": 0.9117,
      "step": 3480
    },
    {
      "epoch": 0.6469057795948708,
      "grad_norm": 0.6520745158195496,
      "learning_rate": 0.00015309880808477862,
      "loss": 1.0439,
      "step": 3481
    },
    {
      "epoch": 0.6470916186582419,
      "grad_norm": 0.6552932262420654,
      "learning_rate": 0.00015307402345956115,
      "loss": 1.1353,
      "step": 3482
    },
    {
      "epoch": 0.647277457721613,
      "grad_norm": 0.7491993308067322,
      "learning_rate": 0.00015304923429480772,
      "loss": 1.3769,
      "step": 3483
    },
    {
      "epoch": 0.6474632967849842,
      "grad_norm": 0.4848712980747223,
      "learning_rate": 0.0001530244405926386,
      "loss": 0.8271,
      "step": 3484
    },
    {
      "epoch": 0.6476491358483554,
      "grad_norm": 0.5544674396514893,
      "learning_rate": 0.00015299964235517455,
      "loss": 1.1254,
      "step": 3485
    },
    {
      "epoch": 0.6478349749117265,
      "grad_norm": 0.5810096859931946,
      "learning_rate": 0.00015297483958453647,
      "loss": 1.2283,
      "step": 3486
    },
    {
      "epoch": 0.6480208139750976,
      "grad_norm": 0.5955312252044678,
      "learning_rate": 0.0001529500322828459,
      "loss": 1.3159,
      "step": 3487
    },
    {
      "epoch": 0.6482066530384687,
      "grad_norm": 0.6579847931861877,
      "learning_rate": 0.00015292522045222463,
      "loss": 0.9444,
      "step": 3488
    },
    {
      "epoch": 0.6483924921018398,
      "grad_norm": 0.5204761028289795,
      "learning_rate": 0.00015290040409479487,
      "loss": 0.8009,
      "step": 3489
    },
    {
      "epoch": 0.6485783311652109,
      "grad_norm": 0.6247485876083374,
      "learning_rate": 0.00015287558321267918,
      "loss": 1.1415,
      "step": 3490
    },
    {
      "epoch": 0.648764170228582,
      "grad_norm": 0.6254748106002808,
      "learning_rate": 0.00015285075780800062,
      "loss": 1.142,
      "step": 3491
    },
    {
      "epoch": 0.6489500092919531,
      "grad_norm": 0.7298083305358887,
      "learning_rate": 0.00015282592788288252,
      "loss": 1.0529,
      "step": 3492
    },
    {
      "epoch": 0.6491358483553243,
      "grad_norm": 0.6158937215805054,
      "learning_rate": 0.0001528010934394486,
      "loss": 0.9719,
      "step": 3493
    },
    {
      "epoch": 0.6493216874186954,
      "grad_norm": 0.5640282034873962,
      "learning_rate": 0.00015277625447982307,
      "loss": 1.1164,
      "step": 3494
    },
    {
      "epoch": 0.6495075264820666,
      "grad_norm": 0.5224336981773376,
      "learning_rate": 0.00015275141100613043,
      "loss": 0.8594,
      "step": 3495
    },
    {
      "epoch": 0.6496933655454377,
      "grad_norm": 0.5831015706062317,
      "learning_rate": 0.00015272656302049557,
      "loss": 0.9334,
      "step": 3496
    },
    {
      "epoch": 0.6498792046088088,
      "grad_norm": 0.6398945450782776,
      "learning_rate": 0.00015270171052504382,
      "loss": 1.2502,
      "step": 3497
    },
    {
      "epoch": 0.6500650436721799,
      "grad_norm": 0.5970938205718994,
      "learning_rate": 0.0001526768535219009,
      "loss": 1.2101,
      "step": 3498
    },
    {
      "epoch": 0.650250882735551,
      "grad_norm": 0.6261694431304932,
      "learning_rate": 0.00015265199201319285,
      "loss": 0.5574,
      "step": 3499
    },
    {
      "epoch": 0.6504367217989221,
      "grad_norm": 0.5698041915893555,
      "learning_rate": 0.0001526271260010461,
      "loss": 1.0594,
      "step": 3500
    },
    {
      "epoch": 0.6506225608622933,
      "grad_norm": 0.5439754724502563,
      "learning_rate": 0.00015260225548758752,
      "loss": 0.8891,
      "step": 3501
    },
    {
      "epoch": 0.6508083999256644,
      "grad_norm": 0.7733392119407654,
      "learning_rate": 0.00015257738047494432,
      "loss": 1.0131,
      "step": 3502
    },
    {
      "epoch": 0.6509942389890355,
      "grad_norm": 0.6281485557556152,
      "learning_rate": 0.0001525525009652442,
      "loss": 1.1565,
      "step": 3503
    },
    {
      "epoch": 0.6511800780524066,
      "grad_norm": 0.633788526058197,
      "learning_rate": 0.00015252761696061501,
      "loss": 1.2533,
      "step": 3504
    },
    {
      "epoch": 0.6513659171157777,
      "grad_norm": 0.554947018623352,
      "learning_rate": 0.00015250272846318528,
      "loss": 1.1795,
      "step": 3505
    },
    {
      "epoch": 0.6515517561791488,
      "grad_norm": 0.6328675746917725,
      "learning_rate": 0.00015247783547508368,
      "loss": 0.8386,
      "step": 3506
    },
    {
      "epoch": 0.65173759524252,
      "grad_norm": 0.6410210132598877,
      "learning_rate": 0.00015245293799843942,
      "loss": 1.1097,
      "step": 3507
    },
    {
      "epoch": 0.6519234343058911,
      "grad_norm": 0.6602080464363098,
      "learning_rate": 0.00015242803603538195,
      "loss": 1.3585,
      "step": 3508
    },
    {
      "epoch": 0.6521092733692622,
      "grad_norm": 0.5675439834594727,
      "learning_rate": 0.00015240312958804132,
      "loss": 1.135,
      "step": 3509
    },
    {
      "epoch": 0.6522951124326334,
      "grad_norm": 0.6612635850906372,
      "learning_rate": 0.0001523782186585477,
      "loss": 1.3385,
      "step": 3510
    },
    {
      "epoch": 0.6524809514960045,
      "grad_norm": 0.6298087239265442,
      "learning_rate": 0.00015235330324903185,
      "loss": 1.2031,
      "step": 3511
    },
    {
      "epoch": 0.6526667905593756,
      "grad_norm": 0.6352697610855103,
      "learning_rate": 0.00015232838336162483,
      "loss": 0.9642,
      "step": 3512
    },
    {
      "epoch": 0.6528526296227467,
      "grad_norm": 0.5610973834991455,
      "learning_rate": 0.0001523034589984581,
      "loss": 1.1033,
      "step": 3513
    },
    {
      "epoch": 0.6530384686861178,
      "grad_norm": 0.598053514957428,
      "learning_rate": 0.00015227853016166345,
      "loss": 1.0022,
      "step": 3514
    },
    {
      "epoch": 0.6532243077494889,
      "grad_norm": 0.6255698204040527,
      "learning_rate": 0.00015225359685337313,
      "loss": 1.1182,
      "step": 3515
    },
    {
      "epoch": 0.65341014681286,
      "grad_norm": 0.5280336737632751,
      "learning_rate": 0.00015222865907571972,
      "loss": 1.1277,
      "step": 3516
    },
    {
      "epoch": 0.6535959858762311,
      "grad_norm": 0.6419905424118042,
      "learning_rate": 0.00015220371683083629,
      "loss": 1.0459,
      "step": 3517
    },
    {
      "epoch": 0.6537818249396024,
      "grad_norm": 0.6171340942382812,
      "learning_rate": 0.00015217877012085605,
      "loss": 1.1594,
      "step": 3518
    },
    {
      "epoch": 0.6539676640029735,
      "grad_norm": 0.6222787499427795,
      "learning_rate": 0.00015215381894791285,
      "loss": 1.1122,
      "step": 3519
    },
    {
      "epoch": 0.6541535030663446,
      "grad_norm": 0.6655737161636353,
      "learning_rate": 0.00015212886331414084,
      "loss": 1.0083,
      "step": 3520
    },
    {
      "epoch": 0.6543393421297157,
      "grad_norm": 0.6613302230834961,
      "learning_rate": 0.00015210390322167446,
      "loss": 1.0373,
      "step": 3521
    },
    {
      "epoch": 0.6545251811930868,
      "grad_norm": 0.6962905526161194,
      "learning_rate": 0.00015207893867264863,
      "loss": 1.0902,
      "step": 3522
    },
    {
      "epoch": 0.6547110202564579,
      "grad_norm": 0.581489622592926,
      "learning_rate": 0.0001520539696691986,
      "loss": 1.1598,
      "step": 3523
    },
    {
      "epoch": 0.654896859319829,
      "grad_norm": 0.660275399684906,
      "learning_rate": 0.00015202899621346005,
      "loss": 1.0715,
      "step": 3524
    },
    {
      "epoch": 0.6550826983832001,
      "grad_norm": 0.5727256536483765,
      "learning_rate": 0.00015200401830756897,
      "loss": 1.151,
      "step": 3525
    },
    {
      "epoch": 0.6552685374465713,
      "grad_norm": 0.8974356651306152,
      "learning_rate": 0.00015197903595366184,
      "loss": 1.3013,
      "step": 3526
    },
    {
      "epoch": 0.6554543765099424,
      "grad_norm": 0.6946527361869812,
      "learning_rate": 0.00015195404915387543,
      "loss": 1.1257,
      "step": 3527
    },
    {
      "epoch": 0.6556402155733135,
      "grad_norm": 0.5919240713119507,
      "learning_rate": 0.00015192905791034693,
      "loss": 1.2153,
      "step": 3528
    },
    {
      "epoch": 0.6558260546366846,
      "grad_norm": 0.5989977121353149,
      "learning_rate": 0.00015190406222521385,
      "loss": 1.2052,
      "step": 3529
    },
    {
      "epoch": 0.6560118937000557,
      "grad_norm": 0.5734437108039856,
      "learning_rate": 0.00015187906210061412,
      "loss": 0.9683,
      "step": 3530
    },
    {
      "epoch": 0.6561977327634269,
      "grad_norm": 0.6741088032722473,
      "learning_rate": 0.00015185405753868612,
      "loss": 0.9762,
      "step": 3531
    },
    {
      "epoch": 0.656383571826798,
      "grad_norm": 0.5712679624557495,
      "learning_rate": 0.0001518290485415685,
      "loss": 1.2456,
      "step": 3532
    },
    {
      "epoch": 0.6565694108901691,
      "grad_norm": 0.5627811551094055,
      "learning_rate": 0.00015180403511140034,
      "loss": 1.3626,
      "step": 3533
    },
    {
      "epoch": 0.6567552499535403,
      "grad_norm": 0.5959043502807617,
      "learning_rate": 0.0001517790172503211,
      "loss": 1.1376,
      "step": 3534
    },
    {
      "epoch": 0.6569410890169114,
      "grad_norm": 0.5870547294616699,
      "learning_rate": 0.0001517539949604706,
      "loss": 1.0926,
      "step": 3535
    },
    {
      "epoch": 0.6571269280802825,
      "grad_norm": 0.5044032335281372,
      "learning_rate": 0.00015172896824398904,
      "loss": 0.989,
      "step": 3536
    },
    {
      "epoch": 0.6573127671436536,
      "grad_norm": 0.5552332401275635,
      "learning_rate": 0.00015170393710301703,
      "loss": 1.1129,
      "step": 3537
    },
    {
      "epoch": 0.6574986062070247,
      "grad_norm": 0.6285333633422852,
      "learning_rate": 0.00015167890153969555,
      "loss": 1.3035,
      "step": 3538
    },
    {
      "epoch": 0.6576844452703958,
      "grad_norm": 0.6018518805503845,
      "learning_rate": 0.0001516538615561659,
      "loss": 1.1757,
      "step": 3539
    },
    {
      "epoch": 0.6578702843337669,
      "grad_norm": 0.6804785132408142,
      "learning_rate": 0.00015162881715456985,
      "loss": 1.1304,
      "step": 3540
    },
    {
      "epoch": 0.658056123397138,
      "grad_norm": 0.5028323531150818,
      "learning_rate": 0.00015160376833704947,
      "loss": 0.8834,
      "step": 3541
    },
    {
      "epoch": 0.6582419624605091,
      "grad_norm": 0.6550874710083008,
      "learning_rate": 0.00015157871510574727,
      "loss": 1.1537,
      "step": 3542
    },
    {
      "epoch": 0.6584278015238804,
      "grad_norm": 0.6116010546684265,
      "learning_rate": 0.0001515536574628061,
      "loss": 1.1274,
      "step": 3543
    },
    {
      "epoch": 0.6586136405872515,
      "grad_norm": 0.649966299533844,
      "learning_rate": 0.0001515285954103692,
      "loss": 0.7512,
      "step": 3544
    },
    {
      "epoch": 0.6587994796506226,
      "grad_norm": 0.7527809739112854,
      "learning_rate": 0.00015150352895058013,
      "loss": 1.164,
      "step": 3545
    },
    {
      "epoch": 0.6589853187139937,
      "grad_norm": 0.6256502866744995,
      "learning_rate": 0.0001514784580855829,
      "loss": 1.2635,
      "step": 3546
    },
    {
      "epoch": 0.6591711577773648,
      "grad_norm": 0.5845198631286621,
      "learning_rate": 0.00015145338281752192,
      "loss": 1.026,
      "step": 3547
    },
    {
      "epoch": 0.6593569968407359,
      "grad_norm": 0.6609519720077515,
      "learning_rate": 0.00015142830314854191,
      "loss": 1.1224,
      "step": 3548
    },
    {
      "epoch": 0.659542835904107,
      "grad_norm": 0.6253378987312317,
      "learning_rate": 0.00015140321908078798,
      "loss": 1.1397,
      "step": 3549
    },
    {
      "epoch": 0.6597286749674781,
      "grad_norm": 0.6805525422096252,
      "learning_rate": 0.00015137813061640563,
      "loss": 1.1877,
      "step": 3550
    },
    {
      "epoch": 0.6599145140308493,
      "grad_norm": 0.5571733713150024,
      "learning_rate": 0.0001513530377575407,
      "loss": 0.9482,
      "step": 3551
    },
    {
      "epoch": 0.6601003530942204,
      "grad_norm": 0.6012757420539856,
      "learning_rate": 0.00015132794050633949,
      "loss": 1.062,
      "step": 3552
    },
    {
      "epoch": 0.6602861921575915,
      "grad_norm": 0.6188012361526489,
      "learning_rate": 0.00015130283886494853,
      "loss": 0.9913,
      "step": 3553
    },
    {
      "epoch": 0.6604720312209627,
      "grad_norm": 0.7151011228561401,
      "learning_rate": 0.00015127773283551494,
      "loss": 1.2644,
      "step": 3554
    },
    {
      "epoch": 0.6606578702843338,
      "grad_norm": 0.5433036684989929,
      "learning_rate": 0.00015125262242018602,
      "loss": 1.1957,
      "step": 3555
    },
    {
      "epoch": 0.6608437093477049,
      "grad_norm": 0.6361022591590881,
      "learning_rate": 0.00015122750762110954,
      "loss": 1.1375,
      "step": 3556
    },
    {
      "epoch": 0.661029548411076,
      "grad_norm": 0.6053920984268188,
      "learning_rate": 0.00015120238844043362,
      "loss": 1.096,
      "step": 3557
    },
    {
      "epoch": 0.6612153874744471,
      "grad_norm": 0.5946423411369324,
      "learning_rate": 0.00015117726488030677,
      "loss": 0.9535,
      "step": 3558
    },
    {
      "epoch": 0.6614012265378183,
      "grad_norm": 0.6494511961936951,
      "learning_rate": 0.0001511521369428778,
      "loss": 0.9476,
      "step": 3559
    },
    {
      "epoch": 0.6615870656011894,
      "grad_norm": 0.7746506929397583,
      "learning_rate": 0.00015112700463029602,
      "loss": 0.7522,
      "step": 3560
    },
    {
      "epoch": 0.6617729046645605,
      "grad_norm": 0.6443954110145569,
      "learning_rate": 0.00015110186794471103,
      "loss": 1.2498,
      "step": 3561
    },
    {
      "epoch": 0.6619587437279316,
      "grad_norm": 0.6012477278709412,
      "learning_rate": 0.00015107672688827287,
      "loss": 0.8012,
      "step": 3562
    },
    {
      "epoch": 0.6621445827913027,
      "grad_norm": 0.6210688948631287,
      "learning_rate": 0.00015105158146313182,
      "loss": 1.0039,
      "step": 3563
    },
    {
      "epoch": 0.6623304218546738,
      "grad_norm": 0.621110737323761,
      "learning_rate": 0.0001510264316714387,
      "loss": 1.063,
      "step": 3564
    },
    {
      "epoch": 0.6625162609180449,
      "grad_norm": 0.5509716272354126,
      "learning_rate": 0.00015100127751534458,
      "loss": 1.0168,
      "step": 3565
    },
    {
      "epoch": 0.662702099981416,
      "grad_norm": 0.7248783707618713,
      "learning_rate": 0.00015097611899700096,
      "loss": 1.0515,
      "step": 3566
    },
    {
      "epoch": 0.6628879390447873,
      "grad_norm": 0.5176343321800232,
      "learning_rate": 0.00015095095611855971,
      "loss": 0.8917,
      "step": 3567
    },
    {
      "epoch": 0.6630737781081584,
      "grad_norm": 0.6484975814819336,
      "learning_rate": 0.00015092578888217307,
      "loss": 1.0518,
      "step": 3568
    },
    {
      "epoch": 0.6632596171715295,
      "grad_norm": 0.608376145362854,
      "learning_rate": 0.00015090061728999365,
      "loss": 1.1361,
      "step": 3569
    },
    {
      "epoch": 0.6634454562349006,
      "grad_norm": 0.5764702558517456,
      "learning_rate": 0.00015087544134417442,
      "loss": 0.8332,
      "step": 3570
    },
    {
      "epoch": 0.6636312952982717,
      "grad_norm": 0.5989474654197693,
      "learning_rate": 0.00015085026104686873,
      "loss": 0.9523,
      "step": 3571
    },
    {
      "epoch": 0.6638171343616428,
      "grad_norm": 0.5990676879882812,
      "learning_rate": 0.00015082507640023032,
      "loss": 0.941,
      "step": 3572
    },
    {
      "epoch": 0.6640029734250139,
      "grad_norm": 0.504645824432373,
      "learning_rate": 0.00015079988740641326,
      "loss": 0.8054,
      "step": 3573
    },
    {
      "epoch": 0.664188812488385,
      "grad_norm": 0.6226556897163391,
      "learning_rate": 0.00015077469406757206,
      "loss": 1.1862,
      "step": 3574
    },
    {
      "epoch": 0.6643746515517562,
      "grad_norm": 0.6347982287406921,
      "learning_rate": 0.0001507494963858615,
      "loss": 1.4854,
      "step": 3575
    },
    {
      "epoch": 0.6645604906151273,
      "grad_norm": 0.570537805557251,
      "learning_rate": 0.00015072429436343687,
      "loss": 0.9307,
      "step": 3576
    },
    {
      "epoch": 0.6647463296784984,
      "grad_norm": 0.6247116923332214,
      "learning_rate": 0.0001506990880024537,
      "loss": 1.2029,
      "step": 3577
    },
    {
      "epoch": 0.6649321687418696,
      "grad_norm": 0.7195250988006592,
      "learning_rate": 0.00015067387730506793,
      "loss": 1.0774,
      "step": 3578
    },
    {
      "epoch": 0.6651180078052407,
      "grad_norm": 0.5821033120155334,
      "learning_rate": 0.00015064866227343596,
      "loss": 1.1147,
      "step": 3579
    },
    {
      "epoch": 0.6653038468686118,
      "grad_norm": 0.6598196625709534,
      "learning_rate": 0.00015062344290971442,
      "loss": 1.2274,
      "step": 3580
    },
    {
      "epoch": 0.6654896859319829,
      "grad_norm": 0.5154291987419128,
      "learning_rate": 0.0001505982192160604,
      "loss": 1.0469,
      "step": 3581
    },
    {
      "epoch": 0.665675524995354,
      "grad_norm": 0.5266774892807007,
      "learning_rate": 0.00015057299119463128,
      "loss": 0.9102,
      "step": 3582
    },
    {
      "epoch": 0.6658613640587251,
      "grad_norm": 0.6214563250541687,
      "learning_rate": 0.00015054775884758498,
      "loss": 1.0344,
      "step": 3583
    },
    {
      "epoch": 0.6660472031220963,
      "grad_norm": 0.6299375891685486,
      "learning_rate": 0.00015052252217707962,
      "loss": 1.1272,
      "step": 3584
    },
    {
      "epoch": 0.6662330421854674,
      "grad_norm": 3.086122989654541,
      "learning_rate": 0.00015049728118527375,
      "loss": 2.7928,
      "step": 3585
    },
    {
      "epoch": 0.6664188812488385,
      "grad_norm": 0.556920051574707,
      "learning_rate": 0.00015047203587432626,
      "loss": 1.1483,
      "step": 3586
    },
    {
      "epoch": 0.6666047203122096,
      "grad_norm": 0.5242355465888977,
      "learning_rate": 0.00015044678624639648,
      "loss": 0.9973,
      "step": 3587
    },
    {
      "epoch": 0.6667905593755807,
      "grad_norm": 0.6420267820358276,
      "learning_rate": 0.000150421532303644,
      "loss": 1.1282,
      "step": 3588
    },
    {
      "epoch": 0.6669763984389518,
      "grad_norm": 0.6128627061843872,
      "learning_rate": 0.00015039627404822892,
      "loss": 1.0652,
      "step": 3589
    },
    {
      "epoch": 0.667162237502323,
      "grad_norm": 0.5869479775428772,
      "learning_rate": 0.00015037101148231158,
      "loss": 0.9901,
      "step": 3590
    },
    {
      "epoch": 0.6673480765656941,
      "grad_norm": 0.5982796549797058,
      "learning_rate": 0.00015034574460805279,
      "loss": 0.9696,
      "step": 3591
    },
    {
      "epoch": 0.6675339156290653,
      "grad_norm": 0.608452320098877,
      "learning_rate": 0.0001503204734276136,
      "loss": 1.2884,
      "step": 3592
    },
    {
      "epoch": 0.6677197546924364,
      "grad_norm": 0.6519098281860352,
      "learning_rate": 0.0001502951979431556,
      "loss": 1.132,
      "step": 3593
    },
    {
      "epoch": 0.6679055937558075,
      "grad_norm": 0.606254518032074,
      "learning_rate": 0.00015026991815684064,
      "loss": 1.1658,
      "step": 3594
    },
    {
      "epoch": 0.6680914328191786,
      "grad_norm": 0.6207895278930664,
      "learning_rate": 0.00015024463407083087,
      "loss": 1.0944,
      "step": 3595
    },
    {
      "epoch": 0.6682772718825497,
      "grad_norm": 0.6666253209114075,
      "learning_rate": 0.00015021934568728898,
      "loss": 1.2198,
      "step": 3596
    },
    {
      "epoch": 0.6684631109459208,
      "grad_norm": 0.5867999792098999,
      "learning_rate": 0.00015019405300837794,
      "loss": 1.0747,
      "step": 3597
    },
    {
      "epoch": 0.6686489500092919,
      "grad_norm": 0.6101366281509399,
      "learning_rate": 0.00015016875603626105,
      "loss": 1.2563,
      "step": 3598
    },
    {
      "epoch": 0.668834789072663,
      "grad_norm": 0.6357227563858032,
      "learning_rate": 0.00015014345477310203,
      "loss": 1.0178,
      "step": 3599
    },
    {
      "epoch": 0.6690206281360342,
      "grad_norm": 0.6158996224403381,
      "learning_rate": 0.00015011814922106495,
      "loss": 1.2432,
      "step": 3600
    },
    {
      "epoch": 0.6692064671994054,
      "grad_norm": 0.5483404994010925,
      "learning_rate": 0.00015009283938231422,
      "loss": 1.1968,
      "step": 3601
    },
    {
      "epoch": 0.6693923062627765,
      "grad_norm": 0.5520810484886169,
      "learning_rate": 0.00015006752525901468,
      "loss": 1.1107,
      "step": 3602
    },
    {
      "epoch": 0.6695781453261476,
      "grad_norm": 0.5998095870018005,
      "learning_rate": 0.00015004220685333153,
      "loss": 0.9836,
      "step": 3603
    },
    {
      "epoch": 0.6697639843895187,
      "grad_norm": 0.48578593134880066,
      "learning_rate": 0.00015001688416743024,
      "loss": 0.7812,
      "step": 3604
    },
    {
      "epoch": 0.6699498234528898,
      "grad_norm": 0.5815861821174622,
      "learning_rate": 0.00014999155720347675,
      "loss": 1.2315,
      "step": 3605
    },
    {
      "epoch": 0.6701356625162609,
      "grad_norm": 0.6169844269752502,
      "learning_rate": 0.00014996622596363733,
      "loss": 0.8476,
      "step": 3606
    },
    {
      "epoch": 0.670321501579632,
      "grad_norm": 0.5464169979095459,
      "learning_rate": 0.00014994089045007865,
      "loss": 0.9376,
      "step": 3607
    },
    {
      "epoch": 0.6705073406430032,
      "grad_norm": 0.6786235570907593,
      "learning_rate": 0.00014991555066496762,
      "loss": 1.0984,
      "step": 3608
    },
    {
      "epoch": 0.6706931797063743,
      "grad_norm": 0.7276574969291687,
      "learning_rate": 0.00014989020661047167,
      "loss": 1.0315,
      "step": 3609
    },
    {
      "epoch": 0.6708790187697454,
      "grad_norm": 0.60429847240448,
      "learning_rate": 0.00014986485828875855,
      "loss": 1.0041,
      "step": 3610
    },
    {
      "epoch": 0.6710648578331165,
      "grad_norm": 0.548711895942688,
      "learning_rate": 0.0001498395057019963,
      "loss": 0.9941,
      "step": 3611
    },
    {
      "epoch": 0.6712506968964876,
      "grad_norm": 0.610514760017395,
      "learning_rate": 0.00014981414885235342,
      "loss": 1.2205,
      "step": 3612
    },
    {
      "epoch": 0.6714365359598587,
      "grad_norm": 0.5238599181175232,
      "learning_rate": 0.0001497887877419987,
      "loss": 1.0296,
      "step": 3613
    },
    {
      "epoch": 0.6716223750232299,
      "grad_norm": 0.5102425217628479,
      "learning_rate": 0.0001497634223731014,
      "loss": 0.9913,
      "step": 3614
    },
    {
      "epoch": 0.671808214086601,
      "grad_norm": 0.6130759119987488,
      "learning_rate": 0.00014973805274783101,
      "loss": 1.0176,
      "step": 3615
    },
    {
      "epoch": 0.6719940531499721,
      "grad_norm": 0.5231581926345825,
      "learning_rate": 0.00014971267886835748,
      "loss": 0.969,
      "step": 3616
    },
    {
      "epoch": 0.6721798922133433,
      "grad_norm": 0.5673276782035828,
      "learning_rate": 0.00014968730073685108,
      "loss": 1.0535,
      "step": 3617
    },
    {
      "epoch": 0.6723657312767144,
      "grad_norm": 0.6071367263793945,
      "learning_rate": 0.00014966191835548247,
      "loss": 1.0199,
      "step": 3618
    },
    {
      "epoch": 0.6725515703400855,
      "grad_norm": 0.6172305345535278,
      "learning_rate": 0.0001496365317264226,
      "loss": 1.099,
      "step": 3619
    },
    {
      "epoch": 0.6727374094034566,
      "grad_norm": 0.5747281312942505,
      "learning_rate": 0.00014961114085184293,
      "loss": 1.2114,
      "step": 3620
    },
    {
      "epoch": 0.6729232484668277,
      "grad_norm": 0.5879877805709839,
      "learning_rate": 0.00014958574573391517,
      "loss": 1.2082,
      "step": 3621
    },
    {
      "epoch": 0.6731090875301988,
      "grad_norm": 0.5075231194496155,
      "learning_rate": 0.0001495603463748114,
      "loss": 0.8392,
      "step": 3622
    },
    {
      "epoch": 0.6732949265935699,
      "grad_norm": 0.6247887015342712,
      "learning_rate": 0.00014953494277670408,
      "loss": 1.0629,
      "step": 3623
    },
    {
      "epoch": 0.673480765656941,
      "grad_norm": 0.5654303431510925,
      "learning_rate": 0.00014950953494176605,
      "loss": 1.0257,
      "step": 3624
    },
    {
      "epoch": 0.6736666047203123,
      "grad_norm": 0.5614563226699829,
      "learning_rate": 0.0001494841228721705,
      "loss": 1.0373,
      "step": 3625
    },
    {
      "epoch": 0.6738524437836834,
      "grad_norm": 0.47630125284194946,
      "learning_rate": 0.00014945870657009095,
      "loss": 0.9697,
      "step": 3626
    },
    {
      "epoch": 0.6740382828470545,
      "grad_norm": 0.6390812397003174,
      "learning_rate": 0.00014943328603770137,
      "loss": 1.0709,
      "step": 3627
    },
    {
      "epoch": 0.6742241219104256,
      "grad_norm": 0.6576070189476013,
      "learning_rate": 0.00014940786127717596,
      "loss": 1.1442,
      "step": 3628
    },
    {
      "epoch": 0.6744099609737967,
      "grad_norm": 0.6565425395965576,
      "learning_rate": 0.00014938243229068942,
      "loss": 1.2,
      "step": 3629
    },
    {
      "epoch": 0.6745958000371678,
      "grad_norm": 0.6493493318557739,
      "learning_rate": 0.00014935699908041666,
      "loss": 1.0959,
      "step": 3630
    },
    {
      "epoch": 0.6747816391005389,
      "grad_norm": 0.5119420886039734,
      "learning_rate": 0.00014933156164853318,
      "loss": 1.0431,
      "step": 3631
    },
    {
      "epoch": 0.67496747816391,
      "grad_norm": 0.7568409442901611,
      "learning_rate": 0.00014930611999721457,
      "loss": 1.3266,
      "step": 3632
    },
    {
      "epoch": 0.6751533172272812,
      "grad_norm": 0.596426248550415,
      "learning_rate": 0.0001492806741286369,
      "loss": 1.1316,
      "step": 3633
    },
    {
      "epoch": 0.6753391562906523,
      "grad_norm": 0.6918880939483643,
      "learning_rate": 0.00014925522404497674,
      "loss": 1.286,
      "step": 3634
    },
    {
      "epoch": 0.6755249953540234,
      "grad_norm": 0.5288131237030029,
      "learning_rate": 0.00014922976974841078,
      "loss": 0.965,
      "step": 3635
    },
    {
      "epoch": 0.6757108344173945,
      "grad_norm": 0.6257370114326477,
      "learning_rate": 0.00014920431124111624,
      "loss": 1.0397,
      "step": 3636
    },
    {
      "epoch": 0.6758966734807657,
      "grad_norm": 0.581321656703949,
      "learning_rate": 0.00014917884852527058,
      "loss": 1.1978,
      "step": 3637
    },
    {
      "epoch": 0.6760825125441368,
      "grad_norm": 0.602165162563324,
      "learning_rate": 0.00014915338160305174,
      "loss": 1.2811,
      "step": 3638
    },
    {
      "epoch": 0.6762683516075079,
      "grad_norm": 0.4515889286994934,
      "learning_rate": 0.00014912791047663791,
      "loss": 0.9584,
      "step": 3639
    },
    {
      "epoch": 0.676454190670879,
      "grad_norm": 0.6313900351524353,
      "learning_rate": 0.00014910243514820777,
      "loss": 1.1116,
      "step": 3640
    },
    {
      "epoch": 0.6766400297342502,
      "grad_norm": 0.6517963409423828,
      "learning_rate": 0.00014907695561994016,
      "loss": 1.1079,
      "step": 3641
    },
    {
      "epoch": 0.6768258687976213,
      "grad_norm": 0.5934468507766724,
      "learning_rate": 0.00014905147189401449,
      "loss": 1.0876,
      "step": 3642
    },
    {
      "epoch": 0.6770117078609924,
      "grad_norm": 0.5406286120414734,
      "learning_rate": 0.00014902598397261043,
      "loss": 0.9613,
      "step": 3643
    },
    {
      "epoch": 0.6771975469243635,
      "grad_norm": 0.6092156767845154,
      "learning_rate": 0.000149000491857908,
      "loss": 0.8997,
      "step": 3644
    },
    {
      "epoch": 0.6773833859877346,
      "grad_norm": 0.5583072304725647,
      "learning_rate": 0.00014897499555208761,
      "loss": 1.2435,
      "step": 3645
    },
    {
      "epoch": 0.6775692250511057,
      "grad_norm": 0.6321995854377747,
      "learning_rate": 0.00014894949505732998,
      "loss": 1.0082,
      "step": 3646
    },
    {
      "epoch": 0.6777550641144768,
      "grad_norm": 0.5312910079956055,
      "learning_rate": 0.00014892399037581627,
      "loss": 1.0634,
      "step": 3647
    },
    {
      "epoch": 0.6779409031778479,
      "grad_norm": 0.6095939874649048,
      "learning_rate": 0.00014889848150972788,
      "loss": 0.9725,
      "step": 3648
    },
    {
      "epoch": 0.678126742241219,
      "grad_norm": 0.7284290790557861,
      "learning_rate": 0.0001488729684612467,
      "loss": 1.1883,
      "step": 3649
    },
    {
      "epoch": 0.6783125813045903,
      "grad_norm": 0.5818365216255188,
      "learning_rate": 0.00014884745123255487,
      "loss": 1.0793,
      "step": 3650
    },
    {
      "epoch": 0.6784984203679614,
      "grad_norm": 0.6459710001945496,
      "learning_rate": 0.00014882192982583502,
      "loss": 1.0533,
      "step": 3651
    },
    {
      "epoch": 0.6786842594313325,
      "grad_norm": 0.5360872745513916,
      "learning_rate": 0.00014879640424327,
      "loss": 0.9852,
      "step": 3652
    },
    {
      "epoch": 0.6788700984947036,
      "grad_norm": 0.5609398484230042,
      "learning_rate": 0.00014877087448704303,
      "loss": 1.0946,
      "step": 3653
    },
    {
      "epoch": 0.6790559375580747,
      "grad_norm": 0.7220416069030762,
      "learning_rate": 0.00014874534055933774,
      "loss": 0.8899,
      "step": 3654
    },
    {
      "epoch": 0.6792417766214458,
      "grad_norm": 0.5624337196350098,
      "learning_rate": 0.00014871980246233815,
      "loss": 1.2113,
      "step": 3655
    },
    {
      "epoch": 0.6794276156848169,
      "grad_norm": 0.5820707082748413,
      "learning_rate": 0.00014869426019822855,
      "loss": 1.1378,
      "step": 3656
    },
    {
      "epoch": 0.679613454748188,
      "grad_norm": 0.6045162677764893,
      "learning_rate": 0.0001486687137691936,
      "loss": 1.2984,
      "step": 3657
    },
    {
      "epoch": 0.6797992938115592,
      "grad_norm": 0.6777940392494202,
      "learning_rate": 0.00014864316317741842,
      "loss": 1.2248,
      "step": 3658
    },
    {
      "epoch": 0.6799851328749303,
      "grad_norm": 0.6275390386581421,
      "learning_rate": 0.00014861760842508836,
      "loss": 1.2587,
      "step": 3659
    },
    {
      "epoch": 0.6801709719383014,
      "grad_norm": 0.5438305139541626,
      "learning_rate": 0.00014859204951438916,
      "loss": 1.104,
      "step": 3660
    },
    {
      "epoch": 0.6803568110016726,
      "grad_norm": 0.7176220417022705,
      "learning_rate": 0.00014856648644750696,
      "loss": 1.2366,
      "step": 3661
    },
    {
      "epoch": 0.6805426500650437,
      "grad_norm": 0.5285633206367493,
      "learning_rate": 0.00014854091922662818,
      "loss": 0.9695,
      "step": 3662
    },
    {
      "epoch": 0.6807284891284148,
      "grad_norm": 0.5242528915405273,
      "learning_rate": 0.0001485153478539397,
      "loss": 1.0871,
      "step": 3663
    },
    {
      "epoch": 0.6809143281917859,
      "grad_norm": 0.6045339107513428,
      "learning_rate": 0.00014848977233162863,
      "loss": 1.1065,
      "step": 3664
    },
    {
      "epoch": 0.681100167255157,
      "grad_norm": 0.6746090650558472,
      "learning_rate": 0.00014846419266188255,
      "loss": 1.1891,
      "step": 3665
    },
    {
      "epoch": 0.6812860063185282,
      "grad_norm": 0.6344028115272522,
      "learning_rate": 0.0001484386088468893,
      "loss": 1.2639,
      "step": 3666
    },
    {
      "epoch": 0.6814718453818993,
      "grad_norm": 0.7116642594337463,
      "learning_rate": 0.0001484130208888372,
      "loss": 1.0406,
      "step": 3667
    },
    {
      "epoch": 0.6816576844452704,
      "grad_norm": 0.5471962094306946,
      "learning_rate": 0.00014838742878991475,
      "loss": 1.1569,
      "step": 3668
    },
    {
      "epoch": 0.6818435235086415,
      "grad_norm": 0.6500441431999207,
      "learning_rate": 0.00014836183255231096,
      "loss": 0.8941,
      "step": 3669
    },
    {
      "epoch": 0.6820293625720126,
      "grad_norm": 0.7653695940971375,
      "learning_rate": 0.00014833623217821507,
      "loss": 1.2929,
      "step": 3670
    },
    {
      "epoch": 0.6822152016353837,
      "grad_norm": 0.7275819182395935,
      "learning_rate": 0.00014831062766981678,
      "loss": 1.0589,
      "step": 3671
    },
    {
      "epoch": 0.6824010406987548,
      "grad_norm": 0.4899131655693054,
      "learning_rate": 0.00014828501902930613,
      "loss": 0.6845,
      "step": 3672
    },
    {
      "epoch": 0.682586879762126,
      "grad_norm": 0.5876761674880981,
      "learning_rate": 0.00014825940625887342,
      "loss": 1.0859,
      "step": 3673
    },
    {
      "epoch": 0.6827727188254972,
      "grad_norm": 0.5021051168441772,
      "learning_rate": 0.0001482337893607094,
      "loss": 0.9498,
      "step": 3674
    },
    {
      "epoch": 0.6829585578888683,
      "grad_norm": 0.8564853072166443,
      "learning_rate": 0.00014820816833700513,
      "loss": 1.3126,
      "step": 3675
    },
    {
      "epoch": 0.6831443969522394,
      "grad_norm": 0.5066491365432739,
      "learning_rate": 0.00014818254318995204,
      "loss": 0.9441,
      "step": 3676
    },
    {
      "epoch": 0.6833302360156105,
      "grad_norm": 0.6260133981704712,
      "learning_rate": 0.00014815691392174187,
      "loss": 1.203,
      "step": 3677
    },
    {
      "epoch": 0.6835160750789816,
      "grad_norm": 0.67519611120224,
      "learning_rate": 0.00014813128053456676,
      "loss": 1.2433,
      "step": 3678
    },
    {
      "epoch": 0.6837019141423527,
      "grad_norm": 0.5620138049125671,
      "learning_rate": 0.00014810564303061922,
      "loss": 1.0496,
      "step": 3679
    },
    {
      "epoch": 0.6838877532057238,
      "grad_norm": 0.6018763780593872,
      "learning_rate": 0.00014808000141209207,
      "loss": 1.0698,
      "step": 3680
    },
    {
      "epoch": 0.6840735922690949,
      "grad_norm": 0.6768195629119873,
      "learning_rate": 0.0001480543556811785,
      "loss": 1.1439,
      "step": 3681
    },
    {
      "epoch": 0.6842594313324661,
      "grad_norm": 0.4748595654964447,
      "learning_rate": 0.000148028705840072,
      "loss": 1.0552,
      "step": 3682
    },
    {
      "epoch": 0.6844452703958372,
      "grad_norm": 0.5992226004600525,
      "learning_rate": 0.00014800305189096647,
      "loss": 1.1218,
      "step": 3683
    },
    {
      "epoch": 0.6846311094592084,
      "grad_norm": 0.5371170043945312,
      "learning_rate": 0.0001479773938360562,
      "loss": 0.6813,
      "step": 3684
    },
    {
      "epoch": 0.6848169485225795,
      "grad_norm": 0.6463387608528137,
      "learning_rate": 0.00014795173167753573,
      "loss": 0.9385,
      "step": 3685
    },
    {
      "epoch": 0.6850027875859506,
      "grad_norm": 0.6344730257987976,
      "learning_rate": 0.0001479260654176,
      "loss": 0.9097,
      "step": 3686
    },
    {
      "epoch": 0.6851886266493217,
      "grad_norm": 0.7050276398658752,
      "learning_rate": 0.00014790039505844434,
      "loss": 1.0801,
      "step": 3687
    },
    {
      "epoch": 0.6853744657126928,
      "grad_norm": 0.6179599761962891,
      "learning_rate": 0.00014787472060226432,
      "loss": 1.1076,
      "step": 3688
    },
    {
      "epoch": 0.6855603047760639,
      "grad_norm": 0.587262749671936,
      "learning_rate": 0.00014784904205125603,
      "loss": 1.0408,
      "step": 3689
    },
    {
      "epoch": 0.685746143839435,
      "grad_norm": 0.6305316686630249,
      "learning_rate": 0.00014782335940761573,
      "loss": 1.2521,
      "step": 3690
    },
    {
      "epoch": 0.6859319829028062,
      "grad_norm": 0.6304357647895813,
      "learning_rate": 0.00014779767267354016,
      "loss": 1.1457,
      "step": 3691
    },
    {
      "epoch": 0.6861178219661773,
      "grad_norm": 0.5930676460266113,
      "learning_rate": 0.0001477719818512263,
      "loss": 0.9276,
      "step": 3692
    },
    {
      "epoch": 0.6863036610295484,
      "grad_norm": 0.6224808096885681,
      "learning_rate": 0.0001477462869428716,
      "loss": 1.0416,
      "step": 3693
    },
    {
      "epoch": 0.6864895000929195,
      "grad_norm": 0.7327170372009277,
      "learning_rate": 0.00014772058795067377,
      "loss": 1.255,
      "step": 3694
    },
    {
      "epoch": 0.6866753391562906,
      "grad_norm": 3.124260187149048,
      "learning_rate": 0.00014769488487683096,
      "loss": 1.9742,
      "step": 3695
    },
    {
      "epoch": 0.6868611782196617,
      "grad_norm": 0.6538242101669312,
      "learning_rate": 0.00014766917772354154,
      "loss": 1.1323,
      "step": 3696
    },
    {
      "epoch": 0.6870470172830329,
      "grad_norm": 0.5959080457687378,
      "learning_rate": 0.00014764346649300434,
      "loss": 1.0653,
      "step": 3697
    },
    {
      "epoch": 0.687232856346404,
      "grad_norm": 0.6007588505744934,
      "learning_rate": 0.00014761775118741848,
      "loss": 1.1666,
      "step": 3698
    },
    {
      "epoch": 0.6874186954097752,
      "grad_norm": 0.6197354793548584,
      "learning_rate": 0.0001475920318089834,
      "loss": 1.0461,
      "step": 3699
    },
    {
      "epoch": 0.6876045344731463,
      "grad_norm": 0.6801838874816895,
      "learning_rate": 0.000147566308359899,
      "loss": 1.061,
      "step": 3700
    },
    {
      "epoch": 0.6877903735365174,
      "grad_norm": 0.662079393863678,
      "learning_rate": 0.00014754058084236546,
      "loss": 1.1442,
      "step": 3701
    },
    {
      "epoch": 0.6879762125998885,
      "grad_norm": 0.6772270202636719,
      "learning_rate": 0.00014751484925858328,
      "loss": 1.1368,
      "step": 3702
    },
    {
      "epoch": 0.6881620516632596,
      "grad_norm": 0.9411020874977112,
      "learning_rate": 0.00014748911361075337,
      "loss": 1.209,
      "step": 3703
    },
    {
      "epoch": 0.6883478907266307,
      "grad_norm": 0.5544502139091492,
      "learning_rate": 0.00014746337390107692,
      "loss": 0.9643,
      "step": 3704
    },
    {
      "epoch": 0.6885337297900018,
      "grad_norm": 0.6655852198600769,
      "learning_rate": 0.00014743763013175552,
      "loss": 1.1832,
      "step": 3705
    },
    {
      "epoch": 0.6887195688533729,
      "grad_norm": 0.5229987502098083,
      "learning_rate": 0.00014741188230499105,
      "loss": 0.9048,
      "step": 3706
    },
    {
      "epoch": 0.6889054079167442,
      "grad_norm": 0.5792685747146606,
      "learning_rate": 0.00014738613042298586,
      "loss": 1.1681,
      "step": 3707
    },
    {
      "epoch": 0.6890912469801153,
      "grad_norm": 0.6217467784881592,
      "learning_rate": 0.0001473603744879425,
      "loss": 0.9995,
      "step": 3708
    },
    {
      "epoch": 0.6892770860434864,
      "grad_norm": 0.5169404745101929,
      "learning_rate": 0.00014733461450206395,
      "loss": 0.9465,
      "step": 3709
    },
    {
      "epoch": 0.6894629251068575,
      "grad_norm": 0.5788629055023193,
      "learning_rate": 0.00014730885046755355,
      "loss": 1.092,
      "step": 3710
    },
    {
      "epoch": 0.6896487641702286,
      "grad_norm": 0.5894346833229065,
      "learning_rate": 0.0001472830823866149,
      "loss": 1.16,
      "step": 3711
    },
    {
      "epoch": 0.6898346032335997,
      "grad_norm": 0.597175121307373,
      "learning_rate": 0.00014725731026145203,
      "loss": 1.2234,
      "step": 3712
    },
    {
      "epoch": 0.6900204422969708,
      "grad_norm": 0.6156522035598755,
      "learning_rate": 0.00014723153409426927,
      "loss": 1.1736,
      "step": 3713
    },
    {
      "epoch": 0.6902062813603419,
      "grad_norm": 0.5588281154632568,
      "learning_rate": 0.00014720575388727132,
      "loss": 1.1153,
      "step": 3714
    },
    {
      "epoch": 0.6903921204237131,
      "grad_norm": 0.5758021473884583,
      "learning_rate": 0.0001471799696426632,
      "loss": 1.0781,
      "step": 3715
    },
    {
      "epoch": 0.6905779594870842,
      "grad_norm": 0.6536004543304443,
      "learning_rate": 0.0001471541813626503,
      "loss": 1.0397,
      "step": 3716
    },
    {
      "epoch": 0.6907637985504553,
      "grad_norm": 0.6770461797714233,
      "learning_rate": 0.0001471283890494384,
      "loss": 0.9885,
      "step": 3717
    },
    {
      "epoch": 0.6909496376138264,
      "grad_norm": 0.5764928460121155,
      "learning_rate": 0.0001471025927052335,
      "loss": 1.0299,
      "step": 3718
    },
    {
      "epoch": 0.6911354766771975,
      "grad_norm": 0.6257376074790955,
      "learning_rate": 0.00014707679233224205,
      "loss": 1.0675,
      "step": 3719
    },
    {
      "epoch": 0.6913213157405687,
      "grad_norm": 0.606505811214447,
      "learning_rate": 0.00014705098793267078,
      "loss": 1.0614,
      "step": 3720
    },
    {
      "epoch": 0.6915071548039398,
      "grad_norm": 0.6218563914299011,
      "learning_rate": 0.00014702517950872685,
      "loss": 1.1444,
      "step": 3721
    },
    {
      "epoch": 0.6916929938673109,
      "grad_norm": 0.591198205947876,
      "learning_rate": 0.00014699936706261767,
      "loss": 1.1453,
      "step": 3722
    },
    {
      "epoch": 0.691878832930682,
      "grad_norm": 0.657878041267395,
      "learning_rate": 0.00014697355059655105,
      "loss": 1.1282,
      "step": 3723
    },
    {
      "epoch": 0.6920646719940532,
      "grad_norm": 0.5806160569190979,
      "learning_rate": 0.00014694773011273514,
      "loss": 1.1001,
      "step": 3724
    },
    {
      "epoch": 0.6922505110574243,
      "grad_norm": 0.5884582996368408,
      "learning_rate": 0.00014692190561337837,
      "loss": 1.0167,
      "step": 3725
    },
    {
      "epoch": 0.6924363501207954,
      "grad_norm": 0.6860553622245789,
      "learning_rate": 0.0001468960771006896,
      "loss": 1.3315,
      "step": 3726
    },
    {
      "epoch": 0.6926221891841665,
      "grad_norm": 0.5938162207603455,
      "learning_rate": 0.00014687024457687805,
      "loss": 0.9759,
      "step": 3727
    },
    {
      "epoch": 0.6928080282475376,
      "grad_norm": 0.5152567028999329,
      "learning_rate": 0.00014684440804415313,
      "loss": 1.0112,
      "step": 3728
    },
    {
      "epoch": 0.6929938673109087,
      "grad_norm": 0.5616927742958069,
      "learning_rate": 0.00014681856750472475,
      "loss": 1.1259,
      "step": 3729
    },
    {
      "epoch": 0.6931797063742798,
      "grad_norm": 0.6486595273017883,
      "learning_rate": 0.00014679272296080312,
      "loss": 1.0842,
      "step": 3730
    },
    {
      "epoch": 0.6933655454376509,
      "grad_norm": 0.7352533340454102,
      "learning_rate": 0.0001467668744145988,
      "loss": 1.1292,
      "step": 3731
    },
    {
      "epoch": 0.6935513845010222,
      "grad_norm": 0.6770464181900024,
      "learning_rate": 0.0001467410218683226,
      "loss": 0.9304,
      "step": 3732
    },
    {
      "epoch": 0.6937372235643933,
      "grad_norm": 0.9361321330070496,
      "learning_rate": 0.0001467151653241858,
      "loss": 1.1675,
      "step": 3733
    },
    {
      "epoch": 0.6939230626277644,
      "grad_norm": 0.6761358976364136,
      "learning_rate": 0.00014668930478439987,
      "loss": 0.8866,
      "step": 3734
    },
    {
      "epoch": 0.6941089016911355,
      "grad_norm": 0.5960943698883057,
      "learning_rate": 0.00014666344025117685,
      "loss": 0.9362,
      "step": 3735
    },
    {
      "epoch": 0.6942947407545066,
      "grad_norm": 0.4672621488571167,
      "learning_rate": 0.00014663757172672895,
      "loss": 0.923,
      "step": 3736
    },
    {
      "epoch": 0.6944805798178777,
      "grad_norm": 0.9298095107078552,
      "learning_rate": 0.00014661169921326872,
      "loss": 1.1617,
      "step": 3737
    },
    {
      "epoch": 0.6946664188812488,
      "grad_norm": 0.47797614336013794,
      "learning_rate": 0.00014658582271300912,
      "loss": 0.973,
      "step": 3738
    },
    {
      "epoch": 0.6948522579446199,
      "grad_norm": 0.6770762801170349,
      "learning_rate": 0.0001465599422281634,
      "loss": 1.0214,
      "step": 3739
    },
    {
      "epoch": 0.6950380970079911,
      "grad_norm": 0.6708875894546509,
      "learning_rate": 0.00014653405776094523,
      "loss": 1.0464,
      "step": 3740
    },
    {
      "epoch": 0.6952239360713622,
      "grad_norm": 0.7865105271339417,
      "learning_rate": 0.00014650816931356848,
      "loss": 1.3261,
      "step": 3741
    },
    {
      "epoch": 0.6954097751347333,
      "grad_norm": 0.5648350119590759,
      "learning_rate": 0.00014648227688824752,
      "loss": 0.882,
      "step": 3742
    },
    {
      "epoch": 0.6955956141981045,
      "grad_norm": 0.6764323711395264,
      "learning_rate": 0.00014645638048719695,
      "loss": 1.0985,
      "step": 3743
    },
    {
      "epoch": 0.6957814532614756,
      "grad_norm": 0.5773393511772156,
      "learning_rate": 0.00014643048011263175,
      "loss": 1.0442,
      "step": 3744
    },
    {
      "epoch": 0.6959672923248467,
      "grad_norm": 0.6438664793968201,
      "learning_rate": 0.00014640457576676723,
      "loss": 1.2884,
      "step": 3745
    },
    {
      "epoch": 0.6961531313882178,
      "grad_norm": 0.5628366470336914,
      "learning_rate": 0.00014637866745181906,
      "loss": 1.0877,
      "step": 3746
    },
    {
      "epoch": 0.6963389704515889,
      "grad_norm": 0.6594421863555908,
      "learning_rate": 0.00014635275517000323,
      "loss": 1.0294,
      "step": 3747
    },
    {
      "epoch": 0.6965248095149601,
      "grad_norm": 0.5636132955551147,
      "learning_rate": 0.00014632683892353604,
      "loss": 1.1675,
      "step": 3748
    },
    {
      "epoch": 0.6967106485783312,
      "grad_norm": 0.551912784576416,
      "learning_rate": 0.00014630091871463424,
      "loss": 0.9442,
      "step": 3749
    },
    {
      "epoch": 0.6968964876417023,
      "grad_norm": 0.6579510569572449,
      "learning_rate": 0.00014627499454551477,
      "loss": 1.1139,
      "step": 3750
    },
    {
      "epoch": 0.6970823267050734,
      "grad_norm": 0.7005996108055115,
      "learning_rate": 0.00014624906641839497,
      "loss": 1.1257,
      "step": 3751
    },
    {
      "epoch": 0.6972681657684445,
      "grad_norm": 0.6416681408882141,
      "learning_rate": 0.0001462231343354926,
      "loss": 0.9766,
      "step": 3752
    },
    {
      "epoch": 0.6974540048318156,
      "grad_norm": 0.6373915672302246,
      "learning_rate": 0.00014619719829902565,
      "loss": 1.3631,
      "step": 3753
    },
    {
      "epoch": 0.6976398438951867,
      "grad_norm": 0.6103551387786865,
      "learning_rate": 0.0001461712583112125,
      "loss": 1.0276,
      "step": 3754
    },
    {
      "epoch": 0.6978256829585578,
      "grad_norm": 0.5814418196678162,
      "learning_rate": 0.0001461453143742718,
      "loss": 1.0218,
      "step": 3755
    },
    {
      "epoch": 0.698011522021929,
      "grad_norm": 0.6234370470046997,
      "learning_rate": 0.00014611936649042265,
      "loss": 1.1922,
      "step": 3756
    },
    {
      "epoch": 0.6981973610853002,
      "grad_norm": 0.6188172698020935,
      "learning_rate": 0.00014609341466188436,
      "loss": 1.1766,
      "step": 3757
    },
    {
      "epoch": 0.6983832001486713,
      "grad_norm": 0.7053810954093933,
      "learning_rate": 0.00014606745889087674,
      "loss": 1.2396,
      "step": 3758
    },
    {
      "epoch": 0.6985690392120424,
      "grad_norm": 0.553032636642456,
      "learning_rate": 0.0001460414991796198,
      "loss": 1.1424,
      "step": 3759
    },
    {
      "epoch": 0.6987548782754135,
      "grad_norm": 0.6534532308578491,
      "learning_rate": 0.0001460155355303339,
      "loss": 1.2171,
      "step": 3760
    },
    {
      "epoch": 0.6989407173387846,
      "grad_norm": 0.595773458480835,
      "learning_rate": 0.00014598956794523984,
      "loss": 1.35,
      "step": 3761
    },
    {
      "epoch": 0.6991265564021557,
      "grad_norm": 0.5927870869636536,
      "learning_rate": 0.00014596359642655857,
      "loss": 0.8717,
      "step": 3762
    },
    {
      "epoch": 0.6993123954655268,
      "grad_norm": 0.5296638607978821,
      "learning_rate": 0.0001459376209765116,
      "loss": 1.2105,
      "step": 3763
    },
    {
      "epoch": 0.6994982345288979,
      "grad_norm": 0.6530826687812805,
      "learning_rate": 0.00014591164159732063,
      "loss": 1.1223,
      "step": 3764
    },
    {
      "epoch": 0.6996840735922691,
      "grad_norm": 0.725963294506073,
      "learning_rate": 0.00014588565829120774,
      "loss": 0.9904,
      "step": 3765
    },
    {
      "epoch": 0.6998699126556402,
      "grad_norm": 0.6792590022087097,
      "learning_rate": 0.00014585967106039527,
      "loss": 1.0106,
      "step": 3766
    },
    {
      "epoch": 0.7000557517190114,
      "grad_norm": 0.657814621925354,
      "learning_rate": 0.00014583367990710604,
      "loss": 0.9676,
      "step": 3767
    },
    {
      "epoch": 0.7002415907823825,
      "grad_norm": 0.7891011834144592,
      "learning_rate": 0.00014580768483356312,
      "loss": 0.7333,
      "step": 3768
    },
    {
      "epoch": 0.7004274298457536,
      "grad_norm": 0.6137871742248535,
      "learning_rate": 0.00014578168584198988,
      "loss": 1.0334,
      "step": 3769
    },
    {
      "epoch": 0.7006132689091247,
      "grad_norm": 0.6025509238243103,
      "learning_rate": 0.00014575568293461014,
      "loss": 0.7372,
      "step": 3770
    },
    {
      "epoch": 0.7007991079724958,
      "grad_norm": 0.5863919854164124,
      "learning_rate": 0.00014572967611364794,
      "loss": 1.1867,
      "step": 3771
    },
    {
      "epoch": 0.7009849470358669,
      "grad_norm": 0.6096488833427429,
      "learning_rate": 0.0001457036653813277,
      "loss": 1.1034,
      "step": 3772
    },
    {
      "epoch": 0.7011707860992381,
      "grad_norm": 0.6400436162948608,
      "learning_rate": 0.00014567765073987419,
      "loss": 1.0334,
      "step": 3773
    },
    {
      "epoch": 0.7013566251626092,
      "grad_norm": 0.6100161075592041,
      "learning_rate": 0.00014565163219151247,
      "loss": 1.186,
      "step": 3774
    },
    {
      "epoch": 0.7015424642259803,
      "grad_norm": 0.5866125822067261,
      "learning_rate": 0.00014562560973846802,
      "loss": 1.0076,
      "step": 3775
    },
    {
      "epoch": 0.7017283032893514,
      "grad_norm": 0.5028868913650513,
      "learning_rate": 0.0001455995833829665,
      "loss": 0.9723,
      "step": 3776
    },
    {
      "epoch": 0.7019141423527225,
      "grad_norm": 0.5891440510749817,
      "learning_rate": 0.00014557355312723412,
      "loss": 0.8673,
      "step": 3777
    },
    {
      "epoch": 0.7020999814160936,
      "grad_norm": 0.7121338248252869,
      "learning_rate": 0.0001455475189734972,
      "loss": 1.2381,
      "step": 3778
    },
    {
      "epoch": 0.7022858204794648,
      "grad_norm": 0.5803287625312805,
      "learning_rate": 0.00014552148092398257,
      "loss": 0.9524,
      "step": 3779
    },
    {
      "epoch": 0.7024716595428359,
      "grad_norm": 0.6246668696403503,
      "learning_rate": 0.00014549543898091728,
      "loss": 1.0676,
      "step": 3780
    },
    {
      "epoch": 0.7026574986062071,
      "grad_norm": 0.638159453868866,
      "learning_rate": 0.00014546939314652877,
      "loss": 1.1794,
      "step": 3781
    },
    {
      "epoch": 0.7028433376695782,
      "grad_norm": 0.7397798299789429,
      "learning_rate": 0.00014544334342304479,
      "loss": 1.1995,
      "step": 3782
    },
    {
      "epoch": 0.7030291767329493,
      "grad_norm": 0.6656100749969482,
      "learning_rate": 0.00014541728981269344,
      "loss": 1.2414,
      "step": 3783
    },
    {
      "epoch": 0.7032150157963204,
      "grad_norm": 0.6307308673858643,
      "learning_rate": 0.0001453912323177031,
      "loss": 1.2007,
      "step": 3784
    },
    {
      "epoch": 0.7034008548596915,
      "grad_norm": 0.6740713715553284,
      "learning_rate": 0.0001453651709403026,
      "loss": 1.1527,
      "step": 3785
    },
    {
      "epoch": 0.7035866939230626,
      "grad_norm": 0.5341428518295288,
      "learning_rate": 0.00014533910568272096,
      "loss": 1.0795,
      "step": 3786
    },
    {
      "epoch": 0.7037725329864337,
      "grad_norm": 0.5600630640983582,
      "learning_rate": 0.0001453130365471876,
      "loss": 1.1013,
      "step": 3787
    },
    {
      "epoch": 0.7039583720498048,
      "grad_norm": 0.5706908106803894,
      "learning_rate": 0.0001452869635359323,
      "loss": 1.0242,
      "step": 3788
    },
    {
      "epoch": 0.704144211113176,
      "grad_norm": 0.7494182586669922,
      "learning_rate": 0.00014526088665118517,
      "loss": 1.24,
      "step": 3789
    },
    {
      "epoch": 0.7043300501765472,
      "grad_norm": 0.860882580280304,
      "learning_rate": 0.00014523480589517652,
      "loss": 1.2362,
      "step": 3790
    },
    {
      "epoch": 0.7045158892399183,
      "grad_norm": 0.573000431060791,
      "learning_rate": 0.00014520872127013723,
      "loss": 1.3139,
      "step": 3791
    },
    {
      "epoch": 0.7047017283032894,
      "grad_norm": 0.621658980846405,
      "learning_rate": 0.00014518263277829823,
      "loss": 1.0566,
      "step": 3792
    },
    {
      "epoch": 0.7048875673666605,
      "grad_norm": 0.6506567001342773,
      "learning_rate": 0.00014515654042189107,
      "loss": 1.1673,
      "step": 3793
    },
    {
      "epoch": 0.7050734064300316,
      "grad_norm": 0.5923455357551575,
      "learning_rate": 0.00014513044420314737,
      "loss": 0.8684,
      "step": 3794
    },
    {
      "epoch": 0.7052592454934027,
      "grad_norm": 0.6548746228218079,
      "learning_rate": 0.00014510434412429922,
      "loss": 1.0784,
      "step": 3795
    },
    {
      "epoch": 0.7054450845567738,
      "grad_norm": 0.5360590815544128,
      "learning_rate": 0.00014507824018757906,
      "loss": 1.0521,
      "step": 3796
    },
    {
      "epoch": 0.7056309236201449,
      "grad_norm": 0.5973278284072876,
      "learning_rate": 0.00014505213239521955,
      "loss": 0.9411,
      "step": 3797
    },
    {
      "epoch": 0.7058167626835161,
      "grad_norm": 0.6145676970481873,
      "learning_rate": 0.0001450260207494538,
      "loss": 1.0747,
      "step": 3798
    },
    {
      "epoch": 0.7060026017468872,
      "grad_norm": 0.5279918313026428,
      "learning_rate": 0.00014499990525251524,
      "loss": 1.1325,
      "step": 3799
    },
    {
      "epoch": 0.7061884408102583,
      "grad_norm": 0.7617853283882141,
      "learning_rate": 0.0001449737859066375,
      "loss": 1.2461,
      "step": 3800
    },
    {
      "epoch": 0.7063742798736294,
      "grad_norm": 0.535610020160675,
      "learning_rate": 0.00014494766271405462,
      "loss": 0.873,
      "step": 3801
    },
    {
      "epoch": 0.7065601189370005,
      "grad_norm": 0.7088138461112976,
      "learning_rate": 0.00014492153567700102,
      "loss": 1.329,
      "step": 3802
    },
    {
      "epoch": 0.7067459580003717,
      "grad_norm": 0.5399606227874756,
      "learning_rate": 0.00014489540479771138,
      "loss": 0.8983,
      "step": 3803
    },
    {
      "epoch": 0.7069317970637428,
      "grad_norm": 0.6456025838851929,
      "learning_rate": 0.00014486927007842073,
      "loss": 1.3894,
      "step": 3804
    },
    {
      "epoch": 0.7071176361271139,
      "grad_norm": 0.5173892378807068,
      "learning_rate": 0.00014484313152136443,
      "loss": 1.0707,
      "step": 3805
    },
    {
      "epoch": 0.7073034751904851,
      "grad_norm": 0.6266672015190125,
      "learning_rate": 0.00014481698912877818,
      "loss": 1.0166,
      "step": 3806
    },
    {
      "epoch": 0.7074893142538562,
      "grad_norm": 0.6050211787223816,
      "learning_rate": 0.000144790842902898,
      "loss": 1.2041,
      "step": 3807
    },
    {
      "epoch": 0.7076751533172273,
      "grad_norm": 0.6445580124855042,
      "learning_rate": 0.00014476469284596024,
      "loss": 1.2802,
      "step": 3808
    },
    {
      "epoch": 0.7078609923805984,
      "grad_norm": 0.6184760332107544,
      "learning_rate": 0.0001447385389602015,
      "loss": 1.1176,
      "step": 3809
    },
    {
      "epoch": 0.7080468314439695,
      "grad_norm": 0.6994800567626953,
      "learning_rate": 0.00014471238124785888,
      "loss": 0.9874,
      "step": 3810
    },
    {
      "epoch": 0.7082326705073406,
      "grad_norm": 0.6110354065895081,
      "learning_rate": 0.00014468621971116963,
      "loss": 0.9592,
      "step": 3811
    },
    {
      "epoch": 0.7084185095707117,
      "grad_norm": 0.5898203253746033,
      "learning_rate": 0.0001446600543523714,
      "loss": 1.1581,
      "step": 3812
    },
    {
      "epoch": 0.7086043486340828,
      "grad_norm": 0.6678897142410278,
      "learning_rate": 0.00014463388517370224,
      "loss": 1.1353,
      "step": 3813
    },
    {
      "epoch": 0.708790187697454,
      "grad_norm": 0.601813018321991,
      "learning_rate": 0.0001446077121774004,
      "loss": 1.1991,
      "step": 3814
    },
    {
      "epoch": 0.7089760267608252,
      "grad_norm": 0.48424074053764343,
      "learning_rate": 0.00014458153536570456,
      "loss": 0.985,
      "step": 3815
    },
    {
      "epoch": 0.7091618658241963,
      "grad_norm": 11.190242767333984,
      "learning_rate": 0.00014455535474085362,
      "loss": 2.6148,
      "step": 3816
    },
    {
      "epoch": 0.7093477048875674,
      "grad_norm": 0.5649991035461426,
      "learning_rate": 0.00014452917030508688,
      "loss": 1.0995,
      "step": 3817
    },
    {
      "epoch": 0.7095335439509385,
      "grad_norm": 0.5765203237533569,
      "learning_rate": 0.00014450298206064395,
      "loss": 0.879,
      "step": 3818
    },
    {
      "epoch": 0.7097193830143096,
      "grad_norm": 0.6534028649330139,
      "learning_rate": 0.00014447679000976482,
      "loss": 1.0186,
      "step": 3819
    },
    {
      "epoch": 0.7099052220776807,
      "grad_norm": 0.6540850400924683,
      "learning_rate": 0.00014445059415468968,
      "loss": 1.3179,
      "step": 3820
    },
    {
      "epoch": 0.7100910611410518,
      "grad_norm": 0.5826019644737244,
      "learning_rate": 0.0001444243944976592,
      "loss": 1.0053,
      "step": 3821
    },
    {
      "epoch": 0.710276900204423,
      "grad_norm": 0.5010527968406677,
      "learning_rate": 0.00014439819104091422,
      "loss": 0.8363,
      "step": 3822
    },
    {
      "epoch": 0.7104627392677941,
      "grad_norm": 0.470480740070343,
      "learning_rate": 0.00014437198378669598,
      "loss": 0.8458,
      "step": 3823
    },
    {
      "epoch": 0.7106485783311652,
      "grad_norm": 0.5956453680992126,
      "learning_rate": 0.00014434577273724609,
      "loss": 1.2399,
      "step": 3824
    },
    {
      "epoch": 0.7108344173945363,
      "grad_norm": 0.7176646590232849,
      "learning_rate": 0.00014431955789480637,
      "loss": 0.9899,
      "step": 3825
    },
    {
      "epoch": 0.7110202564579075,
      "grad_norm": 0.4969988167285919,
      "learning_rate": 0.00014429333926161913,
      "loss": 0.7034,
      "step": 3826
    },
    {
      "epoch": 0.7112060955212786,
      "grad_norm": 0.4980323314666748,
      "learning_rate": 0.00014426711683992685,
      "loss": 1.0628,
      "step": 3827
    },
    {
      "epoch": 0.7113919345846497,
      "grad_norm": 0.6527779698371887,
      "learning_rate": 0.00014424089063197238,
      "loss": 1.0731,
      "step": 3828
    },
    {
      "epoch": 0.7115777736480208,
      "grad_norm": 0.6407890319824219,
      "learning_rate": 0.0001442146606399989,
      "loss": 1.126,
      "step": 3829
    },
    {
      "epoch": 0.7117636127113919,
      "grad_norm": 0.6298221349716187,
      "learning_rate": 0.00014418842686624998,
      "loss": 0.9776,
      "step": 3830
    },
    {
      "epoch": 0.7119494517747631,
      "grad_norm": 0.6135900616645813,
      "learning_rate": 0.00014416218931296936,
      "loss": 1.2445,
      "step": 3831
    },
    {
      "epoch": 0.7121352908381342,
      "grad_norm": 0.5972301959991455,
      "learning_rate": 0.00014413594798240125,
      "loss": 1.3075,
      "step": 3832
    },
    {
      "epoch": 0.7123211299015053,
      "grad_norm": 0.47998619079589844,
      "learning_rate": 0.00014410970287679012,
      "loss": 0.9496,
      "step": 3833
    },
    {
      "epoch": 0.7125069689648764,
      "grad_norm": 0.6807659268379211,
      "learning_rate": 0.00014408345399838078,
      "loss": 1.0152,
      "step": 3834
    },
    {
      "epoch": 0.7126928080282475,
      "grad_norm": 0.5685128569602966,
      "learning_rate": 0.00014405720134941835,
      "loss": 0.911,
      "step": 3835
    },
    {
      "epoch": 0.7128786470916186,
      "grad_norm": 0.636781632900238,
      "learning_rate": 0.00014403094493214825,
      "loss": 0.9629,
      "step": 3836
    },
    {
      "epoch": 0.7130644861549897,
      "grad_norm": 0.570308268070221,
      "learning_rate": 0.0001440046847488163,
      "loss": 1.2221,
      "step": 3837
    },
    {
      "epoch": 0.7132503252183608,
      "grad_norm": 0.5886622071266174,
      "learning_rate": 0.00014397842080166853,
      "loss": 1.202,
      "step": 3838
    },
    {
      "epoch": 0.7134361642817321,
      "grad_norm": 0.5327900648117065,
      "learning_rate": 0.00014395215309295138,
      "loss": 1.1774,
      "step": 3839
    },
    {
      "epoch": 0.7136220033451032,
      "grad_norm": 0.5286857485771179,
      "learning_rate": 0.00014392588162491157,
      "loss": 1.0491,
      "step": 3840
    },
    {
      "epoch": 0.7138078424084743,
      "grad_norm": 0.5948933362960815,
      "learning_rate": 0.00014389960639979617,
      "loss": 1.0008,
      "step": 3841
    },
    {
      "epoch": 0.7139936814718454,
      "grad_norm": 0.5479618906974792,
      "learning_rate": 0.00014387332741985258,
      "loss": 0.9369,
      "step": 3842
    },
    {
      "epoch": 0.7141795205352165,
      "grad_norm": 0.6125556826591492,
      "learning_rate": 0.00014384704468732847,
      "loss": 1.299,
      "step": 3843
    },
    {
      "epoch": 0.7143653595985876,
      "grad_norm": 0.615397036075592,
      "learning_rate": 0.00014382075820447185,
      "loss": 0.8941,
      "step": 3844
    },
    {
      "epoch": 0.7145511986619587,
      "grad_norm": 0.5345596671104431,
      "learning_rate": 0.00014379446797353106,
      "loss": 1.0561,
      "step": 3845
    },
    {
      "epoch": 0.7147370377253298,
      "grad_norm": 0.600506603717804,
      "learning_rate": 0.00014376817399675479,
      "loss": 1.0501,
      "step": 3846
    },
    {
      "epoch": 0.714922876788701,
      "grad_norm": 0.6428175568580627,
      "learning_rate": 0.00014374187627639196,
      "loss": 1.2755,
      "step": 3847
    },
    {
      "epoch": 0.7151087158520721,
      "grad_norm": 0.6733381152153015,
      "learning_rate": 0.00014371557481469198,
      "loss": 1.2535,
      "step": 3848
    },
    {
      "epoch": 0.7152945549154432,
      "grad_norm": 0.7081629633903503,
      "learning_rate": 0.00014368926961390438,
      "loss": 1.1677,
      "step": 3849
    },
    {
      "epoch": 0.7154803939788144,
      "grad_norm": 0.5885323286056519,
      "learning_rate": 0.0001436629606762791,
      "loss": 1.1147,
      "step": 3850
    },
    {
      "epoch": 0.7156662330421855,
      "grad_norm": 0.6808738112449646,
      "learning_rate": 0.00014363664800406647,
      "loss": 1.1371,
      "step": 3851
    },
    {
      "epoch": 0.7158520721055566,
      "grad_norm": 0.6646922826766968,
      "learning_rate": 0.00014361033159951699,
      "loss": 1.1832,
      "step": 3852
    },
    {
      "epoch": 0.7160379111689277,
      "grad_norm": 0.6628801226615906,
      "learning_rate": 0.0001435840114648816,
      "loss": 1.1926,
      "step": 3853
    },
    {
      "epoch": 0.7162237502322988,
      "grad_norm": 0.7184207439422607,
      "learning_rate": 0.0001435576876024115,
      "loss": 1.304,
      "step": 3854
    },
    {
      "epoch": 0.71640958929567,
      "grad_norm": 0.623140275478363,
      "learning_rate": 0.00014353136001435827,
      "loss": 1.1043,
      "step": 3855
    },
    {
      "epoch": 0.7165954283590411,
      "grad_norm": 0.639744222164154,
      "learning_rate": 0.00014350502870297374,
      "loss": 1.1718,
      "step": 3856
    },
    {
      "epoch": 0.7167812674224122,
      "grad_norm": 0.7367560267448425,
      "learning_rate": 0.00014347869367051005,
      "loss": 1.2005,
      "step": 3857
    },
    {
      "epoch": 0.7169671064857833,
      "grad_norm": 0.6177274584770203,
      "learning_rate": 0.00014345235491921974,
      "loss": 1.1259,
      "step": 3858
    },
    {
      "epoch": 0.7171529455491544,
      "grad_norm": 0.5447367429733276,
      "learning_rate": 0.00014342601245135556,
      "loss": 1.0504,
      "step": 3859
    },
    {
      "epoch": 0.7173387846125255,
      "grad_norm": 0.596401572227478,
      "learning_rate": 0.00014339966626917072,
      "loss": 1.0481,
      "step": 3860
    },
    {
      "epoch": 0.7175246236758966,
      "grad_norm": 0.5914924144744873,
      "learning_rate": 0.0001433733163749186,
      "loss": 1.2006,
      "step": 3861
    },
    {
      "epoch": 0.7177104627392678,
      "grad_norm": 0.6990320682525635,
      "learning_rate": 0.000143346962770853,
      "loss": 1.1984,
      "step": 3862
    },
    {
      "epoch": 0.717896301802639,
      "grad_norm": 0.5579658150672913,
      "learning_rate": 0.00014332060545922798,
      "loss": 0.996,
      "step": 3863
    },
    {
      "epoch": 0.7180821408660101,
      "grad_norm": 0.6074368357658386,
      "learning_rate": 0.00014329424444229797,
      "loss": 1.1579,
      "step": 3864
    },
    {
      "epoch": 0.7182679799293812,
      "grad_norm": 0.5289203524589539,
      "learning_rate": 0.00014326787972231765,
      "loss": 0.8867,
      "step": 3865
    },
    {
      "epoch": 0.7184538189927523,
      "grad_norm": 0.6656838059425354,
      "learning_rate": 0.00014324151130154207,
      "loss": 1.2723,
      "step": 3866
    },
    {
      "epoch": 0.7186396580561234,
      "grad_norm": 1.7440506219863892,
      "learning_rate": 0.00014321513918222657,
      "loss": 1.8616,
      "step": 3867
    },
    {
      "epoch": 0.7188254971194945,
      "grad_norm": 0.5146296620368958,
      "learning_rate": 0.0001431887633666268,
      "loss": 0.958,
      "step": 3868
    },
    {
      "epoch": 0.7190113361828656,
      "grad_norm": 0.6164444088935852,
      "learning_rate": 0.00014316238385699877,
      "loss": 1.2226,
      "step": 3869
    },
    {
      "epoch": 0.7191971752462367,
      "grad_norm": 0.6518514752388,
      "learning_rate": 0.00014313600065559876,
      "loss": 1.117,
      "step": 3870
    },
    {
      "epoch": 0.7193830143096078,
      "grad_norm": 0.5010316967964172,
      "learning_rate": 0.00014310961376468338,
      "loss": 0.9683,
      "step": 3871
    },
    {
      "epoch": 0.719568853372979,
      "grad_norm": 0.6240313649177551,
      "learning_rate": 0.00014308322318650957,
      "loss": 1.129,
      "step": 3872
    },
    {
      "epoch": 0.7197546924363502,
      "grad_norm": 0.5419994592666626,
      "learning_rate": 0.00014305682892333453,
      "loss": 1.1179,
      "step": 3873
    },
    {
      "epoch": 0.7199405314997213,
      "grad_norm": 0.6009571552276611,
      "learning_rate": 0.0001430304309774159,
      "loss": 1.2359,
      "step": 3874
    },
    {
      "epoch": 0.7201263705630924,
      "grad_norm": 0.7590610384941101,
      "learning_rate": 0.0001430040293510115,
      "loss": 1.1678,
      "step": 3875
    },
    {
      "epoch": 0.7203122096264635,
      "grad_norm": 0.6211830973625183,
      "learning_rate": 0.0001429776240463795,
      "loss": 1.2744,
      "step": 3876
    },
    {
      "epoch": 0.7204980486898346,
      "grad_norm": 0.7066710591316223,
      "learning_rate": 0.00014295121506577844,
      "loss": 1.1257,
      "step": 3877
    },
    {
      "epoch": 0.7206838877532057,
      "grad_norm": 0.716054379940033,
      "learning_rate": 0.00014292480241146716,
      "loss": 0.941,
      "step": 3878
    },
    {
      "epoch": 0.7208697268165768,
      "grad_norm": 0.6112728714942932,
      "learning_rate": 0.00014289838608570472,
      "loss": 1.3149,
      "step": 3879
    },
    {
      "epoch": 0.721055565879948,
      "grad_norm": 0.5588074326515198,
      "learning_rate": 0.00014287196609075062,
      "loss": 0.8184,
      "step": 3880
    },
    {
      "epoch": 0.7212414049433191,
      "grad_norm": 0.646625280380249,
      "learning_rate": 0.00014284554242886458,
      "loss": 0.9852,
      "step": 3881
    },
    {
      "epoch": 0.7214272440066902,
      "grad_norm": 0.5708473324775696,
      "learning_rate": 0.00014281911510230673,
      "loss": 0.8922,
      "step": 3882
    },
    {
      "epoch": 0.7216130830700613,
      "grad_norm": 0.5914095044136047,
      "learning_rate": 0.0001427926841133374,
      "loss": 1.1814,
      "step": 3883
    },
    {
      "epoch": 0.7217989221334324,
      "grad_norm": 0.7340452671051025,
      "learning_rate": 0.00014276624946421728,
      "loss": 1.09,
      "step": 3884
    },
    {
      "epoch": 0.7219847611968035,
      "grad_norm": 0.6473243832588196,
      "learning_rate": 0.00014273981115720745,
      "loss": 1.1023,
      "step": 3885
    },
    {
      "epoch": 0.7221706002601747,
      "grad_norm": 0.6255431175231934,
      "learning_rate": 0.0001427133691945692,
      "loss": 0.8885,
      "step": 3886
    },
    {
      "epoch": 0.7223564393235458,
      "grad_norm": 0.5867379307746887,
      "learning_rate": 0.00014268692357856415,
      "loss": 0.9642,
      "step": 3887
    },
    {
      "epoch": 0.722542278386917,
      "grad_norm": 0.5964205861091614,
      "learning_rate": 0.00014266047431145428,
      "loss": 1.1225,
      "step": 3888
    },
    {
      "epoch": 0.7227281174502881,
      "grad_norm": 0.6257895827293396,
      "learning_rate": 0.00014263402139550184,
      "loss": 1.0778,
      "step": 3889
    },
    {
      "epoch": 0.7229139565136592,
      "grad_norm": 0.7534006237983704,
      "learning_rate": 0.0001426075648329694,
      "loss": 1.1555,
      "step": 3890
    },
    {
      "epoch": 0.7230997955770303,
      "grad_norm": 0.7284505367279053,
      "learning_rate": 0.00014258110462611986,
      "loss": 0.8882,
      "step": 3891
    },
    {
      "epoch": 0.7232856346404014,
      "grad_norm": 0.54058837890625,
      "learning_rate": 0.00014255464077721642,
      "loss": 0.6677,
      "step": 3892
    },
    {
      "epoch": 0.7234714737037725,
      "grad_norm": 0.5655109882354736,
      "learning_rate": 0.00014252817328852256,
      "loss": 0.9088,
      "step": 3893
    },
    {
      "epoch": 0.7236573127671436,
      "grad_norm": 0.6212193369865417,
      "learning_rate": 0.0001425017021623021,
      "loss": 1.1465,
      "step": 3894
    },
    {
      "epoch": 0.7238431518305147,
      "grad_norm": 0.6020740866661072,
      "learning_rate": 0.00014247522740081924,
      "loss": 0.9097,
      "step": 3895
    },
    {
      "epoch": 0.724028990893886,
      "grad_norm": 0.5451498627662659,
      "learning_rate": 0.00014244874900633839,
      "loss": 0.8835,
      "step": 3896
    },
    {
      "epoch": 0.7242148299572571,
      "grad_norm": 0.6161134243011475,
      "learning_rate": 0.00014242226698112426,
      "loss": 1.1958,
      "step": 3897
    },
    {
      "epoch": 0.7244006690206282,
      "grad_norm": 0.6352266669273376,
      "learning_rate": 0.00014239578132744195,
      "loss": 1.1156,
      "step": 3898
    },
    {
      "epoch": 0.7245865080839993,
      "grad_norm": 0.6710106134414673,
      "learning_rate": 0.00014236929204755684,
      "loss": 0.9735,
      "step": 3899
    },
    {
      "epoch": 0.7247723471473704,
      "grad_norm": 0.6463992595672607,
      "learning_rate": 0.00014234279914373464,
      "loss": 1.0622,
      "step": 3900
    },
    {
      "epoch": 0.7249581862107415,
      "grad_norm": 0.6164476275444031,
      "learning_rate": 0.00014231630261824128,
      "loss": 1.0894,
      "step": 3901
    },
    {
      "epoch": 0.7251440252741126,
      "grad_norm": 0.5363966226577759,
      "learning_rate": 0.0001422898024733431,
      "loss": 1.2831,
      "step": 3902
    },
    {
      "epoch": 0.7253298643374837,
      "grad_norm": 0.533882200717926,
      "learning_rate": 0.00014226329871130672,
      "loss": 1.0128,
      "step": 3903
    },
    {
      "epoch": 0.7255157034008548,
      "grad_norm": 0.5951628088951111,
      "learning_rate": 0.00014223679133439906,
      "loss": 1.0399,
      "step": 3904
    },
    {
      "epoch": 0.725701542464226,
      "grad_norm": 0.5895074605941772,
      "learning_rate": 0.00014221028034488734,
      "loss": 1.0381,
      "step": 3905
    },
    {
      "epoch": 0.7258873815275971,
      "grad_norm": 0.740982174873352,
      "learning_rate": 0.00014218376574503915,
      "loss": 1.1248,
      "step": 3906
    },
    {
      "epoch": 0.7260732205909682,
      "grad_norm": 0.6732906103134155,
      "learning_rate": 0.00014215724753712227,
      "loss": 1.0312,
      "step": 3907
    },
    {
      "epoch": 0.7262590596543393,
      "grad_norm": 0.5585139393806458,
      "learning_rate": 0.0001421307257234049,
      "loss": 0.9479,
      "step": 3908
    },
    {
      "epoch": 0.7264448987177105,
      "grad_norm": 0.5866591334342957,
      "learning_rate": 0.0001421042003061555,
      "loss": 0.929,
      "step": 3909
    },
    {
      "epoch": 0.7266307377810816,
      "grad_norm": 0.6339684128761292,
      "learning_rate": 0.00014207767128764288,
      "loss": 1.0042,
      "step": 3910
    },
    {
      "epoch": 0.7268165768444527,
      "grad_norm": 0.5119479894638062,
      "learning_rate": 0.00014205113867013603,
      "loss": 1.0338,
      "step": 3911
    },
    {
      "epoch": 0.7270024159078238,
      "grad_norm": 0.7797014713287354,
      "learning_rate": 0.00014202460245590445,
      "loss": 1.1913,
      "step": 3912
    },
    {
      "epoch": 0.727188254971195,
      "grad_norm": 0.5340668559074402,
      "learning_rate": 0.00014199806264721777,
      "loss": 1.1246,
      "step": 3913
    },
    {
      "epoch": 0.7273740940345661,
      "grad_norm": 0.5588401556015015,
      "learning_rate": 0.00014197151924634607,
      "loss": 1.1095,
      "step": 3914
    },
    {
      "epoch": 0.7275599330979372,
      "grad_norm": 0.5594502687454224,
      "learning_rate": 0.00014194497225555956,
      "loss": 1.0379,
      "step": 3915
    },
    {
      "epoch": 0.7277457721613083,
      "grad_norm": 0.6729309558868408,
      "learning_rate": 0.00014191842167712894,
      "loss": 1.0472,
      "step": 3916
    },
    {
      "epoch": 0.7279316112246794,
      "grad_norm": 0.5249831676483154,
      "learning_rate": 0.00014189186751332512,
      "loss": 0.7952,
      "step": 3917
    },
    {
      "epoch": 0.7281174502880505,
      "grad_norm": 0.9038906693458557,
      "learning_rate": 0.00014186530976641935,
      "loss": 1.0048,
      "step": 3918
    },
    {
      "epoch": 0.7283032893514216,
      "grad_norm": 0.8185818791389465,
      "learning_rate": 0.00014183874843868313,
      "loss": 1.2085,
      "step": 3919
    },
    {
      "epoch": 0.7284891284147927,
      "grad_norm": 0.6493542194366455,
      "learning_rate": 0.00014181218353238832,
      "loss": 1.3127,
      "step": 3920
    },
    {
      "epoch": 0.728674967478164,
      "grad_norm": 0.5967090129852295,
      "learning_rate": 0.00014178561504980713,
      "loss": 1.0868,
      "step": 3921
    },
    {
      "epoch": 0.7288608065415351,
      "grad_norm": 0.6878150701522827,
      "learning_rate": 0.0001417590429932119,
      "loss": 1.2544,
      "step": 3922
    },
    {
      "epoch": 0.7290466456049062,
      "grad_norm": 0.598172128200531,
      "learning_rate": 0.00014173246736487552,
      "loss": 1.0597,
      "step": 3923
    },
    {
      "epoch": 0.7292324846682773,
      "grad_norm": 0.5575773119926453,
      "learning_rate": 0.00014170588816707103,
      "loss": 0.9363,
      "step": 3924
    },
    {
      "epoch": 0.7294183237316484,
      "grad_norm": 0.6191855072975159,
      "learning_rate": 0.00014167930540207174,
      "loss": 1.1353,
      "step": 3925
    },
    {
      "epoch": 0.7296041627950195,
      "grad_norm": 0.635525643825531,
      "learning_rate": 0.0001416527190721514,
      "loss": 1.23,
      "step": 3926
    },
    {
      "epoch": 0.7297900018583906,
      "grad_norm": 0.5543552041053772,
      "learning_rate": 0.00014162612917958397,
      "loss": 1.1228,
      "step": 3927
    },
    {
      "epoch": 0.7299758409217617,
      "grad_norm": 0.4938865602016449,
      "learning_rate": 0.00014159953572664375,
      "loss": 1.0233,
      "step": 3928
    },
    {
      "epoch": 0.7301616799851329,
      "grad_norm": 0.523423433303833,
      "learning_rate": 0.00014157293871560532,
      "loss": 0.8668,
      "step": 3929
    },
    {
      "epoch": 0.730347519048504,
      "grad_norm": 0.6982086300849915,
      "learning_rate": 0.00014154633814874361,
      "loss": 1.3362,
      "step": 3930
    },
    {
      "epoch": 0.7305333581118751,
      "grad_norm": 0.5337725281715393,
      "learning_rate": 0.0001415197340283338,
      "loss": 0.9793,
      "step": 3931
    },
    {
      "epoch": 0.7307191971752462,
      "grad_norm": 0.5501512885093689,
      "learning_rate": 0.0001414931263566514,
      "loss": 1.0592,
      "step": 3932
    },
    {
      "epoch": 0.7309050362386174,
      "grad_norm": 0.5574057698249817,
      "learning_rate": 0.0001414665151359722,
      "loss": 1.1248,
      "step": 3933
    },
    {
      "epoch": 0.7310908753019885,
      "grad_norm": 0.5722219347953796,
      "learning_rate": 0.0001414399003685724,
      "loss": 1.0139,
      "step": 3934
    },
    {
      "epoch": 0.7312767143653596,
      "grad_norm": 0.599181056022644,
      "learning_rate": 0.0001414132820567283,
      "loss": 1.0779,
      "step": 3935
    },
    {
      "epoch": 0.7314625534287307,
      "grad_norm": 0.6454370021820068,
      "learning_rate": 0.00014138666020271668,
      "loss": 0.847,
      "step": 3936
    },
    {
      "epoch": 0.7316483924921018,
      "grad_norm": 0.5219981074333191,
      "learning_rate": 0.0001413600348088146,
      "loss": 0.9294,
      "step": 3937
    },
    {
      "epoch": 0.731834231555473,
      "grad_norm": 0.5672059059143066,
      "learning_rate": 0.00014133340587729932,
      "loss": 1.0441,
      "step": 3938
    },
    {
      "epoch": 0.7320200706188441,
      "grad_norm": 0.6195349097251892,
      "learning_rate": 0.00014130677341044854,
      "loss": 0.8654,
      "step": 3939
    },
    {
      "epoch": 0.7322059096822152,
      "grad_norm": 0.6063309907913208,
      "learning_rate": 0.00014128013741054014,
      "loss": 1.2963,
      "step": 3940
    },
    {
      "epoch": 0.7323917487455863,
      "grad_norm": 0.5881887078285217,
      "learning_rate": 0.00014125349787985235,
      "loss": 1.1092,
      "step": 3941
    },
    {
      "epoch": 0.7325775878089574,
      "grad_norm": 0.6515393257141113,
      "learning_rate": 0.0001412268548206637,
      "loss": 1.2598,
      "step": 3942
    },
    {
      "epoch": 0.7327634268723285,
      "grad_norm": 0.6397451162338257,
      "learning_rate": 0.00014120020823525311,
      "loss": 1.4133,
      "step": 3943
    },
    {
      "epoch": 0.7329492659356996,
      "grad_norm": 0.6065839529037476,
      "learning_rate": 0.00014117355812589963,
      "loss": 1.0174,
      "step": 3944
    },
    {
      "epoch": 0.7331351049990708,
      "grad_norm": 0.6057631373405457,
      "learning_rate": 0.00014114690449488278,
      "loss": 1.2133,
      "step": 3945
    },
    {
      "epoch": 0.733320944062442,
      "grad_norm": 0.6335853934288025,
      "learning_rate": 0.00014112024734448222,
      "loss": 1.3003,
      "step": 3946
    },
    {
      "epoch": 0.7335067831258131,
      "grad_norm": 0.763322651386261,
      "learning_rate": 0.00014109358667697806,
      "loss": 1.2127,
      "step": 3947
    },
    {
      "epoch": 0.7336926221891842,
      "grad_norm": 0.7434218525886536,
      "learning_rate": 0.0001410669224946506,
      "loss": 1.3435,
      "step": 3948
    },
    {
      "epoch": 0.7338784612525553,
      "grad_norm": 0.843048095703125,
      "learning_rate": 0.00014104025479978047,
      "loss": 1.0375,
      "step": 3949
    },
    {
      "epoch": 0.7340643003159264,
      "grad_norm": 0.5597147941589355,
      "learning_rate": 0.00014101358359464868,
      "loss": 1.191,
      "step": 3950
    },
    {
      "epoch": 0.7342501393792975,
      "grad_norm": 0.6737159490585327,
      "learning_rate": 0.00014098690888153642,
      "loss": 1.1248,
      "step": 3951
    },
    {
      "epoch": 0.7344359784426686,
      "grad_norm": 0.6314411759376526,
      "learning_rate": 0.0001409602306627253,
      "loss": 1.0922,
      "step": 3952
    },
    {
      "epoch": 0.7346218175060397,
      "grad_norm": 0.6439374089241028,
      "learning_rate": 0.0001409335489404971,
      "loss": 1.0697,
      "step": 3953
    },
    {
      "epoch": 0.7348076565694109,
      "grad_norm": 0.6825480461120605,
      "learning_rate": 0.00014090686371713402,
      "loss": 1.2774,
      "step": 3954
    },
    {
      "epoch": 0.734993495632782,
      "grad_norm": 0.5277664661407471,
      "learning_rate": 0.00014088017499491842,
      "loss": 1.0449,
      "step": 3955
    },
    {
      "epoch": 0.7351793346961532,
      "grad_norm": 0.6415640115737915,
      "learning_rate": 0.00014085348277613312,
      "loss": 1.1013,
      "step": 3956
    },
    {
      "epoch": 0.7353651737595243,
      "grad_norm": 0.8472128510475159,
      "learning_rate": 0.00014082678706306117,
      "loss": 1.1158,
      "step": 3957
    },
    {
      "epoch": 0.7355510128228954,
      "grad_norm": 0.5654361248016357,
      "learning_rate": 0.00014080008785798584,
      "loss": 1.1237,
      "step": 3958
    },
    {
      "epoch": 0.7357368518862665,
      "grad_norm": 0.5454433560371399,
      "learning_rate": 0.00014077338516319085,
      "loss": 1.1398,
      "step": 3959
    },
    {
      "epoch": 0.7359226909496376,
      "grad_norm": 0.5983814597129822,
      "learning_rate": 0.0001407466789809601,
      "loss": 1.3361,
      "step": 3960
    },
    {
      "epoch": 0.7361085300130087,
      "grad_norm": 0.6201050281524658,
      "learning_rate": 0.00014071996931357781,
      "loss": 1.0744,
      "step": 3961
    },
    {
      "epoch": 0.7362943690763799,
      "grad_norm": 0.5997368693351746,
      "learning_rate": 0.0001406932561633286,
      "loss": 1.0716,
      "step": 3962
    },
    {
      "epoch": 0.736480208139751,
      "grad_norm": 0.6071730852127075,
      "learning_rate": 0.00014066653953249717,
      "loss": 1.2947,
      "step": 3963
    },
    {
      "epoch": 0.7366660472031221,
      "grad_norm": 0.5586613416671753,
      "learning_rate": 0.00014063981942336872,
      "loss": 1.0362,
      "step": 3964
    },
    {
      "epoch": 0.7368518862664932,
      "grad_norm": 0.5747679471969604,
      "learning_rate": 0.0001406130958382287,
      "loss": 1.1044,
      "step": 3965
    },
    {
      "epoch": 0.7370377253298643,
      "grad_norm": 0.6741050481796265,
      "learning_rate": 0.00014058636877936284,
      "loss": 1.5444,
      "step": 3966
    },
    {
      "epoch": 0.7372235643932354,
      "grad_norm": 0.7054770588874817,
      "learning_rate": 0.0001405596382490571,
      "loss": 1.1852,
      "step": 3967
    },
    {
      "epoch": 0.7374094034566065,
      "grad_norm": 0.6902971863746643,
      "learning_rate": 0.00014053290424959788,
      "loss": 1.1828,
      "step": 3968
    },
    {
      "epoch": 0.7375952425199777,
      "grad_norm": 0.6020306944847107,
      "learning_rate": 0.00014050616678327175,
      "loss": 1.2144,
      "step": 3969
    },
    {
      "epoch": 0.7377810815833489,
      "grad_norm": 0.5189749598503113,
      "learning_rate": 0.00014047942585236563,
      "loss": 1.1501,
      "step": 3970
    },
    {
      "epoch": 0.73796692064672,
      "grad_norm": 0.5975673198699951,
      "learning_rate": 0.0001404526814591667,
      "loss": 1.0416,
      "step": 3971
    },
    {
      "epoch": 0.7381527597100911,
      "grad_norm": 0.571418046951294,
      "learning_rate": 0.0001404259336059625,
      "loss": 0.9414,
      "step": 3972
    },
    {
      "epoch": 0.7383385987734622,
      "grad_norm": 0.7188184261322021,
      "learning_rate": 0.00014039918229504083,
      "loss": 1.2489,
      "step": 3973
    },
    {
      "epoch": 0.7385244378368333,
      "grad_norm": 0.7322155833244324,
      "learning_rate": 0.0001403724275286898,
      "loss": 1.0685,
      "step": 3974
    },
    {
      "epoch": 0.7387102769002044,
      "grad_norm": 0.5024031400680542,
      "learning_rate": 0.0001403456693091978,
      "loss": 0.8796,
      "step": 3975
    },
    {
      "epoch": 0.7388961159635755,
      "grad_norm": 0.5071271061897278,
      "learning_rate": 0.00014031890763885346,
      "loss": 0.9038,
      "step": 3976
    },
    {
      "epoch": 0.7390819550269466,
      "grad_norm": 0.5381253957748413,
      "learning_rate": 0.00014029214251994585,
      "loss": 0.8357,
      "step": 3977
    },
    {
      "epoch": 0.7392677940903177,
      "grad_norm": 0.7139803767204285,
      "learning_rate": 0.00014026537395476417,
      "loss": 1.0646,
      "step": 3978
    },
    {
      "epoch": 0.739453633153689,
      "grad_norm": 0.5862656831741333,
      "learning_rate": 0.00014023860194559807,
      "loss": 1.3679,
      "step": 3979
    },
    {
      "epoch": 0.7396394722170601,
      "grad_norm": 0.5597640872001648,
      "learning_rate": 0.00014021182649473734,
      "loss": 1.0942,
      "step": 3980
    },
    {
      "epoch": 0.7398253112804312,
      "grad_norm": 0.7823113203048706,
      "learning_rate": 0.0001401850476044722,
      "loss": 1.0347,
      "step": 3981
    },
    {
      "epoch": 0.7400111503438023,
      "grad_norm": 0.5648866295814514,
      "learning_rate": 0.0001401582652770931,
      "loss": 0.9956,
      "step": 3982
    },
    {
      "epoch": 0.7401969894071734,
      "grad_norm": 0.5973963141441345,
      "learning_rate": 0.00014013147951489076,
      "loss": 0.7815,
      "step": 3983
    },
    {
      "epoch": 0.7403828284705445,
      "grad_norm": 0.7022433876991272,
      "learning_rate": 0.00014010469032015626,
      "loss": 1.0487,
      "step": 3984
    },
    {
      "epoch": 0.7405686675339156,
      "grad_norm": 0.7050195336341858,
      "learning_rate": 0.0001400778976951809,
      "loss": 1.0073,
      "step": 3985
    },
    {
      "epoch": 0.7407545065972867,
      "grad_norm": 0.6723904013633728,
      "learning_rate": 0.00014005110164225636,
      "loss": 0.962,
      "step": 3986
    },
    {
      "epoch": 0.7409403456606579,
      "grad_norm": 0.5963743329048157,
      "learning_rate": 0.0001400243021636745,
      "loss": 1.1638,
      "step": 3987
    },
    {
      "epoch": 0.741126184724029,
      "grad_norm": 0.6054471731185913,
      "learning_rate": 0.00013999749926172755,
      "loss": 1.1227,
      "step": 3988
    },
    {
      "epoch": 0.7413120237874001,
      "grad_norm": 0.578852117061615,
      "learning_rate": 0.00013997069293870807,
      "loss": 1.087,
      "step": 3989
    },
    {
      "epoch": 0.7414978628507712,
      "grad_norm": 0.7209727764129639,
      "learning_rate": 0.00013994388319690888,
      "loss": 1.1167,
      "step": 3990
    },
    {
      "epoch": 0.7416837019141423,
      "grad_norm": 0.7866323590278625,
      "learning_rate": 0.00013991707003862297,
      "loss": 1.1326,
      "step": 3991
    },
    {
      "epoch": 0.7418695409775135,
      "grad_norm": 0.6872072815895081,
      "learning_rate": 0.00013989025346614382,
      "loss": 0.8651,
      "step": 3992
    },
    {
      "epoch": 0.7420553800408846,
      "grad_norm": 0.568540096282959,
      "learning_rate": 0.00013986343348176505,
      "loss": 1.1095,
      "step": 3993
    },
    {
      "epoch": 0.7422412191042557,
      "grad_norm": 0.6199185252189636,
      "learning_rate": 0.00013983661008778065,
      "loss": 1.0038,
      "step": 3994
    },
    {
      "epoch": 0.7424270581676269,
      "grad_norm": 0.6343973875045776,
      "learning_rate": 0.00013980978328648492,
      "loss": 1.2641,
      "step": 3995
    },
    {
      "epoch": 0.742612897230998,
      "grad_norm": 0.723204493522644,
      "learning_rate": 0.0001397829530801724,
      "loss": 1.283,
      "step": 3996
    },
    {
      "epoch": 0.7427987362943691,
      "grad_norm": 0.5674862861633301,
      "learning_rate": 0.00013975611947113792,
      "loss": 1.1703,
      "step": 3997
    },
    {
      "epoch": 0.7429845753577402,
      "grad_norm": 0.5449051856994629,
      "learning_rate": 0.00013972928246167658,
      "loss": 0.8722,
      "step": 3998
    },
    {
      "epoch": 0.7431704144211113,
      "grad_norm": 0.6609678864479065,
      "learning_rate": 0.00013970244205408386,
      "loss": 1.2703,
      "step": 3999
    },
    {
      "epoch": 0.7433562534844824,
      "grad_norm": 0.6188348531723022,
      "learning_rate": 0.00013967559825065547,
      "loss": 1.3705,
      "step": 4000
    },
    {
      "epoch": 0.7435420925478535,
      "grad_norm": 0.5565700531005859,
      "learning_rate": 0.0001396487510536874,
      "loss": 0.821,
      "step": 4001
    },
    {
      "epoch": 0.7437279316112246,
      "grad_norm": 0.5666722655296326,
      "learning_rate": 0.00013962190046547596,
      "loss": 0.9848,
      "step": 4002
    },
    {
      "epoch": 0.7439137706745959,
      "grad_norm": 0.5390679836273193,
      "learning_rate": 0.00013959504648831776,
      "loss": 1.1785,
      "step": 4003
    },
    {
      "epoch": 0.744099609737967,
      "grad_norm": 0.5793556571006775,
      "learning_rate": 0.00013956818912450963,
      "loss": 1.204,
      "step": 4004
    },
    {
      "epoch": 0.7442854488013381,
      "grad_norm": 0.689157247543335,
      "learning_rate": 0.0001395413283763488,
      "loss": 1.4102,
      "step": 4005
    },
    {
      "epoch": 0.7444712878647092,
      "grad_norm": 0.6170220375061035,
      "learning_rate": 0.00013951446424613267,
      "loss": 1.0307,
      "step": 4006
    },
    {
      "epoch": 0.7446571269280803,
      "grad_norm": 0.7064149379730225,
      "learning_rate": 0.000139487596736159,
      "loss": 1.1184,
      "step": 4007
    },
    {
      "epoch": 0.7448429659914514,
      "grad_norm": 0.5319786667823792,
      "learning_rate": 0.00013946072584872587,
      "loss": 0.9286,
      "step": 4008
    },
    {
      "epoch": 0.7450288050548225,
      "grad_norm": 0.6800196766853333,
      "learning_rate": 0.00013943385158613153,
      "loss": 1.0468,
      "step": 4009
    },
    {
      "epoch": 0.7452146441181936,
      "grad_norm": 0.5505868196487427,
      "learning_rate": 0.00013940697395067465,
      "loss": 0.8411,
      "step": 4010
    },
    {
      "epoch": 0.7454004831815647,
      "grad_norm": 0.5862759351730347,
      "learning_rate": 0.00013938009294465415,
      "loss": 1.2071,
      "step": 4011
    },
    {
      "epoch": 0.7455863222449359,
      "grad_norm": 0.5666061639785767,
      "learning_rate": 0.00013935320857036917,
      "loss": 1.0593,
      "step": 4012
    },
    {
      "epoch": 0.745772161308307,
      "grad_norm": 0.6835244297981262,
      "learning_rate": 0.0001393263208301192,
      "loss": 1.0247,
      "step": 4013
    },
    {
      "epoch": 0.7459580003716781,
      "grad_norm": 0.5994834899902344,
      "learning_rate": 0.000139299429726204,
      "loss": 0.9067,
      "step": 4014
    },
    {
      "epoch": 0.7461438394350492,
      "grad_norm": 0.6289177536964417,
      "learning_rate": 0.00013927253526092368,
      "loss": 1.2112,
      "step": 4015
    },
    {
      "epoch": 0.7463296784984204,
      "grad_norm": 0.655208170413971,
      "learning_rate": 0.0001392456374365785,
      "loss": 1.1013,
      "step": 4016
    },
    {
      "epoch": 0.7465155175617915,
      "grad_norm": 0.5383544564247131,
      "learning_rate": 0.00013921873625546914,
      "loss": 1.1126,
      "step": 4017
    },
    {
      "epoch": 0.7467013566251626,
      "grad_norm": 0.4992688000202179,
      "learning_rate": 0.00013919183171989653,
      "loss": 0.6822,
      "step": 4018
    },
    {
      "epoch": 0.7468871956885337,
      "grad_norm": 0.6615790128707886,
      "learning_rate": 0.00013916492383216184,
      "loss": 0.9034,
      "step": 4019
    },
    {
      "epoch": 0.7470730347519049,
      "grad_norm": 0.6461586356163025,
      "learning_rate": 0.00013913801259456653,
      "loss": 1.0132,
      "step": 4020
    },
    {
      "epoch": 0.747258873815276,
      "grad_norm": 0.5311331152915955,
      "learning_rate": 0.00013911109800941244,
      "loss": 0.907,
      "step": 4021
    },
    {
      "epoch": 0.7474447128786471,
      "grad_norm": 0.6406325101852417,
      "learning_rate": 0.00013908418007900164,
      "loss": 0.9742,
      "step": 4022
    },
    {
      "epoch": 0.7476305519420182,
      "grad_norm": 0.6342041492462158,
      "learning_rate": 0.0001390572588056364,
      "loss": 1.2472,
      "step": 4023
    },
    {
      "epoch": 0.7478163910053893,
      "grad_norm": 0.6052384972572327,
      "learning_rate": 0.00013903033419161942,
      "loss": 1.1535,
      "step": 4024
    },
    {
      "epoch": 0.7480022300687604,
      "grad_norm": 0.7497447729110718,
      "learning_rate": 0.0001390034062392536,
      "loss": 1.0163,
      "step": 4025
    },
    {
      "epoch": 0.7481880691321315,
      "grad_norm": 0.5682515501976013,
      "learning_rate": 0.00013897647495084213,
      "loss": 0.935,
      "step": 4026
    },
    {
      "epoch": 0.7483739081955026,
      "grad_norm": 0.6002129912376404,
      "learning_rate": 0.00013894954032868853,
      "loss": 0.9931,
      "step": 4027
    },
    {
      "epoch": 0.7485597472588739,
      "grad_norm": 0.5856931209564209,
      "learning_rate": 0.0001389226023750966,
      "loss": 0.9945,
      "step": 4028
    },
    {
      "epoch": 0.748745586322245,
      "grad_norm": 0.6775208115577698,
      "learning_rate": 0.00013889566109237032,
      "loss": 1.2281,
      "step": 4029
    },
    {
      "epoch": 0.7489314253856161,
      "grad_norm": 0.5849632024765015,
      "learning_rate": 0.0001388687164828141,
      "loss": 1.3146,
      "step": 4030
    },
    {
      "epoch": 0.7491172644489872,
      "grad_norm": 0.6817806363105774,
      "learning_rate": 0.00013884176854873253,
      "loss": 1.1723,
      "step": 4031
    },
    {
      "epoch": 0.7493031035123583,
      "grad_norm": 0.6314253211021423,
      "learning_rate": 0.00013881481729243055,
      "loss": 0.9815,
      "step": 4032
    },
    {
      "epoch": 0.7494889425757294,
      "grad_norm": 0.5545886158943176,
      "learning_rate": 0.0001387878627162134,
      "loss": 1.1997,
      "step": 4033
    },
    {
      "epoch": 0.7496747816391005,
      "grad_norm": 0.627219557762146,
      "learning_rate": 0.00013876090482238646,
      "loss": 0.8301,
      "step": 4034
    },
    {
      "epoch": 0.7498606207024716,
      "grad_norm": 0.5215731859207153,
      "learning_rate": 0.0001387339436132556,
      "loss": 0.9639,
      "step": 4035
    },
    {
      "epoch": 0.7500464597658428,
      "grad_norm": 0.5535773634910583,
      "learning_rate": 0.00013870697909112683,
      "loss": 0.9157,
      "step": 4036
    },
    {
      "epoch": 0.7502322988292139,
      "grad_norm": 0.6636160612106323,
      "learning_rate": 0.00013868001125830648,
      "loss": 0.9123,
      "step": 4037
    },
    {
      "epoch": 0.750418137892585,
      "grad_norm": 0.5647329688072205,
      "learning_rate": 0.00013865304011710113,
      "loss": 0.8311,
      "step": 4038
    },
    {
      "epoch": 0.750418137892585,
      "eval_loss": 1.0377047061920166,
      "eval_runtime": 23.2105,
      "eval_samples_per_second": 47.048,
      "eval_steps_per_second": 23.524,
      "step": 4038
    },
    {
      "epoch": 0.7506039769559562,
      "grad_norm": 0.6448737382888794,
      "learning_rate": 0.00013862606566981774,
      "loss": 1.054,
      "step": 4039
    },
    {
      "epoch": 0.7507898160193273,
      "grad_norm": 0.6743420958518982,
      "learning_rate": 0.00013859908791876352,
      "loss": 1.2603,
      "step": 4040
    },
    {
      "epoch": 0.7509756550826984,
      "grad_norm": 0.5818408131599426,
      "learning_rate": 0.0001385721068662458,
      "loss": 1.0092,
      "step": 4041
    },
    {
      "epoch": 0.7511614941460695,
      "grad_norm": 0.6513164043426514,
      "learning_rate": 0.00013854512251457247,
      "loss": 1.0878,
      "step": 4042
    },
    {
      "epoch": 0.7513473332094406,
      "grad_norm": 0.6026800274848938,
      "learning_rate": 0.0001385181348660515,
      "loss": 0.8527,
      "step": 4043
    },
    {
      "epoch": 0.7515331722728118,
      "grad_norm": 0.6424375176429749,
      "learning_rate": 0.0001384911439229912,
      "loss": 1.0664,
      "step": 4044
    },
    {
      "epoch": 0.7517190113361829,
      "grad_norm": 0.7439508438110352,
      "learning_rate": 0.00013846414968770015,
      "loss": 1.2037,
      "step": 4045
    },
    {
      "epoch": 0.751904850399554,
      "grad_norm": 0.6531807780265808,
      "learning_rate": 0.00013843715216248727,
      "loss": 1.1257,
      "step": 4046
    },
    {
      "epoch": 0.7520906894629251,
      "grad_norm": 0.6059077978134155,
      "learning_rate": 0.0001384101513496617,
      "loss": 0.9629,
      "step": 4047
    },
    {
      "epoch": 0.7522765285262962,
      "grad_norm": 0.5011289119720459,
      "learning_rate": 0.00013838314725153286,
      "loss": 0.6582,
      "step": 4048
    },
    {
      "epoch": 0.7524623675896673,
      "grad_norm": 0.894359827041626,
      "learning_rate": 0.00013835613987041046,
      "loss": 1.2763,
      "step": 4049
    },
    {
      "epoch": 0.7526482066530384,
      "grad_norm": 0.5608694553375244,
      "learning_rate": 0.00013832912920860456,
      "loss": 1.0542,
      "step": 4050
    },
    {
      "epoch": 0.7528340457164096,
      "grad_norm": 0.6905021667480469,
      "learning_rate": 0.00013830211526842537,
      "loss": 1.0222,
      "step": 4051
    },
    {
      "epoch": 0.7530198847797807,
      "grad_norm": 0.5437237620353699,
      "learning_rate": 0.00013827509805218347,
      "loss": 1.0084,
      "step": 4052
    },
    {
      "epoch": 0.7532057238431519,
      "grad_norm": 0.665327787399292,
      "learning_rate": 0.0001382480775621897,
      "loss": 1.1029,
      "step": 4053
    },
    {
      "epoch": 0.753391562906523,
      "grad_norm": 0.6304662823677063,
      "learning_rate": 0.0001382210538007552,
      "loss": 1.3091,
      "step": 4054
    },
    {
      "epoch": 0.7535774019698941,
      "grad_norm": 0.736091136932373,
      "learning_rate": 0.00013819402677019138,
      "loss": 1.0107,
      "step": 4055
    },
    {
      "epoch": 0.7537632410332652,
      "grad_norm": 0.5813354849815369,
      "learning_rate": 0.0001381669964728099,
      "loss": 1.12,
      "step": 4056
    },
    {
      "epoch": 0.7539490800966363,
      "grad_norm": 0.629216194152832,
      "learning_rate": 0.00013813996291092275,
      "loss": 0.9777,
      "step": 4057
    },
    {
      "epoch": 0.7541349191600074,
      "grad_norm": 0.7185510993003845,
      "learning_rate": 0.0001381129260868421,
      "loss": 1.0516,
      "step": 4058
    },
    {
      "epoch": 0.7543207582233785,
      "grad_norm": 0.5125520825386047,
      "learning_rate": 0.00013808588600288052,
      "loss": 1.0353,
      "step": 4059
    },
    {
      "epoch": 0.7545065972867496,
      "grad_norm": 0.5914701819419861,
      "learning_rate": 0.00013805884266135078,
      "loss": 1.0866,
      "step": 4060
    },
    {
      "epoch": 0.7546924363501208,
      "grad_norm": 0.5655017495155334,
      "learning_rate": 0.00013803179606456599,
      "loss": 1.0189,
      "step": 4061
    },
    {
      "epoch": 0.754878275413492,
      "grad_norm": 0.5555375814437866,
      "learning_rate": 0.0001380047462148395,
      "loss": 1.0671,
      "step": 4062
    },
    {
      "epoch": 0.7550641144768631,
      "grad_norm": 0.5526087284088135,
      "learning_rate": 0.0001379776931144849,
      "loss": 0.9678,
      "step": 4063
    },
    {
      "epoch": 0.7552499535402342,
      "grad_norm": 0.6122016906738281,
      "learning_rate": 0.00013795063676581612,
      "loss": 1.1348,
      "step": 4064
    },
    {
      "epoch": 0.7554357926036053,
      "grad_norm": 0.5869781970977783,
      "learning_rate": 0.00013792357717114738,
      "loss": 1.0977,
      "step": 4065
    },
    {
      "epoch": 0.7556216316669764,
      "grad_norm": 0.619445264339447,
      "learning_rate": 0.0001378965143327931,
      "loss": 1.1654,
      "step": 4066
    },
    {
      "epoch": 0.7558074707303475,
      "grad_norm": 0.7364014387130737,
      "learning_rate": 0.00013786944825306806,
      "loss": 1.3876,
      "step": 4067
    },
    {
      "epoch": 0.7559933097937186,
      "grad_norm": 0.5614713430404663,
      "learning_rate": 0.00013784237893428724,
      "loss": 1.0382,
      "step": 4068
    },
    {
      "epoch": 0.7561791488570898,
      "grad_norm": 0.7402245402336121,
      "learning_rate": 0.00013781530637876598,
      "loss": 1.1889,
      "step": 4069
    },
    {
      "epoch": 0.7563649879204609,
      "grad_norm": 0.5417977571487427,
      "learning_rate": 0.00013778823058881984,
      "loss": 0.9613,
      "step": 4070
    },
    {
      "epoch": 0.756550826983832,
      "grad_norm": 0.6280809640884399,
      "learning_rate": 0.00013776115156676467,
      "loss": 0.9137,
      "step": 4071
    },
    {
      "epoch": 0.7567366660472031,
      "grad_norm": 0.5001559257507324,
      "learning_rate": 0.00013773406931491658,
      "loss": 0.9083,
      "step": 4072
    },
    {
      "epoch": 0.7569225051105742,
      "grad_norm": 0.7508971691131592,
      "learning_rate": 0.00013770698383559201,
      "loss": 1.0583,
      "step": 4073
    },
    {
      "epoch": 0.7571083441739453,
      "grad_norm": 0.642376720905304,
      "learning_rate": 0.0001376798951311076,
      "loss": 1.1538,
      "step": 4074
    },
    {
      "epoch": 0.7572941832373165,
      "grad_norm": 0.5976815223693848,
      "learning_rate": 0.00013765280320378033,
      "loss": 0.8483,
      "step": 4075
    },
    {
      "epoch": 0.7574800223006876,
      "grad_norm": 0.6116734743118286,
      "learning_rate": 0.00013762570805592742,
      "loss": 1.204,
      "step": 4076
    },
    {
      "epoch": 0.7576658613640588,
      "grad_norm": 0.7391072511672974,
      "learning_rate": 0.00013759860968986636,
      "loss": 1.0961,
      "step": 4077
    },
    {
      "epoch": 0.7578517004274299,
      "grad_norm": 0.6154507398605347,
      "learning_rate": 0.00013757150810791497,
      "loss": 1.2466,
      "step": 4078
    },
    {
      "epoch": 0.758037539490801,
      "grad_norm": 0.5805959105491638,
      "learning_rate": 0.00013754440331239132,
      "loss": 1.2261,
      "step": 4079
    },
    {
      "epoch": 0.7582233785541721,
      "grad_norm": 0.6683850288391113,
      "learning_rate": 0.00013751729530561368,
      "loss": 1.1683,
      "step": 4080
    },
    {
      "epoch": 0.7584092176175432,
      "grad_norm": 0.5736557245254517,
      "learning_rate": 0.0001374901840899007,
      "loss": 0.6839,
      "step": 4081
    },
    {
      "epoch": 0.7585950566809143,
      "grad_norm": 0.599035918712616,
      "learning_rate": 0.0001374630696675712,
      "loss": 0.8846,
      "step": 4082
    },
    {
      "epoch": 0.7587808957442854,
      "grad_norm": 0.6209955215454102,
      "learning_rate": 0.0001374359520409444,
      "loss": 1.1019,
      "step": 4083
    },
    {
      "epoch": 0.7589667348076565,
      "grad_norm": 0.5841337442398071,
      "learning_rate": 0.00013740883121233973,
      "loss": 1.1662,
      "step": 4084
    },
    {
      "epoch": 0.7591525738710276,
      "grad_norm": 0.6548193693161011,
      "learning_rate": 0.00013738170718407687,
      "loss": 1.0279,
      "step": 4085
    },
    {
      "epoch": 0.7593384129343989,
      "grad_norm": 0.5952484011650085,
      "learning_rate": 0.00013735457995847582,
      "loss": 1.1303,
      "step": 4086
    },
    {
      "epoch": 0.75952425199777,
      "grad_norm": 0.6196330189704895,
      "learning_rate": 0.0001373274495378568,
      "loss": 1.3022,
      "step": 4087
    },
    {
      "epoch": 0.7597100910611411,
      "grad_norm": 0.5573399066925049,
      "learning_rate": 0.00013730031592454036,
      "loss": 0.8631,
      "step": 4088
    },
    {
      "epoch": 0.7598959301245122,
      "grad_norm": 0.48187312483787537,
      "learning_rate": 0.00013727317912084726,
      "loss": 0.7207,
      "step": 4089
    },
    {
      "epoch": 0.7600817691878833,
      "grad_norm": 0.5117285847663879,
      "learning_rate": 0.00013724603912909858,
      "loss": 0.9147,
      "step": 4090
    },
    {
      "epoch": 0.7602676082512544,
      "grad_norm": 0.6108545660972595,
      "learning_rate": 0.0001372188959516157,
      "loss": 1.0902,
      "step": 4091
    },
    {
      "epoch": 0.7604534473146255,
      "grad_norm": 0.5700968503952026,
      "learning_rate": 0.00013719174959072022,
      "loss": 1.1427,
      "step": 4092
    },
    {
      "epoch": 0.7606392863779966,
      "grad_norm": 0.6507374048233032,
      "learning_rate": 0.00013716460004873398,
      "loss": 0.9902,
      "step": 4093
    },
    {
      "epoch": 0.7608251254413678,
      "grad_norm": 0.5942047834396362,
      "learning_rate": 0.00013713744732797922,
      "loss": 0.7129,
      "step": 4094
    },
    {
      "epoch": 0.7610109645047389,
      "grad_norm": 0.7793191075325012,
      "learning_rate": 0.0001371102914307783,
      "loss": 0.9558,
      "step": 4095
    },
    {
      "epoch": 0.76119680356811,
      "grad_norm": 0.5532881021499634,
      "learning_rate": 0.0001370831323594539,
      "loss": 0.9851,
      "step": 4096
    },
    {
      "epoch": 0.7613826426314811,
      "grad_norm": 0.6629320979118347,
      "learning_rate": 0.0001370559701163291,
      "loss": 0.9608,
      "step": 4097
    },
    {
      "epoch": 0.7615684816948523,
      "grad_norm": 0.6066349744796753,
      "learning_rate": 0.00013702880470372708,
      "loss": 1.1944,
      "step": 4098
    },
    {
      "epoch": 0.7617543207582234,
      "grad_norm": 0.5362082719802856,
      "learning_rate": 0.00013700163612397135,
      "loss": 0.9813,
      "step": 4099
    },
    {
      "epoch": 0.7619401598215945,
      "grad_norm": 0.6526452302932739,
      "learning_rate": 0.00013697446437938571,
      "loss": 1.29,
      "step": 4100
    },
    {
      "epoch": 0.7621259988849656,
      "grad_norm": 0.553182065486908,
      "learning_rate": 0.00013694728947229425,
      "loss": 1.1332,
      "step": 4101
    },
    {
      "epoch": 0.7623118379483368,
      "grad_norm": 0.5968033671379089,
      "learning_rate": 0.00013692011140502126,
      "loss": 1.0957,
      "step": 4102
    },
    {
      "epoch": 0.7624976770117079,
      "grad_norm": 0.7517340779304504,
      "learning_rate": 0.0001368929301798913,
      "loss": 1.2294,
      "step": 4103
    },
    {
      "epoch": 0.762683516075079,
      "grad_norm": 0.5200783014297485,
      "learning_rate": 0.00013686574579922935,
      "loss": 1.1681,
      "step": 4104
    },
    {
      "epoch": 0.7628693551384501,
      "grad_norm": 0.6289031505584717,
      "learning_rate": 0.00013683855826536042,
      "loss": 1.1485,
      "step": 4105
    },
    {
      "epoch": 0.7630551942018212,
      "grad_norm": 0.6423609256744385,
      "learning_rate": 0.00013681136758061005,
      "loss": 1.1549,
      "step": 4106
    },
    {
      "epoch": 0.7632410332651923,
      "grad_norm": 0.7103132605552673,
      "learning_rate": 0.0001367841737473038,
      "loss": 0.9711,
      "step": 4107
    },
    {
      "epoch": 0.7634268723285634,
      "grad_norm": 0.553540050983429,
      "learning_rate": 0.00013675697676776772,
      "loss": 1.1464,
      "step": 4108
    },
    {
      "epoch": 0.7636127113919345,
      "grad_norm": 0.5573582053184509,
      "learning_rate": 0.00013672977664432794,
      "loss": 1.0289,
      "step": 4109
    },
    {
      "epoch": 0.7637985504553058,
      "grad_norm": 0.6105955243110657,
      "learning_rate": 0.00013670257337931093,
      "loss": 1.1219,
      "step": 4110
    },
    {
      "epoch": 0.7639843895186769,
      "grad_norm": 0.5916879177093506,
      "learning_rate": 0.00013667536697504356,
      "loss": 1.1944,
      "step": 4111
    },
    {
      "epoch": 0.764170228582048,
      "grad_norm": 0.5850638151168823,
      "learning_rate": 0.00013664815743385278,
      "loss": 1.1136,
      "step": 4112
    },
    {
      "epoch": 0.7643560676454191,
      "grad_norm": 0.6730114817619324,
      "learning_rate": 0.0001366209447580659,
      "loss": 1.3724,
      "step": 4113
    },
    {
      "epoch": 0.7645419067087902,
      "grad_norm": 0.5630990266799927,
      "learning_rate": 0.00013659372895001042,
      "loss": 0.9657,
      "step": 4114
    },
    {
      "epoch": 0.7647277457721613,
      "grad_norm": 3.9506986141204834,
      "learning_rate": 0.00013656651001201425,
      "loss": 2.3019,
      "step": 4115
    },
    {
      "epoch": 0.7649135848355324,
      "grad_norm": 0.5540075898170471,
      "learning_rate": 0.00013653928794640544,
      "loss": 0.9883,
      "step": 4116
    },
    {
      "epoch": 0.7650994238989035,
      "grad_norm": 0.6030392050743103,
      "learning_rate": 0.00013651206275551233,
      "loss": 1.0255,
      "step": 4117
    },
    {
      "epoch": 0.7652852629622746,
      "grad_norm": 0.6220700144767761,
      "learning_rate": 0.0001364848344416636,
      "loss": 1.1406,
      "step": 4118
    },
    {
      "epoch": 0.7654711020256458,
      "grad_norm": 0.5235512256622314,
      "learning_rate": 0.00013645760300718814,
      "loss": 1.1161,
      "step": 4119
    },
    {
      "epoch": 0.7656569410890169,
      "grad_norm": 0.5732916593551636,
      "learning_rate": 0.0001364303684544151,
      "loss": 1.074,
      "step": 4120
    },
    {
      "epoch": 0.765842780152388,
      "grad_norm": 0.5863698124885559,
      "learning_rate": 0.0001364031307856739,
      "loss": 1.1758,
      "step": 4121
    },
    {
      "epoch": 0.7660286192157592,
      "grad_norm": 0.500783383846283,
      "learning_rate": 0.00013637589000329428,
      "loss": 1.0494,
      "step": 4122
    },
    {
      "epoch": 0.7662144582791303,
      "grad_norm": 0.5018875002861023,
      "learning_rate": 0.00013634864610960614,
      "loss": 1.0457,
      "step": 4123
    },
    {
      "epoch": 0.7664002973425014,
      "grad_norm": 0.6175665855407715,
      "learning_rate": 0.0001363213991069397,
      "loss": 1.0481,
      "step": 4124
    },
    {
      "epoch": 0.7665861364058725,
      "grad_norm": 0.5728722214698792,
      "learning_rate": 0.00013629414899762554,
      "loss": 1.0837,
      "step": 4125
    },
    {
      "epoch": 0.7667719754692436,
      "grad_norm": 0.5949525237083435,
      "learning_rate": 0.00013626689578399438,
      "loss": 0.9858,
      "step": 4126
    },
    {
      "epoch": 0.7669578145326148,
      "grad_norm": 0.6486298441886902,
      "learning_rate": 0.00013623963946837722,
      "loss": 0.849,
      "step": 4127
    },
    {
      "epoch": 0.7671436535959859,
      "grad_norm": 0.6250497102737427,
      "learning_rate": 0.00013621238005310536,
      "loss": 1.2374,
      "step": 4128
    },
    {
      "epoch": 0.767329492659357,
      "grad_norm": 0.6734606027603149,
      "learning_rate": 0.0001361851175405104,
      "loss": 1.2558,
      "step": 4129
    },
    {
      "epoch": 0.7675153317227281,
      "grad_norm": 0.811181902885437,
      "learning_rate": 0.0001361578519329241,
      "loss": 1.2633,
      "step": 4130
    },
    {
      "epoch": 0.7677011707860992,
      "grad_norm": 0.6883793473243713,
      "learning_rate": 0.0001361305832326786,
      "loss": 1.2465,
      "step": 4131
    },
    {
      "epoch": 0.7678870098494703,
      "grad_norm": 0.5060800313949585,
      "learning_rate": 0.00013610331144210623,
      "loss": 0.9639,
      "step": 4132
    },
    {
      "epoch": 0.7680728489128414,
      "grad_norm": 0.49882808327674866,
      "learning_rate": 0.0001360760365635396,
      "loss": 0.9654,
      "step": 4133
    },
    {
      "epoch": 0.7682586879762126,
      "grad_norm": 0.6372555494308472,
      "learning_rate": 0.00013604875859931156,
      "loss": 1.1524,
      "step": 4134
    },
    {
      "epoch": 0.7684445270395838,
      "grad_norm": 0.48308369517326355,
      "learning_rate": 0.00013602147755175533,
      "loss": 0.5496,
      "step": 4135
    },
    {
      "epoch": 0.7686303661029549,
      "grad_norm": 0.6130794286727905,
      "learning_rate": 0.00013599419342320425,
      "loss": 0.7401,
      "step": 4136
    },
    {
      "epoch": 0.768816205166326,
      "grad_norm": 0.6539893746376038,
      "learning_rate": 0.000135966906215992,
      "loss": 1.0096,
      "step": 4137
    },
    {
      "epoch": 0.7690020442296971,
      "grad_norm": 0.537489652633667,
      "learning_rate": 0.00013593961593245254,
      "loss": 1.0363,
      "step": 4138
    },
    {
      "epoch": 0.7691878832930682,
      "grad_norm": 0.6301590204238892,
      "learning_rate": 0.00013591232257492009,
      "loss": 0.9674,
      "step": 4139
    },
    {
      "epoch": 0.7693737223564393,
      "grad_norm": 0.5909221768379211,
      "learning_rate": 0.00013588502614572903,
      "loss": 1.0572,
      "step": 4140
    },
    {
      "epoch": 0.7695595614198104,
      "grad_norm": 0.7259015440940857,
      "learning_rate": 0.00013585772664721413,
      "loss": 1.2155,
      "step": 4141
    },
    {
      "epoch": 0.7697454004831815,
      "grad_norm": 0.59591144323349,
      "learning_rate": 0.00013583042408171037,
      "loss": 0.8733,
      "step": 4142
    },
    {
      "epoch": 0.7699312395465527,
      "grad_norm": 1.9292752742767334,
      "learning_rate": 0.00013580311845155302,
      "loss": 2.0443,
      "step": 4143
    },
    {
      "epoch": 0.7701170786099238,
      "grad_norm": 0.5641506910324097,
      "learning_rate": 0.00013577580975907756,
      "loss": 1.0628,
      "step": 4144
    },
    {
      "epoch": 0.770302917673295,
      "grad_norm": 0.6879433989524841,
      "learning_rate": 0.00013574849800661976,
      "loss": 1.5132,
      "step": 4145
    },
    {
      "epoch": 0.7704887567366661,
      "grad_norm": 0.5869988799095154,
      "learning_rate": 0.00013572118319651565,
      "loss": 1.0678,
      "step": 4146
    },
    {
      "epoch": 0.7706745958000372,
      "grad_norm": 0.5482993721961975,
      "learning_rate": 0.00013569386533110155,
      "loss": 0.9921,
      "step": 4147
    },
    {
      "epoch": 0.7708604348634083,
      "grad_norm": 0.6312058568000793,
      "learning_rate": 0.000135666544412714,
      "loss": 1.0242,
      "step": 4148
    },
    {
      "epoch": 0.7710462739267794,
      "grad_norm": 0.6670210361480713,
      "learning_rate": 0.00013563922044368984,
      "loss": 1.0649,
      "step": 4149
    },
    {
      "epoch": 0.7712321129901505,
      "grad_norm": 0.6085960865020752,
      "learning_rate": 0.00013561189342636615,
      "loss": 0.9379,
      "step": 4150
    },
    {
      "epoch": 0.7714179520535217,
      "grad_norm": 0.5336554050445557,
      "learning_rate": 0.0001355845633630802,
      "loss": 0.869,
      "step": 4151
    },
    {
      "epoch": 0.7716037911168928,
      "grad_norm": 0.5766181945800781,
      "learning_rate": 0.00013555723025616964,
      "loss": 0.9011,
      "step": 4152
    },
    {
      "epoch": 0.7717896301802639,
      "grad_norm": 0.589514434337616,
      "learning_rate": 0.00013552989410797235,
      "loss": 1.0555,
      "step": 4153
    },
    {
      "epoch": 0.771975469243635,
      "grad_norm": 0.6114048361778259,
      "learning_rate": 0.00013550255492082643,
      "loss": 1.1759,
      "step": 4154
    },
    {
      "epoch": 0.7721613083070061,
      "grad_norm": 0.5874496698379517,
      "learning_rate": 0.00013547521269707024,
      "loss": 1.1696,
      "step": 4155
    },
    {
      "epoch": 0.7723471473703772,
      "grad_norm": 0.6765058040618896,
      "learning_rate": 0.0001354478674390424,
      "loss": 1.0971,
      "step": 4156
    },
    {
      "epoch": 0.7725329864337483,
      "grad_norm": 0.5729826092720032,
      "learning_rate": 0.0001354205191490819,
      "loss": 1.1221,
      "step": 4157
    },
    {
      "epoch": 0.7727188254971195,
      "grad_norm": 0.6108171343803406,
      "learning_rate": 0.00013539316782952785,
      "loss": 0.9023,
      "step": 4158
    },
    {
      "epoch": 0.7729046645604906,
      "grad_norm": 0.4902413487434387,
      "learning_rate": 0.00013536581348271958,
      "loss": 0.7808,
      "step": 4159
    },
    {
      "epoch": 0.7730905036238618,
      "grad_norm": 0.5325641632080078,
      "learning_rate": 0.00013533845611099694,
      "loss": 0.8381,
      "step": 4160
    },
    {
      "epoch": 0.7732763426872329,
      "grad_norm": 0.8668013215065002,
      "learning_rate": 0.00013531109571669972,
      "loss": 1.2558,
      "step": 4161
    },
    {
      "epoch": 0.773462181750604,
      "grad_norm": 0.6640474796295166,
      "learning_rate": 0.0001352837323021682,
      "loss": 1.246,
      "step": 4162
    },
    {
      "epoch": 0.7736480208139751,
      "grad_norm": 0.6278790235519409,
      "learning_rate": 0.00013525636586974278,
      "loss": 0.9943,
      "step": 4163
    },
    {
      "epoch": 0.7738338598773462,
      "grad_norm": 0.5972362756729126,
      "learning_rate": 0.00013522899642176418,
      "loss": 1.1249,
      "step": 4164
    },
    {
      "epoch": 0.7740196989407173,
      "grad_norm": 0.4734021723270416,
      "learning_rate": 0.00013520162396057342,
      "loss": 0.8517,
      "step": 4165
    },
    {
      "epoch": 0.7742055380040884,
      "grad_norm": 0.6646032333374023,
      "learning_rate": 0.00013517424848851162,
      "loss": 1.387,
      "step": 4166
    },
    {
      "epoch": 0.7743913770674595,
      "grad_norm": 0.6340314745903015,
      "learning_rate": 0.00013514687000792036,
      "loss": 1.1976,
      "step": 4167
    },
    {
      "epoch": 0.7745772161308307,
      "grad_norm": 0.4301804304122925,
      "learning_rate": 0.00013511948852114132,
      "loss": 0.5702,
      "step": 4168
    },
    {
      "epoch": 0.7747630551942019,
      "grad_norm": 0.5795063972473145,
      "learning_rate": 0.00013509210403051656,
      "loss": 1.0796,
      "step": 4169
    },
    {
      "epoch": 0.774948894257573,
      "grad_norm": 0.6031147241592407,
      "learning_rate": 0.00013506471653838825,
      "loss": 1.2061,
      "step": 4170
    },
    {
      "epoch": 0.7751347333209441,
      "grad_norm": 0.6553006172180176,
      "learning_rate": 0.000135037326047099,
      "loss": 0.9015,
      "step": 4171
    },
    {
      "epoch": 0.7753205723843152,
      "grad_norm": 0.7210426330566406,
      "learning_rate": 0.00013500993255899152,
      "loss": 1.1031,
      "step": 4172
    },
    {
      "epoch": 0.7755064114476863,
      "grad_norm": 0.5744313597679138,
      "learning_rate": 0.0001349825360764088,
      "loss": 0.9968,
      "step": 4173
    },
    {
      "epoch": 0.7756922505110574,
      "grad_norm": 0.7675306797027588,
      "learning_rate": 0.0001349551366016942,
      "loss": 1.0349,
      "step": 4174
    },
    {
      "epoch": 0.7758780895744285,
      "grad_norm": 0.6847807765007019,
      "learning_rate": 0.00013492773413719121,
      "loss": 0.9701,
      "step": 4175
    },
    {
      "epoch": 0.7760639286377997,
      "grad_norm": 0.6042717695236206,
      "learning_rate": 0.00013490032868524365,
      "loss": 1.1616,
      "step": 4176
    },
    {
      "epoch": 0.7762497677011708,
      "grad_norm": 0.6150924563407898,
      "learning_rate": 0.00013487292024819554,
      "loss": 0.8984,
      "step": 4177
    },
    {
      "epoch": 0.7764356067645419,
      "grad_norm": 0.5927572250366211,
      "learning_rate": 0.00013484550882839115,
      "loss": 1.2649,
      "step": 4178
    },
    {
      "epoch": 0.776621445827913,
      "grad_norm": 0.5676189064979553,
      "learning_rate": 0.00013481809442817511,
      "loss": 0.979,
      "step": 4179
    },
    {
      "epoch": 0.7768072848912841,
      "grad_norm": 0.48436301946640015,
      "learning_rate": 0.0001347906770498922,
      "loss": 0.8431,
      "step": 4180
    },
    {
      "epoch": 0.7769931239546553,
      "grad_norm": 0.6204550862312317,
      "learning_rate": 0.00013476325669588748,
      "loss": 1.1561,
      "step": 4181
    },
    {
      "epoch": 0.7771789630180264,
      "grad_norm": 0.512209415435791,
      "learning_rate": 0.00013473583336850631,
      "loss": 1.0287,
      "step": 4182
    },
    {
      "epoch": 0.7773648020813975,
      "grad_norm": 0.6418192386627197,
      "learning_rate": 0.00013470840707009422,
      "loss": 1.1377,
      "step": 4183
    },
    {
      "epoch": 0.7775506411447687,
      "grad_norm": 0.6207935214042664,
      "learning_rate": 0.00013468097780299708,
      "loss": 0.9253,
      "step": 4184
    },
    {
      "epoch": 0.7777364802081398,
      "grad_norm": 0.6537068486213684,
      "learning_rate": 0.00013465354556956092,
      "loss": 1.3023,
      "step": 4185
    },
    {
      "epoch": 0.7779223192715109,
      "grad_norm": 0.5210212469100952,
      "learning_rate": 0.0001346261103721321,
      "loss": 0.9315,
      "step": 4186
    },
    {
      "epoch": 0.778108158334882,
      "grad_norm": 0.4942516088485718,
      "learning_rate": 0.00013459867221305723,
      "loss": 0.8469,
      "step": 4187
    },
    {
      "epoch": 0.7782939973982531,
      "grad_norm": 0.6087897419929504,
      "learning_rate": 0.00013457123109468318,
      "loss": 0.954,
      "step": 4188
    },
    {
      "epoch": 0.7784798364616242,
      "grad_norm": 0.5794795155525208,
      "learning_rate": 0.000134543787019357,
      "loss": 1.1739,
      "step": 4189
    },
    {
      "epoch": 0.7786656755249953,
      "grad_norm": 0.6037955284118652,
      "learning_rate": 0.00013451633998942607,
      "loss": 1.064,
      "step": 4190
    },
    {
      "epoch": 0.7788515145883664,
      "grad_norm": 0.8131178021430969,
      "learning_rate": 0.00013448889000723796,
      "loss": 1.0455,
      "step": 4191
    },
    {
      "epoch": 0.7790373536517375,
      "grad_norm": 0.6027346253395081,
      "learning_rate": 0.00013446143707514056,
      "loss": 1.0113,
      "step": 4192
    },
    {
      "epoch": 0.7792231927151088,
      "grad_norm": 0.6223542094230652,
      "learning_rate": 0.00013443398119548195,
      "loss": 0.9856,
      "step": 4193
    },
    {
      "epoch": 0.7794090317784799,
      "grad_norm": 0.5715983510017395,
      "learning_rate": 0.0001344065223706105,
      "loss": 1.0353,
      "step": 4194
    },
    {
      "epoch": 0.779594870841851,
      "grad_norm": 0.5402487516403198,
      "learning_rate": 0.00013437906060287486,
      "loss": 1.0995,
      "step": 4195
    },
    {
      "epoch": 0.7797807099052221,
      "grad_norm": 0.6275144815444946,
      "learning_rate": 0.00013435159589462385,
      "loss": 1.2073,
      "step": 4196
    },
    {
      "epoch": 0.7799665489685932,
      "grad_norm": 0.6028453707695007,
      "learning_rate": 0.0001343241282482066,
      "loss": 1.2034,
      "step": 4197
    },
    {
      "epoch": 0.7801523880319643,
      "grad_norm": 0.6037322878837585,
      "learning_rate": 0.0001342966576659725,
      "loss": 0.9155,
      "step": 4198
    },
    {
      "epoch": 0.7803382270953354,
      "grad_norm": 0.5991843938827515,
      "learning_rate": 0.0001342691841502711,
      "loss": 0.9529,
      "step": 4199
    },
    {
      "epoch": 0.7805240661587065,
      "grad_norm": 0.6946239471435547,
      "learning_rate": 0.00013424170770345235,
      "loss": 1.349,
      "step": 4200
    },
    {
      "epoch": 0.7807099052220777,
      "grad_norm": 0.5432325601577759,
      "learning_rate": 0.0001342142283278663,
      "loss": 0.9882,
      "step": 4201
    },
    {
      "epoch": 0.7808957442854488,
      "grad_norm": 0.9075375199317932,
      "learning_rate": 0.00013418674602586338,
      "loss": 1.2515,
      "step": 4202
    },
    {
      "epoch": 0.7810815833488199,
      "grad_norm": 0.599430501461029,
      "learning_rate": 0.00013415926079979418,
      "loss": 0.9525,
      "step": 4203
    },
    {
      "epoch": 0.781267422412191,
      "grad_norm": 0.6083348989486694,
      "learning_rate": 0.00013413177265200957,
      "loss": 1.0576,
      "step": 4204
    },
    {
      "epoch": 0.7814532614755622,
      "grad_norm": 0.5876964926719666,
      "learning_rate": 0.0001341042815848607,
      "loss": 1.0482,
      "step": 4205
    },
    {
      "epoch": 0.7816391005389333,
      "grad_norm": 0.8127071857452393,
      "learning_rate": 0.00013407678760069891,
      "loss": 0.9777,
      "step": 4206
    },
    {
      "epoch": 0.7818249396023044,
      "grad_norm": 0.6084738969802856,
      "learning_rate": 0.00013404929070187578,
      "loss": 0.9079,
      "step": 4207
    },
    {
      "epoch": 0.7820107786656755,
      "grad_norm": 0.5878585577011108,
      "learning_rate": 0.00013402179089074326,
      "loss": 1.2447,
      "step": 4208
    },
    {
      "epoch": 0.7821966177290467,
      "grad_norm": 0.6077066659927368,
      "learning_rate": 0.00013399428816965344,
      "loss": 1.2663,
      "step": 4209
    },
    {
      "epoch": 0.7823824567924178,
      "grad_norm": 0.7702139616012573,
      "learning_rate": 0.0001339667825409587,
      "loss": 1.3534,
      "step": 4210
    },
    {
      "epoch": 0.7825682958557889,
      "grad_norm": 0.6338496208190918,
      "learning_rate": 0.0001339392740070116,
      "loss": 1.0933,
      "step": 4211
    },
    {
      "epoch": 0.78275413491916,
      "grad_norm": 0.6644789576530457,
      "learning_rate": 0.0001339117625701651,
      "loss": 1.2102,
      "step": 4212
    },
    {
      "epoch": 0.7829399739825311,
      "grad_norm": 0.6646575927734375,
      "learning_rate": 0.0001338842482327722,
      "loss": 1.1216,
      "step": 4213
    },
    {
      "epoch": 0.7831258130459022,
      "grad_norm": 0.6456634998321533,
      "learning_rate": 0.0001338567309971863,
      "loss": 0.988,
      "step": 4214
    },
    {
      "epoch": 0.7833116521092733,
      "grad_norm": 0.625597357749939,
      "learning_rate": 0.00013382921086576105,
      "loss": 1.2121,
      "step": 4215
    },
    {
      "epoch": 0.7834974911726444,
      "grad_norm": 0.5920417308807373,
      "learning_rate": 0.00013380168784085027,
      "loss": 1.1116,
      "step": 4216
    },
    {
      "epoch": 0.7836833302360157,
      "grad_norm": 0.7123020887374878,
      "learning_rate": 0.00013377416192480809,
      "loss": 1.0922,
      "step": 4217
    },
    {
      "epoch": 0.7838691692993868,
      "grad_norm": 0.5441176295280457,
      "learning_rate": 0.00013374663311998885,
      "loss": 1.0325,
      "step": 4218
    },
    {
      "epoch": 0.7840550083627579,
      "grad_norm": 0.5781022906303406,
      "learning_rate": 0.00013371910142874715,
      "loss": 1.0484,
      "step": 4219
    },
    {
      "epoch": 0.784240847426129,
      "grad_norm": 0.6541292667388916,
      "learning_rate": 0.0001336915668534378,
      "loss": 1.1349,
      "step": 4220
    },
    {
      "epoch": 0.7844266864895001,
      "grad_norm": 0.6895448565483093,
      "learning_rate": 0.00013366402939641594,
      "loss": 1.1096,
      "step": 4221
    },
    {
      "epoch": 0.7846125255528712,
      "grad_norm": 0.5254290699958801,
      "learning_rate": 0.00013363648906003687,
      "loss": 0.9549,
      "step": 4222
    },
    {
      "epoch": 0.7847983646162423,
      "grad_norm": 0.5432946681976318,
      "learning_rate": 0.0001336089458466562,
      "loss": 1.0367,
      "step": 4223
    },
    {
      "epoch": 0.7849842036796134,
      "grad_norm": 0.5595942735671997,
      "learning_rate": 0.00013358139975862976,
      "loss": 1.0784,
      "step": 4224
    },
    {
      "epoch": 0.7851700427429846,
      "grad_norm": 0.5201842784881592,
      "learning_rate": 0.00013355385079831362,
      "loss": 0.9406,
      "step": 4225
    },
    {
      "epoch": 0.7853558818063557,
      "grad_norm": 0.5573832988739014,
      "learning_rate": 0.00013352629896806409,
      "loss": 1.1733,
      "step": 4226
    },
    {
      "epoch": 0.7855417208697268,
      "grad_norm": 0.6541811227798462,
      "learning_rate": 0.00013349874427023777,
      "loss": 1.1657,
      "step": 4227
    },
    {
      "epoch": 0.785727559933098,
      "grad_norm": 0.5593380928039551,
      "learning_rate": 0.00013347118670719142,
      "loss": 0.8143,
      "step": 4228
    },
    {
      "epoch": 0.7859133989964691,
      "grad_norm": 0.6524107456207275,
      "learning_rate": 0.00013344362628128214,
      "loss": 0.9966,
      "step": 4229
    },
    {
      "epoch": 0.7860992380598402,
      "grad_norm": 0.5087578296661377,
      "learning_rate": 0.00013341606299486721,
      "loss": 0.9258,
      "step": 4230
    },
    {
      "epoch": 0.7862850771232113,
      "grad_norm": 0.5728389620780945,
      "learning_rate": 0.00013338849685030424,
      "loss": 1.0884,
      "step": 4231
    },
    {
      "epoch": 0.7864709161865824,
      "grad_norm": 0.7049938440322876,
      "learning_rate": 0.00013336092784995093,
      "loss": 1.2024,
      "step": 4232
    },
    {
      "epoch": 0.7866567552499535,
      "grad_norm": 0.5433733463287354,
      "learning_rate": 0.0001333333559961654,
      "loss": 0.9988,
      "step": 4233
    },
    {
      "epoch": 0.7868425943133247,
      "grad_norm": 0.5832632184028625,
      "learning_rate": 0.00013330578129130583,
      "loss": 1.0733,
      "step": 4234
    },
    {
      "epoch": 0.7870284333766958,
      "grad_norm": 0.48426976799964905,
      "learning_rate": 0.0001332782037377309,
      "loss": 0.778,
      "step": 4235
    },
    {
      "epoch": 0.7872142724400669,
      "grad_norm": 0.6360462307929993,
      "learning_rate": 0.00013325062333779917,
      "loss": 1.2133,
      "step": 4236
    },
    {
      "epoch": 0.787400111503438,
      "grad_norm": 0.6086122393608093,
      "learning_rate": 0.0001332230400938698,
      "loss": 1.0383,
      "step": 4237
    },
    {
      "epoch": 0.7875859505668091,
      "grad_norm": 0.5904986262321472,
      "learning_rate": 0.000133195454008302,
      "loss": 1.2318,
      "step": 4238
    },
    {
      "epoch": 0.7877717896301802,
      "grad_norm": 0.9779626131057739,
      "learning_rate": 0.00013316786508345532,
      "loss": 1.4215,
      "step": 4239
    },
    {
      "epoch": 0.7879576286935513,
      "grad_norm": 0.5626901388168335,
      "learning_rate": 0.0001331402733216894,
      "loss": 1.0571,
      "step": 4240
    },
    {
      "epoch": 0.7881434677569225,
      "grad_norm": 0.5035316348075867,
      "learning_rate": 0.0001331126787253643,
      "loss": 0.9109,
      "step": 4241
    },
    {
      "epoch": 0.7883293068202937,
      "grad_norm": 0.6667996644973755,
      "learning_rate": 0.00013308508129684024,
      "loss": 1.2582,
      "step": 4242
    },
    {
      "epoch": 0.7885151458836648,
      "grad_norm": 0.5744646191596985,
      "learning_rate": 0.00013305748103847764,
      "loss": 1.1911,
      "step": 4243
    },
    {
      "epoch": 0.7887009849470359,
      "grad_norm": 0.6686093211174011,
      "learning_rate": 0.00013302987795263724,
      "loss": 1.2849,
      "step": 4244
    },
    {
      "epoch": 0.788886824010407,
      "grad_norm": 0.5513818860054016,
      "learning_rate": 0.00013300227204168002,
      "loss": 0.836,
      "step": 4245
    },
    {
      "epoch": 0.7890726630737781,
      "grad_norm": 0.4881809949874878,
      "learning_rate": 0.0001329746633079671,
      "loss": 0.7554,
      "step": 4246
    },
    {
      "epoch": 0.7892585021371492,
      "grad_norm": 0.5863307118415833,
      "learning_rate": 0.00013294705175386003,
      "loss": 1.2168,
      "step": 4247
    },
    {
      "epoch": 0.7894443412005203,
      "grad_norm": 0.5727426409721375,
      "learning_rate": 0.00013291943738172037,
      "loss": 1.2246,
      "step": 4248
    },
    {
      "epoch": 0.7896301802638914,
      "grad_norm": 0.562233567237854,
      "learning_rate": 0.00013289182019391008,
      "loss": 0.9127,
      "step": 4249
    },
    {
      "epoch": 0.7898160193272626,
      "grad_norm": 0.7686207890510559,
      "learning_rate": 0.0001328642001927913,
      "loss": 1.1063,
      "step": 4250
    },
    {
      "epoch": 0.7900018583906337,
      "grad_norm": 0.5708333849906921,
      "learning_rate": 0.00013283657738072648,
      "loss": 1.1547,
      "step": 4251
    },
    {
      "epoch": 0.7901876974540049,
      "grad_norm": 0.6480849981307983,
      "learning_rate": 0.0001328089517600782,
      "loss": 1.0502,
      "step": 4252
    },
    {
      "epoch": 0.790373536517376,
      "grad_norm": 0.5901347398757935,
      "learning_rate": 0.0001327813233332094,
      "loss": 1.0596,
      "step": 4253
    },
    {
      "epoch": 0.7905593755807471,
      "grad_norm": 0.597624659538269,
      "learning_rate": 0.00013275369210248317,
      "loss": 0.9652,
      "step": 4254
    },
    {
      "epoch": 0.7907452146441182,
      "grad_norm": 0.5878126621246338,
      "learning_rate": 0.00013272605807026285,
      "loss": 1.0158,
      "step": 4255
    },
    {
      "epoch": 0.7909310537074893,
      "grad_norm": 0.5607028007507324,
      "learning_rate": 0.00013269842123891205,
      "loss": 1.1119,
      "step": 4256
    },
    {
      "epoch": 0.7911168927708604,
      "grad_norm": 0.6098755598068237,
      "learning_rate": 0.0001326707816107946,
      "loss": 1.2164,
      "step": 4257
    },
    {
      "epoch": 0.7913027318342316,
      "grad_norm": 0.640153706073761,
      "learning_rate": 0.00013264313918827462,
      "loss": 0.9233,
      "step": 4258
    },
    {
      "epoch": 0.7914885708976027,
      "grad_norm": 0.6072023510932922,
      "learning_rate": 0.00013261549397371636,
      "loss": 0.9476,
      "step": 4259
    },
    {
      "epoch": 0.7916744099609738,
      "grad_norm": 0.6095691919326782,
      "learning_rate": 0.00013258784596948444,
      "loss": 1.1407,
      "step": 4260
    },
    {
      "epoch": 0.7918602490243449,
      "grad_norm": 0.8312792181968689,
      "learning_rate": 0.0001325601951779436,
      "loss": 0.9949,
      "step": 4261
    },
    {
      "epoch": 0.792046088087716,
      "grad_norm": 0.7518773674964905,
      "learning_rate": 0.00013253254160145892,
      "loss": 1.1936,
      "step": 4262
    },
    {
      "epoch": 0.7922319271510871,
      "grad_norm": 0.5493016242980957,
      "learning_rate": 0.00013250488524239559,
      "loss": 1.0946,
      "step": 4263
    },
    {
      "epoch": 0.7924177662144583,
      "grad_norm": 0.5355292558670044,
      "learning_rate": 0.0001324772261031192,
      "loss": 0.9688,
      "step": 4264
    },
    {
      "epoch": 0.7926036052778294,
      "grad_norm": 0.7269943356513977,
      "learning_rate": 0.0001324495641859955,
      "loss": 0.8984,
      "step": 4265
    },
    {
      "epoch": 0.7927894443412005,
      "grad_norm": 0.5987350940704346,
      "learning_rate": 0.00013242189949339041,
      "loss": 0.967,
      "step": 4266
    },
    {
      "epoch": 0.7929752834045717,
      "grad_norm": 0.7158120274543762,
      "learning_rate": 0.0001323942320276702,
      "loss": 1.1093,
      "step": 4267
    },
    {
      "epoch": 0.7931611224679428,
      "grad_norm": 0.6944933533668518,
      "learning_rate": 0.0001323665617912013,
      "loss": 1.2342,
      "step": 4268
    },
    {
      "epoch": 0.7933469615313139,
      "grad_norm": 0.5208589434623718,
      "learning_rate": 0.00013233888878635044,
      "loss": 1.0631,
      "step": 4269
    },
    {
      "epoch": 0.793532800594685,
      "grad_norm": 0.588266134262085,
      "learning_rate": 0.00013231121301548448,
      "loss": 1.0709,
      "step": 4270
    },
    {
      "epoch": 0.7937186396580561,
      "grad_norm": 0.7578076124191284,
      "learning_rate": 0.00013228353448097068,
      "loss": 1.0585,
      "step": 4271
    },
    {
      "epoch": 0.7939044787214272,
      "grad_norm": 0.573829710483551,
      "learning_rate": 0.00013225585318517638,
      "loss": 1.0457,
      "step": 4272
    },
    {
      "epoch": 0.7940903177847983,
      "grad_norm": 0.6806961297988892,
      "learning_rate": 0.00013222816913046928,
      "loss": 1.0456,
      "step": 4273
    },
    {
      "epoch": 0.7942761568481694,
      "grad_norm": 0.57918381690979,
      "learning_rate": 0.0001322004823192172,
      "loss": 1.066,
      "step": 4274
    },
    {
      "epoch": 0.7944619959115407,
      "grad_norm": 0.6446574926376343,
      "learning_rate": 0.00013217279275378824,
      "loss": 1.0582,
      "step": 4275
    },
    {
      "epoch": 0.7946478349749118,
      "grad_norm": 0.6489240527153015,
      "learning_rate": 0.00013214510043655083,
      "loss": 1.0116,
      "step": 4276
    },
    {
      "epoch": 0.7948336740382829,
      "grad_norm": 0.5904452800750732,
      "learning_rate": 0.00013211740536987345,
      "loss": 1.1452,
      "step": 4277
    },
    {
      "epoch": 0.795019513101654,
      "grad_norm": 0.7266445159912109,
      "learning_rate": 0.00013208970755612502,
      "loss": 1.0409,
      "step": 4278
    },
    {
      "epoch": 0.7952053521650251,
      "grad_norm": 0.6511578559875488,
      "learning_rate": 0.00013206200699767453,
      "loss": 1.0528,
      "step": 4279
    },
    {
      "epoch": 0.7953911912283962,
      "grad_norm": 0.6173616051673889,
      "learning_rate": 0.0001320343036968913,
      "loss": 1.0912,
      "step": 4280
    },
    {
      "epoch": 0.7955770302917673,
      "grad_norm": 0.5258021354675293,
      "learning_rate": 0.00013200659765614484,
      "loss": 0.8425,
      "step": 4281
    },
    {
      "epoch": 0.7957628693551384,
      "grad_norm": 0.6236848831176758,
      "learning_rate": 0.00013197888887780487,
      "loss": 1.1233,
      "step": 4282
    },
    {
      "epoch": 0.7959487084185096,
      "grad_norm": 0.7987383008003235,
      "learning_rate": 0.00013195117736424146,
      "loss": 1.2049,
      "step": 4283
    },
    {
      "epoch": 0.7961345474818807,
      "grad_norm": 0.6573415994644165,
      "learning_rate": 0.00013192346311782476,
      "loss": 1.2696,
      "step": 4284
    },
    {
      "epoch": 0.7963203865452518,
      "grad_norm": 0.5358014106750488,
      "learning_rate": 0.0001318957461409253,
      "loss": 1.1063,
      "step": 4285
    },
    {
      "epoch": 0.7965062256086229,
      "grad_norm": 0.6531534194946289,
      "learning_rate": 0.00013186802643591373,
      "loss": 1.3224,
      "step": 4286
    },
    {
      "epoch": 0.796692064671994,
      "grad_norm": 0.5929547548294067,
      "learning_rate": 0.00013184030400516095,
      "loss": 1.0583,
      "step": 4287
    },
    {
      "epoch": 0.7968779037353652,
      "grad_norm": 0.5735682249069214,
      "learning_rate": 0.00013181257885103818,
      "loss": 1.0209,
      "step": 4288
    },
    {
      "epoch": 0.7970637427987363,
      "grad_norm": 0.5685105323791504,
      "learning_rate": 0.00013178485097591676,
      "loss": 0.8461,
      "step": 4289
    },
    {
      "epoch": 0.7972495818621074,
      "grad_norm": 2.1148364543914795,
      "learning_rate": 0.00013175712038216838,
      "loss": 2.0309,
      "step": 4290
    },
    {
      "epoch": 0.7974354209254786,
      "grad_norm": 0.6161953210830688,
      "learning_rate": 0.00013172938707216482,
      "loss": 1.0573,
      "step": 4291
    },
    {
      "epoch": 0.7976212599888497,
      "grad_norm": 0.559108316898346,
      "learning_rate": 0.00013170165104827824,
      "loss": 0.9276,
      "step": 4292
    },
    {
      "epoch": 0.7978070990522208,
      "grad_norm": 0.6372076869010925,
      "learning_rate": 0.0001316739123128809,
      "loss": 1.0523,
      "step": 4293
    },
    {
      "epoch": 0.7979929381155919,
      "grad_norm": 0.5434962511062622,
      "learning_rate": 0.00013164617086834543,
      "loss": 0.9008,
      "step": 4294
    },
    {
      "epoch": 0.798178777178963,
      "grad_norm": 0.5048145651817322,
      "learning_rate": 0.00013161842671704452,
      "loss": 0.9488,
      "step": 4295
    },
    {
      "epoch": 0.7983646162423341,
      "grad_norm": 0.56045001745224,
      "learning_rate": 0.00013159067986135125,
      "loss": 1.1348,
      "step": 4296
    },
    {
      "epoch": 0.7985504553057052,
      "grad_norm": 0.6737070083618164,
      "learning_rate": 0.00013156293030363886,
      "loss": 1.0575,
      "step": 4297
    },
    {
      "epoch": 0.7987362943690763,
      "grad_norm": 0.6539118885993958,
      "learning_rate": 0.00013153517804628082,
      "loss": 1.0982,
      "step": 4298
    },
    {
      "epoch": 0.7989221334324474,
      "grad_norm": 0.5994445085525513,
      "learning_rate": 0.00013150742309165085,
      "loss": 1.3265,
      "step": 4299
    },
    {
      "epoch": 0.7991079724958187,
      "grad_norm": 0.7722712755203247,
      "learning_rate": 0.0001314796654421229,
      "loss": 1.0636,
      "step": 4300
    },
    {
      "epoch": 0.7992938115591898,
      "grad_norm": 0.6560238003730774,
      "learning_rate": 0.0001314519051000711,
      "loss": 1.244,
      "step": 4301
    },
    {
      "epoch": 0.7994796506225609,
      "grad_norm": 0.5150291919708252,
      "learning_rate": 0.0001314241420678699,
      "loss": 1.0826,
      "step": 4302
    },
    {
      "epoch": 0.799665489685932,
      "grad_norm": 0.6745650768280029,
      "learning_rate": 0.0001313963763478939,
      "loss": 1.1411,
      "step": 4303
    },
    {
      "epoch": 0.7998513287493031,
      "grad_norm": 0.5777543187141418,
      "learning_rate": 0.000131368607942518,
      "loss": 0.9485,
      "step": 4304
    },
    {
      "epoch": 0.8000371678126742,
      "grad_norm": 0.6721982955932617,
      "learning_rate": 0.00013134083685411723,
      "loss": 1.2325,
      "step": 4305
    },
    {
      "epoch": 0.8002230068760453,
      "grad_norm": 0.5857947468757629,
      "learning_rate": 0.00013131306308506696,
      "loss": 0.8454,
      "step": 4306
    },
    {
      "epoch": 0.8004088459394164,
      "grad_norm": 0.6334595680236816,
      "learning_rate": 0.00013128528663774276,
      "loss": 1.1562,
      "step": 4307
    },
    {
      "epoch": 0.8005946850027876,
      "grad_norm": 0.6741576194763184,
      "learning_rate": 0.00013125750751452034,
      "loss": 0.9842,
      "step": 4308
    },
    {
      "epoch": 0.8007805240661587,
      "grad_norm": 0.5686050057411194,
      "learning_rate": 0.00013122972571777576,
      "loss": 1.0821,
      "step": 4309
    },
    {
      "epoch": 0.8009663631295298,
      "grad_norm": 0.6299645304679871,
      "learning_rate": 0.00013120194124988522,
      "loss": 1.1056,
      "step": 4310
    },
    {
      "epoch": 0.801152202192901,
      "grad_norm": 3.2690067291259766,
      "learning_rate": 0.00013117415411322523,
      "loss": 1.9315,
      "step": 4311
    },
    {
      "epoch": 0.8013380412562721,
      "grad_norm": 0.7266578078269958,
      "learning_rate": 0.00013114636431017248,
      "loss": 1.2857,
      "step": 4312
    },
    {
      "epoch": 0.8015238803196432,
      "grad_norm": 0.5830214619636536,
      "learning_rate": 0.00013111857184310382,
      "loss": 0.8759,
      "step": 4313
    },
    {
      "epoch": 0.8017097193830143,
      "grad_norm": 0.5984452962875366,
      "learning_rate": 0.00013109077671439652,
      "loss": 1.2159,
      "step": 4314
    },
    {
      "epoch": 0.8018955584463854,
      "grad_norm": 0.6634993553161621,
      "learning_rate": 0.00013106297892642786,
      "loss": 1.0644,
      "step": 4315
    },
    {
      "epoch": 0.8020813975097566,
      "grad_norm": 0.5921594500541687,
      "learning_rate": 0.00013103517848157546,
      "loss": 0.9128,
      "step": 4316
    },
    {
      "epoch": 0.8022672365731277,
      "grad_norm": 0.501189112663269,
      "learning_rate": 0.0001310073753822172,
      "loss": 0.8117,
      "step": 4317
    },
    {
      "epoch": 0.8024530756364988,
      "grad_norm": 0.5862453579902649,
      "learning_rate": 0.00013097956963073107,
      "loss": 1.0291,
      "step": 4318
    },
    {
      "epoch": 0.8026389146998699,
      "grad_norm": 0.5692201852798462,
      "learning_rate": 0.00013095176122949543,
      "loss": 0.9335,
      "step": 4319
    },
    {
      "epoch": 0.802824753763241,
      "grad_norm": 0.5870519280433655,
      "learning_rate": 0.00013092395018088872,
      "loss": 1.1602,
      "step": 4320
    },
    {
      "epoch": 0.8030105928266121,
      "grad_norm": 0.532107412815094,
      "learning_rate": 0.00013089613648728976,
      "loss": 1.0165,
      "step": 4321
    },
    {
      "epoch": 0.8031964318899832,
      "grad_norm": 0.7711302042007446,
      "learning_rate": 0.00013086832015107746,
      "loss": 1.2256,
      "step": 4322
    },
    {
      "epoch": 0.8033822709533543,
      "grad_norm": 0.5713910460472107,
      "learning_rate": 0.000130840501174631,
      "loss": 1.064,
      "step": 4323
    },
    {
      "epoch": 0.8035681100167256,
      "grad_norm": 0.5301079154014587,
      "learning_rate": 0.00013081267956032986,
      "loss": 0.9982,
      "step": 4324
    },
    {
      "epoch": 0.8037539490800967,
      "grad_norm": 0.47264859080314636,
      "learning_rate": 0.0001307848553105536,
      "loss": 0.6809,
      "step": 4325
    },
    {
      "epoch": 0.8039397881434678,
      "grad_norm": 0.7126772403717041,
      "learning_rate": 0.00013075702842768212,
      "loss": 1.3468,
      "step": 4326
    },
    {
      "epoch": 0.8041256272068389,
      "grad_norm": 0.5951479077339172,
      "learning_rate": 0.00013072919891409556,
      "loss": 0.8848,
      "step": 4327
    },
    {
      "epoch": 0.80431146627021,
      "grad_norm": 0.8132142424583435,
      "learning_rate": 0.0001307013667721742,
      "loss": 1.1889,
      "step": 4328
    },
    {
      "epoch": 0.8044973053335811,
      "grad_norm": 0.6405729055404663,
      "learning_rate": 0.00013067353200429857,
      "loss": 1.2084,
      "step": 4329
    },
    {
      "epoch": 0.8046831443969522,
      "grad_norm": 0.5916075110435486,
      "learning_rate": 0.00013064569461284945,
      "loss": 0.9809,
      "step": 4330
    },
    {
      "epoch": 0.8048689834603233,
      "grad_norm": 0.5530146956443787,
      "learning_rate": 0.00013061785460020785,
      "loss": 1.0646,
      "step": 4331
    },
    {
      "epoch": 0.8050548225236945,
      "grad_norm": 0.5294923186302185,
      "learning_rate": 0.00013059001196875496,
      "loss": 0.8103,
      "step": 4332
    },
    {
      "epoch": 0.8052406615870656,
      "grad_norm": 0.6035967469215393,
      "learning_rate": 0.00013056216672087222,
      "loss": 1.0413,
      "step": 4333
    },
    {
      "epoch": 0.8054265006504367,
      "grad_norm": 0.720556378364563,
      "learning_rate": 0.00013053431885894133,
      "loss": 1.2473,
      "step": 4334
    },
    {
      "epoch": 0.8056123397138079,
      "grad_norm": 0.6400815844535828,
      "learning_rate": 0.00013050646838534417,
      "loss": 0.9757,
      "step": 4335
    },
    {
      "epoch": 0.805798178777179,
      "grad_norm": 0.5894456505775452,
      "learning_rate": 0.00013047861530246282,
      "loss": 1.1899,
      "step": 4336
    },
    {
      "epoch": 0.8059840178405501,
      "grad_norm": 0.6233096122741699,
      "learning_rate": 0.00013045075961267964,
      "loss": 1.0853,
      "step": 4337
    },
    {
      "epoch": 0.8061698569039212,
      "grad_norm": 0.5251320004463196,
      "learning_rate": 0.00013042290131837718,
      "loss": 1.0457,
      "step": 4338
    },
    {
      "epoch": 0.8063556959672923,
      "grad_norm": 0.6254708170890808,
      "learning_rate": 0.00013039504042193818,
      "loss": 0.9582,
      "step": 4339
    },
    {
      "epoch": 0.8065415350306634,
      "grad_norm": 0.5413371920585632,
      "learning_rate": 0.0001303671769257457,
      "loss": 0.9513,
      "step": 4340
    },
    {
      "epoch": 0.8067273740940346,
      "grad_norm": 0.5372425317764282,
      "learning_rate": 0.00013033931083218295,
      "loss": 1.0867,
      "step": 4341
    },
    {
      "epoch": 0.8069132131574057,
      "grad_norm": 0.6267648935317993,
      "learning_rate": 0.00013031144214363337,
      "loss": 1.256,
      "step": 4342
    },
    {
      "epoch": 0.8070990522207768,
      "grad_norm": 0.9133261442184448,
      "learning_rate": 0.00013028357086248063,
      "loss": 0.8639,
      "step": 4343
    },
    {
      "epoch": 0.8072848912841479,
      "grad_norm": 0.4667738378047943,
      "learning_rate": 0.00013025569699110862,
      "loss": 0.9185,
      "step": 4344
    },
    {
      "epoch": 0.807470730347519,
      "grad_norm": 0.5194012522697449,
      "learning_rate": 0.00013022782053190147,
      "loss": 0.8947,
      "step": 4345
    },
    {
      "epoch": 0.8076565694108901,
      "grad_norm": 1.2933731079101562,
      "learning_rate": 0.00013019994148724348,
      "loss": 1.4954,
      "step": 4346
    },
    {
      "epoch": 0.8078424084742613,
      "grad_norm": 0.686096727848053,
      "learning_rate": 0.00013017205985951926,
      "loss": 1.2909,
      "step": 4347
    },
    {
      "epoch": 0.8080282475376324,
      "grad_norm": 0.6402808427810669,
      "learning_rate": 0.0001301441756511135,
      "loss": 1.1911,
      "step": 4348
    },
    {
      "epoch": 0.8082140866010036,
      "grad_norm": 0.5459903478622437,
      "learning_rate": 0.00013011628886441129,
      "loss": 0.9079,
      "step": 4349
    },
    {
      "epoch": 0.8083999256643747,
      "grad_norm": 0.6057734489440918,
      "learning_rate": 0.00013008839950179776,
      "loss": 1.1206,
      "step": 4350
    },
    {
      "epoch": 0.8085857647277458,
      "grad_norm": 0.6145203113555908,
      "learning_rate": 0.0001300605075656584,
      "loss": 1.2271,
      "step": 4351
    },
    {
      "epoch": 0.8087716037911169,
      "grad_norm": 0.5688883662223816,
      "learning_rate": 0.00013003261305837891,
      "loss": 0.6687,
      "step": 4352
    },
    {
      "epoch": 0.808957442854488,
      "grad_norm": 0.5707416534423828,
      "learning_rate": 0.00013000471598234508,
      "loss": 1.0749,
      "step": 4353
    },
    {
      "epoch": 0.8091432819178591,
      "grad_norm": 0.5867960453033447,
      "learning_rate": 0.00012997681633994304,
      "loss": 1.0557,
      "step": 4354
    },
    {
      "epoch": 0.8093291209812302,
      "grad_norm": 0.575100302696228,
      "learning_rate": 0.0001299489141335591,
      "loss": 1.0123,
      "step": 4355
    },
    {
      "epoch": 0.8095149600446013,
      "grad_norm": 0.5517423152923584,
      "learning_rate": 0.00012992100936557984,
      "loss": 1.0448,
      "step": 4356
    },
    {
      "epoch": 0.8097007991079725,
      "grad_norm": 0.552761971950531,
      "learning_rate": 0.00012989310203839193,
      "loss": 0.6827,
      "step": 4357
    },
    {
      "epoch": 0.8098866381713437,
      "grad_norm": 0.531324565410614,
      "learning_rate": 0.00012986519215438244,
      "loss": 1.0561,
      "step": 4358
    },
    {
      "epoch": 0.8100724772347148,
      "grad_norm": 0.6074504852294922,
      "learning_rate": 0.0001298372797159385,
      "loss": 0.9044,
      "step": 4359
    },
    {
      "epoch": 0.8102583162980859,
      "grad_norm": 0.6233542561531067,
      "learning_rate": 0.00012980936472544756,
      "loss": 1.2859,
      "step": 4360
    },
    {
      "epoch": 0.810444155361457,
      "grad_norm": 0.6285319924354553,
      "learning_rate": 0.00012978144718529718,
      "loss": 1.2288,
      "step": 4361
    },
    {
      "epoch": 0.8106299944248281,
      "grad_norm": 0.5883091688156128,
      "learning_rate": 0.00012975352709787528,
      "loss": 0.8492,
      "step": 4362
    },
    {
      "epoch": 0.8108158334881992,
      "grad_norm": 0.6085577011108398,
      "learning_rate": 0.0001297256044655699,
      "loss": 1.0563,
      "step": 4363
    },
    {
      "epoch": 0.8110016725515703,
      "grad_norm": 0.552869439125061,
      "learning_rate": 0.00012969767929076935,
      "loss": 0.8394,
      "step": 4364
    },
    {
      "epoch": 0.8111875116149415,
      "grad_norm": 0.6783784627914429,
      "learning_rate": 0.0001296697515758621,
      "loss": 1.2056,
      "step": 4365
    },
    {
      "epoch": 0.8113733506783126,
      "grad_norm": 0.6433395743370056,
      "learning_rate": 0.00012964182132323688,
      "loss": 1.1556,
      "step": 4366
    },
    {
      "epoch": 0.8115591897416837,
      "grad_norm": 0.7210237979888916,
      "learning_rate": 0.0001296138885352826,
      "loss": 1.1526,
      "step": 4367
    },
    {
      "epoch": 0.8117450288050548,
      "grad_norm": 0.5519304275512695,
      "learning_rate": 0.0001295859532143884,
      "loss": 0.6599,
      "step": 4368
    },
    {
      "epoch": 0.8119308678684259,
      "grad_norm": 0.6399482488632202,
      "learning_rate": 0.00012955801536294371,
      "loss": 1.2954,
      "step": 4369
    },
    {
      "epoch": 0.812116706931797,
      "grad_norm": 0.6463430523872375,
      "learning_rate": 0.00012953007498333808,
      "loss": 0.9671,
      "step": 4370
    },
    {
      "epoch": 0.8123025459951682,
      "grad_norm": 0.6632814407348633,
      "learning_rate": 0.0001295021320779613,
      "loss": 1.2656,
      "step": 4371
    },
    {
      "epoch": 0.8124883850585393,
      "grad_norm": 0.585106611251831,
      "learning_rate": 0.00012947418664920345,
      "loss": 1.0554,
      "step": 4372
    },
    {
      "epoch": 0.8126742241219104,
      "grad_norm": 0.6526263356208801,
      "learning_rate": 0.00012944623869945465,
      "loss": 1.1726,
      "step": 4373
    },
    {
      "epoch": 0.8128600631852816,
      "grad_norm": 0.6378617286682129,
      "learning_rate": 0.00012941828823110545,
      "loss": 1.031,
      "step": 4374
    },
    {
      "epoch": 0.8130459022486527,
      "grad_norm": 2.6388754844665527,
      "learning_rate": 0.00012939033524654646,
      "loss": 2.0343,
      "step": 4375
    },
    {
      "epoch": 0.8132317413120238,
      "grad_norm": 0.5578187704086304,
      "learning_rate": 0.0001293623797481686,
      "loss": 1.0333,
      "step": 4376
    },
    {
      "epoch": 0.8134175803753949,
      "grad_norm": 1.881386637687683,
      "learning_rate": 0.00012933442173836287,
      "loss": 1.8941,
      "step": 4377
    },
    {
      "epoch": 0.813603419438766,
      "grad_norm": 0.6024147272109985,
      "learning_rate": 0.00012930646121952073,
      "loss": 1.0478,
      "step": 4378
    },
    {
      "epoch": 0.8137892585021371,
      "grad_norm": 0.6307503581047058,
      "learning_rate": 0.00012927849819403358,
      "loss": 1.0855,
      "step": 4379
    },
    {
      "epoch": 0.8139750975655082,
      "grad_norm": 0.6026185154914856,
      "learning_rate": 0.00012925053266429323,
      "loss": 1.0046,
      "step": 4380
    },
    {
      "epoch": 0.8141609366288793,
      "grad_norm": 0.7271862626075745,
      "learning_rate": 0.00012922256463269159,
      "loss": 0.9055,
      "step": 4381
    },
    {
      "epoch": 0.8143467756922506,
      "grad_norm": 0.5896093249320984,
      "learning_rate": 0.00012919459410162083,
      "loss": 1.064,
      "step": 4382
    },
    {
      "epoch": 0.8145326147556217,
      "grad_norm": 0.5362374186515808,
      "learning_rate": 0.00012916662107347333,
      "loss": 0.9173,
      "step": 4383
    },
    {
      "epoch": 0.8147184538189928,
      "grad_norm": 0.5842663645744324,
      "learning_rate": 0.0001291386455506417,
      "loss": 0.9661,
      "step": 4384
    },
    {
      "epoch": 0.8149042928823639,
      "grad_norm": 0.7104371190071106,
      "learning_rate": 0.00012911066753551874,
      "loss": 1.162,
      "step": 4385
    },
    {
      "epoch": 0.815090131945735,
      "grad_norm": 0.6280097365379333,
      "learning_rate": 0.0001290826870304975,
      "loss": 1.0367,
      "step": 4386
    },
    {
      "epoch": 0.8152759710091061,
      "grad_norm": 0.5821822285652161,
      "learning_rate": 0.00012905470403797113,
      "loss": 1.2102,
      "step": 4387
    },
    {
      "epoch": 0.8154618100724772,
      "grad_norm": 0.658132791519165,
      "learning_rate": 0.00012902671856033316,
      "loss": 1.1359,
      "step": 4388
    },
    {
      "epoch": 0.8156476491358483,
      "grad_norm": 0.563515841960907,
      "learning_rate": 0.0001289987305999772,
      "loss": 1.3327,
      "step": 4389
    },
    {
      "epoch": 0.8158334881992195,
      "grad_norm": 0.6515486240386963,
      "learning_rate": 0.00012897074015929717,
      "loss": 1.3089,
      "step": 4390
    },
    {
      "epoch": 0.8160193272625906,
      "grad_norm": 0.6316953897476196,
      "learning_rate": 0.00012894274724068708,
      "loss": 1.0272,
      "step": 4391
    },
    {
      "epoch": 0.8162051663259617,
      "grad_norm": 0.5534893274307251,
      "learning_rate": 0.00012891475184654126,
      "loss": 1.1062,
      "step": 4392
    },
    {
      "epoch": 0.8163910053893328,
      "grad_norm": 0.5467495918273926,
      "learning_rate": 0.00012888675397925426,
      "loss": 1.2019,
      "step": 4393
    },
    {
      "epoch": 0.816576844452704,
      "grad_norm": 0.6885834336280823,
      "learning_rate": 0.00012885875364122075,
      "loss": 1.1425,
      "step": 4394
    },
    {
      "epoch": 0.8167626835160751,
      "grad_norm": 0.508720874786377,
      "learning_rate": 0.00012883075083483566,
      "loss": 0.9065,
      "step": 4395
    },
    {
      "epoch": 0.8169485225794462,
      "grad_norm": 0.6153755187988281,
      "learning_rate": 0.00012880274556249414,
      "loss": 1.1685,
      "step": 4396
    },
    {
      "epoch": 0.8171343616428173,
      "grad_norm": 0.6323347091674805,
      "learning_rate": 0.00012877473782659156,
      "loss": 1.2422,
      "step": 4397
    },
    {
      "epoch": 0.8173202007061885,
      "grad_norm": 0.6620142459869385,
      "learning_rate": 0.00012874672762952345,
      "loss": 0.8893,
      "step": 4398
    },
    {
      "epoch": 0.8175060397695596,
      "grad_norm": 0.5536710023880005,
      "learning_rate": 0.00012871871497368559,
      "loss": 1.0329,
      "step": 4399
    },
    {
      "epoch": 0.8176918788329307,
      "grad_norm": 0.6210238933563232,
      "learning_rate": 0.00012869069986147394,
      "loss": 1.1554,
      "step": 4400
    },
    {
      "epoch": 0.8178777178963018,
      "grad_norm": 0.5708968043327332,
      "learning_rate": 0.0001286626822952848,
      "loss": 1.0671,
      "step": 4401
    },
    {
      "epoch": 0.8180635569596729,
      "grad_norm": 0.5646460652351379,
      "learning_rate": 0.00012863466227751443,
      "loss": 1.1507,
      "step": 4402
    },
    {
      "epoch": 0.818249396023044,
      "grad_norm": 0.7381437420845032,
      "learning_rate": 0.0001286066398105595,
      "loss": 1.179,
      "step": 4403
    },
    {
      "epoch": 0.8184352350864151,
      "grad_norm": 0.5559968948364258,
      "learning_rate": 0.00012857861489681688,
      "loss": 0.9457,
      "step": 4404
    },
    {
      "epoch": 0.8186210741497862,
      "grad_norm": 0.6272557377815247,
      "learning_rate": 0.00012855058753868357,
      "loss": 1.0855,
      "step": 4405
    },
    {
      "epoch": 0.8188069132131574,
      "grad_norm": 0.5753204822540283,
      "learning_rate": 0.00012852255773855676,
      "loss": 0.8327,
      "step": 4406
    },
    {
      "epoch": 0.8189927522765286,
      "grad_norm": 0.6473789811134338,
      "learning_rate": 0.00012849452549883398,
      "loss": 1.1054,
      "step": 4407
    },
    {
      "epoch": 0.8191785913398997,
      "grad_norm": 0.6232612729072571,
      "learning_rate": 0.00012846649082191283,
      "loss": 1.2546,
      "step": 4408
    },
    {
      "epoch": 0.8193644304032708,
      "grad_norm": 0.4950093626976013,
      "learning_rate": 0.0001284384537101912,
      "loss": 1.1192,
      "step": 4409
    },
    {
      "epoch": 0.8195502694666419,
      "grad_norm": 0.6465705633163452,
      "learning_rate": 0.00012841041416606714,
      "loss": 1.0576,
      "step": 4410
    },
    {
      "epoch": 0.819736108530013,
      "grad_norm": 0.7886804342269897,
      "learning_rate": 0.00012838237219193896,
      "loss": 1.2127,
      "step": 4411
    },
    {
      "epoch": 0.8199219475933841,
      "grad_norm": 0.5513986349105835,
      "learning_rate": 0.00012835432779020515,
      "loss": 0.897,
      "step": 4412
    },
    {
      "epoch": 0.8201077866567552,
      "grad_norm": 0.5260704755783081,
      "learning_rate": 0.0001283262809632644,
      "loss": 0.8837,
      "step": 4413
    },
    {
      "epoch": 0.8202936257201263,
      "grad_norm": 0.701880693435669,
      "learning_rate": 0.00012829823171351564,
      "loss": 1.0562,
      "step": 4414
    },
    {
      "epoch": 0.8204794647834975,
      "grad_norm": 0.5578733086585999,
      "learning_rate": 0.00012827018004335793,
      "loss": 1.0216,
      "step": 4415
    },
    {
      "epoch": 0.8206653038468686,
      "grad_norm": 0.5386905670166016,
      "learning_rate": 0.00012824212595519065,
      "loss": 0.7909,
      "step": 4416
    },
    {
      "epoch": 0.8208511429102398,
      "grad_norm": 0.7121390700340271,
      "learning_rate": 0.00012821406945141326,
      "loss": 1.1079,
      "step": 4417
    },
    {
      "epoch": 0.8210369819736109,
      "grad_norm": 0.5872541666030884,
      "learning_rate": 0.0001281860105344256,
      "loss": 1.0154,
      "step": 4418
    },
    {
      "epoch": 0.821222821036982,
      "grad_norm": 0.8177608847618103,
      "learning_rate": 0.00012815794920662747,
      "loss": 1.0999,
      "step": 4419
    },
    {
      "epoch": 0.8214086601003531,
      "grad_norm": 0.5959897637367249,
      "learning_rate": 0.00012812988547041912,
      "loss": 1.1842,
      "step": 4420
    },
    {
      "epoch": 0.8215944991637242,
      "grad_norm": 0.5651900768280029,
      "learning_rate": 0.00012810181932820086,
      "loss": 1.2421,
      "step": 4421
    },
    {
      "epoch": 0.8217803382270953,
      "grad_norm": 0.5698773264884949,
      "learning_rate": 0.00012807375078237326,
      "loss": 1.1008,
      "step": 4422
    },
    {
      "epoch": 0.8219661772904665,
      "grad_norm": 0.5098702907562256,
      "learning_rate": 0.00012804567983533708,
      "loss": 0.9125,
      "step": 4423
    },
    {
      "epoch": 0.8221520163538376,
      "grad_norm": 0.6269254088401794,
      "learning_rate": 0.00012801760648949327,
      "loss": 0.7964,
      "step": 4424
    },
    {
      "epoch": 0.8223378554172087,
      "grad_norm": 6.160284996032715,
      "learning_rate": 0.00012798953074724307,
      "loss": 1.8079,
      "step": 4425
    },
    {
      "epoch": 0.8225236944805798,
      "grad_norm": 0.5508794188499451,
      "learning_rate": 0.00012796145261098778,
      "loss": 1.0323,
      "step": 4426
    },
    {
      "epoch": 0.8227095335439509,
      "grad_norm": 0.8076217174530029,
      "learning_rate": 0.00012793337208312903,
      "loss": 1.226,
      "step": 4427
    },
    {
      "epoch": 0.822895372607322,
      "grad_norm": 0.6926615834236145,
      "learning_rate": 0.00012790528916606857,
      "loss": 0.9492,
      "step": 4428
    },
    {
      "epoch": 0.8230812116706931,
      "grad_norm": 0.5568311214447021,
      "learning_rate": 0.00012787720386220843,
      "loss": 0.8471,
      "step": 4429
    },
    {
      "epoch": 0.8232670507340643,
      "grad_norm": 0.5556146502494812,
      "learning_rate": 0.00012784911617395078,
      "loss": 1.1586,
      "step": 4430
    },
    {
      "epoch": 0.8234528897974355,
      "grad_norm": 0.6666334867477417,
      "learning_rate": 0.00012782102610369803,
      "loss": 1.175,
      "step": 4431
    },
    {
      "epoch": 0.8236387288608066,
      "grad_norm": 0.6975376009941101,
      "learning_rate": 0.00012779293365385282,
      "loss": 1.1044,
      "step": 4432
    },
    {
      "epoch": 0.8238245679241777,
      "grad_norm": 0.5512378215789795,
      "learning_rate": 0.00012776483882681792,
      "loss": 1.0993,
      "step": 4433
    },
    {
      "epoch": 0.8240104069875488,
      "grad_norm": 0.5872717499732971,
      "learning_rate": 0.00012773674162499634,
      "loss": 0.885,
      "step": 4434
    },
    {
      "epoch": 0.8241962460509199,
      "grad_norm": 0.6425865292549133,
      "learning_rate": 0.00012770864205079128,
      "loss": 1.1616,
      "step": 4435
    },
    {
      "epoch": 0.824382085114291,
      "grad_norm": 0.5967339873313904,
      "learning_rate": 0.00012768054010660622,
      "loss": 1.1622,
      "step": 4436
    },
    {
      "epoch": 0.8245679241776621,
      "grad_norm": 0.5792629718780518,
      "learning_rate": 0.00012765243579484473,
      "loss": 0.9509,
      "step": 4437
    },
    {
      "epoch": 0.8247537632410332,
      "grad_norm": 0.5237547159194946,
      "learning_rate": 0.00012762432911791061,
      "loss": 1.0155,
      "step": 4438
    },
    {
      "epoch": 0.8249396023044044,
      "grad_norm": 0.6082087755203247,
      "learning_rate": 0.00012759622007820794,
      "loss": 1.1278,
      "step": 4439
    },
    {
      "epoch": 0.8251254413677755,
      "grad_norm": 0.5648608803749084,
      "learning_rate": 0.00012756810867814092,
      "loss": 1.2205,
      "step": 4440
    },
    {
      "epoch": 0.8253112804311467,
      "grad_norm": 0.6407959461212158,
      "learning_rate": 0.000127539994920114,
      "loss": 0.8724,
      "step": 4441
    },
    {
      "epoch": 0.8254971194945178,
      "grad_norm": 0.5947341322898865,
      "learning_rate": 0.00012751187880653174,
      "loss": 1.0559,
      "step": 4442
    },
    {
      "epoch": 0.8256829585578889,
      "grad_norm": 0.5259736180305481,
      "learning_rate": 0.00012748376033979908,
      "loss": 1.1746,
      "step": 4443
    },
    {
      "epoch": 0.82586879762126,
      "grad_norm": 0.5010737180709839,
      "learning_rate": 0.00012745563952232098,
      "loss": 0.8902,
      "step": 4444
    },
    {
      "epoch": 0.8260546366846311,
      "grad_norm": 0.5889286398887634,
      "learning_rate": 0.00012742751635650267,
      "loss": 1.0432,
      "step": 4445
    },
    {
      "epoch": 0.8262404757480022,
      "grad_norm": 2.0364248752593994,
      "learning_rate": 0.00012739939084474962,
      "loss": 1.4645,
      "step": 4446
    },
    {
      "epoch": 0.8264263148113733,
      "grad_norm": 0.7073319554328918,
      "learning_rate": 0.00012737126298946748,
      "loss": 0.9539,
      "step": 4447
    },
    {
      "epoch": 0.8266121538747445,
      "grad_norm": 0.5749005079269409,
      "learning_rate": 0.00012734313279306206,
      "loss": 1.0574,
      "step": 4448
    },
    {
      "epoch": 0.8267979929381156,
      "grad_norm": 0.6676498651504517,
      "learning_rate": 0.00012731500025793936,
      "loss": 1.2505,
      "step": 4449
    },
    {
      "epoch": 0.8269838320014867,
      "grad_norm": 0.6630710959434509,
      "learning_rate": 0.00012728686538650565,
      "loss": 1.2037,
      "step": 4450
    },
    {
      "epoch": 0.8271696710648578,
      "grad_norm": 0.8507979512214661,
      "learning_rate": 0.0001272587281811674,
      "loss": 1.3893,
      "step": 4451
    },
    {
      "epoch": 0.8273555101282289,
      "grad_norm": 0.4904545247554779,
      "learning_rate": 0.00012723058864433118,
      "loss": 1.1238,
      "step": 4452
    },
    {
      "epoch": 0.8275413491916,
      "grad_norm": 0.5739341378211975,
      "learning_rate": 0.00012720244677840388,
      "loss": 0.9467,
      "step": 4453
    },
    {
      "epoch": 0.8277271882549712,
      "grad_norm": 0.49247053265571594,
      "learning_rate": 0.00012717430258579252,
      "loss": 0.781,
      "step": 4454
    },
    {
      "epoch": 0.8279130273183423,
      "grad_norm": 0.6822273135185242,
      "learning_rate": 0.00012714615606890435,
      "loss": 1.0274,
      "step": 4455
    },
    {
      "epoch": 0.8280988663817135,
      "grad_norm": 0.5292199850082397,
      "learning_rate": 0.00012711800723014675,
      "loss": 1.2045,
      "step": 4456
    },
    {
      "epoch": 0.8282847054450846,
      "grad_norm": 0.5441524982452393,
      "learning_rate": 0.0001270898560719274,
      "loss": 1.0811,
      "step": 4457
    },
    {
      "epoch": 0.8284705445084557,
      "grad_norm": 0.6144747138023376,
      "learning_rate": 0.0001270617025966541,
      "loss": 1.1179,
      "step": 4458
    },
    {
      "epoch": 0.8286563835718268,
      "grad_norm": 0.6470378041267395,
      "learning_rate": 0.00012703354680673485,
      "loss": 1.0405,
      "step": 4459
    },
    {
      "epoch": 0.8288422226351979,
      "grad_norm": 0.8334536552429199,
      "learning_rate": 0.00012700538870457796,
      "loss": 1.1096,
      "step": 4460
    },
    {
      "epoch": 0.829028061698569,
      "grad_norm": 0.6929413080215454,
      "learning_rate": 0.0001269772282925918,
      "loss": 0.9405,
      "step": 4461
    },
    {
      "epoch": 0.8292139007619401,
      "grad_norm": 0.7080973982810974,
      "learning_rate": 0.000126949065573185,
      "loss": 1.0241,
      "step": 4462
    },
    {
      "epoch": 0.8293997398253112,
      "grad_norm": 0.5871732234954834,
      "learning_rate": 0.0001269209005487664,
      "loss": 0.9457,
      "step": 4463
    },
    {
      "epoch": 0.8295855788886825,
      "grad_norm": 3.319326639175415,
      "learning_rate": 0.00012689273322174493,
      "loss": 2.2102,
      "step": 4464
    },
    {
      "epoch": 0.8297714179520536,
      "grad_norm": 0.7650028467178345,
      "learning_rate": 0.00012686456359452993,
      "loss": 0.8271,
      "step": 4465
    },
    {
      "epoch": 0.8299572570154247,
      "grad_norm": 0.7382581830024719,
      "learning_rate": 0.0001268363916695307,
      "loss": 1.135,
      "step": 4466
    },
    {
      "epoch": 0.8301430960787958,
      "grad_norm": 0.5419159531593323,
      "learning_rate": 0.0001268082174491569,
      "loss": 0.6879,
      "step": 4467
    },
    {
      "epoch": 0.8303289351421669,
      "grad_norm": 0.5916446447372437,
      "learning_rate": 0.00012678004093581835,
      "loss": 1.0661,
      "step": 4468
    },
    {
      "epoch": 0.830514774205538,
      "grad_norm": 0.6891621351242065,
      "learning_rate": 0.000126751862131925,
      "loss": 1.1768,
      "step": 4469
    },
    {
      "epoch": 0.8307006132689091,
      "grad_norm": 0.5724967122077942,
      "learning_rate": 0.00012672368103988705,
      "loss": 1.1755,
      "step": 4470
    },
    {
      "epoch": 0.8308864523322802,
      "grad_norm": 0.6919474601745605,
      "learning_rate": 0.00012669549766211493,
      "loss": 1.6258,
      "step": 4471
    },
    {
      "epoch": 0.8310722913956514,
      "grad_norm": 0.5278167724609375,
      "learning_rate": 0.00012666731200101916,
      "loss": 0.9755,
      "step": 4472
    },
    {
      "epoch": 0.8312581304590225,
      "grad_norm": 0.6204642653465271,
      "learning_rate": 0.00012663912405901057,
      "loss": 1.1226,
      "step": 4473
    },
    {
      "epoch": 0.8314439695223936,
      "grad_norm": 0.5770162343978882,
      "learning_rate": 0.00012661093383850014,
      "loss": 1.0097,
      "step": 4474
    },
    {
      "epoch": 0.8316298085857647,
      "grad_norm": 0.6649433970451355,
      "learning_rate": 0.00012658274134189903,
      "loss": 1.1369,
      "step": 4475
    },
    {
      "epoch": 0.8318156476491358,
      "grad_norm": 0.70816570520401,
      "learning_rate": 0.00012655454657161858,
      "loss": 1.0995,
      "step": 4476
    },
    {
      "epoch": 0.832001486712507,
      "grad_norm": 0.5342292785644531,
      "learning_rate": 0.00012652634953007037,
      "loss": 1.1424,
      "step": 4477
    },
    {
      "epoch": 0.8321873257758781,
      "grad_norm": 0.47927406430244446,
      "learning_rate": 0.0001264981502196662,
      "loss": 0.9822,
      "step": 4478
    },
    {
      "epoch": 0.8323731648392492,
      "grad_norm": 0.5756170153617859,
      "learning_rate": 0.0001264699486428179,
      "loss": 1.1123,
      "step": 4479
    },
    {
      "epoch": 0.8325590039026203,
      "grad_norm": 0.5887073874473572,
      "learning_rate": 0.0001264417448019377,
      "loss": 1.3087,
      "step": 4480
    },
    {
      "epoch": 0.8327448429659915,
      "grad_norm": 0.6383137702941895,
      "learning_rate": 0.00012641353869943797,
      "loss": 1.0229,
      "step": 4481
    },
    {
      "epoch": 0.8329306820293626,
      "grad_norm": 0.4909496605396271,
      "learning_rate": 0.00012638533033773114,
      "loss": 1.0681,
      "step": 4482
    },
    {
      "epoch": 0.8331165210927337,
      "grad_norm": 0.6577247977256775,
      "learning_rate": 0.00012635711971923,
      "loss": 1.0942,
      "step": 4483
    },
    {
      "epoch": 0.8333023601561048,
      "grad_norm": 0.5949928760528564,
      "learning_rate": 0.00012632890684634746,
      "loss": 1.0045,
      "step": 4484
    },
    {
      "epoch": 0.8334881992194759,
      "grad_norm": 0.6063514351844788,
      "learning_rate": 0.0001263006917214966,
      "loss": 0.9075,
      "step": 4485
    },
    {
      "epoch": 0.833674038282847,
      "grad_norm": 0.774755597114563,
      "learning_rate": 0.0001262724743470907,
      "loss": 1.1189,
      "step": 4486
    },
    {
      "epoch": 0.8338598773462181,
      "grad_norm": 0.5430833697319031,
      "learning_rate": 0.0001262442547255433,
      "loss": 1.0773,
      "step": 4487
    },
    {
      "epoch": 0.8340457164095892,
      "grad_norm": 0.5545639395713806,
      "learning_rate": 0.0001262160328592681,
      "loss": 0.8883,
      "step": 4488
    },
    {
      "epoch": 0.8342315554729605,
      "grad_norm": 0.5910719633102417,
      "learning_rate": 0.00012618780875067894,
      "loss": 0.9673,
      "step": 4489
    },
    {
      "epoch": 0.8344173945363316,
      "grad_norm": 0.5268061757087708,
      "learning_rate": 0.0001261595824021899,
      "loss": 0.9255,
      "step": 4490
    },
    {
      "epoch": 0.8346032335997027,
      "grad_norm": 0.7021967768669128,
      "learning_rate": 0.00012613135381621526,
      "loss": 1.0996,
      "step": 4491
    },
    {
      "epoch": 0.8347890726630738,
      "grad_norm": 0.4780430793762207,
      "learning_rate": 0.0001261031229951694,
      "loss": 0.8551,
      "step": 4492
    },
    {
      "epoch": 0.8349749117264449,
      "grad_norm": 0.5368556976318359,
      "learning_rate": 0.00012607488994146704,
      "loss": 0.9341,
      "step": 4493
    },
    {
      "epoch": 0.835160750789816,
      "grad_norm": 0.6084364056587219,
      "learning_rate": 0.00012604665465752298,
      "loss": 1.0467,
      "step": 4494
    },
    {
      "epoch": 0.8353465898531871,
      "grad_norm": 0.5446452498435974,
      "learning_rate": 0.00012601841714575228,
      "loss": 1.1537,
      "step": 4495
    },
    {
      "epoch": 0.8355324289165582,
      "grad_norm": 0.5627488493919373,
      "learning_rate": 0.0001259901774085701,
      "loss": 0.9802,
      "step": 4496
    },
    {
      "epoch": 0.8357182679799294,
      "grad_norm": 0.6170056462287903,
      "learning_rate": 0.00012596193544839188,
      "loss": 1.1555,
      "step": 4497
    },
    {
      "epoch": 0.8359041070433005,
      "grad_norm": 0.6171839237213135,
      "learning_rate": 0.00012593369126763322,
      "loss": 1.1423,
      "step": 4498
    },
    {
      "epoch": 0.8360899461066716,
      "grad_norm": 0.6114042401313782,
      "learning_rate": 0.00012590544486870987,
      "loss": 0.9804,
      "step": 4499
    },
    {
      "epoch": 0.8362757851700428,
      "grad_norm": 0.6916267275810242,
      "learning_rate": 0.00012587719625403786,
      "loss": 1.1785,
      "step": 4500
    },
    {
      "epoch": 0.8364616242334139,
      "grad_norm": 0.650570273399353,
      "learning_rate": 0.0001258489454260333,
      "loss": 0.9031,
      "step": 4501
    },
    {
      "epoch": 0.836647463296785,
      "grad_norm": 0.6784061789512634,
      "learning_rate": 0.00012582069238711258,
      "loss": 0.9206,
      "step": 4502
    },
    {
      "epoch": 0.8368333023601561,
      "grad_norm": 0.66778564453125,
      "learning_rate": 0.00012579243713969223,
      "loss": 0.9456,
      "step": 4503
    },
    {
      "epoch": 0.8370191414235272,
      "grad_norm": 0.6047230362892151,
      "learning_rate": 0.00012576417968618902,
      "loss": 1.2266,
      "step": 4504
    },
    {
      "epoch": 0.8372049804868984,
      "grad_norm": 0.5893747806549072,
      "learning_rate": 0.00012573592002901982,
      "loss": 1.2147,
      "step": 4505
    },
    {
      "epoch": 0.8373908195502695,
      "grad_norm": 0.5346536636352539,
      "learning_rate": 0.00012570765817060175,
      "loss": 1.0172,
      "step": 4506
    },
    {
      "epoch": 0.8375766586136406,
      "grad_norm": 0.6562202572822571,
      "learning_rate": 0.0001256793941133521,
      "loss": 0.8101,
      "step": 4507
    },
    {
      "epoch": 0.8377624976770117,
      "grad_norm": 0.59455806016922,
      "learning_rate": 0.0001256511278596884,
      "loss": 1.0672,
      "step": 4508
    },
    {
      "epoch": 0.8379483367403828,
      "grad_norm": 0.6395320892333984,
      "learning_rate": 0.00012562285941202826,
      "loss": 1.0752,
      "step": 4509
    },
    {
      "epoch": 0.8381341758037539,
      "grad_norm": 0.594071626663208,
      "learning_rate": 0.00012559458877278961,
      "loss": 1.1381,
      "step": 4510
    },
    {
      "epoch": 0.838320014867125,
      "grad_norm": 0.7788839936256409,
      "learning_rate": 0.00012556631594439047,
      "loss": 1.0234,
      "step": 4511
    },
    {
      "epoch": 0.8385058539304961,
      "grad_norm": 0.6716021299362183,
      "learning_rate": 0.00012553804092924905,
      "loss": 1.3516,
      "step": 4512
    },
    {
      "epoch": 0.8386916929938674,
      "grad_norm": 0.5156703591346741,
      "learning_rate": 0.0001255097637297838,
      "loss": 0.8481,
      "step": 4513
    },
    {
      "epoch": 0.8388775320572385,
      "grad_norm": 0.5573685169219971,
      "learning_rate": 0.0001254814843484133,
      "loss": 0.9987,
      "step": 4514
    },
    {
      "epoch": 0.8390633711206096,
      "grad_norm": 0.5675413608551025,
      "learning_rate": 0.00012545320278755642,
      "loss": 1.0343,
      "step": 4515
    },
    {
      "epoch": 0.8392492101839807,
      "grad_norm": 0.6181820631027222,
      "learning_rate": 0.00012542491904963206,
      "loss": 1.3209,
      "step": 4516
    },
    {
      "epoch": 0.8394350492473518,
      "grad_norm": 0.6297406554222107,
      "learning_rate": 0.00012539663313705944,
      "loss": 1.0885,
      "step": 4517
    },
    {
      "epoch": 0.8396208883107229,
      "grad_norm": 0.5502396821975708,
      "learning_rate": 0.00012536834505225787,
      "loss": 0.9801,
      "step": 4518
    },
    {
      "epoch": 0.839806727374094,
      "grad_norm": 0.6094892024993896,
      "learning_rate": 0.00012534005479764696,
      "loss": 0.9609,
      "step": 4519
    },
    {
      "epoch": 0.8399925664374651,
      "grad_norm": 0.48954108357429504,
      "learning_rate": 0.00012531176237564636,
      "loss": 0.8465,
      "step": 4520
    },
    {
      "epoch": 0.8401784055008362,
      "grad_norm": 0.5575180649757385,
      "learning_rate": 0.00012528346778867603,
      "loss": 1.0366,
      "step": 4521
    },
    {
      "epoch": 0.8403642445642074,
      "grad_norm": 0.6245620250701904,
      "learning_rate": 0.00012525517103915607,
      "loss": 1.0974,
      "step": 4522
    },
    {
      "epoch": 0.8405500836275785,
      "grad_norm": 0.7433310747146606,
      "learning_rate": 0.00012522687212950675,
      "loss": 1.2692,
      "step": 4523
    },
    {
      "epoch": 0.8407359226909497,
      "grad_norm": 0.5833187699317932,
      "learning_rate": 0.0001251985710621485,
      "loss": 1.2388,
      "step": 4524
    },
    {
      "epoch": 0.8409217617543208,
      "grad_norm": 0.6372852921485901,
      "learning_rate": 0.00012517026783950204,
      "loss": 1.0083,
      "step": 4525
    },
    {
      "epoch": 0.8411076008176919,
      "grad_norm": 0.5090671181678772,
      "learning_rate": 0.00012514196246398815,
      "loss": 0.8843,
      "step": 4526
    },
    {
      "epoch": 0.841293439881063,
      "grad_norm": 0.6583060026168823,
      "learning_rate": 0.0001251136549380279,
      "loss": 1.0564,
      "step": 4527
    },
    {
      "epoch": 0.8414792789444341,
      "grad_norm": 0.6386400461196899,
      "learning_rate": 0.0001250853452640424,
      "loss": 1.0424,
      "step": 4528
    },
    {
      "epoch": 0.8416651180078052,
      "grad_norm": 0.576964795589447,
      "learning_rate": 0.00012505703344445316,
      "loss": 1.1201,
      "step": 4529
    },
    {
      "epoch": 0.8418509570711764,
      "grad_norm": 0.6124757528305054,
      "learning_rate": 0.0001250287194816817,
      "loss": 1.1534,
      "step": 4530
    },
    {
      "epoch": 0.8420367961345475,
      "grad_norm": 0.5833020210266113,
      "learning_rate": 0.00012500040337814973,
      "loss": 0.9472,
      "step": 4531
    },
    {
      "epoch": 0.8422226351979186,
      "grad_norm": 0.556035041809082,
      "learning_rate": 0.00012497208513627926,
      "loss": 1.0616,
      "step": 4532
    },
    {
      "epoch": 0.8424084742612897,
      "grad_norm": 0.5945399403572083,
      "learning_rate": 0.00012494376475849237,
      "loss": 1.2011,
      "step": 4533
    },
    {
      "epoch": 0.8425943133246608,
      "grad_norm": 0.5541200041770935,
      "learning_rate": 0.00012491544224721136,
      "loss": 1.0238,
      "step": 4534
    },
    {
      "epoch": 0.842780152388032,
      "grad_norm": 0.5658981204032898,
      "learning_rate": 0.00012488711760485873,
      "loss": 1.1086,
      "step": 4535
    },
    {
      "epoch": 0.842965991451403,
      "grad_norm": 0.5613319277763367,
      "learning_rate": 0.00012485879083385713,
      "loss": 1.2952,
      "step": 4536
    },
    {
      "epoch": 0.8431518305147742,
      "grad_norm": 0.5648307204246521,
      "learning_rate": 0.00012483046193662945,
      "loss": 1.1365,
      "step": 4537
    },
    {
      "epoch": 0.8433376695781454,
      "grad_norm": 0.5618223547935486,
      "learning_rate": 0.00012480213091559868,
      "loss": 1.2456,
      "step": 4538
    },
    {
      "epoch": 0.8435235086415165,
      "grad_norm": 0.4825553297996521,
      "learning_rate": 0.00012477379777318804,
      "loss": 0.8803,
      "step": 4539
    },
    {
      "epoch": 0.8437093477048876,
      "grad_norm": 0.5788162350654602,
      "learning_rate": 0.00012474546251182094,
      "loss": 0.9082,
      "step": 4540
    },
    {
      "epoch": 0.8438951867682587,
      "grad_norm": 0.5924403667449951,
      "learning_rate": 0.00012471712513392096,
      "loss": 0.9885,
      "step": 4541
    },
    {
      "epoch": 0.8440810258316298,
      "grad_norm": 0.5519680976867676,
      "learning_rate": 0.00012468878564191182,
      "loss": 0.9727,
      "step": 4542
    },
    {
      "epoch": 0.8442668648950009,
      "grad_norm": 0.594264030456543,
      "learning_rate": 0.0001246604440382175,
      "loss": 0.9876,
      "step": 4543
    },
    {
      "epoch": 0.844452703958372,
      "grad_norm": 0.5373499989509583,
      "learning_rate": 0.0001246321003252621,
      "loss": 1.0743,
      "step": 4544
    },
    {
      "epoch": 0.8446385430217431,
      "grad_norm": 0.5786815881729126,
      "learning_rate": 0.00012460375450546994,
      "loss": 1.1302,
      "step": 4545
    },
    {
      "epoch": 0.8448243820851143,
      "grad_norm": 0.7555760145187378,
      "learning_rate": 0.00012457540658126546,
      "loss": 1.288,
      "step": 4546
    },
    {
      "epoch": 0.8450102211484855,
      "grad_norm": 0.7258514165878296,
      "learning_rate": 0.00012454705655507334,
      "loss": 1.1075,
      "step": 4547
    },
    {
      "epoch": 0.8451960602118566,
      "grad_norm": 0.49702054262161255,
      "learning_rate": 0.00012451870442931845,
      "loss": 0.9364,
      "step": 4548
    },
    {
      "epoch": 0.8453818992752277,
      "grad_norm": 0.5207595825195312,
      "learning_rate": 0.00012449035020642573,
      "loss": 0.957,
      "step": 4549
    },
    {
      "epoch": 0.8455677383385988,
      "grad_norm": 0.5674048662185669,
      "learning_rate": 0.00012446199388882047,
      "loss": 1.0289,
      "step": 4550
    },
    {
      "epoch": 0.8457535774019699,
      "grad_norm": 0.5538954734802246,
      "learning_rate": 0.000124433635478928,
      "loss": 1.0743,
      "step": 4551
    },
    {
      "epoch": 0.845939416465341,
      "grad_norm": 0.6047095060348511,
      "learning_rate": 0.00012440527497917386,
      "loss": 1.2751,
      "step": 4552
    },
    {
      "epoch": 0.8461252555287121,
      "grad_norm": 0.5031073689460754,
      "learning_rate": 0.00012437691239198379,
      "loss": 0.9235,
      "step": 4553
    },
    {
      "epoch": 0.8463110945920832,
      "grad_norm": 0.7293369770050049,
      "learning_rate": 0.00012434854771978374,
      "loss": 1.1129,
      "step": 4554
    },
    {
      "epoch": 0.8464969336554544,
      "grad_norm": 0.7080215215682983,
      "learning_rate": 0.0001243201809649998,
      "loss": 1.1998,
      "step": 4555
    },
    {
      "epoch": 0.8466827727188255,
      "grad_norm": 0.6878696084022522,
      "learning_rate": 0.00012429181213005817,
      "loss": 1.146,
      "step": 4556
    },
    {
      "epoch": 0.8468686117821966,
      "grad_norm": 0.6264361143112183,
      "learning_rate": 0.00012426344121738538,
      "loss": 1.1465,
      "step": 4557
    },
    {
      "epoch": 0.8470544508455677,
      "grad_norm": 0.8152109384536743,
      "learning_rate": 0.00012423506822940804,
      "loss": 1.2251,
      "step": 4558
    },
    {
      "epoch": 0.8472402899089388,
      "grad_norm": 0.5891522765159607,
      "learning_rate": 0.0001242066931685529,
      "loss": 0.9574,
      "step": 4559
    },
    {
      "epoch": 0.84742612897231,
      "grad_norm": 0.524802029132843,
      "learning_rate": 0.00012417831603724698,
      "loss": 0.9342,
      "step": 4560
    },
    {
      "epoch": 0.8476119680356811,
      "grad_norm": 0.56324702501297,
      "learning_rate": 0.00012414993683791746,
      "loss": 1.0783,
      "step": 4561
    },
    {
      "epoch": 0.8477978070990522,
      "grad_norm": 0.5322441458702087,
      "learning_rate": 0.00012412155557299162,
      "loss": 0.8267,
      "step": 4562
    },
    {
      "epoch": 0.8479836461624234,
      "grad_norm": 0.6256717443466187,
      "learning_rate": 0.00012409317224489698,
      "loss": 1.0611,
      "step": 4563
    },
    {
      "epoch": 0.8481694852257945,
      "grad_norm": 0.5989834666252136,
      "learning_rate": 0.00012406478685606128,
      "loss": 1.1289,
      "step": 4564
    },
    {
      "epoch": 0.8483553242891656,
      "grad_norm": 0.5230849385261536,
      "learning_rate": 0.00012403639940891233,
      "loss": 1.0077,
      "step": 4565
    },
    {
      "epoch": 0.8485411633525367,
      "grad_norm": 0.6140810251235962,
      "learning_rate": 0.00012400800990587822,
      "loss": 0.97,
      "step": 4566
    },
    {
      "epoch": 0.8487270024159078,
      "grad_norm": 0.5292965769767761,
      "learning_rate": 0.00012397961834938713,
      "loss": 0.9602,
      "step": 4567
    },
    {
      "epoch": 0.8489128414792789,
      "grad_norm": 0.7063047289848328,
      "learning_rate": 0.00012395122474186742,
      "loss": 1.0562,
      "step": 4568
    },
    {
      "epoch": 0.84909868054265,
      "grad_norm": 0.6304416060447693,
      "learning_rate": 0.0001239228290857477,
      "loss": 0.9928,
      "step": 4569
    },
    {
      "epoch": 0.8492845196060211,
      "grad_norm": 0.6376837491989136,
      "learning_rate": 0.00012389443138345672,
      "loss": 1.3453,
      "step": 4570
    },
    {
      "epoch": 0.8494703586693924,
      "grad_norm": 0.5010297298431396,
      "learning_rate": 0.00012386603163742338,
      "loss": 0.7436,
      "step": 4571
    },
    {
      "epoch": 0.8496561977327635,
      "grad_norm": 0.5087937116622925,
      "learning_rate": 0.00012383762985007678,
      "loss": 1.0635,
      "step": 4572
    },
    {
      "epoch": 0.8498420367961346,
      "grad_norm": 0.49811092019081116,
      "learning_rate": 0.00012380922602384616,
      "loss": 1.0109,
      "step": 4573
    },
    {
      "epoch": 0.8500278758595057,
      "grad_norm": 0.6603782176971436,
      "learning_rate": 0.000123780820161161,
      "loss": 1.0893,
      "step": 4574
    },
    {
      "epoch": 0.8502137149228768,
      "grad_norm": 0.6798657178878784,
      "learning_rate": 0.00012375241226445088,
      "loss": 1.2919,
      "step": 4575
    },
    {
      "epoch": 0.8503995539862479,
      "grad_norm": 0.6117194294929504,
      "learning_rate": 0.0001237240023361456,
      "loss": 1.1965,
      "step": 4576
    },
    {
      "epoch": 0.850585393049619,
      "grad_norm": 0.621590793132782,
      "learning_rate": 0.00012369559037867512,
      "loss": 1.0154,
      "step": 4577
    },
    {
      "epoch": 0.8507712321129901,
      "grad_norm": 0.6007041335105896,
      "learning_rate": 0.0001236671763944696,
      "loss": 1.2543,
      "step": 4578
    },
    {
      "epoch": 0.8509570711763613,
      "grad_norm": 0.5366314649581909,
      "learning_rate": 0.00012363876038595933,
      "loss": 1.0525,
      "step": 4579
    },
    {
      "epoch": 0.8511429102397324,
      "grad_norm": 0.5456633567810059,
      "learning_rate": 0.00012361034235557482,
      "loss": 1.0457,
      "step": 4580
    },
    {
      "epoch": 0.8513287493031035,
      "grad_norm": 0.5707237720489502,
      "learning_rate": 0.00012358192230574666,
      "loss": 1.1579,
      "step": 4581
    },
    {
      "epoch": 0.8515145883664746,
      "grad_norm": 2.736707925796509,
      "learning_rate": 0.00012355350023890573,
      "loss": 2.0639,
      "step": 4582
    },
    {
      "epoch": 0.8517004274298458,
      "grad_norm": 0.7945477962493896,
      "learning_rate": 0.00012352507615748302,
      "loss": 0.9384,
      "step": 4583
    },
    {
      "epoch": 0.8518862664932169,
      "grad_norm": 0.5095336437225342,
      "learning_rate": 0.00012349665006390967,
      "loss": 1.0477,
      "step": 4584
    },
    {
      "epoch": 0.852072105556588,
      "grad_norm": 0.6436726450920105,
      "learning_rate": 0.0001234682219606171,
      "loss": 1.2062,
      "step": 4585
    },
    {
      "epoch": 0.8522579446199591,
      "grad_norm": 0.625449001789093,
      "learning_rate": 0.00012343979185003678,
      "loss": 1.1338,
      "step": 4586
    },
    {
      "epoch": 0.8524437836833302,
      "grad_norm": 0.6556345224380493,
      "learning_rate": 0.0001234113597346004,
      "loss": 1.0076,
      "step": 4587
    },
    {
      "epoch": 0.8526296227467014,
      "grad_norm": 0.6119154095649719,
      "learning_rate": 0.0001233829256167398,
      "loss": 1.1119,
      "step": 4588
    },
    {
      "epoch": 0.8528154618100725,
      "grad_norm": 0.7122162580490112,
      "learning_rate": 0.00012335448949888706,
      "loss": 1.0975,
      "step": 4589
    },
    {
      "epoch": 0.8530013008734436,
      "grad_norm": 0.472249835729599,
      "learning_rate": 0.00012332605138347434,
      "loss": 0.7964,
      "step": 4590
    },
    {
      "epoch": 0.8531871399368147,
      "grad_norm": 0.5455360412597656,
      "learning_rate": 0.00012329761127293404,
      "loss": 1.0847,
      "step": 4591
    },
    {
      "epoch": 0.8533729790001858,
      "grad_norm": 0.5112693309783936,
      "learning_rate": 0.0001232691691696987,
      "loss": 0.9491,
      "step": 4592
    },
    {
      "epoch": 0.8535588180635569,
      "grad_norm": 0.5319565534591675,
      "learning_rate": 0.000123240725076201,
      "loss": 1.2175,
      "step": 4593
    },
    {
      "epoch": 0.853744657126928,
      "grad_norm": 0.5167549252510071,
      "learning_rate": 0.0001232122789948739,
      "loss": 0.7037,
      "step": 4594
    },
    {
      "epoch": 0.8539304961902991,
      "grad_norm": 0.6013134121894836,
      "learning_rate": 0.0001231838309281504,
      "loss": 1.1435,
      "step": 4595
    },
    {
      "epoch": 0.8541163352536704,
      "grad_norm": 0.5791240334510803,
      "learning_rate": 0.00012315538087846373,
      "loss": 1.2345,
      "step": 4596
    },
    {
      "epoch": 0.8543021743170415,
      "grad_norm": 0.5475298166275024,
      "learning_rate": 0.00012312692884824725,
      "loss": 1.1015,
      "step": 4597
    },
    {
      "epoch": 0.8544880133804126,
      "grad_norm": 0.6367332935333252,
      "learning_rate": 0.0001230984748399346,
      "loss": 0.9356,
      "step": 4598
    },
    {
      "epoch": 0.8546738524437837,
      "grad_norm": 0.6773303151130676,
      "learning_rate": 0.00012307001885595946,
      "loss": 1.1413,
      "step": 4599
    },
    {
      "epoch": 0.8548596915071548,
      "grad_norm": 0.4587186276912689,
      "learning_rate": 0.00012304156089875575,
      "loss": 0.8376,
      "step": 4600
    },
    {
      "epoch": 0.8550455305705259,
      "grad_norm": 0.6566428542137146,
      "learning_rate": 0.00012301310097075755,
      "loss": 1.0847,
      "step": 4601
    },
    {
      "epoch": 0.855231369633897,
      "grad_norm": 0.5402763485908508,
      "learning_rate": 0.00012298463907439909,
      "loss": 0.841,
      "step": 4602
    },
    {
      "epoch": 0.8554172086972681,
      "grad_norm": 0.6253241896629333,
      "learning_rate": 0.00012295617521211478,
      "loss": 0.9708,
      "step": 4603
    },
    {
      "epoch": 0.8556030477606393,
      "grad_norm": 0.5683613419532776,
      "learning_rate": 0.00012292770938633916,
      "loss": 0.928,
      "step": 4604
    },
    {
      "epoch": 0.8557888868240104,
      "grad_norm": 0.5675630569458008,
      "learning_rate": 0.000122899241599507,
      "loss": 1.0236,
      "step": 4605
    },
    {
      "epoch": 0.8559747258873815,
      "grad_norm": 0.5932213068008423,
      "learning_rate": 0.00012287077185405321,
      "loss": 1.2021,
      "step": 4606
    },
    {
      "epoch": 0.8561605649507527,
      "grad_norm": 0.7319698929786682,
      "learning_rate": 0.0001228423001524129,
      "loss": 0.9715,
      "step": 4607
    },
    {
      "epoch": 0.8563464040141238,
      "grad_norm": 0.6660357713699341,
      "learning_rate": 0.00012281382649702129,
      "loss": 1.1203,
      "step": 4608
    },
    {
      "epoch": 0.8565322430774949,
      "grad_norm": 0.6667255163192749,
      "learning_rate": 0.00012278535089031378,
      "loss": 1.2861,
      "step": 4609
    },
    {
      "epoch": 0.856718082140866,
      "grad_norm": 0.707119345664978,
      "learning_rate": 0.00012275687333472598,
      "loss": 1.2471,
      "step": 4610
    },
    {
      "epoch": 0.8569039212042371,
      "grad_norm": 0.6665619015693665,
      "learning_rate": 0.00012272839383269357,
      "loss": 1.1468,
      "step": 4611
    },
    {
      "epoch": 0.8570897602676083,
      "grad_norm": 0.6399442553520203,
      "learning_rate": 0.00012269991238665254,
      "loss": 1.0455,
      "step": 4612
    },
    {
      "epoch": 0.8572755993309794,
      "grad_norm": 0.607695996761322,
      "learning_rate": 0.00012267142899903893,
      "loss": 1.1208,
      "step": 4613
    },
    {
      "epoch": 0.8574614383943505,
      "grad_norm": 0.6243822574615479,
      "learning_rate": 0.00012264294367228903,
      "loss": 1.0788,
      "step": 4614
    },
    {
      "epoch": 0.8576472774577216,
      "grad_norm": 0.5382381677627563,
      "learning_rate": 0.00012261445640883922,
      "loss": 0.8672,
      "step": 4615
    },
    {
      "epoch": 0.8578331165210927,
      "grad_norm": 0.5765889883041382,
      "learning_rate": 0.00012258596721112608,
      "loss": 0.9701,
      "step": 4616
    },
    {
      "epoch": 0.8580189555844638,
      "grad_norm": 0.577860951423645,
      "learning_rate": 0.00012255747608158634,
      "loss": 1.1989,
      "step": 4617
    },
    {
      "epoch": 0.858204794647835,
      "grad_norm": 0.7148776054382324,
      "learning_rate": 0.00012252898302265693,
      "loss": 1.0507,
      "step": 4618
    },
    {
      "epoch": 0.858390633711206,
      "grad_norm": 0.6425639390945435,
      "learning_rate": 0.00012250048803677486,
      "loss": 0.9726,
      "step": 4619
    },
    {
      "epoch": 0.8585764727745773,
      "grad_norm": 0.5930686593055725,
      "learning_rate": 0.00012247199112637745,
      "loss": 1.0762,
      "step": 4620
    },
    {
      "epoch": 0.8587623118379484,
      "grad_norm": 0.5590258240699768,
      "learning_rate": 0.00012244349229390206,
      "loss": 1.0442,
      "step": 4621
    },
    {
      "epoch": 0.8589481509013195,
      "grad_norm": 0.5646161437034607,
      "learning_rate": 0.00012241499154178628,
      "loss": 1.048,
      "step": 4622
    },
    {
      "epoch": 0.8591339899646906,
      "grad_norm": 0.674748957157135,
      "learning_rate": 0.00012238648887246783,
      "loss": 1.1122,
      "step": 4623
    },
    {
      "epoch": 0.8593198290280617,
      "grad_norm": 0.811205267906189,
      "learning_rate": 0.0001223579842883846,
      "loss": 1.111,
      "step": 4624
    },
    {
      "epoch": 0.8595056680914328,
      "grad_norm": 0.6467451453208923,
      "learning_rate": 0.00012232947779197467,
      "loss": 1.0201,
      "step": 4625
    },
    {
      "epoch": 0.8596915071548039,
      "grad_norm": 0.5541864633560181,
      "learning_rate": 0.00012230096938567621,
      "loss": 1.0174,
      "step": 4626
    },
    {
      "epoch": 0.859877346218175,
      "grad_norm": 0.5226763486862183,
      "learning_rate": 0.00012227245907192765,
      "loss": 0.7749,
      "step": 4627
    },
    {
      "epoch": 0.8600631852815461,
      "grad_norm": 0.5585047602653503,
      "learning_rate": 0.00012224394685316753,
      "loss": 1.1309,
      "step": 4628
    },
    {
      "epoch": 0.8602490243449173,
      "grad_norm": 0.5343335866928101,
      "learning_rate": 0.00012221543273183459,
      "loss": 0.9149,
      "step": 4629
    },
    {
      "epoch": 0.8604348634082885,
      "grad_norm": 0.6110146045684814,
      "learning_rate": 0.00012218691671036765,
      "loss": 1.2287,
      "step": 4630
    },
    {
      "epoch": 0.8606207024716596,
      "grad_norm": 0.6890124678611755,
      "learning_rate": 0.00012215839879120577,
      "loss": 1.1387,
      "step": 4631
    },
    {
      "epoch": 0.8608065415350307,
      "grad_norm": 0.511184573173523,
      "learning_rate": 0.00012212987897678815,
      "loss": 0.9056,
      "step": 4632
    },
    {
      "epoch": 0.8609923805984018,
      "grad_norm": 0.6041712760925293,
      "learning_rate": 0.00012210135726955414,
      "loss": 1.0268,
      "step": 4633
    },
    {
      "epoch": 0.8611782196617729,
      "grad_norm": 0.7157412171363831,
      "learning_rate": 0.00012207283367194325,
      "loss": 1.3321,
      "step": 4634
    },
    {
      "epoch": 0.861364058725144,
      "grad_norm": 0.6423292756080627,
      "learning_rate": 0.00012204430818639525,
      "loss": 1.1673,
      "step": 4635
    },
    {
      "epoch": 0.8615498977885151,
      "grad_norm": 0.6783196926116943,
      "learning_rate": 0.00012201578081534992,
      "loss": 0.8651,
      "step": 4636
    },
    {
      "epoch": 0.8617357368518863,
      "grad_norm": 0.6569127440452576,
      "learning_rate": 0.00012198725156124727,
      "loss": 1.0608,
      "step": 4637
    },
    {
      "epoch": 0.8619215759152574,
      "grad_norm": 0.5302185416221619,
      "learning_rate": 0.00012195872042652745,
      "loss": 0.7633,
      "step": 4638
    },
    {
      "epoch": 0.8621074149786285,
      "grad_norm": 0.5998936295509338,
      "learning_rate": 0.00012193018741363084,
      "loss": 1.1865,
      "step": 4639
    },
    {
      "epoch": 0.8622932540419996,
      "grad_norm": 0.5633741617202759,
      "learning_rate": 0.00012190165252499787,
      "loss": 1.2283,
      "step": 4640
    },
    {
      "epoch": 0.8624790931053707,
      "grad_norm": 0.5973196625709534,
      "learning_rate": 0.00012187311576306926,
      "loss": 1.1307,
      "step": 4641
    },
    {
      "epoch": 0.8626649321687418,
      "grad_norm": 0.6969146132469177,
      "learning_rate": 0.00012184457713028575,
      "loss": 1.0946,
      "step": 4642
    },
    {
      "epoch": 0.862850771232113,
      "grad_norm": 0.5954970717430115,
      "learning_rate": 0.00012181603662908835,
      "loss": 1.2309,
      "step": 4643
    },
    {
      "epoch": 0.8630366102954841,
      "grad_norm": 0.5562779307365417,
      "learning_rate": 0.00012178749426191822,
      "loss": 1.0062,
      "step": 4644
    },
    {
      "epoch": 0.8632224493588553,
      "grad_norm": 0.5769143104553223,
      "learning_rate": 0.0001217589500312166,
      "loss": 1.0062,
      "step": 4645
    },
    {
      "epoch": 0.8634082884222264,
      "grad_norm": 0.5674699544906616,
      "learning_rate": 0.00012173040393942496,
      "loss": 1.0515,
      "step": 4646
    },
    {
      "epoch": 0.8635941274855975,
      "grad_norm": 0.5838675498962402,
      "learning_rate": 0.00012170185598898489,
      "loss": 1.2049,
      "step": 4647
    },
    {
      "epoch": 0.8637799665489686,
      "grad_norm": 0.6601033210754395,
      "learning_rate": 0.00012167330618233819,
      "loss": 0.896,
      "step": 4648
    },
    {
      "epoch": 0.8639658056123397,
      "grad_norm": 0.6087871789932251,
      "learning_rate": 0.00012164475452192674,
      "loss": 1.0509,
      "step": 4649
    },
    {
      "epoch": 0.8641516446757108,
      "grad_norm": 0.6676885485649109,
      "learning_rate": 0.00012161620101019269,
      "loss": 0.9619,
      "step": 4650
    },
    {
      "epoch": 0.8643374837390819,
      "grad_norm": 0.7278332710266113,
      "learning_rate": 0.00012158764564957823,
      "loss": 1.2801,
      "step": 4651
    },
    {
      "epoch": 0.864523322802453,
      "grad_norm": 0.6235469579696655,
      "learning_rate": 0.0001215590884425258,
      "loss": 1.1447,
      "step": 4652
    },
    {
      "epoch": 0.8647091618658242,
      "grad_norm": 0.6712271571159363,
      "learning_rate": 0.00012153052939147793,
      "loss": 0.9625,
      "step": 4653
    },
    {
      "epoch": 0.8648950009291954,
      "grad_norm": 0.6502339839935303,
      "learning_rate": 0.00012150196849887734,
      "loss": 1.2098,
      "step": 4654
    },
    {
      "epoch": 0.8650808399925665,
      "grad_norm": 0.6269952058792114,
      "learning_rate": 0.00012147340576716693,
      "loss": 1.0694,
      "step": 4655
    },
    {
      "epoch": 0.8652666790559376,
      "grad_norm": 0.6714789271354675,
      "learning_rate": 0.00012144484119878967,
      "loss": 1.0536,
      "step": 4656
    },
    {
      "epoch": 0.8654525181193087,
      "grad_norm": 0.602345883846283,
      "learning_rate": 0.00012141627479618885,
      "loss": 0.8036,
      "step": 4657
    },
    {
      "epoch": 0.8656383571826798,
      "grad_norm": 0.5539709329605103,
      "learning_rate": 0.00012138770656180774,
      "loss": 1.0422,
      "step": 4658
    },
    {
      "epoch": 0.8658241962460509,
      "grad_norm": 0.5680656433105469,
      "learning_rate": 0.00012135913649808985,
      "loss": 1.0184,
      "step": 4659
    },
    {
      "epoch": 0.866010035309422,
      "grad_norm": 0.6097427606582642,
      "learning_rate": 0.00012133056460747887,
      "loss": 0.7894,
      "step": 4660
    },
    {
      "epoch": 0.8661958743727931,
      "grad_norm": 0.4813399612903595,
      "learning_rate": 0.00012130199089241856,
      "loss": 0.8799,
      "step": 4661
    },
    {
      "epoch": 0.8663817134361643,
      "grad_norm": 0.6142000555992126,
      "learning_rate": 0.00012127341535535297,
      "loss": 1.0662,
      "step": 4662
    },
    {
      "epoch": 0.8665675524995354,
      "grad_norm": 0.6486777663230896,
      "learning_rate": 0.00012124483799872619,
      "loss": 0.9835,
      "step": 4663
    },
    {
      "epoch": 0.8667533915629065,
      "grad_norm": 0.6400505900382996,
      "learning_rate": 0.00012121625882498247,
      "loss": 1.1977,
      "step": 4664
    },
    {
      "epoch": 0.8669392306262776,
      "grad_norm": 0.6632579565048218,
      "learning_rate": 0.00012118767783656629,
      "loss": 1.1879,
      "step": 4665
    },
    {
      "epoch": 0.8671250696896488,
      "grad_norm": 0.4944020211696625,
      "learning_rate": 0.00012115909503592224,
      "loss": 0.8731,
      "step": 4666
    },
    {
      "epoch": 0.8673109087530199,
      "grad_norm": 0.5768524408340454,
      "learning_rate": 0.00012113051042549506,
      "loss": 1.1558,
      "step": 4667
    },
    {
      "epoch": 0.867496747816391,
      "grad_norm": 0.647330105304718,
      "learning_rate": 0.00012110192400772964,
      "loss": 1.038,
      "step": 4668
    },
    {
      "epoch": 0.8676825868797621,
      "grad_norm": 0.6742308139801025,
      "learning_rate": 0.00012107333578507108,
      "loss": 1.2654,
      "step": 4669
    },
    {
      "epoch": 0.8678684259431333,
      "grad_norm": 0.5209482312202454,
      "learning_rate": 0.00012104474575996454,
      "loss": 1.0273,
      "step": 4670
    },
    {
      "epoch": 0.8680542650065044,
      "grad_norm": 0.6833703517913818,
      "learning_rate": 0.00012101615393485539,
      "loss": 1.0157,
      "step": 4671
    },
    {
      "epoch": 0.8682401040698755,
      "grad_norm": 0.583544909954071,
      "learning_rate": 0.00012098756031218921,
      "loss": 0.957,
      "step": 4672
    },
    {
      "epoch": 0.8684259431332466,
      "grad_norm": 0.5707135796546936,
      "learning_rate": 0.00012095896489441163,
      "loss": 1.0744,
      "step": 4673
    },
    {
      "epoch": 0.8686117821966177,
      "grad_norm": 0.6191301941871643,
      "learning_rate": 0.00012093036768396849,
      "loss": 1.2617,
      "step": 4674
    },
    {
      "epoch": 0.8687976212599888,
      "grad_norm": 3.5594120025634766,
      "learning_rate": 0.00012090176868330574,
      "loss": 1.8595,
      "step": 4675
    },
    {
      "epoch": 0.8689834603233599,
      "grad_norm": 0.5981330871582031,
      "learning_rate": 0.00012087316789486958,
      "loss": 1.1521,
      "step": 4676
    },
    {
      "epoch": 0.869169299386731,
      "grad_norm": 0.5382730960845947,
      "learning_rate": 0.00012084456532110624,
      "loss": 0.8385,
      "step": 4677
    },
    {
      "epoch": 0.8693551384501023,
      "grad_norm": 0.599388599395752,
      "learning_rate": 0.00012081596096446215,
      "loss": 1.1152,
      "step": 4678
    },
    {
      "epoch": 0.8695409775134734,
      "grad_norm": 0.5467658042907715,
      "learning_rate": 0.00012078735482738395,
      "loss": 1.1479,
      "step": 4679
    },
    {
      "epoch": 0.8697268165768445,
      "grad_norm": 0.5545910000801086,
      "learning_rate": 0.00012075874691231837,
      "loss": 1.158,
      "step": 4680
    },
    {
      "epoch": 0.8699126556402156,
      "grad_norm": 0.5862640738487244,
      "learning_rate": 0.00012073013722171227,
      "loss": 1.2503,
      "step": 4681
    },
    {
      "epoch": 0.8700984947035867,
      "grad_norm": 0.6473658084869385,
      "learning_rate": 0.00012070152575801276,
      "loss": 0.9587,
      "step": 4682
    },
    {
      "epoch": 0.8702843337669578,
      "grad_norm": 0.48189878463745117,
      "learning_rate": 0.00012067291252366702,
      "loss": 0.9056,
      "step": 4683
    },
    {
      "epoch": 0.8704701728303289,
      "grad_norm": 0.6109098196029663,
      "learning_rate": 0.00012064429752112238,
      "loss": 1.1096,
      "step": 4684
    },
    {
      "epoch": 0.8706560118937,
      "grad_norm": 1.7112163305282593,
      "learning_rate": 0.00012061568075282634,
      "loss": 1.7075,
      "step": 4685
    },
    {
      "epoch": 0.8708418509570712,
      "grad_norm": 0.5883646011352539,
      "learning_rate": 0.00012058706222122657,
      "loss": 0.9264,
      "step": 4686
    },
    {
      "epoch": 0.8710276900204423,
      "grad_norm": 0.5947024822235107,
      "learning_rate": 0.0001205584419287709,
      "loss": 0.9626,
      "step": 4687
    },
    {
      "epoch": 0.8712135290838134,
      "grad_norm": 0.7193747758865356,
      "learning_rate": 0.00012052981987790724,
      "loss": 1.4634,
      "step": 4688
    },
    {
      "epoch": 0.8713993681471845,
      "grad_norm": 0.5391495823860168,
      "learning_rate": 0.00012050119607108372,
      "loss": 1.0638,
      "step": 4689
    },
    {
      "epoch": 0.8715852072105557,
      "grad_norm": 0.5702479481697083,
      "learning_rate": 0.00012047257051074861,
      "loss": 1.0789,
      "step": 4690
    },
    {
      "epoch": 0.8717710462739268,
      "grad_norm": 0.6023498177528381,
      "learning_rate": 0.00012044394319935026,
      "loss": 1.0356,
      "step": 4691
    },
    {
      "epoch": 0.8719568853372979,
      "grad_norm": 0.6121519804000854,
      "learning_rate": 0.00012041531413933728,
      "loss": 1.028,
      "step": 4692
    },
    {
      "epoch": 0.872142724400669,
      "grad_norm": 0.5752167701721191,
      "learning_rate": 0.00012038668333315835,
      "loss": 1.0593,
      "step": 4693
    },
    {
      "epoch": 0.8723285634640402,
      "grad_norm": 0.5538478493690491,
      "learning_rate": 0.00012035805078326235,
      "loss": 1.1715,
      "step": 4694
    },
    {
      "epoch": 0.8725144025274113,
      "grad_norm": 0.5828766822814941,
      "learning_rate": 0.00012032941649209824,
      "loss": 1.1091,
      "step": 4695
    },
    {
      "epoch": 0.8727002415907824,
      "grad_norm": 0.7071311473846436,
      "learning_rate": 0.00012030078046211523,
      "loss": 0.8833,
      "step": 4696
    },
    {
      "epoch": 0.8728860806541535,
      "grad_norm": 0.5653672218322754,
      "learning_rate": 0.00012027214269576258,
      "loss": 0.9883,
      "step": 4697
    },
    {
      "epoch": 0.8730719197175246,
      "grad_norm": 0.6679391860961914,
      "learning_rate": 0.00012024350319548976,
      "loss": 1.2291,
      "step": 4698
    },
    {
      "epoch": 0.8732577587808957,
      "grad_norm": 0.6757986545562744,
      "learning_rate": 0.00012021486196374636,
      "loss": 1.0122,
      "step": 4699
    },
    {
      "epoch": 0.8734435978442668,
      "grad_norm": 0.5338380932807922,
      "learning_rate": 0.00012018621900298211,
      "loss": 0.9106,
      "step": 4700
    },
    {
      "epoch": 0.873629436907638,
      "grad_norm": 0.5537553429603577,
      "learning_rate": 0.00012015757431564691,
      "loss": 1.0227,
      "step": 4701
    },
    {
      "epoch": 0.873815275971009,
      "grad_norm": 0.5925634503364563,
      "learning_rate": 0.00012012892790419083,
      "loss": 0.861,
      "step": 4702
    },
    {
      "epoch": 0.8740011150343803,
      "grad_norm": 0.7933574914932251,
      "learning_rate": 0.00012010027977106404,
      "loss": 1.3481,
      "step": 4703
    },
    {
      "epoch": 0.8741869540977514,
      "grad_norm": 0.6480244398117065,
      "learning_rate": 0.0001200716299187169,
      "loss": 1.0297,
      "step": 4704
    },
    {
      "epoch": 0.8743727931611225,
      "grad_norm": 0.5648460388183594,
      "learning_rate": 0.00012004297834959986,
      "loss": 1.3159,
      "step": 4705
    },
    {
      "epoch": 0.8745586322244936,
      "grad_norm": 0.6062881350517273,
      "learning_rate": 0.00012001432506616354,
      "loss": 0.954,
      "step": 4706
    },
    {
      "epoch": 0.8747444712878647,
      "grad_norm": 0.5937121510505676,
      "learning_rate": 0.00011998567007085874,
      "loss": 1.0219,
      "step": 4707
    },
    {
      "epoch": 0.8749303103512358,
      "grad_norm": 0.6274353265762329,
      "learning_rate": 0.00011995701336613643,
      "loss": 1.0806,
      "step": 4708
    },
    {
      "epoch": 0.8751161494146069,
      "grad_norm": 0.6611014008522034,
      "learning_rate": 0.00011992835495444758,
      "loss": 1.177,
      "step": 4709
    },
    {
      "epoch": 0.875301988477978,
      "grad_norm": 0.7381565570831299,
      "learning_rate": 0.00011989969483824349,
      "loss": 1.0864,
      "step": 4710
    },
    {
      "epoch": 0.8754878275413492,
      "grad_norm": 0.7044957876205444,
      "learning_rate": 0.00011987103301997548,
      "loss": 1.16,
      "step": 4711
    },
    {
      "epoch": 0.8756736666047203,
      "grad_norm": 0.617867112159729,
      "learning_rate": 0.00011984236950209508,
      "loss": 1.2796,
      "step": 4712
    },
    {
      "epoch": 0.8758595056680915,
      "grad_norm": 0.7515525817871094,
      "learning_rate": 0.00011981370428705394,
      "loss": 1.1917,
      "step": 4713
    },
    {
      "epoch": 0.8760453447314626,
      "grad_norm": 0.5283318161964417,
      "learning_rate": 0.00011978503737730384,
      "loss": 0.9867,
      "step": 4714
    },
    {
      "epoch": 0.8762311837948337,
      "grad_norm": 0.6197617650032043,
      "learning_rate": 0.00011975636877529673,
      "loss": 1.0821,
      "step": 4715
    },
    {
      "epoch": 0.8764170228582048,
      "grad_norm": 0.547167956829071,
      "learning_rate": 0.00011972769848348466,
      "loss": 1.1527,
      "step": 4716
    },
    {
      "epoch": 0.8766028619215759,
      "grad_norm": 0.614820122718811,
      "learning_rate": 0.00011969902650431995,
      "loss": 1.0837,
      "step": 4717
    },
    {
      "epoch": 0.876788700984947,
      "grad_norm": 0.5309334993362427,
      "learning_rate": 0.00011967035284025492,
      "loss": 0.8339,
      "step": 4718
    },
    {
      "epoch": 0.8769745400483182,
      "grad_norm": 0.6812418103218079,
      "learning_rate": 0.00011964167749374211,
      "loss": 1.1013,
      "step": 4719
    },
    {
      "epoch": 0.8771603791116893,
      "grad_norm": 0.5131886005401611,
      "learning_rate": 0.00011961300046723416,
      "loss": 0.961,
      "step": 4720
    },
    {
      "epoch": 0.8773462181750604,
      "grad_norm": 0.5880168676376343,
      "learning_rate": 0.0001195843217631839,
      "loss": 1.1434,
      "step": 4721
    },
    {
      "epoch": 0.8775320572384315,
      "grad_norm": 0.6584733724594116,
      "learning_rate": 0.00011955564138404425,
      "loss": 0.9723,
      "step": 4722
    },
    {
      "epoch": 0.8777178963018026,
      "grad_norm": 0.592173159122467,
      "learning_rate": 0.00011952695933226832,
      "loss": 1.375,
      "step": 4723
    },
    {
      "epoch": 0.8779037353651737,
      "grad_norm": 0.6685372591018677,
      "learning_rate": 0.00011949827561030937,
      "loss": 0.9787,
      "step": 4724
    },
    {
      "epoch": 0.8780895744285449,
      "grad_norm": 0.6100190281867981,
      "learning_rate": 0.00011946959022062078,
      "loss": 1.146,
      "step": 4725
    },
    {
      "epoch": 0.878275413491916,
      "grad_norm": 0.5250088572502136,
      "learning_rate": 0.00011944090316565605,
      "loss": 0.9508,
      "step": 4726
    },
    {
      "epoch": 0.8784612525552872,
      "grad_norm": 0.4799655079841614,
      "learning_rate": 0.00011941221444786887,
      "loss": 0.8568,
      "step": 4727
    },
    {
      "epoch": 0.8786470916186583,
      "grad_norm": 0.6235849857330322,
      "learning_rate": 0.00011938352406971303,
      "loss": 1.1144,
      "step": 4728
    },
    {
      "epoch": 0.8788329306820294,
      "grad_norm": 0.6182804107666016,
      "learning_rate": 0.00011935483203364243,
      "loss": 1.1426,
      "step": 4729
    },
    {
      "epoch": 0.8790187697454005,
      "grad_norm": 0.601972758769989,
      "learning_rate": 0.00011932613834211123,
      "loss": 1.0688,
      "step": 4730
    },
    {
      "epoch": 0.8792046088087716,
      "grad_norm": 0.6925641894340515,
      "learning_rate": 0.00011929744299757369,
      "loss": 0.9577,
      "step": 4731
    },
    {
      "epoch": 0.8793904478721427,
      "grad_norm": 0.6421496272087097,
      "learning_rate": 0.00011926874600248415,
      "loss": 1.1401,
      "step": 4732
    },
    {
      "epoch": 0.8795762869355138,
      "grad_norm": 0.5793421268463135,
      "learning_rate": 0.00011924004735929708,
      "loss": 1.1798,
      "step": 4733
    },
    {
      "epoch": 0.8797621259988849,
      "grad_norm": 0.5658807158470154,
      "learning_rate": 0.0001192113470704672,
      "loss": 0.894,
      "step": 4734
    },
    {
      "epoch": 0.879947965062256,
      "grad_norm": 0.6436983942985535,
      "learning_rate": 0.00011918264513844928,
      "loss": 1.3174,
      "step": 4735
    },
    {
      "epoch": 0.8801338041256273,
      "grad_norm": 0.6653681993484497,
      "learning_rate": 0.00011915394156569824,
      "loss": 1.2076,
      "step": 4736
    },
    {
      "epoch": 0.8803196431889984,
      "grad_norm": 0.6940290927886963,
      "learning_rate": 0.00011912523635466922,
      "loss": 1.2932,
      "step": 4737
    },
    {
      "epoch": 0.8805054822523695,
      "grad_norm": 0.5589834451675415,
      "learning_rate": 0.00011909652950781735,
      "loss": 1.0923,
      "step": 4738
    },
    {
      "epoch": 0.8806913213157406,
      "grad_norm": 0.7990158200263977,
      "learning_rate": 0.00011906782102759808,
      "loss": 0.9076,
      "step": 4739
    },
    {
      "epoch": 0.8808771603791117,
      "grad_norm": 0.6897647380828857,
      "learning_rate": 0.00011903911091646684,
      "loss": 1.1487,
      "step": 4740
    },
    {
      "epoch": 0.8810629994424828,
      "grad_norm": 0.6083943247795105,
      "learning_rate": 0.00011901039917687932,
      "loss": 1.106,
      "step": 4741
    },
    {
      "epoch": 0.8812488385058539,
      "grad_norm": 0.7685713171958923,
      "learning_rate": 0.00011898168581129125,
      "loss": 1.1485,
      "step": 4742
    },
    {
      "epoch": 0.881434677569225,
      "grad_norm": 0.6682976484298706,
      "learning_rate": 0.0001189529708221586,
      "loss": 1.1954,
      "step": 4743
    },
    {
      "epoch": 0.8816205166325962,
      "grad_norm": 0.6300533413887024,
      "learning_rate": 0.00011892425421193734,
      "loss": 0.9926,
      "step": 4744
    },
    {
      "epoch": 0.8818063556959673,
      "grad_norm": 0.6247344613075256,
      "learning_rate": 0.00011889553598308374,
      "loss": 1.1316,
      "step": 4745
    },
    {
      "epoch": 0.8819921947593384,
      "grad_norm": 0.6410258412361145,
      "learning_rate": 0.00011886681613805411,
      "loss": 1.2052,
      "step": 4746
    },
    {
      "epoch": 0.8821780338227095,
      "grad_norm": 0.7149096727371216,
      "learning_rate": 0.00011883809467930494,
      "loss": 0.9866,
      "step": 4747
    },
    {
      "epoch": 0.8823638728860806,
      "grad_norm": 0.5473355650901794,
      "learning_rate": 0.0001188093716092928,
      "loss": 0.9347,
      "step": 4748
    },
    {
      "epoch": 0.8825497119494518,
      "grad_norm": 0.7847073078155518,
      "learning_rate": 0.00011878064693047447,
      "loss": 1.3707,
      "step": 4749
    },
    {
      "epoch": 0.8827355510128229,
      "grad_norm": 0.6489022374153137,
      "learning_rate": 0.0001187519206453068,
      "loss": 1.1489,
      "step": 4750
    },
    {
      "epoch": 0.882921390076194,
      "grad_norm": 0.6962729096412659,
      "learning_rate": 0.00011872319275624682,
      "loss": 1.0444,
      "step": 4751
    },
    {
      "epoch": 0.8831072291395652,
      "grad_norm": 0.5032536387443542,
      "learning_rate": 0.00011869446326575169,
      "loss": 0.9388,
      "step": 4752
    },
    {
      "epoch": 0.8832930682029363,
      "grad_norm": 0.5279998779296875,
      "learning_rate": 0.00011866573217627872,
      "loss": 1.1328,
      "step": 4753
    },
    {
      "epoch": 0.8834789072663074,
      "grad_norm": 0.6550498008728027,
      "learning_rate": 0.00011863699949028534,
      "loss": 1.0532,
      "step": 4754
    },
    {
      "epoch": 0.8836647463296785,
      "grad_norm": 0.5503681898117065,
      "learning_rate": 0.00011860826521022914,
      "loss": 1.1704,
      "step": 4755
    },
    {
      "epoch": 0.8838505853930496,
      "grad_norm": 0.7000329494476318,
      "learning_rate": 0.0001185795293385678,
      "loss": 0.8628,
      "step": 4756
    },
    {
      "epoch": 0.8840364244564207,
      "grad_norm": 0.597890317440033,
      "learning_rate": 0.00011855079187775912,
      "loss": 1.1687,
      "step": 4757
    },
    {
      "epoch": 0.8842222635197918,
      "grad_norm": 0.6725625395774841,
      "learning_rate": 0.00011852205283026115,
      "loss": 1.0957,
      "step": 4758
    },
    {
      "epoch": 0.8844081025831629,
      "grad_norm": 0.5943846106529236,
      "learning_rate": 0.00011849331219853195,
      "loss": 1.0617,
      "step": 4759
    },
    {
      "epoch": 0.8845939416465342,
      "grad_norm": 0.6239001750946045,
      "learning_rate": 0.00011846456998502977,
      "loss": 1.1119,
      "step": 4760
    },
    {
      "epoch": 0.8847797807099053,
      "grad_norm": 2.7174880504608154,
      "learning_rate": 0.00011843582619221305,
      "loss": 2.1165,
      "step": 4761
    },
    {
      "epoch": 0.8849656197732764,
      "grad_norm": 0.5697281360626221,
      "learning_rate": 0.00011840708082254026,
      "loss": 1.1697,
      "step": 4762
    },
    {
      "epoch": 0.8851514588366475,
      "grad_norm": 0.6207005977630615,
      "learning_rate": 0.00011837833387847006,
      "loss": 1.0138,
      "step": 4763
    },
    {
      "epoch": 0.8853372979000186,
      "grad_norm": 0.83244788646698,
      "learning_rate": 0.00011834958536246126,
      "loss": 1.237,
      "step": 4764
    },
    {
      "epoch": 0.8855231369633897,
      "grad_norm": 0.6972323060035706,
      "learning_rate": 0.00011832083527697273,
      "loss": 1.122,
      "step": 4765
    },
    {
      "epoch": 0.8857089760267608,
      "grad_norm": 0.5799198150634766,
      "learning_rate": 0.00011829208362446358,
      "loss": 0.9897,
      "step": 4766
    },
    {
      "epoch": 0.8858948150901319,
      "grad_norm": 0.6393404603004456,
      "learning_rate": 0.00011826333040739296,
      "loss": 0.9211,
      "step": 4767
    },
    {
      "epoch": 0.886080654153503,
      "grad_norm": 0.5956047773361206,
      "learning_rate": 0.00011823457562822027,
      "loss": 1.0445,
      "step": 4768
    },
    {
      "epoch": 0.8862664932168742,
      "grad_norm": 0.6870486736297607,
      "learning_rate": 0.00011820581928940488,
      "loss": 1.0243,
      "step": 4769
    },
    {
      "epoch": 0.8864523322802453,
      "grad_norm": 0.7557746171951294,
      "learning_rate": 0.00011817706139340643,
      "loss": 1.0572,
      "step": 4770
    },
    {
      "epoch": 0.8866381713436164,
      "grad_norm": 0.652820348739624,
      "learning_rate": 0.00011814830194268465,
      "loss": 1.0466,
      "step": 4771
    },
    {
      "epoch": 0.8868240104069876,
      "grad_norm": 0.6921366453170776,
      "learning_rate": 0.00011811954093969935,
      "loss": 1.1726,
      "step": 4772
    },
    {
      "epoch": 0.8870098494703587,
      "grad_norm": 0.6798288226127625,
      "learning_rate": 0.00011809077838691057,
      "loss": 1.1054,
      "step": 4773
    },
    {
      "epoch": 0.8871956885337298,
      "grad_norm": 0.5644552111625671,
      "learning_rate": 0.0001180620142867784,
      "loss": 1.0747,
      "step": 4774
    },
    {
      "epoch": 0.8873815275971009,
      "grad_norm": 0.7290164828300476,
      "learning_rate": 0.0001180332486417631,
      "loss": 1.4222,
      "step": 4775
    },
    {
      "epoch": 0.887567366660472,
      "grad_norm": 0.5924285650253296,
      "learning_rate": 0.0001180044814543251,
      "loss": 1.1481,
      "step": 4776
    },
    {
      "epoch": 0.8877532057238432,
      "grad_norm": 0.5664063096046448,
      "learning_rate": 0.00011797571272692489,
      "loss": 1.0801,
      "step": 4777
    },
    {
      "epoch": 0.8879390447872143,
      "grad_norm": 0.620488166809082,
      "learning_rate": 0.00011794694246202311,
      "loss": 0.9766,
      "step": 4778
    },
    {
      "epoch": 0.8881248838505854,
      "grad_norm": 0.6897662281990051,
      "learning_rate": 0.00011791817066208053,
      "loss": 1.0818,
      "step": 4779
    },
    {
      "epoch": 0.8883107229139565,
      "grad_norm": 0.6509714126586914,
      "learning_rate": 0.0001178893973295581,
      "loss": 1.0346,
      "step": 4780
    },
    {
      "epoch": 0.8884965619773276,
      "grad_norm": 0.615665853023529,
      "learning_rate": 0.00011786062246691685,
      "loss": 1.0627,
      "step": 4781
    },
    {
      "epoch": 0.8886824010406987,
      "grad_norm": 0.5649377107620239,
      "learning_rate": 0.00011783184607661799,
      "loss": 0.8559,
      "step": 4782
    },
    {
      "epoch": 0.8888682401040698,
      "grad_norm": 0.6220218539237976,
      "learning_rate": 0.00011780306816112277,
      "loss": 1.0661,
      "step": 4783
    },
    {
      "epoch": 0.889054079167441,
      "grad_norm": 0.641806960105896,
      "learning_rate": 0.00011777428872289267,
      "loss": 1.0956,
      "step": 4784
    },
    {
      "epoch": 0.8892399182308122,
      "grad_norm": 0.6103681921958923,
      "learning_rate": 0.00011774550776438922,
      "loss": 1.1282,
      "step": 4785
    },
    {
      "epoch": 0.8894257572941833,
      "grad_norm": 0.6140033602714539,
      "learning_rate": 0.00011771672528807414,
      "loss": 1.0339,
      "step": 4786
    },
    {
      "epoch": 0.8896115963575544,
      "grad_norm": 0.6332442164421082,
      "learning_rate": 0.00011768794129640925,
      "loss": 1.1501,
      "step": 4787
    },
    {
      "epoch": 0.8897974354209255,
      "grad_norm": 0.6983381509780884,
      "learning_rate": 0.00011765915579185654,
      "loss": 1.1905,
      "step": 4788
    },
    {
      "epoch": 0.8899832744842966,
      "grad_norm": 0.7200410962104797,
      "learning_rate": 0.00011763036877687804,
      "loss": 1.4506,
      "step": 4789
    },
    {
      "epoch": 0.8901691135476677,
      "grad_norm": 0.6325451731681824,
      "learning_rate": 0.00011760158025393599,
      "loss": 1.0697,
      "step": 4790
    },
    {
      "epoch": 0.8903549526110388,
      "grad_norm": 0.5252197980880737,
      "learning_rate": 0.00011757279022549277,
      "loss": 0.9902,
      "step": 4791
    },
    {
      "epoch": 0.8905407916744099,
      "grad_norm": 0.5685787796974182,
      "learning_rate": 0.0001175439986940108,
      "loss": 1.1247,
      "step": 4792
    },
    {
      "epoch": 0.8907266307377811,
      "grad_norm": 0.5685696005821228,
      "learning_rate": 0.0001175152056619527,
      "loss": 1.0047,
      "step": 4793
    },
    {
      "epoch": 0.8909124698011522,
      "grad_norm": 0.5405831933021545,
      "learning_rate": 0.0001174864111317812,
      "loss": 1.0133,
      "step": 4794
    },
    {
      "epoch": 0.8910983088645233,
      "grad_norm": 0.6638238430023193,
      "learning_rate": 0.0001174576151059592,
      "loss": 1.0997,
      "step": 4795
    },
    {
      "epoch": 0.8912841479278945,
      "grad_norm": 0.6298685669898987,
      "learning_rate": 0.00011742881758694961,
      "loss": 1.1055,
      "step": 4796
    },
    {
      "epoch": 0.8914699869912656,
      "grad_norm": 0.6404034495353699,
      "learning_rate": 0.00011740001857721559,
      "loss": 1.057,
      "step": 4797
    },
    {
      "epoch": 0.8916558260546367,
      "grad_norm": 0.547170102596283,
      "learning_rate": 0.0001173712180792204,
      "loss": 1.072,
      "step": 4798
    },
    {
      "epoch": 0.8918416651180078,
      "grad_norm": 0.5974907279014587,
      "learning_rate": 0.00011734241609542738,
      "loss": 1.046,
      "step": 4799
    },
    {
      "epoch": 0.8920275041813789,
      "grad_norm": 0.6565002202987671,
      "learning_rate": 0.00011731361262829998,
      "loss": 1.1398,
      "step": 4800
    },
    {
      "epoch": 0.8922133432447501,
      "grad_norm": 0.6506026387214661,
      "learning_rate": 0.00011728480768030192,
      "loss": 1.0015,
      "step": 4801
    },
    {
      "epoch": 0.8923991823081212,
      "grad_norm": 0.6083167195320129,
      "learning_rate": 0.0001172560012538969,
      "loss": 0.9871,
      "step": 4802
    },
    {
      "epoch": 0.8925850213714923,
      "grad_norm": 0.5694814920425415,
      "learning_rate": 0.00011722719335154877,
      "loss": 1.1955,
      "step": 4803
    },
    {
      "epoch": 0.8927708604348634,
      "grad_norm": 0.5358350872993469,
      "learning_rate": 0.00011719838397572158,
      "loss": 1.0018,
      "step": 4804
    },
    {
      "epoch": 0.8929566994982345,
      "grad_norm": 0.6407806873321533,
      "learning_rate": 0.00011716957312887944,
      "loss": 1.1107,
      "step": 4805
    },
    {
      "epoch": 0.8931425385616056,
      "grad_norm": 0.6780750155448914,
      "learning_rate": 0.00011714076081348661,
      "loss": 1.382,
      "step": 4806
    },
    {
      "epoch": 0.8933283776249767,
      "grad_norm": 0.5691828727722168,
      "learning_rate": 0.00011711194703200743,
      "loss": 1.012,
      "step": 4807
    },
    {
      "epoch": 0.8935142166883479,
      "grad_norm": 0.6187720894813538,
      "learning_rate": 0.00011708313178690646,
      "loss": 1.2116,
      "step": 4808
    },
    {
      "epoch": 0.893700055751719,
      "grad_norm": 0.5498912930488586,
      "learning_rate": 0.00011705431508064831,
      "loss": 0.9507,
      "step": 4809
    },
    {
      "epoch": 0.8938858948150902,
      "grad_norm": 0.5937941670417786,
      "learning_rate": 0.00011702549691569773,
      "loss": 1.0066,
      "step": 4810
    },
    {
      "epoch": 0.8940717338784613,
      "grad_norm": 3.8109848499298096,
      "learning_rate": 0.00011699667729451957,
      "loss": 2.2158,
      "step": 4811
    },
    {
      "epoch": 0.8942575729418324,
      "grad_norm": 0.7385560870170593,
      "learning_rate": 0.00011696785621957891,
      "loss": 1.125,
      "step": 4812
    },
    {
      "epoch": 0.8944434120052035,
      "grad_norm": 0.5790786743164062,
      "learning_rate": 0.00011693903369334083,
      "loss": 0.8837,
      "step": 4813
    },
    {
      "epoch": 0.8946292510685746,
      "grad_norm": 0.6423563957214355,
      "learning_rate": 0.00011691020971827056,
      "loss": 1.0456,
      "step": 4814
    },
    {
      "epoch": 0.8948150901319457,
      "grad_norm": 0.6066467761993408,
      "learning_rate": 0.00011688138429683353,
      "loss": 1.0813,
      "step": 4815
    },
    {
      "epoch": 0.8950009291953168,
      "grad_norm": 0.7395979166030884,
      "learning_rate": 0.0001168525574314952,
      "loss": 1.1017,
      "step": 4816
    },
    {
      "epoch": 0.8951867682586879,
      "grad_norm": 0.7192192077636719,
      "learning_rate": 0.00011682372912472121,
      "loss": 0.8663,
      "step": 4817
    },
    {
      "epoch": 0.8953726073220591,
      "grad_norm": 0.6796870827674866,
      "learning_rate": 0.00011679489937897732,
      "loss": 1.1221,
      "step": 4818
    },
    {
      "epoch": 0.8955584463854303,
      "grad_norm": 0.6365312337875366,
      "learning_rate": 0.00011676606819672938,
      "loss": 1.2095,
      "step": 4819
    },
    {
      "epoch": 0.8957442854488014,
      "grad_norm": 0.6555750370025635,
      "learning_rate": 0.00011673723558044341,
      "loss": 0.9862,
      "step": 4820
    },
    {
      "epoch": 0.8959301245121725,
      "grad_norm": 0.6303730010986328,
      "learning_rate": 0.00011670840153258547,
      "loss": 1.2718,
      "step": 4821
    },
    {
      "epoch": 0.8961159635755436,
      "grad_norm": 0.6512117385864258,
      "learning_rate": 0.00011667956605562185,
      "loss": 1.0232,
      "step": 4822
    },
    {
      "epoch": 0.8963018026389147,
      "grad_norm": 0.6685510277748108,
      "learning_rate": 0.00011665072915201892,
      "loss": 0.8669,
      "step": 4823
    },
    {
      "epoch": 0.8964876417022858,
      "grad_norm": 0.5733498930931091,
      "learning_rate": 0.00011662189082424311,
      "loss": 0.9483,
      "step": 4824
    },
    {
      "epoch": 0.8966734807656569,
      "grad_norm": 0.6234797835350037,
      "learning_rate": 0.00011659305107476104,
      "loss": 1.0886,
      "step": 4825
    },
    {
      "epoch": 0.8968593198290281,
      "grad_norm": 0.6397889852523804,
      "learning_rate": 0.00011656420990603947,
      "loss": 1.1993,
      "step": 4826
    },
    {
      "epoch": 0.8970451588923992,
      "grad_norm": 0.5726931691169739,
      "learning_rate": 0.00011653536732054525,
      "loss": 1.1539,
      "step": 4827
    },
    {
      "epoch": 0.8972309979557703,
      "grad_norm": 0.5924124121665955,
      "learning_rate": 0.00011650652332074527,
      "loss": 1.0823,
      "step": 4828
    },
    {
      "epoch": 0.8974168370191414,
      "grad_norm": 0.5201457738876343,
      "learning_rate": 0.00011647767790910673,
      "loss": 0.925,
      "step": 4829
    },
    {
      "epoch": 0.8976026760825125,
      "grad_norm": 0.6730798482894897,
      "learning_rate": 0.00011644883108809676,
      "loss": 1.306,
      "step": 4830
    },
    {
      "epoch": 0.8977885151458836,
      "grad_norm": 0.6653159856796265,
      "learning_rate": 0.00011641998286018272,
      "loss": 1.1408,
      "step": 4831
    },
    {
      "epoch": 0.8979743542092548,
      "grad_norm": 0.5166033506393433,
      "learning_rate": 0.00011639113322783206,
      "loss": 0.7456,
      "step": 4832
    },
    {
      "epoch": 0.8981601932726259,
      "grad_norm": 0.5565785765647888,
      "learning_rate": 0.00011636228219351235,
      "loss": 1.1999,
      "step": 4833
    },
    {
      "epoch": 0.8983460323359971,
      "grad_norm": 0.5805851817131042,
      "learning_rate": 0.00011633342975969128,
      "loss": 1.134,
      "step": 4834
    },
    {
      "epoch": 0.8985318713993682,
      "grad_norm": 0.5019768476486206,
      "learning_rate": 0.00011630457592883663,
      "loss": 0.9231,
      "step": 4835
    },
    {
      "epoch": 0.8987177104627393,
      "grad_norm": 0.7813116312026978,
      "learning_rate": 0.00011627572070341641,
      "loss": 1.2229,
      "step": 4836
    },
    {
      "epoch": 0.8989035495261104,
      "grad_norm": 0.6496708989143372,
      "learning_rate": 0.0001162468640858986,
      "loss": 0.9698,
      "step": 4837
    },
    {
      "epoch": 0.8990893885894815,
      "grad_norm": 0.5925761461257935,
      "learning_rate": 0.0001162180060787514,
      "loss": 1.0714,
      "step": 4838
    },
    {
      "epoch": 0.8992752276528526,
      "grad_norm": 0.7705389261245728,
      "learning_rate": 0.0001161891466844431,
      "loss": 1.2492,
      "step": 4839
    },
    {
      "epoch": 0.8994610667162237,
      "grad_norm": 2.839663028717041,
      "learning_rate": 0.00011616028590544206,
      "loss": 1.8876,
      "step": 4840
    },
    {
      "epoch": 0.8996469057795948,
      "grad_norm": 0.5334593057632446,
      "learning_rate": 0.00011613142374421686,
      "loss": 1.017,
      "step": 4841
    },
    {
      "epoch": 0.8998327448429659,
      "grad_norm": 0.6567533612251282,
      "learning_rate": 0.00011610256020323609,
      "loss": 1.0719,
      "step": 4842
    },
    {
      "epoch": 0.9000185839063372,
      "grad_norm": 0.6591853499412537,
      "learning_rate": 0.00011607369528496858,
      "loss": 1.0512,
      "step": 4843
    },
    {
      "epoch": 0.9002044229697083,
      "grad_norm": 0.6436012387275696,
      "learning_rate": 0.00011604482899188318,
      "loss": 1.2593,
      "step": 4844
    },
    {
      "epoch": 0.9003902620330794,
      "grad_norm": 0.5583667159080505,
      "learning_rate": 0.00011601596132644887,
      "loss": 1.197,
      "step": 4845
    },
    {
      "epoch": 0.9005761010964505,
      "grad_norm": 0.7685452103614807,
      "learning_rate": 0.00011598709229113476,
      "loss": 1.0777,
      "step": 4846
    },
    {
      "epoch": 0.9007619401598216,
      "grad_norm": 0.560444176197052,
      "learning_rate": 0.0001159582218884101,
      "loss": 0.8902,
      "step": 4847
    },
    {
      "epoch": 0.9009477792231927,
      "grad_norm": 0.5692372918128967,
      "learning_rate": 0.00011592935012074421,
      "loss": 1.107,
      "step": 4848
    },
    {
      "epoch": 0.9011336182865638,
      "grad_norm": 0.5790915489196777,
      "learning_rate": 0.0001159004769906066,
      "loss": 1.0338,
      "step": 4849
    },
    {
      "epoch": 0.9013194573499349,
      "grad_norm": 0.6954670548439026,
      "learning_rate": 0.00011587160250046682,
      "loss": 1.3241,
      "step": 4850
    },
    {
      "epoch": 0.9015052964133061,
      "grad_norm": 0.7290377020835876,
      "learning_rate": 0.00011584272665279461,
      "loss": 1.012,
      "step": 4851
    },
    {
      "epoch": 0.9016911354766772,
      "grad_norm": 0.5656014084815979,
      "learning_rate": 0.00011581384945005971,
      "loss": 0.9303,
      "step": 4852
    },
    {
      "epoch": 0.9018769745400483,
      "grad_norm": 0.7426737546920776,
      "learning_rate": 0.0001157849708947321,
      "loss": 1.1145,
      "step": 4853
    },
    {
      "epoch": 0.9020628136034194,
      "grad_norm": 0.6836064457893372,
      "learning_rate": 0.00011575609098928179,
      "loss": 1.3366,
      "step": 4854
    },
    {
      "epoch": 0.9022486526667906,
      "grad_norm": 0.5508992671966553,
      "learning_rate": 0.00011572720973617901,
      "loss": 0.9922,
      "step": 4855
    },
    {
      "epoch": 0.9024344917301617,
      "grad_norm": 0.6973763108253479,
      "learning_rate": 0.00011569832713789395,
      "loss": 1.1187,
      "step": 4856
    },
    {
      "epoch": 0.9026203307935328,
      "grad_norm": 0.607064962387085,
      "learning_rate": 0.00011566944319689705,
      "loss": 1.0759,
      "step": 4857
    },
    {
      "epoch": 0.9028061698569039,
      "grad_norm": 0.6120737791061401,
      "learning_rate": 0.00011564055791565882,
      "loss": 1.0567,
      "step": 4858
    },
    {
      "epoch": 0.9029920089202751,
      "grad_norm": 0.5953623056411743,
      "learning_rate": 0.00011561167129664987,
      "loss": 1.0374,
      "step": 4859
    },
    {
      "epoch": 0.9031778479836462,
      "grad_norm": 0.6712900400161743,
      "learning_rate": 0.00011558278334234094,
      "loss": 1.3432,
      "step": 4860
    },
    {
      "epoch": 0.9033636870470173,
      "grad_norm": 2.0790538787841797,
      "learning_rate": 0.00011555389405520285,
      "loss": 1.9842,
      "step": 4861
    },
    {
      "epoch": 0.9035495261103884,
      "grad_norm": 0.6319670081138611,
      "learning_rate": 0.00011552500343770658,
      "loss": 1.022,
      "step": 4862
    },
    {
      "epoch": 0.9037353651737595,
      "grad_norm": 0.6280139088630676,
      "learning_rate": 0.0001154961114923232,
      "loss": 0.8457,
      "step": 4863
    },
    {
      "epoch": 0.9039212042371306,
      "grad_norm": 0.5777485370635986,
      "learning_rate": 0.00011546721822152395,
      "loss": 0.8473,
      "step": 4864
    },
    {
      "epoch": 0.9041070433005017,
      "grad_norm": 0.7556338310241699,
      "learning_rate": 0.0001154383236277801,
      "loss": 0.9622,
      "step": 4865
    },
    {
      "epoch": 0.9042928823638728,
      "grad_norm": 0.5529469847679138,
      "learning_rate": 0.00011540942771356305,
      "loss": 1.38,
      "step": 4866
    },
    {
      "epoch": 0.9044787214272441,
      "grad_norm": 0.55649334192276,
      "learning_rate": 0.00011538053048134434,
      "loss": 1.2111,
      "step": 4867
    },
    {
      "epoch": 0.9046645604906152,
      "grad_norm": 0.4702974259853363,
      "learning_rate": 0.00011535163193359562,
      "loss": 0.8591,
      "step": 4868
    },
    {
      "epoch": 0.9048503995539863,
      "grad_norm": 0.673514187335968,
      "learning_rate": 0.00011532273207278862,
      "loss": 1.1053,
      "step": 4869
    },
    {
      "epoch": 0.9050362386173574,
      "grad_norm": 0.47570809721946716,
      "learning_rate": 0.00011529383090139522,
      "loss": 0.7493,
      "step": 4870
    },
    {
      "epoch": 0.9052220776807285,
      "grad_norm": 0.5453622937202454,
      "learning_rate": 0.00011526492842188745,
      "loss": 0.9597,
      "step": 4871
    },
    {
      "epoch": 0.9054079167440996,
      "grad_norm": 0.6614927649497986,
      "learning_rate": 0.00011523602463673734,
      "loss": 0.9449,
      "step": 4872
    },
    {
      "epoch": 0.9055937558074707,
      "grad_norm": 0.5573998093605042,
      "learning_rate": 0.00011520711954841714,
      "loss": 0.9135,
      "step": 4873
    },
    {
      "epoch": 0.9057795948708418,
      "grad_norm": 0.6102240681648254,
      "learning_rate": 0.00011517821315939912,
      "loss": 1.0506,
      "step": 4874
    },
    {
      "epoch": 0.905965433934213,
      "grad_norm": 0.7087007164955139,
      "learning_rate": 0.00011514930547215573,
      "loss": 0.9975,
      "step": 4875
    },
    {
      "epoch": 0.9061512729975841,
      "grad_norm": 0.6605527400970459,
      "learning_rate": 0.00011512039648915947,
      "loss": 1.0797,
      "step": 4876
    },
    {
      "epoch": 0.9063371120609552,
      "grad_norm": 0.5774627327919006,
      "learning_rate": 0.00011509148621288304,
      "loss": 1.0385,
      "step": 4877
    },
    {
      "epoch": 0.9065229511243263,
      "grad_norm": 0.6432042717933655,
      "learning_rate": 0.00011506257464579917,
      "loss": 1.0954,
      "step": 4878
    },
    {
      "epoch": 0.9067087901876975,
      "grad_norm": 0.7714658975601196,
      "learning_rate": 0.00011503366179038075,
      "loss": 1.0423,
      "step": 4879
    },
    {
      "epoch": 0.9068946292510686,
      "grad_norm": 0.5681400299072266,
      "learning_rate": 0.00011500474764910074,
      "loss": 0.9482,
      "step": 4880
    },
    {
      "epoch": 0.9070804683144397,
      "grad_norm": 0.6567079424858093,
      "learning_rate": 0.00011497583222443225,
      "loss": 0.9561,
      "step": 4881
    },
    {
      "epoch": 0.9072663073778108,
      "grad_norm": 0.6021323204040527,
      "learning_rate": 0.00011494691551884844,
      "loss": 0.9268,
      "step": 4882
    },
    {
      "epoch": 0.9074521464411819,
      "grad_norm": 0.6198539733886719,
      "learning_rate": 0.00011491799753482265,
      "loss": 1.1215,
      "step": 4883
    },
    {
      "epoch": 0.9076379855045531,
      "grad_norm": 0.5479195713996887,
      "learning_rate": 0.00011488907827482829,
      "loss": 0.5888,
      "step": 4884
    },
    {
      "epoch": 0.9078238245679242,
      "grad_norm": 0.6176124811172485,
      "learning_rate": 0.00011486015774133885,
      "loss": 1.1016,
      "step": 4885
    },
    {
      "epoch": 0.9080096636312953,
      "grad_norm": 0.49869436025619507,
      "learning_rate": 0.00011483123593682806,
      "loss": 1.005,
      "step": 4886
    },
    {
      "epoch": 0.9081955026946664,
      "grad_norm": 0.6559669971466064,
      "learning_rate": 0.00011480231286376956,
      "loss": 1.0295,
      "step": 4887
    },
    {
      "epoch": 0.9083813417580375,
      "grad_norm": 0.8178464770317078,
      "learning_rate": 0.00011477338852463729,
      "loss": 1.2055,
      "step": 4888
    },
    {
      "epoch": 0.9085671808214086,
      "grad_norm": 0.6112135052680969,
      "learning_rate": 0.00011474446292190515,
      "loss": 1.1929,
      "step": 4889
    },
    {
      "epoch": 0.9087530198847797,
      "grad_norm": 0.6317768096923828,
      "learning_rate": 0.00011471553605804721,
      "loss": 1.1349,
      "step": 4890
    },
    {
      "epoch": 0.9089388589481509,
      "grad_norm": 12.977190017700195,
      "learning_rate": 0.00011468660793553769,
      "loss": 2.2248,
      "step": 4891
    },
    {
      "epoch": 0.9091246980115221,
      "grad_norm": 0.614732563495636,
      "learning_rate": 0.00011465767855685081,
      "loss": 1.1119,
      "step": 4892
    },
    {
      "epoch": 0.9093105370748932,
      "grad_norm": 0.5960726737976074,
      "learning_rate": 0.00011462874792446105,
      "loss": 0.9675,
      "step": 4893
    },
    {
      "epoch": 0.9094963761382643,
      "grad_norm": 0.5411145687103271,
      "learning_rate": 0.00011459981604084287,
      "loss": 0.8035,
      "step": 4894
    },
    {
      "epoch": 0.9096822152016354,
      "grad_norm": 0.7136313319206238,
      "learning_rate": 0.00011457088290847084,
      "loss": 0.9104,
      "step": 4895
    },
    {
      "epoch": 0.9098680542650065,
      "grad_norm": 0.56341552734375,
      "learning_rate": 0.00011454194852981972,
      "loss": 0.9796,
      "step": 4896
    },
    {
      "epoch": 0.9100538933283776,
      "grad_norm": 0.863733172416687,
      "learning_rate": 0.0001145130129073643,
      "loss": 1.2648,
      "step": 4897
    },
    {
      "epoch": 0.9102397323917487,
      "grad_norm": 0.4566417634487152,
      "learning_rate": 0.00011448407604357954,
      "loss": 0.9554,
      "step": 4898
    },
    {
      "epoch": 0.9104255714551198,
      "grad_norm": 0.5524247288703918,
      "learning_rate": 0.00011445513794094042,
      "loss": 1.0031,
      "step": 4899
    },
    {
      "epoch": 0.910611410518491,
      "grad_norm": 0.5676782727241516,
      "learning_rate": 0.00011442619860192212,
      "loss": 1.0623,
      "step": 4900
    },
    {
      "epoch": 0.9107972495818621,
      "grad_norm": 0.4784383475780487,
      "learning_rate": 0.0001143972580289999,
      "loss": 0.6645,
      "step": 4901
    },
    {
      "epoch": 0.9109830886452333,
      "grad_norm": 0.5696948766708374,
      "learning_rate": 0.00011436831622464907,
      "loss": 0.8933,
      "step": 4902
    },
    {
      "epoch": 0.9111689277086044,
      "grad_norm": 0.6131220459938049,
      "learning_rate": 0.00011433937319134511,
      "loss": 1.0181,
      "step": 4903
    },
    {
      "epoch": 0.9113547667719755,
      "grad_norm": 0.5868662595748901,
      "learning_rate": 0.00011431042893156353,
      "loss": 0.9421,
      "step": 4904
    },
    {
      "epoch": 0.9115406058353466,
      "grad_norm": 0.6456340551376343,
      "learning_rate": 0.00011428148344778007,
      "loss": 0.9885,
      "step": 4905
    },
    {
      "epoch": 0.9117264448987177,
      "grad_norm": 0.5382920503616333,
      "learning_rate": 0.00011425253674247045,
      "loss": 0.85,
      "step": 4906
    },
    {
      "epoch": 0.9119122839620888,
      "grad_norm": 0.7104405760765076,
      "learning_rate": 0.00011422358881811056,
      "loss": 1.1623,
      "step": 4907
    },
    {
      "epoch": 0.91209812302546,
      "grad_norm": 0.6414424777030945,
      "learning_rate": 0.00011419463967717638,
      "loss": 1.2485,
      "step": 4908
    },
    {
      "epoch": 0.9122839620888311,
      "grad_norm": 0.6622775197029114,
      "learning_rate": 0.000114165689322144,
      "loss": 0.7905,
      "step": 4909
    },
    {
      "epoch": 0.9124698011522022,
      "grad_norm": 0.5562495589256287,
      "learning_rate": 0.00011413673775548959,
      "loss": 0.9664,
      "step": 4910
    },
    {
      "epoch": 0.9126556402155733,
      "grad_norm": 0.5962749123573303,
      "learning_rate": 0.00011410778497968944,
      "loss": 1.1049,
      "step": 4911
    },
    {
      "epoch": 0.9128414792789444,
      "grad_norm": 0.5652308464050293,
      "learning_rate": 0.00011407883099721994,
      "loss": 0.792,
      "step": 4912
    },
    {
      "epoch": 0.9130273183423155,
      "grad_norm": 0.5293900966644287,
      "learning_rate": 0.00011404987581055761,
      "loss": 1.0509,
      "step": 4913
    },
    {
      "epoch": 0.9132131574056866,
      "grad_norm": 0.6027974486351013,
      "learning_rate": 0.00011402091942217903,
      "loss": 0.9318,
      "step": 4914
    },
    {
      "epoch": 0.9133989964690578,
      "grad_norm": 0.6635562181472778,
      "learning_rate": 0.00011399196183456092,
      "loss": 0.9232,
      "step": 4915
    },
    {
      "epoch": 0.9135848355324289,
      "grad_norm": 0.6220401525497437,
      "learning_rate": 0.00011396300305018008,
      "loss": 1.1271,
      "step": 4916
    },
    {
      "epoch": 0.9137706745958001,
      "grad_norm": 0.6705507040023804,
      "learning_rate": 0.00011393404307151341,
      "loss": 1.15,
      "step": 4917
    },
    {
      "epoch": 0.9139565136591712,
      "grad_norm": 0.6669270992279053,
      "learning_rate": 0.00011390508190103787,
      "loss": 1.0627,
      "step": 4918
    },
    {
      "epoch": 0.9141423527225423,
      "grad_norm": 0.5870175957679749,
      "learning_rate": 0.00011387611954123068,
      "loss": 1.2826,
      "step": 4919
    },
    {
      "epoch": 0.9143281917859134,
      "grad_norm": 0.5145399570465088,
      "learning_rate": 0.00011384715599456898,
      "loss": 0.6593,
      "step": 4920
    },
    {
      "epoch": 0.9145140308492845,
      "grad_norm": 0.6042763590812683,
      "learning_rate": 0.00011381819126353008,
      "loss": 1.2208,
      "step": 4921
    },
    {
      "epoch": 0.9146998699126556,
      "grad_norm": 0.655913770198822,
      "learning_rate": 0.00011378922535059145,
      "loss": 1.1689,
      "step": 4922
    },
    {
      "epoch": 0.9148857089760267,
      "grad_norm": 0.5491055846214294,
      "learning_rate": 0.00011376025825823055,
      "loss": 0.8107,
      "step": 4923
    },
    {
      "epoch": 0.9150715480393978,
      "grad_norm": 0.6189053654670715,
      "learning_rate": 0.00011373128998892504,
      "loss": 1.1629,
      "step": 4924
    },
    {
      "epoch": 0.915257387102769,
      "grad_norm": 0.6688449382781982,
      "learning_rate": 0.0001137023205451526,
      "loss": 1.2761,
      "step": 4925
    },
    {
      "epoch": 0.9154432261661402,
      "grad_norm": 0.6299698352813721,
      "learning_rate": 0.00011367334992939107,
      "loss": 1.113,
      "step": 4926
    },
    {
      "epoch": 0.9156290652295113,
      "grad_norm": 0.60389643907547,
      "learning_rate": 0.00011364437814411839,
      "loss": 1.1595,
      "step": 4927
    },
    {
      "epoch": 0.9158149042928824,
      "grad_norm": 0.49639827013015747,
      "learning_rate": 0.00011361540519181253,
      "loss": 1.085,
      "step": 4928
    },
    {
      "epoch": 0.9160007433562535,
      "grad_norm": 0.5939173102378845,
      "learning_rate": 0.00011358643107495163,
      "loss": 1.3346,
      "step": 4929
    },
    {
      "epoch": 0.9161865824196246,
      "grad_norm": 0.5937512516975403,
      "learning_rate": 0.00011355745579601395,
      "loss": 0.9499,
      "step": 4930
    },
    {
      "epoch": 0.9163724214829957,
      "grad_norm": 0.6250744462013245,
      "learning_rate": 0.00011352847935747777,
      "loss": 1.0898,
      "step": 4931
    },
    {
      "epoch": 0.9165582605463668,
      "grad_norm": 0.5503117442131042,
      "learning_rate": 0.0001134995017618215,
      "loss": 1.0074,
      "step": 4932
    },
    {
      "epoch": 0.916744099609738,
      "grad_norm": 0.7058550119400024,
      "learning_rate": 0.00011347052301152368,
      "loss": 1.1851,
      "step": 4933
    },
    {
      "epoch": 0.9169299386731091,
      "grad_norm": 0.6272091865539551,
      "learning_rate": 0.0001134415431090629,
      "loss": 1.0675,
      "step": 4934
    },
    {
      "epoch": 0.9171157777364802,
      "grad_norm": 0.5527566075325012,
      "learning_rate": 0.0001134125620569179,
      "loss": 1.0778,
      "step": 4935
    },
    {
      "epoch": 0.9173016167998513,
      "grad_norm": 0.65137779712677,
      "learning_rate": 0.00011338357985756745,
      "loss": 0.9924,
      "step": 4936
    },
    {
      "epoch": 0.9174874558632224,
      "grad_norm": 0.5755317807197571,
      "learning_rate": 0.00011335459651349053,
      "loss": 1.0271,
      "step": 4937
    },
    {
      "epoch": 0.9176732949265936,
      "grad_norm": 0.5749362707138062,
      "learning_rate": 0.00011332561202716608,
      "loss": 1.1052,
      "step": 4938
    },
    {
      "epoch": 0.9178591339899647,
      "grad_norm": 0.6218981146812439,
      "learning_rate": 0.00011329662640107322,
      "loss": 0.9382,
      "step": 4939
    },
    {
      "epoch": 0.9180449730533358,
      "grad_norm": 0.5158394575119019,
      "learning_rate": 0.00011326763963769119,
      "loss": 1.2352,
      "step": 4940
    },
    {
      "epoch": 0.918230812116707,
      "grad_norm": 0.6450227499008179,
      "learning_rate": 0.00011323865173949926,
      "loss": 1.0512,
      "step": 4941
    },
    {
      "epoch": 0.9184166511800781,
      "grad_norm": 0.5990210771560669,
      "learning_rate": 0.00011320966270897684,
      "loss": 0.853,
      "step": 4942
    },
    {
      "epoch": 0.9186024902434492,
      "grad_norm": 0.5773190855979919,
      "learning_rate": 0.00011318067254860341,
      "loss": 1.1168,
      "step": 4943
    },
    {
      "epoch": 0.9187883293068203,
      "grad_norm": 0.6192017197608948,
      "learning_rate": 0.00011315168126085857,
      "loss": 0.899,
      "step": 4944
    },
    {
      "epoch": 0.9189741683701914,
      "grad_norm": 0.6629911661148071,
      "learning_rate": 0.00011312268884822201,
      "loss": 1.128,
      "step": 4945
    },
    {
      "epoch": 0.9191600074335625,
      "grad_norm": 0.7068147659301758,
      "learning_rate": 0.0001130936953131735,
      "loss": 1.4354,
      "step": 4946
    },
    {
      "epoch": 0.9193458464969336,
      "grad_norm": 0.6272721886634827,
      "learning_rate": 0.00011306470065819295,
      "loss": 1.0152,
      "step": 4947
    },
    {
      "epoch": 0.9195316855603047,
      "grad_norm": 0.5584651827812195,
      "learning_rate": 0.00011303570488576032,
      "loss": 0.8876,
      "step": 4948
    },
    {
      "epoch": 0.9197175246236758,
      "grad_norm": 0.6266582608222961,
      "learning_rate": 0.00011300670799835567,
      "loss": 0.9633,
      "step": 4949
    },
    {
      "epoch": 0.9199033636870471,
      "grad_norm": 0.5822092294692993,
      "learning_rate": 0.00011297770999845915,
      "loss": 1.1593,
      "step": 4950
    },
    {
      "epoch": 0.9200892027504182,
      "grad_norm": 0.5857718586921692,
      "learning_rate": 0.00011294871088855107,
      "loss": 0.9418,
      "step": 4951
    },
    {
      "epoch": 0.9202750418137893,
      "grad_norm": 0.5828437209129333,
      "learning_rate": 0.00011291971067111178,
      "loss": 0.8059,
      "step": 4952
    },
    {
      "epoch": 0.9204608808771604,
      "grad_norm": 0.5660130977630615,
      "learning_rate": 0.00011289070934862167,
      "loss": 1.1642,
      "step": 4953
    },
    {
      "epoch": 0.9206467199405315,
      "grad_norm": 0.5477927923202515,
      "learning_rate": 0.00011286170692356134,
      "loss": 0.9768,
      "step": 4954
    },
    {
      "epoch": 0.9208325590039026,
      "grad_norm": 0.5076239109039307,
      "learning_rate": 0.00011283270339841144,
      "loss": 0.7493,
      "step": 4955
    },
    {
      "epoch": 0.9210183980672737,
      "grad_norm": 0.5687392354011536,
      "learning_rate": 0.0001128036987756527,
      "loss": 0.9457,
      "step": 4956
    },
    {
      "epoch": 0.9212042371306448,
      "grad_norm": 0.5765700936317444,
      "learning_rate": 0.0001127746930577659,
      "loss": 1.0497,
      "step": 4957
    },
    {
      "epoch": 0.921390076194016,
      "grad_norm": 0.5494324564933777,
      "learning_rate": 0.00011274568624723197,
      "loss": 1.0566,
      "step": 4958
    },
    {
      "epoch": 0.9215759152573871,
      "grad_norm": 0.615958034992218,
      "learning_rate": 0.00011271667834653196,
      "loss": 1.216,
      "step": 4959
    },
    {
      "epoch": 0.9217617543207582,
      "grad_norm": 0.8911369442939758,
      "learning_rate": 0.00011268766935814698,
      "loss": 1.3917,
      "step": 4960
    },
    {
      "epoch": 0.9219475933841293,
      "grad_norm": 0.6073392033576965,
      "learning_rate": 0.00011265865928455824,
      "loss": 1.051,
      "step": 4961
    },
    {
      "epoch": 0.9221334324475005,
      "grad_norm": 0.6068910956382751,
      "learning_rate": 0.00011262964812824699,
      "loss": 0.9606,
      "step": 4962
    },
    {
      "epoch": 0.9223192715108716,
      "grad_norm": 0.5832070708274841,
      "learning_rate": 0.00011260063589169465,
      "loss": 1.1758,
      "step": 4963
    },
    {
      "epoch": 0.9225051105742427,
      "grad_norm": 0.5867427587509155,
      "learning_rate": 0.00011257162257738268,
      "loss": 1.1755,
      "step": 4964
    },
    {
      "epoch": 0.9226909496376138,
      "grad_norm": 0.7116524577140808,
      "learning_rate": 0.00011254260818779263,
      "loss": 1.0624,
      "step": 4965
    },
    {
      "epoch": 0.922876788700985,
      "grad_norm": 0.6223874688148499,
      "learning_rate": 0.00011251359272540624,
      "loss": 1.2023,
      "step": 4966
    },
    {
      "epoch": 0.9230626277643561,
      "grad_norm": 0.6258384585380554,
      "learning_rate": 0.00011248457619270518,
      "loss": 0.9243,
      "step": 4967
    },
    {
      "epoch": 0.9232484668277272,
      "grad_norm": 0.5584441423416138,
      "learning_rate": 0.00011245555859217136,
      "loss": 1.1783,
      "step": 4968
    },
    {
      "epoch": 0.9234343058910983,
      "grad_norm": 0.6257378458976746,
      "learning_rate": 0.00011242653992628672,
      "loss": 1.1093,
      "step": 4969
    },
    {
      "epoch": 0.9236201449544694,
      "grad_norm": 0.6018193364143372,
      "learning_rate": 0.00011239752019753323,
      "loss": 0.9174,
      "step": 4970
    },
    {
      "epoch": 0.9238059840178405,
      "grad_norm": 0.5290696620941162,
      "learning_rate": 0.00011236849940839305,
      "loss": 1.0616,
      "step": 4971
    },
    {
      "epoch": 0.9239918230812116,
      "grad_norm": 0.5852720141410828,
      "learning_rate": 0.00011233947756134835,
      "loss": 1.0018,
      "step": 4972
    },
    {
      "epoch": 0.9241776621445827,
      "grad_norm": 0.5290955901145935,
      "learning_rate": 0.00011231045465888151,
      "loss": 1.2372,
      "step": 4973
    },
    {
      "epoch": 0.924363501207954,
      "grad_norm": 0.674185574054718,
      "learning_rate": 0.00011228143070347486,
      "loss": 0.9144,
      "step": 4974
    },
    {
      "epoch": 0.9245493402713251,
      "grad_norm": 0.7288258671760559,
      "learning_rate": 0.00011225240569761089,
      "loss": 0.9673,
      "step": 4975
    },
    {
      "epoch": 0.9247351793346962,
      "grad_norm": 0.5527795553207397,
      "learning_rate": 0.0001122233796437722,
      "loss": 1.0264,
      "step": 4976
    },
    {
      "epoch": 0.9249210183980673,
      "grad_norm": 0.5867929458618164,
      "learning_rate": 0.00011219435254444141,
      "loss": 1.3047,
      "step": 4977
    },
    {
      "epoch": 0.9251068574614384,
      "grad_norm": 0.6463168859481812,
      "learning_rate": 0.00011216532440210133,
      "loss": 1.1629,
      "step": 4978
    },
    {
      "epoch": 0.9252926965248095,
      "grad_norm": 0.6795656085014343,
      "learning_rate": 0.00011213629521923472,
      "loss": 1.4026,
      "step": 4979
    },
    {
      "epoch": 0.9254785355881806,
      "grad_norm": 0.8002610802650452,
      "learning_rate": 0.00011210726499832456,
      "loss": 1.1641,
      "step": 4980
    },
    {
      "epoch": 0.9256643746515517,
      "grad_norm": 0.6516452431678772,
      "learning_rate": 0.00011207823374185384,
      "loss": 1.3039,
      "step": 4981
    },
    {
      "epoch": 0.9258502137149229,
      "grad_norm": 0.6306900382041931,
      "learning_rate": 0.00011204920145230571,
      "loss": 1.2405,
      "step": 4982
    },
    {
      "epoch": 0.926036052778294,
      "grad_norm": 0.560518741607666,
      "learning_rate": 0.00011202016813216334,
      "loss": 0.8887,
      "step": 4983
    },
    {
      "epoch": 0.9262218918416651,
      "grad_norm": 0.5832896828651428,
      "learning_rate": 0.00011199113378391002,
      "loss": 1.0638,
      "step": 4984
    },
    {
      "epoch": 0.9264077309050363,
      "grad_norm": 0.5987124443054199,
      "learning_rate": 0.00011196209841002909,
      "loss": 1.0041,
      "step": 4985
    },
    {
      "epoch": 0.9265935699684074,
      "grad_norm": 0.5337582230567932,
      "learning_rate": 0.00011193306201300404,
      "loss": 0.8817,
      "step": 4986
    },
    {
      "epoch": 0.9267794090317785,
      "grad_norm": 0.644491970539093,
      "learning_rate": 0.0001119040245953184,
      "loss": 1.319,
      "step": 4987
    },
    {
      "epoch": 0.9269652480951496,
      "grad_norm": 0.7071481347084045,
      "learning_rate": 0.00011187498615945578,
      "loss": 0.9309,
      "step": 4988
    },
    {
      "epoch": 0.9271510871585207,
      "grad_norm": 0.5608466267585754,
      "learning_rate": 0.00011184594670789996,
      "loss": 1.2428,
      "step": 4989
    },
    {
      "epoch": 0.9273369262218918,
      "grad_norm": 0.6157172322273254,
      "learning_rate": 0.00011181690624313472,
      "loss": 0.9522,
      "step": 4990
    },
    {
      "epoch": 0.927522765285263,
      "grad_norm": 0.5183684229850769,
      "learning_rate": 0.00011178786476764393,
      "loss": 0.8553,
      "step": 4991
    },
    {
      "epoch": 0.9277086043486341,
      "grad_norm": 0.6620053052902222,
      "learning_rate": 0.00011175882228391161,
      "loss": 1.3053,
      "step": 4992
    },
    {
      "epoch": 0.9278944434120052,
      "grad_norm": 0.6395905017852783,
      "learning_rate": 0.00011172977879442182,
      "loss": 1.1837,
      "step": 4993
    },
    {
      "epoch": 0.9280802824753763,
      "grad_norm": 0.5325583815574646,
      "learning_rate": 0.00011170073430165864,
      "loss": 1.0923,
      "step": 4994
    },
    {
      "epoch": 0.9282661215387474,
      "grad_norm": 0.5025383830070496,
      "learning_rate": 0.00011167168880810639,
      "loss": 0.8869,
      "step": 4995
    },
    {
      "epoch": 0.9284519606021185,
      "grad_norm": 0.5871690511703491,
      "learning_rate": 0.0001116426423162494,
      "loss": 1.1298,
      "step": 4996
    },
    {
      "epoch": 0.9286377996654896,
      "grad_norm": 0.6282622218132019,
      "learning_rate": 0.00011161359482857204,
      "loss": 1.0211,
      "step": 4997
    },
    {
      "epoch": 0.9288236387288608,
      "grad_norm": 0.6937336325645447,
      "learning_rate": 0.0001115845463475588,
      "loss": 1.1674,
      "step": 4998
    },
    {
      "epoch": 0.929009477792232,
      "grad_norm": 0.6456194519996643,
      "learning_rate": 0.00011155549687569429,
      "loss": 1.1737,
      "step": 4999
    },
    {
      "epoch": 0.9291953168556031,
      "grad_norm": 0.6184737682342529,
      "learning_rate": 0.00011152644641546315,
      "loss": 1.0585,
      "step": 5000
    },
    {
      "epoch": 0.9293811559189742,
      "grad_norm": 0.5586627125740051,
      "learning_rate": 0.00011149739496935012,
      "loss": 1.1264,
      "step": 5001
    },
    {
      "epoch": 0.9295669949823453,
      "grad_norm": 0.6428977251052856,
      "learning_rate": 0.00011146834253984006,
      "loss": 0.9552,
      "step": 5002
    },
    {
      "epoch": 0.9297528340457164,
      "grad_norm": 1.159934401512146,
      "learning_rate": 0.00011143928912941785,
      "loss": 1.123,
      "step": 5003
    },
    {
      "epoch": 0.9299386731090875,
      "grad_norm": 0.552125871181488,
      "learning_rate": 0.00011141023474056856,
      "loss": 1.0911,
      "step": 5004
    },
    {
      "epoch": 0.9301245121724586,
      "grad_norm": 0.685393214225769,
      "learning_rate": 0.00011138117937577721,
      "loss": 1.2168,
      "step": 5005
    },
    {
      "epoch": 0.9303103512358297,
      "grad_norm": 0.550447404384613,
      "learning_rate": 0.00011135212303752902,
      "loss": 1.2536,
      "step": 5006
    },
    {
      "epoch": 0.9304961902992009,
      "grad_norm": 0.5898290872573853,
      "learning_rate": 0.0001113230657283092,
      "loss": 0.9594,
      "step": 5007
    },
    {
      "epoch": 0.930682029362572,
      "grad_norm": 0.6642700433731079,
      "learning_rate": 0.00011129400745060308,
      "loss": 1.4289,
      "step": 5008
    },
    {
      "epoch": 0.9308678684259432,
      "grad_norm": 2.138993501663208,
      "learning_rate": 0.00011126494820689608,
      "loss": 1.4129,
      "step": 5009
    },
    {
      "epoch": 0.9310537074893143,
      "grad_norm": 0.6226675510406494,
      "learning_rate": 0.0001112358879996737,
      "loss": 1.0456,
      "step": 5010
    },
    {
      "epoch": 0.9312395465526854,
      "grad_norm": 0.705071210861206,
      "learning_rate": 0.00011120682683142159,
      "loss": 1.0938,
      "step": 5011
    },
    {
      "epoch": 0.9314253856160565,
      "grad_norm": 0.6553571820259094,
      "learning_rate": 0.00011117776470462533,
      "loss": 1.0687,
      "step": 5012
    },
    {
      "epoch": 0.9316112246794276,
      "grad_norm": 0.6681797504425049,
      "learning_rate": 0.0001111487016217707,
      "loss": 1.1589,
      "step": 5013
    },
    {
      "epoch": 0.9317970637427987,
      "grad_norm": 0.6915502548217773,
      "learning_rate": 0.00011111963758534353,
      "loss": 1.0384,
      "step": 5014
    },
    {
      "epoch": 0.9319829028061699,
      "grad_norm": 0.6186158657073975,
      "learning_rate": 0.00011109057259782976,
      "loss": 0.964,
      "step": 5015
    },
    {
      "epoch": 0.932168741869541,
      "grad_norm": 0.5619001388549805,
      "learning_rate": 0.00011106150666171528,
      "loss": 1.1913,
      "step": 5016
    },
    {
      "epoch": 0.9323545809329121,
      "grad_norm": 0.5783942937850952,
      "learning_rate": 0.00011103243977948623,
      "loss": 1.0368,
      "step": 5017
    },
    {
      "epoch": 0.9325404199962832,
      "grad_norm": 0.6250287294387817,
      "learning_rate": 0.0001110033719536288,
      "loss": 0.9352,
      "step": 5018
    },
    {
      "epoch": 0.9327262590596543,
      "grad_norm": 0.6437892317771912,
      "learning_rate": 0.00011097430318662918,
      "loss": 1.2505,
      "step": 5019
    },
    {
      "epoch": 0.9329120981230254,
      "grad_norm": 0.6726879477500916,
      "learning_rate": 0.00011094523348097369,
      "loss": 1.1019,
      "step": 5020
    },
    {
      "epoch": 0.9330979371863966,
      "grad_norm": 0.5041570663452148,
      "learning_rate": 0.00011091616283914872,
      "loss": 1.0098,
      "step": 5021
    },
    {
      "epoch": 0.9332837762497677,
      "grad_norm": 0.5859299302101135,
      "learning_rate": 0.00011088709126364074,
      "loss": 1.0304,
      "step": 5022
    },
    {
      "epoch": 0.9334696153131388,
      "grad_norm": 0.5126475691795349,
      "learning_rate": 0.00011085801875693631,
      "loss": 0.9202,
      "step": 5023
    },
    {
      "epoch": 0.93365545437651,
      "grad_norm": 0.5984481573104858,
      "learning_rate": 0.00011082894532152206,
      "loss": 1.1288,
      "step": 5024
    },
    {
      "epoch": 0.9338412934398811,
      "grad_norm": 0.6451727151870728,
      "learning_rate": 0.00011079987095988476,
      "loss": 1.0977,
      "step": 5025
    },
    {
      "epoch": 0.9340271325032522,
      "grad_norm": 0.6106451153755188,
      "learning_rate": 0.00011077079567451111,
      "loss": 1.0946,
      "step": 5026
    },
    {
      "epoch": 0.9342129715666233,
      "grad_norm": 1.9733521938323975,
      "learning_rate": 0.00011074171946788805,
      "loss": 1.9512,
      "step": 5027
    },
    {
      "epoch": 0.9343988106299944,
      "grad_norm": 0.6016564965248108,
      "learning_rate": 0.00011071264234250252,
      "loss": 1.1447,
      "step": 5028
    },
    {
      "epoch": 0.9345846496933655,
      "grad_norm": 0.5035417675971985,
      "learning_rate": 0.00011068356430084153,
      "loss": 0.8372,
      "step": 5029
    },
    {
      "epoch": 0.9347704887567366,
      "grad_norm": 0.6445589661598206,
      "learning_rate": 0.00011065448534539216,
      "loss": 1.2813,
      "step": 5030
    },
    {
      "epoch": 0.9349563278201077,
      "grad_norm": 0.5566248297691345,
      "learning_rate": 0.00011062540547864166,
      "loss": 1.1497,
      "step": 5031
    },
    {
      "epoch": 0.935142166883479,
      "grad_norm": 0.6071885824203491,
      "learning_rate": 0.00011059632470307723,
      "loss": 1.0647,
      "step": 5032
    },
    {
      "epoch": 0.9353280059468501,
      "grad_norm": 0.703399658203125,
      "learning_rate": 0.00011056724302118628,
      "loss": 1.1353,
      "step": 5033
    },
    {
      "epoch": 0.9355138450102212,
      "grad_norm": 0.6310158371925354,
      "learning_rate": 0.00011053816043545621,
      "loss": 1.1485,
      "step": 5034
    },
    {
      "epoch": 0.9356996840735923,
      "grad_norm": 0.4967382848262787,
      "learning_rate": 0.00011050907694837449,
      "loss": 0.7314,
      "step": 5035
    },
    {
      "epoch": 0.9358855231369634,
      "grad_norm": 0.5645965933799744,
      "learning_rate": 0.00011047999256242872,
      "loss": 1.1769,
      "step": 5036
    },
    {
      "epoch": 0.9360713622003345,
      "grad_norm": 0.6344399452209473,
      "learning_rate": 0.0001104509072801065,
      "loss": 0.9947,
      "step": 5037
    },
    {
      "epoch": 0.9362572012637056,
      "grad_norm": 0.6954240202903748,
      "learning_rate": 0.0001104218211038956,
      "loss": 1.043,
      "step": 5038
    },
    {
      "epoch": 0.9364430403270767,
      "grad_norm": 0.5661229491233826,
      "learning_rate": 0.0001103927340362838,
      "loss": 1.1193,
      "step": 5039
    },
    {
      "epoch": 0.9366288793904479,
      "grad_norm": 2.2776856422424316,
      "learning_rate": 0.00011036364607975904,
      "loss": 1.9404,
      "step": 5040
    },
    {
      "epoch": 0.936814718453819,
      "grad_norm": 1.624625563621521,
      "learning_rate": 0.00011033455723680922,
      "loss": 1.8305,
      "step": 5041
    },
    {
      "epoch": 0.9370005575171901,
      "grad_norm": 0.6585975885391235,
      "learning_rate": 0.00011030546750992239,
      "loss": 0.9251,
      "step": 5042
    },
    {
      "epoch": 0.9371863965805612,
      "grad_norm": 0.5694805383682251,
      "learning_rate": 0.00011027637690158663,
      "loss": 1.1026,
      "step": 5043
    },
    {
      "epoch": 0.9373722356439323,
      "grad_norm": 0.5340913534164429,
      "learning_rate": 0.00011024728541429015,
      "loss": 0.9748,
      "step": 5044
    },
    {
      "epoch": 0.9375580747073035,
      "grad_norm": 0.6588858366012573,
      "learning_rate": 0.00011021819305052119,
      "loss": 1.1756,
      "step": 5045
    },
    {
      "epoch": 0.9377439137706746,
      "grad_norm": 0.5693681836128235,
      "learning_rate": 0.0001101890998127681,
      "loss": 1.047,
      "step": 5046
    },
    {
      "epoch": 0.9379297528340457,
      "grad_norm": 0.7040513753890991,
      "learning_rate": 0.0001101600057035193,
      "loss": 1.1306,
      "step": 5047
    },
    {
      "epoch": 0.9381155918974169,
      "grad_norm": 0.5445013046264648,
      "learning_rate": 0.00011013091072526322,
      "loss": 1.0133,
      "step": 5048
    },
    {
      "epoch": 0.938301430960788,
      "grad_norm": 0.6809645891189575,
      "learning_rate": 0.0001101018148804885,
      "loss": 1.1443,
      "step": 5049
    },
    {
      "epoch": 0.9384872700241591,
      "grad_norm": 0.659029483795166,
      "learning_rate": 0.0001100727181716837,
      "loss": 1.1365,
      "step": 5050
    },
    {
      "epoch": 0.9386731090875302,
      "grad_norm": 0.6074685454368591,
      "learning_rate": 0.00011004362060133751,
      "loss": 0.9799,
      "step": 5051
    },
    {
      "epoch": 0.9388589481509013,
      "grad_norm": 0.5053578615188599,
      "learning_rate": 0.00011001452217193877,
      "loss": 0.9078,
      "step": 5052
    },
    {
      "epoch": 0.9390447872142724,
      "grad_norm": 0.77981036901474,
      "learning_rate": 0.0001099854228859763,
      "loss": 1.3562,
      "step": 5053
    },
    {
      "epoch": 0.9392306262776435,
      "grad_norm": 0.5708858966827393,
      "learning_rate": 0.00010995632274593904,
      "loss": 1.2978,
      "step": 5054
    },
    {
      "epoch": 0.9394164653410146,
      "grad_norm": 0.661683976650238,
      "learning_rate": 0.00010992722175431595,
      "loss": 1.1618,
      "step": 5055
    },
    {
      "epoch": 0.9396023044043857,
      "grad_norm": 0.631465494632721,
      "learning_rate": 0.00010989811991359618,
      "loss": 0.9292,
      "step": 5056
    },
    {
      "epoch": 0.939788143467757,
      "grad_norm": 0.57979816198349,
      "learning_rate": 0.00010986901722626879,
      "loss": 1.1043,
      "step": 5057
    },
    {
      "epoch": 0.9399739825311281,
      "grad_norm": 0.5770615935325623,
      "learning_rate": 0.00010983991369482301,
      "loss": 1.1172,
      "step": 5058
    },
    {
      "epoch": 0.9401598215944992,
      "grad_norm": 0.5192406177520752,
      "learning_rate": 0.00010981080932174816,
      "loss": 1.0143,
      "step": 5059
    },
    {
      "epoch": 0.9403456606578703,
      "grad_norm": 4.092175006866455,
      "learning_rate": 0.00010978170410953359,
      "loss": 1.7,
      "step": 5060
    },
    {
      "epoch": 0.9405314997212414,
      "grad_norm": 0.6853275895118713,
      "learning_rate": 0.00010975259806066871,
      "loss": 1.109,
      "step": 5061
    },
    {
      "epoch": 0.9407173387846125,
      "grad_norm": 0.6233556866645813,
      "learning_rate": 0.00010972349117764304,
      "loss": 1.2134,
      "step": 5062
    },
    {
      "epoch": 0.9409031778479836,
      "grad_norm": 0.6518190503120422,
      "learning_rate": 0.00010969438346294618,
      "loss": 1.3395,
      "step": 5063
    },
    {
      "epoch": 0.9410890169113547,
      "grad_norm": 0.5534566044807434,
      "learning_rate": 0.00010966527491906775,
      "loss": 1.0615,
      "step": 5064
    },
    {
      "epoch": 0.9412748559747259,
      "grad_norm": 0.5693931579589844,
      "learning_rate": 0.00010963616554849743,
      "loss": 0.9612,
      "step": 5065
    },
    {
      "epoch": 0.941460695038097,
      "grad_norm": 0.6822101473808289,
      "learning_rate": 0.00010960705535372509,
      "loss": 1.1194,
      "step": 5066
    },
    {
      "epoch": 0.9416465341014681,
      "grad_norm": 0.626566469669342,
      "learning_rate": 0.00010957794433724051,
      "loss": 1.1189,
      "step": 5067
    },
    {
      "epoch": 0.9418323731648393,
      "grad_norm": 0.5476789474487305,
      "learning_rate": 0.00010954883250153365,
      "loss": 0.967,
      "step": 5068
    },
    {
      "epoch": 0.9420182122282104,
      "grad_norm": 1.7039079666137695,
      "learning_rate": 0.00010951971984909452,
      "loss": 1.4795,
      "step": 5069
    },
    {
      "epoch": 0.9422040512915815,
      "grad_norm": 0.6259586811065674,
      "learning_rate": 0.00010949060638241318,
      "loss": 1.0563,
      "step": 5070
    },
    {
      "epoch": 0.9423898903549526,
      "grad_norm": 0.6064488291740417,
      "learning_rate": 0.00010946149210397976,
      "loss": 0.9932,
      "step": 5071
    },
    {
      "epoch": 0.9425757294183237,
      "grad_norm": 0.5755789279937744,
      "learning_rate": 0.00010943237701628446,
      "loss": 0.9768,
      "step": 5072
    },
    {
      "epoch": 0.9427615684816949,
      "grad_norm": 0.6496974229812622,
      "learning_rate": 0.00010940326112181758,
      "loss": 1.28,
      "step": 5073
    },
    {
      "epoch": 0.942947407545066,
      "grad_norm": 0.5570834875106812,
      "learning_rate": 0.00010937414442306946,
      "loss": 1.1327,
      "step": 5074
    },
    {
      "epoch": 0.9431332466084371,
      "grad_norm": 0.5175633430480957,
      "learning_rate": 0.00010934502692253048,
      "loss": 0.8864,
      "step": 5075
    },
    {
      "epoch": 0.9433190856718082,
      "grad_norm": 0.4660826027393341,
      "learning_rate": 0.00010931590862269117,
      "loss": 0.7515,
      "step": 5076
    },
    {
      "epoch": 0.9435049247351793,
      "grad_norm": 0.6771041750907898,
      "learning_rate": 0.00010928678952604205,
      "loss": 1.2769,
      "step": 5077
    },
    {
      "epoch": 0.9436907637985504,
      "grad_norm": 0.6460321545600891,
      "learning_rate": 0.00010925766963507378,
      "loss": 0.8766,
      "step": 5078
    },
    {
      "epoch": 0.9438766028619215,
      "grad_norm": 0.7063780426979065,
      "learning_rate": 0.00010922854895227699,
      "loss": 1.1556,
      "step": 5079
    },
    {
      "epoch": 0.9440624419252927,
      "grad_norm": 0.5997153520584106,
      "learning_rate": 0.00010919942748014247,
      "loss": 1.1245,
      "step": 5080
    },
    {
      "epoch": 0.9442482809886639,
      "grad_norm": 0.6136290431022644,
      "learning_rate": 0.00010917030522116104,
      "loss": 0.9105,
      "step": 5081
    },
    {
      "epoch": 0.944434120052035,
      "grad_norm": 0.6878195405006409,
      "learning_rate": 0.00010914118217782358,
      "loss": 0.9943,
      "step": 5082
    },
    {
      "epoch": 0.9446199591154061,
      "grad_norm": 0.5871633291244507,
      "learning_rate": 0.00010911205835262104,
      "loss": 1.198,
      "step": 5083
    },
    {
      "epoch": 0.9448057981787772,
      "grad_norm": 0.7107055187225342,
      "learning_rate": 0.00010908293374804446,
      "loss": 1.2383,
      "step": 5084
    },
    {
      "epoch": 0.9449916372421483,
      "grad_norm": 0.7144885659217834,
      "learning_rate": 0.00010905380836658495,
      "loss": 1.1298,
      "step": 5085
    },
    {
      "epoch": 0.9451774763055194,
      "grad_norm": 0.6252016425132751,
      "learning_rate": 0.0001090246822107336,
      "loss": 1.0093,
      "step": 5086
    },
    {
      "epoch": 0.9453633153688905,
      "grad_norm": 0.5267205834388733,
      "learning_rate": 0.00010899555528298171,
      "loss": 1.0455,
      "step": 5087
    },
    {
      "epoch": 0.9455491544322616,
      "grad_norm": 0.7700502872467041,
      "learning_rate": 0.00010896642758582052,
      "loss": 1.0634,
      "step": 5088
    },
    {
      "epoch": 0.9457349934956328,
      "grad_norm": 0.6101939678192139,
      "learning_rate": 0.00010893729912174142,
      "loss": 0.8266,
      "step": 5089
    },
    {
      "epoch": 0.9459208325590039,
      "grad_norm": 0.6290589570999146,
      "learning_rate": 0.00010890816989323577,
      "loss": 1.0146,
      "step": 5090
    },
    {
      "epoch": 0.946106671622375,
      "grad_norm": 0.5946857929229736,
      "learning_rate": 0.00010887903990279514,
      "loss": 0.8321,
      "step": 5091
    },
    {
      "epoch": 0.9462925106857462,
      "grad_norm": 0.583081066608429,
      "learning_rate": 0.000108849909152911,
      "loss": 0.8929,
      "step": 5092
    },
    {
      "epoch": 0.9464783497491173,
      "grad_norm": 0.5924783945083618,
      "learning_rate": 0.00010882077764607501,
      "loss": 0.8518,
      "step": 5093
    },
    {
      "epoch": 0.9466641888124884,
      "grad_norm": 0.5845556259155273,
      "learning_rate": 0.00010879164538477886,
      "loss": 1.0834,
      "step": 5094
    },
    {
      "epoch": 0.9468500278758595,
      "grad_norm": 0.6627733707427979,
      "learning_rate": 0.00010876251237151429,
      "loss": 1.0807,
      "step": 5095
    },
    {
      "epoch": 0.9470358669392306,
      "grad_norm": 0.5590043663978577,
      "learning_rate": 0.00010873337860877308,
      "loss": 1.1807,
      "step": 5096
    },
    {
      "epoch": 0.9472217060026017,
      "grad_norm": 0.4764077961444855,
      "learning_rate": 0.00010870424409904711,
      "loss": 1.0865,
      "step": 5097
    },
    {
      "epoch": 0.9474075450659729,
      "grad_norm": 0.5997889637947083,
      "learning_rate": 0.00010867510884482837,
      "loss": 1.3383,
      "step": 5098
    },
    {
      "epoch": 0.947593384129344,
      "grad_norm": 0.5681039094924927,
      "learning_rate": 0.00010864597284860876,
      "loss": 1.0977,
      "step": 5099
    },
    {
      "epoch": 0.9477792231927151,
      "grad_norm": 0.6533553004264832,
      "learning_rate": 0.00010861683611288045,
      "loss": 1.1249,
      "step": 5100
    },
    {
      "epoch": 0.9479650622560862,
      "grad_norm": 0.6179550886154175,
      "learning_rate": 0.00010858769864013551,
      "loss": 1.0567,
      "step": 5101
    },
    {
      "epoch": 0.9481509013194573,
      "grad_norm": 0.5242693424224854,
      "learning_rate": 0.00010855856043286616,
      "loss": 0.7278,
      "step": 5102
    },
    {
      "epoch": 0.9483367403828284,
      "grad_norm": 0.579545795917511,
      "learning_rate": 0.00010852942149356464,
      "loss": 1.1227,
      "step": 5103
    },
    {
      "epoch": 0.9485225794461996,
      "grad_norm": 0.6305685639381409,
      "learning_rate": 0.00010850028182472328,
      "loss": 1.2344,
      "step": 5104
    },
    {
      "epoch": 0.9487084185095707,
      "grad_norm": 0.6166912317276001,
      "learning_rate": 0.0001084711414288344,
      "loss": 1.1048,
      "step": 5105
    },
    {
      "epoch": 0.9488942575729419,
      "grad_norm": 0.549769401550293,
      "learning_rate": 0.00010844200030839047,
      "loss": 0.9258,
      "step": 5106
    },
    {
      "epoch": 0.949080096636313,
      "grad_norm": 0.6159671545028687,
      "learning_rate": 0.00010841285846588405,
      "loss": 1.0702,
      "step": 5107
    },
    {
      "epoch": 0.9492659356996841,
      "grad_norm": 0.6680212616920471,
      "learning_rate": 0.00010838371590380765,
      "loss": 1.0217,
      "step": 5108
    },
    {
      "epoch": 0.9494517747630552,
      "grad_norm": 3.235112428665161,
      "learning_rate": 0.00010835457262465391,
      "loss": 1.5417,
      "step": 5109
    },
    {
      "epoch": 0.9496376138264263,
      "grad_norm": 0.6026039123535156,
      "learning_rate": 0.00010832542863091553,
      "loss": 1.0344,
      "step": 5110
    },
    {
      "epoch": 0.9498234528897974,
      "grad_norm": 0.660992443561554,
      "learning_rate": 0.00010829628392508525,
      "loss": 1.3435,
      "step": 5111
    },
    {
      "epoch": 0.9500092919531685,
      "grad_norm": 0.5971868634223938,
      "learning_rate": 0.00010826713850965584,
      "loss": 1.0527,
      "step": 5112
    },
    {
      "epoch": 0.9501951310165396,
      "grad_norm": 0.7074155807495117,
      "learning_rate": 0.00010823799238712022,
      "loss": 0.9946,
      "step": 5113
    },
    {
      "epoch": 0.9503809700799108,
      "grad_norm": 0.5157049298286438,
      "learning_rate": 0.0001082088455599713,
      "loss": 1.0142,
      "step": 5114
    },
    {
      "epoch": 0.950566809143282,
      "grad_norm": 0.6906478404998779,
      "learning_rate": 0.00010817969803070211,
      "loss": 1.3297,
      "step": 5115
    },
    {
      "epoch": 0.9507526482066531,
      "grad_norm": 0.5686500072479248,
      "learning_rate": 0.00010815054980180566,
      "loss": 0.9322,
      "step": 5116
    },
    {
      "epoch": 0.9509384872700242,
      "grad_norm": 0.6842989921569824,
      "learning_rate": 0.00010812140087577504,
      "loss": 1.0405,
      "step": 5117
    },
    {
      "epoch": 0.9511243263333953,
      "grad_norm": 0.6816885471343994,
      "learning_rate": 0.0001080922512551035,
      "loss": 1.3582,
      "step": 5118
    },
    {
      "epoch": 0.9513101653967664,
      "grad_norm": 0.5930526256561279,
      "learning_rate": 0.00010806310094228415,
      "loss": 1.0377,
      "step": 5119
    },
    {
      "epoch": 0.9514960044601375,
      "grad_norm": 0.6771032214164734,
      "learning_rate": 0.00010803394993981038,
      "loss": 1.2365,
      "step": 5120
    },
    {
      "epoch": 0.9516818435235086,
      "grad_norm": 0.5323301553726196,
      "learning_rate": 0.00010800479825017553,
      "loss": 0.8862,
      "step": 5121
    },
    {
      "epoch": 0.9518676825868798,
      "grad_norm": 0.5637294054031372,
      "learning_rate": 0.00010797564587587299,
      "loss": 1.1621,
      "step": 5122
    },
    {
      "epoch": 0.9520535216502509,
      "grad_norm": 0.6822587251663208,
      "learning_rate": 0.0001079464928193962,
      "loss": 0.9469,
      "step": 5123
    },
    {
      "epoch": 0.952239360713622,
      "grad_norm": 0.7079898118972778,
      "learning_rate": 0.0001079173390832387,
      "loss": 1.1282,
      "step": 5124
    },
    {
      "epoch": 0.9524251997769931,
      "grad_norm": 0.6180583238601685,
      "learning_rate": 0.0001078881846698941,
      "loss": 1.2146,
      "step": 5125
    },
    {
      "epoch": 0.9526110388403642,
      "grad_norm": 0.5927160978317261,
      "learning_rate": 0.00010785902958185599,
      "loss": 1.0453,
      "step": 5126
    },
    {
      "epoch": 0.9527968779037354,
      "grad_norm": 0.690459668636322,
      "learning_rate": 0.0001078298738216181,
      "loss": 1.105,
      "step": 5127
    },
    {
      "epoch": 0.9529827169671065,
      "grad_norm": 0.5838513374328613,
      "learning_rate": 0.00010780071739167416,
      "loss": 1.0425,
      "step": 5128
    },
    {
      "epoch": 0.9531685560304776,
      "grad_norm": 0.5430728793144226,
      "learning_rate": 0.00010777156029451805,
      "loss": 0.9739,
      "step": 5129
    },
    {
      "epoch": 0.9533543950938487,
      "grad_norm": 0.5774003863334656,
      "learning_rate": 0.00010774240253264356,
      "loss": 0.9745,
      "step": 5130
    },
    {
      "epoch": 0.9535402341572199,
      "grad_norm": 0.6548106670379639,
      "learning_rate": 0.00010771324410854466,
      "loss": 0.9721,
      "step": 5131
    },
    {
      "epoch": 0.953726073220591,
      "grad_norm": 0.5925551056861877,
      "learning_rate": 0.00010768408502471534,
      "loss": 1.1294,
      "step": 5132
    },
    {
      "epoch": 0.9539119122839621,
      "grad_norm": 0.66984623670578,
      "learning_rate": 0.0001076549252836496,
      "loss": 1.2196,
      "step": 5133
    },
    {
      "epoch": 0.9540977513473332,
      "grad_norm": 0.6303706765174866,
      "learning_rate": 0.00010762576488784153,
      "loss": 1.09,
      "step": 5134
    },
    {
      "epoch": 0.9542835904107043,
      "grad_norm": 0.5988698601722717,
      "learning_rate": 0.00010759660383978532,
      "loss": 1.0056,
      "step": 5135
    },
    {
      "epoch": 0.9544694294740754,
      "grad_norm": 0.5174127221107483,
      "learning_rate": 0.00010756744214197518,
      "loss": 0.8531,
      "step": 5136
    },
    {
      "epoch": 0.9546552685374465,
      "grad_norm": 0.6051040291786194,
      "learning_rate": 0.00010753827979690536,
      "loss": 0.9742,
      "step": 5137
    },
    {
      "epoch": 0.9548411076008176,
      "grad_norm": 0.6123926043510437,
      "learning_rate": 0.00010750911680707016,
      "loss": 1.13,
      "step": 5138
    },
    {
      "epoch": 0.9550269466641889,
      "grad_norm": 0.9708890914916992,
      "learning_rate": 0.000107479953174964,
      "loss": 1.2774,
      "step": 5139
    },
    {
      "epoch": 0.95521278572756,
      "grad_norm": 0.47982341051101685,
      "learning_rate": 0.00010745078890308127,
      "loss": 0.9506,
      "step": 5140
    },
    {
      "epoch": 0.9553986247909311,
      "grad_norm": 0.5771783590316772,
      "learning_rate": 0.00010742162399391642,
      "loss": 1.2626,
      "step": 5141
    },
    {
      "epoch": 0.9555844638543022,
      "grad_norm": 0.5633402466773987,
      "learning_rate": 0.00010739245844996406,
      "loss": 1.2364,
      "step": 5142
    },
    {
      "epoch": 0.9557703029176733,
      "grad_norm": 0.6553140878677368,
      "learning_rate": 0.00010736329227371876,
      "loss": 1.1984,
      "step": 5143
    },
    {
      "epoch": 0.9559561419810444,
      "grad_norm": 0.5633707642555237,
      "learning_rate": 0.00010733412546767513,
      "loss": 0.966,
      "step": 5144
    },
    {
      "epoch": 0.9561419810444155,
      "grad_norm": 0.5353180766105652,
      "learning_rate": 0.00010730495803432794,
      "loss": 0.9972,
      "step": 5145
    },
    {
      "epoch": 0.9563278201077866,
      "grad_norm": 0.6058616638183594,
      "learning_rate": 0.00010727578997617189,
      "loss": 1.0613,
      "step": 5146
    },
    {
      "epoch": 0.9565136591711578,
      "grad_norm": 0.7414457201957703,
      "learning_rate": 0.00010724662129570177,
      "loss": 1.1391,
      "step": 5147
    },
    {
      "epoch": 0.9566994982345289,
      "grad_norm": 0.6592400670051575,
      "learning_rate": 0.0001072174519954125,
      "loss": 1.1756,
      "step": 5148
    },
    {
      "epoch": 0.9568853372979,
      "grad_norm": 0.5891792178153992,
      "learning_rate": 0.00010718828207779894,
      "loss": 1.1063,
      "step": 5149
    },
    {
      "epoch": 0.9570711763612711,
      "grad_norm": 0.7033983469009399,
      "learning_rate": 0.00010715911154535608,
      "loss": 1.2711,
      "step": 5150
    },
    {
      "epoch": 0.9572570154246423,
      "grad_norm": 0.5749421715736389,
      "learning_rate": 0.00010712994040057894,
      "loss": 1.3341,
      "step": 5151
    },
    {
      "epoch": 0.9574428544880134,
      "grad_norm": 0.5675082802772522,
      "learning_rate": 0.0001071007686459626,
      "loss": 0.9953,
      "step": 5152
    },
    {
      "epoch": 0.9576286935513845,
      "grad_norm": 0.5799729824066162,
      "learning_rate": 0.00010707159628400216,
      "loss": 1.0317,
      "step": 5153
    },
    {
      "epoch": 0.9578145326147556,
      "grad_norm": 0.5823427438735962,
      "learning_rate": 0.0001070424233171928,
      "loss": 1.1268,
      "step": 5154
    },
    {
      "epoch": 0.9580003716781268,
      "grad_norm": 0.6185470819473267,
      "learning_rate": 0.00010701324974802975,
      "loss": 1.0075,
      "step": 5155
    },
    {
      "epoch": 0.9581862107414979,
      "grad_norm": 1.9274375438690186,
      "learning_rate": 0.00010698407557900832,
      "loss": 1.1037,
      "step": 5156
    },
    {
      "epoch": 0.958372049804869,
      "grad_norm": 0.875440239906311,
      "learning_rate": 0.00010695490081262376,
      "loss": 0.9812,
      "step": 5157
    },
    {
      "epoch": 0.9585578888682401,
      "grad_norm": 0.5173926949501038,
      "learning_rate": 0.00010692572545137154,
      "loss": 0.8942,
      "step": 5158
    },
    {
      "epoch": 0.9587437279316112,
      "grad_norm": 0.6853187084197998,
      "learning_rate": 0.00010689654949774706,
      "loss": 0.9427,
      "step": 5159
    },
    {
      "epoch": 0.9589295669949823,
      "grad_norm": 0.5588276386260986,
      "learning_rate": 0.00010686737295424578,
      "loss": 1.1392,
      "step": 5160
    },
    {
      "epoch": 0.9591154060583534,
      "grad_norm": 0.7493889331817627,
      "learning_rate": 0.00010683819582336327,
      "loss": 1.1085,
      "step": 5161
    },
    {
      "epoch": 0.9593012451217245,
      "grad_norm": 0.6834607720375061,
      "learning_rate": 0.00010680901810759505,
      "loss": 1.3353,
      "step": 5162
    },
    {
      "epoch": 0.9594870841850958,
      "grad_norm": 0.7147015929222107,
      "learning_rate": 0.00010677983980943684,
      "loss": 0.8831,
      "step": 5163
    },
    {
      "epoch": 0.9596729232484669,
      "grad_norm": 0.5647204518318176,
      "learning_rate": 0.00010675066093138424,
      "loss": 0.9929,
      "step": 5164
    },
    {
      "epoch": 0.959858762311838,
      "grad_norm": 0.5069839954376221,
      "learning_rate": 0.00010672148147593304,
      "loss": 1.0218,
      "step": 5165
    },
    {
      "epoch": 0.9600446013752091,
      "grad_norm": 0.5167731046676636,
      "learning_rate": 0.000106692301445579,
      "loss": 0.9308,
      "step": 5166
    },
    {
      "epoch": 0.9602304404385802,
      "grad_norm": 0.5342850089073181,
      "learning_rate": 0.00010666312084281795,
      "loss": 1.1225,
      "step": 5167
    },
    {
      "epoch": 0.9604162795019513,
      "grad_norm": 0.726222038269043,
      "learning_rate": 0.00010663393967014579,
      "loss": 1.2616,
      "step": 5168
    },
    {
      "epoch": 0.9606021185653224,
      "grad_norm": 0.45744410157203674,
      "learning_rate": 0.00010660475793005842,
      "loss": 0.6474,
      "step": 5169
    },
    {
      "epoch": 0.9607879576286935,
      "grad_norm": 0.677802324295044,
      "learning_rate": 0.00010657557562505183,
      "loss": 1.092,
      "step": 5170
    },
    {
      "epoch": 0.9609737966920646,
      "grad_norm": 0.571209728717804,
      "learning_rate": 0.00010654639275762202,
      "loss": 1.031,
      "step": 5171
    },
    {
      "epoch": 0.9611596357554358,
      "grad_norm": 0.6979965567588806,
      "learning_rate": 0.00010651720933026513,
      "loss": 1.1168,
      "step": 5172
    },
    {
      "epoch": 0.9613454748188069,
      "grad_norm": 0.5759708881378174,
      "learning_rate": 0.00010648802534547721,
      "loss": 1.1165,
      "step": 5173
    },
    {
      "epoch": 0.961531313882178,
      "grad_norm": 1.40087890625,
      "learning_rate": 0.00010645884080575448,
      "loss": 1.6289,
      "step": 5174
    },
    {
      "epoch": 0.9617171529455492,
      "grad_norm": 0.559787392616272,
      "learning_rate": 0.00010642965571359313,
      "loss": 1.1658,
      "step": 5175
    },
    {
      "epoch": 0.9619029920089203,
      "grad_norm": 0.7585298418998718,
      "learning_rate": 0.00010640047007148944,
      "loss": 0.9917,
      "step": 5176
    },
    {
      "epoch": 0.9620888310722914,
      "grad_norm": 0.6009484529495239,
      "learning_rate": 0.00010637128388193969,
      "loss": 0.9554,
      "step": 5177
    },
    {
      "epoch": 0.9622746701356625,
      "grad_norm": 0.5697795748710632,
      "learning_rate": 0.00010634209714744027,
      "loss": 1.0104,
      "step": 5178
    },
    {
      "epoch": 0.9624605091990336,
      "grad_norm": 0.5680558681488037,
      "learning_rate": 0.00010631290987048757,
      "loss": 0.991,
      "step": 5179
    },
    {
      "epoch": 0.9626463482624048,
      "grad_norm": 0.645099937915802,
      "learning_rate": 0.00010628372205357806,
      "loss": 1.2512,
      "step": 5180
    },
    {
      "epoch": 0.9628321873257759,
      "grad_norm": 0.6746792793273926,
      "learning_rate": 0.00010625453369920822,
      "loss": 1.0832,
      "step": 5181
    },
    {
      "epoch": 0.963018026389147,
      "grad_norm": 0.6117158532142639,
      "learning_rate": 0.00010622534480987459,
      "loss": 1.1984,
      "step": 5182
    },
    {
      "epoch": 0.9632038654525181,
      "grad_norm": 0.6414322853088379,
      "learning_rate": 0.00010619615538807374,
      "loss": 1.1448,
      "step": 5183
    },
    {
      "epoch": 0.9633897045158892,
      "grad_norm": 0.6798393726348877,
      "learning_rate": 0.00010616696543630235,
      "loss": 1.073,
      "step": 5184
    },
    {
      "epoch": 0.9635755435792603,
      "grad_norm": 0.5420655608177185,
      "learning_rate": 0.00010613777495705708,
      "loss": 1.0895,
      "step": 5185
    },
    {
      "epoch": 0.9637613826426314,
      "grad_norm": 0.7268361449241638,
      "learning_rate": 0.00010610858395283462,
      "loss": 1.0711,
      "step": 5186
    },
    {
      "epoch": 0.9639472217060026,
      "grad_norm": 0.6947235465049744,
      "learning_rate": 0.00010607939242613183,
      "loss": 1.0781,
      "step": 5187
    },
    {
      "epoch": 0.9641330607693738,
      "grad_norm": 0.5997654795646667,
      "learning_rate": 0.00010605020037944545,
      "loss": 1.2603,
      "step": 5188
    },
    {
      "epoch": 0.9643188998327449,
      "grad_norm": 0.6570544838905334,
      "learning_rate": 0.00010602100781527233,
      "loss": 1.0687,
      "step": 5189
    },
    {
      "epoch": 0.964504738896116,
      "grad_norm": 0.5399557948112488,
      "learning_rate": 0.0001059918147361094,
      "loss": 1.1065,
      "step": 5190
    },
    {
      "epoch": 0.9646905779594871,
      "grad_norm": 0.581305742263794,
      "learning_rate": 0.00010596262114445363,
      "loss": 1.0601,
      "step": 5191
    },
    {
      "epoch": 0.9648764170228582,
      "grad_norm": 0.6789758801460266,
      "learning_rate": 0.00010593342704280198,
      "loss": 0.9875,
      "step": 5192
    },
    {
      "epoch": 0.9650622560862293,
      "grad_norm": 0.6596348881721497,
      "learning_rate": 0.00010590423243365146,
      "loss": 1.1663,
      "step": 5193
    },
    {
      "epoch": 0.9652480951496004,
      "grad_norm": 0.5195457339286804,
      "learning_rate": 0.00010587503731949921,
      "loss": 0.6786,
      "step": 5194
    },
    {
      "epoch": 0.9654339342129715,
      "grad_norm": 0.4248422682285309,
      "learning_rate": 0.00010584584170284233,
      "loss": 0.7107,
      "step": 5195
    },
    {
      "epoch": 0.9656197732763427,
      "grad_norm": 0.5786765813827515,
      "learning_rate": 0.00010581664558617795,
      "loss": 1.0612,
      "step": 5196
    },
    {
      "epoch": 0.9658056123397138,
      "grad_norm": 0.5303447842597961,
      "learning_rate": 0.0001057874489720033,
      "loss": 0.8196,
      "step": 5197
    },
    {
      "epoch": 0.965991451403085,
      "grad_norm": 0.5511763095855713,
      "learning_rate": 0.00010575825186281565,
      "loss": 0.8464,
      "step": 5198
    },
    {
      "epoch": 0.9661772904664561,
      "grad_norm": 0.698489248752594,
      "learning_rate": 0.00010572905426111226,
      "loss": 1.0296,
      "step": 5199
    },
    {
      "epoch": 0.9663631295298272,
      "grad_norm": 0.6211963891983032,
      "learning_rate": 0.00010569985616939049,
      "loss": 1.0672,
      "step": 5200
    },
    {
      "epoch": 0.9665489685931983,
      "grad_norm": 0.5647724270820618,
      "learning_rate": 0.00010567065759014767,
      "loss": 1.2416,
      "step": 5201
    },
    {
      "epoch": 0.9667348076565694,
      "grad_norm": 0.5544220805168152,
      "learning_rate": 0.0001056414585258813,
      "loss": 1.1426,
      "step": 5202
    },
    {
      "epoch": 0.9669206467199405,
      "grad_norm": 0.6381922960281372,
      "learning_rate": 0.00010561225897908875,
      "loss": 1.2999,
      "step": 5203
    },
    {
      "epoch": 0.9671064857833116,
      "grad_norm": 0.5554980635643005,
      "learning_rate": 0.00010558305895226757,
      "loss": 1.0449,
      "step": 5204
    },
    {
      "epoch": 0.9672923248466828,
      "grad_norm": 0.5876613855361938,
      "learning_rate": 0.0001055538584479153,
      "loss": 1.0988,
      "step": 5205
    },
    {
      "epoch": 0.9674781639100539,
      "grad_norm": 0.6700899004936218,
      "learning_rate": 0.0001055246574685295,
      "loss": 1.1554,
      "step": 5206
    },
    {
      "epoch": 0.967664002973425,
      "grad_norm": 0.605531632900238,
      "learning_rate": 0.00010549545601660784,
      "loss": 1.1517,
      "step": 5207
    },
    {
      "epoch": 0.9678498420367961,
      "grad_norm": 0.5939875245094299,
      "learning_rate": 0.00010546625409464789,
      "loss": 0.7665,
      "step": 5208
    },
    {
      "epoch": 0.9680356811001672,
      "grad_norm": 0.7355973124504089,
      "learning_rate": 0.00010543705170514746,
      "loss": 0.9617,
      "step": 5209
    },
    {
      "epoch": 0.9682215201635384,
      "grad_norm": 0.6694480776786804,
      "learning_rate": 0.00010540784885060424,
      "loss": 1.0366,
      "step": 5210
    },
    {
      "epoch": 0.9684073592269095,
      "grad_norm": 0.580572247505188,
      "learning_rate": 0.000105378645533516,
      "loss": 0.9349,
      "step": 5211
    },
    {
      "epoch": 0.9685931982902806,
      "grad_norm": 0.5605287551879883,
      "learning_rate": 0.00010534944175638059,
      "loss": 1.1615,
      "step": 5212
    },
    {
      "epoch": 0.9687790373536518,
      "grad_norm": 0.6726819276809692,
      "learning_rate": 0.00010532023752169589,
      "loss": 1.249,
      "step": 5213
    },
    {
      "epoch": 0.9689648764170229,
      "grad_norm": 0.6232327222824097,
      "learning_rate": 0.00010529103283195976,
      "loss": 1.0537,
      "step": 5214
    },
    {
      "epoch": 0.969150715480394,
      "grad_norm": 0.6396980285644531,
      "learning_rate": 0.00010526182768967015,
      "loss": 0.9684,
      "step": 5215
    },
    {
      "epoch": 0.9693365545437651,
      "grad_norm": 0.6493716239929199,
      "learning_rate": 0.00010523262209732505,
      "loss": 1.1652,
      "step": 5216
    },
    {
      "epoch": 0.9695223936071362,
      "grad_norm": 0.7912707924842834,
      "learning_rate": 0.00010520341605742247,
      "loss": 1.598,
      "step": 5217
    },
    {
      "epoch": 0.9697082326705073,
      "grad_norm": 0.6064230799674988,
      "learning_rate": 0.00010517420957246047,
      "loss": 0.9873,
      "step": 5218
    },
    {
      "epoch": 0.9698940717338784,
      "grad_norm": 0.7344752550125122,
      "learning_rate": 0.00010514500264493714,
      "loss": 1.1776,
      "step": 5219
    },
    {
      "epoch": 0.9700799107972495,
      "grad_norm": 0.5158562660217285,
      "learning_rate": 0.00010511579527735063,
      "loss": 0.9558,
      "step": 5220
    },
    {
      "epoch": 0.9702657498606208,
      "grad_norm": 0.5321223735809326,
      "learning_rate": 0.00010508658747219908,
      "loss": 0.9554,
      "step": 5221
    },
    {
      "epoch": 0.9704515889239919,
      "grad_norm": 0.5637287497520447,
      "learning_rate": 0.00010505737923198069,
      "loss": 0.9914,
      "step": 5222
    },
    {
      "epoch": 0.970637427987363,
      "grad_norm": 0.5985487699508667,
      "learning_rate": 0.00010502817055919376,
      "loss": 1.0695,
      "step": 5223
    },
    {
      "epoch": 0.9708232670507341,
      "grad_norm": 0.6146731972694397,
      "learning_rate": 0.00010499896145633652,
      "loss": 1.19,
      "step": 5224
    },
    {
      "epoch": 0.9710091061141052,
      "grad_norm": 0.6370542049407959,
      "learning_rate": 0.00010496975192590728,
      "loss": 1.047,
      "step": 5225
    },
    {
      "epoch": 0.9711949451774763,
      "grad_norm": 0.574313223361969,
      "learning_rate": 0.00010494054197040444,
      "loss": 1.2502,
      "step": 5226
    },
    {
      "epoch": 0.9713807842408474,
      "grad_norm": 0.5866028666496277,
      "learning_rate": 0.00010491133159232638,
      "loss": 0.9155,
      "step": 5227
    },
    {
      "epoch": 0.9715666233042185,
      "grad_norm": 0.5359747409820557,
      "learning_rate": 0.0001048821207941715,
      "loss": 1.0809,
      "step": 5228
    },
    {
      "epoch": 0.9717524623675897,
      "grad_norm": 0.5584895610809326,
      "learning_rate": 0.00010485290957843829,
      "loss": 1.1471,
      "step": 5229
    },
    {
      "epoch": 0.9719383014309608,
      "grad_norm": 0.5170494318008423,
      "learning_rate": 0.00010482369794762521,
      "loss": 1.0803,
      "step": 5230
    },
    {
      "epoch": 0.9721241404943319,
      "grad_norm": 0.6326431035995483,
      "learning_rate": 0.00010479448590423082,
      "loss": 0.8558,
      "step": 5231
    },
    {
      "epoch": 0.972309979557703,
      "grad_norm": 0.5903434753417969,
      "learning_rate": 0.00010476527345075369,
      "loss": 1.0103,
      "step": 5232
    },
    {
      "epoch": 0.9724958186210741,
      "grad_norm": 0.6057657599449158,
      "learning_rate": 0.00010473606058969242,
      "loss": 1.1428,
      "step": 5233
    },
    {
      "epoch": 0.9726816576844453,
      "grad_norm": 0.5869147777557373,
      "learning_rate": 0.0001047068473235457,
      "loss": 1.2471,
      "step": 5234
    },
    {
      "epoch": 0.9728674967478164,
      "grad_norm": 0.6794689893722534,
      "learning_rate": 0.00010467763365481211,
      "loss": 1.1385,
      "step": 5235
    },
    {
      "epoch": 0.9730533358111875,
      "grad_norm": 0.6516013741493225,
      "learning_rate": 0.00010464841958599043,
      "loss": 1.177,
      "step": 5236
    },
    {
      "epoch": 0.9732391748745586,
      "grad_norm": 0.6525418758392334,
      "learning_rate": 0.00010461920511957934,
      "loss": 1.1904,
      "step": 5237
    },
    {
      "epoch": 0.9734250139379298,
      "grad_norm": 0.5970218777656555,
      "learning_rate": 0.00010458999025807767,
      "loss": 0.95,
      "step": 5238
    },
    {
      "epoch": 0.9736108530013009,
      "grad_norm": 0.5643072128295898,
      "learning_rate": 0.00010456077500398421,
      "loss": 1.1804,
      "step": 5239
    },
    {
      "epoch": 0.973796692064672,
      "grad_norm": 0.7211046814918518,
      "learning_rate": 0.00010453155935979781,
      "loss": 1.0568,
      "step": 5240
    },
    {
      "epoch": 0.9739825311280431,
      "grad_norm": 0.5928149223327637,
      "learning_rate": 0.00010450234332801737,
      "loss": 0.9441,
      "step": 5241
    },
    {
      "epoch": 0.9741683701914142,
      "grad_norm": 0.49844658374786377,
      "learning_rate": 0.00010447312691114175,
      "loss": 0.7909,
      "step": 5242
    },
    {
      "epoch": 0.9743542092547853,
      "grad_norm": 0.5963571667671204,
      "learning_rate": 0.00010444391011166992,
      "loss": 1.0928,
      "step": 5243
    },
    {
      "epoch": 0.9745400483181564,
      "grad_norm": 0.6606438755989075,
      "learning_rate": 0.00010441469293210083,
      "loss": 0.775,
      "step": 5244
    },
    {
      "epoch": 0.9747258873815275,
      "grad_norm": 0.40258845686912537,
      "learning_rate": 0.00010438547537493355,
      "loss": 0.555,
      "step": 5245
    },
    {
      "epoch": 0.9749117264448988,
      "grad_norm": 0.7338322401046753,
      "learning_rate": 0.00010435625744266704,
      "loss": 1.0914,
      "step": 5246
    },
    {
      "epoch": 0.9750975655082699,
      "grad_norm": 0.5825830698013306,
      "learning_rate": 0.00010432703913780044,
      "loss": 1.0547,
      "step": 5247
    },
    {
      "epoch": 0.975283404571641,
      "grad_norm": 0.7490541338920593,
      "learning_rate": 0.00010429782046283285,
      "loss": 1.3439,
      "step": 5248
    },
    {
      "epoch": 0.9754692436350121,
      "grad_norm": 0.5554697513580322,
      "learning_rate": 0.00010426860142026338,
      "loss": 1.0599,
      "step": 5249
    },
    {
      "epoch": 0.9756550826983832,
      "grad_norm": 0.6733309030532837,
      "learning_rate": 0.0001042393820125912,
      "loss": 1.0955,
      "step": 5250
    },
    {
      "epoch": 0.9758409217617543,
      "grad_norm": 0.6073945164680481,
      "learning_rate": 0.0001042101622423155,
      "loss": 1.1162,
      "step": 5251
    },
    {
      "epoch": 0.9760267608251254,
      "grad_norm": 0.5888856053352356,
      "learning_rate": 0.00010418094211193552,
      "loss": 1.1074,
      "step": 5252
    },
    {
      "epoch": 0.9762125998884965,
      "grad_norm": 0.7521282434463501,
      "learning_rate": 0.00010415172162395052,
      "loss": 1.1723,
      "step": 5253
    },
    {
      "epoch": 0.9763984389518677,
      "grad_norm": 0.7571302652359009,
      "learning_rate": 0.00010412250078085983,
      "loss": 1.207,
      "step": 5254
    },
    {
      "epoch": 0.9765842780152388,
      "grad_norm": 0.5502439141273499,
      "learning_rate": 0.00010409327958516273,
      "loss": 0.8738,
      "step": 5255
    },
    {
      "epoch": 0.97677011707861,
      "grad_norm": 0.6101023554801941,
      "learning_rate": 0.00010406405803935857,
      "loss": 0.955,
      "step": 5256
    },
    {
      "epoch": 0.976955956141981,
      "grad_norm": 0.5006557106971741,
      "learning_rate": 0.00010403483614594675,
      "loss": 1.0782,
      "step": 5257
    },
    {
      "epoch": 0.9771417952053522,
      "grad_norm": 0.5799928307533264,
      "learning_rate": 0.00010400561390742668,
      "loss": 0.893,
      "step": 5258
    },
    {
      "epoch": 0.9773276342687233,
      "grad_norm": 0.5734309554100037,
      "learning_rate": 0.00010397639132629778,
      "loss": 1.1346,
      "step": 5259
    },
    {
      "epoch": 0.9775134733320944,
      "grad_norm": 0.6255081295967102,
      "learning_rate": 0.00010394716840505951,
      "loss": 1.0596,
      "step": 5260
    },
    {
      "epoch": 0.9776993123954655,
      "grad_norm": 0.6332784295082092,
      "learning_rate": 0.00010391794514621143,
      "loss": 1.1199,
      "step": 5261
    },
    {
      "epoch": 0.9778851514588367,
      "grad_norm": 0.524943470954895,
      "learning_rate": 0.00010388872155225304,
      "loss": 0.811,
      "step": 5262
    },
    {
      "epoch": 0.9780709905222078,
      "grad_norm": 0.5607344508171082,
      "learning_rate": 0.00010385949762568387,
      "loss": 1.035,
      "step": 5263
    },
    {
      "epoch": 0.9782568295855789,
      "grad_norm": 0.737356960773468,
      "learning_rate": 0.00010383027336900355,
      "loss": 1.2161,
      "step": 5264
    },
    {
      "epoch": 0.97844266864895,
      "grad_norm": 0.5566937923431396,
      "learning_rate": 0.00010380104878471167,
      "loss": 1.0571,
      "step": 5265
    },
    {
      "epoch": 0.9786285077123211,
      "grad_norm": 0.7953934073448181,
      "learning_rate": 0.00010377182387530784,
      "loss": 1.1642,
      "step": 5266
    },
    {
      "epoch": 0.9788143467756922,
      "grad_norm": 0.5225697755813599,
      "learning_rate": 0.00010374259864329177,
      "loss": 1.0756,
      "step": 5267
    },
    {
      "epoch": 0.9790001858390633,
      "grad_norm": 0.5242747664451599,
      "learning_rate": 0.00010371337309116316,
      "loss": 1.1096,
      "step": 5268
    },
    {
      "epoch": 0.9791860249024344,
      "grad_norm": 0.6449241042137146,
      "learning_rate": 0.00010368414722142172,
      "loss": 0.9631,
      "step": 5269
    },
    {
      "epoch": 0.9793718639658057,
      "grad_norm": 0.7144589424133301,
      "learning_rate": 0.00010365492103656723,
      "loss": 1.0049,
      "step": 5270
    },
    {
      "epoch": 0.9795577030291768,
      "grad_norm": 0.6562042832374573,
      "learning_rate": 0.00010362569453909942,
      "loss": 0.8826,
      "step": 5271
    },
    {
      "epoch": 0.9797435420925479,
      "grad_norm": 0.44900819659233093,
      "learning_rate": 0.00010359646773151814,
      "loss": 0.8152,
      "step": 5272
    },
    {
      "epoch": 0.979929381155919,
      "grad_norm": 0.4567967355251312,
      "learning_rate": 0.00010356724061632316,
      "loss": 0.8494,
      "step": 5273
    },
    {
      "epoch": 0.9801152202192901,
      "grad_norm": 0.7190163135528564,
      "learning_rate": 0.00010353801319601441,
      "loss": 1.0996,
      "step": 5274
    },
    {
      "epoch": 0.9803010592826612,
      "grad_norm": 0.6902177333831787,
      "learning_rate": 0.00010350878547309173,
      "loss": 1.2026,
      "step": 5275
    },
    {
      "epoch": 0.9804868983460323,
      "grad_norm": 0.5994812250137329,
      "learning_rate": 0.00010347955745005506,
      "loss": 0.8673,
      "step": 5276
    },
    {
      "epoch": 0.9806727374094034,
      "grad_norm": 0.5680616497993469,
      "learning_rate": 0.00010345032912940433,
      "loss": 1.1063,
      "step": 5277
    },
    {
      "epoch": 0.9808585764727745,
      "grad_norm": 0.5536043643951416,
      "learning_rate": 0.0001034211005136395,
      "loss": 1.1424,
      "step": 5278
    },
    {
      "epoch": 0.9810444155361457,
      "grad_norm": 0.687205970287323,
      "learning_rate": 0.00010339187160526051,
      "loss": 0.9672,
      "step": 5279
    },
    {
      "epoch": 0.9812302545995168,
      "grad_norm": 0.6523383259773254,
      "learning_rate": 0.00010336264240676745,
      "loss": 0.9638,
      "step": 5280
    },
    {
      "epoch": 0.981416093662888,
      "grad_norm": 1.3020282983779907,
      "learning_rate": 0.00010333341292066028,
      "loss": 1.6678,
      "step": 5281
    },
    {
      "epoch": 0.9816019327262591,
      "grad_norm": 0.6111271977424622,
      "learning_rate": 0.00010330418314943911,
      "loss": 1.0743,
      "step": 5282
    },
    {
      "epoch": 0.9817877717896302,
      "grad_norm": 0.6352726817131042,
      "learning_rate": 0.00010327495309560403,
      "loss": 1.1428,
      "step": 5283
    },
    {
      "epoch": 0.9819736108530013,
      "grad_norm": 2.6539978981018066,
      "learning_rate": 0.00010324572276165512,
      "loss": 2.0701,
      "step": 5284
    },
    {
      "epoch": 0.9821594499163724,
      "grad_norm": 0.5675703287124634,
      "learning_rate": 0.00010321649215009253,
      "loss": 0.9753,
      "step": 5285
    },
    {
      "epoch": 0.9823452889797435,
      "grad_norm": 0.634197473526001,
      "learning_rate": 0.00010318726126341642,
      "loss": 1.1443,
      "step": 5286
    },
    {
      "epoch": 0.9825311280431147,
      "grad_norm": 0.6586654186248779,
      "learning_rate": 0.00010315803010412697,
      "loss": 1.0275,
      "step": 5287
    },
    {
      "epoch": 0.9827169671064858,
      "grad_norm": 0.657075047492981,
      "learning_rate": 0.00010312879867472436,
      "loss": 1.2704,
      "step": 5288
    },
    {
      "epoch": 0.9829028061698569,
      "grad_norm": 0.6138443946838379,
      "learning_rate": 0.00010309956697770881,
      "loss": 0.9674,
      "step": 5289
    },
    {
      "epoch": 0.983088645233228,
      "grad_norm": 0.8176442384719849,
      "learning_rate": 0.00010307033501558065,
      "loss": 1.3409,
      "step": 5290
    },
    {
      "epoch": 0.9832744842965991,
      "grad_norm": 0.5863398909568787,
      "learning_rate": 0.0001030411027908401,
      "loss": 0.8171,
      "step": 5291
    },
    {
      "epoch": 0.9834603233599702,
      "grad_norm": 0.5948399305343628,
      "learning_rate": 0.00010301187030598744,
      "loss": 0.9919,
      "step": 5292
    },
    {
      "epoch": 0.9836461624233414,
      "grad_norm": 0.5358024835586548,
      "learning_rate": 0.00010298263756352302,
      "loss": 1.0402,
      "step": 5293
    },
    {
      "epoch": 0.9838320014867125,
      "grad_norm": 0.7168769836425781,
      "learning_rate": 0.00010295340456594714,
      "loss": 0.9947,
      "step": 5294
    },
    {
      "epoch": 0.9840178405500837,
      "grad_norm": 0.6695325374603271,
      "learning_rate": 0.00010292417131576018,
      "loss": 1.2622,
      "step": 5295
    },
    {
      "epoch": 0.9842036796134548,
      "grad_norm": 0.6197385191917419,
      "learning_rate": 0.00010289493781546256,
      "loss": 1.1881,
      "step": 5296
    },
    {
      "epoch": 0.9843895186768259,
      "grad_norm": 0.6674443483352661,
      "learning_rate": 0.00010286570406755461,
      "loss": 1.0887,
      "step": 5297
    },
    {
      "epoch": 0.984575357740197,
      "grad_norm": 0.6339353322982788,
      "learning_rate": 0.00010283647007453681,
      "loss": 1.1297,
      "step": 5298
    },
    {
      "epoch": 0.9847611968035681,
      "grad_norm": 0.5957179069519043,
      "learning_rate": 0.0001028072358389096,
      "loss": 0.9305,
      "step": 5299
    },
    {
      "epoch": 0.9849470358669392,
      "grad_norm": 0.782689094543457,
      "learning_rate": 0.00010277800136317346,
      "loss": 1.2162,
      "step": 5300
    },
    {
      "epoch": 0.9851328749303103,
      "grad_norm": 0.5918034315109253,
      "learning_rate": 0.00010274876664982883,
      "loss": 0.8853,
      "step": 5301
    },
    {
      "epoch": 0.9853187139936814,
      "grad_norm": 0.5832744836807251,
      "learning_rate": 0.00010271953170137628,
      "loss": 1.0583,
      "step": 5302
    },
    {
      "epoch": 0.9855045530570526,
      "grad_norm": 0.6906222105026245,
      "learning_rate": 0.00010269029652031627,
      "loss": 0.8884,
      "step": 5303
    },
    {
      "epoch": 0.9856903921204238,
      "grad_norm": 0.6617158651351929,
      "learning_rate": 0.0001026610611091494,
      "loss": 1.2843,
      "step": 5304
    },
    {
      "epoch": 0.9858762311837949,
      "grad_norm": 0.7408176064491272,
      "learning_rate": 0.00010263182547037622,
      "loss": 1.0618,
      "step": 5305
    },
    {
      "epoch": 0.986062070247166,
      "grad_norm": 0.6209059357643127,
      "learning_rate": 0.00010260258960649734,
      "loss": 1.1427,
      "step": 5306
    },
    {
      "epoch": 0.9862479093105371,
      "grad_norm": 0.6133381128311157,
      "learning_rate": 0.00010257335352001332,
      "loss": 1.1346,
      "step": 5307
    },
    {
      "epoch": 0.9864337483739082,
      "grad_norm": 0.7502332925796509,
      "learning_rate": 0.00010254411721342481,
      "loss": 1.2057,
      "step": 5308
    },
    {
      "epoch": 0.9866195874372793,
      "grad_norm": 0.6478795409202576,
      "learning_rate": 0.00010251488068923248,
      "loss": 1.1604,
      "step": 5309
    },
    {
      "epoch": 0.9868054265006504,
      "grad_norm": 0.6442230939865112,
      "learning_rate": 0.00010248564394993694,
      "loss": 1.0143,
      "step": 5310
    },
    {
      "epoch": 0.9869912655640215,
      "grad_norm": 0.7198289036750793,
      "learning_rate": 0.00010245640699803892,
      "loss": 1.1354,
      "step": 5311
    },
    {
      "epoch": 0.9871771046273927,
      "grad_norm": 0.6207783818244934,
      "learning_rate": 0.00010242716983603909,
      "loss": 1.184,
      "step": 5312
    },
    {
      "epoch": 0.9873629436907638,
      "grad_norm": 0.6826710104942322,
      "learning_rate": 0.00010239793246643819,
      "loss": 1.2905,
      "step": 5313
    },
    {
      "epoch": 0.9875487827541349,
      "grad_norm": 0.6749565005302429,
      "learning_rate": 0.00010236869489173695,
      "loss": 1.2629,
      "step": 5314
    },
    {
      "epoch": 0.987734621817506,
      "grad_norm": 0.6928743124008179,
      "learning_rate": 0.00010233945711443609,
      "loss": 1.0613,
      "step": 5315
    },
    {
      "epoch": 0.9879204608808771,
      "grad_norm": 0.5754216909408569,
      "learning_rate": 0.00010231021913703643,
      "loss": 1.119,
      "step": 5316
    },
    {
      "epoch": 0.9881062999442483,
      "grad_norm": 0.5343156456947327,
      "learning_rate": 0.00010228098096203871,
      "loss": 0.9631,
      "step": 5317
    },
    {
      "epoch": 0.9882921390076194,
      "grad_norm": 0.526121973991394,
      "learning_rate": 0.00010225174259194377,
      "loss": 0.9348,
      "step": 5318
    },
    {
      "epoch": 0.9884779780709905,
      "grad_norm": 0.7248528003692627,
      "learning_rate": 0.00010222250402925242,
      "loss": 1.204,
      "step": 5319
    },
    {
      "epoch": 0.9886638171343617,
      "grad_norm": 0.7399142980575562,
      "learning_rate": 0.00010219326527646551,
      "loss": 1.2729,
      "step": 5320
    },
    {
      "epoch": 0.9888496561977328,
      "grad_norm": 0.5740293860435486,
      "learning_rate": 0.00010216402633608389,
      "loss": 1.1019,
      "step": 5321
    },
    {
      "epoch": 0.9890354952611039,
      "grad_norm": 0.6210380792617798,
      "learning_rate": 0.00010213478721060838,
      "loss": 1.1636,
      "step": 5322
    },
    {
      "epoch": 0.989221334324475,
      "grad_norm": 0.5840253233909607,
      "learning_rate": 0.00010210554790253992,
      "loss": 1.1467,
      "step": 5323
    },
    {
      "epoch": 0.9894071733878461,
      "grad_norm": 0.5882552266120911,
      "learning_rate": 0.0001020763084143794,
      "loss": 1.0356,
      "step": 5324
    },
    {
      "epoch": 0.9895930124512172,
      "grad_norm": 0.5915971398353577,
      "learning_rate": 0.00010204706874862777,
      "loss": 1.3001,
      "step": 5325
    },
    {
      "epoch": 0.9897788515145883,
      "grad_norm": 0.6104952692985535,
      "learning_rate": 0.00010201782890778586,
      "loss": 1.1029,
      "step": 5326
    },
    {
      "epoch": 0.9899646905779594,
      "grad_norm": 0.6219309568405151,
      "learning_rate": 0.00010198858889435475,
      "loss": 1.1958,
      "step": 5327
    },
    {
      "epoch": 0.9901505296413307,
      "grad_norm": 0.748560905456543,
      "learning_rate": 0.00010195934871083528,
      "loss": 0.8558,
      "step": 5328
    },
    {
      "epoch": 0.9903363687047018,
      "grad_norm": 0.6281746625900269,
      "learning_rate": 0.00010193010835972849,
      "loss": 0.9296,
      "step": 5329
    },
    {
      "epoch": 0.9905222077680729,
      "grad_norm": 0.5284553170204163,
      "learning_rate": 0.00010190086784353538,
      "loss": 1.2851,
      "step": 5330
    },
    {
      "epoch": 0.990708046831444,
      "grad_norm": 0.5357339382171631,
      "learning_rate": 0.00010187162716475694,
      "loss": 0.9738,
      "step": 5331
    },
    {
      "epoch": 0.9908938858948151,
      "grad_norm": 0.7003484964370728,
      "learning_rate": 0.00010184238632589416,
      "loss": 1.1544,
      "step": 5332
    },
    {
      "epoch": 0.9910797249581862,
      "grad_norm": 2.5034542083740234,
      "learning_rate": 0.0001018131453294481,
      "loss": 1.7585,
      "step": 5333
    },
    {
      "epoch": 0.9912655640215573,
      "grad_norm": 0.5779796242713928,
      "learning_rate": 0.0001017839041779198,
      "loss": 1.2848,
      "step": 5334
    },
    {
      "epoch": 0.9914514030849284,
      "grad_norm": 0.5366520881652832,
      "learning_rate": 0.00010175466287381034,
      "loss": 0.9984,
      "step": 5335
    },
    {
      "epoch": 0.9916372421482996,
      "grad_norm": 0.6219021081924438,
      "learning_rate": 0.00010172542141962074,
      "loss": 1.137,
      "step": 5336
    },
    {
      "epoch": 0.9918230812116707,
      "grad_norm": 0.5030076503753662,
      "learning_rate": 0.00010169617981785213,
      "loss": 0.8376,
      "step": 5337
    },
    {
      "epoch": 0.9920089202750418,
      "grad_norm": 0.5434117317199707,
      "learning_rate": 0.00010166693807100562,
      "loss": 1.0558,
      "step": 5338
    },
    {
      "epoch": 0.992194759338413,
      "grad_norm": 0.5399218201637268,
      "learning_rate": 0.00010163769618158227,
      "loss": 1.0514,
      "step": 5339
    },
    {
      "epoch": 0.992380598401784,
      "grad_norm": 0.7620237469673157,
      "learning_rate": 0.00010160845415208322,
      "loss": 0.9099,
      "step": 5340
    },
    {
      "epoch": 0.9925664374651552,
      "grad_norm": 0.6571434736251831,
      "learning_rate": 0.00010157921198500961,
      "loss": 1.2444,
      "step": 5341
    },
    {
      "epoch": 0.9927522765285263,
      "grad_norm": 0.5245940089225769,
      "learning_rate": 0.00010154996968286261,
      "loss": 0.9128,
      "step": 5342
    },
    {
      "epoch": 0.9929381155918974,
      "grad_norm": 0.49775126576423645,
      "learning_rate": 0.00010152072724814332,
      "loss": 0.589,
      "step": 5343
    },
    {
      "epoch": 0.9931239546552686,
      "grad_norm": 0.5503200888633728,
      "learning_rate": 0.00010149148468335296,
      "loss": 1.0956,
      "step": 5344
    },
    {
      "epoch": 0.9933097937186397,
      "grad_norm": 0.599502444267273,
      "learning_rate": 0.00010146224199099273,
      "loss": 1.0305,
      "step": 5345
    },
    {
      "epoch": 0.9934956327820108,
      "grad_norm": 0.5318445563316345,
      "learning_rate": 0.00010143299917356373,
      "loss": 1.1173,
      "step": 5346
    },
    {
      "epoch": 0.9936814718453819,
      "grad_norm": 2.255668878555298,
      "learning_rate": 0.00010140375623356724,
      "loss": 2.0431,
      "step": 5347
    },
    {
      "epoch": 0.993867310908753,
      "grad_norm": 0.7480182647705078,
      "learning_rate": 0.00010137451317350442,
      "loss": 1.2049,
      "step": 5348
    },
    {
      "epoch": 0.9940531499721241,
      "grad_norm": 0.6217489838600159,
      "learning_rate": 0.00010134526999587653,
      "loss": 0.8339,
      "step": 5349
    },
    {
      "epoch": 0.9942389890354952,
      "grad_norm": 0.7519838213920593,
      "learning_rate": 0.00010131602670318477,
      "loss": 0.9099,
      "step": 5350
    },
    {
      "epoch": 0.9944248280988663,
      "grad_norm": 0.5862378478050232,
      "learning_rate": 0.00010128678329793042,
      "loss": 1.1724,
      "step": 5351
    },
    {
      "epoch": 0.9946106671622374,
      "grad_norm": 0.6280035376548767,
      "learning_rate": 0.00010125753978261475,
      "loss": 1.1678,
      "step": 5352
    },
    {
      "epoch": 0.9947965062256087,
      "grad_norm": 0.6822145581245422,
      "learning_rate": 0.00010122829615973895,
      "loss": 1.2307,
      "step": 5353
    },
    {
      "epoch": 0.9949823452889798,
      "grad_norm": 0.5909119844436646,
      "learning_rate": 0.00010119905243180432,
      "loss": 1.1142,
      "step": 5354
    },
    {
      "epoch": 0.9951681843523509,
      "grad_norm": 0.6381353139877319,
      "learning_rate": 0.00010116980860131214,
      "loss": 1.0991,
      "step": 5355
    },
    {
      "epoch": 0.995354023415722,
      "grad_norm": 0.5839842557907104,
      "learning_rate": 0.00010114056467076372,
      "loss": 1.1673,
      "step": 5356
    },
    {
      "epoch": 0.9955398624790931,
      "grad_norm": 0.49389171600341797,
      "learning_rate": 0.0001011113206426603,
      "loss": 0.9878,
      "step": 5357
    },
    {
      "epoch": 0.9957257015424642,
      "grad_norm": 0.6194879412651062,
      "learning_rate": 0.00010108207651950325,
      "loss": 1.15,
      "step": 5358
    },
    {
      "epoch": 0.9959115406058353,
      "grad_norm": 0.6395465731620789,
      "learning_rate": 0.00010105283230379387,
      "loss": 0.9558,
      "step": 5359
    },
    {
      "epoch": 0.9960973796692064,
      "grad_norm": 0.5701009631156921,
      "learning_rate": 0.00010102358799803347,
      "loss": 0.9805,
      "step": 5360
    },
    {
      "epoch": 0.9962832187325776,
      "grad_norm": 0.5622742772102356,
      "learning_rate": 0.00010099434360472337,
      "loss": 1.3543,
      "step": 5361
    },
    {
      "epoch": 0.9964690577959487,
      "grad_norm": 0.5801060199737549,
      "learning_rate": 0.00010096509912636489,
      "loss": 0.8837,
      "step": 5362
    },
    {
      "epoch": 0.9966548968593198,
      "grad_norm": 0.6651684045791626,
      "learning_rate": 0.00010093585456545942,
      "loss": 1.1386,
      "step": 5363
    },
    {
      "epoch": 0.996840735922691,
      "grad_norm": 0.6739777326583862,
      "learning_rate": 0.00010090660992450826,
      "loss": 1.2695,
      "step": 5364
    },
    {
      "epoch": 0.9970265749860621,
      "grad_norm": 0.5893535614013672,
      "learning_rate": 0.00010087736520601281,
      "loss": 0.8915,
      "step": 5365
    },
    {
      "epoch": 0.9972124140494332,
      "grad_norm": 0.6335767507553101,
      "learning_rate": 0.00010084812041247442,
      "loss": 0.9015,
      "step": 5366
    },
    {
      "epoch": 0.9973982531128043,
      "grad_norm": 0.5435164570808411,
      "learning_rate": 0.00010081887554639447,
      "loss": 0.9378,
      "step": 5367
    },
    {
      "epoch": 0.9975840921761754,
      "grad_norm": 0.6234378814697266,
      "learning_rate": 0.00010078963061027433,
      "loss": 0.9928,
      "step": 5368
    },
    {
      "epoch": 0.9977699312395466,
      "grad_norm": 0.727533221244812,
      "learning_rate": 0.00010076038560661534,
      "loss": 1.1077,
      "step": 5369
    },
    {
      "epoch": 0.9979557703029177,
      "grad_norm": 0.5350732207298279,
      "learning_rate": 0.00010073114053791894,
      "loss": 1.0918,
      "step": 5370
    },
    {
      "epoch": 0.9981416093662888,
      "grad_norm": 0.6955908536911011,
      "learning_rate": 0.00010070189540668651,
      "loss": 1.0372,
      "step": 5371
    },
    {
      "epoch": 0.9983274484296599,
      "grad_norm": 0.7087992429733276,
      "learning_rate": 0.00010067265021541944,
      "loss": 1.1127,
      "step": 5372
    },
    {
      "epoch": 0.998513287493031,
      "grad_norm": 0.646903932094574,
      "learning_rate": 0.00010064340496661918,
      "loss": 1.074,
      "step": 5373
    },
    {
      "epoch": 0.9986991265564021,
      "grad_norm": 0.5559451580047607,
      "learning_rate": 0.00010061415966278708,
      "loss": 1.0225,
      "step": 5374
    },
    {
      "epoch": 0.9988849656197732,
      "grad_norm": 0.6143019795417786,
      "learning_rate": 0.00010058491430642458,
      "loss": 1.1226,
      "step": 5375
    },
    {
      "epoch": 0.9990708046831444,
      "grad_norm": 0.6290916204452515,
      "learning_rate": 0.00010055566890003309,
      "loss": 1.2038,
      "step": 5376
    },
    {
      "epoch": 0.9992566437465156,
      "grad_norm": 0.6341618299484253,
      "learning_rate": 0.00010052642344611402,
      "loss": 1.0323,
      "step": 5377
    },
    {
      "epoch": 0.9994424828098867,
      "grad_norm": 0.5217851996421814,
      "learning_rate": 0.00010049717794716884,
      "loss": 1.0468,
      "step": 5378
    },
    {
      "epoch": 0.9996283218732578,
      "grad_norm": 0.6314026117324829,
      "learning_rate": 0.00010046793240569894,
      "loss": 1.0798,
      "step": 5379
    },
    {
      "epoch": 0.9998141609366289,
      "grad_norm": 0.5676242709159851,
      "learning_rate": 0.00010043868682420579,
      "loss": 1.239,
      "step": 5380
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6378072500228882,
      "learning_rate": 0.0001004094412051908,
      "loss": 1.1327,
      "step": 5381
    },
    {
      "epoch": 1.000185839063371,
      "grad_norm": 0.6097329258918762,
      "learning_rate": 0.00010038019555115543,
      "loss": 1.0793,
      "step": 5382
    },
    {
      "epoch": 1.0003716781267422,
      "grad_norm": 0.6979326605796814,
      "learning_rate": 0.00010035094986460112,
      "loss": 1.4729,
      "step": 5383
    },
    {
      "epoch": 1.0005575171901133,
      "grad_norm": 0.4561534523963928,
      "learning_rate": 0.00010032170414802926,
      "loss": 0.9795,
      "step": 5384
    },
    {
      "epoch": 1.0005575171901133,
      "eval_loss": 1.0240511894226074,
      "eval_runtime": 23.0165,
      "eval_samples_per_second": 47.444,
      "eval_steps_per_second": 23.722,
      "step": 5384
    },
    {
      "epoch": 1.0007433562534844,
      "grad_norm": 0.76151442527771,
      "learning_rate": 0.00010029245840394137,
      "loss": 1.1695,
      "step": 5385
    },
    {
      "epoch": 1.0009291953168555,
      "grad_norm": 0.5457819104194641,
      "learning_rate": 0.0001002632126348389,
      "loss": 0.8923,
      "step": 5386
    },
    {
      "epoch": 1.0011150343802266,
      "grad_norm": 0.57408207654953,
      "learning_rate": 0.00010023396684322325,
      "loss": 0.7989,
      "step": 5387
    },
    {
      "epoch": 1.0013008734435977,
      "grad_norm": 0.5285831093788147,
      "learning_rate": 0.00010020472103159595,
      "loss": 0.9254,
      "step": 5388
    },
    {
      "epoch": 1.0014867125069689,
      "grad_norm": 0.5257111191749573,
      "learning_rate": 0.0001001754752024584,
      "loss": 0.9928,
      "step": 5389
    },
    {
      "epoch": 1.0016725515703402,
      "grad_norm": 0.5929964184761047,
      "learning_rate": 0.0001001462293583121,
      "loss": 1.17,
      "step": 5390
    },
    {
      "epoch": 1.0018583906337113,
      "grad_norm": 0.6292116641998291,
      "learning_rate": 0.00010011698350165844,
      "loss": 1.2623,
      "step": 5391
    },
    {
      "epoch": 1.0020442296970824,
      "grad_norm": 0.9348997473716736,
      "learning_rate": 0.00010008773763499897,
      "loss": 1.1079,
      "step": 5392
    },
    {
      "epoch": 1.0022300687604535,
      "grad_norm": 0.6319088339805603,
      "learning_rate": 0.00010005849176083507,
      "loss": 1.253,
      "step": 5393
    },
    {
      "epoch": 1.0024159078238246,
      "grad_norm": Infinity,
      "learning_rate": 0.00010005849176083507,
      "loss": 2.0546,
      "step": 5394
    },
    {
      "epoch": 1.0026017468871957,
      "grad_norm": 0.6206772327423096,
      "learning_rate": 0.00010002924588166827,
      "loss": 1.1599,
      "step": 5395
    },
    {
      "epoch": 1.0027875859505668,
      "grad_norm": 0.6307805776596069,
      "learning_rate": 0.0001,
      "loss": 1.1913,
      "step": 5396
    },
    {
      "epoch": 1.002973425013938,
      "grad_norm": 0.6121634840965271,
      "learning_rate": 9.997075411833174e-05,
      "loss": 1.3004,
      "step": 5397
    },
    {
      "epoch": 1.003159264077309,
      "grad_norm": 0.5531421899795532,
      "learning_rate": 9.994150823916494e-05,
      "loss": 0.9743,
      "step": 5398
    },
    {
      "epoch": 1.0033451031406802,
      "grad_norm": 0.49836012721061707,
      "learning_rate": 9.991226236500106e-05,
      "loss": 1.1118,
      "step": 5399
    },
    {
      "epoch": 1.0035309422040513,
      "grad_norm": 0.6006326079368591,
      "learning_rate": 9.988301649834157e-05,
      "loss": 1.15,
      "step": 5400
    },
    {
      "epoch": 1.0037167812674224,
      "grad_norm": 0.5552299618721008,
      "learning_rate": 9.985377064168792e-05,
      "loss": 0.7896,
      "step": 5401
    },
    {
      "epoch": 1.0039026203307935,
      "grad_norm": 0.5896706581115723,
      "learning_rate": 9.982452479754163e-05,
      "loss": 1.0079,
      "step": 5402
    },
    {
      "epoch": 1.0040884593941646,
      "grad_norm": 0.5791441202163696,
      "learning_rate": 9.979527896840407e-05,
      "loss": 1.0957,
      "step": 5403
    },
    {
      "epoch": 1.0042742984575357,
      "grad_norm": 0.6279758214950562,
      "learning_rate": 9.976603315677674e-05,
      "loss": 1.0818,
      "step": 5404
    },
    {
      "epoch": 1.0044601375209068,
      "grad_norm": 0.6163389086723328,
      "learning_rate": 9.973678736516113e-05,
      "loss": 0.7001,
      "step": 5405
    },
    {
      "epoch": 1.0046459765842781,
      "grad_norm": 0.7090188264846802,
      "learning_rate": 9.970754159605864e-05,
      "loss": 1.1302,
      "step": 5406
    },
    {
      "epoch": 1.0048318156476492,
      "grad_norm": 0.5731585025787354,
      "learning_rate": 9.967829585197076e-05,
      "loss": 1.1057,
      "step": 5407
    },
    {
      "epoch": 1.0050176547110203,
      "grad_norm": 0.5643604397773743,
      "learning_rate": 9.964905013539892e-05,
      "loss": 1.0489,
      "step": 5408
    },
    {
      "epoch": 1.0052034937743914,
      "grad_norm": 0.6144115924835205,
      "learning_rate": 9.961980444884461e-05,
      "loss": 1.0398,
      "step": 5409
    },
    {
      "epoch": 1.0053893328377626,
      "grad_norm": 0.6871153116226196,
      "learning_rate": 9.959055879480921e-05,
      "loss": 0.9326,
      "step": 5410
    },
    {
      "epoch": 1.0055751719011337,
      "grad_norm": 0.6981655359268188,
      "learning_rate": 9.956131317579421e-05,
      "loss": 1.0339,
      "step": 5411
    },
    {
      "epoch": 1.0057610109645048,
      "grad_norm": 0.5825377702713013,
      "learning_rate": 9.953206759430107e-05,
      "loss": 0.906,
      "step": 5412
    },
    {
      "epoch": 1.0059468500278759,
      "grad_norm": 0.6913666129112244,
      "learning_rate": 9.950282205283117e-05,
      "loss": 1.0416,
      "step": 5413
    },
    {
      "epoch": 1.006132689091247,
      "grad_norm": 0.6037838459014893,
      "learning_rate": 9.947357655388599e-05,
      "loss": 1.0389,
      "step": 5414
    },
    {
      "epoch": 1.006318528154618,
      "grad_norm": 0.6384328603744507,
      "learning_rate": 9.944433109996692e-05,
      "loss": 1.1716,
      "step": 5415
    },
    {
      "epoch": 1.0065043672179892,
      "grad_norm": 0.6024790406227112,
      "learning_rate": 9.941508569357545e-05,
      "loss": 0.9292,
      "step": 5416
    },
    {
      "epoch": 1.0066902062813603,
      "grad_norm": 0.5940973162651062,
      "learning_rate": 9.938584033721294e-05,
      "loss": 1.002,
      "step": 5417
    },
    {
      "epoch": 1.0068760453447314,
      "grad_norm": 0.5688101053237915,
      "learning_rate": 9.935659503338082e-05,
      "loss": 0.9342,
      "step": 5418
    },
    {
      "epoch": 1.0070618844081025,
      "grad_norm": 0.6713395118713379,
      "learning_rate": 9.932734978458057e-05,
      "loss": 1.1313,
      "step": 5419
    },
    {
      "epoch": 1.0072477234714736,
      "grad_norm": 0.597057580947876,
      "learning_rate": 9.929810459331349e-05,
      "loss": 1.0405,
      "step": 5420
    },
    {
      "epoch": 1.0074335625348447,
      "grad_norm": 0.5775242447853088,
      "learning_rate": 9.926885946208108e-05,
      "loss": 1.1132,
      "step": 5421
    },
    {
      "epoch": 1.0076194015982158,
      "grad_norm": 0.4729931950569153,
      "learning_rate": 9.923961439338468e-05,
      "loss": 0.8111,
      "step": 5422
    },
    {
      "epoch": 1.0078052406615872,
      "grad_norm": 0.7457094192504883,
      "learning_rate": 9.921036938972572e-05,
      "loss": 1.4792,
      "step": 5423
    },
    {
      "epoch": 1.0079910797249583,
      "grad_norm": 0.5729731321334839,
      "learning_rate": 9.918112445360554e-05,
      "loss": 1.0344,
      "step": 5424
    },
    {
      "epoch": 1.0081769187883294,
      "grad_norm": 0.6670306324958801,
      "learning_rate": 9.915187958752556e-05,
      "loss": 1.0718,
      "step": 5425
    },
    {
      "epoch": 1.0083627578517005,
      "grad_norm": 0.5350519418716431,
      "learning_rate": 9.91226347939872e-05,
      "loss": 1.0075,
      "step": 5426
    },
    {
      "epoch": 1.0085485969150716,
      "grad_norm": 0.5211275219917297,
      "learning_rate": 9.909339007549175e-05,
      "loss": 1.0576,
      "step": 5427
    },
    {
      "epoch": 1.0087344359784427,
      "grad_norm": 0.6536908149719238,
      "learning_rate": 9.906414543454063e-05,
      "loss": 1.0691,
      "step": 5428
    },
    {
      "epoch": 1.0089202750418138,
      "grad_norm": 0.5251681208610535,
      "learning_rate": 9.903490087363514e-05,
      "loss": 0.8627,
      "step": 5429
    },
    {
      "epoch": 1.009106114105185,
      "grad_norm": 0.5836920142173767,
      "learning_rate": 9.900565639527668e-05,
      "loss": 1.0218,
      "step": 5430
    },
    {
      "epoch": 1.009291953168556,
      "grad_norm": 0.5761560797691345,
      "learning_rate": 9.897641200196656e-05,
      "loss": 1.1952,
      "step": 5431
    },
    {
      "epoch": 1.0094777922319271,
      "grad_norm": 0.5624791383743286,
      "learning_rate": 9.894716769620612e-05,
      "loss": 1.0669,
      "step": 5432
    },
    {
      "epoch": 1.0096636312952982,
      "grad_norm": 0.5359252095222473,
      "learning_rate": 9.891792348049676e-05,
      "loss": 1.156,
      "step": 5433
    },
    {
      "epoch": 1.0098494703586693,
      "grad_norm": 0.6264462471008301,
      "learning_rate": 9.88886793573397e-05,
      "loss": 0.8701,
      "step": 5434
    },
    {
      "epoch": 1.0100353094220405,
      "grad_norm": 0.5481414198875427,
      "learning_rate": 9.885943532923633e-05,
      "loss": 0.9676,
      "step": 5435
    },
    {
      "epoch": 1.0102211484854116,
      "grad_norm": 0.5599638819694519,
      "learning_rate": 9.883019139868789e-05,
      "loss": 1.0764,
      "step": 5436
    },
    {
      "epoch": 1.0104069875487827,
      "grad_norm": 0.5427635312080383,
      "learning_rate": 9.880094756819572e-05,
      "loss": 1.1117,
      "step": 5437
    },
    {
      "epoch": 1.0105928266121538,
      "grad_norm": 0.56276535987854,
      "learning_rate": 9.877170384026109e-05,
      "loss": 1.2726,
      "step": 5438
    },
    {
      "epoch": 1.010778665675525,
      "grad_norm": 0.6589742302894592,
      "learning_rate": 9.874246021738526e-05,
      "loss": 1.0081,
      "step": 5439
    },
    {
      "epoch": 1.0109645047388962,
      "grad_norm": 0.6330572366714478,
      "learning_rate": 9.87132167020696e-05,
      "loss": 0.9855,
      "step": 5440
    },
    {
      "epoch": 1.0111503438022673,
      "grad_norm": 0.5472434163093567,
      "learning_rate": 9.868397329681523e-05,
      "loss": 1.0102,
      "step": 5441
    },
    {
      "epoch": 1.0113361828656384,
      "grad_norm": 0.6067285537719727,
      "learning_rate": 9.865473000412352e-05,
      "loss": 0.7558,
      "step": 5442
    },
    {
      "epoch": 1.0115220219290095,
      "grad_norm": 0.7218596339225769,
      "learning_rate": 9.86254868264956e-05,
      "loss": 1.2293,
      "step": 5443
    },
    {
      "epoch": 1.0117078609923806,
      "grad_norm": 9.477654457092285,
      "learning_rate": 9.859624376643281e-05,
      "loss": 1.7758,
      "step": 5444
    },
    {
      "epoch": 1.0118937000557517,
      "grad_norm": 0.6567401885986328,
      "learning_rate": 9.856700082643628e-05,
      "loss": 0.9737,
      "step": 5445
    },
    {
      "epoch": 1.0120795391191229,
      "grad_norm": 0.6458907127380371,
      "learning_rate": 9.853775800900728e-05,
      "loss": 1.2239,
      "step": 5446
    },
    {
      "epoch": 1.012265378182494,
      "grad_norm": 0.5758485198020935,
      "learning_rate": 9.850851531664705e-05,
      "loss": 0.8985,
      "step": 5447
    },
    {
      "epoch": 1.012451217245865,
      "grad_norm": 0.7790505290031433,
      "learning_rate": 9.847927275185667e-05,
      "loss": 1.1216,
      "step": 5448
    },
    {
      "epoch": 1.0126370563092362,
      "grad_norm": 0.6475933194160461,
      "learning_rate": 9.845003031713743e-05,
      "loss": 1.2642,
      "step": 5449
    },
    {
      "epoch": 1.0128228953726073,
      "grad_norm": 0.585686981678009,
      "learning_rate": 9.84207880149904e-05,
      "loss": 1.2002,
      "step": 5450
    },
    {
      "epoch": 1.0130087344359784,
      "grad_norm": 0.5584164261817932,
      "learning_rate": 9.83915458479168e-05,
      "loss": 0.9101,
      "step": 5451
    },
    {
      "epoch": 1.0131945734993495,
      "grad_norm": 0.6801714897155762,
      "learning_rate": 9.836230381841775e-05,
      "loss": 1.1562,
      "step": 5452
    },
    {
      "epoch": 1.0133804125627206,
      "grad_norm": 0.6504145860671997,
      "learning_rate": 9.833306192899438e-05,
      "loss": 1.0963,
      "step": 5453
    },
    {
      "epoch": 1.0135662516260917,
      "grad_norm": 0.651237428188324,
      "learning_rate": 9.830382018214788e-05,
      "loss": 1.3058,
      "step": 5454
    },
    {
      "epoch": 1.0137520906894628,
      "grad_norm": 0.5830672979354858,
      "learning_rate": 9.827457858037926e-05,
      "loss": 0.9947,
      "step": 5455
    },
    {
      "epoch": 1.0139379297528341,
      "grad_norm": 0.6010664701461792,
      "learning_rate": 9.824533712618971e-05,
      "loss": 1.2405,
      "step": 5456
    },
    {
      "epoch": 1.0141237688162053,
      "grad_norm": 0.5090926885604858,
      "learning_rate": 9.82160958220802e-05,
      "loss": 0.8201,
      "step": 5457
    },
    {
      "epoch": 1.0143096078795764,
      "grad_norm": 0.6615139245986938,
      "learning_rate": 9.818685467055193e-05,
      "loss": 1.3562,
      "step": 5458
    },
    {
      "epoch": 1.0144954469429475,
      "grad_norm": 0.5601671934127808,
      "learning_rate": 9.815761367410585e-05,
      "loss": 0.9291,
      "step": 5459
    },
    {
      "epoch": 1.0146812860063186,
      "grad_norm": 0.6100068688392639,
      "learning_rate": 9.812837283524306e-05,
      "loss": 1.0619,
      "step": 5460
    },
    {
      "epoch": 1.0148671250696897,
      "grad_norm": 0.7463274598121643,
      "learning_rate": 9.809913215646464e-05,
      "loss": 0.9276,
      "step": 5461
    },
    {
      "epoch": 1.0150529641330608,
      "grad_norm": 0.6056248545646667,
      "learning_rate": 9.80698916402715e-05,
      "loss": 0.9859,
      "step": 5462
    },
    {
      "epoch": 1.015238803196432,
      "grad_norm": 0.5375638008117676,
      "learning_rate": 9.804065128916475e-05,
      "loss": 0.8173,
      "step": 5463
    },
    {
      "epoch": 1.015424642259803,
      "grad_norm": 0.6325738430023193,
      "learning_rate": 9.801141110564529e-05,
      "loss": 0.9961,
      "step": 5464
    },
    {
      "epoch": 1.015610481323174,
      "grad_norm": 0.5336713790893555,
      "learning_rate": 9.798217109221415e-05,
      "loss": 1.0304,
      "step": 5465
    },
    {
      "epoch": 1.0157963203865452,
      "grad_norm": 0.6309736371040344,
      "learning_rate": 9.795293125137225e-05,
      "loss": 0.8847,
      "step": 5466
    },
    {
      "epoch": 1.0159821594499163,
      "grad_norm": 0.633857250213623,
      "learning_rate": 9.79236915856206e-05,
      "loss": 0.842,
      "step": 5467
    },
    {
      "epoch": 1.0161679985132874,
      "grad_norm": 0.689595639705658,
      "learning_rate": 9.78944520974601e-05,
      "loss": 0.9321,
      "step": 5468
    },
    {
      "epoch": 1.0163538375766585,
      "grad_norm": 0.6718122959136963,
      "learning_rate": 9.786521278939163e-05,
      "loss": 1.0637,
      "step": 5469
    },
    {
      "epoch": 1.0165396766400296,
      "grad_norm": 0.6352799534797668,
      "learning_rate": 9.783597366391616e-05,
      "loss": 1.1004,
      "step": 5470
    },
    {
      "epoch": 1.0167255157034008,
      "grad_norm": 0.6477722525596619,
      "learning_rate": 9.780673472353451e-05,
      "loss": 1.145,
      "step": 5471
    },
    {
      "epoch": 1.016911354766772,
      "grad_norm": 0.5410536527633667,
      "learning_rate": 9.777749597074761e-05,
      "loss": 1.0556,
      "step": 5472
    },
    {
      "epoch": 1.0170971938301432,
      "grad_norm": 0.5520496368408203,
      "learning_rate": 9.774825740805626e-05,
      "loss": 0.8118,
      "step": 5473
    },
    {
      "epoch": 1.0172830328935143,
      "grad_norm": 0.8518608808517456,
      "learning_rate": 9.77190190379613e-05,
      "loss": 1.0747,
      "step": 5474
    },
    {
      "epoch": 1.000185839063371,
      "grad_norm": 0.561589241027832,
      "learning_rate": 9.768978086296361e-05,
      "loss": 1.0477,
      "step": 5475
    },
    {
      "epoch": 1.0003716781267422,
      "grad_norm": 0.6613037586212158,
      "learning_rate": 9.766054288556392e-05,
      "loss": 1.1642,
      "step": 5476
    },
    {
      "epoch": 1.0005575171901133,
      "grad_norm": 0.5221126079559326,
      "learning_rate": 9.76313051082631e-05,
      "loss": 0.8971,
      "step": 5477
    },
    {
      "epoch": 1.0007433562534844,
      "grad_norm": 0.5021414756774902,
      "learning_rate": 9.760206753356184e-05,
      "loss": 0.6635,
      "step": 5478
    },
    {
      "epoch": 1.0009291953168555,
      "grad_norm": 0.6210930943489075,
      "learning_rate": 9.757283016396096e-05,
      "loss": 0.8806,
      "step": 5479
    },
    {
      "epoch": 1.0011150343802266,
      "grad_norm": 0.5278712511062622,
      "learning_rate": 9.754359300196112e-05,
      "loss": 0.81,
      "step": 5480
    },
    {
      "epoch": 1.0013008734435977,
      "grad_norm": 0.6328251957893372,
      "learning_rate": 9.751435605006306e-05,
      "loss": 1.0801,
      "step": 5481
    },
    {
      "epoch": 1.0014867125069689,
      "grad_norm": 0.5093181133270264,
      "learning_rate": 9.748511931076756e-05,
      "loss": 0.9716,
      "step": 5482
    },
    {
      "epoch": 1.0016725515703402,
      "grad_norm": 0.5930023193359375,
      "learning_rate": 9.74558827865752e-05,
      "loss": 1.1575,
      "step": 5483
    },
    {
      "epoch": 1.0018583906337113,
      "grad_norm": 0.5147334337234497,
      "learning_rate": 9.742664647998673e-05,
      "loss": 0.9622,
      "step": 5484
    },
    {
      "epoch": 1.0020442296970824,
      "grad_norm": 0.7055521011352539,
      "learning_rate": 9.73974103935027e-05,
      "loss": 0.9228,
      "step": 5485
    },
    {
      "epoch": 1.0022300687604535,
      "grad_norm": 0.550630509853363,
      "learning_rate": 9.736817452962382e-05,
      "loss": 1.0508,
      "step": 5486
    },
    {
      "epoch": 1.0024159078238246,
      "grad_norm": 0.5721721053123474,
      "learning_rate": 9.733893889085063e-05,
      "loss": 0.7281,
      "step": 5487
    },
    {
      "epoch": 1.0026017468871957,
      "grad_norm": 0.49507665634155273,
      "learning_rate": 9.730970347968374e-05,
      "loss": 1.0323,
      "step": 5488
    },
    {
      "epoch": 1.0027875859505668,
      "grad_norm": 0.5477926731109619,
      "learning_rate": 9.728046829862376e-05,
      "loss": 0.9139,
      "step": 5489
    },
    {
      "epoch": 1.002973425013938,
      "grad_norm": 0.5683122873306274,
      "learning_rate": 9.725123335017117e-05,
      "loss": 0.8201,
      "step": 5490
    },
    {
      "epoch": 1.003159264077309,
      "grad_norm": 0.7958745360374451,
      "learning_rate": 9.722199863682659e-05,
      "loss": 1.2047,
      "step": 5491
    },
    {
      "epoch": 1.0033451031406802,
      "grad_norm": 0.6453753113746643,
      "learning_rate": 9.719276416109042e-05,
      "loss": 1.2061,
      "step": 5492
    },
    {
      "epoch": 1.0035309422040513,
      "grad_norm": 0.5566391944885254,
      "learning_rate": 9.716352992546323e-05,
      "loss": 1.0988,
      "step": 5493
    },
    {
      "epoch": 1.0037167812674224,
      "grad_norm": 0.5664240717887878,
      "learning_rate": 9.713429593244541e-05,
      "loss": 1.0277,
      "step": 5494
    },
    {
      "epoch": 1.0039026203307935,
      "grad_norm": 0.5620260238647461,
      "learning_rate": 9.710506218453748e-05,
      "loss": 0.9684,
      "step": 5495
    },
    {
      "epoch": 1.0040884593941646,
      "grad_norm": 0.5455220341682434,
      "learning_rate": 9.707582868423984e-05,
      "loss": 0.6991,
      "step": 5496
    },
    {
      "epoch": 1.0042742984575357,
      "grad_norm": 0.5864872336387634,
      "learning_rate": 9.704659543405287e-05,
      "loss": 0.9655,
      "step": 5497
    },
    {
      "epoch": 1.0044601375209068,
      "grad_norm": 0.49677610397338867,
      "learning_rate": 9.701736243647703e-05,
      "loss": 0.7792,
      "step": 5498
    },
    {
      "epoch": 1.0046459765842781,
      "grad_norm": 0.517552375793457,
      "learning_rate": 9.698812969401257e-05,
      "loss": 0.6622,
      "step": 5499
    },
    {
      "epoch": 1.0048318156476492,
      "grad_norm": 0.6072260141372681,
      "learning_rate": 9.695889720915994e-05,
      "loss": 0.9556,
      "step": 5500
    },
    {
      "epoch": 1.0050176547110203,
      "grad_norm": 0.6447886824607849,
      "learning_rate": 9.692966498441937e-05,
      "loss": 0.8409,
      "step": 5501
    },
    {
      "epoch": 1.0052034937743914,
      "grad_norm": 0.6072793006896973,
      "learning_rate": 9.690043302229117e-05,
      "loss": 0.8845,
      "step": 5502
    },
    {
      "epoch": 1.0053893328377626,
      "grad_norm": 0.5621128082275391,
      "learning_rate": 9.687120132527568e-05,
      "loss": 1.0741,
      "step": 5503
    },
    {
      "epoch": 1.0055751719011337,
      "grad_norm": 0.5688579678535461,
      "learning_rate": 9.684196989587305e-05,
      "loss": 0.9511,
      "step": 5504
    },
    {
      "epoch": 1.0057610109645048,
      "grad_norm": 0.5930265188217163,
      "learning_rate": 9.681273873658362e-05,
      "loss": 0.9649,
      "step": 5505
    },
    {
      "epoch": 1.0059468500278759,
      "grad_norm": 0.575786292552948,
      "learning_rate": 9.678350784990748e-05,
      "loss": 1.1046,
      "step": 5506
    },
    {
      "epoch": 1.006132689091247,
      "grad_norm": 0.5332211256027222,
      "learning_rate": 9.675427723834493e-05,
      "loss": 0.9617,
      "step": 5507
    },
    {
      "epoch": 1.006318528154618,
      "grad_norm": 0.6209401488304138,
      "learning_rate": 9.672504690439601e-05,
      "loss": 0.8199,
      "step": 5508
    },
    {
      "epoch": 1.0065043672179892,
      "grad_norm": 0.5827673673629761,
      "learning_rate": 9.669581685056089e-05,
      "loss": 0.9387,
      "step": 5509
    },
    {
      "epoch": 1.0066902062813603,
      "grad_norm": 0.5273739695549011,
      "learning_rate": 9.666658707933974e-05,
      "loss": 0.8929,
      "step": 5510
    },
    {
      "epoch": 1.0068760453447314,
      "grad_norm": 0.6448649168014526,
      "learning_rate": 9.663735759323258e-05,
      "loss": 0.9621,
      "step": 5511
    },
    {
      "epoch": 1.0070618844081025,
      "grad_norm": 0.5527610182762146,
      "learning_rate": 9.660812839473952e-05,
      "loss": 0.9894,
      "step": 5512
    },
    {
      "epoch": 1.0072477234714736,
      "grad_norm": 0.6742172837257385,
      "learning_rate": 9.657889948636053e-05,
      "loss": 0.9834,
      "step": 5513
    },
    {
      "epoch": 1.0074335625348447,
      "grad_norm": 0.7824978828430176,
      "learning_rate": 9.654967087059572e-05,
      "loss": 0.9771,
      "step": 5514
    },
    {
      "epoch": 1.0076194015982158,
      "grad_norm": 0.5018360614776611,
      "learning_rate": 9.652044254994495e-05,
      "loss": 0.4235,
      "step": 5515
    },
    {
      "epoch": 1.0078052406615872,
      "grad_norm": 0.5719367265701294,
      "learning_rate": 9.649121452690826e-05,
      "loss": 1.044,
      "step": 5516
    },
    {
      "epoch": 1.0079910797249583,
      "grad_norm": 0.5926535725593567,
      "learning_rate": 9.64619868039856e-05,
      "loss": 0.8208,
      "step": 5517
    },
    {
      "epoch": 1.0081769187883294,
      "grad_norm": 0.5106927752494812,
      "learning_rate": 9.643275938367685e-05,
      "loss": 0.8282,
      "step": 5518
    },
    {
      "epoch": 1.0083627578517005,
      "grad_norm": 0.5779476165771484,
      "learning_rate": 9.64035322684819e-05,
      "loss": 1.0108,
      "step": 5519
    },
    {
      "epoch": 1.0085485969150716,
      "grad_norm": 0.8551607131958008,
      "learning_rate": 9.637430546090059e-05,
      "loss": 0.9728,
      "step": 5520
    },
    {
      "epoch": 1.0087344359784427,
      "grad_norm": 0.5459856986999512,
      "learning_rate": 9.634507896343281e-05,
      "loss": 0.9183,
      "step": 5521
    },
    {
      "epoch": 1.0089202750418138,
      "grad_norm": 0.5723996758460999,
      "learning_rate": 9.631585277857829e-05,
      "loss": 1.1975,
      "step": 5522
    },
    {
      "epoch": 1.009106114105185,
      "grad_norm": 0.5988956093788147,
      "learning_rate": 9.628662690883685e-05,
      "loss": 0.9579,
      "step": 5523
    },
    {
      "epoch": 1.009291953168556,
      "grad_norm": 0.6050418615341187,
      "learning_rate": 9.625740135670825e-05,
      "loss": 0.8478,
      "step": 5524
    },
    {
      "epoch": 1.0094777922319271,
      "grad_norm": 0.6499052047729492,
      "learning_rate": 9.622817612469218e-05,
      "loss": 0.9527,
      "step": 5525
    },
    {
      "epoch": 1.0096636312952982,
      "grad_norm": 0.6234305500984192,
      "learning_rate": 9.619895121528838e-05,
      "loss": 0.9708,
      "step": 5526
    },
    {
      "epoch": 1.0098494703586693,
      "grad_norm": 0.6313818097114563,
      "learning_rate": 9.616972663099647e-05,
      "loss": 1.0714,
      "step": 5527
    },
    {
      "epoch": 1.0100353094220405,
      "grad_norm": 0.634307861328125,
      "learning_rate": 9.614050237431616e-05,
      "loss": 0.9302,
      "step": 5528
    },
    {
      "epoch": 1.0102211484854116,
      "grad_norm": 0.5377451181411743,
      "learning_rate": 9.611127844774699e-05,
      "loss": 0.8374,
      "step": 5529
    },
    {
      "epoch": 1.0104069875487827,
      "grad_norm": 0.5915167927742004,
      "learning_rate": 9.608205485378858e-05,
      "loss": 0.9394,
      "step": 5530
    },
    {
      "epoch": 1.0105928266121538,
      "grad_norm": 0.764411211013794,
      "learning_rate": 9.605283159494051e-05,
      "loss": 0.9691,
      "step": 5531
    },
    {
      "epoch": 1.010778665675525,
      "grad_norm": 0.608314037322998,
      "learning_rate": 9.602360867370226e-05,
      "loss": 0.965,
      "step": 5532
    },
    {
      "epoch": 1.0109645047388962,
      "grad_norm": 0.5562364459037781,
      "learning_rate": 9.599438609257336e-05,
      "loss": 0.9165,
      "step": 5533
    },
    {
      "epoch": 1.0111503438022673,
      "grad_norm": 0.6700636148452759,
      "learning_rate": 9.596516385405327e-05,
      "loss": 1.0618,
      "step": 5534
    },
    {
      "epoch": 1.0113361828656384,
      "grad_norm": 0.5265184044837952,
      "learning_rate": 9.593594196064147e-05,
      "loss": 1.036,
      "step": 5535
    },
    {
      "epoch": 1.0115220219290095,
      "grad_norm": 0.5445021390914917,
      "learning_rate": 9.590672041483731e-05,
      "loss": 1.0627,
      "step": 5536
    },
    {
      "epoch": 1.0117078609923806,
      "grad_norm": 0.859940767288208,
      "learning_rate": 9.587749921914018e-05,
      "loss": 1.1261,
      "step": 5537
    },
    {
      "epoch": 1.0118937000557517,
      "grad_norm": 1.5261372327804565,
      "learning_rate": 9.584827837604949e-05,
      "loss": 1.7484,
      "step": 5538
    },
    {
      "epoch": 1.0120795391191229,
      "grad_norm": 0.5151029825210571,
      "learning_rate": 9.58190578880645e-05,
      "loss": 0.7694,
      "step": 5539
    },
    {
      "epoch": 1.012265378182494,
      "grad_norm": 0.5031622052192688,
      "learning_rate": 9.578983775768454e-05,
      "loss": 0.6745,
      "step": 5540
    },
    {
      "epoch": 1.012451217245865,
      "grad_norm": 0.5906434059143066,
      "learning_rate": 9.576061798740882e-05,
      "loss": 0.6502,
      "step": 5541
    },
    {
      "epoch": 1.0126370563092362,
      "grad_norm": 0.6443628072738647,
      "learning_rate": 9.573139857973668e-05,
      "loss": 0.8002,
      "step": 5542
    },
    {
      "epoch": 1.0128228953726073,
      "grad_norm": 0.6574161052703857,
      "learning_rate": 9.570217953716718e-05,
      "loss": 0.8415,
      "step": 5543
    },
    {
      "epoch": 1.0130087344359784,
      "grad_norm": 0.6678338646888733,
      "learning_rate": 9.567296086219955e-05,
      "loss": 1.129,
      "step": 5544
    },
    {
      "epoch": 1.0131945734993495,
      "grad_norm": 0.6403939723968506,
      "learning_rate": 9.564374255733298e-05,
      "loss": 0.9736,
      "step": 5545
    },
    {
      "epoch": 1.0133804125627206,
      "grad_norm": 0.7221897840499878,
      "learning_rate": 9.561452462506648e-05,
      "loss": 0.9191,
      "step": 5546
    },
    {
      "epoch": 1.0135662516260917,
      "grad_norm": 0.6155973672866821,
      "learning_rate": 9.558530706789918e-05,
      "loss": 0.8291,
      "step": 5547
    },
    {
      "epoch": 1.0137520906894628,
      "grad_norm": 0.5788849592208862,
      "learning_rate": 9.55560898883301e-05,
      "loss": 0.8192,
      "step": 5548
    },
    {
      "epoch": 1.0139379297528341,
      "grad_norm": 0.5767815709114075,
      "learning_rate": 9.55268730888583e-05,
      "loss": 0.8011,
      "step": 5549
    },
    {
      "epoch": 1.0141237688162053,
      "grad_norm": 0.5603548884391785,
      "learning_rate": 9.549765667198267e-05,
      "loss": 0.8195,
      "step": 5550
    },
    {
      "epoch": 1.0143096078795764,
      "grad_norm": 0.6302038431167603,
      "learning_rate": 9.546844064020218e-05,
      "loss": 1.0481,
      "step": 5551
    },
    {
      "epoch": 1.0144954469429475,
      "grad_norm": 0.5809621214866638,
      "learning_rate": 9.543922499601582e-05,
      "loss": 0.739,
      "step": 5552
    },
    {
      "epoch": 1.0146812860063186,
      "grad_norm": 0.5936046838760376,
      "learning_rate": 9.541000974192234e-05,
      "loss": 1.0261,
      "step": 5553
    },
    {
      "epoch": 1.0148671250696897,
      "grad_norm": 0.6521549820899963,
      "learning_rate": 9.538079488042069e-05,
      "loss": 1.1009,
      "step": 5554
    },
    {
      "epoch": 1.0150529641330608,
      "grad_norm": 0.6870414614677429,
      "learning_rate": 9.53515804140096e-05,
      "loss": 0.8791,
      "step": 5555
    },
    {
      "epoch": 1.015238803196432,
      "grad_norm": 0.5661901235580444,
      "learning_rate": 9.532236634518793e-05,
      "loss": 1.0703,
      "step": 5556
    },
    {
      "epoch": 1.015424642259803,
      "grad_norm": 0.7870084047317505,
      "learning_rate": 9.529315267645433e-05,
      "loss": 1.0597,
      "step": 5557
    },
    {
      "epoch": 1.015610481323174,
      "grad_norm": 0.7044654488563538,
      "learning_rate": 9.526393941030756e-05,
      "loss": 0.7709,
      "step": 5558
    },
    {
      "epoch": 1.0157963203865452,
      "grad_norm": 0.6198689341545105,
      "learning_rate": 9.523472654924633e-05,
      "loss": 0.7839,
      "step": 5559
    },
    {
      "epoch": 1.0159821594499163,
      "grad_norm": 0.5579044222831726,
      "learning_rate": 9.520551409576919e-05,
      "loss": 0.7988,
      "step": 5560
    },
    {
      "epoch": 1.0161679985132874,
      "grad_norm": 0.47388043999671936,
      "learning_rate": 9.517630205237482e-05,
      "loss": 0.7251,
      "step": 5561
    },
    {
      "epoch": 1.0163538375766585,
      "grad_norm": 0.6593843698501587,
      "learning_rate": 9.514709042156173e-05,
      "loss": 0.8527,
      "step": 5562
    },
    {
      "epoch": 1.0165396766400296,
      "grad_norm": 0.6089546084403992,
      "learning_rate": 9.511787920582855e-05,
      "loss": 0.8445,
      "step": 5563
    },
    {
      "epoch": 1.0167255157034008,
      "grad_norm": 0.7765074372291565,
      "learning_rate": 9.508866840767365e-05,
      "loss": 0.942,
      "step": 5564
    },
    {
      "epoch": 1.016911354766772,
      "grad_norm": 0.5820897221565247,
      "learning_rate": 9.505945802959555e-05,
      "loss": 0.8503,
      "step": 5565
    },
    {
      "epoch": 1.0170971938301432,
      "grad_norm": 1.7710691690444946,
      "learning_rate": 9.503024807409274e-05,
      "loss": 1.5448,
      "step": 5566
    },
    {
      "epoch": 1.0172830328935143,
      "grad_norm": 0.637711763381958,
      "learning_rate": 9.500103854366351e-05,
      "loss": 0.8379,
      "step": 5567
    },
    {
      "epoch": 1.0174688719568854,
      "grad_norm": 0.5511047840118408,
      "learning_rate": 9.497182944080628e-05,
      "loss": 0.9107,
      "step": 5568
    },
    {
      "epoch": 1.0176547110202565,
      "grad_norm": 0.6024526953697205,
      "learning_rate": 9.494262076801933e-05,
      "loss": 0.9868,
      "step": 5569
    },
    {
      "epoch": 1.0178405500836276,
      "grad_norm": 1.2761352062225342,
      "learning_rate": 9.491341252780097e-05,
      "loss": 1.3302,
      "step": 5570
    },
    {
      "epoch": 1.0180263891469987,
      "grad_norm": 0.6507533192634583,
      "learning_rate": 9.48842047226494e-05,
      "loss": 1.0497,
      "step": 5571
    },
    {
      "epoch": 1.0182122282103698,
      "grad_norm": 0.6174860596656799,
      "learning_rate": 9.485499735506286e-05,
      "loss": 0.8212,
      "step": 5572
    },
    {
      "epoch": 1.018398067273741,
      "grad_norm": 0.5975897312164307,
      "learning_rate": 9.482579042753956e-05,
      "loss": 0.7139,
      "step": 5573
    },
    {
      "epoch": 1.018583906337112,
      "grad_norm": 0.6369990706443787,
      "learning_rate": 9.479658394257754e-05,
      "loss": 1.1584,
      "step": 5574
    },
    {
      "epoch": 1.0187697454004832,
      "grad_norm": 0.6980920433998108,
      "learning_rate": 9.476737790267501e-05,
      "loss": 0.9404,
      "step": 5575
    },
    {
      "epoch": 1.0189555844638543,
      "grad_norm": 0.7139065861701965,
      "learning_rate": 9.473817231032989e-05,
      "loss": 1.0755,
      "step": 5576
    },
    {
      "epoch": 1.0191414235272254,
      "grad_norm": 0.7761198878288269,
      "learning_rate": 9.470896716804029e-05,
      "loss": 1.0533,
      "step": 5577
    },
    {
      "epoch": 1.0193272625905965,
      "grad_norm": 0.545531153678894,
      "learning_rate": 9.467976247830413e-05,
      "loss": 1.0633,
      "step": 5578
    },
    {
      "epoch": 1.0195131016539676,
      "grad_norm": 0.6415326595306396,
      "learning_rate": 9.46505582436194e-05,
      "loss": 1.209,
      "step": 5579
    },
    {
      "epoch": 1.0196989407173387,
      "grad_norm": 0.6412090063095093,
      "learning_rate": 9.462135446648402e-05,
      "loss": 0.9798,
      "step": 5580
    },
    {
      "epoch": 1.01988477978071,
      "grad_norm": 0.6819164156913757,
      "learning_rate": 9.459215114939578e-05,
      "loss": 0.991,
      "step": 5581
    },
    {
      "epoch": 1.0200706188440811,
      "grad_norm": 0.6056860089302063,
      "learning_rate": 9.456294829485258e-05,
      "loss": 0.6767,
      "step": 5582
    },
    {
      "epoch": 1.0202564579074522,
      "grad_norm": 0.7711376547813416,
      "learning_rate": 9.453374590535212e-05,
      "loss": 0.9817,
      "step": 5583
    },
    {
      "epoch": 1.0204422969708233,
      "grad_norm": 0.5763733983039856,
      "learning_rate": 9.450454398339221e-05,
      "loss": 1.2615,
      "step": 5584
    },
    {
      "epoch": 1.0206281360341944,
      "grad_norm": 0.631025493144989,
      "learning_rate": 9.44753425314705e-05,
      "loss": 1.0366,
      "step": 5585
    },
    {
      "epoch": 1.0208139750975656,
      "grad_norm": 0.6333885192871094,
      "learning_rate": 9.44461415520847e-05,
      "loss": 0.9011,
      "step": 5586
    },
    {
      "epoch": 1.0209998141609367,
      "grad_norm": 0.621618926525116,
      "learning_rate": 9.441694104773245e-05,
      "loss": 0.8351,
      "step": 5587
    },
    {
      "epoch": 1.0211856532243078,
      "grad_norm": 0.5549609661102295,
      "learning_rate": 9.438774102091127e-05,
      "loss": 0.9983,
      "step": 5588
    },
    {
      "epoch": 1.0213714922876789,
      "grad_norm": 0.668755829334259,
      "learning_rate": 9.435854147411875e-05,
      "loss": 1.006,
      "step": 5589
    },
    {
      "epoch": 1.02155733135105,
      "grad_norm": 0.6217575073242188,
      "learning_rate": 9.432934240985234e-05,
      "loss": 0.7929,
      "step": 5590
    },
    {
      "epoch": 1.021743170414421,
      "grad_norm": 0.609810471534729,
      "learning_rate": 9.430014383060956e-05,
      "loss": 0.863,
      "step": 5591
    },
    {
      "epoch": 1.0219290094777922,
      "grad_norm": 0.5248240828514099,
      "learning_rate": 9.427094573888776e-05,
      "loss": 0.8328,
      "step": 5592
    },
    {
      "epoch": 1.0221148485411633,
      "grad_norm": 0.7486118078231812,
      "learning_rate": 9.424174813718436e-05,
      "loss": 1.1021,
      "step": 5593
    },
    {
      "epoch": 1.0223006876045344,
      "grad_norm": 0.6330525875091553,
      "learning_rate": 9.421255102799672e-05,
      "loss": 0.9775,
      "step": 5594
    },
    {
      "epoch": 1.0224865266679055,
      "grad_norm": 0.7093864679336548,
      "learning_rate": 9.418335441382206e-05,
      "loss": 0.9141,
      "step": 5595
    },
    {
      "epoch": 1.0226723657312766,
      "grad_norm": 0.5771816968917847,
      "learning_rate": 9.415415829715772e-05,
      "loss": 0.6835,
      "step": 5596
    },
    {
      "epoch": 1.0228582047946477,
      "grad_norm": 0.7342430949211121,
      "learning_rate": 9.412496268050082e-05,
      "loss": 0.8209,
      "step": 5597
    },
    {
      "epoch": 1.023044043858019,
      "grad_norm": 0.6433789730072021,
      "learning_rate": 9.409576756634856e-05,
      "loss": 1.0434,
      "step": 5598
    },
    {
      "epoch": 1.0232298829213902,
      "grad_norm": 0.513936460018158,
      "learning_rate": 9.406657295719805e-05,
      "loss": 0.6459,
      "step": 5599
    },
    {
      "epoch": 1.0234157219847613,
      "grad_norm": 0.6257773041725159,
      "learning_rate": 9.403737885554638e-05,
      "loss": 1.1122,
      "step": 5600
    },
    {
      "epoch": 1.0236015610481324,
      "grad_norm": 0.736271858215332,
      "learning_rate": 9.400818526389063e-05,
      "loss": 1.0869,
      "step": 5601
    },
    {
      "epoch": 1.0237874001115035,
      "grad_norm": 0.5881569385528564,
      "learning_rate": 9.397899218472769e-05,
      "loss": 1.034,
      "step": 5602
    },
    {
      "epoch": 1.0239732391748746,
      "grad_norm": 0.6881264448165894,
      "learning_rate": 9.39497996205546e-05,
      "loss": 0.8233,
      "step": 5603
    },
    {
      "epoch": 1.0241590782382457,
      "grad_norm": 0.5099987983703613,
      "learning_rate": 9.392060757386819e-05,
      "loss": 0.8129,
      "step": 5604
    },
    {
      "epoch": 1.0243449173016168,
      "grad_norm": 0.7611659169197083,
      "learning_rate": 9.389141604716536e-05,
      "loss": 1.0172,
      "step": 5605
    },
    {
      "epoch": 1.024530756364988,
      "grad_norm": 0.49309414625167847,
      "learning_rate": 9.386222504294293e-05,
      "loss": 0.7155,
      "step": 5606
    },
    {
      "epoch": 1.024716595428359,
      "grad_norm": 0.6789926886558533,
      "learning_rate": 9.383303456369764e-05,
      "loss": 1.3201,
      "step": 5607
    },
    {
      "epoch": 1.0249024344917301,
      "grad_norm": 0.5713966488838196,
      "learning_rate": 9.380384461192628e-05,
      "loss": 0.9373,
      "step": 5608
    },
    {
      "epoch": 1.0250882735551012,
      "grad_norm": 0.6206234693527222,
      "learning_rate": 9.377465519012543e-05,
      "loss": 0.896,
      "step": 5609
    },
    {
      "epoch": 1.0252741126184723,
      "grad_norm": 0.6701042652130127,
      "learning_rate": 9.374546630079183e-05,
      "loss": 0.8984,
      "step": 5610
    },
    {
      "epoch": 1.0254599516818435,
      "grad_norm": 0.5585264563560486,
      "learning_rate": 9.371627794642196e-05,
      "loss": 1.0708,
      "step": 5611
    },
    {
      "epoch": 1.0256457907452146,
      "grad_norm": 0.6523463726043701,
      "learning_rate": 9.368709012951243e-05,
      "loss": 1.0648,
      "step": 5612
    },
    {
      "epoch": 1.0258316298085857,
      "grad_norm": 0.6141130924224854,
      "learning_rate": 9.365790285255974e-05,
      "loss": 0.9465,
      "step": 5613
    },
    {
      "epoch": 1.0260174688719568,
      "grad_norm": 0.6018891334533691,
      "learning_rate": 9.362871611806032e-05,
      "loss": 0.9618,
      "step": 5614
    },
    {
      "epoch": 1.026203307935328,
      "grad_norm": 0.5785146951675415,
      "learning_rate": 9.35995299285106e-05,
      "loss": 0.6976,
      "step": 5615
    },
    {
      "epoch": 1.0263891469986992,
      "grad_norm": 0.6223296523094177,
      "learning_rate": 9.357034428640688e-05,
      "loss": 1.0349,
      "step": 5616
    },
    {
      "epoch": 1.0265749860620703,
      "grad_norm": 0.6218807697296143,
      "learning_rate": 9.354115919424556e-05,
      "loss": 0.9761,
      "step": 5617
    },
    {
      "epoch": 1.0267608251254414,
      "grad_norm": 0.6906150579452515,
      "learning_rate": 9.35119746545228e-05,
      "loss": 0.8762,
      "step": 5618
    },
    {
      "epoch": 1.0269466641888125,
      "grad_norm": 0.732918918132782,
      "learning_rate": 9.348279066973488e-05,
      "loss": 1.1839,
      "step": 5619
    },
    {
      "epoch": 1.0271325032521836,
      "grad_norm": 0.6837500333786011,
      "learning_rate": 9.345360724237799e-05,
      "loss": 0.8754,
      "step": 5620
    },
    {
      "epoch": 1.0273183423155547,
      "grad_norm": 0.6494518518447876,
      "learning_rate": 9.342442437494819e-05,
      "loss": 1.1035,
      "step": 5621
    },
    {
      "epoch": 1.0275041813789259,
      "grad_norm": 0.5419272184371948,
      "learning_rate": 9.339524206994162e-05,
      "loss": 0.9875,
      "step": 5622
    },
    {
      "epoch": 1.027690020442297,
      "grad_norm": 0.6595784425735474,
      "learning_rate": 9.336606032985422e-05,
      "loss": 0.9096,
      "step": 5623
    },
    {
      "epoch": 1.027875859505668,
      "grad_norm": 0.5965380668640137,
      "learning_rate": 9.333687915718207e-05,
      "loss": 0.8697,
      "step": 5624
    },
    {
      "epoch": 1.0280616985690392,
      "grad_norm": 0.6148014664649963,
      "learning_rate": 9.330769855442102e-05,
      "loss": 0.9502,
      "step": 5625
    },
    {
      "epoch": 1.0282475376324103,
      "grad_norm": 0.5834745764732361,
      "learning_rate": 9.327851852406696e-05,
      "loss": 0.8174,
      "step": 5626
    },
    {
      "epoch": 1.0284333766957814,
      "grad_norm": 0.5997938513755798,
      "learning_rate": 9.32493390686158e-05,
      "loss": 0.8045,
      "step": 5627
    },
    {
      "epoch": 1.0286192157591525,
      "grad_norm": 0.5949342846870422,
      "learning_rate": 9.322016019056319e-05,
      "loss": 0.9448,
      "step": 5628
    },
    {
      "epoch": 1.0288050548225236,
      "grad_norm": 0.6385858654975891,
      "learning_rate": 9.319098189240496e-05,
      "loss": 1.0089,
      "step": 5629
    },
    {
      "epoch": 1.0289908938858947,
      "grad_norm": 0.5652914047241211,
      "learning_rate": 9.316180417663675e-05,
      "loss": 0.9545,
      "step": 5630
    },
    {
      "epoch": 1.029176732949266,
      "grad_norm": 0.5521414279937744,
      "learning_rate": 9.313262704575426e-05,
      "loss": 0.9898,
      "step": 5631
    },
    {
      "epoch": 1.0293625720126371,
      "grad_norm": 0.6309528946876526,
      "learning_rate": 9.310345050225297e-05,
      "loss": 0.9947,
      "step": 5632
    },
    {
      "epoch": 1.0295484110760083,
      "grad_norm": 0.6038761138916016,
      "learning_rate": 9.307427454862844e-05,
      "loss": 0.9867,
      "step": 5633
    },
    {
      "epoch": 1.0297342501393794,
      "grad_norm": 0.720798909664154,
      "learning_rate": 9.304509918737626e-05,
      "loss": 0.9726,
      "step": 5634
    },
    {
      "epoch": 1.0299200892027505,
      "grad_norm": 0.5871129035949707,
      "learning_rate": 9.301592442099172e-05,
      "loss": 1.1046,
      "step": 5635
    },
    {
      "epoch": 1.0301059282661216,
      "grad_norm": 0.5038182735443115,
      "learning_rate": 9.298675025197027e-05,
      "loss": 0.8985,
      "step": 5636
    },
    {
      "epoch": 1.0302917673294927,
      "grad_norm": 0.4841303527355194,
      "learning_rate": 9.295757668280722e-05,
      "loss": 0.5612,
      "step": 5637
    },
    {
      "epoch": 1.0304776063928638,
      "grad_norm": 0.7170572876930237,
      "learning_rate": 9.292840371599789e-05,
      "loss": 1.0366,
      "step": 5638
    },
    {
      "epoch": 1.030663445456235,
      "grad_norm": 0.6779294610023499,
      "learning_rate": 9.289923135403743e-05,
      "loss": 1.0401,
      "step": 5639
    },
    {
      "epoch": 1.030849284519606,
      "grad_norm": 0.615813136100769,
      "learning_rate": 9.287005959942107e-05,
      "loss": 1.0795,
      "step": 5640
    },
    {
      "epoch": 1.031035123582977,
      "grad_norm": 0.7855369448661804,
      "learning_rate": 9.284088845464396e-05,
      "loss": 1.3407,
      "step": 5641
    },
    {
      "epoch": 1.0312209626463482,
      "grad_norm": 0.6328584551811218,
      "learning_rate": 9.281171792220107e-05,
      "loss": 0.8972,
      "step": 5642
    },
    {
      "epoch": 1.0314068017097193,
      "grad_norm": 0.6222116351127625,
      "learning_rate": 9.278254800458753e-05,
      "loss": 0.7151,
      "step": 5643
    },
    {
      "epoch": 1.0315926407730904,
      "grad_norm": 0.5513842105865479,
      "learning_rate": 9.275337870429824e-05,
      "loss": 0.9599,
      "step": 5644
    },
    {
      "epoch": 1.0317784798364615,
      "grad_norm": 0.683917224407196,
      "learning_rate": 9.272421002382816e-05,
      "loss": 1.2183,
      "step": 5645
    },
    {
      "epoch": 1.0319643188998326,
      "grad_norm": 0.5576063990592957,
      "learning_rate": 9.269504196567209e-05,
      "loss": 0.9097,
      "step": 5646
    },
    {
      "epoch": 1.032150157963204,
      "grad_norm": 0.7101729512214661,
      "learning_rate": 9.266587453232486e-05,
      "loss": 0.8636,
      "step": 5647
    },
    {
      "epoch": 1.032335997026575,
      "grad_norm": 0.6904899477958679,
      "learning_rate": 9.263670772628126e-05,
      "loss": 1.136,
      "step": 5648
    },
    {
      "epoch": 1.0325218360899462,
      "grad_norm": 0.6535913348197937,
      "learning_rate": 9.260754155003595e-05,
      "loss": 1.0645,
      "step": 5649
    },
    {
      "epoch": 1.0327076751533173,
      "grad_norm": 0.7509424090385437,
      "learning_rate": 9.257837600608359e-05,
      "loss": 1.1562,
      "step": 5650
    },
    {
      "epoch": 1.0328935142166884,
      "grad_norm": 0.6997819542884827,
      "learning_rate": 9.254921109691877e-05,
      "loss": 1.1857,
      "step": 5651
    },
    {
      "epoch": 1.0330793532800595,
      "grad_norm": 0.573625922203064,
      "learning_rate": 9.252004682503604e-05,
      "loss": 0.863,
      "step": 5652
    },
    {
      "epoch": 1.0332651923434306,
      "grad_norm": 0.5840150117874146,
      "learning_rate": 9.249088319292986e-05,
      "loss": 0.8821,
      "step": 5653
    },
    {
      "epoch": 1.0334510314068017,
      "grad_norm": 0.7319598197937012,
      "learning_rate": 9.246172020309465e-05,
      "loss": 1.0052,
      "step": 5654
    },
    {
      "epoch": 1.0336368704701728,
      "grad_norm": 0.6403792500495911,
      "learning_rate": 9.243255785802484e-05,
      "loss": 0.8859,
      "step": 5655
    },
    {
      "epoch": 1.033822709533544,
      "grad_norm": 0.6094220280647278,
      "learning_rate": 9.240339616021469e-05,
      "loss": 1.0,
      "step": 5656
    },
    {
      "epoch": 1.034008548596915,
      "grad_norm": 0.5595227479934692,
      "learning_rate": 9.23742351121585e-05,
      "loss": 0.9861,
      "step": 5657
    },
    {
      "epoch": 1.0341943876602862,
      "grad_norm": 0.688662588596344,
      "learning_rate": 9.234507471635043e-05,
      "loss": 1.0202,
      "step": 5658
    },
    {
      "epoch": 1.0343802267236573,
      "grad_norm": 0.6431695818901062,
      "learning_rate": 9.23159149752847e-05,
      "loss": 0.9423,
      "step": 5659
    },
    {
      "epoch": 1.0345660657870284,
      "grad_norm": 0.6177623867988586,
      "learning_rate": 9.228675589145536e-05,
      "loss": 1.0328,
      "step": 5660
    },
    {
      "epoch": 1.0347519048503995,
      "grad_norm": 0.6663259863853455,
      "learning_rate": 9.225759746735643e-05,
      "loss": 0.9991,
      "step": 5661
    },
    {
      "epoch": 1.0349377439137706,
      "grad_norm": 0.7278993129730225,
      "learning_rate": 9.222843970548198e-05,
      "loss": 0.9877,
      "step": 5662
    },
    {
      "epoch": 1.0351235829771417,
      "grad_norm": 0.6169386506080627,
      "learning_rate": 9.219928260832584e-05,
      "loss": 1.06,
      "step": 5663
    },
    {
      "epoch": 1.035309422040513,
      "grad_norm": 0.7969826459884644,
      "learning_rate": 9.217012617838192e-05,
      "loss": 0.8952,
      "step": 5664
    },
    {
      "epoch": 1.0354952611038841,
      "grad_norm": 0.6824052333831787,
      "learning_rate": 9.214097041814403e-05,
      "loss": 0.8959,
      "step": 5665
    },
    {
      "epoch": 1.0356811001672552,
      "grad_norm": 0.657074511051178,
      "learning_rate": 9.211181533010595e-05,
      "loss": 0.8222,
      "step": 5666
    },
    {
      "epoch": 1.0358669392306263,
      "grad_norm": 0.658909022808075,
      "learning_rate": 9.20826609167613e-05,
      "loss": 1.0181,
      "step": 5667
    },
    {
      "epoch": 1.0360527782939974,
      "grad_norm": 0.9237680435180664,
      "learning_rate": 9.205350718060381e-05,
      "loss": 1.093,
      "step": 5668
    },
    {
      "epoch": 1.0362386173573686,
      "grad_norm": 0.6500981450080872,
      "learning_rate": 9.202435412412705e-05,
      "loss": 1.0164,
      "step": 5669
    },
    {
      "epoch": 1.0364244564207397,
      "grad_norm": 0.5961865186691284,
      "learning_rate": 9.199520174982446e-05,
      "loss": 0.9181,
      "step": 5670
    },
    {
      "epoch": 1.0366102954841108,
      "grad_norm": 0.5907772183418274,
      "learning_rate": 9.196605006018963e-05,
      "loss": 0.9159,
      "step": 5671
    },
    {
      "epoch": 1.0367961345474819,
      "grad_norm": 0.61663419008255,
      "learning_rate": 9.193689905771586e-05,
      "loss": 0.9584,
      "step": 5672
    },
    {
      "epoch": 1.036981973610853,
      "grad_norm": 0.6415398716926575,
      "learning_rate": 9.190774874489655e-05,
      "loss": 1.1932,
      "step": 5673
    },
    {
      "epoch": 1.037167812674224,
      "grad_norm": 0.5706160664558411,
      "learning_rate": 9.187859912422497e-05,
      "loss": 0.882,
      "step": 5674
    },
    {
      "epoch": 1.0373536517375952,
      "grad_norm": 0.6299116015434265,
      "learning_rate": 9.184945019819435e-05,
      "loss": 0.6879,
      "step": 5675
    },
    {
      "epoch": 1.0375394908009663,
      "grad_norm": 0.6006621718406677,
      "learning_rate": 9.182030196929791e-05,
      "loss": 0.9731,
      "step": 5676
    },
    {
      "epoch": 1.0377253298643374,
      "grad_norm": 0.6359907388687134,
      "learning_rate": 9.17911544400287e-05,
      "loss": 0.9504,
      "step": 5677
    },
    {
      "epoch": 1.0379111689277085,
      "grad_norm": 0.6705753803253174,
      "learning_rate": 9.17620076128798e-05,
      "loss": 0.9331,
      "step": 5678
    },
    {
      "epoch": 1.0380970079910796,
      "grad_norm": 0.733623743057251,
      "learning_rate": 9.173286149034419e-05,
      "loss": 1.0244,
      "step": 5679
    },
    {
      "epoch": 1.038282847054451,
      "grad_norm": 0.5766941905021667,
      "learning_rate": 9.170371607491479e-05,
      "loss": 0.9199,
      "step": 5680
    },
    {
      "epoch": 1.038468686117822,
      "grad_norm": 0.5844184756278992,
      "learning_rate": 9.167457136908448e-05,
      "loss": 0.927,
      "step": 5681
    },
    {
      "epoch": 1.0386545251811932,
      "grad_norm": 0.6267839670181274,
      "learning_rate": 9.164542737534607e-05,
      "loss": 0.8806,
      "step": 5682
    },
    {
      "epoch": 1.0388403642445643,
      "grad_norm": 0.6894923448562622,
      "learning_rate": 9.161628409619236e-05,
      "loss": 0.7935,
      "step": 5683
    },
    {
      "epoch": 1.0390262033079354,
      "grad_norm": 0.5978687405586243,
      "learning_rate": 9.158714153411595e-05,
      "loss": 0.854,
      "step": 5684
    },
    {
      "epoch": 1.0392120423713065,
      "grad_norm": 0.8842291235923767,
      "learning_rate": 9.155799969160955e-05,
      "loss": 0.8898,
      "step": 5685
    },
    {
      "epoch": 1.0393978814346776,
      "grad_norm": 0.6014558672904968,
      "learning_rate": 9.152885857116564e-05,
      "loss": 0.9935,
      "step": 5686
    },
    {
      "epoch": 1.0395837204980487,
      "grad_norm": 0.6135658025741577,
      "learning_rate": 9.149971817527679e-05,
      "loss": 1.0825,
      "step": 5687
    },
    {
      "epoch": 1.0397695595614198,
      "grad_norm": 0.56199049949646,
      "learning_rate": 9.147057850643537e-05,
      "loss": 0.8605,
      "step": 5688
    },
    {
      "epoch": 1.039955398624791,
      "grad_norm": 0.5822799205780029,
      "learning_rate": 9.144143956713384e-05,
      "loss": 0.903,
      "step": 5689
    },
    {
      "epoch": 1.040141237688162,
      "grad_norm": 0.6802158355712891,
      "learning_rate": 9.14123013598645e-05,
      "loss": 0.9905,
      "step": 5690
    },
    {
      "epoch": 1.0403270767515331,
      "grad_norm": 0.6485140919685364,
      "learning_rate": 9.138316388711955e-05,
      "loss": 1.0835,
      "step": 5691
    },
    {
      "epoch": 1.0405129158149042,
      "grad_norm": 0.5066680312156677,
      "learning_rate": 9.135402715139125e-05,
      "loss": 0.7497,
      "step": 5692
    },
    {
      "epoch": 1.0406987548782753,
      "grad_norm": 0.7781651616096497,
      "learning_rate": 9.132489115517166e-05,
      "loss": 1.0902,
      "step": 5693
    },
    {
      "epoch": 1.0408845939416465,
      "grad_norm": 0.6752843260765076,
      "learning_rate": 9.12957559009529e-05,
      "loss": 0.837,
      "step": 5694
    },
    {
      "epoch": 1.0410704330050176,
      "grad_norm": 0.7520736455917358,
      "learning_rate": 9.126662139122694e-05,
      "loss": 1.125,
      "step": 5695
    },
    {
      "epoch": 1.0412562720683889,
      "grad_norm": 0.6787281036376953,
      "learning_rate": 9.123748762848571e-05,
      "loss": 1.1425,
      "step": 5696
    },
    {
      "epoch": 1.04144211113176,
      "grad_norm": 0.5874269604682922,
      "learning_rate": 9.120835461522115e-05,
      "loss": 1.002,
      "step": 5697
    },
    {
      "epoch": 1.041627950195131,
      "grad_norm": 0.6762674450874329,
      "learning_rate": 9.117922235392497e-05,
      "loss": 0.8577,
      "step": 5698
    },
    {
      "epoch": 1.0418137892585022,
      "grad_norm": 0.7589949369430542,
      "learning_rate": 9.115009084708903e-05,
      "loss": 1.3228,
      "step": 5699
    },
    {
      "epoch": 1.0419996283218733,
      "grad_norm": 0.5638782978057861,
      "learning_rate": 9.112096009720489e-05,
      "loss": 0.778,
      "step": 5700
    },
    {
      "epoch": 1.0421854673852444,
      "grad_norm": 0.6516910791397095,
      "learning_rate": 9.109183010676424e-05,
      "loss": 0.9716,
      "step": 5701
    },
    {
      "epoch": 1.0423713064486155,
      "grad_norm": 0.6196531653404236,
      "learning_rate": 9.10627008782586e-05,
      "loss": 0.8716,
      "step": 5702
    },
    {
      "epoch": 1.0425571455119866,
      "grad_norm": 0.5779221057891846,
      "learning_rate": 9.103357241417947e-05,
      "loss": 0.931,
      "step": 5703
    },
    {
      "epoch": 1.0427429845753577,
      "grad_norm": 0.7300266623497009,
      "learning_rate": 9.100444471701831e-05,
      "loss": 0.955,
      "step": 5704
    },
    {
      "epoch": 1.0429288236387289,
      "grad_norm": 0.5698464512825012,
      "learning_rate": 9.097531778926641e-05,
      "loss": 0.891,
      "step": 5705
    },
    {
      "epoch": 1.0431146627021,
      "grad_norm": 0.6704815030097961,
      "learning_rate": 9.09461916334151e-05,
      "loss": 0.8461,
      "step": 5706
    },
    {
      "epoch": 1.043300501765471,
      "grad_norm": 0.6242592334747314,
      "learning_rate": 9.091706625195555e-05,
      "loss": 0.8964,
      "step": 5707
    },
    {
      "epoch": 1.0434863408288422,
      "grad_norm": 0.6542682647705078,
      "learning_rate": 9.0887941647379e-05,
      "loss": 0.9589,
      "step": 5708
    },
    {
      "epoch": 1.0436721798922133,
      "grad_norm": 0.655167281627655,
      "learning_rate": 9.085881782217645e-05,
      "loss": 0.8934,
      "step": 5709
    },
    {
      "epoch": 1.0438580189555844,
      "grad_norm": 0.7744964957237244,
      "learning_rate": 9.082969477883899e-05,
      "loss": 0.9864,
      "step": 5710
    },
    {
      "epoch": 1.0440438580189555,
      "grad_norm": 0.6355542540550232,
      "learning_rate": 9.080057251985756e-05,
      "loss": 1.0679,
      "step": 5711
    },
    {
      "epoch": 1.0442296970823266,
      "grad_norm": 0.5906370878219604,
      "learning_rate": 9.077145104772302e-05,
      "loss": 0.9372,
      "step": 5712
    },
    {
      "epoch": 1.044415536145698,
      "grad_norm": 0.6469417214393616,
      "learning_rate": 9.074233036492626e-05,
      "loss": 1.0032,
      "step": 5713
    },
    {
      "epoch": 1.044601375209069,
      "grad_norm": 0.771095871925354,
      "learning_rate": 9.071321047395796e-05,
      "loss": 1.0268,
      "step": 5714
    },
    {
      "epoch": 1.0447872142724401,
      "grad_norm": 0.796539843082428,
      "learning_rate": 9.068409137730885e-05,
      "loss": 1.1293,
      "step": 5715
    },
    {
      "epoch": 1.0449730533358113,
      "grad_norm": 0.6017162799835205,
      "learning_rate": 9.065497307746956e-05,
      "loss": 0.9915,
      "step": 5716
    },
    {
      "epoch": 1.0451588923991824,
      "grad_norm": 0.5909117460250854,
      "learning_rate": 9.062585557693057e-05,
      "loss": 0.8477,
      "step": 5717
    },
    {
      "epoch": 1.0453447314625535,
      "grad_norm": 0.5846933722496033,
      "learning_rate": 9.059673887818245e-05,
      "loss": 0.7227,
      "step": 5718
    },
    {
      "epoch": 1.0455305705259246,
      "grad_norm": 0.5901908278465271,
      "learning_rate": 9.056762298371555e-05,
      "loss": 0.8948,
      "step": 5719
    },
    {
      "epoch": 1.0457164095892957,
      "grad_norm": 0.6081728339195251,
      "learning_rate": 9.053850789602029e-05,
      "loss": 1.0266,
      "step": 5720
    },
    {
      "epoch": 1.0459022486526668,
      "grad_norm": 0.6434999704360962,
      "learning_rate": 9.050939361758685e-05,
      "loss": 0.89,
      "step": 5721
    },
    {
      "epoch": 1.046088087716038,
      "grad_norm": 0.6816750168800354,
      "learning_rate": 9.048028015090553e-05,
      "loss": 0.9765,
      "step": 5722
    },
    {
      "epoch": 1.046273926779409,
      "grad_norm": 0.6385290026664734,
      "learning_rate": 9.045116749846638e-05,
      "loss": 0.8714,
      "step": 5723
    },
    {
      "epoch": 1.04645976584278,
      "grad_norm": 0.7308647036552429,
      "learning_rate": 9.042205566275951e-05,
      "loss": 0.9376,
      "step": 5724
    },
    {
      "epoch": 1.0466456049061512,
      "grad_norm": 0.6339771747589111,
      "learning_rate": 9.039294464627495e-05,
      "loss": 0.9761,
      "step": 5725
    },
    {
      "epoch": 1.0468314439695223,
      "grad_norm": 0.6732003688812256,
      "learning_rate": 9.036383445150255e-05,
      "loss": 0.9374,
      "step": 5726
    },
    {
      "epoch": 1.0470172830328934,
      "grad_norm": 0.6548773050308228,
      "learning_rate": 9.033472508093229e-05,
      "loss": 1.0606,
      "step": 5727
    },
    {
      "epoch": 1.0472031220962645,
      "grad_norm": 0.8904339671134949,
      "learning_rate": 9.030561653705384e-05,
      "loss": 1.0021,
      "step": 5728
    },
    {
      "epoch": 1.0473889611596356,
      "grad_norm": 0.6712539792060852,
      "learning_rate": 9.027650882235698e-05,
      "loss": 1.0839,
      "step": 5729
    },
    {
      "epoch": 1.047574800223007,
      "grad_norm": 0.6670980453491211,
      "learning_rate": 9.024740193933131e-05,
      "loss": 0.7674,
      "step": 5730
    },
    {
      "epoch": 1.047760639286378,
      "grad_norm": 0.5966160297393799,
      "learning_rate": 9.021829589046643e-05,
      "loss": 0.9512,
      "step": 5731
    },
    {
      "epoch": 1.0479464783497492,
      "grad_norm": 0.7879358530044556,
      "learning_rate": 9.018919067825186e-05,
      "loss": 1.1443,
      "step": 5732
    },
    {
      "epoch": 1.0481323174131203,
      "grad_norm": 0.5953929424285889,
      "learning_rate": 9.0160086305177e-05,
      "loss": 0.9683,
      "step": 5733
    },
    {
      "epoch": 1.0483181564764914,
      "grad_norm": 0.6400251388549805,
      "learning_rate": 9.013098277373127e-05,
      "loss": 0.8876,
      "step": 5734
    },
    {
      "epoch": 1.0485039955398625,
      "grad_norm": 1.7397582530975342,
      "learning_rate": 9.010188008640386e-05,
      "loss": 1.4435,
      "step": 5735
    },
    {
      "epoch": 1.0486898346032336,
      "grad_norm": 0.6329631209373474,
      "learning_rate": 9.007277824568407e-05,
      "loss": 0.9844,
      "step": 5736
    },
    {
      "epoch": 1.0488756736666047,
      "grad_norm": 0.7148013710975647,
      "learning_rate": 9.004367725406098e-05,
      "loss": 0.9352,
      "step": 5737
    },
    {
      "epoch": 1.0490615127299758,
      "grad_norm": 0.644572377204895,
      "learning_rate": 9.001457711402371e-05,
      "loss": 0.8407,
      "step": 5738
    },
    {
      "epoch": 1.049247351793347,
      "grad_norm": 0.585382342338562,
      "learning_rate": 8.998547782806124e-05,
      "loss": 0.7984,
      "step": 5739
    },
    {
      "epoch": 1.049433190856718,
      "grad_norm": 0.6851204037666321,
      "learning_rate": 8.995637939866249e-05,
      "loss": 0.9389,
      "step": 5740
    },
    {
      "epoch": 1.0496190299200892,
      "grad_norm": 0.6227951049804688,
      "learning_rate": 8.992728182831636e-05,
      "loss": 1.0817,
      "step": 5741
    },
    {
      "epoch": 1.0498048689834603,
      "grad_norm": 0.4151616394519806,
      "learning_rate": 8.989818511951153e-05,
      "loss": 0.3853,
      "step": 5742
    },
    {
      "epoch": 1.0499907080468314,
      "grad_norm": 0.7411801218986511,
      "learning_rate": 8.98690892747368e-05,
      "loss": 1.0406,
      "step": 5743
    },
    {
      "epoch": 1.0501765471102025,
      "grad_norm": 0.6021711826324463,
      "learning_rate": 8.983999429648074e-05,
      "loss": 0.5269,
      "step": 5744
    },
    {
      "epoch": 1.0503623861735736,
      "grad_norm": 0.6130529642105103,
      "learning_rate": 8.981090018723191e-05,
      "loss": 0.9034,
      "step": 5745
    },
    {
      "epoch": 1.050548225236945,
      "grad_norm": 0.605331540107727,
      "learning_rate": 8.978180694947884e-05,
      "loss": 0.7755,
      "step": 5746
    },
    {
      "epoch": 1.050734064300316,
      "grad_norm": 0.6502954959869385,
      "learning_rate": 8.975271458570986e-05,
      "loss": 0.9871,
      "step": 5747
    },
    {
      "epoch": 1.0509199033636871,
      "grad_norm": 0.6627545356750488,
      "learning_rate": 8.972362309841341e-05,
      "loss": 0.9986,
      "step": 5748
    },
    {
      "epoch": 1.0511057424270582,
      "grad_norm": 1.8269603252410889,
      "learning_rate": 8.969453249007764e-05,
      "loss": 1.5947,
      "step": 5749
    },
    {
      "epoch": 1.0512915814904293,
      "grad_norm": 0.6196630597114563,
      "learning_rate": 8.966544276319083e-05,
      "loss": 0.9685,
      "step": 5750
    },
    {
      "epoch": 1.0514774205538004,
      "grad_norm": 0.7198757529258728,
      "learning_rate": 8.9636353920241e-05,
      "loss": 0.9335,
      "step": 5751
    },
    {
      "epoch": 1.0516632596171716,
      "grad_norm": 0.6331172585487366,
      "learning_rate": 8.960726596371619e-05,
      "loss": 0.934,
      "step": 5752
    },
    {
      "epoch": 1.0518490986805427,
      "grad_norm": 0.5823332667350769,
      "learning_rate": 8.957817889610442e-05,
      "loss": 1.1294,
      "step": 5753
    },
    {
      "epoch": 1.0520349377439138,
      "grad_norm": 0.7582717537879944,
      "learning_rate": 8.954909271989351e-05,
      "loss": 1.0328,
      "step": 5754
    },
    {
      "epoch": 1.0522207768072849,
      "grad_norm": 0.598598301410675,
      "learning_rate": 8.952000743757133e-05,
      "loss": 0.7204,
      "step": 5755
    },
    {
      "epoch": 1.052406615870656,
      "grad_norm": 0.720652163028717,
      "learning_rate": 8.949092305162553e-05,
      "loss": 0.9257,
      "step": 5756
    },
    {
      "epoch": 1.052592454934027,
      "grad_norm": 0.610378623008728,
      "learning_rate": 8.946183956454384e-05,
      "loss": 0.9195,
      "step": 5757
    },
    {
      "epoch": 1.0527782939973982,
      "grad_norm": 0.6636180281639099,
      "learning_rate": 8.943275697881373e-05,
      "loss": 1.0782,
      "step": 5758
    },
    {
      "epoch": 1.0529641330607693,
      "grad_norm": 0.7740611433982849,
      "learning_rate": 8.940367529692275e-05,
      "loss": 0.954,
      "step": 5759
    },
    {
      "epoch": 1.0531499721241404,
      "grad_norm": 0.5274054408073425,
      "learning_rate": 8.937459452135837e-05,
      "loss": 0.6807,
      "step": 5760
    },
    {
      "epoch": 1.0533358111875115,
      "grad_norm": 0.659879744052887,
      "learning_rate": 8.934551465460786e-05,
      "loss": 1.052,
      "step": 5761
    },
    {
      "epoch": 1.0535216502508828,
      "grad_norm": 0.667706310749054,
      "learning_rate": 8.931643569915852e-05,
      "loss": 1.0115,
      "step": 5762
    },
    {
      "epoch": 1.053707489314254,
      "grad_norm": 0.6441750526428223,
      "learning_rate": 8.92873576574975e-05,
      "loss": 1.1946,
      "step": 5763
    },
    {
      "epoch": 1.053893328377625,
      "grad_norm": 0.5615370869636536,
      "learning_rate": 8.925828053211198e-05,
      "loss": 0.829,
      "step": 5764
    },
    {
      "epoch": 1.0540791674409962,
      "grad_norm": 0.5814454555511475,
      "learning_rate": 8.92292043254889e-05,
      "loss": 0.8492,
      "step": 5765
    },
    {
      "epoch": 1.0542650065043673,
      "grad_norm": 0.7165383100509644,
      "learning_rate": 8.920012904011525e-05,
      "loss": 0.976,
      "step": 5766
    },
    {
      "epoch": 1.0544508455677384,
      "grad_norm": 0.6098819971084595,
      "learning_rate": 8.917105467847795e-05,
      "loss": 0.826,
      "step": 5767
    },
    {
      "epoch": 1.0546366846311095,
      "grad_norm": 0.6676499843597412,
      "learning_rate": 8.914198124306371e-05,
      "loss": 0.8987,
      "step": 5768
    },
    {
      "epoch": 1.0548225236944806,
      "grad_norm": 0.679675281047821,
      "learning_rate": 8.911290873635928e-05,
      "loss": 1.0335,
      "step": 5769
    },
    {
      "epoch": 1.0550083627578517,
      "grad_norm": 0.6107022166252136,
      "learning_rate": 8.90838371608513e-05,
      "loss": 0.9001,
      "step": 5770
    },
    {
      "epoch": 1.0551942018212228,
      "grad_norm": 0.6810942888259888,
      "learning_rate": 8.905476651902636e-05,
      "loss": 1.0118,
      "step": 5771
    },
    {
      "epoch": 1.055380040884594,
      "grad_norm": 0.6379483342170715,
      "learning_rate": 8.902569681337083e-05,
      "loss": 1.0096,
      "step": 5772
    },
    {
      "epoch": 1.055565879947965,
      "grad_norm": 0.842753529548645,
      "learning_rate": 8.89966280463712e-05,
      "loss": 0.9862,
      "step": 5773
    },
    {
      "epoch": 1.0557517190113361,
      "grad_norm": 0.7353532910346985,
      "learning_rate": 8.896756022051378e-05,
      "loss": 1.2512,
      "step": 5774
    },
    {
      "epoch": 1.0559375580747072,
      "grad_norm": 0.6169674396514893,
      "learning_rate": 8.893849333828474e-05,
      "loss": 0.9022,
      "step": 5775
    },
    {
      "epoch": 1.0561233971380783,
      "grad_norm": 0.5542333126068115,
      "learning_rate": 8.89094274021703e-05,
      "loss": 0.9013,
      "step": 5776
    },
    {
      "epoch": 1.0563092362014495,
      "grad_norm": 2.630976915359497,
      "learning_rate": 8.888036241465647e-05,
      "loss": 1.7921,
      "step": 5777
    },
    {
      "epoch": 1.0564950752648206,
      "grad_norm": 0.6785982251167297,
      "learning_rate": 8.885129837822933e-05,
      "loss": 1.0074,
      "step": 5778
    },
    {
      "epoch": 1.056680914328192,
      "grad_norm": 0.6288561820983887,
      "learning_rate": 8.882223529537468e-05,
      "loss": 0.9505,
      "step": 5779
    },
    {
      "epoch": 1.056866753391563,
      "grad_norm": 1.0694950819015503,
      "learning_rate": 8.879317316857841e-05,
      "loss": 1.1911,
      "step": 5780
    },
    {
      "epoch": 1.057052592454934,
      "grad_norm": 0.6061035990715027,
      "learning_rate": 8.876411200032631e-05,
      "loss": 0.9956,
      "step": 5781
    },
    {
      "epoch": 1.0572384315183052,
      "grad_norm": 0.6882989406585693,
      "learning_rate": 8.873505179310394e-05,
      "loss": 0.9454,
      "step": 5782
    },
    {
      "epoch": 1.0574242705816763,
      "grad_norm": 0.6553504467010498,
      "learning_rate": 8.870599254939696e-05,
      "loss": 0.9591,
      "step": 5783
    },
    {
      "epoch": 1.0576101096450474,
      "grad_norm": 0.7941915392875671,
      "learning_rate": 8.867693427169084e-05,
      "loss": 0.9654,
      "step": 5784
    },
    {
      "epoch": 1.0577959487084185,
      "grad_norm": 0.6306182742118835,
      "learning_rate": 8.864787696247103e-05,
      "loss": 0.896,
      "step": 5785
    },
    {
      "epoch": 1.0579817877717896,
      "grad_norm": 0.8002976775169373,
      "learning_rate": 8.86188206242228e-05,
      "loss": 1.2667,
      "step": 5786
    },
    {
      "epoch": 1.0581676268351607,
      "grad_norm": 0.6710218191146851,
      "learning_rate": 8.858976525943144e-05,
      "loss": 0.9895,
      "step": 5787
    },
    {
      "epoch": 1.0583534658985319,
      "grad_norm": 1.6976367235183716,
      "learning_rate": 8.856071087058216e-05,
      "loss": 1.5448,
      "step": 5788
    },
    {
      "epoch": 1.058539304961903,
      "grad_norm": 0.581591784954071,
      "learning_rate": 8.853165746015997e-05,
      "loss": 0.8481,
      "step": 5789
    },
    {
      "epoch": 1.058725144025274,
      "grad_norm": 0.7091907262802124,
      "learning_rate": 8.850260503064992e-05,
      "loss": 0.9044,
      "step": 5790
    },
    {
      "epoch": 1.0589109830886452,
      "grad_norm": 0.6425524353981018,
      "learning_rate": 8.847355358453687e-05,
      "loss": 0.9289,
      "step": 5791
    },
    {
      "epoch": 1.0590968221520163,
      "grad_norm": 0.5482159852981567,
      "learning_rate": 8.844450312430576e-05,
      "loss": 0.9106,
      "step": 5792
    },
    {
      "epoch": 1.0592826612153874,
      "grad_norm": 0.5962523221969604,
      "learning_rate": 8.841545365244123e-05,
      "loss": 0.9134,
      "step": 5793
    },
    {
      "epoch": 1.0594685002787585,
      "grad_norm": 0.6635329723358154,
      "learning_rate": 8.838640517142797e-05,
      "loss": 0.8846,
      "step": 5794
    },
    {
      "epoch": 1.0596543393421296,
      "grad_norm": 0.6011704206466675,
      "learning_rate": 8.835735768375063e-05,
      "loss": 0.8404,
      "step": 5795
    },
    {
      "epoch": 1.059840178405501,
      "grad_norm": 0.6340507864952087,
      "learning_rate": 8.832831119189361e-05,
      "loss": 0.8327,
      "step": 5796
    },
    {
      "epoch": 1.060026017468872,
      "grad_norm": 0.6549215316772461,
      "learning_rate": 8.829926569834137e-05,
      "loss": 1.1531,
      "step": 5797
    },
    {
      "epoch": 1.0602118565322431,
      "grad_norm": 0.7567423582077026,
      "learning_rate": 8.827022120557822e-05,
      "loss": 1.0995,
      "step": 5798
    },
    {
      "epoch": 1.0603976955956143,
      "grad_norm": 0.6789305210113525,
      "learning_rate": 8.824117771608842e-05,
      "loss": 1.0125,
      "step": 5799
    },
    {
      "epoch": 1.0605835346589854,
      "grad_norm": 0.8210663795471191,
      "learning_rate": 8.821213523235608e-05,
      "loss": 1.1365,
      "step": 5800
    },
    {
      "epoch": 1.0607693737223565,
      "grad_norm": 0.752582848072052,
      "learning_rate": 8.818309375686529e-05,
      "loss": 0.9847,
      "step": 5801
    },
    {
      "epoch": 1.0609552127857276,
      "grad_norm": 0.561202347278595,
      "learning_rate": 8.815405329210007e-05,
      "loss": 0.7168,
      "step": 5802
    },
    {
      "epoch": 1.0611410518490987,
      "grad_norm": 0.5668363571166992,
      "learning_rate": 8.812501384054422e-05,
      "loss": 0.9466,
      "step": 5803
    },
    {
      "epoch": 1.0613268909124698,
      "grad_norm": 0.7196042537689209,
      "learning_rate": 8.809597540468164e-05,
      "loss": 0.9273,
      "step": 5804
    },
    {
      "epoch": 1.061512729975841,
      "grad_norm": 0.6628761887550354,
      "learning_rate": 8.806693798699598e-05,
      "loss": 1.0514,
      "step": 5805
    },
    {
      "epoch": 1.061698569039212,
      "grad_norm": 0.6018509864807129,
      "learning_rate": 8.803790158997095e-05,
      "loss": 1.0525,
      "step": 5806
    },
    {
      "epoch": 1.061884408102583,
      "grad_norm": 0.7145560383796692,
      "learning_rate": 8.800886621609001e-05,
      "loss": 1.0421,
      "step": 5807
    },
    {
      "epoch": 1.0620702471659542,
      "grad_norm": 0.5981913805007935,
      "learning_rate": 8.797983186783666e-05,
      "loss": 0.793,
      "step": 5808
    },
    {
      "epoch": 1.0622560862293253,
      "grad_norm": 0.6121430993080139,
      "learning_rate": 8.79507985476943e-05,
      "loss": 0.8296,
      "step": 5809
    },
    {
      "epoch": 1.0624419252926964,
      "grad_norm": 0.6061347723007202,
      "learning_rate": 8.792176625814615e-05,
      "loss": 0.924,
      "step": 5810
    },
    {
      "epoch": 1.0626277643560678,
      "grad_norm": 0.6806914806365967,
      "learning_rate": 8.789273500167547e-05,
      "loss": 1.1121,
      "step": 5811
    },
    {
      "epoch": 1.0628136034194389,
      "grad_norm": 0.6414180397987366,
      "learning_rate": 8.78637047807653e-05,
      "loss": 0.9736,
      "step": 5812
    },
    {
      "epoch": 1.06299944248281,
      "grad_norm": 0.7345332503318787,
      "learning_rate": 8.783467559789872e-05,
      "loss": 0.9447,
      "step": 5813
    },
    {
      "epoch": 1.063185281546181,
      "grad_norm": 0.6870766282081604,
      "learning_rate": 8.78056474555586e-05,
      "loss": 1.0822,
      "step": 5814
    },
    {
      "epoch": 1.0633711206095522,
      "grad_norm": 0.6061105132102966,
      "learning_rate": 8.777662035622781e-05,
      "loss": 0.7237,
      "step": 5815
    },
    {
      "epoch": 1.0635569596729233,
      "grad_norm": 0.5764292478561401,
      "learning_rate": 8.774759430238912e-05,
      "loss": 0.9691,
      "step": 5816
    },
    {
      "epoch": 1.0637427987362944,
      "grad_norm": 0.6762557029724121,
      "learning_rate": 8.771856929652515e-05,
      "loss": 1.107,
      "step": 5817
    },
    {
      "epoch": 1.0639286377996655,
      "grad_norm": 0.7746602296829224,
      "learning_rate": 8.768954534111853e-05,
      "loss": 0.8857,
      "step": 5818
    },
    {
      "epoch": 1.0641144768630366,
      "grad_norm": 0.5895851850509644,
      "learning_rate": 8.766052243865166e-05,
      "loss": 0.9329,
      "step": 5819
    },
    {
      "epoch": 1.0643003159264077,
      "grad_norm": 0.6938430666923523,
      "learning_rate": 8.7631500591607e-05,
      "loss": 0.9406,
      "step": 5820
    },
    {
      "epoch": 1.0644861549897788,
      "grad_norm": 0.6056532263755798,
      "learning_rate": 8.760247980246679e-05,
      "loss": 0.8533,
      "step": 5821
    },
    {
      "epoch": 1.06467199405315,
      "grad_norm": 0.6963773965835571,
      "learning_rate": 8.75734600737133e-05,
      "loss": 1.1039,
      "step": 5822
    },
    {
      "epoch": 1.064857833116521,
      "grad_norm": 1.3031995296478271,
      "learning_rate": 8.754444140782866e-05,
      "loss": 1.4601,
      "step": 5823
    },
    {
      "epoch": 1.0650436721798922,
      "grad_norm": 0.6234591603279114,
      "learning_rate": 8.751542380729482e-05,
      "loss": 0.9837,
      "step": 5824
    },
    {
      "epoch": 1.0652295112432633,
      "grad_norm": 0.6526420712471008,
      "learning_rate": 8.74864072745938e-05,
      "loss": 0.7511,
      "step": 5825
    },
    {
      "epoch": 1.0654153503066344,
      "grad_norm": 0.651283323764801,
      "learning_rate": 8.745739181220738e-05,
      "loss": 1.0071,
      "step": 5826
    },
    {
      "epoch": 1.0656011893700055,
      "grad_norm": 0.5934510231018066,
      "learning_rate": 8.742837742261737e-05,
      "loss": 0.9859,
      "step": 5827
    },
    {
      "epoch": 1.0657870284333768,
      "grad_norm": 0.5294697880744934,
      "learning_rate": 8.739936410830538e-05,
      "loss": 0.8086,
      "step": 5828
    },
    {
      "epoch": 1.065972867496748,
      "grad_norm": 0.6745396256446838,
      "learning_rate": 8.737035187175302e-05,
      "loss": 1.0394,
      "step": 5829
    },
    {
      "epoch": 1.066158706560119,
      "grad_norm": 0.6572603583335876,
      "learning_rate": 8.73413407154418e-05,
      "loss": 0.7895,
      "step": 5830
    },
    {
      "epoch": 1.0663445456234901,
      "grad_norm": 0.7055622339248657,
      "learning_rate": 8.731233064185302e-05,
      "loss": 1.028,
      "step": 5831
    },
    {
      "epoch": 1.0665303846868612,
      "grad_norm": 0.6637229323387146,
      "learning_rate": 8.728332165346806e-05,
      "loss": 1.1012,
      "step": 5832
    },
    {
      "epoch": 1.0667162237502323,
      "grad_norm": 0.5706199407577515,
      "learning_rate": 8.725431375276804e-05,
      "loss": 0.9943,
      "step": 5833
    },
    {
      "epoch": 1.0669020628136034,
      "grad_norm": 0.6235327124595642,
      "learning_rate": 8.722530694223415e-05,
      "loss": 0.9364,
      "step": 5834
    },
    {
      "epoch": 1.0670879018769746,
      "grad_norm": 0.5939970016479492,
      "learning_rate": 8.719630122434734e-05,
      "loss": 0.8755,
      "step": 5835
    },
    {
      "epoch": 1.0672737409403457,
      "grad_norm": 0.6441448330879211,
      "learning_rate": 8.716729660158856e-05,
      "loss": 0.865,
      "step": 5836
    },
    {
      "epoch": 1.0674595800037168,
      "grad_norm": 0.5464093685150146,
      "learning_rate": 8.713829307643867e-05,
      "loss": 0.8899,
      "step": 5837
    },
    {
      "epoch": 1.0676454190670879,
      "grad_norm": 0.6502255797386169,
      "learning_rate": 8.710929065137834e-05,
      "loss": 1.046,
      "step": 5838
    },
    {
      "epoch": 1.067831258130459,
      "grad_norm": 0.6012847423553467,
      "learning_rate": 8.708028932888828e-05,
      "loss": 0.7964,
      "step": 5839
    },
    {
      "epoch": 1.06801709719383,
      "grad_norm": 0.64566969871521,
      "learning_rate": 8.705128911144894e-05,
      "loss": 1.147,
      "step": 5840
    },
    {
      "epoch": 1.0682029362572012,
      "grad_norm": 0.73211669921875,
      "learning_rate": 8.702229000154087e-05,
      "loss": 1.2799,
      "step": 5841
    },
    {
      "epoch": 1.0683887753205723,
      "grad_norm": 0.5920413136482239,
      "learning_rate": 8.699329200164434e-05,
      "loss": 0.9163,
      "step": 5842
    },
    {
      "epoch": 1.0685746143839434,
      "grad_norm": 0.6712371706962585,
      "learning_rate": 8.696429511423968e-05,
      "loss": 1.1072,
      "step": 5843
    },
    {
      "epoch": 1.0687604534473145,
      "grad_norm": 0.6476579308509827,
      "learning_rate": 8.693529934180706e-05,
      "loss": 0.9671,
      "step": 5844
    },
    {
      "epoch": 1.0689462925106858,
      "grad_norm": 0.6458027362823486,
      "learning_rate": 8.69063046868265e-05,
      "loss": 0.9178,
      "step": 5845
    },
    {
      "epoch": 1.069132131574057,
      "grad_norm": 0.6158638000488281,
      "learning_rate": 8.687731115177802e-05,
      "loss": 0.9185,
      "step": 5846
    },
    {
      "epoch": 1.069317970637428,
      "grad_norm": 0.6359545588493347,
      "learning_rate": 8.684831873914145e-05,
      "loss": 0.913,
      "step": 5847
    },
    {
      "epoch": 1.0695038097007992,
      "grad_norm": 0.5854082107543945,
      "learning_rate": 8.681932745139662e-05,
      "loss": 0.6984,
      "step": 5848
    },
    {
      "epoch": 1.0696896487641703,
      "grad_norm": 0.5653736591339111,
      "learning_rate": 8.679033729102317e-05,
      "loss": 0.8022,
      "step": 5849
    },
    {
      "epoch": 1.0698754878275414,
      "grad_norm": 0.7520171403884888,
      "learning_rate": 8.676134826050072e-05,
      "loss": 1.1837,
      "step": 5850
    },
    {
      "epoch": 1.0700613268909125,
      "grad_norm": 0.6621032953262329,
      "learning_rate": 8.673236036230882e-05,
      "loss": 0.971,
      "step": 5851
    },
    {
      "epoch": 1.0702471659542836,
      "grad_norm": 0.6689890027046204,
      "learning_rate": 8.670337359892678e-05,
      "loss": 1.07,
      "step": 5852
    },
    {
      "epoch": 1.0704330050176547,
      "grad_norm": 0.6319896578788757,
      "learning_rate": 8.667438797283397e-05,
      "loss": 1.1009,
      "step": 5853
    },
    {
      "epoch": 1.0706188440810258,
      "grad_norm": 0.843813419342041,
      "learning_rate": 8.664540348650951e-05,
      "loss": 0.9687,
      "step": 5854
    },
    {
      "epoch": 1.070804683144397,
      "grad_norm": 0.5611274838447571,
      "learning_rate": 8.661642014243258e-05,
      "loss": 0.7719,
      "step": 5855
    },
    {
      "epoch": 1.070990522207768,
      "grad_norm": 0.6595000624656677,
      "learning_rate": 8.658743794308212e-05,
      "loss": 0.7983,
      "step": 5856
    },
    {
      "epoch": 1.0711763612711391,
      "grad_norm": 0.7374562621116638,
      "learning_rate": 8.655845689093712e-05,
      "loss": 0.869,
      "step": 5857
    },
    {
      "epoch": 1.0713622003345102,
      "grad_norm": 0.7460797429084778,
      "learning_rate": 8.652947698847636e-05,
      "loss": 0.9378,
      "step": 5858
    },
    {
      "epoch": 1.0715480393978813,
      "grad_norm": 0.6679945588111877,
      "learning_rate": 8.650049823817851e-05,
      "loss": 0.9014,
      "step": 5859
    },
    {
      "epoch": 1.0717338784612525,
      "grad_norm": 0.7798948884010315,
      "learning_rate": 8.647152064252226e-05,
      "loss": 1.0417,
      "step": 5860
    },
    {
      "epoch": 1.0719197175246236,
      "grad_norm": 0.7319615483283997,
      "learning_rate": 8.644254420398607e-05,
      "loss": 0.894,
      "step": 5861
    },
    {
      "epoch": 1.072105556587995,
      "grad_norm": 0.7500199675559998,
      "learning_rate": 8.641356892504838e-05,
      "loss": 1.1208,
      "step": 5862
    },
    {
      "epoch": 1.072291395651366,
      "grad_norm": 0.7801706194877625,
      "learning_rate": 8.63845948081875e-05,
      "loss": 1.1542,
      "step": 5863
    },
    {
      "epoch": 1.072477234714737,
      "grad_norm": 0.6065161228179932,
      "learning_rate": 8.635562185588164e-05,
      "loss": 0.8436,
      "step": 5864
    },
    {
      "epoch": 1.0726630737781082,
      "grad_norm": 0.7084574103355408,
      "learning_rate": 8.632665007060896e-05,
      "loss": 1.0358,
      "step": 5865
    },
    {
      "epoch": 1.0728489128414793,
      "grad_norm": 0.5427306294441223,
      "learning_rate": 8.629767945484741e-05,
      "loss": 0.6025,
      "step": 5866
    },
    {
      "epoch": 1.0730347519048504,
      "grad_norm": 0.5843170285224915,
      "learning_rate": 8.626871001107501e-05,
      "loss": 0.9039,
      "step": 5867
    },
    {
      "epoch": 1.0732205909682215,
      "grad_norm": 0.6646682024002075,
      "learning_rate": 8.623974174176947e-05,
      "loss": 0.7304,
      "step": 5868
    },
    {
      "epoch": 1.0734064300315926,
      "grad_norm": 0.4983295500278473,
      "learning_rate": 8.62107746494086e-05,
      "loss": 0.6743,
      "step": 5869
    },
    {
      "epoch": 1.0735922690949637,
      "grad_norm": 0.8558180332183838,
      "learning_rate": 8.618180873646994e-05,
      "loss": 1.0591,
      "step": 5870
    },
    {
      "epoch": 1.0737781081583349,
      "grad_norm": 0.6745973229408264,
      "learning_rate": 8.615284400543104e-05,
      "loss": 0.9655,
      "step": 5871
    },
    {
      "epoch": 1.073963947221706,
      "grad_norm": 0.6001264452934265,
      "learning_rate": 8.612388045876934e-05,
      "loss": 1.064,
      "step": 5872
    },
    {
      "epoch": 1.074149786285077,
      "grad_norm": 0.6128737330436707,
      "learning_rate": 8.609491809896212e-05,
      "loss": 0.8209,
      "step": 5873
    },
    {
      "epoch": 1.0743356253484482,
      "grad_norm": 0.7345498204231262,
      "learning_rate": 8.606595692848664e-05,
      "loss": 0.9531,
      "step": 5874
    },
    {
      "epoch": 1.0745214644118193,
      "grad_norm": 0.6948199272155762,
      "learning_rate": 8.603699694981995e-05,
      "loss": 0.942,
      "step": 5875
    },
    {
      "epoch": 1.0747073034751904,
      "grad_norm": 0.568913996219635,
      "learning_rate": 8.600803816543912e-05,
      "loss": 1.0326,
      "step": 5876
    },
    {
      "epoch": 1.0748931425385617,
      "grad_norm": 0.6175299882888794,
      "learning_rate": 8.5979080577821e-05,
      "loss": 1.0009,
      "step": 5877
    },
    {
      "epoch": 1.0750789816019328,
      "grad_norm": 2.101738452911377,
      "learning_rate": 8.59501241894424e-05,
      "loss": 1.8329,
      "step": 5878
    },
    {
      "epoch": 1.075264820665304,
      "grad_norm": 0.6964477896690369,
      "learning_rate": 8.592116900278008e-05,
      "loss": 0.9948,
      "step": 5879
    },
    {
      "epoch": 1.075450659728675,
      "grad_norm": 0.6877242922782898,
      "learning_rate": 8.589221502031057e-05,
      "loss": 0.8998,
      "step": 5880
    },
    {
      "epoch": 1.0756364987920461,
      "grad_norm": 0.7231934666633606,
      "learning_rate": 8.586326224451045e-05,
      "loss": 0.8386,
      "step": 5881
    },
    {
      "epoch": 1.0758223378554173,
      "grad_norm": 0.6604142189025879,
      "learning_rate": 8.583431067785602e-05,
      "loss": 0.9283,
      "step": 5882
    },
    {
      "epoch": 1.0760081769187884,
      "grad_norm": 0.6713509559631348,
      "learning_rate": 8.580536032282366e-05,
      "loss": 1.0227,
      "step": 5883
    },
    {
      "epoch": 1.0761940159821595,
      "grad_norm": 0.5723233819007874,
      "learning_rate": 8.577641118188945e-05,
      "loss": 0.8204,
      "step": 5884
    },
    {
      "epoch": 1.0763798550455306,
      "grad_norm": 0.6917603611946106,
      "learning_rate": 8.574746325752956e-05,
      "loss": 1.1986,
      "step": 5885
    },
    {
      "epoch": 1.0765656941089017,
      "grad_norm": 0.6363825798034668,
      "learning_rate": 8.571851655221995e-05,
      "loss": 0.9761,
      "step": 5886
    },
    {
      "epoch": 1.0767515331722728,
      "grad_norm": 0.6058530807495117,
      "learning_rate": 8.568957106843648e-05,
      "loss": 0.997,
      "step": 5887
    },
    {
      "epoch": 1.076937372235644,
      "grad_norm": 0.7039767503738403,
      "learning_rate": 8.566062680865494e-05,
      "loss": 1.1381,
      "step": 5888
    },
    {
      "epoch": 1.077123211299015,
      "grad_norm": 0.7130062580108643,
      "learning_rate": 8.563168377535095e-05,
      "loss": 1.0521,
      "step": 5889
    },
    {
      "epoch": 1.0773090503623861,
      "grad_norm": 0.6804975867271423,
      "learning_rate": 8.560274197100014e-05,
      "loss": 0.9519,
      "step": 5890
    },
    {
      "epoch": 1.0774948894257572,
      "grad_norm": 0.6732782125473022,
      "learning_rate": 8.557380139807789e-05,
      "loss": 0.8481,
      "step": 5891
    },
    {
      "epoch": 1.0776807284891283,
      "grad_norm": 0.6793938279151917,
      "learning_rate": 8.554486205905959e-05,
      "loss": 0.8197,
      "step": 5892
    },
    {
      "epoch": 1.0778665675524994,
      "grad_norm": 0.6705108880996704,
      "learning_rate": 8.551592395642048e-05,
      "loss": 0.9954,
      "step": 5893
    },
    {
      "epoch": 1.0780524066158708,
      "grad_norm": 0.6985551714897156,
      "learning_rate": 8.54869870926357e-05,
      "loss": 0.9462,
      "step": 5894
    },
    {
      "epoch": 1.0782382456792419,
      "grad_norm": 0.6636293530464172,
      "learning_rate": 8.545805147018031e-05,
      "loss": 0.9449,
      "step": 5895
    },
    {
      "epoch": 1.078424084742613,
      "grad_norm": 0.7172894477844238,
      "learning_rate": 8.542911709152917e-05,
      "loss": 1.0889,
      "step": 5896
    },
    {
      "epoch": 1.078609923805984,
      "grad_norm": 0.7706103920936584,
      "learning_rate": 8.540018395915718e-05,
      "loss": 1.0417,
      "step": 5897
    },
    {
      "epoch": 1.0787957628693552,
      "grad_norm": 0.6476448774337769,
      "learning_rate": 8.537125207553897e-05,
      "loss": 0.999,
      "step": 5898
    },
    {
      "epoch": 1.0789816019327263,
      "grad_norm": 0.572585940361023,
      "learning_rate": 8.534232144314918e-05,
      "loss": 0.759,
      "step": 5899
    },
    {
      "epoch": 1.0791674409960974,
      "grad_norm": 0.630120038986206,
      "learning_rate": 8.531339206446234e-05,
      "loss": 0.8345,
      "step": 5900
    },
    {
      "epoch": 1.0793532800594685,
      "grad_norm": 0.5526821613311768,
      "learning_rate": 8.52844639419528e-05,
      "loss": 0.9357,
      "step": 5901
    },
    {
      "epoch": 1.0795391191228396,
      "grad_norm": 0.6959087252616882,
      "learning_rate": 8.525553707809491e-05,
      "loss": 0.8057,
      "step": 5902
    },
    {
      "epoch": 1.0797249581862107,
      "grad_norm": 0.75086510181427,
      "learning_rate": 8.522661147536274e-05,
      "loss": 1.0193,
      "step": 5903
    },
    {
      "epoch": 1.0799107972495818,
      "grad_norm": 0.6601825952529907,
      "learning_rate": 8.519768713623047e-05,
      "loss": 1.0793,
      "step": 5904
    },
    {
      "epoch": 1.080096636312953,
      "grad_norm": 0.5780887603759766,
      "learning_rate": 8.516876406317198e-05,
      "loss": 0.9708,
      "step": 5905
    },
    {
      "epoch": 1.080282475376324,
      "grad_norm": 0.6262563467025757,
      "learning_rate": 8.513984225866114e-05,
      "loss": 0.8962,
      "step": 5906
    },
    {
      "epoch": 1.0804683144396952,
      "grad_norm": 0.6281941533088684,
      "learning_rate": 8.511092172517175e-05,
      "loss": 0.9141,
      "step": 5907
    },
    {
      "epoch": 1.0806541535030663,
      "grad_norm": 0.8933135867118835,
      "learning_rate": 8.508200246517739e-05,
      "loss": 1.0007,
      "step": 5908
    },
    {
      "epoch": 1.0808399925664374,
      "grad_norm": 0.6440215706825256,
      "learning_rate": 8.50530844811516e-05,
      "loss": 1.1193,
      "step": 5909
    },
    {
      "epoch": 1.0810258316298085,
      "grad_norm": 0.618796706199646,
      "learning_rate": 8.502416777556778e-05,
      "loss": 0.8557,
      "step": 5910
    },
    {
      "epoch": 1.0812116706931798,
      "grad_norm": 0.7517762780189514,
      "learning_rate": 8.49952523508993e-05,
      "loss": 1.0686,
      "step": 5911
    },
    {
      "epoch": 1.081397509756551,
      "grad_norm": 0.6386401057243347,
      "learning_rate": 8.496633820961927e-05,
      "loss": 0.956,
      "step": 5912
    },
    {
      "epoch": 1.081583348819922,
      "grad_norm": 0.6620604991912842,
      "learning_rate": 8.493742535420083e-05,
      "loss": 0.8032,
      "step": 5913
    },
    {
      "epoch": 1.0817691878832931,
      "grad_norm": 0.6698114275932312,
      "learning_rate": 8.4908513787117e-05,
      "loss": 0.8991,
      "step": 5914
    },
    {
      "epoch": 1.0819550269466642,
      "grad_norm": 0.6888457536697388,
      "learning_rate": 8.487960351084054e-05,
      "loss": 0.8746,
      "step": 5915
    },
    {
      "epoch": 1.0821408660100353,
      "grad_norm": 0.6484079957008362,
      "learning_rate": 8.485069452784431e-05,
      "loss": 0.822,
      "step": 5916
    },
    {
      "epoch": 1.0823267050734064,
      "grad_norm": 0.743229329586029,
      "learning_rate": 8.48217868406009e-05,
      "loss": 0.8274,
      "step": 5917
    },
    {
      "epoch": 1.0825125441367776,
      "grad_norm": 0.6204001307487488,
      "learning_rate": 8.479288045158291e-05,
      "loss": 0.8803,
      "step": 5918
    },
    {
      "epoch": 1.0826983832001487,
      "grad_norm": 0.7072806358337402,
      "learning_rate": 8.476397536326266e-05,
      "loss": 1.1312,
      "step": 5919
    },
    {
      "epoch": 1.0828842222635198,
      "grad_norm": 0.6361244320869446,
      "learning_rate": 8.473507157811254e-05,
      "loss": 0.7503,
      "step": 5920
    },
    {
      "epoch": 1.0830700613268909,
      "grad_norm": 0.6259986758232117,
      "learning_rate": 8.470616909860479e-05,
      "loss": 1.0263,
      "step": 5921
    },
    {
      "epoch": 1.083255900390262,
      "grad_norm": 0.7605748176574707,
      "learning_rate": 8.46772679272114e-05,
      "loss": 0.9734,
      "step": 5922
    },
    {
      "epoch": 1.083441739453633,
      "grad_norm": 0.6772533059120178,
      "learning_rate": 8.464836806640443e-05,
      "loss": 0.9361,
      "step": 5923
    },
    {
      "epoch": 1.0836275785170042,
      "grad_norm": 0.5974377393722534,
      "learning_rate": 8.46194695186557e-05,
      "loss": 0.9405,
      "step": 5924
    },
    {
      "epoch": 1.0838134175803753,
      "grad_norm": 0.7039512991905212,
      "learning_rate": 8.4590572286437e-05,
      "loss": 1.1505,
      "step": 5925
    },
    {
      "epoch": 1.0839992566437464,
      "grad_norm": 0.6239109635353088,
      "learning_rate": 8.456167637221993e-05,
      "loss": 0.9038,
      "step": 5926
    },
    {
      "epoch": 1.0841850957071175,
      "grad_norm": 0.6761141419410706,
      "learning_rate": 8.453278177847606e-05,
      "loss": 0.8879,
      "step": 5927
    },
    {
      "epoch": 1.0843709347704888,
      "grad_norm": 0.580737829208374,
      "learning_rate": 8.45038885076768e-05,
      "loss": 1.1171,
      "step": 5928
    },
    {
      "epoch": 1.08455677383386,
      "grad_norm": 0.702734112739563,
      "learning_rate": 8.447499656229344e-05,
      "loss": 0.8489,
      "step": 5929
    },
    {
      "epoch": 1.084742612897231,
      "grad_norm": 0.664533257484436,
      "learning_rate": 8.444610594479718e-05,
      "loss": 0.9631,
      "step": 5930
    },
    {
      "epoch": 1.0849284519606022,
      "grad_norm": 0.887567400932312,
      "learning_rate": 8.441721665765909e-05,
      "loss": 0.9973,
      "step": 5931
    },
    {
      "epoch": 1.0851142910239733,
      "grad_norm": 0.6015354990959167,
      "learning_rate": 8.438832870335018e-05,
      "loss": 0.901,
      "step": 5932
    },
    {
      "epoch": 1.0853001300873444,
      "grad_norm": 0.7177890539169312,
      "learning_rate": 8.43594420843412e-05,
      "loss": 1.0315,
      "step": 5933
    },
    {
      "epoch": 1.0854859691507155,
      "grad_norm": 0.667842447757721,
      "learning_rate": 8.433055680310295e-05,
      "loss": 1.1921,
      "step": 5934
    },
    {
      "epoch": 1.0856718082140866,
      "grad_norm": 0.7782358527183533,
      "learning_rate": 8.430167286210608e-05,
      "loss": 0.9981,
      "step": 5935
    },
    {
      "epoch": 1.0858576472774577,
      "grad_norm": 0.6504238247871399,
      "learning_rate": 8.427279026382103e-05,
      "loss": 0.8409,
      "step": 5936
    },
    {
      "epoch": 1.0860434863408288,
      "grad_norm": 0.6968525648117065,
      "learning_rate": 8.424390901071822e-05,
      "loss": 0.9398,
      "step": 5937
    },
    {
      "epoch": 1.0862293254042,
      "grad_norm": 0.6084544658660889,
      "learning_rate": 8.421502910526791e-05,
      "loss": 1.0434,
      "step": 5938
    },
    {
      "epoch": 1.086415164467571,
      "grad_norm": 0.5817170143127441,
      "learning_rate": 8.418615054994034e-05,
      "loss": 0.8075,
      "step": 5939
    },
    {
      "epoch": 1.0866010035309421,
      "grad_norm": 0.708842396736145,
      "learning_rate": 8.415727334720543e-05,
      "loss": 1.01,
      "step": 5940
    },
    {
      "epoch": 1.0867868425943132,
      "grad_norm": 0.6967892050743103,
      "learning_rate": 8.412839749953316e-05,
      "loss": 1.1844,
      "step": 5941
    },
    {
      "epoch": 1.0869726816576843,
      "grad_norm": 0.5187710523605347,
      "learning_rate": 8.409952300939343e-05,
      "loss": 0.709,
      "step": 5942
    },
    {
      "epoch": 1.0871585207210557,
      "grad_norm": 0.7596275210380554,
      "learning_rate": 8.40706498792558e-05,
      "loss": 0.9373,
      "step": 5943
    },
    {
      "epoch": 1.0873443597844268,
      "grad_norm": 0.4423999786376953,
      "learning_rate": 8.404177811158994e-05,
      "loss": 0.4896,
      "step": 5944
    },
    {
      "epoch": 1.087530198847798,
      "grad_norm": 0.6767209768295288,
      "learning_rate": 8.401290770886525e-05,
      "loss": 1.0141,
      "step": 5945
    },
    {
      "epoch": 1.087716037911169,
      "grad_norm": 0.5728704333305359,
      "learning_rate": 8.398403867355118e-05,
      "loss": 0.7946,
      "step": 5946
    },
    {
      "epoch": 1.08790187697454,
      "grad_norm": 0.6678792238235474,
      "learning_rate": 8.395517100811685e-05,
      "loss": 0.8289,
      "step": 5947
    },
    {
      "epoch": 1.0880877160379112,
      "grad_norm": 0.6508834362030029,
      "learning_rate": 8.392630471503141e-05,
      "loss": 0.9429,
      "step": 5948
    },
    {
      "epoch": 1.0882735551012823,
      "grad_norm": 0.6657683253288269,
      "learning_rate": 8.389743979676392e-05,
      "loss": 0.8843,
      "step": 5949
    },
    {
      "epoch": 1.0884593941646534,
      "grad_norm": 0.5776364803314209,
      "learning_rate": 8.386857625578317e-05,
      "loss": 0.7137,
      "step": 5950
    },
    {
      "epoch": 1.0886452332280245,
      "grad_norm": 0.6650852560997009,
      "learning_rate": 8.383971409455798e-05,
      "loss": 0.896,
      "step": 5951
    },
    {
      "epoch": 1.0888310722913956,
      "grad_norm": 0.6167944073677063,
      "learning_rate": 8.381085331555693e-05,
      "loss": 0.8864,
      "step": 5952
    },
    {
      "epoch": 1.0890169113547667,
      "grad_norm": 0.7045409679412842,
      "learning_rate": 8.378199392124864e-05,
      "loss": 0.8547,
      "step": 5953
    },
    {
      "epoch": 1.0892027504181379,
      "grad_norm": 0.5888659954071045,
      "learning_rate": 8.375313591410141e-05,
      "loss": 0.8933,
      "step": 5954
    },
    {
      "epoch": 1.089388589481509,
      "grad_norm": 0.6576387882232666,
      "learning_rate": 8.37242792965836e-05,
      "loss": 0.7533,
      "step": 5955
    },
    {
      "epoch": 1.08957442854488,
      "grad_norm": 0.5648308992385864,
      "learning_rate": 8.369542407116338e-05,
      "loss": 0.9638,
      "step": 5956
    },
    {
      "epoch": 1.0897602676082512,
      "grad_norm": 0.6234437227249146,
      "learning_rate": 8.366657024030876e-05,
      "loss": 0.9858,
      "step": 5957
    },
    {
      "epoch": 1.0899461066716223,
      "grad_norm": 1.7195453643798828,
      "learning_rate": 8.363771780648768e-05,
      "loss": 1.3238,
      "step": 5958
    },
    {
      "epoch": 1.0901319457349934,
      "grad_norm": 0.6707358360290527,
      "learning_rate": 8.360886677216797e-05,
      "loss": 0.7908,
      "step": 5959
    },
    {
      "epoch": 1.0903177847983647,
      "grad_norm": 0.657151997089386,
      "learning_rate": 8.358001713981731e-05,
      "loss": 1.1087,
      "step": 5960
    },
    {
      "epoch": 1.0905036238617358,
      "grad_norm": 0.7293577790260315,
      "learning_rate": 8.355116891190325e-05,
      "loss": 0.8512,
      "step": 5961
    },
    {
      "epoch": 1.090689462925107,
      "grad_norm": 0.5471261739730835,
      "learning_rate": 8.352232209089328e-05,
      "loss": 0.6389,
      "step": 5962
    },
    {
      "epoch": 1.090875301988478,
      "grad_norm": 0.679717481136322,
      "learning_rate": 8.349347667925474e-05,
      "loss": 0.8887,
      "step": 5963
    },
    {
      "epoch": 1.0910611410518491,
      "grad_norm": 0.6497548222541809,
      "learning_rate": 8.346463267945478e-05,
      "loss": 0.7635,
      "step": 5964
    },
    {
      "epoch": 1.0912469801152203,
      "grad_norm": 0.6662367582321167,
      "learning_rate": 8.343579009396057e-05,
      "loss": 0.8766,
      "step": 5965
    },
    {
      "epoch": 1.0914328191785914,
      "grad_norm": 0.7309990525245667,
      "learning_rate": 8.340694892523898e-05,
      "loss": 1.0396,
      "step": 5966
    },
    {
      "epoch": 1.0916186582419625,
      "grad_norm": 0.6899405717849731,
      "learning_rate": 8.337810917575694e-05,
      "loss": 1.0736,
      "step": 5967
    },
    {
      "epoch": 1.0918044973053336,
      "grad_norm": 0.7755197286605835,
      "learning_rate": 8.334927084798112e-05,
      "loss": 0.8966,
      "step": 5968
    },
    {
      "epoch": 1.0919903363687047,
      "grad_norm": 0.996249258518219,
      "learning_rate": 8.332043394437815e-05,
      "loss": 1.0628,
      "step": 5969
    },
    {
      "epoch": 1.0921761754320758,
      "grad_norm": 0.6849657893180847,
      "learning_rate": 8.329159846741457e-05,
      "loss": 0.7763,
      "step": 5970
    },
    {
      "epoch": 1.092362014495447,
      "grad_norm": 0.5944156646728516,
      "learning_rate": 8.326276441955663e-05,
      "loss": 0.9079,
      "step": 5971
    },
    {
      "epoch": 1.092547853558818,
      "grad_norm": 0.6766306757926941,
      "learning_rate": 8.323393180327066e-05,
      "loss": 1.0014,
      "step": 5972
    },
    {
      "epoch": 1.0927336926221891,
      "grad_norm": 0.6476278901100159,
      "learning_rate": 8.320510062102272e-05,
      "loss": 1.1559,
      "step": 5973
    },
    {
      "epoch": 1.0929195316855602,
      "grad_norm": 0.6700665354728699,
      "learning_rate": 8.317627087527882e-05,
      "loss": 0.892,
      "step": 5974
    },
    {
      "epoch": 1.0931053707489313,
      "grad_norm": 0.6307082176208496,
      "learning_rate": 8.314744256850482e-05,
      "loss": 0.8528,
      "step": 5975
    },
    {
      "epoch": 1.0932912098123024,
      "grad_norm": 0.6499921083450317,
      "learning_rate": 8.311861570316648e-05,
      "loss": 1.0078,
      "step": 5976
    },
    {
      "epoch": 1.0934770488756738,
      "grad_norm": 0.5335561633110046,
      "learning_rate": 8.308979028172948e-05,
      "loss": 0.6987,
      "step": 5977
    },
    {
      "epoch": 1.0936628879390449,
      "grad_norm": 1.7582954168319702,
      "learning_rate": 8.306096630665921e-05,
      "loss": 1.4374,
      "step": 5978
    },
    {
      "epoch": 1.093848727002416,
      "grad_norm": 0.8077002763748169,
      "learning_rate": 8.303214378042114e-05,
      "loss": 0.8452,
      "step": 5979
    },
    {
      "epoch": 1.094034566065787,
      "grad_norm": 0.6156245470046997,
      "learning_rate": 8.300332270548044e-05,
      "loss": 0.9902,
      "step": 5980
    },
    {
      "epoch": 1.0942204051291582,
      "grad_norm": 0.6083616018295288,
      "learning_rate": 8.297450308430231e-05,
      "loss": 0.9913,
      "step": 5981
    },
    {
      "epoch": 1.0944062441925293,
      "grad_norm": 0.5292848348617554,
      "learning_rate": 8.29456849193517e-05,
      "loss": 0.8474,
      "step": 5982
    },
    {
      "epoch": 1.0945920832559004,
      "grad_norm": 0.6098827123641968,
      "learning_rate": 8.291686821309353e-05,
      "loss": 0.8643,
      "step": 5983
    },
    {
      "epoch": 1.0947779223192715,
      "grad_norm": 0.7073376178741455,
      "learning_rate": 8.288805296799259e-05,
      "loss": 0.9915,
      "step": 5984
    },
    {
      "epoch": 1.0949637613826426,
      "grad_norm": 0.6034481525421143,
      "learning_rate": 8.285923918651341e-05,
      "loss": 1.0649,
      "step": 5985
    },
    {
      "epoch": 1.0951496004460137,
      "grad_norm": 0.6212545037269592,
      "learning_rate": 8.28304268711206e-05,
      "loss": 0.7996,
      "step": 5986
    },
    {
      "epoch": 1.0953354395093848,
      "grad_norm": 0.6276280283927917,
      "learning_rate": 8.280161602427845e-05,
      "loss": 0.945,
      "step": 5987
    },
    {
      "epoch": 1.095521278572756,
      "grad_norm": 0.7413529753684998,
      "learning_rate": 8.277280664845125e-05,
      "loss": 0.8979,
      "step": 5988
    },
    {
      "epoch": 1.095707117636127,
      "grad_norm": 0.6977653503417969,
      "learning_rate": 8.274399874610312e-05,
      "loss": 1.0445,
      "step": 5989
    },
    {
      "epoch": 1.0958929566994982,
      "grad_norm": 1.6293609142303467,
      "learning_rate": 8.271519231969807e-05,
      "loss": 1.4387,
      "step": 5990
    },
    {
      "epoch": 1.0960787957628693,
      "grad_norm": 0.5372196435928345,
      "learning_rate": 8.268638737170004e-05,
      "loss": 0.7948,
      "step": 5991
    },
    {
      "epoch": 1.0962646348262406,
      "grad_norm": 0.6748941540718079,
      "learning_rate": 8.265758390457266e-05,
      "loss": 0.6746,
      "step": 5992
    },
    {
      "epoch": 1.0964504738896117,
      "grad_norm": 0.8666142225265503,
      "learning_rate": 8.262878192077965e-05,
      "loss": 1.3074,
      "step": 5993
    },
    {
      "epoch": 1.0966363129529828,
      "grad_norm": 0.6674154996871948,
      "learning_rate": 8.259998142278442e-05,
      "loss": 1.1204,
      "step": 5994
    },
    {
      "epoch": 1.096822152016354,
      "grad_norm": 0.6468871235847473,
      "learning_rate": 8.257118241305042e-05,
      "loss": 1.0995,
      "step": 5995
    },
    {
      "epoch": 1.097007991079725,
      "grad_norm": 0.7309514880180359,
      "learning_rate": 8.254238489404083e-05,
      "loss": 1.0042,
      "step": 5996
    },
    {
      "epoch": 1.0971938301430961,
      "grad_norm": 0.7083085179328918,
      "learning_rate": 8.251358886821878e-05,
      "loss": 1.0292,
      "step": 5997
    },
    {
      "epoch": 1.0973796692064672,
      "grad_norm": 0.6642683744430542,
      "learning_rate": 8.248479433804732e-05,
      "loss": 1.1411,
      "step": 5998
    },
    {
      "epoch": 1.0975655082698383,
      "grad_norm": 0.7239875793457031,
      "learning_rate": 8.245600130598922e-05,
      "loss": 0.7915,
      "step": 5999
    },
    {
      "epoch": 1.0977513473332094,
      "grad_norm": 0.8348709940910339,
      "learning_rate": 8.242720977450728e-05,
      "loss": 1.1414,
      "step": 6000
    },
    {
      "epoch": 1.0979371863965806,
      "grad_norm": 0.6298574209213257,
      "learning_rate": 8.239841974606403e-05,
      "loss": 1.1092,
      "step": 6001
    },
    {
      "epoch": 1.0981230254599517,
      "grad_norm": 1.916271686553955,
      "learning_rate": 8.2369631223122e-05,
      "loss": 1.7966,
      "step": 6002
    },
    {
      "epoch": 1.0983088645233228,
      "grad_norm": 0.7313844561576843,
      "learning_rate": 8.23408442081435e-05,
      "loss": 0.9692,
      "step": 6003
    },
    {
      "epoch": 1.0984947035866939,
      "grad_norm": 0.5213794112205505,
      "learning_rate": 8.231205870359077e-05,
      "loss": 0.7192,
      "step": 6004
    },
    {
      "epoch": 1.098680542650065,
      "grad_norm": 0.6529609560966492,
      "learning_rate": 8.22832747119259e-05,
      "loss": 1.0331,
      "step": 6005
    },
    {
      "epoch": 1.098866381713436,
      "grad_norm": 0.6834220290184021,
      "learning_rate": 8.225449223561081e-05,
      "loss": 0.9095,
      "step": 6006
    },
    {
      "epoch": 1.0990522207768072,
      "grad_norm": 0.6474689841270447,
      "learning_rate": 8.222571127710738e-05,
      "loss": 1.1353,
      "step": 6007
    },
    {
      "epoch": 1.0992380598401783,
      "grad_norm": 0.8416957855224609,
      "learning_rate": 8.219693183887726e-05,
      "loss": 1.3315,
      "step": 6008
    },
    {
      "epoch": 1.0994238989035496,
      "grad_norm": 0.631695568561554,
      "learning_rate": 8.216815392338205e-05,
      "loss": 0.9463,
      "step": 6009
    },
    {
      "epoch": 1.0996097379669207,
      "grad_norm": 0.6356896758079529,
      "learning_rate": 8.213937753308316e-05,
      "loss": 1.0267,
      "step": 6010
    },
    {
      "epoch": 1.0997955770302918,
      "grad_norm": 0.6175408363342285,
      "learning_rate": 8.211060267044191e-05,
      "loss": 0.9336,
      "step": 6011
    },
    {
      "epoch": 1.099981416093663,
      "grad_norm": 1.0049728155136108,
      "learning_rate": 8.20818293379195e-05,
      "loss": 1.0377,
      "step": 6012
    },
    {
      "epoch": 1.100167255157034,
      "grad_norm": 0.5931586027145386,
      "learning_rate": 8.205305753797691e-05,
      "loss": 0.9247,
      "step": 6013
    },
    {
      "epoch": 1.1003530942204052,
      "grad_norm": 0.7019885778427124,
      "learning_rate": 8.202428727307516e-05,
      "loss": 0.9673,
      "step": 6014
    },
    {
      "epoch": 1.1005389332837763,
      "grad_norm": 0.666867196559906,
      "learning_rate": 8.199551854567492e-05,
      "loss": 0.7624,
      "step": 6015
    },
    {
      "epoch": 1.1007247723471474,
      "grad_norm": 0.6581511497497559,
      "learning_rate": 8.196675135823689e-05,
      "loss": 0.7796,
      "step": 6016
    },
    {
      "epoch": 1.1009106114105185,
      "grad_norm": 0.6427851319313049,
      "learning_rate": 8.193798571322164e-05,
      "loss": 0.9612,
      "step": 6017
    },
    {
      "epoch": 1.1010964504738896,
      "grad_norm": 0.7291167378425598,
      "learning_rate": 8.190922161308946e-05,
      "loss": 1.08,
      "step": 6018
    },
    {
      "epoch": 1.1012822895372607,
      "grad_norm": 0.6328938007354736,
      "learning_rate": 8.188045906030069e-05,
      "loss": 0.9169,
      "step": 6019
    },
    {
      "epoch": 1.1014681286006318,
      "grad_norm": 0.7824090123176575,
      "learning_rate": 8.185169805731537e-05,
      "loss": 0.9103,
      "step": 6020
    },
    {
      "epoch": 1.101653967664003,
      "grad_norm": 0.6461734771728516,
      "learning_rate": 8.18229386065936e-05,
      "loss": 0.8457,
      "step": 6021
    },
    {
      "epoch": 1.101839806727374,
      "grad_norm": 0.6358814239501953,
      "learning_rate": 8.179418071059514e-05,
      "loss": 0.8985,
      "step": 6022
    },
    {
      "epoch": 1.1020256457907451,
      "grad_norm": 0.6370760202407837,
      "learning_rate": 8.176542437177974e-05,
      "loss": 0.8347,
      "step": 6023
    },
    {
      "epoch": 1.1022114848541162,
      "grad_norm": 0.7202227115631104,
      "learning_rate": 8.173666959260705e-05,
      "loss": 1.3205,
      "step": 6024
    },
    {
      "epoch": 1.1023973239174873,
      "grad_norm": 0.611150324344635,
      "learning_rate": 8.170791637553643e-05,
      "loss": 0.8695,
      "step": 6025
    },
    {
      "epoch": 1.1025831629808587,
      "grad_norm": 0.5846618413925171,
      "learning_rate": 8.16791647230273e-05,
      "loss": 0.8868,
      "step": 6026
    },
    {
      "epoch": 1.1027690020442298,
      "grad_norm": 0.7279630303382874,
      "learning_rate": 8.165041463753876e-05,
      "loss": 1.0683,
      "step": 6027
    },
    {
      "epoch": 1.102954841107601,
      "grad_norm": 0.6632223725318909,
      "learning_rate": 8.162166612152998e-05,
      "loss": 0.9749,
      "step": 6028
    },
    {
      "epoch": 1.103140680170972,
      "grad_norm": 0.7467746138572693,
      "learning_rate": 8.159291917745976e-05,
      "loss": 0.8073,
      "step": 6029
    },
    {
      "epoch": 1.103326519234343,
      "grad_norm": 0.6717825531959534,
      "learning_rate": 8.156417380778696e-05,
      "loss": 0.827,
      "step": 6030
    },
    {
      "epoch": 1.1035123582977142,
      "grad_norm": 0.7068509459495544,
      "learning_rate": 8.153543001497024e-05,
      "loss": 0.837,
      "step": 6031
    },
    {
      "epoch": 1.1036981973610853,
      "grad_norm": 0.6751839518547058,
      "learning_rate": 8.150668780146807e-05,
      "loss": 0.975,
      "step": 6032
    },
    {
      "epoch": 1.1038840364244564,
      "grad_norm": 0.5865058898925781,
      "learning_rate": 8.147794716973889e-05,
      "loss": 0.6932,
      "step": 6033
    },
    {
      "epoch": 1.1040698754878275,
      "grad_norm": 0.6836567521095276,
      "learning_rate": 8.144920812224089e-05,
      "loss": 0.9753,
      "step": 6034
    },
    {
      "epoch": 1.1042557145511986,
      "grad_norm": 0.6463656425476074,
      "learning_rate": 8.142047066143226e-05,
      "loss": 1.0225,
      "step": 6035
    },
    {
      "epoch": 1.1044415536145697,
      "grad_norm": 0.6912021636962891,
      "learning_rate": 8.139173478977087e-05,
      "loss": 1.0064,
      "step": 6036
    },
    {
      "epoch": 1.1046273926779409,
      "grad_norm": 0.6893657445907593,
      "learning_rate": 8.136300050971464e-05,
      "loss": 0.8931,
      "step": 6037
    },
    {
      "epoch": 1.104813231741312,
      "grad_norm": 0.811796247959137,
      "learning_rate": 8.13342678237213e-05,
      "loss": 0.6038,
      "step": 6038
    },
    {
      "epoch": 1.104999070804683,
      "grad_norm": 0.6037810444831848,
      "learning_rate": 8.130553673424832e-05,
      "loss": 0.9233,
      "step": 6039
    },
    {
      "epoch": 1.1051849098680542,
      "grad_norm": 0.730668842792511,
      "learning_rate": 8.127680724375322e-05,
      "loss": 0.8656,
      "step": 6040
    },
    {
      "epoch": 1.1053707489314253,
      "grad_norm": 0.7074486017227173,
      "learning_rate": 8.124807935469323e-05,
      "loss": 0.9637,
      "step": 6041
    },
    {
      "epoch": 1.1055565879947964,
      "grad_norm": 0.6678810715675354,
      "learning_rate": 8.121935306952558e-05,
      "loss": 0.9592,
      "step": 6042
    },
    {
      "epoch": 1.1057424270581677,
      "grad_norm": 0.6933256983757019,
      "learning_rate": 8.119062839070723e-05,
      "loss": 1.1061,
      "step": 6043
    },
    {
      "epoch": 1.1059282661215388,
      "grad_norm": 1.7148810625076294,
      "learning_rate": 8.116190532069507e-05,
      "loss": 1.7415,
      "step": 6044
    },
    {
      "epoch": 1.10611410518491,
      "grad_norm": 0.7014167308807373,
      "learning_rate": 8.11331838619459e-05,
      "loss": 1.0122,
      "step": 6045
    },
    {
      "epoch": 1.106299944248281,
      "grad_norm": 0.8506425619125366,
      "learning_rate": 8.110446401691627e-05,
      "loss": 1.0112,
      "step": 6046
    },
    {
      "epoch": 1.1064857833116521,
      "grad_norm": 0.7145703434944153,
      "learning_rate": 8.107574578806267e-05,
      "loss": 0.841,
      "step": 6047
    },
    {
      "epoch": 1.1066716223750233,
      "grad_norm": 0.6292847990989685,
      "learning_rate": 8.104702917784143e-05,
      "loss": 0.898,
      "step": 6048
    },
    {
      "epoch": 1.1068574614383944,
      "grad_norm": 0.709343671798706,
      "learning_rate": 8.101831418870877e-05,
      "loss": 0.8223,
      "step": 6049
    },
    {
      "epoch": 1.1070433005017655,
      "grad_norm": 0.6488339900970459,
      "learning_rate": 8.09896008231207e-05,
      "loss": 0.7187,
      "step": 6050
    },
    {
      "epoch": 1.1072291395651366,
      "grad_norm": 0.6357553005218506,
      "learning_rate": 8.096088908353315e-05,
      "loss": 0.7479,
      "step": 6051
    },
    {
      "epoch": 1.1074149786285077,
      "grad_norm": 0.6017008423805237,
      "learning_rate": 8.093217897240195e-05,
      "loss": 0.9325,
      "step": 6052
    },
    {
      "epoch": 1.1076008176918788,
      "grad_norm": 0.613470196723938,
      "learning_rate": 8.090347049218266e-05,
      "loss": 1.0354,
      "step": 6053
    },
    {
      "epoch": 1.10778665675525,
      "grad_norm": 0.7553649544715881,
      "learning_rate": 8.087476364533082e-05,
      "loss": 1.0809,
      "step": 6054
    },
    {
      "epoch": 1.107972495818621,
      "grad_norm": 0.6359599828720093,
      "learning_rate": 8.084605843430177e-05,
      "loss": 0.9372,
      "step": 6055
    },
    {
      "epoch": 1.1081583348819921,
      "grad_norm": 0.8213351964950562,
      "learning_rate": 8.081735486155076e-05,
      "loss": 0.9785,
      "step": 6056
    },
    {
      "epoch": 1.1083441739453632,
      "grad_norm": 0.6647172570228577,
      "learning_rate": 8.078865292953281e-05,
      "loss": 0.7293,
      "step": 6057
    },
    {
      "epoch": 1.1085300130087345,
      "grad_norm": 0.5401493310928345,
      "learning_rate": 8.075995264070291e-05,
      "loss": 0.6513,
      "step": 6058
    },
    {
      "epoch": 1.1087158520721057,
      "grad_norm": 0.4942319691181183,
      "learning_rate": 8.073125399751589e-05,
      "loss": 0.6459,
      "step": 6059
    },
    {
      "epoch": 1.1089016911354768,
      "grad_norm": 0.6275574564933777,
      "learning_rate": 8.07025570024263e-05,
      "loss": 0.7735,
      "step": 6060
    },
    {
      "epoch": 1.1090875301988479,
      "grad_norm": 0.5724532008171082,
      "learning_rate": 8.067386165788878e-05,
      "loss": 0.8382,
      "step": 6061
    },
    {
      "epoch": 1.109273369262219,
      "grad_norm": 0.7397722601890564,
      "learning_rate": 8.064516796635758e-05,
      "loss": 0.9636,
      "step": 6062
    },
    {
      "epoch": 1.10945920832559,
      "grad_norm": 0.629734456539154,
      "learning_rate": 8.061647593028703e-05,
      "loss": 1.06,
      "step": 6063
    },
    {
      "epoch": 1.1096450473889612,
      "grad_norm": 0.6586220264434814,
      "learning_rate": 8.058778555213116e-05,
      "loss": 1.0629,
      "step": 6064
    },
    {
      "epoch": 1.1098308864523323,
      "grad_norm": 0.5675712823867798,
      "learning_rate": 8.055909683434395e-05,
      "loss": 0.9236,
      "step": 6065
    },
    {
      "epoch": 1.1100167255157034,
      "grad_norm": 0.6634374260902405,
      "learning_rate": 8.053040977937924e-05,
      "loss": 0.8846,
      "step": 6066
    },
    {
      "epoch": 1.1102025645790745,
      "grad_norm": 0.6590657234191895,
      "learning_rate": 8.050172438969063e-05,
      "loss": 0.8831,
      "step": 6067
    },
    {
      "epoch": 1.1103884036424456,
      "grad_norm": 0.6578685641288757,
      "learning_rate": 8.04730406677317e-05,
      "loss": 0.9238,
      "step": 6068
    },
    {
      "epoch": 1.1105742427058167,
      "grad_norm": 0.7250857949256897,
      "learning_rate": 8.044435861595578e-05,
      "loss": 0.9559,
      "step": 6069
    },
    {
      "epoch": 1.1107600817691878,
      "grad_norm": 0.6505029797554016,
      "learning_rate": 8.041567823681615e-05,
      "loss": 0.8535,
      "step": 6070
    },
    {
      "epoch": 1.110945920832559,
      "grad_norm": 0.7013585567474365,
      "learning_rate": 8.038699953276587e-05,
      "loss": 0.7633,
      "step": 6071
    },
    {
      "epoch": 1.11113175989593,
      "grad_norm": 0.754403293132782,
      "learning_rate": 8.035832250625788e-05,
      "loss": 1.1925,
      "step": 6072
    },
    {
      "epoch": 1.1113175989593012,
      "grad_norm": 0.7387733459472656,
      "learning_rate": 8.032964715974509e-05,
      "loss": 1.0599,
      "step": 6073
    },
    {
      "epoch": 1.1115034380226723,
      "grad_norm": 0.6414195895195007,
      "learning_rate": 8.030097349568004e-05,
      "loss": 0.9555,
      "step": 6074
    },
    {
      "epoch": 1.1116892770860436,
      "grad_norm": 0.6676441431045532,
      "learning_rate": 8.027230151651535e-05,
      "loss": 1.1428,
      "step": 6075
    },
    {
      "epoch": 1.1118751161494147,
      "grad_norm": 0.6780575513839722,
      "learning_rate": 8.024363122470331e-05,
      "loss": 0.9103,
      "step": 6076
    },
    {
      "epoch": 1.1120609552127858,
      "grad_norm": 0.6453208923339844,
      "learning_rate": 8.02149626226962e-05,
      "loss": 1.0003,
      "step": 6077
    },
    {
      "epoch": 1.112246794276157,
      "grad_norm": 0.702014148235321,
      "learning_rate": 8.018629571294607e-05,
      "loss": 1.0396,
      "step": 6078
    },
    {
      "epoch": 1.112432633339528,
      "grad_norm": 0.6062483191490173,
      "learning_rate": 8.01576304979049e-05,
      "loss": 1.0062,
      "step": 6079
    },
    {
      "epoch": 1.1126184724028991,
      "grad_norm": 0.7950749397277832,
      "learning_rate": 8.012896698002454e-05,
      "loss": 0.8965,
      "step": 6080
    },
    {
      "epoch": 1.1128043114662702,
      "grad_norm": 0.5671384334564209,
      "learning_rate": 8.010030516175651e-05,
      "loss": 0.6552,
      "step": 6081
    },
    {
      "epoch": 1.1129901505296413,
      "grad_norm": 0.7059739232063293,
      "learning_rate": 8.007164504555244e-05,
      "loss": 1.2306,
      "step": 6082
    },
    {
      "epoch": 1.1131759895930124,
      "grad_norm": 0.8960155844688416,
      "learning_rate": 8.00429866338636e-05,
      "loss": 1.0406,
      "step": 6083
    },
    {
      "epoch": 1.1133618286563836,
      "grad_norm": 0.7356314063072205,
      "learning_rate": 8.001432992914127e-05,
      "loss": 1.0919,
      "step": 6084
    },
    {
      "epoch": 1.1135476677197547,
      "grad_norm": 0.6466111540794373,
      "learning_rate": 7.998567493383647e-05,
      "loss": 0.9251,
      "step": 6085
    },
    {
      "epoch": 1.1137335067831258,
      "grad_norm": 3.527097463607788,
      "learning_rate": 7.995702165040015e-05,
      "loss": 1.8559,
      "step": 6086
    },
    {
      "epoch": 1.1139193458464969,
      "grad_norm": 0.5798239707946777,
      "learning_rate": 7.992837008128313e-05,
      "loss": 0.9931,
      "step": 6087
    },
    {
      "epoch": 1.114105184909868,
      "grad_norm": 0.580335795879364,
      "learning_rate": 7.989972022893595e-05,
      "loss": 0.9003,
      "step": 6088
    },
    {
      "epoch": 1.114291023973239,
      "grad_norm": 0.5507724285125732,
      "learning_rate": 7.98710720958092e-05,
      "loss": 0.8943,
      "step": 6089
    },
    {
      "epoch": 1.1144768630366102,
      "grad_norm": 0.6352560520172119,
      "learning_rate": 7.98424256843531e-05,
      "loss": 0.9323,
      "step": 6090
    },
    {
      "epoch": 1.1146627020999813,
      "grad_norm": 0.7881081700325012,
      "learning_rate": 7.981378099701793e-05,
      "loss": 1.044,
      "step": 6091
    },
    {
      "epoch": 1.1148485411633526,
      "grad_norm": 0.7111803293228149,
      "learning_rate": 7.978513803625366e-05,
      "loss": 1.0815,
      "step": 6092
    },
    {
      "epoch": 1.1150343802267237,
      "grad_norm": 0.714192807674408,
      "learning_rate": 7.975649680451024e-05,
      "loss": 1.2705,
      "step": 6093
    },
    {
      "epoch": 1.1152202192900948,
      "grad_norm": 0.6725999116897583,
      "learning_rate": 7.972785730423744e-05,
      "loss": 0.9721,
      "step": 6094
    },
    {
      "epoch": 1.115406058353466,
      "grad_norm": 0.6231698393821716,
      "learning_rate": 7.969921953788478e-05,
      "loss": 0.9319,
      "step": 6095
    },
    {
      "epoch": 1.115591897416837,
      "grad_norm": 0.632483959197998,
      "learning_rate": 7.967058350790177e-05,
      "loss": 0.8918,
      "step": 6096
    },
    {
      "epoch": 1.1157777364802082,
      "grad_norm": 2.3974478244781494,
      "learning_rate": 7.964194921673766e-05,
      "loss": 1.8149,
      "step": 6097
    },
    {
      "epoch": 1.1159635755435793,
      "grad_norm": 0.5865818858146667,
      "learning_rate": 7.961331666684166e-05,
      "loss": 1.0894,
      "step": 6098
    },
    {
      "epoch": 1.1161494146069504,
      "grad_norm": 0.8524091243743896,
      "learning_rate": 7.958468586066273e-05,
      "loss": 1.0211,
      "step": 6099
    },
    {
      "epoch": 1.1163352536703215,
      "grad_norm": 0.7423559427261353,
      "learning_rate": 7.955605680064975e-05,
      "loss": 1.1801,
      "step": 6100
    },
    {
      "epoch": 1.1165210927336926,
      "grad_norm": 0.7492449283599854,
      "learning_rate": 7.952742948925143e-05,
      "loss": 0.906,
      "step": 6101
    },
    {
      "epoch": 1.1167069317970637,
      "grad_norm": 0.5786088705062866,
      "learning_rate": 7.949880392891628e-05,
      "loss": 0.7568,
      "step": 6102
    },
    {
      "epoch": 1.1168927708604348,
      "grad_norm": 0.536382257938385,
      "learning_rate": 7.947018012209279e-05,
      "loss": 0.8471,
      "step": 6103
    },
    {
      "epoch": 1.117078609923806,
      "grad_norm": 0.6602856516838074,
      "learning_rate": 7.944155807122912e-05,
      "loss": 0.8139,
      "step": 6104
    },
    {
      "epoch": 1.117264448987177,
      "grad_norm": 0.6715288162231445,
      "learning_rate": 7.941293777877344e-05,
      "loss": 0.8542,
      "step": 6105
    },
    {
      "epoch": 1.1174502880505481,
      "grad_norm": 0.7064169645309448,
      "learning_rate": 7.938431924717368e-05,
      "loss": 1.1756,
      "step": 6106
    },
    {
      "epoch": 1.1176361271139192,
      "grad_norm": 0.6816471815109253,
      "learning_rate": 7.935570247887764e-05,
      "loss": 0.9043,
      "step": 6107
    },
    {
      "epoch": 1.1178219661772903,
      "grad_norm": 0.5369113087654114,
      "learning_rate": 7.932708747633299e-05,
      "loss": 0.8338,
      "step": 6108
    },
    {
      "epoch": 1.1180078052406617,
      "grad_norm": 0.6738365292549133,
      "learning_rate": 7.929847424198723e-05,
      "loss": 0.8163,
      "step": 6109
    },
    {
      "epoch": 1.1181936443040328,
      "grad_norm": 0.7385087609291077,
      "learning_rate": 7.926986277828774e-05,
      "loss": 1.0322,
      "step": 6110
    },
    {
      "epoch": 1.118379483367404,
      "grad_norm": 0.6074032783508301,
      "learning_rate": 7.924125308768165e-05,
      "loss": 0.8479,
      "step": 6111
    },
    {
      "epoch": 1.118565322430775,
      "grad_norm": 0.6593636870384216,
      "learning_rate": 7.92126451726161e-05,
      "loss": 0.8993,
      "step": 6112
    },
    {
      "epoch": 1.118751161494146,
      "grad_norm": 0.7033354043960571,
      "learning_rate": 7.918403903553788e-05,
      "loss": 1.056,
      "step": 6113
    },
    {
      "epoch": 1.1189370005575172,
      "grad_norm": 0.6170438528060913,
      "learning_rate": 7.915543467889379e-05,
      "loss": 1.0231,
      "step": 6114
    },
    {
      "epoch": 1.1191228396208883,
      "grad_norm": 0.6546794176101685,
      "learning_rate": 7.912683210513046e-05,
      "loss": 0.9556,
      "step": 6115
    },
    {
      "epoch": 1.1193086786842594,
      "grad_norm": 0.6962534785270691,
      "learning_rate": 7.909823131669425e-05,
      "loss": 1.077,
      "step": 6116
    },
    {
      "epoch": 1.1194945177476305,
      "grad_norm": 0.7631762623786926,
      "learning_rate": 7.906963231603155e-05,
      "loss": 0.9765,
      "step": 6117
    },
    {
      "epoch": 1.1196803568110016,
      "grad_norm": 0.6790252327919006,
      "learning_rate": 7.904103510558838e-05,
      "loss": 0.9533,
      "step": 6118
    },
    {
      "epoch": 1.1198661958743727,
      "grad_norm": 0.774306058883667,
      "learning_rate": 7.901243968781083e-05,
      "loss": 1.3034,
      "step": 6119
    },
    {
      "epoch": 1.1200520349377439,
      "grad_norm": 0.6652829051017761,
      "learning_rate": 7.898384606514462e-05,
      "loss": 0.7831,
      "step": 6120
    },
    {
      "epoch": 1.120237874001115,
      "grad_norm": 0.7548511624336243,
      "learning_rate": 7.89552542400355e-05,
      "loss": 1.0368,
      "step": 6121
    },
    {
      "epoch": 1.120423713064486,
      "grad_norm": 0.7231112122535706,
      "learning_rate": 7.892666421492896e-05,
      "loss": 1.1899,
      "step": 6122
    },
    {
      "epoch": 1.1206095521278572,
      "grad_norm": 0.7107078433036804,
      "learning_rate": 7.889807599227037e-05,
      "loss": 1.0542,
      "step": 6123
    },
    {
      "epoch": 1.1207953911912285,
      "grad_norm": 0.5880662798881531,
      "learning_rate": 7.886948957450498e-05,
      "loss": 1.0688,
      "step": 6124
    },
    {
      "epoch": 1.1209812302545996,
      "grad_norm": 0.6817053556442261,
      "learning_rate": 7.884090496407779e-05,
      "loss": 1.0867,
      "step": 6125
    },
    {
      "epoch": 1.1211670693179707,
      "grad_norm": 0.6306520104408264,
      "learning_rate": 7.881232216343375e-05,
      "loss": 0.8813,
      "step": 6126
    },
    {
      "epoch": 1.1213529083813418,
      "grad_norm": 0.5859300494194031,
      "learning_rate": 7.878374117501757e-05,
      "loss": 0.7759,
      "step": 6127
    },
    {
      "epoch": 1.121538747444713,
      "grad_norm": 1.0821082592010498,
      "learning_rate": 7.875516200127385e-05,
      "loss": 0.7981,
      "step": 6128
    },
    {
      "epoch": 1.121724586508084,
      "grad_norm": 0.5714467167854309,
      "learning_rate": 7.872658464464704e-05,
      "loss": 1.0366,
      "step": 6129
    },
    {
      "epoch": 1.1219104255714551,
      "grad_norm": 0.6259623169898987,
      "learning_rate": 7.869800910758143e-05,
      "loss": 0.7744,
      "step": 6130
    },
    {
      "epoch": 1.1220962646348263,
      "grad_norm": 0.5901505947113037,
      "learning_rate": 7.866943539252118e-05,
      "loss": 0.9946,
      "step": 6131
    },
    {
      "epoch": 1.1222821036981974,
      "grad_norm": 0.5895738005638123,
      "learning_rate": 7.864086350191017e-05,
      "loss": 0.7583,
      "step": 6132
    },
    {
      "epoch": 1.1224679427615685,
      "grad_norm": 0.6195260882377625,
      "learning_rate": 7.861229343819232e-05,
      "loss": 0.811,
      "step": 6133
    },
    {
      "epoch": 1.1226537818249396,
      "grad_norm": 0.614778995513916,
      "learning_rate": 7.858372520381119e-05,
      "loss": 1.1254,
      "step": 6134
    },
    {
      "epoch": 1.1228396208883107,
      "grad_norm": 0.5922778844833374,
      "learning_rate": 7.855515880121032e-05,
      "loss": 0.9307,
      "step": 6135
    },
    {
      "epoch": 1.1230254599516818,
      "grad_norm": 0.5865144729614258,
      "learning_rate": 7.85265942328331e-05,
      "loss": 1.0438,
      "step": 6136
    },
    {
      "epoch": 1.123211299015053,
      "grad_norm": 0.6751108765602112,
      "learning_rate": 7.849803150112267e-05,
      "loss": 0.9653,
      "step": 6137
    },
    {
      "epoch": 1.123397138078424,
      "grad_norm": 0.7323153614997864,
      "learning_rate": 7.84694706085221e-05,
      "loss": 0.9208,
      "step": 6138
    },
    {
      "epoch": 1.1235829771417951,
      "grad_norm": 0.6893547773361206,
      "learning_rate": 7.844091155747421e-05,
      "loss": 1.108,
      "step": 6139
    },
    {
      "epoch": 1.1237688162051662,
      "grad_norm": 0.5926496982574463,
      "learning_rate": 7.84123543504218e-05,
      "loss": 0.8949,
      "step": 6140
    },
    {
      "epoch": 1.1239546552685376,
      "grad_norm": 0.6670359373092651,
      "learning_rate": 7.838379898980734e-05,
      "loss": 1.0495,
      "step": 6141
    },
    {
      "epoch": 1.1241404943319087,
      "grad_norm": 0.7171518802642822,
      "learning_rate": 7.835524547807326e-05,
      "loss": 0.9813,
      "step": 6142
    },
    {
      "epoch": 1.1243263333952798,
      "grad_norm": 0.6894052028656006,
      "learning_rate": 7.832669381766183e-05,
      "loss": 0.9703,
      "step": 6143
    },
    {
      "epoch": 1.1245121724586509,
      "grad_norm": 0.639323890209198,
      "learning_rate": 7.829814401101512e-05,
      "loss": 0.9347,
      "step": 6144
    },
    {
      "epoch": 1.124698011522022,
      "grad_norm": 0.783460259437561,
      "learning_rate": 7.826959606057508e-05,
      "loss": 0.8925,
      "step": 6145
    },
    {
      "epoch": 1.124883850585393,
      "grad_norm": 0.605365514755249,
      "learning_rate": 7.824104996878342e-05,
      "loss": 0.9144,
      "step": 6146
    },
    {
      "epoch": 1.1250696896487642,
      "grad_norm": 0.6294218301773071,
      "learning_rate": 7.821250573808181e-05,
      "loss": 0.8606,
      "step": 6147
    },
    {
      "epoch": 1.1252555287121353,
      "grad_norm": 0.6887063384056091,
      "learning_rate": 7.818396337091166e-05,
      "loss": 0.6776,
      "step": 6148
    },
    {
      "epoch": 1.1254413677755064,
      "grad_norm": 0.6241545677185059,
      "learning_rate": 7.815542286971426e-05,
      "loss": 0.9171,
      "step": 6149
    },
    {
      "epoch": 1.1256272068388775,
      "grad_norm": 0.6879689693450928,
      "learning_rate": 7.812688423693078e-05,
      "loss": 1.0305,
      "step": 6150
    },
    {
      "epoch": 1.1258130459022486,
      "grad_norm": 0.7353564500808716,
      "learning_rate": 7.809834747500215e-05,
      "loss": 1.0796,
      "step": 6151
    },
    {
      "epoch": 1.1259988849656197,
      "grad_norm": 0.6647047400474548,
      "learning_rate": 7.80698125863692e-05,
      "loss": 0.7962,
      "step": 6152
    },
    {
      "epoch": 1.1261847240289908,
      "grad_norm": 0.7037851810455322,
      "learning_rate": 7.804127957347256e-05,
      "loss": 0.8475,
      "step": 6153
    },
    {
      "epoch": 1.126370563092362,
      "grad_norm": 0.7323387265205383,
      "learning_rate": 7.80127484387528e-05,
      "loss": 0.9171,
      "step": 6154
    },
    {
      "epoch": 1.126556402155733,
      "grad_norm": 0.6864535808563232,
      "learning_rate": 7.79842191846501e-05,
      "loss": 0.7967,
      "step": 6155
    },
    {
      "epoch": 1.1267422412191044,
      "grad_norm": 0.6619288921356201,
      "learning_rate": 7.795569181360474e-05,
      "loss": 0.836,
      "step": 6156
    },
    {
      "epoch": 1.1269280802824753,
      "grad_norm": 0.6207095980644226,
      "learning_rate": 7.792716632805676e-05,
      "loss": 0.9877,
      "step": 6157
    },
    {
      "epoch": 1.1271139193458466,
      "grad_norm": 0.6751691699028015,
      "learning_rate": 7.78986427304459e-05,
      "loss": 0.8719,
      "step": 6158
    },
    {
      "epoch": 1.1272997584092177,
      "grad_norm": 0.6725651025772095,
      "learning_rate": 7.787012102321189e-05,
      "loss": 1.0775,
      "step": 6159
    },
    {
      "epoch": 1.1274855974725888,
      "grad_norm": 0.6328722238540649,
      "learning_rate": 7.784160120879427e-05,
      "loss": 1.1374,
      "step": 6160
    },
    {
      "epoch": 1.12767143653596,
      "grad_norm": 0.6430485844612122,
      "learning_rate": 7.781308328963242e-05,
      "loss": 1.0459,
      "step": 6161
    },
    {
      "epoch": 1.127857275599331,
      "grad_norm": 0.835422158241272,
      "learning_rate": 7.778456726816545e-05,
      "loss": 1.0023,
      "step": 6162
    },
    {
      "epoch": 1.1280431146627021,
      "grad_norm": 0.6023844480514526,
      "learning_rate": 7.775605314683248e-05,
      "loss": 1.0418,
      "step": 6163
    },
    {
      "epoch": 1.1282289537260732,
      "grad_norm": 0.6241095662117004,
      "learning_rate": 7.772754092807239e-05,
      "loss": 0.8114,
      "step": 6164
    },
    {
      "epoch": 1.1284147927894443,
      "grad_norm": 0.6519508361816406,
      "learning_rate": 7.769903061432381e-05,
      "loss": 0.9586,
      "step": 6165
    },
    {
      "epoch": 1.1286006318528155,
      "grad_norm": 0.7015917301177979,
      "learning_rate": 7.767052220802538e-05,
      "loss": 0.8807,
      "step": 6166
    },
    {
      "epoch": 1.1287864709161866,
      "grad_norm": 0.6102322340011597,
      "learning_rate": 7.764201571161541e-05,
      "loss": 0.7752,
      "step": 6167
    },
    {
      "epoch": 1.1289723099795577,
      "grad_norm": 0.6366463899612427,
      "learning_rate": 7.761351112753222e-05,
      "loss": 1.0049,
      "step": 6168
    },
    {
      "epoch": 1.1291581490429288,
      "grad_norm": 0.646867036819458,
      "learning_rate": 7.758500845821374e-05,
      "loss": 1.0766,
      "step": 6169
    },
    {
      "epoch": 1.1293439881062999,
      "grad_norm": 0.745267391204834,
      "learning_rate": 7.755650770609793e-05,
      "loss": 0.9621,
      "step": 6170
    },
    {
      "epoch": 1.129529827169671,
      "grad_norm": 0.7232451438903809,
      "learning_rate": 7.752800887362259e-05,
      "loss": 0.9204,
      "step": 6171
    },
    {
      "epoch": 1.129715666233042,
      "grad_norm": 0.6933814883232117,
      "learning_rate": 7.749951196322516e-05,
      "loss": 0.8385,
      "step": 6172
    },
    {
      "epoch": 1.1299015052964134,
      "grad_norm": 0.6276084780693054,
      "learning_rate": 7.747101697734313e-05,
      "loss": 0.8876,
      "step": 6173
    },
    {
      "epoch": 1.1300873443597843,
      "grad_norm": 0.6545588970184326,
      "learning_rate": 7.74425239184137e-05,
      "loss": 1.0254,
      "step": 6174
    },
    {
      "epoch": 1.1302731834231556,
      "grad_norm": 0.6565459966659546,
      "learning_rate": 7.741403278887397e-05,
      "loss": 0.9012,
      "step": 6175
    },
    {
      "epoch": 1.1304590224865267,
      "grad_norm": 0.6083407998085022,
      "learning_rate": 7.738554359116079e-05,
      "loss": 0.8095,
      "step": 6176
    },
    {
      "epoch": 1.1306448615498979,
      "grad_norm": 0.6915955543518066,
      "learning_rate": 7.735705632771095e-05,
      "loss": 0.9657,
      "step": 6177
    },
    {
      "epoch": 1.130830700613269,
      "grad_norm": 0.8749962449073792,
      "learning_rate": 7.732857100096107e-05,
      "loss": 0.866,
      "step": 6178
    },
    {
      "epoch": 1.13101653967664,
      "grad_norm": 0.7088390588760376,
      "learning_rate": 7.730008761334747e-05,
      "loss": 0.8856,
      "step": 6179
    },
    {
      "epoch": 1.1312023787400112,
      "grad_norm": 0.7398287653923035,
      "learning_rate": 7.727160616730644e-05,
      "loss": 1.0043,
      "step": 6180
    },
    {
      "epoch": 1.1313882178033823,
      "grad_norm": 0.7247969508171082,
      "learning_rate": 7.724312666527406e-05,
      "loss": 0.8159,
      "step": 6181
    },
    {
      "epoch": 1.1315740568667534,
      "grad_norm": 0.9901308417320251,
      "learning_rate": 7.721464910968627e-05,
      "loss": 1.0269,
      "step": 6182
    },
    {
      "epoch": 1.1317598959301245,
      "grad_norm": 0.7300354242324829,
      "learning_rate": 7.718617350297874e-05,
      "loss": 0.9958,
      "step": 6183
    },
    {
      "epoch": 1.1319457349934956,
      "grad_norm": 0.7227690815925598,
      "learning_rate": 7.71576998475871e-05,
      "loss": 0.8632,
      "step": 6184
    },
    {
      "epoch": 1.1321315740568667,
      "grad_norm": 0.7816814184188843,
      "learning_rate": 7.712922814594681e-05,
      "loss": 1.0998,
      "step": 6185
    },
    {
      "epoch": 1.1323174131202378,
      "grad_norm": 0.625962495803833,
      "learning_rate": 7.710075840049302e-05,
      "loss": 0.9012,
      "step": 6186
    },
    {
      "epoch": 1.132503252183609,
      "grad_norm": 0.6541265845298767,
      "learning_rate": 7.707229061366089e-05,
      "loss": 0.9436,
      "step": 6187
    },
    {
      "epoch": 1.13268909124698,
      "grad_norm": 0.6720447540283203,
      "learning_rate": 7.704382478788526e-05,
      "loss": 1.2808,
      "step": 6188
    },
    {
      "epoch": 1.1328749303103511,
      "grad_norm": 0.6707186698913574,
      "learning_rate": 7.701536092560095e-05,
      "loss": 0.8202,
      "step": 6189
    },
    {
      "epoch": 1.1330607693737225,
      "grad_norm": 0.560600996017456,
      "learning_rate": 7.698689902924246e-05,
      "loss": 0.6307,
      "step": 6190
    },
    {
      "epoch": 1.1332466084370936,
      "grad_norm": 0.8436470031738281,
      "learning_rate": 7.695843910124423e-05,
      "loss": 0.9579,
      "step": 6191
    },
    {
      "epoch": 1.1334324475004647,
      "grad_norm": 0.5533195734024048,
      "learning_rate": 7.692998114404055e-05,
      "loss": 0.6506,
      "step": 6192
    },
    {
      "epoch": 1.1336182865638358,
      "grad_norm": 0.7694109678268433,
      "learning_rate": 7.690152516006542e-05,
      "loss": 0.7243,
      "step": 6193
    },
    {
      "epoch": 1.133804125627207,
      "grad_norm": 0.7304584980010986,
      "learning_rate": 7.687307115175275e-05,
      "loss": 1.1268,
      "step": 6194
    },
    {
      "epoch": 1.133989964690578,
      "grad_norm": 0.6806715130805969,
      "learning_rate": 7.684461912153629e-05,
      "loss": 1.0232,
      "step": 6195
    },
    {
      "epoch": 1.134175803753949,
      "grad_norm": 0.7814947962760925,
      "learning_rate": 7.681616907184964e-05,
      "loss": 1.0942,
      "step": 6196
    },
    {
      "epoch": 1.1343616428173202,
      "grad_norm": 0.6447503566741943,
      "learning_rate": 7.678772100512611e-05,
      "loss": 1.0311,
      "step": 6197
    },
    {
      "epoch": 1.1345474818806913,
      "grad_norm": 0.5829337239265442,
      "learning_rate": 7.675927492379898e-05,
      "loss": 0.9055,
      "step": 6198
    },
    {
      "epoch": 1.1347333209440624,
      "grad_norm": 0.6542435884475708,
      "learning_rate": 7.673083083030133e-05,
      "loss": 1.0454,
      "step": 6199
    },
    {
      "epoch": 1.1349191600074335,
      "grad_norm": 0.5979851484298706,
      "learning_rate": 7.670238872706597e-05,
      "loss": 1.196,
      "step": 6200
    },
    {
      "epoch": 1.1351049990708046,
      "grad_norm": 0.7708069682121277,
      "learning_rate": 7.667394861652567e-05,
      "loss": 1.039,
      "step": 6201
    },
    {
      "epoch": 1.1352908381341758,
      "grad_norm": 0.8553258180618286,
      "learning_rate": 7.664551050111298e-05,
      "loss": 1.1342,
      "step": 6202
    },
    {
      "epoch": 1.1354766771975469,
      "grad_norm": 0.8040032386779785,
      "learning_rate": 7.661707438326023e-05,
      "loss": 1.1618,
      "step": 6203
    },
    {
      "epoch": 1.135662516260918,
      "grad_norm": 0.6485744118690491,
      "learning_rate": 7.658864026539962e-05,
      "loss": 1.0778,
      "step": 6204
    },
    {
      "epoch": 1.135848355324289,
      "grad_norm": 0.6490263342857361,
      "learning_rate": 7.656020814996323e-05,
      "loss": 0.8483,
      "step": 6205
    },
    {
      "epoch": 1.1360341943876602,
      "grad_norm": 0.6252707839012146,
      "learning_rate": 7.653177803938292e-05,
      "loss": 0.9277,
      "step": 6206
    },
    {
      "epoch": 1.1362200334510315,
      "grad_norm": 0.7012428045272827,
      "learning_rate": 7.650334993609031e-05,
      "loss": 0.8191,
      "step": 6207
    },
    {
      "epoch": 1.1364058725144026,
      "grad_norm": 0.7016780376434326,
      "learning_rate": 7.647492384251703e-05,
      "loss": 0.9852,
      "step": 6208
    },
    {
      "epoch": 1.1365917115777737,
      "grad_norm": 0.5683523416519165,
      "learning_rate": 7.64464997610943e-05,
      "loss": 0.8534,
      "step": 6209
    },
    {
      "epoch": 1.1367775506411448,
      "grad_norm": 0.8635490536689758,
      "learning_rate": 7.641807769425337e-05,
      "loss": 1.1012,
      "step": 6210
    },
    {
      "epoch": 1.136963389704516,
      "grad_norm": 0.5886553525924683,
      "learning_rate": 7.638965764442521e-05,
      "loss": 0.9735,
      "step": 6211
    },
    {
      "epoch": 1.137149228767887,
      "grad_norm": 0.7273619174957275,
      "learning_rate": 7.636123961404065e-05,
      "loss": 0.9081,
      "step": 6212
    },
    {
      "epoch": 1.1373350678312582,
      "grad_norm": 0.5543389320373535,
      "learning_rate": 7.633282360553041e-05,
      "loss": 0.747,
      "step": 6213
    },
    {
      "epoch": 1.1375209068946293,
      "grad_norm": 0.6617942452430725,
      "learning_rate": 7.630440962132486e-05,
      "loss": 0.8867,
      "step": 6214
    },
    {
      "epoch": 1.1377067459580004,
      "grad_norm": 0.5964873433113098,
      "learning_rate": 7.627599766385443e-05,
      "loss": 0.9283,
      "step": 6215
    },
    {
      "epoch": 1.1378925850213715,
      "grad_norm": 0.6347508430480957,
      "learning_rate": 7.624758773554914e-05,
      "loss": 0.8036,
      "step": 6216
    },
    {
      "epoch": 1.1380784240847426,
      "grad_norm": 0.6257861852645874,
      "learning_rate": 7.621917983883903e-05,
      "loss": 0.8181,
      "step": 6217
    },
    {
      "epoch": 1.1382642631481137,
      "grad_norm": 0.6525983810424805,
      "learning_rate": 7.619077397615384e-05,
      "loss": 0.9738,
      "step": 6218
    },
    {
      "epoch": 1.1384501022114848,
      "grad_norm": 0.6202676296234131,
      "learning_rate": 7.616237014992322e-05,
      "loss": 1.0333,
      "step": 6219
    },
    {
      "epoch": 1.138635941274856,
      "grad_norm": 0.8427987694740295,
      "learning_rate": 7.613396836257663e-05,
      "loss": 1.0282,
      "step": 6220
    },
    {
      "epoch": 1.138821780338227,
      "grad_norm": 0.7445005774497986,
      "learning_rate": 7.610556861654327e-05,
      "loss": 1.2364,
      "step": 6221
    },
    {
      "epoch": 1.1390076194015983,
      "grad_norm": 0.5864232182502747,
      "learning_rate": 7.607717091425233e-05,
      "loss": 0.9195,
      "step": 6222
    },
    {
      "epoch": 1.1391934584649692,
      "grad_norm": 0.5960657596588135,
      "learning_rate": 7.604877525813261e-05,
      "loss": 0.6124,
      "step": 6223
    },
    {
      "epoch": 1.1393792975283406,
      "grad_norm": 0.8077784776687622,
      "learning_rate": 7.602038165061292e-05,
      "loss": 0.8756,
      "step": 6224
    },
    {
      "epoch": 1.1395651365917117,
      "grad_norm": 0.6285901665687561,
      "learning_rate": 7.59919900941218e-05,
      "loss": 0.9791,
      "step": 6225
    },
    {
      "epoch": 1.1397509756550828,
      "grad_norm": 0.5540950298309326,
      "learning_rate": 7.596360059108765e-05,
      "loss": 0.6467,
      "step": 6226
    },
    {
      "epoch": 1.1399368147184539,
      "grad_norm": 0.6270549297332764,
      "learning_rate": 7.593521314393875e-05,
      "loss": 1.1263,
      "step": 6227
    },
    {
      "epoch": 1.140122653781825,
      "grad_norm": 0.6695981621742249,
      "learning_rate": 7.590682775510302e-05,
      "loss": 0.8308,
      "step": 6228
    },
    {
      "epoch": 1.140308492845196,
      "grad_norm": 0.7288331389427185,
      "learning_rate": 7.587844442700843e-05,
      "loss": 0.9318,
      "step": 6229
    },
    {
      "epoch": 1.1404943319085672,
      "grad_norm": 0.657455325126648,
      "learning_rate": 7.585006316208258e-05,
      "loss": 0.9138,
      "step": 6230
    },
    {
      "epoch": 1.1406801709719383,
      "grad_norm": 0.7142748832702637,
      "learning_rate": 7.582168396275304e-05,
      "loss": 1.02,
      "step": 6231
    },
    {
      "epoch": 1.1408660100353094,
      "grad_norm": 0.5973772406578064,
      "learning_rate": 7.579330683144712e-05,
      "loss": 1.0011,
      "step": 6232
    },
    {
      "epoch": 1.1410518490986805,
      "grad_norm": 0.6535288095474243,
      "learning_rate": 7.576493177059197e-05,
      "loss": 1.1361,
      "step": 6233
    },
    {
      "epoch": 1.1412376881620516,
      "grad_norm": 0.7035378217697144,
      "learning_rate": 7.573655878261464e-05,
      "loss": 0.964,
      "step": 6234
    },
    {
      "epoch": 1.1414235272254227,
      "grad_norm": 0.7315672636032104,
      "learning_rate": 7.570818786994183e-05,
      "loss": 1.0567,
      "step": 6235
    },
    {
      "epoch": 1.1416093662887938,
      "grad_norm": 0.7773432731628418,
      "learning_rate": 7.567981903500025e-05,
      "loss": 1.069,
      "step": 6236
    },
    {
      "epoch": 1.141795205352165,
      "grad_norm": 0.6205809116363525,
      "learning_rate": 7.565145228021627e-05,
      "loss": 0.7578,
      "step": 6237
    },
    {
      "epoch": 1.141981044415536,
      "grad_norm": 0.6082891225814819,
      "learning_rate": 7.562308760801622e-05,
      "loss": 0.6793,
      "step": 6238
    },
    {
      "epoch": 1.1421668834789074,
      "grad_norm": 0.6637825965881348,
      "learning_rate": 7.559472502082617e-05,
      "loss": 0.8839,
      "step": 6239
    },
    {
      "epoch": 1.1423527225422783,
      "grad_norm": 0.7240639925003052,
      "learning_rate": 7.5566364521072e-05,
      "loss": 1.0576,
      "step": 6240
    },
    {
      "epoch": 1.1425385616056496,
      "grad_norm": 0.6650069355964661,
      "learning_rate": 7.553800611117956e-05,
      "loss": 0.8786,
      "step": 6241
    },
    {
      "epoch": 1.1427244006690207,
      "grad_norm": 0.7037933468818665,
      "learning_rate": 7.550964979357427e-05,
      "loss": 1.0313,
      "step": 6242
    },
    {
      "epoch": 1.1429102397323918,
      "grad_norm": 0.6801198720932007,
      "learning_rate": 7.54812955706816e-05,
      "loss": 1.0486,
      "step": 6243
    },
    {
      "epoch": 1.143096078795763,
      "grad_norm": 0.6622380018234253,
      "learning_rate": 7.545294344492667e-05,
      "loss": 0.9577,
      "step": 6244
    },
    {
      "epoch": 1.143281917859134,
      "grad_norm": 0.8831017017364502,
      "learning_rate": 7.542459341873456e-05,
      "loss": 1.0314,
      "step": 6245
    },
    {
      "epoch": 1.1434677569225051,
      "grad_norm": 0.6996762752532959,
      "learning_rate": 7.539624549453008e-05,
      "loss": 0.9683,
      "step": 6246
    },
    {
      "epoch": 1.1436535959858762,
      "grad_norm": 0.7951368093490601,
      "learning_rate": 7.53678996747379e-05,
      "loss": 1.0881,
      "step": 6247
    },
    {
      "epoch": 1.1438394350492473,
      "grad_norm": 0.5819055438041687,
      "learning_rate": 7.533955596178252e-05,
      "loss": 1.0429,
      "step": 6248
    },
    {
      "epoch": 1.1440252741126185,
      "grad_norm": 0.654751718044281,
      "learning_rate": 7.531121435808819e-05,
      "loss": 1.0643,
      "step": 6249
    },
    {
      "epoch": 1.1442111131759896,
      "grad_norm": 0.6705732345581055,
      "learning_rate": 7.528287486607909e-05,
      "loss": 1.1292,
      "step": 6250
    },
    {
      "epoch": 1.1443969522393607,
      "grad_norm": 0.5722493529319763,
      "learning_rate": 7.525453748817908e-05,
      "loss": 0.9427,
      "step": 6251
    },
    {
      "epoch": 1.1445827913027318,
      "grad_norm": 0.660951554775238,
      "learning_rate": 7.5226202226812e-05,
      "loss": 0.9269,
      "step": 6252
    },
    {
      "epoch": 1.1447686303661029,
      "grad_norm": 0.7391307950019836,
      "learning_rate": 7.519786908440136e-05,
      "loss": 1.01,
      "step": 6253
    },
    {
      "epoch": 1.144954469429474,
      "grad_norm": 0.6185791492462158,
      "learning_rate": 7.516953806337058e-05,
      "loss": 0.9853,
      "step": 6254
    },
    {
      "epoch": 1.145140308492845,
      "grad_norm": 0.7332004904747009,
      "learning_rate": 7.514120916614289e-05,
      "loss": 1.1023,
      "step": 6255
    },
    {
      "epoch": 1.1453261475562164,
      "grad_norm": 0.5954209566116333,
      "learning_rate": 7.511288239514128e-05,
      "loss": 0.9286,
      "step": 6256
    },
    {
      "epoch": 1.1455119866195875,
      "grad_norm": 0.6706781983375549,
      "learning_rate": 7.508455775278867e-05,
      "loss": 1.0266,
      "step": 6257
    },
    {
      "epoch": 1.1456978256829586,
      "grad_norm": 0.6400855779647827,
      "learning_rate": 7.505623524150765e-05,
      "loss": 1.0605,
      "step": 6258
    },
    {
      "epoch": 1.1458836647463297,
      "grad_norm": 0.6486907005310059,
      "learning_rate": 7.502791486372079e-05,
      "loss": 0.8922,
      "step": 6259
    },
    {
      "epoch": 1.1460695038097009,
      "grad_norm": 0.7294444441795349,
      "learning_rate": 7.499959662185029e-05,
      "loss": 1.0198,
      "step": 6260
    },
    {
      "epoch": 1.146255342873072,
      "grad_norm": 0.5519627332687378,
      "learning_rate": 7.497128051831832e-05,
      "loss": 0.5194,
      "step": 6261
    },
    {
      "epoch": 1.146441181936443,
      "grad_norm": 0.7215742468833923,
      "learning_rate": 7.494296655554685e-05,
      "loss": 0.8087,
      "step": 6262
    },
    {
      "epoch": 1.1466270209998142,
      "grad_norm": 0.6595014929771423,
      "learning_rate": 7.491465473595758e-05,
      "loss": 1.1404,
      "step": 6263
    },
    {
      "epoch": 1.1468128600631853,
      "grad_norm": 0.6985663771629333,
      "learning_rate": 7.488634506197216e-05,
      "loss": 0.8644,
      "step": 6264
    },
    {
      "epoch": 1.1469986991265564,
      "grad_norm": 0.744734525680542,
      "learning_rate": 7.485803753601188e-05,
      "loss": 0.819,
      "step": 6265
    },
    {
      "epoch": 1.1471845381899275,
      "grad_norm": 0.7613779902458191,
      "learning_rate": 7.482973216049803e-05,
      "loss": 0.9787,
      "step": 6266
    },
    {
      "epoch": 1.1473703772532986,
      "grad_norm": 0.8032128810882568,
      "learning_rate": 7.480142893785152e-05,
      "loss": 0.9408,
      "step": 6267
    },
    {
      "epoch": 1.1475562163166697,
      "grad_norm": 0.6917648911476135,
      "learning_rate": 7.477312787049328e-05,
      "loss": 0.9462,
      "step": 6268
    },
    {
      "epoch": 1.1477420553800408,
      "grad_norm": 0.7514396905899048,
      "learning_rate": 7.474482896084397e-05,
      "loss": 1.0154,
      "step": 6269
    },
    {
      "epoch": 1.147927894443412,
      "grad_norm": 0.6715410351753235,
      "learning_rate": 7.471653221132397e-05,
      "loss": 0.7154,
      "step": 6270
    },
    {
      "epoch": 1.148113733506783,
      "grad_norm": 0.6557510495185852,
      "learning_rate": 7.468823762435367e-05,
      "loss": 0.81,
      "step": 6271
    },
    {
      "epoch": 1.1482995725701541,
      "grad_norm": 0.7080697417259216,
      "learning_rate": 7.465994520235306e-05,
      "loss": 0.914,
      "step": 6272
    },
    {
      "epoch": 1.1484854116335255,
      "grad_norm": 0.6797797083854675,
      "learning_rate": 7.463165494774216e-05,
      "loss": 0.9553,
      "step": 6273
    },
    {
      "epoch": 1.1486712506968966,
      "grad_norm": 0.5941304564476013,
      "learning_rate": 7.460336686294061e-05,
      "loss": 1.0521,
      "step": 6274
    },
    {
      "epoch": 1.1488570897602677,
      "grad_norm": 0.7724999189376831,
      "learning_rate": 7.457508095036796e-05,
      "loss": 1.052,
      "step": 6275
    },
    {
      "epoch": 1.1490429288236388,
      "grad_norm": 0.6262985467910767,
      "learning_rate": 7.45467972124436e-05,
      "loss": 0.892,
      "step": 6276
    },
    {
      "epoch": 1.14922876788701,
      "grad_norm": 0.6632989645004272,
      "learning_rate": 7.451851565158669e-05,
      "loss": 0.8847,
      "step": 6277
    },
    {
      "epoch": 1.149414606950381,
      "grad_norm": 0.6198441386222839,
      "learning_rate": 7.449023627021624e-05,
      "loss": 0.9052,
      "step": 6278
    },
    {
      "epoch": 1.149600446013752,
      "grad_norm": 0.6501045823097229,
      "learning_rate": 7.446195907075098e-05,
      "loss": 1.0629,
      "step": 6279
    },
    {
      "epoch": 1.1497862850771232,
      "grad_norm": 0.6493845582008362,
      "learning_rate": 7.443368405560958e-05,
      "loss": 0.6935,
      "step": 6280
    },
    {
      "epoch": 1.1499721241404943,
      "grad_norm": 0.6532225012779236,
      "learning_rate": 7.440541122721041e-05,
      "loss": 0.923,
      "step": 6281
    },
    {
      "epoch": 1.1501579632038654,
      "grad_norm": 0.6956079006195068,
      "learning_rate": 7.437714058797173e-05,
      "loss": 0.8909,
      "step": 6282
    },
    {
      "epoch": 1.1503438022672365,
      "grad_norm": 0.6475423574447632,
      "learning_rate": 7.434887214031163e-05,
      "loss": 0.7268,
      "step": 6283
    },
    {
      "epoch": 1.1505296413306076,
      "grad_norm": 0.6966208815574646,
      "learning_rate": 7.43206058866479e-05,
      "loss": 0.9642,
      "step": 6284
    },
    {
      "epoch": 1.1507154803939788,
      "grad_norm": 0.7085195183753967,
      "learning_rate": 7.42923418293983e-05,
      "loss": 0.8918,
      "step": 6285
    },
    {
      "epoch": 1.1509013194573499,
      "grad_norm": 0.6598019003868103,
      "learning_rate": 7.42640799709802e-05,
      "loss": 0.8213,
      "step": 6286
    },
    {
      "epoch": 1.151087158520721,
      "grad_norm": 0.6603427529335022,
      "learning_rate": 7.423582031381102e-05,
      "loss": 0.9549,
      "step": 6287
    },
    {
      "epoch": 1.1512729975840923,
      "grad_norm": 0.7119370698928833,
      "learning_rate": 7.420756286030778e-05,
      "loss": 1.2205,
      "step": 6288
    },
    {
      "epoch": 1.1514588366474632,
      "grad_norm": 0.7540510296821594,
      "learning_rate": 7.417930761288743e-05,
      "loss": 0.9597,
      "step": 6289
    },
    {
      "epoch": 1.1516446757108345,
      "grad_norm": 0.6168947219848633,
      "learning_rate": 7.415105457396671e-05,
      "loss": 0.8439,
      "step": 6290
    },
    {
      "epoch": 1.1518305147742056,
      "grad_norm": 0.7342690825462341,
      "learning_rate": 7.412280374596215e-05,
      "loss": 0.9151,
      "step": 6291
    },
    {
      "epoch": 1.1520163538375767,
      "grad_norm": 0.6422035694122314,
      "learning_rate": 7.409455513129015e-05,
      "loss": 0.933,
      "step": 6292
    },
    {
      "epoch": 1.1522021929009478,
      "grad_norm": 0.5854395627975464,
      "learning_rate": 7.406630873236681e-05,
      "loss": 0.9118,
      "step": 6293
    },
    {
      "epoch": 1.152388031964319,
      "grad_norm": 0.7802146077156067,
      "learning_rate": 7.403806455160817e-05,
      "loss": 0.8353,
      "step": 6294
    },
    {
      "epoch": 1.15257387102769,
      "grad_norm": 0.7460753917694092,
      "learning_rate": 7.400982259142994e-05,
      "loss": 0.8547,
      "step": 6295
    },
    {
      "epoch": 1.1527597100910612,
      "grad_norm": 0.7687903642654419,
      "learning_rate": 7.398158285424774e-05,
      "loss": 0.9714,
      "step": 6296
    },
    {
      "epoch": 1.1529455491544323,
      "grad_norm": 0.6683840155601501,
      "learning_rate": 7.395334534247703e-05,
      "loss": 0.9438,
      "step": 6297
    },
    {
      "epoch": 1.1531313882178034,
      "grad_norm": 0.6488221883773804,
      "learning_rate": 7.392511005853297e-05,
      "loss": 1.0556,
      "step": 6298
    },
    {
      "epoch": 1.1533172272811745,
      "grad_norm": 0.6396840810775757,
      "learning_rate": 7.389687700483064e-05,
      "loss": 0.9994,
      "step": 6299
    },
    {
      "epoch": 1.1535030663445456,
      "grad_norm": 0.693667471408844,
      "learning_rate": 7.386864618378478e-05,
      "loss": 1.134,
      "step": 6300
    },
    {
      "epoch": 1.1536889054079167,
      "grad_norm": 0.6322534084320068,
      "learning_rate": 7.384041759781013e-05,
      "loss": 0.8274,
      "step": 6301
    },
    {
      "epoch": 1.1538747444712878,
      "grad_norm": 0.840552806854248,
      "learning_rate": 7.381219124932107e-05,
      "loss": 0.9961,
      "step": 6302
    },
    {
      "epoch": 1.154060583534659,
      "grad_norm": 0.7284782528877258,
      "learning_rate": 7.37839671407319e-05,
      "loss": 0.9661,
      "step": 6303
    },
    {
      "epoch": 1.15424642259803,
      "grad_norm": 0.681910514831543,
      "learning_rate": 7.37557452744567e-05,
      "loss": 1.0821,
      "step": 6304
    },
    {
      "epoch": 1.1544322616614013,
      "grad_norm": 0.6733909249305725,
      "learning_rate": 7.37275256529093e-05,
      "loss": 0.9578,
      "step": 6305
    },
    {
      "epoch": 1.1546181007247722,
      "grad_norm": 0.6376176476478577,
      "learning_rate": 7.369930827850345e-05,
      "loss": 0.9386,
      "step": 6306
    },
    {
      "epoch": 1.1548039397881436,
      "grad_norm": 0.6968693733215332,
      "learning_rate": 7.367109315365256e-05,
      "loss": 1.0013,
      "step": 6307
    },
    {
      "epoch": 1.1549897788515147,
      "grad_norm": 0.6921547651290894,
      "learning_rate": 7.364288028077003e-05,
      "loss": 0.8702,
      "step": 6308
    },
    {
      "epoch": 1.1551756179148858,
      "grad_norm": 0.7634596824645996,
      "learning_rate": 7.361466966226886e-05,
      "loss": 0.9821,
      "step": 6309
    },
    {
      "epoch": 1.1553614569782569,
      "grad_norm": 0.594638466835022,
      "learning_rate": 7.358646130056204e-05,
      "loss": 0.9611,
      "step": 6310
    },
    {
      "epoch": 1.155547296041628,
      "grad_norm": 0.6549757122993469,
      "learning_rate": 7.35582551980623e-05,
      "loss": 0.9084,
      "step": 6311
    },
    {
      "epoch": 1.155733135104999,
      "grad_norm": 0.6684489250183105,
      "learning_rate": 7.35300513571821e-05,
      "loss": 1.0863,
      "step": 6312
    },
    {
      "epoch": 1.1559189741683702,
      "grad_norm": 0.6731480956077576,
      "learning_rate": 7.350184978033386e-05,
      "loss": 0.9022,
      "step": 6313
    },
    {
      "epoch": 1.1561048132317413,
      "grad_norm": 0.6278926730155945,
      "learning_rate": 7.347365046992962e-05,
      "loss": 1.1561,
      "step": 6314
    },
    {
      "epoch": 1.1562906522951124,
      "grad_norm": 1.9784194231033325,
      "learning_rate": 7.344545342838146e-05,
      "loss": 1.6575,
      "step": 6315
    },
    {
      "epoch": 1.1564764913584835,
      "grad_norm": 0.6113159656524658,
      "learning_rate": 7.341725865810099e-05,
      "loss": 0.7758,
      "step": 6316
    },
    {
      "epoch": 1.1566623304218546,
      "grad_norm": 0.7391144633293152,
      "learning_rate": 7.338906616149986e-05,
      "loss": 1.129,
      "step": 6317
    },
    {
      "epoch": 1.1568481694852257,
      "grad_norm": 0.6916064023971558,
      "learning_rate": 7.336087594098946e-05,
      "loss": 0.9741,
      "step": 6318
    },
    {
      "epoch": 1.1570340085485968,
      "grad_norm": 0.6826386451721191,
      "learning_rate": 7.333268799898085e-05,
      "loss": 1.2518,
      "step": 6319
    },
    {
      "epoch": 1.157219847611968,
      "grad_norm": 0.6638817191123962,
      "learning_rate": 7.330450233788512e-05,
      "loss": 0.9266,
      "step": 6320
    },
    {
      "epoch": 1.157405686675339,
      "grad_norm": 0.7738638520240784,
      "learning_rate": 7.327631896011297e-05,
      "loss": 1.1382,
      "step": 6321
    },
    {
      "epoch": 1.1575915257387104,
      "grad_norm": 0.7370550036430359,
      "learning_rate": 7.324813786807506e-05,
      "loss": 1.0067,
      "step": 6322
    },
    {
      "epoch": 1.1577773648020815,
      "grad_norm": 0.625015377998352,
      "learning_rate": 7.321995906418169e-05,
      "loss": 1.0133,
      "step": 6323
    },
    {
      "epoch": 1.1579632038654526,
      "grad_norm": 0.5457345247268677,
      "learning_rate": 7.31917825508431e-05,
      "loss": 0.6307,
      "step": 6324
    },
    {
      "epoch": 1.1581490429288237,
      "grad_norm": 0.7099806070327759,
      "learning_rate": 7.316360833046934e-05,
      "loss": 1.1707,
      "step": 6325
    },
    {
      "epoch": 1.1583348819921948,
      "grad_norm": 0.7791039347648621,
      "learning_rate": 7.31354364054701e-05,
      "loss": 0.9305,
      "step": 6326
    },
    {
      "epoch": 1.158520721055566,
      "grad_norm": 0.6316432356834412,
      "learning_rate": 7.310726677825508e-05,
      "loss": 0.9449,
      "step": 6327
    },
    {
      "epoch": 1.158706560118937,
      "grad_norm": 1.1281682252883911,
      "learning_rate": 7.307909945123363e-05,
      "loss": 1.1191,
      "step": 6328
    },
    {
      "epoch": 1.1588923991823081,
      "grad_norm": 0.7638713717460632,
      "learning_rate": 7.305093442681503e-05,
      "loss": 1.0541,
      "step": 6329
    },
    {
      "epoch": 1.1590782382456792,
      "grad_norm": 0.554898202419281,
      "learning_rate": 7.302277170740821e-05,
      "loss": 0.8519,
      "step": 6330
    },
    {
      "epoch": 1.1592640773090503,
      "grad_norm": 0.6649084687232971,
      "learning_rate": 7.299461129542205e-05,
      "loss": 0.8424,
      "step": 6331
    },
    {
      "epoch": 1.1594499163724215,
      "grad_norm": 0.5770688652992249,
      "learning_rate": 7.296645319326516e-05,
      "loss": 0.9126,
      "step": 6332
    },
    {
      "epoch": 1.1596357554357926,
      "grad_norm": 0.6516635417938232,
      "learning_rate": 7.293829740334594e-05,
      "loss": 0.8695,
      "step": 6333
    },
    {
      "epoch": 1.1598215944991637,
      "grad_norm": 0.5726830959320068,
      "learning_rate": 7.291014392807264e-05,
      "loss": 0.891,
      "step": 6334
    },
    {
      "epoch": 1.1600074335625348,
      "grad_norm": 0.7533463835716248,
      "learning_rate": 7.288199276985327e-05,
      "loss": 1.1437,
      "step": 6335
    },
    {
      "epoch": 1.1601932726259059,
      "grad_norm": 0.6398054957389832,
      "learning_rate": 7.28538439310957e-05,
      "loss": 0.7641,
      "step": 6336
    },
    {
      "epoch": 1.1603791116892772,
      "grad_norm": 0.8049401640892029,
      "learning_rate": 7.282569741420749e-05,
      "loss": 0.8801,
      "step": 6337
    },
    {
      "epoch": 1.160564950752648,
      "grad_norm": 0.6398137211799622,
      "learning_rate": 7.279755322159611e-05,
      "loss": 0.9682,
      "step": 6338
    },
    {
      "epoch": 1.1607507898160194,
      "grad_norm": 0.6790042519569397,
      "learning_rate": 7.276941135566884e-05,
      "loss": 1.0039,
      "step": 6339
    },
    {
      "epoch": 1.1609366288793905,
      "grad_norm": 0.7095458507537842,
      "learning_rate": 7.274127181883262e-05,
      "loss": 0.756,
      "step": 6340
    },
    {
      "epoch": 1.1611224679427616,
      "grad_norm": 0.7210416197776794,
      "learning_rate": 7.271313461349436e-05,
      "loss": 0.7686,
      "step": 6341
    },
    {
      "epoch": 1.1613083070061327,
      "grad_norm": 0.6349940299987793,
      "learning_rate": 7.268499974206067e-05,
      "loss": 0.4654,
      "step": 6342
    },
    {
      "epoch": 1.1614941460695039,
      "grad_norm": 0.5711683034896851,
      "learning_rate": 7.2656867206938e-05,
      "loss": 0.6471,
      "step": 6343
    },
    {
      "epoch": 1.161679985132875,
      "grad_norm": 0.7057211399078369,
      "learning_rate": 7.262873701053254e-05,
      "loss": 1.0344,
      "step": 6344
    },
    {
      "epoch": 1.161865824196246,
      "grad_norm": 0.8090634942054749,
      "learning_rate": 7.260060915525037e-05,
      "loss": 1.065,
      "step": 6345
    },
    {
      "epoch": 1.1620516632596172,
      "grad_norm": 0.7852050065994263,
      "learning_rate": 7.257248364349736e-05,
      "loss": 0.9761,
      "step": 6346
    },
    {
      "epoch": 1.1622375023229883,
      "grad_norm": 1.0304369926452637,
      "learning_rate": 7.254436047767905e-05,
      "loss": 1.071,
      "step": 6347
    },
    {
      "epoch": 1.1624233413863594,
      "grad_norm": 0.7340937256813049,
      "learning_rate": 7.251623966020095e-05,
      "loss": 1.0406,
      "step": 6348
    },
    {
      "epoch": 1.1626091804497305,
      "grad_norm": 0.6150162220001221,
      "learning_rate": 7.248812119346827e-05,
      "loss": 0.9738,
      "step": 6349
    },
    {
      "epoch": 1.1627950195131016,
      "grad_norm": 0.6172744631767273,
      "learning_rate": 7.246000507988606e-05,
      "loss": 0.7907,
      "step": 6350
    },
    {
      "epoch": 1.1629808585764727,
      "grad_norm": 0.6725603938102722,
      "learning_rate": 7.24318913218591e-05,
      "loss": 0.8283,
      "step": 6351
    },
    {
      "epoch": 1.1631666976398438,
      "grad_norm": 0.6481371521949768,
      "learning_rate": 7.240377992179206e-05,
      "loss": 0.7308,
      "step": 6352
    },
    {
      "epoch": 1.163352536703215,
      "grad_norm": 0.574016273021698,
      "learning_rate": 7.237567088208942e-05,
      "loss": 0.9847,
      "step": 6353
    },
    {
      "epoch": 1.1635383757665863,
      "grad_norm": 0.6835063695907593,
      "learning_rate": 7.234756420515531e-05,
      "loss": 0.8516,
      "step": 6354
    },
    {
      "epoch": 1.1637242148299571,
      "grad_norm": 0.6407470107078552,
      "learning_rate": 7.231945989339383e-05,
      "loss": 0.6269,
      "step": 6355
    },
    {
      "epoch": 1.1639100538933285,
      "grad_norm": 0.7623193264007568,
      "learning_rate": 7.229135794920874e-05,
      "loss": 1.1166,
      "step": 6356
    },
    {
      "epoch": 1.1640958929566996,
      "grad_norm": 0.6505951881408691,
      "learning_rate": 7.226325837500369e-05,
      "loss": 0.8555,
      "step": 6357
    },
    {
      "epoch": 1.1642817320200707,
      "grad_norm": 0.7478532195091248,
      "learning_rate": 7.22351611731821e-05,
      "loss": 1.0839,
      "step": 6358
    },
    {
      "epoch": 1.1644675710834418,
      "grad_norm": 0.6620163321495056,
      "learning_rate": 7.220706634614717e-05,
      "loss": 0.8757,
      "step": 6359
    },
    {
      "epoch": 1.164653410146813,
      "grad_norm": 0.6156098246574402,
      "learning_rate": 7.217897389630197e-05,
      "loss": 0.9406,
      "step": 6360
    },
    {
      "epoch": 1.164839249210184,
      "grad_norm": 0.7431986927986145,
      "learning_rate": 7.215088382604922e-05,
      "loss": 0.9566,
      "step": 6361
    },
    {
      "epoch": 1.165025088273555,
      "grad_norm": 0.797284722328186,
      "learning_rate": 7.212279613779161e-05,
      "loss": 0.8231,
      "step": 6362
    },
    {
      "epoch": 1.1652109273369262,
      "grad_norm": 0.6188994646072388,
      "learning_rate": 7.209471083393146e-05,
      "loss": 0.732,
      "step": 6363
    },
    {
      "epoch": 1.1653967664002973,
      "grad_norm": 0.7715564370155334,
      "learning_rate": 7.206662791687102e-05,
      "loss": 1.188,
      "step": 6364
    },
    {
      "epoch": 1.1655826054636684,
      "grad_norm": 0.7624145746231079,
      "learning_rate": 7.203854738901223e-05,
      "loss": 1.0448,
      "step": 6365
    },
    {
      "epoch": 1.1657684445270395,
      "grad_norm": 0.7245141863822937,
      "learning_rate": 7.201046925275694e-05,
      "loss": 0.9017,
      "step": 6366
    },
    {
      "epoch": 1.1659542835904106,
      "grad_norm": 0.7528111934661865,
      "learning_rate": 7.198239351050673e-05,
      "loss": 0.7053,
      "step": 6367
    },
    {
      "epoch": 1.1661401226537818,
      "grad_norm": 0.6484794616699219,
      "learning_rate": 7.195432016466293e-05,
      "loss": 0.895,
      "step": 6368
    },
    {
      "epoch": 1.1663259617171529,
      "grad_norm": 1.6364943981170654,
      "learning_rate": 7.192624921762679e-05,
      "loss": 1.6132,
      "step": 6369
    },
    {
      "epoch": 1.166511800780524,
      "grad_norm": 0.9956699013710022,
      "learning_rate": 7.189818067179916e-05,
      "loss": 0.9769,
      "step": 6370
    },
    {
      "epoch": 1.1666976398438953,
      "grad_norm": 0.7614436745643616,
      "learning_rate": 7.187011452958092e-05,
      "loss": 0.7649,
      "step": 6371
    },
    {
      "epoch": 1.1668834789072664,
      "grad_norm": 0.6507315039634705,
      "learning_rate": 7.184205079337254e-05,
      "loss": 0.9658,
      "step": 6372
    },
    {
      "epoch": 1.1670693179706375,
      "grad_norm": 0.8169115781784058,
      "learning_rate": 7.181398946557443e-05,
      "loss": 1.0792,
      "step": 6373
    },
    {
      "epoch": 1.1672551570340086,
      "grad_norm": 0.5518690943717957,
      "learning_rate": 7.178593054858675e-05,
      "loss": 0.637,
      "step": 6374
    },
    {
      "epoch": 1.1674409960973797,
      "grad_norm": 0.6700672507286072,
      "learning_rate": 7.175787404480936e-05,
      "loss": 1.0321,
      "step": 6375
    },
    {
      "epoch": 1.1676268351607508,
      "grad_norm": 0.5618696808815002,
      "learning_rate": 7.17298199566421e-05,
      "loss": 0.6381,
      "step": 6376
    },
    {
      "epoch": 1.167812674224122,
      "grad_norm": 0.6601108312606812,
      "learning_rate": 7.170176828648438e-05,
      "loss": 1.104,
      "step": 6377
    },
    {
      "epoch": 1.167998513287493,
      "grad_norm": 0.7242966294288635,
      "learning_rate": 7.16737190367356e-05,
      "loss": 0.8377,
      "step": 6378
    },
    {
      "epoch": 1.1681843523508642,
      "grad_norm": 0.6647973656654358,
      "learning_rate": 7.164567220979486e-05,
      "loss": 0.962,
      "step": 6379
    },
    {
      "epoch": 1.1683701914142353,
      "grad_norm": 0.8982508182525635,
      "learning_rate": 7.161762780806103e-05,
      "loss": 1.087,
      "step": 6380
    },
    {
      "epoch": 1.1685560304776064,
      "grad_norm": 0.5682864189147949,
      "learning_rate": 7.158958583393289e-05,
      "loss": 0.8412,
      "step": 6381
    },
    {
      "epoch": 1.1687418695409775,
      "grad_norm": 0.6696249842643738,
      "learning_rate": 7.156154628980883e-05,
      "loss": 0.7829,
      "step": 6382
    },
    {
      "epoch": 1.1689277086043486,
      "grad_norm": 0.8441759347915649,
      "learning_rate": 7.153350917808723e-05,
      "loss": 1.1044,
      "step": 6383
    },
    {
      "epoch": 1.1691135476677197,
      "grad_norm": 0.5896120667457581,
      "learning_rate": 7.150547450116606e-05,
      "loss": 0.8167,
      "step": 6384
    },
    {
      "epoch": 1.1692993867310908,
      "grad_norm": 0.6768901348114014,
      "learning_rate": 7.147744226144327e-05,
      "loss": 0.8811,
      "step": 6385
    },
    {
      "epoch": 1.169485225794462,
      "grad_norm": 0.6406312584877014,
      "learning_rate": 7.144941246131647e-05,
      "loss": 0.8857,
      "step": 6386
    },
    {
      "epoch": 1.169671064857833,
      "grad_norm": 0.562627911567688,
      "learning_rate": 7.142138510318311e-05,
      "loss": 0.786,
      "step": 6387
    },
    {
      "epoch": 1.1698569039212043,
      "grad_norm": 0.6905426383018494,
      "learning_rate": 7.13933601894405e-05,
      "loss": 0.9298,
      "step": 6388
    },
    {
      "epoch": 1.1700427429845754,
      "grad_norm": 0.6309052109718323,
      "learning_rate": 7.136533772248559e-05,
      "loss": 1.1465,
      "step": 6389
    },
    {
      "epoch": 1.1702285820479466,
      "grad_norm": 0.6853886842727661,
      "learning_rate": 7.133731770471526e-05,
      "loss": 0.7632,
      "step": 6390
    },
    {
      "epoch": 1.1704144211113177,
      "grad_norm": 0.6898705363273621,
      "learning_rate": 7.130930013852606e-05,
      "loss": 0.9551,
      "step": 6391
    },
    {
      "epoch": 1.1706002601746888,
      "grad_norm": 0.6434992551803589,
      "learning_rate": 7.128128502631445e-05,
      "loss": 1.015,
      "step": 6392
    },
    {
      "epoch": 1.1707860992380599,
      "grad_norm": 0.7179650068283081,
      "learning_rate": 7.125327237047657e-05,
      "loss": 0.9536,
      "step": 6393
    },
    {
      "epoch": 1.170971938301431,
      "grad_norm": 0.650525689125061,
      "learning_rate": 7.122526217340847e-05,
      "loss": 1.073,
      "step": 6394
    },
    {
      "epoch": 1.171157777364802,
      "grad_norm": 0.672072172164917,
      "learning_rate": 7.119725443750588e-05,
      "loss": 0.7899,
      "step": 6395
    },
    {
      "epoch": 1.1713436164281732,
      "grad_norm": 0.665378987789154,
      "learning_rate": 7.116924916516434e-05,
      "loss": 1.0183,
      "step": 6396
    },
    {
      "epoch": 1.1715294554915443,
      "grad_norm": 0.7998149991035461,
      "learning_rate": 7.114124635877929e-05,
      "loss": 1.1661,
      "step": 6397
    },
    {
      "epoch": 1.1717152945549154,
      "grad_norm": 0.740707516670227,
      "learning_rate": 7.111324602074576e-05,
      "loss": 1.0706,
      "step": 6398
    },
    {
      "epoch": 1.1719011336182865,
      "grad_norm": 0.7028396725654602,
      "learning_rate": 7.108524815345873e-05,
      "loss": 0.8416,
      "step": 6399
    },
    {
      "epoch": 1.1720869726816576,
      "grad_norm": 0.7408978343009949,
      "learning_rate": 7.105725275931296e-05,
      "loss": 0.9771,
      "step": 6400
    },
    {
      "epoch": 1.1722728117450287,
      "grad_norm": 0.6476548314094543,
      "learning_rate": 7.102925984070286e-05,
      "loss": 0.6686,
      "step": 6401
    },
    {
      "epoch": 1.1724586508083998,
      "grad_norm": 0.703307032585144,
      "learning_rate": 7.100126940002282e-05,
      "loss": 0.9585,
      "step": 6402
    },
    {
      "epoch": 1.1726444898717712,
      "grad_norm": 0.6147748827934265,
      "learning_rate": 7.097328143966685e-05,
      "loss": 0.8826,
      "step": 6403
    },
    {
      "epoch": 1.172830328935142,
      "grad_norm": 0.7942382097244263,
      "learning_rate": 7.09452959620289e-05,
      "loss": 1.0286,
      "step": 6404
    },
    {
      "epoch": 1.1730161679985134,
      "grad_norm": 0.6308249831199646,
      "learning_rate": 7.091731296950255e-05,
      "loss": 0.9641,
      "step": 6405
    },
    {
      "epoch": 1.1732020070618845,
      "grad_norm": 0.7863003015518188,
      "learning_rate": 7.088933246448131e-05,
      "loss": 1.1257,
      "step": 6406
    },
    {
      "epoch": 1.1733878461252556,
      "grad_norm": 0.5810206532478333,
      "learning_rate": 7.086135444935833e-05,
      "loss": 0.9111,
      "step": 6407
    },
    {
      "epoch": 1.1735736851886267,
      "grad_norm": 0.6318856477737427,
      "learning_rate": 7.083337892652669e-05,
      "loss": 0.8775,
      "step": 6408
    },
    {
      "epoch": 1.1737595242519978,
      "grad_norm": 0.711527407169342,
      "learning_rate": 7.080540589837921e-05,
      "loss": 1.0231,
      "step": 6409
    },
    {
      "epoch": 1.173945363315369,
      "grad_norm": 0.7220918536186218,
      "learning_rate": 7.077743536730842e-05,
      "loss": 0.7934,
      "step": 6410
    },
    {
      "epoch": 1.17413120237874,
      "grad_norm": 0.6480425596237183,
      "learning_rate": 7.074946733570682e-05,
      "loss": 0.7943,
      "step": 6411
    },
    {
      "epoch": 1.1743170414421111,
      "grad_norm": 0.6978838443756104,
      "learning_rate": 7.072150180596644e-05,
      "loss": 1.1222,
      "step": 6412
    },
    {
      "epoch": 1.1745028805054822,
      "grad_norm": 0.7103899717330933,
      "learning_rate": 7.069353878047933e-05,
      "loss": 0.8338,
      "step": 6413
    },
    {
      "epoch": 1.1746887195688533,
      "grad_norm": 0.5447167754173279,
      "learning_rate": 7.066557826163713e-05,
      "loss": 0.8314,
      "step": 6414
    },
    {
      "epoch": 1.1748745586322245,
      "grad_norm": 0.6523606777191162,
      "learning_rate": 7.063762025183144e-05,
      "loss": 0.9746,
      "step": 6415
    },
    {
      "epoch": 1.1750603976955956,
      "grad_norm": 0.7402204871177673,
      "learning_rate": 7.060966475345356e-05,
      "loss": 1.1999,
      "step": 6416
    },
    {
      "epoch": 1.1752462367589667,
      "grad_norm": 0.800520122051239,
      "learning_rate": 7.058171176889456e-05,
      "loss": 0.9486,
      "step": 6417
    },
    {
      "epoch": 1.1754320758223378,
      "grad_norm": 0.6214317083358765,
      "learning_rate": 7.055376130054539e-05,
      "loss": 0.6924,
      "step": 6418
    },
    {
      "epoch": 1.1756179148857089,
      "grad_norm": 0.6461222767829895,
      "learning_rate": 7.05258133507966e-05,
      "loss": 0.9349,
      "step": 6419
    },
    {
      "epoch": 1.1758037539490802,
      "grad_norm": 0.6513392329216003,
      "learning_rate": 7.049786792203874e-05,
      "loss": 0.8558,
      "step": 6420
    },
    {
      "epoch": 1.175989593012451,
      "grad_norm": 0.7250276803970337,
      "learning_rate": 7.046992501666195e-05,
      "loss": 0.9061,
      "step": 6421
    },
    {
      "epoch": 1.1761754320758224,
      "grad_norm": 0.8080319166183472,
      "learning_rate": 7.044198463705631e-05,
      "loss": 1.027,
      "step": 6422
    },
    {
      "epoch": 1.1763612711391935,
      "grad_norm": 0.5869935154914856,
      "learning_rate": 7.041404678561162e-05,
      "loss": 1.0799,
      "step": 6423
    },
    {
      "epoch": 1.1765471102025646,
      "grad_norm": 0.7930959463119507,
      "learning_rate": 7.038611146471743e-05,
      "loss": 1.2881,
      "step": 6424
    },
    {
      "epoch": 1.1767329492659357,
      "grad_norm": 0.6239271759986877,
      "learning_rate": 7.035817867676317e-05,
      "loss": 0.9041,
      "step": 6425
    },
    {
      "epoch": 1.1769187883293069,
      "grad_norm": 0.5888556241989136,
      "learning_rate": 7.033024842413792e-05,
      "loss": 0.9099,
      "step": 6426
    },
    {
      "epoch": 1.177104627392678,
      "grad_norm": 0.7566354274749756,
      "learning_rate": 7.030232070923069e-05,
      "loss": 0.9611,
      "step": 6427
    },
    {
      "epoch": 1.177290466456049,
      "grad_norm": 0.7134408950805664,
      "learning_rate": 7.02743955344301e-05,
      "loss": 0.9586,
      "step": 6428
    },
    {
      "epoch": 1.1774763055194202,
      "grad_norm": 0.7655895352363586,
      "learning_rate": 7.024647290212472e-05,
      "loss": 0.9405,
      "step": 6429
    },
    {
      "epoch": 1.1776621445827913,
      "grad_norm": 0.7383896112442017,
      "learning_rate": 7.021855281470283e-05,
      "loss": 1.2127,
      "step": 6430
    },
    {
      "epoch": 1.1778479836461624,
      "grad_norm": 0.7671892642974854,
      "learning_rate": 7.019063527455246e-05,
      "loss": 1.0271,
      "step": 6431
    },
    {
      "epoch": 1.1780338227095335,
      "grad_norm": 0.5795000791549683,
      "learning_rate": 7.016272028406153e-05,
      "loss": 0.8227,
      "step": 6432
    },
    {
      "epoch": 1.1782196617729046,
      "grad_norm": 0.717526912689209,
      "learning_rate": 7.013480784561759e-05,
      "loss": 0.9521,
      "step": 6433
    },
    {
      "epoch": 1.1784055008362757,
      "grad_norm": 0.6924964189529419,
      "learning_rate": 7.010689796160805e-05,
      "loss": 1.1982,
      "step": 6434
    },
    {
      "epoch": 1.1785913398996468,
      "grad_norm": 0.7154371738433838,
      "learning_rate": 7.00789906344202e-05,
      "loss": 1.062,
      "step": 6435
    },
    {
      "epoch": 1.178777178963018,
      "grad_norm": 0.8084208965301514,
      "learning_rate": 7.005108586644091e-05,
      "loss": 1.0382,
      "step": 6436
    },
    {
      "epoch": 1.1789630180263893,
      "grad_norm": 0.7454267740249634,
      "learning_rate": 7.002318366005699e-05,
      "loss": 0.7163,
      "step": 6437
    },
    {
      "epoch": 1.1791488570897604,
      "grad_norm": 0.5913962125778198,
      "learning_rate": 6.999528401765494e-05,
      "loss": 0.7636,
      "step": 6438
    },
    {
      "epoch": 1.1793346961531315,
      "grad_norm": 0.6394737958908081,
      "learning_rate": 6.996738694162112e-05,
      "loss": 0.8579,
      "step": 6439
    },
    {
      "epoch": 1.1795205352165026,
      "grad_norm": 0.712942898273468,
      "learning_rate": 6.993949243434159e-05,
      "loss": 0.8436,
      "step": 6440
    },
    {
      "epoch": 1.1797063742798737,
      "grad_norm": 0.6500268578529358,
      "learning_rate": 6.991160049820222e-05,
      "loss": 0.7198,
      "step": 6441
    },
    {
      "epoch": 1.1798922133432448,
      "grad_norm": 0.7196839451789856,
      "learning_rate": 6.988371113558876e-05,
      "loss": 0.9428,
      "step": 6442
    },
    {
      "epoch": 1.180078052406616,
      "grad_norm": 0.6893028616905212,
      "learning_rate": 6.985582434888652e-05,
      "loss": 0.8055,
      "step": 6443
    },
    {
      "epoch": 1.180263891469987,
      "grad_norm": 0.6877514123916626,
      "learning_rate": 6.982794014048077e-05,
      "loss": 1.1647,
      "step": 6444
    },
    {
      "epoch": 1.180449730533358,
      "grad_norm": 0.8232098817825317,
      "learning_rate": 6.980005851275653e-05,
      "loss": 1.1632,
      "step": 6445
    },
    {
      "epoch": 1.1806355695967292,
      "grad_norm": 0.5898348093032837,
      "learning_rate": 6.977217946809857e-05,
      "loss": 0.9491,
      "step": 6446
    },
    {
      "epoch": 1.1808214086601003,
      "grad_norm": 0.8944123387336731,
      "learning_rate": 6.974430300889139e-05,
      "loss": 1.0263,
      "step": 6447
    },
    {
      "epoch": 1.1810072477234714,
      "grad_norm": 0.5503453612327576,
      "learning_rate": 6.971642913751936e-05,
      "loss": 0.7226,
      "step": 6448
    },
    {
      "epoch": 1.1811930867868425,
      "grad_norm": 3.2862448692321777,
      "learning_rate": 6.968855785636665e-05,
      "loss": 1.5893,
      "step": 6449
    },
    {
      "epoch": 1.1813789258502136,
      "grad_norm": 0.7002162933349609,
      "learning_rate": 6.966068916781706e-05,
      "loss": 0.9923,
      "step": 6450
    },
    {
      "epoch": 1.1815647649135848,
      "grad_norm": 0.6545072197914124,
      "learning_rate": 6.963282307425434e-05,
      "loss": 0.9587,
      "step": 6451
    },
    {
      "epoch": 1.1817506039769559,
      "grad_norm": 0.7973212599754333,
      "learning_rate": 6.960495957806185e-05,
      "loss": 1.0527,
      "step": 6452
    },
    {
      "epoch": 1.181936443040327,
      "grad_norm": 0.6965364813804626,
      "learning_rate": 6.957709868162288e-05,
      "loss": 0.9853,
      "step": 6453
    },
    {
      "epoch": 1.1821222821036983,
      "grad_norm": 0.6737655401229858,
      "learning_rate": 6.954924038732038e-05,
      "loss": 1.1285,
      "step": 6454
    },
    {
      "epoch": 1.1823081211670694,
      "grad_norm": 0.6464536786079407,
      "learning_rate": 6.952138469753717e-05,
      "loss": 0.9844,
      "step": 6455
    },
    {
      "epoch": 1.1824939602304405,
      "grad_norm": 0.6128717660903931,
      "learning_rate": 6.949353161465585e-05,
      "loss": 0.7094,
      "step": 6456
    },
    {
      "epoch": 1.1826797992938116,
      "grad_norm": 0.6797751784324646,
      "learning_rate": 6.946568114105866e-05,
      "loss": 0.9996,
      "step": 6457
    },
    {
      "epoch": 1.1828656383571827,
      "grad_norm": 0.6216451525688171,
      "learning_rate": 6.943783327912778e-05,
      "loss": 0.8072,
      "step": 6458
    },
    {
      "epoch": 1.1830514774205538,
      "grad_norm": 0.6847977638244629,
      "learning_rate": 6.940998803124505e-05,
      "loss": 1.0382,
      "step": 6459
    },
    {
      "epoch": 1.183237316483925,
      "grad_norm": 0.6069141626358032,
      "learning_rate": 6.938214539979218e-05,
      "loss": 0.9816,
      "step": 6460
    },
    {
      "epoch": 1.183423155547296,
      "grad_norm": 0.7009860873222351,
      "learning_rate": 6.935430538715056e-05,
      "loss": 1.0688,
      "step": 6461
    },
    {
      "epoch": 1.1836089946106672,
      "grad_norm": 0.5075829029083252,
      "learning_rate": 6.932646799570144e-05,
      "loss": 0.4973,
      "step": 6462
    },
    {
      "epoch": 1.1837948336740383,
      "grad_norm": 0.6592481136322021,
      "learning_rate": 6.929863322782583e-05,
      "loss": 0.8969,
      "step": 6463
    },
    {
      "epoch": 1.1839806727374094,
      "grad_norm": 0.8401702642440796,
      "learning_rate": 6.927080108590445e-05,
      "loss": 0.9823,
      "step": 6464
    },
    {
      "epoch": 1.1841665118007805,
      "grad_norm": 0.7158241271972656,
      "learning_rate": 6.92429715723179e-05,
      "loss": 0.8727,
      "step": 6465
    },
    {
      "epoch": 1.1843523508641516,
      "grad_norm": 0.7230640053749084,
      "learning_rate": 6.921514468944643e-05,
      "loss": 1.1566,
      "step": 6466
    },
    {
      "epoch": 1.1845381899275227,
      "grad_norm": 0.6751217842102051,
      "learning_rate": 6.91873204396702e-05,
      "loss": 1.094,
      "step": 6467
    },
    {
      "epoch": 1.1847240289908938,
      "grad_norm": 0.7101399302482605,
      "learning_rate": 6.915949882536901e-05,
      "loss": 1.0724,
      "step": 6468
    },
    {
      "epoch": 1.1849098680542651,
      "grad_norm": 0.7720397710800171,
      "learning_rate": 6.913167984892255e-05,
      "loss": 0.9402,
      "step": 6469
    },
    {
      "epoch": 1.185095707117636,
      "grad_norm": 0.7422873973846436,
      "learning_rate": 6.910386351271026e-05,
      "loss": 1.0397,
      "step": 6470
    },
    {
      "epoch": 1.1852815461810073,
      "grad_norm": 0.7033947110176086,
      "learning_rate": 6.907604981911127e-05,
      "loss": 1.0692,
      "step": 6471
    },
    {
      "epoch": 1.1854673852443784,
      "grad_norm": 0.7778051495552063,
      "learning_rate": 6.904823877050461e-05,
      "loss": 0.9041,
      "step": 6472
    },
    {
      "epoch": 1.1856532243077496,
      "grad_norm": 0.6052567958831787,
      "learning_rate": 6.902043036926894e-05,
      "loss": 0.9387,
      "step": 6473
    },
    {
      "epoch": 1.1858390633711207,
      "grad_norm": 0.7672857642173767,
      "learning_rate": 6.899262461778284e-05,
      "loss": 1.0686,
      "step": 6474
    },
    {
      "epoch": 1.1860249024344918,
      "grad_norm": 0.7012307047843933,
      "learning_rate": 6.896482151842455e-05,
      "loss": 1.0861,
      "step": 6475
    },
    {
      "epoch": 1.1862107414978629,
      "grad_norm": 0.6781837940216064,
      "learning_rate": 6.893702107357215e-05,
      "loss": 0.98,
      "step": 6476
    },
    {
      "epoch": 1.186396580561234,
      "grad_norm": 0.7039365768432617,
      "learning_rate": 6.890922328560352e-05,
      "loss": 0.7219,
      "step": 6477
    },
    {
      "epoch": 1.186582419624605,
      "grad_norm": 0.715893030166626,
      "learning_rate": 6.888142815689617e-05,
      "loss": 0.9221,
      "step": 6478
    },
    {
      "epoch": 1.1867682586879762,
      "grad_norm": 0.5723534226417542,
      "learning_rate": 6.885363568982757e-05,
      "loss": 0.8025,
      "step": 6479
    },
    {
      "epoch": 1.1869540977513473,
      "grad_norm": 0.6755521297454834,
      "learning_rate": 6.882584588677479e-05,
      "loss": 0.8706,
      "step": 6480
    },
    {
      "epoch": 1.1871399368147184,
      "grad_norm": 0.6190736889839172,
      "learning_rate": 6.87980587501148e-05,
      "loss": 0.8423,
      "step": 6481
    },
    {
      "epoch": 1.1873257758780895,
      "grad_norm": 0.6882662773132324,
      "learning_rate": 6.877027428222427e-05,
      "loss": 1.1149,
      "step": 6482
    },
    {
      "epoch": 1.1875116149414606,
      "grad_norm": 0.8642654418945312,
      "learning_rate": 6.874249248547966e-05,
      "loss": 0.952,
      "step": 6483
    },
    {
      "epoch": 1.1876974540048317,
      "grad_norm": 0.6334053874015808,
      "learning_rate": 6.871471336225728e-05,
      "loss": 0.9349,
      "step": 6484
    },
    {
      "epoch": 1.1878832930682028,
      "grad_norm": 0.6225823760032654,
      "learning_rate": 6.868693691493305e-05,
      "loss": 0.9393,
      "step": 6485
    },
    {
      "epoch": 1.1880691321315742,
      "grad_norm": 0.6007322072982788,
      "learning_rate": 6.86591631458828e-05,
      "loss": 0.9067,
      "step": 6486
    },
    {
      "epoch": 1.188254971194945,
      "grad_norm": 0.7253145575523376,
      "learning_rate": 6.863139205748203e-05,
      "loss": 0.6779,
      "step": 6487
    },
    {
      "epoch": 1.1884408102583164,
      "grad_norm": 0.6054568290710449,
      "learning_rate": 6.86036236521061e-05,
      "loss": 0.8598,
      "step": 6488
    },
    {
      "epoch": 1.1886266493216875,
      "grad_norm": 0.8909119367599487,
      "learning_rate": 6.857585793213011e-05,
      "loss": 1.0237,
      "step": 6489
    },
    {
      "epoch": 1.1888124883850586,
      "grad_norm": 0.6840725541114807,
      "learning_rate": 6.85480948999289e-05,
      "loss": 0.9126,
      "step": 6490
    },
    {
      "epoch": 1.1889983274484297,
      "grad_norm": 0.6242436766624451,
      "learning_rate": 6.852033455787712e-05,
      "loss": 0.9567,
      "step": 6491
    },
    {
      "epoch": 1.1891841665118008,
      "grad_norm": 0.6949776411056519,
      "learning_rate": 6.849257690834914e-05,
      "loss": 0.9466,
      "step": 6492
    },
    {
      "epoch": 1.189370005575172,
      "grad_norm": 0.7519738674163818,
      "learning_rate": 6.84648219537192e-05,
      "loss": 0.7867,
      "step": 6493
    },
    {
      "epoch": 1.189555844638543,
      "grad_norm": 0.6986894607543945,
      "learning_rate": 6.843706969636115e-05,
      "loss": 1.1686,
      "step": 6494
    },
    {
      "epoch": 1.1897416837019141,
      "grad_norm": 0.6281919479370117,
      "learning_rate": 6.840932013864876e-05,
      "loss": 0.9281,
      "step": 6495
    },
    {
      "epoch": 1.1899275227652852,
      "grad_norm": 1.0236083269119263,
      "learning_rate": 6.83815732829555e-05,
      "loss": 1.3388,
      "step": 6496
    },
    {
      "epoch": 1.1901133618286563,
      "grad_norm": 0.5717867016792297,
      "learning_rate": 6.83538291316546e-05,
      "loss": 0.7795,
      "step": 6497
    },
    {
      "epoch": 1.1902992008920275,
      "grad_norm": 0.6059176325798035,
      "learning_rate": 6.83260876871191e-05,
      "loss": 0.9609,
      "step": 6498
    },
    {
      "epoch": 1.1904850399553986,
      "grad_norm": 0.6473904848098755,
      "learning_rate": 6.829834895172175e-05,
      "loss": 0.9561,
      "step": 6499
    },
    {
      "epoch": 1.1906708790187697,
      "grad_norm": 0.6453015208244324,
      "learning_rate": 6.827061292783519e-05,
      "loss": 0.8863,
      "step": 6500
    },
    {
      "epoch": 1.1908567180821408,
      "grad_norm": 0.5157768726348877,
      "learning_rate": 6.824287961783163e-05,
      "loss": 0.6292,
      "step": 6501
    },
    {
      "epoch": 1.1910425571455119,
      "grad_norm": 0.6397795677185059,
      "learning_rate": 6.821514902408326e-05,
      "loss": 0.7689,
      "step": 6502
    },
    {
      "epoch": 1.1912283962088832,
      "grad_norm": 0.6399672031402588,
      "learning_rate": 6.818742114896184e-05,
      "loss": 0.8271,
      "step": 6503
    },
    {
      "epoch": 1.1914142352722543,
      "grad_norm": 0.5401207208633423,
      "learning_rate": 6.815969599483906e-05,
      "loss": 0.6154,
      "step": 6504
    },
    {
      "epoch": 1.1916000743356254,
      "grad_norm": 0.790274977684021,
      "learning_rate": 6.81319735640863e-05,
      "loss": 0.5777,
      "step": 6505
    },
    {
      "epoch": 1.1917859133989965,
      "grad_norm": 0.7683626413345337,
      "learning_rate": 6.810425385907471e-05,
      "loss": 0.9646,
      "step": 6506
    },
    {
      "epoch": 1.1919717524623676,
      "grad_norm": 0.6598650813102722,
      "learning_rate": 6.807653688217526e-05,
      "loss": 0.9366,
      "step": 6507
    },
    {
      "epoch": 1.1921575915257387,
      "grad_norm": 0.6899055242538452,
      "learning_rate": 6.804882263575855e-05,
      "loss": 0.9253,
      "step": 6508
    },
    {
      "epoch": 1.1923434305891099,
      "grad_norm": 0.7945172190666199,
      "learning_rate": 6.802111112219516e-05,
      "loss": 0.977,
      "step": 6509
    },
    {
      "epoch": 1.192529269652481,
      "grad_norm": 0.6663496494293213,
      "learning_rate": 6.799340234385521e-05,
      "loss": 0.9406,
      "step": 6510
    },
    {
      "epoch": 1.192715108715852,
      "grad_norm": 0.7770522236824036,
      "learning_rate": 6.796569630310872e-05,
      "loss": 1.0315,
      "step": 6511
    },
    {
      "epoch": 1.1929009477792232,
      "grad_norm": 0.7419028282165527,
      "learning_rate": 6.793799300232548e-05,
      "loss": 0.7592,
      "step": 6512
    },
    {
      "epoch": 1.1930867868425943,
      "grad_norm": 0.7410839796066284,
      "learning_rate": 6.791029244387499e-05,
      "loss": 0.752,
      "step": 6513
    },
    {
      "epoch": 1.1932726259059654,
      "grad_norm": 0.7905089259147644,
      "learning_rate": 6.788259463012656e-05,
      "loss": 0.9731,
      "step": 6514
    },
    {
      "epoch": 1.1934584649693365,
      "grad_norm": 1.5963722467422485,
      "learning_rate": 6.785489956344921e-05,
      "loss": 1.6695,
      "step": 6515
    },
    {
      "epoch": 1.1936443040327076,
      "grad_norm": 0.686390221118927,
      "learning_rate": 6.782720724621178e-05,
      "loss": 0.9611,
      "step": 6516
    },
    {
      "epoch": 1.1938301430960787,
      "grad_norm": 0.6483101844787598,
      "learning_rate": 6.779951768078284e-05,
      "loss": 0.8985,
      "step": 6517
    },
    {
      "epoch": 1.19401598215945,
      "grad_norm": 0.6660516262054443,
      "learning_rate": 6.777183086953074e-05,
      "loss": 1.0176,
      "step": 6518
    },
    {
      "epoch": 1.194201821222821,
      "grad_norm": 0.6592452526092529,
      "learning_rate": 6.774414681482361e-05,
      "loss": 0.8554,
      "step": 6519
    },
    {
      "epoch": 1.1943876602861923,
      "grad_norm": 0.7071273922920227,
      "learning_rate": 6.771646551902932e-05,
      "loss": 1.0079,
      "step": 6520
    },
    {
      "epoch": 1.1945734993495634,
      "grad_norm": 0.7174739241600037,
      "learning_rate": 6.768878698451553e-05,
      "loss": 0.9785,
      "step": 6521
    },
    {
      "epoch": 1.1947593384129345,
      "grad_norm": 0.8033270835876465,
      "learning_rate": 6.766111121364958e-05,
      "loss": 1.0207,
      "step": 6522
    },
    {
      "epoch": 1.1949451774763056,
      "grad_norm": 0.6496764421463013,
      "learning_rate": 6.763343820879872e-05,
      "loss": 1.0746,
      "step": 6523
    },
    {
      "epoch": 1.1951310165396767,
      "grad_norm": 0.6587570905685425,
      "learning_rate": 6.760576797232983e-05,
      "loss": 0.9393,
      "step": 6524
    },
    {
      "epoch": 1.1953168556030478,
      "grad_norm": 0.730059027671814,
      "learning_rate": 6.757810050660958e-05,
      "loss": 0.6695,
      "step": 6525
    },
    {
      "epoch": 1.195502694666419,
      "grad_norm": 0.7065173387527466,
      "learning_rate": 6.755043581400452e-05,
      "loss": 0.9532,
      "step": 6526
    },
    {
      "epoch": 1.19568853372979,
      "grad_norm": 0.7613650560379028,
      "learning_rate": 6.752277389688078e-05,
      "loss": 1.1234,
      "step": 6527
    },
    {
      "epoch": 1.1958743727931611,
      "grad_norm": 0.6849968433380127,
      "learning_rate": 6.749511475760441e-05,
      "loss": 0.8637,
      "step": 6528
    },
    {
      "epoch": 1.1960602118565322,
      "grad_norm": 0.7032272219657898,
      "learning_rate": 6.746745839854111e-05,
      "loss": 0.9595,
      "step": 6529
    },
    {
      "epoch": 1.1962460509199033,
      "grad_norm": 0.6895639300346375,
      "learning_rate": 6.743980482205644e-05,
      "loss": 0.9467,
      "step": 6530
    },
    {
      "epoch": 1.1964318899832744,
      "grad_norm": 0.6482696533203125,
      "learning_rate": 6.74121540305156e-05,
      "loss": 0.9559,
      "step": 6531
    },
    {
      "epoch": 1.1966177290466455,
      "grad_norm": 0.6822243332862854,
      "learning_rate": 6.738450602628364e-05,
      "loss": 1.058,
      "step": 6532
    },
    {
      "epoch": 1.1968035681100166,
      "grad_norm": 0.699733555316925,
      "learning_rate": 6.73568608117254e-05,
      "loss": 0.8538,
      "step": 6533
    },
    {
      "epoch": 1.1969894071733878,
      "grad_norm": 0.6649261713027954,
      "learning_rate": 6.732921838920538e-05,
      "loss": 0.7759,
      "step": 6534
    },
    {
      "epoch": 1.197175246236759,
      "grad_norm": 0.8819229006767273,
      "learning_rate": 6.730157876108798e-05,
      "loss": 1.1101,
      "step": 6535
    },
    {
      "epoch": 1.19736108530013,
      "grad_norm": 0.733377993106842,
      "learning_rate": 6.727394192973716e-05,
      "loss": 1.2236,
      "step": 6536
    },
    {
      "epoch": 1.1975469243635013,
      "grad_norm": 0.6071882247924805,
      "learning_rate": 6.724630789751688e-05,
      "loss": 0.9082,
      "step": 6537
    },
    {
      "epoch": 1.1977327634268724,
      "grad_norm": 0.6043367981910706,
      "learning_rate": 6.72186766667906e-05,
      "loss": 0.787,
      "step": 6538
    },
    {
      "epoch": 1.1979186024902435,
      "grad_norm": 0.5902948379516602,
      "learning_rate": 6.719104823992178e-05,
      "loss": 1.0035,
      "step": 6539
    },
    {
      "epoch": 1.1981044415536146,
      "grad_norm": 0.5684782266616821,
      "learning_rate": 6.716342261927353e-05,
      "loss": 0.7045,
      "step": 6540
    },
    {
      "epoch": 1.1982902806169857,
      "grad_norm": 0.6624257564544678,
      "learning_rate": 6.713579980720872e-05,
      "loss": 1.0531,
      "step": 6541
    },
    {
      "epoch": 1.1984761196803568,
      "grad_norm": 0.7116941809654236,
      "learning_rate": 6.710817980608996e-05,
      "loss": 0.8097,
      "step": 6542
    },
    {
      "epoch": 1.198661958743728,
      "grad_norm": 0.730290412902832,
      "learning_rate": 6.708056261827966e-05,
      "loss": 0.9771,
      "step": 6543
    },
    {
      "epoch": 1.198847797807099,
      "grad_norm": 0.6953971982002258,
      "learning_rate": 6.705294824614004e-05,
      "loss": 0.8913,
      "step": 6544
    },
    {
      "epoch": 1.1990336368704702,
      "grad_norm": 0.6694558262825012,
      "learning_rate": 6.70253366920329e-05,
      "loss": 0.9706,
      "step": 6545
    },
    {
      "epoch": 1.1992194759338413,
      "grad_norm": 0.6435080766677856,
      "learning_rate": 6.699772795832e-05,
      "loss": 0.9639,
      "step": 6546
    },
    {
      "epoch": 1.1994053149972124,
      "grad_norm": 0.6081843376159668,
      "learning_rate": 6.69701220473628e-05,
      "loss": 0.7492,
      "step": 6547
    },
    {
      "epoch": 1.1995911540605835,
      "grad_norm": 0.7480099201202393,
      "learning_rate": 6.69425189615224e-05,
      "loss": 0.9639,
      "step": 6548
    },
    {
      "epoch": 1.1997769931239546,
      "grad_norm": 0.5997082591056824,
      "learning_rate": 6.69149187031598e-05,
      "loss": 0.8858,
      "step": 6549
    },
    {
      "epoch": 1.1999628321873257,
      "grad_norm": 0.6469660401344299,
      "learning_rate": 6.68873212746357e-05,
      "loss": 0.7982,
      "step": 6550
    },
    {
      "epoch": 1.2001486712506968,
      "grad_norm": 0.7918186783790588,
      "learning_rate": 6.685972667831064e-05,
      "loss": 1.2226,
      "step": 6551
    },
    {
      "epoch": 1.2003345103140681,
      "grad_norm": 0.7456339001655579,
      "learning_rate": 6.683213491654472e-05,
      "loss": 1.0805,
      "step": 6552
    },
    {
      "epoch": 1.2005203493774392,
      "grad_norm": 0.7502478957176208,
      "learning_rate": 6.680454599169799e-05,
      "loss": 1.0391,
      "step": 6553
    },
    {
      "epoch": 1.2007061884408103,
      "grad_norm": 0.6726301312446594,
      "learning_rate": 6.677695990613024e-05,
      "loss": 1.0379,
      "step": 6554
    },
    {
      "epoch": 1.2008920275041814,
      "grad_norm": 0.7034464478492737,
      "learning_rate": 6.674937666220085e-05,
      "loss": 0.9541,
      "step": 6555
    },
    {
      "epoch": 1.2010778665675526,
      "grad_norm": 0.6492792367935181,
      "learning_rate": 6.672179626226917e-05,
      "loss": 0.717,
      "step": 6556
    },
    {
      "epoch": 1.2012637056309237,
      "grad_norm": 0.7404565215110779,
      "learning_rate": 6.669421870869416e-05,
      "loss": 1.1081,
      "step": 6557
    },
    {
      "epoch": 1.2014495446942948,
      "grad_norm": 0.7572711706161499,
      "learning_rate": 6.666664400383465e-05,
      "loss": 0.9685,
      "step": 6558
    },
    {
      "epoch": 1.2016353837576659,
      "grad_norm": 0.9167105555534363,
      "learning_rate": 6.663907215004907e-05,
      "loss": 1.031,
      "step": 6559
    },
    {
      "epoch": 1.201821222821037,
      "grad_norm": 1.3603627681732178,
      "learning_rate": 6.661150314969578e-05,
      "loss": 1.564,
      "step": 6560
    },
    {
      "epoch": 1.202007061884408,
      "grad_norm": 0.7942409515380859,
      "learning_rate": 6.65839370051328e-05,
      "loss": 1.0709,
      "step": 6561
    },
    {
      "epoch": 1.2021929009477792,
      "grad_norm": 0.6643564105033875,
      "learning_rate": 6.655637371871788e-05,
      "loss": 1.0886,
      "step": 6562
    },
    {
      "epoch": 1.2023787400111503,
      "grad_norm": 0.6733203530311584,
      "learning_rate": 6.65288132928086e-05,
      "loss": 0.9338,
      "step": 6563
    },
    {
      "epoch": 1.2025645790745214,
      "grad_norm": 0.5864598155021667,
      "learning_rate": 6.650125572976227e-05,
      "loss": 0.9149,
      "step": 6564
    },
    {
      "epoch": 1.2027504181378925,
      "grad_norm": 0.659898042678833,
      "learning_rate": 6.647370103193595e-05,
      "loss": 1.0477,
      "step": 6565
    },
    {
      "epoch": 1.2029362572012636,
      "grad_norm": 0.7071530818939209,
      "learning_rate": 6.644614920168641e-05,
      "loss": 0.9551,
      "step": 6566
    },
    {
      "epoch": 1.2031220962646347,
      "grad_norm": 0.6440016627311707,
      "learning_rate": 6.641860024137025e-05,
      "loss": 0.9242,
      "step": 6567
    },
    {
      "epoch": 1.2033079353280058,
      "grad_norm": 0.5903676152229309,
      "learning_rate": 6.639105415334383e-05,
      "loss": 0.9222,
      "step": 6568
    },
    {
      "epoch": 1.2034937743913772,
      "grad_norm": 0.8212350606918335,
      "learning_rate": 6.636351093996316e-05,
      "loss": 0.8822,
      "step": 6569
    },
    {
      "epoch": 1.2036796134547483,
      "grad_norm": 0.6122947335243225,
      "learning_rate": 6.63359706035841e-05,
      "loss": 0.8444,
      "step": 6570
    },
    {
      "epoch": 1.2038654525181194,
      "grad_norm": 0.6683095097541809,
      "learning_rate": 6.630843314656221e-05,
      "loss": 0.9007,
      "step": 6571
    },
    {
      "epoch": 1.2040512915814905,
      "grad_norm": 0.7547433376312256,
      "learning_rate": 6.62808985712529e-05,
      "loss": 0.934,
      "step": 6572
    },
    {
      "epoch": 1.2042371306448616,
      "grad_norm": 0.6892956495285034,
      "learning_rate": 6.625336688001117e-05,
      "loss": 1.1036,
      "step": 6573
    },
    {
      "epoch": 1.2044229697082327,
      "grad_norm": 0.7414408326148987,
      "learning_rate": 6.622583807519191e-05,
      "loss": 1.1518,
      "step": 6574
    },
    {
      "epoch": 1.2046088087716038,
      "grad_norm": 0.737737238407135,
      "learning_rate": 6.619831215914974e-05,
      "loss": 0.877,
      "step": 6575
    },
    {
      "epoch": 1.204794647834975,
      "grad_norm": 0.7120890021324158,
      "learning_rate": 6.617078913423895e-05,
      "loss": 0.9871,
      "step": 6576
    },
    {
      "epoch": 1.204980486898346,
      "grad_norm": 0.6039966344833374,
      "learning_rate": 6.614326900281371e-05,
      "loss": 0.7617,
      "step": 6577
    },
    {
      "epoch": 1.2051663259617171,
      "grad_norm": 0.6865859031677246,
      "learning_rate": 6.611575176722783e-05,
      "loss": 0.7669,
      "step": 6578
    },
    {
      "epoch": 1.2053521650250882,
      "grad_norm": 0.6812560558319092,
      "learning_rate": 6.608823742983497e-05,
      "loss": 0.9179,
      "step": 6579
    },
    {
      "epoch": 1.2055380040884593,
      "grad_norm": 0.8033499121665955,
      "learning_rate": 6.60607259929884e-05,
      "loss": 1.0227,
      "step": 6580
    },
    {
      "epoch": 1.2057238431518305,
      "grad_norm": 0.6142318844795227,
      "learning_rate": 6.60332174590413e-05,
      "loss": 0.6658,
      "step": 6581
    },
    {
      "epoch": 1.2059096822152016,
      "grad_norm": 0.7506161332130432,
      "learning_rate": 6.600571183034657e-05,
      "loss": 0.9,
      "step": 6582
    },
    {
      "epoch": 1.2060955212785727,
      "grad_norm": 0.7655713558197021,
      "learning_rate": 6.597820910925674e-05,
      "loss": 1.2014,
      "step": 6583
    },
    {
      "epoch": 1.206281360341944,
      "grad_norm": 0.8014414310455322,
      "learning_rate": 6.595070929812423e-05,
      "loss": 1.1676,
      "step": 6584
    },
    {
      "epoch": 1.2064671994053149,
      "grad_norm": 0.6293933987617493,
      "learning_rate": 6.592321239930112e-05,
      "loss": 1.0769,
      "step": 6585
    },
    {
      "epoch": 1.2066530384686862,
      "grad_norm": 0.7731804251670837,
      "learning_rate": 6.589571841513934e-05,
      "loss": 1.0054,
      "step": 6586
    },
    {
      "epoch": 1.2068388775320573,
      "grad_norm": 0.686371922492981,
      "learning_rate": 6.586822734799044e-05,
      "loss": 1.0477,
      "step": 6587
    },
    {
      "epoch": 1.2070247165954284,
      "grad_norm": 0.7579498291015625,
      "learning_rate": 6.584073920020581e-05,
      "loss": 1.0027,
      "step": 6588
    },
    {
      "epoch": 1.2072105556587995,
      "grad_norm": 0.7064153552055359,
      "learning_rate": 6.581325397413664e-05,
      "loss": 1.0057,
      "step": 6589
    },
    {
      "epoch": 1.2073963947221706,
      "grad_norm": 0.6257011294364929,
      "learning_rate": 6.57857716721337e-05,
      "loss": 0.8138,
      "step": 6590
    },
    {
      "epoch": 1.2075822337855417,
      "grad_norm": 0.6599844098091125,
      "learning_rate": 6.575829229654769e-05,
      "loss": 0.8032,
      "step": 6591
    },
    {
      "epoch": 1.2077680728489129,
      "grad_norm": 0.7026282548904419,
      "learning_rate": 6.573081584972892e-05,
      "loss": 1.0257,
      "step": 6592
    },
    {
      "epoch": 1.207953911912284,
      "grad_norm": 0.6733138561248779,
      "learning_rate": 6.570334233402756e-05,
      "loss": 0.9201,
      "step": 6593
    },
    {
      "epoch": 1.208139750975655,
      "grad_norm": 0.6270269155502319,
      "learning_rate": 6.567587175179342e-05,
      "loss": 0.8864,
      "step": 6594
    },
    {
      "epoch": 1.2083255900390262,
      "grad_norm": 0.6575441360473633,
      "learning_rate": 6.564840410537613e-05,
      "loss": 1.0867,
      "step": 6595
    },
    {
      "epoch": 1.2085114291023973,
      "grad_norm": 0.6776338815689087,
      "learning_rate": 6.562093939712516e-05,
      "loss": 0.8681,
      "step": 6596
    },
    {
      "epoch": 1.2086972681657684,
      "grad_norm": 0.67679762840271,
      "learning_rate": 6.55934776293895e-05,
      "loss": 0.8977,
      "step": 6597
    },
    {
      "epoch": 1.2088831072291395,
      "grad_norm": 0.7879827618598938,
      "learning_rate": 6.55660188045181e-05,
      "loss": 0.7244,
      "step": 6598
    },
    {
      "epoch": 1.2090689462925106,
      "grad_norm": 0.6360406875610352,
      "learning_rate": 6.553856292485947e-05,
      "loss": 0.9499,
      "step": 6599
    },
    {
      "epoch": 1.2092547853558817,
      "grad_norm": 1.2036761045455933,
      "learning_rate": 6.551110999276207e-05,
      "loss": 1.0049,
      "step": 6600
    },
    {
      "epoch": 1.209440624419253,
      "grad_norm": 0.8532909750938416,
      "learning_rate": 6.548366001057395e-05,
      "loss": 0.926,
      "step": 6601
    },
    {
      "epoch": 1.209626463482624,
      "grad_norm": 0.7498572468757629,
      "learning_rate": 6.5456212980643e-05,
      "loss": 1.0418,
      "step": 6602
    },
    {
      "epoch": 1.2098123025459953,
      "grad_norm": 0.7277694940567017,
      "learning_rate": 6.542876890531684e-05,
      "loss": 1.0088,
      "step": 6603
    },
    {
      "epoch": 1.2099981416093664,
      "grad_norm": 0.6551283597946167,
      "learning_rate": 6.540132778694276e-05,
      "loss": 0.7892,
      "step": 6604
    },
    {
      "epoch": 1.2101839806727375,
      "grad_norm": 0.6652863621711731,
      "learning_rate": 6.537388962786793e-05,
      "loss": 0.8712,
      "step": 6605
    },
    {
      "epoch": 1.2103698197361086,
      "grad_norm": 0.7065383791923523,
      "learning_rate": 6.534645443043911e-05,
      "loss": 1.0349,
      "step": 6606
    },
    {
      "epoch": 1.2105556587994797,
      "grad_norm": 0.6703068017959595,
      "learning_rate": 6.531902219700296e-05,
      "loss": 0.9831,
      "step": 6607
    },
    {
      "epoch": 1.2107414978628508,
      "grad_norm": 0.7374162673950195,
      "learning_rate": 6.52915929299058e-05,
      "loss": 0.95,
      "step": 6608
    },
    {
      "epoch": 1.210927336926222,
      "grad_norm": 0.689177393913269,
      "learning_rate": 6.526416663149368e-05,
      "loss": 0.7996,
      "step": 6609
    },
    {
      "epoch": 1.211113175989593,
      "grad_norm": 0.6918917894363403,
      "learning_rate": 6.523674330411253e-05,
      "loss": 0.8538,
      "step": 6610
    },
    {
      "epoch": 1.2112990150529641,
      "grad_norm": 0.7495622038841248,
      "learning_rate": 6.52093229501078e-05,
      "loss": 1.1756,
      "step": 6611
    },
    {
      "epoch": 1.2114848541163352,
      "grad_norm": 0.6858820915222168,
      "learning_rate": 6.518190557182492e-05,
      "loss": 1.0732,
      "step": 6612
    },
    {
      "epoch": 1.2116706931797063,
      "grad_norm": 1.2775191068649292,
      "learning_rate": 6.515449117160886e-05,
      "loss": 1.497,
      "step": 6613
    },
    {
      "epoch": 1.2118565322430774,
      "grad_norm": 0.741070568561554,
      "learning_rate": 6.51270797518045e-05,
      "loss": 0.8712,
      "step": 6614
    },
    {
      "epoch": 1.2120423713064485,
      "grad_norm": 0.7354996204376221,
      "learning_rate": 6.509967131475637e-05,
      "loss": 1.2991,
      "step": 6615
    },
    {
      "epoch": 1.2122282103698196,
      "grad_norm": 0.7553620934486389,
      "learning_rate": 6.507226586280877e-05,
      "loss": 0.9451,
      "step": 6616
    },
    {
      "epoch": 1.2124140494331908,
      "grad_norm": 0.7338206171989441,
      "learning_rate": 6.504486339830581e-05,
      "loss": 0.9886,
      "step": 6617
    },
    {
      "epoch": 1.212599888496562,
      "grad_norm": 0.7190319895744324,
      "learning_rate": 6.501746392359118e-05,
      "loss": 1.0271,
      "step": 6618
    },
    {
      "epoch": 1.2127857275599332,
      "grad_norm": 0.6434297561645508,
      "learning_rate": 6.499006744100853e-05,
      "loss": 0.9303,
      "step": 6619
    },
    {
      "epoch": 1.2129715666233043,
      "grad_norm": 0.7200321555137634,
      "learning_rate": 6.496267395290101e-05,
      "loss": 1.1825,
      "step": 6620
    },
    {
      "epoch": 1.2131574056866754,
      "grad_norm": 0.6997349262237549,
      "learning_rate": 6.493528346161176e-05,
      "loss": 1.0327,
      "step": 6621
    },
    {
      "epoch": 1.2133432447500465,
      "grad_norm": 0.7475888729095459,
      "learning_rate": 6.490789596948345e-05,
      "loss": 0.904,
      "step": 6622
    },
    {
      "epoch": 1.2135290838134176,
      "grad_norm": 0.8181599974632263,
      "learning_rate": 6.488051147885866e-05,
      "loss": 1.1392,
      "step": 6623
    },
    {
      "epoch": 1.2137149228767887,
      "grad_norm": 0.7052812576293945,
      "learning_rate": 6.485312999207967e-05,
      "loss": 1.1321,
      "step": 6624
    },
    {
      "epoch": 1.2139007619401598,
      "grad_norm": 0.6384223699569702,
      "learning_rate": 6.482575151148839e-05,
      "loss": 0.9949,
      "step": 6625
    },
    {
      "epoch": 1.214086601003531,
      "grad_norm": 0.7056748867034912,
      "learning_rate": 6.479837603942665e-05,
      "loss": 0.8554,
      "step": 6626
    },
    {
      "epoch": 1.214272440066902,
      "grad_norm": 0.7674799561500549,
      "learning_rate": 6.477100357823583e-05,
      "loss": 1.0488,
      "step": 6627
    },
    {
      "epoch": 1.2144582791302732,
      "grad_norm": 0.6128942966461182,
      "learning_rate": 6.474363413025724e-05,
      "loss": 0.8124,
      "step": 6628
    },
    {
      "epoch": 1.2146441181936443,
      "grad_norm": 0.7572517991065979,
      "learning_rate": 6.47162676978318e-05,
      "loss": 1.1974,
      "step": 6629
    },
    {
      "epoch": 1.2148299572570154,
      "grad_norm": 0.7268834114074707,
      "learning_rate": 6.468890428330026e-05,
      "loss": 0.9459,
      "step": 6630
    },
    {
      "epoch": 1.2150157963203865,
      "grad_norm": 0.5437666773796082,
      "learning_rate": 6.466154388900309e-05,
      "loss": 0.8954,
      "step": 6631
    },
    {
      "epoch": 1.2152016353837576,
      "grad_norm": 0.7641748189926147,
      "learning_rate": 6.46341865172804e-05,
      "loss": 0.9749,
      "step": 6632
    },
    {
      "epoch": 1.2153874744471287,
      "grad_norm": 0.6073144674301147,
      "learning_rate": 6.46068321704722e-05,
      "loss": 0.9929,
      "step": 6633
    },
    {
      "epoch": 1.2155733135104998,
      "grad_norm": 0.6687734127044678,
      "learning_rate": 6.457948085091812e-05,
      "loss": 1.0394,
      "step": 6634
    },
    {
      "epoch": 1.2157591525738711,
      "grad_norm": 0.6596730947494507,
      "learning_rate": 6.45521325609576e-05,
      "loss": 0.8353,
      "step": 6635
    },
    {
      "epoch": 1.2159449916372422,
      "grad_norm": 0.6316202282905579,
      "learning_rate": 6.45247873029298e-05,
      "loss": 0.8294,
      "step": 6636
    },
    {
      "epoch": 1.2161308307006133,
      "grad_norm": 0.616620659828186,
      "learning_rate": 6.449744507917362e-05,
      "loss": 0.8058,
      "step": 6637
    },
    {
      "epoch": 1.2163166697639844,
      "grad_norm": 0.7293714880943298,
      "learning_rate": 6.447010589202769e-05,
      "loss": 1.0555,
      "step": 6638
    },
    {
      "epoch": 1.2165025088273556,
      "grad_norm": 0.6899313926696777,
      "learning_rate": 6.444276974383037e-05,
      "loss": 0.9136,
      "step": 6639
    },
    {
      "epoch": 1.2166883478907267,
      "grad_norm": 0.6521621346473694,
      "learning_rate": 6.441543663691984e-05,
      "loss": 0.9844,
      "step": 6640
    },
    {
      "epoch": 1.2168741869540978,
      "grad_norm": 0.6059038639068604,
      "learning_rate": 6.43881065736339e-05,
      "loss": 0.7697,
      "step": 6641
    },
    {
      "epoch": 1.2170600260174689,
      "grad_norm": 0.6877594590187073,
      "learning_rate": 6.436077955631017e-05,
      "loss": 0.8564,
      "step": 6642
    },
    {
      "epoch": 1.21724586508084,
      "grad_norm": 0.7149982452392578,
      "learning_rate": 6.4333455587286e-05,
      "loss": 0.9599,
      "step": 6643
    },
    {
      "epoch": 1.217431704144211,
      "grad_norm": 0.7110626101493835,
      "learning_rate": 6.430613466889846e-05,
      "loss": 0.9073,
      "step": 6644
    },
    {
      "epoch": 1.2176175432075822,
      "grad_norm": 0.7270492315292358,
      "learning_rate": 6.427881680348437e-05,
      "loss": 0.8962,
      "step": 6645
    },
    {
      "epoch": 1.2178033822709533,
      "grad_norm": 0.721595048904419,
      "learning_rate": 6.425150199338026e-05,
      "loss": 0.7376,
      "step": 6646
    },
    {
      "epoch": 1.2179892213343244,
      "grad_norm": 0.6731526255607605,
      "learning_rate": 6.422419024092248e-05,
      "loss": 0.8957,
      "step": 6647
    },
    {
      "epoch": 1.2181750603976955,
      "grad_norm": 0.6578767895698547,
      "learning_rate": 6.419688154844701e-05,
      "loss": 1.0118,
      "step": 6648
    },
    {
      "epoch": 1.2183608994610666,
      "grad_norm": 0.7469388246536255,
      "learning_rate": 6.416957591828968e-05,
      "loss": 1.0622,
      "step": 6649
    },
    {
      "epoch": 1.218546738524438,
      "grad_norm": 0.6205630898475647,
      "learning_rate": 6.414227335278589e-05,
      "loss": 0.891,
      "step": 6650
    },
    {
      "epoch": 1.2187325775878088,
      "grad_norm": 0.810269296169281,
      "learning_rate": 6.411497385427099e-05,
      "loss": 1.2256,
      "step": 6651
    },
    {
      "epoch": 1.2189184166511802,
      "grad_norm": 0.6002628803253174,
      "learning_rate": 6.408767742507995e-05,
      "loss": 0.8674,
      "step": 6652
    },
    {
      "epoch": 1.2191042557145513,
      "grad_norm": 0.6940420269966125,
      "learning_rate": 6.406038406754744e-05,
      "loss": 0.9959,
      "step": 6653
    },
    {
      "epoch": 1.2192900947779224,
      "grad_norm": 0.7242604494094849,
      "learning_rate": 6.403309378400801e-05,
      "loss": 0.9368,
      "step": 6654
    },
    {
      "epoch": 1.2194759338412935,
      "grad_norm": 0.748961865901947,
      "learning_rate": 6.400580657679576e-05,
      "loss": 0.9917,
      "step": 6655
    },
    {
      "epoch": 1.2196617729046646,
      "grad_norm": 0.6754995584487915,
      "learning_rate": 6.397852244824472e-05,
      "loss": 0.9709,
      "step": 6656
    },
    {
      "epoch": 1.2198476119680357,
      "grad_norm": 0.7186061143875122,
      "learning_rate": 6.395124140068845e-05,
      "loss": 1.0488,
      "step": 6657
    },
    {
      "epoch": 1.2200334510314068,
      "grad_norm": 0.7922396063804626,
      "learning_rate": 6.392396343646042e-05,
      "loss": 1.3205,
      "step": 6658
    },
    {
      "epoch": 1.220219290094778,
      "grad_norm": 0.7937091588973999,
      "learning_rate": 6.38966885578938e-05,
      "loss": 0.8431,
      "step": 6659
    },
    {
      "epoch": 1.220405129158149,
      "grad_norm": 0.6483901739120483,
      "learning_rate": 6.386941676732139e-05,
      "loss": 0.986,
      "step": 6660
    },
    {
      "epoch": 1.2205909682215201,
      "grad_norm": 0.6936187744140625,
      "learning_rate": 6.384214806707591e-05,
      "loss": 0.7993,
      "step": 6661
    },
    {
      "epoch": 1.2207768072848912,
      "grad_norm": 0.6259828805923462,
      "learning_rate": 6.38148824594896e-05,
      "loss": 0.894,
      "step": 6662
    },
    {
      "epoch": 1.2209626463482623,
      "grad_norm": 0.6188904047012329,
      "learning_rate": 6.378761994689467e-05,
      "loss": 0.8241,
      "step": 6663
    },
    {
      "epoch": 1.2211484854116335,
      "grad_norm": 0.7406849265098572,
      "learning_rate": 6.376036053162281e-05,
      "loss": 0.9231,
      "step": 6664
    },
    {
      "epoch": 1.2213343244750046,
      "grad_norm": 0.6560678482055664,
      "learning_rate": 6.373310421600565e-05,
      "loss": 1.0361,
      "step": 6665
    },
    {
      "epoch": 1.2215201635383757,
      "grad_norm": 0.7963076233863831,
      "learning_rate": 6.370585100237447e-05,
      "loss": 1.0212,
      "step": 6666
    },
    {
      "epoch": 1.221706002601747,
      "grad_norm": 0.6689628958702087,
      "learning_rate": 6.367860089306028e-05,
      "loss": 0.9403,
      "step": 6667
    },
    {
      "epoch": 1.2218918416651179,
      "grad_norm": 0.7792874574661255,
      "learning_rate": 6.365135389039392e-05,
      "loss": 0.9399,
      "step": 6668
    },
    {
      "epoch": 1.2220776807284892,
      "grad_norm": 0.7756471633911133,
      "learning_rate": 6.362410999670577e-05,
      "loss": 0.9696,
      "step": 6669
    },
    {
      "epoch": 1.2222635197918603,
      "grad_norm": 0.7439038753509521,
      "learning_rate": 6.359686921432613e-05,
      "loss": 1.1564,
      "step": 6670
    },
    {
      "epoch": 1.2224493588552314,
      "grad_norm": 0.6198158264160156,
      "learning_rate": 6.356963154558491e-05,
      "loss": 0.9367,
      "step": 6671
    },
    {
      "epoch": 1.2226351979186025,
      "grad_norm": 0.6812259554862976,
      "learning_rate": 6.354239699281186e-05,
      "loss": 0.8498,
      "step": 6672
    },
    {
      "epoch": 1.2228210369819736,
      "grad_norm": 0.6730685234069824,
      "learning_rate": 6.351516555833639e-05,
      "loss": 0.8412,
      "step": 6673
    },
    {
      "epoch": 1.2230068760453447,
      "grad_norm": 0.7179238200187683,
      "learning_rate": 6.348793724448765e-05,
      "loss": 0.9143,
      "step": 6674
    },
    {
      "epoch": 1.2231927151087159,
      "grad_norm": 0.7099092602729797,
      "learning_rate": 6.34607120535946e-05,
      "loss": 1.1273,
      "step": 6675
    },
    {
      "epoch": 1.223378554172087,
      "grad_norm": 0.8993801474571228,
      "learning_rate": 6.343348998798577e-05,
      "loss": 1.2519,
      "step": 6676
    },
    {
      "epoch": 1.223564393235458,
      "grad_norm": 0.7195106744766235,
      "learning_rate": 6.340627104998962e-05,
      "loss": 0.8705,
      "step": 6677
    },
    {
      "epoch": 1.2237502322988292,
      "grad_norm": 0.6673482060432434,
      "learning_rate": 6.337905524193415e-05,
      "loss": 0.9225,
      "step": 6678
    },
    {
      "epoch": 1.2239360713622003,
      "grad_norm": 0.6686134934425354,
      "learning_rate": 6.335184256614723e-05,
      "loss": 0.9436,
      "step": 6679
    },
    {
      "epoch": 1.2241219104255714,
      "grad_norm": 0.7211525440216064,
      "learning_rate": 6.332463302495645e-05,
      "loss": 0.9579,
      "step": 6680
    },
    {
      "epoch": 1.2243077494889425,
      "grad_norm": 0.6387026309967041,
      "learning_rate": 6.329742662068905e-05,
      "loss": 0.5192,
      "step": 6681
    },
    {
      "epoch": 1.2244935885523136,
      "grad_norm": 0.68216472864151,
      "learning_rate": 6.327022335567212e-05,
      "loss": 0.9198,
      "step": 6682
    },
    {
      "epoch": 1.2246794276156847,
      "grad_norm": 0.6553606390953064,
      "learning_rate": 6.324302323223232e-05,
      "loss": 0.95,
      "step": 6683
    },
    {
      "epoch": 1.224865266679056,
      "grad_norm": 0.6823198795318604,
      "learning_rate": 6.321582625269624e-05,
      "loss": 0.9124,
      "step": 6684
    },
    {
      "epoch": 1.2250511057424271,
      "grad_norm": 0.8079013824462891,
      "learning_rate": 6.318863241938999e-05,
      "loss": 1.1514,
      "step": 6685
    },
    {
      "epoch": 1.2252369448057983,
      "grad_norm": 0.681402862071991,
      "learning_rate": 6.316144173463958e-05,
      "loss": 0.9534,
      "step": 6686
    },
    {
      "epoch": 1.2254227838691694,
      "grad_norm": 0.6464089751243591,
      "learning_rate": 6.313425420077069e-05,
      "loss": 0.9807,
      "step": 6687
    },
    {
      "epoch": 1.2256086229325405,
      "grad_norm": 0.7560293078422546,
      "learning_rate": 6.31070698201087e-05,
      "loss": 1.0426,
      "step": 6688
    },
    {
      "epoch": 1.2257944619959116,
      "grad_norm": 0.6318104267120361,
      "learning_rate": 6.307988859497878e-05,
      "loss": 0.8507,
      "step": 6689
    },
    {
      "epoch": 1.2259803010592827,
      "grad_norm": 0.7459145188331604,
      "learning_rate": 6.305271052770576e-05,
      "loss": 1.145,
      "step": 6690
    },
    {
      "epoch": 1.2261661401226538,
      "grad_norm": 0.7620947957038879,
      "learning_rate": 6.302553562061431e-05,
      "loss": 0.863,
      "step": 6691
    },
    {
      "epoch": 1.226351979186025,
      "grad_norm": 0.8354644775390625,
      "learning_rate": 6.299836387602867e-05,
      "loss": 0.9738,
      "step": 6692
    },
    {
      "epoch": 1.226537818249396,
      "grad_norm": 0.6539764404296875,
      "learning_rate": 6.297119529627293e-05,
      "loss": 1.1549,
      "step": 6693
    },
    {
      "epoch": 1.2267236573127671,
      "grad_norm": 0.7981652617454529,
      "learning_rate": 6.294402988367094e-05,
      "loss": 0.9152,
      "step": 6694
    },
    {
      "epoch": 1.2269094963761382,
      "grad_norm": 0.6764546036720276,
      "learning_rate": 6.29168676405461e-05,
      "loss": 0.8726,
      "step": 6695
    },
    {
      "epoch": 1.2270953354395093,
      "grad_norm": 0.6629418730735779,
      "learning_rate": 6.288970856922176e-05,
      "loss": 0.9232,
      "step": 6696
    },
    {
      "epoch": 1.2272811745028804,
      "grad_norm": 0.8247547149658203,
      "learning_rate": 6.286255267202081e-05,
      "loss": 0.8028,
      "step": 6697
    },
    {
      "epoch": 1.2274670135662515,
      "grad_norm": 0.6946619749069214,
      "learning_rate": 6.283539995126607e-05,
      "loss": 0.9486,
      "step": 6698
    },
    {
      "epoch": 1.2276528526296229,
      "grad_norm": 0.6491488218307495,
      "learning_rate": 6.280825040927982e-05,
      "loss": 0.7698,
      "step": 6699
    },
    {
      "epoch": 1.2278386916929938,
      "grad_norm": 0.7405893206596375,
      "learning_rate": 6.27811040483843e-05,
      "loss": 1.0136,
      "step": 6700
    },
    {
      "epoch": 1.228024530756365,
      "grad_norm": 2.1106338500976562,
      "learning_rate": 6.275396087090144e-05,
      "loss": 1.4144,
      "step": 6701
    },
    {
      "epoch": 1.2282103698197362,
      "grad_norm": 0.6475285887718201,
      "learning_rate": 6.272682087915276e-05,
      "loss": 1.0483,
      "step": 6702
    },
    {
      "epoch": 1.2283962088831073,
      "grad_norm": 0.7684204578399658,
      "learning_rate": 6.269968407545968e-05,
      "loss": 1.1889,
      "step": 6703
    },
    {
      "epoch": 1.2285820479464784,
      "grad_norm": 0.7069278955459595,
      "learning_rate": 6.267255046214319e-05,
      "loss": 0.9133,
      "step": 6704
    },
    {
      "epoch": 1.2287678870098495,
      "grad_norm": 0.6581943035125732,
      "learning_rate": 6.264542004152421e-05,
      "loss": 0.8838,
      "step": 6705
    },
    {
      "epoch": 1.2289537260732206,
      "grad_norm": 0.6479682326316833,
      "learning_rate": 6.261829281592314e-05,
      "loss": 0.9119,
      "step": 6706
    },
    {
      "epoch": 1.2291395651365917,
      "grad_norm": 0.7059997916221619,
      "learning_rate": 6.259116878766026e-05,
      "loss": 1.1096,
      "step": 6707
    },
    {
      "epoch": 1.2293254041999628,
      "grad_norm": 0.7317528128623962,
      "learning_rate": 6.256404795905561e-05,
      "loss": 0.8583,
      "step": 6708
    },
    {
      "epoch": 1.229511243263334,
      "grad_norm": 0.6883766651153564,
      "learning_rate": 6.253693033242882e-05,
      "loss": 0.8437,
      "step": 6709
    },
    {
      "epoch": 1.229697082326705,
      "grad_norm": 0.6823121905326843,
      "learning_rate": 6.250981591009936e-05,
      "loss": 1.1293,
      "step": 6710
    },
    {
      "epoch": 1.2298829213900762,
      "grad_norm": 0.5986813902854919,
      "learning_rate": 6.248270469438636e-05,
      "loss": 0.8529,
      "step": 6711
    },
    {
      "epoch": 1.2300687604534473,
      "grad_norm": 0.7856054902076721,
      "learning_rate": 6.245559668760874e-05,
      "loss": 0.9986,
      "step": 6712
    },
    {
      "epoch": 1.2302545995168184,
      "grad_norm": 0.5989887118339539,
      "learning_rate": 6.242849189208504e-05,
      "loss": 1.002,
      "step": 6713
    },
    {
      "epoch": 1.2304404385801895,
      "grad_norm": 0.6663972735404968,
      "learning_rate": 6.240139031013363e-05,
      "loss": 0.9013,
      "step": 6714
    },
    {
      "epoch": 1.2306262776435606,
      "grad_norm": 0.7706762552261353,
      "learning_rate": 6.237429194407263e-05,
      "loss": 1.0902,
      "step": 6715
    },
    {
      "epoch": 1.230812116706932,
      "grad_norm": 0.9056258797645569,
      "learning_rate": 6.234719679621969e-05,
      "loss": 1.3235,
      "step": 6716
    },
    {
      "epoch": 1.2309979557703028,
      "grad_norm": 0.6869669556617737,
      "learning_rate": 6.232010486889242e-05,
      "loss": 1.0833,
      "step": 6717
    },
    {
      "epoch": 1.2311837948336741,
      "grad_norm": 0.7842996120452881,
      "learning_rate": 6.2293016164408e-05,
      "loss": 1.1114,
      "step": 6718
    },
    {
      "epoch": 1.2313696338970452,
      "grad_norm": 0.6470237970352173,
      "learning_rate": 6.226593068508344e-05,
      "loss": 0.9545,
      "step": 6719
    },
    {
      "epoch": 1.2315554729604163,
      "grad_norm": 0.6029546856880188,
      "learning_rate": 6.223884843323534e-05,
      "loss": 1.0115,
      "step": 6720
    },
    {
      "epoch": 1.2317413120237874,
      "grad_norm": 0.6828630566596985,
      "learning_rate": 6.221176941118014e-05,
      "loss": 1.0884,
      "step": 6721
    },
    {
      "epoch": 1.2319271510871586,
      "grad_norm": 0.7872723937034607,
      "learning_rate": 6.218469362123403e-05,
      "loss": 0.9801,
      "step": 6722
    },
    {
      "epoch": 1.2321129901505297,
      "grad_norm": 0.630489706993103,
      "learning_rate": 6.215762106571276e-05,
      "loss": 0.8454,
      "step": 6723
    },
    {
      "epoch": 1.2322988292139008,
      "grad_norm": 0.7333617806434631,
      "learning_rate": 6.213055174693196e-05,
      "loss": 0.825,
      "step": 6724
    },
    {
      "epoch": 1.2324846682772719,
      "grad_norm": 0.8034706711769104,
      "learning_rate": 6.21034856672069e-05,
      "loss": 1.1837,
      "step": 6725
    },
    {
      "epoch": 1.232670507340643,
      "grad_norm": 0.5724234580993652,
      "learning_rate": 6.207642282885267e-05,
      "loss": 1.0768,
      "step": 6726
    },
    {
      "epoch": 1.232856346404014,
      "grad_norm": 0.6139940619468689,
      "learning_rate": 6.20493632341839e-05,
      "loss": 0.8875,
      "step": 6727
    },
    {
      "epoch": 1.2330421854673852,
      "grad_norm": 0.8388956189155579,
      "learning_rate": 6.202230688551511e-05,
      "loss": 0.7057,
      "step": 6728
    },
    {
      "epoch": 1.2332280245307563,
      "grad_norm": 0.7120481133460999,
      "learning_rate": 6.199525378516056e-05,
      "loss": 1.0757,
      "step": 6729
    },
    {
      "epoch": 1.2334138635941274,
      "grad_norm": 0.6384205222129822,
      "learning_rate": 6.196820393543404e-05,
      "loss": 0.9849,
      "step": 6730
    },
    {
      "epoch": 1.2334138635941274,
      "eval_loss": 1.0142569541931152,
      "eval_runtime": 23.213,
      "eval_samples_per_second": 47.043,
      "eval_steps_per_second": 23.521,
      "step": 6730
    },
    {
      "epoch": 1.2335997026574985,
      "grad_norm": 0.6989933252334595,
      "learning_rate": 6.194115733864924e-05,
      "loss": 1.1173,
      "step": 6731
    },
    {
      "epoch": 1.2337855417208696,
      "grad_norm": 0.7638552784919739,
      "learning_rate": 6.191411399711949e-05,
      "loss": 0.9186,
      "step": 6732
    },
    {
      "epoch": 1.233971380784241,
      "grad_norm": 0.608952522277832,
      "learning_rate": 6.188707391315793e-05,
      "loss": 1.0625,
      "step": 6733
    },
    {
      "epoch": 1.234157219847612,
      "grad_norm": 0.7631494998931885,
      "learning_rate": 6.186003708907729e-05,
      "loss": 0.9277,
      "step": 6734
    },
    {
      "epoch": 1.2343430589109832,
      "grad_norm": 0.6393689513206482,
      "learning_rate": 6.183300352719008e-05,
      "loss": 0.8245,
      "step": 6735
    },
    {
      "epoch": 1.2345288979743543,
      "grad_norm": 0.6334121823310852,
      "learning_rate": 6.180597322980863e-05,
      "loss": 0.8125,
      "step": 6736
    },
    {
      "epoch": 1.2347147370377254,
      "grad_norm": 0.6829781532287598,
      "learning_rate": 6.177894619924479e-05,
      "loss": 0.9776,
      "step": 6737
    },
    {
      "epoch": 1.2349005761010965,
      "grad_norm": 0.6575719118118286,
      "learning_rate": 6.175192243781031e-05,
      "loss": 0.9262,
      "step": 6738
    },
    {
      "epoch": 1.2350864151644676,
      "grad_norm": 0.7077308893203735,
      "learning_rate": 6.172490194781657e-05,
      "loss": 0.955,
      "step": 6739
    },
    {
      "epoch": 1.2352722542278387,
      "grad_norm": 0.6063582301139832,
      "learning_rate": 6.169788473157468e-05,
      "loss": 0.7382,
      "step": 6740
    },
    {
      "epoch": 1.2354580932912098,
      "grad_norm": 0.651313066482544,
      "learning_rate": 6.167087079139549e-05,
      "loss": 0.8344,
      "step": 6741
    },
    {
      "epoch": 1.235643932354581,
      "grad_norm": 0.5918038487434387,
      "learning_rate": 6.164386012958953e-05,
      "loss": 1.015,
      "step": 6742
    },
    {
      "epoch": 1.235829771417952,
      "grad_norm": 0.6994732618331909,
      "learning_rate": 6.161685274846718e-05,
      "loss": 0.9654,
      "step": 6743
    },
    {
      "epoch": 1.2360156104813231,
      "grad_norm": 0.6554886698722839,
      "learning_rate": 6.158984865033833e-05,
      "loss": 0.866,
      "step": 6744
    },
    {
      "epoch": 1.2362014495446942,
      "grad_norm": 0.7816451787948608,
      "learning_rate": 6.156284783751277e-05,
      "loss": 0.7126,
      "step": 6745
    },
    {
      "epoch": 1.2363872886080653,
      "grad_norm": 0.6727638244628906,
      "learning_rate": 6.153585031229987e-05,
      "loss": 0.9787,
      "step": 6746
    },
    {
      "epoch": 1.2365731276714365,
      "grad_norm": 0.7645466923713684,
      "learning_rate": 6.150885607700884e-05,
      "loss": 1.1703,
      "step": 6747
    },
    {
      "epoch": 1.2367589667348076,
      "grad_norm": 0.8393552303314209,
      "learning_rate": 6.148186513394852e-05,
      "loss": 0.7639,
      "step": 6748
    },
    {
      "epoch": 1.2369448057981787,
      "grad_norm": 0.6084762215614319,
      "learning_rate": 6.145487748542753e-05,
      "loss": 0.7537,
      "step": 6749
    },
    {
      "epoch": 1.23713064486155,
      "grad_norm": 0.7074626684188843,
      "learning_rate": 6.142789313375422e-05,
      "loss": 0.8683,
      "step": 6750
    },
    {
      "epoch": 1.237316483924921,
      "grad_norm": 0.8581859469413757,
      "learning_rate": 6.140091208123653e-05,
      "loss": 0.857,
      "step": 6751
    },
    {
      "epoch": 1.2375023229882922,
      "grad_norm": 0.6810663938522339,
      "learning_rate": 6.137393433018228e-05,
      "loss": 1.166,
      "step": 6752
    },
    {
      "epoch": 1.2376881620516633,
      "grad_norm": 0.8637286424636841,
      "learning_rate": 6.134695988289889e-05,
      "loss": 0.9974,
      "step": 6753
    },
    {
      "epoch": 1.2378740011150344,
      "grad_norm": 0.6321572661399841,
      "learning_rate": 6.131998874169358e-05,
      "loss": 0.6844,
      "step": 6754
    },
    {
      "epoch": 1.2380598401784055,
      "grad_norm": 0.7431946396827698,
      "learning_rate": 6.12930209088732e-05,
      "loss": 0.8477,
      "step": 6755
    },
    {
      "epoch": 1.2382456792417766,
      "grad_norm": 0.7147607207298279,
      "learning_rate": 6.12660563867444e-05,
      "loss": 0.9962,
      "step": 6756
    },
    {
      "epoch": 1.2384315183051477,
      "grad_norm": 0.7256709337234497,
      "learning_rate": 6.123909517761355e-05,
      "loss": 1.1241,
      "step": 6757
    },
    {
      "epoch": 1.2386173573685189,
      "grad_norm": 0.8592318892478943,
      "learning_rate": 6.121213728378663e-05,
      "loss": 0.9181,
      "step": 6758
    },
    {
      "epoch": 1.23880319643189,
      "grad_norm": 0.693541407585144,
      "learning_rate": 6.118518270756948e-05,
      "loss": 0.9073,
      "step": 6759
    },
    {
      "epoch": 1.238989035495261,
      "grad_norm": 0.6646521091461182,
      "learning_rate": 6.11582314512675e-05,
      "loss": 0.8998,
      "step": 6760
    },
    {
      "epoch": 1.2391748745586322,
      "grad_norm": 0.6509725451469421,
      "learning_rate": 6.113128351718596e-05,
      "loss": 0.8526,
      "step": 6761
    },
    {
      "epoch": 1.2393607136220033,
      "grad_norm": 0.7442356944084167,
      "learning_rate": 6.11043389076297e-05,
      "loss": 0.9261,
      "step": 6762
    },
    {
      "epoch": 1.2395465526853744,
      "grad_norm": 0.693193793296814,
      "learning_rate": 6.107739762490343e-05,
      "loss": 1.0336,
      "step": 6763
    },
    {
      "epoch": 1.2397323917487455,
      "grad_norm": 0.7592474222183228,
      "learning_rate": 6.105045967131149e-05,
      "loss": 1.2086,
      "step": 6764
    },
    {
      "epoch": 1.2399182308121168,
      "grad_norm": 0.7396900653839111,
      "learning_rate": 6.102352504915787e-05,
      "loss": 1.1158,
      "step": 6765
    },
    {
      "epoch": 1.2401040698754877,
      "grad_norm": 0.7751068472862244,
      "learning_rate": 6.0996593760746445e-05,
      "loss": 0.9525,
      "step": 6766
    },
    {
      "epoch": 1.240289908938859,
      "grad_norm": 0.5859838724136353,
      "learning_rate": 6.0969665808380616e-05,
      "loss": 0.9032,
      "step": 6767
    },
    {
      "epoch": 1.2404757480022301,
      "grad_norm": 0.663303017616272,
      "learning_rate": 6.094274119436363e-05,
      "loss": 0.8507,
      "step": 6768
    },
    {
      "epoch": 1.2406615870656013,
      "grad_norm": 0.6255428791046143,
      "learning_rate": 6.0915819920998394e-05,
      "loss": 1.1066,
      "step": 6769
    },
    {
      "epoch": 1.2408474261289724,
      "grad_norm": 0.6422243714332581,
      "learning_rate": 6.0888901990587544e-05,
      "loss": 0.946,
      "step": 6770
    },
    {
      "epoch": 1.2410332651923435,
      "grad_norm": 0.6783014535903931,
      "learning_rate": 6.086198740543347e-05,
      "loss": 0.8705,
      "step": 6771
    },
    {
      "epoch": 1.2412191042557146,
      "grad_norm": 0.5838634967803955,
      "learning_rate": 6.083507616783819e-05,
      "loss": 0.9581,
      "step": 6772
    },
    {
      "epoch": 1.2414049433190857,
      "grad_norm": 0.9073149561882019,
      "learning_rate": 6.0808168280103513e-05,
      "loss": 0.7539,
      "step": 6773
    },
    {
      "epoch": 1.2415907823824568,
      "grad_norm": 0.6188364624977112,
      "learning_rate": 6.078126374453086e-05,
      "loss": 0.9521,
      "step": 6774
    },
    {
      "epoch": 1.241776621445828,
      "grad_norm": 0.6189103722572327,
      "learning_rate": 6.075436256342152e-05,
      "loss": 0.6746,
      "step": 6775
    },
    {
      "epoch": 1.241962460509199,
      "grad_norm": 0.6246206164360046,
      "learning_rate": 6.0727464739076334e-05,
      "loss": 0.8267,
      "step": 6776
    },
    {
      "epoch": 1.2421482995725701,
      "grad_norm": 0.6606687903404236,
      "learning_rate": 6.0700570273795984e-05,
      "loss": 0.7699,
      "step": 6777
    },
    {
      "epoch": 1.2423341386359412,
      "grad_norm": 0.7769799828529358,
      "learning_rate": 6.0673679169880824e-05,
      "loss": 1.0587,
      "step": 6778
    },
    {
      "epoch": 1.2425199776993123,
      "grad_norm": 0.5887297987937927,
      "learning_rate": 6.0646791429630844e-05,
      "loss": 0.8285,
      "step": 6779
    },
    {
      "epoch": 1.2427058167626834,
      "grad_norm": 2.852731704711914,
      "learning_rate": 6.061990705534588e-05,
      "loss": 1.3253,
      "step": 6780
    },
    {
      "epoch": 1.2428916558260545,
      "grad_norm": 0.8008617162704468,
      "learning_rate": 6.059302604932535e-05,
      "loss": 0.9496,
      "step": 6781
    },
    {
      "epoch": 1.2430774948894259,
      "grad_norm": 0.8640806078910828,
      "learning_rate": 6.05661484138685e-05,
      "loss": 1.0751,
      "step": 6782
    },
    {
      "epoch": 1.2432633339527968,
      "grad_norm": 0.6174668073654175,
      "learning_rate": 6.053927415127416e-05,
      "loss": 0.9893,
      "step": 6783
    },
    {
      "epoch": 1.243449173016168,
      "grad_norm": 0.7041611671447754,
      "learning_rate": 6.051240326384102e-05,
      "loss": 0.8107,
      "step": 6784
    },
    {
      "epoch": 1.2436350120795392,
      "grad_norm": 0.6654350161552429,
      "learning_rate": 6.0485535753867364e-05,
      "loss": 0.9742,
      "step": 6785
    },
    {
      "epoch": 1.2438208511429103,
      "grad_norm": 0.7300111651420593,
      "learning_rate": 6.045867162365122e-05,
      "loss": 0.7188,
      "step": 6786
    },
    {
      "epoch": 1.2440066902062814,
      "grad_norm": 0.7048124074935913,
      "learning_rate": 6.043181087549039e-05,
      "loss": 1.0385,
      "step": 6787
    },
    {
      "epoch": 1.2441925292696525,
      "grad_norm": 0.617351770401001,
      "learning_rate": 6.040495351168227e-05,
      "loss": 0.9001,
      "step": 6788
    },
    {
      "epoch": 1.2443783683330236,
      "grad_norm": 0.6402180194854736,
      "learning_rate": 6.037809953452406e-05,
      "loss": 0.7944,
      "step": 6789
    },
    {
      "epoch": 1.2445642073963947,
      "grad_norm": 0.8734924793243408,
      "learning_rate": 6.035124894631263e-05,
      "loss": 0.9348,
      "step": 6790
    },
    {
      "epoch": 1.2447500464597658,
      "grad_norm": 0.5609917640686035,
      "learning_rate": 6.0324401749344553e-05,
      "loss": 0.9492,
      "step": 6791
    },
    {
      "epoch": 1.244935885523137,
      "grad_norm": 0.6634128093719482,
      "learning_rate": 6.029755794591616e-05,
      "loss": 0.8949,
      "step": 6792
    },
    {
      "epoch": 1.245121724586508,
      "grad_norm": 1.507298469543457,
      "learning_rate": 6.0270717538323426e-05,
      "loss": 1.4974,
      "step": 6793
    },
    {
      "epoch": 1.2453075636498792,
      "grad_norm": 0.6983362436294556,
      "learning_rate": 6.024388052886214e-05,
      "loss": 0.8417,
      "step": 6794
    },
    {
      "epoch": 1.2454934027132503,
      "grad_norm": 0.6252066493034363,
      "learning_rate": 6.021704691982761e-05,
      "loss": 1.0541,
      "step": 6795
    },
    {
      "epoch": 1.2456792417766214,
      "grad_norm": 0.5981398224830627,
      "learning_rate": 6.01902167135151e-05,
      "loss": 0.6388,
      "step": 6796
    },
    {
      "epoch": 1.2458650808399925,
      "grad_norm": 0.6744561791419983,
      "learning_rate": 6.0163389912219346e-05,
      "loss": 1.0082,
      "step": 6797
    },
    {
      "epoch": 1.2460509199033636,
      "grad_norm": 0.7297447919845581,
      "learning_rate": 6.013656651823496e-05,
      "loss": 0.8587,
      "step": 6798
    },
    {
      "epoch": 1.246236758966735,
      "grad_norm": 0.7522355914115906,
      "learning_rate": 6.0109746533856214e-05,
      "loss": 1.1065,
      "step": 6799
    },
    {
      "epoch": 1.246422598030106,
      "grad_norm": 0.7894991636276245,
      "learning_rate": 6.0082929961377035e-05,
      "loss": 1.1286,
      "step": 6800
    },
    {
      "epoch": 1.2466084370934771,
      "grad_norm": 0.6512338519096375,
      "learning_rate": 6.0056116803091175e-05,
      "loss": 0.9291,
      "step": 6801
    },
    {
      "epoch": 1.2467942761568482,
      "grad_norm": 0.5885134935379028,
      "learning_rate": 6.002930706129193e-05,
      "loss": 1.0359,
      "step": 6802
    },
    {
      "epoch": 1.2469801152202193,
      "grad_norm": 0.6664581894874573,
      "learning_rate": 6.0002500738272474e-05,
      "loss": 0.8518,
      "step": 6803
    },
    {
      "epoch": 1.2471659542835904,
      "grad_norm": 0.6249924302101135,
      "learning_rate": 5.997569783632555e-05,
      "loss": 0.9104,
      "step": 6804
    },
    {
      "epoch": 1.2473517933469616,
      "grad_norm": 0.706665575504303,
      "learning_rate": 5.994889835774369e-05,
      "loss": 1.1926,
      "step": 6805
    },
    {
      "epoch": 1.2475376324103327,
      "grad_norm": 0.5533158779144287,
      "learning_rate": 5.992210230481913e-05,
      "loss": 0.7645,
      "step": 6806
    },
    {
      "epoch": 1.2477234714737038,
      "grad_norm": 0.6172327399253845,
      "learning_rate": 5.989530967984376e-05,
      "loss": 0.9258,
      "step": 6807
    },
    {
      "epoch": 1.2479093105370749,
      "grad_norm": 0.6446318626403809,
      "learning_rate": 5.9868520485109266e-05,
      "loss": 0.8431,
      "step": 6808
    },
    {
      "epoch": 1.248095149600446,
      "grad_norm": 0.7662816047668457,
      "learning_rate": 5.984173472290693e-05,
      "loss": 1.0121,
      "step": 6809
    },
    {
      "epoch": 1.248280988663817,
      "grad_norm": 0.9095577001571655,
      "learning_rate": 5.981495239552783e-05,
      "loss": 1.0564,
      "step": 6810
    },
    {
      "epoch": 1.2484668277271882,
      "grad_norm": 0.6445630788803101,
      "learning_rate": 5.978817350526268e-05,
      "loss": 0.8561,
      "step": 6811
    },
    {
      "epoch": 1.2486526667905593,
      "grad_norm": 0.7017901539802551,
      "learning_rate": 5.976139805440195e-05,
      "loss": 0.9249,
      "step": 6812
    },
    {
      "epoch": 1.2488385058539304,
      "grad_norm": 0.7360055446624756,
      "learning_rate": 5.973462604523584e-05,
      "loss": 1.0012,
      "step": 6813
    },
    {
      "epoch": 1.2490243449173015,
      "grad_norm": 0.6232417821884155,
      "learning_rate": 5.970785748005416e-05,
      "loss": 0.9935,
      "step": 6814
    },
    {
      "epoch": 1.2492101839806726,
      "grad_norm": 0.6803663372993469,
      "learning_rate": 5.968109236114656e-05,
      "loss": 0.7987,
      "step": 6815
    },
    {
      "epoch": 1.249396023044044,
      "grad_norm": 0.6657290458679199,
      "learning_rate": 5.965433069080224e-05,
      "loss": 1.0063,
      "step": 6816
    },
    {
      "epoch": 1.249581862107415,
      "grad_norm": 0.6280941367149353,
      "learning_rate": 5.962757247131024e-05,
      "loss": 0.9978,
      "step": 6817
    },
    {
      "epoch": 1.2497677011707862,
      "grad_norm": 0.638014554977417,
      "learning_rate": 5.96008177049592e-05,
      "loss": 0.8691,
      "step": 6818
    },
    {
      "epoch": 1.2499535402341573,
      "grad_norm": 0.7058728933334351,
      "learning_rate": 5.957406639403751e-05,
      "loss": 0.7765,
      "step": 6819
    },
    {
      "epoch": 1.2501393792975284,
      "grad_norm": 0.7721114158630371,
      "learning_rate": 5.954731854083333e-05,
      "loss": 0.782,
      "step": 6820
    },
    {
      "epoch": 1.2503252183608995,
      "grad_norm": 0.6177173852920532,
      "learning_rate": 5.95205741476344e-05,
      "loss": 0.9974,
      "step": 6821
    },
    {
      "epoch": 1.2505110574242706,
      "grad_norm": 0.6779488921165466,
      "learning_rate": 5.94938332167283e-05,
      "loss": 0.8565,
      "step": 6822
    },
    {
      "epoch": 1.2506968964876417,
      "grad_norm": 0.7669054865837097,
      "learning_rate": 5.946709575040213e-05,
      "loss": 1.0469,
      "step": 6823
    },
    {
      "epoch": 1.2508827355510128,
      "grad_norm": 0.8133144974708557,
      "learning_rate": 5.944036175094292e-05,
      "loss": 0.5655,
      "step": 6824
    },
    {
      "epoch": 1.251068574614384,
      "grad_norm": 0.6289818286895752,
      "learning_rate": 5.941363122063719e-05,
      "loss": 0.9136,
      "step": 6825
    },
    {
      "epoch": 1.251254413677755,
      "grad_norm": 0.660622239112854,
      "learning_rate": 5.9386904161771295e-05,
      "loss": 0.9953,
      "step": 6826
    },
    {
      "epoch": 1.2514402527411261,
      "grad_norm": 0.6116739511489868,
      "learning_rate": 5.9360180576631285e-05,
      "loss": 0.7881,
      "step": 6827
    },
    {
      "epoch": 1.2516260918044972,
      "grad_norm": 0.7629313468933105,
      "learning_rate": 5.933346046750284e-05,
      "loss": 0.9287,
      "step": 6828
    },
    {
      "epoch": 1.2518119308678683,
      "grad_norm": 0.7508540749549866,
      "learning_rate": 5.930674383667146e-05,
      "loss": 0.8278,
      "step": 6829
    },
    {
      "epoch": 1.2519977699312395,
      "grad_norm": 0.6341012120246887,
      "learning_rate": 5.928003068642218e-05,
      "loss": 0.9499,
      "step": 6830
    },
    {
      "epoch": 1.2521836089946108,
      "grad_norm": 0.8538436889648438,
      "learning_rate": 5.925332101903994e-05,
      "loss": 1.1914,
      "step": 6831
    },
    {
      "epoch": 1.2523694480579817,
      "grad_norm": 0.7098791003227234,
      "learning_rate": 5.922661483680917e-05,
      "loss": 0.8967,
      "step": 6832
    },
    {
      "epoch": 1.252555287121353,
      "grad_norm": 0.7230139374732971,
      "learning_rate": 5.919991214201415e-05,
      "loss": 1.1333,
      "step": 6833
    },
    {
      "epoch": 1.252741126184724,
      "grad_norm": 0.8208269476890564,
      "learning_rate": 5.917321293693885e-05,
      "loss": 0.9973,
      "step": 6834
    },
    {
      "epoch": 1.2529269652480952,
      "grad_norm": 0.6810266375541687,
      "learning_rate": 5.9146517223866885e-05,
      "loss": 0.8249,
      "step": 6835
    },
    {
      "epoch": 1.2531128043114663,
      "grad_norm": 0.6982502937316895,
      "learning_rate": 5.91198250050816e-05,
      "loss": 0.8615,
      "step": 6836
    },
    {
      "epoch": 1.2532986433748374,
      "grad_norm": 0.8237326741218567,
      "learning_rate": 5.909313628286601e-05,
      "loss": 1.2216,
      "step": 6837
    },
    {
      "epoch": 1.2534844824382085,
      "grad_norm": 0.7369891405105591,
      "learning_rate": 5.906645105950293e-05,
      "loss": 1.0332,
      "step": 6838
    },
    {
      "epoch": 1.2536703215015796,
      "grad_norm": 0.8150144815444946,
      "learning_rate": 5.903976933727473e-05,
      "loss": 0.9814,
      "step": 6839
    },
    {
      "epoch": 1.2538561605649508,
      "grad_norm": 0.6736454963684082,
      "learning_rate": 5.901309111846356e-05,
      "loss": 0.8444,
      "step": 6840
    },
    {
      "epoch": 1.2540419996283219,
      "grad_norm": 0.8185814023017883,
      "learning_rate": 5.898641640535135e-05,
      "loss": 0.9539,
      "step": 6841
    },
    {
      "epoch": 1.254227838691693,
      "grad_norm": 0.6659137606620789,
      "learning_rate": 5.895974520021954e-05,
      "loss": 0.8065,
      "step": 6842
    },
    {
      "epoch": 1.254413677755064,
      "grad_norm": 0.6813466548919678,
      "learning_rate": 5.893307750534944e-05,
      "loss": 0.8804,
      "step": 6843
    },
    {
      "epoch": 1.2545995168184352,
      "grad_norm": 0.7067291140556335,
      "learning_rate": 5.890641332302198e-05,
      "loss": 1.0552,
      "step": 6844
    },
    {
      "epoch": 1.2547853558818063,
      "grad_norm": 0.7727149128913879,
      "learning_rate": 5.8879752655517775e-05,
      "loss": 0.909,
      "step": 6845
    },
    {
      "epoch": 1.2549711949451776,
      "grad_norm": 0.7835125923156738,
      "learning_rate": 5.885309550511726e-05,
      "loss": 0.8913,
      "step": 6846
    },
    {
      "epoch": 1.2551570340085485,
      "grad_norm": 0.6659790873527527,
      "learning_rate": 5.8826441874100355e-05,
      "loss": 0.9017,
      "step": 6847
    },
    {
      "epoch": 1.2553428730719198,
      "grad_norm": 0.9084991216659546,
      "learning_rate": 5.879979176474692e-05,
      "loss": 0.8669,
      "step": 6848
    },
    {
      "epoch": 1.2555287121352907,
      "grad_norm": 0.6898093819618225,
      "learning_rate": 5.87731451793363e-05,
      "loss": 0.8462,
      "step": 6849
    },
    {
      "epoch": 1.255714551198662,
      "grad_norm": 0.6940835118293762,
      "learning_rate": 5.8746502120147696e-05,
      "loss": 0.8334,
      "step": 6850
    },
    {
      "epoch": 1.2559003902620332,
      "grad_norm": 0.7318105101585388,
      "learning_rate": 5.8719862589459894e-05,
      "loss": 0.9161,
      "step": 6851
    },
    {
      "epoch": 1.2560862293254043,
      "grad_norm": 0.8148001432418823,
      "learning_rate": 5.869322658955147e-05,
      "loss": 1.1776,
      "step": 6852
    },
    {
      "epoch": 1.2562720683887754,
      "grad_norm": 0.6155508756637573,
      "learning_rate": 5.86665941227007e-05,
      "loss": 1.0434,
      "step": 6853
    },
    {
      "epoch": 1.2564579074521465,
      "grad_norm": 0.6666076183319092,
      "learning_rate": 5.863996519118542e-05,
      "loss": 0.8688,
      "step": 6854
    },
    {
      "epoch": 1.2566437465155176,
      "grad_norm": 0.8823888897895813,
      "learning_rate": 5.861333979728334e-05,
      "loss": 1.1275,
      "step": 6855
    },
    {
      "epoch": 1.2568295855788887,
      "grad_norm": 0.7270902395248413,
      "learning_rate": 5.858671794327173e-05,
      "loss": 0.9495,
      "step": 6856
    },
    {
      "epoch": 1.2570154246422598,
      "grad_norm": 0.7535642385482788,
      "learning_rate": 5.8560099631427655e-05,
      "loss": 1.0634,
      "step": 6857
    },
    {
      "epoch": 1.257201263705631,
      "grad_norm": 0.8336496949195862,
      "learning_rate": 5.85334848640278e-05,
      "loss": 1.0138,
      "step": 6858
    },
    {
      "epoch": 1.257387102769002,
      "grad_norm": 0.6126893162727356,
      "learning_rate": 5.85068736433486e-05,
      "loss": 0.778,
      "step": 6859
    },
    {
      "epoch": 1.2575729418323731,
      "grad_norm": 0.6397730112075806,
      "learning_rate": 5.848026597166622e-05,
      "loss": 0.8947,
      "step": 6860
    },
    {
      "epoch": 1.2577587808957442,
      "grad_norm": 0.7330944538116455,
      "learning_rate": 5.8453661851256395e-05,
      "loss": 1.0755,
      "step": 6861
    },
    {
      "epoch": 1.2579446199591153,
      "grad_norm": 0.824577808380127,
      "learning_rate": 5.84270612843947e-05,
      "loss": 0.9779,
      "step": 6862
    },
    {
      "epoch": 1.2581304590224867,
      "grad_norm": 0.6477912068367004,
      "learning_rate": 5.840046427335626e-05,
      "loss": 0.7855,
      "step": 6863
    },
    {
      "epoch": 1.2583162980858575,
      "grad_norm": 0.6228084564208984,
      "learning_rate": 5.837387082041605e-05,
      "loss": 0.9414,
      "step": 6864
    },
    {
      "epoch": 1.2585021371492289,
      "grad_norm": 0.6834182143211365,
      "learning_rate": 5.83472809278486e-05,
      "loss": 1.0634,
      "step": 6865
    },
    {
      "epoch": 1.2586879762125998,
      "grad_norm": 0.7031686305999756,
      "learning_rate": 5.832069459792825e-05,
      "loss": 0.9155,
      "step": 6866
    },
    {
      "epoch": 1.258873815275971,
      "grad_norm": 0.6353990435600281,
      "learning_rate": 5.829411183292901e-05,
      "loss": 0.8913,
      "step": 6867
    },
    {
      "epoch": 1.2590596543393422,
      "grad_norm": 0.6703015565872192,
      "learning_rate": 5.826753263512448e-05,
      "loss": 0.8506,
      "step": 6868
    },
    {
      "epoch": 1.2592454934027133,
      "grad_norm": 0.6896878480911255,
      "learning_rate": 5.824095700678811e-05,
      "loss": 1.0111,
      "step": 6869
    },
    {
      "epoch": 1.2594313324660844,
      "grad_norm": 0.7479128241539001,
      "learning_rate": 5.821438495019291e-05,
      "loss": 1.0428,
      "step": 6870
    },
    {
      "epoch": 1.2596171715294555,
      "grad_norm": 0.6056257486343384,
      "learning_rate": 5.818781646761171e-05,
      "loss": 0.7879,
      "step": 6871
    },
    {
      "epoch": 1.2598030105928266,
      "grad_norm": 0.5943814516067505,
      "learning_rate": 5.816125156131691e-05,
      "loss": 0.9069,
      "step": 6872
    },
    {
      "epoch": 1.2599888496561977,
      "grad_norm": 0.7249520421028137,
      "learning_rate": 5.813469023358067e-05,
      "loss": 0.7768,
      "step": 6873
    },
    {
      "epoch": 1.2601746887195688,
      "grad_norm": 0.764151394367218,
      "learning_rate": 5.8108132486674904e-05,
      "loss": 0.9072,
      "step": 6874
    },
    {
      "epoch": 1.26036052778294,
      "grad_norm": 0.671393632888794,
      "learning_rate": 5.8081578322871066e-05,
      "loss": 1.0368,
      "step": 6875
    },
    {
      "epoch": 1.260546366846311,
      "grad_norm": 0.8536176085472107,
      "learning_rate": 5.805502774444045e-05,
      "loss": 1.0285,
      "step": 6876
    },
    {
      "epoch": 1.2607322059096822,
      "grad_norm": 0.6377580761909485,
      "learning_rate": 5.802848075365395e-05,
      "loss": 0.8253,
      "step": 6877
    },
    {
      "epoch": 1.2609180449730533,
      "grad_norm": 0.688213586807251,
      "learning_rate": 5.8001937352782235e-05,
      "loss": 1.047,
      "step": 6878
    },
    {
      "epoch": 1.2611038840364244,
      "grad_norm": 0.6255641579627991,
      "learning_rate": 5.7975397544095556e-05,
      "loss": 0.9427,
      "step": 6879
    },
    {
      "epoch": 1.2612897230997957,
      "grad_norm": 0.6308231353759766,
      "learning_rate": 5.7948861329863954e-05,
      "loss": 1.0536,
      "step": 6880
    },
    {
      "epoch": 1.2614755621631666,
      "grad_norm": 0.67474365234375,
      "learning_rate": 5.7922328712357164e-05,
      "loss": 0.9614,
      "step": 6881
    },
    {
      "epoch": 1.261661401226538,
      "grad_norm": 0.6463435888290405,
      "learning_rate": 5.789579969384451e-05,
      "loss": 0.8507,
      "step": 6882
    },
    {
      "epoch": 1.2618472402899088,
      "grad_norm": 0.832118570804596,
      "learning_rate": 5.786927427659514e-05,
      "loss": 1.0154,
      "step": 6883
    },
    {
      "epoch": 1.2620330793532801,
      "grad_norm": 0.658928394317627,
      "learning_rate": 5.784275246287776e-05,
      "loss": 0.7714,
      "step": 6884
    },
    {
      "epoch": 1.2622189184166512,
      "grad_norm": 0.7069569826126099,
      "learning_rate": 5.781623425496091e-05,
      "loss": 1.034,
      "step": 6885
    },
    {
      "epoch": 1.2624047574800223,
      "grad_norm": 0.7191367745399475,
      "learning_rate": 5.778971965511268e-05,
      "loss": 0.9983,
      "step": 6886
    },
    {
      "epoch": 1.2625905965433935,
      "grad_norm": 0.9764262437820435,
      "learning_rate": 5.7763208665600955e-05,
      "loss": 0.9769,
      "step": 6887
    },
    {
      "epoch": 1.2627764356067646,
      "grad_norm": 0.6354040503501892,
      "learning_rate": 5.773670128869332e-05,
      "loss": 0.824,
      "step": 6888
    },
    {
      "epoch": 1.2629622746701357,
      "grad_norm": 0.6636117100715637,
      "learning_rate": 5.771019752665692e-05,
      "loss": 1.008,
      "step": 6889
    },
    {
      "epoch": 1.2631481137335068,
      "grad_norm": 0.6330132484436035,
      "learning_rate": 5.768369738175874e-05,
      "loss": 0.9138,
      "step": 6890
    },
    {
      "epoch": 1.2633339527968779,
      "grad_norm": 0.6355640888214111,
      "learning_rate": 5.765720085626537e-05,
      "loss": 1.1154,
      "step": 6891
    },
    {
      "epoch": 1.263519791860249,
      "grad_norm": 0.6977789402008057,
      "learning_rate": 5.7630707952443165e-05,
      "loss": 0.9318,
      "step": 6892
    },
    {
      "epoch": 1.26370563092362,
      "grad_norm": 0.6667346954345703,
      "learning_rate": 5.7604218672558054e-05,
      "loss": 0.9054,
      "step": 6893
    },
    {
      "epoch": 1.2638914699869912,
      "grad_norm": 0.7689474821090698,
      "learning_rate": 5.757773301887572e-05,
      "loss": 0.9534,
      "step": 6894
    },
    {
      "epoch": 1.2640773090503623,
      "grad_norm": 0.7714675068855286,
      "learning_rate": 5.755125099366163e-05,
      "loss": 1.107,
      "step": 6895
    },
    {
      "epoch": 1.2642631481137334,
      "grad_norm": 0.613943338394165,
      "learning_rate": 5.7524772599180745e-05,
      "loss": 0.7832,
      "step": 6896
    },
    {
      "epoch": 1.2644489871771047,
      "grad_norm": 0.6987224817276001,
      "learning_rate": 5.74982978376979e-05,
      "loss": 0.9577,
      "step": 6897
    },
    {
      "epoch": 1.2646348262404756,
      "grad_norm": 0.7213063836097717,
      "learning_rate": 5.747182671147746e-05,
      "loss": 1.0495,
      "step": 6898
    },
    {
      "epoch": 1.264820665303847,
      "grad_norm": 0.7993615865707397,
      "learning_rate": 5.7445359222783636e-05,
      "loss": 0.912,
      "step": 6899
    },
    {
      "epoch": 1.265006504367218,
      "grad_norm": 0.6639682054519653,
      "learning_rate": 5.7418895373880166e-05,
      "loss": 0.6514,
      "step": 6900
    },
    {
      "epoch": 1.2651923434305892,
      "grad_norm": 0.643459677696228,
      "learning_rate": 5.739243516703061e-05,
      "loss": 0.7299,
      "step": 6901
    },
    {
      "epoch": 1.2653781824939603,
      "grad_norm": 0.7396531105041504,
      "learning_rate": 5.7365978604498194e-05,
      "loss": 0.9067,
      "step": 6902
    },
    {
      "epoch": 1.2655640215573314,
      "grad_norm": 0.6802812814712524,
      "learning_rate": 5.733952568854575e-05,
      "loss": 1.0299,
      "step": 6903
    },
    {
      "epoch": 1.2657498606207025,
      "grad_norm": 0.6861310601234436,
      "learning_rate": 5.731307642143585e-05,
      "loss": 1.0433,
      "step": 6904
    },
    {
      "epoch": 1.2659356996840736,
      "grad_norm": 0.640153169631958,
      "learning_rate": 5.72866308054308e-05,
      "loss": 0.8867,
      "step": 6905
    },
    {
      "epoch": 1.2661215387474447,
      "grad_norm": 0.6789703965187073,
      "learning_rate": 5.726018884279256e-05,
      "loss": 1.0454,
      "step": 6906
    },
    {
      "epoch": 1.2663073778108158,
      "grad_norm": 0.764994204044342,
      "learning_rate": 5.723375053578271e-05,
      "loss": 0.8304,
      "step": 6907
    },
    {
      "epoch": 1.266493216874187,
      "grad_norm": 0.5752626657485962,
      "learning_rate": 5.7207315886662615e-05,
      "loss": 0.6994,
      "step": 6908
    },
    {
      "epoch": 1.266679055937558,
      "grad_norm": 0.5792833566665649,
      "learning_rate": 5.718088489769331e-05,
      "loss": 0.856,
      "step": 6909
    },
    {
      "epoch": 1.2668648950009291,
      "grad_norm": 0.6783587336540222,
      "learning_rate": 5.715445757113542e-05,
      "loss": 0.9642,
      "step": 6910
    },
    {
      "epoch": 1.2670507340643002,
      "grad_norm": 0.7789669036865234,
      "learning_rate": 5.712803390924941e-05,
      "loss": 0.9248,
      "step": 6911
    },
    {
      "epoch": 1.2672365731276716,
      "grad_norm": 0.6985512375831604,
      "learning_rate": 5.7101613914295296e-05,
      "loss": 1.1153,
      "step": 6912
    },
    {
      "epoch": 1.2674224121910425,
      "grad_norm": 0.7611011862754822,
      "learning_rate": 5.707519758853288e-05,
      "loss": 0.915,
      "step": 6913
    },
    {
      "epoch": 1.2676082512544138,
      "grad_norm": 0.6558472514152527,
      "learning_rate": 5.704878493422156e-05,
      "loss": 0.7643,
      "step": 6914
    },
    {
      "epoch": 1.2677940903177847,
      "grad_norm": 0.6821544766426086,
      "learning_rate": 5.70223759536205e-05,
      "loss": 1.0508,
      "step": 6915
    },
    {
      "epoch": 1.267979929381156,
      "grad_norm": 0.7058600783348083,
      "learning_rate": 5.699597064898854e-05,
      "loss": 0.9486,
      "step": 6916
    },
    {
      "epoch": 1.268165768444527,
      "grad_norm": 0.6448752880096436,
      "learning_rate": 5.696956902258413e-05,
      "loss": 0.7647,
      "step": 6917
    },
    {
      "epoch": 1.2683516075078982,
      "grad_norm": 0.7937239408493042,
      "learning_rate": 5.694317107666549e-05,
      "loss": 0.8255,
      "step": 6918
    },
    {
      "epoch": 1.2685374465712693,
      "grad_norm": 0.6356651186943054,
      "learning_rate": 5.691677681349048e-05,
      "loss": 0.8717,
      "step": 6919
    },
    {
      "epoch": 1.2687232856346404,
      "grad_norm": 0.6263054013252258,
      "learning_rate": 5.689038623531664e-05,
      "loss": 0.8607,
      "step": 6920
    },
    {
      "epoch": 1.2689091246980115,
      "grad_norm": 0.6370097994804382,
      "learning_rate": 5.686399934440125e-05,
      "loss": 0.922,
      "step": 6921
    },
    {
      "epoch": 1.2690949637613826,
      "grad_norm": 0.5742283463478088,
      "learning_rate": 5.683761614300123e-05,
      "loss": 0.7138,
      "step": 6922
    },
    {
      "epoch": 1.2692808028247538,
      "grad_norm": 0.673902690410614,
      "learning_rate": 5.6811236633373224e-05,
      "loss": 1.0157,
      "step": 6923
    },
    {
      "epoch": 1.2694666418881249,
      "grad_norm": 0.7027066349983215,
      "learning_rate": 5.678486081777343e-05,
      "loss": 0.4939,
      "step": 6924
    },
    {
      "epoch": 1.269652480951496,
      "grad_norm": 0.6953238844871521,
      "learning_rate": 5.675848869845796e-05,
      "loss": 0.9613,
      "step": 6925
    },
    {
      "epoch": 1.269838320014867,
      "grad_norm": 0.6747790575027466,
      "learning_rate": 5.673212027768236e-05,
      "loss": 0.8952,
      "step": 6926
    },
    {
      "epoch": 1.2700241590782382,
      "grad_norm": 0.6629246473312378,
      "learning_rate": 5.6705755557702056e-05,
      "loss": 0.7611,
      "step": 6927
    },
    {
      "epoch": 1.2702099981416093,
      "grad_norm": 0.644935667514801,
      "learning_rate": 5.6679394540772026e-05,
      "loss": 1.0273,
      "step": 6928
    },
    {
      "epoch": 1.2703958372049806,
      "grad_norm": 0.6592975854873657,
      "learning_rate": 5.6653037229147e-05,
      "loss": 1.0137,
      "step": 6929
    },
    {
      "epoch": 1.2705816762683515,
      "grad_norm": 0.749384880065918,
      "learning_rate": 5.662668362508143e-05,
      "loss": 0.9143,
      "step": 6930
    },
    {
      "epoch": 1.2707675153317228,
      "grad_norm": 0.8060128092765808,
      "learning_rate": 5.6600333730829315e-05,
      "loss": 0.9492,
      "step": 6931
    },
    {
      "epoch": 1.2709533543950937,
      "grad_norm": 0.7280310392379761,
      "learning_rate": 5.657398754864448e-05,
      "loss": 0.9502,
      "step": 6932
    },
    {
      "epoch": 1.271139193458465,
      "grad_norm": 0.7918939590454102,
      "learning_rate": 5.654764508078032e-05,
      "loss": 0.8695,
      "step": 6933
    },
    {
      "epoch": 1.2713250325218362,
      "grad_norm": 0.5838748216629028,
      "learning_rate": 5.6521306329489976e-05,
      "loss": 0.7413,
      "step": 6934
    },
    {
      "epoch": 1.2715108715852073,
      "grad_norm": 0.7349932193756104,
      "learning_rate": 5.649497129702628e-05,
      "loss": 1.0556,
      "step": 6935
    },
    {
      "epoch": 1.2716967106485784,
      "grad_norm": 0.6360459923744202,
      "learning_rate": 5.6468639985641714e-05,
      "loss": 0.8109,
      "step": 6936
    },
    {
      "epoch": 1.2718825497119495,
      "grad_norm": 0.6897345185279846,
      "learning_rate": 5.64423123975885e-05,
      "loss": 0.8486,
      "step": 6937
    },
    {
      "epoch": 1.2720683887753206,
      "grad_norm": 1.3575351238250732,
      "learning_rate": 5.64159885351184e-05,
      "loss": 1.4462,
      "step": 6938
    },
    {
      "epoch": 1.2722542278386917,
      "grad_norm": 0.7821032404899597,
      "learning_rate": 5.638966840048303e-05,
      "loss": 0.878,
      "step": 6939
    },
    {
      "epoch": 1.2724400669020628,
      "grad_norm": 0.6788009405136108,
      "learning_rate": 5.6363351995933556e-05,
      "loss": 0.863,
      "step": 6940
    },
    {
      "epoch": 1.272625905965434,
      "grad_norm": 0.6580929756164551,
      "learning_rate": 5.633703932372092e-05,
      "loss": 1.1638,
      "step": 6941
    },
    {
      "epoch": 1.272811745028805,
      "grad_norm": 0.7011153101921082,
      "learning_rate": 5.631073038609565e-05,
      "loss": 0.958,
      "step": 6942
    },
    {
      "epoch": 1.2729975840921761,
      "grad_norm": 0.6621928215026855,
      "learning_rate": 5.6284425185308035e-05,
      "loss": 0.9041,
      "step": 6943
    },
    {
      "epoch": 1.2731834231555472,
      "grad_norm": 0.6394334435462952,
      "learning_rate": 5.625812372360805e-05,
      "loss": 0.8612,
      "step": 6944
    },
    {
      "epoch": 1.2733692622189183,
      "grad_norm": 0.7178499102592468,
      "learning_rate": 5.623182600324524e-05,
      "loss": 1.0288,
      "step": 6945
    },
    {
      "epoch": 1.2735551012822897,
      "grad_norm": 0.6983404159545898,
      "learning_rate": 5.620553202646899e-05,
      "loss": 0.7361,
      "step": 6946
    },
    {
      "epoch": 1.2737409403456605,
      "grad_norm": 0.6791266798973083,
      "learning_rate": 5.6179241795528206e-05,
      "loss": 0.995,
      "step": 6947
    },
    {
      "epoch": 1.2739267794090319,
      "grad_norm": 0.761229932308197,
      "learning_rate": 5.615295531267156e-05,
      "loss": 1.0264,
      "step": 6948
    },
    {
      "epoch": 1.274112618472403,
      "grad_norm": 0.6744269728660583,
      "learning_rate": 5.612667258014743e-05,
      "loss": 0.8984,
      "step": 6949
    },
    {
      "epoch": 1.274298457535774,
      "grad_norm": 0.6982648968696594,
      "learning_rate": 5.61003936002038e-05,
      "loss": 0.8923,
      "step": 6950
    },
    {
      "epoch": 1.2744842965991452,
      "grad_norm": 0.6110215783119202,
      "learning_rate": 5.6074118375088444e-05,
      "loss": 0.9046,
      "step": 6951
    },
    {
      "epoch": 1.2746701356625163,
      "grad_norm": 0.7817128300666809,
      "learning_rate": 5.604784690704863e-05,
      "loss": 0.8979,
      "step": 6952
    },
    {
      "epoch": 1.2748559747258874,
      "grad_norm": 0.6842471361160278,
      "learning_rate": 5.6021579198331506e-05,
      "loss": 0.8434,
      "step": 6953
    },
    {
      "epoch": 1.2750418137892585,
      "grad_norm": 0.7033209800720215,
      "learning_rate": 5.5995315251183734e-05,
      "loss": 0.875,
      "step": 6954
    },
    {
      "epoch": 1.2752276528526296,
      "grad_norm": 0.5664873123168945,
      "learning_rate": 5.5969055067851774e-05,
      "loss": 0.7992,
      "step": 6955
    },
    {
      "epoch": 1.2754134919160007,
      "grad_norm": 0.6672083735466003,
      "learning_rate": 5.594279865058168e-05,
      "loss": 0.6178,
      "step": 6956
    },
    {
      "epoch": 1.2755993309793718,
      "grad_norm": 0.6838880777359009,
      "learning_rate": 5.591654600161922e-05,
      "loss": 1.0042,
      "step": 6957
    },
    {
      "epoch": 1.275785170042743,
      "grad_norm": 0.7843611836433411,
      "learning_rate": 5.58902971232099e-05,
      "loss": 0.9151,
      "step": 6958
    },
    {
      "epoch": 1.275971009106114,
      "grad_norm": 0.7135986685752869,
      "learning_rate": 5.586405201759876e-05,
      "loss": 1.1861,
      "step": 6959
    },
    {
      "epoch": 1.2761568481694852,
      "grad_norm": 0.7255387902259827,
      "learning_rate": 5.5837810687030677e-05,
      "loss": 0.9878,
      "step": 6960
    },
    {
      "epoch": 1.2763426872328563,
      "grad_norm": 0.7054786682128906,
      "learning_rate": 5.581157313375006e-05,
      "loss": 0.8925,
      "step": 6961
    },
    {
      "epoch": 1.2765285262962274,
      "grad_norm": 1.7965220212936401,
      "learning_rate": 5.57853393600011e-05,
      "loss": 1.269,
      "step": 6962
    },
    {
      "epoch": 1.2767143653595987,
      "grad_norm": 1.0769907236099243,
      "learning_rate": 5.575910936802766e-05,
      "loss": 1.0617,
      "step": 6963
    },
    {
      "epoch": 1.2769002044229696,
      "grad_norm": 0.7031131386756897,
      "learning_rate": 5.573288316007317e-05,
      "loss": 1.3079,
      "step": 6964
    },
    {
      "epoch": 1.277086043486341,
      "grad_norm": 1.0076839923858643,
      "learning_rate": 5.5706660738380864e-05,
      "loss": 0.8868,
      "step": 6965
    },
    {
      "epoch": 1.277271882549712,
      "grad_norm": 0.681121289730072,
      "learning_rate": 5.5680442105193596e-05,
      "loss": 1.0442,
      "step": 6966
    },
    {
      "epoch": 1.2774577216130831,
      "grad_norm": 0.8337486982345581,
      "learning_rate": 5.565422726275393e-05,
      "loss": 1.0483,
      "step": 6967
    },
    {
      "epoch": 1.2776435606764542,
      "grad_norm": 0.803463339805603,
      "learning_rate": 5.5628016213304025e-05,
      "loss": 0.7463,
      "step": 6968
    },
    {
      "epoch": 1.2778293997398253,
      "grad_norm": 0.7833393216133118,
      "learning_rate": 5.5601808959085833e-05,
      "loss": 1.0673,
      "step": 6969
    },
    {
      "epoch": 1.2780152388031965,
      "grad_norm": 0.7391160726547241,
      "learning_rate": 5.557560550234082e-05,
      "loss": 0.7825,
      "step": 6970
    },
    {
      "epoch": 1.2782010778665676,
      "grad_norm": 0.7240591645240784,
      "learning_rate": 5.55494058453103e-05,
      "loss": 1.225,
      "step": 6971
    },
    {
      "epoch": 1.2783869169299387,
      "grad_norm": 0.6331393122673035,
      "learning_rate": 5.552320999023521e-05,
      "loss": 1.0397,
      "step": 6972
    },
    {
      "epoch": 1.2785727559933098,
      "grad_norm": 0.66135174036026,
      "learning_rate": 5.549701793935604e-05,
      "loss": 0.8896,
      "step": 6973
    },
    {
      "epoch": 1.2787585950566809,
      "grad_norm": 0.6168441772460938,
      "learning_rate": 5.547082969491317e-05,
      "loss": 0.8593,
      "step": 6974
    },
    {
      "epoch": 1.278944434120052,
      "grad_norm": 0.7322728633880615,
      "learning_rate": 5.544464525914642e-05,
      "loss": 1.1813,
      "step": 6975
    },
    {
      "epoch": 1.279130273183423,
      "grad_norm": 0.5877998471260071,
      "learning_rate": 5.54184646342955e-05,
      "loss": 0.8917,
      "step": 6976
    },
    {
      "epoch": 1.2793161122467942,
      "grad_norm": 0.6502416133880615,
      "learning_rate": 5.539228782259962e-05,
      "loss": 0.8138,
      "step": 6977
    },
    {
      "epoch": 1.2795019513101655,
      "grad_norm": 0.7527049779891968,
      "learning_rate": 5.536611482629777e-05,
      "loss": 1.1709,
      "step": 6978
    },
    {
      "epoch": 1.2796877903735364,
      "grad_norm": 0.7156680226325989,
      "learning_rate": 5.533994564762859e-05,
      "loss": 0.8171,
      "step": 6979
    },
    {
      "epoch": 1.2798736294369077,
      "grad_norm": 0.7617798447608948,
      "learning_rate": 5.5313780288830365e-05,
      "loss": 0.8767,
      "step": 6980
    },
    {
      "epoch": 1.2800594685002786,
      "grad_norm": 0.7092217206954956,
      "learning_rate": 5.5287618752141145e-05,
      "loss": 1.0576,
      "step": 6981
    },
    {
      "epoch": 1.28024530756365,
      "grad_norm": 0.6281307339668274,
      "learning_rate": 5.526146103979849e-05,
      "loss": 0.9587,
      "step": 6982
    },
    {
      "epoch": 1.280431146627021,
      "grad_norm": 0.6523799300193787,
      "learning_rate": 5.5235307154039794e-05,
      "loss": 0.9439,
      "step": 6983
    },
    {
      "epoch": 1.2806169856903922,
      "grad_norm": 0.6482586860656738,
      "learning_rate": 5.5209157097101996e-05,
      "loss": 0.9628,
      "step": 6984
    },
    {
      "epoch": 1.2808028247537633,
      "grad_norm": 0.7241567373275757,
      "learning_rate": 5.518301087122179e-05,
      "loss": 0.9484,
      "step": 6985
    },
    {
      "epoch": 1.2809886638171344,
      "grad_norm": 0.8451587557792664,
      "learning_rate": 5.5156868478635584e-05,
      "loss": 1.1197,
      "step": 6986
    },
    {
      "epoch": 1.2811745028805055,
      "grad_norm": 0.6632991433143616,
      "learning_rate": 5.5130729921579283e-05,
      "loss": 0.8533,
      "step": 6987
    },
    {
      "epoch": 1.2813603419438766,
      "grad_norm": 0.6490793228149414,
      "learning_rate": 5.5104595202288667e-05,
      "loss": 1.086,
      "step": 6988
    },
    {
      "epoch": 1.2815461810072477,
      "grad_norm": 0.705800473690033,
      "learning_rate": 5.507846432299901e-05,
      "loss": 0.8109,
      "step": 6989
    },
    {
      "epoch": 1.2817320200706188,
      "grad_norm": 0.6534713506698608,
      "learning_rate": 5.505233728594543e-05,
      "loss": 0.9947,
      "step": 6990
    },
    {
      "epoch": 1.28191785913399,
      "grad_norm": 0.7074917554855347,
      "learning_rate": 5.5026214093362563e-05,
      "loss": 0.9424,
      "step": 6991
    },
    {
      "epoch": 1.282103698197361,
      "grad_norm": 0.6349173188209534,
      "learning_rate": 5.500009474748479e-05,
      "loss": 0.9836,
      "step": 6992
    },
    {
      "epoch": 1.2822895372607321,
      "grad_norm": 0.8655628561973572,
      "learning_rate": 5.497397925054617e-05,
      "loss": 0.8945,
      "step": 6993
    },
    {
      "epoch": 1.2824753763241032,
      "grad_norm": 0.8330060839653015,
      "learning_rate": 5.4947867604780426e-05,
      "loss": 0.9298,
      "step": 6994
    },
    {
      "epoch": 1.2826612153874746,
      "grad_norm": 0.7203959226608276,
      "learning_rate": 5.492175981242097e-05,
      "loss": 1.025,
      "step": 6995
    },
    {
      "epoch": 1.2828470544508455,
      "grad_norm": 0.6714663505554199,
      "learning_rate": 5.489565587570078e-05,
      "loss": 0.8477,
      "step": 6996
    },
    {
      "epoch": 1.2830328935142168,
      "grad_norm": 0.7011078596115112,
      "learning_rate": 5.486955579685268e-05,
      "loss": 0.953,
      "step": 6997
    },
    {
      "epoch": 1.2832187325775877,
      "grad_norm": 0.6785321831703186,
      "learning_rate": 5.484345957810897e-05,
      "loss": 1.0602,
      "step": 6998
    },
    {
      "epoch": 1.283404571640959,
      "grad_norm": 0.6448400616645813,
      "learning_rate": 5.4817367221701744e-05,
      "loss": 0.9111,
      "step": 6999
    },
    {
      "epoch": 1.28359041070433,
      "grad_norm": 0.7090216279029846,
      "learning_rate": 5.47912787298628e-05,
      "loss": 0.8572,
      "step": 7000
    },
    {
      "epoch": 1.2837762497677012,
      "grad_norm": 0.7504491209983826,
      "learning_rate": 5.476519410482346e-05,
      "loss": 0.8119,
      "step": 7001
    },
    {
      "epoch": 1.2839620888310723,
      "grad_norm": 0.5962739586830139,
      "learning_rate": 5.473911334881488e-05,
      "loss": 0.6943,
      "step": 7002
    },
    {
      "epoch": 1.2841479278944434,
      "grad_norm": 0.8006013035774231,
      "learning_rate": 5.47130364640677e-05,
      "loss": 1.0141,
      "step": 7003
    },
    {
      "epoch": 1.2843337669578145,
      "grad_norm": 0.718031644821167,
      "learning_rate": 5.4686963452812435e-05,
      "loss": 1.0099,
      "step": 7004
    },
    {
      "epoch": 1.2845196060211856,
      "grad_norm": 0.6614823937416077,
      "learning_rate": 5.466089431727909e-05,
      "loss": 0.915,
      "step": 7005
    },
    {
      "epoch": 1.2847054450845568,
      "grad_norm": 0.6416723132133484,
      "learning_rate": 5.463482905969743e-05,
      "loss": 0.9487,
      "step": 7006
    },
    {
      "epoch": 1.2848912841479279,
      "grad_norm": 0.6111795902252197,
      "learning_rate": 5.46087676822969e-05,
      "loss": 0.876,
      "step": 7007
    },
    {
      "epoch": 1.285077123211299,
      "grad_norm": 0.8280202150344849,
      "learning_rate": 5.458271018730661e-05,
      "loss": 1.1683,
      "step": 7008
    },
    {
      "epoch": 1.28526296227467,
      "grad_norm": 0.7306656241416931,
      "learning_rate": 5.455665657695524e-05,
      "loss": 1.0487,
      "step": 7009
    },
    {
      "epoch": 1.2854488013380412,
      "grad_norm": 0.6812098622322083,
      "learning_rate": 5.453060685347123e-05,
      "loss": 0.9802,
      "step": 7010
    },
    {
      "epoch": 1.2856346404014123,
      "grad_norm": 0.7024182081222534,
      "learning_rate": 5.4504561019082744e-05,
      "loss": 0.9556,
      "step": 7011
    },
    {
      "epoch": 1.2858204794647836,
      "grad_norm": 0.6535548567771912,
      "learning_rate": 5.447851907601744e-05,
      "loss": 1.0563,
      "step": 7012
    },
    {
      "epoch": 1.2860063185281545,
      "grad_norm": 0.6057627201080322,
      "learning_rate": 5.445248102650279e-05,
      "loss": 0.831,
      "step": 7013
    },
    {
      "epoch": 1.2861921575915258,
      "grad_norm": 0.7243484258651733,
      "learning_rate": 5.4426446872765914e-05,
      "loss": 1.258,
      "step": 7014
    },
    {
      "epoch": 1.286377996654897,
      "grad_norm": 0.6408053636550903,
      "learning_rate": 5.440041661703349e-05,
      "loss": 0.8989,
      "step": 7015
    },
    {
      "epoch": 1.286563835718268,
      "grad_norm": 0.6918326020240784,
      "learning_rate": 5.437439026153204e-05,
      "loss": 0.8399,
      "step": 7016
    },
    {
      "epoch": 1.2867496747816392,
      "grad_norm": 0.6725391149520874,
      "learning_rate": 5.434836780848755e-05,
      "loss": 1.0227,
      "step": 7017
    },
    {
      "epoch": 1.2869355138450103,
      "grad_norm": 0.8167325258255005,
      "learning_rate": 5.432234926012586e-05,
      "loss": 0.9492,
      "step": 7018
    },
    {
      "epoch": 1.2871213529083814,
      "grad_norm": 0.6119495630264282,
      "learning_rate": 5.429633461867234e-05,
      "loss": 0.9002,
      "step": 7019
    },
    {
      "epoch": 1.2873071919717525,
      "grad_norm": 0.5399630665779114,
      "learning_rate": 5.427032388635208e-05,
      "loss": 0.7491,
      "step": 7020
    },
    {
      "epoch": 1.2874930310351236,
      "grad_norm": 0.7530606389045715,
      "learning_rate": 5.4244317065389905e-05,
      "loss": 0.9089,
      "step": 7021
    },
    {
      "epoch": 1.2876788700984947,
      "grad_norm": 0.6615011096000671,
      "learning_rate": 5.421831415801013e-05,
      "loss": 0.6998,
      "step": 7022
    },
    {
      "epoch": 1.2878647091618658,
      "grad_norm": 0.7424286007881165,
      "learning_rate": 5.41923151664369e-05,
      "loss": 1.0016,
      "step": 7023
    },
    {
      "epoch": 1.288050548225237,
      "grad_norm": 0.7346042394638062,
      "learning_rate": 5.4166320092893966e-05,
      "loss": 0.8475,
      "step": 7024
    },
    {
      "epoch": 1.288236387288608,
      "grad_norm": 0.6558862924575806,
      "learning_rate": 5.414032893960477e-05,
      "loss": 0.7523,
      "step": 7025
    },
    {
      "epoch": 1.2884222263519791,
      "grad_norm": 0.6409276723861694,
      "learning_rate": 5.41143417087923e-05,
      "loss": 1.0147,
      "step": 7026
    },
    {
      "epoch": 1.2886080654153502,
      "grad_norm": 0.6892153024673462,
      "learning_rate": 5.4088358402679376e-05,
      "loss": 0.9573,
      "step": 7027
    },
    {
      "epoch": 1.2887939044787213,
      "grad_norm": 0.6437703967094421,
      "learning_rate": 5.406237902348841e-05,
      "loss": 0.9284,
      "step": 7028
    },
    {
      "epoch": 1.2889797435420927,
      "grad_norm": 0.637093186378479,
      "learning_rate": 5.403640357344142e-05,
      "loss": 0.7124,
      "step": 7029
    },
    {
      "epoch": 1.2891655826054635,
      "grad_norm": 0.6690963506698608,
      "learning_rate": 5.401043205476021e-05,
      "loss": 0.862,
      "step": 7030
    },
    {
      "epoch": 1.2893514216688349,
      "grad_norm": 0.756441056728363,
      "learning_rate": 5.3984464469666094e-05,
      "loss": 0.9872,
      "step": 7031
    },
    {
      "epoch": 1.289537260732206,
      "grad_norm": 0.6815423369407654,
      "learning_rate": 5.3958500820380244e-05,
      "loss": 1.0385,
      "step": 7032
    },
    {
      "epoch": 1.289723099795577,
      "grad_norm": 0.7535609602928162,
      "learning_rate": 5.3932541109123266e-05,
      "loss": 1.1624,
      "step": 7033
    },
    {
      "epoch": 1.2899089388589482,
      "grad_norm": 1.311153769493103,
      "learning_rate": 5.390658533811563e-05,
      "loss": 1.3717,
      "step": 7034
    },
    {
      "epoch": 1.2900947779223193,
      "grad_norm": 0.6109252572059631,
      "learning_rate": 5.3880633509577404e-05,
      "loss": 0.8028,
      "step": 7035
    },
    {
      "epoch": 1.2902806169856904,
      "grad_norm": 0.7281230092048645,
      "learning_rate": 5.385468562572823e-05,
      "loss": 1.0965,
      "step": 7036
    },
    {
      "epoch": 1.2904664560490615,
      "grad_norm": 0.6285109519958496,
      "learning_rate": 5.382874168878753e-05,
      "loss": 0.8566,
      "step": 7037
    },
    {
      "epoch": 1.2906522951124326,
      "grad_norm": 0.7154156565666199,
      "learning_rate": 5.3802801700974335e-05,
      "loss": 1.0242,
      "step": 7038
    },
    {
      "epoch": 1.2908381341758037,
      "grad_norm": 0.7235013246536255,
      "learning_rate": 5.377686566450741e-05,
      "loss": 1.2717,
      "step": 7039
    },
    {
      "epoch": 1.2910239732391748,
      "grad_norm": 0.7788733243942261,
      "learning_rate": 5.375093358160501e-05,
      "loss": 1.1329,
      "step": 7040
    },
    {
      "epoch": 1.291209812302546,
      "grad_norm": 0.6813734769821167,
      "learning_rate": 5.372500545448523e-05,
      "loss": 1.0213,
      "step": 7041
    },
    {
      "epoch": 1.291395651365917,
      "grad_norm": 0.6252562403678894,
      "learning_rate": 5.369908128536578e-05,
      "loss": 0.883,
      "step": 7042
    },
    {
      "epoch": 1.2915814904292882,
      "grad_norm": 0.7074937224388123,
      "learning_rate": 5.367316107646394e-05,
      "loss": 0.8458,
      "step": 7043
    },
    {
      "epoch": 1.2917673294926595,
      "grad_norm": 0.628176748752594,
      "learning_rate": 5.3647244829996815e-05,
      "loss": 0.9994,
      "step": 7044
    },
    {
      "epoch": 1.2919531685560304,
      "grad_norm": 0.6965638399124146,
      "learning_rate": 5.3621332548180955e-05,
      "loss": 0.9085,
      "step": 7045
    },
    {
      "epoch": 1.2921390076194017,
      "grad_norm": 0.7106294631958008,
      "learning_rate": 5.35954242332328e-05,
      "loss": 1.1325,
      "step": 7046
    },
    {
      "epoch": 1.2923248466827726,
      "grad_norm": 0.623037576675415,
      "learning_rate": 5.356951988736828e-05,
      "loss": 0.9159,
      "step": 7047
    },
    {
      "epoch": 1.292510685746144,
      "grad_norm": 0.6270564794540405,
      "learning_rate": 5.354361951280307e-05,
      "loss": 0.9833,
      "step": 7048
    },
    {
      "epoch": 1.292696524809515,
      "grad_norm": 0.6747012734413147,
      "learning_rate": 5.351772311175253e-05,
      "loss": 0.9795,
      "step": 7049
    },
    {
      "epoch": 1.2928823638728861,
      "grad_norm": 0.7606438994407654,
      "learning_rate": 5.349183068643154e-05,
      "loss": 1.018,
      "step": 7050
    },
    {
      "epoch": 1.2930682029362572,
      "grad_norm": 0.7697432041168213,
      "learning_rate": 5.34659422390548e-05,
      "loss": 1.0029,
      "step": 7051
    },
    {
      "epoch": 1.2932540419996283,
      "grad_norm": 0.7591094374656677,
      "learning_rate": 5.3440057771836594e-05,
      "loss": 1.0419,
      "step": 7052
    },
    {
      "epoch": 1.2934398810629995,
      "grad_norm": 0.6181617379188538,
      "learning_rate": 5.3414177286990915e-05,
      "loss": 0.8152,
      "step": 7053
    },
    {
      "epoch": 1.2936257201263706,
      "grad_norm": 0.7263724207878113,
      "learning_rate": 5.3388300786731295e-05,
      "loss": 0.9096,
      "step": 7054
    },
    {
      "epoch": 1.2938115591897417,
      "grad_norm": 0.8252272605895996,
      "learning_rate": 5.336242827327105e-05,
      "loss": 1.0229,
      "step": 7055
    },
    {
      "epoch": 1.2939973982531128,
      "grad_norm": 0.7154651880264282,
      "learning_rate": 5.333655974882316e-05,
      "loss": 1.0228,
      "step": 7056
    },
    {
      "epoch": 1.2941832373164839,
      "grad_norm": 0.7881542444229126,
      "learning_rate": 5.331069521560012e-05,
      "loss": 1.0496,
      "step": 7057
    },
    {
      "epoch": 1.294369076379855,
      "grad_norm": 0.6441651582717896,
      "learning_rate": 5.3284834675814264e-05,
      "loss": 0.829,
      "step": 7058
    },
    {
      "epoch": 1.294554915443226,
      "grad_norm": 0.7924808859825134,
      "learning_rate": 5.325897813167742e-05,
      "loss": 0.9419,
      "step": 7059
    },
    {
      "epoch": 1.2947407545065972,
      "grad_norm": 0.7063006162643433,
      "learning_rate": 5.323312558540125e-05,
      "loss": 1.0371,
      "step": 7060
    },
    {
      "epoch": 1.2949265935699685,
      "grad_norm": 0.655428409576416,
      "learning_rate": 5.320727703919688e-05,
      "loss": 0.8054,
      "step": 7061
    },
    {
      "epoch": 1.2951124326333394,
      "grad_norm": 0.6543022394180298,
      "learning_rate": 5.318143249527523e-05,
      "loss": 0.8574,
      "step": 7062
    },
    {
      "epoch": 1.2952982716967107,
      "grad_norm": 0.6304433345794678,
      "learning_rate": 5.31555919558469e-05,
      "loss": 0.6496,
      "step": 7063
    },
    {
      "epoch": 1.2954841107600816,
      "grad_norm": 0.7286331653594971,
      "learning_rate": 5.3129755423121984e-05,
      "loss": 1.1228,
      "step": 7064
    },
    {
      "epoch": 1.295669949823453,
      "grad_norm": 0.6797538995742798,
      "learning_rate": 5.310392289931043e-05,
      "loss": 0.9312,
      "step": 7065
    },
    {
      "epoch": 1.295855788886824,
      "grad_norm": 0.708625316619873,
      "learning_rate": 5.307809438662167e-05,
      "loss": 1.0594,
      "step": 7066
    },
    {
      "epoch": 1.2960416279501952,
      "grad_norm": 0.7643126249313354,
      "learning_rate": 5.305226988726491e-05,
      "loss": 0.9271,
      "step": 7067
    },
    {
      "epoch": 1.2962274670135663,
      "grad_norm": 0.5726079940795898,
      "learning_rate": 5.302644940344896e-05,
      "loss": 0.8641,
      "step": 7068
    },
    {
      "epoch": 1.2964133060769374,
      "grad_norm": 0.7435072064399719,
      "learning_rate": 5.300063293738232e-05,
      "loss": 1.2463,
      "step": 7069
    },
    {
      "epoch": 1.2965991451403085,
      "grad_norm": 0.7080186009407043,
      "learning_rate": 5.297482049127317e-05,
      "loss": 0.8211,
      "step": 7070
    },
    {
      "epoch": 1.2967849842036796,
      "grad_norm": 0.7281845808029175,
      "learning_rate": 5.2949012067329216e-05,
      "loss": 1.0874,
      "step": 7071
    },
    {
      "epoch": 1.2969708232670507,
      "grad_norm": 0.6594234704971313,
      "learning_rate": 5.2923207667758e-05,
      "loss": 0.7659,
      "step": 7072
    },
    {
      "epoch": 1.2971566623304218,
      "grad_norm": 0.6785892248153687,
      "learning_rate": 5.289740729476652e-05,
      "loss": 0.636,
      "step": 7073
    },
    {
      "epoch": 1.297342501393793,
      "grad_norm": 0.7195522785186768,
      "learning_rate": 5.287161095056165e-05,
      "loss": 0.9129,
      "step": 7074
    },
    {
      "epoch": 1.297528340457164,
      "grad_norm": 0.8044241666793823,
      "learning_rate": 5.284581863734971e-05,
      "loss": 0.9998,
      "step": 7075
    },
    {
      "epoch": 1.2977141795205351,
      "grad_norm": 0.6575559377670288,
      "learning_rate": 5.2820030357336815e-05,
      "loss": 1.0605,
      "step": 7076
    },
    {
      "epoch": 1.2979000185839062,
      "grad_norm": 0.9150829315185547,
      "learning_rate": 5.279424611272873e-05,
      "loss": 1.1433,
      "step": 7077
    },
    {
      "epoch": 1.2980858576472776,
      "grad_norm": 0.6312849521636963,
      "learning_rate": 5.276846590573077e-05,
      "loss": 0.9186,
      "step": 7078
    },
    {
      "epoch": 1.2982716967106485,
      "grad_norm": 0.6796217560768127,
      "learning_rate": 5.2742689738548026e-05,
      "loss": 0.8256,
      "step": 7079
    },
    {
      "epoch": 1.2984575357740198,
      "grad_norm": 0.6361716985702515,
      "learning_rate": 5.271691761338513e-05,
      "loss": 0.9632,
      "step": 7080
    },
    {
      "epoch": 1.298643374837391,
      "grad_norm": 0.8558089733123779,
      "learning_rate": 5.269114953244647e-05,
      "loss": 1.1539,
      "step": 7081
    },
    {
      "epoch": 1.298829213900762,
      "grad_norm": 0.6186767220497131,
      "learning_rate": 5.266538549793603e-05,
      "loss": 0.8281,
      "step": 7082
    },
    {
      "epoch": 1.299015052964133,
      "grad_norm": 0.6998415589332581,
      "learning_rate": 5.2639625512057476e-05,
      "loss": 1.1099,
      "step": 7083
    },
    {
      "epoch": 1.2992008920275042,
      "grad_norm": 0.753054141998291,
      "learning_rate": 5.2613869577014154e-05,
      "loss": 1.0645,
      "step": 7084
    },
    {
      "epoch": 1.2993867310908753,
      "grad_norm": 0.7787800431251526,
      "learning_rate": 5.258811769500893e-05,
      "loss": 0.9746,
      "step": 7085
    },
    {
      "epoch": 1.2995725701542464,
      "grad_norm": 0.7633613348007202,
      "learning_rate": 5.256236986824452e-05,
      "loss": 0.8247,
      "step": 7086
    },
    {
      "epoch": 1.2997584092176175,
      "grad_norm": 0.6192800998687744,
      "learning_rate": 5.253662609892311e-05,
      "loss": 0.9555,
      "step": 7087
    },
    {
      "epoch": 1.2999442482809886,
      "grad_norm": 0.8007282018661499,
      "learning_rate": 5.251088638924668e-05,
      "loss": 0.944,
      "step": 7088
    },
    {
      "epoch": 1.3001300873443598,
      "grad_norm": 0.6033044457435608,
      "learning_rate": 5.2485150741416734e-05,
      "loss": 0.8671,
      "step": 7089
    },
    {
      "epoch": 1.3003159264077309,
      "grad_norm": 0.865083634853363,
      "learning_rate": 5.2459419157634546e-05,
      "loss": 0.9268,
      "step": 7090
    },
    {
      "epoch": 1.300501765471102,
      "grad_norm": 0.8101431131362915,
      "learning_rate": 5.243369164010101e-05,
      "loss": 1.2567,
      "step": 7091
    },
    {
      "epoch": 1.300687604534473,
      "grad_norm": 0.6751659512519836,
      "learning_rate": 5.240796819101661e-05,
      "loss": 0.9784,
      "step": 7092
    },
    {
      "epoch": 1.3008734435978444,
      "grad_norm": 0.7335007190704346,
      "learning_rate": 5.238224881258159e-05,
      "loss": 0.8022,
      "step": 7093
    },
    {
      "epoch": 1.3010592826612153,
      "grad_norm": 0.6479902267456055,
      "learning_rate": 5.23565335069957e-05,
      "loss": 0.926,
      "step": 7094
    },
    {
      "epoch": 1.3012451217245866,
      "grad_norm": 0.8442709445953369,
      "learning_rate": 5.2330822276458466e-05,
      "loss": 0.9154,
      "step": 7095
    },
    {
      "epoch": 1.3014309607879575,
      "grad_norm": 0.5718405246734619,
      "learning_rate": 5.2305115123169036e-05,
      "loss": 0.6149,
      "step": 7096
    },
    {
      "epoch": 1.3016167998513288,
      "grad_norm": 0.6708202362060547,
      "learning_rate": 5.2279412049326183e-05,
      "loss": 1.0,
      "step": 7097
    },
    {
      "epoch": 1.3018026389147,
      "grad_norm": 0.6423093676567078,
      "learning_rate": 5.225371305712841e-05,
      "loss": 0.9844,
      "step": 7098
    },
    {
      "epoch": 1.301988477978071,
      "grad_norm": 0.7350624799728394,
      "learning_rate": 5.222801814877369e-05,
      "loss": 1.2815,
      "step": 7099
    },
    {
      "epoch": 1.3021743170414422,
      "grad_norm": 0.8177218437194824,
      "learning_rate": 5.220232732645989e-05,
      "loss": 0.9045,
      "step": 7100
    },
    {
      "epoch": 1.3023601561048133,
      "grad_norm": 0.6446688175201416,
      "learning_rate": 5.217664059238428e-05,
      "loss": 0.8688,
      "step": 7101
    },
    {
      "epoch": 1.3025459951681844,
      "grad_norm": 0.690547525882721,
      "learning_rate": 5.2150957948744005e-05,
      "loss": 1.0068,
      "step": 7102
    },
    {
      "epoch": 1.3027318342315555,
      "grad_norm": 0.6597347259521484,
      "learning_rate": 5.212527939773568e-05,
      "loss": 0.8027,
      "step": 7103
    },
    {
      "epoch": 1.3029176732949266,
      "grad_norm": 0.656406044960022,
      "learning_rate": 5.209960494155568e-05,
      "loss": 0.999,
      "step": 7104
    },
    {
      "epoch": 1.3031035123582977,
      "grad_norm": 0.7051059007644653,
      "learning_rate": 5.2073934582400016e-05,
      "loss": 1.2237,
      "step": 7105
    },
    {
      "epoch": 1.3032893514216688,
      "grad_norm": 0.8668679594993591,
      "learning_rate": 5.204826832246429e-05,
      "loss": 1.1876,
      "step": 7106
    },
    {
      "epoch": 1.30347519048504,
      "grad_norm": 0.7734251022338867,
      "learning_rate": 5.2022606163943846e-05,
      "loss": 0.8181,
      "step": 7107
    },
    {
      "epoch": 1.303661029548411,
      "grad_norm": 0.6949523091316223,
      "learning_rate": 5.199694810903355e-05,
      "loss": 1.1134,
      "step": 7108
    },
    {
      "epoch": 1.3038468686117821,
      "grad_norm": 0.6487858891487122,
      "learning_rate": 5.197129415992803e-05,
      "loss": 0.811,
      "step": 7109
    },
    {
      "epoch": 1.3040327076751534,
      "grad_norm": 0.82206791639328,
      "learning_rate": 5.194564431882156e-05,
      "loss": 1.0355,
      "step": 7110
    },
    {
      "epoch": 1.3042185467385243,
      "grad_norm": 0.7289284467697144,
      "learning_rate": 5.191999858790795e-05,
      "loss": 0.8096,
      "step": 7111
    },
    {
      "epoch": 1.3044043858018957,
      "grad_norm": 0.6622374653816223,
      "learning_rate": 5.1894356969380773e-05,
      "loss": 0.9715,
      "step": 7112
    },
    {
      "epoch": 1.3045902248652665,
      "grad_norm": 0.7061960101127625,
      "learning_rate": 5.1868719465433226e-05,
      "loss": 1.1033,
      "step": 7113
    },
    {
      "epoch": 1.3047760639286379,
      "grad_norm": 0.6868487000465393,
      "learning_rate": 5.184308607825816e-05,
      "loss": 0.9259,
      "step": 7114
    },
    {
      "epoch": 1.304961902992009,
      "grad_norm": 0.6865556240081787,
      "learning_rate": 5.181745681004798e-05,
      "loss": 1.057,
      "step": 7115
    },
    {
      "epoch": 1.30514774205538,
      "grad_norm": 0.5928111672401428,
      "learning_rate": 5.1791831662994906e-05,
      "loss": 0.9176,
      "step": 7116
    },
    {
      "epoch": 1.3053335811187512,
      "grad_norm": 0.6681034564971924,
      "learning_rate": 5.1766210639290615e-05,
      "loss": 0.993,
      "step": 7117
    },
    {
      "epoch": 1.3055194201821223,
      "grad_norm": 0.7697048783302307,
      "learning_rate": 5.174059374112657e-05,
      "loss": 1.0491,
      "step": 7118
    },
    {
      "epoch": 1.3057052592454934,
      "grad_norm": 0.682702898979187,
      "learning_rate": 5.17149809706939e-05,
      "loss": 0.9057,
      "step": 7119
    },
    {
      "epoch": 1.3058910983088645,
      "grad_norm": 0.7286297082901001,
      "learning_rate": 5.1689372330183204e-05,
      "loss": 1.0486,
      "step": 7120
    },
    {
      "epoch": 1.3060769373722356,
      "grad_norm": 0.7333587408065796,
      "learning_rate": 5.1663767821784956e-05,
      "loss": 1.1101,
      "step": 7121
    },
    {
      "epoch": 1.3062627764356067,
      "grad_norm": 0.7103135585784912,
      "learning_rate": 5.163816744768908e-05,
      "loss": 1.0523,
      "step": 7122
    },
    {
      "epoch": 1.3064486154989778,
      "grad_norm": 0.7001137137413025,
      "learning_rate": 5.1612571210085294e-05,
      "loss": 0.9445,
      "step": 7123
    },
    {
      "epoch": 1.306634454562349,
      "grad_norm": 0.7391849756240845,
      "learning_rate": 5.158697911116284e-05,
      "loss": 1.0721,
      "step": 7124
    },
    {
      "epoch": 1.30682029362572,
      "grad_norm": 0.7259404063224792,
      "learning_rate": 5.156139115311069e-05,
      "loss": 0.9623,
      "step": 7125
    },
    {
      "epoch": 1.3070061326890912,
      "grad_norm": 0.6957717537879944,
      "learning_rate": 5.153580733811746e-05,
      "loss": 0.9799,
      "step": 7126
    },
    {
      "epoch": 1.3071919717524625,
      "grad_norm": 0.7240549921989441,
      "learning_rate": 5.151022766837137e-05,
      "loss": 1.1996,
      "step": 7127
    },
    {
      "epoch": 1.3073778108158334,
      "grad_norm": 0.7606905698776245,
      "learning_rate": 5.148465214606033e-05,
      "loss": 0.8597,
      "step": 7128
    },
    {
      "epoch": 1.3075636498792047,
      "grad_norm": 0.5958693027496338,
      "learning_rate": 5.145908077337183e-05,
      "loss": 0.9355,
      "step": 7129
    },
    {
      "epoch": 1.3077494889425758,
      "grad_norm": 1.5879743099212646,
      "learning_rate": 5.1433513552493086e-05,
      "loss": 1.204,
      "step": 7130
    },
    {
      "epoch": 1.307935328005947,
      "grad_norm": 0.6255626082420349,
      "learning_rate": 5.1407950485610856e-05,
      "loss": 0.8538,
      "step": 7131
    },
    {
      "epoch": 1.308121167069318,
      "grad_norm": 0.7526214718818665,
      "learning_rate": 5.1382391574911646e-05,
      "loss": 1.0025,
      "step": 7132
    },
    {
      "epoch": 1.3083070061326891,
      "grad_norm": 0.7632701396942139,
      "learning_rate": 5.135683682258159e-05,
      "loss": 0.6711,
      "step": 7133
    },
    {
      "epoch": 1.3084928451960602,
      "grad_norm": 0.7910258769989014,
      "learning_rate": 5.1331286230806384e-05,
      "loss": 0.9427,
      "step": 7134
    },
    {
      "epoch": 1.3086786842594313,
      "grad_norm": 0.753558874130249,
      "learning_rate": 5.130573980177149e-05,
      "loss": 0.7711,
      "step": 7135
    },
    {
      "epoch": 1.3088645233228025,
      "grad_norm": 0.8264387845993042,
      "learning_rate": 5.128019753766188e-05,
      "loss": 1.1235,
      "step": 7136
    },
    {
      "epoch": 1.3090503623861736,
      "grad_norm": 0.7292147278785706,
      "learning_rate": 5.12546594406623e-05,
      "loss": 1.0792,
      "step": 7137
    },
    {
      "epoch": 1.3092362014495447,
      "grad_norm": 0.7105168104171753,
      "learning_rate": 5.122912551295702e-05,
      "loss": 1.074,
      "step": 7138
    },
    {
      "epoch": 1.3094220405129158,
      "grad_norm": 0.7290421724319458,
      "learning_rate": 5.120359575673004e-05,
      "loss": 1.1185,
      "step": 7139
    },
    {
      "epoch": 1.3096078795762869,
      "grad_norm": 0.588783323764801,
      "learning_rate": 5.1178070174164985e-05,
      "loss": 0.8375,
      "step": 7140
    },
    {
      "epoch": 1.309793718639658,
      "grad_norm": 0.692352294921875,
      "learning_rate": 5.115254876744509e-05,
      "loss": 1.0739,
      "step": 7141
    },
    {
      "epoch": 1.309979557703029,
      "grad_norm": 0.8384915590286255,
      "learning_rate": 5.1127031538753326e-05,
      "loss": 0.8474,
      "step": 7142
    },
    {
      "epoch": 1.3101653967664002,
      "grad_norm": 0.532124936580658,
      "learning_rate": 5.110151849027214e-05,
      "loss": 0.6912,
      "step": 7143
    },
    {
      "epoch": 1.3103512358297715,
      "grad_norm": 0.7722443342208862,
      "learning_rate": 5.10760096241838e-05,
      "loss": 0.9938,
      "step": 7144
    },
    {
      "epoch": 1.3105370748931424,
      "grad_norm": 0.7080844044685364,
      "learning_rate": 5.1050504942670054e-05,
      "loss": 1.0352,
      "step": 7145
    },
    {
      "epoch": 1.3107229139565137,
      "grad_norm": 0.6958823204040527,
      "learning_rate": 5.10250044479124e-05,
      "loss": 1.1444,
      "step": 7146
    },
    {
      "epoch": 1.3109087530198849,
      "grad_norm": 0.6618379950523376,
      "learning_rate": 5.099950814209202e-05,
      "loss": 0.7356,
      "step": 7147
    },
    {
      "epoch": 1.311094592083256,
      "grad_norm": 0.7903035879135132,
      "learning_rate": 5.0974016027389583e-05,
      "loss": 1.1493,
      "step": 7148
    },
    {
      "epoch": 1.311280431146627,
      "grad_norm": 0.6709766983985901,
      "learning_rate": 5.094852810598554e-05,
      "loss": 1.0465,
      "step": 7149
    },
    {
      "epoch": 1.3114662702099982,
      "grad_norm": 1.7481948137283325,
      "learning_rate": 5.092304438005986e-05,
      "loss": 1.3618,
      "step": 7150
    },
    {
      "epoch": 1.3116521092733693,
      "grad_norm": 0.707204282283783,
      "learning_rate": 5.089756485179231e-05,
      "loss": 1.0906,
      "step": 7151
    },
    {
      "epoch": 1.3118379483367404,
      "grad_norm": 0.6673077940940857,
      "learning_rate": 5.087208952336211e-05,
      "loss": 1.0719,
      "step": 7152
    },
    {
      "epoch": 1.3120237874001115,
      "grad_norm": 0.812818706035614,
      "learning_rate": 5.084661839694829e-05,
      "loss": 1.0051,
      "step": 7153
    },
    {
      "epoch": 1.3122096264634826,
      "grad_norm": 0.7848548889160156,
      "learning_rate": 5.082115147472942e-05,
      "loss": 0.9807,
      "step": 7154
    },
    {
      "epoch": 1.3123954655268537,
      "grad_norm": 0.9212191104888916,
      "learning_rate": 5.079568875888381e-05,
      "loss": 0.8138,
      "step": 7155
    },
    {
      "epoch": 1.3125813045902248,
      "grad_norm": 0.6669211983680725,
      "learning_rate": 5.077023025158922e-05,
      "loss": 0.9154,
      "step": 7156
    },
    {
      "epoch": 1.312767143653596,
      "grad_norm": 0.9354314208030701,
      "learning_rate": 5.074477595502326e-05,
      "loss": 1.0927,
      "step": 7157
    },
    {
      "epoch": 1.312952982716967,
      "grad_norm": 0.6543704271316528,
      "learning_rate": 5.07193258713631e-05,
      "loss": 0.962,
      "step": 7158
    },
    {
      "epoch": 1.3131388217803384,
      "grad_norm": 0.9563025832176208,
      "learning_rate": 5.0693880002785456e-05,
      "loss": 0.8167,
      "step": 7159
    },
    {
      "epoch": 1.3133246608437092,
      "grad_norm": 0.6064286231994629,
      "learning_rate": 5.066843835146684e-05,
      "loss": 0.9669,
      "step": 7160
    },
    {
      "epoch": 1.3135104999070806,
      "grad_norm": 0.7774001955986023,
      "learning_rate": 5.064300091958334e-05,
      "loss": 1.1365,
      "step": 7161
    },
    {
      "epoch": 1.3136963389704515,
      "grad_norm": 0.6404112577438354,
      "learning_rate": 5.061756770931061e-05,
      "loss": 0.8782,
      "step": 7162
    },
    {
      "epoch": 1.3138821780338228,
      "grad_norm": 0.6021904945373535,
      "learning_rate": 5.059213872282407e-05,
      "loss": 0.8356,
      "step": 7163
    },
    {
      "epoch": 1.314068017097194,
      "grad_norm": 0.6289204955101013,
      "learning_rate": 5.056671396229866e-05,
      "loss": 0.8957,
      "step": 7164
    },
    {
      "epoch": 1.314253856160565,
      "grad_norm": 0.7028182744979858,
      "learning_rate": 5.054129342990909e-05,
      "loss": 0.9555,
      "step": 7165
    },
    {
      "epoch": 1.314439695223936,
      "grad_norm": 0.5707007050514221,
      "learning_rate": 5.0515877127829536e-05,
      "loss": 0.9037,
      "step": 7166
    },
    {
      "epoch": 1.3146255342873072,
      "grad_norm": 1.013501763343811,
      "learning_rate": 5.049046505823396e-05,
      "loss": 1.1444,
      "step": 7167
    },
    {
      "epoch": 1.3148113733506783,
      "grad_norm": 0.5975896716117859,
      "learning_rate": 5.0465057223295965e-05,
      "loss": 0.9372,
      "step": 7168
    },
    {
      "epoch": 1.3149972124140494,
      "grad_norm": 0.7597402334213257,
      "learning_rate": 5.043965362518863e-05,
      "loss": 1.2744,
      "step": 7169
    },
    {
      "epoch": 1.3151830514774205,
      "grad_norm": 0.6761762499809265,
      "learning_rate": 5.041425426608485e-05,
      "loss": 1.2111,
      "step": 7170
    },
    {
      "epoch": 1.3153688905407916,
      "grad_norm": 0.6263930201530457,
      "learning_rate": 5.038885914815705e-05,
      "loss": 1.016,
      "step": 7171
    },
    {
      "epoch": 1.3155547296041628,
      "grad_norm": 0.6915165781974792,
      "learning_rate": 5.036346827357741e-05,
      "loss": 0.869,
      "step": 7172
    },
    {
      "epoch": 1.3157405686675339,
      "grad_norm": 0.6968695521354675,
      "learning_rate": 5.033808164451757e-05,
      "loss": 0.9908,
      "step": 7173
    },
    {
      "epoch": 1.315926407730905,
      "grad_norm": 0.7895475625991821,
      "learning_rate": 5.031269926314892e-05,
      "loss": 0.9928,
      "step": 7174
    },
    {
      "epoch": 1.316112246794276,
      "grad_norm": 0.6815813779830933,
      "learning_rate": 5.0287321131642553e-05,
      "loss": 0.7665,
      "step": 7175
    },
    {
      "epoch": 1.3162980858576474,
      "grad_norm": 0.8258618712425232,
      "learning_rate": 5.0261947252168995e-05,
      "loss": 0.9358,
      "step": 7176
    },
    {
      "epoch": 1.3164839249210183,
      "grad_norm": 0.7376391291618347,
      "learning_rate": 5.023657762689864e-05,
      "loss": 0.7476,
      "step": 7177
    },
    {
      "epoch": 1.3166697639843896,
      "grad_norm": 0.6556116938591003,
      "learning_rate": 5.021121225800129e-05,
      "loss": 0.6715,
      "step": 7178
    },
    {
      "epoch": 1.3168556030477605,
      "grad_norm": 0.8220980167388916,
      "learning_rate": 5.018585114764662e-05,
      "loss": 1.163,
      "step": 7179
    },
    {
      "epoch": 1.3170414421111318,
      "grad_norm": 0.8304170966148376,
      "learning_rate": 5.0160494298003733e-05,
      "loss": 0.9653,
      "step": 7180
    },
    {
      "epoch": 1.317227281174503,
      "grad_norm": 0.672730565071106,
      "learning_rate": 5.0135141711241476e-05,
      "loss": 0.872,
      "step": 7181
    },
    {
      "epoch": 1.317413120237874,
      "grad_norm": 0.6483734846115112,
      "learning_rate": 5.010979338952837e-05,
      "loss": 0.9913,
      "step": 7182
    },
    {
      "epoch": 1.3175989593012452,
      "grad_norm": 0.7637404203414917,
      "learning_rate": 5.00844493350324e-05,
      "loss": 1.0753,
      "step": 7183
    },
    {
      "epoch": 1.3177847983646163,
      "grad_norm": 0.8622449040412903,
      "learning_rate": 5.005910954992139e-05,
      "loss": 1.1496,
      "step": 7184
    },
    {
      "epoch": 1.3179706374279874,
      "grad_norm": 0.6058513522148132,
      "learning_rate": 5.0033774036362654e-05,
      "loss": 0.9682,
      "step": 7185
    },
    {
      "epoch": 1.3181564764913585,
      "grad_norm": 0.6887873411178589,
      "learning_rate": 5.0008442796523257e-05,
      "loss": 1.0055,
      "step": 7186
    },
    {
      "epoch": 1.3183423155547296,
      "grad_norm": 0.6620786786079407,
      "learning_rate": 4.998311583256976e-05,
      "loss": 0.8581,
      "step": 7187
    },
    {
      "epoch": 1.3185281546181007,
      "grad_norm": 0.6699737906455994,
      "learning_rate": 4.995779314666848e-05,
      "loss": 0.9666,
      "step": 7188
    },
    {
      "epoch": 1.3187139936814718,
      "grad_norm": 0.6416510343551636,
      "learning_rate": 4.993247474098532e-05,
      "loss": 0.9825,
      "step": 7189
    },
    {
      "epoch": 1.318899832744843,
      "grad_norm": 0.6661123633384705,
      "learning_rate": 4.9907160617685786e-05,
      "loss": 0.6764,
      "step": 7190
    },
    {
      "epoch": 1.319085671808214,
      "grad_norm": 0.7946585416793823,
      "learning_rate": 4.98818507789351e-05,
      "loss": 0.9549,
      "step": 7191
    },
    {
      "epoch": 1.3192715108715851,
      "grad_norm": 0.7565823793411255,
      "learning_rate": 4.9856545226898e-05,
      "loss": 1.103,
      "step": 7192
    },
    {
      "epoch": 1.3194573499349564,
      "grad_norm": 0.7292872071266174,
      "learning_rate": 4.983124396373899e-05,
      "loss": 1.0208,
      "step": 7193
    },
    {
      "epoch": 1.3196431889983273,
      "grad_norm": 0.6677314639091492,
      "learning_rate": 4.980594699162209e-05,
      "loss": 1.0857,
      "step": 7194
    },
    {
      "epoch": 1.3198290280616987,
      "grad_norm": 0.67109614610672,
      "learning_rate": 4.9780654312711015e-05,
      "loss": 1.1204,
      "step": 7195
    },
    {
      "epoch": 1.3200148671250698,
      "grad_norm": 0.8349084854125977,
      "learning_rate": 4.975536592916916e-05,
      "loss": 1.1697,
      "step": 7196
    },
    {
      "epoch": 1.3202007061884409,
      "grad_norm": 0.7957627773284912,
      "learning_rate": 4.973008184315941e-05,
      "loss": 0.9635,
      "step": 7197
    },
    {
      "epoch": 1.320386545251812,
      "grad_norm": 0.6814358234405518,
      "learning_rate": 4.970480205684441e-05,
      "loss": 1.1606,
      "step": 7198
    },
    {
      "epoch": 1.320572384315183,
      "grad_norm": 0.6425552368164062,
      "learning_rate": 4.967952657238639e-05,
      "loss": 0.8352,
      "step": 7199
    },
    {
      "epoch": 1.3207582233785542,
      "grad_norm": 0.8170636296272278,
      "learning_rate": 4.965425539194726e-05,
      "loss": 0.9484,
      "step": 7200
    },
    {
      "epoch": 1.3209440624419253,
      "grad_norm": 0.659493088722229,
      "learning_rate": 4.9628988517688434e-05,
      "loss": 0.7609,
      "step": 7201
    },
    {
      "epoch": 1.3211299015052964,
      "grad_norm": 0.705998957157135,
      "learning_rate": 4.9603725951771094e-05,
      "loss": 1.082,
      "step": 7202
    },
    {
      "epoch": 1.3213157405686675,
      "grad_norm": 0.7331377863883972,
      "learning_rate": 4.957846769635602e-05,
      "loss": 0.9501,
      "step": 7203
    },
    {
      "epoch": 1.3215015796320386,
      "grad_norm": 0.6525837779045105,
      "learning_rate": 4.9553213753603554e-05,
      "loss": 1.0953,
      "step": 7204
    },
    {
      "epoch": 1.3216874186954097,
      "grad_norm": 0.7075824737548828,
      "learning_rate": 4.9527964125673776e-05,
      "loss": 1.0025,
      "step": 7205
    },
    {
      "epoch": 1.3218732577587808,
      "grad_norm": 0.5878762006759644,
      "learning_rate": 4.9502718814726276e-05,
      "loss": 0.8849,
      "step": 7206
    },
    {
      "epoch": 1.322059096822152,
      "grad_norm": 0.6121537089347839,
      "learning_rate": 4.947747782292041e-05,
      "loss": 0.9235,
      "step": 7207
    },
    {
      "epoch": 1.322244935885523,
      "grad_norm": 0.6786346435546875,
      "learning_rate": 4.945224115241503e-05,
      "loss": 0.8459,
      "step": 7208
    },
    {
      "epoch": 1.3224307749488942,
      "grad_norm": 0.8413114547729492,
      "learning_rate": 4.94270088053687e-05,
      "loss": 0.9905,
      "step": 7209
    },
    {
      "epoch": 1.3226166140122655,
      "grad_norm": 0.6944953799247742,
      "learning_rate": 4.9401780783939656e-05,
      "loss": 1.0481,
      "step": 7210
    },
    {
      "epoch": 1.3228024530756364,
      "grad_norm": 0.6693060398101807,
      "learning_rate": 4.9376557090285614e-05,
      "loss": 0.9468,
      "step": 7211
    },
    {
      "epoch": 1.3229882921390077,
      "grad_norm": 0.6698997616767883,
      "learning_rate": 4.93513377265641e-05,
      "loss": 1.0794,
      "step": 7212
    },
    {
      "epoch": 1.3231741312023788,
      "grad_norm": 0.8010924458503723,
      "learning_rate": 4.932612269493208e-05,
      "loss": 0.995,
      "step": 7213
    },
    {
      "epoch": 1.32335997026575,
      "grad_norm": 0.699874997138977,
      "learning_rate": 4.930091199754633e-05,
      "loss": 0.9846,
      "step": 7214
    },
    {
      "epoch": 1.323545809329121,
      "grad_norm": 0.6646803617477417,
      "learning_rate": 4.927570563656315e-05,
      "loss": 0.741,
      "step": 7215
    },
    {
      "epoch": 1.3237316483924921,
      "grad_norm": 0.7552995085716248,
      "learning_rate": 4.925050361413849e-05,
      "loss": 1.048,
      "step": 7216
    },
    {
      "epoch": 1.3239174874558632,
      "grad_norm": 0.6564279198646545,
      "learning_rate": 4.922530593242798e-05,
      "loss": 0.9939,
      "step": 7217
    },
    {
      "epoch": 1.3241033265192343,
      "grad_norm": 0.641197144985199,
      "learning_rate": 4.920011259358675e-05,
      "loss": 0.7629,
      "step": 7218
    },
    {
      "epoch": 1.3242891655826055,
      "grad_norm": 0.6285445094108582,
      "learning_rate": 4.917492359976973e-05,
      "loss": 0.9564,
      "step": 7219
    },
    {
      "epoch": 1.3244750046459766,
      "grad_norm": 0.6213375329971313,
      "learning_rate": 4.914973895313128e-05,
      "loss": 0.8366,
      "step": 7220
    },
    {
      "epoch": 1.3246608437093477,
      "grad_norm": 0.7667694091796875,
      "learning_rate": 4.9124558655825626e-05,
      "loss": 0.9716,
      "step": 7221
    },
    {
      "epoch": 1.3248466827727188,
      "grad_norm": 0.8510693311691284,
      "learning_rate": 4.909938271000638e-05,
      "loss": 0.9444,
      "step": 7222
    },
    {
      "epoch": 1.3250325218360899,
      "grad_norm": 0.6002587080001831,
      "learning_rate": 4.907421111782694e-05,
      "loss": 0.8624,
      "step": 7223
    },
    {
      "epoch": 1.325218360899461,
      "grad_norm": 0.8189764022827148,
      "learning_rate": 4.9049043881440315e-05,
      "loss": 0.8753,
      "step": 7224
    },
    {
      "epoch": 1.3254041999628323,
      "grad_norm": 0.673364520072937,
      "learning_rate": 4.9023881002999064e-05,
      "loss": 0.9,
      "step": 7225
    },
    {
      "epoch": 1.3255900390262032,
      "grad_norm": 0.9202880859375,
      "learning_rate": 4.899872248465548e-05,
      "loss": 1.0446,
      "step": 7226
    },
    {
      "epoch": 1.3257758780895745,
      "grad_norm": 0.6275797486305237,
      "learning_rate": 4.897356832856135e-05,
      "loss": 0.8593,
      "step": 7227
    },
    {
      "epoch": 1.3259617171529454,
      "grad_norm": 0.7230714559555054,
      "learning_rate": 4.89484185368682e-05,
      "loss": 0.8682,
      "step": 7228
    },
    {
      "epoch": 1.3261475562163167,
      "grad_norm": 0.6627464294433594,
      "learning_rate": 4.8923273111727155e-05,
      "loss": 0.8705,
      "step": 7229
    },
    {
      "epoch": 1.3263333952796879,
      "grad_norm": 0.7997526526451111,
      "learning_rate": 4.889813205528895e-05,
      "loss": 0.8549,
      "step": 7230
    },
    {
      "epoch": 1.326519234343059,
      "grad_norm": 0.7096716165542603,
      "learning_rate": 4.887299536970399e-05,
      "loss": 0.9922,
      "step": 7231
    },
    {
      "epoch": 1.32670507340643,
      "grad_norm": 0.7961505651473999,
      "learning_rate": 4.8847863057122204e-05,
      "loss": 1.1905,
      "step": 7232
    },
    {
      "epoch": 1.3268909124698012,
      "grad_norm": 0.7194200754165649,
      "learning_rate": 4.8822735119693275e-05,
      "loss": 1.1028,
      "step": 7233
    },
    {
      "epoch": 1.3270767515331723,
      "grad_norm": 0.715369462966919,
      "learning_rate": 4.879761155956639e-05,
      "loss": 0.7564,
      "step": 7234
    },
    {
      "epoch": 1.3272625905965434,
      "grad_norm": 0.7175028324127197,
      "learning_rate": 4.877249237889049e-05,
      "loss": 1.0886,
      "step": 7235
    },
    {
      "epoch": 1.3274484296599145,
      "grad_norm": 0.6594383120536804,
      "learning_rate": 4.874737757981399e-05,
      "loss": 0.9096,
      "step": 7236
    },
    {
      "epoch": 1.3276342687232856,
      "grad_norm": 0.67848140001297,
      "learning_rate": 4.872226716448506e-05,
      "loss": 1.2045,
      "step": 7237
    },
    {
      "epoch": 1.3278201077866567,
      "grad_norm": 0.6310969591140747,
      "learning_rate": 4.869716113505148e-05,
      "loss": 0.7236,
      "step": 7238
    },
    {
      "epoch": 1.3280059468500278,
      "grad_norm": 0.5502176284790039,
      "learning_rate": 4.867205949366055e-05,
      "loss": 0.6613,
      "step": 7239
    },
    {
      "epoch": 1.328191785913399,
      "grad_norm": 0.5823808908462524,
      "learning_rate": 4.8646962242459345e-05,
      "loss": 0.9206,
      "step": 7240
    },
    {
      "epoch": 1.32837762497677,
      "grad_norm": 0.7270284295082092,
      "learning_rate": 4.8621869383594406e-05,
      "loss": 1.1637,
      "step": 7241
    },
    {
      "epoch": 1.3285634640401414,
      "grad_norm": 0.7297576665878296,
      "learning_rate": 4.8596780919212034e-05,
      "loss": 0.895,
      "step": 7242
    },
    {
      "epoch": 1.3287493031035122,
      "grad_norm": 0.6301212310791016,
      "learning_rate": 4.857169685145808e-05,
      "loss": 0.9788,
      "step": 7243
    },
    {
      "epoch": 1.3289351421668836,
      "grad_norm": 0.6360543966293335,
      "learning_rate": 4.854661718247805e-05,
      "loss": 1.2043,
      "step": 7244
    },
    {
      "epoch": 1.3291209812302545,
      "grad_norm": 0.8445108532905579,
      "learning_rate": 4.852154191441709e-05,
      "loss": 1.1311,
      "step": 7245
    },
    {
      "epoch": 1.3293068202936258,
      "grad_norm": 0.7138316035270691,
      "learning_rate": 4.8496471049419866e-05,
      "loss": 1.1371,
      "step": 7246
    },
    {
      "epoch": 1.329492659356997,
      "grad_norm": 0.6623853445053101,
      "learning_rate": 4.847140458963083e-05,
      "loss": 0.9835,
      "step": 7247
    },
    {
      "epoch": 1.329678498420368,
      "grad_norm": 0.6935982704162598,
      "learning_rate": 4.84463425371939e-05,
      "loss": 0.9587,
      "step": 7248
    },
    {
      "epoch": 1.3298643374837391,
      "grad_norm": 0.8760302066802979,
      "learning_rate": 4.8421284894252737e-05,
      "loss": 1.1497,
      "step": 7249
    },
    {
      "epoch": 1.3300501765471102,
      "grad_norm": 0.6761434078216553,
      "learning_rate": 4.839623166295053e-05,
      "loss": 0.7356,
      "step": 7250
    },
    {
      "epoch": 1.3302360156104813,
      "grad_norm": 0.6983981728553772,
      "learning_rate": 4.837118284543015e-05,
      "loss": 0.933,
      "step": 7251
    },
    {
      "epoch": 1.3304218546738524,
      "grad_norm": 0.7780141830444336,
      "learning_rate": 4.834613844383412e-05,
      "loss": 1.2209,
      "step": 7252
    },
    {
      "epoch": 1.3306076937372235,
      "grad_norm": 0.7106468677520752,
      "learning_rate": 4.832109846030448e-05,
      "loss": 1.0441,
      "step": 7253
    },
    {
      "epoch": 1.3307935328005946,
      "grad_norm": 0.8280313611030579,
      "learning_rate": 4.8296062896983005e-05,
      "loss": 1.0316,
      "step": 7254
    },
    {
      "epoch": 1.3309793718639658,
      "grad_norm": 0.7968834042549133,
      "learning_rate": 4.8271031756011e-05,
      "loss": 0.8459,
      "step": 7255
    },
    {
      "epoch": 1.3311652109273369,
      "grad_norm": 0.716829776763916,
      "learning_rate": 4.824600503952943e-05,
      "loss": 0.8083,
      "step": 7256
    },
    {
      "epoch": 1.331351049990708,
      "grad_norm": 0.6207119822502136,
      "learning_rate": 4.822098274967895e-05,
      "loss": 0.9451,
      "step": 7257
    },
    {
      "epoch": 1.331536889054079,
      "grad_norm": 0.8303805589675903,
      "learning_rate": 4.8195964888599685e-05,
      "loss": 1.814,
      "step": 7258
    },
    {
      "epoch": 1.3317227281174504,
      "grad_norm": 0.6936972141265869,
      "learning_rate": 4.817095145843151e-05,
      "loss": 0.8158,
      "step": 7259
    },
    {
      "epoch": 1.3319085671808213,
      "grad_norm": 0.840417206287384,
      "learning_rate": 4.814594246131387e-05,
      "loss": 1.052,
      "step": 7260
    },
    {
      "epoch": 1.3320944062441926,
      "grad_norm": 0.6627435684204102,
      "learning_rate": 4.812093789938589e-05,
      "loss": 0.8051,
      "step": 7261
    },
    {
      "epoch": 1.3322802453075637,
      "grad_norm": 0.8406475186347961,
      "learning_rate": 4.809593777478617e-05,
      "loss": 1.1079,
      "step": 7262
    },
    {
      "epoch": 1.3324660843709348,
      "grad_norm": 0.6551578044891357,
      "learning_rate": 4.807094208965308e-05,
      "loss": 0.6396,
      "step": 7263
    },
    {
      "epoch": 1.332651923434306,
      "grad_norm": 0.7471421957015991,
      "learning_rate": 4.804595084612458e-05,
      "loss": 0.5304,
      "step": 7264
    },
    {
      "epoch": 1.332837762497677,
      "grad_norm": 0.8838663697242737,
      "learning_rate": 4.8020964046338145e-05,
      "loss": 1.1116,
      "step": 7265
    },
    {
      "epoch": 1.3330236015610482,
      "grad_norm": 0.8529516458511353,
      "learning_rate": 4.7995981692431044e-05,
      "loss": 1.0741,
      "step": 7266
    },
    {
      "epoch": 1.3332094406244193,
      "grad_norm": 0.7180158495903015,
      "learning_rate": 4.797100378653998e-05,
      "loss": 0.8825,
      "step": 7267
    },
    {
      "epoch": 1.3333952796877904,
      "grad_norm": 0.8214242458343506,
      "learning_rate": 4.794603033080146e-05,
      "loss": 0.9276,
      "step": 7268
    },
    {
      "epoch": 1.3335811187511615,
      "grad_norm": 0.7666495442390442,
      "learning_rate": 4.7921061327351425e-05,
      "loss": 0.8363,
      "step": 7269
    },
    {
      "epoch": 1.3337669578145326,
      "grad_norm": 0.7022069096565247,
      "learning_rate": 4.789609677832557e-05,
      "loss": 0.8717,
      "step": 7270
    },
    {
      "epoch": 1.3339527968779037,
      "grad_norm": 0.7741091847419739,
      "learning_rate": 4.787113668585921e-05,
      "loss": 0.9096,
      "step": 7271
    },
    {
      "epoch": 1.3341386359412748,
      "grad_norm": 0.9301775097846985,
      "learning_rate": 4.784618105208716e-05,
      "loss": 1.0445,
      "step": 7272
    },
    {
      "epoch": 1.334324475004646,
      "grad_norm": 0.6503124833106995,
      "learning_rate": 4.782122987914395e-05,
      "loss": 0.9453,
      "step": 7273
    },
    {
      "epoch": 1.3345103140680172,
      "grad_norm": 0.6455255746841431,
      "learning_rate": 4.779628316916374e-05,
      "loss": 0.7312,
      "step": 7274
    },
    {
      "epoch": 1.3346961531313881,
      "grad_norm": 0.7131131887435913,
      "learning_rate": 4.777134092428028e-05,
      "loss": 1.0174,
      "step": 7275
    },
    {
      "epoch": 1.3348819921947594,
      "grad_norm": 0.703403651714325,
      "learning_rate": 4.774640314662688e-05,
      "loss": 0.9192,
      "step": 7276
    },
    {
      "epoch": 1.3350678312581303,
      "grad_norm": 0.7222033143043518,
      "learning_rate": 4.772146983833655e-05,
      "loss": 0.8505,
      "step": 7277
    },
    {
      "epoch": 1.3352536703215017,
      "grad_norm": 0.6918283104896545,
      "learning_rate": 4.769654100154194e-05,
      "loss": 0.9286,
      "step": 7278
    },
    {
      "epoch": 1.3354395093848728,
      "grad_norm": 0.8968819975852966,
      "learning_rate": 4.767161663837517e-05,
      "loss": 0.9518,
      "step": 7279
    },
    {
      "epoch": 1.3356253484482439,
      "grad_norm": 0.6316736936569214,
      "learning_rate": 4.764669675096818e-05,
      "loss": 0.865,
      "step": 7280
    },
    {
      "epoch": 1.335811187511615,
      "grad_norm": 0.701098620891571,
      "learning_rate": 4.7621781341452324e-05,
      "loss": 1.0819,
      "step": 7281
    },
    {
      "epoch": 1.335997026574986,
      "grad_norm": 0.8178427815437317,
      "learning_rate": 4.759687041195874e-05,
      "loss": 0.8916,
      "step": 7282
    },
    {
      "epoch": 1.3361828656383572,
      "grad_norm": 0.6333885788917542,
      "learning_rate": 4.757196396461806e-05,
      "loss": 0.9393,
      "step": 7283
    },
    {
      "epoch": 1.3363687047017283,
      "grad_norm": 0.7877507209777832,
      "learning_rate": 4.7547062001560614e-05,
      "loss": 1.1123,
      "step": 7284
    },
    {
      "epoch": 1.3365545437650994,
      "grad_norm": 0.5933407545089722,
      "learning_rate": 4.752216452491636e-05,
      "loss": 0.8123,
      "step": 7285
    },
    {
      "epoch": 1.3367403828284705,
      "grad_norm": 0.7034826278686523,
      "learning_rate": 4.7497271536814744e-05,
      "loss": 0.9516,
      "step": 7286
    },
    {
      "epoch": 1.3369262218918416,
      "grad_norm": 0.7386837005615234,
      "learning_rate": 4.747238303938498e-05,
      "loss": 0.6906,
      "step": 7287
    },
    {
      "epoch": 1.3371120609552127,
      "grad_norm": 0.64260333776474,
      "learning_rate": 4.7447499034755815e-05,
      "loss": 0.8486,
      "step": 7288
    },
    {
      "epoch": 1.3372979000185838,
      "grad_norm": 0.9089800119400024,
      "learning_rate": 4.742261952505568e-05,
      "loss": 1.0577,
      "step": 7289
    },
    {
      "epoch": 1.337483739081955,
      "grad_norm": 0.5729821920394897,
      "learning_rate": 4.73977445124125e-05,
      "loss": 0.7278,
      "step": 7290
    },
    {
      "epoch": 1.3376695781453263,
      "grad_norm": 0.6363554000854492,
      "learning_rate": 4.7372873998953905e-05,
      "loss": 1.052,
      "step": 7291
    },
    {
      "epoch": 1.3378554172086972,
      "grad_norm": 0.5810597538948059,
      "learning_rate": 4.734800798680719e-05,
      "loss": 0.7631,
      "step": 7292
    },
    {
      "epoch": 1.3380412562720685,
      "grad_norm": 0.779283881187439,
      "learning_rate": 4.73231464780991e-05,
      "loss": 0.8645,
      "step": 7293
    },
    {
      "epoch": 1.3382270953354394,
      "grad_norm": 0.7239099740982056,
      "learning_rate": 4.729828947495619e-05,
      "loss": 1.0195,
      "step": 7294
    },
    {
      "epoch": 1.3384129343988107,
      "grad_norm": 0.7280988693237305,
      "learning_rate": 4.727343697950444e-05,
      "loss": 0.9562,
      "step": 7295
    },
    {
      "epoch": 1.3385987734621818,
      "grad_norm": 0.7767034769058228,
      "learning_rate": 4.7248588993869626e-05,
      "loss": 0.895,
      "step": 7296
    },
    {
      "epoch": 1.338784612525553,
      "grad_norm": 0.7110447883605957,
      "learning_rate": 4.7223745520176956e-05,
      "loss": 1.0914,
      "step": 7297
    },
    {
      "epoch": 1.338970451588924,
      "grad_norm": 0.7331178188323975,
      "learning_rate": 4.7198906560551405e-05,
      "loss": 1.2067,
      "step": 7298
    },
    {
      "epoch": 1.3391562906522951,
      "grad_norm": 0.7957321405410767,
      "learning_rate": 4.717407211711753e-05,
      "loss": 1.0934,
      "step": 7299
    },
    {
      "epoch": 1.3393421297156662,
      "grad_norm": 0.5885205864906311,
      "learning_rate": 4.71492421919994e-05,
      "loss": 0.913,
      "step": 7300
    },
    {
      "epoch": 1.3395279687790373,
      "grad_norm": 0.7285014986991882,
      "learning_rate": 4.7124416787320814e-05,
      "loss": 1.1694,
      "step": 7301
    },
    {
      "epoch": 1.3397138078424085,
      "grad_norm": 0.7373610734939575,
      "learning_rate": 4.7099595905205176e-05,
      "loss": 0.9373,
      "step": 7302
    },
    {
      "epoch": 1.3398996469057796,
      "grad_norm": 0.7105932831764221,
      "learning_rate": 4.7074779547775395e-05,
      "loss": 0.8162,
      "step": 7303
    },
    {
      "epoch": 1.3400854859691507,
      "grad_norm": 0.6319199204444885,
      "learning_rate": 4.70499677171541e-05,
      "loss": 0.9777,
      "step": 7304
    },
    {
      "epoch": 1.3402713250325218,
      "grad_norm": 0.5829048752784729,
      "learning_rate": 4.702516041546351e-05,
      "loss": 0.9162,
      "step": 7305
    },
    {
      "epoch": 1.3404571640958929,
      "grad_norm": 0.7003679871559143,
      "learning_rate": 4.700035764482549e-05,
      "loss": 0.9545,
      "step": 7306
    },
    {
      "epoch": 1.340643003159264,
      "grad_norm": 0.8310719728469849,
      "learning_rate": 4.697555940736138e-05,
      "loss": 1.0496,
      "step": 7307
    },
    {
      "epoch": 1.3408288422226353,
      "grad_norm": 0.6480697393417358,
      "learning_rate": 4.6950765705192315e-05,
      "loss": 0.9615,
      "step": 7308
    },
    {
      "epoch": 1.3410146812860062,
      "grad_norm": 1.5479086637496948,
      "learning_rate": 4.6925976540438874e-05,
      "loss": 1.5403,
      "step": 7309
    },
    {
      "epoch": 1.3412005203493775,
      "grad_norm": 0.6437709331512451,
      "learning_rate": 4.690119191522141e-05,
      "loss": 0.9907,
      "step": 7310
    },
    {
      "epoch": 1.3413863594127486,
      "grad_norm": 0.6723558306694031,
      "learning_rate": 4.687641183165971e-05,
      "loss": 0.8375,
      "step": 7311
    },
    {
      "epoch": 1.3415721984761197,
      "grad_norm": 0.7868484854698181,
      "learning_rate": 4.6851636291873346e-05,
      "loss": 0.9949,
      "step": 7312
    },
    {
      "epoch": 1.3417580375394909,
      "grad_norm": 0.5984890460968018,
      "learning_rate": 4.682686529798143e-05,
      "loss": 0.9826,
      "step": 7313
    },
    {
      "epoch": 1.341943876602862,
      "grad_norm": 0.7483819127082825,
      "learning_rate": 4.6802098852102593e-05,
      "loss": 1.0567,
      "step": 7314
    },
    {
      "epoch": 1.342129715666233,
      "grad_norm": 0.7057022452354431,
      "learning_rate": 4.6777336956355263e-05,
      "loss": 0.9281,
      "step": 7315
    },
    {
      "epoch": 1.3423155547296042,
      "grad_norm": 0.7427000403404236,
      "learning_rate": 4.6752579612857305e-05,
      "loss": 0.8853,
      "step": 7316
    },
    {
      "epoch": 1.3425013937929753,
      "grad_norm": 0.5980744361877441,
      "learning_rate": 4.672782682372628e-05,
      "loss": 0.6722,
      "step": 7317
    },
    {
      "epoch": 1.3426872328563464,
      "grad_norm": 0.833619236946106,
      "learning_rate": 4.6703078591079374e-05,
      "loss": 1.0043,
      "step": 7318
    },
    {
      "epoch": 1.3428730719197175,
      "grad_norm": 0.6096035838127136,
      "learning_rate": 4.667833491703335e-05,
      "loss": 0.9514,
      "step": 7319
    },
    {
      "epoch": 1.3430589109830886,
      "grad_norm": 0.5686629414558411,
      "learning_rate": 4.6653595803704606e-05,
      "loss": 0.7055,
      "step": 7320
    },
    {
      "epoch": 1.3432447500464597,
      "grad_norm": 0.8507375121116638,
      "learning_rate": 4.662886125320909e-05,
      "loss": 1.0759,
      "step": 7321
    },
    {
      "epoch": 1.3434305891098308,
      "grad_norm": 0.6726318597793579,
      "learning_rate": 4.660413126766244e-05,
      "loss": 0.66,
      "step": 7322
    },
    {
      "epoch": 1.343616428173202,
      "grad_norm": 0.6763295531272888,
      "learning_rate": 4.657940584917983e-05,
      "loss": 1.0332,
      "step": 7323
    },
    {
      "epoch": 1.343802267236573,
      "grad_norm": 0.7093154191970825,
      "learning_rate": 4.655468499987612e-05,
      "loss": 1.0181,
      "step": 7324
    },
    {
      "epoch": 1.3439881062999444,
      "grad_norm": 0.7465590238571167,
      "learning_rate": 4.652996872186567e-05,
      "loss": 0.8733,
      "step": 7325
    },
    {
      "epoch": 1.3441739453633152,
      "grad_norm": 0.7101390361785889,
      "learning_rate": 4.650525701726256e-05,
      "loss": 0.8824,
      "step": 7326
    },
    {
      "epoch": 1.3443597844266866,
      "grad_norm": 0.6595645546913147,
      "learning_rate": 4.648054988818048e-05,
      "loss": 0.9001,
      "step": 7327
    },
    {
      "epoch": 1.3445456234900577,
      "grad_norm": 0.6411036849021912,
      "learning_rate": 4.6455847336732593e-05,
      "loss": 0.8999,
      "step": 7328
    },
    {
      "epoch": 1.3447314625534288,
      "grad_norm": 0.6612711548805237,
      "learning_rate": 4.6431149365031855e-05,
      "loss": 1.0747,
      "step": 7329
    },
    {
      "epoch": 1.3449173016168,
      "grad_norm": 0.7382752299308777,
      "learning_rate": 4.6406455975190645e-05,
      "loss": 0.899,
      "step": 7330
    },
    {
      "epoch": 1.345103140680171,
      "grad_norm": 0.8373868465423584,
      "learning_rate": 4.638176716932109e-05,
      "loss": 1.1021,
      "step": 7331
    },
    {
      "epoch": 1.3452889797435421,
      "grad_norm": 0.5916196703910828,
      "learning_rate": 4.635708294953487e-05,
      "loss": 0.6864,
      "step": 7332
    },
    {
      "epoch": 1.3454748188069132,
      "grad_norm": 0.6688162088394165,
      "learning_rate": 4.633240331794328e-05,
      "loss": 1.0403,
      "step": 7333
    },
    {
      "epoch": 1.3456606578702843,
      "grad_norm": 0.8366694450378418,
      "learning_rate": 4.630772827665727e-05,
      "loss": 0.9681,
      "step": 7334
    },
    {
      "epoch": 1.3458464969336554,
      "grad_norm": 0.7941098809242249,
      "learning_rate": 4.628305782778727e-05,
      "loss": 0.9695,
      "step": 7335
    },
    {
      "epoch": 1.3460323359970265,
      "grad_norm": 0.6894522309303284,
      "learning_rate": 4.625839197344346e-05,
      "loss": 0.7946,
      "step": 7336
    },
    {
      "epoch": 1.3462181750603976,
      "grad_norm": 0.6391467452049255,
      "learning_rate": 4.62337307157355e-05,
      "loss": 0.8445,
      "step": 7337
    },
    {
      "epoch": 1.3464040141237688,
      "grad_norm": 0.6794978976249695,
      "learning_rate": 4.62090740567728e-05,
      "loss": 0.9014,
      "step": 7338
    },
    {
      "epoch": 1.3465898531871399,
      "grad_norm": 0.6898692846298218,
      "learning_rate": 4.618442199866421e-05,
      "loss": 0.8583,
      "step": 7339
    },
    {
      "epoch": 1.3467756922505112,
      "grad_norm": 1.0231659412384033,
      "learning_rate": 4.615977454351832e-05,
      "loss": 1.1721,
      "step": 7340
    },
    {
      "epoch": 1.346961531313882,
      "grad_norm": 0.6773861646652222,
      "learning_rate": 4.613513169344331e-05,
      "loss": 0.8633,
      "step": 7341
    },
    {
      "epoch": 1.3471473703772534,
      "grad_norm": 0.7153002619743347,
      "learning_rate": 4.6110493450546866e-05,
      "loss": 0.777,
      "step": 7342
    },
    {
      "epoch": 1.3473332094406243,
      "grad_norm": 0.7849166989326477,
      "learning_rate": 4.608585981693643e-05,
      "loss": 0.866,
      "step": 7343
    },
    {
      "epoch": 1.3475190485039956,
      "grad_norm": 0.6351103782653809,
      "learning_rate": 4.606123079471889e-05,
      "loss": 1.0098,
      "step": 7344
    },
    {
      "epoch": 1.3477048875673667,
      "grad_norm": 0.6069339513778687,
      "learning_rate": 4.6036606386000844e-05,
      "loss": 1.0055,
      "step": 7345
    },
    {
      "epoch": 1.3478907266307378,
      "grad_norm": 0.7701793313026428,
      "learning_rate": 4.60119865928885e-05,
      "loss": 0.7519,
      "step": 7346
    },
    {
      "epoch": 1.348076565694109,
      "grad_norm": 0.6905173063278198,
      "learning_rate": 4.598737141748766e-05,
      "loss": 1.008,
      "step": 7347
    },
    {
      "epoch": 1.34826240475748,
      "grad_norm": 0.6580630540847778,
      "learning_rate": 4.5962760861903644e-05,
      "loss": 0.8601,
      "step": 7348
    },
    {
      "epoch": 1.3484482438208512,
      "grad_norm": 0.6992924213409424,
      "learning_rate": 4.593815492824147e-05,
      "loss": 0.9277,
      "step": 7349
    },
    {
      "epoch": 1.3486340828842223,
      "grad_norm": 0.7928568124771118,
      "learning_rate": 4.59135536186058e-05,
      "loss": 0.9202,
      "step": 7350
    },
    {
      "epoch": 1.3488199219475934,
      "grad_norm": 0.910819947719574,
      "learning_rate": 4.5888956935100745e-05,
      "loss": 1.0964,
      "step": 7351
    },
    {
      "epoch": 1.3490057610109645,
      "grad_norm": 0.6011480689048767,
      "learning_rate": 4.586436487983018e-05,
      "loss": 0.7151,
      "step": 7352
    },
    {
      "epoch": 1.3491916000743356,
      "grad_norm": 0.67581707239151,
      "learning_rate": 4.583977745489746e-05,
      "loss": 0.8922,
      "step": 7353
    },
    {
      "epoch": 1.3493774391377067,
      "grad_norm": 0.6327905654907227,
      "learning_rate": 4.5815194662405635e-05,
      "loss": 0.8786,
      "step": 7354
    },
    {
      "epoch": 1.3495632782010778,
      "grad_norm": 0.6018652319908142,
      "learning_rate": 4.579061650445735e-05,
      "loss": 0.9744,
      "step": 7355
    },
    {
      "epoch": 1.349749117264449,
      "grad_norm": 0.9206721186637878,
      "learning_rate": 4.576604298315476e-05,
      "loss": 0.8008,
      "step": 7356
    },
    {
      "epoch": 1.3499349563278202,
      "grad_norm": 0.6685934662818909,
      "learning_rate": 4.5741474100599756e-05,
      "loss": 0.8541,
      "step": 7357
    },
    {
      "epoch": 1.3501207953911911,
      "grad_norm": 0.7667933702468872,
      "learning_rate": 4.571690985889372e-05,
      "loss": 0.9087,
      "step": 7358
    },
    {
      "epoch": 1.3503066344545624,
      "grad_norm": 0.6506810784339905,
      "learning_rate": 4.569235026013773e-05,
      "loss": 0.8424,
      "step": 7359
    },
    {
      "epoch": 1.3504924735179333,
      "grad_norm": 0.8654145002365112,
      "learning_rate": 4.566779530643237e-05,
      "loss": 0.8636,
      "step": 7360
    },
    {
      "epoch": 1.3506783125813047,
      "grad_norm": 0.7595291137695312,
      "learning_rate": 4.56432449998779e-05,
      "loss": 1.2996,
      "step": 7361
    },
    {
      "epoch": 1.3508641516446758,
      "grad_norm": 0.6884271502494812,
      "learning_rate": 4.561869934257417e-05,
      "loss": 1.0624,
      "step": 7362
    },
    {
      "epoch": 1.3510499907080469,
      "grad_norm": 0.6496449708938599,
      "learning_rate": 4.559415833662061e-05,
      "loss": 0.9485,
      "step": 7363
    },
    {
      "epoch": 1.351235829771418,
      "grad_norm": 0.7063813805580139,
      "learning_rate": 4.556962198411631e-05,
      "loss": 1.0707,
      "step": 7364
    },
    {
      "epoch": 1.351421668834789,
      "grad_norm": 0.7109731435775757,
      "learning_rate": 4.5545090287159854e-05,
      "loss": 1.2407,
      "step": 7365
    },
    {
      "epoch": 1.3516075078981602,
      "grad_norm": 0.7253541350364685,
      "learning_rate": 4.552056324784954e-05,
      "loss": 1.0913,
      "step": 7366
    },
    {
      "epoch": 1.3517933469615313,
      "grad_norm": 0.7168422937393188,
      "learning_rate": 4.5496040868283177e-05,
      "loss": 0.8611,
      "step": 7367
    },
    {
      "epoch": 1.3519791860249024,
      "grad_norm": 0.7601984143257141,
      "learning_rate": 4.5471523150558225e-05,
      "loss": 1.1782,
      "step": 7368
    },
    {
      "epoch": 1.3521650250882735,
      "grad_norm": 0.7291967272758484,
      "learning_rate": 4.5447010096771794e-05,
      "loss": 1.1117,
      "step": 7369
    },
    {
      "epoch": 1.3523508641516446,
      "grad_norm": 0.7122645378112793,
      "learning_rate": 4.542250170902045e-05,
      "loss": 1.0315,
      "step": 7370
    },
    {
      "epoch": 1.3525367032150157,
      "grad_norm": 0.800542414188385,
      "learning_rate": 4.5397997989400544e-05,
      "loss": 0.9357,
      "step": 7371
    },
    {
      "epoch": 1.3527225422783868,
      "grad_norm": 0.6745566129684448,
      "learning_rate": 4.5373498940007834e-05,
      "loss": 1.1384,
      "step": 7372
    },
    {
      "epoch": 1.352908381341758,
      "grad_norm": 0.6796730160713196,
      "learning_rate": 4.534900456293786e-05,
      "loss": 0.8074,
      "step": 7373
    },
    {
      "epoch": 1.3530942204051293,
      "grad_norm": 0.7584159970283508,
      "learning_rate": 4.532451486028559e-05,
      "loss": 1.0569,
      "step": 7374
    },
    {
      "epoch": 1.3532800594685002,
      "grad_norm": 0.6289095878601074,
      "learning_rate": 4.530002983414575e-05,
      "loss": 0.7696,
      "step": 7375
    },
    {
      "epoch": 1.3534658985318715,
      "grad_norm": 0.6373142600059509,
      "learning_rate": 4.527554948661258e-05,
      "loss": 1.1354,
      "step": 7376
    },
    {
      "epoch": 1.3536517375952426,
      "grad_norm": 0.7420815229415894,
      "learning_rate": 4.525107381977993e-05,
      "loss": 0.947,
      "step": 7377
    },
    {
      "epoch": 1.3538375766586137,
      "grad_norm": 0.6928457617759705,
      "learning_rate": 4.522660283574132e-05,
      "loss": 0.8624,
      "step": 7378
    },
    {
      "epoch": 1.3540234157219848,
      "grad_norm": 0.7249012589454651,
      "learning_rate": 4.520213653658969e-05,
      "loss": 0.8162,
      "step": 7379
    },
    {
      "epoch": 1.354209254785356,
      "grad_norm": 0.6938693523406982,
      "learning_rate": 4.517767492441781e-05,
      "loss": 0.9954,
      "step": 7380
    },
    {
      "epoch": 1.354395093848727,
      "grad_norm": 0.9390332698822021,
      "learning_rate": 4.515321800131783e-05,
      "loss": 0.988,
      "step": 7381
    },
    {
      "epoch": 1.3545809329120981,
      "grad_norm": 0.5386117100715637,
      "learning_rate": 4.512876576938167e-05,
      "loss": 0.6488,
      "step": 7382
    },
    {
      "epoch": 1.3547667719754692,
      "grad_norm": 0.615216076374054,
      "learning_rate": 4.5104318230700784e-05,
      "loss": 0.7681,
      "step": 7383
    },
    {
      "epoch": 1.3549526110388403,
      "grad_norm": 0.8385409116744995,
      "learning_rate": 4.507987538736618e-05,
      "loss": 1.171,
      "step": 7384
    },
    {
      "epoch": 1.3551384501022115,
      "grad_norm": 0.7255057692527771,
      "learning_rate": 4.5055437241468565e-05,
      "loss": 0.958,
      "step": 7385
    },
    {
      "epoch": 1.3553242891655826,
      "grad_norm": 0.7279541492462158,
      "learning_rate": 4.5031003795098116e-05,
      "loss": 0.9985,
      "step": 7386
    },
    {
      "epoch": 1.3555101282289537,
      "grad_norm": 0.6622303128242493,
      "learning_rate": 4.5006575050344754e-05,
      "loss": 1.1728,
      "step": 7387
    },
    {
      "epoch": 1.3556959672923248,
      "grad_norm": 0.7048065662384033,
      "learning_rate": 4.498215100929786e-05,
      "loss": 0.9017,
      "step": 7388
    },
    {
      "epoch": 1.3558818063556959,
      "grad_norm": 0.6078386902809143,
      "learning_rate": 4.4957731674046474e-05,
      "loss": 0.9736,
      "step": 7389
    },
    {
      "epoch": 1.356067645419067,
      "grad_norm": 0.7244036197662354,
      "learning_rate": 4.493331704667928e-05,
      "loss": 0.943,
      "step": 7390
    },
    {
      "epoch": 1.3562534844824383,
      "grad_norm": 0.7068463563919067,
      "learning_rate": 4.490890712928448e-05,
      "loss": 0.8434,
      "step": 7391
    },
    {
      "epoch": 1.3564393235458092,
      "grad_norm": 0.7234067320823669,
      "learning_rate": 4.488450192394997e-05,
      "loss": 1.2149,
      "step": 7392
    },
    {
      "epoch": 1.3566251626091805,
      "grad_norm": 0.6822995543479919,
      "learning_rate": 4.486010143276309e-05,
      "loss": 0.8057,
      "step": 7393
    },
    {
      "epoch": 1.3568110016725516,
      "grad_norm": 0.6011866331100464,
      "learning_rate": 4.4835705657810946e-05,
      "loss": 0.9052,
      "step": 7394
    },
    {
      "epoch": 1.3569968407359227,
      "grad_norm": 0.6928417086601257,
      "learning_rate": 4.4811314601180076e-05,
      "loss": 1.017,
      "step": 7395
    },
    {
      "epoch": 1.3571826797992939,
      "grad_norm": 0.6541099548339844,
      "learning_rate": 4.4786928264956764e-05,
      "loss": 0.7197,
      "step": 7396
    },
    {
      "epoch": 1.357368518862665,
      "grad_norm": 0.7261728644371033,
      "learning_rate": 4.476254665122683e-05,
      "loss": 0.9874,
      "step": 7397
    },
    {
      "epoch": 1.357554357926036,
      "grad_norm": 0.7361965775489807,
      "learning_rate": 4.473816976207563e-05,
      "loss": 0.9259,
      "step": 7398
    },
    {
      "epoch": 1.3577401969894072,
      "grad_norm": 0.672639787197113,
      "learning_rate": 4.471379759958825e-05,
      "loss": 0.8932,
      "step": 7399
    },
    {
      "epoch": 1.3579260360527783,
      "grad_norm": 0.7823107838630676,
      "learning_rate": 4.468943016584921e-05,
      "loss": 1.0944,
      "step": 7400
    },
    {
      "epoch": 1.3581118751161494,
      "grad_norm": 0.7286608219146729,
      "learning_rate": 4.466506746294279e-05,
      "loss": 1.1196,
      "step": 7401
    },
    {
      "epoch": 1.3582977141795205,
      "grad_norm": 0.9249753355979919,
      "learning_rate": 4.46407094929527e-05,
      "loss": 0.989,
      "step": 7402
    },
    {
      "epoch": 1.3584835532428916,
      "grad_norm": 0.7424798607826233,
      "learning_rate": 4.4616356257962375e-05,
      "loss": 0.8082,
      "step": 7403
    },
    {
      "epoch": 1.3586693923062627,
      "grad_norm": 0.5924530029296875,
      "learning_rate": 4.459200776005484e-05,
      "loss": 0.7535,
      "step": 7404
    },
    {
      "epoch": 1.3588552313696338,
      "grad_norm": 0.763904333114624,
      "learning_rate": 4.45676640013126e-05,
      "loss": 0.895,
      "step": 7405
    },
    {
      "epoch": 1.3590410704330051,
      "grad_norm": 0.6832556128501892,
      "learning_rate": 4.4543324983817855e-05,
      "loss": 0.8652,
      "step": 7406
    },
    {
      "epoch": 1.359226909496376,
      "grad_norm": 0.7321500182151794,
      "learning_rate": 4.451899070965239e-05,
      "loss": 0.9739,
      "step": 7407
    },
    {
      "epoch": 1.3594127485597474,
      "grad_norm": 0.7232654094696045,
      "learning_rate": 4.449466118089759e-05,
      "loss": 0.96,
      "step": 7408
    },
    {
      "epoch": 1.3595985876231182,
      "grad_norm": 0.7119808793067932,
      "learning_rate": 4.447033639963435e-05,
      "loss": 1.087,
      "step": 7409
    },
    {
      "epoch": 1.3597844266864896,
      "grad_norm": 0.7229185104370117,
      "learning_rate": 4.444601636794326e-05,
      "loss": 0.8998,
      "step": 7410
    },
    {
      "epoch": 1.3599702657498607,
      "grad_norm": 0.8261592388153076,
      "learning_rate": 4.4421701087904474e-05,
      "loss": 0.8719,
      "step": 7411
    },
    {
      "epoch": 1.3601561048132318,
      "grad_norm": 0.756256639957428,
      "learning_rate": 4.439739056159769e-05,
      "loss": 0.8449,
      "step": 7412
    },
    {
      "epoch": 1.360341943876603,
      "grad_norm": 0.6536577343940735,
      "learning_rate": 4.4373084791102305e-05,
      "loss": 0.74,
      "step": 7413
    },
    {
      "epoch": 1.360527782939974,
      "grad_norm": 0.7058019638061523,
      "learning_rate": 4.434878377849716e-05,
      "loss": 0.9292,
      "step": 7414
    },
    {
      "epoch": 1.3607136220033451,
      "grad_norm": 0.6998428702354431,
      "learning_rate": 4.432448752586086e-05,
      "loss": 1.025,
      "step": 7415
    },
    {
      "epoch": 1.3608994610667162,
      "grad_norm": 0.8001590967178345,
      "learning_rate": 4.430019603527144e-05,
      "loss": 0.8781,
      "step": 7416
    },
    {
      "epoch": 1.3610853001300873,
      "grad_norm": 0.695641815662384,
      "learning_rate": 4.4275909308806615e-05,
      "loss": 1.0285,
      "step": 7417
    },
    {
      "epoch": 1.3612711391934584,
      "grad_norm": 0.7073414921760559,
      "learning_rate": 4.425162734854375e-05,
      "loss": 0.9246,
      "step": 7418
    },
    {
      "epoch": 1.3614569782568295,
      "grad_norm": 0.7373897433280945,
      "learning_rate": 4.4227350156559654e-05,
      "loss": 1.0972,
      "step": 7419
    },
    {
      "epoch": 1.3616428173202006,
      "grad_norm": 0.8083869814872742,
      "learning_rate": 4.420307773493084e-05,
      "loss": 0.928,
      "step": 7420
    },
    {
      "epoch": 1.3618286563835718,
      "grad_norm": 0.6736962795257568,
      "learning_rate": 4.417881008573338e-05,
      "loss": 0.9647,
      "step": 7421
    },
    {
      "epoch": 1.3620144954469429,
      "grad_norm": 0.783660888671875,
      "learning_rate": 4.4154547211042976e-05,
      "loss": 0.9698,
      "step": 7422
    },
    {
      "epoch": 1.3622003345103142,
      "grad_norm": 0.7382241487503052,
      "learning_rate": 4.41302891129348e-05,
      "loss": 1.017,
      "step": 7423
    },
    {
      "epoch": 1.362386173573685,
      "grad_norm": 0.7947919368743896,
      "learning_rate": 4.410603579348376e-05,
      "loss": 0.807,
      "step": 7424
    },
    {
      "epoch": 1.3625720126370564,
      "grad_norm": 0.629784882068634,
      "learning_rate": 4.408178725476431e-05,
      "loss": 0.853,
      "step": 7425
    },
    {
      "epoch": 1.3627578517004273,
      "grad_norm": 0.6849907636642456,
      "learning_rate": 4.405754349885042e-05,
      "loss": 0.935,
      "step": 7426
    },
    {
      "epoch": 1.3629436907637986,
      "grad_norm": 0.7976436614990234,
      "learning_rate": 4.403330452781577e-05,
      "loss": 0.9449,
      "step": 7427
    },
    {
      "epoch": 1.3631295298271697,
      "grad_norm": 0.6531306505203247,
      "learning_rate": 4.400907034373351e-05,
      "loss": 0.8535,
      "step": 7428
    },
    {
      "epoch": 1.3633153688905408,
      "grad_norm": 0.7673757076263428,
      "learning_rate": 4.398484094867652e-05,
      "loss": 0.8225,
      "step": 7429
    },
    {
      "epoch": 1.363501207953912,
      "grad_norm": 0.683928906917572,
      "learning_rate": 4.39606163447171e-05,
      "loss": 0.6675,
      "step": 7430
    },
    {
      "epoch": 1.363687047017283,
      "grad_norm": 0.7557547092437744,
      "learning_rate": 4.393639653392729e-05,
      "loss": 0.9763,
      "step": 7431
    },
    {
      "epoch": 1.3638728860806542,
      "grad_norm": 0.8278974294662476,
      "learning_rate": 4.391218151837868e-05,
      "loss": 1.0336,
      "step": 7432
    },
    {
      "epoch": 1.3640587251440253,
      "grad_norm": 0.8780465722084045,
      "learning_rate": 4.388797130014238e-05,
      "loss": 1.1456,
      "step": 7433
    },
    {
      "epoch": 1.3642445642073964,
      "grad_norm": 0.7346441149711609,
      "learning_rate": 4.3863765881289165e-05,
      "loss": 0.9235,
      "step": 7434
    },
    {
      "epoch": 1.3644304032707675,
      "grad_norm": 1.674783706665039,
      "learning_rate": 4.3839565263889385e-05,
      "loss": 1.0321,
      "step": 7435
    },
    {
      "epoch": 1.3646162423341386,
      "grad_norm": 0.7685444355010986,
      "learning_rate": 4.3815369450013e-05,
      "loss": 0.8211,
      "step": 7436
    },
    {
      "epoch": 1.3648020813975097,
      "grad_norm": 0.6860784292221069,
      "learning_rate": 4.3791178441729466e-05,
      "loss": 1.0005,
      "step": 7437
    },
    {
      "epoch": 1.3649879204608808,
      "grad_norm": 0.7961957454681396,
      "learning_rate": 4.3766992241107916e-05,
      "loss": 1.0806,
      "step": 7438
    },
    {
      "epoch": 1.365173759524252,
      "grad_norm": 0.7249303460121155,
      "learning_rate": 4.3742810850217084e-05,
      "loss": 0.9695,
      "step": 7439
    },
    {
      "epoch": 1.3653595985876232,
      "grad_norm": 0.8331009149551392,
      "learning_rate": 4.37186342711252e-05,
      "loss": 0.7105,
      "step": 7440
    },
    {
      "epoch": 1.3655454376509941,
      "grad_norm": 0.8023040890693665,
      "learning_rate": 4.3694462505900215e-05,
      "loss": 0.984,
      "step": 7441
    },
    {
      "epoch": 1.3657312767143654,
      "grad_norm": 0.8293659090995789,
      "learning_rate": 4.367029555660949e-05,
      "loss": 1.0359,
      "step": 7442
    },
    {
      "epoch": 1.3659171157777366,
      "grad_norm": 0.6645398736000061,
      "learning_rate": 4.364613342532018e-05,
      "loss": 0.6134,
      "step": 7443
    },
    {
      "epoch": 1.3661029548411077,
      "grad_norm": 0.8250023126602173,
      "learning_rate": 4.3621976114098826e-05,
      "loss": 0.9576,
      "step": 7444
    },
    {
      "epoch": 1.3662887939044788,
      "grad_norm": 0.6180599927902222,
      "learning_rate": 4.3597823625011714e-05,
      "loss": 0.9442,
      "step": 7445
    },
    {
      "epoch": 1.3664746329678499,
      "grad_norm": 0.765338659286499,
      "learning_rate": 4.3573675960124684e-05,
      "loss": 0.8555,
      "step": 7446
    },
    {
      "epoch": 1.366660472031221,
      "grad_norm": 0.6484029293060303,
      "learning_rate": 4.354953312150305e-05,
      "loss": 0.7575,
      "step": 7447
    },
    {
      "epoch": 1.366846311094592,
      "grad_norm": 0.625566303730011,
      "learning_rate": 4.3525395111211877e-05,
      "loss": 0.9547,
      "step": 7448
    },
    {
      "epoch": 1.3670321501579632,
      "grad_norm": 0.6785605549812317,
      "learning_rate": 4.350126193131574e-05,
      "loss": 1.0458,
      "step": 7449
    },
    {
      "epoch": 1.3672179892213343,
      "grad_norm": 0.762402355670929,
      "learning_rate": 4.347713358387875e-05,
      "loss": 0.9349,
      "step": 7450
    },
    {
      "epoch": 1.3674038282847054,
      "grad_norm": 0.6979005336761475,
      "learning_rate": 4.3453010070964684e-05,
      "loss": 1.0819,
      "step": 7451
    },
    {
      "epoch": 1.3675896673480765,
      "grad_norm": 0.8139050006866455,
      "learning_rate": 4.3428891394636874e-05,
      "loss": 0.9195,
      "step": 7452
    },
    {
      "epoch": 1.3677755064114476,
      "grad_norm": 0.7950904369354248,
      "learning_rate": 4.34047775569583e-05,
      "loss": 0.8346,
      "step": 7453
    },
    {
      "epoch": 1.3679613454748187,
      "grad_norm": 0.6743336915969849,
      "learning_rate": 4.3380668559991366e-05,
      "loss": 0.9345,
      "step": 7454
    },
    {
      "epoch": 1.36814718453819,
      "grad_norm": 0.7096421718597412,
      "learning_rate": 4.335656440579827e-05,
      "loss": 0.7283,
      "step": 7455
    },
    {
      "epoch": 1.368333023601561,
      "grad_norm": 0.5938029885292053,
      "learning_rate": 4.33324650964406e-05,
      "loss": 0.825,
      "step": 7456
    },
    {
      "epoch": 1.3685188626649323,
      "grad_norm": 0.7945120930671692,
      "learning_rate": 4.330837063397971e-05,
      "loss": 1.104,
      "step": 7457
    },
    {
      "epoch": 1.3687047017283032,
      "grad_norm": 0.7108601331710815,
      "learning_rate": 4.328428102047636e-05,
      "loss": 0.8969,
      "step": 7458
    },
    {
      "epoch": 1.3688905407916745,
      "grad_norm": 0.6471177935600281,
      "learning_rate": 4.3260196257991035e-05,
      "loss": 0.9276,
      "step": 7459
    },
    {
      "epoch": 1.3690763798550456,
      "grad_norm": 0.8287071585655212,
      "learning_rate": 4.32361163485838e-05,
      "loss": 0.9662,
      "step": 7460
    },
    {
      "epoch": 1.3692622189184167,
      "grad_norm": 0.7040210366249084,
      "learning_rate": 4.321204129431418e-05,
      "loss": 0.9346,
      "step": 7461
    },
    {
      "epoch": 1.3694480579817878,
      "grad_norm": 0.7015113234519958,
      "learning_rate": 4.3187971097241416e-05,
      "loss": 1.0136,
      "step": 7462
    },
    {
      "epoch": 1.369633897045159,
      "grad_norm": 0.656891942024231,
      "learning_rate": 4.3163905759424264e-05,
      "loss": 0.9282,
      "step": 7463
    },
    {
      "epoch": 1.36981973610853,
      "grad_norm": 0.6929020285606384,
      "learning_rate": 4.313984528292108e-05,
      "loss": 0.958,
      "step": 7464
    },
    {
      "epoch": 1.3700055751719011,
      "grad_norm": 0.7964670062065125,
      "learning_rate": 4.311578966978982e-05,
      "loss": 0.9406,
      "step": 7465
    },
    {
      "epoch": 1.3701914142352722,
      "grad_norm": 0.7206087708473206,
      "learning_rate": 4.309173892208801e-05,
      "loss": 0.9252,
      "step": 7466
    },
    {
      "epoch": 1.3703772532986433,
      "grad_norm": 0.7929943203926086,
      "learning_rate": 4.3067693041872805e-05,
      "loss": 1.1453,
      "step": 7467
    },
    {
      "epoch": 1.3705630923620145,
      "grad_norm": 0.7283685207366943,
      "learning_rate": 4.304365203120082e-05,
      "loss": 1.1698,
      "step": 7468
    },
    {
      "epoch": 1.3707489314253856,
      "grad_norm": 0.7081707715988159,
      "learning_rate": 4.301961589212842e-05,
      "loss": 0.9548,
      "step": 7469
    },
    {
      "epoch": 1.3709347704887567,
      "grad_norm": 0.7527585029602051,
      "learning_rate": 4.299558462671137e-05,
      "loss": 1.0448,
      "step": 7470
    },
    {
      "epoch": 1.3711206095521278,
      "grad_norm": 0.7582692503929138,
      "learning_rate": 4.2971558237005225e-05,
      "loss": 0.901,
      "step": 7471
    },
    {
      "epoch": 1.371306448615499,
      "grad_norm": 0.7193834781646729,
      "learning_rate": 4.2947536725064906e-05,
      "loss": 1.0733,
      "step": 7472
    },
    {
      "epoch": 1.37149228767887,
      "grad_norm": 0.6494482755661011,
      "learning_rate": 4.292352009294509e-05,
      "loss": 0.7626,
      "step": 7473
    },
    {
      "epoch": 1.3716781267422413,
      "grad_norm": 0.49988317489624023,
      "learning_rate": 4.289950834269999e-05,
      "loss": 0.4652,
      "step": 7474
    },
    {
      "epoch": 1.3718639658056122,
      "grad_norm": 0.7491035461425781,
      "learning_rate": 4.287550147638332e-05,
      "loss": 1.0236,
      "step": 7475
    },
    {
      "epoch": 1.3720498048689835,
      "grad_norm": 0.6134287714958191,
      "learning_rate": 4.2851499496048507e-05,
      "loss": 1.1179,
      "step": 7476
    },
    {
      "epoch": 1.3722356439323546,
      "grad_norm": 0.7895690202713013,
      "learning_rate": 4.282750240374842e-05,
      "loss": 0.8788,
      "step": 7477
    },
    {
      "epoch": 1.3724214829957257,
      "grad_norm": 0.61039799451828,
      "learning_rate": 4.280351020153561e-05,
      "loss": 0.9806,
      "step": 7478
    },
    {
      "epoch": 1.3726073220590969,
      "grad_norm": 0.7818472981452942,
      "learning_rate": 4.2779522891462196e-05,
      "loss": 1.1705,
      "step": 7479
    },
    {
      "epoch": 1.372793161122468,
      "grad_norm": 0.6983256340026855,
      "learning_rate": 4.275554047557987e-05,
      "loss": 0.9489,
      "step": 7480
    },
    {
      "epoch": 1.372979000185839,
      "grad_norm": 0.6386864185333252,
      "learning_rate": 4.273156295593991e-05,
      "loss": 0.8734,
      "step": 7481
    },
    {
      "epoch": 1.3731648392492102,
      "grad_norm": 0.6794342994689941,
      "learning_rate": 4.270759033459311e-05,
      "loss": 0.9008,
      "step": 7482
    },
    {
      "epoch": 1.3733506783125813,
      "grad_norm": 0.7258062362670898,
      "learning_rate": 4.268362261358997e-05,
      "loss": 0.9507,
      "step": 7483
    },
    {
      "epoch": 1.3735365173759524,
      "grad_norm": 0.9591363072395325,
      "learning_rate": 4.265965979498043e-05,
      "loss": 0.8545,
      "step": 7484
    },
    {
      "epoch": 1.3737223564393235,
      "grad_norm": 0.6767122745513916,
      "learning_rate": 4.263570188081416e-05,
      "loss": 1.0834,
      "step": 7485
    },
    {
      "epoch": 1.3739081955026946,
      "grad_norm": 0.63570636510849,
      "learning_rate": 4.2611748873140246e-05,
      "loss": 0.7912,
      "step": 7486
    },
    {
      "epoch": 1.3740940345660657,
      "grad_norm": 0.6445006132125854,
      "learning_rate": 4.258780077400748e-05,
      "loss": 0.8576,
      "step": 7487
    },
    {
      "epoch": 1.3742798736294368,
      "grad_norm": 0.711561918258667,
      "learning_rate": 4.2563857585464237e-05,
      "loss": 1.0954,
      "step": 7488
    },
    {
      "epoch": 1.3744657126928082,
      "grad_norm": 0.7400350570678711,
      "learning_rate": 4.253991930955835e-05,
      "loss": 0.8498,
      "step": 7489
    },
    {
      "epoch": 1.374651551756179,
      "grad_norm": 0.71037358045578,
      "learning_rate": 4.25159859483374e-05,
      "loss": 1.0236,
      "step": 7490
    },
    {
      "epoch": 1.3748373908195504,
      "grad_norm": 0.7470990419387817,
      "learning_rate": 4.249205750384836e-05,
      "loss": 1.0152,
      "step": 7491
    },
    {
      "epoch": 1.3750232298829215,
      "grad_norm": 0.7151445150375366,
      "learning_rate": 4.2468133978137945e-05,
      "loss": 0.8797,
      "step": 7492
    },
    {
      "epoch": 1.3752090689462926,
      "grad_norm": 1.017393946647644,
      "learning_rate": 4.244421537325237e-05,
      "loss": 1.1923,
      "step": 7493
    },
    {
      "epoch": 1.3753949080096637,
      "grad_norm": 0.7744883894920349,
      "learning_rate": 4.2420301691237487e-05,
      "loss": 1.015,
      "step": 7494
    },
    {
      "epoch": 1.3755807470730348,
      "grad_norm": 0.6467261910438538,
      "learning_rate": 4.239639293413861e-05,
      "loss": 1.0064,
      "step": 7495
    },
    {
      "epoch": 1.375766586136406,
      "grad_norm": 0.6293275952339172,
      "learning_rate": 4.2372489104000736e-05,
      "loss": 0.7665,
      "step": 7496
    },
    {
      "epoch": 1.375952425199777,
      "grad_norm": 1.479631781578064,
      "learning_rate": 4.234859020286845e-05,
      "loss": 1.2332,
      "step": 7497
    },
    {
      "epoch": 1.3761382642631481,
      "grad_norm": 0.6856657266616821,
      "learning_rate": 4.2324696232785824e-05,
      "loss": 1.0395,
      "step": 7498
    },
    {
      "epoch": 1.3763241033265192,
      "grad_norm": 0.670069694519043,
      "learning_rate": 4.230080719579661e-05,
      "loss": 0.9213,
      "step": 7499
    },
    {
      "epoch": 1.3765099423898903,
      "grad_norm": 0.79466313123703,
      "learning_rate": 4.227692309394402e-05,
      "loss": 1.0373,
      "step": 7500
    },
    {
      "epoch": 1.3766957814532614,
      "grad_norm": 0.9797833561897278,
      "learning_rate": 4.2253043929270964e-05,
      "loss": 1.1574,
      "step": 7501
    },
    {
      "epoch": 1.3768816205166325,
      "grad_norm": 0.5774651169776917,
      "learning_rate": 4.2229169703819904e-05,
      "loss": 0.8187,
      "step": 7502
    },
    {
      "epoch": 1.3770674595800036,
      "grad_norm": 1.4220142364501953,
      "learning_rate": 4.2205300419632774e-05,
      "loss": 1.1914,
      "step": 7503
    },
    {
      "epoch": 1.3772532986433748,
      "grad_norm": 0.6203631162643433,
      "learning_rate": 4.2181436078751256e-05,
      "loss": 0.9958,
      "step": 7504
    },
    {
      "epoch": 1.3774391377067459,
      "grad_norm": 0.915024995803833,
      "learning_rate": 4.215757668321643e-05,
      "loss": 0.951,
      "step": 7505
    },
    {
      "epoch": 1.3776249767701172,
      "grad_norm": 0.7773443460464478,
      "learning_rate": 4.213372223506912e-05,
      "loss": 1.0363,
      "step": 7506
    },
    {
      "epoch": 1.377810815833488,
      "grad_norm": 0.6697326898574829,
      "learning_rate": 4.210987273634958e-05,
      "loss": 0.7743,
      "step": 7507
    },
    {
      "epoch": 1.3779966548968594,
      "grad_norm": 0.7244952917098999,
      "learning_rate": 4.208602818909774e-05,
      "loss": 0.9596,
      "step": 7508
    },
    {
      "epoch": 1.3781824939602305,
      "grad_norm": 0.6577768921852112,
      "learning_rate": 4.206218859535308e-05,
      "loss": 0.6985,
      "step": 7509
    },
    {
      "epoch": 1.3783683330236016,
      "grad_norm": 0.6292741298675537,
      "learning_rate": 4.203835395715464e-05,
      "loss": 0.9942,
      "step": 7510
    },
    {
      "epoch": 1.3785541720869727,
      "grad_norm": 0.7271589636802673,
      "learning_rate": 4.20145242765411e-05,
      "loss": 1.03,
      "step": 7511
    },
    {
      "epoch": 1.3787400111503438,
      "grad_norm": 0.7603896856307983,
      "learning_rate": 4.1990699555550595e-05,
      "loss": 0.8599,
      "step": 7512
    },
    {
      "epoch": 1.378925850213715,
      "grad_norm": 0.6246262192726135,
      "learning_rate": 4.196687979622096e-05,
      "loss": 1.0653,
      "step": 7513
    },
    {
      "epoch": 1.379111689277086,
      "grad_norm": 0.6769149899482727,
      "learning_rate": 4.194306500058947e-05,
      "loss": 0.8994,
      "step": 7514
    },
    {
      "epoch": 1.3792975283404572,
      "grad_norm": 0.6372380256652832,
      "learning_rate": 4.191925517069314e-05,
      "loss": 0.8837,
      "step": 7515
    },
    {
      "epoch": 1.3794833674038283,
      "grad_norm": 0.6261329054832458,
      "learning_rate": 4.189545030856846e-05,
      "loss": 0.9174,
      "step": 7516
    },
    {
      "epoch": 1.3796692064671994,
      "grad_norm": 0.6543997526168823,
      "learning_rate": 4.187165041625146e-05,
      "loss": 1.073,
      "step": 7517
    },
    {
      "epoch": 1.3798550455305705,
      "grad_norm": 0.664309561252594,
      "learning_rate": 4.184785549577788e-05,
      "loss": 0.7874,
      "step": 7518
    },
    {
      "epoch": 1.3800408845939416,
      "grad_norm": 0.7015162110328674,
      "learning_rate": 4.182406554918285e-05,
      "loss": 0.9236,
      "step": 7519
    },
    {
      "epoch": 1.3802267236573127,
      "grad_norm": 0.6500989198684692,
      "learning_rate": 4.180028057850127e-05,
      "loss": 0.9888,
      "step": 7520
    },
    {
      "epoch": 1.380412562720684,
      "grad_norm": 0.9309687614440918,
      "learning_rate": 4.177650058576744e-05,
      "loss": 1.1334,
      "step": 7521
    },
    {
      "epoch": 1.380598401784055,
      "grad_norm": 0.7783293724060059,
      "learning_rate": 4.175272557301536e-05,
      "loss": 0.9452,
      "step": 7522
    },
    {
      "epoch": 1.3807842408474262,
      "grad_norm": 0.6977812647819519,
      "learning_rate": 4.1728955542278546e-05,
      "loss": 1.08,
      "step": 7523
    },
    {
      "epoch": 1.3809700799107971,
      "grad_norm": 0.7367052435874939,
      "learning_rate": 4.170519049559011e-05,
      "loss": 0.8711,
      "step": 7524
    },
    {
      "epoch": 1.3811559189741685,
      "grad_norm": 0.7824680805206299,
      "learning_rate": 4.168143043498276e-05,
      "loss": 0.8856,
      "step": 7525
    },
    {
      "epoch": 1.3813417580375396,
      "grad_norm": 0.6573517322540283,
      "learning_rate": 4.1657675362488655e-05,
      "loss": 1.0912,
      "step": 7526
    },
    {
      "epoch": 1.3815275971009107,
      "grad_norm": 0.7205615043640137,
      "learning_rate": 4.163392528013973e-05,
      "loss": 0.6739,
      "step": 7527
    },
    {
      "epoch": 1.3817134361642818,
      "grad_norm": 0.7122405171394348,
      "learning_rate": 4.161018018996727e-05,
      "loss": 1.0519,
      "step": 7528
    },
    {
      "epoch": 1.3818992752276529,
      "grad_norm": 0.6043539047241211,
      "learning_rate": 4.15864400940023e-05,
      "loss": 1.0287,
      "step": 7529
    },
    {
      "epoch": 1.382085114291024,
      "grad_norm": 0.7427747845649719,
      "learning_rate": 4.1562704994275394e-05,
      "loss": 0.7662,
      "step": 7530
    },
    {
      "epoch": 1.382270953354395,
      "grad_norm": 0.6464288234710693,
      "learning_rate": 4.153897489281658e-05,
      "loss": 0.8072,
      "step": 7531
    },
    {
      "epoch": 1.3824567924177662,
      "grad_norm": 0.7093892693519592,
      "learning_rate": 4.1515249791655654e-05,
      "loss": 0.9883,
      "step": 7532
    },
    {
      "epoch": 1.3826426314811373,
      "grad_norm": 0.7397565841674805,
      "learning_rate": 4.149152969282176e-05,
      "loss": 0.9869,
      "step": 7533
    },
    {
      "epoch": 1.3828284705445084,
      "grad_norm": 0.9709092378616333,
      "learning_rate": 4.146781459834384e-05,
      "loss": 1.0185,
      "step": 7534
    },
    {
      "epoch": 1.3830143096078795,
      "grad_norm": 0.6332142949104309,
      "learning_rate": 4.1444104510250184e-05,
      "loss": 0.9538,
      "step": 7535
    },
    {
      "epoch": 1.3832001486712506,
      "grad_norm": 0.7227572798728943,
      "learning_rate": 4.1420399430568834e-05,
      "loss": 0.974,
      "step": 7536
    },
    {
      "epoch": 1.3833859877346217,
      "grad_norm": 0.7384819984436035,
      "learning_rate": 4.139669936132733e-05,
      "loss": 0.8992,
      "step": 7537
    },
    {
      "epoch": 1.383571826797993,
      "grad_norm": 0.7241960167884827,
      "learning_rate": 4.1373004304552786e-05,
      "loss": 0.9805,
      "step": 7538
    },
    {
      "epoch": 1.383757665861364,
      "grad_norm": 0.6905572414398193,
      "learning_rate": 4.134931426227193e-05,
      "loss": 1.0361,
      "step": 7539
    },
    {
      "epoch": 1.3839435049247353,
      "grad_norm": 0.5935276746749878,
      "learning_rate": 4.1325629236510954e-05,
      "loss": 0.9253,
      "step": 7540
    },
    {
      "epoch": 1.3841293439881062,
      "grad_norm": 0.7509950399398804,
      "learning_rate": 4.1301949229295755e-05,
      "loss": 0.7917,
      "step": 7541
    },
    {
      "epoch": 1.3843151830514775,
      "grad_norm": 0.6783537268638611,
      "learning_rate": 4.127827424265166e-05,
      "loss": 0.8705,
      "step": 7542
    },
    {
      "epoch": 1.3845010221148486,
      "grad_norm": 0.7498492002487183,
      "learning_rate": 4.125460427860369e-05,
      "loss": 0.8801,
      "step": 7543
    },
    {
      "epoch": 1.3846868611782197,
      "grad_norm": 0.7158042192459106,
      "learning_rate": 4.123093933917642e-05,
      "loss": 0.9504,
      "step": 7544
    },
    {
      "epoch": 1.3848727002415908,
      "grad_norm": 0.7044070363044739,
      "learning_rate": 4.120727942639389e-05,
      "loss": 0.9132,
      "step": 7545
    },
    {
      "epoch": 1.385058539304962,
      "grad_norm": 0.7871155738830566,
      "learning_rate": 4.118362454227984e-05,
      "loss": 0.9689,
      "step": 7546
    },
    {
      "epoch": 1.385244378368333,
      "grad_norm": 0.7555184960365295,
      "learning_rate": 4.115997468885748e-05,
      "loss": 0.9615,
      "step": 7547
    },
    {
      "epoch": 1.3854302174317041,
      "grad_norm": 0.6864029765129089,
      "learning_rate": 4.113632986814968e-05,
      "loss": 0.7999,
      "step": 7548
    },
    {
      "epoch": 1.3856160564950752,
      "grad_norm": 0.812187135219574,
      "learning_rate": 4.111269008217879e-05,
      "loss": 0.9588,
      "step": 7549
    },
    {
      "epoch": 1.3858018955584464,
      "grad_norm": 0.8208745121955872,
      "learning_rate": 4.1089055332966764e-05,
      "loss": 1.1081,
      "step": 7550
    },
    {
      "epoch": 1.3859877346218175,
      "grad_norm": 0.8858060240745544,
      "learning_rate": 4.1065425622535216e-05,
      "loss": 0.8471,
      "step": 7551
    },
    {
      "epoch": 1.3861735736851886,
      "grad_norm": 0.751556396484375,
      "learning_rate": 4.104180095290514e-05,
      "loss": 1.0127,
      "step": 7552
    },
    {
      "epoch": 1.3863594127485597,
      "grad_norm": 0.6057085990905762,
      "learning_rate": 4.101818132609726e-05,
      "loss": 0.9103,
      "step": 7553
    },
    {
      "epoch": 1.3865452518119308,
      "grad_norm": 0.5512748956680298,
      "learning_rate": 4.09945667441318e-05,
      "loss": 0.7052,
      "step": 7554
    },
    {
      "epoch": 1.386731090875302,
      "grad_norm": 0.6736949682235718,
      "learning_rate": 4.0970957209028616e-05,
      "loss": 0.8159,
      "step": 7555
    },
    {
      "epoch": 1.386916929938673,
      "grad_norm": 0.7596806287765503,
      "learning_rate": 4.094735272280701e-05,
      "loss": 0.8295,
      "step": 7556
    },
    {
      "epoch": 1.3871027690020443,
      "grad_norm": 0.7188926935195923,
      "learning_rate": 4.0923753287485944e-05,
      "loss": 1.0136,
      "step": 7557
    },
    {
      "epoch": 1.3872886080654154,
      "grad_norm": 0.7635893821716309,
      "learning_rate": 4.090015890508399e-05,
      "loss": 0.9305,
      "step": 7558
    },
    {
      "epoch": 1.3874744471287865,
      "grad_norm": 0.8576083183288574,
      "learning_rate": 4.0876569577619125e-05,
      "loss": 0.9918,
      "step": 7559
    },
    {
      "epoch": 1.3876602861921576,
      "grad_norm": 0.7648732662200928,
      "learning_rate": 4.085298530710909e-05,
      "loss": 1.0151,
      "step": 7560
    },
    {
      "epoch": 1.3878461252555288,
      "grad_norm": 0.735910177230835,
      "learning_rate": 4.082940609557102e-05,
      "loss": 0.966,
      "step": 7561
    },
    {
      "epoch": 1.3880319643188999,
      "grad_norm": 0.5369407534599304,
      "learning_rate": 4.0805831945021764e-05,
      "loss": 0.5685,
      "step": 7562
    },
    {
      "epoch": 1.388217803382271,
      "grad_norm": 0.7792626619338989,
      "learning_rate": 4.07822628574776e-05,
      "loss": 1.1068,
      "step": 7563
    },
    {
      "epoch": 1.388403642445642,
      "grad_norm": 0.7470696568489075,
      "learning_rate": 4.075869883495449e-05,
      "loss": 0.9176,
      "step": 7564
    },
    {
      "epoch": 1.3885894815090132,
      "grad_norm": 0.6785575151443481,
      "learning_rate": 4.0735139879467935e-05,
      "loss": 0.9002,
      "step": 7565
    },
    {
      "epoch": 1.3887753205723843,
      "grad_norm": 0.8073457479476929,
      "learning_rate": 4.071158599303292e-05,
      "loss": 1.1224,
      "step": 7566
    },
    {
      "epoch": 1.3889611596357554,
      "grad_norm": 0.7951982021331787,
      "learning_rate": 4.0688037177664085e-05,
      "loss": 1.0072,
      "step": 7567
    },
    {
      "epoch": 1.3891469986991265,
      "grad_norm": 0.856749415397644,
      "learning_rate": 4.066449343537563e-05,
      "loss": 0.9278,
      "step": 7568
    },
    {
      "epoch": 1.3893328377624976,
      "grad_norm": 0.71700119972229,
      "learning_rate": 4.064095476818133e-05,
      "loss": 1.1046,
      "step": 7569
    },
    {
      "epoch": 1.3895186768258687,
      "grad_norm": 0.645438015460968,
      "learning_rate": 4.0617421178094416e-05,
      "loss": 0.89,
      "step": 7570
    },
    {
      "epoch": 1.3897045158892398,
      "grad_norm": 0.7050654292106628,
      "learning_rate": 4.059389266712782e-05,
      "loss": 0.8633,
      "step": 7571
    },
    {
      "epoch": 1.3898903549526112,
      "grad_norm": 0.9105086922645569,
      "learning_rate": 4.0570369237294006e-05,
      "loss": 1.1659,
      "step": 7572
    },
    {
      "epoch": 1.390076194015982,
      "grad_norm": 0.7280207276344299,
      "learning_rate": 4.0546850890604925e-05,
      "loss": 1.0042,
      "step": 7573
    },
    {
      "epoch": 1.3902620330793534,
      "grad_norm": 0.6948641538619995,
      "learning_rate": 4.0523337629072213e-05,
      "loss": 0.9946,
      "step": 7574
    },
    {
      "epoch": 1.3904478721427245,
      "grad_norm": 0.62689608335495,
      "learning_rate": 4.049982945470695e-05,
      "loss": 0.9436,
      "step": 7575
    },
    {
      "epoch": 1.3906337112060956,
      "grad_norm": 0.6683322191238403,
      "learning_rate": 4.04763263695199e-05,
      "loss": 0.8834,
      "step": 7576
    },
    {
      "epoch": 1.3908195502694667,
      "grad_norm": 0.7566063404083252,
      "learning_rate": 4.045282837552128e-05,
      "loss": 1.025,
      "step": 7577
    },
    {
      "epoch": 1.3910053893328378,
      "grad_norm": 0.7480912804603577,
      "learning_rate": 4.042933547472093e-05,
      "loss": 0.9125,
      "step": 7578
    },
    {
      "epoch": 1.391191228396209,
      "grad_norm": 0.797137439250946,
      "learning_rate": 4.0405847669128314e-05,
      "loss": 0.8781,
      "step": 7579
    },
    {
      "epoch": 1.39137706745958,
      "grad_norm": 0.6337542533874512,
      "learning_rate": 4.038236496075231e-05,
      "loss": 1.0563,
      "step": 7580
    },
    {
      "epoch": 1.3915629065229511,
      "grad_norm": 0.772243082523346,
      "learning_rate": 4.0358887351601474e-05,
      "loss": 1.061,
      "step": 7581
    },
    {
      "epoch": 1.3917487455863222,
      "grad_norm": 0.6702725887298584,
      "learning_rate": 4.0335414843683906e-05,
      "loss": 1.0178,
      "step": 7582
    },
    {
      "epoch": 1.3919345846496933,
      "grad_norm": 0.6551902294158936,
      "learning_rate": 4.03119474390073e-05,
      "loss": 1.0351,
      "step": 7583
    },
    {
      "epoch": 1.3921204237130644,
      "grad_norm": 0.7127566933631897,
      "learning_rate": 4.028848513957879e-05,
      "loss": 0.9753,
      "step": 7584
    },
    {
      "epoch": 1.3923062627764355,
      "grad_norm": 0.7440797686576843,
      "learning_rate": 4.026502794740521e-05,
      "loss": 0.8079,
      "step": 7585
    },
    {
      "epoch": 1.3924921018398067,
      "grad_norm": 0.8096966743469238,
      "learning_rate": 4.024157586449292e-05,
      "loss": 1.1204,
      "step": 7586
    },
    {
      "epoch": 1.392677940903178,
      "grad_norm": 0.6678999662399292,
      "learning_rate": 4.021812889284777e-05,
      "loss": 1.038,
      "step": 7587
    },
    {
      "epoch": 1.3928637799665489,
      "grad_norm": 0.6104394197463989,
      "learning_rate": 4.01946870344753e-05,
      "loss": 0.8797,
      "step": 7588
    },
    {
      "epoch": 1.3930496190299202,
      "grad_norm": 0.6613131761550903,
      "learning_rate": 4.0171250291380455e-05,
      "loss": 1.0774,
      "step": 7589
    },
    {
      "epoch": 1.393235458093291,
      "grad_norm": 0.7873337864875793,
      "learning_rate": 4.014781866556793e-05,
      "loss": 1.1021,
      "step": 7590
    },
    {
      "epoch": 1.3934212971566624,
      "grad_norm": 0.6655529141426086,
      "learning_rate": 4.012439215904179e-05,
      "loss": 0.9414,
      "step": 7591
    },
    {
      "epoch": 1.3936071362200335,
      "grad_norm": 0.7219845652580261,
      "learning_rate": 4.010097077380579e-05,
      "loss": 0.9931,
      "step": 7592
    },
    {
      "epoch": 1.3937929752834046,
      "grad_norm": 0.9080620408058167,
      "learning_rate": 4.007755451186327e-05,
      "loss": 0.8985,
      "step": 7593
    },
    {
      "epoch": 1.3939788143467757,
      "grad_norm": 0.6404428482055664,
      "learning_rate": 4.005414337521698e-05,
      "loss": 0.8768,
      "step": 7594
    },
    {
      "epoch": 1.3941646534101468,
      "grad_norm": 1.5202301740646362,
      "learning_rate": 4.003073736586937e-05,
      "loss": 1.1338,
      "step": 7595
    },
    {
      "epoch": 1.394350492473518,
      "grad_norm": 0.7384157180786133,
      "learning_rate": 4.000733648582242e-05,
      "loss": 1.0007,
      "step": 7596
    },
    {
      "epoch": 1.394536331536889,
      "grad_norm": 0.6847105026245117,
      "learning_rate": 3.998394073707763e-05,
      "loss": 0.9388,
      "step": 7597
    },
    {
      "epoch": 1.3947221706002602,
      "grad_norm": 0.6476550102233887,
      "learning_rate": 3.996055012163609e-05,
      "loss": 1.0399,
      "step": 7598
    },
    {
      "epoch": 1.3949080096636313,
      "grad_norm": 0.6664227247238159,
      "learning_rate": 3.993716464149846e-05,
      "loss": 0.7564,
      "step": 7599
    },
    {
      "epoch": 1.3950938487270024,
      "grad_norm": 0.7326140403747559,
      "learning_rate": 3.991378429866499e-05,
      "loss": 0.883,
      "step": 7600
    },
    {
      "epoch": 1.3952796877903735,
      "grad_norm": 0.6408746242523193,
      "learning_rate": 3.989040909513537e-05,
      "loss": 0.7608,
      "step": 7601
    },
    {
      "epoch": 1.3954655268537446,
      "grad_norm": 0.670562207698822,
      "learning_rate": 3.9867039032909015e-05,
      "loss": 0.9599,
      "step": 7602
    },
    {
      "epoch": 1.3956513659171157,
      "grad_norm": 0.6591134667396545,
      "learning_rate": 3.9843674113984744e-05,
      "loss": 0.8028,
      "step": 7603
    },
    {
      "epoch": 1.395837204980487,
      "grad_norm": 0.6754502654075623,
      "learning_rate": 3.9820314340361075e-05,
      "loss": 0.8671,
      "step": 7604
    },
    {
      "epoch": 1.396023044043858,
      "grad_norm": 0.713962972164154,
      "learning_rate": 3.979695971403594e-05,
      "loss": 1.078,
      "step": 7605
    },
    {
      "epoch": 1.3962088831072292,
      "grad_norm": 0.8512964248657227,
      "learning_rate": 3.977361023700696e-05,
      "loss": 1.0837,
      "step": 7606
    },
    {
      "epoch": 1.3963947221706001,
      "grad_norm": 0.697532057762146,
      "learning_rate": 3.975026591127131e-05,
      "loss": 1.0582,
      "step": 7607
    },
    {
      "epoch": 1.3965805612339715,
      "grad_norm": 0.866722822189331,
      "learning_rate": 3.9726926738825585e-05,
      "loss": 1.1474,
      "step": 7608
    },
    {
      "epoch": 1.3967664002973426,
      "grad_norm": 0.7146264910697937,
      "learning_rate": 3.970359272166613e-05,
      "loss": 0.9921,
      "step": 7609
    },
    {
      "epoch": 1.3969522393607137,
      "grad_norm": 0.6930286884307861,
      "learning_rate": 3.968026386178867e-05,
      "loss": 0.9954,
      "step": 7610
    },
    {
      "epoch": 1.3971380784240848,
      "grad_norm": 0.6407985687255859,
      "learning_rate": 3.965694016118861e-05,
      "loss": 0.9258,
      "step": 7611
    },
    {
      "epoch": 1.3973239174874559,
      "grad_norm": 0.6215234994888306,
      "learning_rate": 3.963362162186086e-05,
      "loss": 0.8348,
      "step": 7612
    },
    {
      "epoch": 1.397509756550827,
      "grad_norm": 0.7461937665939331,
      "learning_rate": 3.961030824579994e-05,
      "loss": 0.9097,
      "step": 7613
    },
    {
      "epoch": 1.397695595614198,
      "grad_norm": 0.6643327474594116,
      "learning_rate": 3.958700003499991e-05,
      "loss": 0.9566,
      "step": 7614
    },
    {
      "epoch": 1.3978814346775692,
      "grad_norm": 0.89393150806427,
      "learning_rate": 3.956369699145429e-05,
      "loss": 0.9473,
      "step": 7615
    },
    {
      "epoch": 1.3980672737409403,
      "grad_norm": 0.6691911220550537,
      "learning_rate": 3.954039911715632e-05,
      "loss": 1.0246,
      "step": 7616
    },
    {
      "epoch": 1.3982531128043114,
      "grad_norm": 0.7593314051628113,
      "learning_rate": 3.951710641409867e-05,
      "loss": 0.9746,
      "step": 7617
    },
    {
      "epoch": 1.3984389518676825,
      "grad_norm": 0.7830350995063782,
      "learning_rate": 3.9493818884273646e-05,
      "loss": 1.0272,
      "step": 7618
    },
    {
      "epoch": 1.3986247909310536,
      "grad_norm": 0.7209092974662781,
      "learning_rate": 3.947053652967303e-05,
      "loss": 0.9343,
      "step": 7619
    },
    {
      "epoch": 1.3988106299944247,
      "grad_norm": 0.6709997057914734,
      "learning_rate": 3.944725935228826e-05,
      "loss": 0.7919,
      "step": 7620
    },
    {
      "epoch": 1.398996469057796,
      "grad_norm": 0.6931282877922058,
      "learning_rate": 3.942398735411028e-05,
      "loss": 0.8118,
      "step": 7621
    },
    {
      "epoch": 1.399182308121167,
      "grad_norm": 0.6827849745750427,
      "learning_rate": 3.9400720537129564e-05,
      "loss": 1.0486,
      "step": 7622
    },
    {
      "epoch": 1.3993681471845383,
      "grad_norm": 0.6376016139984131,
      "learning_rate": 3.937745890333623e-05,
      "loss": 0.787,
      "step": 7623
    },
    {
      "epoch": 1.3995539862479094,
      "grad_norm": 0.820344090461731,
      "learning_rate": 3.935420245471982e-05,
      "loss": 0.94,
      "step": 7624
    },
    {
      "epoch": 1.3997398253112805,
      "grad_norm": 0.6794884204864502,
      "learning_rate": 3.9330951193269547e-05,
      "loss": 0.905,
      "step": 7625
    },
    {
      "epoch": 1.3999256643746516,
      "grad_norm": 0.748024582862854,
      "learning_rate": 3.930770512097414e-05,
      "loss": 0.9723,
      "step": 7626
    },
    {
      "epoch": 1.4001115034380227,
      "grad_norm": 0.7801584005355835,
      "learning_rate": 3.9284464239821897e-05,
      "loss": 0.8886,
      "step": 7627
    },
    {
      "epoch": 1.4002973425013938,
      "grad_norm": 0.6269378066062927,
      "learning_rate": 3.926122855180066e-05,
      "loss": 0.8735,
      "step": 7628
    },
    {
      "epoch": 1.400483181564765,
      "grad_norm": 0.8104212880134583,
      "learning_rate": 3.9237998058897806e-05,
      "loss": 1.1047,
      "step": 7629
    },
    {
      "epoch": 1.400669020628136,
      "grad_norm": 0.5820780992507935,
      "learning_rate": 3.921477276310034e-05,
      "loss": 0.7459,
      "step": 7630
    },
    {
      "epoch": 1.4008548596915071,
      "grad_norm": 0.8262017369270325,
      "learning_rate": 3.9191552666394695e-05,
      "loss": 1.1253,
      "step": 7631
    },
    {
      "epoch": 1.4010406987548782,
      "grad_norm": 0.7182964086532593,
      "learning_rate": 3.916833777076699e-05,
      "loss": 1.0012,
      "step": 7632
    },
    {
      "epoch": 1.4012265378182494,
      "grad_norm": 0.6813958883285522,
      "learning_rate": 3.914512807820282e-05,
      "loss": 0.6448,
      "step": 7633
    },
    {
      "epoch": 1.4014123768816205,
      "grad_norm": 0.7764411568641663,
      "learning_rate": 3.912192359068736e-05,
      "loss": 0.9384,
      "step": 7634
    },
    {
      "epoch": 1.4015982159449916,
      "grad_norm": 0.6565015316009521,
      "learning_rate": 3.9098724310205384e-05,
      "loss": 0.9253,
      "step": 7635
    },
    {
      "epoch": 1.401784055008363,
      "grad_norm": 0.8751726746559143,
      "learning_rate": 3.907553023874111e-05,
      "loss": 0.897,
      "step": 7636
    },
    {
      "epoch": 1.4019698940717338,
      "grad_norm": 0.7254313826560974,
      "learning_rate": 3.905234137827845e-05,
      "loss": 1.0225,
      "step": 7637
    },
    {
      "epoch": 1.402155733135105,
      "grad_norm": 0.7876704335212708,
      "learning_rate": 3.902915773080071e-05,
      "loss": 1.1144,
      "step": 7638
    },
    {
      "epoch": 1.402341572198476,
      "grad_norm": 0.665797233581543,
      "learning_rate": 3.9005979298290894e-05,
      "loss": 0.9349,
      "step": 7639
    },
    {
      "epoch": 1.4025274112618473,
      "grad_norm": 0.7425183653831482,
      "learning_rate": 3.8982806082731495e-05,
      "loss": 0.905,
      "step": 7640
    },
    {
      "epoch": 1.4027132503252184,
      "grad_norm": 0.7434195280075073,
      "learning_rate": 3.895963808610459e-05,
      "loss": 1.0622,
      "step": 7641
    },
    {
      "epoch": 1.4028990893885895,
      "grad_norm": 0.7471517324447632,
      "learning_rate": 3.893647531039174e-05,
      "loss": 1.0465,
      "step": 7642
    },
    {
      "epoch": 1.4030849284519606,
      "grad_norm": 0.7446435689926147,
      "learning_rate": 3.891331775757413e-05,
      "loss": 0.9978,
      "step": 7643
    },
    {
      "epoch": 1.4032707675153318,
      "grad_norm": 0.709337055683136,
      "learning_rate": 3.889016542963252e-05,
      "loss": 0.7238,
      "step": 7644
    },
    {
      "epoch": 1.4034566065787029,
      "grad_norm": 0.6386134028434753,
      "learning_rate": 3.886701832854709e-05,
      "loss": 0.8757,
      "step": 7645
    },
    {
      "epoch": 1.403642445642074,
      "grad_norm": 0.6459241509437561,
      "learning_rate": 3.884387645629774e-05,
      "loss": 1.0497,
      "step": 7646
    },
    {
      "epoch": 1.403828284705445,
      "grad_norm": 0.7791414260864258,
      "learning_rate": 3.8820739814863794e-05,
      "loss": 0.8849,
      "step": 7647
    },
    {
      "epoch": 1.4040141237688162,
      "grad_norm": 0.7343874573707581,
      "learning_rate": 3.879760840622419e-05,
      "loss": 1.0813,
      "step": 7648
    },
    {
      "epoch": 1.4041999628321873,
      "grad_norm": 0.6905683875083923,
      "learning_rate": 3.877448223235746e-05,
      "loss": 1.0526,
      "step": 7649
    },
    {
      "epoch": 1.4043858018955584,
      "grad_norm": 0.6315404176712036,
      "learning_rate": 3.8751361295241536e-05,
      "loss": 0.7624,
      "step": 7650
    },
    {
      "epoch": 1.4045716409589295,
      "grad_norm": 0.625384509563446,
      "learning_rate": 3.87282455968541e-05,
      "loss": 0.9346,
      "step": 7651
    },
    {
      "epoch": 1.4047574800223006,
      "grad_norm": 0.7913904190063477,
      "learning_rate": 3.870513513917221e-05,
      "loss": 1.1093,
      "step": 7652
    },
    {
      "epoch": 1.404943319085672,
      "grad_norm": 0.7879133820533752,
      "learning_rate": 3.8682029924172617e-05,
      "loss": 1.0585,
      "step": 7653
    },
    {
      "epoch": 1.4051291581490428,
      "grad_norm": 0.8064400553703308,
      "learning_rate": 3.86589299538315e-05,
      "loss": 1.0688,
      "step": 7654
    },
    {
      "epoch": 1.4053149972124142,
      "grad_norm": 0.7478057146072388,
      "learning_rate": 3.863583523012467e-05,
      "loss": 1.0208,
      "step": 7655
    },
    {
      "epoch": 1.405500836275785,
      "grad_norm": 0.6603885889053345,
      "learning_rate": 3.861274575502749e-05,
      "loss": 0.934,
      "step": 7656
    },
    {
      "epoch": 1.4056866753391564,
      "grad_norm": 0.7641489505767822,
      "learning_rate": 3.858966153051484e-05,
      "loss": 0.8131,
      "step": 7657
    },
    {
      "epoch": 1.4058725144025275,
      "grad_norm": 0.6927787661552429,
      "learning_rate": 3.85665825585612e-05,
      "loss": 0.9083,
      "step": 7658
    },
    {
      "epoch": 1.4060583534658986,
      "grad_norm": 0.6479927897453308,
      "learning_rate": 3.854350884114049e-05,
      "loss": 0.9776,
      "step": 7659
    },
    {
      "epoch": 1.4062441925292697,
      "grad_norm": 0.804818868637085,
      "learning_rate": 3.852044038022633e-05,
      "loss": 0.9168,
      "step": 7660
    },
    {
      "epoch": 1.4064300315926408,
      "grad_norm": 0.7272639274597168,
      "learning_rate": 3.849737717779174e-05,
      "loss": 1.0321,
      "step": 7661
    },
    {
      "epoch": 1.406615870656012,
      "grad_norm": 0.7542232871055603,
      "learning_rate": 3.847431923580941e-05,
      "loss": 0.9362,
      "step": 7662
    },
    {
      "epoch": 1.406801709719383,
      "grad_norm": 0.8392438292503357,
      "learning_rate": 3.845126655625157e-05,
      "loss": 0.7717,
      "step": 7663
    },
    {
      "epoch": 1.4069875487827541,
      "grad_norm": 0.756179928779602,
      "learning_rate": 3.842821914108988e-05,
      "loss": 0.8892,
      "step": 7664
    },
    {
      "epoch": 1.4071733878461252,
      "grad_norm": 0.7532819509506226,
      "learning_rate": 3.840517699229572e-05,
      "loss": 1.3075,
      "step": 7665
    },
    {
      "epoch": 1.4073592269094963,
      "grad_norm": 0.8892289996147156,
      "learning_rate": 3.838214011183986e-05,
      "loss": 1.123,
      "step": 7666
    },
    {
      "epoch": 1.4075450659728674,
      "grad_norm": 0.5668398141860962,
      "learning_rate": 3.835910850169276e-05,
      "loss": 0.8134,
      "step": 7667
    },
    {
      "epoch": 1.4077309050362385,
      "grad_norm": 0.7089663147926331,
      "learning_rate": 3.833608216382431e-05,
      "loss": 1.1587,
      "step": 7668
    },
    {
      "epoch": 1.4079167440996097,
      "grad_norm": 0.6543086171150208,
      "learning_rate": 3.8313061100204015e-05,
      "loss": 0.8773,
      "step": 7669
    },
    {
      "epoch": 1.408102583162981,
      "grad_norm": 0.7265071868896484,
      "learning_rate": 3.829004531280094e-05,
      "loss": 0.8128,
      "step": 7670
    },
    {
      "epoch": 1.4082884222263519,
      "grad_norm": 0.6952806115150452,
      "learning_rate": 3.826703480358366e-05,
      "loss": 1.0041,
      "step": 7671
    },
    {
      "epoch": 1.4084742612897232,
      "grad_norm": 0.8172613382339478,
      "learning_rate": 3.824402957452033e-05,
      "loss": 1.0239,
      "step": 7672
    },
    {
      "epoch": 1.4086601003530943,
      "grad_norm": 0.6866876482963562,
      "learning_rate": 3.822102962757861e-05,
      "loss": 0.7633,
      "step": 7673
    },
    {
      "epoch": 1.4088459394164654,
      "grad_norm": 0.8350356221199036,
      "learning_rate": 3.819803496472574e-05,
      "loss": 1.117,
      "step": 7674
    },
    {
      "epoch": 1.4090317784798365,
      "grad_norm": 0.7774361371994019,
      "learning_rate": 3.8175045587928546e-05,
      "loss": 0.9895,
      "step": 7675
    },
    {
      "epoch": 1.4092176175432076,
      "grad_norm": 0.6360732913017273,
      "learning_rate": 3.8152061499153294e-05,
      "loss": 0.9731,
      "step": 7676
    },
    {
      "epoch": 1.4094034566065787,
      "grad_norm": 0.7634127140045166,
      "learning_rate": 3.8129082700365924e-05,
      "loss": 1.0542,
      "step": 7677
    },
    {
      "epoch": 1.4095892956699498,
      "grad_norm": 0.659530758857727,
      "learning_rate": 3.810610919353179e-05,
      "loss": 0.9855,
      "step": 7678
    },
    {
      "epoch": 1.409775134733321,
      "grad_norm": 0.7658821940422058,
      "learning_rate": 3.808314098061595e-05,
      "loss": 1.0321,
      "step": 7679
    },
    {
      "epoch": 1.409960973796692,
      "grad_norm": 0.6265006065368652,
      "learning_rate": 3.806017806358284e-05,
      "loss": 0.9206,
      "step": 7680
    },
    {
      "epoch": 1.4101468128600632,
      "grad_norm": 0.7096154093742371,
      "learning_rate": 3.8037220444396585e-05,
      "loss": 1.0367,
      "step": 7681
    },
    {
      "epoch": 1.4103326519234343,
      "grad_norm": 0.8711316585540771,
      "learning_rate": 3.80142681250208e-05,
      "loss": 1.2819,
      "step": 7682
    },
    {
      "epoch": 1.4105184909868054,
      "grad_norm": 0.536262571811676,
      "learning_rate": 3.799132110741861e-05,
      "loss": 0.6076,
      "step": 7683
    },
    {
      "epoch": 1.4107043300501765,
      "grad_norm": 0.6677773594856262,
      "learning_rate": 3.796837939355275e-05,
      "loss": 0.9487,
      "step": 7684
    },
    {
      "epoch": 1.4108901691135476,
      "grad_norm": 0.9733144044876099,
      "learning_rate": 3.794544298538545e-05,
      "loss": 1.2239,
      "step": 7685
    },
    {
      "epoch": 1.4110760081769187,
      "grad_norm": 0.8237594962120056,
      "learning_rate": 3.792251188487857e-05,
      "loss": 0.7716,
      "step": 7686
    },
    {
      "epoch": 1.41126184724029,
      "grad_norm": 0.6180957555770874,
      "learning_rate": 3.7899586093993376e-05,
      "loss": 0.9659,
      "step": 7687
    },
    {
      "epoch": 1.411447686303661,
      "grad_norm": 0.6425844430923462,
      "learning_rate": 3.787666561469081e-05,
      "loss": 0.9726,
      "step": 7688
    },
    {
      "epoch": 1.4116335253670322,
      "grad_norm": 0.7346746921539307,
      "learning_rate": 3.7853750448931335e-05,
      "loss": 1.0017,
      "step": 7689
    },
    {
      "epoch": 1.4118193644304033,
      "grad_norm": 0.6934552192687988,
      "learning_rate": 3.7830840598674864e-05,
      "loss": 1.0284,
      "step": 7690
    },
    {
      "epoch": 1.4120052034937745,
      "grad_norm": 0.6242930889129639,
      "learning_rate": 3.780793606588099e-05,
      "loss": 0.5696,
      "step": 7691
    },
    {
      "epoch": 1.4121910425571456,
      "grad_norm": 0.6234291791915894,
      "learning_rate": 3.778503685250873e-05,
      "loss": 0.891,
      "step": 7692
    },
    {
      "epoch": 1.4123768816205167,
      "grad_norm": 0.6656005382537842,
      "learning_rate": 3.7762142960516765e-05,
      "loss": 0.8927,
      "step": 7693
    },
    {
      "epoch": 1.4125627206838878,
      "grad_norm": 0.7903649806976318,
      "learning_rate": 3.7739254391863185e-05,
      "loss": 0.9798,
      "step": 7694
    },
    {
      "epoch": 1.4127485597472589,
      "grad_norm": 0.6335200071334839,
      "learning_rate": 3.771637114850576e-05,
      "loss": 0.8193,
      "step": 7695
    },
    {
      "epoch": 1.41293439881063,
      "grad_norm": 0.7716221809387207,
      "learning_rate": 3.7693493232401746e-05,
      "loss": 1.0676,
      "step": 7696
    },
    {
      "epoch": 1.413120237874001,
      "grad_norm": 0.8504044413566589,
      "learning_rate": 3.7670620645507884e-05,
      "loss": 1.1039,
      "step": 7697
    },
    {
      "epoch": 1.4133060769373722,
      "grad_norm": 0.6971676349639893,
      "learning_rate": 3.764775338978057e-05,
      "loss": 1.0126,
      "step": 7698
    },
    {
      "epoch": 1.4134919160007433,
      "grad_norm": 0.6730667948722839,
      "learning_rate": 3.762489146717565e-05,
      "loss": 1.0122,
      "step": 7699
    },
    {
      "epoch": 1.4136777550641144,
      "grad_norm": 0.6071676015853882,
      "learning_rate": 3.760203487964857e-05,
      "loss": 0.9624,
      "step": 7700
    },
    {
      "epoch": 1.4138635941274855,
      "grad_norm": 0.6284236311912537,
      "learning_rate": 3.7579183629154316e-05,
      "loss": 0.7958,
      "step": 7701
    },
    {
      "epoch": 1.4140494331908569,
      "grad_norm": 0.755490779876709,
      "learning_rate": 3.755633771764738e-05,
      "loss": 0.9701,
      "step": 7702
    },
    {
      "epoch": 1.4142352722542277,
      "grad_norm": 0.7259854078292847,
      "learning_rate": 3.753349714708188e-05,
      "loss": 0.863,
      "step": 7703
    },
    {
      "epoch": 1.414421111317599,
      "grad_norm": 0.6856536865234375,
      "learning_rate": 3.7510661919411336e-05,
      "loss": 0.8325,
      "step": 7704
    },
    {
      "epoch": 1.41460695038097,
      "grad_norm": 0.6991071701049805,
      "learning_rate": 3.7487832036588975e-05,
      "loss": 1.0775,
      "step": 7705
    },
    {
      "epoch": 1.4147927894443413,
      "grad_norm": 0.7490563988685608,
      "learning_rate": 3.7465007500567416e-05,
      "loss": 0.8755,
      "step": 7706
    },
    {
      "epoch": 1.4149786285077124,
      "grad_norm": 0.637843906879425,
      "learning_rate": 3.744218831329895e-05,
      "loss": 0.9338,
      "step": 7707
    },
    {
      "epoch": 1.4151644675710835,
      "grad_norm": 0.6465867757797241,
      "learning_rate": 3.7419374476735295e-05,
      "loss": 0.9167,
      "step": 7708
    },
    {
      "epoch": 1.4153503066344546,
      "grad_norm": 0.732671856880188,
      "learning_rate": 3.739656599282779e-05,
      "loss": 1.0712,
      "step": 7709
    },
    {
      "epoch": 1.4155361456978257,
      "grad_norm": 0.7095444202423096,
      "learning_rate": 3.737376286352733e-05,
      "loss": 0.5797,
      "step": 7710
    },
    {
      "epoch": 1.4157219847611968,
      "grad_norm": 0.6976941227912903,
      "learning_rate": 3.735096509078425e-05,
      "loss": 0.9708,
      "step": 7711
    },
    {
      "epoch": 1.415907823824568,
      "grad_norm": 0.7448633313179016,
      "learning_rate": 3.7328172676548576e-05,
      "loss": 0.8581,
      "step": 7712
    },
    {
      "epoch": 1.416093662887939,
      "grad_norm": 0.705802321434021,
      "learning_rate": 3.7305385622769695e-05,
      "loss": 0.8562,
      "step": 7713
    },
    {
      "epoch": 1.4162795019513101,
      "grad_norm": 0.7125275135040283,
      "learning_rate": 3.72826039313967e-05,
      "loss": 0.8646,
      "step": 7714
    },
    {
      "epoch": 1.4164653410146812,
      "grad_norm": 0.6283400654792786,
      "learning_rate": 3.7259827604378126e-05,
      "loss": 0.9295,
      "step": 7715
    },
    {
      "epoch": 1.4166511800780524,
      "grad_norm": 0.6857552528381348,
      "learning_rate": 3.723705664366211e-05,
      "loss": 1.1276,
      "step": 7716
    },
    {
      "epoch": 1.4168370191414235,
      "grad_norm": 0.7191590666770935,
      "learning_rate": 3.721429105119631e-05,
      "loss": 0.9671,
      "step": 7717
    },
    {
      "epoch": 1.4170228582047946,
      "grad_norm": 0.621395468711853,
      "learning_rate": 3.7191530828927855e-05,
      "loss": 0.8513,
      "step": 7718
    },
    {
      "epoch": 1.417208697268166,
      "grad_norm": 0.6662892699241638,
      "learning_rate": 3.716877597880356e-05,
      "loss": 0.7902,
      "step": 7719
    },
    {
      "epoch": 1.4173945363315368,
      "grad_norm": 0.7611411809921265,
      "learning_rate": 3.714602650276962e-05,
      "loss": 1.2273,
      "step": 7720
    },
    {
      "epoch": 1.417580375394908,
      "grad_norm": 0.6876758337020874,
      "learning_rate": 3.71232824027719e-05,
      "loss": 1.0438,
      "step": 7721
    },
    {
      "epoch": 1.417766214458279,
      "grad_norm": 0.7688302397727966,
      "learning_rate": 3.71005436807557e-05,
      "loss": 1.0141,
      "step": 7722
    },
    {
      "epoch": 1.4179520535216503,
      "grad_norm": 0.5873725414276123,
      "learning_rate": 3.707781033866595e-05,
      "loss": 0.6356,
      "step": 7723
    },
    {
      "epoch": 1.4181378925850214,
      "grad_norm": 0.600767970085144,
      "learning_rate": 3.7055082378447105e-05,
      "loss": 0.9036,
      "step": 7724
    },
    {
      "epoch": 1.4183237316483925,
      "grad_norm": 0.7185942530632019,
      "learning_rate": 3.703235980204307e-05,
      "loss": 1.0613,
      "step": 7725
    },
    {
      "epoch": 1.4185095707117636,
      "grad_norm": 0.7245935201644897,
      "learning_rate": 3.7009642611397435e-05,
      "loss": 0.8785,
      "step": 7726
    },
    {
      "epoch": 1.4186954097751348,
      "grad_norm": 1.004029631614685,
      "learning_rate": 3.698693080845317e-05,
      "loss": 0.9072,
      "step": 7727
    },
    {
      "epoch": 1.4188812488385059,
      "grad_norm": 0.7187013030052185,
      "learning_rate": 3.696422439515289e-05,
      "loss": 0.8739,
      "step": 7728
    },
    {
      "epoch": 1.419067087901877,
      "grad_norm": 0.7088367342948914,
      "learning_rate": 3.694152337343875e-05,
      "loss": 0.8068,
      "step": 7729
    },
    {
      "epoch": 1.419252926965248,
      "grad_norm": 0.6215872764587402,
      "learning_rate": 3.6918827745252394e-05,
      "loss": 0.9045,
      "step": 7730
    },
    {
      "epoch": 1.4194387660286192,
      "grad_norm": 0.6514158844947815,
      "learning_rate": 3.689613751253508e-05,
      "loss": 0.8785,
      "step": 7731
    },
    {
      "epoch": 1.4196246050919903,
      "grad_norm": 0.7987901568412781,
      "learning_rate": 3.6873452677227474e-05,
      "loss": 1.1281,
      "step": 7732
    },
    {
      "epoch": 1.4198104441553614,
      "grad_norm": 0.7000588178634644,
      "learning_rate": 3.685077324126992e-05,
      "loss": 1.0876,
      "step": 7733
    },
    {
      "epoch": 1.4199962832187325,
      "grad_norm": 0.8126161098480225,
      "learning_rate": 3.682809920660218e-05,
      "loss": 0.9953,
      "step": 7734
    },
    {
      "epoch": 1.4201821222821036,
      "grad_norm": 0.6847161054611206,
      "learning_rate": 3.680543057516369e-05,
      "loss": 0.7406,
      "step": 7735
    },
    {
      "epoch": 1.420367961345475,
      "grad_norm": 0.7768139243125916,
      "learning_rate": 3.678276734889326e-05,
      "loss": 0.932,
      "step": 7736
    },
    {
      "epoch": 1.4205538004088458,
      "grad_norm": 0.8145734071731567,
      "learning_rate": 3.676010952972938e-05,
      "loss": 1.161,
      "step": 7737
    },
    {
      "epoch": 1.4207396394722172,
      "grad_norm": 0.6331034898757935,
      "learning_rate": 3.673745711961005e-05,
      "loss": 0.9342,
      "step": 7738
    },
    {
      "epoch": 1.4209254785355883,
      "grad_norm": 0.7511361241340637,
      "learning_rate": 3.67148101204727e-05,
      "loss": 1.1509,
      "step": 7739
    },
    {
      "epoch": 1.4211113175989594,
      "grad_norm": 0.8429722189903259,
      "learning_rate": 3.6692168534254444e-05,
      "loss": 1.0753,
      "step": 7740
    },
    {
      "epoch": 1.4212971566623305,
      "grad_norm": 0.7080596089363098,
      "learning_rate": 3.666953236289181e-05,
      "loss": 0.8775,
      "step": 7741
    },
    {
      "epoch": 1.4214829957257016,
      "grad_norm": 0.7115619778633118,
      "learning_rate": 3.664690160832096e-05,
      "loss": 1.0145,
      "step": 7742
    },
    {
      "epoch": 1.4216688347890727,
      "grad_norm": 0.6405045986175537,
      "learning_rate": 3.6624276272477565e-05,
      "loss": 1.0048,
      "step": 7743
    },
    {
      "epoch": 1.4218546738524438,
      "grad_norm": 0.691527247428894,
      "learning_rate": 3.660165635729675e-05,
      "loss": 1.0007,
      "step": 7744
    },
    {
      "epoch": 1.422040512915815,
      "grad_norm": 0.6654130220413208,
      "learning_rate": 3.6579041864713305e-05,
      "loss": 0.8106,
      "step": 7745
    },
    {
      "epoch": 1.422226351979186,
      "grad_norm": 0.8092135787010193,
      "learning_rate": 3.6556432796661465e-05,
      "loss": 0.8508,
      "step": 7746
    },
    {
      "epoch": 1.4224121910425571,
      "grad_norm": 0.7909111976623535,
      "learning_rate": 3.6533829155075095e-05,
      "loss": 0.9582,
      "step": 7747
    },
    {
      "epoch": 1.4225980301059282,
      "grad_norm": 0.7206133604049683,
      "learning_rate": 3.651123094188744e-05,
      "loss": 0.7781,
      "step": 7748
    },
    {
      "epoch": 1.4227838691692993,
      "grad_norm": 0.665252149105072,
      "learning_rate": 3.6488638159031465e-05,
      "loss": 0.8502,
      "step": 7749
    },
    {
      "epoch": 1.4229697082326704,
      "grad_norm": 0.6059937477111816,
      "learning_rate": 3.6466050808439486e-05,
      "loss": 1.0482,
      "step": 7750
    },
    {
      "epoch": 1.4231555472960415,
      "grad_norm": 0.9111020565032959,
      "learning_rate": 3.644346889204351e-05,
      "loss": 0.9285,
      "step": 7751
    },
    {
      "epoch": 1.4233413863594127,
      "grad_norm": 0.7593806385993958,
      "learning_rate": 3.642089241177503e-05,
      "loss": 0.8047,
      "step": 7752
    },
    {
      "epoch": 1.423527225422784,
      "grad_norm": 0.5961986780166626,
      "learning_rate": 3.6398321369565006e-05,
      "loss": 0.8097,
      "step": 7753
    },
    {
      "epoch": 1.4237130644861549,
      "grad_norm": 0.6476420164108276,
      "learning_rate": 3.637575576734404e-05,
      "loss": 0.9329,
      "step": 7754
    },
    {
      "epoch": 1.4238989035495262,
      "grad_norm": 0.6606789231300354,
      "learning_rate": 3.635319560704217e-05,
      "loss": 0.9704,
      "step": 7755
    },
    {
      "epoch": 1.4240847426128973,
      "grad_norm": 0.7609673142433167,
      "learning_rate": 3.6330640890589065e-05,
      "loss": 1.0531,
      "step": 7756
    },
    {
      "epoch": 1.4242705816762684,
      "grad_norm": 0.8701117634773254,
      "learning_rate": 3.630809161991382e-05,
      "loss": 1.4154,
      "step": 7757
    },
    {
      "epoch": 1.4244564207396395,
      "grad_norm": 0.653571605682373,
      "learning_rate": 3.6285547796945164e-05,
      "loss": 0.9824,
      "step": 7758
    },
    {
      "epoch": 1.4246422598030106,
      "grad_norm": 0.7411727905273438,
      "learning_rate": 3.626300942361131e-05,
      "loss": 0.8998,
      "step": 7759
    },
    {
      "epoch": 1.4248280988663817,
      "grad_norm": 0.7065626978874207,
      "learning_rate": 3.624047650183999e-05,
      "loss": 0.875,
      "step": 7760
    },
    {
      "epoch": 1.4250139379297528,
      "grad_norm": 0.7236730456352234,
      "learning_rate": 3.6217949033558576e-05,
      "loss": 1.0839,
      "step": 7761
    },
    {
      "epoch": 1.425199776993124,
      "grad_norm": 0.8109464049339294,
      "learning_rate": 3.61954270206938e-05,
      "loss": 0.9164,
      "step": 7762
    },
    {
      "epoch": 1.425385616056495,
      "grad_norm": 0.8932875394821167,
      "learning_rate": 3.617291046517206e-05,
      "loss": 0.9789,
      "step": 7763
    },
    {
      "epoch": 1.4255714551198662,
      "grad_norm": 0.8092448115348816,
      "learning_rate": 3.615039936891922e-05,
      "loss": 1.1064,
      "step": 7764
    },
    {
      "epoch": 1.4257572941832373,
      "grad_norm": 0.726791262626648,
      "learning_rate": 3.612789373386072e-05,
      "loss": 0.8891,
      "step": 7765
    },
    {
      "epoch": 1.4259431332466084,
      "grad_norm": 0.5402032136917114,
      "learning_rate": 3.610539356192154e-05,
      "loss": 0.825,
      "step": 7766
    },
    {
      "epoch": 1.4261289723099795,
      "grad_norm": 0.5635693073272705,
      "learning_rate": 3.6082898855026114e-05,
      "loss": 0.7979,
      "step": 7767
    },
    {
      "epoch": 1.4263148113733508,
      "grad_norm": 0.7631365060806274,
      "learning_rate": 3.606040961509853e-05,
      "loss": 0.8993,
      "step": 7768
    },
    {
      "epoch": 1.4265006504367217,
      "grad_norm": 0.8184545040130615,
      "learning_rate": 3.603792584406227e-05,
      "loss": 1.0553,
      "step": 7769
    },
    {
      "epoch": 1.426686489500093,
      "grad_norm": 0.7019795775413513,
      "learning_rate": 3.601544754384049e-05,
      "loss": 0.9577,
      "step": 7770
    },
    {
      "epoch": 1.426872328563464,
      "grad_norm": 0.8127749562263489,
      "learning_rate": 3.599297471635572e-05,
      "loss": 1.1451,
      "step": 7771
    },
    {
      "epoch": 1.4270581676268352,
      "grad_norm": 0.795525312423706,
      "learning_rate": 3.597050736353017e-05,
      "loss": 0.8848,
      "step": 7772
    },
    {
      "epoch": 1.4272440066902063,
      "grad_norm": 0.6658499240875244,
      "learning_rate": 3.594804548728551e-05,
      "loss": 0.962,
      "step": 7773
    },
    {
      "epoch": 1.4274298457535775,
      "grad_norm": 0.7584779858589172,
      "learning_rate": 3.592558908954295e-05,
      "loss": 0.9173,
      "step": 7774
    },
    {
      "epoch": 1.4276156848169486,
      "grad_norm": 0.665018618106842,
      "learning_rate": 3.590313817222328e-05,
      "loss": 0.8334,
      "step": 7775
    },
    {
      "epoch": 1.4278015238803197,
      "grad_norm": 0.7542497515678406,
      "learning_rate": 3.588069273724669e-05,
      "loss": 0.9799,
      "step": 7776
    },
    {
      "epoch": 1.4279873629436908,
      "grad_norm": 0.7203447818756104,
      "learning_rate": 3.5858252786533074e-05,
      "loss": 1.0654,
      "step": 7777
    },
    {
      "epoch": 1.4281732020070619,
      "grad_norm": 0.6802030801773071,
      "learning_rate": 3.583581832200169e-05,
      "loss": 0.9758,
      "step": 7778
    },
    {
      "epoch": 1.428359041070433,
      "grad_norm": 0.7026258707046509,
      "learning_rate": 3.581338934557143e-05,
      "loss": 0.9786,
      "step": 7779
    },
    {
      "epoch": 1.428544880133804,
      "grad_norm": 0.802281379699707,
      "learning_rate": 3.5790965859160766e-05,
      "loss": 0.967,
      "step": 7780
    },
    {
      "epoch": 1.4287307191971752,
      "grad_norm": 0.7380053400993347,
      "learning_rate": 3.576854786468751e-05,
      "loss": 0.9983,
      "step": 7781
    },
    {
      "epoch": 1.4289165582605463,
      "grad_norm": 0.642246663570404,
      "learning_rate": 3.574613536406921e-05,
      "loss": 0.8811,
      "step": 7782
    },
    {
      "epoch": 1.4291023973239174,
      "grad_norm": 0.6161243915557861,
      "learning_rate": 3.57237283592228e-05,
      "loss": 0.7155,
      "step": 7783
    },
    {
      "epoch": 1.4292882363872885,
      "grad_norm": 0.7315623760223389,
      "learning_rate": 3.570132685206485e-05,
      "loss": 1.0834,
      "step": 7784
    },
    {
      "epoch": 1.4294740754506599,
      "grad_norm": 0.634521484375,
      "learning_rate": 3.567893084451135e-05,
      "loss": 0.7328,
      "step": 7785
    },
    {
      "epoch": 1.4296599145140307,
      "grad_norm": 0.7404763102531433,
      "learning_rate": 3.5656540338477906e-05,
      "loss": 0.9951,
      "step": 7786
    },
    {
      "epoch": 1.429845753577402,
      "grad_norm": 0.7833670973777771,
      "learning_rate": 3.5634155335879635e-05,
      "loss": 0.9775,
      "step": 7787
    },
    {
      "epoch": 1.430031592640773,
      "grad_norm": 0.7117665410041809,
      "learning_rate": 3.561177583863121e-05,
      "loss": 0.8914,
      "step": 7788
    },
    {
      "epoch": 1.4302174317041443,
      "grad_norm": 0.7110545039176941,
      "learning_rate": 3.558940184864672e-05,
      "loss": 1.0174,
      "step": 7789
    },
    {
      "epoch": 1.4304032707675154,
      "grad_norm": 0.7871912121772766,
      "learning_rate": 3.556703336783991e-05,
      "loss": 0.8585,
      "step": 7790
    },
    {
      "epoch": 1.4305891098308865,
      "grad_norm": 0.5926689505577087,
      "learning_rate": 3.5544670398124015e-05,
      "loss": 0.6127,
      "step": 7791
    },
    {
      "epoch": 1.4307749488942576,
      "grad_norm": 0.7476575374603271,
      "learning_rate": 3.552231294141174e-05,
      "loss": 1.1246,
      "step": 7792
    },
    {
      "epoch": 1.4309607879576287,
      "grad_norm": 0.9882261157035828,
      "learning_rate": 3.549996099961539e-05,
      "loss": 0.9569,
      "step": 7793
    },
    {
      "epoch": 1.4311466270209998,
      "grad_norm": 0.7017543911933899,
      "learning_rate": 3.547761457464681e-05,
      "loss": 0.6167,
      "step": 7794
    },
    {
      "epoch": 1.431332466084371,
      "grad_norm": 0.6456220149993896,
      "learning_rate": 3.545527366841728e-05,
      "loss": 1.0104,
      "step": 7795
    },
    {
      "epoch": 1.431518305147742,
      "grad_norm": 0.8840920329093933,
      "learning_rate": 3.5432938282837726e-05,
      "loss": 1.032,
      "step": 7796
    },
    {
      "epoch": 1.4317041442111131,
      "grad_norm": 0.5940844416618347,
      "learning_rate": 3.541060841981847e-05,
      "loss": 0.8061,
      "step": 7797
    },
    {
      "epoch": 1.4318899832744842,
      "grad_norm": 0.7961482405662537,
      "learning_rate": 3.5388284081269495e-05,
      "loss": 1.0628,
      "step": 7798
    },
    {
      "epoch": 1.4320758223378554,
      "grad_norm": 0.6357424259185791,
      "learning_rate": 3.536596526910021e-05,
      "loss": 0.7603,
      "step": 7799
    },
    {
      "epoch": 1.4322616614012265,
      "grad_norm": 0.6908395886421204,
      "learning_rate": 3.534365198521961e-05,
      "loss": 0.9484,
      "step": 7800
    },
    {
      "epoch": 1.4324475004645976,
      "grad_norm": 0.7572526931762695,
      "learning_rate": 3.5321344231536216e-05,
      "loss": 0.9961,
      "step": 7801
    },
    {
      "epoch": 1.432633339527969,
      "grad_norm": 0.7261570692062378,
      "learning_rate": 3.5299042009958014e-05,
      "loss": 0.833,
      "step": 7802
    },
    {
      "epoch": 1.4328191785913398,
      "grad_norm": 0.6793763041496277,
      "learning_rate": 3.527674532239258e-05,
      "loss": 0.8265,
      "step": 7803
    },
    {
      "epoch": 1.433005017654711,
      "grad_norm": 0.6620837450027466,
      "learning_rate": 3.525445417074701e-05,
      "loss": 0.7635,
      "step": 7804
    },
    {
      "epoch": 1.4331908567180822,
      "grad_norm": 0.7269986271858215,
      "learning_rate": 3.523216855692795e-05,
      "loss": 0.9392,
      "step": 7805
    },
    {
      "epoch": 1.4333766957814533,
      "grad_norm": 0.6936744451522827,
      "learning_rate": 3.5209888482841445e-05,
      "loss": 0.7935,
      "step": 7806
    },
    {
      "epoch": 1.4335625348448244,
      "grad_norm": 0.6548713445663452,
      "learning_rate": 3.518761395039322e-05,
      "loss": 0.9881,
      "step": 7807
    },
    {
      "epoch": 1.4337483739081955,
      "grad_norm": 0.7093619704246521,
      "learning_rate": 3.516534496148848e-05,
      "loss": 1.0224,
      "step": 7808
    },
    {
      "epoch": 1.4339342129715666,
      "grad_norm": 0.7175012826919556,
      "learning_rate": 3.514308151803187e-05,
      "loss": 0.8043,
      "step": 7809
    },
    {
      "epoch": 1.4341200520349378,
      "grad_norm": 0.8724030256271362,
      "learning_rate": 3.5120823621927715e-05,
      "loss": 1.0353,
      "step": 7810
    },
    {
      "epoch": 1.4343058910983089,
      "grad_norm": 0.6106569766998291,
      "learning_rate": 3.5098571275079706e-05,
      "loss": 0.7952,
      "step": 7811
    },
    {
      "epoch": 1.43449173016168,
      "grad_norm": 0.6657244563102722,
      "learning_rate": 3.50763244793912e-05,
      "loss": 0.9032,
      "step": 7812
    },
    {
      "epoch": 1.434677569225051,
      "grad_norm": 0.7090873122215271,
      "learning_rate": 3.5054083236764934e-05,
      "loss": 0.952,
      "step": 7813
    },
    {
      "epoch": 1.4348634082884222,
      "grad_norm": 0.7857263684272766,
      "learning_rate": 3.5031847549103305e-05,
      "loss": 1.0373,
      "step": 7814
    },
    {
      "epoch": 1.4350492473517933,
      "grad_norm": 0.7097312808036804,
      "learning_rate": 3.500961741830821e-05,
      "loss": 0.9285,
      "step": 7815
    },
    {
      "epoch": 1.4352350864151644,
      "grad_norm": 0.7621490955352783,
      "learning_rate": 3.498739284628095e-05,
      "loss": 1.0428,
      "step": 7816
    },
    {
      "epoch": 1.4354209254785357,
      "grad_norm": 0.6671808958053589,
      "learning_rate": 3.4965173834922505e-05,
      "loss": 0.8277,
      "step": 7817
    },
    {
      "epoch": 1.4356067645419066,
      "grad_norm": 0.889748215675354,
      "learning_rate": 3.4942960386133296e-05,
      "loss": 0.9986,
      "step": 7818
    },
    {
      "epoch": 1.435792603605278,
      "grad_norm": 0.7396422028541565,
      "learning_rate": 3.492075250181333e-05,
      "loss": 0.9544,
      "step": 7819
    },
    {
      "epoch": 1.4359784426686488,
      "grad_norm": 0.7094883918762207,
      "learning_rate": 3.489855018386201e-05,
      "loss": 1.0928,
      "step": 7820
    },
    {
      "epoch": 1.4361642817320202,
      "grad_norm": 0.7435691952705383,
      "learning_rate": 3.487635343417841e-05,
      "loss": 0.9684,
      "step": 7821
    },
    {
      "epoch": 1.4363501207953913,
      "grad_norm": 0.6778486967086792,
      "learning_rate": 3.485416225466107e-05,
      "loss": 0.9725,
      "step": 7822
    },
    {
      "epoch": 1.4365359598587624,
      "grad_norm": 0.635127067565918,
      "learning_rate": 3.483197664720801e-05,
      "loss": 0.7622,
      "step": 7823
    },
    {
      "epoch": 1.4367217989221335,
      "grad_norm": 0.7089655995368958,
      "learning_rate": 3.4809796613716864e-05,
      "loss": 1.0226,
      "step": 7824
    },
    {
      "epoch": 1.4369076379855046,
      "grad_norm": 0.8211172819137573,
      "learning_rate": 3.478762215608468e-05,
      "loss": 1.0076,
      "step": 7825
    },
    {
      "epoch": 1.4370934770488757,
      "grad_norm": 0.6057694554328918,
      "learning_rate": 3.4765453276208136e-05,
      "loss": 0.8723,
      "step": 7826
    },
    {
      "epoch": 1.4372793161122468,
      "grad_norm": 0.609369695186615,
      "learning_rate": 3.474328997598334e-05,
      "loss": 0.9602,
      "step": 7827
    },
    {
      "epoch": 1.437465155175618,
      "grad_norm": 0.7126015424728394,
      "learning_rate": 3.4721132257305986e-05,
      "loss": 1.2145,
      "step": 7828
    },
    {
      "epoch": 1.437650994238989,
      "grad_norm": 0.7071675658226013,
      "learning_rate": 3.469898012207131e-05,
      "loss": 1.0776,
      "step": 7829
    },
    {
      "epoch": 1.4378368333023601,
      "grad_norm": 0.6722445487976074,
      "learning_rate": 3.467683357217397e-05,
      "loss": 1.0917,
      "step": 7830
    },
    {
      "epoch": 1.4380226723657312,
      "grad_norm": 0.6246117353439331,
      "learning_rate": 3.465469260950823e-05,
      "loss": 1.1885,
      "step": 7831
    },
    {
      "epoch": 1.4382085114291023,
      "grad_norm": 0.6887637376785278,
      "learning_rate": 3.4632557235967875e-05,
      "loss": 0.9325,
      "step": 7832
    },
    {
      "epoch": 1.4383943504924734,
      "grad_norm": 0.6999205350875854,
      "learning_rate": 3.461042745344619e-05,
      "loss": 0.9148,
      "step": 7833
    },
    {
      "epoch": 1.4385801895558448,
      "grad_norm": 0.6207687854766846,
      "learning_rate": 3.4588303263835954e-05,
      "loss": 0.7277,
      "step": 7834
    },
    {
      "epoch": 1.4387660286192157,
      "grad_norm": 0.6546527147293091,
      "learning_rate": 3.456618466902951e-05,
      "loss": 0.8376,
      "step": 7835
    },
    {
      "epoch": 1.438951867682587,
      "grad_norm": 0.7431588768959045,
      "learning_rate": 3.4544071670918745e-05,
      "loss": 0.967,
      "step": 7836
    },
    {
      "epoch": 1.4391377067459579,
      "grad_norm": 0.8065693378448486,
      "learning_rate": 3.452196427139496e-05,
      "loss": 0.9605,
      "step": 7837
    },
    {
      "epoch": 1.4393235458093292,
      "grad_norm": 2.3515830039978027,
      "learning_rate": 3.449986247234912e-05,
      "loss": 1.3481,
      "step": 7838
    },
    {
      "epoch": 1.4395093848727003,
      "grad_norm": 0.7496092319488525,
      "learning_rate": 3.447776627567157e-05,
      "loss": 1.0978,
      "step": 7839
    },
    {
      "epoch": 1.4396952239360714,
      "grad_norm": 0.7271265983581543,
      "learning_rate": 3.445567568325233e-05,
      "loss": 0.9726,
      "step": 7840
    },
    {
      "epoch": 1.4398810629994425,
      "grad_norm": 0.5585505962371826,
      "learning_rate": 3.4433590696980764e-05,
      "loss": 0.6408,
      "step": 7841
    },
    {
      "epoch": 1.4400669020628136,
      "grad_norm": 0.9318434000015259,
      "learning_rate": 3.44115113187459e-05,
      "loss": 1.1037,
      "step": 7842
    },
    {
      "epoch": 1.4402527411261847,
      "grad_norm": 0.6693130731582642,
      "learning_rate": 3.4389437550436264e-05,
      "loss": 0.9345,
      "step": 7843
    },
    {
      "epoch": 1.4404385801895558,
      "grad_norm": 0.6681878566741943,
      "learning_rate": 3.43673693939398e-05,
      "loss": 1.0068,
      "step": 7844
    },
    {
      "epoch": 1.440624419252927,
      "grad_norm": 2.8995208740234375,
      "learning_rate": 3.434530685114411e-05,
      "loss": 1.2537,
      "step": 7845
    },
    {
      "epoch": 1.440810258316298,
      "grad_norm": 0.6880825161933899,
      "learning_rate": 3.43232499239362e-05,
      "loss": 1.0368,
      "step": 7846
    },
    {
      "epoch": 1.4409960973796692,
      "grad_norm": 0.8894131183624268,
      "learning_rate": 3.430119861420267e-05,
      "loss": 1.0654,
      "step": 7847
    },
    {
      "epoch": 1.4411819364430403,
      "grad_norm": 0.7213008403778076,
      "learning_rate": 3.427915292382961e-05,
      "loss": 0.9789,
      "step": 7848
    },
    {
      "epoch": 1.4413677755064114,
      "grad_norm": 1.57517671585083,
      "learning_rate": 3.425711285470264e-05,
      "loss": 1.6136,
      "step": 7849
    },
    {
      "epoch": 1.4415536145697825,
      "grad_norm": 0.7459681034088135,
      "learning_rate": 3.4235078408706936e-05,
      "loss": 1.1973,
      "step": 7850
    },
    {
      "epoch": 1.4417394536331538,
      "grad_norm": 0.646812379360199,
      "learning_rate": 3.421304958772708e-05,
      "loss": 0.932,
      "step": 7851
    },
    {
      "epoch": 1.4419252926965247,
      "grad_norm": 0.8054854273796082,
      "learning_rate": 3.419102639364731e-05,
      "loss": 0.9733,
      "step": 7852
    },
    {
      "epoch": 1.442111131759896,
      "grad_norm": 0.705011248588562,
      "learning_rate": 3.416900882835125e-05,
      "loss": 1.0736,
      "step": 7853
    },
    {
      "epoch": 1.4422969708232671,
      "grad_norm": 0.6439553499221802,
      "learning_rate": 3.4146996893722175e-05,
      "loss": 0.987,
      "step": 7854
    },
    {
      "epoch": 1.4424828098866382,
      "grad_norm": 0.6906138062477112,
      "learning_rate": 3.4124990591642744e-05,
      "loss": 0.8643,
      "step": 7855
    },
    {
      "epoch": 1.4426686489500093,
      "grad_norm": 0.7618701457977295,
      "learning_rate": 3.410298992399524e-05,
      "loss": 1.0197,
      "step": 7856
    },
    {
      "epoch": 1.4428544880133805,
      "grad_norm": 0.6181477904319763,
      "learning_rate": 3.408099489266148e-05,
      "loss": 0.8519,
      "step": 7857
    },
    {
      "epoch": 1.4430403270767516,
      "grad_norm": 0.6817232370376587,
      "learning_rate": 3.405900549952266e-05,
      "loss": 1.0308,
      "step": 7858
    },
    {
      "epoch": 1.4432261661401227,
      "grad_norm": 0.7186017632484436,
      "learning_rate": 3.403702174645964e-05,
      "loss": 1.0243,
      "step": 7859
    },
    {
      "epoch": 1.4434120052034938,
      "grad_norm": 0.6884559988975525,
      "learning_rate": 3.401504363535268e-05,
      "loss": 0.9924,
      "step": 7860
    },
    {
      "epoch": 1.4435978442668649,
      "grad_norm": 0.7954242825508118,
      "learning_rate": 3.3993071168081666e-05,
      "loss": 1.1681,
      "step": 7861
    },
    {
      "epoch": 1.443783683330236,
      "grad_norm": 0.5953904986381531,
      "learning_rate": 3.397110434652593e-05,
      "loss": 0.8374,
      "step": 7862
    },
    {
      "epoch": 1.443969522393607,
      "grad_norm": 0.6842502355575562,
      "learning_rate": 3.3949143172564336e-05,
      "loss": 0.8099,
      "step": 7863
    },
    {
      "epoch": 1.4441553614569782,
      "grad_norm": 0.7532383799552917,
      "learning_rate": 3.3927187648075324e-05,
      "loss": 1.1474,
      "step": 7864
    },
    {
      "epoch": 1.4443412005203493,
      "grad_norm": 0.7515235543251038,
      "learning_rate": 3.390523777493672e-05,
      "loss": 1.0207,
      "step": 7865
    },
    {
      "epoch": 1.4445270395837204,
      "grad_norm": 0.6641031503677368,
      "learning_rate": 3.3883293555026005e-05,
      "loss": 0.8668,
      "step": 7866
    },
    {
      "epoch": 1.4447128786470915,
      "grad_norm": 0.951924741268158,
      "learning_rate": 3.386135499022006e-05,
      "loss": 1.0705,
      "step": 7867
    },
    {
      "epoch": 1.4448987177104629,
      "grad_norm": 0.6455573439598083,
      "learning_rate": 3.38394220823954e-05,
      "loss": 0.946,
      "step": 7868
    },
    {
      "epoch": 1.4450845567738337,
      "grad_norm": 0.6811406016349792,
      "learning_rate": 3.3817494833427934e-05,
      "loss": 0.9002,
      "step": 7869
    },
    {
      "epoch": 1.445270395837205,
      "grad_norm": 0.6838234066963196,
      "learning_rate": 3.3795573245193167e-05,
      "loss": 0.8727,
      "step": 7870
    },
    {
      "epoch": 1.4454562349005762,
      "grad_norm": 0.9390320777893066,
      "learning_rate": 3.3773657319566145e-05,
      "loss": 1.2102,
      "step": 7871
    },
    {
      "epoch": 1.4456420739639473,
      "grad_norm": 0.6732348799705505,
      "learning_rate": 3.375174705842131e-05,
      "loss": 0.873,
      "step": 7872
    },
    {
      "epoch": 1.4458279130273184,
      "grad_norm": 0.5763305425643921,
      "learning_rate": 3.3729842463632766e-05,
      "loss": 0.9149,
      "step": 7873
    },
    {
      "epoch": 1.4460137520906895,
      "grad_norm": 0.7089604139328003,
      "learning_rate": 3.3707943537073994e-05,
      "loss": 0.929,
      "step": 7874
    },
    {
      "epoch": 1.4461995911540606,
      "grad_norm": 0.6530745625495911,
      "learning_rate": 3.368605028061808e-05,
      "loss": 1.0375,
      "step": 7875
    },
    {
      "epoch": 1.4463854302174317,
      "grad_norm": 0.726687490940094,
      "learning_rate": 3.366416269613761e-05,
      "loss": 0.8975,
      "step": 7876
    },
    {
      "epoch": 1.4465712692808028,
      "grad_norm": 0.7921562194824219,
      "learning_rate": 3.364228078550468e-05,
      "loss": 1.1033,
      "step": 7877
    },
    {
      "epoch": 1.446757108344174,
      "grad_norm": 1.9331955909729004,
      "learning_rate": 3.362040455059092e-05,
      "loss": 1.4329,
      "step": 7878
    },
    {
      "epoch": 1.446942947407545,
      "grad_norm": 0.7026411294937134,
      "learning_rate": 3.359853399326739e-05,
      "loss": 1.1135,
      "step": 7879
    },
    {
      "epoch": 1.4471287864709161,
      "grad_norm": 0.7438810467720032,
      "learning_rate": 3.35766691154048e-05,
      "loss": 0.9522,
      "step": 7880
    },
    {
      "epoch": 1.4473146255342872,
      "grad_norm": 0.7092804908752441,
      "learning_rate": 3.355480991887322e-05,
      "loss": 1.0578,
      "step": 7881
    },
    {
      "epoch": 1.4475004645976584,
      "grad_norm": 0.6761428713798523,
      "learning_rate": 3.353295640554239e-05,
      "loss": 0.9218,
      "step": 7882
    },
    {
      "epoch": 1.4476863036610297,
      "grad_norm": 0.6848258376121521,
      "learning_rate": 3.351110857728143e-05,
      "loss": 0.9947,
      "step": 7883
    },
    {
      "epoch": 1.4478721427244006,
      "grad_norm": 0.8061132431030273,
      "learning_rate": 3.348926643595904e-05,
      "loss": 1.1676,
      "step": 7884
    },
    {
      "epoch": 1.448057981787772,
      "grad_norm": 0.7291356325149536,
      "learning_rate": 3.346742998344348e-05,
      "loss": 1.153,
      "step": 7885
    },
    {
      "epoch": 1.4482438208511428,
      "grad_norm": 0.8108109831809998,
      "learning_rate": 3.34455992216024e-05,
      "loss": 1.1443,
      "step": 7886
    },
    {
      "epoch": 1.4484296599145141,
      "grad_norm": 1.0187431573867798,
      "learning_rate": 3.342377415230309e-05,
      "loss": 0.9213,
      "step": 7887
    },
    {
      "epoch": 1.4486154989778852,
      "grad_norm": 0.819898247718811,
      "learning_rate": 3.340195477741224e-05,
      "loss": 1.1785,
      "step": 7888
    },
    {
      "epoch": 1.4488013380412563,
      "grad_norm": 0.6514377593994141,
      "learning_rate": 3.338014109879612e-05,
      "loss": 0.916,
      "step": 7889
    },
    {
      "epoch": 1.4489871771046274,
      "grad_norm": 0.6867848038673401,
      "learning_rate": 3.335833311832056e-05,
      "loss": 0.9396,
      "step": 7890
    },
    {
      "epoch": 1.4491730161679985,
      "grad_norm": 0.6328677535057068,
      "learning_rate": 3.333653083785077e-05,
      "loss": 0.8447,
      "step": 7891
    },
    {
      "epoch": 1.4493588552313696,
      "grad_norm": 0.6908021569252014,
      "learning_rate": 3.3314734259251565e-05,
      "loss": 1.023,
      "step": 7892
    },
    {
      "epoch": 1.4495446942947408,
      "grad_norm": 0.6348336338996887,
      "learning_rate": 3.329294338438728e-05,
      "loss": 0.9131,
      "step": 7893
    },
    {
      "epoch": 1.4497305333581119,
      "grad_norm": 1.490304946899414,
      "learning_rate": 3.3271158215121736e-05,
      "loss": 1.4871,
      "step": 7894
    },
    {
      "epoch": 1.449916372421483,
      "grad_norm": 0.7178076505661011,
      "learning_rate": 3.3249378753318225e-05,
      "loss": 0.9315,
      "step": 7895
    },
    {
      "epoch": 1.450102211484854,
      "grad_norm": 0.6275186538696289,
      "learning_rate": 3.3227605000839654e-05,
      "loss": 1.0205,
      "step": 7896
    },
    {
      "epoch": 1.4502880505482252,
      "grad_norm": 0.5495182275772095,
      "learning_rate": 3.3205836959548296e-05,
      "loss": 0.6146,
      "step": 7897
    },
    {
      "epoch": 1.4504738896115963,
      "grad_norm": 0.844357967376709,
      "learning_rate": 3.3184074631306085e-05,
      "loss": 0.6343,
      "step": 7898
    },
    {
      "epoch": 1.4506597286749674,
      "grad_norm": 0.7961918115615845,
      "learning_rate": 3.316231801797439e-05,
      "loss": 0.8678,
      "step": 7899
    },
    {
      "epoch": 1.4508455677383387,
      "grad_norm": 0.7114526629447937,
      "learning_rate": 3.314056712141407e-05,
      "loss": 1.0124,
      "step": 7900
    },
    {
      "epoch": 1.4510314068017096,
      "grad_norm": 0.7751607298851013,
      "learning_rate": 3.311882194348558e-05,
      "loss": 0.984,
      "step": 7901
    },
    {
      "epoch": 1.451217245865081,
      "grad_norm": 0.6312512159347534,
      "learning_rate": 3.309708248604876e-05,
      "loss": 0.8796,
      "step": 7902
    },
    {
      "epoch": 1.4514030849284518,
      "grad_norm": 0.6945473551750183,
      "learning_rate": 3.30753487509631e-05,
      "loss": 0.9218,
      "step": 7903
    },
    {
      "epoch": 1.4515889239918232,
      "grad_norm": 0.7502307295799255,
      "learning_rate": 3.3053620740087475e-05,
      "loss": 1.1365,
      "step": 7904
    },
    {
      "epoch": 1.4517747630551943,
      "grad_norm": 0.6966648697853088,
      "learning_rate": 3.3031898455280364e-05,
      "loss": 0.9969,
      "step": 7905
    },
    {
      "epoch": 1.4519606021185654,
      "grad_norm": 0.7188398838043213,
      "learning_rate": 3.3010181898399715e-05,
      "loss": 1.0415,
      "step": 7906
    },
    {
      "epoch": 1.4521464411819365,
      "grad_norm": 1.1256226301193237,
      "learning_rate": 3.298847107130298e-05,
      "loss": 1.0763,
      "step": 7907
    },
    {
      "epoch": 1.4523322802453076,
      "grad_norm": 0.7524394392967224,
      "learning_rate": 3.296676597584719e-05,
      "loss": 0.8179,
      "step": 7908
    },
    {
      "epoch": 1.4525181193086787,
      "grad_norm": 0.6933947801589966,
      "learning_rate": 3.294506661388875e-05,
      "loss": 1.0581,
      "step": 7909
    },
    {
      "epoch": 1.4527039583720498,
      "grad_norm": 0.7244120240211487,
      "learning_rate": 3.292337298728372e-05,
      "loss": 0.84,
      "step": 7910
    },
    {
      "epoch": 1.452889797435421,
      "grad_norm": 0.6906845569610596,
      "learning_rate": 3.2901685097887535e-05,
      "loss": 0.9544,
      "step": 7911
    },
    {
      "epoch": 1.453075636498792,
      "grad_norm": 0.7629197239875793,
      "learning_rate": 3.288000294755524e-05,
      "loss": 0.9505,
      "step": 7912
    },
    {
      "epoch": 1.4532614755621631,
      "grad_norm": 0.6772804260253906,
      "learning_rate": 3.285832653814139e-05,
      "loss": 0.9279,
      "step": 7913
    },
    {
      "epoch": 1.4534473146255342,
      "grad_norm": 0.691500723361969,
      "learning_rate": 3.2836655871499955e-05,
      "loss": 0.6032,
      "step": 7914
    },
    {
      "epoch": 1.4536331536889053,
      "grad_norm": 0.7557622194290161,
      "learning_rate": 3.281499094948454e-05,
      "loss": 0.8728,
      "step": 7915
    },
    {
      "epoch": 1.4538189927522764,
      "grad_norm": 0.7844099998474121,
      "learning_rate": 3.2793331773948113e-05,
      "loss": 1.0298,
      "step": 7916
    },
    {
      "epoch": 1.4540048318156478,
      "grad_norm": 1.0272523164749146,
      "learning_rate": 3.277167834674331e-05,
      "loss": 1.2326,
      "step": 7917
    },
    {
      "epoch": 1.4541906708790187,
      "grad_norm": 0.6383036971092224,
      "learning_rate": 3.275003066972212e-05,
      "loss": 0.838,
      "step": 7918
    },
    {
      "epoch": 1.45437650994239,
      "grad_norm": 0.7620444893836975,
      "learning_rate": 3.272838874473616e-05,
      "loss": 0.9757,
      "step": 7919
    },
    {
      "epoch": 1.454562349005761,
      "grad_norm": 0.7140588760375977,
      "learning_rate": 3.27067525736365e-05,
      "loss": 1.0327,
      "step": 7920
    },
    {
      "epoch": 1.4547481880691322,
      "grad_norm": 0.6811936497688293,
      "learning_rate": 3.268512215827373e-05,
      "loss": 0.9918,
      "step": 7921
    },
    {
      "epoch": 1.4549340271325033,
      "grad_norm": 0.7551881074905396,
      "learning_rate": 3.266349750049799e-05,
      "loss": 1.1241,
      "step": 7922
    },
    {
      "epoch": 1.4551198661958744,
      "grad_norm": 0.6850900650024414,
      "learning_rate": 3.2641878602158795e-05,
      "loss": 1.0948,
      "step": 7923
    },
    {
      "epoch": 1.4553057052592455,
      "grad_norm": 0.6953625679016113,
      "learning_rate": 3.262026546510535e-05,
      "loss": 1.1129,
      "step": 7924
    },
    {
      "epoch": 1.4554915443226166,
      "grad_norm": 0.5916890501976013,
      "learning_rate": 3.259865809118618e-05,
      "loss": 0.9081,
      "step": 7925
    },
    {
      "epoch": 1.4556773833859877,
      "grad_norm": 0.6639631390571594,
      "learning_rate": 3.257705648224947e-05,
      "loss": 1.0106,
      "step": 7926
    },
    {
      "epoch": 1.4558632224493588,
      "grad_norm": 0.853246808052063,
      "learning_rate": 3.255546064014287e-05,
      "loss": 1.0606,
      "step": 7927
    },
    {
      "epoch": 1.45604906151273,
      "grad_norm": 0.6534708142280579,
      "learning_rate": 3.253387056671344e-05,
      "loss": 0.7146,
      "step": 7928
    },
    {
      "epoch": 1.456234900576101,
      "grad_norm": 0.807653546333313,
      "learning_rate": 3.251228626380792e-05,
      "loss": 0.8833,
      "step": 7929
    },
    {
      "epoch": 1.4564207396394722,
      "grad_norm": 0.7415690422058105,
      "learning_rate": 3.249070773327239e-05,
      "loss": 1.0653,
      "step": 7930
    },
    {
      "epoch": 1.4566065787028433,
      "grad_norm": 0.6846359372138977,
      "learning_rate": 3.246913497695255e-05,
      "loss": 1.0067,
      "step": 7931
    },
    {
      "epoch": 1.4567924177662144,
      "grad_norm": 0.7008286118507385,
      "learning_rate": 3.2447567996693526e-05,
      "loss": 1.1567,
      "step": 7932
    },
    {
      "epoch": 1.4569782568295855,
      "grad_norm": 0.6331776976585388,
      "learning_rate": 3.242600679434001e-05,
      "loss": 0.7786,
      "step": 7933
    },
    {
      "epoch": 1.4571640958929568,
      "grad_norm": 0.8562239408493042,
      "learning_rate": 3.240445137173619e-05,
      "loss": 0.8782,
      "step": 7934
    },
    {
      "epoch": 1.4573499349563277,
      "grad_norm": 0.6675928235054016,
      "learning_rate": 3.2382901730725756e-05,
      "loss": 0.968,
      "step": 7935
    },
    {
      "epoch": 1.457535774019699,
      "grad_norm": 0.7753133773803711,
      "learning_rate": 3.2361357873151846e-05,
      "loss": 0.8592,
      "step": 7936
    },
    {
      "epoch": 1.4577216130830701,
      "grad_norm": 0.7975282669067383,
      "learning_rate": 3.233981980085719e-05,
      "loss": 1.0158,
      "step": 7937
    },
    {
      "epoch": 1.4579074521464412,
      "grad_norm": 0.8110864758491516,
      "learning_rate": 3.231828751568401e-05,
      "loss": 1.0904,
      "step": 7938
    },
    {
      "epoch": 1.4580932912098123,
      "grad_norm": 0.5796823501586914,
      "learning_rate": 3.2296761019473944e-05,
      "loss": 0.8856,
      "step": 7939
    },
    {
      "epoch": 1.4582791302731835,
      "grad_norm": 0.7160946130752563,
      "learning_rate": 3.2275240314068224e-05,
      "loss": 0.9291,
      "step": 7940
    },
    {
      "epoch": 1.4584649693365546,
      "grad_norm": 0.6326502561569214,
      "learning_rate": 3.225372540130762e-05,
      "loss": 1.0077,
      "step": 7941
    },
    {
      "epoch": 1.4586508083999257,
      "grad_norm": 0.7014022469520569,
      "learning_rate": 3.2232216283032254e-05,
      "loss": 0.9311,
      "step": 7942
    },
    {
      "epoch": 1.4588366474632968,
      "grad_norm": 0.8314329981803894,
      "learning_rate": 3.221071296108195e-05,
      "loss": 0.9909,
      "step": 7943
    },
    {
      "epoch": 1.4590224865266679,
      "grad_norm": 0.7300158143043518,
      "learning_rate": 3.218921543729583e-05,
      "loss": 0.8944,
      "step": 7944
    },
    {
      "epoch": 1.459208325590039,
      "grad_norm": 0.6036564111709595,
      "learning_rate": 3.21677237135127e-05,
      "loss": 0.9688,
      "step": 7945
    },
    {
      "epoch": 1.45939416465341,
      "grad_norm": 0.8187325596809387,
      "learning_rate": 3.214623779157074e-05,
      "loss": 0.9268,
      "step": 7946
    },
    {
      "epoch": 1.4595800037167812,
      "grad_norm": 0.7486228346824646,
      "learning_rate": 3.212475767330771e-05,
      "loss": 0.9888,
      "step": 7947
    },
    {
      "epoch": 1.4597658427801523,
      "grad_norm": 0.6301973462104797,
      "learning_rate": 3.210328336056089e-05,
      "loss": 0.64,
      "step": 7948
    },
    {
      "epoch": 1.4599516818435236,
      "grad_norm": 0.6302767395973206,
      "learning_rate": 3.208181485516696e-05,
      "loss": 0.9719,
      "step": 7949
    },
    {
      "epoch": 1.4601375209068945,
      "grad_norm": 0.6822143197059631,
      "learning_rate": 3.206035215896218e-05,
      "loss": 1.0381,
      "step": 7950
    },
    {
      "epoch": 1.4603233599702659,
      "grad_norm": 0.7191839814186096,
      "learning_rate": 3.2038895273782335e-05,
      "loss": 1.1134,
      "step": 7951
    },
    {
      "epoch": 1.4605091990336367,
      "grad_norm": 0.8182570934295654,
      "learning_rate": 3.201744420146269e-05,
      "loss": 0.8333,
      "step": 7952
    },
    {
      "epoch": 1.460695038097008,
      "grad_norm": 0.8387598395347595,
      "learning_rate": 3.1995998943837935e-05,
      "loss": 1.0758,
      "step": 7953
    },
    {
      "epoch": 1.4608808771603792,
      "grad_norm": 0.6082673668861389,
      "learning_rate": 3.1974559502742364e-05,
      "loss": 0.7804,
      "step": 7954
    },
    {
      "epoch": 1.4610667162237503,
      "grad_norm": 0.6701902151107788,
      "learning_rate": 3.195312588000977e-05,
      "loss": 1.039,
      "step": 7955
    },
    {
      "epoch": 1.4612525552871214,
      "grad_norm": 0.7884292602539062,
      "learning_rate": 3.193169807747336e-05,
      "loss": 1.0115,
      "step": 7956
    },
    {
      "epoch": 1.4614383943504925,
      "grad_norm": 0.725053071975708,
      "learning_rate": 3.1910276096965966e-05,
      "loss": 1.0554,
      "step": 7957
    },
    {
      "epoch": 1.4616242334138636,
      "grad_norm": 0.8168944120407104,
      "learning_rate": 3.188885994031978e-05,
      "loss": 0.9558,
      "step": 7958
    },
    {
      "epoch": 1.4618100724772347,
      "grad_norm": 0.726453959941864,
      "learning_rate": 3.186744960936665e-05,
      "loss": 0.9404,
      "step": 7959
    },
    {
      "epoch": 1.4619959115406058,
      "grad_norm": 0.7123615741729736,
      "learning_rate": 3.184604510593777e-05,
      "loss": 0.8578,
      "step": 7960
    },
    {
      "epoch": 1.462181750603977,
      "grad_norm": 0.6957084536552429,
      "learning_rate": 3.1824646431863956e-05,
      "loss": 0.9583,
      "step": 7961
    },
    {
      "epoch": 1.462367589667348,
      "grad_norm": 0.7854464054107666,
      "learning_rate": 3.180325358897551e-05,
      "loss": 0.6127,
      "step": 7962
    },
    {
      "epoch": 1.4625534287307191,
      "grad_norm": 0.799656331539154,
      "learning_rate": 3.178186657910215e-05,
      "loss": 0.9282,
      "step": 7963
    },
    {
      "epoch": 1.4627392677940902,
      "grad_norm": 0.7751187682151794,
      "learning_rate": 3.176048540407317e-05,
      "loss": 0.9911,
      "step": 7964
    },
    {
      "epoch": 1.4629251068574614,
      "grad_norm": 0.8648492097854614,
      "learning_rate": 3.173911006571736e-05,
      "loss": 0.8404,
      "step": 7965
    },
    {
      "epoch": 1.4631109459208327,
      "grad_norm": 0.6226553320884705,
      "learning_rate": 3.171774056586303e-05,
      "loss": 0.8417,
      "step": 7966
    },
    {
      "epoch": 1.4632967849842036,
      "grad_norm": 0.6965428590774536,
      "learning_rate": 3.1696376906337896e-05,
      "loss": 1.175,
      "step": 7967
    },
    {
      "epoch": 1.463482624047575,
      "grad_norm": 0.6485991477966309,
      "learning_rate": 3.1675019088969274e-05,
      "loss": 0.964,
      "step": 7968
    },
    {
      "epoch": 1.4636684631109458,
      "grad_norm": 0.7091854214668274,
      "learning_rate": 3.1653667115583976e-05,
      "loss": 0.9875,
      "step": 7969
    },
    {
      "epoch": 1.4638543021743171,
      "grad_norm": 0.7788587808609009,
      "learning_rate": 3.16323209880082e-05,
      "loss": 0.9208,
      "step": 7970
    },
    {
      "epoch": 1.4640401412376882,
      "grad_norm": 0.6741541028022766,
      "learning_rate": 3.1610980708067805e-05,
      "loss": 0.8609,
      "step": 7971
    },
    {
      "epoch": 1.4642259803010593,
      "grad_norm": 0.855179488658905,
      "learning_rate": 3.158964627758802e-05,
      "loss": 0.9807,
      "step": 7972
    },
    {
      "epoch": 1.4644118193644304,
      "grad_norm": 0.6979156136512756,
      "learning_rate": 3.1568317698393655e-05,
      "loss": 1.2114,
      "step": 7973
    },
    {
      "epoch": 1.4645976584278015,
      "grad_norm": 0.6675369739532471,
      "learning_rate": 3.154699497230895e-05,
      "loss": 0.887,
      "step": 7974
    },
    {
      "epoch": 1.4647834974911726,
      "grad_norm": 0.7123900651931763,
      "learning_rate": 3.152567810115772e-05,
      "loss": 0.9494,
      "step": 7975
    },
    {
      "epoch": 1.4649693365545438,
      "grad_norm": 0.6881269812583923,
      "learning_rate": 3.1504367086763264e-05,
      "loss": 0.9702,
      "step": 7976
    },
    {
      "epoch": 1.4651551756179149,
      "grad_norm": 0.6381533741950989,
      "learning_rate": 3.14830619309483e-05,
      "loss": 1.0109,
      "step": 7977
    },
    {
      "epoch": 1.465341014681286,
      "grad_norm": 0.7267858982086182,
      "learning_rate": 3.146176263553513e-05,
      "loss": 1.0232,
      "step": 7978
    },
    {
      "epoch": 1.465526853744657,
      "grad_norm": 0.8178361654281616,
      "learning_rate": 3.144046920234553e-05,
      "loss": 0.9091,
      "step": 7979
    },
    {
      "epoch": 1.4657126928080282,
      "grad_norm": 0.7100738286972046,
      "learning_rate": 3.141918163320081e-05,
      "loss": 0.8937,
      "step": 7980
    },
    {
      "epoch": 1.4658985318713993,
      "grad_norm": 0.6343532204627991,
      "learning_rate": 3.1397899929921684e-05,
      "loss": 0.8325,
      "step": 7981
    },
    {
      "epoch": 1.4660843709347704,
      "grad_norm": 0.6798224449157715,
      "learning_rate": 3.1376624094328434e-05,
      "loss": 0.8343,
      "step": 7982
    },
    {
      "epoch": 1.4662702099981417,
      "grad_norm": 0.739434540271759,
      "learning_rate": 3.135535412824088e-05,
      "loss": 0.9174,
      "step": 7983
    },
    {
      "epoch": 1.4664560490615126,
      "grad_norm": 0.7073687314987183,
      "learning_rate": 3.1334090033478206e-05,
      "loss": 0.8722,
      "step": 7984
    },
    {
      "epoch": 1.466641888124884,
      "grad_norm": 0.8394288420677185,
      "learning_rate": 3.1312831811859255e-05,
      "loss": 0.8673,
      "step": 7985
    },
    {
      "epoch": 1.466827727188255,
      "grad_norm": 0.614472508430481,
      "learning_rate": 3.1291579465202224e-05,
      "loss": 0.8653,
      "step": 7986
    },
    {
      "epoch": 1.4670135662516262,
      "grad_norm": 0.6752113103866577,
      "learning_rate": 3.127033299532492e-05,
      "loss": 1.1083,
      "step": 7987
    },
    {
      "epoch": 1.4671994053149973,
      "grad_norm": 0.7037453651428223,
      "learning_rate": 3.124909240404456e-05,
      "loss": 1.0695,
      "step": 7988
    },
    {
      "epoch": 1.4673852443783684,
      "grad_norm": 0.643002986907959,
      "learning_rate": 3.122785769317791e-05,
      "loss": 0.9207,
      "step": 7989
    },
    {
      "epoch": 1.4675710834417395,
      "grad_norm": 0.6749702095985413,
      "learning_rate": 3.1206628864541264e-05,
      "loss": 0.9807,
      "step": 7990
    },
    {
      "epoch": 1.4677569225051106,
      "grad_norm": 0.8223366737365723,
      "learning_rate": 3.1185405919950295e-05,
      "loss": 0.9541,
      "step": 7991
    },
    {
      "epoch": 1.4679427615684817,
      "grad_norm": 0.6743607521057129,
      "learning_rate": 3.1164188861220325e-05,
      "loss": 0.9941,
      "step": 7992
    },
    {
      "epoch": 1.4681286006318528,
      "grad_norm": 0.773655354976654,
      "learning_rate": 3.114297769016602e-05,
      "loss": 1.0346,
      "step": 7993
    },
    {
      "epoch": 1.468314439695224,
      "grad_norm": 0.7370700836181641,
      "learning_rate": 3.1121772408601654e-05,
      "loss": 0.9967,
      "step": 7994
    },
    {
      "epoch": 1.468500278758595,
      "grad_norm": 0.6398366093635559,
      "learning_rate": 3.1100573018340965e-05,
      "loss": 0.8079,
      "step": 7995
    },
    {
      "epoch": 1.4686861178219661,
      "grad_norm": 0.6904419660568237,
      "learning_rate": 3.107937952119716e-05,
      "loss": 1.2773,
      "step": 7996
    },
    {
      "epoch": 1.4688719568853372,
      "grad_norm": 0.6900709271430969,
      "learning_rate": 3.1058191918983025e-05,
      "loss": 0.9101,
      "step": 7997
    },
    {
      "epoch": 1.4690577959487086,
      "grad_norm": 0.7680274248123169,
      "learning_rate": 3.103701021351071e-05,
      "loss": 1.0101,
      "step": 7998
    },
    {
      "epoch": 1.4692436350120794,
      "grad_norm": 0.6433159112930298,
      "learning_rate": 3.1015834406591985e-05,
      "loss": 1.0083,
      "step": 7999
    },
    {
      "epoch": 1.4694294740754508,
      "grad_norm": 0.8103683590888977,
      "learning_rate": 3.0994664500038004e-05,
      "loss": 0.8678,
      "step": 8000
    },
    {
      "epoch": 1.4696153131388217,
      "grad_norm": 0.7142993807792664,
      "learning_rate": 3.097350049565955e-05,
      "loss": 0.9821,
      "step": 8001
    },
    {
      "epoch": 1.469801152202193,
      "grad_norm": 0.659246027469635,
      "learning_rate": 3.0952342395266746e-05,
      "loss": 1.0788,
      "step": 8002
    },
    {
      "epoch": 1.469986991265564,
      "grad_norm": 0.7127649188041687,
      "learning_rate": 3.093119020066933e-05,
      "loss": 1.1004,
      "step": 8003
    },
    {
      "epoch": 1.4701728303289352,
      "grad_norm": 0.6787214279174805,
      "learning_rate": 3.091004391367652e-05,
      "loss": 0.9848,
      "step": 8004
    },
    {
      "epoch": 1.4703586693923063,
      "grad_norm": 0.6760038733482361,
      "learning_rate": 3.088890353609695e-05,
      "loss": 0.8568,
      "step": 8005
    },
    {
      "epoch": 1.4705445084556774,
      "grad_norm": 0.7517935633659363,
      "learning_rate": 3.086776906973886e-05,
      "loss": 0.9261,
      "step": 8006
    },
    {
      "epoch": 1.4707303475190485,
      "grad_norm": 0.7414070963859558,
      "learning_rate": 3.0846640516409865e-05,
      "loss": 1.121,
      "step": 8007
    },
    {
      "epoch": 1.4709161865824196,
      "grad_norm": 0.6196426153182983,
      "learning_rate": 3.082551787791716e-05,
      "loss": 1.0212,
      "step": 8008
    },
    {
      "epoch": 1.4711020256457907,
      "grad_norm": 0.7787929773330688,
      "learning_rate": 3.080440115606743e-05,
      "loss": 0.8718,
      "step": 8009
    },
    {
      "epoch": 1.4712878647091618,
      "grad_norm": 0.9043441414833069,
      "learning_rate": 3.0783290352666816e-05,
      "loss": 1.1491,
      "step": 8010
    },
    {
      "epoch": 1.471473703772533,
      "grad_norm": 0.6080015897750854,
      "learning_rate": 3.076218546952101e-05,
      "loss": 0.7824,
      "step": 8011
    },
    {
      "epoch": 1.471659542835904,
      "grad_norm": 0.7211466431617737,
      "learning_rate": 3.074108650843508e-05,
      "loss": 0.7902,
      "step": 8012
    },
    {
      "epoch": 1.4718453818992752,
      "grad_norm": 0.6852525472640991,
      "learning_rate": 3.071999347121375e-05,
      "loss": 0.8488,
      "step": 8013
    },
    {
      "epoch": 1.4720312209626463,
      "grad_norm": 0.7389121651649475,
      "learning_rate": 3.0698906359661074e-05,
      "loss": 0.7997,
      "step": 8014
    },
    {
      "epoch": 1.4722170600260176,
      "grad_norm": 0.6455549001693726,
      "learning_rate": 3.0677825175580756e-05,
      "loss": 0.718,
      "step": 8015
    },
    {
      "epoch": 1.4724028990893885,
      "grad_norm": 0.5886526703834534,
      "learning_rate": 3.065674992077584e-05,
      "loss": 0.9358,
      "step": 8016
    },
    {
      "epoch": 1.4725887381527598,
      "grad_norm": 0.6826702952384949,
      "learning_rate": 3.0635680597048975e-05,
      "loss": 1.0897,
      "step": 8017
    },
    {
      "epoch": 1.4727745772161307,
      "grad_norm": 0.7578597664833069,
      "learning_rate": 3.0614617206202287e-05,
      "loss": 1.0939,
      "step": 8018
    },
    {
      "epoch": 1.472960416279502,
      "grad_norm": 0.7338199615478516,
      "learning_rate": 3.0593559750037327e-05,
      "loss": 1.2138,
      "step": 8019
    },
    {
      "epoch": 1.4731462553428731,
      "grad_norm": 0.9254752993583679,
      "learning_rate": 3.0572508230355246e-05,
      "loss": 0.8234,
      "step": 8020
    },
    {
      "epoch": 1.4733320944062442,
      "grad_norm": 0.6886142492294312,
      "learning_rate": 3.055146264895655e-05,
      "loss": 1.069,
      "step": 8021
    },
    {
      "epoch": 1.4735179334696153,
      "grad_norm": 0.7945106625556946,
      "learning_rate": 3.053042300764135e-05,
      "loss": 0.9822,
      "step": 8022
    },
    {
      "epoch": 1.4737037725329865,
      "grad_norm": 0.6745129823684692,
      "learning_rate": 3.0509389308209224e-05,
      "loss": 0.9658,
      "step": 8023
    },
    {
      "epoch": 1.4738896115963576,
      "grad_norm": 0.6760376691818237,
      "learning_rate": 3.048836155245921e-05,
      "loss": 1.0209,
      "step": 8024
    },
    {
      "epoch": 1.4740754506597287,
      "grad_norm": 0.6404924988746643,
      "learning_rate": 3.0467339742189895e-05,
      "loss": 1.0087,
      "step": 8025
    },
    {
      "epoch": 1.4742612897230998,
      "grad_norm": 0.6808585524559021,
      "learning_rate": 3.0446323879199267e-05,
      "loss": 1.1226,
      "step": 8026
    },
    {
      "epoch": 1.4744471287864709,
      "grad_norm": 0.7933690547943115,
      "learning_rate": 3.0425313965284917e-05,
      "loss": 0.8736,
      "step": 8027
    },
    {
      "epoch": 1.474632967849842,
      "grad_norm": 0.5835078954696655,
      "learning_rate": 3.0404310002243797e-05,
      "loss": 0.8837,
      "step": 8028
    },
    {
      "epoch": 1.474818806913213,
      "grad_norm": 0.6240862607955933,
      "learning_rate": 3.038331199187249e-05,
      "loss": 0.9098,
      "step": 8029
    },
    {
      "epoch": 1.4750046459765842,
      "grad_norm": 0.6900718212127686,
      "learning_rate": 3.0362319935966942e-05,
      "loss": 1.0717,
      "step": 8030
    },
    {
      "epoch": 1.4751904850399553,
      "grad_norm": 0.6570308804512024,
      "learning_rate": 3.034133383632267e-05,
      "loss": 0.9055,
      "step": 8031
    },
    {
      "epoch": 1.4753763241033266,
      "grad_norm": 0.8068198561668396,
      "learning_rate": 3.0320353694734706e-05,
      "loss": 0.7794,
      "step": 8032
    },
    {
      "epoch": 1.4755621631666975,
      "grad_norm": 0.7877241969108582,
      "learning_rate": 3.029937951299746e-05,
      "loss": 0.971,
      "step": 8033
    },
    {
      "epoch": 1.4757480022300689,
      "grad_norm": 0.686204195022583,
      "learning_rate": 3.027841129290495e-05,
      "loss": 0.9194,
      "step": 8034
    },
    {
      "epoch": 1.47593384129344,
      "grad_norm": 0.7279642820358276,
      "learning_rate": 3.0257449036250584e-05,
      "loss": 0.8231,
      "step": 8035
    },
    {
      "epoch": 1.476119680356811,
      "grad_norm": 0.6372054219245911,
      "learning_rate": 3.023649274482733e-05,
      "loss": 0.9844,
      "step": 8036
    },
    {
      "epoch": 1.4763055194201822,
      "grad_norm": 0.6370877623558044,
      "learning_rate": 3.0215542420427667e-05,
      "loss": 0.8887,
      "step": 8037
    },
    {
      "epoch": 1.4764913584835533,
      "grad_norm": 0.7689259052276611,
      "learning_rate": 3.019459806484345e-05,
      "loss": 0.7539,
      "step": 8038
    },
    {
      "epoch": 1.4766771975469244,
      "grad_norm": 0.6394402384757996,
      "learning_rate": 3.017365967986613e-05,
      "loss": 0.737,
      "step": 8039
    },
    {
      "epoch": 1.4768630366102955,
      "grad_norm": 0.8527523875236511,
      "learning_rate": 3.0152727267286607e-05,
      "loss": 0.9706,
      "step": 8040
    },
    {
      "epoch": 1.4770488756736666,
      "grad_norm": 0.6314235329627991,
      "learning_rate": 3.0131800828895317e-05,
      "loss": 0.7901,
      "step": 8041
    },
    {
      "epoch": 1.4772347147370377,
      "grad_norm": 0.6281957626342773,
      "learning_rate": 3.0110880366482085e-05,
      "loss": 0.9776,
      "step": 8042
    },
    {
      "epoch": 1.4774205538004088,
      "grad_norm": 0.69016432762146,
      "learning_rate": 3.0089965881836324e-05,
      "loss": 1.026,
      "step": 8043
    },
    {
      "epoch": 1.47760639286378,
      "grad_norm": 0.6415174603462219,
      "learning_rate": 3.0069057376746857e-05,
      "loss": 0.998,
      "step": 8044
    },
    {
      "epoch": 1.477792231927151,
      "grad_norm": 0.7496092319488525,
      "learning_rate": 3.004815485300204e-05,
      "loss": 0.6161,
      "step": 8045
    },
    {
      "epoch": 1.4779780709905221,
      "grad_norm": 0.7724894881248474,
      "learning_rate": 3.002725831238976e-05,
      "loss": 1.1819,
      "step": 8046
    },
    {
      "epoch": 1.4781639100538932,
      "grad_norm": 0.7565277814865112,
      "learning_rate": 3.0006367756697284e-05,
      "loss": 0.9707,
      "step": 8047
    },
    {
      "epoch": 1.4783497491172644,
      "grad_norm": 0.7986220121383667,
      "learning_rate": 2.9985483187711473e-05,
      "loss": 1.0053,
      "step": 8048
    },
    {
      "epoch": 1.4785355881806357,
      "grad_norm": 0.6957052946090698,
      "learning_rate": 2.9964604607218572e-05,
      "loss": 0.984,
      "step": 8049
    },
    {
      "epoch": 1.4787214272440066,
      "grad_norm": 0.6831666231155396,
      "learning_rate": 2.9943732017004444e-05,
      "loss": 0.9922,
      "step": 8050
    },
    {
      "epoch": 1.478907266307378,
      "grad_norm": 0.7440987229347229,
      "learning_rate": 2.9922865418854295e-05,
      "loss": 0.813,
      "step": 8051
    },
    {
      "epoch": 1.479093105370749,
      "grad_norm": 0.5783529877662659,
      "learning_rate": 2.990200481455292e-05,
      "loss": 0.8253,
      "step": 8052
    },
    {
      "epoch": 1.4792789444341201,
      "grad_norm": 0.7621669769287109,
      "learning_rate": 2.988115020588458e-05,
      "loss": 0.8967,
      "step": 8053
    },
    {
      "epoch": 1.4794647834974912,
      "grad_norm": 0.6287006735801697,
      "learning_rate": 2.986030159463301e-05,
      "loss": 0.7993,
      "step": 8054
    },
    {
      "epoch": 1.4796506225608623,
      "grad_norm": 0.9010291695594788,
      "learning_rate": 2.983945898258147e-05,
      "loss": 0.6416,
      "step": 8055
    },
    {
      "epoch": 1.4798364616242334,
      "grad_norm": 0.9841485023498535,
      "learning_rate": 2.9818622371512594e-05,
      "loss": 0.9354,
      "step": 8056
    },
    {
      "epoch": 1.4800223006876045,
      "grad_norm": 0.6324443221092224,
      "learning_rate": 2.9797791763208673e-05,
      "loss": 0.7948,
      "step": 8057
    },
    {
      "epoch": 1.4802081397509756,
      "grad_norm": 0.7640272378921509,
      "learning_rate": 2.977696715945132e-05,
      "loss": 0.835,
      "step": 8058
    },
    {
      "epoch": 1.4803939788143468,
      "grad_norm": 0.654093861579895,
      "learning_rate": 2.9756148562021735e-05,
      "loss": 0.9433,
      "step": 8059
    },
    {
      "epoch": 1.4805798178777179,
      "grad_norm": 0.7237150073051453,
      "learning_rate": 2.9735335972700607e-05,
      "loss": 0.8982,
      "step": 8060
    },
    {
      "epoch": 1.480765656941089,
      "grad_norm": 0.7748257517814636,
      "learning_rate": 2.971452939326802e-05,
      "loss": 0.9941,
      "step": 8061
    },
    {
      "epoch": 1.48095149600446,
      "grad_norm": 0.609855055809021,
      "learning_rate": 2.9693728825503687e-05,
      "loss": 0.9444,
      "step": 8062
    },
    {
      "epoch": 1.4811373350678312,
      "grad_norm": 0.7217187285423279,
      "learning_rate": 2.9672934271186636e-05,
      "loss": 0.7118,
      "step": 8063
    },
    {
      "epoch": 1.4813231741312025,
      "grad_norm": 0.8199107050895691,
      "learning_rate": 2.965214573209555e-05,
      "loss": 0.9445,
      "step": 8064
    },
    {
      "epoch": 1.4815090131945734,
      "grad_norm": 0.7062032222747803,
      "learning_rate": 2.9631363210008444e-05,
      "loss": 0.9306,
      "step": 8065
    },
    {
      "epoch": 1.4816948522579447,
      "grad_norm": 0.7301066517829895,
      "learning_rate": 2.9610586706702938e-05,
      "loss": 1.0229,
      "step": 8066
    },
    {
      "epoch": 1.4818806913213156,
      "grad_norm": 0.6758785247802734,
      "learning_rate": 2.958981622395608e-05,
      "loss": 0.9138,
      "step": 8067
    },
    {
      "epoch": 1.482066530384687,
      "grad_norm": 0.8148382902145386,
      "learning_rate": 2.956905176354441e-05,
      "loss": 0.9845,
      "step": 8068
    },
    {
      "epoch": 1.482252369448058,
      "grad_norm": 0.6515848636627197,
      "learning_rate": 2.9548293327244004e-05,
      "loss": 0.8787,
      "step": 8069
    },
    {
      "epoch": 1.4824382085114292,
      "grad_norm": 0.753722071647644,
      "learning_rate": 2.95275409168303e-05,
      "loss": 0.8843,
      "step": 8070
    },
    {
      "epoch": 1.4826240475748003,
      "grad_norm": 0.6378787159919739,
      "learning_rate": 2.9506794534078363e-05,
      "loss": 1.0363,
      "step": 8071
    },
    {
      "epoch": 1.4828098866381714,
      "grad_norm": 0.7277052998542786,
      "learning_rate": 2.948605418076261e-05,
      "loss": 0.9157,
      "step": 8072
    },
    {
      "epoch": 1.4829957257015425,
      "grad_norm": 0.4940520226955414,
      "learning_rate": 2.9465319858657057e-05,
      "loss": 0.5199,
      "step": 8073
    },
    {
      "epoch": 1.4831815647649136,
      "grad_norm": 0.8421851396560669,
      "learning_rate": 2.9444591569535163e-05,
      "loss": 1.0781,
      "step": 8074
    },
    {
      "epoch": 1.4833674038282847,
      "grad_norm": 0.6914893984794617,
      "learning_rate": 2.9423869315169816e-05,
      "loss": 0.8737,
      "step": 8075
    },
    {
      "epoch": 1.4835532428916558,
      "grad_norm": 0.7011006474494934,
      "learning_rate": 2.9403153097333493e-05,
      "loss": 1.1134,
      "step": 8076
    },
    {
      "epoch": 1.4835532428916558,
      "eval_loss": 1.0098494291305542,
      "eval_runtime": 23.2152,
      "eval_samples_per_second": 47.038,
      "eval_steps_per_second": 23.519,
      "step": 8076
    },
    {
      "epoch": 1.483739081955027,
      "grad_norm": 0.6604907512664795,
      "learning_rate": 2.9382442917798027e-05,
      "loss": 1.0299,
      "step": 8077
    },
    {
      "epoch": 1.483924921018398,
      "grad_norm": 0.7371731996536255,
      "learning_rate": 2.936173877833488e-05,
      "loss": 1.2304,
      "step": 8078
    },
    {
      "epoch": 1.4841107600817691,
      "grad_norm": 0.6601764559745789,
      "learning_rate": 2.934104068071485e-05,
      "loss": 0.9159,
      "step": 8079
    },
    {
      "epoch": 1.4842965991451402,
      "grad_norm": 0.6797536015510559,
      "learning_rate": 2.9320348626708327e-05,
      "loss": 0.9202,
      "step": 8080
    },
    {
      "epoch": 1.4844824382085116,
      "grad_norm": 1.3544020652770996,
      "learning_rate": 2.9299662618085145e-05,
      "loss": 1.1437,
      "step": 8081
    },
    {
      "epoch": 1.4846682772718824,
      "grad_norm": 0.686285674571991,
      "learning_rate": 2.927898265661465e-05,
      "loss": 0.838,
      "step": 8082
    },
    {
      "epoch": 1.4848541163352538,
      "grad_norm": 0.7199298739433289,
      "learning_rate": 2.92583087440656e-05,
      "loss": 1.045,
      "step": 8083
    },
    {
      "epoch": 1.4850399553986247,
      "grad_norm": 0.7211536169052124,
      "learning_rate": 2.9237640882206274e-05,
      "loss": 1.203,
      "step": 8084
    },
    {
      "epoch": 1.485225794461996,
      "grad_norm": 0.7082626223564148,
      "learning_rate": 2.9216979072804475e-05,
      "loss": 0.9541,
      "step": 8085
    },
    {
      "epoch": 1.485411633525367,
      "grad_norm": 0.6888993978500366,
      "learning_rate": 2.9196323317627472e-05,
      "loss": 0.7679,
      "step": 8086
    },
    {
      "epoch": 1.4855974725887382,
      "grad_norm": 0.5987149477005005,
      "learning_rate": 2.9175673618441925e-05,
      "loss": 0.8992,
      "step": 8087
    },
    {
      "epoch": 1.4857833116521093,
      "grad_norm": 0.7250107526779175,
      "learning_rate": 2.9155029977014114e-05,
      "loss": 1.1304,
      "step": 8088
    },
    {
      "epoch": 1.4859691507154804,
      "grad_norm": 0.7596735954284668,
      "learning_rate": 2.9134392395109677e-05,
      "loss": 1.1595,
      "step": 8089
    },
    {
      "epoch": 1.4861549897788515,
      "grad_norm": 0.7681040167808533,
      "learning_rate": 2.9113760874493846e-05,
      "loss": 1.0206,
      "step": 8090
    },
    {
      "epoch": 1.4863408288422226,
      "grad_norm": 0.7400950193405151,
      "learning_rate": 2.9093135416931226e-05,
      "loss": 0.6833,
      "step": 8091
    },
    {
      "epoch": 1.4865266679055937,
      "grad_norm": 0.674723207950592,
      "learning_rate": 2.9072516024185993e-05,
      "loss": 1.0275,
      "step": 8092
    },
    {
      "epoch": 1.4867125069689648,
      "grad_norm": 0.7911410331726074,
      "learning_rate": 2.905190269802177e-05,
      "loss": 0.9139,
      "step": 8093
    },
    {
      "epoch": 1.486898346032336,
      "grad_norm": 0.8201619386672974,
      "learning_rate": 2.9031295440201634e-05,
      "loss": 0.9108,
      "step": 8094
    },
    {
      "epoch": 1.487084185095707,
      "grad_norm": 0.5533270835876465,
      "learning_rate": 2.9010694252488203e-05,
      "loss": 0.7583,
      "step": 8095
    },
    {
      "epoch": 1.4872700241590782,
      "grad_norm": 0.7166458368301392,
      "learning_rate": 2.89900991366435e-05,
      "loss": 0.9202,
      "step": 8096
    },
    {
      "epoch": 1.4874558632224493,
      "grad_norm": 0.6119358539581299,
      "learning_rate": 2.8969510094429074e-05,
      "loss": 0.7437,
      "step": 8097
    },
    {
      "epoch": 1.4876417022858206,
      "grad_norm": 0.6777541041374207,
      "learning_rate": 2.8948927127605985e-05,
      "loss": 1.013,
      "step": 8098
    },
    {
      "epoch": 1.4878275413491915,
      "grad_norm": 0.6690256595611572,
      "learning_rate": 2.8928350237934708e-05,
      "loss": 0.9622,
      "step": 8099
    },
    {
      "epoch": 1.4880133804125628,
      "grad_norm": 0.6234597563743591,
      "learning_rate": 2.8907779427175274e-05,
      "loss": 0.7835,
      "step": 8100
    },
    {
      "epoch": 1.488199219475934,
      "grad_norm": 0.6435613036155701,
      "learning_rate": 2.8887214697087094e-05,
      "loss": 1.0462,
      "step": 8101
    },
    {
      "epoch": 1.488385058539305,
      "grad_norm": 0.6958020329475403,
      "learning_rate": 2.8866656049429162e-05,
      "loss": 0.8945,
      "step": 8102
    },
    {
      "epoch": 1.4885708976026761,
      "grad_norm": 0.5903449654579163,
      "learning_rate": 2.884610348595984e-05,
      "loss": 0.9441,
      "step": 8103
    },
    {
      "epoch": 1.4887567366660472,
      "grad_norm": 0.7021283507347107,
      "learning_rate": 2.882555700843711e-05,
      "loss": 0.9256,
      "step": 8104
    },
    {
      "epoch": 1.4889425757294183,
      "grad_norm": 0.8339864611625671,
      "learning_rate": 2.8805016618618285e-05,
      "loss": 1.0403,
      "step": 8105
    },
    {
      "epoch": 1.4891284147927895,
      "grad_norm": 0.6968022584915161,
      "learning_rate": 2.8784482318260252e-05,
      "loss": 0.9723,
      "step": 8106
    },
    {
      "epoch": 1.4893142538561606,
      "grad_norm": 0.7342457175254822,
      "learning_rate": 2.8763954109119408e-05,
      "loss": 0.9621,
      "step": 8107
    },
    {
      "epoch": 1.4895000929195317,
      "grad_norm": 0.6052370071411133,
      "learning_rate": 2.8743431992951487e-05,
      "loss": 1.0057,
      "step": 8108
    },
    {
      "epoch": 1.4896859319829028,
      "grad_norm": 0.6225925087928772,
      "learning_rate": 2.8722915971511876e-05,
      "loss": 0.9445,
      "step": 8109
    },
    {
      "epoch": 1.4898717710462739,
      "grad_norm": 0.7247329354286194,
      "learning_rate": 2.8702406046555273e-05,
      "loss": 0.9245,
      "step": 8110
    },
    {
      "epoch": 1.490057610109645,
      "grad_norm": 0.8592934012413025,
      "learning_rate": 2.8681902219835976e-05,
      "loss": 0.8827,
      "step": 8111
    },
    {
      "epoch": 1.490243449173016,
      "grad_norm": 0.8161135911941528,
      "learning_rate": 2.8661404493107725e-05,
      "loss": 1.0164,
      "step": 8112
    },
    {
      "epoch": 1.4904292882363872,
      "grad_norm": 0.6986614465713501,
      "learning_rate": 2.8640912868123727e-05,
      "loss": 0.9613,
      "step": 8113
    },
    {
      "epoch": 1.4906151272997583,
      "grad_norm": 0.6560515761375427,
      "learning_rate": 2.8620427346636712e-05,
      "loss": 0.8074,
      "step": 8114
    },
    {
      "epoch": 1.4908009663631296,
      "grad_norm": 0.6758663058280945,
      "learning_rate": 2.8599947930398786e-05,
      "loss": 1.0231,
      "step": 8115
    },
    {
      "epoch": 1.4909868054265005,
      "grad_norm": 0.7396555542945862,
      "learning_rate": 2.8579474621161663e-05,
      "loss": 0.9426,
      "step": 8116
    },
    {
      "epoch": 1.4911726444898719,
      "grad_norm": 0.7931793928146362,
      "learning_rate": 2.8559007420676397e-05,
      "loss": 1.0995,
      "step": 8117
    },
    {
      "epoch": 1.491358483553243,
      "grad_norm": 0.5825560688972473,
      "learning_rate": 2.853854633069366e-05,
      "loss": 0.679,
      "step": 8118
    },
    {
      "epoch": 1.491544322616614,
      "grad_norm": 0.6996251344680786,
      "learning_rate": 2.8518091352963484e-05,
      "loss": 0.9056,
      "step": 8119
    },
    {
      "epoch": 1.4917301616799852,
      "grad_norm": 0.7999283671379089,
      "learning_rate": 2.8497642489235444e-05,
      "loss": 0.8565,
      "step": 8120
    },
    {
      "epoch": 1.4919160007433563,
      "grad_norm": 0.7158589363098145,
      "learning_rate": 2.847719974125861e-05,
      "loss": 0.7917,
      "step": 8121
    },
    {
      "epoch": 1.4921018398067274,
      "grad_norm": 0.7307506203651428,
      "learning_rate": 2.8456763110781427e-05,
      "loss": 0.9391,
      "step": 8122
    },
    {
      "epoch": 1.4922876788700985,
      "grad_norm": 0.7652497887611389,
      "learning_rate": 2.843633259955195e-05,
      "loss": 1.058,
      "step": 8123
    },
    {
      "epoch": 1.4924735179334696,
      "grad_norm": 0.6631175875663757,
      "learning_rate": 2.8415908209317588e-05,
      "loss": 0.9038,
      "step": 8124
    },
    {
      "epoch": 1.4926593569968407,
      "grad_norm": 0.6075304746627808,
      "learning_rate": 2.8395489941825315e-05,
      "loss": 0.759,
      "step": 8125
    },
    {
      "epoch": 1.4928451960602118,
      "grad_norm": 0.812389612197876,
      "learning_rate": 2.8375077798821535e-05,
      "loss": 0.995,
      "step": 8126
    },
    {
      "epoch": 1.493031035123583,
      "grad_norm": 0.6336919069290161,
      "learning_rate": 2.83546717820522e-05,
      "loss": 0.9026,
      "step": 8127
    },
    {
      "epoch": 1.493216874186954,
      "grad_norm": 0.7257450222969055,
      "learning_rate": 2.8334271893262587e-05,
      "loss": 1.0456,
      "step": 8128
    },
    {
      "epoch": 1.4934027132503251,
      "grad_norm": 0.6064552664756775,
      "learning_rate": 2.831387813419759e-05,
      "loss": 0.9039,
      "step": 8129
    },
    {
      "epoch": 1.4935885523136965,
      "grad_norm": 0.7207791209220886,
      "learning_rate": 2.829349050660157e-05,
      "loss": 0.8323,
      "step": 8130
    },
    {
      "epoch": 1.4937743913770674,
      "grad_norm": 0.6437431573867798,
      "learning_rate": 2.8273109012218247e-05,
      "loss": 0.8694,
      "step": 8131
    },
    {
      "epoch": 1.4939602304404387,
      "grad_norm": 0.6493000388145447,
      "learning_rate": 2.8252733652790974e-05,
      "loss": 0.7337,
      "step": 8132
    },
    {
      "epoch": 1.4941460695038096,
      "grad_norm": 0.7215288877487183,
      "learning_rate": 2.8232364430062418e-05,
      "loss": 0.8804,
      "step": 8133
    },
    {
      "epoch": 1.494331908567181,
      "grad_norm": 0.665878415107727,
      "learning_rate": 2.821200134577484e-05,
      "loss": 1.1581,
      "step": 8134
    },
    {
      "epoch": 1.494517747630552,
      "grad_norm": 0.6766164302825928,
      "learning_rate": 2.819164440166998e-05,
      "loss": 0.9617,
      "step": 8135
    },
    {
      "epoch": 1.4947035866939231,
      "grad_norm": 0.5960833430290222,
      "learning_rate": 2.8171293599488934e-05,
      "loss": 0.8801,
      "step": 8136
    },
    {
      "epoch": 1.4948894257572942,
      "grad_norm": 0.6401859521865845,
      "learning_rate": 2.815094894097242e-05,
      "loss": 0.9309,
      "step": 8137
    },
    {
      "epoch": 1.4950752648206653,
      "grad_norm": 1.31476628780365,
      "learning_rate": 2.8130610427860504e-05,
      "loss": 1.1537,
      "step": 8138
    },
    {
      "epoch": 1.4952611038840364,
      "grad_norm": 0.6927815675735474,
      "learning_rate": 2.8110278061892837e-05,
      "loss": 1.0044,
      "step": 8139
    },
    {
      "epoch": 1.4954469429474075,
      "grad_norm": 0.7193856835365295,
      "learning_rate": 2.8089951844808425e-05,
      "loss": 0.81,
      "step": 8140
    },
    {
      "epoch": 1.4956327820107786,
      "grad_norm": 0.7486124038696289,
      "learning_rate": 2.806963177834586e-05,
      "loss": 0.9009,
      "step": 8141
    },
    {
      "epoch": 1.4958186210741498,
      "grad_norm": 0.7283627390861511,
      "learning_rate": 2.804931786424314e-05,
      "loss": 0.9371,
      "step": 8142
    },
    {
      "epoch": 1.4960044601375209,
      "grad_norm": 0.6429521441459656,
      "learning_rate": 2.8029010104237785e-05,
      "loss": 0.873,
      "step": 8143
    },
    {
      "epoch": 1.496190299200892,
      "grad_norm": 0.9424156546592712,
      "learning_rate": 2.800870850006676e-05,
      "loss": 0.9181,
      "step": 8144
    },
    {
      "epoch": 1.496376138264263,
      "grad_norm": 0.7225376963615417,
      "learning_rate": 2.7988413053466468e-05,
      "loss": 0.9672,
      "step": 8145
    },
    {
      "epoch": 1.4965619773276342,
      "grad_norm": 0.7944941520690918,
      "learning_rate": 2.7968123766172872e-05,
      "loss": 1.1031,
      "step": 8146
    },
    {
      "epoch": 1.4967478163910055,
      "grad_norm": 0.677739679813385,
      "learning_rate": 2.794784063992131e-05,
      "loss": 0.7279,
      "step": 8147
    },
    {
      "epoch": 1.4969336554543764,
      "grad_norm": 0.6650786399841309,
      "learning_rate": 2.7927563676446657e-05,
      "loss": 0.9218,
      "step": 8148
    },
    {
      "epoch": 1.4971194945177477,
      "grad_norm": 0.6516730785369873,
      "learning_rate": 2.790729287748327e-05,
      "loss": 1.008,
      "step": 8149
    },
    {
      "epoch": 1.4973053335811186,
      "grad_norm": 0.8765578866004944,
      "learning_rate": 2.7887028244764924e-05,
      "loss": 0.9222,
      "step": 8150
    },
    {
      "epoch": 1.49749117264449,
      "grad_norm": 0.6291196942329407,
      "learning_rate": 2.7866769780024927e-05,
      "loss": 1.0059,
      "step": 8151
    },
    {
      "epoch": 1.497677011707861,
      "grad_norm": 0.8652521371841431,
      "learning_rate": 2.7846517484995993e-05,
      "loss": 1.0825,
      "step": 8152
    },
    {
      "epoch": 1.4978628507712322,
      "grad_norm": 0.7405718564987183,
      "learning_rate": 2.7826271361410384e-05,
      "loss": 0.9718,
      "step": 8153
    },
    {
      "epoch": 1.4980486898346033,
      "grad_norm": 0.6694837808609009,
      "learning_rate": 2.7806031410999745e-05,
      "loss": 0.9055,
      "step": 8154
    },
    {
      "epoch": 1.4982345288979744,
      "grad_norm": 0.7980131506919861,
      "learning_rate": 2.778579763549527e-05,
      "loss": 1.2031,
      "step": 8155
    },
    {
      "epoch": 1.4984203679613455,
      "grad_norm": 0.6876955032348633,
      "learning_rate": 2.7765570036627597e-05,
      "loss": 0.7122,
      "step": 8156
    },
    {
      "epoch": 1.4986062070247166,
      "grad_norm": 0.8370168805122375,
      "learning_rate": 2.7745348616126844e-05,
      "loss": 0.9301,
      "step": 8157
    },
    {
      "epoch": 1.4987920460880877,
      "grad_norm": 0.6756365895271301,
      "learning_rate": 2.7725133375722613e-05,
      "loss": 1.1112,
      "step": 8158
    },
    {
      "epoch": 1.4989778851514588,
      "grad_norm": 0.6710642576217651,
      "learning_rate": 2.77049243171439e-05,
      "loss": 0.8111,
      "step": 8159
    },
    {
      "epoch": 1.49916372421483,
      "grad_norm": 0.7266205549240112,
      "learning_rate": 2.76847214421193e-05,
      "loss": 1.0749,
      "step": 8160
    },
    {
      "epoch": 1.499349563278201,
      "grad_norm": 0.6995110511779785,
      "learning_rate": 2.766452475237672e-05,
      "loss": 1.0062,
      "step": 8161
    },
    {
      "epoch": 1.4995354023415721,
      "grad_norm": 0.6293860673904419,
      "learning_rate": 2.7644334249643688e-05,
      "loss": 0.9042,
      "step": 8162
    },
    {
      "epoch": 1.4997212414049432,
      "grad_norm": 0.7614185810089111,
      "learning_rate": 2.762414993564716e-05,
      "loss": 0.8192,
      "step": 8163
    },
    {
      "epoch": 1.4999070804683146,
      "grad_norm": 0.7335909008979797,
      "learning_rate": 2.760397181211348e-05,
      "loss": 1.0214,
      "step": 8164
    },
    {
      "epoch": 1.5000929195316854,
      "grad_norm": 0.5928364992141724,
      "learning_rate": 2.758379988076859e-05,
      "loss": 0.7934,
      "step": 8165
    },
    {
      "epoch": 1.5002787585950568,
      "grad_norm": 0.8057161569595337,
      "learning_rate": 2.756363414333778e-05,
      "loss": 0.951,
      "step": 8166
    },
    {
      "epoch": 1.5004645976584277,
      "grad_norm": 1.572473406791687,
      "learning_rate": 2.7543474601545927e-05,
      "loss": 1.5114,
      "step": 8167
    },
    {
      "epoch": 1.500650436721799,
      "grad_norm": 0.5864894390106201,
      "learning_rate": 2.7523321257117253e-05,
      "loss": 0.7276,
      "step": 8168
    },
    {
      "epoch": 1.50083627578517,
      "grad_norm": 0.6349899172782898,
      "learning_rate": 2.7503174111775566e-05,
      "loss": 0.9229,
      "step": 8169
    },
    {
      "epoch": 1.5010221148485412,
      "grad_norm": 0.723136842250824,
      "learning_rate": 2.7483033167244065e-05,
      "loss": 1.0168,
      "step": 8170
    },
    {
      "epoch": 1.5012079539119123,
      "grad_norm": 0.7174968719482422,
      "learning_rate": 2.7462898425245487e-05,
      "loss": 1.0505,
      "step": 8171
    },
    {
      "epoch": 1.5013937929752834,
      "grad_norm": 0.7736288905143738,
      "learning_rate": 2.7442769887501996e-05,
      "loss": 1.0762,
      "step": 8172
    },
    {
      "epoch": 1.5015796320386545,
      "grad_norm": 0.6830939054489136,
      "learning_rate": 2.7422647555735192e-05,
      "loss": 0.8832,
      "step": 8173
    },
    {
      "epoch": 1.5017654711020256,
      "grad_norm": 0.6052866578102112,
      "learning_rate": 2.7402531431666234e-05,
      "loss": 0.8339,
      "step": 8174
    },
    {
      "epoch": 1.5019513101653967,
      "grad_norm": 0.8002974987030029,
      "learning_rate": 2.7382421517015634e-05,
      "loss": 0.9862,
      "step": 8175
    },
    {
      "epoch": 1.5021371492287678,
      "grad_norm": 0.6755656003952026,
      "learning_rate": 2.7362317813503467e-05,
      "loss": 0.7707,
      "step": 8176
    },
    {
      "epoch": 1.5023229882921392,
      "grad_norm": 0.7527087330818176,
      "learning_rate": 2.7342220322849276e-05,
      "loss": 1.0019,
      "step": 8177
    },
    {
      "epoch": 1.50250882735551,
      "grad_norm": 0.7766450643539429,
      "learning_rate": 2.7322129046771994e-05,
      "loss": 0.8555,
      "step": 8178
    },
    {
      "epoch": 1.5026946664188814,
      "grad_norm": 0.6567835211753845,
      "learning_rate": 2.730204398699011e-05,
      "loss": 1.0564,
      "step": 8179
    },
    {
      "epoch": 1.5028805054822523,
      "grad_norm": 0.7293635010719299,
      "learning_rate": 2.72819651452215e-05,
      "loss": 0.9468,
      "step": 8180
    },
    {
      "epoch": 1.5030663445456236,
      "grad_norm": 0.7716689109802246,
      "learning_rate": 2.7261892523183607e-05,
      "loss": 0.9835,
      "step": 8181
    },
    {
      "epoch": 1.5032521836089945,
      "grad_norm": 0.6577703356742859,
      "learning_rate": 2.724182612259323e-05,
      "loss": 0.9563,
      "step": 8182
    },
    {
      "epoch": 1.5034380226723658,
      "grad_norm": 0.6561897993087769,
      "learning_rate": 2.72217659451667e-05,
      "loss": 1.1897,
      "step": 8183
    },
    {
      "epoch": 1.5036238617357367,
      "grad_norm": 0.728302538394928,
      "learning_rate": 2.720171199261987e-05,
      "loss": 0.8003,
      "step": 8184
    },
    {
      "epoch": 1.503809700799108,
      "grad_norm": 0.7979081273078918,
      "learning_rate": 2.7181664266667906e-05,
      "loss": 1.0519,
      "step": 8185
    },
    {
      "epoch": 1.5039955398624791,
      "grad_norm": 0.7118033170700073,
      "learning_rate": 2.7161622769025585e-05,
      "loss": 0.9791,
      "step": 8186
    },
    {
      "epoch": 1.5041813789258502,
      "grad_norm": 0.7213948965072632,
      "learning_rate": 2.7141587501407095e-05,
      "loss": 0.9177,
      "step": 8187
    },
    {
      "epoch": 1.5043672179892214,
      "grad_norm": 0.8359013199806213,
      "learning_rate": 2.7121558465526133e-05,
      "loss": 1.0279,
      "step": 8188
    },
    {
      "epoch": 1.5045530570525925,
      "grad_norm": 0.6666085720062256,
      "learning_rate": 2.7101535663095735e-05,
      "loss": 1.1037,
      "step": 8189
    },
    {
      "epoch": 1.5047388961159636,
      "grad_norm": 0.7200797200202942,
      "learning_rate": 2.7081519095828568e-05,
      "loss": 0.9053,
      "step": 8190
    },
    {
      "epoch": 1.5049247351793347,
      "grad_norm": 0.6248354911804199,
      "learning_rate": 2.706150876543668e-05,
      "loss": 1.0788,
      "step": 8191
    },
    {
      "epoch": 1.5051105742427058,
      "grad_norm": 0.6507927775382996,
      "learning_rate": 2.7041504673631567e-05,
      "loss": 0.8718,
      "step": 8192
    },
    {
      "epoch": 1.5052964133060769,
      "grad_norm": 0.7430590391159058,
      "learning_rate": 2.7021506822124275e-05,
      "loss": 0.8365,
      "step": 8193
    },
    {
      "epoch": 1.5054822523694482,
      "grad_norm": 0.7140336632728577,
      "learning_rate": 2.7001515212625205e-05,
      "loss": 0.9323,
      "step": 8194
    },
    {
      "epoch": 1.505668091432819,
      "grad_norm": 0.8088352084159851,
      "learning_rate": 2.6981529846844332e-05,
      "loss": 1.1455,
      "step": 8195
    },
    {
      "epoch": 1.5058539304961904,
      "grad_norm": 0.8508310914039612,
      "learning_rate": 2.6961550726491004e-05,
      "loss": 1.084,
      "step": 8196
    },
    {
      "epoch": 1.5060397695595613,
      "grad_norm": 0.6638537645339966,
      "learning_rate": 2.6941577853274093e-05,
      "loss": 1.0291,
      "step": 8197
    },
    {
      "epoch": 1.5062256086229326,
      "grad_norm": 0.597263753414154,
      "learning_rate": 2.6921611228901957e-05,
      "loss": 0.9016,
      "step": 8198
    },
    {
      "epoch": 1.5064114476863035,
      "grad_norm": 0.6490717530250549,
      "learning_rate": 2.690165085508233e-05,
      "loss": 0.8965,
      "step": 8199
    },
    {
      "epoch": 1.5065972867496749,
      "grad_norm": 0.6764369010925293,
      "learning_rate": 2.688169673352249e-05,
      "loss": 0.8949,
      "step": 8200
    },
    {
      "epoch": 1.5067831258130457,
      "grad_norm": 0.746873140335083,
      "learning_rate": 2.686174886592915e-05,
      "loss": 0.9743,
      "step": 8201
    },
    {
      "epoch": 1.506968964876417,
      "grad_norm": 0.5665560960769653,
      "learning_rate": 2.6841807254008545e-05,
      "loss": 0.6897,
      "step": 8202
    },
    {
      "epoch": 1.5071548039397882,
      "grad_norm": 0.7615552544593811,
      "learning_rate": 2.6821871899466234e-05,
      "loss": 0.9747,
      "step": 8203
    },
    {
      "epoch": 1.5073406430031593,
      "grad_norm": 0.850107729434967,
      "learning_rate": 2.6801942804007397e-05,
      "loss": 0.7636,
      "step": 8204
    },
    {
      "epoch": 1.5075264820665304,
      "grad_norm": 0.6412014365196228,
      "learning_rate": 2.6782019969336615e-05,
      "loss": 0.843,
      "step": 8205
    },
    {
      "epoch": 1.5077123211299015,
      "grad_norm": 0.7340335249900818,
      "learning_rate": 2.676210339715788e-05,
      "loss": 0.9309,
      "step": 8206
    },
    {
      "epoch": 1.5078981601932726,
      "grad_norm": 0.6725056767463684,
      "learning_rate": 2.6742193089174762e-05,
      "loss": 0.7102,
      "step": 8207
    },
    {
      "epoch": 1.5080839992566437,
      "grad_norm": 0.6415348649024963,
      "learning_rate": 2.6722289047090166e-05,
      "loss": 0.9333,
      "step": 8208
    },
    {
      "epoch": 1.5082698383200148,
      "grad_norm": 0.8024607300758362,
      "learning_rate": 2.6702391272606596e-05,
      "loss": 0.9718,
      "step": 8209
    },
    {
      "epoch": 1.508455677383386,
      "grad_norm": 0.6357980370521545,
      "learning_rate": 2.6682499767425885e-05,
      "loss": 0.919,
      "step": 8210
    },
    {
      "epoch": 1.5086415164467573,
      "grad_norm": 0.7079041004180908,
      "learning_rate": 2.666261453324943e-05,
      "loss": 0.9,
      "step": 8211
    },
    {
      "epoch": 1.5088273555101281,
      "grad_norm": 0.8778452277183533,
      "learning_rate": 2.664273557177809e-05,
      "loss": 1.0977,
      "step": 8212
    },
    {
      "epoch": 1.5090131945734995,
      "grad_norm": 0.5249161124229431,
      "learning_rate": 2.6622862884712085e-05,
      "loss": 0.7411,
      "step": 8213
    },
    {
      "epoch": 1.5091990336368704,
      "grad_norm": 0.670387327671051,
      "learning_rate": 2.6602996473751208e-05,
      "loss": 0.9845,
      "step": 8214
    },
    {
      "epoch": 1.5093848727002417,
      "grad_norm": 0.7721664309501648,
      "learning_rate": 2.6583136340594674e-05,
      "loss": 0.9434,
      "step": 8215
    },
    {
      "epoch": 1.5095707117636126,
      "grad_norm": 0.7793735861778259,
      "learning_rate": 2.6563282486941197e-05,
      "loss": 0.8901,
      "step": 8216
    },
    {
      "epoch": 1.509756550826984,
      "grad_norm": 0.6147542595863342,
      "learning_rate": 2.6543434914488852e-05,
      "loss": 0.8793,
      "step": 8217
    },
    {
      "epoch": 1.509942389890355,
      "grad_norm": 0.686143159866333,
      "learning_rate": 2.6523593624935283e-05,
      "loss": 0.9777,
      "step": 8218
    },
    {
      "epoch": 1.5101282289537261,
      "grad_norm": 0.6980476379394531,
      "learning_rate": 2.6503758619977594e-05,
      "loss": 0.9811,
      "step": 8219
    },
    {
      "epoch": 1.5103140680170972,
      "grad_norm": 0.6905124187469482,
      "learning_rate": 2.6483929901312233e-05,
      "loss": 0.8508,
      "step": 8220
    },
    {
      "epoch": 1.5104999070804683,
      "grad_norm": 0.6223216652870178,
      "learning_rate": 2.646410747063527e-05,
      "loss": 1.1032,
      "step": 8221
    },
    {
      "epoch": 1.5106857461438394,
      "grad_norm": 0.7925272583961487,
      "learning_rate": 2.644429132964209e-05,
      "loss": 1.1465,
      "step": 8222
    },
    {
      "epoch": 1.5108715852072105,
      "grad_norm": 0.6701057553291321,
      "learning_rate": 2.6424481480027695e-05,
      "loss": 0.9851,
      "step": 8223
    },
    {
      "epoch": 1.5110574242705817,
      "grad_norm": 0.7091572284698486,
      "learning_rate": 2.6404677923486377e-05,
      "loss": 0.9173,
      "step": 8224
    },
    {
      "epoch": 1.5112432633339528,
      "grad_norm": 0.7232625484466553,
      "learning_rate": 2.638488066171201e-05,
      "loss": 1.1431,
      "step": 8225
    },
    {
      "epoch": 1.5114291023973239,
      "grad_norm": 0.6791695356369019,
      "learning_rate": 2.636508969639794e-05,
      "loss": 0.9333,
      "step": 8226
    },
    {
      "epoch": 1.511614941460695,
      "grad_norm": 0.7125614881515503,
      "learning_rate": 2.6345305029236867e-05,
      "loss": 0.8559,
      "step": 8227
    },
    {
      "epoch": 1.5118007805240663,
      "grad_norm": 0.6903975605964661,
      "learning_rate": 2.632552666192104e-05,
      "loss": 0.8817,
      "step": 8228
    },
    {
      "epoch": 1.5119866195874372,
      "grad_norm": 0.8414456844329834,
      "learning_rate": 2.6305754596142173e-05,
      "loss": 1.0211,
      "step": 8229
    },
    {
      "epoch": 1.5121724586508085,
      "grad_norm": 0.6724753975868225,
      "learning_rate": 2.6285988833591367e-05,
      "loss": 0.9309,
      "step": 8230
    },
    {
      "epoch": 1.5123582977141794,
      "grad_norm": 0.6871885061264038,
      "learning_rate": 2.6266229375959252e-05,
      "loss": 0.884,
      "step": 8231
    },
    {
      "epoch": 1.5125441367775507,
      "grad_norm": 0.7753848433494568,
      "learning_rate": 2.6246476224935902e-05,
      "loss": 0.9779,
      "step": 8232
    },
    {
      "epoch": 1.5127299758409216,
      "grad_norm": 0.8011336922645569,
      "learning_rate": 2.6226729382210868e-05,
      "loss": 1.0863,
      "step": 8233
    },
    {
      "epoch": 1.512915814904293,
      "grad_norm": 0.74050372838974,
      "learning_rate": 2.6206988849473092e-05,
      "loss": 0.9114,
      "step": 8234
    },
    {
      "epoch": 1.513101653967664,
      "grad_norm": 1.0021766424179077,
      "learning_rate": 2.6187254628411064e-05,
      "loss": 1.1928,
      "step": 8235
    },
    {
      "epoch": 1.5132874930310352,
      "grad_norm": 0.706662654876709,
      "learning_rate": 2.6167526720712655e-05,
      "loss": 1.0503,
      "step": 8236
    },
    {
      "epoch": 1.5134733320944063,
      "grad_norm": 0.7461012005805969,
      "learning_rate": 2.6147805128065294e-05,
      "loss": 1.2067,
      "step": 8237
    },
    {
      "epoch": 1.5136591711577774,
      "grad_norm": 0.6484167575836182,
      "learning_rate": 2.6128089852155736e-05,
      "loss": 0.9687,
      "step": 8238
    },
    {
      "epoch": 1.5138450102211485,
      "grad_norm": 0.7335364818572998,
      "learning_rate": 2.6108380894670315e-05,
      "loss": 0.9708,
      "step": 8239
    },
    {
      "epoch": 1.5140308492845196,
      "grad_norm": 0.8530272841453552,
      "learning_rate": 2.608867825729481e-05,
      "loss": 0.9567,
      "step": 8240
    },
    {
      "epoch": 1.5142166883478907,
      "grad_norm": 0.6858327984809875,
      "learning_rate": 2.6068981941714367e-05,
      "loss": 0.9842,
      "step": 8241
    },
    {
      "epoch": 1.5144025274112618,
      "grad_norm": 0.6089559197425842,
      "learning_rate": 2.604929194961371e-05,
      "loss": 0.9545,
      "step": 8242
    },
    {
      "epoch": 1.5145883664746331,
      "grad_norm": 0.6408554911613464,
      "learning_rate": 2.6029608282676908e-05,
      "loss": 0.7563,
      "step": 8243
    },
    {
      "epoch": 1.514774205538004,
      "grad_norm": 0.6375374794006348,
      "learning_rate": 2.600993094258759e-05,
      "loss": 0.9235,
      "step": 8244
    },
    {
      "epoch": 1.5149600446013753,
      "grad_norm": 0.6979017853736877,
      "learning_rate": 2.5990259931028792e-05,
      "loss": 0.9803,
      "step": 8245
    },
    {
      "epoch": 1.5151458836647462,
      "grad_norm": 1.0037568807601929,
      "learning_rate": 2.5970595249683017e-05,
      "loss": 1.0065,
      "step": 8246
    },
    {
      "epoch": 1.5153317227281176,
      "grad_norm": 0.764148473739624,
      "learning_rate": 2.595093690023225e-05,
      "loss": 1.068,
      "step": 8247
    },
    {
      "epoch": 1.5155175617914884,
      "grad_norm": 0.7334862351417542,
      "learning_rate": 2.5931284884357864e-05,
      "loss": 1.2625,
      "step": 8248
    },
    {
      "epoch": 1.5157034008548598,
      "grad_norm": 0.734988272190094,
      "learning_rate": 2.59116392037408e-05,
      "loss": 0.9587,
      "step": 8249
    },
    {
      "epoch": 1.5158892399182307,
      "grad_norm": 0.6814414858818054,
      "learning_rate": 2.5891999860061333e-05,
      "loss": 0.9823,
      "step": 8250
    },
    {
      "epoch": 1.516075078981602,
      "grad_norm": 0.7905431985855103,
      "learning_rate": 2.587236685499931e-05,
      "loss": 0.8889,
      "step": 8251
    },
    {
      "epoch": 1.516260918044973,
      "grad_norm": 0.8326538801193237,
      "learning_rate": 2.5852740190233937e-05,
      "loss": 0.8816,
      "step": 8252
    },
    {
      "epoch": 1.5164467571083442,
      "grad_norm": 0.787573516368866,
      "learning_rate": 2.5833119867443945e-05,
      "loss": 0.8138,
      "step": 8253
    },
    {
      "epoch": 1.5166325961717153,
      "grad_norm": 0.820881187915802,
      "learning_rate": 2.581350588830753e-05,
      "loss": 0.792,
      "step": 8254
    },
    {
      "epoch": 1.5168184352350864,
      "grad_norm": 0.6634076833724976,
      "learning_rate": 2.579389825450228e-05,
      "loss": 0.9024,
      "step": 8255
    },
    {
      "epoch": 1.5170042742984575,
      "grad_norm": 0.660010576248169,
      "learning_rate": 2.5774296967705314e-05,
      "loss": 1.0527,
      "step": 8256
    },
    {
      "epoch": 1.5171901133618286,
      "grad_norm": 0.8844035267829895,
      "learning_rate": 2.5754702029593113e-05,
      "loss": 1.1888,
      "step": 8257
    },
    {
      "epoch": 1.5173759524251997,
      "grad_norm": 0.6765933036804199,
      "learning_rate": 2.573511344184173e-05,
      "loss": 1.0193,
      "step": 8258
    },
    {
      "epoch": 1.5175617914885708,
      "grad_norm": 0.725508451461792,
      "learning_rate": 2.5715531206126585e-05,
      "loss": 0.9188,
      "step": 8259
    },
    {
      "epoch": 1.5177476305519422,
      "grad_norm": 0.6848968267440796,
      "learning_rate": 2.5695955324122622e-05,
      "loss": 1.0983,
      "step": 8260
    },
    {
      "epoch": 1.517933469615313,
      "grad_norm": 0.8235938549041748,
      "learning_rate": 2.567638579750421e-05,
      "loss": 0.8204,
      "step": 8261
    },
    {
      "epoch": 1.5181193086786844,
      "grad_norm": 0.6256136894226074,
      "learning_rate": 2.5656822627945133e-05,
      "loss": 0.9667,
      "step": 8262
    },
    {
      "epoch": 1.5183051477420553,
      "grad_norm": 0.7181174755096436,
      "learning_rate": 2.5637265817118715e-05,
      "loss": 0.9107,
      "step": 8263
    },
    {
      "epoch": 1.5184909868054266,
      "grad_norm": 0.6005790829658508,
      "learning_rate": 2.5617715366697647e-05,
      "loss": 0.8373,
      "step": 8264
    },
    {
      "epoch": 1.5186768258687975,
      "grad_norm": 0.7009453773498535,
      "learning_rate": 2.559817127835418e-05,
      "loss": 0.9135,
      "step": 8265
    },
    {
      "epoch": 1.5188626649321688,
      "grad_norm": 0.7299928665161133,
      "learning_rate": 2.5578633553759878e-05,
      "loss": 1.1003,
      "step": 8266
    },
    {
      "epoch": 1.5190485039955397,
      "grad_norm": 0.7495889663696289,
      "learning_rate": 2.5559102194585917e-05,
      "loss": 1.2302,
      "step": 8267
    },
    {
      "epoch": 1.519234343058911,
      "grad_norm": 0.7094706296920776,
      "learning_rate": 2.5539577202502852e-05,
      "loss": 0.8065,
      "step": 8268
    },
    {
      "epoch": 1.5194201821222821,
      "grad_norm": 0.651636004447937,
      "learning_rate": 2.552005857918065e-05,
      "loss": 0.8975,
      "step": 8269
    },
    {
      "epoch": 1.5196060211856532,
      "grad_norm": 0.7161694765090942,
      "learning_rate": 2.550054632628883e-05,
      "loss": 1.0802,
      "step": 8270
    },
    {
      "epoch": 1.5197918602490244,
      "grad_norm": 0.6569831371307373,
      "learning_rate": 2.5481040445496286e-05,
      "loss": 1.0261,
      "step": 8271
    },
    {
      "epoch": 1.5199776993123955,
      "grad_norm": 0.7755568623542786,
      "learning_rate": 2.5461540938471397e-05,
      "loss": 1.0336,
      "step": 8272
    },
    {
      "epoch": 1.5201635383757666,
      "grad_norm": 0.6399084329605103,
      "learning_rate": 2.5442047806882007e-05,
      "loss": 0.9518,
      "step": 8273
    },
    {
      "epoch": 1.5203493774391377,
      "grad_norm": 0.7283554077148438,
      "learning_rate": 2.5422561052395455e-05,
      "loss": 0.9542,
      "step": 8274
    },
    {
      "epoch": 1.5205352165025088,
      "grad_norm": 0.7275633811950684,
      "learning_rate": 2.5403080676678392e-05,
      "loss": 0.982,
      "step": 8275
    },
    {
      "epoch": 1.5207210555658799,
      "grad_norm": 0.6447848677635193,
      "learning_rate": 2.5383606681397076e-05,
      "loss": 0.701,
      "step": 8276
    },
    {
      "epoch": 1.5209068946292512,
      "grad_norm": 0.801145613193512,
      "learning_rate": 2.5364139068217174e-05,
      "loss": 0.891,
      "step": 8277
    },
    {
      "epoch": 1.521092733692622,
      "grad_norm": 0.8931204676628113,
      "learning_rate": 2.5344677838803733e-05,
      "loss": 1.1393,
      "step": 8278
    },
    {
      "epoch": 1.5212785727559934,
      "grad_norm": 0.721251368522644,
      "learning_rate": 2.5325222994821385e-05,
      "loss": 0.9685,
      "step": 8279
    },
    {
      "epoch": 1.5214644118193643,
      "grad_norm": 0.7849665284156799,
      "learning_rate": 2.5305774537934067e-05,
      "loss": 1.1303,
      "step": 8280
    },
    {
      "epoch": 1.5216502508827356,
      "grad_norm": 0.6657546758651733,
      "learning_rate": 2.5286332469805296e-05,
      "loss": 0.8404,
      "step": 8281
    },
    {
      "epoch": 1.5218360899461065,
      "grad_norm": 0.6446507573127747,
      "learning_rate": 2.5266896792098015e-05,
      "loss": 0.8706,
      "step": 8282
    },
    {
      "epoch": 1.5220219290094779,
      "grad_norm": 0.6058208346366882,
      "learning_rate": 2.5247467506474543e-05,
      "loss": 1.0097,
      "step": 8283
    },
    {
      "epoch": 1.522207768072849,
      "grad_norm": 0.6460893154144287,
      "learning_rate": 2.522804461459677e-05,
      "loss": 0.8407,
      "step": 8284
    },
    {
      "epoch": 1.52239360713622,
      "grad_norm": 0.7795756459236145,
      "learning_rate": 2.520862811812591e-05,
      "loss": 0.8684,
      "step": 8285
    },
    {
      "epoch": 1.5225794461995912,
      "grad_norm": 0.6236745715141296,
      "learning_rate": 2.518921801872276e-05,
      "loss": 0.739,
      "step": 8286
    },
    {
      "epoch": 1.5227652852629623,
      "grad_norm": 0.698995053768158,
      "learning_rate": 2.5169814318047447e-05,
      "loss": 1.0715,
      "step": 8287
    },
    {
      "epoch": 1.5229511243263334,
      "grad_norm": 0.6841782331466675,
      "learning_rate": 2.5150417017759654e-05,
      "loss": 0.9162,
      "step": 8288
    },
    {
      "epoch": 1.5231369633897045,
      "grad_norm": 0.7248424291610718,
      "learning_rate": 2.513102611951845e-05,
      "loss": 0.9685,
      "step": 8289
    },
    {
      "epoch": 1.5233228024530756,
      "grad_norm": 0.6345195174217224,
      "learning_rate": 2.5111641624982417e-05,
      "loss": 0.9229,
      "step": 8290
    },
    {
      "epoch": 1.5235086415164467,
      "grad_norm": 0.6780673861503601,
      "learning_rate": 2.5092263535809545e-05,
      "loss": 1.0447,
      "step": 8291
    },
    {
      "epoch": 1.5236944805798178,
      "grad_norm": 0.7004578113555908,
      "learning_rate": 2.5072891853657243e-05,
      "loss": 0.9548,
      "step": 8292
    },
    {
      "epoch": 1.523880319643189,
      "grad_norm": 0.7394935488700867,
      "learning_rate": 2.5053526580182472e-05,
      "loss": 0.889,
      "step": 8293
    },
    {
      "epoch": 1.5240661587065603,
      "grad_norm": 0.8624938130378723,
      "learning_rate": 2.5034167717041523e-05,
      "loss": 1.1627,
      "step": 8294
    },
    {
      "epoch": 1.5242519977699311,
      "grad_norm": 0.7264074683189392,
      "learning_rate": 2.5014815265890236e-05,
      "loss": 1.0638,
      "step": 8295
    },
    {
      "epoch": 1.5244378368333025,
      "grad_norm": 0.8741801381111145,
      "learning_rate": 2.4995469228383884e-05,
      "loss": 0.8824,
      "step": 8296
    },
    {
      "epoch": 1.5246236758966734,
      "grad_norm": 0.8204243779182434,
      "learning_rate": 2.4976129606177122e-05,
      "loss": 0.894,
      "step": 8297
    },
    {
      "epoch": 1.5248095149600447,
      "grad_norm": 0.7651588320732117,
      "learning_rate": 2.4956796400924188e-05,
      "loss": 1.0573,
      "step": 8298
    },
    {
      "epoch": 1.5249953540234156,
      "grad_norm": 0.7579216361045837,
      "learning_rate": 2.493746961427861e-05,
      "loss": 1.1599,
      "step": 8299
    },
    {
      "epoch": 1.525181193086787,
      "grad_norm": 0.7527614235877991,
      "learning_rate": 2.4918149247893508e-05,
      "loss": 1.167,
      "step": 8300
    },
    {
      "epoch": 1.525367032150158,
      "grad_norm": 0.7581583261489868,
      "learning_rate": 2.4898835303421353e-05,
      "loss": 0.9118,
      "step": 8301
    },
    {
      "epoch": 1.5255528712135291,
      "grad_norm": 0.9576455354690552,
      "learning_rate": 2.4879527782514134e-05,
      "loss": 1.1397,
      "step": 8302
    },
    {
      "epoch": 1.5257387102769002,
      "grad_norm": 0.756462812423706,
      "learning_rate": 2.4860226686823253e-05,
      "loss": 1.0269,
      "step": 8303
    },
    {
      "epoch": 1.5259245493402713,
      "grad_norm": 0.935738205909729,
      "learning_rate": 2.4840932017999584e-05,
      "loss": 1.1624,
      "step": 8304
    },
    {
      "epoch": 1.5261103884036424,
      "grad_norm": 0.8938685655593872,
      "learning_rate": 2.4821643777693472e-05,
      "loss": 0.89,
      "step": 8305
    },
    {
      "epoch": 1.5262962274670135,
      "grad_norm": 0.7011599540710449,
      "learning_rate": 2.480236196755462e-05,
      "loss": 0.7417,
      "step": 8306
    },
    {
      "epoch": 1.5264820665303847,
      "grad_norm": 0.6875777244567871,
      "learning_rate": 2.4783086589232295e-05,
      "loss": 1.086,
      "step": 8307
    },
    {
      "epoch": 1.5266679055937558,
      "grad_norm": 0.7056373357772827,
      "learning_rate": 2.476381764437512e-05,
      "loss": 0.8894,
      "step": 8308
    },
    {
      "epoch": 1.526853744657127,
      "grad_norm": 0.7568773627281189,
      "learning_rate": 2.474455513463122e-05,
      "loss": 1.0199,
      "step": 8309
    },
    {
      "epoch": 1.527039583720498,
      "grad_norm": 0.7980220913887024,
      "learning_rate": 2.47252990616482e-05,
      "loss": 0.874,
      "step": 8310
    },
    {
      "epoch": 1.5272254227838693,
      "grad_norm": 0.7311957478523254,
      "learning_rate": 2.470604942707302e-05,
      "loss": 0.979,
      "step": 8311
    },
    {
      "epoch": 1.5274112618472402,
      "grad_norm": 1.1911845207214355,
      "learning_rate": 2.4686806232552183e-05,
      "loss": 0.8673,
      "step": 8312
    },
    {
      "epoch": 1.5275971009106115,
      "grad_norm": 0.748319685459137,
      "learning_rate": 2.4667569479731556e-05,
      "loss": 1.1138,
      "step": 8313
    },
    {
      "epoch": 1.5277829399739824,
      "grad_norm": 0.7561832666397095,
      "learning_rate": 2.4648339170256552e-05,
      "loss": 1.1001,
      "step": 8314
    },
    {
      "epoch": 1.5279687790373537,
      "grad_norm": 0.7505941987037659,
      "learning_rate": 2.4629115305771933e-05,
      "loss": 1.1796,
      "step": 8315
    },
    {
      "epoch": 1.5281546181007246,
      "grad_norm": 0.6446735858917236,
      "learning_rate": 2.4609897887921973e-05,
      "loss": 0.954,
      "step": 8316
    },
    {
      "epoch": 1.528340457164096,
      "grad_norm": 0.6511173248291016,
      "learning_rate": 2.459068691835039e-05,
      "loss": 0.8117,
      "step": 8317
    },
    {
      "epoch": 1.528526296227467,
      "grad_norm": 0.7788544297218323,
      "learning_rate": 2.457148239870033e-05,
      "loss": 0.9103,
      "step": 8318
    },
    {
      "epoch": 1.5287121352908382,
      "grad_norm": 0.7949376702308655,
      "learning_rate": 2.455228433061444e-05,
      "loss": 1.0756,
      "step": 8319
    },
    {
      "epoch": 1.5288979743542093,
      "grad_norm": 0.7452298402786255,
      "learning_rate": 2.4533092715734708e-05,
      "loss": 0.8614,
      "step": 8320
    },
    {
      "epoch": 1.5290838134175804,
      "grad_norm": 0.6368029713630676,
      "learning_rate": 2.4513907555702675e-05,
      "loss": 0.7188,
      "step": 8321
    },
    {
      "epoch": 1.5292696524809515,
      "grad_norm": 0.8253371119499207,
      "learning_rate": 2.4494728852159266e-05,
      "loss": 0.829,
      "step": 8322
    },
    {
      "epoch": 1.5294554915443226,
      "grad_norm": 0.7835356593132019,
      "learning_rate": 2.4475556606744875e-05,
      "loss": 0.9303,
      "step": 8323
    },
    {
      "epoch": 1.5296413306076937,
      "grad_norm": 0.7281311750411987,
      "learning_rate": 2.4456390821099385e-05,
      "loss": 1.0951,
      "step": 8324
    },
    {
      "epoch": 1.5298271696710648,
      "grad_norm": 0.6802740097045898,
      "learning_rate": 2.4437231496862034e-05,
      "loss": 0.9621,
      "step": 8325
    },
    {
      "epoch": 1.5300130087344361,
      "grad_norm": 0.6554965376853943,
      "learning_rate": 2.4418078635671616e-05,
      "loss": 1.1749,
      "step": 8326
    },
    {
      "epoch": 1.530198847797807,
      "grad_norm": 0.7031880021095276,
      "learning_rate": 2.439893223916626e-05,
      "loss": 1.1384,
      "step": 8327
    },
    {
      "epoch": 1.5303846868611783,
      "grad_norm": 0.682519257068634,
      "learning_rate": 2.437979230898365e-05,
      "loss": 0.7771,
      "step": 8328
    },
    {
      "epoch": 1.5305705259245492,
      "grad_norm": 0.6459039449691772,
      "learning_rate": 2.436065884676081e-05,
      "loss": 1.0385,
      "step": 8329
    },
    {
      "epoch": 1.5307563649879206,
      "grad_norm": 0.777291476726532,
      "learning_rate": 2.4341531854134304e-05,
      "loss": 1.21,
      "step": 8330
    },
    {
      "epoch": 1.5309422040512914,
      "grad_norm": 0.7058006525039673,
      "learning_rate": 2.4322411332740115e-05,
      "loss": 0.8476,
      "step": 8331
    },
    {
      "epoch": 1.5311280431146628,
      "grad_norm": 0.689065158367157,
      "learning_rate": 2.430329728421362e-05,
      "loss": 0.6523,
      "step": 8332
    },
    {
      "epoch": 1.5313138821780337,
      "grad_norm": 0.6234003305435181,
      "learning_rate": 2.428418971018972e-05,
      "loss": 1.0589,
      "step": 8333
    },
    {
      "epoch": 1.531499721241405,
      "grad_norm": 0.6710896492004395,
      "learning_rate": 2.4265088612302712e-05,
      "loss": 0.9926,
      "step": 8334
    },
    {
      "epoch": 1.531685560304776,
      "grad_norm": 0.698925793170929,
      "learning_rate": 2.424599399218639e-05,
      "loss": 1.093,
      "step": 8335
    },
    {
      "epoch": 1.5318713993681472,
      "grad_norm": 0.6863391399383545,
      "learning_rate": 2.422690585147389e-05,
      "loss": 0.7161,
      "step": 8336
    },
    {
      "epoch": 1.5320572384315183,
      "grad_norm": 0.7835180759429932,
      "learning_rate": 2.4207824191797913e-05,
      "loss": 0.895,
      "step": 8337
    },
    {
      "epoch": 1.5322430774948894,
      "grad_norm": 0.8524803519248962,
      "learning_rate": 2.4188749014790556e-05,
      "loss": 0.9524,
      "step": 8338
    },
    {
      "epoch": 1.5324289165582605,
      "grad_norm": 0.6551930904388428,
      "learning_rate": 2.416968032208332e-05,
      "loss": 0.9428,
      "step": 8339
    },
    {
      "epoch": 1.5326147556216316,
      "grad_norm": 0.7103485465049744,
      "learning_rate": 2.4150618115307254e-05,
      "loss": 1.0382,
      "step": 8340
    },
    {
      "epoch": 1.5328005946850027,
      "grad_norm": 0.6723774075508118,
      "learning_rate": 2.4131562396092722e-05,
      "loss": 0.9524,
      "step": 8341
    },
    {
      "epoch": 1.5329864337483738,
      "grad_norm": 0.6750916242599487,
      "learning_rate": 2.411251316606966e-05,
      "loss": 0.8356,
      "step": 8342
    },
    {
      "epoch": 1.5331722728117452,
      "grad_norm": 0.7004226446151733,
      "learning_rate": 2.4093470426867326e-05,
      "loss": 0.9496,
      "step": 8343
    },
    {
      "epoch": 1.533358111875116,
      "grad_norm": 0.6192967295646667,
      "learning_rate": 2.4074434180114523e-05,
      "loss": 0.7493,
      "step": 8344
    },
    {
      "epoch": 1.5335439509384874,
      "grad_norm": 0.8744343519210815,
      "learning_rate": 2.4055404427439498e-05,
      "loss": 1.0087,
      "step": 8345
    },
    {
      "epoch": 1.5337297900018583,
      "grad_norm": 0.6814777851104736,
      "learning_rate": 2.4036381170469837e-05,
      "loss": 1.0026,
      "step": 8346
    },
    {
      "epoch": 1.5339156290652296,
      "grad_norm": 0.8172118067741394,
      "learning_rate": 2.4017364410832677e-05,
      "loss": 0.7954,
      "step": 8347
    },
    {
      "epoch": 1.5341014681286005,
      "grad_norm": 0.733949601650238,
      "learning_rate": 2.3998354150154555e-05,
      "loss": 1.0086,
      "step": 8348
    },
    {
      "epoch": 1.5342873071919718,
      "grad_norm": 0.6402868628501892,
      "learning_rate": 2.3979350390061494e-05,
      "loss": 0.6177,
      "step": 8349
    },
    {
      "epoch": 1.534473146255343,
      "grad_norm": 0.671047031879425,
      "learning_rate": 2.396035313217887e-05,
      "loss": 0.7657,
      "step": 8350
    },
    {
      "epoch": 1.534658985318714,
      "grad_norm": 0.8322077989578247,
      "learning_rate": 2.394136237813158e-05,
      "loss": 0.9858,
      "step": 8351
    },
    {
      "epoch": 1.5348448243820851,
      "grad_norm": 0.6607497930526733,
      "learning_rate": 2.3922378129543975e-05,
      "loss": 0.9245,
      "step": 8352
    },
    {
      "epoch": 1.5350306634454562,
      "grad_norm": 0.6483280062675476,
      "learning_rate": 2.3903400388039775e-05,
      "loss": 0.8515,
      "step": 8353
    },
    {
      "epoch": 1.5352165025088274,
      "grad_norm": 0.7380024790763855,
      "learning_rate": 2.3884429155242227e-05,
      "loss": 0.8586,
      "step": 8354
    },
    {
      "epoch": 1.5354023415721985,
      "grad_norm": 0.6636247634887695,
      "learning_rate": 2.3865464432773943e-05,
      "loss": 0.7131,
      "step": 8355
    },
    {
      "epoch": 1.5355881806355696,
      "grad_norm": 0.6486872434616089,
      "learning_rate": 2.384650622225706e-05,
      "loss": 0.8961,
      "step": 8356
    },
    {
      "epoch": 1.5357740196989407,
      "grad_norm": 0.7080100178718567,
      "learning_rate": 2.3827554525313055e-05,
      "loss": 1.1486,
      "step": 8357
    },
    {
      "epoch": 1.535959858762312,
      "grad_norm": 0.7490073442459106,
      "learning_rate": 2.380860934356294e-05,
      "loss": 0.8438,
      "step": 8358
    },
    {
      "epoch": 1.5361456978256829,
      "grad_norm": 0.6263805627822876,
      "learning_rate": 2.3789670678627173e-05,
      "loss": 1.0868,
      "step": 8359
    },
    {
      "epoch": 1.5363315368890542,
      "grad_norm": 0.6247264742851257,
      "learning_rate": 2.377073853212556e-05,
      "loss": 0.7488,
      "step": 8360
    },
    {
      "epoch": 1.536517375952425,
      "grad_norm": 0.805173397064209,
      "learning_rate": 2.3751812905677427e-05,
      "loss": 1.1677,
      "step": 8361
    },
    {
      "epoch": 1.5367032150157964,
      "grad_norm": 0.6882678270339966,
      "learning_rate": 2.3732893800901522e-05,
      "loss": 0.9059,
      "step": 8362
    },
    {
      "epoch": 1.5368890540791673,
      "grad_norm": 0.6723858714103699,
      "learning_rate": 2.3713981219416094e-05,
      "loss": 0.8269,
      "step": 8363
    },
    {
      "epoch": 1.5370748931425386,
      "grad_norm": 0.7437129020690918,
      "learning_rate": 2.3695075162838687e-05,
      "loss": 1.0252,
      "step": 8364
    },
    {
      "epoch": 1.5372607322059095,
      "grad_norm": 0.6647806763648987,
      "learning_rate": 2.3676175632786423e-05,
      "loss": 0.9327,
      "step": 8365
    },
    {
      "epoch": 1.5374465712692809,
      "grad_norm": 0.8653480410575867,
      "learning_rate": 2.3657282630875833e-05,
      "loss": 1.3636,
      "step": 8366
    },
    {
      "epoch": 1.537632410332652,
      "grad_norm": 0.7383580207824707,
      "learning_rate": 2.363839615872284e-05,
      "loss": 0.7505,
      "step": 8367
    },
    {
      "epoch": 1.537818249396023,
      "grad_norm": 0.810131847858429,
      "learning_rate": 2.3619516217942882e-05,
      "loss": 1.0403,
      "step": 8368
    },
    {
      "epoch": 1.5380040884593942,
      "grad_norm": 0.8347377181053162,
      "learning_rate": 2.3600642810150765e-05,
      "loss": 1.0137,
      "step": 8369
    },
    {
      "epoch": 1.5381899275227653,
      "grad_norm": 0.7918209433555603,
      "learning_rate": 2.3581775936960816e-05,
      "loss": 0.8376,
      "step": 8370
    },
    {
      "epoch": 1.5383757665861364,
      "grad_norm": 0.706166684627533,
      "learning_rate": 2.356291559998669e-05,
      "loss": 1.0973,
      "step": 8371
    },
    {
      "epoch": 1.5385616056495075,
      "grad_norm": 0.6235726475715637,
      "learning_rate": 2.3544061800841612e-05,
      "loss": 0.931,
      "step": 8372
    },
    {
      "epoch": 1.5387474447128786,
      "grad_norm": 0.6441529393196106,
      "learning_rate": 2.3525214541138184e-05,
      "loss": 0.9873,
      "step": 8373
    },
    {
      "epoch": 1.5389332837762497,
      "grad_norm": 0.6860111355781555,
      "learning_rate": 2.3506373822488425e-05,
      "loss": 0.7699,
      "step": 8374
    },
    {
      "epoch": 1.539119122839621,
      "grad_norm": 0.7331928014755249,
      "learning_rate": 2.348753964650383e-05,
      "loss": 0.8729,
      "step": 8375
    },
    {
      "epoch": 1.539304961902992,
      "grad_norm": 0.6296803951263428,
      "learning_rate": 2.346871201479537e-05,
      "loss": 0.8982,
      "step": 8376
    },
    {
      "epoch": 1.5394908009663633,
      "grad_norm": 0.7057747840881348,
      "learning_rate": 2.344989092897334e-05,
      "loss": 0.938,
      "step": 8377
    },
    {
      "epoch": 1.5396766400297341,
      "grad_norm": 0.7245149612426758,
      "learning_rate": 2.3431076390647588e-05,
      "loss": 0.878,
      "step": 8378
    },
    {
      "epoch": 1.5398624790931055,
      "grad_norm": 0.684192955493927,
      "learning_rate": 2.341226840142736e-05,
      "loss": 1.0114,
      "step": 8379
    },
    {
      "epoch": 1.5400483181564764,
      "grad_norm": 0.7038612961769104,
      "learning_rate": 2.3393466962921385e-05,
      "loss": 1.0632,
      "step": 8380
    },
    {
      "epoch": 1.5402341572198477,
      "grad_norm": 0.7834692001342773,
      "learning_rate": 2.3374672076737704e-05,
      "loss": 0.9348,
      "step": 8381
    },
    {
      "epoch": 1.5404199962832186,
      "grad_norm": 0.712929904460907,
      "learning_rate": 2.3355883744483964e-05,
      "loss": 0.7863,
      "step": 8382
    },
    {
      "epoch": 1.54060583534659,
      "grad_norm": 0.9249239563941956,
      "learning_rate": 2.3337101967767106e-05,
      "loss": 1.0305,
      "step": 8383
    },
    {
      "epoch": 1.540791674409961,
      "grad_norm": 0.6518446803092957,
      "learning_rate": 2.331832674819363e-05,
      "loss": 0.8175,
      "step": 8384
    },
    {
      "epoch": 1.5409775134733321,
      "grad_norm": 0.7660650014877319,
      "learning_rate": 2.3299558087369376e-05,
      "loss": 0.8638,
      "step": 8385
    },
    {
      "epoch": 1.5411633525367032,
      "grad_norm": 0.6161681413650513,
      "learning_rate": 2.3280795986899695e-05,
      "loss": 0.8675,
      "step": 8386
    },
    {
      "epoch": 1.5413491916000743,
      "grad_norm": 0.6976705193519592,
      "learning_rate": 2.3262040448389367e-05,
      "loss": 0.8803,
      "step": 8387
    },
    {
      "epoch": 1.5415350306634454,
      "grad_norm": 0.683444082736969,
      "learning_rate": 2.324329147344255e-05,
      "loss": 0.972,
      "step": 8388
    },
    {
      "epoch": 1.5417208697268165,
      "grad_norm": 0.7143980264663696,
      "learning_rate": 2.3224549063662927e-05,
      "loss": 1.1016,
      "step": 8389
    },
    {
      "epoch": 1.5419067087901877,
      "grad_norm": 0.7508431077003479,
      "learning_rate": 2.320581322065354e-05,
      "loss": 0.9667,
      "step": 8390
    },
    {
      "epoch": 1.5420925478535588,
      "grad_norm": 0.6859725713729858,
      "learning_rate": 2.3187083946016906e-05,
      "loss": 0.8693,
      "step": 8391
    },
    {
      "epoch": 1.54227838691693,
      "grad_norm": 0.6826413869857788,
      "learning_rate": 2.3168361241355008e-05,
      "loss": 0.89,
      "step": 8392
    },
    {
      "epoch": 1.542464225980301,
      "grad_norm": 0.6446660757064819,
      "learning_rate": 2.3149645108269224e-05,
      "loss": 0.9055,
      "step": 8393
    },
    {
      "epoch": 1.5426500650436723,
      "grad_norm": 0.842805027961731,
      "learning_rate": 2.313093554836041e-05,
      "loss": 0.9147,
      "step": 8394
    },
    {
      "epoch": 1.5428359041070432,
      "grad_norm": 0.7529627084732056,
      "learning_rate": 2.3112232563228798e-05,
      "loss": 1.0043,
      "step": 8395
    },
    {
      "epoch": 1.5430217431704145,
      "grad_norm": 0.6439473628997803,
      "learning_rate": 2.3093536154474137e-05,
      "loss": 0.7722,
      "step": 8396
    },
    {
      "epoch": 1.5432075822337854,
      "grad_norm": 0.6234472990036011,
      "learning_rate": 2.3074846323695522e-05,
      "loss": 1.039,
      "step": 8397
    },
    {
      "epoch": 1.5433934212971567,
      "grad_norm": 0.6632897257804871,
      "learning_rate": 2.3056163072491577e-05,
      "loss": 0.9647,
      "step": 8398
    },
    {
      "epoch": 1.5435792603605278,
      "grad_norm": 0.688230574131012,
      "learning_rate": 2.3037486402460274e-05,
      "loss": 0.9006,
      "step": 8399
    },
    {
      "epoch": 1.543765099423899,
      "grad_norm": 0.7446749210357666,
      "learning_rate": 2.3018816315199097e-05,
      "loss": 0.9976,
      "step": 8400
    },
    {
      "epoch": 1.54395093848727,
      "grad_norm": 0.6614901423454285,
      "learning_rate": 2.3000152812304975e-05,
      "loss": 0.94,
      "step": 8401
    },
    {
      "epoch": 1.5441367775506412,
      "grad_norm": 0.6514410376548767,
      "learning_rate": 2.298149589537416e-05,
      "loss": 0.8233,
      "step": 8402
    },
    {
      "epoch": 1.5443226166140123,
      "grad_norm": 0.8451980352401733,
      "learning_rate": 2.29628455660025e-05,
      "loss": 1.1003,
      "step": 8403
    },
    {
      "epoch": 1.5445084556773834,
      "grad_norm": 0.7540441155433655,
      "learning_rate": 2.2944201825785128e-05,
      "loss": 0.9058,
      "step": 8404
    },
    {
      "epoch": 1.5446942947407545,
      "grad_norm": 0.6781880259513855,
      "learning_rate": 2.2925564676316714e-05,
      "loss": 0.958,
      "step": 8405
    },
    {
      "epoch": 1.5448801338041256,
      "grad_norm": 0.7639658451080322,
      "learning_rate": 2.2906934119191324e-05,
      "loss": 0.9928,
      "step": 8406
    },
    {
      "epoch": 1.5450659728674967,
      "grad_norm": 0.7071417570114136,
      "learning_rate": 2.288831015600249e-05,
      "loss": 0.9922,
      "step": 8407
    },
    {
      "epoch": 1.5452518119308678,
      "grad_norm": 0.7705291509628296,
      "learning_rate": 2.2869692788343166e-05,
      "loss": 0.8055,
      "step": 8408
    },
    {
      "epoch": 1.5454376509942391,
      "grad_norm": 0.6227234601974487,
      "learning_rate": 2.2851082017805703e-05,
      "loss": 0.9943,
      "step": 8409
    },
    {
      "epoch": 1.54562349005761,
      "grad_norm": 0.787907063961029,
      "learning_rate": 2.2832477845981958e-05,
      "loss": 1.0025,
      "step": 8410
    },
    {
      "epoch": 1.5458093291209813,
      "grad_norm": 0.5764409303665161,
      "learning_rate": 2.2813880274463128e-05,
      "loss": 0.7052,
      "step": 8411
    },
    {
      "epoch": 1.5459951681843522,
      "grad_norm": 0.7593272924423218,
      "learning_rate": 2.2795289304839973e-05,
      "loss": 0.8737,
      "step": 8412
    },
    {
      "epoch": 1.5461810072477236,
      "grad_norm": 0.7019006013870239,
      "learning_rate": 2.2776704938702565e-05,
      "loss": 0.9578,
      "step": 8413
    },
    {
      "epoch": 1.5463668463110944,
      "grad_norm": 0.7341148853302002,
      "learning_rate": 2.2758127177640477e-05,
      "loss": 1.0765,
      "step": 8414
    },
    {
      "epoch": 1.5465526853744658,
      "grad_norm": 0.748848557472229,
      "learning_rate": 2.2739556023242738e-05,
      "loss": 0.8449,
      "step": 8415
    },
    {
      "epoch": 1.5467385244378369,
      "grad_norm": 0.6858508586883545,
      "learning_rate": 2.272099147709773e-05,
      "loss": 0.9085,
      "step": 8416
    },
    {
      "epoch": 1.546924363501208,
      "grad_norm": 0.6726716160774231,
      "learning_rate": 2.2702433540793377e-05,
      "loss": 0.9671,
      "step": 8417
    },
    {
      "epoch": 1.547110202564579,
      "grad_norm": 0.7550967335700989,
      "learning_rate": 2.2683882215916907e-05,
      "loss": 1.1665,
      "step": 8418
    },
    {
      "epoch": 1.5472960416279502,
      "grad_norm": 0.898966908454895,
      "learning_rate": 2.2665337504055096e-05,
      "loss": 1.0144,
      "step": 8419
    },
    {
      "epoch": 1.5474818806913213,
      "grad_norm": 0.779742956161499,
      "learning_rate": 2.2646799406794117e-05,
      "loss": 0.9829,
      "step": 8420
    },
    {
      "epoch": 1.5476677197546924,
      "grad_norm": 1.3177577257156372,
      "learning_rate": 2.2628267925719592e-05,
      "loss": 1.559,
      "step": 8421
    },
    {
      "epoch": 1.5478535588180635,
      "grad_norm": 0.7216274738311768,
      "learning_rate": 2.2609743062416498e-05,
      "loss": 0.8658,
      "step": 8422
    },
    {
      "epoch": 1.5480393978814346,
      "grad_norm": 0.7669714689254761,
      "learning_rate": 2.2591224818469347e-05,
      "loss": 0.9549,
      "step": 8423
    },
    {
      "epoch": 1.548225236944806,
      "grad_norm": 0.5377376675605774,
      "learning_rate": 2.257271319546207e-05,
      "loss": 0.7196,
      "step": 8424
    },
    {
      "epoch": 1.5484110760081768,
      "grad_norm": 0.6631278991699219,
      "learning_rate": 2.2554208194977955e-05,
      "loss": 0.7867,
      "step": 8425
    },
    {
      "epoch": 1.5485969150715482,
      "grad_norm": 0.9313716292381287,
      "learning_rate": 2.2535709818599816e-05,
      "loss": 1.1775,
      "step": 8426
    },
    {
      "epoch": 1.548782754134919,
      "grad_norm": 0.7080629467964172,
      "learning_rate": 2.2517218067909807e-05,
      "loss": 0.4182,
      "step": 8427
    },
    {
      "epoch": 1.5489685931982904,
      "grad_norm": 0.6456342339515686,
      "learning_rate": 2.2498732944489596e-05,
      "loss": 0.7099,
      "step": 8428
    },
    {
      "epoch": 1.5491544322616613,
      "grad_norm": 0.7400659322738647,
      "learning_rate": 2.248025444992029e-05,
      "loss": 0.9625,
      "step": 8429
    },
    {
      "epoch": 1.5493402713250326,
      "grad_norm": 1.3561872243881226,
      "learning_rate": 2.246178258578234e-05,
      "loss": 1.3351,
      "step": 8430
    },
    {
      "epoch": 1.5495261103884035,
      "grad_norm": 0.7631298303604126,
      "learning_rate": 2.2443317353655736e-05,
      "loss": 0.8268,
      "step": 8431
    },
    {
      "epoch": 1.5497119494517748,
      "grad_norm": 0.7139994502067566,
      "learning_rate": 2.2424858755119794e-05,
      "loss": 0.9721,
      "step": 8432
    },
    {
      "epoch": 1.549897788515146,
      "grad_norm": 0.6325114965438843,
      "learning_rate": 2.2406406791753364e-05,
      "loss": 0.9429,
      "step": 8433
    },
    {
      "epoch": 1.550083627578517,
      "grad_norm": 0.7448883056640625,
      "learning_rate": 2.238796146513463e-05,
      "loss": 0.8453,
      "step": 8434
    },
    {
      "epoch": 1.5502694666418881,
      "grad_norm": 0.7192485332489014,
      "learning_rate": 2.2369522776841312e-05,
      "loss": 0.9864,
      "step": 8435
    },
    {
      "epoch": 1.5504553057052592,
      "grad_norm": 0.720230758190155,
      "learning_rate": 2.2351090728450486e-05,
      "loss": 0.8496,
      "step": 8436
    },
    {
      "epoch": 1.5506411447686304,
      "grad_norm": 0.7473940849304199,
      "learning_rate": 2.2332665321538682e-05,
      "loss": 0.9731,
      "step": 8437
    },
    {
      "epoch": 1.5508269838320015,
      "grad_norm": 0.7922482490539551,
      "learning_rate": 2.2314246557681905e-05,
      "loss": 1.0866,
      "step": 8438
    },
    {
      "epoch": 1.5510128228953726,
      "grad_norm": 0.6319082379341125,
      "learning_rate": 2.22958344384555e-05,
      "loss": 0.892,
      "step": 8439
    },
    {
      "epoch": 1.5511986619587437,
      "grad_norm": 0.6750452518463135,
      "learning_rate": 2.2277428965434344e-05,
      "loss": 0.842,
      "step": 8440
    },
    {
      "epoch": 1.551384501022115,
      "grad_norm": 0.6215194463729858,
      "learning_rate": 2.2259030140192637e-05,
      "loss": 0.714,
      "step": 8441
    },
    {
      "epoch": 1.5515703400854859,
      "grad_norm": 0.6646790504455566,
      "learning_rate": 2.22406379643041e-05,
      "loss": 0.9337,
      "step": 8442
    },
    {
      "epoch": 1.5517561791488572,
      "grad_norm": 0.5718748569488525,
      "learning_rate": 2.2222252439341896e-05,
      "loss": 0.885,
      "step": 8443
    },
    {
      "epoch": 1.551942018212228,
      "grad_norm": 0.7178288102149963,
      "learning_rate": 2.220387356687851e-05,
      "loss": 1.1043,
      "step": 8444
    },
    {
      "epoch": 1.5521278572755994,
      "grad_norm": 0.8014031648635864,
      "learning_rate": 2.2185501348485972e-05,
      "loss": 0.8161,
      "step": 8445
    },
    {
      "epoch": 1.5523136963389703,
      "grad_norm": 0.7333347201347351,
      "learning_rate": 2.2167135785735672e-05,
      "loss": 0.8641,
      "step": 8446
    },
    {
      "epoch": 1.5524995354023416,
      "grad_norm": 0.7811396718025208,
      "learning_rate": 2.2148776880198487e-05,
      "loss": 1.0327,
      "step": 8447
    },
    {
      "epoch": 1.5526853744657125,
      "grad_norm": 0.6206461191177368,
      "learning_rate": 2.2130424633444645e-05,
      "loss": 0.9963,
      "step": 8448
    },
    {
      "epoch": 1.5528712135290839,
      "grad_norm": 0.8549916744232178,
      "learning_rate": 2.211207904704389e-05,
      "loss": 0.8474,
      "step": 8449
    },
    {
      "epoch": 1.553057052592455,
      "grad_norm": 0.7330931425094604,
      "learning_rate": 2.2093740122565343e-05,
      "loss": 1.0992,
      "step": 8450
    },
    {
      "epoch": 1.553242891655826,
      "grad_norm": 0.8338735103607178,
      "learning_rate": 2.2075407861577578e-05,
      "loss": 0.9663,
      "step": 8451
    },
    {
      "epoch": 1.5534287307191972,
      "grad_norm": 0.6526317000389099,
      "learning_rate": 2.205708226564863e-05,
      "loss": 0.8416,
      "step": 8452
    },
    {
      "epoch": 1.5536145697825683,
      "grad_norm": 0.7737939357757568,
      "learning_rate": 2.203876333634587e-05,
      "loss": 1.0954,
      "step": 8453
    },
    {
      "epoch": 1.5538004088459394,
      "grad_norm": 0.6432002186775208,
      "learning_rate": 2.2020451075236193e-05,
      "loss": 0.7378,
      "step": 8454
    },
    {
      "epoch": 1.5539862479093105,
      "grad_norm": 0.6858252882957458,
      "learning_rate": 2.200214548388586e-05,
      "loss": 1.2002,
      "step": 8455
    },
    {
      "epoch": 1.5541720869726816,
      "grad_norm": 0.8318917155265808,
      "learning_rate": 2.1983846563860587e-05,
      "loss": 0.8824,
      "step": 8456
    },
    {
      "epoch": 1.5543579260360527,
      "grad_norm": 0.8037974238395691,
      "learning_rate": 2.1965554316725568e-05,
      "loss": 1.1095,
      "step": 8457
    },
    {
      "epoch": 1.554543765099424,
      "grad_norm": 1.7428077459335327,
      "learning_rate": 2.1947268744045303e-05,
      "loss": 1.7246,
      "step": 8458
    },
    {
      "epoch": 1.554729604162795,
      "grad_norm": 0.6576951146125793,
      "learning_rate": 2.1928989847383873e-05,
      "loss": 1.05,
      "step": 8459
    },
    {
      "epoch": 1.5549154432261663,
      "grad_norm": 0.7942461371421814,
      "learning_rate": 2.1910717628304656e-05,
      "loss": 0.8262,
      "step": 8460
    },
    {
      "epoch": 1.5551012822895371,
      "grad_norm": 0.7093661427497864,
      "learning_rate": 2.189245208837054e-05,
      "loss": 0.9478,
      "step": 8461
    },
    {
      "epoch": 1.5552871213529085,
      "grad_norm": 0.7666981816291809,
      "learning_rate": 2.1874193229143804e-05,
      "loss": 0.9724,
      "step": 8462
    },
    {
      "epoch": 1.5554729604162794,
      "grad_norm": 0.7784303426742554,
      "learning_rate": 2.1855941052186158e-05,
      "loss": 1.0842,
      "step": 8463
    },
    {
      "epoch": 1.5556587994796507,
      "grad_norm": 0.7566025853157043,
      "learning_rate": 2.183769555905877e-05,
      "loss": 0.9761,
      "step": 8464
    },
    {
      "epoch": 1.5558446385430218,
      "grad_norm": 0.6337560415267944,
      "learning_rate": 2.181945675132221e-05,
      "loss": 0.7571,
      "step": 8465
    },
    {
      "epoch": 1.556030477606393,
      "grad_norm": 0.8436826467514038,
      "learning_rate": 2.180122463053652e-05,
      "loss": 1.0056,
      "step": 8466
    },
    {
      "epoch": 1.556216316669764,
      "grad_norm": 0.7065210342407227,
      "learning_rate": 2.1782999198261057e-05,
      "loss": 1.0622,
      "step": 8467
    },
    {
      "epoch": 1.5564021557331351,
      "grad_norm": 0.6758416295051575,
      "learning_rate": 2.1764780456054757e-05,
      "loss": 0.9907,
      "step": 8468
    },
    {
      "epoch": 1.5565879947965062,
      "grad_norm": 0.6660606265068054,
      "learning_rate": 2.1746568405475843e-05,
      "loss": 0.8577,
      "step": 8469
    },
    {
      "epoch": 1.5567738338598773,
      "grad_norm": 0.7996534705162048,
      "learning_rate": 2.1728363048082068e-05,
      "loss": 0.9277,
      "step": 8470
    },
    {
      "epoch": 1.5569596729232484,
      "grad_norm": 0.8163617253303528,
      "learning_rate": 2.171016438543059e-05,
      "loss": 1.0313,
      "step": 8471
    },
    {
      "epoch": 1.5571455119866195,
      "grad_norm": 0.6959372758865356,
      "learning_rate": 2.169197241907793e-05,
      "loss": 1.0311,
      "step": 8472
    },
    {
      "epoch": 1.5573313510499907,
      "grad_norm": 0.8374831080436707,
      "learning_rate": 2.167378715058015e-05,
      "loss": 1.03,
      "step": 8473
    },
    {
      "epoch": 1.5575171901133618,
      "grad_norm": 0.6212435960769653,
      "learning_rate": 2.1655608581492603e-05,
      "loss": 1.0169,
      "step": 8474
    },
    {
      "epoch": 1.557703029176733,
      "grad_norm": 0.8012619614601135,
      "learning_rate": 2.163743671337021e-05,
      "loss": 1.0254,
      "step": 8475
    },
    {
      "epoch": 1.557888868240104,
      "grad_norm": 0.7814256548881531,
      "learning_rate": 2.1619271547767193e-05,
      "loss": 1.1478,
      "step": 8476
    },
    {
      "epoch": 1.5580747073034753,
      "grad_norm": 0.6881610751152039,
      "learning_rate": 2.160111308623728e-05,
      "loss": 0.8462,
      "step": 8477
    },
    {
      "epoch": 1.5582605463668462,
      "grad_norm": 0.7728042602539062,
      "learning_rate": 2.158296133033364e-05,
      "loss": 1.1767,
      "step": 8478
    },
    {
      "epoch": 1.5584463854302175,
      "grad_norm": 0.7264332175254822,
      "learning_rate": 2.156481628160877e-05,
      "loss": 0.8743,
      "step": 8479
    },
    {
      "epoch": 1.5586322244935884,
      "grad_norm": 0.5601229667663574,
      "learning_rate": 2.1546677941614678e-05,
      "loss": 0.8995,
      "step": 8480
    },
    {
      "epoch": 1.5588180635569597,
      "grad_norm": 0.7997213006019592,
      "learning_rate": 2.1528546311902788e-05,
      "loss": 0.9521,
      "step": 8481
    },
    {
      "epoch": 1.5590039026203308,
      "grad_norm": 0.6687659025192261,
      "learning_rate": 2.151042139402395e-05,
      "loss": 0.9774,
      "step": 8482
    },
    {
      "epoch": 1.559189741683702,
      "grad_norm": 0.6728366017341614,
      "learning_rate": 2.1492303189528395e-05,
      "loss": 0.8514,
      "step": 8483
    },
    {
      "epoch": 1.559375580747073,
      "grad_norm": 0.621809720993042,
      "learning_rate": 2.147419169996582e-05,
      "loss": 0.8785,
      "step": 8484
    },
    {
      "epoch": 1.5595614198104442,
      "grad_norm": 0.8290244340896606,
      "learning_rate": 2.1456086926885367e-05,
      "loss": 1.2388,
      "step": 8485
    },
    {
      "epoch": 1.5597472588738153,
      "grad_norm": 0.6844221353530884,
      "learning_rate": 2.1437988871835523e-05,
      "loss": 1.0081,
      "step": 8486
    },
    {
      "epoch": 1.5599330979371864,
      "grad_norm": 0.8001043796539307,
      "learning_rate": 2.1419897536364332e-05,
      "loss": 1.0609,
      "step": 8487
    },
    {
      "epoch": 1.5601189370005575,
      "grad_norm": 0.7094721794128418,
      "learning_rate": 2.1401812922019092e-05,
      "loss": 0.9187,
      "step": 8488
    },
    {
      "epoch": 1.5603047760639286,
      "grad_norm": 0.6323491930961609,
      "learning_rate": 2.1383735030346706e-05,
      "loss": 0.7464,
      "step": 8489
    },
    {
      "epoch": 1.5604906151273,
      "grad_norm": 0.5895442962646484,
      "learning_rate": 2.1365663862893348e-05,
      "loss": 0.7452,
      "step": 8490
    },
    {
      "epoch": 1.5606764541906708,
      "grad_norm": 0.6722056269645691,
      "learning_rate": 2.13475994212047e-05,
      "loss": 0.9545,
      "step": 8491
    },
    {
      "epoch": 1.5608622932540421,
      "grad_norm": 0.7342957854270935,
      "learning_rate": 2.1329541706825896e-05,
      "loss": 0.9049,
      "step": 8492
    },
    {
      "epoch": 1.561048132317413,
      "grad_norm": 0.9032208919525146,
      "learning_rate": 2.1311490721301396e-05,
      "loss": 0.7619,
      "step": 8493
    },
    {
      "epoch": 1.5612339713807843,
      "grad_norm": 0.5408475399017334,
      "learning_rate": 2.129344646617516e-05,
      "loss": 0.4899,
      "step": 8494
    },
    {
      "epoch": 1.5614198104441552,
      "grad_norm": 0.7296645641326904,
      "learning_rate": 2.127540894299056e-05,
      "loss": 0.5888,
      "step": 8495
    },
    {
      "epoch": 1.5616056495075266,
      "grad_norm": 0.7951775193214417,
      "learning_rate": 2.1257378153290375e-05,
      "loss": 1.2182,
      "step": 8496
    },
    {
      "epoch": 1.5617914885708974,
      "grad_norm": 0.6165066361427307,
      "learning_rate": 2.1239354098616847e-05,
      "loss": 1.0031,
      "step": 8497
    },
    {
      "epoch": 1.5619773276342688,
      "grad_norm": 0.6538174748420715,
      "learning_rate": 2.122133678051156e-05,
      "loss": 0.9143,
      "step": 8498
    },
    {
      "epoch": 1.5621631666976399,
      "grad_norm": 0.7776987552642822,
      "learning_rate": 2.1203326200515628e-05,
      "loss": 1.191,
      "step": 8499
    },
    {
      "epoch": 1.562349005761011,
      "grad_norm": 0.7507323026657104,
      "learning_rate": 2.118532236016948e-05,
      "loss": 1.0686,
      "step": 8500
    },
    {
      "epoch": 1.562534844824382,
      "grad_norm": 0.6666802763938904,
      "learning_rate": 2.1167325261013082e-05,
      "loss": 0.8877,
      "step": 8501
    },
    {
      "epoch": 1.5627206838877532,
      "grad_norm": 0.5777565240859985,
      "learning_rate": 2.1149334904585706e-05,
      "loss": 0.8839,
      "step": 8502
    },
    {
      "epoch": 1.5629065229511243,
      "grad_norm": 0.8229918479919434,
      "learning_rate": 2.113135129242614e-05,
      "loss": 0.8825,
      "step": 8503
    },
    {
      "epoch": 1.5630923620144954,
      "grad_norm": 0.6364744305610657,
      "learning_rate": 2.1113374426072574e-05,
      "loss": 0.9031,
      "step": 8504
    },
    {
      "epoch": 1.5632782010778665,
      "grad_norm": 0.6380570530891418,
      "learning_rate": 2.109540430706256e-05,
      "loss": 0.7958,
      "step": 8505
    },
    {
      "epoch": 1.5634640401412376,
      "grad_norm": 0.7195653319358826,
      "learning_rate": 2.107744093693319e-05,
      "loss": 0.9536,
      "step": 8506
    },
    {
      "epoch": 1.563649879204609,
      "grad_norm": 0.7349860072135925,
      "learning_rate": 2.1059484317220835e-05,
      "loss": 1.0266,
      "step": 8507
    },
    {
      "epoch": 1.5638357182679798,
      "grad_norm": 0.7431777119636536,
      "learning_rate": 2.104153444946141e-05,
      "loss": 0.9168,
      "step": 8508
    },
    {
      "epoch": 1.5640215573313512,
      "grad_norm": 0.7292587757110596,
      "learning_rate": 2.1023591335190184e-05,
      "loss": 0.9093,
      "step": 8509
    },
    {
      "epoch": 1.564207396394722,
      "grad_norm": 0.6694021224975586,
      "learning_rate": 2.100565497594188e-05,
      "loss": 0.9558,
      "step": 8510
    },
    {
      "epoch": 1.5643932354580934,
      "grad_norm": 0.6532893776893616,
      "learning_rate": 2.0987725373250666e-05,
      "loss": 0.934,
      "step": 8511
    },
    {
      "epoch": 1.5645790745214643,
      "grad_norm": 0.7578188180923462,
      "learning_rate": 2.096980252865005e-05,
      "loss": 1.0388,
      "step": 8512
    },
    {
      "epoch": 1.5647649135848356,
      "grad_norm": 0.7167041301727295,
      "learning_rate": 2.0951886443673042e-05,
      "loss": 0.9336,
      "step": 8513
    },
    {
      "epoch": 1.5649507526482065,
      "grad_norm": 0.7971643209457397,
      "learning_rate": 2.093397711985202e-05,
      "loss": 1.1815,
      "step": 8514
    },
    {
      "epoch": 1.5651365917115778,
      "grad_norm": 0.8955301642417908,
      "learning_rate": 2.0916074558718836e-05,
      "loss": 1.2098,
      "step": 8515
    },
    {
      "epoch": 1.565322430774949,
      "grad_norm": 0.6152545809745789,
      "learning_rate": 2.0898178761804698e-05,
      "loss": 1.0297,
      "step": 8516
    },
    {
      "epoch": 1.56550826983832,
      "grad_norm": 0.6585174202919006,
      "learning_rate": 2.088028973064028e-05,
      "loss": 0.6759,
      "step": 8517
    },
    {
      "epoch": 1.5656941089016911,
      "grad_norm": 0.7088993191719055,
      "learning_rate": 2.0862407466755718e-05,
      "loss": 1.0041,
      "step": 8518
    },
    {
      "epoch": 1.5658799479650622,
      "grad_norm": 0.7710633277893066,
      "learning_rate": 2.0844531971680457e-05,
      "loss": 1.0349,
      "step": 8519
    },
    {
      "epoch": 1.5660657870284334,
      "grad_norm": 0.7347123622894287,
      "learning_rate": 2.0826663246943478e-05,
      "loss": 0.8464,
      "step": 8520
    },
    {
      "epoch": 1.5662516260918045,
      "grad_norm": 0.635280191898346,
      "learning_rate": 2.0808801294073087e-05,
      "loss": 0.9236,
      "step": 8521
    },
    {
      "epoch": 1.5664374651551756,
      "grad_norm": 0.8302204608917236,
      "learning_rate": 2.0790946114597066e-05,
      "loss": 1.0667,
      "step": 8522
    },
    {
      "epoch": 1.5666233042185467,
      "grad_norm": 0.8665578961372375,
      "learning_rate": 2.0773097710042655e-05,
      "loss": 1.2012,
      "step": 8523
    },
    {
      "epoch": 1.566809143281918,
      "grad_norm": 0.722381055355072,
      "learning_rate": 2.075525608193639e-05,
      "loss": 0.9888,
      "step": 8524
    },
    {
      "epoch": 1.566994982345289,
      "grad_norm": 0.699985682964325,
      "learning_rate": 2.0737421231804356e-05,
      "loss": 0.8727,
      "step": 8525
    },
    {
      "epoch": 1.5671808214086602,
      "grad_norm": 0.5751911997795105,
      "learning_rate": 2.071959316117198e-05,
      "loss": 0.7623,
      "step": 8526
    },
    {
      "epoch": 1.567366660472031,
      "grad_norm": 0.6195750832557678,
      "learning_rate": 2.070177187156418e-05,
      "loss": 0.9094,
      "step": 8527
    },
    {
      "epoch": 1.5675524995354024,
      "grad_norm": 0.6845465898513794,
      "learning_rate": 2.0683957364505202e-05,
      "loss": 0.9303,
      "step": 8528
    },
    {
      "epoch": 1.5677383385987733,
      "grad_norm": 0.7471318244934082,
      "learning_rate": 2.0666149641518782e-05,
      "loss": 1.1109,
      "step": 8529
    },
    {
      "epoch": 1.5679241776621446,
      "grad_norm": 0.6891338229179382,
      "learning_rate": 2.064834870412804e-05,
      "loss": 0.9626,
      "step": 8530
    },
    {
      "epoch": 1.5681100167255158,
      "grad_norm": 0.6418676376342773,
      "learning_rate": 2.0630554553855518e-05,
      "loss": 0.8029,
      "step": 8531
    },
    {
      "epoch": 1.5682958557888869,
      "grad_norm": 0.6934593915939331,
      "learning_rate": 2.0612767192223238e-05,
      "loss": 1.1254,
      "step": 8532
    },
    {
      "epoch": 1.568481694852258,
      "grad_norm": 0.893159806728363,
      "learning_rate": 2.059498662075253e-05,
      "loss": 1.0269,
      "step": 8533
    },
    {
      "epoch": 1.568667533915629,
      "grad_norm": 0.6825253963470459,
      "learning_rate": 2.0577212840964267e-05,
      "loss": 1.0025,
      "step": 8534
    },
    {
      "epoch": 1.5688533729790002,
      "grad_norm": 0.6435436606407166,
      "learning_rate": 2.055944585437861e-05,
      "loss": 0.8664,
      "step": 8535
    },
    {
      "epoch": 1.5690392120423713,
      "grad_norm": 0.789273738861084,
      "learning_rate": 2.0541685662515276e-05,
      "loss": 1.0906,
      "step": 8536
    },
    {
      "epoch": 1.5692250511057424,
      "grad_norm": 0.8879076838493347,
      "learning_rate": 2.0523932266893264e-05,
      "loss": 1.1946,
      "step": 8537
    },
    {
      "epoch": 1.5694108901691135,
      "grad_norm": 0.6618274450302124,
      "learning_rate": 2.0506185669031107e-05,
      "loss": 1.1427,
      "step": 8538
    },
    {
      "epoch": 1.5695967292324848,
      "grad_norm": 0.855046272277832,
      "learning_rate": 2.0488445870446695e-05,
      "loss": 0.8346,
      "step": 8539
    },
    {
      "epoch": 1.5697825682958557,
      "grad_norm": 0.8858674764633179,
      "learning_rate": 2.047071287265735e-05,
      "loss": 1.101,
      "step": 8540
    },
    {
      "epoch": 1.569968407359227,
      "grad_norm": 0.732211709022522,
      "learning_rate": 2.0452986677179843e-05,
      "loss": 1.0229,
      "step": 8541
    },
    {
      "epoch": 1.570154246422598,
      "grad_norm": 0.7777327299118042,
      "learning_rate": 2.043526728553028e-05,
      "loss": 0.854,
      "step": 8542
    },
    {
      "epoch": 1.5703400854859693,
      "grad_norm": 0.6771265268325806,
      "learning_rate": 2.0417554699224305e-05,
      "loss": 0.927,
      "step": 8543
    },
    {
      "epoch": 1.5705259245493401,
      "grad_norm": 0.8816447257995605,
      "learning_rate": 2.0399848919776844e-05,
      "loss": 1.3677,
      "step": 8544
    },
    {
      "epoch": 1.5707117636127115,
      "grad_norm": 0.5746204257011414,
      "learning_rate": 2.0382149948702346e-05,
      "loss": 0.8165,
      "step": 8545
    },
    {
      "epoch": 1.5708976026760824,
      "grad_norm": 0.8437845706939697,
      "learning_rate": 2.0364457787514666e-05,
      "loss": 1.0448,
      "step": 8546
    },
    {
      "epoch": 1.5710834417394537,
      "grad_norm": 0.8124422430992126,
      "learning_rate": 2.0346772437727003e-05,
      "loss": 0.8906,
      "step": 8547
    },
    {
      "epoch": 1.5712692808028248,
      "grad_norm": 0.7039830088615417,
      "learning_rate": 2.032909390085207e-05,
      "loss": 0.839,
      "step": 8548
    },
    {
      "epoch": 1.571455119866196,
      "grad_norm": 0.7771579623222351,
      "learning_rate": 2.0311422178401907e-05,
      "loss": 1.0007,
      "step": 8549
    },
    {
      "epoch": 1.571640958929567,
      "grad_norm": 0.6714681386947632,
      "learning_rate": 2.0293757271888046e-05,
      "loss": 1.0287,
      "step": 8550
    },
    {
      "epoch": 1.5718267979929381,
      "grad_norm": 0.7209590077400208,
      "learning_rate": 2.0276099182821383e-05,
      "loss": 0.8024,
      "step": 8551
    },
    {
      "epoch": 1.5720126370563092,
      "grad_norm": 0.7411814332008362,
      "learning_rate": 2.0258447912712252e-05,
      "loss": 1.0593,
      "step": 8552
    },
    {
      "epoch": 1.5721984761196803,
      "grad_norm": 0.7381874918937683,
      "learning_rate": 2.0240803463070425e-05,
      "loss": 0.9287,
      "step": 8553
    },
    {
      "epoch": 1.5723843151830514,
      "grad_norm": 0.6632939577102661,
      "learning_rate": 2.0223165835405046e-05,
      "loss": 0.9132,
      "step": 8554
    },
    {
      "epoch": 1.5725701542464225,
      "grad_norm": 0.7423511743545532,
      "learning_rate": 2.0205535031224752e-05,
      "loss": 0.8735,
      "step": 8555
    },
    {
      "epoch": 1.5727559933097939,
      "grad_norm": 0.7011168003082275,
      "learning_rate": 2.018791105203748e-05,
      "loss": 0.7841,
      "step": 8556
    },
    {
      "epoch": 1.5729418323731648,
      "grad_norm": 0.8723081350326538,
      "learning_rate": 2.017029389935069e-05,
      "loss": 1.0542,
      "step": 8557
    },
    {
      "epoch": 1.573127671436536,
      "grad_norm": 0.7361454963684082,
      "learning_rate": 2.015268357467117e-05,
      "loss": 0.8223,
      "step": 8558
    },
    {
      "epoch": 1.573313510499907,
      "grad_norm": 0.8173145055770874,
      "learning_rate": 2.0135080079505208e-05,
      "loss": 0.9669,
      "step": 8559
    },
    {
      "epoch": 1.5734993495632783,
      "grad_norm": 1.3938632011413574,
      "learning_rate": 2.0117483415358474e-05,
      "loss": 1.2551,
      "step": 8560
    },
    {
      "epoch": 1.5736851886266492,
      "grad_norm": 0.7503166198730469,
      "learning_rate": 2.0099893583736006e-05,
      "loss": 0.9042,
      "step": 8561
    },
    {
      "epoch": 1.5738710276900205,
      "grad_norm": 0.8426245450973511,
      "learning_rate": 2.008231058614236e-05,
      "loss": 0.8681,
      "step": 8562
    },
    {
      "epoch": 1.5740568667533914,
      "grad_norm": 0.6863853931427002,
      "learning_rate": 2.0064734424081388e-05,
      "loss": 1.0267,
      "step": 8563
    },
    {
      "epoch": 1.5742427058167627,
      "grad_norm": 0.6541019678115845,
      "learning_rate": 2.0047165099056466e-05,
      "loss": 0.943,
      "step": 8564
    },
    {
      "epoch": 1.5744285448801338,
      "grad_norm": 0.6452354788780212,
      "learning_rate": 2.0029602612570285e-05,
      "loss": 0.852,
      "step": 8565
    },
    {
      "epoch": 1.574614383943505,
      "grad_norm": 0.7347366809844971,
      "learning_rate": 2.001204696612503e-05,
      "loss": 0.8893,
      "step": 8566
    },
    {
      "epoch": 1.574800223006876,
      "grad_norm": 0.7106027007102966,
      "learning_rate": 1.9994498161222287e-05,
      "loss": 1.0119,
      "step": 8567
    },
    {
      "epoch": 1.5749860620702472,
      "grad_norm": 0.7813401818275452,
      "learning_rate": 1.997695619936305e-05,
      "loss": 0.9425,
      "step": 8568
    },
    {
      "epoch": 1.5751719011336183,
      "grad_norm": 0.8719832897186279,
      "learning_rate": 1.995942108204767e-05,
      "loss": 0.9462,
      "step": 8569
    },
    {
      "epoch": 1.5753577401969894,
      "grad_norm": 0.6365001201629639,
      "learning_rate": 1.9941892810775998e-05,
      "loss": 1.0615,
      "step": 8570
    },
    {
      "epoch": 1.5755435792603605,
      "grad_norm": 0.7773917317390442,
      "learning_rate": 1.9924371387047293e-05,
      "loss": 0.9165,
      "step": 8571
    },
    {
      "epoch": 1.5757294183237316,
      "grad_norm": 0.7087615132331848,
      "learning_rate": 1.9906856812360142e-05,
      "loss": 1.1179,
      "step": 8572
    },
    {
      "epoch": 1.575915257387103,
      "grad_norm": 0.6763791441917419,
      "learning_rate": 1.9889349088212627e-05,
      "loss": 0.8996,
      "step": 8573
    },
    {
      "epoch": 1.5761010964504738,
      "grad_norm": 0.722274124622345,
      "learning_rate": 1.9871848216102262e-05,
      "loss": 0.8501,
      "step": 8574
    },
    {
      "epoch": 1.5762869355138451,
      "grad_norm": 0.7045106887817383,
      "learning_rate": 1.985435419752586e-05,
      "loss": 0.8928,
      "step": 8575
    },
    {
      "epoch": 1.576472774577216,
      "grad_norm": 0.6729766726493835,
      "learning_rate": 1.98368670339798e-05,
      "loss": 1.0161,
      "step": 8576
    },
    {
      "epoch": 1.5766586136405873,
      "grad_norm": 0.7530122399330139,
      "learning_rate": 1.9819386726959722e-05,
      "loss": 0.9384,
      "step": 8577
    },
    {
      "epoch": 1.5768444527039582,
      "grad_norm": 0.7287601232528687,
      "learning_rate": 1.9801913277960827e-05,
      "loss": 0.9684,
      "step": 8578
    },
    {
      "epoch": 1.5770302917673296,
      "grad_norm": 0.6725710034370422,
      "learning_rate": 1.9784446688477587e-05,
      "loss": 0.8374,
      "step": 8579
    },
    {
      "epoch": 1.5772161308307004,
      "grad_norm": 0.8536375164985657,
      "learning_rate": 1.9766986960003985e-05,
      "loss": 1.0045,
      "step": 8580
    },
    {
      "epoch": 1.5774019698940718,
      "grad_norm": 0.6646410226821899,
      "learning_rate": 1.9749534094033428e-05,
      "loss": 0.7725,
      "step": 8581
    },
    {
      "epoch": 1.5775878089574429,
      "grad_norm": 0.73421710729599,
      "learning_rate": 1.973208809205863e-05,
      "loss": 0.8398,
      "step": 8582
    },
    {
      "epoch": 1.577773648020814,
      "grad_norm": 0.6640996336936951,
      "learning_rate": 1.9714648955571834e-05,
      "loss": 0.8321,
      "step": 8583
    },
    {
      "epoch": 1.577959487084185,
      "grad_norm": 0.7355279922485352,
      "learning_rate": 1.9697216686064624e-05,
      "loss": 1.1926,
      "step": 8584
    },
    {
      "epoch": 1.5781453261475562,
      "grad_norm": 0.892637312412262,
      "learning_rate": 1.967979128502805e-05,
      "loss": 0.9099,
      "step": 8585
    },
    {
      "epoch": 1.5783311652109273,
      "grad_norm": 0.6400608420372009,
      "learning_rate": 1.9662372753952497e-05,
      "loss": 0.8335,
      "step": 8586
    },
    {
      "epoch": 1.5785170042742984,
      "grad_norm": 0.8645890355110168,
      "learning_rate": 1.964496109432784e-05,
      "loss": 0.9008,
      "step": 8587
    },
    {
      "epoch": 1.5787028433376695,
      "grad_norm": 0.6955536007881165,
      "learning_rate": 1.9627556307643357e-05,
      "loss": 0.847,
      "step": 8588
    },
    {
      "epoch": 1.5788886824010406,
      "grad_norm": 0.7045804262161255,
      "learning_rate": 1.9610158395387657e-05,
      "loss": 0.944,
      "step": 8589
    },
    {
      "epoch": 1.579074521464412,
      "grad_norm": 0.642744243144989,
      "learning_rate": 1.959276735904889e-05,
      "loss": 0.9423,
      "step": 8590
    },
    {
      "epoch": 1.5792603605277828,
      "grad_norm": 0.758800745010376,
      "learning_rate": 1.9575383200114484e-05,
      "loss": 1.1133,
      "step": 8591
    },
    {
      "epoch": 1.5794461995911542,
      "grad_norm": 0.8545302152633667,
      "learning_rate": 1.9558005920071397e-05,
      "loss": 1.1038,
      "step": 8592
    },
    {
      "epoch": 1.579632038654525,
      "grad_norm": 0.6158139705657959,
      "learning_rate": 1.9540635520405902e-05,
      "loss": 0.9974,
      "step": 8593
    },
    {
      "epoch": 1.5798178777178964,
      "grad_norm": 0.6586875319480896,
      "learning_rate": 1.9523272002603742e-05,
      "loss": 0.7796,
      "step": 8594
    },
    {
      "epoch": 1.5800037167812673,
      "grad_norm": 0.6694611310958862,
      "learning_rate": 1.9505915368150084e-05,
      "loss": 1.0665,
      "step": 8595
    },
    {
      "epoch": 1.5801895558446386,
      "grad_norm": 0.7861499190330505,
      "learning_rate": 1.948856561852943e-05,
      "loss": 1.2053,
      "step": 8596
    },
    {
      "epoch": 1.5803753949080097,
      "grad_norm": 0.7392147779464722,
      "learning_rate": 1.947122275522577e-05,
      "loss": 0.8328,
      "step": 8597
    },
    {
      "epoch": 1.5805612339713808,
      "grad_norm": 3.2451038360595703,
      "learning_rate": 1.9453886779722473e-05,
      "loss": 1.4786,
      "step": 8598
    },
    {
      "epoch": 1.580747073034752,
      "grad_norm": 0.6445174813270569,
      "learning_rate": 1.9436557693502356e-05,
      "loss": 0.7439,
      "step": 8599
    },
    {
      "epoch": 1.580932912098123,
      "grad_norm": 0.9785395860671997,
      "learning_rate": 1.9419235498047537e-05,
      "loss": 0.9491,
      "step": 8600
    },
    {
      "epoch": 1.5811187511614941,
      "grad_norm": 0.7265476584434509,
      "learning_rate": 1.9401920194839675e-05,
      "loss": 0.9351,
      "step": 8601
    },
    {
      "epoch": 1.5813045902248652,
      "grad_norm": 0.7562678456306458,
      "learning_rate": 1.9384611785359795e-05,
      "loss": 0.9766,
      "step": 8602
    },
    {
      "epoch": 1.5814904292882364,
      "grad_norm": 0.6438634395599365,
      "learning_rate": 1.936731027108828e-05,
      "loss": 0.805,
      "step": 8603
    },
    {
      "epoch": 1.5816762683516075,
      "grad_norm": 0.6853559613227844,
      "learning_rate": 1.9350015653505015e-05,
      "loss": 0.9047,
      "step": 8604
    },
    {
      "epoch": 1.5818621074149788,
      "grad_norm": 0.8060786724090576,
      "learning_rate": 1.933272793408919e-05,
      "loss": 1.1083,
      "step": 8605
    },
    {
      "epoch": 1.5820479464783497,
      "grad_norm": 0.5869574546813965,
      "learning_rate": 1.9315447114319517e-05,
      "loss": 0.7251,
      "step": 8606
    },
    {
      "epoch": 1.582233785541721,
      "grad_norm": 0.7552639842033386,
      "learning_rate": 1.9298173195674007e-05,
      "loss": 0.935,
      "step": 8607
    },
    {
      "epoch": 1.582419624605092,
      "grad_norm": 0.7403124570846558,
      "learning_rate": 1.9280906179630165e-05,
      "loss": 1.0466,
      "step": 8608
    },
    {
      "epoch": 1.5826054636684632,
      "grad_norm": 0.753761887550354,
      "learning_rate": 1.9263646067664897e-05,
      "loss": 0.8971,
      "step": 8609
    },
    {
      "epoch": 1.582791302731834,
      "grad_norm": 0.7540832161903381,
      "learning_rate": 1.924639286125446e-05,
      "loss": 0.8634,
      "step": 8610
    },
    {
      "epoch": 1.5829771417952054,
      "grad_norm": 0.6291332840919495,
      "learning_rate": 1.9229146561874557e-05,
      "loss": 0.9713,
      "step": 8611
    },
    {
      "epoch": 1.5831629808585763,
      "grad_norm": 0.6404367089271545,
      "learning_rate": 1.921190717100032e-05,
      "loss": 0.8434,
      "step": 8612
    },
    {
      "epoch": 1.5833488199219476,
      "grad_norm": 0.7628955245018005,
      "learning_rate": 1.9194674690106307e-05,
      "loss": 0.8384,
      "step": 8613
    },
    {
      "epoch": 1.5835346589853188,
      "grad_norm": 0.7761968970298767,
      "learning_rate": 1.917744912066638e-05,
      "loss": 1.0999,
      "step": 8614
    },
    {
      "epoch": 1.5837204980486899,
      "grad_norm": 0.7242310047149658,
      "learning_rate": 1.9160230464153905e-05,
      "loss": 1.045,
      "step": 8615
    },
    {
      "epoch": 1.583906337112061,
      "grad_norm": 0.8341673612594604,
      "learning_rate": 1.914301872204166e-05,
      "loss": 0.967,
      "step": 8616
    },
    {
      "epoch": 1.584092176175432,
      "grad_norm": 0.799604058265686,
      "learning_rate": 1.9125813895801758e-05,
      "loss": 0.8067,
      "step": 8617
    },
    {
      "epoch": 1.5842780152388032,
      "grad_norm": 0.7231751084327698,
      "learning_rate": 1.9108615986905808e-05,
      "loss": 1.1896,
      "step": 8618
    },
    {
      "epoch": 1.5844638543021743,
      "grad_norm": 0.7168472409248352,
      "learning_rate": 1.9091424996824748e-05,
      "loss": 0.9404,
      "step": 8619
    },
    {
      "epoch": 1.5846496933655454,
      "grad_norm": 0.5969322919845581,
      "learning_rate": 1.9074240927028986e-05,
      "loss": 0.7992,
      "step": 8620
    },
    {
      "epoch": 1.5848355324289165,
      "grad_norm": 0.7880811095237732,
      "learning_rate": 1.9057063778988283e-05,
      "loss": 0.9778,
      "step": 8621
    },
    {
      "epoch": 1.5850213714922878,
      "grad_norm": 0.6685348749160767,
      "learning_rate": 1.903989355417185e-05,
      "loss": 1.0805,
      "step": 8622
    },
    {
      "epoch": 1.5852072105556587,
      "grad_norm": 0.7681272625923157,
      "learning_rate": 1.9022730254048327e-05,
      "loss": 1.0139,
      "step": 8623
    },
    {
      "epoch": 1.58539304961903,
      "grad_norm": 0.7222515940666199,
      "learning_rate": 1.900557388008568e-05,
      "loss": 0.8886,
      "step": 8624
    },
    {
      "epoch": 1.585578888682401,
      "grad_norm": 0.6568890810012817,
      "learning_rate": 1.8988424433751374e-05,
      "loss": 0.8081,
      "step": 8625
    },
    {
      "epoch": 1.5857647277457723,
      "grad_norm": 0.7260691523551941,
      "learning_rate": 1.8971281916512184e-05,
      "loss": 0.9272,
      "step": 8626
    },
    {
      "epoch": 1.5859505668091431,
      "grad_norm": 0.6106563210487366,
      "learning_rate": 1.8954146329834377e-05,
      "loss": 0.9614,
      "step": 8627
    },
    {
      "epoch": 1.5861364058725145,
      "grad_norm": 0.7054546475410461,
      "learning_rate": 1.893701767518361e-05,
      "loss": 0.8133,
      "step": 8628
    },
    {
      "epoch": 1.5863222449358854,
      "grad_norm": 0.6412302255630493,
      "learning_rate": 1.8919895954024912e-05,
      "loss": 1.0168,
      "step": 8629
    },
    {
      "epoch": 1.5865080839992567,
      "grad_norm": 0.8782892227172852,
      "learning_rate": 1.890278116782277e-05,
      "loss": 1.1363,
      "step": 8630
    },
    {
      "epoch": 1.5866939230626278,
      "grad_norm": 0.8776052594184875,
      "learning_rate": 1.8885673318041018e-05,
      "loss": 0.8191,
      "step": 8631
    },
    {
      "epoch": 1.586879762125999,
      "grad_norm": 0.6485216617584229,
      "learning_rate": 1.8868572406142958e-05,
      "loss": 0.6533,
      "step": 8632
    },
    {
      "epoch": 1.58706560118937,
      "grad_norm": 0.8401525616645813,
      "learning_rate": 1.885147843359122e-05,
      "loss": 1.0349,
      "step": 8633
    },
    {
      "epoch": 1.5872514402527411,
      "grad_norm": 0.7597665786743164,
      "learning_rate": 1.883439140184794e-05,
      "loss": 1.0918,
      "step": 8634
    },
    {
      "epoch": 1.5874372793161122,
      "grad_norm": 0.762586236000061,
      "learning_rate": 1.8817311312374564e-05,
      "loss": 1.1152,
      "step": 8635
    },
    {
      "epoch": 1.5876231183794833,
      "grad_norm": 0.73943692445755,
      "learning_rate": 1.8800238166632012e-05,
      "loss": 0.6863,
      "step": 8636
    },
    {
      "epoch": 1.5878089574428544,
      "grad_norm": 0.6723772883415222,
      "learning_rate": 1.8783171966080603e-05,
      "loss": 0.8738,
      "step": 8637
    },
    {
      "epoch": 1.5879947965062255,
      "grad_norm": 0.7524980306625366,
      "learning_rate": 1.8766112712180006e-05,
      "loss": 0.8884,
      "step": 8638
    },
    {
      "epoch": 1.5881806355695969,
      "grad_norm": 0.8099347352981567,
      "learning_rate": 1.8749060406389386e-05,
      "loss": 1.1606,
      "step": 8639
    },
    {
      "epoch": 1.5883664746329678,
      "grad_norm": 0.756376326084137,
      "learning_rate": 1.8732015050167218e-05,
      "loss": 0.9519,
      "step": 8640
    },
    {
      "epoch": 1.588552313696339,
      "grad_norm": 0.7101845741271973,
      "learning_rate": 1.871497664497144e-05,
      "loss": 0.7999,
      "step": 8641
    },
    {
      "epoch": 1.58873815275971,
      "grad_norm": 0.8314061164855957,
      "learning_rate": 1.869794519225938e-05,
      "loss": 1.0423,
      "step": 8642
    },
    {
      "epoch": 1.5889239918230813,
      "grad_norm": 0.7229037880897522,
      "learning_rate": 1.8680920693487792e-05,
      "loss": 0.9616,
      "step": 8643
    },
    {
      "epoch": 1.5891098308864522,
      "grad_norm": 0.6967963576316833,
      "learning_rate": 1.8663903150112837e-05,
      "loss": 0.9671,
      "step": 8644
    },
    {
      "epoch": 1.5892956699498235,
      "grad_norm": 0.7232875227928162,
      "learning_rate": 1.864689256359001e-05,
      "loss": 1.0798,
      "step": 8645
    },
    {
      "epoch": 1.5894815090131946,
      "grad_norm": 0.7937799096107483,
      "learning_rate": 1.862988893537432e-05,
      "loss": 0.9915,
      "step": 8646
    },
    {
      "epoch": 1.5896673480765657,
      "grad_norm": 0.7361637949943542,
      "learning_rate": 1.8612892266920055e-05,
      "loss": 0.733,
      "step": 8647
    },
    {
      "epoch": 1.5898531871399368,
      "grad_norm": 0.6371137499809265,
      "learning_rate": 1.8595902559681057e-05,
      "loss": 0.8012,
      "step": 8648
    },
    {
      "epoch": 1.590039026203308,
      "grad_norm": 0.7666060328483582,
      "learning_rate": 1.8578919815110408e-05,
      "loss": 1.1355,
      "step": 8649
    },
    {
      "epoch": 1.590224865266679,
      "grad_norm": 0.8657146692276001,
      "learning_rate": 1.856194403466074e-05,
      "loss": 1.0821,
      "step": 8650
    },
    {
      "epoch": 1.5904107043300502,
      "grad_norm": 0.8458912968635559,
      "learning_rate": 1.854497521978402e-05,
      "loss": 1.0808,
      "step": 8651
    },
    {
      "epoch": 1.5905965433934213,
      "grad_norm": 0.7622255086898804,
      "learning_rate": 1.8528013371931596e-05,
      "loss": 0.9853,
      "step": 8652
    },
    {
      "epoch": 1.5907823824567924,
      "grad_norm": 0.7003671526908875,
      "learning_rate": 1.8511058492554302e-05,
      "loss": 1.0668,
      "step": 8653
    },
    {
      "epoch": 1.5909682215201635,
      "grad_norm": 0.7001639604568481,
      "learning_rate": 1.8494110583102265e-05,
      "loss": 0.9536,
      "step": 8654
    },
    {
      "epoch": 1.5911540605835346,
      "grad_norm": 0.8250663876533508,
      "learning_rate": 1.8477169645025105e-05,
      "loss": 0.8961,
      "step": 8655
    },
    {
      "epoch": 1.591339899646906,
      "grad_norm": 0.6515169739723206,
      "learning_rate": 1.846023567977182e-05,
      "loss": 0.8208,
      "step": 8656
    },
    {
      "epoch": 1.5915257387102768,
      "grad_norm": 0.8317230939865112,
      "learning_rate": 1.8443308688790816e-05,
      "loss": 1.1036,
      "step": 8657
    },
    {
      "epoch": 1.5917115777736481,
      "grad_norm": 0.7945295572280884,
      "learning_rate": 1.8426388673529904e-05,
      "loss": 0.9185,
      "step": 8658
    },
    {
      "epoch": 1.591897416837019,
      "grad_norm": 0.5760872960090637,
      "learning_rate": 1.8409475635436245e-05,
      "loss": 0.7793,
      "step": 8659
    },
    {
      "epoch": 1.5920832559003903,
      "grad_norm": 0.627590537071228,
      "learning_rate": 1.8392569575956508e-05,
      "loss": 0.7388,
      "step": 8660
    },
    {
      "epoch": 1.5922690949637612,
      "grad_norm": 0.7489169836044312,
      "learning_rate": 1.8375670496536645e-05,
      "loss": 0.8907,
      "step": 8661
    },
    {
      "epoch": 1.5924549340271326,
      "grad_norm": 0.7438281178474426,
      "learning_rate": 1.8358778398622122e-05,
      "loss": 1.0574,
      "step": 8662
    },
    {
      "epoch": 1.5926407730905037,
      "grad_norm": 0.7566865086555481,
      "learning_rate": 1.8341893283657706e-05,
      "loss": 0.9673,
      "step": 8663
    },
    {
      "epoch": 1.5928266121538748,
      "grad_norm": 0.7408047318458557,
      "learning_rate": 1.8325015153087655e-05,
      "loss": 0.8386,
      "step": 8664
    },
    {
      "epoch": 1.5930124512172459,
      "grad_norm": 0.7485256791114807,
      "learning_rate": 1.83081440083556e-05,
      "loss": 0.8719,
      "step": 8665
    },
    {
      "epoch": 1.593198290280617,
      "grad_norm": 0.713358461856842,
      "learning_rate": 1.829127985090451e-05,
      "loss": 1.0221,
      "step": 8666
    },
    {
      "epoch": 1.593384129343988,
      "grad_norm": 0.6457721590995789,
      "learning_rate": 1.8274422682176882e-05,
      "loss": 0.7824,
      "step": 8667
    },
    {
      "epoch": 1.5935699684073592,
      "grad_norm": 0.6470180153846741,
      "learning_rate": 1.8257572503614494e-05,
      "loss": 0.9674,
      "step": 8668
    },
    {
      "epoch": 1.5937558074707303,
      "grad_norm": 0.654979407787323,
      "learning_rate": 1.82407293166586e-05,
      "loss": 0.976,
      "step": 8669
    },
    {
      "epoch": 1.5939416465341014,
      "grad_norm": 1.354627251625061,
      "learning_rate": 1.8223893122749857e-05,
      "loss": 1.4556,
      "step": 8670
    },
    {
      "epoch": 1.5941274855974727,
      "grad_norm": 0.8423448204994202,
      "learning_rate": 1.8207063923328237e-05,
      "loss": 0.9121,
      "step": 8671
    },
    {
      "epoch": 1.5943133246608436,
      "grad_norm": 0.7790369391441345,
      "learning_rate": 1.8190241719833233e-05,
      "loss": 0.8746,
      "step": 8672
    },
    {
      "epoch": 1.594499163724215,
      "grad_norm": 0.7903871536254883,
      "learning_rate": 1.8173426513703662e-05,
      "loss": 1.0795,
      "step": 8673
    },
    {
      "epoch": 1.5946850027875858,
      "grad_norm": 0.8038315773010254,
      "learning_rate": 1.8156618306377805e-05,
      "loss": 1.0508,
      "step": 8674
    },
    {
      "epoch": 1.5948708418509572,
      "grad_norm": 0.6397636532783508,
      "learning_rate": 1.8139817099293232e-05,
      "loss": 0.8527,
      "step": 8675
    },
    {
      "epoch": 1.595056680914328,
      "grad_norm": 0.774939775466919,
      "learning_rate": 1.8123022893887065e-05,
      "loss": 1.0397,
      "step": 8676
    },
    {
      "epoch": 1.5952425199776994,
      "grad_norm": 0.7472224235534668,
      "learning_rate": 1.8106235691595675e-05,
      "loss": 1.1034,
      "step": 8677
    },
    {
      "epoch": 1.5954283590410703,
      "grad_norm": 0.7666073441505432,
      "learning_rate": 1.8089455493854944e-05,
      "loss": 0.9883,
      "step": 8678
    },
    {
      "epoch": 1.5956141981044416,
      "grad_norm": 0.6919713020324707,
      "learning_rate": 1.807268230210014e-05,
      "loss": 0.8407,
      "step": 8679
    },
    {
      "epoch": 1.5958000371678127,
      "grad_norm": 0.6459718942642212,
      "learning_rate": 1.805591611776587e-05,
      "loss": 0.9556,
      "step": 8680
    },
    {
      "epoch": 1.5959858762311838,
      "grad_norm": 0.7357549667358398,
      "learning_rate": 1.8039156942286218e-05,
      "loss": 0.928,
      "step": 8681
    },
    {
      "epoch": 1.596171715294555,
      "grad_norm": 0.6708377599716187,
      "learning_rate": 1.8022404777094592e-05,
      "loss": 0.7842,
      "step": 8682
    },
    {
      "epoch": 1.596357554357926,
      "grad_norm": 0.7238865494728088,
      "learning_rate": 1.800565962362388e-05,
      "loss": 0.845,
      "step": 8683
    },
    {
      "epoch": 1.5965433934212971,
      "grad_norm": 0.5772403478622437,
      "learning_rate": 1.7988921483306297e-05,
      "loss": 0.9731,
      "step": 8684
    },
    {
      "epoch": 1.5967292324846682,
      "grad_norm": 0.628963828086853,
      "learning_rate": 1.7972190357573505e-05,
      "loss": 0.9534,
      "step": 8685
    },
    {
      "epoch": 1.5969150715480394,
      "grad_norm": 0.6648304462432861,
      "learning_rate": 1.7955466247856556e-05,
      "loss": 0.8157,
      "step": 8686
    },
    {
      "epoch": 1.5971009106114105,
      "grad_norm": 0.7793940305709839,
      "learning_rate": 1.7938749155585897e-05,
      "loss": 1.2967,
      "step": 8687
    },
    {
      "epoch": 1.5972867496747818,
      "grad_norm": 0.6746123433113098,
      "learning_rate": 1.792203908219141e-05,
      "loss": 1.0104,
      "step": 8688
    },
    {
      "epoch": 1.5974725887381527,
      "grad_norm": 0.8871963024139404,
      "learning_rate": 1.790533602910229e-05,
      "loss": 1.184,
      "step": 8689
    },
    {
      "epoch": 1.597658427801524,
      "grad_norm": 0.7723110318183899,
      "learning_rate": 1.7888639997747237e-05,
      "loss": 0.9134,
      "step": 8690
    },
    {
      "epoch": 1.597844266864895,
      "grad_norm": 0.6560113430023193,
      "learning_rate": 1.787195098955424e-05,
      "loss": 0.9556,
      "step": 8691
    },
    {
      "epoch": 1.5980301059282662,
      "grad_norm": 0.846978485584259,
      "learning_rate": 1.785526900595078e-05,
      "loss": 1.0675,
      "step": 8692
    },
    {
      "epoch": 1.598215944991637,
      "grad_norm": 0.7356848120689392,
      "learning_rate": 1.783859404836372e-05,
      "loss": 0.8517,
      "step": 8693
    },
    {
      "epoch": 1.5984017840550084,
      "grad_norm": 0.837843656539917,
      "learning_rate": 1.7821926118219268e-05,
      "loss": 1.2275,
      "step": 8694
    },
    {
      "epoch": 1.5985876231183793,
      "grad_norm": 0.6126946806907654,
      "learning_rate": 1.78052652169431e-05,
      "loss": 0.8591,
      "step": 8695
    },
    {
      "epoch": 1.5987734621817506,
      "grad_norm": 0.7435805201530457,
      "learning_rate": 1.7788611345960227e-05,
      "loss": 1.0836,
      "step": 8696
    },
    {
      "epoch": 1.5989593012451218,
      "grad_norm": 0.7877734899520874,
      "learning_rate": 1.7771964506695136e-05,
      "loss": 1.2115,
      "step": 8697
    },
    {
      "epoch": 1.5991451403084929,
      "grad_norm": 0.5992589592933655,
      "learning_rate": 1.7755324700571608e-05,
      "loss": 0.7697,
      "step": 8698
    },
    {
      "epoch": 1.599330979371864,
      "grad_norm": 1.7316218614578247,
      "learning_rate": 1.7738691929012918e-05,
      "loss": 1.7008,
      "step": 8699
    },
    {
      "epoch": 1.599516818435235,
      "grad_norm": 0.7186190485954285,
      "learning_rate": 1.772206619344169e-05,
      "loss": 0.8947,
      "step": 8700
    },
    {
      "epoch": 1.5997026574986062,
      "grad_norm": 0.6659690141677856,
      "learning_rate": 1.770544749527997e-05,
      "loss": 0.7263,
      "step": 8701
    },
    {
      "epoch": 1.5998884965619773,
      "grad_norm": 0.9670864343643188,
      "learning_rate": 1.7688835835949214e-05,
      "loss": 1.0106,
      "step": 8702
    },
    {
      "epoch": 1.6000743356253484,
      "grad_norm": 0.7367480397224426,
      "learning_rate": 1.7672231216870194e-05,
      "loss": 1.1188,
      "step": 8703
    },
    {
      "epoch": 1.6002601746887195,
      "grad_norm": 0.6136513948440552,
      "learning_rate": 1.7655633639463208e-05,
      "loss": 0.7875,
      "step": 8704
    },
    {
      "epoch": 1.6004460137520908,
      "grad_norm": 0.6350746750831604,
      "learning_rate": 1.7639043105147813e-05,
      "loss": 1.0765,
      "step": 8705
    },
    {
      "epoch": 1.6006318528154617,
      "grad_norm": 0.8280909657478333,
      "learning_rate": 1.7622459615343067e-05,
      "loss": 0.9999,
      "step": 8706
    },
    {
      "epoch": 1.600817691878833,
      "grad_norm": 0.6341865658760071,
      "learning_rate": 1.7605883171467418e-05,
      "loss": 0.8677,
      "step": 8707
    },
    {
      "epoch": 1.601003530942204,
      "grad_norm": 0.6041312217712402,
      "learning_rate": 1.7589313774938643e-05,
      "loss": 0.7505,
      "step": 8708
    },
    {
      "epoch": 1.6011893700055753,
      "grad_norm": 0.6812422871589661,
      "learning_rate": 1.7572751427173984e-05,
      "loss": 0.9407,
      "step": 8709
    },
    {
      "epoch": 1.6013752090689461,
      "grad_norm": 0.7144443392753601,
      "learning_rate": 1.755619612959003e-05,
      "loss": 0.9346,
      "step": 8710
    },
    {
      "epoch": 1.6015610481323175,
      "grad_norm": 0.6252550482749939,
      "learning_rate": 1.7539647883602828e-05,
      "loss": 0.7634,
      "step": 8711
    },
    {
      "epoch": 1.6017468871956886,
      "grad_norm": 0.9227977395057678,
      "learning_rate": 1.7523106690627732e-05,
      "loss": 0.6466,
      "step": 8712
    },
    {
      "epoch": 1.6019327262590597,
      "grad_norm": 0.6825462579727173,
      "learning_rate": 1.750657255207958e-05,
      "loss": 0.9107,
      "step": 8713
    },
    {
      "epoch": 1.6021185653224308,
      "grad_norm": 0.6466217041015625,
      "learning_rate": 1.7490045469372574e-05,
      "loss": 0.8781,
      "step": 8714
    },
    {
      "epoch": 1.602304404385802,
      "grad_norm": 0.766028642654419,
      "learning_rate": 1.7473525443920325e-05,
      "loss": 0.7986,
      "step": 8715
    },
    {
      "epoch": 1.602490243449173,
      "grad_norm": 0.8478003740310669,
      "learning_rate": 1.7457012477135782e-05,
      "loss": 0.7937,
      "step": 8716
    },
    {
      "epoch": 1.6026760825125441,
      "grad_norm": 0.6838301420211792,
      "learning_rate": 1.744050657043137e-05,
      "loss": 1.0052,
      "step": 8717
    },
    {
      "epoch": 1.6028619215759152,
      "grad_norm": 0.6403393149375916,
      "learning_rate": 1.7424007725218883e-05,
      "loss": 0.6411,
      "step": 8718
    },
    {
      "epoch": 1.6030477606392863,
      "grad_norm": 0.7843559980392456,
      "learning_rate": 1.7407515942909457e-05,
      "loss": 0.9943,
      "step": 8719
    },
    {
      "epoch": 1.6032335997026577,
      "grad_norm": 0.69202721118927,
      "learning_rate": 1.739103122491371e-05,
      "loss": 0.8886,
      "step": 8720
    },
    {
      "epoch": 1.6034194387660285,
      "grad_norm": 0.6292579770088196,
      "learning_rate": 1.7374553572641617e-05,
      "loss": 0.9584,
      "step": 8721
    },
    {
      "epoch": 1.6036052778293999,
      "grad_norm": 0.707538902759552,
      "learning_rate": 1.7358082987502523e-05,
      "loss": 0.9153,
      "step": 8722
    },
    {
      "epoch": 1.6037911168927708,
      "grad_norm": 0.7223339080810547,
      "learning_rate": 1.734161947090521e-05,
      "loss": 0.8471,
      "step": 8723
    },
    {
      "epoch": 1.603976955956142,
      "grad_norm": 0.9097388982772827,
      "learning_rate": 1.7325163024257818e-05,
      "loss": 0.9756,
      "step": 8724
    },
    {
      "epoch": 1.604162795019513,
      "grad_norm": 0.6092236638069153,
      "learning_rate": 1.7308713648967945e-05,
      "loss": 0.8956,
      "step": 8725
    },
    {
      "epoch": 1.6043486340828843,
      "grad_norm": 0.7708209753036499,
      "learning_rate": 1.729227134644248e-05,
      "loss": 0.9583,
      "step": 8726
    },
    {
      "epoch": 1.6045344731462552,
      "grad_norm": 0.5948147177696228,
      "learning_rate": 1.7275836118087807e-05,
      "loss": 0.9365,
      "step": 8727
    },
    {
      "epoch": 1.6047203122096265,
      "grad_norm": 0.679939866065979,
      "learning_rate": 1.7259407965309682e-05,
      "loss": 0.963,
      "step": 8728
    },
    {
      "epoch": 1.6049061512729976,
      "grad_norm": 0.6386947631835938,
      "learning_rate": 1.72429868895132e-05,
      "loss": 1.0086,
      "step": 8729
    },
    {
      "epoch": 1.6050919903363687,
      "grad_norm": 0.6231483221054077,
      "learning_rate": 1.7226572892102912e-05,
      "loss": 0.9308,
      "step": 8730
    },
    {
      "epoch": 1.6052778293997398,
      "grad_norm": 0.688547670841217,
      "learning_rate": 1.721016597448274e-05,
      "loss": 0.9231,
      "step": 8731
    },
    {
      "epoch": 1.605463668463111,
      "grad_norm": 0.8977246880531311,
      "learning_rate": 1.719376613805602e-05,
      "loss": 1.176,
      "step": 8732
    },
    {
      "epoch": 1.605649507526482,
      "grad_norm": 0.6980913281440735,
      "learning_rate": 1.7177373384225436e-05,
      "loss": 0.8452,
      "step": 8733
    },
    {
      "epoch": 1.6058353465898532,
      "grad_norm": 0.9629896879196167,
      "learning_rate": 1.7160987714393105e-05,
      "loss": 0.7985,
      "step": 8734
    },
    {
      "epoch": 1.6060211856532243,
      "grad_norm": 0.7529435157775879,
      "learning_rate": 1.7144609129960553e-05,
      "loss": 1.2252,
      "step": 8735
    },
    {
      "epoch": 1.6062070247165954,
      "grad_norm": 0.663002610206604,
      "learning_rate": 1.7128237632328636e-05,
      "loss": 0.8357,
      "step": 8736
    },
    {
      "epoch": 1.6063928637799667,
      "grad_norm": 0.729487419128418,
      "learning_rate": 1.711187322289769e-05,
      "loss": 1.0419,
      "step": 8737
    },
    {
      "epoch": 1.6065787028433376,
      "grad_norm": 0.8559988737106323,
      "learning_rate": 1.7095515903067348e-05,
      "loss": 1.0194,
      "step": 8738
    },
    {
      "epoch": 1.606764541906709,
      "grad_norm": 0.8755013942718506,
      "learning_rate": 1.707916567423674e-05,
      "loss": 1.1847,
      "step": 8739
    },
    {
      "epoch": 1.6069503809700798,
      "grad_norm": 0.6425202488899231,
      "learning_rate": 1.706282253780428e-05,
      "loss": 0.9878,
      "step": 8740
    },
    {
      "epoch": 1.6071362200334511,
      "grad_norm": 0.6389315724372864,
      "learning_rate": 1.7046486495167857e-05,
      "loss": 0.9392,
      "step": 8741
    },
    {
      "epoch": 1.607322059096822,
      "grad_norm": 0.7063748240470886,
      "learning_rate": 1.7030157547724745e-05,
      "loss": 1.0674,
      "step": 8742
    },
    {
      "epoch": 1.6075078981601933,
      "grad_norm": 0.7330775260925293,
      "learning_rate": 1.7013835696871572e-05,
      "loss": 0.8779,
      "step": 8743
    },
    {
      "epoch": 1.6076937372235642,
      "grad_norm": 0.7162792682647705,
      "learning_rate": 1.6997520944004375e-05,
      "loss": 1.028,
      "step": 8744
    },
    {
      "epoch": 1.6078795762869356,
      "grad_norm": 0.7317261695861816,
      "learning_rate": 1.6981213290518614e-05,
      "loss": 1.0238,
      "step": 8745
    },
    {
      "epoch": 1.6080654153503067,
      "grad_norm": 0.6344571113586426,
      "learning_rate": 1.6964912737809123e-05,
      "loss": 0.8266,
      "step": 8746
    },
    {
      "epoch": 1.6082512544136778,
      "grad_norm": 0.6402305960655212,
      "learning_rate": 1.6948619287270097e-05,
      "loss": 0.9155,
      "step": 8747
    },
    {
      "epoch": 1.6084370934770489,
      "grad_norm": 0.7457547187805176,
      "learning_rate": 1.693233294029515e-05,
      "loss": 1.1292,
      "step": 8748
    },
    {
      "epoch": 1.60862293254042,
      "grad_norm": 0.6838110089302063,
      "learning_rate": 1.6916053698277323e-05,
      "loss": 0.6868,
      "step": 8749
    },
    {
      "epoch": 1.608808771603791,
      "grad_norm": 1.871200442314148,
      "learning_rate": 1.6899781562608962e-05,
      "loss": 1.2426,
      "step": 8750
    },
    {
      "epoch": 1.6089946106671622,
      "grad_norm": 0.7761055827140808,
      "learning_rate": 1.688351653468192e-05,
      "loss": 1.0335,
      "step": 8751
    },
    {
      "epoch": 1.6091804497305333,
      "grad_norm": 0.5921274423599243,
      "learning_rate": 1.6867258615887328e-05,
      "loss": 0.7222,
      "step": 8752
    },
    {
      "epoch": 1.6093662887939044,
      "grad_norm": 0.6779144406318665,
      "learning_rate": 1.6851007807615792e-05,
      "loss": 0.873,
      "step": 8753
    },
    {
      "epoch": 1.6095521278572757,
      "grad_norm": 0.6916807889938354,
      "learning_rate": 1.683476411125725e-05,
      "loss": 1.0378,
      "step": 8754
    },
    {
      "epoch": 1.6097379669206466,
      "grad_norm": 0.6583566069602966,
      "learning_rate": 1.6818527528201078e-05,
      "loss": 0.9381,
      "step": 8755
    },
    {
      "epoch": 1.609923805984018,
      "grad_norm": 0.6733114719390869,
      "learning_rate": 1.6802298059836052e-05,
      "loss": 0.7155,
      "step": 8756
    },
    {
      "epoch": 1.6101096450473888,
      "grad_norm": 0.5884139537811279,
      "learning_rate": 1.6786075707550253e-05,
      "loss": 0.8258,
      "step": 8757
    },
    {
      "epoch": 1.6102954841107602,
      "grad_norm": 0.6669282913208008,
      "learning_rate": 1.6769860472731257e-05,
      "loss": 0.8458,
      "step": 8758
    },
    {
      "epoch": 1.610481323174131,
      "grad_norm": 0.7951620221138,
      "learning_rate": 1.6753652356765968e-05,
      "loss": 1.1442,
      "step": 8759
    },
    {
      "epoch": 1.6106671622375024,
      "grad_norm": 0.6817410588264465,
      "learning_rate": 1.6737451361040746e-05,
      "loss": 0.9404,
      "step": 8760
    },
    {
      "epoch": 1.6108530013008733,
      "grad_norm": 0.7657042741775513,
      "learning_rate": 1.6721257486941234e-05,
      "loss": 0.9618,
      "step": 8761
    },
    {
      "epoch": 1.6110388403642446,
      "grad_norm": 0.7671672105789185,
      "learning_rate": 1.670507073585257e-05,
      "loss": 0.88,
      "step": 8762
    },
    {
      "epoch": 1.6112246794276157,
      "grad_norm": 0.7470976114273071,
      "learning_rate": 1.6688891109159244e-05,
      "loss": 1.0605,
      "step": 8763
    },
    {
      "epoch": 1.6114105184909868,
      "grad_norm": 0.7541417479515076,
      "learning_rate": 1.6672718608245086e-05,
      "loss": 1.0119,
      "step": 8764
    },
    {
      "epoch": 1.611596357554358,
      "grad_norm": 0.69011390209198,
      "learning_rate": 1.6656553234493445e-05,
      "loss": 0.8798,
      "step": 8765
    },
    {
      "epoch": 1.611782196617729,
      "grad_norm": 0.7282363176345825,
      "learning_rate": 1.664039498928689e-05,
      "loss": 1.0463,
      "step": 8766
    },
    {
      "epoch": 1.6119680356811001,
      "grad_norm": 0.6672340035438538,
      "learning_rate": 1.662424387400755e-05,
      "loss": 1.0736,
      "step": 8767
    },
    {
      "epoch": 1.6121538747444712,
      "grad_norm": 0.9124537706375122,
      "learning_rate": 1.6608099890036812e-05,
      "loss": 0.8491,
      "step": 8768
    },
    {
      "epoch": 1.6123397138078424,
      "grad_norm": 0.7714374661445618,
      "learning_rate": 1.659196303875551e-05,
      "loss": 0.8433,
      "step": 8769
    },
    {
      "epoch": 1.6125255528712135,
      "grad_norm": 0.6858008503913879,
      "learning_rate": 1.657583332154391e-05,
      "loss": 1.0529,
      "step": 8770
    },
    {
      "epoch": 1.6127113919345848,
      "grad_norm": 0.5979937314987183,
      "learning_rate": 1.6559710739781565e-05,
      "loss": 0.6754,
      "step": 8771
    },
    {
      "epoch": 1.6128972309979557,
      "grad_norm": 0.6006745100021362,
      "learning_rate": 1.6543595294847525e-05,
      "loss": 0.8297,
      "step": 8772
    },
    {
      "epoch": 1.613083070061327,
      "grad_norm": 0.7025749683380127,
      "learning_rate": 1.652748698812012e-05,
      "loss": 0.7941,
      "step": 8773
    },
    {
      "epoch": 1.613268909124698,
      "grad_norm": 0.7484373450279236,
      "learning_rate": 1.6511385820977165e-05,
      "loss": 0.8617,
      "step": 8774
    },
    {
      "epoch": 1.6134547481880692,
      "grad_norm": 0.6823762059211731,
      "learning_rate": 1.6495291794795832e-05,
      "loss": 0.8227,
      "step": 8775
    },
    {
      "epoch": 1.61364058725144,
      "grad_norm": 0.6824286580085754,
      "learning_rate": 1.6479204910952663e-05,
      "loss": 0.979,
      "step": 8776
    },
    {
      "epoch": 1.6138264263148114,
      "grad_norm": 0.7116637229919434,
      "learning_rate": 1.6463125170823635e-05,
      "loss": 0.7134,
      "step": 8777
    },
    {
      "epoch": 1.6140122653781825,
      "grad_norm": 0.716037929058075,
      "learning_rate": 1.6447052575784028e-05,
      "loss": 0.9194,
      "step": 8778
    },
    {
      "epoch": 1.6141981044415536,
      "grad_norm": 0.6647517681121826,
      "learning_rate": 1.6430987127208618e-05,
      "loss": 0.8919,
      "step": 8779
    },
    {
      "epoch": 1.6143839435049248,
      "grad_norm": 0.8217048645019531,
      "learning_rate": 1.6414928826471488e-05,
      "loss": 0.9364,
      "step": 8780
    },
    {
      "epoch": 1.6145697825682959,
      "grad_norm": 0.8008784055709839,
      "learning_rate": 1.6398877674946156e-05,
      "loss": 0.9451,
      "step": 8781
    },
    {
      "epoch": 1.614755621631667,
      "grad_norm": 0.7590863108634949,
      "learning_rate": 1.6382833674005484e-05,
      "loss": 0.7621,
      "step": 8782
    },
    {
      "epoch": 1.614941460695038,
      "grad_norm": 0.6634281277656555,
      "learning_rate": 1.636679682502176e-05,
      "loss": 1.0038,
      "step": 8783
    },
    {
      "epoch": 1.6151272997584092,
      "grad_norm": 0.767975926399231,
      "learning_rate": 1.6350767129366694e-05,
      "loss": 0.8915,
      "step": 8784
    },
    {
      "epoch": 1.6153131388217803,
      "grad_norm": 0.6990622282028198,
      "learning_rate": 1.6334744588411276e-05,
      "loss": 1.0527,
      "step": 8785
    },
    {
      "epoch": 1.6154989778851516,
      "grad_norm": 0.5986868739128113,
      "learning_rate": 1.6318729203525994e-05,
      "loss": 0.9237,
      "step": 8786
    },
    {
      "epoch": 1.6156848169485225,
      "grad_norm": 0.7767715454101562,
      "learning_rate": 1.6302720976080653e-05,
      "loss": 1.0121,
      "step": 8787
    },
    {
      "epoch": 1.6158706560118938,
      "grad_norm": 0.7090464234352112,
      "learning_rate": 1.628671990744447e-05,
      "loss": 0.89,
      "step": 8788
    },
    {
      "epoch": 1.6160564950752647,
      "grad_norm": 0.7036499977111816,
      "learning_rate": 1.627072599898607e-05,
      "loss": 1.0243,
      "step": 8789
    },
    {
      "epoch": 1.616242334138636,
      "grad_norm": 0.7550042867660522,
      "learning_rate": 1.6254739252073436e-05,
      "loss": 0.9792,
      "step": 8790
    },
    {
      "epoch": 1.616428173202007,
      "grad_norm": 0.8685239553451538,
      "learning_rate": 1.6238759668073966e-05,
      "loss": 0.902,
      "step": 8791
    },
    {
      "epoch": 1.6166140122653783,
      "grad_norm": 0.741919755935669,
      "learning_rate": 1.6222787248354398e-05,
      "loss": 0.8697,
      "step": 8792
    },
    {
      "epoch": 1.6167998513287491,
      "grad_norm": 0.6833041906356812,
      "learning_rate": 1.620682199428093e-05,
      "loss": 1.2719,
      "step": 8793
    },
    {
      "epoch": 1.6169856903921205,
      "grad_norm": 0.6062202453613281,
      "learning_rate": 1.6190863907219046e-05,
      "loss": 0.8007,
      "step": 8794
    },
    {
      "epoch": 1.6171715294554916,
      "grad_norm": 0.741337239742279,
      "learning_rate": 1.617491298853373e-05,
      "loss": 0.9497,
      "step": 8795
    },
    {
      "epoch": 1.6173573685188627,
      "grad_norm": 0.683061957359314,
      "learning_rate": 1.615896923958926e-05,
      "loss": 0.8604,
      "step": 8796
    },
    {
      "epoch": 1.6175432075822338,
      "grad_norm": 0.7003892064094543,
      "learning_rate": 1.6143032661749347e-05,
      "loss": 1.0212,
      "step": 8797
    },
    {
      "epoch": 1.617729046645605,
      "grad_norm": 0.6905108690261841,
      "learning_rate": 1.612710325637712e-05,
      "loss": 1.0571,
      "step": 8798
    },
    {
      "epoch": 1.617914885708976,
      "grad_norm": 0.5632252097129822,
      "learning_rate": 1.6111181024835e-05,
      "loss": 0.9349,
      "step": 8799
    },
    {
      "epoch": 1.6181007247723471,
      "grad_norm": 0.6871153116226196,
      "learning_rate": 1.6095265968484898e-05,
      "loss": 0.9384,
      "step": 8800
    },
    {
      "epoch": 1.6182865638357182,
      "grad_norm": 0.8183594346046448,
      "learning_rate": 1.607935808868801e-05,
      "loss": 0.8518,
      "step": 8801
    },
    {
      "epoch": 1.6184724028990893,
      "grad_norm": 0.8559496402740479,
      "learning_rate": 1.6063457386805004e-05,
      "loss": 1.1082,
      "step": 8802
    },
    {
      "epoch": 1.6186582419624607,
      "grad_norm": 0.7051379680633545,
      "learning_rate": 1.6047563864195902e-05,
      "loss": 0.8697,
      "step": 8803
    },
    {
      "epoch": 1.6188440810258315,
      "grad_norm": 0.6523978114128113,
      "learning_rate": 1.6031677522220113e-05,
      "loss": 1.1111,
      "step": 8804
    },
    {
      "epoch": 1.6190299200892029,
      "grad_norm": 0.6645488739013672,
      "learning_rate": 1.6015798362236433e-05,
      "loss": 0.9398,
      "step": 8805
    },
    {
      "epoch": 1.6192157591525738,
      "grad_norm": 0.7267053723335266,
      "learning_rate": 1.5999926385603025e-05,
      "loss": 0.8864,
      "step": 8806
    },
    {
      "epoch": 1.619401598215945,
      "grad_norm": 0.7057207226753235,
      "learning_rate": 1.598406159367748e-05,
      "loss": 0.8944,
      "step": 8807
    },
    {
      "epoch": 1.619587437279316,
      "grad_norm": 0.7058221101760864,
      "learning_rate": 1.59682039878167e-05,
      "loss": 0.7088,
      "step": 8808
    },
    {
      "epoch": 1.6197732763426873,
      "grad_norm": 0.6227111220359802,
      "learning_rate": 1.5952353569377077e-05,
      "loss": 0.9605,
      "step": 8809
    },
    {
      "epoch": 1.6199591154060582,
      "grad_norm": 0.7384781241416931,
      "learning_rate": 1.5936510339714273e-05,
      "loss": 1.1215,
      "step": 8810
    },
    {
      "epoch": 1.6201449544694295,
      "grad_norm": 0.5976701378822327,
      "learning_rate": 1.5920674300183435e-05,
      "loss": 0.6481,
      "step": 8811
    },
    {
      "epoch": 1.6203307935328006,
      "grad_norm": 0.6821479201316833,
      "learning_rate": 1.5904845452139062e-05,
      "loss": 0.9243,
      "step": 8812
    },
    {
      "epoch": 1.6205166325961717,
      "grad_norm": 0.7828944325447083,
      "learning_rate": 1.5889023796934986e-05,
      "loss": 1.0344,
      "step": 8813
    },
    {
      "epoch": 1.6207024716595428,
      "grad_norm": 0.6303045153617859,
      "learning_rate": 1.5873209335924522e-05,
      "loss": 0.8991,
      "step": 8814
    },
    {
      "epoch": 1.620888310722914,
      "grad_norm": 0.8067341446876526,
      "learning_rate": 1.585740207046026e-05,
      "loss": 0.961,
      "step": 8815
    },
    {
      "epoch": 1.621074149786285,
      "grad_norm": 0.7798840403556824,
      "learning_rate": 1.584160200189425e-05,
      "loss": 0.9915,
      "step": 8816
    },
    {
      "epoch": 1.6212599888496562,
      "grad_norm": 0.6932112574577332,
      "learning_rate": 1.582580913157794e-05,
      "loss": 1.0362,
      "step": 8817
    },
    {
      "epoch": 1.6214458279130273,
      "grad_norm": 0.7943875193595886,
      "learning_rate": 1.5810023460862077e-05,
      "loss": 1.2271,
      "step": 8818
    },
    {
      "epoch": 1.6216316669763984,
      "grad_norm": 0.7108856439590454,
      "learning_rate": 1.5794244991096862e-05,
      "loss": 0.959,
      "step": 8819
    },
    {
      "epoch": 1.6218175060397697,
      "grad_norm": 0.683498203754425,
      "learning_rate": 1.577847372363187e-05,
      "loss": 0.8458,
      "step": 8820
    },
    {
      "epoch": 1.6220033451031406,
      "grad_norm": 0.8767088055610657,
      "learning_rate": 1.5762709659816066e-05,
      "loss": 0.8359,
      "step": 8821
    },
    {
      "epoch": 1.622189184166512,
      "grad_norm": 0.6798695921897888,
      "learning_rate": 1.574695280099774e-05,
      "loss": 0.8094,
      "step": 8822
    },
    {
      "epoch": 1.6223750232298828,
      "grad_norm": 0.5580926537513733,
      "learning_rate": 1.5731203148524663e-05,
      "loss": 0.6944,
      "step": 8823
    },
    {
      "epoch": 1.6225608622932541,
      "grad_norm": 0.7183531522750854,
      "learning_rate": 1.571546070374389e-05,
      "loss": 0.9789,
      "step": 8824
    },
    {
      "epoch": 1.622746701356625,
      "grad_norm": 0.5126298666000366,
      "learning_rate": 1.5699725468001912e-05,
      "loss": 0.5138,
      "step": 8825
    },
    {
      "epoch": 1.6229325404199964,
      "grad_norm": 0.6757166981697083,
      "learning_rate": 1.568399744264465e-05,
      "loss": 0.8028,
      "step": 8826
    },
    {
      "epoch": 1.6231183794833675,
      "grad_norm": 0.6396664381027222,
      "learning_rate": 1.5668276629017286e-05,
      "loss": 1.0848,
      "step": 8827
    },
    {
      "epoch": 1.6233042185467386,
      "grad_norm": 0.7553821206092834,
      "learning_rate": 1.5652563028464517e-05,
      "loss": 0.8485,
      "step": 8828
    },
    {
      "epoch": 1.6234900576101097,
      "grad_norm": 0.7239199876785278,
      "learning_rate": 1.56368566423303e-05,
      "loss": 0.9049,
      "step": 8829
    },
    {
      "epoch": 1.6236758966734808,
      "grad_norm": 0.6508596539497375,
      "learning_rate": 1.5621157471958093e-05,
      "loss": 0.8735,
      "step": 8830
    },
    {
      "epoch": 1.6238617357368519,
      "grad_norm": 0.6568943858146667,
      "learning_rate": 1.5605465518690633e-05,
      "loss": 0.9732,
      "step": 8831
    },
    {
      "epoch": 1.624047574800223,
      "grad_norm": 0.6617598533630371,
      "learning_rate": 1.558978078387011e-05,
      "loss": 0.8283,
      "step": 8832
    },
    {
      "epoch": 1.624233413863594,
      "grad_norm": 0.7476336359977722,
      "learning_rate": 1.5574103268838058e-05,
      "loss": 0.8835,
      "step": 8833
    },
    {
      "epoch": 1.6244192529269652,
      "grad_norm": 0.744114339351654,
      "learning_rate": 1.555843297493543e-05,
      "loss": 0.9546,
      "step": 8834
    },
    {
      "epoch": 1.6246050919903363,
      "grad_norm": 0.5931903123855591,
      "learning_rate": 1.5542769903502562e-05,
      "loss": 0.7503,
      "step": 8835
    },
    {
      "epoch": 1.6247909310537074,
      "grad_norm": 0.7275744080543518,
      "learning_rate": 1.552711405587909e-05,
      "loss": 1.0665,
      "step": 8836
    },
    {
      "epoch": 1.6249767701170788,
      "grad_norm": 0.8302237391471863,
      "learning_rate": 1.551146543340415e-05,
      "loss": 0.8374,
      "step": 8837
    },
    {
      "epoch": 1.6251626091804496,
      "grad_norm": 0.7517722249031067,
      "learning_rate": 1.549582403741615e-05,
      "loss": 1.0305,
      "step": 8838
    },
    {
      "epoch": 1.625348448243821,
      "grad_norm": 0.7293110489845276,
      "learning_rate": 1.548018986925296e-05,
      "loss": 0.8774,
      "step": 8839
    },
    {
      "epoch": 1.6255342873071918,
      "grad_norm": 0.7916002869606018,
      "learning_rate": 1.5464562930251814e-05,
      "loss": 1.0857,
      "step": 8840
    },
    {
      "epoch": 1.6257201263705632,
      "grad_norm": 0.7529808282852173,
      "learning_rate": 1.544894322174929e-05,
      "loss": 0.9024,
      "step": 8841
    },
    {
      "epoch": 1.625905965433934,
      "grad_norm": 0.6958862543106079,
      "learning_rate": 1.543333074508142e-05,
      "loss": 0.8539,
      "step": 8842
    },
    {
      "epoch": 1.6260918044973054,
      "grad_norm": 0.7076324820518494,
      "learning_rate": 1.5417725501583514e-05,
      "loss": 0.9758,
      "step": 8843
    },
    {
      "epoch": 1.6262776435606765,
      "grad_norm": 0.6581704020500183,
      "learning_rate": 1.540212749259038e-05,
      "loss": 0.9227,
      "step": 8844
    },
    {
      "epoch": 1.6264634826240476,
      "grad_norm": 0.9186629056930542,
      "learning_rate": 1.5386536719436095e-05,
      "loss": 1.1539,
      "step": 8845
    },
    {
      "epoch": 1.6266493216874187,
      "grad_norm": 0.6017464995384216,
      "learning_rate": 1.5370953183454194e-05,
      "loss": 0.8596,
      "step": 8846
    },
    {
      "epoch": 1.6268351607507898,
      "grad_norm": 0.8916236758232117,
      "learning_rate": 1.535537688597758e-05,
      "loss": 1.0041,
      "step": 8847
    },
    {
      "epoch": 1.627020999814161,
      "grad_norm": 0.7704942226409912,
      "learning_rate": 1.5339807828338524e-05,
      "loss": 0.9189,
      "step": 8848
    },
    {
      "epoch": 1.627206838877532,
      "grad_norm": 0.7622827887535095,
      "learning_rate": 1.5324246011868694e-05,
      "loss": 0.9924,
      "step": 8849
    },
    {
      "epoch": 1.6273926779409031,
      "grad_norm": 0.6575504541397095,
      "learning_rate": 1.5308691437899093e-05,
      "loss": 0.8895,
      "step": 8850
    },
    {
      "epoch": 1.6275785170042743,
      "grad_norm": 0.7674857974052429,
      "learning_rate": 1.5293144107760183e-05,
      "loss": 0.9631,
      "step": 8851
    },
    {
      "epoch": 1.6277643560676456,
      "grad_norm": 0.7199091911315918,
      "learning_rate": 1.5277604022781712e-05,
      "loss": 1.0325,
      "step": 8852
    },
    {
      "epoch": 1.6279501951310165,
      "grad_norm": 0.7748641967773438,
      "learning_rate": 1.526207118429287e-05,
      "loss": 1.1489,
      "step": 8853
    },
    {
      "epoch": 1.6281360341943878,
      "grad_norm": 0.8894672989845276,
      "learning_rate": 1.5246545593622242e-05,
      "loss": 0.9587,
      "step": 8854
    },
    {
      "epoch": 1.6283218732577587,
      "grad_norm": 0.6617799997329712,
      "learning_rate": 1.523102725209773e-05,
      "loss": 1.0071,
      "step": 8855
    },
    {
      "epoch": 1.62850771232113,
      "grad_norm": 0.6675786375999451,
      "learning_rate": 1.5215516161046683e-05,
      "loss": 1.0015,
      "step": 8856
    },
    {
      "epoch": 1.628693551384501,
      "grad_norm": 0.6378850340843201,
      "learning_rate": 1.5200012321795765e-05,
      "loss": 0.8807,
      "step": 8857
    },
    {
      "epoch": 1.6288793904478722,
      "grad_norm": 0.8859041929244995,
      "learning_rate": 1.518451573567109e-05,
      "loss": 1.2239,
      "step": 8858
    },
    {
      "epoch": 1.629065229511243,
      "grad_norm": 0.8571634888648987,
      "learning_rate": 1.5169026403998076e-05,
      "loss": 1.0145,
      "step": 8859
    },
    {
      "epoch": 1.6292510685746144,
      "grad_norm": 1.0119373798370361,
      "learning_rate": 1.5153544328101577e-05,
      "loss": 1.3056,
      "step": 8860
    },
    {
      "epoch": 1.6294369076379855,
      "grad_norm": 0.6872910261154175,
      "learning_rate": 1.5138069509305807e-05,
      "loss": 0.6696,
      "step": 8861
    },
    {
      "epoch": 1.6296227467013567,
      "grad_norm": 0.6504276990890503,
      "learning_rate": 1.512260194893439e-05,
      "loss": 0.9549,
      "step": 8862
    },
    {
      "epoch": 1.6298085857647278,
      "grad_norm": 0.6661650538444519,
      "learning_rate": 1.5107141648310241e-05,
      "loss": 0.8873,
      "step": 8863
    },
    {
      "epoch": 1.6299944248280989,
      "grad_norm": 0.8729071617126465,
      "learning_rate": 1.5091688608755749e-05,
      "loss": 0.847,
      "step": 8864
    },
    {
      "epoch": 1.63018026389147,
      "grad_norm": 0.8262352347373962,
      "learning_rate": 1.5076242831592668e-05,
      "loss": 0.8209,
      "step": 8865
    },
    {
      "epoch": 1.630366102954841,
      "grad_norm": 0.6832601428031921,
      "learning_rate": 1.5060804318142052e-05,
      "loss": 0.8971,
      "step": 8866
    },
    {
      "epoch": 1.6305519420182122,
      "grad_norm": 0.7103747725486755,
      "learning_rate": 1.5045373069724423e-05,
      "loss": 0.8639,
      "step": 8867
    },
    {
      "epoch": 1.6307377810815833,
      "grad_norm": 0.7560636401176453,
      "learning_rate": 1.5029949087659667e-05,
      "loss": 0.941,
      "step": 8868
    },
    {
      "epoch": 1.6309236201449546,
      "grad_norm": 0.7987247109413147,
      "learning_rate": 1.5014532373266987e-05,
      "loss": 0.76,
      "step": 8869
    },
    {
      "epoch": 1.6311094592083255,
      "grad_norm": 0.7744994759559631,
      "learning_rate": 1.4999122927865061e-05,
      "loss": 0.7802,
      "step": 8870
    },
    {
      "epoch": 1.6312952982716968,
      "grad_norm": 0.7106496095657349,
      "learning_rate": 1.4983720752771834e-05,
      "loss": 0.8834,
      "step": 8871
    },
    {
      "epoch": 1.6314811373350677,
      "grad_norm": 0.7209110260009766,
      "learning_rate": 1.4968325849304743e-05,
      "loss": 1.0118,
      "step": 8872
    },
    {
      "epoch": 1.631666976398439,
      "grad_norm": 0.6772007346153259,
      "learning_rate": 1.4952938218780488e-05,
      "loss": 0.6785,
      "step": 8873
    },
    {
      "epoch": 1.63185281546181,
      "grad_norm": 0.6663495302200317,
      "learning_rate": 1.4937557862515239e-05,
      "loss": 1.0211,
      "step": 8874
    },
    {
      "epoch": 1.6320386545251813,
      "grad_norm": 0.6624771356582642,
      "learning_rate": 1.4922184781824545e-05,
      "loss": 0.9342,
      "step": 8875
    },
    {
      "epoch": 1.6322244935885521,
      "grad_norm": 0.656817615032196,
      "learning_rate": 1.490681897802323e-05,
      "loss": 1.0656,
      "step": 8876
    },
    {
      "epoch": 1.6324103326519235,
      "grad_norm": 0.7161831259727478,
      "learning_rate": 1.4891460452425598e-05,
      "loss": 1.0012,
      "step": 8877
    },
    {
      "epoch": 1.6325961717152946,
      "grad_norm": 0.6671724915504456,
      "learning_rate": 1.4876109206345301e-05,
      "loss": 1.1334,
      "step": 8878
    },
    {
      "epoch": 1.6327820107786657,
      "grad_norm": 1.1625293493270874,
      "learning_rate": 1.4860765241095387e-05,
      "loss": 0.7399,
      "step": 8879
    },
    {
      "epoch": 1.6329678498420368,
      "grad_norm": 0.7702340483665466,
      "learning_rate": 1.4845428557988206e-05,
      "loss": 0.823,
      "step": 8880
    },
    {
      "epoch": 1.633153688905408,
      "grad_norm": 0.7329745888710022,
      "learning_rate": 1.4830099158335563e-05,
      "loss": 1.1374,
      "step": 8881
    },
    {
      "epoch": 1.633339527968779,
      "grad_norm": 0.6118117570877075,
      "learning_rate": 1.4814777043448636e-05,
      "loss": 0.6717,
      "step": 8882
    },
    {
      "epoch": 1.6335253670321501,
      "grad_norm": 0.7152222990989685,
      "learning_rate": 1.4799462214637916e-05,
      "loss": 0.9746,
      "step": 8883
    },
    {
      "epoch": 1.6337112060955212,
      "grad_norm": 0.674939751625061,
      "learning_rate": 1.4784154673213357e-05,
      "loss": 0.9757,
      "step": 8884
    },
    {
      "epoch": 1.6338970451588923,
      "grad_norm": 0.7836318016052246,
      "learning_rate": 1.4768854420484214e-05,
      "loss": 1.039,
      "step": 8885
    },
    {
      "epoch": 1.6340828842222637,
      "grad_norm": 0.7620517611503601,
      "learning_rate": 1.4753561457759168e-05,
      "loss": 0.9907,
      "step": 8886
    },
    {
      "epoch": 1.6342687232856346,
      "grad_norm": 0.7664940357208252,
      "learning_rate": 1.4738275786346234e-05,
      "loss": 0.8479,
      "step": 8887
    },
    {
      "epoch": 1.6344545623490059,
      "grad_norm": 0.8321072459220886,
      "learning_rate": 1.4722997407552852e-05,
      "loss": 1.0961,
      "step": 8888
    },
    {
      "epoch": 1.6346404014123768,
      "grad_norm": 0.7598660588264465,
      "learning_rate": 1.4707726322685822e-05,
      "loss": 1.2227,
      "step": 8889
    },
    {
      "epoch": 1.634826240475748,
      "grad_norm": 0.7904730439186096,
      "learning_rate": 1.4692462533051288e-05,
      "loss": 0.7229,
      "step": 8890
    },
    {
      "epoch": 1.635012079539119,
      "grad_norm": 0.8430718183517456,
      "learning_rate": 1.4677206039954805e-05,
      "loss": 0.8934,
      "step": 8891
    },
    {
      "epoch": 1.6351979186024903,
      "grad_norm": 0.6781235337257385,
      "learning_rate": 1.4661956844701297e-05,
      "loss": 0.7438,
      "step": 8892
    },
    {
      "epoch": 1.6353837576658614,
      "grad_norm": 0.7655029892921448,
      "learning_rate": 1.4646714948595076e-05,
      "loss": 0.9364,
      "step": 8893
    },
    {
      "epoch": 1.6355695967292325,
      "grad_norm": 0.809462308883667,
      "learning_rate": 1.4631480352939787e-05,
      "loss": 0.8887,
      "step": 8894
    },
    {
      "epoch": 1.6357554357926036,
      "grad_norm": 0.6403674483299255,
      "learning_rate": 1.4616253059038477e-05,
      "loss": 1.1736,
      "step": 8895
    },
    {
      "epoch": 1.6359412748559747,
      "grad_norm": 0.7233485579490662,
      "learning_rate": 1.4601033068193615e-05,
      "loss": 0.8797,
      "step": 8896
    },
    {
      "epoch": 1.6361271139193458,
      "grad_norm": 1.15740966796875,
      "learning_rate": 1.4585820381706927e-05,
      "loss": 1.3284,
      "step": 8897
    },
    {
      "epoch": 1.636312952982717,
      "grad_norm": 0.6786386966705322,
      "learning_rate": 1.457061500087966e-05,
      "loss": 0.7029,
      "step": 8898
    },
    {
      "epoch": 1.636498792046088,
      "grad_norm": 0.8133037090301514,
      "learning_rate": 1.455541692701231e-05,
      "loss": 1.1436,
      "step": 8899
    },
    {
      "epoch": 1.6366846311094592,
      "grad_norm": 0.6569185256958008,
      "learning_rate": 1.4540226161404835e-05,
      "loss": 0.8784,
      "step": 8900
    },
    {
      "epoch": 1.6368704701728305,
      "grad_norm": 0.6427562832832336,
      "learning_rate": 1.4525042705356506e-05,
      "loss": 0.9798,
      "step": 8901
    },
    {
      "epoch": 1.6370563092362014,
      "grad_norm": 0.616054117679596,
      "learning_rate": 1.4509866560166007e-05,
      "loss": 0.9621,
      "step": 8902
    },
    {
      "epoch": 1.6372421482995727,
      "grad_norm": 0.6826965808868408,
      "learning_rate": 1.44946977271314e-05,
      "loss": 0.9178,
      "step": 8903
    },
    {
      "epoch": 1.6374279873629436,
      "grad_norm": 0.8229233026504517,
      "learning_rate": 1.4479536207550093e-05,
      "loss": 1.1039,
      "step": 8904
    },
    {
      "epoch": 1.637613826426315,
      "grad_norm": 0.7446808815002441,
      "learning_rate": 1.446438200271888e-05,
      "loss": 1.1094,
      "step": 8905
    },
    {
      "epoch": 1.6377996654896858,
      "grad_norm": 0.7734174132347107,
      "learning_rate": 1.4449235113933934e-05,
      "loss": 0.8696,
      "step": 8906
    },
    {
      "epoch": 1.6379855045530571,
      "grad_norm": 0.6430695652961731,
      "learning_rate": 1.4434095542490844e-05,
      "loss": 1.0186,
      "step": 8907
    },
    {
      "epoch": 1.638171343616428,
      "grad_norm": 0.7133775353431702,
      "learning_rate": 1.4418963289684462e-05,
      "loss": 1.1213,
      "step": 8908
    },
    {
      "epoch": 1.6383571826797994,
      "grad_norm": 0.5039517283439636,
      "learning_rate": 1.4403838356809119e-05,
      "loss": 0.6748,
      "step": 8909
    },
    {
      "epoch": 1.6385430217431705,
      "grad_norm": 0.6650997996330261,
      "learning_rate": 1.4388720745158502e-05,
      "loss": 0.96,
      "step": 8910
    },
    {
      "epoch": 1.6387288608065416,
      "grad_norm": 0.6760223507881165,
      "learning_rate": 1.437361045602561e-05,
      "loss": 0.9497,
      "step": 8911
    },
    {
      "epoch": 1.6389146998699127,
      "grad_norm": 0.7307868599891663,
      "learning_rate": 1.4358507490702899e-05,
      "loss": 1.1226,
      "step": 8912
    },
    {
      "epoch": 1.6391005389332838,
      "grad_norm": 0.7235276103019714,
      "learning_rate": 1.4343411850482114e-05,
      "loss": 0.9746,
      "step": 8913
    },
    {
      "epoch": 1.6392863779966549,
      "grad_norm": 0.716784656047821,
      "learning_rate": 1.4328323536654442e-05,
      "loss": 1.049,
      "step": 8914
    },
    {
      "epoch": 1.639472217060026,
      "grad_norm": 0.8350142240524292,
      "learning_rate": 1.4313242550510442e-05,
      "loss": 0.9188,
      "step": 8915
    },
    {
      "epoch": 1.639658056123397,
      "grad_norm": 0.7000654339790344,
      "learning_rate": 1.429816889333997e-05,
      "loss": 0.9902,
      "step": 8916
    },
    {
      "epoch": 1.6398438951867682,
      "grad_norm": 0.6660363674163818,
      "learning_rate": 1.4283102566432371e-05,
      "loss": 0.9285,
      "step": 8917
    },
    {
      "epoch": 1.6400297342501395,
      "grad_norm": 0.6869511008262634,
      "learning_rate": 1.4268043571076239e-05,
      "loss": 1.1581,
      "step": 8918
    },
    {
      "epoch": 1.6402155733135104,
      "grad_norm": 0.7386857271194458,
      "learning_rate": 1.4252991908559665e-05,
      "loss": 1.1359,
      "step": 8919
    },
    {
      "epoch": 1.6404014123768818,
      "grad_norm": 0.7131685018539429,
      "learning_rate": 1.4237947580169986e-05,
      "loss": 0.9634,
      "step": 8920
    },
    {
      "epoch": 1.6405872514402526,
      "grad_norm": 0.6731940507888794,
      "learning_rate": 1.4222910587194005e-05,
      "loss": 0.9535,
      "step": 8921
    },
    {
      "epoch": 1.640773090503624,
      "grad_norm": 0.7020549774169922,
      "learning_rate": 1.4207880930917871e-05,
      "loss": 0.9597,
      "step": 8922
    },
    {
      "epoch": 1.6409589295669949,
      "grad_norm": 0.6156516075134277,
      "learning_rate": 1.4192858612627102e-05,
      "loss": 1.0362,
      "step": 8923
    },
    {
      "epoch": 1.6411447686303662,
      "grad_norm": 0.7188947200775146,
      "learning_rate": 1.4177843633606614e-05,
      "loss": 1.0978,
      "step": 8924
    },
    {
      "epoch": 1.641330607693737,
      "grad_norm": 0.6286084055900574,
      "learning_rate": 1.4162835995140622e-05,
      "loss": 0.998,
      "step": 8925
    },
    {
      "epoch": 1.6415164467571084,
      "grad_norm": 0.6593037843704224,
      "learning_rate": 1.4147835698512812e-05,
      "loss": 0.9243,
      "step": 8926
    },
    {
      "epoch": 1.6417022858204795,
      "grad_norm": 0.6840078234672546,
      "learning_rate": 1.4132842745006148e-05,
      "loss": 0.8513,
      "step": 8927
    },
    {
      "epoch": 1.6418881248838506,
      "grad_norm": 0.7872484922409058,
      "learning_rate": 1.4117857135903013e-05,
      "loss": 0.8972,
      "step": 8928
    },
    {
      "epoch": 1.6420739639472217,
      "grad_norm": 0.7854396104812622,
      "learning_rate": 1.4102878872485203e-05,
      "loss": 0.9316,
      "step": 8929
    },
    {
      "epoch": 1.6422598030105928,
      "grad_norm": 0.7035794854164124,
      "learning_rate": 1.4087907956033785e-05,
      "loss": 1.0664,
      "step": 8930
    },
    {
      "epoch": 1.642445642073964,
      "grad_norm": 0.7821934819221497,
      "learning_rate": 1.4072944387829313e-05,
      "loss": 0.9837,
      "step": 8931
    },
    {
      "epoch": 1.642631481137335,
      "grad_norm": 0.7449780106544495,
      "learning_rate": 1.4057988169151592e-05,
      "loss": 1.0677,
      "step": 8932
    },
    {
      "epoch": 1.6428173202007061,
      "grad_norm": 0.7577154636383057,
      "learning_rate": 1.4043039301279903e-05,
      "loss": 1.1414,
      "step": 8933
    },
    {
      "epoch": 1.6430031592640773,
      "grad_norm": 0.6866757273674011,
      "learning_rate": 1.4028097785492823e-05,
      "loss": 0.9267,
      "step": 8934
    },
    {
      "epoch": 1.6431889983274486,
      "grad_norm": 0.7813276648521423,
      "learning_rate": 1.4013163623068348e-05,
      "loss": 0.8713,
      "step": 8935
    },
    {
      "epoch": 1.6433748373908195,
      "grad_norm": 0.5878521800041199,
      "learning_rate": 1.3998236815283838e-05,
      "loss": 0.7959,
      "step": 8936
    },
    {
      "epoch": 1.6435606764541908,
      "grad_norm": 0.6542325615882874,
      "learning_rate": 1.3983317363416005e-05,
      "loss": 0.8847,
      "step": 8937
    },
    {
      "epoch": 1.6437465155175617,
      "grad_norm": 0.7386040687561035,
      "learning_rate": 1.3968405268740959e-05,
      "loss": 0.7502,
      "step": 8938
    },
    {
      "epoch": 1.643932354580933,
      "grad_norm": 0.657832145690918,
      "learning_rate": 1.3953500532534137e-05,
      "loss": 0.9111,
      "step": 8939
    },
    {
      "epoch": 1.644118193644304,
      "grad_norm": 0.8460830450057983,
      "learning_rate": 1.3938603156070395e-05,
      "loss": 1.0726,
      "step": 8940
    },
    {
      "epoch": 1.6443040327076752,
      "grad_norm": 0.6688886880874634,
      "learning_rate": 1.3923713140623918e-05,
      "loss": 0.6267,
      "step": 8941
    },
    {
      "epoch": 1.644489871771046,
      "grad_norm": 0.7037612199783325,
      "learning_rate": 1.3908830487468283e-05,
      "loss": 0.9604,
      "step": 8942
    },
    {
      "epoch": 1.6446757108344174,
      "grad_norm": 0.7139345407485962,
      "learning_rate": 1.3893955197876474e-05,
      "loss": 0.6997,
      "step": 8943
    },
    {
      "epoch": 1.6448615498977885,
      "grad_norm": 0.7229803800582886,
      "learning_rate": 1.387908727312074e-05,
      "loss": 0.9263,
      "step": 8944
    },
    {
      "epoch": 1.6450473889611597,
      "grad_norm": 0.6166016459465027,
      "learning_rate": 1.386422671447284e-05,
      "loss": 0.794,
      "step": 8945
    },
    {
      "epoch": 1.6452332280245308,
      "grad_norm": 0.6664713621139526,
      "learning_rate": 1.3849373523203769e-05,
      "loss": 0.8972,
      "step": 8946
    },
    {
      "epoch": 1.6454190670879019,
      "grad_norm": 0.7948108911514282,
      "learning_rate": 1.3834527700583988e-05,
      "loss": 0.8877,
      "step": 8947
    },
    {
      "epoch": 1.645604906151273,
      "grad_norm": 1.114028811454773,
      "learning_rate": 1.3819689247883271e-05,
      "loss": 0.9342,
      "step": 8948
    },
    {
      "epoch": 1.645790745214644,
      "grad_norm": 0.7399374842643738,
      "learning_rate": 1.3804858166370782e-05,
      "loss": 0.9073,
      "step": 8949
    },
    {
      "epoch": 1.6459765842780152,
      "grad_norm": 0.7469755411148071,
      "learning_rate": 1.3790034457315082e-05,
      "loss": 1.078,
      "step": 8950
    },
    {
      "epoch": 1.6461624233413863,
      "grad_norm": 0.6429402828216553,
      "learning_rate": 1.3775218121984046e-05,
      "loss": 0.8276,
      "step": 8951
    },
    {
      "epoch": 1.6463482624047576,
      "grad_norm": 0.6140831112861633,
      "learning_rate": 1.3760409161644982e-05,
      "loss": 0.952,
      "step": 8952
    },
    {
      "epoch": 1.6465341014681285,
      "grad_norm": 0.6000019907951355,
      "learning_rate": 1.3745607577564501e-05,
      "loss": 0.6117,
      "step": 8953
    },
    {
      "epoch": 1.6467199405314998,
      "grad_norm": 0.6426273584365845,
      "learning_rate": 1.3730813371008633e-05,
      "loss": 1.0371,
      "step": 8954
    },
    {
      "epoch": 1.6469057795948707,
      "grad_norm": 0.7724543809890747,
      "learning_rate": 1.3716026543242732e-05,
      "loss": 0.9657,
      "step": 8955
    },
    {
      "epoch": 1.647091618658242,
      "grad_norm": 0.6861767768859863,
      "learning_rate": 1.3701247095531566e-05,
      "loss": 0.9033,
      "step": 8956
    },
    {
      "epoch": 1.647277457721613,
      "grad_norm": 0.7680606842041016,
      "learning_rate": 1.3686475029139267e-05,
      "loss": 0.8494,
      "step": 8957
    },
    {
      "epoch": 1.6474632967849843,
      "grad_norm": 0.7061701416969299,
      "learning_rate": 1.3671710345329292e-05,
      "loss": 0.9876,
      "step": 8958
    },
    {
      "epoch": 1.6476491358483554,
      "grad_norm": 0.6801935434341431,
      "learning_rate": 1.365695304536454e-05,
      "loss": 0.8746,
      "step": 8959
    },
    {
      "epoch": 1.6478349749117265,
      "grad_norm": 0.6824638247489929,
      "learning_rate": 1.364220313050717e-05,
      "loss": 0.7854,
      "step": 8960
    },
    {
      "epoch": 1.6480208139750976,
      "grad_norm": 0.6700693368911743,
      "learning_rate": 1.3627460602018838e-05,
      "loss": 0.8267,
      "step": 8961
    },
    {
      "epoch": 1.6482066530384687,
      "grad_norm": 0.7284310460090637,
      "learning_rate": 1.3612725461160459e-05,
      "loss": 1.0115,
      "step": 8962
    },
    {
      "epoch": 1.6483924921018398,
      "grad_norm": 0.6551657915115356,
      "learning_rate": 1.3597997709192378e-05,
      "loss": 1.084,
      "step": 8963
    },
    {
      "epoch": 1.648578331165211,
      "grad_norm": 0.7529725432395935,
      "learning_rate": 1.3583277347374301e-05,
      "loss": 0.8888,
      "step": 8964
    },
    {
      "epoch": 1.648764170228582,
      "grad_norm": 0.848233699798584,
      "learning_rate": 1.3568564376965277e-05,
      "loss": 0.8482,
      "step": 8965
    },
    {
      "epoch": 1.6489500092919531,
      "grad_norm": 0.6673817038536072,
      "learning_rate": 1.355385879922374e-05,
      "loss": 0.8099,
      "step": 8966
    },
    {
      "epoch": 1.6491358483553245,
      "grad_norm": 0.7535316348075867,
      "learning_rate": 1.3539160615407486e-05,
      "loss": 1.0279,
      "step": 8967
    },
    {
      "epoch": 1.6493216874186953,
      "grad_norm": 0.5885194540023804,
      "learning_rate": 1.3524469826773723e-05,
      "loss": 0.869,
      "step": 8968
    },
    {
      "epoch": 1.6495075264820667,
      "grad_norm": 0.7876520752906799,
      "learning_rate": 1.3509786434578931e-05,
      "loss": 1.0675,
      "step": 8969
    },
    {
      "epoch": 1.6496933655454376,
      "grad_norm": 0.6128770709037781,
      "learning_rate": 1.349511044007904e-05,
      "loss": 0.8816,
      "step": 8970
    },
    {
      "epoch": 1.6498792046088089,
      "grad_norm": 0.6121591925621033,
      "learning_rate": 1.3480441844529324e-05,
      "loss": 0.716,
      "step": 8971
    },
    {
      "epoch": 1.6500650436721798,
      "grad_norm": 0.7618534564971924,
      "learning_rate": 1.34657806491844e-05,
      "loss": 0.9837,
      "step": 8972
    },
    {
      "epoch": 1.650250882735551,
      "grad_norm": 0.7061711549758911,
      "learning_rate": 1.345112685529829e-05,
      "loss": 0.9989,
      "step": 8973
    },
    {
      "epoch": 1.650436721798922,
      "grad_norm": 0.6352358460426331,
      "learning_rate": 1.3436480464124346e-05,
      "loss": 0.9719,
      "step": 8974
    },
    {
      "epoch": 1.6506225608622933,
      "grad_norm": 0.6654130220413208,
      "learning_rate": 1.3421841476915343e-05,
      "loss": 1.0551,
      "step": 8975
    },
    {
      "epoch": 1.6508083999256644,
      "grad_norm": 0.7092660665512085,
      "learning_rate": 1.3407209894923323e-05,
      "loss": 1.1038,
      "step": 8976
    },
    {
      "epoch": 1.6509942389890355,
      "grad_norm": 0.7314668297767639,
      "learning_rate": 1.3392585719399797e-05,
      "loss": 0.9794,
      "step": 8977
    },
    {
      "epoch": 1.6511800780524066,
      "grad_norm": 0.739293098449707,
      "learning_rate": 1.3377968951595621e-05,
      "loss": 1.0016,
      "step": 8978
    },
    {
      "epoch": 1.6513659171157777,
      "grad_norm": 0.6878502368927002,
      "learning_rate": 1.336335959276096e-05,
      "loss": 0.9889,
      "step": 8979
    },
    {
      "epoch": 1.6515517561791488,
      "grad_norm": 0.7072255611419678,
      "learning_rate": 1.3348757644145392e-05,
      "loss": 0.8155,
      "step": 8980
    },
    {
      "epoch": 1.65173759524252,
      "grad_norm": 0.7043284177780151,
      "learning_rate": 1.3334163106997865e-05,
      "loss": 1.0011,
      "step": 8981
    },
    {
      "epoch": 1.651923434305891,
      "grad_norm": 0.8223550915718079,
      "learning_rate": 1.331957598256669e-05,
      "loss": 0.8955,
      "step": 8982
    },
    {
      "epoch": 1.6521092733692622,
      "grad_norm": 0.6612815856933594,
      "learning_rate": 1.3304996272099512e-05,
      "loss": 0.8959,
      "step": 8983
    },
    {
      "epoch": 1.6522951124326335,
      "grad_norm": 0.7884023785591125,
      "learning_rate": 1.3290423976843369e-05,
      "loss": 0.8742,
      "step": 8984
    },
    {
      "epoch": 1.6524809514960044,
      "grad_norm": 0.837375819683075,
      "learning_rate": 1.3275859098044684e-05,
      "loss": 0.9496,
      "step": 8985
    },
    {
      "epoch": 1.6526667905593757,
      "grad_norm": 0.6119824051856995,
      "learning_rate": 1.3261301636949197e-05,
      "loss": 0.7175,
      "step": 8986
    },
    {
      "epoch": 1.6528526296227466,
      "grad_norm": 0.7192531824111938,
      "learning_rate": 1.3246751594802064e-05,
      "loss": 0.9797,
      "step": 8987
    },
    {
      "epoch": 1.653038468686118,
      "grad_norm": 0.9993611574172974,
      "learning_rate": 1.3232208972847737e-05,
      "loss": 1.2924,
      "step": 8988
    },
    {
      "epoch": 1.6532243077494888,
      "grad_norm": 0.6902458667755127,
      "learning_rate": 1.321767377233013e-05,
      "loss": 0.8782,
      "step": 8989
    },
    {
      "epoch": 1.6534101468128601,
      "grad_norm": 0.6710980534553528,
      "learning_rate": 1.320314599449244e-05,
      "loss": 0.9699,
      "step": 8990
    },
    {
      "epoch": 1.653595985876231,
      "grad_norm": 0.6646018028259277,
      "learning_rate": 1.3188625640577246e-05,
      "loss": 1.1329,
      "step": 8991
    },
    {
      "epoch": 1.6537818249396024,
      "grad_norm": 0.6741959452629089,
      "learning_rate": 1.3174112711826559e-05,
      "loss": 0.8457,
      "step": 8992
    },
    {
      "epoch": 1.6539676640029735,
      "grad_norm": 0.8614079356193542,
      "learning_rate": 1.3159607209481639e-05,
      "loss": 1.0179,
      "step": 8993
    },
    {
      "epoch": 1.6541535030663446,
      "grad_norm": 0.7563652992248535,
      "learning_rate": 1.3145109134783207e-05,
      "loss": 1.1809,
      "step": 8994
    },
    {
      "epoch": 1.6543393421297157,
      "grad_norm": 0.7481111884117126,
      "learning_rate": 1.3130618488971302e-05,
      "loss": 0.9286,
      "step": 8995
    },
    {
      "epoch": 1.6545251811930868,
      "grad_norm": 0.8062092065811157,
      "learning_rate": 1.3116135273285369e-05,
      "loss": 0.8791,
      "step": 8996
    },
    {
      "epoch": 1.6547110202564579,
      "grad_norm": 0.7598450183868408,
      "learning_rate": 1.3101659488964146e-05,
      "loss": 0.9577,
      "step": 8997
    },
    {
      "epoch": 1.654896859319829,
      "grad_norm": 0.7781131267547607,
      "learning_rate": 1.3087191137245802e-05,
      "loss": 0.972,
      "step": 8998
    },
    {
      "epoch": 1.6550826983832,
      "grad_norm": 0.6933823823928833,
      "learning_rate": 1.3072730219367869e-05,
      "loss": 0.8878,
      "step": 8999
    },
    {
      "epoch": 1.6552685374465712,
      "grad_norm": 0.6313965320587158,
      "learning_rate": 1.3058276736567165e-05,
      "loss": 0.8923,
      "step": 9000
    },
    {
      "epoch": 1.6554543765099425,
      "grad_norm": 0.7283336520195007,
      "learning_rate": 1.3043830690079984e-05,
      "loss": 0.885,
      "step": 9001
    },
    {
      "epoch": 1.6556402155733134,
      "grad_norm": 0.6406028270721436,
      "learning_rate": 1.3029392081141878e-05,
      "loss": 1.0113,
      "step": 9002
    },
    {
      "epoch": 1.6558260546366848,
      "grad_norm": 0.8105238080024719,
      "learning_rate": 1.301496091098786e-05,
      "loss": 1.1788,
      "step": 9003
    },
    {
      "epoch": 1.6560118937000556,
      "grad_norm": 0.6627447605133057,
      "learning_rate": 1.3000537180852212e-05,
      "loss": 0.8417,
      "step": 9004
    },
    {
      "epoch": 1.656197732763427,
      "grad_norm": 0.6995481848716736,
      "learning_rate": 1.2986120891968645e-05,
      "loss": 0.9624,
      "step": 9005
    },
    {
      "epoch": 1.6563835718267979,
      "grad_norm": 0.8995388150215149,
      "learning_rate": 1.297171204557025e-05,
      "loss": 1.0961,
      "step": 9006
    },
    {
      "epoch": 1.6565694108901692,
      "grad_norm": 0.6590665578842163,
      "learning_rate": 1.2957310642889387e-05,
      "loss": 0.9818,
      "step": 9007
    },
    {
      "epoch": 1.6567552499535403,
      "grad_norm": 0.6268938183784485,
      "learning_rate": 1.2942916685157868e-05,
      "loss": 0.8111,
      "step": 9008
    },
    {
      "epoch": 1.6569410890169114,
      "grad_norm": 0.7919785976409912,
      "learning_rate": 1.2928530173606867e-05,
      "loss": 0.9393,
      "step": 9009
    },
    {
      "epoch": 1.6571269280802825,
      "grad_norm": 0.7027867436408997,
      "learning_rate": 1.2914151109466843e-05,
      "loss": 0.6645,
      "step": 9010
    },
    {
      "epoch": 1.6573127671436536,
      "grad_norm": 0.6374213695526123,
      "learning_rate": 1.289977949396769e-05,
      "loss": 0.8168,
      "step": 9011
    },
    {
      "epoch": 1.6574986062070247,
      "grad_norm": 0.7896588444709778,
      "learning_rate": 1.2885415328338645e-05,
      "loss": 0.9752,
      "step": 9012
    },
    {
      "epoch": 1.6576844452703958,
      "grad_norm": 0.7199686765670776,
      "learning_rate": 1.2871058613808329e-05,
      "loss": 0.8953,
      "step": 9013
    },
    {
      "epoch": 1.657870284333767,
      "grad_norm": 0.8703217506408691,
      "learning_rate": 1.2856709351604657e-05,
      "loss": 0.7636,
      "step": 9014
    },
    {
      "epoch": 1.658056123397138,
      "grad_norm": 0.6465247869491577,
      "learning_rate": 1.2842367542955003e-05,
      "loss": 0.8154,
      "step": 9015
    },
    {
      "epoch": 1.6582419624605091,
      "grad_norm": 0.6402195692062378,
      "learning_rate": 1.2828033189085997e-05,
      "loss": 0.7662,
      "step": 9016
    },
    {
      "epoch": 1.6584278015238803,
      "grad_norm": 0.7437441945075989,
      "learning_rate": 1.2813706291223748e-05,
      "loss": 0.9204,
      "step": 9017
    },
    {
      "epoch": 1.6586136405872516,
      "grad_norm": 0.7095872759819031,
      "learning_rate": 1.2799386850593597e-05,
      "loss": 0.9336,
      "step": 9018
    },
    {
      "epoch": 1.6587994796506225,
      "grad_norm": 0.7508102655410767,
      "learning_rate": 1.2785074868420366e-05,
      "loss": 1.1321,
      "step": 9019
    },
    {
      "epoch": 1.6589853187139938,
      "grad_norm": 0.6868159770965576,
      "learning_rate": 1.2770770345928196e-05,
      "loss": 0.9345,
      "step": 9020
    },
    {
      "epoch": 1.6591711577773647,
      "grad_norm": 0.6406965851783752,
      "learning_rate": 1.2756473284340543e-05,
      "loss": 1.02,
      "step": 9021
    },
    {
      "epoch": 1.659356996840736,
      "grad_norm": 0.6748953461647034,
      "learning_rate": 1.2742183684880304e-05,
      "loss": 0.959,
      "step": 9022
    },
    {
      "epoch": 1.659542835904107,
      "grad_norm": 0.7843619585037231,
      "learning_rate": 1.2727901548769672e-05,
      "loss": 0.8909,
      "step": 9023
    },
    {
      "epoch": 1.6597286749674782,
      "grad_norm": 0.9022567272186279,
      "learning_rate": 1.2713626877230233e-05,
      "loss": 0.9762,
      "step": 9024
    },
    {
      "epoch": 1.6599145140308493,
      "grad_norm": 0.6865164041519165,
      "learning_rate": 1.2699359671482947e-05,
      "loss": 0.9494,
      "step": 9025
    },
    {
      "epoch": 1.6601003530942204,
      "grad_norm": 0.8706424236297607,
      "learning_rate": 1.2685099932748101e-05,
      "loss": 1.1028,
      "step": 9026
    },
    {
      "epoch": 1.6602861921575915,
      "grad_norm": 0.576554536819458,
      "learning_rate": 1.26708476622454e-05,
      "loss": 0.6907,
      "step": 9027
    },
    {
      "epoch": 1.6604720312209627,
      "grad_norm": 0.6635376811027527,
      "learning_rate": 1.265660286119381e-05,
      "loss": 0.9428,
      "step": 9028
    },
    {
      "epoch": 1.6606578702843338,
      "grad_norm": 0.8346258401870728,
      "learning_rate": 1.2642365530811772e-05,
      "loss": 0.976,
      "step": 9029
    },
    {
      "epoch": 1.6608437093477049,
      "grad_norm": 0.779428243637085,
      "learning_rate": 1.2628135672317009e-05,
      "loss": 0.887,
      "step": 9030
    },
    {
      "epoch": 1.661029548411076,
      "grad_norm": 0.7409526109695435,
      "learning_rate": 1.2613913286926648e-05,
      "loss": 0.9256,
      "step": 9031
    },
    {
      "epoch": 1.661215387474447,
      "grad_norm": 0.7076467871665955,
      "learning_rate": 1.2599698375857127e-05,
      "loss": 0.9611,
      "step": 9032
    },
    {
      "epoch": 1.6614012265378184,
      "grad_norm": 0.796972393989563,
      "learning_rate": 1.2585490940324297e-05,
      "loss": 1.0653,
      "step": 9033
    },
    {
      "epoch": 1.6615870656011893,
      "grad_norm": 0.7058619260787964,
      "learning_rate": 1.2571290981543382e-05,
      "loss": 1.1194,
      "step": 9034
    },
    {
      "epoch": 1.6617729046645606,
      "grad_norm": 0.6679715514183044,
      "learning_rate": 1.2557098500728892e-05,
      "loss": 0.7929,
      "step": 9035
    },
    {
      "epoch": 1.6619587437279315,
      "grad_norm": 0.6468355655670166,
      "learning_rate": 1.2542913499094766e-05,
      "loss": 1.1127,
      "step": 9036
    },
    {
      "epoch": 1.6621445827913028,
      "grad_norm": 0.854033350944519,
      "learning_rate": 1.2528735977854255e-05,
      "loss": 1.2172,
      "step": 9037
    },
    {
      "epoch": 1.6623304218546737,
      "grad_norm": 0.7941082715988159,
      "learning_rate": 1.2514565938220003e-05,
      "loss": 0.7856,
      "step": 9038
    },
    {
      "epoch": 1.662516260918045,
      "grad_norm": 1.2629947662353516,
      "learning_rate": 1.2500403381404003e-05,
      "loss": 1.1651,
      "step": 9039
    },
    {
      "epoch": 1.662702099981416,
      "grad_norm": 0.6334301829338074,
      "learning_rate": 1.248624830861762e-05,
      "loss": 1.0095,
      "step": 9040
    },
    {
      "epoch": 1.6628879390447873,
      "grad_norm": 0.8614044785499573,
      "learning_rate": 1.2472100721071578e-05,
      "loss": 1.084,
      "step": 9041
    },
    {
      "epoch": 1.6630737781081584,
      "grad_norm": 0.6715635061264038,
      "learning_rate": 1.2457960619975917e-05,
      "loss": 0.8696,
      "step": 9042
    },
    {
      "epoch": 1.6632596171715295,
      "grad_norm": 0.6359160542488098,
      "learning_rate": 1.2443828006540103e-05,
      "loss": 1.0494,
      "step": 9043
    },
    {
      "epoch": 1.6634454562349006,
      "grad_norm": 0.8149633407592773,
      "learning_rate": 1.2429702881972905e-05,
      "loss": 1.1387,
      "step": 9044
    },
    {
      "epoch": 1.6636312952982717,
      "grad_norm": 0.7104032635688782,
      "learning_rate": 1.2415585247482498e-05,
      "loss": 1.1578,
      "step": 9045
    },
    {
      "epoch": 1.6638171343616428,
      "grad_norm": 0.8964525461196899,
      "learning_rate": 1.240147510427635e-05,
      "loss": 1.0038,
      "step": 9046
    },
    {
      "epoch": 1.664002973425014,
      "grad_norm": 0.6363485455513,
      "learning_rate": 1.2387372453561364e-05,
      "loss": 0.9393,
      "step": 9047
    },
    {
      "epoch": 1.664188812488385,
      "grad_norm": 0.6944891810417175,
      "learning_rate": 1.2373277296543795e-05,
      "loss": 0.8978,
      "step": 9048
    },
    {
      "epoch": 1.6643746515517561,
      "grad_norm": 0.766639769077301,
      "learning_rate": 1.235918963442918e-05,
      "loss": 0.9532,
      "step": 9049
    },
    {
      "epoch": 1.6645604906151275,
      "grad_norm": 0.8323293328285217,
      "learning_rate": 1.2345109468422523e-05,
      "loss": 0.8493,
      "step": 9050
    },
    {
      "epoch": 1.6647463296784983,
      "grad_norm": 0.7760543823242188,
      "learning_rate": 1.233103679972807e-05,
      "loss": 0.7718,
      "step": 9051
    },
    {
      "epoch": 1.6649321687418697,
      "grad_norm": 0.7688063979148865,
      "learning_rate": 1.2316971629549511e-05,
      "loss": 1.2442,
      "step": 9052
    },
    {
      "epoch": 1.6651180078052406,
      "grad_norm": 0.783732533454895,
      "learning_rate": 1.230291395908989e-05,
      "loss": 1.0332,
      "step": 9053
    },
    {
      "epoch": 1.6653038468686119,
      "grad_norm": 0.7472023367881775,
      "learning_rate": 1.2288863789551586e-05,
      "loss": 0.9761,
      "step": 9054
    },
    {
      "epoch": 1.6654896859319828,
      "grad_norm": 0.6275432705879211,
      "learning_rate": 1.2274821122136316e-05,
      "loss": 0.8392,
      "step": 9055
    },
    {
      "epoch": 1.665675524995354,
      "grad_norm": 0.7407889366149902,
      "learning_rate": 1.2260785958045196e-05,
      "loss": 1.0087,
      "step": 9056
    },
    {
      "epoch": 1.665861364058725,
      "grad_norm": 0.6514436602592468,
      "learning_rate": 1.22467582984787e-05,
      "loss": 0.8457,
      "step": 9057
    },
    {
      "epoch": 1.6660472031220963,
      "grad_norm": 0.666793167591095,
      "learning_rate": 1.2232738144636613e-05,
      "loss": 0.8491,
      "step": 9058
    },
    {
      "epoch": 1.6662330421854674,
      "grad_norm": 0.708659827709198,
      "learning_rate": 1.221872549771813e-05,
      "loss": 1.1277,
      "step": 9059
    },
    {
      "epoch": 1.6664188812488385,
      "grad_norm": 0.7047358155250549,
      "learning_rate": 1.2204720358921761e-05,
      "loss": 1.0066,
      "step": 9060
    },
    {
      "epoch": 1.6666047203122096,
      "grad_norm": 0.7291968464851379,
      "learning_rate": 1.2190722729445415e-05,
      "loss": 1.1258,
      "step": 9061
    },
    {
      "epoch": 1.6667905593755807,
      "grad_norm": 0.7509036064147949,
      "learning_rate": 1.2176732610486353e-05,
      "loss": 0.7721,
      "step": 9062
    },
    {
      "epoch": 1.6669763984389518,
      "grad_norm": 0.6720606684684753,
      "learning_rate": 1.216275000324114e-05,
      "loss": 1.0531,
      "step": 9063
    },
    {
      "epoch": 1.667162237502323,
      "grad_norm": 0.6274673938751221,
      "learning_rate": 1.214877490890578e-05,
      "loss": 0.8732,
      "step": 9064
    },
    {
      "epoch": 1.667348076565694,
      "grad_norm": 0.7732565402984619,
      "learning_rate": 1.2134807328675557e-05,
      "loss": 1.1978,
      "step": 9065
    },
    {
      "epoch": 1.6675339156290652,
      "grad_norm": 0.6812934279441833,
      "learning_rate": 1.2120847263745183e-05,
      "loss": 0.9908,
      "step": 9066
    },
    {
      "epoch": 1.6677197546924365,
      "grad_norm": 0.7481977939605713,
      "learning_rate": 1.2106894715308647e-05,
      "loss": 0.9184,
      "step": 9067
    },
    {
      "epoch": 1.6679055937558074,
      "grad_norm": 0.6654353141784668,
      "learning_rate": 1.2092949684559373e-05,
      "loss": 0.9962,
      "step": 9068
    },
    {
      "epoch": 1.6680914328191787,
      "grad_norm": 0.8279258608818054,
      "learning_rate": 1.2079012172690108e-05,
      "loss": 0.9949,
      "step": 9069
    },
    {
      "epoch": 1.6682772718825496,
      "grad_norm": 0.6551133990287781,
      "learning_rate": 1.206508218089295e-05,
      "loss": 0.7166,
      "step": 9070
    },
    {
      "epoch": 1.668463110945921,
      "grad_norm": 0.6765016913414001,
      "learning_rate": 1.2051159710359384e-05,
      "loss": 0.8067,
      "step": 9071
    },
    {
      "epoch": 1.6686489500092918,
      "grad_norm": 0.8082672953605652,
      "learning_rate": 1.2037244762280197e-05,
      "loss": 0.932,
      "step": 9072
    },
    {
      "epoch": 1.6688347890726631,
      "grad_norm": 0.7176616191864014,
      "learning_rate": 1.2023337337845596e-05,
      "loss": 0.8466,
      "step": 9073
    },
    {
      "epoch": 1.6690206281360342,
      "grad_norm": 0.7027639150619507,
      "learning_rate": 1.2009437438245074e-05,
      "loss": 0.8016,
      "step": 9074
    },
    {
      "epoch": 1.6692064671994054,
      "grad_norm": 0.5039098858833313,
      "learning_rate": 1.1995545064667535e-05,
      "loss": 0.6857,
      "step": 9075
    },
    {
      "epoch": 1.6693923062627765,
      "grad_norm": 0.7166251540184021,
      "learning_rate": 1.1981660218301261e-05,
      "loss": 0.9291,
      "step": 9076
    },
    {
      "epoch": 1.6695781453261476,
      "grad_norm": 0.6661301851272583,
      "learning_rate": 1.196778290033379e-05,
      "loss": 0.9587,
      "step": 9077
    },
    {
      "epoch": 1.6697639843895187,
      "grad_norm": 0.8017847537994385,
      "learning_rate": 1.1953913111952141e-05,
      "loss": 1.0492,
      "step": 9078
    },
    {
      "epoch": 1.6699498234528898,
      "grad_norm": 0.6611772179603577,
      "learning_rate": 1.1940050854342566e-05,
      "loss": 0.7787,
      "step": 9079
    },
    {
      "epoch": 1.6701356625162609,
      "grad_norm": 0.7013022899627686,
      "learning_rate": 1.1926196128690792e-05,
      "loss": 1.048,
      "step": 9080
    },
    {
      "epoch": 1.670321501579632,
      "grad_norm": 0.7118967175483704,
      "learning_rate": 1.1912348936181794e-05,
      "loss": 1.0242,
      "step": 9081
    },
    {
      "epoch": 1.6705073406430033,
      "grad_norm": 1.904453158378601,
      "learning_rate": 1.189850927799997e-05,
      "loss": 1.505,
      "step": 9082
    },
    {
      "epoch": 1.6706931797063742,
      "grad_norm": 0.6158105731010437,
      "learning_rate": 1.188467715532906e-05,
      "loss": 1.0896,
      "step": 9083
    },
    {
      "epoch": 1.6708790187697455,
      "grad_norm": 0.6351505517959595,
      "learning_rate": 1.1870852569352153e-05,
      "loss": 0.8779,
      "step": 9084
    },
    {
      "epoch": 1.6710648578331164,
      "grad_norm": 0.8218066692352295,
      "learning_rate": 1.1857035521251725e-05,
      "loss": 1.052,
      "step": 9085
    },
    {
      "epoch": 1.6712506968964878,
      "grad_norm": 0.7728452682495117,
      "learning_rate": 1.1843226012209529e-05,
      "loss": 0.9847,
      "step": 9086
    },
    {
      "epoch": 1.6714365359598586,
      "grad_norm": 0.7268964052200317,
      "learning_rate": 1.1829424043406755e-05,
      "loss": 0.7735,
      "step": 9087
    },
    {
      "epoch": 1.67162237502323,
      "grad_norm": 0.6986159682273865,
      "learning_rate": 1.1815629616023893e-05,
      "loss": 0.8668,
      "step": 9088
    },
    {
      "epoch": 1.6718082140866009,
      "grad_norm": 0.6351314783096313,
      "learning_rate": 1.180184273124082e-05,
      "loss": 0.7405,
      "step": 9089
    },
    {
      "epoch": 1.6719940531499722,
      "grad_norm": 0.6102209091186523,
      "learning_rate": 1.1788063390236782e-05,
      "loss": 0.9222,
      "step": 9090
    },
    {
      "epoch": 1.6721798922133433,
      "grad_norm": 0.7918252348899841,
      "learning_rate": 1.1774291594190311e-05,
      "loss": 0.7727,
      "step": 9091
    },
    {
      "epoch": 1.6723657312767144,
      "grad_norm": 0.8020595908164978,
      "learning_rate": 1.1760527344279382e-05,
      "loss": 0.9602,
      "step": 9092
    },
    {
      "epoch": 1.6725515703400855,
      "grad_norm": 0.6822522878646851,
      "learning_rate": 1.1746770641681238e-05,
      "loss": 0.9458,
      "step": 9093
    },
    {
      "epoch": 1.6727374094034566,
      "grad_norm": 0.6773095726966858,
      "learning_rate": 1.173302148757256e-05,
      "loss": 0.8478,
      "step": 9094
    },
    {
      "epoch": 1.6729232484668277,
      "grad_norm": 0.6347144246101379,
      "learning_rate": 1.1719279883129308e-05,
      "loss": 0.8156,
      "step": 9095
    },
    {
      "epoch": 1.6731090875301988,
      "grad_norm": 0.7613920569419861,
      "learning_rate": 1.1705545829526843e-05,
      "loss": 1.0156,
      "step": 9096
    },
    {
      "epoch": 1.67329492659357,
      "grad_norm": 0.6074971556663513,
      "learning_rate": 1.1691819327939879e-05,
      "loss": 0.8857,
      "step": 9097
    },
    {
      "epoch": 1.673480765656941,
      "grad_norm": 0.5004672408103943,
      "learning_rate": 1.1678100379542468e-05,
      "loss": 0.5623,
      "step": 9098
    },
    {
      "epoch": 1.6736666047203124,
      "grad_norm": 0.7880104184150696,
      "learning_rate": 1.1664388985508035e-05,
      "loss": 1.0569,
      "step": 9099
    },
    {
      "epoch": 1.6738524437836833,
      "grad_norm": 0.6186123490333557,
      "learning_rate": 1.1650685147009311e-05,
      "loss": 0.8835,
      "step": 9100
    },
    {
      "epoch": 1.6740382828470546,
      "grad_norm": 0.6890327334403992,
      "learning_rate": 1.1636988865218446e-05,
      "loss": 0.99,
      "step": 9101
    },
    {
      "epoch": 1.6742241219104255,
      "grad_norm": 0.677280604839325,
      "learning_rate": 1.1623300141306892e-05,
      "loss": 1.0252,
      "step": 9102
    },
    {
      "epoch": 1.6744099609737968,
      "grad_norm": 0.6223238110542297,
      "learning_rate": 1.160961897644548e-05,
      "loss": 1.0804,
      "step": 9103
    },
    {
      "epoch": 1.6745958000371677,
      "grad_norm": 0.6181697845458984,
      "learning_rate": 1.1595945371804418e-05,
      "loss": 0.822,
      "step": 9104
    },
    {
      "epoch": 1.674781639100539,
      "grad_norm": 0.7645830512046814,
      "learning_rate": 1.1582279328553191e-05,
      "loss": 0.8882,
      "step": 9105
    },
    {
      "epoch": 1.67496747816391,
      "grad_norm": 0.7457630038261414,
      "learning_rate": 1.1568620847860733e-05,
      "loss": 0.716,
      "step": 9106
    },
    {
      "epoch": 1.6751533172272812,
      "grad_norm": 0.6209438443183899,
      "learning_rate": 1.1554969930895232e-05,
      "loss": 0.8928,
      "step": 9107
    },
    {
      "epoch": 1.6753391562906523,
      "grad_norm": 0.7704802751541138,
      "learning_rate": 1.154132657882433e-05,
      "loss": 0.845,
      "step": 9108
    },
    {
      "epoch": 1.6755249953540234,
      "grad_norm": 0.9186254739761353,
      "learning_rate": 1.1527690792814926e-05,
      "loss": 0.8236,
      "step": 9109
    },
    {
      "epoch": 1.6757108344173945,
      "grad_norm": 0.5914714932441711,
      "learning_rate": 1.1514062574033357e-05,
      "loss": 0.7576,
      "step": 9110
    },
    {
      "epoch": 1.6758966734807657,
      "grad_norm": 0.772611141204834,
      "learning_rate": 1.1500441923645266e-05,
      "loss": 1.0004,
      "step": 9111
    },
    {
      "epoch": 1.6760825125441368,
      "grad_norm": 0.6939578652381897,
      "learning_rate": 1.1486828842815633e-05,
      "loss": 0.9088,
      "step": 9112
    },
    {
      "epoch": 1.6762683516075079,
      "grad_norm": 0.989822268486023,
      "learning_rate": 1.1473223332708838e-05,
      "loss": 0.9906,
      "step": 9113
    },
    {
      "epoch": 1.676454190670879,
      "grad_norm": 0.8016864061355591,
      "learning_rate": 1.1459625394488571e-05,
      "loss": 0.9294,
      "step": 9114
    },
    {
      "epoch": 1.67664002973425,
      "grad_norm": 0.8364852666854858,
      "learning_rate": 1.1446035029317936e-05,
      "loss": 1.044,
      "step": 9115
    },
    {
      "epoch": 1.6768258687976214,
      "grad_norm": 0.7144214510917664,
      "learning_rate": 1.1432452238359293e-05,
      "loss": 1.0231,
      "step": 9116
    },
    {
      "epoch": 1.6770117078609923,
      "grad_norm": 0.6551642417907715,
      "learning_rate": 1.1418877022774421e-05,
      "loss": 0.9845,
      "step": 9117
    },
    {
      "epoch": 1.6771975469243636,
      "grad_norm": 0.7367211580276489,
      "learning_rate": 1.1405309383724472e-05,
      "loss": 1.0699,
      "step": 9118
    },
    {
      "epoch": 1.6773833859877345,
      "grad_norm": 0.676918625831604,
      "learning_rate": 1.1391749322369871e-05,
      "loss": 0.9828,
      "step": 9119
    },
    {
      "epoch": 1.6775692250511058,
      "grad_norm": 0.7982406616210938,
      "learning_rate": 1.137819683987048e-05,
      "loss": 1.1131,
      "step": 9120
    },
    {
      "epoch": 1.6777550641144767,
      "grad_norm": 0.8516080379486084,
      "learning_rate": 1.1364651937385429e-05,
      "loss": 1.1943,
      "step": 9121
    },
    {
      "epoch": 1.677940903177848,
      "grad_norm": 0.7958751320838928,
      "learning_rate": 1.1351114616073277e-05,
      "loss": 1.1384,
      "step": 9122
    },
    {
      "epoch": 1.678126742241219,
      "grad_norm": 0.8014817237854004,
      "learning_rate": 1.1337584877091877e-05,
      "loss": 0.7247,
      "step": 9123
    },
    {
      "epoch": 1.6783125813045903,
      "grad_norm": 0.6827468872070312,
      "learning_rate": 1.132406272159846e-05,
      "loss": 0.7683,
      "step": 9124
    },
    {
      "epoch": 1.6784984203679614,
      "grad_norm": 0.6997562646865845,
      "learning_rate": 1.1310548150749634e-05,
      "loss": 0.6016,
      "step": 9125
    },
    {
      "epoch": 1.6786842594313325,
      "grad_norm": 0.7690675854682922,
      "learning_rate": 1.1297041165701294e-05,
      "loss": 0.9419,
      "step": 9126
    },
    {
      "epoch": 1.6788700984947036,
      "grad_norm": 0.6297846436500549,
      "learning_rate": 1.128354176760873e-05,
      "loss": 0.767,
      "step": 9127
    },
    {
      "epoch": 1.6790559375580747,
      "grad_norm": 0.7651461362838745,
      "learning_rate": 1.1270049957626571e-05,
      "loss": 0.8214,
      "step": 9128
    },
    {
      "epoch": 1.6792417766214458,
      "grad_norm": 0.6811181902885437,
      "learning_rate": 1.1256565736908841e-05,
      "loss": 0.6638,
      "step": 9129
    },
    {
      "epoch": 1.679427615684817,
      "grad_norm": 0.8382944464683533,
      "learning_rate": 1.1243089106608828e-05,
      "loss": 0.8391,
      "step": 9130
    },
    {
      "epoch": 1.679613454748188,
      "grad_norm": 0.6987102031707764,
      "learning_rate": 1.1229620067879221e-05,
      "loss": 0.9216,
      "step": 9131
    },
    {
      "epoch": 1.6797992938115591,
      "grad_norm": 0.7947177886962891,
      "learning_rate": 1.1216158621872098e-05,
      "loss": 1.0208,
      "step": 9132
    },
    {
      "epoch": 1.6799851328749305,
      "grad_norm": 0.6815338134765625,
      "learning_rate": 1.1202704769738792e-05,
      "loss": 0.7046,
      "step": 9133
    },
    {
      "epoch": 1.6801709719383013,
      "grad_norm": 0.6513886451721191,
      "learning_rate": 1.1189258512630074e-05,
      "loss": 1.1137,
      "step": 9134
    },
    {
      "epoch": 1.6803568110016727,
      "grad_norm": 0.5718039274215698,
      "learning_rate": 1.1175819851696012e-05,
      "loss": 0.7564,
      "step": 9135
    },
    {
      "epoch": 1.6805426500650436,
      "grad_norm": 0.7113762497901917,
      "learning_rate": 1.1162388788086065e-05,
      "loss": 1.0238,
      "step": 9136
    },
    {
      "epoch": 1.6807284891284149,
      "grad_norm": 0.6775386333465576,
      "learning_rate": 1.1148965322948989e-05,
      "loss": 0.9812,
      "step": 9137
    },
    {
      "epoch": 1.6809143281917858,
      "grad_norm": 0.8017896413803101,
      "learning_rate": 1.1135549457432948e-05,
      "loss": 0.8781,
      "step": 9138
    },
    {
      "epoch": 1.681100167255157,
      "grad_norm": 0.6371380686759949,
      "learning_rate": 1.1122141192685442e-05,
      "loss": 0.9729,
      "step": 9139
    },
    {
      "epoch": 1.6812860063185282,
      "grad_norm": 0.7196660041809082,
      "learning_rate": 1.1108740529853268e-05,
      "loss": 0.8344,
      "step": 9140
    },
    {
      "epoch": 1.6814718453818993,
      "grad_norm": 0.6226027607917786,
      "learning_rate": 1.1095347470082629e-05,
      "loss": 0.8421,
      "step": 9141
    },
    {
      "epoch": 1.6816576844452704,
      "grad_norm": 0.6938750743865967,
      "learning_rate": 1.1081962014519076e-05,
      "loss": 0.7527,
      "step": 9142
    },
    {
      "epoch": 1.6818435235086415,
      "grad_norm": 0.9057941436767578,
      "learning_rate": 1.1068584164307504e-05,
      "loss": 1.1757,
      "step": 9143
    },
    {
      "epoch": 1.6820293625720126,
      "grad_norm": 0.6205626726150513,
      "learning_rate": 1.1055213920592122e-05,
      "loss": 1.0075,
      "step": 9144
    },
    {
      "epoch": 1.6822152016353837,
      "grad_norm": 0.742459237575531,
      "learning_rate": 1.1041851284516525e-05,
      "loss": 0.8579,
      "step": 9145
    },
    {
      "epoch": 1.6824010406987548,
      "grad_norm": 0.6259186863899231,
      "learning_rate": 1.1028496257223676e-05,
      "loss": 0.8867,
      "step": 9146
    },
    {
      "epoch": 1.682586879762126,
      "grad_norm": 0.6648344993591309,
      "learning_rate": 1.1015148839855805e-05,
      "loss": 0.8293,
      "step": 9147
    },
    {
      "epoch": 1.6827727188254973,
      "grad_norm": 0.609562337398529,
      "learning_rate": 1.10018090335546e-05,
      "loss": 0.9139,
      "step": 9148
    },
    {
      "epoch": 1.6829585578888682,
      "grad_norm": 0.6610622406005859,
      "learning_rate": 1.0988476839461004e-05,
      "loss": 0.9723,
      "step": 9149
    },
    {
      "epoch": 1.6831443969522395,
      "grad_norm": 0.825636088848114,
      "learning_rate": 1.0975152258715382e-05,
      "loss": 0.9169,
      "step": 9150
    },
    {
      "epoch": 1.6833302360156104,
      "grad_norm": 0.6621133685112,
      "learning_rate": 1.0961835292457368e-05,
      "loss": 0.9953,
      "step": 9151
    },
    {
      "epoch": 1.6835160750789817,
      "grad_norm": 0.7959614396095276,
      "learning_rate": 1.0948525941826026e-05,
      "loss": 0.9995,
      "step": 9152
    },
    {
      "epoch": 1.6837019141423526,
      "grad_norm": 0.7150267362594604,
      "learning_rate": 1.0935224207959738e-05,
      "loss": 1.0144,
      "step": 9153
    },
    {
      "epoch": 1.683887753205724,
      "grad_norm": 0.7913087606430054,
      "learning_rate": 1.0921930091996203e-05,
      "loss": 1.1451,
      "step": 9154
    },
    {
      "epoch": 1.6840735922690948,
      "grad_norm": 0.6784157156944275,
      "learning_rate": 1.090864359507251e-05,
      "loss": 1.0383,
      "step": 9155
    },
    {
      "epoch": 1.6842594313324661,
      "grad_norm": 0.9016114473342896,
      "learning_rate": 1.0895364718325096e-05,
      "loss": 1.107,
      "step": 9156
    },
    {
      "epoch": 1.6844452703958372,
      "grad_norm": 0.6144252419471741,
      "learning_rate": 1.0882093462889709e-05,
      "loss": 0.9491,
      "step": 9157
    },
    {
      "epoch": 1.6846311094592084,
      "grad_norm": 0.7540306448936462,
      "learning_rate": 1.0868829829901473e-05,
      "loss": 1.031,
      "step": 9158
    },
    {
      "epoch": 1.6848169485225795,
      "grad_norm": 0.665734589099884,
      "learning_rate": 1.0855573820494857e-05,
      "loss": 1.0527,
      "step": 9159
    },
    {
      "epoch": 1.6850027875859506,
      "grad_norm": 0.8586148619651794,
      "learning_rate": 1.084232543580369e-05,
      "loss": 1.0122,
      "step": 9160
    },
    {
      "epoch": 1.6851886266493217,
      "grad_norm": 0.6013672947883606,
      "learning_rate": 1.082908467696112e-05,
      "loss": 0.8935,
      "step": 9161
    },
    {
      "epoch": 1.6853744657126928,
      "grad_norm": 0.7050352692604065,
      "learning_rate": 1.0815851545099676e-05,
      "loss": 0.8401,
      "step": 9162
    },
    {
      "epoch": 1.685560304776064,
      "grad_norm": 0.7062950730323792,
      "learning_rate": 1.0802626041351183e-05,
      "loss": 1.0707,
      "step": 9163
    },
    {
      "epoch": 1.685746143839435,
      "grad_norm": 0.8225484490394592,
      "learning_rate": 1.0789408166846882e-05,
      "loss": 0.9885,
      "step": 9164
    },
    {
      "epoch": 1.6859319829028063,
      "grad_norm": 0.8079144358634949,
      "learning_rate": 1.077619792271729e-05,
      "loss": 1.1399,
      "step": 9165
    },
    {
      "epoch": 1.6861178219661772,
      "grad_norm": 0.6949532628059387,
      "learning_rate": 1.0762995310092328e-05,
      "loss": 0.8332,
      "step": 9166
    },
    {
      "epoch": 1.6863036610295485,
      "grad_norm": 0.7433616518974304,
      "learning_rate": 1.0749800330101257e-05,
      "loss": 1.0674,
      "step": 9167
    },
    {
      "epoch": 1.6864895000929194,
      "grad_norm": 0.7382586598396301,
      "learning_rate": 1.073661298387265e-05,
      "loss": 0.7192,
      "step": 9168
    },
    {
      "epoch": 1.6866753391562908,
      "grad_norm": 0.8430142402648926,
      "learning_rate": 1.0723433272534467e-05,
      "loss": 0.9512,
      "step": 9169
    },
    {
      "epoch": 1.6868611782196616,
      "grad_norm": 0.6655109524726868,
      "learning_rate": 1.0710261197213978e-05,
      "loss": 0.8721,
      "step": 9170
    },
    {
      "epoch": 1.687047017283033,
      "grad_norm": 0.6514587998390198,
      "learning_rate": 1.0697096759037829e-05,
      "loss": 0.8485,
      "step": 9171
    },
    {
      "epoch": 1.6872328563464039,
      "grad_norm": 0.734404444694519,
      "learning_rate": 1.0683939959131994e-05,
      "loss": 1.2359,
      "step": 9172
    },
    {
      "epoch": 1.6874186954097752,
      "grad_norm": 0.5610777735710144,
      "learning_rate": 1.0670790798621822e-05,
      "loss": 0.5728,
      "step": 9173
    },
    {
      "epoch": 1.6876045344731463,
      "grad_norm": 0.7563809752464294,
      "learning_rate": 1.0657649278631998e-05,
      "loss": 1.2102,
      "step": 9174
    },
    {
      "epoch": 1.6877903735365174,
      "grad_norm": 0.8163540363311768,
      "learning_rate": 1.0644515400286504e-05,
      "loss": 1.0889,
      "step": 9175
    },
    {
      "epoch": 1.6879762125998885,
      "grad_norm": 0.9675668478012085,
      "learning_rate": 1.063138916470875e-05,
      "loss": 0.7862,
      "step": 9176
    },
    {
      "epoch": 1.6881620516632596,
      "grad_norm": 0.6208799481391907,
      "learning_rate": 1.0618270573021405e-05,
      "loss": 0.892,
      "step": 9177
    },
    {
      "epoch": 1.6883478907266307,
      "grad_norm": 0.6580058932304382,
      "learning_rate": 1.0605159626346584e-05,
      "loss": 0.75,
      "step": 9178
    },
    {
      "epoch": 1.6885337297900018,
      "grad_norm": 0.7946237921714783,
      "learning_rate": 1.0592056325805644e-05,
      "loss": 0.9343,
      "step": 9179
    },
    {
      "epoch": 1.688719568853373,
      "grad_norm": 0.6942155957221985,
      "learning_rate": 1.0578960672519366e-05,
      "loss": 0.8665,
      "step": 9180
    },
    {
      "epoch": 1.688905407916744,
      "grad_norm": 0.693755567073822,
      "learning_rate": 1.0565872667607867e-05,
      "loss": 0.7981,
      "step": 9181
    },
    {
      "epoch": 1.6890912469801154,
      "grad_norm": 0.9935961961746216,
      "learning_rate": 1.0552792312190552e-05,
      "loss": 1.0529,
      "step": 9182
    },
    {
      "epoch": 1.6892770860434863,
      "grad_norm": 0.7871524095535278,
      "learning_rate": 1.0539719607386244e-05,
      "loss": 1.0497,
      "step": 9183
    },
    {
      "epoch": 1.6894629251068576,
      "grad_norm": 0.7142606377601624,
      "learning_rate": 1.0526654554313053e-05,
      "loss": 0.9732,
      "step": 9184
    },
    {
      "epoch": 1.6896487641702285,
      "grad_norm": 0.784253716468811,
      "learning_rate": 1.0513597154088472e-05,
      "loss": 1.0093,
      "step": 9185
    },
    {
      "epoch": 1.6898346032335998,
      "grad_norm": 0.6851341128349304,
      "learning_rate": 1.0500547407829341e-05,
      "loss": 0.9728,
      "step": 9186
    },
    {
      "epoch": 1.6900204422969707,
      "grad_norm": 0.961861789226532,
      "learning_rate": 1.0487505316651813e-05,
      "loss": 0.9666,
      "step": 9187
    },
    {
      "epoch": 1.690206281360342,
      "grad_norm": 0.630499541759491,
      "learning_rate": 1.0474470881671439e-05,
      "loss": 0.9504,
      "step": 9188
    },
    {
      "epoch": 1.6903921204237131,
      "grad_norm": 0.7499755024909973,
      "learning_rate": 1.0461444104003049e-05,
      "loss": 0.9772,
      "step": 9189
    },
    {
      "epoch": 1.6905779594870842,
      "grad_norm": 0.6846833229064941,
      "learning_rate": 1.0448424984760874e-05,
      "loss": 0.9679,
      "step": 9190
    },
    {
      "epoch": 1.6907637985504553,
      "grad_norm": 0.6661667823791504,
      "learning_rate": 1.0435413525058435e-05,
      "loss": 0.9704,
      "step": 9191
    },
    {
      "epoch": 1.6909496376138264,
      "grad_norm": 0.6650241613388062,
      "learning_rate": 1.0422409726008675e-05,
      "loss": 1.0795,
      "step": 9192
    },
    {
      "epoch": 1.6911354766771975,
      "grad_norm": 0.7499749660491943,
      "learning_rate": 1.0409413588723804e-05,
      "loss": 0.7986,
      "step": 9193
    },
    {
      "epoch": 1.6913213157405687,
      "grad_norm": 0.8803886771202087,
      "learning_rate": 1.0396425114315412e-05,
      "loss": 0.9701,
      "step": 9194
    },
    {
      "epoch": 1.6915071548039398,
      "grad_norm": 0.692290186882019,
      "learning_rate": 1.0383444303894452e-05,
      "loss": 0.8647,
      "step": 9195
    },
    {
      "epoch": 1.6916929938673109,
      "grad_norm": 0.7082966566085815,
      "learning_rate": 1.0370471158571172e-05,
      "loss": 0.916,
      "step": 9196
    },
    {
      "epoch": 1.691878832930682,
      "grad_norm": 0.7729305028915405,
      "learning_rate": 1.0357505679455226e-05,
      "loss": 1.0942,
      "step": 9197
    },
    {
      "epoch": 1.692064671994053,
      "grad_norm": 0.7437100410461426,
      "learning_rate": 1.0344547867655553e-05,
      "loss": 0.821,
      "step": 9198
    },
    {
      "epoch": 1.6922505110574244,
      "grad_norm": 0.6189757585525513,
      "learning_rate": 1.0331597724280461e-05,
      "loss": 0.8888,
      "step": 9199
    },
    {
      "epoch": 1.6924363501207953,
      "grad_norm": 0.6114462614059448,
      "learning_rate": 1.0318655250437614e-05,
      "loss": 0.8383,
      "step": 9200
    },
    {
      "epoch": 1.6926221891841666,
      "grad_norm": 0.6665554642677307,
      "learning_rate": 1.0305720447234035e-05,
      "loss": 0.9874,
      "step": 9201
    },
    {
      "epoch": 1.6928080282475375,
      "grad_norm": 0.7060142159461975,
      "learning_rate": 1.0292793315776027e-05,
      "loss": 0.9165,
      "step": 9202
    },
    {
      "epoch": 1.6929938673109088,
      "grad_norm": 0.7609959244728088,
      "learning_rate": 1.0279873857169276e-05,
      "loss": 0.9382,
      "step": 9203
    },
    {
      "epoch": 1.6931797063742797,
      "grad_norm": 0.7125695943832397,
      "learning_rate": 1.0266962072518849e-05,
      "loss": 1.0135,
      "step": 9204
    },
    {
      "epoch": 1.693365545437651,
      "grad_norm": 0.7978722453117371,
      "learning_rate": 1.0254057962929087e-05,
      "loss": 0.9271,
      "step": 9205
    },
    {
      "epoch": 1.6935513845010222,
      "grad_norm": 1.0889793634414673,
      "learning_rate": 1.0241161529503717e-05,
      "loss": 1.3928,
      "step": 9206
    },
    {
      "epoch": 1.6937372235643933,
      "grad_norm": 0.6475967764854431,
      "learning_rate": 1.0228272773345782e-05,
      "loss": 0.971,
      "step": 9207
    },
    {
      "epoch": 1.6939230626277644,
      "grad_norm": 0.7573670148849487,
      "learning_rate": 1.0215391695557707e-05,
      "loss": 1.1833,
      "step": 9208
    },
    {
      "epoch": 1.6941089016911355,
      "grad_norm": 0.7922182083129883,
      "learning_rate": 1.0202518297241237e-05,
      "loss": 1.004,
      "step": 9209
    },
    {
      "epoch": 1.6942947407545066,
      "grad_norm": 0.7527633309364319,
      "learning_rate": 1.0189652579497444e-05,
      "loss": 0.9222,
      "step": 9210
    },
    {
      "epoch": 1.6944805798178777,
      "grad_norm": 0.6256292462348938,
      "learning_rate": 1.0176794543426782e-05,
      "loss": 0.9044,
      "step": 9211
    },
    {
      "epoch": 1.6946664188812488,
      "grad_norm": 0.8159219622612,
      "learning_rate": 1.0163944190129005e-05,
      "loss": 0.9947,
      "step": 9212
    },
    {
      "epoch": 1.69485225794462,
      "grad_norm": 0.6840503811836243,
      "learning_rate": 1.0151101520703265e-05,
      "loss": 1.0449,
      "step": 9213
    },
    {
      "epoch": 1.6950380970079912,
      "grad_norm": 0.6659444570541382,
      "learning_rate": 1.0138266536247976e-05,
      "loss": 1.0373,
      "step": 9214
    },
    {
      "epoch": 1.6952239360713621,
      "grad_norm": 0.7139298915863037,
      "learning_rate": 1.0125439237860968e-05,
      "loss": 0.9433,
      "step": 9215
    },
    {
      "epoch": 1.6954097751347335,
      "grad_norm": 0.6589419841766357,
      "learning_rate": 1.0112619626639386e-05,
      "loss": 0.8481,
      "step": 9216
    },
    {
      "epoch": 1.6955956141981043,
      "grad_norm": 0.6323763132095337,
      "learning_rate": 1.009980770367972e-05,
      "loss": 0.8495,
      "step": 9217
    },
    {
      "epoch": 1.6957814532614757,
      "grad_norm": 0.920706033706665,
      "learning_rate": 1.0087003470077815e-05,
      "loss": 1.1459,
      "step": 9218
    },
    {
      "epoch": 1.6959672923248466,
      "grad_norm": 0.7430810332298279,
      "learning_rate": 1.0074206926928819e-05,
      "loss": 0.9022,
      "step": 9219
    },
    {
      "epoch": 1.6961531313882179,
      "grad_norm": 0.724772572517395,
      "learning_rate": 1.0061418075327278e-05,
      "loss": 0.6118,
      "step": 9220
    },
    {
      "epoch": 1.6963389704515888,
      "grad_norm": 0.7099050879478455,
      "learning_rate": 1.0048636916367004e-05,
      "loss": 0.8845,
      "step": 9221
    },
    {
      "epoch": 1.69652480951496,
      "grad_norm": 0.7855052947998047,
      "learning_rate": 1.0035863451141236e-05,
      "loss": 0.9102,
      "step": 9222
    },
    {
      "epoch": 1.6967106485783312,
      "grad_norm": 0.7682072520256042,
      "learning_rate": 1.0023097680742522e-05,
      "loss": 0.8534,
      "step": 9223
    },
    {
      "epoch": 1.6968964876417023,
      "grad_norm": 0.7189264297485352,
      "learning_rate": 1.0010339606262709e-05,
      "loss": 1.0848,
      "step": 9224
    },
    {
      "epoch": 1.6970823267050734,
      "grad_norm": 1.1096469163894653,
      "learning_rate": 9.997589228793058e-06,
      "loss": 1.1886,
      "step": 9225
    },
    {
      "epoch": 1.6972681657684445,
      "grad_norm": 0.7639279961585999,
      "learning_rate": 9.98484654942411e-06,
      "loss": 0.7868,
      "step": 9226
    },
    {
      "epoch": 1.6974540048318156,
      "grad_norm": 0.5577638745307922,
      "learning_rate": 9.972111569245791e-06,
      "loss": 0.7305,
      "step": 9227
    },
    {
      "epoch": 1.6976398438951867,
      "grad_norm": 0.6836424469947815,
      "learning_rate": 9.95938428934734e-06,
      "loss": 0.773,
      "step": 9228
    },
    {
      "epoch": 1.6978256829585578,
      "grad_norm": 0.7009959816932678,
      "learning_rate": 9.946664710817345e-06,
      "loss": 0.9731,
      "step": 9229
    },
    {
      "epoch": 1.698011522021929,
      "grad_norm": 0.7553769946098328,
      "learning_rate": 9.933952834743753e-06,
      "loss": 1.0267,
      "step": 9230
    },
    {
      "epoch": 1.6981973610853003,
      "grad_norm": 0.8114926815032959,
      "learning_rate": 9.921248662213833e-06,
      "loss": 1.0416,
      "step": 9231
    },
    {
      "epoch": 1.6983832001486712,
      "grad_norm": 0.7146091461181641,
      "learning_rate": 9.908552194314214e-06,
      "loss": 1.08,
      "step": 9232
    },
    {
      "epoch": 1.6985690392120425,
      "grad_norm": 0.7939289212226868,
      "learning_rate": 9.895863432130826e-06,
      "loss": 0.9978,
      "step": 9233
    },
    {
      "epoch": 1.6987548782754134,
      "grad_norm": 0.7090169787406921,
      "learning_rate": 9.883182376749001e-06,
      "loss": 0.9042,
      "step": 9234
    },
    {
      "epoch": 1.6989407173387847,
      "grad_norm": 0.736428439617157,
      "learning_rate": 9.870509029253327e-06,
      "loss": 1.0081,
      "step": 9235
    },
    {
      "epoch": 1.6991265564021556,
      "grad_norm": 0.6867282390594482,
      "learning_rate": 9.857843390727816e-06,
      "loss": 0.9255,
      "step": 9236
    },
    {
      "epoch": 1.699312395465527,
      "grad_norm": 0.6552194356918335,
      "learning_rate": 9.8451854622558e-06,
      "loss": 0.9218,
      "step": 9237
    },
    {
      "epoch": 1.6994982345288978,
      "grad_norm": 1.0081390142440796,
      "learning_rate": 9.832535244919906e-06,
      "loss": 0.9419,
      "step": 9238
    },
    {
      "epoch": 1.6996840735922691,
      "grad_norm": 0.8012736439704895,
      "learning_rate": 9.819892739802173e-06,
      "loss": 1.052,
      "step": 9239
    },
    {
      "epoch": 1.6998699126556402,
      "grad_norm": 0.6890170574188232,
      "learning_rate": 9.807257947983895e-06,
      "loss": 0.9941,
      "step": 9240
    },
    {
      "epoch": 1.7000557517190114,
      "grad_norm": 0.7510611414909363,
      "learning_rate": 9.794630870545796e-06,
      "loss": 1.0552,
      "step": 9241
    },
    {
      "epoch": 1.7002415907823825,
      "grad_norm": 0.833249032497406,
      "learning_rate": 9.782011508567857e-06,
      "loss": 1.1007,
      "step": 9242
    },
    {
      "epoch": 1.7004274298457536,
      "grad_norm": 1.0734540224075317,
      "learning_rate": 9.769399863129459e-06,
      "loss": 1.3107,
      "step": 9243
    },
    {
      "epoch": 1.7006132689091247,
      "grad_norm": 0.6381735801696777,
      "learning_rate": 9.756795935309316e-06,
      "loss": 1.0896,
      "step": 9244
    },
    {
      "epoch": 1.7007991079724958,
      "grad_norm": 0.6753941178321838,
      "learning_rate": 9.744199726185443e-06,
      "loss": 1.0672,
      "step": 9245
    },
    {
      "epoch": 1.700984947035867,
      "grad_norm": 0.825974702835083,
      "learning_rate": 9.731611236835259e-06,
      "loss": 0.9139,
      "step": 9246
    },
    {
      "epoch": 1.701170786099238,
      "grad_norm": 0.700402021408081,
      "learning_rate": 9.719030468335443e-06,
      "loss": 0.7945,
      "step": 9247
    },
    {
      "epoch": 1.7013566251626093,
      "grad_norm": 0.7434213757514954,
      "learning_rate": 9.706457421762106e-06,
      "loss": 1.1943,
      "step": 9248
    },
    {
      "epoch": 1.7015424642259802,
      "grad_norm": 0.5882927775382996,
      "learning_rate": 9.693892098190582e-06,
      "loss": 0.8897,
      "step": 9249
    },
    {
      "epoch": 1.7017283032893515,
      "grad_norm": 0.627251148223877,
      "learning_rate": 9.681334498695648e-06,
      "loss": 0.9317,
      "step": 9250
    },
    {
      "epoch": 1.7019141423527224,
      "grad_norm": 0.74644935131073,
      "learning_rate": 9.66878462435139e-06,
      "loss": 1.091,
      "step": 9251
    },
    {
      "epoch": 1.7020999814160938,
      "grad_norm": 0.7172256708145142,
      "learning_rate": 9.656242476231204e-06,
      "loss": 0.7462,
      "step": 9252
    },
    {
      "epoch": 1.7022858204794646,
      "grad_norm": 0.761581540107727,
      "learning_rate": 9.643708055407873e-06,
      "loss": 0.8465,
      "step": 9253
    },
    {
      "epoch": 1.702471659542836,
      "grad_norm": 0.7488076090812683,
      "learning_rate": 9.631181362953467e-06,
      "loss": 0.7814,
      "step": 9254
    },
    {
      "epoch": 1.702657498606207,
      "grad_norm": 0.803229033946991,
      "learning_rate": 9.618662399939448e-06,
      "loss": 0.9648,
      "step": 9255
    },
    {
      "epoch": 1.7028433376695782,
      "grad_norm": 0.7162587642669678,
      "learning_rate": 9.60615116743655e-06,
      "loss": 0.8486,
      "step": 9256
    },
    {
      "epoch": 1.7030291767329493,
      "grad_norm": 0.8364415764808655,
      "learning_rate": 9.593647666514904e-06,
      "loss": 0.9365,
      "step": 9257
    },
    {
      "epoch": 1.7032150157963204,
      "grad_norm": 0.6324708461761475,
      "learning_rate": 9.581151898244e-06,
      "loss": 0.9032,
      "step": 9258
    },
    {
      "epoch": 1.7034008548596915,
      "grad_norm": 0.6390442252159119,
      "learning_rate": 9.568663863692562e-06,
      "loss": 0.877,
      "step": 9259
    },
    {
      "epoch": 1.7035866939230626,
      "grad_norm": 0.6136648058891296,
      "learning_rate": 9.556183563928766e-06,
      "loss": 0.941,
      "step": 9260
    },
    {
      "epoch": 1.7037725329864337,
      "grad_norm": 0.6558448672294617,
      "learning_rate": 9.54371100002005e-06,
      "loss": 0.9333,
      "step": 9261
    },
    {
      "epoch": 1.7039583720498048,
      "grad_norm": 0.8663216829299927,
      "learning_rate": 9.53124617303326e-06,
      "loss": 1.1077,
      "step": 9262
    },
    {
      "epoch": 1.7041442111131762,
      "grad_norm": 0.9772462844848633,
      "learning_rate": 9.518789084034495e-06,
      "loss": 1.0044,
      "step": 9263
    },
    {
      "epoch": 1.704330050176547,
      "grad_norm": 0.7501927614212036,
      "learning_rate": 9.506339734089253e-06,
      "loss": 0.8734,
      "step": 9264
    },
    {
      "epoch": 1.7045158892399184,
      "grad_norm": 0.8446945548057556,
      "learning_rate": 9.49389812426238e-06,
      "loss": 1.0656,
      "step": 9265
    },
    {
      "epoch": 1.7047017283032893,
      "grad_norm": 0.7121186852455139,
      "learning_rate": 9.48146425561799e-06,
      "loss": 1.0447,
      "step": 9266
    },
    {
      "epoch": 1.7048875673666606,
      "grad_norm": 0.7923415303230286,
      "learning_rate": 9.46903812921962e-06,
      "loss": 0.7783,
      "step": 9267
    },
    {
      "epoch": 1.7050734064300315,
      "grad_norm": 0.7992509603500366,
      "learning_rate": 9.456619746130058e-06,
      "loss": 0.854,
      "step": 9268
    },
    {
      "epoch": 1.7052592454934028,
      "grad_norm": 0.8346595764160156,
      "learning_rate": 9.444209107411518e-06,
      "loss": 1.1072,
      "step": 9269
    },
    {
      "epoch": 1.7054450845567737,
      "grad_norm": 0.6362949013710022,
      "learning_rate": 9.431806214125472e-06,
      "loss": 0.688,
      "step": 9270
    },
    {
      "epoch": 1.705630923620145,
      "grad_norm": 0.9164164662361145,
      "learning_rate": 9.419411067332773e-06,
      "loss": 1.0372,
      "step": 9271
    },
    {
      "epoch": 1.7058167626835161,
      "grad_norm": 0.6323965191841125,
      "learning_rate": 9.40702366809365e-06,
      "loss": 0.7713,
      "step": 9272
    },
    {
      "epoch": 1.7060026017468872,
      "grad_norm": 0.7026657462120056,
      "learning_rate": 9.39464401746757e-06,
      "loss": 1.0552,
      "step": 9273
    },
    {
      "epoch": 1.7061884408102583,
      "grad_norm": 0.8523253202438354,
      "learning_rate": 9.382272116513402e-06,
      "loss": 0.9925,
      "step": 9274
    },
    {
      "epoch": 1.7063742798736294,
      "grad_norm": 0.7092315554618835,
      "learning_rate": 9.36990796628936e-06,
      "loss": 0.9838,
      "step": 9275
    },
    {
      "epoch": 1.7065601189370005,
      "grad_norm": 0.7713611721992493,
      "learning_rate": 9.357551567852974e-06,
      "loss": 0.6708,
      "step": 9276
    },
    {
      "epoch": 1.7067459580003717,
      "grad_norm": 0.6664693355560303,
      "learning_rate": 9.345202922261098e-06,
      "loss": 1.0176,
      "step": 9277
    },
    {
      "epoch": 1.7069317970637428,
      "grad_norm": 0.7604504823684692,
      "learning_rate": 9.332862030569944e-06,
      "loss": 1.0009,
      "step": 9278
    },
    {
      "epoch": 1.7071176361271139,
      "grad_norm": 0.7931390404701233,
      "learning_rate": 9.320528893835078e-06,
      "loss": 0.9902,
      "step": 9279
    },
    {
      "epoch": 1.7073034751904852,
      "grad_norm": 0.7187214493751526,
      "learning_rate": 9.308203513111336e-06,
      "loss": 0.9574,
      "step": 9280
    },
    {
      "epoch": 1.707489314253856,
      "grad_norm": 0.7779387831687927,
      "learning_rate": 9.295885889452982e-06,
      "loss": 1.0485,
      "step": 9281
    },
    {
      "epoch": 1.7076751533172274,
      "grad_norm": 0.7711448073387146,
      "learning_rate": 9.283576023913532e-06,
      "loss": 1.0164,
      "step": 9282
    },
    {
      "epoch": 1.7078609923805983,
      "grad_norm": 0.6007513403892517,
      "learning_rate": 9.271273917545897e-06,
      "loss": 0.8938,
      "step": 9283
    },
    {
      "epoch": 1.7080468314439696,
      "grad_norm": 0.8609693646430969,
      "learning_rate": 9.258979571402282e-06,
      "loss": 1.0425,
      "step": 9284
    },
    {
      "epoch": 1.7082326705073405,
      "grad_norm": 0.6705703139305115,
      "learning_rate": 9.246692986534255e-06,
      "loss": 1.1867,
      "step": 9285
    },
    {
      "epoch": 1.7084185095707118,
      "grad_norm": 0.8482201099395752,
      "learning_rate": 9.234414163992744e-06,
      "loss": 1.0239,
      "step": 9286
    },
    {
      "epoch": 1.7086043486340827,
      "grad_norm": 0.7028303146362305,
      "learning_rate": 9.22214310482794e-06,
      "loss": 1.054,
      "step": 9287
    },
    {
      "epoch": 1.708790187697454,
      "grad_norm": 0.6620766520500183,
      "learning_rate": 9.20987981008944e-06,
      "loss": 0.7547,
      "step": 9288
    },
    {
      "epoch": 1.7089760267608252,
      "grad_norm": 0.685680627822876,
      "learning_rate": 9.197624280826134e-06,
      "loss": 0.9087,
      "step": 9289
    },
    {
      "epoch": 1.7091618658241963,
      "grad_norm": 0.8493560552597046,
      "learning_rate": 9.185376518086297e-06,
      "loss": 1.0277,
      "step": 9290
    },
    {
      "epoch": 1.7093477048875674,
      "grad_norm": 0.7470026016235352,
      "learning_rate": 9.173136522917457e-06,
      "loss": 0.867,
      "step": 9291
    },
    {
      "epoch": 1.7095335439509385,
      "grad_norm": 0.6782554388046265,
      "learning_rate": 9.160904296366557e-06,
      "loss": 0.9899,
      "step": 9292
    },
    {
      "epoch": 1.7097193830143096,
      "grad_norm": 0.7193986177444458,
      "learning_rate": 9.148679839479856e-06,
      "loss": 0.7549,
      "step": 9293
    },
    {
      "epoch": 1.7099052220776807,
      "grad_norm": 1.4221397638320923,
      "learning_rate": 9.13646315330291e-06,
      "loss": 0.9677,
      "step": 9294
    },
    {
      "epoch": 1.7100910611410518,
      "grad_norm": 0.7279139161109924,
      "learning_rate": 9.12425423888067e-06,
      "loss": 0.7372,
      "step": 9295
    },
    {
      "epoch": 1.710276900204423,
      "grad_norm": 2.2297890186309814,
      "learning_rate": 9.112053097257357e-06,
      "loss": 1.1687,
      "step": 9296
    },
    {
      "epoch": 1.7104627392677942,
      "grad_norm": 0.7337605953216553,
      "learning_rate": 9.0998597294766e-06,
      "loss": 0.9823,
      "step": 9297
    },
    {
      "epoch": 1.7106485783311651,
      "grad_norm": 0.7120336890220642,
      "learning_rate": 9.087674136581271e-06,
      "loss": 0.7,
      "step": 9298
    },
    {
      "epoch": 1.7108344173945365,
      "grad_norm": 0.6502007842063904,
      "learning_rate": 9.075496319613663e-06,
      "loss": 0.9573,
      "step": 9299
    },
    {
      "epoch": 1.7110202564579073,
      "grad_norm": 0.7360041737556458,
      "learning_rate": 9.063326279615391e-06,
      "loss": 0.9621,
      "step": 9300
    },
    {
      "epoch": 1.7112060955212787,
      "grad_norm": 0.6855154037475586,
      "learning_rate": 9.051164017627346e-06,
      "loss": 0.8322,
      "step": 9301
    },
    {
      "epoch": 1.7113919345846496,
      "grad_norm": 0.8629727959632874,
      "learning_rate": 9.039009534689813e-06,
      "loss": 0.9323,
      "step": 9302
    },
    {
      "epoch": 1.7115777736480209,
      "grad_norm": 0.7231772541999817,
      "learning_rate": 9.026862831842398e-06,
      "loss": 0.7956,
      "step": 9303
    },
    {
      "epoch": 1.7117636127113918,
      "grad_norm": 0.6572983860969543,
      "learning_rate": 9.014723910124001e-06,
      "loss": 0.8395,
      "step": 9304
    },
    {
      "epoch": 1.711949451774763,
      "grad_norm": 0.8665408492088318,
      "learning_rate": 9.002592770572915e-06,
      "loss": 1.0307,
      "step": 9305
    },
    {
      "epoch": 1.7121352908381342,
      "grad_norm": 0.8959839940071106,
      "learning_rate": 8.990469414226743e-06,
      "loss": 0.8463,
      "step": 9306
    },
    {
      "epoch": 1.7123211299015053,
      "grad_norm": 0.6686840057373047,
      "learning_rate": 8.978353842122445e-06,
      "loss": 0.8437,
      "step": 9307
    },
    {
      "epoch": 1.7125069689648764,
      "grad_norm": 0.655849277973175,
      "learning_rate": 8.966246055296235e-06,
      "loss": 1.0538,
      "step": 9308
    },
    {
      "epoch": 1.7126928080282475,
      "grad_norm": 0.6408183574676514,
      "learning_rate": 8.954146054783774e-06,
      "loss": 0.8483,
      "step": 9309
    },
    {
      "epoch": 1.7128786470916186,
      "grad_norm": 0.7304595112800598,
      "learning_rate": 8.942053841619968e-06,
      "loss": 0.8701,
      "step": 9310
    },
    {
      "epoch": 1.7130644861549897,
      "grad_norm": 0.7849457263946533,
      "learning_rate": 8.9299694168391e-06,
      "loss": 0.9448,
      "step": 9311
    },
    {
      "epoch": 1.7132503252183608,
      "grad_norm": 0.7002692818641663,
      "learning_rate": 8.917892781474768e-06,
      "loss": 1.1745,
      "step": 9312
    },
    {
      "epoch": 1.713436164281732,
      "grad_norm": 0.6553815007209778,
      "learning_rate": 8.90582393655992e-06,
      "loss": 1.0872,
      "step": 9313
    },
    {
      "epoch": 1.7136220033451033,
      "grad_norm": 0.885659396648407,
      "learning_rate": 8.893762883126855e-06,
      "loss": 0.8797,
      "step": 9314
    },
    {
      "epoch": 1.7138078424084742,
      "grad_norm": 0.6541244983673096,
      "learning_rate": 8.881709622207125e-06,
      "loss": 0.9569,
      "step": 9315
    },
    {
      "epoch": 1.7139936814718455,
      "grad_norm": 0.7587239742279053,
      "learning_rate": 8.869664154831737e-06,
      "loss": 0.9818,
      "step": 9316
    },
    {
      "epoch": 1.7141795205352164,
      "grad_norm": 0.8520368337631226,
      "learning_rate": 8.857626482030901e-06,
      "loss": 0.8596,
      "step": 9317
    },
    {
      "epoch": 1.7143653595985877,
      "grad_norm": 0.6669599413871765,
      "learning_rate": 8.84559660483426e-06,
      "loss": 0.9869,
      "step": 9318
    },
    {
      "epoch": 1.7145511986619586,
      "grad_norm": 0.6341535449028015,
      "learning_rate": 8.83357452427076e-06,
      "loss": 1.0055,
      "step": 9319
    },
    {
      "epoch": 1.71473703772533,
      "grad_norm": 0.8398535251617432,
      "learning_rate": 8.821560241368655e-06,
      "loss": 0.9017,
      "step": 9320
    },
    {
      "epoch": 1.714922876788701,
      "grad_norm": 1.0017404556274414,
      "learning_rate": 8.809553757155597e-06,
      "loss": 0.8948,
      "step": 9321
    },
    {
      "epoch": 1.7151087158520721,
      "grad_norm": 0.7218863368034363,
      "learning_rate": 8.79755507265847e-06,
      "loss": 0.8497,
      "step": 9322
    },
    {
      "epoch": 1.7152945549154432,
      "grad_norm": 0.7040542364120483,
      "learning_rate": 8.785564188903583e-06,
      "loss": 1.0778,
      "step": 9323
    },
    {
      "epoch": 1.7154803939788144,
      "grad_norm": 0.609380841255188,
      "learning_rate": 8.773581106916517e-06,
      "loss": 0.7377,
      "step": 9324
    },
    {
      "epoch": 1.7156662330421855,
      "grad_norm": 0.6335259675979614,
      "learning_rate": 8.76160582772223e-06,
      "loss": 0.8535,
      "step": 9325
    },
    {
      "epoch": 1.7158520721055566,
      "grad_norm": 0.6981773376464844,
      "learning_rate": 8.749638352345002e-06,
      "loss": 0.8354,
      "step": 9326
    },
    {
      "epoch": 1.7160379111689277,
      "grad_norm": 0.7336592078208923,
      "learning_rate": 8.737678681808414e-06,
      "loss": 0.9337,
      "step": 9327
    },
    {
      "epoch": 1.7162237502322988,
      "grad_norm": 0.7219322919845581,
      "learning_rate": 8.725726817135427e-06,
      "loss": 0.9018,
      "step": 9328
    },
    {
      "epoch": 1.7164095892956701,
      "grad_norm": 0.7079262137413025,
      "learning_rate": 8.713782759348266e-06,
      "loss": 0.8605,
      "step": 9329
    },
    {
      "epoch": 1.716595428359041,
      "grad_norm": 0.6500135660171509,
      "learning_rate": 8.701846509468592e-06,
      "loss": 0.798,
      "step": 9330
    },
    {
      "epoch": 1.7167812674224123,
      "grad_norm": 0.8792296051979065,
      "learning_rate": 8.689918068517278e-06,
      "loss": 1.222,
      "step": 9331
    },
    {
      "epoch": 1.7169671064857832,
      "grad_norm": 0.720497727394104,
      "learning_rate": 8.677997437514629e-06,
      "loss": 0.8851,
      "step": 9332
    },
    {
      "epoch": 1.7171529455491545,
      "grad_norm": 0.6603461503982544,
      "learning_rate": 8.66608461748023e-06,
      "loss": 0.8619,
      "step": 9333
    },
    {
      "epoch": 1.7173387846125254,
      "grad_norm": 0.731478214263916,
      "learning_rate": 8.654179609433011e-06,
      "loss": 0.9271,
      "step": 9334
    },
    {
      "epoch": 1.7175246236758968,
      "grad_norm": 1.2854375839233398,
      "learning_rate": 8.642282414391257e-06,
      "loss": 1.5186,
      "step": 9335
    },
    {
      "epoch": 1.7177104627392676,
      "grad_norm": 0.7765483260154724,
      "learning_rate": 8.630393033372507e-06,
      "loss": 0.7269,
      "step": 9336
    },
    {
      "epoch": 1.717896301802639,
      "grad_norm": 0.6998369097709656,
      "learning_rate": 8.618511467393741e-06,
      "loss": 1.0292,
      "step": 9337
    },
    {
      "epoch": 1.71808214086601,
      "grad_norm": 0.7568623423576355,
      "learning_rate": 8.606637717471178e-06,
      "loss": 1.1241,
      "step": 9338
    },
    {
      "epoch": 1.7182679799293812,
      "grad_norm": 0.8563328981399536,
      "learning_rate": 8.594771784620414e-06,
      "loss": 0.9679,
      "step": 9339
    },
    {
      "epoch": 1.7184538189927523,
      "grad_norm": 0.6528421640396118,
      "learning_rate": 8.582913669856385e-06,
      "loss": 0.8435,
      "step": 9340
    },
    {
      "epoch": 1.7186396580561234,
      "grad_norm": 0.7107359170913696,
      "learning_rate": 8.571063374193322e-06,
      "loss": 0.9204,
      "step": 9341
    },
    {
      "epoch": 1.7188254971194945,
      "grad_norm": 0.725544273853302,
      "learning_rate": 8.559220898644827e-06,
      "loss": 0.9239,
      "step": 9342
    },
    {
      "epoch": 1.7190113361828656,
      "grad_norm": 0.7376357913017273,
      "learning_rate": 8.547386244223777e-06,
      "loss": 1.0024,
      "step": 9343
    },
    {
      "epoch": 1.7191971752462367,
      "grad_norm": 0.8160426616668701,
      "learning_rate": 8.53555941194245e-06,
      "loss": 0.9461,
      "step": 9344
    },
    {
      "epoch": 1.7193830143096078,
      "grad_norm": 0.6708709597587585,
      "learning_rate": 8.523740402812397e-06,
      "loss": 0.7973,
      "step": 9345
    },
    {
      "epoch": 1.7195688533729792,
      "grad_norm": 0.8642181158065796,
      "learning_rate": 8.511929217844528e-06,
      "loss": 0.9434,
      "step": 9346
    },
    {
      "epoch": 1.71975469243635,
      "grad_norm": 0.6233113408088684,
      "learning_rate": 8.50012585804908e-06,
      "loss": 0.8648,
      "step": 9347
    },
    {
      "epoch": 1.7199405314997214,
      "grad_norm": 0.6707333922386169,
      "learning_rate": 8.488330324435655e-06,
      "loss": 0.9194,
      "step": 9348
    },
    {
      "epoch": 1.7201263705630923,
      "grad_norm": 0.6645792722702026,
      "learning_rate": 8.476542618013095e-06,
      "loss": 0.8238,
      "step": 9349
    },
    {
      "epoch": 1.7203122096264636,
      "grad_norm": 0.6607288718223572,
      "learning_rate": 8.464762739789655e-06,
      "loss": 0.7656,
      "step": 9350
    },
    {
      "epoch": 1.7204980486898345,
      "grad_norm": 0.7014643549919128,
      "learning_rate": 8.45299069077291e-06,
      "loss": 1.0564,
      "step": 9351
    },
    {
      "epoch": 1.7206838877532058,
      "grad_norm": 0.7516298890113831,
      "learning_rate": 8.44122647196971e-06,
      "loss": 1.0525,
      "step": 9352
    },
    {
      "epoch": 1.7208697268165767,
      "grad_norm": 0.8982193470001221,
      "learning_rate": 8.429470084386293e-06,
      "loss": 1.1706,
      "step": 9353
    },
    {
      "epoch": 1.721055565879948,
      "grad_norm": 0.7839781045913696,
      "learning_rate": 8.417721529028234e-06,
      "loss": 1.0219,
      "step": 9354
    },
    {
      "epoch": 1.7212414049433191,
      "grad_norm": 0.8507671356201172,
      "learning_rate": 8.405980806900371e-06,
      "loss": 0.9497,
      "step": 9355
    },
    {
      "epoch": 1.7214272440066902,
      "grad_norm": 1.1558749675750732,
      "learning_rate": 8.394247919006936e-06,
      "loss": 1.34,
      "step": 9356
    },
    {
      "epoch": 1.7216130830700613,
      "grad_norm": 0.9968247413635254,
      "learning_rate": 8.38252286635145e-06,
      "loss": 1.3205,
      "step": 9357
    },
    {
      "epoch": 1.7217989221334324,
      "grad_norm": 0.938660740852356,
      "learning_rate": 8.370805649936818e-06,
      "loss": 1.102,
      "step": 9358
    },
    {
      "epoch": 1.7219847611968035,
      "grad_norm": 0.6969743967056274,
      "learning_rate": 8.359096270765198e-06,
      "loss": 0.8952,
      "step": 9359
    },
    {
      "epoch": 1.7221706002601747,
      "grad_norm": 0.7880121469497681,
      "learning_rate": 8.347394729838132e-06,
      "loss": 1.1787,
      "step": 9360
    },
    {
      "epoch": 1.7223564393235458,
      "grad_norm": 0.7151280045509338,
      "learning_rate": 8.335701028156495e-06,
      "loss": 0.9412,
      "step": 9361
    },
    {
      "epoch": 1.7225422783869169,
      "grad_norm": 0.6870322823524475,
      "learning_rate": 8.324015166720456e-06,
      "loss": 0.9034,
      "step": 9362
    },
    {
      "epoch": 1.7227281174502882,
      "grad_norm": 0.7795487642288208,
      "learning_rate": 8.312337146529525e-06,
      "loss": 1.3137,
      "step": 9363
    },
    {
      "epoch": 1.722913956513659,
      "grad_norm": 0.7532526254653931,
      "learning_rate": 8.300666968582572e-06,
      "loss": 0.8896,
      "step": 9364
    },
    {
      "epoch": 1.7230997955770304,
      "grad_norm": 0.77845299243927,
      "learning_rate": 8.289004633877773e-06,
      "loss": 1.2301,
      "step": 9365
    },
    {
      "epoch": 1.7232856346404013,
      "grad_norm": 0.7793643474578857,
      "learning_rate": 8.27735014341261e-06,
      "loss": 1.0974,
      "step": 9366
    },
    {
      "epoch": 1.7234714737037726,
      "grad_norm": 0.6873371005058289,
      "learning_rate": 8.265703498183918e-06,
      "loss": 0.8727,
      "step": 9367
    },
    {
      "epoch": 1.7236573127671435,
      "grad_norm": 0.7508918046951294,
      "learning_rate": 8.25406469918789e-06,
      "loss": 1.0924,
      "step": 9368
    },
    {
      "epoch": 1.7238431518305148,
      "grad_norm": 0.8036147356033325,
      "learning_rate": 8.242433747419987e-06,
      "loss": 1.0435,
      "step": 9369
    },
    {
      "epoch": 1.724028990893886,
      "grad_norm": 0.7148988842964172,
      "learning_rate": 8.230810643875053e-06,
      "loss": 1.0826,
      "step": 9370
    },
    {
      "epoch": 1.724214829957257,
      "grad_norm": 0.8428753018379211,
      "learning_rate": 8.219195389547207e-06,
      "loss": 0.9271,
      "step": 9371
    },
    {
      "epoch": 1.7244006690206282,
      "grad_norm": 0.65247642993927,
      "learning_rate": 8.207587985429954e-06,
      "loss": 0.8675,
      "step": 9372
    },
    {
      "epoch": 1.7245865080839993,
      "grad_norm": 0.9677809476852417,
      "learning_rate": 8.195988432516078e-06,
      "loss": 0.9372,
      "step": 9373
    },
    {
      "epoch": 1.7247723471473704,
      "grad_norm": 0.7008064389228821,
      "learning_rate": 8.184396731797728e-06,
      "loss": 0.7741,
      "step": 9374
    },
    {
      "epoch": 1.7249581862107415,
      "grad_norm": 0.6639923453330994,
      "learning_rate": 8.172812884266378e-06,
      "loss": 0.813,
      "step": 9375
    },
    {
      "epoch": 1.7251440252741126,
      "grad_norm": 0.7582129240036011,
      "learning_rate": 8.161236890912794e-06,
      "loss": 0.8838,
      "step": 9376
    },
    {
      "epoch": 1.7253298643374837,
      "grad_norm": 0.6991216540336609,
      "learning_rate": 8.149668752727113e-06,
      "loss": 0.9785,
      "step": 9377
    },
    {
      "epoch": 1.7255157034008548,
      "grad_norm": 0.6826270222663879,
      "learning_rate": 8.138108470698768e-06,
      "loss": 0.9951,
      "step": 9378
    },
    {
      "epoch": 1.725701542464226,
      "grad_norm": 0.6590842604637146,
      "learning_rate": 8.126556045816569e-06,
      "loss": 0.8817,
      "step": 9379
    },
    {
      "epoch": 1.7258873815275972,
      "grad_norm": 0.7262789011001587,
      "learning_rate": 8.115011479068568e-06,
      "loss": 1.0486,
      "step": 9380
    },
    {
      "epoch": 1.7260732205909681,
      "grad_norm": 0.6248814463615417,
      "learning_rate": 8.103474771442243e-06,
      "loss": 0.9154,
      "step": 9381
    },
    {
      "epoch": 1.7262590596543395,
      "grad_norm": 0.7736847400665283,
      "learning_rate": 8.091945923924338e-06,
      "loss": 0.9123,
      "step": 9382
    },
    {
      "epoch": 1.7264448987177103,
      "grad_norm": 0.6092911958694458,
      "learning_rate": 8.08042493750093e-06,
      "loss": 1.0612,
      "step": 9383
    },
    {
      "epoch": 1.7266307377810817,
      "grad_norm": 0.6939818263053894,
      "learning_rate": 8.068911813157454e-06,
      "loss": 1.0582,
      "step": 9384
    },
    {
      "epoch": 1.7268165768444526,
      "grad_norm": 0.7486193776130676,
      "learning_rate": 8.057406551878621e-06,
      "loss": 1.0236,
      "step": 9385
    },
    {
      "epoch": 1.7270024159078239,
      "grad_norm": 0.6597424149513245,
      "learning_rate": 8.045909154648534e-06,
      "loss": 0.9871,
      "step": 9386
    },
    {
      "epoch": 1.727188254971195,
      "grad_norm": 2.9619107246398926,
      "learning_rate": 8.034419622450561e-06,
      "loss": 1.189,
      "step": 9387
    },
    {
      "epoch": 1.727374094034566,
      "grad_norm": 0.838079571723938,
      "learning_rate": 8.02293795626744e-06,
      "loss": 0.8031,
      "step": 9388
    },
    {
      "epoch": 1.7275599330979372,
      "grad_norm": 0.6807136535644531,
      "learning_rate": 8.01146415708125e-06,
      "loss": 0.9412,
      "step": 9389
    },
    {
      "epoch": 1.7277457721613083,
      "grad_norm": 0.641676664352417,
      "learning_rate": 7.999998225873317e-06,
      "loss": 0.7303,
      "step": 9390
    },
    {
      "epoch": 1.7279316112246794,
      "grad_norm": 0.7298771142959595,
      "learning_rate": 7.988540163624381e-06,
      "loss": 0.8572,
      "step": 9391
    },
    {
      "epoch": 1.7281174502880505,
      "grad_norm": 0.8123074173927307,
      "learning_rate": 7.977089971314466e-06,
      "loss": 0.9141,
      "step": 9392
    },
    {
      "epoch": 1.7283032893514216,
      "grad_norm": 0.6712712049484253,
      "learning_rate": 7.965647649922947e-06,
      "loss": 0.8206,
      "step": 9393
    },
    {
      "epoch": 1.7284891284147927,
      "grad_norm": 0.7178130745887756,
      "learning_rate": 7.954213200428485e-06,
      "loss": 0.9454,
      "step": 9394
    },
    {
      "epoch": 1.728674967478164,
      "grad_norm": 0.6644273996353149,
      "learning_rate": 7.942786623809106e-06,
      "loss": 1.0341,
      "step": 9395
    },
    {
      "epoch": 1.728860806541535,
      "grad_norm": 0.7427142262458801,
      "learning_rate": 7.931367921042166e-06,
      "loss": 0.9111,
      "step": 9396
    },
    {
      "epoch": 1.7290466456049063,
      "grad_norm": 0.7232275009155273,
      "learning_rate": 7.91995709310429e-06,
      "loss": 0.8069,
      "step": 9397
    },
    {
      "epoch": 1.7292324846682772,
      "grad_norm": 0.6869901418685913,
      "learning_rate": 7.908554140971514e-06,
      "loss": 0.8375,
      "step": 9398
    },
    {
      "epoch": 1.7294183237316485,
      "grad_norm": 0.8033495545387268,
      "learning_rate": 7.89715906561912e-06,
      "loss": 0.9688,
      "step": 9399
    },
    {
      "epoch": 1.7296041627950194,
      "grad_norm": 0.8616623878479004,
      "learning_rate": 7.885771868021796e-06,
      "loss": 1.1344,
      "step": 9400
    },
    {
      "epoch": 1.7297900018583907,
      "grad_norm": 0.6528728008270264,
      "learning_rate": 7.874392549153464e-06,
      "loss": 0.9157,
      "step": 9401
    },
    {
      "epoch": 1.7299758409217616,
      "grad_norm": 0.7244764566421509,
      "learning_rate": 7.863021109987457e-06,
      "loss": 0.571,
      "step": 9402
    },
    {
      "epoch": 1.730161679985133,
      "grad_norm": 0.8218297362327576,
      "learning_rate": 7.851657551496395e-06,
      "loss": 1.0371,
      "step": 9403
    },
    {
      "epoch": 1.730347519048504,
      "grad_norm": 0.6427557468414307,
      "learning_rate": 7.840301874652211e-06,
      "loss": 1.0065,
      "step": 9404
    },
    {
      "epoch": 1.7305333581118751,
      "grad_norm": 0.6553307175636292,
      "learning_rate": 7.828954080426198e-06,
      "loss": 0.8694,
      "step": 9405
    },
    {
      "epoch": 1.7307191971752462,
      "grad_norm": 1.0670928955078125,
      "learning_rate": 7.817614169788945e-06,
      "loss": 1.3094,
      "step": 9406
    },
    {
      "epoch": 1.7309050362386174,
      "grad_norm": 0.6958805918693542,
      "learning_rate": 7.806282143710375e-06,
      "loss": 0.9212,
      "step": 9407
    },
    {
      "epoch": 1.7310908753019885,
      "grad_norm": 0.6197575926780701,
      "learning_rate": 7.794958003159757e-06,
      "loss": 0.7294,
      "step": 9408
    },
    {
      "epoch": 1.7312767143653596,
      "grad_norm": 0.698625922203064,
      "learning_rate": 7.783641749105664e-06,
      "loss": 1.0314,
      "step": 9409
    },
    {
      "epoch": 1.7314625534287307,
      "grad_norm": 0.6479657888412476,
      "learning_rate": 7.772333382516017e-06,
      "loss": 0.971,
      "step": 9410
    },
    {
      "epoch": 1.7316483924921018,
      "grad_norm": 0.7665457129478455,
      "learning_rate": 7.761032904358e-06,
      "loss": 1.1693,
      "step": 9411
    },
    {
      "epoch": 1.7318342315554731,
      "grad_norm": 0.7469654083251953,
      "learning_rate": 7.749740315598209e-06,
      "loss": 0.9795,
      "step": 9412
    },
    {
      "epoch": 1.732020070618844,
      "grad_norm": 0.6912213563919067,
      "learning_rate": 7.738455617202501e-06,
      "loss": 0.9352,
      "step": 9413
    },
    {
      "epoch": 1.7322059096822153,
      "grad_norm": 0.6912152171134949,
      "learning_rate": 7.727178810136093e-06,
      "loss": 1.0428,
      "step": 9414
    },
    {
      "epoch": 1.7323917487455862,
      "grad_norm": 0.9005116820335388,
      "learning_rate": 7.715909895363505e-06,
      "loss": 1.0998,
      "step": 9415
    },
    {
      "epoch": 1.7325775878089575,
      "grad_norm": 0.6274826526641846,
      "learning_rate": 7.704648873848586e-06,
      "loss": 1.1338,
      "step": 9416
    },
    {
      "epoch": 1.7327634268723284,
      "grad_norm": 0.8490490913391113,
      "learning_rate": 7.693395746554544e-06,
      "loss": 0.9429,
      "step": 9417
    },
    {
      "epoch": 1.7329492659356998,
      "grad_norm": 1.9930272102355957,
      "learning_rate": 7.682150514443853e-06,
      "loss": 1.04,
      "step": 9418
    },
    {
      "epoch": 1.7331351049990706,
      "grad_norm": 0.6957984566688538,
      "learning_rate": 7.670913178478367e-06,
      "loss": 0.9502,
      "step": 9419
    },
    {
      "epoch": 1.733320944062442,
      "grad_norm": 0.7939738631248474,
      "learning_rate": 7.659683739619206e-06,
      "loss": 1.1698,
      "step": 9420
    },
    {
      "epoch": 1.733506783125813,
      "grad_norm": 0.6218949556350708,
      "learning_rate": 7.648462198826879e-06,
      "loss": 0.952,
      "step": 9421
    },
    {
      "epoch": 1.7336926221891842,
      "grad_norm": 0.6665553450584412,
      "learning_rate": 7.63724855706116e-06,
      "loss": 0.916,
      "step": 9422
    },
    {
      "epoch": 1.7336926221891842,
      "eval_loss": 1.0072797536849976,
      "eval_runtime": 23.2133,
      "eval_samples_per_second": 47.042,
      "eval_steps_per_second": 23.521,
      "step": 9422
    },
    {
      "epoch": 1.7338784612525553,
      "grad_norm": 0.7004141211509705,
      "learning_rate": 7.626042815281209e-06,
      "loss": 0.8868,
      "step": 9423
    },
    {
      "epoch": 1.7340643003159264,
      "grad_norm": 0.7327351570129395,
      "learning_rate": 7.6148449744454786e-06,
      "loss": 0.9734,
      "step": 9424
    },
    {
      "epoch": 1.7342501393792975,
      "grad_norm": 0.6216403841972351,
      "learning_rate": 7.603655035511703e-06,
      "loss": 0.9558,
      "step": 9425
    },
    {
      "epoch": 1.7344359784426686,
      "grad_norm": 0.6237961649894714,
      "learning_rate": 7.592472999437028e-06,
      "loss": 0.8549,
      "step": 9426
    },
    {
      "epoch": 1.7346218175060397,
      "grad_norm": 0.7453010082244873,
      "learning_rate": 7.581298867177844e-06,
      "loss": 0.9747,
      "step": 9427
    },
    {
      "epoch": 1.7348076565694108,
      "grad_norm": 0.6449342966079712,
      "learning_rate": 7.5701326396899194e-06,
      "loss": 0.8981,
      "step": 9428
    },
    {
      "epoch": 1.7349934956327822,
      "grad_norm": 0.7956316471099854,
      "learning_rate": 7.558974317928313e-06,
      "loss": 0.8466,
      "step": 9429
    },
    {
      "epoch": 1.735179334696153,
      "grad_norm": 0.7923126220703125,
      "learning_rate": 7.547823902847406e-06,
      "loss": 1.1053,
      "step": 9430
    },
    {
      "epoch": 1.7353651737595244,
      "grad_norm": 0.7033422589302063,
      "learning_rate": 7.536681395400968e-06,
      "loss": 0.8778,
      "step": 9431
    },
    {
      "epoch": 1.7355510128228953,
      "grad_norm": 0.6055691242218018,
      "learning_rate": 7.525546796541994e-06,
      "loss": 0.6723,
      "step": 9432
    },
    {
      "epoch": 1.7357368518862666,
      "grad_norm": 0.6363259553909302,
      "learning_rate": 7.5144201072228745e-06,
      "loss": 0.859,
      "step": 9433
    },
    {
      "epoch": 1.7359226909496375,
      "grad_norm": 0.6942827105522156,
      "learning_rate": 7.503301328395273e-06,
      "loss": 1.1544,
      "step": 9434
    },
    {
      "epoch": 1.7361085300130088,
      "grad_norm": 0.6959713101387024,
      "learning_rate": 7.492190461010229e-06,
      "loss": 0.8484,
      "step": 9435
    },
    {
      "epoch": 1.73629436907638,
      "grad_norm": 0.7058919072151184,
      "learning_rate": 7.481087506018058e-06,
      "loss": 0.9904,
      "step": 9436
    },
    {
      "epoch": 1.736480208139751,
      "grad_norm": 0.7573116421699524,
      "learning_rate": 7.469992464368447e-06,
      "loss": 0.8349,
      "step": 9437
    },
    {
      "epoch": 1.7366660472031221,
      "grad_norm": 0.6595671772956848,
      "learning_rate": 7.458905337010369e-06,
      "loss": 0.8104,
      "step": 9438
    },
    {
      "epoch": 1.7368518862664932,
      "grad_norm": 0.7107803821563721,
      "learning_rate": 7.447826124892121e-06,
      "loss": 0.8094,
      "step": 9439
    },
    {
      "epoch": 1.7370377253298643,
      "grad_norm": 0.7347820997238159,
      "learning_rate": 7.436754828961346e-06,
      "loss": 0.8601,
      "step": 9440
    },
    {
      "epoch": 1.7372235643932354,
      "grad_norm": 0.6467185020446777,
      "learning_rate": 7.425691450164962e-06,
      "loss": 0.9238,
      "step": 9441
    },
    {
      "epoch": 1.7374094034566065,
      "grad_norm": 4.698081016540527,
      "learning_rate": 7.414635989449293e-06,
      "loss": 1.2099,
      "step": 9442
    },
    {
      "epoch": 1.7375952425199777,
      "grad_norm": 0.6623215079307556,
      "learning_rate": 7.40358844775989e-06,
      "loss": 0.8568,
      "step": 9443
    },
    {
      "epoch": 1.737781081583349,
      "grad_norm": 0.6400848031044006,
      "learning_rate": 7.392548826041701e-06,
      "loss": 1.0292,
      "step": 9444
    },
    {
      "epoch": 1.7379669206467199,
      "grad_norm": 0.7742132544517517,
      "learning_rate": 7.381517125238968e-06,
      "loss": 1.0895,
      "step": 9445
    },
    {
      "epoch": 1.7381527597100912,
      "grad_norm": 0.6758570671081543,
      "learning_rate": 7.370493346295249e-06,
      "loss": 0.8089,
      "step": 9446
    },
    {
      "epoch": 1.738338598773462,
      "grad_norm": 0.7959402203559875,
      "learning_rate": 7.359477490153432e-06,
      "loss": 1.1993,
      "step": 9447
    },
    {
      "epoch": 1.7385244378368334,
      "grad_norm": 0.7615758776664734,
      "learning_rate": 7.3484695577557216e-06,
      "loss": 1.0477,
      "step": 9448
    },
    {
      "epoch": 1.7387102769002043,
      "grad_norm": 0.8708173036575317,
      "learning_rate": 7.337469550043652e-06,
      "loss": 1.5399,
      "step": 9449
    },
    {
      "epoch": 1.7388961159635756,
      "grad_norm": 0.6313830614089966,
      "learning_rate": 7.326477467958104e-06,
      "loss": 0.8139,
      "step": 9450
    },
    {
      "epoch": 1.7390819550269465,
      "grad_norm": 0.6188602447509766,
      "learning_rate": 7.3154933124392145e-06,
      "loss": 0.7236,
      "step": 9451
    },
    {
      "epoch": 1.7392677940903178,
      "grad_norm": 0.6811180114746094,
      "learning_rate": 7.304517084426488e-06,
      "loss": 1.0196,
      "step": 9452
    },
    {
      "epoch": 1.739453633153689,
      "grad_norm": 0.7616562247276306,
      "learning_rate": 7.2935487848587725e-06,
      "loss": 1.1128,
      "step": 9453
    },
    {
      "epoch": 1.73963947221706,
      "grad_norm": 0.7264105677604675,
      "learning_rate": 7.282588414674196e-06,
      "loss": 1.0773,
      "step": 9454
    },
    {
      "epoch": 1.7398253112804312,
      "grad_norm": 0.7808012366294861,
      "learning_rate": 7.27163597481022e-06,
      "loss": 1.0436,
      "step": 9455
    },
    {
      "epoch": 1.7400111503438023,
      "grad_norm": 0.7362692356109619,
      "learning_rate": 7.260691466203639e-06,
      "loss": 1.033,
      "step": 9456
    },
    {
      "epoch": 1.7401969894071734,
      "grad_norm": 0.8960447907447815,
      "learning_rate": 7.249754889790539e-06,
      "loss": 0.893,
      "step": 9457
    },
    {
      "epoch": 1.7403828284705445,
      "grad_norm": 0.7096341252326965,
      "learning_rate": 7.238826246506359e-06,
      "loss": 0.9255,
      "step": 9458
    },
    {
      "epoch": 1.7405686675339156,
      "grad_norm": 0.6668227314949036,
      "learning_rate": 7.227905537285873e-06,
      "loss": 1.0548,
      "step": 9459
    },
    {
      "epoch": 1.7407545065972867,
      "grad_norm": 0.567503035068512,
      "learning_rate": 7.216992763063113e-06,
      "loss": 0.8955,
      "step": 9460
    },
    {
      "epoch": 1.740940345660658,
      "grad_norm": 0.698553204536438,
      "learning_rate": 7.206087924771521e-06,
      "loss": 1.1543,
      "step": 9461
    },
    {
      "epoch": 1.741126184724029,
      "grad_norm": 0.6601186394691467,
      "learning_rate": 7.19519102334375e-06,
      "loss": 1.0774,
      "step": 9462
    },
    {
      "epoch": 1.7413120237874002,
      "grad_norm": 0.7196704149246216,
      "learning_rate": 7.184302059711901e-06,
      "loss": 0.832,
      "step": 9463
    },
    {
      "epoch": 1.7414978628507711,
      "grad_norm": 0.6979762315750122,
      "learning_rate": 7.17342103480727e-06,
      "loss": 0.8598,
      "step": 9464
    },
    {
      "epoch": 1.7416837019141425,
      "grad_norm": 0.7031317353248596,
      "learning_rate": 7.162547949560561e-06,
      "loss": 0.9583,
      "step": 9465
    },
    {
      "epoch": 1.7418695409775133,
      "grad_norm": 0.7397237420082092,
      "learning_rate": 7.151682804901782e-06,
      "loss": 0.9233,
      "step": 9466
    },
    {
      "epoch": 1.7420553800408847,
      "grad_norm": 0.6605537533760071,
      "learning_rate": 7.140825601760237e-06,
      "loss": 0.8789,
      "step": 9467
    },
    {
      "epoch": 1.7422412191042556,
      "grad_norm": 0.7009168863296509,
      "learning_rate": 7.129976341064592e-06,
      "loss": 0.9078,
      "step": 9468
    },
    {
      "epoch": 1.7424270581676269,
      "grad_norm": 0.7297293543815613,
      "learning_rate": 7.119135023742773e-06,
      "loss": 0.8503,
      "step": 9469
    },
    {
      "epoch": 1.742612897230998,
      "grad_norm": 0.6891294717788696,
      "learning_rate": 7.108301650722104e-06,
      "loss": 0.9163,
      "step": 9470
    },
    {
      "epoch": 1.742798736294369,
      "grad_norm": 0.749011754989624,
      "learning_rate": 7.097476222929134e-06,
      "loss": 0.9002,
      "step": 9471
    },
    {
      "epoch": 1.7429845753577402,
      "grad_norm": 0.7376906871795654,
      "learning_rate": 7.0866587412898085e-06,
      "loss": 0.8309,
      "step": 9472
    },
    {
      "epoch": 1.7431704144211113,
      "grad_norm": 0.6205510497093201,
      "learning_rate": 7.07584920672939e-06,
      "loss": 0.901,
      "step": 9473
    },
    {
      "epoch": 1.7433562534844824,
      "grad_norm": 0.8700895309448242,
      "learning_rate": 7.065047620172416e-06,
      "loss": 0.7299,
      "step": 9474
    },
    {
      "epoch": 1.7435420925478535,
      "grad_norm": 1.9028913974761963,
      "learning_rate": 7.054253982542791e-06,
      "loss": 1.5785,
      "step": 9475
    },
    {
      "epoch": 1.7437279316112246,
      "grad_norm": 0.7316854000091553,
      "learning_rate": 7.043468294763689e-06,
      "loss": 0.8147,
      "step": 9476
    },
    {
      "epoch": 1.7439137706745957,
      "grad_norm": 0.561488687992096,
      "learning_rate": 7.03269055775766e-06,
      "loss": 0.8449,
      "step": 9477
    },
    {
      "epoch": 1.744099609737967,
      "grad_norm": 0.73995041847229,
      "learning_rate": 7.021920772446522e-06,
      "loss": 0.9434,
      "step": 9478
    },
    {
      "epoch": 1.744285448801338,
      "grad_norm": 0.6605014801025391,
      "learning_rate": 7.011158939751461e-06,
      "loss": 0.8919,
      "step": 9479
    },
    {
      "epoch": 1.7444712878647093,
      "grad_norm": 0.6619118452072144,
      "learning_rate": 7.000405060592941e-06,
      "loss": 0.9865,
      "step": 9480
    },
    {
      "epoch": 1.7446571269280802,
      "grad_norm": 0.7531654834747314,
      "learning_rate": 6.98965913589078e-06,
      "loss": 0.8969,
      "step": 9481
    },
    {
      "epoch": 1.7448429659914515,
      "grad_norm": 0.6228139400482178,
      "learning_rate": 6.978921166564112e-06,
      "loss": 0.7863,
      "step": 9482
    },
    {
      "epoch": 1.7450288050548224,
      "grad_norm": 0.7277988791465759,
      "learning_rate": 6.968191153531334e-06,
      "loss": 1.0369,
      "step": 9483
    },
    {
      "epoch": 1.7452146441181937,
      "grad_norm": 0.6712504029273987,
      "learning_rate": 6.9574690977102675e-06,
      "loss": 1.0008,
      "step": 9484
    },
    {
      "epoch": 1.7454004831815646,
      "grad_norm": 0.7697523832321167,
      "learning_rate": 6.946755000017935e-06,
      "loss": 0.9245,
      "step": 9485
    },
    {
      "epoch": 1.745586322244936,
      "grad_norm": 0.5483777523040771,
      "learning_rate": 6.9360488613707585e-06,
      "loss": 0.5716,
      "step": 9486
    },
    {
      "epoch": 1.745772161308307,
      "grad_norm": 0.725399911403656,
      "learning_rate": 6.925350682684484e-06,
      "loss": 0.9928,
      "step": 9487
    },
    {
      "epoch": 1.7459580003716781,
      "grad_norm": 0.7769641876220703,
      "learning_rate": 6.914660464874112e-06,
      "loss": 0.9754,
      "step": 9488
    },
    {
      "epoch": 1.7461438394350492,
      "grad_norm": 0.5875201225280762,
      "learning_rate": 6.903978208854023e-06,
      "loss": 0.6502,
      "step": 9489
    },
    {
      "epoch": 1.7463296784984204,
      "grad_norm": 0.6539204120635986,
      "learning_rate": 6.893303915537874e-06,
      "loss": 0.9873,
      "step": 9490
    },
    {
      "epoch": 1.7465155175617915,
      "grad_norm": 0.6799665689468384,
      "learning_rate": 6.882637585838681e-06,
      "loss": 0.9155,
      "step": 9491
    },
    {
      "epoch": 1.7467013566251626,
      "grad_norm": 0.7157257199287415,
      "learning_rate": 6.871979220668734e-06,
      "loss": 1.1283,
      "step": 9492
    },
    {
      "epoch": 1.7468871956885337,
      "grad_norm": 0.7215269804000854,
      "learning_rate": 6.861328820939683e-06,
      "loss": 0.8819,
      "step": 9493
    },
    {
      "epoch": 1.7470730347519048,
      "grad_norm": 0.7681776285171509,
      "learning_rate": 6.8506863875624775e-06,
      "loss": 1.0661,
      "step": 9494
    },
    {
      "epoch": 1.7472588738152761,
      "grad_norm": 0.7559865713119507,
      "learning_rate": 6.84005192144741e-06,
      "loss": 0.821,
      "step": 9495
    },
    {
      "epoch": 1.747444712878647,
      "grad_norm": 0.8904591798782349,
      "learning_rate": 6.829425423504021e-06,
      "loss": 0.9272,
      "step": 9496
    },
    {
      "epoch": 1.7476305519420183,
      "grad_norm": 0.6811634302139282,
      "learning_rate": 6.81880689464125e-06,
      "loss": 0.6446,
      "step": 9497
    },
    {
      "epoch": 1.7478163910053892,
      "grad_norm": 0.6657458543777466,
      "learning_rate": 6.808196335767325e-06,
      "loss": 0.9042,
      "step": 9498
    },
    {
      "epoch": 1.7480022300687605,
      "grad_norm": 0.6674712300300598,
      "learning_rate": 6.797593747789765e-06,
      "loss": 0.9185,
      "step": 9499
    },
    {
      "epoch": 1.7481880691321314,
      "grad_norm": 0.5958244800567627,
      "learning_rate": 6.786999131615457e-06,
      "loss": 0.8436,
      "step": 9500
    },
    {
      "epoch": 1.7483739081955028,
      "grad_norm": 0.7511781454086304,
      "learning_rate": 6.776412488150585e-06,
      "loss": 1.1611,
      "step": 9501
    },
    {
      "epoch": 1.7485597472588739,
      "grad_norm": 0.6515342593193054,
      "learning_rate": 6.7658338183006155e-06,
      "loss": 1.1498,
      "step": 9502
    },
    {
      "epoch": 1.748745586322245,
      "grad_norm": 0.7776544094085693,
      "learning_rate": 6.755263122970412e-06,
      "loss": 0.8307,
      "step": 9503
    },
    {
      "epoch": 1.748931425385616,
      "grad_norm": 0.5926681160926819,
      "learning_rate": 6.7447004030640525e-06,
      "loss": 0.9145,
      "step": 9504
    },
    {
      "epoch": 1.7491172644489872,
      "grad_norm": 0.707554042339325,
      "learning_rate": 6.734145659485047e-06,
      "loss": 1.0083,
      "step": 9505
    },
    {
      "epoch": 1.7493031035123583,
      "grad_norm": 0.6766769289970398,
      "learning_rate": 6.723598893136119e-06,
      "loss": 1.0921,
      "step": 9506
    },
    {
      "epoch": 1.7494889425757294,
      "grad_norm": 0.6683719158172607,
      "learning_rate": 6.7130601049193664e-06,
      "loss": 0.9509,
      "step": 9507
    },
    {
      "epoch": 1.7496747816391005,
      "grad_norm": 0.6655029058456421,
      "learning_rate": 6.7025292957362265e-06,
      "loss": 0.6811,
      "step": 9508
    },
    {
      "epoch": 1.7498606207024716,
      "grad_norm": 0.6449025869369507,
      "learning_rate": 6.692006466487377e-06,
      "loss": 0.8224,
      "step": 9509
    },
    {
      "epoch": 1.750046459765843,
      "grad_norm": 0.588196873664856,
      "learning_rate": 6.681491618072877e-06,
      "loss": 0.6702,
      "step": 9510
    },
    {
      "epoch": 1.7502322988292138,
      "grad_norm": 0.7115856409072876,
      "learning_rate": 6.670984751392095e-06,
      "loss": 1.03,
      "step": 9511
    },
    {
      "epoch": 1.7504181378925852,
      "grad_norm": 0.6933878064155579,
      "learning_rate": 6.660485867343713e-06,
      "loss": 0.7458,
      "step": 9512
    },
    {
      "epoch": 1.750603976955956,
      "grad_norm": 0.8777282238006592,
      "learning_rate": 6.649994966825679e-06,
      "loss": 0.8825,
      "step": 9513
    },
    {
      "epoch": 1.7507898160193274,
      "grad_norm": 0.951866090297699,
      "learning_rate": 6.6395120507353415e-06,
      "loss": 0.9348,
      "step": 9514
    },
    {
      "epoch": 1.7509756550826983,
      "grad_norm": 0.6491940021514893,
      "learning_rate": 6.6290371199693395e-06,
      "loss": 0.9793,
      "step": 9515
    },
    {
      "epoch": 1.7511614941460696,
      "grad_norm": 0.6472504734992981,
      "learning_rate": 6.618570175423566e-06,
      "loss": 0.8149,
      "step": 9516
    },
    {
      "epoch": 1.7513473332094405,
      "grad_norm": 0.7329473495483398,
      "learning_rate": 6.608111217993329e-06,
      "loss": 0.9426,
      "step": 9517
    },
    {
      "epoch": 1.7515331722728118,
      "grad_norm": 0.6990534067153931,
      "learning_rate": 6.597660248573179e-06,
      "loss": 0.9525,
      "step": 9518
    },
    {
      "epoch": 1.751719011336183,
      "grad_norm": 0.7106050848960876,
      "learning_rate": 6.587217268057022e-06,
      "loss": 1.1132,
      "step": 9519
    },
    {
      "epoch": 1.751904850399554,
      "grad_norm": 0.7301816344261169,
      "learning_rate": 6.576782277338056e-06,
      "loss": 1.0255,
      "step": 9520
    },
    {
      "epoch": 1.7520906894629251,
      "grad_norm": 0.8306465148925781,
      "learning_rate": 6.566355277308811e-06,
      "loss": 0.9773,
      "step": 9521
    },
    {
      "epoch": 1.7522765285262962,
      "grad_norm": 0.6874080300331116,
      "learning_rate": 6.555936268861151e-06,
      "loss": 1.1446,
      "step": 9522
    },
    {
      "epoch": 1.7524623675896673,
      "grad_norm": 0.5944158434867859,
      "learning_rate": 6.545525252886198e-06,
      "loss": 0.8901,
      "step": 9523
    },
    {
      "epoch": 1.7526482066530384,
      "grad_norm": 0.7529304027557373,
      "learning_rate": 6.53512223027446e-06,
      "loss": 1.0603,
      "step": 9524
    },
    {
      "epoch": 1.7528340457164096,
      "grad_norm": 0.6817454695701599,
      "learning_rate": 6.524727201915726e-06,
      "loss": 1.0123,
      "step": 9525
    },
    {
      "epoch": 1.7530198847797807,
      "grad_norm": 0.6408984065055847,
      "learning_rate": 6.514340168699107e-06,
      "loss": 0.8478,
      "step": 9526
    },
    {
      "epoch": 1.753205723843152,
      "grad_norm": 0.8707150816917419,
      "learning_rate": 6.5039611315130144e-06,
      "loss": 1.3053,
      "step": 9527
    },
    {
      "epoch": 1.7533915629065229,
      "grad_norm": 0.6480703353881836,
      "learning_rate": 6.493590091245194e-06,
      "loss": 1.0935,
      "step": 9528
    },
    {
      "epoch": 1.7535774019698942,
      "grad_norm": 0.6530660390853882,
      "learning_rate": 6.483227048782725e-06,
      "loss": 0.8194,
      "step": 9529
    },
    {
      "epoch": 1.753763241033265,
      "grad_norm": 1.7306596040725708,
      "learning_rate": 6.472872005011954e-06,
      "loss": 1.2105,
      "step": 9530
    },
    {
      "epoch": 1.7539490800966364,
      "grad_norm": 0.7162639498710632,
      "learning_rate": 6.462524960818583e-06,
      "loss": 1.0676,
      "step": 9531
    },
    {
      "epoch": 1.7541349191600073,
      "grad_norm": 0.6799525022506714,
      "learning_rate": 6.4521859170876055e-06,
      "loss": 0.9085,
      "step": 9532
    },
    {
      "epoch": 1.7543207582233786,
      "grad_norm": 0.8963913321495056,
      "learning_rate": 6.441854874703357e-06,
      "loss": 1.1685,
      "step": 9533
    },
    {
      "epoch": 1.7545065972867495,
      "grad_norm": 0.7606403231620789,
      "learning_rate": 6.4315318345494536e-06,
      "loss": 1.0158,
      "step": 9534
    },
    {
      "epoch": 1.7546924363501208,
      "grad_norm": 0.7327620983123779,
      "learning_rate": 6.421216797508866e-06,
      "loss": 0.9107,
      "step": 9535
    },
    {
      "epoch": 1.754878275413492,
      "grad_norm": 0.7143133878707886,
      "learning_rate": 6.4109097644638684e-06,
      "loss": 0.9722,
      "step": 9536
    },
    {
      "epoch": 1.755064114476863,
      "grad_norm": 0.8020583391189575,
      "learning_rate": 6.4006107362960195e-06,
      "loss": 1.2301,
      "step": 9537
    },
    {
      "epoch": 1.7552499535402342,
      "grad_norm": 0.6440826058387756,
      "learning_rate": 6.390319713886217e-06,
      "loss": 1.0175,
      "step": 9538
    },
    {
      "epoch": 1.7554357926036053,
      "grad_norm": 0.6304287910461426,
      "learning_rate": 6.3800366981147e-06,
      "loss": 0.9864,
      "step": 9539
    },
    {
      "epoch": 1.7556216316669764,
      "grad_norm": 0.7574819922447205,
      "learning_rate": 6.369761689860998e-06,
      "loss": 0.9718,
      "step": 9540
    },
    {
      "epoch": 1.7558074707303475,
      "grad_norm": 0.6932011246681213,
      "learning_rate": 6.35949469000392e-06,
      "loss": 0.96,
      "step": 9541
    },
    {
      "epoch": 1.7559933097937186,
      "grad_norm": 0.6913305521011353,
      "learning_rate": 6.3492356994216404e-06,
      "loss": 1.1127,
      "step": 9542
    },
    {
      "epoch": 1.7561791488570897,
      "grad_norm": 0.766208291053772,
      "learning_rate": 6.338984718991658e-06,
      "loss": 1.0068,
      "step": 9543
    },
    {
      "epoch": 1.756364987920461,
      "grad_norm": 0.6946398615837097,
      "learning_rate": 6.328741749590727e-06,
      "loss": 1.0393,
      "step": 9544
    },
    {
      "epoch": 1.756550826983832,
      "grad_norm": 0.6677290797233582,
      "learning_rate": 6.31850679209498e-06,
      "loss": 0.7805,
      "step": 9545
    },
    {
      "epoch": 1.7567366660472032,
      "grad_norm": 0.6138538718223572,
      "learning_rate": 6.308279847379806e-06,
      "loss": 0.8485,
      "step": 9546
    },
    {
      "epoch": 1.7569225051105741,
      "grad_norm": 0.733422577381134,
      "learning_rate": 6.298060916319959e-06,
      "loss": 0.9421,
      "step": 9547
    },
    {
      "epoch": 1.7571083441739455,
      "grad_norm": 0.6412607431411743,
      "learning_rate": 6.2878499997894635e-06,
      "loss": 0.9485,
      "step": 9548
    },
    {
      "epoch": 1.7572941832373163,
      "grad_norm": 0.6196644902229309,
      "learning_rate": 6.277647098661699e-06,
      "loss": 1.0333,
      "step": 9549
    },
    {
      "epoch": 1.7574800223006877,
      "grad_norm": 0.734294056892395,
      "learning_rate": 6.267452213809355e-06,
      "loss": 0.9258,
      "step": 9550
    },
    {
      "epoch": 1.7576658613640588,
      "grad_norm": 0.6560178995132446,
      "learning_rate": 6.2572653461043905e-06,
      "loss": 0.8529,
      "step": 9551
    },
    {
      "epoch": 1.7578517004274299,
      "grad_norm": 0.6958460211753845,
      "learning_rate": 6.247086496418142e-06,
      "loss": 1.0363,
      "step": 9552
    },
    {
      "epoch": 1.758037539490801,
      "grad_norm": 0.629332959651947,
      "learning_rate": 6.236915665621201e-06,
      "loss": 0.8877,
      "step": 9553
    },
    {
      "epoch": 1.758223378554172,
      "grad_norm": 0.8279165625572205,
      "learning_rate": 6.226752854583506e-06,
      "loss": 1.0771,
      "step": 9554
    },
    {
      "epoch": 1.7584092176175432,
      "grad_norm": 0.6355517506599426,
      "learning_rate": 6.216598064174306e-06,
      "loss": 0.8392,
      "step": 9555
    },
    {
      "epoch": 1.7585950566809143,
      "grad_norm": 0.7394723892211914,
      "learning_rate": 6.206451295262173e-06,
      "loss": 0.9233,
      "step": 9556
    },
    {
      "epoch": 1.7587808957442854,
      "grad_norm": 0.7708925604820251,
      "learning_rate": 6.196312548714989e-06,
      "loss": 1.1978,
      "step": 9557
    },
    {
      "epoch": 1.7589667348076565,
      "grad_norm": 0.7235886454582214,
      "learning_rate": 6.186181825399906e-06,
      "loss": 1.0615,
      "step": 9558
    },
    {
      "epoch": 1.7591525738710276,
      "grad_norm": 0.7010600566864014,
      "learning_rate": 6.176059126183475e-06,
      "loss": 0.8606,
      "step": 9559
    },
    {
      "epoch": 1.7593384129343987,
      "grad_norm": 0.7215830683708191,
      "learning_rate": 6.165944451931471e-06,
      "loss": 0.9703,
      "step": 9560
    },
    {
      "epoch": 1.75952425199777,
      "grad_norm": 0.6933256387710571,
      "learning_rate": 6.155837803509046e-06,
      "loss": 0.9035,
      "step": 9561
    },
    {
      "epoch": 1.759710091061141,
      "grad_norm": 0.823757529258728,
      "learning_rate": 6.14573918178063e-06,
      "loss": 0.9078,
      "step": 9562
    },
    {
      "epoch": 1.7598959301245123,
      "grad_norm": 0.5969639420509338,
      "learning_rate": 6.135648587609988e-06,
      "loss": 0.845,
      "step": 9563
    },
    {
      "epoch": 1.7600817691878832,
      "grad_norm": 0.6157955527305603,
      "learning_rate": 6.1255660218602074e-06,
      "loss": 1.0596,
      "step": 9564
    },
    {
      "epoch": 1.7602676082512545,
      "grad_norm": 0.723203718662262,
      "learning_rate": 6.115491485393632e-06,
      "loss": 0.9388,
      "step": 9565
    },
    {
      "epoch": 1.7604534473146254,
      "grad_norm": 0.8322511911392212,
      "learning_rate": 6.1054249790720055e-06,
      "loss": 0.9494,
      "step": 9566
    },
    {
      "epoch": 1.7606392863779967,
      "grad_norm": 0.6444146037101746,
      "learning_rate": 6.095366503756295e-06,
      "loss": 0.8682,
      "step": 9567
    },
    {
      "epoch": 1.7608251254413678,
      "grad_norm": 0.8847242593765259,
      "learning_rate": 6.085316060306834e-06,
      "loss": 1.0478,
      "step": 9568
    },
    {
      "epoch": 1.761010964504739,
      "grad_norm": 0.6584118008613586,
      "learning_rate": 6.075273649583268e-06,
      "loss": 0.7521,
      "step": 9569
    },
    {
      "epoch": 1.76119680356811,
      "grad_norm": 0.6482993364334106,
      "learning_rate": 6.065239272444556e-06,
      "loss": 1.0018,
      "step": 9570
    },
    {
      "epoch": 1.7613826426314811,
      "grad_norm": 0.6682214736938477,
      "learning_rate": 6.055212929748955e-06,
      "loss": 0.9601,
      "step": 9571
    },
    {
      "epoch": 1.7615684816948523,
      "grad_norm": 0.6621394157409668,
      "learning_rate": 6.045194622354022e-06,
      "loss": 0.8458,
      "step": 9572
    },
    {
      "epoch": 1.7617543207582234,
      "grad_norm": 0.7191212177276611,
      "learning_rate": 6.035184351116663e-06,
      "loss": 0.9726,
      "step": 9573
    },
    {
      "epoch": 1.7619401598215945,
      "grad_norm": 0.6445138454437256,
      "learning_rate": 6.025182116893069e-06,
      "loss": 0.9465,
      "step": 9574
    },
    {
      "epoch": 1.7621259988849656,
      "grad_norm": 0.6561879515647888,
      "learning_rate": 6.015187920538767e-06,
      "loss": 0.8458,
      "step": 9575
    },
    {
      "epoch": 1.762311837948337,
      "grad_norm": 0.7514671683311462,
      "learning_rate": 6.005201762908552e-06,
      "loss": 1.0257,
      "step": 9576
    },
    {
      "epoch": 1.7624976770117078,
      "grad_norm": 0.6483860015869141,
      "learning_rate": 5.9952236448565735e-06,
      "loss": 0.9824,
      "step": 9577
    },
    {
      "epoch": 1.7626835160750791,
      "grad_norm": 0.6980741024017334,
      "learning_rate": 5.985253567236304e-06,
      "loss": 1.1466,
      "step": 9578
    },
    {
      "epoch": 1.76286935513845,
      "grad_norm": 0.6964401602745056,
      "learning_rate": 5.975291530900484e-06,
      "loss": 0.801,
      "step": 9579
    },
    {
      "epoch": 1.7630551942018213,
      "grad_norm": 0.6684138774871826,
      "learning_rate": 5.9653375367011985e-06,
      "loss": 0.7537,
      "step": 9580
    },
    {
      "epoch": 1.7632410332651922,
      "grad_norm": 0.7552790641784668,
      "learning_rate": 5.955391585489822e-06,
      "loss": 0.9026,
      "step": 9581
    },
    {
      "epoch": 1.7634268723285635,
      "grad_norm": 0.6640274524688721,
      "learning_rate": 5.945453678117063e-06,
      "loss": 1.0204,
      "step": 9582
    },
    {
      "epoch": 1.7636127113919344,
      "grad_norm": 0.6475071310997009,
      "learning_rate": 5.935523815432919e-06,
      "loss": 0.8581,
      "step": 9583
    },
    {
      "epoch": 1.7637985504553058,
      "grad_norm": 0.7806620001792908,
      "learning_rate": 5.925601998286734e-06,
      "loss": 1.1294,
      "step": 9584
    },
    {
      "epoch": 1.7639843895186769,
      "grad_norm": 0.7511613965034485,
      "learning_rate": 5.915688227527139e-06,
      "loss": 1.1108,
      "step": 9585
    },
    {
      "epoch": 1.764170228582048,
      "grad_norm": 0.6426988840103149,
      "learning_rate": 5.905782504002055e-06,
      "loss": 0.6887,
      "step": 9586
    },
    {
      "epoch": 1.764356067645419,
      "grad_norm": 0.8487985730171204,
      "learning_rate": 5.8958848285587844e-06,
      "loss": 0.9581,
      "step": 9587
    },
    {
      "epoch": 1.7645419067087902,
      "grad_norm": 0.7105994820594788,
      "learning_rate": 5.885995202043848e-06,
      "loss": 0.9261,
      "step": 9588
    },
    {
      "epoch": 1.7647277457721613,
      "grad_norm": 0.8497649431228638,
      "learning_rate": 5.876113625303159e-06,
      "loss": 1.0863,
      "step": 9589
    },
    {
      "epoch": 1.7649135848355324,
      "grad_norm": 0.6333727240562439,
      "learning_rate": 5.866240099181886e-06,
      "loss": 0.8614,
      "step": 9590
    },
    {
      "epoch": 1.7650994238989035,
      "grad_norm": 1.783128261566162,
      "learning_rate": 5.856374624524552e-06,
      "loss": 1.5696,
      "step": 9591
    },
    {
      "epoch": 1.7652852629622746,
      "grad_norm": 0.6523680090904236,
      "learning_rate": 5.846517202174984e-06,
      "loss": 0.9532,
      "step": 9592
    },
    {
      "epoch": 1.765471102025646,
      "grad_norm": 1.0413211584091187,
      "learning_rate": 5.836667832976272e-06,
      "loss": 0.8771,
      "step": 9593
    },
    {
      "epoch": 1.7656569410890168,
      "grad_norm": 0.6950120329856873,
      "learning_rate": 5.8268265177709e-06,
      "loss": 0.7809,
      "step": 9594
    },
    {
      "epoch": 1.7658427801523882,
      "grad_norm": 0.7298243641853333,
      "learning_rate": 5.81699325740056e-06,
      "loss": 0.9626,
      "step": 9595
    },
    {
      "epoch": 1.766028619215759,
      "grad_norm": 0.7584399580955505,
      "learning_rate": 5.8071680527063575e-06,
      "loss": 1.0026,
      "step": 9596
    },
    {
      "epoch": 1.7662144582791304,
      "grad_norm": 0.6953753232955933,
      "learning_rate": 5.797350904528653e-06,
      "loss": 0.9389,
      "step": 9597
    },
    {
      "epoch": 1.7664002973425013,
      "grad_norm": 0.6708144545555115,
      "learning_rate": 5.78754181370712e-06,
      "loss": 1.1463,
      "step": 9598
    },
    {
      "epoch": 1.7665861364058726,
      "grad_norm": 0.6775829195976257,
      "learning_rate": 5.777740781080743e-06,
      "loss": 1.1084,
      "step": 9599
    },
    {
      "epoch": 1.7667719754692435,
      "grad_norm": 0.837083637714386,
      "learning_rate": 5.76794780748785e-06,
      "loss": 0.9402,
      "step": 9600
    },
    {
      "epoch": 1.7669578145326148,
      "grad_norm": 0.7145852446556091,
      "learning_rate": 5.75816289376605e-06,
      "loss": 1.1851,
      "step": 9601
    },
    {
      "epoch": 1.767143653595986,
      "grad_norm": 0.7356708645820618,
      "learning_rate": 5.748386040752252e-06,
      "loss": 0.9147,
      "step": 9602
    },
    {
      "epoch": 1.767329492659357,
      "grad_norm": 0.8315185904502869,
      "learning_rate": 5.738617249282707e-06,
      "loss": 1.1358,
      "step": 9603
    },
    {
      "epoch": 1.7675153317227281,
      "grad_norm": 0.6617395877838135,
      "learning_rate": 5.728856520192949e-06,
      "loss": 1.068,
      "step": 9604
    },
    {
      "epoch": 1.7677011707860992,
      "grad_norm": 0.8139854669570923,
      "learning_rate": 5.719103854317831e-06,
      "loss": 1.0914,
      "step": 9605
    },
    {
      "epoch": 1.7678870098494703,
      "grad_norm": 0.6905813813209534,
      "learning_rate": 5.709359252491553e-06,
      "loss": 0.9102,
      "step": 9606
    },
    {
      "epoch": 1.7680728489128414,
      "grad_norm": 0.7273473143577576,
      "learning_rate": 5.699622715547548e-06,
      "loss": 1.16,
      "step": 9607
    },
    {
      "epoch": 1.7682586879762126,
      "grad_norm": 0.7004194259643555,
      "learning_rate": 5.689894244318628e-06,
      "loss": 1.0497,
      "step": 9608
    },
    {
      "epoch": 1.7684445270395837,
      "grad_norm": 0.6936884522438049,
      "learning_rate": 5.680173839636882e-06,
      "loss": 0.9429,
      "step": 9609
    },
    {
      "epoch": 1.768630366102955,
      "grad_norm": 0.916488766670227,
      "learning_rate": 5.670461502333724e-06,
      "loss": 0.9076,
      "step": 9610
    },
    {
      "epoch": 1.7688162051663259,
      "grad_norm": 0.7607235908508301,
      "learning_rate": 5.6607572332398554e-06,
      "loss": 1.1614,
      "step": 9611
    },
    {
      "epoch": 1.7690020442296972,
      "grad_norm": 0.7022486925125122,
      "learning_rate": 5.651061033185323e-06,
      "loss": 0.8073,
      "step": 9612
    },
    {
      "epoch": 1.769187883293068,
      "grad_norm": 0.7221197485923767,
      "learning_rate": 5.641372902999442e-06,
      "loss": 0.8381,
      "step": 9613
    },
    {
      "epoch": 1.7693737223564394,
      "grad_norm": 0.8069917559623718,
      "learning_rate": 5.631692843510883e-06,
      "loss": 1.2511,
      "step": 9614
    },
    {
      "epoch": 1.7695595614198103,
      "grad_norm": 0.7064328193664551,
      "learning_rate": 5.622020855547605e-06,
      "loss": 1.1579,
      "step": 9615
    },
    {
      "epoch": 1.7697454004831816,
      "grad_norm": 0.8496193885803223,
      "learning_rate": 5.612356939936836e-06,
      "loss": 0.8325,
      "step": 9616
    },
    {
      "epoch": 1.7699312395465527,
      "grad_norm": 0.6425907015800476,
      "learning_rate": 5.602701097505203e-06,
      "loss": 1.0847,
      "step": 9617
    },
    {
      "epoch": 1.7701170786099238,
      "grad_norm": 0.7497144341468811,
      "learning_rate": 5.593053329078546e-06,
      "loss": 0.6886,
      "step": 9618
    },
    {
      "epoch": 1.770302917673295,
      "grad_norm": 0.6739358305931091,
      "learning_rate": 5.583413635482082e-06,
      "loss": 0.9889,
      "step": 9619
    },
    {
      "epoch": 1.770488756736666,
      "grad_norm": 0.5923432111740112,
      "learning_rate": 5.573782017540319e-06,
      "loss": 0.8585,
      "step": 9620
    },
    {
      "epoch": 1.7706745958000372,
      "grad_norm": 0.7298094630241394,
      "learning_rate": 5.564158476077053e-06,
      "loss": 0.8495,
      "step": 9621
    },
    {
      "epoch": 1.7708604348634083,
      "grad_norm": 0.6663339138031006,
      "learning_rate": 5.554543011915425e-06,
      "loss": 0.828,
      "step": 9622
    },
    {
      "epoch": 1.7710462739267794,
      "grad_norm": 0.7471387982368469,
      "learning_rate": 5.544935625877834e-06,
      "loss": 1.0049,
      "step": 9623
    },
    {
      "epoch": 1.7712321129901505,
      "grad_norm": 0.6680027842521667,
      "learning_rate": 5.535336318786077e-06,
      "loss": 1.1446,
      "step": 9624
    },
    {
      "epoch": 1.7714179520535218,
      "grad_norm": 0.8819547295570374,
      "learning_rate": 5.525745091461143e-06,
      "loss": 1.2994,
      "step": 9625
    },
    {
      "epoch": 1.7716037911168927,
      "grad_norm": 0.7097487449645996,
      "learning_rate": 5.516161944723419e-06,
      "loss": 0.9665,
      "step": 9626
    },
    {
      "epoch": 1.771789630180264,
      "grad_norm": 0.8968320488929749,
      "learning_rate": 5.506586879392572e-06,
      "loss": 1.1205,
      "step": 9627
    },
    {
      "epoch": 1.771975469243635,
      "grad_norm": 0.7016741633415222,
      "learning_rate": 5.49701989628757e-06,
      "loss": 1.1043,
      "step": 9628
    },
    {
      "epoch": 1.7721613083070062,
      "grad_norm": 0.7549055814743042,
      "learning_rate": 5.4874609962267236e-06,
      "loss": 0.809,
      "step": 9629
    },
    {
      "epoch": 1.7723471473703771,
      "grad_norm": 0.6078971028327942,
      "learning_rate": 5.47791018002759e-06,
      "loss": 0.7538,
      "step": 9630
    },
    {
      "epoch": 1.7725329864337485,
      "grad_norm": 1.518277883529663,
      "learning_rate": 5.468367448507106e-06,
      "loss": 1.4723,
      "step": 9631
    },
    {
      "epoch": 1.7727188254971193,
      "grad_norm": 0.8478793501853943,
      "learning_rate": 5.4588328024814505e-06,
      "loss": 0.928,
      "step": 9632
    },
    {
      "epoch": 1.7729046645604907,
      "grad_norm": 0.874586284160614,
      "learning_rate": 5.44930624276615e-06,
      "loss": 0.8689,
      "step": 9633
    },
    {
      "epoch": 1.7730905036238618,
      "grad_norm": 0.7951120138168335,
      "learning_rate": 5.439787770176053e-06,
      "loss": 0.8361,
      "step": 9634
    },
    {
      "epoch": 1.7732763426872329,
      "grad_norm": 0.7073873281478882,
      "learning_rate": 5.430277385525273e-06,
      "loss": 1.2187,
      "step": 9635
    },
    {
      "epoch": 1.773462181750604,
      "grad_norm": 0.6185064315795898,
      "learning_rate": 5.420775089627272e-06,
      "loss": 0.8071,
      "step": 9636
    },
    {
      "epoch": 1.773648020813975,
      "grad_norm": 0.7715813517570496,
      "learning_rate": 5.411280883294767e-06,
      "loss": 0.9601,
      "step": 9637
    },
    {
      "epoch": 1.7738338598773462,
      "grad_norm": 0.6617690324783325,
      "learning_rate": 5.4017947673398625e-06,
      "loss": 1.1743,
      "step": 9638
    },
    {
      "epoch": 1.7740196989407173,
      "grad_norm": 0.7935189008712769,
      "learning_rate": 5.3923167425738885e-06,
      "loss": 1.0538,
      "step": 9639
    },
    {
      "epoch": 1.7742055380040884,
      "grad_norm": 0.7197004556655884,
      "learning_rate": 5.38284680980754e-06,
      "loss": 1.0795,
      "step": 9640
    },
    {
      "epoch": 1.7743913770674595,
      "grad_norm": 0.6420508027076721,
      "learning_rate": 5.373384969850792e-06,
      "loss": 1.0606,
      "step": 9641
    },
    {
      "epoch": 1.7745772161308309,
      "grad_norm": 0.9372122883796692,
      "learning_rate": 5.363931223512964e-06,
      "loss": 0.8252,
      "step": 9642
    },
    {
      "epoch": 1.7747630551942017,
      "grad_norm": 0.6339817643165588,
      "learning_rate": 5.354485571602619e-06,
      "loss": 0.7044,
      "step": 9643
    },
    {
      "epoch": 1.774948894257573,
      "grad_norm": 0.6962215900421143,
      "learning_rate": 5.3450480149276785e-06,
      "loss": 0.9662,
      "step": 9644
    },
    {
      "epoch": 1.775134733320944,
      "grad_norm": 0.7110217213630676,
      "learning_rate": 5.335618554295374e-06,
      "loss": 0.9173,
      "step": 9645
    },
    {
      "epoch": 1.7753205723843153,
      "grad_norm": 0.7352878451347351,
      "learning_rate": 5.326197190512194e-06,
      "loss": 1.0549,
      "step": 9646
    },
    {
      "epoch": 1.7755064114476862,
      "grad_norm": 0.6918430328369141,
      "learning_rate": 5.316783924383983e-06,
      "loss": 0.8215,
      "step": 9647
    },
    {
      "epoch": 1.7756922505110575,
      "grad_norm": 0.7566372156143188,
      "learning_rate": 5.307378756715898e-06,
      "loss": 1.1456,
      "step": 9648
    },
    {
      "epoch": 1.7758780895744284,
      "grad_norm": 0.5607185363769531,
      "learning_rate": 5.297981688312348e-06,
      "loss": 0.5717,
      "step": 9649
    },
    {
      "epoch": 1.7760639286377997,
      "grad_norm": 0.788888156414032,
      "learning_rate": 5.288592719977114e-06,
      "loss": 0.8542,
      "step": 9650
    },
    {
      "epoch": 1.7762497677011708,
      "grad_norm": 0.648224949836731,
      "learning_rate": 5.279211852513233e-06,
      "loss": 0.5628,
      "step": 9651
    },
    {
      "epoch": 1.776435606764542,
      "grad_norm": 0.6493148803710938,
      "learning_rate": 5.269839086723094e-06,
      "loss": 0.8436,
      "step": 9652
    },
    {
      "epoch": 1.776621445827913,
      "grad_norm": 0.7417361736297607,
      "learning_rate": 5.260474423408346e-06,
      "loss": 0.9803,
      "step": 9653
    },
    {
      "epoch": 1.7768072848912841,
      "grad_norm": 0.8001540303230286,
      "learning_rate": 5.251117863369981e-06,
      "loss": 0.8131,
      "step": 9654
    },
    {
      "epoch": 1.7769931239546553,
      "grad_norm": 0.5939586162567139,
      "learning_rate": 5.241769407408293e-06,
      "loss": 1.0488,
      "step": 9655
    },
    {
      "epoch": 1.7771789630180264,
      "grad_norm": 0.7578622698783875,
      "learning_rate": 5.232429056322852e-06,
      "loss": 0.971,
      "step": 9656
    },
    {
      "epoch": 1.7773648020813975,
      "grad_norm": 0.7863082885742188,
      "learning_rate": 5.223096810912576e-06,
      "loss": 0.9085,
      "step": 9657
    },
    {
      "epoch": 1.7775506411447686,
      "grad_norm": 0.8486718535423279,
      "learning_rate": 5.21377267197567e-06,
      "loss": 0.8571,
      "step": 9658
    },
    {
      "epoch": 1.77773648020814,
      "grad_norm": 0.97194504737854,
      "learning_rate": 5.204456640309663e-06,
      "loss": 0.6344,
      "step": 9659
    },
    {
      "epoch": 1.7779223192715108,
      "grad_norm": 0.7148550748825073,
      "learning_rate": 5.19514871671134e-06,
      "loss": 1.2341,
      "step": 9660
    },
    {
      "epoch": 1.7781081583348821,
      "grad_norm": 0.9144877195358276,
      "learning_rate": 5.185848901976853e-06,
      "loss": 1.1233,
      "step": 9661
    },
    {
      "epoch": 1.778293997398253,
      "grad_norm": 0.7061643004417419,
      "learning_rate": 5.1765571969016436e-06,
      "loss": 0.8772,
      "step": 9662
    },
    {
      "epoch": 1.7784798364616243,
      "grad_norm": 0.5985503196716309,
      "learning_rate": 5.167273602280421e-06,
      "loss": 1.1218,
      "step": 9663
    },
    {
      "epoch": 1.7786656755249952,
      "grad_norm": 0.7457888126373291,
      "learning_rate": 5.15799811890727e-06,
      "loss": 1.3084,
      "step": 9664
    },
    {
      "epoch": 1.7788515145883665,
      "grad_norm": 0.7234838008880615,
      "learning_rate": 5.148730747575492e-06,
      "loss": 0.8696,
      "step": 9665
    },
    {
      "epoch": 1.7790373536517374,
      "grad_norm": 0.7006774544715881,
      "learning_rate": 5.1394714890777965e-06,
      "loss": 0.8938,
      "step": 9666
    },
    {
      "epoch": 1.7792231927151088,
      "grad_norm": 0.6671582460403442,
      "learning_rate": 5.130220344206116e-06,
      "loss": 0.7923,
      "step": 9667
    },
    {
      "epoch": 1.7794090317784799,
      "grad_norm": 0.7441495060920715,
      "learning_rate": 5.120977313751718e-06,
      "loss": 0.8206,
      "step": 9668
    },
    {
      "epoch": 1.779594870841851,
      "grad_norm": 0.7187232971191406,
      "learning_rate": 5.111742398505215e-06,
      "loss": 0.7165,
      "step": 9669
    },
    {
      "epoch": 1.779780709905222,
      "grad_norm": 0.6727659702301025,
      "learning_rate": 5.102515599256441e-06,
      "loss": 0.7269,
      "step": 9670
    },
    {
      "epoch": 1.7799665489685932,
      "grad_norm": 0.6532821655273438,
      "learning_rate": 5.0932969167946095e-06,
      "loss": 0.8986,
      "step": 9671
    },
    {
      "epoch": 1.7801523880319643,
      "grad_norm": 0.7378084659576416,
      "learning_rate": 5.084086351908224e-06,
      "loss": 1.0243,
      "step": 9672
    },
    {
      "epoch": 1.7803382270953354,
      "grad_norm": 0.7145583033561707,
      "learning_rate": 5.074883905385075e-06,
      "loss": 1.0555,
      "step": 9673
    },
    {
      "epoch": 1.7805240661587065,
      "grad_norm": 0.7014021873474121,
      "learning_rate": 5.065689578012267e-06,
      "loss": 0.9485,
      "step": 9674
    },
    {
      "epoch": 1.7807099052220776,
      "grad_norm": 0.6306982636451721,
      "learning_rate": 5.056503370576193e-06,
      "loss": 1.0009,
      "step": 9675
    },
    {
      "epoch": 1.780895744285449,
      "grad_norm": 0.7286544442176819,
      "learning_rate": 5.0473252838626025e-06,
      "loss": 1.0114,
      "step": 9676
    },
    {
      "epoch": 1.7810815833488198,
      "grad_norm": 0.7763290405273438,
      "learning_rate": 5.038155318656501e-06,
      "loss": 0.9,
      "step": 9677
    },
    {
      "epoch": 1.7812674224121912,
      "grad_norm": 0.6970409154891968,
      "learning_rate": 5.028993475742217e-06,
      "loss": 0.8089,
      "step": 9678
    },
    {
      "epoch": 1.781453261475562,
      "grad_norm": 0.6255539059638977,
      "learning_rate": 5.019839755903366e-06,
      "loss": 0.9643,
      "step": 9679
    },
    {
      "epoch": 1.7816391005389334,
      "grad_norm": 0.7349768877029419,
      "learning_rate": 5.0106941599229125e-06,
      "loss": 0.9415,
      "step": 9680
    },
    {
      "epoch": 1.7818249396023043,
      "grad_norm": 0.6985937356948853,
      "learning_rate": 5.0015566885830756e-06,
      "loss": 0.8242,
      "step": 9681
    },
    {
      "epoch": 1.7820107786656756,
      "grad_norm": 0.7383229732513428,
      "learning_rate": 4.992427342665406e-06,
      "loss": 1.0944,
      "step": 9682
    },
    {
      "epoch": 1.7821966177290467,
      "grad_norm": 0.7543425559997559,
      "learning_rate": 4.983306122950793e-06,
      "loss": 1.0926,
      "step": 9683
    },
    {
      "epoch": 1.7823824567924178,
      "grad_norm": 0.699912965297699,
      "learning_rate": 4.974193030219343e-06,
      "loss": 0.7168,
      "step": 9684
    },
    {
      "epoch": 1.782568295855789,
      "grad_norm": 0.6963754296302795,
      "learning_rate": 4.965088065250545e-06,
      "loss": 0.9581,
      "step": 9685
    },
    {
      "epoch": 1.78275413491916,
      "grad_norm": 0.8270363211631775,
      "learning_rate": 4.955991228823154e-06,
      "loss": 0.9795,
      "step": 9686
    },
    {
      "epoch": 1.7829399739825311,
      "grad_norm": 0.6962025761604309,
      "learning_rate": 4.946902521715269e-06,
      "loss": 0.9428,
      "step": 9687
    },
    {
      "epoch": 1.7831258130459022,
      "grad_norm": 0.6201028823852539,
      "learning_rate": 4.9378219447042354e-06,
      "loss": 0.7519,
      "step": 9688
    },
    {
      "epoch": 1.7833116521092733,
      "grad_norm": 0.7408379316329956,
      "learning_rate": 4.928749498566743e-06,
      "loss": 1.0062,
      "step": 9689
    },
    {
      "epoch": 1.7834974911726444,
      "grad_norm": 0.6847137212753296,
      "learning_rate": 4.919685184078804e-06,
      "loss": 1.0721,
      "step": 9690
    },
    {
      "epoch": 1.7836833302360158,
      "grad_norm": 0.7941872477531433,
      "learning_rate": 4.910629002015665e-06,
      "loss": 0.7529,
      "step": 9691
    },
    {
      "epoch": 1.7838691692993867,
      "grad_norm": 0.7792496085166931,
      "learning_rate": 4.901580953151963e-06,
      "loss": 0.9921,
      "step": 9692
    },
    {
      "epoch": 1.784055008362758,
      "grad_norm": 0.5640096068382263,
      "learning_rate": 4.8925410382615555e-06,
      "loss": 0.7529,
      "step": 9693
    },
    {
      "epoch": 1.7842408474261289,
      "grad_norm": 0.8288540244102478,
      "learning_rate": 4.88350925811768e-06,
      "loss": 1.0118,
      "step": 9694
    },
    {
      "epoch": 1.7844266864895002,
      "grad_norm": 0.7848095297813416,
      "learning_rate": 4.87448561349283e-06,
      "loss": 1.1714,
      "step": 9695
    },
    {
      "epoch": 1.784612525552871,
      "grad_norm": 0.715578019618988,
      "learning_rate": 4.865470105158809e-06,
      "loss": 0.9999,
      "step": 9696
    },
    {
      "epoch": 1.7847983646162424,
      "grad_norm": 0.6751718521118164,
      "learning_rate": 4.856462733886757e-06,
      "loss": 0.8101,
      "step": 9697
    },
    {
      "epoch": 1.7849842036796133,
      "grad_norm": 0.7342599630355835,
      "learning_rate": 4.847463500447069e-06,
      "loss": 0.9857,
      "step": 9698
    },
    {
      "epoch": 1.7851700427429846,
      "grad_norm": 0.7690227627754211,
      "learning_rate": 4.838472405609495e-06,
      "loss": 1.0378,
      "step": 9699
    },
    {
      "epoch": 1.7853558818063557,
      "grad_norm": 0.6772790551185608,
      "learning_rate": 4.829489450143021e-06,
      "loss": 1.0879,
      "step": 9700
    },
    {
      "epoch": 1.7855417208697268,
      "grad_norm": 0.8678627014160156,
      "learning_rate": 4.82051463481602e-06,
      "loss": 1.079,
      "step": 9701
    },
    {
      "epoch": 1.785727559933098,
      "grad_norm": 0.6281694769859314,
      "learning_rate": 4.811547960396101e-06,
      "loss": 0.8617,
      "step": 9702
    },
    {
      "epoch": 1.785913398996469,
      "grad_norm": 0.7710700035095215,
      "learning_rate": 4.8025894276502285e-06,
      "loss": 0.7237,
      "step": 9703
    },
    {
      "epoch": 1.7860992380598402,
      "grad_norm": 0.7536906003952026,
      "learning_rate": 4.7936390373446325e-06,
      "loss": 0.6335,
      "step": 9704
    },
    {
      "epoch": 1.7862850771232113,
      "grad_norm": 0.93522709608078,
      "learning_rate": 4.784696790244847e-06,
      "loss": 0.979,
      "step": 9705
    },
    {
      "epoch": 1.7864709161865824,
      "grad_norm": 0.6946282386779785,
      "learning_rate": 4.775762687115748e-06,
      "loss": 0.9663,
      "step": 9706
    },
    {
      "epoch": 1.7866567552499535,
      "grad_norm": 0.6784142255783081,
      "learning_rate": 4.766836728721458e-06,
      "loss": 0.7839,
      "step": 9707
    },
    {
      "epoch": 1.7868425943133248,
      "grad_norm": 0.7739217281341553,
      "learning_rate": 4.757918915825455e-06,
      "loss": 0.9678,
      "step": 9708
    },
    {
      "epoch": 1.7870284333766957,
      "grad_norm": 0.6688413023948669,
      "learning_rate": 4.749009249190495e-06,
      "loss": 0.9067,
      "step": 9709
    },
    {
      "epoch": 1.787214272440067,
      "grad_norm": 0.7805230617523193,
      "learning_rate": 4.7401077295786244e-06,
      "loss": 0.8597,
      "step": 9710
    },
    {
      "epoch": 1.787400111503438,
      "grad_norm": 0.7463875412940979,
      "learning_rate": 4.731214357751245e-06,
      "loss": 1.0316,
      "step": 9711
    },
    {
      "epoch": 1.7875859505668092,
      "grad_norm": 0.7301071286201477,
      "learning_rate": 4.722329134468983e-06,
      "loss": 0.9194,
      "step": 9712
    },
    {
      "epoch": 1.7877717896301801,
      "grad_norm": 0.702477216720581,
      "learning_rate": 4.713452060491852e-06,
      "loss": 1.176,
      "step": 9713
    },
    {
      "epoch": 1.7879576286935515,
      "grad_norm": 0.6391510963439941,
      "learning_rate": 4.704583136579088e-06,
      "loss": 0.8577,
      "step": 9714
    },
    {
      "epoch": 1.7881434677569223,
      "grad_norm": 0.71745765209198,
      "learning_rate": 4.695722363489297e-06,
      "loss": 0.9414,
      "step": 9715
    },
    {
      "epoch": 1.7883293068202937,
      "grad_norm": 0.7245863676071167,
      "learning_rate": 4.6868697419803485e-06,
      "loss": 0.9282,
      "step": 9716
    },
    {
      "epoch": 1.7885151458836648,
      "grad_norm": 0.6648089289665222,
      "learning_rate": 4.678025272809427e-06,
      "loss": 0.9492,
      "step": 9717
    },
    {
      "epoch": 1.7887009849470359,
      "grad_norm": 0.7349907159805298,
      "learning_rate": 4.6691889567330395e-06,
      "loss": 1.0022,
      "step": 9718
    },
    {
      "epoch": 1.788886824010407,
      "grad_norm": 0.713378369808197,
      "learning_rate": 4.660360794506946e-06,
      "loss": 1.0444,
      "step": 9719
    },
    {
      "epoch": 1.789072663073778,
      "grad_norm": 0.8368687629699707,
      "learning_rate": 4.651540786886255e-06,
      "loss": 1.0523,
      "step": 9720
    },
    {
      "epoch": 1.7892585021371492,
      "grad_norm": 0.6994075775146484,
      "learning_rate": 4.6427289346253515e-06,
      "loss": 1.0559,
      "step": 9721
    },
    {
      "epoch": 1.7894443412005203,
      "grad_norm": 0.8998470306396484,
      "learning_rate": 4.633925238477943e-06,
      "loss": 1.0073,
      "step": 9722
    },
    {
      "epoch": 1.7896301802638914,
      "grad_norm": 0.7161475419998169,
      "learning_rate": 4.625129699197006e-06,
      "loss": 0.8975,
      "step": 9723
    },
    {
      "epoch": 1.7898160193272625,
      "grad_norm": 0.8403297662734985,
      "learning_rate": 4.616342317534861e-06,
      "loss": 0.8977,
      "step": 9724
    },
    {
      "epoch": 1.7900018583906339,
      "grad_norm": 0.6233807802200317,
      "learning_rate": 4.607563094243128e-06,
      "loss": 0.8097,
      "step": 9725
    },
    {
      "epoch": 1.7901876974540047,
      "grad_norm": 0.6614106297492981,
      "learning_rate": 4.598792030072674e-06,
      "loss": 1.1261,
      "step": 9726
    },
    {
      "epoch": 1.790373536517376,
      "grad_norm": 0.7223928570747375,
      "learning_rate": 4.590029125773742e-06,
      "loss": 0.748,
      "step": 9727
    },
    {
      "epoch": 1.790559375580747,
      "grad_norm": 0.6953086853027344,
      "learning_rate": 4.581274382095812e-06,
      "loss": 0.7081,
      "step": 9728
    },
    {
      "epoch": 1.7907452146441183,
      "grad_norm": 0.7670634984970093,
      "learning_rate": 4.572527799787718e-06,
      "loss": 0.954,
      "step": 9729
    },
    {
      "epoch": 1.7909310537074892,
      "grad_norm": 0.6398134827613831,
      "learning_rate": 4.5637893795975605e-06,
      "loss": 0.9984,
      "step": 9730
    },
    {
      "epoch": 1.7911168927708605,
      "grad_norm": 0.5937197208404541,
      "learning_rate": 4.5550591222727645e-06,
      "loss": 0.8642,
      "step": 9731
    },
    {
      "epoch": 1.7913027318342316,
      "grad_norm": 1.3300977945327759,
      "learning_rate": 4.546337028560066e-06,
      "loss": 1.1381,
      "step": 9732
    },
    {
      "epoch": 1.7914885708976027,
      "grad_norm": 0.7279658317565918,
      "learning_rate": 4.537623099205435e-06,
      "loss": 0.819,
      "step": 9733
    },
    {
      "epoch": 1.7916744099609738,
      "grad_norm": 0.7543433308601379,
      "learning_rate": 4.5289173349542414e-06,
      "loss": 0.7391,
      "step": 9734
    },
    {
      "epoch": 1.791860249024345,
      "grad_norm": 0.678759753704071,
      "learning_rate": 4.520219736551079e-06,
      "loss": 1.0428,
      "step": 9735
    },
    {
      "epoch": 1.792046088087716,
      "grad_norm": 0.6893246173858643,
      "learning_rate": 4.5115303047398745e-06,
      "loss": 0.808,
      "step": 9736
    },
    {
      "epoch": 1.7922319271510871,
      "grad_norm": 0.6843549013137817,
      "learning_rate": 4.502849040263879e-06,
      "loss": 0.9483,
      "step": 9737
    },
    {
      "epoch": 1.7924177662144583,
      "grad_norm": 0.7032724618911743,
      "learning_rate": 4.494175943865597e-06,
      "loss": 1.0616,
      "step": 9738
    },
    {
      "epoch": 1.7926036052778294,
      "grad_norm": 0.6670497059822083,
      "learning_rate": 4.4855110162868585e-06,
      "loss": 0.9504,
      "step": 9739
    },
    {
      "epoch": 1.7927894443412005,
      "grad_norm": 0.6596825122833252,
      "learning_rate": 4.4768542582687925e-06,
      "loss": 0.992,
      "step": 9740
    },
    {
      "epoch": 1.7929752834045716,
      "grad_norm": 0.7123975157737732,
      "learning_rate": 4.468205670551851e-06,
      "loss": 0.8786,
      "step": 9741
    },
    {
      "epoch": 1.793161122467943,
      "grad_norm": 0.6945950984954834,
      "learning_rate": 4.45956525387573e-06,
      "loss": 0.967,
      "step": 9742
    },
    {
      "epoch": 1.7933469615313138,
      "grad_norm": 1.4450714588165283,
      "learning_rate": 4.4509330089794855e-06,
      "loss": 1.2641,
      "step": 9743
    },
    {
      "epoch": 1.7935328005946851,
      "grad_norm": 0.7171867489814758,
      "learning_rate": 4.442308936601458e-06,
      "loss": 0.9805,
      "step": 9744
    },
    {
      "epoch": 1.793718639658056,
      "grad_norm": 0.737531840801239,
      "learning_rate": 4.4336930374792695e-06,
      "loss": 0.9138,
      "step": 9745
    },
    {
      "epoch": 1.7939044787214273,
      "grad_norm": 0.7215433120727539,
      "learning_rate": 4.425085312349853e-06,
      "loss": 0.9392,
      "step": 9746
    },
    {
      "epoch": 1.7940903177847982,
      "grad_norm": 0.752029299736023,
      "learning_rate": 4.416485761949462e-06,
      "loss": 0.997,
      "step": 9747
    },
    {
      "epoch": 1.7942761568481695,
      "grad_norm": 0.8215382695198059,
      "learning_rate": 4.407894387013634e-06,
      "loss": 0.8914,
      "step": 9748
    },
    {
      "epoch": 1.7944619959115407,
      "grad_norm": 0.7880039215087891,
      "learning_rate": 4.399311188277189e-06,
      "loss": 0.9539,
      "step": 9749
    },
    {
      "epoch": 1.7946478349749118,
      "grad_norm": 1.2010185718536377,
      "learning_rate": 4.390736166474285e-06,
      "loss": 1.1393,
      "step": 9750
    },
    {
      "epoch": 1.7948336740382829,
      "grad_norm": 0.5865445137023926,
      "learning_rate": 4.382169322338359e-06,
      "loss": 0.8907,
      "step": 9751
    },
    {
      "epoch": 1.795019513101654,
      "grad_norm": 0.7465519309043884,
      "learning_rate": 4.373610656602134e-06,
      "loss": 0.8727,
      "step": 9752
    },
    {
      "epoch": 1.795205352165025,
      "grad_norm": 0.5926641821861267,
      "learning_rate": 4.365060169997692e-06,
      "loss": 1.0208,
      "step": 9753
    },
    {
      "epoch": 1.7953911912283962,
      "grad_norm": 0.6432677507400513,
      "learning_rate": 4.356517863256326e-06,
      "loss": 0.9708,
      "step": 9754
    },
    {
      "epoch": 1.7955770302917673,
      "grad_norm": 0.7736765742301941,
      "learning_rate": 4.347983737108708e-06,
      "loss": 1.0175,
      "step": 9755
    },
    {
      "epoch": 1.7957628693551384,
      "grad_norm": 0.8660035729408264,
      "learning_rate": 4.339457792284762e-06,
      "loss": 0.8954,
      "step": 9756
    },
    {
      "epoch": 1.7959487084185097,
      "grad_norm": 0.7590531706809998,
      "learning_rate": 4.330940029513741e-06,
      "loss": 0.9476,
      "step": 9757
    },
    {
      "epoch": 1.7961345474818806,
      "grad_norm": 0.7696019411087036,
      "learning_rate": 4.322430449524206e-06,
      "loss": 1.1094,
      "step": 9758
    },
    {
      "epoch": 1.796320386545252,
      "grad_norm": 0.6668081879615784,
      "learning_rate": 4.313929053043963e-06,
      "loss": 0.8396,
      "step": 9759
    },
    {
      "epoch": 1.7965062256086228,
      "grad_norm": 0.8675522804260254,
      "learning_rate": 4.305435840800176e-06,
      "loss": 0.8775,
      "step": 9760
    },
    {
      "epoch": 1.7966920646719942,
      "grad_norm": 0.8108303546905518,
      "learning_rate": 4.2969508135192844e-06,
      "loss": 0.8486,
      "step": 9761
    },
    {
      "epoch": 1.796877903735365,
      "grad_norm": 0.638525664806366,
      "learning_rate": 4.288473971927043e-06,
      "loss": 0.7874,
      "step": 9762
    },
    {
      "epoch": 1.7970637427987364,
      "grad_norm": 0.6676344871520996,
      "learning_rate": 4.2800053167484696e-06,
      "loss": 0.5854,
      "step": 9763
    },
    {
      "epoch": 1.7972495818621073,
      "grad_norm": 0.6011619567871094,
      "learning_rate": 4.27154484870792e-06,
      "loss": 0.8661,
      "step": 9764
    },
    {
      "epoch": 1.7974354209254786,
      "grad_norm": 1.2263431549072266,
      "learning_rate": 4.263092568529048e-06,
      "loss": 1.3577,
      "step": 9765
    },
    {
      "epoch": 1.7976212599888497,
      "grad_norm": 0.7756538391113281,
      "learning_rate": 4.254648476934775e-06,
      "loss": 0.9552,
      "step": 9766
    },
    {
      "epoch": 1.7978070990522208,
      "grad_norm": 1.5570555925369263,
      "learning_rate": 4.246212574647357e-06,
      "loss": 1.2383,
      "step": 9767
    },
    {
      "epoch": 1.797992938115592,
      "grad_norm": 0.8159297108650208,
      "learning_rate": 4.237784862388316e-06,
      "loss": 1.101,
      "step": 9768
    },
    {
      "epoch": 1.798178777178963,
      "grad_norm": 0.6934140920639038,
      "learning_rate": 4.229365340878522e-06,
      "loss": 1.1076,
      "step": 9769
    },
    {
      "epoch": 1.7983646162423341,
      "grad_norm": 0.7720450758934021,
      "learning_rate": 4.2209540108380855e-06,
      "loss": 0.9178,
      "step": 9770
    },
    {
      "epoch": 1.7985504553057052,
      "grad_norm": 1.274947166442871,
      "learning_rate": 4.2125508729864535e-06,
      "loss": 1.5571,
      "step": 9771
    },
    {
      "epoch": 1.7987362943690763,
      "grad_norm": 1.9692296981811523,
      "learning_rate": 4.204155928042386e-06,
      "loss": 1.3165,
      "step": 9772
    },
    {
      "epoch": 1.7989221334324474,
      "grad_norm": 0.726324200630188,
      "learning_rate": 4.195769176723885e-06,
      "loss": 1.0163,
      "step": 9773
    },
    {
      "epoch": 1.7991079724958188,
      "grad_norm": 1.0148183107376099,
      "learning_rate": 4.18739061974831e-06,
      "loss": 0.9745,
      "step": 9774
    },
    {
      "epoch": 1.7992938115591897,
      "grad_norm": 0.6964415907859802,
      "learning_rate": 4.1790202578323e-06,
      "loss": 1.0187,
      "step": 9775
    },
    {
      "epoch": 1.799479650622561,
      "grad_norm": 0.852936863899231,
      "learning_rate": 4.170658091691792e-06,
      "loss": 0.9487,
      "step": 9776
    },
    {
      "epoch": 1.7996654896859319,
      "grad_norm": 0.7964418530464172,
      "learning_rate": 4.162304122041994e-06,
      "loss": 1.15,
      "step": 9777
    },
    {
      "epoch": 1.7998513287493032,
      "grad_norm": 0.8701530694961548,
      "learning_rate": 4.153958349597464e-06,
      "loss": 1.078,
      "step": 9778
    },
    {
      "epoch": 1.800037167812674,
      "grad_norm": 1.0016974210739136,
      "learning_rate": 4.1456207750720345e-06,
      "loss": 0.9747,
      "step": 9779
    },
    {
      "epoch": 1.8002230068760454,
      "grad_norm": 0.7461830377578735,
      "learning_rate": 4.137291399178822e-06,
      "loss": 1.0308,
      "step": 9780
    },
    {
      "epoch": 1.8004088459394163,
      "grad_norm": 0.6315507292747498,
      "learning_rate": 4.128970222630268e-06,
      "loss": 1.0419,
      "step": 9781
    },
    {
      "epoch": 1.8005946850027876,
      "grad_norm": 0.7336019277572632,
      "learning_rate": 4.120657246138093e-06,
      "loss": 0.7364,
      "step": 9782
    },
    {
      "epoch": 1.8007805240661587,
      "grad_norm": 0.8575476408004761,
      "learning_rate": 4.112352470413328e-06,
      "loss": 1.1506,
      "step": 9783
    },
    {
      "epoch": 1.8009663631295298,
      "grad_norm": 0.6964014768600464,
      "learning_rate": 4.104055896166292e-06,
      "loss": 1.0312,
      "step": 9784
    },
    {
      "epoch": 1.801152202192901,
      "grad_norm": 0.6191890239715576,
      "learning_rate": 4.095767524106609e-06,
      "loss": 0.8681,
      "step": 9785
    },
    {
      "epoch": 1.801338041256272,
      "grad_norm": 0.74542635679245,
      "learning_rate": 4.08748735494322e-06,
      "loss": 0.783,
      "step": 9786
    },
    {
      "epoch": 1.8015238803196432,
      "grad_norm": 0.7950938940048218,
      "learning_rate": 4.079215389384316e-06,
      "loss": 1.082,
      "step": 9787
    },
    {
      "epoch": 1.8017097193830143,
      "grad_norm": 0.6605604887008667,
      "learning_rate": 4.070951628137443e-06,
      "loss": 1.0312,
      "step": 9788
    },
    {
      "epoch": 1.8018955584463854,
      "grad_norm": 0.6296193599700928,
      "learning_rate": 4.062696071909411e-06,
      "loss": 0.5754,
      "step": 9789
    },
    {
      "epoch": 1.8020813975097565,
      "grad_norm": 0.6792443990707397,
      "learning_rate": 4.054448721406312e-06,
      "loss": 0.6941,
      "step": 9790
    },
    {
      "epoch": 1.8022672365731278,
      "grad_norm": 0.7101089954376221,
      "learning_rate": 4.046209577333593e-06,
      "loss": 0.8607,
      "step": 9791
    },
    {
      "epoch": 1.8024530756364987,
      "grad_norm": 0.6400145888328552,
      "learning_rate": 4.037978640395945e-06,
      "loss": 0.7641,
      "step": 9792
    },
    {
      "epoch": 1.80263891469987,
      "grad_norm": 0.6590439677238464,
      "learning_rate": 4.029755911297395e-06,
      "loss": 1.0334,
      "step": 9793
    },
    {
      "epoch": 1.802824753763241,
      "grad_norm": 0.8242725133895874,
      "learning_rate": 4.021541390741235e-06,
      "loss": 0.9977,
      "step": 9794
    },
    {
      "epoch": 1.8030105928266122,
      "grad_norm": 0.7872090935707092,
      "learning_rate": 4.013335079430092e-06,
      "loss": 0.8787,
      "step": 9795
    },
    {
      "epoch": 1.8031964318899831,
      "grad_norm": 0.8022633194923401,
      "learning_rate": 4.005136978065837e-06,
      "loss": 1.0688,
      "step": 9796
    },
    {
      "epoch": 1.8033822709533545,
      "grad_norm": 0.7343721389770508,
      "learning_rate": 3.996947087349701e-06,
      "loss": 1.0336,
      "step": 9797
    },
    {
      "epoch": 1.8035681100167256,
      "grad_norm": 0.684739351272583,
      "learning_rate": 3.9887654079821536e-06,
      "loss": 1.0992,
      "step": 9798
    },
    {
      "epoch": 1.8037539490800967,
      "grad_norm": 0.6524844169616699,
      "learning_rate": 3.980591940663015e-06,
      "loss": 0.9778,
      "step": 9799
    },
    {
      "epoch": 1.8039397881434678,
      "grad_norm": 0.7692164778709412,
      "learning_rate": 3.9724266860913795e-06,
      "loss": 0.7201,
      "step": 9800
    },
    {
      "epoch": 1.8041256272068389,
      "grad_norm": 0.7136658430099487,
      "learning_rate": 3.964269644965623e-06,
      "loss": 0.9806,
      "step": 9801
    },
    {
      "epoch": 1.80431146627021,
      "grad_norm": 0.6302587985992432,
      "learning_rate": 3.956120817983455e-06,
      "loss": 0.9819,
      "step": 9802
    },
    {
      "epoch": 1.804497305333581,
      "grad_norm": 0.8235353827476501,
      "learning_rate": 3.947980205841839e-06,
      "loss": 0.9021,
      "step": 9803
    },
    {
      "epoch": 1.8046831443969522,
      "grad_norm": 0.6519436836242676,
      "learning_rate": 3.9398478092370736e-06,
      "loss": 0.95,
      "step": 9804
    },
    {
      "epoch": 1.8048689834603233,
      "grad_norm": 0.6274929046630859,
      "learning_rate": 3.9317236288647365e-06,
      "loss": 1.0314,
      "step": 9805
    },
    {
      "epoch": 1.8050548225236946,
      "grad_norm": 0.9879053235054016,
      "learning_rate": 3.923607665419715e-06,
      "loss": 0.9081,
      "step": 9806
    },
    {
      "epoch": 1.8052406615870655,
      "grad_norm": 0.7791699767112732,
      "learning_rate": 3.915499919596189e-06,
      "loss": 0.7968,
      "step": 9807
    },
    {
      "epoch": 1.8054265006504369,
      "grad_norm": 0.7316449284553528,
      "learning_rate": 3.907400392087602e-06,
      "loss": 0.9871,
      "step": 9808
    },
    {
      "epoch": 1.8056123397138077,
      "grad_norm": 0.7337790131568909,
      "learning_rate": 3.899309083586755e-06,
      "loss": 1.1018,
      "step": 9809
    },
    {
      "epoch": 1.805798178777179,
      "grad_norm": 0.766534149646759,
      "learning_rate": 3.891225994785697e-06,
      "loss": 1.0126,
      "step": 9810
    },
    {
      "epoch": 1.80598401784055,
      "grad_norm": 0.6639066338539124,
      "learning_rate": 3.883151126375806e-06,
      "loss": 0.9444,
      "step": 9811
    },
    {
      "epoch": 1.8061698569039213,
      "grad_norm": 0.7706999182701111,
      "learning_rate": 3.87508447904773e-06,
      "loss": 1.094,
      "step": 9812
    },
    {
      "epoch": 1.8063556959672922,
      "grad_norm": 0.5971929430961609,
      "learning_rate": 3.867026053491418e-06,
      "loss": 0.8445,
      "step": 9813
    },
    {
      "epoch": 1.8065415350306635,
      "grad_norm": 0.7434088587760925,
      "learning_rate": 3.8589758503961625e-06,
      "loss": 0.7663,
      "step": 9814
    },
    {
      "epoch": 1.8067273740940346,
      "grad_norm": 0.9167700409889221,
      "learning_rate": 3.850933870450479e-06,
      "loss": 1.1226,
      "step": 9815
    },
    {
      "epoch": 1.8069132131574057,
      "grad_norm": 0.6300912499427795,
      "learning_rate": 3.8429001143422296e-06,
      "loss": 0.8259,
      "step": 9816
    },
    {
      "epoch": 1.8070990522207768,
      "grad_norm": 0.6689139008522034,
      "learning_rate": 3.834874582758552e-06,
      "loss": 1.0018,
      "step": 9817
    },
    {
      "epoch": 1.807284891284148,
      "grad_norm": 0.7586479187011719,
      "learning_rate": 3.826857276385886e-06,
      "loss": 1.0716,
      "step": 9818
    },
    {
      "epoch": 1.807470730347519,
      "grad_norm": 0.7082575559616089,
      "learning_rate": 3.8188481959099745e-06,
      "loss": 1.0471,
      "step": 9819
    },
    {
      "epoch": 1.8076565694108901,
      "grad_norm": 0.7369147539138794,
      "learning_rate": 3.810847342015855e-06,
      "loss": 0.769,
      "step": 9820
    },
    {
      "epoch": 1.8078424084742613,
      "grad_norm": 0.686251699924469,
      "learning_rate": 3.8028547153878714e-06,
      "loss": 0.736,
      "step": 9821
    },
    {
      "epoch": 1.8080282475376324,
      "grad_norm": 0.7864113450050354,
      "learning_rate": 3.79487031670962e-06,
      "loss": 0.7086,
      "step": 9822
    },
    {
      "epoch": 1.8082140866010037,
      "grad_norm": 0.6438609957695007,
      "learning_rate": 3.786894146664044e-06,
      "loss": 0.7527,
      "step": 9823
    },
    {
      "epoch": 1.8083999256643746,
      "grad_norm": 0.6810690760612488,
      "learning_rate": 3.778926205933342e-06,
      "loss": 1.1615,
      "step": 9824
    },
    {
      "epoch": 1.808585764727746,
      "grad_norm": 0.653982400894165,
      "learning_rate": 3.7709664951990575e-06,
      "loss": 0.7498,
      "step": 9825
    },
    {
      "epoch": 1.8087716037911168,
      "grad_norm": 0.7026686072349548,
      "learning_rate": 3.7630150151419684e-06,
      "loss": 1.0246,
      "step": 9826
    },
    {
      "epoch": 1.8089574428544881,
      "grad_norm": 0.7423397302627563,
      "learning_rate": 3.7550717664422085e-06,
      "loss": 1.0985,
      "step": 9827
    },
    {
      "epoch": 1.809143281917859,
      "grad_norm": 0.5977106094360352,
      "learning_rate": 3.7471367497791897e-06,
      "loss": 0.8143,
      "step": 9828
    },
    {
      "epoch": 1.8093291209812303,
      "grad_norm": 0.6815504431724548,
      "learning_rate": 3.7392099658315693e-06,
      "loss": 1.1765,
      "step": 9829
    },
    {
      "epoch": 1.8095149600446012,
      "grad_norm": 0.8144717216491699,
      "learning_rate": 3.7312914152773936e-06,
      "loss": 1.1033,
      "step": 9830
    },
    {
      "epoch": 1.8097007991079725,
      "grad_norm": 0.6317072510719299,
      "learning_rate": 3.72338109879391e-06,
      "loss": 0.7315,
      "step": 9831
    },
    {
      "epoch": 1.8098866381713437,
      "grad_norm": 0.69041907787323,
      "learning_rate": 3.7154790170577103e-06,
      "loss": 0.8003,
      "step": 9832
    },
    {
      "epoch": 1.8100724772347148,
      "grad_norm": 0.7883708477020264,
      "learning_rate": 3.7075851707446983e-06,
      "loss": 0.7498,
      "step": 9833
    },
    {
      "epoch": 1.8102583162980859,
      "grad_norm": 0.6976882815361023,
      "learning_rate": 3.6996995605300567e-06,
      "loss": 0.9999,
      "step": 9834
    },
    {
      "epoch": 1.810444155361457,
      "grad_norm": 0.7721940279006958,
      "learning_rate": 3.6918221870882342e-06,
      "loss": 0.801,
      "step": 9835
    },
    {
      "epoch": 1.810629994424828,
      "grad_norm": 0.6680358052253723,
      "learning_rate": 3.6839530510930032e-06,
      "loss": 0.8228,
      "step": 9836
    },
    {
      "epoch": 1.8108158334881992,
      "grad_norm": 0.7693148255348206,
      "learning_rate": 3.6760921532174474e-06,
      "loss": 0.9878,
      "step": 9837
    },
    {
      "epoch": 1.8110016725515703,
      "grad_norm": 0.7048506140708923,
      "learning_rate": 3.668239494133896e-06,
      "loss": 0.9883,
      "step": 9838
    },
    {
      "epoch": 1.8111875116149414,
      "grad_norm": 0.6518003940582275,
      "learning_rate": 3.6603950745140447e-06,
      "loss": 0.9176,
      "step": 9839
    },
    {
      "epoch": 1.8113733506783127,
      "grad_norm": 0.6439449787139893,
      "learning_rate": 3.6525588950288013e-06,
      "loss": 0.7398,
      "step": 9840
    },
    {
      "epoch": 1.8115591897416836,
      "grad_norm": 1.1814019680023193,
      "learning_rate": 3.64473095634843e-06,
      "loss": 1.2524,
      "step": 9841
    },
    {
      "epoch": 1.811745028805055,
      "grad_norm": 0.6845316290855408,
      "learning_rate": 3.636911259142484e-06,
      "loss": 0.9213,
      "step": 9842
    },
    {
      "epoch": 1.8119308678684258,
      "grad_norm": 0.9367323517799377,
      "learning_rate": 3.6290998040797717e-06,
      "loss": 1.0116,
      "step": 9843
    },
    {
      "epoch": 1.8121167069317972,
      "grad_norm": 0.7118209004402161,
      "learning_rate": 3.621296591828449e-06,
      "loss": 1.0242,
      "step": 9844
    },
    {
      "epoch": 1.812302545995168,
      "grad_norm": 0.7774336934089661,
      "learning_rate": 3.613501623055926e-06,
      "loss": 0.734,
      "step": 9845
    },
    {
      "epoch": 1.8124883850585394,
      "grad_norm": 0.6717697978019714,
      "learning_rate": 3.605714898428936e-06,
      "loss": 1.1021,
      "step": 9846
    },
    {
      "epoch": 1.8126742241219103,
      "grad_norm": 0.7385362386703491,
      "learning_rate": 3.597936418613468e-06,
      "loss": 0.8702,
      "step": 9847
    },
    {
      "epoch": 1.8128600631852816,
      "grad_norm": 0.7260059714317322,
      "learning_rate": 3.5901661842748568e-06,
      "loss": 0.9295,
      "step": 9848
    },
    {
      "epoch": 1.8130459022486527,
      "grad_norm": 0.6407851576805115,
      "learning_rate": 3.5824041960777044e-06,
      "loss": 0.8742,
      "step": 9849
    },
    {
      "epoch": 1.8132317413120238,
      "grad_norm": 0.7481780052185059,
      "learning_rate": 3.5746504546859018e-06,
      "loss": 0.9837,
      "step": 9850
    },
    {
      "epoch": 1.813417580375395,
      "grad_norm": 0.694119930267334,
      "learning_rate": 3.5669049607626514e-06,
      "loss": 1.2178,
      "step": 9851
    },
    {
      "epoch": 1.813603419438766,
      "grad_norm": 0.8012346029281616,
      "learning_rate": 3.5591677149704462e-06,
      "loss": 1.0699,
      "step": 9852
    },
    {
      "epoch": 1.8137892585021371,
      "grad_norm": 0.7269610166549683,
      "learning_rate": 3.551438717971056e-06,
      "loss": 0.9231,
      "step": 9853
    },
    {
      "epoch": 1.8139750975655082,
      "grad_norm": 0.7896559834480286,
      "learning_rate": 3.543717970425564e-06,
      "loss": 0.9874,
      "step": 9854
    },
    {
      "epoch": 1.8141609366288793,
      "grad_norm": 0.6833961606025696,
      "learning_rate": 3.5360054729943413e-06,
      "loss": 1.008,
      "step": 9855
    },
    {
      "epoch": 1.8143467756922504,
      "grad_norm": 0.6516629457473755,
      "learning_rate": 3.528301226337072e-06,
      "loss": 0.9614,
      "step": 9856
    },
    {
      "epoch": 1.8145326147556218,
      "grad_norm": 0.8365222215652466,
      "learning_rate": 3.5206052311126836e-06,
      "loss": 1.0754,
      "step": 9857
    },
    {
      "epoch": 1.8147184538189927,
      "grad_norm": 0.7373721599578857,
      "learning_rate": 3.5129174879794613e-06,
      "loss": 1.0566,
      "step": 9858
    },
    {
      "epoch": 1.814904292882364,
      "grad_norm": 0.7686201333999634,
      "learning_rate": 3.5052379975949343e-06,
      "loss": 1.1492,
      "step": 9859
    },
    {
      "epoch": 1.8150901319457349,
      "grad_norm": 0.7509794235229492,
      "learning_rate": 3.497566760615967e-06,
      "loss": 1.1087,
      "step": 9860
    },
    {
      "epoch": 1.8152759710091062,
      "grad_norm": 1.1377304792404175,
      "learning_rate": 3.4899037776986665e-06,
      "loss": 1.2943,
      "step": 9861
    },
    {
      "epoch": 1.815461810072477,
      "grad_norm": 0.789901077747345,
      "learning_rate": 3.482249049498487e-06,
      "loss": 1.0101,
      "step": 9862
    },
    {
      "epoch": 1.8156476491358484,
      "grad_norm": 0.6904792785644531,
      "learning_rate": 3.4746025766701494e-06,
      "loss": 0.7317,
      "step": 9863
    },
    {
      "epoch": 1.8158334881992195,
      "grad_norm": 0.7771925926208496,
      "learning_rate": 3.4669643598676637e-06,
      "loss": 0.9515,
      "step": 9864
    },
    {
      "epoch": 1.8160193272625906,
      "grad_norm": 0.8037922978401184,
      "learning_rate": 3.459334399744374e-06,
      "loss": 0.9012,
      "step": 9865
    },
    {
      "epoch": 1.8162051663259617,
      "grad_norm": 0.8191158175468445,
      "learning_rate": 3.4517126969528467e-06,
      "loss": 1.1604,
      "step": 9866
    },
    {
      "epoch": 1.8163910053893328,
      "grad_norm": 0.6841832399368286,
      "learning_rate": 3.4440992521450057e-06,
      "loss": 1.0259,
      "step": 9867
    },
    {
      "epoch": 1.816576844452704,
      "grad_norm": 0.6980661749839783,
      "learning_rate": 3.43649406597204e-06,
      "loss": 0.7243,
      "step": 9868
    },
    {
      "epoch": 1.816762683516075,
      "grad_norm": 0.7261941432952881,
      "learning_rate": 3.4288971390844303e-06,
      "loss": 1.0282,
      "step": 9869
    },
    {
      "epoch": 1.8169485225794462,
      "grad_norm": 0.7119371891021729,
      "learning_rate": 3.4213084721319897e-06,
      "loss": 1.15,
      "step": 9870
    },
    {
      "epoch": 1.8171343616428173,
      "grad_norm": 0.7503767013549805,
      "learning_rate": 3.413728065763744e-06,
      "loss": 0.8418,
      "step": 9871
    },
    {
      "epoch": 1.8173202007061886,
      "grad_norm": 0.6611210703849792,
      "learning_rate": 3.406155920628107e-06,
      "loss": 0.8486,
      "step": 9872
    },
    {
      "epoch": 1.8175060397695595,
      "grad_norm": 0.689490020275116,
      "learning_rate": 3.3985920373727053e-06,
      "loss": 0.8954,
      "step": 9873
    },
    {
      "epoch": 1.8176918788329308,
      "grad_norm": 0.6549000144004822,
      "learning_rate": 3.391036416644522e-06,
      "loss": 0.9076,
      "step": 9874
    },
    {
      "epoch": 1.8178777178963017,
      "grad_norm": 0.8170062899589539,
      "learning_rate": 3.383489059089784e-06,
      "loss": 1.0304,
      "step": 9875
    },
    {
      "epoch": 1.818063556959673,
      "grad_norm": 0.572019636631012,
      "learning_rate": 3.3759499653540416e-06,
      "loss": 0.7253,
      "step": 9876
    },
    {
      "epoch": 1.818249396023044,
      "grad_norm": 0.8576362729072571,
      "learning_rate": 3.368419136082135e-06,
      "loss": 0.9564,
      "step": 9877
    },
    {
      "epoch": 1.8184352350864152,
      "grad_norm": 0.7417992353439331,
      "learning_rate": 3.360896571918182e-06,
      "loss": 1.0617,
      "step": 9878
    },
    {
      "epoch": 1.8186210741497861,
      "grad_norm": 0.6697123050689697,
      "learning_rate": 3.353382273505623e-06,
      "loss": 0.7978,
      "step": 9879
    },
    {
      "epoch": 1.8188069132131575,
      "grad_norm": 0.6738744378089905,
      "learning_rate": 3.345876241487145e-06,
      "loss": 0.8131,
      "step": 9880
    },
    {
      "epoch": 1.8189927522765286,
      "grad_norm": 0.6736224889755249,
      "learning_rate": 3.338378476504789e-06,
      "loss": 0.966,
      "step": 9881
    },
    {
      "epoch": 1.8191785913398997,
      "grad_norm": 0.656453549861908,
      "learning_rate": 3.330888979199809e-06,
      "loss": 0.8023,
      "step": 9882
    },
    {
      "epoch": 1.8193644304032708,
      "grad_norm": 0.5946424603462219,
      "learning_rate": 3.3234077502128367e-06,
      "loss": 0.7506,
      "step": 9883
    },
    {
      "epoch": 1.819550269466642,
      "grad_norm": 0.813317596912384,
      "learning_rate": 3.31593479018375e-06,
      "loss": 1.0371,
      "step": 9884
    },
    {
      "epoch": 1.819736108530013,
      "grad_norm": 0.6244372129440308,
      "learning_rate": 3.308470099751715e-06,
      "loss": 0.8708,
      "step": 9885
    },
    {
      "epoch": 1.819921947593384,
      "grad_norm": 0.8620980978012085,
      "learning_rate": 3.3010136795552204e-06,
      "loss": 1.0695,
      "step": 9886
    },
    {
      "epoch": 1.8201077866567552,
      "grad_norm": 0.7761400938034058,
      "learning_rate": 3.2935655302320123e-06,
      "loss": 1.0793,
      "step": 9887
    },
    {
      "epoch": 1.8202936257201263,
      "grad_norm": 0.6685424447059631,
      "learning_rate": 3.2861256524191586e-06,
      "loss": 0.9213,
      "step": 9888
    },
    {
      "epoch": 1.8204794647834976,
      "grad_norm": 0.6758712530136108,
      "learning_rate": 3.2786940467529948e-06,
      "loss": 0.9071,
      "step": 9889
    },
    {
      "epoch": 1.8206653038468685,
      "grad_norm": 0.6849507689476013,
      "learning_rate": 3.271270713869179e-06,
      "loss": 0.9324,
      "step": 9890
    },
    {
      "epoch": 1.8208511429102399,
      "grad_norm": 0.8326185345649719,
      "learning_rate": 3.2638556544026368e-06,
      "loss": 1.1884,
      "step": 9891
    },
    {
      "epoch": 1.8210369819736107,
      "grad_norm": 0.6872798800468445,
      "learning_rate": 3.256448868987594e-06,
      "loss": 1.0787,
      "step": 9892
    },
    {
      "epoch": 1.821222821036982,
      "grad_norm": 0.6884906888008118,
      "learning_rate": 3.249050358257566e-06,
      "loss": 0.8761,
      "step": 9893
    },
    {
      "epoch": 1.821408660100353,
      "grad_norm": 0.8486963510513306,
      "learning_rate": 3.241660122845358e-06,
      "loss": 1.1537,
      "step": 9894
    },
    {
      "epoch": 1.8215944991637243,
      "grad_norm": 0.6911030411720276,
      "learning_rate": 3.2342781633831085e-06,
      "loss": 1.03,
      "step": 9895
    },
    {
      "epoch": 1.8217803382270952,
      "grad_norm": 0.7132589817047119,
      "learning_rate": 3.2269044805021577e-06,
      "loss": 0.7436,
      "step": 9896
    },
    {
      "epoch": 1.8219661772904665,
      "grad_norm": 0.6569371223449707,
      "learning_rate": 3.2195390748332334e-06,
      "loss": 0.9463,
      "step": 9897
    },
    {
      "epoch": 1.8221520163538376,
      "grad_norm": 0.6883752942085266,
      "learning_rate": 3.2121819470063097e-06,
      "loss": 0.9598,
      "step": 9898
    },
    {
      "epoch": 1.8223378554172087,
      "grad_norm": 0.7526514530181885,
      "learning_rate": 3.204833097650639e-06,
      "loss": 1.0459,
      "step": 9899
    },
    {
      "epoch": 1.8225236944805798,
      "grad_norm": 0.7786809206008911,
      "learning_rate": 3.1974925273948075e-06,
      "loss": 0.9149,
      "step": 9900
    },
    {
      "epoch": 1.822709533543951,
      "grad_norm": 0.6994304060935974,
      "learning_rate": 3.190160236866646e-06,
      "loss": 1.1353,
      "step": 9901
    },
    {
      "epoch": 1.822895372607322,
      "grad_norm": 0.6753819584846497,
      "learning_rate": 3.182836226693331e-06,
      "loss": 0.6098,
      "step": 9902
    },
    {
      "epoch": 1.8230812116706931,
      "grad_norm": 0.6040153503417969,
      "learning_rate": 3.175520497501261e-06,
      "loss": 1.0412,
      "step": 9903
    },
    {
      "epoch": 1.8232670507340643,
      "grad_norm": 0.7435270547866821,
      "learning_rate": 3.168213049916191e-06,
      "loss": 0.9567,
      "step": 9904
    },
    {
      "epoch": 1.8234528897974354,
      "grad_norm": 0.6494448184967041,
      "learning_rate": 3.1609138845631546e-06,
      "loss": 0.9767,
      "step": 9905
    },
    {
      "epoch": 1.8236387288608067,
      "grad_norm": 0.7443634271621704,
      "learning_rate": 3.1536230020664417e-06,
      "loss": 1.1566,
      "step": 9906
    },
    {
      "epoch": 1.8238245679241776,
      "grad_norm": 0.7014687657356262,
      "learning_rate": 3.146340403049652e-06,
      "loss": 0.9532,
      "step": 9907
    },
    {
      "epoch": 1.824010406987549,
      "grad_norm": 0.6410042643547058,
      "learning_rate": 3.1390660881357116e-06,
      "loss": 0.7681,
      "step": 9908
    },
    {
      "epoch": 1.8241962460509198,
      "grad_norm": 0.6442815065383911,
      "learning_rate": 3.131800057946799e-06,
      "loss": 0.8326,
      "step": 9909
    },
    {
      "epoch": 1.8243820851142911,
      "grad_norm": 0.801694929599762,
      "learning_rate": 3.1245423131043726e-06,
      "loss": 1.0109,
      "step": 9910
    },
    {
      "epoch": 1.824567924177662,
      "grad_norm": 0.6718851327896118,
      "learning_rate": 3.1172928542292146e-06,
      "loss": 0.7381,
      "step": 9911
    },
    {
      "epoch": 1.8247537632410333,
      "grad_norm": 0.6806093454360962,
      "learning_rate": 3.110051681941406e-06,
      "loss": 1.0502,
      "step": 9912
    },
    {
      "epoch": 1.8249396023044044,
      "grad_norm": 0.654414176940918,
      "learning_rate": 3.102818796860274e-06,
      "loss": 0.9833,
      "step": 9913
    },
    {
      "epoch": 1.8251254413677755,
      "grad_norm": 0.8000136017799377,
      "learning_rate": 3.09559419960449e-06,
      "loss": 0.9632,
      "step": 9914
    },
    {
      "epoch": 1.8253112804311467,
      "grad_norm": 0.7904422283172607,
      "learning_rate": 3.088377890791949e-06,
      "loss": 1.1118,
      "step": 9915
    },
    {
      "epoch": 1.8254971194945178,
      "grad_norm": 0.8170812726020813,
      "learning_rate": 3.081169871039913e-06,
      "loss": 0.862,
      "step": 9916
    },
    {
      "epoch": 1.8256829585578889,
      "grad_norm": 0.7572363615036011,
      "learning_rate": 3.073970140964877e-06,
      "loss": 0.9306,
      "step": 9917
    },
    {
      "epoch": 1.82586879762126,
      "grad_norm": 0.7189555764198303,
      "learning_rate": 3.066778701182671e-06,
      "loss": 1.0578,
      "step": 9918
    },
    {
      "epoch": 1.826054636684631,
      "grad_norm": 0.7715607285499573,
      "learning_rate": 3.059595552308381e-06,
      "loss": 0.9815,
      "step": 9919
    },
    {
      "epoch": 1.8262404757480022,
      "grad_norm": 0.5737166404724121,
      "learning_rate": 3.0524206949563926e-06,
      "loss": 0.7259,
      "step": 9920
    },
    {
      "epoch": 1.8264263148113733,
      "grad_norm": 0.7895978093147278,
      "learning_rate": 3.0452541297403934e-06,
      "loss": 1.0323,
      "step": 9921
    },
    {
      "epoch": 1.8266121538747444,
      "grad_norm": 0.8224542140960693,
      "learning_rate": 3.038095857273349e-06,
      "loss": 1.0439,
      "step": 9922
    },
    {
      "epoch": 1.8267979929381157,
      "grad_norm": 0.7269798517227173,
      "learning_rate": 3.030945878167546e-06,
      "loss": 0.983,
      "step": 9923
    },
    {
      "epoch": 1.8269838320014866,
      "grad_norm": 0.6962180137634277,
      "learning_rate": 3.0238041930345073e-06,
      "loss": 0.9832,
      "step": 9924
    },
    {
      "epoch": 1.827169671064858,
      "grad_norm": 0.8408883810043335,
      "learning_rate": 3.0166708024850886e-06,
      "loss": 1.1521,
      "step": 9925
    },
    {
      "epoch": 1.8273555101282288,
      "grad_norm": 0.6679197549819946,
      "learning_rate": 3.009545707129435e-06,
      "loss": 0.7788,
      "step": 9926
    },
    {
      "epoch": 1.8275413491916002,
      "grad_norm": 0.6073801517486572,
      "learning_rate": 3.0024289075769596e-06,
      "loss": 0.9097,
      "step": 9927
    },
    {
      "epoch": 1.827727188254971,
      "grad_norm": 0.6985334157943726,
      "learning_rate": 2.9953204044363747e-06,
      "loss": 0.8337,
      "step": 9928
    },
    {
      "epoch": 1.8279130273183424,
      "grad_norm": 0.6692867279052734,
      "learning_rate": 2.988220198315683e-06,
      "loss": 0.8872,
      "step": 9929
    },
    {
      "epoch": 1.8280988663817135,
      "grad_norm": 0.6330978274345398,
      "learning_rate": 2.9811282898222104e-06,
      "loss": 0.9072,
      "step": 9930
    },
    {
      "epoch": 1.8282847054450846,
      "grad_norm": 0.7436206340789795,
      "learning_rate": 2.974044679562493e-06,
      "loss": 1.1705,
      "step": 9931
    },
    {
      "epoch": 1.8284705445084557,
      "grad_norm": 0.6733962893486023,
      "learning_rate": 2.966969368142447e-06,
      "loss": 0.7876,
      "step": 9932
    },
    {
      "epoch": 1.8286563835718268,
      "grad_norm": 0.7703623175621033,
      "learning_rate": 2.959902356167232e-06,
      "loss": 0.9789,
      "step": 9933
    },
    {
      "epoch": 1.828842222635198,
      "grad_norm": 0.6624138355255127,
      "learning_rate": 2.9528436442412988e-06,
      "loss": 0.9694,
      "step": 9934
    },
    {
      "epoch": 1.829028061698569,
      "grad_norm": 0.666533887386322,
      "learning_rate": 2.945793232968386e-06,
      "loss": 1.1035,
      "step": 9935
    },
    {
      "epoch": 1.8292139007619401,
      "grad_norm": 0.7200380563735962,
      "learning_rate": 2.938751122951544e-06,
      "loss": 0.9698,
      "step": 9936
    },
    {
      "epoch": 1.8293997398253112,
      "grad_norm": 0.7723873853683472,
      "learning_rate": 2.931717314793092e-06,
      "loss": 0.7781,
      "step": 9937
    },
    {
      "epoch": 1.8295855788886826,
      "grad_norm": 0.7704063653945923,
      "learning_rate": 2.9246918090946597e-06,
      "loss": 1.1694,
      "step": 9938
    },
    {
      "epoch": 1.8297714179520534,
      "grad_norm": 0.6921645998954773,
      "learning_rate": 2.9176746064571327e-06,
      "loss": 1.096,
      "step": 9939
    },
    {
      "epoch": 1.8299572570154248,
      "grad_norm": 0.6649049520492554,
      "learning_rate": 2.9106657074807307e-06,
      "loss": 0.5551,
      "step": 9940
    },
    {
      "epoch": 1.8301430960787957,
      "grad_norm": 0.7728575468063354,
      "learning_rate": 2.9036651127649194e-06,
      "loss": 0.9851,
      "step": 9941
    },
    {
      "epoch": 1.830328935142167,
      "grad_norm": 0.6400600671768188,
      "learning_rate": 2.896672822908497e-06,
      "loss": 0.8924,
      "step": 9942
    },
    {
      "epoch": 1.8305147742055379,
      "grad_norm": 0.8135813474655151,
      "learning_rate": 2.8896888385094967e-06,
      "loss": 1.0684,
      "step": 9943
    },
    {
      "epoch": 1.8307006132689092,
      "grad_norm": 0.6735711693763733,
      "learning_rate": 2.8827131601653178e-06,
      "loss": 1.0137,
      "step": 9944
    },
    {
      "epoch": 1.83088645233228,
      "grad_norm": 0.6968306303024292,
      "learning_rate": 2.8757457884725613e-06,
      "loss": 1.0254,
      "step": 9945
    },
    {
      "epoch": 1.8310722913956514,
      "grad_norm": 0.7960559725761414,
      "learning_rate": 2.8687867240271725e-06,
      "loss": 0.8744,
      "step": 9946
    },
    {
      "epoch": 1.8312581304590225,
      "grad_norm": 0.8474821448326111,
      "learning_rate": 2.861835967424409e-06,
      "loss": 1.1025,
      "step": 9947
    },
    {
      "epoch": 1.8314439695223936,
      "grad_norm": 0.6721470355987549,
      "learning_rate": 2.8548935192587388e-06,
      "loss": 0.8398,
      "step": 9948
    },
    {
      "epoch": 1.8316298085857647,
      "grad_norm": 0.9784024357795715,
      "learning_rate": 2.847959380123999e-06,
      "loss": 0.7279,
      "step": 9949
    },
    {
      "epoch": 1.8318156476491358,
      "grad_norm": 0.6895228624343872,
      "learning_rate": 2.841033550613259e-06,
      "loss": 1.0313,
      "step": 9950
    },
    {
      "epoch": 1.832001486712507,
      "grad_norm": 0.6543493270874023,
      "learning_rate": 2.8341160313189006e-06,
      "loss": 0.5196,
      "step": 9951
    },
    {
      "epoch": 1.832187325775878,
      "grad_norm": 0.6853628158569336,
      "learning_rate": 2.8272068228325955e-06,
      "loss": 0.9682,
      "step": 9952
    },
    {
      "epoch": 1.8323731648392492,
      "grad_norm": 0.8264276385307312,
      "learning_rate": 2.8203059257453256e-06,
      "loss": 1.2192,
      "step": 9953
    },
    {
      "epoch": 1.8325590039026203,
      "grad_norm": 0.7728658318519592,
      "learning_rate": 2.8134133406473197e-06,
      "loss": 1.0588,
      "step": 9954
    },
    {
      "epoch": 1.8327448429659916,
      "grad_norm": 0.8136929273605347,
      "learning_rate": 2.8065290681281164e-06,
      "loss": 0.8082,
      "step": 9955
    },
    {
      "epoch": 1.8329306820293625,
      "grad_norm": 0.7946676015853882,
      "learning_rate": 2.799653108776545e-06,
      "loss": 0.8515,
      "step": 9956
    },
    {
      "epoch": 1.8331165210927338,
      "grad_norm": 0.5548973083496094,
      "learning_rate": 2.7927854631807247e-06,
      "loss": 0.6857,
      "step": 9957
    },
    {
      "epoch": 1.8333023601561047,
      "grad_norm": 0.7363007664680481,
      "learning_rate": 2.7859261319280515e-06,
      "loss": 0.8401,
      "step": 9958
    },
    {
      "epoch": 1.833488199219476,
      "grad_norm": 0.6322947144508362,
      "learning_rate": 2.779075115605223e-06,
      "loss": 0.8207,
      "step": 9959
    },
    {
      "epoch": 1.833674038282847,
      "grad_norm": 0.7248316407203674,
      "learning_rate": 2.772232414798226e-06,
      "loss": 1.0542,
      "step": 9960
    },
    {
      "epoch": 1.8338598773462182,
      "grad_norm": 0.8802450299263,
      "learning_rate": 2.7653980300923254e-06,
      "loss": 1.0695,
      "step": 9961
    },
    {
      "epoch": 1.8340457164095891,
      "grad_norm": 0.7538386583328247,
      "learning_rate": 2.7585719620720765e-06,
      "loss": 0.9221,
      "step": 9962
    },
    {
      "epoch": 1.8342315554729605,
      "grad_norm": 0.7812655568122864,
      "learning_rate": 2.7517542113213445e-06,
      "loss": 0.9761,
      "step": 9963
    },
    {
      "epoch": 1.8344173945363316,
      "grad_norm": 0.6933215260505676,
      "learning_rate": 2.744944778423253e-06,
      "loss": 0.8904,
      "step": 9964
    },
    {
      "epoch": 1.8346032335997027,
      "grad_norm": 0.6729377508163452,
      "learning_rate": 2.7381436639602243e-06,
      "loss": 0.9443,
      "step": 9965
    },
    {
      "epoch": 1.8347890726630738,
      "grad_norm": 0.7871366739273071,
      "learning_rate": 2.731350868513982e-06,
      "loss": 1.0416,
      "step": 9966
    },
    {
      "epoch": 1.834974911726445,
      "grad_norm": 0.679682731628418,
      "learning_rate": 2.7245663926655174e-06,
      "loss": 1.0339,
      "step": 9967
    },
    {
      "epoch": 1.835160750789816,
      "grad_norm": 0.7425009608268738,
      "learning_rate": 2.7177902369951436e-06,
      "loss": 0.9609,
      "step": 9968
    },
    {
      "epoch": 1.835346589853187,
      "grad_norm": 0.8477035760879517,
      "learning_rate": 2.7110224020824194e-06,
      "loss": 1.0835,
      "step": 9969
    },
    {
      "epoch": 1.8355324289165582,
      "grad_norm": 0.8120266795158386,
      "learning_rate": 2.704262888506226e-06,
      "loss": 0.875,
      "step": 9970
    },
    {
      "epoch": 1.8357182679799293,
      "grad_norm": 0.7462162375450134,
      "learning_rate": 2.697511696844701e-06,
      "loss": 0.9639,
      "step": 9971
    },
    {
      "epoch": 1.8359041070433006,
      "grad_norm": 0.777911365032196,
      "learning_rate": 2.6907688276753053e-06,
      "loss": 0.9999,
      "step": 9972
    },
    {
      "epoch": 1.8360899461066715,
      "grad_norm": 0.6171411871910095,
      "learning_rate": 2.6840342815747544e-06,
      "loss": 0.7703,
      "step": 9973
    },
    {
      "epoch": 1.8362757851700429,
      "grad_norm": 1.0107492208480835,
      "learning_rate": 2.677308059119088e-06,
      "loss": 1.3989,
      "step": 9974
    },
    {
      "epoch": 1.8364616242334137,
      "grad_norm": 0.7087421417236328,
      "learning_rate": 2.6705901608836017e-06,
      "loss": 0.9501,
      "step": 9975
    },
    {
      "epoch": 1.836647463296785,
      "grad_norm": 0.7753424048423767,
      "learning_rate": 2.6638805874428907e-06,
      "loss": 0.803,
      "step": 9976
    },
    {
      "epoch": 1.836833302360156,
      "grad_norm": 1.1736202239990234,
      "learning_rate": 2.657179339370863e-06,
      "loss": 0.6303,
      "step": 9977
    },
    {
      "epoch": 1.8370191414235273,
      "grad_norm": 0.7461515665054321,
      "learning_rate": 2.6504864172406497e-06,
      "loss": 0.9204,
      "step": 9978
    },
    {
      "epoch": 1.8372049804868984,
      "grad_norm": 0.664656937122345,
      "learning_rate": 2.6438018216247475e-06,
      "loss": 0.8246,
      "step": 9979
    },
    {
      "epoch": 1.8373908195502695,
      "grad_norm": 0.7617626786231995,
      "learning_rate": 2.637125553094877e-06,
      "loss": 1.0333,
      "step": 9980
    },
    {
      "epoch": 1.8375766586136406,
      "grad_norm": 0.7563158869743347,
      "learning_rate": 2.6304576122221035e-06,
      "loss": 0.9878,
      "step": 9981
    },
    {
      "epoch": 1.8377624976770117,
      "grad_norm": 0.6394928693771362,
      "learning_rate": 2.6237979995767157e-06,
      "loss": 1.0379,
      "step": 9982
    },
    {
      "epoch": 1.8379483367403828,
      "grad_norm": 0.7006260752677917,
      "learning_rate": 2.6171467157283467e-06,
      "loss": 1.0053,
      "step": 9983
    },
    {
      "epoch": 1.838134175803754,
      "grad_norm": 0.8882328271865845,
      "learning_rate": 2.6105037612459083e-06,
      "loss": 1.1598,
      "step": 9984
    },
    {
      "epoch": 1.838320014867125,
      "grad_norm": 1.0692800283432007,
      "learning_rate": 2.6038691366975566e-06,
      "loss": 0.8886,
      "step": 9985
    },
    {
      "epoch": 1.8385058539304961,
      "grad_norm": 0.5800960063934326,
      "learning_rate": 2.5972428426507823e-06,
      "loss": 0.9888,
      "step": 9986
    },
    {
      "epoch": 1.8386916929938675,
      "grad_norm": 0.7504416108131409,
      "learning_rate": 2.590624879672332e-06,
      "loss": 0.9861,
      "step": 9987
    },
    {
      "epoch": 1.8388775320572384,
      "grad_norm": 1.5852302312850952,
      "learning_rate": 2.5840152483282752e-06,
      "loss": 1.3195,
      "step": 9988
    },
    {
      "epoch": 1.8390633711206097,
      "grad_norm": 0.7464103102684021,
      "learning_rate": 2.5774139491839376e-06,
      "loss": 0.9331,
      "step": 9989
    },
    {
      "epoch": 1.8392492101839806,
      "grad_norm": 0.6646602153778076,
      "learning_rate": 2.5708209828039343e-06,
      "loss": 0.9741,
      "step": 9990
    },
    {
      "epoch": 1.839435049247352,
      "grad_norm": 0.6763034462928772,
      "learning_rate": 2.5642363497522025e-06,
      "loss": 0.9705,
      "step": 9991
    },
    {
      "epoch": 1.8396208883107228,
      "grad_norm": 0.6655982136726379,
      "learning_rate": 2.5576600505919035e-06,
      "loss": 0.8737,
      "step": 9992
    },
    {
      "epoch": 1.8398067273740941,
      "grad_norm": 0.6391465663909912,
      "learning_rate": 2.5510920858855535e-06,
      "loss": 0.7289,
      "step": 9993
    },
    {
      "epoch": 1.839992566437465,
      "grad_norm": 0.7988792061805725,
      "learning_rate": 2.5445324561949035e-06,
      "loss": 0.8365,
      "step": 9994
    },
    {
      "epoch": 1.8401784055008363,
      "grad_norm": 0.7054262161254883,
      "learning_rate": 2.537981162081027e-06,
      "loss": 0.8371,
      "step": 9995
    },
    {
      "epoch": 1.8403642445642074,
      "grad_norm": 0.6635208129882812,
      "learning_rate": 2.531438204104253e-06,
      "loss": 0.8408,
      "step": 9996
    },
    {
      "epoch": 1.8405500836275785,
      "grad_norm": 0.7707086801528931,
      "learning_rate": 2.524903582824234e-06,
      "loss": 0.9308,
      "step": 9997
    },
    {
      "epoch": 1.8407359226909497,
      "grad_norm": 0.6752921938896179,
      "learning_rate": 2.51837729879989e-06,
      "loss": 1.0843,
      "step": 9998
    },
    {
      "epoch": 1.8409217617543208,
      "grad_norm": 0.6770755052566528,
      "learning_rate": 2.5118593525894184e-06,
      "loss": 0.7495,
      "step": 9999
    },
    {
      "epoch": 1.8411076008176919,
      "grad_norm": 0.5796447992324829,
      "learning_rate": 2.5053497447503183e-06,
      "loss": 0.6934,
      "step": 10000
    },
    {
      "epoch": 1.841293439881063,
      "grad_norm": 0.6556915640830994,
      "learning_rate": 2.498848475839355e-06,
      "loss": 0.9931,
      "step": 10001
    },
    {
      "epoch": 1.841479278944434,
      "grad_norm": 0.734487771987915,
      "learning_rate": 2.492355546412617e-06,
      "loss": 0.8293,
      "step": 10002
    },
    {
      "epoch": 1.8416651180078052,
      "grad_norm": 1.2783101797103882,
      "learning_rate": 2.4858709570254603e-06,
      "loss": 1.2197,
      "step": 10003
    },
    {
      "epoch": 1.8418509570711765,
      "grad_norm": 0.8434581756591797,
      "learning_rate": 2.4793947082324964e-06,
      "loss": 0.8842,
      "step": 10004
    },
    {
      "epoch": 1.8420367961345474,
      "grad_norm": 0.8138322234153748,
      "learning_rate": 2.4729268005876938e-06,
      "loss": 1.1372,
      "step": 10005
    },
    {
      "epoch": 1.8422226351979187,
      "grad_norm": 0.69370436668396,
      "learning_rate": 2.466467234644232e-06,
      "loss": 0.9559,
      "step": 10006
    },
    {
      "epoch": 1.8424084742612896,
      "grad_norm": 0.7971329689025879,
      "learning_rate": 2.4600160109546354e-06,
      "loss": 0.9585,
      "step": 10007
    },
    {
      "epoch": 1.842594313324661,
      "grad_norm": 0.8003334999084473,
      "learning_rate": 2.453573130070674e-06,
      "loss": 0.7161,
      "step": 10008
    },
    {
      "epoch": 1.8427801523880318,
      "grad_norm": 0.6565207839012146,
      "learning_rate": 2.4471385925434297e-06,
      "loss": 0.916,
      "step": 10009
    },
    {
      "epoch": 1.8429659914514032,
      "grad_norm": 0.6413012742996216,
      "learning_rate": 2.4407123989232507e-06,
      "loss": 0.9176,
      "step": 10010
    },
    {
      "epoch": 1.843151830514774,
      "grad_norm": 0.587217390537262,
      "learning_rate": 2.4342945497598078e-06,
      "loss": 0.7268,
      "step": 10011
    },
    {
      "epoch": 1.8433376695781454,
      "grad_norm": 0.9451626539230347,
      "learning_rate": 2.4278850456020184e-06,
      "loss": 1.0721,
      "step": 10012
    },
    {
      "epoch": 1.8435235086415165,
      "grad_norm": 0.8284534215927124,
      "learning_rate": 2.4214838869980994e-06,
      "loss": 0.8944,
      "step": 10013
    },
    {
      "epoch": 1.8437093477048876,
      "grad_norm": 0.8125157952308655,
      "learning_rate": 2.415091074495557e-06,
      "loss": 1.2481,
      "step": 10014
    },
    {
      "epoch": 1.8438951867682587,
      "grad_norm": 1.239067792892456,
      "learning_rate": 2.4087066086411867e-06,
      "loss": 1.1899,
      "step": 10015
    },
    {
      "epoch": 1.8440810258316298,
      "grad_norm": 0.6294758915901184,
      "learning_rate": 2.402330489981053e-06,
      "loss": 0.7415,
      "step": 10016
    },
    {
      "epoch": 1.844266864895001,
      "grad_norm": 0.6115023493766785,
      "learning_rate": 2.3959627190605404e-06,
      "loss": 0.8013,
      "step": 10017
    },
    {
      "epoch": 1.844452703958372,
      "grad_norm": 0.6057900786399841,
      "learning_rate": 2.389603296424281e-06,
      "loss": 0.7154,
      "step": 10018
    },
    {
      "epoch": 1.8446385430217431,
      "grad_norm": 0.6940528154373169,
      "learning_rate": 2.383252222616228e-06,
      "loss": 0.759,
      "step": 10019
    },
    {
      "epoch": 1.8448243820851142,
      "grad_norm": 1.202257513999939,
      "learning_rate": 2.3769094981795693e-06,
      "loss": 1.3982,
      "step": 10020
    },
    {
      "epoch": 1.8450102211484856,
      "grad_norm": 0.6999146342277527,
      "learning_rate": 2.3705751236568485e-06,
      "loss": 0.8744,
      "step": 10021
    },
    {
      "epoch": 1.8451960602118564,
      "grad_norm": 0.7464430332183838,
      "learning_rate": 2.3642490995898327e-06,
      "loss": 1.4102,
      "step": 10022
    },
    {
      "epoch": 1.8453818992752278,
      "grad_norm": 0.7458065748214722,
      "learning_rate": 2.3579314265196104e-06,
      "loss": 0.9248,
      "step": 10023
    },
    {
      "epoch": 1.8455677383385987,
      "grad_norm": 0.5382136702537537,
      "learning_rate": 2.3516221049865505e-06,
      "loss": 0.5963,
      "step": 10024
    },
    {
      "epoch": 1.84575357740197,
      "grad_norm": 0.7321708798408508,
      "learning_rate": 2.3453211355302874e-06,
      "loss": 0.8908,
      "step": 10025
    },
    {
      "epoch": 1.8459394164653409,
      "grad_norm": 0.6686316132545471,
      "learning_rate": 2.3390285186897896e-06,
      "loss": 0.8337,
      "step": 10026
    },
    {
      "epoch": 1.8461252555287122,
      "grad_norm": 0.7071788311004639,
      "learning_rate": 2.3327442550032385e-06,
      "loss": 0.938,
      "step": 10027
    },
    {
      "epoch": 1.846311094592083,
      "grad_norm": 0.6526303291320801,
      "learning_rate": 2.3264683450081594e-06,
      "loss": 0.7184,
      "step": 10028
    },
    {
      "epoch": 1.8464969336554544,
      "grad_norm": 0.736538827419281,
      "learning_rate": 2.3202007892413447e-06,
      "loss": 0.8133,
      "step": 10029
    },
    {
      "epoch": 1.8466827727188255,
      "grad_norm": 0.7265169024467468,
      "learning_rate": 2.313941588238866e-06,
      "loss": 0.9624,
      "step": 10030
    },
    {
      "epoch": 1.8468686117821966,
      "grad_norm": 0.8249513506889343,
      "learning_rate": 2.3076907425361062e-06,
      "loss": 1.13,
      "step": 10031
    },
    {
      "epoch": 1.8470544508455677,
      "grad_norm": 0.6018231511116028,
      "learning_rate": 2.3014482526676816e-06,
      "loss": 0.678,
      "step": 10032
    },
    {
      "epoch": 1.8472402899089388,
      "grad_norm": 0.6337421536445618,
      "learning_rate": 2.2952141191675547e-06,
      "loss": 1.1592,
      "step": 10033
    },
    {
      "epoch": 1.84742612897231,
      "grad_norm": 0.7989595532417297,
      "learning_rate": 2.2889883425689206e-06,
      "loss": 0.9037,
      "step": 10034
    },
    {
      "epoch": 1.847611968035681,
      "grad_norm": 0.6029747128486633,
      "learning_rate": 2.2827709234043092e-06,
      "loss": 0.7932,
      "step": 10035
    },
    {
      "epoch": 1.8477978070990522,
      "grad_norm": 0.6751794815063477,
      "learning_rate": 2.276561862205473e-06,
      "loss": 0.8837,
      "step": 10036
    },
    {
      "epoch": 1.8479836461624233,
      "grad_norm": 0.7136726379394531,
      "learning_rate": 2.27036115950352e-06,
      "loss": 0.8024,
      "step": 10037
    },
    {
      "epoch": 1.8481694852257946,
      "grad_norm": 0.6392678618431091,
      "learning_rate": 2.2641688158288043e-06,
      "loss": 0.7938,
      "step": 10038
    },
    {
      "epoch": 1.8483553242891655,
      "grad_norm": 0.6552032232284546,
      "learning_rate": 2.2579848317109575e-06,
      "loss": 0.8467,
      "step": 10039
    },
    {
      "epoch": 1.8485411633525368,
      "grad_norm": 0.6480730772018433,
      "learning_rate": 2.2518092076789234e-06,
      "loss": 0.9873,
      "step": 10040
    },
    {
      "epoch": 1.8487270024159077,
      "grad_norm": 0.6986692547798157,
      "learning_rate": 2.2456419442609012e-06,
      "loss": 0.9678,
      "step": 10041
    },
    {
      "epoch": 1.848912841479279,
      "grad_norm": 0.7117049098014832,
      "learning_rate": 2.2394830419844027e-06,
      "loss": 0.9666,
      "step": 10042
    },
    {
      "epoch": 1.84909868054265,
      "grad_norm": 0.6958870887756348,
      "learning_rate": 2.233332501376206e-06,
      "loss": 0.9391,
      "step": 10043
    },
    {
      "epoch": 1.8492845196060212,
      "grad_norm": 0.6736997961997986,
      "learning_rate": 2.22719032296238e-06,
      "loss": 0.7636,
      "step": 10044
    },
    {
      "epoch": 1.8494703586693924,
      "grad_norm": 0.5893682241439819,
      "learning_rate": 2.221056507268293e-06,
      "loss": 0.6087,
      "step": 10045
    },
    {
      "epoch": 1.8496561977327635,
      "grad_norm": 0.7475652694702148,
      "learning_rate": 2.2149310548185586e-06,
      "loss": 1.0171,
      "step": 10046
    },
    {
      "epoch": 1.8498420367961346,
      "grad_norm": 0.580328643321991,
      "learning_rate": 2.2088139661371244e-06,
      "loss": 0.9603,
      "step": 10047
    },
    {
      "epoch": 1.8500278758595057,
      "grad_norm": 0.8390524983406067,
      "learning_rate": 2.2027052417471715e-06,
      "loss": 1.0896,
      "step": 10048
    },
    {
      "epoch": 1.8502137149228768,
      "grad_norm": 0.7339285612106323,
      "learning_rate": 2.196604882171216e-06,
      "loss": 1.2479,
      "step": 10049
    },
    {
      "epoch": 1.850399553986248,
      "grad_norm": 0.6563944816589355,
      "learning_rate": 2.190512887931018e-06,
      "loss": 0.908,
      "step": 10050
    },
    {
      "epoch": 1.850585393049619,
      "grad_norm": 0.6774222254753113,
      "learning_rate": 2.184429259547649e-06,
      "loss": 0.7921,
      "step": 10051
    },
    {
      "epoch": 1.85077123211299,
      "grad_norm": 0.7838877439498901,
      "learning_rate": 2.1783539975414714e-06,
      "loss": 0.8798,
      "step": 10052
    },
    {
      "epoch": 1.8509570711763614,
      "grad_norm": 0.6545433402061462,
      "learning_rate": 2.172287102432069e-06,
      "loss": 0.9829,
      "step": 10053
    },
    {
      "epoch": 1.8511429102397323,
      "grad_norm": 0.7077997326850891,
      "learning_rate": 2.1662285747383936e-06,
      "loss": 1.0176,
      "step": 10054
    },
    {
      "epoch": 1.8513287493031036,
      "grad_norm": 0.6887453198432922,
      "learning_rate": 2.160178414978642e-06,
      "loss": 0.826,
      "step": 10055
    },
    {
      "epoch": 1.8515145883664745,
      "grad_norm": 0.7139452695846558,
      "learning_rate": 2.1541366236702887e-06,
      "loss": 0.8449,
      "step": 10056
    },
    {
      "epoch": 1.8517004274298459,
      "grad_norm": 0.8103654980659485,
      "learning_rate": 2.1481032013300984e-06,
      "loss": 1.1509,
      "step": 10057
    },
    {
      "epoch": 1.8518862664932167,
      "grad_norm": 0.634610116481781,
      "learning_rate": 2.1420781484741247e-06,
      "loss": 0.9645,
      "step": 10058
    },
    {
      "epoch": 1.852072105556588,
      "grad_norm": 0.7717024683952332,
      "learning_rate": 2.1360614656177115e-06,
      "loss": 0.841,
      "step": 10059
    },
    {
      "epoch": 1.852257944619959,
      "grad_norm": 0.6939625144004822,
      "learning_rate": 2.1300531532754686e-06,
      "loss": 0.8912,
      "step": 10060
    },
    {
      "epoch": 1.8524437836833303,
      "grad_norm": 0.5884217023849487,
      "learning_rate": 2.1240532119613078e-06,
      "loss": 0.7692,
      "step": 10061
    },
    {
      "epoch": 1.8526296227467014,
      "grad_norm": 0.5974510312080383,
      "learning_rate": 2.118061642188418e-06,
      "loss": 0.7782,
      "step": 10062
    },
    {
      "epoch": 1.8528154618100725,
      "grad_norm": 0.5739233493804932,
      "learning_rate": 2.1120784444692675e-06,
      "loss": 0.7937,
      "step": 10063
    },
    {
      "epoch": 1.8530013008734436,
      "grad_norm": 0.7020597457885742,
      "learning_rate": 2.1061036193156025e-06,
      "loss": 0.9993,
      "step": 10064
    },
    {
      "epoch": 1.8531871399368147,
      "grad_norm": 0.9987313151359558,
      "learning_rate": 2.1001371672384697e-06,
      "loss": 1.2808,
      "step": 10065
    },
    {
      "epoch": 1.8533729790001858,
      "grad_norm": 0.7735684514045715,
      "learning_rate": 2.0941790887482047e-06,
      "loss": 1.062,
      "step": 10066
    },
    {
      "epoch": 1.853558818063557,
      "grad_norm": 0.6970610618591309,
      "learning_rate": 2.0882293843543897e-06,
      "loss": 0.7506,
      "step": 10067
    },
    {
      "epoch": 1.853744657126928,
      "grad_norm": 0.6600317358970642,
      "learning_rate": 2.0822880545659396e-06,
      "loss": 0.8805,
      "step": 10068
    },
    {
      "epoch": 1.8539304961902991,
      "grad_norm": 0.7148259878158569,
      "learning_rate": 2.0763550998910143e-06,
      "loss": 1.0151,
      "step": 10069
    },
    {
      "epoch": 1.8541163352536705,
      "grad_norm": 0.720138430595398,
      "learning_rate": 2.0704305208370857e-06,
      "loss": 1.0692,
      "step": 10070
    },
    {
      "epoch": 1.8543021743170414,
      "grad_norm": 0.9224472045898438,
      "learning_rate": 2.064514317910893e-06,
      "loss": 1.1559,
      "step": 10071
    },
    {
      "epoch": 1.8544880133804127,
      "grad_norm": 0.9005986452102661,
      "learning_rate": 2.058606491618442e-06,
      "loss": 1.0002,
      "step": 10072
    },
    {
      "epoch": 1.8546738524437836,
      "grad_norm": 0.6716923713684082,
      "learning_rate": 2.0527070424650742e-06,
      "loss": 0.8971,
      "step": 10073
    },
    {
      "epoch": 1.854859691507155,
      "grad_norm": 0.8251639604568481,
      "learning_rate": 2.0468159709553513e-06,
      "loss": 0.8688,
      "step": 10074
    },
    {
      "epoch": 1.8550455305705258,
      "grad_norm": 0.7305351495742798,
      "learning_rate": 2.040933277593171e-06,
      "loss": 0.9406,
      "step": 10075
    },
    {
      "epoch": 1.8552313696338971,
      "grad_norm": 0.8185998797416687,
      "learning_rate": 2.035058962881675e-06,
      "loss": 0.9181,
      "step": 10076
    },
    {
      "epoch": 1.855417208697268,
      "grad_norm": 0.5533907413482666,
      "learning_rate": 2.0291930273233394e-06,
      "loss": 0.7119,
      "step": 10077
    },
    {
      "epoch": 1.8556030477606393,
      "grad_norm": 0.7022377848625183,
      "learning_rate": 2.0233354714198406e-06,
      "loss": 1.1283,
      "step": 10078
    },
    {
      "epoch": 1.8557888868240104,
      "grad_norm": 0.7600376605987549,
      "learning_rate": 2.017486295672222e-06,
      "loss": 1.0736,
      "step": 10079
    },
    {
      "epoch": 1.8559747258873815,
      "grad_norm": 0.7274320721626282,
      "learning_rate": 2.011645500580772e-06,
      "loss": 1.1651,
      "step": 10080
    },
    {
      "epoch": 1.8561605649507527,
      "grad_norm": 0.8164176940917969,
      "learning_rate": 2.0058130866450695e-06,
      "loss": 1.2344,
      "step": 10081
    },
    {
      "epoch": 1.8563464040141238,
      "grad_norm": 0.8189756870269775,
      "learning_rate": 1.9999890543639487e-06,
      "loss": 1.0574,
      "step": 10082
    },
    {
      "epoch": 1.8565322430774949,
      "grad_norm": 0.5922825932502747,
      "learning_rate": 1.9941734042355886e-06,
      "loss": 0.6764,
      "step": 10083
    },
    {
      "epoch": 1.856718082140866,
      "grad_norm": 0.7183148860931396,
      "learning_rate": 1.9883661367573803e-06,
      "loss": 1.0216,
      "step": 10084
    },
    {
      "epoch": 1.856903921204237,
      "grad_norm": 0.6784137487411499,
      "learning_rate": 1.9825672524260598e-06,
      "loss": 0.8048,
      "step": 10085
    },
    {
      "epoch": 1.8570897602676082,
      "grad_norm": 0.6123964786529541,
      "learning_rate": 1.9767767517375966e-06,
      "loss": 1.0757,
      "step": 10086
    },
    {
      "epoch": 1.8572755993309795,
      "grad_norm": 0.9567644000053406,
      "learning_rate": 1.9709946351872845e-06,
      "loss": 0.8913,
      "step": 10087
    },
    {
      "epoch": 1.8574614383943504,
      "grad_norm": 0.6350728869438171,
      "learning_rate": 1.96522090326966e-06,
      "loss": 0.8999,
      "step": 10088
    },
    {
      "epoch": 1.8576472774577217,
      "grad_norm": 0.6837267279624939,
      "learning_rate": 1.9594555564785844e-06,
      "loss": 0.8036,
      "step": 10089
    },
    {
      "epoch": 1.8578331165210926,
      "grad_norm": 0.7758045196533203,
      "learning_rate": 1.9536985953071516e-06,
      "loss": 1.1897,
      "step": 10090
    },
    {
      "epoch": 1.858018955584464,
      "grad_norm": 0.7234291434288025,
      "learning_rate": 1.947950020247802e-06,
      "loss": 0.9989,
      "step": 10091
    },
    {
      "epoch": 1.8582047946478348,
      "grad_norm": 0.7876737117767334,
      "learning_rate": 1.9422098317921854e-06,
      "loss": 0.9718,
      "step": 10092
    },
    {
      "epoch": 1.8583906337112062,
      "grad_norm": 0.8611382842063904,
      "learning_rate": 1.936478030431299e-06,
      "loss": 0.927,
      "step": 10093
    },
    {
      "epoch": 1.8585764727745773,
      "grad_norm": 0.6509695649147034,
      "learning_rate": 1.9307546166554056e-06,
      "loss": 1.0372,
      "step": 10094
    },
    {
      "epoch": 1.8587623118379484,
      "grad_norm": 0.7223415970802307,
      "learning_rate": 1.925039590954014e-06,
      "loss": 0.988,
      "step": 10095
    },
    {
      "epoch": 1.8589481509013195,
      "grad_norm": 0.6797177195549011,
      "learning_rate": 1.9193329538159553e-06,
      "loss": 0.8761,
      "step": 10096
    },
    {
      "epoch": 1.8591339899646906,
      "grad_norm": 0.5940880179405212,
      "learning_rate": 1.913634705729328e-06,
      "loss": 0.7282,
      "step": 10097
    },
    {
      "epoch": 1.8593198290280617,
      "grad_norm": 0.7168911695480347,
      "learning_rate": 1.9079448471815087e-06,
      "loss": 1.055,
      "step": 10098
    },
    {
      "epoch": 1.8595056680914328,
      "grad_norm": 0.5971741080284119,
      "learning_rate": 1.9022633786591747e-06,
      "loss": 0.9821,
      "step": 10099
    },
    {
      "epoch": 1.859691507154804,
      "grad_norm": 0.7200136780738831,
      "learning_rate": 1.896590300648271e-06,
      "loss": 1.1599,
      "step": 10100
    },
    {
      "epoch": 1.859877346218175,
      "grad_norm": 0.676017701625824,
      "learning_rate": 1.8909256136340425e-06,
      "loss": 1.0941,
      "step": 10101
    },
    {
      "epoch": 1.8600631852815461,
      "grad_norm": 0.7950423359870911,
      "learning_rate": 1.8852693181009795e-06,
      "loss": 1.0326,
      "step": 10102
    },
    {
      "epoch": 1.8602490243449172,
      "grad_norm": 0.6871570348739624,
      "learning_rate": 1.8796214145328838e-06,
      "loss": 0.9171,
      "step": 10103
    },
    {
      "epoch": 1.8604348634082886,
      "grad_norm": 0.7822986841201782,
      "learning_rate": 1.8739819034128359e-06,
      "loss": 1.051,
      "step": 10104
    },
    {
      "epoch": 1.8606207024716594,
      "grad_norm": 0.6993398666381836,
      "learning_rate": 1.868350785223194e-06,
      "loss": 0.9718,
      "step": 10105
    },
    {
      "epoch": 1.8608065415350308,
      "grad_norm": 1.4518990516662598,
      "learning_rate": 1.8627280604456066e-06,
      "loss": 0.7944,
      "step": 10106
    },
    {
      "epoch": 1.8609923805984017,
      "grad_norm": 0.7145583629608154,
      "learning_rate": 1.8571137295609775e-06,
      "loss": 0.8091,
      "step": 10107
    },
    {
      "epoch": 1.861178219661773,
      "grad_norm": 0.6150624752044678,
      "learning_rate": 1.8515077930495338e-06,
      "loss": 0.9405,
      "step": 10108
    },
    {
      "epoch": 1.8613640587251439,
      "grad_norm": 0.9233468770980835,
      "learning_rate": 1.8459102513907477e-06,
      "loss": 0.9038,
      "step": 10109
    },
    {
      "epoch": 1.8615498977885152,
      "grad_norm": 0.6704568862915039,
      "learning_rate": 1.840321105063414e-06,
      "loss": 0.8806,
      "step": 10110
    },
    {
      "epoch": 1.8617357368518863,
      "grad_norm": 0.9922787547111511,
      "learning_rate": 1.83474035454555e-06,
      "loss": 0.9688,
      "step": 10111
    },
    {
      "epoch": 1.8619215759152574,
      "grad_norm": 0.7160792946815491,
      "learning_rate": 1.8291680003145073e-06,
      "loss": 0.9976,
      "step": 10112
    },
    {
      "epoch": 1.8621074149786285,
      "grad_norm": 0.6046891212463379,
      "learning_rate": 1.8236040428469049e-06,
      "loss": 0.8351,
      "step": 10113
    },
    {
      "epoch": 1.8622932540419996,
      "grad_norm": 0.640130877494812,
      "learning_rate": 1.8180484826186283e-06,
      "loss": 0.8284,
      "step": 10114
    },
    {
      "epoch": 1.8624790931053707,
      "grad_norm": 0.7319105267524719,
      "learning_rate": 1.8125013201048757e-06,
      "loss": 1.0137,
      "step": 10115
    },
    {
      "epoch": 1.8626649321687418,
      "grad_norm": 0.6713131666183472,
      "learning_rate": 1.8069625557800896e-06,
      "loss": 0.758,
      "step": 10116
    },
    {
      "epoch": 1.862850771232113,
      "grad_norm": 0.5534182190895081,
      "learning_rate": 1.8014321901180354e-06,
      "loss": 0.772,
      "step": 10117
    },
    {
      "epoch": 1.863036610295484,
      "grad_norm": 0.7840726375579834,
      "learning_rate": 1.7959102235917014e-06,
      "loss": 1.123,
      "step": 10118
    },
    {
      "epoch": 1.8632224493588554,
      "grad_norm": 0.8231391310691833,
      "learning_rate": 1.790396656673421e-06,
      "loss": 0.977,
      "step": 10119
    },
    {
      "epoch": 1.8634082884222263,
      "grad_norm": 0.6878448128700256,
      "learning_rate": 1.7848914898347723e-06,
      "loss": 0.9395,
      "step": 10120
    },
    {
      "epoch": 1.8635941274855976,
      "grad_norm": 0.6510032415390015,
      "learning_rate": 1.7793947235466235e-06,
      "loss": 1.0007,
      "step": 10121
    },
    {
      "epoch": 1.8637799665489685,
      "grad_norm": 0.845956027507782,
      "learning_rate": 1.7739063582791315e-06,
      "loss": 1.0199,
      "step": 10122
    },
    {
      "epoch": 1.8639658056123398,
      "grad_norm": 0.837614119052887,
      "learning_rate": 1.768426394501721e-06,
      "loss": 0.9004,
      "step": 10123
    },
    {
      "epoch": 1.8641516446757107,
      "grad_norm": 0.6333421468734741,
      "learning_rate": 1.7629548326831058e-06,
      "loss": 0.8635,
      "step": 10124
    },
    {
      "epoch": 1.864337483739082,
      "grad_norm": 0.7224404215812683,
      "learning_rate": 1.7574916732912782e-06,
      "loss": 0.9852,
      "step": 10125
    },
    {
      "epoch": 1.864523322802453,
      "grad_norm": 0.7380586266517639,
      "learning_rate": 1.7520369167935202e-06,
      "loss": 0.9746,
      "step": 10126
    },
    {
      "epoch": 1.8647091618658242,
      "grad_norm": 0.8007980585098267,
      "learning_rate": 1.7465905636563806e-06,
      "loss": 0.8674,
      "step": 10127
    },
    {
      "epoch": 1.8648950009291954,
      "grad_norm": 0.6576514840126038,
      "learning_rate": 1.74115261434572e-06,
      "loss": 1.1316,
      "step": 10128
    },
    {
      "epoch": 1.8650808399925665,
      "grad_norm": 0.6236616969108582,
      "learning_rate": 1.7357230693266336e-06,
      "loss": 0.9738,
      "step": 10129
    },
    {
      "epoch": 1.8652666790559376,
      "grad_norm": 0.7905383706092834,
      "learning_rate": 1.730301929063527e-06,
      "loss": 1.2255,
      "step": 10130
    },
    {
      "epoch": 1.8654525181193087,
      "grad_norm": 0.6649918556213379,
      "learning_rate": 1.724889194020085e-06,
      "loss": 1.1122,
      "step": 10131
    },
    {
      "epoch": 1.8656383571826798,
      "grad_norm": 1.586954116821289,
      "learning_rate": 1.7194848646592709e-06,
      "loss": 1.3224,
      "step": 10132
    },
    {
      "epoch": 1.865824196246051,
      "grad_norm": 0.6258449554443359,
      "learning_rate": 1.714088941443337e-06,
      "loss": 0.938,
      "step": 10133
    },
    {
      "epoch": 1.866010035309422,
      "grad_norm": 0.873955488204956,
      "learning_rate": 1.7087014248337918e-06,
      "loss": 0.9731,
      "step": 10134
    },
    {
      "epoch": 1.866195874372793,
      "grad_norm": 0.7023456692695618,
      "learning_rate": 1.7033223152914446e-06,
      "loss": 0.937,
      "step": 10135
    },
    {
      "epoch": 1.8663817134361644,
      "grad_norm": 0.9610904455184937,
      "learning_rate": 1.697951613276405e-06,
      "loss": 1.1295,
      "step": 10136
    },
    {
      "epoch": 1.8665675524995353,
      "grad_norm": 0.6218596696853638,
      "learning_rate": 1.6925893192480058e-06,
      "loss": 0.8515,
      "step": 10137
    },
    {
      "epoch": 1.8667533915629066,
      "grad_norm": 0.7436656951904297,
      "learning_rate": 1.6872354336649243e-06,
      "loss": 0.9952,
      "step": 10138
    },
    {
      "epoch": 1.8669392306262775,
      "grad_norm": 0.6895730495452881,
      "learning_rate": 1.681889956985061e-06,
      "loss": 0.6323,
      "step": 10139
    },
    {
      "epoch": 1.8671250696896489,
      "grad_norm": 0.6715397238731384,
      "learning_rate": 1.6765528896656612e-06,
      "loss": 0.9762,
      "step": 10140
    },
    {
      "epoch": 1.8673109087530197,
      "grad_norm": 0.8727614283561707,
      "learning_rate": 1.6712242321631822e-06,
      "loss": 0.9921,
      "step": 10141
    },
    {
      "epoch": 1.867496747816391,
      "grad_norm": 0.6065548658370972,
      "learning_rate": 1.6659039849334147e-06,
      "loss": 0.8066,
      "step": 10142
    },
    {
      "epoch": 1.867682586879762,
      "grad_norm": 0.7665624618530273,
      "learning_rate": 1.6605921484314169e-06,
      "loss": 0.7634,
      "step": 10143
    },
    {
      "epoch": 1.8678684259431333,
      "grad_norm": 0.6623754501342773,
      "learning_rate": 1.6552887231115033e-06,
      "loss": 1.0655,
      "step": 10144
    },
    {
      "epoch": 1.8680542650065044,
      "grad_norm": 0.6355829238891602,
      "learning_rate": 1.6499937094273e-06,
      "loss": 0.7108,
      "step": 10145
    },
    {
      "epoch": 1.8682401040698755,
      "grad_norm": 0.7509677410125732,
      "learning_rate": 1.644707107831689e-06,
      "loss": 1.041,
      "step": 10146
    },
    {
      "epoch": 1.8684259431332466,
      "grad_norm": 0.8060480952262878,
      "learning_rate": 1.639428918776864e-06,
      "loss": 0.7989,
      "step": 10147
    },
    {
      "epoch": 1.8686117821966177,
      "grad_norm": 0.8885887861251831,
      "learning_rate": 1.6341591427142643e-06,
      "loss": 1.0574,
      "step": 10148
    },
    {
      "epoch": 1.8687976212599888,
      "grad_norm": 0.7218427062034607,
      "learning_rate": 1.6288977800946292e-06,
      "loss": 1.2323,
      "step": 10149
    },
    {
      "epoch": 1.86898346032336,
      "grad_norm": 0.668627142906189,
      "learning_rate": 1.6236448313679764e-06,
      "loss": 0.8974,
      "step": 10150
    },
    {
      "epoch": 1.869169299386731,
      "grad_norm": 0.6256469488143921,
      "learning_rate": 1.618400296983602e-06,
      "loss": 0.972,
      "step": 10151
    },
    {
      "epoch": 1.8693551384501021,
      "grad_norm": 0.679287314414978,
      "learning_rate": 1.6131641773900807e-06,
      "loss": 0.7333,
      "step": 10152
    },
    {
      "epoch": 1.8695409775134735,
      "grad_norm": 0.6528913974761963,
      "learning_rate": 1.6079364730352653e-06,
      "loss": 0.9249,
      "step": 10153
    },
    {
      "epoch": 1.8697268165768444,
      "grad_norm": 0.7103220224380493,
      "learning_rate": 1.6027171843662981e-06,
      "loss": 1.0212,
      "step": 10154
    },
    {
      "epoch": 1.8699126556402157,
      "grad_norm": 0.7032877206802368,
      "learning_rate": 1.5975063118295996e-06,
      "loss": 0.9606,
      "step": 10155
    },
    {
      "epoch": 1.8700984947035866,
      "grad_norm": 0.6555808782577515,
      "learning_rate": 1.592303855870847e-06,
      "loss": 0.894,
      "step": 10156
    },
    {
      "epoch": 1.870284333766958,
      "grad_norm": 0.6237680315971375,
      "learning_rate": 1.5871098169350395e-06,
      "loss": 0.9904,
      "step": 10157
    },
    {
      "epoch": 1.8704701728303288,
      "grad_norm": 0.7430229187011719,
      "learning_rate": 1.5819241954664331e-06,
      "loss": 0.9049,
      "step": 10158
    },
    {
      "epoch": 1.8706560118937001,
      "grad_norm": 0.695476770401001,
      "learning_rate": 1.576746991908562e-06,
      "loss": 1.1796,
      "step": 10159
    },
    {
      "epoch": 1.8708418509570712,
      "grad_norm": 0.6034365296363831,
      "learning_rate": 1.5715782067042385e-06,
      "loss": 0.8092,
      "step": 10160
    },
    {
      "epoch": 1.8710276900204423,
      "grad_norm": 0.7450648546218872,
      "learning_rate": 1.5664178402955532e-06,
      "loss": 0.9053,
      "step": 10161
    },
    {
      "epoch": 1.8712135290838134,
      "grad_norm": 0.6403368711471558,
      "learning_rate": 1.5612658931239089e-06,
      "loss": 1.0559,
      "step": 10162
    },
    {
      "epoch": 1.8713993681471845,
      "grad_norm": 0.6581562757492065,
      "learning_rate": 1.5561223656299307e-06,
      "loss": 0.9156,
      "step": 10163
    },
    {
      "epoch": 1.8715852072105557,
      "grad_norm": 0.8338717818260193,
      "learning_rate": 1.550987258253589e-06,
      "loss": 0.9417,
      "step": 10164
    },
    {
      "epoch": 1.8717710462739268,
      "grad_norm": 0.6185001730918884,
      "learning_rate": 1.5458605714340657e-06,
      "loss": 0.5992,
      "step": 10165
    },
    {
      "epoch": 1.8719568853372979,
      "grad_norm": 0.7387946248054504,
      "learning_rate": 1.540742305609899e-06,
      "loss": 0.9353,
      "step": 10166
    },
    {
      "epoch": 1.872142724400669,
      "grad_norm": 0.762082576751709,
      "learning_rate": 1.5356324612188166e-06,
      "loss": 1.1015,
      "step": 10167
    },
    {
      "epoch": 1.8723285634640403,
      "grad_norm": 0.7067580819129944,
      "learning_rate": 1.5305310386979132e-06,
      "loss": 1.0188,
      "step": 10168
    },
    {
      "epoch": 1.8725144025274112,
      "grad_norm": 0.7750047445297241,
      "learning_rate": 1.5254380384835065e-06,
      "loss": 0.9803,
      "step": 10169
    },
    {
      "epoch": 1.8727002415907825,
      "grad_norm": 0.7149769067764282,
      "learning_rate": 1.5203534610112147e-06,
      "loss": 1.1311,
      "step": 10170
    },
    {
      "epoch": 1.8728860806541534,
      "grad_norm": 0.8082043528556824,
      "learning_rate": 1.5152773067159342e-06,
      "loss": 0.9219,
      "step": 10171
    },
    {
      "epoch": 1.8730719197175247,
      "grad_norm": 0.6081134676933289,
      "learning_rate": 1.5102095760318291e-06,
      "loss": 0.9132,
      "step": 10172
    },
    {
      "epoch": 1.8732577587808956,
      "grad_norm": 0.6561916470527649,
      "learning_rate": 1.5051502693923858e-06,
      "loss": 1.0227,
      "step": 10173
    },
    {
      "epoch": 1.873443597844267,
      "grad_norm": 0.7679175734519958,
      "learning_rate": 1.5000993872302916e-06,
      "loss": 0.789,
      "step": 10174
    },
    {
      "epoch": 1.8736294369076378,
      "grad_norm": 0.6609779596328735,
      "learning_rate": 1.4950569299776008e-06,
      "loss": 0.979,
      "step": 10175
    },
    {
      "epoch": 1.8738152759710092,
      "grad_norm": 0.6817535161972046,
      "learning_rate": 1.4900228980655794e-06,
      "loss": 0.8825,
      "step": 10176
    },
    {
      "epoch": 1.8740011150343803,
      "grad_norm": 0.7659928202629089,
      "learning_rate": 1.4849972919248168e-06,
      "loss": 1.1114,
      "step": 10177
    },
    {
      "epoch": 1.8741869540977514,
      "grad_norm": 0.7996124029159546,
      "learning_rate": 1.4799801119851465e-06,
      "loss": 1.3768,
      "step": 10178
    },
    {
      "epoch": 1.8743727931611225,
      "grad_norm": 0.7944355607032776,
      "learning_rate": 1.474971358675703e-06,
      "loss": 0.8329,
      "step": 10179
    },
    {
      "epoch": 1.8745586322244936,
      "grad_norm": 0.6426302194595337,
      "learning_rate": 1.4699710324249105e-06,
      "loss": 0.9689,
      "step": 10180
    },
    {
      "epoch": 1.8747444712878647,
      "grad_norm": 0.7049875855445862,
      "learning_rate": 1.4649791336604491e-06,
      "loss": 0.8881,
      "step": 10181
    },
    {
      "epoch": 1.8749303103512358,
      "grad_norm": 0.7229358553886414,
      "learning_rate": 1.459995662809266e-06,
      "loss": 0.7566,
      "step": 10182
    },
    {
      "epoch": 1.875116149414607,
      "grad_norm": 0.5669488310813904,
      "learning_rate": 1.4550206202976536e-06,
      "loss": 0.8043,
      "step": 10183
    },
    {
      "epoch": 1.875301988477978,
      "grad_norm": 0.9962398409843445,
      "learning_rate": 1.4500540065510825e-06,
      "loss": 1.1309,
      "step": 10184
    },
    {
      "epoch": 1.8754878275413494,
      "grad_norm": 0.7182316184043884,
      "learning_rate": 1.4450958219944132e-06,
      "loss": 0.9019,
      "step": 10185
    },
    {
      "epoch": 1.8756736666047202,
      "grad_norm": 0.7526625990867615,
      "learning_rate": 1.4401460670516842e-06,
      "loss": 1.0304,
      "step": 10186
    },
    {
      "epoch": 1.8758595056680916,
      "grad_norm": 0.7577337026596069,
      "learning_rate": 1.4352047421462789e-06,
      "loss": 0.9453,
      "step": 10187
    },
    {
      "epoch": 1.8760453447314624,
      "grad_norm": 0.6579355597496033,
      "learning_rate": 1.4302718477008481e-06,
      "loss": 0.9844,
      "step": 10188
    },
    {
      "epoch": 1.8762311837948338,
      "grad_norm": 0.8321214318275452,
      "learning_rate": 1.4253473841372988e-06,
      "loss": 0.9747,
      "step": 10189
    },
    {
      "epoch": 1.8764170228582047,
      "grad_norm": 0.7083122730255127,
      "learning_rate": 1.4204313518768387e-06,
      "loss": 0.8278,
      "step": 10190
    },
    {
      "epoch": 1.876602861921576,
      "grad_norm": 0.7506430745124817,
      "learning_rate": 1.4155237513399312e-06,
      "loss": 0.9235,
      "step": 10191
    },
    {
      "epoch": 1.8767887009849469,
      "grad_norm": 0.6202881336212158,
      "learning_rate": 1.4106245829463627e-06,
      "loss": 0.7342,
      "step": 10192
    },
    {
      "epoch": 1.8769745400483182,
      "grad_norm": 0.7362320423126221,
      "learning_rate": 1.4057338471151427e-06,
      "loss": 1.0821,
      "step": 10193
    },
    {
      "epoch": 1.8771603791116893,
      "grad_norm": 0.6716258525848389,
      "learning_rate": 1.4008515442646141e-06,
      "loss": 0.8926,
      "step": 10194
    },
    {
      "epoch": 1.8773462181750604,
      "grad_norm": 0.727524995803833,
      "learning_rate": 1.395977674812332e-06,
      "loss": 0.9095,
      "step": 10195
    },
    {
      "epoch": 1.8775320572384315,
      "grad_norm": 0.6297793984413147,
      "learning_rate": 1.3911122391751962e-06,
      "loss": 0.9232,
      "step": 10196
    },
    {
      "epoch": 1.8777178963018026,
      "grad_norm": 0.7759661674499512,
      "learning_rate": 1.3862552377693627e-06,
      "loss": 0.8585,
      "step": 10197
    },
    {
      "epoch": 1.8779037353651737,
      "grad_norm": 0.7787834405899048,
      "learning_rate": 1.3814066710102436e-06,
      "loss": 1.0092,
      "step": 10198
    },
    {
      "epoch": 1.8780895744285449,
      "grad_norm": 0.6704865097999573,
      "learning_rate": 1.3765665393125515e-06,
      "loss": 0.9321,
      "step": 10199
    },
    {
      "epoch": 1.878275413491916,
      "grad_norm": 0.8808336853981018,
      "learning_rate": 1.3717348430902776e-06,
      "loss": 0.8639,
      "step": 10200
    },
    {
      "epoch": 1.878461252555287,
      "grad_norm": 0.7640100717544556,
      "learning_rate": 1.3669115827566915e-06,
      "loss": 1.0329,
      "step": 10201
    },
    {
      "epoch": 1.8786470916186584,
      "grad_norm": 0.7684122323989868,
      "learning_rate": 1.3620967587243294e-06,
      "loss": 1.0354,
      "step": 10202
    },
    {
      "epoch": 1.8788329306820293,
      "grad_norm": 0.7215694785118103,
      "learning_rate": 1.3572903714050067e-06,
      "loss": 1.1981,
      "step": 10203
    },
    {
      "epoch": 1.8790187697454006,
      "grad_norm": 0.6940367817878723,
      "learning_rate": 1.35249242120985e-06,
      "loss": 0.9857,
      "step": 10204
    },
    {
      "epoch": 1.8792046088087715,
      "grad_norm": 0.6102963089942932,
      "learning_rate": 1.3477029085492087e-06,
      "loss": 0.9679,
      "step": 10205
    },
    {
      "epoch": 1.8793904478721428,
      "grad_norm": 0.7456520795822144,
      "learning_rate": 1.3429218338327666e-06,
      "loss": 0.5382,
      "step": 10206
    },
    {
      "epoch": 1.8795762869355137,
      "grad_norm": 0.7495307922363281,
      "learning_rate": 1.3381491974694405e-06,
      "loss": 1.1192,
      "step": 10207
    },
    {
      "epoch": 1.879762125998885,
      "grad_norm": 0.7574533224105835,
      "learning_rate": 1.3333849998674487e-06,
      "loss": 1.0553,
      "step": 10208
    },
    {
      "epoch": 1.879947965062256,
      "grad_norm": 0.6794142723083496,
      "learning_rate": 1.328629241434276e-06,
      "loss": 0.9301,
      "step": 10209
    },
    {
      "epoch": 1.8801338041256273,
      "grad_norm": 0.7537904381752014,
      "learning_rate": 1.3238819225767085e-06,
      "loss": 0.8018,
      "step": 10210
    },
    {
      "epoch": 1.8803196431889984,
      "grad_norm": 0.8281949758529663,
      "learning_rate": 1.319143043700799e-06,
      "loss": 0.9391,
      "step": 10211
    },
    {
      "epoch": 1.8805054822523695,
      "grad_norm": 0.7075900435447693,
      "learning_rate": 1.3144126052118454e-06,
      "loss": 0.95,
      "step": 10212
    },
    {
      "epoch": 1.8806913213157406,
      "grad_norm": 0.7371687293052673,
      "learning_rate": 1.3096906075144798e-06,
      "loss": 0.973,
      "step": 10213
    },
    {
      "epoch": 1.8808771603791117,
      "grad_norm": 0.7950626611709595,
      "learning_rate": 1.3049770510125681e-06,
      "loss": 1.0202,
      "step": 10214
    },
    {
      "epoch": 1.8810629994424828,
      "grad_norm": 0.6449244618415833,
      "learning_rate": 1.3002719361092763e-06,
      "loss": 0.9921,
      "step": 10215
    },
    {
      "epoch": 1.881248838505854,
      "grad_norm": 0.7738901972770691,
      "learning_rate": 1.2955752632070384e-06,
      "loss": 1.0622,
      "step": 10216
    },
    {
      "epoch": 1.881434677569225,
      "grad_norm": 0.7098965644836426,
      "learning_rate": 1.2908870327075884e-06,
      "loss": 0.8252,
      "step": 10217
    },
    {
      "epoch": 1.881620516632596,
      "grad_norm": 0.707387387752533,
      "learning_rate": 1.2862072450119057e-06,
      "loss": 1.0608,
      "step": 10218
    },
    {
      "epoch": 1.8818063556959674,
      "grad_norm": 0.6619812846183777,
      "learning_rate": 1.2815359005202588e-06,
      "loss": 0.9372,
      "step": 10219
    },
    {
      "epoch": 1.8819921947593383,
      "grad_norm": 0.7271058559417725,
      "learning_rate": 1.2768729996322171e-06,
      "loss": 0.8793,
      "step": 10220
    },
    {
      "epoch": 1.8821780338227097,
      "grad_norm": 0.6920179128646851,
      "learning_rate": 1.272218542746595e-06,
      "loss": 0.8097,
      "step": 10221
    },
    {
      "epoch": 1.8823638728860805,
      "grad_norm": 0.7369257807731628,
      "learning_rate": 1.267572530261496e-06,
      "loss": 1.0333,
      "step": 10222
    },
    {
      "epoch": 1.8825497119494519,
      "grad_norm": 0.7454902529716492,
      "learning_rate": 1.2629349625743025e-06,
      "loss": 0.9823,
      "step": 10223
    },
    {
      "epoch": 1.8827355510128228,
      "grad_norm": 0.639519989490509,
      "learning_rate": 1.258305840081686e-06,
      "loss": 0.8006,
      "step": 10224
    },
    {
      "epoch": 1.882921390076194,
      "grad_norm": 0.7165234684944153,
      "learning_rate": 1.2536851631795854e-06,
      "loss": 0.9338,
      "step": 10225
    },
    {
      "epoch": 1.8831072291395652,
      "grad_norm": 0.602321445941925,
      "learning_rate": 1.2490729322631955e-06,
      "loss": 0.7861,
      "step": 10226
    },
    {
      "epoch": 1.8832930682029363,
      "grad_norm": 0.8231889605522156,
      "learning_rate": 1.2444691477270453e-06,
      "loss": 1.0127,
      "step": 10227
    },
    {
      "epoch": 1.8834789072663074,
      "grad_norm": 0.7236772179603577,
      "learning_rate": 1.2398738099648755e-06,
      "loss": 1.0205,
      "step": 10228
    },
    {
      "epoch": 1.8836647463296785,
      "grad_norm": 0.686307966709137,
      "learning_rate": 1.235286919369749e-06,
      "loss": 0.9445,
      "step": 10229
    },
    {
      "epoch": 1.8838505853930496,
      "grad_norm": 0.8536903262138367,
      "learning_rate": 1.2307084763339971e-06,
      "loss": 1.0062,
      "step": 10230
    },
    {
      "epoch": 1.8840364244564207,
      "grad_norm": 0.778286337852478,
      "learning_rate": 1.2261384812492061e-06,
      "loss": 1.0921,
      "step": 10231
    },
    {
      "epoch": 1.8842222635197918,
      "grad_norm": 0.740349531173706,
      "learning_rate": 1.2215769345062744e-06,
      "loss": 1.0735,
      "step": 10232
    },
    {
      "epoch": 1.884408102583163,
      "grad_norm": 0.671109676361084,
      "learning_rate": 1.2170238364953567e-06,
      "loss": 0.8181,
      "step": 10233
    },
    {
      "epoch": 1.8845939416465343,
      "grad_norm": 0.7762554287910461,
      "learning_rate": 1.212479187605897e-06,
      "loss": 0.8365,
      "step": 10234
    },
    {
      "epoch": 1.8847797807099052,
      "grad_norm": 0.699923574924469,
      "learning_rate": 1.2079429882265847e-06,
      "loss": 0.8449,
      "step": 10235
    },
    {
      "epoch": 1.8849656197732765,
      "grad_norm": 0.7666675448417664,
      "learning_rate": 1.2034152387454424e-06,
      "loss": 0.9722,
      "step": 10236
    },
    {
      "epoch": 1.8851514588366474,
      "grad_norm": 0.6274758577346802,
      "learning_rate": 1.1988959395497045e-06,
      "loss": 0.8508,
      "step": 10237
    },
    {
      "epoch": 1.8853372979000187,
      "grad_norm": 0.716448962688446,
      "learning_rate": 1.1943850910259401e-06,
      "loss": 0.9753,
      "step": 10238
    },
    {
      "epoch": 1.8855231369633896,
      "grad_norm": 0.739844560623169,
      "learning_rate": 1.1898826935599737e-06,
      "loss": 0.9353,
      "step": 10239
    },
    {
      "epoch": 1.885708976026761,
      "grad_norm": 0.66637122631073,
      "learning_rate": 1.1853887475368974e-06,
      "loss": 0.7805,
      "step": 10240
    },
    {
      "epoch": 1.8858948150901318,
      "grad_norm": 0.8586833477020264,
      "learning_rate": 1.1809032533410814e-06,
      "loss": 0.8905,
      "step": 10241
    },
    {
      "epoch": 1.8860806541535031,
      "grad_norm": 0.7887431383132935,
      "learning_rate": 1.1764262113561964e-06,
      "loss": 0.9023,
      "step": 10242
    },
    {
      "epoch": 1.8862664932168742,
      "grad_norm": 0.676100492477417,
      "learning_rate": 1.1719576219651585e-06,
      "loss": 0.8355,
      "step": 10243
    },
    {
      "epoch": 1.8864523322802453,
      "grad_norm": 0.8734824657440186,
      "learning_rate": 1.167497485550173e-06,
      "loss": 0.9403,
      "step": 10244
    },
    {
      "epoch": 1.8866381713436164,
      "grad_norm": 0.7353190779685974,
      "learning_rate": 1.1630458024927348e-06,
      "loss": 0.9986,
      "step": 10245
    },
    {
      "epoch": 1.8868240104069876,
      "grad_norm": 0.7474417090415955,
      "learning_rate": 1.158602573173606e-06,
      "loss": 0.9251,
      "step": 10246
    },
    {
      "epoch": 1.8870098494703587,
      "grad_norm": 0.6567177176475525,
      "learning_rate": 1.1541677979728161e-06,
      "loss": 0.78,
      "step": 10247
    },
    {
      "epoch": 1.8871956885337298,
      "grad_norm": 0.6639741659164429,
      "learning_rate": 1.1497414772697056e-06,
      "loss": 1.071,
      "step": 10248
    },
    {
      "epoch": 1.8873815275971009,
      "grad_norm": 0.656317949295044,
      "learning_rate": 1.145323611442839e-06,
      "loss": 0.7967,
      "step": 10249
    },
    {
      "epoch": 1.887567366660472,
      "grad_norm": 0.6598522067070007,
      "learning_rate": 1.1409142008701023e-06,
      "loss": 0.9534,
      "step": 10250
    },
    {
      "epoch": 1.8877532057238433,
      "grad_norm": 0.9313237071037292,
      "learning_rate": 1.1365132459286275e-06,
      "loss": 0.8162,
      "step": 10251
    },
    {
      "epoch": 1.8879390447872142,
      "grad_norm": 0.739664614200592,
      "learning_rate": 1.132120746994847e-06,
      "loss": 0.8651,
      "step": 10252
    },
    {
      "epoch": 1.8881248838505855,
      "grad_norm": 0.7578397989273071,
      "learning_rate": 1.127736704444471e-06,
      "loss": 0.8895,
      "step": 10253
    },
    {
      "epoch": 1.8883107229139564,
      "grad_norm": 0.7636153101921082,
      "learning_rate": 1.123361118652455e-06,
      "loss": 1.0874,
      "step": 10254
    },
    {
      "epoch": 1.8884965619773277,
      "grad_norm": 0.687780499458313,
      "learning_rate": 1.1189939899930778e-06,
      "loss": 0.7707,
      "step": 10255
    },
    {
      "epoch": 1.8886824010406986,
      "grad_norm": 0.6946448087692261,
      "learning_rate": 1.1146353188398407e-06,
      "loss": 0.9095,
      "step": 10256
    },
    {
      "epoch": 1.88886824010407,
      "grad_norm": 0.6907515525817871,
      "learning_rate": 1.1102851055655784e-06,
      "loss": 0.9238,
      "step": 10257
    },
    {
      "epoch": 1.8890540791674408,
      "grad_norm": 0.7136023044586182,
      "learning_rate": 1.1059433505423488e-06,
      "loss": 0.9172,
      "step": 10258
    },
    {
      "epoch": 1.8892399182308122,
      "grad_norm": 0.630096971988678,
      "learning_rate": 1.101610054141522e-06,
      "loss": 0.9887,
      "step": 10259
    },
    {
      "epoch": 1.8894257572941833,
      "grad_norm": 0.6554904580116272,
      "learning_rate": 1.0972852167337345e-06,
      "loss": 1.0012,
      "step": 10260
    },
    {
      "epoch": 1.8896115963575544,
      "grad_norm": 0.5938538908958435,
      "learning_rate": 1.0929688386889014e-06,
      "loss": 0.961,
      "step": 10261
    },
    {
      "epoch": 1.8897974354209255,
      "grad_norm": 0.6572749018669128,
      "learning_rate": 1.0886609203762054e-06,
      "loss": 0.9109,
      "step": 10262
    },
    {
      "epoch": 1.8899832744842966,
      "grad_norm": 0.7717788219451904,
      "learning_rate": 1.0843614621641185e-06,
      "loss": 1.2538,
      "step": 10263
    },
    {
      "epoch": 1.8901691135476677,
      "grad_norm": 0.8543664216995239,
      "learning_rate": 1.0800704644203908e-06,
      "loss": 0.8828,
      "step": 10264
    },
    {
      "epoch": 1.8903549526110388,
      "grad_norm": 0.6619085669517517,
      "learning_rate": 1.0757879275120175e-06,
      "loss": 0.8544,
      "step": 10265
    },
    {
      "epoch": 1.89054079167441,
      "grad_norm": 0.6682559251785278,
      "learning_rate": 1.071513851805317e-06,
      "loss": 0.7595,
      "step": 10266
    },
    {
      "epoch": 1.890726630737781,
      "grad_norm": 0.6358002424240112,
      "learning_rate": 1.067248237665841e-06,
      "loss": 0.7431,
      "step": 10267
    },
    {
      "epoch": 1.8909124698011524,
      "grad_norm": 0.6468208432197571,
      "learning_rate": 1.0629910854584534e-06,
      "loss": 0.7797,
      "step": 10268
    },
    {
      "epoch": 1.8910983088645232,
      "grad_norm": 0.7336880564689636,
      "learning_rate": 1.058742395547263e-06,
      "loss": 0.9174,
      "step": 10269
    },
    {
      "epoch": 1.8912841479278946,
      "grad_norm": 0.9050149917602539,
      "learning_rate": 1.0545021682956791e-06,
      "loss": 0.8339,
      "step": 10270
    },
    {
      "epoch": 1.8914699869912655,
      "grad_norm": 0.6562024354934692,
      "learning_rate": 1.0502704040663669e-06,
      "loss": 0.9984,
      "step": 10271
    },
    {
      "epoch": 1.8916558260546368,
      "grad_norm": 0.6586121916770935,
      "learning_rate": 1.0460471032212925e-06,
      "loss": 0.8628,
      "step": 10272
    },
    {
      "epoch": 1.8918416651180077,
      "grad_norm": 0.7983189821243286,
      "learning_rate": 1.0418322661216673e-06,
      "loss": 1.1281,
      "step": 10273
    },
    {
      "epoch": 1.892027504181379,
      "grad_norm": 0.6309989094734192,
      "learning_rate": 1.0376258931280136e-06,
      "loss": 0.8158,
      "step": 10274
    },
    {
      "epoch": 1.89221334324475,
      "grad_norm": 0.8750993013381958,
      "learning_rate": 1.0334279846001106e-06,
      "loss": 1.1648,
      "step": 10275
    },
    {
      "epoch": 1.8923991823081212,
      "grad_norm": 0.7219237089157104,
      "learning_rate": 1.029238540896993e-06,
      "loss": 0.9757,
      "step": 10276
    },
    {
      "epoch": 1.8925850213714923,
      "grad_norm": 0.7669157981872559,
      "learning_rate": 1.0250575623770075e-06,
      "loss": 0.9233,
      "step": 10277
    },
    {
      "epoch": 1.8927708604348634,
      "grad_norm": 0.7733719348907471,
      "learning_rate": 1.0208850493977685e-06,
      "loss": 0.849,
      "step": 10278
    },
    {
      "epoch": 1.8929566994982345,
      "grad_norm": 0.7132909297943115,
      "learning_rate": 1.0167210023161566e-06,
      "loss": 1.1725,
      "step": 10279
    },
    {
      "epoch": 1.8931425385616056,
      "grad_norm": 0.7796614766120911,
      "learning_rate": 1.0125654214883207e-06,
      "loss": 1.1139,
      "step": 10280
    },
    {
      "epoch": 1.8933283776249767,
      "grad_norm": 0.7235081791877747,
      "learning_rate": 1.0084183072697095e-06,
      "loss": 0.8145,
      "step": 10281
    },
    {
      "epoch": 1.8935142166883479,
      "grad_norm": 0.7745675444602966,
      "learning_rate": 1.0042796600150172e-06,
      "loss": 1.0418,
      "step": 10282
    },
    {
      "epoch": 1.893700055751719,
      "grad_norm": 0.6154912710189819,
      "learning_rate": 1.0001494800782606e-06,
      "loss": 0.9604,
      "step": 10283
    },
    {
      "epoch": 1.89388589481509,
      "grad_norm": 0.7308815717697144,
      "learning_rate": 9.960277678126795e-07,
      "loss": 0.9988,
      "step": 10284
    },
    {
      "epoch": 1.8940717338784614,
      "grad_norm": 0.7246875762939453,
      "learning_rate": 9.919145235708139e-07,
      "loss": 1.0016,
      "step": 10285
    },
    {
      "epoch": 1.8942575729418323,
      "grad_norm": 0.7031250596046448,
      "learning_rate": 9.878097477044934e-07,
      "loss": 0.9951,
      "step": 10286
    },
    {
      "epoch": 1.8944434120052036,
      "grad_norm": 0.6622384786605835,
      "learning_rate": 9.83713440564782e-07,
      "loss": 1.0912,
      "step": 10287
    },
    {
      "epoch": 1.8946292510685745,
      "grad_norm": 0.6151398420333862,
      "learning_rate": 9.796256025020767e-07,
      "loss": 0.9752,
      "step": 10288
    },
    {
      "epoch": 1.8948150901319458,
      "grad_norm": 0.6820587515830994,
      "learning_rate": 9.755462338659981e-07,
      "loss": 0.9326,
      "step": 10289
    },
    {
      "epoch": 1.8950009291953167,
      "grad_norm": 0.5994402766227722,
      "learning_rate": 9.714753350054783e-07,
      "loss": 0.7013,
      "step": 10290
    },
    {
      "epoch": 1.895186768258688,
      "grad_norm": 0.7843385338783264,
      "learning_rate": 9.674129062686944e-07,
      "loss": 1.1573,
      "step": 10291
    },
    {
      "epoch": 1.8953726073220591,
      "grad_norm": 0.7633580565452576,
      "learning_rate": 9.633589480031236e-07,
      "loss": 0.7809,
      "step": 10292
    },
    {
      "epoch": 1.8955584463854303,
      "grad_norm": 0.7467591762542725,
      "learning_rate": 9.593134605555109e-07,
      "loss": 0.868,
      "step": 10293
    },
    {
      "epoch": 1.8957442854488014,
      "grad_norm": 0.9395126104354858,
      "learning_rate": 9.552764442718687e-07,
      "loss": 0.9111,
      "step": 10294
    },
    {
      "epoch": 1.8959301245121725,
      "grad_norm": 1.2283921241760254,
      "learning_rate": 9.512478994974982e-07,
      "loss": 1.4538,
      "step": 10295
    },
    {
      "epoch": 1.8961159635755436,
      "grad_norm": 0.685003399848938,
      "learning_rate": 9.472278265769685e-07,
      "loss": 1.1466,
      "step": 10296
    },
    {
      "epoch": 1.8963018026389147,
      "grad_norm": 0.7409753799438477,
      "learning_rate": 9.432162258541267e-07,
      "loss": 0.9852,
      "step": 10297
    },
    {
      "epoch": 1.8964876417022858,
      "grad_norm": 0.7501311302185059,
      "learning_rate": 9.39213097672087e-07,
      "loss": 1.0421,
      "step": 10298
    },
    {
      "epoch": 1.896673480765657,
      "grad_norm": 0.6653595566749573,
      "learning_rate": 9.352184423732535e-07,
      "loss": 1.1197,
      "step": 10299
    },
    {
      "epoch": 1.8968593198290282,
      "grad_norm": 0.6663728356361389,
      "learning_rate": 9.312322602992863e-07,
      "loss": 0.8264,
      "step": 10300
    },
    {
      "epoch": 1.897045158892399,
      "grad_norm": 0.697401225566864,
      "learning_rate": 9.272545517911457e-07,
      "loss": 1.1152,
      "step": 10301
    },
    {
      "epoch": 1.8972309979557704,
      "grad_norm": 0.8602698445320129,
      "learning_rate": 9.232853171890488e-07,
      "loss": 0.7819,
      "step": 10302
    },
    {
      "epoch": 1.8974168370191413,
      "grad_norm": 0.8105998635292053,
      "learning_rate": 9.193245568324904e-07,
      "loss": 0.9317,
      "step": 10303
    },
    {
      "epoch": 1.8976026760825127,
      "grad_norm": 0.700174868106842,
      "learning_rate": 9.15372271060233e-07,
      "loss": 0.9864,
      "step": 10304
    },
    {
      "epoch": 1.8977885151458835,
      "grad_norm": 0.6726325154304504,
      "learning_rate": 9.114284602103507e-07,
      "loss": 0.726,
      "step": 10305
    },
    {
      "epoch": 1.8979743542092549,
      "grad_norm": 0.7302104830741882,
      "learning_rate": 9.074931246201402e-07,
      "loss": 1.1052,
      "step": 10306
    },
    {
      "epoch": 1.8981601932726258,
      "grad_norm": 0.5432365536689758,
      "learning_rate": 9.03566264626221e-07,
      "loss": 0.6084,
      "step": 10307
    },
    {
      "epoch": 1.898346032335997,
      "grad_norm": 0.6986281871795654,
      "learning_rate": 8.996478805644471e-07,
      "loss": 0.8696,
      "step": 10308
    },
    {
      "epoch": 1.8985318713993682,
      "grad_norm": 0.9530007243156433,
      "learning_rate": 8.957379727699722e-07,
      "loss": 1.023,
      "step": 10309
    },
    {
      "epoch": 1.8987177104627393,
      "grad_norm": 0.6925898194313049,
      "learning_rate": 8.91836541577229e-07,
      "loss": 1.1046,
      "step": 10310
    },
    {
      "epoch": 1.8989035495261104,
      "grad_norm": 0.8129331469535828,
      "learning_rate": 8.879435873199061e-07,
      "loss": 0.9929,
      "step": 10311
    },
    {
      "epoch": 1.8990893885894815,
      "grad_norm": 0.7528035044670105,
      "learning_rate": 8.840591103309703e-07,
      "loss": 1.167,
      "step": 10312
    },
    {
      "epoch": 1.8992752276528526,
      "grad_norm": 0.7556763887405396,
      "learning_rate": 8.801831109426783e-07,
      "loss": 1.006,
      "step": 10313
    },
    {
      "epoch": 1.8994610667162237,
      "grad_norm": 0.6110503077507019,
      "learning_rate": 8.763155894865538e-07,
      "loss": 0.7358,
      "step": 10314
    },
    {
      "epoch": 1.8996469057795948,
      "grad_norm": 0.8103258609771729,
      "learning_rate": 8.724565462933876e-07,
      "loss": 1.0752,
      "step": 10315
    },
    {
      "epoch": 1.899832744842966,
      "grad_norm": 0.7095519304275513,
      "learning_rate": 8.686059816932602e-07,
      "loss": 1.0222,
      "step": 10316
    },
    {
      "epoch": 1.9000185839063373,
      "grad_norm": 0.705930233001709,
      "learning_rate": 8.64763896015508e-07,
      "loss": 0.9961,
      "step": 10317
    },
    {
      "epoch": 1.9002044229697082,
      "grad_norm": 0.667270302772522,
      "learning_rate": 8.609302895887683e-07,
      "loss": 0.8224,
      "step": 10318
    },
    {
      "epoch": 1.9003902620330795,
      "grad_norm": 0.6715710163116455,
      "learning_rate": 8.571051627409232e-07,
      "loss": 0.9866,
      "step": 10319
    },
    {
      "epoch": 1.9005761010964504,
      "grad_norm": 0.6669120788574219,
      "learning_rate": 8.532885157991555e-07,
      "loss": 0.8995,
      "step": 10320
    },
    {
      "epoch": 1.9007619401598217,
      "grad_norm": 0.7272221446037292,
      "learning_rate": 8.494803490898928e-07,
      "loss": 1.0071,
      "step": 10321
    },
    {
      "epoch": 1.9009477792231926,
      "grad_norm": 0.7126994132995605,
      "learning_rate": 8.456806629388747e-07,
      "loss": 0.8916,
      "step": 10322
    },
    {
      "epoch": 1.901133618286564,
      "grad_norm": 0.6916284561157227,
      "learning_rate": 8.418894576710967e-07,
      "loss": 1.0381,
      "step": 10323
    },
    {
      "epoch": 1.9013194573499348,
      "grad_norm": 0.7423785328865051,
      "learning_rate": 8.381067336108106e-07,
      "loss": 0.9574,
      "step": 10324
    },
    {
      "epoch": 1.9015052964133061,
      "grad_norm": 0.8324457406997681,
      "learning_rate": 8.343324910815798e-07,
      "loss": 0.7473,
      "step": 10325
    },
    {
      "epoch": 1.9016911354766772,
      "grad_norm": 0.6474454998970032,
      "learning_rate": 8.305667304062237e-07,
      "loss": 0.8059,
      "step": 10326
    },
    {
      "epoch": 1.9018769745400483,
      "grad_norm": 0.6828518509864807,
      "learning_rate": 8.268094519068182e-07,
      "loss": 1.0419,
      "step": 10327
    },
    {
      "epoch": 1.9020628136034194,
      "grad_norm": 0.7035902142524719,
      "learning_rate": 8.230606559047394e-07,
      "loss": 0.8835,
      "step": 10328
    },
    {
      "epoch": 1.9022486526667906,
      "grad_norm": 0.6972468495368958,
      "learning_rate": 8.19320342720642e-07,
      "loss": 1.0148,
      "step": 10329
    },
    {
      "epoch": 1.9024344917301617,
      "grad_norm": 0.8360022902488708,
      "learning_rate": 8.155885126744256e-07,
      "loss": 1.222,
      "step": 10330
    },
    {
      "epoch": 1.9026203307935328,
      "grad_norm": 0.7290945649147034,
      "learning_rate": 8.118651660852905e-07,
      "loss": 0.9121,
      "step": 10331
    },
    {
      "epoch": 1.9028061698569039,
      "grad_norm": 0.7698082327842712,
      "learning_rate": 8.081503032717041e-07,
      "loss": 0.9743,
      "step": 10332
    },
    {
      "epoch": 1.902992008920275,
      "grad_norm": 0.6450352072715759,
      "learning_rate": 8.044439245514013e-07,
      "loss": 0.8837,
      "step": 10333
    },
    {
      "epoch": 1.9031778479836463,
      "grad_norm": 0.7240906953811646,
      "learning_rate": 8.00746030241395e-07,
      "loss": 0.8972,
      "step": 10334
    },
    {
      "epoch": 1.9033636870470172,
      "grad_norm": 0.6440247297286987,
      "learning_rate": 7.970566206579877e-07,
      "loss": 0.8675,
      "step": 10335
    },
    {
      "epoch": 1.9035495261103885,
      "grad_norm": 0.780107855796814,
      "learning_rate": 7.933756961167271e-07,
      "loss": 1.0229,
      "step": 10336
    },
    {
      "epoch": 1.9037353651737594,
      "grad_norm": 0.65886390209198,
      "learning_rate": 7.897032569324614e-07,
      "loss": 1.0398,
      "step": 10337
    },
    {
      "epoch": 1.9039212042371307,
      "grad_norm": 0.6943807601928711,
      "learning_rate": 7.860393034192948e-07,
      "loss": 0.8795,
      "step": 10338
    },
    {
      "epoch": 1.9041070433005016,
      "grad_norm": 0.7381815314292908,
      "learning_rate": 7.82383835890621e-07,
      "loss": 0.8438,
      "step": 10339
    },
    {
      "epoch": 1.904292882363873,
      "grad_norm": 0.7263152003288269,
      "learning_rate": 7.787368546590901e-07,
      "loss": 0.8213,
      "step": 10340
    },
    {
      "epoch": 1.904478721427244,
      "grad_norm": 1.1103852987289429,
      "learning_rate": 7.750983600366524e-07,
      "loss": 0.8701,
      "step": 10341
    },
    {
      "epoch": 1.9046645604906152,
      "grad_norm": 0.752496600151062,
      "learning_rate": 7.714683523344924e-07,
      "loss": 1.1805,
      "step": 10342
    },
    {
      "epoch": 1.9048503995539863,
      "grad_norm": 0.8111420273780823,
      "learning_rate": 7.678468318631061e-07,
      "loss": 1.162,
      "step": 10343
    },
    {
      "epoch": 1.9050362386173574,
      "grad_norm": 0.6829381585121155,
      "learning_rate": 7.642337989322568e-07,
      "loss": 1.0697,
      "step": 10344
    },
    {
      "epoch": 1.9052220776807285,
      "grad_norm": 0.7154696583747864,
      "learning_rate": 7.60629253850953e-07,
      "loss": 0.9418,
      "step": 10345
    },
    {
      "epoch": 1.9054079167440996,
      "grad_norm": 0.7428658604621887,
      "learning_rate": 7.570331969275257e-07,
      "loss": 0.8949,
      "step": 10346
    },
    {
      "epoch": 1.9055937558074707,
      "grad_norm": 0.7169175148010254,
      "learning_rate": 7.534456284695401e-07,
      "loss": 1.0155,
      "step": 10347
    },
    {
      "epoch": 1.9057795948708418,
      "grad_norm": 0.5764283537864685,
      "learning_rate": 7.498665487838508e-07,
      "loss": 0.6472,
      "step": 10348
    },
    {
      "epoch": 1.9059654339342131,
      "grad_norm": 0.657071590423584,
      "learning_rate": 7.462959581765794e-07,
      "loss": 1.0341,
      "step": 10349
    },
    {
      "epoch": 1.906151272997584,
      "grad_norm": 0.7069972157478333,
      "learning_rate": 7.427338569531261e-07,
      "loss": 0.8275,
      "step": 10350
    },
    {
      "epoch": 1.9063371120609554,
      "grad_norm": 0.7730807065963745,
      "learning_rate": 7.391802454181807e-07,
      "loss": 0.9245,
      "step": 10351
    },
    {
      "epoch": 1.9065229511243262,
      "grad_norm": 0.6439422965049744,
      "learning_rate": 7.356351238756665e-07,
      "loss": 1.1224,
      "step": 10352
    },
    {
      "epoch": 1.9067087901876976,
      "grad_norm": 0.7238260507583618,
      "learning_rate": 7.320984926288299e-07,
      "loss": 1.0158,
      "step": 10353
    },
    {
      "epoch": 1.9068946292510685,
      "grad_norm": 0.6328291893005371,
      "learning_rate": 7.285703519801512e-07,
      "loss": 1.0939,
      "step": 10354
    },
    {
      "epoch": 1.9070804683144398,
      "grad_norm": 0.7213243842124939,
      "learning_rate": 7.250507022314113e-07,
      "loss": 1.0515,
      "step": 10355
    },
    {
      "epoch": 1.9072663073778107,
      "grad_norm": 0.6419424414634705,
      "learning_rate": 7.215395436836358e-07,
      "loss": 0.7971,
      "step": 10356
    },
    {
      "epoch": 1.907452146441182,
      "grad_norm": 0.6070999503135681,
      "learning_rate": 7.180368766371515e-07,
      "loss": 0.8184,
      "step": 10357
    },
    {
      "epoch": 1.907637985504553,
      "grad_norm": 0.6605267524719238,
      "learning_rate": 7.145427013915629e-07,
      "loss": 0.8258,
      "step": 10358
    },
    {
      "epoch": 1.9078238245679242,
      "grad_norm": 0.761379599571228,
      "learning_rate": 7.110570182457088e-07,
      "loss": 0.9409,
      "step": 10359
    },
    {
      "epoch": 1.9080096636312953,
      "grad_norm": 0.7137780785560608,
      "learning_rate": 7.075798274977396e-07,
      "loss": 0.8647,
      "step": 10360
    },
    {
      "epoch": 1.9081955026946664,
      "grad_norm": 0.800717294216156,
      "learning_rate": 7.04111129445073e-07,
      "loss": 0.9653,
      "step": 10361
    },
    {
      "epoch": 1.9083813417580375,
      "grad_norm": 0.8079236149787903,
      "learning_rate": 7.006509243843829e-07,
      "loss": 1.1265,
      "step": 10362
    },
    {
      "epoch": 1.9085671808214086,
      "grad_norm": 0.8685330748558044,
      "learning_rate": 6.971992126116322e-07,
      "loss": 1.1229,
      "step": 10363
    },
    {
      "epoch": 1.9087530198847797,
      "grad_norm": 0.6397227048873901,
      "learning_rate": 6.937559944220628e-07,
      "loss": 0.9112,
      "step": 10364
    },
    {
      "epoch": 1.9089388589481509,
      "grad_norm": 0.6759904623031616,
      "learning_rate": 6.903212701101725e-07,
      "loss": 0.8865,
      "step": 10365
    },
    {
      "epoch": 1.9091246980115222,
      "grad_norm": 0.6716403961181641,
      "learning_rate": 6.86895039969726e-07,
      "loss": 1.143,
      "step": 10366
    },
    {
      "epoch": 1.909310537074893,
      "grad_norm": 0.6793673038482666,
      "learning_rate": 6.834773042938114e-07,
      "loss": 0.8889,
      "step": 10367
    },
    {
      "epoch": 1.9094963761382644,
      "grad_norm": 0.7783335447311401,
      "learning_rate": 6.800680633747169e-07,
      "loss": 1.1913,
      "step": 10368
    },
    {
      "epoch": 1.9096822152016353,
      "grad_norm": 0.7612161636352539,
      "learning_rate": 6.76667317504065e-07,
      "loss": 0.9476,
      "step": 10369
    },
    {
      "epoch": 1.9098680542650066,
      "grad_norm": 0.7343637347221375,
      "learning_rate": 6.732750669727228e-07,
      "loss": 1.0106,
      "step": 10370
    },
    {
      "epoch": 1.9100538933283775,
      "grad_norm": 1.6431206464767456,
      "learning_rate": 6.698913120708361e-07,
      "loss": 1.6607,
      "step": 10371
    },
    {
      "epoch": 1.9102397323917488,
      "grad_norm": 0.6783691644668579,
      "learning_rate": 6.665160530878401e-07,
      "loss": 0.8022,
      "step": 10372
    },
    {
      "epoch": 1.9104255714551197,
      "grad_norm": 2.257593870162964,
      "learning_rate": 6.631492903123926e-07,
      "loss": 1.3108,
      "step": 10373
    },
    {
      "epoch": 1.910611410518491,
      "grad_norm": 0.7017894983291626,
      "learning_rate": 6.597910240324967e-07,
      "loss": 1.1359,
      "step": 10374
    },
    {
      "epoch": 1.9107972495818621,
      "grad_norm": 0.6423133611679077,
      "learning_rate": 6.564412545353671e-07,
      "loss": 1.0166,
      "step": 10375
    },
    {
      "epoch": 1.9109830886452333,
      "grad_norm": 0.6490014791488647,
      "learning_rate": 6.530999821075301e-07,
      "loss": 0.9365,
      "step": 10376
    },
    {
      "epoch": 1.9111689277086044,
      "grad_norm": 0.7735306620597839,
      "learning_rate": 6.497672070347793e-07,
      "loss": 1.0894,
      "step": 10377
    },
    {
      "epoch": 1.9113547667719755,
      "grad_norm": 0.7858856916427612,
      "learning_rate": 6.464429296021534e-07,
      "loss": 1.1268,
      "step": 10378
    },
    {
      "epoch": 1.9115406058353466,
      "grad_norm": 0.655606746673584,
      "learning_rate": 6.431271500939917e-07,
      "loss": 1.1135,
      "step": 10379
    },
    {
      "epoch": 1.9117264448987177,
      "grad_norm": 0.8696974515914917,
      "learning_rate": 6.398198687939005e-07,
      "loss": 1.0957,
      "step": 10380
    },
    {
      "epoch": 1.9119122839620888,
      "grad_norm": 0.7006300687789917,
      "learning_rate": 6.365210859847759e-07,
      "loss": 0.9136,
      "step": 10381
    },
    {
      "epoch": 1.91209812302546,
      "grad_norm": 0.687654435634613,
      "learning_rate": 6.332308019487476e-07,
      "loss": 0.6121,
      "step": 10382
    },
    {
      "epoch": 1.9122839620888312,
      "grad_norm": 0.6635216474533081,
      "learning_rate": 6.299490169672462e-07,
      "loss": 0.9353,
      "step": 10383
    },
    {
      "epoch": 1.912469801152202,
      "grad_norm": 0.6450825333595276,
      "learning_rate": 6.266757313209693e-07,
      "loss": 0.8709,
      "step": 10384
    },
    {
      "epoch": 1.9126556402155734,
      "grad_norm": 0.6535255312919617,
      "learning_rate": 6.234109452899039e-07,
      "loss": 0.9271,
      "step": 10385
    },
    {
      "epoch": 1.9128414792789443,
      "grad_norm": 0.6958978176116943,
      "learning_rate": 6.201546591532714e-07,
      "loss": 0.6958,
      "step": 10386
    },
    {
      "epoch": 1.9130273183423157,
      "grad_norm": 0.6085829138755798,
      "learning_rate": 6.169068731896044e-07,
      "loss": 0.5802,
      "step": 10387
    },
    {
      "epoch": 1.9132131574056865,
      "grad_norm": 0.7713790535926819,
      "learning_rate": 6.136675876766806e-07,
      "loss": 1.1302,
      "step": 10388
    },
    {
      "epoch": 1.9133989964690579,
      "grad_norm": 0.6537017822265625,
      "learning_rate": 6.104368028915785e-07,
      "loss": 0.8586,
      "step": 10389
    },
    {
      "epoch": 1.9135848355324288,
      "grad_norm": 0.6550865769386292,
      "learning_rate": 6.072145191106216e-07,
      "loss": 0.8799,
      "step": 10390
    },
    {
      "epoch": 1.9137706745958,
      "grad_norm": 0.7534873485565186,
      "learning_rate": 6.040007366094225e-07,
      "loss": 0.998,
      "step": 10391
    },
    {
      "epoch": 1.9139565136591712,
      "grad_norm": 0.756750226020813,
      "learning_rate": 6.007954556628615e-07,
      "loss": 1.0923,
      "step": 10392
    },
    {
      "epoch": 1.9141423527225423,
      "grad_norm": 0.8632221221923828,
      "learning_rate": 5.975986765450969e-07,
      "loss": 1.1357,
      "step": 10393
    },
    {
      "epoch": 1.9143281917859134,
      "grad_norm": 0.73847496509552,
      "learning_rate": 5.944103995295547e-07,
      "loss": 0.9223,
      "step": 10394
    },
    {
      "epoch": 1.9145140308492845,
      "grad_norm": 0.6389723420143127,
      "learning_rate": 5.912306248889388e-07,
      "loss": 0.9463,
      "step": 10395
    },
    {
      "epoch": 1.9146998699126556,
      "grad_norm": 0.6227095723152161,
      "learning_rate": 5.880593528952094e-07,
      "loss": 0.8493,
      "step": 10396
    },
    {
      "epoch": 1.9148857089760267,
      "grad_norm": 0.6578342914581299,
      "learning_rate": 5.848965838196274e-07,
      "loss": 0.9054,
      "step": 10397
    },
    {
      "epoch": 1.9150715480393978,
      "grad_norm": 0.8469536900520325,
      "learning_rate": 5.817423179327098e-07,
      "loss": 1.0138,
      "step": 10398
    },
    {
      "epoch": 1.915257387102769,
      "grad_norm": 0.6048383712768555,
      "learning_rate": 5.785965555042295e-07,
      "loss": 0.85,
      "step": 10399
    },
    {
      "epoch": 1.9154432261661403,
      "grad_norm": 1.3355565071105957,
      "learning_rate": 5.754592968032713e-07,
      "loss": 1.393,
      "step": 10400
    },
    {
      "epoch": 1.9156290652295112,
      "grad_norm": 1.1654173135757446,
      "learning_rate": 5.72330542098165e-07,
      "loss": 1.057,
      "step": 10401
    },
    {
      "epoch": 1.9158149042928825,
      "grad_norm": 1.7299017906188965,
      "learning_rate": 5.69210291656519e-07,
      "loss": 1.3309,
      "step": 10402
    },
    {
      "epoch": 1.9160007433562534,
      "grad_norm": 0.7381700873374939,
      "learning_rate": 5.660985457452195e-07,
      "loss": 0.9086,
      "step": 10403
    },
    {
      "epoch": 1.9161865824196247,
      "grad_norm": 0.9093272686004639,
      "learning_rate": 5.629953046304093e-07,
      "loss": 0.8974,
      "step": 10404
    },
    {
      "epoch": 1.9163724214829956,
      "grad_norm": 0.7489191293716431,
      "learning_rate": 5.599005685775205e-07,
      "loss": 0.9781,
      "step": 10405
    },
    {
      "epoch": 1.916558260546367,
      "grad_norm": 0.7251028418540955,
      "learning_rate": 5.568143378512525e-07,
      "loss": 1.0954,
      "step": 10406
    },
    {
      "epoch": 1.916744099609738,
      "grad_norm": 0.6643712520599365,
      "learning_rate": 5.53736612715583e-07,
      "loss": 0.8945,
      "step": 10407
    },
    {
      "epoch": 1.9169299386731091,
      "grad_norm": 0.7415669560432434,
      "learning_rate": 5.50667393433757e-07,
      "loss": 1.152,
      "step": 10408
    },
    {
      "epoch": 1.9171157777364802,
      "grad_norm": 0.6884954571723938,
      "learning_rate": 5.476066802682866e-07,
      "loss": 1.1658,
      "step": 10409
    },
    {
      "epoch": 1.9173016167998513,
      "grad_norm": 0.6170604825019836,
      "learning_rate": 5.445544734809626e-07,
      "loss": 0.9808,
      "step": 10410
    },
    {
      "epoch": 1.9174874558632224,
      "grad_norm": 0.6698222160339355,
      "learning_rate": 5.415107733328428e-07,
      "loss": 1.1359,
      "step": 10411
    },
    {
      "epoch": 1.9176732949265936,
      "grad_norm": 0.741590142250061,
      "learning_rate": 5.384755800842633e-07,
      "loss": 0.7777,
      "step": 10412
    },
    {
      "epoch": 1.9178591339899647,
      "grad_norm": 0.6918681263923645,
      "learning_rate": 5.354488939948277e-07,
      "loss": 0.9641,
      "step": 10413
    },
    {
      "epoch": 1.9180449730533358,
      "grad_norm": 0.6736308336257935,
      "learning_rate": 5.324307153234287e-07,
      "loss": 0.8739,
      "step": 10414
    },
    {
      "epoch": 1.918230812116707,
      "grad_norm": 0.7565740346908569,
      "learning_rate": 5.294210443282044e-07,
      "loss": 0.6995,
      "step": 10415
    },
    {
      "epoch": 1.918416651180078,
      "grad_norm": 0.7019480466842651,
      "learning_rate": 5.264198812665822e-07,
      "loss": 0.6801,
      "step": 10416
    },
    {
      "epoch": 1.9186024902434493,
      "grad_norm": 0.7393890619277954,
      "learning_rate": 5.234272263952566e-07,
      "loss": 1.0164,
      "step": 10417
    },
    {
      "epoch": 1.9187883293068202,
      "grad_norm": 0.93512362241745,
      "learning_rate": 5.204430799702009e-07,
      "loss": 1.0921,
      "step": 10418
    },
    {
      "epoch": 1.9189741683701915,
      "grad_norm": 0.7040289640426636,
      "learning_rate": 5.174674422466552e-07,
      "loss": 0.8494,
      "step": 10419
    },
    {
      "epoch": 1.9191600074335624,
      "grad_norm": 0.6942346096038818,
      "learning_rate": 5.14500313479116e-07,
      "loss": 1.0021,
      "step": 10420
    },
    {
      "epoch": 1.9193458464969337,
      "grad_norm": 0.6451189517974854,
      "learning_rate": 5.115416939213913e-07,
      "loss": 0.8916,
      "step": 10421
    },
    {
      "epoch": 1.9195316855603046,
      "grad_norm": 1.1839951276779175,
      "learning_rate": 5.085915838265343e-07,
      "loss": 1.3439,
      "step": 10422
    },
    {
      "epoch": 1.919717524623676,
      "grad_norm": 0.7084904313087463,
      "learning_rate": 5.056499834468543e-07,
      "loss": 0.9234,
      "step": 10423
    },
    {
      "epoch": 1.919903363687047,
      "grad_norm": 0.6282612681388855,
      "learning_rate": 5.027168930339721e-07,
      "loss": 0.7671,
      "step": 10424
    },
    {
      "epoch": 1.9200892027504182,
      "grad_norm": 0.7924585938453674,
      "learning_rate": 4.99792312838765e-07,
      "loss": 0.8562,
      "step": 10425
    },
    {
      "epoch": 1.9202750418137893,
      "grad_norm": 0.746904730796814,
      "learning_rate": 4.968762431113549e-07,
      "loss": 1.0235,
      "step": 10426
    },
    {
      "epoch": 1.9204608808771604,
      "grad_norm": 0.7356643080711365,
      "learning_rate": 4.93968684101187e-07,
      "loss": 1.0711,
      "step": 10427
    },
    {
      "epoch": 1.9206467199405315,
      "grad_norm": 0.7611639499664307,
      "learning_rate": 4.910696360569289e-07,
      "loss": 0.9852,
      "step": 10428
    },
    {
      "epoch": 1.9208325590039026,
      "grad_norm": 0.7935344576835632,
      "learning_rate": 4.881790992265601e-07,
      "loss": 0.9326,
      "step": 10429
    },
    {
      "epoch": 1.9210183980672737,
      "grad_norm": 0.6445058584213257,
      "learning_rate": 4.852970738573048e-07,
      "loss": 0.8003,
      "step": 10430
    },
    {
      "epoch": 1.9212042371306448,
      "grad_norm": 0.6539177894592285,
      "learning_rate": 4.824235601956661e-07,
      "loss": 0.797,
      "step": 10431
    },
    {
      "epoch": 1.9213900761940161,
      "grad_norm": 0.7449811100959778,
      "learning_rate": 4.795585584874252e-07,
      "loss": 0.9627,
      "step": 10432
    },
    {
      "epoch": 1.921575915257387,
      "grad_norm": 0.7212844491004944,
      "learning_rate": 4.7670206897763025e-07,
      "loss": 0.9828,
      "step": 10433
    },
    {
      "epoch": 1.9217617543207584,
      "grad_norm": 0.6266747713088989,
      "learning_rate": 4.7385409191060826e-07,
      "loss": 0.5924,
      "step": 10434
    },
    {
      "epoch": 1.9219475933841292,
      "grad_norm": 0.6358219385147095,
      "learning_rate": 4.710146275299643e-07,
      "loss": 0.8659,
      "step": 10435
    },
    {
      "epoch": 1.9221334324475006,
      "grad_norm": 0.7808411121368408,
      "learning_rate": 4.681836760785374e-07,
      "loss": 0.9741,
      "step": 10436
    },
    {
      "epoch": 1.9223192715108715,
      "grad_norm": 0.9361091256141663,
      "learning_rate": 4.653612377984673e-07,
      "loss": 1.1164,
      "step": 10437
    },
    {
      "epoch": 1.9225051105742428,
      "grad_norm": 0.6802321076393127,
      "learning_rate": 4.6254731293118304e-07,
      "loss": 0.9483,
      "step": 10438
    },
    {
      "epoch": 1.9226909496376137,
      "grad_norm": 0.722303032875061,
      "learning_rate": 4.5974190171735874e-07,
      "loss": 1.0042,
      "step": 10439
    },
    {
      "epoch": 1.922876788700985,
      "grad_norm": 0.6165829300880432,
      "learning_rate": 4.5694500439694696e-07,
      "loss": 0.9669,
      "step": 10440
    },
    {
      "epoch": 1.923062627764356,
      "grad_norm": 0.73440021276474,
      "learning_rate": 4.5415662120915637e-07,
      "loss": 0.9426,
      "step": 10441
    },
    {
      "epoch": 1.9232484668277272,
      "grad_norm": 0.7627027630805969,
      "learning_rate": 4.513767523925072e-07,
      "loss": 0.9235,
      "step": 10442
    },
    {
      "epoch": 1.9234343058910983,
      "grad_norm": 0.6763946413993835,
      "learning_rate": 4.486053981847538e-07,
      "loss": 0.863,
      "step": 10443
    },
    {
      "epoch": 1.9236201449544694,
      "grad_norm": 0.7742003202438354,
      "learning_rate": 4.458425588229509e-07,
      "loss": 0.8294,
      "step": 10444
    },
    {
      "epoch": 1.9238059840178405,
      "grad_norm": 0.860358476638794,
      "learning_rate": 4.4308823454338734e-07,
      "loss": 0.9598,
      "step": 10445
    },
    {
      "epoch": 1.9239918230812116,
      "grad_norm": 0.8054762482643127,
      "learning_rate": 4.403424255816635e-07,
      "loss": 1.0493,
      "step": 10446
    },
    {
      "epoch": 1.9241776621445827,
      "grad_norm": 0.7892675399780273,
      "learning_rate": 4.376051321726138e-07,
      "loss": 0.9371,
      "step": 10447
    },
    {
      "epoch": 1.9243635012079539,
      "grad_norm": 0.9061946868896484,
      "learning_rate": 4.3487635455039533e-07,
      "loss": 0.9521,
      "step": 10448
    },
    {
      "epoch": 1.9245493402713252,
      "grad_norm": 0.6500113606452942,
      "learning_rate": 4.321560929483881e-07,
      "loss": 1.1488,
      "step": 10449
    },
    {
      "epoch": 1.924735179334696,
      "grad_norm": 0.6235157251358032,
      "learning_rate": 4.294443475992616e-07,
      "loss": 0.8004,
      "step": 10450
    },
    {
      "epoch": 1.9249210183980674,
      "grad_norm": 0.753731369972229,
      "learning_rate": 4.2674111873496345e-07,
      "loss": 1.1275,
      "step": 10451
    },
    {
      "epoch": 1.9251068574614383,
      "grad_norm": 0.7848681211471558,
      "learning_rate": 4.240464065866978e-07,
      "loss": 0.9641,
      "step": 10452
    },
    {
      "epoch": 1.9252926965248096,
      "grad_norm": 0.7322390079498291,
      "learning_rate": 4.21360211384958e-07,
      "loss": 0.9026,
      "step": 10453
    },
    {
      "epoch": 1.9254785355881805,
      "grad_norm": 0.8562449812889099,
      "learning_rate": 4.1868253335949345e-07,
      "loss": 1.0697,
      "step": 10454
    },
    {
      "epoch": 1.9256643746515518,
      "grad_norm": 0.6389977335929871,
      "learning_rate": 4.160133727393434e-07,
      "loss": 1.0883,
      "step": 10455
    },
    {
      "epoch": 1.925850213714923,
      "grad_norm": 0.7520100474357605,
      "learning_rate": 4.133527297527917e-07,
      "loss": 0.8914,
      "step": 10456
    },
    {
      "epoch": 1.926036052778294,
      "grad_norm": 0.6367651224136353,
      "learning_rate": 4.107006046274231e-07,
      "loss": 0.8606,
      "step": 10457
    },
    {
      "epoch": 1.9262218918416651,
      "grad_norm": 0.7314348220825195,
      "learning_rate": 4.080569975900672e-07,
      "loss": 1.2232,
      "step": 10458
    },
    {
      "epoch": 1.9264077309050363,
      "grad_norm": 0.571670413017273,
      "learning_rate": 4.0542190886684317e-07,
      "loss": 0.9089,
      "step": 10459
    },
    {
      "epoch": 1.9265935699684074,
      "grad_norm": 0.6684804558753967,
      "learning_rate": 4.0279533868314844e-07,
      "loss": 0.7138,
      "step": 10460
    },
    {
      "epoch": 1.9267794090317785,
      "grad_norm": 0.6759762763977051,
      "learning_rate": 4.0017728726360336e-07,
      "loss": 1.0231,
      "step": 10461
    },
    {
      "epoch": 1.9269652480951496,
      "grad_norm": 1.505415439605713,
      "learning_rate": 3.975677548321732e-07,
      "loss": 1.5421,
      "step": 10462
    },
    {
      "epoch": 1.9271510871585207,
      "grad_norm": 0.7095223665237427,
      "learning_rate": 3.949667416120462e-07,
      "loss": 0.8065,
      "step": 10463
    },
    {
      "epoch": 1.9273369262218918,
      "grad_norm": 0.6022650599479675,
      "learning_rate": 3.9237424782567757e-07,
      "loss": 0.8043,
      "step": 10464
    },
    {
      "epoch": 1.927522765285263,
      "grad_norm": 0.6729022264480591,
      "learning_rate": 3.8979027369482333e-07,
      "loss": 0.668,
      "step": 10465
    },
    {
      "epoch": 1.9277086043486342,
      "grad_norm": 0.635015606880188,
      "learning_rate": 3.8721481944048455e-07,
      "loss": 0.8435,
      "step": 10466
    },
    {
      "epoch": 1.927894443412005,
      "grad_norm": 0.8524276614189148,
      "learning_rate": 3.846478852829738e-07,
      "loss": 1.204,
      "step": 10467
    },
    {
      "epoch": 1.9280802824753764,
      "grad_norm": 1.4829657077789307,
      "learning_rate": 3.820894714418044e-07,
      "loss": 1.5791,
      "step": 10468
    },
    {
      "epoch": 1.9282661215387473,
      "grad_norm": 0.7150884866714478,
      "learning_rate": 3.795395781358346e-07,
      "loss": 1.0709,
      "step": 10469
    },
    {
      "epoch": 1.9284519606021187,
      "grad_norm": 0.8258856534957886,
      "learning_rate": 3.7699820558315667e-07,
      "loss": 0.7993,
      "step": 10470
    },
    {
      "epoch": 1.9286377996654895,
      "grad_norm": 0.7824717164039612,
      "learning_rate": 3.7446535400114114e-07,
      "loss": 1.0424,
      "step": 10471
    },
    {
      "epoch": 1.9288236387288609,
      "grad_norm": 0.6867206692695618,
      "learning_rate": 3.719410236064147e-07,
      "loss": 0.978,
      "step": 10472
    },
    {
      "epoch": 1.929009477792232,
      "grad_norm": 0.8157885670661926,
      "learning_rate": 3.6942521461489356e-07,
      "loss": 1.032,
      "step": 10473
    },
    {
      "epoch": 1.929195316855603,
      "grad_norm": 0.7037714123725891,
      "learning_rate": 3.669179272417722e-07,
      "loss": 0.89,
      "step": 10474
    },
    {
      "epoch": 1.9293811559189742,
      "grad_norm": 0.7749755382537842,
      "learning_rate": 3.6441916170149024e-07,
      "loss": 0.9459,
      "step": 10475
    },
    {
      "epoch": 1.9295669949823453,
      "grad_norm": 0.6908509135246277,
      "learning_rate": 3.6192891820777676e-07,
      "loss": 0.9343,
      "step": 10476
    },
    {
      "epoch": 1.9297528340457164,
      "grad_norm": 0.711149275302887,
      "learning_rate": 3.5944719697363907e-07,
      "loss": 0.9214,
      "step": 10477
    },
    {
      "epoch": 1.9299386731090875,
      "grad_norm": 0.7808820009231567,
      "learning_rate": 3.569739982113185e-07,
      "loss": 1.0175,
      "step": 10478
    },
    {
      "epoch": 1.9301245121724586,
      "grad_norm": 0.7938927412033081,
      "learning_rate": 3.5450932213237917e-07,
      "loss": 1.0029,
      "step": 10479
    },
    {
      "epoch": 1.9303103512358297,
      "grad_norm": 0.7212532758712769,
      "learning_rate": 3.520531689476192e-07,
      "loss": 0.9368,
      "step": 10480
    },
    {
      "epoch": 1.930496190299201,
      "grad_norm": 0.7534226179122925,
      "learning_rate": 3.496055388671149e-07,
      "loss": 0.9919,
      "step": 10481
    },
    {
      "epoch": 1.930682029362572,
      "grad_norm": 0.6845839619636536,
      "learning_rate": 3.47166432100221e-07,
      "loss": 0.8314,
      "step": 10482
    },
    {
      "epoch": 1.9308678684259433,
      "grad_norm": 0.7597885131835938,
      "learning_rate": 3.4473584885557075e-07,
      "loss": 1.0507,
      "step": 10483
    },
    {
      "epoch": 1.9310537074893142,
      "grad_norm": 0.6701219081878662,
      "learning_rate": 3.423137893410422e-07,
      "loss": 0.8769,
      "step": 10484
    },
    {
      "epoch": 1.9312395465526855,
      "grad_norm": 0.8467074036598206,
      "learning_rate": 3.399002537637919e-07,
      "loss": 0.788,
      "step": 10485
    },
    {
      "epoch": 1.9314253856160564,
      "grad_norm": 0.823009729385376,
      "learning_rate": 3.3749524233028795e-07,
      "loss": 0.9358,
      "step": 10486
    },
    {
      "epoch": 1.9316112246794277,
      "grad_norm": 0.6456196904182434,
      "learning_rate": 3.3509875524619924e-07,
      "loss": 0.843,
      "step": 10487
    },
    {
      "epoch": 1.9317970637427986,
      "grad_norm": 0.7399866580963135,
      "learning_rate": 3.327107927165174e-07,
      "loss": 0.9594,
      "step": 10488
    },
    {
      "epoch": 1.93198290280617,
      "grad_norm": 0.6855292916297913,
      "learning_rate": 3.3033135494550117e-07,
      "loss": 0.8529,
      "step": 10489
    },
    {
      "epoch": 1.932168741869541,
      "grad_norm": 0.8369126915931702,
      "learning_rate": 3.2796044213665447e-07,
      "loss": 0.7957,
      "step": 10490
    },
    {
      "epoch": 1.9323545809329121,
      "grad_norm": 0.6412609219551086,
      "learning_rate": 3.2559805449277056e-07,
      "loss": 0.8879,
      "step": 10491
    },
    {
      "epoch": 1.9325404199962832,
      "grad_norm": 0.908142626285553,
      "learning_rate": 3.2324419221591017e-07,
      "loss": 1.1453,
      "step": 10492
    },
    {
      "epoch": 1.9327262590596543,
      "grad_norm": 0.6138076186180115,
      "learning_rate": 3.2089885550740106e-07,
      "loss": 0.9155,
      "step": 10493
    },
    {
      "epoch": 1.9329120981230254,
      "grad_norm": 0.7081475853919983,
      "learning_rate": 3.185620445678494e-07,
      "loss": 0.8805,
      "step": 10494
    },
    {
      "epoch": 1.9330979371863966,
      "grad_norm": 0.7727792263031006,
      "learning_rate": 3.1623375959711767e-07,
      "loss": 1.0354,
      "step": 10495
    },
    {
      "epoch": 1.9332837762497677,
      "grad_norm": 0.7606066465377808,
      "learning_rate": 3.139140007943686e-07,
      "loss": 1.1053,
      "step": 10496
    },
    {
      "epoch": 1.9334696153131388,
      "grad_norm": 0.6166344881057739,
      "learning_rate": 3.1160276835799916e-07,
      "loss": 0.8652,
      "step": 10497
    },
    {
      "epoch": 1.93365545437651,
      "grad_norm": 0.6693359613418579,
      "learning_rate": 3.0930006248570675e-07,
      "loss": 0.9455,
      "step": 10498
    },
    {
      "epoch": 1.933841293439881,
      "grad_norm": 0.9655212163925171,
      "learning_rate": 3.070058833744338e-07,
      "loss": 1.1453,
      "step": 10499
    },
    {
      "epoch": 1.9340271325032523,
      "grad_norm": 0.767994225025177,
      "learning_rate": 3.0472023122041225e-07,
      "loss": 0.768,
      "step": 10500
    },
    {
      "epoch": 1.9342129715666232,
      "grad_norm": 0.8077344298362732,
      "learning_rate": 3.024431062191413e-07,
      "loss": 1.139,
      "step": 10501
    },
    {
      "epoch": 1.9343988106299945,
      "grad_norm": 0.7413500547409058,
      "learning_rate": 3.001745085653873e-07,
      "loss": 1.0268,
      "step": 10502
    },
    {
      "epoch": 1.9345846496933654,
      "grad_norm": 1.358109474182129,
      "learning_rate": 2.9791443845319514e-07,
      "loss": 1.4329,
      "step": 10503
    },
    {
      "epoch": 1.9347704887567367,
      "grad_norm": 0.674161970615387,
      "learning_rate": 2.956628960758545e-07,
      "loss": 0.8224,
      "step": 10504
    },
    {
      "epoch": 1.9349563278201076,
      "grad_norm": 0.9352040886878967,
      "learning_rate": 2.934198816259559e-07,
      "loss": 0.8062,
      "step": 10505
    },
    {
      "epoch": 1.935142166883479,
      "grad_norm": 0.8096712231636047,
      "learning_rate": 2.9118539529534586e-07,
      "loss": 0.9979,
      "step": 10506
    },
    {
      "epoch": 1.93532800594685,
      "grad_norm": 0.708527684211731,
      "learning_rate": 2.889594372751603e-07,
      "loss": 1.0437,
      "step": 10507
    },
    {
      "epoch": 1.9355138450102212,
      "grad_norm": 0.7208974361419678,
      "learning_rate": 2.867420077557692e-07,
      "loss": 1.1676,
      "step": 10508
    },
    {
      "epoch": 1.9356996840735923,
      "grad_norm": 0.8035686016082764,
      "learning_rate": 2.845331069268431e-07,
      "loss": 1.0673,
      "step": 10509
    },
    {
      "epoch": 1.9358855231369634,
      "grad_norm": 0.7524265050888062,
      "learning_rate": 2.823327349773197e-07,
      "loss": 0.8933,
      "step": 10510
    },
    {
      "epoch": 1.9360713622003345,
      "grad_norm": 0.5901182293891907,
      "learning_rate": 2.8014089209538186e-07,
      "loss": 0.9226,
      "step": 10511
    },
    {
      "epoch": 1.9362572012637056,
      "grad_norm": 0.8229266405105591,
      "learning_rate": 2.779575784685351e-07,
      "loss": 1.1362,
      "step": 10512
    },
    {
      "epoch": 1.9364430403270767,
      "grad_norm": 0.8021316528320312,
      "learning_rate": 2.7578279428349676e-07,
      "loss": 1.1076,
      "step": 10513
    },
    {
      "epoch": 1.9366288793904478,
      "grad_norm": 0.6480070948600769,
      "learning_rate": 2.7361653972628467e-07,
      "loss": 0.9693,
      "step": 10514
    },
    {
      "epoch": 1.9368147184538191,
      "grad_norm": 0.6520128846168518,
      "learning_rate": 2.7145881498218395e-07,
      "loss": 0.9093,
      "step": 10515
    },
    {
      "epoch": 1.93700055751719,
      "grad_norm": 0.7292173504829407,
      "learning_rate": 2.693096202357692e-07,
      "loss": 1.0411,
      "step": 10516
    },
    {
      "epoch": 1.9371863965805614,
      "grad_norm": 0.8277835249900818,
      "learning_rate": 2.6716895567083786e-07,
      "loss": 0.9591,
      "step": 10517
    },
    {
      "epoch": 1.9373722356439322,
      "grad_norm": 0.7037623524665833,
      "learning_rate": 2.650368214704879e-07,
      "loss": 0.8856,
      "step": 10518
    },
    {
      "epoch": 1.9375580747073036,
      "grad_norm": 0.9497572779655457,
      "learning_rate": 2.629132178171068e-07,
      "loss": 1.115,
      "step": 10519
    },
    {
      "epoch": 1.9377439137706745,
      "grad_norm": 0.686296284198761,
      "learning_rate": 2.607981448923158e-07,
      "loss": 0.9223,
      "step": 10520
    },
    {
      "epoch": 1.9379297528340458,
      "grad_norm": 0.7315409779548645,
      "learning_rate": 2.586916028770259e-07,
      "loss": 0.7553,
      "step": 10521
    },
    {
      "epoch": 1.938115591897417,
      "grad_norm": 0.7983646392822266,
      "learning_rate": 2.56593591951404e-07,
      "loss": 0.9392,
      "step": 10522
    },
    {
      "epoch": 1.938301430960788,
      "grad_norm": 0.6945982575416565,
      "learning_rate": 2.5450411229491765e-07,
      "loss": 0.5498,
      "step": 10523
    },
    {
      "epoch": 1.938487270024159,
      "grad_norm": 0.5849335789680481,
      "learning_rate": 2.524231640862684e-07,
      "loss": 0.8544,
      "step": 10524
    },
    {
      "epoch": 1.9386731090875302,
      "grad_norm": 0.568228542804718,
      "learning_rate": 2.5035074750344724e-07,
      "loss": 0.6332,
      "step": 10525
    },
    {
      "epoch": 1.9388589481509013,
      "grad_norm": 0.7203419804573059,
      "learning_rate": 2.4828686272371226e-07,
      "loss": 1.0255,
      "step": 10526
    },
    {
      "epoch": 1.9390447872142724,
      "grad_norm": 0.741192102432251,
      "learning_rate": 2.46231509923589e-07,
      "loss": 0.8389,
      "step": 10527
    },
    {
      "epoch": 1.9392306262776435,
      "grad_norm": 0.6687784194946289,
      "learning_rate": 2.441846892788924e-07,
      "loss": 0.9354,
      "step": 10528
    },
    {
      "epoch": 1.9394164653410146,
      "grad_norm": 0.639785885810852,
      "learning_rate": 2.421464009646712e-07,
      "loss": 0.9714,
      "step": 10529
    },
    {
      "epoch": 1.9396023044043857,
      "grad_norm": 0.8041900992393494,
      "learning_rate": 2.401166451552861e-07,
      "loss": 0.8889,
      "step": 10530
    },
    {
      "epoch": 1.9397881434677569,
      "grad_norm": 0.8194205164909363,
      "learning_rate": 2.3809542202432034e-07,
      "loss": 1.0706,
      "step": 10531
    },
    {
      "epoch": 1.9399739825311282,
      "grad_norm": 0.6742299199104309,
      "learning_rate": 2.36082731744669e-07,
      "loss": 1.085,
      "step": 10532
    },
    {
      "epoch": 1.940159821594499,
      "grad_norm": 0.9145841598510742,
      "learning_rate": 2.3407857448848324e-07,
      "loss": 1.016,
      "step": 10533
    },
    {
      "epoch": 1.9403456606578704,
      "grad_norm": 0.7550665140151978,
      "learning_rate": 2.3208295042718154e-07,
      "loss": 1.0904,
      "step": 10534
    },
    {
      "epoch": 1.9405314997212413,
      "grad_norm": 0.7545983791351318,
      "learning_rate": 2.3009585973144955e-07,
      "loss": 0.8752,
      "step": 10535
    },
    {
      "epoch": 1.9407173387846126,
      "grad_norm": 0.8406251072883606,
      "learning_rate": 2.281173025712513e-07,
      "loss": 0.9596,
      "step": 10536
    },
    {
      "epoch": 1.9409031778479835,
      "grad_norm": 0.6592369079589844,
      "learning_rate": 2.2614727911581812e-07,
      "loss": 0.928,
      "step": 10537
    },
    {
      "epoch": 1.9410890169113548,
      "grad_norm": 0.7806558012962341,
      "learning_rate": 2.2418578953363745e-07,
      "loss": 1.0567,
      "step": 10538
    },
    {
      "epoch": 1.941274855974726,
      "grad_norm": 0.6726809740066528,
      "learning_rate": 2.2223283399249728e-07,
      "loss": 1.0418,
      "step": 10539
    },
    {
      "epoch": 1.941460695038097,
      "grad_norm": 0.7796017527580261,
      "learning_rate": 2.2028841265943067e-07,
      "loss": 1.0132,
      "step": 10540
    },
    {
      "epoch": 1.9416465341014681,
      "grad_norm": 0.5700896382331848,
      "learning_rate": 2.1835252570074905e-07,
      "loss": 0.7862,
      "step": 10541
    },
    {
      "epoch": 1.9418323731648393,
      "grad_norm": 0.7132280468940735,
      "learning_rate": 2.1642517328203105e-07,
      "loss": 0.9926,
      "step": 10542
    },
    {
      "epoch": 1.9420182122282104,
      "grad_norm": 0.7122735977172852,
      "learning_rate": 2.145063555681226e-07,
      "loss": 0.771,
      "step": 10543
    },
    {
      "epoch": 1.9422040512915815,
      "grad_norm": 0.6890891194343567,
      "learning_rate": 2.1259607272315907e-07,
      "loss": 0.9131,
      "step": 10544
    },
    {
      "epoch": 1.9423898903549526,
      "grad_norm": 0.7969406843185425,
      "learning_rate": 2.1069432491052088e-07,
      "loss": 0.7925,
      "step": 10545
    },
    {
      "epoch": 1.9425757294183237,
      "grad_norm": 0.6591358780860901,
      "learning_rate": 2.088011122928668e-07,
      "loss": 0.957,
      "step": 10546
    },
    {
      "epoch": 1.942761568481695,
      "grad_norm": 0.7907562851905823,
      "learning_rate": 2.0691643503213398e-07,
      "loss": 0.983,
      "step": 10547
    },
    {
      "epoch": 1.942947407545066,
      "grad_norm": 0.5965955853462219,
      "learning_rate": 2.050402932895157e-07,
      "loss": 0.7991,
      "step": 10548
    },
    {
      "epoch": 1.9431332466084372,
      "grad_norm": 0.6739276051521301,
      "learning_rate": 2.0317268722549464e-07,
      "loss": 0.943,
      "step": 10549
    },
    {
      "epoch": 1.943319085671808,
      "grad_norm": 0.6748188138008118,
      "learning_rate": 2.0131361699979868e-07,
      "loss": 0.9612,
      "step": 10550
    },
    {
      "epoch": 1.9435049247351794,
      "grad_norm": 0.7416154146194458,
      "learning_rate": 1.99463082771445e-07,
      "loss": 0.8746,
      "step": 10551
    },
    {
      "epoch": 1.9436907637985503,
      "grad_norm": 0.7586497068405151,
      "learning_rate": 1.97621084698707e-07,
      "loss": 0.8507,
      "step": 10552
    },
    {
      "epoch": 1.9438766028619217,
      "grad_norm": 0.788551926612854,
      "learning_rate": 1.9578762293913644e-07,
      "loss": 0.9915,
      "step": 10553
    },
    {
      "epoch": 1.9440624419252925,
      "grad_norm": 0.7596179246902466,
      "learning_rate": 1.9396269764956343e-07,
      "loss": 0.8605,
      "step": 10554
    },
    {
      "epoch": 1.9442482809886639,
      "grad_norm": 0.7364116311073303,
      "learning_rate": 1.9214630898606312e-07,
      "loss": 0.8314,
      "step": 10555
    },
    {
      "epoch": 1.944434120052035,
      "grad_norm": 2.9015426635742188,
      "learning_rate": 1.9033845710400012e-07,
      "loss": 1.5259,
      "step": 10556
    },
    {
      "epoch": 1.944619959115406,
      "grad_norm": 0.6181079149246216,
      "learning_rate": 1.885391421580174e-07,
      "loss": 0.7021,
      "step": 10557
    },
    {
      "epoch": 1.9448057981787772,
      "grad_norm": 0.670671284198761,
      "learning_rate": 1.8674836430199183e-07,
      "loss": 1.0022,
      "step": 10558
    },
    {
      "epoch": 1.9449916372421483,
      "grad_norm": 0.7670441269874573,
      "learning_rate": 1.8496612368911203e-07,
      "loss": 0.8757,
      "step": 10559
    },
    {
      "epoch": 1.9451774763055194,
      "grad_norm": 0.7375046014785767,
      "learning_rate": 1.831924204718005e-07,
      "loss": 0.8354,
      "step": 10560
    },
    {
      "epoch": 1.9453633153688905,
      "grad_norm": 0.7295529842376709,
      "learning_rate": 1.8142725480178037e-07,
      "loss": 0.9892,
      "step": 10561
    },
    {
      "epoch": 1.9455491544322616,
      "grad_norm": 0.7388021945953369,
      "learning_rate": 1.7967062683001967e-07,
      "loss": 1.0299,
      "step": 10562
    },
    {
      "epoch": 1.9457349934956327,
      "grad_norm": 0.5903481245040894,
      "learning_rate": 1.7792253670676496e-07,
      "loss": 0.9985,
      "step": 10563
    },
    {
      "epoch": 1.945920832559004,
      "grad_norm": 0.5798445343971252,
      "learning_rate": 1.7618298458155213e-07,
      "loss": 0.5386,
      "step": 10564
    },
    {
      "epoch": 1.946106671622375,
      "grad_norm": 0.6595876812934875,
      "learning_rate": 1.744519706031511e-07,
      "loss": 0.8605,
      "step": 10565
    },
    {
      "epoch": 1.9462925106857463,
      "grad_norm": 0.5831649303436279,
      "learning_rate": 1.727294949196212e-07,
      "loss": 0.8318,
      "step": 10566
    },
    {
      "epoch": 1.9464783497491172,
      "grad_norm": 0.7746573686599731,
      "learning_rate": 1.7101555767828904e-07,
      "loss": 1.048,
      "step": 10567
    },
    {
      "epoch": 1.9466641888124885,
      "grad_norm": 0.7500365376472473,
      "learning_rate": 1.6931015902575954e-07,
      "loss": 1.1274,
      "step": 10568
    },
    {
      "epoch": 1.9468500278758594,
      "grad_norm": 0.7787057757377625,
      "learning_rate": 1.676132991079049e-07,
      "loss": 0.918,
      "step": 10569
    },
    {
      "epoch": 1.9470358669392307,
      "grad_norm": 0.6510689854621887,
      "learning_rate": 1.6592497806983133e-07,
      "loss": 0.7985,
      "step": 10570
    },
    {
      "epoch": 1.9472217060026016,
      "grad_norm": 0.704802393913269,
      "learning_rate": 1.6424519605597876e-07,
      "loss": 0.8693,
      "step": 10571
    },
    {
      "epoch": 1.947407545065973,
      "grad_norm": 0.7769961953163147,
      "learning_rate": 1.6257395320999903e-07,
      "loss": 0.9472,
      "step": 10572
    },
    {
      "epoch": 1.947593384129344,
      "grad_norm": 0.7937042713165283,
      "learning_rate": 1.6091124967485548e-07,
      "loss": 0.7564,
      "step": 10573
    },
    {
      "epoch": 1.9477792231927151,
      "grad_norm": 0.7213950157165527,
      "learning_rate": 1.5925708559274556e-07,
      "loss": 1.1245,
      "step": 10574
    },
    {
      "epoch": 1.9479650622560862,
      "grad_norm": 0.7393320798873901,
      "learning_rate": 1.576114611051671e-07,
      "loss": 1.0531,
      "step": 10575
    },
    {
      "epoch": 1.9481509013194573,
      "grad_norm": 0.6105191111564636,
      "learning_rate": 1.5597437635286316e-07,
      "loss": 0.8292,
      "step": 10576
    },
    {
      "epoch": 1.9483367403828284,
      "grad_norm": 0.7339911460876465,
      "learning_rate": 1.5434583147586612e-07,
      "loss": 0.9979,
      "step": 10577
    },
    {
      "epoch": 1.9485225794461996,
      "grad_norm": 0.7356317043304443,
      "learning_rate": 1.5272582661346458e-07,
      "loss": 0.9202,
      "step": 10578
    },
    {
      "epoch": 1.9487084185095707,
      "grad_norm": 0.6233305335044861,
      "learning_rate": 1.5111436190422546e-07,
      "loss": 0.7786,
      "step": 10579
    },
    {
      "epoch": 1.9488942575729418,
      "grad_norm": 0.6189817786216736,
      "learning_rate": 1.4951143748597186e-07,
      "loss": 0.7014,
      "step": 10580
    },
    {
      "epoch": 1.949080096636313,
      "grad_norm": 0.7941526174545288,
      "learning_rate": 1.479170534958052e-07,
      "loss": 0.9507,
      "step": 10581
    },
    {
      "epoch": 1.949265935699684,
      "grad_norm": 0.6824877262115479,
      "learning_rate": 1.4633121007011642e-07,
      "loss": 1.1483,
      "step": 10582
    },
    {
      "epoch": 1.9494517747630553,
      "grad_norm": 0.8881486654281616,
      "learning_rate": 1.4475390734451922e-07,
      "loss": 0.9177,
      "step": 10583
    },
    {
      "epoch": 1.9496376138264262,
      "grad_norm": 0.6779103875160217,
      "learning_rate": 1.4318514545393903e-07,
      "loss": 1.0484,
      "step": 10584
    },
    {
      "epoch": 1.9498234528897975,
      "grad_norm": 0.66915363073349,
      "learning_rate": 1.4162492453254628e-07,
      "loss": 0.6762,
      "step": 10585
    },
    {
      "epoch": 1.9500092919531684,
      "grad_norm": 0.6980171799659729,
      "learning_rate": 1.400732447138009e-07,
      "loss": 0.8928,
      "step": 10586
    },
    {
      "epoch": 1.9501951310165397,
      "grad_norm": 0.7655068039894104,
      "learning_rate": 1.3853010613041896e-07,
      "loss": 0.9839,
      "step": 10587
    },
    {
      "epoch": 1.9503809700799108,
      "grad_norm": 0.7359232902526855,
      "learning_rate": 1.3699550891438372e-07,
      "loss": 1.0138,
      "step": 10588
    },
    {
      "epoch": 1.950566809143282,
      "grad_norm": 0.6629403829574585,
      "learning_rate": 1.3546945319695693e-07,
      "loss": 0.8499,
      "step": 10589
    },
    {
      "epoch": 1.950752648206653,
      "grad_norm": 0.8161008954048157,
      "learning_rate": 1.3395193910866744e-07,
      "loss": 1.0079,
      "step": 10590
    },
    {
      "epoch": 1.9509384872700242,
      "grad_norm": 0.640409529209137,
      "learning_rate": 1.3244296677930035e-07,
      "loss": 0.8732,
      "step": 10591
    },
    {
      "epoch": 1.9511243263333953,
      "grad_norm": 0.6880654692649841,
      "learning_rate": 1.309425363379302e-07,
      "loss": 1.0767,
      "step": 10592
    },
    {
      "epoch": 1.9513101653967664,
      "grad_norm": 0.8354215621948242,
      "learning_rate": 1.2945064791288764e-07,
      "loss": 1.0845,
      "step": 10593
    },
    {
      "epoch": 1.9514960044601375,
      "grad_norm": 0.6961938738822937,
      "learning_rate": 1.279673016317817e-07,
      "loss": 0.9278,
      "step": 10594
    },
    {
      "epoch": 1.9516818435235086,
      "grad_norm": 1.2774786949157715,
      "learning_rate": 1.2649249762148874e-07,
      "loss": 1.3529,
      "step": 10595
    },
    {
      "epoch": 1.95186768258688,
      "grad_norm": 0.7045060992240906,
      "learning_rate": 1.250262360081411e-07,
      "loss": 0.9794,
      "step": 10596
    },
    {
      "epoch": 1.9520535216502508,
      "grad_norm": 0.7024220824241638,
      "learning_rate": 1.2356851691716077e-07,
      "loss": 1.1276,
      "step": 10597
    },
    {
      "epoch": 1.9522393607136221,
      "grad_norm": 0.6957845091819763,
      "learning_rate": 1.221193404732257e-07,
      "loss": 0.8501,
      "step": 10598
    },
    {
      "epoch": 1.952425199776993,
      "grad_norm": 0.7334973216056824,
      "learning_rate": 1.2067870680029236e-07,
      "loss": 1.0198,
      "step": 10599
    },
    {
      "epoch": 1.9526110388403644,
      "grad_norm": 0.8500604033470154,
      "learning_rate": 1.1924661602157329e-07,
      "loss": 0.8819,
      "step": 10600
    },
    {
      "epoch": 1.9527968779037352,
      "grad_norm": 0.7799627780914307,
      "learning_rate": 1.1782306825955935e-07,
      "loss": 1.0159,
      "step": 10601
    },
    {
      "epoch": 1.9529827169671066,
      "grad_norm": 0.6954518556594849,
      "learning_rate": 1.1640806363601986e-07,
      "loss": 0.9773,
      "step": 10602
    },
    {
      "epoch": 1.9531685560304775,
      "grad_norm": 0.6808642745018005,
      "learning_rate": 1.150016022719691e-07,
      "loss": 0.9727,
      "step": 10603
    },
    {
      "epoch": 1.9533543950938488,
      "grad_norm": 0.7429071068763733,
      "learning_rate": 1.1360368428771084e-07,
      "loss": 0.8981,
      "step": 10604
    },
    {
      "epoch": 1.95354023415722,
      "grad_norm": 0.7698872089385986,
      "learning_rate": 1.122143098028161e-07,
      "loss": 0.8609,
      "step": 10605
    },
    {
      "epoch": 1.953726073220591,
      "grad_norm": 0.6500043272972107,
      "learning_rate": 1.1083347893611206e-07,
      "loss": 0.7626,
      "step": 10606
    },
    {
      "epoch": 1.953911912283962,
      "grad_norm": 0.7074955105781555,
      "learning_rate": 1.0946119180571535e-07,
      "loss": 1.18,
      "step": 10607
    },
    {
      "epoch": 1.9540977513473332,
      "grad_norm": 0.7872495055198669,
      "learning_rate": 1.0809744852898762e-07,
      "loss": 1.2688,
      "step": 10608
    },
    {
      "epoch": 1.9542835904107043,
      "grad_norm": 0.7019426822662354,
      "learning_rate": 1.0674224922258003e-07,
      "loss": 0.9837,
      "step": 10609
    },
    {
      "epoch": 1.9544694294740754,
      "grad_norm": 0.6763778328895569,
      "learning_rate": 1.0539559400242205e-07,
      "loss": 0.8063,
      "step": 10610
    },
    {
      "epoch": 1.9546552685374465,
      "grad_norm": 1.0150338411331177,
      "learning_rate": 1.0405748298366602e-07,
      "loss": 0.9567,
      "step": 10611
    },
    {
      "epoch": 1.9548411076008176,
      "grad_norm": 0.8957339525222778,
      "learning_rate": 1.0272791628077594e-07,
      "loss": 1.0751,
      "step": 10612
    },
    {
      "epoch": 1.955026946664189,
      "grad_norm": 0.6664177775382996,
      "learning_rate": 1.0140689400747194e-07,
      "loss": 1.121,
      "step": 10613
    },
    {
      "epoch": 1.9552127857275599,
      "grad_norm": 0.6629258394241333,
      "learning_rate": 1.0009441627675254e-07,
      "loss": 0.7483,
      "step": 10614
    },
    {
      "epoch": 1.9553986247909312,
      "grad_norm": 0.8004807233810425,
      "learning_rate": 9.879048320086126e-08,
      "loss": 0.9074,
      "step": 10615
    },
    {
      "epoch": 1.955584463854302,
      "grad_norm": 0.5369228720664978,
      "learning_rate": 9.749509489134223e-08,
      "loss": 0.4898,
      "step": 10616
    },
    {
      "epoch": 1.9557703029176734,
      "grad_norm": 0.7646516561508179,
      "learning_rate": 9.62082514589846e-08,
      "loss": 0.8515,
      "step": 10617
    },
    {
      "epoch": 1.9559561419810443,
      "grad_norm": 0.7015865445137024,
      "learning_rate": 9.492995301385588e-08,
      "loss": 0.9903,
      "step": 10618
    },
    {
      "epoch": 1.9561419810444156,
      "grad_norm": 1.3769493103027344,
      "learning_rate": 9.366019966529083e-08,
      "loss": 1.3636,
      "step": 10619
    },
    {
      "epoch": 1.9563278201077865,
      "grad_norm": 0.6428963541984558,
      "learning_rate": 9.239899152189146e-08,
      "loss": 0.9718,
      "step": 10620
    },
    {
      "epoch": 1.9565136591711578,
      "grad_norm": 0.6724200248718262,
      "learning_rate": 9.114632869153816e-08,
      "loss": 1.0401,
      "step": 10621
    },
    {
      "epoch": 1.956699498234529,
      "grad_norm": 0.8284958600997925,
      "learning_rate": 8.990221128137854e-08,
      "loss": 1.0784,
      "step": 10622
    },
    {
      "epoch": 1.9568853372979,
      "grad_norm": 0.8442558646202087,
      "learning_rate": 8.866663939780529e-08,
      "loss": 0.9158,
      "step": 10623
    },
    {
      "epoch": 1.9570711763612711,
      "grad_norm": 0.8485099673271179,
      "learning_rate": 8.743961314651161e-08,
      "loss": 1.0308,
      "step": 10624
    },
    {
      "epoch": 1.9572570154246423,
      "grad_norm": 0.7097102999687195,
      "learning_rate": 8.622113263245801e-08,
      "loss": 0.8732,
      "step": 10625
    },
    {
      "epoch": 1.9574428544880134,
      "grad_norm": 0.7192796468734741,
      "learning_rate": 8.501119795983892e-08,
      "loss": 0.9586,
      "step": 10626
    },
    {
      "epoch": 1.9576286935513845,
      "grad_norm": 0.7567042112350464,
      "learning_rate": 8.380980923217152e-08,
      "loss": 1.0679,
      "step": 10627
    },
    {
      "epoch": 1.9578145326147556,
      "grad_norm": 0.7260935306549072,
      "learning_rate": 8.261696655218476e-08,
      "loss": 0.9328,
      "step": 10628
    },
    {
      "epoch": 1.9580003716781267,
      "grad_norm": 0.6080493330955505,
      "learning_rate": 8.143267002193034e-08,
      "loss": 0.8633,
      "step": 10629
    },
    {
      "epoch": 1.958186210741498,
      "grad_norm": 0.6445833444595337,
      "learning_rate": 8.025691974268279e-08,
      "loss": 0.9214,
      "step": 10630
    },
    {
      "epoch": 1.958372049804869,
      "grad_norm": 0.640957236289978,
      "learning_rate": 7.908971581501723e-08,
      "loss": 0.8789,
      "step": 10631
    },
    {
      "epoch": 1.9585578888682402,
      "grad_norm": 0.7331262230873108,
      "learning_rate": 7.793105833876491e-08,
      "loss": 1.0869,
      "step": 10632
    },
    {
      "epoch": 1.9587437279316111,
      "grad_norm": 0.9871970415115356,
      "learning_rate": 7.678094741303543e-08,
      "loss": 0.8037,
      "step": 10633
    },
    {
      "epoch": 1.9589295669949824,
      "grad_norm": 0.6546254754066467,
      "learning_rate": 7.563938313619457e-08,
      "loss": 0.9437,
      "step": 10634
    },
    {
      "epoch": 1.9591154060583533,
      "grad_norm": 0.6918784976005554,
      "learning_rate": 7.450636560587531e-08,
      "loss": 0.8262,
      "step": 10635
    },
    {
      "epoch": 1.9593012451217247,
      "grad_norm": 0.6856305599212646,
      "learning_rate": 7.338189491900016e-08,
      "loss": 0.7126,
      "step": 10636
    },
    {
      "epoch": 1.9594870841850958,
      "grad_norm": 0.6383512616157532,
      "learning_rate": 7.226597117173661e-08,
      "loss": 0.8585,
      "step": 10637
    },
    {
      "epoch": 1.9596729232484669,
      "grad_norm": 1.4464577436447144,
      "learning_rate": 7.115859445954165e-08,
      "loss": 1.3592,
      "step": 10638
    },
    {
      "epoch": 1.959858762311838,
      "grad_norm": 0.7710501551628113,
      "learning_rate": 7.005976487711729e-08,
      "loss": 1.1809,
      "step": 10639
    },
    {
      "epoch": 1.960044601375209,
      "grad_norm": 0.6691771149635315,
      "learning_rate": 6.896948251846613e-08,
      "loss": 0.9255,
      "step": 10640
    },
    {
      "epoch": 1.9602304404385802,
      "grad_norm": 0.7149869203567505,
      "learning_rate": 6.78877474768358e-08,
      "loss": 0.9247,
      "step": 10641
    },
    {
      "epoch": 1.9604162795019513,
      "grad_norm": 0.8322938084602356,
      "learning_rate": 6.681455984474116e-08,
      "loss": 1.0775,
      "step": 10642
    },
    {
      "epoch": 1.9606021185653224,
      "grad_norm": 0.7335876822471619,
      "learning_rate": 6.574991971398658e-08,
      "loss": 0.7026,
      "step": 10643
    },
    {
      "epoch": 1.9607879576286935,
      "grad_norm": 0.7936712503433228,
      "learning_rate": 6.469382717563255e-08,
      "loss": 0.8784,
      "step": 10644
    },
    {
      "epoch": 1.9609737966920646,
      "grad_norm": 0.7369674444198608,
      "learning_rate": 6.36462823199957e-08,
      "loss": 1.0884,
      "step": 10645
    },
    {
      "epoch": 1.9611596357554357,
      "grad_norm": 0.8778194189071655,
      "learning_rate": 6.260728523668213e-08,
      "loss": 0.8975,
      "step": 10646
    },
    {
      "epoch": 1.961345474818807,
      "grad_norm": 0.6566877961158752,
      "learning_rate": 6.15768360145652e-08,
      "loss": 0.8484,
      "step": 10647
    },
    {
      "epoch": 1.961531313882178,
      "grad_norm": 0.690839409828186,
      "learning_rate": 6.055493474177443e-08,
      "loss": 0.9666,
      "step": 10648
    },
    {
      "epoch": 1.9617171529455493,
      "grad_norm": 0.7134891748428345,
      "learning_rate": 5.9541581505717646e-08,
      "loss": 0.722,
      "step": 10649
    },
    {
      "epoch": 1.9619029920089202,
      "grad_norm": 0.7135552763938904,
      "learning_rate": 5.853677639306998e-08,
      "loss": 0.8176,
      "step": 10650
    },
    {
      "epoch": 1.9620888310722915,
      "grad_norm": 0.7070262432098389,
      "learning_rate": 5.7540519489773794e-08,
      "loss": 0.8338,
      "step": 10651
    },
    {
      "epoch": 1.9622746701356624,
      "grad_norm": 0.7059808373451233,
      "learning_rate": 5.65528108810498e-08,
      "loss": 0.7998,
      "step": 10652
    },
    {
      "epoch": 1.9624605091990337,
      "grad_norm": 0.6993285417556763,
      "learning_rate": 5.557365065135267e-08,
      "loss": 0.9615,
      "step": 10653
    },
    {
      "epoch": 1.9626463482624048,
      "grad_norm": 0.782971203327179,
      "learning_rate": 5.4603038884459834e-08,
      "loss": 0.9966,
      "step": 10654
    },
    {
      "epoch": 1.962832187325776,
      "grad_norm": 0.9900389313697815,
      "learning_rate": 5.3640975663382664e-08,
      "loss": 1.1481,
      "step": 10655
    },
    {
      "epoch": 1.963018026389147,
      "grad_norm": 0.7023151516914368,
      "learning_rate": 5.268746107039979e-08,
      "loss": 1.0139,
      "step": 10656
    },
    {
      "epoch": 1.9632038654525181,
      "grad_norm": 0.6240903735160828,
      "learning_rate": 5.17424951870793e-08,
      "loss": 1.055,
      "step": 10657
    },
    {
      "epoch": 1.9633897045158892,
      "grad_norm": 0.7413076162338257,
      "learning_rate": 5.0806078094234325e-08,
      "loss": 0.8608,
      "step": 10658
    },
    {
      "epoch": 1.9635755435792603,
      "grad_norm": 0.6608368158340454,
      "learning_rate": 4.987820987195635e-08,
      "loss": 0.9219,
      "step": 10659
    },
    {
      "epoch": 1.9637613826426314,
      "grad_norm": 1.0145299434661865,
      "learning_rate": 4.895889059962633e-08,
      "loss": 1.0964,
      "step": 10660
    },
    {
      "epoch": 1.9639472217060026,
      "grad_norm": 0.6843851208686829,
      "learning_rate": 4.804812035585915e-08,
      "loss": 0.9641,
      "step": 10661
    },
    {
      "epoch": 1.9641330607693739,
      "grad_norm": 0.665683925151825,
      "learning_rate": 4.714589921857027e-08,
      "loss": 0.8979,
      "step": 10662
    },
    {
      "epoch": 1.9643188998327448,
      "grad_norm": 1.0688481330871582,
      "learning_rate": 4.625222726490908e-08,
      "loss": 1.2587,
      "step": 10663
    },
    {
      "epoch": 1.964504738896116,
      "grad_norm": 0.6237012147903442,
      "learning_rate": 4.5367104571325535e-08,
      "loss": 0.6693,
      "step": 10664
    },
    {
      "epoch": 1.964690577959487,
      "grad_norm": 0.6066529750823975,
      "learning_rate": 4.4490531213525754e-08,
      "loss": 0.7855,
      "step": 10665
    },
    {
      "epoch": 1.9648764170228583,
      "grad_norm": 0.761724054813385,
      "learning_rate": 4.3622507266483095e-08,
      "loss": 0.7746,
      "step": 10666
    },
    {
      "epoch": 1.9650622560862292,
      "grad_norm": 0.7549482583999634,
      "learning_rate": 4.2763032804449267e-08,
      "loss": 1.0191,
      "step": 10667
    },
    {
      "epoch": 1.9652480951496005,
      "grad_norm": 0.8386363387107849,
      "learning_rate": 4.191210790092104e-08,
      "loss": 0.8702,
      "step": 10668
    },
    {
      "epoch": 1.9654339342129714,
      "grad_norm": 0.7027541995048523,
      "learning_rate": 4.106973262868463e-08,
      "loss": 1.0075,
      "step": 10669
    },
    {
      "epoch": 1.9656197732763427,
      "grad_norm": 0.6551834940910339,
      "learning_rate": 4.023590705979352e-08,
      "loss": 0.8276,
      "step": 10670
    },
    {
      "epoch": 1.9658056123397138,
      "grad_norm": 0.6328686475753784,
      "learning_rate": 3.941063126557954e-08,
      "loss": 0.9006,
      "step": 10671
    },
    {
      "epoch": 1.965991451403085,
      "grad_norm": 0.6727689504623413,
      "learning_rate": 3.859390531660845e-08,
      "loss": 0.9283,
      "step": 10672
    },
    {
      "epoch": 1.966177290466456,
      "grad_norm": 0.6957010626792908,
      "learning_rate": 3.7785729282746594e-08,
      "loss": 0.9045,
      "step": 10673
    },
    {
      "epoch": 1.9663631295298272,
      "grad_norm": 0.6626486778259277,
      "learning_rate": 3.6986103233116465e-08,
      "loss": 0.9211,
      "step": 10674
    },
    {
      "epoch": 1.9665489685931983,
      "grad_norm": 0.6777610182762146,
      "learning_rate": 3.619502723610779e-08,
      "loss": 0.9478,
      "step": 10675
    },
    {
      "epoch": 1.9667348076565694,
      "grad_norm": 0.6472916603088379,
      "learning_rate": 3.5412501359399776e-08,
      "loss": 0.9344,
      "step": 10676
    },
    {
      "epoch": 1.9669206467199405,
      "grad_norm": 0.7133098840713501,
      "learning_rate": 3.4638525669905554e-08,
      "loss": 1.0665,
      "step": 10677
    },
    {
      "epoch": 1.9671064857833116,
      "grad_norm": 0.703316867351532,
      "learning_rate": 3.387310023383883e-08,
      "loss": 0.9378,
      "step": 10678
    },
    {
      "epoch": 1.967292324846683,
      "grad_norm": 0.777379035949707,
      "learning_rate": 3.311622511665835e-08,
      "loss": 0.8881,
      "step": 10679
    },
    {
      "epoch": 1.9674781639100538,
      "grad_norm": 0.8485547304153442,
      "learning_rate": 3.2367900383101227e-08,
      "loss": 1.0767,
      "step": 10680
    },
    {
      "epoch": 1.9676640029734251,
      "grad_norm": 0.7185204029083252,
      "learning_rate": 3.162812609717181e-08,
      "loss": 0.8991,
      "step": 10681
    },
    {
      "epoch": 1.967849842036796,
      "grad_norm": 0.774558424949646,
      "learning_rate": 3.089690232215281e-08,
      "loss": 0.7413,
      "step": 10682
    },
    {
      "epoch": 1.9680356811001674,
      "grad_norm": 0.6125521659851074,
      "learning_rate": 3.01742291205831e-08,
      "loss": 0.8836,
      "step": 10683
    },
    {
      "epoch": 1.9682215201635382,
      "grad_norm": 0.6851577162742615,
      "learning_rate": 2.9460106554279886e-08,
      "loss": 0.7599,
      "step": 10684
    },
    {
      "epoch": 1.9684073592269096,
      "grad_norm": 0.8051230311393738,
      "learning_rate": 2.8754534684316547e-08,
      "loss": 1.0482,
      "step": 10685
    },
    {
      "epoch": 1.9685931982902805,
      "grad_norm": 0.737567126750946,
      "learning_rate": 2.8057513571044803e-08,
      "loss": 0.9012,
      "step": 10686
    },
    {
      "epoch": 1.9687790373536518,
      "grad_norm": 0.7131343483924866,
      "learning_rate": 2.7369043274072524e-08,
      "loss": 1.1263,
      "step": 10687
    },
    {
      "epoch": 1.968964876417023,
      "grad_norm": 0.74207603931427,
      "learning_rate": 2.668912385230815e-08,
      "loss": 0.8822,
      "step": 10688
    },
    {
      "epoch": 1.969150715480394,
      "grad_norm": 0.6574434638023376,
      "learning_rate": 2.6017755363882955e-08,
      "loss": 0.6997,
      "step": 10689
    },
    {
      "epoch": 1.969336554543765,
      "grad_norm": 0.8549362421035767,
      "learning_rate": 2.535493786623988e-08,
      "loss": 1.0949,
      "step": 10690
    },
    {
      "epoch": 1.9695223936071362,
      "grad_norm": 0.9490185976028442,
      "learning_rate": 2.470067141605581e-08,
      "loss": 0.9981,
      "step": 10691
    },
    {
      "epoch": 1.9697082326705073,
      "grad_norm": 0.8517225384712219,
      "learning_rate": 2.4054956069308188e-08,
      "loss": 1.2965,
      "step": 10692
    },
    {
      "epoch": 1.9698940717338784,
      "grad_norm": 0.638239324092865,
      "learning_rate": 2.3417791881208405e-08,
      "loss": 0.7109,
      "step": 10693
    },
    {
      "epoch": 1.9700799107972495,
      "grad_norm": 0.6345565319061279,
      "learning_rate": 2.2789178906257313e-08,
      "loss": 0.9631,
      "step": 10694
    },
    {
      "epoch": 1.9702657498606206,
      "grad_norm": 0.7187051177024841,
      "learning_rate": 2.2169117198234114e-08,
      "loss": 0.9652,
      "step": 10695
    },
    {
      "epoch": 1.970451588923992,
      "grad_norm": 0.6116509437561035,
      "learning_rate": 2.1557606810174157e-08,
      "loss": 0.9015,
      "step": 10696
    },
    {
      "epoch": 1.9706374279873629,
      "grad_norm": 0.8735626339912415,
      "learning_rate": 2.0954647794357852e-08,
      "loss": 0.9417,
      "step": 10697
    },
    {
      "epoch": 1.9708232670507342,
      "grad_norm": 0.6325920224189758,
      "learning_rate": 2.0360240202388357e-08,
      "loss": 1.0079,
      "step": 10698
    },
    {
      "epoch": 1.971009106114105,
      "grad_norm": 0.687492847442627,
      "learning_rate": 1.9774384085080587e-08,
      "loss": 1.0389,
      "step": 10699
    },
    {
      "epoch": 1.9711949451774764,
      "grad_norm": 0.971679151058197,
      "learning_rate": 1.919707949256111e-08,
      "loss": 1.0921,
      "step": 10700
    },
    {
      "epoch": 1.9713807842408473,
      "grad_norm": 0.7148098945617676,
      "learning_rate": 1.862832647420154e-08,
      "loss": 0.7552,
      "step": 10701
    },
    {
      "epoch": 1.9715666233042186,
      "grad_norm": 0.9046821594238281,
      "learning_rate": 1.8068125078651853e-08,
      "loss": 0.9696,
      "step": 10702
    },
    {
      "epoch": 1.9717524623675897,
      "grad_norm": 0.7230017781257629,
      "learning_rate": 1.7516475353829275e-08,
      "loss": 0.9414,
      "step": 10703
    },
    {
      "epoch": 1.9719383014309608,
      "grad_norm": 0.7333557605743408,
      "learning_rate": 1.697337734689608e-08,
      "loss": 0.874,
      "step": 10704
    },
    {
      "epoch": 1.972124140494332,
      "grad_norm": 0.6274997591972351,
      "learning_rate": 1.6438831104337305e-08,
      "loss": 0.6356,
      "step": 10705
    },
    {
      "epoch": 1.972309979557703,
      "grad_norm": 0.7487014532089233,
      "learning_rate": 1.5912836671849728e-08,
      "loss": 1.1066,
      "step": 10706
    },
    {
      "epoch": 1.9724958186210741,
      "grad_norm": 0.7841073274612427,
      "learning_rate": 1.5395394094441794e-08,
      "loss": 1.0279,
      "step": 10707
    },
    {
      "epoch": 1.9726816576844453,
      "grad_norm": 0.6704834699630737,
      "learning_rate": 1.488650341635589e-08,
      "loss": 0.7724,
      "step": 10708
    },
    {
      "epoch": 1.9728674967478164,
      "grad_norm": 0.7334095239639282,
      "learning_rate": 1.4386164681112757e-08,
      "loss": 1.0464,
      "step": 10709
    },
    {
      "epoch": 1.9730533358111875,
      "grad_norm": 0.6330432891845703,
      "learning_rate": 1.3894377931533698e-08,
      "loss": 0.8495,
      "step": 10710
    },
    {
      "epoch": 1.9732391748745586,
      "grad_norm": 0.7090950608253479,
      "learning_rate": 1.3411143209651755e-08,
      "loss": 1.1233,
      "step": 10711
    },
    {
      "epoch": 1.9734250139379297,
      "grad_norm": 0.6479860544204712,
      "learning_rate": 1.2936460556822737e-08,
      "loss": 0.878,
      "step": 10712
    },
    {
      "epoch": 1.973610853001301,
      "grad_norm": 0.7400798201560974,
      "learning_rate": 1.2470330013647502e-08,
      "loss": 0.8981,
      "step": 10713
    },
    {
      "epoch": 1.973796692064672,
      "grad_norm": 0.7641975283622742,
      "learning_rate": 1.2012751619971952e-08,
      "loss": 1.0427,
      "step": 10714
    },
    {
      "epoch": 1.9739825311280432,
      "grad_norm": 0.7831242680549622,
      "learning_rate": 1.1563725414953652e-08,
      "loss": 1.0748,
      "step": 10715
    },
    {
      "epoch": 1.9741683701914141,
      "grad_norm": 0.6629639863967896,
      "learning_rate": 1.1123251436995218e-08,
      "loss": 0.7366,
      "step": 10716
    },
    {
      "epoch": 1.9743542092547854,
      "grad_norm": 0.6962116956710815,
      "learning_rate": 1.0691329723766519e-08,
      "loss": 1.065,
      "step": 10717
    },
    {
      "epoch": 1.9745400483181563,
      "grad_norm": 0.6824777722358704,
      "learning_rate": 1.0267960312215774e-08,
      "loss": 1.0652,
      "step": 10718
    },
    {
      "epoch": 1.9747258873815277,
      "grad_norm": 0.7710967659950256,
      "learning_rate": 9.853143238547358e-09,
      "loss": 1.0013,
      "step": 10719
    },
    {
      "epoch": 1.9749117264448988,
      "grad_norm": 0.8690904378890991,
      "learning_rate": 9.4468785382551e-09,
      "loss": 1.057,
      "step": 10720
    },
    {
      "epoch": 1.9750975655082699,
      "grad_norm": 0.8146247863769531,
      "learning_rate": 9.049166246077878e-09,
      "loss": 1.0114,
      "step": 10721
    },
    {
      "epoch": 1.975283404571641,
      "grad_norm": 0.6535385251045227,
      "learning_rate": 8.660006396032928e-09,
      "loss": 0.9387,
      "step": 10722
    },
    {
      "epoch": 1.975469243635012,
      "grad_norm": 0.5985758304595947,
      "learning_rate": 8.279399021404732e-09,
      "loss": 0.8463,
      "step": 10723
    },
    {
      "epoch": 1.9756550826983832,
      "grad_norm": 0.9042931199073792,
      "learning_rate": 7.907344154756135e-09,
      "loss": 0.66,
      "step": 10724
    },
    {
      "epoch": 1.9758409217617543,
      "grad_norm": 0.7004178762435913,
      "learning_rate": 7.543841827895025e-09,
      "loss": 0.8676,
      "step": 10725
    },
    {
      "epoch": 1.9760267608251254,
      "grad_norm": 2.0802855491638184,
      "learning_rate": 7.188892071929854e-09,
      "loss": 1.6243,
      "step": 10726
    },
    {
      "epoch": 1.9762125998884965,
      "grad_norm": 0.8206648826599121,
      "learning_rate": 6.8424949172030124e-09,
      "loss": 1.0054,
      "step": 10727
    },
    {
      "epoch": 1.9763984389518678,
      "grad_norm": 0.6644618511199951,
      "learning_rate": 6.504650393357459e-09,
      "loss": 1.0766,
      "step": 10728
    },
    {
      "epoch": 1.9765842780152387,
      "grad_norm": 0.5854812860488892,
      "learning_rate": 6.1753585292811945e-09,
      "loss": 1.0447,
      "step": 10729
    },
    {
      "epoch": 1.97677011707861,
      "grad_norm": 0.6894762516021729,
      "learning_rate": 5.854619353140578e-09,
      "loss": 0.7763,
      "step": 10730
    },
    {
      "epoch": 1.976955956141981,
      "grad_norm": 0.6703664064407349,
      "learning_rate": 5.542432892369221e-09,
      "loss": 0.9355,
      "step": 10731
    },
    {
      "epoch": 1.9771417952053523,
      "grad_norm": 0.720122218132019,
      "learning_rate": 5.2387991736790876e-09,
      "loss": 1.1274,
      "step": 10732
    },
    {
      "epoch": 1.9773276342687232,
      "grad_norm": 0.7998895049095154,
      "learning_rate": 4.943718223016092e-09,
      "loss": 1.2719,
      "step": 10733
    },
    {
      "epoch": 1.9775134733320945,
      "grad_norm": 0.7410762310028076,
      "learning_rate": 4.657190065648909e-09,
      "loss": 1.0627,
      "step": 10734
    },
    {
      "epoch": 1.9776993123954654,
      "grad_norm": 0.842653751373291,
      "learning_rate": 4.379214726069058e-09,
      "loss": 0.9684,
      "step": 10735
    },
    {
      "epoch": 1.9778851514588367,
      "grad_norm": 0.6367177963256836,
      "learning_rate": 4.1097922280464165e-09,
      "loss": 0.9476,
      "step": 10736
    },
    {
      "epoch": 1.9780709905222078,
      "grad_norm": 0.6847108006477356,
      "learning_rate": 3.848922594640314e-09,
      "loss": 0.9157,
      "step": 10737
    },
    {
      "epoch": 1.978256829585579,
      "grad_norm": 0.7617043852806091,
      "learning_rate": 3.59660584814403e-09,
      "loss": 0.9627,
      "step": 10738
    },
    {
      "epoch": 1.97844266864895,
      "grad_norm": 0.6675279140472412,
      "learning_rate": 3.352842010162505e-09,
      "loss": 0.9792,
      "step": 10739
    },
    {
      "epoch": 1.9786285077123211,
      "grad_norm": 0.7527198195457458,
      "learning_rate": 3.117631101534624e-09,
      "loss": 0.942,
      "step": 10740
    },
    {
      "epoch": 1.9788143467756922,
      "grad_norm": 0.8000282049179077,
      "learning_rate": 2.8909731423665264e-09,
      "loss": 0.9577,
      "step": 10741
    },
    {
      "epoch": 1.9790001858390633,
      "grad_norm": 0.6555179953575134,
      "learning_rate": 2.672868152064911e-09,
      "loss": 0.9841,
      "step": 10742
    },
    {
      "epoch": 1.9791860249024344,
      "grad_norm": 0.753463089466095,
      "learning_rate": 2.4633161492704227e-09,
      "loss": 0.9208,
      "step": 10743
    },
    {
      "epoch": 1.9793718639658056,
      "grad_norm": 0.8542004227638245,
      "learning_rate": 2.2623171519131625e-09,
      "loss": 1.1224,
      "step": 10744
    },
    {
      "epoch": 1.9795577030291769,
      "grad_norm": 0.7783091068267822,
      "learning_rate": 2.0698711771793835e-09,
      "loss": 0.7907,
      "step": 10745
    },
    {
      "epoch": 1.9797435420925478,
      "grad_norm": 0.6484434604644775,
      "learning_rate": 1.8859782415447947e-09,
      "loss": 1.1133,
      "step": 10746
    },
    {
      "epoch": 1.979929381155919,
      "grad_norm": 1.1288589239120483,
      "learning_rate": 1.7106383607190523e-09,
      "loss": 1.3652,
      "step": 10747
    },
    {
      "epoch": 1.98011522021929,
      "grad_norm": 0.7378808856010437,
      "learning_rate": 1.5438515497123718e-09,
      "loss": 0.7895,
      "step": 10748
    },
    {
      "epoch": 1.9803010592826613,
      "grad_norm": 0.706842839717865,
      "learning_rate": 1.3856178227800165e-09,
      "loss": 0.8849,
      "step": 10749
    },
    {
      "epoch": 1.9804868983460322,
      "grad_norm": 0.6956677436828613,
      "learning_rate": 1.2359371934556052e-09,
      "loss": 1.0329,
      "step": 10750
    },
    {
      "epoch": 1.9806727374094035,
      "grad_norm": 0.7249075174331665,
      "learning_rate": 1.0948096745622138e-09,
      "loss": 0.9499,
      "step": 10751
    },
    {
      "epoch": 1.9808585764727744,
      "grad_norm": 0.7836699485778809,
      "learning_rate": 9.622352781457622e-10,
      "loss": 1.2175,
      "step": 10752
    },
    {
      "epoch": 1.9810444155361457,
      "grad_norm": 0.7368754148483276,
      "learning_rate": 8.382140155527296e-10,
      "loss": 1.0044,
      "step": 10753
    },
    {
      "epoch": 1.9812302545995168,
      "grad_norm": 0.6586326360702515,
      "learning_rate": 7.227458973968482e-10,
      "loss": 1.1634,
      "step": 10754
    },
    {
      "epoch": 1.981416093662888,
      "grad_norm": 0.7742688655853271,
      "learning_rate": 6.158309335591028e-10,
      "loss": 0.8664,
      "step": 10755
    },
    {
      "epoch": 1.981601932726259,
      "grad_norm": 0.703048050403595,
      "learning_rate": 5.174691331655268e-10,
      "loss": 0.8754,
      "step": 10756
    },
    {
      "epoch": 1.9817877717896302,
      "grad_norm": 0.6949231028556824,
      "learning_rate": 4.276605046427129e-10,
      "loss": 1.0135,
      "step": 10757
    },
    {
      "epoch": 1.9819736108530013,
      "grad_norm": 0.8362816572189331,
      "learning_rate": 3.464050556734044e-10,
      "loss": 0.9027,
      "step": 10758
    },
    {
      "epoch": 1.9821594499163724,
      "grad_norm": 0.6021760106086731,
      "learning_rate": 2.737027932075975e-10,
      "loss": 0.9221,
      "step": 10759
    },
    {
      "epoch": 1.9823452889797435,
      "grad_norm": 0.7097657322883606,
      "learning_rate": 2.0955372345143886e-10,
      "loss": 0.945,
      "step": 10760
    },
    {
      "epoch": 1.9825311280431146,
      "grad_norm": 0.713502824306488,
      "learning_rate": 1.5395785190053247e-10,
      "loss": 0.7171,
      "step": 10761
    },
    {
      "epoch": 1.982716967106486,
      "grad_norm": 0.7117783427238464,
      "learning_rate": 1.0691518331773509e-10,
      "loss": 0.8693,
      "step": 10762
    },
    {
      "epoch": 2.0001858390633713,
      "grad_norm": 0.6326022744178772,
      "learning_rate": 0.00010020453062911643,
      "loss": 0.9353,
      "step": 10763
    },
    {
      "epoch": 2.000371678126742,
      "grad_norm": 0.615979790687561,
      "learning_rate": 0.0001001899213167062,
      "loss": 0.941,
      "step": 10764
    },
    {
      "epoch": 2.0005575171901135,
      "grad_norm": 0.6112083196640015,
      "learning_rate": 0.00010017531200024244,
      "loss": 0.8806,
      "step": 10765
    },
    {
      "epoch": 2.0007433562534844,
      "grad_norm": 0.7036938071250916,
      "learning_rate": 0.00010016070268003695,
      "loss": 0.7532,
      "step": 10766
    },
    {
      "epoch": 2.0009291953168558,
      "grad_norm": 0.6403826475143433,
      "learning_rate": 0.00010014609335640152,
      "loss": 0.6336,
      "step": 10767
    },
    {
      "epoch": 2.0011150343802266,
      "grad_norm": 0.6601222157478333,
      "learning_rate": 0.00010013148402964798,
      "loss": 0.8791,
      "step": 10768
    },
    {
      "epoch": 2.0011150343802266,
      "eval_loss": 1.0075899362564087,
      "eval_runtime": 23.187,
      "eval_samples_per_second": 47.095,
      "eval_steps_per_second": 23.548,
      "step": 10768
    },
    {
      "epoch": 2.001300873443598,
      "grad_norm": 0.7029190063476562,
      "learning_rate": 0.00010011687470008817,
      "loss": 0.8273,
      "step": 10769
    },
    {
      "epoch": 2.001486712506969,
      "grad_norm": 0.630573570728302,
      "learning_rate": 0.00010010226536803385,
      "loss": 0.9447,
      "step": 10770
    },
    {
      "epoch": 2.00167255157034,
      "grad_norm": 0.6786049008369446,
      "learning_rate": 0.00010008765603379685,
      "loss": 0.9115,
      "step": 10771
    },
    {
      "epoch": 2.001858390633711,
      "grad_norm": 0.680065929889679,
      "learning_rate": 0.00010007304669768897,
      "loss": 0.8726,
      "step": 10772
    },
    {
      "epoch": 2.0020442296970824,
      "grad_norm": 0.6689335107803345,
      "learning_rate": 0.00010005843736002206,
      "loss": 0.9695,
      "step": 10773
    },
    {
      "epoch": 2.0022300687604533,
      "grad_norm": 0.7676763534545898,
      "learning_rate": 0.00010004382802110787,
      "loss": 1.04,
      "step": 10774
    },
    {
      "epoch": 2.0024159078238246,
      "grad_norm": 0.7226296663284302,
      "learning_rate": 0.00010002921868125827,
      "loss": 0.6713,
      "step": 10775
    },
    {
      "epoch": 2.0026017468871955,
      "grad_norm": 0.6010726094245911,
      "learning_rate": 0.00010001460934078506,
      "loss": 0.6302,
      "step": 10776
    },
    {
      "epoch": 2.002787585950567,
      "grad_norm": 0.8062921762466431,
      "learning_rate": 0.0001,
      "loss": 0.9336,
      "step": 10777
    },
    {
      "epoch": 2.0029734250139377,
      "grad_norm": 0.8343218564987183,
      "learning_rate": 9.998539065921496e-05,
      "loss": 1.0856,
      "step": 10778
    },
    {
      "epoch": 2.003159264077309,
      "grad_norm": 0.6121909618377686,
      "learning_rate": 9.997078131874175e-05,
      "loss": 0.9131,
      "step": 10779
    },
    {
      "epoch": 2.0033451031406804,
      "grad_norm": 0.7555389404296875,
      "learning_rate": 9.995617197889215e-05,
      "loss": 1.0778,
      "step": 10780
    },
    {
      "epoch": 2.0035309422040513,
      "grad_norm": 0.6045559048652649,
      "learning_rate": 9.9941562639978e-05,
      "loss": 0.7465,
      "step": 10781
    },
    {
      "epoch": 2.0037167812674226,
      "grad_norm": 0.7431402802467346,
      "learning_rate": 9.992695330231105e-05,
      "loss": 0.9095,
      "step": 10782
    },
    {
      "epoch": 2.0039026203307935,
      "grad_norm": 0.7544237375259399,
      "learning_rate": 9.991234396620316e-05,
      "loss": 0.9414,
      "step": 10783
    },
    {
      "epoch": 2.004088459394165,
      "grad_norm": 0.6956236362457275,
      "learning_rate": 9.989773463196618e-05,
      "loss": 1.1008,
      "step": 10784
    },
    {
      "epoch": 2.0042742984575357,
      "grad_norm": 0.7116724252700806,
      "learning_rate": 9.988312529991185e-05,
      "loss": 0.8229,
      "step": 10785
    },
    {
      "epoch": 2.004460137520907,
      "grad_norm": 0.7476662993431091,
      "learning_rate": 9.986851597035203e-05,
      "loss": 0.7782,
      "step": 10786
    },
    {
      "epoch": 2.004645976584278,
      "grad_norm": 0.7127363681793213,
      "learning_rate": 9.985390664359849e-05,
      "loss": 1.1201,
      "step": 10787
    },
    {
      "epoch": 2.0048318156476492,
      "grad_norm": 0.6719833612442017,
      "learning_rate": 9.983929731996306e-05,
      "loss": 0.7797,
      "step": 10788
    },
    {
      "epoch": 2.00501765471102,
      "grad_norm": 0.7116879820823669,
      "learning_rate": 9.98246879997576e-05,
      "loss": 0.7709,
      "step": 10789
    },
    {
      "epoch": 2.0052034937743914,
      "grad_norm": 0.7485224008560181,
      "learning_rate": 9.981007868329382e-05,
      "loss": 1.1297,
      "step": 10790
    },
    {
      "epoch": 2.0053893328377623,
      "grad_norm": 0.7865563631057739,
      "learning_rate": 9.979546937088362e-05,
      "loss": 1.0207,
      "step": 10791
    },
    {
      "epoch": 2.0055751719011337,
      "grad_norm": 0.65020352602005,
      "learning_rate": 9.978086006283875e-05,
      "loss": 0.6663,
      "step": 10792
    },
    {
      "epoch": 2.0057610109645045,
      "grad_norm": 0.6295920014381409,
      "learning_rate": 9.976625075947103e-05,
      "loss": 0.6556,
      "step": 10793
    },
    {
      "epoch": 2.005946850027876,
      "grad_norm": 0.8350384831428528,
      "learning_rate": 9.975164146109232e-05,
      "loss": 0.9925,
      "step": 10794
    },
    {
      "epoch": 2.006132689091247,
      "grad_norm": 0.7139639854431152,
      "learning_rate": 9.973703216801437e-05,
      "loss": 0.7801,
      "step": 10795
    },
    {
      "epoch": 2.006318528154618,
      "grad_norm": 0.8106564283370972,
      "learning_rate": 9.972242288054906e-05,
      "loss": 0.962,
      "step": 10796
    },
    {
      "epoch": 2.0065043672179894,
      "grad_norm": 0.7551612257957458,
      "learning_rate": 9.970781359900812e-05,
      "loss": 0.8711,
      "step": 10797
    },
    {
      "epoch": 2.0066902062813603,
      "grad_norm": 0.6887959837913513,
      "learning_rate": 9.96932043237034e-05,
      "loss": 0.8263,
      "step": 10798
    },
    {
      "epoch": 2.0068760453447316,
      "grad_norm": 0.7183730006217957,
      "learning_rate": 9.967859505494673e-05,
      "loss": 0.7888,
      "step": 10799
    },
    {
      "epoch": 2.0070618844081025,
      "grad_norm": 0.7318005561828613,
      "learning_rate": 9.966398579304989e-05,
      "loss": 0.9993,
      "step": 10800
    },
    {
      "epoch": 2.007247723471474,
      "grad_norm": 0.9438154101371765,
      "learning_rate": 9.964937653832468e-05,
      "loss": 1.1902,
      "step": 10801
    },
    {
      "epoch": 2.0074335625348447,
      "grad_norm": 0.8429304361343384,
      "learning_rate": 9.963476729108294e-05,
      "loss": 0.9782,
      "step": 10802
    },
    {
      "epoch": 2.007619401598216,
      "grad_norm": 0.7255978584289551,
      "learning_rate": 9.962015805163647e-05,
      "loss": 0.7803,
      "step": 10803
    },
    {
      "epoch": 2.007805240661587,
      "grad_norm": 0.7246456742286682,
      "learning_rate": 9.960554882029705e-05,
      "loss": 0.8352,
      "step": 10804
    },
    {
      "epoch": 2.0079910797249583,
      "grad_norm": 0.6792916059494019,
      "learning_rate": 9.959093959737652e-05,
      "loss": 0.795,
      "step": 10805
    },
    {
      "epoch": 2.008176918788329,
      "grad_norm": 0.8336760997772217,
      "learning_rate": 9.95763303831867e-05,
      "loss": 0.9368,
      "step": 10806
    },
    {
      "epoch": 2.0083627578517005,
      "grad_norm": 0.7357673645019531,
      "learning_rate": 9.956172117803937e-05,
      "loss": 1.0218,
      "step": 10807
    },
    {
      "epoch": 2.0085485969150714,
      "grad_norm": 0.9112713932991028,
      "learning_rate": 9.954711198224634e-05,
      "loss": 1.193,
      "step": 10808
    },
    {
      "epoch": 2.0087344359784427,
      "grad_norm": 0.8156577348709106,
      "learning_rate": 9.953250279611946e-05,
      "loss": 0.9637,
      "step": 10809
    },
    {
      "epoch": 2.0089202750418136,
      "grad_norm": 0.6362747550010681,
      "learning_rate": 9.951789361997048e-05,
      "loss": 0.717,
      "step": 10810
    },
    {
      "epoch": 2.009106114105185,
      "grad_norm": 0.6513655781745911,
      "learning_rate": 9.950328445411125e-05,
      "loss": 0.8537,
      "step": 10811
    },
    {
      "epoch": 2.0092919531685562,
      "grad_norm": 0.65449059009552,
      "learning_rate": 9.948867529885355e-05,
      "loss": 0.9545,
      "step": 10812
    },
    {
      "epoch": 2.009477792231927,
      "grad_norm": 0.8565689921379089,
      "learning_rate": 9.94740661545092e-05,
      "loss": 1.092,
      "step": 10813
    },
    {
      "epoch": 2.0096636312952985,
      "grad_norm": 0.6790142059326172,
      "learning_rate": 9.945945702138998e-05,
      "loss": 0.9912,
      "step": 10814
    },
    {
      "epoch": 2.0098494703586693,
      "grad_norm": 0.6726483106613159,
      "learning_rate": 9.944484789980773e-05,
      "loss": 0.7884,
      "step": 10815
    },
    {
      "epoch": 2.0100353094220407,
      "grad_norm": 0.7265700697898865,
      "learning_rate": 9.943023879007428e-05,
      "loss": 0.8606,
      "step": 10816
    },
    {
      "epoch": 2.0102211484854116,
      "grad_norm": 0.7459362149238586,
      "learning_rate": 9.941562969250138e-05,
      "loss": 0.8296,
      "step": 10817
    },
    {
      "epoch": 2.010406987548783,
      "grad_norm": 0.6986656188964844,
      "learning_rate": 9.940102060740087e-05,
      "loss": 0.8882,
      "step": 10818
    },
    {
      "epoch": 2.0105928266121538,
      "grad_norm": 0.8597057461738586,
      "learning_rate": 9.938641153508456e-05,
      "loss": 1.0597,
      "step": 10819
    },
    {
      "epoch": 2.010778665675525,
      "grad_norm": 0.6738445162773132,
      "learning_rate": 9.937180247586422e-05,
      "loss": 0.7268,
      "step": 10820
    },
    {
      "epoch": 2.010964504738896,
      "grad_norm": 0.6111264228820801,
      "learning_rate": 9.935719343005171e-05,
      "loss": 0.6557,
      "step": 10821
    },
    {
      "epoch": 2.0111503438022673,
      "grad_norm": 0.7664129734039307,
      "learning_rate": 9.934258439795879e-05,
      "loss": 0.8225,
      "step": 10822
    },
    {
      "epoch": 2.011336182865638,
      "grad_norm": 0.7452059984207153,
      "learning_rate": 9.932797537989727e-05,
      "loss": 0.8836,
      "step": 10823
    },
    {
      "epoch": 2.0115220219290095,
      "grad_norm": 0.6665897965431213,
      "learning_rate": 9.931336637617896e-05,
      "loss": 0.8776,
      "step": 10824
    },
    {
      "epoch": 2.0117078609923804,
      "grad_norm": 0.7272871732711792,
      "learning_rate": 9.929875738711564e-05,
      "loss": 0.7879,
      "step": 10825
    },
    {
      "epoch": 2.0118937000557517,
      "grad_norm": 0.6861128807067871,
      "learning_rate": 9.928414841301919e-05,
      "loss": 0.9416,
      "step": 10826
    },
    {
      "epoch": 2.0120795391191226,
      "grad_norm": 0.7545861005783081,
      "learning_rate": 9.926953945420132e-05,
      "loss": 1.0873,
      "step": 10827
    },
    {
      "epoch": 2.012265378182494,
      "grad_norm": 0.7821633219718933,
      "learning_rate": 9.925493051097389e-05,
      "loss": 0.9366,
      "step": 10828
    },
    {
      "epoch": 2.0124512172458653,
      "grad_norm": 0.6889685392379761,
      "learning_rate": 9.924032158364873e-05,
      "loss": 0.8754,
      "step": 10829
    },
    {
      "epoch": 2.012637056309236,
      "grad_norm": 1.621429443359375,
      "learning_rate": 9.922571267253755e-05,
      "loss": 1.0026,
      "step": 10830
    },
    {
      "epoch": 2.0128228953726075,
      "grad_norm": 0.7398176193237305,
      "learning_rate": 9.921110377795225e-05,
      "loss": 0.8988,
      "step": 10831
    },
    {
      "epoch": 2.0130087344359784,
      "grad_norm": 0.6716428995132446,
      "learning_rate": 9.919649490020455e-05,
      "loss": 0.9864,
      "step": 10832
    },
    {
      "epoch": 2.0131945734993497,
      "grad_norm": 0.7361804246902466,
      "learning_rate": 9.918188603960632e-05,
      "loss": 1.0162,
      "step": 10833
    },
    {
      "epoch": 2.0133804125627206,
      "grad_norm": 0.8350973725318909,
      "learning_rate": 9.91672771964693e-05,
      "loss": 0.9043,
      "step": 10834
    },
    {
      "epoch": 2.013566251626092,
      "grad_norm": 0.869459867477417,
      "learning_rate": 9.915266837110529e-05,
      "loss": 0.9351,
      "step": 10835
    },
    {
      "epoch": 2.013752090689463,
      "grad_norm": 0.6775286793708801,
      "learning_rate": 9.913805956382617e-05,
      "loss": 0.7345,
      "step": 10836
    },
    {
      "epoch": 2.013937929752834,
      "grad_norm": 0.7324315309524536,
      "learning_rate": 9.912345077494368e-05,
      "loss": 0.8199,
      "step": 10837
    },
    {
      "epoch": 2.014123768816205,
      "grad_norm": 0.8208001852035522,
      "learning_rate": 9.91088420047696e-05,
      "loss": 1.1714,
      "step": 10838
    },
    {
      "epoch": 2.0143096078795764,
      "grad_norm": 0.8033028244972229,
      "learning_rate": 9.909423325361581e-05,
      "loss": 0.9778,
      "step": 10839
    },
    {
      "epoch": 2.0144954469429472,
      "grad_norm": 0.7496377825737,
      "learning_rate": 9.907962452179403e-05,
      "loss": 0.718,
      "step": 10840
    },
    {
      "epoch": 2.0146812860063186,
      "grad_norm": 0.7925049662590027,
      "learning_rate": 9.90650158096161e-05,
      "loss": 1.1434,
      "step": 10841
    },
    {
      "epoch": 2.0148671250696895,
      "grad_norm": 0.8028472065925598,
      "learning_rate": 9.905040711739377e-05,
      "loss": 0.8863,
      "step": 10842
    },
    {
      "epoch": 2.015052964133061,
      "grad_norm": 0.8392669558525085,
      "learning_rate": 9.90357984454389e-05,
      "loss": 1.0166,
      "step": 10843
    },
    {
      "epoch": 2.0152388031964317,
      "grad_norm": 0.7445780634880066,
      "learning_rate": 9.902118979406324e-05,
      "loss": 0.8275,
      "step": 10844
    },
    {
      "epoch": 2.015424642259803,
      "grad_norm": 0.638634204864502,
      "learning_rate": 9.900658116357863e-05,
      "loss": 0.9946,
      "step": 10845
    },
    {
      "epoch": 2.0156104813231743,
      "grad_norm": 0.776951014995575,
      "learning_rate": 9.899197255429678e-05,
      "loss": 0.7251,
      "step": 10846
    },
    {
      "epoch": 2.015796320386545,
      "grad_norm": 0.7218950986862183,
      "learning_rate": 9.897736396652959e-05,
      "loss": 0.875,
      "step": 10847
    },
    {
      "epoch": 2.0159821594499165,
      "grad_norm": 0.7573806047439575,
      "learning_rate": 9.896275540058881e-05,
      "loss": 1.0281,
      "step": 10848
    },
    {
      "epoch": 2.0161679985132874,
      "grad_norm": 0.7256792783737183,
      "learning_rate": 9.894814685678624e-05,
      "loss": 0.8202,
      "step": 10849
    },
    {
      "epoch": 2.0163538375766588,
      "grad_norm": 0.7149482369422913,
      "learning_rate": 9.893353833543365e-05,
      "loss": 0.7834,
      "step": 10850
    },
    {
      "epoch": 2.0165396766400296,
      "grad_norm": 0.6294644474983215,
      "learning_rate": 9.891892983684289e-05,
      "loss": 0.7256,
      "step": 10851
    },
    {
      "epoch": 2.016725515703401,
      "grad_norm": 0.7436536550521851,
      "learning_rate": 9.89043213613257e-05,
      "loss": 0.8414,
      "step": 10852
    },
    {
      "epoch": 2.016911354766772,
      "grad_norm": 0.708697497844696,
      "learning_rate": 9.888971290919389e-05,
      "loss": 0.9798,
      "step": 10853
    },
    {
      "epoch": 2.017097193830143,
      "grad_norm": 0.6918801665306091,
      "learning_rate": 9.887510448075924e-05,
      "loss": 0.7837,
      "step": 10854
    },
    {
      "epoch": 2.017283032893514,
      "grad_norm": 0.7205170392990112,
      "learning_rate": 9.886049607633357e-05,
      "loss": 0.6071,
      "step": 10855
    },
    {
      "epoch": 2.0174688719568854,
      "grad_norm": 0.735811710357666,
      "learning_rate": 9.884588769622862e-05,
      "loss": 0.9584,
      "step": 10856
    },
    {
      "epoch": 2.0176547110202563,
      "grad_norm": 0.6744986772537231,
      "learning_rate": 9.883127934075624e-05,
      "loss": 0.7276,
      "step": 10857
    },
    {
      "epoch": 2.0178405500836276,
      "grad_norm": 0.8203528523445129,
      "learning_rate": 9.88166710102282e-05,
      "loss": 0.8217,
      "step": 10858
    },
    {
      "epoch": 2.0180263891469985,
      "grad_norm": 0.718228816986084,
      "learning_rate": 9.88020627049563e-05,
      "loss": 0.905,
      "step": 10859
    },
    {
      "epoch": 2.01821222821037,
      "grad_norm": 1.0625849962234497,
      "learning_rate": 9.87874544252523e-05,
      "loss": 1.0942,
      "step": 10860
    },
    {
      "epoch": 2.018398067273741,
      "grad_norm": 0.6921961903572083,
      "learning_rate": 9.877284617142802e-05,
      "loss": 0.8311,
      "step": 10861
    },
    {
      "epoch": 2.018583906337112,
      "grad_norm": 0.7733319997787476,
      "learning_rate": 9.875823794379522e-05,
      "loss": 0.9363,
      "step": 10862
    },
    {
      "epoch": 2.0187697454004834,
      "grad_norm": 0.8737978935241699,
      "learning_rate": 9.874362974266572e-05,
      "loss": 0.7882,
      "step": 10863
    },
    {
      "epoch": 2.0189555844638543,
      "grad_norm": 0.6428980231285095,
      "learning_rate": 9.872902156835125e-05,
      "loss": 0.8382,
      "step": 10864
    },
    {
      "epoch": 2.0191414235272256,
      "grad_norm": 0.8537091612815857,
      "learning_rate": 9.871441342116368e-05,
      "loss": 1.1804,
      "step": 10865
    },
    {
      "epoch": 2.0193272625905965,
      "grad_norm": 0.8355739116668701,
      "learning_rate": 9.86998053014147e-05,
      "loss": 1.0267,
      "step": 10866
    },
    {
      "epoch": 2.019513101653968,
      "grad_norm": 0.6342544555664062,
      "learning_rate": 9.868519720941615e-05,
      "loss": 0.5791,
      "step": 10867
    },
    {
      "epoch": 2.0196989407173387,
      "grad_norm": 0.7603575587272644,
      "learning_rate": 9.867058914547981e-05,
      "loss": 1.0268,
      "step": 10868
    },
    {
      "epoch": 2.01988477978071,
      "grad_norm": 0.7468525767326355,
      "learning_rate": 9.86559811099175e-05,
      "loss": 1.032,
      "step": 10869
    },
    {
      "epoch": 2.020070618844081,
      "grad_norm": 0.6898447275161743,
      "learning_rate": 9.864137310304094e-05,
      "loss": 0.757,
      "step": 10870
    },
    {
      "epoch": 2.0202564579074522,
      "grad_norm": 0.5959223508834839,
      "learning_rate": 9.862676512516196e-05,
      "loss": 0.7033,
      "step": 10871
    },
    {
      "epoch": 2.020442296970823,
      "grad_norm": 0.7550606727600098,
      "learning_rate": 9.86121571765923e-05,
      "loss": 0.8592,
      "step": 10872
    },
    {
      "epoch": 2.0206281360341944,
      "grad_norm": 0.7787352800369263,
      "learning_rate": 9.859754925764378e-05,
      "loss": 1.0283,
      "step": 10873
    },
    {
      "epoch": 2.0208139750975653,
      "grad_norm": 0.848624587059021,
      "learning_rate": 9.858294136862815e-05,
      "loss": 0.8106,
      "step": 10874
    },
    {
      "epoch": 2.0209998141609367,
      "grad_norm": 0.6989394426345825,
      "learning_rate": 9.856833350985722e-05,
      "loss": 0.8153,
      "step": 10875
    },
    {
      "epoch": 2.0211856532243075,
      "grad_norm": 0.7675639986991882,
      "learning_rate": 9.855372568164274e-05,
      "loss": 0.9524,
      "step": 10876
    },
    {
      "epoch": 2.021371492287679,
      "grad_norm": 0.9018722176551819,
      "learning_rate": 9.853911788429648e-05,
      "loss": 1.0343,
      "step": 10877
    },
    {
      "epoch": 2.02155733135105,
      "grad_norm": 0.7960389852523804,
      "learning_rate": 9.852451011813027e-05,
      "loss": 0.8321,
      "step": 10878
    },
    {
      "epoch": 2.021743170414421,
      "grad_norm": 0.7187342643737793,
      "learning_rate": 9.850990238345587e-05,
      "loss": 0.9727,
      "step": 10879
    },
    {
      "epoch": 2.0219290094777924,
      "grad_norm": 0.7807068228721619,
      "learning_rate": 9.849529468058503e-05,
      "loss": 1.0024,
      "step": 10880
    },
    {
      "epoch": 2.0221148485411633,
      "grad_norm": 0.6860795021057129,
      "learning_rate": 9.848068700982955e-05,
      "loss": 0.9321,
      "step": 10881
    },
    {
      "epoch": 2.0223006876045346,
      "grad_norm": 1.0080658197402954,
      "learning_rate": 9.84660793715012e-05,
      "loss": 0.9219,
      "step": 10882
    },
    {
      "epoch": 2.0224865266679055,
      "grad_norm": 2.826295852661133,
      "learning_rate": 9.845147176591176e-05,
      "loss": 1.5738,
      "step": 10883
    },
    {
      "epoch": 2.022672365731277,
      "grad_norm": 0.5995877981185913,
      "learning_rate": 9.843686419337298e-05,
      "loss": 0.4572,
      "step": 10884
    },
    {
      "epoch": 2.0228582047946477,
      "grad_norm": 0.7833349108695984,
      "learning_rate": 9.842225665419668e-05,
      "loss": 0.9323,
      "step": 10885
    },
    {
      "epoch": 2.023044043858019,
      "grad_norm": 0.827555775642395,
      "learning_rate": 9.840764914869456e-05,
      "loss": 0.9816,
      "step": 10886
    },
    {
      "epoch": 2.02322988292139,
      "grad_norm": 0.7901204824447632,
      "learning_rate": 9.839304167717849e-05,
      "loss": 0.7856,
      "step": 10887
    },
    {
      "epoch": 2.0234157219847613,
      "grad_norm": 0.8344346284866333,
      "learning_rate": 9.837843423996011e-05,
      "loss": 1.0287,
      "step": 10888
    },
    {
      "epoch": 2.023601561048132,
      "grad_norm": 0.7958934307098389,
      "learning_rate": 9.836382683735132e-05,
      "loss": 0.9216,
      "step": 10889
    },
    {
      "epoch": 2.0237874001115035,
      "grad_norm": 0.7749890089035034,
      "learning_rate": 9.834921946966383e-05,
      "loss": 0.9817,
      "step": 10890
    },
    {
      "epoch": 2.0239732391748744,
      "grad_norm": 0.6827292442321777,
      "learning_rate": 9.833461213720944e-05,
      "loss": 1.0178,
      "step": 10891
    },
    {
      "epoch": 2.0241590782382457,
      "grad_norm": 0.8337804675102234,
      "learning_rate": 9.832000484029987e-05,
      "loss": 0.8763,
      "step": 10892
    },
    {
      "epoch": 2.0243449173016166,
      "grad_norm": 0.8186933398246765,
      "learning_rate": 9.830539757924694e-05,
      "loss": 0.9928,
      "step": 10893
    },
    {
      "epoch": 2.024530756364988,
      "grad_norm": 0.91987144947052,
      "learning_rate": 9.829079035436236e-05,
      "loss": 1.1445,
      "step": 10894
    },
    {
      "epoch": 2.0247165954283592,
      "grad_norm": 0.7778676748275757,
      "learning_rate": 9.827618316595795e-05,
      "loss": 0.9972,
      "step": 10895
    },
    {
      "epoch": 2.02490243449173,
      "grad_norm": 0.8444634079933167,
      "learning_rate": 9.826157601434543e-05,
      "loss": 0.7407,
      "step": 10896
    },
    {
      "epoch": 2.0250882735551015,
      "grad_norm": 0.8590813279151917,
      "learning_rate": 9.824696889983662e-05,
      "loss": 1.0552,
      "step": 10897
    },
    {
      "epoch": 2.0252741126184723,
      "grad_norm": 0.765149712562561,
      "learning_rate": 9.823236182274318e-05,
      "loss": 0.9949,
      "step": 10898
    },
    {
      "epoch": 2.0254599516818437,
      "grad_norm": 0.8012186288833618,
      "learning_rate": 9.8217754783377e-05,
      "loss": 1.0422,
      "step": 10899
    },
    {
      "epoch": 2.0256457907452146,
      "grad_norm": 0.9006714820861816,
      "learning_rate": 9.820314778204978e-05,
      "loss": 1.1215,
      "step": 10900
    },
    {
      "epoch": 2.025831629808586,
      "grad_norm": 0.9209079146385193,
      "learning_rate": 9.81885408190733e-05,
      "loss": 1.1306,
      "step": 10901
    },
    {
      "epoch": 2.0260174688719568,
      "grad_norm": 0.6512848734855652,
      "learning_rate": 9.817393389475928e-05,
      "loss": 0.7758,
      "step": 10902
    },
    {
      "epoch": 2.026203307935328,
      "grad_norm": 0.8860459327697754,
      "learning_rate": 9.815932700941954e-05,
      "loss": 0.9835,
      "step": 10903
    },
    {
      "epoch": 2.026389146998699,
      "grad_norm": 0.7635270953178406,
      "learning_rate": 9.814472016336578e-05,
      "loss": 0.8204,
      "step": 10904
    },
    {
      "epoch": 2.0265749860620703,
      "grad_norm": 0.7491157054901123,
      "learning_rate": 9.813011335690981e-05,
      "loss": 0.7987,
      "step": 10905
    },
    {
      "epoch": 2.026760825125441,
      "grad_norm": 0.6436562538146973,
      "learning_rate": 9.811550659036333e-05,
      "loss": 0.9272,
      "step": 10906
    },
    {
      "epoch": 2.0269466641888125,
      "grad_norm": 0.804854691028595,
      "learning_rate": 9.810089986403816e-05,
      "loss": 0.9982,
      "step": 10907
    },
    {
      "epoch": 2.0271325032521834,
      "grad_norm": 0.6464484333992004,
      "learning_rate": 9.808629317824599e-05,
      "loss": 0.7298,
      "step": 10908
    },
    {
      "epoch": 2.0273183423155547,
      "grad_norm": 0.7032682299613953,
      "learning_rate": 9.807168653329858e-05,
      "loss": 0.8478,
      "step": 10909
    },
    {
      "epoch": 2.0275041813789256,
      "grad_norm": 0.7089880108833313,
      "learning_rate": 9.805707992950776e-05,
      "loss": 0.8488,
      "step": 10910
    },
    {
      "epoch": 2.027690020442297,
      "grad_norm": 1.3038439750671387,
      "learning_rate": 9.804247336718523e-05,
      "loss": 0.7484,
      "step": 10911
    },
    {
      "epoch": 2.0278758595056683,
      "grad_norm": 0.7218604683876038,
      "learning_rate": 9.802786684664274e-05,
      "loss": 0.7819,
      "step": 10912
    },
    {
      "epoch": 2.028061698569039,
      "grad_norm": 0.9532296061515808,
      "learning_rate": 9.801326036819208e-05,
      "loss": 1.137,
      "step": 10913
    },
    {
      "epoch": 2.0282475376324105,
      "grad_norm": 0.6444713473320007,
      "learning_rate": 9.799865393214492e-05,
      "loss": 0.838,
      "step": 10914
    },
    {
      "epoch": 2.0284333766957814,
      "grad_norm": 0.8020101189613342,
      "learning_rate": 9.79840475388131e-05,
      "loss": 0.8373,
      "step": 10915
    },
    {
      "epoch": 2.0286192157591527,
      "grad_norm": 0.8423665165901184,
      "learning_rate": 9.796944118850829e-05,
      "loss": 0.7276,
      "step": 10916
    },
    {
      "epoch": 2.0288050548225236,
      "grad_norm": 0.8215285539627075,
      "learning_rate": 9.79548348815423e-05,
      "loss": 0.9521,
      "step": 10917
    },
    {
      "epoch": 2.028990893885895,
      "grad_norm": 0.7488442659378052,
      "learning_rate": 9.794022861822684e-05,
      "loss": 0.8316,
      "step": 10918
    },
    {
      "epoch": 2.029176732949266,
      "grad_norm": 0.6915283799171448,
      "learning_rate": 9.792562239887363e-05,
      "loss": 0.5913,
      "step": 10919
    },
    {
      "epoch": 2.029362572012637,
      "grad_norm": 0.8294501304626465,
      "learning_rate": 9.791101622379448e-05,
      "loss": 1.0033,
      "step": 10920
    },
    {
      "epoch": 2.029548411076008,
      "grad_norm": 0.7450535297393799,
      "learning_rate": 9.789641009330111e-05,
      "loss": 0.8725,
      "step": 10921
    },
    {
      "epoch": 2.0297342501393794,
      "grad_norm": 0.7984308004379272,
      "learning_rate": 9.788180400770524e-05,
      "loss": 1.1632,
      "step": 10922
    },
    {
      "epoch": 2.0299200892027502,
      "grad_norm": 0.7514494061470032,
      "learning_rate": 9.786719796731866e-05,
      "loss": 1.0312,
      "step": 10923
    },
    {
      "epoch": 2.0301059282661216,
      "grad_norm": 0.7737922668457031,
      "learning_rate": 9.785259197245306e-05,
      "loss": 0.8648,
      "step": 10924
    },
    {
      "epoch": 2.0302917673294925,
      "grad_norm": 0.8706908822059631,
      "learning_rate": 9.783798602342021e-05,
      "loss": 1.1641,
      "step": 10925
    },
    {
      "epoch": 2.030477606392864,
      "grad_norm": 1.0741982460021973,
      "learning_rate": 9.782338012053183e-05,
      "loss": 0.8855,
      "step": 10926
    },
    {
      "epoch": 2.030663445456235,
      "grad_norm": 1.0242948532104492,
      "learning_rate": 9.780877426409966e-05,
      "loss": 1.0979,
      "step": 10927
    },
    {
      "epoch": 2.030849284519606,
      "grad_norm": 0.718450665473938,
      "learning_rate": 9.779416845443545e-05,
      "loss": 0.9463,
      "step": 10928
    },
    {
      "epoch": 2.0310351235829773,
      "grad_norm": 0.7844233512878418,
      "learning_rate": 9.777956269185092e-05,
      "loss": 0.859,
      "step": 10929
    },
    {
      "epoch": 2.031220962646348,
      "grad_norm": 0.820450484752655,
      "learning_rate": 9.77649569766578e-05,
      "loss": 0.8462,
      "step": 10930
    },
    {
      "epoch": 2.0314068017097195,
      "grad_norm": 0.7150939106941223,
      "learning_rate": 9.775035130916786e-05,
      "loss": 0.8532,
      "step": 10931
    },
    {
      "epoch": 2.0315926407730904,
      "grad_norm": 0.9015347957611084,
      "learning_rate": 9.77357456896928e-05,
      "loss": 0.8909,
      "step": 10932
    },
    {
      "epoch": 2.0317784798364618,
      "grad_norm": 0.7773807644844055,
      "learning_rate": 9.772114011854438e-05,
      "loss": 0.7619,
      "step": 10933
    },
    {
      "epoch": 2.0319643188998326,
      "grad_norm": 0.7356576323509216,
      "learning_rate": 9.77065345960343e-05,
      "loss": 1.0733,
      "step": 10934
    },
    {
      "epoch": 2.032150157963204,
      "grad_norm": 0.7767458558082581,
      "learning_rate": 9.769192912247434e-05,
      "loss": 0.9063,
      "step": 10935
    },
    {
      "epoch": 2.032335997026575,
      "grad_norm": 0.7118449211120605,
      "learning_rate": 9.767732369817615e-05,
      "loss": 0.8264,
      "step": 10936
    },
    {
      "epoch": 2.032521836089946,
      "grad_norm": 0.7526161670684814,
      "learning_rate": 9.766271832345152e-05,
      "loss": 0.836,
      "step": 10937
    },
    {
      "epoch": 2.032707675153317,
      "grad_norm": 0.7727689146995544,
      "learning_rate": 9.764811299861215e-05,
      "loss": 1.0169,
      "step": 10938
    },
    {
      "epoch": 2.0328935142166884,
      "grad_norm": 0.6880142092704773,
      "learning_rate": 9.763350772396975e-05,
      "loss": 0.8605,
      "step": 10939
    },
    {
      "epoch": 2.0330793532800593,
      "grad_norm": 0.7179078459739685,
      "learning_rate": 9.761890249983605e-05,
      "loss": 0.9297,
      "step": 10940
    },
    {
      "epoch": 2.0332651923434306,
      "grad_norm": 0.6645086407661438,
      "learning_rate": 9.760429732652284e-05,
      "loss": 0.941,
      "step": 10941
    },
    {
      "epoch": 2.0334510314068015,
      "grad_norm": 0.8322639465332031,
      "learning_rate": 9.758969220434178e-05,
      "loss": 0.9808,
      "step": 10942
    },
    {
      "epoch": 2.033636870470173,
      "grad_norm": 0.6776291728019714,
      "learning_rate": 9.757508713360461e-05,
      "loss": 0.7789,
      "step": 10943
    },
    {
      "epoch": 2.033822709533544,
      "grad_norm": 0.6935237646102905,
      "learning_rate": 9.756048211462303e-05,
      "loss": 0.8107,
      "step": 10944
    },
    {
      "epoch": 2.034008548596915,
      "grad_norm": 0.8299887180328369,
      "learning_rate": 9.754587714770878e-05,
      "loss": 0.8408,
      "step": 10945
    },
    {
      "epoch": 2.0341943876602864,
      "grad_norm": 0.652435839176178,
      "learning_rate": 9.753127223317357e-05,
      "loss": 0.961,
      "step": 10946
    },
    {
      "epoch": 2.0343802267236573,
      "grad_norm": 0.8443545699119568,
      "learning_rate": 9.751666737132913e-05,
      "loss": 1.1084,
      "step": 10947
    },
    {
      "epoch": 2.0345660657870286,
      "grad_norm": 0.6825294494628906,
      "learning_rate": 9.750206256248715e-05,
      "loss": 0.7776,
      "step": 10948
    },
    {
      "epoch": 2.0347519048503995,
      "grad_norm": 0.9838349223136902,
      "learning_rate": 9.748745780695934e-05,
      "loss": 0.9621,
      "step": 10949
    },
    {
      "epoch": 2.034937743913771,
      "grad_norm": 0.6445854902267456,
      "learning_rate": 9.747285310505747e-05,
      "loss": 0.8213,
      "step": 10950
    },
    {
      "epoch": 2.0351235829771417,
      "grad_norm": 0.8864513635635376,
      "learning_rate": 9.745824845709316e-05,
      "loss": 1.0444,
      "step": 10951
    },
    {
      "epoch": 2.035309422040513,
      "grad_norm": 0.5171553492546082,
      "learning_rate": 9.74436438633782e-05,
      "loss": 0.5602,
      "step": 10952
    },
    {
      "epoch": 2.035495261103884,
      "grad_norm": 0.6739189028739929,
      "learning_rate": 9.74290393242243e-05,
      "loss": 0.803,
      "step": 10953
    },
    {
      "epoch": 2.0356811001672552,
      "grad_norm": 0.7470421195030212,
      "learning_rate": 9.741443483994312e-05,
      "loss": 0.7937,
      "step": 10954
    },
    {
      "epoch": 2.035866939230626,
      "grad_norm": 0.7404161095619202,
      "learning_rate": 9.739983041084642e-05,
      "loss": 1.0081,
      "step": 10955
    },
    {
      "epoch": 2.0360527782939974,
      "grad_norm": 0.6297615766525269,
      "learning_rate": 9.738522603724584e-05,
      "loss": 0.6757,
      "step": 10956
    },
    {
      "epoch": 2.0362386173573683,
      "grad_norm": 0.6591044068336487,
      "learning_rate": 9.737062171945315e-05,
      "loss": 0.6592,
      "step": 10957
    },
    {
      "epoch": 2.0364244564207397,
      "grad_norm": 0.7246942520141602,
      "learning_rate": 9.735601745778001e-05,
      "loss": 0.8395,
      "step": 10958
    },
    {
      "epoch": 2.0366102954841105,
      "grad_norm": 0.8555113673210144,
      "learning_rate": 9.734141325253813e-05,
      "loss": 1.0245,
      "step": 10959
    },
    {
      "epoch": 2.036796134547482,
      "grad_norm": 0.8811112642288208,
      "learning_rate": 9.732680910403926e-05,
      "loss": 0.9338,
      "step": 10960
    },
    {
      "epoch": 2.036981973610853,
      "grad_norm": 0.7802111506462097,
      "learning_rate": 9.731220501259501e-05,
      "loss": 1.1103,
      "step": 10961
    },
    {
      "epoch": 2.037167812674224,
      "grad_norm": 0.8129925727844238,
      "learning_rate": 9.729760097851716e-05,
      "loss": 0.9776,
      "step": 10962
    },
    {
      "epoch": 2.0373536517375954,
      "grad_norm": 0.7644342184066772,
      "learning_rate": 9.728299700211739e-05,
      "loss": 0.835,
      "step": 10963
    },
    {
      "epoch": 2.0375394908009663,
      "grad_norm": 0.7132039070129395,
      "learning_rate": 9.726839308370738e-05,
      "loss": 0.9298,
      "step": 10964
    },
    {
      "epoch": 2.0377253298643376,
      "grad_norm": 0.7377307415008545,
      "learning_rate": 9.725378922359884e-05,
      "loss": 0.9102,
      "step": 10965
    },
    {
      "epoch": 2.0379111689277085,
      "grad_norm": 0.852078914642334,
      "learning_rate": 9.723918542210343e-05,
      "loss": 1.0642,
      "step": 10966
    },
    {
      "epoch": 2.03809700799108,
      "grad_norm": 0.6840757727622986,
      "learning_rate": 9.72245816795329e-05,
      "loss": 0.8916,
      "step": 10967
    },
    {
      "epoch": 2.0382828470544507,
      "grad_norm": 0.6280407905578613,
      "learning_rate": 9.72099779961989e-05,
      "loss": 0.9485,
      "step": 10968
    },
    {
      "epoch": 2.038468686117822,
      "grad_norm": 0.7283921241760254,
      "learning_rate": 9.719537437241312e-05,
      "loss": 0.9284,
      "step": 10969
    },
    {
      "epoch": 2.038654525181193,
      "grad_norm": 0.8217094540596008,
      "learning_rate": 9.718077080848727e-05,
      "loss": 0.7122,
      "step": 10970
    },
    {
      "epoch": 2.0388403642445643,
      "grad_norm": 0.7073930501937866,
      "learning_rate": 9.716616730473304e-05,
      "loss": 0.7266,
      "step": 10971
    },
    {
      "epoch": 2.039026203307935,
      "grad_norm": 0.6238510608673096,
      "learning_rate": 9.715156386146206e-05,
      "loss": 0.7414,
      "step": 10972
    },
    {
      "epoch": 2.0392120423713065,
      "grad_norm": 0.7381787300109863,
      "learning_rate": 9.71369604789861e-05,
      "loss": 0.9392,
      "step": 10973
    },
    {
      "epoch": 2.0393978814346774,
      "grad_norm": 0.8915388584136963,
      "learning_rate": 9.712235715761679e-05,
      "loss": 0.8417,
      "step": 10974
    },
    {
      "epoch": 2.0395837204980487,
      "grad_norm": 0.7346033453941345,
      "learning_rate": 9.710775389766587e-05,
      "loss": 1.0207,
      "step": 10975
    },
    {
      "epoch": 2.03976955956142,
      "grad_norm": 0.9085640907287598,
      "learning_rate": 9.709315069944493e-05,
      "loss": 0.9353,
      "step": 10976
    },
    {
      "epoch": 2.039955398624791,
      "grad_norm": 0.7272212505340576,
      "learning_rate": 9.707854756326575e-05,
      "loss": 1.0059,
      "step": 10977
    },
    {
      "epoch": 2.0401412376881622,
      "grad_norm": 0.8539851307868958,
      "learning_rate": 9.706394448943991e-05,
      "loss": 0.9126,
      "step": 10978
    },
    {
      "epoch": 2.040327076751533,
      "grad_norm": 0.696914792060852,
      "learning_rate": 9.704934147827914e-05,
      "loss": 0.7602,
      "step": 10979
    },
    {
      "epoch": 2.0405129158149045,
      "grad_norm": 0.858230710029602,
      "learning_rate": 9.703473853009514e-05,
      "loss": 1.1857,
      "step": 10980
    },
    {
      "epoch": 2.0406987548782753,
      "grad_norm": 0.8481776118278503,
      "learning_rate": 9.702013564519954e-05,
      "loss": 0.7737,
      "step": 10981
    },
    {
      "epoch": 2.0408845939416467,
      "grad_norm": 0.7733168601989746,
      "learning_rate": 9.700553282390401e-05,
      "loss": 0.9721,
      "step": 10982
    },
    {
      "epoch": 2.0410704330050176,
      "grad_norm": 0.8131083250045776,
      "learning_rate": 9.699093006652028e-05,
      "loss": 0.813,
      "step": 10983
    },
    {
      "epoch": 2.041256272068389,
      "grad_norm": 0.8533946871757507,
      "learning_rate": 9.697632737335996e-05,
      "loss": 1.1391,
      "step": 10984
    },
    {
      "epoch": 2.0414421111317598,
      "grad_norm": 0.7576166391372681,
      "learning_rate": 9.696172474473479e-05,
      "loss": 0.7313,
      "step": 10985
    },
    {
      "epoch": 2.041627950195131,
      "grad_norm": 0.664013147354126,
      "learning_rate": 9.694712218095634e-05,
      "loss": 0.9421,
      "step": 10986
    },
    {
      "epoch": 2.041813789258502,
      "grad_norm": 0.7320838570594788,
      "learning_rate": 9.693251968233635e-05,
      "loss": 0.7804,
      "step": 10987
    },
    {
      "epoch": 2.0419996283218733,
      "grad_norm": 0.8353871703147888,
      "learning_rate": 9.691791724918646e-05,
      "loss": 0.9285,
      "step": 10988
    },
    {
      "epoch": 2.042185467385244,
      "grad_norm": 0.7641454935073853,
      "learning_rate": 9.690331488181834e-05,
      "loss": 0.9363,
      "step": 10989
    },
    {
      "epoch": 2.0423713064486155,
      "grad_norm": 0.8495318293571472,
      "learning_rate": 9.688871258054366e-05,
      "loss": 0.9949,
      "step": 10990
    },
    {
      "epoch": 2.0425571455119864,
      "grad_norm": 0.8309895396232605,
      "learning_rate": 9.687411034567404e-05,
      "loss": 0.7591,
      "step": 10991
    },
    {
      "epoch": 2.0427429845753577,
      "grad_norm": 0.7717214822769165,
      "learning_rate": 9.68595081775212e-05,
      "loss": 1.1372,
      "step": 10992
    },
    {
      "epoch": 2.042928823638729,
      "grad_norm": 0.756056010723114,
      "learning_rate": 9.684490607639673e-05,
      "loss": 0.8602,
      "step": 10993
    },
    {
      "epoch": 2.0431146627021,
      "grad_norm": 0.7627288699150085,
      "learning_rate": 9.683030404261234e-05,
      "loss": 1.0895,
      "step": 10994
    },
    {
      "epoch": 2.0433005017654713,
      "grad_norm": 0.7354733943939209,
      "learning_rate": 9.68157020764797e-05,
      "loss": 0.9717,
      "step": 10995
    },
    {
      "epoch": 2.043486340828842,
      "grad_norm": 0.8014447689056396,
      "learning_rate": 9.68011001783104e-05,
      "loss": 1.1493,
      "step": 10996
    },
    {
      "epoch": 2.0436721798922135,
      "grad_norm": 0.6164006590843201,
      "learning_rate": 9.678649834841617e-05,
      "loss": 0.8292,
      "step": 10997
    },
    {
      "epoch": 2.0438580189555844,
      "grad_norm": 0.9240174889564514,
      "learning_rate": 9.677189658710858e-05,
      "loss": 0.8912,
      "step": 10998
    },
    {
      "epoch": 2.0440438580189557,
      "grad_norm": 0.7147821187973022,
      "learning_rate": 9.675729489469933e-05,
      "loss": 0.7899,
      "step": 10999
    },
    {
      "epoch": 2.0442296970823266,
      "grad_norm": 0.8751394152641296,
      "learning_rate": 9.674269327150007e-05,
      "loss": 1.1711,
      "step": 11000
    },
    {
      "epoch": 2.044415536145698,
      "grad_norm": 0.8109802603721619,
      "learning_rate": 9.67280917178224e-05,
      "loss": 0.8241,
      "step": 11001
    },
    {
      "epoch": 2.044601375209069,
      "grad_norm": 0.7171081900596619,
      "learning_rate": 9.671349023397803e-05,
      "loss": 0.8985,
      "step": 11002
    },
    {
      "epoch": 2.04478721427244,
      "grad_norm": 0.6551255583763123,
      "learning_rate": 9.669888882027853e-05,
      "loss": 0.8839,
      "step": 11003
    },
    {
      "epoch": 2.044973053335811,
      "grad_norm": 0.8627475500106812,
      "learning_rate": 9.66842874770356e-05,
      "loss": 0.9054,
      "step": 11004
    },
    {
      "epoch": 2.0451588923991824,
      "grad_norm": 0.8003649115562439,
      "learning_rate": 9.666968620456088e-05,
      "loss": 0.8746,
      "step": 11005
    },
    {
      "epoch": 2.0453447314625532,
      "grad_norm": 0.8886460065841675,
      "learning_rate": 9.6655085003166e-05,
      "loss": 1.2039,
      "step": 11006
    },
    {
      "epoch": 2.0455305705259246,
      "grad_norm": 0.8739449381828308,
      "learning_rate": 9.664048387316258e-05,
      "loss": 0.93,
      "step": 11007
    },
    {
      "epoch": 2.0457164095892955,
      "grad_norm": 0.8335513472557068,
      "learning_rate": 9.662588281486226e-05,
      "loss": 0.8358,
      "step": 11008
    },
    {
      "epoch": 2.045902248652667,
      "grad_norm": 0.7394416928291321,
      "learning_rate": 9.661128182857667e-05,
      "loss": 0.8069,
      "step": 11009
    },
    {
      "epoch": 2.046088087716038,
      "grad_norm": 0.7699256539344788,
      "learning_rate": 9.659668091461749e-05,
      "loss": 1.0213,
      "step": 11010
    },
    {
      "epoch": 2.046273926779409,
      "grad_norm": 0.739697277545929,
      "learning_rate": 9.658208007329628e-05,
      "loss": 0.9363,
      "step": 11011
    },
    {
      "epoch": 2.0464597658427803,
      "grad_norm": 0.7034657597541809,
      "learning_rate": 9.656747930492474e-05,
      "loss": 0.9542,
      "step": 11012
    },
    {
      "epoch": 2.046645604906151,
      "grad_norm": 0.633341908454895,
      "learning_rate": 9.655287860981444e-05,
      "loss": 0.8322,
      "step": 11013
    },
    {
      "epoch": 2.0468314439695225,
      "grad_norm": 0.7756519913673401,
      "learning_rate": 9.653827798827699e-05,
      "loss": 0.9888,
      "step": 11014
    },
    {
      "epoch": 2.0470172830328934,
      "grad_norm": 0.7440750002861023,
      "learning_rate": 9.652367744062412e-05,
      "loss": 0.9726,
      "step": 11015
    },
    {
      "epoch": 2.0472031220962648,
      "grad_norm": 0.7013724446296692,
      "learning_rate": 9.650907696716736e-05,
      "loss": 1.0856,
      "step": 11016
    },
    {
      "epoch": 2.0473889611596356,
      "grad_norm": 0.8220937252044678,
      "learning_rate": 9.649447656821839e-05,
      "loss": 1.0899,
      "step": 11017
    },
    {
      "epoch": 2.047574800223007,
      "grad_norm": 0.7048841118812561,
      "learning_rate": 9.647987624408877e-05,
      "loss": 0.8282,
      "step": 11018
    },
    {
      "epoch": 2.047760639286378,
      "grad_norm": 0.7343522906303406,
      "learning_rate": 9.646527599509015e-05,
      "loss": 0.8971,
      "step": 11019
    },
    {
      "epoch": 2.047946478349749,
      "grad_norm": 0.6640952825546265,
      "learning_rate": 9.645067582153418e-05,
      "loss": 0.7383,
      "step": 11020
    },
    {
      "epoch": 2.04813231741312,
      "grad_norm": 0.864271879196167,
      "learning_rate": 9.643607572373242e-05,
      "loss": 0.8369,
      "step": 11021
    },
    {
      "epoch": 2.0483181564764914,
      "grad_norm": 1.0475118160247803,
      "learning_rate": 9.642147570199652e-05,
      "loss": 0.8565,
      "step": 11022
    },
    {
      "epoch": 2.0485039955398623,
      "grad_norm": 0.6887284517288208,
      "learning_rate": 9.640687575663808e-05,
      "loss": 0.8654,
      "step": 11023
    },
    {
      "epoch": 2.0486898346032336,
      "grad_norm": 0.7000967264175415,
      "learning_rate": 9.639227588796866e-05,
      "loss": 0.8847,
      "step": 11024
    },
    {
      "epoch": 2.0488756736666045,
      "grad_norm": 0.832517147064209,
      "learning_rate": 9.637767609629999e-05,
      "loss": 1.0639,
      "step": 11025
    },
    {
      "epoch": 2.049061512729976,
      "grad_norm": 0.7449747920036316,
      "learning_rate": 9.636307638194357e-05,
      "loss": 0.9122,
      "step": 11026
    },
    {
      "epoch": 2.049247351793347,
      "grad_norm": 0.6630158424377441,
      "learning_rate": 9.634847674521106e-05,
      "loss": 0.8657,
      "step": 11027
    },
    {
      "epoch": 2.049433190856718,
      "grad_norm": 0.8289523720741272,
      "learning_rate": 9.633387718641404e-05,
      "loss": 0.9851,
      "step": 11028
    },
    {
      "epoch": 2.0496190299200894,
      "grad_norm": 0.7368113994598389,
      "learning_rate": 9.631927770586412e-05,
      "loss": 0.9755,
      "step": 11029
    },
    {
      "epoch": 2.0498048689834603,
      "grad_norm": 0.7921997904777527,
      "learning_rate": 9.630467830387291e-05,
      "loss": 1.1219,
      "step": 11030
    },
    {
      "epoch": 2.0499907080468316,
      "grad_norm": 0.9077422618865967,
      "learning_rate": 9.6290078980752e-05,
      "loss": 1.1224,
      "step": 11031
    },
    {
      "epoch": 2.0501765471102025,
      "grad_norm": 0.7921608090400696,
      "learning_rate": 9.627547973681299e-05,
      "loss": 0.8099,
      "step": 11032
    },
    {
      "epoch": 2.050362386173574,
      "grad_norm": 0.6978192925453186,
      "learning_rate": 9.626088057236745e-05,
      "loss": 0.9829,
      "step": 11033
    },
    {
      "epoch": 2.0505482252369447,
      "grad_norm": 0.6289072632789612,
      "learning_rate": 9.624628148772703e-05,
      "loss": 0.6646,
      "step": 11034
    },
    {
      "epoch": 2.050734064300316,
      "grad_norm": 0.6852868795394897,
      "learning_rate": 9.623168248320325e-05,
      "loss": 0.7644,
      "step": 11035
    },
    {
      "epoch": 2.050919903363687,
      "grad_norm": Infinity,
      "learning_rate": 9.623168248320325e-05,
      "loss": 1.7865,
      "step": 11036
    },
    {
      "epoch": 2.0511057424270582,
      "grad_norm": 0.774402379989624,
      "learning_rate": 9.621708355910776e-05,
      "loss": 0.8294,
      "step": 11037
    },
    {
      "epoch": 2.051291581490429,
      "grad_norm": 0.7371591329574585,
      "learning_rate": 9.620248471575214e-05,
      "loss": 0.7436,
      "step": 11038
    },
    {
      "epoch": 2.0514774205538004,
      "grad_norm": 0.7850064039230347,
      "learning_rate": 9.618788595344795e-05,
      "loss": 0.8809,
      "step": 11039
    },
    {
      "epoch": 2.0516632596171713,
      "grad_norm": 0.865877628326416,
      "learning_rate": 9.61732872725068e-05,
      "loss": 0.78,
      "step": 11040
    },
    {
      "epoch": 2.0518490986805427,
      "grad_norm": 0.663486659526825,
      "learning_rate": 9.615868867324028e-05,
      "loss": 0.7357,
      "step": 11041
    },
    {
      "epoch": 2.0520349377439135,
      "grad_norm": 0.7983995676040649,
      "learning_rate": 9.614409015595995e-05,
      "loss": 0.9285,
      "step": 11042
    },
    {
      "epoch": 2.052220776807285,
      "grad_norm": 0.7850358486175537,
      "learning_rate": 9.61294917209774e-05,
      "loss": 0.8746,
      "step": 11043
    },
    {
      "epoch": 2.052406615870656,
      "grad_norm": 0.866909384727478,
      "learning_rate": 9.61148933686042e-05,
      "loss": 1.0918,
      "step": 11044
    },
    {
      "epoch": 2.052592454934027,
      "grad_norm": 0.8137707710266113,
      "learning_rate": 9.610029509915195e-05,
      "loss": 1.0047,
      "step": 11045
    },
    {
      "epoch": 2.0527782939973984,
      "grad_norm": 0.6748443245887756,
      "learning_rate": 9.608569691293216e-05,
      "loss": 0.7952,
      "step": 11046
    },
    {
      "epoch": 2.0529641330607693,
      "grad_norm": 0.8559642434120178,
      "learning_rate": 9.607109881025648e-05,
      "loss": 1.0124,
      "step": 11047
    },
    {
      "epoch": 2.0531499721241406,
      "grad_norm": 0.8332653641700745,
      "learning_rate": 9.605650079143648e-05,
      "loss": 0.9663,
      "step": 11048
    },
    {
      "epoch": 2.0533358111875115,
      "grad_norm": 0.7888073921203613,
      "learning_rate": 9.604190285678368e-05,
      "loss": 0.7784,
      "step": 11049
    },
    {
      "epoch": 2.053521650250883,
      "grad_norm": 0.7318856120109558,
      "learning_rate": 9.602730500660967e-05,
      "loss": 0.6779,
      "step": 11050
    },
    {
      "epoch": 2.0537074893142537,
      "grad_norm": 0.8940858840942383,
      "learning_rate": 9.601270724122603e-05,
      "loss": 0.6288,
      "step": 11051
    },
    {
      "epoch": 2.053893328377625,
      "grad_norm": 0.8630791306495667,
      "learning_rate": 9.59981095609443e-05,
      "loss": 0.9083,
      "step": 11052
    },
    {
      "epoch": 2.054079167440996,
      "grad_norm": 0.7749754190444946,
      "learning_rate": 9.598351196607607e-05,
      "loss": 0.9646,
      "step": 11053
    },
    {
      "epoch": 2.0542650065043673,
      "grad_norm": 0.8278696537017822,
      "learning_rate": 9.596891445693287e-05,
      "loss": 0.7448,
      "step": 11054
    },
    {
      "epoch": 2.054450845567738,
      "grad_norm": 0.7356969714164734,
      "learning_rate": 9.595431703382628e-05,
      "loss": 1.0189,
      "step": 11055
    },
    {
      "epoch": 2.0546366846311095,
      "grad_norm": 0.8219599723815918,
      "learning_rate": 9.593971969706786e-05,
      "loss": 0.8709,
      "step": 11056
    },
    {
      "epoch": 2.0548225236944804,
      "grad_norm": 0.9181432127952576,
      "learning_rate": 9.59251224469691e-05,
      "loss": 1.0932,
      "step": 11057
    },
    {
      "epoch": 2.0550083627578517,
      "grad_norm": 0.6893907189369202,
      "learning_rate": 9.591052528384167e-05,
      "loss": 0.9475,
      "step": 11058
    },
    {
      "epoch": 2.055194201821223,
      "grad_norm": 0.7080320715904236,
      "learning_rate": 9.589592820799704e-05,
      "loss": 0.8971,
      "step": 11059
    },
    {
      "epoch": 2.055380040884594,
      "grad_norm": 0.5905109643936157,
      "learning_rate": 9.588133121974676e-05,
      "loss": 0.3737,
      "step": 11060
    },
    {
      "epoch": 2.0555658799479652,
      "grad_norm": 0.7928087711334229,
      "learning_rate": 9.586673431940242e-05,
      "loss": 0.7819,
      "step": 11061
    },
    {
      "epoch": 2.055751719011336,
      "grad_norm": 0.9699975848197937,
      "learning_rate": 9.585213750727553e-05,
      "loss": 0.8992,
      "step": 11062
    },
    {
      "epoch": 2.0559375580747075,
      "grad_norm": 1.4009530544281006,
      "learning_rate": 9.583754078367768e-05,
      "loss": 1.2486,
      "step": 11063
    },
    {
      "epoch": 2.0561233971380783,
      "grad_norm": 0.7332651615142822,
      "learning_rate": 9.582294414892036e-05,
      "loss": 1.0257,
      "step": 11064
    },
    {
      "epoch": 2.0563092362014497,
      "grad_norm": 0.9999315142631531,
      "learning_rate": 9.580834760331514e-05,
      "loss": 0.7446,
      "step": 11065
    },
    {
      "epoch": 2.0564950752648206,
      "grad_norm": 0.8058012127876282,
      "learning_rate": 9.579375114717351e-05,
      "loss": 1.0821,
      "step": 11066
    },
    {
      "epoch": 2.056680914328192,
      "grad_norm": 0.8406280279159546,
      "learning_rate": 9.577915478080705e-05,
      "loss": 0.9408,
      "step": 11067
    },
    {
      "epoch": 2.0568667533915628,
      "grad_norm": 0.6476280093193054,
      "learning_rate": 9.576455850452731e-05,
      "loss": 0.5505,
      "step": 11068
    },
    {
      "epoch": 2.057052592454934,
      "grad_norm": 0.9520523548126221,
      "learning_rate": 9.57499623186458e-05,
      "loss": 0.98,
      "step": 11069
    },
    {
      "epoch": 2.057238431518305,
      "grad_norm": 0.7537355422973633,
      "learning_rate": 9.573536622347404e-05,
      "loss": 0.9191,
      "step": 11070
    },
    {
      "epoch": 2.0574242705816763,
      "grad_norm": 0.8616262078285217,
      "learning_rate": 9.572077021932358e-05,
      "loss": 0.8966,
      "step": 11071
    },
    {
      "epoch": 2.057610109645047,
      "grad_norm": 0.7534152865409851,
      "learning_rate": 9.570617430650592e-05,
      "loss": 0.7814,
      "step": 11072
    },
    {
      "epoch": 2.0577959487084185,
      "grad_norm": 0.6804052591323853,
      "learning_rate": 9.569157848533262e-05,
      "loss": 0.8825,
      "step": 11073
    },
    {
      "epoch": 2.0579817877717894,
      "grad_norm": 0.7233737707138062,
      "learning_rate": 9.567698275611517e-05,
      "loss": 0.9317,
      "step": 11074
    },
    {
      "epoch": 2.0581676268351607,
      "grad_norm": 0.7929511666297913,
      "learning_rate": 9.566238711916513e-05,
      "loss": 0.8986,
      "step": 11075
    },
    {
      "epoch": 2.058353465898532,
      "grad_norm": 0.7047044634819031,
      "learning_rate": 9.564779157479397e-05,
      "loss": 1.0629,
      "step": 11076
    },
    {
      "epoch": 2.058539304961903,
      "grad_norm": 0.8283855319023132,
      "learning_rate": 9.563319612331324e-05,
      "loss": 1.1195,
      "step": 11077
    },
    {
      "epoch": 2.0587251440252743,
      "grad_norm": 0.8268432021141052,
      "learning_rate": 9.561860076503442e-05,
      "loss": 0.8749,
      "step": 11078
    },
    {
      "epoch": 2.058910983088645,
      "grad_norm": 0.7072595953941345,
      "learning_rate": 9.560400550026905e-05,
      "loss": 0.9702,
      "step": 11079
    },
    {
      "epoch": 2.0590968221520165,
      "grad_norm": 0.9471309185028076,
      "learning_rate": 9.558941032932864e-05,
      "loss": 0.9403,
      "step": 11080
    },
    {
      "epoch": 2.0592826612153874,
      "grad_norm": 0.8507811427116394,
      "learning_rate": 9.557481525252474e-05,
      "loss": 0.8969,
      "step": 11081
    },
    {
      "epoch": 2.0594685002787587,
      "grad_norm": 0.8363974690437317,
      "learning_rate": 9.556022027016878e-05,
      "loss": 1.0464,
      "step": 11082
    },
    {
      "epoch": 2.0596543393421296,
      "grad_norm": 0.9051037430763245,
      "learning_rate": 9.554562538257231e-05,
      "loss": 0.7263,
      "step": 11083
    },
    {
      "epoch": 2.059840178405501,
      "grad_norm": 1.8966175317764282,
      "learning_rate": 9.553103059004682e-05,
      "loss": 1.4728,
      "step": 11084
    },
    {
      "epoch": 2.060026017468872,
      "grad_norm": 0.829832136631012,
      "learning_rate": 9.551643589290383e-05,
      "loss": 0.9924,
      "step": 11085
    },
    {
      "epoch": 2.060211856532243,
      "grad_norm": 0.6673998236656189,
      "learning_rate": 9.550184129145479e-05,
      "loss": 0.8441,
      "step": 11086
    },
    {
      "epoch": 2.060397695595614,
      "grad_norm": 0.7418094873428345,
      "learning_rate": 9.548724678601125e-05,
      "loss": 1.1444,
      "step": 11087
    },
    {
      "epoch": 2.0605835346589854,
      "grad_norm": 0.8789692521095276,
      "learning_rate": 9.547265237688465e-05,
      "loss": 1.012,
      "step": 11088
    },
    {
      "epoch": 2.0607693737223562,
      "grad_norm": 0.7394304871559143,
      "learning_rate": 9.545805806438653e-05,
      "loss": 0.8909,
      "step": 11089
    },
    {
      "epoch": 2.0609552127857276,
      "grad_norm": 0.9010868072509766,
      "learning_rate": 9.544346384882838e-05,
      "loss": 1.042,
      "step": 11090
    },
    {
      "epoch": 2.0611410518490985,
      "grad_norm": 0.7490463852882385,
      "learning_rate": 9.542886973052169e-05,
      "loss": 0.796,
      "step": 11091
    },
    {
      "epoch": 2.06132689091247,
      "grad_norm": 0.6933886408805847,
      "learning_rate": 9.54142757097779e-05,
      "loss": 0.6588,
      "step": 11092
    },
    {
      "epoch": 2.061512729975841,
      "grad_norm": 0.7842268347740173,
      "learning_rate": 9.539968178690856e-05,
      "loss": 0.8955,
      "step": 11093
    },
    {
      "epoch": 2.061698569039212,
      "grad_norm": 0.7727258801460266,
      "learning_rate": 9.538508796222508e-05,
      "loss": 0.8295,
      "step": 11094
    },
    {
      "epoch": 2.0618844081025833,
      "grad_norm": 0.7276997566223145,
      "learning_rate": 9.537049423603901e-05,
      "loss": 1.0724,
      "step": 11095
    },
    {
      "epoch": 2.062070247165954,
      "grad_norm": 0.6727124452590942,
      "learning_rate": 9.535590060866177e-05,
      "loss": 1.0115,
      "step": 11096
    },
    {
      "epoch": 2.0622560862293255,
      "grad_norm": 0.956548273563385,
      "learning_rate": 9.53413070804049e-05,
      "loss": 1.1568,
      "step": 11097
    },
    {
      "epoch": 2.0624419252926964,
      "grad_norm": 0.7817228436470032,
      "learning_rate": 9.53267136515798e-05,
      "loss": 0.8645,
      "step": 11098
    },
    {
      "epoch": 2.0626277643560678,
      "grad_norm": 0.6998949646949768,
      "learning_rate": 9.531212032249795e-05,
      "loss": 0.7166,
      "step": 11099
    },
    {
      "epoch": 2.0628136034194386,
      "grad_norm": 0.8000270128250122,
      "learning_rate": 9.529752709347087e-05,
      "loss": 0.7585,
      "step": 11100
    },
    {
      "epoch": 2.06299944248281,
      "grad_norm": 0.801515519618988,
      "learning_rate": 9.528293396481005e-05,
      "loss": 0.7732,
      "step": 11101
    },
    {
      "epoch": 2.063185281546181,
      "grad_norm": 0.68977290391922,
      "learning_rate": 9.526834093682685e-05,
      "loss": 0.7811,
      "step": 11102
    },
    {
      "epoch": 2.063371120609552,
      "grad_norm": 0.7006689310073853,
      "learning_rate": 9.525374800983284e-05,
      "loss": 0.8202,
      "step": 11103
    },
    {
      "epoch": 2.063556959672923,
      "grad_norm": 0.7314426302909851,
      "learning_rate": 9.523915518413941e-05,
      "loss": 0.9052,
      "step": 11104
    },
    {
      "epoch": 2.0637427987362944,
      "grad_norm": 0.7322511672973633,
      "learning_rate": 9.522456246005806e-05,
      "loss": 0.9001,
      "step": 11105
    },
    {
      "epoch": 2.0639286377996653,
      "grad_norm": 0.7833496332168579,
      "learning_rate": 9.52099698379002e-05,
      "loss": 0.8426,
      "step": 11106
    },
    {
      "epoch": 2.0641144768630366,
      "grad_norm": 0.888405978679657,
      "learning_rate": 9.519537731797736e-05,
      "loss": 0.9917,
      "step": 11107
    },
    {
      "epoch": 2.064300315926408,
      "grad_norm": 0.7309452295303345,
      "learning_rate": 9.518078490060089e-05,
      "loss": 0.7145,
      "step": 11108
    },
    {
      "epoch": 2.064486154989779,
      "grad_norm": 0.690657377243042,
      "learning_rate": 9.51661925860823e-05,
      "loss": 0.7821,
      "step": 11109
    },
    {
      "epoch": 2.06467199405315,
      "grad_norm": 0.7759823203086853,
      "learning_rate": 9.515160037473304e-05,
      "loss": 0.9312,
      "step": 11110
    },
    {
      "epoch": 2.064857833116521,
      "grad_norm": 0.9517083764076233,
      "learning_rate": 9.513700826686458e-05,
      "loss": 1.0287,
      "step": 11111
    },
    {
      "epoch": 2.0650436721798924,
      "grad_norm": 0.9148969054222107,
      "learning_rate": 9.512241626278832e-05,
      "loss": 0.9626,
      "step": 11112
    },
    {
      "epoch": 2.0652295112432633,
      "grad_norm": 1.4553234577178955,
      "learning_rate": 9.510782436281572e-05,
      "loss": 1.398,
      "step": 11113
    },
    {
      "epoch": 2.0654153503066346,
      "grad_norm": 0.7028954029083252,
      "learning_rate": 9.509323256725821e-05,
      "loss": 0.8219,
      "step": 11114
    },
    {
      "epoch": 2.0656011893700055,
      "grad_norm": 0.7765356302261353,
      "learning_rate": 9.507864087642724e-05,
      "loss": 0.9148,
      "step": 11115
    },
    {
      "epoch": 2.065787028433377,
      "grad_norm": 0.7603330612182617,
      "learning_rate": 9.506404929063422e-05,
      "loss": 1.0486,
      "step": 11116
    },
    {
      "epoch": 2.0659728674967477,
      "grad_norm": 0.8426289558410645,
      "learning_rate": 9.50494578101906e-05,
      "loss": 1.0035,
      "step": 11117
    },
    {
      "epoch": 2.066158706560119,
      "grad_norm": 0.6924977898597717,
      "learning_rate": 9.503486643540781e-05,
      "loss": 0.7859,
      "step": 11118
    },
    {
      "epoch": 2.06634454562349,
      "grad_norm": 0.7882113456726074,
      "learning_rate": 9.502027516659728e-05,
      "loss": 0.8989,
      "step": 11119
    },
    {
      "epoch": 2.0665303846868612,
      "grad_norm": 0.6670895218849182,
      "learning_rate": 9.500568400407037e-05,
      "loss": 0.8255,
      "step": 11120
    },
    {
      "epoch": 2.066716223750232,
      "grad_norm": 0.7035497426986694,
      "learning_rate": 9.499109294813863e-05,
      "loss": 0.7877,
      "step": 11121
    },
    {
      "epoch": 2.0669020628136034,
      "grad_norm": 0.7977902889251709,
      "learning_rate": 9.497650199911341e-05,
      "loss": 0.9922,
      "step": 11122
    },
    {
      "epoch": 2.0670879018769743,
      "grad_norm": 0.6651736497879028,
      "learning_rate": 9.496191115730614e-05,
      "loss": 0.8098,
      "step": 11123
    },
    {
      "epoch": 2.0672737409403457,
      "grad_norm": 0.9503287076950073,
      "learning_rate": 9.494732042302821e-05,
      "loss": 0.91,
      "step": 11124
    },
    {
      "epoch": 2.067459580003717,
      "grad_norm": 0.6866540908813477,
      "learning_rate": 9.493272979659108e-05,
      "loss": 0.5869,
      "step": 11125
    },
    {
      "epoch": 2.067645419067088,
      "grad_norm": 0.8645631670951843,
      "learning_rate": 9.491813927830611e-05,
      "loss": 0.8996,
      "step": 11126
    },
    {
      "epoch": 2.067831258130459,
      "grad_norm": 0.7809982895851135,
      "learning_rate": 9.490354886848476e-05,
      "loss": 0.8874,
      "step": 11127
    },
    {
      "epoch": 2.06801709719383,
      "grad_norm": 0.7561979293823242,
      "learning_rate": 9.488895856743839e-05,
      "loss": 0.8648,
      "step": 11128
    },
    {
      "epoch": 2.0682029362572014,
      "grad_norm": 0.8173332810401917,
      "learning_rate": 9.487436837547846e-05,
      "loss": 1.2458,
      "step": 11129
    },
    {
      "epoch": 2.0683887753205723,
      "grad_norm": 0.8098888397216797,
      "learning_rate": 9.485977829291627e-05,
      "loss": 0.9267,
      "step": 11130
    },
    {
      "epoch": 2.0685746143839436,
      "grad_norm": 0.8903951048851013,
      "learning_rate": 9.484518832006335e-05,
      "loss": 0.8468,
      "step": 11131
    },
    {
      "epoch": 2.0687604534473145,
      "grad_norm": 0.7828354835510254,
      "learning_rate": 9.483059845723102e-05,
      "loss": 0.8767,
      "step": 11132
    },
    {
      "epoch": 2.068946292510686,
      "grad_norm": 0.7997280955314636,
      "learning_rate": 9.481600870473072e-05,
      "loss": 1.0065,
      "step": 11133
    },
    {
      "epoch": 2.0691321315740567,
      "grad_norm": 0.8885520100593567,
      "learning_rate": 9.48014190628738e-05,
      "loss": 1.0292,
      "step": 11134
    },
    {
      "epoch": 2.069317970637428,
      "grad_norm": 0.6353083848953247,
      "learning_rate": 9.478682953197169e-05,
      "loss": 0.7653,
      "step": 11135
    },
    {
      "epoch": 2.069503809700799,
      "grad_norm": 0.7411783933639526,
      "learning_rate": 9.477224011233573e-05,
      "loss": 0.7281,
      "step": 11136
    },
    {
      "epoch": 2.0696896487641703,
      "grad_norm": 0.6759365200996399,
      "learning_rate": 9.475765080427736e-05,
      "loss": 0.8458,
      "step": 11137
    },
    {
      "epoch": 2.069875487827541,
      "grad_norm": 0.7651349306106567,
      "learning_rate": 9.474306160810791e-05,
      "loss": 0.9969,
      "step": 11138
    },
    {
      "epoch": 2.0700613268909125,
      "grad_norm": 0.6856313347816467,
      "learning_rate": 9.472847252413881e-05,
      "loss": 0.8236,
      "step": 11139
    },
    {
      "epoch": 2.0702471659542834,
      "grad_norm": 0.7677355408668518,
      "learning_rate": 9.47138835526814e-05,
      "loss": 0.888,
      "step": 11140
    },
    {
      "epoch": 2.0704330050176547,
      "grad_norm": 0.9559133052825928,
      "learning_rate": 9.469929469404706e-05,
      "loss": 0.9105,
      "step": 11141
    },
    {
      "epoch": 2.070618844081026,
      "grad_norm": 0.7924931645393372,
      "learning_rate": 9.468470594854718e-05,
      "loss": 1.0257,
      "step": 11142
    },
    {
      "epoch": 2.070804683144397,
      "grad_norm": 0.6944732666015625,
      "learning_rate": 9.467011731649316e-05,
      "loss": 0.8962,
      "step": 11143
    },
    {
      "epoch": 2.0709905222077682,
      "grad_norm": 0.9529390335083008,
      "learning_rate": 9.465552879819633e-05,
      "loss": 0.8416,
      "step": 11144
    },
    {
      "epoch": 2.071176361271139,
      "grad_norm": 0.760055422782898,
      "learning_rate": 9.464094039396807e-05,
      "loss": 0.8923,
      "step": 11145
    },
    {
      "epoch": 2.0713622003345105,
      "grad_norm": 0.7828572988510132,
      "learning_rate": 9.462635210411974e-05,
      "loss": 0.9872,
      "step": 11146
    },
    {
      "epoch": 2.0715480393978813,
      "grad_norm": 0.746248185634613,
      "learning_rate": 9.461176392896271e-05,
      "loss": 0.9141,
      "step": 11147
    },
    {
      "epoch": 2.0717338784612527,
      "grad_norm": 0.8269466161727905,
      "learning_rate": 9.459717586880832e-05,
      "loss": 0.78,
      "step": 11148
    },
    {
      "epoch": 2.0719197175246236,
      "grad_norm": 0.7321226596832275,
      "learning_rate": 9.458258792396796e-05,
      "loss": 1.0112,
      "step": 11149
    },
    {
      "epoch": 2.072105556587995,
      "grad_norm": 0.762174129486084,
      "learning_rate": 9.456800009475293e-05,
      "loss": 1.0043,
      "step": 11150
    },
    {
      "epoch": 2.0722913956513658,
      "grad_norm": 0.737138569355011,
      "learning_rate": 9.45534123814746e-05,
      "loss": 0.7323,
      "step": 11151
    },
    {
      "epoch": 2.072477234714737,
      "grad_norm": 0.8031544089317322,
      "learning_rate": 9.453882478444436e-05,
      "loss": 1.0911,
      "step": 11152
    },
    {
      "epoch": 2.072663073778108,
      "grad_norm": 0.9331284761428833,
      "learning_rate": 9.452423730397357e-05,
      "loss": 1.1431,
      "step": 11153
    },
    {
      "epoch": 2.0728489128414793,
      "grad_norm": 0.6606505513191223,
      "learning_rate": 9.450964994037349e-05,
      "loss": 0.8519,
      "step": 11154
    },
    {
      "epoch": 2.07303475190485,
      "grad_norm": 0.7464000582695007,
      "learning_rate": 9.449506269395553e-05,
      "loss": 0.7497,
      "step": 11155
    },
    {
      "epoch": 2.0732205909682215,
      "grad_norm": 0.7550594806671143,
      "learning_rate": 9.448047556503102e-05,
      "loss": 0.9044,
      "step": 11156
    },
    {
      "epoch": 2.073406430031593,
      "grad_norm": 0.7202317118644714,
      "learning_rate": 9.446588855391129e-05,
      "loss": 0.6572,
      "step": 11157
    },
    {
      "epoch": 2.0735922690949637,
      "grad_norm": 0.751534104347229,
      "learning_rate": 9.445130166090765e-05,
      "loss": 1.0057,
      "step": 11158
    },
    {
      "epoch": 2.073778108158335,
      "grad_norm": 0.7617736458778381,
      "learning_rate": 9.443671488633146e-05,
      "loss": 0.8516,
      "step": 11159
    },
    {
      "epoch": 2.073963947221706,
      "grad_norm": 0.8996736407279968,
      "learning_rate": 9.442212823049404e-05,
      "loss": 0.8691,
      "step": 11160
    },
    {
      "epoch": 2.0741497862850773,
      "grad_norm": 1.1209017038345337,
      "learning_rate": 9.440754169370673e-05,
      "loss": 1.1305,
      "step": 11161
    },
    {
      "epoch": 2.074335625348448,
      "grad_norm": 0.9391574859619141,
      "learning_rate": 9.439295527628081e-05,
      "loss": 1.184,
      "step": 11162
    },
    {
      "epoch": 2.0745214644118195,
      "grad_norm": 0.8048052787780762,
      "learning_rate": 9.437836897852769e-05,
      "loss": 0.7383,
      "step": 11163
    },
    {
      "epoch": 2.0747073034751904,
      "grad_norm": 0.7512253522872925,
      "learning_rate": 9.43637828007586e-05,
      "loss": 0.8006,
      "step": 11164
    },
    {
      "epoch": 2.0748931425385617,
      "grad_norm": 0.7666751742362976,
      "learning_rate": 9.434919674328493e-05,
      "loss": 0.6582,
      "step": 11165
    },
    {
      "epoch": 2.0750789816019326,
      "grad_norm": 0.7968409061431885,
      "learning_rate": 9.433461080641793e-05,
      "loss": 0.8598,
      "step": 11166
    },
    {
      "epoch": 2.075264820665304,
      "grad_norm": 0.8389952182769775,
      "learning_rate": 9.432002499046897e-05,
      "loss": 0.7596,
      "step": 11167
    },
    {
      "epoch": 2.075450659728675,
      "grad_norm": 0.8382617235183716,
      "learning_rate": 9.43054392957493e-05,
      "loss": 0.9456,
      "step": 11168
    },
    {
      "epoch": 2.075636498792046,
      "grad_norm": 0.7901649475097656,
      "learning_rate": 9.429085372257027e-05,
      "loss": 0.7259,
      "step": 11169
    },
    {
      "epoch": 2.075822337855417,
      "grad_norm": 0.6912295818328857,
      "learning_rate": 9.427626827124317e-05,
      "loss": 0.742,
      "step": 11170
    },
    {
      "epoch": 2.0760081769187884,
      "grad_norm": 0.7385955452919006,
      "learning_rate": 9.426168294207928e-05,
      "loss": 0.7213,
      "step": 11171
    },
    {
      "epoch": 2.0761940159821592,
      "grad_norm": 0.7250804901123047,
      "learning_rate": 9.42470977353899e-05,
      "loss": 0.8584,
      "step": 11172
    },
    {
      "epoch": 2.0763798550455306,
      "grad_norm": 0.8665145635604858,
      "learning_rate": 9.423251265148641e-05,
      "loss": 0.972,
      "step": 11173
    },
    {
      "epoch": 2.076565694108902,
      "grad_norm": 0.9986973404884338,
      "learning_rate": 9.421792769068e-05,
      "loss": 1.2396,
      "step": 11174
    },
    {
      "epoch": 2.076751533172273,
      "grad_norm": 0.737506628036499,
      "learning_rate": 9.420334285328203e-05,
      "loss": 1.1461,
      "step": 11175
    },
    {
      "epoch": 2.076937372235644,
      "grad_norm": 0.8702325224876404,
      "learning_rate": 9.418875813960372e-05,
      "loss": 0.9769,
      "step": 11176
    },
    {
      "epoch": 2.077123211299015,
      "grad_norm": 1.0117233991622925,
      "learning_rate": 9.417417354995645e-05,
      "loss": 1.0833,
      "step": 11177
    },
    {
      "epoch": 2.0773090503623863,
      "grad_norm": 0.8385712504386902,
      "learning_rate": 9.41595890846514e-05,
      "loss": 0.8416,
      "step": 11178
    },
    {
      "epoch": 2.077494889425757,
      "grad_norm": 0.8949857950210571,
      "learning_rate": 9.414500474399993e-05,
      "loss": 0.8885,
      "step": 11179
    },
    {
      "epoch": 2.0776807284891285,
      "grad_norm": 0.7184500098228455,
      "learning_rate": 9.413042052831327e-05,
      "loss": 0.946,
      "step": 11180
    },
    {
      "epoch": 2.0778665675524994,
      "grad_norm": 0.6816924214363098,
      "learning_rate": 9.41158364379027e-05,
      "loss": 0.7566,
      "step": 11181
    },
    {
      "epoch": 2.0780524066158708,
      "grad_norm": 0.7060255408287048,
      "learning_rate": 9.410125247307952e-05,
      "loss": 0.6378,
      "step": 11182
    },
    {
      "epoch": 2.0782382456792416,
      "grad_norm": 0.7339649796485901,
      "learning_rate": 9.408666863415497e-05,
      "loss": 1.0362,
      "step": 11183
    },
    {
      "epoch": 2.078424084742613,
      "grad_norm": 0.815864622592926,
      "learning_rate": 9.407208492144031e-05,
      "loss": 0.9541,
      "step": 11184
    },
    {
      "epoch": 2.078609923805984,
      "grad_norm": 0.746654212474823,
      "learning_rate": 9.405750133524687e-05,
      "loss": 0.9796,
      "step": 11185
    },
    {
      "epoch": 2.078795762869355,
      "grad_norm": 0.7695572972297668,
      "learning_rate": 9.404291787588584e-05,
      "loss": 0.8375,
      "step": 11186
    },
    {
      "epoch": 2.078981601932726,
      "grad_norm": 0.7486796379089355,
      "learning_rate": 9.402833454366854e-05,
      "loss": 0.5969,
      "step": 11187
    },
    {
      "epoch": 2.0791674409960974,
      "grad_norm": 0.6751930713653564,
      "learning_rate": 9.401375133890615e-05,
      "loss": 0.9364,
      "step": 11188
    },
    {
      "epoch": 2.0793532800594683,
      "grad_norm": 0.9874372482299805,
      "learning_rate": 9.399916826191001e-05,
      "loss": 1.09,
      "step": 11189
    },
    {
      "epoch": 2.0795391191228396,
      "grad_norm": 0.7594035267829895,
      "learning_rate": 9.398458531299128e-05,
      "loss": 1.1015,
      "step": 11190
    },
    {
      "epoch": 2.079724958186211,
      "grad_norm": 6.832021713256836,
      "learning_rate": 9.397000249246126e-05,
      "loss": 1.9706,
      "step": 11191
    },
    {
      "epoch": 2.079910797249582,
      "grad_norm": 0.893206775188446,
      "learning_rate": 9.395541980063122e-05,
      "loss": 1.2385,
      "step": 11192
    },
    {
      "epoch": 2.080096636312953,
      "grad_norm": 0.7244065999984741,
      "learning_rate": 9.394083723781232e-05,
      "loss": 0.8494,
      "step": 11193
    },
    {
      "epoch": 2.080282475376324,
      "grad_norm": 0.8250991702079773,
      "learning_rate": 9.392625480431587e-05,
      "loss": 1.0322,
      "step": 11194
    },
    {
      "epoch": 2.0804683144396954,
      "grad_norm": 0.64857017993927,
      "learning_rate": 9.391167250045311e-05,
      "loss": 0.828,
      "step": 11195
    },
    {
      "epoch": 2.0806541535030663,
      "grad_norm": 0.7038189172744751,
      "learning_rate": 9.389709032653526e-05,
      "loss": 0.9225,
      "step": 11196
    },
    {
      "epoch": 2.0808399925664376,
      "grad_norm": 0.7618677020072937,
      "learning_rate": 9.388250828287353e-05,
      "loss": 0.9054,
      "step": 11197
    },
    {
      "epoch": 2.0810258316298085,
      "grad_norm": 0.9729681015014648,
      "learning_rate": 9.386792636977915e-05,
      "loss": 0.9982,
      "step": 11198
    },
    {
      "epoch": 2.08121167069318,
      "grad_norm": 0.7468367218971252,
      "learning_rate": 9.385334458756338e-05,
      "loss": 0.7856,
      "step": 11199
    },
    {
      "epoch": 2.0813975097565507,
      "grad_norm": 0.9042201638221741,
      "learning_rate": 9.383876293653741e-05,
      "loss": 0.7959,
      "step": 11200
    },
    {
      "epoch": 2.081583348819922,
      "grad_norm": 0.8591930270195007,
      "learning_rate": 9.382418141701246e-05,
      "loss": 0.9576,
      "step": 11201
    },
    {
      "epoch": 2.081769187883293,
      "grad_norm": 0.8453245162963867,
      "learning_rate": 9.380960002929979e-05,
      "loss": 1.1545,
      "step": 11202
    },
    {
      "epoch": 2.0819550269466642,
      "grad_norm": 0.9259535670280457,
      "learning_rate": 9.379501877371055e-05,
      "loss": 0.8565,
      "step": 11203
    },
    {
      "epoch": 2.082140866010035,
      "grad_norm": 0.7732197046279907,
      "learning_rate": 9.378043765055597e-05,
      "loss": 0.6425,
      "step": 11204
    },
    {
      "epoch": 2.0823267050734064,
      "grad_norm": 0.7024487257003784,
      "learning_rate": 9.376585666014733e-05,
      "loss": 0.8923,
      "step": 11205
    },
    {
      "epoch": 2.0825125441367778,
      "grad_norm": 0.739058256149292,
      "learning_rate": 9.375127580279574e-05,
      "loss": 0.8377,
      "step": 11206
    },
    {
      "epoch": 2.0826983832001487,
      "grad_norm": 0.7641466856002808,
      "learning_rate": 9.373669507881249e-05,
      "loss": 0.953,
      "step": 11207
    },
    {
      "epoch": 2.08288422226352,
      "grad_norm": 0.6840345859527588,
      "learning_rate": 9.372211448850869e-05,
      "loss": 1.0087,
      "step": 11208
    },
    {
      "epoch": 2.083070061326891,
      "grad_norm": 0.8499522805213928,
      "learning_rate": 9.370753403219562e-05,
      "loss": 0.9861,
      "step": 11209
    },
    {
      "epoch": 2.083255900390262,
      "grad_norm": 0.8159723281860352,
      "learning_rate": 9.369295371018442e-05,
      "loss": 1.0673,
      "step": 11210
    },
    {
      "epoch": 2.083441739453633,
      "grad_norm": 0.6602420806884766,
      "learning_rate": 9.367837352278629e-05,
      "loss": 0.703,
      "step": 11211
    },
    {
      "epoch": 2.0836275785170044,
      "grad_norm": 0.8844058513641357,
      "learning_rate": 9.366379347031245e-05,
      "loss": 0.9038,
      "step": 11212
    },
    {
      "epoch": 2.0838134175803753,
      "grad_norm": 0.7467980980873108,
      "learning_rate": 9.364921355307404e-05,
      "loss": 0.6782,
      "step": 11213
    },
    {
      "epoch": 2.0839992566437466,
      "grad_norm": 0.6584466099739075,
      "learning_rate": 9.363463377138225e-05,
      "loss": 0.8225,
      "step": 11214
    },
    {
      "epoch": 2.0841850957071175,
      "grad_norm": 0.7740834951400757,
      "learning_rate": 9.362005412554832e-05,
      "loss": 0.8503,
      "step": 11215
    },
    {
      "epoch": 2.084370934770489,
      "grad_norm": 0.7651553153991699,
      "learning_rate": 9.360547461588339e-05,
      "loss": 1.0257,
      "step": 11216
    },
    {
      "epoch": 2.0845567738338597,
      "grad_norm": 0.8948690295219421,
      "learning_rate": 9.359089524269862e-05,
      "loss": 1.1734,
      "step": 11217
    },
    {
      "epoch": 2.084742612897231,
      "grad_norm": 0.8161410689353943,
      "learning_rate": 9.357631600630518e-05,
      "loss": 1.2433,
      "step": 11218
    },
    {
      "epoch": 2.084928451960602,
      "grad_norm": 0.7995330095291138,
      "learning_rate": 9.356173690701427e-05,
      "loss": 1.157,
      "step": 11219
    },
    {
      "epoch": 2.0851142910239733,
      "grad_norm": 0.7776280045509338,
      "learning_rate": 9.354715794513702e-05,
      "loss": 0.8992,
      "step": 11220
    },
    {
      "epoch": 2.085300130087344,
      "grad_norm": 0.8435629606246948,
      "learning_rate": 9.353257912098462e-05,
      "loss": 0.7887,
      "step": 11221
    },
    {
      "epoch": 2.0854859691507155,
      "grad_norm": 0.7178513407707214,
      "learning_rate": 9.351800043486823e-05,
      "loss": 0.9152,
      "step": 11222
    },
    {
      "epoch": 2.0856718082140864,
      "grad_norm": 0.8578248620033264,
      "learning_rate": 9.350342188709898e-05,
      "loss": 0.9122,
      "step": 11223
    },
    {
      "epoch": 2.0858576472774577,
      "grad_norm": 0.7572755217552185,
      "learning_rate": 9.348884347798805e-05,
      "loss": 0.8292,
      "step": 11224
    },
    {
      "epoch": 2.086043486340829,
      "grad_norm": 0.8700441718101501,
      "learning_rate": 9.347426520784656e-05,
      "loss": 1.1506,
      "step": 11225
    },
    {
      "epoch": 2.0862293254042,
      "grad_norm": 0.8453944325447083,
      "learning_rate": 9.345968707698569e-05,
      "loss": 1.0707,
      "step": 11226
    },
    {
      "epoch": 2.0864151644675712,
      "grad_norm": 0.7037955522537231,
      "learning_rate": 9.34451090857166e-05,
      "loss": 0.8334,
      "step": 11227
    },
    {
      "epoch": 2.086601003530942,
      "grad_norm": 0.8542226552963257,
      "learning_rate": 9.343053123435038e-05,
      "loss": 0.7795,
      "step": 11228
    },
    {
      "epoch": 2.0867868425943135,
      "grad_norm": 0.895807147026062,
      "learning_rate": 9.341595352319822e-05,
      "loss": 0.7925,
      "step": 11229
    },
    {
      "epoch": 2.0869726816576843,
      "grad_norm": 0.8979935050010681,
      "learning_rate": 9.340137595257122e-05,
      "loss": 1.0552,
      "step": 11230
    },
    {
      "epoch": 2.0871585207210557,
      "grad_norm": 0.7536588311195374,
      "learning_rate": 9.338679852278052e-05,
      "loss": 0.9038,
      "step": 11231
    },
    {
      "epoch": 2.0873443597844266,
      "grad_norm": 0.7497516870498657,
      "learning_rate": 9.337222123413729e-05,
      "loss": 0.8808,
      "step": 11232
    },
    {
      "epoch": 2.087530198847798,
      "grad_norm": 0.810566782951355,
      "learning_rate": 9.335764408695259e-05,
      "loss": 0.9248,
      "step": 11233
    },
    {
      "epoch": 2.0877160379111688,
      "grad_norm": 0.7648537158966064,
      "learning_rate": 9.334306708153761e-05,
      "loss": 0.984,
      "step": 11234
    },
    {
      "epoch": 2.08790187697454,
      "grad_norm": 0.9524139165878296,
      "learning_rate": 9.332849021820339e-05,
      "loss": 0.8884,
      "step": 11235
    },
    {
      "epoch": 2.088087716037911,
      "grad_norm": 0.6822572350502014,
      "learning_rate": 9.331391349726112e-05,
      "loss": 0.869,
      "step": 11236
    },
    {
      "epoch": 2.0882735551012823,
      "grad_norm": 0.8357948064804077,
      "learning_rate": 9.329933691902191e-05,
      "loss": 1.0129,
      "step": 11237
    },
    {
      "epoch": 2.088459394164653,
      "grad_norm": 0.8944813013076782,
      "learning_rate": 9.328476048379684e-05,
      "loss": 0.9413,
      "step": 11238
    },
    {
      "epoch": 2.0886452332280245,
      "grad_norm": 0.9181443452835083,
      "learning_rate": 9.327018419189705e-05,
      "loss": 1.0597,
      "step": 11239
    },
    {
      "epoch": 2.088831072291396,
      "grad_norm": 0.8279556632041931,
      "learning_rate": 9.32556080436336e-05,
      "loss": 0.8327,
      "step": 11240
    },
    {
      "epoch": 2.0890169113547667,
      "grad_norm": 0.7568429708480835,
      "learning_rate": 9.324103203931763e-05,
      "loss": 0.7845,
      "step": 11241
    },
    {
      "epoch": 2.089202750418138,
      "grad_norm": 0.8147261738777161,
      "learning_rate": 9.322645617926026e-05,
      "loss": 0.8945,
      "step": 11242
    },
    {
      "epoch": 2.089388589481509,
      "grad_norm": 0.7661022543907166,
      "learning_rate": 9.321188046377254e-05,
      "loss": 0.8606,
      "step": 11243
    },
    {
      "epoch": 2.0895744285448803,
      "grad_norm": 0.7411051392555237,
      "learning_rate": 9.319730489316559e-05,
      "loss": 0.7371,
      "step": 11244
    },
    {
      "epoch": 2.089760267608251,
      "grad_norm": 0.9050106406211853,
      "learning_rate": 9.318272946775049e-05,
      "loss": 0.9799,
      "step": 11245
    },
    {
      "epoch": 2.0899461066716225,
      "grad_norm": 0.7047238349914551,
      "learning_rate": 9.316815418783829e-05,
      "loss": 0.9825,
      "step": 11246
    },
    {
      "epoch": 2.0901319457349934,
      "grad_norm": 0.8626230955123901,
      "learning_rate": 9.315357905374017e-05,
      "loss": 1.0136,
      "step": 11247
    },
    {
      "epoch": 2.0903177847983647,
      "grad_norm": 0.7386853098869324,
      "learning_rate": 9.313900406576712e-05,
      "loss": 0.8804,
      "step": 11248
    },
    {
      "epoch": 2.0905036238617356,
      "grad_norm": 0.7710316777229309,
      "learning_rate": 9.31244292242303e-05,
      "loss": 0.7352,
      "step": 11249
    },
    {
      "epoch": 2.090689462925107,
      "grad_norm": 0.7478527426719666,
      "learning_rate": 9.31098545294407e-05,
      "loss": 0.9791,
      "step": 11250
    },
    {
      "epoch": 2.090875301988478,
      "grad_norm": 0.7317497730255127,
      "learning_rate": 9.309527998170944e-05,
      "loss": 0.935,
      "step": 11251
    },
    {
      "epoch": 2.091061141051849,
      "grad_norm": 0.8391066789627075,
      "learning_rate": 9.308070558134759e-05,
      "loss": 0.9382,
      "step": 11252
    },
    {
      "epoch": 2.09124698011522,
      "grad_norm": 0.7186779975891113,
      "learning_rate": 9.30661313286662e-05,
      "loss": 0.9443,
      "step": 11253
    },
    {
      "epoch": 2.0914328191785914,
      "grad_norm": 0.6773986220359802,
      "learning_rate": 9.305155722397634e-05,
      "loss": 0.8425,
      "step": 11254
    },
    {
      "epoch": 2.0916186582419622,
      "grad_norm": 0.7712244391441345,
      "learning_rate": 9.303698326758905e-05,
      "loss": 1.0549,
      "step": 11255
    },
    {
      "epoch": 2.0918044973053336,
      "grad_norm": 0.7316671013832092,
      "learning_rate": 9.30224094598154e-05,
      "loss": 0.9495,
      "step": 11256
    },
    {
      "epoch": 2.091990336368705,
      "grad_norm": 0.7815166115760803,
      "learning_rate": 9.300783580096647e-05,
      "loss": 1.0787,
      "step": 11257
    },
    {
      "epoch": 2.092176175432076,
      "grad_norm": 0.7134209275245667,
      "learning_rate": 9.299326229135326e-05,
      "loss": 0.7159,
      "step": 11258
    },
    {
      "epoch": 2.092362014495447,
      "grad_norm": 0.8529147505760193,
      "learning_rate": 9.297868893128689e-05,
      "loss": 0.8936,
      "step": 11259
    },
    {
      "epoch": 2.092547853558818,
      "grad_norm": 0.8664206266403198,
      "learning_rate": 9.296411572107831e-05,
      "loss": 0.953,
      "step": 11260
    },
    {
      "epoch": 2.0927336926221893,
      "grad_norm": 0.7702050805091858,
      "learning_rate": 9.294954266103862e-05,
      "loss": 0.8955,
      "step": 11261
    },
    {
      "epoch": 2.09291953168556,
      "grad_norm": 0.7006514072418213,
      "learning_rate": 9.293496975147885e-05,
      "loss": 0.8528,
      "step": 11262
    },
    {
      "epoch": 2.0931053707489315,
      "grad_norm": 0.7274358868598938,
      "learning_rate": 9.292039699271003e-05,
      "loss": 0.8714,
      "step": 11263
    },
    {
      "epoch": 2.0932912098123024,
      "grad_norm": 0.717570960521698,
      "learning_rate": 9.290582438504319e-05,
      "loss": 1.0414,
      "step": 11264
    },
    {
      "epoch": 2.0934770488756738,
      "grad_norm": 0.7226032018661499,
      "learning_rate": 9.289125192878933e-05,
      "loss": 0.9059,
      "step": 11265
    },
    {
      "epoch": 2.0936628879390446,
      "grad_norm": 0.7209280729293823,
      "learning_rate": 9.287667962425952e-05,
      "loss": 0.9195,
      "step": 11266
    },
    {
      "epoch": 2.093848727002416,
      "grad_norm": 0.7089038491249084,
      "learning_rate": 9.286210747176473e-05,
      "loss": 0.8997,
      "step": 11267
    },
    {
      "epoch": 2.094034566065787,
      "grad_norm": 0.8167850971221924,
      "learning_rate": 9.284753547161601e-05,
      "loss": 0.9131,
      "step": 11268
    },
    {
      "epoch": 2.094220405129158,
      "grad_norm": 0.6800757050514221,
      "learning_rate": 9.28329636241244e-05,
      "loss": 0.7618,
      "step": 11269
    },
    {
      "epoch": 2.094406244192529,
      "grad_norm": 0.8122108578681946,
      "learning_rate": 9.281839192960087e-05,
      "loss": 0.791,
      "step": 11270
    },
    {
      "epoch": 2.0945920832559004,
      "grad_norm": 0.7946428656578064,
      "learning_rate": 9.280382038835644e-05,
      "loss": 0.8722,
      "step": 11271
    },
    {
      "epoch": 2.0947779223192713,
      "grad_norm": 0.8282971382141113,
      "learning_rate": 9.278924900070212e-05,
      "loss": 0.959,
      "step": 11272
    },
    {
      "epoch": 2.0949637613826426,
      "grad_norm": 0.7812013030052185,
      "learning_rate": 9.277467776694889e-05,
      "loss": 0.9049,
      "step": 11273
    },
    {
      "epoch": 2.095149600446014,
      "grad_norm": 0.7257639169692993,
      "learning_rate": 9.27601066874078e-05,
      "loss": 0.9538,
      "step": 11274
    },
    {
      "epoch": 2.095335439509385,
      "grad_norm": 0.7436121702194214,
      "learning_rate": 9.274553576238977e-05,
      "loss": 1.0246,
      "step": 11275
    },
    {
      "epoch": 2.095521278572756,
      "grad_norm": 0.7253244519233704,
      "learning_rate": 9.273096499220585e-05,
      "loss": 0.7533,
      "step": 11276
    },
    {
      "epoch": 2.095707117636127,
      "grad_norm": 1.0669963359832764,
      "learning_rate": 9.271639437716698e-05,
      "loss": 1.0058,
      "step": 11277
    },
    {
      "epoch": 2.0958929566994984,
      "grad_norm": 0.8048694729804993,
      "learning_rate": 9.270182391758419e-05,
      "loss": 0.958,
      "step": 11278
    },
    {
      "epoch": 2.0960787957628693,
      "grad_norm": 0.7488383054733276,
      "learning_rate": 9.268725361376847e-05,
      "loss": 0.835,
      "step": 11279
    },
    {
      "epoch": 2.0962646348262406,
      "grad_norm": 0.8129644393920898,
      "learning_rate": 9.267268346603074e-05,
      "loss": 0.8263,
      "step": 11280
    },
    {
      "epoch": 2.0964504738896115,
      "grad_norm": 0.8920313119888306,
      "learning_rate": 9.265811347468201e-05,
      "loss": 1.0749,
      "step": 11281
    },
    {
      "epoch": 2.096636312952983,
      "grad_norm": 0.7410893440246582,
      "learning_rate": 9.264354364003327e-05,
      "loss": 0.8763,
      "step": 11282
    },
    {
      "epoch": 2.0968221520163537,
      "grad_norm": 0.7437509298324585,
      "learning_rate": 9.262897396239544e-05,
      "loss": 1.0144,
      "step": 11283
    },
    {
      "epoch": 2.097007991079725,
      "grad_norm": 0.8066478967666626,
      "learning_rate": 9.261440444207955e-05,
      "loss": 0.9315,
      "step": 11284
    },
    {
      "epoch": 2.097193830143096,
      "grad_norm": 0.8381285667419434,
      "learning_rate": 9.259983507939649e-05,
      "loss": 0.9115,
      "step": 11285
    },
    {
      "epoch": 2.0973796692064672,
      "grad_norm": 0.8904730081558228,
      "learning_rate": 9.258526587465727e-05,
      "loss": 0.8368,
      "step": 11286
    },
    {
      "epoch": 2.097565508269838,
      "grad_norm": 0.895017683506012,
      "learning_rate": 9.25706968281728e-05,
      "loss": 1.0122,
      "step": 11287
    },
    {
      "epoch": 2.0977513473332094,
      "grad_norm": 0.8561277389526367,
      "learning_rate": 9.255612794025405e-05,
      "loss": 0.8293,
      "step": 11288
    },
    {
      "epoch": 2.0979371863965808,
      "grad_norm": 0.9393472671508789,
      "learning_rate": 9.254155921121201e-05,
      "loss": 0.7817,
      "step": 11289
    },
    {
      "epoch": 2.0981230254599517,
      "grad_norm": 0.7481895089149475,
      "learning_rate": 9.252699064135758e-05,
      "loss": 0.9282,
      "step": 11290
    },
    {
      "epoch": 2.098308864523323,
      "grad_norm": 0.7104239463806152,
      "learning_rate": 9.251242223100169e-05,
      "loss": 0.8032,
      "step": 11291
    },
    {
      "epoch": 2.098494703586694,
      "grad_norm": 0.8819231986999512,
      "learning_rate": 9.249785398045534e-05,
      "loss": 0.9462,
      "step": 11292
    },
    {
      "epoch": 2.098680542650065,
      "grad_norm": 0.7585166096687317,
      "learning_rate": 9.24832858900294e-05,
      "loss": 0.8988,
      "step": 11293
    },
    {
      "epoch": 2.098866381713436,
      "grad_norm": 0.7590588927268982,
      "learning_rate": 9.246871796003482e-05,
      "loss": 1.0465,
      "step": 11294
    },
    {
      "epoch": 2.0990522207768074,
      "grad_norm": 0.6837874054908752,
      "learning_rate": 9.245415019078253e-05,
      "loss": 0.8946,
      "step": 11295
    },
    {
      "epoch": 2.0992380598401783,
      "grad_norm": 0.9009872674942017,
      "learning_rate": 9.243958258258347e-05,
      "loss": 1.042,
      "step": 11296
    },
    {
      "epoch": 2.0994238989035496,
      "grad_norm": 0.7737190127372742,
      "learning_rate": 9.242501513574853e-05,
      "loss": 0.8884,
      "step": 11297
    },
    {
      "epoch": 2.0996097379669205,
      "grad_norm": 0.9149159789085388,
      "learning_rate": 9.241044785058862e-05,
      "loss": 1.0078,
      "step": 11298
    },
    {
      "epoch": 2.099795577030292,
      "grad_norm": 0.7528471350669861,
      "learning_rate": 9.239588072741472e-05,
      "loss": 1.0133,
      "step": 11299
    },
    {
      "epoch": 2.0999814160936627,
      "grad_norm": 0.9609968066215515,
      "learning_rate": 9.238131376653766e-05,
      "loss": 1.2004,
      "step": 11300
    },
    {
      "epoch": 2.100167255157034,
      "grad_norm": 0.9802888035774231,
      "learning_rate": 9.23667469682684e-05,
      "loss": 0.8518,
      "step": 11301
    },
    {
      "epoch": 2.100353094220405,
      "grad_norm": 0.7507895231246948,
      "learning_rate": 9.235218033291784e-05,
      "loss": 0.9317,
      "step": 11302
    },
    {
      "epoch": 2.1005389332837763,
      "grad_norm": 0.764111340045929,
      "learning_rate": 9.233761386079685e-05,
      "loss": 0.9227,
      "step": 11303
    },
    {
      "epoch": 2.100724772347147,
      "grad_norm": 0.5499566793441772,
      "learning_rate": 9.232304755221637e-05,
      "loss": 0.4425,
      "step": 11304
    },
    {
      "epoch": 2.1009106114105185,
      "grad_norm": 0.6732088923454285,
      "learning_rate": 9.230848140748724e-05,
      "loss": 0.8442,
      "step": 11305
    },
    {
      "epoch": 2.10109645047389,
      "grad_norm": 0.8917862772941589,
      "learning_rate": 9.229391542692039e-05,
      "loss": 1.037,
      "step": 11306
    },
    {
      "epoch": 2.1012822895372607,
      "grad_norm": 0.7546885013580322,
      "learning_rate": 9.227934961082667e-05,
      "loss": 1.1376,
      "step": 11307
    },
    {
      "epoch": 2.101468128600632,
      "grad_norm": 0.9907255172729492,
      "learning_rate": 9.226478395951701e-05,
      "loss": 0.9869,
      "step": 11308
    },
    {
      "epoch": 2.101653967664003,
      "grad_norm": 0.7944941520690918,
      "learning_rate": 9.22502184733022e-05,
      "loss": 1.002,
      "step": 11309
    },
    {
      "epoch": 2.1018398067273742,
      "grad_norm": 0.8978254795074463,
      "learning_rate": 9.223565315249325e-05,
      "loss": 0.9644,
      "step": 11310
    },
    {
      "epoch": 2.102025645790745,
      "grad_norm": 0.7816819548606873,
      "learning_rate": 9.222108799740091e-05,
      "loss": 0.8697,
      "step": 11311
    },
    {
      "epoch": 2.1022114848541165,
      "grad_norm": 0.7533339262008667,
      "learning_rate": 9.220652300833615e-05,
      "loss": 0.9458,
      "step": 11312
    },
    {
      "epoch": 2.1023973239174873,
      "grad_norm": 0.7051428556442261,
      "learning_rate": 9.219195818560974e-05,
      "loss": 0.8269,
      "step": 11313
    },
    {
      "epoch": 2.1025831629808587,
      "grad_norm": 0.7213872671127319,
      "learning_rate": 9.21773935295326e-05,
      "loss": 0.8971,
      "step": 11314
    },
    {
      "epoch": 2.1027690020442296,
      "grad_norm": 1.5018362998962402,
      "learning_rate": 9.216282904041557e-05,
      "loss": 0.8616,
      "step": 11315
    },
    {
      "epoch": 2.102954841107601,
      "grad_norm": 0.6886687278747559,
      "learning_rate": 9.21482647185695e-05,
      "loss": 0.888,
      "step": 11316
    },
    {
      "epoch": 2.1031406801709718,
      "grad_norm": 0.9783439040184021,
      "learning_rate": 9.213370056430526e-05,
      "loss": 0.8997,
      "step": 11317
    },
    {
      "epoch": 2.103326519234343,
      "grad_norm": 0.761658251285553,
      "learning_rate": 9.211913657793369e-05,
      "loss": 1.1431,
      "step": 11318
    },
    {
      "epoch": 2.103512358297714,
      "grad_norm": 0.7639985680580139,
      "learning_rate": 9.210457275976554e-05,
      "loss": 0.8484,
      "step": 11319
    },
    {
      "epoch": 2.1036981973610853,
      "grad_norm": 0.7667630910873413,
      "learning_rate": 9.209000911011182e-05,
      "loss": 0.838,
      "step": 11320
    },
    {
      "epoch": 2.103884036424456,
      "grad_norm": 0.8004581332206726,
      "learning_rate": 9.207544562928325e-05,
      "loss": 0.815,
      "step": 11321
    },
    {
      "epoch": 2.1040698754878275,
      "grad_norm": 0.8453937768936157,
      "learning_rate": 9.206088231759073e-05,
      "loss": 0.8801,
      "step": 11322
    },
    {
      "epoch": 2.104255714551199,
      "grad_norm": 0.7058165669441223,
      "learning_rate": 9.204631917534501e-05,
      "loss": 0.7879,
      "step": 11323
    },
    {
      "epoch": 2.1044415536145697,
      "grad_norm": 0.7437944412231445,
      "learning_rate": 9.2031756202857e-05,
      "loss": 0.8152,
      "step": 11324
    },
    {
      "epoch": 2.104627392677941,
      "grad_norm": 0.9589625597000122,
      "learning_rate": 9.201719340043745e-05,
      "loss": 0.8667,
      "step": 11325
    },
    {
      "epoch": 2.104813231741312,
      "grad_norm": 0.7726141810417175,
      "learning_rate": 9.200263076839724e-05,
      "loss": 0.756,
      "step": 11326
    },
    {
      "epoch": 2.1049990708046833,
      "grad_norm": 0.7532119154930115,
      "learning_rate": 9.198806830704711e-05,
      "loss": 0.8386,
      "step": 11327
    },
    {
      "epoch": 2.105184909868054,
      "grad_norm": 2.2922170162200928,
      "learning_rate": 9.197350601669795e-05,
      "loss": 1.3785,
      "step": 11328
    },
    {
      "epoch": 2.1053707489314255,
      "grad_norm": 0.7087710499763489,
      "learning_rate": 9.19589438976605e-05,
      "loss": 0.8279,
      "step": 11329
    },
    {
      "epoch": 2.1055565879947964,
      "grad_norm": 0.8486080169677734,
      "learning_rate": 9.194438195024557e-05,
      "loss": 0.9904,
      "step": 11330
    },
    {
      "epoch": 2.1057424270581677,
      "grad_norm": 0.7624375224113464,
      "learning_rate": 9.192982017476401e-05,
      "loss": 0.9955,
      "step": 11331
    },
    {
      "epoch": 2.1059282661215386,
      "grad_norm": 0.8084495663642883,
      "learning_rate": 9.191525857152662e-05,
      "loss": 0.9191,
      "step": 11332
    },
    {
      "epoch": 2.10611410518491,
      "grad_norm": 0.6322254538536072,
      "learning_rate": 9.190069714084413e-05,
      "loss": 0.7792,
      "step": 11333
    },
    {
      "epoch": 2.106299944248281,
      "grad_norm": 0.8094192743301392,
      "learning_rate": 9.188613588302739e-05,
      "loss": 0.8863,
      "step": 11334
    },
    {
      "epoch": 2.106485783311652,
      "grad_norm": 0.7641638517379761,
      "learning_rate": 9.187157479838713e-05,
      "loss": 0.6735,
      "step": 11335
    },
    {
      "epoch": 2.106671622375023,
      "grad_norm": 0.765892505645752,
      "learning_rate": 9.185701388723416e-05,
      "loss": 0.8556,
      "step": 11336
    },
    {
      "epoch": 2.1068574614383944,
      "grad_norm": 0.8530939817428589,
      "learning_rate": 9.184245314987925e-05,
      "loss": 0.8466,
      "step": 11337
    },
    {
      "epoch": 2.1070433005017657,
      "grad_norm": 0.7106696367263794,
      "learning_rate": 9.182789258663321e-05,
      "loss": 0.9071,
      "step": 11338
    },
    {
      "epoch": 2.1072291395651366,
      "grad_norm": 0.9117116928100586,
      "learning_rate": 9.181333219780675e-05,
      "loss": 1.0345,
      "step": 11339
    },
    {
      "epoch": 2.107414978628508,
      "grad_norm": 0.7420015335083008,
      "learning_rate": 9.179877198371063e-05,
      "loss": 0.639,
      "step": 11340
    },
    {
      "epoch": 2.107600817691879,
      "grad_norm": 0.8371978402137756,
      "learning_rate": 9.178421194465569e-05,
      "loss": 0.9271,
      "step": 11341
    },
    {
      "epoch": 2.10778665675525,
      "grad_norm": 0.684158444404602,
      "learning_rate": 9.176965208095265e-05,
      "loss": 0.8825,
      "step": 11342
    },
    {
      "epoch": 2.107972495818621,
      "grad_norm": 0.6936484575271606,
      "learning_rate": 9.175509239291225e-05,
      "loss": 0.8442,
      "step": 11343
    },
    {
      "epoch": 2.1081583348819923,
      "grad_norm": 0.8441296815872192,
      "learning_rate": 9.174053288084527e-05,
      "loss": 0.7672,
      "step": 11344
    },
    {
      "epoch": 2.108344173945363,
      "grad_norm": 0.954045832157135,
      "learning_rate": 9.172597354506244e-05,
      "loss": 0.9207,
      "step": 11345
    },
    {
      "epoch": 2.1085300130087345,
      "grad_norm": 0.7127572298049927,
      "learning_rate": 9.171141438587451e-05,
      "loss": 0.7678,
      "step": 11346
    },
    {
      "epoch": 2.1087158520721054,
      "grad_norm": 0.6595544815063477,
      "learning_rate": 9.16968554035922e-05,
      "loss": 0.7733,
      "step": 11347
    },
    {
      "epoch": 2.1089016911354768,
      "grad_norm": 0.7821077704429626,
      "learning_rate": 9.168229659852629e-05,
      "loss": 0.8252,
      "step": 11348
    },
    {
      "epoch": 2.1090875301988476,
      "grad_norm": 0.8677456378936768,
      "learning_rate": 9.166773797098745e-05,
      "loss": 1.0355,
      "step": 11349
    },
    {
      "epoch": 2.109273369262219,
      "grad_norm": 0.7058128118515015,
      "learning_rate": 9.165317952128647e-05,
      "loss": 0.7044,
      "step": 11350
    },
    {
      "epoch": 2.10945920832559,
      "grad_norm": 0.7050992846488953,
      "learning_rate": 9.163862124973398e-05,
      "loss": 0.9244,
      "step": 11351
    },
    {
      "epoch": 2.109645047388961,
      "grad_norm": 0.7171057462692261,
      "learning_rate": 9.162406315664085e-05,
      "loss": 0.7257,
      "step": 11352
    },
    {
      "epoch": 2.109830886452332,
      "grad_norm": 0.8750724196434021,
      "learning_rate": 9.160950524231768e-05,
      "loss": 0.9358,
      "step": 11353
    },
    {
      "epoch": 2.1100167255157034,
      "grad_norm": 0.6821286678314209,
      "learning_rate": 9.159494750707526e-05,
      "loss": 0.8675,
      "step": 11354
    },
    {
      "epoch": 2.1102025645790747,
      "grad_norm": 0.7473365068435669,
      "learning_rate": 9.158038995122424e-05,
      "loss": 0.753,
      "step": 11355
    },
    {
      "epoch": 2.1103884036424456,
      "grad_norm": 0.7012847065925598,
      "learning_rate": 9.156583257507535e-05,
      "loss": 0.9714,
      "step": 11356
    },
    {
      "epoch": 2.110574242705817,
      "grad_norm": 1.2019333839416504,
      "learning_rate": 9.155127537893929e-05,
      "loss": 1.156,
      "step": 11357
    },
    {
      "epoch": 2.110760081769188,
      "grad_norm": 1.0187358856201172,
      "learning_rate": 9.153671836312675e-05,
      "loss": 0.9881,
      "step": 11358
    },
    {
      "epoch": 2.110945920832559,
      "grad_norm": 0.7755599617958069,
      "learning_rate": 9.152216152794844e-05,
      "loss": 0.9068,
      "step": 11359
    },
    {
      "epoch": 2.11113175989593,
      "grad_norm": 0.7259166836738586,
      "learning_rate": 9.150760487371506e-05,
      "loss": 0.969,
      "step": 11360
    },
    {
      "epoch": 2.1113175989593014,
      "grad_norm": 0.6794169545173645,
      "learning_rate": 9.149304840073721e-05,
      "loss": 0.8936,
      "step": 11361
    },
    {
      "epoch": 2.1115034380226723,
      "grad_norm": 0.8772898316383362,
      "learning_rate": 9.147849210932571e-05,
      "loss": 0.887,
      "step": 11362
    },
    {
      "epoch": 2.1116892770860436,
      "grad_norm": 0.860237181186676,
      "learning_rate": 9.146393599979114e-05,
      "loss": 0.9114,
      "step": 11363
    },
    {
      "epoch": 2.1118751161494145,
      "grad_norm": 0.6794553995132446,
      "learning_rate": 9.144938007244426e-05,
      "loss": 0.8011,
      "step": 11364
    },
    {
      "epoch": 2.112060955212786,
      "grad_norm": 0.7240217924118042,
      "learning_rate": 9.143482432759565e-05,
      "loss": 0.6517,
      "step": 11365
    },
    {
      "epoch": 2.1122467942761567,
      "grad_norm": 0.7571320533752441,
      "learning_rate": 9.142026876555602e-05,
      "loss": 1.0215,
      "step": 11366
    },
    {
      "epoch": 2.112432633339528,
      "grad_norm": 0.7616672515869141,
      "learning_rate": 9.140571338663604e-05,
      "loss": 0.959,
      "step": 11367
    },
    {
      "epoch": 2.112618472402899,
      "grad_norm": 0.7437677383422852,
      "learning_rate": 9.139115819114636e-05,
      "loss": 0.809,
      "step": 11368
    },
    {
      "epoch": 2.1128043114662702,
      "grad_norm": 0.8062520027160645,
      "learning_rate": 9.137660317939761e-05,
      "loss": 1.0213,
      "step": 11369
    },
    {
      "epoch": 2.112990150529641,
      "grad_norm": 0.8090141415596008,
      "learning_rate": 9.13620483517005e-05,
      "loss": 1.038,
      "step": 11370
    },
    {
      "epoch": 2.1131759895930124,
      "grad_norm": 0.8203197717666626,
      "learning_rate": 9.134749370836563e-05,
      "loss": 0.9441,
      "step": 11371
    },
    {
      "epoch": 2.113361828656384,
      "grad_norm": 0.8296011686325073,
      "learning_rate": 9.133293924970362e-05,
      "loss": 0.9218,
      "step": 11372
    },
    {
      "epoch": 2.1135476677197547,
      "grad_norm": 0.7578732967376709,
      "learning_rate": 9.131838497602517e-05,
      "loss": 0.9063,
      "step": 11373
    },
    {
      "epoch": 2.113733506783126,
      "grad_norm": 1.003522276878357,
      "learning_rate": 9.130383088764093e-05,
      "loss": 0.8269,
      "step": 11374
    },
    {
      "epoch": 2.113919345846497,
      "grad_norm": 0.7684604525566101,
      "learning_rate": 9.128927698486147e-05,
      "loss": 0.9037,
      "step": 11375
    },
    {
      "epoch": 2.114105184909868,
      "grad_norm": 0.8644822239875793,
      "learning_rate": 9.127472326799746e-05,
      "loss": 1.0589,
      "step": 11376
    },
    {
      "epoch": 2.114291023973239,
      "grad_norm": 0.7624018788337708,
      "learning_rate": 9.12601697373595e-05,
      "loss": 1.0758,
      "step": 11377
    },
    {
      "epoch": 2.1144768630366104,
      "grad_norm": 0.7607561349868774,
      "learning_rate": 9.124561639325822e-05,
      "loss": 1.087,
      "step": 11378
    },
    {
      "epoch": 2.1146627020999813,
      "grad_norm": 0.7460099458694458,
      "learning_rate": 9.123106323600424e-05,
      "loss": 1.0461,
      "step": 11379
    },
    {
      "epoch": 2.1148485411633526,
      "grad_norm": 0.7677852511405945,
      "learning_rate": 9.121651026590816e-05,
      "loss": 1.0069,
      "step": 11380
    },
    {
      "epoch": 2.1150343802267235,
      "grad_norm": 0.7719981074333191,
      "learning_rate": 9.120195748328059e-05,
      "loss": 0.9425,
      "step": 11381
    },
    {
      "epoch": 2.115220219290095,
      "grad_norm": 1.7005661725997925,
      "learning_rate": 9.118740488843211e-05,
      "loss": 1.5553,
      "step": 11382
    },
    {
      "epoch": 2.1154060583534657,
      "grad_norm": 0.7657036781311035,
      "learning_rate": 9.117285248167337e-05,
      "loss": 0.9203,
      "step": 11383
    },
    {
      "epoch": 2.115591897416837,
      "grad_norm": 0.8619498610496521,
      "learning_rate": 9.115830026331498e-05,
      "loss": 1.0513,
      "step": 11384
    },
    {
      "epoch": 2.115777736480208,
      "grad_norm": 0.8409205079078674,
      "learning_rate": 9.114374823366747e-05,
      "loss": 1.0542,
      "step": 11385
    },
    {
      "epoch": 2.1159635755435793,
      "grad_norm": 0.7962708473205566,
      "learning_rate": 9.112919639304148e-05,
      "loss": 0.5205,
      "step": 11386
    },
    {
      "epoch": 2.1161494146069506,
      "grad_norm": 0.6844388842582703,
      "learning_rate": 9.111464474174755e-05,
      "loss": 0.7801,
      "step": 11387
    },
    {
      "epoch": 2.1163352536703215,
      "grad_norm": 0.9194725155830383,
      "learning_rate": 9.11000932800963e-05,
      "loss": 1.1574,
      "step": 11388
    },
    {
      "epoch": 2.116521092733693,
      "grad_norm": 0.8045440316200256,
      "learning_rate": 9.108554200839826e-05,
      "loss": 0.7407,
      "step": 11389
    },
    {
      "epoch": 2.1167069317970637,
      "grad_norm": 0.6948972344398499,
      "learning_rate": 9.107099092696405e-05,
      "loss": 0.8281,
      "step": 11390
    },
    {
      "epoch": 2.116892770860435,
      "grad_norm": 0.8075729608535767,
      "learning_rate": 9.10564400361042e-05,
      "loss": 1.0748,
      "step": 11391
    },
    {
      "epoch": 2.117078609923806,
      "grad_norm": 0.8684819340705872,
      "learning_rate": 9.104188933612929e-05,
      "loss": 0.8818,
      "step": 11392
    },
    {
      "epoch": 2.1172644489871772,
      "grad_norm": 0.7302981019020081,
      "learning_rate": 9.102733882734984e-05,
      "loss": 1.0826,
      "step": 11393
    },
    {
      "epoch": 2.117450288050548,
      "grad_norm": 0.7284327149391174,
      "learning_rate": 9.10127885100765e-05,
      "loss": 0.6441,
      "step": 11394
    },
    {
      "epoch": 2.1176361271139195,
      "grad_norm": 0.8038931488990784,
      "learning_rate": 9.099823838461975e-05,
      "loss": 0.9264,
      "step": 11395
    },
    {
      "epoch": 2.1178219661772903,
      "grad_norm": 0.7592558264732361,
      "learning_rate": 9.098368845129018e-05,
      "loss": 1.1874,
      "step": 11396
    },
    {
      "epoch": 2.1180078052406617,
      "grad_norm": 0.7145334482192993,
      "learning_rate": 9.096913871039827e-05,
      "loss": 0.9178,
      "step": 11397
    },
    {
      "epoch": 2.1181936443040326,
      "grad_norm": 0.8046518564224243,
      "learning_rate": 9.095458916225464e-05,
      "loss": 1.0908,
      "step": 11398
    },
    {
      "epoch": 2.118379483367404,
      "grad_norm": 0.7807143330574036,
      "learning_rate": 9.094003980716975e-05,
      "loss": 0.9177,
      "step": 11399
    },
    {
      "epoch": 2.1185653224307748,
      "grad_norm": 0.7822641134262085,
      "learning_rate": 9.092549064545417e-05,
      "loss": 0.9126,
      "step": 11400
    },
    {
      "epoch": 2.118751161494146,
      "grad_norm": 0.7133731245994568,
      "learning_rate": 9.091094167741841e-05,
      "loss": 0.823,
      "step": 11401
    },
    {
      "epoch": 2.118937000557517,
      "grad_norm": 0.9369903206825256,
      "learning_rate": 9.0896392903373e-05,
      "loss": 0.728,
      "step": 11402
    },
    {
      "epoch": 2.1191228396208883,
      "grad_norm": 0.7016496062278748,
      "learning_rate": 9.088184432362844e-05,
      "loss": 1.0928,
      "step": 11403
    },
    {
      "epoch": 2.119308678684259,
      "grad_norm": 0.6865460276603699,
      "learning_rate": 9.086729593849532e-05,
      "loss": 0.749,
      "step": 11404
    },
    {
      "epoch": 2.1194945177476305,
      "grad_norm": 0.808489978313446,
      "learning_rate": 9.085274774828404e-05,
      "loss": 0.8495,
      "step": 11405
    },
    {
      "epoch": 2.119680356811002,
      "grad_norm": 0.7700434923171997,
      "learning_rate": 9.083819975330522e-05,
      "loss": 0.8739,
      "step": 11406
    },
    {
      "epoch": 2.1198661958743727,
      "grad_norm": 0.7213497161865234,
      "learning_rate": 9.082365195386928e-05,
      "loss": 0.9642,
      "step": 11407
    },
    {
      "epoch": 2.120052034937744,
      "grad_norm": 0.8258712887763977,
      "learning_rate": 9.080910435028674e-05,
      "loss": 0.9336,
      "step": 11408
    },
    {
      "epoch": 2.120237874001115,
      "grad_norm": 0.6518661975860596,
      "learning_rate": 9.07945569428681e-05,
      "loss": 0.6723,
      "step": 11409
    },
    {
      "epoch": 2.1204237130644863,
      "grad_norm": 0.7806102633476257,
      "learning_rate": 9.078000973192384e-05,
      "loss": 0.8546,
      "step": 11410
    },
    {
      "epoch": 2.120609552127857,
      "grad_norm": 0.7937559485435486,
      "learning_rate": 9.076546271776444e-05,
      "loss": 1.052,
      "step": 11411
    },
    {
      "epoch": 2.1207953911912285,
      "grad_norm": 0.8053615093231201,
      "learning_rate": 9.075091590070041e-05,
      "loss": 0.8929,
      "step": 11412
    },
    {
      "epoch": 2.1209812302545994,
      "grad_norm": 0.8069961667060852,
      "learning_rate": 9.073636928104221e-05,
      "loss": 0.8843,
      "step": 11413
    },
    {
      "epoch": 2.1211670693179707,
      "grad_norm": 0.7899468541145325,
      "learning_rate": 9.072182285910028e-05,
      "loss": 0.8003,
      "step": 11414
    },
    {
      "epoch": 2.1213529083813416,
      "grad_norm": 0.7648258805274963,
      "learning_rate": 9.070727663518512e-05,
      "loss": 1.1836,
      "step": 11415
    },
    {
      "epoch": 2.121538747444713,
      "grad_norm": 0.7307213544845581,
      "learning_rate": 9.069273060960723e-05,
      "loss": 0.9682,
      "step": 11416
    },
    {
      "epoch": 2.121724586508084,
      "grad_norm": 0.8227604627609253,
      "learning_rate": 9.0678184782677e-05,
      "loss": 0.7594,
      "step": 11417
    },
    {
      "epoch": 2.121910425571455,
      "grad_norm": 0.9569815993309021,
      "learning_rate": 9.066363915470495e-05,
      "loss": 1.0488,
      "step": 11418
    },
    {
      "epoch": 2.122096264634826,
      "grad_norm": 0.7983200550079346,
      "learning_rate": 9.064909372600148e-05,
      "loss": 0.8448,
      "step": 11419
    },
    {
      "epoch": 2.1222821036981974,
      "grad_norm": 0.905168890953064,
      "learning_rate": 9.063454849687707e-05,
      "loss": 0.9164,
      "step": 11420
    },
    {
      "epoch": 2.1224679427615687,
      "grad_norm": 1.0643441677093506,
      "learning_rate": 9.062000346764214e-05,
      "loss": 0.8715,
      "step": 11421
    },
    {
      "epoch": 2.1226537818249396,
      "grad_norm": 0.8178336024284363,
      "learning_rate": 9.060545863860713e-05,
      "loss": 1.1267,
      "step": 11422
    },
    {
      "epoch": 2.122839620888311,
      "grad_norm": 0.7979642748832703,
      "learning_rate": 9.059091401008252e-05,
      "loss": 1.002,
      "step": 11423
    },
    {
      "epoch": 2.123025459951682,
      "grad_norm": 0.7845909595489502,
      "learning_rate": 9.057636958237864e-05,
      "loss": 0.864,
      "step": 11424
    },
    {
      "epoch": 2.123211299015053,
      "grad_norm": 0.7184850573539734,
      "learning_rate": 9.056182535580601e-05,
      "loss": 0.9398,
      "step": 11425
    },
    {
      "epoch": 2.123397138078424,
      "grad_norm": 0.734954297542572,
      "learning_rate": 9.054728133067505e-05,
      "loss": 0.8397,
      "step": 11426
    },
    {
      "epoch": 2.1235829771417953,
      "grad_norm": 0.8179150819778442,
      "learning_rate": 9.053273750729611e-05,
      "loss": 0.7084,
      "step": 11427
    },
    {
      "epoch": 2.123768816205166,
      "grad_norm": 1.0683249235153198,
      "learning_rate": 9.051819388597969e-05,
      "loss": 1.2287,
      "step": 11428
    },
    {
      "epoch": 2.1239546552685376,
      "grad_norm": 0.9108082056045532,
      "learning_rate": 9.050365046703611e-05,
      "loss": 1.1409,
      "step": 11429
    },
    {
      "epoch": 2.1241404943319084,
      "grad_norm": 0.9073103666305542,
      "learning_rate": 9.048910725077584e-05,
      "loss": 0.7285,
      "step": 11430
    },
    {
      "epoch": 2.1243263333952798,
      "grad_norm": 0.8368109464645386,
      "learning_rate": 9.047456423750923e-05,
      "loss": 1.0234,
      "step": 11431
    },
    {
      "epoch": 2.1245121724586506,
      "grad_norm": 0.8634905815124512,
      "learning_rate": 9.04600214275467e-05,
      "loss": 0.977,
      "step": 11432
    },
    {
      "epoch": 2.124698011522022,
      "grad_norm": 0.7714248895645142,
      "learning_rate": 9.044547882119867e-05,
      "loss": 0.9311,
      "step": 11433
    },
    {
      "epoch": 2.124883850585393,
      "grad_norm": 0.6275684833526611,
      "learning_rate": 9.043093641877547e-05,
      "loss": 0.7582,
      "step": 11434
    },
    {
      "epoch": 2.125069689648764,
      "grad_norm": 0.7640289068222046,
      "learning_rate": 9.041639422058749e-05,
      "loss": 0.9363,
      "step": 11435
    },
    {
      "epoch": 2.1252555287121355,
      "grad_norm": 0.9009206295013428,
      "learning_rate": 9.040185222694517e-05,
      "loss": 1.1269,
      "step": 11436
    },
    {
      "epoch": 2.1254413677755064,
      "grad_norm": 0.8125781416893005,
      "learning_rate": 9.038731043815882e-05,
      "loss": 0.8613,
      "step": 11437
    },
    {
      "epoch": 2.1256272068388777,
      "grad_norm": 0.8105628490447998,
      "learning_rate": 9.037276885453887e-05,
      "loss": 0.9688,
      "step": 11438
    },
    {
      "epoch": 2.1258130459022486,
      "grad_norm": 0.8331906199455261,
      "learning_rate": 9.035822747639561e-05,
      "loss": 0.7517,
      "step": 11439
    },
    {
      "epoch": 2.12599888496562,
      "grad_norm": 0.7814489006996155,
      "learning_rate": 9.034368630403946e-05,
      "loss": 1.1615,
      "step": 11440
    },
    {
      "epoch": 2.126184724028991,
      "grad_norm": 0.9096186757087708,
      "learning_rate": 9.032914533778075e-05,
      "loss": 1.0034,
      "step": 11441
    },
    {
      "epoch": 2.126370563092362,
      "grad_norm": 2.2994253635406494,
      "learning_rate": 9.031460457792982e-05,
      "loss": 1.5315,
      "step": 11442
    },
    {
      "epoch": 2.126556402155733,
      "grad_norm": 0.7273047566413879,
      "learning_rate": 9.030006402479706e-05,
      "loss": 0.7698,
      "step": 11443
    },
    {
      "epoch": 2.1267422412191044,
      "grad_norm": 0.9945558905601501,
      "learning_rate": 9.028552367869277e-05,
      "loss": 0.7612,
      "step": 11444
    },
    {
      "epoch": 2.1269280802824753,
      "grad_norm": 0.8034189939498901,
      "learning_rate": 9.027098353992729e-05,
      "loss": 0.8135,
      "step": 11445
    },
    {
      "epoch": 2.1271139193458466,
      "grad_norm": 0.8340957760810852,
      "learning_rate": 9.0256443608811e-05,
      "loss": 0.8855,
      "step": 11446
    },
    {
      "epoch": 2.1272997584092175,
      "grad_norm": 0.9442667365074158,
      "learning_rate": 9.024190388565418e-05,
      "loss": 1.0958,
      "step": 11447
    },
    {
      "epoch": 2.127485597472589,
      "grad_norm": 0.9203138947486877,
      "learning_rate": 9.02273643707672e-05,
      "loss": 1.161,
      "step": 11448
    },
    {
      "epoch": 2.1276714365359597,
      "grad_norm": 0.6896242499351501,
      "learning_rate": 9.021282506446035e-05,
      "loss": 0.8245,
      "step": 11449
    },
    {
      "epoch": 2.127857275599331,
      "grad_norm": 0.7537946701049805,
      "learning_rate": 9.019828596704394e-05,
      "loss": 0.7225,
      "step": 11450
    },
    {
      "epoch": 2.128043114662702,
      "grad_norm": 0.9126191139221191,
      "learning_rate": 9.018374707882831e-05,
      "loss": 1.0069,
      "step": 11451
    },
    {
      "epoch": 2.1282289537260732,
      "grad_norm": 0.7200766205787659,
      "learning_rate": 9.016920840012375e-05,
      "loss": 0.8518,
      "step": 11452
    },
    {
      "epoch": 2.128414792789444,
      "grad_norm": 0.8545762896537781,
      "learning_rate": 9.015466993124057e-05,
      "loss": 0.8795,
      "step": 11453
    },
    {
      "epoch": 2.1286006318528155,
      "grad_norm": 0.6515056490898132,
      "learning_rate": 9.014013167248906e-05,
      "loss": 0.7824,
      "step": 11454
    },
    {
      "epoch": 2.128786470916187,
      "grad_norm": 0.7449262142181396,
      "learning_rate": 9.012559362417954e-05,
      "loss": 1.0393,
      "step": 11455
    },
    {
      "epoch": 2.1289723099795577,
      "grad_norm": 0.8179948329925537,
      "learning_rate": 9.011105578662222e-05,
      "loss": 1.0184,
      "step": 11456
    },
    {
      "epoch": 2.129158149042929,
      "grad_norm": 0.653978168964386,
      "learning_rate": 9.009651816012748e-05,
      "loss": 0.7403,
      "step": 11457
    },
    {
      "epoch": 2.1293439881063,
      "grad_norm": 0.7826240658760071,
      "learning_rate": 9.008198074500557e-05,
      "loss": 0.8306,
      "step": 11458
    },
    {
      "epoch": 2.129529827169671,
      "grad_norm": 0.8257392644882202,
      "learning_rate": 9.006744354156676e-05,
      "loss": 0.8867,
      "step": 11459
    },
    {
      "epoch": 2.129715666233042,
      "grad_norm": 0.8371952176094055,
      "learning_rate": 9.00529065501213e-05,
      "loss": 0.9432,
      "step": 11460
    },
    {
      "epoch": 2.1299015052964134,
      "grad_norm": 0.7463301420211792,
      "learning_rate": 9.003836977097952e-05,
      "loss": 0.7961,
      "step": 11461
    },
    {
      "epoch": 2.1300873443597843,
      "grad_norm": 0.7014195919036865,
      "learning_rate": 9.002383320445163e-05,
      "loss": 0.9354,
      "step": 11462
    },
    {
      "epoch": 2.1302731834231556,
      "grad_norm": 0.7329978346824646,
      "learning_rate": 9.00092968508479e-05,
      "loss": 0.8252,
      "step": 11463
    },
    {
      "epoch": 2.1304590224865265,
      "grad_norm": 0.7446506023406982,
      "learning_rate": 8.999476071047857e-05,
      "loss": 0.8878,
      "step": 11464
    },
    {
      "epoch": 2.130644861549898,
      "grad_norm": 0.8863813877105713,
      "learning_rate": 8.998022478365392e-05,
      "loss": 0.8377,
      "step": 11465
    },
    {
      "epoch": 2.1308307006132687,
      "grad_norm": 0.7592625617980957,
      "learning_rate": 8.996568907068415e-05,
      "loss": 0.9176,
      "step": 11466
    },
    {
      "epoch": 2.13101653967664,
      "grad_norm": 0.8758193254470825,
      "learning_rate": 8.995115357187953e-05,
      "loss": 1.2571,
      "step": 11467
    },
    {
      "epoch": 2.131202378740011,
      "grad_norm": 0.7715368866920471,
      "learning_rate": 8.993661828755032e-05,
      "loss": 0.8266,
      "step": 11468
    },
    {
      "epoch": 2.1313882178033823,
      "grad_norm": 0.6676689982414246,
      "learning_rate": 8.99220832180067e-05,
      "loss": 0.7584,
      "step": 11469
    },
    {
      "epoch": 2.1315740568667536,
      "grad_norm": 0.9252825379371643,
      "learning_rate": 8.990754836355893e-05,
      "loss": 0.8657,
      "step": 11470
    },
    {
      "epoch": 2.1317598959301245,
      "grad_norm": 0.876563310623169,
      "learning_rate": 8.989301372451722e-05,
      "loss": 0.9531,
      "step": 11471
    },
    {
      "epoch": 2.131945734993496,
      "grad_norm": 0.7079026699066162,
      "learning_rate": 8.987847930119179e-05,
      "loss": 0.9201,
      "step": 11472
    },
    {
      "epoch": 2.1321315740568667,
      "grad_norm": 0.7698578238487244,
      "learning_rate": 8.986394509389287e-05,
      "loss": 0.7739,
      "step": 11473
    },
    {
      "epoch": 2.132317413120238,
      "grad_norm": 0.9522943496704102,
      "learning_rate": 8.984941110293061e-05,
      "loss": 0.9106,
      "step": 11474
    },
    {
      "epoch": 2.132503252183609,
      "grad_norm": 0.8080538511276245,
      "learning_rate": 8.983487732861528e-05,
      "loss": 1.0263,
      "step": 11475
    },
    {
      "epoch": 2.1326890912469803,
      "grad_norm": 0.8311910033226013,
      "learning_rate": 8.982034377125703e-05,
      "loss": 0.8961,
      "step": 11476
    },
    {
      "epoch": 2.132874930310351,
      "grad_norm": 0.7102888822555542,
      "learning_rate": 8.980581043116605e-05,
      "loss": 0.8197,
      "step": 11477
    },
    {
      "epoch": 2.1330607693737225,
      "grad_norm": 0.8097278475761414,
      "learning_rate": 8.97912773086526e-05,
      "loss": 0.9794,
      "step": 11478
    },
    {
      "epoch": 2.1332466084370934,
      "grad_norm": 0.7649460434913635,
      "learning_rate": 8.977674440402679e-05,
      "loss": 0.8794,
      "step": 11479
    },
    {
      "epoch": 2.1334324475004647,
      "grad_norm": 0.730686366558075,
      "learning_rate": 8.976221171759883e-05,
      "loss": 1.0425,
      "step": 11480
    },
    {
      "epoch": 2.1336182865638356,
      "grad_norm": 0.7751401662826538,
      "learning_rate": 8.97476792496789e-05,
      "loss": 0.971,
      "step": 11481
    },
    {
      "epoch": 2.133804125627207,
      "grad_norm": 0.7385703921318054,
      "learning_rate": 8.973314700057717e-05,
      "loss": 0.7994,
      "step": 11482
    },
    {
      "epoch": 2.133989964690578,
      "grad_norm": 0.777972400188446,
      "learning_rate": 8.97186149706038e-05,
      "loss": 1.1205,
      "step": 11483
    },
    {
      "epoch": 2.134175803753949,
      "grad_norm": 0.7550100684165955,
      "learning_rate": 8.970408316006893e-05,
      "loss": 1.0834,
      "step": 11484
    },
    {
      "epoch": 2.13436164281732,
      "grad_norm": 0.6919218301773071,
      "learning_rate": 8.968955156928276e-05,
      "loss": 1.0027,
      "step": 11485
    },
    {
      "epoch": 2.1345474818806913,
      "grad_norm": 0.8418641090393066,
      "learning_rate": 8.967502019855539e-05,
      "loss": 0.8558,
      "step": 11486
    },
    {
      "epoch": 2.134733320944062,
      "grad_norm": 0.8520260453224182,
      "learning_rate": 8.966048904819698e-05,
      "loss": 0.7657,
      "step": 11487
    },
    {
      "epoch": 2.1349191600074335,
      "grad_norm": 0.7584142088890076,
      "learning_rate": 8.964595811851774e-05,
      "loss": 0.9419,
      "step": 11488
    },
    {
      "epoch": 2.135104999070805,
      "grad_norm": 0.7584882378578186,
      "learning_rate": 8.963142740982773e-05,
      "loss": 0.806,
      "step": 11489
    },
    {
      "epoch": 2.1352908381341758,
      "grad_norm": 0.7900401949882507,
      "learning_rate": 8.96168969224371e-05,
      "loss": 1.0048,
      "step": 11490
    },
    {
      "epoch": 2.135476677197547,
      "grad_norm": 0.7943800091743469,
      "learning_rate": 8.9602366656656e-05,
      "loss": 1.0302,
      "step": 11491
    },
    {
      "epoch": 2.135662516260918,
      "grad_norm": 0.7397233843803406,
      "learning_rate": 8.958783661279454e-05,
      "loss": 0.6135,
      "step": 11492
    },
    {
      "epoch": 2.1358483553242893,
      "grad_norm": 0.7478030323982239,
      "learning_rate": 8.957330679116286e-05,
      "loss": 0.8393,
      "step": 11493
    },
    {
      "epoch": 2.13603419438766,
      "grad_norm": 0.7902356386184692,
      "learning_rate": 8.955877719207102e-05,
      "loss": 0.8388,
      "step": 11494
    },
    {
      "epoch": 2.1362200334510315,
      "grad_norm": 0.8481513261795044,
      "learning_rate": 8.95442478158292e-05,
      "loss": 1.074,
      "step": 11495
    },
    {
      "epoch": 2.1364058725144024,
      "grad_norm": 0.8785560727119446,
      "learning_rate": 8.952971866274742e-05,
      "loss": 0.6792,
      "step": 11496
    },
    {
      "epoch": 2.1365917115777737,
      "grad_norm": 0.7510362267494202,
      "learning_rate": 8.951518973313587e-05,
      "loss": 0.8089,
      "step": 11497
    },
    {
      "epoch": 2.1367775506411446,
      "grad_norm": 0.7218495011329651,
      "learning_rate": 8.950066102730456e-05,
      "loss": 0.85,
      "step": 11498
    },
    {
      "epoch": 2.136963389704516,
      "grad_norm": 0.857275664806366,
      "learning_rate": 8.948613254556364e-05,
      "loss": 1.0204,
      "step": 11499
    },
    {
      "epoch": 2.137149228767887,
      "grad_norm": 0.7544958591461182,
      "learning_rate": 8.947160428822318e-05,
      "loss": 0.7236,
      "step": 11500
    },
    {
      "epoch": 2.137335067831258,
      "grad_norm": 0.7592241168022156,
      "learning_rate": 8.945707625559328e-05,
      "loss": 0.8866,
      "step": 11501
    },
    {
      "epoch": 2.137520906894629,
      "grad_norm": 0.860626220703125,
      "learning_rate": 8.944254844798397e-05,
      "loss": 1.198,
      "step": 11502
    },
    {
      "epoch": 2.1377067459580004,
      "grad_norm": 0.819098174571991,
      "learning_rate": 8.942802086570536e-05,
      "loss": 0.8842,
      "step": 11503
    },
    {
      "epoch": 2.1378925850213717,
      "grad_norm": 0.7637498378753662,
      "learning_rate": 8.94134935090675e-05,
      "loss": 0.7908,
      "step": 11504
    },
    {
      "epoch": 2.1380784240847426,
      "grad_norm": 0.9611912369728088,
      "learning_rate": 8.939896637838045e-05,
      "loss": 0.9452,
      "step": 11505
    },
    {
      "epoch": 2.138264263148114,
      "grad_norm": 0.7857555747032166,
      "learning_rate": 8.938443947395427e-05,
      "loss": 1.004,
      "step": 11506
    },
    {
      "epoch": 2.138450102211485,
      "grad_norm": 0.7934672236442566,
      "learning_rate": 8.936991279609903e-05,
      "loss": 0.9511,
      "step": 11507
    },
    {
      "epoch": 2.138635941274856,
      "grad_norm": 0.868075430393219,
      "learning_rate": 8.935538634512472e-05,
      "loss": 1.0373,
      "step": 11508
    },
    {
      "epoch": 2.138821780338227,
      "grad_norm": 0.6859819889068604,
      "learning_rate": 8.934086012134143e-05,
      "loss": 0.8933,
      "step": 11509
    },
    {
      "epoch": 2.1390076194015983,
      "grad_norm": 0.7351570129394531,
      "learning_rate": 8.93263341250592e-05,
      "loss": 1.0215,
      "step": 11510
    },
    {
      "epoch": 2.139193458464969,
      "grad_norm": 0.7704660892486572,
      "learning_rate": 8.931180835658807e-05,
      "loss": 0.8473,
      "step": 11511
    },
    {
      "epoch": 2.1393792975283406,
      "grad_norm": 0.8092166781425476,
      "learning_rate": 8.929728281623802e-05,
      "loss": 0.6342,
      "step": 11512
    },
    {
      "epoch": 2.1395651365917114,
      "grad_norm": 0.7497720122337341,
      "learning_rate": 8.928275750431912e-05,
      "loss": 0.9206,
      "step": 11513
    },
    {
      "epoch": 2.1397509756550828,
      "grad_norm": 0.875147819519043,
      "learning_rate": 8.926823242114136e-05,
      "loss": 0.9346,
      "step": 11514
    },
    {
      "epoch": 2.1399368147184537,
      "grad_norm": 0.724993884563446,
      "learning_rate": 8.92537075670148e-05,
      "loss": 0.8848,
      "step": 11515
    },
    {
      "epoch": 2.140122653781825,
      "grad_norm": 0.9631668925285339,
      "learning_rate": 8.923918294224935e-05,
      "loss": 0.8877,
      "step": 11516
    },
    {
      "epoch": 2.140308492845196,
      "grad_norm": 0.8639618158340454,
      "learning_rate": 8.922465854715512e-05,
      "loss": 0.7687,
      "step": 11517
    },
    {
      "epoch": 2.140494331908567,
      "grad_norm": 0.845238208770752,
      "learning_rate": 8.921013438204204e-05,
      "loss": 0.9183,
      "step": 11518
    },
    {
      "epoch": 2.1406801709719385,
      "grad_norm": 0.9043660759925842,
      "learning_rate": 8.91956104472201e-05,
      "loss": 0.9652,
      "step": 11519
    },
    {
      "epoch": 2.1408660100353094,
      "grad_norm": 0.6800969243049622,
      "learning_rate": 8.918108674299934e-05,
      "loss": 0.8477,
      "step": 11520
    },
    {
      "epoch": 2.1410518490986807,
      "grad_norm": 0.8428097367286682,
      "learning_rate": 8.916656326968973e-05,
      "loss": 1.132,
      "step": 11521
    },
    {
      "epoch": 2.1412376881620516,
      "grad_norm": 0.6179491877555847,
      "learning_rate": 8.915204002760122e-05,
      "loss": 0.6547,
      "step": 11522
    },
    {
      "epoch": 2.141423527225423,
      "grad_norm": 0.738825261592865,
      "learning_rate": 8.913751701704383e-05,
      "loss": 0.9981,
      "step": 11523
    },
    {
      "epoch": 2.141609366288794,
      "grad_norm": 0.7048845291137695,
      "learning_rate": 8.912299423832746e-05,
      "loss": 0.7569,
      "step": 11524
    },
    {
      "epoch": 2.141795205352165,
      "grad_norm": 0.790368914604187,
      "learning_rate": 8.910847169176213e-05,
      "loss": 0.7607,
      "step": 11525
    },
    {
      "epoch": 2.141981044415536,
      "grad_norm": 0.9649472832679749,
      "learning_rate": 8.909394937765778e-05,
      "loss": 0.8596,
      "step": 11526
    },
    {
      "epoch": 2.1421668834789074,
      "grad_norm": 0.7964535355567932,
      "learning_rate": 8.907942729632439e-05,
      "loss": 0.8862,
      "step": 11527
    },
    {
      "epoch": 2.1423527225422783,
      "grad_norm": 0.8461882472038269,
      "learning_rate": 8.906490544807184e-05,
      "loss": 0.842,
      "step": 11528
    },
    {
      "epoch": 2.1425385616056496,
      "grad_norm": 0.9009879231452942,
      "learning_rate": 8.90503838332101e-05,
      "loss": 0.8015,
      "step": 11529
    },
    {
      "epoch": 2.1427244006690205,
      "grad_norm": 0.6781339049339294,
      "learning_rate": 8.903586245204917e-05,
      "loss": 0.9277,
      "step": 11530
    },
    {
      "epoch": 2.142910239732392,
      "grad_norm": 0.8017725944519043,
      "learning_rate": 8.902134130489895e-05,
      "loss": 0.8699,
      "step": 11531
    },
    {
      "epoch": 2.1430960787957627,
      "grad_norm": 0.863451361656189,
      "learning_rate": 8.900682039206932e-05,
      "loss": 0.8769,
      "step": 11532
    },
    {
      "epoch": 2.143281917859134,
      "grad_norm": 0.692290723323822,
      "learning_rate": 8.899229971387029e-05,
      "loss": 0.8921,
      "step": 11533
    },
    {
      "epoch": 2.143467756922505,
      "grad_norm": 1.104231834411621,
      "learning_rate": 8.897777927061169e-05,
      "loss": 1.2087,
      "step": 11534
    },
    {
      "epoch": 2.1436535959858762,
      "grad_norm": 0.6479054689407349,
      "learning_rate": 8.896325906260352e-05,
      "loss": 0.5637,
      "step": 11535
    },
    {
      "epoch": 2.143839435049247,
      "grad_norm": 0.854199230670929,
      "learning_rate": 8.89487390901556e-05,
      "loss": 1.0063,
      "step": 11536
    },
    {
      "epoch": 2.1440252741126185,
      "grad_norm": 0.896887481212616,
      "learning_rate": 8.893421935357793e-05,
      "loss": 1.0112,
      "step": 11537
    },
    {
      "epoch": 2.14421111317599,
      "grad_norm": 0.8437681794166565,
      "learning_rate": 8.891969985318031e-05,
      "loss": 1.0294,
      "step": 11538
    },
    {
      "epoch": 2.1443969522393607,
      "grad_norm": 0.767591118812561,
      "learning_rate": 8.890518058927272e-05,
      "loss": 0.972,
      "step": 11539
    },
    {
      "epoch": 2.144582791302732,
      "grad_norm": 0.8373448848724365,
      "learning_rate": 8.889066156216495e-05,
      "loss": 1.0073,
      "step": 11540
    },
    {
      "epoch": 2.144768630366103,
      "grad_norm": 0.7533522844314575,
      "learning_rate": 8.8876142772167e-05,
      "loss": 0.7912,
      "step": 11541
    },
    {
      "epoch": 2.144954469429474,
      "grad_norm": 0.9369121789932251,
      "learning_rate": 8.886162421958867e-05,
      "loss": 0.8993,
      "step": 11542
    },
    {
      "epoch": 2.145140308492845,
      "grad_norm": 0.7979742884635925,
      "learning_rate": 8.884710590473988e-05,
      "loss": 0.8112,
      "step": 11543
    },
    {
      "epoch": 2.1453261475562164,
      "grad_norm": 0.9218032360076904,
      "learning_rate": 8.883258782793045e-05,
      "loss": 0.9041,
      "step": 11544
    },
    {
      "epoch": 2.1455119866195873,
      "grad_norm": 0.894201397895813,
      "learning_rate": 8.88180699894703e-05,
      "loss": 0.81,
      "step": 11545
    },
    {
      "epoch": 2.1456978256829586,
      "grad_norm": 0.8262656331062317,
      "learning_rate": 8.880355238966923e-05,
      "loss": 1.0715,
      "step": 11546
    },
    {
      "epoch": 2.1458836647463295,
      "grad_norm": 1.118515133857727,
      "learning_rate": 8.878903502883714e-05,
      "loss": 0.7873,
      "step": 11547
    },
    {
      "epoch": 2.146069503809701,
      "grad_norm": 0.6342157125473022,
      "learning_rate": 8.877451790728383e-05,
      "loss": 0.7688,
      "step": 11548
    },
    {
      "epoch": 2.1462553428730717,
      "grad_norm": 0.8411121368408203,
      "learning_rate": 8.87600010253192e-05,
      "loss": 0.8673,
      "step": 11549
    },
    {
      "epoch": 2.146441181936443,
      "grad_norm": 0.7336524128913879,
      "learning_rate": 8.874548438325301e-05,
      "loss": 0.6193,
      "step": 11550
    },
    {
      "epoch": 2.146627020999814,
      "grad_norm": 0.8948276042938232,
      "learning_rate": 8.873096798139518e-05,
      "loss": 0.9816,
      "step": 11551
    },
    {
      "epoch": 2.1468128600631853,
      "grad_norm": 0.7167133688926697,
      "learning_rate": 8.87164518200555e-05,
      "loss": 0.9412,
      "step": 11552
    },
    {
      "epoch": 2.1469986991265566,
      "grad_norm": 0.7607763409614563,
      "learning_rate": 8.870193589954381e-05,
      "loss": 0.8501,
      "step": 11553
    },
    {
      "epoch": 2.1471845381899275,
      "grad_norm": 0.8126594424247742,
      "learning_rate": 8.868742022016989e-05,
      "loss": 0.8756,
      "step": 11554
    },
    {
      "epoch": 2.147370377253299,
      "grad_norm": 0.8860375285148621,
      "learning_rate": 8.867290478224359e-05,
      "loss": 0.7168,
      "step": 11555
    },
    {
      "epoch": 2.1475562163166697,
      "grad_norm": 0.901677131652832,
      "learning_rate": 8.865838958607467e-05,
      "loss": 0.8255,
      "step": 11556
    },
    {
      "epoch": 2.147742055380041,
      "grad_norm": 0.7999879121780396,
      "learning_rate": 8.864387463197299e-05,
      "loss": 0.8414,
      "step": 11557
    },
    {
      "epoch": 2.147927894443412,
      "grad_norm": 0.7665722370147705,
      "learning_rate": 8.862935992024829e-05,
      "loss": 0.9855,
      "step": 11558
    },
    {
      "epoch": 2.1481137335067833,
      "grad_norm": 0.7725609540939331,
      "learning_rate": 8.861484545121041e-05,
      "loss": 0.8761,
      "step": 11559
    },
    {
      "epoch": 2.148299572570154,
      "grad_norm": 0.6915417909622192,
      "learning_rate": 8.86003312251691e-05,
      "loss": 0.7849,
      "step": 11560
    },
    {
      "epoch": 2.1484854116335255,
      "grad_norm": 0.7465013265609741,
      "learning_rate": 8.858581724243415e-05,
      "loss": 1.0637,
      "step": 11561
    },
    {
      "epoch": 2.1486712506968964,
      "grad_norm": 0.766620397567749,
      "learning_rate": 8.857130350331535e-05,
      "loss": 0.839,
      "step": 11562
    },
    {
      "epoch": 2.1488570897602677,
      "grad_norm": 0.7727715969085693,
      "learning_rate": 8.855679000812249e-05,
      "loss": 0.9321,
      "step": 11563
    },
    {
      "epoch": 2.1490429288236386,
      "grad_norm": 0.8486520051956177,
      "learning_rate": 8.854227675716529e-05,
      "loss": 0.8663,
      "step": 11564
    },
    {
      "epoch": 2.14922876788701,
      "grad_norm": 0.7445151805877686,
      "learning_rate": 8.852776375075355e-05,
      "loss": 0.761,
      "step": 11565
    },
    {
      "epoch": 2.149414606950381,
      "grad_norm": 0.9629629850387573,
      "learning_rate": 8.851325098919698e-05,
      "loss": 1.2405,
      "step": 11566
    },
    {
      "epoch": 2.149600446013752,
      "grad_norm": 0.9040130972862244,
      "learning_rate": 8.849873847280537e-05,
      "loss": 0.9975,
      "step": 11567
    },
    {
      "epoch": 2.1497862850771234,
      "grad_norm": 0.8581045866012573,
      "learning_rate": 8.848422620188844e-05,
      "loss": 0.9953,
      "step": 11568
    },
    {
      "epoch": 2.1499721241404943,
      "grad_norm": 0.6524201035499573,
      "learning_rate": 8.846971417675596e-05,
      "loss": 0.6089,
      "step": 11569
    },
    {
      "epoch": 2.1501579632038657,
      "grad_norm": 0.9643622636795044,
      "learning_rate": 8.845520239771762e-05,
      "loss": 1.1088,
      "step": 11570
    },
    {
      "epoch": 2.1503438022672365,
      "grad_norm": 0.8635513782501221,
      "learning_rate": 8.844069086508316e-05,
      "loss": 0.9776,
      "step": 11571
    },
    {
      "epoch": 2.150529641330608,
      "grad_norm": 0.6922736763954163,
      "learning_rate": 8.842617957916234e-05,
      "loss": 0.8603,
      "step": 11572
    },
    {
      "epoch": 2.1507154803939788,
      "grad_norm": 0.8901512026786804,
      "learning_rate": 8.841166854026487e-05,
      "loss": 1.153,
      "step": 11573
    },
    {
      "epoch": 2.15090131945735,
      "grad_norm": 0.8341671228408813,
      "learning_rate": 8.839715774870042e-05,
      "loss": 1.0115,
      "step": 11574
    },
    {
      "epoch": 2.151087158520721,
      "grad_norm": 0.7496020197868347,
      "learning_rate": 8.838264720477875e-05,
      "loss": 1.0316,
      "step": 11575
    },
    {
      "epoch": 2.1512729975840923,
      "grad_norm": 0.7692998051643372,
      "learning_rate": 8.836813690880953e-05,
      "loss": 0.9599,
      "step": 11576
    },
    {
      "epoch": 2.151458836647463,
      "grad_norm": 0.7583287358283997,
      "learning_rate": 8.835362686110248e-05,
      "loss": 0.8289,
      "step": 11577
    },
    {
      "epoch": 2.1516446757108345,
      "grad_norm": 0.7585510015487671,
      "learning_rate": 8.833911706196725e-05,
      "loss": 0.8314,
      "step": 11578
    },
    {
      "epoch": 2.1518305147742054,
      "grad_norm": 0.8208407759666443,
      "learning_rate": 8.83246075117136e-05,
      "loss": 0.8846,
      "step": 11579
    },
    {
      "epoch": 2.1520163538375767,
      "grad_norm": 0.6847831010818481,
      "learning_rate": 8.831009821065113e-05,
      "loss": 0.884,
      "step": 11580
    },
    {
      "epoch": 2.1522021929009476,
      "grad_norm": 0.8710030913352966,
      "learning_rate": 8.829558915908956e-05,
      "loss": 1.1368,
      "step": 11581
    },
    {
      "epoch": 2.152388031964319,
      "grad_norm": 0.7906744480133057,
      "learning_rate": 8.82810803573385e-05,
      "loss": 0.8698,
      "step": 11582
    },
    {
      "epoch": 2.15257387102769,
      "grad_norm": 0.667944610118866,
      "learning_rate": 8.826657180570775e-05,
      "loss": 0.6606,
      "step": 11583
    },
    {
      "epoch": 2.152759710091061,
      "grad_norm": 0.7219695448875427,
      "learning_rate": 8.825206350450685e-05,
      "loss": 0.7216,
      "step": 11584
    },
    {
      "epoch": 2.152945549154432,
      "grad_norm": 0.7834694981575012,
      "learning_rate": 8.823755545404553e-05,
      "loss": 0.9002,
      "step": 11585
    },
    {
      "epoch": 2.1531313882178034,
      "grad_norm": 0.7436776757240295,
      "learning_rate": 8.822304765463337e-05,
      "loss": 0.7823,
      "step": 11586
    },
    {
      "epoch": 2.1533172272811747,
      "grad_norm": 0.7721044421195984,
      "learning_rate": 8.820854010658007e-05,
      "loss": 0.7665,
      "step": 11587
    },
    {
      "epoch": 2.1535030663445456,
      "grad_norm": 0.899067759513855,
      "learning_rate": 8.819403281019522e-05,
      "loss": 1.1261,
      "step": 11588
    },
    {
      "epoch": 2.153688905407917,
      "grad_norm": 0.7515740394592285,
      "learning_rate": 8.817952576578851e-05,
      "loss": 1.1386,
      "step": 11589
    },
    {
      "epoch": 2.153874744471288,
      "grad_norm": 0.7487910985946655,
      "learning_rate": 8.816501897366953e-05,
      "loss": 0.8617,
      "step": 11590
    },
    {
      "epoch": 2.154060583534659,
      "grad_norm": 0.7791520953178406,
      "learning_rate": 8.815051243414792e-05,
      "loss": 0.7485,
      "step": 11591
    },
    {
      "epoch": 2.15424642259803,
      "grad_norm": 0.7022900581359863,
      "learning_rate": 8.813600614753322e-05,
      "loss": 0.8004,
      "step": 11592
    },
    {
      "epoch": 2.1544322616614013,
      "grad_norm": 0.7686887383460999,
      "learning_rate": 8.812150011413517e-05,
      "loss": 0.9525,
      "step": 11593
    },
    {
      "epoch": 2.1546181007247722,
      "grad_norm": 0.8644038438796997,
      "learning_rate": 8.81069943342633e-05,
      "loss": 0.9486,
      "step": 11594
    },
    {
      "epoch": 2.1548039397881436,
      "grad_norm": 1.0755417346954346,
      "learning_rate": 8.809248880822725e-05,
      "loss": 1.0545,
      "step": 11595
    },
    {
      "epoch": 2.1549897788515144,
      "grad_norm": 0.9233502149581909,
      "learning_rate": 8.807798353633657e-05,
      "loss": 0.9696,
      "step": 11596
    },
    {
      "epoch": 2.1551756179148858,
      "grad_norm": 0.885118305683136,
      "learning_rate": 8.806347851890088e-05,
      "loss": 0.9493,
      "step": 11597
    },
    {
      "epoch": 2.1553614569782567,
      "grad_norm": 0.7692207098007202,
      "learning_rate": 8.804897375622975e-05,
      "loss": 1.1325,
      "step": 11598
    },
    {
      "epoch": 2.155547296041628,
      "grad_norm": 0.8364658951759338,
      "learning_rate": 8.80344692486328e-05,
      "loss": 0.8861,
      "step": 11599
    },
    {
      "epoch": 2.155733135104999,
      "grad_norm": 0.6783618330955505,
      "learning_rate": 8.801996499641951e-05,
      "loss": 0.7697,
      "step": 11600
    },
    {
      "epoch": 2.15591897416837,
      "grad_norm": 0.8993368148803711,
      "learning_rate": 8.800546099989953e-05,
      "loss": 0.7748,
      "step": 11601
    },
    {
      "epoch": 2.1561048132317415,
      "grad_norm": 0.9230395555496216,
      "learning_rate": 8.799095725938243e-05,
      "loss": 1.2362,
      "step": 11602
    },
    {
      "epoch": 2.1562906522951124,
      "grad_norm": 0.8871796727180481,
      "learning_rate": 8.79764537751777e-05,
      "loss": 0.9197,
      "step": 11603
    },
    {
      "epoch": 2.1564764913584837,
      "grad_norm": 0.6526246070861816,
      "learning_rate": 8.796195054759493e-05,
      "loss": 0.5867,
      "step": 11604
    },
    {
      "epoch": 2.1566623304218546,
      "grad_norm": 0.8340665102005005,
      "learning_rate": 8.79474475769437e-05,
      "loss": 1.0749,
      "step": 11605
    },
    {
      "epoch": 2.156848169485226,
      "grad_norm": 0.8240909576416016,
      "learning_rate": 8.793294486353351e-05,
      "loss": 0.8532,
      "step": 11606
    },
    {
      "epoch": 2.157034008548597,
      "grad_norm": 0.7156884074211121,
      "learning_rate": 8.791844240767391e-05,
      "loss": 0.8128,
      "step": 11607
    },
    {
      "epoch": 2.157219847611968,
      "grad_norm": 0.7586245536804199,
      "learning_rate": 8.790394020967441e-05,
      "loss": 0.8983,
      "step": 11608
    },
    {
      "epoch": 2.157405686675339,
      "grad_norm": 0.7270494103431702,
      "learning_rate": 8.788943826984457e-05,
      "loss": 0.8398,
      "step": 11609
    },
    {
      "epoch": 2.1575915257387104,
      "grad_norm": 0.9726952314376831,
      "learning_rate": 8.787493658849386e-05,
      "loss": 0.958,
      "step": 11610
    },
    {
      "epoch": 2.1577773648020813,
      "grad_norm": 0.7318455576896667,
      "learning_rate": 8.786043516593183e-05,
      "loss": 0.7498,
      "step": 11611
    },
    {
      "epoch": 2.1579632038654526,
      "grad_norm": 0.9342050552368164,
      "learning_rate": 8.784593400246799e-05,
      "loss": 0.9865,
      "step": 11612
    },
    {
      "epoch": 2.1581490429288235,
      "grad_norm": 0.735383152961731,
      "learning_rate": 8.78314330984118e-05,
      "loss": 0.795,
      "step": 11613
    },
    {
      "epoch": 2.158334881992195,
      "grad_norm": 0.8356888890266418,
      "learning_rate": 8.781693245407279e-05,
      "loss": 1.1769,
      "step": 11614
    },
    {
      "epoch": 2.1585207210555657,
      "grad_norm": 0.790277898311615,
      "learning_rate": 8.78024320697605e-05,
      "loss": 0.7551,
      "step": 11615
    },
    {
      "epoch": 2.158706560118937,
      "grad_norm": 0.8385562896728516,
      "learning_rate": 8.778793194578433e-05,
      "loss": 1.0738,
      "step": 11616
    },
    {
      "epoch": 2.1588923991823084,
      "grad_norm": 0.7916375398635864,
      "learning_rate": 8.777343208245382e-05,
      "loss": 0.9502,
      "step": 11617
    },
    {
      "epoch": 2.1590782382456792,
      "grad_norm": 0.7239101529121399,
      "learning_rate": 8.775893248007839e-05,
      "loss": 0.9905,
      "step": 11618
    },
    {
      "epoch": 2.1592640773090506,
      "grad_norm": 0.6555399894714355,
      "learning_rate": 8.774443313896757e-05,
      "loss": 0.7378,
      "step": 11619
    },
    {
      "epoch": 2.1594499163724215,
      "grad_norm": 0.8354867696762085,
      "learning_rate": 8.772993405943077e-05,
      "loss": 0.9485,
      "step": 11620
    },
    {
      "epoch": 2.159635755435793,
      "grad_norm": 0.7708397507667542,
      "learning_rate": 8.771543524177747e-05,
      "loss": 0.9601,
      "step": 11621
    },
    {
      "epoch": 2.1598215944991637,
      "grad_norm": 0.7791001200675964,
      "learning_rate": 8.770093668631714e-05,
      "loss": 0.7209,
      "step": 11622
    },
    {
      "epoch": 2.160007433562535,
      "grad_norm": 0.7207884788513184,
      "learning_rate": 8.76864383933592e-05,
      "loss": 0.8228,
      "step": 11623
    },
    {
      "epoch": 2.160193272625906,
      "grad_norm": 1.162050724029541,
      "learning_rate": 8.767194036321307e-05,
      "loss": 0.888,
      "step": 11624
    },
    {
      "epoch": 2.160379111689277,
      "grad_norm": 0.6758953332901001,
      "learning_rate": 8.765744259618826e-05,
      "loss": 0.8817,
      "step": 11625
    },
    {
      "epoch": 2.160564950752648,
      "grad_norm": 0.9050666689872742,
      "learning_rate": 8.764294509259414e-05,
      "loss": 0.8469,
      "step": 11626
    },
    {
      "epoch": 2.1607507898160194,
      "grad_norm": 0.8308858871459961,
      "learning_rate": 8.762844785274017e-05,
      "loss": 0.7436,
      "step": 11627
    },
    {
      "epoch": 2.1609366288793903,
      "grad_norm": 0.7234518527984619,
      "learning_rate": 8.761395087693573e-05,
      "loss": 0.9538,
      "step": 11628
    },
    {
      "epoch": 2.1611224679427616,
      "grad_norm": 0.6110588908195496,
      "learning_rate": 8.759945416549028e-05,
      "loss": 0.6267,
      "step": 11629
    },
    {
      "epoch": 2.1613083070061325,
      "grad_norm": 0.7216711640357971,
      "learning_rate": 8.758495771871318e-05,
      "loss": 0.7065,
      "step": 11630
    },
    {
      "epoch": 2.161494146069504,
      "grad_norm": 1.0335111618041992,
      "learning_rate": 8.757046153691383e-05,
      "loss": 0.8654,
      "step": 11631
    },
    {
      "epoch": 2.1616799851328747,
      "grad_norm": 0.7840709090232849,
      "learning_rate": 8.75559656204017e-05,
      "loss": 0.8205,
      "step": 11632
    },
    {
      "epoch": 2.161865824196246,
      "grad_norm": 0.8022106289863586,
      "learning_rate": 8.754146996948609e-05,
      "loss": 0.7341,
      "step": 11633
    },
    {
      "epoch": 2.162051663259617,
      "grad_norm": 0.7449596524238586,
      "learning_rate": 8.752697458447641e-05,
      "loss": 0.7446,
      "step": 11634
    },
    {
      "epoch": 2.1622375023229883,
      "grad_norm": 0.6985468864440918,
      "learning_rate": 8.75124794656821e-05,
      "loss": 0.9942,
      "step": 11635
    },
    {
      "epoch": 2.1624233413863596,
      "grad_norm": 0.740747332572937,
      "learning_rate": 8.749798461341246e-05,
      "loss": 0.8125,
      "step": 11636
    },
    {
      "epoch": 2.1626091804497305,
      "grad_norm": 0.6656926274299622,
      "learning_rate": 8.748349002797688e-05,
      "loss": 0.8239,
      "step": 11637
    },
    {
      "epoch": 2.162795019513102,
      "grad_norm": 0.7795909643173218,
      "learning_rate": 8.746899570968474e-05,
      "loss": 0.843,
      "step": 11638
    },
    {
      "epoch": 2.1629808585764727,
      "grad_norm": 0.8801382184028625,
      "learning_rate": 8.74545016588454e-05,
      "loss": 0.7779,
      "step": 11639
    },
    {
      "epoch": 2.163166697639844,
      "grad_norm": 0.7868742346763611,
      "learning_rate": 8.744000787576816e-05,
      "loss": 0.8968,
      "step": 11640
    },
    {
      "epoch": 2.163352536703215,
      "grad_norm": 0.8084865212440491,
      "learning_rate": 8.742551436076241e-05,
      "loss": 0.8565,
      "step": 11641
    },
    {
      "epoch": 2.1635383757665863,
      "grad_norm": 0.7272623777389526,
      "learning_rate": 8.741102111413748e-05,
      "loss": 0.8322,
      "step": 11642
    },
    {
      "epoch": 2.163724214829957,
      "grad_norm": 0.7027956247329712,
      "learning_rate": 8.739652813620269e-05,
      "loss": 0.8649,
      "step": 11643
    },
    {
      "epoch": 2.1639100538933285,
      "grad_norm": 0.879515528678894,
      "learning_rate": 8.73820354272674e-05,
      "loss": 1.1277,
      "step": 11644
    },
    {
      "epoch": 2.1640958929566994,
      "grad_norm": 0.8235723972320557,
      "learning_rate": 8.736754298764087e-05,
      "loss": 0.9575,
      "step": 11645
    },
    {
      "epoch": 2.1642817320200707,
      "grad_norm": 0.8570378422737122,
      "learning_rate": 8.735305081763248e-05,
      "loss": 1.0092,
      "step": 11646
    },
    {
      "epoch": 2.1644675710834416,
      "grad_norm": 0.8653255701065063,
      "learning_rate": 8.733855891755153e-05,
      "loss": 0.8659,
      "step": 11647
    },
    {
      "epoch": 2.164653410146813,
      "grad_norm": 0.6897451877593994,
      "learning_rate": 8.73240672877073e-05,
      "loss": 1.0281,
      "step": 11648
    },
    {
      "epoch": 2.164839249210184,
      "grad_norm": 0.8468342423439026,
      "learning_rate": 8.730957592840912e-05,
      "loss": 0.9191,
      "step": 11649
    },
    {
      "epoch": 2.165025088273555,
      "grad_norm": 0.8642510771751404,
      "learning_rate": 8.729508483996624e-05,
      "loss": 0.9842,
      "step": 11650
    },
    {
      "epoch": 2.1652109273369264,
      "grad_norm": 0.8552669286727905,
      "learning_rate": 8.728059402268796e-05,
      "loss": 0.9799,
      "step": 11651
    },
    {
      "epoch": 2.1653967664002973,
      "grad_norm": 0.7763219475746155,
      "learning_rate": 8.726610347688361e-05,
      "loss": 1.0787,
      "step": 11652
    },
    {
      "epoch": 2.1655826054636687,
      "grad_norm": 1.0020298957824707,
      "learning_rate": 8.72516132028624e-05,
      "loss": 0.8941,
      "step": 11653
    },
    {
      "epoch": 2.1657684445270395,
      "grad_norm": 1.0689773559570312,
      "learning_rate": 8.723712320093367e-05,
      "loss": 0.9945,
      "step": 11654
    },
    {
      "epoch": 2.165954283590411,
      "grad_norm": 0.9250777363777161,
      "learning_rate": 8.722263347140658e-05,
      "loss": 0.8263,
      "step": 11655
    },
    {
      "epoch": 2.1661401226537818,
      "grad_norm": 0.8608018159866333,
      "learning_rate": 8.720814401459048e-05,
      "loss": 0.8399,
      "step": 11656
    },
    {
      "epoch": 2.166325961717153,
      "grad_norm": 0.9532923698425293,
      "learning_rate": 8.719365483079461e-05,
      "loss": 0.917,
      "step": 11657
    },
    {
      "epoch": 2.166511800780524,
      "grad_norm": 0.8235387802124023,
      "learning_rate": 8.717916592032818e-05,
      "loss": 0.8097,
      "step": 11658
    },
    {
      "epoch": 2.1666976398438953,
      "grad_norm": 0.7674156427383423,
      "learning_rate": 8.716467728350049e-05,
      "loss": 0.7582,
      "step": 11659
    },
    {
      "epoch": 2.166883478907266,
      "grad_norm": 0.7772092819213867,
      "learning_rate": 8.715018892062069e-05,
      "loss": 0.9003,
      "step": 11660
    },
    {
      "epoch": 2.1670693179706375,
      "grad_norm": 0.7887098789215088,
      "learning_rate": 8.713570083199806e-05,
      "loss": 0.9104,
      "step": 11661
    },
    {
      "epoch": 2.1672551570340084,
      "grad_norm": 0.8114263415336609,
      "learning_rate": 8.712121301794186e-05,
      "loss": 0.8308,
      "step": 11662
    },
    {
      "epoch": 2.1674409960973797,
      "grad_norm": 0.7965273261070251,
      "learning_rate": 8.710672547876124e-05,
      "loss": 0.7873,
      "step": 11663
    },
    {
      "epoch": 2.1676268351607506,
      "grad_norm": 0.800265908241272,
      "learning_rate": 8.709223821476546e-05,
      "loss": 0.9765,
      "step": 11664
    },
    {
      "epoch": 2.167812674224122,
      "grad_norm": 0.7654651403427124,
      "learning_rate": 8.707775122626367e-05,
      "loss": 0.9037,
      "step": 11665
    },
    {
      "epoch": 2.167998513287493,
      "grad_norm": 0.9573057293891907,
      "learning_rate": 8.706326451356511e-05,
      "loss": 0.8409,
      "step": 11666
    },
    {
      "epoch": 2.168184352350864,
      "grad_norm": 0.9364465475082397,
      "learning_rate": 8.704877807697897e-05,
      "loss": 0.9115,
      "step": 11667
    },
    {
      "epoch": 2.168370191414235,
      "grad_norm": 0.8293829560279846,
      "learning_rate": 8.703429191681445e-05,
      "loss": 0.8936,
      "step": 11668
    },
    {
      "epoch": 2.1685560304776064,
      "grad_norm": 0.8080455660820007,
      "learning_rate": 8.701980603338073e-05,
      "loss": 0.8978,
      "step": 11669
    },
    {
      "epoch": 2.1687418695409777,
      "grad_norm": 1.0206828117370605,
      "learning_rate": 8.700532042698696e-05,
      "loss": 0.9648,
      "step": 11670
    },
    {
      "epoch": 2.1689277086043486,
      "grad_norm": 0.8141977190971375,
      "learning_rate": 8.699083509794231e-05,
      "loss": 0.9385,
      "step": 11671
    },
    {
      "epoch": 2.16911354766772,
      "grad_norm": 0.8308578729629517,
      "learning_rate": 8.697635004655599e-05,
      "loss": 1.0552,
      "step": 11672
    },
    {
      "epoch": 2.169299386731091,
      "grad_norm": 0.8184531331062317,
      "learning_rate": 8.696186527313711e-05,
      "loss": 1.0682,
      "step": 11673
    },
    {
      "epoch": 2.169485225794462,
      "grad_norm": 0.7495091557502747,
      "learning_rate": 8.694738077799488e-05,
      "loss": 0.9417,
      "step": 11674
    },
    {
      "epoch": 2.169671064857833,
      "grad_norm": 0.9168591499328613,
      "learning_rate": 8.693289656143835e-05,
      "loss": 0.9295,
      "step": 11675
    },
    {
      "epoch": 2.1698569039212043,
      "grad_norm": 0.6868381500244141,
      "learning_rate": 8.691841262377671e-05,
      "loss": 0.6573,
      "step": 11676
    },
    {
      "epoch": 2.1700427429845752,
      "grad_norm": 0.7707226276397705,
      "learning_rate": 8.690392896531916e-05,
      "loss": 0.9796,
      "step": 11677
    },
    {
      "epoch": 2.1702285820479466,
      "grad_norm": 1.0843822956085205,
      "learning_rate": 8.688944558637474e-05,
      "loss": 0.939,
      "step": 11678
    },
    {
      "epoch": 2.1704144211113174,
      "grad_norm": 0.8778970241546631,
      "learning_rate": 8.687496248725263e-05,
      "loss": 1.0477,
      "step": 11679
    },
    {
      "epoch": 2.1706002601746888,
      "grad_norm": 0.8354715704917908,
      "learning_rate": 8.68604796682619e-05,
      "loss": 0.9234,
      "step": 11680
    },
    {
      "epoch": 2.1707860992380597,
      "grad_norm": 0.8660401105880737,
      "learning_rate": 8.684599712971168e-05,
      "loss": 0.8527,
      "step": 11681
    },
    {
      "epoch": 2.170971938301431,
      "grad_norm": 0.772363007068634,
      "learning_rate": 8.68315148719111e-05,
      "loss": 0.9042,
      "step": 11682
    },
    {
      "epoch": 2.171157777364802,
      "grad_norm": 0.7308756709098816,
      "learning_rate": 8.681703289516922e-05,
      "loss": 1.0071,
      "step": 11683
    },
    {
      "epoch": 2.171343616428173,
      "grad_norm": 0.7892624139785767,
      "learning_rate": 8.680255119979518e-05,
      "loss": 0.8131,
      "step": 11684
    },
    {
      "epoch": 2.1715294554915445,
      "grad_norm": 1.0934081077575684,
      "learning_rate": 8.678806978609801e-05,
      "loss": 1.3285,
      "step": 11685
    },
    {
      "epoch": 2.1717152945549154,
      "grad_norm": 0.7644238471984863,
      "learning_rate": 8.677358865438682e-05,
      "loss": 0.8815,
      "step": 11686
    },
    {
      "epoch": 2.1719011336182867,
      "grad_norm": 0.8351262807846069,
      "learning_rate": 8.675910780497068e-05,
      "loss": 0.9617,
      "step": 11687
    },
    {
      "epoch": 2.1720869726816576,
      "grad_norm": 0.8664665818214417,
      "learning_rate": 8.674462723815865e-05,
      "loss": 1.0228,
      "step": 11688
    },
    {
      "epoch": 2.172272811745029,
      "grad_norm": 0.82393479347229,
      "learning_rate": 8.673014695425983e-05,
      "loss": 1.0277,
      "step": 11689
    },
    {
      "epoch": 2.1724586508084,
      "grad_norm": 0.7539329528808594,
      "learning_rate": 8.671566695358324e-05,
      "loss": 0.8537,
      "step": 11690
    },
    {
      "epoch": 2.172644489871771,
      "grad_norm": 0.9268426299095154,
      "learning_rate": 8.670118723643792e-05,
      "loss": 0.8671,
      "step": 11691
    },
    {
      "epoch": 2.172830328935142,
      "grad_norm": 0.8412514328956604,
      "learning_rate": 8.668670780313299e-05,
      "loss": 0.9128,
      "step": 11692
    },
    {
      "epoch": 2.1730161679985134,
      "grad_norm": 0.7689191102981567,
      "learning_rate": 8.66722286539774e-05,
      "loss": 0.8982,
      "step": 11693
    },
    {
      "epoch": 2.1732020070618843,
      "grad_norm": 0.8509365320205688,
      "learning_rate": 8.665774978928023e-05,
      "loss": 0.7322,
      "step": 11694
    },
    {
      "epoch": 2.1733878461252556,
      "grad_norm": 0.6554549932479858,
      "learning_rate": 8.664327120935047e-05,
      "loss": 0.8867,
      "step": 11695
    },
    {
      "epoch": 2.1735736851886265,
      "grad_norm": 0.8374941349029541,
      "learning_rate": 8.662879291449722e-05,
      "loss": 0.7539,
      "step": 11696
    },
    {
      "epoch": 2.173759524251998,
      "grad_norm": 0.8129752278327942,
      "learning_rate": 8.661431490502936e-05,
      "loss": 0.8304,
      "step": 11697
    },
    {
      "epoch": 2.1739453633153687,
      "grad_norm": 0.7586342096328735,
      "learning_rate": 8.659983718125602e-05,
      "loss": 0.8621,
      "step": 11698
    },
    {
      "epoch": 2.17413120237874,
      "grad_norm": 0.7914800047874451,
      "learning_rate": 8.658535974348617e-05,
      "loss": 0.8289,
      "step": 11699
    },
    {
      "epoch": 2.1743170414421114,
      "grad_norm": 1.0254422426223755,
      "learning_rate": 8.657088259202876e-05,
      "loss": 1.0067,
      "step": 11700
    },
    {
      "epoch": 2.1745028805054822,
      "grad_norm": 0.8690272569656372,
      "learning_rate": 8.655640572719285e-05,
      "loss": 1.1076,
      "step": 11701
    },
    {
      "epoch": 2.1746887195688536,
      "grad_norm": 0.9043143391609192,
      "learning_rate": 8.654192914928739e-05,
      "loss": 0.8491,
      "step": 11702
    },
    {
      "epoch": 2.1748745586322245,
      "grad_norm": 0.7068991661071777,
      "learning_rate": 8.652745285862134e-05,
      "loss": 0.8558,
      "step": 11703
    },
    {
      "epoch": 2.175060397695596,
      "grad_norm": 1.0964959859848022,
      "learning_rate": 8.651297685550371e-05,
      "loss": 1.084,
      "step": 11704
    },
    {
      "epoch": 2.1752462367589667,
      "grad_norm": 0.906252920627594,
      "learning_rate": 8.649850114024342e-05,
      "loss": 0.8628,
      "step": 11705
    },
    {
      "epoch": 2.175432075822338,
      "grad_norm": 0.9718541502952576,
      "learning_rate": 8.648402571314949e-05,
      "loss": 0.903,
      "step": 11706
    },
    {
      "epoch": 2.175617914885709,
      "grad_norm": 0.7787262201309204,
      "learning_rate": 8.64695505745308e-05,
      "loss": 1.0466,
      "step": 11707
    },
    {
      "epoch": 2.17580375394908,
      "grad_norm": 0.8103688955307007,
      "learning_rate": 8.645507572469633e-05,
      "loss": 1.0094,
      "step": 11708
    },
    {
      "epoch": 2.175989593012451,
      "grad_norm": 0.668194591999054,
      "learning_rate": 8.644060116395505e-05,
      "loss": 1.0853,
      "step": 11709
    },
    {
      "epoch": 2.1761754320758224,
      "grad_norm": 0.8936452269554138,
      "learning_rate": 8.642612689261584e-05,
      "loss": 1.0407,
      "step": 11710
    },
    {
      "epoch": 2.1763612711391933,
      "grad_norm": 0.8345995545387268,
      "learning_rate": 8.641165291098767e-05,
      "loss": 0.9775,
      "step": 11711
    },
    {
      "epoch": 2.1765471102025646,
      "grad_norm": 0.6872609853744507,
      "learning_rate": 8.639717921937947e-05,
      "loss": 0.763,
      "step": 11712
    },
    {
      "epoch": 2.1767329492659355,
      "grad_norm": 0.7668678760528564,
      "learning_rate": 8.638270581810013e-05,
      "loss": 0.8762,
      "step": 11713
    },
    {
      "epoch": 2.176918788329307,
      "grad_norm": 0.6786547303199768,
      "learning_rate": 8.636823270745858e-05,
      "loss": 0.5868,
      "step": 11714
    },
    {
      "epoch": 2.1771046273926777,
      "grad_norm": 0.7277447581291199,
      "learning_rate": 8.635375988776368e-05,
      "loss": 0.7244,
      "step": 11715
    },
    {
      "epoch": 2.177290466456049,
      "grad_norm": 0.7102178931236267,
      "learning_rate": 8.633928735932437e-05,
      "loss": 1.1194,
      "step": 11716
    },
    {
      "epoch": 2.17747630551942,
      "grad_norm": 0.8310472965240479,
      "learning_rate": 8.632481512244951e-05,
      "loss": 1.1853,
      "step": 11717
    },
    {
      "epoch": 2.1776621445827913,
      "grad_norm": 0.7447524070739746,
      "learning_rate": 8.6310343177448e-05,
      "loss": 0.7727,
      "step": 11718
    },
    {
      "epoch": 2.1778479836461626,
      "grad_norm": 0.7981902360916138,
      "learning_rate": 8.629587152462876e-05,
      "loss": 0.9509,
      "step": 11719
    },
    {
      "epoch": 2.1780338227095335,
      "grad_norm": 1.008462905883789,
      "learning_rate": 8.628140016430059e-05,
      "loss": 0.9314,
      "step": 11720
    },
    {
      "epoch": 2.178219661772905,
      "grad_norm": 0.8005197048187256,
      "learning_rate": 8.626692909677241e-05,
      "loss": 0.7536,
      "step": 11721
    },
    {
      "epoch": 2.1784055008362757,
      "grad_norm": 0.9703549146652222,
      "learning_rate": 8.625245832235307e-05,
      "loss": 1.0033,
      "step": 11722
    },
    {
      "epoch": 2.178591339899647,
      "grad_norm": 0.788756251335144,
      "learning_rate": 8.62379878413514e-05,
      "loss": 0.9943,
      "step": 11723
    },
    {
      "epoch": 2.178777178963018,
      "grad_norm": 1.0320338010787964,
      "learning_rate": 8.622351765407628e-05,
      "loss": 0.9527,
      "step": 11724
    },
    {
      "epoch": 2.1789630180263893,
      "grad_norm": 0.9259473085403442,
      "learning_rate": 8.620904776083652e-05,
      "loss": 1.1026,
      "step": 11725
    },
    {
      "epoch": 2.17914885708976,
      "grad_norm": 0.7380974888801575,
      "learning_rate": 8.619457816194099e-05,
      "loss": 0.823,
      "step": 11726
    },
    {
      "epoch": 2.1793346961531315,
      "grad_norm": 0.9026634693145752,
      "learning_rate": 8.618010885769845e-05,
      "loss": 0.7946,
      "step": 11727
    },
    {
      "epoch": 2.1795205352165024,
      "grad_norm": 0.82573002576828,
      "learning_rate": 8.616563984841783e-05,
      "loss": 1.0932,
      "step": 11728
    },
    {
      "epoch": 2.1797063742798737,
      "grad_norm": 0.8773999214172363,
      "learning_rate": 8.615117113440783e-05,
      "loss": 0.9676,
      "step": 11729
    },
    {
      "epoch": 2.1798922133432446,
      "grad_norm": 0.9673879742622375,
      "learning_rate": 8.613670271597733e-05,
      "loss": 0.9729,
      "step": 11730
    },
    {
      "epoch": 2.180078052406616,
      "grad_norm": 0.774804413318634,
      "learning_rate": 8.612223459343513e-05,
      "loss": 1.0304,
      "step": 11731
    },
    {
      "epoch": 2.180263891469987,
      "grad_norm": 0.8201828002929688,
      "learning_rate": 8.610776676709004e-05,
      "loss": 0.987,
      "step": 11732
    },
    {
      "epoch": 2.180449730533358,
      "grad_norm": 0.7700895667076111,
      "learning_rate": 8.60932992372508e-05,
      "loss": 0.9923,
      "step": 11733
    },
    {
      "epoch": 2.1806355695967294,
      "grad_norm": 0.7555466890335083,
      "learning_rate": 8.607883200422624e-05,
      "loss": 0.8199,
      "step": 11734
    },
    {
      "epoch": 2.1808214086601003,
      "grad_norm": 0.8858521580696106,
      "learning_rate": 8.606436506832512e-05,
      "loss": 0.9766,
      "step": 11735
    },
    {
      "epoch": 2.1810072477234717,
      "grad_norm": 0.6815047264099121,
      "learning_rate": 8.604989842985622e-05,
      "loss": 0.8501,
      "step": 11736
    },
    {
      "epoch": 2.1811930867868425,
      "grad_norm": 0.7316385507583618,
      "learning_rate": 8.603543208912827e-05,
      "loss": 0.8795,
      "step": 11737
    },
    {
      "epoch": 2.181378925850214,
      "grad_norm": 0.8119329214096069,
      "learning_rate": 8.602096604645009e-05,
      "loss": 1.1118,
      "step": 11738
    },
    {
      "epoch": 2.1815647649135848,
      "grad_norm": 0.8801705837249756,
      "learning_rate": 8.600650030213036e-05,
      "loss": 0.891,
      "step": 11739
    },
    {
      "epoch": 2.181750603976956,
      "grad_norm": 0.9159716963768005,
      "learning_rate": 8.599203485647788e-05,
      "loss": 0.8022,
      "step": 11740
    },
    {
      "epoch": 2.181936443040327,
      "grad_norm": 0.810187578201294,
      "learning_rate": 8.597756970980138e-05,
      "loss": 0.8708,
      "step": 11741
    },
    {
      "epoch": 2.1821222821036983,
      "grad_norm": 0.743595540523529,
      "learning_rate": 8.59631048624096e-05,
      "loss": 0.8445,
      "step": 11742
    },
    {
      "epoch": 2.182308121167069,
      "grad_norm": 0.8353040218353271,
      "learning_rate": 8.594864031461127e-05,
      "loss": 1.0315,
      "step": 11743
    },
    {
      "epoch": 2.1824939602304405,
      "grad_norm": 0.7551995515823364,
      "learning_rate": 8.593417606671509e-05,
      "loss": 1.0222,
      "step": 11744
    },
    {
      "epoch": 2.1826797992938114,
      "grad_norm": 0.7523811459541321,
      "learning_rate": 8.591971211902978e-05,
      "loss": 1.0328,
      "step": 11745
    },
    {
      "epoch": 2.1828656383571827,
      "grad_norm": 0.8597369194030762,
      "learning_rate": 8.590524847186407e-05,
      "loss": 0.689,
      "step": 11746
    },
    {
      "epoch": 2.1830514774205536,
      "grad_norm": 0.9100225567817688,
      "learning_rate": 8.589078512552662e-05,
      "loss": 0.8001,
      "step": 11747
    },
    {
      "epoch": 2.183237316483925,
      "grad_norm": 0.8944543600082397,
      "learning_rate": 8.587632208032617e-05,
      "loss": 1.2127,
      "step": 11748
    },
    {
      "epoch": 2.1834231555472963,
      "grad_norm": 0.7308274507522583,
      "learning_rate": 8.586185933657135e-05,
      "loss": 0.947,
      "step": 11749
    },
    {
      "epoch": 2.183608994610667,
      "grad_norm": 0.7620413303375244,
      "learning_rate": 8.584739689457089e-05,
      "loss": 0.7689,
      "step": 11750
    },
    {
      "epoch": 2.1837948336740385,
      "grad_norm": 0.7631767392158508,
      "learning_rate": 8.583293475463345e-05,
      "loss": 1.1856,
      "step": 11751
    },
    {
      "epoch": 2.1839806727374094,
      "grad_norm": 0.8975901007652283,
      "learning_rate": 8.581847291706775e-05,
      "loss": 0.9497,
      "step": 11752
    },
    {
      "epoch": 2.1841665118007807,
      "grad_norm": 0.8111382722854614,
      "learning_rate": 8.580401138218238e-05,
      "loss": 0.7698,
      "step": 11753
    },
    {
      "epoch": 2.1843523508641516,
      "grad_norm": 0.9419172406196594,
      "learning_rate": 8.578955015028605e-05,
      "loss": 0.7565,
      "step": 11754
    },
    {
      "epoch": 2.184538189927523,
      "grad_norm": 0.8097640872001648,
      "learning_rate": 8.577508922168735e-05,
      "loss": 1.0693,
      "step": 11755
    },
    {
      "epoch": 2.184724028990894,
      "grad_norm": 0.6869693994522095,
      "learning_rate": 8.5760628596695e-05,
      "loss": 0.8488,
      "step": 11756
    },
    {
      "epoch": 2.184909868054265,
      "grad_norm": 0.7901400327682495,
      "learning_rate": 8.574616827561756e-05,
      "loss": 0.96,
      "step": 11757
    },
    {
      "epoch": 2.185095707117636,
      "grad_norm": 0.8940836191177368,
      "learning_rate": 8.573170825876373e-05,
      "loss": 0.9835,
      "step": 11758
    },
    {
      "epoch": 2.1852815461810073,
      "grad_norm": 0.8824992179870605,
      "learning_rate": 8.571724854644207e-05,
      "loss": 1.007,
      "step": 11759
    },
    {
      "epoch": 2.1854673852443782,
      "grad_norm": 1.7196069955825806,
      "learning_rate": 8.570278913896122e-05,
      "loss": 1.3019,
      "step": 11760
    },
    {
      "epoch": 2.1856532243077496,
      "grad_norm": 0.7538241744041443,
      "learning_rate": 8.568833003662982e-05,
      "loss": 0.8869,
      "step": 11761
    },
    {
      "epoch": 2.1858390633711204,
      "grad_norm": 0.8255589604377747,
      "learning_rate": 8.567387123975648e-05,
      "loss": 0.9831,
      "step": 11762
    },
    {
      "epoch": 2.1860249024344918,
      "grad_norm": 0.8417258262634277,
      "learning_rate": 8.565941274864975e-05,
      "loss": 0.9011,
      "step": 11763
    },
    {
      "epoch": 2.1862107414978627,
      "grad_norm": 0.7021176218986511,
      "learning_rate": 8.564495456361825e-05,
      "loss": 0.8793,
      "step": 11764
    },
    {
      "epoch": 2.186396580561234,
      "grad_norm": 0.9574293494224548,
      "learning_rate": 8.563049668497057e-05,
      "loss": 0.9614,
      "step": 11765
    },
    {
      "epoch": 2.186582419624605,
      "grad_norm": 0.6656575202941895,
      "learning_rate": 8.561603911301528e-05,
      "loss": 0.8067,
      "step": 11766
    },
    {
      "epoch": 2.186768258687976,
      "grad_norm": 0.7005726099014282,
      "learning_rate": 8.560158184806093e-05,
      "loss": 0.8465,
      "step": 11767
    },
    {
      "epoch": 2.1869540977513475,
      "grad_norm": 1.0161550045013428,
      "learning_rate": 8.558712489041613e-05,
      "loss": 0.7472,
      "step": 11768
    },
    {
      "epoch": 2.1871399368147184,
      "grad_norm": 0.953964352607727,
      "learning_rate": 8.557266824038939e-05,
      "loss": 1.036,
      "step": 11769
    },
    {
      "epoch": 2.1873257758780897,
      "grad_norm": 0.7425580620765686,
      "learning_rate": 8.555821189828932e-05,
      "loss": 0.8413,
      "step": 11770
    },
    {
      "epoch": 2.1875116149414606,
      "grad_norm": 0.7482516765594482,
      "learning_rate": 8.554375586442436e-05,
      "loss": 0.9099,
      "step": 11771
    },
    {
      "epoch": 2.187697454004832,
      "grad_norm": 0.6497737765312195,
      "learning_rate": 8.55293001391032e-05,
      "loss": 0.7929,
      "step": 11772
    },
    {
      "epoch": 2.187883293068203,
      "grad_norm": 0.8159965872764587,
      "learning_rate": 8.551484472263426e-05,
      "loss": 0.9298,
      "step": 11773
    },
    {
      "epoch": 2.188069132131574,
      "grad_norm": 1.116498351097107,
      "learning_rate": 8.550038961532613e-05,
      "loss": 1.0621,
      "step": 11774
    },
    {
      "epoch": 2.188254971194945,
      "grad_norm": 0.7577797174453735,
      "learning_rate": 8.548593481748728e-05,
      "loss": 0.9997,
      "step": 11775
    },
    {
      "epoch": 2.1884408102583164,
      "grad_norm": 0.7481876015663147,
      "learning_rate": 8.547148032942627e-05,
      "loss": 0.6183,
      "step": 11776
    },
    {
      "epoch": 2.1886266493216873,
      "grad_norm": 0.8192367553710938,
      "learning_rate": 8.545702615145155e-05,
      "loss": 0.7456,
      "step": 11777
    },
    {
      "epoch": 2.1888124883850586,
      "grad_norm": 0.7634150385856628,
      "learning_rate": 8.544257228387167e-05,
      "loss": 0.8861,
      "step": 11778
    },
    {
      "epoch": 2.1889983274484295,
      "grad_norm": 0.9598495364189148,
      "learning_rate": 8.54281187269951e-05,
      "loss": 0.8434,
      "step": 11779
    },
    {
      "epoch": 2.189184166511801,
      "grad_norm": 0.8291441798210144,
      "learning_rate": 8.541366548113033e-05,
      "loss": 0.7626,
      "step": 11780
    },
    {
      "epoch": 2.1893700055751717,
      "grad_norm": 0.9502770304679871,
      "learning_rate": 8.539921254658579e-05,
      "loss": 0.9791,
      "step": 11781
    },
    {
      "epoch": 2.189555844638543,
      "grad_norm": 0.715624213218689,
      "learning_rate": 8.538475992367006e-05,
      "loss": 0.8788,
      "step": 11782
    },
    {
      "epoch": 2.1897416837019144,
      "grad_norm": 0.7813842296600342,
      "learning_rate": 8.537030761269151e-05,
      "loss": 1.1406,
      "step": 11783
    },
    {
      "epoch": 2.1899275227652852,
      "grad_norm": 0.7807679772377014,
      "learning_rate": 8.53558556139587e-05,
      "loss": 1.0152,
      "step": 11784
    },
    {
      "epoch": 2.1901133618286566,
      "grad_norm": 0.8997833728790283,
      "learning_rate": 8.534140392777996e-05,
      "loss": 1.0855,
      "step": 11785
    },
    {
      "epoch": 2.1902992008920275,
      "grad_norm": 0.7192310690879822,
      "learning_rate": 8.532695255446383e-05,
      "loss": 0.806,
      "step": 11786
    },
    {
      "epoch": 2.190485039955399,
      "grad_norm": 0.674193799495697,
      "learning_rate": 8.531250149431871e-05,
      "loss": 0.7599,
      "step": 11787
    },
    {
      "epoch": 2.1906708790187697,
      "grad_norm": 0.7934219241142273,
      "learning_rate": 8.529805074765306e-05,
      "loss": 0.8466,
      "step": 11788
    },
    {
      "epoch": 2.190856718082141,
      "grad_norm": 0.7016252875328064,
      "learning_rate": 8.528360031477526e-05,
      "loss": 0.6994,
      "step": 11789
    },
    {
      "epoch": 2.191042557145512,
      "grad_norm": 0.9743391275405884,
      "learning_rate": 8.526915019599377e-05,
      "loss": 1.0064,
      "step": 11790
    },
    {
      "epoch": 2.191228396208883,
      "grad_norm": 0.7312706708908081,
      "learning_rate": 8.525470039161698e-05,
      "loss": 0.7806,
      "step": 11791
    },
    {
      "epoch": 2.191414235272254,
      "grad_norm": 0.7978386878967285,
      "learning_rate": 8.524025090195328e-05,
      "loss": 0.8825,
      "step": 11792
    },
    {
      "epoch": 2.1916000743356254,
      "grad_norm": 0.7949557900428772,
      "learning_rate": 8.522580172731112e-05,
      "loss": 1.1221,
      "step": 11793
    },
    {
      "epoch": 2.1917859133989963,
      "grad_norm": 0.7279219031333923,
      "learning_rate": 8.52113528679989e-05,
      "loss": 0.8345,
      "step": 11794
    },
    {
      "epoch": 2.1919717524623676,
      "grad_norm": 0.8030551075935364,
      "learning_rate": 8.519690432432494e-05,
      "loss": 0.9106,
      "step": 11795
    },
    {
      "epoch": 2.1921575915257385,
      "grad_norm": 0.8001941442489624,
      "learning_rate": 8.518245609659768e-05,
      "loss": 0.8612,
      "step": 11796
    },
    {
      "epoch": 2.19234343058911,
      "grad_norm": 2.0981667041778564,
      "learning_rate": 8.516800818512545e-05,
      "loss": 1.374,
      "step": 11797
    },
    {
      "epoch": 2.192529269652481,
      "grad_norm": 0.8190371990203857,
      "learning_rate": 8.515356059021664e-05,
      "loss": 1.0411,
      "step": 11798
    },
    {
      "epoch": 2.192715108715852,
      "grad_norm": 0.86910080909729,
      "learning_rate": 8.513911331217958e-05,
      "loss": 1.0112,
      "step": 11799
    },
    {
      "epoch": 2.1929009477792234,
      "grad_norm": 0.9143921136856079,
      "learning_rate": 8.512466635132268e-05,
      "loss": 0.9417,
      "step": 11800
    },
    {
      "epoch": 2.1930867868425943,
      "grad_norm": 0.6698209047317505,
      "learning_rate": 8.511021970795423e-05,
      "loss": 0.8751,
      "step": 11801
    },
    {
      "epoch": 2.1932726259059656,
      "grad_norm": 0.8188903331756592,
      "learning_rate": 8.509577338238255e-05,
      "loss": 0.8438,
      "step": 11802
    },
    {
      "epoch": 2.1934584649693365,
      "grad_norm": 0.7005814909934998,
      "learning_rate": 8.508132737491604e-05,
      "loss": 0.8891,
      "step": 11803
    },
    {
      "epoch": 2.193644304032708,
      "grad_norm": 0.7396237850189209,
      "learning_rate": 8.506688168586302e-05,
      "loss": 0.9306,
      "step": 11804
    },
    {
      "epoch": 2.1938301430960787,
      "grad_norm": 0.9246975183486938,
      "learning_rate": 8.505243631553176e-05,
      "loss": 1.0293,
      "step": 11805
    },
    {
      "epoch": 2.19401598215945,
      "grad_norm": 0.7621631622314453,
      "learning_rate": 8.503799126423062e-05,
      "loss": 0.8078,
      "step": 11806
    },
    {
      "epoch": 2.194201821222821,
      "grad_norm": 0.7225520014762878,
      "learning_rate": 8.502354653226785e-05,
      "loss": 0.8706,
      "step": 11807
    },
    {
      "epoch": 2.1943876602861923,
      "grad_norm": 0.7985264658927917,
      "learning_rate": 8.500910211995182e-05,
      "loss": 0.9305,
      "step": 11808
    },
    {
      "epoch": 2.194573499349563,
      "grad_norm": 0.877153217792511,
      "learning_rate": 8.499465802759073e-05,
      "loss": 1.0499,
      "step": 11809
    },
    {
      "epoch": 2.1947593384129345,
      "grad_norm": 0.7561744451522827,
      "learning_rate": 8.498021425549296e-05,
      "loss": 0.8551,
      "step": 11810
    },
    {
      "epoch": 2.1949451774763054,
      "grad_norm": 0.6741968393325806,
      "learning_rate": 8.49657708039667e-05,
      "loss": 0.798,
      "step": 11811
    },
    {
      "epoch": 2.1951310165396767,
      "grad_norm": 0.8386529088020325,
      "learning_rate": 8.495132767332027e-05,
      "loss": 0.9944,
      "step": 11812
    },
    {
      "epoch": 2.1953168556030476,
      "grad_norm": 0.8982092142105103,
      "learning_rate": 8.493688486386193e-05,
      "loss": 1.129,
      "step": 11813
    },
    {
      "epoch": 2.195502694666419,
      "grad_norm": 1.0913642644882202,
      "learning_rate": 8.492244237589995e-05,
      "loss": 1.0555,
      "step": 11814
    },
    {
      "epoch": 2.19568853372979,
      "grad_norm": 0.7127829194068909,
      "learning_rate": 8.490800020974255e-05,
      "loss": 1.0229,
      "step": 11815
    },
    {
      "epoch": 2.195874372793161,
      "grad_norm": 0.8113505840301514,
      "learning_rate": 8.4893558365698e-05,
      "loss": 0.9756,
      "step": 11816
    },
    {
      "epoch": 2.1960602118565324,
      "grad_norm": 0.8548054695129395,
      "learning_rate": 8.487911684407451e-05,
      "loss": 0.799,
      "step": 11817
    },
    {
      "epoch": 2.1962460509199033,
      "grad_norm": 0.7963377833366394,
      "learning_rate": 8.486467564518034e-05,
      "loss": 0.7255,
      "step": 11818
    },
    {
      "epoch": 2.1964318899832747,
      "grad_norm": 1.93429696559906,
      "learning_rate": 8.485023476932367e-05,
      "loss": 1.2792,
      "step": 11819
    },
    {
      "epoch": 2.1966177290466455,
      "grad_norm": 0.8027183413505554,
      "learning_rate": 8.483579421681276e-05,
      "loss": 0.7442,
      "step": 11820
    },
    {
      "epoch": 2.196803568110017,
      "grad_norm": 0.8489840030670166,
      "learning_rate": 8.482135398795579e-05,
      "loss": 0.8223,
      "step": 11821
    },
    {
      "epoch": 2.1969894071733878,
      "grad_norm": 0.9878379106521606,
      "learning_rate": 8.480691408306097e-05,
      "loss": 0.8635,
      "step": 11822
    },
    {
      "epoch": 2.197175246236759,
      "grad_norm": 0.7123323678970337,
      "learning_rate": 8.479247450243646e-05,
      "loss": 0.9169,
      "step": 11823
    },
    {
      "epoch": 2.19736108530013,
      "grad_norm": 0.7710812091827393,
      "learning_rate": 8.477803524639054e-05,
      "loss": 0.896,
      "step": 11824
    },
    {
      "epoch": 2.1975469243635013,
      "grad_norm": 0.7759889364242554,
      "learning_rate": 8.476359631523133e-05,
      "loss": 0.9463,
      "step": 11825
    },
    {
      "epoch": 2.197732763426872,
      "grad_norm": 0.7710733413696289,
      "learning_rate": 8.474915770926699e-05,
      "loss": 0.7537,
      "step": 11826
    },
    {
      "epoch": 2.1979186024902435,
      "grad_norm": 1.081716775894165,
      "learning_rate": 8.473471942880572e-05,
      "loss": 0.786,
      "step": 11827
    },
    {
      "epoch": 2.1981044415536144,
      "grad_norm": 0.698901891708374,
      "learning_rate": 8.472028147415567e-05,
      "loss": 0.7743,
      "step": 11828
    },
    {
      "epoch": 2.1982902806169857,
      "grad_norm": 0.8639124631881714,
      "learning_rate": 8.470584384562499e-05,
      "loss": 0.786,
      "step": 11829
    },
    {
      "epoch": 2.1984761196803566,
      "grad_norm": 0.8688544631004333,
      "learning_rate": 8.469140654352183e-05,
      "loss": 0.9876,
      "step": 11830
    },
    {
      "epoch": 2.198661958743728,
      "grad_norm": 0.8796912431716919,
      "learning_rate": 8.467696956815432e-05,
      "loss": 0.6324,
      "step": 11831
    },
    {
      "epoch": 2.1988477978070993,
      "grad_norm": 0.7779879570007324,
      "learning_rate": 8.466253291983059e-05,
      "loss": 1.011,
      "step": 11832
    },
    {
      "epoch": 2.19903363687047,
      "grad_norm": 0.8026193976402283,
      "learning_rate": 8.46480965988588e-05,
      "loss": 0.7093,
      "step": 11833
    },
    {
      "epoch": 2.1992194759338415,
      "grad_norm": 1.2052098512649536,
      "learning_rate": 8.463366060554698e-05,
      "loss": 1.4435,
      "step": 11834
    },
    {
      "epoch": 2.1994053149972124,
      "grad_norm": 0.7946503758430481,
      "learning_rate": 8.461922494020336e-05,
      "loss": 0.8539,
      "step": 11835
    },
    {
      "epoch": 2.1995911540605837,
      "grad_norm": 0.721773087978363,
      "learning_rate": 8.460478960313599e-05,
      "loss": 0.6973,
      "step": 11836
    },
    {
      "epoch": 2.1997769931239546,
      "grad_norm": 0.7838326096534729,
      "learning_rate": 8.459035459465294e-05,
      "loss": 0.8626,
      "step": 11837
    },
    {
      "epoch": 2.199962832187326,
      "grad_norm": 0.6588775515556335,
      "learning_rate": 8.457591991506236e-05,
      "loss": 0.6709,
      "step": 11838
    },
    {
      "epoch": 2.200148671250697,
      "grad_norm": 1.0023447275161743,
      "learning_rate": 8.456148556467228e-05,
      "loss": 0.8973,
      "step": 11839
    },
    {
      "epoch": 2.200334510314068,
      "grad_norm": 0.9219651222229004,
      "learning_rate": 8.454705154379082e-05,
      "loss": 0.9682,
      "step": 11840
    },
    {
      "epoch": 2.200520349377439,
      "grad_norm": 1.0128252506256104,
      "learning_rate": 8.4532617852726e-05,
      "loss": 0.8887,
      "step": 11841
    },
    {
      "epoch": 2.2007061884408103,
      "grad_norm": 0.8471603989601135,
      "learning_rate": 8.451818449178591e-05,
      "loss": 0.8781,
      "step": 11842
    },
    {
      "epoch": 2.2008920275041812,
      "grad_norm": 0.6617082357406616,
      "learning_rate": 8.450375146127862e-05,
      "loss": 0.6619,
      "step": 11843
    },
    {
      "epoch": 2.2010778665675526,
      "grad_norm": 0.9384303092956543,
      "learning_rate": 8.448931876151212e-05,
      "loss": 1.0131,
      "step": 11844
    },
    {
      "epoch": 2.2012637056309234,
      "grad_norm": 0.716192901134491,
      "learning_rate": 8.447488639279452e-05,
      "loss": 0.702,
      "step": 11845
    },
    {
      "epoch": 2.2014495446942948,
      "grad_norm": 0.7688436508178711,
      "learning_rate": 8.446045435543387e-05,
      "loss": 0.8177,
      "step": 11846
    },
    {
      "epoch": 2.2016353837576657,
      "grad_norm": 0.7544126510620117,
      "learning_rate": 8.44460226497381e-05,
      "loss": 0.9157,
      "step": 11847
    },
    {
      "epoch": 2.201821222821037,
      "grad_norm": 0.7286648154258728,
      "learning_rate": 8.443159127601532e-05,
      "loss": 0.6314,
      "step": 11848
    },
    {
      "epoch": 2.202007061884408,
      "grad_norm": 0.8427985906600952,
      "learning_rate": 8.44171602345735e-05,
      "loss": 0.6621,
      "step": 11849
    },
    {
      "epoch": 2.202192900947779,
      "grad_norm": 0.7273639440536499,
      "learning_rate": 8.440272952572064e-05,
      "loss": 0.9898,
      "step": 11850
    },
    {
      "epoch": 2.2023787400111505,
      "grad_norm": 0.7268412709236145,
      "learning_rate": 8.438829914976477e-05,
      "loss": 0.8822,
      "step": 11851
    },
    {
      "epoch": 2.2025645790745214,
      "grad_norm": 0.6997602581977844,
      "learning_rate": 8.437386910701384e-05,
      "loss": 0.6818,
      "step": 11852
    },
    {
      "epoch": 2.2027504181378927,
      "grad_norm": 0.9023566842079163,
      "learning_rate": 8.435943939777588e-05,
      "loss": 0.6727,
      "step": 11853
    },
    {
      "epoch": 2.2029362572012636,
      "grad_norm": 0.72446209192276,
      "learning_rate": 8.434501002235883e-05,
      "loss": 0.8095,
      "step": 11854
    },
    {
      "epoch": 2.203122096264635,
      "grad_norm": 0.7783926725387573,
      "learning_rate": 8.433058098107065e-05,
      "loss": 0.8789,
      "step": 11855
    },
    {
      "epoch": 2.203307935328006,
      "grad_norm": 0.8605186939239502,
      "learning_rate": 8.431615227421937e-05,
      "loss": 0.9955,
      "step": 11856
    },
    {
      "epoch": 2.203493774391377,
      "grad_norm": 0.8575723171234131,
      "learning_rate": 8.43017239021129e-05,
      "loss": 0.7633,
      "step": 11857
    },
    {
      "epoch": 2.203679613454748,
      "grad_norm": 0.8678473830223083,
      "learning_rate": 8.42872958650592e-05,
      "loss": 0.9923,
      "step": 11858
    },
    {
      "epoch": 2.2038654525181194,
      "grad_norm": 0.9308450818061829,
      "learning_rate": 8.42728681633662e-05,
      "loss": 1.1472,
      "step": 11859
    },
    {
      "epoch": 2.2040512915814903,
      "grad_norm": 0.7468540668487549,
      "learning_rate": 8.425844079734184e-05,
      "loss": 0.7736,
      "step": 11860
    },
    {
      "epoch": 2.2042371306448616,
      "grad_norm": 0.7660921216011047,
      "learning_rate": 8.424401376729404e-05,
      "loss": 0.7215,
      "step": 11861
    },
    {
      "epoch": 2.2044229697082325,
      "grad_norm": 0.8048068284988403,
      "learning_rate": 8.422958707353072e-05,
      "loss": 0.8567,
      "step": 11862
    },
    {
      "epoch": 2.204608808771604,
      "grad_norm": 0.927282452583313,
      "learning_rate": 8.421516071635984e-05,
      "loss": 1.0192,
      "step": 11863
    },
    {
      "epoch": 2.2047946478349747,
      "grad_norm": 0.8508273959159851,
      "learning_rate": 8.420073469608922e-05,
      "loss": 1.0097,
      "step": 11864
    },
    {
      "epoch": 2.204980486898346,
      "grad_norm": 0.8396299481391907,
      "learning_rate": 8.418630901302679e-05,
      "loss": 0.8578,
      "step": 11865
    },
    {
      "epoch": 2.2051663259617174,
      "grad_norm": 0.9400133490562439,
      "learning_rate": 8.417188366748052e-05,
      "loss": 0.9306,
      "step": 11866
    },
    {
      "epoch": 2.2053521650250882,
      "grad_norm": 0.7686029672622681,
      "learning_rate": 8.415745865975819e-05,
      "loss": 0.8702,
      "step": 11867
    },
    {
      "epoch": 2.2055380040884596,
      "grad_norm": 0.9369570016860962,
      "learning_rate": 8.414303399016775e-05,
      "loss": 0.656,
      "step": 11868
    },
    {
      "epoch": 2.2057238431518305,
      "grad_norm": 0.7464059591293335,
      "learning_rate": 8.4128609659017e-05,
      "loss": 0.9271,
      "step": 11869
    },
    {
      "epoch": 2.205909682215202,
      "grad_norm": 0.8918207883834839,
      "learning_rate": 8.411418566661388e-05,
      "loss": 0.9687,
      "step": 11870
    },
    {
      "epoch": 2.2060955212785727,
      "grad_norm": 0.7785696983337402,
      "learning_rate": 8.409976201326618e-05,
      "loss": 0.9655,
      "step": 11871
    },
    {
      "epoch": 2.206281360341944,
      "grad_norm": 0.6803451776504517,
      "learning_rate": 8.408533869928177e-05,
      "loss": 0.6949,
      "step": 11872
    },
    {
      "epoch": 2.206467199405315,
      "grad_norm": 0.7219688296318054,
      "learning_rate": 8.407091572496851e-05,
      "loss": 0.9155,
      "step": 11873
    },
    {
      "epoch": 2.206653038468686,
      "grad_norm": 0.7794553637504578,
      "learning_rate": 8.40564930906342e-05,
      "loss": 1.0356,
      "step": 11874
    },
    {
      "epoch": 2.206838877532057,
      "grad_norm": 1.1876341104507446,
      "learning_rate": 8.404207079658671e-05,
      "loss": 1.0767,
      "step": 11875
    },
    {
      "epoch": 2.2070247165954284,
      "grad_norm": 0.7047664523124695,
      "learning_rate": 8.40276488431338e-05,
      "loss": 0.8224,
      "step": 11876
    },
    {
      "epoch": 2.2072105556587993,
      "grad_norm": 1.0113282203674316,
      "learning_rate": 8.401322723058332e-05,
      "loss": 0.8664,
      "step": 11877
    },
    {
      "epoch": 2.2073963947221706,
      "grad_norm": 0.7891967296600342,
      "learning_rate": 8.399880595924312e-05,
      "loss": 0.7399,
      "step": 11878
    },
    {
      "epoch": 2.2075822337855415,
      "grad_norm": 0.7502462267875671,
      "learning_rate": 8.39843850294209e-05,
      "loss": 0.9752,
      "step": 11879
    },
    {
      "epoch": 2.207768072848913,
      "grad_norm": 0.700694739818573,
      "learning_rate": 8.396996444142452e-05,
      "loss": 0.6745,
      "step": 11880
    },
    {
      "epoch": 2.207953911912284,
      "grad_norm": 0.8352717757225037,
      "learning_rate": 8.395554419556173e-05,
      "loss": 0.9155,
      "step": 11881
    },
    {
      "epoch": 2.208139750975655,
      "grad_norm": 0.8278815746307373,
      "learning_rate": 8.39411242921403e-05,
      "loss": 0.9406,
      "step": 11882
    },
    {
      "epoch": 2.2083255900390264,
      "grad_norm": 0.7651520371437073,
      "learning_rate": 8.392670473146806e-05,
      "loss": 0.9727,
      "step": 11883
    },
    {
      "epoch": 2.2085114291023973,
      "grad_norm": 0.8754087686538696,
      "learning_rate": 8.391228551385268e-05,
      "loss": 0.9176,
      "step": 11884
    },
    {
      "epoch": 2.2086972681657686,
      "grad_norm": 0.8060709238052368,
      "learning_rate": 8.3897866639602e-05,
      "loss": 0.8943,
      "step": 11885
    },
    {
      "epoch": 2.2088831072291395,
      "grad_norm": 0.9044803977012634,
      "learning_rate": 8.388344810902367e-05,
      "loss": 0.8597,
      "step": 11886
    },
    {
      "epoch": 2.209068946292511,
      "grad_norm": 0.6913041472434998,
      "learning_rate": 8.386902992242549e-05,
      "loss": 0.9207,
      "step": 11887
    },
    {
      "epoch": 2.2092547853558817,
      "grad_norm": 1.0214216709136963,
      "learning_rate": 8.385461208011523e-05,
      "loss": 1.1811,
      "step": 11888
    },
    {
      "epoch": 2.209440624419253,
      "grad_norm": 0.8380892872810364,
      "learning_rate": 8.384019458240054e-05,
      "loss": 0.9377,
      "step": 11889
    },
    {
      "epoch": 2.209626463482624,
      "grad_norm": 0.898802638053894,
      "learning_rate": 8.382577742958917e-05,
      "loss": 0.7937,
      "step": 11890
    },
    {
      "epoch": 2.2098123025459953,
      "grad_norm": 0.8506227731704712,
      "learning_rate": 8.381136062198881e-05,
      "loss": 1.0291,
      "step": 11891
    },
    {
      "epoch": 2.209998141609366,
      "grad_norm": 0.7427432537078857,
      "learning_rate": 8.37969441599072e-05,
      "loss": 0.6678,
      "step": 11892
    },
    {
      "epoch": 2.2101839806727375,
      "grad_norm": 0.7741548418998718,
      "learning_rate": 8.378252804365202e-05,
      "loss": 0.9471,
      "step": 11893
    },
    {
      "epoch": 2.2103698197361084,
      "grad_norm": 0.7084181904792786,
      "learning_rate": 8.376811227353091e-05,
      "loss": 0.9541,
      "step": 11894
    },
    {
      "epoch": 2.2105556587994797,
      "grad_norm": 0.7543783187866211,
      "learning_rate": 8.375369684985164e-05,
      "loss": 0.8969,
      "step": 11895
    },
    {
      "epoch": 2.2107414978628506,
      "grad_norm": 0.9766399264335632,
      "learning_rate": 8.373928177292179e-05,
      "loss": 0.8219,
      "step": 11896
    },
    {
      "epoch": 2.210927336926222,
      "grad_norm": 1.0114517211914062,
      "learning_rate": 8.372486704304905e-05,
      "loss": 1.0136,
      "step": 11897
    },
    {
      "epoch": 2.211113175989593,
      "grad_norm": 0.6716600656509399,
      "learning_rate": 8.371045266054114e-05,
      "loss": 0.6813,
      "step": 11898
    },
    {
      "epoch": 2.211299015052964,
      "grad_norm": 0.7623451352119446,
      "learning_rate": 8.369603862570563e-05,
      "loss": 0.9137,
      "step": 11899
    },
    {
      "epoch": 2.2114848541163354,
      "grad_norm": 0.8574510812759399,
      "learning_rate": 8.368162493885022e-05,
      "loss": 0.7934,
      "step": 11900
    },
    {
      "epoch": 2.2116706931797063,
      "grad_norm": 0.8110435605049133,
      "learning_rate": 8.366721160028251e-05,
      "loss": 0.7885,
      "step": 11901
    },
    {
      "epoch": 2.2118565322430777,
      "grad_norm": 0.8771747946739197,
      "learning_rate": 8.365279861031014e-05,
      "loss": 0.9261,
      "step": 11902
    },
    {
      "epoch": 2.2120423713064485,
      "grad_norm": 0.9254234433174133,
      "learning_rate": 8.363838596924075e-05,
      "loss": 1.0125,
      "step": 11903
    },
    {
      "epoch": 2.21222821036982,
      "grad_norm": 0.7689984440803528,
      "learning_rate": 8.362397367738191e-05,
      "loss": 1.0572,
      "step": 11904
    },
    {
      "epoch": 2.2124140494331908,
      "grad_norm": 0.7987253665924072,
      "learning_rate": 8.360956173504127e-05,
      "loss": 0.9679,
      "step": 11905
    },
    {
      "epoch": 2.212599888496562,
      "grad_norm": 0.8854193687438965,
      "learning_rate": 8.359515014252638e-05,
      "loss": 0.9194,
      "step": 11906
    },
    {
      "epoch": 2.212785727559933,
      "grad_norm": 0.7638492584228516,
      "learning_rate": 8.358073890014486e-05,
      "loss": 0.9595,
      "step": 11907
    },
    {
      "epoch": 2.2129715666233043,
      "grad_norm": 0.866855263710022,
      "learning_rate": 8.356632800820432e-05,
      "loss": 0.9976,
      "step": 11908
    },
    {
      "epoch": 2.213157405686675,
      "grad_norm": 0.9702017307281494,
      "learning_rate": 8.355191746701227e-05,
      "loss": 0.9449,
      "step": 11909
    },
    {
      "epoch": 2.2133432447500465,
      "grad_norm": 0.7269165515899658,
      "learning_rate": 8.353750727687637e-05,
      "loss": 0.7667,
      "step": 11910
    },
    {
      "epoch": 2.2135290838134174,
      "grad_norm": 0.781199038028717,
      "learning_rate": 8.352309743810409e-05,
      "loss": 0.8012,
      "step": 11911
    },
    {
      "epoch": 2.2137149228767887,
      "grad_norm": 0.8521917462348938,
      "learning_rate": 8.350868795100301e-05,
      "loss": 0.9337,
      "step": 11912
    },
    {
      "epoch": 2.2139007619401596,
      "grad_norm": 0.9267933368682861,
      "learning_rate": 8.349427881588071e-05,
      "loss": 0.8166,
      "step": 11913
    },
    {
      "epoch": 2.214086601003531,
      "grad_norm": 0.865480363368988,
      "learning_rate": 8.347987003304469e-05,
      "loss": 1.0724,
      "step": 11914
    },
    {
      "epoch": 2.2142724400669023,
      "grad_norm": 0.6637668609619141,
      "learning_rate": 8.34654616028025e-05,
      "loss": 0.7964,
      "step": 11915
    },
    {
      "epoch": 2.214458279130273,
      "grad_norm": 0.838745653629303,
      "learning_rate": 8.345105352546166e-05,
      "loss": 0.9489,
      "step": 11916
    },
    {
      "epoch": 2.2146441181936445,
      "grad_norm": 0.9586402177810669,
      "learning_rate": 8.34366458013297e-05,
      "loss": 0.8424,
      "step": 11917
    },
    {
      "epoch": 2.2148299572570154,
      "grad_norm": 0.7502764463424683,
      "learning_rate": 8.342223843071406e-05,
      "loss": 1.0838,
      "step": 11918
    },
    {
      "epoch": 2.2150157963203867,
      "grad_norm": 0.6431674957275391,
      "learning_rate": 8.340783141392231e-05,
      "loss": 0.6327,
      "step": 11919
    },
    {
      "epoch": 2.2152016353837576,
      "grad_norm": 0.7192468047142029,
      "learning_rate": 8.339342475126195e-05,
      "loss": 0.8298,
      "step": 11920
    },
    {
      "epoch": 2.215387474447129,
      "grad_norm": 0.7980626821517944,
      "learning_rate": 8.337901844304042e-05,
      "loss": 1.006,
      "step": 11921
    },
    {
      "epoch": 2.2155733135105,
      "grad_norm": 0.7548335194587708,
      "learning_rate": 8.336461248956522e-05,
      "loss": 0.6331,
      "step": 11922
    },
    {
      "epoch": 2.215759152573871,
      "grad_norm": 1.0166386365890503,
      "learning_rate": 8.335020689114384e-05,
      "loss": 0.9778,
      "step": 11923
    },
    {
      "epoch": 2.215944991637242,
      "grad_norm": 0.7799232602119446,
      "learning_rate": 8.333580164808372e-05,
      "loss": 0.7998,
      "step": 11924
    },
    {
      "epoch": 2.2161308307006133,
      "grad_norm": 0.838235080242157,
      "learning_rate": 8.33213967606923e-05,
      "loss": 0.865,
      "step": 11925
    },
    {
      "epoch": 2.2163166697639842,
      "grad_norm": 0.9255481958389282,
      "learning_rate": 8.330699222927705e-05,
      "loss": 1.018,
      "step": 11926
    },
    {
      "epoch": 2.2165025088273556,
      "grad_norm": 0.8905596733093262,
      "learning_rate": 8.329258805414542e-05,
      "loss": 1.0686,
      "step": 11927
    },
    {
      "epoch": 2.2166883478907264,
      "grad_norm": 0.836381733417511,
      "learning_rate": 8.327818423560478e-05,
      "loss": 1.023,
      "step": 11928
    },
    {
      "epoch": 2.2168741869540978,
      "grad_norm": 1.0391117334365845,
      "learning_rate": 8.326378077396262e-05,
      "loss": 1.118,
      "step": 11929
    },
    {
      "epoch": 2.217060026017469,
      "grad_norm": 0.7728176712989807,
      "learning_rate": 8.324937766952638e-05,
      "loss": 0.9211,
      "step": 11930
    },
    {
      "epoch": 2.21724586508084,
      "grad_norm": 0.9985128045082092,
      "learning_rate": 8.323497492260338e-05,
      "loss": 1.2685,
      "step": 11931
    },
    {
      "epoch": 2.2174317041442113,
      "grad_norm": 0.7964622378349304,
      "learning_rate": 8.322057253350108e-05,
      "loss": 0.9126,
      "step": 11932
    },
    {
      "epoch": 2.217617543207582,
      "grad_norm": 0.8635339736938477,
      "learning_rate": 8.32061705025269e-05,
      "loss": 0.9854,
      "step": 11933
    },
    {
      "epoch": 2.2178033822709535,
      "grad_norm": 0.8185827136039734,
      "learning_rate": 8.319176882998816e-05,
      "loss": 1.0503,
      "step": 11934
    },
    {
      "epoch": 2.2179892213343244,
      "grad_norm": 0.8222116827964783,
      "learning_rate": 8.317736751619229e-05,
      "loss": 0.948,
      "step": 11935
    },
    {
      "epoch": 2.2181750603976957,
      "grad_norm": 0.7733019590377808,
      "learning_rate": 8.316296656144662e-05,
      "loss": 1.0669,
      "step": 11936
    },
    {
      "epoch": 2.2183608994610666,
      "grad_norm": 0.7140665650367737,
      "learning_rate": 8.314856596605856e-05,
      "loss": 0.8587,
      "step": 11937
    },
    {
      "epoch": 2.218546738524438,
      "grad_norm": 0.7716907858848572,
      "learning_rate": 8.313416573033541e-05,
      "loss": 0.9409,
      "step": 11938
    },
    {
      "epoch": 2.218732577587809,
      "grad_norm": 0.9706321358680725,
      "learning_rate": 8.311976585458453e-05,
      "loss": 0.8897,
      "step": 11939
    },
    {
      "epoch": 2.21891841665118,
      "grad_norm": 0.6067124009132385,
      "learning_rate": 8.310536633911334e-05,
      "loss": 0.6413,
      "step": 11940
    },
    {
      "epoch": 2.219104255714551,
      "grad_norm": 0.7392117977142334,
      "learning_rate": 8.309096718422907e-05,
      "loss": 0.6049,
      "step": 11941
    },
    {
      "epoch": 2.2192900947779224,
      "grad_norm": 0.7874842286109924,
      "learning_rate": 8.307656839023909e-05,
      "loss": 1.0683,
      "step": 11942
    },
    {
      "epoch": 2.2194759338412933,
      "grad_norm": 0.8187624216079712,
      "learning_rate": 8.306216995745075e-05,
      "loss": 0.8404,
      "step": 11943
    },
    {
      "epoch": 2.2196617729046646,
      "grad_norm": 1.0005868673324585,
      "learning_rate": 8.30477718861713e-05,
      "loss": 0.9145,
      "step": 11944
    },
    {
      "epoch": 2.2198476119680355,
      "grad_norm": 0.7779518961906433,
      "learning_rate": 8.303337417670811e-05,
      "loss": 0.8756,
      "step": 11945
    },
    {
      "epoch": 2.220033451031407,
      "grad_norm": 0.9289587140083313,
      "learning_rate": 8.301897682936838e-05,
      "loss": 0.9345,
      "step": 11946
    },
    {
      "epoch": 2.2202192900947777,
      "grad_norm": 0.8238828182220459,
      "learning_rate": 8.300457984445948e-05,
      "loss": 0.8322,
      "step": 11947
    },
    {
      "epoch": 2.220405129158149,
      "grad_norm": 0.911462664604187,
      "learning_rate": 8.299018322228864e-05,
      "loss": 0.9052,
      "step": 11948
    },
    {
      "epoch": 2.2205909682215204,
      "grad_norm": 0.8210474848747253,
      "learning_rate": 8.297578696316312e-05,
      "loss": 0.9744,
      "step": 11949
    },
    {
      "epoch": 2.2207768072848912,
      "grad_norm": 0.7710187435150146,
      "learning_rate": 8.296139106739026e-05,
      "loss": 1.0271,
      "step": 11950
    },
    {
      "epoch": 2.2209626463482626,
      "grad_norm": 0.8088672757148743,
      "learning_rate": 8.294699553527725e-05,
      "loss": 1.0682,
      "step": 11951
    },
    {
      "epoch": 2.2211484854116335,
      "grad_norm": 0.7803736925125122,
      "learning_rate": 8.293260036713135e-05,
      "loss": 0.8144,
      "step": 11952
    },
    {
      "epoch": 2.221334324475005,
      "grad_norm": 0.6293572187423706,
      "learning_rate": 8.291820556325984e-05,
      "loss": 0.8965,
      "step": 11953
    },
    {
      "epoch": 2.2215201635383757,
      "grad_norm": 0.6929018497467041,
      "learning_rate": 8.290381112396987e-05,
      "loss": 0.71,
      "step": 11954
    },
    {
      "epoch": 2.221706002601747,
      "grad_norm": 1.1523988246917725,
      "learning_rate": 8.288941704956876e-05,
      "loss": 0.9437,
      "step": 11955
    },
    {
      "epoch": 2.221891841665118,
      "grad_norm": 0.7967273592948914,
      "learning_rate": 8.287502334036365e-05,
      "loss": 1.0549,
      "step": 11956
    },
    {
      "epoch": 2.222077680728489,
      "grad_norm": 0.8078708052635193,
      "learning_rate": 8.286062999666182e-05,
      "loss": 0.8772,
      "step": 11957
    },
    {
      "epoch": 2.22226351979186,
      "grad_norm": 0.726443350315094,
      "learning_rate": 8.284623701877037e-05,
      "loss": 0.9252,
      "step": 11958
    },
    {
      "epoch": 2.2224493588552314,
      "grad_norm": 0.8753724694252014,
      "learning_rate": 8.28318444069966e-05,
      "loss": 0.8713,
      "step": 11959
    },
    {
      "epoch": 2.2226351979186023,
      "grad_norm": 0.5907177925109863,
      "learning_rate": 8.28174521616476e-05,
      "loss": 0.6124,
      "step": 11960
    },
    {
      "epoch": 2.2228210369819736,
      "grad_norm": 0.9451390504837036,
      "learning_rate": 8.280306028303063e-05,
      "loss": 1.117,
      "step": 11961
    },
    {
      "epoch": 2.2230068760453445,
      "grad_norm": 0.8585532903671265,
      "learning_rate": 8.278866877145282e-05,
      "loss": 1.0065,
      "step": 11962
    },
    {
      "epoch": 2.223192715108716,
      "grad_norm": 0.843278706073761,
      "learning_rate": 8.277427762722136e-05,
      "loss": 1.2527,
      "step": 11963
    },
    {
      "epoch": 2.223378554172087,
      "grad_norm": 0.831453800201416,
      "learning_rate": 8.275988685064335e-05,
      "loss": 0.9158,
      "step": 11964
    },
    {
      "epoch": 2.223564393235458,
      "grad_norm": 0.757862389087677,
      "learning_rate": 8.274549644202602e-05,
      "loss": 0.7903,
      "step": 11965
    },
    {
      "epoch": 2.2237502322988294,
      "grad_norm": 0.7283931970596313,
      "learning_rate": 8.273110640167641e-05,
      "loss": 0.9407,
      "step": 11966
    },
    {
      "epoch": 2.2239360713622003,
      "grad_norm": 0.6937702894210815,
      "learning_rate": 8.271671672990173e-05,
      "loss": 0.8566,
      "step": 11967
    },
    {
      "epoch": 2.2241219104255716,
      "grad_norm": 0.9456591606140137,
      "learning_rate": 8.270232742700906e-05,
      "loss": 1.024,
      "step": 11968
    },
    {
      "epoch": 2.2243077494889425,
      "grad_norm": 0.9929325580596924,
      "learning_rate": 8.268793849330553e-05,
      "loss": 0.7111,
      "step": 11969
    },
    {
      "epoch": 2.224493588552314,
      "grad_norm": 1.037474274635315,
      "learning_rate": 8.267354992909822e-05,
      "loss": 0.7245,
      "step": 11970
    },
    {
      "epoch": 2.2246794276156847,
      "grad_norm": 0.7578388452529907,
      "learning_rate": 8.265916173469426e-05,
      "loss": 0.9581,
      "step": 11971
    },
    {
      "epoch": 2.224865266679056,
      "grad_norm": 0.9178577661514282,
      "learning_rate": 8.264477391040075e-05,
      "loss": 0.9945,
      "step": 11972
    },
    {
      "epoch": 2.225051105742427,
      "grad_norm": 1.0197036266326904,
      "learning_rate": 8.263038645652477e-05,
      "loss": 0.8752,
      "step": 11973
    },
    {
      "epoch": 2.2252369448057983,
      "grad_norm": 0.8248927593231201,
      "learning_rate": 8.261599937337336e-05,
      "loss": 0.8712,
      "step": 11974
    },
    {
      "epoch": 2.225422783869169,
      "grad_norm": 0.7531763315200806,
      "learning_rate": 8.260161266125363e-05,
      "loss": 0.8023,
      "step": 11975
    },
    {
      "epoch": 2.2256086229325405,
      "grad_norm": 1.0819270610809326,
      "learning_rate": 8.25872263204726e-05,
      "loss": 1.0796,
      "step": 11976
    },
    {
      "epoch": 2.2257944619959114,
      "grad_norm": 2.0839169025421143,
      "learning_rate": 8.257284035133738e-05,
      "loss": 1.4436,
      "step": 11977
    },
    {
      "epoch": 2.2259803010592827,
      "grad_norm": 0.856492817401886,
      "learning_rate": 8.255845475415494e-05,
      "loss": 0.9693,
      "step": 11978
    },
    {
      "epoch": 2.226166140122654,
      "grad_norm": 0.6808337569236755,
      "learning_rate": 8.254406952923238e-05,
      "loss": 0.698,
      "step": 11979
    },
    {
      "epoch": 2.226351979186025,
      "grad_norm": 0.8147267699241638,
      "learning_rate": 8.252968467687667e-05,
      "loss": 0.9628,
      "step": 11980
    },
    {
      "epoch": 2.2265378182493962,
      "grad_norm": 0.9136236310005188,
      "learning_rate": 8.251530019739485e-05,
      "loss": 0.6957,
      "step": 11981
    },
    {
      "epoch": 2.226723657312767,
      "grad_norm": 0.802736222743988,
      "learning_rate": 8.250091609109396e-05,
      "loss": 0.9213,
      "step": 11982
    },
    {
      "epoch": 2.2269094963761384,
      "grad_norm": 1.0352932214736938,
      "learning_rate": 8.248653235828102e-05,
      "loss": 1.1422,
      "step": 11983
    },
    {
      "epoch": 2.2270953354395093,
      "grad_norm": 0.8228843808174133,
      "learning_rate": 8.247214899926294e-05,
      "loss": 0.8079,
      "step": 11984
    },
    {
      "epoch": 2.2272811745028807,
      "grad_norm": 0.7414042949676514,
      "learning_rate": 8.24577660143468e-05,
      "loss": 0.9815,
      "step": 11985
    },
    {
      "epoch": 2.2274670135662515,
      "grad_norm": 0.8866422772407532,
      "learning_rate": 8.244338340383953e-05,
      "loss": 0.9617,
      "step": 11986
    },
    {
      "epoch": 2.227652852629623,
      "grad_norm": 0.894111156463623,
      "learning_rate": 8.242900116804811e-05,
      "loss": 1.0187,
      "step": 11987
    },
    {
      "epoch": 2.2278386916929938,
      "grad_norm": 0.8763172626495361,
      "learning_rate": 8.24146193072795e-05,
      "loss": 0.7959,
      "step": 11988
    },
    {
      "epoch": 2.228024530756365,
      "grad_norm": 0.8017054796218872,
      "learning_rate": 8.240023782184068e-05,
      "loss": 0.8997,
      "step": 11989
    },
    {
      "epoch": 2.228210369819736,
      "grad_norm": 0.7158930897712708,
      "learning_rate": 8.238585671203858e-05,
      "loss": 0.6309,
      "step": 11990
    },
    {
      "epoch": 2.2283962088831073,
      "grad_norm": 0.7901464104652405,
      "learning_rate": 8.237147597818009e-05,
      "loss": 0.7189,
      "step": 11991
    },
    {
      "epoch": 2.228582047946478,
      "grad_norm": 0.9525620937347412,
      "learning_rate": 8.235709562057224e-05,
      "loss": 0.7027,
      "step": 11992
    },
    {
      "epoch": 2.2287678870098495,
      "grad_norm": 0.9726958274841309,
      "learning_rate": 8.234271563952192e-05,
      "loss": 0.9302,
      "step": 11993
    },
    {
      "epoch": 2.2289537260732204,
      "grad_norm": 0.7856528759002686,
      "learning_rate": 8.2328336035336e-05,
      "loss": 0.8738,
      "step": 11994
    },
    {
      "epoch": 2.2291395651365917,
      "grad_norm": 0.8742666840553284,
      "learning_rate": 8.231395680832145e-05,
      "loss": 0.8732,
      "step": 11995
    },
    {
      "epoch": 2.2293254041999626,
      "grad_norm": 0.7303487658500671,
      "learning_rate": 8.229957795878514e-05,
      "loss": 0.623,
      "step": 11996
    },
    {
      "epoch": 2.229511243263334,
      "grad_norm": 0.8090687990188599,
      "learning_rate": 8.228519948703396e-05,
      "loss": 0.8562,
      "step": 11997
    },
    {
      "epoch": 2.2296970823267053,
      "grad_norm": 0.7455593943595886,
      "learning_rate": 8.227082139337478e-05,
      "loss": 0.9119,
      "step": 11998
    },
    {
      "epoch": 2.229882921390076,
      "grad_norm": 0.7629585266113281,
      "learning_rate": 8.225644367811452e-05,
      "loss": 0.9862,
      "step": 11999
    },
    {
      "epoch": 2.2300687604534475,
      "grad_norm": 1.1065359115600586,
      "learning_rate": 8.224206634156e-05,
      "loss": 0.9766,
      "step": 12000
    },
    {
      "epoch": 2.2302545995168184,
      "grad_norm": 0.7593920826911926,
      "learning_rate": 8.222768938401812e-05,
      "loss": 0.9066,
      "step": 12001
    },
    {
      "epoch": 2.2304404385801897,
      "grad_norm": 0.9303902983665466,
      "learning_rate": 8.221331280579564e-05,
      "loss": 0.842,
      "step": 12002
    },
    {
      "epoch": 2.2306262776435606,
      "grad_norm": 0.811173141002655,
      "learning_rate": 8.219893660719954e-05,
      "loss": 0.7842,
      "step": 12003
    },
    {
      "epoch": 2.230812116706932,
      "grad_norm": 0.7823946475982666,
      "learning_rate": 8.218456078853658e-05,
      "loss": 0.7958,
      "step": 12004
    },
    {
      "epoch": 2.230997955770303,
      "grad_norm": 0.9735191464424133,
      "learning_rate": 8.21701853501136e-05,
      "loss": 0.8981,
      "step": 12005
    },
    {
      "epoch": 2.231183794833674,
      "grad_norm": 0.7602159380912781,
      "learning_rate": 8.215581029223742e-05,
      "loss": 0.8651,
      "step": 12006
    },
    {
      "epoch": 2.231369633897045,
      "grad_norm": 0.7160788774490356,
      "learning_rate": 8.214143561521485e-05,
      "loss": 0.9801,
      "step": 12007
    },
    {
      "epoch": 2.2315554729604163,
      "grad_norm": 0.8263550400733948,
      "learning_rate": 8.212706131935268e-05,
      "loss": 0.9681,
      "step": 12008
    },
    {
      "epoch": 2.2317413120237872,
      "grad_norm": 1.2644280195236206,
      "learning_rate": 8.211268740495773e-05,
      "loss": 1.2091,
      "step": 12009
    },
    {
      "epoch": 2.2319271510871586,
      "grad_norm": 0.983466386795044,
      "learning_rate": 8.209831387233676e-05,
      "loss": 0.8649,
      "step": 12010
    },
    {
      "epoch": 2.2321129901505294,
      "grad_norm": 0.8372595906257629,
      "learning_rate": 8.208394072179657e-05,
      "loss": 0.7551,
      "step": 12011
    },
    {
      "epoch": 2.2322988292139008,
      "grad_norm": 0.815680205821991,
      "learning_rate": 8.206956795364386e-05,
      "loss": 1.0987,
      "step": 12012
    },
    {
      "epoch": 2.232484668277272,
      "grad_norm": 0.9907569885253906,
      "learning_rate": 8.20551955681855e-05,
      "loss": 0.9979,
      "step": 12013
    },
    {
      "epoch": 2.232670507340643,
      "grad_norm": 0.8694087862968445,
      "learning_rate": 8.20408235657282e-05,
      "loss": 1.0583,
      "step": 12014
    },
    {
      "epoch": 2.2328563464040143,
      "grad_norm": 0.9345015287399292,
      "learning_rate": 8.20264519465787e-05,
      "loss": 0.9702,
      "step": 12015
    },
    {
      "epoch": 2.233042185467385,
      "grad_norm": 0.7867563962936401,
      "learning_rate": 8.201208071104374e-05,
      "loss": 0.7381,
      "step": 12016
    },
    {
      "epoch": 2.2332280245307565,
      "grad_norm": 0.9041784405708313,
      "learning_rate": 8.199770985943006e-05,
      "loss": 1.0081,
      "step": 12017
    },
    {
      "epoch": 2.2334138635941274,
      "grad_norm": 0.842856764793396,
      "learning_rate": 8.198333939204435e-05,
      "loss": 1.0738,
      "step": 12018
    },
    {
      "epoch": 2.2335997026574987,
      "grad_norm": 1.0508366823196411,
      "learning_rate": 8.196896930919336e-05,
      "loss": 0.8267,
      "step": 12019
    },
    {
      "epoch": 2.2337855417208696,
      "grad_norm": 0.8856883645057678,
      "learning_rate": 8.195459961118377e-05,
      "loss": 0.9387,
      "step": 12020
    },
    {
      "epoch": 2.233971380784241,
      "grad_norm": 0.874148428440094,
      "learning_rate": 8.19402302983223e-05,
      "loss": 1.0397,
      "step": 12021
    },
    {
      "epoch": 2.234157219847612,
      "grad_norm": 0.9051022529602051,
      "learning_rate": 8.192586137091559e-05,
      "loss": 0.8769,
      "step": 12022
    },
    {
      "epoch": 2.234343058910983,
      "grad_norm": 0.9347072839736938,
      "learning_rate": 8.191149282927034e-05,
      "loss": 0.9741,
      "step": 12023
    },
    {
      "epoch": 2.234528897974354,
      "grad_norm": 0.9115281701087952,
      "learning_rate": 8.189712467369325e-05,
      "loss": 1.0689,
      "step": 12024
    },
    {
      "epoch": 2.2347147370377254,
      "grad_norm": 0.8097609877586365,
      "learning_rate": 8.1882756904491e-05,
      "loss": 0.8385,
      "step": 12025
    },
    {
      "epoch": 2.2349005761010963,
      "grad_norm": 0.8436894416809082,
      "learning_rate": 8.186838952197018e-05,
      "loss": 0.9244,
      "step": 12026
    },
    {
      "epoch": 2.2350864151644676,
      "grad_norm": 0.788254976272583,
      "learning_rate": 8.18540225264375e-05,
      "loss": 0.9671,
      "step": 12027
    },
    {
      "epoch": 2.2352722542278385,
      "grad_norm": 0.885251522064209,
      "learning_rate": 8.183965591819954e-05,
      "loss": 0.8399,
      "step": 12028
    },
    {
      "epoch": 2.23545809329121,
      "grad_norm": 0.7178447842597961,
      "learning_rate": 8.182528969756299e-05,
      "loss": 0.8732,
      "step": 12029
    },
    {
      "epoch": 2.2356439323545807,
      "grad_norm": 0.8282201290130615,
      "learning_rate": 8.181092386483442e-05,
      "loss": 0.7439,
      "step": 12030
    },
    {
      "epoch": 2.235829771417952,
      "grad_norm": 0.6753260493278503,
      "learning_rate": 8.179655842032048e-05,
      "loss": 0.745,
      "step": 12031
    },
    {
      "epoch": 2.2360156104813234,
      "grad_norm": 0.7672291994094849,
      "learning_rate": 8.178219336432775e-05,
      "loss": 1.0459,
      "step": 12032
    },
    {
      "epoch": 2.2362014495446942,
      "grad_norm": 0.7237166166305542,
      "learning_rate": 8.176782869716282e-05,
      "loss": 0.8665,
      "step": 12033
    },
    {
      "epoch": 2.2363872886080656,
      "grad_norm": 0.8562415838241577,
      "learning_rate": 8.17534644191323e-05,
      "loss": 1.0771,
      "step": 12034
    },
    {
      "epoch": 2.2365731276714365,
      "grad_norm": 0.869492769241333,
      "learning_rate": 8.17391005305428e-05,
      "loss": 1.0328,
      "step": 12035
    },
    {
      "epoch": 2.236758966734808,
      "grad_norm": 0.7791122794151306,
      "learning_rate": 8.172473703170085e-05,
      "loss": 0.9982,
      "step": 12036
    },
    {
      "epoch": 2.2369448057981787,
      "grad_norm": 0.947382926940918,
      "learning_rate": 8.171037392291305e-05,
      "loss": 1.0431,
      "step": 12037
    },
    {
      "epoch": 2.23713064486155,
      "grad_norm": 0.7841407656669617,
      "learning_rate": 8.16960112044859e-05,
      "loss": 0.8743,
      "step": 12038
    },
    {
      "epoch": 2.237316483924921,
      "grad_norm": 0.8942570090293884,
      "learning_rate": 8.168164887672602e-05,
      "loss": 1.0514,
      "step": 12039
    },
    {
      "epoch": 2.237502322988292,
      "grad_norm": 0.8452247977256775,
      "learning_rate": 8.166728693993987e-05,
      "loss": 0.7703,
      "step": 12040
    },
    {
      "epoch": 2.237688162051663,
      "grad_norm": 0.9405089616775513,
      "learning_rate": 8.165292539443405e-05,
      "loss": 1.0186,
      "step": 12041
    },
    {
      "epoch": 2.2378740011150344,
      "grad_norm": 0.8000966310501099,
      "learning_rate": 8.163856424051502e-05,
      "loss": 0.9675,
      "step": 12042
    },
    {
      "epoch": 2.2380598401784053,
      "grad_norm": 0.7403755784034729,
      "learning_rate": 8.162420347848934e-05,
      "loss": 0.9268,
      "step": 12043
    },
    {
      "epoch": 2.2382456792417766,
      "grad_norm": 0.81078040599823,
      "learning_rate": 8.160984310866348e-05,
      "loss": 0.8997,
      "step": 12044
    },
    {
      "epoch": 2.2384315183051475,
      "grad_norm": 0.8180494904518127,
      "learning_rate": 8.1595483131344e-05,
      "loss": 0.9303,
      "step": 12045
    },
    {
      "epoch": 2.238617357368519,
      "grad_norm": 0.8312895894050598,
      "learning_rate": 8.158112354683732e-05,
      "loss": 0.9163,
      "step": 12046
    },
    {
      "epoch": 2.23880319643189,
      "grad_norm": 0.916143536567688,
      "learning_rate": 8.156676435544997e-05,
      "loss": 0.9956,
      "step": 12047
    },
    {
      "epoch": 2.238989035495261,
      "grad_norm": 0.7745480537414551,
      "learning_rate": 8.15524055574884e-05,
      "loss": 0.9158,
      "step": 12048
    },
    {
      "epoch": 2.2391748745586324,
      "grad_norm": 0.6603816747665405,
      "learning_rate": 8.153804715325908e-05,
      "loss": 0.6452,
      "step": 12049
    },
    {
      "epoch": 2.2393607136220033,
      "grad_norm": 0.8387047648429871,
      "learning_rate": 8.152368914306846e-05,
      "loss": 1.0962,
      "step": 12050
    },
    {
      "epoch": 2.2395465526853746,
      "grad_norm": 0.7379850745201111,
      "learning_rate": 8.150933152722297e-05,
      "loss": 0.8961,
      "step": 12051
    },
    {
      "epoch": 2.2397323917487455,
      "grad_norm": 0.8029947280883789,
      "learning_rate": 8.149497430602909e-05,
      "loss": 1.0276,
      "step": 12052
    },
    {
      "epoch": 2.239918230812117,
      "grad_norm": 0.9665631651878357,
      "learning_rate": 8.14806174797932e-05,
      "loss": 1.0227,
      "step": 12053
    },
    {
      "epoch": 2.2401040698754877,
      "grad_norm": 0.9259684085845947,
      "learning_rate": 8.146626104882173e-05,
      "loss": 0.943,
      "step": 12054
    },
    {
      "epoch": 2.240289908938859,
      "grad_norm": 0.8665300607681274,
      "learning_rate": 8.145190501342117e-05,
      "loss": 0.9626,
      "step": 12055
    },
    {
      "epoch": 2.24047574800223,
      "grad_norm": 0.7480776309967041,
      "learning_rate": 8.143754937389784e-05,
      "loss": 0.9297,
      "step": 12056
    },
    {
      "epoch": 2.2406615870656013,
      "grad_norm": 1.044670581817627,
      "learning_rate": 8.142319413055818e-05,
      "loss": 0.9313,
      "step": 12057
    },
    {
      "epoch": 2.240847426128972,
      "grad_norm": 0.81687331199646,
      "learning_rate": 8.140883928370855e-05,
      "loss": 0.8503,
      "step": 12058
    },
    {
      "epoch": 2.2410332651923435,
      "grad_norm": 0.8070717453956604,
      "learning_rate": 8.139448483365535e-05,
      "loss": 0.8644,
      "step": 12059
    },
    {
      "epoch": 2.2412191042557144,
      "grad_norm": 0.7749050855636597,
      "learning_rate": 8.138013078070493e-05,
      "loss": 0.8437,
      "step": 12060
    },
    {
      "epoch": 2.2414049433190857,
      "grad_norm": 0.8389307856559753,
      "learning_rate": 8.136577712516369e-05,
      "loss": 0.5763,
      "step": 12061
    },
    {
      "epoch": 2.241590782382457,
      "grad_norm": 0.7573163509368896,
      "learning_rate": 8.135142386733794e-05,
      "loss": 0.7237,
      "step": 12062
    },
    {
      "epoch": 2.241776621445828,
      "grad_norm": 0.5418093204498291,
      "learning_rate": 8.133707100753404e-05,
      "loss": 0.4378,
      "step": 12063
    },
    {
      "epoch": 2.2419624605091992,
      "grad_norm": 0.784127414226532,
      "learning_rate": 8.132271854605835e-05,
      "loss": 0.963,
      "step": 12064
    },
    {
      "epoch": 2.24214829957257,
      "grad_norm": 0.7751089930534363,
      "learning_rate": 8.130836648321713e-05,
      "loss": 1.0521,
      "step": 12065
    },
    {
      "epoch": 2.2423341386359414,
      "grad_norm": 0.6985954642295837,
      "learning_rate": 8.12940148193168e-05,
      "loss": 0.7062,
      "step": 12066
    },
    {
      "epoch": 2.2425199776993123,
      "grad_norm": 0.8636794090270996,
      "learning_rate": 8.127966355466363e-05,
      "loss": 0.975,
      "step": 12067
    },
    {
      "epoch": 2.2427058167626837,
      "grad_norm": 0.9296517968177795,
      "learning_rate": 8.126531268956389e-05,
      "loss": 0.992,
      "step": 12068
    },
    {
      "epoch": 2.2428916558260545,
      "grad_norm": 0.9428527355194092,
      "learning_rate": 8.125096222432392e-05,
      "loss": 0.8024,
      "step": 12069
    },
    {
      "epoch": 2.243077494889426,
      "grad_norm": 0.786514937877655,
      "learning_rate": 8.123661215924999e-05,
      "loss": 0.9604,
      "step": 12070
    },
    {
      "epoch": 2.2432633339527968,
      "grad_norm": 0.8809694051742554,
      "learning_rate": 8.122226249464836e-05,
      "loss": 1.2671,
      "step": 12071
    },
    {
      "epoch": 2.243449173016168,
      "grad_norm": 0.791553258895874,
      "learning_rate": 8.120791323082531e-05,
      "loss": 0.9265,
      "step": 12072
    },
    {
      "epoch": 2.243635012079539,
      "grad_norm": 0.769401490688324,
      "learning_rate": 8.11935643680871e-05,
      "loss": 0.9175,
      "step": 12073
    },
    {
      "epoch": 2.2438208511429103,
      "grad_norm": 0.9083631038665771,
      "learning_rate": 8.117921590674001e-05,
      "loss": 1.0466,
      "step": 12074
    },
    {
      "epoch": 2.244006690206281,
      "grad_norm": 0.8304582238197327,
      "learning_rate": 8.116486784709022e-05,
      "loss": 1.1085,
      "step": 12075
    },
    {
      "epoch": 2.2441925292696525,
      "grad_norm": 1.6944912672042847,
      "learning_rate": 8.115052018944402e-05,
      "loss": 1.3513,
      "step": 12076
    },
    {
      "epoch": 2.2443783683330234,
      "grad_norm": 0.9561991095542908,
      "learning_rate": 8.113617293410763e-05,
      "loss": 1.0121,
      "step": 12077
    },
    {
      "epoch": 2.2445642073963947,
      "grad_norm": 0.8028138875961304,
      "learning_rate": 8.112182608138725e-05,
      "loss": 0.9769,
      "step": 12078
    },
    {
      "epoch": 2.2447500464597656,
      "grad_norm": 1.1675058603286743,
      "learning_rate": 8.11074796315891e-05,
      "loss": 1.0301,
      "step": 12079
    },
    {
      "epoch": 2.244935885523137,
      "grad_norm": 0.7154922485351562,
      "learning_rate": 8.10931335850194e-05,
      "loss": 0.9621,
      "step": 12080
    },
    {
      "epoch": 2.2451217245865083,
      "grad_norm": 0.9804053902626038,
      "learning_rate": 8.10787879419843e-05,
      "loss": 0.8526,
      "step": 12081
    },
    {
      "epoch": 2.245307563649879,
      "grad_norm": 0.859544038772583,
      "learning_rate": 8.106444270278999e-05,
      "loss": 0.9144,
      "step": 12082
    },
    {
      "epoch": 2.2454934027132505,
      "grad_norm": 0.6977066397666931,
      "learning_rate": 8.105009786774264e-05,
      "loss": 0.5816,
      "step": 12083
    },
    {
      "epoch": 2.2456792417766214,
      "grad_norm": 1.0285512208938599,
      "learning_rate": 8.103575343714847e-05,
      "loss": 0.9562,
      "step": 12084
    },
    {
      "epoch": 2.2458650808399927,
      "grad_norm": 0.9995384216308594,
      "learning_rate": 8.102140941131358e-05,
      "loss": 0.9575,
      "step": 12085
    },
    {
      "epoch": 2.2460509199033636,
      "grad_norm": 1.0273016691207886,
      "learning_rate": 8.10070657905441e-05,
      "loss": 0.895,
      "step": 12086
    },
    {
      "epoch": 2.246236758966735,
      "grad_norm": 0.8160095810890198,
      "learning_rate": 8.099272257514627e-05,
      "loss": 1.02,
      "step": 12087
    },
    {
      "epoch": 2.246422598030106,
      "grad_norm": 0.8777551651000977,
      "learning_rate": 8.097837976542612e-05,
      "loss": 1.1007,
      "step": 12088
    },
    {
      "epoch": 2.246608437093477,
      "grad_norm": 0.7665696740150452,
      "learning_rate": 8.096403736168985e-05,
      "loss": 0.9315,
      "step": 12089
    },
    {
      "epoch": 2.246794276156848,
      "grad_norm": 0.8144046664237976,
      "learning_rate": 8.094969536424351e-05,
      "loss": 0.9009,
      "step": 12090
    },
    {
      "epoch": 2.2469801152202193,
      "grad_norm": 0.7302137613296509,
      "learning_rate": 8.093535377339325e-05,
      "loss": 0.8593,
      "step": 12091
    },
    {
      "epoch": 2.2471659542835902,
      "grad_norm": 0.6795136332511902,
      "learning_rate": 8.092101258944512e-05,
      "loss": 0.9363,
      "step": 12092
    },
    {
      "epoch": 2.2473517933469616,
      "grad_norm": 0.7593479752540588,
      "learning_rate": 8.090667181270524e-05,
      "loss": 0.89,
      "step": 12093
    },
    {
      "epoch": 2.2475376324103324,
      "grad_norm": 0.832933247089386,
      "learning_rate": 8.08923314434797e-05,
      "loss": 0.755,
      "step": 12094
    },
    {
      "epoch": 2.2477234714737038,
      "grad_norm": 0.8301267623901367,
      "learning_rate": 8.087799148207455e-05,
      "loss": 0.9751,
      "step": 12095
    },
    {
      "epoch": 2.247909310537075,
      "grad_norm": 1.8187836408615112,
      "learning_rate": 8.086365192879582e-05,
      "loss": 1.3092,
      "step": 12096
    },
    {
      "epoch": 2.248095149600446,
      "grad_norm": 0.767377495765686,
      "learning_rate": 8.084931278394964e-05,
      "loss": 0.9277,
      "step": 12097
    },
    {
      "epoch": 2.2482809886638173,
      "grad_norm": 0.9382076263427734,
      "learning_rate": 8.083497404784201e-05,
      "loss": 1.0645,
      "step": 12098
    },
    {
      "epoch": 2.248466827727188,
      "grad_norm": 0.824059247970581,
      "learning_rate": 8.082063572077899e-05,
      "loss": 0.8328,
      "step": 12099
    },
    {
      "epoch": 2.2486526667905595,
      "grad_norm": 1.642223596572876,
      "learning_rate": 8.080629780306655e-05,
      "loss": 1.2618,
      "step": 12100
    },
    {
      "epoch": 2.2488385058539304,
      "grad_norm": 0.7809373140335083,
      "learning_rate": 8.079196029501077e-05,
      "loss": 0.9178,
      "step": 12101
    },
    {
      "epoch": 2.2490243449173017,
      "grad_norm": 0.7867723703384399,
      "learning_rate": 8.077762319691763e-05,
      "loss": 1.1984,
      "step": 12102
    },
    {
      "epoch": 2.2492101839806726,
      "grad_norm": 0.8529563546180725,
      "learning_rate": 8.076328650909313e-05,
      "loss": 0.8242,
      "step": 12103
    },
    {
      "epoch": 2.249396023044044,
      "grad_norm": 0.5668296813964844,
      "learning_rate": 8.074895023184328e-05,
      "loss": 0.4841,
      "step": 12104
    },
    {
      "epoch": 2.249581862107415,
      "grad_norm": 0.8013913631439209,
      "learning_rate": 8.073461436547402e-05,
      "loss": 1.1196,
      "step": 12105
    },
    {
      "epoch": 2.249767701170786,
      "grad_norm": 0.8628795146942139,
      "learning_rate": 8.07202789102914e-05,
      "loss": 0.9863,
      "step": 12106
    },
    {
      "epoch": 2.249953540234157,
      "grad_norm": 0.9259904623031616,
      "learning_rate": 8.07059438666013e-05,
      "loss": 1.2337,
      "step": 12107
    },
    {
      "epoch": 2.2501393792975284,
      "grad_norm": 1.2544859647750854,
      "learning_rate": 8.069160923470974e-05,
      "loss": 0.7573,
      "step": 12108
    },
    {
      "epoch": 2.2503252183608993,
      "grad_norm": 0.6399009227752686,
      "learning_rate": 8.067727501492265e-05,
      "loss": 0.6587,
      "step": 12109
    },
    {
      "epoch": 2.2505110574242706,
      "grad_norm": 0.7670935988426208,
      "learning_rate": 8.066294120754595e-05,
      "loss": 0.89,
      "step": 12110
    },
    {
      "epoch": 2.250696896487642,
      "grad_norm": 0.695530891418457,
      "learning_rate": 8.064860781288561e-05,
      "loss": 0.7293,
      "step": 12111
    },
    {
      "epoch": 2.250882735551013,
      "grad_norm": 0.7325321435928345,
      "learning_rate": 8.063427483124751e-05,
      "loss": 0.8391,
      "step": 12112
    },
    {
      "epoch": 2.2510685746143837,
      "grad_norm": 0.9947556257247925,
      "learning_rate": 8.061994226293757e-05,
      "loss": 1.0669,
      "step": 12113
    },
    {
      "epoch": 2.251254413677755,
      "grad_norm": 0.8754369616508484,
      "learning_rate": 8.060561010826173e-05,
      "loss": 1.1143,
      "step": 12114
    },
    {
      "epoch": 2.251254413677755,
      "eval_loss": 1.0257036685943604,
      "eval_runtime": 23.5312,
      "eval_samples_per_second": 46.406,
      "eval_steps_per_second": 23.203,
      "step": 12114
    },
    {
      "epoch": 2.2514402527411264,
      "grad_norm": 0.8003126978874207,
      "learning_rate": 8.059127836752585e-05,
      "loss": 1.0135,
      "step": 12115
    },
    {
      "epoch": 2.2516260918044972,
      "grad_norm": 0.815754771232605,
      "learning_rate": 8.057694704103584e-05,
      "loss": 0.911,
      "step": 12116
    },
    {
      "epoch": 2.2518119308678686,
      "grad_norm": 0.7302402257919312,
      "learning_rate": 8.056261612909752e-05,
      "loss": 0.9386,
      "step": 12117
    },
    {
      "epoch": 2.2519977699312395,
      "grad_norm": 0.810268223285675,
      "learning_rate": 8.054828563201682e-05,
      "loss": 0.6335,
      "step": 12118
    },
    {
      "epoch": 2.252183608994611,
      "grad_norm": 1.025160551071167,
      "learning_rate": 8.053395555009961e-05,
      "loss": 0.9549,
      "step": 12119
    },
    {
      "epoch": 2.2523694480579817,
      "grad_norm": 1.0502477884292603,
      "learning_rate": 8.051962588365169e-05,
      "loss": 0.9445,
      "step": 12120
    },
    {
      "epoch": 2.252555287121353,
      "grad_norm": 0.6322656273841858,
      "learning_rate": 8.050529663297895e-05,
      "loss": 0.6752,
      "step": 12121
    },
    {
      "epoch": 2.252741126184724,
      "grad_norm": 0.8192923069000244,
      "learning_rate": 8.049096779838719e-05,
      "loss": 0.9635,
      "step": 12122
    },
    {
      "epoch": 2.252926965248095,
      "grad_norm": 0.7621162533760071,
      "learning_rate": 8.047663938018223e-05,
      "loss": 0.8157,
      "step": 12123
    },
    {
      "epoch": 2.253112804311466,
      "grad_norm": 0.818742573261261,
      "learning_rate": 8.046231137866992e-05,
      "loss": 0.9787,
      "step": 12124
    },
    {
      "epoch": 2.2532986433748374,
      "grad_norm": 0.7273044586181641,
      "learning_rate": 8.044798379415603e-05,
      "loss": 0.737,
      "step": 12125
    },
    {
      "epoch": 2.2534844824382088,
      "grad_norm": 0.7540285587310791,
      "learning_rate": 8.043365662694639e-05,
      "loss": 0.7866,
      "step": 12126
    },
    {
      "epoch": 2.2536703215015796,
      "grad_norm": 0.7194929122924805,
      "learning_rate": 8.041932987734674e-05,
      "loss": 0.8063,
      "step": 12127
    },
    {
      "epoch": 2.2538561605649505,
      "grad_norm": 0.7326306104660034,
      "learning_rate": 8.040500354566288e-05,
      "loss": 0.8546,
      "step": 12128
    },
    {
      "epoch": 2.254041999628322,
      "grad_norm": 0.7915225625038147,
      "learning_rate": 8.039067763220064e-05,
      "loss": 0.979,
      "step": 12129
    },
    {
      "epoch": 2.254227838691693,
      "grad_norm": 0.671522319316864,
      "learning_rate": 8.037635213726571e-05,
      "loss": 0.6696,
      "step": 12130
    },
    {
      "epoch": 2.254413677755064,
      "grad_norm": 0.8632599711418152,
      "learning_rate": 8.03620270611639e-05,
      "loss": 1.0085,
      "step": 12131
    },
    {
      "epoch": 2.2545995168184354,
      "grad_norm": 0.9222419857978821,
      "learning_rate": 8.034770240420089e-05,
      "loss": 0.7619,
      "step": 12132
    },
    {
      "epoch": 2.2547853558818063,
      "grad_norm": 0.804999828338623,
      "learning_rate": 8.033337816668244e-05,
      "loss": 1.0261,
      "step": 12133
    },
    {
      "epoch": 2.2549711949451776,
      "grad_norm": 0.7770805358886719,
      "learning_rate": 8.031905434891431e-05,
      "loss": 0.7145,
      "step": 12134
    },
    {
      "epoch": 2.2551570340085485,
      "grad_norm": 0.9589446187019348,
      "learning_rate": 8.030473095120218e-05,
      "loss": 1.1274,
      "step": 12135
    },
    {
      "epoch": 2.25534287307192,
      "grad_norm": 0.7501167058944702,
      "learning_rate": 8.029040797385178e-05,
      "loss": 0.7728,
      "step": 12136
    },
    {
      "epoch": 2.2555287121352907,
      "grad_norm": 0.8118953704833984,
      "learning_rate": 8.027608541716878e-05,
      "loss": 1.0878,
      "step": 12137
    },
    {
      "epoch": 2.255714551198662,
      "grad_norm": 0.7665377259254456,
      "learning_rate": 8.026176328145887e-05,
      "loss": 0.9035,
      "step": 12138
    },
    {
      "epoch": 2.255900390262033,
      "grad_norm": 0.7800799012184143,
      "learning_rate": 8.02474415670278e-05,
      "loss": 1.0653,
      "step": 12139
    },
    {
      "epoch": 2.2560862293254043,
      "grad_norm": 3.0759878158569336,
      "learning_rate": 8.023312027418116e-05,
      "loss": 0.9193,
      "step": 12140
    },
    {
      "epoch": 2.256272068388775,
      "grad_norm": 0.7537910342216492,
      "learning_rate": 8.021879940322467e-05,
      "loss": 1.0038,
      "step": 12141
    },
    {
      "epoch": 2.2564579074521465,
      "grad_norm": 0.6949830651283264,
      "learning_rate": 8.020447895446395e-05,
      "loss": 0.9341,
      "step": 12142
    },
    {
      "epoch": 2.2566437465155174,
      "grad_norm": 0.7067804336547852,
      "learning_rate": 8.019015892820464e-05,
      "loss": 0.8213,
      "step": 12143
    },
    {
      "epoch": 2.2568295855788887,
      "grad_norm": 1.0574980974197388,
      "learning_rate": 8.017583932475242e-05,
      "loss": 1.038,
      "step": 12144
    },
    {
      "epoch": 2.25701542464226,
      "grad_norm": 0.8189867734909058,
      "learning_rate": 8.016152014441287e-05,
      "loss": 0.9441,
      "step": 12145
    },
    {
      "epoch": 2.257201263705631,
      "grad_norm": 0.790252685546875,
      "learning_rate": 8.014720138749166e-05,
      "loss": 1.026,
      "step": 12146
    },
    {
      "epoch": 2.2573871027690022,
      "grad_norm": 1.8028156757354736,
      "learning_rate": 8.013288305429433e-05,
      "loss": 1.297,
      "step": 12147
    },
    {
      "epoch": 2.257572941832373,
      "grad_norm": 0.8649793863296509,
      "learning_rate": 8.011856514512655e-05,
      "loss": 1.0208,
      "step": 12148
    },
    {
      "epoch": 2.2577587808957444,
      "grad_norm": 0.9388594627380371,
      "learning_rate": 8.010424766029383e-05,
      "loss": 0.8088,
      "step": 12149
    },
    {
      "epoch": 2.2579446199591153,
      "grad_norm": 0.8206169009208679,
      "learning_rate": 8.008993060010183e-05,
      "loss": 1.0357,
      "step": 12150
    },
    {
      "epoch": 2.2581304590224867,
      "grad_norm": 0.7279551029205322,
      "learning_rate": 8.00756139648561e-05,
      "loss": 0.7968,
      "step": 12151
    },
    {
      "epoch": 2.2583162980858575,
      "grad_norm": 0.8391129970550537,
      "learning_rate": 8.00612977548622e-05,
      "loss": 0.7751,
      "step": 12152
    },
    {
      "epoch": 2.258502137149229,
      "grad_norm": 0.907163143157959,
      "learning_rate": 8.004698197042567e-05,
      "loss": 0.8934,
      "step": 12153
    },
    {
      "epoch": 2.2586879762125998,
      "grad_norm": 0.7368512153625488,
      "learning_rate": 8.003266661185209e-05,
      "loss": 0.8552,
      "step": 12154
    },
    {
      "epoch": 2.258873815275971,
      "grad_norm": 0.7704903483390808,
      "learning_rate": 8.001835167944695e-05,
      "loss": 0.8249,
      "step": 12155
    },
    {
      "epoch": 2.259059654339342,
      "grad_norm": 0.6785067915916443,
      "learning_rate": 8.000403717351583e-05,
      "loss": 0.9551,
      "step": 12156
    },
    {
      "epoch": 2.2592454934027133,
      "grad_norm": 0.799774706363678,
      "learning_rate": 7.99897230943642e-05,
      "loss": 0.776,
      "step": 12157
    },
    {
      "epoch": 2.259431332466084,
      "grad_norm": 0.7252311110496521,
      "learning_rate": 7.99754094422976e-05,
      "loss": 0.8019,
      "step": 12158
    },
    {
      "epoch": 2.2596171715294555,
      "grad_norm": 0.7100863456726074,
      "learning_rate": 7.99610962176215e-05,
      "loss": 0.8433,
      "step": 12159
    },
    {
      "epoch": 2.259803010592827,
      "grad_norm": 0.7952927350997925,
      "learning_rate": 7.994678342064144e-05,
      "loss": 1.2275,
      "step": 12160
    },
    {
      "epoch": 2.2599888496561977,
      "grad_norm": 0.7643435001373291,
      "learning_rate": 7.993247105166288e-05,
      "loss": 0.9501,
      "step": 12161
    },
    {
      "epoch": 2.2601746887195686,
      "grad_norm": 0.7532644271850586,
      "learning_rate": 7.991815911099126e-05,
      "loss": 0.8428,
      "step": 12162
    },
    {
      "epoch": 2.26036052778294,
      "grad_norm": 0.8832160234451294,
      "learning_rate": 7.990384759893209e-05,
      "loss": 1.0004,
      "step": 12163
    },
    {
      "epoch": 2.2605463668463113,
      "grad_norm": 0.7995880842208862,
      "learning_rate": 7.988953651579081e-05,
      "loss": 0.8839,
      "step": 12164
    },
    {
      "epoch": 2.260732205909682,
      "grad_norm": 0.9771887063980103,
      "learning_rate": 7.987522586187285e-05,
      "loss": 1.157,
      "step": 12165
    },
    {
      "epoch": 2.2609180449730535,
      "grad_norm": 0.743340790271759,
      "learning_rate": 7.986091563748369e-05,
      "loss": 0.7109,
      "step": 12166
    },
    {
      "epoch": 2.2611038840364244,
      "grad_norm": 0.7299709320068359,
      "learning_rate": 7.98466058429287e-05,
      "loss": 0.7499,
      "step": 12167
    },
    {
      "epoch": 2.2612897230997957,
      "grad_norm": 0.9508938789367676,
      "learning_rate": 7.983229647851333e-05,
      "loss": 1.0407,
      "step": 12168
    },
    {
      "epoch": 2.2614755621631666,
      "grad_norm": 0.8278660178184509,
      "learning_rate": 7.981798754454298e-05,
      "loss": 0.9318,
      "step": 12169
    },
    {
      "epoch": 2.261661401226538,
      "grad_norm": 1.05748450756073,
      "learning_rate": 7.980367904132302e-05,
      "loss": 1.0355,
      "step": 12170
    },
    {
      "epoch": 2.261847240289909,
      "grad_norm": 0.7254590392112732,
      "learning_rate": 7.978937096915893e-05,
      "loss": 0.8755,
      "step": 12171
    },
    {
      "epoch": 2.26203307935328,
      "grad_norm": 0.8196592926979065,
      "learning_rate": 7.977506332835599e-05,
      "loss": 1.0426,
      "step": 12172
    },
    {
      "epoch": 2.262218918416651,
      "grad_norm": 1.0252938270568848,
      "learning_rate": 7.976075611921963e-05,
      "loss": 1.1224,
      "step": 12173
    },
    {
      "epoch": 2.2624047574800223,
      "grad_norm": 0.7816571593284607,
      "learning_rate": 7.974644934205521e-05,
      "loss": 0.7767,
      "step": 12174
    },
    {
      "epoch": 2.2625905965433932,
      "grad_norm": 0.7756083607673645,
      "learning_rate": 7.973214299716806e-05,
      "loss": 0.9505,
      "step": 12175
    },
    {
      "epoch": 2.2627764356067646,
      "grad_norm": 0.6845076084136963,
      "learning_rate": 7.971783708486355e-05,
      "loss": 0.8793,
      "step": 12176
    },
    {
      "epoch": 2.2629622746701354,
      "grad_norm": 0.8913975358009338,
      "learning_rate": 7.970353160544699e-05,
      "loss": 0.9497,
      "step": 12177
    },
    {
      "epoch": 2.2631481137335068,
      "grad_norm": 0.718268871307373,
      "learning_rate": 7.968922655922374e-05,
      "loss": 1.0769,
      "step": 12178
    },
    {
      "epoch": 2.263333952796878,
      "grad_norm": 0.7820630073547363,
      "learning_rate": 7.967492194649907e-05,
      "loss": 0.8911,
      "step": 12179
    },
    {
      "epoch": 2.263519791860249,
      "grad_norm": 0.7482863664627075,
      "learning_rate": 7.966061776757828e-05,
      "loss": 1.0707,
      "step": 12180
    },
    {
      "epoch": 2.2637056309236203,
      "grad_norm": 0.9600237607955933,
      "learning_rate": 7.964631402276675e-05,
      "loss": 1.035,
      "step": 12181
    },
    {
      "epoch": 2.263891469986991,
      "grad_norm": 0.7653817534446716,
      "learning_rate": 7.963201071236971e-05,
      "loss": 0.8424,
      "step": 12182
    },
    {
      "epoch": 2.2640773090503625,
      "grad_norm": 1.5581787824630737,
      "learning_rate": 7.961770783669244e-05,
      "loss": 1.3041,
      "step": 12183
    },
    {
      "epoch": 2.2642631481137334,
      "grad_norm": 0.693136990070343,
      "learning_rate": 7.960340539604023e-05,
      "loss": 0.9574,
      "step": 12184
    },
    {
      "epoch": 2.2644489871771047,
      "grad_norm": 0.7554642558097839,
      "learning_rate": 7.958910339071831e-05,
      "loss": 0.7301,
      "step": 12185
    },
    {
      "epoch": 2.2646348262404756,
      "grad_norm": 0.9046366810798645,
      "learning_rate": 7.957480182103198e-05,
      "loss": 0.9854,
      "step": 12186
    },
    {
      "epoch": 2.264820665303847,
      "grad_norm": 0.791377067565918,
      "learning_rate": 7.956050068728643e-05,
      "loss": 0.7111,
      "step": 12187
    },
    {
      "epoch": 2.265006504367218,
      "grad_norm": 0.8362942337989807,
      "learning_rate": 7.954619998978694e-05,
      "loss": 1.0304,
      "step": 12188
    },
    {
      "epoch": 2.265192343430589,
      "grad_norm": 0.7706637978553772,
      "learning_rate": 7.95318997288387e-05,
      "loss": 0.8028,
      "step": 12189
    },
    {
      "epoch": 2.26537818249396,
      "grad_norm": 0.8431205749511719,
      "learning_rate": 7.951759990474694e-05,
      "loss": 0.9803,
      "step": 12190
    },
    {
      "epoch": 2.2655640215573314,
      "grad_norm": 0.7928107976913452,
      "learning_rate": 7.950330051781682e-05,
      "loss": 1.0411,
      "step": 12191
    },
    {
      "epoch": 2.2657498606207023,
      "grad_norm": 0.7658857107162476,
      "learning_rate": 7.948900156835361e-05,
      "loss": 0.6491,
      "step": 12192
    },
    {
      "epoch": 2.2659356996840736,
      "grad_norm": 0.9177638292312622,
      "learning_rate": 7.947470305666245e-05,
      "loss": 0.8517,
      "step": 12193
    },
    {
      "epoch": 2.266121538747445,
      "grad_norm": 0.6885226368904114,
      "learning_rate": 7.946040498304856e-05,
      "loss": 0.7414,
      "step": 12194
    },
    {
      "epoch": 2.266307377810816,
      "grad_norm": 0.8419578075408936,
      "learning_rate": 7.944610734781706e-05,
      "loss": 0.816,
      "step": 12195
    },
    {
      "epoch": 2.266493216874187,
      "grad_norm": 0.8276805281639099,
      "learning_rate": 7.943181015127314e-05,
      "loss": 0.9183,
      "step": 12196
    },
    {
      "epoch": 2.266679055937558,
      "grad_norm": 0.7674410343170166,
      "learning_rate": 7.941751339372192e-05,
      "loss": 1.0516,
      "step": 12197
    },
    {
      "epoch": 2.2668648950009294,
      "grad_norm": 0.9558389186859131,
      "learning_rate": 7.940321707546857e-05,
      "loss": 1.1491,
      "step": 12198
    },
    {
      "epoch": 2.2670507340643002,
      "grad_norm": 0.8432926535606384,
      "learning_rate": 7.93889211968182e-05,
      "loss": 0.9333,
      "step": 12199
    },
    {
      "epoch": 2.2672365731276716,
      "grad_norm": 0.7996000051498413,
      "learning_rate": 7.937462575807593e-05,
      "loss": 0.8055,
      "step": 12200
    },
    {
      "epoch": 2.2674224121910425,
      "grad_norm": 0.7567355036735535,
      "learning_rate": 7.936033075954687e-05,
      "loss": 1.0259,
      "step": 12201
    },
    {
      "epoch": 2.267608251254414,
      "grad_norm": 1.0439573526382446,
      "learning_rate": 7.934603620153614e-05,
      "loss": 1.1349,
      "step": 12202
    },
    {
      "epoch": 2.2677940903177847,
      "grad_norm": 1.0130488872528076,
      "learning_rate": 7.933174208434882e-05,
      "loss": 1.0292,
      "step": 12203
    },
    {
      "epoch": 2.267979929381156,
      "grad_norm": 0.7859103083610535,
      "learning_rate": 7.931744840829e-05,
      "loss": 0.9531,
      "step": 12204
    },
    {
      "epoch": 2.268165768444527,
      "grad_norm": 0.7681362628936768,
      "learning_rate": 7.930315517366475e-05,
      "loss": 0.8845,
      "step": 12205
    },
    {
      "epoch": 2.268351607507898,
      "grad_norm": 0.7114256024360657,
      "learning_rate": 7.928886238077816e-05,
      "loss": 0.9216,
      "step": 12206
    },
    {
      "epoch": 2.268537446571269,
      "grad_norm": 0.935510516166687,
      "learning_rate": 7.927457002993524e-05,
      "loss": 0.881,
      "step": 12207
    },
    {
      "epoch": 2.2687232856346404,
      "grad_norm": 0.7839011549949646,
      "learning_rate": 7.926027812144106e-05,
      "loss": 0.9851,
      "step": 12208
    },
    {
      "epoch": 2.2689091246980118,
      "grad_norm": 0.8583346605300903,
      "learning_rate": 7.924598665560065e-05,
      "loss": 1.053,
      "step": 12209
    },
    {
      "epoch": 2.2690949637613826,
      "grad_norm": 0.811122477054596,
      "learning_rate": 7.923169563271904e-05,
      "loss": 0.8436,
      "step": 12210
    },
    {
      "epoch": 2.2692808028247535,
      "grad_norm": 0.7919806838035583,
      "learning_rate": 7.921740505310125e-05,
      "loss": 0.7359,
      "step": 12211
    },
    {
      "epoch": 2.269466641888125,
      "grad_norm": 0.686648428440094,
      "learning_rate": 7.920311491705225e-05,
      "loss": 0.7526,
      "step": 12212
    },
    {
      "epoch": 2.269652480951496,
      "grad_norm": 0.8840171098709106,
      "learning_rate": 7.91888252248771e-05,
      "loss": 1.2036,
      "step": 12213
    },
    {
      "epoch": 2.269838320014867,
      "grad_norm": 0.8836767673492432,
      "learning_rate": 7.917453597688078e-05,
      "loss": 0.876,
      "step": 12214
    },
    {
      "epoch": 2.2700241590782384,
      "grad_norm": 0.8597715497016907,
      "learning_rate": 7.916024717336823e-05,
      "loss": 0.6955,
      "step": 12215
    },
    {
      "epoch": 2.2702099981416093,
      "grad_norm": 1.0432536602020264,
      "learning_rate": 7.914595881464448e-05,
      "loss": 0.9834,
      "step": 12216
    },
    {
      "epoch": 2.2703958372049806,
      "grad_norm": 0.930818498134613,
      "learning_rate": 7.913167090101442e-05,
      "loss": 0.9366,
      "step": 12217
    },
    {
      "epoch": 2.2705816762683515,
      "grad_norm": 0.8075294494628906,
      "learning_rate": 7.911738343278304e-05,
      "loss": 1.0056,
      "step": 12218
    },
    {
      "epoch": 2.270767515331723,
      "grad_norm": 0.7768161296844482,
      "learning_rate": 7.910309641025526e-05,
      "loss": 0.9891,
      "step": 12219
    },
    {
      "epoch": 2.2709533543950937,
      "grad_norm": 0.797738254070282,
      "learning_rate": 7.908880983373604e-05,
      "loss": 1.0823,
      "step": 12220
    },
    {
      "epoch": 2.271139193458465,
      "grad_norm": 0.7448236346244812,
      "learning_rate": 7.907452370353028e-05,
      "loss": 0.7997,
      "step": 12221
    },
    {
      "epoch": 2.271325032521836,
      "grad_norm": 0.9213894605636597,
      "learning_rate": 7.906023801994289e-05,
      "loss": 0.8196,
      "step": 12222
    },
    {
      "epoch": 2.2715108715852073,
      "grad_norm": 0.8545947074890137,
      "learning_rate": 7.904595278327879e-05,
      "loss": 0.9532,
      "step": 12223
    },
    {
      "epoch": 2.271696710648578,
      "grad_norm": 0.9589551687240601,
      "learning_rate": 7.90316679938429e-05,
      "loss": 0.9656,
      "step": 12224
    },
    {
      "epoch": 2.2718825497119495,
      "grad_norm": 0.8454769849777222,
      "learning_rate": 7.901738365194004e-05,
      "loss": 0.7572,
      "step": 12225
    },
    {
      "epoch": 2.2720683887753204,
      "grad_norm": 0.8186737895011902,
      "learning_rate": 7.900309975787515e-05,
      "loss": 1.0801,
      "step": 12226
    },
    {
      "epoch": 2.2722542278386917,
      "grad_norm": 0.7751138210296631,
      "learning_rate": 7.898881631195303e-05,
      "loss": 0.8403,
      "step": 12227
    },
    {
      "epoch": 2.272440066902063,
      "grad_norm": 0.7126848101615906,
      "learning_rate": 7.897453331447862e-05,
      "loss": 0.8724,
      "step": 12228
    },
    {
      "epoch": 2.272625905965434,
      "grad_norm": 0.8206993341445923,
      "learning_rate": 7.896025076575667e-05,
      "loss": 0.7512,
      "step": 12229
    },
    {
      "epoch": 2.2728117450288052,
      "grad_norm": 0.7583099007606506,
      "learning_rate": 7.894596866609208e-05,
      "loss": 0.8338,
      "step": 12230
    },
    {
      "epoch": 2.272997584092176,
      "grad_norm": 0.8556515574455261,
      "learning_rate": 7.893168701578966e-05,
      "loss": 1.1067,
      "step": 12231
    },
    {
      "epoch": 2.2731834231555474,
      "grad_norm": 0.9765307307243347,
      "learning_rate": 7.891740581515422e-05,
      "loss": 0.9025,
      "step": 12232
    },
    {
      "epoch": 2.2733692622189183,
      "grad_norm": 0.7528972029685974,
      "learning_rate": 7.890312506449053e-05,
      "loss": 1.1078,
      "step": 12233
    },
    {
      "epoch": 2.2735551012822897,
      "grad_norm": 0.8102426528930664,
      "learning_rate": 7.888884476410348e-05,
      "loss": 0.8538,
      "step": 12234
    },
    {
      "epoch": 2.2737409403456605,
      "grad_norm": 1.0095503330230713,
      "learning_rate": 7.88745649142978e-05,
      "loss": 0.9943,
      "step": 12235
    },
    {
      "epoch": 2.273926779409032,
      "grad_norm": 0.7305726408958435,
      "learning_rate": 7.88602855153783e-05,
      "loss": 0.8046,
      "step": 12236
    },
    {
      "epoch": 2.2741126184724028,
      "grad_norm": 0.6932523846626282,
      "learning_rate": 7.884600656764971e-05,
      "loss": 0.9641,
      "step": 12237
    },
    {
      "epoch": 2.274298457535774,
      "grad_norm": 0.92442786693573,
      "learning_rate": 7.883172807141683e-05,
      "loss": 1.1832,
      "step": 12238
    },
    {
      "epoch": 2.274484296599145,
      "grad_norm": 1.3693289756774902,
      "learning_rate": 7.881745002698435e-05,
      "loss": 1.3648,
      "step": 12239
    },
    {
      "epoch": 2.2746701356625163,
      "grad_norm": 0.7918336391448975,
      "learning_rate": 7.880317243465709e-05,
      "loss": 0.9941,
      "step": 12240
    },
    {
      "epoch": 2.274855974725887,
      "grad_norm": 0.8530664443969727,
      "learning_rate": 7.878889529473969e-05,
      "loss": 0.7045,
      "step": 12241
    },
    {
      "epoch": 2.2750418137892585,
      "grad_norm": 0.8111917972564697,
      "learning_rate": 7.877461860753697e-05,
      "loss": 0.9872,
      "step": 12242
    },
    {
      "epoch": 2.27522765285263,
      "grad_norm": 0.7255557775497437,
      "learning_rate": 7.876034237335352e-05,
      "loss": 0.9302,
      "step": 12243
    },
    {
      "epoch": 2.2754134919160007,
      "grad_norm": 1.0161799192428589,
      "learning_rate": 7.874606659249417e-05,
      "loss": 0.9581,
      "step": 12244
    },
    {
      "epoch": 2.2755993309793716,
      "grad_norm": 0.7238139510154724,
      "learning_rate": 7.873179126526355e-05,
      "loss": 0.9064,
      "step": 12245
    },
    {
      "epoch": 2.275785170042743,
      "grad_norm": 0.7794341444969177,
      "learning_rate": 7.871751639196635e-05,
      "loss": 0.6845,
      "step": 12246
    },
    {
      "epoch": 2.2759710091061143,
      "grad_norm": 0.8446567058563232,
      "learning_rate": 7.870324197290723e-05,
      "loss": 1.1763,
      "step": 12247
    },
    {
      "epoch": 2.276156848169485,
      "grad_norm": 0.875368595123291,
      "learning_rate": 7.868896800839087e-05,
      "loss": 0.84,
      "step": 12248
    },
    {
      "epoch": 2.2763426872328565,
      "grad_norm": 0.9106482863426208,
      "learning_rate": 7.867469449872192e-05,
      "loss": 1.1002,
      "step": 12249
    },
    {
      "epoch": 2.2765285262962274,
      "grad_norm": 0.9384245276451111,
      "learning_rate": 7.866042144420502e-05,
      "loss": 1.0101,
      "step": 12250
    },
    {
      "epoch": 2.2767143653595987,
      "grad_norm": 0.748377799987793,
      "learning_rate": 7.864614884514479e-05,
      "loss": 0.9531,
      "step": 12251
    },
    {
      "epoch": 2.2769002044229696,
      "grad_norm": 0.7939218878746033,
      "learning_rate": 7.86318767018459e-05,
      "loss": 0.8427,
      "step": 12252
    },
    {
      "epoch": 2.277086043486341,
      "grad_norm": 0.7586506605148315,
      "learning_rate": 7.861760501461291e-05,
      "loss": 0.8141,
      "step": 12253
    },
    {
      "epoch": 2.277271882549712,
      "grad_norm": 0.9688478112220764,
      "learning_rate": 7.860333378375041e-05,
      "loss": 0.7653,
      "step": 12254
    },
    {
      "epoch": 2.277457721613083,
      "grad_norm": 0.7218331098556519,
      "learning_rate": 7.858906300956305e-05,
      "loss": 0.9217,
      "step": 12255
    },
    {
      "epoch": 2.277643560676454,
      "grad_norm": 0.9971310496330261,
      "learning_rate": 7.857479269235544e-05,
      "loss": 1.1914,
      "step": 12256
    },
    {
      "epoch": 2.2778293997398253,
      "grad_norm": 0.9666123986244202,
      "learning_rate": 7.856052283243209e-05,
      "loss": 0.898,
      "step": 12257
    },
    {
      "epoch": 2.2780152388031967,
      "grad_norm": 0.6891591548919678,
      "learning_rate": 7.854625343009759e-05,
      "loss": 0.8644,
      "step": 12258
    },
    {
      "epoch": 2.2782010778665676,
      "grad_norm": 0.6942464113235474,
      "learning_rate": 7.85319844856565e-05,
      "loss": 0.8322,
      "step": 12259
    },
    {
      "epoch": 2.2783869169299384,
      "grad_norm": 0.715089738368988,
      "learning_rate": 7.851771599941336e-05,
      "loss": 0.733,
      "step": 12260
    },
    {
      "epoch": 2.2785727559933098,
      "grad_norm": 0.7570328116416931,
      "learning_rate": 7.85034479716727e-05,
      "loss": 1.1289,
      "step": 12261
    },
    {
      "epoch": 2.278758595056681,
      "grad_norm": 1.1029595136642456,
      "learning_rate": 7.848918040273907e-05,
      "loss": 1.0443,
      "step": 12262
    },
    {
      "epoch": 2.278944434120052,
      "grad_norm": 0.7635982036590576,
      "learning_rate": 7.847491329291694e-05,
      "loss": 1.0012,
      "step": 12263
    },
    {
      "epoch": 2.2791302731834233,
      "grad_norm": 0.8406252264976501,
      "learning_rate": 7.846064664251082e-05,
      "loss": 1.0415,
      "step": 12264
    },
    {
      "epoch": 2.279316112246794,
      "grad_norm": 1.0755679607391357,
      "learning_rate": 7.844638045182528e-05,
      "loss": 1.1597,
      "step": 12265
    },
    {
      "epoch": 2.2795019513101655,
      "grad_norm": 0.808347225189209,
      "learning_rate": 7.843211472116476e-05,
      "loss": 0.906,
      "step": 12266
    },
    {
      "epoch": 2.2796877903735364,
      "grad_norm": 0.9315565228462219,
      "learning_rate": 7.841784945083373e-05,
      "loss": 0.7202,
      "step": 12267
    },
    {
      "epoch": 2.2798736294369077,
      "grad_norm": 0.6871508955955505,
      "learning_rate": 7.840358464113669e-05,
      "loss": 0.8551,
      "step": 12268
    },
    {
      "epoch": 2.2800594685002786,
      "grad_norm": 0.7517482042312622,
      "learning_rate": 7.838932029237805e-05,
      "loss": 1.0232,
      "step": 12269
    },
    {
      "epoch": 2.28024530756365,
      "grad_norm": 0.8469917178153992,
      "learning_rate": 7.837505640486232e-05,
      "loss": 0.9563,
      "step": 12270
    },
    {
      "epoch": 2.280431146627021,
      "grad_norm": 0.7821633815765381,
      "learning_rate": 7.836079297889385e-05,
      "loss": 0.8611,
      "step": 12271
    },
    {
      "epoch": 2.280616985690392,
      "grad_norm": 0.9212172031402588,
      "learning_rate": 7.834653001477717e-05,
      "loss": 0.9973,
      "step": 12272
    },
    {
      "epoch": 2.280802824753763,
      "grad_norm": 0.7685173153877258,
      "learning_rate": 7.833226751281661e-05,
      "loss": 1.1228,
      "step": 12273
    },
    {
      "epoch": 2.2809886638171344,
      "grad_norm": 0.8296868205070496,
      "learning_rate": 7.831800547331662e-05,
      "loss": 1.0455,
      "step": 12274
    },
    {
      "epoch": 2.2811745028805053,
      "grad_norm": 0.7512198686599731,
      "learning_rate": 7.830374389658157e-05,
      "loss": 0.8252,
      "step": 12275
    },
    {
      "epoch": 2.2813603419438766,
      "grad_norm": 0.7423361539840698,
      "learning_rate": 7.828948278291592e-05,
      "loss": 0.8986,
      "step": 12276
    },
    {
      "epoch": 2.281546181007248,
      "grad_norm": Infinity,
      "learning_rate": 7.828948278291592e-05,
      "loss": 1.8963,
      "step": 12277
    },
    {
      "epoch": 2.281732020070619,
      "grad_norm": 0.7230324149131775,
      "learning_rate": 7.827522213262399e-05,
      "loss": 0.8992,
      "step": 12278
    },
    {
      "epoch": 2.28191785913399,
      "grad_norm": 0.8100959062576294,
      "learning_rate": 7.826096194601017e-05,
      "loss": 0.7857,
      "step": 12279
    },
    {
      "epoch": 2.282103698197361,
      "grad_norm": 0.8222119212150574,
      "learning_rate": 7.82467022233788e-05,
      "loss": 0.9166,
      "step": 12280
    },
    {
      "epoch": 2.2822895372607324,
      "grad_norm": 0.7102981805801392,
      "learning_rate": 7.823244296503425e-05,
      "loss": 0.9988,
      "step": 12281
    },
    {
      "epoch": 2.2824753763241032,
      "grad_norm": 0.8669292330741882,
      "learning_rate": 7.821818417128084e-05,
      "loss": 0.9622,
      "step": 12282
    },
    {
      "epoch": 2.2826612153874746,
      "grad_norm": 0.7663910984992981,
      "learning_rate": 7.820392584242293e-05,
      "loss": 0.9651,
      "step": 12283
    },
    {
      "epoch": 2.2828470544508455,
      "grad_norm": 1.1469777822494507,
      "learning_rate": 7.81896679787648e-05,
      "loss": 1.0575,
      "step": 12284
    },
    {
      "epoch": 2.283032893514217,
      "grad_norm": 0.7489597201347351,
      "learning_rate": 7.817541058061077e-05,
      "loss": 0.857,
      "step": 12285
    },
    {
      "epoch": 2.2832187325775877,
      "grad_norm": 0.8339454531669617,
      "learning_rate": 7.816115364826513e-05,
      "loss": 0.8734,
      "step": 12286
    },
    {
      "epoch": 2.283404571640959,
      "grad_norm": 0.6416560411453247,
      "learning_rate": 7.814689718203223e-05,
      "loss": 0.6614,
      "step": 12287
    },
    {
      "epoch": 2.28359041070433,
      "grad_norm": 0.7036505937576294,
      "learning_rate": 7.813264118221629e-05,
      "loss": 0.8057,
      "step": 12288
    },
    {
      "epoch": 2.283776249767701,
      "grad_norm": 0.7501203417778015,
      "learning_rate": 7.811838564912163e-05,
      "loss": 1.0344,
      "step": 12289
    },
    {
      "epoch": 2.283962088831072,
      "grad_norm": 0.7263859510421753,
      "learning_rate": 7.810413058305244e-05,
      "loss": 1.0007,
      "step": 12290
    },
    {
      "epoch": 2.2841479278944434,
      "grad_norm": 0.6111660003662109,
      "learning_rate": 7.808987598431303e-05,
      "loss": 0.4128,
      "step": 12291
    },
    {
      "epoch": 2.2843337669578148,
      "grad_norm": 0.762751042842865,
      "learning_rate": 7.807562185320761e-05,
      "loss": 1.1027,
      "step": 12292
    },
    {
      "epoch": 2.2845196060211856,
      "grad_norm": 0.8497124910354614,
      "learning_rate": 7.806136819004042e-05,
      "loss": 0.9507,
      "step": 12293
    },
    {
      "epoch": 2.2847054450845565,
      "grad_norm": 0.8142184019088745,
      "learning_rate": 7.804711499511568e-05,
      "loss": 0.7366,
      "step": 12294
    },
    {
      "epoch": 2.284891284147928,
      "grad_norm": 0.953407347202301,
      "learning_rate": 7.803286226873759e-05,
      "loss": 0.8777,
      "step": 12295
    },
    {
      "epoch": 2.285077123211299,
      "grad_norm": 0.8734500408172607,
      "learning_rate": 7.801861001121037e-05,
      "loss": 0.9468,
      "step": 12296
    },
    {
      "epoch": 2.28526296227467,
      "grad_norm": 0.7192512154579163,
      "learning_rate": 7.800435822283816e-05,
      "loss": 1.0027,
      "step": 12297
    },
    {
      "epoch": 2.2854488013380414,
      "grad_norm": 0.8476136922836304,
      "learning_rate": 7.79901069039252e-05,
      "loss": 1.0746,
      "step": 12298
    },
    {
      "epoch": 2.2856346404014123,
      "grad_norm": 0.7300472259521484,
      "learning_rate": 7.797585605477566e-05,
      "loss": 0.9394,
      "step": 12299
    },
    {
      "epoch": 2.2858204794647836,
      "grad_norm": 0.8123537302017212,
      "learning_rate": 7.796160567569365e-05,
      "loss": 0.749,
      "step": 12300
    },
    {
      "epoch": 2.2860063185281545,
      "grad_norm": 0.8489190340042114,
      "learning_rate": 7.794735576698338e-05,
      "loss": 0.9906,
      "step": 12301
    },
    {
      "epoch": 2.286192157591526,
      "grad_norm": 0.8736398816108704,
      "learning_rate": 7.793310632894893e-05,
      "loss": 1.219,
      "step": 12302
    },
    {
      "epoch": 2.2863779966548967,
      "grad_norm": 0.8092479109764099,
      "learning_rate": 7.791885736189447e-05,
      "loss": 1.19,
      "step": 12303
    },
    {
      "epoch": 2.286563835718268,
      "grad_norm": 0.9571067690849304,
      "learning_rate": 7.79046088661241e-05,
      "loss": 1.0245,
      "step": 12304
    },
    {
      "epoch": 2.286749674781639,
      "grad_norm": 0.7240368127822876,
      "learning_rate": 7.789036084194193e-05,
      "loss": 0.9194,
      "step": 12305
    },
    {
      "epoch": 2.2869355138450103,
      "grad_norm": 0.7796837091445923,
      "learning_rate": 7.787611328965207e-05,
      "loss": 0.9709,
      "step": 12306
    },
    {
      "epoch": 2.2871213529083816,
      "grad_norm": 0.7277730703353882,
      "learning_rate": 7.78618662095586e-05,
      "loss": 0.8694,
      "step": 12307
    },
    {
      "epoch": 2.2873071919717525,
      "grad_norm": 1.0038048028945923,
      "learning_rate": 7.78476196019656e-05,
      "loss": 1.2055,
      "step": 12308
    },
    {
      "epoch": 2.2874930310351234,
      "grad_norm": 0.7962849736213684,
      "learning_rate": 7.783337346717717e-05,
      "loss": 0.8344,
      "step": 12309
    },
    {
      "epoch": 2.2876788700984947,
      "grad_norm": 0.7871695756912231,
      "learning_rate": 7.781912780549733e-05,
      "loss": 0.9471,
      "step": 12310
    },
    {
      "epoch": 2.287864709161866,
      "grad_norm": 0.8273323774337769,
      "learning_rate": 7.780488261723016e-05,
      "loss": 0.8913,
      "step": 12311
    },
    {
      "epoch": 2.288050548225237,
      "grad_norm": 1.6526739597320557,
      "learning_rate": 7.779063790267965e-05,
      "loss": 1.4423,
      "step": 12312
    },
    {
      "epoch": 2.2882363872886082,
      "grad_norm": 0.6899594068527222,
      "learning_rate": 7.777639366214991e-05,
      "loss": 0.8627,
      "step": 12313
    },
    {
      "epoch": 2.288422226351979,
      "grad_norm": 0.8037352561950684,
      "learning_rate": 7.776214989594486e-05,
      "loss": 0.7692,
      "step": 12314
    },
    {
      "epoch": 2.2886080654153504,
      "grad_norm": 0.7869132161140442,
      "learning_rate": 7.774790660436858e-05,
      "loss": 0.9398,
      "step": 12315
    },
    {
      "epoch": 2.2887939044787213,
      "grad_norm": 0.7735188603401184,
      "learning_rate": 7.773366378772506e-05,
      "loss": 0.7763,
      "step": 12316
    },
    {
      "epoch": 2.2889797435420927,
      "grad_norm": 0.6020045876502991,
      "learning_rate": 7.771942144631824e-05,
      "loss": 0.619,
      "step": 12317
    },
    {
      "epoch": 2.2891655826054635,
      "grad_norm": 0.6839711666107178,
      "learning_rate": 7.770517958045213e-05,
      "loss": 0.8986,
      "step": 12318
    },
    {
      "epoch": 2.289351421668835,
      "grad_norm": 0.9535199999809265,
      "learning_rate": 7.769093819043074e-05,
      "loss": 0.8784,
      "step": 12319
    },
    {
      "epoch": 2.2895372607322058,
      "grad_norm": 0.8104950189590454,
      "learning_rate": 7.767669727655798e-05,
      "loss": 0.8394,
      "step": 12320
    },
    {
      "epoch": 2.289723099795577,
      "grad_norm": 0.7972855567932129,
      "learning_rate": 7.766245683913781e-05,
      "loss": 0.7895,
      "step": 12321
    },
    {
      "epoch": 2.289908938858948,
      "grad_norm": 0.8431624174118042,
      "learning_rate": 7.764821687847415e-05,
      "loss": 0.8213,
      "step": 12322
    },
    {
      "epoch": 2.2900947779223193,
      "grad_norm": 0.7957172393798828,
      "learning_rate": 7.763397739487098e-05,
      "loss": 0.8179,
      "step": 12323
    },
    {
      "epoch": 2.29028061698569,
      "grad_norm": 0.707563579082489,
      "learning_rate": 7.761973838863214e-05,
      "loss": 0.6829,
      "step": 12324
    },
    {
      "epoch": 2.2904664560490615,
      "grad_norm": 0.6948530077934265,
      "learning_rate": 7.760549986006159e-05,
      "loss": 0.649,
      "step": 12325
    },
    {
      "epoch": 2.290652295112433,
      "grad_norm": 0.7895127534866333,
      "learning_rate": 7.759126180946324e-05,
      "loss": 0.7972,
      "step": 12326
    },
    {
      "epoch": 2.2908381341758037,
      "grad_norm": 0.7186713218688965,
      "learning_rate": 7.757702423714093e-05,
      "loss": 0.8915,
      "step": 12327
    },
    {
      "epoch": 2.291023973239175,
      "grad_norm": 0.7776643633842468,
      "learning_rate": 7.756278714339852e-05,
      "loss": 0.7304,
      "step": 12328
    },
    {
      "epoch": 2.291209812302546,
      "grad_norm": 0.893409013748169,
      "learning_rate": 7.754855052853997e-05,
      "loss": 0.8193,
      "step": 12329
    },
    {
      "epoch": 2.2913956513659173,
      "grad_norm": 0.7297701239585876,
      "learning_rate": 7.753431439286906e-05,
      "loss": 0.8881,
      "step": 12330
    },
    {
      "epoch": 2.291581490429288,
      "grad_norm": 0.910232663154602,
      "learning_rate": 7.752007873668967e-05,
      "loss": 1.1218,
      "step": 12331
    },
    {
      "epoch": 2.2917673294926595,
      "grad_norm": 0.8789218664169312,
      "learning_rate": 7.750584356030562e-05,
      "loss": 0.9486,
      "step": 12332
    },
    {
      "epoch": 2.2919531685560304,
      "grad_norm": 0.7561124563217163,
      "learning_rate": 7.749160886402073e-05,
      "loss": 1.1138,
      "step": 12333
    },
    {
      "epoch": 2.2921390076194017,
      "grad_norm": 0.7757100462913513,
      "learning_rate": 7.747737464813882e-05,
      "loss": 0.9899,
      "step": 12334
    },
    {
      "epoch": 2.2923248466827726,
      "grad_norm": 0.8278506994247437,
      "learning_rate": 7.74631409129637e-05,
      "loss": 0.5483,
      "step": 12335
    },
    {
      "epoch": 2.292510685746144,
      "grad_norm": 0.8181744813919067,
      "learning_rate": 7.744890765879918e-05,
      "loss": 0.9717,
      "step": 12336
    },
    {
      "epoch": 2.292696524809515,
      "grad_norm": 0.9865650534629822,
      "learning_rate": 7.7434674885949e-05,
      "loss": 1.0432,
      "step": 12337
    },
    {
      "epoch": 2.292882363872886,
      "grad_norm": 0.6678515076637268,
      "learning_rate": 7.7420442594717e-05,
      "loss": 0.6523,
      "step": 12338
    },
    {
      "epoch": 2.293068202936257,
      "grad_norm": 0.81447434425354,
      "learning_rate": 7.740621078540684e-05,
      "loss": 0.9622,
      "step": 12339
    },
    {
      "epoch": 2.2932540419996283,
      "grad_norm": 0.8288367986679077,
      "learning_rate": 7.739197945832235e-05,
      "loss": 0.9989,
      "step": 12340
    },
    {
      "epoch": 2.2934398810629997,
      "grad_norm": 0.7778246998786926,
      "learning_rate": 7.737774861376732e-05,
      "loss": 0.8636,
      "step": 12341
    },
    {
      "epoch": 2.2936257201263706,
      "grad_norm": 1.3834868669509888,
      "learning_rate": 7.736351825204537e-05,
      "loss": 1.2447,
      "step": 12342
    },
    {
      "epoch": 2.2938115591897414,
      "grad_norm": 0.9833600521087646,
      "learning_rate": 7.734928837346033e-05,
      "loss": 1.0867,
      "step": 12343
    },
    {
      "epoch": 2.2939973982531128,
      "grad_norm": 0.941429853439331,
      "learning_rate": 7.73350589783158e-05,
      "loss": 0.7508,
      "step": 12344
    },
    {
      "epoch": 2.294183237316484,
      "grad_norm": 0.9358783960342407,
      "learning_rate": 7.732083006691558e-05,
      "loss": 1.2586,
      "step": 12345
    },
    {
      "epoch": 2.294369076379855,
      "grad_norm": 0.6414960026741028,
      "learning_rate": 7.730660163956333e-05,
      "loss": 0.8383,
      "step": 12346
    },
    {
      "epoch": 2.2945549154432263,
      "grad_norm": 1.0815060138702393,
      "learning_rate": 7.729237369656269e-05,
      "loss": 1.139,
      "step": 12347
    },
    {
      "epoch": 2.294740754506597,
      "grad_norm": 0.8384090065956116,
      "learning_rate": 7.727814623821739e-05,
      "loss": 0.8974,
      "step": 12348
    },
    {
      "epoch": 2.2949265935699685,
      "grad_norm": 0.9051550626754761,
      "learning_rate": 7.726391926483104e-05,
      "loss": 1.0349,
      "step": 12349
    },
    {
      "epoch": 2.2951124326333394,
      "grad_norm": 0.8306860327720642,
      "learning_rate": 7.724969277670733e-05,
      "loss": 0.91,
      "step": 12350
    },
    {
      "epoch": 2.2952982716967107,
      "grad_norm": 0.7919532060623169,
      "learning_rate": 7.72354667741499e-05,
      "loss": 0.7608,
      "step": 12351
    },
    {
      "epoch": 2.2954841107600816,
      "grad_norm": 0.6512516736984253,
      "learning_rate": 7.722124125746236e-05,
      "loss": 0.6876,
      "step": 12352
    },
    {
      "epoch": 2.295669949823453,
      "grad_norm": 0.7694550156593323,
      "learning_rate": 7.720701622694834e-05,
      "loss": 0.9659,
      "step": 12353
    },
    {
      "epoch": 2.295855788886824,
      "grad_norm": 0.9084151983261108,
      "learning_rate": 7.719279168291144e-05,
      "loss": 0.9069,
      "step": 12354
    },
    {
      "epoch": 2.296041627950195,
      "grad_norm": 0.8145677447319031,
      "learning_rate": 7.717856762565524e-05,
      "loss": 0.9747,
      "step": 12355
    },
    {
      "epoch": 2.296227467013566,
      "grad_norm": 0.7414378523826599,
      "learning_rate": 7.71643440554834e-05,
      "loss": 0.9986,
      "step": 12356
    },
    {
      "epoch": 2.2964133060769374,
      "grad_norm": 0.7389624118804932,
      "learning_rate": 7.715012097269942e-05,
      "loss": 0.9096,
      "step": 12357
    },
    {
      "epoch": 2.2965991451403083,
      "grad_norm": 1.1509134769439697,
      "learning_rate": 7.713589837760691e-05,
      "loss": 1.1409,
      "step": 12358
    },
    {
      "epoch": 2.2967849842036796,
      "grad_norm": 0.8112145066261292,
      "learning_rate": 7.712167627050938e-05,
      "loss": 1.1119,
      "step": 12359
    },
    {
      "epoch": 2.296970823267051,
      "grad_norm": 0.8990470170974731,
      "learning_rate": 7.710745465171041e-05,
      "loss": 1.0636,
      "step": 12360
    },
    {
      "epoch": 2.297156662330422,
      "grad_norm": 0.7694247364997864,
      "learning_rate": 7.709323352151357e-05,
      "loss": 0.8278,
      "step": 12361
    },
    {
      "epoch": 2.297342501393793,
      "grad_norm": 1.0254145860671997,
      "learning_rate": 7.707901288022232e-05,
      "loss": 1.1192,
      "step": 12362
    },
    {
      "epoch": 2.297528340457164,
      "grad_norm": 0.7462365627288818,
      "learning_rate": 7.706479272814023e-05,
      "loss": 1.0382,
      "step": 12363
    },
    {
      "epoch": 2.2977141795205354,
      "grad_norm": 0.9076231122016907,
      "learning_rate": 7.705057306557076e-05,
      "loss": 0.9178,
      "step": 12364
    },
    {
      "epoch": 2.2979000185839062,
      "grad_norm": 0.8459599614143372,
      "learning_rate": 7.703635389281743e-05,
      "loss": 0.9922,
      "step": 12365
    },
    {
      "epoch": 2.2980858576472776,
      "grad_norm": 0.6892786026000977,
      "learning_rate": 7.702213521018374e-05,
      "loss": 0.7726,
      "step": 12366
    },
    {
      "epoch": 2.2982716967106485,
      "grad_norm": 1.467746615409851,
      "learning_rate": 7.700791701797313e-05,
      "loss": 1.4936,
      "step": 12367
    },
    {
      "epoch": 2.29845753577402,
      "grad_norm": 0.7365753054618835,
      "learning_rate": 7.699369931648909e-05,
      "loss": 0.9974,
      "step": 12368
    },
    {
      "epoch": 2.2986433748373907,
      "grad_norm": 0.8061040043830872,
      "learning_rate": 7.697948210603503e-05,
      "loss": 0.9074,
      "step": 12369
    },
    {
      "epoch": 2.298829213900762,
      "grad_norm": 0.7725257277488708,
      "learning_rate": 7.69652653869144e-05,
      "loss": 0.9085,
      "step": 12370
    },
    {
      "epoch": 2.299015052964133,
      "grad_norm": 0.8355392217636108,
      "learning_rate": 7.695104915943069e-05,
      "loss": 0.9589,
      "step": 12371
    },
    {
      "epoch": 2.299200892027504,
      "grad_norm": 0.7052400708198547,
      "learning_rate": 7.693683342388727e-05,
      "loss": 0.791,
      "step": 12372
    },
    {
      "epoch": 2.299386731090875,
      "grad_norm": 0.6995886564254761,
      "learning_rate": 7.692261818058758e-05,
      "loss": 0.6648,
      "step": 12373
    },
    {
      "epoch": 2.2995725701542464,
      "grad_norm": 1.0227339267730713,
      "learning_rate": 7.690840342983498e-05,
      "loss": 1.1262,
      "step": 12374
    },
    {
      "epoch": 2.2997584092176178,
      "grad_norm": 0.8556922078132629,
      "learning_rate": 7.689418917193289e-05,
      "loss": 0.8543,
      "step": 12375
    },
    {
      "epoch": 2.2999442482809886,
      "grad_norm": 0.8244157433509827,
      "learning_rate": 7.68799754071847e-05,
      "loss": 0.7228,
      "step": 12376
    },
    {
      "epoch": 2.30013008734436,
      "grad_norm": 0.8170839548110962,
      "learning_rate": 7.686576213589373e-05,
      "loss": 1.1313,
      "step": 12377
    },
    {
      "epoch": 2.300315926407731,
      "grad_norm": 0.8408236503601074,
      "learning_rate": 7.685154935836339e-05,
      "loss": 0.9342,
      "step": 12378
    },
    {
      "epoch": 2.300501765471102,
      "grad_norm": 0.75736004114151,
      "learning_rate": 7.683733707489699e-05,
      "loss": 0.7792,
      "step": 12379
    },
    {
      "epoch": 2.300687604534473,
      "grad_norm": 0.7825633883476257,
      "learning_rate": 7.68231252857979e-05,
      "loss": 1.0248,
      "step": 12380
    },
    {
      "epoch": 2.3008734435978444,
      "grad_norm": 0.7606642842292786,
      "learning_rate": 7.680891399136938e-05,
      "loss": 0.9121,
      "step": 12381
    },
    {
      "epoch": 2.3010592826612153,
      "grad_norm": 0.7969626784324646,
      "learning_rate": 7.679470319191482e-05,
      "loss": 1.0324,
      "step": 12382
    },
    {
      "epoch": 2.3012451217245866,
      "grad_norm": 0.857466459274292,
      "learning_rate": 7.678049288773752e-05,
      "loss": 1.1447,
      "step": 12383
    },
    {
      "epoch": 2.3014309607879575,
      "grad_norm": 0.757928192615509,
      "learning_rate": 7.676628307914074e-05,
      "loss": 0.7483,
      "step": 12384
    },
    {
      "epoch": 2.301616799851329,
      "grad_norm": 0.7650940418243408,
      "learning_rate": 7.675207376642775e-05,
      "loss": 0.96,
      "step": 12385
    },
    {
      "epoch": 2.3018026389146997,
      "grad_norm": 0.8665147423744202,
      "learning_rate": 7.67378649499019e-05,
      "loss": 0.8273,
      "step": 12386
    },
    {
      "epoch": 2.301988477978071,
      "grad_norm": 0.8380240797996521,
      "learning_rate": 7.672365662986636e-05,
      "loss": 0.763,
      "step": 12387
    },
    {
      "epoch": 2.302174317041442,
      "grad_norm": 0.7871870398521423,
      "learning_rate": 7.670944880662446e-05,
      "loss": 0.8913,
      "step": 12388
    },
    {
      "epoch": 2.3023601561048133,
      "grad_norm": 0.8932545781135559,
      "learning_rate": 7.669524148047938e-05,
      "loss": 0.9449,
      "step": 12389
    },
    {
      "epoch": 2.3025459951681846,
      "grad_norm": 0.8677648901939392,
      "learning_rate": 7.668103465173441e-05,
      "loss": 1.0888,
      "step": 12390
    },
    {
      "epoch": 2.3027318342315555,
      "grad_norm": 0.8020843267440796,
      "learning_rate": 7.666682832069269e-05,
      "loss": 1.0817,
      "step": 12391
    },
    {
      "epoch": 2.3029176732949264,
      "grad_norm": 0.8647310733795166,
      "learning_rate": 7.665262248765749e-05,
      "loss": 0.8104,
      "step": 12392
    },
    {
      "epoch": 2.3031035123582977,
      "grad_norm": 0.775503396987915,
      "learning_rate": 7.663841715293202e-05,
      "loss": 0.9919,
      "step": 12393
    },
    {
      "epoch": 2.303289351421669,
      "grad_norm": 0.7088800072669983,
      "learning_rate": 7.662421231681944e-05,
      "loss": 0.7559,
      "step": 12394
    },
    {
      "epoch": 2.30347519048504,
      "grad_norm": 0.8725742101669312,
      "learning_rate": 7.661000797962292e-05,
      "loss": 1.0528,
      "step": 12395
    },
    {
      "epoch": 2.3036610295484112,
      "grad_norm": 0.889268159866333,
      "learning_rate": 7.659580414164567e-05,
      "loss": 1.1647,
      "step": 12396
    },
    {
      "epoch": 2.303846868611782,
      "grad_norm": 0.7021028399467468,
      "learning_rate": 7.658160080319078e-05,
      "loss": 0.7145,
      "step": 12397
    },
    {
      "epoch": 2.3040327076751534,
      "grad_norm": 0.835035502910614,
      "learning_rate": 7.656739796456148e-05,
      "loss": 0.7709,
      "step": 12398
    },
    {
      "epoch": 2.3042185467385243,
      "grad_norm": 0.8327295184135437,
      "learning_rate": 7.655319562606083e-05,
      "loss": 0.8802,
      "step": 12399
    },
    {
      "epoch": 2.3044043858018957,
      "grad_norm": 0.7869976162910461,
      "learning_rate": 7.6538993787992e-05,
      "loss": 1.0701,
      "step": 12400
    },
    {
      "epoch": 2.3045902248652665,
      "grad_norm": 0.762888491153717,
      "learning_rate": 7.652479245065806e-05,
      "loss": 0.9929,
      "step": 12401
    },
    {
      "epoch": 2.304776063928638,
      "grad_norm": 0.8556522727012634,
      "learning_rate": 7.651059161436211e-05,
      "loss": 0.9351,
      "step": 12402
    },
    {
      "epoch": 2.3049619029920088,
      "grad_norm": 0.844419538974762,
      "learning_rate": 7.649639127940735e-05,
      "loss": 0.8421,
      "step": 12403
    },
    {
      "epoch": 2.30514774205538,
      "grad_norm": 3.3211510181427,
      "learning_rate": 7.648219144609673e-05,
      "loss": 1.0601,
      "step": 12404
    },
    {
      "epoch": 2.305333581118751,
      "grad_norm": 0.7639872431755066,
      "learning_rate": 7.646799211473338e-05,
      "loss": 0.9178,
      "step": 12405
    },
    {
      "epoch": 2.3055194201821223,
      "grad_norm": 0.8685746788978577,
      "learning_rate": 7.645379328562039e-05,
      "loss": 0.712,
      "step": 12406
    },
    {
      "epoch": 2.305705259245493,
      "grad_norm": 0.7220452427864075,
      "learning_rate": 7.643959495906074e-05,
      "loss": 0.806,
      "step": 12407
    },
    {
      "epoch": 2.3058910983088645,
      "grad_norm": 0.6752744317054749,
      "learning_rate": 7.642539713535755e-05,
      "loss": 0.8161,
      "step": 12408
    },
    {
      "epoch": 2.306076937372236,
      "grad_norm": 0.7289901971817017,
      "learning_rate": 7.641119981481377e-05,
      "loss": 0.8479,
      "step": 12409
    },
    {
      "epoch": 2.3062627764356067,
      "grad_norm": 0.6853960752487183,
      "learning_rate": 7.639700299773245e-05,
      "loss": 0.6681,
      "step": 12410
    },
    {
      "epoch": 2.306448615498978,
      "grad_norm": 0.7992112636566162,
      "learning_rate": 7.63828066844166e-05,
      "loss": 0.8624,
      "step": 12411
    },
    {
      "epoch": 2.306634454562349,
      "grad_norm": 0.7620840668678284,
      "learning_rate": 7.636861087516917e-05,
      "loss": 1.0049,
      "step": 12412
    },
    {
      "epoch": 2.3068202936257203,
      "grad_norm": 0.7466810345649719,
      "learning_rate": 7.635441557029325e-05,
      "loss": 0.8561,
      "step": 12413
    },
    {
      "epoch": 2.307006132689091,
      "grad_norm": 0.9495031833648682,
      "learning_rate": 7.634022077009173e-05,
      "loss": 1.1464,
      "step": 12414
    },
    {
      "epoch": 2.3071919717524625,
      "grad_norm": 0.8608010411262512,
      "learning_rate": 7.632602647486759e-05,
      "loss": 1.1267,
      "step": 12415
    },
    {
      "epoch": 2.3073778108158334,
      "grad_norm": 0.8829839825630188,
      "learning_rate": 7.63118326849238e-05,
      "loss": 1.0529,
      "step": 12416
    },
    {
      "epoch": 2.3075636498792047,
      "grad_norm": 0.6954786777496338,
      "learning_rate": 7.629763940056328e-05,
      "loss": 0.8939,
      "step": 12417
    },
    {
      "epoch": 2.3077494889425756,
      "grad_norm": 0.7862223386764526,
      "learning_rate": 7.628344662208899e-05,
      "loss": 0.8202,
      "step": 12418
    },
    {
      "epoch": 2.307935328005947,
      "grad_norm": 0.6856497526168823,
      "learning_rate": 7.626925434980382e-05,
      "loss": 0.7252,
      "step": 12419
    },
    {
      "epoch": 2.308121167069318,
      "grad_norm": 0.8154091835021973,
      "learning_rate": 7.625506258401071e-05,
      "loss": 0.8622,
      "step": 12420
    },
    {
      "epoch": 2.308307006132689,
      "grad_norm": 0.8779199719429016,
      "learning_rate": 7.624087132501252e-05,
      "loss": 1.0541,
      "step": 12421
    },
    {
      "epoch": 2.30849284519606,
      "grad_norm": 0.810685932636261,
      "learning_rate": 7.622668057311217e-05,
      "loss": 0.9654,
      "step": 12422
    },
    {
      "epoch": 2.3086786842594313,
      "grad_norm": 0.8009283542633057,
      "learning_rate": 7.621249032861249e-05,
      "loss": 1.0143,
      "step": 12423
    },
    {
      "epoch": 2.3088645233228027,
      "grad_norm": 0.7413025498390198,
      "learning_rate": 7.61983005918164e-05,
      "loss": 0.9063,
      "step": 12424
    },
    {
      "epoch": 2.3090503623861736,
      "grad_norm": 0.7271872162818909,
      "learning_rate": 7.618411136302675e-05,
      "loss": 0.7407,
      "step": 12425
    },
    {
      "epoch": 2.3092362014495444,
      "grad_norm": 0.972961962223053,
      "learning_rate": 7.61699226425464e-05,
      "loss": 0.9242,
      "step": 12426
    },
    {
      "epoch": 2.3094220405129158,
      "grad_norm": 0.756356954574585,
      "learning_rate": 7.615573443067812e-05,
      "loss": 0.9317,
      "step": 12427
    },
    {
      "epoch": 2.309607879576287,
      "grad_norm": 0.898516058921814,
      "learning_rate": 7.61415467277248e-05,
      "loss": 0.9378,
      "step": 12428
    },
    {
      "epoch": 2.309793718639658,
      "grad_norm": 1.0278419256210327,
      "learning_rate": 7.61273595339892e-05,
      "loss": 0.843,
      "step": 12429
    },
    {
      "epoch": 2.3099795577030293,
      "grad_norm": 0.9227488040924072,
      "learning_rate": 7.611317284977415e-05,
      "loss": 1.0285,
      "step": 12430
    },
    {
      "epoch": 2.3101653967664,
      "grad_norm": 0.8172624707221985,
      "learning_rate": 7.609898667538243e-05,
      "loss": 0.9617,
      "step": 12431
    },
    {
      "epoch": 2.3103512358297715,
      "grad_norm": 0.879382848739624,
      "learning_rate": 7.608480101111685e-05,
      "loss": 0.9802,
      "step": 12432
    },
    {
      "epoch": 2.3105370748931424,
      "grad_norm": 0.7681975364685059,
      "learning_rate": 7.60706158572801e-05,
      "loss": 0.9792,
      "step": 12433
    },
    {
      "epoch": 2.3107229139565137,
      "grad_norm": 0.8928554058074951,
      "learning_rate": 7.605643121417501e-05,
      "loss": 0.9939,
      "step": 12434
    },
    {
      "epoch": 2.3109087530198846,
      "grad_norm": 0.7671574950218201,
      "learning_rate": 7.60422470821043e-05,
      "loss": 0.8958,
      "step": 12435
    },
    {
      "epoch": 2.311094592083256,
      "grad_norm": 0.9336884021759033,
      "learning_rate": 7.602806346137075e-05,
      "loss": 0.7966,
      "step": 12436
    },
    {
      "epoch": 2.311280431146627,
      "grad_norm": 0.8644281625747681,
      "learning_rate": 7.601388035227702e-05,
      "loss": 1.1457,
      "step": 12437
    },
    {
      "epoch": 2.311466270209998,
      "grad_norm": 0.7973626852035522,
      "learning_rate": 7.599969775512587e-05,
      "loss": 0.9548,
      "step": 12438
    },
    {
      "epoch": 2.3116521092733695,
      "grad_norm": 1.0643774271011353,
      "learning_rate": 7.598551567021997e-05,
      "loss": 0.9208,
      "step": 12439
    },
    {
      "epoch": 2.3118379483367404,
      "grad_norm": 0.8160991668701172,
      "learning_rate": 7.597133409786203e-05,
      "loss": 0.8189,
      "step": 12440
    },
    {
      "epoch": 2.3120237874001113,
      "grad_norm": 0.7943512201309204,
      "learning_rate": 7.595715303835474e-05,
      "loss": 0.8455,
      "step": 12441
    },
    {
      "epoch": 2.3122096264634826,
      "grad_norm": 0.7578389048576355,
      "learning_rate": 7.594297249200076e-05,
      "loss": 0.933,
      "step": 12442
    },
    {
      "epoch": 2.312395465526854,
      "grad_norm": 0.7750895023345947,
      "learning_rate": 7.592879245910273e-05,
      "loss": 0.8472,
      "step": 12443
    },
    {
      "epoch": 2.312581304590225,
      "grad_norm": 0.81695556640625,
      "learning_rate": 7.591461293996329e-05,
      "loss": 0.9109,
      "step": 12444
    },
    {
      "epoch": 2.312767143653596,
      "grad_norm": 0.6965035796165466,
      "learning_rate": 7.590043393488512e-05,
      "loss": 0.9366,
      "step": 12445
    },
    {
      "epoch": 2.312952982716967,
      "grad_norm": 0.7417106032371521,
      "learning_rate": 7.588625544417087e-05,
      "loss": 0.7527,
      "step": 12446
    },
    {
      "epoch": 2.3131388217803384,
      "grad_norm": 0.8987021446228027,
      "learning_rate": 7.58720774681231e-05,
      "loss": 0.801,
      "step": 12447
    },
    {
      "epoch": 2.3133246608437092,
      "grad_norm": 0.8437244296073914,
      "learning_rate": 7.585790000704445e-05,
      "loss": 0.8314,
      "step": 12448
    },
    {
      "epoch": 2.3135104999070806,
      "grad_norm": 0.8858640789985657,
      "learning_rate": 7.584372306123746e-05,
      "loss": 0.9603,
      "step": 12449
    },
    {
      "epoch": 2.3136963389704515,
      "grad_norm": 0.8065061569213867,
      "learning_rate": 7.582954663100477e-05,
      "loss": 0.9499,
      "step": 12450
    },
    {
      "epoch": 2.313882178033823,
      "grad_norm": 0.7675348520278931,
      "learning_rate": 7.581537071664891e-05,
      "loss": 0.9969,
      "step": 12451
    },
    {
      "epoch": 2.3140680170971937,
      "grad_norm": 0.8798878192901611,
      "learning_rate": 7.58011953184725e-05,
      "loss": 0.9539,
      "step": 12452
    },
    {
      "epoch": 2.314253856160565,
      "grad_norm": 0.8737105131149292,
      "learning_rate": 7.5787020436778e-05,
      "loss": 0.9449,
      "step": 12453
    },
    {
      "epoch": 2.314439695223936,
      "grad_norm": 0.7421537041664124,
      "learning_rate": 7.577284607186798e-05,
      "loss": 0.9947,
      "step": 12454
    },
    {
      "epoch": 2.314625534287307,
      "grad_norm": 0.8042177557945251,
      "learning_rate": 7.575867222404503e-05,
      "loss": 0.9808,
      "step": 12455
    },
    {
      "epoch": 2.314811373350678,
      "grad_norm": 0.7402068376541138,
      "learning_rate": 7.574449889361162e-05,
      "loss": 0.7764,
      "step": 12456
    },
    {
      "epoch": 2.3149972124140494,
      "grad_norm": 0.8809178471565247,
      "learning_rate": 7.573032608087024e-05,
      "loss": 0.9017,
      "step": 12457
    },
    {
      "epoch": 2.3151830514774208,
      "grad_norm": 0.9781234264373779,
      "learning_rate": 7.571615378612342e-05,
      "loss": 1.0225,
      "step": 12458
    },
    {
      "epoch": 2.3153688905407916,
      "grad_norm": 0.7336781024932861,
      "learning_rate": 7.570198200967362e-05,
      "loss": 0.7536,
      "step": 12459
    },
    {
      "epoch": 2.315554729604163,
      "grad_norm": 0.8988581895828247,
      "learning_rate": 7.568781075182332e-05,
      "loss": 1.04,
      "step": 12460
    },
    {
      "epoch": 2.315740568667534,
      "grad_norm": 0.9049788117408752,
      "learning_rate": 7.567364001287496e-05,
      "loss": 0.8962,
      "step": 12461
    },
    {
      "epoch": 2.315926407730905,
      "grad_norm": 0.9063324928283691,
      "learning_rate": 7.565946979313105e-05,
      "loss": 0.8652,
      "step": 12462
    },
    {
      "epoch": 2.316112246794276,
      "grad_norm": 1.0230926275253296,
      "learning_rate": 7.564530009289395e-05,
      "loss": 0.8893,
      "step": 12463
    },
    {
      "epoch": 2.3162980858576474,
      "grad_norm": 0.8703257441520691,
      "learning_rate": 7.563113091246614e-05,
      "loss": 1.0691,
      "step": 12464
    },
    {
      "epoch": 2.3164839249210183,
      "grad_norm": 0.8154070377349854,
      "learning_rate": 7.561696225214995e-05,
      "loss": 0.8211,
      "step": 12465
    },
    {
      "epoch": 2.3166697639843896,
      "grad_norm": 0.68623948097229,
      "learning_rate": 7.560279411224795e-05,
      "loss": 0.5873,
      "step": 12466
    },
    {
      "epoch": 2.3168556030477605,
      "grad_norm": 0.8234708309173584,
      "learning_rate": 7.558862649306239e-05,
      "loss": 0.9179,
      "step": 12467
    },
    {
      "epoch": 2.317041442111132,
      "grad_norm": 0.8765814900398254,
      "learning_rate": 7.557445939489576e-05,
      "loss": 1.0854,
      "step": 12468
    },
    {
      "epoch": 2.3172272811745027,
      "grad_norm": 0.8760924935340881,
      "learning_rate": 7.556029281805034e-05,
      "loss": 1.1933,
      "step": 12469
    },
    {
      "epoch": 2.317413120237874,
      "grad_norm": 0.7699716687202454,
      "learning_rate": 7.554612676282854e-05,
      "loss": 0.854,
      "step": 12470
    },
    {
      "epoch": 2.317598959301245,
      "grad_norm": 0.7311826348304749,
      "learning_rate": 7.55319612295327e-05,
      "loss": 0.9411,
      "step": 12471
    },
    {
      "epoch": 2.3177847983646163,
      "grad_norm": 0.7913200855255127,
      "learning_rate": 7.551779621846517e-05,
      "loss": 1.0598,
      "step": 12472
    },
    {
      "epoch": 2.3179706374279876,
      "grad_norm": 0.6416594386100769,
      "learning_rate": 7.550363172992824e-05,
      "loss": 0.7361,
      "step": 12473
    },
    {
      "epoch": 2.3181564764913585,
      "grad_norm": 0.8257509469985962,
      "learning_rate": 7.548946776422428e-05,
      "loss": 1.0523,
      "step": 12474
    },
    {
      "epoch": 2.3183423155547294,
      "grad_norm": 0.9835213422775269,
      "learning_rate": 7.54753043216555e-05,
      "loss": 0.8885,
      "step": 12475
    },
    {
      "epoch": 2.3185281546181007,
      "grad_norm": 1.0134283304214478,
      "learning_rate": 7.546114140252433e-05,
      "loss": 0.9571,
      "step": 12476
    },
    {
      "epoch": 2.318713993681472,
      "grad_norm": 0.855441153049469,
      "learning_rate": 7.544697900713298e-05,
      "loss": 0.8827,
      "step": 12477
    },
    {
      "epoch": 2.318899832744843,
      "grad_norm": 0.8818331956863403,
      "learning_rate": 7.543281713578374e-05,
      "loss": 1.0948,
      "step": 12478
    },
    {
      "epoch": 2.3190856718082142,
      "grad_norm": 1.4963585138320923,
      "learning_rate": 7.541865578877882e-05,
      "loss": 1.4983,
      "step": 12479
    },
    {
      "epoch": 2.319271510871585,
      "grad_norm": 0.7784662246704102,
      "learning_rate": 7.540449496642056e-05,
      "loss": 0.9539,
      "step": 12480
    },
    {
      "epoch": 2.3194573499349564,
      "grad_norm": 0.7249751091003418,
      "learning_rate": 7.539033466901111e-05,
      "loss": 0.9508,
      "step": 12481
    },
    {
      "epoch": 2.3196431889983273,
      "grad_norm": 0.8944977521896362,
      "learning_rate": 7.537617489685276e-05,
      "loss": 0.912,
      "step": 12482
    },
    {
      "epoch": 2.3198290280616987,
      "grad_norm": 0.7689292430877686,
      "learning_rate": 7.536201565024767e-05,
      "loss": 0.8769,
      "step": 12483
    },
    {
      "epoch": 2.3200148671250695,
      "grad_norm": 0.9170524477958679,
      "learning_rate": 7.53478569294981e-05,
      "loss": 0.7936,
      "step": 12484
    },
    {
      "epoch": 2.320200706188441,
      "grad_norm": 0.7663047313690186,
      "learning_rate": 7.533369873490622e-05,
      "loss": 0.7094,
      "step": 12485
    },
    {
      "epoch": 2.3203865452518118,
      "grad_norm": 0.9506276845932007,
      "learning_rate": 7.531954106677416e-05,
      "loss": 0.7563,
      "step": 12486
    },
    {
      "epoch": 2.320572384315183,
      "grad_norm": 0.8071599006652832,
      "learning_rate": 7.530538392540418e-05,
      "loss": 0.8353,
      "step": 12487
    },
    {
      "epoch": 2.3207582233785544,
      "grad_norm": 0.87018883228302,
      "learning_rate": 7.52912273110984e-05,
      "loss": 1.0631,
      "step": 12488
    },
    {
      "epoch": 2.3209440624419253,
      "grad_norm": 0.9310821890830994,
      "learning_rate": 7.527707122415898e-05,
      "loss": 0.9587,
      "step": 12489
    },
    {
      "epoch": 2.321129901505296,
      "grad_norm": 0.8222516775131226,
      "learning_rate": 7.526291566488805e-05,
      "loss": 1.0424,
      "step": 12490
    },
    {
      "epoch": 2.3213157405686675,
      "grad_norm": 0.9288331270217896,
      "learning_rate": 7.524876063358773e-05,
      "loss": 0.7647,
      "step": 12491
    },
    {
      "epoch": 2.321501579632039,
      "grad_norm": 0.8477649688720703,
      "learning_rate": 7.523460613056015e-05,
      "loss": 0.8382,
      "step": 12492
    },
    {
      "epoch": 2.3216874186954097,
      "grad_norm": 0.9424648880958557,
      "learning_rate": 7.522045215610738e-05,
      "loss": 1.1071,
      "step": 12493
    },
    {
      "epoch": 2.321873257758781,
      "grad_norm": 0.7217926979064941,
      "learning_rate": 7.520629871053157e-05,
      "loss": 0.7972,
      "step": 12494
    },
    {
      "epoch": 2.322059096822152,
      "grad_norm": 0.7429036498069763,
      "learning_rate": 7.519214579413472e-05,
      "loss": 0.8802,
      "step": 12495
    },
    {
      "epoch": 2.3222449358855233,
      "grad_norm": 1.660508394241333,
      "learning_rate": 7.517799340721895e-05,
      "loss": 1.2279,
      "step": 12496
    },
    {
      "epoch": 2.322430774948894,
      "grad_norm": 0.7576091885566711,
      "learning_rate": 7.516384155008633e-05,
      "loss": 0.8146,
      "step": 12497
    },
    {
      "epoch": 2.3226166140122655,
      "grad_norm": 0.8176612257957458,
      "learning_rate": 7.51496902230389e-05,
      "loss": 0.889,
      "step": 12498
    },
    {
      "epoch": 2.3228024530756364,
      "grad_norm": 0.8361815810203552,
      "learning_rate": 7.513553942637868e-05,
      "loss": 0.8844,
      "step": 12499
    },
    {
      "epoch": 2.3229882921390077,
      "grad_norm": 0.8504658937454224,
      "learning_rate": 7.512138916040772e-05,
      "loss": 0.73,
      "step": 12500
    },
    {
      "epoch": 2.3231741312023786,
      "grad_norm": 0.8520758152008057,
      "learning_rate": 7.5107239425428e-05,
      "loss": 0.8937,
      "step": 12501
    },
    {
      "epoch": 2.32335997026575,
      "grad_norm": 0.8601714968681335,
      "learning_rate": 7.509309022174153e-05,
      "loss": 0.8883,
      "step": 12502
    },
    {
      "epoch": 2.323545809329121,
      "grad_norm": 0.8504960536956787,
      "learning_rate": 7.507894154965031e-05,
      "loss": 0.9509,
      "step": 12503
    },
    {
      "epoch": 2.323731648392492,
      "grad_norm": 0.7272476553916931,
      "learning_rate": 7.506479340945634e-05,
      "loss": 0.9195,
      "step": 12504
    },
    {
      "epoch": 2.323917487455863,
      "grad_norm": 0.834908127784729,
      "learning_rate": 7.505064580146154e-05,
      "loss": 0.8367,
      "step": 12505
    },
    {
      "epoch": 2.3241033265192343,
      "grad_norm": 0.7319338917732239,
      "learning_rate": 7.503649872596789e-05,
      "loss": 0.8302,
      "step": 12506
    },
    {
      "epoch": 2.3242891655826057,
      "grad_norm": 0.7511029243469238,
      "learning_rate": 7.502235218327731e-05,
      "loss": 0.7277,
      "step": 12507
    },
    {
      "epoch": 2.3244750046459766,
      "grad_norm": 0.8672081232070923,
      "learning_rate": 7.500820617369179e-05,
      "loss": 0.9919,
      "step": 12508
    },
    {
      "epoch": 2.324660843709348,
      "grad_norm": 0.742408812046051,
      "learning_rate": 7.499406069751323e-05,
      "loss": 0.8629,
      "step": 12509
    },
    {
      "epoch": 2.3248466827727188,
      "grad_norm": 1.0467314720153809,
      "learning_rate": 7.497991575504353e-05,
      "loss": 1.0455,
      "step": 12510
    },
    {
      "epoch": 2.32503252183609,
      "grad_norm": 0.8941596746444702,
      "learning_rate": 7.496577134658458e-05,
      "loss": 1.0436,
      "step": 12511
    },
    {
      "epoch": 2.325218360899461,
      "grad_norm": 0.8611307144165039,
      "learning_rate": 7.495162747243829e-05,
      "loss": 0.9668,
      "step": 12512
    },
    {
      "epoch": 2.3254041999628323,
      "grad_norm": 0.651538074016571,
      "learning_rate": 7.493748413290653e-05,
      "loss": 0.8183,
      "step": 12513
    },
    {
      "epoch": 2.325590039026203,
      "grad_norm": 0.9411389827728271,
      "learning_rate": 7.492334132829115e-05,
      "loss": 0.8401,
      "step": 12514
    },
    {
      "epoch": 2.3257758780895745,
      "grad_norm": 0.7687920928001404,
      "learning_rate": 7.490919905889403e-05,
      "loss": 0.9558,
      "step": 12515
    },
    {
      "epoch": 2.3259617171529454,
      "grad_norm": 0.8409010767936707,
      "learning_rate": 7.489505732501699e-05,
      "loss": 1.0783,
      "step": 12516
    },
    {
      "epoch": 2.3261475562163167,
      "grad_norm": 0.8515502214431763,
      "learning_rate": 7.488091612696182e-05,
      "loss": 0.9425,
      "step": 12517
    },
    {
      "epoch": 2.3263333952796876,
      "grad_norm": 0.7506030797958374,
      "learning_rate": 7.486677546503044e-05,
      "loss": 0.8782,
      "step": 12518
    },
    {
      "epoch": 2.326519234343059,
      "grad_norm": 0.6447235345840454,
      "learning_rate": 7.485263533952461e-05,
      "loss": 0.7478,
      "step": 12519
    },
    {
      "epoch": 2.32670507340643,
      "grad_norm": 0.8545085191726685,
      "learning_rate": 7.483849575074612e-05,
      "loss": 0.8679,
      "step": 12520
    },
    {
      "epoch": 2.326890912469801,
      "grad_norm": 0.7526487708091736,
      "learning_rate": 7.482435669899677e-05,
      "loss": 0.9467,
      "step": 12521
    },
    {
      "epoch": 2.3270767515331725,
      "grad_norm": 0.7747771739959717,
      "learning_rate": 7.481021818457831e-05,
      "loss": 0.9128,
      "step": 12522
    },
    {
      "epoch": 2.3272625905965434,
      "grad_norm": 1.074554681777954,
      "learning_rate": 7.479608020779252e-05,
      "loss": 0.8633,
      "step": 12523
    },
    {
      "epoch": 2.3274484296599143,
      "grad_norm": 0.5868602991104126,
      "learning_rate": 7.478194276894116e-05,
      "loss": 0.5134,
      "step": 12524
    },
    {
      "epoch": 2.3276342687232856,
      "grad_norm": 0.8135141730308533,
      "learning_rate": 7.476780586832594e-05,
      "loss": 0.8191,
      "step": 12525
    },
    {
      "epoch": 2.327820107786657,
      "grad_norm": 0.7354833483695984,
      "learning_rate": 7.475366950624858e-05,
      "loss": 0.9799,
      "step": 12526
    },
    {
      "epoch": 2.328005946850028,
      "grad_norm": 0.9085583090782166,
      "learning_rate": 7.473953368301086e-05,
      "loss": 1.0275,
      "step": 12527
    },
    {
      "epoch": 2.328191785913399,
      "grad_norm": 0.8576337695121765,
      "learning_rate": 7.47253983989144e-05,
      "loss": 1.0876,
      "step": 12528
    },
    {
      "epoch": 2.32837762497677,
      "grad_norm": 0.6999117136001587,
      "learning_rate": 7.471126365426095e-05,
      "loss": 0.871,
      "step": 12529
    },
    {
      "epoch": 2.3285634640401414,
      "grad_norm": 0.9206585884094238,
      "learning_rate": 7.469712944935222e-05,
      "loss": 0.8432,
      "step": 12530
    },
    {
      "epoch": 2.3287493031035122,
      "grad_norm": 0.8157764673233032,
      "learning_rate": 7.46829957844898e-05,
      "loss": 0.9118,
      "step": 12531
    },
    {
      "epoch": 2.3289351421668836,
      "grad_norm": 0.9429844617843628,
      "learning_rate": 7.466886265997543e-05,
      "loss": 1.0772,
      "step": 12532
    },
    {
      "epoch": 2.3291209812302545,
      "grad_norm": 0.6998193860054016,
      "learning_rate": 7.465473007611068e-05,
      "loss": 0.7093,
      "step": 12533
    },
    {
      "epoch": 2.329306820293626,
      "grad_norm": 1.0119681358337402,
      "learning_rate": 7.464059803319725e-05,
      "loss": 0.9456,
      "step": 12534
    },
    {
      "epoch": 2.3294926593569967,
      "grad_norm": 0.9467625021934509,
      "learning_rate": 7.462646653153671e-05,
      "loss": 1.052,
      "step": 12535
    },
    {
      "epoch": 2.329678498420368,
      "grad_norm": 0.8921098709106445,
      "learning_rate": 7.46123355714307e-05,
      "loss": 0.8632,
      "step": 12536
    },
    {
      "epoch": 2.329864337483739,
      "grad_norm": 0.8844832181930542,
      "learning_rate": 7.459820515318085e-05,
      "loss": 0.8857,
      "step": 12537
    },
    {
      "epoch": 2.33005017654711,
      "grad_norm": 1.073271632194519,
      "learning_rate": 7.458407527708868e-05,
      "loss": 0.8916,
      "step": 12538
    },
    {
      "epoch": 2.330236015610481,
      "grad_norm": 0.814613938331604,
      "learning_rate": 7.45699459434558e-05,
      "loss": 0.942,
      "step": 12539
    },
    {
      "epoch": 2.3304218546738524,
      "grad_norm": 0.8630864024162292,
      "learning_rate": 7.455581715258385e-05,
      "loss": 1.1416,
      "step": 12540
    },
    {
      "epoch": 2.3306076937372238,
      "grad_norm": 0.9294757843017578,
      "learning_rate": 7.454168890477427e-05,
      "loss": 1.0254,
      "step": 12541
    },
    {
      "epoch": 2.3307935328005946,
      "grad_norm": 0.8275464177131653,
      "learning_rate": 7.452756120032867e-05,
      "loss": 0.6494,
      "step": 12542
    },
    {
      "epoch": 2.330979371863966,
      "grad_norm": 0.7378981113433838,
      "learning_rate": 7.451343403954856e-05,
      "loss": 0.595,
      "step": 12543
    },
    {
      "epoch": 2.331165210927337,
      "grad_norm": 0.6890910267829895,
      "learning_rate": 7.449930742273547e-05,
      "loss": 0.739,
      "step": 12544
    },
    {
      "epoch": 2.331351049990708,
      "grad_norm": 0.77701735496521,
      "learning_rate": 7.448518135019088e-05,
      "loss": 0.8901,
      "step": 12545
    },
    {
      "epoch": 2.331536889054079,
      "grad_norm": 0.8840491771697998,
      "learning_rate": 7.447105582221631e-05,
      "loss": 0.8107,
      "step": 12546
    },
    {
      "epoch": 2.3317227281174504,
      "grad_norm": 0.7704744338989258,
      "learning_rate": 7.445693083911327e-05,
      "loss": 0.8986,
      "step": 12547
    },
    {
      "epoch": 2.3319085671808213,
      "grad_norm": 0.7400922775268555,
      "learning_rate": 7.444280640118319e-05,
      "loss": 0.8463,
      "step": 12548
    },
    {
      "epoch": 2.3320944062441926,
      "grad_norm": 0.7758222818374634,
      "learning_rate": 7.442868250872753e-05,
      "loss": 0.7794,
      "step": 12549
    },
    {
      "epoch": 2.3322802453075635,
      "grad_norm": 0.7377352118492126,
      "learning_rate": 7.44145591620478e-05,
      "loss": 0.8337,
      "step": 12550
    },
    {
      "epoch": 2.332466084370935,
      "grad_norm": 0.9961763620376587,
      "learning_rate": 7.440043636144536e-05,
      "loss": 0.8354,
      "step": 12551
    },
    {
      "epoch": 2.3326519234343057,
      "grad_norm": 0.8004982471466064,
      "learning_rate": 7.438631410722172e-05,
      "loss": 0.8029,
      "step": 12552
    },
    {
      "epoch": 2.332837762497677,
      "grad_norm": 0.8638790845870972,
      "learning_rate": 7.43721923996782e-05,
      "loss": 0.7802,
      "step": 12553
    },
    {
      "epoch": 2.333023601561048,
      "grad_norm": 0.8142632842063904,
      "learning_rate": 7.435807123911631e-05,
      "loss": 0.8276,
      "step": 12554
    },
    {
      "epoch": 2.3332094406244193,
      "grad_norm": 0.8604399561882019,
      "learning_rate": 7.434395062583734e-05,
      "loss": 0.8149,
      "step": 12555
    },
    {
      "epoch": 2.3333952796877906,
      "grad_norm": 0.7470417618751526,
      "learning_rate": 7.432983056014273e-05,
      "loss": 0.7667,
      "step": 12556
    },
    {
      "epoch": 2.3335811187511615,
      "grad_norm": 0.8007526397705078,
      "learning_rate": 7.431571104233385e-05,
      "loss": 0.8139,
      "step": 12557
    },
    {
      "epoch": 2.333766957814533,
      "grad_norm": 0.8846230506896973,
      "learning_rate": 7.430159207271203e-05,
      "loss": 0.8382,
      "step": 12558
    },
    {
      "epoch": 2.3339527968779037,
      "grad_norm": 0.7674434185028076,
      "learning_rate": 7.42874736515786e-05,
      "loss": 1.01,
      "step": 12559
    },
    {
      "epoch": 2.334138635941275,
      "grad_norm": 0.7947673797607422,
      "learning_rate": 7.427335577923496e-05,
      "loss": 0.9574,
      "step": 12560
    },
    {
      "epoch": 2.334324475004646,
      "grad_norm": 0.8595078587532043,
      "learning_rate": 7.425923845598237e-05,
      "loss": 0.9556,
      "step": 12561
    },
    {
      "epoch": 2.3345103140680172,
      "grad_norm": 0.9428784251213074,
      "learning_rate": 7.424512168212219e-05,
      "loss": 0.7944,
      "step": 12562
    },
    {
      "epoch": 2.334696153131388,
      "grad_norm": 0.8632462620735168,
      "learning_rate": 7.423100545795565e-05,
      "loss": 0.9847,
      "step": 12563
    },
    {
      "epoch": 2.3348819921947594,
      "grad_norm": 0.90449059009552,
      "learning_rate": 7.421688978378411e-05,
      "loss": 0.9834,
      "step": 12564
    },
    {
      "epoch": 2.3350678312581303,
      "grad_norm": 0.9151463508605957,
      "learning_rate": 7.42027746599088e-05,
      "loss": 0.9398,
      "step": 12565
    },
    {
      "epoch": 2.3352536703215017,
      "grad_norm": 0.9773748517036438,
      "learning_rate": 7.418866008663099e-05,
      "loss": 1.0023,
      "step": 12566
    },
    {
      "epoch": 2.3354395093848725,
      "grad_norm": 0.7529578804969788,
      "learning_rate": 7.417454606425196e-05,
      "loss": 0.8994,
      "step": 12567
    },
    {
      "epoch": 2.335625348448244,
      "grad_norm": 0.9450651407241821,
      "learning_rate": 7.41604325930729e-05,
      "loss": 1.1175,
      "step": 12568
    },
    {
      "epoch": 2.3358111875116148,
      "grad_norm": 0.8100607395172119,
      "learning_rate": 7.414631967339509e-05,
      "loss": 0.9399,
      "step": 12569
    },
    {
      "epoch": 2.335997026574986,
      "grad_norm": 1.0437417030334473,
      "learning_rate": 7.413220730551966e-05,
      "loss": 0.9058,
      "step": 12570
    },
    {
      "epoch": 2.3361828656383574,
      "grad_norm": 0.7519281506538391,
      "learning_rate": 7.411809548974792e-05,
      "loss": 0.9539,
      "step": 12571
    },
    {
      "epoch": 2.3363687047017283,
      "grad_norm": 1.2798535823822021,
      "learning_rate": 7.410398422638102e-05,
      "loss": 1.2466,
      "step": 12572
    },
    {
      "epoch": 2.336554543765099,
      "grad_norm": 0.7466521859169006,
      "learning_rate": 7.408987351572013e-05,
      "loss": 1.0244,
      "step": 12573
    },
    {
      "epoch": 2.3367403828284705,
      "grad_norm": 0.6776727437973022,
      "learning_rate": 7.407576335806644e-05,
      "loss": 0.7165,
      "step": 12574
    },
    {
      "epoch": 2.336926221891842,
      "grad_norm": 0.9132217168807983,
      "learning_rate": 7.406165375372107e-05,
      "loss": 1.2458,
      "step": 12575
    },
    {
      "epoch": 2.3371120609552127,
      "grad_norm": 0.8114913105964661,
      "learning_rate": 7.40475447029852e-05,
      "loss": 0.9963,
      "step": 12576
    },
    {
      "epoch": 2.337297900018584,
      "grad_norm": 1.0077768564224243,
      "learning_rate": 7.403343620615997e-05,
      "loss": 0.946,
      "step": 12577
    },
    {
      "epoch": 2.337483739081955,
      "grad_norm": 0.749036431312561,
      "learning_rate": 7.401932826354646e-05,
      "loss": 0.676,
      "step": 12578
    },
    {
      "epoch": 2.3376695781453263,
      "grad_norm": 0.7417488694190979,
      "learning_rate": 7.400522087544582e-05,
      "loss": 0.9087,
      "step": 12579
    },
    {
      "epoch": 2.337855417208697,
      "grad_norm": 0.8185071349143982,
      "learning_rate": 7.399111404215909e-05,
      "loss": 0.9881,
      "step": 12580
    },
    {
      "epoch": 2.3380412562720685,
      "grad_norm": 0.8285645842552185,
      "learning_rate": 7.397700776398741e-05,
      "loss": 1.1504,
      "step": 12581
    },
    {
      "epoch": 2.3382270953354394,
      "grad_norm": 0.7775911092758179,
      "learning_rate": 7.396290204123188e-05,
      "loss": 0.97,
      "step": 12582
    },
    {
      "epoch": 2.3384129343988107,
      "grad_norm": 0.7964914441108704,
      "learning_rate": 7.39487968741935e-05,
      "loss": 1.1304,
      "step": 12583
    },
    {
      "epoch": 2.3385987734621816,
      "grad_norm": 0.8669666051864624,
      "learning_rate": 7.393469226317335e-05,
      "loss": 0.8356,
      "step": 12584
    },
    {
      "epoch": 2.338784612525553,
      "grad_norm": 0.9136351346969604,
      "learning_rate": 7.392058820847245e-05,
      "loss": 0.7691,
      "step": 12585
    },
    {
      "epoch": 2.338970451588924,
      "grad_norm": 0.8523462414741516,
      "learning_rate": 7.390648471039185e-05,
      "loss": 0.8044,
      "step": 12586
    },
    {
      "epoch": 2.339156290652295,
      "grad_norm": 0.9019622206687927,
      "learning_rate": 7.389238176923258e-05,
      "loss": 0.8436,
      "step": 12587
    },
    {
      "epoch": 2.339342129715666,
      "grad_norm": 0.8432104587554932,
      "learning_rate": 7.387827938529559e-05,
      "loss": 0.7952,
      "step": 12588
    },
    {
      "epoch": 2.3395279687790373,
      "grad_norm": 0.714282214641571,
      "learning_rate": 7.386417755888191e-05,
      "loss": 0.7451,
      "step": 12589
    },
    {
      "epoch": 2.3397138078424087,
      "grad_norm": 0.8890557289123535,
      "learning_rate": 7.38500762902925e-05,
      "loss": 0.8865,
      "step": 12590
    },
    {
      "epoch": 2.3398996469057796,
      "grad_norm": 0.6751204133033752,
      "learning_rate": 7.383597557982833e-05,
      "loss": 0.8756,
      "step": 12591
    },
    {
      "epoch": 2.340085485969151,
      "grad_norm": 0.7461575865745544,
      "learning_rate": 7.382187542779038e-05,
      "loss": 0.8794,
      "step": 12592
    },
    {
      "epoch": 2.3402713250325218,
      "grad_norm": 0.8565900325775146,
      "learning_rate": 7.380777583447957e-05,
      "loss": 1.0726,
      "step": 12593
    },
    {
      "epoch": 2.340457164095893,
      "grad_norm": 0.7575398087501526,
      "learning_rate": 7.379367680019685e-05,
      "loss": 0.9309,
      "step": 12594
    },
    {
      "epoch": 2.340643003159264,
      "grad_norm": 1.0048532485961914,
      "learning_rate": 7.377957832524311e-05,
      "loss": 0.9448,
      "step": 12595
    },
    {
      "epoch": 2.3408288422226353,
      "grad_norm": 0.8079115152359009,
      "learning_rate": 7.376548040991927e-05,
      "loss": 0.8584,
      "step": 12596
    },
    {
      "epoch": 2.341014681286006,
      "grad_norm": 0.8296436071395874,
      "learning_rate": 7.375138305452624e-05,
      "loss": 0.9667,
      "step": 12597
    },
    {
      "epoch": 2.3412005203493775,
      "grad_norm": 0.8004963994026184,
      "learning_rate": 7.37372862593649e-05,
      "loss": 0.9276,
      "step": 12598
    },
    {
      "epoch": 2.3413863594127484,
      "grad_norm": 0.7355859279632568,
      "learning_rate": 7.372319002473613e-05,
      "loss": 0.7174,
      "step": 12599
    },
    {
      "epoch": 2.3415721984761197,
      "grad_norm": 0.8082651495933533,
      "learning_rate": 7.370909435094074e-05,
      "loss": 0.9901,
      "step": 12600
    },
    {
      "epoch": 2.3417580375394906,
      "grad_norm": 0.893548846244812,
      "learning_rate": 7.36949992382796e-05,
      "loss": 1.0624,
      "step": 12601
    },
    {
      "epoch": 2.341943876602862,
      "grad_norm": 0.8404073119163513,
      "learning_rate": 7.368090468705361e-05,
      "loss": 0.8835,
      "step": 12602
    },
    {
      "epoch": 2.342129715666233,
      "grad_norm": 0.8230233192443848,
      "learning_rate": 7.366681069756352e-05,
      "loss": 1.0559,
      "step": 12603
    },
    {
      "epoch": 2.342315554729604,
      "grad_norm": 1.057384729385376,
      "learning_rate": 7.365271727011019e-05,
      "loss": 1.1518,
      "step": 12604
    },
    {
      "epoch": 2.3425013937929755,
      "grad_norm": 0.9644550681114197,
      "learning_rate": 7.363862440499436e-05,
      "loss": 0.8966,
      "step": 12605
    },
    {
      "epoch": 2.3426872328563464,
      "grad_norm": 0.7263677716255188,
      "learning_rate": 7.362453210251686e-05,
      "loss": 0.7729,
      "step": 12606
    },
    {
      "epoch": 2.3428730719197173,
      "grad_norm": 0.7990922331809998,
      "learning_rate": 7.361044036297849e-05,
      "loss": 0.9411,
      "step": 12607
    },
    {
      "epoch": 2.3430589109830886,
      "grad_norm": 0.7927680015563965,
      "learning_rate": 7.359634918667996e-05,
      "loss": 0.891,
      "step": 12608
    },
    {
      "epoch": 2.34324475004646,
      "grad_norm": 0.8109414577484131,
      "learning_rate": 7.358225857392207e-05,
      "loss": 1.0352,
      "step": 12609
    },
    {
      "epoch": 2.343430589109831,
      "grad_norm": 0.8968274593353271,
      "learning_rate": 7.35681685250055e-05,
      "loss": 0.9016,
      "step": 12610
    },
    {
      "epoch": 2.343616428173202,
      "grad_norm": 0.8033596873283386,
      "learning_rate": 7.355407904023103e-05,
      "loss": 0.8873,
      "step": 12611
    },
    {
      "epoch": 2.343802267236573,
      "grad_norm": 1.0283641815185547,
      "learning_rate": 7.353999011989935e-05,
      "loss": 1.1681,
      "step": 12612
    },
    {
      "epoch": 2.3439881062999444,
      "grad_norm": 0.7543270587921143,
      "learning_rate": 7.352590176431118e-05,
      "loss": 0.6605,
      "step": 12613
    },
    {
      "epoch": 2.3441739453633152,
      "grad_norm": 0.8932667374610901,
      "learning_rate": 7.351181397376724e-05,
      "loss": 0.895,
      "step": 12614
    },
    {
      "epoch": 2.3443597844266866,
      "grad_norm": 0.8471357822418213,
      "learning_rate": 7.349772674856812e-05,
      "loss": 1.0288,
      "step": 12615
    },
    {
      "epoch": 2.3445456234900575,
      "grad_norm": 0.865290105342865,
      "learning_rate": 7.348364008901456e-05,
      "loss": 0.7652,
      "step": 12616
    },
    {
      "epoch": 2.344731462553429,
      "grad_norm": 0.8844984769821167,
      "learning_rate": 7.346955399540723e-05,
      "loss": 0.8963,
      "step": 12617
    },
    {
      "epoch": 2.3449173016167997,
      "grad_norm": 0.8787597417831421,
      "learning_rate": 7.34554684680467e-05,
      "loss": 1.081,
      "step": 12618
    },
    {
      "epoch": 2.345103140680171,
      "grad_norm": 0.7446838021278381,
      "learning_rate": 7.344138350723369e-05,
      "loss": 0.9378,
      "step": 12619
    },
    {
      "epoch": 2.3452889797435423,
      "grad_norm": 0.8385235667228699,
      "learning_rate": 7.342729911326873e-05,
      "loss": 0.9505,
      "step": 12620
    },
    {
      "epoch": 2.345474818806913,
      "grad_norm": 0.6965145468711853,
      "learning_rate": 7.34132152864525e-05,
      "loss": 0.7226,
      "step": 12621
    },
    {
      "epoch": 2.345660657870284,
      "grad_norm": 0.8368469476699829,
      "learning_rate": 7.339913202708552e-05,
      "loss": 0.8274,
      "step": 12622
    },
    {
      "epoch": 2.3458464969336554,
      "grad_norm": 0.7920130491256714,
      "learning_rate": 7.338504933546844e-05,
      "loss": 0.8632,
      "step": 12623
    },
    {
      "epoch": 2.3460323359970268,
      "grad_norm": 0.7846779227256775,
      "learning_rate": 7.337096721190182e-05,
      "loss": 0.9491,
      "step": 12624
    },
    {
      "epoch": 2.3462181750603976,
      "grad_norm": 0.8140245676040649,
      "learning_rate": 7.335688565668621e-05,
      "loss": 0.8609,
      "step": 12625
    },
    {
      "epoch": 2.346404014123769,
      "grad_norm": 0.7807095646858215,
      "learning_rate": 7.334280467012214e-05,
      "loss": 0.8984,
      "step": 12626
    },
    {
      "epoch": 2.34658985318714,
      "grad_norm": 0.8692532181739807,
      "learning_rate": 7.332872425251018e-05,
      "loss": 1.0086,
      "step": 12627
    },
    {
      "epoch": 2.346775692250511,
      "grad_norm": 0.7482457160949707,
      "learning_rate": 7.331464440415082e-05,
      "loss": 0.8269,
      "step": 12628
    },
    {
      "epoch": 2.346961531313882,
      "grad_norm": 0.9138236045837402,
      "learning_rate": 7.330056512534458e-05,
      "loss": 1.0416,
      "step": 12629
    },
    {
      "epoch": 2.3471473703772534,
      "grad_norm": 0.8924627304077148,
      "learning_rate": 7.328648641639195e-05,
      "loss": 0.8862,
      "step": 12630
    },
    {
      "epoch": 2.3473332094406243,
      "grad_norm": 0.8058710098266602,
      "learning_rate": 7.327240827759345e-05,
      "loss": 0.7509,
      "step": 12631
    },
    {
      "epoch": 2.3475190485039956,
      "grad_norm": 0.9041599631309509,
      "learning_rate": 7.32583307092495e-05,
      "loss": 0.7448,
      "step": 12632
    },
    {
      "epoch": 2.3477048875673665,
      "grad_norm": 0.8715543150901794,
      "learning_rate": 7.324425371166057e-05,
      "loss": 1.1062,
      "step": 12633
    },
    {
      "epoch": 2.347890726630738,
      "grad_norm": 0.7531129717826843,
      "learning_rate": 7.323017728512717e-05,
      "loss": 0.8701,
      "step": 12634
    },
    {
      "epoch": 2.3480765656941087,
      "grad_norm": 1.0183310508728027,
      "learning_rate": 7.32161014299497e-05,
      "loss": 0.8623,
      "step": 12635
    },
    {
      "epoch": 2.34826240475748,
      "grad_norm": 0.7323870658874512,
      "learning_rate": 7.320202614642856e-05,
      "loss": 0.973,
      "step": 12636
    },
    {
      "epoch": 2.348448243820851,
      "grad_norm": 1.0801042318344116,
      "learning_rate": 7.31879514348642e-05,
      "loss": 0.9972,
      "step": 12637
    },
    {
      "epoch": 2.3486340828842223,
      "grad_norm": 0.7815359234809875,
      "learning_rate": 7.3173877295557e-05,
      "loss": 0.9521,
      "step": 12638
    },
    {
      "epoch": 2.3488199219475936,
      "grad_norm": 0.7192123532295227,
      "learning_rate": 7.315980372880738e-05,
      "loss": 0.9116,
      "step": 12639
    },
    {
      "epoch": 2.3490057610109645,
      "grad_norm": 0.7962202429771423,
      "learning_rate": 7.314573073491566e-05,
      "loss": 0.6876,
      "step": 12640
    },
    {
      "epoch": 2.349191600074336,
      "grad_norm": 0.9301812052726746,
      "learning_rate": 7.313165831418225e-05,
      "loss": 1.0148,
      "step": 12641
    },
    {
      "epoch": 2.3493774391377067,
      "grad_norm": 0.751960277557373,
      "learning_rate": 7.311758646690748e-05,
      "loss": 0.8908,
      "step": 12642
    },
    {
      "epoch": 2.349563278201078,
      "grad_norm": 0.9244655966758728,
      "learning_rate": 7.310351519339165e-05,
      "loss": 0.9406,
      "step": 12643
    },
    {
      "epoch": 2.349749117264449,
      "grad_norm": 0.8816783428192139,
      "learning_rate": 7.308944449393518e-05,
      "loss": 1.0295,
      "step": 12644
    },
    {
      "epoch": 2.3499349563278202,
      "grad_norm": 0.8948429226875305,
      "learning_rate": 7.307537436883833e-05,
      "loss": 0.8946,
      "step": 12645
    },
    {
      "epoch": 2.350120795391191,
      "grad_norm": 0.5853556990623474,
      "learning_rate": 7.306130481840139e-05,
      "loss": 0.5547,
      "step": 12646
    },
    {
      "epoch": 2.3503066344545624,
      "grad_norm": 0.7878490686416626,
      "learning_rate": 7.30472358429247e-05,
      "loss": 0.9347,
      "step": 12647
    },
    {
      "epoch": 2.3504924735179333,
      "grad_norm": 0.8565953373908997,
      "learning_rate": 7.303316744270849e-05,
      "loss": 0.8626,
      "step": 12648
    },
    {
      "epoch": 2.3506783125813047,
      "grad_norm": 0.8676490783691406,
      "learning_rate": 7.301909961805306e-05,
      "loss": 0.9516,
      "step": 12649
    },
    {
      "epoch": 2.3508641516446755,
      "grad_norm": 0.8519862294197083,
      "learning_rate": 7.300503236925864e-05,
      "loss": 0.9389,
      "step": 12650
    },
    {
      "epoch": 2.351049990708047,
      "grad_norm": 0.7775579690933228,
      "learning_rate": 7.29909656966255e-05,
      "loss": 0.7831,
      "step": 12651
    },
    {
      "epoch": 2.3512358297714178,
      "grad_norm": 0.8417463302612305,
      "learning_rate": 7.29768996004538e-05,
      "loss": 0.918,
      "step": 12652
    },
    {
      "epoch": 2.351421668834789,
      "grad_norm": 0.7603926062583923,
      "learning_rate": 7.296283408104385e-05,
      "loss": 1.0112,
      "step": 12653
    },
    {
      "epoch": 2.3516075078981604,
      "grad_norm": 0.6868374943733215,
      "learning_rate": 7.294876913869573e-05,
      "loss": 0.6459,
      "step": 12654
    },
    {
      "epoch": 2.3517933469615313,
      "grad_norm": 0.8233698010444641,
      "learning_rate": 7.293470477370979e-05,
      "loss": 1.1302,
      "step": 12655
    },
    {
      "epoch": 2.351979186024902,
      "grad_norm": 0.9082133173942566,
      "learning_rate": 7.29206409863861e-05,
      "loss": 1.0395,
      "step": 12656
    },
    {
      "epoch": 2.3521650250882735,
      "grad_norm": 0.7943674921989441,
      "learning_rate": 7.290657777702488e-05,
      "loss": 0.6356,
      "step": 12657
    },
    {
      "epoch": 2.352350864151645,
      "grad_norm": 0.9202367067337036,
      "learning_rate": 7.289251514592625e-05,
      "loss": 1.0519,
      "step": 12658
    },
    {
      "epoch": 2.3525367032150157,
      "grad_norm": 0.8511589169502258,
      "learning_rate": 7.287845309339038e-05,
      "loss": 1.1179,
      "step": 12659
    },
    {
      "epoch": 2.352722542278387,
      "grad_norm": 0.796493411064148,
      "learning_rate": 7.286439161971737e-05,
      "loss": 0.8966,
      "step": 12660
    },
    {
      "epoch": 2.352908381341758,
      "grad_norm": 0.7692650556564331,
      "learning_rate": 7.285033072520736e-05,
      "loss": 0.9399,
      "step": 12661
    },
    {
      "epoch": 2.3530942204051293,
      "grad_norm": 0.8270449042320251,
      "learning_rate": 7.283627041016043e-05,
      "loss": 0.9297,
      "step": 12662
    },
    {
      "epoch": 2.3532800594685,
      "grad_norm": 1.3305835723876953,
      "learning_rate": 7.282221067487673e-05,
      "loss": 0.759,
      "step": 12663
    },
    {
      "epoch": 2.3534658985318715,
      "grad_norm": 0.8420296311378479,
      "learning_rate": 7.280815151965624e-05,
      "loss": 0.9815,
      "step": 12664
    },
    {
      "epoch": 2.3536517375952424,
      "grad_norm": 0.7621555924415588,
      "learning_rate": 7.279409294479915e-05,
      "loss": 0.8782,
      "step": 12665
    },
    {
      "epoch": 2.3538375766586137,
      "grad_norm": 0.8693684339523315,
      "learning_rate": 7.278003495060543e-05,
      "loss": 0.915,
      "step": 12666
    },
    {
      "epoch": 2.3540234157219846,
      "grad_norm": 0.8019158840179443,
      "learning_rate": 7.27659775373752e-05,
      "loss": 0.7045,
      "step": 12667
    },
    {
      "epoch": 2.354209254785356,
      "grad_norm": 0.8997321724891663,
      "learning_rate": 7.27519207054084e-05,
      "loss": 0.8401,
      "step": 12668
    },
    {
      "epoch": 2.3543950938487272,
      "grad_norm": 0.7859323024749756,
      "learning_rate": 7.273786445500513e-05,
      "loss": 0.7796,
      "step": 12669
    },
    {
      "epoch": 2.354580932912098,
      "grad_norm": 0.8584409952163696,
      "learning_rate": 7.272380878646532e-05,
      "loss": 1.0528,
      "step": 12670
    },
    {
      "epoch": 2.354766771975469,
      "grad_norm": 0.8414707779884338,
      "learning_rate": 7.270975370008903e-05,
      "loss": 1.0077,
      "step": 12671
    },
    {
      "epoch": 2.3549526110388403,
      "grad_norm": 0.8825092911720276,
      "learning_rate": 7.26956991961762e-05,
      "loss": 1.1716,
      "step": 12672
    },
    {
      "epoch": 2.3551384501022117,
      "grad_norm": 0.8332867622375488,
      "learning_rate": 7.268164527502683e-05,
      "loss": 0.8462,
      "step": 12673
    },
    {
      "epoch": 2.3553242891655826,
      "grad_norm": 0.7624945044517517,
      "learning_rate": 7.266759193694084e-05,
      "loss": 0.696,
      "step": 12674
    },
    {
      "epoch": 2.355510128228954,
      "grad_norm": 0.8629702925682068,
      "learning_rate": 7.265353918221817e-05,
      "loss": 1.1784,
      "step": 12675
    },
    {
      "epoch": 2.3556959672923248,
      "grad_norm": 0.6932694315910339,
      "learning_rate": 7.263948701115881e-05,
      "loss": 0.8929,
      "step": 12676
    },
    {
      "epoch": 2.355881806355696,
      "grad_norm": 0.7592587471008301,
      "learning_rate": 7.262543542406267e-05,
      "loss": 0.8846,
      "step": 12677
    },
    {
      "epoch": 2.356067645419067,
      "grad_norm": 0.7431928515434265,
      "learning_rate": 7.261138442122962e-05,
      "loss": 0.8915,
      "step": 12678
    },
    {
      "epoch": 2.3562534844824383,
      "grad_norm": 0.9140413999557495,
      "learning_rate": 7.259733400295956e-05,
      "loss": 1.0119,
      "step": 12679
    },
    {
      "epoch": 2.356439323545809,
      "grad_norm": 0.7461313009262085,
      "learning_rate": 7.258328416955238e-05,
      "loss": 0.9977,
      "step": 12680
    },
    {
      "epoch": 2.3566251626091805,
      "grad_norm": 0.9505020976066589,
      "learning_rate": 7.256923492130797e-05,
      "loss": 0.8069,
      "step": 12681
    },
    {
      "epoch": 2.3568110016725514,
      "grad_norm": 0.8287774920463562,
      "learning_rate": 7.255518625852615e-05,
      "loss": 0.9404,
      "step": 12682
    },
    {
      "epoch": 2.3569968407359227,
      "grad_norm": 0.8033880591392517,
      "learning_rate": 7.25411381815068e-05,
      "loss": 0.9273,
      "step": 12683
    },
    {
      "epoch": 2.3571826797992936,
      "grad_norm": 0.7415333986282349,
      "learning_rate": 7.252709069054971e-05,
      "loss": 0.8007,
      "step": 12684
    },
    {
      "epoch": 2.357368518862665,
      "grad_norm": 0.8875011801719666,
      "learning_rate": 7.251304378595471e-05,
      "loss": 0.8221,
      "step": 12685
    },
    {
      "epoch": 2.357554357926036,
      "grad_norm": 0.8783387541770935,
      "learning_rate": 7.249899746802163e-05,
      "loss": 0.9782,
      "step": 12686
    },
    {
      "epoch": 2.357740196989407,
      "grad_norm": 0.7870514988899231,
      "learning_rate": 7.248495173705027e-05,
      "loss": 0.869,
      "step": 12687
    },
    {
      "epoch": 2.3579260360527785,
      "grad_norm": 0.8032506704330444,
      "learning_rate": 7.247090659334039e-05,
      "loss": 1.1387,
      "step": 12688
    },
    {
      "epoch": 2.3581118751161494,
      "grad_norm": 0.823181688785553,
      "learning_rate": 7.245686203719179e-05,
      "loss": 0.8583,
      "step": 12689
    },
    {
      "epoch": 2.3582977141795207,
      "grad_norm": 0.9091300964355469,
      "learning_rate": 7.244281806890419e-05,
      "loss": 0.9424,
      "step": 12690
    },
    {
      "epoch": 2.3584835532428916,
      "grad_norm": 0.6754640936851501,
      "learning_rate": 7.242877468877735e-05,
      "loss": 0.7311,
      "step": 12691
    },
    {
      "epoch": 2.358669392306263,
      "grad_norm": 0.984846293926239,
      "learning_rate": 7.241473189711099e-05,
      "loss": 1.1976,
      "step": 12692
    },
    {
      "epoch": 2.358855231369634,
      "grad_norm": 0.9455552101135254,
      "learning_rate": 7.240068969420486e-05,
      "loss": 0.8349,
      "step": 12693
    },
    {
      "epoch": 2.359041070433005,
      "grad_norm": 0.8178308010101318,
      "learning_rate": 7.238664808035863e-05,
      "loss": 1.069,
      "step": 12694
    },
    {
      "epoch": 2.359226909496376,
      "grad_norm": 0.8013490438461304,
      "learning_rate": 7.237260705587203e-05,
      "loss": 0.8185,
      "step": 12695
    },
    {
      "epoch": 2.3594127485597474,
      "grad_norm": 0.8036800622940063,
      "learning_rate": 7.235856662104465e-05,
      "loss": 0.8689,
      "step": 12696
    },
    {
      "epoch": 2.3595985876231182,
      "grad_norm": 0.8035423755645752,
      "learning_rate": 7.23445267761763e-05,
      "loss": 1.0646,
      "step": 12697
    },
    {
      "epoch": 2.3597844266864896,
      "grad_norm": 0.872822105884552,
      "learning_rate": 7.233048752156655e-05,
      "loss": 0.9459,
      "step": 12698
    },
    {
      "epoch": 2.3599702657498605,
      "grad_norm": 0.7758422493934631,
      "learning_rate": 7.231644885751507e-05,
      "loss": 0.8219,
      "step": 12699
    },
    {
      "epoch": 2.360156104813232,
      "grad_norm": 0.882146418094635,
      "learning_rate": 7.230241078432148e-05,
      "loss": 0.8086,
      "step": 12700
    },
    {
      "epoch": 2.3603419438766027,
      "grad_norm": 0.7079089879989624,
      "learning_rate": 7.22883733022854e-05,
      "loss": 0.9594,
      "step": 12701
    },
    {
      "epoch": 2.360527782939974,
      "grad_norm": 0.8441744446754456,
      "learning_rate": 7.227433641170643e-05,
      "loss": 0.9397,
      "step": 12702
    },
    {
      "epoch": 2.3607136220033453,
      "grad_norm": 0.6709832549095154,
      "learning_rate": 7.226030011288419e-05,
      "loss": 0.7539,
      "step": 12703
    },
    {
      "epoch": 2.360899461066716,
      "grad_norm": 0.8459264636039734,
      "learning_rate": 7.224626440611821e-05,
      "loss": 0.9691,
      "step": 12704
    },
    {
      "epoch": 2.361085300130087,
      "grad_norm": 0.7479331493377686,
      "learning_rate": 7.223222929170812e-05,
      "loss": 0.7687,
      "step": 12705
    },
    {
      "epoch": 2.3612711391934584,
      "grad_norm": 0.9345939755439758,
      "learning_rate": 7.221819476995337e-05,
      "loss": 1.0044,
      "step": 12706
    },
    {
      "epoch": 2.3614569782568298,
      "grad_norm": 0.7808400988578796,
      "learning_rate": 7.220416084115364e-05,
      "loss": 0.9883,
      "step": 12707
    },
    {
      "epoch": 2.3616428173202006,
      "grad_norm": 0.7628604173660278,
      "learning_rate": 7.219012750560839e-05,
      "loss": 0.7652,
      "step": 12708
    },
    {
      "epoch": 2.361828656383572,
      "grad_norm": 0.7046692371368408,
      "learning_rate": 7.217609476361717e-05,
      "loss": 0.7371,
      "step": 12709
    },
    {
      "epoch": 2.362014495446943,
      "grad_norm": 0.7415536046028137,
      "learning_rate": 7.216206261547942e-05,
      "loss": 0.7985,
      "step": 12710
    },
    {
      "epoch": 2.362200334510314,
      "grad_norm": 0.7339528799057007,
      "learning_rate": 7.21480310614947e-05,
      "loss": 0.8325,
      "step": 12711
    },
    {
      "epoch": 2.362386173573685,
      "grad_norm": 0.7695592045783997,
      "learning_rate": 7.213400010196245e-05,
      "loss": 0.6426,
      "step": 12712
    },
    {
      "epoch": 2.3625720126370564,
      "grad_norm": 0.9707488417625427,
      "learning_rate": 7.211996973718216e-05,
      "loss": 0.9557,
      "step": 12713
    },
    {
      "epoch": 2.3627578517004273,
      "grad_norm": 0.7279425263404846,
      "learning_rate": 7.210593996745325e-05,
      "loss": 0.9252,
      "step": 12714
    },
    {
      "epoch": 2.3629436907637986,
      "grad_norm": 0.7701457142829895,
      "learning_rate": 7.209191079307521e-05,
      "loss": 0.9995,
      "step": 12715
    },
    {
      "epoch": 2.3631295298271695,
      "grad_norm": 0.706351101398468,
      "learning_rate": 7.207788221434741e-05,
      "loss": 0.8566,
      "step": 12716
    },
    {
      "epoch": 2.363315368890541,
      "grad_norm": 0.9042361974716187,
      "learning_rate": 7.206385423156929e-05,
      "loss": 1.0993,
      "step": 12717
    },
    {
      "epoch": 2.3635012079539117,
      "grad_norm": 0.820608377456665,
      "learning_rate": 7.204982684504027e-05,
      "loss": 0.6605,
      "step": 12718
    },
    {
      "epoch": 2.363687047017283,
      "grad_norm": 0.8894959092140198,
      "learning_rate": 7.203580005505977e-05,
      "loss": 0.9441,
      "step": 12719
    },
    {
      "epoch": 2.363872886080654,
      "grad_norm": 0.7809706926345825,
      "learning_rate": 7.202177386192707e-05,
      "loss": 0.8342,
      "step": 12720
    },
    {
      "epoch": 2.3640587251440253,
      "grad_norm": 0.9790495038032532,
      "learning_rate": 7.200774826594166e-05,
      "loss": 0.9595,
      "step": 12721
    },
    {
      "epoch": 2.3642445642073966,
      "grad_norm": 0.7445716261863708,
      "learning_rate": 7.19937232674028e-05,
      "loss": 1.1687,
      "step": 12722
    },
    {
      "epoch": 2.3644304032707675,
      "grad_norm": 0.80829256772995,
      "learning_rate": 7.197969886660984e-05,
      "loss": 0.8791,
      "step": 12723
    },
    {
      "epoch": 2.364616242334139,
      "grad_norm": 0.8042446970939636,
      "learning_rate": 7.196567506386214e-05,
      "loss": 1.124,
      "step": 12724
    },
    {
      "epoch": 2.3648020813975097,
      "grad_norm": 0.8191190361976624,
      "learning_rate": 7.1951651859459e-05,
      "loss": 1.0811,
      "step": 12725
    },
    {
      "epoch": 2.364987920460881,
      "grad_norm": 0.8683209419250488,
      "learning_rate": 7.19376292536997e-05,
      "loss": 0.7933,
      "step": 12726
    },
    {
      "epoch": 2.365173759524252,
      "grad_norm": 0.6210349798202515,
      "learning_rate": 7.192360724688352e-05,
      "loss": 0.5691,
      "step": 12727
    },
    {
      "epoch": 2.3653595985876232,
      "grad_norm": 0.7325350642204285,
      "learning_rate": 7.19095858393098e-05,
      "loss": 0.6499,
      "step": 12728
    },
    {
      "epoch": 2.365545437650994,
      "grad_norm": 0.8471894264221191,
      "learning_rate": 7.189556503127776e-05,
      "loss": 0.756,
      "step": 12729
    },
    {
      "epoch": 2.3657312767143654,
      "grad_norm": 1.4356576204299927,
      "learning_rate": 7.188154482308667e-05,
      "loss": 1.328,
      "step": 12730
    },
    {
      "epoch": 2.3659171157777363,
      "grad_norm": 0.7856905460357666,
      "learning_rate": 7.186752521503575e-05,
      "loss": 0.8839,
      "step": 12731
    },
    {
      "epoch": 2.3661029548411077,
      "grad_norm": 0.7553587555885315,
      "learning_rate": 7.185350620742421e-05,
      "loss": 0.9472,
      "step": 12732
    },
    {
      "epoch": 2.3662887939044785,
      "grad_norm": 0.9476107358932495,
      "learning_rate": 7.18394878005513e-05,
      "loss": 0.7734,
      "step": 12733
    },
    {
      "epoch": 2.36647463296785,
      "grad_norm": 0.9446849822998047,
      "learning_rate": 7.182546999471617e-05,
      "loss": 1.077,
      "step": 12734
    },
    {
      "epoch": 2.3666604720312208,
      "grad_norm": 0.7738652229309082,
      "learning_rate": 7.181145279021806e-05,
      "loss": 0.8729,
      "step": 12735
    },
    {
      "epoch": 2.366846311094592,
      "grad_norm": 0.8181040287017822,
      "learning_rate": 7.179743618735611e-05,
      "loss": 1.0301,
      "step": 12736
    },
    {
      "epoch": 2.3670321501579634,
      "grad_norm": 0.89509516954422,
      "learning_rate": 7.178342018642947e-05,
      "loss": 1.0886,
      "step": 12737
    },
    {
      "epoch": 2.3672179892213343,
      "grad_norm": 0.8342881202697754,
      "learning_rate": 7.176940478773729e-05,
      "loss": 1.0007,
      "step": 12738
    },
    {
      "epoch": 2.3674038282847056,
      "grad_norm": 0.81234210729599,
      "learning_rate": 7.175538999157876e-05,
      "loss": 1.1919,
      "step": 12739
    },
    {
      "epoch": 2.3675896673480765,
      "grad_norm": 0.806109607219696,
      "learning_rate": 7.174137579825293e-05,
      "loss": 0.9236,
      "step": 12740
    },
    {
      "epoch": 2.367775506411448,
      "grad_norm": 0.8120869398117065,
      "learning_rate": 7.172736220805898e-05,
      "loss": 0.8751,
      "step": 12741
    },
    {
      "epoch": 2.3679613454748187,
      "grad_norm": 0.756284773349762,
      "learning_rate": 7.171334922129593e-05,
      "loss": 0.9141,
      "step": 12742
    },
    {
      "epoch": 2.36814718453819,
      "grad_norm": 0.9308738112449646,
      "learning_rate": 7.169933683826292e-05,
      "loss": 0.8915,
      "step": 12743
    },
    {
      "epoch": 2.368333023601561,
      "grad_norm": 0.8234387636184692,
      "learning_rate": 7.168532505925899e-05,
      "loss": 0.8855,
      "step": 12744
    },
    {
      "epoch": 2.3685188626649323,
      "grad_norm": 0.8203461170196533,
      "learning_rate": 7.167131388458321e-05,
      "loss": 0.6825,
      "step": 12745
    },
    {
      "epoch": 2.368704701728303,
      "grad_norm": 0.9258405566215515,
      "learning_rate": 7.165730331453462e-05,
      "loss": 0.9386,
      "step": 12746
    },
    {
      "epoch": 2.3688905407916745,
      "grad_norm": 1.023603081703186,
      "learning_rate": 7.164329334941224e-05,
      "loss": 1.0062,
      "step": 12747
    },
    {
      "epoch": 2.3690763798550454,
      "grad_norm": 0.7846123576164246,
      "learning_rate": 7.162928398951508e-05,
      "loss": 0.9227,
      "step": 12748
    },
    {
      "epoch": 2.3692622189184167,
      "grad_norm": 0.836129903793335,
      "learning_rate": 7.161527523514221e-05,
      "loss": 0.6194,
      "step": 12749
    },
    {
      "epoch": 2.3694480579817876,
      "grad_norm": 0.8389571905136108,
      "learning_rate": 7.160126708659256e-05,
      "loss": 0.8251,
      "step": 12750
    },
    {
      "epoch": 2.369633897045159,
      "grad_norm": 0.9334648847579956,
      "learning_rate": 7.158725954416516e-05,
      "loss": 0.8305,
      "step": 12751
    },
    {
      "epoch": 2.3698197361085303,
      "grad_norm": 0.7365804314613342,
      "learning_rate": 7.157325260815892e-05,
      "loss": 0.6032,
      "step": 12752
    },
    {
      "epoch": 2.370005575171901,
      "grad_norm": 1.0147955417633057,
      "learning_rate": 7.155924627887282e-05,
      "loss": 0.9548,
      "step": 12753
    },
    {
      "epoch": 2.370191414235272,
      "grad_norm": 0.7993833422660828,
      "learning_rate": 7.154524055660581e-05,
      "loss": 0.7528,
      "step": 12754
    },
    {
      "epoch": 2.3703772532986433,
      "grad_norm": 0.8479723334312439,
      "learning_rate": 7.153123544165683e-05,
      "loss": 0.9525,
      "step": 12755
    },
    {
      "epoch": 2.3705630923620147,
      "grad_norm": 0.7671613097190857,
      "learning_rate": 7.151723093432474e-05,
      "loss": 0.8971,
      "step": 12756
    },
    {
      "epoch": 2.3707489314253856,
      "grad_norm": 0.9010011553764343,
      "learning_rate": 7.150322703490849e-05,
      "loss": 0.9733,
      "step": 12757
    },
    {
      "epoch": 2.370934770488757,
      "grad_norm": 0.9608262181282043,
      "learning_rate": 7.148922374370697e-05,
      "loss": 1.0727,
      "step": 12758
    },
    {
      "epoch": 2.371120609552128,
      "grad_norm": 0.6904570460319519,
      "learning_rate": 7.147522106101901e-05,
      "loss": 0.819,
      "step": 12759
    },
    {
      "epoch": 2.371306448615499,
      "grad_norm": 0.8582159280776978,
      "learning_rate": 7.146121898714351e-05,
      "loss": 0.8103,
      "step": 12760
    },
    {
      "epoch": 2.37149228767887,
      "grad_norm": 0.8501172065734863,
      "learning_rate": 7.144721752237935e-05,
      "loss": 1.0177,
      "step": 12761
    },
    {
      "epoch": 2.3716781267422413,
      "grad_norm": 0.9760304093360901,
      "learning_rate": 7.143321666702531e-05,
      "loss": 0.9418,
      "step": 12762
    },
    {
      "epoch": 2.371863965805612,
      "grad_norm": 0.557565450668335,
      "learning_rate": 7.141921642138025e-05,
      "loss": 0.5414,
      "step": 12763
    },
    {
      "epoch": 2.3720498048689835,
      "grad_norm": 0.7802746891975403,
      "learning_rate": 7.140521678574296e-05,
      "loss": 0.9173,
      "step": 12764
    },
    {
      "epoch": 2.3722356439323544,
      "grad_norm": 0.8309347033500671,
      "learning_rate": 7.139121776041226e-05,
      "loss": 0.8513,
      "step": 12765
    },
    {
      "epoch": 2.3724214829957257,
      "grad_norm": 0.865175187587738,
      "learning_rate": 7.137721934568691e-05,
      "loss": 0.958,
      "step": 12766
    },
    {
      "epoch": 2.3726073220590966,
      "grad_norm": 1.1070252656936646,
      "learning_rate": 7.136322154186568e-05,
      "loss": 0.9722,
      "step": 12767
    },
    {
      "epoch": 2.372793161122468,
      "grad_norm": 0.9914840459823608,
      "learning_rate": 7.134922434924737e-05,
      "loss": 1.0765,
      "step": 12768
    },
    {
      "epoch": 2.372979000185839,
      "grad_norm": 0.751986563205719,
      "learning_rate": 7.133522776813066e-05,
      "loss": 1.1364,
      "step": 12769
    },
    {
      "epoch": 2.37316483924921,
      "grad_norm": 0.7722371220588684,
      "learning_rate": 7.132123179881435e-05,
      "loss": 0.7617,
      "step": 12770
    },
    {
      "epoch": 2.3733506783125815,
      "grad_norm": 0.8863736391067505,
      "learning_rate": 7.130723644159715e-05,
      "loss": 0.6999,
      "step": 12771
    },
    {
      "epoch": 2.3735365173759524,
      "grad_norm": 0.8677679896354675,
      "learning_rate": 7.129324169677773e-05,
      "loss": 1.1685,
      "step": 12772
    },
    {
      "epoch": 2.3737223564393237,
      "grad_norm": 0.8729515671730042,
      "learning_rate": 7.12792475646548e-05,
      "loss": 1.0509,
      "step": 12773
    },
    {
      "epoch": 2.3739081955026946,
      "grad_norm": 0.849473774433136,
      "learning_rate": 7.126525404552706e-05,
      "loss": 0.9959,
      "step": 12774
    },
    {
      "epoch": 2.374094034566066,
      "grad_norm": 0.9366466999053955,
      "learning_rate": 7.125126113969316e-05,
      "loss": 0.8327,
      "step": 12775
    },
    {
      "epoch": 2.374279873629437,
      "grad_norm": 0.8395025730133057,
      "learning_rate": 7.123726884745174e-05,
      "loss": 0.8906,
      "step": 12776
    },
    {
      "epoch": 2.374465712692808,
      "grad_norm": 0.7203903794288635,
      "learning_rate": 7.122327716910146e-05,
      "loss": 0.8016,
      "step": 12777
    },
    {
      "epoch": 2.374651551756179,
      "grad_norm": 0.8004384636878967,
      "learning_rate": 7.120928610494097e-05,
      "loss": 0.9767,
      "step": 12778
    },
    {
      "epoch": 2.3748373908195504,
      "grad_norm": 0.8686206340789795,
      "learning_rate": 7.119529565526883e-05,
      "loss": 1.0413,
      "step": 12779
    },
    {
      "epoch": 2.3750232298829212,
      "grad_norm": 0.7640944123268127,
      "learning_rate": 7.118130582038367e-05,
      "loss": 0.8518,
      "step": 12780
    },
    {
      "epoch": 2.3752090689462926,
      "grad_norm": 0.7188209891319275,
      "learning_rate": 7.11673166005841e-05,
      "loss": 0.8153,
      "step": 12781
    },
    {
      "epoch": 2.3753949080096635,
      "grad_norm": 0.8678622841835022,
      "learning_rate": 7.115332799616868e-05,
      "loss": 1.1128,
      "step": 12782
    },
    {
      "epoch": 2.375580747073035,
      "grad_norm": 0.8177844882011414,
      "learning_rate": 7.113934000743598e-05,
      "loss": 0.8343,
      "step": 12783
    },
    {
      "epoch": 2.3757665861364057,
      "grad_norm": 0.8202971816062927,
      "learning_rate": 7.112535263468452e-05,
      "loss": 0.9535,
      "step": 12784
    },
    {
      "epoch": 2.375952425199777,
      "grad_norm": 0.9193758368492126,
      "learning_rate": 7.111136587821288e-05,
      "loss": 0.6666,
      "step": 12785
    },
    {
      "epoch": 2.3761382642631483,
      "grad_norm": 0.7091536521911621,
      "learning_rate": 7.109737973831954e-05,
      "loss": 0.9264,
      "step": 12786
    },
    {
      "epoch": 2.376324103326519,
      "grad_norm": 0.8636837601661682,
      "learning_rate": 7.108339421530303e-05,
      "loss": 0.7951,
      "step": 12787
    },
    {
      "epoch": 2.37650994238989,
      "grad_norm": 0.9927158355712891,
      "learning_rate": 7.106940930946186e-05,
      "loss": 1.1423,
      "step": 12788
    },
    {
      "epoch": 2.3766957814532614,
      "grad_norm": 0.7743449211120605,
      "learning_rate": 7.105542502109451e-05,
      "loss": 1.0022,
      "step": 12789
    },
    {
      "epoch": 2.3768816205166328,
      "grad_norm": 0.8231563568115234,
      "learning_rate": 7.104144135049939e-05,
      "loss": 0.9358,
      "step": 12790
    },
    {
      "epoch": 2.3770674595800036,
      "grad_norm": 0.702055037021637,
      "learning_rate": 7.102745829797506e-05,
      "loss": 0.809,
      "step": 12791
    },
    {
      "epoch": 2.377253298643375,
      "grad_norm": 0.7115417122840881,
      "learning_rate": 7.101347586381989e-05,
      "loss": 0.8033,
      "step": 12792
    },
    {
      "epoch": 2.377439137706746,
      "grad_norm": 0.7683055996894836,
      "learning_rate": 7.099949404833236e-05,
      "loss": 0.9871,
      "step": 12793
    },
    {
      "epoch": 2.377624976770117,
      "grad_norm": 0.7212742567062378,
      "learning_rate": 7.098551285181084e-05,
      "loss": 0.7506,
      "step": 12794
    },
    {
      "epoch": 2.377810815833488,
      "grad_norm": 1.4942293167114258,
      "learning_rate": 7.097153227455379e-05,
      "loss": 1.0214,
      "step": 12795
    },
    {
      "epoch": 2.3779966548968594,
      "grad_norm": 0.8741627335548401,
      "learning_rate": 7.095755231685953e-05,
      "loss": 0.8677,
      "step": 12796
    },
    {
      "epoch": 2.3781824939602303,
      "grad_norm": 0.833321750164032,
      "learning_rate": 7.094357297902649e-05,
      "loss": 0.8144,
      "step": 12797
    },
    {
      "epoch": 2.3783683330236016,
      "grad_norm": 0.7516381144523621,
      "learning_rate": 7.092959426135304e-05,
      "loss": 1.2043,
      "step": 12798
    },
    {
      "epoch": 2.3785541720869725,
      "grad_norm": 1.009989619255066,
      "learning_rate": 7.091561616413749e-05,
      "loss": 1.1505,
      "step": 12799
    },
    {
      "epoch": 2.378740011150344,
      "grad_norm": 0.8444727659225464,
      "learning_rate": 7.090163868767822e-05,
      "loss": 1.0908,
      "step": 12800
    },
    {
      "epoch": 2.378925850213715,
      "grad_norm": 0.7950802445411682,
      "learning_rate": 7.08876618322735e-05,
      "loss": 0.8567,
      "step": 12801
    },
    {
      "epoch": 2.379111689277086,
      "grad_norm": 0.7582051753997803,
      "learning_rate": 7.087368559822169e-05,
      "loss": 0.9553,
      "step": 12802
    },
    {
      "epoch": 2.379297528340457,
      "grad_norm": 0.7837277054786682,
      "learning_rate": 7.085970998582112e-05,
      "loss": 0.9817,
      "step": 12803
    },
    {
      "epoch": 2.3794833674038283,
      "grad_norm": 1.0931239128112793,
      "learning_rate": 7.084573499536999e-05,
      "loss": 0.8585,
      "step": 12804
    },
    {
      "epoch": 2.3796692064671996,
      "grad_norm": 0.7584730982780457,
      "learning_rate": 7.083176062716664e-05,
      "loss": 0.9564,
      "step": 12805
    },
    {
      "epoch": 2.3798550455305705,
      "grad_norm": 1.0813666582107544,
      "learning_rate": 7.08177868815093e-05,
      "loss": 1.1069,
      "step": 12806
    },
    {
      "epoch": 2.380040884593942,
      "grad_norm": 0.7736077308654785,
      "learning_rate": 7.080381375869623e-05,
      "loss": 1.0861,
      "step": 12807
    },
    {
      "epoch": 2.3802267236573127,
      "grad_norm": 0.8041399121284485,
      "learning_rate": 7.078984125902565e-05,
      "loss": 0.8578,
      "step": 12808
    },
    {
      "epoch": 2.380412562720684,
      "grad_norm": 0.8879541754722595,
      "learning_rate": 7.077586938279576e-05,
      "loss": 0.9983,
      "step": 12809
    },
    {
      "epoch": 2.380598401784055,
      "grad_norm": 0.8086082935333252,
      "learning_rate": 7.076189813030481e-05,
      "loss": 0.9249,
      "step": 12810
    },
    {
      "epoch": 2.3807842408474262,
      "grad_norm": 0.7989996671676636,
      "learning_rate": 7.074792750185094e-05,
      "loss": 0.9374,
      "step": 12811
    },
    {
      "epoch": 2.380970079910797,
      "grad_norm": 0.7615584135055542,
      "learning_rate": 7.073395749773238e-05,
      "loss": 0.695,
      "step": 12812
    },
    {
      "epoch": 2.3811559189741685,
      "grad_norm": 0.9058272838592529,
      "learning_rate": 7.071998811824727e-05,
      "loss": 0.9893,
      "step": 12813
    },
    {
      "epoch": 2.3813417580375393,
      "grad_norm": 0.9969415664672852,
      "learning_rate": 7.070601936369376e-05,
      "loss": 1.0421,
      "step": 12814
    },
    {
      "epoch": 2.3815275971009107,
      "grad_norm": 1.01531982421875,
      "learning_rate": 7.069205123437001e-05,
      "loss": 1.0108,
      "step": 12815
    },
    {
      "epoch": 2.3817134361642815,
      "grad_norm": 0.9354668259620667,
      "learning_rate": 7.067808373057414e-05,
      "loss": 0.9271,
      "step": 12816
    },
    {
      "epoch": 2.381899275227653,
      "grad_norm": 0.8690527081489563,
      "learning_rate": 7.066411685260424e-05,
      "loss": 0.8625,
      "step": 12817
    },
    {
      "epoch": 2.3820851142910238,
      "grad_norm": 0.7677668929100037,
      "learning_rate": 7.065015060075845e-05,
      "loss": 1.0756,
      "step": 12818
    },
    {
      "epoch": 2.382270953354395,
      "grad_norm": 0.6953485608100891,
      "learning_rate": 7.063618497533479e-05,
      "loss": 0.8802,
      "step": 12819
    },
    {
      "epoch": 2.3824567924177664,
      "grad_norm": 0.8176363110542297,
      "learning_rate": 7.062221997663141e-05,
      "loss": 1.1312,
      "step": 12820
    },
    {
      "epoch": 2.3826426314811373,
      "grad_norm": 0.9096832275390625,
      "learning_rate": 7.060825560494629e-05,
      "loss": 1.027,
      "step": 12821
    },
    {
      "epoch": 2.3828284705445086,
      "grad_norm": 0.7821842432022095,
      "learning_rate": 7.059429186057752e-05,
      "loss": 1.0861,
      "step": 12822
    },
    {
      "epoch": 2.3830143096078795,
      "grad_norm": 0.9728854894638062,
      "learning_rate": 7.058032874382314e-05,
      "loss": 0.978,
      "step": 12823
    },
    {
      "epoch": 2.383200148671251,
      "grad_norm": 0.8028355240821838,
      "learning_rate": 7.056636625498117e-05,
      "loss": 0.7403,
      "step": 12824
    },
    {
      "epoch": 2.3833859877346217,
      "grad_norm": 0.8021082878112793,
      "learning_rate": 7.055240439434958e-05,
      "loss": 0.9734,
      "step": 12825
    },
    {
      "epoch": 2.383571826797993,
      "grad_norm": 0.7374281287193298,
      "learning_rate": 7.053844316222639e-05,
      "loss": 0.8877,
      "step": 12826
    },
    {
      "epoch": 2.383757665861364,
      "grad_norm": 0.8111301064491272,
      "learning_rate": 7.052448255890957e-05,
      "loss": 1.1204,
      "step": 12827
    },
    {
      "epoch": 2.3839435049247353,
      "grad_norm": 0.7399295568466187,
      "learning_rate": 7.051052258469713e-05,
      "loss": 0.9523,
      "step": 12828
    },
    {
      "epoch": 2.384129343988106,
      "grad_norm": 0.8343197703361511,
      "learning_rate": 7.049656323988692e-05,
      "loss": 1.0284,
      "step": 12829
    },
    {
      "epoch": 2.3843151830514775,
      "grad_norm": 0.87831711769104,
      "learning_rate": 7.048260452477699e-05,
      "loss": 0.914,
      "step": 12830
    },
    {
      "epoch": 2.3845010221148484,
      "grad_norm": 0.7254018783569336,
      "learning_rate": 7.046864643966517e-05,
      "loss": 0.7349,
      "step": 12831
    },
    {
      "epoch": 2.3846868611782197,
      "grad_norm": 0.7538672685623169,
      "learning_rate": 7.04546889848494e-05,
      "loss": 0.9437,
      "step": 12832
    },
    {
      "epoch": 2.3848727002415906,
      "grad_norm": 1.0284329652786255,
      "learning_rate": 7.044073216062763e-05,
      "loss": 1.0807,
      "step": 12833
    },
    {
      "epoch": 2.385058539304962,
      "grad_norm": 0.7986030578613281,
      "learning_rate": 7.04267759672977e-05,
      "loss": 0.8098,
      "step": 12834
    },
    {
      "epoch": 2.3852443783683333,
      "grad_norm": 0.958909273147583,
      "learning_rate": 7.041282040515748e-05,
      "loss": 1.0232,
      "step": 12835
    },
    {
      "epoch": 2.385430217431704,
      "grad_norm": 0.6816046833992004,
      "learning_rate": 7.039886547450485e-05,
      "loss": 0.659,
      "step": 12836
    },
    {
      "epoch": 2.385616056495075,
      "grad_norm": 0.8010440468788147,
      "learning_rate": 7.038491117563761e-05,
      "loss": 0.9317,
      "step": 12837
    },
    {
      "epoch": 2.3858018955584464,
      "grad_norm": 0.8590589761734009,
      "learning_rate": 7.037095750885366e-05,
      "loss": 0.8401,
      "step": 12838
    },
    {
      "epoch": 2.3859877346218177,
      "grad_norm": 0.8526574969291687,
      "learning_rate": 7.035700447445075e-05,
      "loss": 0.9058,
      "step": 12839
    },
    {
      "epoch": 2.3861735736851886,
      "grad_norm": 0.7676329016685486,
      "learning_rate": 7.034305207272673e-05,
      "loss": 0.757,
      "step": 12840
    },
    {
      "epoch": 2.38635941274856,
      "grad_norm": 0.7403881549835205,
      "learning_rate": 7.032910030397934e-05,
      "loss": 0.8974,
      "step": 12841
    },
    {
      "epoch": 2.386545251811931,
      "grad_norm": 0.7914829850196838,
      "learning_rate": 7.031514916850641e-05,
      "loss": 0.9674,
      "step": 12842
    },
    {
      "epoch": 2.386731090875302,
      "grad_norm": 0.8582102656364441,
      "learning_rate": 7.030119866660564e-05,
      "loss": 0.8858,
      "step": 12843
    },
    {
      "epoch": 2.386916929938673,
      "grad_norm": 0.737777590751648,
      "learning_rate": 7.028724879857484e-05,
      "loss": 0.9357,
      "step": 12844
    },
    {
      "epoch": 2.3871027690020443,
      "grad_norm": 0.7970533967018127,
      "learning_rate": 7.027329956471174e-05,
      "loss": 0.759,
      "step": 12845
    },
    {
      "epoch": 2.387288608065415,
      "grad_norm": 0.8642406463623047,
      "learning_rate": 7.025935096531407e-05,
      "loss": 0.9972,
      "step": 12846
    },
    {
      "epoch": 2.3874744471287865,
      "grad_norm": 0.8043685555458069,
      "learning_rate": 7.024540300067947e-05,
      "loss": 0.8481,
      "step": 12847
    },
    {
      "epoch": 2.3876602861921574,
      "grad_norm": 0.6988628506660461,
      "learning_rate": 7.023145567110572e-05,
      "loss": 0.9632,
      "step": 12848
    },
    {
      "epoch": 2.3878461252555288,
      "grad_norm": 0.7697068452835083,
      "learning_rate": 7.021750897689043e-05,
      "loss": 1.008,
      "step": 12849
    },
    {
      "epoch": 2.3880319643189,
      "grad_norm": 0.7987003326416016,
      "learning_rate": 7.020356291833134e-05,
      "loss": 0.806,
      "step": 12850
    },
    {
      "epoch": 2.388217803382271,
      "grad_norm": 0.7947388291358948,
      "learning_rate": 7.018961749572604e-05,
      "loss": 0.9329,
      "step": 12851
    },
    {
      "epoch": 2.388403642445642,
      "grad_norm": 0.7829242944717407,
      "learning_rate": 7.017567270937221e-05,
      "loss": 0.8627,
      "step": 12852
    },
    {
      "epoch": 2.388589481509013,
      "grad_norm": 0.7357932925224304,
      "learning_rate": 7.016172855956743e-05,
      "loss": 0.9987,
      "step": 12853
    },
    {
      "epoch": 2.3887753205723845,
      "grad_norm": 0.7504284381866455,
      "learning_rate": 7.014778504660938e-05,
      "loss": 0.8692,
      "step": 12854
    },
    {
      "epoch": 2.3889611596357554,
      "grad_norm": 0.7871151566505432,
      "learning_rate": 7.01338421707956e-05,
      "loss": 0.8557,
      "step": 12855
    },
    {
      "epoch": 2.3891469986991267,
      "grad_norm": 0.8007663488388062,
      "learning_rate": 7.011989993242376e-05,
      "loss": 0.8101,
      "step": 12856
    },
    {
      "epoch": 2.3893328377624976,
      "grad_norm": 0.7916004061698914,
      "learning_rate": 7.010595833179134e-05,
      "loss": 0.8924,
      "step": 12857
    },
    {
      "epoch": 2.389518676825869,
      "grad_norm": 0.8495494723320007,
      "learning_rate": 7.009201736919597e-05,
      "loss": 1.0092,
      "step": 12858
    },
    {
      "epoch": 2.38970451588924,
      "grad_norm": 0.7786046862602234,
      "learning_rate": 7.007807704493514e-05,
      "loss": 0.7221,
      "step": 12859
    },
    {
      "epoch": 2.389890354952611,
      "grad_norm": 0.9809713959693909,
      "learning_rate": 7.006413735930642e-05,
      "loss": 1.0681,
      "step": 12860
    },
    {
      "epoch": 2.390076194015982,
      "grad_norm": 0.7984169125556946,
      "learning_rate": 7.005019831260731e-05,
      "loss": 0.8321,
      "step": 12861
    },
    {
      "epoch": 2.3902620330793534,
      "grad_norm": 0.82598876953125,
      "learning_rate": 7.003625990513534e-05,
      "loss": 1.0514,
      "step": 12862
    },
    {
      "epoch": 2.3904478721427243,
      "grad_norm": 0.7637234926223755,
      "learning_rate": 7.002232213718795e-05,
      "loss": 1.1037,
      "step": 12863
    },
    {
      "epoch": 2.3906337112060956,
      "grad_norm": 0.7955513596534729,
      "learning_rate": 7.000838500906263e-05,
      "loss": 0.9274,
      "step": 12864
    },
    {
      "epoch": 2.3908195502694665,
      "grad_norm": 0.894230306148529,
      "learning_rate": 6.99944485210569e-05,
      "loss": 0.914,
      "step": 12865
    },
    {
      "epoch": 2.391005389332838,
      "grad_norm": 0.9638434052467346,
      "learning_rate": 6.998051267346819e-05,
      "loss": 0.9538,
      "step": 12866
    },
    {
      "epoch": 2.3911912283962087,
      "grad_norm": 0.7865784764289856,
      "learning_rate": 6.99665774665939e-05,
      "loss": 0.8582,
      "step": 12867
    },
    {
      "epoch": 2.39137706745958,
      "grad_norm": 0.7800189256668091,
      "learning_rate": 6.99526429007315e-05,
      "loss": 0.8443,
      "step": 12868
    },
    {
      "epoch": 2.3915629065229513,
      "grad_norm": 0.9118089079856873,
      "learning_rate": 6.993870897617834e-05,
      "loss": 0.9001,
      "step": 12869
    },
    {
      "epoch": 2.3917487455863222,
      "grad_norm": 0.8228194713592529,
      "learning_rate": 6.992477569323187e-05,
      "loss": 0.7806,
      "step": 12870
    },
    {
      "epoch": 2.3919345846496936,
      "grad_norm": 0.7540860772132874,
      "learning_rate": 6.991084305218943e-05,
      "loss": 0.9232,
      "step": 12871
    },
    {
      "epoch": 2.3921204237130644,
      "grad_norm": 0.8330655694007874,
      "learning_rate": 6.989691105334844e-05,
      "loss": 0.8367,
      "step": 12872
    },
    {
      "epoch": 2.3923062627764358,
      "grad_norm": 1.076634407043457,
      "learning_rate": 6.988297969700621e-05,
      "loss": 1.0661,
      "step": 12873
    },
    {
      "epoch": 2.3924921018398067,
      "grad_norm": 0.6492782831192017,
      "learning_rate": 6.986904898346006e-05,
      "loss": 0.6156,
      "step": 12874
    },
    {
      "epoch": 2.392677940903178,
      "grad_norm": 0.7901709675788879,
      "learning_rate": 6.985511891300737e-05,
      "loss": 1.0402,
      "step": 12875
    },
    {
      "epoch": 2.392863779966549,
      "grad_norm": 0.9432883858680725,
      "learning_rate": 6.984118948594547e-05,
      "loss": 1.0928,
      "step": 12876
    },
    {
      "epoch": 2.39304961902992,
      "grad_norm": 0.8093352913856506,
      "learning_rate": 6.982726070257161e-05,
      "loss": 1.0158,
      "step": 12877
    },
    {
      "epoch": 2.393235458093291,
      "grad_norm": 0.8153324723243713,
      "learning_rate": 6.98133325631831e-05,
      "loss": 0.8456,
      "step": 12878
    },
    {
      "epoch": 2.3934212971566624,
      "grad_norm": 0.8256557583808899,
      "learning_rate": 6.97994050680772e-05,
      "loss": 1.0952,
      "step": 12879
    },
    {
      "epoch": 2.3936071362200333,
      "grad_norm": 0.8582428693771362,
      "learning_rate": 6.978547821755118e-05,
      "loss": 0.9545,
      "step": 12880
    },
    {
      "epoch": 2.3937929752834046,
      "grad_norm": 0.7831152677536011,
      "learning_rate": 6.977155201190226e-05,
      "loss": 0.8716,
      "step": 12881
    },
    {
      "epoch": 2.3939788143467755,
      "grad_norm": 0.893572986125946,
      "learning_rate": 6.975762645142771e-05,
      "loss": 0.9879,
      "step": 12882
    },
    {
      "epoch": 2.394164653410147,
      "grad_norm": 0.8478510975837708,
      "learning_rate": 6.974370153642468e-05,
      "loss": 0.9668,
      "step": 12883
    },
    {
      "epoch": 2.394350492473518,
      "grad_norm": 0.831895112991333,
      "learning_rate": 6.972977726719048e-05,
      "loss": 0.7504,
      "step": 12884
    },
    {
      "epoch": 2.394536331536889,
      "grad_norm": 0.865291953086853,
      "learning_rate": 6.971585364402216e-05,
      "loss": 1.0494,
      "step": 12885
    },
    {
      "epoch": 2.39472217060026,
      "grad_norm": 0.809294581413269,
      "learning_rate": 6.970193066721701e-05,
      "loss": 0.9431,
      "step": 12886
    },
    {
      "epoch": 2.3949080096636313,
      "grad_norm": 0.7376503348350525,
      "learning_rate": 6.968800833707218e-05,
      "loss": 0.7132,
      "step": 12887
    },
    {
      "epoch": 2.3950938487270026,
      "grad_norm": 0.9075321555137634,
      "learning_rate": 6.967408665388479e-05,
      "loss": 1.1907,
      "step": 12888
    },
    {
      "epoch": 2.3952796877903735,
      "grad_norm": 0.8884609937667847,
      "learning_rate": 6.966016561795196e-05,
      "loss": 0.8951,
      "step": 12889
    },
    {
      "epoch": 2.395465526853745,
      "grad_norm": 0.8289088606834412,
      "learning_rate": 6.964624522957087e-05,
      "loss": 1.1564,
      "step": 12890
    },
    {
      "epoch": 2.3956513659171157,
      "grad_norm": 1.029945969581604,
      "learning_rate": 6.963232548903853e-05,
      "loss": 0.9681,
      "step": 12891
    },
    {
      "epoch": 2.395837204980487,
      "grad_norm": 0.8445291519165039,
      "learning_rate": 6.961840639665214e-05,
      "loss": 0.7358,
      "step": 12892
    },
    {
      "epoch": 2.396023044043858,
      "grad_norm": 0.8968092799186707,
      "learning_rate": 6.960448795270868e-05,
      "loss": 1.0552,
      "step": 12893
    },
    {
      "epoch": 2.3962088831072292,
      "grad_norm": 0.8325434327125549,
      "learning_rate": 6.959057015750532e-05,
      "loss": 0.7296,
      "step": 12894
    },
    {
      "epoch": 2.3963947221706,
      "grad_norm": 0.8439461588859558,
      "learning_rate": 6.957665301133897e-05,
      "loss": 0.769,
      "step": 12895
    },
    {
      "epoch": 2.3965805612339715,
      "grad_norm": 1.0753918886184692,
      "learning_rate": 6.956273651450682e-05,
      "loss": 0.8952,
      "step": 12896
    },
    {
      "epoch": 2.3967664002973423,
      "grad_norm": 0.8330729007720947,
      "learning_rate": 6.95488206673058e-05,
      "loss": 0.9998,
      "step": 12897
    },
    {
      "epoch": 2.3969522393607137,
      "grad_norm": 0.824369490146637,
      "learning_rate": 6.953490547003298e-05,
      "loss": 1.0072,
      "step": 12898
    },
    {
      "epoch": 2.3971380784240846,
      "grad_norm": 0.7380400896072388,
      "learning_rate": 6.95209909229853e-05,
      "loss": 0.8426,
      "step": 12899
    },
    {
      "epoch": 2.397323917487456,
      "grad_norm": 0.8798367977142334,
      "learning_rate": 6.950707702645978e-05,
      "loss": 1.0807,
      "step": 12900
    },
    {
      "epoch": 2.3975097565508268,
      "grad_norm": 0.8785592913627625,
      "learning_rate": 6.949316378075334e-05,
      "loss": 0.7244,
      "step": 12901
    },
    {
      "epoch": 2.397695595614198,
      "grad_norm": 0.7741357684135437,
      "learning_rate": 6.9479251186163e-05,
      "loss": 1.1072,
      "step": 12902
    },
    {
      "epoch": 2.3978814346775694,
      "grad_norm": 0.8932114243507385,
      "learning_rate": 6.946533924298566e-05,
      "loss": 0.9851,
      "step": 12903
    },
    {
      "epoch": 2.3980672737409403,
      "grad_norm": 0.8525800108909607,
      "learning_rate": 6.945142795151827e-05,
      "loss": 0.9463,
      "step": 12904
    },
    {
      "epoch": 2.3982531128043116,
      "grad_norm": 0.9864116311073303,
      "learning_rate": 6.943751731205769e-05,
      "loss": 0.9916,
      "step": 12905
    },
    {
      "epoch": 2.3984389518676825,
      "grad_norm": 0.8220023512840271,
      "learning_rate": 6.942360732490084e-05,
      "loss": 1.0048,
      "step": 12906
    },
    {
      "epoch": 2.398624790931054,
      "grad_norm": 0.956392228603363,
      "learning_rate": 6.940969799034465e-05,
      "loss": 1.0739,
      "step": 12907
    },
    {
      "epoch": 2.3988106299944247,
      "grad_norm": 0.8336243629455566,
      "learning_rate": 6.939578930868598e-05,
      "loss": 1.1104,
      "step": 12908
    },
    {
      "epoch": 2.398996469057796,
      "grad_norm": 0.7887905240058899,
      "learning_rate": 6.938188128022164e-05,
      "loss": 0.9338,
      "step": 12909
    },
    {
      "epoch": 2.399182308121167,
      "grad_norm": 0.7147557139396667,
      "learning_rate": 6.936797390524852e-05,
      "loss": 0.7779,
      "step": 12910
    },
    {
      "epoch": 2.3993681471845383,
      "grad_norm": 0.7383899688720703,
      "learning_rate": 6.935406718406342e-05,
      "loss": 0.8503,
      "step": 12911
    },
    {
      "epoch": 2.399553986247909,
      "grad_norm": 0.8463869094848633,
      "learning_rate": 6.934016111696318e-05,
      "loss": 0.9774,
      "step": 12912
    },
    {
      "epoch": 2.3997398253112805,
      "grad_norm": 0.769568681716919,
      "learning_rate": 6.932625570424457e-05,
      "loss": 0.9911,
      "step": 12913
    },
    {
      "epoch": 2.3999256643746514,
      "grad_norm": 0.772621750831604,
      "learning_rate": 6.93123509462044e-05,
      "loss": 0.8799,
      "step": 12914
    },
    {
      "epoch": 2.4001115034380227,
      "grad_norm": 0.7122842073440552,
      "learning_rate": 6.929844684313945e-05,
      "loss": 0.7471,
      "step": 12915
    },
    {
      "epoch": 2.4002973425013936,
      "grad_norm": 0.7505543828010559,
      "learning_rate": 6.928454339534642e-05,
      "loss": 1.0051,
      "step": 12916
    },
    {
      "epoch": 2.400483181564765,
      "grad_norm": 0.9213287830352783,
      "learning_rate": 6.927064060312212e-05,
      "loss": 0.9575,
      "step": 12917
    },
    {
      "epoch": 2.4006690206281363,
      "grad_norm": 1.6840814352035522,
      "learning_rate": 6.925673846676329e-05,
      "loss": 1.3871,
      "step": 12918
    },
    {
      "epoch": 2.400854859691507,
      "grad_norm": 0.835517406463623,
      "learning_rate": 6.92428369865666e-05,
      "loss": 0.924,
      "step": 12919
    },
    {
      "epoch": 2.4010406987548785,
      "grad_norm": 0.7431739568710327,
      "learning_rate": 6.92289361628288e-05,
      "loss": 0.8386,
      "step": 12920
    },
    {
      "epoch": 2.4012265378182494,
      "grad_norm": 0.730377197265625,
      "learning_rate": 6.921503599584652e-05,
      "loss": 0.8848,
      "step": 12921
    },
    {
      "epoch": 2.4014123768816207,
      "grad_norm": 0.7416588664054871,
      "learning_rate": 6.920113648591649e-05,
      "loss": 0.733,
      "step": 12922
    },
    {
      "epoch": 2.4015982159449916,
      "grad_norm": 0.7663400769233704,
      "learning_rate": 6.918723763333532e-05,
      "loss": 0.9261,
      "step": 12923
    },
    {
      "epoch": 2.401784055008363,
      "grad_norm": 0.9053961038589478,
      "learning_rate": 6.917333943839972e-05,
      "loss": 0.7423,
      "step": 12924
    },
    {
      "epoch": 2.401969894071734,
      "grad_norm": 0.9448252320289612,
      "learning_rate": 6.915944190140628e-05,
      "loss": 0.7936,
      "step": 12925
    },
    {
      "epoch": 2.402155733135105,
      "grad_norm": 0.8270347118377686,
      "learning_rate": 6.914554502265162e-05,
      "loss": 1.041,
      "step": 12926
    },
    {
      "epoch": 2.402341572198476,
      "grad_norm": 0.7604723572731018,
      "learning_rate": 6.913164880243231e-05,
      "loss": 0.9407,
      "step": 12927
    },
    {
      "epoch": 2.4025274112618473,
      "grad_norm": 0.8789168000221252,
      "learning_rate": 6.911775324104504e-05,
      "loss": 0.9699,
      "step": 12928
    },
    {
      "epoch": 2.402713250325218,
      "grad_norm": 0.777075469493866,
      "learning_rate": 6.910385833878632e-05,
      "loss": 0.9532,
      "step": 12929
    },
    {
      "epoch": 2.4028990893885895,
      "grad_norm": 0.6907746195793152,
      "learning_rate": 6.908996409595273e-05,
      "loss": 0.9357,
      "step": 12930
    },
    {
      "epoch": 2.4030849284519604,
      "grad_norm": 0.7531406283378601,
      "learning_rate": 6.907607051284079e-05,
      "loss": 0.7451,
      "step": 12931
    },
    {
      "epoch": 2.4032707675153318,
      "grad_norm": 0.8290823101997375,
      "learning_rate": 6.906217758974709e-05,
      "loss": 0.9023,
      "step": 12932
    },
    {
      "epoch": 2.403456606578703,
      "grad_norm": 0.8053004145622253,
      "learning_rate": 6.90482853269681e-05,
      "loss": 0.9673,
      "step": 12933
    },
    {
      "epoch": 2.403642445642074,
      "grad_norm": 0.795650839805603,
      "learning_rate": 6.903439372480037e-05,
      "loss": 0.7834,
      "step": 12934
    },
    {
      "epoch": 2.403828284705445,
      "grad_norm": 0.7404899597167969,
      "learning_rate": 6.902050278354033e-05,
      "loss": 0.7151,
      "step": 12935
    },
    {
      "epoch": 2.404014123768816,
      "grad_norm": 0.7955520749092102,
      "learning_rate": 6.900661250348452e-05,
      "loss": 0.9404,
      "step": 12936
    },
    {
      "epoch": 2.4041999628321875,
      "grad_norm": 0.8237014412879944,
      "learning_rate": 6.89927228849293e-05,
      "loss": 0.8605,
      "step": 12937
    },
    {
      "epoch": 2.4043858018955584,
      "grad_norm": 0.8399513363838196,
      "learning_rate": 6.897883392817128e-05,
      "loss": 0.7486,
      "step": 12938
    },
    {
      "epoch": 2.4045716409589297,
      "grad_norm": 0.6245372295379639,
      "learning_rate": 6.89649456335068e-05,
      "loss": 0.5818,
      "step": 12939
    },
    {
      "epoch": 2.4047574800223006,
      "grad_norm": 0.7267877459526062,
      "learning_rate": 6.895105800123231e-05,
      "loss": 0.7855,
      "step": 12940
    },
    {
      "epoch": 2.404943319085672,
      "grad_norm": 0.7312634587287903,
      "learning_rate": 6.893717103164419e-05,
      "loss": 0.9549,
      "step": 12941
    },
    {
      "epoch": 2.405129158149043,
      "grad_norm": 0.7766819596290588,
      "learning_rate": 6.892328472503887e-05,
      "loss": 0.784,
      "step": 12942
    },
    {
      "epoch": 2.405314997212414,
      "grad_norm": 0.738930881023407,
      "learning_rate": 6.890939908171268e-05,
      "loss": 0.7427,
      "step": 12943
    },
    {
      "epoch": 2.405500836275785,
      "grad_norm": 0.7638840079307556,
      "learning_rate": 6.889551410196203e-05,
      "loss": 0.973,
      "step": 12944
    },
    {
      "epoch": 2.4056866753391564,
      "grad_norm": 0.7316983938217163,
      "learning_rate": 6.888162978608325e-05,
      "loss": 0.9748,
      "step": 12945
    },
    {
      "epoch": 2.4058725144025273,
      "grad_norm": 0.8835756182670593,
      "learning_rate": 6.886774613437266e-05,
      "loss": 0.7174,
      "step": 12946
    },
    {
      "epoch": 2.4060583534658986,
      "grad_norm": 0.7415609955787659,
      "learning_rate": 6.885386314712663e-05,
      "loss": 0.8063,
      "step": 12947
    },
    {
      "epoch": 2.4062441925292695,
      "grad_norm": 0.7552850842475891,
      "learning_rate": 6.883998082464142e-05,
      "loss": 0.8563,
      "step": 12948
    },
    {
      "epoch": 2.406430031592641,
      "grad_norm": 0.7759203314781189,
      "learning_rate": 6.882609916721335e-05,
      "loss": 0.7608,
      "step": 12949
    },
    {
      "epoch": 2.4066158706560117,
      "grad_norm": 0.7742679715156555,
      "learning_rate": 6.881221817513873e-05,
      "loss": 1.0519,
      "step": 12950
    },
    {
      "epoch": 2.406801709719383,
      "grad_norm": 0.8516238331794739,
      "learning_rate": 6.879833784871374e-05,
      "loss": 0.9311,
      "step": 12951
    },
    {
      "epoch": 2.4069875487827543,
      "grad_norm": 0.6422767639160156,
      "learning_rate": 6.878445818823473e-05,
      "loss": 0.7316,
      "step": 12952
    },
    {
      "epoch": 2.4071733878461252,
      "grad_norm": 0.7644221782684326,
      "learning_rate": 6.877057919399788e-05,
      "loss": 0.7009,
      "step": 12953
    },
    {
      "epoch": 2.4073592269094966,
      "grad_norm": 0.7887259721755981,
      "learning_rate": 6.875670086629942e-05,
      "loss": 0.9655,
      "step": 12954
    },
    {
      "epoch": 2.4075450659728674,
      "grad_norm": 0.9073466062545776,
      "learning_rate": 6.874282320543557e-05,
      "loss": 1.0732,
      "step": 12955
    },
    {
      "epoch": 2.4077309050362388,
      "grad_norm": 0.8829473853111267,
      "learning_rate": 6.87289462117025e-05,
      "loss": 1.0061,
      "step": 12956
    },
    {
      "epoch": 2.4079167440996097,
      "grad_norm": 0.7792870998382568,
      "learning_rate": 6.871506988539644e-05,
      "loss": 0.7244,
      "step": 12957
    },
    {
      "epoch": 2.408102583162981,
      "grad_norm": 0.9756807088851929,
      "learning_rate": 6.87011942268135e-05,
      "loss": 1.0455,
      "step": 12958
    },
    {
      "epoch": 2.408288422226352,
      "grad_norm": 0.9650059342384338,
      "learning_rate": 6.868731923624985e-05,
      "loss": 1.1628,
      "step": 12959
    },
    {
      "epoch": 2.408474261289723,
      "grad_norm": 0.8020375370979309,
      "learning_rate": 6.867344491400168e-05,
      "loss": 1.0403,
      "step": 12960
    },
    {
      "epoch": 2.408660100353094,
      "grad_norm": 0.9928075075149536,
      "learning_rate": 6.865957126036502e-05,
      "loss": 0.9012,
      "step": 12961
    },
    {
      "epoch": 2.4088459394164654,
      "grad_norm": 0.6184268593788147,
      "learning_rate": 6.864569827563607e-05,
      "loss": 0.5636,
      "step": 12962
    },
    {
      "epoch": 2.4090317784798363,
      "grad_norm": 0.893981397151947,
      "learning_rate": 6.863182596011087e-05,
      "loss": 1.0563,
      "step": 12963
    },
    {
      "epoch": 2.4092176175432076,
      "grad_norm": 0.8809736967086792,
      "learning_rate": 6.86179543140855e-05,
      "loss": 0.9638,
      "step": 12964
    },
    {
      "epoch": 2.4094034566065785,
      "grad_norm": 0.9104132652282715,
      "learning_rate": 6.860408333785606e-05,
      "loss": 0.8908,
      "step": 12965
    },
    {
      "epoch": 2.40958929566995,
      "grad_norm": 0.7690507769584656,
      "learning_rate": 6.859021303171857e-05,
      "loss": 0.8789,
      "step": 12966
    },
    {
      "epoch": 2.409775134733321,
      "grad_norm": 0.7475899457931519,
      "learning_rate": 6.85763433959691e-05,
      "loss": 0.9408,
      "step": 12967
    },
    {
      "epoch": 2.409960973796692,
      "grad_norm": 0.8373811841011047,
      "learning_rate": 6.856247443090365e-05,
      "loss": 0.8776,
      "step": 12968
    },
    {
      "epoch": 2.410146812860063,
      "grad_norm": 0.9541831016540527,
      "learning_rate": 6.85486061368182e-05,
      "loss": 0.9184,
      "step": 12969
    },
    {
      "epoch": 2.4103326519234343,
      "grad_norm": 0.759236752986908,
      "learning_rate": 6.853473851400882e-05,
      "loss": 0.7704,
      "step": 12970
    },
    {
      "epoch": 2.4105184909868056,
      "grad_norm": 0.8918933272361755,
      "learning_rate": 6.852087156277143e-05,
      "loss": 0.9457,
      "step": 12971
    },
    {
      "epoch": 2.4107043300501765,
      "grad_norm": 1.1025477647781372,
      "learning_rate": 6.850700528340204e-05,
      "loss": 0.8847,
      "step": 12972
    },
    {
      "epoch": 2.410890169113548,
      "grad_norm": 0.7837464213371277,
      "learning_rate": 6.849313967619655e-05,
      "loss": 0.6502,
      "step": 12973
    },
    {
      "epoch": 2.4110760081769187,
      "grad_norm": 0.7813313007354736,
      "learning_rate": 6.847927474145095e-05,
      "loss": 1.0479,
      "step": 12974
    },
    {
      "epoch": 2.41126184724029,
      "grad_norm": 0.8003170490264893,
      "learning_rate": 6.846541047946112e-05,
      "loss": 0.9041,
      "step": 12975
    },
    {
      "epoch": 2.411447686303661,
      "grad_norm": 0.7718076705932617,
      "learning_rate": 6.845154689052299e-05,
      "loss": 0.9193,
      "step": 12976
    },
    {
      "epoch": 2.4116335253670322,
      "grad_norm": 0.810656726360321,
      "learning_rate": 6.843768397493246e-05,
      "loss": 0.9488,
      "step": 12977
    },
    {
      "epoch": 2.411819364430403,
      "grad_norm": 0.7257324457168579,
      "learning_rate": 6.84238217329854e-05,
      "loss": 1.0075,
      "step": 12978
    },
    {
      "epoch": 2.4120052034937745,
      "grad_norm": 0.770806074142456,
      "learning_rate": 6.840996016497765e-05,
      "loss": 0.9499,
      "step": 12979
    },
    {
      "epoch": 2.4121910425571453,
      "grad_norm": 0.6285768747329712,
      "learning_rate": 6.839609927120513e-05,
      "loss": 0.7179,
      "step": 12980
    },
    {
      "epoch": 2.4123768816205167,
      "grad_norm": 0.9031039476394653,
      "learning_rate": 6.838223905196362e-05,
      "loss": 0.9258,
      "step": 12981
    },
    {
      "epoch": 2.412562720683888,
      "grad_norm": 1.1146539449691772,
      "learning_rate": 6.836837950754897e-05,
      "loss": 0.9575,
      "step": 12982
    },
    {
      "epoch": 2.412748559747259,
      "grad_norm": 0.6355863809585571,
      "learning_rate": 6.835452063825697e-05,
      "loss": 0.7045,
      "step": 12983
    },
    {
      "epoch": 2.4129343988106298,
      "grad_norm": 0.8352602124214172,
      "learning_rate": 6.834066244438343e-05,
      "loss": 0.9539,
      "step": 12984
    },
    {
      "epoch": 2.413120237874001,
      "grad_norm": 0.7984040975570679,
      "learning_rate": 6.832680492622411e-05,
      "loss": 0.9267,
      "step": 12985
    },
    {
      "epoch": 2.4133060769373724,
      "grad_norm": 0.8858015537261963,
      "learning_rate": 6.831294808407478e-05,
      "loss": 0.9607,
      "step": 12986
    },
    {
      "epoch": 2.4134919160007433,
      "grad_norm": 0.818161129951477,
      "learning_rate": 6.829909191823121e-05,
      "loss": 0.9118,
      "step": 12987
    },
    {
      "epoch": 2.4136777550641146,
      "grad_norm": 0.9285476207733154,
      "learning_rate": 6.828523642898911e-05,
      "loss": 0.8921,
      "step": 12988
    },
    {
      "epoch": 2.4138635941274855,
      "grad_norm": 0.6942713856697083,
      "learning_rate": 6.827138161664423e-05,
      "loss": 0.9212,
      "step": 12989
    },
    {
      "epoch": 2.414049433190857,
      "grad_norm": 0.8579042553901672,
      "learning_rate": 6.825752748149224e-05,
      "loss": 0.8859,
      "step": 12990
    },
    {
      "epoch": 2.4142352722542277,
      "grad_norm": 0.8150649070739746,
      "learning_rate": 6.824367402382885e-05,
      "loss": 0.95,
      "step": 12991
    },
    {
      "epoch": 2.414421111317599,
      "grad_norm": 0.8967882394790649,
      "learning_rate": 6.822982124394977e-05,
      "loss": 0.8815,
      "step": 12992
    },
    {
      "epoch": 2.41460695038097,
      "grad_norm": 0.6919797658920288,
      "learning_rate": 6.82159691421506e-05,
      "loss": 0.7887,
      "step": 12993
    },
    {
      "epoch": 2.4147927894443413,
      "grad_norm": 0.8287318348884583,
      "learning_rate": 6.820211771872707e-05,
      "loss": 0.8545,
      "step": 12994
    },
    {
      "epoch": 2.414978628507712,
      "grad_norm": 0.7475675940513611,
      "learning_rate": 6.818826697397474e-05,
      "loss": 0.9978,
      "step": 12995
    },
    {
      "epoch": 2.4151644675710835,
      "grad_norm": 0.9870370626449585,
      "learning_rate": 6.817441690818925e-05,
      "loss": 0.9869,
      "step": 12996
    },
    {
      "epoch": 2.4153503066344544,
      "grad_norm": 0.8511314988136292,
      "learning_rate": 6.816056752166626e-05,
      "loss": 0.8065,
      "step": 12997
    },
    {
      "epoch": 2.4155361456978257,
      "grad_norm": 0.8184236884117126,
      "learning_rate": 6.814671881470128e-05,
      "loss": 1.0464,
      "step": 12998
    },
    {
      "epoch": 2.4157219847611966,
      "grad_norm": 0.7355121374130249,
      "learning_rate": 6.813287078758995e-05,
      "loss": 0.8323,
      "step": 12999
    },
    {
      "epoch": 2.415907823824568,
      "grad_norm": 0.859524667263031,
      "learning_rate": 6.811902344062775e-05,
      "loss": 0.9679,
      "step": 13000
    },
    {
      "epoch": 2.4160936628879393,
      "grad_norm": 0.6515125036239624,
      "learning_rate": 6.810517677411032e-05,
      "loss": 0.8337,
      "step": 13001
    },
    {
      "epoch": 2.41627950195131,
      "grad_norm": 1.004290223121643,
      "learning_rate": 6.809133078833319e-05,
      "loss": 1.0798,
      "step": 13002
    },
    {
      "epoch": 2.4164653410146815,
      "grad_norm": 0.7877489328384399,
      "learning_rate": 6.807748548359181e-05,
      "loss": 1.0417,
      "step": 13003
    },
    {
      "epoch": 2.4166511800780524,
      "grad_norm": 0.6995466351509094,
      "learning_rate": 6.806364086018174e-05,
      "loss": 0.9388,
      "step": 13004
    },
    {
      "epoch": 2.4168370191414237,
      "grad_norm": 0.7970275282859802,
      "learning_rate": 6.804979691839843e-05,
      "loss": 0.9569,
      "step": 13005
    },
    {
      "epoch": 2.4170228582047946,
      "grad_norm": 0.8142886757850647,
      "learning_rate": 6.803595365853739e-05,
      "loss": 0.9285,
      "step": 13006
    },
    {
      "epoch": 2.417208697268166,
      "grad_norm": 2.7952983379364014,
      "learning_rate": 6.80221110808941e-05,
      "loss": 1.3252,
      "step": 13007
    },
    {
      "epoch": 2.417394536331537,
      "grad_norm": 0.7739753723144531,
      "learning_rate": 6.800826918576392e-05,
      "loss": 0.856,
      "step": 13008
    },
    {
      "epoch": 2.417580375394908,
      "grad_norm": 0.6938930749893188,
      "learning_rate": 6.799442797344236e-05,
      "loss": 0.8958,
      "step": 13009
    },
    {
      "epoch": 2.417766214458279,
      "grad_norm": 0.7906118631362915,
      "learning_rate": 6.79805874442248e-05,
      "loss": 0.9118,
      "step": 13010
    },
    {
      "epoch": 2.4179520535216503,
      "grad_norm": 0.9179455637931824,
      "learning_rate": 6.796674759840662e-05,
      "loss": 0.9237,
      "step": 13011
    },
    {
      "epoch": 2.418137892585021,
      "grad_norm": 0.9764026403427124,
      "learning_rate": 6.79529084362833e-05,
      "loss": 1.0781,
      "step": 13012
    },
    {
      "epoch": 2.4183237316483925,
      "grad_norm": 0.7278144955635071,
      "learning_rate": 6.793906995815013e-05,
      "loss": 0.7157,
      "step": 13013
    },
    {
      "epoch": 2.4185095707117634,
      "grad_norm": 0.8127259016036987,
      "learning_rate": 6.792523216430251e-05,
      "loss": 0.9512,
      "step": 13014
    },
    {
      "epoch": 2.4186954097751348,
      "grad_norm": 0.866875171661377,
      "learning_rate": 6.791139505503574e-05,
      "loss": 1.025,
      "step": 13015
    },
    {
      "epoch": 2.418881248838506,
      "grad_norm": 0.7558534145355225,
      "learning_rate": 6.789755863064519e-05,
      "loss": 1.0285,
      "step": 13016
    },
    {
      "epoch": 2.419067087901877,
      "grad_norm": 0.8152698278427124,
      "learning_rate": 6.788372289142618e-05,
      "loss": 0.8862,
      "step": 13017
    },
    {
      "epoch": 2.419252926965248,
      "grad_norm": 0.7945411801338196,
      "learning_rate": 6.786988783767397e-05,
      "loss": 0.9398,
      "step": 13018
    },
    {
      "epoch": 2.419438766028619,
      "grad_norm": 0.9634117484092712,
      "learning_rate": 6.785605346968386e-05,
      "loss": 0.8342,
      "step": 13019
    },
    {
      "epoch": 2.4196246050919905,
      "grad_norm": 0.7568840384483337,
      "learning_rate": 6.784221978775114e-05,
      "loss": 0.8844,
      "step": 13020
    },
    {
      "epoch": 2.4198104441553614,
      "grad_norm": 0.9052173495292664,
      "learning_rate": 6.782838679217102e-05,
      "loss": 0.8794,
      "step": 13021
    },
    {
      "epoch": 2.4199962832187327,
      "grad_norm": 0.7473658323287964,
      "learning_rate": 6.781455448323881e-05,
      "loss": 0.6503,
      "step": 13022
    },
    {
      "epoch": 2.4201821222821036,
      "grad_norm": 0.8628986477851868,
      "learning_rate": 6.78007228612497e-05,
      "loss": 0.8545,
      "step": 13023
    },
    {
      "epoch": 2.420367961345475,
      "grad_norm": 0.8227683901786804,
      "learning_rate": 6.77868919264989e-05,
      "loss": 1.0456,
      "step": 13024
    },
    {
      "epoch": 2.420553800408846,
      "grad_norm": 0.7157145738601685,
      "learning_rate": 6.777306167928161e-05,
      "loss": 0.6423,
      "step": 13025
    },
    {
      "epoch": 2.420739639472217,
      "grad_norm": 0.9058414697647095,
      "learning_rate": 6.775923211989301e-05,
      "loss": 1.0976,
      "step": 13026
    },
    {
      "epoch": 2.420925478535588,
      "grad_norm": 0.8798935413360596,
      "learning_rate": 6.77454032486283e-05,
      "loss": 1.0868,
      "step": 13027
    },
    {
      "epoch": 2.4211113175989594,
      "grad_norm": 0.8619986176490784,
      "learning_rate": 6.773157506578258e-05,
      "loss": 0.9352,
      "step": 13028
    },
    {
      "epoch": 2.4212971566623303,
      "grad_norm": 0.8048561811447144,
      "learning_rate": 6.771774757165103e-05,
      "loss": 0.9036,
      "step": 13029
    },
    {
      "epoch": 2.4214829957257016,
      "grad_norm": 0.6234624981880188,
      "learning_rate": 6.770392076652875e-05,
      "loss": 0.7566,
      "step": 13030
    },
    {
      "epoch": 2.421668834789073,
      "grad_norm": 0.7719530463218689,
      "learning_rate": 6.769009465071088e-05,
      "loss": 1.0081,
      "step": 13031
    },
    {
      "epoch": 2.421854673852444,
      "grad_norm": 0.9368285536766052,
      "learning_rate": 6.767626922449244e-05,
      "loss": 0.9753,
      "step": 13032
    },
    {
      "epoch": 2.4220405129158147,
      "grad_norm": 0.876006007194519,
      "learning_rate": 6.766244448816859e-05,
      "loss": 0.9372,
      "step": 13033
    },
    {
      "epoch": 2.422226351979186,
      "grad_norm": 0.9135903716087341,
      "learning_rate": 6.764862044203439e-05,
      "loss": 0.9031,
      "step": 13034
    },
    {
      "epoch": 2.4224121910425573,
      "grad_norm": 0.8877546787261963,
      "learning_rate": 6.763479708638485e-05,
      "loss": 0.9847,
      "step": 13035
    },
    {
      "epoch": 2.4225980301059282,
      "grad_norm": 0.7598859071731567,
      "learning_rate": 6.762097442151503e-05,
      "loss": 0.9687,
      "step": 13036
    },
    {
      "epoch": 2.4227838691692996,
      "grad_norm": 1.9608190059661865,
      "learning_rate": 6.760715244771995e-05,
      "loss": 1.5244,
      "step": 13037
    },
    {
      "epoch": 2.4229697082326704,
      "grad_norm": 0.7144572734832764,
      "learning_rate": 6.759333116529461e-05,
      "loss": 0.8079,
      "step": 13038
    },
    {
      "epoch": 2.4231555472960418,
      "grad_norm": 0.8622708916664124,
      "learning_rate": 6.757951057453402e-05,
      "loss": 0.9636,
      "step": 13039
    },
    {
      "epoch": 2.4233413863594127,
      "grad_norm": 0.9091222286224365,
      "learning_rate": 6.756569067573313e-05,
      "loss": 0.8942,
      "step": 13040
    },
    {
      "epoch": 2.423527225422784,
      "grad_norm": 0.8898884057998657,
      "learning_rate": 6.755187146918691e-05,
      "loss": 0.9219,
      "step": 13041
    },
    {
      "epoch": 2.423713064486155,
      "grad_norm": 0.8423311114311218,
      "learning_rate": 6.753805295519028e-05,
      "loss": 0.9356,
      "step": 13042
    },
    {
      "epoch": 2.423898903549526,
      "grad_norm": 0.7890492677688599,
      "learning_rate": 6.752423513403824e-05,
      "loss": 0.7311,
      "step": 13043
    },
    {
      "epoch": 2.424084742612897,
      "grad_norm": 0.8877571225166321,
      "learning_rate": 6.751041800602566e-05,
      "loss": 0.86,
      "step": 13044
    },
    {
      "epoch": 2.4242705816762684,
      "grad_norm": 0.7152068018913269,
      "learning_rate": 6.749660157144746e-05,
      "loss": 0.9282,
      "step": 13045
    },
    {
      "epoch": 2.4244564207396393,
      "grad_norm": 0.7144151329994202,
      "learning_rate": 6.74827858305985e-05,
      "loss": 0.7036,
      "step": 13046
    },
    {
      "epoch": 2.4246422598030106,
      "grad_norm": 0.8931900858879089,
      "learning_rate": 6.746897078377372e-05,
      "loss": 1.0599,
      "step": 13047
    },
    {
      "epoch": 2.4248280988663815,
      "grad_norm": 0.7475230097770691,
      "learning_rate": 6.745515643126789e-05,
      "loss": 0.6593,
      "step": 13048
    },
    {
      "epoch": 2.425013937929753,
      "grad_norm": 0.744490921497345,
      "learning_rate": 6.744134277337592e-05,
      "loss": 0.9392,
      "step": 13049
    },
    {
      "epoch": 2.425199776993124,
      "grad_norm": 0.7806805968284607,
      "learning_rate": 6.742752981039262e-05,
      "loss": 1.1194,
      "step": 13050
    },
    {
      "epoch": 2.425385616056495,
      "grad_norm": 0.787963330745697,
      "learning_rate": 6.741371754261278e-05,
      "loss": 0.934,
      "step": 13051
    },
    {
      "epoch": 2.4255714551198664,
      "grad_norm": 1.080190896987915,
      "learning_rate": 6.739990597033124e-05,
      "loss": 0.8801,
      "step": 13052
    },
    {
      "epoch": 2.4257572941832373,
      "grad_norm": 0.8024685978889465,
      "learning_rate": 6.738609509384272e-05,
      "loss": 0.8261,
      "step": 13053
    },
    {
      "epoch": 2.4259431332466086,
      "grad_norm": 0.7282275557518005,
      "learning_rate": 6.737228491344207e-05,
      "loss": 0.9063,
      "step": 13054
    },
    {
      "epoch": 2.4261289723099795,
      "grad_norm": 0.8026573061943054,
      "learning_rate": 6.7358475429424e-05,
      "loss": 0.9185,
      "step": 13055
    },
    {
      "epoch": 2.426314811373351,
      "grad_norm": 0.8870404362678528,
      "learning_rate": 6.734466664208324e-05,
      "loss": 1.0354,
      "step": 13056
    },
    {
      "epoch": 2.4265006504367217,
      "grad_norm": 0.7788059115409851,
      "learning_rate": 6.733085855171457e-05,
      "loss": 0.8882,
      "step": 13057
    },
    {
      "epoch": 2.426686489500093,
      "grad_norm": 0.8050462007522583,
      "learning_rate": 6.731705115861262e-05,
      "loss": 0.8745,
      "step": 13058
    },
    {
      "epoch": 2.426872328563464,
      "grad_norm": 0.9904711246490479,
      "learning_rate": 6.730324446307217e-05,
      "loss": 1.1831,
      "step": 13059
    },
    {
      "epoch": 2.4270581676268352,
      "grad_norm": 0.8144270181655884,
      "learning_rate": 6.728943846538782e-05,
      "loss": 0.9835,
      "step": 13060
    },
    {
      "epoch": 2.427244006690206,
      "grad_norm": 0.7658560872077942,
      "learning_rate": 6.72756331658543e-05,
      "loss": 0.849,
      "step": 13061
    },
    {
      "epoch": 2.4274298457535775,
      "grad_norm": 0.7519936561584473,
      "learning_rate": 6.726182856476621e-05,
      "loss": 0.8053,
      "step": 13062
    },
    {
      "epoch": 2.4276156848169483,
      "grad_norm": 0.7167922854423523,
      "learning_rate": 6.724802466241819e-05,
      "loss": 0.8097,
      "step": 13063
    },
    {
      "epoch": 2.4278015238803197,
      "grad_norm": 0.8672042489051819,
      "learning_rate": 6.723422145910492e-05,
      "loss": 0.9202,
      "step": 13064
    },
    {
      "epoch": 2.427987362943691,
      "grad_norm": 0.9349759817123413,
      "learning_rate": 6.722041895512094e-05,
      "loss": 1.0381,
      "step": 13065
    },
    {
      "epoch": 2.428173202007062,
      "grad_norm": 0.8431971073150635,
      "learning_rate": 6.720661715076087e-05,
      "loss": 1.1296,
      "step": 13066
    },
    {
      "epoch": 2.4283590410704328,
      "grad_norm": 0.7165058851242065,
      "learning_rate": 6.71928160463193e-05,
      "loss": 0.9111,
      "step": 13067
    },
    {
      "epoch": 2.428544880133804,
      "grad_norm": 0.9309887290000916,
      "learning_rate": 6.717901564209075e-05,
      "loss": 1.0847,
      "step": 13068
    },
    {
      "epoch": 2.4287307191971754,
      "grad_norm": 0.7920277118682861,
      "learning_rate": 6.71652159383698e-05,
      "loss": 1.044,
      "step": 13069
    },
    {
      "epoch": 2.4289165582605463,
      "grad_norm": 0.9909117221832275,
      "learning_rate": 6.715141693545098e-05,
      "loss": 1.2691,
      "step": 13070
    },
    {
      "epoch": 2.4291023973239176,
      "grad_norm": 1.0344980955123901,
      "learning_rate": 6.713761863362878e-05,
      "loss": 0.7816,
      "step": 13071
    },
    {
      "epoch": 2.4292882363872885,
      "grad_norm": 0.8635357022285461,
      "learning_rate": 6.712382103319771e-05,
      "loss": 1.0432,
      "step": 13072
    },
    {
      "epoch": 2.42947407545066,
      "grad_norm": 0.8340578675270081,
      "learning_rate": 6.711002413445229e-05,
      "loss": 0.7909,
      "step": 13073
    },
    {
      "epoch": 2.4296599145140307,
      "grad_norm": 0.6920415759086609,
      "learning_rate": 6.709622793768691e-05,
      "loss": 0.9075,
      "step": 13074
    },
    {
      "epoch": 2.429845753577402,
      "grad_norm": 0.7110925316810608,
      "learning_rate": 6.708243244319611e-05,
      "loss": 0.8178,
      "step": 13075
    },
    {
      "epoch": 2.430031592640773,
      "grad_norm": 0.8715147376060486,
      "learning_rate": 6.706863765127429e-05,
      "loss": 0.9181,
      "step": 13076
    },
    {
      "epoch": 2.4302174317041443,
      "grad_norm": 0.7485066652297974,
      "learning_rate": 6.705484356221591e-05,
      "loss": 0.9947,
      "step": 13077
    },
    {
      "epoch": 2.430403270767515,
      "grad_norm": 0.83921217918396,
      "learning_rate": 6.704105017631534e-05,
      "loss": 0.8474,
      "step": 13078
    },
    {
      "epoch": 2.4305891098308865,
      "grad_norm": 0.8253395557403564,
      "learning_rate": 6.702725749386703e-05,
      "loss": 0.9027,
      "step": 13079
    },
    {
      "epoch": 2.4307749488942574,
      "grad_norm": 0.8697446584701538,
      "learning_rate": 6.701346551516528e-05,
      "loss": 1.0753,
      "step": 13080
    },
    {
      "epoch": 2.4309607879576287,
      "grad_norm": 0.7758882641792297,
      "learning_rate": 6.699967424050453e-05,
      "loss": 1.0756,
      "step": 13081
    },
    {
      "epoch": 2.4311466270209996,
      "grad_norm": 0.7584431767463684,
      "learning_rate": 6.698588367017909e-05,
      "loss": 0.9126,
      "step": 13082
    },
    {
      "epoch": 2.431332466084371,
      "grad_norm": 0.7834359407424927,
      "learning_rate": 6.697209380448333e-05,
      "loss": 0.8498,
      "step": 13083
    },
    {
      "epoch": 2.4315183051477423,
      "grad_norm": 0.6805732250213623,
      "learning_rate": 6.69583046437115e-05,
      "loss": 0.7568,
      "step": 13084
    },
    {
      "epoch": 2.431704144211113,
      "grad_norm": 0.9775819778442383,
      "learning_rate": 6.694451618815797e-05,
      "loss": 0.9638,
      "step": 13085
    },
    {
      "epoch": 2.4318899832744845,
      "grad_norm": 1.0010170936584473,
      "learning_rate": 6.693072843811703e-05,
      "loss": 1.1458,
      "step": 13086
    },
    {
      "epoch": 2.4320758223378554,
      "grad_norm": 0.8021188974380493,
      "learning_rate": 6.691694139388295e-05,
      "loss": 1.0033,
      "step": 13087
    },
    {
      "epoch": 2.4322616614012267,
      "grad_norm": 0.834514856338501,
      "learning_rate": 6.690315505574997e-05,
      "loss": 0.8317,
      "step": 13088
    },
    {
      "epoch": 2.4324475004645976,
      "grad_norm": 0.8626028299331665,
      "learning_rate": 6.688936942401237e-05,
      "loss": 0.9276,
      "step": 13089
    },
    {
      "epoch": 2.432633339527969,
      "grad_norm": 0.9956173300743103,
      "learning_rate": 6.687558449896434e-05,
      "loss": 0.7991,
      "step": 13090
    },
    {
      "epoch": 2.43281917859134,
      "grad_norm": 0.7554524540901184,
      "learning_rate": 6.686180028090014e-05,
      "loss": 0.8468,
      "step": 13091
    },
    {
      "epoch": 2.433005017654711,
      "grad_norm": 0.9718710780143738,
      "learning_rate": 6.684801677011392e-05,
      "loss": 1.2913,
      "step": 13092
    },
    {
      "epoch": 2.433190856718082,
      "grad_norm": 0.7081608772277832,
      "learning_rate": 6.683423396689991e-05,
      "loss": 0.6683,
      "step": 13093
    },
    {
      "epoch": 2.4333766957814533,
      "grad_norm": 0.7922672033309937,
      "learning_rate": 6.682045187155223e-05,
      "loss": 0.8012,
      "step": 13094
    },
    {
      "epoch": 2.433562534844824,
      "grad_norm": 0.7638914585113525,
      "learning_rate": 6.680667048436506e-05,
      "loss": 0.8774,
      "step": 13095
    },
    {
      "epoch": 2.4337483739081955,
      "grad_norm": 0.931211531162262,
      "learning_rate": 6.679288980563256e-05,
      "loss": 0.9557,
      "step": 13096
    },
    {
      "epoch": 2.4339342129715664,
      "grad_norm": 0.8707103729248047,
      "learning_rate": 6.677910983564888e-05,
      "loss": 0.908,
      "step": 13097
    },
    {
      "epoch": 2.4341200520349378,
      "grad_norm": 0.8395317196846008,
      "learning_rate": 6.676533057470806e-05,
      "loss": 0.9026,
      "step": 13098
    },
    {
      "epoch": 2.434305891098309,
      "grad_norm": 0.7544784545898438,
      "learning_rate": 6.675155202310424e-05,
      "loss": 0.9046,
      "step": 13099
    },
    {
      "epoch": 2.43449173016168,
      "grad_norm": 0.9859960675239563,
      "learning_rate": 6.673777418113147e-05,
      "loss": 0.8512,
      "step": 13100
    },
    {
      "epoch": 2.4346775692250513,
      "grad_norm": 0.7868154048919678,
      "learning_rate": 6.672399704908384e-05,
      "loss": 0.7508,
      "step": 13101
    },
    {
      "epoch": 2.434863408288422,
      "grad_norm": 0.866687536239624,
      "learning_rate": 6.671022062725539e-05,
      "loss": 0.9324,
      "step": 13102
    },
    {
      "epoch": 2.4350492473517935,
      "grad_norm": 0.9121086597442627,
      "learning_rate": 6.669644491594015e-05,
      "loss": 1.1329,
      "step": 13103
    },
    {
      "epoch": 2.4352350864151644,
      "grad_norm": 0.811203122138977,
      "learning_rate": 6.668266991543212e-05,
      "loss": 0.8548,
      "step": 13104
    },
    {
      "epoch": 2.4354209254785357,
      "grad_norm": 0.830508291721344,
      "learning_rate": 6.666889562602532e-05,
      "loss": 0.9622,
      "step": 13105
    },
    {
      "epoch": 2.4356067645419066,
      "grad_norm": 0.9460077881813049,
      "learning_rate": 6.665512204801376e-05,
      "loss": 0.9933,
      "step": 13106
    },
    {
      "epoch": 2.435792603605278,
      "grad_norm": 1.0581727027893066,
      "learning_rate": 6.664134918169143e-05,
      "loss": 0.9589,
      "step": 13107
    },
    {
      "epoch": 2.435978442668649,
      "grad_norm": 0.9143171906471252,
      "learning_rate": 6.662757702735223e-05,
      "loss": 0.9227,
      "step": 13108
    },
    {
      "epoch": 2.43616428173202,
      "grad_norm": 0.8102787733078003,
      "learning_rate": 6.661380558529013e-05,
      "loss": 0.8932,
      "step": 13109
    },
    {
      "epoch": 2.436350120795391,
      "grad_norm": 0.7931554317474365,
      "learning_rate": 6.660003485579907e-05,
      "loss": 0.8818,
      "step": 13110
    },
    {
      "epoch": 2.4365359598587624,
      "grad_norm": 0.886027455329895,
      "learning_rate": 6.658626483917295e-05,
      "loss": 0.9567,
      "step": 13111
    },
    {
      "epoch": 2.4367217989221333,
      "grad_norm": 0.7048546671867371,
      "learning_rate": 6.657249553570563e-05,
      "loss": 0.8424,
      "step": 13112
    },
    {
      "epoch": 2.4369076379855046,
      "grad_norm": 0.8832733035087585,
      "learning_rate": 6.655872694569107e-05,
      "loss": 0.966,
      "step": 13113
    },
    {
      "epoch": 2.437093477048876,
      "grad_norm": 0.87441486120224,
      "learning_rate": 6.654495906942309e-05,
      "loss": 0.8732,
      "step": 13114
    },
    {
      "epoch": 2.437279316112247,
      "grad_norm": 0.8636944890022278,
      "learning_rate": 6.653119190719554e-05,
      "loss": 0.9355,
      "step": 13115
    },
    {
      "epoch": 2.4374651551756177,
      "grad_norm": 0.7891425490379333,
      "learning_rate": 6.651742545930222e-05,
      "loss": 0.9178,
      "step": 13116
    },
    {
      "epoch": 2.437650994238989,
      "grad_norm": 0.9344954490661621,
      "learning_rate": 6.650365972603705e-05,
      "loss": 1.0686,
      "step": 13117
    },
    {
      "epoch": 2.4378368333023603,
      "grad_norm": 1.0192102193832397,
      "learning_rate": 6.648989470769376e-05,
      "loss": 1.1044,
      "step": 13118
    },
    {
      "epoch": 2.4380226723657312,
      "grad_norm": 0.8289742469787598,
      "learning_rate": 6.647613040456618e-05,
      "loss": 1.0682,
      "step": 13119
    },
    {
      "epoch": 2.4382085114291026,
      "grad_norm": 0.8244996070861816,
      "learning_rate": 6.646236681694805e-05,
      "loss": 1.0694,
      "step": 13120
    },
    {
      "epoch": 2.4383943504924734,
      "grad_norm": 0.857039213180542,
      "learning_rate": 6.644860394513317e-05,
      "loss": 0.889,
      "step": 13121
    },
    {
      "epoch": 2.4385801895558448,
      "grad_norm": 0.971135139465332,
      "learning_rate": 6.643484178941524e-05,
      "loss": 1.0055,
      "step": 13122
    },
    {
      "epoch": 2.4387660286192157,
      "grad_norm": 0.8699045777320862,
      "learning_rate": 6.642108035008803e-05,
      "loss": 0.8433,
      "step": 13123
    },
    {
      "epoch": 2.438951867682587,
      "grad_norm": 1.0612807273864746,
      "learning_rate": 6.640731962744521e-05,
      "loss": 0.8707,
      "step": 13124
    },
    {
      "epoch": 2.439137706745958,
      "grad_norm": 0.8168467283248901,
      "learning_rate": 6.639355962178053e-05,
      "loss": 0.8866,
      "step": 13125
    },
    {
      "epoch": 2.439323545809329,
      "grad_norm": 0.8367739319801331,
      "learning_rate": 6.63798003333876e-05,
      "loss": 0.9637,
      "step": 13126
    },
    {
      "epoch": 2.4395093848727,
      "grad_norm": 0.7842310667037964,
      "learning_rate": 6.636604176256019e-05,
      "loss": 0.903,
      "step": 13127
    },
    {
      "epoch": 2.4396952239360714,
      "grad_norm": 2.4124133586883545,
      "learning_rate": 6.635228390959187e-05,
      "loss": 1.5411,
      "step": 13128
    },
    {
      "epoch": 2.4398810629994423,
      "grad_norm": 0.7578110098838806,
      "learning_rate": 6.633852677477635e-05,
      "loss": 0.881,
      "step": 13129
    },
    {
      "epoch": 2.4400669020628136,
      "grad_norm": 0.8522405624389648,
      "learning_rate": 6.632477035840718e-05,
      "loss": 1.0018,
      "step": 13130
    },
    {
      "epoch": 2.4402527411261845,
      "grad_norm": 0.6682462692260742,
      "learning_rate": 6.6311014660778e-05,
      "loss": 0.81,
      "step": 13131
    },
    {
      "epoch": 2.440438580189556,
      "grad_norm": 1.0970356464385986,
      "learning_rate": 6.62972596821824e-05,
      "loss": 0.7741,
      "step": 13132
    },
    {
      "epoch": 2.440624419252927,
      "grad_norm": 0.9331226944923401,
      "learning_rate": 6.628350542291397e-05,
      "loss": 0.9969,
      "step": 13133
    },
    {
      "epoch": 2.440810258316298,
      "grad_norm": 0.7452312707901001,
      "learning_rate": 6.626975188326624e-05,
      "loss": 0.7214,
      "step": 13134
    },
    {
      "epoch": 2.4409960973796694,
      "grad_norm": 0.740334689617157,
      "learning_rate": 6.62559990635328e-05,
      "loss": 0.8288,
      "step": 13135
    },
    {
      "epoch": 2.4411819364430403,
      "grad_norm": 0.8428611755371094,
      "learning_rate": 6.624224696400711e-05,
      "loss": 0.9081,
      "step": 13136
    },
    {
      "epoch": 2.4413677755064116,
      "grad_norm": 0.7663208246231079,
      "learning_rate": 6.622849558498274e-05,
      "loss": 0.9875,
      "step": 13137
    },
    {
      "epoch": 2.4415536145697825,
      "grad_norm": 1.5824609994888306,
      "learning_rate": 6.621474492675318e-05,
      "loss": 1.1943,
      "step": 13138
    },
    {
      "epoch": 2.441739453633154,
      "grad_norm": 0.8739747405052185,
      "learning_rate": 6.620099498961193e-05,
      "loss": 1.0732,
      "step": 13139
    },
    {
      "epoch": 2.4419252926965247,
      "grad_norm": 0.931546688079834,
      "learning_rate": 6.618724577385241e-05,
      "loss": 1.037,
      "step": 13140
    },
    {
      "epoch": 2.442111131759896,
      "grad_norm": 0.7744889259338379,
      "learning_rate": 6.617349727976814e-05,
      "loss": 0.8209,
      "step": 13141
    },
    {
      "epoch": 2.442296970823267,
      "grad_norm": 0.8622698783874512,
      "learning_rate": 6.615974950765249e-05,
      "loss": 0.638,
      "step": 13142
    },
    {
      "epoch": 2.4424828098866382,
      "grad_norm": 1.1232050657272339,
      "learning_rate": 6.614600245779894e-05,
      "loss": 1.1343,
      "step": 13143
    },
    {
      "epoch": 2.442668648950009,
      "grad_norm": 0.8128316402435303,
      "learning_rate": 6.613225613050085e-05,
      "loss": 0.9589,
      "step": 13144
    },
    {
      "epoch": 2.4428544880133805,
      "grad_norm": 0.8051408529281616,
      "learning_rate": 6.611851052605167e-05,
      "loss": 1.0296,
      "step": 13145
    },
    {
      "epoch": 2.4430403270767513,
      "grad_norm": 0.8059528470039368,
      "learning_rate": 6.610476564474471e-05,
      "loss": 0.9228,
      "step": 13146
    },
    {
      "epoch": 2.4432261661401227,
      "grad_norm": 0.8602081537246704,
      "learning_rate": 6.609102148687333e-05,
      "loss": 1.0438,
      "step": 13147
    },
    {
      "epoch": 2.443412005203494,
      "grad_norm": 0.8533175587654114,
      "learning_rate": 6.607727805273094e-05,
      "loss": 0.979,
      "step": 13148
    },
    {
      "epoch": 2.443597844266865,
      "grad_norm": 0.7846186757087708,
      "learning_rate": 6.606353534261085e-05,
      "loss": 0.9215,
      "step": 13149
    },
    {
      "epoch": 2.4437836833302358,
      "grad_norm": 0.7387604117393494,
      "learning_rate": 6.604979335680633e-05,
      "loss": 0.5422,
      "step": 13150
    },
    {
      "epoch": 2.443969522393607,
      "grad_norm": 0.8741371631622314,
      "learning_rate": 6.603605209561075e-05,
      "loss": 0.9177,
      "step": 13151
    },
    {
      "epoch": 2.4441553614569784,
      "grad_norm": 0.8448536992073059,
      "learning_rate": 6.602231155931731e-05,
      "loss": 0.9546,
      "step": 13152
    },
    {
      "epoch": 2.4443412005203493,
      "grad_norm": 0.8914835453033447,
      "learning_rate": 6.600857174821938e-05,
      "loss": 0.8356,
      "step": 13153
    },
    {
      "epoch": 2.4445270395837206,
      "grad_norm": 0.7318701148033142,
      "learning_rate": 6.599483266261009e-05,
      "loss": 0.855,
      "step": 13154
    },
    {
      "epoch": 2.4447128786470915,
      "grad_norm": 1.010781168937683,
      "learning_rate": 6.598109430278279e-05,
      "loss": 0.7821,
      "step": 13155
    },
    {
      "epoch": 2.444898717710463,
      "grad_norm": 0.8632301688194275,
      "learning_rate": 6.596735666903063e-05,
      "loss": 0.6817,
      "step": 13156
    },
    {
      "epoch": 2.4450845567738337,
      "grad_norm": 0.9101744294166565,
      "learning_rate": 6.595361976164683e-05,
      "loss": 0.8722,
      "step": 13157
    },
    {
      "epoch": 2.445270395837205,
      "grad_norm": 0.7859084010124207,
      "learning_rate": 6.593988358092458e-05,
      "loss": 1.0306,
      "step": 13158
    },
    {
      "epoch": 2.445456234900576,
      "grad_norm": 0.7916283011436462,
      "learning_rate": 6.59261481271571e-05,
      "loss": 0.9579,
      "step": 13159
    },
    {
      "epoch": 2.4456420739639473,
      "grad_norm": 0.7866398692131042,
      "learning_rate": 6.59124134006375e-05,
      "loss": 0.963,
      "step": 13160
    },
    {
      "epoch": 2.445827913027318,
      "grad_norm": 0.8245794773101807,
      "learning_rate": 6.589867940165895e-05,
      "loss": 0.9524,
      "step": 13161
    },
    {
      "epoch": 2.4460137520906895,
      "grad_norm": 0.7416488528251648,
      "learning_rate": 6.588494613051455e-05,
      "loss": 1.0268,
      "step": 13162
    },
    {
      "epoch": 2.446199591154061,
      "grad_norm": 0.8672488927841187,
      "learning_rate": 6.587121358749745e-05,
      "loss": 0.9448,
      "step": 13163
    },
    {
      "epoch": 2.4463854302174317,
      "grad_norm": 0.8468174934387207,
      "learning_rate": 6.585748177290072e-05,
      "loss": 0.9216,
      "step": 13164
    },
    {
      "epoch": 2.4465712692808026,
      "grad_norm": 0.6494128704071045,
      "learning_rate": 6.584375068701746e-05,
      "loss": 0.6241,
      "step": 13165
    },
    {
      "epoch": 2.446757108344174,
      "grad_norm": 0.67799973487854,
      "learning_rate": 6.583002033014071e-05,
      "loss": 0.7008,
      "step": 13166
    },
    {
      "epoch": 2.4469429474075453,
      "grad_norm": 0.7948530912399292,
      "learning_rate": 6.581629070256354e-05,
      "loss": 0.8765,
      "step": 13167
    },
    {
      "epoch": 2.447128786470916,
      "grad_norm": 0.641555905342102,
      "learning_rate": 6.580256180457895e-05,
      "loss": 0.7714,
      "step": 13168
    },
    {
      "epoch": 2.4473146255342875,
      "grad_norm": 0.8227325081825256,
      "learning_rate": 6.578883363648005e-05,
      "loss": 1.0227,
      "step": 13169
    },
    {
      "epoch": 2.4475004645976584,
      "grad_norm": 0.7780269384384155,
      "learning_rate": 6.577510619855977e-05,
      "loss": 0.9948,
      "step": 13170
    },
    {
      "epoch": 2.4476863036610297,
      "grad_norm": 0.9374097585678101,
      "learning_rate": 6.576137949111113e-05,
      "loss": 1.0773,
      "step": 13171
    },
    {
      "epoch": 2.4478721427244006,
      "grad_norm": 0.7884427905082703,
      "learning_rate": 6.574765351442707e-05,
      "loss": 0.9597,
      "step": 13172
    },
    {
      "epoch": 2.448057981787772,
      "grad_norm": 0.782262921333313,
      "learning_rate": 6.573392826880059e-05,
      "loss": 0.7153,
      "step": 13173
    },
    {
      "epoch": 2.448243820851143,
      "grad_norm": 0.8168174624443054,
      "learning_rate": 6.572020375452457e-05,
      "loss": 0.7016,
      "step": 13174
    },
    {
      "epoch": 2.448429659914514,
      "grad_norm": 0.8480477929115295,
      "learning_rate": 6.570647997189201e-05,
      "loss": 0.823,
      "step": 13175
    },
    {
      "epoch": 2.448615498977885,
      "grad_norm": 0.8476222157478333,
      "learning_rate": 6.569275692119576e-05,
      "loss": 0.7272,
      "step": 13176
    },
    {
      "epoch": 2.4488013380412563,
      "grad_norm": 0.8779712915420532,
      "learning_rate": 6.567903460272874e-05,
      "loss": 0.9677,
      "step": 13177
    },
    {
      "epoch": 2.448987177104627,
      "grad_norm": 0.8764821290969849,
      "learning_rate": 6.566531301678384e-05,
      "loss": 1.1981,
      "step": 13178
    },
    {
      "epoch": 2.4491730161679985,
      "grad_norm": 0.8151456117630005,
      "learning_rate": 6.565159216365389e-05,
      "loss": 0.9573,
      "step": 13179
    },
    {
      "epoch": 2.4493588552313694,
      "grad_norm": 0.8219358325004578,
      "learning_rate": 6.563787204363177e-05,
      "loss": 0.8288,
      "step": 13180
    },
    {
      "epoch": 2.4495446942947408,
      "grad_norm": 0.9229908585548401,
      "learning_rate": 6.562415265701034e-05,
      "loss": 1.1221,
      "step": 13181
    },
    {
      "epoch": 2.449730533358112,
      "grad_norm": 0.8629090189933777,
      "learning_rate": 6.561043400408234e-05,
      "loss": 0.9825,
      "step": 13182
    },
    {
      "epoch": 2.449916372421483,
      "grad_norm": 0.8018071055412292,
      "learning_rate": 6.559671608514065e-05,
      "loss": 1.011,
      "step": 13183
    },
    {
      "epoch": 2.4501022114848543,
      "grad_norm": 0.8450409173965454,
      "learning_rate": 6.558299890047798e-05,
      "loss": 0.8215,
      "step": 13184
    },
    {
      "epoch": 2.450288050548225,
      "grad_norm": 0.7382701635360718,
      "learning_rate": 6.556928245038716e-05,
      "loss": 0.8025,
      "step": 13185
    },
    {
      "epoch": 2.4504738896115965,
      "grad_norm": 0.7963736057281494,
      "learning_rate": 6.55555667351609e-05,
      "loss": 0.818,
      "step": 13186
    },
    {
      "epoch": 2.4506597286749674,
      "grad_norm": 1.4558244943618774,
      "learning_rate": 6.554185175509197e-05,
      "loss": 1.3749,
      "step": 13187
    },
    {
      "epoch": 2.4508455677383387,
      "grad_norm": 1.0389723777770996,
      "learning_rate": 6.552813751047309e-05,
      "loss": 1.1356,
      "step": 13188
    },
    {
      "epoch": 2.4510314068017096,
      "grad_norm": 0.6506426930427551,
      "learning_rate": 6.551442400159692e-05,
      "loss": 0.6469,
      "step": 13189
    },
    {
      "epoch": 2.451217245865081,
      "grad_norm": 0.739996612071991,
      "learning_rate": 6.550071122875621e-05,
      "loss": 0.8171,
      "step": 13190
    },
    {
      "epoch": 2.451403084928452,
      "grad_norm": 0.821476399898529,
      "learning_rate": 6.548699919224364e-05,
      "loss": 0.9817,
      "step": 13191
    },
    {
      "epoch": 2.451588923991823,
      "grad_norm": 1.0348422527313232,
      "learning_rate": 6.547328789235183e-05,
      "loss": 1.2025,
      "step": 13192
    },
    {
      "epoch": 2.451774763055194,
      "grad_norm": 0.7092103958129883,
      "learning_rate": 6.545957732937345e-05,
      "loss": 0.7236,
      "step": 13193
    },
    {
      "epoch": 2.4519606021185654,
      "grad_norm": 1.1522830724716187,
      "learning_rate": 6.54458675036011e-05,
      "loss": 1.1493,
      "step": 13194
    },
    {
      "epoch": 2.4521464411819363,
      "grad_norm": 0.945174515247345,
      "learning_rate": 6.543215841532744e-05,
      "loss": 1.1024,
      "step": 13195
    },
    {
      "epoch": 2.4523322802453076,
      "grad_norm": 0.9081987738609314,
      "learning_rate": 6.5418450064845e-05,
      "loss": 1.2248,
      "step": 13196
    },
    {
      "epoch": 2.452518119308679,
      "grad_norm": 0.8504126667976379,
      "learning_rate": 6.54047424524464e-05,
      "loss": 1.0508,
      "step": 13197
    },
    {
      "epoch": 2.45270395837205,
      "grad_norm": 0.8007256388664246,
      "learning_rate": 6.539103557842425e-05,
      "loss": 0.8587,
      "step": 13198
    },
    {
      "epoch": 2.4528897974354207,
      "grad_norm": 1.0808576345443726,
      "learning_rate": 6.537732944307102e-05,
      "loss": 0.8679,
      "step": 13199
    },
    {
      "epoch": 2.453075636498792,
      "grad_norm": 0.7359915971755981,
      "learning_rate": 6.536362404667924e-05,
      "loss": 0.7308,
      "step": 13200
    },
    {
      "epoch": 2.4532614755621633,
      "grad_norm": 0.7679325342178345,
      "learning_rate": 6.534991938954154e-05,
      "loss": 0.9018,
      "step": 13201
    },
    {
      "epoch": 2.4534473146255342,
      "grad_norm": 0.8206538558006287,
      "learning_rate": 6.53362154719503e-05,
      "loss": 0.7636,
      "step": 13202
    },
    {
      "epoch": 2.4536331536889056,
      "grad_norm": 0.9894808530807495,
      "learning_rate": 6.53225122941981e-05,
      "loss": 0.9901,
      "step": 13203
    },
    {
      "epoch": 2.4538189927522764,
      "grad_norm": 0.8593584895133972,
      "learning_rate": 6.530880985657733e-05,
      "loss": 0.7279,
      "step": 13204
    },
    {
      "epoch": 2.4540048318156478,
      "grad_norm": 1.3301892280578613,
      "learning_rate": 6.52951081593805e-05,
      "loss": 1.1438,
      "step": 13205
    },
    {
      "epoch": 2.4541906708790187,
      "grad_norm": 0.8623514175415039,
      "learning_rate": 6.528140720290001e-05,
      "loss": 0.8775,
      "step": 13206
    },
    {
      "epoch": 2.45437650994239,
      "grad_norm": 0.7796410322189331,
      "learning_rate": 6.526770698742831e-05,
      "loss": 0.8452,
      "step": 13207
    },
    {
      "epoch": 2.454562349005761,
      "grad_norm": 0.8087790608406067,
      "learning_rate": 6.525400751325783e-05,
      "loss": 0.8721,
      "step": 13208
    },
    {
      "epoch": 2.454748188069132,
      "grad_norm": 0.8593664765357971,
      "learning_rate": 6.524030878068091e-05,
      "loss": 1.0969,
      "step": 13209
    },
    {
      "epoch": 2.454934027132503,
      "grad_norm": 0.7332558631896973,
      "learning_rate": 6.52266107899899e-05,
      "loss": 0.9816,
      "step": 13210
    },
    {
      "epoch": 2.4551198661958744,
      "grad_norm": 0.9154184460639954,
      "learning_rate": 6.521291354147727e-05,
      "loss": 1.099,
      "step": 13211
    },
    {
      "epoch": 2.4553057052592457,
      "grad_norm": 0.8243273496627808,
      "learning_rate": 6.51992170354353e-05,
      "loss": 1.0562,
      "step": 13212
    },
    {
      "epoch": 2.4554915443226166,
      "grad_norm": 0.9075212478637695,
      "learning_rate": 6.518552127215632e-05,
      "loss": 1.0337,
      "step": 13213
    },
    {
      "epoch": 2.4556773833859875,
      "grad_norm": 0.7624395489692688,
      "learning_rate": 6.517182625193263e-05,
      "loss": 1.1125,
      "step": 13214
    },
    {
      "epoch": 2.455863222449359,
      "grad_norm": 1.781527042388916,
      "learning_rate": 6.515813197505656e-05,
      "loss": 1.4969,
      "step": 13215
    },
    {
      "epoch": 2.45604906151273,
      "grad_norm": 0.8518239855766296,
      "learning_rate": 6.514443844182035e-05,
      "loss": 1.0883,
      "step": 13216
    },
    {
      "epoch": 2.456234900576101,
      "grad_norm": 0.7943968176841736,
      "learning_rate": 6.513074565251629e-05,
      "loss": 0.8723,
      "step": 13217
    },
    {
      "epoch": 2.4564207396394724,
      "grad_norm": 0.8362912535667419,
      "learning_rate": 6.511705360743665e-05,
      "loss": 1.0535,
      "step": 13218
    },
    {
      "epoch": 2.4566065787028433,
      "grad_norm": 0.8788190484046936,
      "learning_rate": 6.510336230687362e-05,
      "loss": 0.9461,
      "step": 13219
    },
    {
      "epoch": 2.4567924177662146,
      "grad_norm": 1.0436252355575562,
      "learning_rate": 6.508967175111945e-05,
      "loss": 1.0679,
      "step": 13220
    },
    {
      "epoch": 2.4569782568295855,
      "grad_norm": 0.8631687760353088,
      "learning_rate": 6.507598194046629e-05,
      "loss": 0.9971,
      "step": 13221
    },
    {
      "epoch": 2.457164095892957,
      "grad_norm": 0.816724956035614,
      "learning_rate": 6.506229287520638e-05,
      "loss": 0.939,
      "step": 13222
    },
    {
      "epoch": 2.4573499349563277,
      "grad_norm": 0.8277599215507507,
      "learning_rate": 6.50486045556319e-05,
      "loss": 0.9906,
      "step": 13223
    },
    {
      "epoch": 2.457535774019699,
      "grad_norm": 0.7382860779762268,
      "learning_rate": 6.503491698203496e-05,
      "loss": 0.8693,
      "step": 13224
    },
    {
      "epoch": 2.45772161308307,
      "grad_norm": 0.767471194267273,
      "learning_rate": 6.502123015470772e-05,
      "loss": 0.6254,
      "step": 13225
    },
    {
      "epoch": 2.4579074521464412,
      "grad_norm": 0.8623015880584717,
      "learning_rate": 6.50075440739423e-05,
      "loss": 0.7919,
      "step": 13226
    },
    {
      "epoch": 2.458093291209812,
      "grad_norm": 0.7773959040641785,
      "learning_rate": 6.499385874003076e-05,
      "loss": 0.9748,
      "step": 13227
    },
    {
      "epoch": 2.4582791302731835,
      "grad_norm": 0.8545732498168945,
      "learning_rate": 6.49801741532653e-05,
      "loss": 1.1325,
      "step": 13228
    },
    {
      "epoch": 2.4584649693365543,
      "grad_norm": 0.7644531726837158,
      "learning_rate": 6.496649031393788e-05,
      "loss": 0.9449,
      "step": 13229
    },
    {
      "epoch": 2.4586508083999257,
      "grad_norm": 0.8056524991989136,
      "learning_rate": 6.495280722234063e-05,
      "loss": 1.0851,
      "step": 13230
    },
    {
      "epoch": 2.458836647463297,
      "grad_norm": 0.6984783411026001,
      "learning_rate": 6.493912487876553e-05,
      "loss": 1.0835,
      "step": 13231
    },
    {
      "epoch": 2.459022486526668,
      "grad_norm": 0.7677348256111145,
      "learning_rate": 6.492544328350465e-05,
      "loss": 0.8166,
      "step": 13232
    },
    {
      "epoch": 2.459208325590039,
      "grad_norm": 0.6787754893302917,
      "learning_rate": 6.491176243685003e-05,
      "loss": 0.6079,
      "step": 13233
    },
    {
      "epoch": 2.45939416465341,
      "grad_norm": 1.002785086631775,
      "learning_rate": 6.489808233909359e-05,
      "loss": 0.9857,
      "step": 13234
    },
    {
      "epoch": 2.4595800037167814,
      "grad_norm": 0.7710455060005188,
      "learning_rate": 6.488440299052738e-05,
      "loss": 1.0792,
      "step": 13235
    },
    {
      "epoch": 2.4597658427801523,
      "grad_norm": 0.8439406752586365,
      "learning_rate": 6.487072439144331e-05,
      "loss": 0.9279,
      "step": 13236
    },
    {
      "epoch": 2.4599516818435236,
      "grad_norm": 0.6731829643249512,
      "learning_rate": 6.485704654213331e-05,
      "loss": 0.8066,
      "step": 13237
    },
    {
      "epoch": 2.4601375209068945,
      "grad_norm": 0.922251284122467,
      "learning_rate": 6.48433694428894e-05,
      "loss": 0.9569,
      "step": 13238
    },
    {
      "epoch": 2.460323359970266,
      "grad_norm": 0.8056465983390808,
      "learning_rate": 6.482969309400341e-05,
      "loss": 0.8648,
      "step": 13239
    },
    {
      "epoch": 2.4605091990336367,
      "grad_norm": 0.8848257660865784,
      "learning_rate": 6.481601749576728e-05,
      "loss": 0.7394,
      "step": 13240
    },
    {
      "epoch": 2.460695038097008,
      "grad_norm": 0.7655457854270935,
      "learning_rate": 6.480234264847286e-05,
      "loss": 0.938,
      "step": 13241
    },
    {
      "epoch": 2.460880877160379,
      "grad_norm": 0.8142226934432983,
      "learning_rate": 6.478866855241202e-05,
      "loss": 0.912,
      "step": 13242
    },
    {
      "epoch": 2.4610667162237503,
      "grad_norm": 0.9077260494232178,
      "learning_rate": 6.477499520787665e-05,
      "loss": 0.9718,
      "step": 13243
    },
    {
      "epoch": 2.461252555287121,
      "grad_norm": 0.8445703387260437,
      "learning_rate": 6.476132261515854e-05,
      "loss": 0.9174,
      "step": 13244
    },
    {
      "epoch": 2.4614383943504925,
      "grad_norm": 0.8891257643699646,
      "learning_rate": 6.474765077454957e-05,
      "loss": 0.9767,
      "step": 13245
    },
    {
      "epoch": 2.461624233413864,
      "grad_norm": 0.787733793258667,
      "learning_rate": 6.473397968634147e-05,
      "loss": 0.9266,
      "step": 13246
    },
    {
      "epoch": 2.4618100724772347,
      "grad_norm": 0.8732960224151611,
      "learning_rate": 6.472030935082605e-05,
      "loss": 0.8442,
      "step": 13247
    },
    {
      "epoch": 2.4619959115406056,
      "grad_norm": 0.9979159235954285,
      "learning_rate": 6.470663976829509e-05,
      "loss": 1.0616,
      "step": 13248
    },
    {
      "epoch": 2.462181750603977,
      "grad_norm": 0.8794299960136414,
      "learning_rate": 6.469297093904035e-05,
      "loss": 0.9818,
      "step": 13249
    },
    {
      "epoch": 2.4623675896673483,
      "grad_norm": 0.9490398168563843,
      "learning_rate": 6.467930286335356e-05,
      "loss": 0.9454,
      "step": 13250
    },
    {
      "epoch": 2.462553428730719,
      "grad_norm": 0.8701783418655396,
      "learning_rate": 6.466563554152641e-05,
      "loss": 0.7952,
      "step": 13251
    },
    {
      "epoch": 2.4627392677940905,
      "grad_norm": 0.7783467769622803,
      "learning_rate": 6.465196897385062e-05,
      "loss": 0.9286,
      "step": 13252
    },
    {
      "epoch": 2.4629251068574614,
      "grad_norm": 0.7469505667686462,
      "learning_rate": 6.463830316061792e-05,
      "loss": 0.7903,
      "step": 13253
    },
    {
      "epoch": 2.4631109459208327,
      "grad_norm": 0.8126850724220276,
      "learning_rate": 6.462463810211995e-05,
      "loss": 0.9425,
      "step": 13254
    },
    {
      "epoch": 2.4632967849842036,
      "grad_norm": 0.7326277494430542,
      "learning_rate": 6.46109737986484e-05,
      "loss": 0.718,
      "step": 13255
    },
    {
      "epoch": 2.463482624047575,
      "grad_norm": 0.8058909773826599,
      "learning_rate": 6.459731025049484e-05,
      "loss": 1.518,
      "step": 13256
    },
    {
      "epoch": 2.463668463110946,
      "grad_norm": 0.8176577091217041,
      "learning_rate": 6.458364745795096e-05,
      "loss": 1.1653,
      "step": 13257
    },
    {
      "epoch": 2.463854302174317,
      "grad_norm": 0.6765969395637512,
      "learning_rate": 6.456998542130837e-05,
      "loss": 0.7163,
      "step": 13258
    },
    {
      "epoch": 2.464040141237688,
      "grad_norm": 0.7259505391120911,
      "learning_rate": 6.455632414085861e-05,
      "loss": 0.9289,
      "step": 13259
    },
    {
      "epoch": 2.4642259803010593,
      "grad_norm": 0.8830696940422058,
      "learning_rate": 6.454266361689331e-05,
      "loss": 1.0349,
      "step": 13260
    },
    {
      "epoch": 2.46441181936443,
      "grad_norm": 0.9436715245246887,
      "learning_rate": 6.4529003849704e-05,
      "loss": 0.9728,
      "step": 13261
    },
    {
      "epoch": 2.4645976584278015,
      "grad_norm": 0.7746622562408447,
      "learning_rate": 6.451534483958222e-05,
      "loss": 0.9486,
      "step": 13262
    },
    {
      "epoch": 2.4647834974911724,
      "grad_norm": 0.7761104702949524,
      "learning_rate": 6.45016865868195e-05,
      "loss": 1.0427,
      "step": 13263
    },
    {
      "epoch": 2.4649693365545438,
      "grad_norm": 0.7415323853492737,
      "learning_rate": 6.448802909170737e-05,
      "loss": 1.0287,
      "step": 13264
    },
    {
      "epoch": 2.465155175617915,
      "grad_norm": 0.7377958297729492,
      "learning_rate": 6.447437235453735e-05,
      "loss": 0.9365,
      "step": 13265
    },
    {
      "epoch": 2.465341014681286,
      "grad_norm": 0.9537144303321838,
      "learning_rate": 6.446071637560085e-05,
      "loss": 0.9043,
      "step": 13266
    },
    {
      "epoch": 2.4655268537446573,
      "grad_norm": 0.6393646597862244,
      "learning_rate": 6.444706115518938e-05,
      "loss": 0.9902,
      "step": 13267
    },
    {
      "epoch": 2.465712692808028,
      "grad_norm": 0.8335622549057007,
      "learning_rate": 6.44334066935944e-05,
      "loss": 0.7365,
      "step": 13268
    },
    {
      "epoch": 2.4658985318713995,
      "grad_norm": 0.8195977210998535,
      "learning_rate": 6.44197529911073e-05,
      "loss": 0.9792,
      "step": 13269
    },
    {
      "epoch": 2.4660843709347704,
      "grad_norm": 0.8189404606819153,
      "learning_rate": 6.440610004801953e-05,
      "loss": 1.0515,
      "step": 13270
    },
    {
      "epoch": 2.4662702099981417,
      "grad_norm": 0.8037553429603577,
      "learning_rate": 6.439244786462245e-05,
      "loss": 1.0898,
      "step": 13271
    },
    {
      "epoch": 2.4664560490615126,
      "grad_norm": 0.7144148945808411,
      "learning_rate": 6.43787964412075e-05,
      "loss": 0.8562,
      "step": 13272
    },
    {
      "epoch": 2.466641888124884,
      "grad_norm": 0.8356459736824036,
      "learning_rate": 6.436514577806595e-05,
      "loss": 0.9447,
      "step": 13273
    },
    {
      "epoch": 2.466827727188255,
      "grad_norm": 0.8574797511100769,
      "learning_rate": 6.435149587548925e-05,
      "loss": 0.9169,
      "step": 13274
    },
    {
      "epoch": 2.467013566251626,
      "grad_norm": 0.7633848190307617,
      "learning_rate": 6.43378467337687e-05,
      "loss": 1.0534,
      "step": 13275
    },
    {
      "epoch": 2.467199405314997,
      "grad_norm": 0.9943694472312927,
      "learning_rate": 6.43241983531956e-05,
      "loss": 0.9129,
      "step": 13276
    },
    {
      "epoch": 2.4673852443783684,
      "grad_norm": 0.779023289680481,
      "learning_rate": 6.431055073406127e-05,
      "loss": 0.9427,
      "step": 13277
    },
    {
      "epoch": 2.4675710834417393,
      "grad_norm": 0.832300066947937,
      "learning_rate": 6.429690387665702e-05,
      "loss": 0.8809,
      "step": 13278
    },
    {
      "epoch": 2.4677569225051106,
      "grad_norm": 0.8126018047332764,
      "learning_rate": 6.428325778127404e-05,
      "loss": 1.1159,
      "step": 13279
    },
    {
      "epoch": 2.467942761568482,
      "grad_norm": 0.8588595986366272,
      "learning_rate": 6.426961244820366e-05,
      "loss": 0.8322,
      "step": 13280
    },
    {
      "epoch": 2.468128600631853,
      "grad_norm": 0.8416522145271301,
      "learning_rate": 6.425596787773706e-05,
      "loss": 0.9026,
      "step": 13281
    },
    {
      "epoch": 2.468314439695224,
      "grad_norm": 0.756829023361206,
      "learning_rate": 6.424232407016553e-05,
      "loss": 0.8206,
      "step": 13282
    },
    {
      "epoch": 2.468500278758595,
      "grad_norm": 0.7184683084487915,
      "learning_rate": 6.422868102578018e-05,
      "loss": 0.6891,
      "step": 13283
    },
    {
      "epoch": 2.4686861178219663,
      "grad_norm": 0.9127999544143677,
      "learning_rate": 6.421503874487224e-05,
      "loss": 0.8975,
      "step": 13284
    },
    {
      "epoch": 2.4688719568853372,
      "grad_norm": 0.7633183002471924,
      "learning_rate": 6.420139722773296e-05,
      "loss": 0.9248,
      "step": 13285
    },
    {
      "epoch": 2.4690577959487086,
      "grad_norm": 0.8028041124343872,
      "learning_rate": 6.418775647465336e-05,
      "loss": 0.8665,
      "step": 13286
    },
    {
      "epoch": 2.4692436350120794,
      "grad_norm": 1.1864566802978516,
      "learning_rate": 6.417411648592466e-05,
      "loss": 1.0498,
      "step": 13287
    },
    {
      "epoch": 2.4694294740754508,
      "grad_norm": 0.7055113911628723,
      "learning_rate": 6.416047726183798e-05,
      "loss": 0.7556,
      "step": 13288
    },
    {
      "epoch": 2.4696153131388217,
      "grad_norm": 0.8412120938301086,
      "learning_rate": 6.41468388026844e-05,
      "loss": 0.9165,
      "step": 13289
    },
    {
      "epoch": 2.469801152202193,
      "grad_norm": 0.9094467759132385,
      "learning_rate": 6.413320110875502e-05,
      "loss": 0.9182,
      "step": 13290
    },
    {
      "epoch": 2.469986991265564,
      "grad_norm": 0.9734574556350708,
      "learning_rate": 6.41195641803409e-05,
      "loss": 1.0071,
      "step": 13291
    },
    {
      "epoch": 2.470172830328935,
      "grad_norm": 0.7539117336273193,
      "learning_rate": 6.410592801773313e-05,
      "loss": 0.9826,
      "step": 13292
    },
    {
      "epoch": 2.470358669392306,
      "grad_norm": 0.8051267266273499,
      "learning_rate": 6.409229262122271e-05,
      "loss": 0.7357,
      "step": 13293
    },
    {
      "epoch": 2.4705445084556774,
      "grad_norm": 1.0722055435180664,
      "learning_rate": 6.407865799110067e-05,
      "loss": 1.0554,
      "step": 13294
    },
    {
      "epoch": 2.4707303475190487,
      "grad_norm": 0.8269297480583191,
      "learning_rate": 6.406502412765807e-05,
      "loss": 0.8793,
      "step": 13295
    },
    {
      "epoch": 2.4709161865824196,
      "grad_norm": 0.8753021955490112,
      "learning_rate": 6.405139103118583e-05,
      "loss": 0.8462,
      "step": 13296
    },
    {
      "epoch": 2.4711020256457905,
      "grad_norm": 0.7838700413703918,
      "learning_rate": 6.403775870197497e-05,
      "loss": 0.973,
      "step": 13297
    },
    {
      "epoch": 2.471287864709162,
      "grad_norm": 0.7521423697471619,
      "learning_rate": 6.402412714031645e-05,
      "loss": 0.9985,
      "step": 13298
    },
    {
      "epoch": 2.471473703772533,
      "grad_norm": 0.8879348039627075,
      "learning_rate": 6.401049634650118e-05,
      "loss": 0.932,
      "step": 13299
    },
    {
      "epoch": 2.471659542835904,
      "grad_norm": 0.8718003034591675,
      "learning_rate": 6.399686632082014e-05,
      "loss": 0.9369,
      "step": 13300
    },
    {
      "epoch": 2.4718453818992754,
      "grad_norm": 1.2705172300338745,
      "learning_rate": 6.398323706356417e-05,
      "loss": 0.9817,
      "step": 13301
    },
    {
      "epoch": 2.4720312209626463,
      "grad_norm": 0.7764171361923218,
      "learning_rate": 6.396960857502422e-05,
      "loss": 0.8831,
      "step": 13302
    },
    {
      "epoch": 2.4722170600260176,
      "grad_norm": 0.8133841156959534,
      "learning_rate": 6.395598085549113e-05,
      "loss": 1.0403,
      "step": 13303
    },
    {
      "epoch": 2.4724028990893885,
      "grad_norm": 0.7242906093597412,
      "learning_rate": 6.394235390525579e-05,
      "loss": 0.8429,
      "step": 13304
    },
    {
      "epoch": 2.47258873815276,
      "grad_norm": 0.7564440965652466,
      "learning_rate": 6.392872772460898e-05,
      "loss": 1.0826,
      "step": 13305
    },
    {
      "epoch": 2.4727745772161307,
      "grad_norm": 0.7228856682777405,
      "learning_rate": 6.391510231384162e-05,
      "loss": 0.7916,
      "step": 13306
    },
    {
      "epoch": 2.472960416279502,
      "grad_norm": 0.7421514987945557,
      "learning_rate": 6.390147767324445e-05,
      "loss": 0.852,
      "step": 13307
    },
    {
      "epoch": 2.473146255342873,
      "grad_norm": 0.8280564546585083,
      "learning_rate": 6.388785380310833e-05,
      "loss": 0.8453,
      "step": 13308
    },
    {
      "epoch": 2.4733320944062442,
      "grad_norm": 0.8708766102790833,
      "learning_rate": 6.387423070372395e-05,
      "loss": 0.7161,
      "step": 13309
    },
    {
      "epoch": 2.473517933469615,
      "grad_norm": 0.9055882096290588,
      "learning_rate": 6.386060837538218e-05,
      "loss": 1.1629,
      "step": 13310
    },
    {
      "epoch": 2.4737037725329865,
      "grad_norm": 0.8641484975814819,
      "learning_rate": 6.384698681837366e-05,
      "loss": 0.8631,
      "step": 13311
    },
    {
      "epoch": 2.4738896115963573,
      "grad_norm": 0.9759421944618225,
      "learning_rate": 6.383336603298921e-05,
      "loss": 0.9113,
      "step": 13312
    },
    {
      "epoch": 2.4740754506597287,
      "grad_norm": 0.7598199844360352,
      "learning_rate": 6.381974601951945e-05,
      "loss": 0.9642,
      "step": 13313
    },
    {
      "epoch": 2.4742612897231,
      "grad_norm": 0.8624680638313293,
      "learning_rate": 6.380612677825518e-05,
      "loss": 0.9527,
      "step": 13314
    },
    {
      "epoch": 2.474447128786471,
      "grad_norm": 0.8249783515930176,
      "learning_rate": 6.379250830948694e-05,
      "loss": 0.8295,
      "step": 13315
    },
    {
      "epoch": 2.474632967849842,
      "grad_norm": 0.9170358777046204,
      "learning_rate": 6.377889061350551e-05,
      "loss": 0.8197,
      "step": 13316
    },
    {
      "epoch": 2.474818806913213,
      "grad_norm": 0.766761302947998,
      "learning_rate": 6.376527369060151e-05,
      "loss": 0.9006,
      "step": 13317
    },
    {
      "epoch": 2.4750046459765844,
      "grad_norm": 0.8031441569328308,
      "learning_rate": 6.375165754106558e-05,
      "loss": 0.9433,
      "step": 13318
    },
    {
      "epoch": 2.4751904850399553,
      "grad_norm": 0.8161746263504028,
      "learning_rate": 6.37380421651883e-05,
      "loss": 0.5266,
      "step": 13319
    },
    {
      "epoch": 2.4753763241033266,
      "grad_norm": 0.9765466451644897,
      "learning_rate": 6.37244275632603e-05,
      "loss": 0.965,
      "step": 13320
    },
    {
      "epoch": 2.4755621631666975,
      "grad_norm": 0.8644874691963196,
      "learning_rate": 6.37108137355721e-05,
      "loss": 0.9189,
      "step": 13321
    },
    {
      "epoch": 2.475748002230069,
      "grad_norm": 0.8705654740333557,
      "learning_rate": 6.369720068241435e-05,
      "loss": 0.9405,
      "step": 13322
    },
    {
      "epoch": 2.4759338412934397,
      "grad_norm": 0.8883621692657471,
      "learning_rate": 6.368358840407753e-05,
      "loss": 0.8004,
      "step": 13323
    },
    {
      "epoch": 2.476119680356811,
      "grad_norm": 0.6940230131149292,
      "learning_rate": 6.366997690085221e-05,
      "loss": 0.9775,
      "step": 13324
    },
    {
      "epoch": 2.476305519420182,
      "grad_norm": 0.9831171631813049,
      "learning_rate": 6.365636617302888e-05,
      "loss": 0.8483,
      "step": 13325
    },
    {
      "epoch": 2.4764913584835533,
      "grad_norm": 0.8517470359802246,
      "learning_rate": 6.364275622089802e-05,
      "loss": 0.9926,
      "step": 13326
    },
    {
      "epoch": 2.476677197546924,
      "grad_norm": 0.962731659412384,
      "learning_rate": 6.362914704475014e-05,
      "loss": 0.923,
      "step": 13327
    },
    {
      "epoch": 2.4768630366102955,
      "grad_norm": 0.8018013834953308,
      "learning_rate": 6.361553864487573e-05,
      "loss": 0.6676,
      "step": 13328
    },
    {
      "epoch": 2.477048875673667,
      "grad_norm": 0.8119571208953857,
      "learning_rate": 6.36019310215652e-05,
      "loss": 0.8344,
      "step": 13329
    },
    {
      "epoch": 2.4772347147370377,
      "grad_norm": 0.999383807182312,
      "learning_rate": 6.3588324175109e-05,
      "loss": 0.671,
      "step": 13330
    },
    {
      "epoch": 2.4774205538004086,
      "grad_norm": 0.6964300870895386,
      "learning_rate": 6.357471810579752e-05,
      "loss": 0.7833,
      "step": 13331
    },
    {
      "epoch": 2.47760639286378,
      "grad_norm": 0.7667035460472107,
      "learning_rate": 6.356111281392119e-05,
      "loss": 0.9719,
      "step": 13332
    },
    {
      "epoch": 2.4777922319271513,
      "grad_norm": 0.9297795295715332,
      "learning_rate": 6.354750829977036e-05,
      "loss": 1.0311,
      "step": 13333
    },
    {
      "epoch": 2.477978070990522,
      "grad_norm": 0.9912102818489075,
      "learning_rate": 6.353390456363544e-05,
      "loss": 0.9136,
      "step": 13334
    },
    {
      "epoch": 2.4781639100538935,
      "grad_norm": 0.8039162158966064,
      "learning_rate": 6.352030160580671e-05,
      "loss": 0.7175,
      "step": 13335
    },
    {
      "epoch": 2.4783497491172644,
      "grad_norm": 0.7512396574020386,
      "learning_rate": 6.350669942657452e-05,
      "loss": 0.7518,
      "step": 13336
    },
    {
      "epoch": 2.4785355881806357,
      "grad_norm": 0.9079875349998474,
      "learning_rate": 6.349309802622925e-05,
      "loss": 1.1857,
      "step": 13337
    },
    {
      "epoch": 2.4787214272440066,
      "grad_norm": 0.9301310777664185,
      "learning_rate": 6.347949740506116e-05,
      "loss": 1.211,
      "step": 13338
    },
    {
      "epoch": 2.478907266307378,
      "grad_norm": 0.8027583360671997,
      "learning_rate": 6.34658975633605e-05,
      "loss": 0.8181,
      "step": 13339
    },
    {
      "epoch": 2.479093105370749,
      "grad_norm": 0.7249844670295715,
      "learning_rate": 6.345229850141758e-05,
      "loss": 1.0283,
      "step": 13340
    },
    {
      "epoch": 2.47927894443412,
      "grad_norm": 0.8149129748344421,
      "learning_rate": 6.343870021952262e-05,
      "loss": 0.9977,
      "step": 13341
    },
    {
      "epoch": 2.479464783497491,
      "grad_norm": 0.6895486116409302,
      "learning_rate": 6.342510271796588e-05,
      "loss": 0.8257,
      "step": 13342
    },
    {
      "epoch": 2.4796506225608623,
      "grad_norm": 0.8426530957221985,
      "learning_rate": 6.341150599703753e-05,
      "loss": 0.9625,
      "step": 13343
    },
    {
      "epoch": 2.4798364616242337,
      "grad_norm": 0.6276504397392273,
      "learning_rate": 6.339791005702783e-05,
      "loss": 0.4804,
      "step": 13344
    },
    {
      "epoch": 2.4800223006876045,
      "grad_norm": 0.7405399680137634,
      "learning_rate": 6.33843148982269e-05,
      "loss": 0.9845,
      "step": 13345
    },
    {
      "epoch": 2.4802081397509754,
      "grad_norm": 0.788058340549469,
      "learning_rate": 6.337072052092495e-05,
      "loss": 0.9082,
      "step": 13346
    },
    {
      "epoch": 2.4803939788143468,
      "grad_norm": 0.829477846622467,
      "learning_rate": 6.335712692541204e-05,
      "loss": 0.7689,
      "step": 13347
    },
    {
      "epoch": 2.480579817877718,
      "grad_norm": 0.8082651495933533,
      "learning_rate": 6.334353411197846e-05,
      "loss": 0.8273,
      "step": 13348
    },
    {
      "epoch": 2.480765656941089,
      "grad_norm": 0.8245391845703125,
      "learning_rate": 6.33299420809142e-05,
      "loss": 0.9454,
      "step": 13349
    },
    {
      "epoch": 2.4809514960044603,
      "grad_norm": 0.7668597102165222,
      "learning_rate": 6.331635083250943e-05,
      "loss": 0.9144,
      "step": 13350
    },
    {
      "epoch": 2.481137335067831,
      "grad_norm": 0.7859448790550232,
      "learning_rate": 6.330276036705418e-05,
      "loss": 0.8892,
      "step": 13351
    },
    {
      "epoch": 2.4813231741312025,
      "grad_norm": 0.8485981225967407,
      "learning_rate": 6.328917068483855e-05,
      "loss": 0.9379,
      "step": 13352
    },
    {
      "epoch": 2.4815090131945734,
      "grad_norm": 0.6385952234268188,
      "learning_rate": 6.327558178615256e-05,
      "loss": 0.6124,
      "step": 13353
    },
    {
      "epoch": 2.4816948522579447,
      "grad_norm": 0.851843535900116,
      "learning_rate": 6.326199367128627e-05,
      "loss": 1.0969,
      "step": 13354
    },
    {
      "epoch": 2.4818806913213156,
      "grad_norm": 0.7018309235572815,
      "learning_rate": 6.324840634052967e-05,
      "loss": 0.7291,
      "step": 13355
    },
    {
      "epoch": 2.482066530384687,
      "grad_norm": 0.9294320344924927,
      "learning_rate": 6.323481979417277e-05,
      "loss": 0.9707,
      "step": 13356
    },
    {
      "epoch": 2.482252369448058,
      "grad_norm": 0.8005899786949158,
      "learning_rate": 6.32212340325055e-05,
      "loss": 1.074,
      "step": 13357
    },
    {
      "epoch": 2.482438208511429,
      "grad_norm": 0.8252381682395935,
      "learning_rate": 6.320764905581793e-05,
      "loss": 0.8933,
      "step": 13358
    },
    {
      "epoch": 2.4826240475748,
      "grad_norm": 0.7771022319793701,
      "learning_rate": 6.319406486439993e-05,
      "loss": 0.94,
      "step": 13359
    },
    {
      "epoch": 2.4828098866381714,
      "grad_norm": 0.8648090958595276,
      "learning_rate": 6.31804814585415e-05,
      "loss": 0.9789,
      "step": 13360
    },
    {
      "epoch": 2.4829957257015423,
      "grad_norm": 0.7643998265266418,
      "learning_rate": 6.316689883853246e-05,
      "loss": 0.8146,
      "step": 13361
    },
    {
      "epoch": 2.4831815647649136,
      "grad_norm": 0.8901168704032898,
      "learning_rate": 6.315331700466278e-05,
      "loss": 1.0039,
      "step": 13362
    },
    {
      "epoch": 2.483367403828285,
      "grad_norm": 0.8452150821685791,
      "learning_rate": 6.31397359572223e-05,
      "loss": 0.909,
      "step": 13363
    },
    {
      "epoch": 2.483553242891656,
      "grad_norm": 0.7983315587043762,
      "learning_rate": 6.312615569650093e-05,
      "loss": 0.9486,
      "step": 13364
    },
    {
      "epoch": 2.483739081955027,
      "grad_norm": 0.7734251618385315,
      "learning_rate": 6.311257622278845e-05,
      "loss": 0.8991,
      "step": 13365
    },
    {
      "epoch": 2.483924921018398,
      "grad_norm": 0.8092442154884338,
      "learning_rate": 6.309899753637478e-05,
      "loss": 0.9363,
      "step": 13366
    },
    {
      "epoch": 2.4841107600817693,
      "grad_norm": 0.779928982257843,
      "learning_rate": 6.308541963754964e-05,
      "loss": 1.0153,
      "step": 13367
    },
    {
      "epoch": 2.4842965991451402,
      "grad_norm": 0.8639279007911682,
      "learning_rate": 6.307184252660285e-05,
      "loss": 0.9031,
      "step": 13368
    },
    {
      "epoch": 2.4844824382085116,
      "grad_norm": 0.7608044147491455,
      "learning_rate": 6.305826620382422e-05,
      "loss": 0.9893,
      "step": 13369
    },
    {
      "epoch": 2.4846682772718824,
      "grad_norm": 0.8655931949615479,
      "learning_rate": 6.304469066950355e-05,
      "loss": 1.0699,
      "step": 13370
    },
    {
      "epoch": 2.4848541163352538,
      "grad_norm": 0.7753689289093018,
      "learning_rate": 6.30311159239305e-05,
      "loss": 0.8803,
      "step": 13371
    },
    {
      "epoch": 2.4850399553986247,
      "grad_norm": 0.8257173895835876,
      "learning_rate": 6.301754196739487e-05,
      "loss": 0.7816,
      "step": 13372
    },
    {
      "epoch": 2.485225794461996,
      "grad_norm": 0.9522544145584106,
      "learning_rate": 6.300396880018632e-05,
      "loss": 1.0108,
      "step": 13373
    },
    {
      "epoch": 2.485411633525367,
      "grad_norm": 0.9205360412597656,
      "learning_rate": 6.299039642259458e-05,
      "loss": 0.9166,
      "step": 13374
    },
    {
      "epoch": 2.485597472588738,
      "grad_norm": 0.9840677380561829,
      "learning_rate": 6.29768248349093e-05,
      "loss": 0.9388,
      "step": 13375
    },
    {
      "epoch": 2.485783311652109,
      "grad_norm": 0.8334775567054749,
      "learning_rate": 6.29632540374202e-05,
      "loss": 0.9968,
      "step": 13376
    },
    {
      "epoch": 2.4859691507154804,
      "grad_norm": 0.6557118892669678,
      "learning_rate": 6.294968403041685e-05,
      "loss": 0.6664,
      "step": 13377
    },
    {
      "epoch": 2.4861549897788517,
      "grad_norm": 0.7033586502075195,
      "learning_rate": 6.293611481418889e-05,
      "loss": 0.7885,
      "step": 13378
    },
    {
      "epoch": 2.4863408288422226,
      "grad_norm": 0.8240175247192383,
      "learning_rate": 6.292254638902597e-05,
      "loss": 0.7301,
      "step": 13379
    },
    {
      "epoch": 2.4865266679055935,
      "grad_norm": 0.7722638249397278,
      "learning_rate": 6.290897875521769e-05,
      "loss": 0.8163,
      "step": 13380
    },
    {
      "epoch": 2.486712506968965,
      "grad_norm": 0.7567647099494934,
      "learning_rate": 6.28954119130536e-05,
      "loss": 0.915,
      "step": 13381
    },
    {
      "epoch": 2.486898346032336,
      "grad_norm": 0.9332913160324097,
      "learning_rate": 6.288184586282329e-05,
      "loss": 0.9552,
      "step": 13382
    },
    {
      "epoch": 2.487084185095707,
      "grad_norm": 0.756186306476593,
      "learning_rate": 6.286828060481626e-05,
      "loss": 0.9971,
      "step": 13383
    },
    {
      "epoch": 2.4872700241590784,
      "grad_norm": 0.8372758626937866,
      "learning_rate": 6.285471613932207e-05,
      "loss": 1.0718,
      "step": 13384
    },
    {
      "epoch": 2.4874558632224493,
      "grad_norm": 0.9867591857910156,
      "learning_rate": 6.28411524666302e-05,
      "loss": 1.1196,
      "step": 13385
    },
    {
      "epoch": 2.4876417022858206,
      "grad_norm": 0.789208173751831,
      "learning_rate": 6.282758958703018e-05,
      "loss": 0.6183,
      "step": 13386
    },
    {
      "epoch": 2.4878275413491915,
      "grad_norm": 0.9247452020645142,
      "learning_rate": 6.281402750081147e-05,
      "loss": 0.8366,
      "step": 13387
    },
    {
      "epoch": 2.488013380412563,
      "grad_norm": 0.7863283157348633,
      "learning_rate": 6.280046620826352e-05,
      "loss": 0.9512,
      "step": 13388
    },
    {
      "epoch": 2.4881992194759337,
      "grad_norm": 0.8000710010528564,
      "learning_rate": 6.278690570967576e-05,
      "loss": 0.8256,
      "step": 13389
    },
    {
      "epoch": 2.488385058539305,
      "grad_norm": 0.7943741083145142,
      "learning_rate": 6.277334600533768e-05,
      "loss": 0.9896,
      "step": 13390
    },
    {
      "epoch": 2.488570897602676,
      "grad_norm": 0.7985944151878357,
      "learning_rate": 6.275978709553862e-05,
      "loss": 1.0321,
      "step": 13391
    },
    {
      "epoch": 2.4887567366660472,
      "grad_norm": 0.8724450469017029,
      "learning_rate": 6.274622898056804e-05,
      "loss": 1.058,
      "step": 13392
    },
    {
      "epoch": 2.4889425757294186,
      "grad_norm": 0.6923696994781494,
      "learning_rate": 6.273267166071521e-05,
      "loss": 0.8085,
      "step": 13393
    },
    {
      "epoch": 2.4891284147927895,
      "grad_norm": 0.9188840389251709,
      "learning_rate": 6.27191151362696e-05,
      "loss": 1.063,
      "step": 13394
    },
    {
      "epoch": 2.4893142538561603,
      "grad_norm": 0.7417453527450562,
      "learning_rate": 6.270555940752047e-05,
      "loss": 0.834,
      "step": 13395
    },
    {
      "epoch": 2.4895000929195317,
      "grad_norm": 0.7921097278594971,
      "learning_rate": 6.269200447475721e-05,
      "loss": 1.0216,
      "step": 13396
    },
    {
      "epoch": 2.489685931982903,
      "grad_norm": 0.8512747883796692,
      "learning_rate": 6.267845033826906e-05,
      "loss": 0.9431,
      "step": 13397
    },
    {
      "epoch": 2.489871771046274,
      "grad_norm": 0.7549306154251099,
      "learning_rate": 6.266489699834535e-05,
      "loss": 0.9412,
      "step": 13398
    },
    {
      "epoch": 2.490057610109645,
      "grad_norm": 0.8000445365905762,
      "learning_rate": 6.26513444552753e-05,
      "loss": 0.7382,
      "step": 13399
    },
    {
      "epoch": 2.490243449173016,
      "grad_norm": 0.7849511504173279,
      "learning_rate": 6.263779270934828e-05,
      "loss": 0.9786,
      "step": 13400
    },
    {
      "epoch": 2.4904292882363874,
      "grad_norm": 0.7787767052650452,
      "learning_rate": 6.262424176085342e-05,
      "loss": 0.9891,
      "step": 13401
    },
    {
      "epoch": 2.4906151272997583,
      "grad_norm": 0.8131736516952515,
      "learning_rate": 6.261069161008001e-05,
      "loss": 1.0321,
      "step": 13402
    },
    {
      "epoch": 2.4908009663631296,
      "grad_norm": 0.6768377423286438,
      "learning_rate": 6.259714225731718e-05,
      "loss": 0.7298,
      "step": 13403
    },
    {
      "epoch": 2.4909868054265005,
      "grad_norm": 0.7723026871681213,
      "learning_rate": 6.258359370285422e-05,
      "loss": 0.8469,
      "step": 13404
    },
    {
      "epoch": 2.491172644489872,
      "grad_norm": 0.7771666646003723,
      "learning_rate": 6.257004594698021e-05,
      "loss": 0.9502,
      "step": 13405
    },
    {
      "epoch": 2.4913584835532427,
      "grad_norm": 0.8538346886634827,
      "learning_rate": 6.255649898998434e-05,
      "loss": 0.9947,
      "step": 13406
    },
    {
      "epoch": 2.491544322616614,
      "grad_norm": 0.7655419111251831,
      "learning_rate": 6.254295283215574e-05,
      "loss": 1.0506,
      "step": 13407
    },
    {
      "epoch": 2.491730161679985,
      "grad_norm": 0.8938736319541931,
      "learning_rate": 6.252940747378353e-05,
      "loss": 0.8542,
      "step": 13408
    },
    {
      "epoch": 2.4919160007433563,
      "grad_norm": 0.8296077847480774,
      "learning_rate": 6.251586291515683e-05,
      "loss": 0.8791,
      "step": 13409
    },
    {
      "epoch": 2.492101839806727,
      "grad_norm": 0.8586382269859314,
      "learning_rate": 6.250231915656468e-05,
      "loss": 0.8881,
      "step": 13410
    },
    {
      "epoch": 2.4922876788700985,
      "grad_norm": 0.8820834755897522,
      "learning_rate": 6.248877619829619e-05,
      "loss": 1.1859,
      "step": 13411
    },
    {
      "epoch": 2.49247351793347,
      "grad_norm": 0.7467595934867859,
      "learning_rate": 6.247523404064042e-05,
      "loss": 0.7304,
      "step": 13412
    },
    {
      "epoch": 2.4926593569968407,
      "grad_norm": 0.9176929593086243,
      "learning_rate": 6.246169268388638e-05,
      "loss": 0.8659,
      "step": 13413
    },
    {
      "epoch": 2.492845196060212,
      "grad_norm": 0.705633819103241,
      "learning_rate": 6.244815212832309e-05,
      "loss": 0.8766,
      "step": 13414
    },
    {
      "epoch": 2.493031035123583,
      "grad_norm": 0.7432262897491455,
      "learning_rate": 6.243461237423955e-05,
      "loss": 1.0674,
      "step": 13415
    },
    {
      "epoch": 2.4932168741869543,
      "grad_norm": 0.7909014225006104,
      "learning_rate": 6.242107342192475e-05,
      "loss": 1.0911,
      "step": 13416
    },
    {
      "epoch": 2.493402713250325,
      "grad_norm": 0.9045495986938477,
      "learning_rate": 6.240753527166763e-05,
      "loss": 0.9914,
      "step": 13417
    },
    {
      "epoch": 2.4935885523136965,
      "grad_norm": 0.9623762369155884,
      "learning_rate": 6.239399792375718e-05,
      "loss": 1.0743,
      "step": 13418
    },
    {
      "epoch": 2.4937743913770674,
      "grad_norm": 0.7921001315116882,
      "learning_rate": 6.23804613784823e-05,
      "loss": 0.906,
      "step": 13419
    },
    {
      "epoch": 2.4939602304404387,
      "grad_norm": 0.6968930959701538,
      "learning_rate": 6.236692563613189e-05,
      "loss": 0.6455,
      "step": 13420
    },
    {
      "epoch": 2.4941460695038096,
      "grad_norm": 1.0765269994735718,
      "learning_rate": 6.235339069699489e-05,
      "loss": 0.9525,
      "step": 13421
    },
    {
      "epoch": 2.494331908567181,
      "grad_norm": 0.8552760481834412,
      "learning_rate": 6.233985656136019e-05,
      "loss": 0.9474,
      "step": 13422
    },
    {
      "epoch": 2.494517747630552,
      "grad_norm": 0.8925338387489319,
      "learning_rate": 6.23263232295166e-05,
      "loss": 1.0033,
      "step": 13423
    },
    {
      "epoch": 2.494703586693923,
      "grad_norm": 0.8355891108512878,
      "learning_rate": 6.231279070175298e-05,
      "loss": 0.8004,
      "step": 13424
    },
    {
      "epoch": 2.494889425757294,
      "grad_norm": 0.7731959819793701,
      "learning_rate": 6.229925897835818e-05,
      "loss": 0.8918,
      "step": 13425
    },
    {
      "epoch": 2.4950752648206653,
      "grad_norm": 1.1086838245391846,
      "learning_rate": 6.228572805962102e-05,
      "loss": 0.9063,
      "step": 13426
    },
    {
      "epoch": 2.4952611038840367,
      "grad_norm": 0.8130460977554321,
      "learning_rate": 6.227219794583023e-05,
      "loss": 1.0371,
      "step": 13427
    },
    {
      "epoch": 2.4954469429474075,
      "grad_norm": 0.815039873123169,
      "learning_rate": 6.225866863727467e-05,
      "loss": 1.1554,
      "step": 13428
    },
    {
      "epoch": 2.4956327820107784,
      "grad_norm": 0.9628624320030212,
      "learning_rate": 6.224514013424305e-05,
      "loss": 0.8759,
      "step": 13429
    },
    {
      "epoch": 2.4958186210741498,
      "grad_norm": 0.7636185884475708,
      "learning_rate": 6.223161243702412e-05,
      "loss": 0.8707,
      "step": 13430
    },
    {
      "epoch": 2.496004460137521,
      "grad_norm": 0.8707903027534485,
      "learning_rate": 6.221808554590657e-05,
      "loss": 1.0159,
      "step": 13431
    },
    {
      "epoch": 2.496190299200892,
      "grad_norm": 0.7642168402671814,
      "learning_rate": 6.22045594611792e-05,
      "loss": 0.7965,
      "step": 13432
    },
    {
      "epoch": 2.4963761382642633,
      "grad_norm": 0.8629317283630371,
      "learning_rate": 6.219103418313064e-05,
      "loss": 1.0283,
      "step": 13433
    },
    {
      "epoch": 2.496561977327634,
      "grad_norm": 0.9485316872596741,
      "learning_rate": 6.217750971204958e-05,
      "loss": 1.0264,
      "step": 13434
    },
    {
      "epoch": 2.4967478163910055,
      "grad_norm": 0.7349181771278381,
      "learning_rate": 6.216398604822466e-05,
      "loss": 0.8394,
      "step": 13435
    },
    {
      "epoch": 2.4969336554543764,
      "grad_norm": 0.7712982892990112,
      "learning_rate": 6.215046319194454e-05,
      "loss": 0.8971,
      "step": 13436
    },
    {
      "epoch": 2.4971194945177477,
      "grad_norm": 0.8418555855751038,
      "learning_rate": 6.21369411434978e-05,
      "loss": 0.9512,
      "step": 13437
    },
    {
      "epoch": 2.4973053335811186,
      "grad_norm": 0.7971872091293335,
      "learning_rate": 6.212341990317309e-05,
      "loss": 0.939,
      "step": 13438
    },
    {
      "epoch": 2.49749117264449,
      "grad_norm": 0.7714715003967285,
      "learning_rate": 6.2109899471259e-05,
      "loss": 1.1059,
      "step": 13439
    },
    {
      "epoch": 2.497677011707861,
      "grad_norm": 0.8538257479667664,
      "learning_rate": 6.209637984804406e-05,
      "loss": 0.8529,
      "step": 13440
    },
    {
      "epoch": 2.497862850771232,
      "grad_norm": 0.7392212748527527,
      "learning_rate": 6.208286103381684e-05,
      "loss": 0.7408,
      "step": 13441
    },
    {
      "epoch": 2.498048689834603,
      "grad_norm": 1.0642915964126587,
      "learning_rate": 6.206934302886591e-05,
      "loss": 0.7987,
      "step": 13442
    },
    {
      "epoch": 2.4982345288979744,
      "grad_norm": 0.7957573533058167,
      "learning_rate": 6.205582583347974e-05,
      "loss": 0.9889,
      "step": 13443
    },
    {
      "epoch": 2.4984203679613453,
      "grad_norm": 0.7173166275024414,
      "learning_rate": 6.204230944794688e-05,
      "loss": 0.9598,
      "step": 13444
    },
    {
      "epoch": 2.4986062070247166,
      "grad_norm": 0.9373538494110107,
      "learning_rate": 6.202879387255576e-05,
      "loss": 1.0222,
      "step": 13445
    },
    {
      "epoch": 2.498792046088088,
      "grad_norm": 0.8266227841377258,
      "learning_rate": 6.20152791075949e-05,
      "loss": 0.9327,
      "step": 13446
    },
    {
      "epoch": 2.498977885151459,
      "grad_norm": 0.9247116446495056,
      "learning_rate": 6.20017651533527e-05,
      "loss": 0.7081,
      "step": 13447
    },
    {
      "epoch": 2.49916372421483,
      "grad_norm": 0.7575215697288513,
      "learning_rate": 6.198825201011763e-05,
      "loss": 0.824,
      "step": 13448
    },
    {
      "epoch": 2.499349563278201,
      "grad_norm": 0.9333925843238831,
      "learning_rate": 6.197473967817808e-05,
      "loss": 0.8717,
      "step": 13449
    },
    {
      "epoch": 2.4995354023415723,
      "grad_norm": 0.8797184824943542,
      "learning_rate": 6.196122815782245e-05,
      "loss": 0.9035,
      "step": 13450
    },
    {
      "epoch": 2.4997212414049432,
      "grad_norm": 0.8298222422599792,
      "learning_rate": 6.194771744933916e-05,
      "loss": 0.8492,
      "step": 13451
    },
    {
      "epoch": 2.4999070804683146,
      "grad_norm": 0.9510545134544373,
      "learning_rate": 6.193420755301649e-05,
      "loss": 0.9619,
      "step": 13452
    },
    {
      "epoch": 2.5000929195316854,
      "grad_norm": 0.7924848794937134,
      "learning_rate": 6.192069846914285e-05,
      "loss": 0.8717,
      "step": 13453
    },
    {
      "epoch": 2.5002787585950568,
      "grad_norm": 0.8562452793121338,
      "learning_rate": 6.19071901980066e-05,
      "loss": 0.9483,
      "step": 13454
    },
    {
      "epoch": 2.5004645976584277,
      "grad_norm": 0.7453281283378601,
      "learning_rate": 6.189368273989596e-05,
      "loss": 0.863,
      "step": 13455
    },
    {
      "epoch": 2.500650436721799,
      "grad_norm": 0.8021848797798157,
      "learning_rate": 6.18801760950993e-05,
      "loss": 1.1081,
      "step": 13456
    },
    {
      "epoch": 2.5008362757851703,
      "grad_norm": 0.8074384331703186,
      "learning_rate": 6.186667026390485e-05,
      "loss": 0.8636,
      "step": 13457
    },
    {
      "epoch": 2.501022114848541,
      "grad_norm": 0.8265329003334045,
      "learning_rate": 6.185316524660087e-05,
      "loss": 0.8431,
      "step": 13458
    },
    {
      "epoch": 2.501207953911912,
      "grad_norm": 0.7328630089759827,
      "learning_rate": 6.183966104347564e-05,
      "loss": 1.054,
      "step": 13459
    },
    {
      "epoch": 2.5013937929752834,
      "grad_norm": 0.9158023595809937,
      "learning_rate": 6.182615765481734e-05,
      "loss": 1.1426,
      "step": 13460
    },
    {
      "epoch": 2.5013937929752834,
      "eval_loss": 1.0169224739074707,
      "eval_runtime": 23.4685,
      "eval_samples_per_second": 46.53,
      "eval_steps_per_second": 23.265,
      "step": 13460
    },
    {
      "epoch": 2.5015796320386547,
      "grad_norm": 0.8677700757980347,
      "learning_rate": 6.181265508091422e-05,
      "loss": 1.0208,
      "step": 13461
    },
    {
      "epoch": 2.5017654711020256,
      "grad_norm": 0.9003196954727173,
      "learning_rate": 6.179915332205439e-05,
      "loss": 0.9279,
      "step": 13462
    },
    {
      "epoch": 2.5019513101653965,
      "grad_norm": 0.8088935613632202,
      "learning_rate": 6.178565237852611e-05,
      "loss": 0.8697,
      "step": 13463
    },
    {
      "epoch": 2.502137149228768,
      "grad_norm": 0.7889655828475952,
      "learning_rate": 6.177215225061753e-05,
      "loss": 1.1017,
      "step": 13464
    },
    {
      "epoch": 2.502322988292139,
      "grad_norm": 0.7348492741584778,
      "learning_rate": 6.175865293861674e-05,
      "loss": 0.9419,
      "step": 13465
    },
    {
      "epoch": 2.50250882735551,
      "grad_norm": 0.7974409461021423,
      "learning_rate": 6.174515444281188e-05,
      "loss": 0.9685,
      "step": 13466
    },
    {
      "epoch": 2.5026946664188814,
      "grad_norm": 0.7407426238059998,
      "learning_rate": 6.173165676349103e-05,
      "loss": 0.8596,
      "step": 13467
    },
    {
      "epoch": 2.5028805054822523,
      "grad_norm": 0.9617146849632263,
      "learning_rate": 6.171815990094231e-05,
      "loss": 0.986,
      "step": 13468
    },
    {
      "epoch": 2.5030663445456236,
      "grad_norm": 0.7930120229721069,
      "learning_rate": 6.170466385545378e-05,
      "loss": 1.0472,
      "step": 13469
    },
    {
      "epoch": 2.5032521836089945,
      "grad_norm": 0.8628350496292114,
      "learning_rate": 6.169116862731346e-05,
      "loss": 0.8275,
      "step": 13470
    },
    {
      "epoch": 2.503438022672366,
      "grad_norm": 0.8005587458610535,
      "learning_rate": 6.167767421680944e-05,
      "loss": 0.8874,
      "step": 13471
    },
    {
      "epoch": 2.5036238617357367,
      "grad_norm": 0.8890085816383362,
      "learning_rate": 6.166418062422967e-05,
      "loss": 0.9465,
      "step": 13472
    },
    {
      "epoch": 2.503809700799108,
      "grad_norm": 0.7766950130462646,
      "learning_rate": 6.165068784986216e-05,
      "loss": 1.0052,
      "step": 13473
    },
    {
      "epoch": 2.503995539862479,
      "grad_norm": 0.8336291909217834,
      "learning_rate": 6.163719589399495e-05,
      "loss": 1.1179,
      "step": 13474
    },
    {
      "epoch": 2.5041813789258502,
      "grad_norm": 0.8241628408432007,
      "learning_rate": 6.162370475691594e-05,
      "loss": 1.2073,
      "step": 13475
    },
    {
      "epoch": 2.5043672179892216,
      "grad_norm": 1.0504337549209595,
      "learning_rate": 6.161021443891313e-05,
      "loss": 0.991,
      "step": 13476
    },
    {
      "epoch": 2.5045530570525925,
      "grad_norm": 0.8310957551002502,
      "learning_rate": 6.159672494027438e-05,
      "loss": 0.9675,
      "step": 13477
    },
    {
      "epoch": 2.5047388961159633,
      "grad_norm": 1.0969948768615723,
      "learning_rate": 6.158323626128762e-05,
      "loss": 1.1299,
      "step": 13478
    },
    {
      "epoch": 2.5049247351793347,
      "grad_norm": 0.9421222805976868,
      "learning_rate": 6.15697484022408e-05,
      "loss": 0.8132,
      "step": 13479
    },
    {
      "epoch": 2.505110574242706,
      "grad_norm": 0.7072616815567017,
      "learning_rate": 6.155626136342171e-05,
      "loss": 0.8952,
      "step": 13480
    },
    {
      "epoch": 2.505296413306077,
      "grad_norm": 0.7259114384651184,
      "learning_rate": 6.154277514511829e-05,
      "loss": 0.8849,
      "step": 13481
    },
    {
      "epoch": 2.505482252369448,
      "grad_norm": 0.8621964454650879,
      "learning_rate": 6.152928974761829e-05,
      "loss": 0.8402,
      "step": 13482
    },
    {
      "epoch": 2.505668091432819,
      "grad_norm": 1.2493312358856201,
      "learning_rate": 6.151580517120958e-05,
      "loss": 1.0745,
      "step": 13483
    },
    {
      "epoch": 2.5058539304961904,
      "grad_norm": 0.8947202563285828,
      "learning_rate": 6.150232141618e-05,
      "loss": 0.948,
      "step": 13484
    },
    {
      "epoch": 2.5060397695595613,
      "grad_norm": 0.9812045693397522,
      "learning_rate": 6.14888384828173e-05,
      "loss": 0.9176,
      "step": 13485
    },
    {
      "epoch": 2.5062256086229326,
      "grad_norm": 0.8159515857696533,
      "learning_rate": 6.147535637140926e-05,
      "loss": 1.0679,
      "step": 13486
    },
    {
      "epoch": 2.5064114476863035,
      "grad_norm": 0.8644919991493225,
      "learning_rate": 6.146187508224362e-05,
      "loss": 0.9617,
      "step": 13487
    },
    {
      "epoch": 2.506597286749675,
      "grad_norm": 0.7351353764533997,
      "learning_rate": 6.144839461560812e-05,
      "loss": 0.9063,
      "step": 13488
    },
    {
      "epoch": 2.5067831258130457,
      "grad_norm": 0.8311001658439636,
      "learning_rate": 6.14349149717905e-05,
      "loss": 0.8582,
      "step": 13489
    },
    {
      "epoch": 2.506968964876417,
      "grad_norm": 0.7653113007545471,
      "learning_rate": 6.142143615107841e-05,
      "loss": 0.8755,
      "step": 13490
    },
    {
      "epoch": 2.5071548039397884,
      "grad_norm": 0.8149527907371521,
      "learning_rate": 6.140795815375958e-05,
      "loss": 0.9695,
      "step": 13491
    },
    {
      "epoch": 2.5073406430031593,
      "grad_norm": 0.8174289464950562,
      "learning_rate": 6.139448098012163e-05,
      "loss": 0.9273,
      "step": 13492
    },
    {
      "epoch": 2.50752648206653,
      "grad_norm": 1.8553173542022705,
      "learning_rate": 6.138100463045229e-05,
      "loss": 1.341,
      "step": 13493
    },
    {
      "epoch": 2.5077123211299015,
      "grad_norm": 0.729652464389801,
      "learning_rate": 6.136752910503906e-05,
      "loss": 0.9325,
      "step": 13494
    },
    {
      "epoch": 2.507898160193273,
      "grad_norm": 0.8200655579566956,
      "learning_rate": 6.135405440416966e-05,
      "loss": 0.9185,
      "step": 13495
    },
    {
      "epoch": 2.5080839992566437,
      "grad_norm": 0.8505041599273682,
      "learning_rate": 6.134058052813166e-05,
      "loss": 0.9047,
      "step": 13496
    },
    {
      "epoch": 2.5082698383200146,
      "grad_norm": 0.9282901287078857,
      "learning_rate": 6.132710747721261e-05,
      "loss": 0.8199,
      "step": 13497
    },
    {
      "epoch": 2.508455677383386,
      "grad_norm": 0.8327956199645996,
      "learning_rate": 6.131363525170008e-05,
      "loss": 0.9152,
      "step": 13498
    },
    {
      "epoch": 2.5086415164467573,
      "grad_norm": 0.8083164095878601,
      "learning_rate": 6.130016385188166e-05,
      "loss": 0.9533,
      "step": 13499
    },
    {
      "epoch": 2.508827355510128,
      "grad_norm": 0.8243703246116638,
      "learning_rate": 6.128669327804479e-05,
      "loss": 1.1127,
      "step": 13500
    },
    {
      "epoch": 2.5090131945734995,
      "grad_norm": 0.97088623046875,
      "learning_rate": 6.127322353047704e-05,
      "loss": 0.8968,
      "step": 13501
    },
    {
      "epoch": 2.5091990336368704,
      "grad_norm": 0.7824402451515198,
      "learning_rate": 6.125975460946587e-05,
      "loss": 0.8047,
      "step": 13502
    },
    {
      "epoch": 2.5093848727002417,
      "grad_norm": 0.9820107817649841,
      "learning_rate": 6.124628651529875e-05,
      "loss": 0.8793,
      "step": 13503
    },
    {
      "epoch": 2.5095707117636126,
      "grad_norm": 0.8072025179862976,
      "learning_rate": 6.123281924826313e-05,
      "loss": 0.73,
      "step": 13504
    },
    {
      "epoch": 2.509756550826984,
      "grad_norm": 1.0725669860839844,
      "learning_rate": 6.121935280864646e-05,
      "loss": 0.9829,
      "step": 13505
    },
    {
      "epoch": 2.5099423898903552,
      "grad_norm": 1.0136836767196655,
      "learning_rate": 6.120588719673617e-05,
      "loss": 1.1713,
      "step": 13506
    },
    {
      "epoch": 2.510128228953726,
      "grad_norm": 0.8955618739128113,
      "learning_rate": 6.119242241281964e-05,
      "loss": 0.8974,
      "step": 13507
    },
    {
      "epoch": 2.510314068017097,
      "grad_norm": 0.816618800163269,
      "learning_rate": 6.117895845718424e-05,
      "loss": 0.934,
      "step": 13508
    },
    {
      "epoch": 2.5104999070804683,
      "grad_norm": 0.8638303279876709,
      "learning_rate": 6.116549533011738e-05,
      "loss": 0.944,
      "step": 13509
    },
    {
      "epoch": 2.5106857461438397,
      "grad_norm": 0.7912139892578125,
      "learning_rate": 6.115203303190637e-05,
      "loss": 1.0496,
      "step": 13510
    },
    {
      "epoch": 2.5108715852072105,
      "grad_norm": 0.8715540766716003,
      "learning_rate": 6.113857156283856e-05,
      "loss": 0.937,
      "step": 13511
    },
    {
      "epoch": 2.5110574242705814,
      "grad_norm": 0.8288502097129822,
      "learning_rate": 6.112511092320122e-05,
      "loss": 0.8559,
      "step": 13512
    },
    {
      "epoch": 2.5112432633339528,
      "grad_norm": 0.8309054970741272,
      "learning_rate": 6.11116511132817e-05,
      "loss": 0.9803,
      "step": 13513
    },
    {
      "epoch": 2.511429102397324,
      "grad_norm": 0.755159854888916,
      "learning_rate": 6.109819213336724e-05,
      "loss": 0.8886,
      "step": 13514
    },
    {
      "epoch": 2.511614941460695,
      "grad_norm": 0.884611964225769,
      "learning_rate": 6.10847339837451e-05,
      "loss": 1.0456,
      "step": 13515
    },
    {
      "epoch": 2.5118007805240663,
      "grad_norm": 0.8449623584747314,
      "learning_rate": 6.107127666470255e-05,
      "loss": 0.7734,
      "step": 13516
    },
    {
      "epoch": 2.511986619587437,
      "grad_norm": 0.9435641169548035,
      "learning_rate": 6.105782017652679e-05,
      "loss": 1.0571,
      "step": 13517
    },
    {
      "epoch": 2.5121724586508085,
      "grad_norm": 1.208054780960083,
      "learning_rate": 6.104436451950501e-05,
      "loss": 0.8751,
      "step": 13518
    },
    {
      "epoch": 2.5123582977141794,
      "grad_norm": 0.6422603726387024,
      "learning_rate": 6.103090969392447e-05,
      "loss": 0.5961,
      "step": 13519
    },
    {
      "epoch": 2.5125441367775507,
      "grad_norm": 0.770844042301178,
      "learning_rate": 6.101745570007224e-05,
      "loss": 1.1031,
      "step": 13520
    },
    {
      "epoch": 2.5127299758409216,
      "grad_norm": 0.7535857558250427,
      "learning_rate": 6.100400253823556e-05,
      "loss": 0.8568,
      "step": 13521
    },
    {
      "epoch": 2.512915814904293,
      "grad_norm": 0.7936834692955017,
      "learning_rate": 6.09905502087015e-05,
      "loss": 0.9994,
      "step": 13522
    },
    {
      "epoch": 2.513101653967664,
      "grad_norm": 0.9213391542434692,
      "learning_rate": 6.097709871175723e-05,
      "loss": 0.8465,
      "step": 13523
    },
    {
      "epoch": 2.513287493031035,
      "grad_norm": 0.8148267269134521,
      "learning_rate": 6.096364804768979e-05,
      "loss": 0.7718,
      "step": 13524
    },
    {
      "epoch": 2.5134733320944065,
      "grad_norm": 1.343180775642395,
      "learning_rate": 6.095019821678627e-05,
      "loss": 0.9792,
      "step": 13525
    },
    {
      "epoch": 2.5136591711577774,
      "grad_norm": 0.8396459221839905,
      "learning_rate": 6.093674921933381e-05,
      "loss": 1.22,
      "step": 13526
    },
    {
      "epoch": 2.5138450102211483,
      "grad_norm": 0.724976658821106,
      "learning_rate": 6.0923301055619384e-05,
      "loss": 0.9506,
      "step": 13527
    },
    {
      "epoch": 2.5140308492845196,
      "grad_norm": 0.9246887564659119,
      "learning_rate": 6.0909853725930034e-05,
      "loss": 1.0247,
      "step": 13528
    },
    {
      "epoch": 2.514216688347891,
      "grad_norm": 0.8594968318939209,
      "learning_rate": 6.089640723055279e-05,
      "loss": 0.9865,
      "step": 13529
    },
    {
      "epoch": 2.514402527411262,
      "grad_norm": 0.734356701374054,
      "learning_rate": 6.088296156977462e-05,
      "loss": 1.0941,
      "step": 13530
    },
    {
      "epoch": 2.514588366474633,
      "grad_norm": 0.7757397890090942,
      "learning_rate": 6.0869516743882516e-05,
      "loss": 1.0349,
      "step": 13531
    },
    {
      "epoch": 2.514774205538004,
      "grad_norm": 0.6831685304641724,
      "learning_rate": 6.0856072753163406e-05,
      "loss": 0.635,
      "step": 13532
    },
    {
      "epoch": 2.5149600446013753,
      "grad_norm": 1.8188354969024658,
      "learning_rate": 6.084262959790428e-05,
      "loss": 1.5332,
      "step": 13533
    },
    {
      "epoch": 2.5151458836647462,
      "grad_norm": 0.8049079775810242,
      "learning_rate": 6.0829187278391996e-05,
      "loss": 0.9258,
      "step": 13534
    },
    {
      "epoch": 2.5153317227281176,
      "grad_norm": 0.8711287379264832,
      "learning_rate": 6.0815745794913524e-05,
      "loss": 1.0228,
      "step": 13535
    },
    {
      "epoch": 2.5155175617914884,
      "grad_norm": 0.8077500462532043,
      "learning_rate": 6.080230514775567e-05,
      "loss": 0.9122,
      "step": 13536
    },
    {
      "epoch": 2.5157034008548598,
      "grad_norm": 1.6004647016525269,
      "learning_rate": 6.0788865337205355e-05,
      "loss": 1.438,
      "step": 13537
    },
    {
      "epoch": 2.5158892399182307,
      "grad_norm": 1.7043529748916626,
      "learning_rate": 6.077542636354944e-05,
      "loss": 1.2784,
      "step": 13538
    },
    {
      "epoch": 2.516075078981602,
      "grad_norm": 0.8830113410949707,
      "learning_rate": 6.076198822707475e-05,
      "loss": 0.8335,
      "step": 13539
    },
    {
      "epoch": 2.5162609180449733,
      "grad_norm": 0.7613457441329956,
      "learning_rate": 6.074855092806806e-05,
      "loss": 1.0871,
      "step": 13540
    },
    {
      "epoch": 2.516446757108344,
      "grad_norm": 0.8486202359199524,
      "learning_rate": 6.0735114466816214e-05,
      "loss": 0.8161,
      "step": 13541
    },
    {
      "epoch": 2.516632596171715,
      "grad_norm": 0.7651652097702026,
      "learning_rate": 6.072167884360596e-05,
      "loss": 0.8114,
      "step": 13542
    },
    {
      "epoch": 2.5168184352350864,
      "grad_norm": 0.6826961636543274,
      "learning_rate": 6.070824405872406e-05,
      "loss": 0.7817,
      "step": 13543
    },
    {
      "epoch": 2.5170042742984577,
      "grad_norm": 0.8031132817268372,
      "learning_rate": 6.0694810112457265e-05,
      "loss": 0.9292,
      "step": 13544
    },
    {
      "epoch": 2.5171901133618286,
      "grad_norm": 1.1389261484146118,
      "learning_rate": 6.068137700509231e-05,
      "loss": 1.064,
      "step": 13545
    },
    {
      "epoch": 2.5173759524251995,
      "grad_norm": 0.7853626012802124,
      "learning_rate": 6.066794473691586e-05,
      "loss": 0.9105,
      "step": 13546
    },
    {
      "epoch": 2.517561791488571,
      "grad_norm": 0.8402011394500732,
      "learning_rate": 6.0654513308214645e-05,
      "loss": 0.964,
      "step": 13547
    },
    {
      "epoch": 2.517747630551942,
      "grad_norm": 0.9154853820800781,
      "learning_rate": 6.064108271927532e-05,
      "loss": 1.1316,
      "step": 13548
    },
    {
      "epoch": 2.517933469615313,
      "grad_norm": 0.792783260345459,
      "learning_rate": 6.062765297038457e-05,
      "loss": 0.7286,
      "step": 13549
    },
    {
      "epoch": 2.5181193086786844,
      "grad_norm": 0.9570148587226868,
      "learning_rate": 6.061422406182897e-05,
      "loss": 0.9536,
      "step": 13550
    },
    {
      "epoch": 2.5183051477420553,
      "grad_norm": 0.837431788444519,
      "learning_rate": 6.060079599389521e-05,
      "loss": 0.8922,
      "step": 13551
    },
    {
      "epoch": 2.5184909868054266,
      "grad_norm": 0.7448437213897705,
      "learning_rate": 6.058736876686981e-05,
      "loss": 0.8407,
      "step": 13552
    },
    {
      "epoch": 2.5186768258687975,
      "grad_norm": 0.7879940271377563,
      "learning_rate": 6.057394238103942e-05,
      "loss": 1.0672,
      "step": 13553
    },
    {
      "epoch": 2.518862664932169,
      "grad_norm": 0.7715732455253601,
      "learning_rate": 6.056051683669053e-05,
      "loss": 0.7682,
      "step": 13554
    },
    {
      "epoch": 2.5190485039955397,
      "grad_norm": 0.7592453360557556,
      "learning_rate": 6.0547092134109764e-05,
      "loss": 0.6859,
      "step": 13555
    },
    {
      "epoch": 2.519234343058911,
      "grad_norm": 0.8092079162597656,
      "learning_rate": 6.05336682735836e-05,
      "loss": 0.7248,
      "step": 13556
    },
    {
      "epoch": 2.519420182122282,
      "grad_norm": 0.8283814191818237,
      "learning_rate": 6.052024525539852e-05,
      "loss": 0.8203,
      "step": 13557
    },
    {
      "epoch": 2.5196060211856532,
      "grad_norm": 0.7173612713813782,
      "learning_rate": 6.050682307984109e-05,
      "loss": 0.9486,
      "step": 13558
    },
    {
      "epoch": 2.5197918602490246,
      "grad_norm": 0.9390692114830017,
      "learning_rate": 6.049340174719776e-05,
      "loss": 0.9158,
      "step": 13559
    },
    {
      "epoch": 2.5199776993123955,
      "grad_norm": 0.7687232494354248,
      "learning_rate": 6.047998125775496e-05,
      "loss": 0.8785,
      "step": 13560
    },
    {
      "epoch": 2.5201635383757663,
      "grad_norm": 0.8472660183906555,
      "learning_rate": 6.046656161179915e-05,
      "loss": 1.0567,
      "step": 13561
    },
    {
      "epoch": 2.5203493774391377,
      "grad_norm": 0.7931621074676514,
      "learning_rate": 6.045314280961674e-05,
      "loss": 0.9247,
      "step": 13562
    },
    {
      "epoch": 2.520535216502509,
      "grad_norm": 1.0868077278137207,
      "learning_rate": 6.043972485149414e-05,
      "loss": 1.158,
      "step": 13563
    },
    {
      "epoch": 2.52072105556588,
      "grad_norm": 0.8607794642448425,
      "learning_rate": 6.0426307737717715e-05,
      "loss": 0.9757,
      "step": 13564
    },
    {
      "epoch": 2.520906894629251,
      "grad_norm": 0.7563510537147522,
      "learning_rate": 6.041289146857384e-05,
      "loss": 0.8482,
      "step": 13565
    },
    {
      "epoch": 2.521092733692622,
      "grad_norm": 0.8587755560874939,
      "learning_rate": 6.039947604434886e-05,
      "loss": 0.9595,
      "step": 13566
    },
    {
      "epoch": 2.5212785727559934,
      "grad_norm": 0.7921454310417175,
      "learning_rate": 6.038606146532908e-05,
      "loss": 1.0287,
      "step": 13567
    },
    {
      "epoch": 2.5214644118193643,
      "grad_norm": 0.8255019187927246,
      "learning_rate": 6.0372647731800844e-05,
      "loss": 1.0019,
      "step": 13568
    },
    {
      "epoch": 2.5216502508827356,
      "grad_norm": 0.8133623003959656,
      "learning_rate": 6.0359234844050486e-05,
      "loss": 1.0032,
      "step": 13569
    },
    {
      "epoch": 2.5218360899461065,
      "grad_norm": 0.8292799592018127,
      "learning_rate": 6.034582280236419e-05,
      "loss": 1.0846,
      "step": 13570
    },
    {
      "epoch": 2.522021929009478,
      "grad_norm": 0.8180259466171265,
      "learning_rate": 6.033241160702829e-05,
      "loss": 1.0461,
      "step": 13571
    },
    {
      "epoch": 2.5222077680728487,
      "grad_norm": 0.6881454586982727,
      "learning_rate": 6.031900125832897e-05,
      "loss": 0.9085,
      "step": 13572
    },
    {
      "epoch": 2.52239360713622,
      "grad_norm": 0.833112359046936,
      "learning_rate": 6.030559175655249e-05,
      "loss": 1.047,
      "step": 13573
    },
    {
      "epoch": 2.5225794461995914,
      "grad_norm": 0.8363716006278992,
      "learning_rate": 6.029218310198502e-05,
      "loss": 0.9768,
      "step": 13574
    },
    {
      "epoch": 2.5227652852629623,
      "grad_norm": 0.9038662910461426,
      "learning_rate": 6.0278775294912784e-05,
      "loss": 0.9416,
      "step": 13575
    },
    {
      "epoch": 2.522951124326333,
      "grad_norm": 0.743466317653656,
      "learning_rate": 6.02653683356219e-05,
      "loss": 0.7738,
      "step": 13576
    },
    {
      "epoch": 2.5231369633897045,
      "grad_norm": 0.9426493644714355,
      "learning_rate": 6.025196222439855e-05,
      "loss": 1.1298,
      "step": 13577
    },
    {
      "epoch": 2.523322802453076,
      "grad_norm": 0.8024913668632507,
      "learning_rate": 6.023855696152881e-05,
      "loss": 0.9108,
      "step": 13578
    },
    {
      "epoch": 2.5235086415164467,
      "grad_norm": 0.9854022860527039,
      "learning_rate": 6.02251525472989e-05,
      "loss": 1.063,
      "step": 13579
    },
    {
      "epoch": 2.5236944805798176,
      "grad_norm": 0.942895233631134,
      "learning_rate": 6.021174898199482e-05,
      "loss": 1.2723,
      "step": 13580
    },
    {
      "epoch": 2.523880319643189,
      "grad_norm": 0.7434505224227905,
      "learning_rate": 6.0198346265902706e-05,
      "loss": 0.9639,
      "step": 13581
    },
    {
      "epoch": 2.5240661587065603,
      "grad_norm": 0.7559301853179932,
      "learning_rate": 6.018494439930857e-05,
      "loss": 0.9286,
      "step": 13582
    },
    {
      "epoch": 2.524251997769931,
      "grad_norm": 1.4655334949493408,
      "learning_rate": 6.0171543382498485e-05,
      "loss": 1.1281,
      "step": 13583
    },
    {
      "epoch": 2.5244378368333025,
      "grad_norm": 0.7436730265617371,
      "learning_rate": 6.0158143215758455e-05,
      "loss": 0.9599,
      "step": 13584
    },
    {
      "epoch": 2.5246236758966734,
      "grad_norm": 0.8883718252182007,
      "learning_rate": 6.014474389937449e-05,
      "loss": 0.9478,
      "step": 13585
    },
    {
      "epoch": 2.5248095149600447,
      "grad_norm": 0.7029927372932434,
      "learning_rate": 6.013134543363255e-05,
      "loss": 0.6995,
      "step": 13586
    },
    {
      "epoch": 2.5249953540234156,
      "grad_norm": 0.8948057889938354,
      "learning_rate": 6.011794781881866e-05,
      "loss": 0.9246,
      "step": 13587
    },
    {
      "epoch": 2.525181193086787,
      "grad_norm": 0.8800473809242249,
      "learning_rate": 6.010455105521866e-05,
      "loss": 0.9495,
      "step": 13588
    },
    {
      "epoch": 2.5253670321501582,
      "grad_norm": 0.6592245101928711,
      "learning_rate": 6.009115514311862e-05,
      "loss": 0.7117,
      "step": 13589
    },
    {
      "epoch": 2.525552871213529,
      "grad_norm": 0.9519256353378296,
      "learning_rate": 6.0077760082804355e-05,
      "loss": 0.8683,
      "step": 13590
    },
    {
      "epoch": 2.5257387102769,
      "grad_norm": 0.9765269160270691,
      "learning_rate": 6.0064365874561835e-05,
      "loss": 0.8311,
      "step": 13591
    },
    {
      "epoch": 2.5259245493402713,
      "grad_norm": 0.7550574541091919,
      "learning_rate": 6.005097251867685e-05,
      "loss": 0.9183,
      "step": 13592
    },
    {
      "epoch": 2.5261103884036427,
      "grad_norm": 0.8850460648536682,
      "learning_rate": 6.0037580015435334e-05,
      "loss": 0.9423,
      "step": 13593
    },
    {
      "epoch": 2.5262962274670135,
      "grad_norm": 0.9110420942306519,
      "learning_rate": 6.002418836512307e-05,
      "loss": 0.9126,
      "step": 13594
    },
    {
      "epoch": 2.5264820665303844,
      "grad_norm": 0.9628833532333374,
      "learning_rate": 6.001079756802592e-05,
      "loss": 1.0606,
      "step": 13595
    },
    {
      "epoch": 2.5266679055937558,
      "grad_norm": 1.0309127569198608,
      "learning_rate": 5.999740762442967e-05,
      "loss": 0.8266,
      "step": 13596
    },
    {
      "epoch": 2.526853744657127,
      "grad_norm": 0.7594256401062012,
      "learning_rate": 5.99840185346201e-05,
      "loss": 0.9296,
      "step": 13597
    },
    {
      "epoch": 2.527039583720498,
      "grad_norm": 0.7407244443893433,
      "learning_rate": 5.997063029888299e-05,
      "loss": 0.8898,
      "step": 13598
    },
    {
      "epoch": 2.5272254227838693,
      "grad_norm": 0.7121025919914246,
      "learning_rate": 5.995724291750404e-05,
      "loss": 0.848,
      "step": 13599
    },
    {
      "epoch": 2.52741126184724,
      "grad_norm": 0.8437877893447876,
      "learning_rate": 5.994385639076907e-05,
      "loss": 1.2426,
      "step": 13600
    },
    {
      "epoch": 2.5275971009106115,
      "grad_norm": 0.8755210041999817,
      "learning_rate": 5.993047071896374e-05,
      "loss": 0.9305,
      "step": 13601
    },
    {
      "epoch": 2.5277829399739824,
      "grad_norm": 0.9839112162590027,
      "learning_rate": 5.991708590237375e-05,
      "loss": 1.1434,
      "step": 13602
    },
    {
      "epoch": 2.5279687790373537,
      "grad_norm": 0.8365402817726135,
      "learning_rate": 5.990370194128479e-05,
      "loss": 1.0356,
      "step": 13603
    },
    {
      "epoch": 2.5281546181007246,
      "grad_norm": 0.8370087146759033,
      "learning_rate": 5.989031883598248e-05,
      "loss": 0.809,
      "step": 13604
    },
    {
      "epoch": 2.528340457164096,
      "grad_norm": 0.7712308168411255,
      "learning_rate": 5.987693658675251e-05,
      "loss": 0.8797,
      "step": 13605
    },
    {
      "epoch": 2.528526296227467,
      "grad_norm": 0.7896856665611267,
      "learning_rate": 5.986355519388045e-05,
      "loss": 0.8417,
      "step": 13606
    },
    {
      "epoch": 2.528712135290838,
      "grad_norm": 1.0379797220230103,
      "learning_rate": 5.9850174657651946e-05,
      "loss": 1.0399,
      "step": 13607
    },
    {
      "epoch": 2.5288979743542095,
      "grad_norm": 1.0011177062988281,
      "learning_rate": 5.983679497835255e-05,
      "loss": 1.0312,
      "step": 13608
    },
    {
      "epoch": 2.5290838134175804,
      "grad_norm": 0.777677059173584,
      "learning_rate": 5.982341615626781e-05,
      "loss": 0.6413,
      "step": 13609
    },
    {
      "epoch": 2.5292696524809513,
      "grad_norm": 0.8527577519416809,
      "learning_rate": 5.9810038191683335e-05,
      "loss": 0.8086,
      "step": 13610
    },
    {
      "epoch": 2.5294554915443226,
      "grad_norm": 0.8974331617355347,
      "learning_rate": 5.9796661084884644e-05,
      "loss": 0.8363,
      "step": 13611
    },
    {
      "epoch": 2.529641330607694,
      "grad_norm": 0.9144704341888428,
      "learning_rate": 5.978328483615722e-05,
      "loss": 0.9432,
      "step": 13612
    },
    {
      "epoch": 2.529827169671065,
      "grad_norm": 0.9022623896598816,
      "learning_rate": 5.976990944578658e-05,
      "loss": 0.7475,
      "step": 13613
    },
    {
      "epoch": 2.530013008734436,
      "grad_norm": 0.991865336894989,
      "learning_rate": 5.975653491405816e-05,
      "loss": 0.9885,
      "step": 13614
    },
    {
      "epoch": 2.530198847797807,
      "grad_norm": 0.7925933003425598,
      "learning_rate": 5.9743161241257475e-05,
      "loss": 1.0289,
      "step": 13615
    },
    {
      "epoch": 2.5303846868611783,
      "grad_norm": 0.9255247712135315,
      "learning_rate": 5.9729788427669895e-05,
      "loss": 0.9356,
      "step": 13616
    },
    {
      "epoch": 2.5305705259245492,
      "grad_norm": 0.8493859767913818,
      "learning_rate": 5.9716416473580905e-05,
      "loss": 0.928,
      "step": 13617
    },
    {
      "epoch": 2.5307563649879206,
      "grad_norm": 0.8425911068916321,
      "learning_rate": 5.9703045379275845e-05,
      "loss": 1.0973,
      "step": 13618
    },
    {
      "epoch": 2.5309422040512914,
      "grad_norm": 0.827329158782959,
      "learning_rate": 5.968967514504014e-05,
      "loss": 0.9527,
      "step": 13619
    },
    {
      "epoch": 2.5311280431146628,
      "grad_norm": 0.7627117037773132,
      "learning_rate": 5.9676305771159114e-05,
      "loss": 0.9594,
      "step": 13620
    },
    {
      "epoch": 2.5313138821780337,
      "grad_norm": 1.0649540424346924,
      "learning_rate": 5.966293725791818e-05,
      "loss": 0.7797,
      "step": 13621
    },
    {
      "epoch": 2.531499721241405,
      "grad_norm": 2.9969398975372314,
      "learning_rate": 5.964956960560263e-05,
      "loss": 1.3579,
      "step": 13622
    },
    {
      "epoch": 2.5316855603047763,
      "grad_norm": 0.7865945100784302,
      "learning_rate": 5.963620281449778e-05,
      "loss": 0.7402,
      "step": 13623
    },
    {
      "epoch": 2.531871399368147,
      "grad_norm": 0.7578870058059692,
      "learning_rate": 5.96228368848889e-05,
      "loss": 0.8787,
      "step": 13624
    },
    {
      "epoch": 2.532057238431518,
      "grad_norm": 0.7157632112503052,
      "learning_rate": 5.960947181706128e-05,
      "loss": 0.9353,
      "step": 13625
    },
    {
      "epoch": 2.5322430774948894,
      "grad_norm": 0.9781704545021057,
      "learning_rate": 5.9596107611300164e-05,
      "loss": 1.0551,
      "step": 13626
    },
    {
      "epoch": 2.5324289165582607,
      "grad_norm": 0.8759918212890625,
      "learning_rate": 5.9582744267890814e-05,
      "loss": 1.019,
      "step": 13627
    },
    {
      "epoch": 2.5326147556216316,
      "grad_norm": 0.8140712976455688,
      "learning_rate": 5.9569381787118404e-05,
      "loss": 1.1218,
      "step": 13628
    },
    {
      "epoch": 2.5328005946850025,
      "grad_norm": 0.7885955572128296,
      "learning_rate": 5.9556020169268156e-05,
      "loss": 0.8807,
      "step": 13629
    },
    {
      "epoch": 2.532986433748374,
      "grad_norm": 1.094180941581726,
      "learning_rate": 5.9542659414625246e-05,
      "loss": 1.0929,
      "step": 13630
    },
    {
      "epoch": 2.533172272811745,
      "grad_norm": 0.8771237730979919,
      "learning_rate": 5.952929952347488e-05,
      "loss": 1.0978,
      "step": 13631
    },
    {
      "epoch": 2.533358111875116,
      "grad_norm": 0.7179327011108398,
      "learning_rate": 5.951594049610214e-05,
      "loss": 0.8338,
      "step": 13632
    },
    {
      "epoch": 2.5335439509384874,
      "grad_norm": 0.7645947933197021,
      "learning_rate": 5.9502582332792204e-05,
      "loss": 0.6564,
      "step": 13633
    },
    {
      "epoch": 2.5337297900018583,
      "grad_norm": 0.8140125274658203,
      "learning_rate": 5.948922503383012e-05,
      "loss": 1.1048,
      "step": 13634
    },
    {
      "epoch": 2.5339156290652296,
      "grad_norm": 0.8520426750183105,
      "learning_rate": 5.947586859950103e-05,
      "loss": 0.8477,
      "step": 13635
    },
    {
      "epoch": 2.5341014681286005,
      "grad_norm": 0.8290002346038818,
      "learning_rate": 5.946251303008996e-05,
      "loss": 0.8047,
      "step": 13636
    },
    {
      "epoch": 2.534287307191972,
      "grad_norm": 0.709611177444458,
      "learning_rate": 5.944915832588201e-05,
      "loss": 0.8665,
      "step": 13637
    },
    {
      "epoch": 2.534473146255343,
      "grad_norm": 0.8430138230323792,
      "learning_rate": 5.943580448716217e-05,
      "loss": 1.024,
      "step": 13638
    },
    {
      "epoch": 2.534658985318714,
      "grad_norm": 0.981663167476654,
      "learning_rate": 5.942245151421548e-05,
      "loss": 1.0096,
      "step": 13639
    },
    {
      "epoch": 2.534844824382085,
      "grad_norm": 0.778021514415741,
      "learning_rate": 5.940909940732693e-05,
      "loss": 0.7695,
      "step": 13640
    },
    {
      "epoch": 2.5350306634454562,
      "grad_norm": 0.9181351065635681,
      "learning_rate": 5.939574816678146e-05,
      "loss": 0.9499,
      "step": 13641
    },
    {
      "epoch": 2.5352165025088276,
      "grad_norm": 0.7551867365837097,
      "learning_rate": 5.93823977928641e-05,
      "loss": 1.0542,
      "step": 13642
    },
    {
      "epoch": 2.5354023415721985,
      "grad_norm": 0.8496702909469604,
      "learning_rate": 5.936904828585976e-05,
      "loss": 0.9137,
      "step": 13643
    },
    {
      "epoch": 2.5355881806355693,
      "grad_norm": 0.8048803806304932,
      "learning_rate": 5.9355699646053356e-05,
      "loss": 0.8004,
      "step": 13644
    },
    {
      "epoch": 2.5357740196989407,
      "grad_norm": 0.9019173383712769,
      "learning_rate": 5.934235187372981e-05,
      "loss": 1.0288,
      "step": 13645
    },
    {
      "epoch": 2.535959858762312,
      "grad_norm": 0.8891580700874329,
      "learning_rate": 5.9329004969173965e-05,
      "loss": 1.027,
      "step": 13646
    },
    {
      "epoch": 2.536145697825683,
      "grad_norm": 0.8195191621780396,
      "learning_rate": 5.9315658932670726e-05,
      "loss": 0.9147,
      "step": 13647
    },
    {
      "epoch": 2.536331536889054,
      "grad_norm": 0.7364562153816223,
      "learning_rate": 5.9302313764504927e-05,
      "loss": 0.913,
      "step": 13648
    },
    {
      "epoch": 2.536517375952425,
      "grad_norm": 0.723248302936554,
      "learning_rate": 5.9288969464961385e-05,
      "loss": 0.8202,
      "step": 13649
    },
    {
      "epoch": 2.5367032150157964,
      "grad_norm": 0.9361730813980103,
      "learning_rate": 5.927562603432497e-05,
      "loss": 0.8501,
      "step": 13650
    },
    {
      "epoch": 2.5368890540791673,
      "grad_norm": 0.8435133695602417,
      "learning_rate": 5.9262283472880365e-05,
      "loss": 0.8544,
      "step": 13651
    },
    {
      "epoch": 2.5370748931425386,
      "grad_norm": 0.7979792952537537,
      "learning_rate": 5.924894178091245e-05,
      "loss": 0.9817,
      "step": 13652
    },
    {
      "epoch": 2.5372607322059095,
      "grad_norm": 0.9304459095001221,
      "learning_rate": 5.923560095870595e-05,
      "loss": 0.8531,
      "step": 13653
    },
    {
      "epoch": 2.537446571269281,
      "grad_norm": 0.7247792482376099,
      "learning_rate": 5.9222261006545576e-05,
      "loss": 0.7344,
      "step": 13654
    },
    {
      "epoch": 2.5376324103326517,
      "grad_norm": 0.778704047203064,
      "learning_rate": 5.92089219247161e-05,
      "loss": 0.9272,
      "step": 13655
    },
    {
      "epoch": 2.537818249396023,
      "grad_norm": 0.6862264275550842,
      "learning_rate": 5.919558371350213e-05,
      "loss": 0.9147,
      "step": 13656
    },
    {
      "epoch": 2.5380040884593944,
      "grad_norm": 0.8819536566734314,
      "learning_rate": 5.9182246373188456e-05,
      "loss": 0.882,
      "step": 13657
    },
    {
      "epoch": 2.5381899275227653,
      "grad_norm": 0.780981183052063,
      "learning_rate": 5.916890990405966e-05,
      "loss": 0.8523,
      "step": 13658
    },
    {
      "epoch": 2.538375766586136,
      "grad_norm": 0.828430712223053,
      "learning_rate": 5.9155574306400395e-05,
      "loss": 1.0052,
      "step": 13659
    },
    {
      "epoch": 2.5385616056495075,
      "grad_norm": 1.1634652614593506,
      "learning_rate": 5.914223958049534e-05,
      "loss": 0.9461,
      "step": 13660
    },
    {
      "epoch": 2.538747444712879,
      "grad_norm": 0.7423953413963318,
      "learning_rate": 5.912890572662903e-05,
      "loss": 0.6167,
      "step": 13661
    },
    {
      "epoch": 2.5389332837762497,
      "grad_norm": 0.7293835878372192,
      "learning_rate": 5.911557274508609e-05,
      "loss": 0.8475,
      "step": 13662
    },
    {
      "epoch": 2.539119122839621,
      "grad_norm": 0.8138513565063477,
      "learning_rate": 5.9102240636151126e-05,
      "loss": 0.9412,
      "step": 13663
    },
    {
      "epoch": 2.539304961902992,
      "grad_norm": 0.8363872766494751,
      "learning_rate": 5.908890940010863e-05,
      "loss": 0.9112,
      "step": 13664
    },
    {
      "epoch": 2.5394908009663633,
      "grad_norm": 0.921448290348053,
      "learning_rate": 5.907557903724317e-05,
      "loss": 0.9629,
      "step": 13665
    },
    {
      "epoch": 2.539676640029734,
      "grad_norm": 0.8595923185348511,
      "learning_rate": 5.906224954783923e-05,
      "loss": 0.9787,
      "step": 13666
    },
    {
      "epoch": 2.5398624790931055,
      "grad_norm": 0.8172317147254944,
      "learning_rate": 5.904892093218134e-05,
      "loss": 0.7755,
      "step": 13667
    },
    {
      "epoch": 2.5400483181564764,
      "grad_norm": 0.8472798466682434,
      "learning_rate": 5.903559319055394e-05,
      "loss": 0.752,
      "step": 13668
    },
    {
      "epoch": 2.5402341572198477,
      "grad_norm": 0.7198326587677002,
      "learning_rate": 5.90222663232415e-05,
      "loss": 0.8107,
      "step": 13669
    },
    {
      "epoch": 2.5404199962832186,
      "grad_norm": 0.8129236102104187,
      "learning_rate": 5.9008940330528505e-05,
      "loss": 0.8908,
      "step": 13670
    },
    {
      "epoch": 2.54060583534659,
      "grad_norm": 0.768535852432251,
      "learning_rate": 5.899561521269929e-05,
      "loss": 0.9418,
      "step": 13671
    },
    {
      "epoch": 2.5407916744099612,
      "grad_norm": 0.9059661626815796,
      "learning_rate": 5.898229097003829e-05,
      "loss": 0.8386,
      "step": 13672
    },
    {
      "epoch": 2.540977513473332,
      "grad_norm": 0.8625373840332031,
      "learning_rate": 5.896896760282995e-05,
      "loss": 1.1597,
      "step": 13673
    },
    {
      "epoch": 2.541163352536703,
      "grad_norm": 0.7552303671836853,
      "learning_rate": 5.895564511135856e-05,
      "loss": 0.886,
      "step": 13674
    },
    {
      "epoch": 2.5413491916000743,
      "grad_norm": 0.7569305896759033,
      "learning_rate": 5.89423234959085e-05,
      "loss": 0.8546,
      "step": 13675
    },
    {
      "epoch": 2.5415350306634457,
      "grad_norm": 0.802854597568512,
      "learning_rate": 5.892900275676407e-05,
      "loss": 0.9316,
      "step": 13676
    },
    {
      "epoch": 2.5417208697268165,
      "grad_norm": 0.7180258631706238,
      "learning_rate": 5.8915682894209626e-05,
      "loss": 0.8168,
      "step": 13677
    },
    {
      "epoch": 2.5419067087901874,
      "grad_norm": 0.8847736120223999,
      "learning_rate": 5.89023639085294e-05,
      "loss": 0.8559,
      "step": 13678
    },
    {
      "epoch": 2.5420925478535588,
      "grad_norm": 0.7706083655357361,
      "learning_rate": 5.888904580000769e-05,
      "loss": 0.9055,
      "step": 13679
    },
    {
      "epoch": 2.54227838691693,
      "grad_norm": 0.9170486330986023,
      "learning_rate": 5.887572856892877e-05,
      "loss": 0.8428,
      "step": 13680
    },
    {
      "epoch": 2.542464225980301,
      "grad_norm": 0.9619743227958679,
      "learning_rate": 5.8862412215576824e-05,
      "loss": 1.0794,
      "step": 13681
    },
    {
      "epoch": 2.5426500650436723,
      "grad_norm": 0.8550994396209717,
      "learning_rate": 5.8849096740236134e-05,
      "loss": 0.907,
      "step": 13682
    },
    {
      "epoch": 2.542835904107043,
      "grad_norm": 0.9619942903518677,
      "learning_rate": 5.88357821431908e-05,
      "loss": 0.9376,
      "step": 13683
    },
    {
      "epoch": 2.5430217431704145,
      "grad_norm": 1.087105393409729,
      "learning_rate": 5.8822468424725075e-05,
      "loss": 0.8564,
      "step": 13684
    },
    {
      "epoch": 2.5432075822337854,
      "grad_norm": 9.038790702819824,
      "learning_rate": 5.880915558512312e-05,
      "loss": 1.658,
      "step": 13685
    },
    {
      "epoch": 2.5433934212971567,
      "grad_norm": 0.8297736644744873,
      "learning_rate": 5.879584362466903e-05,
      "loss": 0.9431,
      "step": 13686
    },
    {
      "epoch": 2.543579260360528,
      "grad_norm": 0.9380455017089844,
      "learning_rate": 5.8782532543646985e-05,
      "loss": 1.0081,
      "step": 13687
    },
    {
      "epoch": 2.543765099423899,
      "grad_norm": 0.6814043521881104,
      "learning_rate": 5.876922234234101e-05,
      "loss": 0.8086,
      "step": 13688
    },
    {
      "epoch": 2.54395093848727,
      "grad_norm": 0.9108039140701294,
      "learning_rate": 5.875591302103525e-05,
      "loss": 0.8482,
      "step": 13689
    },
    {
      "epoch": 2.544136777550641,
      "grad_norm": 0.8931623697280884,
      "learning_rate": 5.874260458001375e-05,
      "loss": 0.9589,
      "step": 13690
    },
    {
      "epoch": 2.5443226166140125,
      "grad_norm": 0.7117545008659363,
      "learning_rate": 5.872929701956054e-05,
      "loss": 0.6862,
      "step": 13691
    },
    {
      "epoch": 2.5445084556773834,
      "grad_norm": 1.050232172012329,
      "learning_rate": 5.871599033995968e-05,
      "loss": 0.9403,
      "step": 13692
    },
    {
      "epoch": 2.5446942947407543,
      "grad_norm": 0.9568449854850769,
      "learning_rate": 5.8702684541495125e-05,
      "loss": 0.9937,
      "step": 13693
    },
    {
      "epoch": 2.5448801338041256,
      "grad_norm": 0.7261242270469666,
      "learning_rate": 5.8689379624450914e-05,
      "loss": 0.7272,
      "step": 13694
    },
    {
      "epoch": 2.545065972867497,
      "grad_norm": 0.8362358808517456,
      "learning_rate": 5.867607558911103e-05,
      "loss": 0.949,
      "step": 13695
    },
    {
      "epoch": 2.545251811930868,
      "grad_norm": 0.9631180763244629,
      "learning_rate": 5.866277243575937e-05,
      "loss": 0.9955,
      "step": 13696
    },
    {
      "epoch": 2.545437650994239,
      "grad_norm": 0.8250423073768616,
      "learning_rate": 5.864947016467992e-05,
      "loss": 0.8525,
      "step": 13697
    },
    {
      "epoch": 2.54562349005761,
      "grad_norm": 0.706535279750824,
      "learning_rate": 5.863616877615656e-05,
      "loss": 0.7951,
      "step": 13698
    },
    {
      "epoch": 2.5458093291209813,
      "grad_norm": 0.8374561667442322,
      "learning_rate": 5.862286827047319e-05,
      "loss": 0.8642,
      "step": 13699
    },
    {
      "epoch": 2.5459951681843522,
      "grad_norm": 0.800125241279602,
      "learning_rate": 5.860956864791369e-05,
      "loss": 0.883,
      "step": 13700
    },
    {
      "epoch": 2.5461810072477236,
      "grad_norm": 0.8778433203697205,
      "learning_rate": 5.859626990876191e-05,
      "loss": 1.1478,
      "step": 13701
    },
    {
      "epoch": 2.5463668463110944,
      "grad_norm": 1.111130952835083,
      "learning_rate": 5.858297205330172e-05,
      "loss": 0.9495,
      "step": 13702
    },
    {
      "epoch": 2.5465526853744658,
      "grad_norm": 0.8640955090522766,
      "learning_rate": 5.85696750818169e-05,
      "loss": 0.8719,
      "step": 13703
    },
    {
      "epoch": 2.5467385244378367,
      "grad_norm": 0.7243440747261047,
      "learning_rate": 5.8556378994591244e-05,
      "loss": 0.7848,
      "step": 13704
    },
    {
      "epoch": 2.546924363501208,
      "grad_norm": 0.844635009765625,
      "learning_rate": 5.8543083791908604e-05,
      "loss": 0.8264,
      "step": 13705
    },
    {
      "epoch": 2.5471102025645793,
      "grad_norm": 0.804513156414032,
      "learning_rate": 5.852978947405267e-05,
      "loss": 1.0957,
      "step": 13706
    },
    {
      "epoch": 2.54729604162795,
      "grad_norm": 0.818187415599823,
      "learning_rate": 5.851649604130723e-05,
      "loss": 0.7591,
      "step": 13707
    },
    {
      "epoch": 2.547481880691321,
      "grad_norm": 0.7252411842346191,
      "learning_rate": 5.8503203493955985e-05,
      "loss": 0.9239,
      "step": 13708
    },
    {
      "epoch": 2.5476677197546924,
      "grad_norm": 0.8514246344566345,
      "learning_rate": 5.848991183228264e-05,
      "loss": 0.961,
      "step": 13709
    },
    {
      "epoch": 2.5478535588180637,
      "grad_norm": 0.8979095816612244,
      "learning_rate": 5.8476621056570924e-05,
      "loss": 1.0529,
      "step": 13710
    },
    {
      "epoch": 2.5480393978814346,
      "grad_norm": 0.7347493767738342,
      "learning_rate": 5.8463331167104453e-05,
      "loss": 0.7018,
      "step": 13711
    },
    {
      "epoch": 2.548225236944806,
      "grad_norm": 0.9283837676048279,
      "learning_rate": 5.84500421641669e-05,
      "loss": 1.0879,
      "step": 13712
    },
    {
      "epoch": 2.548411076008177,
      "grad_norm": 0.9232157468795776,
      "learning_rate": 5.8436754048041896e-05,
      "loss": 0.9902,
      "step": 13713
    },
    {
      "epoch": 2.548596915071548,
      "grad_norm": 0.824123740196228,
      "learning_rate": 5.842346681901302e-05,
      "loss": 0.9658,
      "step": 13714
    },
    {
      "epoch": 2.548782754134919,
      "grad_norm": 0.8550252914428711,
      "learning_rate": 5.841018047736394e-05,
      "loss": 0.7703,
      "step": 13715
    },
    {
      "epoch": 2.5489685931982904,
      "grad_norm": 0.7530178427696228,
      "learning_rate": 5.839689502337817e-05,
      "loss": 0.8716,
      "step": 13716
    },
    {
      "epoch": 2.5491544322616613,
      "grad_norm": 0.9273643493652344,
      "learning_rate": 5.838361045733927e-05,
      "loss": 1.0495,
      "step": 13717
    },
    {
      "epoch": 2.5493402713250326,
      "grad_norm": 0.9969702363014221,
      "learning_rate": 5.837032677953083e-05,
      "loss": 0.9738,
      "step": 13718
    },
    {
      "epoch": 2.5495261103884035,
      "grad_norm": 0.8267450928688049,
      "learning_rate": 5.83570439902363e-05,
      "loss": 0.8757,
      "step": 13719
    },
    {
      "epoch": 2.549711949451775,
      "grad_norm": 0.7460461854934692,
      "learning_rate": 5.8343762089739175e-05,
      "loss": 1.0294,
      "step": 13720
    },
    {
      "epoch": 2.549897788515146,
      "grad_norm": 0.7614554762840271,
      "learning_rate": 5.833048107832301e-05,
      "loss": 0.8374,
      "step": 13721
    },
    {
      "epoch": 2.550083627578517,
      "grad_norm": 0.7778684496879578,
      "learning_rate": 5.831720095627121e-05,
      "loss": 1.0021,
      "step": 13722
    },
    {
      "epoch": 2.550269466641888,
      "grad_norm": 0.8780698776245117,
      "learning_rate": 5.830392172386723e-05,
      "loss": 1.0871,
      "step": 13723
    },
    {
      "epoch": 2.5504553057052592,
      "grad_norm": 0.7678472399711609,
      "learning_rate": 5.829064338139445e-05,
      "loss": 0.8444,
      "step": 13724
    },
    {
      "epoch": 2.5506411447686306,
      "grad_norm": 0.805841863155365,
      "learning_rate": 5.827736592913631e-05,
      "loss": 0.6904,
      "step": 13725
    },
    {
      "epoch": 2.5508269838320015,
      "grad_norm": 0.8294493556022644,
      "learning_rate": 5.826408936737622e-05,
      "loss": 0.9244,
      "step": 13726
    },
    {
      "epoch": 2.5510128228953723,
      "grad_norm": 0.7568511962890625,
      "learning_rate": 5.825081369639753e-05,
      "loss": 0.7845,
      "step": 13727
    },
    {
      "epoch": 2.5511986619587437,
      "grad_norm": 0.8128231763839722,
      "learning_rate": 5.823753891648353e-05,
      "loss": 0.9111,
      "step": 13728
    },
    {
      "epoch": 2.551384501022115,
      "grad_norm": 0.8409896492958069,
      "learning_rate": 5.822426502791766e-05,
      "loss": 0.7971,
      "step": 13729
    },
    {
      "epoch": 2.551570340085486,
      "grad_norm": 0.8081522583961487,
      "learning_rate": 5.8210992030983144e-05,
      "loss": 0.8657,
      "step": 13730
    },
    {
      "epoch": 2.551756179148857,
      "grad_norm": 0.65762859582901,
      "learning_rate": 5.81977199259633e-05,
      "loss": 0.4465,
      "step": 13731
    },
    {
      "epoch": 2.551942018212228,
      "grad_norm": 1.1293163299560547,
      "learning_rate": 5.8184448713141334e-05,
      "loss": 1.0436,
      "step": 13732
    },
    {
      "epoch": 2.5521278572755994,
      "grad_norm": 1.0125041007995605,
      "learning_rate": 5.8171178392800595e-05,
      "loss": 1.0815,
      "step": 13733
    },
    {
      "epoch": 2.5523136963389703,
      "grad_norm": 0.8967159986495972,
      "learning_rate": 5.815790896522428e-05,
      "loss": 1.1234,
      "step": 13734
    },
    {
      "epoch": 2.5524995354023416,
      "grad_norm": 0.7436438202857971,
      "learning_rate": 5.814464043069556e-05,
      "loss": 0.7549,
      "step": 13735
    },
    {
      "epoch": 2.5526853744657125,
      "grad_norm": 0.8675150275230408,
      "learning_rate": 5.813137278949769e-05,
      "loss": 0.8948,
      "step": 13736
    },
    {
      "epoch": 2.552871213529084,
      "grad_norm": 0.7287111282348633,
      "learning_rate": 5.811810604191379e-05,
      "loss": 0.6142,
      "step": 13737
    },
    {
      "epoch": 2.5530570525924547,
      "grad_norm": 0.7745053172111511,
      "learning_rate": 5.8104840188227075e-05,
      "loss": 0.8256,
      "step": 13738
    },
    {
      "epoch": 2.553242891655826,
      "grad_norm": 0.8518672585487366,
      "learning_rate": 5.809157522872065e-05,
      "loss": 0.952,
      "step": 13739
    },
    {
      "epoch": 2.5534287307191974,
      "grad_norm": 0.7259352207183838,
      "learning_rate": 5.807831116367759e-05,
      "loss": 0.871,
      "step": 13740
    },
    {
      "epoch": 2.5536145697825683,
      "grad_norm": 0.8650814294815063,
      "learning_rate": 5.806504799338109e-05,
      "loss": 1.0228,
      "step": 13741
    },
    {
      "epoch": 2.553800408845939,
      "grad_norm": 0.8809137940406799,
      "learning_rate": 5.8051785718114185e-05,
      "loss": 0.8819,
      "step": 13742
    },
    {
      "epoch": 2.5539862479093105,
      "grad_norm": 0.7917113304138184,
      "learning_rate": 5.80385243381599e-05,
      "loss": 0.8703,
      "step": 13743
    },
    {
      "epoch": 2.554172086972682,
      "grad_norm": 0.9165493845939636,
      "learning_rate": 5.802526385380128e-05,
      "loss": 1.0186,
      "step": 13744
    },
    {
      "epoch": 2.5543579260360527,
      "grad_norm": 0.8668243288993835,
      "learning_rate": 5.8012004265321406e-05,
      "loss": 0.9955,
      "step": 13745
    },
    {
      "epoch": 2.554543765099424,
      "grad_norm": 1.047468900680542,
      "learning_rate": 5.799874557300322e-05,
      "loss": 1.0813,
      "step": 13746
    },
    {
      "epoch": 2.554729604162795,
      "grad_norm": 0.7318970561027527,
      "learning_rate": 5.7985487777129765e-05,
      "loss": 1.0389,
      "step": 13747
    },
    {
      "epoch": 2.5549154432261663,
      "grad_norm": 0.8629610538482666,
      "learning_rate": 5.7972230877983933e-05,
      "loss": 0.6949,
      "step": 13748
    },
    {
      "epoch": 2.555101282289537,
      "grad_norm": 0.7724053263664246,
      "learning_rate": 5.795897487584875e-05,
      "loss": 0.9542,
      "step": 13749
    },
    {
      "epoch": 2.5552871213529085,
      "grad_norm": 0.8183825016021729,
      "learning_rate": 5.79457197710071e-05,
      "loss": 0.884,
      "step": 13750
    },
    {
      "epoch": 2.5554729604162794,
      "grad_norm": 1.026491403579712,
      "learning_rate": 5.79324655637419e-05,
      "loss": 1.1477,
      "step": 13751
    },
    {
      "epoch": 2.5556587994796507,
      "grad_norm": 0.7449808120727539,
      "learning_rate": 5.7919212254335987e-05,
      "loss": 0.9825,
      "step": 13752
    },
    {
      "epoch": 2.5558446385430216,
      "grad_norm": 0.8822129368782043,
      "learning_rate": 5.790595984307232e-05,
      "loss": 0.9226,
      "step": 13753
    },
    {
      "epoch": 2.556030477606393,
      "grad_norm": 0.8080644011497498,
      "learning_rate": 5.78927083302337e-05,
      "loss": 0.7673,
      "step": 13754
    },
    {
      "epoch": 2.5562163166697642,
      "grad_norm": 0.775949239730835,
      "learning_rate": 5.787945771610296e-05,
      "loss": 1.1468,
      "step": 13755
    },
    {
      "epoch": 2.556402155733135,
      "grad_norm": 0.9412445425987244,
      "learning_rate": 5.7866208000962873e-05,
      "loss": 0.9846,
      "step": 13756
    },
    {
      "epoch": 2.556587994796506,
      "grad_norm": 0.8279641270637512,
      "learning_rate": 5.785295918509628e-05,
      "loss": 1.0843,
      "step": 13757
    },
    {
      "epoch": 2.5567738338598773,
      "grad_norm": 0.7588843107223511,
      "learning_rate": 5.783971126878598e-05,
      "loss": 0.9479,
      "step": 13758
    },
    {
      "epoch": 2.5569596729232487,
      "grad_norm": 1.0775775909423828,
      "learning_rate": 5.7826464252314683e-05,
      "loss": 1.0048,
      "step": 13759
    },
    {
      "epoch": 2.5571455119866195,
      "grad_norm": 0.7340498566627502,
      "learning_rate": 5.7813218135965105e-05,
      "loss": 0.8742,
      "step": 13760
    },
    {
      "epoch": 2.5573313510499904,
      "grad_norm": 0.872485339641571,
      "learning_rate": 5.779997292002004e-05,
      "loss": 0.9653,
      "step": 13761
    },
    {
      "epoch": 2.5575171901133618,
      "grad_norm": 0.7634876370429993,
      "learning_rate": 5.778672860476212e-05,
      "loss": 0.7495,
      "step": 13762
    },
    {
      "epoch": 2.557703029176733,
      "grad_norm": 0.8036033511161804,
      "learning_rate": 5.7773485190474044e-05,
      "loss": 0.8182,
      "step": 13763
    },
    {
      "epoch": 2.557888868240104,
      "grad_norm": 0.9266873598098755,
      "learning_rate": 5.7760242677438416e-05,
      "loss": 1.1838,
      "step": 13764
    },
    {
      "epoch": 2.5580747073034753,
      "grad_norm": 0.8402783870697021,
      "learning_rate": 5.774700106593797e-05,
      "loss": 0.8673,
      "step": 13765
    },
    {
      "epoch": 2.558260546366846,
      "grad_norm": 0.7656075358390808,
      "learning_rate": 5.7733760356255264e-05,
      "loss": 1.1087,
      "step": 13766
    },
    {
      "epoch": 2.5584463854302175,
      "grad_norm": 0.8482323288917542,
      "learning_rate": 5.772052054867289e-05,
      "loss": 1.0285,
      "step": 13767
    },
    {
      "epoch": 2.5586322244935884,
      "grad_norm": 0.8206349611282349,
      "learning_rate": 5.770728164347343e-05,
      "loss": 0.899,
      "step": 13768
    },
    {
      "epoch": 2.5588180635569597,
      "grad_norm": 0.8543358445167542,
      "learning_rate": 5.769404364093952e-05,
      "loss": 0.8919,
      "step": 13769
    },
    {
      "epoch": 2.559003902620331,
      "grad_norm": 1.1699315309524536,
      "learning_rate": 5.768080654135365e-05,
      "loss": 0.7744,
      "step": 13770
    },
    {
      "epoch": 2.559189741683702,
      "grad_norm": 0.8294997215270996,
      "learning_rate": 5.7667570344998343e-05,
      "loss": 0.9095,
      "step": 13771
    },
    {
      "epoch": 2.559375580747073,
      "grad_norm": 0.8407626748085022,
      "learning_rate": 5.765433505215605e-05,
      "loss": 0.9098,
      "step": 13772
    },
    {
      "epoch": 2.559561419810444,
      "grad_norm": 0.927783191204071,
      "learning_rate": 5.764110066310935e-05,
      "loss": 0.7629,
      "step": 13773
    },
    {
      "epoch": 2.5597472588738155,
      "grad_norm": 0.8324963450431824,
      "learning_rate": 5.762786717814067e-05,
      "loss": 0.8243,
      "step": 13774
    },
    {
      "epoch": 2.5599330979371864,
      "grad_norm": 0.8019198775291443,
      "learning_rate": 5.761463459753245e-05,
      "loss": 0.809,
      "step": 13775
    },
    {
      "epoch": 2.5601189370005573,
      "grad_norm": 0.9161101579666138,
      "learning_rate": 5.760140292156708e-05,
      "loss": 0.8882,
      "step": 13776
    },
    {
      "epoch": 2.5603047760639286,
      "grad_norm": 1.7970740795135498,
      "learning_rate": 5.7588172150527006e-05,
      "loss": 1.314,
      "step": 13777
    },
    {
      "epoch": 2.5604906151273,
      "grad_norm": 1.0165661573410034,
      "learning_rate": 5.7574942284694656e-05,
      "loss": 0.9316,
      "step": 13778
    },
    {
      "epoch": 2.560676454190671,
      "grad_norm": 0.8424798250198364,
      "learning_rate": 5.7561713324352365e-05,
      "loss": 1.0966,
      "step": 13779
    },
    {
      "epoch": 2.560862293254042,
      "grad_norm": 0.7895606756210327,
      "learning_rate": 5.754848526978242e-05,
      "loss": 0.7786,
      "step": 13780
    },
    {
      "epoch": 2.561048132317413,
      "grad_norm": 0.6329425573348999,
      "learning_rate": 5.753525812126726e-05,
      "loss": 0.6509,
      "step": 13781
    },
    {
      "epoch": 2.5612339713807843,
      "grad_norm": 0.8896647691726685,
      "learning_rate": 5.7522031879089133e-05,
      "loss": 1.1891,
      "step": 13782
    },
    {
      "epoch": 2.5614198104441552,
      "grad_norm": 0.838678240776062,
      "learning_rate": 5.7508806543530346e-05,
      "loss": 1.099,
      "step": 13783
    },
    {
      "epoch": 2.5616056495075266,
      "grad_norm": 0.7234699130058289,
      "learning_rate": 5.7495582114873114e-05,
      "loss": 0.8925,
      "step": 13784
    },
    {
      "epoch": 2.5617914885708974,
      "grad_norm": 0.9089459180831909,
      "learning_rate": 5.7482358593399796e-05,
      "loss": 0.8512,
      "step": 13785
    },
    {
      "epoch": 2.5619773276342688,
      "grad_norm": 0.8577240705490112,
      "learning_rate": 5.7469135979392576e-05,
      "loss": 0.8033,
      "step": 13786
    },
    {
      "epoch": 2.5621631666976397,
      "grad_norm": 0.7673660516738892,
      "learning_rate": 5.7455914273133646e-05,
      "loss": 1.0081,
      "step": 13787
    },
    {
      "epoch": 2.562349005761011,
      "grad_norm": 0.7493413686752319,
      "learning_rate": 5.744269347490514e-05,
      "loss": 0.7829,
      "step": 13788
    },
    {
      "epoch": 2.5625348448243823,
      "grad_norm": 0.8880299925804138,
      "learning_rate": 5.74294735849894e-05,
      "loss": 0.6736,
      "step": 13789
    },
    {
      "epoch": 2.562720683887753,
      "grad_norm": 0.787715494632721,
      "learning_rate": 5.7416254603668484e-05,
      "loss": 1.0808,
      "step": 13790
    },
    {
      "epoch": 2.562906522951124,
      "grad_norm": 0.8745567798614502,
      "learning_rate": 5.7403036531224544e-05,
      "loss": 1.0884,
      "step": 13791
    },
    {
      "epoch": 2.5630923620144954,
      "grad_norm": 0.8465336561203003,
      "learning_rate": 5.7389819367939656e-05,
      "loss": 0.7757,
      "step": 13792
    },
    {
      "epoch": 2.5632782010778667,
      "grad_norm": 0.9584832787513733,
      "learning_rate": 5.7376603114095986e-05,
      "loss": 1.0241,
      "step": 13793
    },
    {
      "epoch": 2.5634640401412376,
      "grad_norm": 0.758416473865509,
      "learning_rate": 5.736338776997556e-05,
      "loss": 0.972,
      "step": 13794
    },
    {
      "epoch": 2.563649879204609,
      "grad_norm": 1.0920721292495728,
      "learning_rate": 5.735017333586048e-05,
      "loss": 0.9077,
      "step": 13795
    },
    {
      "epoch": 2.56383571826798,
      "grad_norm": 0.7854132652282715,
      "learning_rate": 5.733695981203271e-05,
      "loss": 0.7501,
      "step": 13796
    },
    {
      "epoch": 2.564021557331351,
      "grad_norm": 0.7899761199951172,
      "learning_rate": 5.732374719877436e-05,
      "loss": 0.9115,
      "step": 13797
    },
    {
      "epoch": 2.564207396394722,
      "grad_norm": 0.7128276824951172,
      "learning_rate": 5.731053549636733e-05,
      "loss": 0.7982,
      "step": 13798
    },
    {
      "epoch": 2.5643932354580934,
      "grad_norm": 0.9120762944221497,
      "learning_rate": 5.729732470509372e-05,
      "loss": 0.9454,
      "step": 13799
    },
    {
      "epoch": 2.5645790745214643,
      "grad_norm": 0.9080113768577576,
      "learning_rate": 5.7284114825235416e-05,
      "loss": 0.9263,
      "step": 13800
    },
    {
      "epoch": 2.5647649135848356,
      "grad_norm": 0.7981117963790894,
      "learning_rate": 5.727090585707439e-05,
      "loss": 0.7399,
      "step": 13801
    },
    {
      "epoch": 2.5649507526482065,
      "grad_norm": 1.0038573741912842,
      "learning_rate": 5.725769780089256e-05,
      "loss": 1.034,
      "step": 13802
    },
    {
      "epoch": 2.565136591711578,
      "grad_norm": 1.3188296556472778,
      "learning_rate": 5.7244490656971815e-05,
      "loss": 1.2923,
      "step": 13803
    },
    {
      "epoch": 2.565322430774949,
      "grad_norm": 0.8303518891334534,
      "learning_rate": 5.723128442559401e-05,
      "loss": 0.7213,
      "step": 13804
    },
    {
      "epoch": 2.56550826983832,
      "grad_norm": 0.918181300163269,
      "learning_rate": 5.721807910704109e-05,
      "loss": 0.9924,
      "step": 13805
    },
    {
      "epoch": 2.565694108901691,
      "grad_norm": 0.896000325679779,
      "learning_rate": 5.720487470159484e-05,
      "loss": 0.9482,
      "step": 13806
    },
    {
      "epoch": 2.5658799479650622,
      "grad_norm": 0.8845444321632385,
      "learning_rate": 5.7191671209537124e-05,
      "loss": 1.0871,
      "step": 13807
    },
    {
      "epoch": 2.5660657870284336,
      "grad_norm": 0.8426279425621033,
      "learning_rate": 5.717846863114966e-05,
      "loss": 1.1556,
      "step": 13808
    },
    {
      "epoch": 2.5662516260918045,
      "grad_norm": 0.9557228088378906,
      "learning_rate": 5.716526696671431e-05,
      "loss": 1.0941,
      "step": 13809
    },
    {
      "epoch": 2.5664374651551753,
      "grad_norm": 0.5845919251441956,
      "learning_rate": 5.715206621651288e-05,
      "loss": 0.4136,
      "step": 13810
    },
    {
      "epoch": 2.5666233042185467,
      "grad_norm": 0.8088479042053223,
      "learning_rate": 5.713886638082705e-05,
      "loss": 0.9118,
      "step": 13811
    },
    {
      "epoch": 2.566809143281918,
      "grad_norm": 0.937294602394104,
      "learning_rate": 5.712566745993853e-05,
      "loss": 0.9451,
      "step": 13812
    },
    {
      "epoch": 2.566994982345289,
      "grad_norm": 1.0141209363937378,
      "learning_rate": 5.711246945412909e-05,
      "loss": 1.0489,
      "step": 13813
    },
    {
      "epoch": 2.56718082140866,
      "grad_norm": 0.970757246017456,
      "learning_rate": 5.7099272363680414e-05,
      "loss": 0.9403,
      "step": 13814
    },
    {
      "epoch": 2.567366660472031,
      "grad_norm": 0.8191547989845276,
      "learning_rate": 5.708607618887414e-05,
      "loss": 0.7734,
      "step": 13815
    },
    {
      "epoch": 2.5675524995354024,
      "grad_norm": 1.0061275959014893,
      "learning_rate": 5.707288092999188e-05,
      "loss": 0.9257,
      "step": 13816
    },
    {
      "epoch": 2.5677383385987733,
      "grad_norm": 0.7247336506843567,
      "learning_rate": 5.705968658731535e-05,
      "loss": 0.828,
      "step": 13817
    },
    {
      "epoch": 2.5679241776621446,
      "grad_norm": 0.8073325157165527,
      "learning_rate": 5.7046493161126135e-05,
      "loss": 0.9172,
      "step": 13818
    },
    {
      "epoch": 2.568110016725516,
      "grad_norm": 0.9320276975631714,
      "learning_rate": 5.703330065170577e-05,
      "loss": 1.0238,
      "step": 13819
    },
    {
      "epoch": 2.568295855788887,
      "grad_norm": 0.7512152791023254,
      "learning_rate": 5.7020109059335855e-05,
      "loss": 0.8337,
      "step": 13820
    },
    {
      "epoch": 2.5684816948522577,
      "grad_norm": 0.7233678102493286,
      "learning_rate": 5.7006918384298015e-05,
      "loss": 1.1078,
      "step": 13821
    },
    {
      "epoch": 2.568667533915629,
      "grad_norm": 0.8729320764541626,
      "learning_rate": 5.699372862687371e-05,
      "loss": 1.0297,
      "step": 13822
    },
    {
      "epoch": 2.5688533729790004,
      "grad_norm": 0.7991220355033875,
      "learning_rate": 5.6980539787344454e-05,
      "loss": 0.7869,
      "step": 13823
    },
    {
      "epoch": 2.5690392120423713,
      "grad_norm": 0.8474665880203247,
      "learning_rate": 5.696735186599174e-05,
      "loss": 1.0024,
      "step": 13824
    },
    {
      "epoch": 2.569225051105742,
      "grad_norm": 0.7745363712310791,
      "learning_rate": 5.695416486309707e-05,
      "loss": 0.8602,
      "step": 13825
    },
    {
      "epoch": 2.5694108901691135,
      "grad_norm": 0.7119348049163818,
      "learning_rate": 5.694097877894189e-05,
      "loss": 1.0427,
      "step": 13826
    },
    {
      "epoch": 2.569596729232485,
      "grad_norm": 0.8804404735565186,
      "learning_rate": 5.6927793613807624e-05,
      "loss": 0.9245,
      "step": 13827
    },
    {
      "epoch": 2.5697825682958557,
      "grad_norm": 0.7989339232444763,
      "learning_rate": 5.691460936797566e-05,
      "loss": 0.8493,
      "step": 13828
    },
    {
      "epoch": 2.569968407359227,
      "grad_norm": 0.8049229979515076,
      "learning_rate": 5.6901426041727447e-05,
      "loss": 0.989,
      "step": 13829
    },
    {
      "epoch": 2.570154246422598,
      "grad_norm": 0.7842298746109009,
      "learning_rate": 5.6888243635344295e-05,
      "loss": 0.8367,
      "step": 13830
    },
    {
      "epoch": 2.5703400854859693,
      "grad_norm": 0.72968590259552,
      "learning_rate": 5.687506214910765e-05,
      "loss": 0.9222,
      "step": 13831
    },
    {
      "epoch": 2.57052592454934,
      "grad_norm": 1.1353753805160522,
      "learning_rate": 5.686188158329875e-05,
      "loss": 0.9685,
      "step": 13832
    },
    {
      "epoch": 2.5707117636127115,
      "grad_norm": 0.8593534827232361,
      "learning_rate": 5.6848701938199e-05,
      "loss": 0.9871,
      "step": 13833
    },
    {
      "epoch": 2.5708976026760824,
      "grad_norm": 0.7766969203948975,
      "learning_rate": 5.683552321408967e-05,
      "loss": 0.9892,
      "step": 13834
    },
    {
      "epoch": 2.5710834417394537,
      "grad_norm": 0.7157514095306396,
      "learning_rate": 5.6822345411251995e-05,
      "loss": 0.8832,
      "step": 13835
    },
    {
      "epoch": 2.5712692808028246,
      "grad_norm": 0.8342850804328918,
      "learning_rate": 5.680916852996725e-05,
      "loss": 1.031,
      "step": 13836
    },
    {
      "epoch": 2.571455119866196,
      "grad_norm": 0.8797349333763123,
      "learning_rate": 5.679599257051671e-05,
      "loss": 0.9466,
      "step": 13837
    },
    {
      "epoch": 2.5716409589295672,
      "grad_norm": 0.7610642910003662,
      "learning_rate": 5.678281753318157e-05,
      "loss": 0.8896,
      "step": 13838
    },
    {
      "epoch": 2.571826797992938,
      "grad_norm": 1.056077241897583,
      "learning_rate": 5.676964341824299e-05,
      "loss": 0.8226,
      "step": 13839
    },
    {
      "epoch": 2.572012637056309,
      "grad_norm": 0.898135244846344,
      "learning_rate": 5.6756470225982186e-05,
      "loss": 0.7951,
      "step": 13840
    },
    {
      "epoch": 2.5721984761196803,
      "grad_norm": 0.8857336640357971,
      "learning_rate": 5.674329795668034e-05,
      "loss": 0.8316,
      "step": 13841
    },
    {
      "epoch": 2.5723843151830517,
      "grad_norm": 0.7753831744194031,
      "learning_rate": 5.673012661061858e-05,
      "loss": 0.9243,
      "step": 13842
    },
    {
      "epoch": 2.5725701542464225,
      "grad_norm": 0.8753922581672668,
      "learning_rate": 5.671695618807802e-05,
      "loss": 0.7834,
      "step": 13843
    },
    {
      "epoch": 2.572755993309794,
      "grad_norm": 0.8811227679252625,
      "learning_rate": 5.670378668933971e-05,
      "loss": 0.8982,
      "step": 13844
    },
    {
      "epoch": 2.5729418323731648,
      "grad_norm": 0.912420928478241,
      "learning_rate": 5.669061811468481e-05,
      "loss": 1.0495,
      "step": 13845
    },
    {
      "epoch": 2.573127671436536,
      "grad_norm": 0.7488943338394165,
      "learning_rate": 5.667745046439434e-05,
      "loss": 0.9085,
      "step": 13846
    },
    {
      "epoch": 2.573313510499907,
      "grad_norm": 0.8022164702415466,
      "learning_rate": 5.666428373874936e-05,
      "loss": 0.8339,
      "step": 13847
    },
    {
      "epoch": 2.5734993495632783,
      "grad_norm": 0.8374843001365662,
      "learning_rate": 5.665111793803082e-05,
      "loss": 0.9486,
      "step": 13848
    },
    {
      "epoch": 2.573685188626649,
      "grad_norm": 0.6516796350479126,
      "learning_rate": 5.663795306251982e-05,
      "loss": 0.6117,
      "step": 13849
    },
    {
      "epoch": 2.5738710276900205,
      "grad_norm": 0.8019716739654541,
      "learning_rate": 5.662478911249731e-05,
      "loss": 0.7115,
      "step": 13850
    },
    {
      "epoch": 2.5740568667533914,
      "grad_norm": 0.9151284098625183,
      "learning_rate": 5.6611626088244194e-05,
      "loss": 1.0861,
      "step": 13851
    },
    {
      "epoch": 2.5742427058167627,
      "grad_norm": 0.8689349889755249,
      "learning_rate": 5.659846399004147e-05,
      "loss": 0.9528,
      "step": 13852
    },
    {
      "epoch": 2.574428544880134,
      "grad_norm": 0.7861039042472839,
      "learning_rate": 5.658530281817008e-05,
      "loss": 0.8674,
      "step": 13853
    },
    {
      "epoch": 2.574614383943505,
      "grad_norm": 0.7574406266212463,
      "learning_rate": 5.65721425729109e-05,
      "loss": 0.9153,
      "step": 13854
    },
    {
      "epoch": 2.574800223006876,
      "grad_norm": 0.7896664142608643,
      "learning_rate": 5.655898325454483e-05,
      "loss": 0.8182,
      "step": 13855
    },
    {
      "epoch": 2.574986062070247,
      "grad_norm": 0.8285431265830994,
      "learning_rate": 5.6545824863352645e-05,
      "loss": 1.0559,
      "step": 13856
    },
    {
      "epoch": 2.5751719011336185,
      "grad_norm": 0.7879959940910339,
      "learning_rate": 5.653266739961532e-05,
      "loss": 0.7827,
      "step": 13857
    },
    {
      "epoch": 2.5753577401969894,
      "grad_norm": 0.933560311794281,
      "learning_rate": 5.651951086361361e-05,
      "loss": 1.1124,
      "step": 13858
    },
    {
      "epoch": 2.5755435792603603,
      "grad_norm": 1.057924747467041,
      "learning_rate": 5.6506355255628284e-05,
      "loss": 1.003,
      "step": 13859
    },
    {
      "epoch": 2.5757294183237316,
      "grad_norm": 0.8306922912597656,
      "learning_rate": 5.649320057594021e-05,
      "loss": 0.8607,
      "step": 13860
    },
    {
      "epoch": 2.575915257387103,
      "grad_norm": 0.7214329838752747,
      "learning_rate": 5.648004682483008e-05,
      "loss": 0.5074,
      "step": 13861
    },
    {
      "epoch": 2.576101096450474,
      "grad_norm": 0.9435710310935974,
      "learning_rate": 5.646689400257868e-05,
      "loss": 1.0446,
      "step": 13862
    },
    {
      "epoch": 2.576286935513845,
      "grad_norm": 0.7284060716629028,
      "learning_rate": 5.645374210946674e-05,
      "loss": 0.7227,
      "step": 13863
    },
    {
      "epoch": 2.576472774577216,
      "grad_norm": 0.8803086280822754,
      "learning_rate": 5.644059114577491e-05,
      "loss": 0.9937,
      "step": 13864
    },
    {
      "epoch": 2.5766586136405873,
      "grad_norm": 0.7651980519294739,
      "learning_rate": 5.642744111178394e-05,
      "loss": 0.7798,
      "step": 13865
    },
    {
      "epoch": 2.5768444527039582,
      "grad_norm": 0.836889922618866,
      "learning_rate": 5.641429200777446e-05,
      "loss": 0.8243,
      "step": 13866
    },
    {
      "epoch": 2.5770302917673296,
      "grad_norm": 0.8004657626152039,
      "learning_rate": 5.640114383402714e-05,
      "loss": 1.093,
      "step": 13867
    },
    {
      "epoch": 2.5772161308307004,
      "grad_norm": 1.3861210346221924,
      "learning_rate": 5.6387996590822535e-05,
      "loss": 1.1794,
      "step": 13868
    },
    {
      "epoch": 2.5774019698940718,
      "grad_norm": 0.8555946946144104,
      "learning_rate": 5.637485027844134e-05,
      "loss": 0.8745,
      "step": 13869
    },
    {
      "epoch": 2.5775878089574427,
      "grad_norm": 0.8644457459449768,
      "learning_rate": 5.63617048971641e-05,
      "loss": 1.1022,
      "step": 13870
    },
    {
      "epoch": 2.577773648020814,
      "grad_norm": 0.8424917459487915,
      "learning_rate": 5.634856044727136e-05,
      "loss": 0.9504,
      "step": 13871
    },
    {
      "epoch": 2.5779594870841853,
      "grad_norm": 0.8418530821800232,
      "learning_rate": 5.633541692904367e-05,
      "loss": 0.9645,
      "step": 13872
    },
    {
      "epoch": 2.578145326147556,
      "grad_norm": 0.7905413508415222,
      "learning_rate": 5.632227434276163e-05,
      "loss": 0.9744,
      "step": 13873
    },
    {
      "epoch": 2.578331165210927,
      "grad_norm": 0.9202725887298584,
      "learning_rate": 5.630913268870568e-05,
      "loss": 0.9128,
      "step": 13874
    },
    {
      "epoch": 2.5785170042742984,
      "grad_norm": 0.6845600605010986,
      "learning_rate": 5.6295991967156314e-05,
      "loss": 0.9016,
      "step": 13875
    },
    {
      "epoch": 2.5787028433376697,
      "grad_norm": 1.0831317901611328,
      "learning_rate": 5.6282852178393966e-05,
      "loss": 1.083,
      "step": 13876
    },
    {
      "epoch": 2.5788886824010406,
      "grad_norm": 0.9119095802307129,
      "learning_rate": 5.6269713322699146e-05,
      "loss": 0.9246,
      "step": 13877
    },
    {
      "epoch": 2.579074521464412,
      "grad_norm": 0.8733465671539307,
      "learning_rate": 5.625657540035226e-05,
      "loss": 0.9341,
      "step": 13878
    },
    {
      "epoch": 2.579260360527783,
      "grad_norm": 0.7767495512962341,
      "learning_rate": 5.624343841163367e-05,
      "loss": 0.8646,
      "step": 13879
    },
    {
      "epoch": 2.579446199591154,
      "grad_norm": 1.0635108947753906,
      "learning_rate": 5.6230302356823825e-05,
      "loss": 0.8414,
      "step": 13880
    },
    {
      "epoch": 2.579632038654525,
      "grad_norm": 0.7466744184494019,
      "learning_rate": 5.621716723620307e-05,
      "loss": 0.9437,
      "step": 13881
    },
    {
      "epoch": 2.5798178777178964,
      "grad_norm": 1.0021867752075195,
      "learning_rate": 5.620403305005172e-05,
      "loss": 0.8329,
      "step": 13882
    },
    {
      "epoch": 2.5800037167812673,
      "grad_norm": 0.9675542116165161,
      "learning_rate": 5.6190899798650156e-05,
      "loss": 0.7331,
      "step": 13883
    },
    {
      "epoch": 2.5801895558446386,
      "grad_norm": 0.7637116312980652,
      "learning_rate": 5.617776748227861e-05,
      "loss": 0.8081,
      "step": 13884
    },
    {
      "epoch": 2.5803753949080095,
      "grad_norm": 0.7900305390357971,
      "learning_rate": 5.616463610121746e-05,
      "loss": 0.8334,
      "step": 13885
    },
    {
      "epoch": 2.580561233971381,
      "grad_norm": 0.7600497603416443,
      "learning_rate": 5.615150565574694e-05,
      "loss": 0.9039,
      "step": 13886
    },
    {
      "epoch": 2.580747073034752,
      "grad_norm": 0.8872308135032654,
      "learning_rate": 5.613837614614727e-05,
      "loss": 0.7615,
      "step": 13887
    },
    {
      "epoch": 2.580932912098123,
      "grad_norm": 0.7821314334869385,
      "learning_rate": 5.612524757269866e-05,
      "loss": 0.8217,
      "step": 13888
    },
    {
      "epoch": 2.581118751161494,
      "grad_norm": 0.8883917331695557,
      "learning_rate": 5.6112119935681375e-05,
      "loss": 0.915,
      "step": 13889
    },
    {
      "epoch": 2.5813045902248652,
      "grad_norm": 0.7755342721939087,
      "learning_rate": 5.6098993235375576e-05,
      "loss": 1.0398,
      "step": 13890
    },
    {
      "epoch": 2.5814904292882366,
      "grad_norm": 0.9562380909919739,
      "learning_rate": 5.60858674720614e-05,
      "loss": 0.9848,
      "step": 13891
    },
    {
      "epoch": 2.5816762683516075,
      "grad_norm": 0.9915200471878052,
      "learning_rate": 5.607274264601907e-05,
      "loss": 1.1596,
      "step": 13892
    },
    {
      "epoch": 2.581862107414979,
      "grad_norm": 0.8348504900932312,
      "learning_rate": 5.605961875752861e-05,
      "loss": 1.2279,
      "step": 13893
    },
    {
      "epoch": 2.5820479464783497,
      "grad_norm": 0.7134225368499756,
      "learning_rate": 5.604649580687022e-05,
      "loss": 0.8465,
      "step": 13894
    },
    {
      "epoch": 2.582233785541721,
      "grad_norm": 0.6985647082328796,
      "learning_rate": 5.603337379432396e-05,
      "loss": 0.7941,
      "step": 13895
    },
    {
      "epoch": 2.582419624605092,
      "grad_norm": 0.7442715167999268,
      "learning_rate": 5.602025272016985e-05,
      "loss": 0.908,
      "step": 13896
    },
    {
      "epoch": 2.582605463668463,
      "grad_norm": 0.7957456111907959,
      "learning_rate": 5.6007132584688014e-05,
      "loss": 0.8506,
      "step": 13897
    },
    {
      "epoch": 2.582791302731834,
      "grad_norm": 0.8388959169387817,
      "learning_rate": 5.599401338815844e-05,
      "loss": 0.7865,
      "step": 13898
    },
    {
      "epoch": 2.5829771417952054,
      "grad_norm": 1.0251314640045166,
      "learning_rate": 5.598089513086108e-05,
      "loss": 1.122,
      "step": 13899
    },
    {
      "epoch": 2.5831629808585763,
      "grad_norm": 0.863593339920044,
      "learning_rate": 5.596777781307602e-05,
      "loss": 0.7528,
      "step": 13900
    },
    {
      "epoch": 2.5833488199219476,
      "grad_norm": 0.8443089723587036,
      "learning_rate": 5.595466143508318e-05,
      "loss": 0.8237,
      "step": 13901
    },
    {
      "epoch": 2.583534658985319,
      "grad_norm": 0.8671047687530518,
      "learning_rate": 5.5941545997162506e-05,
      "loss": 0.988,
      "step": 13902
    },
    {
      "epoch": 2.58372049804869,
      "grad_norm": 0.8505552411079407,
      "learning_rate": 5.592843149959388e-05,
      "loss": 0.7937,
      "step": 13903
    },
    {
      "epoch": 2.5839063371120607,
      "grad_norm": 0.7884801626205444,
      "learning_rate": 5.5915317942657264e-05,
      "loss": 0.942,
      "step": 13904
    },
    {
      "epoch": 2.584092176175432,
      "grad_norm": 0.7203455567359924,
      "learning_rate": 5.5902205326632576e-05,
      "loss": 0.8007,
      "step": 13905
    },
    {
      "epoch": 2.5842780152388034,
      "grad_norm": 0.9898548126220703,
      "learning_rate": 5.5889093651799626e-05,
      "loss": 0.9014,
      "step": 13906
    },
    {
      "epoch": 2.5844638543021743,
      "grad_norm": 1.3909528255462646,
      "learning_rate": 5.587598291843829e-05,
      "loss": 1.3544,
      "step": 13907
    },
    {
      "epoch": 2.584649693365545,
      "grad_norm": 0.8734009861946106,
      "learning_rate": 5.586287312682833e-05,
      "loss": 0.8794,
      "step": 13908
    },
    {
      "epoch": 2.5848355324289165,
      "grad_norm": 0.6974748373031616,
      "learning_rate": 5.584976427724964e-05,
      "loss": 0.7835,
      "step": 13909
    },
    {
      "epoch": 2.585021371492288,
      "grad_norm": 0.9306507706642151,
      "learning_rate": 5.5836656369981965e-05,
      "loss": 0.8179,
      "step": 13910
    },
    {
      "epoch": 2.5852072105556587,
      "grad_norm": 0.7667589783668518,
      "learning_rate": 5.582354940530504e-05,
      "loss": 0.9037,
      "step": 13911
    },
    {
      "epoch": 2.58539304961903,
      "grad_norm": 0.7419784069061279,
      "learning_rate": 5.581044338349868e-05,
      "loss": 0.8633,
      "step": 13912
    },
    {
      "epoch": 2.585578888682401,
      "grad_norm": 0.8643168807029724,
      "learning_rate": 5.579733830484257e-05,
      "loss": 1.0308,
      "step": 13913
    },
    {
      "epoch": 2.5857647277457723,
      "grad_norm": 0.7932475805282593,
      "learning_rate": 5.578423416961639e-05,
      "loss": 1.003,
      "step": 13914
    },
    {
      "epoch": 2.585950566809143,
      "grad_norm": 0.8723037838935852,
      "learning_rate": 5.577113097809989e-05,
      "loss": 1.0774,
      "step": 13915
    },
    {
      "epoch": 2.5861364058725145,
      "grad_norm": 0.81297367811203,
      "learning_rate": 5.575802873057265e-05,
      "loss": 0.993,
      "step": 13916
    },
    {
      "epoch": 2.5863222449358854,
      "grad_norm": 0.7683242559432983,
      "learning_rate": 5.574492742731441e-05,
      "loss": 0.8725,
      "step": 13917
    },
    {
      "epoch": 2.5865080839992567,
      "grad_norm": 0.7096510529518127,
      "learning_rate": 5.573182706860476e-05,
      "loss": 0.8755,
      "step": 13918
    },
    {
      "epoch": 2.5866939230626276,
      "grad_norm": 0.8059236407279968,
      "learning_rate": 5.571872765472324e-05,
      "loss": 0.9017,
      "step": 13919
    },
    {
      "epoch": 2.586879762125999,
      "grad_norm": 0.8255543112754822,
      "learning_rate": 5.5705629185949526e-05,
      "loss": 0.8532,
      "step": 13920
    },
    {
      "epoch": 2.5870656011893702,
      "grad_norm": 0.8200182914733887,
      "learning_rate": 5.5692531662563165e-05,
      "loss": 0.8727,
      "step": 13921
    },
    {
      "epoch": 2.587251440252741,
      "grad_norm": 0.9070665240287781,
      "learning_rate": 5.5679435084843655e-05,
      "loss": 0.7226,
      "step": 13922
    },
    {
      "epoch": 2.587437279316112,
      "grad_norm": 0.8500924110412598,
      "learning_rate": 5.566633945307052e-05,
      "loss": 0.9021,
      "step": 13923
    },
    {
      "epoch": 2.5876231183794833,
      "grad_norm": 0.7485762238502502,
      "learning_rate": 5.565324476752329e-05,
      "loss": 0.7641,
      "step": 13924
    },
    {
      "epoch": 2.5878089574428547,
      "grad_norm": 0.9648006558418274,
      "learning_rate": 5.564015102848149e-05,
      "loss": 0.8558,
      "step": 13925
    },
    {
      "epoch": 2.5879947965062255,
      "grad_norm": 0.9895826578140259,
      "learning_rate": 5.562705823622454e-05,
      "loss": 0.8674,
      "step": 13926
    },
    {
      "epoch": 2.588180635569597,
      "grad_norm": 0.773406982421875,
      "learning_rate": 5.561396639103189e-05,
      "loss": 0.8504,
      "step": 13927
    },
    {
      "epoch": 2.5883664746329678,
      "grad_norm": 0.9207121133804321,
      "learning_rate": 5.560087549318291e-05,
      "loss": 0.8726,
      "step": 13928
    },
    {
      "epoch": 2.588552313696339,
      "grad_norm": 0.8121840357780457,
      "learning_rate": 5.558778554295709e-05,
      "loss": 0.8818,
      "step": 13929
    },
    {
      "epoch": 2.58873815275971,
      "grad_norm": 1.0189287662506104,
      "learning_rate": 5.557469654063377e-05,
      "loss": 1.0176,
      "step": 13930
    },
    {
      "epoch": 2.5889239918230813,
      "grad_norm": 0.8649277687072754,
      "learning_rate": 5.556160848649228e-05,
      "loss": 1.0651,
      "step": 13931
    },
    {
      "epoch": 2.589109830886452,
      "grad_norm": 0.9580209851264954,
      "learning_rate": 5.5548521380812036e-05,
      "loss": 0.9707,
      "step": 13932
    },
    {
      "epoch": 2.5892956699498235,
      "grad_norm": 0.760979175567627,
      "learning_rate": 5.5535435223872325e-05,
      "loss": 0.9167,
      "step": 13933
    },
    {
      "epoch": 2.5894815090131944,
      "grad_norm": 0.9788494110107422,
      "learning_rate": 5.552235001595245e-05,
      "loss": 0.9616,
      "step": 13934
    },
    {
      "epoch": 2.5896673480765657,
      "grad_norm": 0.8736420273780823,
      "learning_rate": 5.550926575733164e-05,
      "loss": 0.8962,
      "step": 13935
    },
    {
      "epoch": 2.589853187139937,
      "grad_norm": 0.8266481161117554,
      "learning_rate": 5.54961824482892e-05,
      "loss": 0.9473,
      "step": 13936
    },
    {
      "epoch": 2.590039026203308,
      "grad_norm": 0.9490053057670593,
      "learning_rate": 5.548310008910442e-05,
      "loss": 0.7363,
      "step": 13937
    },
    {
      "epoch": 2.590224865266679,
      "grad_norm": 0.9110895395278931,
      "learning_rate": 5.547001868005648e-05,
      "loss": 1.0962,
      "step": 13938
    },
    {
      "epoch": 2.59041070433005,
      "grad_norm": 0.7902200818061829,
      "learning_rate": 5.545693822142453e-05,
      "loss": 0.8136,
      "step": 13939
    },
    {
      "epoch": 2.5905965433934215,
      "grad_norm": 0.783957302570343,
      "learning_rate": 5.5443858713487827e-05,
      "loss": 1.0766,
      "step": 13940
    },
    {
      "epoch": 2.5907823824567924,
      "grad_norm": 0.8990013003349304,
      "learning_rate": 5.54307801565255e-05,
      "loss": 1.0653,
      "step": 13941
    },
    {
      "epoch": 2.5909682215201633,
      "grad_norm": 0.8461039662361145,
      "learning_rate": 5.5417702550816684e-05,
      "loss": 0.9389,
      "step": 13942
    },
    {
      "epoch": 2.5911540605835346,
      "grad_norm": 0.9367736577987671,
      "learning_rate": 5.540462589664047e-05,
      "loss": 0.832,
      "step": 13943
    },
    {
      "epoch": 2.591339899646906,
      "grad_norm": 0.7142069935798645,
      "learning_rate": 5.539155019427602e-05,
      "loss": 0.9313,
      "step": 13944
    },
    {
      "epoch": 2.591525738710277,
      "grad_norm": 0.814614474773407,
      "learning_rate": 5.537847544400234e-05,
      "loss": 1.0447,
      "step": 13945
    },
    {
      "epoch": 2.591711577773648,
      "grad_norm": 2.7057414054870605,
      "learning_rate": 5.536540164609858e-05,
      "loss": 1.3782,
      "step": 13946
    },
    {
      "epoch": 2.591897416837019,
      "grad_norm": 0.7128685712814331,
      "learning_rate": 5.5352328800843724e-05,
      "loss": 0.8777,
      "step": 13947
    },
    {
      "epoch": 2.5920832559003903,
      "grad_norm": 0.7369137406349182,
      "learning_rate": 5.533925690851673e-05,
      "loss": 0.8638,
      "step": 13948
    },
    {
      "epoch": 2.5922690949637612,
      "grad_norm": 0.902484118938446,
      "learning_rate": 5.5326185969396714e-05,
      "loss": 1.0354,
      "step": 13949
    },
    {
      "epoch": 2.5924549340271326,
      "grad_norm": 0.9577115774154663,
      "learning_rate": 5.531311598376261e-05,
      "loss": 0.8224,
      "step": 13950
    },
    {
      "epoch": 2.592640773090504,
      "grad_norm": 0.8262429237365723,
      "learning_rate": 5.530004695189329e-05,
      "loss": 0.9783,
      "step": 13951
    },
    {
      "epoch": 2.5928266121538748,
      "grad_norm": 0.766438901424408,
      "learning_rate": 5.528697887406782e-05,
      "loss": 0.7031,
      "step": 13952
    },
    {
      "epoch": 2.5930124512172457,
      "grad_norm": 0.6652514934539795,
      "learning_rate": 5.5273911750565045e-05,
      "loss": 0.6129,
      "step": 13953
    },
    {
      "epoch": 2.593198290280617,
      "grad_norm": 0.8506471514701843,
      "learning_rate": 5.526084558166387e-05,
      "loss": 0.738,
      "step": 13954
    },
    {
      "epoch": 2.5933841293439883,
      "grad_norm": 0.9001114964485168,
      "learning_rate": 5.524778036764313e-05,
      "loss": 0.7862,
      "step": 13955
    },
    {
      "epoch": 2.593569968407359,
      "grad_norm": 0.7212256193161011,
      "learning_rate": 5.523471610878172e-05,
      "loss": 0.8856,
      "step": 13956
    },
    {
      "epoch": 2.59375580747073,
      "grad_norm": 0.7791721224784851,
      "learning_rate": 5.522165280535853e-05,
      "loss": 0.8765,
      "step": 13957
    },
    {
      "epoch": 2.5939416465341014,
      "grad_norm": 0.7826201319694519,
      "learning_rate": 5.52085904576523e-05,
      "loss": 1.0375,
      "step": 13958
    },
    {
      "epoch": 2.5941274855974727,
      "grad_norm": 0.8129811882972717,
      "learning_rate": 5.519552906594181e-05,
      "loss": 1.0598,
      "step": 13959
    },
    {
      "epoch": 2.5943133246608436,
      "grad_norm": 0.9655573964118958,
      "learning_rate": 5.51824686305059e-05,
      "loss": 1.03,
      "step": 13960
    },
    {
      "epoch": 2.594499163724215,
      "grad_norm": 0.7806291580200195,
      "learning_rate": 5.5169409151623294e-05,
      "loss": 1.0793,
      "step": 13961
    },
    {
      "epoch": 2.594685002787586,
      "grad_norm": 0.7537856101989746,
      "learning_rate": 5.515635062957272e-05,
      "loss": 0.7853,
      "step": 13962
    },
    {
      "epoch": 2.594870841850957,
      "grad_norm": 0.8779000639915466,
      "learning_rate": 5.514329306463284e-05,
      "loss": 0.8949,
      "step": 13963
    },
    {
      "epoch": 2.595056680914328,
      "grad_norm": 0.9200186133384705,
      "learning_rate": 5.513023645708242e-05,
      "loss": 0.7334,
      "step": 13964
    },
    {
      "epoch": 2.5952425199776994,
      "grad_norm": 0.9138427376747131,
      "learning_rate": 5.511718080720012e-05,
      "loss": 0.9246,
      "step": 13965
    },
    {
      "epoch": 2.5954283590410703,
      "grad_norm": 0.9949511885643005,
      "learning_rate": 5.510412611526452e-05,
      "loss": 1.0521,
      "step": 13966
    },
    {
      "epoch": 2.5956141981044416,
      "grad_norm": 0.7660025358200073,
      "learning_rate": 5.5091072381554355e-05,
      "loss": 1.0903,
      "step": 13967
    },
    {
      "epoch": 2.5958000371678125,
      "grad_norm": 1.9562807083129883,
      "learning_rate": 5.507801960634815e-05,
      "loss": 1.4339,
      "step": 13968
    },
    {
      "epoch": 2.595985876231184,
      "grad_norm": 0.8114579319953918,
      "learning_rate": 5.506496778992456e-05,
      "loss": 0.9345,
      "step": 13969
    },
    {
      "epoch": 2.596171715294555,
      "grad_norm": 0.9337628483772278,
      "learning_rate": 5.505191693256211e-05,
      "loss": 0.731,
      "step": 13970
    },
    {
      "epoch": 2.596357554357926,
      "grad_norm": 0.9045740962028503,
      "learning_rate": 5.503886703453933e-05,
      "loss": 0.7923,
      "step": 13971
    },
    {
      "epoch": 2.596543393421297,
      "grad_norm": 0.7792370319366455,
      "learning_rate": 5.502581809613482e-05,
      "loss": 0.6075,
      "step": 13972
    },
    {
      "epoch": 2.5967292324846682,
      "grad_norm": 0.8924160599708557,
      "learning_rate": 5.5012770117627046e-05,
      "loss": 0.9055,
      "step": 13973
    },
    {
      "epoch": 2.5969150715480396,
      "grad_norm": 0.7485678791999817,
      "learning_rate": 5.499972309929448e-05,
      "loss": 0.9435,
      "step": 13974
    },
    {
      "epoch": 2.5971009106114105,
      "grad_norm": 0.9983780384063721,
      "learning_rate": 5.498667704141556e-05,
      "loss": 0.7541,
      "step": 13975
    },
    {
      "epoch": 2.597286749674782,
      "grad_norm": 0.8822980523109436,
      "learning_rate": 5.497363194426882e-05,
      "loss": 0.9006,
      "step": 13976
    },
    {
      "epoch": 2.5974725887381527,
      "grad_norm": 1.0196903944015503,
      "learning_rate": 5.49605878081326e-05,
      "loss": 1.0334,
      "step": 13977
    },
    {
      "epoch": 2.597658427801524,
      "grad_norm": 0.7188934683799744,
      "learning_rate": 5.494754463328538e-05,
      "loss": 0.8842,
      "step": 13978
    },
    {
      "epoch": 2.597844266864895,
      "grad_norm": 0.9956566095352173,
      "learning_rate": 5.493450242000546e-05,
      "loss": 0.7571,
      "step": 13979
    },
    {
      "epoch": 2.598030105928266,
      "grad_norm": 0.7895113229751587,
      "learning_rate": 5.49214611685713e-05,
      "loss": 0.7521,
      "step": 13980
    },
    {
      "epoch": 2.598215944991637,
      "grad_norm": 0.8534258008003235,
      "learning_rate": 5.490842087926119e-05,
      "loss": 0.875,
      "step": 13981
    },
    {
      "epoch": 2.5984017840550084,
      "grad_norm": 0.8755273222923279,
      "learning_rate": 5.4895381552353455e-05,
      "loss": 0.8375,
      "step": 13982
    },
    {
      "epoch": 2.5985876231183793,
      "grad_norm": 0.96613609790802,
      "learning_rate": 5.488234318812636e-05,
      "loss": 0.972,
      "step": 13983
    },
    {
      "epoch": 2.5987734621817506,
      "grad_norm": 0.8058257102966309,
      "learning_rate": 5.486930578685826e-05,
      "loss": 1.0076,
      "step": 13984
    },
    {
      "epoch": 2.598959301245122,
      "grad_norm": 0.7786903977394104,
      "learning_rate": 5.485626934882739e-05,
      "loss": 0.7411,
      "step": 13985
    },
    {
      "epoch": 2.599145140308493,
      "grad_norm": 1.0404150485992432,
      "learning_rate": 5.484323387431197e-05,
      "loss": 1.0283,
      "step": 13986
    },
    {
      "epoch": 2.5993309793718637,
      "grad_norm": 0.7860246896743774,
      "learning_rate": 5.483019936359019e-05,
      "loss": 0.8234,
      "step": 13987
    },
    {
      "epoch": 2.599516818435235,
      "grad_norm": 0.9711631536483765,
      "learning_rate": 5.48171658169403e-05,
      "loss": 0.9653,
      "step": 13988
    },
    {
      "epoch": 2.5997026574986064,
      "grad_norm": 1.0118674039840698,
      "learning_rate": 5.480413323464051e-05,
      "loss": 1.0696,
      "step": 13989
    },
    {
      "epoch": 2.5998884965619773,
      "grad_norm": 0.7147940397262573,
      "learning_rate": 5.479110161696893e-05,
      "loss": 0.7738,
      "step": 13990
    },
    {
      "epoch": 2.600074335625348,
      "grad_norm": 0.8029753565788269,
      "learning_rate": 5.4778070964203667e-05,
      "loss": 0.9146,
      "step": 13991
    },
    {
      "epoch": 2.6002601746887195,
      "grad_norm": 0.7796580195426941,
      "learning_rate": 5.476504127662292e-05,
      "loss": 0.9461,
      "step": 13992
    },
    {
      "epoch": 2.600446013752091,
      "grad_norm": 0.7540099024772644,
      "learning_rate": 5.4752012554504725e-05,
      "loss": 0.9506,
      "step": 13993
    },
    {
      "epoch": 2.6006318528154617,
      "grad_norm": 1.0693564414978027,
      "learning_rate": 5.473898479812719e-05,
      "loss": 0.7965,
      "step": 13994
    },
    {
      "epoch": 2.600817691878833,
      "grad_norm": 0.730949878692627,
      "learning_rate": 5.4725958007768304e-05,
      "loss": 0.9624,
      "step": 13995
    },
    {
      "epoch": 2.601003530942204,
      "grad_norm": 0.897834062576294,
      "learning_rate": 5.471293218370619e-05,
      "loss": 0.8675,
      "step": 13996
    },
    {
      "epoch": 2.6011893700055753,
      "grad_norm": 0.7791353464126587,
      "learning_rate": 5.4699907326218836e-05,
      "loss": 1.0761,
      "step": 13997
    },
    {
      "epoch": 2.601375209068946,
      "grad_norm": 0.793819010257721,
      "learning_rate": 5.4686883435584166e-05,
      "loss": 1.0897,
      "step": 13998
    },
    {
      "epoch": 2.6015610481323175,
      "grad_norm": 0.8791100978851318,
      "learning_rate": 5.467386051208022e-05,
      "loss": 1.1172,
      "step": 13999
    },
    {
      "epoch": 2.601746887195689,
      "grad_norm": 0.7686331868171692,
      "learning_rate": 5.4660838555984974e-05,
      "loss": 0.8535,
      "step": 14000
    },
    {
      "epoch": 2.6019327262590597,
      "grad_norm": 0.8077011108398438,
      "learning_rate": 5.464781756757633e-05,
      "loss": 0.9649,
      "step": 14001
    },
    {
      "epoch": 2.6021185653224306,
      "grad_norm": 0.8720762729644775,
      "learning_rate": 5.463479754713219e-05,
      "loss": 0.9052,
      "step": 14002
    },
    {
      "epoch": 2.602304404385802,
      "grad_norm": 0.9532802104949951,
      "learning_rate": 5.4621778494930397e-05,
      "loss": 0.9079,
      "step": 14003
    },
    {
      "epoch": 2.6024902434491732,
      "grad_norm": 0.8844078183174133,
      "learning_rate": 5.460876041124892e-05,
      "loss": 0.9669,
      "step": 14004
    },
    {
      "epoch": 2.602676082512544,
      "grad_norm": 0.8178666234016418,
      "learning_rate": 5.459574329636554e-05,
      "loss": 0.8515,
      "step": 14005
    },
    {
      "epoch": 2.602861921575915,
      "grad_norm": 0.7567655444145203,
      "learning_rate": 5.458272715055812e-05,
      "loss": 0.9037,
      "step": 14006
    },
    {
      "epoch": 2.6030477606392863,
      "grad_norm": 1.0607315301895142,
      "learning_rate": 5.456971197410441e-05,
      "loss": 0.6469,
      "step": 14007
    },
    {
      "epoch": 2.6032335997026577,
      "grad_norm": 0.8017802238464355,
      "learning_rate": 5.4556697767282224e-05,
      "loss": 0.9601,
      "step": 14008
    },
    {
      "epoch": 2.6034194387660285,
      "grad_norm": 1.093765377998352,
      "learning_rate": 5.4543684530369396e-05,
      "loss": 0.7734,
      "step": 14009
    },
    {
      "epoch": 2.6036052778294,
      "grad_norm": 0.7641136050224304,
      "learning_rate": 5.453067226364361e-05,
      "loss": 0.9624,
      "step": 14010
    },
    {
      "epoch": 2.6037911168927708,
      "grad_norm": 0.8289857506752014,
      "learning_rate": 5.451766096738256e-05,
      "loss": 0.958,
      "step": 14011
    },
    {
      "epoch": 2.603976955956142,
      "grad_norm": 0.7701115012168884,
      "learning_rate": 5.450465064186401e-05,
      "loss": 0.7542,
      "step": 14012
    },
    {
      "epoch": 2.604162795019513,
      "grad_norm": 0.8174500465393066,
      "learning_rate": 5.4491641287365635e-05,
      "loss": 1.0229,
      "step": 14013
    },
    {
      "epoch": 2.6043486340828843,
      "grad_norm": 0.8088310956954956,
      "learning_rate": 5.4478632904165074e-05,
      "loss": 0.9266,
      "step": 14014
    },
    {
      "epoch": 2.604534473146255,
      "grad_norm": 0.7646427750587463,
      "learning_rate": 5.446562549253993e-05,
      "loss": 1.0236,
      "step": 14015
    },
    {
      "epoch": 2.6047203122096265,
      "grad_norm": 0.7802772521972656,
      "learning_rate": 5.445261905276793e-05,
      "loss": 0.9357,
      "step": 14016
    },
    {
      "epoch": 2.6049061512729974,
      "grad_norm": 0.8911521434783936,
      "learning_rate": 5.4439613585126594e-05,
      "loss": 0.8832,
      "step": 14017
    },
    {
      "epoch": 2.6050919903363687,
      "grad_norm": 0.8874276876449585,
      "learning_rate": 5.442660908989352e-05,
      "loss": 0.8062,
      "step": 14018
    },
    {
      "epoch": 2.60527782939974,
      "grad_norm": 0.7851524353027344,
      "learning_rate": 5.44136055673462e-05,
      "loss": 0.9741,
      "step": 14019
    },
    {
      "epoch": 2.605463668463111,
      "grad_norm": 0.8898999691009521,
      "learning_rate": 5.440060301776232e-05,
      "loss": 0.9239,
      "step": 14020
    },
    {
      "epoch": 2.605649507526482,
      "grad_norm": 0.8409792184829712,
      "learning_rate": 5.438760144141931e-05,
      "loss": 0.901,
      "step": 14021
    },
    {
      "epoch": 2.605835346589853,
      "grad_norm": 0.8695258498191833,
      "learning_rate": 5.437460083859467e-05,
      "loss": 0.9999,
      "step": 14022
    },
    {
      "epoch": 2.6060211856532245,
      "grad_norm": 0.7994899749755859,
      "learning_rate": 5.436160120956585e-05,
      "loss": 0.8466,
      "step": 14023
    },
    {
      "epoch": 2.6062070247165954,
      "grad_norm": 0.84480881690979,
      "learning_rate": 5.434860255461036e-05,
      "loss": 0.8116,
      "step": 14024
    },
    {
      "epoch": 2.6063928637799667,
      "grad_norm": 0.6807081699371338,
      "learning_rate": 5.433560487400562e-05,
      "loss": 0.784,
      "step": 14025
    },
    {
      "epoch": 2.6065787028433376,
      "grad_norm": 0.81355881690979,
      "learning_rate": 5.432260816802903e-05,
      "loss": 1.1106,
      "step": 14026
    },
    {
      "epoch": 2.606764541906709,
      "grad_norm": 1.743600845336914,
      "learning_rate": 5.4309612436957937e-05,
      "loss": 1.1663,
      "step": 14027
    },
    {
      "epoch": 2.60695038097008,
      "grad_norm": 1.0914791822433472,
      "learning_rate": 5.429661768106981e-05,
      "loss": 1.3701,
      "step": 14028
    },
    {
      "epoch": 2.607136220033451,
      "grad_norm": 0.771381139755249,
      "learning_rate": 5.42836239006419e-05,
      "loss": 1.0859,
      "step": 14029
    },
    {
      "epoch": 2.607322059096822,
      "grad_norm": 0.7847346067428589,
      "learning_rate": 5.427063109595162e-05,
      "loss": 0.8319,
      "step": 14030
    },
    {
      "epoch": 2.6075078981601933,
      "grad_norm": 1.1157923936843872,
      "learning_rate": 5.4257639267276205e-05,
      "loss": 1.3161,
      "step": 14031
    },
    {
      "epoch": 2.6076937372235642,
      "grad_norm": 0.726537823677063,
      "learning_rate": 5.424464841489303e-05,
      "loss": 1.032,
      "step": 14032
    },
    {
      "epoch": 2.6078795762869356,
      "grad_norm": 0.7729559540748596,
      "learning_rate": 5.423165853907931e-05,
      "loss": 0.8861,
      "step": 14033
    },
    {
      "epoch": 2.608065415350307,
      "grad_norm": 0.8606934547424316,
      "learning_rate": 5.421866964011231e-05,
      "loss": 0.9862,
      "step": 14034
    },
    {
      "epoch": 2.6082512544136778,
      "grad_norm": 0.6861522793769836,
      "learning_rate": 5.4205681718269184e-05,
      "loss": 0.6266,
      "step": 14035
    },
    {
      "epoch": 2.6084370934770487,
      "grad_norm": 0.7935603260993958,
      "learning_rate": 5.419269477382725e-05,
      "loss": 0.8256,
      "step": 14036
    },
    {
      "epoch": 2.60862293254042,
      "grad_norm": 0.8142755031585693,
      "learning_rate": 5.4179708807063625e-05,
      "loss": 1.0308,
      "step": 14037
    },
    {
      "epoch": 2.6088087716037913,
      "grad_norm": 1.0010840892791748,
      "learning_rate": 5.416672381825549e-05,
      "loss": 0.9534,
      "step": 14038
    },
    {
      "epoch": 2.608994610667162,
      "grad_norm": 0.8254314661026001,
      "learning_rate": 5.415373980767995e-05,
      "loss": 1.0915,
      "step": 14039
    },
    {
      "epoch": 2.609180449730533,
      "grad_norm": 0.799942672252655,
      "learning_rate": 5.4140756775614146e-05,
      "loss": 0.7815,
      "step": 14040
    },
    {
      "epoch": 2.6093662887939044,
      "grad_norm": 0.8121205568313599,
      "learning_rate": 5.412777472233523e-05,
      "loss": 1.0408,
      "step": 14041
    },
    {
      "epoch": 2.6095521278572757,
      "grad_norm": 0.8696049451828003,
      "learning_rate": 5.4114793648120256e-05,
      "loss": 1.0698,
      "step": 14042
    },
    {
      "epoch": 2.6097379669206466,
      "grad_norm": 0.98824143409729,
      "learning_rate": 5.410181355324622e-05,
      "loss": 1.2393,
      "step": 14043
    },
    {
      "epoch": 2.609923805984018,
      "grad_norm": 0.8695021271705627,
      "learning_rate": 5.408883443799026e-05,
      "loss": 1.0092,
      "step": 14044
    },
    {
      "epoch": 2.610109645047389,
      "grad_norm": 0.8957984447479248,
      "learning_rate": 5.407585630262932e-05,
      "loss": 0.9223,
      "step": 14045
    },
    {
      "epoch": 2.61029548411076,
      "grad_norm": 0.7589720487594604,
      "learning_rate": 5.406287914744044e-05,
      "loss": 1.1484,
      "step": 14046
    },
    {
      "epoch": 2.610481323174131,
      "grad_norm": 0.8064348697662354,
      "learning_rate": 5.4049902972700515e-05,
      "loss": 1.0092,
      "step": 14047
    },
    {
      "epoch": 2.6106671622375024,
      "grad_norm": 0.6093027591705322,
      "learning_rate": 5.4036927778686616e-05,
      "loss": 0.575,
      "step": 14048
    },
    {
      "epoch": 2.6108530013008733,
      "grad_norm": 0.7983278036117554,
      "learning_rate": 5.4023953565675614e-05,
      "loss": 0.9588,
      "step": 14049
    },
    {
      "epoch": 2.6110388403642446,
      "grad_norm": 0.8939300179481506,
      "learning_rate": 5.4010980333944386e-05,
      "loss": 0.8152,
      "step": 14050
    },
    {
      "epoch": 2.6112246794276155,
      "grad_norm": 1.0524145364761353,
      "learning_rate": 5.3998008083769846e-05,
      "loss": 0.827,
      "step": 14051
    },
    {
      "epoch": 2.611410518490987,
      "grad_norm": 0.8120452165603638,
      "learning_rate": 5.398503681542893e-05,
      "loss": 1.0297,
      "step": 14052
    },
    {
      "epoch": 2.611596357554358,
      "grad_norm": 0.7002134323120117,
      "learning_rate": 5.397206652919844e-05,
      "loss": 0.9747,
      "step": 14053
    },
    {
      "epoch": 2.611782196617729,
      "grad_norm": 1.1011672019958496,
      "learning_rate": 5.395909722535522e-05,
      "loss": 1.1254,
      "step": 14054
    },
    {
      "epoch": 2.6119680356811,
      "grad_norm": 0.8079476356506348,
      "learning_rate": 5.3946128904176e-05,
      "loss": 0.8523,
      "step": 14055
    },
    {
      "epoch": 2.6121538747444712,
      "grad_norm": 0.7758453488349915,
      "learning_rate": 5.3933161565937665e-05,
      "loss": 1.0984,
      "step": 14056
    },
    {
      "epoch": 2.6123397138078426,
      "grad_norm": 0.8518459796905518,
      "learning_rate": 5.392019521091695e-05,
      "loss": 0.8602,
      "step": 14057
    },
    {
      "epoch": 2.6125255528712135,
      "grad_norm": 0.7195460200309753,
      "learning_rate": 5.390722983939058e-05,
      "loss": 0.6718,
      "step": 14058
    },
    {
      "epoch": 2.612711391934585,
      "grad_norm": 0.820122480392456,
      "learning_rate": 5.389426545163527e-05,
      "loss": 0.9311,
      "step": 14059
    },
    {
      "epoch": 2.6128972309979557,
      "grad_norm": 0.8300378918647766,
      "learning_rate": 5.3881302047927774e-05,
      "loss": 0.7584,
      "step": 14060
    },
    {
      "epoch": 2.613083070061327,
      "grad_norm": 0.8203210830688477,
      "learning_rate": 5.3868339628544695e-05,
      "loss": 1.0066,
      "step": 14061
    },
    {
      "epoch": 2.613268909124698,
      "grad_norm": 1.0901485681533813,
      "learning_rate": 5.38553781937628e-05,
      "loss": 1.0315,
      "step": 14062
    },
    {
      "epoch": 2.613454748188069,
      "grad_norm": 0.7519093155860901,
      "learning_rate": 5.384241774385862e-05,
      "loss": 0.7722,
      "step": 14063
    },
    {
      "epoch": 2.61364058725144,
      "grad_norm": 0.833781898021698,
      "learning_rate": 5.382945827910886e-05,
      "loss": 1.022,
      "step": 14064
    },
    {
      "epoch": 2.6138264263148114,
      "grad_norm": 0.7555166482925415,
      "learning_rate": 5.381649979979009e-05,
      "loss": 0.8817,
      "step": 14065
    },
    {
      "epoch": 2.6140122653781823,
      "grad_norm": 0.7260909676551819,
      "learning_rate": 5.3803542306178865e-05,
      "loss": 0.8799,
      "step": 14066
    },
    {
      "epoch": 2.6141981044415536,
      "grad_norm": 0.7376378178596497,
      "learning_rate": 5.379058579855172e-05,
      "loss": 0.7841,
      "step": 14067
    },
    {
      "epoch": 2.614383943504925,
      "grad_norm": 0.8301506638526917,
      "learning_rate": 5.3777630277185264e-05,
      "loss": 0.846,
      "step": 14068
    },
    {
      "epoch": 2.614569782568296,
      "grad_norm": 0.9465357065200806,
      "learning_rate": 5.3764675742355974e-05,
      "loss": 0.9891,
      "step": 14069
    },
    {
      "epoch": 2.6147556216316667,
      "grad_norm": 0.8111327886581421,
      "learning_rate": 5.375172219434028e-05,
      "loss": 0.945,
      "step": 14070
    },
    {
      "epoch": 2.614941460695038,
      "grad_norm": 0.7847363352775574,
      "learning_rate": 5.373876963341473e-05,
      "loss": 0.8943,
      "step": 14071
    },
    {
      "epoch": 2.6151272997584094,
      "grad_norm": 0.970258891582489,
      "learning_rate": 5.3725818059855796e-05,
      "loss": 1.0219,
      "step": 14072
    },
    {
      "epoch": 2.6153131388217803,
      "grad_norm": 0.8293201327323914,
      "learning_rate": 5.3712867473939845e-05,
      "loss": 1.0821,
      "step": 14073
    },
    {
      "epoch": 2.6154989778851516,
      "grad_norm": 0.8214329481124878,
      "learning_rate": 5.369991787594332e-05,
      "loss": 0.6193,
      "step": 14074
    },
    {
      "epoch": 2.6156848169485225,
      "grad_norm": 1.016649603843689,
      "learning_rate": 5.368696926614255e-05,
      "loss": 1.1271,
      "step": 14075
    },
    {
      "epoch": 2.615870656011894,
      "grad_norm": 0.8211673498153687,
      "learning_rate": 5.367402164481399e-05,
      "loss": 0.9893,
      "step": 14076
    },
    {
      "epoch": 2.6160564950752647,
      "grad_norm": 0.8138982057571411,
      "learning_rate": 5.366107501223393e-05,
      "loss": 0.9162,
      "step": 14077
    },
    {
      "epoch": 2.616242334138636,
      "grad_norm": 0.8277393579483032,
      "learning_rate": 5.364812936867871e-05,
      "loss": 1.2818,
      "step": 14078
    },
    {
      "epoch": 2.616428173202007,
      "grad_norm": 0.9266590476036072,
      "learning_rate": 5.3635184714424594e-05,
      "loss": 0.8863,
      "step": 14079
    },
    {
      "epoch": 2.6166140122653783,
      "grad_norm": 0.8038361072540283,
      "learning_rate": 5.3622241049747914e-05,
      "loss": 1.0169,
      "step": 14080
    },
    {
      "epoch": 2.616799851328749,
      "grad_norm": 0.8010311722755432,
      "learning_rate": 5.360929837492492e-05,
      "loss": 0.7726,
      "step": 14081
    },
    {
      "epoch": 2.6169856903921205,
      "grad_norm": 0.8894407153129578,
      "learning_rate": 5.359635669023181e-05,
      "loss": 0.9489,
      "step": 14082
    },
    {
      "epoch": 2.617171529455492,
      "grad_norm": 0.7788730263710022,
      "learning_rate": 5.358341599594483e-05,
      "loss": 0.866,
      "step": 14083
    },
    {
      "epoch": 2.6173573685188627,
      "grad_norm": 0.9567217826843262,
      "learning_rate": 5.357047629234022e-05,
      "loss": 0.5063,
      "step": 14084
    },
    {
      "epoch": 2.6175432075822336,
      "grad_norm": 0.8853626847267151,
      "learning_rate": 5.3557537579694114e-05,
      "loss": 0.9121,
      "step": 14085
    },
    {
      "epoch": 2.617729046645605,
      "grad_norm": 0.8689543604850769,
      "learning_rate": 5.354459985828268e-05,
      "loss": 0.9747,
      "step": 14086
    },
    {
      "epoch": 2.6179148857089762,
      "grad_norm": 1.1644787788391113,
      "learning_rate": 5.353166312838199e-05,
      "loss": 1.1248,
      "step": 14087
    },
    {
      "epoch": 2.618100724772347,
      "grad_norm": 0.8772037625312805,
      "learning_rate": 5.351872739026826e-05,
      "loss": 0.7357,
      "step": 14088
    },
    {
      "epoch": 2.618286563835718,
      "grad_norm": 0.8058109283447266,
      "learning_rate": 5.3505792644217525e-05,
      "loss": 0.8819,
      "step": 14089
    },
    {
      "epoch": 2.6184724028990893,
      "grad_norm": 0.838029146194458,
      "learning_rate": 5.349285889050582e-05,
      "loss": 0.8949,
      "step": 14090
    },
    {
      "epoch": 2.6186582419624607,
      "grad_norm": 0.8061281442642212,
      "learning_rate": 5.347992612940926e-05,
      "loss": 0.9215,
      "step": 14091
    },
    {
      "epoch": 2.6188440810258315,
      "grad_norm": 0.8494970202445984,
      "learning_rate": 5.3466994361203816e-05,
      "loss": 0.9996,
      "step": 14092
    },
    {
      "epoch": 2.619029920089203,
      "grad_norm": 0.7258545756340027,
      "learning_rate": 5.345406358616556e-05,
      "loss": 0.9192,
      "step": 14093
    },
    {
      "epoch": 2.6192157591525738,
      "grad_norm": 0.7305350303649902,
      "learning_rate": 5.344113380457044e-05,
      "loss": 0.7586,
      "step": 14094
    },
    {
      "epoch": 2.619401598215945,
      "grad_norm": 0.6838800311088562,
      "learning_rate": 5.342820501669439e-05,
      "loss": 0.8023,
      "step": 14095
    },
    {
      "epoch": 2.619587437279316,
      "grad_norm": 0.9066343903541565,
      "learning_rate": 5.3415277222813406e-05,
      "loss": 1.204,
      "step": 14096
    },
    {
      "epoch": 2.6197732763426873,
      "grad_norm": 0.9841775298118591,
      "learning_rate": 5.340235042320341e-05,
      "loss": 1.0951,
      "step": 14097
    },
    {
      "epoch": 2.619959115406058,
      "grad_norm": 0.8012330532073975,
      "learning_rate": 5.338942461814025e-05,
      "loss": 0.8781,
      "step": 14098
    },
    {
      "epoch": 2.6201449544694295,
      "grad_norm": 0.8908490538597107,
      "learning_rate": 5.337649980789982e-05,
      "loss": 0.6783,
      "step": 14099
    },
    {
      "epoch": 2.6203307935328004,
      "grad_norm": 0.8828836679458618,
      "learning_rate": 5.336357599275802e-05,
      "loss": 1.0055,
      "step": 14100
    },
    {
      "epoch": 2.6205166325961717,
      "grad_norm": 0.7191521525382996,
      "learning_rate": 5.335065317299066e-05,
      "loss": 0.6156,
      "step": 14101
    },
    {
      "epoch": 2.620702471659543,
      "grad_norm": 0.6662703156471252,
      "learning_rate": 5.333773134887352e-05,
      "loss": 0.6017,
      "step": 14102
    },
    {
      "epoch": 2.620888310722914,
      "grad_norm": 0.9348257184028625,
      "learning_rate": 5.332481052068243e-05,
      "loss": 1.1609,
      "step": 14103
    },
    {
      "epoch": 2.621074149786285,
      "grad_norm": 0.7789109945297241,
      "learning_rate": 5.33118906886932e-05,
      "loss": 0.7852,
      "step": 14104
    },
    {
      "epoch": 2.621259988849656,
      "grad_norm": 0.8765986561775208,
      "learning_rate": 5.329897185318153e-05,
      "loss": 0.9478,
      "step": 14105
    },
    {
      "epoch": 2.6214458279130275,
      "grad_norm": 0.8632599711418152,
      "learning_rate": 5.328605401442318e-05,
      "loss": 0.9694,
      "step": 14106
    },
    {
      "epoch": 2.6216316669763984,
      "grad_norm": 0.7510929703712463,
      "learning_rate": 5.32731371726938e-05,
      "loss": 0.7101,
      "step": 14107
    },
    {
      "epoch": 2.6218175060397697,
      "grad_norm": 0.9050139784812927,
      "learning_rate": 5.3260221328269156e-05,
      "loss": 0.8079,
      "step": 14108
    },
    {
      "epoch": 2.6220033451031406,
      "grad_norm": 0.7674015164375305,
      "learning_rate": 5.324730648142489e-05,
      "loss": 1.0027,
      "step": 14109
    },
    {
      "epoch": 2.622189184166512,
      "grad_norm": 0.8893373012542725,
      "learning_rate": 5.323439263243657e-05,
      "loss": 0.8087,
      "step": 14110
    },
    {
      "epoch": 2.622375023229883,
      "grad_norm": 0.7976099252700806,
      "learning_rate": 5.3221479781579955e-05,
      "loss": 0.7984,
      "step": 14111
    },
    {
      "epoch": 2.622560862293254,
      "grad_norm": 0.6816142797470093,
      "learning_rate": 5.320856792913056e-05,
      "loss": 0.8604,
      "step": 14112
    },
    {
      "epoch": 2.622746701356625,
      "grad_norm": 0.9564425945281982,
      "learning_rate": 5.3195657075363956e-05,
      "loss": 0.9432,
      "step": 14113
    },
    {
      "epoch": 2.6229325404199964,
      "grad_norm": 0.8268784284591675,
      "learning_rate": 5.318274722055576e-05,
      "loss": 0.8341,
      "step": 14114
    },
    {
      "epoch": 2.6231183794833672,
      "grad_norm": 0.9182666540145874,
      "learning_rate": 5.316983836498146e-05,
      "loss": 1.0505,
      "step": 14115
    },
    {
      "epoch": 2.6233042185467386,
      "grad_norm": 0.7662116885185242,
      "learning_rate": 5.3156930508916634e-05,
      "loss": 0.9504,
      "step": 14116
    },
    {
      "epoch": 2.62349005761011,
      "grad_norm": 0.7458078861236572,
      "learning_rate": 5.314402365263673e-05,
      "loss": 0.8565,
      "step": 14117
    },
    {
      "epoch": 2.623675896673481,
      "grad_norm": 0.7623659372329712,
      "learning_rate": 5.313111779641724e-05,
      "loss": 1.1978,
      "step": 14118
    },
    {
      "epoch": 2.6238617357368517,
      "grad_norm": 0.7925945520401001,
      "learning_rate": 5.311821294053356e-05,
      "loss": 0.7834,
      "step": 14119
    },
    {
      "epoch": 2.624047574800223,
      "grad_norm": 0.9102346301078796,
      "learning_rate": 5.3105309085261224e-05,
      "loss": 0.9621,
      "step": 14120
    },
    {
      "epoch": 2.6242334138635943,
      "grad_norm": 0.992512047290802,
      "learning_rate": 5.309240623087559e-05,
      "loss": 0.8519,
      "step": 14121
    },
    {
      "epoch": 2.624419252926965,
      "grad_norm": 0.777908205986023,
      "learning_rate": 5.3079504377652e-05,
      "loss": 0.7648,
      "step": 14122
    },
    {
      "epoch": 2.624605091990336,
      "grad_norm": 0.7941464185714722,
      "learning_rate": 5.3066603525865924e-05,
      "loss": 0.9803,
      "step": 14123
    },
    {
      "epoch": 2.6247909310537074,
      "grad_norm": 0.8003142476081848,
      "learning_rate": 5.30537036757926e-05,
      "loss": 0.7327,
      "step": 14124
    },
    {
      "epoch": 2.6249767701170788,
      "grad_norm": 1.1307048797607422,
      "learning_rate": 5.3040804827707457e-05,
      "loss": 0.8996,
      "step": 14125
    },
    {
      "epoch": 2.6251626091804496,
      "grad_norm": 0.903079092502594,
      "learning_rate": 5.3027906981885747e-05,
      "loss": 0.7964,
      "step": 14126
    },
    {
      "epoch": 2.625348448243821,
      "grad_norm": 0.8964478969573975,
      "learning_rate": 5.301501013860272e-05,
      "loss": 0.9799,
      "step": 14127
    },
    {
      "epoch": 2.625534287307192,
      "grad_norm": 1.1879934072494507,
      "learning_rate": 5.300211429813371e-05,
      "loss": 1.138,
      "step": 14128
    },
    {
      "epoch": 2.625720126370563,
      "grad_norm": 0.8264209628105164,
      "learning_rate": 5.29892194607539e-05,
      "loss": 0.9496,
      "step": 14129
    },
    {
      "epoch": 2.625905965433934,
      "grad_norm": 0.8963238000869751,
      "learning_rate": 5.297632562673851e-05,
      "loss": 0.8034,
      "step": 14130
    },
    {
      "epoch": 2.6260918044973054,
      "grad_norm": 1.0959327220916748,
      "learning_rate": 5.296343279636279e-05,
      "loss": 0.832,
      "step": 14131
    },
    {
      "epoch": 2.6262776435606767,
      "grad_norm": 1.2209457159042358,
      "learning_rate": 5.2950540969901876e-05,
      "loss": 1.1996,
      "step": 14132
    },
    {
      "epoch": 2.6264634826240476,
      "grad_norm": 1.5622323751449585,
      "learning_rate": 5.293765014763093e-05,
      "loss": 1.3739,
      "step": 14133
    },
    {
      "epoch": 2.6266493216874185,
      "grad_norm": 0.7270522117614746,
      "learning_rate": 5.292476032982503e-05,
      "loss": 0.7721,
      "step": 14134
    },
    {
      "epoch": 2.62683516075079,
      "grad_norm": 0.879499614238739,
      "learning_rate": 5.2911871516759336e-05,
      "loss": 0.9366,
      "step": 14135
    },
    {
      "epoch": 2.627020999814161,
      "grad_norm": 0.8861373662948608,
      "learning_rate": 5.289898370870898e-05,
      "loss": 0.892,
      "step": 14136
    },
    {
      "epoch": 2.627206838877532,
      "grad_norm": 0.7887271046638489,
      "learning_rate": 5.288609690594899e-05,
      "loss": 0.9441,
      "step": 14137
    },
    {
      "epoch": 2.627392677940903,
      "grad_norm": 1.050877332687378,
      "learning_rate": 5.287321110875442e-05,
      "loss": 0.8373,
      "step": 14138
    },
    {
      "epoch": 2.6275785170042743,
      "grad_norm": 0.7968853712081909,
      "learning_rate": 5.286032631740023e-05,
      "loss": 0.7597,
      "step": 14139
    },
    {
      "epoch": 2.6277643560676456,
      "grad_norm": 0.7845934629440308,
      "learning_rate": 5.284744253216154e-05,
      "loss": 0.9509,
      "step": 14140
    },
    {
      "epoch": 2.6279501951310165,
      "grad_norm": 0.7953255772590637,
      "learning_rate": 5.283455975331325e-05,
      "loss": 0.9759,
      "step": 14141
    },
    {
      "epoch": 2.628136034194388,
      "grad_norm": 0.7907785177230835,
      "learning_rate": 5.282167798113031e-05,
      "loss": 0.901,
      "step": 14142
    },
    {
      "epoch": 2.6283218732577587,
      "grad_norm": 0.825691282749176,
      "learning_rate": 5.280879721588774e-05,
      "loss": 0.9301,
      "step": 14143
    },
    {
      "epoch": 2.62850771232113,
      "grad_norm": 0.8010538816452026,
      "learning_rate": 5.2795917457860414e-05,
      "loss": 0.8767,
      "step": 14144
    },
    {
      "epoch": 2.628693551384501,
      "grad_norm": 0.7336350083351135,
      "learning_rate": 5.2783038707323174e-05,
      "loss": 0.759,
      "step": 14145
    },
    {
      "epoch": 2.628879390447872,
      "grad_norm": 0.8091769814491272,
      "learning_rate": 5.2770160964550985e-05,
      "loss": 0.9816,
      "step": 14146
    },
    {
      "epoch": 2.629065229511243,
      "grad_norm": 0.812383770942688,
      "learning_rate": 5.275728422981861e-05,
      "loss": 0.7816,
      "step": 14147
    },
    {
      "epoch": 2.6292510685746144,
      "grad_norm": 0.9698837399482727,
      "learning_rate": 5.274440850340099e-05,
      "loss": 1.1908,
      "step": 14148
    },
    {
      "epoch": 2.6294369076379853,
      "grad_norm": 0.7005797624588013,
      "learning_rate": 5.273153378557287e-05,
      "loss": 0.8051,
      "step": 14149
    },
    {
      "epoch": 2.6296227467013567,
      "grad_norm": 0.7625775933265686,
      "learning_rate": 5.2718660076609005e-05,
      "loss": 0.9715,
      "step": 14150
    },
    {
      "epoch": 2.629808585764728,
      "grad_norm": 0.7771723866462708,
      "learning_rate": 5.270578737678425e-05,
      "loss": 0.8616,
      "step": 14151
    },
    {
      "epoch": 2.629994424828099,
      "grad_norm": 0.8352897763252258,
      "learning_rate": 5.269291568637329e-05,
      "loss": 0.7741,
      "step": 14152
    },
    {
      "epoch": 2.6301802638914697,
      "grad_norm": 0.8637375235557556,
      "learning_rate": 5.268004500565087e-05,
      "loss": 1.0465,
      "step": 14153
    },
    {
      "epoch": 2.630366102954841,
      "grad_norm": 0.882972776889801,
      "learning_rate": 5.266717533489164e-05,
      "loss": 1.052,
      "step": 14154
    },
    {
      "epoch": 2.6305519420182124,
      "grad_norm": 0.8657705187797546,
      "learning_rate": 5.265430667437033e-05,
      "loss": 0.9644,
      "step": 14155
    },
    {
      "epoch": 2.6307377810815833,
      "grad_norm": 0.8730942606925964,
      "learning_rate": 5.264143902436164e-05,
      "loss": 1.0131,
      "step": 14156
    },
    {
      "epoch": 2.6309236201449546,
      "grad_norm": 0.7059460878372192,
      "learning_rate": 5.2628572385140165e-05,
      "loss": 0.711,
      "step": 14157
    },
    {
      "epoch": 2.6311094592083255,
      "grad_norm": 0.9356588125228882,
      "learning_rate": 5.261570675698051e-05,
      "loss": 0.6547,
      "step": 14158
    },
    {
      "epoch": 2.631295298271697,
      "grad_norm": 0.8111671209335327,
      "learning_rate": 5.260284214015724e-05,
      "loss": 0.7721,
      "step": 14159
    },
    {
      "epoch": 2.6314811373350677,
      "grad_norm": 0.9288527369499207,
      "learning_rate": 5.258997853494501e-05,
      "loss": 0.8618,
      "step": 14160
    },
    {
      "epoch": 2.631666976398439,
      "grad_norm": 0.9449626207351685,
      "learning_rate": 5.257711594161833e-05,
      "loss": 0.9077,
      "step": 14161
    },
    {
      "epoch": 2.63185281546181,
      "grad_norm": 0.9024152159690857,
      "learning_rate": 5.2564254360451694e-05,
      "loss": 1.0439,
      "step": 14162
    },
    {
      "epoch": 2.6320386545251813,
      "grad_norm": 0.9024345874786377,
      "learning_rate": 5.255139379171967e-05,
      "loss": 1.0034,
      "step": 14163
    },
    {
      "epoch": 2.632224493588552,
      "grad_norm": 1.0499695539474487,
      "learning_rate": 5.253853423569675e-05,
      "loss": 0.9327,
      "step": 14164
    },
    {
      "epoch": 2.6324103326519235,
      "grad_norm": 0.8915688991546631,
      "learning_rate": 5.252567569265734e-05,
      "loss": 0.8831,
      "step": 14165
    },
    {
      "epoch": 2.632596171715295,
      "grad_norm": 0.8023610711097717,
      "learning_rate": 5.251281816287589e-05,
      "loss": 0.8582,
      "step": 14166
    },
    {
      "epoch": 2.6327820107786657,
      "grad_norm": 0.7795034646987915,
      "learning_rate": 5.2499961646626824e-05,
      "loss": 1.0204,
      "step": 14167
    },
    {
      "epoch": 2.6329678498420366,
      "grad_norm": 0.7348303198814392,
      "learning_rate": 5.248710614418463e-05,
      "loss": 0.7561,
      "step": 14168
    },
    {
      "epoch": 2.633153688905408,
      "grad_norm": 0.7823712229728699,
      "learning_rate": 5.24742516558236e-05,
      "loss": 0.9611,
      "step": 14169
    },
    {
      "epoch": 2.6333395279687792,
      "grad_norm": 0.9677551984786987,
      "learning_rate": 5.246139818181809e-05,
      "loss": 1.1466,
      "step": 14170
    },
    {
      "epoch": 2.63352536703215,
      "grad_norm": 0.8288678526878357,
      "learning_rate": 5.2448545722442486e-05,
      "loss": 1.0228,
      "step": 14171
    },
    {
      "epoch": 2.633711206095521,
      "grad_norm": 0.8328431844711304,
      "learning_rate": 5.243569427797107e-05,
      "loss": 0.7301,
      "step": 14172
    },
    {
      "epoch": 2.6338970451588923,
      "grad_norm": 0.7218412756919861,
      "learning_rate": 5.2422843848678136e-05,
      "loss": 0.8425,
      "step": 14173
    },
    {
      "epoch": 2.6340828842222637,
      "grad_norm": 0.9164881110191345,
      "learning_rate": 5.2409994434837914e-05,
      "loss": 1.1753,
      "step": 14174
    },
    {
      "epoch": 2.6342687232856346,
      "grad_norm": 0.7694655060768127,
      "learning_rate": 5.239714603672474e-05,
      "loss": 0.7652,
      "step": 14175
    },
    {
      "epoch": 2.634454562349006,
      "grad_norm": 0.7973865866661072,
      "learning_rate": 5.238429865461275e-05,
      "loss": 1.0709,
      "step": 14176
    },
    {
      "epoch": 2.6346404014123768,
      "grad_norm": 1.0666766166687012,
      "learning_rate": 5.237145228877623e-05,
      "loss": 0.9486,
      "step": 14177
    },
    {
      "epoch": 2.634826240475748,
      "grad_norm": 0.8147100806236267,
      "learning_rate": 5.235860693948933e-05,
      "loss": 0.8283,
      "step": 14178
    },
    {
      "epoch": 2.635012079539119,
      "grad_norm": 0.7925849556922913,
      "learning_rate": 5.2345762607026174e-05,
      "loss": 0.8076,
      "step": 14179
    },
    {
      "epoch": 2.6351979186024903,
      "grad_norm": 0.7890560626983643,
      "learning_rate": 5.2332919291660976e-05,
      "loss": 0.8417,
      "step": 14180
    },
    {
      "epoch": 2.6353837576658616,
      "grad_norm": 0.7491629123687744,
      "learning_rate": 5.232007699366781e-05,
      "loss": 1.0209,
      "step": 14181
    },
    {
      "epoch": 2.6355695967292325,
      "grad_norm": 0.8427138328552246,
      "learning_rate": 5.230723571332075e-05,
      "loss": 1.0749,
      "step": 14182
    },
    {
      "epoch": 2.6357554357926034,
      "grad_norm": 0.8499711155891418,
      "learning_rate": 5.229439545089392e-05,
      "loss": 0.8501,
      "step": 14183
    },
    {
      "epoch": 2.6359412748559747,
      "grad_norm": 0.8856374621391296,
      "learning_rate": 5.2281556206661366e-05,
      "loss": 1.0512,
      "step": 14184
    },
    {
      "epoch": 2.636127113919346,
      "grad_norm": 0.8373563289642334,
      "learning_rate": 5.2268717980897095e-05,
      "loss": 0.8741,
      "step": 14185
    },
    {
      "epoch": 2.636312952982717,
      "grad_norm": 0.8471721410751343,
      "learning_rate": 5.225588077387511e-05,
      "loss": 0.9665,
      "step": 14186
    },
    {
      "epoch": 2.636498792046088,
      "grad_norm": 0.7628297805786133,
      "learning_rate": 5.2243044585869394e-05,
      "loss": 0.7661,
      "step": 14187
    },
    {
      "epoch": 2.636684631109459,
      "grad_norm": 0.8188031315803528,
      "learning_rate": 5.2230209417153994e-05,
      "loss": 0.9453,
      "step": 14188
    },
    {
      "epoch": 2.6368704701728305,
      "grad_norm": 0.8792202472686768,
      "learning_rate": 5.221737526800279e-05,
      "loss": 0.9707,
      "step": 14189
    },
    {
      "epoch": 2.6370563092362014,
      "grad_norm": 1.096123456954956,
      "learning_rate": 5.2204542138689685e-05,
      "loss": 0.9645,
      "step": 14190
    },
    {
      "epoch": 2.6372421482995727,
      "grad_norm": 0.7871784567832947,
      "learning_rate": 5.219171002948863e-05,
      "loss": 0.9165,
      "step": 14191
    },
    {
      "epoch": 2.6374279873629436,
      "grad_norm": 0.94199538230896,
      "learning_rate": 5.217887894067348e-05,
      "loss": 0.8837,
      "step": 14192
    },
    {
      "epoch": 2.637613826426315,
      "grad_norm": 0.8055230975151062,
      "learning_rate": 5.21660488725181e-05,
      "loss": 0.8327,
      "step": 14193
    },
    {
      "epoch": 2.637799665489686,
      "grad_norm": 0.8414456248283386,
      "learning_rate": 5.215321982529629e-05,
      "loss": 1.0833,
      "step": 14194
    },
    {
      "epoch": 2.637985504553057,
      "grad_norm": 0.8373480439186096,
      "learning_rate": 5.214039179928194e-05,
      "loss": 0.9982,
      "step": 14195
    },
    {
      "epoch": 2.638171343616428,
      "grad_norm": 0.7460327744483948,
      "learning_rate": 5.212756479474878e-05,
      "loss": 0.9654,
      "step": 14196
    },
    {
      "epoch": 2.6383571826797994,
      "grad_norm": 0.8177512288093567,
      "learning_rate": 5.211473881197056e-05,
      "loss": 0.9626,
      "step": 14197
    },
    {
      "epoch": 2.6385430217431702,
      "grad_norm": 0.6966250538825989,
      "learning_rate": 5.210191385122112e-05,
      "loss": 0.7215,
      "step": 14198
    },
    {
      "epoch": 2.6387288608065416,
      "grad_norm": 0.78021240234375,
      "learning_rate": 5.208908991277408e-05,
      "loss": 0.8037,
      "step": 14199
    },
    {
      "epoch": 2.638914699869913,
      "grad_norm": 0.922143280506134,
      "learning_rate": 5.2076266996903225e-05,
      "loss": 1.046,
      "step": 14200
    },
    {
      "epoch": 2.639100538933284,
      "grad_norm": 0.810273289680481,
      "learning_rate": 5.206344510388224e-05,
      "loss": 0.9812,
      "step": 14201
    },
    {
      "epoch": 2.6392863779966547,
      "grad_norm": 0.782341718673706,
      "learning_rate": 5.205062423398469e-05,
      "loss": 0.8912,
      "step": 14202
    },
    {
      "epoch": 2.639472217060026,
      "grad_norm": 1.0330047607421875,
      "learning_rate": 5.203780438748433e-05,
      "loss": 0.8249,
      "step": 14203
    },
    {
      "epoch": 2.6396580561233973,
      "grad_norm": 0.9015677571296692,
      "learning_rate": 5.202498556465474e-05,
      "loss": 1.1167,
      "step": 14204
    },
    {
      "epoch": 2.639843895186768,
      "grad_norm": 0.908305823802948,
      "learning_rate": 5.2012167765769494e-05,
      "loss": 0.9548,
      "step": 14205
    },
    {
      "epoch": 2.6400297342501395,
      "grad_norm": 0.8112418055534363,
      "learning_rate": 5.199935099110214e-05,
      "loss": 1.0671,
      "step": 14206
    },
    {
      "epoch": 2.6402155733135104,
      "grad_norm": 0.8479462265968323,
      "learning_rate": 5.1986535240926294e-05,
      "loss": 0.9551,
      "step": 14207
    },
    {
      "epoch": 2.6404014123768818,
      "grad_norm": 0.6957316398620605,
      "learning_rate": 5.1973720515515434e-05,
      "loss": 0.8288,
      "step": 14208
    },
    {
      "epoch": 2.6405872514402526,
      "grad_norm": 0.8121285438537598,
      "learning_rate": 5.196090681514313e-05,
      "loss": 0.7677,
      "step": 14209
    },
    {
      "epoch": 2.640773090503624,
      "grad_norm": 1.094486117362976,
      "learning_rate": 5.194809414008279e-05,
      "loss": 1.0898,
      "step": 14210
    },
    {
      "epoch": 2.640958929566995,
      "grad_norm": 0.9745634198188782,
      "learning_rate": 5.193528249060796e-05,
      "loss": 1.0696,
      "step": 14211
    },
    {
      "epoch": 2.641144768630366,
      "grad_norm": 0.7873470187187195,
      "learning_rate": 5.192247186699204e-05,
      "loss": 0.8372,
      "step": 14212
    },
    {
      "epoch": 2.641330607693737,
      "grad_norm": 0.8493298888206482,
      "learning_rate": 5.190966226950846e-05,
      "loss": 0.9002,
      "step": 14213
    },
    {
      "epoch": 2.6415164467571084,
      "grad_norm": 0.7814367413520813,
      "learning_rate": 5.189685369843057e-05,
      "loss": 0.8651,
      "step": 14214
    },
    {
      "epoch": 2.6417022858204797,
      "grad_norm": 1.3405762910842896,
      "learning_rate": 5.188404615403183e-05,
      "loss": 0.9166,
      "step": 14215
    },
    {
      "epoch": 2.6418881248838506,
      "grad_norm": 0.8147268295288086,
      "learning_rate": 5.187123963658556e-05,
      "loss": 0.7849,
      "step": 14216
    },
    {
      "epoch": 2.6420739639472215,
      "grad_norm": 0.8468379378318787,
      "learning_rate": 5.1858434146365075e-05,
      "loss": 0.8855,
      "step": 14217
    },
    {
      "epoch": 2.642259803010593,
      "grad_norm": 0.8435200452804565,
      "learning_rate": 5.184562968364366e-05,
      "loss": 1.2382,
      "step": 14218
    },
    {
      "epoch": 2.642445642073964,
      "grad_norm": 1.185233235359192,
      "learning_rate": 5.1832826248694654e-05,
      "loss": 0.9162,
      "step": 14219
    },
    {
      "epoch": 2.642631481137335,
      "grad_norm": 0.8596548438072205,
      "learning_rate": 5.182002384179134e-05,
      "loss": 0.8396,
      "step": 14220
    },
    {
      "epoch": 2.642817320200706,
      "grad_norm": 0.7079363465309143,
      "learning_rate": 5.180722246320695e-05,
      "loss": 0.7973,
      "step": 14221
    },
    {
      "epoch": 2.6430031592640773,
      "grad_norm": 0.7993518114089966,
      "learning_rate": 5.179442211321466e-05,
      "loss": 0.8655,
      "step": 14222
    },
    {
      "epoch": 2.6431889983274486,
      "grad_norm": 0.8697411417961121,
      "learning_rate": 5.1781622792087735e-05,
      "loss": 0.8161,
      "step": 14223
    },
    {
      "epoch": 2.6433748373908195,
      "grad_norm": 1.0144784450531006,
      "learning_rate": 5.176882450009932e-05,
      "loss": 0.9606,
      "step": 14224
    },
    {
      "epoch": 2.643560676454191,
      "grad_norm": 0.7561770677566528,
      "learning_rate": 5.175602723752258e-05,
      "loss": 0.698,
      "step": 14225
    },
    {
      "epoch": 2.6437465155175617,
      "grad_norm": 0.9197513461112976,
      "learning_rate": 5.174323100463061e-05,
      "loss": 0.9161,
      "step": 14226
    },
    {
      "epoch": 2.643932354580933,
      "grad_norm": 0.7830883264541626,
      "learning_rate": 5.173043580169661e-05,
      "loss": 0.8977,
      "step": 14227
    },
    {
      "epoch": 2.644118193644304,
      "grad_norm": 0.7943947911262512,
      "learning_rate": 5.1717641628993615e-05,
      "loss": 1.0571,
      "step": 14228
    },
    {
      "epoch": 2.6443040327076752,
      "grad_norm": 0.800528883934021,
      "learning_rate": 5.1704848486794664e-05,
      "loss": 0.9082,
      "step": 14229
    },
    {
      "epoch": 2.644489871771046,
      "grad_norm": 0.7897242903709412,
      "learning_rate": 5.1692056375372855e-05,
      "loss": 0.8276,
      "step": 14230
    },
    {
      "epoch": 2.6446757108344174,
      "grad_norm": 1.4376215934753418,
      "learning_rate": 5.167926529500124e-05,
      "loss": 1.0305,
      "step": 14231
    },
    {
      "epoch": 2.6448615498977883,
      "grad_norm": 0.7697019577026367,
      "learning_rate": 5.166647524595278e-05,
      "loss": 0.8588,
      "step": 14232
    },
    {
      "epoch": 2.6450473889611597,
      "grad_norm": 0.8492146134376526,
      "learning_rate": 5.1653686228500465e-05,
      "loss": 0.996,
      "step": 14233
    },
    {
      "epoch": 2.645233228024531,
      "grad_norm": 1.0387376546859741,
      "learning_rate": 5.16408982429172e-05,
      "loss": 0.9314,
      "step": 14234
    },
    {
      "epoch": 2.645419067087902,
      "grad_norm": 1.6353427171707153,
      "learning_rate": 5.162811128947602e-05,
      "loss": 1.4163,
      "step": 14235
    },
    {
      "epoch": 2.6456049061512728,
      "grad_norm": 0.9683260321617126,
      "learning_rate": 5.161532536844981e-05,
      "loss": 0.8203,
      "step": 14236
    },
    {
      "epoch": 2.645790745214644,
      "grad_norm": 0.874306321144104,
      "learning_rate": 5.1602540480111436e-05,
      "loss": 0.8829,
      "step": 14237
    },
    {
      "epoch": 2.6459765842780154,
      "grad_norm": 0.7753832936286926,
      "learning_rate": 5.1589756624733756e-05,
      "loss": 0.8694,
      "step": 14238
    },
    {
      "epoch": 2.6461624233413863,
      "grad_norm": 0.7963182330131531,
      "learning_rate": 5.157697380258962e-05,
      "loss": 0.9056,
      "step": 14239
    },
    {
      "epoch": 2.6463482624047576,
      "grad_norm": 1.00355064868927,
      "learning_rate": 5.156419201395194e-05,
      "loss": 1.0158,
      "step": 14240
    },
    {
      "epoch": 2.6465341014681285,
      "grad_norm": 1.5292688608169556,
      "learning_rate": 5.1551411259093466e-05,
      "loss": 1.0719,
      "step": 14241
    },
    {
      "epoch": 2.6467199405315,
      "grad_norm": 0.725684404373169,
      "learning_rate": 5.153863153828692e-05,
      "loss": 0.9253,
      "step": 14242
    },
    {
      "epoch": 2.6469057795948707,
      "grad_norm": 0.6308286786079407,
      "learning_rate": 5.152585285180517e-05,
      "loss": 0.536,
      "step": 14243
    },
    {
      "epoch": 2.647091618658242,
      "grad_norm": 0.7836467623710632,
      "learning_rate": 5.15130751999209e-05,
      "loss": 0.9303,
      "step": 14244
    },
    {
      "epoch": 2.647277457721613,
      "grad_norm": 0.6353241801261902,
      "learning_rate": 5.150029858290684e-05,
      "loss": 0.5537,
      "step": 14245
    },
    {
      "epoch": 2.6474632967849843,
      "grad_norm": 0.959025502204895,
      "learning_rate": 5.148752300103563e-05,
      "loss": 1.0209,
      "step": 14246
    },
    {
      "epoch": 2.647649135848355,
      "grad_norm": 0.9193606972694397,
      "learning_rate": 5.1474748454580047e-05,
      "loss": 1.0573,
      "step": 14247
    },
    {
      "epoch": 2.6478349749117265,
      "grad_norm": 0.7997956871986389,
      "learning_rate": 5.1461974943812674e-05,
      "loss": 0.9159,
      "step": 14248
    },
    {
      "epoch": 2.648020813975098,
      "grad_norm": 0.8786797523498535,
      "learning_rate": 5.1449202469006155e-05,
      "loss": 0.7875,
      "step": 14249
    },
    {
      "epoch": 2.6482066530384687,
      "grad_norm": 0.9094030261039734,
      "learning_rate": 5.1436431030433005e-05,
      "loss": 1.0004,
      "step": 14250
    },
    {
      "epoch": 2.6483924921018396,
      "grad_norm": 0.9047142267227173,
      "learning_rate": 5.142366062836599e-05,
      "loss": 0.9802,
      "step": 14251
    },
    {
      "epoch": 2.648578331165211,
      "grad_norm": 1.208447813987732,
      "learning_rate": 5.1410891263077566e-05,
      "loss": 0.9259,
      "step": 14252
    },
    {
      "epoch": 2.6487641702285822,
      "grad_norm": 0.8681517243385315,
      "learning_rate": 5.13981229348403e-05,
      "loss": 1.0942,
      "step": 14253
    },
    {
      "epoch": 2.648950009291953,
      "grad_norm": 0.6221885681152344,
      "learning_rate": 5.138535564392664e-05,
      "loss": 0.7514,
      "step": 14254
    },
    {
      "epoch": 2.6491358483553245,
      "grad_norm": 0.7742706537246704,
      "learning_rate": 5.137258939060918e-05,
      "loss": 0.9531,
      "step": 14255
    },
    {
      "epoch": 2.6493216874186953,
      "grad_norm": 0.757107138633728,
      "learning_rate": 5.135982417516036e-05,
      "loss": 0.9992,
      "step": 14256
    },
    {
      "epoch": 2.6495075264820667,
      "grad_norm": 0.9197544455528259,
      "learning_rate": 5.134705999785261e-05,
      "loss": 0.9013,
      "step": 14257
    },
    {
      "epoch": 2.6496933655454376,
      "grad_norm": 1.5455429553985596,
      "learning_rate": 5.133429685895834e-05,
      "loss": 1.1803,
      "step": 14258
    },
    {
      "epoch": 2.649879204608809,
      "grad_norm": 0.7492552995681763,
      "learning_rate": 5.1321534758750037e-05,
      "loss": 0.8355,
      "step": 14259
    },
    {
      "epoch": 2.6500650436721798,
      "grad_norm": 0.8414443135261536,
      "learning_rate": 5.130877369749999e-05,
      "loss": 1.0262,
      "step": 14260
    },
    {
      "epoch": 2.650250882735551,
      "grad_norm": 0.9861397743225098,
      "learning_rate": 5.129601367548065e-05,
      "loss": 1.1301,
      "step": 14261
    },
    {
      "epoch": 2.650436721798922,
      "grad_norm": 0.833222508430481,
      "learning_rate": 5.1283254692964286e-05,
      "loss": 1.2049,
      "step": 14262
    },
    {
      "epoch": 2.6506225608622933,
      "grad_norm": 0.7767835855484009,
      "learning_rate": 5.127049675022329e-05,
      "loss": 0.7874,
      "step": 14263
    },
    {
      "epoch": 2.6508083999256646,
      "grad_norm": 0.7894871830940247,
      "learning_rate": 5.125773984752991e-05,
      "loss": 0.8248,
      "step": 14264
    },
    {
      "epoch": 2.6509942389890355,
      "grad_norm": 0.7002455592155457,
      "learning_rate": 5.124498398515642e-05,
      "loss": 0.8009,
      "step": 14265
    },
    {
      "epoch": 2.6511800780524064,
      "grad_norm": 0.8562579154968262,
      "learning_rate": 5.1232229163375064e-05,
      "loss": 1.1361,
      "step": 14266
    },
    {
      "epoch": 2.6513659171157777,
      "grad_norm": 0.7994339466094971,
      "learning_rate": 5.12194753824581e-05,
      "loss": 0.7451,
      "step": 14267
    },
    {
      "epoch": 2.651551756179149,
      "grad_norm": 0.9162044525146484,
      "learning_rate": 5.120672264267775e-05,
      "loss": 0.9215,
      "step": 14268
    },
    {
      "epoch": 2.65173759524252,
      "grad_norm": 0.9385553598403931,
      "learning_rate": 5.119397094430616e-05,
      "loss": 0.9916,
      "step": 14269
    },
    {
      "epoch": 2.651923434305891,
      "grad_norm": 1.3068491220474243,
      "learning_rate": 5.118122028761547e-05,
      "loss": 1.2854,
      "step": 14270
    },
    {
      "epoch": 2.652109273369262,
      "grad_norm": 0.8506157398223877,
      "learning_rate": 5.1168470672877845e-05,
      "loss": 0.9981,
      "step": 14271
    },
    {
      "epoch": 2.6522951124326335,
      "grad_norm": 0.7934330701828003,
      "learning_rate": 5.1155722100365465e-05,
      "loss": 0.989,
      "step": 14272
    },
    {
      "epoch": 2.6524809514960044,
      "grad_norm": 0.8915961384773254,
      "learning_rate": 5.114297457035038e-05,
      "loss": 0.9278,
      "step": 14273
    },
    {
      "epoch": 2.6526667905593757,
      "grad_norm": 0.7989098429679871,
      "learning_rate": 5.1130228083104615e-05,
      "loss": 0.8468,
      "step": 14274
    },
    {
      "epoch": 2.6528526296227466,
      "grad_norm": 0.7344555258750916,
      "learning_rate": 5.111748263890031e-05,
      "loss": 0.8673,
      "step": 14275
    },
    {
      "epoch": 2.653038468686118,
      "grad_norm": 0.7172108888626099,
      "learning_rate": 5.1104738238009444e-05,
      "loss": 0.9453,
      "step": 14276
    },
    {
      "epoch": 2.653224307749489,
      "grad_norm": 0.7355473637580872,
      "learning_rate": 5.109199488070404e-05,
      "loss": 0.9495,
      "step": 14277
    },
    {
      "epoch": 2.65341014681286,
      "grad_norm": 0.8460148572921753,
      "learning_rate": 5.107925256725603e-05,
      "loss": 0.8353,
      "step": 14278
    },
    {
      "epoch": 2.653595985876231,
      "grad_norm": 1.0438544750213623,
      "learning_rate": 5.106651129793747e-05,
      "loss": 0.8366,
      "step": 14279
    },
    {
      "epoch": 2.6537818249396024,
      "grad_norm": 0.8582540154457092,
      "learning_rate": 5.1053771073020254e-05,
      "loss": 1.1757,
      "step": 14280
    },
    {
      "epoch": 2.6539676640029732,
      "grad_norm": 0.7860969305038452,
      "learning_rate": 5.104103189277625e-05,
      "loss": 0.6489,
      "step": 14281
    },
    {
      "epoch": 2.6541535030663446,
      "grad_norm": 0.7694686055183411,
      "learning_rate": 5.1028293757477416e-05,
      "loss": 1.0067,
      "step": 14282
    },
    {
      "epoch": 2.654339342129716,
      "grad_norm": 0.863415539264679,
      "learning_rate": 5.1015556667395636e-05,
      "loss": 0.8971,
      "step": 14283
    },
    {
      "epoch": 2.654525181193087,
      "grad_norm": 1.3249080181121826,
      "learning_rate": 5.100282062280275e-05,
      "loss": 1.3751,
      "step": 14284
    },
    {
      "epoch": 2.6547110202564577,
      "grad_norm": 1.5945967435836792,
      "learning_rate": 5.0990085623970585e-05,
      "loss": 1.4996,
      "step": 14285
    },
    {
      "epoch": 2.654896859319829,
      "grad_norm": 0.8719706535339355,
      "learning_rate": 5.097735167117088e-05,
      "loss": 1.0089,
      "step": 14286
    },
    {
      "epoch": 2.6550826983832003,
      "grad_norm": 0.7747623920440674,
      "learning_rate": 5.096461876467553e-05,
      "loss": 0.8771,
      "step": 14287
    },
    {
      "epoch": 2.655268537446571,
      "grad_norm": 0.7249778509140015,
      "learning_rate": 5.095188690475623e-05,
      "loss": 0.9572,
      "step": 14288
    },
    {
      "epoch": 2.6554543765099425,
      "grad_norm": 0.9162830114364624,
      "learning_rate": 5.093915609168474e-05,
      "loss": 0.9416,
      "step": 14289
    },
    {
      "epoch": 2.6556402155733134,
      "grad_norm": 0.6986264586448669,
      "learning_rate": 5.0926426325732746e-05,
      "loss": 1.0275,
      "step": 14290
    },
    {
      "epoch": 2.6558260546366848,
      "grad_norm": 0.9343037009239197,
      "learning_rate": 5.091369760717199e-05,
      "loss": 1.0384,
      "step": 14291
    },
    {
      "epoch": 2.6560118937000556,
      "grad_norm": 0.8413649797439575,
      "learning_rate": 5.090096993627409e-05,
      "loss": 0.6088,
      "step": 14292
    },
    {
      "epoch": 2.656197732763427,
      "grad_norm": 0.788287341594696,
      "learning_rate": 5.088824331331077e-05,
      "loss": 0.897,
      "step": 14293
    },
    {
      "epoch": 2.656383571826798,
      "grad_norm": 0.8309547305107117,
      "learning_rate": 5.087551773855357e-05,
      "loss": 0.8824,
      "step": 14294
    },
    {
      "epoch": 2.656569410890169,
      "grad_norm": 0.8467617630958557,
      "learning_rate": 5.0862793212274186e-05,
      "loss": 0.7063,
      "step": 14295
    },
    {
      "epoch": 2.65675524995354,
      "grad_norm": 0.8307881355285645,
      "learning_rate": 5.0850069734744166e-05,
      "loss": 0.9054,
      "step": 14296
    },
    {
      "epoch": 2.6569410890169114,
      "grad_norm": 0.9663828015327454,
      "learning_rate": 5.083734730623506e-05,
      "loss": 0.7634,
      "step": 14297
    },
    {
      "epoch": 2.6571269280802827,
      "grad_norm": 0.9168074727058411,
      "learning_rate": 5.0824625927018366e-05,
      "loss": 0.9365,
      "step": 14298
    },
    {
      "epoch": 2.6573127671436536,
      "grad_norm": 0.7665404677391052,
      "learning_rate": 5.0811905597365684e-05,
      "loss": 0.7316,
      "step": 14299
    },
    {
      "epoch": 2.6574986062070245,
      "grad_norm": 0.8714559674263,
      "learning_rate": 5.079918631754847e-05,
      "loss": 0.993,
      "step": 14300
    },
    {
      "epoch": 2.657684445270396,
      "grad_norm": 0.7602294087409973,
      "learning_rate": 5.078646808783815e-05,
      "loss": 0.8304,
      "step": 14301
    },
    {
      "epoch": 2.657870284333767,
      "grad_norm": 0.9733949303627014,
      "learning_rate": 5.077375090850621e-05,
      "loss": 0.7749,
      "step": 14302
    },
    {
      "epoch": 2.658056123397138,
      "grad_norm": 0.9775307178497314,
      "learning_rate": 5.0761034779824126e-05,
      "loss": 0.7289,
      "step": 14303
    },
    {
      "epoch": 2.658241962460509,
      "grad_norm": 0.7890309691429138,
      "learning_rate": 5.074831970206326e-05,
      "loss": 0.9095,
      "step": 14304
    },
    {
      "epoch": 2.6584278015238803,
      "grad_norm": 0.8886024951934814,
      "learning_rate": 5.073560567549498e-05,
      "loss": 0.9967,
      "step": 14305
    },
    {
      "epoch": 2.6586136405872516,
      "grad_norm": 0.7603307962417603,
      "learning_rate": 5.072289270039063e-05,
      "loss": 0.6832,
      "step": 14306
    },
    {
      "epoch": 2.6587994796506225,
      "grad_norm": 0.7396262884140015,
      "learning_rate": 5.071018077702161e-05,
      "loss": 0.9276,
      "step": 14307
    },
    {
      "epoch": 2.658985318713994,
      "grad_norm": 0.7766372561454773,
      "learning_rate": 5.0697469905659176e-05,
      "loss": 0.8897,
      "step": 14308
    },
    {
      "epoch": 2.6591711577773647,
      "grad_norm": 0.79985511302948,
      "learning_rate": 5.0684760086574665e-05,
      "loss": 0.8603,
      "step": 14309
    },
    {
      "epoch": 2.659356996840736,
      "grad_norm": 0.9170798659324646,
      "learning_rate": 5.067205132003927e-05,
      "loss": 0.9559,
      "step": 14310
    },
    {
      "epoch": 2.659542835904107,
      "grad_norm": 0.8943963050842285,
      "learning_rate": 5.065934360632434e-05,
      "loss": 0.8446,
      "step": 14311
    },
    {
      "epoch": 2.6597286749674782,
      "grad_norm": 0.8036159873008728,
      "learning_rate": 5.064663694570104e-05,
      "loss": 0.7597,
      "step": 14312
    },
    {
      "epoch": 2.6599145140308496,
      "grad_norm": 0.7508684396743774,
      "learning_rate": 5.063393133844055e-05,
      "loss": 0.7334,
      "step": 14313
    },
    {
      "epoch": 2.6601003530942204,
      "grad_norm": 0.9858553409576416,
      "learning_rate": 5.062122678481407e-05,
      "loss": 0.7561,
      "step": 14314
    },
    {
      "epoch": 2.6602861921575913,
      "grad_norm": 0.7608938813209534,
      "learning_rate": 5.06085232850928e-05,
      "loss": 0.7628,
      "step": 14315
    },
    {
      "epoch": 2.6604720312209627,
      "grad_norm": 0.8505204319953918,
      "learning_rate": 5.059582083954785e-05,
      "loss": 1.1395,
      "step": 14316
    },
    {
      "epoch": 2.660657870284334,
      "grad_norm": 0.7545738816261292,
      "learning_rate": 5.0583119448450324e-05,
      "loss": 0.8516,
      "step": 14317
    },
    {
      "epoch": 2.660843709347705,
      "grad_norm": 1.0585992336273193,
      "learning_rate": 5.057041911207128e-05,
      "loss": 0.7651,
      "step": 14318
    },
    {
      "epoch": 2.6610295484110758,
      "grad_norm": 0.9138060212135315,
      "learning_rate": 5.055771983068183e-05,
      "loss": 1.0043,
      "step": 14319
    },
    {
      "epoch": 2.661215387474447,
      "grad_norm": 0.801011860370636,
      "learning_rate": 5.0545021604553024e-05,
      "loss": 0.8839,
      "step": 14320
    },
    {
      "epoch": 2.6614012265378184,
      "grad_norm": 0.9903347492218018,
      "learning_rate": 5.053232443395581e-05,
      "loss": 0.8389,
      "step": 14321
    },
    {
      "epoch": 2.6615870656011893,
      "grad_norm": 0.7993159890174866,
      "learning_rate": 5.05196283191613e-05,
      "loss": 1.0068,
      "step": 14322
    },
    {
      "epoch": 2.6617729046645606,
      "grad_norm": 0.9454100728034973,
      "learning_rate": 5.050693326044036e-05,
      "loss": 0.8917,
      "step": 14323
    },
    {
      "epoch": 2.6619587437279315,
      "grad_norm": 0.971154510974884,
      "learning_rate": 5.049423925806403e-05,
      "loss": 0.763,
      "step": 14324
    },
    {
      "epoch": 2.662144582791303,
      "grad_norm": 0.7480639815330505,
      "learning_rate": 5.0481546312303216e-05,
      "loss": 0.8984,
      "step": 14325
    },
    {
      "epoch": 2.6623304218546737,
      "grad_norm": 0.8236550092697144,
      "learning_rate": 5.046885442342879e-05,
      "loss": 0.8049,
      "step": 14326
    },
    {
      "epoch": 2.662516260918045,
      "grad_norm": 0.8084285259246826,
      "learning_rate": 5.045616359171169e-05,
      "loss": 1.0081,
      "step": 14327
    },
    {
      "epoch": 2.662702099981416,
      "grad_norm": 0.7616249322891235,
      "learning_rate": 5.044347381742276e-05,
      "loss": 1.01,
      "step": 14328
    },
    {
      "epoch": 2.6628879390447873,
      "grad_norm": 0.9445034265518188,
      "learning_rate": 5.0430785100832846e-05,
      "loss": 0.9182,
      "step": 14329
    },
    {
      "epoch": 2.663073778108158,
      "grad_norm": 0.7334463596343994,
      "learning_rate": 5.0418097442212716e-05,
      "loss": 0.9515,
      "step": 14330
    },
    {
      "epoch": 2.6632596171715295,
      "grad_norm": 0.8281427621841431,
      "learning_rate": 5.0405410841833253e-05,
      "loss": 0.5546,
      "step": 14331
    },
    {
      "epoch": 2.663445456234901,
      "grad_norm": 0.8622934818267822,
      "learning_rate": 5.039272529996518e-05,
      "loss": 0.9142,
      "step": 14332
    },
    {
      "epoch": 2.6636312952982717,
      "grad_norm": 0.8588988780975342,
      "learning_rate": 5.038004081687922e-05,
      "loss": 0.9058,
      "step": 14333
    },
    {
      "epoch": 2.6638171343616426,
      "grad_norm": 1.0082697868347168,
      "learning_rate": 5.0367357392846146e-05,
      "loss": 0.7629,
      "step": 14334
    },
    {
      "epoch": 2.664002973425014,
      "grad_norm": 0.8957542777061462,
      "learning_rate": 5.0354675028136685e-05,
      "loss": 1.014,
      "step": 14335
    },
    {
      "epoch": 2.6641888124883852,
      "grad_norm": 0.828673243522644,
      "learning_rate": 5.03419937230215e-05,
      "loss": 0.8816,
      "step": 14336
    },
    {
      "epoch": 2.664374651551756,
      "grad_norm": 0.9521352648735046,
      "learning_rate": 5.032931347777123e-05,
      "loss": 1.1619,
      "step": 14337
    },
    {
      "epoch": 2.6645604906151275,
      "grad_norm": 0.7346090078353882,
      "learning_rate": 5.0316634292656495e-05,
      "loss": 0.9334,
      "step": 14338
    },
    {
      "epoch": 2.6647463296784983,
      "grad_norm": 0.8352324962615967,
      "learning_rate": 5.0303956167947976e-05,
      "loss": 0.9386,
      "step": 14339
    },
    {
      "epoch": 2.6649321687418697,
      "grad_norm": 0.9011669754981995,
      "learning_rate": 5.029127910391623e-05,
      "loss": 1.0561,
      "step": 14340
    },
    {
      "epoch": 2.6651180078052406,
      "grad_norm": 0.796430230140686,
      "learning_rate": 5.027860310083179e-05,
      "loss": 1.1265,
      "step": 14341
    },
    {
      "epoch": 2.665303846868612,
      "grad_norm": 0.8729113340377808,
      "learning_rate": 5.0265928158965295e-05,
      "loss": 1.0006,
      "step": 14342
    },
    {
      "epoch": 2.6654896859319828,
      "grad_norm": 0.9211896657943726,
      "learning_rate": 5.0253254278587195e-05,
      "loss": 1.0051,
      "step": 14343
    },
    {
      "epoch": 2.665675524995354,
      "grad_norm": 0.8242248892784119,
      "learning_rate": 5.024058145996797e-05,
      "loss": 1.1681,
      "step": 14344
    },
    {
      "epoch": 2.665861364058725,
      "grad_norm": 1.0670288801193237,
      "learning_rate": 5.0227909703378204e-05,
      "loss": 1.1635,
      "step": 14345
    },
    {
      "epoch": 2.6660472031220963,
      "grad_norm": 0.8117260932922363,
      "learning_rate": 5.021523900908824e-05,
      "loss": 0.8082,
      "step": 14346
    },
    {
      "epoch": 2.6662330421854676,
      "grad_norm": 0.9363516569137573,
      "learning_rate": 5.0202569377368616e-05,
      "loss": 1.0035,
      "step": 14347
    },
    {
      "epoch": 2.6664188812488385,
      "grad_norm": 0.8591864705085754,
      "learning_rate": 5.018990080848969e-05,
      "loss": 0.9849,
      "step": 14348
    },
    {
      "epoch": 2.6666047203122094,
      "grad_norm": 0.7635031342506409,
      "learning_rate": 5.017723330272184e-05,
      "loss": 1.0059,
      "step": 14349
    },
    {
      "epoch": 2.6667905593755807,
      "grad_norm": 0.7889599204063416,
      "learning_rate": 5.0164566860335414e-05,
      "loss": 0.5671,
      "step": 14350
    },
    {
      "epoch": 2.666976398438952,
      "grad_norm": 0.7520024180412292,
      "learning_rate": 5.015190148160083e-05,
      "loss": 0.9146,
      "step": 14351
    },
    {
      "epoch": 2.667162237502323,
      "grad_norm": 0.6558710932731628,
      "learning_rate": 5.013923716678837e-05,
      "loss": 0.7512,
      "step": 14352
    },
    {
      "epoch": 2.667348076565694,
      "grad_norm": 0.8916223645210266,
      "learning_rate": 5.0126573916168286e-05,
      "loss": 0.9032,
      "step": 14353
    },
    {
      "epoch": 2.667533915629065,
      "grad_norm": 0.8456600904464722,
      "learning_rate": 5.011391173001093e-05,
      "loss": 0.8913,
      "step": 14354
    },
    {
      "epoch": 2.6677197546924365,
      "grad_norm": 0.7922993898391724,
      "learning_rate": 5.010125060858648e-05,
      "loss": 1.0398,
      "step": 14355
    },
    {
      "epoch": 2.6679055937558074,
      "grad_norm": 0.8712417483329773,
      "learning_rate": 5.008859055216526e-05,
      "loss": 0.8474,
      "step": 14356
    },
    {
      "epoch": 2.6680914328191787,
      "grad_norm": 0.8451159000396729,
      "learning_rate": 5.0075931561017396e-05,
      "loss": 0.7022,
      "step": 14357
    },
    {
      "epoch": 2.6682772718825496,
      "grad_norm": 0.9817770719528198,
      "learning_rate": 5.006327363541308e-05,
      "loss": 0.9718,
      "step": 14358
    },
    {
      "epoch": 2.668463110945921,
      "grad_norm": 0.8464716076850891,
      "learning_rate": 5.005061677562253e-05,
      "loss": 0.907,
      "step": 14359
    },
    {
      "epoch": 2.668648950009292,
      "grad_norm": 0.8030047416687012,
      "learning_rate": 5.0037960981915844e-05,
      "loss": 0.7352,
      "step": 14360
    },
    {
      "epoch": 2.668834789072663,
      "grad_norm": 0.8626695871353149,
      "learning_rate": 5.002530625456312e-05,
      "loss": 0.8694,
      "step": 14361
    },
    {
      "epoch": 2.6690206281360345,
      "grad_norm": 0.7666325569152832,
      "learning_rate": 5.0012652593834496e-05,
      "loss": 0.796,
      "step": 14362
    },
    {
      "epoch": 2.6692064671994054,
      "grad_norm": 0.7711402773857117,
      "learning_rate": 5.000000000000002e-05,
      "loss": 0.7858,
      "step": 14363
    },
    {
      "epoch": 2.6693923062627762,
      "grad_norm": 0.7479420304298401,
      "learning_rate": 4.998734847332974e-05,
      "loss": 0.9035,
      "step": 14364
    },
    {
      "epoch": 2.6695781453261476,
      "grad_norm": 1.0149599313735962,
      "learning_rate": 4.9974698014093655e-05,
      "loss": 0.9755,
      "step": 14365
    },
    {
      "epoch": 2.669763984389519,
      "grad_norm": 0.8029696941375732,
      "learning_rate": 4.996204862256179e-05,
      "loss": 1.0167,
      "step": 14366
    },
    {
      "epoch": 2.66994982345289,
      "grad_norm": 0.8907744884490967,
      "learning_rate": 4.9949400299004155e-05,
      "loss": 0.9284,
      "step": 14367
    },
    {
      "epoch": 2.6701356625162607,
      "grad_norm": 0.8258806467056274,
      "learning_rate": 4.9936753043690695e-05,
      "loss": 0.8525,
      "step": 14368
    },
    {
      "epoch": 2.670321501579632,
      "grad_norm": 0.7489919066429138,
      "learning_rate": 4.992410685689133e-05,
      "loss": 0.8296,
      "step": 14369
    },
    {
      "epoch": 2.6705073406430033,
      "grad_norm": 0.8124386072158813,
      "learning_rate": 4.991146173887592e-05,
      "loss": 0.7661,
      "step": 14370
    },
    {
      "epoch": 2.670693179706374,
      "grad_norm": 0.9028873443603516,
      "learning_rate": 4.9898817689914445e-05,
      "loss": 0.9848,
      "step": 14371
    },
    {
      "epoch": 2.6708790187697455,
      "grad_norm": 0.892085075378418,
      "learning_rate": 4.988617471027673e-05,
      "loss": 0.8562,
      "step": 14372
    },
    {
      "epoch": 2.6710648578331164,
      "grad_norm": 0.9520072340965271,
      "learning_rate": 4.987353280023259e-05,
      "loss": 1.12,
      "step": 14373
    },
    {
      "epoch": 2.6712506968964878,
      "grad_norm": 0.8968404531478882,
      "learning_rate": 4.98608919600519e-05,
      "loss": 1.0637,
      "step": 14374
    },
    {
      "epoch": 2.6714365359598586,
      "grad_norm": 1.0067741870880127,
      "learning_rate": 4.984825219000443e-05,
      "loss": 1.0221,
      "step": 14375
    },
    {
      "epoch": 2.67162237502323,
      "grad_norm": 0.9988031983375549,
      "learning_rate": 4.983561349035991e-05,
      "loss": 1.0664,
      "step": 14376
    },
    {
      "epoch": 2.671808214086601,
      "grad_norm": 0.7657214999198914,
      "learning_rate": 4.982297586138819e-05,
      "loss": 0.9417,
      "step": 14377
    },
    {
      "epoch": 2.671994053149972,
      "grad_norm": 0.8332259058952332,
      "learning_rate": 4.981033930335889e-05,
      "loss": 0.8141,
      "step": 14378
    },
    {
      "epoch": 2.672179892213343,
      "grad_norm": 0.9903144240379333,
      "learning_rate": 4.979770381654181e-05,
      "loss": 0.9264,
      "step": 14379
    },
    {
      "epoch": 2.6723657312767144,
      "grad_norm": 0.9208203554153442,
      "learning_rate": 4.978506940120659e-05,
      "loss": 1.1661,
      "step": 14380
    },
    {
      "epoch": 2.6725515703400857,
      "grad_norm": 0.9300960898399353,
      "learning_rate": 4.977243605762285e-05,
      "loss": 0.9891,
      "step": 14381
    },
    {
      "epoch": 2.6727374094034566,
      "grad_norm": 0.740096926689148,
      "learning_rate": 4.975980378606031e-05,
      "loss": 0.8299,
      "step": 14382
    },
    {
      "epoch": 2.6729232484668275,
      "grad_norm": 0.8849905133247375,
      "learning_rate": 4.9747172586788546e-05,
      "loss": 0.8402,
      "step": 14383
    },
    {
      "epoch": 2.673109087530199,
      "grad_norm": 0.8056560754776001,
      "learning_rate": 4.973454246007714e-05,
      "loss": 0.9675,
      "step": 14384
    },
    {
      "epoch": 2.67329492659357,
      "grad_norm": 0.8289986252784729,
      "learning_rate": 4.972191340619563e-05,
      "loss": 0.7964,
      "step": 14385
    },
    {
      "epoch": 2.673480765656941,
      "grad_norm": 0.9971672892570496,
      "learning_rate": 4.97092854254136e-05,
      "loss": 0.8966,
      "step": 14386
    },
    {
      "epoch": 2.6736666047203124,
      "grad_norm": 0.9446187019348145,
      "learning_rate": 4.9696658518000613e-05,
      "loss": 0.8894,
      "step": 14387
    },
    {
      "epoch": 2.6738524437836833,
      "grad_norm": 0.7367177605628967,
      "learning_rate": 4.968403268422612e-05,
      "loss": 0.8897,
      "step": 14388
    },
    {
      "epoch": 2.6740382828470546,
      "grad_norm": 0.9521549940109253,
      "learning_rate": 4.96714079243596e-05,
      "loss": 0.9574,
      "step": 14389
    },
    {
      "epoch": 2.6742241219104255,
      "grad_norm": 0.8508783578872681,
      "learning_rate": 4.9658784238670475e-05,
      "loss": 0.9341,
      "step": 14390
    },
    {
      "epoch": 2.674409960973797,
      "grad_norm": 0.9512181282043457,
      "learning_rate": 4.964616162742826e-05,
      "loss": 0.9891,
      "step": 14391
    },
    {
      "epoch": 2.6745958000371677,
      "grad_norm": 0.8128776550292969,
      "learning_rate": 4.9633540090902296e-05,
      "loss": 0.9037,
      "step": 14392
    },
    {
      "epoch": 2.674781639100539,
      "grad_norm": 0.8413349390029907,
      "learning_rate": 4.962091962936196e-05,
      "loss": 0.7678,
      "step": 14393
    },
    {
      "epoch": 2.67496747816391,
      "grad_norm": 0.9310625195503235,
      "learning_rate": 4.9608300243076674e-05,
      "loss": 0.9023,
      "step": 14394
    },
    {
      "epoch": 2.6751533172272812,
      "grad_norm": 0.7659721970558167,
      "learning_rate": 4.959568193231575e-05,
      "loss": 0.9902,
      "step": 14395
    },
    {
      "epoch": 2.6753391562906526,
      "grad_norm": 0.7874768376350403,
      "learning_rate": 4.9583064697348495e-05,
      "loss": 0.9488,
      "step": 14396
    },
    {
      "epoch": 2.6755249953540234,
      "grad_norm": 0.7902805209159851,
      "learning_rate": 4.957044853844417e-05,
      "loss": 0.8912,
      "step": 14397
    },
    {
      "epoch": 2.6757108344173943,
      "grad_norm": 0.7868594527244568,
      "learning_rate": 4.9557833455872074e-05,
      "loss": 1.2056,
      "step": 14398
    },
    {
      "epoch": 2.6758966734807657,
      "grad_norm": 0.7617980241775513,
      "learning_rate": 4.95452194499015e-05,
      "loss": 0.8513,
      "step": 14399
    },
    {
      "epoch": 2.676082512544137,
      "grad_norm": 0.8179283142089844,
      "learning_rate": 4.953260652080163e-05,
      "loss": 0.8543,
      "step": 14400
    },
    {
      "epoch": 2.676268351607508,
      "grad_norm": 0.8407628536224365,
      "learning_rate": 4.951999466884164e-05,
      "loss": 0.9947,
      "step": 14401
    },
    {
      "epoch": 2.6764541906708788,
      "grad_norm": 0.8140806555747986,
      "learning_rate": 4.950738389429077e-05,
      "loss": 0.9707,
      "step": 14402
    },
    {
      "epoch": 2.67664002973425,
      "grad_norm": 0.8304571509361267,
      "learning_rate": 4.949477419741814e-05,
      "loss": 0.8568,
      "step": 14403
    },
    {
      "epoch": 2.6768258687976214,
      "grad_norm": 0.7684166431427002,
      "learning_rate": 4.948216557849288e-05,
      "loss": 0.8984,
      "step": 14404
    },
    {
      "epoch": 2.6770117078609923,
      "grad_norm": 0.8779479265213013,
      "learning_rate": 4.946955803778407e-05,
      "loss": 1.0934,
      "step": 14405
    },
    {
      "epoch": 2.6771975469243636,
      "grad_norm": 0.8864030241966248,
      "learning_rate": 4.945695157556087e-05,
      "loss": 0.6961,
      "step": 14406
    },
    {
      "epoch": 2.6773833859877345,
      "grad_norm": 0.8695092797279358,
      "learning_rate": 4.944434619209226e-05,
      "loss": 0.9414,
      "step": 14407
    },
    {
      "epoch": 2.677569225051106,
      "grad_norm": 0.8751302361488342,
      "learning_rate": 4.943174188764737e-05,
      "loss": 0.9969,
      "step": 14408
    },
    {
      "epoch": 2.6777550641144767,
      "grad_norm": 0.8184906840324402,
      "learning_rate": 4.941913866249517e-05,
      "loss": 0.9043,
      "step": 14409
    },
    {
      "epoch": 2.677940903177848,
      "grad_norm": 0.7370218634605408,
      "learning_rate": 4.9406536516904615e-05,
      "loss": 0.8803,
      "step": 14410
    },
    {
      "epoch": 2.678126742241219,
      "grad_norm": 0.8426598310470581,
      "learning_rate": 4.939393545114475e-05,
      "loss": 0.7561,
      "step": 14411
    },
    {
      "epoch": 2.6783125813045903,
      "grad_norm": 0.9060295820236206,
      "learning_rate": 4.938133546548449e-05,
      "loss": 0.8861,
      "step": 14412
    },
    {
      "epoch": 2.678498420367961,
      "grad_norm": 0.797497570514679,
      "learning_rate": 4.936873656019273e-05,
      "loss": 0.767,
      "step": 14413
    },
    {
      "epoch": 2.6786842594313325,
      "grad_norm": 0.7489145398139954,
      "learning_rate": 4.9356138735538425e-05,
      "loss": 0.8747,
      "step": 14414
    },
    {
      "epoch": 2.678870098494704,
      "grad_norm": 0.7489208579063416,
      "learning_rate": 4.934354199179044e-05,
      "loss": 0.943,
      "step": 14415
    },
    {
      "epoch": 2.6790559375580747,
      "grad_norm": 0.7728361487388611,
      "learning_rate": 4.933094632921762e-05,
      "loss": 0.7232,
      "step": 14416
    },
    {
      "epoch": 2.6792417766214456,
      "grad_norm": 1.0955910682678223,
      "learning_rate": 4.931835174808876e-05,
      "loss": 0.9856,
      "step": 14417
    },
    {
      "epoch": 2.679427615684817,
      "grad_norm": 0.7960167527198792,
      "learning_rate": 4.930575824867271e-05,
      "loss": 0.9901,
      "step": 14418
    },
    {
      "epoch": 2.6796134547481882,
      "grad_norm": 1.1254215240478516,
      "learning_rate": 4.92931658312383e-05,
      "loss": 0.6882,
      "step": 14419
    },
    {
      "epoch": 2.679799293811559,
      "grad_norm": 0.8247450590133667,
      "learning_rate": 4.928057449605423e-05,
      "loss": 0.9769,
      "step": 14420
    },
    {
      "epoch": 2.6799851328749305,
      "grad_norm": 0.8798690438270569,
      "learning_rate": 4.926798424338923e-05,
      "loss": 1.0393,
      "step": 14421
    },
    {
      "epoch": 2.6801709719383013,
      "grad_norm": 0.7442476749420166,
      "learning_rate": 4.925539507351209e-05,
      "loss": 0.8805,
      "step": 14422
    },
    {
      "epoch": 2.6803568110016727,
      "grad_norm": 0.8409537076950073,
      "learning_rate": 4.924280698669146e-05,
      "loss": 1.1698,
      "step": 14423
    },
    {
      "epoch": 2.6805426500650436,
      "grad_norm": 0.7573469281196594,
      "learning_rate": 4.9230219983196015e-05,
      "loss": 0.9185,
      "step": 14424
    },
    {
      "epoch": 2.680728489128415,
      "grad_norm": 0.8142333030700684,
      "learning_rate": 4.921763406329436e-05,
      "loss": 0.997,
      "step": 14425
    },
    {
      "epoch": 2.6809143281917858,
      "grad_norm": 0.7045801281929016,
      "learning_rate": 4.9205049227255194e-05,
      "loss": 0.7366,
      "step": 14426
    },
    {
      "epoch": 2.681100167255157,
      "grad_norm": 0.756362795829773,
      "learning_rate": 4.919246547534708e-05,
      "loss": 0.9306,
      "step": 14427
    },
    {
      "epoch": 2.681286006318528,
      "grad_norm": 0.893324077129364,
      "learning_rate": 4.9179882807838575e-05,
      "loss": 1.0491,
      "step": 14428
    },
    {
      "epoch": 2.6814718453818993,
      "grad_norm": 0.7502169013023376,
      "learning_rate": 4.9167301224998285e-05,
      "loss": 1.0363,
      "step": 14429
    },
    {
      "epoch": 2.6816576844452706,
      "grad_norm": 0.6055135726928711,
      "learning_rate": 4.9154720727094695e-05,
      "loss": 0.7655,
      "step": 14430
    },
    {
      "epoch": 2.6818435235086415,
      "grad_norm": 0.8633196949958801,
      "learning_rate": 4.914214131439636e-05,
      "loss": 0.937,
      "step": 14431
    },
    {
      "epoch": 2.6820293625720124,
      "grad_norm": 0.7565577030181885,
      "learning_rate": 4.912956298717175e-05,
      "loss": 0.8577,
      "step": 14432
    },
    {
      "epoch": 2.6822152016353837,
      "grad_norm": 1.0513862371444702,
      "learning_rate": 4.9116985745689294e-05,
      "loss": 1.1963,
      "step": 14433
    },
    {
      "epoch": 2.682401040698755,
      "grad_norm": 0.7482678890228271,
      "learning_rate": 4.910440959021748e-05,
      "loss": 0.8609,
      "step": 14434
    },
    {
      "epoch": 2.682586879762126,
      "grad_norm": 0.7162086367607117,
      "learning_rate": 4.9091834521024714e-05,
      "loss": 0.8151,
      "step": 14435
    },
    {
      "epoch": 2.6827727188254973,
      "grad_norm": 0.7310633659362793,
      "learning_rate": 4.9079260538379366e-05,
      "loss": 0.7638,
      "step": 14436
    },
    {
      "epoch": 2.682958557888868,
      "grad_norm": 0.8538556098937988,
      "learning_rate": 4.906668764254979e-05,
      "loss": 0.9732,
      "step": 14437
    },
    {
      "epoch": 2.6831443969522395,
      "grad_norm": 0.8188505172729492,
      "learning_rate": 4.90541158338044e-05,
      "loss": 0.9777,
      "step": 14438
    },
    {
      "epoch": 2.6833302360156104,
      "grad_norm": 0.6868131160736084,
      "learning_rate": 4.9041545112411434e-05,
      "loss": 0.7902,
      "step": 14439
    },
    {
      "epoch": 2.6835160750789817,
      "grad_norm": 0.7646917700767517,
      "learning_rate": 4.902897547863928e-05,
      "loss": 0.8853,
      "step": 14440
    },
    {
      "epoch": 2.6837019141423526,
      "grad_norm": 0.8058961033821106,
      "learning_rate": 4.9016406932756155e-05,
      "loss": 0.9065,
      "step": 14441
    },
    {
      "epoch": 2.683887753205724,
      "grad_norm": 0.9321679472923279,
      "learning_rate": 4.900383947503034e-05,
      "loss": 1.0314,
      "step": 14442
    },
    {
      "epoch": 2.684073592269095,
      "grad_norm": 0.786422610282898,
      "learning_rate": 4.899127310573009e-05,
      "loss": 0.7887,
      "step": 14443
    },
    {
      "epoch": 2.684259431332466,
      "grad_norm": 0.874234676361084,
      "learning_rate": 4.897870782512356e-05,
      "loss": 1.1024,
      "step": 14444
    },
    {
      "epoch": 2.6844452703958375,
      "grad_norm": 0.8297612071037292,
      "learning_rate": 4.896614363347892e-05,
      "loss": 0.8138,
      "step": 14445
    },
    {
      "epoch": 2.6846311094592084,
      "grad_norm": 0.9445776343345642,
      "learning_rate": 4.89535805310644e-05,
      "loss": 0.8972,
      "step": 14446
    },
    {
      "epoch": 2.6848169485225792,
      "grad_norm": 0.9599937796592712,
      "learning_rate": 4.894101851814812e-05,
      "loss": 0.7803,
      "step": 14447
    },
    {
      "epoch": 2.6850027875859506,
      "grad_norm": 0.8982528448104858,
      "learning_rate": 4.8928457594998165e-05,
      "loss": 0.8521,
      "step": 14448
    },
    {
      "epoch": 2.685188626649322,
      "grad_norm": 0.9184148907661438,
      "learning_rate": 4.891589776188261e-05,
      "loss": 1.1211,
      "step": 14449
    },
    {
      "epoch": 2.685374465712693,
      "grad_norm": 0.8056932687759399,
      "learning_rate": 4.890333901906955e-05,
      "loss": 0.9148,
      "step": 14450
    },
    {
      "epoch": 2.6855603047760637,
      "grad_norm": 0.847355306148529,
      "learning_rate": 4.889078136682708e-05,
      "loss": 0.9258,
      "step": 14451
    },
    {
      "epoch": 2.685746143839435,
      "grad_norm": 0.9067242741584778,
      "learning_rate": 4.8878224805423167e-05,
      "loss": 0.8694,
      "step": 14452
    },
    {
      "epoch": 2.6859319829028063,
      "grad_norm": 1.0393316745758057,
      "learning_rate": 4.886566933512578e-05,
      "loss": 1.007,
      "step": 14453
    },
    {
      "epoch": 2.686117821966177,
      "grad_norm": 0.8801686763763428,
      "learning_rate": 4.885311495620296e-05,
      "loss": 1.0337,
      "step": 14454
    },
    {
      "epoch": 2.6863036610295485,
      "grad_norm": 0.6567339897155762,
      "learning_rate": 4.884056166892265e-05,
      "loss": 0.5863,
      "step": 14455
    },
    {
      "epoch": 2.6864895000929194,
      "grad_norm": 0.8444393873214722,
      "learning_rate": 4.882800947355274e-05,
      "loss": 1.0625,
      "step": 14456
    },
    {
      "epoch": 2.6866753391562908,
      "grad_norm": 0.8692553043365479,
      "learning_rate": 4.881545837036112e-05,
      "loss": 0.8933,
      "step": 14457
    },
    {
      "epoch": 2.6868611782196616,
      "grad_norm": 0.9198605418205261,
      "learning_rate": 4.880290835961574e-05,
      "loss": 0.9862,
      "step": 14458
    },
    {
      "epoch": 2.687047017283033,
      "grad_norm": 0.8040031790733337,
      "learning_rate": 4.879035944158443e-05,
      "loss": 0.9413,
      "step": 14459
    },
    {
      "epoch": 2.687232856346404,
      "grad_norm": 0.8228282928466797,
      "learning_rate": 4.8777811616534976e-05,
      "loss": 0.7872,
      "step": 14460
    },
    {
      "epoch": 2.687418695409775,
      "grad_norm": 0.8292063474655151,
      "learning_rate": 4.876526488473524e-05,
      "loss": 0.8434,
      "step": 14461
    },
    {
      "epoch": 2.687604534473146,
      "grad_norm": 0.9853034615516663,
      "learning_rate": 4.875271924645303e-05,
      "loss": 0.9802,
      "step": 14462
    },
    {
      "epoch": 2.6877903735365174,
      "grad_norm": 0.7087395191192627,
      "learning_rate": 4.8740174701956085e-05,
      "loss": 0.8545,
      "step": 14463
    },
    {
      "epoch": 2.6879762125998887,
      "grad_norm": 0.8959962725639343,
      "learning_rate": 4.872763125151215e-05,
      "loss": 0.8218,
      "step": 14464
    },
    {
      "epoch": 2.6881620516632596,
      "grad_norm": 1.5733524560928345,
      "learning_rate": 4.8715088895388904e-05,
      "loss": 1.4046,
      "step": 14465
    },
    {
      "epoch": 2.6883478907266305,
      "grad_norm": 0.7405545711517334,
      "learning_rate": 4.870254763385411e-05,
      "loss": 0.9044,
      "step": 14466
    },
    {
      "epoch": 2.688533729790002,
      "grad_norm": 0.7159562706947327,
      "learning_rate": 4.8690007467175414e-05,
      "loss": 0.9171,
      "step": 14467
    },
    {
      "epoch": 2.688719568853373,
      "grad_norm": 0.800383448600769,
      "learning_rate": 4.867746839562045e-05,
      "loss": 0.9989,
      "step": 14468
    },
    {
      "epoch": 2.688905407916744,
      "grad_norm": 0.7202253341674805,
      "learning_rate": 4.866493041945681e-05,
      "loss": 0.7756,
      "step": 14469
    },
    {
      "epoch": 2.6890912469801154,
      "grad_norm": 0.7838736772537231,
      "learning_rate": 4.8652393538952146e-05,
      "loss": 0.9914,
      "step": 14470
    },
    {
      "epoch": 2.6892770860434863,
      "grad_norm": 0.86966872215271,
      "learning_rate": 4.863985775437405e-05,
      "loss": 0.967,
      "step": 14471
    },
    {
      "epoch": 2.6894629251068576,
      "grad_norm": 0.8207091689109802,
      "learning_rate": 4.862732306599006e-05,
      "loss": 0.8466,
      "step": 14472
    },
    {
      "epoch": 2.6896487641702285,
      "grad_norm": 0.8837814331054688,
      "learning_rate": 4.861478947406767e-05,
      "loss": 0.975,
      "step": 14473
    },
    {
      "epoch": 2.6898346032336,
      "grad_norm": 0.8468865156173706,
      "learning_rate": 4.860225697887444e-05,
      "loss": 0.8451,
      "step": 14474
    },
    {
      "epoch": 2.6900204422969707,
      "grad_norm": 0.7788984775543213,
      "learning_rate": 4.8589725580677835e-05,
      "loss": 0.904,
      "step": 14475
    },
    {
      "epoch": 2.690206281360342,
      "grad_norm": 1.0463881492614746,
      "learning_rate": 4.857719527974532e-05,
      "loss": 0.9791,
      "step": 14476
    },
    {
      "epoch": 2.690392120423713,
      "grad_norm": 0.8855894804000854,
      "learning_rate": 4.856466607634429e-05,
      "loss": 1.1418,
      "step": 14477
    },
    {
      "epoch": 2.6905779594870842,
      "grad_norm": 0.7966851592063904,
      "learning_rate": 4.855213797074224e-05,
      "loss": 1.0921,
      "step": 14478
    },
    {
      "epoch": 2.6907637985504556,
      "grad_norm": 0.7545307874679565,
      "learning_rate": 4.8539610963206504e-05,
      "loss": 0.8605,
      "step": 14479
    },
    {
      "epoch": 2.6909496376138264,
      "grad_norm": 0.8396836519241333,
      "learning_rate": 4.852708505400447e-05,
      "loss": 0.6345,
      "step": 14480
    },
    {
      "epoch": 2.6911354766771973,
      "grad_norm": 0.9581857919692993,
      "learning_rate": 4.8514560243403396e-05,
      "loss": 1.1265,
      "step": 14481
    },
    {
      "epoch": 2.6913213157405687,
      "grad_norm": 0.8851617574691772,
      "learning_rate": 4.850203653167076e-05,
      "loss": 0.8745,
      "step": 14482
    },
    {
      "epoch": 2.69150715480394,
      "grad_norm": 0.9810107946395874,
      "learning_rate": 4.848951391907377e-05,
      "loss": 1.0509,
      "step": 14483
    },
    {
      "epoch": 2.691692993867311,
      "grad_norm": 0.8600289821624756,
      "learning_rate": 4.847699240587972e-05,
      "loss": 0.8622,
      "step": 14484
    },
    {
      "epoch": 2.6918788329306818,
      "grad_norm": 0.9160045385360718,
      "learning_rate": 4.8464471992355805e-05,
      "loss": 0.9755,
      "step": 14485
    },
    {
      "epoch": 2.692064671994053,
      "grad_norm": 0.7092505097389221,
      "learning_rate": 4.845195267876934e-05,
      "loss": 0.7566,
      "step": 14486
    },
    {
      "epoch": 2.6922505110574244,
      "grad_norm": 1.0010638236999512,
      "learning_rate": 4.843943446538749e-05,
      "loss": 0.9392,
      "step": 14487
    },
    {
      "epoch": 2.6924363501207953,
      "grad_norm": 0.8743418455123901,
      "learning_rate": 4.842691735247742e-05,
      "loss": 0.9309,
      "step": 14488
    },
    {
      "epoch": 2.6926221891841666,
      "grad_norm": 0.7750649452209473,
      "learning_rate": 4.841440134030627e-05,
      "loss": 0.8315,
      "step": 14489
    },
    {
      "epoch": 2.6928080282475375,
      "grad_norm": 0.8826529383659363,
      "learning_rate": 4.8401886429141216e-05,
      "loss": 0.9026,
      "step": 14490
    },
    {
      "epoch": 2.692993867310909,
      "grad_norm": 0.8705019354820251,
      "learning_rate": 4.8389372619249326e-05,
      "loss": 1.0637,
      "step": 14491
    },
    {
      "epoch": 2.6931797063742797,
      "grad_norm": 0.861504852771759,
      "learning_rate": 4.8376859910897756e-05,
      "loss": 0.9621,
      "step": 14492
    },
    {
      "epoch": 2.693365545437651,
      "grad_norm": 0.8101507425308228,
      "learning_rate": 4.836434830435348e-05,
      "loss": 0.8763,
      "step": 14493
    },
    {
      "epoch": 2.6935513845010224,
      "grad_norm": 0.943341076374054,
      "learning_rate": 4.8351837799883605e-05,
      "loss": 0.7872,
      "step": 14494
    },
    {
      "epoch": 2.6937372235643933,
      "grad_norm": 0.8089012503623962,
      "learning_rate": 4.8339328397755135e-05,
      "loss": 0.7785,
      "step": 14495
    },
    {
      "epoch": 2.693923062627764,
      "grad_norm": 0.9210566878318787,
      "learning_rate": 4.832682009823504e-05,
      "loss": 1.0245,
      "step": 14496
    },
    {
      "epoch": 2.6941089016911355,
      "grad_norm": 0.8864029049873352,
      "learning_rate": 4.831431290159025e-05,
      "loss": 1.0957,
      "step": 14497
    },
    {
      "epoch": 2.694294740754507,
      "grad_norm": 0.8884971737861633,
      "learning_rate": 4.8301806808087804e-05,
      "loss": 0.8855,
      "step": 14498
    },
    {
      "epoch": 2.6944805798178777,
      "grad_norm": 0.8697060942649841,
      "learning_rate": 4.828930181799457e-05,
      "loss": 1.0415,
      "step": 14499
    },
    {
      "epoch": 2.6946664188812486,
      "grad_norm": 0.708972156047821,
      "learning_rate": 4.827679793157746e-05,
      "loss": 0.8836,
      "step": 14500
    },
    {
      "epoch": 2.69485225794462,
      "grad_norm": 0.946927547454834,
      "learning_rate": 4.8264295149103275e-05,
      "loss": 1.2003,
      "step": 14501
    },
    {
      "epoch": 2.6950380970079912,
      "grad_norm": 0.9309900999069214,
      "learning_rate": 4.825179347083895e-05,
      "loss": 0.8666,
      "step": 14502
    },
    {
      "epoch": 2.695223936071362,
      "grad_norm": 0.7605841159820557,
      "learning_rate": 4.823929289705132e-05,
      "loss": 0.7412,
      "step": 14503
    },
    {
      "epoch": 2.6954097751347335,
      "grad_norm": 0.7765486836433411,
      "learning_rate": 4.8226793428007156e-05,
      "loss": 0.9386,
      "step": 14504
    },
    {
      "epoch": 2.6955956141981043,
      "grad_norm": 0.9206398129463196,
      "learning_rate": 4.82142950639732e-05,
      "loss": 0.8607,
      "step": 14505
    },
    {
      "epoch": 2.6957814532614757,
      "grad_norm": 0.7876893281936646,
      "learning_rate": 4.8201797805216294e-05,
      "loss": 0.8987,
      "step": 14506
    },
    {
      "epoch": 2.6959672923248466,
      "grad_norm": 0.8226445913314819,
      "learning_rate": 4.818930165200313e-05,
      "loss": 0.7574,
      "step": 14507
    },
    {
      "epoch": 2.696153131388218,
      "grad_norm": 0.9933581352233887,
      "learning_rate": 4.81768066046004e-05,
      "loss": 0.8114,
      "step": 14508
    },
    {
      "epoch": 2.6963389704515888,
      "grad_norm": 0.7784883975982666,
      "learning_rate": 4.816431266327477e-05,
      "loss": 1.0112,
      "step": 14509
    },
    {
      "epoch": 2.69652480951496,
      "grad_norm": 0.7937835454940796,
      "learning_rate": 4.8151819828292954e-05,
      "loss": 0.5749,
      "step": 14510
    },
    {
      "epoch": 2.696710648578331,
      "grad_norm": 0.7758409380912781,
      "learning_rate": 4.813932809992159e-05,
      "loss": 0.8044,
      "step": 14511
    },
    {
      "epoch": 2.6968964876417023,
      "grad_norm": 0.7109744548797607,
      "learning_rate": 4.8126837478427223e-05,
      "loss": 0.7163,
      "step": 14512
    },
    {
      "epoch": 2.6970823267050736,
      "grad_norm": 0.8645885586738586,
      "learning_rate": 4.811434796407649e-05,
      "loss": 1.0301,
      "step": 14513
    },
    {
      "epoch": 2.6972681657684445,
      "grad_norm": 0.9839845299720764,
      "learning_rate": 4.8101859557136e-05,
      "loss": 0.9307,
      "step": 14514
    },
    {
      "epoch": 2.6974540048318154,
      "grad_norm": 0.8526188135147095,
      "learning_rate": 4.8089372257872266e-05,
      "loss": 0.7929,
      "step": 14515
    },
    {
      "epoch": 2.6976398438951867,
      "grad_norm": 0.746585488319397,
      "learning_rate": 4.807688606655179e-05,
      "loss": 0.9063,
      "step": 14516
    },
    {
      "epoch": 2.697825682958558,
      "grad_norm": 0.8589296936988831,
      "learning_rate": 4.806440098344104e-05,
      "loss": 0.9476,
      "step": 14517
    },
    {
      "epoch": 2.698011522021929,
      "grad_norm": 0.7648781538009644,
      "learning_rate": 4.805191700880657e-05,
      "loss": 0.7917,
      "step": 14518
    },
    {
      "epoch": 2.6981973610853003,
      "grad_norm": 0.8353559374809265,
      "learning_rate": 4.803943414291477e-05,
      "loss": 0.9623,
      "step": 14519
    },
    {
      "epoch": 2.698383200148671,
      "grad_norm": 0.915885329246521,
      "learning_rate": 4.8026952386032096e-05,
      "loss": 0.8834,
      "step": 14520
    },
    {
      "epoch": 2.6985690392120425,
      "grad_norm": 0.9126646518707275,
      "learning_rate": 4.8014471738424885e-05,
      "loss": 0.9102,
      "step": 14521
    },
    {
      "epoch": 2.6987548782754134,
      "grad_norm": 1.121840476989746,
      "learning_rate": 4.8001992200359616e-05,
      "loss": 1.0407,
      "step": 14522
    },
    {
      "epoch": 2.6989407173387847,
      "grad_norm": 0.8188127279281616,
      "learning_rate": 4.7989513772102537e-05,
      "loss": 1.1356,
      "step": 14523
    },
    {
      "epoch": 2.6991265564021556,
      "grad_norm": 0.8660606145858765,
      "learning_rate": 4.7977036453920074e-05,
      "loss": 0.9382,
      "step": 14524
    },
    {
      "epoch": 2.699312395465527,
      "grad_norm": 0.9244834780693054,
      "learning_rate": 4.796456024607846e-05,
      "loss": 0.9308,
      "step": 14525
    },
    {
      "epoch": 2.699498234528898,
      "grad_norm": 0.7831543684005737,
      "learning_rate": 4.795208514884405e-05,
      "loss": 0.9356,
      "step": 14526
    },
    {
      "epoch": 2.699684073592269,
      "grad_norm": 0.7805042862892151,
      "learning_rate": 4.7939611162483065e-05,
      "loss": 0.8295,
      "step": 14527
    },
    {
      "epoch": 2.6998699126556405,
      "grad_norm": 0.9255910515785217,
      "learning_rate": 4.792713828726174e-05,
      "loss": 1.0148,
      "step": 14528
    },
    {
      "epoch": 2.7000557517190114,
      "grad_norm": 0.8160353302955627,
      "learning_rate": 4.791466652344625e-05,
      "loss": 0.8816,
      "step": 14529
    },
    {
      "epoch": 2.7002415907823822,
      "grad_norm": 0.8323808908462524,
      "learning_rate": 4.790219587130286e-05,
      "loss": 0.9375,
      "step": 14530
    },
    {
      "epoch": 2.7004274298457536,
      "grad_norm": 0.8540868163108826,
      "learning_rate": 4.7889726331097686e-05,
      "loss": 0.9339,
      "step": 14531
    },
    {
      "epoch": 2.700613268909125,
      "grad_norm": 0.8291054368019104,
      "learning_rate": 4.787725790309685e-05,
      "loss": 0.8972,
      "step": 14532
    },
    {
      "epoch": 2.700799107972496,
      "grad_norm": 1.0824410915374756,
      "learning_rate": 4.78647905875665e-05,
      "loss": 0.9822,
      "step": 14533
    },
    {
      "epoch": 2.7009849470358667,
      "grad_norm": 0.8491551876068115,
      "learning_rate": 4.785232438477276e-05,
      "loss": 0.9835,
      "step": 14534
    },
    {
      "epoch": 2.701170786099238,
      "grad_norm": 0.8033833503723145,
      "learning_rate": 4.783985929498166e-05,
      "loss": 0.9216,
      "step": 14535
    },
    {
      "epoch": 2.7013566251626093,
      "grad_norm": 1.0135128498077393,
      "learning_rate": 4.782739531845927e-05,
      "loss": 0.8598,
      "step": 14536
    },
    {
      "epoch": 2.70154246422598,
      "grad_norm": 0.950504720211029,
      "learning_rate": 4.781493245547154e-05,
      "loss": 0.8575,
      "step": 14537
    },
    {
      "epoch": 2.7017283032893515,
      "grad_norm": 0.7731863260269165,
      "learning_rate": 4.780247070628457e-05,
      "loss": 0.9419,
      "step": 14538
    },
    {
      "epoch": 2.7019141423527224,
      "grad_norm": 0.8756670951843262,
      "learning_rate": 4.7790010071164295e-05,
      "loss": 1.0453,
      "step": 14539
    },
    {
      "epoch": 2.7020999814160938,
      "grad_norm": 0.9157370924949646,
      "learning_rate": 4.777755055037665e-05,
      "loss": 0.8694,
      "step": 14540
    },
    {
      "epoch": 2.7022858204794646,
      "grad_norm": 0.8106825947761536,
      "learning_rate": 4.776509214418753e-05,
      "loss": 0.903,
      "step": 14541
    },
    {
      "epoch": 2.702471659542836,
      "grad_norm": 0.81466144323349,
      "learning_rate": 4.775263485286292e-05,
      "loss": 0.97,
      "step": 14542
    },
    {
      "epoch": 2.7026574986062073,
      "grad_norm": 0.7218373417854309,
      "learning_rate": 4.774017867666867e-05,
      "loss": 0.7417,
      "step": 14543
    },
    {
      "epoch": 2.702843337669578,
      "grad_norm": 0.6913556456565857,
      "learning_rate": 4.772772361587058e-05,
      "loss": 0.7463,
      "step": 14544
    },
    {
      "epoch": 2.703029176732949,
      "grad_norm": 0.776649534702301,
      "learning_rate": 4.771526967073452e-05,
      "loss": 0.9157,
      "step": 14545
    },
    {
      "epoch": 2.7032150157963204,
      "grad_norm": 0.7431927919387817,
      "learning_rate": 4.770281684152634e-05,
      "loss": 0.9396,
      "step": 14546
    },
    {
      "epoch": 2.7034008548596917,
      "grad_norm": 0.9437835216522217,
      "learning_rate": 4.76903651285118e-05,
      "loss": 0.7785,
      "step": 14547
    },
    {
      "epoch": 2.7035866939230626,
      "grad_norm": 0.7822114825248718,
      "learning_rate": 4.7677914531956645e-05,
      "loss": 0.9485,
      "step": 14548
    },
    {
      "epoch": 2.7037725329864335,
      "grad_norm": 0.7751176357269287,
      "learning_rate": 4.766546505212658e-05,
      "loss": 0.899,
      "step": 14549
    },
    {
      "epoch": 2.703958372049805,
      "grad_norm": 0.8520172834396362,
      "learning_rate": 4.76530166892874e-05,
      "loss": 0.8291,
      "step": 14550
    },
    {
      "epoch": 2.704144211113176,
      "grad_norm": 0.8314445614814758,
      "learning_rate": 4.764056944370473e-05,
      "loss": 0.6832,
      "step": 14551
    },
    {
      "epoch": 2.704330050176547,
      "grad_norm": 0.7597737908363342,
      "learning_rate": 4.762812331564422e-05,
      "loss": 0.9719,
      "step": 14552
    },
    {
      "epoch": 2.7045158892399184,
      "grad_norm": 0.8481391072273254,
      "learning_rate": 4.761567830537159e-05,
      "loss": 0.9278,
      "step": 14553
    },
    {
      "epoch": 2.7047017283032893,
      "grad_norm": 0.8097413182258606,
      "learning_rate": 4.760323441315237e-05,
      "loss": 0.8921,
      "step": 14554
    },
    {
      "epoch": 2.7048875673666606,
      "grad_norm": 0.6693623661994934,
      "learning_rate": 4.759079163925223e-05,
      "loss": 0.6959,
      "step": 14555
    },
    {
      "epoch": 2.7050734064300315,
      "grad_norm": 0.7500324249267578,
      "learning_rate": 4.757834998393671e-05,
      "loss": 0.7453,
      "step": 14556
    },
    {
      "epoch": 2.705259245493403,
      "grad_norm": 0.9771433472633362,
      "learning_rate": 4.7565909447471313e-05,
      "loss": 0.9639,
      "step": 14557
    },
    {
      "epoch": 2.7054450845567737,
      "grad_norm": 0.8323385119438171,
      "learning_rate": 4.7553470030121626e-05,
      "loss": 0.7888,
      "step": 14558
    },
    {
      "epoch": 2.705630923620145,
      "grad_norm": 0.7277812957763672,
      "learning_rate": 4.754103173215313e-05,
      "loss": 0.7471,
      "step": 14559
    },
    {
      "epoch": 2.705816762683516,
      "grad_norm": 0.8828470706939697,
      "learning_rate": 4.752859455383129e-05,
      "loss": 0.9009,
      "step": 14560
    },
    {
      "epoch": 2.7060026017468872,
      "grad_norm": 0.7129109501838684,
      "learning_rate": 4.75161584954215e-05,
      "loss": 0.7362,
      "step": 14561
    },
    {
      "epoch": 2.7061884408102586,
      "grad_norm": 0.7221847772598267,
      "learning_rate": 4.75037235571893e-05,
      "loss": 0.941,
      "step": 14562
    },
    {
      "epoch": 2.7063742798736294,
      "grad_norm": 0.9126443266868591,
      "learning_rate": 4.749128973940001e-05,
      "loss": 0.9747,
      "step": 14563
    },
    {
      "epoch": 2.7065601189370003,
      "grad_norm": 1.0719454288482666,
      "learning_rate": 4.747885704231901e-05,
      "loss": 0.9455,
      "step": 14564
    },
    {
      "epoch": 2.7067459580003717,
      "grad_norm": 0.6202034950256348,
      "learning_rate": 4.746642546621167e-05,
      "loss": 0.7132,
      "step": 14565
    },
    {
      "epoch": 2.706931797063743,
      "grad_norm": 0.8351492881774902,
      "learning_rate": 4.745399501134338e-05,
      "loss": 0.968,
      "step": 14566
    },
    {
      "epoch": 2.707117636127114,
      "grad_norm": 0.8512027859687805,
      "learning_rate": 4.744156567797938e-05,
      "loss": 0.845,
      "step": 14567
    },
    {
      "epoch": 2.707303475190485,
      "grad_norm": 0.8470203876495361,
      "learning_rate": 4.7429137466384964e-05,
      "loss": 0.9657,
      "step": 14568
    },
    {
      "epoch": 2.707489314253856,
      "grad_norm": 0.7456303834915161,
      "learning_rate": 4.741671037682537e-05,
      "loss": 0.8667,
      "step": 14569
    },
    {
      "epoch": 2.7076751533172274,
      "grad_norm": 1.0743417739868164,
      "learning_rate": 4.740428440956588e-05,
      "loss": 1.013,
      "step": 14570
    },
    {
      "epoch": 2.7078609923805983,
      "grad_norm": 0.7952508330345154,
      "learning_rate": 4.739185956487169e-05,
      "loss": 0.846,
      "step": 14571
    },
    {
      "epoch": 2.7080468314439696,
      "grad_norm": 0.9333015084266663,
      "learning_rate": 4.737943584300793e-05,
      "loss": 0.6875,
      "step": 14572
    },
    {
      "epoch": 2.7082326705073405,
      "grad_norm": 0.8915261030197144,
      "learning_rate": 4.736701324423987e-05,
      "loss": 1.0861,
      "step": 14573
    },
    {
      "epoch": 2.708418509570712,
      "grad_norm": 0.8243820071220398,
      "learning_rate": 4.7354591768832566e-05,
      "loss": 0.8534,
      "step": 14574
    },
    {
      "epoch": 2.7086043486340827,
      "grad_norm": 0.88190758228302,
      "learning_rate": 4.734217141705114e-05,
      "loss": 1.1287,
      "step": 14575
    },
    {
      "epoch": 2.708790187697454,
      "grad_norm": 0.9980376362800598,
      "learning_rate": 4.732975218916072e-05,
      "loss": 0.8787,
      "step": 14576
    },
    {
      "epoch": 2.7089760267608254,
      "grad_norm": 0.8127431273460388,
      "learning_rate": 4.731733408542634e-05,
      "loss": 0.9969,
      "step": 14577
    },
    {
      "epoch": 2.7091618658241963,
      "grad_norm": 0.8898497819900513,
      "learning_rate": 4.7304917106113064e-05,
      "loss": 0.7366,
      "step": 14578
    },
    {
      "epoch": 2.709347704887567,
      "grad_norm": 0.861444354057312,
      "learning_rate": 4.729250125148592e-05,
      "loss": 1.0436,
      "step": 14579
    },
    {
      "epoch": 2.7095335439509385,
      "grad_norm": 0.8095132112503052,
      "learning_rate": 4.7280086521809894e-05,
      "loss": 0.8042,
      "step": 14580
    },
    {
      "epoch": 2.70971938301431,
      "grad_norm": 0.8140536546707153,
      "learning_rate": 4.72676729173499e-05,
      "loss": 1.058,
      "step": 14581
    },
    {
      "epoch": 2.7099052220776807,
      "grad_norm": 0.8666590452194214,
      "learning_rate": 4.7255260438370984e-05,
      "loss": 1.0129,
      "step": 14582
    },
    {
      "epoch": 2.7100910611410516,
      "grad_norm": 0.8712512850761414,
      "learning_rate": 4.724284908513802e-05,
      "loss": 1.0081,
      "step": 14583
    },
    {
      "epoch": 2.710276900204423,
      "grad_norm": 0.736910343170166,
      "learning_rate": 4.723043885791586e-05,
      "loss": 0.5821,
      "step": 14584
    },
    {
      "epoch": 2.7104627392677942,
      "grad_norm": 0.7968629002571106,
      "learning_rate": 4.721802975696947e-05,
      "loss": 0.8978,
      "step": 14585
    },
    {
      "epoch": 2.710648578331165,
      "grad_norm": 0.762345552444458,
      "learning_rate": 4.7205621782563614e-05,
      "loss": 0.8701,
      "step": 14586
    },
    {
      "epoch": 2.7108344173945365,
      "grad_norm": 0.9853722453117371,
      "learning_rate": 4.7193214934963206e-05,
      "loss": 0.8083,
      "step": 14587
    },
    {
      "epoch": 2.7110202564579073,
      "grad_norm": 0.8430361151695251,
      "learning_rate": 4.718080921443301e-05,
      "loss": 0.9613,
      "step": 14588
    },
    {
      "epoch": 2.7112060955212787,
      "grad_norm": 0.7686344385147095,
      "learning_rate": 4.7168404621237746e-05,
      "loss": 0.9783,
      "step": 14589
    },
    {
      "epoch": 2.7113919345846496,
      "grad_norm": 0.7804996967315674,
      "learning_rate": 4.7156001155642273e-05,
      "loss": 1.1721,
      "step": 14590
    },
    {
      "epoch": 2.711577773648021,
      "grad_norm": 1.0992109775543213,
      "learning_rate": 4.714359881791127e-05,
      "loss": 1.1223,
      "step": 14591
    },
    {
      "epoch": 2.7117636127113918,
      "grad_norm": 0.7722237706184387,
      "learning_rate": 4.713119760830941e-05,
      "loss": 1.0712,
      "step": 14592
    },
    {
      "epoch": 2.711949451774763,
      "grad_norm": 0.7525615692138672,
      "learning_rate": 4.7118797527101446e-05,
      "loss": 0.846,
      "step": 14593
    },
    {
      "epoch": 2.712135290838134,
      "grad_norm": 0.7216577529907227,
      "learning_rate": 4.710639857455199e-05,
      "loss": 0.711,
      "step": 14594
    },
    {
      "epoch": 2.7123211299015053,
      "grad_norm": 2.185736656188965,
      "learning_rate": 4.70940007509257e-05,
      "loss": 1.5017,
      "step": 14595
    },
    {
      "epoch": 2.7125069689648766,
      "grad_norm": 0.7739629149436951,
      "learning_rate": 4.708160405648713e-05,
      "loss": 0.8468,
      "step": 14596
    },
    {
      "epoch": 2.7126928080282475,
      "grad_norm": 0.8368583917617798,
      "learning_rate": 4.706920849150092e-05,
      "loss": 0.7949,
      "step": 14597
    },
    {
      "epoch": 2.7128786470916184,
      "grad_norm": 0.8572446703910828,
      "learning_rate": 4.705681405623165e-05,
      "loss": 0.9776,
      "step": 14598
    },
    {
      "epoch": 2.7130644861549897,
      "grad_norm": 0.926463782787323,
      "learning_rate": 4.704442075094383e-05,
      "loss": 0.9535,
      "step": 14599
    },
    {
      "epoch": 2.713250325218361,
      "grad_norm": 0.7697812914848328,
      "learning_rate": 4.703202857590198e-05,
      "loss": 1.0249,
      "step": 14600
    },
    {
      "epoch": 2.713436164281732,
      "grad_norm": 0.8938848376274109,
      "learning_rate": 4.701963753137053e-05,
      "loss": 0.8571,
      "step": 14601
    },
    {
      "epoch": 2.7136220033451033,
      "grad_norm": 0.7829850316047668,
      "learning_rate": 4.700724761761405e-05,
      "loss": 1.0279,
      "step": 14602
    },
    {
      "epoch": 2.713807842408474,
      "grad_norm": 0.8015586137771606,
      "learning_rate": 4.699485883489693e-05,
      "loss": 1.1516,
      "step": 14603
    },
    {
      "epoch": 2.7139936814718455,
      "grad_norm": 0.8683645725250244,
      "learning_rate": 4.6982471183483545e-05,
      "loss": 0.7275,
      "step": 14604
    },
    {
      "epoch": 2.7141795205352164,
      "grad_norm": 0.7575713992118835,
      "learning_rate": 4.697008466363837e-05,
      "loss": 0.9761,
      "step": 14605
    },
    {
      "epoch": 2.7143653595985877,
      "grad_norm": 0.7274417877197266,
      "learning_rate": 4.6957699275625744e-05,
      "loss": 1.0053,
      "step": 14606
    },
    {
      "epoch": 2.7145511986619586,
      "grad_norm": 0.9730561375617981,
      "learning_rate": 4.6945315019709956e-05,
      "loss": 1.051,
      "step": 14607
    },
    {
      "epoch": 2.71473703772533,
      "grad_norm": 0.7930693030357361,
      "learning_rate": 4.693293189615541e-05,
      "loss": 1.1249,
      "step": 14608
    },
    {
      "epoch": 2.714922876788701,
      "grad_norm": 0.877459704875946,
      "learning_rate": 4.6920549905226326e-05,
      "loss": 0.9589,
      "step": 14609
    },
    {
      "epoch": 2.715108715852072,
      "grad_norm": 0.6798363924026489,
      "learning_rate": 4.690816904718707e-05,
      "loss": 0.7238,
      "step": 14610
    },
    {
      "epoch": 2.7152945549154435,
      "grad_norm": 1.2394379377365112,
      "learning_rate": 4.689578932230182e-05,
      "loss": 1.0593,
      "step": 14611
    },
    {
      "epoch": 2.7154803939788144,
      "grad_norm": 0.9165498614311218,
      "learning_rate": 4.6883410730834785e-05,
      "loss": 1.0986,
      "step": 14612
    },
    {
      "epoch": 2.7156662330421852,
      "grad_norm": 0.9275784492492676,
      "learning_rate": 4.6871033273050235e-05,
      "loss": 0.8614,
      "step": 14613
    },
    {
      "epoch": 2.7158520721055566,
      "grad_norm": 0.7050016522407532,
      "learning_rate": 4.68586569492123e-05,
      "loss": 0.8167,
      "step": 14614
    },
    {
      "epoch": 2.716037911168928,
      "grad_norm": 0.8385757207870483,
      "learning_rate": 4.6846281759585144e-05,
      "loss": 0.9408,
      "step": 14615
    },
    {
      "epoch": 2.716223750232299,
      "grad_norm": 0.8483341932296753,
      "learning_rate": 4.683390770443284e-05,
      "loss": 0.9463,
      "step": 14616
    },
    {
      "epoch": 2.71640958929567,
      "grad_norm": 1.039903163909912,
      "learning_rate": 4.682153478401956e-05,
      "loss": 0.9554,
      "step": 14617
    },
    {
      "epoch": 2.716595428359041,
      "grad_norm": 0.9295901656150818,
      "learning_rate": 4.680916299860939e-05,
      "loss": 0.8416,
      "step": 14618
    },
    {
      "epoch": 2.7167812674224123,
      "grad_norm": 0.7862735986709595,
      "learning_rate": 4.6796792348466356e-05,
      "loss": 1.0495,
      "step": 14619
    },
    {
      "epoch": 2.716967106485783,
      "grad_norm": 0.8048015236854553,
      "learning_rate": 4.678442283385448e-05,
      "loss": 0.7554,
      "step": 14620
    },
    {
      "epoch": 2.7171529455491545,
      "grad_norm": 0.8309585452079773,
      "learning_rate": 4.677205445503775e-05,
      "loss": 0.9189,
      "step": 14621
    },
    {
      "epoch": 2.7173387846125254,
      "grad_norm": 0.7203308343887329,
      "learning_rate": 4.675968721228021e-05,
      "loss": 1.0408,
      "step": 14622
    },
    {
      "epoch": 2.7175246236758968,
      "grad_norm": 0.8534119725227356,
      "learning_rate": 4.674732110584579e-05,
      "loss": 0.7916,
      "step": 14623
    },
    {
      "epoch": 2.7177104627392676,
      "grad_norm": 0.8853223323822021,
      "learning_rate": 4.673495613599836e-05,
      "loss": 1.0103,
      "step": 14624
    },
    {
      "epoch": 2.717896301802639,
      "grad_norm": 0.8163644671440125,
      "learning_rate": 4.6722592303001944e-05,
      "loss": 1.0153,
      "step": 14625
    },
    {
      "epoch": 2.7180821408660103,
      "grad_norm": 0.8186414241790771,
      "learning_rate": 4.671022960712035e-05,
      "loss": 0.8593,
      "step": 14626
    },
    {
      "epoch": 2.718267979929381,
      "grad_norm": 0.8188729286193848,
      "learning_rate": 4.6697868048617467e-05,
      "loss": 0.8212,
      "step": 14627
    },
    {
      "epoch": 2.718453818992752,
      "grad_norm": 0.8711169958114624,
      "learning_rate": 4.6685507627757074e-05,
      "loss": 1.0026,
      "step": 14628
    },
    {
      "epoch": 2.7186396580561234,
      "grad_norm": 0.8931276798248291,
      "learning_rate": 4.667314834480304e-05,
      "loss": 0.8006,
      "step": 14629
    },
    {
      "epoch": 2.7188254971194947,
      "grad_norm": 0.9605731964111328,
      "learning_rate": 4.6660790200019165e-05,
      "loss": 1.0555,
      "step": 14630
    },
    {
      "epoch": 2.7190113361828656,
      "grad_norm": 0.8598033785820007,
      "learning_rate": 4.664843319366921e-05,
      "loss": 0.8958,
      "step": 14631
    },
    {
      "epoch": 2.7191971752462365,
      "grad_norm": 0.8986445069313049,
      "learning_rate": 4.663607732601684e-05,
      "loss": 0.9925,
      "step": 14632
    },
    {
      "epoch": 2.719383014309608,
      "grad_norm": 0.7430914640426636,
      "learning_rate": 4.662372259732587e-05,
      "loss": 1.013,
      "step": 14633
    },
    {
      "epoch": 2.719568853372979,
      "grad_norm": 0.7768040299415588,
      "learning_rate": 4.661136900785993e-05,
      "loss": 0.6541,
      "step": 14634
    },
    {
      "epoch": 2.71975469243635,
      "grad_norm": 0.8109592795372009,
      "learning_rate": 4.6599016557882715e-05,
      "loss": 0.8924,
      "step": 14635
    },
    {
      "epoch": 2.7199405314997214,
      "grad_norm": 0.8567116260528564,
      "learning_rate": 4.65866652476578e-05,
      "loss": 0.5437,
      "step": 14636
    },
    {
      "epoch": 2.7201263705630923,
      "grad_norm": 0.8742349743843079,
      "learning_rate": 4.6574315077448905e-05,
      "loss": 0.9358,
      "step": 14637
    },
    {
      "epoch": 2.7203122096264636,
      "grad_norm": 0.8746808767318726,
      "learning_rate": 4.656196604751954e-05,
      "loss": 0.9686,
      "step": 14638
    },
    {
      "epoch": 2.7204980486898345,
      "grad_norm": 0.9990546107292175,
      "learning_rate": 4.654961815813332e-05,
      "loss": 0.896,
      "step": 14639
    },
    {
      "epoch": 2.720683887753206,
      "grad_norm": 0.7727985382080078,
      "learning_rate": 4.653727140955379e-05,
      "loss": 0.9727,
      "step": 14640
    },
    {
      "epoch": 2.7208697268165767,
      "grad_norm": 0.6992377042770386,
      "learning_rate": 4.652492580204442e-05,
      "loss": 0.7702,
      "step": 14641
    },
    {
      "epoch": 2.721055565879948,
      "grad_norm": 0.8959800601005554,
      "learning_rate": 4.651258133586878e-05,
      "loss": 0.97,
      "step": 14642
    },
    {
      "epoch": 2.721241404943319,
      "grad_norm": 0.8036636710166931,
      "learning_rate": 4.6500238011290295e-05,
      "loss": 1.0769,
      "step": 14643
    },
    {
      "epoch": 2.7214272440066902,
      "grad_norm": 0.8741715550422668,
      "learning_rate": 4.648789582857239e-05,
      "loss": 1.1427,
      "step": 14644
    },
    {
      "epoch": 2.7216130830700616,
      "grad_norm": 0.8315646648406982,
      "learning_rate": 4.6475554787978556e-05,
      "loss": 0.8081,
      "step": 14645
    },
    {
      "epoch": 2.7217989221334324,
      "grad_norm": 0.9681646823883057,
      "learning_rate": 4.6463214889772154e-05,
      "loss": 0.8694,
      "step": 14646
    },
    {
      "epoch": 2.7219847611968033,
      "grad_norm": 0.8161425590515137,
      "learning_rate": 4.645087613421656e-05,
      "loss": 0.9062,
      "step": 14647
    },
    {
      "epoch": 2.7221706002601747,
      "grad_norm": 0.8496885299682617,
      "learning_rate": 4.643853852157507e-05,
      "loss": 0.767,
      "step": 14648
    },
    {
      "epoch": 2.722356439323546,
      "grad_norm": 1.285094141960144,
      "learning_rate": 4.6426202052111064e-05,
      "loss": 1.2346,
      "step": 14649
    },
    {
      "epoch": 2.722542278386917,
      "grad_norm": 0.8886316418647766,
      "learning_rate": 4.6413866726087884e-05,
      "loss": 0.7106,
      "step": 14650
    },
    {
      "epoch": 2.722728117450288,
      "grad_norm": 1.214694619178772,
      "learning_rate": 4.6401532543768766e-05,
      "loss": 0.8667,
      "step": 14651
    },
    {
      "epoch": 2.722913956513659,
      "grad_norm": 0.8256045579910278,
      "learning_rate": 4.638919950541691e-05,
      "loss": 1.0791,
      "step": 14652
    },
    {
      "epoch": 2.7230997955770304,
      "grad_norm": 0.8462839126586914,
      "learning_rate": 4.6376867611295624e-05,
      "loss": 0.927,
      "step": 14653
    },
    {
      "epoch": 2.7232856346404013,
      "grad_norm": 0.8731264472007751,
      "learning_rate": 4.636453686166808e-05,
      "loss": 1.161,
      "step": 14654
    },
    {
      "epoch": 2.7234714737037726,
      "grad_norm": 1.0182912349700928,
      "learning_rate": 4.6352207256797464e-05,
      "loss": 0.9539,
      "step": 14655
    },
    {
      "epoch": 2.7236573127671435,
      "grad_norm": 0.8832132816314697,
      "learning_rate": 4.633987879694688e-05,
      "loss": 1.0213,
      "step": 14656
    },
    {
      "epoch": 2.723843151830515,
      "grad_norm": 0.8857216835021973,
      "learning_rate": 4.632755148237954e-05,
      "loss": 1.0354,
      "step": 14657
    },
    {
      "epoch": 2.7240289908938857,
      "grad_norm": 0.7460151314735413,
      "learning_rate": 4.63152253133585e-05,
      "loss": 0.8688,
      "step": 14658
    },
    {
      "epoch": 2.724214829957257,
      "grad_norm": 0.8710841536521912,
      "learning_rate": 4.630290029014681e-05,
      "loss": 0.8975,
      "step": 14659
    },
    {
      "epoch": 2.7244006690206284,
      "grad_norm": 0.7464547753334045,
      "learning_rate": 4.62905764130076e-05,
      "loss": 0.5561,
      "step": 14660
    },
    {
      "epoch": 2.7245865080839993,
      "grad_norm": 1.018832802772522,
      "learning_rate": 4.6278253682203844e-05,
      "loss": 0.9404,
      "step": 14661
    },
    {
      "epoch": 2.72477234714737,
      "grad_norm": 0.7565674781799316,
      "learning_rate": 4.6265932097998596e-05,
      "loss": 0.8725,
      "step": 14662
    },
    {
      "epoch": 2.7249581862107415,
      "grad_norm": 0.9171304106712341,
      "learning_rate": 4.625361166065483e-05,
      "loss": 0.8651,
      "step": 14663
    },
    {
      "epoch": 2.725144025274113,
      "grad_norm": 0.9045332074165344,
      "learning_rate": 4.624129237043544e-05,
      "loss": 0.9935,
      "step": 14664
    },
    {
      "epoch": 2.7253298643374837,
      "grad_norm": 0.8097667098045349,
      "learning_rate": 4.622897422760346e-05,
      "loss": 0.9527,
      "step": 14665
    },
    {
      "epoch": 2.7255157034008546,
      "grad_norm": 0.8182275891304016,
      "learning_rate": 4.6216657232421744e-05,
      "loss": 0.9202,
      "step": 14666
    },
    {
      "epoch": 2.725701542464226,
      "grad_norm": 0.9292824864387512,
      "learning_rate": 4.6204341385153186e-05,
      "loss": 0.7853,
      "step": 14667
    },
    {
      "epoch": 2.7258873815275972,
      "grad_norm": 0.900060772895813,
      "learning_rate": 4.6192026686060596e-05,
      "loss": 1.0285,
      "step": 14668
    },
    {
      "epoch": 2.726073220590968,
      "grad_norm": 0.8207162022590637,
      "learning_rate": 4.617971313540691e-05,
      "loss": 0.884,
      "step": 14669
    },
    {
      "epoch": 2.7262590596543395,
      "grad_norm": 0.6432229280471802,
      "learning_rate": 4.616740073345485e-05,
      "loss": 0.6324,
      "step": 14670
    },
    {
      "epoch": 2.7264448987177103,
      "grad_norm": 0.7894888520240784,
      "learning_rate": 4.615508948046726e-05,
      "loss": 0.8774,
      "step": 14671
    },
    {
      "epoch": 2.7266307377810817,
      "grad_norm": 0.7836933135986328,
      "learning_rate": 4.614277937670687e-05,
      "loss": 0.7802,
      "step": 14672
    },
    {
      "epoch": 2.7268165768444526,
      "grad_norm": 0.7634031176567078,
      "learning_rate": 4.613047042243646e-05,
      "loss": 0.7226,
      "step": 14673
    },
    {
      "epoch": 2.727002415907824,
      "grad_norm": 0.7131878733634949,
      "learning_rate": 4.6118162617918705e-05,
      "loss": 0.8122,
      "step": 14674
    },
    {
      "epoch": 2.727188254971195,
      "grad_norm": 0.6382018327713013,
      "learning_rate": 4.610585596341631e-05,
      "loss": 0.7506,
      "step": 14675
    },
    {
      "epoch": 2.727374094034566,
      "grad_norm": 0.8915993571281433,
      "learning_rate": 4.60935504591919e-05,
      "loss": 0.8766,
      "step": 14676
    },
    {
      "epoch": 2.727559933097937,
      "grad_norm": 0.7661280035972595,
      "learning_rate": 4.608124610550818e-05,
      "loss": 1.0099,
      "step": 14677
    },
    {
      "epoch": 2.7277457721613083,
      "grad_norm": 0.9038292765617371,
      "learning_rate": 4.606894290262774e-05,
      "loss": 1.0057,
      "step": 14678
    },
    {
      "epoch": 2.7279316112246796,
      "grad_norm": 0.833634078502655,
      "learning_rate": 4.605664085081316e-05,
      "loss": 0.7077,
      "step": 14679
    },
    {
      "epoch": 2.7281174502880505,
      "grad_norm": 0.7558378577232361,
      "learning_rate": 4.6044339950326944e-05,
      "loss": 0.5531,
      "step": 14680
    },
    {
      "epoch": 2.7283032893514214,
      "grad_norm": 0.877690315246582,
      "learning_rate": 4.603204020143176e-05,
      "loss": 0.8854,
      "step": 14681
    },
    {
      "epoch": 2.7284891284147927,
      "grad_norm": 0.8856509923934937,
      "learning_rate": 4.6019741604390075e-05,
      "loss": 0.7802,
      "step": 14682
    },
    {
      "epoch": 2.728674967478164,
      "grad_norm": 0.7319670915603638,
      "learning_rate": 4.600744415946438e-05,
      "loss": 1.0037,
      "step": 14683
    },
    {
      "epoch": 2.728860806541535,
      "grad_norm": 0.8687145709991455,
      "learning_rate": 4.599514786691708e-05,
      "loss": 1.1824,
      "step": 14684
    },
    {
      "epoch": 2.7290466456049063,
      "grad_norm": 0.8814767599105835,
      "learning_rate": 4.5982852727010715e-05,
      "loss": 1.0184,
      "step": 14685
    },
    {
      "epoch": 2.729232484668277,
      "grad_norm": 0.7699080109596252,
      "learning_rate": 4.5970558740007666e-05,
      "loss": 0.9656,
      "step": 14686
    },
    {
      "epoch": 2.7294183237316485,
      "grad_norm": 0.6471914052963257,
      "learning_rate": 4.5958265906170315e-05,
      "loss": 0.6988,
      "step": 14687
    },
    {
      "epoch": 2.7296041627950194,
      "grad_norm": 0.8503636121749878,
      "learning_rate": 4.594597422576101e-05,
      "loss": 0.7991,
      "step": 14688
    },
    {
      "epoch": 2.7297900018583907,
      "grad_norm": 0.8441890478134155,
      "learning_rate": 4.593368369904216e-05,
      "loss": 0.9067,
      "step": 14689
    },
    {
      "epoch": 2.7299758409217616,
      "grad_norm": 0.7589361071586609,
      "learning_rate": 4.592139432627605e-05,
      "loss": 0.8271,
      "step": 14690
    },
    {
      "epoch": 2.730161679985133,
      "grad_norm": 0.6977306604385376,
      "learning_rate": 4.590910610772493e-05,
      "loss": 0.76,
      "step": 14691
    },
    {
      "epoch": 2.730347519048504,
      "grad_norm": 0.8234983086585999,
      "learning_rate": 4.589681904365113e-05,
      "loss": 0.8081,
      "step": 14692
    },
    {
      "epoch": 2.730533358111875,
      "grad_norm": 0.7815526127815247,
      "learning_rate": 4.588453313431691e-05,
      "loss": 0.7782,
      "step": 14693
    },
    {
      "epoch": 2.7307191971752465,
      "grad_norm": 0.8292824029922485,
      "learning_rate": 4.5872248379984464e-05,
      "loss": 0.6498,
      "step": 14694
    },
    {
      "epoch": 2.7309050362386174,
      "grad_norm": 0.8337152600288391,
      "learning_rate": 4.5859964780916e-05,
      "loss": 0.9102,
      "step": 14695
    },
    {
      "epoch": 2.7310908753019882,
      "grad_norm": 0.9309200048446655,
      "learning_rate": 4.584768233737363e-05,
      "loss": 1.03,
      "step": 14696
    },
    {
      "epoch": 2.7312767143653596,
      "grad_norm": 0.779767632484436,
      "learning_rate": 4.58354010496196e-05,
      "loss": 0.9953,
      "step": 14697
    },
    {
      "epoch": 2.731462553428731,
      "grad_norm": 0.8021158576011658,
      "learning_rate": 4.582312091791597e-05,
      "loss": 0.8515,
      "step": 14698
    },
    {
      "epoch": 2.731648392492102,
      "grad_norm": 0.9271323084831238,
      "learning_rate": 4.5810841942524864e-05,
      "loss": 0.8251,
      "step": 14699
    },
    {
      "epoch": 2.731834231555473,
      "grad_norm": 0.8800750970840454,
      "learning_rate": 4.579856412370831e-05,
      "loss": 0.8703,
      "step": 14700
    },
    {
      "epoch": 2.732020070618844,
      "grad_norm": 0.8031773567199707,
      "learning_rate": 4.578628746172838e-05,
      "loss": 0.8584,
      "step": 14701
    },
    {
      "epoch": 2.7322059096822153,
      "grad_norm": 0.8233345150947571,
      "learning_rate": 4.577401195684715e-05,
      "loss": 1.0454,
      "step": 14702
    },
    {
      "epoch": 2.732391748745586,
      "grad_norm": 0.8056541681289673,
      "learning_rate": 4.576173760932658e-05,
      "loss": 1.0582,
      "step": 14703
    },
    {
      "epoch": 2.7325775878089575,
      "grad_norm": 0.9316117763519287,
      "learning_rate": 4.574946441942861e-05,
      "loss": 1.0211,
      "step": 14704
    },
    {
      "epoch": 2.7327634268723284,
      "grad_norm": 0.7821735739707947,
      "learning_rate": 4.573719238741526e-05,
      "loss": 0.6886,
      "step": 14705
    },
    {
      "epoch": 2.7329492659356998,
      "grad_norm": 0.8058688640594482,
      "learning_rate": 4.572492151354842e-05,
      "loss": 1.0934,
      "step": 14706
    },
    {
      "epoch": 2.7331351049990706,
      "grad_norm": 0.8328232765197754,
      "learning_rate": 4.571265179808999e-05,
      "loss": 0.8818,
      "step": 14707
    },
    {
      "epoch": 2.733320944062442,
      "grad_norm": 0.7413308620452881,
      "learning_rate": 4.570038324130181e-05,
      "loss": 1.0653,
      "step": 14708
    },
    {
      "epoch": 2.7335067831258133,
      "grad_norm": 0.8222902417182922,
      "learning_rate": 4.568811584344581e-05,
      "loss": 1.0763,
      "step": 14709
    },
    {
      "epoch": 2.733692622189184,
      "grad_norm": 0.8213867545127869,
      "learning_rate": 4.567584960478377e-05,
      "loss": 0.8748,
      "step": 14710
    },
    {
      "epoch": 2.733878461252555,
      "grad_norm": 0.9049264788627625,
      "learning_rate": 4.56635845255775e-05,
      "loss": 1.0071,
      "step": 14711
    },
    {
      "epoch": 2.7340643003159264,
      "grad_norm": 0.9140104651451111,
      "learning_rate": 4.565132060608869e-05,
      "loss": 0.9517,
      "step": 14712
    },
    {
      "epoch": 2.7342501393792977,
      "grad_norm": 0.8153535723686218,
      "learning_rate": 4.5639057846579246e-05,
      "loss": 0.8449,
      "step": 14713
    },
    {
      "epoch": 2.7344359784426686,
      "grad_norm": 0.9286594390869141,
      "learning_rate": 4.562679624731084e-05,
      "loss": 0.8305,
      "step": 14714
    },
    {
      "epoch": 2.7346218175060395,
      "grad_norm": 0.7901486158370972,
      "learning_rate": 4.561453580854516e-05,
      "loss": 0.7905,
      "step": 14715
    },
    {
      "epoch": 2.734807656569411,
      "grad_norm": 0.7989631295204163,
      "learning_rate": 4.560227653054383e-05,
      "loss": 0.7111,
      "step": 14716
    },
    {
      "epoch": 2.734993495632782,
      "grad_norm": 0.8573057651519775,
      "learning_rate": 4.559001841356861e-05,
      "loss": 1.0795,
      "step": 14717
    },
    {
      "epoch": 2.735179334696153,
      "grad_norm": 0.9220576286315918,
      "learning_rate": 4.557776145788108e-05,
      "loss": 1.1402,
      "step": 14718
    },
    {
      "epoch": 2.7353651737595244,
      "grad_norm": 0.8109697699546814,
      "learning_rate": 4.556550566374283e-05,
      "loss": 1.1164,
      "step": 14719
    },
    {
      "epoch": 2.7355510128228953,
      "grad_norm": 0.8072250485420227,
      "learning_rate": 4.55532510314154e-05,
      "loss": 0.8904,
      "step": 14720
    },
    {
      "epoch": 2.7357368518862666,
      "grad_norm": 0.8466371893882751,
      "learning_rate": 4.5540997561160434e-05,
      "loss": 0.9383,
      "step": 14721
    },
    {
      "epoch": 2.7359226909496375,
      "grad_norm": 0.8313689231872559,
      "learning_rate": 4.552874525323939e-05,
      "loss": 0.8661,
      "step": 14722
    },
    {
      "epoch": 2.736108530013009,
      "grad_norm": 0.724134087562561,
      "learning_rate": 4.551649410791384e-05,
      "loss": 0.7092,
      "step": 14723
    },
    {
      "epoch": 2.73629436907638,
      "grad_norm": 0.7286176681518555,
      "learning_rate": 4.55042441254452e-05,
      "loss": 0.848,
      "step": 14724
    },
    {
      "epoch": 2.736480208139751,
      "grad_norm": 0.8972499966621399,
      "learning_rate": 4.549199530609497e-05,
      "loss": 1.1611,
      "step": 14725
    },
    {
      "epoch": 2.736666047203122,
      "grad_norm": 0.8000126481056213,
      "learning_rate": 4.547974765012456e-05,
      "loss": 0.8529,
      "step": 14726
    },
    {
      "epoch": 2.7368518862664932,
      "grad_norm": 0.7437418699264526,
      "learning_rate": 4.546750115779538e-05,
      "loss": 0.873,
      "step": 14727
    },
    {
      "epoch": 2.7370377253298646,
      "grad_norm": 0.7979970574378967,
      "learning_rate": 4.545525582936877e-05,
      "loss": 0.8908,
      "step": 14728
    },
    {
      "epoch": 2.7372235643932354,
      "grad_norm": 0.9213851690292358,
      "learning_rate": 4.544301166510616e-05,
      "loss": 0.9213,
      "step": 14729
    },
    {
      "epoch": 2.7374094034566063,
      "grad_norm": 0.9173702001571655,
      "learning_rate": 4.543076866526884e-05,
      "loss": 0.886,
      "step": 14730
    },
    {
      "epoch": 2.7375952425199777,
      "grad_norm": 0.7364941835403442,
      "learning_rate": 4.541852683011812e-05,
      "loss": 0.9047,
      "step": 14731
    },
    {
      "epoch": 2.737781081583349,
      "grad_norm": 0.8463900089263916,
      "learning_rate": 4.540628615991525e-05,
      "loss": 0.82,
      "step": 14732
    },
    {
      "epoch": 2.73796692064672,
      "grad_norm": 0.7539997100830078,
      "learning_rate": 4.539404665492151e-05,
      "loss": 0.8949,
      "step": 14733
    },
    {
      "epoch": 2.738152759710091,
      "grad_norm": 0.8457027077674866,
      "learning_rate": 4.538180831539817e-05,
      "loss": 1.0184,
      "step": 14734
    },
    {
      "epoch": 2.738338598773462,
      "grad_norm": 1.0152959823608398,
      "learning_rate": 4.5369571141606406e-05,
      "loss": 0.9442,
      "step": 14735
    },
    {
      "epoch": 2.7385244378368334,
      "grad_norm": 0.7157995104789734,
      "learning_rate": 4.5357335133807356e-05,
      "loss": 0.882,
      "step": 14736
    },
    {
      "epoch": 2.7387102769002043,
      "grad_norm": 0.8495806455612183,
      "learning_rate": 4.5345100292262254e-05,
      "loss": 0.8718,
      "step": 14737
    },
    {
      "epoch": 2.7388961159635756,
      "grad_norm": 0.876046895980835,
      "learning_rate": 4.533286661723219e-05,
      "loss": 0.9261,
      "step": 14738
    },
    {
      "epoch": 2.7390819550269465,
      "grad_norm": 0.7684153914451599,
      "learning_rate": 4.532063410897829e-05,
      "loss": 0.9394,
      "step": 14739
    },
    {
      "epoch": 2.739267794090318,
      "grad_norm": 0.8016115427017212,
      "learning_rate": 4.5308402767761573e-05,
      "loss": 0.8162,
      "step": 14740
    },
    {
      "epoch": 2.7394536331536887,
      "grad_norm": 0.7203810811042786,
      "learning_rate": 4.5296172593843177e-05,
      "loss": 0.7493,
      "step": 14741
    },
    {
      "epoch": 2.73963947221706,
      "grad_norm": 0.8666025996208191,
      "learning_rate": 4.5283943587484115e-05,
      "loss": 0.9386,
      "step": 14742
    },
    {
      "epoch": 2.7398253112804314,
      "grad_norm": 0.8360776901245117,
      "learning_rate": 4.527171574894534e-05,
      "loss": 0.812,
      "step": 14743
    },
    {
      "epoch": 2.7400111503438023,
      "grad_norm": 0.817794919013977,
      "learning_rate": 4.525948907848787e-05,
      "loss": 1.0972,
      "step": 14744
    },
    {
      "epoch": 2.740196989407173,
      "grad_norm": 0.8460525870323181,
      "learning_rate": 4.524726357637271e-05,
      "loss": 1.0712,
      "step": 14745
    },
    {
      "epoch": 2.7403828284705445,
      "grad_norm": 0.943310022354126,
      "learning_rate": 4.523503924286074e-05,
      "loss": 0.9896,
      "step": 14746
    },
    {
      "epoch": 2.740568667533916,
      "grad_norm": 0.7673540115356445,
      "learning_rate": 4.522281607821288e-05,
      "loss": 0.8957,
      "step": 14747
    },
    {
      "epoch": 2.7407545065972867,
      "grad_norm": 0.9330706000328064,
      "learning_rate": 4.5210594082689986e-05,
      "loss": 0.9737,
      "step": 14748
    },
    {
      "epoch": 2.740940345660658,
      "grad_norm": 0.9254998564720154,
      "learning_rate": 4.519837325655296e-05,
      "loss": 0.5841,
      "step": 14749
    },
    {
      "epoch": 2.741126184724029,
      "grad_norm": 0.7357127070426941,
      "learning_rate": 4.518615360006262e-05,
      "loss": 0.8598,
      "step": 14750
    },
    {
      "epoch": 2.7413120237874002,
      "grad_norm": 0.8163831233978271,
      "learning_rate": 4.517393511347976e-05,
      "loss": 1.0535,
      "step": 14751
    },
    {
      "epoch": 2.741497862850771,
      "grad_norm": 0.8442891836166382,
      "learning_rate": 4.516171779706514e-05,
      "loss": 0.9516,
      "step": 14752
    },
    {
      "epoch": 2.7416837019141425,
      "grad_norm": 0.9410925507545471,
      "learning_rate": 4.514950165107959e-05,
      "loss": 0.9114,
      "step": 14753
    },
    {
      "epoch": 2.7418695409775133,
      "grad_norm": 0.6992850303649902,
      "learning_rate": 4.513728667578375e-05,
      "loss": 0.9611,
      "step": 14754
    },
    {
      "epoch": 2.7420553800408847,
      "grad_norm": 0.894448459148407,
      "learning_rate": 4.512507287143842e-05,
      "loss": 0.9718,
      "step": 14755
    },
    {
      "epoch": 2.7422412191042556,
      "grad_norm": 1.4124360084533691,
      "learning_rate": 4.511286023830421e-05,
      "loss": 1.4758,
      "step": 14756
    },
    {
      "epoch": 2.742427058167627,
      "grad_norm": 0.8501431941986084,
      "learning_rate": 4.5100648776641844e-05,
      "loss": 0.6057,
      "step": 14757
    },
    {
      "epoch": 2.742612897230998,
      "grad_norm": 0.7781663537025452,
      "learning_rate": 4.5088438486711905e-05,
      "loss": 0.831,
      "step": 14758
    },
    {
      "epoch": 2.742798736294369,
      "grad_norm": 0.8618481755256653,
      "learning_rate": 4.5076229368775024e-05,
      "loss": 0.7601,
      "step": 14759
    },
    {
      "epoch": 2.74298457535774,
      "grad_norm": 0.7881844639778137,
      "learning_rate": 4.506402142309174e-05,
      "loss": 0.9385,
      "step": 14760
    },
    {
      "epoch": 2.7431704144211113,
      "grad_norm": 0.9128740429878235,
      "learning_rate": 4.5051814649922674e-05,
      "loss": 0.9802,
      "step": 14761
    },
    {
      "epoch": 2.7433562534844826,
      "grad_norm": 0.892898440361023,
      "learning_rate": 4.5039609049528333e-05,
      "loss": 0.8683,
      "step": 14762
    },
    {
      "epoch": 2.7435420925478535,
      "grad_norm": 0.8797706365585327,
      "learning_rate": 4.502740462216919e-05,
      "loss": 1.0483,
      "step": 14763
    },
    {
      "epoch": 2.7437279316112244,
      "grad_norm": 0.8321303725242615,
      "learning_rate": 4.501520136810575e-05,
      "loss": 0.9829,
      "step": 14764
    },
    {
      "epoch": 2.7439137706745957,
      "grad_norm": 1.2347742319107056,
      "learning_rate": 4.5002999287598524e-05,
      "loss": 1.357,
      "step": 14765
    },
    {
      "epoch": 2.744099609737967,
      "grad_norm": 0.7855826020240784,
      "learning_rate": 4.4990798380907895e-05,
      "loss": 0.8295,
      "step": 14766
    },
    {
      "epoch": 2.744285448801338,
      "grad_norm": 0.7561647891998291,
      "learning_rate": 4.497859864829427e-05,
      "loss": 0.879,
      "step": 14767
    },
    {
      "epoch": 2.7444712878647093,
      "grad_norm": 0.9717286229133606,
      "learning_rate": 4.4966400090018e-05,
      "loss": 0.7701,
      "step": 14768
    },
    {
      "epoch": 2.74465712692808,
      "grad_norm": 0.8584936857223511,
      "learning_rate": 4.495420270633953e-05,
      "loss": 0.7957,
      "step": 14769
    },
    {
      "epoch": 2.7448429659914515,
      "grad_norm": 0.7288717031478882,
      "learning_rate": 4.494200649751913e-05,
      "loss": 0.8592,
      "step": 14770
    },
    {
      "epoch": 2.7450288050548224,
      "grad_norm": 1.1725331544876099,
      "learning_rate": 4.492981146381712e-05,
      "loss": 0.9022,
      "step": 14771
    },
    {
      "epoch": 2.7452146441181937,
      "grad_norm": 0.8128632307052612,
      "learning_rate": 4.491761760549374e-05,
      "loss": 0.7259,
      "step": 14772
    },
    {
      "epoch": 2.7454004831815646,
      "grad_norm": 0.9310014843940735,
      "learning_rate": 4.4905424922809326e-05,
      "loss": 0.8722,
      "step": 14773
    },
    {
      "epoch": 2.745586322244936,
      "grad_norm": 0.790547788143158,
      "learning_rate": 4.489323341602407e-05,
      "loss": 1.0479,
      "step": 14774
    },
    {
      "epoch": 2.745772161308307,
      "grad_norm": 0.8988275527954102,
      "learning_rate": 4.488104308539814e-05,
      "loss": 0.9611,
      "step": 14775
    },
    {
      "epoch": 2.745958000371678,
      "grad_norm": 0.7555850744247437,
      "learning_rate": 4.486885393119177e-05,
      "loss": 0.9676,
      "step": 14776
    },
    {
      "epoch": 2.7461438394350495,
      "grad_norm": 0.7913243770599365,
      "learning_rate": 4.485666595366512e-05,
      "loss": 0.8562,
      "step": 14777
    },
    {
      "epoch": 2.7463296784984204,
      "grad_norm": 0.8432217836380005,
      "learning_rate": 4.4844479153078325e-05,
      "loss": 1.0425,
      "step": 14778
    },
    {
      "epoch": 2.7465155175617912,
      "grad_norm": 0.7686410546302795,
      "learning_rate": 4.483229352969146e-05,
      "loss": 0.9279,
      "step": 14779
    },
    {
      "epoch": 2.7467013566251626,
      "grad_norm": 0.8736889362335205,
      "learning_rate": 4.482010908376459e-05,
      "loss": 1.0946,
      "step": 14780
    },
    {
      "epoch": 2.746887195688534,
      "grad_norm": 0.8477604389190674,
      "learning_rate": 4.480792581555783e-05,
      "loss": 0.9413,
      "step": 14781
    },
    {
      "epoch": 2.747073034751905,
      "grad_norm": 0.881502628326416,
      "learning_rate": 4.4795743725331185e-05,
      "loss": 0.757,
      "step": 14782
    },
    {
      "epoch": 2.747258873815276,
      "grad_norm": 0.7617281675338745,
      "learning_rate": 4.4783562813344616e-05,
      "loss": 0.7723,
      "step": 14783
    },
    {
      "epoch": 2.747444712878647,
      "grad_norm": 0.7542743682861328,
      "learning_rate": 4.4771383079858175e-05,
      "loss": 0.8692,
      "step": 14784
    },
    {
      "epoch": 2.7476305519420183,
      "grad_norm": 0.8307807445526123,
      "learning_rate": 4.475920452513176e-05,
      "loss": 0.7277,
      "step": 14785
    },
    {
      "epoch": 2.747816391005389,
      "grad_norm": 0.954990804195404,
      "learning_rate": 4.474702714942536e-05,
      "loss": 0.968,
      "step": 14786
    },
    {
      "epoch": 2.7480022300687605,
      "grad_norm": 0.774532675743103,
      "learning_rate": 4.473485095299884e-05,
      "loss": 0.8424,
      "step": 14787
    },
    {
      "epoch": 2.7481880691321314,
      "grad_norm": 0.8852267265319824,
      "learning_rate": 4.472267593611206e-05,
      "loss": 1.0926,
      "step": 14788
    },
    {
      "epoch": 2.7483739081955028,
      "grad_norm": 0.7718761563301086,
      "learning_rate": 4.471050209902493e-05,
      "loss": 1.0268,
      "step": 14789
    },
    {
      "epoch": 2.7485597472588736,
      "grad_norm": 0.897894561290741,
      "learning_rate": 4.469832944199727e-05,
      "loss": 0.8493,
      "step": 14790
    },
    {
      "epoch": 2.748745586322245,
      "grad_norm": 0.7675634622573853,
      "learning_rate": 4.4686157965288845e-05,
      "loss": 0.9872,
      "step": 14791
    },
    {
      "epoch": 2.7489314253856163,
      "grad_norm": 0.7812278866767883,
      "learning_rate": 4.4673987669159434e-05,
      "loss": 0.9965,
      "step": 14792
    },
    {
      "epoch": 2.749117264448987,
      "grad_norm": 0.9022442698478699,
      "learning_rate": 4.4661818553868836e-05,
      "loss": 0.7828,
      "step": 14793
    },
    {
      "epoch": 2.749303103512358,
      "grad_norm": 0.8825806379318237,
      "learning_rate": 4.464965061967677e-05,
      "loss": 0.859,
      "step": 14794
    },
    {
      "epoch": 2.7494889425757294,
      "grad_norm": 1.4941855669021606,
      "learning_rate": 4.463748386684289e-05,
      "loss": 1.2713,
      "step": 14795
    },
    {
      "epoch": 2.7496747816391007,
      "grad_norm": 0.8661191463470459,
      "learning_rate": 4.46253182956269e-05,
      "loss": 0.9672,
      "step": 14796
    },
    {
      "epoch": 2.7498606207024716,
      "grad_norm": 0.7032533288002014,
      "learning_rate": 4.4613153906288506e-05,
      "loss": 0.866,
      "step": 14797
    },
    {
      "epoch": 2.750046459765843,
      "grad_norm": 0.8711912631988525,
      "learning_rate": 4.4600990699087294e-05,
      "loss": 0.8178,
      "step": 14798
    },
    {
      "epoch": 2.750232298829214,
      "grad_norm": 0.6879599094390869,
      "learning_rate": 4.458882867428288e-05,
      "loss": 0.784,
      "step": 14799
    },
    {
      "epoch": 2.750418137892585,
      "grad_norm": 0.8086994290351868,
      "learning_rate": 4.4576667832134775e-05,
      "loss": 0.811,
      "step": 14800
    },
    {
      "epoch": 2.750603976955956,
      "grad_norm": 0.7266994714736938,
      "learning_rate": 4.456450817290263e-05,
      "loss": 0.8126,
      "step": 14801
    },
    {
      "epoch": 2.7507898160193274,
      "grad_norm": 1.0289912223815918,
      "learning_rate": 4.455234969684592e-05,
      "loss": 0.8967,
      "step": 14802
    },
    {
      "epoch": 2.7509756550826983,
      "grad_norm": 1.1989860534667969,
      "learning_rate": 4.454019240422412e-05,
      "loss": 1.0713,
      "step": 14803
    },
    {
      "epoch": 2.7511614941460696,
      "grad_norm": 0.8081581592559814,
      "learning_rate": 4.452803629529678e-05,
      "loss": 0.7533,
      "step": 14804
    },
    {
      "epoch": 2.7513473332094405,
      "grad_norm": 0.8733363151550293,
      "learning_rate": 4.4515881370323306e-05,
      "loss": 0.914,
      "step": 14805
    },
    {
      "epoch": 2.751533172272812,
      "grad_norm": 0.8047726154327393,
      "learning_rate": 4.45037276295631e-05,
      "loss": 1.0163,
      "step": 14806
    },
    {
      "epoch": 2.751533172272812,
      "eval_loss": 1.016861081123352,
      "eval_runtime": 23.1696,
      "eval_samples_per_second": 47.131,
      "eval_steps_per_second": 23.565,
      "step": 14806
    },
    {
      "epoch": 2.751719011336183,
      "grad_norm": 0.8058808445930481,
      "learning_rate": 4.4491575073275625e-05,
      "loss": 0.9496,
      "step": 14807
    },
    {
      "epoch": 2.751904850399554,
      "grad_norm": 0.9865077137947083,
      "learning_rate": 4.447942370172019e-05,
      "loss": 0.73,
      "step": 14808
    },
    {
      "epoch": 2.752090689462925,
      "grad_norm": 1.0397051572799683,
      "learning_rate": 4.4467273515156216e-05,
      "loss": 0.9428,
      "step": 14809
    },
    {
      "epoch": 2.7522765285262962,
      "grad_norm": 1.0134010314941406,
      "learning_rate": 4.445512451384298e-05,
      "loss": 0.9326,
      "step": 14810
    },
    {
      "epoch": 2.7524623675896676,
      "grad_norm": 0.6431830525398254,
      "learning_rate": 4.444297669803981e-05,
      "loss": 0.6813,
      "step": 14811
    },
    {
      "epoch": 2.7526482066530384,
      "grad_norm": 0.7811623215675354,
      "learning_rate": 4.4430830068005914e-05,
      "loss": 0.8558,
      "step": 14812
    },
    {
      "epoch": 2.7528340457164093,
      "grad_norm": 0.8130550980567932,
      "learning_rate": 4.4418684624000626e-05,
      "loss": 0.8384,
      "step": 14813
    },
    {
      "epoch": 2.7530198847797807,
      "grad_norm": 0.9185358881950378,
      "learning_rate": 4.4406540366283134e-05,
      "loss": 1.1363,
      "step": 14814
    },
    {
      "epoch": 2.753205723843152,
      "grad_norm": 0.9589024186134338,
      "learning_rate": 4.43943972951126e-05,
      "loss": 0.799,
      "step": 14815
    },
    {
      "epoch": 2.753391562906523,
      "grad_norm": 0.8474277853965759,
      "learning_rate": 4.438225541074827e-05,
      "loss": 0.8946,
      "step": 14816
    },
    {
      "epoch": 2.753577401969894,
      "grad_norm": 1.0162084102630615,
      "learning_rate": 4.437011471344922e-05,
      "loss": 0.8196,
      "step": 14817
    },
    {
      "epoch": 2.753763241033265,
      "grad_norm": 0.7180842161178589,
      "learning_rate": 4.4357975203474644e-05,
      "loss": 0.9129,
      "step": 14818
    },
    {
      "epoch": 2.7539490800966364,
      "grad_norm": 0.729688286781311,
      "learning_rate": 4.43458368810836e-05,
      "loss": 0.903,
      "step": 14819
    },
    {
      "epoch": 2.7541349191600073,
      "grad_norm": 0.9457212090492249,
      "learning_rate": 4.433369974653512e-05,
      "loss": 1.0448,
      "step": 14820
    },
    {
      "epoch": 2.7543207582233786,
      "grad_norm": 0.7897746562957764,
      "learning_rate": 4.432156380008834e-05,
      "loss": 0.8628,
      "step": 14821
    },
    {
      "epoch": 2.7545065972867495,
      "grad_norm": 0.8315415978431702,
      "learning_rate": 4.430942904200223e-05,
      "loss": 1.0471,
      "step": 14822
    },
    {
      "epoch": 2.754692436350121,
      "grad_norm": 0.8469882011413574,
      "learning_rate": 4.429729547253574e-05,
      "loss": 0.8774,
      "step": 14823
    },
    {
      "epoch": 2.7548782754134917,
      "grad_norm": 0.785725474357605,
      "learning_rate": 4.428516309194793e-05,
      "loss": 1.0214,
      "step": 14824
    },
    {
      "epoch": 2.755064114476863,
      "grad_norm": 0.8006011843681335,
      "learning_rate": 4.427303190049771e-05,
      "loss": 1.0044,
      "step": 14825
    },
    {
      "epoch": 2.7552499535402344,
      "grad_norm": 0.6976585388183594,
      "learning_rate": 4.426090189844398e-05,
      "loss": 0.8216,
      "step": 14826
    },
    {
      "epoch": 2.7554357926036053,
      "grad_norm": 1.048567533493042,
      "learning_rate": 4.424877308604563e-05,
      "loss": 1.1218,
      "step": 14827
    },
    {
      "epoch": 2.755621631666976,
      "grad_norm": 0.8750372529029846,
      "learning_rate": 4.423664546356152e-05,
      "loss": 0.9318,
      "step": 14828
    },
    {
      "epoch": 2.7558074707303475,
      "grad_norm": 0.7322916984558105,
      "learning_rate": 4.422451903125058e-05,
      "loss": 0.7277,
      "step": 14829
    },
    {
      "epoch": 2.755993309793719,
      "grad_norm": 0.9944577813148499,
      "learning_rate": 4.421239378937156e-05,
      "loss": 1.1098,
      "step": 14830
    },
    {
      "epoch": 2.7561791488570897,
      "grad_norm": 0.8471137881278992,
      "learning_rate": 4.420026973818321e-05,
      "loss": 1.0242,
      "step": 14831
    },
    {
      "epoch": 2.756364987920461,
      "grad_norm": 0.8469953536987305,
      "learning_rate": 4.418814687794439e-05,
      "loss": 0.9368,
      "step": 14832
    },
    {
      "epoch": 2.756550826983832,
      "grad_norm": 0.7500290274620056,
      "learning_rate": 4.41760252089138e-05,
      "loss": 0.5646,
      "step": 14833
    },
    {
      "epoch": 2.7567366660472032,
      "grad_norm": 0.9665776491165161,
      "learning_rate": 4.4163904731350144e-05,
      "loss": 0.9747,
      "step": 14834
    },
    {
      "epoch": 2.756922505110574,
      "grad_norm": 0.6673862934112549,
      "learning_rate": 4.415178544551211e-05,
      "loss": 0.6564,
      "step": 14835
    },
    {
      "epoch": 2.7571083441739455,
      "grad_norm": 0.7555088400840759,
      "learning_rate": 4.413966735165839e-05,
      "loss": 1.0257,
      "step": 14836
    },
    {
      "epoch": 2.7572941832373163,
      "grad_norm": 0.8664906024932861,
      "learning_rate": 4.4127550450047616e-05,
      "loss": 0.8143,
      "step": 14837
    },
    {
      "epoch": 2.7574800223006877,
      "grad_norm": 0.9580231308937073,
      "learning_rate": 4.4115434740938365e-05,
      "loss": 1.1836,
      "step": 14838
    },
    {
      "epoch": 2.7576658613640586,
      "grad_norm": 1.0043007135391235,
      "learning_rate": 4.41033202245893e-05,
      "loss": 0.8902,
      "step": 14839
    },
    {
      "epoch": 2.75785170042743,
      "grad_norm": 0.8290864825248718,
      "learning_rate": 4.40912069012589e-05,
      "loss": 0.9894,
      "step": 14840
    },
    {
      "epoch": 2.758037539490801,
      "grad_norm": 1.0630974769592285,
      "learning_rate": 4.407909477120578e-05,
      "loss": 0.7752,
      "step": 14841
    },
    {
      "epoch": 2.758223378554172,
      "grad_norm": 0.760078489780426,
      "learning_rate": 4.4066983834688415e-05,
      "loss": 0.8939,
      "step": 14842
    },
    {
      "epoch": 2.758409217617543,
      "grad_norm": 1.1572614908218384,
      "learning_rate": 4.405487409196526e-05,
      "loss": 1.0579,
      "step": 14843
    },
    {
      "epoch": 2.7585950566809143,
      "grad_norm": 0.8342961668968201,
      "learning_rate": 4.404276554329485e-05,
      "loss": 0.9826,
      "step": 14844
    },
    {
      "epoch": 2.7587808957442856,
      "grad_norm": 0.9802063703536987,
      "learning_rate": 4.403065818893558e-05,
      "loss": 1.1177,
      "step": 14845
    },
    {
      "epoch": 2.7589667348076565,
      "grad_norm": 0.8134041428565979,
      "learning_rate": 4.401855202914586e-05,
      "loss": 0.9372,
      "step": 14846
    },
    {
      "epoch": 2.7591525738710274,
      "grad_norm": 0.7945948243141174,
      "learning_rate": 4.400644706418405e-05,
      "loss": 0.9862,
      "step": 14847
    },
    {
      "epoch": 2.7593384129343987,
      "grad_norm": 0.7153193354606628,
      "learning_rate": 4.399434329430854e-05,
      "loss": 0.7187,
      "step": 14848
    },
    {
      "epoch": 2.75952425199777,
      "grad_norm": 0.9038686752319336,
      "learning_rate": 4.39822407197777e-05,
      "loss": 0.7832,
      "step": 14849
    },
    {
      "epoch": 2.759710091061141,
      "grad_norm": 0.7421132326126099,
      "learning_rate": 4.397013934084979e-05,
      "loss": 0.8534,
      "step": 14850
    },
    {
      "epoch": 2.7598959301245123,
      "grad_norm": 0.8442568778991699,
      "learning_rate": 4.395803915778308e-05,
      "loss": 0.6722,
      "step": 14851
    },
    {
      "epoch": 2.760081769187883,
      "grad_norm": 0.9152247309684753,
      "learning_rate": 4.3945940170835885e-05,
      "loss": 0.9758,
      "step": 14852
    },
    {
      "epoch": 2.7602676082512545,
      "grad_norm": 0.8400849103927612,
      "learning_rate": 4.393384238026641e-05,
      "loss": 0.9794,
      "step": 14853
    },
    {
      "epoch": 2.7604534473146254,
      "grad_norm": 0.7551645636558533,
      "learning_rate": 4.3921745786332845e-05,
      "loss": 0.9922,
      "step": 14854
    },
    {
      "epoch": 2.7606392863779967,
      "grad_norm": 0.8360217213630676,
      "learning_rate": 4.3909650389293355e-05,
      "loss": 1.1151,
      "step": 14855
    },
    {
      "epoch": 2.760825125441368,
      "grad_norm": 0.9101220965385437,
      "learning_rate": 4.389755618940615e-05,
      "loss": 1.1536,
      "step": 14856
    },
    {
      "epoch": 2.761010964504739,
      "grad_norm": 0.7217719554901123,
      "learning_rate": 4.388546318692934e-05,
      "loss": 0.6163,
      "step": 14857
    },
    {
      "epoch": 2.76119680356811,
      "grad_norm": 0.9773128032684326,
      "learning_rate": 4.387337138212102e-05,
      "loss": 1.1546,
      "step": 14858
    },
    {
      "epoch": 2.761382642631481,
      "grad_norm": 0.7525717616081238,
      "learning_rate": 4.386128077523923e-05,
      "loss": 0.7095,
      "step": 14859
    },
    {
      "epoch": 2.7615684816948525,
      "grad_norm": 0.8013903498649597,
      "learning_rate": 4.384919136654207e-05,
      "loss": 1.0295,
      "step": 14860
    },
    {
      "epoch": 2.7617543207582234,
      "grad_norm": 1.510123372077942,
      "learning_rate": 4.3837103156287596e-05,
      "loss": 1.1998,
      "step": 14861
    },
    {
      "epoch": 2.7619401598215942,
      "grad_norm": 0.8579853773117065,
      "learning_rate": 4.382501614473378e-05,
      "loss": 1.1587,
      "step": 14862
    },
    {
      "epoch": 2.7621259988849656,
      "grad_norm": 0.8998013138771057,
      "learning_rate": 4.3812930332138557e-05,
      "loss": 0.9156,
      "step": 14863
    },
    {
      "epoch": 2.762311837948337,
      "grad_norm": 1.0269873142242432,
      "learning_rate": 4.3800845718759944e-05,
      "loss": 1.0224,
      "step": 14864
    },
    {
      "epoch": 2.762497677011708,
      "grad_norm": 0.7870360016822815,
      "learning_rate": 4.3788762304855844e-05,
      "loss": 1.1499,
      "step": 14865
    },
    {
      "epoch": 2.762683516075079,
      "grad_norm": 0.8245033025741577,
      "learning_rate": 4.377668009068415e-05,
      "loss": 0.9342,
      "step": 14866
    },
    {
      "epoch": 2.76286935513845,
      "grad_norm": 0.6947269439697266,
      "learning_rate": 4.3764599076502696e-05,
      "loss": 0.8029,
      "step": 14867
    },
    {
      "epoch": 2.7630551942018213,
      "grad_norm": 0.75066077709198,
      "learning_rate": 4.37525192625694e-05,
      "loss": 0.8048,
      "step": 14868
    },
    {
      "epoch": 2.763241033265192,
      "grad_norm": 0.7691546678543091,
      "learning_rate": 4.374044064914203e-05,
      "loss": 0.93,
      "step": 14869
    },
    {
      "epoch": 2.7634268723285635,
      "grad_norm": 0.8516491055488586,
      "learning_rate": 4.372836323647845e-05,
      "loss": 0.8159,
      "step": 14870
    },
    {
      "epoch": 2.7636127113919344,
      "grad_norm": 0.6997004151344299,
      "learning_rate": 4.3716287024836346e-05,
      "loss": 0.7847,
      "step": 14871
    },
    {
      "epoch": 2.7637985504553058,
      "grad_norm": 0.7898983359336853,
      "learning_rate": 4.370421201447356e-05,
      "loss": 0.5884,
      "step": 14872
    },
    {
      "epoch": 2.7639843895186766,
      "grad_norm": 0.8238096833229065,
      "learning_rate": 4.369213820564774e-05,
      "loss": 1.0009,
      "step": 14873
    },
    {
      "epoch": 2.764170228582048,
      "grad_norm": 0.9211301803588867,
      "learning_rate": 4.36800655986166e-05,
      "loss": 0.9284,
      "step": 14874
    },
    {
      "epoch": 2.7643560676454193,
      "grad_norm": 0.7392138242721558,
      "learning_rate": 4.3667994193637796e-05,
      "loss": 0.6207,
      "step": 14875
    },
    {
      "epoch": 2.76454190670879,
      "grad_norm": 0.7954480051994324,
      "learning_rate": 4.3655923990969005e-05,
      "loss": 0.9053,
      "step": 14876
    },
    {
      "epoch": 2.764727745772161,
      "grad_norm": 0.7361290454864502,
      "learning_rate": 4.364385499086784e-05,
      "loss": 0.9016,
      "step": 14877
    },
    {
      "epoch": 2.7649135848355324,
      "grad_norm": 0.9881463646888733,
      "learning_rate": 4.363178719359187e-05,
      "loss": 1.0199,
      "step": 14878
    },
    {
      "epoch": 2.7650994238989037,
      "grad_norm": 0.7867943048477173,
      "learning_rate": 4.3619720599398637e-05,
      "loss": 0.756,
      "step": 14879
    },
    {
      "epoch": 2.7652852629622746,
      "grad_norm": 1.01762855052948,
      "learning_rate": 4.3607655208545715e-05,
      "loss": 0.8002,
      "step": 14880
    },
    {
      "epoch": 2.765471102025646,
      "grad_norm": 0.8283891081809998,
      "learning_rate": 4.359559102129065e-05,
      "loss": 0.8995,
      "step": 14881
    },
    {
      "epoch": 2.765656941089017,
      "grad_norm": 0.9196315407752991,
      "learning_rate": 4.358352803789091e-05,
      "loss": 0.9448,
      "step": 14882
    },
    {
      "epoch": 2.765842780152388,
      "grad_norm": 0.6811094284057617,
      "learning_rate": 4.357146625860391e-05,
      "loss": 0.8073,
      "step": 14883
    },
    {
      "epoch": 2.766028619215759,
      "grad_norm": 0.7988422513008118,
      "learning_rate": 4.355940568368716e-05,
      "loss": 0.9822,
      "step": 14884
    },
    {
      "epoch": 2.7662144582791304,
      "grad_norm": 0.9253214001655579,
      "learning_rate": 4.354734631339804e-05,
      "loss": 0.8889,
      "step": 14885
    },
    {
      "epoch": 2.7664002973425013,
      "grad_norm": 1.163099765777588,
      "learning_rate": 4.3535288147993946e-05,
      "loss": 1.0403,
      "step": 14886
    },
    {
      "epoch": 2.7665861364058726,
      "grad_norm": 0.729446530342102,
      "learning_rate": 4.352323118773218e-05,
      "loss": 0.9152,
      "step": 14887
    },
    {
      "epoch": 2.7667719754692435,
      "grad_norm": 1.0046440362930298,
      "learning_rate": 4.351117543287018e-05,
      "loss": 0.7264,
      "step": 14888
    },
    {
      "epoch": 2.766957814532615,
      "grad_norm": 0.8194682598114014,
      "learning_rate": 4.349912088366518e-05,
      "loss": 0.9762,
      "step": 14889
    },
    {
      "epoch": 2.767143653595986,
      "grad_norm": 0.8480716347694397,
      "learning_rate": 4.348706754037446e-05,
      "loss": 0.7538,
      "step": 14890
    },
    {
      "epoch": 2.767329492659357,
      "grad_norm": 0.7657939195632935,
      "learning_rate": 4.3475015403255304e-05,
      "loss": 0.7417,
      "step": 14891
    },
    {
      "epoch": 2.767515331722728,
      "grad_norm": 0.7754166126251221,
      "learning_rate": 4.346296447256498e-05,
      "loss": 0.7323,
      "step": 14892
    },
    {
      "epoch": 2.7677011707860992,
      "grad_norm": 0.8909863233566284,
      "learning_rate": 4.345091474856066e-05,
      "loss": 1.0459,
      "step": 14893
    },
    {
      "epoch": 2.7678870098494706,
      "grad_norm": 0.9239513874053955,
      "learning_rate": 4.343886623149952e-05,
      "loss": 1.0778,
      "step": 14894
    },
    {
      "epoch": 2.7680728489128414,
      "grad_norm": 0.7537717819213867,
      "learning_rate": 4.342681892163868e-05,
      "loss": 0.8128,
      "step": 14895
    },
    {
      "epoch": 2.7682586879762123,
      "grad_norm": 0.8553297519683838,
      "learning_rate": 4.341477281923534e-05,
      "loss": 0.8233,
      "step": 14896
    },
    {
      "epoch": 2.7684445270395837,
      "grad_norm": 0.9643924832344055,
      "learning_rate": 4.340272792454657e-05,
      "loss": 1.0147,
      "step": 14897
    },
    {
      "epoch": 2.768630366102955,
      "grad_norm": 0.9254254698753357,
      "learning_rate": 4.339068423782945e-05,
      "loss": 0.9021,
      "step": 14898
    },
    {
      "epoch": 2.768816205166326,
      "grad_norm": 0.7231565117835999,
      "learning_rate": 4.3378641759341e-05,
      "loss": 0.7963,
      "step": 14899
    },
    {
      "epoch": 2.769002044229697,
      "grad_norm": 0.8661108016967773,
      "learning_rate": 4.3366600489338305e-05,
      "loss": 1.2158,
      "step": 14900
    },
    {
      "epoch": 2.769187883293068,
      "grad_norm": 0.7713779211044312,
      "learning_rate": 4.335456042807829e-05,
      "loss": 0.7526,
      "step": 14901
    },
    {
      "epoch": 2.7693737223564394,
      "grad_norm": 0.8019841313362122,
      "learning_rate": 4.334252157581803e-05,
      "loss": 0.8314,
      "step": 14902
    },
    {
      "epoch": 2.7695595614198103,
      "grad_norm": 0.8363486528396606,
      "learning_rate": 4.333048393281437e-05,
      "loss": 0.9437,
      "step": 14903
    },
    {
      "epoch": 2.7697454004831816,
      "grad_norm": 0.7701266407966614,
      "learning_rate": 4.331844749932432e-05,
      "loss": 0.9542,
      "step": 14904
    },
    {
      "epoch": 2.769931239546553,
      "grad_norm": 0.7874062657356262,
      "learning_rate": 4.330641227560474e-05,
      "loss": 0.6885,
      "step": 14905
    },
    {
      "epoch": 2.770117078609924,
      "grad_norm": 0.9805448651313782,
      "learning_rate": 4.3294378261912504e-05,
      "loss": 1.0802,
      "step": 14906
    },
    {
      "epoch": 2.7703029176732947,
      "grad_norm": 0.6709483861923218,
      "learning_rate": 4.328234545850442e-05,
      "loss": 0.9273,
      "step": 14907
    },
    {
      "epoch": 2.770488756736666,
      "grad_norm": 0.8253301382064819,
      "learning_rate": 4.327031386563737e-05,
      "loss": 1.024,
      "step": 14908
    },
    {
      "epoch": 2.7706745958000374,
      "grad_norm": 0.907292902469635,
      "learning_rate": 4.325828348356812e-05,
      "loss": 1.0564,
      "step": 14909
    },
    {
      "epoch": 2.7708604348634083,
      "grad_norm": 0.7271608114242554,
      "learning_rate": 4.324625431255342e-05,
      "loss": 0.7091,
      "step": 14910
    },
    {
      "epoch": 2.771046273926779,
      "grad_norm": 0.9254564642906189,
      "learning_rate": 4.323422635284998e-05,
      "loss": 0.8868,
      "step": 14911
    },
    {
      "epoch": 2.7712321129901505,
      "grad_norm": 0.8518773913383484,
      "learning_rate": 4.322219960471463e-05,
      "loss": 0.7656,
      "step": 14912
    },
    {
      "epoch": 2.771417952053522,
      "grad_norm": 0.7480016946792603,
      "learning_rate": 4.3210174068404e-05,
      "loss": 0.6123,
      "step": 14913
    },
    {
      "epoch": 2.7716037911168927,
      "grad_norm": 0.8283432126045227,
      "learning_rate": 4.319814974417473e-05,
      "loss": 0.9516,
      "step": 14914
    },
    {
      "epoch": 2.771789630180264,
      "grad_norm": 0.8425053954124451,
      "learning_rate": 4.3186126632283465e-05,
      "loss": 0.9492,
      "step": 14915
    },
    {
      "epoch": 2.771975469243635,
      "grad_norm": 0.7290083169937134,
      "learning_rate": 4.3174104732986854e-05,
      "loss": 0.6701,
      "step": 14916
    },
    {
      "epoch": 2.7721613083070062,
      "grad_norm": 0.8720309734344482,
      "learning_rate": 4.316208404654146e-05,
      "loss": 0.9944,
      "step": 14917
    },
    {
      "epoch": 2.772347147370377,
      "grad_norm": 0.8075978755950928,
      "learning_rate": 4.315006457320384e-05,
      "loss": 0.7447,
      "step": 14918
    },
    {
      "epoch": 2.7725329864337485,
      "grad_norm": 0.797562301158905,
      "learning_rate": 4.3138046313230506e-05,
      "loss": 0.9379,
      "step": 14919
    },
    {
      "epoch": 2.7727188254971193,
      "grad_norm": 1.662497878074646,
      "learning_rate": 4.312602926687803e-05,
      "loss": 1.0968,
      "step": 14920
    },
    {
      "epoch": 2.7729046645604907,
      "grad_norm": 0.7986350655555725,
      "learning_rate": 4.3114013434402856e-05,
      "loss": 0.7646,
      "step": 14921
    },
    {
      "epoch": 2.7730905036238616,
      "grad_norm": 0.7795454859733582,
      "learning_rate": 4.310199881606141e-05,
      "loss": 0.8714,
      "step": 14922
    },
    {
      "epoch": 2.773276342687233,
      "grad_norm": 0.8627483248710632,
      "learning_rate": 4.308998541211015e-05,
      "loss": 0.9569,
      "step": 14923
    },
    {
      "epoch": 2.773462181750604,
      "grad_norm": 1.0427031517028809,
      "learning_rate": 4.307797322280555e-05,
      "loss": 1.386,
      "step": 14924
    },
    {
      "epoch": 2.773648020813975,
      "grad_norm": 0.7577645778656006,
      "learning_rate": 4.306596224840391e-05,
      "loss": 0.8991,
      "step": 14925
    },
    {
      "epoch": 2.773833859877346,
      "grad_norm": 0.8506353497505188,
      "learning_rate": 4.305395248916161e-05,
      "loss": 0.7597,
      "step": 14926
    },
    {
      "epoch": 2.7740196989407173,
      "grad_norm": 1.018517017364502,
      "learning_rate": 4.3041943945334925e-05,
      "loss": 0.8695,
      "step": 14927
    },
    {
      "epoch": 2.7742055380040886,
      "grad_norm": 0.7532228231430054,
      "learning_rate": 4.3029936617180255e-05,
      "loss": 0.6172,
      "step": 14928
    },
    {
      "epoch": 2.7743913770674595,
      "grad_norm": 0.7622318267822266,
      "learning_rate": 4.301793050495381e-05,
      "loss": 0.8329,
      "step": 14929
    },
    {
      "epoch": 2.774577216130831,
      "grad_norm": 0.7729710340499878,
      "learning_rate": 4.3005925608911856e-05,
      "loss": 0.9368,
      "step": 14930
    },
    {
      "epoch": 2.7747630551942017,
      "grad_norm": 1.1952792406082153,
      "learning_rate": 4.2993921929310576e-05,
      "loss": 1.0395,
      "step": 14931
    },
    {
      "epoch": 2.774948894257573,
      "grad_norm": 0.8073039650917053,
      "learning_rate": 4.298191946640622e-05,
      "loss": 0.8195,
      "step": 14932
    },
    {
      "epoch": 2.775134733320944,
      "grad_norm": 0.9328490495681763,
      "learning_rate": 4.296991822045496e-05,
      "loss": 0.8152,
      "step": 14933
    },
    {
      "epoch": 2.7753205723843153,
      "grad_norm": 3.2860829830169678,
      "learning_rate": 4.2957918191712945e-05,
      "loss": 1.2947,
      "step": 14934
    },
    {
      "epoch": 2.775506411447686,
      "grad_norm": 0.9785019159317017,
      "learning_rate": 4.294591938043624e-05,
      "loss": 0.635,
      "step": 14935
    },
    {
      "epoch": 2.7756922505110575,
      "grad_norm": 0.9144778251647949,
      "learning_rate": 4.293392178688101e-05,
      "loss": 0.8483,
      "step": 14936
    },
    {
      "epoch": 2.7758780895744284,
      "grad_norm": 0.9231557846069336,
      "learning_rate": 4.292192541130329e-05,
      "loss": 1.1413,
      "step": 14937
    },
    {
      "epoch": 2.7760639286377997,
      "grad_norm": 0.7811604142189026,
      "learning_rate": 4.290993025395912e-05,
      "loss": 0.8274,
      "step": 14938
    },
    {
      "epoch": 2.776249767701171,
      "grad_norm": 0.9679073691368103,
      "learning_rate": 4.289793631510448e-05,
      "loss": 1.0898,
      "step": 14939
    },
    {
      "epoch": 2.776435606764542,
      "grad_norm": 0.7433283925056458,
      "learning_rate": 4.2885943594995434e-05,
      "loss": 0.7807,
      "step": 14940
    },
    {
      "epoch": 2.776621445827913,
      "grad_norm": 0.696416437625885,
      "learning_rate": 4.287395209388793e-05,
      "loss": 0.6271,
      "step": 14941
    },
    {
      "epoch": 2.776807284891284,
      "grad_norm": 0.7764953374862671,
      "learning_rate": 4.2861961812037866e-05,
      "loss": 0.8496,
      "step": 14942
    },
    {
      "epoch": 2.7769931239546555,
      "grad_norm": 0.7799713611602783,
      "learning_rate": 4.2849972749701104e-05,
      "loss": 0.8081,
      "step": 14943
    },
    {
      "epoch": 2.7771789630180264,
      "grad_norm": 0.7540068626403809,
      "learning_rate": 4.283798490713368e-05,
      "loss": 0.8996,
      "step": 14944
    },
    {
      "epoch": 2.7773648020813972,
      "grad_norm": 0.7690991163253784,
      "learning_rate": 4.282599828459136e-05,
      "loss": 0.9814,
      "step": 14945
    },
    {
      "epoch": 2.7775506411447686,
      "grad_norm": 0.7605803608894348,
      "learning_rate": 4.281401288233001e-05,
      "loss": 0.7908,
      "step": 14946
    },
    {
      "epoch": 2.77773648020814,
      "grad_norm": 0.7501049637794495,
      "learning_rate": 4.280202870060538e-05,
      "loss": 0.865,
      "step": 14947
    },
    {
      "epoch": 2.777922319271511,
      "grad_norm": 1.371169090270996,
      "learning_rate": 4.279004573967332e-05,
      "loss": 1.3352,
      "step": 14948
    },
    {
      "epoch": 2.778108158334882,
      "grad_norm": 0.9031583070755005,
      "learning_rate": 4.277806399978956e-05,
      "loss": 1.0714,
      "step": 14949
    },
    {
      "epoch": 2.778293997398253,
      "grad_norm": 0.8056634664535522,
      "learning_rate": 4.276608348120983e-05,
      "loss": 1.1175,
      "step": 14950
    },
    {
      "epoch": 2.7784798364616243,
      "grad_norm": 0.7620245814323425,
      "learning_rate": 4.275410418418979e-05,
      "loss": 0.9606,
      "step": 14951
    },
    {
      "epoch": 2.778665675524995,
      "grad_norm": 1.1092839241027832,
      "learning_rate": 4.2742126108985205e-05,
      "loss": 1.157,
      "step": 14952
    },
    {
      "epoch": 2.7788515145883665,
      "grad_norm": 0.829846203327179,
      "learning_rate": 4.273014925585163e-05,
      "loss": 0.7582,
      "step": 14953
    },
    {
      "epoch": 2.7790373536517374,
      "grad_norm": 0.9030061960220337,
      "learning_rate": 4.271817362504479e-05,
      "loss": 0.9176,
      "step": 14954
    },
    {
      "epoch": 2.7792231927151088,
      "grad_norm": 0.824874997138977,
      "learning_rate": 4.2706199216820185e-05,
      "loss": 1.064,
      "step": 14955
    },
    {
      "epoch": 2.7794090317784796,
      "grad_norm": 0.6992202401161194,
      "learning_rate": 4.269422603143348e-05,
      "loss": 0.6227,
      "step": 14956
    },
    {
      "epoch": 2.779594870841851,
      "grad_norm": 0.887396514415741,
      "learning_rate": 4.268225406914018e-05,
      "loss": 1.094,
      "step": 14957
    },
    {
      "epoch": 2.7797807099052223,
      "grad_norm": 0.7713488340377808,
      "learning_rate": 4.26702833301958e-05,
      "loss": 0.9086,
      "step": 14958
    },
    {
      "epoch": 2.779966548968593,
      "grad_norm": 0.9397330284118652,
      "learning_rate": 4.265831381485581e-05,
      "loss": 0.9952,
      "step": 14959
    },
    {
      "epoch": 2.780152388031964,
      "grad_norm": 0.8007112145423889,
      "learning_rate": 4.2646345523375744e-05,
      "loss": 0.9232,
      "step": 14960
    },
    {
      "epoch": 2.7803382270953354,
      "grad_norm": 0.8054376244544983,
      "learning_rate": 4.2634378456011015e-05,
      "loss": 0.9057,
      "step": 14961
    },
    {
      "epoch": 2.7805240661587067,
      "grad_norm": 0.9074475765228271,
      "learning_rate": 4.2622412613016995e-05,
      "loss": 1.0884,
      "step": 14962
    },
    {
      "epoch": 2.7807099052220776,
      "grad_norm": 0.7887566089630127,
      "learning_rate": 4.261044799464915e-05,
      "loss": 0.8381,
      "step": 14963
    },
    {
      "epoch": 2.780895744285449,
      "grad_norm": 0.842220664024353,
      "learning_rate": 4.259848460116277e-05,
      "loss": 1.0426,
      "step": 14964
    },
    {
      "epoch": 2.78108158334882,
      "grad_norm": 0.7846879363059998,
      "learning_rate": 4.2586522432813266e-05,
      "loss": 0.7901,
      "step": 14965
    },
    {
      "epoch": 2.781267422412191,
      "grad_norm": 0.9481580853462219,
      "learning_rate": 4.257456148985592e-05,
      "loss": 1.0801,
      "step": 14966
    },
    {
      "epoch": 2.781453261475562,
      "grad_norm": 0.9254017472267151,
      "learning_rate": 4.2562601772545975e-05,
      "loss": 1.0059,
      "step": 14967
    },
    {
      "epoch": 2.7816391005389334,
      "grad_norm": 0.984853982925415,
      "learning_rate": 4.2550643281138756e-05,
      "loss": 1.0964,
      "step": 14968
    },
    {
      "epoch": 2.7818249396023043,
      "grad_norm": 0.8644289970397949,
      "learning_rate": 4.2538686015889486e-05,
      "loss": 0.9188,
      "step": 14969
    },
    {
      "epoch": 2.7820107786656756,
      "grad_norm": 0.8592866063117981,
      "learning_rate": 4.2526729977053345e-05,
      "loss": 0.9523,
      "step": 14970
    },
    {
      "epoch": 2.7821966177290465,
      "grad_norm": 0.7572512030601501,
      "learning_rate": 4.2514775164885487e-05,
      "loss": 1.0383,
      "step": 14971
    },
    {
      "epoch": 2.782382456792418,
      "grad_norm": 0.8783193826675415,
      "learning_rate": 4.250282157964115e-05,
      "loss": 0.8557,
      "step": 14972
    },
    {
      "epoch": 2.782568295855789,
      "grad_norm": 1.02633535861969,
      "learning_rate": 4.2490869221575415e-05,
      "loss": 1.2725,
      "step": 14973
    },
    {
      "epoch": 2.78275413491916,
      "grad_norm": 0.7333510518074036,
      "learning_rate": 4.2478918090943346e-05,
      "loss": 1.0315,
      "step": 14974
    },
    {
      "epoch": 2.782939973982531,
      "grad_norm": 1.0128086805343628,
      "learning_rate": 4.246696818800006e-05,
      "loss": 0.8859,
      "step": 14975
    },
    {
      "epoch": 2.7831258130459022,
      "grad_norm": 0.8416959047317505,
      "learning_rate": 4.245501951300064e-05,
      "loss": 0.7682,
      "step": 14976
    },
    {
      "epoch": 2.7833116521092736,
      "grad_norm": 0.6340311169624329,
      "learning_rate": 4.244307206620009e-05,
      "loss": 0.7206,
      "step": 14977
    },
    {
      "epoch": 2.7834974911726444,
      "grad_norm": 0.8692113757133484,
      "learning_rate": 4.243112584785338e-05,
      "loss": 1.1894,
      "step": 14978
    },
    {
      "epoch": 2.7836833302360158,
      "grad_norm": 0.8255258202552795,
      "learning_rate": 4.241918085821547e-05,
      "loss": 0.8858,
      "step": 14979
    },
    {
      "epoch": 2.7838691692993867,
      "grad_norm": 0.9497602581977844,
      "learning_rate": 4.2407237097541364e-05,
      "loss": 1.0581,
      "step": 14980
    },
    {
      "epoch": 2.784055008362758,
      "grad_norm": 0.6905065178871155,
      "learning_rate": 4.239529456608595e-05,
      "loss": 0.8081,
      "step": 14981
    },
    {
      "epoch": 2.784240847426129,
      "grad_norm": 0.7528558373451233,
      "learning_rate": 4.238335326410408e-05,
      "loss": 1.0204,
      "step": 14982
    },
    {
      "epoch": 2.7844266864895,
      "grad_norm": 0.7711358666419983,
      "learning_rate": 4.237141319185071e-05,
      "loss": 0.869,
      "step": 14983
    },
    {
      "epoch": 2.784612525552871,
      "grad_norm": 0.9386207461357117,
      "learning_rate": 4.2359474349580616e-05,
      "loss": 0.9466,
      "step": 14984
    },
    {
      "epoch": 2.7847983646162424,
      "grad_norm": 0.8581230044364929,
      "learning_rate": 4.23475367375486e-05,
      "loss": 0.8827,
      "step": 14985
    },
    {
      "epoch": 2.7849842036796133,
      "grad_norm": 0.920850396156311,
      "learning_rate": 4.233560035600951e-05,
      "loss": 1.0693,
      "step": 14986
    },
    {
      "epoch": 2.7851700427429846,
      "grad_norm": 0.7897260785102844,
      "learning_rate": 4.232366520521803e-05,
      "loss": 0.8717,
      "step": 14987
    },
    {
      "epoch": 2.785355881806356,
      "grad_norm": 0.9491976499557495,
      "learning_rate": 4.231173128542899e-05,
      "loss": 0.9194,
      "step": 14988
    },
    {
      "epoch": 2.785541720869727,
      "grad_norm": 0.9895254373550415,
      "learning_rate": 4.229979859689704e-05,
      "loss": 0.9675,
      "step": 14989
    },
    {
      "epoch": 2.7857275599330977,
      "grad_norm": 0.8803383111953735,
      "learning_rate": 4.228786713987688e-05,
      "loss": 0.8465,
      "step": 14990
    },
    {
      "epoch": 2.785913398996469,
      "grad_norm": 0.9793798327445984,
      "learning_rate": 4.2275936914623105e-05,
      "loss": 1.2113,
      "step": 14991
    },
    {
      "epoch": 2.7860992380598404,
      "grad_norm": 0.8835542798042297,
      "learning_rate": 4.2264007921390446e-05,
      "loss": 0.9977,
      "step": 14992
    },
    {
      "epoch": 2.7862850771232113,
      "grad_norm": 0.7891272902488708,
      "learning_rate": 4.225208016043346e-05,
      "loss": 0.7701,
      "step": 14993
    },
    {
      "epoch": 2.786470916186582,
      "grad_norm": 0.8491330742835999,
      "learning_rate": 4.224015363200668e-05,
      "loss": 0.7749,
      "step": 14994
    },
    {
      "epoch": 2.7866567552499535,
      "grad_norm": 0.8505903482437134,
      "learning_rate": 4.222822833636472e-05,
      "loss": 1.1129,
      "step": 14995
    },
    {
      "epoch": 2.786842594313325,
      "grad_norm": 0.9466049671173096,
      "learning_rate": 4.2216304273762094e-05,
      "loss": 1.1278,
      "step": 14996
    },
    {
      "epoch": 2.7870284333766957,
      "grad_norm": 0.8390737771987915,
      "learning_rate": 4.220438144445331e-05,
      "loss": 0.9739,
      "step": 14997
    },
    {
      "epoch": 2.787214272440067,
      "grad_norm": 0.8704586625099182,
      "learning_rate": 4.219245984869282e-05,
      "loss": 0.9557,
      "step": 14998
    },
    {
      "epoch": 2.787400111503438,
      "grad_norm": 0.9225991368293762,
      "learning_rate": 4.2180539486735036e-05,
      "loss": 0.7677,
      "step": 14999
    },
    {
      "epoch": 2.7875859505668092,
      "grad_norm": 0.7355827689170837,
      "learning_rate": 4.216862035883445e-05,
      "loss": 0.8258,
      "step": 15000
    },
    {
      "epoch": 2.78777178963018,
      "grad_norm": 0.896327793598175,
      "learning_rate": 4.2156702465245424e-05,
      "loss": 0.9988,
      "step": 15001
    },
    {
      "epoch": 2.7879576286935515,
      "grad_norm": 0.7767103910446167,
      "learning_rate": 4.214478580622228e-05,
      "loss": 0.9715,
      "step": 15002
    },
    {
      "epoch": 2.7881434677569223,
      "grad_norm": 0.9789737462997437,
      "learning_rate": 4.213287038201943e-05,
      "loss": 0.957,
      "step": 15003
    },
    {
      "epoch": 2.7883293068202937,
      "grad_norm": 0.8596428632736206,
      "learning_rate": 4.212095619289117e-05,
      "loss": 1.0102,
      "step": 15004
    },
    {
      "epoch": 2.7885151458836646,
      "grad_norm": 0.7422680258750916,
      "learning_rate": 4.210904323909177e-05,
      "loss": 1.1146,
      "step": 15005
    },
    {
      "epoch": 2.788700984947036,
      "grad_norm": 0.6503537893295288,
      "learning_rate": 4.2097131520875456e-05,
      "loss": 0.5747,
      "step": 15006
    },
    {
      "epoch": 2.788886824010407,
      "grad_norm": 0.7762205600738525,
      "learning_rate": 4.2085221038496515e-05,
      "loss": 0.9154,
      "step": 15007
    },
    {
      "epoch": 2.789072663073778,
      "grad_norm": 0.9213268160820007,
      "learning_rate": 4.2073311792209166e-05,
      "loss": 0.739,
      "step": 15008
    },
    {
      "epoch": 2.789258502137149,
      "grad_norm": 0.8095465302467346,
      "learning_rate": 4.206140378226758e-05,
      "loss": 1.0868,
      "step": 15009
    },
    {
      "epoch": 2.7894443412005203,
      "grad_norm": 0.8037760257720947,
      "learning_rate": 4.204949700892591e-05,
      "loss": 0.8164,
      "step": 15010
    },
    {
      "epoch": 2.7896301802638916,
      "grad_norm": 0.9405101537704468,
      "learning_rate": 4.2037591472438245e-05,
      "loss": 0.9941,
      "step": 15011
    },
    {
      "epoch": 2.7898160193272625,
      "grad_norm": 0.8144403696060181,
      "learning_rate": 4.202568717305875e-05,
      "loss": 0.9442,
      "step": 15012
    },
    {
      "epoch": 2.790001858390634,
      "grad_norm": 0.905034065246582,
      "learning_rate": 4.201378411104146e-05,
      "loss": 0.9174,
      "step": 15013
    },
    {
      "epoch": 2.7901876974540047,
      "grad_norm": 0.8545786142349243,
      "learning_rate": 4.200188228664043e-05,
      "loss": 0.7051,
      "step": 15014
    },
    {
      "epoch": 2.790373536517376,
      "grad_norm": 0.9535547494888306,
      "learning_rate": 4.198998170010971e-05,
      "loss": 0.8842,
      "step": 15015
    },
    {
      "epoch": 2.790559375580747,
      "grad_norm": 0.9396203756332397,
      "learning_rate": 4.197808235170326e-05,
      "loss": 0.7345,
      "step": 15016
    },
    {
      "epoch": 2.7907452146441183,
      "grad_norm": 1.2389658689498901,
      "learning_rate": 4.19661842416751e-05,
      "loss": 1.2049,
      "step": 15017
    },
    {
      "epoch": 2.790931053707489,
      "grad_norm": 0.7988381385803223,
      "learning_rate": 4.1954287370279155e-05,
      "loss": 0.8315,
      "step": 15018
    },
    {
      "epoch": 2.7911168927708605,
      "grad_norm": 0.8601673245429993,
      "learning_rate": 4.194239173776929e-05,
      "loss": 1.1607,
      "step": 15019
    },
    {
      "epoch": 2.7913027318342314,
      "grad_norm": 0.9635416269302368,
      "learning_rate": 4.193049734439949e-05,
      "loss": 0.7352,
      "step": 15020
    },
    {
      "epoch": 2.7914885708976027,
      "grad_norm": 0.886246383190155,
      "learning_rate": 4.1918604190423564e-05,
      "loss": 1.0577,
      "step": 15021
    },
    {
      "epoch": 2.791674409960974,
      "grad_norm": 0.7917489409446716,
      "learning_rate": 4.1906712276095325e-05,
      "loss": 0.8374,
      "step": 15022
    },
    {
      "epoch": 2.791860249024345,
      "grad_norm": 0.9550040364265442,
      "learning_rate": 4.189482160166865e-05,
      "loss": 0.7641,
      "step": 15023
    },
    {
      "epoch": 2.792046088087716,
      "grad_norm": 0.8474428653717041,
      "learning_rate": 4.188293216739732e-05,
      "loss": 1.126,
      "step": 15024
    },
    {
      "epoch": 2.792231927151087,
      "grad_norm": 0.7299398183822632,
      "learning_rate": 4.187104397353504e-05,
      "loss": 0.7369,
      "step": 15025
    },
    {
      "epoch": 2.7924177662144585,
      "grad_norm": 0.9598284363746643,
      "learning_rate": 4.185915702033555e-05,
      "loss": 0.6783,
      "step": 15026
    },
    {
      "epoch": 2.7926036052778294,
      "grad_norm": 0.8690744638442993,
      "learning_rate": 4.1847271308052583e-05,
      "loss": 0.8888,
      "step": 15027
    },
    {
      "epoch": 2.7927894443412002,
      "grad_norm": 0.8142648935317993,
      "learning_rate": 4.183538683693985e-05,
      "loss": 1.0293,
      "step": 15028
    },
    {
      "epoch": 2.7929752834045716,
      "grad_norm": 0.9416887164115906,
      "learning_rate": 4.182350360725097e-05,
      "loss": 0.8944,
      "step": 15029
    },
    {
      "epoch": 2.793161122467943,
      "grad_norm": 1.0445160865783691,
      "learning_rate": 4.181162161923957e-05,
      "loss": 1.0746,
      "step": 15030
    },
    {
      "epoch": 2.793346961531314,
      "grad_norm": 0.922220766544342,
      "learning_rate": 4.179974087315921e-05,
      "loss": 0.9097,
      "step": 15031
    },
    {
      "epoch": 2.793532800594685,
      "grad_norm": 0.7769144773483276,
      "learning_rate": 4.1787861369263546e-05,
      "loss": 0.8806,
      "step": 15032
    },
    {
      "epoch": 2.793718639658056,
      "grad_norm": 1.0614155530929565,
      "learning_rate": 4.1775983107806086e-05,
      "loss": 0.8616,
      "step": 15033
    },
    {
      "epoch": 2.7939044787214273,
      "grad_norm": 0.803954541683197,
      "learning_rate": 4.176410608904031e-05,
      "loss": 0.8816,
      "step": 15034
    },
    {
      "epoch": 2.794090317784798,
      "grad_norm": 0.7549547553062439,
      "learning_rate": 4.17522303132198e-05,
      "loss": 0.8736,
      "step": 15035
    },
    {
      "epoch": 2.7942761568481695,
      "grad_norm": 0.9446969032287598,
      "learning_rate": 4.174035578059796e-05,
      "loss": 0.889,
      "step": 15036
    },
    {
      "epoch": 2.794461995911541,
      "grad_norm": 0.8554919362068176,
      "learning_rate": 4.1728482491428224e-05,
      "loss": 0.9947,
      "step": 15037
    },
    {
      "epoch": 2.7946478349749118,
      "grad_norm": 0.8129143714904785,
      "learning_rate": 4.171661044596407e-05,
      "loss": 0.8565,
      "step": 15038
    },
    {
      "epoch": 2.7948336740382826,
      "grad_norm": 0.8183847069740295,
      "learning_rate": 4.17047396444588e-05,
      "loss": 0.9955,
      "step": 15039
    },
    {
      "epoch": 2.795019513101654,
      "grad_norm": 0.7391417622566223,
      "learning_rate": 4.169287008716587e-05,
      "loss": 0.9115,
      "step": 15040
    },
    {
      "epoch": 2.7952053521650253,
      "grad_norm": 0.9141026139259338,
      "learning_rate": 4.1681001774338566e-05,
      "loss": 0.8662,
      "step": 15041
    },
    {
      "epoch": 2.795391191228396,
      "grad_norm": 0.9632161259651184,
      "learning_rate": 4.166913470623016e-05,
      "loss": 0.8851,
      "step": 15042
    },
    {
      "epoch": 2.795577030291767,
      "grad_norm": 0.8051280975341797,
      "learning_rate": 4.165726888309402e-05,
      "loss": 0.9795,
      "step": 15043
    },
    {
      "epoch": 2.7957628693551384,
      "grad_norm": 0.8322173953056335,
      "learning_rate": 4.1645404305183345e-05,
      "loss": 0.7878,
      "step": 15044
    },
    {
      "epoch": 2.7959487084185097,
      "grad_norm": 0.8591897487640381,
      "learning_rate": 4.163354097275137e-05,
      "loss": 1.0312,
      "step": 15045
    },
    {
      "epoch": 2.7961345474818806,
      "grad_norm": 0.9375728368759155,
      "learning_rate": 4.162167888605127e-05,
      "loss": 1.0866,
      "step": 15046
    },
    {
      "epoch": 2.796320386545252,
      "grad_norm": 0.8724905252456665,
      "learning_rate": 4.16098180453363e-05,
      "loss": 0.784,
      "step": 15047
    },
    {
      "epoch": 2.796506225608623,
      "grad_norm": 0.8089430928230286,
      "learning_rate": 4.159795845085952e-05,
      "loss": 0.6377,
      "step": 15048
    },
    {
      "epoch": 2.796692064671994,
      "grad_norm": 0.838665246963501,
      "learning_rate": 4.158610010287413e-05,
      "loss": 0.967,
      "step": 15049
    },
    {
      "epoch": 2.796877903735365,
      "grad_norm": 0.7745414972305298,
      "learning_rate": 4.15742430016332e-05,
      "loss": 0.848,
      "step": 15050
    },
    {
      "epoch": 2.7970637427987364,
      "grad_norm": 0.8479868769645691,
      "learning_rate": 4.1562387147389735e-05,
      "loss": 0.9451,
      "step": 15051
    },
    {
      "epoch": 2.7972495818621073,
      "grad_norm": 0.9081116914749146,
      "learning_rate": 4.155053254039688e-05,
      "loss": 0.9511,
      "step": 15052
    },
    {
      "epoch": 2.7974354209254786,
      "grad_norm": 0.7565174102783203,
      "learning_rate": 4.15386791809076e-05,
      "loss": 0.8633,
      "step": 15053
    },
    {
      "epoch": 2.7976212599888495,
      "grad_norm": 0.7815892100334167,
      "learning_rate": 4.152682706917487e-05,
      "loss": 0.9766,
      "step": 15054
    },
    {
      "epoch": 2.797807099052221,
      "grad_norm": 0.9136144518852234,
      "learning_rate": 4.1514976205451696e-05,
      "loss": 1.0633,
      "step": 15055
    },
    {
      "epoch": 2.797992938115592,
      "grad_norm": 0.9465920329093933,
      "learning_rate": 4.1503126589990984e-05,
      "loss": 1.0764,
      "step": 15056
    },
    {
      "epoch": 2.798178777178963,
      "grad_norm": 0.9290378093719482,
      "learning_rate": 4.149127822304566e-05,
      "loss": 0.9421,
      "step": 15057
    },
    {
      "epoch": 2.798364616242334,
      "grad_norm": 0.7892298698425293,
      "learning_rate": 4.147943110486856e-05,
      "loss": 1.0366,
      "step": 15058
    },
    {
      "epoch": 2.7985504553057052,
      "grad_norm": 0.9133583307266235,
      "learning_rate": 4.146758523571257e-05,
      "loss": 0.966,
      "step": 15059
    },
    {
      "epoch": 2.7987362943690766,
      "grad_norm": 0.9047538042068481,
      "learning_rate": 4.1455740615830574e-05,
      "loss": 1.0607,
      "step": 15060
    },
    {
      "epoch": 2.7989221334324474,
      "grad_norm": 1.4756627082824707,
      "learning_rate": 4.1443897245475315e-05,
      "loss": 1.1325,
      "step": 15061
    },
    {
      "epoch": 2.7991079724958188,
      "grad_norm": 0.9655285477638245,
      "learning_rate": 4.143205512489956e-05,
      "loss": 0.7489,
      "step": 15062
    },
    {
      "epoch": 2.7992938115591897,
      "grad_norm": 0.73797208070755,
      "learning_rate": 4.142021425435612e-05,
      "loss": 0.6195,
      "step": 15063
    },
    {
      "epoch": 2.799479650622561,
      "grad_norm": 0.8950634598731995,
      "learning_rate": 4.140837463409767e-05,
      "loss": 0.9471,
      "step": 15064
    },
    {
      "epoch": 2.799665489685932,
      "grad_norm": 0.909694492816925,
      "learning_rate": 4.139653626437692e-05,
      "loss": 0.8666,
      "step": 15065
    },
    {
      "epoch": 2.799851328749303,
      "grad_norm": 0.7249810099601746,
      "learning_rate": 4.1384699145446495e-05,
      "loss": 0.6832,
      "step": 15066
    },
    {
      "epoch": 2.800037167812674,
      "grad_norm": 0.869342029094696,
      "learning_rate": 4.137286327755913e-05,
      "loss": 0.984,
      "step": 15067
    },
    {
      "epoch": 2.8002230068760454,
      "grad_norm": 0.6247124671936035,
      "learning_rate": 4.136102866096737e-05,
      "loss": 0.6172,
      "step": 15068
    },
    {
      "epoch": 2.8004088459394163,
      "grad_norm": 0.7755012512207031,
      "learning_rate": 4.13491952959238e-05,
      "loss": 0.9734,
      "step": 15069
    },
    {
      "epoch": 2.8005946850027876,
      "grad_norm": 0.719447910785675,
      "learning_rate": 4.1337363182681046e-05,
      "loss": 0.8898,
      "step": 15070
    },
    {
      "epoch": 2.800780524066159,
      "grad_norm": 0.8762292861938477,
      "learning_rate": 4.1325532321491555e-05,
      "loss": 0.997,
      "step": 15071
    },
    {
      "epoch": 2.80096636312953,
      "grad_norm": 0.982386589050293,
      "learning_rate": 4.131370271260794e-05,
      "loss": 0.993,
      "step": 15072
    },
    {
      "epoch": 2.8011522021929007,
      "grad_norm": 0.7873516082763672,
      "learning_rate": 4.1301874356282624e-05,
      "loss": 1.0242,
      "step": 15073
    },
    {
      "epoch": 2.801338041256272,
      "grad_norm": 0.8276214003562927,
      "learning_rate": 4.1290047252768036e-05,
      "loss": 0.9155,
      "step": 15074
    },
    {
      "epoch": 2.8015238803196434,
      "grad_norm": 0.9610852599143982,
      "learning_rate": 4.1278221402316674e-05,
      "loss": 0.9898,
      "step": 15075
    },
    {
      "epoch": 2.8017097193830143,
      "grad_norm": 0.762342631816864,
      "learning_rate": 4.1266396805180905e-05,
      "loss": 1.0075,
      "step": 15076
    },
    {
      "epoch": 2.801895558446385,
      "grad_norm": 1.317216396331787,
      "learning_rate": 4.12545734616131e-05,
      "loss": 1.2906,
      "step": 15077
    },
    {
      "epoch": 2.8020813975097565,
      "grad_norm": 0.8017038106918335,
      "learning_rate": 4.124275137186559e-05,
      "loss": 0.7105,
      "step": 15078
    },
    {
      "epoch": 2.802267236573128,
      "grad_norm": 1.9257043600082397,
      "learning_rate": 4.1230930536190715e-05,
      "loss": 1.3639,
      "step": 15079
    },
    {
      "epoch": 2.8024530756364987,
      "grad_norm": 0.8237961530685425,
      "learning_rate": 4.12191109548408e-05,
      "loss": 0.8367,
      "step": 15080
    },
    {
      "epoch": 2.80263891469987,
      "grad_norm": 0.7258276343345642,
      "learning_rate": 4.1207292628068106e-05,
      "loss": 0.7606,
      "step": 15081
    },
    {
      "epoch": 2.802824753763241,
      "grad_norm": 0.8935003280639648,
      "learning_rate": 4.119547555612482e-05,
      "loss": 0.97,
      "step": 15082
    },
    {
      "epoch": 2.8030105928266122,
      "grad_norm": 0.7853168845176697,
      "learning_rate": 4.118365973926322e-05,
      "loss": 0.781,
      "step": 15083
    },
    {
      "epoch": 2.803196431889983,
      "grad_norm": 0.853007435798645,
      "learning_rate": 4.117184517773548e-05,
      "loss": 0.9727,
      "step": 15084
    },
    {
      "epoch": 2.8033822709533545,
      "grad_norm": 0.9908416867256165,
      "learning_rate": 4.116003187179375e-05,
      "loss": 0.9773,
      "step": 15085
    },
    {
      "epoch": 2.803568110016726,
      "grad_norm": 0.8514252305030823,
      "learning_rate": 4.114821982169013e-05,
      "loss": 1.0044,
      "step": 15086
    },
    {
      "epoch": 2.8037539490800967,
      "grad_norm": 0.9149060249328613,
      "learning_rate": 4.11364090276768e-05,
      "loss": 1.1324,
      "step": 15087
    },
    {
      "epoch": 2.8039397881434676,
      "grad_norm": 0.9491459727287292,
      "learning_rate": 4.11245994900058e-05,
      "loss": 0.9364,
      "step": 15088
    },
    {
      "epoch": 2.804125627206839,
      "grad_norm": 0.7908619046211243,
      "learning_rate": 4.11127912089292e-05,
      "loss": 0.8661,
      "step": 15089
    },
    {
      "epoch": 2.80431146627021,
      "grad_norm": 0.775545060634613,
      "learning_rate": 4.110098418469898e-05,
      "loss": 0.8741,
      "step": 15090
    },
    {
      "epoch": 2.804497305333581,
      "grad_norm": 0.7879462242126465,
      "learning_rate": 4.1089178417567164e-05,
      "loss": 0.8878,
      "step": 15091
    },
    {
      "epoch": 2.804683144396952,
      "grad_norm": 0.9746586084365845,
      "learning_rate": 4.107737390778579e-05,
      "loss": 0.6941,
      "step": 15092
    },
    {
      "epoch": 2.8048689834603233,
      "grad_norm": 0.7224279046058655,
      "learning_rate": 4.1065570655606754e-05,
      "loss": 0.5672,
      "step": 15093
    },
    {
      "epoch": 2.8050548225236946,
      "grad_norm": 0.7029162049293518,
      "learning_rate": 4.105376866128193e-05,
      "loss": 0.8019,
      "step": 15094
    },
    {
      "epoch": 2.8052406615870655,
      "grad_norm": 0.9751745462417603,
      "learning_rate": 4.104196792506331e-05,
      "loss": 0.7743,
      "step": 15095
    },
    {
      "epoch": 2.805426500650437,
      "grad_norm": 0.9375293850898743,
      "learning_rate": 4.10301684472027e-05,
      "loss": 0.8651,
      "step": 15096
    },
    {
      "epoch": 2.8056123397138077,
      "grad_norm": 0.9242647886276245,
      "learning_rate": 4.101837022795194e-05,
      "loss": 1.0177,
      "step": 15097
    },
    {
      "epoch": 2.805798178777179,
      "grad_norm": 0.8099009394645691,
      "learning_rate": 4.100657326756282e-05,
      "loss": 0.8403,
      "step": 15098
    },
    {
      "epoch": 2.80598401784055,
      "grad_norm": 0.811262309551239,
      "learning_rate": 4.0994777566287204e-05,
      "loss": 0.9109,
      "step": 15099
    },
    {
      "epoch": 2.8061698569039213,
      "grad_norm": 0.8253323435783386,
      "learning_rate": 4.0982983124376763e-05,
      "loss": 0.708,
      "step": 15100
    },
    {
      "epoch": 2.806355695967292,
      "grad_norm": 0.8106323480606079,
      "learning_rate": 4.09711899420833e-05,
      "loss": 1.1688,
      "step": 15101
    },
    {
      "epoch": 2.8065415350306635,
      "grad_norm": 0.7565474510192871,
      "learning_rate": 4.0959398019658456e-05,
      "loss": 0.9421,
      "step": 15102
    },
    {
      "epoch": 2.8067273740940344,
      "grad_norm": 0.8208668828010559,
      "learning_rate": 4.094760735735398e-05,
      "loss": 0.8534,
      "step": 15103
    },
    {
      "epoch": 2.8069132131574057,
      "grad_norm": 0.8241255879402161,
      "learning_rate": 4.093581795542149e-05,
      "loss": 1.0367,
      "step": 15104
    },
    {
      "epoch": 2.807099052220777,
      "grad_norm": 0.9445957541465759,
      "learning_rate": 4.0924029814112594e-05,
      "loss": 0.9277,
      "step": 15105
    },
    {
      "epoch": 2.807284891284148,
      "grad_norm": 0.8766359686851501,
      "learning_rate": 4.091224293367888e-05,
      "loss": 0.8135,
      "step": 15106
    },
    {
      "epoch": 2.807470730347519,
      "grad_norm": 0.8228170275688171,
      "learning_rate": 4.090045731437198e-05,
      "loss": 0.7023,
      "step": 15107
    },
    {
      "epoch": 2.80765656941089,
      "grad_norm": 0.8613119721412659,
      "learning_rate": 4.08886729564434e-05,
      "loss": 0.9599,
      "step": 15108
    },
    {
      "epoch": 2.8078424084742615,
      "grad_norm": 0.8965815901756287,
      "learning_rate": 4.087688986014464e-05,
      "loss": 0.9747,
      "step": 15109
    },
    {
      "epoch": 2.8080282475376324,
      "grad_norm": 0.8584528565406799,
      "learning_rate": 4.086510802572718e-05,
      "loss": 0.9173,
      "step": 15110
    },
    {
      "epoch": 2.8082140866010037,
      "grad_norm": 0.7747320532798767,
      "learning_rate": 4.0853327453442525e-05,
      "loss": 0.9965,
      "step": 15111
    },
    {
      "epoch": 2.8083999256643746,
      "grad_norm": 0.8218334913253784,
      "learning_rate": 4.0841548143542116e-05,
      "loss": 0.936,
      "step": 15112
    },
    {
      "epoch": 2.808585764727746,
      "grad_norm": 0.8838158845901489,
      "learning_rate": 4.082977009627734e-05,
      "loss": 0.9228,
      "step": 15113
    },
    {
      "epoch": 2.808771603791117,
      "grad_norm": 0.8490532040596008,
      "learning_rate": 4.081799331189955e-05,
      "loss": 0.9187,
      "step": 15114
    },
    {
      "epoch": 2.808957442854488,
      "grad_norm": 0.7764754295349121,
      "learning_rate": 4.0806217790660176e-05,
      "loss": 0.8403,
      "step": 15115
    },
    {
      "epoch": 2.809143281917859,
      "grad_norm": 0.7498995661735535,
      "learning_rate": 4.079444353281049e-05,
      "loss": 0.8031,
      "step": 15116
    },
    {
      "epoch": 2.8093291209812303,
      "grad_norm": 1.1498351097106934,
      "learning_rate": 4.0782670538601806e-05,
      "loss": 1.1147,
      "step": 15117
    },
    {
      "epoch": 2.809514960044601,
      "grad_norm": 0.9433811902999878,
      "learning_rate": 4.077089880828537e-05,
      "loss": 0.5933,
      "step": 15118
    },
    {
      "epoch": 2.8097007991079725,
      "grad_norm": 0.7371313571929932,
      "learning_rate": 4.07591283421125e-05,
      "loss": 0.8163,
      "step": 15119
    },
    {
      "epoch": 2.809886638171344,
      "grad_norm": 1.1072052717208862,
      "learning_rate": 4.074735914033436e-05,
      "loss": 0.8688,
      "step": 15120
    },
    {
      "epoch": 2.8100724772347148,
      "grad_norm": 0.973744809627533,
      "learning_rate": 4.0735591203202125e-05,
      "loss": 0.9479,
      "step": 15121
    },
    {
      "epoch": 2.8102583162980856,
      "grad_norm": 0.7982321977615356,
      "learning_rate": 4.0723824530967004e-05,
      "loss": 0.8084,
      "step": 15122
    },
    {
      "epoch": 2.810444155361457,
      "grad_norm": 0.8149759769439697,
      "learning_rate": 4.0712059123880155e-05,
      "loss": 0.8018,
      "step": 15123
    },
    {
      "epoch": 2.8106299944248283,
      "grad_norm": 0.8275374174118042,
      "learning_rate": 4.070029498219266e-05,
      "loss": 0.9292,
      "step": 15124
    },
    {
      "epoch": 2.810815833488199,
      "grad_norm": 0.7239959836006165,
      "learning_rate": 4.0688532106155606e-05,
      "loss": 0.9979,
      "step": 15125
    },
    {
      "epoch": 2.81100167255157,
      "grad_norm": 0.8658648133277893,
      "learning_rate": 4.067677049602001e-05,
      "loss": 1.0328,
      "step": 15126
    },
    {
      "epoch": 2.8111875116149414,
      "grad_norm": 0.8098410964012146,
      "learning_rate": 4.0665010152036985e-05,
      "loss": 0.7847,
      "step": 15127
    },
    {
      "epoch": 2.8113733506783127,
      "grad_norm": 0.722276508808136,
      "learning_rate": 4.065325107445751e-05,
      "loss": 0.8967,
      "step": 15128
    },
    {
      "epoch": 2.8115591897416836,
      "grad_norm": 0.9402121305465698,
      "learning_rate": 4.0641493263532524e-05,
      "loss": 0.8615,
      "step": 15129
    },
    {
      "epoch": 2.811745028805055,
      "grad_norm": 0.8405276536941528,
      "learning_rate": 4.062973671951299e-05,
      "loss": 0.933,
      "step": 15130
    },
    {
      "epoch": 2.811930867868426,
      "grad_norm": 0.943343997001648,
      "learning_rate": 4.0617981442649855e-05,
      "loss": 0.9931,
      "step": 15131
    },
    {
      "epoch": 2.812116706931797,
      "grad_norm": 0.8641842007637024,
      "learning_rate": 4.060622743319398e-05,
      "loss": 0.9604,
      "step": 15132
    },
    {
      "epoch": 2.812302545995168,
      "grad_norm": 0.8120639324188232,
      "learning_rate": 4.059447469139629e-05,
      "loss": 0.9545,
      "step": 15133
    },
    {
      "epoch": 2.8124883850585394,
      "grad_norm": 0.9698015451431274,
      "learning_rate": 4.0582723217507554e-05,
      "loss": 0.9675,
      "step": 15134
    },
    {
      "epoch": 2.8126742241219103,
      "grad_norm": 0.8321253061294556,
      "learning_rate": 4.0570973011778665e-05,
      "loss": 1.0917,
      "step": 15135
    },
    {
      "epoch": 2.8128600631852816,
      "grad_norm": 0.7853554487228394,
      "learning_rate": 4.055922407446038e-05,
      "loss": 0.9034,
      "step": 15136
    },
    {
      "epoch": 2.8130459022486525,
      "grad_norm": 1.1400892734527588,
      "learning_rate": 4.054747640580344e-05,
      "loss": 1.0301,
      "step": 15137
    },
    {
      "epoch": 2.813231741312024,
      "grad_norm": 0.802152156829834,
      "learning_rate": 4.053573000605857e-05,
      "loss": 0.8063,
      "step": 15138
    },
    {
      "epoch": 2.813417580375395,
      "grad_norm": 0.7391362190246582,
      "learning_rate": 4.052398487547653e-05,
      "loss": 0.8216,
      "step": 15139
    },
    {
      "epoch": 2.813603419438766,
      "grad_norm": 0.8531056046485901,
      "learning_rate": 4.051224101430796e-05,
      "loss": 0.7364,
      "step": 15140
    },
    {
      "epoch": 2.813789258502137,
      "grad_norm": 0.7853450179100037,
      "learning_rate": 4.050049842280352e-05,
      "loss": 0.8467,
      "step": 15141
    },
    {
      "epoch": 2.8139750975655082,
      "grad_norm": 0.738269567489624,
      "learning_rate": 4.048875710121377e-05,
      "loss": 0.9664,
      "step": 15142
    },
    {
      "epoch": 2.8141609366288796,
      "grad_norm": 0.8238461017608643,
      "learning_rate": 4.047701704978944e-05,
      "loss": 1.1125,
      "step": 15143
    },
    {
      "epoch": 2.8143467756922504,
      "grad_norm": 0.897844672203064,
      "learning_rate": 4.046527826878105e-05,
      "loss": 0.8101,
      "step": 15144
    },
    {
      "epoch": 2.8145326147556218,
      "grad_norm": 0.7138451933860779,
      "learning_rate": 4.045354075843911e-05,
      "loss": 0.8195,
      "step": 15145
    },
    {
      "epoch": 2.8147184538189927,
      "grad_norm": 0.8560433983802795,
      "learning_rate": 4.044180451901414e-05,
      "loss": 0.7273,
      "step": 15146
    },
    {
      "epoch": 2.814904292882364,
      "grad_norm": 0.7499582171440125,
      "learning_rate": 4.0430069550756665e-05,
      "loss": 0.953,
      "step": 15147
    },
    {
      "epoch": 2.815090131945735,
      "grad_norm": 0.95637446641922,
      "learning_rate": 4.0418335853917145e-05,
      "loss": 1.0183,
      "step": 15148
    },
    {
      "epoch": 2.815275971009106,
      "grad_norm": 0.8454667329788208,
      "learning_rate": 4.0406603428745996e-05,
      "loss": 0.8944,
      "step": 15149
    },
    {
      "epoch": 2.815461810072477,
      "grad_norm": 0.7893767356872559,
      "learning_rate": 4.039487227549359e-05,
      "loss": 0.9952,
      "step": 15150
    },
    {
      "epoch": 2.8156476491358484,
      "grad_norm": 0.8175316452980042,
      "learning_rate": 4.038314239441038e-05,
      "loss": 0.8697,
      "step": 15151
    },
    {
      "epoch": 2.8158334881992193,
      "grad_norm": 0.7143104672431946,
      "learning_rate": 4.037141378574669e-05,
      "loss": 0.6834,
      "step": 15152
    },
    {
      "epoch": 2.8160193272625906,
      "grad_norm": 0.7967771291732788,
      "learning_rate": 4.035968644975282e-05,
      "loss": 0.8128,
      "step": 15153
    },
    {
      "epoch": 2.816205166325962,
      "grad_norm": 0.878592312335968,
      "learning_rate": 4.03479603866791e-05,
      "loss": 0.9127,
      "step": 15154
    },
    {
      "epoch": 2.816391005389333,
      "grad_norm": 0.8325759172439575,
      "learning_rate": 4.033623559677582e-05,
      "loss": 1.088,
      "step": 15155
    },
    {
      "epoch": 2.8165768444527037,
      "grad_norm": 0.9697619080543518,
      "learning_rate": 4.032451208029321e-05,
      "loss": 0.9117,
      "step": 15156
    },
    {
      "epoch": 2.816762683516075,
      "grad_norm": 0.7757670283317566,
      "learning_rate": 4.031278983748148e-05,
      "loss": 0.7649,
      "step": 15157
    },
    {
      "epoch": 2.8169485225794464,
      "grad_norm": 0.842522144317627,
      "learning_rate": 4.030106886859079e-05,
      "loss": 0.9733,
      "step": 15158
    },
    {
      "epoch": 2.8171343616428173,
      "grad_norm": 0.7370792031288147,
      "learning_rate": 4.028934917387137e-05,
      "loss": 0.8486,
      "step": 15159
    },
    {
      "epoch": 2.8173202007061886,
      "grad_norm": 0.8743408918380737,
      "learning_rate": 4.027763075357333e-05,
      "loss": 0.8942,
      "step": 15160
    },
    {
      "epoch": 2.8175060397695595,
      "grad_norm": 0.8407097458839417,
      "learning_rate": 4.026591360794677e-05,
      "loss": 0.979,
      "step": 15161
    },
    {
      "epoch": 2.817691878832931,
      "grad_norm": 0.8373445272445679,
      "learning_rate": 4.025419773724174e-05,
      "loss": 0.9727,
      "step": 15162
    },
    {
      "epoch": 2.8178777178963017,
      "grad_norm": 0.7689995765686035,
      "learning_rate": 4.0242483141708334e-05,
      "loss": 0.8183,
      "step": 15163
    },
    {
      "epoch": 2.818063556959673,
      "grad_norm": 0.959061861038208,
      "learning_rate": 4.0230769821596605e-05,
      "loss": 1.0507,
      "step": 15164
    },
    {
      "epoch": 2.818249396023044,
      "grad_norm": 0.756231963634491,
      "learning_rate": 4.0219057777156536e-05,
      "loss": 0.9184,
      "step": 15165
    },
    {
      "epoch": 2.8184352350864152,
      "grad_norm": 0.864276111125946,
      "learning_rate": 4.0207347008638055e-05,
      "loss": 0.8817,
      "step": 15166
    },
    {
      "epoch": 2.818621074149786,
      "grad_norm": 0.6302124261856079,
      "learning_rate": 4.019563751629117e-05,
      "loss": 0.6543,
      "step": 15167
    },
    {
      "epoch": 2.8188069132131575,
      "grad_norm": 0.8549239039421082,
      "learning_rate": 4.0183929300365783e-05,
      "loss": 0.8403,
      "step": 15168
    },
    {
      "epoch": 2.818992752276529,
      "grad_norm": 0.8719046711921692,
      "learning_rate": 4.017222236111178e-05,
      "loss": 1.0681,
      "step": 15169
    },
    {
      "epoch": 2.8191785913398997,
      "grad_norm": 0.7814198136329651,
      "learning_rate": 4.016051669877897e-05,
      "loss": 0.9229,
      "step": 15170
    },
    {
      "epoch": 2.8193644304032706,
      "grad_norm": 0.8007107377052307,
      "learning_rate": 4.014881231361729e-05,
      "loss": 0.8217,
      "step": 15171
    },
    {
      "epoch": 2.819550269466642,
      "grad_norm": 0.6752849817276001,
      "learning_rate": 4.01371092058765e-05,
      "loss": 0.684,
      "step": 15172
    },
    {
      "epoch": 2.819736108530013,
      "grad_norm": 0.7651057839393616,
      "learning_rate": 4.012540737580635e-05,
      "loss": 0.7586,
      "step": 15173
    },
    {
      "epoch": 2.819921947593384,
      "grad_norm": 0.7388724684715271,
      "learning_rate": 4.0113706823656625e-05,
      "loss": 0.9782,
      "step": 15174
    },
    {
      "epoch": 2.820107786656755,
      "grad_norm": 1.1076356172561646,
      "learning_rate": 4.0102007549677114e-05,
      "loss": 0.7971,
      "step": 15175
    },
    {
      "epoch": 2.8202936257201263,
      "grad_norm": 0.8896591663360596,
      "learning_rate": 4.0090309554117444e-05,
      "loss": 0.6954,
      "step": 15176
    },
    {
      "epoch": 2.8204794647834976,
      "grad_norm": 0.782341718673706,
      "learning_rate": 4.007861283722731e-05,
      "loss": 0.8519,
      "step": 15177
    },
    {
      "epoch": 2.8206653038468685,
      "grad_norm": 1.0588573217391968,
      "learning_rate": 4.006691739925632e-05,
      "loss": 0.6504,
      "step": 15178
    },
    {
      "epoch": 2.82085114291024,
      "grad_norm": 0.7600374817848206,
      "learning_rate": 4.0055223240454167e-05,
      "loss": 0.8186,
      "step": 15179
    },
    {
      "epoch": 2.8210369819736107,
      "grad_norm": 0.7226833701133728,
      "learning_rate": 4.0043530361070405e-05,
      "loss": 1.042,
      "step": 15180
    },
    {
      "epoch": 2.821222821036982,
      "grad_norm": 1.0568703413009644,
      "learning_rate": 4.00318387613546e-05,
      "loss": 0.929,
      "step": 15181
    },
    {
      "epoch": 2.821408660100353,
      "grad_norm": 0.8555261492729187,
      "learning_rate": 4.002014844155625e-05,
      "loss": 0.7783,
      "step": 15182
    },
    {
      "epoch": 2.8215944991637243,
      "grad_norm": 0.9463980793952942,
      "learning_rate": 4.0008459401924936e-05,
      "loss": 0.7808,
      "step": 15183
    },
    {
      "epoch": 2.821780338227095,
      "grad_norm": 0.8411297798156738,
      "learning_rate": 3.999677164271007e-05,
      "loss": 0.9684,
      "step": 15184
    },
    {
      "epoch": 2.8219661772904665,
      "grad_norm": 0.7953760623931885,
      "learning_rate": 3.998508516416118e-05,
      "loss": 0.8714,
      "step": 15185
    },
    {
      "epoch": 2.8221520163538374,
      "grad_norm": 0.90069979429245,
      "learning_rate": 3.997339996652762e-05,
      "loss": 0.8948,
      "step": 15186
    },
    {
      "epoch": 2.8223378554172087,
      "grad_norm": 0.9424738883972168,
      "learning_rate": 3.996171605005887e-05,
      "loss": 1.0742,
      "step": 15187
    },
    {
      "epoch": 2.82252369448058,
      "grad_norm": 0.9908680319786072,
      "learning_rate": 3.995003341500426e-05,
      "loss": 0.9633,
      "step": 15188
    },
    {
      "epoch": 2.822709533543951,
      "grad_norm": 0.718686580657959,
      "learning_rate": 3.993835206161313e-05,
      "loss": 0.7854,
      "step": 15189
    },
    {
      "epoch": 2.822895372607322,
      "grad_norm": 0.7781935334205627,
      "learning_rate": 3.992667199013478e-05,
      "loss": 0.9473,
      "step": 15190
    },
    {
      "epoch": 2.823081211670693,
      "grad_norm": 0.7772121429443359,
      "learning_rate": 3.991499320081855e-05,
      "loss": 0.7848,
      "step": 15191
    },
    {
      "epoch": 2.8232670507340645,
      "grad_norm": 0.8744450807571411,
      "learning_rate": 3.990331569391369e-05,
      "loss": 0.8523,
      "step": 15192
    },
    {
      "epoch": 2.8234528897974354,
      "grad_norm": 0.751013457775116,
      "learning_rate": 3.9891639469669384e-05,
      "loss": 0.8262,
      "step": 15193
    },
    {
      "epoch": 2.8236387288608067,
      "grad_norm": 0.8620365262031555,
      "learning_rate": 3.987996452833492e-05,
      "loss": 0.5675,
      "step": 15194
    },
    {
      "epoch": 2.8238245679241776,
      "grad_norm": 1.0090669393539429,
      "learning_rate": 3.9868290870159405e-05,
      "loss": 1.1258,
      "step": 15195
    },
    {
      "epoch": 2.824010406987549,
      "grad_norm": 0.9945322871208191,
      "learning_rate": 3.9856618495392074e-05,
      "loss": 0.7712,
      "step": 15196
    },
    {
      "epoch": 2.82419624605092,
      "grad_norm": 0.9460313320159912,
      "learning_rate": 3.9844947404282006e-05,
      "loss": 0.9587,
      "step": 15197
    },
    {
      "epoch": 2.824382085114291,
      "grad_norm": 0.6281357407569885,
      "learning_rate": 3.9833277597078275e-05,
      "loss": 0.5446,
      "step": 15198
    },
    {
      "epoch": 2.824567924177662,
      "grad_norm": 0.9807770848274231,
      "learning_rate": 3.982160907403002e-05,
      "loss": 1.1381,
      "step": 15199
    },
    {
      "epoch": 2.8247537632410333,
      "grad_norm": 1.112540602684021,
      "learning_rate": 3.980994183538624e-05,
      "loss": 1.2307,
      "step": 15200
    },
    {
      "epoch": 2.824939602304404,
      "grad_norm": 0.8312342762947083,
      "learning_rate": 3.979827588139596e-05,
      "loss": 1.1071,
      "step": 15201
    },
    {
      "epoch": 2.8251254413677755,
      "grad_norm": 0.8704702854156494,
      "learning_rate": 3.9786611212308144e-05,
      "loss": 1.0509,
      "step": 15202
    },
    {
      "epoch": 2.825311280431147,
      "grad_norm": 0.7412673234939575,
      "learning_rate": 3.977494782837182e-05,
      "loss": 1.0044,
      "step": 15203
    },
    {
      "epoch": 2.8254971194945178,
      "grad_norm": 0.9747570753097534,
      "learning_rate": 3.9763285729835874e-05,
      "loss": 1.1114,
      "step": 15204
    },
    {
      "epoch": 2.8256829585578886,
      "grad_norm": 0.79355788230896,
      "learning_rate": 3.97516249169492e-05,
      "loss": 0.8749,
      "step": 15205
    },
    {
      "epoch": 2.82586879762126,
      "grad_norm": 0.9315033555030823,
      "learning_rate": 3.97399653899607e-05,
      "loss": 1.0057,
      "step": 15206
    },
    {
      "epoch": 2.8260546366846313,
      "grad_norm": 0.7607800364494324,
      "learning_rate": 3.9728307149119245e-05,
      "loss": 0.9749,
      "step": 15207
    },
    {
      "epoch": 2.826240475748002,
      "grad_norm": 0.7286087870597839,
      "learning_rate": 3.9716650194673666e-05,
      "loss": 1.038,
      "step": 15208
    },
    {
      "epoch": 2.826426314811373,
      "grad_norm": 0.7711015939712524,
      "learning_rate": 3.970499452687273e-05,
      "loss": 1.0462,
      "step": 15209
    },
    {
      "epoch": 2.8266121538747444,
      "grad_norm": 0.7511913180351257,
      "learning_rate": 3.9693340145965174e-05,
      "loss": 0.8697,
      "step": 15210
    },
    {
      "epoch": 2.8267979929381157,
      "grad_norm": 0.7339666485786438,
      "learning_rate": 3.968168705219982e-05,
      "loss": 0.7829,
      "step": 15211
    },
    {
      "epoch": 2.8269838320014866,
      "grad_norm": 0.6857532858848572,
      "learning_rate": 3.9670035245825353e-05,
      "loss": 0.7155,
      "step": 15212
    },
    {
      "epoch": 2.827169671064858,
      "grad_norm": 0.7902857065200806,
      "learning_rate": 3.9658384727090415e-05,
      "loss": 0.901,
      "step": 15213
    },
    {
      "epoch": 2.827355510128229,
      "grad_norm": 0.738915205001831,
      "learning_rate": 3.9646735496243736e-05,
      "loss": 0.86,
      "step": 15214
    },
    {
      "epoch": 2.8275413491916,
      "grad_norm": 0.7594788074493408,
      "learning_rate": 3.9635087553533924e-05,
      "loss": 0.8544,
      "step": 15215
    },
    {
      "epoch": 2.827727188254971,
      "grad_norm": 0.7334456443786621,
      "learning_rate": 3.962344089920954e-05,
      "loss": 0.9809,
      "step": 15216
    },
    {
      "epoch": 2.8279130273183424,
      "grad_norm": 0.7943280935287476,
      "learning_rate": 3.961179553351924e-05,
      "loss": 0.7557,
      "step": 15217
    },
    {
      "epoch": 2.8280988663817137,
      "grad_norm": 1.2078486680984497,
      "learning_rate": 3.9600151456711485e-05,
      "loss": 1.3622,
      "step": 15218
    },
    {
      "epoch": 2.8282847054450846,
      "grad_norm": 0.7896971702575684,
      "learning_rate": 3.9588508669034886e-05,
      "loss": 0.8955,
      "step": 15219
    },
    {
      "epoch": 2.8284705445084555,
      "grad_norm": 0.8287630081176758,
      "learning_rate": 3.957686717073791e-05,
      "loss": 0.9633,
      "step": 15220
    },
    {
      "epoch": 2.828656383571827,
      "grad_norm": 0.7887197732925415,
      "learning_rate": 3.9565226962069e-05,
      "loss": 0.9236,
      "step": 15221
    },
    {
      "epoch": 2.828842222635198,
      "grad_norm": 1.049709439277649,
      "learning_rate": 3.955358804327658e-05,
      "loss": 1.019,
      "step": 15222
    },
    {
      "epoch": 2.829028061698569,
      "grad_norm": 0.7993926405906677,
      "learning_rate": 3.954195041460913e-05,
      "loss": 0.9172,
      "step": 15223
    },
    {
      "epoch": 2.82921390076194,
      "grad_norm": 0.781864583492279,
      "learning_rate": 3.9530314076315e-05,
      "loss": 0.9311,
      "step": 15224
    },
    {
      "epoch": 2.8293997398253112,
      "grad_norm": 0.7717763185501099,
      "learning_rate": 3.9518679028642515e-05,
      "loss": 0.9092,
      "step": 15225
    },
    {
      "epoch": 2.8295855788886826,
      "grad_norm": 0.7482540011405945,
      "learning_rate": 3.950704527184003e-05,
      "loss": 0.8378,
      "step": 15226
    },
    {
      "epoch": 2.8297714179520534,
      "grad_norm": 0.791333794593811,
      "learning_rate": 3.9495412806155883e-05,
      "loss": 1.0969,
      "step": 15227
    },
    {
      "epoch": 2.8299572570154248,
      "grad_norm": 0.801101565361023,
      "learning_rate": 3.948378163183832e-05,
      "loss": 0.9298,
      "step": 15228
    },
    {
      "epoch": 2.8301430960787957,
      "grad_norm": 1.3604888916015625,
      "learning_rate": 3.94721517491356e-05,
      "loss": 1.3488,
      "step": 15229
    },
    {
      "epoch": 2.830328935142167,
      "grad_norm": 0.8929975628852844,
      "learning_rate": 3.946052315829589e-05,
      "loss": 0.9312,
      "step": 15230
    },
    {
      "epoch": 2.830514774205538,
      "grad_norm": 1.0351811647415161,
      "learning_rate": 3.944889585956746e-05,
      "loss": 0.996,
      "step": 15231
    },
    {
      "epoch": 2.830700613268909,
      "grad_norm": 0.8826019167900085,
      "learning_rate": 3.943726985319843e-05,
      "loss": 1.0848,
      "step": 15232
    },
    {
      "epoch": 2.83088645233228,
      "grad_norm": 0.8329673409461975,
      "learning_rate": 3.942564513943692e-05,
      "loss": 0.943,
      "step": 15233
    },
    {
      "epoch": 2.8310722913956514,
      "grad_norm": 0.7803268432617188,
      "learning_rate": 3.941402171853109e-05,
      "loss": 0.8656,
      "step": 15234
    },
    {
      "epoch": 2.8312581304590223,
      "grad_norm": 0.9162929654121399,
      "learning_rate": 3.9402399590729e-05,
      "loss": 1.0986,
      "step": 15235
    },
    {
      "epoch": 2.8314439695223936,
      "grad_norm": 0.6823164820671082,
      "learning_rate": 3.9390778756278716e-05,
      "loss": 0.7321,
      "step": 15236
    },
    {
      "epoch": 2.831629808585765,
      "grad_norm": 0.7181358933448792,
      "learning_rate": 3.93791592154282e-05,
      "loss": 0.7124,
      "step": 15237
    },
    {
      "epoch": 2.831815647649136,
      "grad_norm": 0.8273493647575378,
      "learning_rate": 3.9367540968425496e-05,
      "loss": 0.9114,
      "step": 15238
    },
    {
      "epoch": 2.8320014867125067,
      "grad_norm": 0.9327648282051086,
      "learning_rate": 3.935592401551862e-05,
      "loss": 1.1341,
      "step": 15239
    },
    {
      "epoch": 2.832187325775878,
      "grad_norm": 0.8144948482513428,
      "learning_rate": 3.9344308356955483e-05,
      "loss": 0.8627,
      "step": 15240
    },
    {
      "epoch": 2.8323731648392494,
      "grad_norm": 0.849429190158844,
      "learning_rate": 3.933269399298399e-05,
      "loss": 0.8588,
      "step": 15241
    },
    {
      "epoch": 2.8325590039026203,
      "grad_norm": 0.804368793964386,
      "learning_rate": 3.9321080923852e-05,
      "loss": 0.8856,
      "step": 15242
    },
    {
      "epoch": 2.8327448429659916,
      "grad_norm": 0.7444977760314941,
      "learning_rate": 3.930946914980744e-05,
      "loss": 0.8866,
      "step": 15243
    },
    {
      "epoch": 2.8329306820293625,
      "grad_norm": 0.8025500774383545,
      "learning_rate": 3.929785867109812e-05,
      "loss": 0.9125,
      "step": 15244
    },
    {
      "epoch": 2.833116521092734,
      "grad_norm": 0.8177288770675659,
      "learning_rate": 3.92862494879718e-05,
      "loss": 0.8098,
      "step": 15245
    },
    {
      "epoch": 2.8333023601561047,
      "grad_norm": 0.7735792994499207,
      "learning_rate": 3.927464160067633e-05,
      "loss": 0.7662,
      "step": 15246
    },
    {
      "epoch": 2.833488199219476,
      "grad_norm": 0.8327000141143799,
      "learning_rate": 3.9263035009459395e-05,
      "loss": 0.9096,
      "step": 15247
    },
    {
      "epoch": 2.833674038282847,
      "grad_norm": 0.8771273493766785,
      "learning_rate": 3.925142971456878e-05,
      "loss": 0.9426,
      "step": 15248
    },
    {
      "epoch": 2.8338598773462182,
      "grad_norm": 0.7832034230232239,
      "learning_rate": 3.923982571625214e-05,
      "loss": 0.8878,
      "step": 15249
    },
    {
      "epoch": 2.834045716409589,
      "grad_norm": 0.7822767496109009,
      "learning_rate": 3.9228223014757124e-05,
      "loss": 0.797,
      "step": 15250
    },
    {
      "epoch": 2.8342315554729605,
      "grad_norm": 0.8614217638969421,
      "learning_rate": 3.921662161033144e-05,
      "loss": 1.123,
      "step": 15251
    },
    {
      "epoch": 2.834417394536332,
      "grad_norm": 0.8838312029838562,
      "learning_rate": 3.920502150322265e-05,
      "loss": 0.9017,
      "step": 15252
    },
    {
      "epoch": 2.8346032335997027,
      "grad_norm": 0.7245998978614807,
      "learning_rate": 3.919342269367832e-05,
      "loss": 0.8462,
      "step": 15253
    },
    {
      "epoch": 2.8347890726630736,
      "grad_norm": 0.8327683806419373,
      "learning_rate": 3.9181825181946065e-05,
      "loss": 0.9388,
      "step": 15254
    },
    {
      "epoch": 2.834974911726445,
      "grad_norm": 1.6923487186431885,
      "learning_rate": 3.9170228968273384e-05,
      "loss": 1.4495,
      "step": 15255
    },
    {
      "epoch": 2.835160750789816,
      "grad_norm": 1.0795648097991943,
      "learning_rate": 3.915863405290777e-05,
      "loss": 0.9367,
      "step": 15256
    },
    {
      "epoch": 2.835346589853187,
      "grad_norm": 1.0243604183197021,
      "learning_rate": 3.914704043609668e-05,
      "loss": 0.9431,
      "step": 15257
    },
    {
      "epoch": 2.835532428916558,
      "grad_norm": 0.8284094929695129,
      "learning_rate": 3.913544811808758e-05,
      "loss": 1.0057,
      "step": 15258
    },
    {
      "epoch": 2.8357182679799293,
      "grad_norm": 0.8010451793670654,
      "learning_rate": 3.9123857099127936e-05,
      "loss": 0.8132,
      "step": 15259
    },
    {
      "epoch": 2.8359041070433006,
      "grad_norm": 0.93331378698349,
      "learning_rate": 3.9112267379465087e-05,
      "loss": 1.0041,
      "step": 15260
    },
    {
      "epoch": 2.8360899461066715,
      "grad_norm": 0.8245121836662292,
      "learning_rate": 3.91006789593464e-05,
      "loss": 0.7767,
      "step": 15261
    },
    {
      "epoch": 2.836275785170043,
      "grad_norm": 0.9565784335136414,
      "learning_rate": 3.9089091839019176e-05,
      "loss": 0.8926,
      "step": 15262
    },
    {
      "epoch": 2.8364616242334137,
      "grad_norm": 0.801906406879425,
      "learning_rate": 3.90775060187308e-05,
      "loss": 0.8542,
      "step": 15263
    },
    {
      "epoch": 2.836647463296785,
      "grad_norm": 0.814352810382843,
      "learning_rate": 3.906592149872852e-05,
      "loss": 0.8702,
      "step": 15264
    },
    {
      "epoch": 2.836833302360156,
      "grad_norm": 0.7679254412651062,
      "learning_rate": 3.9054338279259526e-05,
      "loss": 0.6711,
      "step": 15265
    },
    {
      "epoch": 2.8370191414235273,
      "grad_norm": 1.0212868452072144,
      "learning_rate": 3.904275636057114e-05,
      "loss": 1.1643,
      "step": 15266
    },
    {
      "epoch": 2.8372049804868986,
      "grad_norm": 1.1639509201049805,
      "learning_rate": 3.9031175742910506e-05,
      "loss": 1.2237,
      "step": 15267
    },
    {
      "epoch": 2.8373908195502695,
      "grad_norm": 0.8592068552970886,
      "learning_rate": 3.901959642652477e-05,
      "loss": 0.9922,
      "step": 15268
    },
    {
      "epoch": 2.8375766586136404,
      "grad_norm": 0.7639787793159485,
      "learning_rate": 3.900801841166114e-05,
      "loss": 0.8276,
      "step": 15269
    },
    {
      "epoch": 2.8377624976770117,
      "grad_norm": 0.8858164548873901,
      "learning_rate": 3.8996441698566656e-05,
      "loss": 1.1325,
      "step": 15270
    },
    {
      "epoch": 2.837948336740383,
      "grad_norm": 0.9063674807548523,
      "learning_rate": 3.8984866287488465e-05,
      "loss": 0.972,
      "step": 15271
    },
    {
      "epoch": 2.838134175803754,
      "grad_norm": 0.8093922138214111,
      "learning_rate": 3.897329217867359e-05,
      "loss": 0.7104,
      "step": 15272
    },
    {
      "epoch": 2.838320014867125,
      "grad_norm": 0.8616423010826111,
      "learning_rate": 3.896171937236904e-05,
      "loss": 1.0722,
      "step": 15273
    },
    {
      "epoch": 2.838505853930496,
      "grad_norm": 1.0238964557647705,
      "learning_rate": 3.895014786882189e-05,
      "loss": 1.0427,
      "step": 15274
    },
    {
      "epoch": 2.8386916929938675,
      "grad_norm": 0.893201470375061,
      "learning_rate": 3.8938577668279055e-05,
      "loss": 1.1476,
      "step": 15275
    },
    {
      "epoch": 2.8388775320572384,
      "grad_norm": 0.9454589486122131,
      "learning_rate": 3.892700877098749e-05,
      "loss": 1.1824,
      "step": 15276
    },
    {
      "epoch": 2.8390633711206097,
      "grad_norm": 1.279524326324463,
      "learning_rate": 3.89154411771941e-05,
      "loss": 1.1519,
      "step": 15277
    },
    {
      "epoch": 2.8392492101839806,
      "grad_norm": 0.8839907050132751,
      "learning_rate": 3.890387488714582e-05,
      "loss": 0.9033,
      "step": 15278
    },
    {
      "epoch": 2.839435049247352,
      "grad_norm": 0.9178303480148315,
      "learning_rate": 3.889230990108944e-05,
      "loss": 1.024,
      "step": 15279
    },
    {
      "epoch": 2.839620888310723,
      "grad_norm": 0.6783154606819153,
      "learning_rate": 3.888074621927189e-05,
      "loss": 0.7864,
      "step": 15280
    },
    {
      "epoch": 2.839806727374094,
      "grad_norm": 0.8283354640007019,
      "learning_rate": 3.886918384193992e-05,
      "loss": 0.9406,
      "step": 15281
    },
    {
      "epoch": 2.839992566437465,
      "grad_norm": 0.8283498287200928,
      "learning_rate": 3.8857622769340286e-05,
      "loss": 0.8384,
      "step": 15282
    },
    {
      "epoch": 2.8401784055008363,
      "grad_norm": 0.8397684693336487,
      "learning_rate": 3.884606300171979e-05,
      "loss": 0.9565,
      "step": 15283
    },
    {
      "epoch": 2.840364244564207,
      "grad_norm": 1.0575993061065674,
      "learning_rate": 3.883450453932514e-05,
      "loss": 1.0169,
      "step": 15284
    },
    {
      "epoch": 2.8405500836275785,
      "grad_norm": 0.7142122387886047,
      "learning_rate": 3.8822947382403005e-05,
      "loss": 0.953,
      "step": 15285
    },
    {
      "epoch": 2.84073592269095,
      "grad_norm": 0.7820481061935425,
      "learning_rate": 3.8811391531200105e-05,
      "loss": 0.9879,
      "step": 15286
    },
    {
      "epoch": 2.8409217617543208,
      "grad_norm": 0.7898126840591431,
      "learning_rate": 3.8799836985963047e-05,
      "loss": 0.9148,
      "step": 15287
    },
    {
      "epoch": 2.8411076008176916,
      "grad_norm": 0.854949951171875,
      "learning_rate": 3.8788283746938437e-05,
      "loss": 0.9617,
      "step": 15288
    },
    {
      "epoch": 2.841293439881063,
      "grad_norm": 0.854562520980835,
      "learning_rate": 3.8776731814372836e-05,
      "loss": 0.8635,
      "step": 15289
    },
    {
      "epoch": 2.8414792789444343,
      "grad_norm": 0.9364863038063049,
      "learning_rate": 3.8765181188512835e-05,
      "loss": 0.858,
      "step": 15290
    },
    {
      "epoch": 2.841665118007805,
      "grad_norm": 0.7348250150680542,
      "learning_rate": 3.875363186960499e-05,
      "loss": 0.5607,
      "step": 15291
    },
    {
      "epoch": 2.8418509570711765,
      "grad_norm": 0.817834198474884,
      "learning_rate": 3.8742083857895775e-05,
      "loss": 0.877,
      "step": 15292
    },
    {
      "epoch": 2.8420367961345474,
      "grad_norm": 0.9872136116027832,
      "learning_rate": 3.873053715363161e-05,
      "loss": 1.1492,
      "step": 15293
    },
    {
      "epoch": 2.8422226351979187,
      "grad_norm": 0.9656583070755005,
      "learning_rate": 3.871899175705903e-05,
      "loss": 1.1006,
      "step": 15294
    },
    {
      "epoch": 2.8424084742612896,
      "grad_norm": 0.7747172713279724,
      "learning_rate": 3.870744766842441e-05,
      "loss": 1.0136,
      "step": 15295
    },
    {
      "epoch": 2.842594313324661,
      "grad_norm": 0.7914901375770569,
      "learning_rate": 3.869590488797413e-05,
      "loss": 0.7739,
      "step": 15296
    },
    {
      "epoch": 2.842780152388032,
      "grad_norm": 0.8331795930862427,
      "learning_rate": 3.868436341595453e-05,
      "loss": 0.7103,
      "step": 15297
    },
    {
      "epoch": 2.842965991451403,
      "grad_norm": 0.7876693606376648,
      "learning_rate": 3.8672823252612015e-05,
      "loss": 0.8948,
      "step": 15298
    },
    {
      "epoch": 2.843151830514774,
      "grad_norm": 0.967292308807373,
      "learning_rate": 3.866128439819283e-05,
      "loss": 1.1008,
      "step": 15299
    },
    {
      "epoch": 2.8433376695781454,
      "grad_norm": 0.772185742855072,
      "learning_rate": 3.8649746852943234e-05,
      "loss": 0.9245,
      "step": 15300
    },
    {
      "epoch": 2.8435235086415167,
      "grad_norm": 0.9624976515769958,
      "learning_rate": 3.8638210617109546e-05,
      "loss": 0.8139,
      "step": 15301
    },
    {
      "epoch": 2.8437093477048876,
      "grad_norm": 0.8755714297294617,
      "learning_rate": 3.8626675690937934e-05,
      "loss": 0.7875,
      "step": 15302
    },
    {
      "epoch": 2.8438951867682585,
      "grad_norm": 3.687150001525879,
      "learning_rate": 3.8615142074674625e-05,
      "loss": 1.5541,
      "step": 15303
    },
    {
      "epoch": 2.84408102583163,
      "grad_norm": 0.9187508821487427,
      "learning_rate": 3.8603609768565776e-05,
      "loss": 1.1961,
      "step": 15304
    },
    {
      "epoch": 2.844266864895001,
      "grad_norm": 0.8098819255828857,
      "learning_rate": 3.8592078772857465e-05,
      "loss": 0.8885,
      "step": 15305
    },
    {
      "epoch": 2.844452703958372,
      "grad_norm": 0.9316312074661255,
      "learning_rate": 3.85805490877959e-05,
      "loss": 1.0535,
      "step": 15306
    },
    {
      "epoch": 2.844638543021743,
      "grad_norm": 0.8622249960899353,
      "learning_rate": 3.8569020713627126e-05,
      "loss": 1.0789,
      "step": 15307
    },
    {
      "epoch": 2.8448243820851142,
      "grad_norm": 0.9714615345001221,
      "learning_rate": 3.855749365059716e-05,
      "loss": 0.8778,
      "step": 15308
    },
    {
      "epoch": 2.8450102211484856,
      "grad_norm": 1.0000050067901611,
      "learning_rate": 3.854596789895204e-05,
      "loss": 1.0138,
      "step": 15309
    },
    {
      "epoch": 2.8451960602118564,
      "grad_norm": 0.8454744815826416,
      "learning_rate": 3.8534443458937764e-05,
      "loss": 0.7292,
      "step": 15310
    },
    {
      "epoch": 2.8453818992752278,
      "grad_norm": 0.7739028334617615,
      "learning_rate": 3.852292033080035e-05,
      "loss": 0.4697,
      "step": 15311
    },
    {
      "epoch": 2.8455677383385987,
      "grad_norm": 1.3835629224777222,
      "learning_rate": 3.85113985147857e-05,
      "loss": 1.4786,
      "step": 15312
    },
    {
      "epoch": 2.84575357740197,
      "grad_norm": 0.7709473967552185,
      "learning_rate": 3.84998780111397e-05,
      "loss": 0.842,
      "step": 15313
    },
    {
      "epoch": 2.845939416465341,
      "grad_norm": 0.718707263469696,
      "learning_rate": 3.8488358820108305e-05,
      "loss": 0.772,
      "step": 15314
    },
    {
      "epoch": 2.846125255528712,
      "grad_norm": 0.9819291234016418,
      "learning_rate": 3.847684094193733e-05,
      "loss": 1.1089,
      "step": 15315
    },
    {
      "epoch": 2.846311094592083,
      "grad_norm": 0.7201307415962219,
      "learning_rate": 3.8465324376872604e-05,
      "loss": 0.8921,
      "step": 15316
    },
    {
      "epoch": 2.8464969336554544,
      "grad_norm": 0.8110470771789551,
      "learning_rate": 3.845380912515989e-05,
      "loss": 1.0143,
      "step": 15317
    },
    {
      "epoch": 2.8466827727188253,
      "grad_norm": 0.9493751525878906,
      "learning_rate": 3.844229518704505e-05,
      "loss": 0.9172,
      "step": 15318
    },
    {
      "epoch": 2.8468686117821966,
      "grad_norm": 1.0635316371917725,
      "learning_rate": 3.843078256277377e-05,
      "loss": 0.8767,
      "step": 15319
    },
    {
      "epoch": 2.847054450845568,
      "grad_norm": 0.9986346960067749,
      "learning_rate": 3.841927125259179e-05,
      "loss": 0.908,
      "step": 15320
    },
    {
      "epoch": 2.847240289908939,
      "grad_norm": 0.7902888059616089,
      "learning_rate": 3.840776125674475e-05,
      "loss": 0.7851,
      "step": 15321
    },
    {
      "epoch": 2.8474261289723097,
      "grad_norm": 0.8674366474151611,
      "learning_rate": 3.8396252575478344e-05,
      "loss": 0.9204,
      "step": 15322
    },
    {
      "epoch": 2.847611968035681,
      "grad_norm": 0.8738417029380798,
      "learning_rate": 3.838474520903825e-05,
      "loss": 0.9319,
      "step": 15323
    },
    {
      "epoch": 2.8477978070990524,
      "grad_norm": 0.7828598022460938,
      "learning_rate": 3.837323915767004e-05,
      "loss": 0.8096,
      "step": 15324
    },
    {
      "epoch": 2.8479836461624233,
      "grad_norm": 0.8735978603363037,
      "learning_rate": 3.836173442161923e-05,
      "loss": 1.2256,
      "step": 15325
    },
    {
      "epoch": 2.8481694852257946,
      "grad_norm": 0.7538908123970032,
      "learning_rate": 3.8350231001131476e-05,
      "loss": 0.8915,
      "step": 15326
    },
    {
      "epoch": 2.8483553242891655,
      "grad_norm": 0.7566878795623779,
      "learning_rate": 3.833872889645225e-05,
      "loss": 0.9013,
      "step": 15327
    },
    {
      "epoch": 2.848541163352537,
      "grad_norm": 0.7990183234214783,
      "learning_rate": 3.832722810782703e-05,
      "loss": 1.1669,
      "step": 15328
    },
    {
      "epoch": 2.8487270024159077,
      "grad_norm": 0.8590530157089233,
      "learning_rate": 3.831572863550126e-05,
      "loss": 0.9329,
      "step": 15329
    },
    {
      "epoch": 2.848912841479279,
      "grad_norm": 0.819938600063324,
      "learning_rate": 3.830423047972044e-05,
      "loss": 1.0213,
      "step": 15330
    },
    {
      "epoch": 2.84909868054265,
      "grad_norm": 0.8118523955345154,
      "learning_rate": 3.8292733640729905e-05,
      "loss": 0.9333,
      "step": 15331
    },
    {
      "epoch": 2.8492845196060212,
      "grad_norm": 0.7628287076950073,
      "learning_rate": 3.828123811877512e-05,
      "loss": 0.6968,
      "step": 15332
    },
    {
      "epoch": 2.849470358669392,
      "grad_norm": 0.956183910369873,
      "learning_rate": 3.826974391410135e-05,
      "loss": 1.0995,
      "step": 15333
    },
    {
      "epoch": 2.8496561977327635,
      "grad_norm": 0.7814002633094788,
      "learning_rate": 3.825825102695402e-05,
      "loss": 0.957,
      "step": 15334
    },
    {
      "epoch": 2.849842036796135,
      "grad_norm": 0.9299039244651794,
      "learning_rate": 3.8246759457578354e-05,
      "loss": 0.9723,
      "step": 15335
    },
    {
      "epoch": 2.8500278758595057,
      "grad_norm": 0.7808316946029663,
      "learning_rate": 3.823526920621963e-05,
      "loss": 0.7032,
      "step": 15336
    },
    {
      "epoch": 2.8502137149228766,
      "grad_norm": 0.9004837274551392,
      "learning_rate": 3.822378027312305e-05,
      "loss": 0.8531,
      "step": 15337
    },
    {
      "epoch": 2.850399553986248,
      "grad_norm": 1.0717692375183105,
      "learning_rate": 3.821229265853392e-05,
      "loss": 0.9096,
      "step": 15338
    },
    {
      "epoch": 2.850585393049619,
      "grad_norm": 0.9275893568992615,
      "learning_rate": 3.820080636269737e-05,
      "loss": 0.965,
      "step": 15339
    },
    {
      "epoch": 2.85077123211299,
      "grad_norm": 0.7810984253883362,
      "learning_rate": 3.8189321385858556e-05,
      "loss": 0.8819,
      "step": 15340
    },
    {
      "epoch": 2.8509570711763614,
      "grad_norm": 0.8521530032157898,
      "learning_rate": 3.8177837728262564e-05,
      "loss": 1.0317,
      "step": 15341
    },
    {
      "epoch": 2.8511429102397323,
      "grad_norm": 0.9492397904396057,
      "learning_rate": 3.816635539015454e-05,
      "loss": 0.8219,
      "step": 15342
    },
    {
      "epoch": 2.8513287493031036,
      "grad_norm": 0.751935601234436,
      "learning_rate": 3.815487437177958e-05,
      "loss": 0.7021,
      "step": 15343
    },
    {
      "epoch": 2.8515145883664745,
      "grad_norm": 0.7729502320289612,
      "learning_rate": 3.814339467338269e-05,
      "loss": 0.8726,
      "step": 15344
    },
    {
      "epoch": 2.851700427429846,
      "grad_norm": 0.7897372841835022,
      "learning_rate": 3.813191629520887e-05,
      "loss": 1.0875,
      "step": 15345
    },
    {
      "epoch": 2.8518862664932167,
      "grad_norm": 1.0642775297164917,
      "learning_rate": 3.812043923750316e-05,
      "loss": 1.1055,
      "step": 15346
    },
    {
      "epoch": 2.852072105556588,
      "grad_norm": 0.7688366770744324,
      "learning_rate": 3.810896350051048e-05,
      "loss": 0.8611,
      "step": 15347
    },
    {
      "epoch": 2.852257944619959,
      "grad_norm": 0.8489544987678528,
      "learning_rate": 3.809748908447576e-05,
      "loss": 0.9513,
      "step": 15348
    },
    {
      "epoch": 2.8524437836833303,
      "grad_norm": 0.7277908325195312,
      "learning_rate": 3.8086015989643875e-05,
      "loss": 0.9195,
      "step": 15349
    },
    {
      "epoch": 2.8526296227467016,
      "grad_norm": 0.8600225448608398,
      "learning_rate": 3.8074544216259766e-05,
      "loss": 0.9291,
      "step": 15350
    },
    {
      "epoch": 2.8528154618100725,
      "grad_norm": 0.8020973205566406,
      "learning_rate": 3.806307376456825e-05,
      "loss": 0.7507,
      "step": 15351
    },
    {
      "epoch": 2.8530013008734434,
      "grad_norm": 1.6078366041183472,
      "learning_rate": 3.8051604634814085e-05,
      "loss": 1.2141,
      "step": 15352
    },
    {
      "epoch": 2.8531871399368147,
      "grad_norm": 0.8335413336753845,
      "learning_rate": 3.8040136827242124e-05,
      "loss": 0.9712,
      "step": 15353
    },
    {
      "epoch": 2.853372979000186,
      "grad_norm": 0.8307401537895203,
      "learning_rate": 3.802867034209715e-05,
      "loss": 0.8872,
      "step": 15354
    },
    {
      "epoch": 2.853558818063557,
      "grad_norm": 0.8642200827598572,
      "learning_rate": 3.801720517962386e-05,
      "loss": 0.9417,
      "step": 15355
    },
    {
      "epoch": 2.853744657126928,
      "grad_norm": 0.8002011179924011,
      "learning_rate": 3.800574134006695e-05,
      "loss": 0.8183,
      "step": 15356
    },
    {
      "epoch": 2.853930496190299,
      "grad_norm": 0.8602607250213623,
      "learning_rate": 3.7994278823671084e-05,
      "loss": 1.0025,
      "step": 15357
    },
    {
      "epoch": 2.8541163352536705,
      "grad_norm": 0.9555697441101074,
      "learning_rate": 3.798281763068096e-05,
      "loss": 0.7349,
      "step": 15358
    },
    {
      "epoch": 2.8543021743170414,
      "grad_norm": 0.7865543365478516,
      "learning_rate": 3.7971357761341154e-05,
      "loss": 1.0243,
      "step": 15359
    },
    {
      "epoch": 2.8544880133804127,
      "grad_norm": 0.7438862919807434,
      "learning_rate": 3.795989921589629e-05,
      "loss": 0.8489,
      "step": 15360
    },
    {
      "epoch": 2.8546738524437836,
      "grad_norm": 0.9199408888816833,
      "learning_rate": 3.794844199459087e-05,
      "loss": 0.9497,
      "step": 15361
    },
    {
      "epoch": 2.854859691507155,
      "grad_norm": 0.8037041425704956,
      "learning_rate": 3.79369860976695e-05,
      "loss": 0.7723,
      "step": 15362
    },
    {
      "epoch": 2.855045530570526,
      "grad_norm": 0.9081737995147705,
      "learning_rate": 3.7925531525376623e-05,
      "loss": 0.9376,
      "step": 15363
    },
    {
      "epoch": 2.855231369633897,
      "grad_norm": 0.8575348258018494,
      "learning_rate": 3.7914078277956786e-05,
      "loss": 1.0563,
      "step": 15364
    },
    {
      "epoch": 2.855417208697268,
      "grad_norm": 0.8420919179916382,
      "learning_rate": 3.790262635565438e-05,
      "loss": 0.9279,
      "step": 15365
    },
    {
      "epoch": 2.8556030477606393,
      "grad_norm": 0.8386310935020447,
      "learning_rate": 3.7891175758713885e-05,
      "loss": 0.7817,
      "step": 15366
    },
    {
      "epoch": 2.85578888682401,
      "grad_norm": 1.0895872116088867,
      "learning_rate": 3.7879726487379655e-05,
      "loss": 1.1757,
      "step": 15367
    },
    {
      "epoch": 2.8559747258873815,
      "grad_norm": 0.6934189796447754,
      "learning_rate": 3.786827854189606e-05,
      "loss": 0.6941,
      "step": 15368
    },
    {
      "epoch": 2.856160564950753,
      "grad_norm": 0.744887113571167,
      "learning_rate": 3.785683192250741e-05,
      "loss": 0.8822,
      "step": 15369
    },
    {
      "epoch": 2.8563464040141238,
      "grad_norm": 0.8233052492141724,
      "learning_rate": 3.784538662945807e-05,
      "loss": 1.0015,
      "step": 15370
    },
    {
      "epoch": 2.8565322430774946,
      "grad_norm": 0.9672362208366394,
      "learning_rate": 3.783394266299228e-05,
      "loss": 0.9964,
      "step": 15371
    },
    {
      "epoch": 2.856718082140866,
      "grad_norm": 0.8399665951728821,
      "learning_rate": 3.782250002335431e-05,
      "loss": 0.8387,
      "step": 15372
    },
    {
      "epoch": 2.8569039212042373,
      "grad_norm": 0.9578357934951782,
      "learning_rate": 3.7811058710788314e-05,
      "loss": 1.1728,
      "step": 15373
    },
    {
      "epoch": 2.857089760267608,
      "grad_norm": 0.8220307230949402,
      "learning_rate": 3.779961872553862e-05,
      "loss": 0.9484,
      "step": 15374
    },
    {
      "epoch": 2.8572755993309795,
      "grad_norm": 0.84305739402771,
      "learning_rate": 3.778818006784931e-05,
      "loss": 1.0166,
      "step": 15375
    },
    {
      "epoch": 2.8574614383943504,
      "grad_norm": 0.8652834296226501,
      "learning_rate": 3.7776742737964555e-05,
      "loss": 1.0987,
      "step": 15376
    },
    {
      "epoch": 2.8576472774577217,
      "grad_norm": 0.8326297402381897,
      "learning_rate": 3.7765306736128395e-05,
      "loss": 0.9049,
      "step": 15377
    },
    {
      "epoch": 2.8578331165210926,
      "grad_norm": 0.73166823387146,
      "learning_rate": 3.7753872062585e-05,
      "loss": 0.9023,
      "step": 15378
    },
    {
      "epoch": 2.858018955584464,
      "grad_norm": 0.9456663131713867,
      "learning_rate": 3.77424387175784e-05,
      "loss": 0.9886,
      "step": 15379
    },
    {
      "epoch": 2.858204794647835,
      "grad_norm": 0.9489964246749878,
      "learning_rate": 3.77310067013526e-05,
      "loss": 0.7923,
      "step": 15380
    },
    {
      "epoch": 2.858390633711206,
      "grad_norm": 0.8790481686592102,
      "learning_rate": 3.771957601415157e-05,
      "loss": 0.9042,
      "step": 15381
    },
    {
      "epoch": 2.858576472774577,
      "grad_norm": 0.7280595302581787,
      "learning_rate": 3.7708146656219346e-05,
      "loss": 1.0003,
      "step": 15382
    },
    {
      "epoch": 2.8587623118379484,
      "grad_norm": 0.8281516432762146,
      "learning_rate": 3.769671862779983e-05,
      "loss": 0.9558,
      "step": 15383
    },
    {
      "epoch": 2.8589481509013197,
      "grad_norm": 0.8291042447090149,
      "learning_rate": 3.768529192913691e-05,
      "loss": 0.943,
      "step": 15384
    },
    {
      "epoch": 2.8591339899646906,
      "grad_norm": 0.8003760576248169,
      "learning_rate": 3.7673866560474504e-05,
      "loss": 0.8982,
      "step": 15385
    },
    {
      "epoch": 2.8593198290280615,
      "grad_norm": 0.716830313205719,
      "learning_rate": 3.766244252205649e-05,
      "loss": 0.8642,
      "step": 15386
    },
    {
      "epoch": 2.859505668091433,
      "grad_norm": 0.9099149703979492,
      "learning_rate": 3.7651019814126654e-05,
      "loss": 0.8758,
      "step": 15387
    },
    {
      "epoch": 2.859691507154804,
      "grad_norm": 0.8813703060150146,
      "learning_rate": 3.763959843692881e-05,
      "loss": 0.9723,
      "step": 15388
    },
    {
      "epoch": 2.859877346218175,
      "grad_norm": 0.6747955679893494,
      "learning_rate": 3.762817839070669e-05,
      "loss": 0.6139,
      "step": 15389
    },
    {
      "epoch": 2.860063185281546,
      "grad_norm": 0.8541521430015564,
      "learning_rate": 3.7616759675704105e-05,
      "loss": 1.0557,
      "step": 15390
    },
    {
      "epoch": 2.8602490243449172,
      "grad_norm": 0.7706393599510193,
      "learning_rate": 3.7605342292164735e-05,
      "loss": 0.7235,
      "step": 15391
    },
    {
      "epoch": 2.8604348634082886,
      "grad_norm": 0.8833755254745483,
      "learning_rate": 3.7593926240332253e-05,
      "loss": 1.1436,
      "step": 15392
    },
    {
      "epoch": 2.8606207024716594,
      "grad_norm": 0.8103173971176147,
      "learning_rate": 3.758251152045029e-05,
      "loss": 0.7348,
      "step": 15393
    },
    {
      "epoch": 2.860806541535031,
      "grad_norm": 0.7386962175369263,
      "learning_rate": 3.7571098132762506e-05,
      "loss": 0.7346,
      "step": 15394
    },
    {
      "epoch": 2.8609923805984017,
      "grad_norm": 0.9515390396118164,
      "learning_rate": 3.7559686077512534e-05,
      "loss": 0.9968,
      "step": 15395
    },
    {
      "epoch": 2.861178219661773,
      "grad_norm": 0.793876588344574,
      "learning_rate": 3.7548275354943904e-05,
      "loss": 0.6786,
      "step": 15396
    },
    {
      "epoch": 2.861364058725144,
      "grad_norm": 0.8353383541107178,
      "learning_rate": 3.753686596530014e-05,
      "loss": 1.0705,
      "step": 15397
    },
    {
      "epoch": 2.861549897788515,
      "grad_norm": 0.7873873114585876,
      "learning_rate": 3.7525457908824815e-05,
      "loss": 0.7793,
      "step": 15398
    },
    {
      "epoch": 2.8617357368518865,
      "grad_norm": 0.8067834973335266,
      "learning_rate": 3.751405118576138e-05,
      "loss": 1.2097,
      "step": 15399
    },
    {
      "epoch": 2.8619215759152574,
      "grad_norm": 0.7366782426834106,
      "learning_rate": 3.7502645796353284e-05,
      "loss": 0.974,
      "step": 15400
    },
    {
      "epoch": 2.8621074149786283,
      "grad_norm": 0.7902079820632935,
      "learning_rate": 3.749124174084394e-05,
      "loss": 1.0888,
      "step": 15401
    },
    {
      "epoch": 2.8622932540419996,
      "grad_norm": 0.8420895338058472,
      "learning_rate": 3.7479839019476804e-05,
      "loss": 0.8503,
      "step": 15402
    },
    {
      "epoch": 2.862479093105371,
      "grad_norm": 1.4798964262008667,
      "learning_rate": 3.746843763249521e-05,
      "loss": 1.1554,
      "step": 15403
    },
    {
      "epoch": 2.862664932168742,
      "grad_norm": 0.8021959066390991,
      "learning_rate": 3.745703758014247e-05,
      "loss": 0.8877,
      "step": 15404
    },
    {
      "epoch": 2.8628507712321127,
      "grad_norm": 0.9752201437950134,
      "learning_rate": 3.744563886266195e-05,
      "loss": 0.833,
      "step": 15405
    },
    {
      "epoch": 2.863036610295484,
      "grad_norm": 0.8889212608337402,
      "learning_rate": 3.743424148029694e-05,
      "loss": 0.8776,
      "step": 15406
    },
    {
      "epoch": 2.8632224493588554,
      "grad_norm": 0.8550772070884705,
      "learning_rate": 3.742284543329068e-05,
      "loss": 0.8245,
      "step": 15407
    },
    {
      "epoch": 2.8634082884222263,
      "grad_norm": 0.8788866996765137,
      "learning_rate": 3.74114507218864e-05,
      "loss": 1.0853,
      "step": 15408
    },
    {
      "epoch": 2.8635941274855976,
      "grad_norm": 0.8350381255149841,
      "learning_rate": 3.7400057346327266e-05,
      "loss": 1.0919,
      "step": 15409
    },
    {
      "epoch": 2.8637799665489685,
      "grad_norm": 0.8196921944618225,
      "learning_rate": 3.738866530685652e-05,
      "loss": 0.9017,
      "step": 15410
    },
    {
      "epoch": 2.86396580561234,
      "grad_norm": 0.7616583108901978,
      "learning_rate": 3.737727460371726e-05,
      "loss": 0.9222,
      "step": 15411
    },
    {
      "epoch": 2.8641516446757107,
      "grad_norm": 0.8864631652832031,
      "learning_rate": 3.736588523715261e-05,
      "loss": 0.9886,
      "step": 15412
    },
    {
      "epoch": 2.864337483739082,
      "grad_norm": 0.7534281015396118,
      "learning_rate": 3.7354497207405614e-05,
      "loss": 0.965,
      "step": 15413
    },
    {
      "epoch": 2.864523322802453,
      "grad_norm": 0.7649170756340027,
      "learning_rate": 3.734311051471942e-05,
      "loss": 0.9695,
      "step": 15414
    },
    {
      "epoch": 2.8647091618658242,
      "grad_norm": 0.848669171333313,
      "learning_rate": 3.733172515933697e-05,
      "loss": 1.0583,
      "step": 15415
    },
    {
      "epoch": 2.864895000929195,
      "grad_norm": 0.858026921749115,
      "learning_rate": 3.7320341141501334e-05,
      "loss": 0.6962,
      "step": 15416
    },
    {
      "epoch": 2.8650808399925665,
      "grad_norm": 0.8865530490875244,
      "learning_rate": 3.730895846145542e-05,
      "loss": 0.8461,
      "step": 15417
    },
    {
      "epoch": 2.865266679055938,
      "grad_norm": 0.7353289723396301,
      "learning_rate": 3.729757711944224e-05,
      "loss": 0.6562,
      "step": 15418
    },
    {
      "epoch": 2.8654525181193087,
      "grad_norm": 0.9045048952102661,
      "learning_rate": 3.728619711570468e-05,
      "loss": 1.0692,
      "step": 15419
    },
    {
      "epoch": 2.8656383571826796,
      "grad_norm": 0.9202983975410461,
      "learning_rate": 3.727481845048562e-05,
      "loss": 0.872,
      "step": 15420
    },
    {
      "epoch": 2.865824196246051,
      "grad_norm": 0.7834516167640686,
      "learning_rate": 3.7263441124027875e-05,
      "loss": 0.9966,
      "step": 15421
    },
    {
      "epoch": 2.866010035309422,
      "grad_norm": 0.8796038031578064,
      "learning_rate": 3.725206513657435e-05,
      "loss": 0.9245,
      "step": 15422
    },
    {
      "epoch": 2.866195874372793,
      "grad_norm": 1.0300281047821045,
      "learning_rate": 3.7240690488367833e-05,
      "loss": 0.9943,
      "step": 15423
    },
    {
      "epoch": 2.8663817134361644,
      "grad_norm": 0.7049465179443359,
      "learning_rate": 3.722931717965103e-05,
      "loss": 0.7929,
      "step": 15424
    },
    {
      "epoch": 2.8665675524995353,
      "grad_norm": 0.9132517576217651,
      "learning_rate": 3.721794521066678e-05,
      "loss": 1.234,
      "step": 15425
    },
    {
      "epoch": 2.8667533915629066,
      "grad_norm": 0.8524289727210999,
      "learning_rate": 3.720657458165771e-05,
      "loss": 1.0903,
      "step": 15426
    },
    {
      "epoch": 2.8669392306262775,
      "grad_norm": 0.8202354907989502,
      "learning_rate": 3.719520529286659e-05,
      "loss": 0.9081,
      "step": 15427
    },
    {
      "epoch": 2.867125069689649,
      "grad_norm": 0.9034220576286316,
      "learning_rate": 3.718383734453602e-05,
      "loss": 1.0881,
      "step": 15428
    },
    {
      "epoch": 2.8673109087530197,
      "grad_norm": 0.7171639204025269,
      "learning_rate": 3.7172470736908616e-05,
      "loss": 1.0301,
      "step": 15429
    },
    {
      "epoch": 2.867496747816391,
      "grad_norm": 0.9133208394050598,
      "learning_rate": 3.716110547022704e-05,
      "loss": 0.9448,
      "step": 15430
    },
    {
      "epoch": 2.867682586879762,
      "grad_norm": 0.9423163533210754,
      "learning_rate": 3.714974154473384e-05,
      "loss": 1.2942,
      "step": 15431
    },
    {
      "epoch": 2.8678684259431333,
      "grad_norm": 0.8688696622848511,
      "learning_rate": 3.713837896067155e-05,
      "loss": 0.8479,
      "step": 15432
    },
    {
      "epoch": 2.8680542650065046,
      "grad_norm": 0.8275406360626221,
      "learning_rate": 3.712701771828265e-05,
      "loss": 0.8357,
      "step": 15433
    },
    {
      "epoch": 2.8682401040698755,
      "grad_norm": 0.9139199256896973,
      "learning_rate": 3.7115657817809704e-05,
      "loss": 1.305,
      "step": 15434
    },
    {
      "epoch": 2.8684259431332464,
      "grad_norm": 0.9804993271827698,
      "learning_rate": 3.710429925949511e-05,
      "loss": 0.6615,
      "step": 15435
    },
    {
      "epoch": 2.8686117821966177,
      "grad_norm": 0.821039080619812,
      "learning_rate": 3.709294204358129e-05,
      "loss": 0.9444,
      "step": 15436
    },
    {
      "epoch": 2.868797621259989,
      "grad_norm": 0.8548663854598999,
      "learning_rate": 3.708158617031068e-05,
      "loss": 0.9661,
      "step": 15437
    },
    {
      "epoch": 2.86898346032336,
      "grad_norm": 0.7277716994285583,
      "learning_rate": 3.7070231639925654e-05,
      "loss": 0.8455,
      "step": 15438
    },
    {
      "epoch": 2.869169299386731,
      "grad_norm": 0.8620544672012329,
      "learning_rate": 3.705887845266857e-05,
      "loss": 0.9819,
      "step": 15439
    },
    {
      "epoch": 2.869355138450102,
      "grad_norm": 0.7607631087303162,
      "learning_rate": 3.7047526608781693e-05,
      "loss": 0.952,
      "step": 15440
    },
    {
      "epoch": 2.8695409775134735,
      "grad_norm": 0.8347303867340088,
      "learning_rate": 3.703617610850729e-05,
      "loss": 0.8776,
      "step": 15441
    },
    {
      "epoch": 2.8697268165768444,
      "grad_norm": 0.9303619265556335,
      "learning_rate": 3.702482695208771e-05,
      "loss": 0.9928,
      "step": 15442
    },
    {
      "epoch": 2.8699126556402157,
      "grad_norm": 0.9514495730400085,
      "learning_rate": 3.7013479139765115e-05,
      "loss": 1.0331,
      "step": 15443
    },
    {
      "epoch": 2.8700984947035866,
      "grad_norm": 0.6765170693397522,
      "learning_rate": 3.70021326717817e-05,
      "loss": 0.7428,
      "step": 15444
    },
    {
      "epoch": 2.870284333766958,
      "grad_norm": 0.6782026886940002,
      "learning_rate": 3.699078754837967e-05,
      "loss": 0.6353,
      "step": 15445
    },
    {
      "epoch": 2.870470172830329,
      "grad_norm": 0.769478976726532,
      "learning_rate": 3.697944376980116e-05,
      "loss": 1.0179,
      "step": 15446
    },
    {
      "epoch": 2.8706560118937,
      "grad_norm": 0.7449698448181152,
      "learning_rate": 3.696810133628824e-05,
      "loss": 0.7441,
      "step": 15447
    },
    {
      "epoch": 2.8708418509570715,
      "grad_norm": 0.7663121819496155,
      "learning_rate": 3.6956760248083064e-05,
      "loss": 0.9675,
      "step": 15448
    },
    {
      "epoch": 2.8710276900204423,
      "grad_norm": 0.8151096105575562,
      "learning_rate": 3.694542050542763e-05,
      "loss": 0.9595,
      "step": 15449
    },
    {
      "epoch": 2.871213529083813,
      "grad_norm": 0.9838616847991943,
      "learning_rate": 3.6934082108564016e-05,
      "loss": 0.8339,
      "step": 15450
    },
    {
      "epoch": 2.8713993681471845,
      "grad_norm": 0.8464575409889221,
      "learning_rate": 3.692274505773419e-05,
      "loss": 1.0164,
      "step": 15451
    },
    {
      "epoch": 2.871585207210556,
      "grad_norm": 0.8795459270477295,
      "learning_rate": 3.691140935318012e-05,
      "loss": 0.9412,
      "step": 15452
    },
    {
      "epoch": 2.8717710462739268,
      "grad_norm": 0.864625871181488,
      "learning_rate": 3.690007499514374e-05,
      "loss": 0.7772,
      "step": 15453
    },
    {
      "epoch": 2.8719568853372976,
      "grad_norm": 0.8079337477684021,
      "learning_rate": 3.688874198386699e-05,
      "loss": 1.065,
      "step": 15454
    },
    {
      "epoch": 2.872142724400669,
      "grad_norm": 0.7533987164497375,
      "learning_rate": 3.687741031959175e-05,
      "loss": 0.942,
      "step": 15455
    },
    {
      "epoch": 2.8723285634640403,
      "grad_norm": 0.7962412238121033,
      "learning_rate": 3.686608000255982e-05,
      "loss": 1.1564,
      "step": 15456
    },
    {
      "epoch": 2.872514402527411,
      "grad_norm": 0.8227654099464417,
      "learning_rate": 3.685475103301308e-05,
      "loss": 0.8466,
      "step": 15457
    },
    {
      "epoch": 2.8727002415907825,
      "grad_norm": 0.8999118208885193,
      "learning_rate": 3.684342341119335e-05,
      "loss": 0.938,
      "step": 15458
    },
    {
      "epoch": 2.8728860806541534,
      "grad_norm": 1.2973206043243408,
      "learning_rate": 3.683209713734237e-05,
      "loss": 1.1469,
      "step": 15459
    },
    {
      "epoch": 2.8730719197175247,
      "grad_norm": 1.1290135383605957,
      "learning_rate": 3.682077221170186e-05,
      "loss": 0.9415,
      "step": 15460
    },
    {
      "epoch": 2.8732577587808956,
      "grad_norm": 0.8910056948661804,
      "learning_rate": 3.680944863451353e-05,
      "loss": 1.1199,
      "step": 15461
    },
    {
      "epoch": 2.873443597844267,
      "grad_norm": 0.7031981348991394,
      "learning_rate": 3.6798126406019104e-05,
      "loss": 0.7767,
      "step": 15462
    },
    {
      "epoch": 2.873629436907638,
      "grad_norm": 0.757603645324707,
      "learning_rate": 3.678680552646021e-05,
      "loss": 0.8649,
      "step": 15463
    },
    {
      "epoch": 2.873815275971009,
      "grad_norm": 0.8971279263496399,
      "learning_rate": 3.677548599607845e-05,
      "loss": 0.9413,
      "step": 15464
    },
    {
      "epoch": 2.87400111503438,
      "grad_norm": 0.867313027381897,
      "learning_rate": 3.676416781511547e-05,
      "loss": 0.8164,
      "step": 15465
    },
    {
      "epoch": 2.8741869540977514,
      "grad_norm": 0.9275748133659363,
      "learning_rate": 3.675285098381281e-05,
      "loss": 0.9774,
      "step": 15466
    },
    {
      "epoch": 2.8743727931611227,
      "grad_norm": 0.8772784471511841,
      "learning_rate": 3.674153550241202e-05,
      "loss": 0.726,
      "step": 15467
    },
    {
      "epoch": 2.8745586322244936,
      "grad_norm": 0.8420283794403076,
      "learning_rate": 3.673022137115455e-05,
      "loss": 1.0059,
      "step": 15468
    },
    {
      "epoch": 2.8747444712878645,
      "grad_norm": 0.8634923696517944,
      "learning_rate": 3.671890859028194e-05,
      "loss": 0.7513,
      "step": 15469
    },
    {
      "epoch": 2.874930310351236,
      "grad_norm": 0.9102967381477356,
      "learning_rate": 3.6707597160035667e-05,
      "loss": 0.9892,
      "step": 15470
    },
    {
      "epoch": 2.875116149414607,
      "grad_norm": 0.8464215397834778,
      "learning_rate": 3.669628708065711e-05,
      "loss": 1.0069,
      "step": 15471
    },
    {
      "epoch": 2.875301988477978,
      "grad_norm": 0.8910304307937622,
      "learning_rate": 3.668497835238768e-05,
      "loss": 1.0988,
      "step": 15472
    },
    {
      "epoch": 2.8754878275413494,
      "grad_norm": 0.874662458896637,
      "learning_rate": 3.667367097546869e-05,
      "loss": 0.8517,
      "step": 15473
    },
    {
      "epoch": 2.8756736666047202,
      "grad_norm": 0.841926634311676,
      "learning_rate": 3.666236495014156e-05,
      "loss": 0.7573,
      "step": 15474
    },
    {
      "epoch": 2.8758595056680916,
      "grad_norm": 0.7800859212875366,
      "learning_rate": 3.665106027664755e-05,
      "loss": 0.9069,
      "step": 15475
    },
    {
      "epoch": 2.8760453447314624,
      "grad_norm": 0.7146313190460205,
      "learning_rate": 3.6639756955227936e-05,
      "loss": 0.8104,
      "step": 15476
    },
    {
      "epoch": 2.876231183794834,
      "grad_norm": 0.6743584275245667,
      "learning_rate": 3.6628454986123995e-05,
      "loss": 0.7703,
      "step": 15477
    },
    {
      "epoch": 2.8764170228582047,
      "grad_norm": 0.878092885017395,
      "learning_rate": 3.66171543695769e-05,
      "loss": 0.9727,
      "step": 15478
    },
    {
      "epoch": 2.876602861921576,
      "grad_norm": 0.8362070322036743,
      "learning_rate": 3.6605855105827915e-05,
      "loss": 1.1399,
      "step": 15479
    },
    {
      "epoch": 2.876788700984947,
      "grad_norm": 0.9773759245872498,
      "learning_rate": 3.6594557195118165e-05,
      "loss": 0.9773,
      "step": 15480
    },
    {
      "epoch": 2.876974540048318,
      "grad_norm": 0.8178821802139282,
      "learning_rate": 3.6583260637688744e-05,
      "loss": 1.218,
      "step": 15481
    },
    {
      "epoch": 2.8771603791116895,
      "grad_norm": 0.6931003928184509,
      "learning_rate": 3.657196543378083e-05,
      "loss": 0.6433,
      "step": 15482
    },
    {
      "epoch": 2.8773462181750604,
      "grad_norm": 0.8900542855262756,
      "learning_rate": 3.6560671583635467e-05,
      "loss": 0.7365,
      "step": 15483
    },
    {
      "epoch": 2.8775320572384313,
      "grad_norm": 0.8439195156097412,
      "learning_rate": 3.6549379087493674e-05,
      "loss": 0.8774,
      "step": 15484
    },
    {
      "epoch": 2.8777178963018026,
      "grad_norm": 0.7627041935920715,
      "learning_rate": 3.653808794559653e-05,
      "loss": 0.8925,
      "step": 15485
    },
    {
      "epoch": 2.877903735365174,
      "grad_norm": 0.8097158074378967,
      "learning_rate": 3.6526798158184994e-05,
      "loss": 1.0422,
      "step": 15486
    },
    {
      "epoch": 2.878089574428545,
      "grad_norm": 0.7690940499305725,
      "learning_rate": 3.651550972550003e-05,
      "loss": 0.8416,
      "step": 15487
    },
    {
      "epoch": 2.8782754134919157,
      "grad_norm": 0.9093706607818604,
      "learning_rate": 3.650422264778253e-05,
      "loss": 0.9186,
      "step": 15488
    },
    {
      "epoch": 2.878461252555287,
      "grad_norm": 0.8120034337043762,
      "learning_rate": 3.649293692527344e-05,
      "loss": 0.949,
      "step": 15489
    },
    {
      "epoch": 2.8786470916186584,
      "grad_norm": 0.9720838069915771,
      "learning_rate": 3.6481652558213654e-05,
      "loss": 0.9473,
      "step": 15490
    },
    {
      "epoch": 2.8788329306820293,
      "grad_norm": 0.7634152173995972,
      "learning_rate": 3.647036954684399e-05,
      "loss": 0.9627,
      "step": 15491
    },
    {
      "epoch": 2.8790187697454006,
      "grad_norm": 0.7963048219680786,
      "learning_rate": 3.645908789140527e-05,
      "loss": 0.8999,
      "step": 15492
    },
    {
      "epoch": 2.8792046088087715,
      "grad_norm": 0.798642098903656,
      "learning_rate": 3.644780759213825e-05,
      "loss": 0.8625,
      "step": 15493
    },
    {
      "epoch": 2.879390447872143,
      "grad_norm": 0.859175443649292,
      "learning_rate": 3.6436528649283754e-05,
      "loss": 0.9735,
      "step": 15494
    },
    {
      "epoch": 2.8795762869355137,
      "grad_norm": 0.9950864315032959,
      "learning_rate": 3.6425251063082465e-05,
      "loss": 0.956,
      "step": 15495
    },
    {
      "epoch": 2.879762125998885,
      "grad_norm": 0.7690659761428833,
      "learning_rate": 3.6413974833775055e-05,
      "loss": 0.8988,
      "step": 15496
    },
    {
      "epoch": 2.879947965062256,
      "grad_norm": 1.670119285583496,
      "learning_rate": 3.6402699961602285e-05,
      "loss": 1.1537,
      "step": 15497
    },
    {
      "epoch": 2.8801338041256273,
      "grad_norm": 0.7463253140449524,
      "learning_rate": 3.639142644680473e-05,
      "loss": 0.6221,
      "step": 15498
    },
    {
      "epoch": 2.880319643188998,
      "grad_norm": 0.7164505124092102,
      "learning_rate": 3.6380154289623e-05,
      "loss": 0.97,
      "step": 15499
    },
    {
      "epoch": 2.8805054822523695,
      "grad_norm": 0.8878457546234131,
      "learning_rate": 3.6368883490297734e-05,
      "loss": 0.9194,
      "step": 15500
    },
    {
      "epoch": 2.880691321315741,
      "grad_norm": 0.733720600605011,
      "learning_rate": 3.635761404906941e-05,
      "loss": 0.7988,
      "step": 15501
    },
    {
      "epoch": 2.8808771603791117,
      "grad_norm": 0.839289665222168,
      "learning_rate": 3.634634596617864e-05,
      "loss": 0.8863,
      "step": 15502
    },
    {
      "epoch": 2.8810629994424826,
      "grad_norm": 0.6451913118362427,
      "learning_rate": 3.633507924186588e-05,
      "loss": 0.6838,
      "step": 15503
    },
    {
      "epoch": 2.881248838505854,
      "grad_norm": 0.7698342204093933,
      "learning_rate": 3.632381387637157e-05,
      "loss": 0.7794,
      "step": 15504
    },
    {
      "epoch": 2.8814346775692252,
      "grad_norm": 0.6710625886917114,
      "learning_rate": 3.6312549869936206e-05,
      "loss": 0.735,
      "step": 15505
    },
    {
      "epoch": 2.881620516632596,
      "grad_norm": 0.8364967107772827,
      "learning_rate": 3.630128722280019e-05,
      "loss": 0.7442,
      "step": 15506
    },
    {
      "epoch": 2.8818063556959674,
      "grad_norm": 0.6898694634437561,
      "learning_rate": 3.629002593520386e-05,
      "loss": 0.8518,
      "step": 15507
    },
    {
      "epoch": 2.8819921947593383,
      "grad_norm": 0.9124313592910767,
      "learning_rate": 3.627876600738759e-05,
      "loss": 0.8588,
      "step": 15508
    },
    {
      "epoch": 2.8821780338227097,
      "grad_norm": 0.8680217266082764,
      "learning_rate": 3.626750743959173e-05,
      "loss": 0.9466,
      "step": 15509
    },
    {
      "epoch": 2.8823638728860805,
      "grad_norm": 0.8426185250282288,
      "learning_rate": 3.625625023205652e-05,
      "loss": 0.8538,
      "step": 15510
    },
    {
      "epoch": 2.882549711949452,
      "grad_norm": 0.9768823385238647,
      "learning_rate": 3.624499438502229e-05,
      "loss": 0.8646,
      "step": 15511
    },
    {
      "epoch": 2.8827355510128228,
      "grad_norm": 0.7641948461532593,
      "learning_rate": 3.623373989872926e-05,
      "loss": 0.8495,
      "step": 15512
    },
    {
      "epoch": 2.882921390076194,
      "grad_norm": 0.8565607666969299,
      "learning_rate": 3.6222486773417574e-05,
      "loss": 0.9625,
      "step": 15513
    },
    {
      "epoch": 2.883107229139565,
      "grad_norm": 0.7964957356452942,
      "learning_rate": 3.6211235009327505e-05,
      "loss": 0.9454,
      "step": 15514
    },
    {
      "epoch": 2.8832930682029363,
      "grad_norm": 0.8411273956298828,
      "learning_rate": 3.6199984606699155e-05,
      "loss": 0.7864,
      "step": 15515
    },
    {
      "epoch": 2.8834789072663076,
      "grad_norm": 1.5076212882995605,
      "learning_rate": 3.618873556577261e-05,
      "loss": 1.1191,
      "step": 15516
    },
    {
      "epoch": 2.8836647463296785,
      "grad_norm": 0.8100115656852722,
      "learning_rate": 3.6177487886788054e-05,
      "loss": 0.8513,
      "step": 15517
    },
    {
      "epoch": 2.8838505853930494,
      "grad_norm": 0.8636518716812134,
      "learning_rate": 3.6166241569985484e-05,
      "loss": 0.8855,
      "step": 15518
    },
    {
      "epoch": 2.8840364244564207,
      "grad_norm": 0.7322862148284912,
      "learning_rate": 3.615499661560493e-05,
      "loss": 0.9125,
      "step": 15519
    },
    {
      "epoch": 2.884222263519792,
      "grad_norm": 0.7911383509635925,
      "learning_rate": 3.614375302388639e-05,
      "loss": 1.0129,
      "step": 15520
    },
    {
      "epoch": 2.884408102583163,
      "grad_norm": 0.9409505724906921,
      "learning_rate": 3.613251079506986e-05,
      "loss": 0.7448,
      "step": 15521
    },
    {
      "epoch": 2.8845939416465343,
      "grad_norm": 0.8631063103675842,
      "learning_rate": 3.612126992939532e-05,
      "loss": 0.7665,
      "step": 15522
    },
    {
      "epoch": 2.884779780709905,
      "grad_norm": 0.7590117454528809,
      "learning_rate": 3.611003042710266e-05,
      "loss": 1.132,
      "step": 15523
    },
    {
      "epoch": 2.8849656197732765,
      "grad_norm": 1.4974132776260376,
      "learning_rate": 3.609879228843171e-05,
      "loss": 1.4162,
      "step": 15524
    },
    {
      "epoch": 2.8851514588366474,
      "grad_norm": 0.9238195419311523,
      "learning_rate": 3.608755551362243e-05,
      "loss": 0.8645,
      "step": 15525
    },
    {
      "epoch": 2.8853372979000187,
      "grad_norm": 0.7798268795013428,
      "learning_rate": 3.6076320102914585e-05,
      "loss": 0.8418,
      "step": 15526
    },
    {
      "epoch": 2.8855231369633896,
      "grad_norm": 0.9914414286613464,
      "learning_rate": 3.6065086056548004e-05,
      "loss": 1.0875,
      "step": 15527
    },
    {
      "epoch": 2.885708976026761,
      "grad_norm": 0.885212779045105,
      "learning_rate": 3.6053853374762404e-05,
      "loss": 1.2031,
      "step": 15528
    },
    {
      "epoch": 2.885894815090132,
      "grad_norm": 0.7570885419845581,
      "learning_rate": 3.6042622057797605e-05,
      "loss": 0.7165,
      "step": 15529
    },
    {
      "epoch": 2.886080654153503,
      "grad_norm": 0.8846122026443481,
      "learning_rate": 3.603139210589327e-05,
      "loss": 1.0756,
      "step": 15530
    },
    {
      "epoch": 2.8862664932168745,
      "grad_norm": 0.8745748996734619,
      "learning_rate": 3.6020163519289076e-05,
      "loss": 0.9672,
      "step": 15531
    },
    {
      "epoch": 2.8864523322802453,
      "grad_norm": 0.6768541932106018,
      "learning_rate": 3.600893629822473e-05,
      "loss": 0.8916,
      "step": 15532
    },
    {
      "epoch": 2.886638171343616,
      "grad_norm": 0.9766746163368225,
      "learning_rate": 3.599771044293978e-05,
      "loss": 0.5867,
      "step": 15533
    },
    {
      "epoch": 2.8868240104069876,
      "grad_norm": 0.8952239751815796,
      "learning_rate": 3.598648595367392e-05,
      "loss": 0.9723,
      "step": 15534
    },
    {
      "epoch": 2.887009849470359,
      "grad_norm": 0.7633703351020813,
      "learning_rate": 3.5975262830666655e-05,
      "loss": 1.0225,
      "step": 15535
    },
    {
      "epoch": 2.8871956885337298,
      "grad_norm": 0.6744645833969116,
      "learning_rate": 3.596404107415749e-05,
      "loss": 0.7406,
      "step": 15536
    },
    {
      "epoch": 2.8873815275971006,
      "grad_norm": 1.0360220670700073,
      "learning_rate": 3.5952820684386034e-05,
      "loss": 1.0136,
      "step": 15537
    },
    {
      "epoch": 2.887567366660472,
      "grad_norm": 0.8803971409797668,
      "learning_rate": 3.59416016615917e-05,
      "loss": 1.083,
      "step": 15538
    },
    {
      "epoch": 2.8877532057238433,
      "grad_norm": 0.8622626066207886,
      "learning_rate": 3.593038400601395e-05,
      "loss": 1.179,
      "step": 15539
    },
    {
      "epoch": 2.887939044787214,
      "grad_norm": 0.8891432285308838,
      "learning_rate": 3.591916771789218e-05,
      "loss": 0.8029,
      "step": 15540
    },
    {
      "epoch": 2.8881248838505855,
      "grad_norm": 0.7373136281967163,
      "learning_rate": 3.59079527974658e-05,
      "loss": 0.7517,
      "step": 15541
    },
    {
      "epoch": 2.8883107229139564,
      "grad_norm": 0.9604713320732117,
      "learning_rate": 3.589673924497421e-05,
      "loss": 0.9143,
      "step": 15542
    },
    {
      "epoch": 2.8884965619773277,
      "grad_norm": 0.9270872473716736,
      "learning_rate": 3.588552706065672e-05,
      "loss": 0.9747,
      "step": 15543
    },
    {
      "epoch": 2.8886824010406986,
      "grad_norm": 0.7779958844184875,
      "learning_rate": 3.58743162447526e-05,
      "loss": 0.8861,
      "step": 15544
    },
    {
      "epoch": 2.88886824010407,
      "grad_norm": 0.78629070520401,
      "learning_rate": 3.5863106797501187e-05,
      "loss": 0.745,
      "step": 15545
    },
    {
      "epoch": 2.889054079167441,
      "grad_norm": 1.6500821113586426,
      "learning_rate": 3.5851898719141695e-05,
      "loss": 1.2476,
      "step": 15546
    },
    {
      "epoch": 2.889239918230812,
      "grad_norm": 0.9943754076957703,
      "learning_rate": 3.5840692009913344e-05,
      "loss": 0.9949,
      "step": 15547
    },
    {
      "epoch": 2.889425757294183,
      "grad_norm": 0.8301264643669128,
      "learning_rate": 3.582948667005529e-05,
      "loss": 1.0129,
      "step": 15548
    },
    {
      "epoch": 2.8896115963575544,
      "grad_norm": 0.7343300580978394,
      "learning_rate": 3.581828269980675e-05,
      "loss": 0.446,
      "step": 15549
    },
    {
      "epoch": 2.8897974354209257,
      "grad_norm": 0.7584952712059021,
      "learning_rate": 3.580708009940682e-05,
      "loss": 0.9541,
      "step": 15550
    },
    {
      "epoch": 2.8899832744842966,
      "grad_norm": 0.7859680652618408,
      "learning_rate": 3.579587886909461e-05,
      "loss": 0.9432,
      "step": 15551
    },
    {
      "epoch": 2.8901691135476675,
      "grad_norm": 1.3147470951080322,
      "learning_rate": 3.5784679009109154e-05,
      "loss": 0.6812,
      "step": 15552
    },
    {
      "epoch": 2.890354952611039,
      "grad_norm": 0.7729043364524841,
      "learning_rate": 3.577348051968954e-05,
      "loss": 0.8775,
      "step": 15553
    },
    {
      "epoch": 2.89054079167441,
      "grad_norm": 0.8663438558578491,
      "learning_rate": 3.576228340107478e-05,
      "loss": 0.7809,
      "step": 15554
    },
    {
      "epoch": 2.890726630737781,
      "grad_norm": 0.8728376030921936,
      "learning_rate": 3.575108765350386e-05,
      "loss": 1.004,
      "step": 15555
    },
    {
      "epoch": 2.8909124698011524,
      "grad_norm": 0.7745139598846436,
      "learning_rate": 3.5739893277215676e-05,
      "loss": 0.803,
      "step": 15556
    },
    {
      "epoch": 2.8910983088645232,
      "grad_norm": 0.7463011741638184,
      "learning_rate": 3.5728700272449234e-05,
      "loss": 0.8841,
      "step": 15557
    },
    {
      "epoch": 2.8912841479278946,
      "grad_norm": 1.0205782651901245,
      "learning_rate": 3.5717508639443387e-05,
      "loss": 0.9652,
      "step": 15558
    },
    {
      "epoch": 2.8914699869912655,
      "grad_norm": 0.8964250683784485,
      "learning_rate": 3.570631837843701e-05,
      "loss": 0.9396,
      "step": 15559
    },
    {
      "epoch": 2.891655826054637,
      "grad_norm": 0.7447481751441956,
      "learning_rate": 3.569512948966889e-05,
      "loss": 0.8611,
      "step": 15560
    },
    {
      "epoch": 2.8918416651180077,
      "grad_norm": 0.8456870913505554,
      "learning_rate": 3.5683941973377924e-05,
      "loss": 0.9574,
      "step": 15561
    },
    {
      "epoch": 2.892027504181379,
      "grad_norm": 0.7943773865699768,
      "learning_rate": 3.5672755829802805e-05,
      "loss": 0.959,
      "step": 15562
    },
    {
      "epoch": 2.89221334324475,
      "grad_norm": 1.0943312644958496,
      "learning_rate": 3.5661571059182366e-05,
      "loss": 1.127,
      "step": 15563
    },
    {
      "epoch": 2.892399182308121,
      "grad_norm": 0.7443644404411316,
      "learning_rate": 3.565038766175525e-05,
      "loss": 0.728,
      "step": 15564
    },
    {
      "epoch": 2.8925850213714925,
      "grad_norm": 0.8033445477485657,
      "learning_rate": 3.56392056377602e-05,
      "loss": 0.6451,
      "step": 15565
    },
    {
      "epoch": 2.8927708604348634,
      "grad_norm": 0.8828470706939697,
      "learning_rate": 3.562802498743587e-05,
      "loss": 0.8077,
      "step": 15566
    },
    {
      "epoch": 2.8929566994982343,
      "grad_norm": 0.7708657383918762,
      "learning_rate": 3.561684571102087e-05,
      "loss": 0.8274,
      "step": 15567
    },
    {
      "epoch": 2.8931425385616056,
      "grad_norm": 0.8346666693687439,
      "learning_rate": 3.560566780875379e-05,
      "loss": 1.02,
      "step": 15568
    },
    {
      "epoch": 2.893328377624977,
      "grad_norm": 0.8248069882392883,
      "learning_rate": 3.5594491280873254e-05,
      "loss": 1.0995,
      "step": 15569
    },
    {
      "epoch": 2.893514216688348,
      "grad_norm": 0.7798810005187988,
      "learning_rate": 3.5583316127617784e-05,
      "loss": 0.9831,
      "step": 15570
    },
    {
      "epoch": 2.8937000557517187,
      "grad_norm": 0.8893752694129944,
      "learning_rate": 3.5572142349225876e-05,
      "loss": 0.9212,
      "step": 15571
    },
    {
      "epoch": 2.89388589481509,
      "grad_norm": 0.8199259042739868,
      "learning_rate": 3.556096994593599e-05,
      "loss": 0.6263,
      "step": 15572
    },
    {
      "epoch": 2.8940717338784614,
      "grad_norm": 0.8772500157356262,
      "learning_rate": 3.554979891798663e-05,
      "loss": 1.0401,
      "step": 15573
    },
    {
      "epoch": 2.8942575729418323,
      "grad_norm": 1.0153173208236694,
      "learning_rate": 3.553862926561625e-05,
      "loss": 0.7838,
      "step": 15574
    },
    {
      "epoch": 2.8944434120052036,
      "grad_norm": 0.8423423171043396,
      "learning_rate": 3.55274609890632e-05,
      "loss": 0.9374,
      "step": 15575
    },
    {
      "epoch": 2.8946292510685745,
      "grad_norm": 1.033982515335083,
      "learning_rate": 3.5516294088565813e-05,
      "loss": 1.1189,
      "step": 15576
    },
    {
      "epoch": 2.894815090131946,
      "grad_norm": 0.791053056716919,
      "learning_rate": 3.5505128564362525e-05,
      "loss": 0.9697,
      "step": 15577
    },
    {
      "epoch": 2.8950009291953167,
      "grad_norm": 0.827377438545227,
      "learning_rate": 3.5493964416691574e-05,
      "loss": 0.9403,
      "step": 15578
    },
    {
      "epoch": 2.895186768258688,
      "grad_norm": 0.7587912678718567,
      "learning_rate": 3.548280164579126e-05,
      "loss": 1.0134,
      "step": 15579
    },
    {
      "epoch": 2.8953726073220594,
      "grad_norm": 0.7401148080825806,
      "learning_rate": 3.5471640251899796e-05,
      "loss": 0.9301,
      "step": 15580
    },
    {
      "epoch": 2.8955584463854303,
      "grad_norm": 0.8820634484291077,
      "learning_rate": 3.546048023525547e-05,
      "loss": 1.0362,
      "step": 15581
    },
    {
      "epoch": 2.895744285448801,
      "grad_norm": 0.8928791284561157,
      "learning_rate": 3.544932159609644e-05,
      "loss": 0.9027,
      "step": 15582
    },
    {
      "epoch": 2.8959301245121725,
      "grad_norm": 0.8220198750495911,
      "learning_rate": 3.543816433466083e-05,
      "loss": 0.741,
      "step": 15583
    },
    {
      "epoch": 2.896115963575544,
      "grad_norm": 0.8550474047660828,
      "learning_rate": 3.542700845118682e-05,
      "loss": 0.786,
      "step": 15584
    },
    {
      "epoch": 2.8963018026389147,
      "grad_norm": 0.8651540875434875,
      "learning_rate": 3.541585394591252e-05,
      "loss": 1.1396,
      "step": 15585
    },
    {
      "epoch": 2.8964876417022856,
      "grad_norm": 0.8764159679412842,
      "learning_rate": 3.5404700819076e-05,
      "loss": 0.9171,
      "step": 15586
    },
    {
      "epoch": 2.896673480765657,
      "grad_norm": 0.7330661416053772,
      "learning_rate": 3.539354907091528e-05,
      "loss": 0.8973,
      "step": 15587
    },
    {
      "epoch": 2.8968593198290282,
      "grad_norm": 0.7530513405799866,
      "learning_rate": 3.538239870166836e-05,
      "loss": 0.7893,
      "step": 15588
    },
    {
      "epoch": 2.897045158892399,
      "grad_norm": 0.8874174356460571,
      "learning_rate": 3.537124971157328e-05,
      "loss": 1.1772,
      "step": 15589
    },
    {
      "epoch": 2.8972309979557704,
      "grad_norm": 0.9068132638931274,
      "learning_rate": 3.5360102100867976e-05,
      "loss": 1.0861,
      "step": 15590
    },
    {
      "epoch": 2.8974168370191413,
      "grad_norm": 0.7956612706184387,
      "learning_rate": 3.534895586979036e-05,
      "loss": 0.7394,
      "step": 15591
    },
    {
      "epoch": 2.8976026760825127,
      "grad_norm": 0.8411120176315308,
      "learning_rate": 3.53378110185783e-05,
      "loss": 0.9235,
      "step": 15592
    },
    {
      "epoch": 2.8977885151458835,
      "grad_norm": 0.9352555871009827,
      "learning_rate": 3.5326667547469735e-05,
      "loss": 0.9848,
      "step": 15593
    },
    {
      "epoch": 2.897974354209255,
      "grad_norm": 0.6587279438972473,
      "learning_rate": 3.531552545670244e-05,
      "loss": 0.8386,
      "step": 15594
    },
    {
      "epoch": 2.8981601932726258,
      "grad_norm": 0.8666529059410095,
      "learning_rate": 3.530438474651428e-05,
      "loss": 0.9736,
      "step": 15595
    },
    {
      "epoch": 2.898346032335997,
      "grad_norm": 0.8803017139434814,
      "learning_rate": 3.529324541714296e-05,
      "loss": 1.0456,
      "step": 15596
    },
    {
      "epoch": 2.898531871399368,
      "grad_norm": 0.830962598323822,
      "learning_rate": 3.528210746882632e-05,
      "loss": 0.8811,
      "step": 15597
    },
    {
      "epoch": 2.8987177104627393,
      "grad_norm": 0.7424424290657043,
      "learning_rate": 3.5270970901802036e-05,
      "loss": 0.6607,
      "step": 15598
    },
    {
      "epoch": 2.8989035495261106,
      "grad_norm": 0.9809721112251282,
      "learning_rate": 3.52598357163078e-05,
      "loss": 1.2557,
      "step": 15599
    },
    {
      "epoch": 2.8990893885894815,
      "grad_norm": 0.8323695063591003,
      "learning_rate": 3.524870191258124e-05,
      "loss": 0.6972,
      "step": 15600
    },
    {
      "epoch": 2.8992752276528524,
      "grad_norm": 0.7893000245094299,
      "learning_rate": 3.523756949086004e-05,
      "loss": 0.8696,
      "step": 15601
    },
    {
      "epoch": 2.8994610667162237,
      "grad_norm": 0.7958328127861023,
      "learning_rate": 3.5226438451381794e-05,
      "loss": 0.8583,
      "step": 15602
    },
    {
      "epoch": 2.899646905779595,
      "grad_norm": 0.7533215880393982,
      "learning_rate": 3.521530879438407e-05,
      "loss": 0.7273,
      "step": 15603
    },
    {
      "epoch": 2.899832744842966,
      "grad_norm": 0.790684163570404,
      "learning_rate": 3.520418052010432e-05,
      "loss": 0.9137,
      "step": 15604
    },
    {
      "epoch": 2.9000185839063373,
      "grad_norm": 0.6913897395133972,
      "learning_rate": 3.519305362878023e-05,
      "loss": 0.8058,
      "step": 15605
    },
    {
      "epoch": 2.900204422969708,
      "grad_norm": 0.8906787633895874,
      "learning_rate": 3.518192812064918e-05,
      "loss": 0.964,
      "step": 15606
    },
    {
      "epoch": 2.9003902620330795,
      "grad_norm": 0.9065820574760437,
      "learning_rate": 3.517080399594865e-05,
      "loss": 0.8591,
      "step": 15607
    },
    {
      "epoch": 2.9005761010964504,
      "grad_norm": 0.7903452515602112,
      "learning_rate": 3.515968125491602e-05,
      "loss": 0.7845,
      "step": 15608
    },
    {
      "epoch": 2.9007619401598217,
      "grad_norm": 1.014837622642517,
      "learning_rate": 3.514855989778876e-05,
      "loss": 1.0081,
      "step": 15609
    },
    {
      "epoch": 2.9009477792231926,
      "grad_norm": 0.8896788954734802,
      "learning_rate": 3.513743992480419e-05,
      "loss": 0.9189,
      "step": 15610
    },
    {
      "epoch": 2.901133618286564,
      "grad_norm": 0.7536845803260803,
      "learning_rate": 3.5126321336199676e-05,
      "loss": 0.9054,
      "step": 15611
    },
    {
      "epoch": 2.901319457349935,
      "grad_norm": 0.7198872566223145,
      "learning_rate": 3.5115204132212454e-05,
      "loss": 0.7971,
      "step": 15612
    },
    {
      "epoch": 2.901505296413306,
      "grad_norm": 0.8007821440696716,
      "learning_rate": 3.51040883130799e-05,
      "loss": 0.9311,
      "step": 15613
    },
    {
      "epoch": 2.9016911354766775,
      "grad_norm": 0.8197160363197327,
      "learning_rate": 3.50929738790392e-05,
      "loss": 0.991,
      "step": 15614
    },
    {
      "epoch": 2.9018769745400483,
      "grad_norm": 0.9037057757377625,
      "learning_rate": 3.5081860830327564e-05,
      "loss": 1.0113,
      "step": 15615
    },
    {
      "epoch": 2.9020628136034192,
      "grad_norm": 0.9215180277824402,
      "learning_rate": 3.50707491671822e-05,
      "loss": 0.966,
      "step": 15616
    },
    {
      "epoch": 2.9022486526667906,
      "grad_norm": 0.7795912027359009,
      "learning_rate": 3.505963888984031e-05,
      "loss": 0.9497,
      "step": 15617
    },
    {
      "epoch": 2.902434491730162,
      "grad_norm": 0.7926895618438721,
      "learning_rate": 3.504852999853898e-05,
      "loss": 1.0439,
      "step": 15618
    },
    {
      "epoch": 2.9026203307935328,
      "grad_norm": 0.8523248434066772,
      "learning_rate": 3.503742249351533e-05,
      "loss": 0.918,
      "step": 15619
    },
    {
      "epoch": 2.9028061698569037,
      "grad_norm": 0.9220716953277588,
      "learning_rate": 3.502631637500637e-05,
      "loss": 1.1795,
      "step": 15620
    },
    {
      "epoch": 2.902992008920275,
      "grad_norm": 0.9029595255851746,
      "learning_rate": 3.501521164324922e-05,
      "loss": 0.8578,
      "step": 15621
    },
    {
      "epoch": 2.9031778479836463,
      "grad_norm": 0.9522722363471985,
      "learning_rate": 3.5004108298480866e-05,
      "loss": 0.8879,
      "step": 15622
    },
    {
      "epoch": 2.903363687047017,
      "grad_norm": 0.6594275236129761,
      "learning_rate": 3.499300634093828e-05,
      "loss": 0.6583,
      "step": 15623
    },
    {
      "epoch": 2.9035495261103885,
      "grad_norm": 0.7711548805236816,
      "learning_rate": 3.498190577085838e-05,
      "loss": 0.8813,
      "step": 15624
    },
    {
      "epoch": 2.9037353651737594,
      "grad_norm": 0.8120653629302979,
      "learning_rate": 3.4970806588478136e-05,
      "loss": 0.8379,
      "step": 15625
    },
    {
      "epoch": 2.9039212042371307,
      "grad_norm": 0.7879056930541992,
      "learning_rate": 3.4959708794034453e-05,
      "loss": 0.8443,
      "step": 15626
    },
    {
      "epoch": 2.9041070433005016,
      "grad_norm": 0.6453332901000977,
      "learning_rate": 3.494861238776418e-05,
      "loss": 0.6438,
      "step": 15627
    },
    {
      "epoch": 2.904292882363873,
      "grad_norm": 0.7373093962669373,
      "learning_rate": 3.493751736990411e-05,
      "loss": 0.7821,
      "step": 15628
    },
    {
      "epoch": 2.9044787214272443,
      "grad_norm": 0.9499757885932922,
      "learning_rate": 3.4926423740691106e-05,
      "loss": 1.0659,
      "step": 15629
    },
    {
      "epoch": 2.904664560490615,
      "grad_norm": 0.8392454385757446,
      "learning_rate": 3.4915331500361924e-05,
      "loss": 0.9771,
      "step": 15630
    },
    {
      "epoch": 2.904850399553986,
      "grad_norm": 0.9138320684432983,
      "learning_rate": 3.49042406491533e-05,
      "loss": 1.0365,
      "step": 15631
    },
    {
      "epoch": 2.9050362386173574,
      "grad_norm": 0.8812173008918762,
      "learning_rate": 3.489315118730191e-05,
      "loss": 1.1716,
      "step": 15632
    },
    {
      "epoch": 2.9052220776807287,
      "grad_norm": 0.9005640745162964,
      "learning_rate": 3.488206311504452e-05,
      "loss": 0.9469,
      "step": 15633
    },
    {
      "epoch": 2.9054079167440996,
      "grad_norm": 0.8597601652145386,
      "learning_rate": 3.487097643261774e-05,
      "loss": 0.9493,
      "step": 15634
    },
    {
      "epoch": 2.9055937558074705,
      "grad_norm": 0.8719737529754639,
      "learning_rate": 3.4859891140258183e-05,
      "loss": 0.7799,
      "step": 15635
    },
    {
      "epoch": 2.905779594870842,
      "grad_norm": 0.7709887623786926,
      "learning_rate": 3.484880723820245e-05,
      "loss": 0.8098,
      "step": 15636
    },
    {
      "epoch": 2.905965433934213,
      "grad_norm": 0.7473528385162354,
      "learning_rate": 3.483772472668716e-05,
      "loss": 1.0006,
      "step": 15637
    },
    {
      "epoch": 2.906151272997584,
      "grad_norm": 0.7472079396247864,
      "learning_rate": 3.482664360594882e-05,
      "loss": 0.8065,
      "step": 15638
    },
    {
      "epoch": 2.9063371120609554,
      "grad_norm": 0.7801439762115479,
      "learning_rate": 3.481556387622392e-05,
      "loss": 0.7806,
      "step": 15639
    },
    {
      "epoch": 2.9065229511243262,
      "grad_norm": 0.704574465751648,
      "learning_rate": 3.4804485537748923e-05,
      "loss": 0.9096,
      "step": 15640
    },
    {
      "epoch": 2.9067087901876976,
      "grad_norm": 0.8833271861076355,
      "learning_rate": 3.4793408590760324e-05,
      "loss": 0.8006,
      "step": 15641
    },
    {
      "epoch": 2.9068946292510685,
      "grad_norm": 1.2138185501098633,
      "learning_rate": 3.478233303549453e-05,
      "loss": 0.8233,
      "step": 15642
    },
    {
      "epoch": 2.90708046831444,
      "grad_norm": 0.7799171209335327,
      "learning_rate": 3.477125887218792e-05,
      "loss": 0.8199,
      "step": 15643
    },
    {
      "epoch": 2.9072663073778107,
      "grad_norm": 1.0028904676437378,
      "learning_rate": 3.476018610107681e-05,
      "loss": 1.0786,
      "step": 15644
    },
    {
      "epoch": 2.907452146441182,
      "grad_norm": 1.0555802583694458,
      "learning_rate": 3.474911472239761e-05,
      "loss": 1.0305,
      "step": 15645
    },
    {
      "epoch": 2.907637985504553,
      "grad_norm": 0.7084522843360901,
      "learning_rate": 3.473804473638654e-05,
      "loss": 0.8662,
      "step": 15646
    },
    {
      "epoch": 2.907823824567924,
      "grad_norm": 0.8872708082199097,
      "learning_rate": 3.472697614327996e-05,
      "loss": 1.1082,
      "step": 15647
    },
    {
      "epoch": 2.9080096636312955,
      "grad_norm": 0.8006177544593811,
      "learning_rate": 3.471590894331402e-05,
      "loss": 0.9288,
      "step": 15648
    },
    {
      "epoch": 2.9081955026946664,
      "grad_norm": 0.8129321336746216,
      "learning_rate": 3.470484313672501e-05,
      "loss": 0.988,
      "step": 15649
    },
    {
      "epoch": 2.9083813417580373,
      "grad_norm": 0.7988606691360474,
      "learning_rate": 3.469377872374907e-05,
      "loss": 0.8885,
      "step": 15650
    },
    {
      "epoch": 2.9085671808214086,
      "grad_norm": 1.041595697402954,
      "learning_rate": 3.468271570462235e-05,
      "loss": 1.1708,
      "step": 15651
    },
    {
      "epoch": 2.90875301988478,
      "grad_norm": 0.9497601389884949,
      "learning_rate": 3.467165407958094e-05,
      "loss": 0.8956,
      "step": 15652
    },
    {
      "epoch": 2.908938858948151,
      "grad_norm": 0.9359521865844727,
      "learning_rate": 3.4660593848861e-05,
      "loss": 0.8024,
      "step": 15653
    },
    {
      "epoch": 2.909124698011522,
      "grad_norm": 0.8633369207382202,
      "learning_rate": 3.464953501269855e-05,
      "loss": 0.8987,
      "step": 15654
    },
    {
      "epoch": 2.909310537074893,
      "grad_norm": 0.725227415561676,
      "learning_rate": 3.4638477571329606e-05,
      "loss": 0.9908,
      "step": 15655
    },
    {
      "epoch": 2.9094963761382644,
      "grad_norm": 0.8891089558601379,
      "learning_rate": 3.462742152499023e-05,
      "loss": 1.2938,
      "step": 15656
    },
    {
      "epoch": 2.9096822152016353,
      "grad_norm": 0.8465229868888855,
      "learning_rate": 3.461636687391632e-05,
      "loss": 0.8997,
      "step": 15657
    },
    {
      "epoch": 2.9098680542650066,
      "grad_norm": 0.7925376892089844,
      "learning_rate": 3.460531361834389e-05,
      "loss": 0.6933,
      "step": 15658
    },
    {
      "epoch": 2.9100538933283775,
      "grad_norm": 0.7975295782089233,
      "learning_rate": 3.459426175850882e-05,
      "loss": 0.9303,
      "step": 15659
    },
    {
      "epoch": 2.910239732391749,
      "grad_norm": 0.7541177272796631,
      "learning_rate": 3.458321129464696e-05,
      "loss": 0.8467,
      "step": 15660
    },
    {
      "epoch": 2.9104255714551197,
      "grad_norm": 0.784887969493866,
      "learning_rate": 3.457216222699422e-05,
      "loss": 0.9571,
      "step": 15661
    },
    {
      "epoch": 2.910611410518491,
      "grad_norm": 0.856224000453949,
      "learning_rate": 3.45611145557864e-05,
      "loss": 0.9387,
      "step": 15662
    },
    {
      "epoch": 2.9107972495818624,
      "grad_norm": 0.8426495790481567,
      "learning_rate": 3.4550068281259295e-05,
      "loss": 1.0054,
      "step": 15663
    },
    {
      "epoch": 2.9109830886452333,
      "grad_norm": 0.8616632223129272,
      "learning_rate": 3.4539023403648644e-05,
      "loss": 0.7377,
      "step": 15664
    },
    {
      "epoch": 2.911168927708604,
      "grad_norm": 0.8532987833023071,
      "learning_rate": 3.452797992319022e-05,
      "loss": 1.0197,
      "step": 15665
    },
    {
      "epoch": 2.9113547667719755,
      "grad_norm": 0.776501476764679,
      "learning_rate": 3.451693784011971e-05,
      "loss": 0.6837,
      "step": 15666
    },
    {
      "epoch": 2.911540605835347,
      "grad_norm": 0.8846834301948547,
      "learning_rate": 3.450589715467276e-05,
      "loss": 0.8795,
      "step": 15667
    },
    {
      "epoch": 2.9117264448987177,
      "grad_norm": 0.8329945206642151,
      "learning_rate": 3.449485786708505e-05,
      "loss": 0.9378,
      "step": 15668
    },
    {
      "epoch": 2.9119122839620886,
      "grad_norm": 0.8448494076728821,
      "learning_rate": 3.4483819977592216e-05,
      "loss": 1.0225,
      "step": 15669
    },
    {
      "epoch": 2.91209812302546,
      "grad_norm": 0.7854666709899902,
      "learning_rate": 3.447278348642982e-05,
      "loss": 0.8708,
      "step": 15670
    },
    {
      "epoch": 2.9122839620888312,
      "grad_norm": 0.8177938461303711,
      "learning_rate": 3.446174839383339e-05,
      "loss": 0.9076,
      "step": 15671
    },
    {
      "epoch": 2.912469801152202,
      "grad_norm": 0.7455102205276489,
      "learning_rate": 3.4450714700038453e-05,
      "loss": 0.8464,
      "step": 15672
    },
    {
      "epoch": 2.9126556402155734,
      "grad_norm": 0.8319911956787109,
      "learning_rate": 3.443968240528056e-05,
      "loss": 0.9907,
      "step": 15673
    },
    {
      "epoch": 2.9128414792789443,
      "grad_norm": 0.7769563794136047,
      "learning_rate": 3.442865150979514e-05,
      "loss": 0.9798,
      "step": 15674
    },
    {
      "epoch": 2.9130273183423157,
      "grad_norm": 0.6890460252761841,
      "learning_rate": 3.4417622013817595e-05,
      "loss": 0.8935,
      "step": 15675
    },
    {
      "epoch": 2.9132131574056865,
      "grad_norm": 0.9530012607574463,
      "learning_rate": 3.4406593917583397e-05,
      "loss": 0.7713,
      "step": 15676
    },
    {
      "epoch": 2.913398996469058,
      "grad_norm": 0.9716929197311401,
      "learning_rate": 3.439556722132788e-05,
      "loss": 0.9907,
      "step": 15677
    },
    {
      "epoch": 2.9135848355324288,
      "grad_norm": 0.7098780870437622,
      "learning_rate": 3.438454192528637e-05,
      "loss": 0.6473,
      "step": 15678
    },
    {
      "epoch": 2.9137706745958,
      "grad_norm": 0.800399661064148,
      "learning_rate": 3.437351802969425e-05,
      "loss": 0.9786,
      "step": 15679
    },
    {
      "epoch": 2.913956513659171,
      "grad_norm": 0.7401872873306274,
      "learning_rate": 3.436249553478674e-05,
      "loss": 0.7292,
      "step": 15680
    },
    {
      "epoch": 2.9141423527225423,
      "grad_norm": 1.1086809635162354,
      "learning_rate": 3.4351474440799144e-05,
      "loss": 1.0502,
      "step": 15681
    },
    {
      "epoch": 2.9143281917859136,
      "grad_norm": 1.0315526723861694,
      "learning_rate": 3.434045474796668e-05,
      "loss": 0.8773,
      "step": 15682
    },
    {
      "epoch": 2.9145140308492845,
      "grad_norm": 0.9345428943634033,
      "learning_rate": 3.432943645652453e-05,
      "loss": 0.9042,
      "step": 15683
    },
    {
      "epoch": 2.9146998699126554,
      "grad_norm": 0.8518444895744324,
      "learning_rate": 3.4318419566707826e-05,
      "loss": 0.9686,
      "step": 15684
    },
    {
      "epoch": 2.9148857089760267,
      "grad_norm": 0.8447026610374451,
      "learning_rate": 3.4307404078751784e-05,
      "loss": 1.016,
      "step": 15685
    },
    {
      "epoch": 2.915071548039398,
      "grad_norm": 0.8735696077346802,
      "learning_rate": 3.4296389992891454e-05,
      "loss": 1.0496,
      "step": 15686
    },
    {
      "epoch": 2.915257387102769,
      "grad_norm": 0.7049415707588196,
      "learning_rate": 3.428537730936191e-05,
      "loss": 0.8717,
      "step": 15687
    },
    {
      "epoch": 2.9154432261661403,
      "grad_norm": 0.7728877663612366,
      "learning_rate": 3.427436602839821e-05,
      "loss": 0.7514,
      "step": 15688
    },
    {
      "epoch": 2.915629065229511,
      "grad_norm": 4.035832405090332,
      "learning_rate": 3.42633561502354e-05,
      "loss": 1.3198,
      "step": 15689
    },
    {
      "epoch": 2.9158149042928825,
      "grad_norm": 0.9382774829864502,
      "learning_rate": 3.425234767510847e-05,
      "loss": 0.7971,
      "step": 15690
    },
    {
      "epoch": 2.9160007433562534,
      "grad_norm": 0.7563207149505615,
      "learning_rate": 3.424134060325233e-05,
      "loss": 0.7422,
      "step": 15691
    },
    {
      "epoch": 2.9161865824196247,
      "grad_norm": 0.8817317485809326,
      "learning_rate": 3.42303349349019e-05,
      "loss": 1.0201,
      "step": 15692
    },
    {
      "epoch": 2.9163724214829956,
      "grad_norm": 1.0669676065444946,
      "learning_rate": 3.421933067029214e-05,
      "loss": 0.9374,
      "step": 15693
    },
    {
      "epoch": 2.916558260546367,
      "grad_norm": 0.8332189321517944,
      "learning_rate": 3.420832780965788e-05,
      "loss": 0.7491,
      "step": 15694
    },
    {
      "epoch": 2.916744099609738,
      "grad_norm": 0.9186729788780212,
      "learning_rate": 3.419732635323394e-05,
      "loss": 1.0503,
      "step": 15695
    },
    {
      "epoch": 2.916929938673109,
      "grad_norm": 0.7349416017532349,
      "learning_rate": 3.418632630125517e-05,
      "loss": 0.768,
      "step": 15696
    },
    {
      "epoch": 2.9171157777364805,
      "grad_norm": 0.8888520002365112,
      "learning_rate": 3.4175327653956325e-05,
      "loss": 1.0243,
      "step": 15697
    },
    {
      "epoch": 2.9173016167998513,
      "grad_norm": 0.8811149597167969,
      "learning_rate": 3.416433041157214e-05,
      "loss": 1.1016,
      "step": 15698
    },
    {
      "epoch": 2.9174874558632222,
      "grad_norm": 0.8639944195747375,
      "learning_rate": 3.415333457433733e-05,
      "loss": 0.8876,
      "step": 15699
    },
    {
      "epoch": 2.9176732949265936,
      "grad_norm": 0.8154263496398926,
      "learning_rate": 3.4142340142486584e-05,
      "loss": 0.8681,
      "step": 15700
    },
    {
      "epoch": 2.917859133989965,
      "grad_norm": 0.6924809813499451,
      "learning_rate": 3.4131347116254606e-05,
      "loss": 0.6653,
      "step": 15701
    },
    {
      "epoch": 2.9180449730533358,
      "grad_norm": 0.7547911405563354,
      "learning_rate": 3.412035549587599e-05,
      "loss": 0.7428,
      "step": 15702
    },
    {
      "epoch": 2.918230812116707,
      "grad_norm": 0.7436400055885315,
      "learning_rate": 3.4109365281585336e-05,
      "loss": 0.85,
      "step": 15703
    },
    {
      "epoch": 2.918416651180078,
      "grad_norm": 0.9104697108268738,
      "learning_rate": 3.4098376473617164e-05,
      "loss": 1.0624,
      "step": 15704
    },
    {
      "epoch": 2.9186024902434493,
      "grad_norm": 0.7633656859397888,
      "learning_rate": 3.40873890722061e-05,
      "loss": 0.7758,
      "step": 15705
    },
    {
      "epoch": 2.91878832930682,
      "grad_norm": 0.7920777201652527,
      "learning_rate": 3.4076403077586605e-05,
      "loss": 0.9387,
      "step": 15706
    },
    {
      "epoch": 2.9189741683701915,
      "grad_norm": 0.9203588366508484,
      "learning_rate": 3.406541848999312e-05,
      "loss": 0.9305,
      "step": 15707
    },
    {
      "epoch": 2.9191600074335624,
      "grad_norm": 0.7177546620368958,
      "learning_rate": 3.405443530966016e-05,
      "loss": 0.9136,
      "step": 15708
    },
    {
      "epoch": 2.9193458464969337,
      "grad_norm": 0.8345839977264404,
      "learning_rate": 3.404345353682208e-05,
      "loss": 1.0656,
      "step": 15709
    },
    {
      "epoch": 2.9195316855603046,
      "grad_norm": 0.755546510219574,
      "learning_rate": 3.403247317171333e-05,
      "loss": 0.8703,
      "step": 15710
    },
    {
      "epoch": 2.919717524623676,
      "grad_norm": 0.9971104264259338,
      "learning_rate": 3.402149421456823e-05,
      "loss": 0.9118,
      "step": 15711
    },
    {
      "epoch": 2.9199033636870473,
      "grad_norm": 0.8281334638595581,
      "learning_rate": 3.401051666562108e-05,
      "loss": 0.9542,
      "step": 15712
    },
    {
      "epoch": 2.920089202750418,
      "grad_norm": 0.8721912503242493,
      "learning_rate": 3.399954052510625e-05,
      "loss": 0.804,
      "step": 15713
    },
    {
      "epoch": 2.920275041813789,
      "grad_norm": 0.8998188972473145,
      "learning_rate": 3.398856579325796e-05,
      "loss": 1.0668,
      "step": 15714
    },
    {
      "epoch": 2.9204608808771604,
      "grad_norm": 0.6928837895393372,
      "learning_rate": 3.397759247031044e-05,
      "loss": 0.8659,
      "step": 15715
    },
    {
      "epoch": 2.9206467199405317,
      "grad_norm": 0.926973283290863,
      "learning_rate": 3.396662055649792e-05,
      "loss": 0.9504,
      "step": 15716
    },
    {
      "epoch": 2.9208325590039026,
      "grad_norm": 0.7991766929626465,
      "learning_rate": 3.395565005205458e-05,
      "loss": 1.0994,
      "step": 15717
    },
    {
      "epoch": 2.9210183980672735,
      "grad_norm": 0.8260103464126587,
      "learning_rate": 3.394468095721454e-05,
      "loss": 0.9118,
      "step": 15718
    },
    {
      "epoch": 2.921204237130645,
      "grad_norm": 0.8974000215530396,
      "learning_rate": 3.39337132722119e-05,
      "loss": 0.8529,
      "step": 15719
    },
    {
      "epoch": 2.921390076194016,
      "grad_norm": 0.8190984725952148,
      "learning_rate": 3.3922746997280775e-05,
      "loss": 0.9201,
      "step": 15720
    },
    {
      "epoch": 2.921575915257387,
      "grad_norm": 0.807974636554718,
      "learning_rate": 3.391178213265525e-05,
      "loss": 1.111,
      "step": 15721
    },
    {
      "epoch": 2.9217617543207584,
      "grad_norm": 0.7600648403167725,
      "learning_rate": 3.3900818678569344e-05,
      "loss": 0.8193,
      "step": 15722
    },
    {
      "epoch": 2.9219475933841292,
      "grad_norm": 0.8470858335494995,
      "learning_rate": 3.388985663525702e-05,
      "loss": 1.1329,
      "step": 15723
    },
    {
      "epoch": 2.9221334324475006,
      "grad_norm": 0.9734567403793335,
      "learning_rate": 3.387889600295222e-05,
      "loss": 0.9613,
      "step": 15724
    },
    {
      "epoch": 2.9223192715108715,
      "grad_norm": 0.8636355996131897,
      "learning_rate": 3.386793678188895e-05,
      "loss": 1.0745,
      "step": 15725
    },
    {
      "epoch": 2.922505110574243,
      "grad_norm": 0.7585934996604919,
      "learning_rate": 3.385697897230109e-05,
      "loss": 1.0608,
      "step": 15726
    },
    {
      "epoch": 2.9226909496376137,
      "grad_norm": 0.8040744066238403,
      "learning_rate": 3.3846022574422475e-05,
      "loss": 0.9901,
      "step": 15727
    },
    {
      "epoch": 2.922876788700985,
      "grad_norm": 0.7978103756904602,
      "learning_rate": 3.383506758848701e-05,
      "loss": 0.6268,
      "step": 15728
    },
    {
      "epoch": 2.923062627764356,
      "grad_norm": 0.9896499514579773,
      "learning_rate": 3.382411401472848e-05,
      "loss": 1.1853,
      "step": 15729
    },
    {
      "epoch": 2.923248466827727,
      "grad_norm": 0.8281686305999756,
      "learning_rate": 3.381316185338065e-05,
      "loss": 0.9363,
      "step": 15730
    },
    {
      "epoch": 2.9234343058910985,
      "grad_norm": 0.753558337688446,
      "learning_rate": 3.380221110467733e-05,
      "loss": 0.6371,
      "step": 15731
    },
    {
      "epoch": 2.9236201449544694,
      "grad_norm": 1.2504048347473145,
      "learning_rate": 3.3791261768852175e-05,
      "loss": 1.08,
      "step": 15732
    },
    {
      "epoch": 2.9238059840178403,
      "grad_norm": 0.9097568988800049,
      "learning_rate": 3.378031384613897e-05,
      "loss": 1.1595,
      "step": 15733
    },
    {
      "epoch": 2.9239918230812116,
      "grad_norm": 0.7802842855453491,
      "learning_rate": 3.3769367336771306e-05,
      "loss": 0.9175,
      "step": 15734
    },
    {
      "epoch": 2.924177662144583,
      "grad_norm": 0.9235023856163025,
      "learning_rate": 3.375842224098281e-05,
      "loss": 1.0529,
      "step": 15735
    },
    {
      "epoch": 2.924363501207954,
      "grad_norm": 0.7967664003372192,
      "learning_rate": 3.374747855900716e-05,
      "loss": 0.8076,
      "step": 15736
    },
    {
      "epoch": 2.924549340271325,
      "grad_norm": 0.8209742307662964,
      "learning_rate": 3.373653629107788e-05,
      "loss": 1.1509,
      "step": 15737
    },
    {
      "epoch": 2.924735179334696,
      "grad_norm": 0.7549864649772644,
      "learning_rate": 3.372559543742852e-05,
      "loss": 0.7769,
      "step": 15738
    },
    {
      "epoch": 2.9249210183980674,
      "grad_norm": 0.8510519862174988,
      "learning_rate": 3.371465599829257e-05,
      "loss": 0.8683,
      "step": 15739
    },
    {
      "epoch": 2.9251068574614383,
      "grad_norm": 0.7822895050048828,
      "learning_rate": 3.370371797390356e-05,
      "loss": 0.9195,
      "step": 15740
    },
    {
      "epoch": 2.9252926965248096,
      "grad_norm": 0.845230758190155,
      "learning_rate": 3.3692781364494884e-05,
      "loss": 1.0649,
      "step": 15741
    },
    {
      "epoch": 2.9254785355881805,
      "grad_norm": 0.8666579127311707,
      "learning_rate": 3.3681846170300046e-05,
      "loss": 0.9302,
      "step": 15742
    },
    {
      "epoch": 2.925664374651552,
      "grad_norm": 0.7660261988639832,
      "learning_rate": 3.367091239155239e-05,
      "loss": 0.8797,
      "step": 15743
    },
    {
      "epoch": 2.9258502137149227,
      "grad_norm": 0.8672879934310913,
      "learning_rate": 3.365998002848525e-05,
      "loss": 1.0502,
      "step": 15744
    },
    {
      "epoch": 2.926036052778294,
      "grad_norm": 0.9607042074203491,
      "learning_rate": 3.364904908133203e-05,
      "loss": 0.9121,
      "step": 15745
    },
    {
      "epoch": 2.9262218918416654,
      "grad_norm": 1.217624545097351,
      "learning_rate": 3.3638119550326e-05,
      "loss": 0.9755,
      "step": 15746
    },
    {
      "epoch": 2.9264077309050363,
      "grad_norm": 0.8371081948280334,
      "learning_rate": 3.3627191435700376e-05,
      "loss": 0.9684,
      "step": 15747
    },
    {
      "epoch": 2.926593569968407,
      "grad_norm": 0.6598268747329712,
      "learning_rate": 3.3616264737688495e-05,
      "loss": 0.7122,
      "step": 15748
    },
    {
      "epoch": 2.9267794090317785,
      "grad_norm": 0.7115182876586914,
      "learning_rate": 3.360533945652352e-05,
      "loss": 0.9374,
      "step": 15749
    },
    {
      "epoch": 2.92696524809515,
      "grad_norm": 0.9041739106178284,
      "learning_rate": 3.359441559243864e-05,
      "loss": 0.8403,
      "step": 15750
    },
    {
      "epoch": 2.9271510871585207,
      "grad_norm": 0.706041157245636,
      "learning_rate": 3.358349314566697e-05,
      "loss": 0.7009,
      "step": 15751
    },
    {
      "epoch": 2.9273369262218916,
      "grad_norm": 0.7268288135528564,
      "learning_rate": 3.357257211644166e-05,
      "loss": 0.985,
      "step": 15752
    },
    {
      "epoch": 2.927522765285263,
      "grad_norm": 0.7373469471931458,
      "learning_rate": 3.3561652504995835e-05,
      "loss": 0.8849,
      "step": 15753
    },
    {
      "epoch": 2.9277086043486342,
      "grad_norm": 0.7528462409973145,
      "learning_rate": 3.3550734311562527e-05,
      "loss": 0.928,
      "step": 15754
    },
    {
      "epoch": 2.927894443412005,
      "grad_norm": 0.6918595433235168,
      "learning_rate": 3.353981753637473e-05,
      "loss": 0.8758,
      "step": 15755
    },
    {
      "epoch": 2.9280802824753764,
      "grad_norm": 0.7775110602378845,
      "learning_rate": 3.352890217966551e-05,
      "loss": 1.1246,
      "step": 15756
    },
    {
      "epoch": 2.9282661215387473,
      "grad_norm": 0.6666045188903809,
      "learning_rate": 3.351798824166781e-05,
      "loss": 0.7464,
      "step": 15757
    },
    {
      "epoch": 2.9284519606021187,
      "grad_norm": 0.8970504403114319,
      "learning_rate": 3.3507075722614556e-05,
      "loss": 0.9118,
      "step": 15758
    },
    {
      "epoch": 2.9286377996654895,
      "grad_norm": 0.8148529529571533,
      "learning_rate": 3.349616462273864e-05,
      "loss": 0.8598,
      "step": 15759
    },
    {
      "epoch": 2.928823638728861,
      "grad_norm": 0.7585864663124084,
      "learning_rate": 3.348525494227299e-05,
      "loss": 0.6569,
      "step": 15760
    },
    {
      "epoch": 2.929009477792232,
      "grad_norm": 0.6061311960220337,
      "learning_rate": 3.347434668145043e-05,
      "loss": 0.6898,
      "step": 15761
    },
    {
      "epoch": 2.929195316855603,
      "grad_norm": 0.8349879384040833,
      "learning_rate": 3.3463439840503755e-05,
      "loss": 1.2111,
      "step": 15762
    },
    {
      "epoch": 2.929381155918974,
      "grad_norm": 0.8286142945289612,
      "learning_rate": 3.345253441966579e-05,
      "loss": 1.0443,
      "step": 15763
    },
    {
      "epoch": 2.9295669949823453,
      "grad_norm": 0.8316406011581421,
      "learning_rate": 3.3441630419169254e-05,
      "loss": 0.8959,
      "step": 15764
    },
    {
      "epoch": 2.9297528340457166,
      "grad_norm": 0.9590152502059937,
      "learning_rate": 3.343072783924693e-05,
      "loss": 0.9913,
      "step": 15765
    },
    {
      "epoch": 2.9299386731090875,
      "grad_norm": 0.8900766372680664,
      "learning_rate": 3.3419826680131494e-05,
      "loss": 0.8112,
      "step": 15766
    },
    {
      "epoch": 2.9301245121724584,
      "grad_norm": 0.7802647352218628,
      "learning_rate": 3.340892694205555e-05,
      "loss": 1.1222,
      "step": 15767
    },
    {
      "epoch": 2.9303103512358297,
      "grad_norm": 1.0942567586898804,
      "learning_rate": 3.339802862525183e-05,
      "loss": 0.8967,
      "step": 15768
    },
    {
      "epoch": 2.930496190299201,
      "grad_norm": 0.776422381401062,
      "learning_rate": 3.338713172995289e-05,
      "loss": 0.8029,
      "step": 15769
    },
    {
      "epoch": 2.930682029362572,
      "grad_norm": 0.8660294413566589,
      "learning_rate": 3.3376236256391316e-05,
      "loss": 0.8022,
      "step": 15770
    },
    {
      "epoch": 2.9308678684259433,
      "grad_norm": 0.7707787156105042,
      "learning_rate": 3.336534220479961e-05,
      "loss": 0.8491,
      "step": 15771
    },
    {
      "epoch": 2.931053707489314,
      "grad_norm": 0.8675243258476257,
      "learning_rate": 3.3354449575410326e-05,
      "loss": 0.7361,
      "step": 15772
    },
    {
      "epoch": 2.9312395465526855,
      "grad_norm": 0.6914379596710205,
      "learning_rate": 3.334355836845599e-05,
      "loss": 0.7739,
      "step": 15773
    },
    {
      "epoch": 2.9314253856160564,
      "grad_norm": 0.911323070526123,
      "learning_rate": 3.3332668584169004e-05,
      "loss": 1.0789,
      "step": 15774
    },
    {
      "epoch": 2.9316112246794277,
      "grad_norm": 0.7367120385169983,
      "learning_rate": 3.3321780222781776e-05,
      "loss": 0.8948,
      "step": 15775
    },
    {
      "epoch": 2.9317970637427986,
      "grad_norm": 1.1031914949417114,
      "learning_rate": 3.331089328452674e-05,
      "loss": 1.0156,
      "step": 15776
    },
    {
      "epoch": 2.93198290280617,
      "grad_norm": 0.7308126091957092,
      "learning_rate": 3.330000776963626e-05,
      "loss": 0.8076,
      "step": 15777
    },
    {
      "epoch": 2.932168741869541,
      "grad_norm": 1.0267692804336548,
      "learning_rate": 3.3289123678342645e-05,
      "loss": 0.8587,
      "step": 15778
    },
    {
      "epoch": 2.932354580932912,
      "grad_norm": 0.75275057554245,
      "learning_rate": 3.3278241010878175e-05,
      "loss": 0.8548,
      "step": 15779
    },
    {
      "epoch": 2.9325404199962835,
      "grad_norm": 0.9832603335380554,
      "learning_rate": 3.326735976747518e-05,
      "loss": 1.0699,
      "step": 15780
    },
    {
      "epoch": 2.9327262590596543,
      "grad_norm": 0.7555899024009705,
      "learning_rate": 3.325647994836588e-05,
      "loss": 0.8503,
      "step": 15781
    },
    {
      "epoch": 2.9329120981230252,
      "grad_norm": 1.0316232442855835,
      "learning_rate": 3.324560155378248e-05,
      "loss": 0.8765,
      "step": 15782
    },
    {
      "epoch": 2.9330979371863966,
      "grad_norm": 0.8232181668281555,
      "learning_rate": 3.323472458395712e-05,
      "loss": 0.6197,
      "step": 15783
    },
    {
      "epoch": 2.933283776249768,
      "grad_norm": 0.8313225507736206,
      "learning_rate": 3.322384903912199e-05,
      "loss": 1.0768,
      "step": 15784
    },
    {
      "epoch": 2.9334696153131388,
      "grad_norm": 0.7981607913970947,
      "learning_rate": 3.321297491950925e-05,
      "loss": 1.0084,
      "step": 15785
    },
    {
      "epoch": 2.93365545437651,
      "grad_norm": 0.8880338668823242,
      "learning_rate": 3.320210222535094e-05,
      "loss": 0.6865,
      "step": 15786
    },
    {
      "epoch": 2.933841293439881,
      "grad_norm": 0.8243213891983032,
      "learning_rate": 3.3191230956879104e-05,
      "loss": 0.7799,
      "step": 15787
    },
    {
      "epoch": 2.9340271325032523,
      "grad_norm": 0.7768036723136902,
      "learning_rate": 3.318036111432582e-05,
      "loss": 1.0689,
      "step": 15788
    },
    {
      "epoch": 2.934212971566623,
      "grad_norm": 0.9177983999252319,
      "learning_rate": 3.316949269792306e-05,
      "loss": 0.7647,
      "step": 15789
    },
    {
      "epoch": 2.9343988106299945,
      "grad_norm": 0.8567959070205688,
      "learning_rate": 3.3158625707902805e-05,
      "loss": 0.9281,
      "step": 15790
    },
    {
      "epoch": 2.9345846496933654,
      "grad_norm": 0.800028383731842,
      "learning_rate": 3.314776014449694e-05,
      "loss": 1.0502,
      "step": 15791
    },
    {
      "epoch": 2.9347704887567367,
      "grad_norm": 0.8020050525665283,
      "learning_rate": 3.3136896007937444e-05,
      "loss": 0.9888,
      "step": 15792
    },
    {
      "epoch": 2.9349563278201076,
      "grad_norm": 2.0786726474761963,
      "learning_rate": 3.312603329845613e-05,
      "loss": 1.2364,
      "step": 15793
    },
    {
      "epoch": 2.935142166883479,
      "grad_norm": 0.8580712676048279,
      "learning_rate": 3.31151720162849e-05,
      "loss": 0.7469,
      "step": 15794
    },
    {
      "epoch": 2.9353280059468503,
      "grad_norm": 0.8118255138397217,
      "learning_rate": 3.310431216165551e-05,
      "loss": 0.7951,
      "step": 15795
    },
    {
      "epoch": 2.935513845010221,
      "grad_norm": 0.7357614040374756,
      "learning_rate": 3.309345373479983e-05,
      "loss": 0.658,
      "step": 15796
    },
    {
      "epoch": 2.935699684073592,
      "grad_norm": 1.0954209566116333,
      "learning_rate": 3.308259673594955e-05,
      "loss": 0.971,
      "step": 15797
    },
    {
      "epoch": 2.9358855231369634,
      "grad_norm": 0.95086270570755,
      "learning_rate": 3.307174116533641e-05,
      "loss": 0.7394,
      "step": 15798
    },
    {
      "epoch": 2.9360713622003347,
      "grad_norm": 0.8866342902183533,
      "learning_rate": 3.306088702319208e-05,
      "loss": 0.9712,
      "step": 15799
    },
    {
      "epoch": 2.9362572012637056,
      "grad_norm": 0.9677961468696594,
      "learning_rate": 3.3050034309748256e-05,
      "loss": 0.6273,
      "step": 15800
    },
    {
      "epoch": 2.9364430403270765,
      "grad_norm": 0.9282647371292114,
      "learning_rate": 3.3039183025236575e-05,
      "loss": 1.0787,
      "step": 15801
    },
    {
      "epoch": 2.936628879390448,
      "grad_norm": 1.0624250173568726,
      "learning_rate": 3.302833316988861e-05,
      "loss": 1.1045,
      "step": 15802
    },
    {
      "epoch": 2.936814718453819,
      "grad_norm": 0.7853368520736694,
      "learning_rate": 3.301748474393592e-05,
      "loss": 0.8321,
      "step": 15803
    },
    {
      "epoch": 2.93700055751719,
      "grad_norm": 0.7858495116233826,
      "learning_rate": 3.300663774761008e-05,
      "loss": 0.9149,
      "step": 15804
    },
    {
      "epoch": 2.9371863965805614,
      "grad_norm": 0.831453800201416,
      "learning_rate": 3.2995792181142615e-05,
      "loss": 1.0512,
      "step": 15805
    },
    {
      "epoch": 2.9373722356439322,
      "grad_norm": 0.8120449781417847,
      "learning_rate": 3.298494804476499e-05,
      "loss": 1.2116,
      "step": 15806
    },
    {
      "epoch": 2.9375580747073036,
      "grad_norm": 1.215599775314331,
      "learning_rate": 3.2974105338708594e-05,
      "loss": 0.7631,
      "step": 15807
    },
    {
      "epoch": 2.9377439137706745,
      "grad_norm": 0.47633785009384155,
      "learning_rate": 3.2963264063204946e-05,
      "loss": 0.4033,
      "step": 15808
    },
    {
      "epoch": 2.937929752834046,
      "grad_norm": 0.9493878483772278,
      "learning_rate": 3.295242421848539e-05,
      "loss": 0.8563,
      "step": 15809
    },
    {
      "epoch": 2.938115591897417,
      "grad_norm": 0.8112297654151917,
      "learning_rate": 3.294158580478127e-05,
      "loss": 0.9625,
      "step": 15810
    },
    {
      "epoch": 2.938301430960788,
      "grad_norm": 0.7687706351280212,
      "learning_rate": 3.2930748822323896e-05,
      "loss": 0.9907,
      "step": 15811
    },
    {
      "epoch": 2.938487270024159,
      "grad_norm": 0.643135666847229,
      "learning_rate": 3.291991327134464e-05,
      "loss": 0.7125,
      "step": 15812
    },
    {
      "epoch": 2.93867310908753,
      "grad_norm": 0.8122084736824036,
      "learning_rate": 3.2909079152074705e-05,
      "loss": 0.9092,
      "step": 15813
    },
    {
      "epoch": 2.9388589481509015,
      "grad_norm": 0.84363853931427,
      "learning_rate": 3.289824646474531e-05,
      "loss": 0.892,
      "step": 15814
    },
    {
      "epoch": 2.9390447872142724,
      "grad_norm": 0.7943958044052124,
      "learning_rate": 3.28874152095877e-05,
      "loss": 0.9036,
      "step": 15815
    },
    {
      "epoch": 2.9392306262776433,
      "grad_norm": 0.8364081978797913,
      "learning_rate": 3.287658538683307e-05,
      "loss": 1.013,
      "step": 15816
    },
    {
      "epoch": 2.9394164653410146,
      "grad_norm": 0.9474763870239258,
      "learning_rate": 3.286575699671254e-05,
      "loss": 0.8919,
      "step": 15817
    },
    {
      "epoch": 2.939602304404386,
      "grad_norm": 0.8956844210624695,
      "learning_rate": 3.285493003945722e-05,
      "loss": 1.0745,
      "step": 15818
    },
    {
      "epoch": 2.939788143467757,
      "grad_norm": 0.8403872847557068,
      "learning_rate": 3.2844104515298155e-05,
      "loss": 0.8356,
      "step": 15819
    },
    {
      "epoch": 2.939973982531128,
      "grad_norm": 1.1269252300262451,
      "learning_rate": 3.2833280424466473e-05,
      "loss": 0.9814,
      "step": 15820
    },
    {
      "epoch": 2.940159821594499,
      "grad_norm": 0.8535260558128357,
      "learning_rate": 3.282245776719316e-05,
      "loss": 0.93,
      "step": 15821
    },
    {
      "epoch": 2.9403456606578704,
      "grad_norm": 0.8071843385696411,
      "learning_rate": 3.281163654370919e-05,
      "loss": 1.0905,
      "step": 15822
    },
    {
      "epoch": 2.9405314997212413,
      "grad_norm": 0.7468648552894592,
      "learning_rate": 3.280081675424552e-05,
      "loss": 0.7606,
      "step": 15823
    },
    {
      "epoch": 2.9407173387846126,
      "grad_norm": 0.7991985082626343,
      "learning_rate": 3.2789998399033126e-05,
      "loss": 0.816,
      "step": 15824
    },
    {
      "epoch": 2.9409031778479835,
      "grad_norm": 0.7739672660827637,
      "learning_rate": 3.277918147830284e-05,
      "loss": 1.0061,
      "step": 15825
    },
    {
      "epoch": 2.941089016911355,
      "grad_norm": 0.8021376729011536,
      "learning_rate": 3.276836599228562e-05,
      "loss": 1.2006,
      "step": 15826
    },
    {
      "epoch": 2.9412748559747257,
      "grad_norm": 0.8987312912940979,
      "learning_rate": 3.275755194121221e-05,
      "loss": 0.9632,
      "step": 15827
    },
    {
      "epoch": 2.941460695038097,
      "grad_norm": 0.8584547638893127,
      "learning_rate": 3.274673932531349e-05,
      "loss": 0.9651,
      "step": 15828
    },
    {
      "epoch": 2.9416465341014684,
      "grad_norm": 0.7346953749656677,
      "learning_rate": 3.2735928144820214e-05,
      "loss": 0.7253,
      "step": 15829
    },
    {
      "epoch": 2.9418323731648393,
      "grad_norm": 0.9900451898574829,
      "learning_rate": 3.2725118399963115e-05,
      "loss": 0.972,
      "step": 15830
    },
    {
      "epoch": 2.94201821222821,
      "grad_norm": 0.7965803742408752,
      "learning_rate": 3.27143100909729e-05,
      "loss": 0.7316,
      "step": 15831
    },
    {
      "epoch": 2.9422040512915815,
      "grad_norm": 0.9513272643089294,
      "learning_rate": 3.270350321808029e-05,
      "loss": 0.7201,
      "step": 15832
    },
    {
      "epoch": 2.942389890354953,
      "grad_norm": 1.0336984395980835,
      "learning_rate": 3.2692697781515916e-05,
      "loss": 1.0326,
      "step": 15833
    },
    {
      "epoch": 2.9425757294183237,
      "grad_norm": 0.9182641506195068,
      "learning_rate": 3.2681893781510406e-05,
      "loss": 0.8409,
      "step": 15834
    },
    {
      "epoch": 2.942761568481695,
      "grad_norm": 0.7116926312446594,
      "learning_rate": 3.2671091218294284e-05,
      "loss": 0.8546,
      "step": 15835
    },
    {
      "epoch": 2.942947407545066,
      "grad_norm": 0.7742061614990234,
      "learning_rate": 3.266029009209825e-05,
      "loss": 0.9241,
      "step": 15836
    },
    {
      "epoch": 2.9431332466084372,
      "grad_norm": 0.9186986088752747,
      "learning_rate": 3.264949040315276e-05,
      "loss": 1.1348,
      "step": 15837
    },
    {
      "epoch": 2.943319085671808,
      "grad_norm": 0.9015552401542664,
      "learning_rate": 3.263869215168831e-05,
      "loss": 0.9576,
      "step": 15838
    },
    {
      "epoch": 2.9435049247351794,
      "grad_norm": 0.9571332335472107,
      "learning_rate": 3.262789533793537e-05,
      "loss": 0.7288,
      "step": 15839
    },
    {
      "epoch": 2.9436907637985503,
      "grad_norm": 0.8588224649429321,
      "learning_rate": 3.2617099962124396e-05,
      "loss": 1.0207,
      "step": 15840
    },
    {
      "epoch": 2.9438766028619217,
      "grad_norm": 0.8299174904823303,
      "learning_rate": 3.26063060244858e-05,
      "loss": 0.7644,
      "step": 15841
    },
    {
      "epoch": 2.9440624419252925,
      "grad_norm": 0.7573933005332947,
      "learning_rate": 3.2595513525249954e-05,
      "loss": 0.9171,
      "step": 15842
    },
    {
      "epoch": 2.944248280988664,
      "grad_norm": 0.9382292628288269,
      "learning_rate": 3.258472246464717e-05,
      "loss": 0.828,
      "step": 15843
    },
    {
      "epoch": 2.944434120052035,
      "grad_norm": 0.9817286133766174,
      "learning_rate": 3.257393284290782e-05,
      "loss": 0.9116,
      "step": 15844
    },
    {
      "epoch": 2.944619959115406,
      "grad_norm": 0.8646917939186096,
      "learning_rate": 3.256314466026217e-05,
      "loss": 1.0139,
      "step": 15845
    },
    {
      "epoch": 2.944805798178777,
      "grad_norm": 0.7684429883956909,
      "learning_rate": 3.255235791694043e-05,
      "loss": 1.1085,
      "step": 15846
    },
    {
      "epoch": 2.9449916372421483,
      "grad_norm": 0.8010138273239136,
      "learning_rate": 3.254157261317288e-05,
      "loss": 0.99,
      "step": 15847
    },
    {
      "epoch": 2.9451774763055196,
      "grad_norm": 0.9764358401298523,
      "learning_rate": 3.253078874918972e-05,
      "loss": 1.0624,
      "step": 15848
    },
    {
      "epoch": 2.9453633153688905,
      "grad_norm": 0.8024470210075378,
      "learning_rate": 3.252000632522111e-05,
      "loss": 0.8637,
      "step": 15849
    },
    {
      "epoch": 2.9455491544322614,
      "grad_norm": 0.8670867085456848,
      "learning_rate": 3.2509225341497154e-05,
      "loss": 1.1131,
      "step": 15850
    },
    {
      "epoch": 2.9457349934956327,
      "grad_norm": 0.8533880710601807,
      "learning_rate": 3.2498445798247926e-05,
      "loss": 0.8523,
      "step": 15851
    },
    {
      "epoch": 2.945920832559004,
      "grad_norm": 0.9057621359825134,
      "learning_rate": 3.248766769570357e-05,
      "loss": 0.9767,
      "step": 15852
    },
    {
      "epoch": 2.946106671622375,
      "grad_norm": 0.8551714420318604,
      "learning_rate": 3.2476891034094095e-05,
      "loss": 1.0522,
      "step": 15853
    },
    {
      "epoch": 2.9462925106857463,
      "grad_norm": 1.0271611213684082,
      "learning_rate": 3.24661158136495e-05,
      "loss": 1.0574,
      "step": 15854
    },
    {
      "epoch": 2.946478349749117,
      "grad_norm": 0.8021514415740967,
      "learning_rate": 3.2455342034599746e-05,
      "loss": 0.9811,
      "step": 15855
    },
    {
      "epoch": 2.9466641888124885,
      "grad_norm": 0.8042370080947876,
      "learning_rate": 3.24445696971748e-05,
      "loss": 0.6625,
      "step": 15856
    },
    {
      "epoch": 2.9468500278758594,
      "grad_norm": 0.8870751857757568,
      "learning_rate": 3.243379880160462e-05,
      "loss": 1.0134,
      "step": 15857
    },
    {
      "epoch": 2.9470358669392307,
      "grad_norm": 0.9629341959953308,
      "learning_rate": 3.242302934811905e-05,
      "loss": 1.0757,
      "step": 15858
    },
    {
      "epoch": 2.9472217060026016,
      "grad_norm": 0.811947762966156,
      "learning_rate": 3.2412261336947926e-05,
      "loss": 0.765,
      "step": 15859
    },
    {
      "epoch": 2.947407545065973,
      "grad_norm": 0.964412271976471,
      "learning_rate": 3.2401494768321125e-05,
      "loss": 0.9061,
      "step": 15860
    },
    {
      "epoch": 2.947593384129344,
      "grad_norm": 0.8239457607269287,
      "learning_rate": 3.239072964246842e-05,
      "loss": 0.9501,
      "step": 15861
    },
    {
      "epoch": 2.947779223192715,
      "grad_norm": 0.7884500026702881,
      "learning_rate": 3.237996595961956e-05,
      "loss": 0.8559,
      "step": 15862
    },
    {
      "epoch": 2.9479650622560865,
      "grad_norm": 0.7907791137695312,
      "learning_rate": 3.236920372000426e-05,
      "loss": 1.0445,
      "step": 15863
    },
    {
      "epoch": 2.9481509013194573,
      "grad_norm": 0.9507573246955872,
      "learning_rate": 3.235844292385228e-05,
      "loss": 0.963,
      "step": 15864
    },
    {
      "epoch": 2.9483367403828282,
      "grad_norm": 0.9838759303092957,
      "learning_rate": 3.234768357139326e-05,
      "loss": 0.8269,
      "step": 15865
    },
    {
      "epoch": 2.9485225794461996,
      "grad_norm": 1.0180253982543945,
      "learning_rate": 3.233692566285681e-05,
      "loss": 0.9124,
      "step": 15866
    },
    {
      "epoch": 2.948708418509571,
      "grad_norm": 0.8494012355804443,
      "learning_rate": 3.2326169198472556e-05,
      "loss": 1.0001,
      "step": 15867
    },
    {
      "epoch": 2.9488942575729418,
      "grad_norm": 0.9028781652450562,
      "learning_rate": 3.2315414178470124e-05,
      "loss": 0.7656,
      "step": 15868
    },
    {
      "epoch": 2.949080096636313,
      "grad_norm": 0.8880323767662048,
      "learning_rate": 3.230466060307903e-05,
      "loss": 0.9977,
      "step": 15869
    },
    {
      "epoch": 2.949265935699684,
      "grad_norm": 0.9243189692497253,
      "learning_rate": 3.2293908472528776e-05,
      "loss": 0.9917,
      "step": 15870
    },
    {
      "epoch": 2.9494517747630553,
      "grad_norm": 0.9479065537452698,
      "learning_rate": 3.228315778704884e-05,
      "loss": 1.0309,
      "step": 15871
    },
    {
      "epoch": 2.949637613826426,
      "grad_norm": 0.7647424936294556,
      "learning_rate": 3.227240854686872e-05,
      "loss": 0.8696,
      "step": 15872
    },
    {
      "epoch": 2.9498234528897975,
      "grad_norm": 1.0016605854034424,
      "learning_rate": 3.22616607522178e-05,
      "loss": 1.1685,
      "step": 15873
    },
    {
      "epoch": 2.9500092919531684,
      "grad_norm": 0.6829125881195068,
      "learning_rate": 3.2250914403325503e-05,
      "loss": 0.7039,
      "step": 15874
    },
    {
      "epoch": 2.9501951310165397,
      "grad_norm": 0.7340703010559082,
      "learning_rate": 3.2240169500421135e-05,
      "loss": 0.9722,
      "step": 15875
    },
    {
      "epoch": 2.9503809700799106,
      "grad_norm": 0.796754002571106,
      "learning_rate": 3.222942604373411e-05,
      "loss": 0.982,
      "step": 15876
    },
    {
      "epoch": 2.950566809143282,
      "grad_norm": 0.7887791991233826,
      "learning_rate": 3.2218684033493655e-05,
      "loss": 0.9354,
      "step": 15877
    },
    {
      "epoch": 2.9507526482066533,
      "grad_norm": 0.9478221535682678,
      "learning_rate": 3.220794346992909e-05,
      "loss": 1.0203,
      "step": 15878
    },
    {
      "epoch": 2.950938487270024,
      "grad_norm": 0.8265958428382874,
      "learning_rate": 3.2197204353269614e-05,
      "loss": 0.9327,
      "step": 15879
    },
    {
      "epoch": 2.951124326333395,
      "grad_norm": 0.8037706613540649,
      "learning_rate": 3.2186466683744485e-05,
      "loss": 0.9537,
      "step": 15880
    },
    {
      "epoch": 2.9513101653967664,
      "grad_norm": 0.9107820987701416,
      "learning_rate": 3.2175730461582866e-05,
      "loss": 1.0413,
      "step": 15881
    },
    {
      "epoch": 2.9514960044601377,
      "grad_norm": 0.893482506275177,
      "learning_rate": 3.2164995687013876e-05,
      "loss": 0.8939,
      "step": 15882
    },
    {
      "epoch": 2.9516818435235086,
      "grad_norm": 0.7663919925689697,
      "learning_rate": 3.2154262360266607e-05,
      "loss": 0.8662,
      "step": 15883
    },
    {
      "epoch": 2.95186768258688,
      "grad_norm": 0.7317687273025513,
      "learning_rate": 3.2143530481570225e-05,
      "loss": 0.9311,
      "step": 15884
    },
    {
      "epoch": 2.952053521650251,
      "grad_norm": 0.8877572417259216,
      "learning_rate": 3.2132800051153734e-05,
      "loss": 0.9503,
      "step": 15885
    },
    {
      "epoch": 2.952239360713622,
      "grad_norm": 1.0320180654525757,
      "learning_rate": 3.212207106924613e-05,
      "loss": 1.1679,
      "step": 15886
    },
    {
      "epoch": 2.952425199776993,
      "grad_norm": 0.8952203989028931,
      "learning_rate": 3.2111343536076456e-05,
      "loss": 0.9192,
      "step": 15887
    },
    {
      "epoch": 2.9526110388403644,
      "grad_norm": 0.8200172185897827,
      "learning_rate": 3.210061745187363e-05,
      "loss": 0.8682,
      "step": 15888
    },
    {
      "epoch": 2.9527968779037352,
      "grad_norm": 0.8985316753387451,
      "learning_rate": 3.208989281686664e-05,
      "loss": 1.077,
      "step": 15889
    },
    {
      "epoch": 2.9529827169671066,
      "grad_norm": 0.8635002374649048,
      "learning_rate": 3.207916963128434e-05,
      "loss": 0.9769,
      "step": 15890
    },
    {
      "epoch": 2.9531685560304775,
      "grad_norm": 0.8763348460197449,
      "learning_rate": 3.206844789535558e-05,
      "loss": 1.0203,
      "step": 15891
    },
    {
      "epoch": 2.953354395093849,
      "grad_norm": 0.7320058345794678,
      "learning_rate": 3.2057727609309265e-05,
      "loss": 1.0473,
      "step": 15892
    },
    {
      "epoch": 2.95354023415722,
      "grad_norm": 0.7867861390113831,
      "learning_rate": 3.204700877337415e-05,
      "loss": 0.9808,
      "step": 15893
    },
    {
      "epoch": 2.953726073220591,
      "grad_norm": 0.9463543891906738,
      "learning_rate": 3.203629138777903e-05,
      "loss": 0.9315,
      "step": 15894
    },
    {
      "epoch": 2.953911912283962,
      "grad_norm": 1.0253965854644775,
      "learning_rate": 3.20255754527526e-05,
      "loss": 1.0547,
      "step": 15895
    },
    {
      "epoch": 2.954097751347333,
      "grad_norm": 0.8799330592155457,
      "learning_rate": 3.201486096852364e-05,
      "loss": 0.8729,
      "step": 15896
    },
    {
      "epoch": 2.9542835904107045,
      "grad_norm": 0.8454210758209229,
      "learning_rate": 3.2004147935320814e-05,
      "loss": 0.9261,
      "step": 15897
    },
    {
      "epoch": 2.9544694294740754,
      "grad_norm": 0.889337956905365,
      "learning_rate": 3.199343635337273e-05,
      "loss": 1.1,
      "step": 15898
    },
    {
      "epoch": 2.9546552685374463,
      "grad_norm": 0.8433546423912048,
      "learning_rate": 3.198272622290804e-05,
      "loss": 0.7996,
      "step": 15899
    },
    {
      "epoch": 2.9548411076008176,
      "grad_norm": 0.8294208645820618,
      "learning_rate": 3.1972017544155376e-05,
      "loss": 0.7894,
      "step": 15900
    },
    {
      "epoch": 2.955026946664189,
      "grad_norm": 0.7913797497749329,
      "learning_rate": 3.196131031734326e-05,
      "loss": 0.8688,
      "step": 15901
    },
    {
      "epoch": 2.95521278572756,
      "grad_norm": 0.9375088810920715,
      "learning_rate": 3.195060454270021e-05,
      "loss": 0.9925,
      "step": 15902
    },
    {
      "epoch": 2.955398624790931,
      "grad_norm": 0.7612003087997437,
      "learning_rate": 3.19399002204547e-05,
      "loss": 0.8621,
      "step": 15903
    },
    {
      "epoch": 2.955584463854302,
      "grad_norm": 0.7709554433822632,
      "learning_rate": 3.192919735083525e-05,
      "loss": 0.7884,
      "step": 15904
    },
    {
      "epoch": 2.9557703029176734,
      "grad_norm": 0.7612934708595276,
      "learning_rate": 3.191849593407027e-05,
      "loss": 0.7994,
      "step": 15905
    },
    {
      "epoch": 2.9559561419810443,
      "grad_norm": 1.7417844533920288,
      "learning_rate": 3.190779597038813e-05,
      "loss": 1.4456,
      "step": 15906
    },
    {
      "epoch": 2.9561419810444156,
      "grad_norm": 0.9167274832725525,
      "learning_rate": 3.1897097460017266e-05,
      "loss": 0.9625,
      "step": 15907
    },
    {
      "epoch": 2.9563278201077865,
      "grad_norm": 0.7483566999435425,
      "learning_rate": 3.188640040318599e-05,
      "loss": 1.0241,
      "step": 15908
    },
    {
      "epoch": 2.956513659171158,
      "grad_norm": 0.7958605289459229,
      "learning_rate": 3.187570480012258e-05,
      "loss": 0.8882,
      "step": 15909
    },
    {
      "epoch": 2.9566994982345287,
      "grad_norm": 0.84869784116745,
      "learning_rate": 3.186501065105536e-05,
      "loss": 0.7438,
      "step": 15910
    },
    {
      "epoch": 2.9568853372979,
      "grad_norm": 0.8469640612602234,
      "learning_rate": 3.185431795621255e-05,
      "loss": 0.8025,
      "step": 15911
    },
    {
      "epoch": 2.9570711763612714,
      "grad_norm": 0.8365806341171265,
      "learning_rate": 3.1843626715822406e-05,
      "loss": 0.9082,
      "step": 15912
    },
    {
      "epoch": 2.9572570154246423,
      "grad_norm": 0.8219962120056152,
      "learning_rate": 3.1832936930113086e-05,
      "loss": 0.7641,
      "step": 15913
    },
    {
      "epoch": 2.957442854488013,
      "grad_norm": 0.7901747822761536,
      "learning_rate": 3.1822248599312754e-05,
      "loss": 0.9285,
      "step": 15914
    },
    {
      "epoch": 2.9576286935513845,
      "grad_norm": 0.8131659626960754,
      "learning_rate": 3.18115617236495e-05,
      "loss": 0.6014,
      "step": 15915
    },
    {
      "epoch": 2.957814532614756,
      "grad_norm": 1.0069524049758911,
      "learning_rate": 3.180087630335147e-05,
      "loss": 0.9341,
      "step": 15916
    },
    {
      "epoch": 2.9580003716781267,
      "grad_norm": 0.9531908631324768,
      "learning_rate": 3.17901923386467e-05,
      "loss": 0.8465,
      "step": 15917
    },
    {
      "epoch": 2.958186210741498,
      "grad_norm": 0.7890156507492065,
      "learning_rate": 3.1779509829763186e-05,
      "loss": 0.9076,
      "step": 15918
    },
    {
      "epoch": 2.958372049804869,
      "grad_norm": 0.9298142194747925,
      "learning_rate": 3.176882877692897e-05,
      "loss": 0.9867,
      "step": 15919
    },
    {
      "epoch": 2.9585578888682402,
      "grad_norm": 0.8551667928695679,
      "learning_rate": 3.1758149180372045e-05,
      "loss": 0.9891,
      "step": 15920
    },
    {
      "epoch": 2.958743727931611,
      "grad_norm": 0.7418187260627747,
      "learning_rate": 3.174747104032032e-05,
      "loss": 1.0218,
      "step": 15921
    },
    {
      "epoch": 2.9589295669949824,
      "grad_norm": 0.7891413569450378,
      "learning_rate": 3.1736794357001695e-05,
      "loss": 1.0136,
      "step": 15922
    },
    {
      "epoch": 2.9591154060583533,
      "grad_norm": 0.9421678185462952,
      "learning_rate": 3.172611913064402e-05,
      "loss": 1.0678,
      "step": 15923
    },
    {
      "epoch": 2.9593012451217247,
      "grad_norm": 1.0176401138305664,
      "learning_rate": 3.171544536147521e-05,
      "loss": 0.9857,
      "step": 15924
    },
    {
      "epoch": 2.9594870841850955,
      "grad_norm": 0.8667479753494263,
      "learning_rate": 3.1704773049723024e-05,
      "loss": 0.8421,
      "step": 15925
    },
    {
      "epoch": 2.959672923248467,
      "grad_norm": 0.8144848346710205,
      "learning_rate": 3.169410219561523e-05,
      "loss": 0.9336,
      "step": 15926
    },
    {
      "epoch": 2.959858762311838,
      "grad_norm": 0.7537040114402771,
      "learning_rate": 3.168343279937964e-05,
      "loss": 0.8843,
      "step": 15927
    },
    {
      "epoch": 2.960044601375209,
      "grad_norm": 0.880525529384613,
      "learning_rate": 3.1672764861243945e-05,
      "loss": 1.0492,
      "step": 15928
    },
    {
      "epoch": 2.96023044043858,
      "grad_norm": 0.9077096581459045,
      "learning_rate": 3.166209838143582e-05,
      "loss": 0.968,
      "step": 15929
    },
    {
      "epoch": 2.9604162795019513,
      "grad_norm": 0.8353539109230042,
      "learning_rate": 3.1651433360182916e-05,
      "loss": 0.8478,
      "step": 15930
    },
    {
      "epoch": 2.9606021185653226,
      "grad_norm": 0.8167479038238525,
      "learning_rate": 3.164076979771287e-05,
      "loss": 0.9171,
      "step": 15931
    },
    {
      "epoch": 2.9607879576286935,
      "grad_norm": 0.6608850359916687,
      "learning_rate": 3.163010769425331e-05,
      "loss": 0.7729,
      "step": 15932
    },
    {
      "epoch": 2.9609737966920644,
      "grad_norm": 0.8036105036735535,
      "learning_rate": 3.161944705003177e-05,
      "loss": 0.9901,
      "step": 15933
    },
    {
      "epoch": 2.9611596357554357,
      "grad_norm": 0.9866622090339661,
      "learning_rate": 3.160878786527579e-05,
      "loss": 1.1158,
      "step": 15934
    },
    {
      "epoch": 2.961345474818807,
      "grad_norm": 0.8332897424697876,
      "learning_rate": 3.159813014021285e-05,
      "loss": 0.9942,
      "step": 15935
    },
    {
      "epoch": 2.961531313882178,
      "grad_norm": 0.840764045715332,
      "learning_rate": 3.1587473875070464e-05,
      "loss": 1.075,
      "step": 15936
    },
    {
      "epoch": 2.9617171529455493,
      "grad_norm": 0.9720584750175476,
      "learning_rate": 3.1576819070076036e-05,
      "loss": 0.9517,
      "step": 15937
    },
    {
      "epoch": 2.96190299200892,
      "grad_norm": 0.7495701909065247,
      "learning_rate": 3.156616572545697e-05,
      "loss": 0.9145,
      "step": 15938
    },
    {
      "epoch": 2.9620888310722915,
      "grad_norm": 0.8397412896156311,
      "learning_rate": 3.155551384144069e-05,
      "loss": 0.9287,
      "step": 15939
    },
    {
      "epoch": 2.9622746701356624,
      "grad_norm": 0.7724320888519287,
      "learning_rate": 3.154486341825446e-05,
      "loss": 0.9588,
      "step": 15940
    },
    {
      "epoch": 2.9624605091990337,
      "grad_norm": 0.956950306892395,
      "learning_rate": 3.1534214456125696e-05,
      "loss": 0.7432,
      "step": 15941
    },
    {
      "epoch": 2.962646348262405,
      "grad_norm": 1.2833693027496338,
      "learning_rate": 3.1523566955281635e-05,
      "loss": 1.2552,
      "step": 15942
    },
    {
      "epoch": 2.962832187325776,
      "grad_norm": 1.0043977499008179,
      "learning_rate": 3.1512920915949495e-05,
      "loss": 0.8906,
      "step": 15943
    },
    {
      "epoch": 2.963018026389147,
      "grad_norm": 0.7243955731391907,
      "learning_rate": 3.150227633835655e-05,
      "loss": 0.7975,
      "step": 15944
    },
    {
      "epoch": 2.963203865452518,
      "grad_norm": 0.6837208271026611,
      "learning_rate": 3.1491633222729975e-05,
      "loss": 0.7359,
      "step": 15945
    },
    {
      "epoch": 2.9633897045158895,
      "grad_norm": 0.7101500630378723,
      "learning_rate": 3.148099156929689e-05,
      "loss": 0.8906,
      "step": 15946
    },
    {
      "epoch": 2.9635755435792603,
      "grad_norm": 0.9120703339576721,
      "learning_rate": 3.14703513782845e-05,
      "loss": 0.8411,
      "step": 15947
    },
    {
      "epoch": 2.9637613826426312,
      "grad_norm": 0.8641577363014221,
      "learning_rate": 3.145971264991985e-05,
      "loss": 0.9366,
      "step": 15948
    },
    {
      "epoch": 2.9639472217060026,
      "grad_norm": 0.7627655863761902,
      "learning_rate": 3.144907538443002e-05,
      "loss": 0.6819,
      "step": 15949
    },
    {
      "epoch": 2.964133060769374,
      "grad_norm": 0.7648568749427795,
      "learning_rate": 3.143843958204199e-05,
      "loss": 1.1172,
      "step": 15950
    },
    {
      "epoch": 2.9643188998327448,
      "grad_norm": 0.8323633074760437,
      "learning_rate": 3.1427805242982824e-05,
      "loss": 1.0325,
      "step": 15951
    },
    {
      "epoch": 2.964504738896116,
      "grad_norm": 0.7272006273269653,
      "learning_rate": 3.141717236747951e-05,
      "loss": 0.7491,
      "step": 15952
    },
    {
      "epoch": 2.964690577959487,
      "grad_norm": 0.8330870270729065,
      "learning_rate": 3.140654095575896e-05,
      "loss": 0.681,
      "step": 15953
    },
    {
      "epoch": 2.9648764170228583,
      "grad_norm": 0.899566650390625,
      "learning_rate": 3.139591100804807e-05,
      "loss": 1.084,
      "step": 15954
    },
    {
      "epoch": 2.965062256086229,
      "grad_norm": 0.7420375943183899,
      "learning_rate": 3.1385282524573715e-05,
      "loss": 0.673,
      "step": 15955
    },
    {
      "epoch": 2.9652480951496005,
      "grad_norm": 0.890329897403717,
      "learning_rate": 3.1374655505562775e-05,
      "loss": 1.0518,
      "step": 15956
    },
    {
      "epoch": 2.9654339342129714,
      "grad_norm": 0.6901466250419617,
      "learning_rate": 3.1364029951242055e-05,
      "loss": 1.1218,
      "step": 15957
    },
    {
      "epoch": 2.9656197732763427,
      "grad_norm": 0.8333992958068848,
      "learning_rate": 3.13534058618383e-05,
      "loss": 0.9538,
      "step": 15958
    },
    {
      "epoch": 2.9658056123397136,
      "grad_norm": 0.8375001549720764,
      "learning_rate": 3.134278323757833e-05,
      "loss": 1.0051,
      "step": 15959
    },
    {
      "epoch": 2.965991451403085,
      "grad_norm": 0.8724488019943237,
      "learning_rate": 3.133216207868883e-05,
      "loss": 0.8164,
      "step": 15960
    },
    {
      "epoch": 2.9661772904664563,
      "grad_norm": 0.7335485219955444,
      "learning_rate": 3.132154238539645e-05,
      "loss": 0.9836,
      "step": 15961
    },
    {
      "epoch": 2.966363129529827,
      "grad_norm": 0.957584798336029,
      "learning_rate": 3.1310924157927935e-05,
      "loss": 0.996,
      "step": 15962
    },
    {
      "epoch": 2.966548968593198,
      "grad_norm": 1.464234471321106,
      "learning_rate": 3.130030739650983e-05,
      "loss": 1.5019,
      "step": 15963
    },
    {
      "epoch": 2.9667348076565694,
      "grad_norm": 0.7681525945663452,
      "learning_rate": 3.128969210136881e-05,
      "loss": 0.7493,
      "step": 15964
    },
    {
      "epoch": 2.9669206467199407,
      "grad_norm": 0.7735837697982788,
      "learning_rate": 3.1279078272731385e-05,
      "loss": 0.6888,
      "step": 15965
    },
    {
      "epoch": 2.9671064857833116,
      "grad_norm": 0.7694737911224365,
      "learning_rate": 3.126846591082409e-05,
      "loss": 0.8867,
      "step": 15966
    },
    {
      "epoch": 2.967292324846683,
      "grad_norm": 0.7423407435417175,
      "learning_rate": 3.125785501587346e-05,
      "loss": 0.7762,
      "step": 15967
    },
    {
      "epoch": 2.967478163910054,
      "grad_norm": 0.734919011592865,
      "learning_rate": 3.124724558810596e-05,
      "loss": 0.637,
      "step": 15968
    },
    {
      "epoch": 2.967664002973425,
      "grad_norm": 0.9352990984916687,
      "learning_rate": 3.1236637627748e-05,
      "loss": 0.8245,
      "step": 15969
    },
    {
      "epoch": 2.967849842036796,
      "grad_norm": 0.7590056657791138,
      "learning_rate": 3.122603113502599e-05,
      "loss": 0.9774,
      "step": 15970
    },
    {
      "epoch": 2.9680356811001674,
      "grad_norm": 0.817415177822113,
      "learning_rate": 3.1215426110166356e-05,
      "loss": 0.9022,
      "step": 15971
    },
    {
      "epoch": 2.9682215201635382,
      "grad_norm": 1.0079842805862427,
      "learning_rate": 3.120482255339537e-05,
      "loss": 1.0907,
      "step": 15972
    },
    {
      "epoch": 2.9684073592269096,
      "grad_norm": 0.8362964987754822,
      "learning_rate": 3.1194220464939436e-05,
      "loss": 0.6905,
      "step": 15973
    },
    {
      "epoch": 2.9685931982902805,
      "grad_norm": 0.7991756200790405,
      "learning_rate": 3.118361984502479e-05,
      "loss": 0.8582,
      "step": 15974
    },
    {
      "epoch": 2.968779037353652,
      "grad_norm": 0.8549006581306458,
      "learning_rate": 3.117302069387764e-05,
      "loss": 1.0008,
      "step": 15975
    },
    {
      "epoch": 2.968964876417023,
      "grad_norm": 0.7779741883277893,
      "learning_rate": 3.116242301172429e-05,
      "loss": 0.76,
      "step": 15976
    },
    {
      "epoch": 2.969150715480394,
      "grad_norm": 1.0254641771316528,
      "learning_rate": 3.11518267987909e-05,
      "loss": 1.0068,
      "step": 15977
    },
    {
      "epoch": 2.969336554543765,
      "grad_norm": 0.8137785196304321,
      "learning_rate": 3.114123205530358e-05,
      "loss": 0.7319,
      "step": 15978
    },
    {
      "epoch": 2.969522393607136,
      "grad_norm": 0.7890313267707825,
      "learning_rate": 3.113063878148852e-05,
      "loss": 0.7448,
      "step": 15979
    },
    {
      "epoch": 2.9697082326705075,
      "grad_norm": 0.862103283405304,
      "learning_rate": 3.11200469775718e-05,
      "loss": 0.8145,
      "step": 15980
    },
    {
      "epoch": 2.9698940717338784,
      "grad_norm": 0.7743937969207764,
      "learning_rate": 3.110945664377947e-05,
      "loss": 0.8747,
      "step": 15981
    },
    {
      "epoch": 2.9700799107972493,
      "grad_norm": 0.800082266330719,
      "learning_rate": 3.109886778033754e-05,
      "loss": 0.8537,
      "step": 15982
    },
    {
      "epoch": 2.9702657498606206,
      "grad_norm": 0.798474133014679,
      "learning_rate": 3.1088280387472034e-05,
      "loss": 0.9413,
      "step": 15983
    },
    {
      "epoch": 2.970451588923992,
      "grad_norm": 0.7497609853744507,
      "learning_rate": 3.107769446540896e-05,
      "loss": 1.0148,
      "step": 15984
    },
    {
      "epoch": 2.970637427987363,
      "grad_norm": 0.9664625525474548,
      "learning_rate": 3.106711001437421e-05,
      "loss": 0.9903,
      "step": 15985
    },
    {
      "epoch": 2.970823267050734,
      "grad_norm": 1.123875617980957,
      "learning_rate": 3.105652703459368e-05,
      "loss": 0.7681,
      "step": 15986
    },
    {
      "epoch": 2.971009106114105,
      "grad_norm": 3.0108494758605957,
      "learning_rate": 3.104594552629331e-05,
      "loss": 1.2648,
      "step": 15987
    },
    {
      "epoch": 2.9711949451774764,
      "grad_norm": 1.2039192914962769,
      "learning_rate": 3.10353654896989e-05,
      "loss": 1.2072,
      "step": 15988
    },
    {
      "epoch": 2.9713807842408473,
      "grad_norm": 0.818681538105011,
      "learning_rate": 3.1024786925036255e-05,
      "loss": 0.9656,
      "step": 15989
    },
    {
      "epoch": 2.9715666233042186,
      "grad_norm": 0.7975749373435974,
      "learning_rate": 3.101420983253114e-05,
      "loss": 0.9338,
      "step": 15990
    },
    {
      "epoch": 2.97175246236759,
      "grad_norm": 0.8213855028152466,
      "learning_rate": 3.100363421240936e-05,
      "loss": 0.8867,
      "step": 15991
    },
    {
      "epoch": 2.971938301430961,
      "grad_norm": 0.9206775426864624,
      "learning_rate": 3.099306006489662e-05,
      "loss": 0.9823,
      "step": 15992
    },
    {
      "epoch": 2.9721241404943317,
      "grad_norm": 0.6859269738197327,
      "learning_rate": 3.098248739021854e-05,
      "loss": 0.6746,
      "step": 15993
    },
    {
      "epoch": 2.972309979557703,
      "grad_norm": 0.7936283349990845,
      "learning_rate": 3.097191618860087e-05,
      "loss": 0.9326,
      "step": 15994
    },
    {
      "epoch": 2.9724958186210744,
      "grad_norm": 0.792716383934021,
      "learning_rate": 3.096134646026917e-05,
      "loss": 0.7887,
      "step": 15995
    },
    {
      "epoch": 2.9726816576844453,
      "grad_norm": 0.9117279648780823,
      "learning_rate": 3.095077820544907e-05,
      "loss": 1.1533,
      "step": 15996
    },
    {
      "epoch": 2.972867496747816,
      "grad_norm": 0.7501056790351868,
      "learning_rate": 3.0940211424366125e-05,
      "loss": 0.6921,
      "step": 15997
    },
    {
      "epoch": 2.9730533358111875,
      "grad_norm": 0.9197656512260437,
      "learning_rate": 3.0929646117245815e-05,
      "loss": 1.0386,
      "step": 15998
    },
    {
      "epoch": 2.973239174874559,
      "grad_norm": 0.8189886212348938,
      "learning_rate": 3.091908228431373e-05,
      "loss": 0.8909,
      "step": 15999
    },
    {
      "epoch": 2.9734250139379297,
      "grad_norm": 0.8167677521705627,
      "learning_rate": 3.0908519925795264e-05,
      "loss": 0.9702,
      "step": 16000
    },
    {
      "epoch": 2.973610853001301,
      "grad_norm": 0.8350397944450378,
      "learning_rate": 3.089795904191588e-05,
      "loss": 0.8225,
      "step": 16001
    },
    {
      "epoch": 2.973796692064672,
      "grad_norm": 0.7315388321876526,
      "learning_rate": 3.088739963290095e-05,
      "loss": 0.8987,
      "step": 16002
    },
    {
      "epoch": 2.9739825311280432,
      "grad_norm": 0.7389110922813416,
      "learning_rate": 3.087684169897588e-05,
      "loss": 0.8792,
      "step": 16003
    },
    {
      "epoch": 2.974168370191414,
      "grad_norm": 0.8871979117393494,
      "learning_rate": 3.086628524036602e-05,
      "loss": 0.7208,
      "step": 16004
    },
    {
      "epoch": 2.9743542092547854,
      "grad_norm": 0.7603043913841248,
      "learning_rate": 3.0855730257296665e-05,
      "loss": 0.8907,
      "step": 16005
    },
    {
      "epoch": 2.9745400483181563,
      "grad_norm": 0.9048733711242676,
      "learning_rate": 3.0845176749993073e-05,
      "loss": 1.0397,
      "step": 16006
    },
    {
      "epoch": 2.9747258873815277,
      "grad_norm": 0.9382114410400391,
      "learning_rate": 3.0834624718680524e-05,
      "loss": 0.9248,
      "step": 16007
    },
    {
      "epoch": 2.9749117264448985,
      "grad_norm": 0.9046305418014526,
      "learning_rate": 3.082407416358424e-05,
      "loss": 0.9245,
      "step": 16008
    },
    {
      "epoch": 2.97509756550827,
      "grad_norm": 0.8209074139595032,
      "learning_rate": 3.0813525084929365e-05,
      "loss": 0.9154,
      "step": 16009
    },
    {
      "epoch": 2.975283404571641,
      "grad_norm": 0.959549605846405,
      "learning_rate": 3.080297748294104e-05,
      "loss": 0.861,
      "step": 16010
    },
    {
      "epoch": 2.975469243635012,
      "grad_norm": 1.0113213062286377,
      "learning_rate": 3.079243135784444e-05,
      "loss": 1.1289,
      "step": 16011
    },
    {
      "epoch": 2.975655082698383,
      "grad_norm": 0.8068660497665405,
      "learning_rate": 3.078188670986464e-05,
      "loss": 0.804,
      "step": 16012
    },
    {
      "epoch": 2.9758409217617543,
      "grad_norm": 0.7963927388191223,
      "learning_rate": 3.077134353922667e-05,
      "loss": 0.7001,
      "step": 16013
    },
    {
      "epoch": 2.9760267608251256,
      "grad_norm": 0.8770447373390198,
      "learning_rate": 3.076080184615552e-05,
      "loss": 1.0307,
      "step": 16014
    },
    {
      "epoch": 2.9762125998884965,
      "grad_norm": 0.8757564425468445,
      "learning_rate": 3.07502616308763e-05,
      "loss": 1.2144,
      "step": 16015
    },
    {
      "epoch": 2.976398438951868,
      "grad_norm": 0.8343798518180847,
      "learning_rate": 3.07397228936139e-05,
      "loss": 0.9809,
      "step": 16016
    },
    {
      "epoch": 2.9765842780152387,
      "grad_norm": 0.7926608920097351,
      "learning_rate": 3.072918563459326e-05,
      "loss": 1.0375,
      "step": 16017
    },
    {
      "epoch": 2.97677011707861,
      "grad_norm": 0.8190401792526245,
      "learning_rate": 3.071864985403925e-05,
      "loss": 0.9349,
      "step": 16018
    },
    {
      "epoch": 2.976955956141981,
      "grad_norm": 1.1282371282577515,
      "learning_rate": 3.07081155521768e-05,
      "loss": 1.0622,
      "step": 16019
    },
    {
      "epoch": 2.9771417952053523,
      "grad_norm": 0.8359345197677612,
      "learning_rate": 3.0697582729230715e-05,
      "loss": 0.9606,
      "step": 16020
    },
    {
      "epoch": 2.977327634268723,
      "grad_norm": 0.87877357006073,
      "learning_rate": 3.0687051385425794e-05,
      "loss": 0.7875,
      "step": 16021
    },
    {
      "epoch": 2.9775134733320945,
      "grad_norm": 0.8610250949859619,
      "learning_rate": 3.067652152098679e-05,
      "loss": 0.8837,
      "step": 16022
    },
    {
      "epoch": 2.9776993123954654,
      "grad_norm": 0.9947982430458069,
      "learning_rate": 3.066599313613849e-05,
      "loss": 0.9258,
      "step": 16023
    },
    {
      "epoch": 2.9778851514588367,
      "grad_norm": 0.836203932762146,
      "learning_rate": 3.0655466231105566e-05,
      "loss": 0.9683,
      "step": 16024
    },
    {
      "epoch": 2.978070990522208,
      "grad_norm": 0.9147794842720032,
      "learning_rate": 3.064494080611274e-05,
      "loss": 0.8862,
      "step": 16025
    },
    {
      "epoch": 2.978256829585579,
      "grad_norm": 0.9989723563194275,
      "learning_rate": 3.063441686138461e-05,
      "loss": 1.014,
      "step": 16026
    },
    {
      "epoch": 2.97844266864895,
      "grad_norm": 0.7151840925216675,
      "learning_rate": 3.062389439714584e-05,
      "loss": 0.7912,
      "step": 16027
    },
    {
      "epoch": 2.978628507712321,
      "grad_norm": 0.8194684982299805,
      "learning_rate": 3.0613373413620994e-05,
      "loss": 0.9364,
      "step": 16028
    },
    {
      "epoch": 2.9788143467756925,
      "grad_norm": 1.0978440046310425,
      "learning_rate": 3.0602853911034624e-05,
      "loss": 1.0119,
      "step": 16029
    },
    {
      "epoch": 2.9790001858390633,
      "grad_norm": 0.9319393634796143,
      "learning_rate": 3.0592335889611204e-05,
      "loss": 0.8991,
      "step": 16030
    },
    {
      "epoch": 2.9791860249024342,
      "grad_norm": 0.8846945762634277,
      "learning_rate": 3.05818193495753e-05,
      "loss": 0.9444,
      "step": 16031
    },
    {
      "epoch": 2.9793718639658056,
      "grad_norm": 0.844276487827301,
      "learning_rate": 3.057130429115135e-05,
      "loss": 0.9255,
      "step": 16032
    },
    {
      "epoch": 2.979557703029177,
      "grad_norm": 0.8965041041374207,
      "learning_rate": 3.056079071456375e-05,
      "loss": 0.9613,
      "step": 16033
    },
    {
      "epoch": 2.9797435420925478,
      "grad_norm": 0.8440106511116028,
      "learning_rate": 3.055027862003689e-05,
      "loss": 0.9324,
      "step": 16034
    },
    {
      "epoch": 2.979929381155919,
      "grad_norm": 0.8329331874847412,
      "learning_rate": 3.053976800779513e-05,
      "loss": 0.9108,
      "step": 16035
    },
    {
      "epoch": 2.98011522021929,
      "grad_norm": 0.8621357083320618,
      "learning_rate": 3.0529258878062886e-05,
      "loss": 0.7843,
      "step": 16036
    },
    {
      "epoch": 2.9803010592826613,
      "grad_norm": 1.0645602941513062,
      "learning_rate": 3.051875123106437e-05,
      "loss": 1.3506,
      "step": 16037
    },
    {
      "epoch": 2.980486898346032,
      "grad_norm": 0.845532238483429,
      "learning_rate": 3.0508245067023854e-05,
      "loss": 0.883,
      "step": 16038
    },
    {
      "epoch": 2.9806727374094035,
      "grad_norm": 0.8710725903511047,
      "learning_rate": 3.0497740386165618e-05,
      "loss": 0.9955,
      "step": 16039
    },
    {
      "epoch": 2.9808585764727744,
      "grad_norm": 0.9422255754470825,
      "learning_rate": 3.0487237188713858e-05,
      "loss": 1.0953,
      "step": 16040
    },
    {
      "epoch": 2.9810444155361457,
      "grad_norm": 0.7569517493247986,
      "learning_rate": 3.0476735474892713e-05,
      "loss": 0.8029,
      "step": 16041
    },
    {
      "epoch": 2.9812302545995166,
      "grad_norm": 0.7387254238128662,
      "learning_rate": 3.046623524492632e-05,
      "loss": 0.6981,
      "step": 16042
    },
    {
      "epoch": 2.981416093662888,
      "grad_norm": 0.8482987880706787,
      "learning_rate": 3.0455736499038845e-05,
      "loss": 0.9245,
      "step": 16043
    },
    {
      "epoch": 2.9816019327262593,
      "grad_norm": 0.8994969129562378,
      "learning_rate": 3.0445239237454338e-05,
      "loss": 0.921,
      "step": 16044
    },
    {
      "epoch": 2.98178777178963,
      "grad_norm": 0.6694866418838501,
      "learning_rate": 3.0434743460396796e-05,
      "loss": 0.7867,
      "step": 16045
    },
    {
      "epoch": 2.981973610853001,
      "grad_norm": 1.0706828832626343,
      "learning_rate": 3.042424916809028e-05,
      "loss": 0.763,
      "step": 16046
    },
    {
      "epoch": 2.9821594499163724,
      "grad_norm": 0.8097942471504211,
      "learning_rate": 3.0413756360758793e-05,
      "loss": 0.8001,
      "step": 16047
    },
    {
      "epoch": 2.9823452889797437,
      "grad_norm": 0.9327067732810974,
      "learning_rate": 3.0403265038626273e-05,
      "loss": 1.0208,
      "step": 16048
    },
    {
      "epoch": 2.9825311280431146,
      "grad_norm": 0.7851989269256592,
      "learning_rate": 3.0392775201916625e-05,
      "loss": 1.0449,
      "step": 16049
    },
    {
      "epoch": 2.982716967106486,
      "grad_norm": 0.7200372219085693,
      "learning_rate": 3.0382286850853703e-05,
      "loss": 1.0121,
      "step": 16050
    },
    {
      "epoch": 2.982902806169857,
      "grad_norm": 1.2710801362991333,
      "learning_rate": 3.0371799985661433e-05,
      "loss": 0.9335,
      "step": 16051
    },
    {
      "epoch": 2.983088645233228,
      "grad_norm": 0.8790804147720337,
      "learning_rate": 3.0361314606563606e-05,
      "loss": 0.901,
      "step": 16052
    },
    {
      "epoch": 2.983274484296599,
      "grad_norm": 0.9105607867240906,
      "learning_rate": 3.035083071378402e-05,
      "loss": 0.9344,
      "step": 16053
    },
    {
      "epoch": 2.9834603233599704,
      "grad_norm": 0.7116270661354065,
      "learning_rate": 3.0340348307546384e-05,
      "loss": 0.6129,
      "step": 16054
    },
    {
      "epoch": 2.9836461624233412,
      "grad_norm": 0.8581184148788452,
      "learning_rate": 3.0329867388074517e-05,
      "loss": 1.1956,
      "step": 16055
    },
    {
      "epoch": 2.9838320014867126,
      "grad_norm": 0.8559527397155762,
      "learning_rate": 3.0319387955592038e-05,
      "loss": 1.0101,
      "step": 16056
    },
    {
      "epoch": 2.9840178405500835,
      "grad_norm": 1.0393134355545044,
      "learning_rate": 3.0308910010322666e-05,
      "loss": 0.8381,
      "step": 16057
    },
    {
      "epoch": 2.984203679613455,
      "grad_norm": 0.76030033826828,
      "learning_rate": 3.029843355249e-05,
      "loss": 0.9155,
      "step": 16058
    },
    {
      "epoch": 2.984389518676826,
      "grad_norm": 0.7992974519729614,
      "learning_rate": 3.0287958582317676e-05,
      "loss": 1.0005,
      "step": 16059
    },
    {
      "epoch": 2.984575357740197,
      "grad_norm": 0.9479563236236572,
      "learning_rate": 3.0277485100029256e-05,
      "loss": 0.9097,
      "step": 16060
    },
    {
      "epoch": 2.984761196803568,
      "grad_norm": 1.0739942789077759,
      "learning_rate": 3.026701310584826e-05,
      "loss": 0.9447,
      "step": 16061
    },
    {
      "epoch": 2.984947035866939,
      "grad_norm": 0.7115394473075867,
      "learning_rate": 3.0256542599998172e-05,
      "loss": 0.7588,
      "step": 16062
    },
    {
      "epoch": 2.9851328749303105,
      "grad_norm": 0.9326103329658508,
      "learning_rate": 3.0246073582702538e-05,
      "loss": 0.8896,
      "step": 16063
    },
    {
      "epoch": 2.9853187139936814,
      "grad_norm": 0.9557722210884094,
      "learning_rate": 3.023560605418475e-05,
      "loss": 0.9496,
      "step": 16064
    },
    {
      "epoch": 2.9855045530570528,
      "grad_norm": 0.9472669363021851,
      "learning_rate": 3.022514001466824e-05,
      "loss": 0.8392,
      "step": 16065
    },
    {
      "epoch": 2.9856903921204236,
      "grad_norm": 1.0637836456298828,
      "learning_rate": 3.021467546437631e-05,
      "loss": 0.7144,
      "step": 16066
    },
    {
      "epoch": 2.985876231183795,
      "grad_norm": 0.705267608165741,
      "learning_rate": 3.020421240353244e-05,
      "loss": 0.7417,
      "step": 16067
    },
    {
      "epoch": 2.986062070247166,
      "grad_norm": 0.856480062007904,
      "learning_rate": 3.019375083235988e-05,
      "loss": 0.7482,
      "step": 16068
    },
    {
      "epoch": 2.986247909310537,
      "grad_norm": 0.8884443640708923,
      "learning_rate": 3.0183290751081918e-05,
      "loss": 1.1324,
      "step": 16069
    },
    {
      "epoch": 2.986433748373908,
      "grad_norm": 0.9075732827186584,
      "learning_rate": 3.017283215992176e-05,
      "loss": 0.9301,
      "step": 16070
    },
    {
      "epoch": 2.9866195874372794,
      "grad_norm": 0.8808230757713318,
      "learning_rate": 3.016237505910272e-05,
      "loss": 0.9391,
      "step": 16071
    },
    {
      "epoch": 2.9868054265006503,
      "grad_norm": 0.8501234650611877,
      "learning_rate": 3.0151919448847933e-05,
      "loss": 1.0081,
      "step": 16072
    },
    {
      "epoch": 2.9869912655640216,
      "grad_norm": 0.7957719564437866,
      "learning_rate": 3.014146532938056e-05,
      "loss": 0.8891,
      "step": 16073
    },
    {
      "epoch": 2.987177104627393,
      "grad_norm": 0.8605754971504211,
      "learning_rate": 3.01310127009237e-05,
      "loss": 0.8632,
      "step": 16074
    },
    {
      "epoch": 2.987362943690764,
      "grad_norm": 0.732107937335968,
      "learning_rate": 3.01205615637005e-05,
      "loss": 0.7144,
      "step": 16075
    },
    {
      "epoch": 2.9875487827541347,
      "grad_norm": 0.8604171276092529,
      "learning_rate": 3.011011191793399e-05,
      "loss": 0.8015,
      "step": 16076
    },
    {
      "epoch": 2.987734621817506,
      "grad_norm": 0.7981947660446167,
      "learning_rate": 3.0099663763847185e-05,
      "loss": 0.9057,
      "step": 16077
    },
    {
      "epoch": 2.9879204608808774,
      "grad_norm": 0.7912887930870056,
      "learning_rate": 3.0089217101663103e-05,
      "loss": 0.8847,
      "step": 16078
    },
    {
      "epoch": 2.9881062999442483,
      "grad_norm": 1.4333757162094116,
      "learning_rate": 3.0078771931604743e-05,
      "loss": 1.3325,
      "step": 16079
    },
    {
      "epoch": 2.988292139007619,
      "grad_norm": 0.8902912139892578,
      "learning_rate": 3.0068328253894997e-05,
      "loss": 0.8626,
      "step": 16080
    },
    {
      "epoch": 2.9884779780709905,
      "grad_norm": 0.838409960269928,
      "learning_rate": 3.005788606875678e-05,
      "loss": 1.0151,
      "step": 16081
    },
    {
      "epoch": 2.988663817134362,
      "grad_norm": 1.026604175567627,
      "learning_rate": 3.004744537641293e-05,
      "loss": 0.9868,
      "step": 16082
    },
    {
      "epoch": 2.9888496561977327,
      "grad_norm": 0.898180365562439,
      "learning_rate": 3.0037006177086346e-05,
      "loss": 1.1465,
      "step": 16083
    },
    {
      "epoch": 2.989035495261104,
      "grad_norm": 0.748232364654541,
      "learning_rate": 3.0026568470999806e-05,
      "loss": 0.8587,
      "step": 16084
    },
    {
      "epoch": 2.989221334324475,
      "grad_norm": 0.80550616979599,
      "learning_rate": 3.0016132258376084e-05,
      "loss": 0.8996,
      "step": 16085
    },
    {
      "epoch": 2.9894071733878462,
      "grad_norm": 0.7416648864746094,
      "learning_rate": 3.0005697539437884e-05,
      "loss": 0.7914,
      "step": 16086
    },
    {
      "epoch": 2.989593012451217,
      "grad_norm": 0.8478348255157471,
      "learning_rate": 2.9995264314407956e-05,
      "loss": 0.9807,
      "step": 16087
    },
    {
      "epoch": 2.9897788515145884,
      "grad_norm": 1.0586156845092773,
      "learning_rate": 2.9984832583509016e-05,
      "loss": 1.0294,
      "step": 16088
    },
    {
      "epoch": 2.9899646905779593,
      "grad_norm": 0.7601550221443176,
      "learning_rate": 2.997440234696367e-05,
      "loss": 0.6927,
      "step": 16089
    },
    {
      "epoch": 2.9901505296413307,
      "grad_norm": 0.9126309156417847,
      "learning_rate": 2.9963973604994498e-05,
      "loss": 0.9581,
      "step": 16090
    },
    {
      "epoch": 2.9903363687047015,
      "grad_norm": 0.8759339451789856,
      "learning_rate": 2.995354635782417e-05,
      "loss": 0.8344,
      "step": 16091
    },
    {
      "epoch": 2.990522207768073,
      "grad_norm": 0.885123610496521,
      "learning_rate": 2.9943120605675177e-05,
      "loss": 0.8952,
      "step": 16092
    },
    {
      "epoch": 2.990708046831444,
      "grad_norm": 0.7887548208236694,
      "learning_rate": 2.9932696348770063e-05,
      "loss": 0.8907,
      "step": 16093
    },
    {
      "epoch": 2.990893885894815,
      "grad_norm": 0.7854909896850586,
      "learning_rate": 2.9922273587331263e-05,
      "loss": 0.8581,
      "step": 16094
    },
    {
      "epoch": 2.991079724958186,
      "grad_norm": 0.9311997294425964,
      "learning_rate": 2.9911852321581313e-05,
      "loss": 0.9449,
      "step": 16095
    },
    {
      "epoch": 2.9912655640215573,
      "grad_norm": 1.2775605916976929,
      "learning_rate": 2.9901432551742594e-05,
      "loss": 1.1392,
      "step": 16096
    },
    {
      "epoch": 2.9914514030849286,
      "grad_norm": 0.8485652208328247,
      "learning_rate": 2.9891014278037478e-05,
      "loss": 0.9056,
      "step": 16097
    },
    {
      "epoch": 2.9916372421482995,
      "grad_norm": 0.9734633564949036,
      "learning_rate": 2.9880597500688346e-05,
      "loss": 0.9759,
      "step": 16098
    },
    {
      "epoch": 2.991823081211671,
      "grad_norm": 0.7715386748313904,
      "learning_rate": 2.9870182219917564e-05,
      "loss": 0.7665,
      "step": 16099
    },
    {
      "epoch": 2.9920089202750417,
      "grad_norm": 0.8023921251296997,
      "learning_rate": 2.9859768435947388e-05,
      "loss": 0.7398,
      "step": 16100
    },
    {
      "epoch": 2.992194759338413,
      "grad_norm": 0.7612713575363159,
      "learning_rate": 2.9849356149000098e-05,
      "loss": 0.8553,
      "step": 16101
    },
    {
      "epoch": 2.992380598401784,
      "grad_norm": 1.0459855794906616,
      "learning_rate": 2.9838945359297888e-05,
      "loss": 0.946,
      "step": 16102
    },
    {
      "epoch": 2.9925664374651553,
      "grad_norm": 0.9531751275062561,
      "learning_rate": 2.9828536067063016e-05,
      "loss": 0.8861,
      "step": 16103
    },
    {
      "epoch": 2.992752276528526,
      "grad_norm": 0.8266820907592773,
      "learning_rate": 2.981812827251762e-05,
      "loss": 0.5946,
      "step": 16104
    },
    {
      "epoch": 2.9929381155918975,
      "grad_norm": 0.8020736575126648,
      "learning_rate": 2.980772197588384e-05,
      "loss": 1.0465,
      "step": 16105
    },
    {
      "epoch": 2.9931239546552684,
      "grad_norm": 0.9104275107383728,
      "learning_rate": 2.9797317177383756e-05,
      "loss": 0.9653,
      "step": 16106
    },
    {
      "epoch": 2.9933097937186397,
      "grad_norm": 1.0445524454116821,
      "learning_rate": 2.9786913877239487e-05,
      "loss": 1.1114,
      "step": 16107
    },
    {
      "epoch": 2.993495632782011,
      "grad_norm": 0.6620239615440369,
      "learning_rate": 2.9776512075673024e-05,
      "loss": 0.6062,
      "step": 16108
    },
    {
      "epoch": 2.993681471845382,
      "grad_norm": 0.8883970975875854,
      "learning_rate": 2.9766111772906424e-05,
      "loss": 1.0232,
      "step": 16109
    },
    {
      "epoch": 2.993867310908753,
      "grad_norm": 0.778900682926178,
      "learning_rate": 2.975571296916162e-05,
      "loss": 0.9295,
      "step": 16110
    },
    {
      "epoch": 2.994053149972124,
      "grad_norm": 0.8331949710845947,
      "learning_rate": 2.974531566466061e-05,
      "loss": 0.9017,
      "step": 16111
    },
    {
      "epoch": 2.9942389890354955,
      "grad_norm": 0.7040413022041321,
      "learning_rate": 2.9734919859625275e-05,
      "loss": 0.8675,
      "step": 16112
    },
    {
      "epoch": 2.9944248280988663,
      "grad_norm": 0.7584128379821777,
      "learning_rate": 2.9724525554277495e-05,
      "loss": 0.8517,
      "step": 16113
    },
    {
      "epoch": 2.9946106671622372,
      "grad_norm": 1.0098402500152588,
      "learning_rate": 2.9714132748839086e-05,
      "loss": 0.7754,
      "step": 16114
    },
    {
      "epoch": 2.9947965062256086,
      "grad_norm": 0.8807827830314636,
      "learning_rate": 2.970374144353193e-05,
      "loss": 0.9387,
      "step": 16115
    },
    {
      "epoch": 2.99498234528898,
      "grad_norm": 0.7473820447921753,
      "learning_rate": 2.9693351638577782e-05,
      "loss": 0.7434,
      "step": 16116
    },
    {
      "epoch": 2.9951681843523508,
      "grad_norm": 0.9319244027137756,
      "learning_rate": 2.9682963334198356e-05,
      "loss": 0.9373,
      "step": 16117
    },
    {
      "epoch": 2.995354023415722,
      "grad_norm": 0.9080926179885864,
      "learning_rate": 2.9672576530615447e-05,
      "loss": 1.0254,
      "step": 16118
    },
    {
      "epoch": 2.995539862479093,
      "grad_norm": 1.1424757242202759,
      "learning_rate": 2.9662191228050672e-05,
      "loss": 0.8289,
      "step": 16119
    },
    {
      "epoch": 2.9957257015424643,
      "grad_norm": 0.8220130205154419,
      "learning_rate": 2.965180742672574e-05,
      "loss": 1.0567,
      "step": 16120
    },
    {
      "epoch": 2.995911540605835,
      "grad_norm": 0.8922242522239685,
      "learning_rate": 2.9641425126862278e-05,
      "loss": 0.8066,
      "step": 16121
    },
    {
      "epoch": 2.9960973796692065,
      "grad_norm": 0.9695261716842651,
      "learning_rate": 2.9631044328681813e-05,
      "loss": 0.7672,
      "step": 16122
    },
    {
      "epoch": 2.996283218732578,
      "grad_norm": 0.8676010370254517,
      "learning_rate": 2.962066503240598e-05,
      "loss": 0.7832,
      "step": 16123
    },
    {
      "epoch": 2.9964690577959487,
      "grad_norm": 0.9023370146751404,
      "learning_rate": 2.9610287238256275e-05,
      "loss": 1.0591,
      "step": 16124
    },
    {
      "epoch": 2.9966548968593196,
      "grad_norm": 0.7452178001403809,
      "learning_rate": 2.95999109464542e-05,
      "loss": 0.9342,
      "step": 16125
    },
    {
      "epoch": 2.996840735922691,
      "grad_norm": 0.8523251414299011,
      "learning_rate": 2.9589536157221177e-05,
      "loss": 0.9896,
      "step": 16126
    },
    {
      "epoch": 2.9970265749860623,
      "grad_norm": 0.7315793633460999,
      "learning_rate": 2.957916287077871e-05,
      "loss": 0.9418,
      "step": 16127
    },
    {
      "epoch": 2.997212414049433,
      "grad_norm": 0.7727538347244263,
      "learning_rate": 2.9568791087348168e-05,
      "loss": 1.2043,
      "step": 16128
    },
    {
      "epoch": 2.997398253112804,
      "grad_norm": 0.6960626840591431,
      "learning_rate": 2.9558420807150878e-05,
      "loss": 0.5409,
      "step": 16129
    },
    {
      "epoch": 2.9975840921761754,
      "grad_norm": 0.8825200796127319,
      "learning_rate": 2.9548052030408213e-05,
      "loss": 0.8736,
      "step": 16130
    },
    {
      "epoch": 2.9977699312395467,
      "grad_norm": 0.8022184371948242,
      "learning_rate": 2.953768475734151e-05,
      "loss": 0.913,
      "step": 16131
    },
    {
      "epoch": 2.9979557703029176,
      "grad_norm": 0.7228624820709229,
      "learning_rate": 2.9527318988172015e-05,
      "loss": 0.9511,
      "step": 16132
    },
    {
      "epoch": 2.998141609366289,
      "grad_norm": 0.8025230169296265,
      "learning_rate": 2.9516954723120948e-05,
      "loss": 0.9299,
      "step": 16133
    },
    {
      "epoch": 2.99832744842966,
      "grad_norm": 0.8989067673683167,
      "learning_rate": 2.9506591962409512e-05,
      "loss": 0.9173,
      "step": 16134
    },
    {
      "epoch": 2.998513287493031,
      "grad_norm": 0.7874870300292969,
      "learning_rate": 2.949623070625892e-05,
      "loss": 0.9975,
      "step": 16135
    },
    {
      "epoch": 2.998699126556402,
      "grad_norm": 0.8301673531532288,
      "learning_rate": 2.9485870954890294e-05,
      "loss": 0.9403,
      "step": 16136
    },
    {
      "epoch": 2.9988849656197734,
      "grad_norm": 0.8750012516975403,
      "learning_rate": 2.9475512708524723e-05,
      "loss": 1.0227,
      "step": 16137
    },
    {
      "epoch": 2.9990708046831442,
      "grad_norm": 0.8098914623260498,
      "learning_rate": 2.946515596738333e-05,
      "loss": 0.9104,
      "step": 16138
    },
    {
      "epoch": 2.9992566437465156,
      "grad_norm": 0.8439404368400574,
      "learning_rate": 2.945480073168714e-05,
      "loss": 1.0937,
      "step": 16139
    },
    {
      "epoch": 2.9994424828098865,
      "grad_norm": 0.7623707056045532,
      "learning_rate": 2.944444700165714e-05,
      "loss": 0.81,
      "step": 16140
    },
    {
      "epoch": 2.999628321873258,
      "grad_norm": 0.8703219294548035,
      "learning_rate": 2.9434094777514375e-05,
      "loss": 1.0107,
      "step": 16141
    },
    {
      "epoch": 2.999814160936629,
      "grad_norm": 0.7967303395271301,
      "learning_rate": 2.942374405947973e-05,
      "loss": 1.0668,
      "step": 16142
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0338119268417358,
      "learning_rate": 2.9413394847774178e-05,
      "loss": 0.8666,
      "step": 16143
    },
    {
      "epoch": 3.000185839063371,
      "grad_norm": 0.8520253300666809,
      "learning_rate": 2.9403047142618588e-05,
      "loss": 0.915,
      "step": 16144
    },
    {
      "epoch": 3.000371678126742,
      "grad_norm": 1.1674602031707764,
      "learning_rate": 2.9392700944233797e-05,
      "loss": 0.7089,
      "step": 16145
    },
    {
      "epoch": 3.0005575171901135,
      "grad_norm": 0.8320359587669373,
      "learning_rate": 2.9382356252840614e-05,
      "loss": 0.9291,
      "step": 16146
    },
    {
      "epoch": 3.0007433562534844,
      "grad_norm": 0.9347818493843079,
      "learning_rate": 2.9372013068659886e-05,
      "loss": 0.7985,
      "step": 16147
    },
    {
      "epoch": 3.0009291953168553,
      "grad_norm": 0.8586578369140625,
      "learning_rate": 2.9361671391912337e-05,
      "loss": 0.9368,
      "step": 16148
    },
    {
      "epoch": 3.0011150343802266,
      "grad_norm": 0.8137461543083191,
      "learning_rate": 2.9351331222818657e-05,
      "loss": 0.8722,
      "step": 16149
    },
    {
      "epoch": 3.001300873443598,
      "grad_norm": 0.8831539750099182,
      "learning_rate": 2.934099256159958e-05,
      "loss": 1.0599,
      "step": 16150
    },
    {
      "epoch": 3.001486712506969,
      "grad_norm": 0.9825401306152344,
      "learning_rate": 2.933065540847578e-05,
      "loss": 1.0112,
      "step": 16151
    },
    {
      "epoch": 3.00167255157034,
      "grad_norm": 0.7601832151412964,
      "learning_rate": 2.932031976366788e-05,
      "loss": 0.8814,
      "step": 16152
    },
    {
      "epoch": 3.00167255157034,
      "eval_loss": 1.0085203647613525,
      "eval_runtime": 23.1453,
      "eval_samples_per_second": 47.18,
      "eval_steps_per_second": 23.59,
      "step": 16152
    },
    {
      "epoch": 3.0018583906337115,
      "grad_norm": 0.8601055145263672,
      "learning_rate": 2.930998562739645e-05,
      "loss": 0.8939,
      "step": 16153
    },
    {
      "epoch": 3.0020442296970824,
      "grad_norm": 0.745703935623169,
      "learning_rate": 2.9299652999882054e-05,
      "loss": 0.8115,
      "step": 16154
    },
    {
      "epoch": 3.0022300687604533,
      "grad_norm": 0.746161162853241,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.836,
      "step": 16155
    },
    {
      "epoch": 3.0024159078238246,
      "grad_norm": 0.9259798526763916,
      "learning_rate": 2.9278992272006545e-05,
      "loss": 0.8179,
      "step": 16156
    },
    {
      "epoch": 3.002601746887196,
      "grad_norm": 0.7527800798416138,
      "learning_rate": 2.9268664172086346e-05,
      "loss": 0.9853,
      "step": 16157
    },
    {
      "epoch": 3.002787585950567,
      "grad_norm": 0.7640870213508606,
      "learning_rate": 2.925833758180516e-05,
      "loss": 0.9494,
      "step": 16158
    },
    {
      "epoch": 3.0029734250139377,
      "grad_norm": 0.7097516059875488,
      "learning_rate": 2.924801250138336e-05,
      "loss": 0.8013,
      "step": 16159
    },
    {
      "epoch": 3.003159264077309,
      "grad_norm": 0.8351288437843323,
      "learning_rate": 2.923768893104133e-05,
      "loss": 0.7751,
      "step": 16160
    },
    {
      "epoch": 3.0033451031406804,
      "grad_norm": 1.039231300354004,
      "learning_rate": 2.9227366870999353e-05,
      "loss": 0.9235,
      "step": 16161
    },
    {
      "epoch": 3.0035309422040513,
      "grad_norm": 0.8639129400253296,
      "learning_rate": 2.921704632147778e-05,
      "loss": 0.6621,
      "step": 16162
    },
    {
      "epoch": 3.003716781267422,
      "grad_norm": 0.8037671446800232,
      "learning_rate": 2.920672728269692e-05,
      "loss": 1.0548,
      "step": 16163
    },
    {
      "epoch": 3.0039026203307935,
      "grad_norm": 0.8230457901954651,
      "learning_rate": 2.9196409754876984e-05,
      "loss": 0.8566,
      "step": 16164
    },
    {
      "epoch": 3.004088459394165,
      "grad_norm": 0.7340700030326843,
      "learning_rate": 2.918609373823814e-05,
      "loss": 0.8596,
      "step": 16165
    },
    {
      "epoch": 3.0042742984575357,
      "grad_norm": 0.8230661749839783,
      "learning_rate": 2.9175779233000633e-05,
      "loss": 1.0987,
      "step": 16166
    },
    {
      "epoch": 3.0044601375209066,
      "grad_norm": 1.0336686372756958,
      "learning_rate": 2.9165466239384587e-05,
      "loss": 0.6387,
      "step": 16167
    },
    {
      "epoch": 3.0046459765842783,
      "grad_norm": 0.7918894290924072,
      "learning_rate": 2.9155154757610103e-05,
      "loss": 0.8485,
      "step": 16168
    },
    {
      "epoch": 3.0048318156476492,
      "grad_norm": 0.8027900457382202,
      "learning_rate": 2.914484478789724e-05,
      "loss": 0.983,
      "step": 16169
    },
    {
      "epoch": 3.00501765471102,
      "grad_norm": 0.9236595034599304,
      "learning_rate": 2.91345363304661e-05,
      "loss": 1.1157,
      "step": 16170
    },
    {
      "epoch": 3.0052034937743914,
      "grad_norm": 0.8126418590545654,
      "learning_rate": 2.912422938553664e-05,
      "loss": 1.062,
      "step": 16171
    },
    {
      "epoch": 3.0053893328377628,
      "grad_norm": 0.7875935435295105,
      "learning_rate": 2.9113923953328903e-05,
      "loss": 0.8832,
      "step": 16172
    },
    {
      "epoch": 3.0055751719011337,
      "grad_norm": 0.7910284996032715,
      "learning_rate": 2.910362003406282e-05,
      "loss": 0.9957,
      "step": 16173
    },
    {
      "epoch": 3.0057610109645045,
      "grad_norm": 0.9798944592475891,
      "learning_rate": 2.909331762795827e-05,
      "loss": 1.0637,
      "step": 16174
    },
    {
      "epoch": 3.005946850027876,
      "grad_norm": 0.9422750473022461,
      "learning_rate": 2.9083016735235203e-05,
      "loss": 1.0967,
      "step": 16175
    },
    {
      "epoch": 3.006132689091247,
      "grad_norm": 0.9282810091972351,
      "learning_rate": 2.9072717356113454e-05,
      "loss": 1.149,
      "step": 16176
    },
    {
      "epoch": 3.006318528154618,
      "grad_norm": 0.8232243657112122,
      "learning_rate": 2.9062419490812798e-05,
      "loss": 1.1191,
      "step": 16177
    },
    {
      "epoch": 3.006504367217989,
      "grad_norm": 0.8733327388763428,
      "learning_rate": 2.9052123139553102e-05,
      "loss": 1.0162,
      "step": 16178
    },
    {
      "epoch": 3.0066902062813603,
      "grad_norm": 0.7954488396644592,
      "learning_rate": 2.9041828302554074e-05,
      "loss": 0.8666,
      "step": 16179
    },
    {
      "epoch": 3.0068760453447316,
      "grad_norm": 0.7659454941749573,
      "learning_rate": 2.9031534980035468e-05,
      "loss": 0.9133,
      "step": 16180
    },
    {
      "epoch": 3.0070618844081025,
      "grad_norm": 0.7625312805175781,
      "learning_rate": 2.9021243172216916e-05,
      "loss": 0.7314,
      "step": 16181
    },
    {
      "epoch": 3.0072477234714734,
      "grad_norm": 1.0805476903915405,
      "learning_rate": 2.9010952879318132e-05,
      "loss": 0.8219,
      "step": 16182
    },
    {
      "epoch": 3.0074335625348447,
      "grad_norm": 0.9051377773284912,
      "learning_rate": 2.9000664101558773e-05,
      "loss": 1.0145,
      "step": 16183
    },
    {
      "epoch": 3.007619401598216,
      "grad_norm": 0.7972425222396851,
      "learning_rate": 2.8990376839158384e-05,
      "loss": 0.89,
      "step": 16184
    },
    {
      "epoch": 3.007805240661587,
      "grad_norm": 0.7707135677337646,
      "learning_rate": 2.8980091092336525e-05,
      "loss": 0.8227,
      "step": 16185
    },
    {
      "epoch": 3.0079910797249583,
      "grad_norm": 0.9423364400863647,
      "learning_rate": 2.8969806861312776e-05,
      "loss": 0.9644,
      "step": 16186
    },
    {
      "epoch": 3.0081769187883296,
      "grad_norm": 0.8737858533859253,
      "learning_rate": 2.8959524146306603e-05,
      "loss": 1.029,
      "step": 16187
    },
    {
      "epoch": 3.0083627578517005,
      "grad_norm": 0.8065149784088135,
      "learning_rate": 2.894924294753748e-05,
      "loss": 1.063,
      "step": 16188
    },
    {
      "epoch": 3.0085485969150714,
      "grad_norm": 0.8207863569259644,
      "learning_rate": 2.8938963265224807e-05,
      "loss": 0.9712,
      "step": 16189
    },
    {
      "epoch": 3.0087344359784427,
      "grad_norm": 0.7644856572151184,
      "learning_rate": 2.8928685099588037e-05,
      "loss": 0.8829,
      "step": 16190
    },
    {
      "epoch": 3.008920275041814,
      "grad_norm": 0.8748577833175659,
      "learning_rate": 2.8918408450846533e-05,
      "loss": 1.2508,
      "step": 16191
    },
    {
      "epoch": 3.009106114105185,
      "grad_norm": 0.909366250038147,
      "learning_rate": 2.8908133319219588e-05,
      "loss": 0.8923,
      "step": 16192
    },
    {
      "epoch": 3.009291953168556,
      "grad_norm": 0.9220009446144104,
      "learning_rate": 2.8897859704926566e-05,
      "loss": 1.0259,
      "step": 16193
    },
    {
      "epoch": 3.009477792231927,
      "grad_norm": 0.7947103381156921,
      "learning_rate": 2.8887587608186684e-05,
      "loss": 0.9017,
      "step": 16194
    },
    {
      "epoch": 3.0096636312952985,
      "grad_norm": 0.7401075959205627,
      "learning_rate": 2.887731702921924e-05,
      "loss": 0.8673,
      "step": 16195
    },
    {
      "epoch": 3.0098494703586693,
      "grad_norm": 0.8753591775894165,
      "learning_rate": 2.886704796824341e-05,
      "loss": 0.8195,
      "step": 16196
    },
    {
      "epoch": 3.0100353094220402,
      "grad_norm": 0.8051326870918274,
      "learning_rate": 2.8856780425478347e-05,
      "loss": 0.9303,
      "step": 16197
    },
    {
      "epoch": 3.0102211484854116,
      "grad_norm": 0.8923689126968384,
      "learning_rate": 2.8846514401143254e-05,
      "loss": 0.7542,
      "step": 16198
    },
    {
      "epoch": 3.010406987548783,
      "grad_norm": 0.943494439125061,
      "learning_rate": 2.88362498954572e-05,
      "loss": 1.071,
      "step": 16199
    },
    {
      "epoch": 3.0105928266121538,
      "grad_norm": 0.855754554271698,
      "learning_rate": 2.882598690863927e-05,
      "loss": 0.9027,
      "step": 16200
    },
    {
      "epoch": 3.010778665675525,
      "grad_norm": 0.8352408409118652,
      "learning_rate": 2.881572544090848e-05,
      "loss": 0.6896,
      "step": 16201
    },
    {
      "epoch": 3.0109645047388964,
      "grad_norm": 0.935312807559967,
      "learning_rate": 2.8805465492483897e-05,
      "loss": 1.0309,
      "step": 16202
    },
    {
      "epoch": 3.0111503438022673,
      "grad_norm": 0.9511630535125732,
      "learning_rate": 2.879520706358446e-05,
      "loss": 0.9513,
      "step": 16203
    },
    {
      "epoch": 3.011336182865638,
      "grad_norm": 0.8023136854171753,
      "learning_rate": 2.878495015442916e-05,
      "loss": 0.8842,
      "step": 16204
    },
    {
      "epoch": 3.0115220219290095,
      "grad_norm": 1.0229847431182861,
      "learning_rate": 2.8774694765236854e-05,
      "loss": 1.0292,
      "step": 16205
    },
    {
      "epoch": 3.011707860992381,
      "grad_norm": 0.809948742389679,
      "learning_rate": 2.8764440896226498e-05,
      "loss": 0.9655,
      "step": 16206
    },
    {
      "epoch": 3.0118937000557517,
      "grad_norm": 0.7976549863815308,
      "learning_rate": 2.875418854761691e-05,
      "loss": 0.8296,
      "step": 16207
    },
    {
      "epoch": 3.0120795391191226,
      "grad_norm": 0.8738260269165039,
      "learning_rate": 2.8743937719626902e-05,
      "loss": 0.938,
      "step": 16208
    },
    {
      "epoch": 3.012265378182494,
      "grad_norm": 0.8011206388473511,
      "learning_rate": 2.8733688412475223e-05,
      "loss": 1.0696,
      "step": 16209
    },
    {
      "epoch": 3.0124512172458653,
      "grad_norm": 0.8072490692138672,
      "learning_rate": 2.8723440626380705e-05,
      "loss": 1.0296,
      "step": 16210
    },
    {
      "epoch": 3.012637056309236,
      "grad_norm": 0.7365478873252869,
      "learning_rate": 2.8713194361562036e-05,
      "loss": 0.9477,
      "step": 16211
    },
    {
      "epoch": 3.012822895372607,
      "grad_norm": 0.7511142492294312,
      "learning_rate": 2.8702949618237896e-05,
      "loss": 0.8417,
      "step": 16212
    },
    {
      "epoch": 3.0130087344359784,
      "grad_norm": 0.8439598679542542,
      "learning_rate": 2.869270639662691e-05,
      "loss": 0.8634,
      "step": 16213
    },
    {
      "epoch": 3.0131945734993497,
      "grad_norm": 0.8430346250534058,
      "learning_rate": 2.8682464696947742e-05,
      "loss": 1.0266,
      "step": 16214
    },
    {
      "epoch": 3.0133804125627206,
      "grad_norm": 0.7551569938659668,
      "learning_rate": 2.867222451941901e-05,
      "loss": 0.8768,
      "step": 16215
    },
    {
      "epoch": 3.0135662516260915,
      "grad_norm": 0.927182674407959,
      "learning_rate": 2.8661985864259243e-05,
      "loss": 0.8485,
      "step": 16216
    },
    {
      "epoch": 3.013752090689463,
      "grad_norm": 0.8453295230865479,
      "learning_rate": 2.8651748731686935e-05,
      "loss": 0.8316,
      "step": 16217
    },
    {
      "epoch": 3.013937929752834,
      "grad_norm": 0.8905909657478333,
      "learning_rate": 2.8641513121920637e-05,
      "loss": 0.9733,
      "step": 16218
    },
    {
      "epoch": 3.014123768816205,
      "grad_norm": 0.9714555144309998,
      "learning_rate": 2.8631279035178793e-05,
      "loss": 1.0372,
      "step": 16219
    },
    {
      "epoch": 3.0143096078795764,
      "grad_norm": 0.7101377248764038,
      "learning_rate": 2.8621046471679826e-05,
      "loss": 0.8919,
      "step": 16220
    },
    {
      "epoch": 3.0144954469429477,
      "grad_norm": 0.9432785511016846,
      "learning_rate": 2.8610815431642103e-05,
      "loss": 0.7906,
      "step": 16221
    },
    {
      "epoch": 3.0146812860063186,
      "grad_norm": 0.8138036727905273,
      "learning_rate": 2.860058591528404e-05,
      "loss": 0.7629,
      "step": 16222
    },
    {
      "epoch": 3.0148671250696895,
      "grad_norm": 0.7100986838340759,
      "learning_rate": 2.859035792282394e-05,
      "loss": 0.9996,
      "step": 16223
    },
    {
      "epoch": 3.015052964133061,
      "grad_norm": 0.8357480764389038,
      "learning_rate": 2.8580131454480074e-05,
      "loss": 0.7076,
      "step": 16224
    },
    {
      "epoch": 3.015238803196432,
      "grad_norm": 0.8065438866615295,
      "learning_rate": 2.8569906510470757e-05,
      "loss": 0.7609,
      "step": 16225
    },
    {
      "epoch": 3.015424642259803,
      "grad_norm": 0.9661011099815369,
      "learning_rate": 2.8559683091014223e-05,
      "loss": 1.002,
      "step": 16226
    },
    {
      "epoch": 3.015610481323174,
      "grad_norm": 0.9474563598632812,
      "learning_rate": 2.8549461196328663e-05,
      "loss": 1.0073,
      "step": 16227
    },
    {
      "epoch": 3.015796320386545,
      "grad_norm": 0.6291988492012024,
      "learning_rate": 2.853924082663223e-05,
      "loss": 0.5213,
      "step": 16228
    },
    {
      "epoch": 3.0159821594499165,
      "grad_norm": 0.7146613001823425,
      "learning_rate": 2.8529021982143056e-05,
      "loss": 0.6329,
      "step": 16229
    },
    {
      "epoch": 3.0161679985132874,
      "grad_norm": 0.780358076095581,
      "learning_rate": 2.8518804663079278e-05,
      "loss": 0.7285,
      "step": 16230
    },
    {
      "epoch": 3.0163538375766583,
      "grad_norm": 0.9360339641571045,
      "learning_rate": 2.8508588869658946e-05,
      "loss": 0.9908,
      "step": 16231
    },
    {
      "epoch": 3.0165396766400296,
      "grad_norm": 0.808726966381073,
      "learning_rate": 2.849837460210011e-05,
      "loss": 0.8783,
      "step": 16232
    },
    {
      "epoch": 3.016725515703401,
      "grad_norm": 0.9102755784988403,
      "learning_rate": 2.848816186062073e-05,
      "loss": 0.852,
      "step": 16233
    },
    {
      "epoch": 3.016911354766772,
      "grad_norm": 0.725265383720398,
      "learning_rate": 2.847795064543881e-05,
      "loss": 0.9368,
      "step": 16234
    },
    {
      "epoch": 3.017097193830143,
      "grad_norm": 0.8580981492996216,
      "learning_rate": 2.846774095677234e-05,
      "loss": 0.974,
      "step": 16235
    },
    {
      "epoch": 3.0172830328935145,
      "grad_norm": 0.7755009531974792,
      "learning_rate": 2.845753279483918e-05,
      "loss": 0.9889,
      "step": 16236
    },
    {
      "epoch": 3.0174688719568854,
      "grad_norm": 1.553571343421936,
      "learning_rate": 2.844732615985718e-05,
      "loss": 1.4221,
      "step": 16237
    },
    {
      "epoch": 3.0176547110202563,
      "grad_norm": 0.7738388180732727,
      "learning_rate": 2.8437121052044235e-05,
      "loss": 0.8568,
      "step": 16238
    },
    {
      "epoch": 3.0178405500836276,
      "grad_norm": 0.7205919027328491,
      "learning_rate": 2.8426917471618144e-05,
      "loss": 0.5961,
      "step": 16239
    },
    {
      "epoch": 3.018026389146999,
      "grad_norm": 1.14812433719635,
      "learning_rate": 2.841671541879668e-05,
      "loss": 0.9657,
      "step": 16240
    },
    {
      "epoch": 3.0001858390633713,
      "grad_norm": 0.7224998474121094,
      "learning_rate": 2.8406514893797543e-05,
      "loss": 0.7203,
      "step": 16241
    },
    {
      "epoch": 3.000371678126742,
      "grad_norm": 0.7414650321006775,
      "learning_rate": 2.839631589683852e-05,
      "loss": 0.7727,
      "step": 16242
    },
    {
      "epoch": 3.0005575171901135,
      "grad_norm": 0.8323161602020264,
      "learning_rate": 2.8386118428137254e-05,
      "loss": 0.8729,
      "step": 16243
    },
    {
      "epoch": 3.0007433562534844,
      "grad_norm": 0.6795046925544739,
      "learning_rate": 2.83759224879114e-05,
      "loss": 0.6143,
      "step": 16244
    },
    {
      "epoch": 3.0009291953168558,
      "grad_norm": 0.9268404245376587,
      "learning_rate": 2.8365728076378508e-05,
      "loss": 0.6701,
      "step": 16245
    },
    {
      "epoch": 3.0011150343802266,
      "grad_norm": 0.7565287351608276,
      "learning_rate": 2.835553519375628e-05,
      "loss": 0.7988,
      "step": 16246
    },
    {
      "epoch": 3.001300873443598,
      "grad_norm": 0.8616061210632324,
      "learning_rate": 2.8345343840262205e-05,
      "loss": 1.0423,
      "step": 16247
    },
    {
      "epoch": 3.001486712506969,
      "grad_norm": 0.7660315632820129,
      "learning_rate": 2.8335154016113806e-05,
      "loss": 0.9305,
      "step": 16248
    },
    {
      "epoch": 3.00167255157034,
      "grad_norm": 0.8274088501930237,
      "learning_rate": 2.8324965721528528e-05,
      "loss": 0.8605,
      "step": 16249
    },
    {
      "epoch": 3.001858390633711,
      "grad_norm": 0.7002495527267456,
      "learning_rate": 2.83147789567239e-05,
      "loss": 0.7109,
      "step": 16250
    },
    {
      "epoch": 3.0020442296970824,
      "grad_norm": 0.7538357973098755,
      "learning_rate": 2.8304593721917285e-05,
      "loss": 0.6587,
      "step": 16251
    },
    {
      "epoch": 3.0022300687604533,
      "grad_norm": 0.7135251760482788,
      "learning_rate": 2.8294410017326088e-05,
      "loss": 0.6873,
      "step": 16252
    },
    {
      "epoch": 3.0024159078238246,
      "grad_norm": 1.008959412574768,
      "learning_rate": 2.828422784316762e-05,
      "loss": 1.0646,
      "step": 16253
    },
    {
      "epoch": 3.0026017468871955,
      "grad_norm": 0.7986328601837158,
      "learning_rate": 2.8274047199659282e-05,
      "loss": 0.8082,
      "step": 16254
    },
    {
      "epoch": 3.002787585950567,
      "grad_norm": 0.7848049998283386,
      "learning_rate": 2.8263868087018274e-05,
      "loss": 0.8706,
      "step": 16255
    },
    {
      "epoch": 3.0029734250139377,
      "grad_norm": 0.9080967307090759,
      "learning_rate": 2.8253690505461937e-05,
      "loss": 0.758,
      "step": 16256
    },
    {
      "epoch": 3.003159264077309,
      "grad_norm": 0.875514805316925,
      "learning_rate": 2.8243514455207432e-05,
      "loss": 0.8892,
      "step": 16257
    },
    {
      "epoch": 3.0033451031406804,
      "grad_norm": 0.7878082394599915,
      "learning_rate": 2.8233339936471993e-05,
      "loss": 0.7364,
      "step": 16258
    },
    {
      "epoch": 3.0035309422040513,
      "grad_norm": 0.8241668343544006,
      "learning_rate": 2.8223166949472758e-05,
      "loss": 0.7358,
      "step": 16259
    },
    {
      "epoch": 3.0037167812674226,
      "grad_norm": 0.8483393788337708,
      "learning_rate": 2.8212995494426842e-05,
      "loss": 0.9541,
      "step": 16260
    },
    {
      "epoch": 3.0039026203307935,
      "grad_norm": 0.8021813035011292,
      "learning_rate": 2.8202825571551326e-05,
      "loss": 0.7795,
      "step": 16261
    },
    {
      "epoch": 3.004088459394165,
      "grad_norm": 0.8833257555961609,
      "learning_rate": 2.819265718106332e-05,
      "loss": 0.9313,
      "step": 16262
    },
    {
      "epoch": 3.0042742984575357,
      "grad_norm": 0.8764292597770691,
      "learning_rate": 2.818249032317981e-05,
      "loss": 0.7534,
      "step": 16263
    },
    {
      "epoch": 3.004460137520907,
      "grad_norm": 0.9224509596824646,
      "learning_rate": 2.8172324998117815e-05,
      "loss": 0.7994,
      "step": 16264
    },
    {
      "epoch": 3.004645976584278,
      "grad_norm": 0.7286317944526672,
      "learning_rate": 2.816216120609424e-05,
      "loss": 0.9616,
      "step": 16265
    },
    {
      "epoch": 3.0048318156476492,
      "grad_norm": 0.9721893072128296,
      "learning_rate": 2.8151998947326065e-05,
      "loss": 0.7647,
      "step": 16266
    },
    {
      "epoch": 3.00501765471102,
      "grad_norm": 0.6974045038223267,
      "learning_rate": 2.814183822203019e-05,
      "loss": 0.5875,
      "step": 16267
    },
    {
      "epoch": 3.0052034937743914,
      "grad_norm": 0.7608116269111633,
      "learning_rate": 2.8131679030423485e-05,
      "loss": 0.8592,
      "step": 16268
    },
    {
      "epoch": 3.0053893328377623,
      "grad_norm": 0.5959421396255493,
      "learning_rate": 2.8121521372722716e-05,
      "loss": 0.438,
      "step": 16269
    },
    {
      "epoch": 3.0055751719011337,
      "grad_norm": 0.8465961813926697,
      "learning_rate": 2.811136524914476e-05,
      "loss": 0.5456,
      "step": 16270
    },
    {
      "epoch": 3.0057610109645045,
      "grad_norm": 0.8137955665588379,
      "learning_rate": 2.8101210659906342e-05,
      "loss": 0.9552,
      "step": 16271
    },
    {
      "epoch": 3.005946850027876,
      "grad_norm": 0.905461847782135,
      "learning_rate": 2.8091057605224203e-05,
      "loss": 0.9942,
      "step": 16272
    },
    {
      "epoch": 3.006132689091247,
      "grad_norm": 0.9759244322776794,
      "learning_rate": 2.808090608531502e-05,
      "loss": 1.0752,
      "step": 16273
    },
    {
      "epoch": 3.006318528154618,
      "grad_norm": 0.9051206707954407,
      "learning_rate": 2.807075610039549e-05,
      "loss": 0.821,
      "step": 16274
    },
    {
      "epoch": 3.0065043672179894,
      "grad_norm": 0.7651662230491638,
      "learning_rate": 2.8060607650682246e-05,
      "loss": 0.7506,
      "step": 16275
    },
    {
      "epoch": 3.0066902062813603,
      "grad_norm": 1.1936324834823608,
      "learning_rate": 2.805046073639185e-05,
      "loss": 1.0375,
      "step": 16276
    },
    {
      "epoch": 3.0068760453447316,
      "grad_norm": 0.7253216505050659,
      "learning_rate": 2.8040315357740887e-05,
      "loss": 0.5289,
      "step": 16277
    },
    {
      "epoch": 3.0070618844081025,
      "grad_norm": 0.9311563968658447,
      "learning_rate": 2.8030171514945947e-05,
      "loss": 0.9183,
      "step": 16278
    },
    {
      "epoch": 3.007247723471474,
      "grad_norm": 0.8920841813087463,
      "learning_rate": 2.8020029208223486e-05,
      "loss": 1.0476,
      "step": 16279
    },
    {
      "epoch": 3.0074335625348447,
      "grad_norm": 0.7824742197990417,
      "learning_rate": 2.8009888437789976e-05,
      "loss": 0.8087,
      "step": 16280
    },
    {
      "epoch": 3.007619401598216,
      "grad_norm": 0.5891439318656921,
      "learning_rate": 2.7999749203861836e-05,
      "loss": 0.5434,
      "step": 16281
    },
    {
      "epoch": 3.007805240661587,
      "grad_norm": 0.9570181965827942,
      "learning_rate": 2.7989611506655513e-05,
      "loss": 0.8865,
      "step": 16282
    },
    {
      "epoch": 3.0079910797249583,
      "grad_norm": 0.8050483465194702,
      "learning_rate": 2.797947534638736e-05,
      "loss": 0.4971,
      "step": 16283
    },
    {
      "epoch": 3.008176918788329,
      "grad_norm": 0.7769497036933899,
      "learning_rate": 2.796934072327372e-05,
      "loss": 0.9063,
      "step": 16284
    },
    {
      "epoch": 3.0083627578517005,
      "grad_norm": 1.0004961490631104,
      "learning_rate": 2.795920763753086e-05,
      "loss": 0.8958,
      "step": 16285
    },
    {
      "epoch": 3.0085485969150714,
      "grad_norm": 0.8578765392303467,
      "learning_rate": 2.7949076089375116e-05,
      "loss": 0.8738,
      "step": 16286
    },
    {
      "epoch": 3.0087344359784427,
      "grad_norm": 0.8630593419075012,
      "learning_rate": 2.7938946079022665e-05,
      "loss": 0.7894,
      "step": 16287
    },
    {
      "epoch": 3.0089202750418136,
      "grad_norm": 0.8347944617271423,
      "learning_rate": 2.7928817606689774e-05,
      "loss": 0.8217,
      "step": 16288
    },
    {
      "epoch": 3.009106114105185,
      "grad_norm": 0.8454095125198364,
      "learning_rate": 2.7918690672592573e-05,
      "loss": 0.7392,
      "step": 16289
    },
    {
      "epoch": 3.0092919531685562,
      "grad_norm": 0.8619915843009949,
      "learning_rate": 2.790856527694724e-05,
      "loss": 0.9189,
      "step": 16290
    },
    {
      "epoch": 3.009477792231927,
      "grad_norm": 0.8524312376976013,
      "learning_rate": 2.7898441419969878e-05,
      "loss": 0.6246,
      "step": 16291
    },
    {
      "epoch": 3.0096636312952985,
      "grad_norm": 0.717739999294281,
      "learning_rate": 2.7888319101876546e-05,
      "loss": 0.8099,
      "step": 16292
    },
    {
      "epoch": 3.0098494703586693,
      "grad_norm": 0.652247965335846,
      "learning_rate": 2.7878198322883276e-05,
      "loss": 0.5795,
      "step": 16293
    },
    {
      "epoch": 3.0100353094220407,
      "grad_norm": 0.9642404913902283,
      "learning_rate": 2.786807908320611e-05,
      "loss": 0.9324,
      "step": 16294
    },
    {
      "epoch": 3.0102211484854116,
      "grad_norm": 0.9150660037994385,
      "learning_rate": 2.7857961383061028e-05,
      "loss": 0.995,
      "step": 16295
    },
    {
      "epoch": 3.010406987548783,
      "grad_norm": 0.7613053917884827,
      "learning_rate": 2.784784522266395e-05,
      "loss": 1.017,
      "step": 16296
    },
    {
      "epoch": 3.0105928266121538,
      "grad_norm": 1.062303066253662,
      "learning_rate": 2.7837730602230737e-05,
      "loss": 0.8231,
      "step": 16297
    },
    {
      "epoch": 3.010778665675525,
      "grad_norm": 0.9443035125732422,
      "learning_rate": 2.78276175219774e-05,
      "loss": 0.8127,
      "step": 16298
    },
    {
      "epoch": 3.010964504738896,
      "grad_norm": 1.8857661485671997,
      "learning_rate": 2.7817505982119706e-05,
      "loss": 1.1149,
      "step": 16299
    },
    {
      "epoch": 3.0111503438022673,
      "grad_norm": 0.7440164685249329,
      "learning_rate": 2.7807395982873475e-05,
      "loss": 0.6358,
      "step": 16300
    },
    {
      "epoch": 3.011336182865638,
      "grad_norm": 0.9950499534606934,
      "learning_rate": 2.7797287524454463e-05,
      "loss": 0.7353,
      "step": 16301
    },
    {
      "epoch": 3.0115220219290095,
      "grad_norm": 0.8099482655525208,
      "learning_rate": 2.7787180607078477e-05,
      "loss": 0.7613,
      "step": 16302
    },
    {
      "epoch": 3.0117078609923804,
      "grad_norm": 0.7734261751174927,
      "learning_rate": 2.7777075230961192e-05,
      "loss": 0.8693,
      "step": 16303
    },
    {
      "epoch": 3.0118937000557517,
      "grad_norm": 0.7725483775138855,
      "learning_rate": 2.7766971396318297e-05,
      "loss": 0.8373,
      "step": 16304
    },
    {
      "epoch": 3.0120795391191226,
      "grad_norm": 0.8226912021636963,
      "learning_rate": 2.7756869103365413e-05,
      "loss": 0.8417,
      "step": 16305
    },
    {
      "epoch": 3.012265378182494,
      "grad_norm": 0.9145111441612244,
      "learning_rate": 2.774676835231822e-05,
      "loss": 0.9199,
      "step": 16306
    },
    {
      "epoch": 3.0124512172458653,
      "grad_norm": 0.8743301033973694,
      "learning_rate": 2.7736669143392256e-05,
      "loss": 0.8631,
      "step": 16307
    },
    {
      "epoch": 3.012637056309236,
      "grad_norm": 0.9417018294334412,
      "learning_rate": 2.7726571476803053e-05,
      "loss": 0.657,
      "step": 16308
    },
    {
      "epoch": 3.0128228953726075,
      "grad_norm": 0.8447418808937073,
      "learning_rate": 2.7716475352766146e-05,
      "loss": 0.7629,
      "step": 16309
    },
    {
      "epoch": 3.0130087344359784,
      "grad_norm": 0.8393754959106445,
      "learning_rate": 2.7706380771497075e-05,
      "loss": 0.7797,
      "step": 16310
    },
    {
      "epoch": 3.0131945734993497,
      "grad_norm": 0.794707715511322,
      "learning_rate": 2.7696287733211234e-05,
      "loss": 0.8391,
      "step": 16311
    },
    {
      "epoch": 3.0133804125627206,
      "grad_norm": 0.8632328510284424,
      "learning_rate": 2.7686196238124063e-05,
      "loss": 0.6809,
      "step": 16312
    },
    {
      "epoch": 3.013566251626092,
      "grad_norm": 0.7959421873092651,
      "learning_rate": 2.767610628645091e-05,
      "loss": 0.6658,
      "step": 16313
    },
    {
      "epoch": 3.013752090689463,
      "grad_norm": 1.0240352153778076,
      "learning_rate": 2.766601787840717e-05,
      "loss": 1.1223,
      "step": 16314
    },
    {
      "epoch": 3.013937929752834,
      "grad_norm": 0.8519721031188965,
      "learning_rate": 2.765593101420816e-05,
      "loss": 0.8447,
      "step": 16315
    },
    {
      "epoch": 3.014123768816205,
      "grad_norm": 0.7613992094993591,
      "learning_rate": 2.7645845694069116e-05,
      "loss": 0.793,
      "step": 16316
    },
    {
      "epoch": 3.0143096078795764,
      "grad_norm": 0.8992709517478943,
      "learning_rate": 2.7635761918205372e-05,
      "loss": 0.8327,
      "step": 16317
    },
    {
      "epoch": 3.0144954469429472,
      "grad_norm": 0.8548943400382996,
      "learning_rate": 2.7625679686832072e-05,
      "loss": 1.0561,
      "step": 16318
    },
    {
      "epoch": 3.0146812860063186,
      "grad_norm": 1.1899662017822266,
      "learning_rate": 2.761559900016447e-05,
      "loss": 0.8694,
      "step": 16319
    },
    {
      "epoch": 3.0148671250696895,
      "grad_norm": 0.8247866630554199,
      "learning_rate": 2.7605519858417694e-05,
      "loss": 0.7833,
      "step": 16320
    },
    {
      "epoch": 3.015052964133061,
      "grad_norm": 0.7799816727638245,
      "learning_rate": 2.7595442261806837e-05,
      "loss": 0.8294,
      "step": 16321
    },
    {
      "epoch": 3.0152388031964317,
      "grad_norm": 0.8495178818702698,
      "learning_rate": 2.7585366210547036e-05,
      "loss": 0.6405,
      "step": 16322
    },
    {
      "epoch": 3.015424642259803,
      "grad_norm": 0.871384859085083,
      "learning_rate": 2.7575291704853323e-05,
      "loss": 0.7246,
      "step": 16323
    },
    {
      "epoch": 3.0156104813231743,
      "grad_norm": 0.8070007562637329,
      "learning_rate": 2.7565218744940724e-05,
      "loss": 0.791,
      "step": 16324
    },
    {
      "epoch": 3.015796320386545,
      "grad_norm": 0.9237125515937805,
      "learning_rate": 2.75551473310242e-05,
      "loss": 0.7915,
      "step": 16325
    },
    {
      "epoch": 3.0159821594499165,
      "grad_norm": 0.8128219246864319,
      "learning_rate": 2.754507746331877e-05,
      "loss": 0.8471,
      "step": 16326
    },
    {
      "epoch": 3.0161679985132874,
      "grad_norm": 0.7762858271598816,
      "learning_rate": 2.7535009142039315e-05,
      "loss": 0.7577,
      "step": 16327
    },
    {
      "epoch": 3.0163538375766588,
      "grad_norm": 1.1461944580078125,
      "learning_rate": 2.7524942367400706e-05,
      "loss": 1.1128,
      "step": 16328
    },
    {
      "epoch": 3.0165396766400296,
      "grad_norm": 1.7739145755767822,
      "learning_rate": 2.7514877139617833e-05,
      "loss": 1.3723,
      "step": 16329
    },
    {
      "epoch": 3.016725515703401,
      "grad_norm": 0.894992470741272,
      "learning_rate": 2.7504813458905554e-05,
      "loss": 0.8777,
      "step": 16330
    },
    {
      "epoch": 3.016911354766772,
      "grad_norm": 0.8429700136184692,
      "learning_rate": 2.749475132547862e-05,
      "loss": 0.9114,
      "step": 16331
    },
    {
      "epoch": 3.017097193830143,
      "grad_norm": 0.9092171788215637,
      "learning_rate": 2.748469073955179e-05,
      "loss": 0.9582,
      "step": 16332
    },
    {
      "epoch": 3.017283032893514,
      "grad_norm": 0.9921269416809082,
      "learning_rate": 2.747463170133977e-05,
      "loss": 0.8154,
      "step": 16333
    },
    {
      "epoch": 3.0174688719568854,
      "grad_norm": 0.9600369930267334,
      "learning_rate": 2.7464574211057304e-05,
      "loss": 0.6221,
      "step": 16334
    },
    {
      "epoch": 3.0176547110202563,
      "grad_norm": 0.8476983904838562,
      "learning_rate": 2.7454518268919027e-05,
      "loss": 0.8533,
      "step": 16335
    },
    {
      "epoch": 3.0178405500836276,
      "grad_norm": 0.706290602684021,
      "learning_rate": 2.7444463875139537e-05,
      "loss": 0.6707,
      "step": 16336
    },
    {
      "epoch": 3.0180263891469985,
      "grad_norm": 0.85426926612854,
      "learning_rate": 2.7434411029933483e-05,
      "loss": 0.7433,
      "step": 16337
    },
    {
      "epoch": 3.01821222821037,
      "grad_norm": 0.9000372290611267,
      "learning_rate": 2.7424359733515403e-05,
      "loss": 0.9114,
      "step": 16338
    },
    {
      "epoch": 3.018398067273741,
      "grad_norm": 0.8626327514648438,
      "learning_rate": 2.7414309986099783e-05,
      "loss": 1.0012,
      "step": 16339
    },
    {
      "epoch": 3.018583906337112,
      "grad_norm": 0.7985569834709167,
      "learning_rate": 2.7404261787901185e-05,
      "loss": 0.8213,
      "step": 16340
    },
    {
      "epoch": 3.0187697454004834,
      "grad_norm": 0.808411717414856,
      "learning_rate": 2.7394215139134015e-05,
      "loss": 0.7976,
      "step": 16341
    },
    {
      "epoch": 3.0189555844638543,
      "grad_norm": 0.8957644701004028,
      "learning_rate": 2.7384170040012758e-05,
      "loss": 0.8949,
      "step": 16342
    },
    {
      "epoch": 3.0191414235272256,
      "grad_norm": 0.8540877103805542,
      "learning_rate": 2.737412649075177e-05,
      "loss": 0.8557,
      "step": 16343
    },
    {
      "epoch": 3.0193272625905965,
      "grad_norm": 0.7954753637313843,
      "learning_rate": 2.7364084491565424e-05,
      "loss": 0.8965,
      "step": 16344
    },
    {
      "epoch": 3.019513101653968,
      "grad_norm": 0.7853621244430542,
      "learning_rate": 2.7354044042668014e-05,
      "loss": 0.7109,
      "step": 16345
    },
    {
      "epoch": 3.0196989407173387,
      "grad_norm": 0.7370253205299377,
      "learning_rate": 2.73440051442739e-05,
      "loss": 0.7888,
      "step": 16346
    },
    {
      "epoch": 3.01988477978071,
      "grad_norm": 0.8112165927886963,
      "learning_rate": 2.7333967796597315e-05,
      "loss": 0.9457,
      "step": 16347
    },
    {
      "epoch": 3.020070618844081,
      "grad_norm": 1.0122783184051514,
      "learning_rate": 2.732393199985246e-05,
      "loss": 0.5817,
      "step": 16348
    },
    {
      "epoch": 3.0202564579074522,
      "grad_norm": 0.8012350797653198,
      "learning_rate": 2.731389775425358e-05,
      "loss": 0.8897,
      "step": 16349
    },
    {
      "epoch": 3.020442296970823,
      "grad_norm": 0.803153932094574,
      "learning_rate": 2.73038650600148e-05,
      "loss": 0.6872,
      "step": 16350
    },
    {
      "epoch": 3.0206281360341944,
      "grad_norm": 1.1143475770950317,
      "learning_rate": 2.72938339173503e-05,
      "loss": 0.7349,
      "step": 16351
    },
    {
      "epoch": 3.0208139750975653,
      "grad_norm": 0.8050698637962341,
      "learning_rate": 2.7283804326474137e-05,
      "loss": 0.697,
      "step": 16352
    },
    {
      "epoch": 3.0209998141609367,
      "grad_norm": 0.8835048675537109,
      "learning_rate": 2.727377628760035e-05,
      "loss": 0.8201,
      "step": 16353
    },
    {
      "epoch": 3.0211856532243075,
      "grad_norm": 0.786395788192749,
      "learning_rate": 2.726374980094305e-05,
      "loss": 0.7965,
      "step": 16354
    },
    {
      "epoch": 3.021371492287679,
      "grad_norm": 0.8563681840896606,
      "learning_rate": 2.725372486671618e-05,
      "loss": 0.839,
      "step": 16355
    },
    {
      "epoch": 3.02155733135105,
      "grad_norm": 0.9203509092330933,
      "learning_rate": 2.724370148513369e-05,
      "loss": 0.603,
      "step": 16356
    },
    {
      "epoch": 3.021743170414421,
      "grad_norm": 0.8235127925872803,
      "learning_rate": 2.723367965640956e-05,
      "loss": 0.9054,
      "step": 16357
    },
    {
      "epoch": 3.0219290094777924,
      "grad_norm": 0.8137462735176086,
      "learning_rate": 2.722365938075767e-05,
      "loss": 0.9626,
      "step": 16358
    },
    {
      "epoch": 3.0221148485411633,
      "grad_norm": 0.9724891185760498,
      "learning_rate": 2.7213640658391882e-05,
      "loss": 0.9267,
      "step": 16359
    },
    {
      "epoch": 3.0223006876045346,
      "grad_norm": 0.9871930480003357,
      "learning_rate": 2.7203623489525996e-05,
      "loss": 0.8444,
      "step": 16360
    },
    {
      "epoch": 3.0224865266679055,
      "grad_norm": 0.7596043348312378,
      "learning_rate": 2.719360787437384e-05,
      "loss": 0.6035,
      "step": 16361
    },
    {
      "epoch": 3.022672365731277,
      "grad_norm": 0.8044273257255554,
      "learning_rate": 2.7183593813149223e-05,
      "loss": 0.9266,
      "step": 16362
    },
    {
      "epoch": 3.0228582047946477,
      "grad_norm": 0.8382756114006042,
      "learning_rate": 2.7173581306065833e-05,
      "loss": 1.0828,
      "step": 16363
    },
    {
      "epoch": 3.023044043858019,
      "grad_norm": 0.9325973987579346,
      "learning_rate": 2.7163570353337375e-05,
      "loss": 0.6733,
      "step": 16364
    },
    {
      "epoch": 3.02322988292139,
      "grad_norm": 0.8788731098175049,
      "learning_rate": 2.7153560955177483e-05,
      "loss": 0.9024,
      "step": 16365
    },
    {
      "epoch": 3.0234157219847613,
      "grad_norm": 0.9045056104660034,
      "learning_rate": 2.7143553111799845e-05,
      "loss": 1.0876,
      "step": 16366
    },
    {
      "epoch": 3.023601561048132,
      "grad_norm": 0.8674336075782776,
      "learning_rate": 2.7133546823418054e-05,
      "loss": 0.827,
      "step": 16367
    },
    {
      "epoch": 3.0237874001115035,
      "grad_norm": 0.7579058408737183,
      "learning_rate": 2.7123542090245625e-05,
      "loss": 0.7541,
      "step": 16368
    },
    {
      "epoch": 3.0239732391748744,
      "grad_norm": 0.7754246592521667,
      "learning_rate": 2.711353891249616e-05,
      "loss": 0.7431,
      "step": 16369
    },
    {
      "epoch": 3.0241590782382457,
      "grad_norm": 0.9136900305747986,
      "learning_rate": 2.710353729038313e-05,
      "loss": 0.9576,
      "step": 16370
    },
    {
      "epoch": 3.0243449173016166,
      "grad_norm": 0.8725305199623108,
      "learning_rate": 2.709353722411997e-05,
      "loss": 0.8939,
      "step": 16371
    },
    {
      "epoch": 3.024530756364988,
      "grad_norm": 0.9331547617912292,
      "learning_rate": 2.7083538713920174e-05,
      "loss": 0.6316,
      "step": 16372
    },
    {
      "epoch": 3.0247165954283592,
      "grad_norm": 0.8981873989105225,
      "learning_rate": 2.707354175999709e-05,
      "loss": 0.71,
      "step": 16373
    },
    {
      "epoch": 3.02490243449173,
      "grad_norm": 0.7866777181625366,
      "learning_rate": 2.706354636256414e-05,
      "loss": 0.7442,
      "step": 16374
    },
    {
      "epoch": 3.0250882735551015,
      "grad_norm": 0.9603359699249268,
      "learning_rate": 2.7053552521834624e-05,
      "loss": 0.9581,
      "step": 16375
    },
    {
      "epoch": 3.0252741126184723,
      "grad_norm": 1.1127327680587769,
      "learning_rate": 2.7043560238021825e-05,
      "loss": 0.7281,
      "step": 16376
    },
    {
      "epoch": 3.0254599516818437,
      "grad_norm": 0.8161758184432983,
      "learning_rate": 2.7033569511339062e-05,
      "loss": 0.8828,
      "step": 16377
    },
    {
      "epoch": 3.0256457907452146,
      "grad_norm": 0.9463825821876526,
      "learning_rate": 2.7023580341999532e-05,
      "loss": 1.1826,
      "step": 16378
    },
    {
      "epoch": 3.025831629808586,
      "grad_norm": 0.8008928298950195,
      "learning_rate": 2.7013592730216465e-05,
      "loss": 0.9116,
      "step": 16379
    },
    {
      "epoch": 3.0260174688719568,
      "grad_norm": 0.7768778204917908,
      "learning_rate": 2.7003606676202963e-05,
      "loss": 0.7778,
      "step": 16380
    },
    {
      "epoch": 3.026203307935328,
      "grad_norm": 0.8135813474655151,
      "learning_rate": 2.6993622180172218e-05,
      "loss": 0.6944,
      "step": 16381
    },
    {
      "epoch": 3.026389146998699,
      "grad_norm": 1.1667437553405762,
      "learning_rate": 2.6983639242337356e-05,
      "loss": 1.0562,
      "step": 16382
    },
    {
      "epoch": 3.0265749860620703,
      "grad_norm": 0.889934778213501,
      "learning_rate": 2.6973657862911418e-05,
      "loss": 0.8491,
      "step": 16383
    },
    {
      "epoch": 3.026760825125441,
      "grad_norm": 0.8528688549995422,
      "learning_rate": 2.696367804210742e-05,
      "loss": 0.7671,
      "step": 16384
    },
    {
      "epoch": 3.0269466641888125,
      "grad_norm": 0.9560061097145081,
      "learning_rate": 2.6953699780138364e-05,
      "loss": 0.8825,
      "step": 16385
    },
    {
      "epoch": 3.0271325032521834,
      "grad_norm": 0.8095105886459351,
      "learning_rate": 2.6943723077217252e-05,
      "loss": 0.7381,
      "step": 16386
    },
    {
      "epoch": 3.0273183423155547,
      "grad_norm": 0.9033812880516052,
      "learning_rate": 2.6933747933557008e-05,
      "loss": 0.866,
      "step": 16387
    },
    {
      "epoch": 3.0275041813789256,
      "grad_norm": 0.8631482720375061,
      "learning_rate": 2.692377434937049e-05,
      "loss": 0.8141,
      "step": 16388
    },
    {
      "epoch": 3.027690020442297,
      "grad_norm": 0.9113109111785889,
      "learning_rate": 2.691380232487064e-05,
      "loss": 0.9179,
      "step": 16389
    },
    {
      "epoch": 3.0278758595056683,
      "grad_norm": 0.7113003134727478,
      "learning_rate": 2.6903831860270256e-05,
      "loss": 0.6443,
      "step": 16390
    },
    {
      "epoch": 3.028061698569039,
      "grad_norm": 0.8057465553283691,
      "learning_rate": 2.6893862955782155e-05,
      "loss": 0.9655,
      "step": 16391
    },
    {
      "epoch": 3.0282475376324105,
      "grad_norm": 0.738502562046051,
      "learning_rate": 2.6883895611619047e-05,
      "loss": 0.5979,
      "step": 16392
    },
    {
      "epoch": 3.0284333766957814,
      "grad_norm": 0.9927711486816406,
      "learning_rate": 2.6873929827993727e-05,
      "loss": 0.8447,
      "step": 16393
    },
    {
      "epoch": 3.0286192157591527,
      "grad_norm": 0.7252799868583679,
      "learning_rate": 2.686396560511891e-05,
      "loss": 0.9221,
      "step": 16394
    },
    {
      "epoch": 3.0288050548225236,
      "grad_norm": 0.8324769139289856,
      "learning_rate": 2.6854002943207246e-05,
      "loss": 0.7464,
      "step": 16395
    },
    {
      "epoch": 3.028990893885895,
      "grad_norm": 0.8967441916465759,
      "learning_rate": 2.6844041842471324e-05,
      "loss": 0.9719,
      "step": 16396
    },
    {
      "epoch": 3.029176732949266,
      "grad_norm": 0.7824270725250244,
      "learning_rate": 2.683408230312383e-05,
      "loss": 0.7379,
      "step": 16397
    },
    {
      "epoch": 3.029362572012637,
      "grad_norm": 0.9299219846725464,
      "learning_rate": 2.6824124325377287e-05,
      "loss": 0.9068,
      "step": 16398
    },
    {
      "epoch": 3.029548411076008,
      "grad_norm": 0.9649413228034973,
      "learning_rate": 2.6814167909444233e-05,
      "loss": 0.7533,
      "step": 16399
    },
    {
      "epoch": 3.0297342501393794,
      "grad_norm": 1.058040738105774,
      "learning_rate": 2.680421305553714e-05,
      "loss": 0.8024,
      "step": 16400
    },
    {
      "epoch": 3.0299200892027502,
      "grad_norm": 0.8325514793395996,
      "learning_rate": 2.6794259763868555e-05,
      "loss": 0.9875,
      "step": 16401
    },
    {
      "epoch": 3.0301059282661216,
      "grad_norm": 0.8647835850715637,
      "learning_rate": 2.678430803465083e-05,
      "loss": 0.8065,
      "step": 16402
    },
    {
      "epoch": 3.0302917673294925,
      "grad_norm": 0.9864528775215149,
      "learning_rate": 2.6774357868096432e-05,
      "loss": 1.0541,
      "step": 16403
    },
    {
      "epoch": 3.030477606392864,
      "grad_norm": 0.8255147933959961,
      "learning_rate": 2.676440926441771e-05,
      "loss": 0.8324,
      "step": 16404
    },
    {
      "epoch": 3.030663445456235,
      "grad_norm": 0.9190768599510193,
      "learning_rate": 2.675446222382697e-05,
      "loss": 0.726,
      "step": 16405
    },
    {
      "epoch": 3.030849284519606,
      "grad_norm": 0.9106069207191467,
      "learning_rate": 2.6744516746536562e-05,
      "loss": 0.6332,
      "step": 16406
    },
    {
      "epoch": 3.0310351235829773,
      "grad_norm": 0.7660915851593018,
      "learning_rate": 2.673457283275873e-05,
      "loss": 0.7403,
      "step": 16407
    },
    {
      "epoch": 3.031220962646348,
      "grad_norm": 1.1061785221099854,
      "learning_rate": 2.6724630482705693e-05,
      "loss": 0.7549,
      "step": 16408
    },
    {
      "epoch": 3.0314068017097195,
      "grad_norm": 0.9986096620559692,
      "learning_rate": 2.671468969658969e-05,
      "loss": 0.9835,
      "step": 16409
    },
    {
      "epoch": 3.0315926407730904,
      "grad_norm": 0.8700722455978394,
      "learning_rate": 2.6704750474622886e-05,
      "loss": 0.7516,
      "step": 16410
    },
    {
      "epoch": 3.0317784798364618,
      "grad_norm": 0.7872580289840698,
      "learning_rate": 2.669481281701739e-05,
      "loss": 0.7264,
      "step": 16411
    },
    {
      "epoch": 3.0319643188998326,
      "grad_norm": 0.8272488117218018,
      "learning_rate": 2.6684876723985298e-05,
      "loss": 0.7823,
      "step": 16412
    },
    {
      "epoch": 3.032150157963204,
      "grad_norm": 0.7371950745582581,
      "learning_rate": 2.6674942195738695e-05,
      "loss": 0.7922,
      "step": 16413
    },
    {
      "epoch": 3.032335997026575,
      "grad_norm": 0.7064793705940247,
      "learning_rate": 2.6665009232489657e-05,
      "loss": 0.7347,
      "step": 16414
    },
    {
      "epoch": 3.032521836089946,
      "grad_norm": 0.8907574415206909,
      "learning_rate": 2.6655077834450136e-05,
      "loss": 0.9183,
      "step": 16415
    },
    {
      "epoch": 3.032707675153317,
      "grad_norm": 0.9381104111671448,
      "learning_rate": 2.6645148001832086e-05,
      "loss": 0.8648,
      "step": 16416
    },
    {
      "epoch": 3.0328935142166884,
      "grad_norm": 1.1486037969589233,
      "learning_rate": 2.6635219734847507e-05,
      "loss": 0.7935,
      "step": 16417
    },
    {
      "epoch": 3.0330793532800593,
      "grad_norm": 0.8872647285461426,
      "learning_rate": 2.6625293033708264e-05,
      "loss": 0.9706,
      "step": 16418
    },
    {
      "epoch": 3.0332651923434306,
      "grad_norm": 0.8415544629096985,
      "learning_rate": 2.661536789862622e-05,
      "loss": 0.8858,
      "step": 16419
    },
    {
      "epoch": 3.0334510314068015,
      "grad_norm": 0.8398575782775879,
      "learning_rate": 2.660544432981319e-05,
      "loss": 0.8452,
      "step": 16420
    },
    {
      "epoch": 3.033636870470173,
      "grad_norm": 0.8813350796699524,
      "learning_rate": 2.6595522327481027e-05,
      "loss": 0.717,
      "step": 16421
    },
    {
      "epoch": 3.033822709533544,
      "grad_norm": 0.9681750535964966,
      "learning_rate": 2.6585601891841472e-05,
      "loss": 1.1187,
      "step": 16422
    },
    {
      "epoch": 3.034008548596915,
      "grad_norm": 0.9663285613059998,
      "learning_rate": 2.6575683023106236e-05,
      "loss": 0.8364,
      "step": 16423
    },
    {
      "epoch": 3.0341943876602864,
      "grad_norm": 0.892166018486023,
      "learning_rate": 2.656576572148707e-05,
      "loss": 0.8615,
      "step": 16424
    },
    {
      "epoch": 3.0343802267236573,
      "grad_norm": 0.7774261832237244,
      "learning_rate": 2.655584998719558e-05,
      "loss": 0.9411,
      "step": 16425
    },
    {
      "epoch": 3.0345660657870286,
      "grad_norm": 0.9727977514266968,
      "learning_rate": 2.654593582044348e-05,
      "loss": 0.7237,
      "step": 16426
    },
    {
      "epoch": 3.0347519048503995,
      "grad_norm": 0.7590813636779785,
      "learning_rate": 2.6536023221442318e-05,
      "loss": 0.6202,
      "step": 16427
    },
    {
      "epoch": 3.034937743913771,
      "grad_norm": 1.093888282775879,
      "learning_rate": 2.652611219040365e-05,
      "loss": 0.712,
      "step": 16428
    },
    {
      "epoch": 3.0351235829771417,
      "grad_norm": 0.8715057969093323,
      "learning_rate": 2.6516202727539052e-05,
      "loss": 0.8436,
      "step": 16429
    },
    {
      "epoch": 3.035309422040513,
      "grad_norm": 0.8790238499641418,
      "learning_rate": 2.6506294833060008e-05,
      "loss": 0.7807,
      "step": 16430
    },
    {
      "epoch": 3.035495261103884,
      "grad_norm": 0.9124879837036133,
      "learning_rate": 2.6496388507177982e-05,
      "loss": 0.9362,
      "step": 16431
    },
    {
      "epoch": 3.0356811001672552,
      "grad_norm": 0.9696473479270935,
      "learning_rate": 2.6486483750104373e-05,
      "loss": 0.7244,
      "step": 16432
    },
    {
      "epoch": 3.035866939230626,
      "grad_norm": 0.836959958076477,
      "learning_rate": 2.6476580562050645e-05,
      "loss": 0.775,
      "step": 16433
    },
    {
      "epoch": 3.0360527782939974,
      "grad_norm": 0.9190247058868408,
      "learning_rate": 2.6466678943228106e-05,
      "loss": 0.7233,
      "step": 16434
    },
    {
      "epoch": 3.0362386173573683,
      "grad_norm": 1.5903725624084473,
      "learning_rate": 2.6456778893848144e-05,
      "loss": 1.0188,
      "step": 16435
    },
    {
      "epoch": 3.0364244564207397,
      "grad_norm": 0.8425945043563843,
      "learning_rate": 2.6446880414122e-05,
      "loss": 0.7854,
      "step": 16436
    },
    {
      "epoch": 3.0366102954841105,
      "grad_norm": 0.8605955243110657,
      "learning_rate": 2.6436983504261005e-05,
      "loss": 0.7152,
      "step": 16437
    },
    {
      "epoch": 3.036796134547482,
      "grad_norm": 1.015659213066101,
      "learning_rate": 2.6427088164476353e-05,
      "loss": 0.8773,
      "step": 16438
    },
    {
      "epoch": 3.036981973610853,
      "grad_norm": 0.7520237565040588,
      "learning_rate": 2.641719439497925e-05,
      "loss": 0.7434,
      "step": 16439
    },
    {
      "epoch": 3.037167812674224,
      "grad_norm": 0.9844590425491333,
      "learning_rate": 2.640730219598083e-05,
      "loss": 0.9365,
      "step": 16440
    },
    {
      "epoch": 3.0373536517375954,
      "grad_norm": 1.0298523902893066,
      "learning_rate": 2.639741156769229e-05,
      "loss": 0.8222,
      "step": 16441
    },
    {
      "epoch": 3.0375394908009663,
      "grad_norm": 0.746608555316925,
      "learning_rate": 2.638752251032468e-05,
      "loss": 0.8316,
      "step": 16442
    },
    {
      "epoch": 3.0377253298643376,
      "grad_norm": 0.9144389033317566,
      "learning_rate": 2.6377635024089087e-05,
      "loss": 0.5609,
      "step": 16443
    },
    {
      "epoch": 3.0379111689277085,
      "grad_norm": 1.2635074853897095,
      "learning_rate": 2.6367749109196505e-05,
      "loss": 0.6612,
      "step": 16444
    },
    {
      "epoch": 3.03809700799108,
      "grad_norm": 0.828485906124115,
      "learning_rate": 2.6357864765857953e-05,
      "loss": 0.5799,
      "step": 16445
    },
    {
      "epoch": 3.0382828470544507,
      "grad_norm": 1.4046440124511719,
      "learning_rate": 2.6347981994284442e-05,
      "loss": 0.9053,
      "step": 16446
    },
    {
      "epoch": 3.038468686117822,
      "grad_norm": 0.800578773021698,
      "learning_rate": 2.633810079468686e-05,
      "loss": 0.8299,
      "step": 16447
    },
    {
      "epoch": 3.038654525181193,
      "grad_norm": 0.8983428478240967,
      "learning_rate": 2.632822116727608e-05,
      "loss": 0.7084,
      "step": 16448
    },
    {
      "epoch": 3.0388403642445643,
      "grad_norm": 0.91361403465271,
      "learning_rate": 2.6318343112263012e-05,
      "loss": 0.6849,
      "step": 16449
    },
    {
      "epoch": 3.039026203307935,
      "grad_norm": 1.0825188159942627,
      "learning_rate": 2.6308466629858474e-05,
      "loss": 0.9871,
      "step": 16450
    },
    {
      "epoch": 3.0392120423713065,
      "grad_norm": 1.1262755393981934,
      "learning_rate": 2.629859172027326e-05,
      "loss": 0.733,
      "step": 16451
    },
    {
      "epoch": 3.0393978814346774,
      "grad_norm": 0.7709237933158875,
      "learning_rate": 2.6288718383718092e-05,
      "loss": 0.6211,
      "step": 16452
    },
    {
      "epoch": 3.0395837204980487,
      "grad_norm": 0.9766125082969666,
      "learning_rate": 2.6278846620403776e-05,
      "loss": 0.4818,
      "step": 16453
    },
    {
      "epoch": 3.03976955956142,
      "grad_norm": 0.9492712020874023,
      "learning_rate": 2.6268976430540958e-05,
      "loss": 0.858,
      "step": 16454
    },
    {
      "epoch": 3.039955398624791,
      "grad_norm": 0.9925112128257751,
      "learning_rate": 2.6259107814340287e-05,
      "loss": 0.6898,
      "step": 16455
    },
    {
      "epoch": 3.0401412376881622,
      "grad_norm": 0.9052692651748657,
      "learning_rate": 2.6249240772012417e-05,
      "loss": 0.7026,
      "step": 16456
    },
    {
      "epoch": 3.040327076751533,
      "grad_norm": 0.8264253735542297,
      "learning_rate": 2.6239375303767966e-05,
      "loss": 0.737,
      "step": 16457
    },
    {
      "epoch": 3.0405129158149045,
      "grad_norm": 0.9099843502044678,
      "learning_rate": 2.6229511409817474e-05,
      "loss": 0.7609,
      "step": 16458
    },
    {
      "epoch": 3.0406987548782753,
      "grad_norm": 0.8074800968170166,
      "learning_rate": 2.6219649090371466e-05,
      "loss": 0.7593,
      "step": 16459
    },
    {
      "epoch": 3.0408845939416467,
      "grad_norm": 1.121508240699768,
      "learning_rate": 2.620978834564041e-05,
      "loss": 0.9156,
      "step": 16460
    },
    {
      "epoch": 3.0410704330050176,
      "grad_norm": 0.8222777843475342,
      "learning_rate": 2.6199929175834813e-05,
      "loss": 0.8353,
      "step": 16461
    },
    {
      "epoch": 3.041256272068389,
      "grad_norm": 1.0044645071029663,
      "learning_rate": 2.6190071581165086e-05,
      "loss": 0.732,
      "step": 16462
    },
    {
      "epoch": 3.0414421111317598,
      "grad_norm": 1.001706600189209,
      "learning_rate": 2.6180215561841614e-05,
      "loss": 0.9493,
      "step": 16463
    },
    {
      "epoch": 3.041627950195131,
      "grad_norm": 0.9047880172729492,
      "learning_rate": 2.617036111807474e-05,
      "loss": 0.8127,
      "step": 16464
    },
    {
      "epoch": 3.041813789258502,
      "grad_norm": 0.8721327185630798,
      "learning_rate": 2.61605082500748e-05,
      "loss": 0.8758,
      "step": 16465
    },
    {
      "epoch": 3.0419996283218733,
      "grad_norm": 1.8152014017105103,
      "learning_rate": 2.6150656958052132e-05,
      "loss": 0.9119,
      "step": 16466
    },
    {
      "epoch": 3.042185467385244,
      "grad_norm": 0.8496637940406799,
      "learning_rate": 2.614080724221697e-05,
      "loss": 0.6909,
      "step": 16467
    },
    {
      "epoch": 3.0423713064486155,
      "grad_norm": 0.9403654932975769,
      "learning_rate": 2.6130959102779495e-05,
      "loss": 0.891,
      "step": 16468
    },
    {
      "epoch": 3.0425571455119864,
      "grad_norm": 0.9045391082763672,
      "learning_rate": 2.6121112539949954e-05,
      "loss": 0.7195,
      "step": 16469
    },
    {
      "epoch": 3.0427429845753577,
      "grad_norm": 0.8190896511077881,
      "learning_rate": 2.6111267553938502e-05,
      "loss": 0.777,
      "step": 16470
    },
    {
      "epoch": 3.042928823638729,
      "grad_norm": 0.892221212387085,
      "learning_rate": 2.6101424144955235e-05,
      "loss": 0.8231,
      "step": 16471
    },
    {
      "epoch": 3.0431146627021,
      "grad_norm": 0.8593822717666626,
      "learning_rate": 2.6091582313210216e-05,
      "loss": 0.8822,
      "step": 16472
    },
    {
      "epoch": 3.0433005017654713,
      "grad_norm": 0.7897011637687683,
      "learning_rate": 2.6081742058913584e-05,
      "loss": 0.6229,
      "step": 16473
    },
    {
      "epoch": 3.043486340828842,
      "grad_norm": 0.9321253299713135,
      "learning_rate": 2.607190338227532e-05,
      "loss": 0.8131,
      "step": 16474
    },
    {
      "epoch": 3.0436721798922135,
      "grad_norm": 0.8023005723953247,
      "learning_rate": 2.6062066283505403e-05,
      "loss": 0.6842,
      "step": 16475
    },
    {
      "epoch": 3.0438580189555844,
      "grad_norm": 1.1247652769088745,
      "learning_rate": 2.6052230762813745e-05,
      "loss": 0.8034,
      "step": 16476
    },
    {
      "epoch": 3.0440438580189557,
      "grad_norm": 0.7327141165733337,
      "learning_rate": 2.6042396820410375e-05,
      "loss": 0.5958,
      "step": 16477
    },
    {
      "epoch": 3.0442296970823266,
      "grad_norm": 1.011448621749878,
      "learning_rate": 2.6032564456505138e-05,
      "loss": 0.9633,
      "step": 16478
    },
    {
      "epoch": 3.044415536145698,
      "grad_norm": 0.7816374897956848,
      "learning_rate": 2.6022733671307876e-05,
      "loss": 0.691,
      "step": 16479
    },
    {
      "epoch": 3.044601375209069,
      "grad_norm": 0.9759527444839478,
      "learning_rate": 2.6012904465028376e-05,
      "loss": 0.8437,
      "step": 16480
    },
    {
      "epoch": 3.04478721427244,
      "grad_norm": 0.7570684552192688,
      "learning_rate": 2.6003076837876507e-05,
      "loss": 0.7463,
      "step": 16481
    },
    {
      "epoch": 3.044973053335811,
      "grad_norm": 0.9799579977989197,
      "learning_rate": 2.5993250790061973e-05,
      "loss": 0.8084,
      "step": 16482
    },
    {
      "epoch": 3.0451588923991824,
      "grad_norm": 1.0092289447784424,
      "learning_rate": 2.5983426321794502e-05,
      "loss": 0.9419,
      "step": 16483
    },
    {
      "epoch": 3.0453447314625532,
      "grad_norm": 0.918048620223999,
      "learning_rate": 2.5973603433283746e-05,
      "loss": 1.0072,
      "step": 16484
    },
    {
      "epoch": 3.0455305705259246,
      "grad_norm": 1.0498931407928467,
      "learning_rate": 2.596378212473941e-05,
      "loss": 0.9556,
      "step": 16485
    },
    {
      "epoch": 3.0457164095892955,
      "grad_norm": 0.6764087080955505,
      "learning_rate": 2.595396239637108e-05,
      "loss": 0.6806,
      "step": 16486
    },
    {
      "epoch": 3.045902248652667,
      "grad_norm": 0.8425265550613403,
      "learning_rate": 2.594414424838838e-05,
      "loss": 0.8192,
      "step": 16487
    },
    {
      "epoch": 3.046088087716038,
      "grad_norm": 0.8224844932556152,
      "learning_rate": 2.5934327681000802e-05,
      "loss": 0.8188,
      "step": 16488
    },
    {
      "epoch": 3.046273926779409,
      "grad_norm": 0.7876110076904297,
      "learning_rate": 2.5924512694417924e-05,
      "loss": 0.7812,
      "step": 16489
    },
    {
      "epoch": 3.0464597658427803,
      "grad_norm": 0.8870657086372375,
      "learning_rate": 2.591469928884922e-05,
      "loss": 0.6744,
      "step": 16490
    },
    {
      "epoch": 3.046645604906151,
      "grad_norm": 0.7624846696853638,
      "learning_rate": 2.5904887464504114e-05,
      "loss": 0.711,
      "step": 16491
    },
    {
      "epoch": 3.0468314439695225,
      "grad_norm": 1.009589672088623,
      "learning_rate": 2.5895077221592e-05,
      "loss": 1.0146,
      "step": 16492
    },
    {
      "epoch": 3.0470172830328934,
      "grad_norm": 0.8577511310577393,
      "learning_rate": 2.588526856032233e-05,
      "loss": 0.8643,
      "step": 16493
    },
    {
      "epoch": 3.0472031220962648,
      "grad_norm": 0.9337430596351624,
      "learning_rate": 2.5875461480904406e-05,
      "loss": 0.8704,
      "step": 16494
    },
    {
      "epoch": 3.0473889611596356,
      "grad_norm": 0.801527738571167,
      "learning_rate": 2.5865655983547566e-05,
      "loss": 0.7768,
      "step": 16495
    },
    {
      "epoch": 3.047574800223007,
      "grad_norm": 0.9945794343948364,
      "learning_rate": 2.585585206846104e-05,
      "loss": 0.8735,
      "step": 16496
    },
    {
      "epoch": 3.047760639286378,
      "grad_norm": 0.8158237338066101,
      "learning_rate": 2.5846049735854117e-05,
      "loss": 0.6968,
      "step": 16497
    },
    {
      "epoch": 3.047946478349749,
      "grad_norm": 0.779066264629364,
      "learning_rate": 2.583624898593605e-05,
      "loss": 0.7956,
      "step": 16498
    },
    {
      "epoch": 3.04813231741312,
      "grad_norm": 0.9122100472450256,
      "learning_rate": 2.5826449818915964e-05,
      "loss": 0.7876,
      "step": 16499
    },
    {
      "epoch": 3.0483181564764914,
      "grad_norm": 1.0245736837387085,
      "learning_rate": 2.5816652235003e-05,
      "loss": 0.7925,
      "step": 16500
    },
    {
      "epoch": 3.0485039955398623,
      "grad_norm": 0.6653180122375488,
      "learning_rate": 2.5806856234406317e-05,
      "loss": 0.457,
      "step": 16501
    },
    {
      "epoch": 3.0486898346032336,
      "grad_norm": 0.9029264450073242,
      "learning_rate": 2.5797061817334967e-05,
      "loss": 0.8034,
      "step": 16502
    },
    {
      "epoch": 3.0488756736666045,
      "grad_norm": 0.8435755372047424,
      "learning_rate": 2.578726898399799e-05,
      "loss": 0.8431,
      "step": 16503
    },
    {
      "epoch": 3.049061512729976,
      "grad_norm": 0.9914752244949341,
      "learning_rate": 2.5777477734604383e-05,
      "loss": 0.9909,
      "step": 16504
    },
    {
      "epoch": 3.049247351793347,
      "grad_norm": 0.776179313659668,
      "learning_rate": 2.576768806936316e-05,
      "loss": 0.7043,
      "step": 16505
    },
    {
      "epoch": 3.049433190856718,
      "grad_norm": 1.0241563320159912,
      "learning_rate": 2.575789998848326e-05,
      "loss": 0.7818,
      "step": 16506
    },
    {
      "epoch": 3.0496190299200894,
      "grad_norm": 0.928352952003479,
      "learning_rate": 2.5748113492173543e-05,
      "loss": 0.9997,
      "step": 16507
    },
    {
      "epoch": 3.0498048689834603,
      "grad_norm": 0.8682227730751038,
      "learning_rate": 2.5738328580642922e-05,
      "loss": 0.9907,
      "step": 16508
    },
    {
      "epoch": 3.0499907080468316,
      "grad_norm": 1.1242249011993408,
      "learning_rate": 2.5728545254100268e-05,
      "loss": 0.728,
      "step": 16509
    },
    {
      "epoch": 3.0501765471102025,
      "grad_norm": 0.9435965418815613,
      "learning_rate": 2.5718763512754363e-05,
      "loss": 0.8136,
      "step": 16510
    },
    {
      "epoch": 3.050362386173574,
      "grad_norm": 0.8476418852806091,
      "learning_rate": 2.5708983356813975e-05,
      "loss": 0.6715,
      "step": 16511
    },
    {
      "epoch": 3.0505482252369447,
      "grad_norm": 0.9711516499519348,
      "learning_rate": 2.569920478648783e-05,
      "loss": 0.7907,
      "step": 16512
    },
    {
      "epoch": 3.050734064300316,
      "grad_norm": 0.8876985907554626,
      "learning_rate": 2.5689427801984678e-05,
      "loss": 0.9151,
      "step": 16513
    },
    {
      "epoch": 3.050919903363687,
      "grad_norm": 0.8093453049659729,
      "learning_rate": 2.5679652403513167e-05,
      "loss": 0.6496,
      "step": 16514
    },
    {
      "epoch": 3.0511057424270582,
      "grad_norm": 1.0686590671539307,
      "learning_rate": 2.5669878591281926e-05,
      "loss": 0.775,
      "step": 16515
    },
    {
      "epoch": 3.051291581490429,
      "grad_norm": 0.7788293361663818,
      "learning_rate": 2.566010636549956e-05,
      "loss": 0.8973,
      "step": 16516
    },
    {
      "epoch": 3.0514774205538004,
      "grad_norm": 0.8514882922172546,
      "learning_rate": 2.5650335726374662e-05,
      "loss": 0.8554,
      "step": 16517
    },
    {
      "epoch": 3.0516632596171713,
      "grad_norm": 0.8660567998886108,
      "learning_rate": 2.5640566674115742e-05,
      "loss": 0.9811,
      "step": 16518
    },
    {
      "epoch": 3.0518490986805427,
      "grad_norm": 1.5931254625320435,
      "learning_rate": 2.5630799208931345e-05,
      "loss": 1.1427,
      "step": 16519
    },
    {
      "epoch": 3.0520349377439135,
      "grad_norm": 1.0193164348602295,
      "learning_rate": 2.56210333310299e-05,
      "loss": 0.9796,
      "step": 16520
    },
    {
      "epoch": 3.052220776807285,
      "grad_norm": 0.851655125617981,
      "learning_rate": 2.5611269040619867e-05,
      "loss": 0.6453,
      "step": 16521
    },
    {
      "epoch": 3.052406615870656,
      "grad_norm": 0.8646299242973328,
      "learning_rate": 2.5601506337909654e-05,
      "loss": 0.8345,
      "step": 16522
    },
    {
      "epoch": 3.052592454934027,
      "grad_norm": 0.863435685634613,
      "learning_rate": 2.559174522310761e-05,
      "loss": 0.8992,
      "step": 16523
    },
    {
      "epoch": 3.0527782939973984,
      "grad_norm": 0.7195289134979248,
      "learning_rate": 2.5581985696422052e-05,
      "loss": 0.545,
      "step": 16524
    },
    {
      "epoch": 3.0529641330607693,
      "grad_norm": 0.9412771463394165,
      "learning_rate": 2.5572227758061328e-05,
      "loss": 0.7509,
      "step": 16525
    },
    {
      "epoch": 3.0531499721241406,
      "grad_norm": 0.9601781964302063,
      "learning_rate": 2.556247140823368e-05,
      "loss": 0.8078,
      "step": 16526
    },
    {
      "epoch": 3.0533358111875115,
      "grad_norm": 0.9324098229408264,
      "learning_rate": 2.55527166471473e-05,
      "loss": 0.7121,
      "step": 16527
    },
    {
      "epoch": 3.053521650250883,
      "grad_norm": 0.7718344926834106,
      "learning_rate": 2.554296347501044e-05,
      "loss": 0.8115,
      "step": 16528
    },
    {
      "epoch": 3.0537074893142537,
      "grad_norm": 0.7441035509109497,
      "learning_rate": 2.5533211892031273e-05,
      "loss": 0.931,
      "step": 16529
    },
    {
      "epoch": 3.053893328377625,
      "grad_norm": 0.9605115056037903,
      "learning_rate": 2.552346189841791e-05,
      "loss": 0.8943,
      "step": 16530
    },
    {
      "epoch": 3.054079167440996,
      "grad_norm": 0.7816125154495239,
      "learning_rate": 2.5513713494378455e-05,
      "loss": 0.6336,
      "step": 16531
    },
    {
      "epoch": 3.0542650065043673,
      "grad_norm": 0.792016863822937,
      "learning_rate": 2.5503966680120916e-05,
      "loss": 0.7136,
      "step": 16532
    },
    {
      "epoch": 3.054450845567738,
      "grad_norm": 0.877753496170044,
      "learning_rate": 2.5494221455853408e-05,
      "loss": 0.596,
      "step": 16533
    },
    {
      "epoch": 3.0546366846311095,
      "grad_norm": 0.9530837535858154,
      "learning_rate": 2.548447782178388e-05,
      "loss": 0.9308,
      "step": 16534
    },
    {
      "epoch": 3.0548225236944804,
      "grad_norm": 0.9567283987998962,
      "learning_rate": 2.5474735778120306e-05,
      "loss": 0.827,
      "step": 16535
    },
    {
      "epoch": 3.0550083627578517,
      "grad_norm": 1.026564359664917,
      "learning_rate": 2.546499532507057e-05,
      "loss": 0.8805,
      "step": 16536
    },
    {
      "epoch": 3.055194201821223,
      "grad_norm": 0.7885897755622864,
      "learning_rate": 2.5455256462842638e-05,
      "loss": 1.095,
      "step": 16537
    },
    {
      "epoch": 3.055380040884594,
      "grad_norm": 0.815269410610199,
      "learning_rate": 2.544551919164433e-05,
      "loss": 0.7272,
      "step": 16538
    },
    {
      "epoch": 3.0555658799479652,
      "grad_norm": 0.9533278942108154,
      "learning_rate": 2.5435783511683443e-05,
      "loss": 0.8395,
      "step": 16539
    },
    {
      "epoch": 3.055751719011336,
      "grad_norm": 0.7652965784072876,
      "learning_rate": 2.542604942316781e-05,
      "loss": 0.8543,
      "step": 16540
    },
    {
      "epoch": 3.0559375580747075,
      "grad_norm": 0.8190576434135437,
      "learning_rate": 2.5416316926305195e-05,
      "loss": 0.6892,
      "step": 16541
    },
    {
      "epoch": 3.0561233971380783,
      "grad_norm": 1.004726529121399,
      "learning_rate": 2.5406586021303302e-05,
      "loss": 1.1388,
      "step": 16542
    },
    {
      "epoch": 3.0563092362014497,
      "grad_norm": 0.9118068218231201,
      "learning_rate": 2.5396856708369832e-05,
      "loss": 0.7668,
      "step": 16543
    },
    {
      "epoch": 3.0564950752648206,
      "grad_norm": 1.0773011445999146,
      "learning_rate": 2.5387128987712404e-05,
      "loss": 0.9329,
      "step": 16544
    },
    {
      "epoch": 3.056680914328192,
      "grad_norm": 0.9110521674156189,
      "learning_rate": 2.537740285953869e-05,
      "loss": 0.9904,
      "step": 16545
    },
    {
      "epoch": 3.0568667533915628,
      "grad_norm": 0.871599018573761,
      "learning_rate": 2.5367678324056253e-05,
      "loss": 0.5775,
      "step": 16546
    },
    {
      "epoch": 3.057052592454934,
      "grad_norm": 0.8627876043319702,
      "learning_rate": 2.535795538147262e-05,
      "loss": 0.8143,
      "step": 16547
    },
    {
      "epoch": 3.057238431518305,
      "grad_norm": 0.7532731294631958,
      "learning_rate": 2.5348234031995377e-05,
      "loss": 0.7186,
      "step": 16548
    },
    {
      "epoch": 3.0574242705816763,
      "grad_norm": 0.8290095329284668,
      "learning_rate": 2.5338514275831925e-05,
      "loss": 0.6237,
      "step": 16549
    },
    {
      "epoch": 3.057610109645047,
      "grad_norm": 0.8998545408248901,
      "learning_rate": 2.5328796113189802e-05,
      "loss": 0.7936,
      "step": 16550
    },
    {
      "epoch": 3.0577959487084185,
      "grad_norm": 0.8460700511932373,
      "learning_rate": 2.531907954427638e-05,
      "loss": 0.919,
      "step": 16551
    },
    {
      "epoch": 3.0579817877717894,
      "grad_norm": 0.9167879819869995,
      "learning_rate": 2.530936456929902e-05,
      "loss": 1.0441,
      "step": 16552
    },
    {
      "epoch": 3.0581676268351607,
      "grad_norm": 0.8317403197288513,
      "learning_rate": 2.5299651188465136e-05,
      "loss": 0.8466,
      "step": 16553
    },
    {
      "epoch": 3.058353465898532,
      "grad_norm": 0.8379568457603455,
      "learning_rate": 2.5289939401982e-05,
      "loss": 0.8954,
      "step": 16554
    },
    {
      "epoch": 3.058539304961903,
      "grad_norm": 0.7911828756332397,
      "learning_rate": 2.5280229210056895e-05,
      "loss": 0.8179,
      "step": 16555
    },
    {
      "epoch": 3.0587251440252743,
      "grad_norm": 0.9084653854370117,
      "learning_rate": 2.527052061289705e-05,
      "loss": 0.8155,
      "step": 16556
    },
    {
      "epoch": 3.058910983088645,
      "grad_norm": 0.8726891875267029,
      "learning_rate": 2.526081361070973e-05,
      "loss": 0.6489,
      "step": 16557
    },
    {
      "epoch": 3.0590968221520165,
      "grad_norm": 0.8376065492630005,
      "learning_rate": 2.525110820370209e-05,
      "loss": 0.7411,
      "step": 16558
    },
    {
      "epoch": 3.0592826612153874,
      "grad_norm": 0.8209545016288757,
      "learning_rate": 2.5241404392081237e-05,
      "loss": 0.7611,
      "step": 16559
    },
    {
      "epoch": 3.0594685002787587,
      "grad_norm": 0.8267958164215088,
      "learning_rate": 2.5231702176054317e-05,
      "loss": 0.6398,
      "step": 16560
    },
    {
      "epoch": 3.0596543393421296,
      "grad_norm": 0.9265934228897095,
      "learning_rate": 2.5222001555828434e-05,
      "loss": 0.9597,
      "step": 16561
    },
    {
      "epoch": 3.059840178405501,
      "grad_norm": 0.832126259803772,
      "learning_rate": 2.5212302531610598e-05,
      "loss": 0.7867,
      "step": 16562
    },
    {
      "epoch": 3.060026017468872,
      "grad_norm": 0.8703610897064209,
      "learning_rate": 2.5202605103607835e-05,
      "loss": 0.911,
      "step": 16563
    },
    {
      "epoch": 3.060211856532243,
      "grad_norm": 0.7699049711227417,
      "learning_rate": 2.5192909272027076e-05,
      "loss": 0.959,
      "step": 16564
    },
    {
      "epoch": 3.060397695595614,
      "grad_norm": 0.9440488815307617,
      "learning_rate": 2.5183215037075324e-05,
      "loss": 0.7962,
      "step": 16565
    },
    {
      "epoch": 3.0605835346589854,
      "grad_norm": 0.8586769104003906,
      "learning_rate": 2.517352239895946e-05,
      "loss": 0.8593,
      "step": 16566
    },
    {
      "epoch": 3.0607693737223562,
      "grad_norm": 0.7995127439498901,
      "learning_rate": 2.5163831357886314e-05,
      "loss": 0.6925,
      "step": 16567
    },
    {
      "epoch": 3.0609552127857276,
      "grad_norm": 0.7316787242889404,
      "learning_rate": 2.51541419140628e-05,
      "loss": 0.8178,
      "step": 16568
    },
    {
      "epoch": 3.0611410518490985,
      "grad_norm": 0.8505775928497314,
      "learning_rate": 2.5144454067695677e-05,
      "loss": 0.8638,
      "step": 16569
    },
    {
      "epoch": 3.06132689091247,
      "grad_norm": 0.7964285612106323,
      "learning_rate": 2.513476781899171e-05,
      "loss": 0.925,
      "step": 16570
    },
    {
      "epoch": 3.061512729975841,
      "grad_norm": 1.5267374515533447,
      "learning_rate": 2.512508316815767e-05,
      "loss": 1.072,
      "step": 16571
    },
    {
      "epoch": 3.061698569039212,
      "grad_norm": 0.8420001864433289,
      "learning_rate": 2.5115400115400213e-05,
      "loss": 0.8227,
      "step": 16572
    },
    {
      "epoch": 3.0618844081025833,
      "grad_norm": 0.9002656936645508,
      "learning_rate": 2.5105718660926058e-05,
      "loss": 0.808,
      "step": 16573
    },
    {
      "epoch": 3.062070247165954,
      "grad_norm": 0.9324301481246948,
      "learning_rate": 2.5096038804941812e-05,
      "loss": 0.6955,
      "step": 16574
    },
    {
      "epoch": 3.0622560862293255,
      "grad_norm": 0.7720271944999695,
      "learning_rate": 2.5086360547654087e-05,
      "loss": 0.7693,
      "step": 16575
    },
    {
      "epoch": 3.0624419252926964,
      "grad_norm": 1.048789620399475,
      "learning_rate": 2.5076683889269405e-05,
      "loss": 0.9154,
      "step": 16576
    },
    {
      "epoch": 3.0626277643560678,
      "grad_norm": 0.9712361097335815,
      "learning_rate": 2.5067008829994365e-05,
      "loss": 0.8935,
      "step": 16577
    },
    {
      "epoch": 3.0628136034194386,
      "grad_norm": 0.9016991257667542,
      "learning_rate": 2.5057335370035427e-05,
      "loss": 0.7707,
      "step": 16578
    },
    {
      "epoch": 3.06299944248281,
      "grad_norm": 0.869917631149292,
      "learning_rate": 2.504766350959903e-05,
      "loss": 0.792,
      "step": 16579
    },
    {
      "epoch": 3.063185281546181,
      "grad_norm": 0.7795620560646057,
      "learning_rate": 2.5037993248891667e-05,
      "loss": 0.8157,
      "step": 16580
    },
    {
      "epoch": 3.063371120609552,
      "grad_norm": 0.8983386754989624,
      "learning_rate": 2.5028324588119656e-05,
      "loss": 0.8888,
      "step": 16581
    },
    {
      "epoch": 3.063556959672923,
      "grad_norm": 0.8761674165725708,
      "learning_rate": 2.5018657527489443e-05,
      "loss": 0.7019,
      "step": 16582
    },
    {
      "epoch": 3.0637427987362944,
      "grad_norm": 0.8494465351104736,
      "learning_rate": 2.5008992067207303e-05,
      "loss": 0.6808,
      "step": 16583
    },
    {
      "epoch": 3.0639286377996653,
      "grad_norm": 0.8243811726570129,
      "learning_rate": 2.499932820747951e-05,
      "loss": 0.8011,
      "step": 16584
    },
    {
      "epoch": 3.0641144768630366,
      "grad_norm": 3.3844828605651855,
      "learning_rate": 2.4989665948512387e-05,
      "loss": 1.0565,
      "step": 16585
    },
    {
      "epoch": 3.064300315926408,
      "grad_norm": 0.9450771808624268,
      "learning_rate": 2.498000529051211e-05,
      "loss": 1.0121,
      "step": 16586
    },
    {
      "epoch": 3.064486154989779,
      "grad_norm": 0.8206013441085815,
      "learning_rate": 2.497034623368486e-05,
      "loss": 0.7842,
      "step": 16587
    },
    {
      "epoch": 3.06467199405315,
      "grad_norm": 0.8935214281082153,
      "learning_rate": 2.496068877823684e-05,
      "loss": 0.6792,
      "step": 16588
    },
    {
      "epoch": 3.064857833116521,
      "grad_norm": 0.9215742945671082,
      "learning_rate": 2.495103292437415e-05,
      "loss": 0.9773,
      "step": 16589
    },
    {
      "epoch": 3.0650436721798924,
      "grad_norm": 0.9969448447227478,
      "learning_rate": 2.4941378672302862e-05,
      "loss": 0.8248,
      "step": 16590
    },
    {
      "epoch": 3.0652295112432633,
      "grad_norm": 1.2787046432495117,
      "learning_rate": 2.493172602222902e-05,
      "loss": 0.8893,
      "step": 16591
    },
    {
      "epoch": 3.0654153503066346,
      "grad_norm": 0.8934160470962524,
      "learning_rate": 2.4922074974358657e-05,
      "loss": 0.7708,
      "step": 16592
    },
    {
      "epoch": 3.0656011893700055,
      "grad_norm": 0.8716941475868225,
      "learning_rate": 2.4912425528897798e-05,
      "loss": 0.9923,
      "step": 16593
    },
    {
      "epoch": 3.065787028433377,
      "grad_norm": 0.911169707775116,
      "learning_rate": 2.490277768605237e-05,
      "loss": 0.7935,
      "step": 16594
    },
    {
      "epoch": 3.0659728674967477,
      "grad_norm": 0.9753298759460449,
      "learning_rate": 2.489313144602826e-05,
      "loss": 0.8333,
      "step": 16595
    },
    {
      "epoch": 3.066158706560119,
      "grad_norm": 0.9423926472663879,
      "learning_rate": 2.4883486809031354e-05,
      "loss": 0.7983,
      "step": 16596
    },
    {
      "epoch": 3.06634454562349,
      "grad_norm": 0.8739957213401794,
      "learning_rate": 2.4873843775267545e-05,
      "loss": 0.9515,
      "step": 16597
    },
    {
      "epoch": 3.0665303846868612,
      "grad_norm": 0.9837615489959717,
      "learning_rate": 2.486420234494262e-05,
      "loss": 0.7153,
      "step": 16598
    },
    {
      "epoch": 3.066716223750232,
      "grad_norm": 0.8995093107223511,
      "learning_rate": 2.4854562518262327e-05,
      "loss": 0.7964,
      "step": 16599
    },
    {
      "epoch": 3.0669020628136034,
      "grad_norm": 0.8065444827079773,
      "learning_rate": 2.484492429543247e-05,
      "loss": 0.5506,
      "step": 16600
    },
    {
      "epoch": 3.0670879018769743,
      "grad_norm": 0.8899737596511841,
      "learning_rate": 2.4835287676658735e-05,
      "loss": 0.7886,
      "step": 16601
    },
    {
      "epoch": 3.0672737409403457,
      "grad_norm": 0.8818018436431885,
      "learning_rate": 2.482565266214677e-05,
      "loss": 0.7062,
      "step": 16602
    },
    {
      "epoch": 3.067459580003717,
      "grad_norm": 1.153317928314209,
      "learning_rate": 2.4816019252102273e-05,
      "loss": 0.9368,
      "step": 16603
    },
    {
      "epoch": 3.067645419067088,
      "grad_norm": 0.8451851010322571,
      "learning_rate": 2.4806387446730795e-05,
      "loss": 0.6308,
      "step": 16604
    },
    {
      "epoch": 3.067831258130459,
      "grad_norm": 1.00625479221344,
      "learning_rate": 2.479675724623797e-05,
      "loss": 0.8113,
      "step": 16605
    },
    {
      "epoch": 3.06801709719383,
      "grad_norm": 1.0481326580047607,
      "learning_rate": 2.4787128650829293e-05,
      "loss": 0.8003,
      "step": 16606
    },
    {
      "epoch": 3.0682029362572014,
      "grad_norm": 0.8781499266624451,
      "learning_rate": 2.4777501660710267e-05,
      "loss": 0.9148,
      "step": 16607
    },
    {
      "epoch": 3.0683887753205723,
      "grad_norm": 1.05372154712677,
      "learning_rate": 2.476787627608641e-05,
      "loss": 0.8686,
      "step": 16608
    },
    {
      "epoch": 3.0685746143839436,
      "grad_norm": 0.9408324956893921,
      "learning_rate": 2.4758252497163137e-05,
      "loss": 0.6715,
      "step": 16609
    },
    {
      "epoch": 3.0687604534473145,
      "grad_norm": 1.0712614059448242,
      "learning_rate": 2.474863032414584e-05,
      "loss": 0.7837,
      "step": 16610
    },
    {
      "epoch": 3.068946292510686,
      "grad_norm": 0.7135563492774963,
      "learning_rate": 2.473900975723985e-05,
      "loss": 0.7282,
      "step": 16611
    },
    {
      "epoch": 3.0691321315740567,
      "grad_norm": 0.9954936504364014,
      "learning_rate": 2.4729390796650564e-05,
      "loss": 0.6655,
      "step": 16612
    },
    {
      "epoch": 3.069317970637428,
      "grad_norm": 0.8331687450408936,
      "learning_rate": 2.4719773442583273e-05,
      "loss": 0.9261,
      "step": 16613
    },
    {
      "epoch": 3.069503809700799,
      "grad_norm": 1.0044938325881958,
      "learning_rate": 2.4710157695243242e-05,
      "loss": 0.8095,
      "step": 16614
    },
    {
      "epoch": 3.0696896487641703,
      "grad_norm": 1.2021206617355347,
      "learning_rate": 2.4700543554835697e-05,
      "loss": 1.1148,
      "step": 16615
    },
    {
      "epoch": 3.069875487827541,
      "grad_norm": 0.7989461421966553,
      "learning_rate": 2.469093102156579e-05,
      "loss": 0.8558,
      "step": 16616
    },
    {
      "epoch": 3.0700613268909125,
      "grad_norm": 0.9222882986068726,
      "learning_rate": 2.468132009563876e-05,
      "loss": 0.732,
      "step": 16617
    },
    {
      "epoch": 3.0702471659542834,
      "grad_norm": 0.9197480082511902,
      "learning_rate": 2.4671710777259703e-05,
      "loss": 0.8164,
      "step": 16618
    },
    {
      "epoch": 3.0704330050176547,
      "grad_norm": 0.8911687731742859,
      "learning_rate": 2.466210306663368e-05,
      "loss": 0.9387,
      "step": 16619
    },
    {
      "epoch": 3.070618844081026,
      "grad_norm": 0.8025472164154053,
      "learning_rate": 2.46524969639658e-05,
      "loss": 0.7469,
      "step": 16620
    },
    {
      "epoch": 3.070804683144397,
      "grad_norm": 0.9530259966850281,
      "learning_rate": 2.4642892469461088e-05,
      "loss": 1.0753,
      "step": 16621
    },
    {
      "epoch": 3.0709905222077682,
      "grad_norm": 0.8317549228668213,
      "learning_rate": 2.4633289583324503e-05,
      "loss": 0.8232,
      "step": 16622
    },
    {
      "epoch": 3.071176361271139,
      "grad_norm": 1.006614327430725,
      "learning_rate": 2.4623688305761005e-05,
      "loss": 0.8355,
      "step": 16623
    },
    {
      "epoch": 3.0713622003345105,
      "grad_norm": 0.7379666566848755,
      "learning_rate": 2.4614088636975518e-05,
      "loss": 0.6482,
      "step": 16624
    },
    {
      "epoch": 3.0715480393978813,
      "grad_norm": 1.028715968132019,
      "learning_rate": 2.4604490577172967e-05,
      "loss": 0.7598,
      "step": 16625
    },
    {
      "epoch": 3.0717338784612527,
      "grad_norm": 0.7389920353889465,
      "learning_rate": 2.459489412655819e-05,
      "loss": 0.6548,
      "step": 16626
    },
    {
      "epoch": 3.0719197175246236,
      "grad_norm": 0.9339531064033508,
      "learning_rate": 2.458529928533596e-05,
      "loss": 0.8555,
      "step": 16627
    },
    {
      "epoch": 3.072105556587995,
      "grad_norm": 1.1104316711425781,
      "learning_rate": 2.4575706053711134e-05,
      "loss": 1.0854,
      "step": 16628
    },
    {
      "epoch": 3.0722913956513658,
      "grad_norm": 0.7918559312820435,
      "learning_rate": 2.4566114431888432e-05,
      "loss": 0.7041,
      "step": 16629
    },
    {
      "epoch": 3.072477234714737,
      "grad_norm": 0.9732975959777832,
      "learning_rate": 2.455652442007257e-05,
      "loss": 0.6959,
      "step": 16630
    },
    {
      "epoch": 3.072663073778108,
      "grad_norm": 0.9053291082382202,
      "learning_rate": 2.454693601846819e-05,
      "loss": 0.8479,
      "step": 16631
    },
    {
      "epoch": 3.0728489128414793,
      "grad_norm": 0.800689160823822,
      "learning_rate": 2.4537349227280014e-05,
      "loss": 0.8431,
      "step": 16632
    },
    {
      "epoch": 3.07303475190485,
      "grad_norm": 0.9349857568740845,
      "learning_rate": 2.4527764046712587e-05,
      "loss": 0.8068,
      "step": 16633
    },
    {
      "epoch": 3.0732205909682215,
      "grad_norm": 1.96149742603302,
      "learning_rate": 2.4518180476970555e-05,
      "loss": 1.4771,
      "step": 16634
    },
    {
      "epoch": 3.073406430031593,
      "grad_norm": 0.8421056270599365,
      "learning_rate": 2.450859851825842e-05,
      "loss": 0.865,
      "step": 16635
    },
    {
      "epoch": 3.0735922690949637,
      "grad_norm": 0.8871978521347046,
      "learning_rate": 2.4499018170780673e-05,
      "loss": 0.8629,
      "step": 16636
    },
    {
      "epoch": 3.073778108158335,
      "grad_norm": 0.8812477588653564,
      "learning_rate": 2.4489439434741855e-05,
      "loss": 0.9481,
      "step": 16637
    },
    {
      "epoch": 3.073963947221706,
      "grad_norm": 0.890427827835083,
      "learning_rate": 2.447986231034636e-05,
      "loss": 0.8507,
      "step": 16638
    },
    {
      "epoch": 3.0741497862850773,
      "grad_norm": 0.9563420414924622,
      "learning_rate": 2.447028679779859e-05,
      "loss": 1.0243,
      "step": 16639
    },
    {
      "epoch": 3.074335625348448,
      "grad_norm": 0.8652082681655884,
      "learning_rate": 2.446071289730295e-05,
      "loss": 0.7468,
      "step": 16640
    },
    {
      "epoch": 3.0745214644118195,
      "grad_norm": 0.9163973331451416,
      "learning_rate": 2.445114060906377e-05,
      "loss": 0.8505,
      "step": 16641
    },
    {
      "epoch": 3.0747073034751904,
      "grad_norm": 0.8107301592826843,
      "learning_rate": 2.4441569933285345e-05,
      "loss": 0.6438,
      "step": 16642
    },
    {
      "epoch": 3.0748931425385617,
      "grad_norm": 0.8502079844474792,
      "learning_rate": 2.443200087017192e-05,
      "loss": 0.7834,
      "step": 16643
    },
    {
      "epoch": 3.0750789816019326,
      "grad_norm": 1.0325417518615723,
      "learning_rate": 2.442243341992776e-05,
      "loss": 0.802,
      "step": 16644
    },
    {
      "epoch": 3.075264820665304,
      "grad_norm": 0.7983143925666809,
      "learning_rate": 2.4412867582757083e-05,
      "loss": 0.8331,
      "step": 16645
    },
    {
      "epoch": 3.075450659728675,
      "grad_norm": 0.9426150321960449,
      "learning_rate": 2.4403303358864037e-05,
      "loss": 0.8964,
      "step": 16646
    },
    {
      "epoch": 3.075636498792046,
      "grad_norm": 1.0253843069076538,
      "learning_rate": 2.439374074845272e-05,
      "loss": 0.907,
      "step": 16647
    },
    {
      "epoch": 3.075822337855417,
      "grad_norm": 0.8924049139022827,
      "learning_rate": 2.43841797517273e-05,
      "loss": 0.9502,
      "step": 16648
    },
    {
      "epoch": 3.0760081769187884,
      "grad_norm": 1.0296672582626343,
      "learning_rate": 2.43746203688918e-05,
      "loss": 0.9336,
      "step": 16649
    },
    {
      "epoch": 3.0761940159821592,
      "grad_norm": 1.3383376598358154,
      "learning_rate": 2.4365062600150234e-05,
      "loss": 0.8846,
      "step": 16650
    },
    {
      "epoch": 3.0763798550455306,
      "grad_norm": 0.789011538028717,
      "learning_rate": 2.43555064457066e-05,
      "loss": 0.8331,
      "step": 16651
    },
    {
      "epoch": 3.076565694108902,
      "grad_norm": 0.9009502530097961,
      "learning_rate": 2.4345951905764897e-05,
      "loss": 0.7493,
      "step": 16652
    },
    {
      "epoch": 3.076751533172273,
      "grad_norm": 0.8864755034446716,
      "learning_rate": 2.4336398980529017e-05,
      "loss": 0.8703,
      "step": 16653
    },
    {
      "epoch": 3.076937372235644,
      "grad_norm": 0.9883394837379456,
      "learning_rate": 2.4326847670202823e-05,
      "loss": 0.9677,
      "step": 16654
    },
    {
      "epoch": 3.077123211299015,
      "grad_norm": 1.1275808811187744,
      "learning_rate": 2.431729797499025e-05,
      "loss": 0.8323,
      "step": 16655
    },
    {
      "epoch": 3.0773090503623863,
      "grad_norm": 0.9039409756660461,
      "learning_rate": 2.430774989509503e-05,
      "loss": 0.9529,
      "step": 16656
    },
    {
      "epoch": 3.077494889425757,
      "grad_norm": 0.9433819651603699,
      "learning_rate": 2.4298203430721033e-05,
      "loss": 0.8014,
      "step": 16657
    },
    {
      "epoch": 3.0776807284891285,
      "grad_norm": 0.8646581172943115,
      "learning_rate": 2.4288658582071977e-05,
      "loss": 0.6687,
      "step": 16658
    },
    {
      "epoch": 3.0778665675524994,
      "grad_norm": 0.8743041753768921,
      "learning_rate": 2.4279115349351543e-05,
      "loss": 0.8719,
      "step": 16659
    },
    {
      "epoch": 3.0780524066158708,
      "grad_norm": 0.8597197532653809,
      "learning_rate": 2.4269573732763475e-05,
      "loss": 0.8298,
      "step": 16660
    },
    {
      "epoch": 3.0782382456792416,
      "grad_norm": 0.7918018102645874,
      "learning_rate": 2.42600337325114e-05,
      "loss": 0.8337,
      "step": 16661
    },
    {
      "epoch": 3.078424084742613,
      "grad_norm": 0.7828530669212341,
      "learning_rate": 2.425049534879894e-05,
      "loss": 0.6693,
      "step": 16662
    },
    {
      "epoch": 3.078609923805984,
      "grad_norm": 0.894785463809967,
      "learning_rate": 2.424095858182962e-05,
      "loss": 0.8944,
      "step": 16663
    },
    {
      "epoch": 3.078795762869355,
      "grad_norm": 1.0368988513946533,
      "learning_rate": 2.4231423431807076e-05,
      "loss": 0.819,
      "step": 16664
    },
    {
      "epoch": 3.078981601932726,
      "grad_norm": 0.82077956199646,
      "learning_rate": 2.4221889898934734e-05,
      "loss": 0.6261,
      "step": 16665
    },
    {
      "epoch": 3.0791674409960974,
      "grad_norm": 0.8031370639801025,
      "learning_rate": 2.4212357983416146e-05,
      "loss": 0.9246,
      "step": 16666
    },
    {
      "epoch": 3.0793532800594683,
      "grad_norm": 0.8558308482170105,
      "learning_rate": 2.420282768545469e-05,
      "loss": 1.0473,
      "step": 16667
    },
    {
      "epoch": 3.0795391191228396,
      "grad_norm": 0.8579029440879822,
      "learning_rate": 2.4193299005253832e-05,
      "loss": 0.8786,
      "step": 16668
    },
    {
      "epoch": 3.079724958186211,
      "grad_norm": 0.8373103737831116,
      "learning_rate": 2.418377194301692e-05,
      "loss": 0.7825,
      "step": 16669
    },
    {
      "epoch": 3.079910797249582,
      "grad_norm": 0.8694421052932739,
      "learning_rate": 2.4174246498947284e-05,
      "loss": 0.7338,
      "step": 16670
    },
    {
      "epoch": 3.080096636312953,
      "grad_norm": 0.9413045644760132,
      "learning_rate": 2.4164722673248207e-05,
      "loss": 0.8307,
      "step": 16671
    },
    {
      "epoch": 3.080282475376324,
      "grad_norm": 0.713191568851471,
      "learning_rate": 2.415520046612302e-05,
      "loss": 0.5577,
      "step": 16672
    },
    {
      "epoch": 3.0804683144396954,
      "grad_norm": 0.9494543671607971,
      "learning_rate": 2.4145679877774928e-05,
      "loss": 0.7002,
      "step": 16673
    },
    {
      "epoch": 3.0806541535030663,
      "grad_norm": 1.0250353813171387,
      "learning_rate": 2.4136160908407113e-05,
      "loss": 0.8437,
      "step": 16674
    },
    {
      "epoch": 3.0808399925664376,
      "grad_norm": 0.9267086982727051,
      "learning_rate": 2.412664355822274e-05,
      "loss": 0.7667,
      "step": 16675
    },
    {
      "epoch": 3.0810258316298085,
      "grad_norm": 0.8907784223556519,
      "learning_rate": 2.4117127827424958e-05,
      "loss": 0.6792,
      "step": 16676
    },
    {
      "epoch": 3.08121167069318,
      "grad_norm": 0.7296506762504578,
      "learning_rate": 2.4107613716216882e-05,
      "loss": 0.9244,
      "step": 16677
    },
    {
      "epoch": 3.0813975097565507,
      "grad_norm": 0.789673388004303,
      "learning_rate": 2.409810122480155e-05,
      "loss": 0.6373,
      "step": 16678
    },
    {
      "epoch": 3.081583348819922,
      "grad_norm": 0.9214822053909302,
      "learning_rate": 2.408859035338198e-05,
      "loss": 0.8431,
      "step": 16679
    },
    {
      "epoch": 3.081769187883293,
      "grad_norm": 0.8458185791969299,
      "learning_rate": 2.4079081102161195e-05,
      "loss": 0.8627,
      "step": 16680
    },
    {
      "epoch": 3.0819550269466642,
      "grad_norm": 0.8897072076797485,
      "learning_rate": 2.4069573471342143e-05,
      "loss": 0.7229,
      "step": 16681
    },
    {
      "epoch": 3.082140866010035,
      "grad_norm": 0.9710602164268494,
      "learning_rate": 2.4060067461127755e-05,
      "loss": 0.5108,
      "step": 16682
    },
    {
      "epoch": 3.0823267050734064,
      "grad_norm": 0.9847256541252136,
      "learning_rate": 2.4050563071720867e-05,
      "loss": 0.623,
      "step": 16683
    },
    {
      "epoch": 3.0825125441367778,
      "grad_norm": 0.8316210508346558,
      "learning_rate": 2.4041060303324402e-05,
      "loss": 0.7586,
      "step": 16684
    },
    {
      "epoch": 3.0826983832001487,
      "grad_norm": 1.0670188665390015,
      "learning_rate": 2.403155915614117e-05,
      "loss": 0.8198,
      "step": 16685
    },
    {
      "epoch": 3.08288422226352,
      "grad_norm": 0.89921635389328,
      "learning_rate": 2.4022059630373896e-05,
      "loss": 0.8845,
      "step": 16686
    },
    {
      "epoch": 3.083070061326891,
      "grad_norm": 0.8962400555610657,
      "learning_rate": 2.401256172622538e-05,
      "loss": 0.9516,
      "step": 16687
    },
    {
      "epoch": 3.083255900390262,
      "grad_norm": 0.8482634425163269,
      "learning_rate": 2.4003065443898365e-05,
      "loss": 0.8793,
      "step": 16688
    },
    {
      "epoch": 3.083441739453633,
      "grad_norm": 1.0588843822479248,
      "learning_rate": 2.399357078359551e-05,
      "loss": 0.8523,
      "step": 16689
    },
    {
      "epoch": 3.0836275785170044,
      "grad_norm": 0.8351184725761414,
      "learning_rate": 2.398407774551944e-05,
      "loss": 0.8478,
      "step": 16690
    },
    {
      "epoch": 3.0838134175803753,
      "grad_norm": 0.8612925410270691,
      "learning_rate": 2.397458632987276e-05,
      "loss": 0.7712,
      "step": 16691
    },
    {
      "epoch": 3.0839992566437466,
      "grad_norm": 1.1683576107025146,
      "learning_rate": 2.39650965368581e-05,
      "loss": 0.8221,
      "step": 16692
    },
    {
      "epoch": 3.0841850957071175,
      "grad_norm": 0.9445899128913879,
      "learning_rate": 2.3955608366677974e-05,
      "loss": 0.8417,
      "step": 16693
    },
    {
      "epoch": 3.084370934770489,
      "grad_norm": 0.690139651298523,
      "learning_rate": 2.3946121819534885e-05,
      "loss": 0.6781,
      "step": 16694
    },
    {
      "epoch": 3.0845567738338597,
      "grad_norm": 0.7465941905975342,
      "learning_rate": 2.3936636895631292e-05,
      "loss": 0.7303,
      "step": 16695
    },
    {
      "epoch": 3.084742612897231,
      "grad_norm": 1.1478382349014282,
      "learning_rate": 2.392715359516965e-05,
      "loss": 0.7927,
      "step": 16696
    },
    {
      "epoch": 3.084928451960602,
      "grad_norm": 0.8253434300422668,
      "learning_rate": 2.391767191835239e-05,
      "loss": 0.7372,
      "step": 16697
    },
    {
      "epoch": 3.0851142910239733,
      "grad_norm": 0.8607159852981567,
      "learning_rate": 2.3908191865381868e-05,
      "loss": 1.0916,
      "step": 16698
    },
    {
      "epoch": 3.085300130087344,
      "grad_norm": 0.8785241842269897,
      "learning_rate": 2.3898713436460375e-05,
      "loss": 0.7048,
      "step": 16699
    },
    {
      "epoch": 3.0854859691507155,
      "grad_norm": 1.2133129835128784,
      "learning_rate": 2.388923663179029e-05,
      "loss": 1.0641,
      "step": 16700
    },
    {
      "epoch": 3.0856718082140864,
      "grad_norm": 0.7245477437973022,
      "learning_rate": 2.3879761451573835e-05,
      "loss": 0.797,
      "step": 16701
    },
    {
      "epoch": 3.0858576472774577,
      "grad_norm": 0.95304274559021,
      "learning_rate": 2.3870287896013232e-05,
      "loss": 0.6126,
      "step": 16702
    },
    {
      "epoch": 3.086043486340829,
      "grad_norm": 0.9752274751663208,
      "learning_rate": 2.3860815965310668e-05,
      "loss": 0.9041,
      "step": 16703
    },
    {
      "epoch": 3.0862293254042,
      "grad_norm": 0.9736945629119873,
      "learning_rate": 2.3851345659668368e-05,
      "loss": 0.8335,
      "step": 16704
    },
    {
      "epoch": 3.0864151644675712,
      "grad_norm": 1.0160160064697266,
      "learning_rate": 2.38418769792884e-05,
      "loss": 1.1237,
      "step": 16705
    },
    {
      "epoch": 3.086601003530942,
      "grad_norm": 0.8461440205574036,
      "learning_rate": 2.383240992437289e-05,
      "loss": 0.6576,
      "step": 16706
    },
    {
      "epoch": 3.0867868425943135,
      "grad_norm": 0.9372764825820923,
      "learning_rate": 2.382294449512381e-05,
      "loss": 0.8287,
      "step": 16707
    },
    {
      "epoch": 3.0869726816576843,
      "grad_norm": 1.056541085243225,
      "learning_rate": 2.3813480691743328e-05,
      "loss": 0.8241,
      "step": 16708
    },
    {
      "epoch": 3.0871585207210557,
      "grad_norm": 0.9410994648933411,
      "learning_rate": 2.3804018514433345e-05,
      "loss": 0.963,
      "step": 16709
    },
    {
      "epoch": 3.0873443597844266,
      "grad_norm": 1.1573671102523804,
      "learning_rate": 2.3794557963395836e-05,
      "loss": 0.7073,
      "step": 16710
    },
    {
      "epoch": 3.087530198847798,
      "grad_norm": 0.9475613236427307,
      "learning_rate": 2.378509903883268e-05,
      "loss": 0.8321,
      "step": 16711
    },
    {
      "epoch": 3.0877160379111688,
      "grad_norm": 1.05289626121521,
      "learning_rate": 2.377564174094583e-05,
      "loss": 0.8266,
      "step": 16712
    },
    {
      "epoch": 3.08790187697454,
      "grad_norm": 0.9490292072296143,
      "learning_rate": 2.3766186069937103e-05,
      "loss": 0.9654,
      "step": 16713
    },
    {
      "epoch": 3.088087716037911,
      "grad_norm": 0.788969874382019,
      "learning_rate": 2.37567320260083e-05,
      "loss": 0.8875,
      "step": 16714
    },
    {
      "epoch": 3.0882735551012823,
      "grad_norm": 0.9245768189430237,
      "learning_rate": 2.3747279609361196e-05,
      "loss": 0.642,
      "step": 16715
    },
    {
      "epoch": 3.088459394164653,
      "grad_norm": 0.9661019444465637,
      "learning_rate": 2.3737828820197583e-05,
      "loss": 0.8747,
      "step": 16716
    },
    {
      "epoch": 3.0886452332280245,
      "grad_norm": 0.8161077499389648,
      "learning_rate": 2.3728379658719102e-05,
      "loss": 0.8601,
      "step": 16717
    },
    {
      "epoch": 3.088831072291396,
      "grad_norm": 0.9887533187866211,
      "learning_rate": 2.3718932125127514e-05,
      "loss": 0.7007,
      "step": 16718
    },
    {
      "epoch": 3.0890169113547667,
      "grad_norm": 0.9117093086242676,
      "learning_rate": 2.3709486219624377e-05,
      "loss": 0.8345,
      "step": 16719
    },
    {
      "epoch": 3.089202750418138,
      "grad_norm": 0.8773633241653442,
      "learning_rate": 2.3700041942411367e-05,
      "loss": 0.8651,
      "step": 16720
    },
    {
      "epoch": 3.089388589481509,
      "grad_norm": 0.9503114819526672,
      "learning_rate": 2.369059929369003e-05,
      "loss": 0.9321,
      "step": 16721
    },
    {
      "epoch": 3.0895744285448803,
      "grad_norm": 0.8335660696029663,
      "learning_rate": 2.3681158273661896e-05,
      "loss": 0.5858,
      "step": 16722
    },
    {
      "epoch": 3.089760267608251,
      "grad_norm": 0.8890185952186584,
      "learning_rate": 2.3671718882528437e-05,
      "loss": 0.7505,
      "step": 16723
    },
    {
      "epoch": 3.0899461066716225,
      "grad_norm": 0.9524630904197693,
      "learning_rate": 2.3662281120491182e-05,
      "loss": 0.6554,
      "step": 16724
    },
    {
      "epoch": 3.0901319457349934,
      "grad_norm": 0.8818069696426392,
      "learning_rate": 2.3652844987751542e-05,
      "loss": 0.7038,
      "step": 16725
    },
    {
      "epoch": 3.0903177847983647,
      "grad_norm": 1.1302887201309204,
      "learning_rate": 2.3643410484510908e-05,
      "loss": 1.0679,
      "step": 16726
    },
    {
      "epoch": 3.0905036238617356,
      "grad_norm": 0.8042498230934143,
      "learning_rate": 2.3633977610970616e-05,
      "loss": 0.886,
      "step": 16727
    },
    {
      "epoch": 3.090689462925107,
      "grad_norm": 1.2275042533874512,
      "learning_rate": 2.3624546367332022e-05,
      "loss": 0.6698,
      "step": 16728
    },
    {
      "epoch": 3.090875301988478,
      "grad_norm": 0.8624935150146484,
      "learning_rate": 2.361511675379645e-05,
      "loss": 0.9283,
      "step": 16729
    },
    {
      "epoch": 3.091061141051849,
      "grad_norm": 0.8328248262405396,
      "learning_rate": 2.3605688770565127e-05,
      "loss": 0.7019,
      "step": 16730
    },
    {
      "epoch": 3.09124698011522,
      "grad_norm": 1.3361327648162842,
      "learning_rate": 2.3596262417839255e-05,
      "loss": 0.9344,
      "step": 16731
    },
    {
      "epoch": 3.0914328191785914,
      "grad_norm": 1.0902270078659058,
      "learning_rate": 2.3586837695820064e-05,
      "loss": 0.822,
      "step": 16732
    },
    {
      "epoch": 3.0916186582419622,
      "grad_norm": 0.9357127547264099,
      "learning_rate": 2.3577414604708704e-05,
      "loss": 0.7635,
      "step": 16733
    },
    {
      "epoch": 3.0918044973053336,
      "grad_norm": 0.9868106245994568,
      "learning_rate": 2.3567993144706278e-05,
      "loss": 0.9043,
      "step": 16734
    },
    {
      "epoch": 3.091990336368705,
      "grad_norm": 1.0099672079086304,
      "learning_rate": 2.3558573316013845e-05,
      "loss": 0.8065,
      "step": 16735
    },
    {
      "epoch": 3.092176175432076,
      "grad_norm": 0.8899115920066833,
      "learning_rate": 2.354915511883252e-05,
      "loss": 0.6263,
      "step": 16736
    },
    {
      "epoch": 3.092362014495447,
      "grad_norm": 0.9453414678573608,
      "learning_rate": 2.353973855336329e-05,
      "loss": 0.7882,
      "step": 16737
    },
    {
      "epoch": 3.092547853558818,
      "grad_norm": 0.9318817257881165,
      "learning_rate": 2.3530323619807083e-05,
      "loss": 0.9171,
      "step": 16738
    },
    {
      "epoch": 3.0927336926221893,
      "grad_norm": 1.0196037292480469,
      "learning_rate": 2.3520910318364907e-05,
      "loss": 1.0691,
      "step": 16739
    },
    {
      "epoch": 3.09291953168556,
      "grad_norm": 0.9235552549362183,
      "learning_rate": 2.3511498649237674e-05,
      "loss": 1.0074,
      "step": 16740
    },
    {
      "epoch": 3.0931053707489315,
      "grad_norm": 0.9589453935623169,
      "learning_rate": 2.350208861262625e-05,
      "loss": 0.8637,
      "step": 16741
    },
    {
      "epoch": 3.0932912098123024,
      "grad_norm": 1.0189878940582275,
      "learning_rate": 2.3492680208731464e-05,
      "loss": 0.8796,
      "step": 16742
    },
    {
      "epoch": 3.0934770488756738,
      "grad_norm": 0.9348618388175964,
      "learning_rate": 2.3483273437754107e-05,
      "loss": 0.9578,
      "step": 16743
    },
    {
      "epoch": 3.0936628879390446,
      "grad_norm": 0.9557995200157166,
      "learning_rate": 2.3473868299894998e-05,
      "loss": 0.7415,
      "step": 16744
    },
    {
      "epoch": 3.093848727002416,
      "grad_norm": 0.7726553678512573,
      "learning_rate": 2.3464464795354834e-05,
      "loss": 0.7415,
      "step": 16745
    },
    {
      "epoch": 3.094034566065787,
      "grad_norm": 0.8676804304122925,
      "learning_rate": 2.3455062924334337e-05,
      "loss": 0.9098,
      "step": 16746
    },
    {
      "epoch": 3.094220405129158,
      "grad_norm": 1.062719464302063,
      "learning_rate": 2.3445662687034144e-05,
      "loss": 0.8672,
      "step": 16747
    },
    {
      "epoch": 3.094406244192529,
      "grad_norm": 0.8634195327758789,
      "learning_rate": 2.3436264083654925e-05,
      "loss": 0.7621,
      "step": 16748
    },
    {
      "epoch": 3.0945920832559004,
      "grad_norm": 0.8746082782745361,
      "learning_rate": 2.342686711439723e-05,
      "loss": 0.9445,
      "step": 16749
    },
    {
      "epoch": 3.0947779223192713,
      "grad_norm": 0.9249411821365356,
      "learning_rate": 2.3417471779461685e-05,
      "loss": 0.9334,
      "step": 16750
    },
    {
      "epoch": 3.0949637613826426,
      "grad_norm": 0.9423879981040955,
      "learning_rate": 2.3408078079048767e-05,
      "loss": 1.103,
      "step": 16751
    },
    {
      "epoch": 3.095149600446014,
      "grad_norm": 0.8189881443977356,
      "learning_rate": 2.3398686013358996e-05,
      "loss": 0.848,
      "step": 16752
    },
    {
      "epoch": 3.095335439509385,
      "grad_norm": 1.0123745203018188,
      "learning_rate": 2.3389295582592828e-05,
      "loss": 0.7669,
      "step": 16753
    },
    {
      "epoch": 3.095521278572756,
      "grad_norm": 0.9926055669784546,
      "learning_rate": 2.3379906786950677e-05,
      "loss": 0.793,
      "step": 16754
    },
    {
      "epoch": 3.095707117636127,
      "grad_norm": 0.7395652532577515,
      "learning_rate": 2.3370519626632902e-05,
      "loss": 0.7581,
      "step": 16755
    },
    {
      "epoch": 3.0958929566994984,
      "grad_norm": 0.826623797416687,
      "learning_rate": 2.336113410183991e-05,
      "loss": 0.7201,
      "step": 16756
    },
    {
      "epoch": 3.0960787957628693,
      "grad_norm": 0.7846941947937012,
      "learning_rate": 2.3351750212771995e-05,
      "loss": 0.7227,
      "step": 16757
    },
    {
      "epoch": 3.0962646348262406,
      "grad_norm": 0.9971020817756653,
      "learning_rate": 2.3342367959629408e-05,
      "loss": 0.9792,
      "step": 16758
    },
    {
      "epoch": 3.0964504738896115,
      "grad_norm": 1.027601718902588,
      "learning_rate": 2.3332987342612423e-05,
      "loss": 0.8237,
      "step": 16759
    },
    {
      "epoch": 3.096636312952983,
      "grad_norm": 0.7158558964729309,
      "learning_rate": 2.332360836192129e-05,
      "loss": 0.7387,
      "step": 16760
    },
    {
      "epoch": 3.0968221520163537,
      "grad_norm": 0.8798509240150452,
      "learning_rate": 2.331423101775616e-05,
      "loss": 0.8709,
      "step": 16761
    },
    {
      "epoch": 3.097007991079725,
      "grad_norm": 0.866066575050354,
      "learning_rate": 2.3304855310317166e-05,
      "loss": 0.7117,
      "step": 16762
    },
    {
      "epoch": 3.097193830143096,
      "grad_norm": 0.8405870795249939,
      "learning_rate": 2.3295481239804384e-05,
      "loss": 0.739,
      "step": 16763
    },
    {
      "epoch": 3.0973796692064672,
      "grad_norm": 1.0137507915496826,
      "learning_rate": 2.328610880641797e-05,
      "loss": 1.0033,
      "step": 16764
    },
    {
      "epoch": 3.097565508269838,
      "grad_norm": 0.8602619171142578,
      "learning_rate": 2.3276738010357903e-05,
      "loss": 0.8401,
      "step": 16765
    },
    {
      "epoch": 3.0977513473332094,
      "grad_norm": 1.2083443403244019,
      "learning_rate": 2.3267368851824212e-05,
      "loss": 0.8599,
      "step": 16766
    },
    {
      "epoch": 3.0979371863965808,
      "grad_norm": 1.0404247045516968,
      "learning_rate": 2.325800133101682e-05,
      "loss": 0.7361,
      "step": 16767
    },
    {
      "epoch": 3.0981230254599517,
      "grad_norm": 0.777709424495697,
      "learning_rate": 2.3248635448135725e-05,
      "loss": 0.7952,
      "step": 16768
    },
    {
      "epoch": 3.098308864523323,
      "grad_norm": 0.9634824991226196,
      "learning_rate": 2.3239271203380797e-05,
      "loss": 0.8092,
      "step": 16769
    },
    {
      "epoch": 3.098494703586694,
      "grad_norm": 0.8834127187728882,
      "learning_rate": 2.3229908596951865e-05,
      "loss": 0.9975,
      "step": 16770
    },
    {
      "epoch": 3.098680542650065,
      "grad_norm": 1.0853376388549805,
      "learning_rate": 2.3220547629048796e-05,
      "loss": 0.9208,
      "step": 16771
    },
    {
      "epoch": 3.098866381713436,
      "grad_norm": 0.8186860084533691,
      "learning_rate": 2.32111882998714e-05,
      "loss": 0.8202,
      "step": 16772
    },
    {
      "epoch": 3.0990522207768074,
      "grad_norm": 0.86734539270401,
      "learning_rate": 2.3201830609619425e-05,
      "loss": 0.8744,
      "step": 16773
    },
    {
      "epoch": 3.0992380598401783,
      "grad_norm": 1.0710790157318115,
      "learning_rate": 2.319247455849257e-05,
      "loss": 0.8189,
      "step": 16774
    },
    {
      "epoch": 3.0994238989035496,
      "grad_norm": 1.0556211471557617,
      "learning_rate": 2.3183120146690528e-05,
      "loss": 0.7788,
      "step": 16775
    },
    {
      "epoch": 3.0996097379669205,
      "grad_norm": 1.0326217412948608,
      "learning_rate": 2.317376737441298e-05,
      "loss": 0.8049,
      "step": 16776
    },
    {
      "epoch": 3.099795577030292,
      "grad_norm": 1.3107826709747314,
      "learning_rate": 2.3164416241859533e-05,
      "loss": 1.2447,
      "step": 16777
    },
    {
      "epoch": 3.0999814160936627,
      "grad_norm": 0.8404430150985718,
      "learning_rate": 2.3155066749229737e-05,
      "loss": 0.6428,
      "step": 16778
    },
    {
      "epoch": 3.100167255157034,
      "grad_norm": 0.903167188167572,
      "learning_rate": 2.31457188967232e-05,
      "loss": 0.8984,
      "step": 16779
    },
    {
      "epoch": 3.100353094220405,
      "grad_norm": 0.7048614025115967,
      "learning_rate": 2.3136372684539374e-05,
      "loss": 0.6329,
      "step": 16780
    },
    {
      "epoch": 3.1005389332837763,
      "grad_norm": 1.206588864326477,
      "learning_rate": 2.3127028112877803e-05,
      "loss": 1.0657,
      "step": 16781
    },
    {
      "epoch": 3.100724772347147,
      "grad_norm": 0.9022656679153442,
      "learning_rate": 2.31176851819379e-05,
      "loss": 0.847,
      "step": 16782
    },
    {
      "epoch": 3.1009106114105185,
      "grad_norm": 1.3397260904312134,
      "learning_rate": 2.3108343891919037e-05,
      "loss": 1.1737,
      "step": 16783
    },
    {
      "epoch": 3.10109645047389,
      "grad_norm": 0.836466372013092,
      "learning_rate": 2.309900424302065e-05,
      "loss": 0.751,
      "step": 16784
    },
    {
      "epoch": 3.1012822895372607,
      "grad_norm": 1.0850869417190552,
      "learning_rate": 2.3089666235442054e-05,
      "loss": 0.8397,
      "step": 16785
    },
    {
      "epoch": 3.101468128600632,
      "grad_norm": 0.8306275010108948,
      "learning_rate": 2.3080329869382544e-05,
      "loss": 0.7889,
      "step": 16786
    },
    {
      "epoch": 3.101653967664003,
      "grad_norm": 0.8751552700996399,
      "learning_rate": 2.3070995145041364e-05,
      "loss": 0.6941,
      "step": 16787
    },
    {
      "epoch": 3.1018398067273742,
      "grad_norm": 1.119346022605896,
      "learning_rate": 2.3061662062617804e-05,
      "loss": 0.9556,
      "step": 16788
    },
    {
      "epoch": 3.102025645790745,
      "grad_norm": 1.1482000350952148,
      "learning_rate": 2.3052330622311046e-05,
      "loss": 0.8685,
      "step": 16789
    },
    {
      "epoch": 3.1022114848541165,
      "grad_norm": 0.9722250699996948,
      "learning_rate": 2.3043000824320216e-05,
      "loss": 0.9658,
      "step": 16790
    },
    {
      "epoch": 3.1023973239174873,
      "grad_norm": 0.9512292146682739,
      "learning_rate": 2.3033672668844465e-05,
      "loss": 1.0272,
      "step": 16791
    },
    {
      "epoch": 3.1025831629808587,
      "grad_norm": 0.9728047847747803,
      "learning_rate": 2.302434615608291e-05,
      "loss": 0.9121,
      "step": 16792
    },
    {
      "epoch": 3.1027690020442296,
      "grad_norm": 0.9288976192474365,
      "learning_rate": 2.3015021286234605e-05,
      "loss": 0.9073,
      "step": 16793
    },
    {
      "epoch": 3.102954841107601,
      "grad_norm": 0.9850062727928162,
      "learning_rate": 2.300569805949855e-05,
      "loss": 0.7848,
      "step": 16794
    },
    {
      "epoch": 3.1031406801709718,
      "grad_norm": 0.9823880791664124,
      "learning_rate": 2.2996376476073723e-05,
      "loss": 0.9452,
      "step": 16795
    },
    {
      "epoch": 3.103326519234343,
      "grad_norm": 0.9922701716423035,
      "learning_rate": 2.298705653615911e-05,
      "loss": 1.1013,
      "step": 16796
    },
    {
      "epoch": 3.103512358297714,
      "grad_norm": 1.0710746049880981,
      "learning_rate": 2.297773823995363e-05,
      "loss": 0.7522,
      "step": 16797
    },
    {
      "epoch": 3.1036981973610853,
      "grad_norm": 0.9887570738792419,
      "learning_rate": 2.296842158765612e-05,
      "loss": 0.5986,
      "step": 16798
    },
    {
      "epoch": 3.103884036424456,
      "grad_norm": 0.8032146096229553,
      "learning_rate": 2.2959106579465483e-05,
      "loss": 0.7118,
      "step": 16799
    },
    {
      "epoch": 3.1040698754878275,
      "grad_norm": 1.0194767713546753,
      "learning_rate": 2.2949793215580516e-05,
      "loss": 0.7177,
      "step": 16800
    },
    {
      "epoch": 3.104255714551199,
      "grad_norm": 0.9036667346954346,
      "learning_rate": 2.294048149619995e-05,
      "loss": 0.9459,
      "step": 16801
    },
    {
      "epoch": 3.1044415536145697,
      "grad_norm": 0.8096848130226135,
      "learning_rate": 2.29311714215226e-05,
      "loss": 0.6868,
      "step": 16802
    },
    {
      "epoch": 3.104627392677941,
      "grad_norm": 0.8234491348266602,
      "learning_rate": 2.292186299174712e-05,
      "loss": 0.7727,
      "step": 16803
    },
    {
      "epoch": 3.104813231741312,
      "grad_norm": 0.9342586994171143,
      "learning_rate": 2.291255620707222e-05,
      "loss": 0.7721,
      "step": 16804
    },
    {
      "epoch": 3.1049990708046833,
      "grad_norm": 0.8771103620529175,
      "learning_rate": 2.290325106769653e-05,
      "loss": 1.2882,
      "step": 16805
    },
    {
      "epoch": 3.105184909868054,
      "grad_norm": 0.98863285779953,
      "learning_rate": 2.289394757381864e-05,
      "loss": 0.8628,
      "step": 16806
    },
    {
      "epoch": 3.1053707489314255,
      "grad_norm": 0.9666051268577576,
      "learning_rate": 2.28846457256371e-05,
      "loss": 0.9621,
      "step": 16807
    },
    {
      "epoch": 3.1055565879947964,
      "grad_norm": 0.9185341000556946,
      "learning_rate": 2.2875345523350478e-05,
      "loss": 0.7908,
      "step": 16808
    },
    {
      "epoch": 3.1057424270581677,
      "grad_norm": 0.9975274801254272,
      "learning_rate": 2.2866046967157252e-05,
      "loss": 0.8881,
      "step": 16809
    },
    {
      "epoch": 3.1059282661215386,
      "grad_norm": 1.0821888446807861,
      "learning_rate": 2.285675005725586e-05,
      "loss": 0.8016,
      "step": 16810
    },
    {
      "epoch": 3.10611410518491,
      "grad_norm": 0.9405296444892883,
      "learning_rate": 2.2847454793844793e-05,
      "loss": 0.8053,
      "step": 16811
    },
    {
      "epoch": 3.106299944248281,
      "grad_norm": 1.0025142431259155,
      "learning_rate": 2.2838161177122363e-05,
      "loss": 0.9043,
      "step": 16812
    },
    {
      "epoch": 3.106485783311652,
      "grad_norm": 0.8691514730453491,
      "learning_rate": 2.2828869207287007e-05,
      "loss": 1.0407,
      "step": 16813
    },
    {
      "epoch": 3.106671622375023,
      "grad_norm": 0.8702809810638428,
      "learning_rate": 2.281957888453701e-05,
      "loss": 0.8272,
      "step": 16814
    },
    {
      "epoch": 3.1068574614383944,
      "grad_norm": 0.7858754992485046,
      "learning_rate": 2.281029020907062e-05,
      "loss": 0.6867,
      "step": 16815
    },
    {
      "epoch": 3.1070433005017657,
      "grad_norm": 0.9576261043548584,
      "learning_rate": 2.2801003181086146e-05,
      "loss": 0.8368,
      "step": 16816
    },
    {
      "epoch": 3.1072291395651366,
      "grad_norm": 0.8192430734634399,
      "learning_rate": 2.2791717800781774e-05,
      "loss": 0.7159,
      "step": 16817
    },
    {
      "epoch": 3.107414978628508,
      "grad_norm": 0.8357921838760376,
      "learning_rate": 2.2782434068355674e-05,
      "loss": 0.5564,
      "step": 16818
    },
    {
      "epoch": 3.107600817691879,
      "grad_norm": 1.038155436515808,
      "learning_rate": 2.2773151984006035e-05,
      "loss": 0.7805,
      "step": 16819
    },
    {
      "epoch": 3.10778665675525,
      "grad_norm": 1.601139783859253,
      "learning_rate": 2.2763871547930948e-05,
      "loss": 1.2576,
      "step": 16820
    },
    {
      "epoch": 3.107972495818621,
      "grad_norm": 0.8837090134620667,
      "learning_rate": 2.2754592760328475e-05,
      "loss": 0.8866,
      "step": 16821
    },
    {
      "epoch": 3.1081583348819923,
      "grad_norm": 0.8853998184204102,
      "learning_rate": 2.2745315621396623e-05,
      "loss": 0.7798,
      "step": 16822
    },
    {
      "epoch": 3.108344173945363,
      "grad_norm": 0.9195745587348938,
      "learning_rate": 2.2736040131333448e-05,
      "loss": 0.7439,
      "step": 16823
    },
    {
      "epoch": 3.1085300130087345,
      "grad_norm": 0.9202739000320435,
      "learning_rate": 2.2726766290336922e-05,
      "loss": 0.6378,
      "step": 16824
    },
    {
      "epoch": 3.1087158520721054,
      "grad_norm": 0.9971712827682495,
      "learning_rate": 2.2717494098604973e-05,
      "loss": 1.0663,
      "step": 16825
    },
    {
      "epoch": 3.1089016911354768,
      "grad_norm": 1.095689296722412,
      "learning_rate": 2.2708223556335485e-05,
      "loss": 1.0218,
      "step": 16826
    },
    {
      "epoch": 3.1090875301988476,
      "grad_norm": 1.0428909063339233,
      "learning_rate": 2.26989546637263e-05,
      "loss": 0.938,
      "step": 16827
    },
    {
      "epoch": 3.109273369262219,
      "grad_norm": 0.8788793087005615,
      "learning_rate": 2.2689687420975302e-05,
      "loss": 0.8181,
      "step": 16828
    },
    {
      "epoch": 3.10945920832559,
      "grad_norm": 0.9384439587593079,
      "learning_rate": 2.2680421828280263e-05,
      "loss": 0.8732,
      "step": 16829
    },
    {
      "epoch": 3.109645047388961,
      "grad_norm": 0.9256169199943542,
      "learning_rate": 2.26711578858389e-05,
      "loss": 0.8838,
      "step": 16830
    },
    {
      "epoch": 3.109830886452332,
      "grad_norm": 0.985909104347229,
      "learning_rate": 2.2661895593849003e-05,
      "loss": 0.8377,
      "step": 16831
    },
    {
      "epoch": 3.1100167255157034,
      "grad_norm": 1.1542880535125732,
      "learning_rate": 2.2652634952508224e-05,
      "loss": 0.8399,
      "step": 16832
    },
    {
      "epoch": 3.1102025645790747,
      "grad_norm": 0.7717205882072449,
      "learning_rate": 2.264337596201419e-05,
      "loss": 0.7881,
      "step": 16833
    },
    {
      "epoch": 3.1103884036424456,
      "grad_norm": 0.8536641597747803,
      "learning_rate": 2.263411862256457e-05,
      "loss": 0.6673,
      "step": 16834
    },
    {
      "epoch": 3.110574242705817,
      "grad_norm": 1.0915049314498901,
      "learning_rate": 2.2624862934356894e-05,
      "loss": 0.7309,
      "step": 16835
    },
    {
      "epoch": 3.110760081769188,
      "grad_norm": 0.7397167086601257,
      "learning_rate": 2.261560889758878e-05,
      "loss": 0.507,
      "step": 16836
    },
    {
      "epoch": 3.110945920832559,
      "grad_norm": 0.8940251469612122,
      "learning_rate": 2.260635651245768e-05,
      "loss": 0.8155,
      "step": 16837
    },
    {
      "epoch": 3.11113175989593,
      "grad_norm": 0.883128821849823,
      "learning_rate": 2.2597105779161076e-05,
      "loss": 0.9584,
      "step": 16838
    },
    {
      "epoch": 3.1113175989593014,
      "grad_norm": 0.8234800696372986,
      "learning_rate": 2.2587856697896436e-05,
      "loss": 0.783,
      "step": 16839
    },
    {
      "epoch": 3.1115034380226723,
      "grad_norm": 0.8749162554740906,
      "learning_rate": 2.257860926886115e-05,
      "loss": 0.7674,
      "step": 16840
    },
    {
      "epoch": 3.1116892770860436,
      "grad_norm": 0.8481289744377136,
      "learning_rate": 2.2569363492252594e-05,
      "loss": 0.7863,
      "step": 16841
    },
    {
      "epoch": 3.1118751161494145,
      "grad_norm": 0.9204713702201843,
      "learning_rate": 2.2560119368268072e-05,
      "loss": 0.8807,
      "step": 16842
    },
    {
      "epoch": 3.112060955212786,
      "grad_norm": 0.9275779724121094,
      "learning_rate": 2.2550876897104913e-05,
      "loss": 0.9018,
      "step": 16843
    },
    {
      "epoch": 3.1122467942761567,
      "grad_norm": 0.9116693139076233,
      "learning_rate": 2.2541636078960394e-05,
      "loss": 0.9594,
      "step": 16844
    },
    {
      "epoch": 3.112432633339528,
      "grad_norm": 0.8437976241111755,
      "learning_rate": 2.253239691403174e-05,
      "loss": 0.7266,
      "step": 16845
    },
    {
      "epoch": 3.112618472402899,
      "grad_norm": 0.8511406183242798,
      "learning_rate": 2.2523159402516124e-05,
      "loss": 0.7824,
      "step": 16846
    },
    {
      "epoch": 3.1128043114662702,
      "grad_norm": 0.8179444670677185,
      "learning_rate": 2.2513923544610693e-05,
      "loss": 0.529,
      "step": 16847
    },
    {
      "epoch": 3.112990150529641,
      "grad_norm": 1.014588475227356,
      "learning_rate": 2.250468934051262e-05,
      "loss": 1.0015,
      "step": 16848
    },
    {
      "epoch": 3.1131759895930124,
      "grad_norm": 1.113190770149231,
      "learning_rate": 2.2495456790418967e-05,
      "loss": 1.1612,
      "step": 16849
    },
    {
      "epoch": 3.113361828656384,
      "grad_norm": 1.0497996807098389,
      "learning_rate": 2.248622589452676e-05,
      "loss": 0.8133,
      "step": 16850
    },
    {
      "epoch": 3.1135476677197547,
      "grad_norm": 0.8338691592216492,
      "learning_rate": 2.2476996653033065e-05,
      "loss": 0.655,
      "step": 16851
    },
    {
      "epoch": 3.113733506783126,
      "grad_norm": 0.8171826004981995,
      "learning_rate": 2.2467769066134835e-05,
      "loss": 0.6691,
      "step": 16852
    },
    {
      "epoch": 3.113919345846497,
      "grad_norm": 0.7760305404663086,
      "learning_rate": 2.245854313402904e-05,
      "loss": 0.5202,
      "step": 16853
    },
    {
      "epoch": 3.114105184909868,
      "grad_norm": 0.8941469192504883,
      "learning_rate": 2.2449318856912538e-05,
      "loss": 0.8038,
      "step": 16854
    },
    {
      "epoch": 3.114291023973239,
      "grad_norm": 0.7653095722198486,
      "learning_rate": 2.2440096234982245e-05,
      "loss": 0.8399,
      "step": 16855
    },
    {
      "epoch": 3.1144768630366104,
      "grad_norm": 0.9781475067138672,
      "learning_rate": 2.2430875268435025e-05,
      "loss": 0.7465,
      "step": 16856
    },
    {
      "epoch": 3.1146627020999813,
      "grad_norm": 0.8678989410400391,
      "learning_rate": 2.242165595746766e-05,
      "loss": 0.8347,
      "step": 16857
    },
    {
      "epoch": 3.1148485411633526,
      "grad_norm": 0.842769980430603,
      "learning_rate": 2.2412438302276894e-05,
      "loss": 0.6127,
      "step": 16858
    },
    {
      "epoch": 3.1150343802267235,
      "grad_norm": 0.9742424488067627,
      "learning_rate": 2.240322230305951e-05,
      "loss": 0.8599,
      "step": 16859
    },
    {
      "epoch": 3.115220219290095,
      "grad_norm": 0.8025090098381042,
      "learning_rate": 2.2394007960012186e-05,
      "loss": 0.6261,
      "step": 16860
    },
    {
      "epoch": 3.1154060583534657,
      "grad_norm": 0.927635908126831,
      "learning_rate": 2.2384795273331582e-05,
      "loss": 0.8964,
      "step": 16861
    },
    {
      "epoch": 3.115591897416837,
      "grad_norm": 0.9228578805923462,
      "learning_rate": 2.237558424321431e-05,
      "loss": 0.7664,
      "step": 16862
    },
    {
      "epoch": 3.115777736480208,
      "grad_norm": 1.0590260028839111,
      "learning_rate": 2.2366374869856998e-05,
      "loss": 0.8339,
      "step": 16863
    },
    {
      "epoch": 3.1159635755435793,
      "grad_norm": 0.9382837414741516,
      "learning_rate": 2.2357167153456173e-05,
      "loss": 0.8449,
      "step": 16864
    },
    {
      "epoch": 3.1161494146069506,
      "grad_norm": 0.8885865807533264,
      "learning_rate": 2.2347961094208404e-05,
      "loss": 0.7328,
      "step": 16865
    },
    {
      "epoch": 3.1163352536703215,
      "grad_norm": 0.8921540975570679,
      "learning_rate": 2.2338756692310148e-05,
      "loss": 0.6661,
      "step": 16866
    },
    {
      "epoch": 3.116521092733693,
      "grad_norm": 0.9339958429336548,
      "learning_rate": 2.2329553947957826e-05,
      "loss": 0.8159,
      "step": 16867
    },
    {
      "epoch": 3.1167069317970637,
      "grad_norm": 1.01011061668396,
      "learning_rate": 2.2320352861347915e-05,
      "loss": 0.9934,
      "step": 16868
    },
    {
      "epoch": 3.116892770860435,
      "grad_norm": 0.9793814420700073,
      "learning_rate": 2.2311153432676768e-05,
      "loss": 0.6934,
      "step": 16869
    },
    {
      "epoch": 3.117078609923806,
      "grad_norm": 0.7937256097793579,
      "learning_rate": 2.2301955662140706e-05,
      "loss": 0.6465,
      "step": 16870
    },
    {
      "epoch": 3.1172644489871772,
      "grad_norm": 0.9259080290794373,
      "learning_rate": 2.2292759549936092e-05,
      "loss": 0.7638,
      "step": 16871
    },
    {
      "epoch": 3.117450288050548,
      "grad_norm": 0.7967198491096497,
      "learning_rate": 2.2283565096259174e-05,
      "loss": 0.7027,
      "step": 16872
    },
    {
      "epoch": 3.1176361271139195,
      "grad_norm": 0.7589902281761169,
      "learning_rate": 2.2274372301306202e-05,
      "loss": 0.7964,
      "step": 16873
    },
    {
      "epoch": 3.1178219661772903,
      "grad_norm": 0.9229668378829956,
      "learning_rate": 2.226518116527333e-05,
      "loss": 0.9169,
      "step": 16874
    },
    {
      "epoch": 3.1180078052406617,
      "grad_norm": 1.1118543148040771,
      "learning_rate": 2.225599168835677e-05,
      "loss": 0.982,
      "step": 16875
    },
    {
      "epoch": 3.1181936443040326,
      "grad_norm": 0.8196911811828613,
      "learning_rate": 2.2246803870752685e-05,
      "loss": 0.9871,
      "step": 16876
    },
    {
      "epoch": 3.118379483367404,
      "grad_norm": 0.851613461971283,
      "learning_rate": 2.2237617712657145e-05,
      "loss": 0.8474,
      "step": 16877
    },
    {
      "epoch": 3.1185653224307748,
      "grad_norm": 0.8437772989273071,
      "learning_rate": 2.2228433214266175e-05,
      "loss": 0.7872,
      "step": 16878
    },
    {
      "epoch": 3.118751161494146,
      "grad_norm": 0.9893875122070312,
      "learning_rate": 2.221925037577587e-05,
      "loss": 0.9883,
      "step": 16879
    },
    {
      "epoch": 3.118937000557517,
      "grad_norm": 1.1424741744995117,
      "learning_rate": 2.2210069197382178e-05,
      "loss": 1.1362,
      "step": 16880
    },
    {
      "epoch": 3.1191228396208883,
      "grad_norm": 1.070717692375183,
      "learning_rate": 2.2200889679281077e-05,
      "loss": 0.7183,
      "step": 16881
    },
    {
      "epoch": 3.119308678684259,
      "grad_norm": 0.8586819767951965,
      "learning_rate": 2.2191711821668447e-05,
      "loss": 0.9298,
      "step": 16882
    },
    {
      "epoch": 3.1194945177476305,
      "grad_norm": 0.9052451848983765,
      "learning_rate": 2.218253562474023e-05,
      "loss": 0.9157,
      "step": 16883
    },
    {
      "epoch": 3.119680356811002,
      "grad_norm": 0.8883751034736633,
      "learning_rate": 2.2173361088692247e-05,
      "loss": 0.5857,
      "step": 16884
    },
    {
      "epoch": 3.1198661958743727,
      "grad_norm": 0.9195517897605896,
      "learning_rate": 2.2164188213720295e-05,
      "loss": 0.7607,
      "step": 16885
    },
    {
      "epoch": 3.120052034937744,
      "grad_norm": 0.9243937730789185,
      "learning_rate": 2.2155017000020194e-05,
      "loss": 0.9328,
      "step": 16886
    },
    {
      "epoch": 3.120237874001115,
      "grad_norm": 0.8645699620246887,
      "learning_rate": 2.2145847447787636e-05,
      "loss": 0.8224,
      "step": 16887
    },
    {
      "epoch": 3.1204237130644863,
      "grad_norm": 1.0877633094787598,
      "learning_rate": 2.21366795572184e-05,
      "loss": 0.6701,
      "step": 16888
    },
    {
      "epoch": 3.120609552127857,
      "grad_norm": 0.8657917976379395,
      "learning_rate": 2.2127513328508122e-05,
      "loss": 0.7049,
      "step": 16889
    },
    {
      "epoch": 3.1207953911912285,
      "grad_norm": 0.8166100978851318,
      "learning_rate": 2.2118348761852403e-05,
      "loss": 0.8629,
      "step": 16890
    },
    {
      "epoch": 3.1209812302545994,
      "grad_norm": 0.7946850657463074,
      "learning_rate": 2.2109185857446903e-05,
      "loss": 0.6853,
      "step": 16891
    },
    {
      "epoch": 3.1211670693179707,
      "grad_norm": 0.9389623403549194,
      "learning_rate": 2.2100024615487168e-05,
      "loss": 0.8606,
      "step": 16892
    },
    {
      "epoch": 3.1213529083813416,
      "grad_norm": 0.6917771697044373,
      "learning_rate": 2.209086503616873e-05,
      "loss": 0.5993,
      "step": 16893
    },
    {
      "epoch": 3.121538747444713,
      "grad_norm": 0.8145844340324402,
      "learning_rate": 2.2081707119687055e-05,
      "loss": 0.6598,
      "step": 16894
    },
    {
      "epoch": 3.121724586508084,
      "grad_norm": 1.6039071083068848,
      "learning_rate": 2.207255086623765e-05,
      "loss": 1.1735,
      "step": 16895
    },
    {
      "epoch": 3.121910425571455,
      "grad_norm": 0.9803751111030579,
      "learning_rate": 2.206339627601589e-05,
      "loss": 0.884,
      "step": 16896
    },
    {
      "epoch": 3.122096264634826,
      "grad_norm": 0.9029519557952881,
      "learning_rate": 2.2054243349217217e-05,
      "loss": 0.7225,
      "step": 16897
    },
    {
      "epoch": 3.1222821036981974,
      "grad_norm": 0.8601951003074646,
      "learning_rate": 2.2045092086036946e-05,
      "loss": 0.8944,
      "step": 16898
    },
    {
      "epoch": 3.1224679427615687,
      "grad_norm": 0.8614652156829834,
      "learning_rate": 2.2035942486670425e-05,
      "loss": 0.6491,
      "step": 16899
    },
    {
      "epoch": 3.1226537818249396,
      "grad_norm": 0.9597631692886353,
      "learning_rate": 2.2026794551312923e-05,
      "loss": 0.9077,
      "step": 16900
    },
    {
      "epoch": 3.122839620888311,
      "grad_norm": 1.0509092807769775,
      "learning_rate": 2.2017648280159676e-05,
      "loss": 0.8838,
      "step": 16901
    },
    {
      "epoch": 3.123025459951682,
      "grad_norm": 0.8761752247810364,
      "learning_rate": 2.200850367340589e-05,
      "loss": 0.7991,
      "step": 16902
    },
    {
      "epoch": 3.123211299015053,
      "grad_norm": 0.8011770844459534,
      "learning_rate": 2.1999360731246764e-05,
      "loss": 0.891,
      "step": 16903
    },
    {
      "epoch": 3.123397138078424,
      "grad_norm": 0.8152168393135071,
      "learning_rate": 2.199021945387745e-05,
      "loss": 0.5233,
      "step": 16904
    },
    {
      "epoch": 3.1235829771417953,
      "grad_norm": 0.9296904802322388,
      "learning_rate": 2.1981079841493014e-05,
      "loss": 0.7773,
      "step": 16905
    },
    {
      "epoch": 3.123768816205166,
      "grad_norm": 0.8891332745552063,
      "learning_rate": 2.1971941894288517e-05,
      "loss": 0.7992,
      "step": 16906
    },
    {
      "epoch": 3.1239546552685376,
      "grad_norm": 1.0827486515045166,
      "learning_rate": 2.1962805612459024e-05,
      "loss": 0.7503,
      "step": 16907
    },
    {
      "epoch": 3.1241404943319084,
      "grad_norm": 0.8078093528747559,
      "learning_rate": 2.195367099619956e-05,
      "loss": 0.8623,
      "step": 16908
    },
    {
      "epoch": 3.1243263333952798,
      "grad_norm": 0.8417673110961914,
      "learning_rate": 2.1944538045705055e-05,
      "loss": 0.8114,
      "step": 16909
    },
    {
      "epoch": 3.1245121724586506,
      "grad_norm": 1.1821131706237793,
      "learning_rate": 2.1935406761170396e-05,
      "loss": 0.8338,
      "step": 16910
    },
    {
      "epoch": 3.124698011522022,
      "grad_norm": 0.895492434501648,
      "learning_rate": 2.1926277142790552e-05,
      "loss": 0.9698,
      "step": 16911
    },
    {
      "epoch": 3.124883850585393,
      "grad_norm": 1.0384591817855835,
      "learning_rate": 2.1917149190760346e-05,
      "loss": 1.0607,
      "step": 16912
    },
    {
      "epoch": 3.125069689648764,
      "grad_norm": 0.7608576416969299,
      "learning_rate": 2.1908022905274595e-05,
      "loss": 0.6077,
      "step": 16913
    },
    {
      "epoch": 3.1252555287121355,
      "grad_norm": 0.9334760904312134,
      "learning_rate": 2.1898898286528047e-05,
      "loss": 0.7666,
      "step": 16914
    },
    {
      "epoch": 3.1254413677755064,
      "grad_norm": 1.024980068206787,
      "learning_rate": 2.1889775334715525e-05,
      "loss": 0.5776,
      "step": 16915
    },
    {
      "epoch": 3.1256272068388777,
      "grad_norm": 0.8938941955566406,
      "learning_rate": 2.18806540500317e-05,
      "loss": 0.8932,
      "step": 16916
    },
    {
      "epoch": 3.1258130459022486,
      "grad_norm": 1.1975553035736084,
      "learning_rate": 2.1871534432671238e-05,
      "loss": 0.8648,
      "step": 16917
    },
    {
      "epoch": 3.12599888496562,
      "grad_norm": 0.8598381280899048,
      "learning_rate": 2.1862416482828783e-05,
      "loss": 0.9019,
      "step": 16918
    },
    {
      "epoch": 3.126184724028991,
      "grad_norm": 1.0694983005523682,
      "learning_rate": 2.1853300200699e-05,
      "loss": 0.7902,
      "step": 16919
    },
    {
      "epoch": 3.126370563092362,
      "grad_norm": 0.959589421749115,
      "learning_rate": 2.1844185586476416e-05,
      "loss": 0.9651,
      "step": 16920
    },
    {
      "epoch": 3.126556402155733,
      "grad_norm": 0.8493366241455078,
      "learning_rate": 2.183507264035557e-05,
      "loss": 0.6951,
      "step": 16921
    },
    {
      "epoch": 3.1267422412191044,
      "grad_norm": 0.893376350402832,
      "learning_rate": 2.1825961362530932e-05,
      "loss": 1.0166,
      "step": 16922
    },
    {
      "epoch": 3.1269280802824753,
      "grad_norm": 0.9279434084892273,
      "learning_rate": 2.181685175319702e-05,
      "loss": 0.6794,
      "step": 16923
    },
    {
      "epoch": 3.1271139193458466,
      "grad_norm": 0.9398536682128906,
      "learning_rate": 2.1807743812548253e-05,
      "loss": 0.6859,
      "step": 16924
    },
    {
      "epoch": 3.1272997584092175,
      "grad_norm": 0.8415723443031311,
      "learning_rate": 2.1798637540779e-05,
      "loss": 0.729,
      "step": 16925
    },
    {
      "epoch": 3.127485597472589,
      "grad_norm": 1.1050769090652466,
      "learning_rate": 2.178953293808361e-05,
      "loss": 0.869,
      "step": 16926
    },
    {
      "epoch": 3.1276714365359597,
      "grad_norm": 1.0179059505462646,
      "learning_rate": 2.1780430004656416e-05,
      "loss": 0.9382,
      "step": 16927
    },
    {
      "epoch": 3.127857275599331,
      "grad_norm": 0.912026047706604,
      "learning_rate": 2.1771328740691745e-05,
      "loss": 0.8454,
      "step": 16928
    },
    {
      "epoch": 3.128043114662702,
      "grad_norm": 1.1212213039398193,
      "learning_rate": 2.1762229146383816e-05,
      "loss": 0.7195,
      "step": 16929
    },
    {
      "epoch": 3.1282289537260732,
      "grad_norm": 0.9466208219528198,
      "learning_rate": 2.1753131221926825e-05,
      "loss": 0.976,
      "step": 16930
    },
    {
      "epoch": 3.128414792789444,
      "grad_norm": 0.9594152569770813,
      "learning_rate": 2.1744034967514992e-05,
      "loss": 0.6365,
      "step": 16931
    },
    {
      "epoch": 3.1286006318528155,
      "grad_norm": 0.9360522627830505,
      "learning_rate": 2.173494038334244e-05,
      "loss": 0.9949,
      "step": 16932
    },
    {
      "epoch": 3.128786470916187,
      "grad_norm": 0.9982490539550781,
      "learning_rate": 2.172584746960328e-05,
      "loss": 0.4391,
      "step": 16933
    },
    {
      "epoch": 3.1289723099795577,
      "grad_norm": 0.9104912877082825,
      "learning_rate": 2.1716756226491543e-05,
      "loss": 0.7869,
      "step": 16934
    },
    {
      "epoch": 3.129158149042929,
      "grad_norm": 0.8269588351249695,
      "learning_rate": 2.1707666654201342e-05,
      "loss": 0.796,
      "step": 16935
    },
    {
      "epoch": 3.1293439881063,
      "grad_norm": 1.1801178455352783,
      "learning_rate": 2.1698578752926634e-05,
      "loss": 0.8755,
      "step": 16936
    },
    {
      "epoch": 3.129529827169671,
      "grad_norm": 0.9644780158996582,
      "learning_rate": 2.1689492522861398e-05,
      "loss": 0.9328,
      "step": 16937
    },
    {
      "epoch": 3.129715666233042,
      "grad_norm": 0.8413048982620239,
      "learning_rate": 2.168040796419951e-05,
      "loss": 0.7658,
      "step": 16938
    },
    {
      "epoch": 3.1299015052964134,
      "grad_norm": 0.9241469502449036,
      "learning_rate": 2.1671325077134963e-05,
      "loss": 0.8426,
      "step": 16939
    },
    {
      "epoch": 3.1300873443597843,
      "grad_norm": 0.9627590179443359,
      "learning_rate": 2.166224386186155e-05,
      "loss": 0.9674,
      "step": 16940
    },
    {
      "epoch": 3.1302731834231556,
      "grad_norm": 0.9975087642669678,
      "learning_rate": 2.1653164318573126e-05,
      "loss": 0.7909,
      "step": 16941
    },
    {
      "epoch": 3.1304590224865265,
      "grad_norm": 0.9133538603782654,
      "learning_rate": 2.164408644746343e-05,
      "loss": 1.1582,
      "step": 16942
    },
    {
      "epoch": 3.130644861549898,
      "grad_norm": 0.8557292222976685,
      "learning_rate": 2.1635010248726274e-05,
      "loss": 0.8368,
      "step": 16943
    },
    {
      "epoch": 3.1308307006132687,
      "grad_norm": 0.9679376482963562,
      "learning_rate": 2.1625935722555335e-05,
      "loss": 0.7238,
      "step": 16944
    },
    {
      "epoch": 3.13101653967664,
      "grad_norm": 1.0886197090148926,
      "learning_rate": 2.1616862869144304e-05,
      "loss": 0.7671,
      "step": 16945
    },
    {
      "epoch": 3.131202378740011,
      "grad_norm": 0.8300682902336121,
      "learning_rate": 2.1607791688686808e-05,
      "loss": 0.7886,
      "step": 16946
    },
    {
      "epoch": 3.1313882178033823,
      "grad_norm": 0.9571086168289185,
      "learning_rate": 2.1598722181376497e-05,
      "loss": 0.829,
      "step": 16947
    },
    {
      "epoch": 3.1315740568667536,
      "grad_norm": 0.9584956169128418,
      "learning_rate": 2.1589654347406886e-05,
      "loss": 0.8305,
      "step": 16948
    },
    {
      "epoch": 3.1317598959301245,
      "grad_norm": 0.9124837517738342,
      "learning_rate": 2.1580588186971575e-05,
      "loss": 0.7435,
      "step": 16949
    },
    {
      "epoch": 3.131945734993496,
      "grad_norm": 0.8917635679244995,
      "learning_rate": 2.1571523700264006e-05,
      "loss": 0.8684,
      "step": 16950
    },
    {
      "epoch": 3.1321315740568667,
      "grad_norm": 0.9225831031799316,
      "learning_rate": 2.15624608874777e-05,
      "loss": 0.537,
      "step": 16951
    },
    {
      "epoch": 3.132317413120238,
      "grad_norm": 0.991547703742981,
      "learning_rate": 2.155339974880607e-05,
      "loss": 0.9213,
      "step": 16952
    },
    {
      "epoch": 3.132503252183609,
      "grad_norm": 0.9788913726806641,
      "learning_rate": 2.15443402844425e-05,
      "loss": 0.8236,
      "step": 16953
    },
    {
      "epoch": 3.1326890912469803,
      "grad_norm": 0.7579008340835571,
      "learning_rate": 2.1535282494580323e-05,
      "loss": 1.0362,
      "step": 16954
    },
    {
      "epoch": 3.132874930310351,
      "grad_norm": 0.9785481691360474,
      "learning_rate": 2.1526226379412906e-05,
      "loss": 0.8357,
      "step": 16955
    },
    {
      "epoch": 3.1330607693737225,
      "grad_norm": 0.907016932964325,
      "learning_rate": 2.1517171939133542e-05,
      "loss": 0.767,
      "step": 16956
    },
    {
      "epoch": 3.1332466084370934,
      "grad_norm": 0.8444396257400513,
      "learning_rate": 2.1508119173935448e-05,
      "loss": 0.8928,
      "step": 16957
    },
    {
      "epoch": 3.1334324475004647,
      "grad_norm": 1.1882593631744385,
      "learning_rate": 2.149906808401183e-05,
      "loss": 0.9047,
      "step": 16958
    },
    {
      "epoch": 3.1336182865638356,
      "grad_norm": 0.8932169675827026,
      "learning_rate": 2.149001866955589e-05,
      "loss": 0.7651,
      "step": 16959
    },
    {
      "epoch": 3.133804125627207,
      "grad_norm": 0.9666036367416382,
      "learning_rate": 2.148097093076079e-05,
      "loss": 0.6123,
      "step": 16960
    },
    {
      "epoch": 3.133989964690578,
      "grad_norm": 0.9772616028785706,
      "learning_rate": 2.1471924867819636e-05,
      "loss": 0.8967,
      "step": 16961
    },
    {
      "epoch": 3.134175803753949,
      "grad_norm": 0.933224081993103,
      "learning_rate": 2.1462880480925452e-05,
      "loss": 0.9886,
      "step": 16962
    },
    {
      "epoch": 3.13436164281732,
      "grad_norm": 0.9614982008934021,
      "learning_rate": 2.1453837770271334e-05,
      "loss": 0.7811,
      "step": 16963
    },
    {
      "epoch": 3.1345474818806913,
      "grad_norm": 0.9009011387825012,
      "learning_rate": 2.1444796736050266e-05,
      "loss": 0.9589,
      "step": 16964
    },
    {
      "epoch": 3.134733320944062,
      "grad_norm": 1.1222648620605469,
      "learning_rate": 2.14357573784552e-05,
      "loss": 0.773,
      "step": 16965
    },
    {
      "epoch": 3.1349191600074335,
      "grad_norm": 0.9041205048561096,
      "learning_rate": 2.1426719697679054e-05,
      "loss": 0.7157,
      "step": 16966
    },
    {
      "epoch": 3.135104999070805,
      "grad_norm": 0.8128951191902161,
      "learning_rate": 2.1417683693914747e-05,
      "loss": 0.7409,
      "step": 16967
    },
    {
      "epoch": 3.1352908381341758,
      "grad_norm": 0.8408454060554504,
      "learning_rate": 2.1408649367355148e-05,
      "loss": 0.5803,
      "step": 16968
    },
    {
      "epoch": 3.135476677197547,
      "grad_norm": 0.951663613319397,
      "learning_rate": 2.1399616718193017e-05,
      "loss": 0.7976,
      "step": 16969
    },
    {
      "epoch": 3.135662516260918,
      "grad_norm": 1.1752151250839233,
      "learning_rate": 2.1390585746621193e-05,
      "loss": 1.0018,
      "step": 16970
    },
    {
      "epoch": 3.1358483553242893,
      "grad_norm": 0.8627622127532959,
      "learning_rate": 2.138155645283244e-05,
      "loss": 0.8013,
      "step": 16971
    },
    {
      "epoch": 3.13603419438766,
      "grad_norm": 0.999226450920105,
      "learning_rate": 2.1372528837019455e-05,
      "loss": 1.0242,
      "step": 16972
    },
    {
      "epoch": 3.1362200334510315,
      "grad_norm": 1.0154447555541992,
      "learning_rate": 2.136350289937491e-05,
      "loss": 1.0382,
      "step": 16973
    },
    {
      "epoch": 3.1364058725144024,
      "grad_norm": 0.9527133703231812,
      "learning_rate": 2.1354478640091424e-05,
      "loss": 0.7691,
      "step": 16974
    },
    {
      "epoch": 3.1365917115777737,
      "grad_norm": 0.778480052947998,
      "learning_rate": 2.134545605936166e-05,
      "loss": 0.7514,
      "step": 16975
    },
    {
      "epoch": 3.1367775506411446,
      "grad_norm": 1.0286147594451904,
      "learning_rate": 2.1336435157378143e-05,
      "loss": 0.6675,
      "step": 16976
    },
    {
      "epoch": 3.136963389704516,
      "grad_norm": 0.9593778252601624,
      "learning_rate": 2.1327415934333438e-05,
      "loss": 0.8536,
      "step": 16977
    },
    {
      "epoch": 3.137149228767887,
      "grad_norm": 0.7049160003662109,
      "learning_rate": 2.1318398390420013e-05,
      "loss": 0.6347,
      "step": 16978
    },
    {
      "epoch": 3.137335067831258,
      "grad_norm": 1.0583542585372925,
      "learning_rate": 2.130938252583037e-05,
      "loss": 0.9364,
      "step": 16979
    },
    {
      "epoch": 3.137520906894629,
      "grad_norm": 0.9129087328910828,
      "learning_rate": 2.130036834075688e-05,
      "loss": 0.9682,
      "step": 16980
    },
    {
      "epoch": 3.1377067459580004,
      "grad_norm": 0.6961577534675598,
      "learning_rate": 2.1291355835392025e-05,
      "loss": 0.5642,
      "step": 16981
    },
    {
      "epoch": 3.1378925850213717,
      "grad_norm": 0.7811499834060669,
      "learning_rate": 2.128234500992806e-05,
      "loss": 0.5969,
      "step": 16982
    },
    {
      "epoch": 3.1380784240847426,
      "grad_norm": 0.8767191767692566,
      "learning_rate": 2.12733358645574e-05,
      "loss": 0.7998,
      "step": 16983
    },
    {
      "epoch": 3.138264263148114,
      "grad_norm": 0.809659481048584,
      "learning_rate": 2.1264328399472277e-05,
      "loss": 0.7883,
      "step": 16984
    },
    {
      "epoch": 3.138450102211485,
      "grad_norm": 0.8992799520492554,
      "learning_rate": 2.1255322614864947e-05,
      "loss": 0.8595,
      "step": 16985
    },
    {
      "epoch": 3.138635941274856,
      "grad_norm": 0.8708093762397766,
      "learning_rate": 2.124631851092761e-05,
      "loss": 0.848,
      "step": 16986
    },
    {
      "epoch": 3.138821780338227,
      "grad_norm": 0.9375101327896118,
      "learning_rate": 2.1237316087852466e-05,
      "loss": 0.8506,
      "step": 16987
    },
    {
      "epoch": 3.1390076194015983,
      "grad_norm": 0.8981173634529114,
      "learning_rate": 2.1228315345831662e-05,
      "loss": 0.8373,
      "step": 16988
    },
    {
      "epoch": 3.139193458464969,
      "grad_norm": 0.8532535433769226,
      "learning_rate": 2.1219316285057255e-05,
      "loss": 0.6353,
      "step": 16989
    },
    {
      "epoch": 3.1393792975283406,
      "grad_norm": 0.9816616177558899,
      "learning_rate": 2.1210318905721348e-05,
      "loss": 1.1142,
      "step": 16990
    },
    {
      "epoch": 3.1395651365917114,
      "grad_norm": 0.8044140934944153,
      "learning_rate": 2.120132320801601e-05,
      "loss": 0.6628,
      "step": 16991
    },
    {
      "epoch": 3.1397509756550828,
      "grad_norm": 0.943625271320343,
      "learning_rate": 2.1192329192133198e-05,
      "loss": 0.6296,
      "step": 16992
    },
    {
      "epoch": 3.1399368147184537,
      "grad_norm": 1.0095001459121704,
      "learning_rate": 2.118333685826489e-05,
      "loss": 1.1645,
      "step": 16993
    },
    {
      "epoch": 3.140122653781825,
      "grad_norm": 0.8924473524093628,
      "learning_rate": 2.1174346206602968e-05,
      "loss": 0.8996,
      "step": 16994
    },
    {
      "epoch": 3.140308492845196,
      "grad_norm": 0.93332839012146,
      "learning_rate": 2.116535723733938e-05,
      "loss": 1.076,
      "step": 16995
    },
    {
      "epoch": 3.140494331908567,
      "grad_norm": 0.9304823279380798,
      "learning_rate": 2.1156369950665956e-05,
      "loss": 0.8909,
      "step": 16996
    },
    {
      "epoch": 3.1406801709719385,
      "grad_norm": 0.8623441457748413,
      "learning_rate": 2.1147384346774523e-05,
      "loss": 0.611,
      "step": 16997
    },
    {
      "epoch": 3.1408660100353094,
      "grad_norm": 0.9617223143577576,
      "learning_rate": 2.113840042585682e-05,
      "loss": 0.9052,
      "step": 16998
    },
    {
      "epoch": 3.1410518490986807,
      "grad_norm": 0.8871750831604004,
      "learning_rate": 2.1129418188104656e-05,
      "loss": 0.6957,
      "step": 16999
    },
    {
      "epoch": 3.1412376881620516,
      "grad_norm": 0.8519712686538696,
      "learning_rate": 2.1120437633709712e-05,
      "loss": 0.5776,
      "step": 17000
    },
    {
      "epoch": 3.141423527225423,
      "grad_norm": 0.8367720246315002,
      "learning_rate": 2.1111458762863633e-05,
      "loss": 0.6709,
      "step": 17001
    },
    {
      "epoch": 3.141609366288794,
      "grad_norm": 0.8938384056091309,
      "learning_rate": 2.1102481575758094e-05,
      "loss": 0.784,
      "step": 17002
    },
    {
      "epoch": 3.141795205352165,
      "grad_norm": 1.1250386238098145,
      "learning_rate": 2.1093506072584724e-05,
      "loss": 0.9777,
      "step": 17003
    },
    {
      "epoch": 3.141981044415536,
      "grad_norm": 0.8990409970283508,
      "learning_rate": 2.108453225353505e-05,
      "loss": 0.9568,
      "step": 17004
    },
    {
      "epoch": 3.1421668834789074,
      "grad_norm": 0.9447553157806396,
      "learning_rate": 2.107556011880062e-05,
      "loss": 0.6788,
      "step": 17005
    },
    {
      "epoch": 3.1423527225422783,
      "grad_norm": 0.8468141555786133,
      "learning_rate": 2.106658966857288e-05,
      "loss": 0.776,
      "step": 17006
    },
    {
      "epoch": 3.1425385616056496,
      "grad_norm": 1.0220496654510498,
      "learning_rate": 2.105762090304336e-05,
      "loss": 0.6239,
      "step": 17007
    },
    {
      "epoch": 3.1427244006690205,
      "grad_norm": 1.0547794103622437,
      "learning_rate": 2.1048653822403454e-05,
      "loss": 0.9287,
      "step": 17008
    },
    {
      "epoch": 3.142910239732392,
      "grad_norm": 0.8928568959236145,
      "learning_rate": 2.103968842684452e-05,
      "loss": 0.768,
      "step": 17009
    },
    {
      "epoch": 3.1430960787957627,
      "grad_norm": 0.9178768992424011,
      "learning_rate": 2.1030724716557958e-05,
      "loss": 0.7972,
      "step": 17010
    },
    {
      "epoch": 3.143281917859134,
      "grad_norm": 0.9317730665206909,
      "learning_rate": 2.1021762691735035e-05,
      "loss": 0.8405,
      "step": 17011
    },
    {
      "epoch": 3.143467756922505,
      "grad_norm": 0.9283566474914551,
      "learning_rate": 2.1012802352567085e-05,
      "loss": 0.7425,
      "step": 17012
    },
    {
      "epoch": 3.1436535959858762,
      "grad_norm": 0.7513983845710754,
      "learning_rate": 2.1003843699245317e-05,
      "loss": 0.6545,
      "step": 17013
    },
    {
      "epoch": 3.143839435049247,
      "grad_norm": 0.9660999774932861,
      "learning_rate": 2.099488673196093e-05,
      "loss": 0.7218,
      "step": 17014
    },
    {
      "epoch": 3.1440252741126185,
      "grad_norm": 0.6714902520179749,
      "learning_rate": 2.0985931450905118e-05,
      "loss": 0.4998,
      "step": 17015
    },
    {
      "epoch": 3.14421111317599,
      "grad_norm": 1.0693867206573486,
      "learning_rate": 2.0976977856269008e-05,
      "loss": 0.8455,
      "step": 17016
    },
    {
      "epoch": 3.1443969522393607,
      "grad_norm": 0.9602853059768677,
      "learning_rate": 2.0968025948243696e-05,
      "loss": 0.8069,
      "step": 17017
    },
    {
      "epoch": 3.144582791302732,
      "grad_norm": 0.8815193772315979,
      "learning_rate": 2.095907572702023e-05,
      "loss": 0.9785,
      "step": 17018
    },
    {
      "epoch": 3.144768630366103,
      "grad_norm": 0.8699519038200378,
      "learning_rate": 2.095012719278966e-05,
      "loss": 0.7088,
      "step": 17019
    },
    {
      "epoch": 3.144954469429474,
      "grad_norm": 0.7912164926528931,
      "learning_rate": 2.0941180345742983e-05,
      "loss": 0.5679,
      "step": 17020
    },
    {
      "epoch": 3.145140308492845,
      "grad_norm": 1.0096666812896729,
      "learning_rate": 2.0932235186071114e-05,
      "loss": 0.8969,
      "step": 17021
    },
    {
      "epoch": 3.1453261475562164,
      "grad_norm": 0.7588926553726196,
      "learning_rate": 2.0923291713965e-05,
      "loss": 0.7405,
      "step": 17022
    },
    {
      "epoch": 3.1455119866195873,
      "grad_norm": 0.8650400638580322,
      "learning_rate": 2.091434992961555e-05,
      "loss": 0.7048,
      "step": 17023
    },
    {
      "epoch": 3.1456978256829586,
      "grad_norm": 1.0307903289794922,
      "learning_rate": 2.0905409833213574e-05,
      "loss": 0.8026,
      "step": 17024
    },
    {
      "epoch": 3.1458836647463295,
      "grad_norm": 1.0307273864746094,
      "learning_rate": 2.0896471424949904e-05,
      "loss": 0.8717,
      "step": 17025
    },
    {
      "epoch": 3.146069503809701,
      "grad_norm": 1.038767695426941,
      "learning_rate": 2.0887534705015276e-05,
      "loss": 0.8306,
      "step": 17026
    },
    {
      "epoch": 3.1462553428730717,
      "grad_norm": 0.9466065764427185,
      "learning_rate": 2.0878599673600486e-05,
      "loss": 0.8489,
      "step": 17027
    },
    {
      "epoch": 3.146441181936443,
      "grad_norm": 0.9983720183372498,
      "learning_rate": 2.086966633089621e-05,
      "loss": 0.8301,
      "step": 17028
    },
    {
      "epoch": 3.146627020999814,
      "grad_norm": 0.836397647857666,
      "learning_rate": 2.0860734677093074e-05,
      "loss": 0.7008,
      "step": 17029
    },
    {
      "epoch": 3.1468128600631853,
      "grad_norm": 0.8518416881561279,
      "learning_rate": 2.0851804712381783e-05,
      "loss": 0.8584,
      "step": 17030
    },
    {
      "epoch": 3.1469986991265566,
      "grad_norm": 0.8676844835281372,
      "learning_rate": 2.0842876436952897e-05,
      "loss": 0.648,
      "step": 17031
    },
    {
      "epoch": 3.1471845381899275,
      "grad_norm": 0.8333600163459778,
      "learning_rate": 2.0833949850996948e-05,
      "loss": 0.8164,
      "step": 17032
    },
    {
      "epoch": 3.147370377253299,
      "grad_norm": 0.8571373820304871,
      "learning_rate": 2.082502495470451e-05,
      "loss": 0.6943,
      "step": 17033
    },
    {
      "epoch": 3.1475562163166697,
      "grad_norm": 0.9203354716300964,
      "learning_rate": 2.0816101748266014e-05,
      "loss": 0.6767,
      "step": 17034
    },
    {
      "epoch": 3.147742055380041,
      "grad_norm": 0.9560515284538269,
      "learning_rate": 2.080718023187198e-05,
      "loss": 0.7109,
      "step": 17035
    },
    {
      "epoch": 3.147927894443412,
      "grad_norm": 0.9986955523490906,
      "learning_rate": 2.0798260405712777e-05,
      "loss": 0.8818,
      "step": 17036
    },
    {
      "epoch": 3.1481137335067833,
      "grad_norm": 0.9421307444572449,
      "learning_rate": 2.0789342269978785e-05,
      "loss": 0.6323,
      "step": 17037
    },
    {
      "epoch": 3.148299572570154,
      "grad_norm": 0.9801344871520996,
      "learning_rate": 2.078042582486033e-05,
      "loss": 0.8406,
      "step": 17038
    },
    {
      "epoch": 3.1484854116335255,
      "grad_norm": 0.8806696534156799,
      "learning_rate": 2.0771511070547768e-05,
      "loss": 0.7143,
      "step": 17039
    },
    {
      "epoch": 3.1486712506968964,
      "grad_norm": 0.9294523000717163,
      "learning_rate": 2.076259800723134e-05,
      "loss": 0.8736,
      "step": 17040
    },
    {
      "epoch": 3.1488570897602677,
      "grad_norm": 0.8931056261062622,
      "learning_rate": 2.0753686635101244e-05,
      "loss": 0.7932,
      "step": 17041
    },
    {
      "epoch": 3.1490429288236386,
      "grad_norm": 0.9533994793891907,
      "learning_rate": 2.0744776954347745e-05,
      "loss": 0.8303,
      "step": 17042
    },
    {
      "epoch": 3.14922876788701,
      "grad_norm": 1.0638116598129272,
      "learning_rate": 2.0735868965160953e-05,
      "loss": 1.0265,
      "step": 17043
    },
    {
      "epoch": 3.149414606950381,
      "grad_norm": 0.8813306093215942,
      "learning_rate": 2.0726962667731044e-05,
      "loss": 1.0306,
      "step": 17044
    },
    {
      "epoch": 3.149600446013752,
      "grad_norm": 0.8358373045921326,
      "learning_rate": 2.0718058062248068e-05,
      "loss": 0.6696,
      "step": 17045
    },
    {
      "epoch": 3.1497862850771234,
      "grad_norm": 0.9361814856529236,
      "learning_rate": 2.070915514890207e-05,
      "loss": 0.839,
      "step": 17046
    },
    {
      "epoch": 3.1499721241404943,
      "grad_norm": 0.9304302334785461,
      "learning_rate": 2.0700253927883117e-05,
      "loss": 0.7693,
      "step": 17047
    },
    {
      "epoch": 3.1501579632038657,
      "grad_norm": 0.9680723547935486,
      "learning_rate": 2.0691354399381158e-05,
      "loss": 0.84,
      "step": 17048
    },
    {
      "epoch": 3.1503438022672365,
      "grad_norm": 0.8648490905761719,
      "learning_rate": 2.0682456563586118e-05,
      "loss": 0.6622,
      "step": 17049
    },
    {
      "epoch": 3.150529641330608,
      "grad_norm": 1.137999415397644,
      "learning_rate": 2.0673560420687945e-05,
      "loss": 1.0689,
      "step": 17050
    },
    {
      "epoch": 3.1507154803939788,
      "grad_norm": 0.9675096869468689,
      "learning_rate": 2.0664665970876496e-05,
      "loss": 0.8624,
      "step": 17051
    },
    {
      "epoch": 3.15090131945735,
      "grad_norm": 0.9901977777481079,
      "learning_rate": 2.0655773214341612e-05,
      "loss": 0.8271,
      "step": 17052
    },
    {
      "epoch": 3.151087158520721,
      "grad_norm": 1.0320415496826172,
      "learning_rate": 2.0646882151273062e-05,
      "loss": 0.7143,
      "step": 17053
    },
    {
      "epoch": 3.1512729975840923,
      "grad_norm": 0.9752329587936401,
      "learning_rate": 2.063799278186064e-05,
      "loss": 0.8849,
      "step": 17054
    },
    {
      "epoch": 3.151458836647463,
      "grad_norm": 0.7743865251541138,
      "learning_rate": 2.062910510629409e-05,
      "loss": 0.7674,
      "step": 17055
    },
    {
      "epoch": 3.1516446757108345,
      "grad_norm": 0.9031053185462952,
      "learning_rate": 2.0620219124763086e-05,
      "loss": 0.9793,
      "step": 17056
    },
    {
      "epoch": 3.1518305147742054,
      "grad_norm": 0.9149486422538757,
      "learning_rate": 2.0611334837457287e-05,
      "loss": 0.857,
      "step": 17057
    },
    {
      "epoch": 3.1520163538375767,
      "grad_norm": 0.8257530927658081,
      "learning_rate": 2.0602452244566283e-05,
      "loss": 0.7835,
      "step": 17058
    },
    {
      "epoch": 3.1522021929009476,
      "grad_norm": 1.0256621837615967,
      "learning_rate": 2.0593571346279716e-05,
      "loss": 0.9413,
      "step": 17059
    },
    {
      "epoch": 3.152388031964319,
      "grad_norm": 0.8315449357032776,
      "learning_rate": 2.0584692142787086e-05,
      "loss": 0.9959,
      "step": 17060
    },
    {
      "epoch": 3.15257387102769,
      "grad_norm": 0.9501693248748779,
      "learning_rate": 2.0575814634277914e-05,
      "loss": 0.6163,
      "step": 17061
    },
    {
      "epoch": 3.152759710091061,
      "grad_norm": 1.0857973098754883,
      "learning_rate": 2.0566938820941695e-05,
      "loss": 0.842,
      "step": 17062
    },
    {
      "epoch": 3.152945549154432,
      "grad_norm": 0.5966393947601318,
      "learning_rate": 2.055806470296785e-05,
      "loss": 0.4318,
      "step": 17063
    },
    {
      "epoch": 3.1531313882178034,
      "grad_norm": 0.8495082855224609,
      "learning_rate": 2.054919228054577e-05,
      "loss": 0.8336,
      "step": 17064
    },
    {
      "epoch": 3.1533172272811747,
      "grad_norm": 1.0625152587890625,
      "learning_rate": 2.054032155386486e-05,
      "loss": 0.8935,
      "step": 17065
    },
    {
      "epoch": 3.1535030663445456,
      "grad_norm": 0.7919192314147949,
      "learning_rate": 2.0531452523114402e-05,
      "loss": 0.9689,
      "step": 17066
    },
    {
      "epoch": 3.153688905407917,
      "grad_norm": 0.9562473297119141,
      "learning_rate": 2.0522585188483745e-05,
      "loss": 0.8727,
      "step": 17067
    },
    {
      "epoch": 3.153874744471288,
      "grad_norm": 0.9182508587837219,
      "learning_rate": 2.0513719550162114e-05,
      "loss": 1.0428,
      "step": 17068
    },
    {
      "epoch": 3.154060583534659,
      "grad_norm": 1.3862417936325073,
      "learning_rate": 2.0504855608338715e-05,
      "loss": 1.1412,
      "step": 17069
    },
    {
      "epoch": 3.15424642259803,
      "grad_norm": 0.8006358742713928,
      "learning_rate": 2.049599336320277e-05,
      "loss": 0.7885,
      "step": 17070
    },
    {
      "epoch": 3.1544322616614013,
      "grad_norm": 1.0290948152542114,
      "learning_rate": 2.0487132814943422e-05,
      "loss": 0.8757,
      "step": 17071
    },
    {
      "epoch": 3.1546181007247722,
      "grad_norm": 0.8608807921409607,
      "learning_rate": 2.0478273963749774e-05,
      "loss": 0.7615,
      "step": 17072
    },
    {
      "epoch": 3.1548039397881436,
      "grad_norm": 0.9114319086074829,
      "learning_rate": 2.0469416809810872e-05,
      "loss": 0.8203,
      "step": 17073
    },
    {
      "epoch": 3.1549897788515144,
      "grad_norm": 0.9723644852638245,
      "learning_rate": 2.0460561353315788e-05,
      "loss": 0.8033,
      "step": 17074
    },
    {
      "epoch": 3.1551756179148858,
      "grad_norm": 0.8336568474769592,
      "learning_rate": 2.0451707594453562e-05,
      "loss": 0.6984,
      "step": 17075
    },
    {
      "epoch": 3.1553614569782567,
      "grad_norm": 0.9868106245994568,
      "learning_rate": 2.0442855533413117e-05,
      "loss": 0.9259,
      "step": 17076
    },
    {
      "epoch": 3.155547296041628,
      "grad_norm": 0.9859785437583923,
      "learning_rate": 2.0434005170383396e-05,
      "loss": 0.8286,
      "step": 17077
    },
    {
      "epoch": 3.155733135104999,
      "grad_norm": 0.8500993847846985,
      "learning_rate": 2.0425156505553277e-05,
      "loss": 0.8179,
      "step": 17078
    },
    {
      "epoch": 3.15591897416837,
      "grad_norm": 1.1688470840454102,
      "learning_rate": 2.0416309539111654e-05,
      "loss": 0.8003,
      "step": 17079
    },
    {
      "epoch": 3.1561048132317415,
      "grad_norm": 0.9138656854629517,
      "learning_rate": 2.0407464271247335e-05,
      "loss": 0.9918,
      "step": 17080
    },
    {
      "epoch": 3.1562906522951124,
      "grad_norm": 1.0699207782745361,
      "learning_rate": 2.0398620702149072e-05,
      "loss": 0.8225,
      "step": 17081
    },
    {
      "epoch": 3.1564764913584837,
      "grad_norm": 1.0861259698867798,
      "learning_rate": 2.038977883200569e-05,
      "loss": 0.6632,
      "step": 17082
    },
    {
      "epoch": 3.1566623304218546,
      "grad_norm": 1.0681793689727783,
      "learning_rate": 2.0380938661005854e-05,
      "loss": 0.7384,
      "step": 17083
    },
    {
      "epoch": 3.156848169485226,
      "grad_norm": 0.9535313248634338,
      "learning_rate": 2.037210018933825e-05,
      "loss": 0.7299,
      "step": 17084
    },
    {
      "epoch": 3.157034008548597,
      "grad_norm": 0.852402925491333,
      "learning_rate": 2.0363263417191493e-05,
      "loss": 0.9142,
      "step": 17085
    },
    {
      "epoch": 3.157219847611968,
      "grad_norm": 1.0568815469741821,
      "learning_rate": 2.0354428344754217e-05,
      "loss": 0.6579,
      "step": 17086
    },
    {
      "epoch": 3.157405686675339,
      "grad_norm": 0.798697292804718,
      "learning_rate": 2.0345594972215022e-05,
      "loss": 0.7432,
      "step": 17087
    },
    {
      "epoch": 3.1575915257387104,
      "grad_norm": 0.9491557478904724,
      "learning_rate": 2.0336763299762406e-05,
      "loss": 0.8381,
      "step": 17088
    },
    {
      "epoch": 3.1577773648020813,
      "grad_norm": 1.0051136016845703,
      "learning_rate": 2.0327933327584835e-05,
      "loss": 0.9975,
      "step": 17089
    },
    {
      "epoch": 3.1579632038654526,
      "grad_norm": 0.9311573505401611,
      "learning_rate": 2.0319105055870846e-05,
      "loss": 0.6724,
      "step": 17090
    },
    {
      "epoch": 3.1581490429288235,
      "grad_norm": 1.0000122785568237,
      "learning_rate": 2.031027848480881e-05,
      "loss": 0.8165,
      "step": 17091
    },
    {
      "epoch": 3.158334881992195,
      "grad_norm": 0.7631297707557678,
      "learning_rate": 2.030145361458714e-05,
      "loss": 0.7885,
      "step": 17092
    },
    {
      "epoch": 3.1585207210555657,
      "grad_norm": 0.937091588973999,
      "learning_rate": 2.0292630445394133e-05,
      "loss": 0.9293,
      "step": 17093
    },
    {
      "epoch": 3.158706560118937,
      "grad_norm": 0.8916681408882141,
      "learning_rate": 2.0283808977418185e-05,
      "loss": 0.7998,
      "step": 17094
    },
    {
      "epoch": 3.1588923991823084,
      "grad_norm": 0.9731296896934509,
      "learning_rate": 2.0274989210847495e-05,
      "loss": 0.8251,
      "step": 17095
    },
    {
      "epoch": 3.1590782382456792,
      "grad_norm": 1.0815353393554688,
      "learning_rate": 2.0266171145870383e-05,
      "loss": 0.8091,
      "step": 17096
    },
    {
      "epoch": 3.1592640773090506,
      "grad_norm": 0.8604963421821594,
      "learning_rate": 2.025735478267502e-05,
      "loss": 0.7841,
      "step": 17097
    },
    {
      "epoch": 3.1594499163724215,
      "grad_norm": 0.9545714855194092,
      "learning_rate": 2.0248540121449543e-05,
      "loss": 0.8389,
      "step": 17098
    },
    {
      "epoch": 3.159635755435793,
      "grad_norm": 0.8550902009010315,
      "learning_rate": 2.0239727162382148e-05,
      "loss": 0.8468,
      "step": 17099
    },
    {
      "epoch": 3.1598215944991637,
      "grad_norm": 0.9193598628044128,
      "learning_rate": 2.0230915905660906e-05,
      "loss": 0.7489,
      "step": 17100
    },
    {
      "epoch": 3.160007433562535,
      "grad_norm": 1.4833147525787354,
      "learning_rate": 2.0222106351473845e-05,
      "loss": 1.2193,
      "step": 17101
    },
    {
      "epoch": 3.160193272625906,
      "grad_norm": 0.7797554135322571,
      "learning_rate": 2.0213298500009047e-05,
      "loss": 0.6539,
      "step": 17102
    },
    {
      "epoch": 3.160379111689277,
      "grad_norm": 1.5775411128997803,
      "learning_rate": 2.0204492351454472e-05,
      "loss": 1.2373,
      "step": 17103
    },
    {
      "epoch": 3.160564950752648,
      "grad_norm": 0.8735769987106323,
      "learning_rate": 2.0195687905998073e-05,
      "loss": 0.7057,
      "step": 17104
    },
    {
      "epoch": 3.1607507898160194,
      "grad_norm": 0.8621777892112732,
      "learning_rate": 2.018688516382774e-05,
      "loss": 0.8648,
      "step": 17105
    },
    {
      "epoch": 3.1609366288793903,
      "grad_norm": 1.0254769325256348,
      "learning_rate": 2.017808412513137e-05,
      "loss": 0.9406,
      "step": 17106
    },
    {
      "epoch": 3.1611224679427616,
      "grad_norm": 1.0010100603103638,
      "learning_rate": 2.0169284790096853e-05,
      "loss": 0.9229,
      "step": 17107
    },
    {
      "epoch": 3.1613083070061325,
      "grad_norm": 0.906528115272522,
      "learning_rate": 2.0160487158911955e-05,
      "loss": 0.6888,
      "step": 17108
    },
    {
      "epoch": 3.161494146069504,
      "grad_norm": 0.8315296173095703,
      "learning_rate": 2.0151691231764414e-05,
      "loss": 0.8592,
      "step": 17109
    },
    {
      "epoch": 3.1616799851328747,
      "grad_norm": 0.8605505228042603,
      "learning_rate": 2.0142897008842033e-05,
      "loss": 0.7481,
      "step": 17110
    },
    {
      "epoch": 3.161865824196246,
      "grad_norm": 0.9833331108093262,
      "learning_rate": 2.0134104490332474e-05,
      "loss": 0.7347,
      "step": 17111
    },
    {
      "epoch": 3.162051663259617,
      "grad_norm": 0.8601307272911072,
      "learning_rate": 2.0125313676423396e-05,
      "loss": 0.7927,
      "step": 17112
    },
    {
      "epoch": 3.1622375023229883,
      "grad_norm": 0.8588778972625732,
      "learning_rate": 2.011652456730241e-05,
      "loss": 0.7516,
      "step": 17113
    },
    {
      "epoch": 3.1624233413863596,
      "grad_norm": 0.8486018180847168,
      "learning_rate": 2.0107737163157138e-05,
      "loss": 0.8462,
      "step": 17114
    },
    {
      "epoch": 3.1626091804497305,
      "grad_norm": 0.9136857986450195,
      "learning_rate": 2.009895146417512e-05,
      "loss": 0.895,
      "step": 17115
    },
    {
      "epoch": 3.162795019513102,
      "grad_norm": 1.0777807235717773,
      "learning_rate": 2.009016747054384e-05,
      "loss": 0.8649,
      "step": 17116
    },
    {
      "epoch": 3.1629808585764727,
      "grad_norm": 1.7526211738586426,
      "learning_rate": 2.008138518245083e-05,
      "loss": 1.0711,
      "step": 17117
    },
    {
      "epoch": 3.163166697639844,
      "grad_norm": 0.8413838148117065,
      "learning_rate": 2.007260460008348e-05,
      "loss": 0.5849,
      "step": 17118
    },
    {
      "epoch": 3.163352536703215,
      "grad_norm": 0.8410135507583618,
      "learning_rate": 2.0063825723629247e-05,
      "loss": 0.6995,
      "step": 17119
    },
    {
      "epoch": 3.1635383757665863,
      "grad_norm": 0.9260984659194946,
      "learning_rate": 2.0055048553275492e-05,
      "loss": 0.8481,
      "step": 17120
    },
    {
      "epoch": 3.163724214829957,
      "grad_norm": 0.7520155310630798,
      "learning_rate": 2.0046273089209487e-05,
      "loss": 0.4878,
      "step": 17121
    },
    {
      "epoch": 3.1639100538933285,
      "grad_norm": 1.036642074584961,
      "learning_rate": 2.0037499331618627e-05,
      "loss": 0.944,
      "step": 17122
    },
    {
      "epoch": 3.1640958929566994,
      "grad_norm": 1.0706443786621094,
      "learning_rate": 2.0028727280690107e-05,
      "loss": 0.6616,
      "step": 17123
    },
    {
      "epoch": 3.1642817320200707,
      "grad_norm": 0.9441154599189758,
      "learning_rate": 2.0019956936611172e-05,
      "loss": 0.8402,
      "step": 17124
    },
    {
      "epoch": 3.1644675710834416,
      "grad_norm": 0.9219971299171448,
      "learning_rate": 2.0011188299568973e-05,
      "loss": 0.8124,
      "step": 17125
    },
    {
      "epoch": 3.164653410146813,
      "grad_norm": 1.2289496660232544,
      "learning_rate": 2.0002421369750723e-05,
      "loss": 0.9161,
      "step": 17126
    },
    {
      "epoch": 3.164839249210184,
      "grad_norm": 0.9382373690605164,
      "learning_rate": 1.9993656147343486e-05,
      "loss": 0.9322,
      "step": 17127
    },
    {
      "epoch": 3.165025088273555,
      "grad_norm": 1.0162302255630493,
      "learning_rate": 1.998489263253438e-05,
      "loss": 0.7234,
      "step": 17128
    },
    {
      "epoch": 3.1652109273369264,
      "grad_norm": 0.8443132638931274,
      "learning_rate": 1.9976130825510408e-05,
      "loss": 0.8587,
      "step": 17129
    },
    {
      "epoch": 3.1653967664002973,
      "grad_norm": 1.1633564233779907,
      "learning_rate": 1.9967370726458623e-05,
      "loss": 0.9895,
      "step": 17130
    },
    {
      "epoch": 3.1655826054636687,
      "grad_norm": 1.0444053411483765,
      "learning_rate": 1.9958612335565975e-05,
      "loss": 0.931,
      "step": 17131
    },
    {
      "epoch": 3.1657684445270395,
      "grad_norm": 1.5743813514709473,
      "learning_rate": 1.9949855653019377e-05,
      "loss": 1.0989,
      "step": 17132
    },
    {
      "epoch": 3.165954283590411,
      "grad_norm": 0.9263360500335693,
      "learning_rate": 1.994110067900572e-05,
      "loss": 0.6137,
      "step": 17133
    },
    {
      "epoch": 3.1661401226537818,
      "grad_norm": 0.979530394077301,
      "learning_rate": 1.9932347413711905e-05,
      "loss": 0.9215,
      "step": 17134
    },
    {
      "epoch": 3.166325961717153,
      "grad_norm": 0.752700924873352,
      "learning_rate": 1.992359585732474e-05,
      "loss": 0.6815,
      "step": 17135
    },
    {
      "epoch": 3.166511800780524,
      "grad_norm": 1.1921418905258179,
      "learning_rate": 1.9914846010030995e-05,
      "loss": 0.8916,
      "step": 17136
    },
    {
      "epoch": 3.1666976398438953,
      "grad_norm": 0.8520069122314453,
      "learning_rate": 1.9906097872017415e-05,
      "loss": 0.7906,
      "step": 17137
    },
    {
      "epoch": 3.166883478907266,
      "grad_norm": 0.8959609866142273,
      "learning_rate": 1.9897351443470723e-05,
      "loss": 0.8337,
      "step": 17138
    },
    {
      "epoch": 3.1670693179706375,
      "grad_norm": 1.0658314228057861,
      "learning_rate": 1.988860672457763e-05,
      "loss": 0.8374,
      "step": 17139
    },
    {
      "epoch": 3.1672551570340084,
      "grad_norm": 0.938890278339386,
      "learning_rate": 1.9879863715524748e-05,
      "loss": 0.7187,
      "step": 17140
    },
    {
      "epoch": 3.1674409960973797,
      "grad_norm": 0.9903023838996887,
      "learning_rate": 1.9871122416498657e-05,
      "loss": 0.7023,
      "step": 17141
    },
    {
      "epoch": 3.1676268351607506,
      "grad_norm": 1.1366755962371826,
      "learning_rate": 1.986238282768598e-05,
      "loss": 0.6451,
      "step": 17142
    },
    {
      "epoch": 3.167812674224122,
      "grad_norm": 0.9639435410499573,
      "learning_rate": 1.9853644949273222e-05,
      "loss": 0.9946,
      "step": 17143
    },
    {
      "epoch": 3.167998513287493,
      "grad_norm": 0.9725199341773987,
      "learning_rate": 1.984490878144687e-05,
      "loss": 1.0356,
      "step": 17144
    },
    {
      "epoch": 3.168184352350864,
      "grad_norm": 0.8418335318565369,
      "learning_rate": 1.9836174324393365e-05,
      "loss": 0.8344,
      "step": 17145
    },
    {
      "epoch": 3.168370191414235,
      "grad_norm": 0.9876839518547058,
      "learning_rate": 1.9827441578299176e-05,
      "loss": 0.7208,
      "step": 17146
    },
    {
      "epoch": 3.1685560304776064,
      "grad_norm": 0.8322383165359497,
      "learning_rate": 1.981871054335067e-05,
      "loss": 0.6992,
      "step": 17147
    },
    {
      "epoch": 3.1687418695409777,
      "grad_norm": 1.0060124397277832,
      "learning_rate": 1.9809981219734163e-05,
      "loss": 0.6777,
      "step": 17148
    },
    {
      "epoch": 3.1689277086043486,
      "grad_norm": 0.933151364326477,
      "learning_rate": 1.9801253607635996e-05,
      "loss": 0.6345,
      "step": 17149
    },
    {
      "epoch": 3.16911354766772,
      "grad_norm": 1.0822348594665527,
      "learning_rate": 1.979252770724247e-05,
      "loss": 0.8977,
      "step": 17150
    },
    {
      "epoch": 3.169299386731091,
      "grad_norm": 0.8258489370346069,
      "learning_rate": 1.9783803518739796e-05,
      "loss": 0.6308,
      "step": 17151
    },
    {
      "epoch": 3.169485225794462,
      "grad_norm": 1.0155291557312012,
      "learning_rate": 1.977508104231418e-05,
      "loss": 0.9677,
      "step": 17152
    },
    {
      "epoch": 3.169671064857833,
      "grad_norm": 0.9031057953834534,
      "learning_rate": 1.9766360278151764e-05,
      "loss": 0.8015,
      "step": 17153
    },
    {
      "epoch": 3.1698569039212043,
      "grad_norm": 1.0259935855865479,
      "learning_rate": 1.9757641226438728e-05,
      "loss": 0.9574,
      "step": 17154
    },
    {
      "epoch": 3.1700427429845752,
      "grad_norm": 0.8801981210708618,
      "learning_rate": 1.9748923887361137e-05,
      "loss": 0.7554,
      "step": 17155
    },
    {
      "epoch": 3.1702285820479466,
      "grad_norm": 0.8430247902870178,
      "learning_rate": 1.974020826110504e-05,
      "loss": 0.7563,
      "step": 17156
    },
    {
      "epoch": 3.1704144211113174,
      "grad_norm": 0.9828110337257385,
      "learning_rate": 1.973149434785645e-05,
      "loss": 0.8258,
      "step": 17157
    },
    {
      "epoch": 3.1706002601746888,
      "grad_norm": 0.935191810131073,
      "learning_rate": 1.9722782147801365e-05,
      "loss": 0.7948,
      "step": 17158
    },
    {
      "epoch": 3.1707860992380597,
      "grad_norm": 1.270983338356018,
      "learning_rate": 1.9714071661125765e-05,
      "loss": 0.9351,
      "step": 17159
    },
    {
      "epoch": 3.170971938301431,
      "grad_norm": 0.9522151350975037,
      "learning_rate": 1.970536288801552e-05,
      "loss": 0.7877,
      "step": 17160
    },
    {
      "epoch": 3.171157777364802,
      "grad_norm": 0.7604349255561829,
      "learning_rate": 1.9696655828656497e-05,
      "loss": 0.5488,
      "step": 17161
    },
    {
      "epoch": 3.171343616428173,
      "grad_norm": 1.1275665760040283,
      "learning_rate": 1.9687950483234575e-05,
      "loss": 0.7563,
      "step": 17162
    },
    {
      "epoch": 3.1715294554915445,
      "grad_norm": 1.044658899307251,
      "learning_rate": 1.967924685193552e-05,
      "loss": 0.8459,
      "step": 17163
    },
    {
      "epoch": 3.1717152945549154,
      "grad_norm": 0.8572118282318115,
      "learning_rate": 1.9670544934945112e-05,
      "loss": 0.7864,
      "step": 17164
    },
    {
      "epoch": 3.1719011336182867,
      "grad_norm": 0.9118345975875854,
      "learning_rate": 1.9661844732449043e-05,
      "loss": 0.7072,
      "step": 17165
    },
    {
      "epoch": 3.1720869726816576,
      "grad_norm": 1.1098875999450684,
      "learning_rate": 1.965314624463306e-05,
      "loss": 1.0089,
      "step": 17166
    },
    {
      "epoch": 3.172272811745029,
      "grad_norm": 1.6386604309082031,
      "learning_rate": 1.964444947168278e-05,
      "loss": 0.7754,
      "step": 17167
    },
    {
      "epoch": 3.1724586508084,
      "grad_norm": 0.8848221898078918,
      "learning_rate": 1.9635754413783835e-05,
      "loss": 0.7577,
      "step": 17168
    },
    {
      "epoch": 3.172644489871771,
      "grad_norm": 1.0232877731323242,
      "learning_rate": 1.9627061071121755e-05,
      "loss": 0.8765,
      "step": 17169
    },
    {
      "epoch": 3.172830328935142,
      "grad_norm": 0.9133501648902893,
      "learning_rate": 1.961836944388218e-05,
      "loss": 0.7751,
      "step": 17170
    },
    {
      "epoch": 3.1730161679985134,
      "grad_norm": 1.5902894735336304,
      "learning_rate": 1.960967953225057e-05,
      "loss": 1.5975,
      "step": 17171
    },
    {
      "epoch": 3.1732020070618843,
      "grad_norm": 0.847976565361023,
      "learning_rate": 1.96009913364124e-05,
      "loss": 0.9335,
      "step": 17172
    },
    {
      "epoch": 3.1733878461252556,
      "grad_norm": 0.9059402942657471,
      "learning_rate": 1.959230485655307e-05,
      "loss": 0.8474,
      "step": 17173
    },
    {
      "epoch": 3.1735736851886265,
      "grad_norm": 0.8859007358551025,
      "learning_rate": 1.9583620092858025e-05,
      "loss": 0.7758,
      "step": 17174
    },
    {
      "epoch": 3.173759524251998,
      "grad_norm": 0.8168259263038635,
      "learning_rate": 1.957493704551262e-05,
      "loss": 0.6757,
      "step": 17175
    },
    {
      "epoch": 3.1739453633153687,
      "grad_norm": 0.82487952709198,
      "learning_rate": 1.9566255714702165e-05,
      "loss": 0.5136,
      "step": 17176
    },
    {
      "epoch": 3.17413120237874,
      "grad_norm": 0.9470896124839783,
      "learning_rate": 1.9557576100611917e-05,
      "loss": 0.767,
      "step": 17177
    },
    {
      "epoch": 3.1743170414421114,
      "grad_norm": 1.0997936725616455,
      "learning_rate": 1.9548898203427203e-05,
      "loss": 0.6879,
      "step": 17178
    },
    {
      "epoch": 3.1745028805054822,
      "grad_norm": 0.9015858173370361,
      "learning_rate": 1.9540222023333166e-05,
      "loss": 0.7374,
      "step": 17179
    },
    {
      "epoch": 3.1746887195688536,
      "grad_norm": 0.8093281388282776,
      "learning_rate": 1.953154756051504e-05,
      "loss": 0.5783,
      "step": 17180
    },
    {
      "epoch": 3.1748745586322245,
      "grad_norm": 1.0351872444152832,
      "learning_rate": 1.952287481515792e-05,
      "loss": 0.9301,
      "step": 17181
    },
    {
      "epoch": 3.175060397695596,
      "grad_norm": 1.540510654449463,
      "learning_rate": 1.951420378744695e-05,
      "loss": 1.0627,
      "step": 17182
    },
    {
      "epoch": 3.1752462367589667,
      "grad_norm": 0.8963640332221985,
      "learning_rate": 1.950553447756719e-05,
      "loss": 0.9015,
      "step": 17183
    },
    {
      "epoch": 3.175432075822338,
      "grad_norm": 0.9509427547454834,
      "learning_rate": 1.9496866885703657e-05,
      "loss": 0.6713,
      "step": 17184
    },
    {
      "epoch": 3.175617914885709,
      "grad_norm": 1.0596952438354492,
      "learning_rate": 1.948820101204133e-05,
      "loss": 0.9011,
      "step": 17185
    },
    {
      "epoch": 3.17580375394908,
      "grad_norm": 0.9688878059387207,
      "learning_rate": 1.9479536856765214e-05,
      "loss": 0.9906,
      "step": 17186
    },
    {
      "epoch": 3.175989593012451,
      "grad_norm": 1.1700866222381592,
      "learning_rate": 1.9470874420060202e-05,
      "loss": 0.8296,
      "step": 17187
    },
    {
      "epoch": 3.1761754320758224,
      "grad_norm": 0.825894296169281,
      "learning_rate": 1.9462213702111177e-05,
      "loss": 0.8445,
      "step": 17188
    },
    {
      "epoch": 3.1763612711391933,
      "grad_norm": 0.8722351789474487,
      "learning_rate": 1.945355470310297e-05,
      "loss": 0.8353,
      "step": 17189
    },
    {
      "epoch": 3.1765471102025646,
      "grad_norm": 0.8865044713020325,
      "learning_rate": 1.944489742322042e-05,
      "loss": 0.8187,
      "step": 17190
    },
    {
      "epoch": 3.1767329492659355,
      "grad_norm": 0.9434642791748047,
      "learning_rate": 1.943624186264832e-05,
      "loss": 0.8013,
      "step": 17191
    },
    {
      "epoch": 3.176918788329307,
      "grad_norm": 0.9115262627601624,
      "learning_rate": 1.9427588021571385e-05,
      "loss": 0.8178,
      "step": 17192
    },
    {
      "epoch": 3.1771046273926777,
      "grad_norm": 1.1289523839950562,
      "learning_rate": 1.9418935900174284e-05,
      "loss": 0.8025,
      "step": 17193
    },
    {
      "epoch": 3.177290466456049,
      "grad_norm": 0.7821084260940552,
      "learning_rate": 1.9410285498641745e-05,
      "loss": 0.7787,
      "step": 17194
    },
    {
      "epoch": 3.17747630551942,
      "grad_norm": 0.9180419445037842,
      "learning_rate": 1.9401636817158365e-05,
      "loss": 0.8752,
      "step": 17195
    },
    {
      "epoch": 3.1776621445827913,
      "grad_norm": 1.1209893226623535,
      "learning_rate": 1.939298985590874e-05,
      "loss": 0.7798,
      "step": 17196
    },
    {
      "epoch": 3.1778479836461626,
      "grad_norm": 0.9026002883911133,
      "learning_rate": 1.9384344615077386e-05,
      "loss": 0.9728,
      "step": 17197
    },
    {
      "epoch": 3.1780338227095335,
      "grad_norm": 0.9963973760604858,
      "learning_rate": 1.9375701094848873e-05,
      "loss": 0.8824,
      "step": 17198
    },
    {
      "epoch": 3.178219661772905,
      "grad_norm": 0.9234552383422852,
      "learning_rate": 1.9367059295407674e-05,
      "loss": 0.7089,
      "step": 17199
    },
    {
      "epoch": 3.1784055008362757,
      "grad_norm": 0.892646849155426,
      "learning_rate": 1.9358419216938195e-05,
      "loss": 0.5419,
      "step": 17200
    },
    {
      "epoch": 3.178591339899647,
      "grad_norm": 0.9919077157974243,
      "learning_rate": 1.9349780859624865e-05,
      "loss": 0.8166,
      "step": 17201
    },
    {
      "epoch": 3.178777178963018,
      "grad_norm": 1.040787935256958,
      "learning_rate": 1.9341144223652098e-05,
      "loss": 0.9313,
      "step": 17202
    },
    {
      "epoch": 3.1789630180263893,
      "grad_norm": 1.034578800201416,
      "learning_rate": 1.9332509309204183e-05,
      "loss": 1.0048,
      "step": 17203
    },
    {
      "epoch": 3.17914885708976,
      "grad_norm": 1.0032830238342285,
      "learning_rate": 1.9323876116465435e-05,
      "loss": 0.9563,
      "step": 17204
    },
    {
      "epoch": 3.1793346961531315,
      "grad_norm": 1.0224652290344238,
      "learning_rate": 1.9315244645620067e-05,
      "loss": 1.0332,
      "step": 17205
    },
    {
      "epoch": 3.1795205352165024,
      "grad_norm": 0.95107501745224,
      "learning_rate": 1.9306614896852383e-05,
      "loss": 0.7288,
      "step": 17206
    },
    {
      "epoch": 3.1797063742798737,
      "grad_norm": 0.8130441904067993,
      "learning_rate": 1.929798687034652e-05,
      "loss": 0.8187,
      "step": 17207
    },
    {
      "epoch": 3.1798922133432446,
      "grad_norm": 1.0477732419967651,
      "learning_rate": 1.928936056628663e-05,
      "loss": 0.858,
      "step": 17208
    },
    {
      "epoch": 3.180078052406616,
      "grad_norm": 0.8282469511032104,
      "learning_rate": 1.928073598485681e-05,
      "loss": 0.858,
      "step": 17209
    },
    {
      "epoch": 3.180263891469987,
      "grad_norm": 0.975362241268158,
      "learning_rate": 1.9272113126241198e-05,
      "loss": 0.8719,
      "step": 17210
    },
    {
      "epoch": 3.180449730533358,
      "grad_norm": 1.0730140209197998,
      "learning_rate": 1.926349199062376e-05,
      "loss": 1.0322,
      "step": 17211
    },
    {
      "epoch": 3.1806355695967294,
      "grad_norm": 0.8283026814460754,
      "learning_rate": 1.9254872578188567e-05,
      "loss": 0.9431,
      "step": 17212
    },
    {
      "epoch": 3.1808214086601003,
      "grad_norm": 1.0434376001358032,
      "learning_rate": 1.924625488911953e-05,
      "loss": 0.9381,
      "step": 17213
    },
    {
      "epoch": 3.1810072477234717,
      "grad_norm": 0.8632153868675232,
      "learning_rate": 1.923763892360062e-05,
      "loss": 0.9113,
      "step": 17214
    },
    {
      "epoch": 3.1811930867868425,
      "grad_norm": 0.7788403630256653,
      "learning_rate": 1.9229024681815722e-05,
      "loss": 0.7617,
      "step": 17215
    },
    {
      "epoch": 3.181378925850214,
      "grad_norm": 0.9866867065429688,
      "learning_rate": 1.9220412163948688e-05,
      "loss": 0.84,
      "step": 17216
    },
    {
      "epoch": 3.1815647649135848,
      "grad_norm": 0.8431900143623352,
      "learning_rate": 1.9211801370183292e-05,
      "loss": 0.7343,
      "step": 17217
    },
    {
      "epoch": 3.181750603976956,
      "grad_norm": 1.0417009592056274,
      "learning_rate": 1.92031923007034e-05,
      "loss": 0.6948,
      "step": 17218
    },
    {
      "epoch": 3.181936443040327,
      "grad_norm": 0.8335167169570923,
      "learning_rate": 1.9194584955692706e-05,
      "loss": 0.6104,
      "step": 17219
    },
    {
      "epoch": 3.1821222821036983,
      "grad_norm": 0.8310338258743286,
      "learning_rate": 1.9185979335334904e-05,
      "loss": 0.5485,
      "step": 17220
    },
    {
      "epoch": 3.182308121167069,
      "grad_norm": 0.9344164729118347,
      "learning_rate": 1.9177375439813684e-05,
      "loss": 0.9334,
      "step": 17221
    },
    {
      "epoch": 3.1824939602304405,
      "grad_norm": 0.9417360424995422,
      "learning_rate": 1.9168773269312724e-05,
      "loss": 0.7755,
      "step": 17222
    },
    {
      "epoch": 3.1826797992938114,
      "grad_norm": 1.168516755104065,
      "learning_rate": 1.9160172824015586e-05,
      "loss": 0.9884,
      "step": 17223
    },
    {
      "epoch": 3.1828656383571827,
      "grad_norm": 0.8864932060241699,
      "learning_rate": 1.9151574104105828e-05,
      "loss": 1.0201,
      "step": 17224
    },
    {
      "epoch": 3.1830514774205536,
      "grad_norm": 0.8255242705345154,
      "learning_rate": 1.914297710976696e-05,
      "loss": 0.8236,
      "step": 17225
    },
    {
      "epoch": 3.183237316483925,
      "grad_norm": 0.8763002157211304,
      "learning_rate": 1.9134381841182503e-05,
      "loss": 0.6465,
      "step": 17226
    },
    {
      "epoch": 3.1834231555472963,
      "grad_norm": 0.9340523481369019,
      "learning_rate": 1.9125788298535908e-05,
      "loss": 0.6152,
      "step": 17227
    },
    {
      "epoch": 3.183608994610667,
      "grad_norm": 1.1670637130737305,
      "learning_rate": 1.911719648201057e-05,
      "loss": 0.8589,
      "step": 17228
    },
    {
      "epoch": 3.1837948336740385,
      "grad_norm": 0.7700712084770203,
      "learning_rate": 1.910860639178984e-05,
      "loss": 0.8393,
      "step": 17229
    },
    {
      "epoch": 3.1839806727374094,
      "grad_norm": 1.044851541519165,
      "learning_rate": 1.9100018028057132e-05,
      "loss": 0.8374,
      "step": 17230
    },
    {
      "epoch": 3.1841665118007807,
      "grad_norm": 1.1065137386322021,
      "learning_rate": 1.9091431390995697e-05,
      "loss": 0.7924,
      "step": 17231
    },
    {
      "epoch": 3.1843523508641516,
      "grad_norm": 1.4392445087432861,
      "learning_rate": 1.9082846480788795e-05,
      "loss": 0.6219,
      "step": 17232
    },
    {
      "epoch": 3.184538189927523,
      "grad_norm": 1.009527325630188,
      "learning_rate": 1.9074263297619676e-05,
      "loss": 0.8973,
      "step": 17233
    },
    {
      "epoch": 3.184724028990894,
      "grad_norm": 0.8259090185165405,
      "learning_rate": 1.9065681841671558e-05,
      "loss": 0.7303,
      "step": 17234
    },
    {
      "epoch": 3.184909868054265,
      "grad_norm": 0.8061724305152893,
      "learning_rate": 1.905710211312757e-05,
      "loss": 0.7925,
      "step": 17235
    },
    {
      "epoch": 3.185095707117636,
      "grad_norm": 0.8986761569976807,
      "learning_rate": 1.904852411217083e-05,
      "loss": 0.7326,
      "step": 17236
    },
    {
      "epoch": 3.1852815461810073,
      "grad_norm": 0.910462498664856,
      "learning_rate": 1.9039947838984405e-05,
      "loss": 0.9161,
      "step": 17237
    },
    {
      "epoch": 3.1854673852443782,
      "grad_norm": 0.901995062828064,
      "learning_rate": 1.9031373293751377e-05,
      "loss": 0.6487,
      "step": 17238
    },
    {
      "epoch": 3.1856532243077496,
      "grad_norm": 0.8945409655570984,
      "learning_rate": 1.9022800476654744e-05,
      "loss": 0.8433,
      "step": 17239
    },
    {
      "epoch": 3.1858390633711204,
      "grad_norm": 0.8279741406440735,
      "learning_rate": 1.9014229387877448e-05,
      "loss": 0.7384,
      "step": 17240
    },
    {
      "epoch": 3.1860249024344918,
      "grad_norm": 0.9453520774841309,
      "learning_rate": 1.900566002760248e-05,
      "loss": 0.7548,
      "step": 17241
    },
    {
      "epoch": 3.1862107414978627,
      "grad_norm": 0.8394404053688049,
      "learning_rate": 1.8997092396012673e-05,
      "loss": 0.831,
      "step": 17242
    },
    {
      "epoch": 3.186396580561234,
      "grad_norm": 0.7381008863449097,
      "learning_rate": 1.898852649329095e-05,
      "loss": 0.661,
      "step": 17243
    },
    {
      "epoch": 3.186582419624605,
      "grad_norm": 0.7975539565086365,
      "learning_rate": 1.8979962319620104e-05,
      "loss": 0.739,
      "step": 17244
    },
    {
      "epoch": 3.186768258687976,
      "grad_norm": 0.9062146544456482,
      "learning_rate": 1.8971399875182905e-05,
      "loss": 0.7843,
      "step": 17245
    },
    {
      "epoch": 3.1869540977513475,
      "grad_norm": 0.8561675548553467,
      "learning_rate": 1.896283916016216e-05,
      "loss": 0.6319,
      "step": 17246
    },
    {
      "epoch": 3.1871399368147184,
      "grad_norm": 1.153652310371399,
      "learning_rate": 1.8954280174740537e-05,
      "loss": 0.8982,
      "step": 17247
    },
    {
      "epoch": 3.1873257758780897,
      "grad_norm": 0.8147569298744202,
      "learning_rate": 1.8945722919100717e-05,
      "loss": 0.6174,
      "step": 17248
    },
    {
      "epoch": 3.1875116149414606,
      "grad_norm": 0.8925120234489441,
      "learning_rate": 1.8937167393425335e-05,
      "loss": 0.8004,
      "step": 17249
    },
    {
      "epoch": 3.187697454004832,
      "grad_norm": 0.9293339848518372,
      "learning_rate": 1.8928613597897026e-05,
      "loss": 0.9017,
      "step": 17250
    },
    {
      "epoch": 3.187883293068203,
      "grad_norm": 0.9332923293113708,
      "learning_rate": 1.892006153269833e-05,
      "loss": 0.9528,
      "step": 17251
    },
    {
      "epoch": 3.188069132131574,
      "grad_norm": 0.9275184273719788,
      "learning_rate": 1.8911511198011766e-05,
      "loss": 1.0236,
      "step": 17252
    },
    {
      "epoch": 3.188254971194945,
      "grad_norm": 0.8863673210144043,
      "learning_rate": 1.8902962594019836e-05,
      "loss": 0.9379,
      "step": 17253
    },
    {
      "epoch": 3.1884408102583164,
      "grad_norm": 0.7908885478973389,
      "learning_rate": 1.8894415720905023e-05,
      "loss": 0.8286,
      "step": 17254
    },
    {
      "epoch": 3.1886266493216873,
      "grad_norm": 0.9230900406837463,
      "learning_rate": 1.888587057884973e-05,
      "loss": 0.9145,
      "step": 17255
    },
    {
      "epoch": 3.1888124883850586,
      "grad_norm": 0.8644905090332031,
      "learning_rate": 1.887732716803633e-05,
      "loss": 0.7901,
      "step": 17256
    },
    {
      "epoch": 3.1889983274484295,
      "grad_norm": 0.8076888918876648,
      "learning_rate": 1.8868785488647145e-05,
      "loss": 0.7581,
      "step": 17257
    },
    {
      "epoch": 3.189184166511801,
      "grad_norm": 0.994051456451416,
      "learning_rate": 1.886024554086453e-05,
      "loss": 0.7661,
      "step": 17258
    },
    {
      "epoch": 3.1893700055751717,
      "grad_norm": 0.8983469605445862,
      "learning_rate": 1.885170732487074e-05,
      "loss": 0.6752,
      "step": 17259
    },
    {
      "epoch": 3.189555844638543,
      "grad_norm": 0.7368454337120056,
      "learning_rate": 1.8843170840847968e-05,
      "loss": 0.6503,
      "step": 17260
    },
    {
      "epoch": 3.1897416837019144,
      "grad_norm": 0.9642093777656555,
      "learning_rate": 1.8834636088978476e-05,
      "loss": 0.8498,
      "step": 17261
    },
    {
      "epoch": 3.1899275227652852,
      "grad_norm": 0.9174643158912659,
      "learning_rate": 1.882610306944438e-05,
      "loss": 0.9256,
      "step": 17262
    },
    {
      "epoch": 3.1901133618286566,
      "grad_norm": 0.9920444488525391,
      "learning_rate": 1.88175717824278e-05,
      "loss": 0.6953,
      "step": 17263
    },
    {
      "epoch": 3.1902992008920275,
      "grad_norm": 0.9611275792121887,
      "learning_rate": 1.8809042228110852e-05,
      "loss": 0.7977,
      "step": 17264
    },
    {
      "epoch": 3.190485039955399,
      "grad_norm": 1.028159737586975,
      "learning_rate": 1.880051440667555e-05,
      "loss": 1.0349,
      "step": 17265
    },
    {
      "epoch": 3.1906708790187697,
      "grad_norm": 0.9539807438850403,
      "learning_rate": 1.8791988318303943e-05,
      "loss": 0.6513,
      "step": 17266
    },
    {
      "epoch": 3.190856718082141,
      "grad_norm": 0.9756878018379211,
      "learning_rate": 1.8783463963177993e-05,
      "loss": 0.6905,
      "step": 17267
    },
    {
      "epoch": 3.191042557145512,
      "grad_norm": 1.1659778356552124,
      "learning_rate": 1.8774941341479623e-05,
      "loss": 0.9444,
      "step": 17268
    },
    {
      "epoch": 3.191228396208883,
      "grad_norm": 0.9837881922721863,
      "learning_rate": 1.876642045339073e-05,
      "loss": 0.9958,
      "step": 17269
    },
    {
      "epoch": 3.191414235272254,
      "grad_norm": 1.0949634313583374,
      "learning_rate": 1.8757901299093207e-05,
      "loss": 0.939,
      "step": 17270
    },
    {
      "epoch": 3.1916000743356254,
      "grad_norm": 1.1174333095550537,
      "learning_rate": 1.8749383878768867e-05,
      "loss": 0.7475,
      "step": 17271
    },
    {
      "epoch": 3.1917859133989963,
      "grad_norm": 0.8240452408790588,
      "learning_rate": 1.874086819259947e-05,
      "loss": 0.9406,
      "step": 17272
    },
    {
      "epoch": 3.1919717524623676,
      "grad_norm": 1.0166107416152954,
      "learning_rate": 1.8732354240766813e-05,
      "loss": 0.9246,
      "step": 17273
    },
    {
      "epoch": 3.1921575915257385,
      "grad_norm": 0.9158414602279663,
      "learning_rate": 1.8723842023452577e-05,
      "loss": 0.8681,
      "step": 17274
    },
    {
      "epoch": 3.19234343058911,
      "grad_norm": 0.9893122911453247,
      "learning_rate": 1.8715331540838487e-05,
      "loss": 0.821,
      "step": 17275
    },
    {
      "epoch": 3.192529269652481,
      "grad_norm": 0.8084926009178162,
      "learning_rate": 1.870682279310615e-05,
      "loss": 0.6835,
      "step": 17276
    },
    {
      "epoch": 3.192715108715852,
      "grad_norm": 1.0281102657318115,
      "learning_rate": 1.8698315780437148e-05,
      "loss": 0.7458,
      "step": 17277
    },
    {
      "epoch": 3.1929009477792234,
      "grad_norm": 0.8300309181213379,
      "learning_rate": 1.8689810503013107e-05,
      "loss": 0.6879,
      "step": 17278
    },
    {
      "epoch": 3.1930867868425943,
      "grad_norm": 0.9366108179092407,
      "learning_rate": 1.868130696101552e-05,
      "loss": 0.7827,
      "step": 17279
    },
    {
      "epoch": 3.1932726259059656,
      "grad_norm": 0.8351370096206665,
      "learning_rate": 1.867280515462586e-05,
      "loss": 0.971,
      "step": 17280
    },
    {
      "epoch": 3.1934584649693365,
      "grad_norm": 0.9481558799743652,
      "learning_rate": 1.8664305084025645e-05,
      "loss": 0.5461,
      "step": 17281
    },
    {
      "epoch": 3.193644304032708,
      "grad_norm": 0.8852013945579529,
      "learning_rate": 1.8655806749396254e-05,
      "loss": 0.8311,
      "step": 17282
    },
    {
      "epoch": 3.1938301430960787,
      "grad_norm": 0.8930564522743225,
      "learning_rate": 1.8647310150919083e-05,
      "loss": 0.6868,
      "step": 17283
    },
    {
      "epoch": 3.19401598215945,
      "grad_norm": 0.9694168567657471,
      "learning_rate": 1.8638815288775436e-05,
      "loss": 0.8869,
      "step": 17284
    },
    {
      "epoch": 3.194201821222821,
      "grad_norm": 1.0066895484924316,
      "learning_rate": 1.8630322163146664e-05,
      "loss": 0.6339,
      "step": 17285
    },
    {
      "epoch": 3.1943876602861923,
      "grad_norm": 1.0214611291885376,
      "learning_rate": 1.8621830774214044e-05,
      "loss": 0.6687,
      "step": 17286
    },
    {
      "epoch": 3.194573499349563,
      "grad_norm": 1.0388553142547607,
      "learning_rate": 1.8613341122158813e-05,
      "loss": 0.7142,
      "step": 17287
    },
    {
      "epoch": 3.1947593384129345,
      "grad_norm": 0.9885718822479248,
      "learning_rate": 1.860485320716213e-05,
      "loss": 0.8672,
      "step": 17288
    },
    {
      "epoch": 3.1949451774763054,
      "grad_norm": 1.0320250988006592,
      "learning_rate": 1.859636702940516e-05,
      "loss": 0.7929,
      "step": 17289
    },
    {
      "epoch": 3.1951310165396767,
      "grad_norm": 1.0271234512329102,
      "learning_rate": 1.8587882589069073e-05,
      "loss": 0.7322,
      "step": 17290
    },
    {
      "epoch": 3.1953168556030476,
      "grad_norm": 0.719939112663269,
      "learning_rate": 1.8579399886334914e-05,
      "loss": 0.6652,
      "step": 17291
    },
    {
      "epoch": 3.195502694666419,
      "grad_norm": 0.9527273178100586,
      "learning_rate": 1.857091892138372e-05,
      "loss": 0.7538,
      "step": 17292
    },
    {
      "epoch": 3.19568853372979,
      "grad_norm": 0.912845253944397,
      "learning_rate": 1.856243969439655e-05,
      "loss": 0.9217,
      "step": 17293
    },
    {
      "epoch": 3.195874372793161,
      "grad_norm": 1.1026488542556763,
      "learning_rate": 1.8553962205554352e-05,
      "loss": 0.8682,
      "step": 17294
    },
    {
      "epoch": 3.1960602118565324,
      "grad_norm": 0.809587299823761,
      "learning_rate": 1.854548645503803e-05,
      "loss": 0.6614,
      "step": 17295
    },
    {
      "epoch": 3.1962460509199033,
      "grad_norm": 0.9222704768180847,
      "learning_rate": 1.8537012443028557e-05,
      "loss": 0.9814,
      "step": 17296
    },
    {
      "epoch": 3.1964318899832747,
      "grad_norm": 0.8720242381095886,
      "learning_rate": 1.852854016970672e-05,
      "loss": 0.8118,
      "step": 17297
    },
    {
      "epoch": 3.1966177290466455,
      "grad_norm": 0.8971609473228455,
      "learning_rate": 1.8520069635253414e-05,
      "loss": 0.7165,
      "step": 17298
    },
    {
      "epoch": 3.196803568110017,
      "grad_norm": 1.086275339126587,
      "learning_rate": 1.8511600839849397e-05,
      "loss": 0.9747,
      "step": 17299
    },
    {
      "epoch": 3.1969894071733878,
      "grad_norm": 0.993175745010376,
      "learning_rate": 1.8503133783675397e-05,
      "loss": 0.9135,
      "step": 17300
    },
    {
      "epoch": 3.197175246236759,
      "grad_norm": 1.087994933128357,
      "learning_rate": 1.8494668466912167e-05,
      "loss": 0.8,
      "step": 17301
    },
    {
      "epoch": 3.19736108530013,
      "grad_norm": 0.9241663813591003,
      "learning_rate": 1.8486204889740376e-05,
      "loss": 0.8348,
      "step": 17302
    },
    {
      "epoch": 3.1975469243635013,
      "grad_norm": 1.007936954498291,
      "learning_rate": 1.8477743052340668e-05,
      "loss": 0.7418,
      "step": 17303
    },
    {
      "epoch": 3.197732763426872,
      "grad_norm": 1.0289450883865356,
      "learning_rate": 1.8469282954893598e-05,
      "loss": 0.7169,
      "step": 17304
    },
    {
      "epoch": 3.1979186024902435,
      "grad_norm": 0.8627272248268127,
      "learning_rate": 1.8460824597579773e-05,
      "loss": 0.8952,
      "step": 17305
    },
    {
      "epoch": 3.1981044415536144,
      "grad_norm": 0.815715491771698,
      "learning_rate": 1.8452367980579756e-05,
      "loss": 0.8715,
      "step": 17306
    },
    {
      "epoch": 3.1982902806169857,
      "grad_norm": 1.161781907081604,
      "learning_rate": 1.8443913104073983e-05,
      "loss": 0.7697,
      "step": 17307
    },
    {
      "epoch": 3.1984761196803566,
      "grad_norm": 0.9430105686187744,
      "learning_rate": 1.8435459968242953e-05,
      "loss": 0.6636,
      "step": 17308
    },
    {
      "epoch": 3.198661958743728,
      "grad_norm": 1.0547893047332764,
      "learning_rate": 1.8427008573267013e-05,
      "loss": 0.7107,
      "step": 17309
    },
    {
      "epoch": 3.1988477978070993,
      "grad_norm": 0.8477012515068054,
      "learning_rate": 1.8418558919326635e-05,
      "loss": 1.013,
      "step": 17310
    },
    {
      "epoch": 3.19903363687047,
      "grad_norm": 0.9253641366958618,
      "learning_rate": 1.8410111006602105e-05,
      "loss": 0.6009,
      "step": 17311
    },
    {
      "epoch": 3.1992194759338415,
      "grad_norm": 1.069978952407837,
      "learning_rate": 1.8401664835273713e-05,
      "loss": 1.046,
      "step": 17312
    },
    {
      "epoch": 3.1994053149972124,
      "grad_norm": 0.9249023795127869,
      "learning_rate": 1.8393220405521794e-05,
      "loss": 0.8473,
      "step": 17313
    },
    {
      "epoch": 3.1995911540605837,
      "grad_norm": 0.9044826030731201,
      "learning_rate": 1.8384777717526537e-05,
      "loss": 0.7975,
      "step": 17314
    },
    {
      "epoch": 3.1997769931239546,
      "grad_norm": 0.8663088083267212,
      "learning_rate": 1.837633677146814e-05,
      "loss": 0.6825,
      "step": 17315
    },
    {
      "epoch": 3.199962832187326,
      "grad_norm": 1.0946561098098755,
      "learning_rate": 1.8367897567526738e-05,
      "loss": 0.994,
      "step": 17316
    },
    {
      "epoch": 3.200148671250697,
      "grad_norm": 1.0189578533172607,
      "learning_rate": 1.8359460105882476e-05,
      "loss": 0.9894,
      "step": 17317
    },
    {
      "epoch": 3.200334510314068,
      "grad_norm": 0.870308518409729,
      "learning_rate": 1.8351024386715464e-05,
      "loss": 0.8479,
      "step": 17318
    },
    {
      "epoch": 3.200520349377439,
      "grad_norm": 1.003595232963562,
      "learning_rate": 1.834259041020572e-05,
      "loss": 0.9592,
      "step": 17319
    },
    {
      "epoch": 3.2007061884408103,
      "grad_norm": 0.8877385258674622,
      "learning_rate": 1.8334158176533233e-05,
      "loss": 0.8161,
      "step": 17320
    },
    {
      "epoch": 3.2008920275041812,
      "grad_norm": 0.989646852016449,
      "learning_rate": 1.8325727685878014e-05,
      "loss": 0.7566,
      "step": 17321
    },
    {
      "epoch": 3.2010778665675526,
      "grad_norm": 0.937359631061554,
      "learning_rate": 1.831729893841998e-05,
      "loss": 0.918,
      "step": 17322
    },
    {
      "epoch": 3.2012637056309234,
      "grad_norm": 0.9940277338027954,
      "learning_rate": 1.8308871934339033e-05,
      "loss": 0.7984,
      "step": 17323
    },
    {
      "epoch": 3.2014495446942948,
      "grad_norm": 0.8262761235237122,
      "learning_rate": 1.8300446673815e-05,
      "loss": 0.6444,
      "step": 17324
    },
    {
      "epoch": 3.2016353837576657,
      "grad_norm": 0.8894662857055664,
      "learning_rate": 1.829202315702775e-05,
      "loss": 0.8029,
      "step": 17325
    },
    {
      "epoch": 3.201821222821037,
      "grad_norm": 1.0533615350723267,
      "learning_rate": 1.8283601384157023e-05,
      "loss": 0.7401,
      "step": 17326
    },
    {
      "epoch": 3.202007061884408,
      "grad_norm": 0.8021373152732849,
      "learning_rate": 1.827518135538263e-05,
      "loss": 0.9263,
      "step": 17327
    },
    {
      "epoch": 3.202192900947779,
      "grad_norm": 0.7998637557029724,
      "learning_rate": 1.826676307088424e-05,
      "loss": 0.555,
      "step": 17328
    },
    {
      "epoch": 3.2023787400111505,
      "grad_norm": 0.956844687461853,
      "learning_rate": 1.8258346530841508e-05,
      "loss": 1.0466,
      "step": 17329
    },
    {
      "epoch": 3.2025645790745214,
      "grad_norm": 1.0209529399871826,
      "learning_rate": 1.824993173543411e-05,
      "loss": 0.8735,
      "step": 17330
    },
    {
      "epoch": 3.2027504181378927,
      "grad_norm": 1.0398613214492798,
      "learning_rate": 1.824151868484164e-05,
      "loss": 0.7566,
      "step": 17331
    },
    {
      "epoch": 3.2029362572012636,
      "grad_norm": 0.9832069277763367,
      "learning_rate": 1.823310737924363e-05,
      "loss": 0.7192,
      "step": 17332
    },
    {
      "epoch": 3.203122096264635,
      "grad_norm": 0.9637802243232727,
      "learning_rate": 1.8224697818819646e-05,
      "loss": 0.6565,
      "step": 17333
    },
    {
      "epoch": 3.203307935328006,
      "grad_norm": 0.9825659990310669,
      "learning_rate": 1.8216290003749147e-05,
      "loss": 0.7692,
      "step": 17334
    },
    {
      "epoch": 3.203493774391377,
      "grad_norm": 0.9774386882781982,
      "learning_rate": 1.82078839342116e-05,
      "loss": 0.9013,
      "step": 17335
    },
    {
      "epoch": 3.203679613454748,
      "grad_norm": 0.9933068752288818,
      "learning_rate": 1.8199479610386382e-05,
      "loss": 0.8352,
      "step": 17336
    },
    {
      "epoch": 3.2038654525181194,
      "grad_norm": 0.9934918880462646,
      "learning_rate": 1.8191077032452898e-05,
      "loss": 0.6283,
      "step": 17337
    },
    {
      "epoch": 3.2040512915814903,
      "grad_norm": 0.9922161102294922,
      "learning_rate": 1.818267620059051e-05,
      "loss": 0.859,
      "step": 17338
    },
    {
      "epoch": 3.2042371306448616,
      "grad_norm": 0.8654912114143372,
      "learning_rate": 1.81742771149785e-05,
      "loss": 0.8402,
      "step": 17339
    },
    {
      "epoch": 3.2044229697082325,
      "grad_norm": 0.9573314189910889,
      "learning_rate": 1.816587977579609e-05,
      "loss": 1.0281,
      "step": 17340
    },
    {
      "epoch": 3.204608808771604,
      "grad_norm": 1.1613332033157349,
      "learning_rate": 1.8157484183222574e-05,
      "loss": 0.9246,
      "step": 17341
    },
    {
      "epoch": 3.2047946478349747,
      "grad_norm": 0.8110087513923645,
      "learning_rate": 1.8149090337437114e-05,
      "loss": 0.778,
      "step": 17342
    },
    {
      "epoch": 3.204980486898346,
      "grad_norm": 1.1747585535049438,
      "learning_rate": 1.8140698238618846e-05,
      "loss": 0.89,
      "step": 17343
    },
    {
      "epoch": 3.2051663259617174,
      "grad_norm": 0.8087934851646423,
      "learning_rate": 1.8132307886946886e-05,
      "loss": 0.7475,
      "step": 17344
    },
    {
      "epoch": 3.2053521650250882,
      "grad_norm": 0.8710451722145081,
      "learning_rate": 1.8123919282600342e-05,
      "loss": 0.7769,
      "step": 17345
    },
    {
      "epoch": 3.2055380040884596,
      "grad_norm": 0.9005964994430542,
      "learning_rate": 1.811553242575824e-05,
      "loss": 0.7139,
      "step": 17346
    },
    {
      "epoch": 3.2057238431518305,
      "grad_norm": 0.7418885827064514,
      "learning_rate": 1.810714731659955e-05,
      "loss": 0.8709,
      "step": 17347
    },
    {
      "epoch": 3.205909682215202,
      "grad_norm": 1.1336352825164795,
      "learning_rate": 1.80987639553033e-05,
      "loss": 1.1367,
      "step": 17348
    },
    {
      "epoch": 3.2060955212785727,
      "grad_norm": 0.9179961085319519,
      "learning_rate": 1.809038234204835e-05,
      "loss": 0.9771,
      "step": 17349
    },
    {
      "epoch": 3.206281360341944,
      "grad_norm": 0.9118735194206238,
      "learning_rate": 1.808200247701366e-05,
      "loss": 0.7601,
      "step": 17350
    },
    {
      "epoch": 3.206467199405315,
      "grad_norm": 1.0031778812408447,
      "learning_rate": 1.807362436037804e-05,
      "loss": 1.1395,
      "step": 17351
    },
    {
      "epoch": 3.206653038468686,
      "grad_norm": 0.9841191172599792,
      "learning_rate": 1.80652479923203e-05,
      "loss": 0.8276,
      "step": 17352
    },
    {
      "epoch": 3.206838877532057,
      "grad_norm": 0.9167536497116089,
      "learning_rate": 1.8056873373019255e-05,
      "loss": 0.9635,
      "step": 17353
    },
    {
      "epoch": 3.2070247165954284,
      "grad_norm": 1.0898497104644775,
      "learning_rate": 1.8048500502653632e-05,
      "loss": 0.706,
      "step": 17354
    },
    {
      "epoch": 3.2072105556587993,
      "grad_norm": 1.0086370706558228,
      "learning_rate": 1.8040129381402137e-05,
      "loss": 0.7848,
      "step": 17355
    },
    {
      "epoch": 3.2073963947221706,
      "grad_norm": 0.8157899975776672,
      "learning_rate": 1.80317600094434e-05,
      "loss": 0.7784,
      "step": 17356
    },
    {
      "epoch": 3.2075822337855415,
      "grad_norm": 0.9179068803787231,
      "learning_rate": 1.8023392386956105e-05,
      "loss": 1.042,
      "step": 17357
    },
    {
      "epoch": 3.207768072848913,
      "grad_norm": 0.9932705163955688,
      "learning_rate": 1.8015026514118794e-05,
      "loss": 0.6414,
      "step": 17358
    },
    {
      "epoch": 3.207953911912284,
      "grad_norm": 1.0313844680786133,
      "learning_rate": 1.8006662391110085e-05,
      "loss": 0.9211,
      "step": 17359
    },
    {
      "epoch": 3.208139750975655,
      "grad_norm": 0.9287518858909607,
      "learning_rate": 1.7998300018108426e-05,
      "loss": 0.8664,
      "step": 17360
    },
    {
      "epoch": 3.2083255900390264,
      "grad_norm": 1.0484979152679443,
      "learning_rate": 1.7989939395292365e-05,
      "loss": 0.7927,
      "step": 17361
    },
    {
      "epoch": 3.2085114291023973,
      "grad_norm": 0.7547088265419006,
      "learning_rate": 1.7981580522840312e-05,
      "loss": 0.7856,
      "step": 17362
    },
    {
      "epoch": 3.2086972681657686,
      "grad_norm": 0.8366743922233582,
      "learning_rate": 1.797322340093067e-05,
      "loss": 0.8571,
      "step": 17363
    },
    {
      "epoch": 3.2088831072291395,
      "grad_norm": 0.9730438590049744,
      "learning_rate": 1.79648680297418e-05,
      "loss": 0.907,
      "step": 17364
    },
    {
      "epoch": 3.209068946292511,
      "grad_norm": 1.0750919580459595,
      "learning_rate": 1.7956514409452052e-05,
      "loss": 0.9299,
      "step": 17365
    },
    {
      "epoch": 3.2092547853558817,
      "grad_norm": 0.958261251449585,
      "learning_rate": 1.7948162540239723e-05,
      "loss": 0.759,
      "step": 17366
    },
    {
      "epoch": 3.209440624419253,
      "grad_norm": 0.8561491370201111,
      "learning_rate": 1.7939812422283055e-05,
      "loss": 0.5556,
      "step": 17367
    },
    {
      "epoch": 3.209626463482624,
      "grad_norm": 1.0215051174163818,
      "learning_rate": 1.7931464055760207e-05,
      "loss": 0.9857,
      "step": 17368
    },
    {
      "epoch": 3.2098123025459953,
      "grad_norm": 1.0156500339508057,
      "learning_rate": 1.792311744084949e-05,
      "loss": 0.7026,
      "step": 17369
    },
    {
      "epoch": 3.209998141609366,
      "grad_norm": 0.9816356301307678,
      "learning_rate": 1.7914772577728966e-05,
      "loss": 0.9134,
      "step": 17370
    },
    {
      "epoch": 3.2101839806727375,
      "grad_norm": 1.0151969194412231,
      "learning_rate": 1.7906429466576767e-05,
      "loss": 0.8126,
      "step": 17371
    },
    {
      "epoch": 3.2103698197361084,
      "grad_norm": 0.9710429906845093,
      "learning_rate": 1.7898088107570932e-05,
      "loss": 0.8628,
      "step": 17372
    },
    {
      "epoch": 3.2105556587994797,
      "grad_norm": 1.1953200101852417,
      "learning_rate": 1.7889748500889536e-05,
      "loss": 0.7935,
      "step": 17373
    },
    {
      "epoch": 3.2107414978628506,
      "grad_norm": 0.781575620174408,
      "learning_rate": 1.788141064671055e-05,
      "loss": 0.9018,
      "step": 17374
    },
    {
      "epoch": 3.210927336926222,
      "grad_norm": 0.9112656712532043,
      "learning_rate": 1.7873074545211932e-05,
      "loss": 0.7,
      "step": 17375
    },
    {
      "epoch": 3.211113175989593,
      "grad_norm": 1.0575007200241089,
      "learning_rate": 1.786474019657157e-05,
      "loss": 0.9902,
      "step": 17376
    },
    {
      "epoch": 3.211299015052964,
      "grad_norm": 0.6355597376823425,
      "learning_rate": 1.7856407600967416e-05,
      "loss": 0.4051,
      "step": 17377
    },
    {
      "epoch": 3.2114848541163354,
      "grad_norm": 0.8127633929252625,
      "learning_rate": 1.784807675857727e-05,
      "loss": 0.7989,
      "step": 17378
    },
    {
      "epoch": 3.2116706931797063,
      "grad_norm": 0.8742978572845459,
      "learning_rate": 1.783974766957893e-05,
      "loss": 0.7368,
      "step": 17379
    },
    {
      "epoch": 3.2118565322430777,
      "grad_norm": 0.8669548630714417,
      "learning_rate": 1.7831420334150183e-05,
      "loss": 0.782,
      "step": 17380
    },
    {
      "epoch": 3.2120423713064485,
      "grad_norm": 0.8861702680587769,
      "learning_rate": 1.7823094752468782e-05,
      "loss": 0.8799,
      "step": 17381
    },
    {
      "epoch": 3.21222821036982,
      "grad_norm": 0.9047160744667053,
      "learning_rate": 1.78147709247124e-05,
      "loss": 0.9262,
      "step": 17382
    },
    {
      "epoch": 3.2124140494331908,
      "grad_norm": 0.8661258220672607,
      "learning_rate": 1.78064488510587e-05,
      "loss": 0.5885,
      "step": 17383
    },
    {
      "epoch": 3.212599888496562,
      "grad_norm": 0.955339789390564,
      "learning_rate": 1.7798128531685276e-05,
      "loss": 1.0327,
      "step": 17384
    },
    {
      "epoch": 3.212785727559933,
      "grad_norm": 0.9303874373435974,
      "learning_rate": 1.7789809966769753e-05,
      "loss": 0.4796,
      "step": 17385
    },
    {
      "epoch": 3.2129715666233043,
      "grad_norm": 0.7944729328155518,
      "learning_rate": 1.778149315648967e-05,
      "loss": 0.8049,
      "step": 17386
    },
    {
      "epoch": 3.213157405686675,
      "grad_norm": 1.2434570789337158,
      "learning_rate": 1.7773178101022514e-05,
      "loss": 0.9347,
      "step": 17387
    },
    {
      "epoch": 3.2133432447500465,
      "grad_norm": 0.9937906861305237,
      "learning_rate": 1.7764864800545744e-05,
      "loss": 0.9397,
      "step": 17388
    },
    {
      "epoch": 3.2135290838134174,
      "grad_norm": 1.1380114555358887,
      "learning_rate": 1.7756553255236807e-05,
      "loss": 0.7016,
      "step": 17389
    },
    {
      "epoch": 3.2137149228767887,
      "grad_norm": 0.9934018850326538,
      "learning_rate": 1.774824346527314e-05,
      "loss": 0.7752,
      "step": 17390
    },
    {
      "epoch": 3.2139007619401596,
      "grad_norm": 1.2391068935394287,
      "learning_rate": 1.7739935430832055e-05,
      "loss": 0.8902,
      "step": 17391
    },
    {
      "epoch": 3.214086601003531,
      "grad_norm": 0.8846229314804077,
      "learning_rate": 1.773162915209087e-05,
      "loss": 0.8731,
      "step": 17392
    },
    {
      "epoch": 3.2142724400669023,
      "grad_norm": 0.8115828633308411,
      "learning_rate": 1.77233246292269e-05,
      "loss": 0.7271,
      "step": 17393
    },
    {
      "epoch": 3.214458279130273,
      "grad_norm": 0.9347731471061707,
      "learning_rate": 1.771502186241738e-05,
      "loss": 0.7558,
      "step": 17394
    },
    {
      "epoch": 3.2146441181936445,
      "grad_norm": 1.0829119682312012,
      "learning_rate": 1.770672085183951e-05,
      "loss": 0.7962,
      "step": 17395
    },
    {
      "epoch": 3.2148299572570154,
      "grad_norm": 0.953552782535553,
      "learning_rate": 1.7698421597670433e-05,
      "loss": 1.0499,
      "step": 17396
    },
    {
      "epoch": 3.2150157963203867,
      "grad_norm": 0.8669760227203369,
      "learning_rate": 1.7690124100087335e-05,
      "loss": 0.8136,
      "step": 17397
    },
    {
      "epoch": 3.2152016353837576,
      "grad_norm": 0.8127499222755432,
      "learning_rate": 1.7681828359267294e-05,
      "loss": 0.9346,
      "step": 17398
    },
    {
      "epoch": 3.215387474447129,
      "grad_norm": 0.8423436880111694,
      "learning_rate": 1.7673534375387356e-05,
      "loss": 0.7928,
      "step": 17399
    },
    {
      "epoch": 3.2155733135105,
      "grad_norm": 0.8691553473472595,
      "learning_rate": 1.7665242148624517e-05,
      "loss": 0.8103,
      "step": 17400
    },
    {
      "epoch": 3.215759152573871,
      "grad_norm": 1.0253678560256958,
      "learning_rate": 1.765695167915582e-05,
      "loss": 0.8424,
      "step": 17401
    },
    {
      "epoch": 3.215944991637242,
      "grad_norm": 0.8192505240440369,
      "learning_rate": 1.7648662967158203e-05,
      "loss": 0.7175,
      "step": 17402
    },
    {
      "epoch": 3.2161308307006133,
      "grad_norm": 0.875633716583252,
      "learning_rate": 1.7640376012808536e-05,
      "loss": 0.953,
      "step": 17403
    },
    {
      "epoch": 3.2163166697639842,
      "grad_norm": 0.8881807923316956,
      "learning_rate": 1.7632090816283696e-05,
      "loss": 0.7815,
      "step": 17404
    },
    {
      "epoch": 3.2165025088273556,
      "grad_norm": 0.9027812480926514,
      "learning_rate": 1.7623807377760548e-05,
      "loss": 0.8665,
      "step": 17405
    },
    {
      "epoch": 3.2166883478907264,
      "grad_norm": 0.943454921245575,
      "learning_rate": 1.761552569741587e-05,
      "loss": 0.823,
      "step": 17406
    },
    {
      "epoch": 3.2168741869540978,
      "grad_norm": 1.0618342161178589,
      "learning_rate": 1.7607245775426418e-05,
      "loss": 0.7578,
      "step": 17407
    },
    {
      "epoch": 3.217060026017469,
      "grad_norm": 0.7595953941345215,
      "learning_rate": 1.7598967611968885e-05,
      "loss": 0.8006,
      "step": 17408
    },
    {
      "epoch": 3.21724586508084,
      "grad_norm": 0.9064504504203796,
      "learning_rate": 1.759069120722001e-05,
      "loss": 0.7277,
      "step": 17409
    },
    {
      "epoch": 3.2174317041442113,
      "grad_norm": 0.959622859954834,
      "learning_rate": 1.7582416561356395e-05,
      "loss": 0.8413,
      "step": 17410
    },
    {
      "epoch": 3.217617543207582,
      "grad_norm": 1.2936711311340332,
      "learning_rate": 1.7574143674554677e-05,
      "loss": 0.6811,
      "step": 17411
    },
    {
      "epoch": 3.2178033822709535,
      "grad_norm": 0.8030010461807251,
      "learning_rate": 1.756587254699139e-05,
      "loss": 0.762,
      "step": 17412
    },
    {
      "epoch": 3.2179892213343244,
      "grad_norm": 1.1652907133102417,
      "learning_rate": 1.755760317884313e-05,
      "loss": 0.8683,
      "step": 17413
    },
    {
      "epoch": 3.2181750603976957,
      "grad_norm": 0.7912318110466003,
      "learning_rate": 1.7549335570286342e-05,
      "loss": 0.6673,
      "step": 17414
    },
    {
      "epoch": 3.2183608994610666,
      "grad_norm": 0.8837103247642517,
      "learning_rate": 1.7541069721497493e-05,
      "loss": 0.9807,
      "step": 17415
    },
    {
      "epoch": 3.218546738524438,
      "grad_norm": 0.8473705053329468,
      "learning_rate": 1.7532805632652984e-05,
      "loss": 0.7719,
      "step": 17416
    },
    {
      "epoch": 3.218732577587809,
      "grad_norm": 0.7680140733718872,
      "learning_rate": 1.752454330392924e-05,
      "loss": 0.7889,
      "step": 17417
    },
    {
      "epoch": 3.21891841665118,
      "grad_norm": 0.7964486479759216,
      "learning_rate": 1.751628273550259e-05,
      "loss": 0.6018,
      "step": 17418
    },
    {
      "epoch": 3.219104255714551,
      "grad_norm": 0.9775687456130981,
      "learning_rate": 1.7508023927549333e-05,
      "loss": 0.8298,
      "step": 17419
    },
    {
      "epoch": 3.2192900947779224,
      "grad_norm": 1.0651310682296753,
      "learning_rate": 1.7499766880245728e-05,
      "loss": 0.6224,
      "step": 17420
    },
    {
      "epoch": 3.2194759338412933,
      "grad_norm": 0.9536681175231934,
      "learning_rate": 1.7491511593768016e-05,
      "loss": 0.7594,
      "step": 17421
    },
    {
      "epoch": 3.2196617729046646,
      "grad_norm": 0.8498244881629944,
      "learning_rate": 1.748325806829242e-05,
      "loss": 0.8059,
      "step": 17422
    },
    {
      "epoch": 3.2198476119680355,
      "grad_norm": 0.96226567029953,
      "learning_rate": 1.7475006303995077e-05,
      "loss": 0.9998,
      "step": 17423
    },
    {
      "epoch": 3.220033451031407,
      "grad_norm": 0.9402596354484558,
      "learning_rate": 1.746675630105209e-05,
      "loss": 0.9033,
      "step": 17424
    },
    {
      "epoch": 3.2202192900947777,
      "grad_norm": 0.9709116220474243,
      "learning_rate": 1.7458508059639577e-05,
      "loss": 0.7782,
      "step": 17425
    },
    {
      "epoch": 3.220405129158149,
      "grad_norm": 1.0609785318374634,
      "learning_rate": 1.7450261579933557e-05,
      "loss": 0.7978,
      "step": 17426
    },
    {
      "epoch": 3.2205909682215204,
      "grad_norm": 0.8852887153625488,
      "learning_rate": 1.7442016862110056e-05,
      "loss": 0.8358,
      "step": 17427
    },
    {
      "epoch": 3.2207768072848912,
      "grad_norm": 0.8573838472366333,
      "learning_rate": 1.7433773906344998e-05,
      "loss": 0.7084,
      "step": 17428
    },
    {
      "epoch": 3.2209626463482626,
      "grad_norm": 0.9779657125473022,
      "learning_rate": 1.742553271281436e-05,
      "loss": 0.7715,
      "step": 17429
    },
    {
      "epoch": 3.2211484854116335,
      "grad_norm": 0.8632031083106995,
      "learning_rate": 1.7417293281694035e-05,
      "loss": 0.7354,
      "step": 17430
    },
    {
      "epoch": 3.221334324475005,
      "grad_norm": 1.1131991147994995,
      "learning_rate": 1.740905561315984e-05,
      "loss": 0.807,
      "step": 17431
    },
    {
      "epoch": 3.2215201635383757,
      "grad_norm": 0.8839072585105896,
      "learning_rate": 1.740081970738763e-05,
      "loss": 0.7896,
      "step": 17432
    },
    {
      "epoch": 3.221706002601747,
      "grad_norm": 0.9753618836402893,
      "learning_rate": 1.7392585564553188e-05,
      "loss": 0.6865,
      "step": 17433
    },
    {
      "epoch": 3.221891841665118,
      "grad_norm": 0.8987646698951721,
      "learning_rate": 1.738435318483226e-05,
      "loss": 0.9049,
      "step": 17434
    },
    {
      "epoch": 3.222077680728489,
      "grad_norm": 0.9076732397079468,
      "learning_rate": 1.7376122568400532e-05,
      "loss": 0.9078,
      "step": 17435
    },
    {
      "epoch": 3.22226351979186,
      "grad_norm": 0.8945940732955933,
      "learning_rate": 1.7367893715433647e-05,
      "loss": 0.8435,
      "step": 17436
    },
    {
      "epoch": 3.2224493588552314,
      "grad_norm": 0.8051013350486755,
      "learning_rate": 1.7359666626107306e-05,
      "loss": 0.6429,
      "step": 17437
    },
    {
      "epoch": 3.2226351979186023,
      "grad_norm": 0.9417064785957336,
      "learning_rate": 1.735144130059706e-05,
      "loss": 0.8148,
      "step": 17438
    },
    {
      "epoch": 3.2228210369819736,
      "grad_norm": 0.9133759140968323,
      "learning_rate": 1.7343217739078466e-05,
      "loss": 1.0047,
      "step": 17439
    },
    {
      "epoch": 3.2230068760453445,
      "grad_norm": 0.8629543781280518,
      "learning_rate": 1.7334995941727018e-05,
      "loss": 0.8391,
      "step": 17440
    },
    {
      "epoch": 3.223192715108716,
      "grad_norm": 0.8090866208076477,
      "learning_rate": 1.732677590871825e-05,
      "loss": 0.8694,
      "step": 17441
    },
    {
      "epoch": 3.223378554172087,
      "grad_norm": 1.018536925315857,
      "learning_rate": 1.7318557640227562e-05,
      "loss": 0.6775,
      "step": 17442
    },
    {
      "epoch": 3.223564393235458,
      "grad_norm": 0.8848704695701599,
      "learning_rate": 1.7310341136430385e-05,
      "loss": 0.8948,
      "step": 17443
    },
    {
      "epoch": 3.2237502322988294,
      "grad_norm": 1.021011471748352,
      "learning_rate": 1.730212639750206e-05,
      "loss": 0.8119,
      "step": 17444
    },
    {
      "epoch": 3.2239360713622003,
      "grad_norm": 0.9703376889228821,
      "learning_rate": 1.7293913423617958e-05,
      "loss": 0.9911,
      "step": 17445
    },
    {
      "epoch": 3.2241219104255716,
      "grad_norm": 0.9496825337409973,
      "learning_rate": 1.7285702214953335e-05,
      "loss": 0.7845,
      "step": 17446
    },
    {
      "epoch": 3.2243077494889425,
      "grad_norm": 1.0321695804595947,
      "learning_rate": 1.7277492771683458e-05,
      "loss": 0.888,
      "step": 17447
    },
    {
      "epoch": 3.224493588552314,
      "grad_norm": 0.8181531429290771,
      "learning_rate": 1.7269285093983523e-05,
      "loss": 0.6862,
      "step": 17448
    },
    {
      "epoch": 3.2246794276156847,
      "grad_norm": 0.8088932037353516,
      "learning_rate": 1.7261079182028738e-05,
      "loss": 0.6089,
      "step": 17449
    },
    {
      "epoch": 3.224865266679056,
      "grad_norm": 0.8229671120643616,
      "learning_rate": 1.7252875035994242e-05,
      "loss": 0.8199,
      "step": 17450
    },
    {
      "epoch": 3.225051105742427,
      "grad_norm": 2.0220842361450195,
      "learning_rate": 1.7244672656055106e-05,
      "loss": 1.2741,
      "step": 17451
    },
    {
      "epoch": 3.2252369448057983,
      "grad_norm": 0.7622953057289124,
      "learning_rate": 1.723647204238641e-05,
      "loss": 0.8076,
      "step": 17452
    },
    {
      "epoch": 3.225422783869169,
      "grad_norm": 0.7097654342651367,
      "learning_rate": 1.7228273195163213e-05,
      "loss": 0.4847,
      "step": 17453
    },
    {
      "epoch": 3.2256086229325405,
      "grad_norm": 0.9323017597198486,
      "learning_rate": 1.722007611456049e-05,
      "loss": 0.8992,
      "step": 17454
    },
    {
      "epoch": 3.2257944619959114,
      "grad_norm": 0.8673163056373596,
      "learning_rate": 1.7211880800753187e-05,
      "loss": 0.7807,
      "step": 17455
    },
    {
      "epoch": 3.2259803010592827,
      "grad_norm": 0.8620083332061768,
      "learning_rate": 1.720368725391619e-05,
      "loss": 0.8554,
      "step": 17456
    },
    {
      "epoch": 3.226166140122654,
      "grad_norm": 1.004608154296875,
      "learning_rate": 1.719549547422443e-05,
      "loss": 0.6373,
      "step": 17457
    },
    {
      "epoch": 3.226351979186025,
      "grad_norm": 0.9062421321868896,
      "learning_rate": 1.718730546185272e-05,
      "loss": 0.8553,
      "step": 17458
    },
    {
      "epoch": 3.2265378182493962,
      "grad_norm": 0.8143224716186523,
      "learning_rate": 1.717911721697586e-05,
      "loss": 0.6425,
      "step": 17459
    },
    {
      "epoch": 3.226723657312767,
      "grad_norm": 1.049385905265808,
      "learning_rate": 1.71709307397686e-05,
      "loss": 0.8966,
      "step": 17460
    },
    {
      "epoch": 3.2269094963761384,
      "grad_norm": 0.9719651937484741,
      "learning_rate": 1.7162746030405708e-05,
      "loss": 1.0292,
      "step": 17461
    },
    {
      "epoch": 3.2270953354395093,
      "grad_norm": 0.8390783071517944,
      "learning_rate": 1.7154563089061837e-05,
      "loss": 0.7897,
      "step": 17462
    },
    {
      "epoch": 3.2272811745028807,
      "grad_norm": 1.06975519657135,
      "learning_rate": 1.7146381915911624e-05,
      "loss": 0.9167,
      "step": 17463
    },
    {
      "epoch": 3.2274670135662515,
      "grad_norm": 0.895122766494751,
      "learning_rate": 1.713820251112972e-05,
      "loss": 0.8312,
      "step": 17464
    },
    {
      "epoch": 3.227652852629623,
      "grad_norm": 0.9486971497535706,
      "learning_rate": 1.7130024874890705e-05,
      "loss": 0.9377,
      "step": 17465
    },
    {
      "epoch": 3.2278386916929938,
      "grad_norm": 0.9098740220069885,
      "learning_rate": 1.7121849007369095e-05,
      "loss": 0.9408,
      "step": 17466
    },
    {
      "epoch": 3.228024530756365,
      "grad_norm": 0.9361106157302856,
      "learning_rate": 1.7113674908739396e-05,
      "loss": 0.5457,
      "step": 17467
    },
    {
      "epoch": 3.228210369819736,
      "grad_norm": 0.8573850989341736,
      "learning_rate": 1.7105502579176036e-05,
      "loss": 0.7005,
      "step": 17468
    },
    {
      "epoch": 3.2283962088831073,
      "grad_norm": 1.0105928182601929,
      "learning_rate": 1.7097332018853507e-05,
      "loss": 0.6945,
      "step": 17469
    },
    {
      "epoch": 3.228582047946478,
      "grad_norm": 0.8553135991096497,
      "learning_rate": 1.708916322794615e-05,
      "loss": 0.6846,
      "step": 17470
    },
    {
      "epoch": 3.2287678870098495,
      "grad_norm": 0.8401904702186584,
      "learning_rate": 1.7080996206628307e-05,
      "loss": 0.8056,
      "step": 17471
    },
    {
      "epoch": 3.2289537260732204,
      "grad_norm": 0.9462745785713196,
      "learning_rate": 1.7072830955074326e-05,
      "loss": 0.9097,
      "step": 17472
    },
    {
      "epoch": 3.2291395651365917,
      "grad_norm": 0.9162221550941467,
      "learning_rate": 1.706466747345843e-05,
      "loss": 0.7836,
      "step": 17473
    },
    {
      "epoch": 3.2293254041999626,
      "grad_norm": 0.9128598570823669,
      "learning_rate": 1.705650576195491e-05,
      "loss": 0.8026,
      "step": 17474
    },
    {
      "epoch": 3.229511243263334,
      "grad_norm": 0.8670196533203125,
      "learning_rate": 1.704834582073794e-05,
      "loss": 0.7359,
      "step": 17475
    },
    {
      "epoch": 3.2296970823267053,
      "grad_norm": 0.8910917639732361,
      "learning_rate": 1.7040187649981653e-05,
      "loss": 0.7666,
      "step": 17476
    },
    {
      "epoch": 3.229882921390076,
      "grad_norm": 0.9486824870109558,
      "learning_rate": 1.7032031249860225e-05,
      "loss": 0.7557,
      "step": 17477
    },
    {
      "epoch": 3.2300687604534475,
      "grad_norm": 1.0442936420440674,
      "learning_rate": 1.7023876620547707e-05,
      "loss": 0.67,
      "step": 17478
    },
    {
      "epoch": 3.2302545995168184,
      "grad_norm": 0.8497734665870667,
      "learning_rate": 1.701572376221815e-05,
      "loss": 0.7017,
      "step": 17479
    },
    {
      "epoch": 3.2304404385801897,
      "grad_norm": 0.8752564787864685,
      "learning_rate": 1.7007572675045536e-05,
      "loss": 0.6539,
      "step": 17480
    },
    {
      "epoch": 3.2306262776435606,
      "grad_norm": 1.0192726850509644,
      "learning_rate": 1.6999423359203893e-05,
      "loss": 0.6568,
      "step": 17481
    },
    {
      "epoch": 3.230812116706932,
      "grad_norm": 0.9239850640296936,
      "learning_rate": 1.6991275814867125e-05,
      "loss": 0.7994,
      "step": 17482
    },
    {
      "epoch": 3.230997955770303,
      "grad_norm": 0.8536369800567627,
      "learning_rate": 1.6983130042209094e-05,
      "loss": 0.7733,
      "step": 17483
    },
    {
      "epoch": 3.231183794833674,
      "grad_norm": 1.1056362390518188,
      "learning_rate": 1.6974986041403683e-05,
      "loss": 0.9172,
      "step": 17484
    },
    {
      "epoch": 3.231369633897045,
      "grad_norm": 1.0050475597381592,
      "learning_rate": 1.6966843812624754e-05,
      "loss": 0.5457,
      "step": 17485
    },
    {
      "epoch": 3.2315554729604163,
      "grad_norm": 0.9066400527954102,
      "learning_rate": 1.6958703356046056e-05,
      "loss": 0.9269,
      "step": 17486
    },
    {
      "epoch": 3.2317413120237872,
      "grad_norm": 1.073598861694336,
      "learning_rate": 1.6950564671841328e-05,
      "loss": 0.935,
      "step": 17487
    },
    {
      "epoch": 3.2319271510871586,
      "grad_norm": 0.8215804100036621,
      "learning_rate": 1.6942427760184243e-05,
      "loss": 0.6632,
      "step": 17488
    },
    {
      "epoch": 3.2321129901505294,
      "grad_norm": 0.8021205067634583,
      "learning_rate": 1.693429262124854e-05,
      "loss": 0.6346,
      "step": 17489
    },
    {
      "epoch": 3.2322988292139008,
      "grad_norm": 0.8880865573883057,
      "learning_rate": 1.6926159255207817e-05,
      "loss": 0.7718,
      "step": 17490
    },
    {
      "epoch": 3.232484668277272,
      "grad_norm": 0.8654438257217407,
      "learning_rate": 1.6918027662235637e-05,
      "loss": 0.6455,
      "step": 17491
    },
    {
      "epoch": 3.232670507340643,
      "grad_norm": 0.9712738990783691,
      "learning_rate": 1.6909897842505608e-05,
      "loss": 0.7434,
      "step": 17492
    },
    {
      "epoch": 3.2328563464040143,
      "grad_norm": 1.083753228187561,
      "learning_rate": 1.6901769796191214e-05,
      "loss": 0.9017,
      "step": 17493
    },
    {
      "epoch": 3.233042185467385,
      "grad_norm": 0.8520892262458801,
      "learning_rate": 1.689364352346592e-05,
      "loss": 0.8221,
      "step": 17494
    },
    {
      "epoch": 3.2332280245307565,
      "grad_norm": 0.869002103805542,
      "learning_rate": 1.6885519024503215e-05,
      "loss": 0.6924,
      "step": 17495
    },
    {
      "epoch": 3.2334138635941274,
      "grad_norm": 1.0325137376785278,
      "learning_rate": 1.6877396299476445e-05,
      "loss": 0.8872,
      "step": 17496
    },
    {
      "epoch": 3.2335997026574987,
      "grad_norm": 0.8123409152030945,
      "learning_rate": 1.6869275348559045e-05,
      "loss": 0.5695,
      "step": 17497
    },
    {
      "epoch": 3.2337855417208696,
      "grad_norm": 0.9622693061828613,
      "learning_rate": 1.6861156171924297e-05,
      "loss": 0.8806,
      "step": 17498
    },
    {
      "epoch": 3.2337855417208696,
      "eval_loss": 1.0438145399093628,
      "eval_runtime": 23.1525,
      "eval_samples_per_second": 47.165,
      "eval_steps_per_second": 23.583,
      "step": 17498
    },
    {
      "epoch": 3.233971380784241,
      "grad_norm": 0.9132378101348877,
      "learning_rate": 1.6853038769745467e-05,
      "loss": 0.7751,
      "step": 17499
    },
    {
      "epoch": 3.234157219847612,
      "grad_norm": 0.8262106776237488,
      "learning_rate": 1.6844923142195878e-05,
      "loss": 0.8341,
      "step": 17500
    },
    {
      "epoch": 3.234343058910983,
      "grad_norm": 0.8283272385597229,
      "learning_rate": 1.6836809289448698e-05,
      "loss": 0.9478,
      "step": 17501
    },
    {
      "epoch": 3.234528897974354,
      "grad_norm": 1.0779932737350464,
      "learning_rate": 1.6828697211677103e-05,
      "loss": 0.9412,
      "step": 17502
    },
    {
      "epoch": 3.2347147370377254,
      "grad_norm": 0.8511762619018555,
      "learning_rate": 1.6820586909054215e-05,
      "loss": 0.6303,
      "step": 17503
    },
    {
      "epoch": 3.2349005761010963,
      "grad_norm": 0.8621987104415894,
      "learning_rate": 1.6812478381753194e-05,
      "loss": 0.8246,
      "step": 17504
    },
    {
      "epoch": 3.2350864151644676,
      "grad_norm": 1.0734256505966187,
      "learning_rate": 1.6804371629947034e-05,
      "loss": 0.878,
      "step": 17505
    },
    {
      "epoch": 3.2352722542278385,
      "grad_norm": 0.904697597026825,
      "learning_rate": 1.6796266653808812e-05,
      "loss": 0.8808,
      "step": 17506
    },
    {
      "epoch": 3.23545809329121,
      "grad_norm": 0.940464973449707,
      "learning_rate": 1.6788163453511508e-05,
      "loss": 0.8067,
      "step": 17507
    },
    {
      "epoch": 3.2356439323545807,
      "grad_norm": 1.1474518775939941,
      "learning_rate": 1.678006202922803e-05,
      "loss": 0.8774,
      "step": 17508
    },
    {
      "epoch": 3.235829771417952,
      "grad_norm": 0.840255081653595,
      "learning_rate": 1.6771962381131335e-05,
      "loss": 0.8111,
      "step": 17509
    },
    {
      "epoch": 3.2360156104813234,
      "grad_norm": 1.0624665021896362,
      "learning_rate": 1.6763864509394277e-05,
      "loss": 0.7288,
      "step": 17510
    },
    {
      "epoch": 3.2362014495446942,
      "grad_norm": 0.8624823689460754,
      "learning_rate": 1.6755768414189664e-05,
      "loss": 0.9116,
      "step": 17511
    },
    {
      "epoch": 3.2363872886080656,
      "grad_norm": 1.5082919597625732,
      "learning_rate": 1.6747674095690357e-05,
      "loss": 1.1363,
      "step": 17512
    },
    {
      "epoch": 3.2365731276714365,
      "grad_norm": 0.9853875041007996,
      "learning_rate": 1.6739581554069072e-05,
      "loss": 0.7867,
      "step": 17513
    },
    {
      "epoch": 3.236758966734808,
      "grad_norm": 0.960754930973053,
      "learning_rate": 1.673149078949855e-05,
      "loss": 0.8707,
      "step": 17514
    },
    {
      "epoch": 3.2369448057981787,
      "grad_norm": 1.0832768678665161,
      "learning_rate": 1.672340180215143e-05,
      "loss": 0.7316,
      "step": 17515
    },
    {
      "epoch": 3.23713064486155,
      "grad_norm": 0.9456191658973694,
      "learning_rate": 1.6715314592200382e-05,
      "loss": 0.8601,
      "step": 17516
    },
    {
      "epoch": 3.237316483924921,
      "grad_norm": 0.8804882168769836,
      "learning_rate": 1.6707229159818062e-05,
      "loss": 0.6816,
      "step": 17517
    },
    {
      "epoch": 3.237502322988292,
      "grad_norm": 0.7851766347885132,
      "learning_rate": 1.6699145505176983e-05,
      "loss": 0.7597,
      "step": 17518
    },
    {
      "epoch": 3.237688162051663,
      "grad_norm": 0.9217889308929443,
      "learning_rate": 1.6691063628449677e-05,
      "loss": 1.0376,
      "step": 17519
    },
    {
      "epoch": 3.2378740011150344,
      "grad_norm": 1.2963614463806152,
      "learning_rate": 1.6682983529808683e-05,
      "loss": 0.8681,
      "step": 17520
    },
    {
      "epoch": 3.2380598401784053,
      "grad_norm": 0.9279267191886902,
      "learning_rate": 1.667490520942643e-05,
      "loss": 0.8838,
      "step": 17521
    },
    {
      "epoch": 3.2382456792417766,
      "grad_norm": 0.8906137347221375,
      "learning_rate": 1.666682866747532e-05,
      "loss": 0.8667,
      "step": 17522
    },
    {
      "epoch": 3.2384315183051475,
      "grad_norm": 0.7467948198318481,
      "learning_rate": 1.6658753904127734e-05,
      "loss": 0.6173,
      "step": 17523
    },
    {
      "epoch": 3.238617357368519,
      "grad_norm": 0.9630430936813354,
      "learning_rate": 1.665068091955604e-05,
      "loss": 0.82,
      "step": 17524
    },
    {
      "epoch": 3.23880319643189,
      "grad_norm": 1.0588127374649048,
      "learning_rate": 1.6642609713932535e-05,
      "loss": 0.9261,
      "step": 17525
    },
    {
      "epoch": 3.238989035495261,
      "grad_norm": 1.1427159309387207,
      "learning_rate": 1.6634540287429457e-05,
      "loss": 0.7037,
      "step": 17526
    },
    {
      "epoch": 3.2391748745586324,
      "grad_norm": 1.0751985311508179,
      "learning_rate": 1.6626472640219083e-05,
      "loss": 0.9218,
      "step": 17527
    },
    {
      "epoch": 3.2393607136220033,
      "grad_norm": 0.9019919633865356,
      "learning_rate": 1.661840677247354e-05,
      "loss": 0.8532,
      "step": 17528
    },
    {
      "epoch": 3.2395465526853746,
      "grad_norm": 1.1597665548324585,
      "learning_rate": 1.6610342684365044e-05,
      "loss": 0.8841,
      "step": 17529
    },
    {
      "epoch": 3.2397323917487455,
      "grad_norm": 0.9410503506660461,
      "learning_rate": 1.6602280376065682e-05,
      "loss": 0.84,
      "step": 17530
    },
    {
      "epoch": 3.239918230812117,
      "grad_norm": 0.9693671464920044,
      "learning_rate": 1.6594219847747504e-05,
      "loss": 0.8363,
      "step": 17531
    },
    {
      "epoch": 3.2401040698754877,
      "grad_norm": 0.992178201675415,
      "learning_rate": 1.6586161099582608e-05,
      "loss": 0.9105,
      "step": 17532
    },
    {
      "epoch": 3.240289908938859,
      "grad_norm": 1.1470305919647217,
      "learning_rate": 1.6578104131742945e-05,
      "loss": 0.8765,
      "step": 17533
    },
    {
      "epoch": 3.24047574800223,
      "grad_norm": 0.852066695690155,
      "learning_rate": 1.6570048944400496e-05,
      "loss": 0.5751,
      "step": 17534
    },
    {
      "epoch": 3.2406615870656013,
      "grad_norm": 1.0298662185668945,
      "learning_rate": 1.656199553772716e-05,
      "loss": 0.7357,
      "step": 17535
    },
    {
      "epoch": 3.240847426128972,
      "grad_norm": 0.9861373901367188,
      "learning_rate": 1.6553943911894843e-05,
      "loss": 0.9414,
      "step": 17536
    },
    {
      "epoch": 3.2410332651923435,
      "grad_norm": 0.864341676235199,
      "learning_rate": 1.6545894067075418e-05,
      "loss": 0.8659,
      "step": 17537
    },
    {
      "epoch": 3.2412191042557144,
      "grad_norm": 0.9241873621940613,
      "learning_rate": 1.653784600344067e-05,
      "loss": 0.794,
      "step": 17538
    },
    {
      "epoch": 3.2414049433190857,
      "grad_norm": 0.9814783930778503,
      "learning_rate": 1.6529799721162343e-05,
      "loss": 0.849,
      "step": 17539
    },
    {
      "epoch": 3.241590782382457,
      "grad_norm": 0.9415474534034729,
      "learning_rate": 1.652175522041224e-05,
      "loss": 0.9087,
      "step": 17540
    },
    {
      "epoch": 3.241776621445828,
      "grad_norm": 0.9838879704475403,
      "learning_rate": 1.6513712501362e-05,
      "loss": 0.8529,
      "step": 17541
    },
    {
      "epoch": 3.2419624605091992,
      "grad_norm": 0.9883627891540527,
      "learning_rate": 1.650567156418331e-05,
      "loss": 1.0156,
      "step": 17542
    },
    {
      "epoch": 3.24214829957257,
      "grad_norm": 0.9032547473907471,
      "learning_rate": 1.649763240904775e-05,
      "loss": 0.8906,
      "step": 17543
    },
    {
      "epoch": 3.2423341386359414,
      "grad_norm": 0.9878135919570923,
      "learning_rate": 1.6489595036126946e-05,
      "loss": 0.7668,
      "step": 17544
    },
    {
      "epoch": 3.2425199776993123,
      "grad_norm": 0.9744989275932312,
      "learning_rate": 1.6481559445592442e-05,
      "loss": 0.9409,
      "step": 17545
    },
    {
      "epoch": 3.2427058167626837,
      "grad_norm": 1.0206999778747559,
      "learning_rate": 1.6473525637615716e-05,
      "loss": 0.8726,
      "step": 17546
    },
    {
      "epoch": 3.2428916558260545,
      "grad_norm": 1.1411030292510986,
      "learning_rate": 1.6465493612368233e-05,
      "loss": 0.9333,
      "step": 17547
    },
    {
      "epoch": 3.243077494889426,
      "grad_norm": 1.3796285390853882,
      "learning_rate": 1.6457463370021432e-05,
      "loss": 0.8776,
      "step": 17548
    },
    {
      "epoch": 3.2432633339527968,
      "grad_norm": 0.7899224162101746,
      "learning_rate": 1.644943491074674e-05,
      "loss": 0.7336,
      "step": 17549
    },
    {
      "epoch": 3.243449173016168,
      "grad_norm": 0.810294508934021,
      "learning_rate": 1.6441408234715483e-05,
      "loss": 0.7711,
      "step": 17550
    },
    {
      "epoch": 3.243635012079539,
      "grad_norm": 0.9372333288192749,
      "learning_rate": 1.643338334209894e-05,
      "loss": 0.9952,
      "step": 17551
    },
    {
      "epoch": 3.2438208511429103,
      "grad_norm": 0.8857443332672119,
      "learning_rate": 1.6425360233068455e-05,
      "loss": 1.014,
      "step": 17552
    },
    {
      "epoch": 3.244006690206281,
      "grad_norm": 1.0264416933059692,
      "learning_rate": 1.6417338907795232e-05,
      "loss": 0.8121,
      "step": 17553
    },
    {
      "epoch": 3.2441925292696525,
      "grad_norm": 0.9513406157493591,
      "learning_rate": 1.6409319366450483e-05,
      "loss": 0.761,
      "step": 17554
    },
    {
      "epoch": 3.2443783683330234,
      "grad_norm": 1.558884620666504,
      "learning_rate": 1.6401301609205335e-05,
      "loss": 1.2663,
      "step": 17555
    },
    {
      "epoch": 3.2445642073963947,
      "grad_norm": 0.9729040265083313,
      "learning_rate": 1.639328563623097e-05,
      "loss": 0.6958,
      "step": 17556
    },
    {
      "epoch": 3.2447500464597656,
      "grad_norm": 0.9089725017547607,
      "learning_rate": 1.6385271447698425e-05,
      "loss": 0.7511,
      "step": 17557
    },
    {
      "epoch": 3.244935885523137,
      "grad_norm": 0.9211695790290833,
      "learning_rate": 1.637725904377879e-05,
      "loss": 0.8387,
      "step": 17558
    },
    {
      "epoch": 3.2451217245865083,
      "grad_norm": 0.8523728847503662,
      "learning_rate": 1.6369248424643035e-05,
      "loss": 0.7116,
      "step": 17559
    },
    {
      "epoch": 3.245307563649879,
      "grad_norm": 1.1124895811080933,
      "learning_rate": 1.6361239590462184e-05,
      "loss": 0.9035,
      "step": 17560
    },
    {
      "epoch": 3.2454934027132505,
      "grad_norm": 1.123880386352539,
      "learning_rate": 1.6353232541407148e-05,
      "loss": 0.8926,
      "step": 17561
    },
    {
      "epoch": 3.2456792417766214,
      "grad_norm": 0.828389048576355,
      "learning_rate": 1.6345227277648812e-05,
      "loss": 0.6941,
      "step": 17562
    },
    {
      "epoch": 3.2458650808399927,
      "grad_norm": 0.9504824280738831,
      "learning_rate": 1.6337223799358026e-05,
      "loss": 0.7694,
      "step": 17563
    },
    {
      "epoch": 3.2460509199033636,
      "grad_norm": 0.8812553882598877,
      "learning_rate": 1.6329222106705643e-05,
      "loss": 0.6938,
      "step": 17564
    },
    {
      "epoch": 3.246236758966735,
      "grad_norm": 1.429211974143982,
      "learning_rate": 1.632122219986244e-05,
      "loss": 0.8115,
      "step": 17565
    },
    {
      "epoch": 3.246422598030106,
      "grad_norm": 1.1453877687454224,
      "learning_rate": 1.6313224078999145e-05,
      "loss": 0.8998,
      "step": 17566
    },
    {
      "epoch": 3.246608437093477,
      "grad_norm": 0.8724762797355652,
      "learning_rate": 1.6305227744286443e-05,
      "loss": 0.8391,
      "step": 17567
    },
    {
      "epoch": 3.246794276156848,
      "grad_norm": 0.8664789199829102,
      "learning_rate": 1.629723319589502e-05,
      "loss": 0.8601,
      "step": 17568
    },
    {
      "epoch": 3.2469801152202193,
      "grad_norm": 0.8182777762413025,
      "learning_rate": 1.6289240433995555e-05,
      "loss": 0.7966,
      "step": 17569
    },
    {
      "epoch": 3.2471659542835902,
      "grad_norm": 0.8636091947555542,
      "learning_rate": 1.6281249458758597e-05,
      "loss": 0.8471,
      "step": 17570
    },
    {
      "epoch": 3.2473517933469616,
      "grad_norm": 0.9435926675796509,
      "learning_rate": 1.6273260270354674e-05,
      "loss": 0.8945,
      "step": 17571
    },
    {
      "epoch": 3.2475376324103324,
      "grad_norm": 0.7587558627128601,
      "learning_rate": 1.626527286895435e-05,
      "loss": 0.6551,
      "step": 17572
    },
    {
      "epoch": 3.2477234714737038,
      "grad_norm": 0.8410842418670654,
      "learning_rate": 1.625728725472809e-05,
      "loss": 0.7667,
      "step": 17573
    },
    {
      "epoch": 3.247909310537075,
      "grad_norm": 0.9617728590965271,
      "learning_rate": 1.624930342784633e-05,
      "loss": 0.6956,
      "step": 17574
    },
    {
      "epoch": 3.248095149600446,
      "grad_norm": 0.8479878902435303,
      "learning_rate": 1.6241321388479424e-05,
      "loss": 0.8241,
      "step": 17575
    },
    {
      "epoch": 3.2482809886638173,
      "grad_norm": 1.0032156705856323,
      "learning_rate": 1.6233341136797818e-05,
      "loss": 0.6864,
      "step": 17576
    },
    {
      "epoch": 3.248466827727188,
      "grad_norm": 0.8712114095687866,
      "learning_rate": 1.6225362672971788e-05,
      "loss": 0.883,
      "step": 17577
    },
    {
      "epoch": 3.2486526667905595,
      "grad_norm": 0.7898356318473816,
      "learning_rate": 1.6217385997171608e-05,
      "loss": 0.7023,
      "step": 17578
    },
    {
      "epoch": 3.2488385058539304,
      "grad_norm": 0.8623623847961426,
      "learning_rate": 1.620941110956754e-05,
      "loss": 0.8952,
      "step": 17579
    },
    {
      "epoch": 3.2490243449173017,
      "grad_norm": 0.857649028301239,
      "learning_rate": 1.620143801032984e-05,
      "loss": 0.7604,
      "step": 17580
    },
    {
      "epoch": 3.2492101839806726,
      "grad_norm": 1.2372016906738281,
      "learning_rate": 1.6193466699628622e-05,
      "loss": 1.1232,
      "step": 17581
    },
    {
      "epoch": 3.249396023044044,
      "grad_norm": 0.9473984241485596,
      "learning_rate": 1.6185497177634056e-05,
      "loss": 0.843,
      "step": 17582
    },
    {
      "epoch": 3.249581862107415,
      "grad_norm": 0.815261960029602,
      "learning_rate": 1.6177529444516194e-05,
      "loss": 0.7389,
      "step": 17583
    },
    {
      "epoch": 3.249767701170786,
      "grad_norm": 0.9083401560783386,
      "learning_rate": 1.6169563500445138e-05,
      "loss": 0.7564,
      "step": 17584
    },
    {
      "epoch": 3.249953540234157,
      "grad_norm": 1.1427556276321411,
      "learning_rate": 1.6161599345590894e-05,
      "loss": 0.8388,
      "step": 17585
    },
    {
      "epoch": 3.2501393792975284,
      "grad_norm": 0.8198236227035522,
      "learning_rate": 1.6153636980123442e-05,
      "loss": 0.8777,
      "step": 17586
    },
    {
      "epoch": 3.2503252183608993,
      "grad_norm": 0.9134224653244019,
      "learning_rate": 1.6145676404212696e-05,
      "loss": 0.846,
      "step": 17587
    },
    {
      "epoch": 3.2505110574242706,
      "grad_norm": 0.9644818305969238,
      "learning_rate": 1.6137717618028614e-05,
      "loss": 0.7818,
      "step": 17588
    },
    {
      "epoch": 3.250696896487642,
      "grad_norm": 0.9705414175987244,
      "learning_rate": 1.6129760621741008e-05,
      "loss": 0.8388,
      "step": 17589
    },
    {
      "epoch": 3.250882735551013,
      "grad_norm": 1.0678094625473022,
      "learning_rate": 1.6121805415519763e-05,
      "loss": 0.8321,
      "step": 17590
    },
    {
      "epoch": 3.2510685746143837,
      "grad_norm": 0.8345924615859985,
      "learning_rate": 1.6113851999534613e-05,
      "loss": 0.7277,
      "step": 17591
    },
    {
      "epoch": 3.251254413677755,
      "grad_norm": 1.1566282510757446,
      "learning_rate": 1.6105900373955364e-05,
      "loss": 0.8567,
      "step": 17592
    },
    {
      "epoch": 3.2514402527411264,
      "grad_norm": 0.8294490575790405,
      "learning_rate": 1.60979505389517e-05,
      "loss": 0.8072,
      "step": 17593
    },
    {
      "epoch": 3.2516260918044972,
      "grad_norm": 1.0158021450042725,
      "learning_rate": 1.60900024946933e-05,
      "loss": 0.9375,
      "step": 17594
    },
    {
      "epoch": 3.2518119308678686,
      "grad_norm": 0.9069027900695801,
      "learning_rate": 1.6082056241349786e-05,
      "loss": 0.9329,
      "step": 17595
    },
    {
      "epoch": 3.2519977699312395,
      "grad_norm": 0.9600441455841064,
      "learning_rate": 1.6074111779090784e-05,
      "loss": 0.8963,
      "step": 17596
    },
    {
      "epoch": 3.252183608994611,
      "grad_norm": 1.4393479824066162,
      "learning_rate": 1.606616910808586e-05,
      "loss": 0.9299,
      "step": 17597
    },
    {
      "epoch": 3.2523694480579817,
      "grad_norm": 0.8488100171089172,
      "learning_rate": 1.6058228228504503e-05,
      "loss": 0.7866,
      "step": 17598
    },
    {
      "epoch": 3.252555287121353,
      "grad_norm": 0.9428489804267883,
      "learning_rate": 1.6050289140516174e-05,
      "loss": 0.8406,
      "step": 17599
    },
    {
      "epoch": 3.252741126184724,
      "grad_norm": 0.8999878168106079,
      "learning_rate": 1.6042351844290404e-05,
      "loss": 0.7001,
      "step": 17600
    },
    {
      "epoch": 3.252926965248095,
      "grad_norm": 0.9167279005050659,
      "learning_rate": 1.6034416339996562e-05,
      "loss": 0.754,
      "step": 17601
    },
    {
      "epoch": 3.253112804311466,
      "grad_norm": 0.9463081359863281,
      "learning_rate": 1.602648262780402e-05,
      "loss": 0.7938,
      "step": 17602
    },
    {
      "epoch": 3.2532986433748374,
      "grad_norm": 0.8923173546791077,
      "learning_rate": 1.6018550707882062e-05,
      "loss": 1.0319,
      "step": 17603
    },
    {
      "epoch": 3.2534844824382088,
      "grad_norm": 0.8322475552558899,
      "learning_rate": 1.6010620580400047e-05,
      "loss": 0.7353,
      "step": 17604
    },
    {
      "epoch": 3.2536703215015796,
      "grad_norm": 0.8827752470970154,
      "learning_rate": 1.6002692245527218e-05,
      "loss": 0.7999,
      "step": 17605
    },
    {
      "epoch": 3.2538561605649505,
      "grad_norm": 0.9471423029899597,
      "learning_rate": 1.5994765703432767e-05,
      "loss": 0.7182,
      "step": 17606
    },
    {
      "epoch": 3.254041999628322,
      "grad_norm": 0.9410121440887451,
      "learning_rate": 1.5986840954285865e-05,
      "loss": 0.939,
      "step": 17607
    },
    {
      "epoch": 3.254227838691693,
      "grad_norm": 0.9756559729576111,
      "learning_rate": 1.597891799825568e-05,
      "loss": 0.8744,
      "step": 17608
    },
    {
      "epoch": 3.254413677755064,
      "grad_norm": 1.0734498500823975,
      "learning_rate": 1.597099683551132e-05,
      "loss": 0.8687,
      "step": 17609
    },
    {
      "epoch": 3.2545995168184354,
      "grad_norm": 1.6099714040756226,
      "learning_rate": 1.5963077466221808e-05,
      "loss": 0.9598,
      "step": 17610
    },
    {
      "epoch": 3.2547853558818063,
      "grad_norm": 0.8606815338134766,
      "learning_rate": 1.595515989055618e-05,
      "loss": 0.8517,
      "step": 17611
    },
    {
      "epoch": 3.2549711949451776,
      "grad_norm": 0.8491156697273254,
      "learning_rate": 1.5947244108683478e-05,
      "loss": 0.7487,
      "step": 17612
    },
    {
      "epoch": 3.2551570340085485,
      "grad_norm": 0.8932136297225952,
      "learning_rate": 1.59393301207726e-05,
      "loss": 0.8061,
      "step": 17613
    },
    {
      "epoch": 3.25534287307192,
      "grad_norm": 0.9115806221961975,
      "learning_rate": 1.5931417926992476e-05,
      "loss": 0.7714,
      "step": 17614
    },
    {
      "epoch": 3.2555287121352907,
      "grad_norm": 0.9725075960159302,
      "learning_rate": 1.592350752751194e-05,
      "loss": 0.9278,
      "step": 17615
    },
    {
      "epoch": 3.255714551198662,
      "grad_norm": 0.9991533160209656,
      "learning_rate": 1.5915598922499875e-05,
      "loss": 0.6981,
      "step": 17616
    },
    {
      "epoch": 3.255900390262033,
      "grad_norm": 0.925247311592102,
      "learning_rate": 1.590769211212506e-05,
      "loss": 0.8375,
      "step": 17617
    },
    {
      "epoch": 3.2560862293254043,
      "grad_norm": 1.1035805940628052,
      "learning_rate": 1.589978709655625e-05,
      "loss": 0.8186,
      "step": 17618
    },
    {
      "epoch": 3.256272068388775,
      "grad_norm": 0.928608238697052,
      "learning_rate": 1.5891883875962134e-05,
      "loss": 0.854,
      "step": 17619
    },
    {
      "epoch": 3.2564579074521465,
      "grad_norm": 0.9559894800186157,
      "learning_rate": 1.5883982450511427e-05,
      "loss": 1.0359,
      "step": 17620
    },
    {
      "epoch": 3.2566437465155174,
      "grad_norm": 1.0768336057662964,
      "learning_rate": 1.5876082820372785e-05,
      "loss": 0.623,
      "step": 17621
    },
    {
      "epoch": 3.2568295855788887,
      "grad_norm": 0.9776822924613953,
      "learning_rate": 1.5868184985714797e-05,
      "loss": 0.8205,
      "step": 17622
    },
    {
      "epoch": 3.25701542464226,
      "grad_norm": 0.8623474836349487,
      "learning_rate": 1.5860288946706003e-05,
      "loss": 0.5707,
      "step": 17623
    },
    {
      "epoch": 3.257201263705631,
      "grad_norm": 0.9123068451881409,
      "learning_rate": 1.5852394703514972e-05,
      "loss": 0.8048,
      "step": 17624
    },
    {
      "epoch": 3.2573871027690022,
      "grad_norm": 1.2158424854278564,
      "learning_rate": 1.5844502256310167e-05,
      "loss": 0.8853,
      "step": 17625
    },
    {
      "epoch": 3.257572941832373,
      "grad_norm": 0.7986825108528137,
      "learning_rate": 1.5836611605260054e-05,
      "loss": 0.7819,
      "step": 17626
    },
    {
      "epoch": 3.2577587808957444,
      "grad_norm": 1.1961933374404907,
      "learning_rate": 1.5828722750533008e-05,
      "loss": 0.9998,
      "step": 17627
    },
    {
      "epoch": 3.2579446199591153,
      "grad_norm": 1.1025460958480835,
      "learning_rate": 1.5820835692297455e-05,
      "loss": 0.6594,
      "step": 17628
    },
    {
      "epoch": 3.2581304590224867,
      "grad_norm": 1.039323091506958,
      "learning_rate": 1.5812950430721708e-05,
      "loss": 0.885,
      "step": 17629
    },
    {
      "epoch": 3.2583162980858575,
      "grad_norm": 0.849334716796875,
      "learning_rate": 1.5805066965974058e-05,
      "loss": 0.7472,
      "step": 17630
    },
    {
      "epoch": 3.258502137149229,
      "grad_norm": 0.9839600920677185,
      "learning_rate": 1.5797185298222727e-05,
      "loss": 0.8077,
      "step": 17631
    },
    {
      "epoch": 3.2586879762125998,
      "grad_norm": 0.8936685919761658,
      "learning_rate": 1.5789305427636027e-05,
      "loss": 0.7506,
      "step": 17632
    },
    {
      "epoch": 3.258873815275971,
      "grad_norm": 1.0029278993606567,
      "learning_rate": 1.5781427354382083e-05,
      "loss": 0.7903,
      "step": 17633
    },
    {
      "epoch": 3.259059654339342,
      "grad_norm": 1.0882457494735718,
      "learning_rate": 1.5773551078629056e-05,
      "loss": 0.7335,
      "step": 17634
    },
    {
      "epoch": 3.2592454934027133,
      "grad_norm": 0.8651626706123352,
      "learning_rate": 1.576567660054502e-05,
      "loss": 0.8915,
      "step": 17635
    },
    {
      "epoch": 3.259431332466084,
      "grad_norm": 1.1022568941116333,
      "learning_rate": 1.5757803920298087e-05,
      "loss": 0.8194,
      "step": 17636
    },
    {
      "epoch": 3.2596171715294555,
      "grad_norm": 0.8831328749656677,
      "learning_rate": 1.5749933038056265e-05,
      "loss": 1.0054,
      "step": 17637
    },
    {
      "epoch": 3.259803010592827,
      "grad_norm": 0.910919725894928,
      "learning_rate": 1.5742063953987528e-05,
      "loss": 0.8578,
      "step": 17638
    },
    {
      "epoch": 3.2599888496561977,
      "grad_norm": 0.8606796264648438,
      "learning_rate": 1.573419666825984e-05,
      "loss": 0.6577,
      "step": 17639
    },
    {
      "epoch": 3.2601746887195686,
      "grad_norm": 0.8369166851043701,
      "learning_rate": 1.5726331181041133e-05,
      "loss": 0.7804,
      "step": 17640
    },
    {
      "epoch": 3.26036052778294,
      "grad_norm": 0.8461362719535828,
      "learning_rate": 1.5718467492499245e-05,
      "loss": 0.7175,
      "step": 17641
    },
    {
      "epoch": 3.2605463668463113,
      "grad_norm": 1.1136691570281982,
      "learning_rate": 1.5710605602802065e-05,
      "loss": 0.9438,
      "step": 17642
    },
    {
      "epoch": 3.260732205909682,
      "grad_norm": 0.7843376994132996,
      "learning_rate": 1.5702745512117324e-05,
      "loss": 0.7915,
      "step": 17643
    },
    {
      "epoch": 3.2609180449730535,
      "grad_norm": 0.9372939467430115,
      "learning_rate": 1.5694887220612854e-05,
      "loss": 0.7961,
      "step": 17644
    },
    {
      "epoch": 3.2611038840364244,
      "grad_norm": 0.9335784316062927,
      "learning_rate": 1.5687030728456344e-05,
      "loss": 0.831,
      "step": 17645
    },
    {
      "epoch": 3.2612897230997957,
      "grad_norm": 0.8994518518447876,
      "learning_rate": 1.567917603581547e-05,
      "loss": 0.6857,
      "step": 17646
    },
    {
      "epoch": 3.2614755621631666,
      "grad_norm": 1.0384033918380737,
      "learning_rate": 1.5671323142857863e-05,
      "loss": 0.8729,
      "step": 17647
    },
    {
      "epoch": 3.261661401226538,
      "grad_norm": 0.9693853855133057,
      "learning_rate": 1.566347204975117e-05,
      "loss": 1.0206,
      "step": 17648
    },
    {
      "epoch": 3.261847240289909,
      "grad_norm": 0.943858802318573,
      "learning_rate": 1.5655622756662948e-05,
      "loss": 0.8711,
      "step": 17649
    },
    {
      "epoch": 3.26203307935328,
      "grad_norm": 0.8209655284881592,
      "learning_rate": 1.5647775263760687e-05,
      "loss": 0.8619,
      "step": 17650
    },
    {
      "epoch": 3.262218918416651,
      "grad_norm": 0.8807152509689331,
      "learning_rate": 1.563992957121193e-05,
      "loss": 0.8328,
      "step": 17651
    },
    {
      "epoch": 3.2624047574800223,
      "grad_norm": 0.8838180303573608,
      "learning_rate": 1.5632085679184093e-05,
      "loss": 0.5347,
      "step": 17652
    },
    {
      "epoch": 3.2625905965433932,
      "grad_norm": 0.8919572234153748,
      "learning_rate": 1.5624243587844622e-05,
      "loss": 1.106,
      "step": 17653
    },
    {
      "epoch": 3.2627764356067646,
      "grad_norm": 0.8996062278747559,
      "learning_rate": 1.561640329736088e-05,
      "loss": 0.8418,
      "step": 17654
    },
    {
      "epoch": 3.2629622746701354,
      "grad_norm": 0.9466404914855957,
      "learning_rate": 1.5608564807900184e-05,
      "loss": 1.0323,
      "step": 17655
    },
    {
      "epoch": 3.2631481137335068,
      "grad_norm": 1.0202292203903198,
      "learning_rate": 1.560072811962987e-05,
      "loss": 0.9618,
      "step": 17656
    },
    {
      "epoch": 3.263333952796878,
      "grad_norm": 0.9965801239013672,
      "learning_rate": 1.5592893232717175e-05,
      "loss": 0.7936,
      "step": 17657
    },
    {
      "epoch": 3.263519791860249,
      "grad_norm": 0.7998269200325012,
      "learning_rate": 1.558506014732932e-05,
      "loss": 0.7537,
      "step": 17658
    },
    {
      "epoch": 3.2637056309236203,
      "grad_norm": 0.9901534914970398,
      "learning_rate": 1.5577228863633486e-05,
      "loss": 0.7394,
      "step": 17659
    },
    {
      "epoch": 3.263891469986991,
      "grad_norm": 0.8547446727752686,
      "learning_rate": 1.5569399381796846e-05,
      "loss": 0.6628,
      "step": 17660
    },
    {
      "epoch": 3.2640773090503625,
      "grad_norm": 1.1768836975097656,
      "learning_rate": 1.5561571701986477e-05,
      "loss": 0.7231,
      "step": 17661
    },
    {
      "epoch": 3.2642631481137334,
      "grad_norm": 0.9869349598884583,
      "learning_rate": 1.555374582436945e-05,
      "loss": 0.6661,
      "step": 17662
    },
    {
      "epoch": 3.2644489871771047,
      "grad_norm": 1.0868933200836182,
      "learning_rate": 1.554592174911279e-05,
      "loss": 0.8784,
      "step": 17663
    },
    {
      "epoch": 3.2646348262404756,
      "grad_norm": 0.8081589341163635,
      "learning_rate": 1.5538099476383527e-05,
      "loss": 0.6805,
      "step": 17664
    },
    {
      "epoch": 3.264820665303847,
      "grad_norm": 0.9574329257011414,
      "learning_rate": 1.5530279006348592e-05,
      "loss": 0.8514,
      "step": 17665
    },
    {
      "epoch": 3.265006504367218,
      "grad_norm": 0.883391797542572,
      "learning_rate": 1.5522460339174892e-05,
      "loss": 0.8386,
      "step": 17666
    },
    {
      "epoch": 3.265192343430589,
      "grad_norm": 0.9138352274894714,
      "learning_rate": 1.551464347502929e-05,
      "loss": 0.8067,
      "step": 17667
    },
    {
      "epoch": 3.26537818249396,
      "grad_norm": 0.8999261260032654,
      "learning_rate": 1.5506828414078655e-05,
      "loss": 0.8958,
      "step": 17668
    },
    {
      "epoch": 3.2655640215573314,
      "grad_norm": 1.483202338218689,
      "learning_rate": 1.5499015156489784e-05,
      "loss": 1.3382,
      "step": 17669
    },
    {
      "epoch": 3.2657498606207023,
      "grad_norm": 0.9040524959564209,
      "learning_rate": 1.549120370242939e-05,
      "loss": 0.9632,
      "step": 17670
    },
    {
      "epoch": 3.2659356996840736,
      "grad_norm": 0.7727354168891907,
      "learning_rate": 1.548339405206426e-05,
      "loss": 0.7169,
      "step": 17671
    },
    {
      "epoch": 3.266121538747445,
      "grad_norm": 0.8451464772224426,
      "learning_rate": 1.5475586205561042e-05,
      "loss": 0.7764,
      "step": 17672
    },
    {
      "epoch": 3.266307377810816,
      "grad_norm": 0.8612467646598816,
      "learning_rate": 1.5467780163086365e-05,
      "loss": 0.9192,
      "step": 17673
    },
    {
      "epoch": 3.266493216874187,
      "grad_norm": 0.7170829772949219,
      "learning_rate": 1.5459975924806878e-05,
      "loss": 0.438,
      "step": 17674
    },
    {
      "epoch": 3.266679055937558,
      "grad_norm": 0.9012123942375183,
      "learning_rate": 1.545217349088911e-05,
      "loss": 0.7104,
      "step": 17675
    },
    {
      "epoch": 3.2668648950009294,
      "grad_norm": 0.8574148416519165,
      "learning_rate": 1.5444372861499634e-05,
      "loss": 0.8781,
      "step": 17676
    },
    {
      "epoch": 3.2670507340643002,
      "grad_norm": 1.2092663049697876,
      "learning_rate": 1.543657403680492e-05,
      "loss": 0.9431,
      "step": 17677
    },
    {
      "epoch": 3.2672365731276716,
      "grad_norm": 0.9819631576538086,
      "learning_rate": 1.5428777016971407e-05,
      "loss": 0.7433,
      "step": 17678
    },
    {
      "epoch": 3.2674224121910425,
      "grad_norm": 0.8612439632415771,
      "learning_rate": 1.5420981802165502e-05,
      "loss": 1.0082,
      "step": 17679
    },
    {
      "epoch": 3.267608251254414,
      "grad_norm": 1.0462713241577148,
      "learning_rate": 1.5413188392553612e-05,
      "loss": 0.8537,
      "step": 17680
    },
    {
      "epoch": 3.2677940903177847,
      "grad_norm": 0.8656796813011169,
      "learning_rate": 1.5405396788302064e-05,
      "loss": 0.8139,
      "step": 17681
    },
    {
      "epoch": 3.267979929381156,
      "grad_norm": 1.0041792392730713,
      "learning_rate": 1.539760698957713e-05,
      "loss": 0.6557,
      "step": 17682
    },
    {
      "epoch": 3.268165768444527,
      "grad_norm": 1.0172066688537598,
      "learning_rate": 1.538981899654508e-05,
      "loss": 0.9778,
      "step": 17683
    },
    {
      "epoch": 3.268351607507898,
      "grad_norm": 0.901366651058197,
      "learning_rate": 1.5382032809372182e-05,
      "loss": 0.7569,
      "step": 17684
    },
    {
      "epoch": 3.268537446571269,
      "grad_norm": 0.8727717399597168,
      "learning_rate": 1.537424842822458e-05,
      "loss": 0.7546,
      "step": 17685
    },
    {
      "epoch": 3.2687232856346404,
      "grad_norm": 0.9215835928916931,
      "learning_rate": 1.5366465853268418e-05,
      "loss": 0.763,
      "step": 17686
    },
    {
      "epoch": 3.2689091246980118,
      "grad_norm": 0.9440967440605164,
      "learning_rate": 1.5358685084669788e-05,
      "loss": 0.8188,
      "step": 17687
    },
    {
      "epoch": 3.2690949637613826,
      "grad_norm": 0.9095939993858337,
      "learning_rate": 1.53509061225948e-05,
      "loss": 0.9214,
      "step": 17688
    },
    {
      "epoch": 3.2692808028247535,
      "grad_norm": 1.0717787742614746,
      "learning_rate": 1.5343128967209452e-05,
      "loss": 1.0171,
      "step": 17689
    },
    {
      "epoch": 3.269466641888125,
      "grad_norm": 1.1467723846435547,
      "learning_rate": 1.5335353618679716e-05,
      "loss": 0.8464,
      "step": 17690
    },
    {
      "epoch": 3.269652480951496,
      "grad_norm": 0.6764289140701294,
      "learning_rate": 1.5327580077171587e-05,
      "loss": 0.4945,
      "step": 17691
    },
    {
      "epoch": 3.269838320014867,
      "grad_norm": 1.0119014978408813,
      "learning_rate": 1.5319808342850962e-05,
      "loss": 0.7477,
      "step": 17692
    },
    {
      "epoch": 3.2700241590782384,
      "grad_norm": 0.9415217638015747,
      "learning_rate": 1.5312038415883713e-05,
      "loss": 0.8812,
      "step": 17693
    },
    {
      "epoch": 3.2702099981416093,
      "grad_norm": 1.0114682912826538,
      "learning_rate": 1.5304270296435643e-05,
      "loss": 0.9665,
      "step": 17694
    },
    {
      "epoch": 3.2703958372049806,
      "grad_norm": 0.8940282464027405,
      "learning_rate": 1.5296503984672582e-05,
      "loss": 0.9342,
      "step": 17695
    },
    {
      "epoch": 3.2705816762683515,
      "grad_norm": 1.0693212747573853,
      "learning_rate": 1.5288739480760317e-05,
      "loss": 0.81,
      "step": 17696
    },
    {
      "epoch": 3.270767515331723,
      "grad_norm": 1.014376163482666,
      "learning_rate": 1.528097678486453e-05,
      "loss": 0.8339,
      "step": 17697
    },
    {
      "epoch": 3.2709533543950937,
      "grad_norm": 0.9386370182037354,
      "learning_rate": 1.5273215897150917e-05,
      "loss": 0.7358,
      "step": 17698
    },
    {
      "epoch": 3.271139193458465,
      "grad_norm": 1.4261482954025269,
      "learning_rate": 1.5265456817785084e-05,
      "loss": 1.314,
      "step": 17699
    },
    {
      "epoch": 3.271325032521836,
      "grad_norm": 1.1061711311340332,
      "learning_rate": 1.5257699546932691e-05,
      "loss": 0.8965,
      "step": 17700
    },
    {
      "epoch": 3.2715108715852073,
      "grad_norm": 1.0086822509765625,
      "learning_rate": 1.5249944084759272e-05,
      "loss": 0.8394,
      "step": 17701
    },
    {
      "epoch": 3.271696710648578,
      "grad_norm": 0.904879093170166,
      "learning_rate": 1.5242190431430347e-05,
      "loss": 0.8432,
      "step": 17702
    },
    {
      "epoch": 3.2718825497119495,
      "grad_norm": 1.4062803983688354,
      "learning_rate": 1.5234438587111433e-05,
      "loss": 0.6721,
      "step": 17703
    },
    {
      "epoch": 3.2720683887753204,
      "grad_norm": 0.9530239105224609,
      "learning_rate": 1.5226688551967937e-05,
      "loss": 0.7427,
      "step": 17704
    },
    {
      "epoch": 3.2722542278386917,
      "grad_norm": 1.1069839000701904,
      "learning_rate": 1.521894032616532e-05,
      "loss": 0.848,
      "step": 17705
    },
    {
      "epoch": 3.272440066902063,
      "grad_norm": 0.9540766477584839,
      "learning_rate": 1.5211193909868936e-05,
      "loss": 0.8806,
      "step": 17706
    },
    {
      "epoch": 3.272625905965434,
      "grad_norm": 0.9043301939964294,
      "learning_rate": 1.5203449303244099e-05,
      "loss": 0.7207,
      "step": 17707
    },
    {
      "epoch": 3.2728117450288052,
      "grad_norm": 1.205649733543396,
      "learning_rate": 1.5195706506456131e-05,
      "loss": 0.8976,
      "step": 17708
    },
    {
      "epoch": 3.272997584092176,
      "grad_norm": 0.8512119054794312,
      "learning_rate": 1.5187965519670289e-05,
      "loss": 0.8219,
      "step": 17709
    },
    {
      "epoch": 3.2731834231555474,
      "grad_norm": 1.1259433031082153,
      "learning_rate": 1.518022634305175e-05,
      "loss": 0.7307,
      "step": 17710
    },
    {
      "epoch": 3.2733692622189183,
      "grad_norm": 0.9172109961509705,
      "learning_rate": 1.5172488976765741e-05,
      "loss": 0.868,
      "step": 17711
    },
    {
      "epoch": 3.2735551012822897,
      "grad_norm": 0.8422805070877075,
      "learning_rate": 1.5164753420977395e-05,
      "loss": 0.9534,
      "step": 17712
    },
    {
      "epoch": 3.2737409403456605,
      "grad_norm": 0.9498412013053894,
      "learning_rate": 1.5157019675851803e-05,
      "loss": 0.763,
      "step": 17713
    },
    {
      "epoch": 3.273926779409032,
      "grad_norm": 0.9260938763618469,
      "learning_rate": 1.5149287741553996e-05,
      "loss": 0.8064,
      "step": 17714
    },
    {
      "epoch": 3.2741126184724028,
      "grad_norm": 0.8831002712249756,
      "learning_rate": 1.5141557618249036e-05,
      "loss": 0.8015,
      "step": 17715
    },
    {
      "epoch": 3.274298457535774,
      "grad_norm": 1.01204252243042,
      "learning_rate": 1.5133829306101932e-05,
      "loss": 0.6571,
      "step": 17716
    },
    {
      "epoch": 3.274484296599145,
      "grad_norm": 1.1708745956420898,
      "learning_rate": 1.5126102805277609e-05,
      "loss": 0.5718,
      "step": 17717
    },
    {
      "epoch": 3.2746701356625163,
      "grad_norm": 0.9031075835227966,
      "learning_rate": 1.5118378115940968e-05,
      "loss": 0.6955,
      "step": 17718
    },
    {
      "epoch": 3.274855974725887,
      "grad_norm": 0.9439001083374023,
      "learning_rate": 1.5110655238256866e-05,
      "loss": 0.8054,
      "step": 17719
    },
    {
      "epoch": 3.2750418137892585,
      "grad_norm": 0.7462693452835083,
      "learning_rate": 1.510293417239017e-05,
      "loss": 0.5632,
      "step": 17720
    },
    {
      "epoch": 3.27522765285263,
      "grad_norm": 1.0460938215255737,
      "learning_rate": 1.509521491850565e-05,
      "loss": 0.7548,
      "step": 17721
    },
    {
      "epoch": 3.2754134919160007,
      "grad_norm": 0.8902409076690674,
      "learning_rate": 1.5087497476768054e-05,
      "loss": 0.6419,
      "step": 17722
    },
    {
      "epoch": 3.2755993309793716,
      "grad_norm": 0.8912127017974854,
      "learning_rate": 1.5079781847342123e-05,
      "loss": 0.8265,
      "step": 17723
    },
    {
      "epoch": 3.275785170042743,
      "grad_norm": 1.0797715187072754,
      "learning_rate": 1.5072068030392516e-05,
      "loss": 0.8039,
      "step": 17724
    },
    {
      "epoch": 3.2759710091061143,
      "grad_norm": 0.9005500674247742,
      "learning_rate": 1.5064356026083858e-05,
      "loss": 0.8328,
      "step": 17725
    },
    {
      "epoch": 3.276156848169485,
      "grad_norm": 1.153085470199585,
      "learning_rate": 1.5056645834580784e-05,
      "loss": 0.8964,
      "step": 17726
    },
    {
      "epoch": 3.2763426872328565,
      "grad_norm": 0.981286346912384,
      "learning_rate": 1.5048937456047818e-05,
      "loss": 0.7963,
      "step": 17727
    },
    {
      "epoch": 3.2765285262962274,
      "grad_norm": 0.9868923425674438,
      "learning_rate": 1.5041230890649526e-05,
      "loss": 0.8792,
      "step": 17728
    },
    {
      "epoch": 3.2767143653595987,
      "grad_norm": 0.8343085646629333,
      "learning_rate": 1.5033526138550359e-05,
      "loss": 0.858,
      "step": 17729
    },
    {
      "epoch": 3.2769002044229696,
      "grad_norm": 0.832370936870575,
      "learning_rate": 1.5025823199914747e-05,
      "loss": 0.6926,
      "step": 17730
    },
    {
      "epoch": 3.277086043486341,
      "grad_norm": 0.901152491569519,
      "learning_rate": 1.5018122074907136e-05,
      "loss": 0.8358,
      "step": 17731
    },
    {
      "epoch": 3.277271882549712,
      "grad_norm": 0.8559943437576294,
      "learning_rate": 1.5010422763691878e-05,
      "loss": 0.5558,
      "step": 17732
    },
    {
      "epoch": 3.277457721613083,
      "grad_norm": 0.8962408304214478,
      "learning_rate": 1.5002725266433303e-05,
      "loss": 0.7495,
      "step": 17733
    },
    {
      "epoch": 3.277643560676454,
      "grad_norm": 0.9491870999336243,
      "learning_rate": 1.4995029583295672e-05,
      "loss": 0.7545,
      "step": 17734
    },
    {
      "epoch": 3.2778293997398253,
      "grad_norm": 0.9699558615684509,
      "learning_rate": 1.4987335714443274e-05,
      "loss": 0.8994,
      "step": 17735
    },
    {
      "epoch": 3.2780152388031967,
      "grad_norm": 0.9728564023971558,
      "learning_rate": 1.49796436600403e-05,
      "loss": 0.8333,
      "step": 17736
    },
    {
      "epoch": 3.2782010778665676,
      "grad_norm": 0.984086275100708,
      "learning_rate": 1.497195342025094e-05,
      "loss": 0.8111,
      "step": 17737
    },
    {
      "epoch": 3.2783869169299384,
      "grad_norm": 1.0738487243652344,
      "learning_rate": 1.496426499523933e-05,
      "loss": 1.048,
      "step": 17738
    },
    {
      "epoch": 3.2785727559933098,
      "grad_norm": 1.0294325351715088,
      "learning_rate": 1.4956578385169528e-05,
      "loss": 0.7576,
      "step": 17739
    },
    {
      "epoch": 3.278758595056681,
      "grad_norm": 0.9861834645271301,
      "learning_rate": 1.4948893590205637e-05,
      "loss": 0.7033,
      "step": 17740
    },
    {
      "epoch": 3.278944434120052,
      "grad_norm": 1.0322951078414917,
      "learning_rate": 1.494121061051167e-05,
      "loss": 0.6938,
      "step": 17741
    },
    {
      "epoch": 3.2791302731834233,
      "grad_norm": 0.7800756692886353,
      "learning_rate": 1.4933529446251571e-05,
      "loss": 0.636,
      "step": 17742
    },
    {
      "epoch": 3.279316112246794,
      "grad_norm": 0.9137340784072876,
      "learning_rate": 1.4925850097589323e-05,
      "loss": 0.8715,
      "step": 17743
    },
    {
      "epoch": 3.2795019513101655,
      "grad_norm": 0.8566145896911621,
      "learning_rate": 1.4918172564688815e-05,
      "loss": 0.7664,
      "step": 17744
    },
    {
      "epoch": 3.2796877903735364,
      "grad_norm": 0.8737754821777344,
      "learning_rate": 1.4910496847713906e-05,
      "loss": 0.7809,
      "step": 17745
    },
    {
      "epoch": 3.2798736294369077,
      "grad_norm": 0.9414018988609314,
      "learning_rate": 1.4902822946828398e-05,
      "loss": 0.7845,
      "step": 17746
    },
    {
      "epoch": 3.2800594685002786,
      "grad_norm": 0.845181405544281,
      "learning_rate": 1.4895150862196105e-05,
      "loss": 0.8409,
      "step": 17747
    },
    {
      "epoch": 3.28024530756365,
      "grad_norm": 0.9034122228622437,
      "learning_rate": 1.4887480593980797e-05,
      "loss": 0.8416,
      "step": 17748
    },
    {
      "epoch": 3.280431146627021,
      "grad_norm": 0.8876949548721313,
      "learning_rate": 1.487981214234615e-05,
      "loss": 0.9379,
      "step": 17749
    },
    {
      "epoch": 3.280616985690392,
      "grad_norm": 0.8697424530982971,
      "learning_rate": 1.4872145507455825e-05,
      "loss": 0.713,
      "step": 17750
    },
    {
      "epoch": 3.280802824753763,
      "grad_norm": 0.8058562278747559,
      "learning_rate": 1.486448068947348e-05,
      "loss": 0.7265,
      "step": 17751
    },
    {
      "epoch": 3.2809886638171344,
      "grad_norm": 0.9465721249580383,
      "learning_rate": 1.4856817688562708e-05,
      "loss": 0.6551,
      "step": 17752
    },
    {
      "epoch": 3.2811745028805053,
      "grad_norm": 1.1225144863128662,
      "learning_rate": 1.4849156504887053e-05,
      "loss": 0.8225,
      "step": 17753
    },
    {
      "epoch": 3.2813603419438766,
      "grad_norm": 1.046244740486145,
      "learning_rate": 1.4841497138609995e-05,
      "loss": 0.7533,
      "step": 17754
    },
    {
      "epoch": 3.281546181007248,
      "grad_norm": 0.9831907749176025,
      "learning_rate": 1.4833839589895071e-05,
      "loss": 0.7502,
      "step": 17755
    },
    {
      "epoch": 3.281732020070619,
      "grad_norm": 0.998605489730835,
      "learning_rate": 1.4826183858905685e-05,
      "loss": 0.7527,
      "step": 17756
    },
    {
      "epoch": 3.28191785913399,
      "grad_norm": 0.9687653183937073,
      "learning_rate": 1.4818529945805227e-05,
      "loss": 0.7545,
      "step": 17757
    },
    {
      "epoch": 3.282103698197361,
      "grad_norm": 1.0033705234527588,
      "learning_rate": 1.4810877850757088e-05,
      "loss": 0.9128,
      "step": 17758
    },
    {
      "epoch": 3.2822895372607324,
      "grad_norm": 0.9830369353294373,
      "learning_rate": 1.4803227573924539e-05,
      "loss": 0.8042,
      "step": 17759
    },
    {
      "epoch": 3.2824753763241032,
      "grad_norm": 0.9130050539970398,
      "learning_rate": 1.4795579115470925e-05,
      "loss": 0.8025,
      "step": 17760
    },
    {
      "epoch": 3.2826612153874746,
      "grad_norm": 0.6415044069290161,
      "learning_rate": 1.4787932475559462e-05,
      "loss": 0.3121,
      "step": 17761
    },
    {
      "epoch": 3.2828470544508455,
      "grad_norm": 0.8551591634750366,
      "learning_rate": 1.478028765435332e-05,
      "loss": 0.8739,
      "step": 17762
    },
    {
      "epoch": 3.283032893514217,
      "grad_norm": 1.0814300775527954,
      "learning_rate": 1.477264465201572e-05,
      "loss": 0.8861,
      "step": 17763
    },
    {
      "epoch": 3.2832187325775877,
      "grad_norm": 0.9738458395004272,
      "learning_rate": 1.476500346870977e-05,
      "loss": 0.8235,
      "step": 17764
    },
    {
      "epoch": 3.283404571640959,
      "grad_norm": 0.9195381999015808,
      "learning_rate": 1.4757364104598547e-05,
      "loss": 0.9812,
      "step": 17765
    },
    {
      "epoch": 3.28359041070433,
      "grad_norm": 0.8763140439987183,
      "learning_rate": 1.4749726559845078e-05,
      "loss": 0.9033,
      "step": 17766
    },
    {
      "epoch": 3.283776249767701,
      "grad_norm": 0.9802149534225464,
      "learning_rate": 1.474209083461241e-05,
      "loss": 0.9211,
      "step": 17767
    },
    {
      "epoch": 3.283962088831072,
      "grad_norm": 2.1971049308776855,
      "learning_rate": 1.4734456929063523e-05,
      "loss": 1.2841,
      "step": 17768
    },
    {
      "epoch": 3.2841479278944434,
      "grad_norm": 1.1380846500396729,
      "learning_rate": 1.472682484336133e-05,
      "loss": 1.0558,
      "step": 17769
    },
    {
      "epoch": 3.2843337669578148,
      "grad_norm": 0.9155475497245789,
      "learning_rate": 1.4719194577668704e-05,
      "loss": 0.8186,
      "step": 17770
    },
    {
      "epoch": 3.2845196060211856,
      "grad_norm": 0.9848891496658325,
      "learning_rate": 1.4711566132148536e-05,
      "loss": 0.7564,
      "step": 17771
    },
    {
      "epoch": 3.2847054450845565,
      "grad_norm": 0.9461994171142578,
      "learning_rate": 1.4703939506963637e-05,
      "loss": 0.8314,
      "step": 17772
    },
    {
      "epoch": 3.284891284147928,
      "grad_norm": 1.004651665687561,
      "learning_rate": 1.469631470227678e-05,
      "loss": 0.8203,
      "step": 17773
    },
    {
      "epoch": 3.285077123211299,
      "grad_norm": 0.7294480800628662,
      "learning_rate": 1.4688691718250669e-05,
      "loss": 0.7021,
      "step": 17774
    },
    {
      "epoch": 3.28526296227467,
      "grad_norm": 1.2479922771453857,
      "learning_rate": 1.468107055504806e-05,
      "loss": 1.0149,
      "step": 17775
    },
    {
      "epoch": 3.2854488013380414,
      "grad_norm": 0.9774560332298279,
      "learning_rate": 1.4673451212831579e-05,
      "loss": 0.8686,
      "step": 17776
    },
    {
      "epoch": 3.2856346404014123,
      "grad_norm": 1.1198142766952515,
      "learning_rate": 1.4665833691763864e-05,
      "loss": 0.8014,
      "step": 17777
    },
    {
      "epoch": 3.2858204794647836,
      "grad_norm": 1.036876916885376,
      "learning_rate": 1.4658217992007462e-05,
      "loss": 0.7655,
      "step": 17778
    },
    {
      "epoch": 3.2860063185281545,
      "grad_norm": 1.0837442874908447,
      "learning_rate": 1.4650604113724953e-05,
      "loss": 0.9843,
      "step": 17779
    },
    {
      "epoch": 3.286192157591526,
      "grad_norm": 0.9364877343177795,
      "learning_rate": 1.4642992057078853e-05,
      "loss": 0.739,
      "step": 17780
    },
    {
      "epoch": 3.2863779966548967,
      "grad_norm": 1.0207990407943726,
      "learning_rate": 1.4635381822231597e-05,
      "loss": 0.8071,
      "step": 17781
    },
    {
      "epoch": 3.286563835718268,
      "grad_norm": 0.8080578446388245,
      "learning_rate": 1.4627773409345613e-05,
      "loss": 0.8426,
      "step": 17782
    },
    {
      "epoch": 3.286749674781639,
      "grad_norm": 0.7918444871902466,
      "learning_rate": 1.4620166818583324e-05,
      "loss": 0.6927,
      "step": 17783
    },
    {
      "epoch": 3.2869355138450103,
      "grad_norm": 1.3243615627288818,
      "learning_rate": 1.4612562050107048e-05,
      "loss": 0.9785,
      "step": 17784
    },
    {
      "epoch": 3.2871213529083816,
      "grad_norm": 0.8511014580726624,
      "learning_rate": 1.4604959104079119e-05,
      "loss": 0.7664,
      "step": 17785
    },
    {
      "epoch": 3.2873071919717525,
      "grad_norm": 1.721552848815918,
      "learning_rate": 1.4597357980661764e-05,
      "loss": 0.8782,
      "step": 17786
    },
    {
      "epoch": 3.2874930310351234,
      "grad_norm": 1.078242301940918,
      "learning_rate": 1.4589758680017263e-05,
      "loss": 0.8383,
      "step": 17787
    },
    {
      "epoch": 3.2876788700984947,
      "grad_norm": 0.9365386962890625,
      "learning_rate": 1.4582161202307776e-05,
      "loss": 0.6878,
      "step": 17788
    },
    {
      "epoch": 3.287864709161866,
      "grad_norm": 0.6306977272033691,
      "learning_rate": 1.4574565547695495e-05,
      "loss": 0.4308,
      "step": 17789
    },
    {
      "epoch": 3.288050548225237,
      "grad_norm": 0.8376436829566956,
      "learning_rate": 1.4566971716342493e-05,
      "loss": 0.9278,
      "step": 17790
    },
    {
      "epoch": 3.2882363872886082,
      "grad_norm": 0.8800222873687744,
      "learning_rate": 1.4559379708410904e-05,
      "loss": 0.7149,
      "step": 17791
    },
    {
      "epoch": 3.288422226351979,
      "grad_norm": 0.9601709842681885,
      "learning_rate": 1.455178952406272e-05,
      "loss": 0.9445,
      "step": 17792
    },
    {
      "epoch": 3.2886080654153504,
      "grad_norm": 1.3182393312454224,
      "learning_rate": 1.4544201163459959e-05,
      "loss": 0.7908,
      "step": 17793
    },
    {
      "epoch": 3.2887939044787213,
      "grad_norm": 0.8644195795059204,
      "learning_rate": 1.4536614626764556e-05,
      "loss": 0.5513,
      "step": 17794
    },
    {
      "epoch": 3.2889797435420927,
      "grad_norm": 1.2175333499908447,
      "learning_rate": 1.452902991413847e-05,
      "loss": 0.7819,
      "step": 17795
    },
    {
      "epoch": 3.2891655826054635,
      "grad_norm": 0.7996311187744141,
      "learning_rate": 1.4521447025743572e-05,
      "loss": 0.9284,
      "step": 17796
    },
    {
      "epoch": 3.289351421668835,
      "grad_norm": 0.7490044236183167,
      "learning_rate": 1.4513865961741702e-05,
      "loss": 0.5596,
      "step": 17797
    },
    {
      "epoch": 3.2895372607322058,
      "grad_norm": 1.0699917078018188,
      "learning_rate": 1.4506286722294627e-05,
      "loss": 0.6434,
      "step": 17798
    },
    {
      "epoch": 3.289723099795577,
      "grad_norm": 0.8811341524124146,
      "learning_rate": 1.4498709307564162e-05,
      "loss": 0.932,
      "step": 17799
    },
    {
      "epoch": 3.289908938858948,
      "grad_norm": 0.9763567447662354,
      "learning_rate": 1.4491133717712036e-05,
      "loss": 0.6995,
      "step": 17800
    },
    {
      "epoch": 3.2900947779223193,
      "grad_norm": 0.79994797706604,
      "learning_rate": 1.4483559952899929e-05,
      "loss": 0.7669,
      "step": 17801
    },
    {
      "epoch": 3.29028061698569,
      "grad_norm": 0.8885137438774109,
      "learning_rate": 1.4475988013289454e-05,
      "loss": 0.9014,
      "step": 17802
    },
    {
      "epoch": 3.2904664560490615,
      "grad_norm": 1.0276567935943604,
      "learning_rate": 1.4468417899042275e-05,
      "loss": 0.8182,
      "step": 17803
    },
    {
      "epoch": 3.290652295112433,
      "grad_norm": 0.9631934762001038,
      "learning_rate": 1.4460849610319937e-05,
      "loss": 0.7452,
      "step": 17804
    },
    {
      "epoch": 3.2908381341758037,
      "grad_norm": 1.0075114965438843,
      "learning_rate": 1.4453283147283969e-05,
      "loss": 0.7062,
      "step": 17805
    },
    {
      "epoch": 3.291023973239175,
      "grad_norm": 0.9544219970703125,
      "learning_rate": 1.4445718510095852e-05,
      "loss": 0.7634,
      "step": 17806
    },
    {
      "epoch": 3.291209812302546,
      "grad_norm": 0.8006826043128967,
      "learning_rate": 1.4438155698917077e-05,
      "loss": 0.7816,
      "step": 17807
    },
    {
      "epoch": 3.2913956513659173,
      "grad_norm": 0.8424093127250671,
      "learning_rate": 1.443059471390904e-05,
      "loss": 0.8006,
      "step": 17808
    },
    {
      "epoch": 3.291581490429288,
      "grad_norm": 0.9989455342292786,
      "learning_rate": 1.4423035555233088e-05,
      "loss": 0.6723,
      "step": 17809
    },
    {
      "epoch": 3.2917673294926595,
      "grad_norm": 1.9413286447525024,
      "learning_rate": 1.4415478223050593e-05,
      "loss": 1.1197,
      "step": 17810
    },
    {
      "epoch": 3.2919531685560304,
      "grad_norm": 1.029294729232788,
      "learning_rate": 1.440792271752287e-05,
      "loss": 1.0826,
      "step": 17811
    },
    {
      "epoch": 3.2921390076194017,
      "grad_norm": 0.8019607663154602,
      "learning_rate": 1.4400369038811145e-05,
      "loss": 0.7248,
      "step": 17812
    },
    {
      "epoch": 3.2923248466827726,
      "grad_norm": 1.039353370666504,
      "learning_rate": 1.4392817187076658e-05,
      "loss": 1.0597,
      "step": 17813
    },
    {
      "epoch": 3.292510685746144,
      "grad_norm": 0.9127644300460815,
      "learning_rate": 1.4385267162480554e-05,
      "loss": 0.771,
      "step": 17814
    },
    {
      "epoch": 3.292696524809515,
      "grad_norm": 0.8197526931762695,
      "learning_rate": 1.4377718965184017e-05,
      "loss": 0.7415,
      "step": 17815
    },
    {
      "epoch": 3.292882363872886,
      "grad_norm": 1.0656996965408325,
      "learning_rate": 1.437017259534814e-05,
      "loss": 1.028,
      "step": 17816
    },
    {
      "epoch": 3.293068202936257,
      "grad_norm": 1.1229791641235352,
      "learning_rate": 1.4362628053133986e-05,
      "loss": 0.7646,
      "step": 17817
    },
    {
      "epoch": 3.2932540419996283,
      "grad_norm": 0.8567220568656921,
      "learning_rate": 1.4355085338702556e-05,
      "loss": 0.8046,
      "step": 17818
    },
    {
      "epoch": 3.2934398810629997,
      "grad_norm": 0.9808992743492126,
      "learning_rate": 1.4347544452214868e-05,
      "loss": 0.8,
      "step": 17819
    },
    {
      "epoch": 3.2936257201263706,
      "grad_norm": 0.7619749903678894,
      "learning_rate": 1.4340005393831845e-05,
      "loss": 0.6574,
      "step": 17820
    },
    {
      "epoch": 3.2938115591897414,
      "grad_norm": 1.034301519393921,
      "learning_rate": 1.433246816371443e-05,
      "loss": 0.9074,
      "step": 17821
    },
    {
      "epoch": 3.2939973982531128,
      "grad_norm": 0.8320847153663635,
      "learning_rate": 1.4324932762023447e-05,
      "loss": 0.728,
      "step": 17822
    },
    {
      "epoch": 3.294183237316484,
      "grad_norm": 0.8412254452705383,
      "learning_rate": 1.4317399188919767e-05,
      "loss": 0.6114,
      "step": 17823
    },
    {
      "epoch": 3.294369076379855,
      "grad_norm": 1.0957362651824951,
      "learning_rate": 1.4309867444564173e-05,
      "loss": 0.8073,
      "step": 17824
    },
    {
      "epoch": 3.2945549154432263,
      "grad_norm": 0.9523847699165344,
      "learning_rate": 1.4302337529117404e-05,
      "loss": 0.6516,
      "step": 17825
    },
    {
      "epoch": 3.294740754506597,
      "grad_norm": 1.0757582187652588,
      "learning_rate": 1.4294809442740153e-05,
      "loss": 0.9843,
      "step": 17826
    },
    {
      "epoch": 3.2949265935699685,
      "grad_norm": 1.077390193939209,
      "learning_rate": 1.4287283185593148e-05,
      "loss": 0.9113,
      "step": 17827
    },
    {
      "epoch": 3.2951124326333394,
      "grad_norm": 0.9525008797645569,
      "learning_rate": 1.4279758757836992e-05,
      "loss": 0.9461,
      "step": 17828
    },
    {
      "epoch": 3.2952982716967107,
      "grad_norm": 0.9313654899597168,
      "learning_rate": 1.427223615963228e-05,
      "loss": 0.7244,
      "step": 17829
    },
    {
      "epoch": 3.2954841107600816,
      "grad_norm": 0.8749706149101257,
      "learning_rate": 1.4264715391139527e-05,
      "loss": 0.8138,
      "step": 17830
    },
    {
      "epoch": 3.295669949823453,
      "grad_norm": 0.92448890209198,
      "learning_rate": 1.425719645251934e-05,
      "loss": 0.8129,
      "step": 17831
    },
    {
      "epoch": 3.295855788886824,
      "grad_norm": 1.0686804056167603,
      "learning_rate": 1.4249679343932154e-05,
      "loss": 0.7959,
      "step": 17832
    },
    {
      "epoch": 3.296041627950195,
      "grad_norm": 0.9681421518325806,
      "learning_rate": 1.424216406553841e-05,
      "loss": 1.0192,
      "step": 17833
    },
    {
      "epoch": 3.296227467013566,
      "grad_norm": 0.9680215120315552,
      "learning_rate": 1.423465061749849e-05,
      "loss": 1.0388,
      "step": 17834
    },
    {
      "epoch": 3.2964133060769374,
      "grad_norm": 1.0418201684951782,
      "learning_rate": 1.42271389999728e-05,
      "loss": 0.923,
      "step": 17835
    },
    {
      "epoch": 3.2965991451403083,
      "grad_norm": 1.0079830884933472,
      "learning_rate": 1.421962921312162e-05,
      "loss": 0.7838,
      "step": 17836
    },
    {
      "epoch": 3.2967849842036796,
      "grad_norm": 0.782057523727417,
      "learning_rate": 1.421212125710526e-05,
      "loss": 0.8256,
      "step": 17837
    },
    {
      "epoch": 3.296970823267051,
      "grad_norm": 0.8839608430862427,
      "learning_rate": 1.4204615132083931e-05,
      "loss": 0.5899,
      "step": 17838
    },
    {
      "epoch": 3.297156662330422,
      "grad_norm": 1.164730429649353,
      "learning_rate": 1.4197110838217886e-05,
      "loss": 1.0913,
      "step": 17839
    },
    {
      "epoch": 3.297342501393793,
      "grad_norm": 0.8485312461853027,
      "learning_rate": 1.4189608375667263e-05,
      "loss": 0.8149,
      "step": 17840
    },
    {
      "epoch": 3.297528340457164,
      "grad_norm": 0.94539874792099,
      "learning_rate": 1.4182107744592166e-05,
      "loss": 0.8123,
      "step": 17841
    },
    {
      "epoch": 3.2977141795205354,
      "grad_norm": 0.8088021278381348,
      "learning_rate": 1.4174608945152712e-05,
      "loss": 0.7983,
      "step": 17842
    },
    {
      "epoch": 3.2979000185839062,
      "grad_norm": 0.8575440049171448,
      "learning_rate": 1.4167111977508973e-05,
      "loss": 0.949,
      "step": 17843
    },
    {
      "epoch": 3.2980858576472776,
      "grad_norm": 0.7935901284217834,
      "learning_rate": 1.4159616841820933e-05,
      "loss": 0.542,
      "step": 17844
    },
    {
      "epoch": 3.2982716967106485,
      "grad_norm": 1.159449577331543,
      "learning_rate": 1.4152123538248563e-05,
      "loss": 1.146,
      "step": 17845
    },
    {
      "epoch": 3.29845753577402,
      "grad_norm": 1.0020778179168701,
      "learning_rate": 1.4144632066951779e-05,
      "loss": 0.702,
      "step": 17846
    },
    {
      "epoch": 3.2986433748373907,
      "grad_norm": 0.9826176762580872,
      "learning_rate": 1.4137142428090511e-05,
      "loss": 0.8788,
      "step": 17847
    },
    {
      "epoch": 3.298829213900762,
      "grad_norm": 0.8390106558799744,
      "learning_rate": 1.4129654621824583e-05,
      "loss": 0.7019,
      "step": 17848
    },
    {
      "epoch": 3.299015052964133,
      "grad_norm": 0.8387085795402527,
      "learning_rate": 1.4122168648313816e-05,
      "loss": 0.6126,
      "step": 17849
    },
    {
      "epoch": 3.299200892027504,
      "grad_norm": 0.884889543056488,
      "learning_rate": 1.411468450771798e-05,
      "loss": 0.7499,
      "step": 17850
    },
    {
      "epoch": 3.299386731090875,
      "grad_norm": 0.8823387026786804,
      "learning_rate": 1.4107202200196801e-05,
      "loss": 0.6584,
      "step": 17851
    },
    {
      "epoch": 3.2995725701542464,
      "grad_norm": 0.9210385084152222,
      "learning_rate": 1.409972172591002e-05,
      "loss": 0.7476,
      "step": 17852
    },
    {
      "epoch": 3.2997584092176178,
      "grad_norm": 0.9333329200744629,
      "learning_rate": 1.4092243085017276e-05,
      "loss": 0.8189,
      "step": 17853
    },
    {
      "epoch": 3.2999442482809886,
      "grad_norm": 0.8348424434661865,
      "learning_rate": 1.4084766277678152e-05,
      "loss": 0.741,
      "step": 17854
    },
    {
      "epoch": 3.30013008734436,
      "grad_norm": 1.09077787399292,
      "learning_rate": 1.4077291304052275e-05,
      "loss": 0.625,
      "step": 17855
    },
    {
      "epoch": 3.300315926407731,
      "grad_norm": 0.8905661702156067,
      "learning_rate": 1.4069818164299165e-05,
      "loss": 0.6849,
      "step": 17856
    },
    {
      "epoch": 3.300501765471102,
      "grad_norm": 0.8002813458442688,
      "learning_rate": 1.4062346858578335e-05,
      "loss": 0.7506,
      "step": 17857
    },
    {
      "epoch": 3.300687604534473,
      "grad_norm": 1.1196472644805908,
      "learning_rate": 1.4054877387049203e-05,
      "loss": 0.9203,
      "step": 17858
    },
    {
      "epoch": 3.3008734435978444,
      "grad_norm": 1.0638034343719482,
      "learning_rate": 1.4047409749871255e-05,
      "loss": 0.99,
      "step": 17859
    },
    {
      "epoch": 3.3010592826612153,
      "grad_norm": 0.917509913444519,
      "learning_rate": 1.4039943947203838e-05,
      "loss": 0.6758,
      "step": 17860
    },
    {
      "epoch": 3.3012451217245866,
      "grad_norm": 0.9278843998908997,
      "learning_rate": 1.4032479979206282e-05,
      "loss": 0.9307,
      "step": 17861
    },
    {
      "epoch": 3.3014309607879575,
      "grad_norm": 0.9846850037574768,
      "learning_rate": 1.4025017846037924e-05,
      "loss": 0.8342,
      "step": 17862
    },
    {
      "epoch": 3.301616799851329,
      "grad_norm": 0.9571072459220886,
      "learning_rate": 1.401755754785804e-05,
      "loss": 1.0341,
      "step": 17863
    },
    {
      "epoch": 3.3018026389146997,
      "grad_norm": 1.0230777263641357,
      "learning_rate": 1.4010099084825834e-05,
      "loss": 0.9664,
      "step": 17864
    },
    {
      "epoch": 3.301988477978071,
      "grad_norm": 1.0085383653640747,
      "learning_rate": 1.4002642457100502e-05,
      "loss": 1.0487,
      "step": 17865
    },
    {
      "epoch": 3.302174317041442,
      "grad_norm": 0.8759761452674866,
      "learning_rate": 1.3995187664841158e-05,
      "loss": 0.8395,
      "step": 17866
    },
    {
      "epoch": 3.3023601561048133,
      "grad_norm": 0.9805045127868652,
      "learning_rate": 1.398773470820698e-05,
      "loss": 0.7442,
      "step": 17867
    },
    {
      "epoch": 3.3025459951681846,
      "grad_norm": 0.8271703124046326,
      "learning_rate": 1.398028358735699e-05,
      "loss": 0.5299,
      "step": 17868
    },
    {
      "epoch": 3.3027318342315555,
      "grad_norm": 0.9561894536018372,
      "learning_rate": 1.3972834302450233e-05,
      "loss": 0.8613,
      "step": 17869
    },
    {
      "epoch": 3.3029176732949264,
      "grad_norm": 0.9207913875579834,
      "learning_rate": 1.396538685364568e-05,
      "loss": 1.0006,
      "step": 17870
    },
    {
      "epoch": 3.3031035123582977,
      "grad_norm": 1.1078895330429077,
      "learning_rate": 1.3957941241102324e-05,
      "loss": 0.9432,
      "step": 17871
    },
    {
      "epoch": 3.303289351421669,
      "grad_norm": 0.8856802582740784,
      "learning_rate": 1.395049746497904e-05,
      "loss": 0.8406,
      "step": 17872
    },
    {
      "epoch": 3.30347519048504,
      "grad_norm": 0.97230464220047,
      "learning_rate": 1.3943055525434745e-05,
      "loss": 0.9904,
      "step": 17873
    },
    {
      "epoch": 3.3036610295484112,
      "grad_norm": 0.9858556985855103,
      "learning_rate": 1.3935615422628223e-05,
      "loss": 0.8607,
      "step": 17874
    },
    {
      "epoch": 3.303846868611782,
      "grad_norm": 0.9251289367675781,
      "learning_rate": 1.3928177156718314e-05,
      "loss": 0.7276,
      "step": 17875
    },
    {
      "epoch": 3.3040327076751534,
      "grad_norm": 1.1109967231750488,
      "learning_rate": 1.3920740727863768e-05,
      "loss": 0.8898,
      "step": 17876
    },
    {
      "epoch": 3.3042185467385243,
      "grad_norm": 0.7776536345481873,
      "learning_rate": 1.3913306136223292e-05,
      "loss": 0.7946,
      "step": 17877
    },
    {
      "epoch": 3.3044043858018957,
      "grad_norm": 1.0654406547546387,
      "learning_rate": 1.3905873381955536e-05,
      "loss": 0.9813,
      "step": 17878
    },
    {
      "epoch": 3.3045902248652665,
      "grad_norm": 0.8269950151443481,
      "learning_rate": 1.3898442465219198e-05,
      "loss": 0.8146,
      "step": 17879
    },
    {
      "epoch": 3.304776063928638,
      "grad_norm": 0.957031786441803,
      "learning_rate": 1.3891013386172846e-05,
      "loss": 0.7226,
      "step": 17880
    },
    {
      "epoch": 3.3049619029920088,
      "grad_norm": 1.144073724746704,
      "learning_rate": 1.3883586144975026e-05,
      "loss": 0.7355,
      "step": 17881
    },
    {
      "epoch": 3.30514774205538,
      "grad_norm": 0.9513867497444153,
      "learning_rate": 1.3876160741784294e-05,
      "loss": 0.9539,
      "step": 17882
    },
    {
      "epoch": 3.305333581118751,
      "grad_norm": 0.991651177406311,
      "learning_rate": 1.3868737176759106e-05,
      "loss": 0.89,
      "step": 17883
    },
    {
      "epoch": 3.3055194201821223,
      "grad_norm": 1.1039535999298096,
      "learning_rate": 1.386131545005792e-05,
      "loss": 0.899,
      "step": 17884
    },
    {
      "epoch": 3.305705259245493,
      "grad_norm": 0.7326856851577759,
      "learning_rate": 1.3853895561839147e-05,
      "loss": 0.7802,
      "step": 17885
    },
    {
      "epoch": 3.3058910983088645,
      "grad_norm": 0.93720942735672,
      "learning_rate": 1.3846477512261124e-05,
      "loss": 0.6857,
      "step": 17886
    },
    {
      "epoch": 3.306076937372236,
      "grad_norm": 0.8614287972450256,
      "learning_rate": 1.3839061301482203e-05,
      "loss": 0.7416,
      "step": 17887
    },
    {
      "epoch": 3.3062627764356067,
      "grad_norm": 0.9519493579864502,
      "learning_rate": 1.3831646929660668e-05,
      "loss": 0.9447,
      "step": 17888
    },
    {
      "epoch": 3.306448615498978,
      "grad_norm": 0.8550201058387756,
      "learning_rate": 1.382423439695476e-05,
      "loss": 0.8227,
      "step": 17889
    },
    {
      "epoch": 3.306634454562349,
      "grad_norm": 0.94979327917099,
      "learning_rate": 1.3816823703522663e-05,
      "loss": 0.9126,
      "step": 17890
    },
    {
      "epoch": 3.3068202936257203,
      "grad_norm": 1.3942769765853882,
      "learning_rate": 1.3809414849522584e-05,
      "loss": 0.785,
      "step": 17891
    },
    {
      "epoch": 3.307006132689091,
      "grad_norm": 1.0408034324645996,
      "learning_rate": 1.3802007835112641e-05,
      "loss": 0.9444,
      "step": 17892
    },
    {
      "epoch": 3.3071919717524625,
      "grad_norm": 1.0535863637924194,
      "learning_rate": 1.3794602660450895e-05,
      "loss": 1.0558,
      "step": 17893
    },
    {
      "epoch": 3.3073778108158334,
      "grad_norm": 0.7811908721923828,
      "learning_rate": 1.3787199325695422e-05,
      "loss": 0.6766,
      "step": 17894
    },
    {
      "epoch": 3.3075636498792047,
      "grad_norm": 0.8901042938232422,
      "learning_rate": 1.3779797831004248e-05,
      "loss": 0.7227,
      "step": 17895
    },
    {
      "epoch": 3.3077494889425756,
      "grad_norm": 0.8555505871772766,
      "learning_rate": 1.3772398176535339e-05,
      "loss": 0.6997,
      "step": 17896
    },
    {
      "epoch": 3.307935328005947,
      "grad_norm": 0.9381975531578064,
      "learning_rate": 1.3765000362446612e-05,
      "loss": 0.8737,
      "step": 17897
    },
    {
      "epoch": 3.308121167069318,
      "grad_norm": 1.1561797857284546,
      "learning_rate": 1.3757604388895951e-05,
      "loss": 0.843,
      "step": 17898
    },
    {
      "epoch": 3.308307006132689,
      "grad_norm": 1.1110519170761108,
      "learning_rate": 1.3750210256041241e-05,
      "loss": 0.8357,
      "step": 17899
    },
    {
      "epoch": 3.30849284519606,
      "grad_norm": 1.0905356407165527,
      "learning_rate": 1.374281796404029e-05,
      "loss": 0.7613,
      "step": 17900
    },
    {
      "epoch": 3.3086786842594313,
      "grad_norm": 0.903249204158783,
      "learning_rate": 1.3735427513050836e-05,
      "loss": 0.6294,
      "step": 17901
    },
    {
      "epoch": 3.3088645233228027,
      "grad_norm": 1.348289132118225,
      "learning_rate": 1.3728038903230666e-05,
      "loss": 0.9079,
      "step": 17902
    },
    {
      "epoch": 3.3090503623861736,
      "grad_norm": 0.9120255708694458,
      "learning_rate": 1.3720652134737466e-05,
      "loss": 0.9507,
      "step": 17903
    },
    {
      "epoch": 3.3092362014495444,
      "grad_norm": 0.9518958330154419,
      "learning_rate": 1.3713267207728863e-05,
      "loss": 0.9064,
      "step": 17904
    },
    {
      "epoch": 3.3094220405129158,
      "grad_norm": 0.8687372207641602,
      "learning_rate": 1.3705884122362512e-05,
      "loss": 0.9743,
      "step": 17905
    },
    {
      "epoch": 3.309607879576287,
      "grad_norm": 0.9766872525215149,
      "learning_rate": 1.3698502878795961e-05,
      "loss": 0.7682,
      "step": 17906
    },
    {
      "epoch": 3.309793718639658,
      "grad_norm": 0.859889566898346,
      "learning_rate": 1.3691123477186785e-05,
      "loss": 0.7043,
      "step": 17907
    },
    {
      "epoch": 3.3099795577030293,
      "grad_norm": 1.0203168392181396,
      "learning_rate": 1.368374591769247e-05,
      "loss": 0.7855,
      "step": 17908
    },
    {
      "epoch": 3.3101653967664,
      "grad_norm": 0.8277616500854492,
      "learning_rate": 1.3676370200470479e-05,
      "loss": 0.7292,
      "step": 17909
    },
    {
      "epoch": 3.3103512358297715,
      "grad_norm": 0.981030285358429,
      "learning_rate": 1.3668996325678196e-05,
      "loss": 0.727,
      "step": 17910
    },
    {
      "epoch": 3.3105370748931424,
      "grad_norm": 0.9350651502609253,
      "learning_rate": 1.3661624293473063e-05,
      "loss": 0.8273,
      "step": 17911
    },
    {
      "epoch": 3.3107229139565137,
      "grad_norm": 0.9313578605651855,
      "learning_rate": 1.3654254104012398e-05,
      "loss": 0.8877,
      "step": 17912
    },
    {
      "epoch": 3.3109087530198846,
      "grad_norm": 0.9104059934616089,
      "learning_rate": 1.3646885757453488e-05,
      "loss": 0.7371,
      "step": 17913
    },
    {
      "epoch": 3.311094592083256,
      "grad_norm": 0.984149694442749,
      "learning_rate": 1.3639519253953604e-05,
      "loss": 1.0117,
      "step": 17914
    },
    {
      "epoch": 3.311280431146627,
      "grad_norm": 1.1708226203918457,
      "learning_rate": 1.363215459367001e-05,
      "loss": 0.7616,
      "step": 17915
    },
    {
      "epoch": 3.311466270209998,
      "grad_norm": 0.8822725415229797,
      "learning_rate": 1.3624791776759849e-05,
      "loss": 0.8812,
      "step": 17916
    },
    {
      "epoch": 3.3116521092733695,
      "grad_norm": 1.0658091306686401,
      "learning_rate": 1.3617430803380293e-05,
      "loss": 1.0615,
      "step": 17917
    },
    {
      "epoch": 3.3118379483367404,
      "grad_norm": 1.0318050384521484,
      "learning_rate": 1.3610071673688418e-05,
      "loss": 1.0045,
      "step": 17918
    },
    {
      "epoch": 3.3120237874001113,
      "grad_norm": 1.094164490699768,
      "learning_rate": 1.360271438784133e-05,
      "loss": 0.8698,
      "step": 17919
    },
    {
      "epoch": 3.3122096264634826,
      "grad_norm": 1.0132442712783813,
      "learning_rate": 1.359535894599604e-05,
      "loss": 0.983,
      "step": 17920
    },
    {
      "epoch": 3.312395465526854,
      "grad_norm": 0.9011918902397156,
      "learning_rate": 1.3588005348309517e-05,
      "loss": 0.7316,
      "step": 17921
    },
    {
      "epoch": 3.312581304590225,
      "grad_norm": 1.044669508934021,
      "learning_rate": 1.3580653594938752e-05,
      "loss": 0.8474,
      "step": 17922
    },
    {
      "epoch": 3.312767143653596,
      "grad_norm": 0.9390353560447693,
      "learning_rate": 1.3573303686040628e-05,
      "loss": 1.1222,
      "step": 17923
    },
    {
      "epoch": 3.312952982716967,
      "grad_norm": 0.7880856990814209,
      "learning_rate": 1.356595562177203e-05,
      "loss": 0.5984,
      "step": 17924
    },
    {
      "epoch": 3.3131388217803384,
      "grad_norm": 1.1461987495422363,
      "learning_rate": 1.3558609402289768e-05,
      "loss": 0.7476,
      "step": 17925
    },
    {
      "epoch": 3.3133246608437092,
      "grad_norm": 0.7984837889671326,
      "learning_rate": 1.3551265027750637e-05,
      "loss": 0.6918,
      "step": 17926
    },
    {
      "epoch": 3.3135104999070806,
      "grad_norm": 0.9764217734336853,
      "learning_rate": 1.3543922498311424e-05,
      "loss": 0.8359,
      "step": 17927
    },
    {
      "epoch": 3.3136963389704515,
      "grad_norm": 0.8988882303237915,
      "learning_rate": 1.3536581814128824e-05,
      "loss": 0.7925,
      "step": 17928
    },
    {
      "epoch": 3.313882178033823,
      "grad_norm": 0.8989973068237305,
      "learning_rate": 1.3529242975359514e-05,
      "loss": 0.7593,
      "step": 17929
    },
    {
      "epoch": 3.3140680170971937,
      "grad_norm": 0.8710231184959412,
      "learning_rate": 1.3521905982160088e-05,
      "loss": 0.7564,
      "step": 17930
    },
    {
      "epoch": 3.314253856160565,
      "grad_norm": 0.914422333240509,
      "learning_rate": 1.3514570834687202e-05,
      "loss": 0.7843,
      "step": 17931
    },
    {
      "epoch": 3.314439695223936,
      "grad_norm": 0.8551025390625,
      "learning_rate": 1.3507237533097395e-05,
      "loss": 0.8548,
      "step": 17932
    },
    {
      "epoch": 3.314625534287307,
      "grad_norm": 1.2345638275146484,
      "learning_rate": 1.3499906077547143e-05,
      "loss": 0.7326,
      "step": 17933
    },
    {
      "epoch": 3.314811373350678,
      "grad_norm": 0.9380419850349426,
      "learning_rate": 1.3492576468192985e-05,
      "loss": 1.0185,
      "step": 17934
    },
    {
      "epoch": 3.3149972124140494,
      "grad_norm": 0.9171674847602844,
      "learning_rate": 1.3485248705191311e-05,
      "loss": 0.7686,
      "step": 17935
    },
    {
      "epoch": 3.3151830514774208,
      "grad_norm": 0.7556156516075134,
      "learning_rate": 1.347792278869855e-05,
      "loss": 0.7326,
      "step": 17936
    },
    {
      "epoch": 3.3153688905407916,
      "grad_norm": 0.8587283492088318,
      "learning_rate": 1.3470598718871052e-05,
      "loss": 1.0934,
      "step": 17937
    },
    {
      "epoch": 3.315554729604163,
      "grad_norm": 0.9215811491012573,
      "learning_rate": 1.3463276495865118e-05,
      "loss": 0.5984,
      "step": 17938
    },
    {
      "epoch": 3.315740568667534,
      "grad_norm": 1.0269794464111328,
      "learning_rate": 1.3455956119837055e-05,
      "loss": 0.7353,
      "step": 17939
    },
    {
      "epoch": 3.315926407730905,
      "grad_norm": 0.8657709360122681,
      "learning_rate": 1.3448637590943103e-05,
      "loss": 0.9945,
      "step": 17940
    },
    {
      "epoch": 3.316112246794276,
      "grad_norm": 1.0170204639434814,
      "learning_rate": 1.3441320909339428e-05,
      "loss": 1.0004,
      "step": 17941
    },
    {
      "epoch": 3.3162980858576474,
      "grad_norm": 0.9796534776687622,
      "learning_rate": 1.3434006075182238e-05,
      "loss": 0.7462,
      "step": 17942
    },
    {
      "epoch": 3.3164839249210183,
      "grad_norm": 0.8594641089439392,
      "learning_rate": 1.342669308862764e-05,
      "loss": 0.6974,
      "step": 17943
    },
    {
      "epoch": 3.3166697639843896,
      "grad_norm": 1.0251960754394531,
      "learning_rate": 1.341938194983171e-05,
      "loss": 0.7658,
      "step": 17944
    },
    {
      "epoch": 3.3168556030477605,
      "grad_norm": 0.8630359172821045,
      "learning_rate": 1.341207265895047e-05,
      "loss": 0.9239,
      "step": 17945
    },
    {
      "epoch": 3.317041442111132,
      "grad_norm": 1.0533076524734497,
      "learning_rate": 1.3404765216139947e-05,
      "loss": 0.9937,
      "step": 17946
    },
    {
      "epoch": 3.3172272811745027,
      "grad_norm": 0.8733925223350525,
      "learning_rate": 1.339745962155613e-05,
      "loss": 0.9519,
      "step": 17947
    },
    {
      "epoch": 3.317413120237874,
      "grad_norm": 1.035334587097168,
      "learning_rate": 1.3390155875354927e-05,
      "loss": 0.6359,
      "step": 17948
    },
    {
      "epoch": 3.317598959301245,
      "grad_norm": 0.7760814428329468,
      "learning_rate": 1.3382853977692211e-05,
      "loss": 0.6602,
      "step": 17949
    },
    {
      "epoch": 3.3177847983646163,
      "grad_norm": 0.8502481579780579,
      "learning_rate": 1.3375553928723816e-05,
      "loss": 0.8501,
      "step": 17950
    },
    {
      "epoch": 3.3179706374279876,
      "grad_norm": 0.7824277877807617,
      "learning_rate": 1.3368255728605594e-05,
      "loss": 0.8327,
      "step": 17951
    },
    {
      "epoch": 3.3181564764913585,
      "grad_norm": 0.9095959663391113,
      "learning_rate": 1.3360959377493287e-05,
      "loss": 0.7192,
      "step": 17952
    },
    {
      "epoch": 3.3183423155547294,
      "grad_norm": 0.9373075366020203,
      "learning_rate": 1.3353664875542605e-05,
      "loss": 0.7329,
      "step": 17953
    },
    {
      "epoch": 3.3185281546181007,
      "grad_norm": 1.0542707443237305,
      "learning_rate": 1.3346372222909265e-05,
      "loss": 0.7624,
      "step": 17954
    },
    {
      "epoch": 3.318713993681472,
      "grad_norm": 0.8361062407493591,
      "learning_rate": 1.3339081419748922e-05,
      "loss": 0.8864,
      "step": 17955
    },
    {
      "epoch": 3.318899832744843,
      "grad_norm": 0.8348095417022705,
      "learning_rate": 1.333179246621713e-05,
      "loss": 0.8173,
      "step": 17956
    },
    {
      "epoch": 3.3190856718082142,
      "grad_norm": 0.9309622645378113,
      "learning_rate": 1.3324505362469542e-05,
      "loss": 0.7088,
      "step": 17957
    },
    {
      "epoch": 3.319271510871585,
      "grad_norm": 1.0715322494506836,
      "learning_rate": 1.331722010866161e-05,
      "loss": 0.7816,
      "step": 17958
    },
    {
      "epoch": 3.3194573499349564,
      "grad_norm": 0.9678760766983032,
      "learning_rate": 1.3309936704948877e-05,
      "loss": 0.8219,
      "step": 17959
    },
    {
      "epoch": 3.3196431889983273,
      "grad_norm": 1.3653862476348877,
      "learning_rate": 1.3302655151486798e-05,
      "loss": 0.8307,
      "step": 17960
    },
    {
      "epoch": 3.3198290280616987,
      "grad_norm": 1.0239980220794678,
      "learning_rate": 1.3295375448430724e-05,
      "loss": 0.8326,
      "step": 17961
    },
    {
      "epoch": 3.3200148671250695,
      "grad_norm": 0.8559108376502991,
      "learning_rate": 1.3288097595936111e-05,
      "loss": 0.6935,
      "step": 17962
    },
    {
      "epoch": 3.320200706188441,
      "grad_norm": 0.8458294868469238,
      "learning_rate": 1.3280821594158243e-05,
      "loss": 0.8023,
      "step": 17963
    },
    {
      "epoch": 3.3203865452518118,
      "grad_norm": 0.9620524644851685,
      "learning_rate": 1.327354744325242e-05,
      "loss": 0.8157,
      "step": 17964
    },
    {
      "epoch": 3.320572384315183,
      "grad_norm": 0.8043466806411743,
      "learning_rate": 1.3266275143373874e-05,
      "loss": 0.6708,
      "step": 17965
    },
    {
      "epoch": 3.3207582233785544,
      "grad_norm": 0.9349076151847839,
      "learning_rate": 1.3259004694677869e-05,
      "loss": 0.9326,
      "step": 17966
    },
    {
      "epoch": 3.3209440624419253,
      "grad_norm": 1.0523160696029663,
      "learning_rate": 1.3251736097319534e-05,
      "loss": 0.7398,
      "step": 17967
    },
    {
      "epoch": 3.321129901505296,
      "grad_norm": 1.1118494272232056,
      "learning_rate": 1.3244469351454048e-05,
      "loss": 0.809,
      "step": 17968
    },
    {
      "epoch": 3.3213157405686675,
      "grad_norm": 0.8730397820472717,
      "learning_rate": 1.3237204457236484e-05,
      "loss": 0.6698,
      "step": 17969
    },
    {
      "epoch": 3.321501579632039,
      "grad_norm": 1.2297297716140747,
      "learning_rate": 1.3229941414821878e-05,
      "loss": 0.8201,
      "step": 17970
    },
    {
      "epoch": 3.3216874186954097,
      "grad_norm": 1.0545740127563477,
      "learning_rate": 1.3222680224365292e-05,
      "loss": 0.8432,
      "step": 17971
    },
    {
      "epoch": 3.321873257758781,
      "grad_norm": 0.8901224136352539,
      "learning_rate": 1.321542088602168e-05,
      "loss": 0.8768,
      "step": 17972
    },
    {
      "epoch": 3.322059096822152,
      "grad_norm": 0.859428882598877,
      "learning_rate": 1.3208163399945961e-05,
      "loss": 0.6758,
      "step": 17973
    },
    {
      "epoch": 3.3222449358855233,
      "grad_norm": 1.1241716146469116,
      "learning_rate": 1.320090776629308e-05,
      "loss": 0.8622,
      "step": 17974
    },
    {
      "epoch": 3.322430774948894,
      "grad_norm": 0.9695386290550232,
      "learning_rate": 1.3193653985217858e-05,
      "loss": 0.7271,
      "step": 17975
    },
    {
      "epoch": 3.3226166140122655,
      "grad_norm": 0.9608593583106995,
      "learning_rate": 1.3186402056875135e-05,
      "loss": 0.6767,
      "step": 17976
    },
    {
      "epoch": 3.3228024530756364,
      "grad_norm": 0.9435471892356873,
      "learning_rate": 1.3179151981419658e-05,
      "loss": 0.7687,
      "step": 17977
    },
    {
      "epoch": 3.3229882921390077,
      "grad_norm": 1.1941502094268799,
      "learning_rate": 1.3171903759006188e-05,
      "loss": 1.0315,
      "step": 17978
    },
    {
      "epoch": 3.3231741312023786,
      "grad_norm": 1.0207300186157227,
      "learning_rate": 1.3164657389789458e-05,
      "loss": 0.6687,
      "step": 17979
    },
    {
      "epoch": 3.32335997026575,
      "grad_norm": 0.8957182765007019,
      "learning_rate": 1.3157412873924092e-05,
      "loss": 0.8879,
      "step": 17980
    },
    {
      "epoch": 3.323545809329121,
      "grad_norm": 0.877617597579956,
      "learning_rate": 1.3150170211564705e-05,
      "loss": 0.728,
      "step": 17981
    },
    {
      "epoch": 3.323731648392492,
      "grad_norm": 0.9174977540969849,
      "learning_rate": 1.314292940286591e-05,
      "loss": 0.672,
      "step": 17982
    },
    {
      "epoch": 3.323917487455863,
      "grad_norm": 1.834579348564148,
      "learning_rate": 1.313569044798224e-05,
      "loss": 1.1762,
      "step": 17983
    },
    {
      "epoch": 3.3241033265192343,
      "grad_norm": 1.1315218210220337,
      "learning_rate": 1.3128453347068192e-05,
      "loss": 0.8405,
      "step": 17984
    },
    {
      "epoch": 3.3242891655826057,
      "grad_norm": 1.1871623992919922,
      "learning_rate": 1.31212181002782e-05,
      "loss": 0.7738,
      "step": 17985
    },
    {
      "epoch": 3.3244750046459766,
      "grad_norm": 0.9807666540145874,
      "learning_rate": 1.3113984707766746e-05,
      "loss": 0.7332,
      "step": 17986
    },
    {
      "epoch": 3.324660843709348,
      "grad_norm": 0.9335218667984009,
      "learning_rate": 1.310675316968819e-05,
      "loss": 0.8143,
      "step": 17987
    },
    {
      "epoch": 3.3248466827727188,
      "grad_norm": 1.12467622756958,
      "learning_rate": 1.3099523486196853e-05,
      "loss": 0.8869,
      "step": 17988
    },
    {
      "epoch": 3.32503252183609,
      "grad_norm": 0.9656354188919067,
      "learning_rate": 1.3092295657447073e-05,
      "loss": 0.8738,
      "step": 17989
    },
    {
      "epoch": 3.325218360899461,
      "grad_norm": 1.1582108736038208,
      "learning_rate": 1.3085069683593076e-05,
      "loss": 0.8042,
      "step": 17990
    },
    {
      "epoch": 3.3254041999628323,
      "grad_norm": 0.8076069951057434,
      "learning_rate": 1.3077845564789148e-05,
      "loss": 0.6007,
      "step": 17991
    },
    {
      "epoch": 3.325590039026203,
      "grad_norm": 0.9304507374763489,
      "learning_rate": 1.3070623301189433e-05,
      "loss": 0.8243,
      "step": 17992
    },
    {
      "epoch": 3.3257758780895745,
      "grad_norm": 0.9668257832527161,
      "learning_rate": 1.3063402892948074e-05,
      "loss": 0.7761,
      "step": 17993
    },
    {
      "epoch": 3.3259617171529454,
      "grad_norm": 0.9958893060684204,
      "learning_rate": 1.3056184340219213e-05,
      "loss": 0.9935,
      "step": 17994
    },
    {
      "epoch": 3.3261475562163167,
      "grad_norm": 0.8639600872993469,
      "learning_rate": 1.3048967643156884e-05,
      "loss": 0.8225,
      "step": 17995
    },
    {
      "epoch": 3.3263333952796876,
      "grad_norm": 0.9271656274795532,
      "learning_rate": 1.304175280191514e-05,
      "loss": 0.8033,
      "step": 17996
    },
    {
      "epoch": 3.326519234343059,
      "grad_norm": 0.9010143280029297,
      "learning_rate": 1.3034539816647928e-05,
      "loss": 0.8416,
      "step": 17997
    },
    {
      "epoch": 3.32670507340643,
      "grad_norm": 1.003106713294983,
      "learning_rate": 1.3027328687509221e-05,
      "loss": 0.7771,
      "step": 17998
    },
    {
      "epoch": 3.326890912469801,
      "grad_norm": 0.7881788015365601,
      "learning_rate": 1.3020119414652965e-05,
      "loss": 0.8576,
      "step": 17999
    },
    {
      "epoch": 3.3270767515331725,
      "grad_norm": 1.155503511428833,
      "learning_rate": 1.3012911998232991e-05,
      "loss": 0.807,
      "step": 18000
    },
    {
      "epoch": 3.3272625905965434,
      "grad_norm": 0.9561105966567993,
      "learning_rate": 1.300570643840311e-05,
      "loss": 0.8275,
      "step": 18001
    },
    {
      "epoch": 3.3274484296599143,
      "grad_norm": 0.8585792183876038,
      "learning_rate": 1.2998502735317153e-05,
      "loss": 0.7732,
      "step": 18002
    },
    {
      "epoch": 3.3276342687232856,
      "grad_norm": 0.8768472075462341,
      "learning_rate": 1.2991300889128866e-05,
      "loss": 0.7609,
      "step": 18003
    },
    {
      "epoch": 3.327820107786657,
      "grad_norm": 0.9516651630401611,
      "learning_rate": 1.2984100899991947e-05,
      "loss": 0.7306,
      "step": 18004
    },
    {
      "epoch": 3.328005946850028,
      "grad_norm": 0.9200682044029236,
      "learning_rate": 1.297690276806004e-05,
      "loss": 0.8011,
      "step": 18005
    },
    {
      "epoch": 3.328191785913399,
      "grad_norm": 1.0520631074905396,
      "learning_rate": 1.2969706493486833e-05,
      "loss": 0.8373,
      "step": 18006
    },
    {
      "epoch": 3.32837762497677,
      "grad_norm": 0.8613325357437134,
      "learning_rate": 1.2962512076425892e-05,
      "loss": 0.6305,
      "step": 18007
    },
    {
      "epoch": 3.3285634640401414,
      "grad_norm": 0.7984744310379028,
      "learning_rate": 1.2955319517030762e-05,
      "loss": 0.605,
      "step": 18008
    },
    {
      "epoch": 3.3287493031035122,
      "grad_norm": 0.8289118409156799,
      "learning_rate": 1.2948128815454951e-05,
      "loss": 0.9193,
      "step": 18009
    },
    {
      "epoch": 3.3289351421668836,
      "grad_norm": 0.7974891662597656,
      "learning_rate": 1.294093997185194e-05,
      "loss": 0.5449,
      "step": 18010
    },
    {
      "epoch": 3.3291209812302545,
      "grad_norm": 0.9495866894721985,
      "learning_rate": 1.293375298637518e-05,
      "loss": 0.8973,
      "step": 18011
    },
    {
      "epoch": 3.329306820293626,
      "grad_norm": 0.9522069096565247,
      "learning_rate": 1.2926567859178052e-05,
      "loss": 0.7878,
      "step": 18012
    },
    {
      "epoch": 3.3294926593569967,
      "grad_norm": 1.059888243675232,
      "learning_rate": 1.2919384590413897e-05,
      "loss": 0.7148,
      "step": 18013
    },
    {
      "epoch": 3.329678498420368,
      "grad_norm": 0.8780260682106018,
      "learning_rate": 1.291220318023606e-05,
      "loss": 0.8033,
      "step": 18014
    },
    {
      "epoch": 3.329864337483739,
      "grad_norm": 1.0775411128997803,
      "learning_rate": 1.2905023628797797e-05,
      "loss": 0.8126,
      "step": 18015
    },
    {
      "epoch": 3.33005017654711,
      "grad_norm": 0.9820963740348816,
      "learning_rate": 1.2897845936252339e-05,
      "loss": 0.927,
      "step": 18016
    },
    {
      "epoch": 3.330236015610481,
      "grad_norm": 0.8544705510139465,
      "learning_rate": 1.2890670102752867e-05,
      "loss": 0.7791,
      "step": 18017
    },
    {
      "epoch": 3.3304218546738524,
      "grad_norm": 1.0104566812515259,
      "learning_rate": 1.2883496128452577e-05,
      "loss": 0.83,
      "step": 18018
    },
    {
      "epoch": 3.3306076937372238,
      "grad_norm": 0.9168938398361206,
      "learning_rate": 1.2876324013504547e-05,
      "loss": 0.8342,
      "step": 18019
    },
    {
      "epoch": 3.3307935328005946,
      "grad_norm": 1.260501503944397,
      "learning_rate": 1.286915375806188e-05,
      "loss": 0.7581,
      "step": 18020
    },
    {
      "epoch": 3.330979371863966,
      "grad_norm": 1.1076463460922241,
      "learning_rate": 1.2861985362277595e-05,
      "loss": 0.8086,
      "step": 18021
    },
    {
      "epoch": 3.331165210927337,
      "grad_norm": 1.2231957912445068,
      "learning_rate": 1.2854818826304715e-05,
      "loss": 0.778,
      "step": 18022
    },
    {
      "epoch": 3.331351049990708,
      "grad_norm": 0.8688556551933289,
      "learning_rate": 1.2847654150296184e-05,
      "loss": 0.8561,
      "step": 18023
    },
    {
      "epoch": 3.331536889054079,
      "grad_norm": 1.01069974899292,
      "learning_rate": 1.2840491334404913e-05,
      "loss": 1.0374,
      "step": 18024
    },
    {
      "epoch": 3.3317227281174504,
      "grad_norm": 0.9949778318405151,
      "learning_rate": 1.2833330378783759e-05,
      "loss": 0.9732,
      "step": 18025
    },
    {
      "epoch": 3.3319085671808213,
      "grad_norm": 0.9987459778785706,
      "learning_rate": 1.282617128358561e-05,
      "loss": 0.84,
      "step": 18026
    },
    {
      "epoch": 3.3320944062441926,
      "grad_norm": 0.8804039359092712,
      "learning_rate": 1.281901404896323e-05,
      "loss": 0.8324,
      "step": 18027
    },
    {
      "epoch": 3.3322802453075635,
      "grad_norm": 0.6644346714019775,
      "learning_rate": 1.2811858675069387e-05,
      "loss": 0.7908,
      "step": 18028
    },
    {
      "epoch": 3.332466084370935,
      "grad_norm": 0.868604302406311,
      "learning_rate": 1.2804705162056785e-05,
      "loss": 0.6634,
      "step": 18029
    },
    {
      "epoch": 3.3326519234343057,
      "grad_norm": 0.9592475891113281,
      "learning_rate": 1.279755351007812e-05,
      "loss": 0.757,
      "step": 18030
    },
    {
      "epoch": 3.332837762497677,
      "grad_norm": 0.9952224493026733,
      "learning_rate": 1.2790403719286049e-05,
      "loss": 1.0982,
      "step": 18031
    },
    {
      "epoch": 3.333023601561048,
      "grad_norm": 0.8424680829048157,
      "learning_rate": 1.2783255789833148e-05,
      "loss": 0.6902,
      "step": 18032
    },
    {
      "epoch": 3.3332094406244193,
      "grad_norm": 0.9481412172317505,
      "learning_rate": 1.2776109721871965e-05,
      "loss": 0.8379,
      "step": 18033
    },
    {
      "epoch": 3.3333952796877906,
      "grad_norm": 0.9572385549545288,
      "learning_rate": 1.2768965515555065e-05,
      "loss": 1.0206,
      "step": 18034
    },
    {
      "epoch": 3.3335811187511615,
      "grad_norm": 0.8761714696884155,
      "learning_rate": 1.2761823171034892e-05,
      "loss": 0.9961,
      "step": 18035
    },
    {
      "epoch": 3.333766957814533,
      "grad_norm": 0.9750075340270996,
      "learning_rate": 1.2754682688463903e-05,
      "loss": 0.8697,
      "step": 18036
    },
    {
      "epoch": 3.3339527968779037,
      "grad_norm": 0.8760260343551636,
      "learning_rate": 1.2747544067994465e-05,
      "loss": 0.8723,
      "step": 18037
    },
    {
      "epoch": 3.334138635941275,
      "grad_norm": 0.9728489518165588,
      "learning_rate": 1.274040730977899e-05,
      "loss": 0.8646,
      "step": 18038
    },
    {
      "epoch": 3.334324475004646,
      "grad_norm": 0.8067655563354492,
      "learning_rate": 1.2733272413969776e-05,
      "loss": 0.6259,
      "step": 18039
    },
    {
      "epoch": 3.3345103140680172,
      "grad_norm": 0.9222126007080078,
      "learning_rate": 1.2726139380719094e-05,
      "loss": 0.8291,
      "step": 18040
    },
    {
      "epoch": 3.334696153131388,
      "grad_norm": 1.1046346426010132,
      "learning_rate": 1.2719008210179195e-05,
      "loss": 0.777,
      "step": 18041
    },
    {
      "epoch": 3.3348819921947594,
      "grad_norm": 0.820184588432312,
      "learning_rate": 1.2711878902502305e-05,
      "loss": 0.7441,
      "step": 18042
    },
    {
      "epoch": 3.3350678312581303,
      "grad_norm": 0.8473408222198486,
      "learning_rate": 1.270475145784057e-05,
      "loss": 0.7787,
      "step": 18043
    },
    {
      "epoch": 3.3352536703215017,
      "grad_norm": 0.6779477596282959,
      "learning_rate": 1.2697625876346108e-05,
      "loss": 0.5674,
      "step": 18044
    },
    {
      "epoch": 3.3354395093848725,
      "grad_norm": 1.0007576942443848,
      "learning_rate": 1.269050215817099e-05,
      "loss": 0.9149,
      "step": 18045
    },
    {
      "epoch": 3.335625348448244,
      "grad_norm": 0.9065996408462524,
      "learning_rate": 1.2683380303467285e-05,
      "loss": 0.8983,
      "step": 18046
    },
    {
      "epoch": 3.3358111875116148,
      "grad_norm": 0.835256040096283,
      "learning_rate": 1.2676260312386989e-05,
      "loss": 1.0209,
      "step": 18047
    },
    {
      "epoch": 3.335997026574986,
      "grad_norm": 0.9279325008392334,
      "learning_rate": 1.2669142185082073e-05,
      "loss": 0.9567,
      "step": 18048
    },
    {
      "epoch": 3.3361828656383574,
      "grad_norm": 1.4079657793045044,
      "learning_rate": 1.2662025921704423e-05,
      "loss": 0.8364,
      "step": 18049
    },
    {
      "epoch": 3.3363687047017283,
      "grad_norm": 0.8377646803855896,
      "learning_rate": 1.2654911522405966e-05,
      "loss": 0.7861,
      "step": 18050
    },
    {
      "epoch": 3.336554543765099,
      "grad_norm": 0.9112058281898499,
      "learning_rate": 1.2647798987338521e-05,
      "loss": 0.9795,
      "step": 18051
    },
    {
      "epoch": 3.3367403828284705,
      "grad_norm": 0.88719242811203,
      "learning_rate": 1.2640688316653926e-05,
      "loss": 0.7678,
      "step": 18052
    },
    {
      "epoch": 3.336926221891842,
      "grad_norm": 0.9839407801628113,
      "learning_rate": 1.26335795105039e-05,
      "loss": 0.8179,
      "step": 18053
    },
    {
      "epoch": 3.3371120609552127,
      "grad_norm": 0.7015066742897034,
      "learning_rate": 1.2626472569040204e-05,
      "loss": 0.6559,
      "step": 18054
    },
    {
      "epoch": 3.337297900018584,
      "grad_norm": 0.9736129641532898,
      "learning_rate": 1.2619367492414524e-05,
      "loss": 0.6797,
      "step": 18055
    },
    {
      "epoch": 3.337483739081955,
      "grad_norm": 1.0731183290481567,
      "learning_rate": 1.2612264280778496e-05,
      "loss": 0.8349,
      "step": 18056
    },
    {
      "epoch": 3.3376695781453263,
      "grad_norm": 1.0453765392303467,
      "learning_rate": 1.2605162934283699e-05,
      "loss": 0.7079,
      "step": 18057
    },
    {
      "epoch": 3.337855417208697,
      "grad_norm": 1.546590805053711,
      "learning_rate": 1.2598063453081743e-05,
      "loss": 1.1752,
      "step": 18058
    },
    {
      "epoch": 3.3380412562720685,
      "grad_norm": 0.7495214343070984,
      "learning_rate": 1.2590965837324131e-05,
      "loss": 0.5539,
      "step": 18059
    },
    {
      "epoch": 3.3382270953354394,
      "grad_norm": 0.9308766722679138,
      "learning_rate": 1.2583870087162353e-05,
      "loss": 0.9218,
      "step": 18060
    },
    {
      "epoch": 3.3384129343988107,
      "grad_norm": 1.0387533903121948,
      "learning_rate": 1.257677620274782e-05,
      "loss": 0.9143,
      "step": 18061
    },
    {
      "epoch": 3.3385987734621816,
      "grad_norm": 0.8280618190765381,
      "learning_rate": 1.2569684184232011e-05,
      "loss": 0.8469,
      "step": 18062
    },
    {
      "epoch": 3.338784612525553,
      "grad_norm": 0.9430438280105591,
      "learning_rate": 1.2562594031766262e-05,
      "loss": 0.8007,
      "step": 18063
    },
    {
      "epoch": 3.338970451588924,
      "grad_norm": 0.8449985980987549,
      "learning_rate": 1.2555505745501894e-05,
      "loss": 0.6529,
      "step": 18064
    },
    {
      "epoch": 3.339156290652295,
      "grad_norm": 0.9311802983283997,
      "learning_rate": 1.2548419325590177e-05,
      "loss": 0.8376,
      "step": 18065
    },
    {
      "epoch": 3.339342129715666,
      "grad_norm": 0.9691295623779297,
      "learning_rate": 1.254133477218239e-05,
      "loss": 0.9887,
      "step": 18066
    },
    {
      "epoch": 3.3395279687790373,
      "grad_norm": 0.9618540406227112,
      "learning_rate": 1.2534252085429732e-05,
      "loss": 0.9186,
      "step": 18067
    },
    {
      "epoch": 3.3397138078424087,
      "grad_norm": 0.947231650352478,
      "learning_rate": 1.2527171265483361e-05,
      "loss": 0.9473,
      "step": 18068
    },
    {
      "epoch": 3.3398996469057796,
      "grad_norm": 0.9473240971565247,
      "learning_rate": 1.2520092312494402e-05,
      "loss": 1.0109,
      "step": 18069
    },
    {
      "epoch": 3.340085485969151,
      "grad_norm": 0.8536008596420288,
      "learning_rate": 1.2513015226613956e-05,
      "loss": 0.8339,
      "step": 18070
    },
    {
      "epoch": 3.3402713250325218,
      "grad_norm": 0.732178270816803,
      "learning_rate": 1.2505940007993078e-05,
      "loss": 0.695,
      "step": 18071
    },
    {
      "epoch": 3.340457164095893,
      "grad_norm": 0.906110405921936,
      "learning_rate": 1.2498866656782737e-05,
      "loss": 0.9163,
      "step": 18072
    },
    {
      "epoch": 3.340643003159264,
      "grad_norm": 0.906360387802124,
      "learning_rate": 1.2491795173133935e-05,
      "loss": 0.6139,
      "step": 18073
    },
    {
      "epoch": 3.3408288422226353,
      "grad_norm": 0.890641987323761,
      "learning_rate": 1.2484725557197608e-05,
      "loss": 0.7674,
      "step": 18074
    },
    {
      "epoch": 3.341014681286006,
      "grad_norm": 1.048957347869873,
      "learning_rate": 1.2477657809124631e-05,
      "loss": 0.7132,
      "step": 18075
    },
    {
      "epoch": 3.3412005203493775,
      "grad_norm": 1.1851656436920166,
      "learning_rate": 1.2470591929065866e-05,
      "loss": 0.8471,
      "step": 18076
    },
    {
      "epoch": 3.3413863594127484,
      "grad_norm": 1.098007082939148,
      "learning_rate": 1.246352791717208e-05,
      "loss": 1.1545,
      "step": 18077
    },
    {
      "epoch": 3.3415721984761197,
      "grad_norm": 0.9009913802146912,
      "learning_rate": 1.2456465773594094e-05,
      "loss": 0.6493,
      "step": 18078
    },
    {
      "epoch": 3.3417580375394906,
      "grad_norm": 0.9068120718002319,
      "learning_rate": 1.2449405498482613e-05,
      "loss": 0.6948,
      "step": 18079
    },
    {
      "epoch": 3.341943876602862,
      "grad_norm": 0.9160207509994507,
      "learning_rate": 1.2442347091988338e-05,
      "loss": 0.8363,
      "step": 18080
    },
    {
      "epoch": 3.342129715666233,
      "grad_norm": 0.9870322942733765,
      "learning_rate": 1.2435290554261881e-05,
      "loss": 0.8309,
      "step": 18081
    },
    {
      "epoch": 3.342315554729604,
      "grad_norm": 0.8470276594161987,
      "learning_rate": 1.2428235885453875e-05,
      "loss": 0.7559,
      "step": 18082
    },
    {
      "epoch": 3.3425013937929755,
      "grad_norm": 0.9534688591957092,
      "learning_rate": 1.2421183085714927e-05,
      "loss": 0.906,
      "step": 18083
    },
    {
      "epoch": 3.3426872328563464,
      "grad_norm": 0.8364575505256653,
      "learning_rate": 1.2414132155195524e-05,
      "loss": 0.9157,
      "step": 18084
    },
    {
      "epoch": 3.3428730719197173,
      "grad_norm": 0.9537498354911804,
      "learning_rate": 1.2407083094046157e-05,
      "loss": 0.7473,
      "step": 18085
    },
    {
      "epoch": 3.3430589109830886,
      "grad_norm": 1.1054604053497314,
      "learning_rate": 1.2400035902417295e-05,
      "loss": 1.2257,
      "step": 18086
    },
    {
      "epoch": 3.34324475004646,
      "grad_norm": 0.9021813273429871,
      "learning_rate": 1.2392990580459352e-05,
      "loss": 0.9291,
      "step": 18087
    },
    {
      "epoch": 3.343430589109831,
      "grad_norm": 1.1228984594345093,
      "learning_rate": 1.2385947128322672e-05,
      "loss": 0.9025,
      "step": 18088
    },
    {
      "epoch": 3.343616428173202,
      "grad_norm": 0.9197649359703064,
      "learning_rate": 1.2378905546157583e-05,
      "loss": 0.698,
      "step": 18089
    },
    {
      "epoch": 3.343802267236573,
      "grad_norm": 0.9107482433319092,
      "learning_rate": 1.2371865834114416e-05,
      "loss": 0.8834,
      "step": 18090
    },
    {
      "epoch": 3.3439881062999444,
      "grad_norm": 0.8883651494979858,
      "learning_rate": 1.2364827992343397e-05,
      "loss": 0.8465,
      "step": 18091
    },
    {
      "epoch": 3.3441739453633152,
      "grad_norm": 1.1927982568740845,
      "learning_rate": 1.2357792020994718e-05,
      "loss": 0.6891,
      "step": 18092
    },
    {
      "epoch": 3.3443597844266866,
      "grad_norm": 0.6591570973396301,
      "learning_rate": 1.2350757920218558e-05,
      "loss": 0.4652,
      "step": 18093
    },
    {
      "epoch": 3.3445456234900575,
      "grad_norm": 0.9535413384437561,
      "learning_rate": 1.2343725690165087e-05,
      "loss": 0.8278,
      "step": 18094
    },
    {
      "epoch": 3.344731462553429,
      "grad_norm": 0.8922446966171265,
      "learning_rate": 1.233669533098436e-05,
      "loss": 0.8075,
      "step": 18095
    },
    {
      "epoch": 3.3449173016167997,
      "grad_norm": 0.9422911405563354,
      "learning_rate": 1.2329666842826438e-05,
      "loss": 1.0,
      "step": 18096
    },
    {
      "epoch": 3.345103140680171,
      "grad_norm": 0.8566906452178955,
      "learning_rate": 1.2322640225841309e-05,
      "loss": 0.6979,
      "step": 18097
    },
    {
      "epoch": 3.3452889797435423,
      "grad_norm": 1.1618760824203491,
      "learning_rate": 1.2315615480178989e-05,
      "loss": 0.804,
      "step": 18098
    },
    {
      "epoch": 3.345474818806913,
      "grad_norm": 1.0303587913513184,
      "learning_rate": 1.2308592605989377e-05,
      "loss": 0.8608,
      "step": 18099
    },
    {
      "epoch": 3.345660657870284,
      "grad_norm": 1.1015151739120483,
      "learning_rate": 1.2301571603422379e-05,
      "loss": 0.8997,
      "step": 18100
    },
    {
      "epoch": 3.3458464969336554,
      "grad_norm": 0.9106659889221191,
      "learning_rate": 1.2294552472627808e-05,
      "loss": 0.7916,
      "step": 18101
    },
    {
      "epoch": 3.3460323359970268,
      "grad_norm": 1.1678972244262695,
      "learning_rate": 1.2287535213755542e-05,
      "loss": 0.9502,
      "step": 18102
    },
    {
      "epoch": 3.3462181750603976,
      "grad_norm": 0.9588977694511414,
      "learning_rate": 1.2280519826955284e-05,
      "loss": 0.7856,
      "step": 18103
    },
    {
      "epoch": 3.346404014123769,
      "grad_norm": 0.8328583836555481,
      "learning_rate": 1.2273506312376815e-05,
      "loss": 0.6244,
      "step": 18104
    },
    {
      "epoch": 3.34658985318714,
      "grad_norm": 0.9519017338752747,
      "learning_rate": 1.2266494670169792e-05,
      "loss": 0.8703,
      "step": 18105
    },
    {
      "epoch": 3.346775692250511,
      "grad_norm": 0.9912613034248352,
      "learning_rate": 1.2259484900483909e-05,
      "loss": 0.9899,
      "step": 18106
    },
    {
      "epoch": 3.346961531313882,
      "grad_norm": 0.9581605195999146,
      "learning_rate": 1.2252477003468743e-05,
      "loss": 0.8696,
      "step": 18107
    },
    {
      "epoch": 3.3471473703772534,
      "grad_norm": 0.8546910285949707,
      "learning_rate": 1.2245470979273887e-05,
      "loss": 0.6573,
      "step": 18108
    },
    {
      "epoch": 3.3473332094406243,
      "grad_norm": 0.848886251449585,
      "learning_rate": 1.2238466828048833e-05,
      "loss": 0.7496,
      "step": 18109
    },
    {
      "epoch": 3.3475190485039956,
      "grad_norm": 0.9090949892997742,
      "learning_rate": 1.2231464549943117e-05,
      "loss": 0.9007,
      "step": 18110
    },
    {
      "epoch": 3.3477048875673665,
      "grad_norm": 1.0155179500579834,
      "learning_rate": 1.2224464145106174e-05,
      "loss": 0.7334,
      "step": 18111
    },
    {
      "epoch": 3.347890726630738,
      "grad_norm": 1.1983330249786377,
      "learning_rate": 1.2217465613687395e-05,
      "loss": 1.0167,
      "step": 18112
    },
    {
      "epoch": 3.3480765656941087,
      "grad_norm": 0.9881544709205627,
      "learning_rate": 1.2210468955836196e-05,
      "loss": 0.805,
      "step": 18113
    },
    {
      "epoch": 3.34826240475748,
      "grad_norm": 0.9015958309173584,
      "learning_rate": 1.2203474171701867e-05,
      "loss": 0.8233,
      "step": 18114
    },
    {
      "epoch": 3.348448243820851,
      "grad_norm": 0.9489085078239441,
      "learning_rate": 1.2196481261433735e-05,
      "loss": 0.9338,
      "step": 18115
    },
    {
      "epoch": 3.3486340828842223,
      "grad_norm": 0.889735758304596,
      "learning_rate": 1.2189490225181022e-05,
      "loss": 0.8285,
      "step": 18116
    },
    {
      "epoch": 3.3488199219475936,
      "grad_norm": 0.9477465748786926,
      "learning_rate": 1.2182501063092943e-05,
      "loss": 0.9435,
      "step": 18117
    },
    {
      "epoch": 3.3490057610109645,
      "grad_norm": 0.956580400466919,
      "learning_rate": 1.2175513775318692e-05,
      "loss": 1.0321,
      "step": 18118
    },
    {
      "epoch": 3.349191600074336,
      "grad_norm": 0.8784446120262146,
      "learning_rate": 1.2168528362007393e-05,
      "loss": 0.8943,
      "step": 18119
    },
    {
      "epoch": 3.3493774391377067,
      "grad_norm": 1.0098607540130615,
      "learning_rate": 1.2161544823308124e-05,
      "loss": 0.8295,
      "step": 18120
    },
    {
      "epoch": 3.349563278201078,
      "grad_norm": 0.890578031539917,
      "learning_rate": 1.2154563159369924e-05,
      "loss": 0.6808,
      "step": 18121
    },
    {
      "epoch": 3.349749117264449,
      "grad_norm": 1.0231043100357056,
      "learning_rate": 1.2147583370341853e-05,
      "loss": 0.7606,
      "step": 18122
    },
    {
      "epoch": 3.3499349563278202,
      "grad_norm": 0.873323380947113,
      "learning_rate": 1.2140605456372855e-05,
      "loss": 0.8282,
      "step": 18123
    },
    {
      "epoch": 3.350120795391191,
      "grad_norm": 0.8876305222511292,
      "learning_rate": 1.2133629417611835e-05,
      "loss": 0.6846,
      "step": 18124
    },
    {
      "epoch": 3.3503066344545624,
      "grad_norm": 0.851346492767334,
      "learning_rate": 1.2126655254207709e-05,
      "loss": 0.922,
      "step": 18125
    },
    {
      "epoch": 3.3504924735179333,
      "grad_norm": 1.0348562002182007,
      "learning_rate": 1.2119682966309353e-05,
      "loss": 0.8517,
      "step": 18126
    },
    {
      "epoch": 3.3506783125813047,
      "grad_norm": 0.8237027525901794,
      "learning_rate": 1.2112712554065553e-05,
      "loss": 0.712,
      "step": 18127
    },
    {
      "epoch": 3.3508641516446755,
      "grad_norm": 0.8335827589035034,
      "learning_rate": 1.2105744017625075e-05,
      "loss": 0.6588,
      "step": 18128
    },
    {
      "epoch": 3.351049990708047,
      "grad_norm": 0.8928906321525574,
      "learning_rate": 1.2098777357136648e-05,
      "loss": 0.8155,
      "step": 18129
    },
    {
      "epoch": 3.3512358297714178,
      "grad_norm": 1.0455178022384644,
      "learning_rate": 1.2091812572748994e-05,
      "loss": 0.7419,
      "step": 18130
    },
    {
      "epoch": 3.351421668834789,
      "grad_norm": 1.686511754989624,
      "learning_rate": 1.2084849664610742e-05,
      "loss": 1.2921,
      "step": 18131
    },
    {
      "epoch": 3.3516075078981604,
      "grad_norm": 0.8682252764701843,
      "learning_rate": 1.2077888632870493e-05,
      "loss": 0.7061,
      "step": 18132
    },
    {
      "epoch": 3.3517933469615313,
      "grad_norm": 0.859376847743988,
      "learning_rate": 1.207092947767684e-05,
      "loss": 0.853,
      "step": 18133
    },
    {
      "epoch": 3.351979186024902,
      "grad_norm": 0.7160893082618713,
      "learning_rate": 1.2063972199178309e-05,
      "loss": 0.5596,
      "step": 18134
    },
    {
      "epoch": 3.3521650250882735,
      "grad_norm": 0.8916295170783997,
      "learning_rate": 1.2057016797523369e-05,
      "loss": 0.7075,
      "step": 18135
    },
    {
      "epoch": 3.352350864151645,
      "grad_norm": 0.7931239008903503,
      "learning_rate": 1.20500632728605e-05,
      "loss": 0.6814,
      "step": 18136
    },
    {
      "epoch": 3.3525367032150157,
      "grad_norm": 0.9768849015235901,
      "learning_rate": 1.20431116253381e-05,
      "loss": 0.638,
      "step": 18137
    },
    {
      "epoch": 3.352722542278387,
      "grad_norm": 1.1019235849380493,
      "learning_rate": 1.2036161855104544e-05,
      "loss": 0.8028,
      "step": 18138
    },
    {
      "epoch": 3.352908381341758,
      "grad_norm": 0.9690166711807251,
      "learning_rate": 1.2029213962308172e-05,
      "loss": 0.7347,
      "step": 18139
    },
    {
      "epoch": 3.3530942204051293,
      "grad_norm": 1.0368572473526,
      "learning_rate": 1.2022267947097266e-05,
      "loss": 1.0224,
      "step": 18140
    },
    {
      "epoch": 3.3532800594685,
      "grad_norm": 0.9114643931388855,
      "learning_rate": 1.2015323809620049e-05,
      "loss": 0.65,
      "step": 18141
    },
    {
      "epoch": 3.3534658985318715,
      "grad_norm": 0.9931135177612305,
      "learning_rate": 1.2008381550024773e-05,
      "loss": 0.6699,
      "step": 18142
    },
    {
      "epoch": 3.3536517375952424,
      "grad_norm": 0.930069625377655,
      "learning_rate": 1.2001441168459604e-05,
      "loss": 0.646,
      "step": 18143
    },
    {
      "epoch": 3.3538375766586137,
      "grad_norm": 0.9435836672782898,
      "learning_rate": 1.1994502665072637e-05,
      "loss": 0.7175,
      "step": 18144
    },
    {
      "epoch": 3.3540234157219846,
      "grad_norm": 0.7887563109397888,
      "learning_rate": 1.1987566040011989e-05,
      "loss": 0.6377,
      "step": 18145
    },
    {
      "epoch": 3.354209254785356,
      "grad_norm": 0.8834825158119202,
      "learning_rate": 1.1980631293425726e-05,
      "loss": 0.8163,
      "step": 18146
    },
    {
      "epoch": 3.3543950938487272,
      "grad_norm": 0.8089690208435059,
      "learning_rate": 1.1973698425461832e-05,
      "loss": 0.7909,
      "step": 18147
    },
    {
      "epoch": 3.354580932912098,
      "grad_norm": 0.8597369194030762,
      "learning_rate": 1.19667674362683e-05,
      "loss": 0.5808,
      "step": 18148
    },
    {
      "epoch": 3.354766771975469,
      "grad_norm": 1.174746036529541,
      "learning_rate": 1.1959838325993023e-05,
      "loss": 0.9286,
      "step": 18149
    },
    {
      "epoch": 3.3549526110388403,
      "grad_norm": 1.0257681608200073,
      "learning_rate": 1.1952911094783926e-05,
      "loss": 0.8775,
      "step": 18150
    },
    {
      "epoch": 3.3551384501022117,
      "grad_norm": 1.017487645149231,
      "learning_rate": 1.1945985742788856e-05,
      "loss": 0.775,
      "step": 18151
    },
    {
      "epoch": 3.3553242891655826,
      "grad_norm": 1.312432885169983,
      "learning_rate": 1.1939062270155588e-05,
      "loss": 0.8469,
      "step": 18152
    },
    {
      "epoch": 3.355510128228954,
      "grad_norm": 1.1309226751327515,
      "learning_rate": 1.1932140677031944e-05,
      "loss": 0.9789,
      "step": 18153
    },
    {
      "epoch": 3.3556959672923248,
      "grad_norm": 0.9536949396133423,
      "learning_rate": 1.192522096356562e-05,
      "loss": 0.9253,
      "step": 18154
    },
    {
      "epoch": 3.355881806355696,
      "grad_norm": 1.0114742517471313,
      "learning_rate": 1.1918303129904317e-05,
      "loss": 0.7059,
      "step": 18155
    },
    {
      "epoch": 3.356067645419067,
      "grad_norm": 0.9156493544578552,
      "learning_rate": 1.1911387176195654e-05,
      "loss": 0.7234,
      "step": 18156
    },
    {
      "epoch": 3.3562534844824383,
      "grad_norm": 0.9517843723297119,
      "learning_rate": 1.1904473102587266e-05,
      "loss": 0.7238,
      "step": 18157
    },
    {
      "epoch": 3.356439323545809,
      "grad_norm": 0.9302890300750732,
      "learning_rate": 1.1897560909226746e-05,
      "loss": 0.9865,
      "step": 18158
    },
    {
      "epoch": 3.3566251626091805,
      "grad_norm": 1.1713268756866455,
      "learning_rate": 1.1890650596261599e-05,
      "loss": 0.9072,
      "step": 18159
    },
    {
      "epoch": 3.3568110016725514,
      "grad_norm": 0.9015994071960449,
      "learning_rate": 1.1883742163839307e-05,
      "loss": 0.7619,
      "step": 18160
    },
    {
      "epoch": 3.3569968407359227,
      "grad_norm": 1.0113537311553955,
      "learning_rate": 1.1876835612107306e-05,
      "loss": 0.7482,
      "step": 18161
    },
    {
      "epoch": 3.3571826797992936,
      "grad_norm": 0.8868255615234375,
      "learning_rate": 1.1869930941213048e-05,
      "loss": 0.9041,
      "step": 18162
    },
    {
      "epoch": 3.357368518862665,
      "grad_norm": 0.7899235486984253,
      "learning_rate": 1.1863028151303879e-05,
      "loss": 0.8324,
      "step": 18163
    },
    {
      "epoch": 3.357554357926036,
      "grad_norm": 1.072371006011963,
      "learning_rate": 1.1856127242527094e-05,
      "loss": 0.9066,
      "step": 18164
    },
    {
      "epoch": 3.357740196989407,
      "grad_norm": 0.8462903499603271,
      "learning_rate": 1.1849228215030039e-05,
      "loss": 0.5531,
      "step": 18165
    },
    {
      "epoch": 3.3579260360527785,
      "grad_norm": 0.9083438515663147,
      "learning_rate": 1.1842331068959922e-05,
      "loss": 0.813,
      "step": 18166
    },
    {
      "epoch": 3.3581118751161494,
      "grad_norm": 0.9696852564811707,
      "learning_rate": 1.1835435804463967e-05,
      "loss": 0.7097,
      "step": 18167
    },
    {
      "epoch": 3.3582977141795207,
      "grad_norm": 0.8321973085403442,
      "learning_rate": 1.1828542421689348e-05,
      "loss": 0.5627,
      "step": 18168
    },
    {
      "epoch": 3.3584835532428916,
      "grad_norm": 1.0764424800872803,
      "learning_rate": 1.1821650920783166e-05,
      "loss": 0.8361,
      "step": 18169
    },
    {
      "epoch": 3.358669392306263,
      "grad_norm": 0.9155271649360657,
      "learning_rate": 1.1814761301892542e-05,
      "loss": 0.8987,
      "step": 18170
    },
    {
      "epoch": 3.358855231369634,
      "grad_norm": 0.9456757307052612,
      "learning_rate": 1.1807873565164506e-05,
      "loss": 0.7765,
      "step": 18171
    },
    {
      "epoch": 3.359041070433005,
      "grad_norm": 0.9458122849464417,
      "learning_rate": 1.1800987710746048e-05,
      "loss": 0.6491,
      "step": 18172
    },
    {
      "epoch": 3.359226909496376,
      "grad_norm": 0.9681454300880432,
      "learning_rate": 1.1794103738784169e-05,
      "loss": 0.8975,
      "step": 18173
    },
    {
      "epoch": 3.3594127485597474,
      "grad_norm": 1.0836167335510254,
      "learning_rate": 1.1787221649425772e-05,
      "loss": 0.9195,
      "step": 18174
    },
    {
      "epoch": 3.3595985876231182,
      "grad_norm": 0.8586798310279846,
      "learning_rate": 1.1780341442817754e-05,
      "loss": 0.7228,
      "step": 18175
    },
    {
      "epoch": 3.3597844266864896,
      "grad_norm": 0.9736157655715942,
      "learning_rate": 1.1773463119106943e-05,
      "loss": 0.7443,
      "step": 18176
    },
    {
      "epoch": 3.3599702657498605,
      "grad_norm": 0.9830484390258789,
      "learning_rate": 1.176658667844015e-05,
      "loss": 0.8453,
      "step": 18177
    },
    {
      "epoch": 3.360156104813232,
      "grad_norm": 0.8664001226425171,
      "learning_rate": 1.1759712120964184e-05,
      "loss": 0.6916,
      "step": 18178
    },
    {
      "epoch": 3.3603419438766027,
      "grad_norm": 0.768081784248352,
      "learning_rate": 1.1752839446825725e-05,
      "loss": 0.7665,
      "step": 18179
    },
    {
      "epoch": 3.360527782939974,
      "grad_norm": 0.8800269961357117,
      "learning_rate": 1.1745968656171469e-05,
      "loss": 0.5951,
      "step": 18180
    },
    {
      "epoch": 3.3607136220033453,
      "grad_norm": 0.9124451875686646,
      "learning_rate": 1.173909974914804e-05,
      "loss": 0.785,
      "step": 18181
    },
    {
      "epoch": 3.360899461066716,
      "grad_norm": 0.9192720055580139,
      "learning_rate": 1.1732232725902093e-05,
      "loss": 0.7034,
      "step": 18182
    },
    {
      "epoch": 3.361085300130087,
      "grad_norm": 1.033612608909607,
      "learning_rate": 1.1725367586580161e-05,
      "loss": 0.9732,
      "step": 18183
    },
    {
      "epoch": 3.3612711391934584,
      "grad_norm": 1.0083500146865845,
      "learning_rate": 1.171850433132875e-05,
      "loss": 0.7537,
      "step": 18184
    },
    {
      "epoch": 3.3614569782568298,
      "grad_norm": 0.8668590784072876,
      "learning_rate": 1.171164296029439e-05,
      "loss": 0.823,
      "step": 18185
    },
    {
      "epoch": 3.3616428173202006,
      "grad_norm": 0.9430564641952515,
      "learning_rate": 1.1704783473623504e-05,
      "loss": 0.8356,
      "step": 18186
    },
    {
      "epoch": 3.361828656383572,
      "grad_norm": 1.1226727962493896,
      "learning_rate": 1.1697925871462467e-05,
      "loss": 0.6784,
      "step": 18187
    },
    {
      "epoch": 3.362014495446943,
      "grad_norm": 1.08354651927948,
      "learning_rate": 1.1691070153957695e-05,
      "loss": 0.724,
      "step": 18188
    },
    {
      "epoch": 3.362200334510314,
      "grad_norm": 0.9508259892463684,
      "learning_rate": 1.168421632125547e-05,
      "loss": 0.8168,
      "step": 18189
    },
    {
      "epoch": 3.362386173573685,
      "grad_norm": 1.0835859775543213,
      "learning_rate": 1.1677364373502108e-05,
      "loss": 0.8825,
      "step": 18190
    },
    {
      "epoch": 3.3625720126370564,
      "grad_norm": 1.0884180068969727,
      "learning_rate": 1.1670514310843839e-05,
      "loss": 0.9111,
      "step": 18191
    },
    {
      "epoch": 3.3627578517004273,
      "grad_norm": 0.9678394794464111,
      "learning_rate": 1.1663666133426832e-05,
      "loss": 0.74,
      "step": 18192
    },
    {
      "epoch": 3.3629436907637986,
      "grad_norm": 0.9171245098114014,
      "learning_rate": 1.1656819841397315e-05,
      "loss": 0.6997,
      "step": 18193
    },
    {
      "epoch": 3.3631295298271695,
      "grad_norm": 0.9149906635284424,
      "learning_rate": 1.164997543490136e-05,
      "loss": 0.7581,
      "step": 18194
    },
    {
      "epoch": 3.363315368890541,
      "grad_norm": 1.517656922340393,
      "learning_rate": 1.1643132914085076e-05,
      "loss": 1.0754,
      "step": 18195
    },
    {
      "epoch": 3.3635012079539117,
      "grad_norm": 0.8923678994178772,
      "learning_rate": 1.1636292279094473e-05,
      "loss": 0.7914,
      "step": 18196
    },
    {
      "epoch": 3.363687047017283,
      "grad_norm": 0.9691092371940613,
      "learning_rate": 1.1629453530075584e-05,
      "loss": 1.1266,
      "step": 18197
    },
    {
      "epoch": 3.363872886080654,
      "grad_norm": 0.9492999911308289,
      "learning_rate": 1.1622616667174346e-05,
      "loss": 0.914,
      "step": 18198
    },
    {
      "epoch": 3.3640587251440253,
      "grad_norm": 0.9340925812721252,
      "learning_rate": 1.161578169053672e-05,
      "loss": 0.7787,
      "step": 18199
    },
    {
      "epoch": 3.3642445642073966,
      "grad_norm": 0.809636652469635,
      "learning_rate": 1.1608948600308545e-05,
      "loss": 0.7348,
      "step": 18200
    },
    {
      "epoch": 3.3644304032707675,
      "grad_norm": 0.9379822015762329,
      "learning_rate": 1.160211739663567e-05,
      "loss": 0.8477,
      "step": 18201
    },
    {
      "epoch": 3.364616242334139,
      "grad_norm": 1.0020337104797363,
      "learning_rate": 1.1595288079663912e-05,
      "loss": 0.8806,
      "step": 18202
    },
    {
      "epoch": 3.3648020813975097,
      "grad_norm": 0.783037006855011,
      "learning_rate": 1.1588460649539035e-05,
      "loss": 0.7428,
      "step": 18203
    },
    {
      "epoch": 3.364987920460881,
      "grad_norm": 0.8872162103652954,
      "learning_rate": 1.1581635106406718e-05,
      "loss": 0.8652,
      "step": 18204
    },
    {
      "epoch": 3.365173759524252,
      "grad_norm": 0.9861376881599426,
      "learning_rate": 1.1574811450412682e-05,
      "loss": 0.7183,
      "step": 18205
    },
    {
      "epoch": 3.3653595985876232,
      "grad_norm": 0.9283874034881592,
      "learning_rate": 1.1567989681702563e-05,
      "loss": 0.8004,
      "step": 18206
    },
    {
      "epoch": 3.365545437650994,
      "grad_norm": 0.8397024273872375,
      "learning_rate": 1.1561169800421944e-05,
      "loss": 0.8229,
      "step": 18207
    },
    {
      "epoch": 3.3657312767143654,
      "grad_norm": 0.9359708428382874,
      "learning_rate": 1.1554351806716358e-05,
      "loss": 0.7928,
      "step": 18208
    },
    {
      "epoch": 3.3659171157777363,
      "grad_norm": 1.1154730319976807,
      "learning_rate": 1.1547535700731372e-05,
      "loss": 0.7391,
      "step": 18209
    },
    {
      "epoch": 3.3661029548411077,
      "grad_norm": 0.8465034365653992,
      "learning_rate": 1.1540721482612448e-05,
      "loss": 0.7974,
      "step": 18210
    },
    {
      "epoch": 3.3662887939044785,
      "grad_norm": 0.9470143914222717,
      "learning_rate": 1.153390915250504e-05,
      "loss": 0.9165,
      "step": 18211
    },
    {
      "epoch": 3.36647463296785,
      "grad_norm": 0.9602369666099548,
      "learning_rate": 1.1527098710554496e-05,
      "loss": 0.8292,
      "step": 18212
    },
    {
      "epoch": 3.3666604720312208,
      "grad_norm": 0.9672537446022034,
      "learning_rate": 1.1520290156906221e-05,
      "loss": 0.7527,
      "step": 18213
    },
    {
      "epoch": 3.366846311094592,
      "grad_norm": 1.0739892721176147,
      "learning_rate": 1.1513483491705524e-05,
      "loss": 0.8147,
      "step": 18214
    },
    {
      "epoch": 3.3670321501579634,
      "grad_norm": 1.022194504737854,
      "learning_rate": 1.1506678715097673e-05,
      "loss": 0.7284,
      "step": 18215
    },
    {
      "epoch": 3.3672179892213343,
      "grad_norm": 1.012466549873352,
      "learning_rate": 1.149987582722788e-05,
      "loss": 0.9523,
      "step": 18216
    },
    {
      "epoch": 3.3674038282847056,
      "grad_norm": 0.8380947113037109,
      "learning_rate": 1.149307482824138e-05,
      "loss": 0.4953,
      "step": 18217
    },
    {
      "epoch": 3.3675896673480765,
      "grad_norm": 0.9516050815582275,
      "learning_rate": 1.1486275718283323e-05,
      "loss": 0.9402,
      "step": 18218
    },
    {
      "epoch": 3.367775506411448,
      "grad_norm": 0.9570687413215637,
      "learning_rate": 1.1479478497498797e-05,
      "loss": 0.7767,
      "step": 18219
    },
    {
      "epoch": 3.3679613454748187,
      "grad_norm": 1.1169215440750122,
      "learning_rate": 1.1472683166032904e-05,
      "loss": 0.9841,
      "step": 18220
    },
    {
      "epoch": 3.36814718453819,
      "grad_norm": 0.8415247797966003,
      "learning_rate": 1.1465889724030654e-05,
      "loss": 0.8199,
      "step": 18221
    },
    {
      "epoch": 3.368333023601561,
      "grad_norm": 0.9767362475395203,
      "learning_rate": 1.1459098171637084e-05,
      "loss": 0.6967,
      "step": 18222
    },
    {
      "epoch": 3.3685188626649323,
      "grad_norm": 1.0055066347122192,
      "learning_rate": 1.1452308508997111e-05,
      "loss": 0.9089,
      "step": 18223
    },
    {
      "epoch": 3.368704701728303,
      "grad_norm": 1.0055992603302002,
      "learning_rate": 1.1445520736255644e-05,
      "loss": 0.9448,
      "step": 18224
    },
    {
      "epoch": 3.3688905407916745,
      "grad_norm": 0.8618344068527222,
      "learning_rate": 1.1438734853557597e-05,
      "loss": 0.7672,
      "step": 18225
    },
    {
      "epoch": 3.3690763798550454,
      "grad_norm": 0.911933958530426,
      "learning_rate": 1.143195086104777e-05,
      "loss": 0.8706,
      "step": 18226
    },
    {
      "epoch": 3.3692622189184167,
      "grad_norm": 1.140630841255188,
      "learning_rate": 1.1425168758870964e-05,
      "loss": 0.9779,
      "step": 18227
    },
    {
      "epoch": 3.3694480579817876,
      "grad_norm": 1.1386034488677979,
      "learning_rate": 1.141838854717191e-05,
      "loss": 0.9669,
      "step": 18228
    },
    {
      "epoch": 3.369633897045159,
      "grad_norm": 0.9301897287368774,
      "learning_rate": 1.1411610226095349e-05,
      "loss": 0.641,
      "step": 18229
    },
    {
      "epoch": 3.3698197361085303,
      "grad_norm": 0.8807746767997742,
      "learning_rate": 1.1404833795785951e-05,
      "loss": 0.6094,
      "step": 18230
    },
    {
      "epoch": 3.370005575171901,
      "grad_norm": 0.8215765357017517,
      "learning_rate": 1.139805925638835e-05,
      "loss": 0.8526,
      "step": 18231
    },
    {
      "epoch": 3.370191414235272,
      "grad_norm": 0.923136293888092,
      "learning_rate": 1.1391286608047103e-05,
      "loss": 0.6587,
      "step": 18232
    },
    {
      "epoch": 3.3703772532986433,
      "grad_norm": 0.9220919013023376,
      "learning_rate": 1.1384515850906807e-05,
      "loss": 0.9753,
      "step": 18233
    },
    {
      "epoch": 3.3705630923620147,
      "grad_norm": 0.8931311964988708,
      "learning_rate": 1.1377746985111947e-05,
      "loss": 0.8369,
      "step": 18234
    },
    {
      "epoch": 3.3707489314253856,
      "grad_norm": 0.9474433660507202,
      "learning_rate": 1.1370980010806997e-05,
      "loss": 0.8126,
      "step": 18235
    },
    {
      "epoch": 3.370934770488757,
      "grad_norm": 0.9365830421447754,
      "learning_rate": 1.1364214928136362e-05,
      "loss": 0.8838,
      "step": 18236
    },
    {
      "epoch": 3.371120609552128,
      "grad_norm": 0.9512014389038086,
      "learning_rate": 1.1357451737244474e-05,
      "loss": 0.7424,
      "step": 18237
    },
    {
      "epoch": 3.371306448615499,
      "grad_norm": 1.041359782218933,
      "learning_rate": 1.135069043827567e-05,
      "loss": 0.6137,
      "step": 18238
    },
    {
      "epoch": 3.37149228767887,
      "grad_norm": 1.0264954566955566,
      "learning_rate": 1.1343931031374244e-05,
      "loss": 0.8591,
      "step": 18239
    },
    {
      "epoch": 3.3716781267422413,
      "grad_norm": 0.995843231678009,
      "learning_rate": 1.1337173516684451e-05,
      "loss": 0.9076,
      "step": 18240
    },
    {
      "epoch": 3.371863965805612,
      "grad_norm": 0.7283825874328613,
      "learning_rate": 1.1330417894350531e-05,
      "loss": 0.598,
      "step": 18241
    },
    {
      "epoch": 3.3720498048689835,
      "grad_norm": 0.9453611969947815,
      "learning_rate": 1.1323664164516712e-05,
      "loss": 0.8658,
      "step": 18242
    },
    {
      "epoch": 3.3722356439323544,
      "grad_norm": 0.9562330842018127,
      "learning_rate": 1.13169123273271e-05,
      "loss": 0.8286,
      "step": 18243
    },
    {
      "epoch": 3.3724214829957257,
      "grad_norm": 0.9837996363639832,
      "learning_rate": 1.1310162382925782e-05,
      "loss": 0.8294,
      "step": 18244
    },
    {
      "epoch": 3.3726073220590966,
      "grad_norm": 0.8887550830841064,
      "learning_rate": 1.1303414331456875e-05,
      "loss": 0.6426,
      "step": 18245
    },
    {
      "epoch": 3.372793161122468,
      "grad_norm": 1.0126186609268188,
      "learning_rate": 1.1296668173064385e-05,
      "loss": 0.6124,
      "step": 18246
    },
    {
      "epoch": 3.372979000185839,
      "grad_norm": 1.0040444135665894,
      "learning_rate": 1.1289923907892286e-05,
      "loss": 0.8762,
      "step": 18247
    },
    {
      "epoch": 3.37316483924921,
      "grad_norm": 1.069229245185852,
      "learning_rate": 1.1283181536084508e-05,
      "loss": 0.7736,
      "step": 18248
    },
    {
      "epoch": 3.3733506783125815,
      "grad_norm": 0.7893381118774414,
      "learning_rate": 1.1276441057785003e-05,
      "loss": 0.654,
      "step": 18249
    },
    {
      "epoch": 3.3735365173759524,
      "grad_norm": 1.025433897972107,
      "learning_rate": 1.1269702473137578e-05,
      "loss": 0.8712,
      "step": 18250
    },
    {
      "epoch": 3.3737223564393237,
      "grad_norm": 1.0625967979431152,
      "learning_rate": 1.1262965782286116e-05,
      "loss": 0.794,
      "step": 18251
    },
    {
      "epoch": 3.3739081955026946,
      "grad_norm": 0.9125096201896667,
      "learning_rate": 1.1256230985374339e-05,
      "loss": 0.6924,
      "step": 18252
    },
    {
      "epoch": 3.374094034566066,
      "grad_norm": 1.0169774293899536,
      "learning_rate": 1.124949808254605e-05,
      "loss": 0.783,
      "step": 18253
    },
    {
      "epoch": 3.374279873629437,
      "grad_norm": 0.9269421696662903,
      "learning_rate": 1.1242767073944926e-05,
      "loss": 0.9538,
      "step": 18254
    },
    {
      "epoch": 3.374465712692808,
      "grad_norm": 0.9135279655456543,
      "learning_rate": 1.1236037959714618e-05,
      "loss": 0.8072,
      "step": 18255
    },
    {
      "epoch": 3.374651551756179,
      "grad_norm": 0.8569638133049011,
      "learning_rate": 1.1229310739998733e-05,
      "loss": 0.7048,
      "step": 18256
    },
    {
      "epoch": 3.3748373908195504,
      "grad_norm": 1.1444765329360962,
      "learning_rate": 1.1222585414940901e-05,
      "loss": 0.815,
      "step": 18257
    },
    {
      "epoch": 3.3750232298829212,
      "grad_norm": 1.0449762344360352,
      "learning_rate": 1.1215861984684628e-05,
      "loss": 0.6666,
      "step": 18258
    },
    {
      "epoch": 3.3752090689462926,
      "grad_norm": 0.8725976943969727,
      "learning_rate": 1.1209140449373423e-05,
      "loss": 0.7466,
      "step": 18259
    },
    {
      "epoch": 3.3753949080096635,
      "grad_norm": 0.9362976551055908,
      "learning_rate": 1.1202420809150726e-05,
      "loss": 0.7889,
      "step": 18260
    },
    {
      "epoch": 3.375580747073035,
      "grad_norm": 0.8548526167869568,
      "learning_rate": 1.119570306415998e-05,
      "loss": 0.7122,
      "step": 18261
    },
    {
      "epoch": 3.3757665861364057,
      "grad_norm": 1.083701252937317,
      "learning_rate": 1.1188987214544565e-05,
      "loss": 0.8872,
      "step": 18262
    },
    {
      "epoch": 3.375952425199777,
      "grad_norm": 0.9338276982307434,
      "learning_rate": 1.1182273260447817e-05,
      "loss": 0.8801,
      "step": 18263
    },
    {
      "epoch": 3.3761382642631483,
      "grad_norm": 0.8710282444953918,
      "learning_rate": 1.1175561202013018e-05,
      "loss": 0.6658,
      "step": 18264
    },
    {
      "epoch": 3.376324103326519,
      "grad_norm": 0.90129154920578,
      "learning_rate": 1.1168851039383444e-05,
      "loss": 0.7791,
      "step": 18265
    },
    {
      "epoch": 3.37650994238989,
      "grad_norm": 1.1231645345687866,
      "learning_rate": 1.116214277270231e-05,
      "loss": 0.9588,
      "step": 18266
    },
    {
      "epoch": 3.3766957814532614,
      "grad_norm": 0.8895449042320251,
      "learning_rate": 1.1155436402112785e-05,
      "loss": 0.9344,
      "step": 18267
    },
    {
      "epoch": 3.3768816205166328,
      "grad_norm": 1.054827094078064,
      "learning_rate": 1.1148731927757983e-05,
      "loss": 0.8257,
      "step": 18268
    },
    {
      "epoch": 3.3770674595800036,
      "grad_norm": 0.7896285057067871,
      "learning_rate": 1.1142029349781046e-05,
      "loss": 0.7109,
      "step": 18269
    },
    {
      "epoch": 3.377253298643375,
      "grad_norm": 0.9272920489311218,
      "learning_rate": 1.1135328668325018e-05,
      "loss": 0.8157,
      "step": 18270
    },
    {
      "epoch": 3.377439137706746,
      "grad_norm": 0.9161113500595093,
      "learning_rate": 1.112862988353287e-05,
      "loss": 1.0153,
      "step": 18271
    },
    {
      "epoch": 3.377624976770117,
      "grad_norm": 0.7954372763633728,
      "learning_rate": 1.1121932995547601e-05,
      "loss": 0.7849,
      "step": 18272
    },
    {
      "epoch": 3.377810815833488,
      "grad_norm": 0.9211541414260864,
      "learning_rate": 1.1115238004512186e-05,
      "loss": 0.887,
      "step": 18273
    },
    {
      "epoch": 3.3779966548968594,
      "grad_norm": 0.8511643409729004,
      "learning_rate": 1.1108544910569474e-05,
      "loss": 0.9017,
      "step": 18274
    },
    {
      "epoch": 3.3781824939602303,
      "grad_norm": 1.4660279750823975,
      "learning_rate": 1.1101853713862332e-05,
      "loss": 1.1466,
      "step": 18275
    },
    {
      "epoch": 3.3783683330236016,
      "grad_norm": 0.8548551201820374,
      "learning_rate": 1.1095164414533543e-05,
      "loss": 0.9776,
      "step": 18276
    },
    {
      "epoch": 3.3785541720869725,
      "grad_norm": 0.8459223508834839,
      "learning_rate": 1.1088477012725917e-05,
      "loss": 0.7552,
      "step": 18277
    },
    {
      "epoch": 3.378740011150344,
      "grad_norm": 1.1502907276153564,
      "learning_rate": 1.1081791508582185e-05,
      "loss": 0.8632,
      "step": 18278
    },
    {
      "epoch": 3.378925850213715,
      "grad_norm": 0.996711790561676,
      "learning_rate": 1.1075107902245019e-05,
      "loss": 0.9776,
      "step": 18279
    },
    {
      "epoch": 3.379111689277086,
      "grad_norm": 0.917891800403595,
      "learning_rate": 1.1068426193857051e-05,
      "loss": 0.807,
      "step": 18280
    },
    {
      "epoch": 3.379297528340457,
      "grad_norm": 0.8591600060462952,
      "learning_rate": 1.1061746383560934e-05,
      "loss": 0.9063,
      "step": 18281
    },
    {
      "epoch": 3.3794833674038283,
      "grad_norm": 0.9120370745658875,
      "learning_rate": 1.1055068471499197e-05,
      "loss": 0.8746,
      "step": 18282
    },
    {
      "epoch": 3.3796692064671996,
      "grad_norm": 0.9349958896636963,
      "learning_rate": 1.1048392457814405e-05,
      "loss": 0.4698,
      "step": 18283
    },
    {
      "epoch": 3.3798550455305705,
      "grad_norm": 1.1120960712432861,
      "learning_rate": 1.104171834264901e-05,
      "loss": 0.874,
      "step": 18284
    },
    {
      "epoch": 3.380040884593942,
      "grad_norm": 0.9761961698532104,
      "learning_rate": 1.1035046126145499e-05,
      "loss": 0.7462,
      "step": 18285
    },
    {
      "epoch": 3.3802267236573127,
      "grad_norm": 1.0015671253204346,
      "learning_rate": 1.1028375808446256e-05,
      "loss": 0.8401,
      "step": 18286
    },
    {
      "epoch": 3.380412562720684,
      "grad_norm": 0.8225238919258118,
      "learning_rate": 1.102170738969366e-05,
      "loss": 0.743,
      "step": 18287
    },
    {
      "epoch": 3.380598401784055,
      "grad_norm": 1.0311946868896484,
      "learning_rate": 1.1015040870029992e-05,
      "loss": 0.8776,
      "step": 18288
    },
    {
      "epoch": 3.3807842408474262,
      "grad_norm": 1.012428641319275,
      "learning_rate": 1.1008376249597596e-05,
      "loss": 0.8297,
      "step": 18289
    },
    {
      "epoch": 3.380970079910797,
      "grad_norm": 1.0129367113113403,
      "learning_rate": 1.1001713528538704e-05,
      "loss": 1.0106,
      "step": 18290
    },
    {
      "epoch": 3.3811559189741685,
      "grad_norm": 0.9160915613174438,
      "learning_rate": 1.0995052706995502e-05,
      "loss": 0.6437,
      "step": 18291
    },
    {
      "epoch": 3.3813417580375393,
      "grad_norm": 0.973933756351471,
      "learning_rate": 1.0988393785110118e-05,
      "loss": 0.885,
      "step": 18292
    },
    {
      "epoch": 3.3815275971009107,
      "grad_norm": 0.8929466605186462,
      "learning_rate": 1.0981736763024764e-05,
      "loss": 0.7792,
      "step": 18293
    },
    {
      "epoch": 3.3817134361642815,
      "grad_norm": 1.1814781427383423,
      "learning_rate": 1.097508164088148e-05,
      "loss": 0.692,
      "step": 18294
    },
    {
      "epoch": 3.381899275227653,
      "grad_norm": 0.9687650203704834,
      "learning_rate": 1.0968428418822296e-05,
      "loss": 0.7951,
      "step": 18295
    },
    {
      "epoch": 3.3820851142910238,
      "grad_norm": 0.8729159235954285,
      "learning_rate": 1.0961777096989212e-05,
      "loss": 0.7703,
      "step": 18296
    },
    {
      "epoch": 3.382270953354395,
      "grad_norm": 0.712313711643219,
      "learning_rate": 1.0955127675524214e-05,
      "loss": 0.7491,
      "step": 18297
    },
    {
      "epoch": 3.3824567924177664,
      "grad_norm": 0.8763603568077087,
      "learning_rate": 1.0948480154569207e-05,
      "loss": 0.7241,
      "step": 18298
    },
    {
      "epoch": 3.3826426314811373,
      "grad_norm": 0.9360529780387878,
      "learning_rate": 1.0941834534266083e-05,
      "loss": 0.7995,
      "step": 18299
    },
    {
      "epoch": 3.3828284705445086,
      "grad_norm": 1.0026065111160278,
      "learning_rate": 1.0935190814756646e-05,
      "loss": 0.8343,
      "step": 18300
    },
    {
      "epoch": 3.3830143096078795,
      "grad_norm": 0.8706707954406738,
      "learning_rate": 1.092854899618273e-05,
      "loss": 0.6912,
      "step": 18301
    },
    {
      "epoch": 3.383200148671251,
      "grad_norm": 0.9272733330726624,
      "learning_rate": 1.0921909078686099e-05,
      "loss": 0.9772,
      "step": 18302
    },
    {
      "epoch": 3.3833859877346217,
      "grad_norm": 0.9085229635238647,
      "learning_rate": 1.0915271062408428e-05,
      "loss": 0.8226,
      "step": 18303
    },
    {
      "epoch": 3.383571826797993,
      "grad_norm": 0.8816424012184143,
      "learning_rate": 1.0908634947491415e-05,
      "loss": 0.813,
      "step": 18304
    },
    {
      "epoch": 3.383757665861364,
      "grad_norm": 0.9400708079338074,
      "learning_rate": 1.0902000734076723e-05,
      "loss": 0.7601,
      "step": 18305
    },
    {
      "epoch": 3.3839435049247353,
      "grad_norm": 0.9915024638175964,
      "learning_rate": 1.089536842230594e-05,
      "loss": 0.7915,
      "step": 18306
    },
    {
      "epoch": 3.384129343988106,
      "grad_norm": 0.8487621545791626,
      "learning_rate": 1.0888738012320598e-05,
      "loss": 0.7124,
      "step": 18307
    },
    {
      "epoch": 3.3843151830514775,
      "grad_norm": 0.9537492394447327,
      "learning_rate": 1.0882109504262195e-05,
      "loss": 0.5165,
      "step": 18308
    },
    {
      "epoch": 3.3845010221148484,
      "grad_norm": 0.9375237822532654,
      "learning_rate": 1.0875482898272261e-05,
      "loss": 0.7454,
      "step": 18309
    },
    {
      "epoch": 3.3846868611782197,
      "grad_norm": 0.8832110166549683,
      "learning_rate": 1.0868858194492204e-05,
      "loss": 0.679,
      "step": 18310
    },
    {
      "epoch": 3.3848727002415906,
      "grad_norm": 0.9369748830795288,
      "learning_rate": 1.0862235393063413e-05,
      "loss": 0.7653,
      "step": 18311
    },
    {
      "epoch": 3.385058539304962,
      "grad_norm": 1.932416558265686,
      "learning_rate": 1.085561449412722e-05,
      "loss": 1.2103,
      "step": 18312
    },
    {
      "epoch": 3.3852443783683333,
      "grad_norm": 0.9251443147659302,
      "learning_rate": 1.0848995497824955e-05,
      "loss": 0.8801,
      "step": 18313
    },
    {
      "epoch": 3.385430217431704,
      "grad_norm": 1.2950562238693237,
      "learning_rate": 1.0842378404297904e-05,
      "loss": 0.8168,
      "step": 18314
    },
    {
      "epoch": 3.385616056495075,
      "grad_norm": 0.8824008703231812,
      "learning_rate": 1.08357632136873e-05,
      "loss": 0.7471,
      "step": 18315
    },
    {
      "epoch": 3.3858018955584464,
      "grad_norm": 1.043350100517273,
      "learning_rate": 1.0829149926134285e-05,
      "loss": 1.1583,
      "step": 18316
    },
    {
      "epoch": 3.3859877346218177,
      "grad_norm": 0.9486417770385742,
      "learning_rate": 1.082253854178007e-05,
      "loss": 0.7595,
      "step": 18317
    },
    {
      "epoch": 3.3861735736851886,
      "grad_norm": 0.7901325225830078,
      "learning_rate": 1.0815929060765739e-05,
      "loss": 0.7761,
      "step": 18318
    },
    {
      "epoch": 3.38635941274856,
      "grad_norm": 0.8888583779335022,
      "learning_rate": 1.0809321483232348e-05,
      "loss": 0.7806,
      "step": 18319
    },
    {
      "epoch": 3.386545251811931,
      "grad_norm": 0.9182053208351135,
      "learning_rate": 1.0802715809320918e-05,
      "loss": 0.9955,
      "step": 18320
    },
    {
      "epoch": 3.386731090875302,
      "grad_norm": 1.0127581357955933,
      "learning_rate": 1.0796112039172468e-05,
      "loss": 0.98,
      "step": 18321
    },
    {
      "epoch": 3.386916929938673,
      "grad_norm": 1.1698704957962036,
      "learning_rate": 1.0789510172927918e-05,
      "loss": 0.7972,
      "step": 18322
    },
    {
      "epoch": 3.3871027690020443,
      "grad_norm": 0.9562602043151855,
      "learning_rate": 1.078291021072817e-05,
      "loss": 0.8788,
      "step": 18323
    },
    {
      "epoch": 3.387288608065415,
      "grad_norm": 0.9870921969413757,
      "learning_rate": 1.0776312152714108e-05,
      "loss": 0.8564,
      "step": 18324
    },
    {
      "epoch": 3.3874744471287865,
      "grad_norm": 0.949755847454071,
      "learning_rate": 1.0769715999026564e-05,
      "loss": 0.7915,
      "step": 18325
    },
    {
      "epoch": 3.3876602861921574,
      "grad_norm": 0.9226321578025818,
      "learning_rate": 1.0763121749806304e-05,
      "loss": 0.8564,
      "step": 18326
    },
    {
      "epoch": 3.3878461252555288,
      "grad_norm": 1.2454584836959839,
      "learning_rate": 1.075652940519407e-05,
      "loss": 0.8097,
      "step": 18327
    },
    {
      "epoch": 3.3880319643189,
      "grad_norm": 0.9788869023323059,
      "learning_rate": 1.0749938965330552e-05,
      "loss": 0.9494,
      "step": 18328
    },
    {
      "epoch": 3.388217803382271,
      "grad_norm": 0.7627193927764893,
      "learning_rate": 1.0743350430356447e-05,
      "loss": 0.6081,
      "step": 18329
    },
    {
      "epoch": 3.388403642445642,
      "grad_norm": 1.7664850950241089,
      "learning_rate": 1.0736763800412365e-05,
      "loss": 0.8987,
      "step": 18330
    },
    {
      "epoch": 3.388589481509013,
      "grad_norm": 0.9788818955421448,
      "learning_rate": 1.0730179075638868e-05,
      "loss": 0.7628,
      "step": 18331
    },
    {
      "epoch": 3.3887753205723845,
      "grad_norm": 0.9811448454856873,
      "learning_rate": 1.0723596256176483e-05,
      "loss": 0.8474,
      "step": 18332
    },
    {
      "epoch": 3.3889611596357554,
      "grad_norm": 0.8933404684066772,
      "learning_rate": 1.071701534216576e-05,
      "loss": 0.8238,
      "step": 18333
    },
    {
      "epoch": 3.3891469986991267,
      "grad_norm": 0.9329827427864075,
      "learning_rate": 1.0710436333747109e-05,
      "loss": 0.841,
      "step": 18334
    },
    {
      "epoch": 3.3893328377624976,
      "grad_norm": 0.7232084274291992,
      "learning_rate": 1.0703859231060975e-05,
      "loss": 0.6748,
      "step": 18335
    },
    {
      "epoch": 3.389518676825869,
      "grad_norm": 1.0448474884033203,
      "learning_rate": 1.0697284034247724e-05,
      "loss": 0.8262,
      "step": 18336
    },
    {
      "epoch": 3.38970451588924,
      "grad_norm": 0.8937177062034607,
      "learning_rate": 1.0690710743447707e-05,
      "loss": 0.9501,
      "step": 18337
    },
    {
      "epoch": 3.389890354952611,
      "grad_norm": 0.8887295126914978,
      "learning_rate": 1.0684139358801205e-05,
      "loss": 0.7002,
      "step": 18338
    },
    {
      "epoch": 3.390076194015982,
      "grad_norm": 0.675439178943634,
      "learning_rate": 1.067756988044848e-05,
      "loss": 0.4412,
      "step": 18339
    },
    {
      "epoch": 3.3902620330793534,
      "grad_norm": NaN,
      "learning_rate": 1.067756988044848e-05,
      "loss": 0.8879,
      "step": 18340
    },
    {
      "epoch": 3.3904478721427243,
      "grad_norm": 0.8766264319419861,
      "learning_rate": 1.067100230852972e-05,
      "loss": 0.787,
      "step": 18341
    },
    {
      "epoch": 3.3906337112060956,
      "grad_norm": 0.9330973029136658,
      "learning_rate": 1.0664436643185149e-05,
      "loss": 0.7302,
      "step": 18342
    },
    {
      "epoch": 3.3908195502694665,
      "grad_norm": 0.7809070348739624,
      "learning_rate": 1.0657872884554865e-05,
      "loss": 0.584,
      "step": 18343
    },
    {
      "epoch": 3.391005389332838,
      "grad_norm": 0.8240885138511658,
      "learning_rate": 1.0651311032778955e-05,
      "loss": 0.7813,
      "step": 18344
    },
    {
      "epoch": 3.3911912283962087,
      "grad_norm": 1.007537603378296,
      "learning_rate": 1.0644751087997495e-05,
      "loss": 0.9514,
      "step": 18345
    },
    {
      "epoch": 3.39137706745958,
      "grad_norm": 0.9993891716003418,
      "learning_rate": 1.0638193050350464e-05,
      "loss": 0.8986,
      "step": 18346
    },
    {
      "epoch": 3.3915629065229513,
      "grad_norm": 0.8804618120193481,
      "learning_rate": 1.0631636919977871e-05,
      "loss": 0.9539,
      "step": 18347
    },
    {
      "epoch": 3.3917487455863222,
      "grad_norm": 1.138956069946289,
      "learning_rate": 1.0625082697019628e-05,
      "loss": 0.9374,
      "step": 18348
    },
    {
      "epoch": 3.3919345846496936,
      "grad_norm": 1.0949984788894653,
      "learning_rate": 1.0618530381615599e-05,
      "loss": 0.8375,
      "step": 18349
    },
    {
      "epoch": 3.3921204237130644,
      "grad_norm": 0.9565363526344299,
      "learning_rate": 1.0611979973905672e-05,
      "loss": 0.7164,
      "step": 18350
    },
    {
      "epoch": 3.3923062627764358,
      "grad_norm": 1.3193212747573853,
      "learning_rate": 1.0605431474029637e-05,
      "loss": 0.8225,
      "step": 18351
    },
    {
      "epoch": 3.3924921018398067,
      "grad_norm": 0.884672224521637,
      "learning_rate": 1.0598884882127258e-05,
      "loss": 0.6619,
      "step": 18352
    },
    {
      "epoch": 3.392677940903178,
      "grad_norm": 1.015413522720337,
      "learning_rate": 1.0592340198338258e-05,
      "loss": 0.8915,
      "step": 18353
    },
    {
      "epoch": 3.392863779966549,
      "grad_norm": 0.8938660025596619,
      "learning_rate": 1.0585797422802335e-05,
      "loss": 0.8037,
      "step": 18354
    },
    {
      "epoch": 3.39304961902992,
      "grad_norm": 0.8355761170387268,
      "learning_rate": 1.0579256555659123e-05,
      "loss": 0.4946,
      "step": 18355
    },
    {
      "epoch": 3.393235458093291,
      "grad_norm": 0.8579012155532837,
      "learning_rate": 1.057271759704821e-05,
      "loss": 0.8542,
      "step": 18356
    },
    {
      "epoch": 3.3934212971566624,
      "grad_norm": 1.0483804941177368,
      "learning_rate": 1.0566180547109173e-05,
      "loss": 0.8588,
      "step": 18357
    },
    {
      "epoch": 3.3936071362200333,
      "grad_norm": 0.9812692403793335,
      "learning_rate": 1.0559645405981566e-05,
      "loss": 0.8812,
      "step": 18358
    },
    {
      "epoch": 3.3937929752834046,
      "grad_norm": 1.053178071975708,
      "learning_rate": 1.0553112173804835e-05,
      "loss": 0.9585,
      "step": 18359
    },
    {
      "epoch": 3.3939788143467755,
      "grad_norm": 0.8852431178092957,
      "learning_rate": 1.0546580850718435e-05,
      "loss": 0.9329,
      "step": 18360
    },
    {
      "epoch": 3.394164653410147,
      "grad_norm": 1.1948872804641724,
      "learning_rate": 1.054005143686173e-05,
      "loss": 0.9735,
      "step": 18361
    },
    {
      "epoch": 3.394350492473518,
      "grad_norm": 1.1322181224822998,
      "learning_rate": 1.0533523932374134e-05,
      "loss": 0.8999,
      "step": 18362
    },
    {
      "epoch": 3.394536331536889,
      "grad_norm": 0.8780480027198792,
      "learning_rate": 1.0526998337394933e-05,
      "loss": 0.9141,
      "step": 18363
    },
    {
      "epoch": 3.39472217060026,
      "grad_norm": 0.8715663552284241,
      "learning_rate": 1.0520474652063394e-05,
      "loss": 0.6362,
      "step": 18364
    },
    {
      "epoch": 3.3949080096636313,
      "grad_norm": 0.8854778409004211,
      "learning_rate": 1.0513952876518796e-05,
      "loss": 0.9201,
      "step": 18365
    },
    {
      "epoch": 3.3950938487270026,
      "grad_norm": 0.9006483554840088,
      "learning_rate": 1.0507433010900303e-05,
      "loss": 0.6453,
      "step": 18366
    },
    {
      "epoch": 3.3952796877903735,
      "grad_norm": 1.0067654848098755,
      "learning_rate": 1.050091505534706e-05,
      "loss": 0.8112,
      "step": 18367
    },
    {
      "epoch": 3.395465526853745,
      "grad_norm": 1.026990294456482,
      "learning_rate": 1.0494399009998212e-05,
      "loss": 0.6235,
      "step": 18368
    },
    {
      "epoch": 3.3956513659171157,
      "grad_norm": 0.9172264933586121,
      "learning_rate": 1.0487884874992803e-05,
      "loss": 0.7043,
      "step": 18369
    },
    {
      "epoch": 3.395837204980487,
      "grad_norm": 0.9033785462379456,
      "learning_rate": 1.0481372650469912e-05,
      "loss": 0.665,
      "step": 18370
    },
    {
      "epoch": 3.396023044043858,
      "grad_norm": 0.8452712297439575,
      "learning_rate": 1.0474862336568492e-05,
      "loss": 0.7495,
      "step": 18371
    },
    {
      "epoch": 3.3962088831072292,
      "grad_norm": 0.9229526519775391,
      "learning_rate": 1.04683539334275e-05,
      "loss": 0.7547,
      "step": 18372
    },
    {
      "epoch": 3.3963947221706,
      "grad_norm": 1.1461782455444336,
      "learning_rate": 1.0461847441185834e-05,
      "loss": 1.07,
      "step": 18373
    },
    {
      "epoch": 3.3965805612339715,
      "grad_norm": 1.0893611907958984,
      "learning_rate": 1.0455342859982409e-05,
      "loss": 0.8133,
      "step": 18374
    },
    {
      "epoch": 3.3967664002973423,
      "grad_norm": 1.0343410968780518,
      "learning_rate": 1.0448840189956022e-05,
      "loss": 0.9544,
      "step": 18375
    },
    {
      "epoch": 3.3969522393607137,
      "grad_norm": 0.784121036529541,
      "learning_rate": 1.0442339431245441e-05,
      "loss": 0.7844,
      "step": 18376
    },
    {
      "epoch": 3.3971380784240846,
      "grad_norm": 1.0187186002731323,
      "learning_rate": 1.0435840583989443e-05,
      "loss": 0.8634,
      "step": 18377
    },
    {
      "epoch": 3.397323917487456,
      "grad_norm": 0.8501676917076111,
      "learning_rate": 1.0429343648326739e-05,
      "loss": 0.7817,
      "step": 18378
    },
    {
      "epoch": 3.3975097565508268,
      "grad_norm": 0.8218252062797546,
      "learning_rate": 1.0422848624395998e-05,
      "loss": 0.7286,
      "step": 18379
    },
    {
      "epoch": 3.397695595614198,
      "grad_norm": 1.3224713802337646,
      "learning_rate": 1.041635551233583e-05,
      "loss": 1.24,
      "step": 18380
    },
    {
      "epoch": 3.3978814346775694,
      "grad_norm": 1.0980743169784546,
      "learning_rate": 1.04098643122848e-05,
      "loss": 0.7951,
      "step": 18381
    },
    {
      "epoch": 3.3980672737409403,
      "grad_norm": 1.0464022159576416,
      "learning_rate": 1.040337502438149e-05,
      "loss": 0.9525,
      "step": 18382
    },
    {
      "epoch": 3.3982531128043116,
      "grad_norm": 0.9397351741790771,
      "learning_rate": 1.0396887648764386e-05,
      "loss": 0.9203,
      "step": 18383
    },
    {
      "epoch": 3.3984389518676825,
      "grad_norm": 1.1105705499649048,
      "learning_rate": 1.0390402185571934e-05,
      "loss": 0.9589,
      "step": 18384
    },
    {
      "epoch": 3.398624790931054,
      "grad_norm": 1.1369887590408325,
      "learning_rate": 1.038391863494259e-05,
      "loss": 0.764,
      "step": 18385
    },
    {
      "epoch": 3.3988106299944247,
      "grad_norm": 1.0813019275665283,
      "learning_rate": 1.0377436997014711e-05,
      "loss": 0.8825,
      "step": 18386
    },
    {
      "epoch": 3.398996469057796,
      "grad_norm": 0.9052541255950928,
      "learning_rate": 1.0370957271926652e-05,
      "loss": 0.6729,
      "step": 18387
    },
    {
      "epoch": 3.399182308121167,
      "grad_norm": 1.7850995063781738,
      "learning_rate": 1.0364479459816667e-05,
      "loss": 1.1384,
      "step": 18388
    },
    {
      "epoch": 3.3993681471845383,
      "grad_norm": 0.8072879314422607,
      "learning_rate": 1.035800356082306e-05,
      "loss": 0.6948,
      "step": 18389
    },
    {
      "epoch": 3.399553986247909,
      "grad_norm": 0.8056290745735168,
      "learning_rate": 1.0351529575084051e-05,
      "loss": 0.4928,
      "step": 18390
    },
    {
      "epoch": 3.3997398253112805,
      "grad_norm": 0.8751406073570251,
      "learning_rate": 1.0345057502737798e-05,
      "loss": 0.732,
      "step": 18391
    },
    {
      "epoch": 3.3999256643746514,
      "grad_norm": 0.9247086048126221,
      "learning_rate": 1.0338587343922435e-05,
      "loss": 0.7336,
      "step": 18392
    },
    {
      "epoch": 3.4001115034380227,
      "grad_norm": 0.8679309487342834,
      "learning_rate": 1.033211909877605e-05,
      "loss": 0.7248,
      "step": 18393
    },
    {
      "epoch": 3.4002973425013936,
      "grad_norm": 0.9449750185012817,
      "learning_rate": 1.0325652767436711e-05,
      "loss": 1.1002,
      "step": 18394
    },
    {
      "epoch": 3.400483181564765,
      "grad_norm": 0.9762586951255798,
      "learning_rate": 1.0319188350042442e-05,
      "loss": 0.7576,
      "step": 18395
    },
    {
      "epoch": 3.4006690206281363,
      "grad_norm": 1.1907624006271362,
      "learning_rate": 1.0312725846731175e-05,
      "loss": 1.0439,
      "step": 18396
    },
    {
      "epoch": 3.400854859691507,
      "grad_norm": 0.8823482394218445,
      "learning_rate": 1.0306265257640879e-05,
      "loss": 0.851,
      "step": 18397
    },
    {
      "epoch": 3.4010406987548785,
      "grad_norm": 0.9875620603561401,
      "learning_rate": 1.029980658290941e-05,
      "loss": 0.8812,
      "step": 18398
    },
    {
      "epoch": 3.4012265378182494,
      "grad_norm": 1.1810065507888794,
      "learning_rate": 1.029334982267467e-05,
      "loss": 0.8582,
      "step": 18399
    },
    {
      "epoch": 3.4014123768816207,
      "grad_norm": 1.1070986986160278,
      "learning_rate": 1.0286894977074424e-05,
      "loss": 0.969,
      "step": 18400
    },
    {
      "epoch": 3.4015982159449916,
      "grad_norm": 0.8918006420135498,
      "learning_rate": 1.0280442046246441e-05,
      "loss": 0.7951,
      "step": 18401
    },
    {
      "epoch": 3.401784055008363,
      "grad_norm": 1.1689788103103638,
      "learning_rate": 1.0273991030328468e-05,
      "loss": 0.9026,
      "step": 18402
    },
    {
      "epoch": 3.401969894071734,
      "grad_norm": 0.9085296988487244,
      "learning_rate": 1.0267541929458179e-05,
      "loss": 0.817,
      "step": 18403
    },
    {
      "epoch": 3.402155733135105,
      "grad_norm": 0.7045767307281494,
      "learning_rate": 1.0261094743773203e-05,
      "loss": 0.5753,
      "step": 18404
    },
    {
      "epoch": 3.402341572198476,
      "grad_norm": 0.9169483184814453,
      "learning_rate": 1.0254649473411192e-05,
      "loss": 0.7561,
      "step": 18405
    },
    {
      "epoch": 3.4025274112618473,
      "grad_norm": 0.9103830456733704,
      "learning_rate": 1.0248206118509663e-05,
      "loss": 0.6749,
      "step": 18406
    },
    {
      "epoch": 3.402713250325218,
      "grad_norm": 1.016221523284912,
      "learning_rate": 1.0241764679206168e-05,
      "loss": 1.0526,
      "step": 18407
    },
    {
      "epoch": 3.4028990893885895,
      "grad_norm": 1.1982204914093018,
      "learning_rate": 1.0235325155638142e-05,
      "loss": 0.5418,
      "step": 18408
    },
    {
      "epoch": 3.4030849284519604,
      "grad_norm": 0.8669753074645996,
      "learning_rate": 1.0228887547943066e-05,
      "loss": 0.8356,
      "step": 18409
    },
    {
      "epoch": 3.4032707675153318,
      "grad_norm": 0.9314633011817932,
      "learning_rate": 1.0222451856258352e-05,
      "loss": 0.7528,
      "step": 18410
    },
    {
      "epoch": 3.403456606578703,
      "grad_norm": 0.9003772139549255,
      "learning_rate": 1.0216018080721335e-05,
      "loss": 0.6818,
      "step": 18411
    },
    {
      "epoch": 3.403642445642074,
      "grad_norm": 1.2028591632843018,
      "learning_rate": 1.0209586221469335e-05,
      "loss": 0.8116,
      "step": 18412
    },
    {
      "epoch": 3.403828284705445,
      "grad_norm": 0.9273260831832886,
      "learning_rate": 1.0203156278639625e-05,
      "loss": 0.5253,
      "step": 18413
    },
    {
      "epoch": 3.404014123768816,
      "grad_norm": 1.0306023359298706,
      "learning_rate": 1.0196728252369447e-05,
      "loss": 1.005,
      "step": 18414
    },
    {
      "epoch": 3.4041999628321875,
      "grad_norm": 1.029457688331604,
      "learning_rate": 1.0190302142796004e-05,
      "loss": 0.7985,
      "step": 18415
    },
    {
      "epoch": 3.4043858018955584,
      "grad_norm": 0.8798620700836182,
      "learning_rate": 1.0183877950056431e-05,
      "loss": 0.7899,
      "step": 18416
    },
    {
      "epoch": 3.4045716409589297,
      "grad_norm": 1.1171305179595947,
      "learning_rate": 1.0177455674287861e-05,
      "loss": 0.8113,
      "step": 18417
    },
    {
      "epoch": 3.4047574800223006,
      "grad_norm": 0.9819057583808899,
      "learning_rate": 1.0171035315627363e-05,
      "loss": 1.0677,
      "step": 18418
    },
    {
      "epoch": 3.404943319085672,
      "grad_norm": 0.8564400672912598,
      "learning_rate": 1.016461687421194e-05,
      "loss": 0.6969,
      "step": 18419
    },
    {
      "epoch": 3.405129158149043,
      "grad_norm": 0.968268871307373,
      "learning_rate": 1.0158200350178627e-05,
      "loss": 0.8084,
      "step": 18420
    },
    {
      "epoch": 3.405314997212414,
      "grad_norm": 0.8687854409217834,
      "learning_rate": 1.0151785743664333e-05,
      "loss": 0.8461,
      "step": 18421
    },
    {
      "epoch": 3.405500836275785,
      "grad_norm": 1.0560195446014404,
      "learning_rate": 1.0145373054806007e-05,
      "loss": 0.5994,
      "step": 18422
    },
    {
      "epoch": 3.4056866753391564,
      "grad_norm": 0.7632457613945007,
      "learning_rate": 1.0138962283740494e-05,
      "loss": 0.8112,
      "step": 18423
    },
    {
      "epoch": 3.4058725144025273,
      "grad_norm": 1.2036640644073486,
      "learning_rate": 1.0132553430604608e-05,
      "loss": 0.7384,
      "step": 18424
    },
    {
      "epoch": 3.4060583534658986,
      "grad_norm": 1.1761847734451294,
      "learning_rate": 1.012614649553517e-05,
      "loss": 0.8114,
      "step": 18425
    },
    {
      "epoch": 3.4062441925292695,
      "grad_norm": 1.1434109210968018,
      "learning_rate": 1.011974147866891e-05,
      "loss": 0.8595,
      "step": 18426
    },
    {
      "epoch": 3.406430031592641,
      "grad_norm": 0.9771609902381897,
      "learning_rate": 1.0113338380142513e-05,
      "loss": 0.9144,
      "step": 18427
    },
    {
      "epoch": 3.4066158706560117,
      "grad_norm": 0.854061484336853,
      "learning_rate": 1.0106937200092648e-05,
      "loss": 0.5983,
      "step": 18428
    },
    {
      "epoch": 3.406801709719383,
      "grad_norm": 0.9027602076530457,
      "learning_rate": 1.0100537938655963e-05,
      "loss": 0.8257,
      "step": 18429
    },
    {
      "epoch": 3.4069875487827543,
      "grad_norm": 0.917335569858551,
      "learning_rate": 1.0094140595969004e-05,
      "loss": 0.8791,
      "step": 18430
    },
    {
      "epoch": 3.4071733878461252,
      "grad_norm": 1.032515287399292,
      "learning_rate": 1.0087745172168338e-05,
      "loss": 0.7861,
      "step": 18431
    },
    {
      "epoch": 3.4073592269094966,
      "grad_norm": 0.8733752965927124,
      "learning_rate": 1.008135166739047e-05,
      "loss": 0.9026,
      "step": 18432
    },
    {
      "epoch": 3.4075450659728674,
      "grad_norm": 1.133104681968689,
      "learning_rate": 1.0074960081771822e-05,
      "loss": 0.981,
      "step": 18433
    },
    {
      "epoch": 3.4077309050362388,
      "grad_norm": 0.8792244791984558,
      "learning_rate": 1.0068570415448852e-05,
      "loss": 0.7714,
      "step": 18434
    },
    {
      "epoch": 3.4079167440996097,
      "grad_norm": 1.0705727338790894,
      "learning_rate": 1.0062182668557907e-05,
      "loss": 0.9903,
      "step": 18435
    },
    {
      "epoch": 3.408102583162981,
      "grad_norm": 1.041522741317749,
      "learning_rate": 1.0055796841235332e-05,
      "loss": 0.9209,
      "step": 18436
    },
    {
      "epoch": 3.408288422226352,
      "grad_norm": 0.8988785743713379,
      "learning_rate": 1.004941293361743e-05,
      "loss": 0.6943,
      "step": 18437
    },
    {
      "epoch": 3.408474261289723,
      "grad_norm": 0.9060993790626526,
      "learning_rate": 1.0043030945840447e-05,
      "loss": 0.7478,
      "step": 18438
    },
    {
      "epoch": 3.408660100353094,
      "grad_norm": 0.9467904567718506,
      "learning_rate": 1.0036650878040599e-05,
      "loss": 0.8402,
      "step": 18439
    },
    {
      "epoch": 3.4088459394164654,
      "grad_norm": 1.0410497188568115,
      "learning_rate": 1.003027273035404e-05,
      "loss": 0.9122,
      "step": 18440
    },
    {
      "epoch": 3.4090317784798363,
      "grad_norm": 0.9075198769569397,
      "learning_rate": 1.0023896502916908e-05,
      "loss": 0.9351,
      "step": 18441
    },
    {
      "epoch": 3.4092176175432076,
      "grad_norm": 0.8967174887657166,
      "learning_rate": 1.0017522195865326e-05,
      "loss": 0.8257,
      "step": 18442
    },
    {
      "epoch": 3.4094034566065785,
      "grad_norm": 0.8995972871780396,
      "learning_rate": 1.0011149809335318e-05,
      "loss": 0.646,
      "step": 18443
    },
    {
      "epoch": 3.40958929566995,
      "grad_norm": 0.8520565032958984,
      "learning_rate": 1.0004779343462867e-05,
      "loss": 0.7053,
      "step": 18444
    },
    {
      "epoch": 3.409775134733321,
      "grad_norm": 1.0125555992126465,
      "learning_rate": 9.998410798383984e-06,
      "loss": 0.8133,
      "step": 18445
    },
    {
      "epoch": 3.409960973796692,
      "grad_norm": 0.9866090416908264,
      "learning_rate": 9.992044174234561e-06,
      "loss": 0.9799,
      "step": 18446
    },
    {
      "epoch": 3.410146812860063,
      "grad_norm": 1.2013903856277466,
      "learning_rate": 9.98567947115051e-06,
      "loss": 0.8826,
      "step": 18447
    },
    {
      "epoch": 3.4103326519234343,
      "grad_norm": 0.846396803855896,
      "learning_rate": 9.979316689267636e-06,
      "loss": 0.8524,
      "step": 18448
    },
    {
      "epoch": 3.4105184909868056,
      "grad_norm": 0.9855815768241882,
      "learning_rate": 9.972955828721787e-06,
      "loss": 0.7695,
      "step": 18449
    },
    {
      "epoch": 3.4107043300501765,
      "grad_norm": 0.931355357170105,
      "learning_rate": 9.966596889648706e-06,
      "loss": 0.7244,
      "step": 18450
    },
    {
      "epoch": 3.410890169113548,
      "grad_norm": 0.9406512379646301,
      "learning_rate": 9.960239872184085e-06,
      "loss": 0.8765,
      "step": 18451
    },
    {
      "epoch": 3.4110760081769187,
      "grad_norm": 0.9696619510650635,
      "learning_rate": 9.953884776463652e-06,
      "loss": 1.0083,
      "step": 18452
    },
    {
      "epoch": 3.41126184724029,
      "grad_norm": 1.0051896572113037,
      "learning_rate": 9.947531602622995e-06,
      "loss": 0.6334,
      "step": 18453
    },
    {
      "epoch": 3.411447686303661,
      "grad_norm": 1.5537279844284058,
      "learning_rate": 9.941180350797774e-06,
      "loss": 1.2872,
      "step": 18454
    },
    {
      "epoch": 3.4116335253670322,
      "grad_norm": 0.8657042980194092,
      "learning_rate": 9.934831021123502e-06,
      "loss": 0.808,
      "step": 18455
    },
    {
      "epoch": 3.411819364430403,
      "grad_norm": 0.9618478417396545,
      "learning_rate": 9.928483613735684e-06,
      "loss": 0.7162,
      "step": 18456
    },
    {
      "epoch": 3.4120052034937745,
      "grad_norm": 0.9999720454216003,
      "learning_rate": 9.922138128769831e-06,
      "loss": 0.9676,
      "step": 18457
    },
    {
      "epoch": 3.4121910425571453,
      "grad_norm": 0.8654031157493591,
      "learning_rate": 9.91579456636137e-06,
      "loss": 0.8429,
      "step": 18458
    },
    {
      "epoch": 3.4123768816205167,
      "grad_norm": 0.9619836807250977,
      "learning_rate": 9.90945292664568e-06,
      "loss": 0.5053,
      "step": 18459
    },
    {
      "epoch": 3.412562720683888,
      "grad_norm": 0.8627220988273621,
      "learning_rate": 9.903113209758096e-06,
      "loss": 0.7697,
      "step": 18460
    },
    {
      "epoch": 3.412748559747259,
      "grad_norm": 0.8880993127822876,
      "learning_rate": 9.896775415833947e-06,
      "loss": 0.5013,
      "step": 18461
    },
    {
      "epoch": 3.4129343988106298,
      "grad_norm": 0.9004693031311035,
      "learning_rate": 9.890439545008523e-06,
      "loss": 0.8842,
      "step": 18462
    },
    {
      "epoch": 3.413120237874001,
      "grad_norm": 1.0259151458740234,
      "learning_rate": 9.884105597417026e-06,
      "loss": 0.8893,
      "step": 18463
    },
    {
      "epoch": 3.4133060769373724,
      "grad_norm": 0.954877495765686,
      "learning_rate": 9.877773573194637e-06,
      "loss": 0.8357,
      "step": 18464
    },
    {
      "epoch": 3.4134919160007433,
      "grad_norm": 1.0130378007888794,
      "learning_rate": 9.871443472476538e-06,
      "loss": 0.8433,
      "step": 18465
    },
    {
      "epoch": 3.4136777550641146,
      "grad_norm": 0.9062729477882385,
      "learning_rate": 9.865115295397808e-06,
      "loss": 0.878,
      "step": 18466
    },
    {
      "epoch": 3.4138635941274855,
      "grad_norm": 0.9700537323951721,
      "learning_rate": 9.858789042093508e-06,
      "loss": 0.9453,
      "step": 18467
    },
    {
      "epoch": 3.414049433190857,
      "grad_norm": 1.3633286952972412,
      "learning_rate": 9.85246471269866e-06,
      "loss": 0.9084,
      "step": 18468
    },
    {
      "epoch": 3.4142352722542277,
      "grad_norm": 0.8684259653091431,
      "learning_rate": 9.84614230734826e-06,
      "loss": 0.7808,
      "step": 18469
    },
    {
      "epoch": 3.414421111317599,
      "grad_norm": 1.0092612504959106,
      "learning_rate": 9.839821826177254e-06,
      "loss": 0.7104,
      "step": 18470
    },
    {
      "epoch": 3.41460695038097,
      "grad_norm": 0.8173282146453857,
      "learning_rate": 9.83350326932052e-06,
      "loss": 0.7987,
      "step": 18471
    },
    {
      "epoch": 3.4147927894443413,
      "grad_norm": 1.0052686929702759,
      "learning_rate": 9.827186636912911e-06,
      "loss": 0.791,
      "step": 18472
    },
    {
      "epoch": 3.414978628507712,
      "grad_norm": 1.0594308376312256,
      "learning_rate": 9.820871929089271e-06,
      "loss": 0.8965,
      "step": 18473
    },
    {
      "epoch": 3.4151644675710835,
      "grad_norm": 1.091590166091919,
      "learning_rate": 9.81455914598437e-06,
      "loss": 0.9273,
      "step": 18474
    },
    {
      "epoch": 3.4153503066344544,
      "grad_norm": 1.0309139490127563,
      "learning_rate": 9.808248287732947e-06,
      "loss": 0.7162,
      "step": 18475
    },
    {
      "epoch": 3.4155361456978257,
      "grad_norm": 0.715144693851471,
      "learning_rate": 9.80193935446967e-06,
      "loss": 0.8441,
      "step": 18476
    },
    {
      "epoch": 3.4157219847611966,
      "grad_norm": 1.0665366649627686,
      "learning_rate": 9.795632346329232e-06,
      "loss": 0.6637,
      "step": 18477
    },
    {
      "epoch": 3.415907823824568,
      "grad_norm": 0.8957089185714722,
      "learning_rate": 9.789327263446213e-06,
      "loss": 0.8185,
      "step": 18478
    },
    {
      "epoch": 3.4160936628879393,
      "grad_norm": 1.088594675064087,
      "learning_rate": 9.783024105955207e-06,
      "loss": 0.6706,
      "step": 18479
    },
    {
      "epoch": 3.41627950195131,
      "grad_norm": 1.17194402217865,
      "learning_rate": 9.776722873990719e-06,
      "loss": 0.8527,
      "step": 18480
    },
    {
      "epoch": 3.4164653410146815,
      "grad_norm": 0.8662586808204651,
      "learning_rate": 9.77042356768726e-06,
      "loss": 0.8623,
      "step": 18481
    },
    {
      "epoch": 3.4166511800780524,
      "grad_norm": 1.0301966667175293,
      "learning_rate": 9.764126187179256e-06,
      "loss": 0.8051,
      "step": 18482
    },
    {
      "epoch": 3.4168370191414237,
      "grad_norm": 0.891620934009552,
      "learning_rate": 9.757830732601136e-06,
      "loss": 0.7495,
      "step": 18483
    },
    {
      "epoch": 3.4170228582047946,
      "grad_norm": 0.9465243816375732,
      "learning_rate": 9.751537204087258e-06,
      "loss": 0.6785,
      "step": 18484
    },
    {
      "epoch": 3.417208697268166,
      "grad_norm": 1.1204577684402466,
      "learning_rate": 9.745245601771946e-06,
      "loss": 0.84,
      "step": 18485
    },
    {
      "epoch": 3.417394536331537,
      "grad_norm": 0.9050443172454834,
      "learning_rate": 9.738955925789494e-06,
      "loss": 0.9034,
      "step": 18486
    },
    {
      "epoch": 3.417580375394908,
      "grad_norm": 0.9914594888687134,
      "learning_rate": 9.73266817627413e-06,
      "loss": 0.9183,
      "step": 18487
    },
    {
      "epoch": 3.417766214458279,
      "grad_norm": 0.9951683878898621,
      "learning_rate": 9.726382353360041e-06,
      "loss": 0.8975,
      "step": 18488
    },
    {
      "epoch": 3.4179520535216503,
      "grad_norm": 1.047358751296997,
      "learning_rate": 9.720098457181425e-06,
      "loss": 0.7869,
      "step": 18489
    },
    {
      "epoch": 3.418137892585021,
      "grad_norm": 1.073513388633728,
      "learning_rate": 9.713816487872374e-06,
      "loss": 0.8416,
      "step": 18490
    },
    {
      "epoch": 3.4183237316483925,
      "grad_norm": 1.0014564990997314,
      "learning_rate": 9.707536445566977e-06,
      "loss": 0.9485,
      "step": 18491
    },
    {
      "epoch": 3.4185095707117634,
      "grad_norm": 0.8295140266418457,
      "learning_rate": 9.701258330399255e-06,
      "loss": 0.7742,
      "step": 18492
    },
    {
      "epoch": 3.4186954097751348,
      "grad_norm": 1.0205795764923096,
      "learning_rate": 9.694982142503205e-06,
      "loss": 0.9397,
      "step": 18493
    },
    {
      "epoch": 3.418881248838506,
      "grad_norm": 1.060734748840332,
      "learning_rate": 9.688707882012814e-06,
      "loss": 0.7891,
      "step": 18494
    },
    {
      "epoch": 3.419067087901877,
      "grad_norm": 1.210156798362732,
      "learning_rate": 9.682435549061974e-06,
      "loss": 0.812,
      "step": 18495
    },
    {
      "epoch": 3.419252926965248,
      "grad_norm": 0.9137862920761108,
      "learning_rate": 9.676165143784544e-06,
      "loss": 0.6196,
      "step": 18496
    },
    {
      "epoch": 3.419438766028619,
      "grad_norm": 1.0834940671920776,
      "learning_rate": 9.669896666314371e-06,
      "loss": 0.8834,
      "step": 18497
    },
    {
      "epoch": 3.4196246050919905,
      "grad_norm": 1.0085303783416748,
      "learning_rate": 9.66363011678525e-06,
      "loss": 0.7188,
      "step": 18498
    },
    {
      "epoch": 3.4198104441553614,
      "grad_norm": 0.9292789697647095,
      "learning_rate": 9.657365495330916e-06,
      "loss": 1.0473,
      "step": 18499
    },
    {
      "epoch": 3.4199962832187327,
      "grad_norm": 0.8212962746620178,
      "learning_rate": 9.651102802085066e-06,
      "loss": 0.5599,
      "step": 18500
    },
    {
      "epoch": 3.4201821222821036,
      "grad_norm": 0.9323970079421997,
      "learning_rate": 9.6448420371814e-06,
      "loss": 0.7903,
      "step": 18501
    },
    {
      "epoch": 3.420367961345475,
      "grad_norm": 0.8843657970428467,
      "learning_rate": 9.638583200753525e-06,
      "loss": 0.9521,
      "step": 18502
    },
    {
      "epoch": 3.420553800408846,
      "grad_norm": 1.1132733821868896,
      "learning_rate": 9.632326292935012e-06,
      "loss": 0.586,
      "step": 18503
    },
    {
      "epoch": 3.420739639472217,
      "grad_norm": 0.9576237797737122,
      "learning_rate": 9.626071313859409e-06,
      "loss": 0.9634,
      "step": 18504
    },
    {
      "epoch": 3.420925478535588,
      "grad_norm": 0.9085745811462402,
      "learning_rate": 9.619818263660252e-06,
      "loss": 0.819,
      "step": 18505
    },
    {
      "epoch": 3.4211113175989594,
      "grad_norm": 0.9066290855407715,
      "learning_rate": 9.613567142470969e-06,
      "loss": 0.6965,
      "step": 18506
    },
    {
      "epoch": 3.4212971566623303,
      "grad_norm": 0.9321801662445068,
      "learning_rate": 9.607317950424987e-06,
      "loss": 0.7295,
      "step": 18507
    },
    {
      "epoch": 3.4214829957257016,
      "grad_norm": 0.9430164694786072,
      "learning_rate": 9.601070687655667e-06,
      "loss": 0.6397,
      "step": 18508
    },
    {
      "epoch": 3.421668834789073,
      "grad_norm": 0.9664836525917053,
      "learning_rate": 9.594825354296377e-06,
      "loss": 0.841,
      "step": 18509
    },
    {
      "epoch": 3.421854673852444,
      "grad_norm": 0.9440019130706787,
      "learning_rate": 9.588581950480401e-06,
      "loss": 1.039,
      "step": 18510
    },
    {
      "epoch": 3.4220405129158147,
      "grad_norm": 1.0869544744491577,
      "learning_rate": 9.58234047634099e-06,
      "loss": 0.8229,
      "step": 18511
    },
    {
      "epoch": 3.422226351979186,
      "grad_norm": 0.9242025017738342,
      "learning_rate": 9.576100932011344e-06,
      "loss": 0.7676,
      "step": 18512
    },
    {
      "epoch": 3.4224121910425573,
      "grad_norm": 0.9432842135429382,
      "learning_rate": 9.569863317624662e-06,
      "loss": 0.6781,
      "step": 18513
    },
    {
      "epoch": 3.4225980301059282,
      "grad_norm": 0.8574406504631042,
      "learning_rate": 9.563627633314054e-06,
      "loss": 0.6487,
      "step": 18514
    },
    {
      "epoch": 3.4227838691692996,
      "grad_norm": 0.942700982093811,
      "learning_rate": 9.557393879212629e-06,
      "loss": 0.7327,
      "step": 18515
    },
    {
      "epoch": 3.4229697082326704,
      "grad_norm": 0.9813656806945801,
      "learning_rate": 9.551162055453411e-06,
      "loss": 0.8764,
      "step": 18516
    },
    {
      "epoch": 3.4231555472960418,
      "grad_norm": 0.7930585145950317,
      "learning_rate": 9.54493216216944e-06,
      "loss": 0.6155,
      "step": 18517
    },
    {
      "epoch": 3.4233413863594127,
      "grad_norm": 1.2245550155639648,
      "learning_rate": 9.538704199493664e-06,
      "loss": 0.813,
      "step": 18518
    },
    {
      "epoch": 3.423527225422784,
      "grad_norm": 0.8570970892906189,
      "learning_rate": 9.532478167559011e-06,
      "loss": 0.922,
      "step": 18519
    },
    {
      "epoch": 3.423713064486155,
      "grad_norm": 0.8624559640884399,
      "learning_rate": 9.52625406649834e-06,
      "loss": 0.6445,
      "step": 18520
    },
    {
      "epoch": 3.423898903549526,
      "grad_norm": 0.8528590798377991,
      "learning_rate": 9.520031896444536e-06,
      "loss": 0.9415,
      "step": 18521
    },
    {
      "epoch": 3.424084742612897,
      "grad_norm": 0.9589774012565613,
      "learning_rate": 9.513811657530391e-06,
      "loss": 0.9034,
      "step": 18522
    },
    {
      "epoch": 3.4242705816762684,
      "grad_norm": 0.9136031270027161,
      "learning_rate": 9.507593349888644e-06,
      "loss": 0.8908,
      "step": 18523
    },
    {
      "epoch": 3.4244564207396393,
      "grad_norm": 1.0127958059310913,
      "learning_rate": 9.501376973651999e-06,
      "loss": 0.6507,
      "step": 18524
    },
    {
      "epoch": 3.4246422598030106,
      "grad_norm": 0.8139373064041138,
      "learning_rate": 9.495162528953195e-06,
      "loss": 0.6984,
      "step": 18525
    },
    {
      "epoch": 3.4248280988663815,
      "grad_norm": 0.9890860319137573,
      "learning_rate": 9.488950015924824e-06,
      "loss": 0.777,
      "step": 18526
    },
    {
      "epoch": 3.425013937929753,
      "grad_norm": 1.1747925281524658,
      "learning_rate": 9.482739434699506e-06,
      "loss": 0.9121,
      "step": 18527
    },
    {
      "epoch": 3.425199776993124,
      "grad_norm": 1.0490977764129639,
      "learning_rate": 9.476530785409754e-06,
      "loss": 0.9954,
      "step": 18528
    },
    {
      "epoch": 3.425385616056495,
      "grad_norm": 1.0559828281402588,
      "learning_rate": 9.47032406818813e-06,
      "loss": 0.741,
      "step": 18529
    },
    {
      "epoch": 3.4255714551198664,
      "grad_norm": 2.027252674102783,
      "learning_rate": 9.464119283167072e-06,
      "loss": 1.3585,
      "step": 18530
    },
    {
      "epoch": 3.4257572941832373,
      "grad_norm": 0.9632944464683533,
      "learning_rate": 9.45791643047904e-06,
      "loss": 0.8285,
      "step": 18531
    },
    {
      "epoch": 3.4259431332466086,
      "grad_norm": 1.0093849897384644,
      "learning_rate": 9.451715510256377e-06,
      "loss": 0.9482,
      "step": 18532
    },
    {
      "epoch": 3.4261289723099795,
      "grad_norm": 0.9287329912185669,
      "learning_rate": 9.445516522631482e-06,
      "loss": 0.8109,
      "step": 18533
    },
    {
      "epoch": 3.426314811373351,
      "grad_norm": 0.957906186580658,
      "learning_rate": 9.43931946773663e-06,
      "loss": 0.7097,
      "step": 18534
    },
    {
      "epoch": 3.4265006504367217,
      "grad_norm": 1.0049121379852295,
      "learning_rate": 9.433124345704092e-06,
      "loss": 0.8023,
      "step": 18535
    },
    {
      "epoch": 3.426686489500093,
      "grad_norm": 0.9933798313140869,
      "learning_rate": 9.426931156666085e-06,
      "loss": 0.7093,
      "step": 18536
    },
    {
      "epoch": 3.426872328563464,
      "grad_norm": 1.0770821571350098,
      "learning_rate": 9.420739900754815e-06,
      "loss": 0.7999,
      "step": 18537
    },
    {
      "epoch": 3.4270581676268352,
      "grad_norm": 0.9103144407272339,
      "learning_rate": 9.414550578102422e-06,
      "loss": 0.7976,
      "step": 18538
    },
    {
      "epoch": 3.427244006690206,
      "grad_norm": 0.9013404250144958,
      "learning_rate": 9.408363188840996e-06,
      "loss": 0.8276,
      "step": 18539
    },
    {
      "epoch": 3.4274298457535775,
      "grad_norm": 0.9676746726036072,
      "learning_rate": 9.402177733102579e-06,
      "loss": 0.539,
      "step": 18540
    },
    {
      "epoch": 3.4276156848169483,
      "grad_norm": 0.9252790808677673,
      "learning_rate": 9.39599421101922e-06,
      "loss": 0.7271,
      "step": 18541
    },
    {
      "epoch": 3.4278015238803197,
      "grad_norm": 1.0059067010879517,
      "learning_rate": 9.389812622722882e-06,
      "loss": 0.8074,
      "step": 18542
    },
    {
      "epoch": 3.427987362943691,
      "grad_norm": 1.0993884801864624,
      "learning_rate": 9.383632968345501e-06,
      "loss": 0.809,
      "step": 18543
    },
    {
      "epoch": 3.428173202007062,
      "grad_norm": 0.864134669303894,
      "learning_rate": 9.377455248018963e-06,
      "loss": 0.8721,
      "step": 18544
    },
    {
      "epoch": 3.4283590410704328,
      "grad_norm": 0.9246398210525513,
      "learning_rate": 9.371279461875115e-06,
      "loss": 0.6925,
      "step": 18545
    },
    {
      "epoch": 3.428544880133804,
      "grad_norm": 0.9502807259559631,
      "learning_rate": 9.365105610045798e-06,
      "loss": 1.0406,
      "step": 18546
    },
    {
      "epoch": 3.4287307191971754,
      "grad_norm": 0.8488006591796875,
      "learning_rate": 9.358933692662775e-06,
      "loss": 0.8352,
      "step": 18547
    },
    {
      "epoch": 3.4289165582605463,
      "grad_norm": 1.1370174884796143,
      "learning_rate": 9.352763709857749e-06,
      "loss": 0.913,
      "step": 18548
    },
    {
      "epoch": 3.4291023973239176,
      "grad_norm": 0.9592936635017395,
      "learning_rate": 9.346595661762436e-06,
      "loss": 0.7774,
      "step": 18549
    },
    {
      "epoch": 3.4292882363872885,
      "grad_norm": 0.841438353061676,
      "learning_rate": 9.340429548508468e-06,
      "loss": 0.6929,
      "step": 18550
    },
    {
      "epoch": 3.42947407545066,
      "grad_norm": 0.9155181050300598,
      "learning_rate": 9.334265370227458e-06,
      "loss": 0.7457,
      "step": 18551
    },
    {
      "epoch": 3.4296599145140307,
      "grad_norm": 0.9380632638931274,
      "learning_rate": 9.328103127050947e-06,
      "loss": 0.9594,
      "step": 18552
    },
    {
      "epoch": 3.429845753577402,
      "grad_norm": 1.0478308200836182,
      "learning_rate": 9.321942819110496e-06,
      "loss": 0.9126,
      "step": 18553
    },
    {
      "epoch": 3.430031592640773,
      "grad_norm": 0.9147072434425354,
      "learning_rate": 9.315784446537568e-06,
      "loss": 0.8514,
      "step": 18554
    },
    {
      "epoch": 3.4302174317041443,
      "grad_norm": 0.9005574584007263,
      "learning_rate": 9.309628009463579e-06,
      "loss": 0.749,
      "step": 18555
    },
    {
      "epoch": 3.430403270767515,
      "grad_norm": 0.9355246424674988,
      "learning_rate": 9.303473508019944e-06,
      "loss": 0.7584,
      "step": 18556
    },
    {
      "epoch": 3.4305891098308865,
      "grad_norm": 2.22025465965271,
      "learning_rate": 9.29732094233805e-06,
      "loss": 1.1015,
      "step": 18557
    },
    {
      "epoch": 3.4307749488942574,
      "grad_norm": 1.0061955451965332,
      "learning_rate": 9.29117031254919e-06,
      "loss": 0.715,
      "step": 18558
    },
    {
      "epoch": 3.4309607879576287,
      "grad_norm": 1.1127967834472656,
      "learning_rate": 9.285021618784628e-06,
      "loss": 0.947,
      "step": 18559
    },
    {
      "epoch": 3.4311466270209996,
      "grad_norm": 0.9937306642532349,
      "learning_rate": 9.278874861175601e-06,
      "loss": 0.9385,
      "step": 18560
    },
    {
      "epoch": 3.431332466084371,
      "grad_norm": 0.901160478591919,
      "learning_rate": 9.272730039853317e-06,
      "loss": 0.8884,
      "step": 18561
    },
    {
      "epoch": 3.4315183051477423,
      "grad_norm": 1.075750708580017,
      "learning_rate": 9.266587154948914e-06,
      "loss": 0.931,
      "step": 18562
    },
    {
      "epoch": 3.431704144211113,
      "grad_norm": 0.9136353135108948,
      "learning_rate": 9.260446206593509e-06,
      "loss": 0.8377,
      "step": 18563
    },
    {
      "epoch": 3.4318899832744845,
      "grad_norm": 1.0850541591644287,
      "learning_rate": 9.254307194918144e-06,
      "loss": 0.8653,
      "step": 18564
    },
    {
      "epoch": 3.4320758223378554,
      "grad_norm": 1.0017285346984863,
      "learning_rate": 9.248170120053878e-06,
      "loss": 0.7573,
      "step": 18565
    },
    {
      "epoch": 3.4322616614012267,
      "grad_norm": 0.9335682988166809,
      "learning_rate": 9.242034982131675e-06,
      "loss": 0.9018,
      "step": 18566
    },
    {
      "epoch": 3.4324475004645976,
      "grad_norm": 0.8629546165466309,
      "learning_rate": 9.235901781282496e-06,
      "loss": 0.4799,
      "step": 18567
    },
    {
      "epoch": 3.432633339527969,
      "grad_norm": 1.1791781187057495,
      "learning_rate": 9.229770517637227e-06,
      "loss": 0.7093,
      "step": 18568
    },
    {
      "epoch": 3.43281917859134,
      "grad_norm": 0.9566700458526611,
      "learning_rate": 9.22364119132676e-06,
      "loss": 0.7499,
      "step": 18569
    },
    {
      "epoch": 3.433005017654711,
      "grad_norm": 1.1813853979110718,
      "learning_rate": 9.21751380248188e-06,
      "loss": 0.8975,
      "step": 18570
    },
    {
      "epoch": 3.433190856718082,
      "grad_norm": 1.0684021711349487,
      "learning_rate": 9.211388351233396e-06,
      "loss": 0.747,
      "step": 18571
    },
    {
      "epoch": 3.4333766957814533,
      "grad_norm": 0.9259321689605713,
      "learning_rate": 9.205264837711991e-06,
      "loss": 0.8761,
      "step": 18572
    },
    {
      "epoch": 3.433562534844824,
      "grad_norm": 1.0204839706420898,
      "learning_rate": 9.199143262048427e-06,
      "loss": 0.8364,
      "step": 18573
    },
    {
      "epoch": 3.4337483739081955,
      "grad_norm": 0.8068413734436035,
      "learning_rate": 9.193023624373332e-06,
      "loss": 0.5637,
      "step": 18574
    },
    {
      "epoch": 3.4339342129715664,
      "grad_norm": 0.8738575577735901,
      "learning_rate": 9.186905924817301e-06,
      "loss": 0.7658,
      "step": 18575
    },
    {
      "epoch": 3.4341200520349378,
      "grad_norm": 0.9638357758522034,
      "learning_rate": 9.180790163510933e-06,
      "loss": 0.8261,
      "step": 18576
    },
    {
      "epoch": 3.434305891098309,
      "grad_norm": 1.031435251235962,
      "learning_rate": 9.174676340584721e-06,
      "loss": 0.9145,
      "step": 18577
    },
    {
      "epoch": 3.43449173016168,
      "grad_norm": 0.877661406993866,
      "learning_rate": 9.168564456169215e-06,
      "loss": 0.7629,
      "step": 18578
    },
    {
      "epoch": 3.4346775692250513,
      "grad_norm": 0.9631086587905884,
      "learning_rate": 9.162454510394813e-06,
      "loss": 0.7251,
      "step": 18579
    },
    {
      "epoch": 3.434863408288422,
      "grad_norm": 1.0152180194854736,
      "learning_rate": 9.156346503391921e-06,
      "loss": 0.791,
      "step": 18580
    },
    {
      "epoch": 3.4350492473517935,
      "grad_norm": 0.9161962866783142,
      "learning_rate": 9.150240435290947e-06,
      "loss": 0.8774,
      "step": 18581
    },
    {
      "epoch": 3.4352350864151644,
      "grad_norm": 0.8945319056510925,
      "learning_rate": 9.144136306222185e-06,
      "loss": 0.8238,
      "step": 18582
    },
    {
      "epoch": 3.4354209254785357,
      "grad_norm": 0.8093655109405518,
      "learning_rate": 9.138034116315907e-06,
      "loss": 0.6699,
      "step": 18583
    },
    {
      "epoch": 3.4356067645419066,
      "grad_norm": 1.2264208793640137,
      "learning_rate": 9.131933865702358e-06,
      "loss": 0.9794,
      "step": 18584
    },
    {
      "epoch": 3.435792603605278,
      "grad_norm": 0.7184557318687439,
      "learning_rate": 9.125835554511753e-06,
      "loss": 0.54,
      "step": 18585
    },
    {
      "epoch": 3.435978442668649,
      "grad_norm": 0.8314578533172607,
      "learning_rate": 9.119739182874254e-06,
      "loss": 0.7339,
      "step": 18586
    },
    {
      "epoch": 3.43616428173202,
      "grad_norm": 0.8828555345535278,
      "learning_rate": 9.113644750919936e-06,
      "loss": 0.9658,
      "step": 18587
    },
    {
      "epoch": 3.436350120795391,
      "grad_norm": 1.0641796588897705,
      "learning_rate": 9.107552258778907e-06,
      "loss": 0.7089,
      "step": 18588
    },
    {
      "epoch": 3.4365359598587624,
      "grad_norm": 0.8992449641227722,
      "learning_rate": 9.101461706581216e-06,
      "loss": 0.8167,
      "step": 18589
    },
    {
      "epoch": 3.4367217989221333,
      "grad_norm": 1.0069942474365234,
      "learning_rate": 9.095373094456839e-06,
      "loss": 0.704,
      "step": 18590
    },
    {
      "epoch": 3.4369076379855046,
      "grad_norm": 0.9273150563240051,
      "learning_rate": 9.089286422535726e-06,
      "loss": 0.8596,
      "step": 18591
    },
    {
      "epoch": 3.437093477048876,
      "grad_norm": 0.8703062534332275,
      "learning_rate": 9.083201690947763e-06,
      "loss": 0.7614,
      "step": 18592
    },
    {
      "epoch": 3.437279316112247,
      "grad_norm": 0.9495863914489746,
      "learning_rate": 9.077118899822857e-06,
      "loss": 0.8382,
      "step": 18593
    },
    {
      "epoch": 3.4374651551756177,
      "grad_norm": 0.9941915273666382,
      "learning_rate": 9.071038049290825e-06,
      "loss": 0.9582,
      "step": 18594
    },
    {
      "epoch": 3.437650994238989,
      "grad_norm": 1.2743667364120483,
      "learning_rate": 9.064959139481422e-06,
      "loss": 0.9139,
      "step": 18595
    },
    {
      "epoch": 3.4378368333023603,
      "grad_norm": 0.9535473585128784,
      "learning_rate": 9.05888217052443e-06,
      "loss": 0.8414,
      "step": 18596
    },
    {
      "epoch": 3.4380226723657312,
      "grad_norm": 0.8331275582313538,
      "learning_rate": 9.052807142549546e-06,
      "loss": 0.7422,
      "step": 18597
    },
    {
      "epoch": 3.4382085114291026,
      "grad_norm": 0.8911383748054504,
      "learning_rate": 9.046734055686401e-06,
      "loss": 0.8685,
      "step": 18598
    },
    {
      "epoch": 3.4383943504924734,
      "grad_norm": 1.0376087427139282,
      "learning_rate": 9.040662910064646e-06,
      "loss": 0.7465,
      "step": 18599
    },
    {
      "epoch": 3.4385801895558448,
      "grad_norm": 1.1285665035247803,
      "learning_rate": 9.034593705813843e-06,
      "loss": 1.2301,
      "step": 18600
    },
    {
      "epoch": 3.4387660286192157,
      "grad_norm": 0.9236640334129333,
      "learning_rate": 9.028526443063546e-06,
      "loss": 0.9906,
      "step": 18601
    },
    {
      "epoch": 3.438951867682587,
      "grad_norm": 1.0847108364105225,
      "learning_rate": 9.022461121943238e-06,
      "loss": 0.8161,
      "step": 18602
    },
    {
      "epoch": 3.439137706745958,
      "grad_norm": 0.729643702507019,
      "learning_rate": 9.016397742582373e-06,
      "loss": 0.3817,
      "step": 18603
    },
    {
      "epoch": 3.439323545809329,
      "grad_norm": 0.861388623714447,
      "learning_rate": 9.010336305110345e-06,
      "loss": 0.9105,
      "step": 18604
    },
    {
      "epoch": 3.4395093848727,
      "grad_norm": 0.6577388048171997,
      "learning_rate": 9.004276809656564e-06,
      "loss": 0.4741,
      "step": 18605
    },
    {
      "epoch": 3.4396952239360714,
      "grad_norm": 0.8057047724723816,
      "learning_rate": 8.998219256350349e-06,
      "loss": 0.6355,
      "step": 18606
    },
    {
      "epoch": 3.4398810629994423,
      "grad_norm": 0.9502322673797607,
      "learning_rate": 8.99216364532095e-06,
      "loss": 0.9192,
      "step": 18607
    },
    {
      "epoch": 3.4400669020628136,
      "grad_norm": 1.1867955923080444,
      "learning_rate": 8.986109976697643e-06,
      "loss": 0.8475,
      "step": 18608
    },
    {
      "epoch": 3.4402527411261845,
      "grad_norm": 0.9152840971946716,
      "learning_rate": 8.980058250609647e-06,
      "loss": 0.8698,
      "step": 18609
    },
    {
      "epoch": 3.440438580189556,
      "grad_norm": 0.9248976707458496,
      "learning_rate": 8.974008467186124e-06,
      "loss": 0.8996,
      "step": 18610
    },
    {
      "epoch": 3.440624419252927,
      "grad_norm": 0.7468507289886475,
      "learning_rate": 8.967960626556172e-06,
      "loss": 0.8113,
      "step": 18611
    },
    {
      "epoch": 3.440810258316298,
      "grad_norm": 1.05239737033844,
      "learning_rate": 8.961914728848864e-06,
      "loss": 0.8636,
      "step": 18612
    },
    {
      "epoch": 3.4409960973796694,
      "grad_norm": 1.0198965072631836,
      "learning_rate": 8.955870774193287e-06,
      "loss": 0.9243,
      "step": 18613
    },
    {
      "epoch": 3.4411819364430403,
      "grad_norm": 0.8686797618865967,
      "learning_rate": 8.949828762718404e-06,
      "loss": 0.6556,
      "step": 18614
    },
    {
      "epoch": 3.4413677755064116,
      "grad_norm": 1.007763385772705,
      "learning_rate": 8.943788694553157e-06,
      "loss": 0.9191,
      "step": 18615
    },
    {
      "epoch": 3.4415536145697825,
      "grad_norm": 1.1565308570861816,
      "learning_rate": 8.937750569826508e-06,
      "loss": 0.7586,
      "step": 18616
    },
    {
      "epoch": 3.441739453633154,
      "grad_norm": 0.7900993227958679,
      "learning_rate": 8.93171438866729e-06,
      "loss": 0.5349,
      "step": 18617
    },
    {
      "epoch": 3.4419252926965247,
      "grad_norm": 0.9284295439720154,
      "learning_rate": 8.925680151204363e-06,
      "loss": 0.8183,
      "step": 18618
    },
    {
      "epoch": 3.442111131759896,
      "grad_norm": 0.9950578808784485,
      "learning_rate": 8.919647857566482e-06,
      "loss": 1.0031,
      "step": 18619
    },
    {
      "epoch": 3.442296970823267,
      "grad_norm": 1.1575837135314941,
      "learning_rate": 8.91361750788241e-06,
      "loss": 0.9175,
      "step": 18620
    },
    {
      "epoch": 3.4424828098866382,
      "grad_norm": 0.8874562382698059,
      "learning_rate": 8.90758910228089e-06,
      "loss": 0.8789,
      "step": 18621
    },
    {
      "epoch": 3.442668648950009,
      "grad_norm": 0.893882691860199,
      "learning_rate": 8.901562640890548e-06,
      "loss": 0.6872,
      "step": 18622
    },
    {
      "epoch": 3.4428544880133805,
      "grad_norm": 0.8668039441108704,
      "learning_rate": 8.89553812384003e-06,
      "loss": 0.7776,
      "step": 18623
    },
    {
      "epoch": 3.4430403270767513,
      "grad_norm": 0.844916820526123,
      "learning_rate": 8.889515551257888e-06,
      "loss": 0.8515,
      "step": 18624
    },
    {
      "epoch": 3.4432261661401227,
      "grad_norm": 0.9719852209091187,
      "learning_rate": 8.883494923272707e-06,
      "loss": 0.7908,
      "step": 18625
    },
    {
      "epoch": 3.443412005203494,
      "grad_norm": 0.9443371891975403,
      "learning_rate": 8.87747624001296e-06,
      "loss": 0.8065,
      "step": 18626
    },
    {
      "epoch": 3.443597844266865,
      "grad_norm": 1.1390016078948975,
      "learning_rate": 8.871459501607093e-06,
      "loss": 0.8421,
      "step": 18627
    },
    {
      "epoch": 3.4437836833302358,
      "grad_norm": 0.8440651297569275,
      "learning_rate": 8.865444708183556e-06,
      "loss": 0.7339,
      "step": 18628
    },
    {
      "epoch": 3.443969522393607,
      "grad_norm": 0.9557360410690308,
      "learning_rate": 8.859431859870692e-06,
      "loss": 0.7506,
      "step": 18629
    },
    {
      "epoch": 3.4441553614569784,
      "grad_norm": 0.8384804129600525,
      "learning_rate": 8.853420956796866e-06,
      "loss": 0.6606,
      "step": 18630
    },
    {
      "epoch": 3.4443412005203493,
      "grad_norm": 1.0732804536819458,
      "learning_rate": 8.84741199909036e-06,
      "loss": 0.8602,
      "step": 18631
    },
    {
      "epoch": 3.4445270395837206,
      "grad_norm": 0.9016420245170593,
      "learning_rate": 8.841404986879409e-06,
      "loss": 0.973,
      "step": 18632
    },
    {
      "epoch": 3.4447128786470915,
      "grad_norm": 0.9232279062271118,
      "learning_rate": 8.835399920292241e-06,
      "loss": 0.7959,
      "step": 18633
    },
    {
      "epoch": 3.444898717710463,
      "grad_norm": 0.9467220306396484,
      "learning_rate": 8.829396799457024e-06,
      "loss": 0.9368,
      "step": 18634
    },
    {
      "epoch": 3.4450845567738337,
      "grad_norm": 0.7945659160614014,
      "learning_rate": 8.823395624501863e-06,
      "loss": 0.7663,
      "step": 18635
    },
    {
      "epoch": 3.445270395837205,
      "grad_norm": 0.8220109343528748,
      "learning_rate": 8.817396395554866e-06,
      "loss": 0.5273,
      "step": 18636
    },
    {
      "epoch": 3.445456234900576,
      "grad_norm": 0.8508674502372742,
      "learning_rate": 8.811399112744067e-06,
      "loss": 0.8074,
      "step": 18637
    },
    {
      "epoch": 3.4456420739639473,
      "grad_norm": 1.1290472745895386,
      "learning_rate": 8.805403776197474e-06,
      "loss": 0.984,
      "step": 18638
    },
    {
      "epoch": 3.445827913027318,
      "grad_norm": 0.9419941306114197,
      "learning_rate": 8.799410386043016e-06,
      "loss": 0.7817,
      "step": 18639
    },
    {
      "epoch": 3.4460137520906895,
      "grad_norm": 0.9528607130050659,
      "learning_rate": 8.79341894240865e-06,
      "loss": 0.9257,
      "step": 18640
    },
    {
      "epoch": 3.446199591154061,
      "grad_norm": 0.969680666923523,
      "learning_rate": 8.787429445422246e-06,
      "loss": 0.9073,
      "step": 18641
    },
    {
      "epoch": 3.4463854302174317,
      "grad_norm": 0.8936437368392944,
      "learning_rate": 8.78144189521164e-06,
      "loss": 0.787,
      "step": 18642
    },
    {
      "epoch": 3.4465712692808026,
      "grad_norm": 0.7957565188407898,
      "learning_rate": 8.775456291904615e-06,
      "loss": 0.7421,
      "step": 18643
    },
    {
      "epoch": 3.446757108344174,
      "grad_norm": 0.9845880270004272,
      "learning_rate": 8.769472635628905e-06,
      "loss": 0.8581,
      "step": 18644
    },
    {
      "epoch": 3.4469429474075453,
      "grad_norm": 1.0251277685165405,
      "learning_rate": 8.763490926512274e-06,
      "loss": 0.7913,
      "step": 18645
    },
    {
      "epoch": 3.447128786470916,
      "grad_norm": 1.0311559438705444,
      "learning_rate": 8.757511164682353e-06,
      "loss": 0.8396,
      "step": 18646
    },
    {
      "epoch": 3.4473146255342875,
      "grad_norm": 1.0363506078720093,
      "learning_rate": 8.751533350266761e-06,
      "loss": 0.8461,
      "step": 18647
    },
    {
      "epoch": 3.4475004645976584,
      "grad_norm": 1.0235755443572998,
      "learning_rate": 8.745557483393118e-06,
      "loss": 0.8727,
      "step": 18648
    },
    {
      "epoch": 3.4476863036610297,
      "grad_norm": 0.8560394644737244,
      "learning_rate": 8.739583564188958e-06,
      "loss": 0.613,
      "step": 18649
    },
    {
      "epoch": 3.4478721427244006,
      "grad_norm": 0.9190382957458496,
      "learning_rate": 8.733611592781755e-06,
      "loss": 0.9174,
      "step": 18650
    },
    {
      "epoch": 3.448057981787772,
      "grad_norm": 1.0788168907165527,
      "learning_rate": 8.727641569299006e-06,
      "loss": 0.8025,
      "step": 18651
    },
    {
      "epoch": 3.448243820851143,
      "grad_norm": 0.9430216550827026,
      "learning_rate": 8.72167349386811e-06,
      "loss": 0.777,
      "step": 18652
    },
    {
      "epoch": 3.448429659914514,
      "grad_norm": 0.917182207107544,
      "learning_rate": 8.715707366616466e-06,
      "loss": 0.9633,
      "step": 18653
    },
    {
      "epoch": 3.448615498977885,
      "grad_norm": 0.9732248783111572,
      "learning_rate": 8.709743187671393e-06,
      "loss": 0.6324,
      "step": 18654
    },
    {
      "epoch": 3.4488013380412563,
      "grad_norm": 0.9532489776611328,
      "learning_rate": 8.70378095716018e-06,
      "loss": 0.7928,
      "step": 18655
    },
    {
      "epoch": 3.448987177104627,
      "grad_norm": 0.8786229491233826,
      "learning_rate": 8.697820675210111e-06,
      "loss": 0.6679,
      "step": 18656
    },
    {
      "epoch": 3.4491730161679985,
      "grad_norm": 1.0036118030548096,
      "learning_rate": 8.691862341948376e-06,
      "loss": 0.8445,
      "step": 18657
    },
    {
      "epoch": 3.4493588552313694,
      "grad_norm": 1.0232371091842651,
      "learning_rate": 8.685905957502149e-06,
      "loss": 1.0843,
      "step": 18658
    },
    {
      "epoch": 3.4495446942947408,
      "grad_norm": 1.1256111860275269,
      "learning_rate": 8.679951521998542e-06,
      "loss": 0.8726,
      "step": 18659
    },
    {
      "epoch": 3.449730533358112,
      "grad_norm": 0.9262232184410095,
      "learning_rate": 8.673999035564673e-06,
      "loss": 0.9612,
      "step": 18660
    },
    {
      "epoch": 3.449916372421483,
      "grad_norm": 0.8881259560585022,
      "learning_rate": 8.668048498327552e-06,
      "loss": 0.7052,
      "step": 18661
    },
    {
      "epoch": 3.4501022114848543,
      "grad_norm": 0.9636815190315247,
      "learning_rate": 8.662099910414222e-06,
      "loss": 0.9367,
      "step": 18662
    },
    {
      "epoch": 3.450288050548225,
      "grad_norm": 1.1980029344558716,
      "learning_rate": 8.656153271951639e-06,
      "loss": 0.8395,
      "step": 18663
    },
    {
      "epoch": 3.4504738896115965,
      "grad_norm": 0.8954828381538391,
      "learning_rate": 8.650208583066689e-06,
      "loss": 0.8351,
      "step": 18664
    },
    {
      "epoch": 3.4506597286749674,
      "grad_norm": 1.0028259754180908,
      "learning_rate": 8.64426584388629e-06,
      "loss": 0.777,
      "step": 18665
    },
    {
      "epoch": 3.4508455677383387,
      "grad_norm": 0.953440248966217,
      "learning_rate": 8.638325054537255e-06,
      "loss": 0.7337,
      "step": 18666
    },
    {
      "epoch": 3.4510314068017096,
      "grad_norm": 0.9255670309066772,
      "learning_rate": 8.632386215146381e-06,
      "loss": 0.8076,
      "step": 18667
    },
    {
      "epoch": 3.451217245865081,
      "grad_norm": 0.9749565720558167,
      "learning_rate": 8.626449325840447e-06,
      "loss": 0.8783,
      "step": 18668
    },
    {
      "epoch": 3.451403084928452,
      "grad_norm": 0.9131873846054077,
      "learning_rate": 8.620514386746148e-06,
      "loss": 0.723,
      "step": 18669
    },
    {
      "epoch": 3.451588923991823,
      "grad_norm": 0.960011899471283,
      "learning_rate": 8.614581397990162e-06,
      "loss": 0.9202,
      "step": 18670
    },
    {
      "epoch": 3.451774763055194,
      "grad_norm": 1.0307480096817017,
      "learning_rate": 8.608650359699088e-06,
      "loss": 0.7317,
      "step": 18671
    },
    {
      "epoch": 3.4519606021185654,
      "grad_norm": 0.8767074346542358,
      "learning_rate": 8.602721271999536e-06,
      "loss": 1.0208,
      "step": 18672
    },
    {
      "epoch": 3.4521464411819363,
      "grad_norm": 0.9889634847640991,
      "learning_rate": 8.59679413501807e-06,
      "loss": 0.7734,
      "step": 18673
    },
    {
      "epoch": 3.4523322802453076,
      "grad_norm": 1.1916424036026,
      "learning_rate": 8.590868948881192e-06,
      "loss": 1.0509,
      "step": 18674
    },
    {
      "epoch": 3.452518119308679,
      "grad_norm": 1.2274255752563477,
      "learning_rate": 8.58494571371532e-06,
      "loss": 0.8215,
      "step": 18675
    },
    {
      "epoch": 3.45270395837205,
      "grad_norm": 1.0868438482284546,
      "learning_rate": 8.579024429646932e-06,
      "loss": 0.9448,
      "step": 18676
    },
    {
      "epoch": 3.4528897974354207,
      "grad_norm": 1.3764984607696533,
      "learning_rate": 8.57310509680238e-06,
      "loss": 1.0186,
      "step": 18677
    },
    {
      "epoch": 3.453075636498792,
      "grad_norm": 0.9937775731086731,
      "learning_rate": 8.567187715308012e-06,
      "loss": 1.0688,
      "step": 18678
    },
    {
      "epoch": 3.4532614755621633,
      "grad_norm": 1.1719824075698853,
      "learning_rate": 8.561272285290101e-06,
      "loss": 0.7771,
      "step": 18679
    },
    {
      "epoch": 3.4534473146255342,
      "grad_norm": 1.0099252462387085,
      "learning_rate": 8.555358806874925e-06,
      "loss": 0.6463,
      "step": 18680
    },
    {
      "epoch": 3.4536331536889056,
      "grad_norm": 1.1827915906906128,
      "learning_rate": 8.549447280188694e-06,
      "loss": 0.693,
      "step": 18681
    },
    {
      "epoch": 3.4538189927522764,
      "grad_norm": 0.9828644394874573,
      "learning_rate": 8.543537705357563e-06,
      "loss": 0.8288,
      "step": 18682
    },
    {
      "epoch": 3.4540048318156478,
      "grad_norm": 0.9095243811607361,
      "learning_rate": 8.537630082507687e-06,
      "loss": 0.7863,
      "step": 18683
    },
    {
      "epoch": 3.4541906708790187,
      "grad_norm": 1.0734424591064453,
      "learning_rate": 8.531724411765119e-06,
      "loss": 0.8124,
      "step": 18684
    },
    {
      "epoch": 3.45437650994239,
      "grad_norm": 1.265278697013855,
      "learning_rate": 8.525820693255949e-06,
      "loss": 1.0153,
      "step": 18685
    },
    {
      "epoch": 3.454562349005761,
      "grad_norm": 0.8495627045631409,
      "learning_rate": 8.519918927106164e-06,
      "loss": 0.6581,
      "step": 18686
    },
    {
      "epoch": 3.454748188069132,
      "grad_norm": 1.0464892387390137,
      "learning_rate": 8.514019113441695e-06,
      "loss": 0.894,
      "step": 18687
    },
    {
      "epoch": 3.454934027132503,
      "grad_norm": 0.9293912053108215,
      "learning_rate": 8.508121252388523e-06,
      "loss": 0.8371,
      "step": 18688
    },
    {
      "epoch": 3.4551198661958744,
      "grad_norm": 0.9331597685813904,
      "learning_rate": 8.50222534407249e-06,
      "loss": 0.8066,
      "step": 18689
    },
    {
      "epoch": 3.4553057052592457,
      "grad_norm": 0.7989200353622437,
      "learning_rate": 8.496331388619438e-06,
      "loss": 0.83,
      "step": 18690
    },
    {
      "epoch": 3.4554915443226166,
      "grad_norm": 0.8642352819442749,
      "learning_rate": 8.490439386155147e-06,
      "loss": 0.7559,
      "step": 18691
    },
    {
      "epoch": 3.4556773833859875,
      "grad_norm": 0.9246635437011719,
      "learning_rate": 8.484549336805403e-06,
      "loss": 0.7547,
      "step": 18692
    },
    {
      "epoch": 3.455863222449359,
      "grad_norm": 0.8502155542373657,
      "learning_rate": 8.478661240695917e-06,
      "loss": 0.6164,
      "step": 18693
    },
    {
      "epoch": 3.45604906151273,
      "grad_norm": 1.032968521118164,
      "learning_rate": 8.472775097952346e-06,
      "loss": 0.8578,
      "step": 18694
    },
    {
      "epoch": 3.456234900576101,
      "grad_norm": 0.8894579410552979,
      "learning_rate": 8.46689090870031e-06,
      "loss": 0.9053,
      "step": 18695
    },
    {
      "epoch": 3.4564207396394724,
      "grad_norm": 0.8251888751983643,
      "learning_rate": 8.461008673065429e-06,
      "loss": 0.7271,
      "step": 18696
    },
    {
      "epoch": 3.4566065787028433,
      "grad_norm": 0.974197506904602,
      "learning_rate": 8.455128391173228e-06,
      "loss": 0.9299,
      "step": 18697
    },
    {
      "epoch": 3.4567924177662146,
      "grad_norm": 0.6621809601783752,
      "learning_rate": 8.449250063149217e-06,
      "loss": 0.2961,
      "step": 18698
    },
    {
      "epoch": 3.4569782568295855,
      "grad_norm": 0.9626237750053406,
      "learning_rate": 8.443373689118827e-06,
      "loss": 1.0561,
      "step": 18699
    },
    {
      "epoch": 3.457164095892957,
      "grad_norm": 0.8487881422042847,
      "learning_rate": 8.437499269207539e-06,
      "loss": 0.781,
      "step": 18700
    },
    {
      "epoch": 3.4573499349563277,
      "grad_norm": 1.360926628112793,
      "learning_rate": 8.431626803540694e-06,
      "loss": 1.249,
      "step": 18701
    },
    {
      "epoch": 3.457535774019699,
      "grad_norm": 0.9287676811218262,
      "learning_rate": 8.425756292243647e-06,
      "loss": 0.9681,
      "step": 18702
    },
    {
      "epoch": 3.45772161308307,
      "grad_norm": 1.0354424715042114,
      "learning_rate": 8.419887735441633e-06,
      "loss": 0.9981,
      "step": 18703
    },
    {
      "epoch": 3.4579074521464412,
      "grad_norm": 0.9920165538787842,
      "learning_rate": 8.414021133260007e-06,
      "loss": 0.7668,
      "step": 18704
    },
    {
      "epoch": 3.458093291209812,
      "grad_norm": 0.9180377125740051,
      "learning_rate": 8.408156485823926e-06,
      "loss": 0.7742,
      "step": 18705
    },
    {
      "epoch": 3.4582791302731835,
      "grad_norm": 1.3005715608596802,
      "learning_rate": 8.402293793258576e-06,
      "loss": 0.9009,
      "step": 18706
    },
    {
      "epoch": 3.4584649693365543,
      "grad_norm": 0.9481584429740906,
      "learning_rate": 8.396433055689045e-06,
      "loss": 0.9433,
      "step": 18707
    },
    {
      "epoch": 3.4586508083999257,
      "grad_norm": 1.0840070247650146,
      "learning_rate": 8.390574273240482e-06,
      "loss": 0.8461,
      "step": 18708
    },
    {
      "epoch": 3.458836647463297,
      "grad_norm": 0.9308413863182068,
      "learning_rate": 8.384717446037893e-06,
      "loss": 0.8624,
      "step": 18709
    },
    {
      "epoch": 3.459022486526668,
      "grad_norm": 0.9320765137672424,
      "learning_rate": 8.378862574206303e-06,
      "loss": 0.814,
      "step": 18710
    },
    {
      "epoch": 3.459208325590039,
      "grad_norm": 0.953338086605072,
      "learning_rate": 8.373009657870635e-06,
      "loss": 0.7742,
      "step": 18711
    },
    {
      "epoch": 3.45939416465341,
      "grad_norm": 0.8855268359184265,
      "learning_rate": 8.367158697155863e-06,
      "loss": 0.8255,
      "step": 18712
    },
    {
      "epoch": 3.4595800037167814,
      "grad_norm": 0.9158693552017212,
      "learning_rate": 8.361309692186814e-06,
      "loss": 0.982,
      "step": 18713
    },
    {
      "epoch": 3.4597658427801523,
      "grad_norm": 0.9135979413986206,
      "learning_rate": 8.355462643088373e-06,
      "loss": 0.8785,
      "step": 18714
    },
    {
      "epoch": 3.4599516818435236,
      "grad_norm": 0.9471499919891357,
      "learning_rate": 8.3496175499853e-06,
      "loss": 0.9211,
      "step": 18715
    },
    {
      "epoch": 3.4601375209068945,
      "grad_norm": 0.8170753717422485,
      "learning_rate": 8.343774413002381e-06,
      "loss": 0.6915,
      "step": 18716
    },
    {
      "epoch": 3.460323359970266,
      "grad_norm": 0.8711549639701843,
      "learning_rate": 8.337933232264305e-06,
      "loss": 0.7263,
      "step": 18717
    },
    {
      "epoch": 3.4605091990336367,
      "grad_norm": 0.8051249980926514,
      "learning_rate": 8.332094007895741e-06,
      "loss": 0.6193,
      "step": 18718
    },
    {
      "epoch": 3.460695038097008,
      "grad_norm": 0.9760210514068604,
      "learning_rate": 8.32625674002131e-06,
      "loss": 0.8647,
      "step": 18719
    },
    {
      "epoch": 3.460880877160379,
      "grad_norm": 0.9609522223472595,
      "learning_rate": 8.320421428765635e-06,
      "loss": 0.7761,
      "step": 18720
    },
    {
      "epoch": 3.4610667162237503,
      "grad_norm": 1.2400734424591064,
      "learning_rate": 8.314588074253226e-06,
      "loss": 0.635,
      "step": 18721
    },
    {
      "epoch": 3.461252555287121,
      "grad_norm": Infinity,
      "learning_rate": 8.314588074253226e-06,
      "loss": 1.1963,
      "step": 18722
    },
    {
      "epoch": 3.4614383943504925,
      "grad_norm": 1.0367873907089233,
      "learning_rate": 8.308756676608597e-06,
      "loss": 0.9893,
      "step": 18723
    },
    {
      "epoch": 3.461624233413864,
      "grad_norm": 1.1819725036621094,
      "learning_rate": 8.302927235956193e-06,
      "loss": 0.7899,
      "step": 18724
    },
    {
      "epoch": 3.4618100724772347,
      "grad_norm": 0.9547193050384521,
      "learning_rate": 8.297099752420446e-06,
      "loss": 0.8314,
      "step": 18725
    },
    {
      "epoch": 3.4619959115406056,
      "grad_norm": 1.0257132053375244,
      "learning_rate": 8.291274226125756e-06,
      "loss": 0.8587,
      "step": 18726
    },
    {
      "epoch": 3.462181750603977,
      "grad_norm": 0.8922994136810303,
      "learning_rate": 8.285450657196437e-06,
      "loss": 0.8294,
      "step": 18727
    },
    {
      "epoch": 3.4623675896673483,
      "grad_norm": 0.8677076697349548,
      "learning_rate": 8.279629045756764e-06,
      "loss": 0.7083,
      "step": 18728
    },
    {
      "epoch": 3.462553428730719,
      "grad_norm": 0.9443361759185791,
      "learning_rate": 8.273809391931042e-06,
      "loss": 0.9847,
      "step": 18729
    },
    {
      "epoch": 3.4627392677940905,
      "grad_norm": 0.8874073028564453,
      "learning_rate": 8.267991695843436e-06,
      "loss": 0.5835,
      "step": 18730
    },
    {
      "epoch": 3.4629251068574614,
      "grad_norm": 1.1664199829101562,
      "learning_rate": 8.262175957618135e-06,
      "loss": 0.9179,
      "step": 18731
    },
    {
      "epoch": 3.4631109459208327,
      "grad_norm": 0.8360967040061951,
      "learning_rate": 8.25636217737924e-06,
      "loss": 0.7487,
      "step": 18732
    },
    {
      "epoch": 3.4632967849842036,
      "grad_norm": 1.129758596420288,
      "learning_rate": 8.250550355250875e-06,
      "loss": 0.9754,
      "step": 18733
    },
    {
      "epoch": 3.463482624047575,
      "grad_norm": 1.4248270988464355,
      "learning_rate": 8.244740491357062e-06,
      "loss": 1.1269,
      "step": 18734
    },
    {
      "epoch": 3.463668463110946,
      "grad_norm": 0.9525567889213562,
      "learning_rate": 8.23893258582179e-06,
      "loss": 0.9248,
      "step": 18735
    },
    {
      "epoch": 3.463854302174317,
      "grad_norm": 1.058853030204773,
      "learning_rate": 8.23312663876904e-06,
      "loss": 0.7135,
      "step": 18736
    },
    {
      "epoch": 3.464040141237688,
      "grad_norm": 0.9106456637382507,
      "learning_rate": 8.227322650322734e-06,
      "loss": 0.6708,
      "step": 18737
    },
    {
      "epoch": 3.4642259803010593,
      "grad_norm": 0.8357923030853271,
      "learning_rate": 8.221520620606738e-06,
      "loss": 0.8337,
      "step": 18738
    },
    {
      "epoch": 3.46441181936443,
      "grad_norm": 1.2621804475784302,
      "learning_rate": 8.215720549744887e-06,
      "loss": 0.939,
      "step": 18739
    },
    {
      "epoch": 3.4645976584278015,
      "grad_norm": 0.8909697532653809,
      "learning_rate": 8.20992243786095e-06,
      "loss": 0.7573,
      "step": 18740
    },
    {
      "epoch": 3.4647834974911724,
      "grad_norm": 0.9684772491455078,
      "learning_rate": 8.204126285078728e-06,
      "loss": 0.8583,
      "step": 18741
    },
    {
      "epoch": 3.4649693365545438,
      "grad_norm": 1.0464305877685547,
      "learning_rate": 8.198332091521898e-06,
      "loss": 0.8594,
      "step": 18742
    },
    {
      "epoch": 3.465155175617915,
      "grad_norm": 0.8310005068778992,
      "learning_rate": 8.19253985731413e-06,
      "loss": 0.8416,
      "step": 18743
    },
    {
      "epoch": 3.465341014681286,
      "grad_norm": 0.8563743829727173,
      "learning_rate": 8.186749582579034e-06,
      "loss": 0.8199,
      "step": 18744
    },
    {
      "epoch": 3.4655268537446573,
      "grad_norm": 0.8466929197311401,
      "learning_rate": 8.180961267440223e-06,
      "loss": 0.8748,
      "step": 18745
    },
    {
      "epoch": 3.465712692808028,
      "grad_norm": 0.9952386617660522,
      "learning_rate": 8.175174912021211e-06,
      "loss": 0.7719,
      "step": 18746
    },
    {
      "epoch": 3.4658985318713995,
      "grad_norm": 0.9680238366127014,
      "learning_rate": 8.16939051644554e-06,
      "loss": 0.8767,
      "step": 18747
    },
    {
      "epoch": 3.4660843709347704,
      "grad_norm": 0.9317744374275208,
      "learning_rate": 8.163608080836616e-06,
      "loss": 0.9406,
      "step": 18748
    },
    {
      "epoch": 3.4662702099981417,
      "grad_norm": 0.9056304693222046,
      "learning_rate": 8.157827605317892e-06,
      "loss": 0.7743,
      "step": 18749
    },
    {
      "epoch": 3.4664560490615126,
      "grad_norm": 0.8400526642799377,
      "learning_rate": 8.152049090012736e-06,
      "loss": 0.89,
      "step": 18750
    },
    {
      "epoch": 3.466641888124884,
      "grad_norm": 0.905646026134491,
      "learning_rate": 8.146272535044475e-06,
      "loss": 1.0205,
      "step": 18751
    },
    {
      "epoch": 3.466827727188255,
      "grad_norm": 0.8235187530517578,
      "learning_rate": 8.140497940536385e-06,
      "loss": 0.5697,
      "step": 18752
    },
    {
      "epoch": 3.467013566251626,
      "grad_norm": 1.1538528203964233,
      "learning_rate": 8.134725306611734e-06,
      "loss": 0.9183,
      "step": 18753
    },
    {
      "epoch": 3.467199405314997,
      "grad_norm": 0.8710090517997742,
      "learning_rate": 8.128954633393737e-06,
      "loss": 0.8054,
      "step": 18754
    },
    {
      "epoch": 3.4673852443783684,
      "grad_norm": 1.0354934930801392,
      "learning_rate": 8.123185921005538e-06,
      "loss": 0.9725,
      "step": 18755
    },
    {
      "epoch": 3.4675710834417393,
      "grad_norm": 1.0618163347244263,
      "learning_rate": 8.11741916957024e-06,
      "loss": 0.973,
      "step": 18756
    },
    {
      "epoch": 3.4677569225051106,
      "grad_norm": 0.6448819637298584,
      "learning_rate": 8.111654379210987e-06,
      "loss": 0.5416,
      "step": 18757
    },
    {
      "epoch": 3.467942761568482,
      "grad_norm": 1.0582246780395508,
      "learning_rate": 8.105891550050782e-06,
      "loss": 0.9356,
      "step": 18758
    },
    {
      "epoch": 3.468128600631853,
      "grad_norm": 0.9536944031715393,
      "learning_rate": 8.100130682212625e-06,
      "loss": 0.7974,
      "step": 18759
    },
    {
      "epoch": 3.468314439695224,
      "grad_norm": 0.9796311259269714,
      "learning_rate": 8.094371775819464e-06,
      "loss": 0.7147,
      "step": 18760
    },
    {
      "epoch": 3.468500278758595,
      "grad_norm": 0.9044296145439148,
      "learning_rate": 8.088614830994223e-06,
      "loss": 0.8041,
      "step": 18761
    },
    {
      "epoch": 3.4686861178219663,
      "grad_norm": 0.8856521844863892,
      "learning_rate": 8.08285984785978e-06,
      "loss": 0.8417,
      "step": 18762
    },
    {
      "epoch": 3.4688719568853372,
      "grad_norm": 0.8639485239982605,
      "learning_rate": 8.07710682653896e-06,
      "loss": 0.6279,
      "step": 18763
    },
    {
      "epoch": 3.4690577959487086,
      "grad_norm": 0.8745250701904297,
      "learning_rate": 8.07135576715453e-06,
      "loss": 0.8151,
      "step": 18764
    },
    {
      "epoch": 3.4692436350120794,
      "grad_norm": 0.9135953187942505,
      "learning_rate": 8.065606669829273e-06,
      "loss": 0.827,
      "step": 18765
    },
    {
      "epoch": 3.4694294740754508,
      "grad_norm": 0.9876458048820496,
      "learning_rate": 8.059859534685877e-06,
      "loss": 0.7074,
      "step": 18766
    },
    {
      "epoch": 3.4696153131388217,
      "grad_norm": 0.9676094055175781,
      "learning_rate": 8.054114361846988e-06,
      "loss": 0.9162,
      "step": 18767
    },
    {
      "epoch": 3.469801152202193,
      "grad_norm": 1.0216379165649414,
      "learning_rate": 8.04837115143523e-06,
      "loss": 0.6916,
      "step": 18768
    },
    {
      "epoch": 3.469986991265564,
      "grad_norm": 1.0474780797958374,
      "learning_rate": 8.042629903573217e-06,
      "loss": 0.7252,
      "step": 18769
    },
    {
      "epoch": 3.470172830328935,
      "grad_norm": 1.016576886177063,
      "learning_rate": 8.036890618383474e-06,
      "loss": 1.1245,
      "step": 18770
    },
    {
      "epoch": 3.470358669392306,
      "grad_norm": 1.0590362548828125,
      "learning_rate": 8.03115329598848e-06,
      "loss": 0.8428,
      "step": 18771
    },
    {
      "epoch": 3.4705445084556774,
      "grad_norm": 0.9816322922706604,
      "learning_rate": 8.025417936510671e-06,
      "loss": 0.8735,
      "step": 18772
    },
    {
      "epoch": 3.4707303475190487,
      "grad_norm": 0.9632866382598877,
      "learning_rate": 8.019684540072503e-06,
      "loss": 0.9617,
      "step": 18773
    },
    {
      "epoch": 3.4709161865824196,
      "grad_norm": 0.9751099944114685,
      "learning_rate": 8.013953106796324e-06,
      "loss": 0.7521,
      "step": 18774
    },
    {
      "epoch": 3.4711020256457905,
      "grad_norm": 0.8639714121818542,
      "learning_rate": 8.008223636804457e-06,
      "loss": 0.8057,
      "step": 18775
    },
    {
      "epoch": 3.471287864709162,
      "grad_norm": 1.7861194610595703,
      "learning_rate": 8.002496130219184e-06,
      "loss": 0.6409,
      "step": 18776
    },
    {
      "epoch": 3.471473703772533,
      "grad_norm": 0.8952376842498779,
      "learning_rate": 7.99677058716275e-06,
      "loss": 0.7854,
      "step": 18777
    },
    {
      "epoch": 3.471659542835904,
      "grad_norm": 0.9320510625839233,
      "learning_rate": 7.991047007757379e-06,
      "loss": 0.7999,
      "step": 18778
    },
    {
      "epoch": 3.4718453818992754,
      "grad_norm": 1.1861048936843872,
      "learning_rate": 7.98532539212522e-06,
      "loss": 0.9218,
      "step": 18779
    },
    {
      "epoch": 3.4720312209626463,
      "grad_norm": 0.8894674777984619,
      "learning_rate": 7.979605740388375e-06,
      "loss": 0.7517,
      "step": 18780
    },
    {
      "epoch": 3.4722170600260176,
      "grad_norm": 1.066564679145813,
      "learning_rate": 7.973888052668943e-06,
      "loss": 0.9006,
      "step": 18781
    },
    {
      "epoch": 3.4724028990893885,
      "grad_norm": 0.9083952307701111,
      "learning_rate": 7.968172329088952e-06,
      "loss": 0.8833,
      "step": 18782
    },
    {
      "epoch": 3.47258873815276,
      "grad_norm": 0.8635721206665039,
      "learning_rate": 7.962458569770382e-06,
      "loss": 0.896,
      "step": 18783
    },
    {
      "epoch": 3.4727745772161307,
      "grad_norm": 0.8096431493759155,
      "learning_rate": 7.956746774835178e-06,
      "loss": 0.7672,
      "step": 18784
    },
    {
      "epoch": 3.472960416279502,
      "grad_norm": 0.7971764802932739,
      "learning_rate": 7.951036944405287e-06,
      "loss": 0.7884,
      "step": 18785
    },
    {
      "epoch": 3.473146255342873,
      "grad_norm": 1.084967851638794,
      "learning_rate": 7.945329078602538e-06,
      "loss": 0.9731,
      "step": 18786
    },
    {
      "epoch": 3.4733320944062442,
      "grad_norm": 0.8574565052986145,
      "learning_rate": 7.93962317754876e-06,
      "loss": 0.7691,
      "step": 18787
    },
    {
      "epoch": 3.473517933469615,
      "grad_norm": 1.0025873184204102,
      "learning_rate": 7.933919241365729e-06,
      "loss": 0.8392,
      "step": 18788
    },
    {
      "epoch": 3.4737037725329865,
      "grad_norm": 1.0289613008499146,
      "learning_rate": 7.92821727017523e-06,
      "loss": 0.7774,
      "step": 18789
    },
    {
      "epoch": 3.4738896115963573,
      "grad_norm": 0.9610695838928223,
      "learning_rate": 7.922517264098927e-06,
      "loss": 0.7482,
      "step": 18790
    },
    {
      "epoch": 3.4740754506597287,
      "grad_norm": 0.8945107460021973,
      "learning_rate": 7.916819223258487e-06,
      "loss": 0.8455,
      "step": 18791
    },
    {
      "epoch": 3.4742612897231,
      "grad_norm": 0.8954695463180542,
      "learning_rate": 7.911123147775502e-06,
      "loss": 0.8472,
      "step": 18792
    },
    {
      "epoch": 3.474447128786471,
      "grad_norm": 1.206834077835083,
      "learning_rate": 7.905429037771572e-06,
      "loss": 0.8113,
      "step": 18793
    },
    {
      "epoch": 3.474632967849842,
      "grad_norm": 0.7065415978431702,
      "learning_rate": 7.899736893368226e-06,
      "loss": 0.4913,
      "step": 18794
    },
    {
      "epoch": 3.474818806913213,
      "grad_norm": 0.9732913374900818,
      "learning_rate": 7.894046714686943e-06,
      "loss": 0.6407,
      "step": 18795
    },
    {
      "epoch": 3.4750046459765844,
      "grad_norm": 0.9556186199188232,
      "learning_rate": 7.888358501849146e-06,
      "loss": 0.8798,
      "step": 18796
    },
    {
      "epoch": 3.4751904850399553,
      "grad_norm": 1.121410608291626,
      "learning_rate": 7.882672254976297e-06,
      "loss": 0.8895,
      "step": 18797
    },
    {
      "epoch": 3.4753763241033266,
      "grad_norm": 0.926697850227356,
      "learning_rate": 7.876987974189708e-06,
      "loss": 0.9799,
      "step": 18798
    },
    {
      "epoch": 3.4755621631666975,
      "grad_norm": 1.3014369010925293,
      "learning_rate": 7.871305659610739e-06,
      "loss": 0.9644,
      "step": 18799
    },
    {
      "epoch": 3.475748002230069,
      "grad_norm": 0.9880854487419128,
      "learning_rate": 7.865625311360636e-06,
      "loss": 0.8778,
      "step": 18800
    },
    {
      "epoch": 3.4759338412934397,
      "grad_norm": 0.7500266432762146,
      "learning_rate": 7.859946929560658e-06,
      "loss": 0.6615,
      "step": 18801
    },
    {
      "epoch": 3.476119680356811,
      "grad_norm": 0.9513152837753296,
      "learning_rate": 7.854270514332007e-06,
      "loss": 0.8724,
      "step": 18802
    },
    {
      "epoch": 3.476305519420182,
      "grad_norm": 0.945457935333252,
      "learning_rate": 7.84859606579582e-06,
      "loss": 0.821,
      "step": 18803
    },
    {
      "epoch": 3.4764913584835533,
      "grad_norm": 1.0324273109436035,
      "learning_rate": 7.842923584073192e-06,
      "loss": 0.8328,
      "step": 18804
    },
    {
      "epoch": 3.476677197546924,
      "grad_norm": 1.029531478881836,
      "learning_rate": 7.837253069285234e-06,
      "loss": 0.9378,
      "step": 18805
    },
    {
      "epoch": 3.4768630366102955,
      "grad_norm": 0.9380819201469421,
      "learning_rate": 7.831584521552949e-06,
      "loss": 0.8479,
      "step": 18806
    },
    {
      "epoch": 3.477048875673667,
      "grad_norm": 0.8309686779975891,
      "learning_rate": 7.825917940997308e-06,
      "loss": 0.7903,
      "step": 18807
    },
    {
      "epoch": 3.4772347147370377,
      "grad_norm": 1.0270386934280396,
      "learning_rate": 7.82025332773928e-06,
      "loss": 0.7759,
      "step": 18808
    },
    {
      "epoch": 3.4774205538004086,
      "grad_norm": 1.0845005512237549,
      "learning_rate": 7.814590681899758e-06,
      "loss": 0.897,
      "step": 18809
    },
    {
      "epoch": 3.47760639286378,
      "grad_norm": 0.9255852103233337,
      "learning_rate": 7.808930003599601e-06,
      "loss": 0.8083,
      "step": 18810
    },
    {
      "epoch": 3.4777922319271513,
      "grad_norm": 0.8624998927116394,
      "learning_rate": 7.803271292959636e-06,
      "loss": 0.6886,
      "step": 18811
    },
    {
      "epoch": 3.477978070990522,
      "grad_norm": 0.8667799830436707,
      "learning_rate": 7.797614550100618e-06,
      "loss": 0.7746,
      "step": 18812
    },
    {
      "epoch": 3.4781639100538935,
      "grad_norm": 1.1160162687301636,
      "learning_rate": 7.791959775143298e-06,
      "loss": 0.8162,
      "step": 18813
    },
    {
      "epoch": 3.4783497491172644,
      "grad_norm": 0.8530917167663574,
      "learning_rate": 7.786306968208368e-06,
      "loss": 0.552,
      "step": 18814
    },
    {
      "epoch": 3.4785355881806357,
      "grad_norm": 0.7741609811782837,
      "learning_rate": 7.780656129416464e-06,
      "loss": 0.8718,
      "step": 18815
    },
    {
      "epoch": 3.4787214272440066,
      "grad_norm": 0.9647586941719055,
      "learning_rate": 7.77500725888819e-06,
      "loss": 0.6975,
      "step": 18816
    },
    {
      "epoch": 3.478907266307378,
      "grad_norm": 0.8000401258468628,
      "learning_rate": 7.769360356744137e-06,
      "loss": 0.7548,
      "step": 18817
    },
    {
      "epoch": 3.479093105370749,
      "grad_norm": 1.4719072580337524,
      "learning_rate": 7.763715423104811e-06,
      "loss": 1.1186,
      "step": 18818
    },
    {
      "epoch": 3.47927894443412,
      "grad_norm": 0.9751482605934143,
      "learning_rate": 7.758072458090681e-06,
      "loss": 0.8613,
      "step": 18819
    },
    {
      "epoch": 3.479464783497491,
      "grad_norm": 0.8374261856079102,
      "learning_rate": 7.752431461822196e-06,
      "loss": 0.7657,
      "step": 18820
    },
    {
      "epoch": 3.4796506225608623,
      "grad_norm": 0.846748948097229,
      "learning_rate": 7.746792434419781e-06,
      "loss": 0.6049,
      "step": 18821
    },
    {
      "epoch": 3.4798364616242337,
      "grad_norm": 0.8745054602622986,
      "learning_rate": 7.741155376003773e-06,
      "loss": 0.8999,
      "step": 18822
    },
    {
      "epoch": 3.4800223006876045,
      "grad_norm": 0.958818256855011,
      "learning_rate": 7.735520286694475e-06,
      "loss": 0.8278,
      "step": 18823
    },
    {
      "epoch": 3.4802081397509754,
      "grad_norm": 0.9613667726516724,
      "learning_rate": 7.729887166612138e-06,
      "loss": 0.7098,
      "step": 18824
    },
    {
      "epoch": 3.4803939788143468,
      "grad_norm": 0.9327418208122253,
      "learning_rate": 7.724256015877051e-06,
      "loss": 0.8451,
      "step": 18825
    },
    {
      "epoch": 3.480579817877718,
      "grad_norm": 0.917344868183136,
      "learning_rate": 7.718626834609354e-06,
      "loss": 0.9286,
      "step": 18826
    },
    {
      "epoch": 3.480765656941089,
      "grad_norm": 0.8673303127288818,
      "learning_rate": 7.712999622929196e-06,
      "loss": 0.4841,
      "step": 18827
    },
    {
      "epoch": 3.4809514960044603,
      "grad_norm": 0.8560457229614258,
      "learning_rate": 7.707374380956711e-06,
      "loss": 0.6967,
      "step": 18828
    },
    {
      "epoch": 3.481137335067831,
      "grad_norm": 1.1645073890686035,
      "learning_rate": 7.701751108811938e-06,
      "loss": 0.753,
      "step": 18829
    },
    {
      "epoch": 3.4813231741312025,
      "grad_norm": 1.1489976644515991,
      "learning_rate": 7.696129806614871e-06,
      "loss": 0.8405,
      "step": 18830
    },
    {
      "epoch": 3.4815090131945734,
      "grad_norm": 0.9245211482048035,
      "learning_rate": 7.690510474485535e-06,
      "loss": 0.6869,
      "step": 18831
    },
    {
      "epoch": 3.4816948522579447,
      "grad_norm": 0.9011597633361816,
      "learning_rate": 7.684893112543834e-06,
      "loss": 0.8236,
      "step": 18832
    },
    {
      "epoch": 3.4818806913213156,
      "grad_norm": 1.0039126873016357,
      "learning_rate": 7.679277720909684e-06,
      "loss": 0.5301,
      "step": 18833
    },
    {
      "epoch": 3.482066530384687,
      "grad_norm": 0.9441542029380798,
      "learning_rate": 7.67366429970292e-06,
      "loss": 0.9177,
      "step": 18834
    },
    {
      "epoch": 3.482252369448058,
      "grad_norm": 0.8128690123558044,
      "learning_rate": 7.66805284904335e-06,
      "loss": 0.6621,
      "step": 18835
    },
    {
      "epoch": 3.482438208511429,
      "grad_norm": 0.8651259541511536,
      "learning_rate": 7.662443369050731e-06,
      "loss": 0.7271,
      "step": 18836
    },
    {
      "epoch": 3.4826240475748,
      "grad_norm": 0.8950327634811401,
      "learning_rate": 7.656835859844814e-06,
      "loss": 0.8193,
      "step": 18837
    },
    {
      "epoch": 3.4828098866381714,
      "grad_norm": 0.9409818649291992,
      "learning_rate": 7.651230321545278e-06,
      "loss": 0.868,
      "step": 18838
    },
    {
      "epoch": 3.4829957257015423,
      "grad_norm": 0.961976170539856,
      "learning_rate": 7.64562675427173e-06,
      "loss": 0.9764,
      "step": 18839
    },
    {
      "epoch": 3.4831815647649136,
      "grad_norm": 1.0851531028747559,
      "learning_rate": 7.640025158143782e-06,
      "loss": 0.6041,
      "step": 18840
    },
    {
      "epoch": 3.483367403828285,
      "grad_norm": 0.9040879011154175,
      "learning_rate": 7.634425533281031e-06,
      "loss": 0.8333,
      "step": 18841
    },
    {
      "epoch": 3.483553242891656,
      "grad_norm": 0.8291379809379578,
      "learning_rate": 7.628827879802958e-06,
      "loss": 0.749,
      "step": 18842
    },
    {
      "epoch": 3.483739081955027,
      "grad_norm": 0.9865771532058716,
      "learning_rate": 7.6232321978290444e-06,
      "loss": 0.9675,
      "step": 18843
    },
    {
      "epoch": 3.483924921018398,
      "grad_norm": 1.0167065858840942,
      "learning_rate": 7.617638487478696e-06,
      "loss": 0.9132,
      "step": 18844
    },
    {
      "epoch": 3.483924921018398,
      "eval_loss": 1.044190764427185,
      "eval_runtime": 23.1506,
      "eval_samples_per_second": 47.169,
      "eval_steps_per_second": 23.585,
      "step": 18844
    },
    {
      "epoch": 3.4841107600817693,
      "grad_norm": 1.0542774200439453,
      "learning_rate": 7.612046748871327e-06,
      "loss": 1.1246,
      "step": 18845
    },
    {
      "epoch": 3.4842965991451402,
      "grad_norm": 0.9042951464653015,
      "learning_rate": 7.606456982126287e-06,
      "loss": 0.8144,
      "step": 18846
    },
    {
      "epoch": 3.4844824382085116,
      "grad_norm": 1.0759397745132446,
      "learning_rate": 7.6008691873628465e-06,
      "loss": 0.7357,
      "step": 18847
    },
    {
      "epoch": 3.4846682772718824,
      "grad_norm": 0.8770076632499695,
      "learning_rate": 7.595283364700301e-06,
      "loss": 0.9929,
      "step": 18848
    },
    {
      "epoch": 3.4848541163352538,
      "grad_norm": 1.0161012411117554,
      "learning_rate": 7.589699514257875e-06,
      "loss": 0.7426,
      "step": 18849
    },
    {
      "epoch": 3.4850399553986247,
      "grad_norm": 0.9345296025276184,
      "learning_rate": 7.58411763615472e-06,
      "loss": 0.8618,
      "step": 18850
    },
    {
      "epoch": 3.485225794461996,
      "grad_norm": 0.8955548405647278,
      "learning_rate": 7.57853773050996e-06,
      "loss": 0.7533,
      "step": 18851
    },
    {
      "epoch": 3.485411633525367,
      "grad_norm": 0.8949218988418579,
      "learning_rate": 7.5729597974427246e-06,
      "loss": 0.6548,
      "step": 18852
    },
    {
      "epoch": 3.485597472588738,
      "grad_norm": 1.1832979917526245,
      "learning_rate": 7.567383837072062e-06,
      "loss": 0.9166,
      "step": 18853
    },
    {
      "epoch": 3.485783311652109,
      "grad_norm": 0.9716928601264954,
      "learning_rate": 7.561809849516965e-06,
      "loss": 0.8581,
      "step": 18854
    },
    {
      "epoch": 3.4859691507154804,
      "grad_norm": 0.9576780796051025,
      "learning_rate": 7.556237834896396e-06,
      "loss": 0.811,
      "step": 18855
    },
    {
      "epoch": 3.4861549897788517,
      "grad_norm": 0.9692073464393616,
      "learning_rate": 7.550667793329303e-06,
      "loss": 0.9054,
      "step": 18856
    },
    {
      "epoch": 3.4863408288422226,
      "grad_norm": 0.7223288416862488,
      "learning_rate": 7.545099724934557e-06,
      "loss": 0.5999,
      "step": 18857
    },
    {
      "epoch": 3.4865266679055935,
      "grad_norm": 1.0299596786499023,
      "learning_rate": 7.539533629831008e-06,
      "loss": 0.806,
      "step": 18858
    },
    {
      "epoch": 3.486712506968965,
      "grad_norm": 0.9294703006744385,
      "learning_rate": 7.533969508137418e-06,
      "loss": 0.945,
      "step": 18859
    },
    {
      "epoch": 3.486898346032336,
      "grad_norm": 0.9691776633262634,
      "learning_rate": 7.528407359972589e-06,
      "loss": 0.8303,
      "step": 18860
    },
    {
      "epoch": 3.487084185095707,
      "grad_norm": 0.9316863417625427,
      "learning_rate": 7.522847185455195e-06,
      "loss": 0.7473,
      "step": 18861
    },
    {
      "epoch": 3.4872700241590784,
      "grad_norm": 0.923642098903656,
      "learning_rate": 7.517288984703963e-06,
      "loss": 0.9339,
      "step": 18862
    },
    {
      "epoch": 3.4874558632224493,
      "grad_norm": 0.9262492656707764,
      "learning_rate": 7.511732757837476e-06,
      "loss": 0.811,
      "step": 18863
    },
    {
      "epoch": 3.4876417022858206,
      "grad_norm": 1.005344271659851,
      "learning_rate": 7.506178504974326e-06,
      "loss": 0.8925,
      "step": 18864
    },
    {
      "epoch": 3.4878275413491915,
      "grad_norm": 0.959369957447052,
      "learning_rate": 7.500626226233087e-06,
      "loss": 0.8635,
      "step": 18865
    },
    {
      "epoch": 3.488013380412563,
      "grad_norm": 1.0304522514343262,
      "learning_rate": 7.495075921732253e-06,
      "loss": 0.7596,
      "step": 18866
    },
    {
      "epoch": 3.4881992194759337,
      "grad_norm": 1.2714794874191284,
      "learning_rate": 7.4895275915902505e-06,
      "loss": 1.1942,
      "step": 18867
    },
    {
      "epoch": 3.488385058539305,
      "grad_norm": 1.0587759017944336,
      "learning_rate": 7.4839812359255525e-06,
      "loss": 0.994,
      "step": 18868
    },
    {
      "epoch": 3.488570897602676,
      "grad_norm": 1.0070648193359375,
      "learning_rate": 7.478436854856507e-06,
      "loss": 0.8278,
      "step": 18869
    },
    {
      "epoch": 3.4887567366660472,
      "grad_norm": 1.0866297483444214,
      "learning_rate": 7.472894448501455e-06,
      "loss": 1.0012,
      "step": 18870
    },
    {
      "epoch": 3.4889425757294186,
      "grad_norm": 0.9275298714637756,
      "learning_rate": 7.467354016978667e-06,
      "loss": 0.8685,
      "step": 18871
    },
    {
      "epoch": 3.4891284147927895,
      "grad_norm": 0.974448025226593,
      "learning_rate": 7.461815560406416e-06,
      "loss": 0.8151,
      "step": 18872
    },
    {
      "epoch": 3.4893142538561603,
      "grad_norm": 1.1058423519134521,
      "learning_rate": 7.456279078902928e-06,
      "loss": 0.8733,
      "step": 18873
    },
    {
      "epoch": 3.4895000929195317,
      "grad_norm": 0.964985191822052,
      "learning_rate": 7.450744572586354e-06,
      "loss": 0.6893,
      "step": 18874
    },
    {
      "epoch": 3.489685931982903,
      "grad_norm": 0.8219852447509766,
      "learning_rate": 7.4452120415748e-06,
      "loss": 0.6868,
      "step": 18875
    },
    {
      "epoch": 3.489871771046274,
      "grad_norm": 0.8574902415275574,
      "learning_rate": 7.439681485986383e-06,
      "loss": 0.7591,
      "step": 18876
    },
    {
      "epoch": 3.490057610109645,
      "grad_norm": 0.7997888326644897,
      "learning_rate": 7.434152905939118e-06,
      "loss": 0.8704,
      "step": 18877
    },
    {
      "epoch": 3.490243449173016,
      "grad_norm": 0.9168525338172913,
      "learning_rate": 7.428626301551012e-06,
      "loss": 1.0332,
      "step": 18878
    },
    {
      "epoch": 3.4904292882363874,
      "grad_norm": 0.937621533870697,
      "learning_rate": 7.423101672940003e-06,
      "loss": 0.7234,
      "step": 18879
    },
    {
      "epoch": 3.4906151272997583,
      "grad_norm": 0.8390883207321167,
      "learning_rate": 7.4175790202240415e-06,
      "loss": 0.7647,
      "step": 18880
    },
    {
      "epoch": 3.4908009663631296,
      "grad_norm": 0.9851588606834412,
      "learning_rate": 7.412058343520967e-06,
      "loss": 0.8092,
      "step": 18881
    },
    {
      "epoch": 3.4909868054265005,
      "grad_norm": 0.9515487551689148,
      "learning_rate": 7.406539642948606e-06,
      "loss": 0.8454,
      "step": 18882
    },
    {
      "epoch": 3.491172644489872,
      "grad_norm": 1.0104717016220093,
      "learning_rate": 7.40102291862479e-06,
      "loss": 0.8593,
      "step": 18883
    },
    {
      "epoch": 3.4913584835532427,
      "grad_norm": 0.7935540676116943,
      "learning_rate": 7.395508170667198e-06,
      "loss": 0.6452,
      "step": 18884
    },
    {
      "epoch": 3.491544322616614,
      "grad_norm": 0.9558344483375549,
      "learning_rate": 7.389995399193595e-06,
      "loss": 0.8742,
      "step": 18885
    },
    {
      "epoch": 3.491730161679985,
      "grad_norm": 0.9744873642921448,
      "learning_rate": 7.384484604321617e-06,
      "loss": 0.818,
      "step": 18886
    },
    {
      "epoch": 3.4919160007433563,
      "grad_norm": 1.1060720682144165,
      "learning_rate": 7.378975786168863e-06,
      "loss": 0.9289,
      "step": 18887
    },
    {
      "epoch": 3.492101839806727,
      "grad_norm": 0.9581206440925598,
      "learning_rate": 7.373468944852946e-06,
      "loss": 1.0307,
      "step": 18888
    },
    {
      "epoch": 3.4922876788700985,
      "grad_norm": 0.9559915661811829,
      "learning_rate": 7.367964080491374e-06,
      "loss": 0.8827,
      "step": 18889
    },
    {
      "epoch": 3.49247351793347,
      "grad_norm": 1.1436219215393066,
      "learning_rate": 7.362461193201642e-06,
      "loss": 0.9701,
      "step": 18890
    },
    {
      "epoch": 3.4926593569968407,
      "grad_norm": 1.0252236127853394,
      "learning_rate": 7.356960283101199e-06,
      "loss": 0.7517,
      "step": 18891
    },
    {
      "epoch": 3.492845196060212,
      "grad_norm": 0.9899161458015442,
      "learning_rate": 7.351461350307465e-06,
      "loss": 0.85,
      "step": 18892
    },
    {
      "epoch": 3.493031035123583,
      "grad_norm": 1.1004449129104614,
      "learning_rate": 7.345964394937787e-06,
      "loss": 0.8956,
      "step": 18893
    },
    {
      "epoch": 3.4932168741869543,
      "grad_norm": 1.0823179483413696,
      "learning_rate": 7.340469417109508e-06,
      "loss": 0.9652,
      "step": 18894
    },
    {
      "epoch": 3.493402713250325,
      "grad_norm": 0.96690833568573,
      "learning_rate": 7.334976416939887e-06,
      "loss": 0.887,
      "step": 18895
    },
    {
      "epoch": 3.4935885523136965,
      "grad_norm": 0.9663548469543457,
      "learning_rate": 7.329485394546187e-06,
      "loss": 0.9396,
      "step": 18896
    },
    {
      "epoch": 3.4937743913770674,
      "grad_norm": 0.9076710939407349,
      "learning_rate": 7.323996350045592e-06,
      "loss": 0.7719,
      "step": 18897
    },
    {
      "epoch": 3.4939602304404387,
      "grad_norm": 1.0663554668426514,
      "learning_rate": 7.318509283555264e-06,
      "loss": 0.9178,
      "step": 18898
    },
    {
      "epoch": 3.4941460695038096,
      "grad_norm": 1.2956310510635376,
      "learning_rate": 7.3130241951922865e-06,
      "loss": 1.1085,
      "step": 18899
    },
    {
      "epoch": 3.494331908567181,
      "grad_norm": 0.8738195300102234,
      "learning_rate": 7.307541085073766e-06,
      "loss": 0.9171,
      "step": 18900
    },
    {
      "epoch": 3.494517747630552,
      "grad_norm": 0.9473695755004883,
      "learning_rate": 7.30205995331672e-06,
      "loss": 0.7803,
      "step": 18901
    },
    {
      "epoch": 3.494703586693923,
      "grad_norm": 0.9984051585197449,
      "learning_rate": 7.296580800038122e-06,
      "loss": 0.9812,
      "step": 18902
    },
    {
      "epoch": 3.494889425757294,
      "grad_norm": 1.0475997924804688,
      "learning_rate": 7.291103625354911e-06,
      "loss": 0.9325,
      "step": 18903
    },
    {
      "epoch": 3.4950752648206653,
      "grad_norm": 0.9377017021179199,
      "learning_rate": 7.285628429383984e-06,
      "loss": 0.7597,
      "step": 18904
    },
    {
      "epoch": 3.4952611038840367,
      "grad_norm": 0.8637597560882568,
      "learning_rate": 7.2801552122422454e-06,
      "loss": 0.5853,
      "step": 18905
    },
    {
      "epoch": 3.4954469429474075,
      "grad_norm": 0.8163447976112366,
      "learning_rate": 7.2746839740464704e-06,
      "loss": 0.6579,
      "step": 18906
    },
    {
      "epoch": 3.4956327820107784,
      "grad_norm": 1.0498013496398926,
      "learning_rate": 7.269214714913431e-06,
      "loss": 0.8269,
      "step": 18907
    },
    {
      "epoch": 3.4958186210741498,
      "grad_norm": 1.0026522874832153,
      "learning_rate": 7.2637474349598886e-06,
      "loss": 0.7059,
      "step": 18908
    },
    {
      "epoch": 3.496004460137521,
      "grad_norm": 1.001114010810852,
      "learning_rate": 7.258282134302519e-06,
      "loss": 0.9834,
      "step": 18909
    },
    {
      "epoch": 3.496190299200892,
      "grad_norm": 0.9880086779594421,
      "learning_rate": 7.25281881305796e-06,
      "loss": 0.8499,
      "step": 18910
    },
    {
      "epoch": 3.4963761382642633,
      "grad_norm": 1.004437804222107,
      "learning_rate": 7.2473574713428085e-06,
      "loss": 0.6975,
      "step": 18911
    },
    {
      "epoch": 3.496561977327634,
      "grad_norm": 0.7537514567375183,
      "learning_rate": 7.241898109273659e-06,
      "loss": 0.6349,
      "step": 18912
    },
    {
      "epoch": 3.4967478163910055,
      "grad_norm": 1.0492370128631592,
      "learning_rate": 7.236440726967009e-06,
      "loss": 1.0626,
      "step": 18913
    },
    {
      "epoch": 3.4969336554543764,
      "grad_norm": 2.7229769229888916,
      "learning_rate": 7.230985324539341e-06,
      "loss": 1.1309,
      "step": 18914
    },
    {
      "epoch": 3.4971194945177477,
      "grad_norm": 1.1682634353637695,
      "learning_rate": 7.225531902107075e-06,
      "loss": 1.0752,
      "step": 18915
    },
    {
      "epoch": 3.4973053335811186,
      "grad_norm": 1.3416955471038818,
      "learning_rate": 7.220080459786649e-06,
      "loss": 0.8116,
      "step": 18916
    },
    {
      "epoch": 3.49749117264449,
      "grad_norm": 0.9258460402488708,
      "learning_rate": 7.214630997694393e-06,
      "loss": 0.993,
      "step": 18917
    },
    {
      "epoch": 3.497677011707861,
      "grad_norm": 0.7989770174026489,
      "learning_rate": 7.209183515946605e-06,
      "loss": 0.5692,
      "step": 18918
    },
    {
      "epoch": 3.497862850771232,
      "grad_norm": 0.8964726328849792,
      "learning_rate": 7.203738014659545e-06,
      "loss": 0.7662,
      "step": 18919
    },
    {
      "epoch": 3.498048689834603,
      "grad_norm": 1.1176609992980957,
      "learning_rate": 7.198294493949475e-06,
      "loss": 0.8345,
      "step": 18920
    },
    {
      "epoch": 3.4982345288979744,
      "grad_norm": 0.9257598519325256,
      "learning_rate": 7.1928529539325495e-06,
      "loss": 0.7902,
      "step": 18921
    },
    {
      "epoch": 3.4984203679613453,
      "grad_norm": 0.97118079662323,
      "learning_rate": 7.187413394724907e-06,
      "loss": 0.7702,
      "step": 18922
    },
    {
      "epoch": 3.4986062070247166,
      "grad_norm": 0.9934542179107666,
      "learning_rate": 7.181975816442643e-06,
      "loss": 0.8258,
      "step": 18923
    },
    {
      "epoch": 3.498792046088088,
      "grad_norm": 0.9483210444450378,
      "learning_rate": 7.176540219201833e-06,
      "loss": 0.8938,
      "step": 18924
    },
    {
      "epoch": 3.498977885151459,
      "grad_norm": 0.8880845904350281,
      "learning_rate": 7.171106603118482e-06,
      "loss": 0.5926,
      "step": 18925
    },
    {
      "epoch": 3.49916372421483,
      "grad_norm": 0.8391692638397217,
      "learning_rate": 7.165674968308567e-06,
      "loss": 0.9354,
      "step": 18926
    },
    {
      "epoch": 3.499349563278201,
      "grad_norm": 0.9406927824020386,
      "learning_rate": 7.160245314887992e-06,
      "loss": 0.6715,
      "step": 18927
    },
    {
      "epoch": 3.4995354023415723,
      "grad_norm": 0.9885954856872559,
      "learning_rate": 7.1548176429726775e-06,
      "loss": 0.807,
      "step": 18928
    },
    {
      "epoch": 3.4997212414049432,
      "grad_norm": 1.1389963626861572,
      "learning_rate": 7.149391952678452e-06,
      "loss": 0.8965,
      "step": 18929
    },
    {
      "epoch": 3.4999070804683146,
      "grad_norm": 1.0039417743682861,
      "learning_rate": 7.143968244121113e-06,
      "loss": 0.7828,
      "step": 18930
    },
    {
      "epoch": 3.5000929195316854,
      "grad_norm": 0.9714101552963257,
      "learning_rate": 7.138546517416411e-06,
      "loss": 0.8252,
      "step": 18931
    },
    {
      "epoch": 3.5002787585950568,
      "grad_norm": 1.0853351354599,
      "learning_rate": 7.1331267726800985e-06,
      "loss": 0.6779,
      "step": 18932
    },
    {
      "epoch": 3.5004645976584277,
      "grad_norm": 0.8246802687644958,
      "learning_rate": 7.127709010027827e-06,
      "loss": 0.8654,
      "step": 18933
    },
    {
      "epoch": 3.500650436721799,
      "grad_norm": 0.7620457410812378,
      "learning_rate": 7.122293229575228e-06,
      "loss": 0.665,
      "step": 18934
    },
    {
      "epoch": 3.5008362757851703,
      "grad_norm": 0.8930896520614624,
      "learning_rate": 7.116879431437862e-06,
      "loss": 0.8025,
      "step": 18935
    },
    {
      "epoch": 3.501022114848541,
      "grad_norm": 1.221191644668579,
      "learning_rate": 7.111467615731349e-06,
      "loss": 1.017,
      "step": 18936
    },
    {
      "epoch": 3.501207953911912,
      "grad_norm": 0.8264509439468384,
      "learning_rate": 7.106057782571152e-06,
      "loss": 0.9468,
      "step": 18937
    },
    {
      "epoch": 3.5013937929752834,
      "grad_norm": 0.8465451002120972,
      "learning_rate": 7.1006499320727336e-06,
      "loss": 0.5654,
      "step": 18938
    },
    {
      "epoch": 3.5015796320386547,
      "grad_norm": 0.8156012296676636,
      "learning_rate": 7.095244064351514e-06,
      "loss": 0.6718,
      "step": 18939
    },
    {
      "epoch": 3.5017654711020256,
      "grad_norm": 1.0571731328964233,
      "learning_rate": 7.089840179522888e-06,
      "loss": 0.9362,
      "step": 18940
    },
    {
      "epoch": 3.5019513101653965,
      "grad_norm": 0.9062555432319641,
      "learning_rate": 7.084438277702188e-06,
      "loss": 0.8661,
      "step": 18941
    },
    {
      "epoch": 3.502137149228768,
      "grad_norm": 1.0284250974655151,
      "learning_rate": 7.079038359004708e-06,
      "loss": 0.9107,
      "step": 18942
    },
    {
      "epoch": 3.502322988292139,
      "grad_norm": 1.419744610786438,
      "learning_rate": 7.073640423545669e-06,
      "loss": 1.0237,
      "step": 18943
    },
    {
      "epoch": 3.50250882735551,
      "grad_norm": 0.9171438813209534,
      "learning_rate": 7.068244471440333e-06,
      "loss": 0.8541,
      "step": 18944
    },
    {
      "epoch": 3.5026946664188814,
      "grad_norm": 0.8876679539680481,
      "learning_rate": 7.06285050280382e-06,
      "loss": 0.7288,
      "step": 18945
    },
    {
      "epoch": 3.5028805054822523,
      "grad_norm": 0.9148408770561218,
      "learning_rate": 7.057458517751292e-06,
      "loss": 0.8729,
      "step": 18946
    },
    {
      "epoch": 3.5030663445456236,
      "grad_norm": 0.8968338966369629,
      "learning_rate": 7.052068516397803e-06,
      "loss": 0.8159,
      "step": 18947
    },
    {
      "epoch": 3.5032521836089945,
      "grad_norm": 0.8965976238250732,
      "learning_rate": 7.046680498858415e-06,
      "loss": 0.6007,
      "step": 18948
    },
    {
      "epoch": 3.503438022672366,
      "grad_norm": 1.0043259859085083,
      "learning_rate": 7.041294465248127e-06,
      "loss": 0.8904,
      "step": 18949
    },
    {
      "epoch": 3.5036238617357367,
      "grad_norm": 0.9085475206375122,
      "learning_rate": 7.035910415681879e-06,
      "loss": 0.7296,
      "step": 18950
    },
    {
      "epoch": 3.503809700799108,
      "grad_norm": 1.0071899890899658,
      "learning_rate": 7.030528350274579e-06,
      "loss": 0.9354,
      "step": 18951
    },
    {
      "epoch": 3.503995539862479,
      "grad_norm": 0.9521633982658386,
      "learning_rate": 7.0251482691411245e-06,
      "loss": 0.9424,
      "step": 18952
    },
    {
      "epoch": 3.5041813789258502,
      "grad_norm": 0.7816515564918518,
      "learning_rate": 7.019770172396334e-06,
      "loss": 0.7411,
      "step": 18953
    },
    {
      "epoch": 3.5043672179892216,
      "grad_norm": 0.8777903318405151,
      "learning_rate": 7.014394060154983e-06,
      "loss": 0.7773,
      "step": 18954
    },
    {
      "epoch": 3.5045530570525925,
      "grad_norm": 0.9105899930000305,
      "learning_rate": 7.009019932531802e-06,
      "loss": 1.0029,
      "step": 18955
    },
    {
      "epoch": 3.5047388961159633,
      "grad_norm": 1.0930300951004028,
      "learning_rate": 7.0036477896415095e-06,
      "loss": 1.1189,
      "step": 18956
    },
    {
      "epoch": 3.5049247351793347,
      "grad_norm": 0.8531712293624878,
      "learning_rate": 6.998277631598793e-06,
      "loss": 0.7549,
      "step": 18957
    },
    {
      "epoch": 3.505110574242706,
      "grad_norm": 0.9285844564437866,
      "learning_rate": 6.992909458518237e-06,
      "loss": 0.6884,
      "step": 18958
    },
    {
      "epoch": 3.505296413306077,
      "grad_norm": 1.1396297216415405,
      "learning_rate": 6.987543270514407e-06,
      "loss": 0.7258,
      "step": 18959
    },
    {
      "epoch": 3.505482252369448,
      "grad_norm": 0.9273241758346558,
      "learning_rate": 6.982179067701866e-06,
      "loss": 0.9317,
      "step": 18960
    },
    {
      "epoch": 3.505668091432819,
      "grad_norm": 0.8453611135482788,
      "learning_rate": 6.976816850195089e-06,
      "loss": 0.6724,
      "step": 18961
    },
    {
      "epoch": 3.5058539304961904,
      "grad_norm": 1.0250625610351562,
      "learning_rate": 6.9714566181085185e-06,
      "loss": 0.9586,
      "step": 18962
    },
    {
      "epoch": 3.5060397695595613,
      "grad_norm": 0.8976748585700989,
      "learning_rate": 6.966098371556551e-06,
      "loss": 0.9126,
      "step": 18963
    },
    {
      "epoch": 3.5062256086229326,
      "grad_norm": 1.01888108253479,
      "learning_rate": 6.960742110653573e-06,
      "loss": 0.6516,
      "step": 18964
    },
    {
      "epoch": 3.5064114476863035,
      "grad_norm": 0.8228797912597656,
      "learning_rate": 6.9553878355138936e-06,
      "loss": 0.7013,
      "step": 18965
    },
    {
      "epoch": 3.506597286749675,
      "grad_norm": 0.8661003112792969,
      "learning_rate": 6.950035546251777e-06,
      "loss": 0.7522,
      "step": 18966
    },
    {
      "epoch": 3.5067831258130457,
      "grad_norm": 0.9044394493103027,
      "learning_rate": 6.944685242981463e-06,
      "loss": 1.0133,
      "step": 18967
    },
    {
      "epoch": 3.506968964876417,
      "grad_norm": 0.8815124034881592,
      "learning_rate": 6.939336925817186e-06,
      "loss": 0.8539,
      "step": 18968
    },
    {
      "epoch": 3.5071548039397884,
      "grad_norm": 0.9001720547676086,
      "learning_rate": 6.9339905948730525e-06,
      "loss": 0.5971,
      "step": 18969
    },
    {
      "epoch": 3.5073406430031593,
      "grad_norm": 1.0033185482025146,
      "learning_rate": 6.928646250263182e-06,
      "loss": 0.7974,
      "step": 18970
    },
    {
      "epoch": 3.50752648206653,
      "grad_norm": 0.9565051794052124,
      "learning_rate": 6.923303892101629e-06,
      "loss": 0.9197,
      "step": 18971
    },
    {
      "epoch": 3.5077123211299015,
      "grad_norm": 1.0655335187911987,
      "learning_rate": 6.917963520502435e-06,
      "loss": 0.9777,
      "step": 18972
    },
    {
      "epoch": 3.507898160193273,
      "grad_norm": 0.938111424446106,
      "learning_rate": 6.9126251355795864e-06,
      "loss": 0.7424,
      "step": 18973
    },
    {
      "epoch": 3.5080839992566437,
      "grad_norm": 1.0866252183914185,
      "learning_rate": 6.907288737447004e-06,
      "loss": 0.7386,
      "step": 18974
    },
    {
      "epoch": 3.5082698383200146,
      "grad_norm": 0.9407966136932373,
      "learning_rate": 6.901954326218573e-06,
      "loss": 0.9777,
      "step": 18975
    },
    {
      "epoch": 3.508455677383386,
      "grad_norm": 1.230141043663025,
      "learning_rate": 6.8966219020081826e-06,
      "loss": 0.9592,
      "step": 18976
    },
    {
      "epoch": 3.5086415164467573,
      "grad_norm": 0.9053949117660522,
      "learning_rate": 6.891291464929606e-06,
      "loss": 0.7945,
      "step": 18977
    },
    {
      "epoch": 3.508827355510128,
      "grad_norm": 0.8104222416877747,
      "learning_rate": 6.885963015096652e-06,
      "loss": 0.9587,
      "step": 18978
    },
    {
      "epoch": 3.5090131945734995,
      "grad_norm": 0.9656167030334473,
      "learning_rate": 6.88063655262301e-06,
      "loss": 0.6239,
      "step": 18979
    },
    {
      "epoch": 3.5091990336368704,
      "grad_norm": 1.8394811153411865,
      "learning_rate": 6.875312077622398e-06,
      "loss": 1.1899,
      "step": 18980
    },
    {
      "epoch": 3.5093848727002417,
      "grad_norm": 0.9846952557563782,
      "learning_rate": 6.869989590208448e-06,
      "loss": 0.9301,
      "step": 18981
    },
    {
      "epoch": 3.5095707117636126,
      "grad_norm": 1.0243431329727173,
      "learning_rate": 6.864669090494746e-06,
      "loss": 0.7457,
      "step": 18982
    },
    {
      "epoch": 3.509756550826984,
      "grad_norm": 0.9600151181221008,
      "learning_rate": 6.8593505785948365e-06,
      "loss": 0.7291,
      "step": 18983
    },
    {
      "epoch": 3.5099423898903552,
      "grad_norm": 0.9003852009773254,
      "learning_rate": 6.854034054622271e-06,
      "loss": 0.8666,
      "step": 18984
    },
    {
      "epoch": 3.510128228953726,
      "grad_norm": 1.0153359174728394,
      "learning_rate": 6.848719518690505e-06,
      "loss": 0.826,
      "step": 18985
    },
    {
      "epoch": 3.510314068017097,
      "grad_norm": 0.861562967300415,
      "learning_rate": 6.843406970912969e-06,
      "loss": 0.6989,
      "step": 18986
    },
    {
      "epoch": 3.5104999070804683,
      "grad_norm": 1.1449443101882935,
      "learning_rate": 6.838096411403017e-06,
      "loss": 0.7711,
      "step": 18987
    },
    {
      "epoch": 3.5106857461438397,
      "grad_norm": 0.8476724028587341,
      "learning_rate": 6.8327878402740465e-06,
      "loss": 0.8695,
      "step": 18988
    },
    {
      "epoch": 3.5108715852072105,
      "grad_norm": 0.8885052800178528,
      "learning_rate": 6.827481257639345e-06,
      "loss": 0.8422,
      "step": 18989
    },
    {
      "epoch": 3.5110574242705814,
      "grad_norm": 0.90379798412323,
      "learning_rate": 6.822176663612168e-06,
      "loss": 0.8569,
      "step": 18990
    },
    {
      "epoch": 3.5112432633339528,
      "grad_norm": 0.9090231657028198,
      "learning_rate": 6.816874058305711e-06,
      "loss": 0.7485,
      "step": 18991
    },
    {
      "epoch": 3.511429102397324,
      "grad_norm": 0.8142905831336975,
      "learning_rate": 6.8115734418331965e-06,
      "loss": 0.7595,
      "step": 18992
    },
    {
      "epoch": 3.511614941460695,
      "grad_norm": 0.9143938422203064,
      "learning_rate": 6.806274814307712e-06,
      "loss": 0.6598,
      "step": 18993
    },
    {
      "epoch": 3.5118007805240663,
      "grad_norm": 0.9759804606437683,
      "learning_rate": 6.800978175842377e-06,
      "loss": 0.7634,
      "step": 18994
    },
    {
      "epoch": 3.511986619587437,
      "grad_norm": 0.9622668623924255,
      "learning_rate": 6.7956835265502e-06,
      "loss": 0.5872,
      "step": 18995
    },
    {
      "epoch": 3.5121724586508085,
      "grad_norm": 0.9753997325897217,
      "learning_rate": 6.790390866544238e-06,
      "loss": 0.687,
      "step": 18996
    },
    {
      "epoch": 3.5123582977141794,
      "grad_norm": 1.0855896472930908,
      "learning_rate": 6.785100195937422e-06,
      "loss": 0.5035,
      "step": 18997
    },
    {
      "epoch": 3.5125441367775507,
      "grad_norm": 1.1820878982543945,
      "learning_rate": 6.7798115148426715e-06,
      "loss": 0.9115,
      "step": 18998
    },
    {
      "epoch": 3.5127299758409216,
      "grad_norm": 0.9932574033737183,
      "learning_rate": 6.774524823372863e-06,
      "loss": 0.8128,
      "step": 18999
    },
    {
      "epoch": 3.512915814904293,
      "grad_norm": 0.9725932478904724,
      "learning_rate": 6.769240121640863e-06,
      "loss": 0.8454,
      "step": 19000
    },
    {
      "epoch": 3.513101653967664,
      "grad_norm": 1.1076157093048096,
      "learning_rate": 6.763957409759447e-06,
      "loss": 0.7596,
      "step": 19001
    },
    {
      "epoch": 3.513287493031035,
      "grad_norm": 0.8583747744560242,
      "learning_rate": 6.758676687841348e-06,
      "loss": 0.8204,
      "step": 19002
    },
    {
      "epoch": 3.5134733320944065,
      "grad_norm": 0.8928229808807373,
      "learning_rate": 6.753397955999274e-06,
      "loss": 0.8113,
      "step": 19003
    },
    {
      "epoch": 3.5136591711577774,
      "grad_norm": 0.9273632764816284,
      "learning_rate": 6.748121214345926e-06,
      "loss": 0.8261,
      "step": 19004
    },
    {
      "epoch": 3.5138450102211483,
      "grad_norm": 0.8253338932991028,
      "learning_rate": 6.742846462993901e-06,
      "loss": 0.6899,
      "step": 19005
    },
    {
      "epoch": 3.5140308492845196,
      "grad_norm": 0.9060574769973755,
      "learning_rate": 6.737573702055766e-06,
      "loss": 0.8283,
      "step": 19006
    },
    {
      "epoch": 3.514216688347891,
      "grad_norm": 0.8999438881874084,
      "learning_rate": 6.732302931644085e-06,
      "loss": 0.888,
      "step": 19007
    },
    {
      "epoch": 3.514402527411262,
      "grad_norm": 1.105154037475586,
      "learning_rate": 6.727034151871336e-06,
      "loss": 0.7137,
      "step": 19008
    },
    {
      "epoch": 3.514588366474633,
      "grad_norm": 0.865492045879364,
      "learning_rate": 6.721767362849996e-06,
      "loss": 0.7081,
      "step": 19009
    },
    {
      "epoch": 3.514774205538004,
      "grad_norm": 0.9106673002243042,
      "learning_rate": 6.716502564692462e-06,
      "loss": 0.739,
      "step": 19010
    },
    {
      "epoch": 3.5149600446013753,
      "grad_norm": 0.8903344869613647,
      "learning_rate": 6.711239757511079e-06,
      "loss": 0.8402,
      "step": 19011
    },
    {
      "epoch": 3.5151458836647462,
      "grad_norm": 1.0108816623687744,
      "learning_rate": 6.705978941418212e-06,
      "loss": 0.8127,
      "step": 19012
    },
    {
      "epoch": 3.5153317227281176,
      "grad_norm": 0.9158901572227478,
      "learning_rate": 6.700720116526116e-06,
      "loss": 0.8192,
      "step": 19013
    },
    {
      "epoch": 3.5155175617914884,
      "grad_norm": 0.9774060845375061,
      "learning_rate": 6.695463282947057e-06,
      "loss": 0.7544,
      "step": 19014
    },
    {
      "epoch": 3.5157034008548598,
      "grad_norm": 1.0111356973648071,
      "learning_rate": 6.690208440793189e-06,
      "loss": 0.9235,
      "step": 19015
    },
    {
      "epoch": 3.5158892399182307,
      "grad_norm": 1.0181794166564941,
      "learning_rate": 6.684955590176711e-06,
      "loss": 0.676,
      "step": 19016
    },
    {
      "epoch": 3.516075078981602,
      "grad_norm": 1.300851583480835,
      "learning_rate": 6.679704731209724e-06,
      "loss": 1.1365,
      "step": 19017
    },
    {
      "epoch": 3.5162609180449733,
      "grad_norm": 0.9330505728721619,
      "learning_rate": 6.674455864004292e-06,
      "loss": 0.8658,
      "step": 19018
    },
    {
      "epoch": 3.516446757108344,
      "grad_norm": 1.028049111366272,
      "learning_rate": 6.669208988672426e-06,
      "loss": 0.8605,
      "step": 19019
    },
    {
      "epoch": 3.516632596171715,
      "grad_norm": 1.0538666248321533,
      "learning_rate": 6.663964105326159e-06,
      "loss": 0.859,
      "step": 19020
    },
    {
      "epoch": 3.5168184352350864,
      "grad_norm": 0.8396360874176025,
      "learning_rate": 6.658721214077412e-06,
      "loss": 0.7899,
      "step": 19021
    },
    {
      "epoch": 3.5170042742984577,
      "grad_norm": 1.0336717367172241,
      "learning_rate": 6.653480315038074e-06,
      "loss": 0.8283,
      "step": 19022
    },
    {
      "epoch": 3.5171901133618286,
      "grad_norm": 0.9230265021324158,
      "learning_rate": 6.648241408319988e-06,
      "loss": 0.7298,
      "step": 19023
    },
    {
      "epoch": 3.5173759524251995,
      "grad_norm": 0.952717125415802,
      "learning_rate": 6.643004494035021e-06,
      "loss": 0.8616,
      "step": 19024
    },
    {
      "epoch": 3.517561791488571,
      "grad_norm": 0.93483567237854,
      "learning_rate": 6.637769572294905e-06,
      "loss": 0.8873,
      "step": 19025
    },
    {
      "epoch": 3.517747630551942,
      "grad_norm": 0.948124349117279,
      "learning_rate": 6.632536643211373e-06,
      "loss": 0.7508,
      "step": 19026
    },
    {
      "epoch": 3.517933469615313,
      "grad_norm": 1.176931619644165,
      "learning_rate": 6.6273057068961255e-06,
      "loss": 0.7108,
      "step": 19027
    },
    {
      "epoch": 3.5181193086786844,
      "grad_norm": 1.1013628244400024,
      "learning_rate": 6.622076763460816e-06,
      "loss": 1.1076,
      "step": 19028
    },
    {
      "epoch": 3.5183051477420553,
      "grad_norm": 1.0013099908828735,
      "learning_rate": 6.616849813017012e-06,
      "loss": 0.8936,
      "step": 19029
    },
    {
      "epoch": 3.5184909868054266,
      "grad_norm": 0.9172407984733582,
      "learning_rate": 6.611624855676313e-06,
      "loss": 0.7041,
      "step": 19030
    },
    {
      "epoch": 3.5186768258687975,
      "grad_norm": 0.9969894886016846,
      "learning_rate": 6.6064018915502065e-06,
      "loss": 0.5888,
      "step": 19031
    },
    {
      "epoch": 3.518862664932169,
      "grad_norm": 1.0015980005264282,
      "learning_rate": 6.601180920750205e-06,
      "loss": 0.7944,
      "step": 19032
    },
    {
      "epoch": 3.5190485039955397,
      "grad_norm": 0.9136693477630615,
      "learning_rate": 6.595961943387707e-06,
      "loss": 0.8752,
      "step": 19033
    },
    {
      "epoch": 3.519234343058911,
      "grad_norm": 0.9295935034751892,
      "learning_rate": 6.5907449595741244e-06,
      "loss": 0.7903,
      "step": 19034
    },
    {
      "epoch": 3.519420182122282,
      "grad_norm": 1.0148817300796509,
      "learning_rate": 6.5855299694207675e-06,
      "loss": 0.8275,
      "step": 19035
    },
    {
      "epoch": 3.5196060211856532,
      "grad_norm": 0.8606860041618347,
      "learning_rate": 6.5803169730389914e-06,
      "loss": 0.6784,
      "step": 19036
    },
    {
      "epoch": 3.5197918602490246,
      "grad_norm": 0.8806768655776978,
      "learning_rate": 6.5751059705400295e-06,
      "loss": 0.6457,
      "step": 19037
    },
    {
      "epoch": 3.5199776993123955,
      "grad_norm": 0.9651481509208679,
      "learning_rate": 6.569896962035094e-06,
      "loss": 0.7849,
      "step": 19038
    },
    {
      "epoch": 3.5201635383757663,
      "grad_norm": 1.0954023599624634,
      "learning_rate": 6.564689947635383e-06,
      "loss": 0.7911,
      "step": 19039
    },
    {
      "epoch": 3.5203493774391377,
      "grad_norm": 0.8198789358139038,
      "learning_rate": 6.55948492745202e-06,
      "loss": 0.5841,
      "step": 19040
    },
    {
      "epoch": 3.520535216502509,
      "grad_norm": 0.9690950512886047,
      "learning_rate": 6.554281901596104e-06,
      "loss": 0.6836,
      "step": 19041
    },
    {
      "epoch": 3.52072105556588,
      "grad_norm": 0.977009117603302,
      "learning_rate": 6.549080870178692e-06,
      "loss": 0.6822,
      "step": 19042
    },
    {
      "epoch": 3.520906894629251,
      "grad_norm": 0.9366753101348877,
      "learning_rate": 6.543881833310761e-06,
      "loss": 0.9865,
      "step": 19043
    },
    {
      "epoch": 3.521092733692622,
      "grad_norm": 1.1933904886245728,
      "learning_rate": 6.538684791103311e-06,
      "loss": 1.0158,
      "step": 19044
    },
    {
      "epoch": 3.5212785727559934,
      "grad_norm": 0.9818659424781799,
      "learning_rate": 6.5334897436672535e-06,
      "loss": 0.7215,
      "step": 19045
    },
    {
      "epoch": 3.5214644118193643,
      "grad_norm": 0.9957472085952759,
      "learning_rate": 6.528296691113445e-06,
      "loss": 0.7268,
      "step": 19046
    },
    {
      "epoch": 3.5216502508827356,
      "grad_norm": 1.0240875482559204,
      "learning_rate": 6.523105633552751e-06,
      "loss": 0.7408,
      "step": 19047
    },
    {
      "epoch": 3.5218360899461065,
      "grad_norm": 0.9679796695709229,
      "learning_rate": 6.51791657109595e-06,
      "loss": 0.9828,
      "step": 19048
    },
    {
      "epoch": 3.522021929009478,
      "grad_norm": 1.1154346466064453,
      "learning_rate": 6.5127295038538095e-06,
      "loss": 0.9706,
      "step": 19049
    },
    {
      "epoch": 3.5222077680728487,
      "grad_norm": 0.858024537563324,
      "learning_rate": 6.507544431937007e-06,
      "loss": 0.8874,
      "step": 19050
    },
    {
      "epoch": 3.52239360713622,
      "grad_norm": 1.121969223022461,
      "learning_rate": 6.5023613554562216e-06,
      "loss": 0.7596,
      "step": 19051
    },
    {
      "epoch": 3.5225794461995914,
      "grad_norm": 0.8989073634147644,
      "learning_rate": 6.497180274522107e-06,
      "loss": 0.8389,
      "step": 19052
    },
    {
      "epoch": 3.5227652852629623,
      "grad_norm": 0.9460635185241699,
      "learning_rate": 6.492001189245223e-06,
      "loss": 1.0833,
      "step": 19053
    },
    {
      "epoch": 3.522951124326333,
      "grad_norm": 0.9267661571502686,
      "learning_rate": 6.48682409973611e-06,
      "loss": 1.0331,
      "step": 19054
    },
    {
      "epoch": 3.5231369633897045,
      "grad_norm": 0.9577577710151672,
      "learning_rate": 6.48164900610524e-06,
      "loss": 0.9354,
      "step": 19055
    },
    {
      "epoch": 3.523322802453076,
      "grad_norm": 1.011656641960144,
      "learning_rate": 6.4764759084631085e-06,
      "loss": 0.809,
      "step": 19056
    },
    {
      "epoch": 3.5235086415164467,
      "grad_norm": 0.9175121784210205,
      "learning_rate": 6.471304806920109e-06,
      "loss": 0.7586,
      "step": 19057
    },
    {
      "epoch": 3.5236944805798176,
      "grad_norm": 1.0583686828613281,
      "learning_rate": 6.466135701586584e-06,
      "loss": 0.8551,
      "step": 19058
    },
    {
      "epoch": 3.523880319643189,
      "grad_norm": 1.392242431640625,
      "learning_rate": 6.460968592572892e-06,
      "loss": 0.7784,
      "step": 19059
    },
    {
      "epoch": 3.5240661587065603,
      "grad_norm": 1.0272305011749268,
      "learning_rate": 6.45580347998932e-06,
      "loss": 0.8824,
      "step": 19060
    },
    {
      "epoch": 3.524251997769931,
      "grad_norm": 0.9726600646972656,
      "learning_rate": 6.45064036394607e-06,
      "loss": 1.2702,
      "step": 19061
    },
    {
      "epoch": 3.5244378368333025,
      "grad_norm": 1.0088410377502441,
      "learning_rate": 6.445479244553376e-06,
      "loss": 0.8307,
      "step": 19062
    },
    {
      "epoch": 3.5246236758966734,
      "grad_norm": 1.0037113428115845,
      "learning_rate": 6.440320121921372e-06,
      "loss": 0.8856,
      "step": 19063
    },
    {
      "epoch": 3.5248095149600447,
      "grad_norm": 1.1829400062561035,
      "learning_rate": 6.435162996160204e-06,
      "loss": 1.0169,
      "step": 19064
    },
    {
      "epoch": 3.5249953540234156,
      "grad_norm": 0.8123091459274292,
      "learning_rate": 6.430007867379905e-06,
      "loss": 0.8608,
      "step": 19065
    },
    {
      "epoch": 3.525181193086787,
      "grad_norm": 1.1172996759414673,
      "learning_rate": 6.4248547356905105e-06,
      "loss": 0.8055,
      "step": 19066
    },
    {
      "epoch": 3.5253670321501582,
      "grad_norm": 0.9727967381477356,
      "learning_rate": 6.41970360120202e-06,
      "loss": 0.8744,
      "step": 19067
    },
    {
      "epoch": 3.525552871213529,
      "grad_norm": 1.1461392641067505,
      "learning_rate": 6.4145544640243585e-06,
      "loss": 0.973,
      "step": 19068
    },
    {
      "epoch": 3.5257387102769,
      "grad_norm": 1.0670104026794434,
      "learning_rate": 6.409407324267447e-06,
      "loss": 0.8479,
      "step": 19069
    },
    {
      "epoch": 3.5259245493402713,
      "grad_norm": 1.036465048789978,
      "learning_rate": 6.404262182041099e-06,
      "loss": 0.7001,
      "step": 19070
    },
    {
      "epoch": 3.5261103884036427,
      "grad_norm": 1.0140867233276367,
      "learning_rate": 6.39911903745517e-06,
      "loss": 0.8693,
      "step": 19071
    },
    {
      "epoch": 3.5262962274670135,
      "grad_norm": 0.7492370009422302,
      "learning_rate": 6.3939778906194294e-06,
      "loss": 0.6441,
      "step": 19072
    },
    {
      "epoch": 3.5264820665303844,
      "grad_norm": 0.8034688234329224,
      "learning_rate": 6.388838741643599e-06,
      "loss": 0.6989,
      "step": 19073
    },
    {
      "epoch": 3.5266679055937558,
      "grad_norm": 0.9098494052886963,
      "learning_rate": 6.383701590637359e-06,
      "loss": 1.0181,
      "step": 19074
    },
    {
      "epoch": 3.526853744657127,
      "grad_norm": 1.013979196548462,
      "learning_rate": 6.3785664377103425e-06,
      "loss": 1.0134,
      "step": 19075
    },
    {
      "epoch": 3.527039583720498,
      "grad_norm": 0.8733248114585876,
      "learning_rate": 6.3734332829721745e-06,
      "loss": 0.7987,
      "step": 19076
    },
    {
      "epoch": 3.5272254227838693,
      "grad_norm": 0.9233043193817139,
      "learning_rate": 6.36830212653241e-06,
      "loss": 0.8204,
      "step": 19077
    },
    {
      "epoch": 3.52741126184724,
      "grad_norm": 0.9699963927268982,
      "learning_rate": 6.363172968500541e-06,
      "loss": 0.8159,
      "step": 19078
    },
    {
      "epoch": 3.5275971009106115,
      "grad_norm": 0.8718108534812927,
      "learning_rate": 6.3580458089860794e-06,
      "loss": 0.8537,
      "step": 19079
    },
    {
      "epoch": 3.5277829399739824,
      "grad_norm": 0.9551896452903748,
      "learning_rate": 6.352920648098426e-06,
      "loss": 0.787,
      "step": 19080
    },
    {
      "epoch": 3.5279687790373537,
      "grad_norm": 0.8582215309143066,
      "learning_rate": 6.3477974859469825e-06,
      "loss": 0.7686,
      "step": 19081
    },
    {
      "epoch": 3.5281546181007246,
      "grad_norm": 1.047671914100647,
      "learning_rate": 6.342676322641072e-06,
      "loss": 0.76,
      "step": 19082
    },
    {
      "epoch": 3.528340457164096,
      "grad_norm": 0.7875006198883057,
      "learning_rate": 6.337557158290009e-06,
      "loss": 0.8089,
      "step": 19083
    },
    {
      "epoch": 3.528526296227467,
      "grad_norm": 1.0197948217391968,
      "learning_rate": 6.33243999300307e-06,
      "loss": 1.2163,
      "step": 19084
    },
    {
      "epoch": 3.528712135290838,
      "grad_norm": 0.9266148805618286,
      "learning_rate": 6.327324826889469e-06,
      "loss": 0.7697,
      "step": 19085
    },
    {
      "epoch": 3.5288979743542095,
      "grad_norm": 1.0414260625839233,
      "learning_rate": 6.322211660058342e-06,
      "loss": 0.8559,
      "step": 19086
    },
    {
      "epoch": 3.5290838134175804,
      "grad_norm": 1.0009326934814453,
      "learning_rate": 6.317100492618877e-06,
      "loss": 0.9016,
      "step": 19087
    },
    {
      "epoch": 3.5292696524809513,
      "grad_norm": 0.91349196434021,
      "learning_rate": 6.311991324680133e-06,
      "loss": 0.6186,
      "step": 19088
    },
    {
      "epoch": 3.5294554915443226,
      "grad_norm": 0.8203105926513672,
      "learning_rate": 6.306884156351156e-06,
      "loss": 0.8004,
      "step": 19089
    },
    {
      "epoch": 3.529641330607694,
      "grad_norm": 0.9701465964317322,
      "learning_rate": 6.301778987740936e-06,
      "loss": 0.8254,
      "step": 19090
    },
    {
      "epoch": 3.529827169671065,
      "grad_norm": 1.039881706237793,
      "learning_rate": 6.296675818958475e-06,
      "loss": 0.7108,
      "step": 19091
    },
    {
      "epoch": 3.530013008734436,
      "grad_norm": 0.9743512272834778,
      "learning_rate": 6.291574650112641e-06,
      "loss": 0.8026,
      "step": 19092
    },
    {
      "epoch": 3.530198847797807,
      "grad_norm": 0.9358527064323425,
      "learning_rate": 6.28647548131236e-06,
      "loss": 0.9723,
      "step": 19093
    },
    {
      "epoch": 3.5303846868611783,
      "grad_norm": 0.8311519622802734,
      "learning_rate": 6.281378312666442e-06,
      "loss": 0.7873,
      "step": 19094
    },
    {
      "epoch": 3.5305705259245492,
      "grad_norm": 0.8295403122901917,
      "learning_rate": 6.276283144283656e-06,
      "loss": 0.7648,
      "step": 19095
    },
    {
      "epoch": 3.5307563649879206,
      "grad_norm": 0.9160231947898865,
      "learning_rate": 6.271189976272785e-06,
      "loss": 0.8451,
      "step": 19096
    },
    {
      "epoch": 3.5309422040512914,
      "grad_norm": 1.2153279781341553,
      "learning_rate": 6.266098808742516e-06,
      "loss": 0.8933,
      "step": 19097
    },
    {
      "epoch": 3.5311280431146628,
      "grad_norm": 0.8746144771575928,
      "learning_rate": 6.261009641801507e-06,
      "loss": 0.7959,
      "step": 19098
    },
    {
      "epoch": 3.5313138821780337,
      "grad_norm": 0.825903594493866,
      "learning_rate": 6.2559224755583955e-06,
      "loss": 0.6744,
      "step": 19099
    },
    {
      "epoch": 3.531499721241405,
      "grad_norm": 0.8564138412475586,
      "learning_rate": 6.250837310121738e-06,
      "loss": 0.7711,
      "step": 19100
    },
    {
      "epoch": 3.5316855603047763,
      "grad_norm": 0.9014472365379333,
      "learning_rate": 6.245754145600091e-06,
      "loss": 0.9041,
      "step": 19101
    },
    {
      "epoch": 3.531871399368147,
      "grad_norm": 0.9554387927055359,
      "learning_rate": 6.240672982101914e-06,
      "loss": 0.8498,
      "step": 19102
    },
    {
      "epoch": 3.532057238431518,
      "grad_norm": 0.8042546510696411,
      "learning_rate": 6.235593819735674e-06,
      "loss": 0.8419,
      "step": 19103
    },
    {
      "epoch": 3.5322430774948894,
      "grad_norm": 0.9801455736160278,
      "learning_rate": 6.230516658609798e-06,
      "loss": 0.834,
      "step": 19104
    },
    {
      "epoch": 3.5324289165582607,
      "grad_norm": 0.9968674778938293,
      "learning_rate": 6.225441498832629e-06,
      "loss": 0.9294,
      "step": 19105
    },
    {
      "epoch": 3.5326147556216316,
      "grad_norm": 0.9890447854995728,
      "learning_rate": 6.220368340512461e-06,
      "loss": 0.7123,
      "step": 19106
    },
    {
      "epoch": 3.5328005946850025,
      "grad_norm": 0.8202847242355347,
      "learning_rate": 6.215297183757629e-06,
      "loss": 0.5153,
      "step": 19107
    },
    {
      "epoch": 3.532986433748374,
      "grad_norm": 0.8917155265808105,
      "learning_rate": 6.2102280286763235e-06,
      "loss": 0.9371,
      "step": 19108
    },
    {
      "epoch": 3.533172272811745,
      "grad_norm": 0.9823048114776611,
      "learning_rate": 6.2051608753767585e-06,
      "loss": 0.7532,
      "step": 19109
    },
    {
      "epoch": 3.533358111875116,
      "grad_norm": 0.8202247023582458,
      "learning_rate": 6.200095723967058e-06,
      "loss": 0.5955,
      "step": 19110
    },
    {
      "epoch": 3.5335439509384874,
      "grad_norm": 0.9594755172729492,
      "learning_rate": 6.1950325745553705e-06,
      "loss": 0.8521,
      "step": 19111
    },
    {
      "epoch": 3.5337297900018583,
      "grad_norm": 1.0144023895263672,
      "learning_rate": 6.18997142724973e-06,
      "loss": 0.7749,
      "step": 19112
    },
    {
      "epoch": 3.5339156290652296,
      "grad_norm": 1.067853331565857,
      "learning_rate": 6.1849122821581505e-06,
      "loss": 0.873,
      "step": 19113
    },
    {
      "epoch": 3.5341014681286005,
      "grad_norm": 1.0349130630493164,
      "learning_rate": 6.179855139388635e-06,
      "loss": 0.821,
      "step": 19114
    },
    {
      "epoch": 3.534287307191972,
      "grad_norm": 0.8972457051277161,
      "learning_rate": 6.174799999049097e-06,
      "loss": 0.7264,
      "step": 19115
    },
    {
      "epoch": 3.534473146255343,
      "grad_norm": 0.8936841487884521,
      "learning_rate": 6.169746861247461e-06,
      "loss": 0.6736,
      "step": 19116
    },
    {
      "epoch": 3.534658985318714,
      "grad_norm": 1.215692400932312,
      "learning_rate": 6.164695726091563e-06,
      "loss": 0.8523,
      "step": 19117
    },
    {
      "epoch": 3.534844824382085,
      "grad_norm": 0.904807448387146,
      "learning_rate": 6.1596465936891835e-06,
      "loss": 0.884,
      "step": 19118
    },
    {
      "epoch": 3.5350306634454562,
      "grad_norm": 1.222141146659851,
      "learning_rate": 6.154599464148136e-06,
      "loss": 1.0758,
      "step": 19119
    },
    {
      "epoch": 3.5352165025088276,
      "grad_norm": 0.9937120079994202,
      "learning_rate": 6.149554337576113e-06,
      "loss": 0.8717,
      "step": 19120
    },
    {
      "epoch": 3.5354023415721985,
      "grad_norm": 1.0441583395004272,
      "learning_rate": 6.144511214080795e-06,
      "loss": 0.7194,
      "step": 19121
    },
    {
      "epoch": 3.5355881806355693,
      "grad_norm": 0.931191086769104,
      "learning_rate": 6.139470093769806e-06,
      "loss": 0.6677,
      "step": 19122
    },
    {
      "epoch": 3.5357740196989407,
      "grad_norm": 0.9039041996002197,
      "learning_rate": 6.134430976750783e-06,
      "loss": 0.827,
      "step": 19123
    },
    {
      "epoch": 3.535959858762312,
      "grad_norm": 0.9199702739715576,
      "learning_rate": 6.129393863131228e-06,
      "loss": 0.7714,
      "step": 19124
    },
    {
      "epoch": 3.536145697825683,
      "grad_norm": 0.9820542335510254,
      "learning_rate": 6.124358753018689e-06,
      "loss": 0.8069,
      "step": 19125
    },
    {
      "epoch": 3.536331536889054,
      "grad_norm": 0.9555262923240662,
      "learning_rate": 6.119325646520591e-06,
      "loss": 0.5713,
      "step": 19126
    },
    {
      "epoch": 3.536517375952425,
      "grad_norm": 1.0285457372665405,
      "learning_rate": 6.114294543744414e-06,
      "loss": 0.7669,
      "step": 19127
    },
    {
      "epoch": 3.5367032150157964,
      "grad_norm": 0.9669595956802368,
      "learning_rate": 6.1092654447974965e-06,
      "loss": 0.8578,
      "step": 19128
    },
    {
      "epoch": 3.5368890540791673,
      "grad_norm": 0.911249041557312,
      "learning_rate": 6.104238349787183e-06,
      "loss": 0.6561,
      "step": 19129
    },
    {
      "epoch": 3.5370748931425386,
      "grad_norm": 0.8711757659912109,
      "learning_rate": 6.0992132588207576e-06,
      "loss": 0.4448,
      "step": 19130
    },
    {
      "epoch": 3.5372607322059095,
      "grad_norm": 0.8461050987243652,
      "learning_rate": 6.0941901720054985e-06,
      "loss": 0.8608,
      "step": 19131
    },
    {
      "epoch": 3.537446571269281,
      "grad_norm": 0.9975296258926392,
      "learning_rate": 6.089169089448599e-06,
      "loss": 0.8095,
      "step": 19132
    },
    {
      "epoch": 3.5376324103326517,
      "grad_norm": 0.9165214896202087,
      "learning_rate": 6.084150011257239e-06,
      "loss": 0.7625,
      "step": 19133
    },
    {
      "epoch": 3.537818249396023,
      "grad_norm": 1.0196541547775269,
      "learning_rate": 6.079132937538501e-06,
      "loss": 0.6671,
      "step": 19134
    },
    {
      "epoch": 3.5380040884593944,
      "grad_norm": 0.9346649050712585,
      "learning_rate": 6.074117868399509e-06,
      "loss": 0.8232,
      "step": 19135
    },
    {
      "epoch": 3.5381899275227653,
      "grad_norm": 0.9664354920387268,
      "learning_rate": 6.069104803947301e-06,
      "loss": 0.771,
      "step": 19136
    },
    {
      "epoch": 3.538375766586136,
      "grad_norm": 1.1240217685699463,
      "learning_rate": 6.064093744288845e-06,
      "loss": 0.8951,
      "step": 19137
    },
    {
      "epoch": 3.5385616056495075,
      "grad_norm": 1.1051549911499023,
      "learning_rate": 6.059084689531113e-06,
      "loss": 0.8819,
      "step": 19138
    },
    {
      "epoch": 3.538747444712879,
      "grad_norm": 0.8375083804130554,
      "learning_rate": 6.0540776397810085e-06,
      "loss": 0.6488,
      "step": 19139
    },
    {
      "epoch": 3.5389332837762497,
      "grad_norm": 0.900913655757904,
      "learning_rate": 6.049072595145411e-06,
      "loss": 0.9445,
      "step": 19140
    },
    {
      "epoch": 3.539119122839621,
      "grad_norm": 1.3367409706115723,
      "learning_rate": 6.044069555731124e-06,
      "loss": 0.9208,
      "step": 19141
    },
    {
      "epoch": 3.539304961902992,
      "grad_norm": 1.1735326051712036,
      "learning_rate": 6.03906852164492e-06,
      "loss": 1.0493,
      "step": 19142
    },
    {
      "epoch": 3.5394908009663633,
      "grad_norm": 0.918788731098175,
      "learning_rate": 6.034069492993577e-06,
      "loss": 0.7548,
      "step": 19143
    },
    {
      "epoch": 3.539676640029734,
      "grad_norm": 0.8883483409881592,
      "learning_rate": 6.0290724698837675e-06,
      "loss": 0.8226,
      "step": 19144
    },
    {
      "epoch": 3.5398624790931055,
      "grad_norm": 0.931384265422821,
      "learning_rate": 6.024077452422128e-06,
      "loss": 0.779,
      "step": 19145
    },
    {
      "epoch": 3.5400483181564764,
      "grad_norm": 0.8548913598060608,
      "learning_rate": 6.019084440715295e-06,
      "loss": 0.6444,
      "step": 19146
    },
    {
      "epoch": 3.5402341572198477,
      "grad_norm": 1.0747677087783813,
      "learning_rate": 6.014093434869828e-06,
      "loss": 0.8846,
      "step": 19147
    },
    {
      "epoch": 3.5404199962832186,
      "grad_norm": 1.1103484630584717,
      "learning_rate": 6.009104434992263e-06,
      "loss": 0.8992,
      "step": 19148
    },
    {
      "epoch": 3.54060583534659,
      "grad_norm": 1.0426290035247803,
      "learning_rate": 6.004117441189061e-06,
      "loss": 0.9561,
      "step": 19149
    },
    {
      "epoch": 3.5407916744099612,
      "grad_norm": 1.2314473390579224,
      "learning_rate": 5.999132453566658e-06,
      "loss": 1.0631,
      "step": 19150
    },
    {
      "epoch": 3.540977513473332,
      "grad_norm": 0.7971338629722595,
      "learning_rate": 5.994149472231469e-06,
      "loss": 0.6471,
      "step": 19151
    },
    {
      "epoch": 3.541163352536703,
      "grad_norm": 1.167417287826538,
      "learning_rate": 5.989168497289843e-06,
      "loss": 0.823,
      "step": 19152
    },
    {
      "epoch": 3.5413491916000743,
      "grad_norm": 0.9355377554893494,
      "learning_rate": 5.984189528848095e-06,
      "loss": 1.0002,
      "step": 19153
    },
    {
      "epoch": 3.5415350306634457,
      "grad_norm": 0.8810675740242004,
      "learning_rate": 5.97921256701246e-06,
      "loss": 0.8025,
      "step": 19154
    },
    {
      "epoch": 3.5417208697268165,
      "grad_norm": 0.9732177257537842,
      "learning_rate": 5.974237611889189e-06,
      "loss": 1.0295,
      "step": 19155
    },
    {
      "epoch": 3.5419067087901874,
      "grad_norm": 0.8081285357475281,
      "learning_rate": 5.969264663584473e-06,
      "loss": 0.3682,
      "step": 19156
    },
    {
      "epoch": 3.5420925478535588,
      "grad_norm": 0.852075457572937,
      "learning_rate": 5.9642937222044396e-06,
      "loss": 0.7055,
      "step": 19157
    },
    {
      "epoch": 3.54227838691693,
      "grad_norm": 0.8370651602745056,
      "learning_rate": 5.95932478785517e-06,
      "loss": 0.7749,
      "step": 19158
    },
    {
      "epoch": 3.542464225980301,
      "grad_norm": 1.0717309713363647,
      "learning_rate": 5.954357860642745e-06,
      "loss": 0.906,
      "step": 19159
    },
    {
      "epoch": 3.5426500650436723,
      "grad_norm": 1.0112475156784058,
      "learning_rate": 5.9493929406731705e-06,
      "loss": 0.9781,
      "step": 19160
    },
    {
      "epoch": 3.542835904107043,
      "grad_norm": 1.2165931463241577,
      "learning_rate": 5.944430028052405e-06,
      "loss": 0.8503,
      "step": 19161
    },
    {
      "epoch": 3.5430217431704145,
      "grad_norm": 0.9976071119308472,
      "learning_rate": 5.939469122886354e-06,
      "loss": 0.7948,
      "step": 19162
    },
    {
      "epoch": 3.5432075822337854,
      "grad_norm": 0.9213830232620239,
      "learning_rate": 5.934510225280943e-06,
      "loss": 1.0859,
      "step": 19163
    },
    {
      "epoch": 3.5433934212971567,
      "grad_norm": 0.8859527111053467,
      "learning_rate": 5.929553335341986e-06,
      "loss": 1.0171,
      "step": 19164
    },
    {
      "epoch": 3.543579260360528,
      "grad_norm": 0.9154654145240784,
      "learning_rate": 5.9245984531752784e-06,
      "loss": 0.5721,
      "step": 19165
    },
    {
      "epoch": 3.543765099423899,
      "grad_norm": 0.8688372373580933,
      "learning_rate": 5.919645578886545e-06,
      "loss": 0.7146,
      "step": 19166
    },
    {
      "epoch": 3.54395093848727,
      "grad_norm": 0.8358892798423767,
      "learning_rate": 5.91469471258157e-06,
      "loss": 0.8985,
      "step": 19167
    },
    {
      "epoch": 3.544136777550641,
      "grad_norm": 1.0070583820343018,
      "learning_rate": 5.9097458543659666e-06,
      "loss": 0.9028,
      "step": 19168
    },
    {
      "epoch": 3.5443226166140125,
      "grad_norm": 1.22250235080719,
      "learning_rate": 5.904799004345385e-06,
      "loss": 1.0275,
      "step": 19169
    },
    {
      "epoch": 3.5445084556773834,
      "grad_norm": 0.8911054730415344,
      "learning_rate": 5.899854162625362e-06,
      "loss": 0.8251,
      "step": 19170
    },
    {
      "epoch": 3.5446942947407543,
      "grad_norm": 1.0135678052902222,
      "learning_rate": 5.894911329311503e-06,
      "loss": 0.9109,
      "step": 19171
    },
    {
      "epoch": 3.5448801338041256,
      "grad_norm": 0.9569916725158691,
      "learning_rate": 5.889970504509257e-06,
      "loss": 0.9177,
      "step": 19172
    },
    {
      "epoch": 3.545065972867497,
      "grad_norm": 1.3330678939819336,
      "learning_rate": 5.8850316883241055e-06,
      "loss": 0.9676,
      "step": 19173
    },
    {
      "epoch": 3.545251811930868,
      "grad_norm": 0.8885575532913208,
      "learning_rate": 5.88009488086142e-06,
      "loss": 0.7955,
      "step": 19174
    },
    {
      "epoch": 3.545437650994239,
      "grad_norm": 0.9844068288803101,
      "learning_rate": 5.875160082226605e-06,
      "loss": 1.0337,
      "step": 19175
    },
    {
      "epoch": 3.54562349005761,
      "grad_norm": 0.8716814517974854,
      "learning_rate": 5.870227292524966e-06,
      "loss": 0.7287,
      "step": 19176
    },
    {
      "epoch": 3.5458093291209813,
      "grad_norm": 1.133697509765625,
      "learning_rate": 5.865296511861806e-06,
      "loss": 0.9023,
      "step": 19177
    },
    {
      "epoch": 3.5459951681843522,
      "grad_norm": 0.9876354932785034,
      "learning_rate": 5.860367740342332e-06,
      "loss": 0.8509,
      "step": 19178
    },
    {
      "epoch": 3.5461810072477236,
      "grad_norm": 1.0449570417404175,
      "learning_rate": 5.855440978071769e-06,
      "loss": 1.0393,
      "step": 19179
    },
    {
      "epoch": 3.5463668463110944,
      "grad_norm": 1.3608499765396118,
      "learning_rate": 5.850516225155267e-06,
      "loss": 0.8978,
      "step": 19180
    },
    {
      "epoch": 3.5465526853744658,
      "grad_norm": 0.9593074321746826,
      "learning_rate": 5.8455934816979305e-06,
      "loss": 0.6206,
      "step": 19181
    },
    {
      "epoch": 3.5467385244378367,
      "grad_norm": 0.9247384071350098,
      "learning_rate": 5.840672747804809e-06,
      "loss": 0.8468,
      "step": 19182
    },
    {
      "epoch": 3.546924363501208,
      "grad_norm": 0.8710890412330627,
      "learning_rate": 5.835754023580964e-06,
      "loss": 1.0445,
      "step": 19183
    },
    {
      "epoch": 3.5471102025645793,
      "grad_norm": 0.9194334149360657,
      "learning_rate": 5.8308373091313425e-06,
      "loss": 0.8244,
      "step": 19184
    },
    {
      "epoch": 3.54729604162795,
      "grad_norm": 0.7997312545776367,
      "learning_rate": 5.825922604560908e-06,
      "loss": 0.7945,
      "step": 19185
    },
    {
      "epoch": 3.547481880691321,
      "grad_norm": 1.069359540939331,
      "learning_rate": 5.8210099099745284e-06,
      "loss": 0.9528,
      "step": 19186
    },
    {
      "epoch": 3.5476677197546924,
      "grad_norm": 0.9582682251930237,
      "learning_rate": 5.816099225477079e-06,
      "loss": 0.9763,
      "step": 19187
    },
    {
      "epoch": 3.5478535588180637,
      "grad_norm": 0.8646688461303711,
      "learning_rate": 5.811190551173373e-06,
      "loss": 0.8443,
      "step": 19188
    },
    {
      "epoch": 3.5480393978814346,
      "grad_norm": 0.9127521514892578,
      "learning_rate": 5.806283887168185e-06,
      "loss": 0.7724,
      "step": 19189
    },
    {
      "epoch": 3.548225236944806,
      "grad_norm": 0.779879093170166,
      "learning_rate": 5.801379233566195e-06,
      "loss": 0.8479,
      "step": 19190
    },
    {
      "epoch": 3.548411076008177,
      "grad_norm": 0.8569574356079102,
      "learning_rate": 5.796476590472133e-06,
      "loss": 0.7031,
      "step": 19191
    },
    {
      "epoch": 3.548596915071548,
      "grad_norm": 1.0802698135375977,
      "learning_rate": 5.791575957990614e-06,
      "loss": 0.8848,
      "step": 19192
    },
    {
      "epoch": 3.548782754134919,
      "grad_norm": 1.2723355293273926,
      "learning_rate": 5.7866773362262425e-06,
      "loss": 1.2078,
      "step": 19193
    },
    {
      "epoch": 3.5489685931982904,
      "grad_norm": 1.1142456531524658,
      "learning_rate": 5.781780725283558e-06,
      "loss": 0.8916,
      "step": 19194
    },
    {
      "epoch": 3.5491544322616613,
      "grad_norm": 0.7921794652938843,
      "learning_rate": 5.776886125267089e-06,
      "loss": 0.6711,
      "step": 19195
    },
    {
      "epoch": 3.5493402713250326,
      "grad_norm": 0.925909161567688,
      "learning_rate": 5.771993536281284e-06,
      "loss": 0.9102,
      "step": 19196
    },
    {
      "epoch": 3.5495261103884035,
      "grad_norm": 0.9324236512184143,
      "learning_rate": 5.76710295843057e-06,
      "loss": 0.8768,
      "step": 19197
    },
    {
      "epoch": 3.549711949451775,
      "grad_norm": 1.0298572778701782,
      "learning_rate": 5.7622143918193314e-06,
      "loss": 0.7518,
      "step": 19198
    },
    {
      "epoch": 3.549897788515146,
      "grad_norm": 0.8958404660224915,
      "learning_rate": 5.757327836551929e-06,
      "loss": 0.6492,
      "step": 19199
    },
    {
      "epoch": 3.550083627578517,
      "grad_norm": 1.1485790014266968,
      "learning_rate": 5.7524432927326235e-06,
      "loss": 0.7827,
      "step": 19200
    },
    {
      "epoch": 3.550269466641888,
      "grad_norm": 0.872681736946106,
      "learning_rate": 5.747560760465687e-06,
      "loss": 0.8896,
      "step": 19201
    },
    {
      "epoch": 3.5504553057052592,
      "grad_norm": 0.8319202065467834,
      "learning_rate": 5.742680239855314e-06,
      "loss": 0.5662,
      "step": 19202
    },
    {
      "epoch": 3.5506411447686306,
      "grad_norm": 0.9280526638031006,
      "learning_rate": 5.737801731005688e-06,
      "loss": 0.7522,
      "step": 19203
    },
    {
      "epoch": 3.5508269838320015,
      "grad_norm": 1.0687966346740723,
      "learning_rate": 5.732925234020925e-06,
      "loss": 0.9029,
      "step": 19204
    },
    {
      "epoch": 3.5510128228953723,
      "grad_norm": 0.9236042499542236,
      "learning_rate": 5.7280507490050985e-06,
      "loss": 0.8034,
      "step": 19205
    },
    {
      "epoch": 3.5511986619587437,
      "grad_norm": 0.9122721552848816,
      "learning_rate": 5.723178276062247e-06,
      "loss": 0.7589,
      "step": 19206
    },
    {
      "epoch": 3.551384501022115,
      "grad_norm": 0.86166912317276,
      "learning_rate": 5.718307815296375e-06,
      "loss": 0.8254,
      "step": 19207
    },
    {
      "epoch": 3.551570340085486,
      "grad_norm": 1.1353827714920044,
      "learning_rate": 5.713439366811413e-06,
      "loss": 0.6789,
      "step": 19208
    },
    {
      "epoch": 3.551756179148857,
      "grad_norm": 0.8476354479789734,
      "learning_rate": 5.708572930711309e-06,
      "loss": 0.6649,
      "step": 19209
    },
    {
      "epoch": 3.551942018212228,
      "grad_norm": 0.8318846225738525,
      "learning_rate": 5.703708507099881e-06,
      "loss": 0.7236,
      "step": 19210
    },
    {
      "epoch": 3.5521278572755994,
      "grad_norm": 0.9746939539909363,
      "learning_rate": 5.698846096081001e-06,
      "loss": 0.8217,
      "step": 19211
    },
    {
      "epoch": 3.5523136963389703,
      "grad_norm": 0.9182035326957703,
      "learning_rate": 5.693985697758419e-06,
      "loss": 0.7409,
      "step": 19212
    },
    {
      "epoch": 3.5524995354023416,
      "grad_norm": 1.098073959350586,
      "learning_rate": 5.689127312235876e-06,
      "loss": 0.8893,
      "step": 19213
    },
    {
      "epoch": 3.5526853744657125,
      "grad_norm": 0.9553743600845337,
      "learning_rate": 5.684270939617054e-06,
      "loss": 0.8132,
      "step": 19214
    },
    {
      "epoch": 3.552871213529084,
      "grad_norm": 1.0023797750473022,
      "learning_rate": 5.679416580005626e-06,
      "loss": 0.8384,
      "step": 19215
    },
    {
      "epoch": 3.5530570525924547,
      "grad_norm": 0.8118361830711365,
      "learning_rate": 5.6745642335051976e-06,
      "loss": 0.6801,
      "step": 19216
    },
    {
      "epoch": 3.553242891655826,
      "grad_norm": 1.0483431816101074,
      "learning_rate": 5.669713900219309e-06,
      "loss": 0.9402,
      "step": 19217
    },
    {
      "epoch": 3.5534287307191974,
      "grad_norm": 0.924910843372345,
      "learning_rate": 5.664865580251499e-06,
      "loss": 0.7887,
      "step": 19218
    },
    {
      "epoch": 3.5536145697825683,
      "grad_norm": 0.9264012575149536,
      "learning_rate": 5.660019273705264e-06,
      "loss": 0.9293,
      "step": 19219
    },
    {
      "epoch": 3.553800408845939,
      "grad_norm": 0.9317333102226257,
      "learning_rate": 5.65517498068402e-06,
      "loss": 0.9858,
      "step": 19220
    },
    {
      "epoch": 3.5539862479093105,
      "grad_norm": 0.9699829816818237,
      "learning_rate": 5.6503327012911725e-06,
      "loss": 0.8917,
      "step": 19221
    },
    {
      "epoch": 3.554172086972682,
      "grad_norm": 0.7067593336105347,
      "learning_rate": 5.64549243563004e-06,
      "loss": 0.5219,
      "step": 19222
    },
    {
      "epoch": 3.5543579260360527,
      "grad_norm": 0.9304884076118469,
      "learning_rate": 5.640654183803962e-06,
      "loss": 0.6241,
      "step": 19223
    },
    {
      "epoch": 3.554543765099424,
      "grad_norm": 0.8964663147926331,
      "learning_rate": 5.635817945916211e-06,
      "loss": 0.7789,
      "step": 19224
    },
    {
      "epoch": 3.554729604162795,
      "grad_norm": 0.8794653415679932,
      "learning_rate": 5.630983722069972e-06,
      "loss": 0.763,
      "step": 19225
    },
    {
      "epoch": 3.5549154432261663,
      "grad_norm": 0.8196675181388855,
      "learning_rate": 5.626151512368438e-06,
      "loss": 0.9579,
      "step": 19226
    },
    {
      "epoch": 3.555101282289537,
      "grad_norm": 0.9369693398475647,
      "learning_rate": 5.621321316914763e-06,
      "loss": 0.8541,
      "step": 19227
    },
    {
      "epoch": 3.5552871213529085,
      "grad_norm": 0.9243240356445312,
      "learning_rate": 5.616493135812018e-06,
      "loss": 0.7995,
      "step": 19228
    },
    {
      "epoch": 3.5554729604162794,
      "grad_norm": 0.8889716863632202,
      "learning_rate": 5.611666969163243e-06,
      "loss": 0.8044,
      "step": 19229
    },
    {
      "epoch": 3.5556587994796507,
      "grad_norm": 0.8678849935531616,
      "learning_rate": 5.606842817071467e-06,
      "loss": 0.8267,
      "step": 19230
    },
    {
      "epoch": 3.5558446385430216,
      "grad_norm": 0.9801349639892578,
      "learning_rate": 5.602020679639663e-06,
      "loss": 0.8801,
      "step": 19231
    },
    {
      "epoch": 3.556030477606393,
      "grad_norm": 0.8349409699440002,
      "learning_rate": 5.597200556970727e-06,
      "loss": 0.8689,
      "step": 19232
    },
    {
      "epoch": 3.5562163166697642,
      "grad_norm": 0.9708721041679382,
      "learning_rate": 5.592382449167543e-06,
      "loss": 0.7498,
      "step": 19233
    },
    {
      "epoch": 3.556402155733135,
      "grad_norm": 0.9165699481964111,
      "learning_rate": 5.58756635633293e-06,
      "loss": 0.892,
      "step": 19234
    },
    {
      "epoch": 3.556587994796506,
      "grad_norm": 0.9796352982521057,
      "learning_rate": 5.5827522785697136e-06,
      "loss": 0.7933,
      "step": 19235
    },
    {
      "epoch": 3.5567738338598773,
      "grad_norm": 0.8175069093704224,
      "learning_rate": 5.577940215980626e-06,
      "loss": 0.8168,
      "step": 19236
    },
    {
      "epoch": 3.5569596729232487,
      "grad_norm": 1.0237641334533691,
      "learning_rate": 5.57313016866835e-06,
      "loss": 0.8044,
      "step": 19237
    },
    {
      "epoch": 3.5571455119866195,
      "grad_norm": 0.9125300645828247,
      "learning_rate": 5.568322136735582e-06,
      "loss": 0.6959,
      "step": 19238
    },
    {
      "epoch": 3.5573313510499904,
      "grad_norm": 1.0325636863708496,
      "learning_rate": 5.5635161202849175e-06,
      "loss": 0.8413,
      "step": 19239
    },
    {
      "epoch": 3.5575171901133618,
      "grad_norm": 0.9485664963722229,
      "learning_rate": 5.558712119418952e-06,
      "loss": 0.7329,
      "step": 19240
    },
    {
      "epoch": 3.557703029176733,
      "grad_norm": 0.8749877214431763,
      "learning_rate": 5.553910134240203e-06,
      "loss": 0.8431,
      "step": 19241
    },
    {
      "epoch": 3.557888868240104,
      "grad_norm": 1.127166986465454,
      "learning_rate": 5.549110164851157e-06,
      "loss": 0.8702,
      "step": 19242
    },
    {
      "epoch": 3.5580747073034753,
      "grad_norm": 0.9633110761642456,
      "learning_rate": 5.544312211354286e-06,
      "loss": 0.9324,
      "step": 19243
    },
    {
      "epoch": 3.558260546366846,
      "grad_norm": 0.9466096758842468,
      "learning_rate": 5.539516273851986e-06,
      "loss": 0.7709,
      "step": 19244
    },
    {
      "epoch": 3.5584463854302175,
      "grad_norm": 0.8669358491897583,
      "learning_rate": 5.534722352446598e-06,
      "loss": 0.8215,
      "step": 19245
    },
    {
      "epoch": 3.5586322244935884,
      "grad_norm": 0.8715214133262634,
      "learning_rate": 5.529930447240439e-06,
      "loss": 0.6906,
      "step": 19246
    },
    {
      "epoch": 3.5588180635569597,
      "grad_norm": 1.3729928731918335,
      "learning_rate": 5.525140558335817e-06,
      "loss": 1.1958,
      "step": 19247
    },
    {
      "epoch": 3.559003902620331,
      "grad_norm": 0.9086440801620483,
      "learning_rate": 5.52035268583494e-06,
      "loss": 0.7287,
      "step": 19248
    },
    {
      "epoch": 3.559189741683702,
      "grad_norm": 1.0862767696380615,
      "learning_rate": 5.515566829839991e-06,
      "loss": 0.7094,
      "step": 19249
    },
    {
      "epoch": 3.559375580747073,
      "grad_norm": 0.8084904551506042,
      "learning_rate": 5.510782990453123e-06,
      "loss": 0.9409,
      "step": 19250
    },
    {
      "epoch": 3.559561419810444,
      "grad_norm": 0.960584819316864,
      "learning_rate": 5.506001167776464e-06,
      "loss": 0.6924,
      "step": 19251
    },
    {
      "epoch": 3.5597472588738155,
      "grad_norm": 1.1462523937225342,
      "learning_rate": 5.501221361912046e-06,
      "loss": 1.0633,
      "step": 19252
    },
    {
      "epoch": 3.5599330979371864,
      "grad_norm": 0.9685887098312378,
      "learning_rate": 5.496443572961895e-06,
      "loss": 0.7463,
      "step": 19253
    },
    {
      "epoch": 3.5601189370005573,
      "grad_norm": 0.8634170889854431,
      "learning_rate": 5.491667801027955e-06,
      "loss": 0.9771,
      "step": 19254
    },
    {
      "epoch": 3.5603047760639286,
      "grad_norm": 0.8998491168022156,
      "learning_rate": 5.4868940462122094e-06,
      "loss": 0.8428,
      "step": 19255
    },
    {
      "epoch": 3.5604906151273,
      "grad_norm": 0.7958370447158813,
      "learning_rate": 5.482122308616522e-06,
      "loss": 0.6143,
      "step": 19256
    },
    {
      "epoch": 3.560676454190671,
      "grad_norm": 1.019345998764038,
      "learning_rate": 5.47735258834271e-06,
      "loss": 0.7629,
      "step": 19257
    },
    {
      "epoch": 3.560862293254042,
      "grad_norm": 0.9218160510063171,
      "learning_rate": 5.472584885492616e-06,
      "loss": 0.8334,
      "step": 19258
    },
    {
      "epoch": 3.561048132317413,
      "grad_norm": 1.104111909866333,
      "learning_rate": 5.467819200167989e-06,
      "loss": 0.9566,
      "step": 19259
    },
    {
      "epoch": 3.5612339713807843,
      "grad_norm": 0.9276748299598694,
      "learning_rate": 5.463055532470518e-06,
      "loss": 0.7049,
      "step": 19260
    },
    {
      "epoch": 3.5614198104441552,
      "grad_norm": 0.9067643880844116,
      "learning_rate": 5.458293882501908e-06,
      "loss": 0.8014,
      "step": 19261
    },
    {
      "epoch": 3.5616056495075266,
      "grad_norm": 1.0602048635482788,
      "learning_rate": 5.4535342503637675e-06,
      "loss": 0.8069,
      "step": 19262
    },
    {
      "epoch": 3.5617914885708974,
      "grad_norm": 0.9947314858436584,
      "learning_rate": 5.448776636157693e-06,
      "loss": 0.8717,
      "step": 19263
    },
    {
      "epoch": 3.5619773276342688,
      "grad_norm": 1.2373080253601074,
      "learning_rate": 5.444021039985236e-06,
      "loss": 0.8202,
      "step": 19264
    },
    {
      "epoch": 3.5621631666976397,
      "grad_norm": 0.9831281304359436,
      "learning_rate": 5.439267461947883e-06,
      "loss": 0.7609,
      "step": 19265
    },
    {
      "epoch": 3.562349005761011,
      "grad_norm": 0.9348733425140381,
      "learning_rate": 5.434515902147075e-06,
      "loss": 0.7215,
      "step": 19266
    },
    {
      "epoch": 3.5625348448243823,
      "grad_norm": 0.9554674029350281,
      "learning_rate": 5.429766360684252e-06,
      "loss": 0.8535,
      "step": 19267
    },
    {
      "epoch": 3.562720683887753,
      "grad_norm": 1.0535904169082642,
      "learning_rate": 5.425018837660789e-06,
      "loss": 1.0426,
      "step": 19268
    },
    {
      "epoch": 3.562906522951124,
      "grad_norm": 1.0819015502929688,
      "learning_rate": 5.420273333177983e-06,
      "loss": 0.8682,
      "step": 19269
    },
    {
      "epoch": 3.5630923620144954,
      "grad_norm": 1.0718095302581787,
      "learning_rate": 5.415529847337153e-06,
      "loss": 1.1112,
      "step": 19270
    },
    {
      "epoch": 3.5632782010778667,
      "grad_norm": 0.8572877049446106,
      "learning_rate": 5.410788380239507e-06,
      "loss": 0.9122,
      "step": 19271
    },
    {
      "epoch": 3.5634640401412376,
      "grad_norm": 1.0377191305160522,
      "learning_rate": 5.406048931986285e-06,
      "loss": 0.4833,
      "step": 19272
    },
    {
      "epoch": 3.563649879204609,
      "grad_norm": 1.1077818870544434,
      "learning_rate": 5.401311502678608e-06,
      "loss": 0.7051,
      "step": 19273
    },
    {
      "epoch": 3.56383571826798,
      "grad_norm": 0.8639101386070251,
      "learning_rate": 5.396576092417593e-06,
      "loss": 0.7289,
      "step": 19274
    },
    {
      "epoch": 3.564021557331351,
      "grad_norm": 1.079602837562561,
      "learning_rate": 5.391842701304328e-06,
      "loss": 0.6417,
      "step": 19275
    },
    {
      "epoch": 3.564207396394722,
      "grad_norm": 0.9743924736976624,
      "learning_rate": 5.38711132943982e-06,
      "loss": 0.9016,
      "step": 19276
    },
    {
      "epoch": 3.5643932354580934,
      "grad_norm": 1.0895919799804688,
      "learning_rate": 5.382381976925044e-06,
      "loss": 0.8508,
      "step": 19277
    },
    {
      "epoch": 3.5645790745214643,
      "grad_norm": 0.9121440052986145,
      "learning_rate": 5.377654643860974e-06,
      "loss": 0.6967,
      "step": 19278
    },
    {
      "epoch": 3.5647649135848356,
      "grad_norm": 1.0038836002349854,
      "learning_rate": 5.3729293303484864e-06,
      "loss": 0.9408,
      "step": 19279
    },
    {
      "epoch": 3.5649507526482065,
      "grad_norm": 0.9119470119476318,
      "learning_rate": 5.368206036488443e-06,
      "loss": 0.8108,
      "step": 19280
    },
    {
      "epoch": 3.565136591711578,
      "grad_norm": 0.9499068856239319,
      "learning_rate": 5.3634847623816205e-06,
      "loss": 0.5967,
      "step": 19281
    },
    {
      "epoch": 3.565322430774949,
      "grad_norm": 0.8114598393440247,
      "learning_rate": 5.358765508128816e-06,
      "loss": 0.6871,
      "step": 19282
    },
    {
      "epoch": 3.56550826983832,
      "grad_norm": 1.0090092420578003,
      "learning_rate": 5.35404827383077e-06,
      "loss": 0.9781,
      "step": 19283
    },
    {
      "epoch": 3.565694108901691,
      "grad_norm": 0.961084246635437,
      "learning_rate": 5.349333059588146e-06,
      "loss": 0.8388,
      "step": 19284
    },
    {
      "epoch": 3.5658799479650622,
      "grad_norm": 0.9709075093269348,
      "learning_rate": 5.3446198655015765e-06,
      "loss": 0.9357,
      "step": 19285
    },
    {
      "epoch": 3.5660657870284336,
      "grad_norm": 0.8814340233802795,
      "learning_rate": 5.339908691671647e-06,
      "loss": 0.7196,
      "step": 19286
    },
    {
      "epoch": 3.5662516260918045,
      "grad_norm": 1.1103253364562988,
      "learning_rate": 5.335199538198932e-06,
      "loss": 0.884,
      "step": 19287
    },
    {
      "epoch": 3.5664374651551753,
      "grad_norm": 0.9542502164840698,
      "learning_rate": 5.33049240518394e-06,
      "loss": 0.8492,
      "step": 19288
    },
    {
      "epoch": 3.5666233042185467,
      "grad_norm": 1.0562374591827393,
      "learning_rate": 5.325787292727113e-06,
      "loss": 0.7911,
      "step": 19289
    },
    {
      "epoch": 3.566809143281918,
      "grad_norm": 1.0418407917022705,
      "learning_rate": 5.3210842009288945e-06,
      "loss": 0.6607,
      "step": 19290
    },
    {
      "epoch": 3.566994982345289,
      "grad_norm": 0.7913893461227417,
      "learning_rate": 5.316383129889668e-06,
      "loss": 0.6634,
      "step": 19291
    },
    {
      "epoch": 3.56718082140866,
      "grad_norm": 0.8778683543205261,
      "learning_rate": 5.311684079709745e-06,
      "loss": 0.7286,
      "step": 19292
    },
    {
      "epoch": 3.567366660472031,
      "grad_norm": 1.1600593328475952,
      "learning_rate": 5.306987050489442e-06,
      "loss": 0.741,
      "step": 19293
    },
    {
      "epoch": 3.5675524995354024,
      "grad_norm": 0.8523426055908203,
      "learning_rate": 5.302292042328982e-06,
      "loss": 0.8115,
      "step": 19294
    },
    {
      "epoch": 3.5677383385987733,
      "grad_norm": 0.8197619318962097,
      "learning_rate": 5.297599055328617e-06,
      "loss": 0.8362,
      "step": 19295
    },
    {
      "epoch": 3.5679241776621446,
      "grad_norm": 1.0496013164520264,
      "learning_rate": 5.292908089588478e-06,
      "loss": 0.8323,
      "step": 19296
    },
    {
      "epoch": 3.568110016725516,
      "grad_norm": 0.9278433322906494,
      "learning_rate": 5.2882191452086834e-06,
      "loss": 0.8625,
      "step": 19297
    },
    {
      "epoch": 3.568295855788887,
      "grad_norm": 1.8732656240463257,
      "learning_rate": 5.283532222289322e-06,
      "loss": 1.3532,
      "step": 19298
    },
    {
      "epoch": 3.5684816948522577,
      "grad_norm": 0.8991494178771973,
      "learning_rate": 5.278847320930436e-06,
      "loss": 0.6995,
      "step": 19299
    },
    {
      "epoch": 3.568667533915629,
      "grad_norm": 1.0465878248214722,
      "learning_rate": 5.274164441231999e-06,
      "loss": 1.0019,
      "step": 19300
    },
    {
      "epoch": 3.5688533729790004,
      "grad_norm": 0.9910255670547485,
      "learning_rate": 5.269483583293966e-06,
      "loss": 0.8754,
      "step": 19301
    },
    {
      "epoch": 3.5690392120423713,
      "grad_norm": 0.9528656005859375,
      "learning_rate": 5.264804747216223e-06,
      "loss": 0.7644,
      "step": 19302
    },
    {
      "epoch": 3.569225051105742,
      "grad_norm": 0.8942814469337463,
      "learning_rate": 5.26012793309868e-06,
      "loss": 0.952,
      "step": 19303
    },
    {
      "epoch": 3.5694108901691135,
      "grad_norm": 1.0535019636154175,
      "learning_rate": 5.255453141041111e-06,
      "loss": 1.0923,
      "step": 19304
    },
    {
      "epoch": 3.569596729232485,
      "grad_norm": 1.106765627861023,
      "learning_rate": 5.250780371143315e-06,
      "loss": 0.7056,
      "step": 19305
    },
    {
      "epoch": 3.5697825682958557,
      "grad_norm": 0.9234157204627991,
      "learning_rate": 5.246109623505002e-06,
      "loss": 0.8472,
      "step": 19306
    },
    {
      "epoch": 3.569968407359227,
      "grad_norm": 1.0718744993209839,
      "learning_rate": 5.241440898225891e-06,
      "loss": 1.0636,
      "step": 19307
    },
    {
      "epoch": 3.570154246422598,
      "grad_norm": 0.8861552476882935,
      "learning_rate": 5.236774195405614e-06,
      "loss": 0.8399,
      "step": 19308
    },
    {
      "epoch": 3.5703400854859693,
      "grad_norm": 0.9062960743904114,
      "learning_rate": 5.232109515143746e-06,
      "loss": 0.6951,
      "step": 19309
    },
    {
      "epoch": 3.57052592454934,
      "grad_norm": 1.8155157566070557,
      "learning_rate": 5.227446857539897e-06,
      "loss": 1.4572,
      "step": 19310
    },
    {
      "epoch": 3.5707117636127115,
      "grad_norm": 0.938761830329895,
      "learning_rate": 5.222786222693554e-06,
      "loss": 0.7112,
      "step": 19311
    },
    {
      "epoch": 3.5708976026760824,
      "grad_norm": 0.9258983135223389,
      "learning_rate": 5.218127610704194e-06,
      "loss": 0.6857,
      "step": 19312
    },
    {
      "epoch": 3.5710834417394537,
      "grad_norm": 0.8706833720207214,
      "learning_rate": 5.213471021671235e-06,
      "loss": 0.622,
      "step": 19313
    },
    {
      "epoch": 3.5712692808028246,
      "grad_norm": 0.9127697348594666,
      "learning_rate": 5.208816455694077e-06,
      "loss": 0.7618,
      "step": 19314
    },
    {
      "epoch": 3.571455119866196,
      "grad_norm": 0.7951032519340515,
      "learning_rate": 5.204163912872073e-06,
      "loss": 0.8307,
      "step": 19315
    },
    {
      "epoch": 3.5716409589295672,
      "grad_norm": 0.9423043727874756,
      "learning_rate": 5.1995133933045225e-06,
      "loss": 0.4945,
      "step": 19316
    },
    {
      "epoch": 3.571826797992938,
      "grad_norm": 0.9046568870544434,
      "learning_rate": 5.194864897090656e-06,
      "loss": 0.8827,
      "step": 19317
    },
    {
      "epoch": 3.572012637056309,
      "grad_norm": 0.9737492203712463,
      "learning_rate": 5.190218424329718e-06,
      "loss": 0.7078,
      "step": 19318
    },
    {
      "epoch": 3.5721984761196803,
      "grad_norm": 0.9893245697021484,
      "learning_rate": 5.185573975120872e-06,
      "loss": 0.8561,
      "step": 19319
    },
    {
      "epoch": 3.5723843151830517,
      "grad_norm": 0.8789792060852051,
      "learning_rate": 5.1809315495632394e-06,
      "loss": 0.713,
      "step": 19320
    },
    {
      "epoch": 3.5725701542464225,
      "grad_norm": 0.9256452918052673,
      "learning_rate": 5.1762911477558965e-06,
      "loss": 0.8957,
      "step": 19321
    },
    {
      "epoch": 3.572755993309794,
      "grad_norm": 0.8583182692527771,
      "learning_rate": 5.171652769797908e-06,
      "loss": 0.8656,
      "step": 19322
    },
    {
      "epoch": 3.5729418323731648,
      "grad_norm": 1.0559611320495605,
      "learning_rate": 5.167016415788239e-06,
      "loss": 0.982,
      "step": 19323
    },
    {
      "epoch": 3.573127671436536,
      "grad_norm": 1.0091091394424438,
      "learning_rate": 5.162382085825878e-06,
      "loss": 1.0239,
      "step": 19324
    },
    {
      "epoch": 3.573313510499907,
      "grad_norm": 0.9123789668083191,
      "learning_rate": 5.157749780009735e-06,
      "loss": 0.5609,
      "step": 19325
    },
    {
      "epoch": 3.5734993495632783,
      "grad_norm": 0.9494024515151978,
      "learning_rate": 5.153119498438641e-06,
      "loss": 0.6981,
      "step": 19326
    },
    {
      "epoch": 3.573685188626649,
      "grad_norm": 0.9123292565345764,
      "learning_rate": 5.148491241211473e-06,
      "loss": 0.7819,
      "step": 19327
    },
    {
      "epoch": 3.5738710276900205,
      "grad_norm": 0.8726586699485779,
      "learning_rate": 5.143865008426973e-06,
      "loss": 0.7431,
      "step": 19328
    },
    {
      "epoch": 3.5740568667533914,
      "grad_norm": 1.0286791324615479,
      "learning_rate": 5.139240800183897e-06,
      "loss": 0.9894,
      "step": 19329
    },
    {
      "epoch": 3.5742427058167627,
      "grad_norm": 0.9350249767303467,
      "learning_rate": 5.134618616580944e-06,
      "loss": 0.8188,
      "step": 19330
    },
    {
      "epoch": 3.574428544880134,
      "grad_norm": 1.010162591934204,
      "learning_rate": 5.129998457716756e-06,
      "loss": 0.8046,
      "step": 19331
    },
    {
      "epoch": 3.574614383943505,
      "grad_norm": 0.9536052346229553,
      "learning_rate": 5.125380323689954e-06,
      "loss": 0.7729,
      "step": 19332
    },
    {
      "epoch": 3.574800223006876,
      "grad_norm": 0.8379716277122498,
      "learning_rate": 5.120764214599083e-06,
      "loss": 0.9196,
      "step": 19333
    },
    {
      "epoch": 3.574986062070247,
      "grad_norm": 1.0392776727676392,
      "learning_rate": 5.116150130542685e-06,
      "loss": 0.8505,
      "step": 19334
    },
    {
      "epoch": 3.5751719011336185,
      "grad_norm": 0.7669402360916138,
      "learning_rate": 5.111538071619249e-06,
      "loss": 0.61,
      "step": 19335
    },
    {
      "epoch": 3.5753577401969894,
      "grad_norm": 0.9669473171234131,
      "learning_rate": 5.106928037927195e-06,
      "loss": 0.9057,
      "step": 19336
    },
    {
      "epoch": 3.5755435792603603,
      "grad_norm": 1.1792256832122803,
      "learning_rate": 5.1023200295649e-06,
      "loss": 0.9097,
      "step": 19337
    },
    {
      "epoch": 3.5757294183237316,
      "grad_norm": 0.934672474861145,
      "learning_rate": 5.097714046630753e-06,
      "loss": 1.0523,
      "step": 19338
    },
    {
      "epoch": 3.575915257387103,
      "grad_norm": 0.9194068908691406,
      "learning_rate": 5.09311008922303e-06,
      "loss": 0.8906,
      "step": 19339
    },
    {
      "epoch": 3.576101096450474,
      "grad_norm": 0.9827641248703003,
      "learning_rate": 5.0885081574400195e-06,
      "loss": 0.985,
      "step": 19340
    },
    {
      "epoch": 3.576286935513845,
      "grad_norm": 1.0495479106903076,
      "learning_rate": 5.08390825137991e-06,
      "loss": 0.8759,
      "step": 19341
    },
    {
      "epoch": 3.576472774577216,
      "grad_norm": 1.0074039697647095,
      "learning_rate": 5.0793103711409e-06,
      "loss": 0.6792,
      "step": 19342
    },
    {
      "epoch": 3.5766586136405873,
      "grad_norm": 0.9039619565010071,
      "learning_rate": 5.074714516821133e-06,
      "loss": 0.6722,
      "step": 19343
    },
    {
      "epoch": 3.5768444527039582,
      "grad_norm": 1.0133075714111328,
      "learning_rate": 5.070120688518665e-06,
      "loss": 0.8542,
      "step": 19344
    },
    {
      "epoch": 3.5770302917673296,
      "grad_norm": 0.8256039023399353,
      "learning_rate": 5.065528886331583e-06,
      "loss": 0.6006,
      "step": 19345
    },
    {
      "epoch": 3.5772161308307004,
      "grad_norm": 1.1857333183288574,
      "learning_rate": 5.060939110357854e-06,
      "loss": 1.0767,
      "step": 19346
    },
    {
      "epoch": 3.5774019698940718,
      "grad_norm": 0.97385573387146,
      "learning_rate": 5.056351360695477e-06,
      "loss": 0.816,
      "step": 19347
    },
    {
      "epoch": 3.5775878089574427,
      "grad_norm": 0.8637109398841858,
      "learning_rate": 5.051765637442351e-06,
      "loss": 0.9806,
      "step": 19348
    },
    {
      "epoch": 3.577773648020814,
      "grad_norm": 0.7983254790306091,
      "learning_rate": 5.047181940696333e-06,
      "loss": 0.5248,
      "step": 19349
    },
    {
      "epoch": 3.5779594870841853,
      "grad_norm": 0.8218321204185486,
      "learning_rate": 5.042600270555287e-06,
      "loss": 0.8291,
      "step": 19350
    },
    {
      "epoch": 3.578145326147556,
      "grad_norm": 0.9524140954017639,
      "learning_rate": 5.03802062711698e-06,
      "loss": 0.8874,
      "step": 19351
    },
    {
      "epoch": 3.578331165210927,
      "grad_norm": 0.8764885067939758,
      "learning_rate": 5.0334430104791664e-06,
      "loss": 0.8199,
      "step": 19352
    },
    {
      "epoch": 3.5785170042742984,
      "grad_norm": 0.986173689365387,
      "learning_rate": 5.028867420739524e-06,
      "loss": 0.7779,
      "step": 19353
    },
    {
      "epoch": 3.5787028433376697,
      "grad_norm": 1.0108602046966553,
      "learning_rate": 5.0242938579957435e-06,
      "loss": 0.7467,
      "step": 19354
    },
    {
      "epoch": 3.5788886824010406,
      "grad_norm": 0.970950186252594,
      "learning_rate": 5.0197223223454106e-06,
      "loss": 0.9147,
      "step": 19355
    },
    {
      "epoch": 3.579074521464412,
      "grad_norm": 0.9849465489387512,
      "learning_rate": 5.015152813886137e-06,
      "loss": 0.9148,
      "step": 19356
    },
    {
      "epoch": 3.579260360527783,
      "grad_norm": 0.9472989439964294,
      "learning_rate": 5.0105853327154e-06,
      "loss": 0.8742,
      "step": 19357
    },
    {
      "epoch": 3.579446199591154,
      "grad_norm": 0.8969159722328186,
      "learning_rate": 5.0060198789307345e-06,
      "loss": 0.7965,
      "step": 19358
    },
    {
      "epoch": 3.579632038654525,
      "grad_norm": 1.0213388204574585,
      "learning_rate": 5.001456452629549e-06,
      "loss": 0.7155,
      "step": 19359
    },
    {
      "epoch": 3.5798178777178964,
      "grad_norm": 0.7637826204299927,
      "learning_rate": 4.996895053909256e-06,
      "loss": 0.5367,
      "step": 19360
    },
    {
      "epoch": 3.5800037167812673,
      "grad_norm": 1.0223419666290283,
      "learning_rate": 4.992335682867188e-06,
      "loss": 0.7434,
      "step": 19361
    },
    {
      "epoch": 3.5801895558446386,
      "grad_norm": 1.0513916015625,
      "learning_rate": 4.9877783396006906e-06,
      "loss": 0.7295,
      "step": 19362
    },
    {
      "epoch": 3.5803753949080095,
      "grad_norm": 0.9246789216995239,
      "learning_rate": 4.983223024207029e-06,
      "loss": 0.6569,
      "step": 19363
    },
    {
      "epoch": 3.580561233971381,
      "grad_norm": 1.0383719205856323,
      "learning_rate": 4.978669736783404e-06,
      "loss": 0.8964,
      "step": 19364
    },
    {
      "epoch": 3.580747073034752,
      "grad_norm": 0.8949868679046631,
      "learning_rate": 4.974118477426992e-06,
      "loss": 0.8108,
      "step": 19365
    },
    {
      "epoch": 3.580932912098123,
      "grad_norm": 0.9759200811386108,
      "learning_rate": 4.969569246234962e-06,
      "loss": 0.7496,
      "step": 19366
    },
    {
      "epoch": 3.581118751161494,
      "grad_norm": 0.9014949202537537,
      "learning_rate": 4.965022043304413e-06,
      "loss": 0.7466,
      "step": 19367
    },
    {
      "epoch": 3.5813045902248652,
      "grad_norm": 0.8902834057807922,
      "learning_rate": 4.9604768687323664e-06,
      "loss": 0.954,
      "step": 19368
    },
    {
      "epoch": 3.5814904292882366,
      "grad_norm": 0.8954604268074036,
      "learning_rate": 4.955933722615847e-06,
      "loss": 0.6673,
      "step": 19369
    },
    {
      "epoch": 3.5816762683516075,
      "grad_norm": 0.9591830372810364,
      "learning_rate": 4.95139260505183e-06,
      "loss": 0.8093,
      "step": 19370
    },
    {
      "epoch": 3.581862107414979,
      "grad_norm": 0.8783988356590271,
      "learning_rate": 4.946853516137229e-06,
      "loss": 0.6318,
      "step": 19371
    },
    {
      "epoch": 3.5820479464783497,
      "grad_norm": 1.4477580785751343,
      "learning_rate": 4.94231645596892e-06,
      "loss": 1.3058,
      "step": 19372
    },
    {
      "epoch": 3.582233785541721,
      "grad_norm": 1.1015466451644897,
      "learning_rate": 4.937781424643728e-06,
      "loss": 0.8462,
      "step": 19373
    },
    {
      "epoch": 3.582419624605092,
      "grad_norm": 0.9507879614830017,
      "learning_rate": 4.933248422258463e-06,
      "loss": 0.8298,
      "step": 19374
    },
    {
      "epoch": 3.582605463668463,
      "grad_norm": 0.9847450852394104,
      "learning_rate": 4.928717448909881e-06,
      "loss": 0.9353,
      "step": 19375
    },
    {
      "epoch": 3.582791302731834,
      "grad_norm": 0.8292127251625061,
      "learning_rate": 4.92418850469466e-06,
      "loss": 0.8207,
      "step": 19376
    },
    {
      "epoch": 3.5829771417952054,
      "grad_norm": 0.8937145471572876,
      "learning_rate": 4.919661589709479e-06,
      "loss": 0.7023,
      "step": 19377
    },
    {
      "epoch": 3.5831629808585763,
      "grad_norm": 1.0915123224258423,
      "learning_rate": 4.915136704050971e-06,
      "loss": 0.7531,
      "step": 19378
    },
    {
      "epoch": 3.5833488199219476,
      "grad_norm": 0.9832667112350464,
      "learning_rate": 4.910613847815693e-06,
      "loss": 0.8312,
      "step": 19379
    },
    {
      "epoch": 3.583534658985319,
      "grad_norm": 1.0424070358276367,
      "learning_rate": 4.906093021100189e-06,
      "loss": 0.7697,
      "step": 19380
    },
    {
      "epoch": 3.58372049804869,
      "grad_norm": 0.8419145941734314,
      "learning_rate": 4.901574224000927e-06,
      "loss": 0.8657,
      "step": 19381
    },
    {
      "epoch": 3.5839063371120607,
      "grad_norm": 0.8672407865524292,
      "learning_rate": 4.897057456614373e-06,
      "loss": 0.8676,
      "step": 19382
    },
    {
      "epoch": 3.584092176175432,
      "grad_norm": 0.8903321623802185,
      "learning_rate": 4.89254271903693e-06,
      "loss": 0.7165,
      "step": 19383
    },
    {
      "epoch": 3.5842780152388034,
      "grad_norm": 0.9509225487709045,
      "learning_rate": 4.888030011364953e-06,
      "loss": 0.6107,
      "step": 19384
    },
    {
      "epoch": 3.5844638543021743,
      "grad_norm": 1.4456324577331543,
      "learning_rate": 4.883519333694742e-06,
      "loss": 0.9916,
      "step": 19385
    },
    {
      "epoch": 3.584649693365545,
      "grad_norm": 1.193587064743042,
      "learning_rate": 4.879010686122587e-06,
      "loss": 0.8602,
      "step": 19386
    },
    {
      "epoch": 3.5848355324289165,
      "grad_norm": 1.0764076709747314,
      "learning_rate": 4.874504068744723e-06,
      "loss": 0.6382,
      "step": 19387
    },
    {
      "epoch": 3.585021371492288,
      "grad_norm": 0.8120892643928528,
      "learning_rate": 4.869999481657339e-06,
      "loss": 0.8553,
      "step": 19388
    },
    {
      "epoch": 3.5852072105556587,
      "grad_norm": 0.7945622205734253,
      "learning_rate": 4.865496924956536e-06,
      "loss": 0.5361,
      "step": 19389
    },
    {
      "epoch": 3.58539304961903,
      "grad_norm": 1.0395240783691406,
      "learning_rate": 4.860996398738471e-06,
      "loss": 0.9424,
      "step": 19390
    },
    {
      "epoch": 3.585578888682401,
      "grad_norm": 1.0619295835494995,
      "learning_rate": 4.856497903099166e-06,
      "loss": 0.7907,
      "step": 19391
    },
    {
      "epoch": 3.5857647277457723,
      "grad_norm": 0.8295533657073975,
      "learning_rate": 4.852001438134646e-06,
      "loss": 0.7913,
      "step": 19392
    },
    {
      "epoch": 3.585950566809143,
      "grad_norm": 1.1208479404449463,
      "learning_rate": 4.847507003940854e-06,
      "loss": 0.8687,
      "step": 19393
    },
    {
      "epoch": 3.5861364058725145,
      "grad_norm": 0.8541909456253052,
      "learning_rate": 4.84301460061376e-06,
      "loss": 0.7109,
      "step": 19394
    },
    {
      "epoch": 3.5863222449358854,
      "grad_norm": 1.2070188522338867,
      "learning_rate": 4.83852422824922e-06,
      "loss": 0.8644,
      "step": 19395
    },
    {
      "epoch": 3.5865080839992567,
      "grad_norm": 0.9395968317985535,
      "learning_rate": 4.834035886943067e-06,
      "loss": 0.7369,
      "step": 19396
    },
    {
      "epoch": 3.5866939230626276,
      "grad_norm": 0.9339488744735718,
      "learning_rate": 4.829549576791093e-06,
      "loss": 0.6236,
      "step": 19397
    },
    {
      "epoch": 3.586879762125999,
      "grad_norm": 1.0937880277633667,
      "learning_rate": 4.825065297889075e-06,
      "loss": 0.8,
      "step": 19398
    },
    {
      "epoch": 3.5870656011893702,
      "grad_norm": 1.1339138746261597,
      "learning_rate": 4.8205830503327274e-06,
      "loss": 0.7031,
      "step": 19399
    },
    {
      "epoch": 3.587251440252741,
      "grad_norm": 0.8786293268203735,
      "learning_rate": 4.816102834217695e-06,
      "loss": 0.7978,
      "step": 19400
    },
    {
      "epoch": 3.587437279316112,
      "grad_norm": 0.8842145204544067,
      "learning_rate": 4.81162464963959e-06,
      "loss": 0.7576,
      "step": 19401
    },
    {
      "epoch": 3.5876231183794833,
      "grad_norm": 1.0113906860351562,
      "learning_rate": 4.807148496694025e-06,
      "loss": 0.8677,
      "step": 19402
    },
    {
      "epoch": 3.5878089574428547,
      "grad_norm": 0.8485032320022583,
      "learning_rate": 4.8026743754765125e-06,
      "loss": 0.9075,
      "step": 19403
    },
    {
      "epoch": 3.5879947965062255,
      "grad_norm": 1.0245674848556519,
      "learning_rate": 4.798202286082554e-06,
      "loss": 0.8218,
      "step": 19404
    },
    {
      "epoch": 3.588180635569597,
      "grad_norm": 0.878632664680481,
      "learning_rate": 4.7937322286075725e-06,
      "loss": 0.8435,
      "step": 19405
    },
    {
      "epoch": 3.5883664746329678,
      "grad_norm": 0.9851412773132324,
      "learning_rate": 4.789264203147026e-06,
      "loss": 0.8201,
      "step": 19406
    },
    {
      "epoch": 3.588552313696339,
      "grad_norm": 0.9448543190956116,
      "learning_rate": 4.784798209796215e-06,
      "loss": 0.6644,
      "step": 19407
    },
    {
      "epoch": 3.58873815275971,
      "grad_norm": 1.1279706954956055,
      "learning_rate": 4.78033424865052e-06,
      "loss": 0.9009,
      "step": 19408
    },
    {
      "epoch": 3.5889239918230813,
      "grad_norm": 0.9069414138793945,
      "learning_rate": 4.775872319805164e-06,
      "loss": 0.7876,
      "step": 19409
    },
    {
      "epoch": 3.589109830886452,
      "grad_norm": 0.9666802287101746,
      "learning_rate": 4.771412423355415e-06,
      "loss": 0.8229,
      "step": 19410
    },
    {
      "epoch": 3.5892956699498235,
      "grad_norm": 1.0100266933441162,
      "learning_rate": 4.766954559396452e-06,
      "loss": 0.9738,
      "step": 19411
    },
    {
      "epoch": 3.5894815090131944,
      "grad_norm": 1.088109016418457,
      "learning_rate": 4.762498728023423e-06,
      "loss": 0.752,
      "step": 19412
    },
    {
      "epoch": 3.5896673480765657,
      "grad_norm": 1.0191460847854614,
      "learning_rate": 4.7580449293314066e-06,
      "loss": 0.8325,
      "step": 19413
    },
    {
      "epoch": 3.589853187139937,
      "grad_norm": 0.9894083142280579,
      "learning_rate": 4.753593163415493e-06,
      "loss": 0.9898,
      "step": 19414
    },
    {
      "epoch": 3.590039026203308,
      "grad_norm": 0.876889169216156,
      "learning_rate": 4.749143430370684e-06,
      "loss": 0.671,
      "step": 19415
    },
    {
      "epoch": 3.590224865266679,
      "grad_norm": 0.7905625700950623,
      "learning_rate": 4.744695730291948e-06,
      "loss": 0.4096,
      "step": 19416
    },
    {
      "epoch": 3.59041070433005,
      "grad_norm": 0.8803296089172363,
      "learning_rate": 4.74025006327421e-06,
      "loss": 0.5567,
      "step": 19417
    },
    {
      "epoch": 3.5905965433934215,
      "grad_norm": 0.9684118032455444,
      "learning_rate": 4.735806429412359e-06,
      "loss": 0.8737,
      "step": 19418
    },
    {
      "epoch": 3.5907823824567924,
      "grad_norm": 0.9677191376686096,
      "learning_rate": 4.731364828801243e-06,
      "loss": 0.7487,
      "step": 19419
    },
    {
      "epoch": 3.5909682215201633,
      "grad_norm": 0.936278760433197,
      "learning_rate": 4.726925261535675e-06,
      "loss": 0.7198,
      "step": 19420
    },
    {
      "epoch": 3.5911540605835346,
      "grad_norm": 1.0864543914794922,
      "learning_rate": 4.722487727710368e-06,
      "loss": 1.0318,
      "step": 19421
    },
    {
      "epoch": 3.591339899646906,
      "grad_norm": 1.2860759496688843,
      "learning_rate": 4.718052227420078e-06,
      "loss": 0.631,
      "step": 19422
    },
    {
      "epoch": 3.591525738710277,
      "grad_norm": 0.8440675735473633,
      "learning_rate": 4.713618760759442e-06,
      "loss": 0.6534,
      "step": 19423
    },
    {
      "epoch": 3.591711577773648,
      "grad_norm": 0.9321073293685913,
      "learning_rate": 4.709187327823106e-06,
      "loss": 0.714,
      "step": 19424
    },
    {
      "epoch": 3.591897416837019,
      "grad_norm": 0.8669753074645996,
      "learning_rate": 4.704757928705617e-06,
      "loss": 0.834,
      "step": 19425
    },
    {
      "epoch": 3.5920832559003903,
      "grad_norm": 1.121113657951355,
      "learning_rate": 4.700330563501565e-06,
      "loss": 0.8264,
      "step": 19426
    },
    {
      "epoch": 3.5922690949637612,
      "grad_norm": 0.8920173645019531,
      "learning_rate": 4.6959052323054085e-06,
      "loss": 1.0234,
      "step": 19427
    },
    {
      "epoch": 3.5924549340271326,
      "grad_norm": 0.9285154938697815,
      "learning_rate": 4.691481935211584e-06,
      "loss": 0.8498,
      "step": 19428
    },
    {
      "epoch": 3.592640773090504,
      "grad_norm": 1.2579076290130615,
      "learning_rate": 4.687060672314536e-06,
      "loss": 1.0118,
      "step": 19429
    },
    {
      "epoch": 3.5928266121538748,
      "grad_norm": 0.8487569689750671,
      "learning_rate": 4.682641443708624e-06,
      "loss": 0.7867,
      "step": 19430
    },
    {
      "epoch": 3.5930124512172457,
      "grad_norm": 0.9112158417701721,
      "learning_rate": 4.67822424948815e-06,
      "loss": 0.6424,
      "step": 19431
    },
    {
      "epoch": 3.593198290280617,
      "grad_norm": 0.9790871143341064,
      "learning_rate": 4.673809089747416e-06,
      "loss": 0.7442,
      "step": 19432
    },
    {
      "epoch": 3.5933841293439883,
      "grad_norm": 1.0553319454193115,
      "learning_rate": 4.669395964580614e-06,
      "loss": 0.8304,
      "step": 19433
    },
    {
      "epoch": 3.593569968407359,
      "grad_norm": 0.9264032244682312,
      "learning_rate": 4.6649848740819794e-06,
      "loss": 0.7993,
      "step": 19434
    },
    {
      "epoch": 3.59375580747073,
      "grad_norm": 0.8428817391395569,
      "learning_rate": 4.660575818345647e-06,
      "loss": 0.7858,
      "step": 19435
    },
    {
      "epoch": 3.5939416465341014,
      "grad_norm": 0.8501147627830505,
      "learning_rate": 4.6561687974657095e-06,
      "loss": 0.8255,
      "step": 19436
    },
    {
      "epoch": 3.5941274855974727,
      "grad_norm": 0.9591187834739685,
      "learning_rate": 4.6517638115362136e-06,
      "loss": 0.7297,
      "step": 19437
    },
    {
      "epoch": 3.5943133246608436,
      "grad_norm": 0.8795649409294128,
      "learning_rate": 4.647360860651217e-06,
      "loss": 0.6012,
      "step": 19438
    },
    {
      "epoch": 3.594499163724215,
      "grad_norm": 1.1258068084716797,
      "learning_rate": 4.642959944904656e-06,
      "loss": 0.8224,
      "step": 19439
    },
    {
      "epoch": 3.594685002787586,
      "grad_norm": 0.9716440439224243,
      "learning_rate": 4.63856106439049e-06,
      "loss": 0.9072,
      "step": 19440
    },
    {
      "epoch": 3.594870841850957,
      "grad_norm": 0.8310915231704712,
      "learning_rate": 4.634164219202575e-06,
      "loss": 0.7235,
      "step": 19441
    },
    {
      "epoch": 3.595056680914328,
      "grad_norm": 0.8197675943374634,
      "learning_rate": 4.629769409434781e-06,
      "loss": 0.7642,
      "step": 19442
    },
    {
      "epoch": 3.5952425199776994,
      "grad_norm": 1.115140438079834,
      "learning_rate": 4.625376635180901e-06,
      "loss": 0.9677,
      "step": 19443
    },
    {
      "epoch": 3.5954283590410703,
      "grad_norm": 0.995252251625061,
      "learning_rate": 4.620985896534679e-06,
      "loss": 0.7594,
      "step": 19444
    },
    {
      "epoch": 3.5956141981044416,
      "grad_norm": 1.0043798685073853,
      "learning_rate": 4.616597193589833e-06,
      "loss": 0.7948,
      "step": 19445
    },
    {
      "epoch": 3.5958000371678125,
      "grad_norm": 0.8082947731018066,
      "learning_rate": 4.612210526440053e-06,
      "loss": 0.6672,
      "step": 19446
    },
    {
      "epoch": 3.595985876231184,
      "grad_norm": 0.9716810584068298,
      "learning_rate": 4.607825895178941e-06,
      "loss": 1.0794,
      "step": 19447
    },
    {
      "epoch": 3.596171715294555,
      "grad_norm": 0.8218140602111816,
      "learning_rate": 4.603443299900068e-06,
      "loss": 0.7711,
      "step": 19448
    },
    {
      "epoch": 3.596357554357926,
      "grad_norm": 1.154445767402649,
      "learning_rate": 4.599062740696991e-06,
      "loss": 0.9637,
      "step": 19449
    },
    {
      "epoch": 3.596543393421297,
      "grad_norm": 0.8171733021736145,
      "learning_rate": 4.594684217663225e-06,
      "loss": 0.6243,
      "step": 19450
    },
    {
      "epoch": 3.5967292324846682,
      "grad_norm": 1.1623929738998413,
      "learning_rate": 4.590307730892196e-06,
      "loss": 0.4765,
      "step": 19451
    },
    {
      "epoch": 3.5969150715480396,
      "grad_norm": 0.8249045014381409,
      "learning_rate": 4.585933280477328e-06,
      "loss": 0.8474,
      "step": 19452
    },
    {
      "epoch": 3.5971009106114105,
      "grad_norm": 0.8545265197753906,
      "learning_rate": 4.581560866511958e-06,
      "loss": 0.6475,
      "step": 19453
    },
    {
      "epoch": 3.597286749674782,
      "grad_norm": 1.0757006406784058,
      "learning_rate": 4.577190489089445e-06,
      "loss": 0.8369,
      "step": 19454
    },
    {
      "epoch": 3.5974725887381527,
      "grad_norm": 0.8231223821640015,
      "learning_rate": 4.572822148303035e-06,
      "loss": 0.6575,
      "step": 19455
    },
    {
      "epoch": 3.597658427801524,
      "grad_norm": 0.9437882900238037,
      "learning_rate": 4.568455844245989e-06,
      "loss": 0.9076,
      "step": 19456
    },
    {
      "epoch": 3.597844266864895,
      "grad_norm": 1.02458655834198,
      "learning_rate": 4.564091577011465e-06,
      "loss": 0.7054,
      "step": 19457
    },
    {
      "epoch": 3.598030105928266,
      "grad_norm": 0.8490471243858337,
      "learning_rate": 4.559729346692654e-06,
      "loss": 0.8083,
      "step": 19458
    },
    {
      "epoch": 3.598215944991637,
      "grad_norm": 1.1624078750610352,
      "learning_rate": 4.5553691533826385e-06,
      "loss": 0.8191,
      "step": 19459
    },
    {
      "epoch": 3.5984017840550084,
      "grad_norm": 0.7939742803573608,
      "learning_rate": 4.551010997174454e-06,
      "loss": 0.6867,
      "step": 19460
    },
    {
      "epoch": 3.5985876231183793,
      "grad_norm": 0.9577316045761108,
      "learning_rate": 4.54665487816115e-06,
      "loss": 0.9839,
      "step": 19461
    },
    {
      "epoch": 3.5987734621817506,
      "grad_norm": 1.0471750497817993,
      "learning_rate": 4.542300796435706e-06,
      "loss": 0.8577,
      "step": 19462
    },
    {
      "epoch": 3.598959301245122,
      "grad_norm": 1.0368680953979492,
      "learning_rate": 4.537948752091037e-06,
      "loss": 0.901,
      "step": 19463
    },
    {
      "epoch": 3.599145140308493,
      "grad_norm": 1.006668210029602,
      "learning_rate": 4.5335987452200355e-06,
      "loss": 0.9178,
      "step": 19464
    },
    {
      "epoch": 3.5993309793718637,
      "grad_norm": 1.0635803937911987,
      "learning_rate": 4.529250775915528e-06,
      "loss": 0.6381,
      "step": 19465
    },
    {
      "epoch": 3.599516818435235,
      "grad_norm": 0.9568383693695068,
      "learning_rate": 4.524904844270339e-06,
      "loss": 0.95,
      "step": 19466
    },
    {
      "epoch": 3.5997026574986064,
      "grad_norm": 1.078012228012085,
      "learning_rate": 4.520560950377217e-06,
      "loss": 0.8683,
      "step": 19467
    },
    {
      "epoch": 3.5998884965619773,
      "grad_norm": 0.9089167714118958,
      "learning_rate": 4.5162190943288665e-06,
      "loss": 0.8525,
      "step": 19468
    },
    {
      "epoch": 3.600074335625348,
      "grad_norm": 1.0891344547271729,
      "learning_rate": 4.5118792762179676e-06,
      "loss": 0.9373,
      "step": 19469
    },
    {
      "epoch": 3.6002601746887195,
      "grad_norm": 0.9258513450622559,
      "learning_rate": 4.507541496137125e-06,
      "loss": 0.5951,
      "step": 19470
    },
    {
      "epoch": 3.600446013752091,
      "grad_norm": 0.9756423830986023,
      "learning_rate": 4.503205754178952e-06,
      "loss": 0.9625,
      "step": 19471
    },
    {
      "epoch": 3.6006318528154617,
      "grad_norm": 1.035539150238037,
      "learning_rate": 4.498872050435976e-06,
      "loss": 0.7596,
      "step": 19472
    },
    {
      "epoch": 3.600817691878833,
      "grad_norm": 1.0090545415878296,
      "learning_rate": 4.494540385000679e-06,
      "loss": 0.8671,
      "step": 19473
    },
    {
      "epoch": 3.601003530942204,
      "grad_norm": 0.8952997326850891,
      "learning_rate": 4.490210757965529e-06,
      "loss": 0.7329,
      "step": 19474
    },
    {
      "epoch": 3.6011893700055753,
      "grad_norm": 0.9174466133117676,
      "learning_rate": 4.485883169422933e-06,
      "loss": 0.8245,
      "step": 19475
    },
    {
      "epoch": 3.601375209068946,
      "grad_norm": 1.4027634859085083,
      "learning_rate": 4.481557619465249e-06,
      "loss": 0.7972,
      "step": 19476
    },
    {
      "epoch": 3.6015610481323175,
      "grad_norm": 1.035872459411621,
      "learning_rate": 4.477234108184791e-06,
      "loss": 0.9884,
      "step": 19477
    },
    {
      "epoch": 3.601746887195689,
      "grad_norm": 0.9323967695236206,
      "learning_rate": 4.472912635673865e-06,
      "loss": 0.7008,
      "step": 19478
    },
    {
      "epoch": 3.6019327262590597,
      "grad_norm": 0.8735101222991943,
      "learning_rate": 4.468593202024685e-06,
      "loss": 0.7825,
      "step": 19479
    },
    {
      "epoch": 3.6021185653224306,
      "grad_norm": 0.9623664617538452,
      "learning_rate": 4.464275807329432e-06,
      "loss": 0.8998,
      "step": 19480
    },
    {
      "epoch": 3.602304404385802,
      "grad_norm": 1.0125689506530762,
      "learning_rate": 4.459960451680268e-06,
      "loss": 1.0694,
      "step": 19481
    },
    {
      "epoch": 3.6024902434491732,
      "grad_norm": 0.981284499168396,
      "learning_rate": 4.4556471351692945e-06,
      "loss": 1.1182,
      "step": 19482
    },
    {
      "epoch": 3.602676082512544,
      "grad_norm": 0.9122440218925476,
      "learning_rate": 4.451335857888583e-06,
      "loss": 0.9736,
      "step": 19483
    },
    {
      "epoch": 3.602861921575915,
      "grad_norm": 1.042223334312439,
      "learning_rate": 4.447026619930139e-06,
      "loss": 0.7919,
      "step": 19484
    },
    {
      "epoch": 3.6030477606392863,
      "grad_norm": 0.9088339805603027,
      "learning_rate": 4.442719421385922e-06,
      "loss": 0.7814,
      "step": 19485
    },
    {
      "epoch": 3.6032335997026577,
      "grad_norm": 1.1977900266647339,
      "learning_rate": 4.43841426234789e-06,
      "loss": 0.7343,
      "step": 19486
    },
    {
      "epoch": 3.6034194387660285,
      "grad_norm": 1.3704267740249634,
      "learning_rate": 4.4341111429079156e-06,
      "loss": 1.1858,
      "step": 19487
    },
    {
      "epoch": 3.6036052778294,
      "grad_norm": 0.9473437666893005,
      "learning_rate": 4.429810063157835e-06,
      "loss": 0.7812,
      "step": 19488
    },
    {
      "epoch": 3.6037911168927708,
      "grad_norm": 1.0082486867904663,
      "learning_rate": 4.425511023189455e-06,
      "loss": 0.7871,
      "step": 19489
    },
    {
      "epoch": 3.603976955956142,
      "grad_norm": 1.0087922811508179,
      "learning_rate": 4.421214023094533e-06,
      "loss": 0.8876,
      "step": 19490
    },
    {
      "epoch": 3.604162795019513,
      "grad_norm": 0.7733488082885742,
      "learning_rate": 4.416919062964764e-06,
      "loss": 0.6503,
      "step": 19491
    },
    {
      "epoch": 3.6043486340828843,
      "grad_norm": 0.8714876770973206,
      "learning_rate": 4.412626142891851e-06,
      "loss": 0.5623,
      "step": 19492
    },
    {
      "epoch": 3.604534473146255,
      "grad_norm": 0.8953797221183777,
      "learning_rate": 4.408335262967378e-06,
      "loss": 0.7119,
      "step": 19493
    },
    {
      "epoch": 3.6047203122096265,
      "grad_norm": 0.7247197031974792,
      "learning_rate": 4.404046423282959e-06,
      "loss": 0.5571,
      "step": 19494
    },
    {
      "epoch": 3.6049061512729974,
      "grad_norm": 1.0153875350952148,
      "learning_rate": 4.399759623930122e-06,
      "loss": 0.8047,
      "step": 19495
    },
    {
      "epoch": 3.6050919903363687,
      "grad_norm": 0.9947652220726013,
      "learning_rate": 4.3954748650003704e-06,
      "loss": 0.9466,
      "step": 19496
    },
    {
      "epoch": 3.60527782939974,
      "grad_norm": 0.8497185707092285,
      "learning_rate": 4.3911921465851105e-06,
      "loss": 0.8645,
      "step": 19497
    },
    {
      "epoch": 3.605463668463111,
      "grad_norm": 0.9185053110122681,
      "learning_rate": 4.386911468775812e-06,
      "loss": 0.6431,
      "step": 19498
    },
    {
      "epoch": 3.605649507526482,
      "grad_norm": 0.8739264011383057,
      "learning_rate": 4.382632831663802e-06,
      "loss": 0.7773,
      "step": 19499
    },
    {
      "epoch": 3.605835346589853,
      "grad_norm": 0.8473137617111206,
      "learning_rate": 4.378356235340397e-06,
      "loss": 0.7008,
      "step": 19500
    },
    {
      "epoch": 3.6060211856532245,
      "grad_norm": 1.0435789823532104,
      "learning_rate": 4.374081679896902e-06,
      "loss": 0.8217,
      "step": 19501
    },
    {
      "epoch": 3.6062070247165954,
      "grad_norm": 0.8870427012443542,
      "learning_rate": 4.36980916542451e-06,
      "loss": 0.8054,
      "step": 19502
    },
    {
      "epoch": 3.6063928637799667,
      "grad_norm": 0.993805468082428,
      "learning_rate": 4.365538692014459e-06,
      "loss": 0.7849,
      "step": 19503
    },
    {
      "epoch": 3.6065787028433376,
      "grad_norm": 1.0244243144989014,
      "learning_rate": 4.361270259757855e-06,
      "loss": 0.8614,
      "step": 19504
    },
    {
      "epoch": 3.606764541906709,
      "grad_norm": 0.9650846719741821,
      "learning_rate": 4.3570038687458125e-06,
      "loss": 0.769,
      "step": 19505
    },
    {
      "epoch": 3.60695038097008,
      "grad_norm": 0.8885460495948792,
      "learning_rate": 4.352739519069404e-06,
      "loss": 0.7117,
      "step": 19506
    },
    {
      "epoch": 3.607136220033451,
      "grad_norm": 1.1355706453323364,
      "learning_rate": 4.348477210819624e-06,
      "loss": 1.001,
      "step": 19507
    },
    {
      "epoch": 3.607322059096822,
      "grad_norm": 0.9262754917144775,
      "learning_rate": 4.344216944087454e-06,
      "loss": 1.0028,
      "step": 19508
    },
    {
      "epoch": 3.6075078981601933,
      "grad_norm": 0.9142513275146484,
      "learning_rate": 4.339958718963821e-06,
      "loss": 0.7278,
      "step": 19509
    },
    {
      "epoch": 3.6076937372235642,
      "grad_norm": 1.113861322402954,
      "learning_rate": 4.335702535539621e-06,
      "loss": 0.9307,
      "step": 19510
    },
    {
      "epoch": 3.6078795762869356,
      "grad_norm": 0.9880869388580322,
      "learning_rate": 4.331448393905679e-06,
      "loss": 0.6321,
      "step": 19511
    },
    {
      "epoch": 3.608065415350307,
      "grad_norm": 1.4004838466644287,
      "learning_rate": 4.32719629415278e-06,
      "loss": 0.9643,
      "step": 19512
    },
    {
      "epoch": 3.6082512544136778,
      "grad_norm": 0.9977883696556091,
      "learning_rate": 4.322946236371705e-06,
      "loss": 0.7023,
      "step": 19513
    },
    {
      "epoch": 3.6084370934770487,
      "grad_norm": 0.9743738174438477,
      "learning_rate": 4.318698220653161e-06,
      "loss": 0.7202,
      "step": 19514
    },
    {
      "epoch": 3.60862293254042,
      "grad_norm": 1.4286434650421143,
      "learning_rate": 4.314452247087808e-06,
      "loss": 0.6513,
      "step": 19515
    },
    {
      "epoch": 3.6088087716037913,
      "grad_norm": 0.9633848667144775,
      "learning_rate": 4.310208315766273e-06,
      "loss": 0.8454,
      "step": 19516
    },
    {
      "epoch": 3.608994610667162,
      "grad_norm": 1.1255860328674316,
      "learning_rate": 4.305966426779118e-06,
      "loss": 0.9287,
      "step": 19517
    },
    {
      "epoch": 3.609180449730533,
      "grad_norm": 1.0300997495651245,
      "learning_rate": 4.301726580216902e-06,
      "loss": 0.8876,
      "step": 19518
    },
    {
      "epoch": 3.6093662887939044,
      "grad_norm": 0.9136091470718384,
      "learning_rate": 4.2974887761701e-06,
      "loss": 0.6974,
      "step": 19519
    },
    {
      "epoch": 3.6095521278572757,
      "grad_norm": 0.8567022681236267,
      "learning_rate": 4.293253014729171e-06,
      "loss": 0.9059,
      "step": 19520
    },
    {
      "epoch": 3.6097379669206466,
      "grad_norm": 1.0704967975616455,
      "learning_rate": 4.28901929598452e-06,
      "loss": 0.5962,
      "step": 19521
    },
    {
      "epoch": 3.609923805984018,
      "grad_norm": 0.9951958060264587,
      "learning_rate": 4.284787620026498e-06,
      "loss": 0.5911,
      "step": 19522
    },
    {
      "epoch": 3.610109645047389,
      "grad_norm": 0.8182536363601685,
      "learning_rate": 4.280557986945433e-06,
      "loss": 0.7981,
      "step": 19523
    },
    {
      "epoch": 3.61029548411076,
      "grad_norm": 0.8957955837249756,
      "learning_rate": 4.276330396831596e-06,
      "loss": 0.8166,
      "step": 19524
    },
    {
      "epoch": 3.610481323174131,
      "grad_norm": 0.9633662700653076,
      "learning_rate": 4.272104849775216e-06,
      "loss": 0.8607,
      "step": 19525
    },
    {
      "epoch": 3.6106671622375024,
      "grad_norm": 1.0764795541763306,
      "learning_rate": 4.2678813458664866e-06,
      "loss": 0.7152,
      "step": 19526
    },
    {
      "epoch": 3.6108530013008733,
      "grad_norm": 0.7728953957557678,
      "learning_rate": 4.263659885195559e-06,
      "loss": 0.5237,
      "step": 19527
    },
    {
      "epoch": 3.6110388403642446,
      "grad_norm": 0.8335964679718018,
      "learning_rate": 4.2594404678524935e-06,
      "loss": 0.6697,
      "step": 19528
    },
    {
      "epoch": 3.6112246794276155,
      "grad_norm": 0.9022690653800964,
      "learning_rate": 4.255223093927396e-06,
      "loss": 0.4068,
      "step": 19529
    },
    {
      "epoch": 3.611410518490987,
      "grad_norm": 1.0087932348251343,
      "learning_rate": 4.251007763510262e-06,
      "loss": 0.7562,
      "step": 19530
    },
    {
      "epoch": 3.611596357554358,
      "grad_norm": 0.8287792205810547,
      "learning_rate": 4.24679447669104e-06,
      "loss": 0.7704,
      "step": 19531
    },
    {
      "epoch": 3.611782196617729,
      "grad_norm": 1.1251407861709595,
      "learning_rate": 4.242583233559671e-06,
      "loss": 0.6351,
      "step": 19532
    },
    {
      "epoch": 3.6119680356811,
      "grad_norm": 1.1125820875167847,
      "learning_rate": 4.238374034206028e-06,
      "loss": 0.8003,
      "step": 19533
    },
    {
      "epoch": 3.6121538747444712,
      "grad_norm": 0.8124536275863647,
      "learning_rate": 4.234166878719981e-06,
      "loss": 0.7484,
      "step": 19534
    },
    {
      "epoch": 3.6123397138078426,
      "grad_norm": 0.8514935970306396,
      "learning_rate": 4.229961767191293e-06,
      "loss": 0.8515,
      "step": 19535
    },
    {
      "epoch": 3.6125255528712135,
      "grad_norm": 1.0911568403244019,
      "learning_rate": 4.225758699709725e-06,
      "loss": 1.0046,
      "step": 19536
    },
    {
      "epoch": 3.612711391934585,
      "grad_norm": 0.823409378528595,
      "learning_rate": 4.2215576763649735e-06,
      "loss": 0.459,
      "step": 19537
    },
    {
      "epoch": 3.6128972309979557,
      "grad_norm": 0.8221675157546997,
      "learning_rate": 4.217358697246709e-06,
      "loss": 0.6495,
      "step": 19538
    },
    {
      "epoch": 3.613083070061327,
      "grad_norm": 0.8527970910072327,
      "learning_rate": 4.213161762444573e-06,
      "loss": 0.8859,
      "step": 19539
    },
    {
      "epoch": 3.613268909124698,
      "grad_norm": 1.006277322769165,
      "learning_rate": 4.208966872048092e-06,
      "loss": 0.6568,
      "step": 19540
    },
    {
      "epoch": 3.613454748188069,
      "grad_norm": 1.1168917417526245,
      "learning_rate": 4.204774026146851e-06,
      "loss": 1.0532,
      "step": 19541
    },
    {
      "epoch": 3.61364058725144,
      "grad_norm": 0.8510649800300598,
      "learning_rate": 4.200583224830324e-06,
      "loss": 0.8008,
      "step": 19542
    },
    {
      "epoch": 3.6138264263148114,
      "grad_norm": 0.9765340089797974,
      "learning_rate": 4.196394468187936e-06,
      "loss": 1.0101,
      "step": 19543
    },
    {
      "epoch": 3.6140122653781823,
      "grad_norm": 0.9470708966255188,
      "learning_rate": 4.192207756309097e-06,
      "loss": 0.7003,
      "step": 19544
    },
    {
      "epoch": 3.6141981044415536,
      "grad_norm": 1.0259088277816772,
      "learning_rate": 4.188023089283177e-06,
      "loss": 0.9788,
      "step": 19545
    },
    {
      "epoch": 3.614383943504925,
      "grad_norm": 0.8592630624771118,
      "learning_rate": 4.183840467199485e-06,
      "loss": 0.762,
      "step": 19546
    },
    {
      "epoch": 3.614569782568296,
      "grad_norm": 1.0951875448226929,
      "learning_rate": 4.179659890147302e-06,
      "loss": 1.1296,
      "step": 19547
    },
    {
      "epoch": 3.6147556216316667,
      "grad_norm": 0.8168365359306335,
      "learning_rate": 4.175481358215827e-06,
      "loss": 0.7834,
      "step": 19548
    },
    {
      "epoch": 3.614941460695038,
      "grad_norm": 1.0954785346984863,
      "learning_rate": 4.171304871494264e-06,
      "loss": 0.6931,
      "step": 19549
    },
    {
      "epoch": 3.6151272997584094,
      "grad_norm": 1.0510135889053345,
      "learning_rate": 4.167130430071764e-06,
      "loss": 1.1022,
      "step": 19550
    },
    {
      "epoch": 3.6153131388217803,
      "grad_norm": 1.0180712938308716,
      "learning_rate": 4.1629580340374005e-06,
      "loss": 1.0358,
      "step": 19551
    },
    {
      "epoch": 3.6154989778851516,
      "grad_norm": 1.0190739631652832,
      "learning_rate": 4.1587876834802255e-06,
      "loss": 0.7687,
      "step": 19552
    },
    {
      "epoch": 3.6156848169485225,
      "grad_norm": 1.0596504211425781,
      "learning_rate": 4.154619378489266e-06,
      "loss": 0.9051,
      "step": 19553
    },
    {
      "epoch": 3.615870656011894,
      "grad_norm": 0.8554338812828064,
      "learning_rate": 4.150453119153475e-06,
      "loss": 0.8167,
      "step": 19554
    },
    {
      "epoch": 3.6160564950752647,
      "grad_norm": 0.8878720998764038,
      "learning_rate": 4.14628890556178e-06,
      "loss": 0.9358,
      "step": 19555
    },
    {
      "epoch": 3.616242334138636,
      "grad_norm": 1.0696130990982056,
      "learning_rate": 4.142126737803065e-06,
      "loss": 0.862,
      "step": 19556
    },
    {
      "epoch": 3.616428173202007,
      "grad_norm": 1.079859733581543,
      "learning_rate": 4.137966615966138e-06,
      "loss": 0.9542,
      "step": 19557
    },
    {
      "epoch": 3.6166140122653783,
      "grad_norm": 0.9978886246681213,
      "learning_rate": 4.1338085401398275e-06,
      "loss": 1.1051,
      "step": 19558
    },
    {
      "epoch": 3.616799851328749,
      "grad_norm": 0.9419093132019043,
      "learning_rate": 4.129652510412851e-06,
      "loss": 0.87,
      "step": 19559
    },
    {
      "epoch": 3.6169856903921205,
      "grad_norm": 0.9488292336463928,
      "learning_rate": 4.125498526873917e-06,
      "loss": 0.9075,
      "step": 19560
    },
    {
      "epoch": 3.617171529455492,
      "grad_norm": 0.9746842980384827,
      "learning_rate": 4.121346589611697e-06,
      "loss": 0.7477,
      "step": 19561
    },
    {
      "epoch": 3.6173573685188627,
      "grad_norm": 0.992132842540741,
      "learning_rate": 4.117196698714798e-06,
      "loss": 0.7586,
      "step": 19562
    },
    {
      "epoch": 3.6175432075822336,
      "grad_norm": 0.9009767174720764,
      "learning_rate": 4.113048854271795e-06,
      "loss": 1.0266,
      "step": 19563
    },
    {
      "epoch": 3.617729046645605,
      "grad_norm": 0.7765776515007019,
      "learning_rate": 4.108903056371216e-06,
      "loss": 0.6627,
      "step": 19564
    },
    {
      "epoch": 3.6179148857089762,
      "grad_norm": 1.0591367483139038,
      "learning_rate": 4.104759305101525e-06,
      "loss": 0.918,
      "step": 19565
    },
    {
      "epoch": 3.618100724772347,
      "grad_norm": 1.149461030960083,
      "learning_rate": 4.100617600551216e-06,
      "loss": 0.6963,
      "step": 19566
    },
    {
      "epoch": 3.618286563835718,
      "grad_norm": 0.8625674247741699,
      "learning_rate": 4.096477942808641e-06,
      "loss": 0.7483,
      "step": 19567
    },
    {
      "epoch": 3.6184724028990893,
      "grad_norm": 0.8393241763114929,
      "learning_rate": 4.092340331962152e-06,
      "loss": 0.9917,
      "step": 19568
    },
    {
      "epoch": 3.6186582419624607,
      "grad_norm": 0.8998364806175232,
      "learning_rate": 4.088204768100101e-06,
      "loss": 0.6585,
      "step": 19569
    },
    {
      "epoch": 3.6188440810258315,
      "grad_norm": 0.9522582292556763,
      "learning_rate": 4.0840712513107175e-06,
      "loss": 0.7368,
      "step": 19570
    },
    {
      "epoch": 3.619029920089203,
      "grad_norm": 0.8371335864067078,
      "learning_rate": 4.07993978168224e-06,
      "loss": 0.6592,
      "step": 19571
    },
    {
      "epoch": 3.6192157591525738,
      "grad_norm": 0.9628044962882996,
      "learning_rate": 4.075810359302823e-06,
      "loss": 1.0382,
      "step": 19572
    },
    {
      "epoch": 3.619401598215945,
      "grad_norm": 1.0300853252410889,
      "learning_rate": 4.071682984260638e-06,
      "loss": 0.798,
      "step": 19573
    },
    {
      "epoch": 3.619587437279316,
      "grad_norm": 0.90743088722229,
      "learning_rate": 4.0675576566437615e-06,
      "loss": 0.7294,
      "step": 19574
    },
    {
      "epoch": 3.6197732763426873,
      "grad_norm": 1.2764816284179688,
      "learning_rate": 4.063434376540232e-06,
      "loss": 0.7791,
      "step": 19575
    },
    {
      "epoch": 3.619959115406058,
      "grad_norm": 1.1036268472671509,
      "learning_rate": 4.059313144038069e-06,
      "loss": 0.7762,
      "step": 19576
    },
    {
      "epoch": 3.6201449544694295,
      "grad_norm": 0.9103646874427795,
      "learning_rate": 4.055193959225212e-06,
      "loss": 0.8537,
      "step": 19577
    },
    {
      "epoch": 3.6203307935328004,
      "grad_norm": 0.8981128334999084,
      "learning_rate": 4.051076822189603e-06,
      "loss": 0.9564,
      "step": 19578
    },
    {
      "epoch": 3.6205166325961717,
      "grad_norm": 0.9731377363204956,
      "learning_rate": 4.046961733019117e-06,
      "loss": 0.6327,
      "step": 19579
    },
    {
      "epoch": 3.620702471659543,
      "grad_norm": 0.83604496717453,
      "learning_rate": 4.042848691801548e-06,
      "loss": 0.6543,
      "step": 19580
    },
    {
      "epoch": 3.620888310722914,
      "grad_norm": 0.8671492338180542,
      "learning_rate": 4.0387376986247174e-06,
      "loss": 0.8174,
      "step": 19581
    },
    {
      "epoch": 3.621074149786285,
      "grad_norm": 0.9103574752807617,
      "learning_rate": 4.0346287535763524e-06,
      "loss": 0.6509,
      "step": 19582
    },
    {
      "epoch": 3.621259988849656,
      "grad_norm": 0.9837188124656677,
      "learning_rate": 4.030521856744163e-06,
      "loss": 0.7643,
      "step": 19583
    },
    {
      "epoch": 3.6214458279130275,
      "grad_norm": 0.8985093235969543,
      "learning_rate": 4.026417008215766e-06,
      "loss": 0.8411,
      "step": 19584
    },
    {
      "epoch": 3.6216316669763984,
      "grad_norm": 0.9094043374061584,
      "learning_rate": 4.022314208078826e-06,
      "loss": 0.7546,
      "step": 19585
    },
    {
      "epoch": 3.6218175060397697,
      "grad_norm": 1.068396806716919,
      "learning_rate": 4.018213456420872e-06,
      "loss": 0.8532,
      "step": 19586
    },
    {
      "epoch": 3.6220033451031406,
      "grad_norm": 0.9189227819442749,
      "learning_rate": 4.014114753329446e-06,
      "loss": 0.6418,
      "step": 19587
    },
    {
      "epoch": 3.622189184166512,
      "grad_norm": 1.0001705884933472,
      "learning_rate": 4.01001809889201e-06,
      "loss": 0.8016,
      "step": 19588
    },
    {
      "epoch": 3.622375023229883,
      "grad_norm": 1.0853298902511597,
      "learning_rate": 4.0059234931960284e-06,
      "loss": 0.9409,
      "step": 19589
    },
    {
      "epoch": 3.622560862293254,
      "grad_norm": 1.1793755292892456,
      "learning_rate": 4.001830936328877e-06,
      "loss": 1.1449,
      "step": 19590
    },
    {
      "epoch": 3.622746701356625,
      "grad_norm": 0.8038926124572754,
      "learning_rate": 3.997740428377905e-06,
      "loss": 0.7745,
      "step": 19591
    },
    {
      "epoch": 3.6229325404199964,
      "grad_norm": 0.9299508929252625,
      "learning_rate": 3.993651969430412e-06,
      "loss": 0.9194,
      "step": 19592
    },
    {
      "epoch": 3.6231183794833672,
      "grad_norm": 1.0494447946548462,
      "learning_rate": 3.989565559573672e-06,
      "loss": 0.9838,
      "step": 19593
    },
    {
      "epoch": 3.6233042185467386,
      "grad_norm": 1.0491498708724976,
      "learning_rate": 3.985481198894891e-06,
      "loss": 0.926,
      "step": 19594
    },
    {
      "epoch": 3.62349005761011,
      "grad_norm": 1.0643278360366821,
      "learning_rate": 3.981398887481247e-06,
      "loss": 0.6892,
      "step": 19595
    },
    {
      "epoch": 3.623675896673481,
      "grad_norm": 0.9665752053260803,
      "learning_rate": 3.977318625419857e-06,
      "loss": 0.994,
      "step": 19596
    },
    {
      "epoch": 3.6238617357368517,
      "grad_norm": 0.9783822298049927,
      "learning_rate": 3.973240412797819e-06,
      "loss": 0.7812,
      "step": 19597
    },
    {
      "epoch": 3.624047574800223,
      "grad_norm": 0.9810153245925903,
      "learning_rate": 3.969164249702184e-06,
      "loss": 0.7105,
      "step": 19598
    },
    {
      "epoch": 3.6242334138635943,
      "grad_norm": 0.8744558095932007,
      "learning_rate": 3.965090136219951e-06,
      "loss": 1.0382,
      "step": 19599
    },
    {
      "epoch": 3.624419252926965,
      "grad_norm": 0.9780319333076477,
      "learning_rate": 3.96101807243805e-06,
      "loss": 0.9367,
      "step": 19600
    },
    {
      "epoch": 3.624605091990336,
      "grad_norm": 0.8120416402816772,
      "learning_rate": 3.9569480584434216e-06,
      "loss": 0.8283,
      "step": 19601
    },
    {
      "epoch": 3.6247909310537074,
      "grad_norm": 0.9405677914619446,
      "learning_rate": 3.952880094322919e-06,
      "loss": 0.7167,
      "step": 19602
    },
    {
      "epoch": 3.6249767701170788,
      "grad_norm": 0.970422625541687,
      "learning_rate": 3.9488141801633736e-06,
      "loss": 0.7446,
      "step": 19603
    },
    {
      "epoch": 3.6251626091804496,
      "grad_norm": 0.9250800609588623,
      "learning_rate": 3.944750316051537e-06,
      "loss": 0.6638,
      "step": 19604
    },
    {
      "epoch": 3.625348448243821,
      "grad_norm": 0.8559107184410095,
      "learning_rate": 3.940688502074186e-06,
      "loss": 0.81,
      "step": 19605
    },
    {
      "epoch": 3.625534287307192,
      "grad_norm": 0.9691064953804016,
      "learning_rate": 3.9366287383179935e-06,
      "loss": 0.9659,
      "step": 19606
    },
    {
      "epoch": 3.625720126370563,
      "grad_norm": 1.0005972385406494,
      "learning_rate": 3.932571024869591e-06,
      "loss": 0.9558,
      "step": 19607
    },
    {
      "epoch": 3.625905965433934,
      "grad_norm": 0.8913658261299133,
      "learning_rate": 3.9285153618156104e-06,
      "loss": 0.8656,
      "step": 19608
    },
    {
      "epoch": 3.6260918044973054,
      "grad_norm": 0.9526522755622864,
      "learning_rate": 3.924461749242614e-06,
      "loss": 0.79,
      "step": 19609
    },
    {
      "epoch": 3.6262776435606767,
      "grad_norm": 0.9458757638931274,
      "learning_rate": 3.920410187237111e-06,
      "loss": 0.8057,
      "step": 19610
    },
    {
      "epoch": 3.6264634826240476,
      "grad_norm": 0.8560062050819397,
      "learning_rate": 3.916360675885566e-06,
      "loss": 0.7014,
      "step": 19611
    },
    {
      "epoch": 3.6266493216874185,
      "grad_norm": 0.9556618332862854,
      "learning_rate": 3.91231321527441e-06,
      "loss": 0.7079,
      "step": 19612
    },
    {
      "epoch": 3.62683516075079,
      "grad_norm": 0.9822280406951904,
      "learning_rate": 3.908267805490051e-06,
      "loss": 0.9576,
      "step": 19613
    },
    {
      "epoch": 3.627020999814161,
      "grad_norm": 1.2236249446868896,
      "learning_rate": 3.904224446618809e-06,
      "loss": 0.9071,
      "step": 19614
    },
    {
      "epoch": 3.627206838877532,
      "grad_norm": 0.9653648138046265,
      "learning_rate": 3.900183138746993e-06,
      "loss": 0.7818,
      "step": 19615
    },
    {
      "epoch": 3.627392677940903,
      "grad_norm": 0.9901843667030334,
      "learning_rate": 3.8961438819608345e-06,
      "loss": 0.8232,
      "step": 19616
    },
    {
      "epoch": 3.6275785170042743,
      "grad_norm": 0.931734561920166,
      "learning_rate": 3.892106676346574e-06,
      "loss": 0.7768,
      "step": 19617
    },
    {
      "epoch": 3.6277643560676456,
      "grad_norm": 1.1918833255767822,
      "learning_rate": 3.888071521990378e-06,
      "loss": 0.7921,
      "step": 19618
    },
    {
      "epoch": 3.6279501951310165,
      "grad_norm": 1.5211327075958252,
      "learning_rate": 3.884038418978364e-06,
      "loss": 1.1896,
      "step": 19619
    },
    {
      "epoch": 3.628136034194388,
      "grad_norm": 1.0748412609100342,
      "learning_rate": 3.880007367396588e-06,
      "loss": 0.7944,
      "step": 19620
    },
    {
      "epoch": 3.6283218732577587,
      "grad_norm": 1.0167418718338013,
      "learning_rate": 3.875978367331124e-06,
      "loss": 0.8971,
      "step": 19621
    },
    {
      "epoch": 3.62850771232113,
      "grad_norm": 0.8878464698791504,
      "learning_rate": 3.871951418867936e-06,
      "loss": 0.8168,
      "step": 19622
    },
    {
      "epoch": 3.628693551384501,
      "grad_norm": 0.9061787128448486,
      "learning_rate": 3.867926522092991e-06,
      "loss": 0.6172,
      "step": 19623
    },
    {
      "epoch": 3.628879390447872,
      "grad_norm": 1.3221555948257446,
      "learning_rate": 3.863903677092173e-06,
      "loss": 0.7848,
      "step": 19624
    },
    {
      "epoch": 3.629065229511243,
      "grad_norm": 0.8736843466758728,
      "learning_rate": 3.859882883951371e-06,
      "loss": 0.7202,
      "step": 19625
    },
    {
      "epoch": 3.6292510685746144,
      "grad_norm": 1.0111111402511597,
      "learning_rate": 3.855864142756382e-06,
      "loss": 0.9331,
      "step": 19626
    },
    {
      "epoch": 3.6294369076379853,
      "grad_norm": 0.9184491038322449,
      "learning_rate": 3.8518474535929826e-06,
      "loss": 0.6702,
      "step": 19627
    },
    {
      "epoch": 3.6296227467013567,
      "grad_norm": 1.0071487426757812,
      "learning_rate": 3.847832816546881e-06,
      "loss": 0.7365,
      "step": 19628
    },
    {
      "epoch": 3.629808585764728,
      "grad_norm": 1.061942458152771,
      "learning_rate": 3.8438202317037986e-06,
      "loss": 0.8551,
      "step": 19629
    },
    {
      "epoch": 3.629994424828099,
      "grad_norm": 0.9342822432518005,
      "learning_rate": 3.8398096991493774e-06,
      "loss": 0.7583,
      "step": 19630
    },
    {
      "epoch": 3.6301802638914697,
      "grad_norm": 0.8872012495994568,
      "learning_rate": 3.835801218969193e-06,
      "loss": 0.7427,
      "step": 19631
    },
    {
      "epoch": 3.630366102954841,
      "grad_norm": 1.1290273666381836,
      "learning_rate": 3.8317947912487886e-06,
      "loss": 0.7622,
      "step": 19632
    },
    {
      "epoch": 3.6305519420182124,
      "grad_norm": 0.9274680614471436,
      "learning_rate": 3.827790416073718e-06,
      "loss": 0.742,
      "step": 19633
    },
    {
      "epoch": 3.6307377810815833,
      "grad_norm": 0.8616178035736084,
      "learning_rate": 3.823788093529412e-06,
      "loss": 0.9817,
      "step": 19634
    },
    {
      "epoch": 3.6309236201449546,
      "grad_norm": 1.890286922454834,
      "learning_rate": 3.819787823701315e-06,
      "loss": 1.2465,
      "step": 19635
    },
    {
      "epoch": 3.6311094592083255,
      "grad_norm": 1.0613394975662231,
      "learning_rate": 3.815789606674769e-06,
      "loss": 0.7922,
      "step": 19636
    },
    {
      "epoch": 3.631295298271697,
      "grad_norm": 0.8788023591041565,
      "learning_rate": 3.8117934425351607e-06,
      "loss": 0.916,
      "step": 19637
    },
    {
      "epoch": 3.6314811373350677,
      "grad_norm": 1.0239942073822021,
      "learning_rate": 3.807799331367734e-06,
      "loss": 0.9387,
      "step": 19638
    },
    {
      "epoch": 3.631666976398439,
      "grad_norm": 1.172844648361206,
      "learning_rate": 3.8038072732577758e-06,
      "loss": 0.8622,
      "step": 19639
    },
    {
      "epoch": 3.63185281546181,
      "grad_norm": 0.8852987885475159,
      "learning_rate": 3.799817268290462e-06,
      "loss": 0.6051,
      "step": 19640
    },
    {
      "epoch": 3.6320386545251813,
      "grad_norm": 0.9584529399871826,
      "learning_rate": 3.7958293165509697e-06,
      "loss": 0.7786,
      "step": 19641
    },
    {
      "epoch": 3.632224493588552,
      "grad_norm": 1.499760627746582,
      "learning_rate": 3.791843418124408e-06,
      "loss": 0.8859,
      "step": 19642
    },
    {
      "epoch": 3.6324103326519235,
      "grad_norm": 0.9277597069740295,
      "learning_rate": 3.787859573095853e-06,
      "loss": 0.8963,
      "step": 19643
    },
    {
      "epoch": 3.632596171715295,
      "grad_norm": 0.8993749022483826,
      "learning_rate": 3.7838777815503047e-06,
      "loss": 0.8914,
      "step": 19644
    },
    {
      "epoch": 3.6327820107786657,
      "grad_norm": 1.0153673887252808,
      "learning_rate": 3.779898043572794e-06,
      "loss": 0.6897,
      "step": 19645
    },
    {
      "epoch": 3.6329678498420366,
      "grad_norm": 0.9434956312179565,
      "learning_rate": 3.7759203592482416e-06,
      "loss": 0.9235,
      "step": 19646
    },
    {
      "epoch": 3.633153688905408,
      "grad_norm": 0.9439644813537598,
      "learning_rate": 3.771944728661536e-06,
      "loss": 0.8132,
      "step": 19647
    },
    {
      "epoch": 3.6333395279687792,
      "grad_norm": 1.0923913717269897,
      "learning_rate": 3.767971151897531e-06,
      "loss": 1.0037,
      "step": 19648
    },
    {
      "epoch": 3.63352536703215,
      "grad_norm": 1.04251229763031,
      "learning_rate": 3.763999629041026e-06,
      "loss": 0.8628,
      "step": 19649
    },
    {
      "epoch": 3.633711206095521,
      "grad_norm": 0.9401928782463074,
      "learning_rate": 3.7600301601768305e-06,
      "loss": 0.9765,
      "step": 19650
    },
    {
      "epoch": 3.6338970451588923,
      "grad_norm": 0.8512163758277893,
      "learning_rate": 3.7560627453896324e-06,
      "loss": 0.6523,
      "step": 19651
    },
    {
      "epoch": 3.6340828842222637,
      "grad_norm": 1.129056453704834,
      "learning_rate": 3.7520973847640973e-06,
      "loss": 0.8549,
      "step": 19652
    },
    {
      "epoch": 3.6342687232856346,
      "grad_norm": 0.9493163824081421,
      "learning_rate": 3.7481340783848904e-06,
      "loss": 0.7761,
      "step": 19653
    },
    {
      "epoch": 3.634454562349006,
      "grad_norm": 0.8012655973434448,
      "learning_rate": 3.7441728263365783e-06,
      "loss": 0.6869,
      "step": 19654
    },
    {
      "epoch": 3.6346404014123768,
      "grad_norm": 0.9964479207992554,
      "learning_rate": 3.740213628703726e-06,
      "loss": 0.7729,
      "step": 19655
    },
    {
      "epoch": 3.634826240475748,
      "grad_norm": 0.9677879810333252,
      "learning_rate": 3.7362564855708106e-06,
      "loss": 0.8043,
      "step": 19656
    },
    {
      "epoch": 3.635012079539119,
      "grad_norm": 1.0772374868392944,
      "learning_rate": 3.73230139702232e-06,
      "loss": 0.875,
      "step": 19657
    },
    {
      "epoch": 3.6351979186024903,
      "grad_norm": 0.9329453110694885,
      "learning_rate": 3.728348363142653e-06,
      "loss": 0.824,
      "step": 19658
    },
    {
      "epoch": 3.6353837576658616,
      "grad_norm": 1.1014840602874756,
      "learning_rate": 3.724397384016165e-06,
      "loss": 0.6482,
      "step": 19659
    },
    {
      "epoch": 3.6355695967292325,
      "grad_norm": 0.9583537578582764,
      "learning_rate": 3.7204484597271995e-06,
      "loss": 0.6224,
      "step": 19660
    },
    {
      "epoch": 3.6357554357926034,
      "grad_norm": 0.9297904968261719,
      "learning_rate": 3.7165015903600555e-06,
      "loss": 0.8584,
      "step": 19661
    },
    {
      "epoch": 3.6359412748559747,
      "grad_norm": 0.7306504845619202,
      "learning_rate": 3.7125567759989545e-06,
      "loss": 0.3111,
      "step": 19662
    },
    {
      "epoch": 3.636127113919346,
      "grad_norm": 0.9969332814216614,
      "learning_rate": 3.708614016728096e-06,
      "loss": 0.6899,
      "step": 19663
    },
    {
      "epoch": 3.636312952982717,
      "grad_norm": 1.0395742654800415,
      "learning_rate": 3.7046733126316126e-06,
      "loss": 0.8837,
      "step": 19664
    },
    {
      "epoch": 3.636498792046088,
      "grad_norm": 0.9263941645622253,
      "learning_rate": 3.700734663793637e-06,
      "loss": 0.7839,
      "step": 19665
    },
    {
      "epoch": 3.636684631109459,
      "grad_norm": 0.901239812374115,
      "learning_rate": 3.696798070298224e-06,
      "loss": 0.7272,
      "step": 19666
    },
    {
      "epoch": 3.6368704701728305,
      "grad_norm": 0.8450802564620972,
      "learning_rate": 3.692863532229407e-06,
      "loss": 0.8683,
      "step": 19667
    },
    {
      "epoch": 3.6370563092362014,
      "grad_norm": 1.290610432624817,
      "learning_rate": 3.6889310496711184e-06,
      "loss": 0.7479,
      "step": 19668
    },
    {
      "epoch": 3.6372421482995727,
      "grad_norm": 0.8974242806434631,
      "learning_rate": 3.6850006227073464e-06,
      "loss": 1.0162,
      "step": 19669
    },
    {
      "epoch": 3.6374279873629436,
      "grad_norm": 0.8204767107963562,
      "learning_rate": 3.6810722514219355e-06,
      "loss": 0.8115,
      "step": 19670
    },
    {
      "epoch": 3.637613826426315,
      "grad_norm": 0.8401081562042236,
      "learning_rate": 3.6771459358987513e-06,
      "loss": 0.6104,
      "step": 19671
    },
    {
      "epoch": 3.637799665489686,
      "grad_norm": 0.9978303909301758,
      "learning_rate": 3.673221676221594e-06,
      "loss": 0.8078,
      "step": 19672
    },
    {
      "epoch": 3.637985504553057,
      "grad_norm": 0.9485687613487244,
      "learning_rate": 3.669299472474219e-06,
      "loss": 0.8662,
      "step": 19673
    },
    {
      "epoch": 3.638171343616428,
      "grad_norm": 1.2545994520187378,
      "learning_rate": 3.665379324740337e-06,
      "loss": 0.9099,
      "step": 19674
    },
    {
      "epoch": 3.6383571826797994,
      "grad_norm": 0.8143329620361328,
      "learning_rate": 3.6614612331036136e-06,
      "loss": 0.8263,
      "step": 19675
    },
    {
      "epoch": 3.6385430217431702,
      "grad_norm": 0.7915576100349426,
      "learning_rate": 3.6575451976476715e-06,
      "loss": 0.7497,
      "step": 19676
    },
    {
      "epoch": 3.6387288608065416,
      "grad_norm": 0.7817553877830505,
      "learning_rate": 3.6536312184560993e-06,
      "loss": 0.7649,
      "step": 19677
    },
    {
      "epoch": 3.638914699869913,
      "grad_norm": 1.7373839616775513,
      "learning_rate": 3.6497192956124414e-06,
      "loss": 1.3472,
      "step": 19678
    },
    {
      "epoch": 3.639100538933284,
      "grad_norm": 0.966449499130249,
      "learning_rate": 3.645809429200164e-06,
      "loss": 0.736,
      "step": 19679
    },
    {
      "epoch": 3.6392863779966547,
      "grad_norm": 1.2942341566085815,
      "learning_rate": 3.6419016193027455e-06,
      "loss": 0.7483,
      "step": 19680
    },
    {
      "epoch": 3.639472217060026,
      "grad_norm": 1.1573847532272339,
      "learning_rate": 3.637995866003574e-06,
      "loss": 0.7297,
      "step": 19681
    },
    {
      "epoch": 3.6396580561233973,
      "grad_norm": 1.0986378192901611,
      "learning_rate": 3.6340921693860276e-06,
      "loss": 1.0598,
      "step": 19682
    },
    {
      "epoch": 3.639843895186768,
      "grad_norm": 1.030543565750122,
      "learning_rate": 3.630190529533417e-06,
      "loss": 0.864,
      "step": 19683
    },
    {
      "epoch": 3.6400297342501395,
      "grad_norm": 0.9853333830833435,
      "learning_rate": 3.626290946528987e-06,
      "loss": 0.6901,
      "step": 19684
    },
    {
      "epoch": 3.6402155733135104,
      "grad_norm": 1.2523033618927002,
      "learning_rate": 3.622393420456016e-06,
      "loss": 0.5345,
      "step": 19685
    },
    {
      "epoch": 3.6404014123768818,
      "grad_norm": 0.8894228935241699,
      "learning_rate": 3.618497951397659e-06,
      "loss": 0.8965,
      "step": 19686
    },
    {
      "epoch": 3.6405872514402526,
      "grad_norm": 1.0515977144241333,
      "learning_rate": 3.614604539437072e-06,
      "loss": 0.8215,
      "step": 19687
    },
    {
      "epoch": 3.640773090503624,
      "grad_norm": 0.8954871892929077,
      "learning_rate": 3.6107131846573327e-06,
      "loss": 0.8407,
      "step": 19688
    },
    {
      "epoch": 3.640958929566995,
      "grad_norm": 0.9122246503829956,
      "learning_rate": 3.6068238871415306e-06,
      "loss": 1.0491,
      "step": 19689
    },
    {
      "epoch": 3.641144768630366,
      "grad_norm": 0.9619038701057434,
      "learning_rate": 3.6029366469726433e-06,
      "loss": 0.8329,
      "step": 19690
    },
    {
      "epoch": 3.641330607693737,
      "grad_norm": 1.3435996770858765,
      "learning_rate": 3.599051464233638e-06,
      "loss": 1.2128,
      "step": 19691
    },
    {
      "epoch": 3.6415164467571084,
      "grad_norm": 0.9356776475906372,
      "learning_rate": 3.5951683390074487e-06,
      "loss": 0.9509,
      "step": 19692
    },
    {
      "epoch": 3.6417022858204797,
      "grad_norm": 1.0035337209701538,
      "learning_rate": 3.5912872713769753e-06,
      "loss": 0.694,
      "step": 19693
    },
    {
      "epoch": 3.6418881248838506,
      "grad_norm": 0.9547224044799805,
      "learning_rate": 3.5874082614250182e-06,
      "loss": 1.0477,
      "step": 19694
    },
    {
      "epoch": 3.6420739639472215,
      "grad_norm": 0.8454679846763611,
      "learning_rate": 3.5835313092343893e-06,
      "loss": 0.9012,
      "step": 19695
    },
    {
      "epoch": 3.642259803010593,
      "grad_norm": 0.962364673614502,
      "learning_rate": 3.579656414887811e-06,
      "loss": 0.7182,
      "step": 19696
    },
    {
      "epoch": 3.642445642073964,
      "grad_norm": 0.9948057532310486,
      "learning_rate": 3.575783578468017e-06,
      "loss": 0.8841,
      "step": 19697
    },
    {
      "epoch": 3.642631481137335,
      "grad_norm": 0.8334029912948608,
      "learning_rate": 3.5719128000576528e-06,
      "loss": 0.7499,
      "step": 19698
    },
    {
      "epoch": 3.642817320200706,
      "grad_norm": 0.984076976776123,
      "learning_rate": 3.56804407973933e-06,
      "loss": 0.8517,
      "step": 19699
    },
    {
      "epoch": 3.6430031592640773,
      "grad_norm": 1.1743614673614502,
      "learning_rate": 3.5641774175956156e-06,
      "loss": 1.0066,
      "step": 19700
    },
    {
      "epoch": 3.6431889983274486,
      "grad_norm": 0.9171334505081177,
      "learning_rate": 3.5603128137090435e-06,
      "loss": 0.903,
      "step": 19701
    },
    {
      "epoch": 3.6433748373908195,
      "grad_norm": 0.8026492595672607,
      "learning_rate": 3.5564502681621037e-06,
      "loss": 0.7917,
      "step": 19702
    },
    {
      "epoch": 3.643560676454191,
      "grad_norm": 1.0340454578399658,
      "learning_rate": 3.55258978103723e-06,
      "loss": 0.8378,
      "step": 19703
    },
    {
      "epoch": 3.6437465155175617,
      "grad_norm": 0.9348717927932739,
      "learning_rate": 3.548731352416812e-06,
      "loss": 0.7884,
      "step": 19704
    },
    {
      "epoch": 3.643932354580933,
      "grad_norm": 0.9344277381896973,
      "learning_rate": 3.5448749823832174e-06,
      "loss": 0.7056,
      "step": 19705
    },
    {
      "epoch": 3.644118193644304,
      "grad_norm": 0.9544888138771057,
      "learning_rate": 3.5410206710187353e-06,
      "loss": 0.8964,
      "step": 19706
    },
    {
      "epoch": 3.6443040327076752,
      "grad_norm": 1.1061723232269287,
      "learning_rate": 3.537168418405634e-06,
      "loss": 0.851,
      "step": 19707
    },
    {
      "epoch": 3.644489871771046,
      "grad_norm": 0.9542726874351501,
      "learning_rate": 3.5333182246261367e-06,
      "loss": 0.8178,
      "step": 19708
    },
    {
      "epoch": 3.6446757108344174,
      "grad_norm": 0.9229307174682617,
      "learning_rate": 3.529470089762421e-06,
      "loss": 0.5897,
      "step": 19709
    },
    {
      "epoch": 3.6448615498977883,
      "grad_norm": 1.1158232688903809,
      "learning_rate": 3.5256240138966224e-06,
      "loss": 0.8865,
      "step": 19710
    },
    {
      "epoch": 3.6450473889611597,
      "grad_norm": 0.9804997444152832,
      "learning_rate": 3.5217799971107966e-06,
      "loss": 0.8154,
      "step": 19711
    },
    {
      "epoch": 3.645233228024531,
      "grad_norm": 0.9301099181175232,
      "learning_rate": 3.5179380394870345e-06,
      "loss": 0.9157,
      "step": 19712
    },
    {
      "epoch": 3.645419067087902,
      "grad_norm": 1.1942648887634277,
      "learning_rate": 3.514098141107314e-06,
      "loss": 0.8761,
      "step": 19713
    },
    {
      "epoch": 3.6456049061512728,
      "grad_norm": 1.1579983234405518,
      "learning_rate": 3.5102603020535916e-06,
      "loss": 0.8602,
      "step": 19714
    },
    {
      "epoch": 3.645790745214644,
      "grad_norm": 0.972847044467926,
      "learning_rate": 3.506424522407781e-06,
      "loss": 0.6708,
      "step": 19715
    },
    {
      "epoch": 3.6459765842780154,
      "grad_norm": 1.035806655883789,
      "learning_rate": 3.5025908022517374e-06,
      "loss": 0.9466,
      "step": 19716
    },
    {
      "epoch": 3.6461624233413863,
      "grad_norm": 0.9446421265602112,
      "learning_rate": 3.498759141667318e-06,
      "loss": 0.9707,
      "step": 19717
    },
    {
      "epoch": 3.6463482624047576,
      "grad_norm": 1.0214331150054932,
      "learning_rate": 3.494929540736269e-06,
      "loss": 0.8919,
      "step": 19718
    },
    {
      "epoch": 3.6465341014681285,
      "grad_norm": 1.0789176225662231,
      "learning_rate": 3.491101999540336e-06,
      "loss": 0.812,
      "step": 19719
    },
    {
      "epoch": 3.6467199405315,
      "grad_norm": 0.8678821921348572,
      "learning_rate": 3.48727651816122e-06,
      "loss": 1.0487,
      "step": 19720
    },
    {
      "epoch": 3.6469057795948707,
      "grad_norm": 1.0004359483718872,
      "learning_rate": 3.4834530966805667e-06,
      "loss": 0.8156,
      "step": 19721
    },
    {
      "epoch": 3.647091618658242,
      "grad_norm": 0.9830246567726135,
      "learning_rate": 3.4796317351799777e-06,
      "loss": 0.5935,
      "step": 19722
    },
    {
      "epoch": 3.647277457721613,
      "grad_norm": 1.0413470268249512,
      "learning_rate": 3.47581243374101e-06,
      "loss": 0.7936,
      "step": 19723
    },
    {
      "epoch": 3.6474632967849843,
      "grad_norm": 0.975347638130188,
      "learning_rate": 3.4719951924451767e-06,
      "loss": 1.0661,
      "step": 19724
    },
    {
      "epoch": 3.647649135848355,
      "grad_norm": 1.0321831703186035,
      "learning_rate": 3.4681800113739783e-06,
      "loss": 0.734,
      "step": 19725
    },
    {
      "epoch": 3.6478349749117265,
      "grad_norm": 1.9831745624542236,
      "learning_rate": 3.4643668906088056e-06,
      "loss": 1.3494,
      "step": 19726
    },
    {
      "epoch": 3.648020813975098,
      "grad_norm": 0.9179535508155823,
      "learning_rate": 3.4605558302310715e-06,
      "loss": 0.8206,
      "step": 19727
    },
    {
      "epoch": 3.6482066530384687,
      "grad_norm": 0.9104013442993164,
      "learning_rate": 3.4567468303220886e-06,
      "loss": 0.7554,
      "step": 19728
    },
    {
      "epoch": 3.6483924921018396,
      "grad_norm": 0.994432806968689,
      "learning_rate": 3.4529398909631803e-06,
      "loss": 0.8357,
      "step": 19729
    },
    {
      "epoch": 3.648578331165211,
      "grad_norm": 0.914111316204071,
      "learning_rate": 3.449135012235594e-06,
      "loss": 0.7204,
      "step": 19730
    },
    {
      "epoch": 3.6487641702285822,
      "grad_norm": 0.9353268146514893,
      "learning_rate": 3.4453321942205186e-06,
      "loss": 0.8731,
      "step": 19731
    },
    {
      "epoch": 3.648950009291953,
      "grad_norm": 0.8555099964141846,
      "learning_rate": 3.4415314369991458e-06,
      "loss": 0.9585,
      "step": 19732
    },
    {
      "epoch": 3.6491358483553245,
      "grad_norm": 0.9863101243972778,
      "learning_rate": 3.437732740652566e-06,
      "loss": 0.8062,
      "step": 19733
    },
    {
      "epoch": 3.6493216874186953,
      "grad_norm": 0.9781383872032166,
      "learning_rate": 3.4339361052618923e-06,
      "loss": 0.7702,
      "step": 19734
    },
    {
      "epoch": 3.6495075264820667,
      "grad_norm": 0.9102252721786499,
      "learning_rate": 3.4301415309081376e-06,
      "loss": 0.8075,
      "step": 19735
    },
    {
      "epoch": 3.6496933655454376,
      "grad_norm": 0.8706068992614746,
      "learning_rate": 3.426349017672281e-06,
      "loss": 0.7584,
      "step": 19736
    },
    {
      "epoch": 3.649879204608809,
      "grad_norm": 1.0591397285461426,
      "learning_rate": 3.422558565635281e-06,
      "loss": 0.7617,
      "step": 19737
    },
    {
      "epoch": 3.6500650436721798,
      "grad_norm": 0.8889842629432678,
      "learning_rate": 3.418770174878039e-06,
      "loss": 0.5825,
      "step": 19738
    },
    {
      "epoch": 3.650250882735551,
      "grad_norm": 1.0698341131210327,
      "learning_rate": 3.414983845481401e-06,
      "loss": 0.7684,
      "step": 19739
    },
    {
      "epoch": 3.650436721798922,
      "grad_norm": 0.9470606446266174,
      "learning_rate": 3.411199577526192e-06,
      "loss": 0.9922,
      "step": 19740
    },
    {
      "epoch": 3.6506225608622933,
      "grad_norm": 0.9126014113426208,
      "learning_rate": 3.40741737109318e-06,
      "loss": 0.8621,
      "step": 19741
    },
    {
      "epoch": 3.6508083999256646,
      "grad_norm": 1.010872483253479,
      "learning_rate": 3.4036372262630787e-06,
      "loss": 0.7708,
      "step": 19742
    },
    {
      "epoch": 3.6509942389890355,
      "grad_norm": 1.04227614402771,
      "learning_rate": 3.3998591431165682e-06,
      "loss": 0.8666,
      "step": 19743
    },
    {
      "epoch": 3.6511800780524064,
      "grad_norm": 1.3853538036346436,
      "learning_rate": 3.3960831217342836e-06,
      "loss": 0.7814,
      "step": 19744
    },
    {
      "epoch": 3.6513659171157777,
      "grad_norm": 0.9948869347572327,
      "learning_rate": 3.3923091621968493e-06,
      "loss": 0.6507,
      "step": 19745
    },
    {
      "epoch": 3.651551756179149,
      "grad_norm": 1.059181809425354,
      "learning_rate": 3.388537264584779e-06,
      "loss": 0.9442,
      "step": 19746
    },
    {
      "epoch": 3.65173759524252,
      "grad_norm": 0.9141784906387329,
      "learning_rate": 3.384767428978597e-06,
      "loss": 0.6792,
      "step": 19747
    },
    {
      "epoch": 3.651923434305891,
      "grad_norm": 0.9963765740394592,
      "learning_rate": 3.3809996554587497e-06,
      "loss": 0.8995,
      "step": 19748
    },
    {
      "epoch": 3.652109273369262,
      "grad_norm": 1.1455966234207153,
      "learning_rate": 3.377233944105662e-06,
      "loss": 0.8618,
      "step": 19749
    },
    {
      "epoch": 3.6522951124326335,
      "grad_norm": 0.9624447822570801,
      "learning_rate": 3.3734702949997034e-06,
      "loss": 0.975,
      "step": 19750
    },
    {
      "epoch": 3.6524809514960044,
      "grad_norm": 1.3912914991378784,
      "learning_rate": 3.3697087082211975e-06,
      "loss": 1.3396,
      "step": 19751
    },
    {
      "epoch": 3.6526667905593757,
      "grad_norm": 1.043426752090454,
      "learning_rate": 3.3659491838504364e-06,
      "loss": 0.8956,
      "step": 19752
    },
    {
      "epoch": 3.6528526296227466,
      "grad_norm": 0.8907557725906372,
      "learning_rate": 3.3621917219676666e-06,
      "loss": 0.7101,
      "step": 19753
    },
    {
      "epoch": 3.653038468686118,
      "grad_norm": 1.01949942111969,
      "learning_rate": 3.358436322653069e-06,
      "loss": 0.392,
      "step": 19754
    },
    {
      "epoch": 3.653224307749489,
      "grad_norm": 1.0524967908859253,
      "learning_rate": 3.3546829859868123e-06,
      "loss": 0.855,
      "step": 19755
    },
    {
      "epoch": 3.65341014681286,
      "grad_norm": 0.8573651313781738,
      "learning_rate": 3.350931712048988e-06,
      "loss": 0.8067,
      "step": 19756
    },
    {
      "epoch": 3.653595985876231,
      "grad_norm": 1.0655800104141235,
      "learning_rate": 3.3471825009196766e-06,
      "loss": 0.6145,
      "step": 19757
    },
    {
      "epoch": 3.6537818249396024,
      "grad_norm": 0.8858900666236877,
      "learning_rate": 3.3434353526789032e-06,
      "loss": 0.906,
      "step": 19758
    },
    {
      "epoch": 3.6539676640029732,
      "grad_norm": 1.1333801746368408,
      "learning_rate": 3.3396902674066033e-06,
      "loss": 1.0519,
      "step": 19759
    },
    {
      "epoch": 3.6541535030663446,
      "grad_norm": 1.1849747896194458,
      "learning_rate": 3.335947245182769e-06,
      "loss": 1.061,
      "step": 19760
    },
    {
      "epoch": 3.654339342129716,
      "grad_norm": 0.9107005596160889,
      "learning_rate": 3.332206286087247e-06,
      "loss": 0.7066,
      "step": 19761
    },
    {
      "epoch": 3.654525181193087,
      "grad_norm": 0.9740599393844604,
      "learning_rate": 3.3284673901998964e-06,
      "loss": 0.6674,
      "step": 19762
    },
    {
      "epoch": 3.6547110202564577,
      "grad_norm": 0.8133429884910583,
      "learning_rate": 3.324730557600497e-06,
      "loss": 0.6233,
      "step": 19763
    },
    {
      "epoch": 3.654896859319829,
      "grad_norm": 0.9082481861114502,
      "learning_rate": 3.3209957883688304e-06,
      "loss": 0.9023,
      "step": 19764
    },
    {
      "epoch": 3.6550826983832003,
      "grad_norm": 0.939697802066803,
      "learning_rate": 3.3172630825846095e-06,
      "loss": 0.7883,
      "step": 19765
    },
    {
      "epoch": 3.655268537446571,
      "grad_norm": 0.9083552360534668,
      "learning_rate": 3.3135324403274935e-06,
      "loss": 1.0788,
      "step": 19766
    },
    {
      "epoch": 3.6554543765099425,
      "grad_norm": 0.9659096598625183,
      "learning_rate": 3.3098038616771075e-06,
      "loss": 0.869,
      "step": 19767
    },
    {
      "epoch": 3.6556402155733134,
      "grad_norm": 1.592389702796936,
      "learning_rate": 3.306077346713021e-06,
      "loss": 1.145,
      "step": 19768
    },
    {
      "epoch": 3.6558260546366848,
      "grad_norm": 1.1005619764328003,
      "learning_rate": 3.302352895514793e-06,
      "loss": 0.9374,
      "step": 19769
    },
    {
      "epoch": 3.6560118937000556,
      "grad_norm": 0.9759562611579895,
      "learning_rate": 3.298630508161904e-06,
      "loss": 0.893,
      "step": 19770
    },
    {
      "epoch": 3.656197732763427,
      "grad_norm": 0.9161255359649658,
      "learning_rate": 3.2949101847337792e-06,
      "loss": 0.7266,
      "step": 19771
    },
    {
      "epoch": 3.656383571826798,
      "grad_norm": 0.9906998872756958,
      "learning_rate": 3.2911919253098664e-06,
      "loss": 0.859,
      "step": 19772
    },
    {
      "epoch": 3.656569410890169,
      "grad_norm": 0.9362435936927795,
      "learning_rate": 3.2874757299694913e-06,
      "loss": 1.032,
      "step": 19773
    },
    {
      "epoch": 3.65675524995354,
      "grad_norm": 1.1768964529037476,
      "learning_rate": 3.2837615987919902e-06,
      "loss": 0.7615,
      "step": 19774
    },
    {
      "epoch": 3.6569410890169114,
      "grad_norm": 0.9677229523658752,
      "learning_rate": 3.2800495318566105e-06,
      "loss": 0.7366,
      "step": 19775
    },
    {
      "epoch": 3.6571269280802827,
      "grad_norm": 1.0703089237213135,
      "learning_rate": 3.2763395292426004e-06,
      "loss": 0.8577,
      "step": 19776
    },
    {
      "epoch": 3.6573127671436536,
      "grad_norm": 0.8955932855606079,
      "learning_rate": 3.2726315910291516e-06,
      "loss": 0.6111,
      "step": 19777
    },
    {
      "epoch": 3.6574986062070245,
      "grad_norm": 1.3545371294021606,
      "learning_rate": 3.268925717295379e-06,
      "loss": 0.8886,
      "step": 19778
    },
    {
      "epoch": 3.657684445270396,
      "grad_norm": 0.9368428587913513,
      "learning_rate": 3.265221908120386e-06,
      "loss": 0.8056,
      "step": 19779
    },
    {
      "epoch": 3.657870284333767,
      "grad_norm": 0.8998544812202454,
      "learning_rate": 3.2615201635832428e-06,
      "loss": 0.7344,
      "step": 19780
    },
    {
      "epoch": 3.658056123397138,
      "grad_norm": 0.9378226399421692,
      "learning_rate": 3.257820483762941e-06,
      "loss": 0.9107,
      "step": 19781
    },
    {
      "epoch": 3.658241962460509,
      "grad_norm": 1.0062965154647827,
      "learning_rate": 3.2541228687384406e-06,
      "loss": 1.3038,
      "step": 19782
    },
    {
      "epoch": 3.6584278015238803,
      "grad_norm": 0.8791436553001404,
      "learning_rate": 3.2504273185886447e-06,
      "loss": 0.9039,
      "step": 19783
    },
    {
      "epoch": 3.6586136405872516,
      "grad_norm": 0.869947075843811,
      "learning_rate": 3.246733833392479e-06,
      "loss": 0.8601,
      "step": 19784
    },
    {
      "epoch": 3.6587994796506225,
      "grad_norm": 0.8931410312652588,
      "learning_rate": 3.2430424132287253e-06,
      "loss": 0.9305,
      "step": 19785
    },
    {
      "epoch": 3.658985318713994,
      "grad_norm": 0.9086544513702393,
      "learning_rate": 3.2393530581761977e-06,
      "loss": 0.9764,
      "step": 19786
    },
    {
      "epoch": 3.6591711577773647,
      "grad_norm": 0.839087188243866,
      "learning_rate": 3.235665768313634e-06,
      "loss": 0.8387,
      "step": 19787
    },
    {
      "epoch": 3.659356996840736,
      "grad_norm": 1.050937533378601,
      "learning_rate": 3.231980543719715e-06,
      "loss": 0.814,
      "step": 19788
    },
    {
      "epoch": 3.659542835904107,
      "grad_norm": 1.0313888788223267,
      "learning_rate": 3.2282973844731224e-06,
      "loss": 0.7997,
      "step": 19789
    },
    {
      "epoch": 3.6597286749674782,
      "grad_norm": 1.1018502712249756,
      "learning_rate": 3.22461629065246e-06,
      "loss": 1.0423,
      "step": 19790
    },
    {
      "epoch": 3.6599145140308496,
      "grad_norm": 1.0087827444076538,
      "learning_rate": 3.2209372623362765e-06,
      "loss": 0.8628,
      "step": 19791
    },
    {
      "epoch": 3.6601003530942204,
      "grad_norm": 0.9059208035469055,
      "learning_rate": 3.21726029960312e-06,
      "loss": 0.7809,
      "step": 19792
    },
    {
      "epoch": 3.6602861921575913,
      "grad_norm": 0.99634850025177,
      "learning_rate": 3.2135854025314495e-06,
      "loss": 0.8574,
      "step": 19793
    },
    {
      "epoch": 3.6604720312209627,
      "grad_norm": 1.172855019569397,
      "learning_rate": 3.2099125711997134e-06,
      "loss": 0.6191,
      "step": 19794
    },
    {
      "epoch": 3.660657870284334,
      "grad_norm": 0.9979076981544495,
      "learning_rate": 3.2062418056862833e-06,
      "loss": 1.0289,
      "step": 19795
    },
    {
      "epoch": 3.660843709347705,
      "grad_norm": 0.9946637153625488,
      "learning_rate": 3.2025731060695174e-06,
      "loss": 0.8253,
      "step": 19796
    },
    {
      "epoch": 3.6610295484110758,
      "grad_norm": 0.9385978579521179,
      "learning_rate": 3.198906472427732e-06,
      "loss": 0.9922,
      "step": 19797
    },
    {
      "epoch": 3.661215387474447,
      "grad_norm": 0.7791514992713928,
      "learning_rate": 3.1952419048391635e-06,
      "loss": 0.6416,
      "step": 19798
    },
    {
      "epoch": 3.6614012265378184,
      "grad_norm": 1.0495576858520508,
      "learning_rate": 3.191579403382028e-06,
      "loss": 0.8827,
      "step": 19799
    },
    {
      "epoch": 3.6615870656011893,
      "grad_norm": 1.1099753379821777,
      "learning_rate": 3.187918968134507e-06,
      "loss": 0.9144,
      "step": 19800
    },
    {
      "epoch": 3.6617729046645606,
      "grad_norm": 1.0003912448883057,
      "learning_rate": 3.1842605991747154e-06,
      "loss": 0.7915,
      "step": 19801
    },
    {
      "epoch": 3.6619587437279315,
      "grad_norm": 0.793580949306488,
      "learning_rate": 3.180604296580747e-06,
      "loss": 0.8196,
      "step": 19802
    },
    {
      "epoch": 3.662144582791303,
      "grad_norm": 0.8666147589683533,
      "learning_rate": 3.1769500604306167e-06,
      "loss": 0.7983,
      "step": 19803
    },
    {
      "epoch": 3.6623304218546737,
      "grad_norm": 1.5029810667037964,
      "learning_rate": 3.1732978908023405e-06,
      "loss": 1.3051,
      "step": 19804
    },
    {
      "epoch": 3.662516260918045,
      "grad_norm": 0.959281861782074,
      "learning_rate": 3.169647787773866e-06,
      "loss": 0.7916,
      "step": 19805
    },
    {
      "epoch": 3.662702099981416,
      "grad_norm": 1.0968326330184937,
      "learning_rate": 3.1659997514230768e-06,
      "loss": 0.9021,
      "step": 19806
    },
    {
      "epoch": 3.6628879390447873,
      "grad_norm": 1.0215089321136475,
      "learning_rate": 3.162353781827865e-06,
      "loss": 0.8865,
      "step": 19807
    },
    {
      "epoch": 3.663073778108158,
      "grad_norm": 1.0578625202178955,
      "learning_rate": 3.158709879066013e-06,
      "loss": 0.8077,
      "step": 19808
    },
    {
      "epoch": 3.6632596171715295,
      "grad_norm": 1.015753149986267,
      "learning_rate": 3.155068043215326e-06,
      "loss": 0.8311,
      "step": 19809
    },
    {
      "epoch": 3.663445456234901,
      "grad_norm": 0.956145703792572,
      "learning_rate": 3.1514282743535184e-06,
      "loss": 0.8834,
      "step": 19810
    },
    {
      "epoch": 3.6636312952982717,
      "grad_norm": 0.9006800055503845,
      "learning_rate": 3.1477905725582623e-06,
      "loss": 0.7183,
      "step": 19811
    },
    {
      "epoch": 3.6638171343616426,
      "grad_norm": 0.8815486431121826,
      "learning_rate": 3.144154937907229e-06,
      "loss": 0.785,
      "step": 19812
    },
    {
      "epoch": 3.664002973425014,
      "grad_norm": 0.9718546867370605,
      "learning_rate": 3.1405213704780002e-06,
      "loss": 0.7752,
      "step": 19813
    },
    {
      "epoch": 3.6641888124883852,
      "grad_norm": 0.9407109022140503,
      "learning_rate": 3.1368898703481143e-06,
      "loss": 0.7941,
      "step": 19814
    },
    {
      "epoch": 3.664374651551756,
      "grad_norm": 0.9799614548683167,
      "learning_rate": 3.133260437595087e-06,
      "loss": 0.8844,
      "step": 19815
    },
    {
      "epoch": 3.6645604906151275,
      "grad_norm": 1.2716703414916992,
      "learning_rate": 3.1296330722963897e-06,
      "loss": 0.7302,
      "step": 19816
    },
    {
      "epoch": 3.6647463296784983,
      "grad_norm": 0.8386351466178894,
      "learning_rate": 3.126007774529438e-06,
      "loss": 0.7081,
      "step": 19817
    },
    {
      "epoch": 3.6649321687418697,
      "grad_norm": 1.0708590745925903,
      "learning_rate": 3.1223845443716037e-06,
      "loss": 0.9288,
      "step": 19818
    },
    {
      "epoch": 3.6651180078052406,
      "grad_norm": 1.1155301332473755,
      "learning_rate": 3.118763381900225e-06,
      "loss": 0.998,
      "step": 19819
    },
    {
      "epoch": 3.665303846868612,
      "grad_norm": 0.9290077090263367,
      "learning_rate": 3.1151442871925952e-06,
      "loss": 0.5783,
      "step": 19820
    },
    {
      "epoch": 3.6654896859319828,
      "grad_norm": 0.8495419025421143,
      "learning_rate": 3.1115272603259416e-06,
      "loss": 0.7152,
      "step": 19821
    },
    {
      "epoch": 3.665675524995354,
      "grad_norm": 0.8739228844642639,
      "learning_rate": 3.10791230137748e-06,
      "loss": 0.6829,
      "step": 19822
    },
    {
      "epoch": 3.665861364058725,
      "grad_norm": 0.8517270088195801,
      "learning_rate": 3.1042994104243494e-06,
      "loss": 0.9774,
      "step": 19823
    },
    {
      "epoch": 3.6660472031220963,
      "grad_norm": 1.1146289110183716,
      "learning_rate": 3.100688587543665e-06,
      "loss": 1.083,
      "step": 19824
    },
    {
      "epoch": 3.6662330421854676,
      "grad_norm": 0.885505735874176,
      "learning_rate": 3.0970798328125105e-06,
      "loss": 0.6914,
      "step": 19825
    },
    {
      "epoch": 3.6664188812488385,
      "grad_norm": 0.7472629547119141,
      "learning_rate": 3.0934731463078904e-06,
      "loss": 0.7286,
      "step": 19826
    },
    {
      "epoch": 3.6666047203122094,
      "grad_norm": 0.8759147524833679,
      "learning_rate": 3.089868528106776e-06,
      "loss": 0.6962,
      "step": 19827
    },
    {
      "epoch": 3.6667905593755807,
      "grad_norm": 0.9485564231872559,
      "learning_rate": 3.086265978286107e-06,
      "loss": 0.9121,
      "step": 19828
    },
    {
      "epoch": 3.666976398438952,
      "grad_norm": 1.0498046875,
      "learning_rate": 3.0826654969227987e-06,
      "loss": 0.788,
      "step": 19829
    },
    {
      "epoch": 3.667162237502323,
      "grad_norm": 1.0976457595825195,
      "learning_rate": 3.0790670840936788e-06,
      "loss": 0.6969,
      "step": 19830
    },
    {
      "epoch": 3.667348076565694,
      "grad_norm": 0.9646531939506531,
      "learning_rate": 3.075470739875541e-06,
      "loss": 0.9687,
      "step": 19831
    },
    {
      "epoch": 3.667533915629065,
      "grad_norm": 0.9509338736534119,
      "learning_rate": 3.0718764643451582e-06,
      "loss": 0.8766,
      "step": 19832
    },
    {
      "epoch": 3.6677197546924365,
      "grad_norm": 1.0596628189086914,
      "learning_rate": 3.068284257579235e-06,
      "loss": 0.8185,
      "step": 19833
    },
    {
      "epoch": 3.6679055937558074,
      "grad_norm": 0.8546445965766907,
      "learning_rate": 3.064694119654443e-06,
      "loss": 0.6911,
      "step": 19834
    },
    {
      "epoch": 3.6680914328191787,
      "grad_norm": 0.8668458461761475,
      "learning_rate": 3.061106050647389e-06,
      "loss": 0.7985,
      "step": 19835
    },
    {
      "epoch": 3.6682772718825496,
      "grad_norm": 0.9155436754226685,
      "learning_rate": 3.057520050634699e-06,
      "loss": 0.8708,
      "step": 19836
    },
    {
      "epoch": 3.668463110945921,
      "grad_norm": 0.913185715675354,
      "learning_rate": 3.0539361196928682e-06,
      "loss": 0.7792,
      "step": 19837
    },
    {
      "epoch": 3.668648950009292,
      "grad_norm": 1.0407506227493286,
      "learning_rate": 3.050354257898402e-06,
      "loss": 0.7501,
      "step": 19838
    },
    {
      "epoch": 3.668834789072663,
      "grad_norm": 1.0762405395507812,
      "learning_rate": 3.0467744653277617e-06,
      "loss": 0.8153,
      "step": 19839
    },
    {
      "epoch": 3.6690206281360345,
      "grad_norm": 0.9334627389907837,
      "learning_rate": 3.0431967420573414e-06,
      "loss": 0.8077,
      "step": 19840
    },
    {
      "epoch": 3.6692064671994054,
      "grad_norm": 1.1207897663116455,
      "learning_rate": 3.0396210881635023e-06,
      "loss": 0.8184,
      "step": 19841
    },
    {
      "epoch": 3.6693923062627762,
      "grad_norm": 1.1474028825759888,
      "learning_rate": 3.036047503722561e-06,
      "loss": 0.945,
      "step": 19842
    },
    {
      "epoch": 3.6695781453261476,
      "grad_norm": 0.8330485820770264,
      "learning_rate": 3.032475988810779e-06,
      "loss": 0.7401,
      "step": 19843
    },
    {
      "epoch": 3.669763984389519,
      "grad_norm": 0.9342332482337952,
      "learning_rate": 3.028906543504406e-06,
      "loss": 0.731,
      "step": 19844
    },
    {
      "epoch": 3.66994982345289,
      "grad_norm": 0.9211150407791138,
      "learning_rate": 3.025339167879615e-06,
      "loss": 0.7433,
      "step": 19845
    },
    {
      "epoch": 3.6701356625162607,
      "grad_norm": 0.8624177575111389,
      "learning_rate": 3.021773862012545e-06,
      "loss": 0.8936,
      "step": 19846
    },
    {
      "epoch": 3.670321501579632,
      "grad_norm": 0.8853157758712769,
      "learning_rate": 3.0182106259792676e-06,
      "loss": 0.9055,
      "step": 19847
    },
    {
      "epoch": 3.6705073406430033,
      "grad_norm": 0.8142578601837158,
      "learning_rate": 3.0146494598558674e-06,
      "loss": 0.6743,
      "step": 19848
    },
    {
      "epoch": 3.670693179706374,
      "grad_norm": 0.9826507568359375,
      "learning_rate": 3.0110903637183497e-06,
      "loss": 0.7625,
      "step": 19849
    },
    {
      "epoch": 3.6708790187697455,
      "grad_norm": 0.9327687621116638,
      "learning_rate": 3.007533337642676e-06,
      "loss": 0.7594,
      "step": 19850
    },
    {
      "epoch": 3.6710648578331164,
      "grad_norm": 0.9218081831932068,
      "learning_rate": 3.0039783817047418e-06,
      "loss": 0.7935,
      "step": 19851
    },
    {
      "epoch": 3.6712506968964878,
      "grad_norm": 1.0486152172088623,
      "learning_rate": 3.000425495980441e-06,
      "loss": 0.8762,
      "step": 19852
    },
    {
      "epoch": 3.6714365359598586,
      "grad_norm": 0.8621472120285034,
      "learning_rate": 2.996874680545603e-06,
      "loss": 0.8295,
      "step": 19853
    },
    {
      "epoch": 3.67162237502323,
      "grad_norm": 1.0741668939590454,
      "learning_rate": 2.99332593547601e-06,
      "loss": 0.9637,
      "step": 19854
    },
    {
      "epoch": 3.671808214086601,
      "grad_norm": 0.8611621856689453,
      "learning_rate": 2.989779260847403e-06,
      "loss": 0.7141,
      "step": 19855
    },
    {
      "epoch": 3.671994053149972,
      "grad_norm": 0.8329762816429138,
      "learning_rate": 2.986234656735487e-06,
      "loss": 0.8697,
      "step": 19856
    },
    {
      "epoch": 3.672179892213343,
      "grad_norm": 0.9581859707832336,
      "learning_rate": 2.982692123215913e-06,
      "loss": 0.9079,
      "step": 19857
    },
    {
      "epoch": 3.6723657312767144,
      "grad_norm": 0.9887629151344299,
      "learning_rate": 2.9791516603642765e-06,
      "loss": 0.8446,
      "step": 19858
    },
    {
      "epoch": 3.6725515703400857,
      "grad_norm": 0.9503438472747803,
      "learning_rate": 2.9756132682561387e-06,
      "loss": 0.9235,
      "step": 19859
    },
    {
      "epoch": 3.6727374094034566,
      "grad_norm": 0.898719847202301,
      "learning_rate": 2.9720769469670617e-06,
      "loss": 0.7815,
      "step": 19860
    },
    {
      "epoch": 3.6729232484668275,
      "grad_norm": 0.975048303604126,
      "learning_rate": 2.968542696572485e-06,
      "loss": 0.8249,
      "step": 19861
    },
    {
      "epoch": 3.673109087530199,
      "grad_norm": 0.8516594171524048,
      "learning_rate": 2.9650105171478593e-06,
      "loss": 0.7859,
      "step": 19862
    },
    {
      "epoch": 3.67329492659357,
      "grad_norm": 1.0244548320770264,
      "learning_rate": 2.961480408768558e-06,
      "loss": 0.8145,
      "step": 19863
    },
    {
      "epoch": 3.673480765656941,
      "grad_norm": 0.9714505672454834,
      "learning_rate": 2.957952371509931e-06,
      "loss": 0.9863,
      "step": 19864
    },
    {
      "epoch": 3.6736666047203124,
      "grad_norm": 0.9386559724807739,
      "learning_rate": 2.954426405447297e-06,
      "loss": 0.6725,
      "step": 19865
    },
    {
      "epoch": 3.6738524437836833,
      "grad_norm": 0.9028299450874329,
      "learning_rate": 2.9509025106558843e-06,
      "loss": 0.9663,
      "step": 19866
    },
    {
      "epoch": 3.6740382828470546,
      "grad_norm": 1.3612467050552368,
      "learning_rate": 2.9473806872108988e-06,
      "loss": 0.951,
      "step": 19867
    },
    {
      "epoch": 3.6742241219104255,
      "grad_norm": 0.9654171466827393,
      "learning_rate": 2.9438609351875366e-06,
      "loss": 0.9266,
      "step": 19868
    },
    {
      "epoch": 3.674409960973797,
      "grad_norm": 0.8086720705032349,
      "learning_rate": 2.9403432546609043e-06,
      "loss": 0.7396,
      "step": 19869
    },
    {
      "epoch": 3.6745958000371677,
      "grad_norm": 0.915579617023468,
      "learning_rate": 2.936827645706086e-06,
      "loss": 0.6547,
      "step": 19870
    },
    {
      "epoch": 3.674781639100539,
      "grad_norm": 0.8911121487617493,
      "learning_rate": 2.933314108398111e-06,
      "loss": 0.7598,
      "step": 19871
    },
    {
      "epoch": 3.67496747816391,
      "grad_norm": 0.9755234718322754,
      "learning_rate": 2.929802642811985e-06,
      "loss": 0.9399,
      "step": 19872
    },
    {
      "epoch": 3.6751533172272812,
      "grad_norm": 0.9139371514320374,
      "learning_rate": 2.926293249022638e-06,
      "loss": 0.9042,
      "step": 19873
    },
    {
      "epoch": 3.6753391562906526,
      "grad_norm": 1.0513211488723755,
      "learning_rate": 2.9227859271049763e-06,
      "loss": 1.0011,
      "step": 19874
    },
    {
      "epoch": 3.6755249953540234,
      "grad_norm": 0.884815514087677,
      "learning_rate": 2.919280677133851e-06,
      "loss": 0.7407,
      "step": 19875
    },
    {
      "epoch": 3.6757108344173943,
      "grad_norm": 0.8955864906311035,
      "learning_rate": 2.9157774991840915e-06,
      "loss": 0.8038,
      "step": 19876
    },
    {
      "epoch": 3.6758966734807657,
      "grad_norm": 1.3236089944839478,
      "learning_rate": 2.9122763933304486e-06,
      "loss": 1.0079,
      "step": 19877
    },
    {
      "epoch": 3.676082512544137,
      "grad_norm": 0.8637065291404724,
      "learning_rate": 2.908777359647674e-06,
      "loss": 0.7272,
      "step": 19878
    },
    {
      "epoch": 3.676268351607508,
      "grad_norm": 1.0538060665130615,
      "learning_rate": 2.9052803982104083e-06,
      "loss": 0.6253,
      "step": 19879
    },
    {
      "epoch": 3.6764541906708788,
      "grad_norm": 1.1004332304000854,
      "learning_rate": 2.901785509093313e-06,
      "loss": 0.816,
      "step": 19880
    },
    {
      "epoch": 3.67664002973425,
      "grad_norm": 1.302672266960144,
      "learning_rate": 2.8982926923709962e-06,
      "loss": 1.2928,
      "step": 19881
    },
    {
      "epoch": 3.6768258687976214,
      "grad_norm": 0.9969615936279297,
      "learning_rate": 2.8948019481179757e-06,
      "loss": 0.9255,
      "step": 19882
    },
    {
      "epoch": 3.6770117078609923,
      "grad_norm": 0.9608601927757263,
      "learning_rate": 2.8913132764087693e-06,
      "loss": 0.8857,
      "step": 19883
    },
    {
      "epoch": 3.6771975469243636,
      "grad_norm": 0.8944962024688721,
      "learning_rate": 2.8878266773178396e-06,
      "loss": 0.9644,
      "step": 19884
    },
    {
      "epoch": 3.6773833859877345,
      "grad_norm": 1.051511526107788,
      "learning_rate": 2.8843421509196057e-06,
      "loss": 1.0011,
      "step": 19885
    },
    {
      "epoch": 3.677569225051106,
      "grad_norm": 0.9785056710243225,
      "learning_rate": 2.8808596972884296e-06,
      "loss": 0.949,
      "step": 19886
    },
    {
      "epoch": 3.6777550641144767,
      "grad_norm": 0.9483036398887634,
      "learning_rate": 2.8773793164986298e-06,
      "loss": 0.8208,
      "step": 19887
    },
    {
      "epoch": 3.677940903177848,
      "grad_norm": 1.0405542850494385,
      "learning_rate": 2.873901008624502e-06,
      "loss": 0.7559,
      "step": 19888
    },
    {
      "epoch": 3.678126742241219,
      "grad_norm": 1.360510230064392,
      "learning_rate": 2.870424773740288e-06,
      "loss": 0.9153,
      "step": 19889
    },
    {
      "epoch": 3.6783125813045903,
      "grad_norm": 0.8900576829910278,
      "learning_rate": 2.8669506119201716e-06,
      "loss": 0.9084,
      "step": 19890
    },
    {
      "epoch": 3.678498420367961,
      "grad_norm": 0.9248895049095154,
      "learning_rate": 2.8634785232382944e-06,
      "loss": 0.9498,
      "step": 19891
    },
    {
      "epoch": 3.6786842594313325,
      "grad_norm": 1.0271313190460205,
      "learning_rate": 2.8600085077687965e-06,
      "loss": 0.7387,
      "step": 19892
    },
    {
      "epoch": 3.678870098494704,
      "grad_norm": 0.8631148934364319,
      "learning_rate": 2.8565405655857192e-06,
      "loss": 0.8698,
      "step": 19893
    },
    {
      "epoch": 3.6790559375580747,
      "grad_norm": 0.9825854897499084,
      "learning_rate": 2.8530746967630695e-06,
      "loss": 0.7327,
      "step": 19894
    },
    {
      "epoch": 3.6792417766214456,
      "grad_norm": 0.8682910203933716,
      "learning_rate": 2.8496109013748217e-06,
      "loss": 0.6029,
      "step": 19895
    },
    {
      "epoch": 3.679427615684817,
      "grad_norm": 0.9333560466766357,
      "learning_rate": 2.846149179494928e-06,
      "loss": 0.7099,
      "step": 19896
    },
    {
      "epoch": 3.6796134547481882,
      "grad_norm": 1.0085049867630005,
      "learning_rate": 2.8426895311972513e-06,
      "loss": 0.6731,
      "step": 19897
    },
    {
      "epoch": 3.679799293811559,
      "grad_norm": 0.9532813429832458,
      "learning_rate": 2.8392319565556436e-06,
      "loss": 0.9296,
      "step": 19898
    },
    {
      "epoch": 3.6799851328749305,
      "grad_norm": 0.8818166851997375,
      "learning_rate": 2.835776455643879e-06,
      "loss": 0.7739,
      "step": 19899
    },
    {
      "epoch": 3.6801709719383013,
      "grad_norm": 0.970205545425415,
      "learning_rate": 2.8323230285357327e-06,
      "loss": 0.7891,
      "step": 19900
    },
    {
      "epoch": 3.6803568110016727,
      "grad_norm": 0.8429179787635803,
      "learning_rate": 2.8288716753049005e-06,
      "loss": 0.7411,
      "step": 19901
    },
    {
      "epoch": 3.6805426500650436,
      "grad_norm": 1.096352219581604,
      "learning_rate": 2.825422396025068e-06,
      "loss": 0.872,
      "step": 19902
    },
    {
      "epoch": 3.680728489128415,
      "grad_norm": 0.8698795437812805,
      "learning_rate": 2.82197519076981e-06,
      "loss": 0.7923,
      "step": 19903
    },
    {
      "epoch": 3.6809143281917858,
      "grad_norm": 0.8768172264099121,
      "learning_rate": 2.818530059612745e-06,
      "loss": 0.8951,
      "step": 19904
    },
    {
      "epoch": 3.681100167255157,
      "grad_norm": 0.9791592359542847,
      "learning_rate": 2.8150870026273922e-06,
      "loss": 0.6247,
      "step": 19905
    },
    {
      "epoch": 3.681286006318528,
      "grad_norm": 0.9714826941490173,
      "learning_rate": 2.8116460198872264e-06,
      "loss": 1.048,
      "step": 19906
    },
    {
      "epoch": 3.6814718453818993,
      "grad_norm": 0.9335835576057434,
      "learning_rate": 2.8082071114656772e-06,
      "loss": 0.7909,
      "step": 19907
    },
    {
      "epoch": 3.6816576844452706,
      "grad_norm": 0.97763991355896,
      "learning_rate": 2.8047702774361863e-06,
      "loss": 0.9408,
      "step": 19908
    },
    {
      "epoch": 3.6818435235086415,
      "grad_norm": 0.8466770052909851,
      "learning_rate": 2.8013355178720612e-06,
      "loss": 0.8742,
      "step": 19909
    },
    {
      "epoch": 3.6820293625720124,
      "grad_norm": 1.0056238174438477,
      "learning_rate": 2.7979028328466328e-06,
      "loss": 0.7738,
      "step": 19910
    },
    {
      "epoch": 3.6822152016353837,
      "grad_norm": 0.8779562711715698,
      "learning_rate": 2.794472222433153e-06,
      "loss": 0.8267,
      "step": 19911
    },
    {
      "epoch": 3.682401040698755,
      "grad_norm": 0.9409157633781433,
      "learning_rate": 2.791043686704875e-06,
      "loss": 0.9228,
      "step": 19912
    },
    {
      "epoch": 3.682586879762126,
      "grad_norm": 0.95030277967453,
      "learning_rate": 2.78761722573494e-06,
      "loss": 0.8736,
      "step": 19913
    },
    {
      "epoch": 3.6827727188254973,
      "grad_norm": 1.142154335975647,
      "learning_rate": 2.7841928395965e-06,
      "loss": 0.902,
      "step": 19914
    },
    {
      "epoch": 3.682958557888868,
      "grad_norm": 0.8686795234680176,
      "learning_rate": 2.7807705283626194e-06,
      "loss": 0.8466,
      "step": 19915
    },
    {
      "epoch": 3.6831443969522395,
      "grad_norm": 1.1532697677612305,
      "learning_rate": 2.7773502921063733e-06,
      "loss": 0.8168,
      "step": 19916
    },
    {
      "epoch": 3.6833302360156104,
      "grad_norm": 1.1650819778442383,
      "learning_rate": 2.773932130900736e-06,
      "loss": 0.758,
      "step": 19917
    },
    {
      "epoch": 3.6835160750789817,
      "grad_norm": 0.8981561064720154,
      "learning_rate": 2.770516044818683e-06,
      "loss": 0.7541,
      "step": 19918
    },
    {
      "epoch": 3.6837019141423526,
      "grad_norm": 1.029568076133728,
      "learning_rate": 2.7671020339330888e-06,
      "loss": 0.7395,
      "step": 19919
    },
    {
      "epoch": 3.683887753205724,
      "grad_norm": 1.0341814756393433,
      "learning_rate": 2.7636900983168625e-06,
      "loss": 0.8342,
      "step": 19920
    },
    {
      "epoch": 3.684073592269095,
      "grad_norm": 1.0072674751281738,
      "learning_rate": 2.7602802380428007e-06,
      "loss": 0.882,
      "step": 19921
    },
    {
      "epoch": 3.684259431332466,
      "grad_norm": 0.8727888464927673,
      "learning_rate": 2.756872453183679e-06,
      "loss": 0.7647,
      "step": 19922
    },
    {
      "epoch": 3.6844452703958375,
      "grad_norm": 0.9565852284431458,
      "learning_rate": 2.7534667438122387e-06,
      "loss": 0.6707,
      "step": 19923
    },
    {
      "epoch": 3.6846311094592084,
      "grad_norm": 0.9715271592140198,
      "learning_rate": 2.7500631100011888e-06,
      "loss": 0.8839,
      "step": 19924
    },
    {
      "epoch": 3.6848169485225792,
      "grad_norm": 0.8984237909317017,
      "learning_rate": 2.7466615518231486e-06,
      "loss": 0.8218,
      "step": 19925
    },
    {
      "epoch": 3.6850027875859506,
      "grad_norm": 1.039727807044983,
      "learning_rate": 2.743262069350716e-06,
      "loss": 0.9666,
      "step": 19926
    },
    {
      "epoch": 3.685188626649322,
      "grad_norm": 1.0353304147720337,
      "learning_rate": 2.739864662656455e-06,
      "loss": 0.7697,
      "step": 19927
    },
    {
      "epoch": 3.685374465712693,
      "grad_norm": 1.000001311302185,
      "learning_rate": 2.7364693318128744e-06,
      "loss": 0.7935,
      "step": 19928
    },
    {
      "epoch": 3.6855603047760637,
      "grad_norm": 0.905925452709198,
      "learning_rate": 2.7330760768924603e-06,
      "loss": 0.7919,
      "step": 19929
    },
    {
      "epoch": 3.685746143839435,
      "grad_norm": 0.8545135855674744,
      "learning_rate": 2.7296848979675992e-06,
      "loss": 0.7409,
      "step": 19930
    },
    {
      "epoch": 3.6859319829028063,
      "grad_norm": 1.0508830547332764,
      "learning_rate": 2.7262957951107116e-06,
      "loss": 0.7794,
      "step": 19931
    },
    {
      "epoch": 3.686117821966177,
      "grad_norm": 0.8892425894737244,
      "learning_rate": 2.7229087683940946e-06,
      "loss": 0.5947,
      "step": 19932
    },
    {
      "epoch": 3.6863036610295485,
      "grad_norm": 1.0860157012939453,
      "learning_rate": 2.7195238178900684e-06,
      "loss": 0.9729,
      "step": 19933
    },
    {
      "epoch": 3.6864895000929194,
      "grad_norm": 0.9292671084403992,
      "learning_rate": 2.716140943670864e-06,
      "loss": 0.9533,
      "step": 19934
    },
    {
      "epoch": 3.6866753391562908,
      "grad_norm": 1.120851993560791,
      "learning_rate": 2.712760145808679e-06,
      "loss": 0.9393,
      "step": 19935
    },
    {
      "epoch": 3.6868611782196616,
      "grad_norm": 1.0049084424972534,
      "learning_rate": 2.70938142437569e-06,
      "loss": 0.8023,
      "step": 19936
    },
    {
      "epoch": 3.687047017283033,
      "grad_norm": 1.0591238737106323,
      "learning_rate": 2.7060047794439936e-06,
      "loss": 0.7781,
      "step": 19937
    },
    {
      "epoch": 3.687232856346404,
      "grad_norm": 1.1058365106582642,
      "learning_rate": 2.702630211085655e-06,
      "loss": 0.8144,
      "step": 19938
    },
    {
      "epoch": 3.687418695409775,
      "grad_norm": 0.868482768535614,
      "learning_rate": 2.6992577193727053e-06,
      "loss": 0.8758,
      "step": 19939
    },
    {
      "epoch": 3.687604534473146,
      "grad_norm": 0.762561023235321,
      "learning_rate": 2.6958873043771317e-06,
      "loss": 0.7921,
      "step": 19940
    },
    {
      "epoch": 3.6877903735365174,
      "grad_norm": 1.0790410041809082,
      "learning_rate": 2.692518966170865e-06,
      "loss": 0.9765,
      "step": 19941
    },
    {
      "epoch": 3.6879762125998887,
      "grad_norm": 0.9293431639671326,
      "learning_rate": 2.6891527048257813e-06,
      "loss": 0.9429,
      "step": 19942
    },
    {
      "epoch": 3.6881620516632596,
      "grad_norm": 1.1230645179748535,
      "learning_rate": 2.6857885204137346e-06,
      "loss": 0.7974,
      "step": 19943
    },
    {
      "epoch": 3.6883478907266305,
      "grad_norm": 0.8476539850234985,
      "learning_rate": 2.6824264130065555e-06,
      "loss": 0.7633,
      "step": 19944
    },
    {
      "epoch": 3.688533729790002,
      "grad_norm": 0.98152756690979,
      "learning_rate": 2.6790663826759765e-06,
      "loss": 0.9192,
      "step": 19945
    },
    {
      "epoch": 3.688719568853373,
      "grad_norm": 1.035069227218628,
      "learning_rate": 2.6757084294937173e-06,
      "loss": 0.8501,
      "step": 19946
    },
    {
      "epoch": 3.688905407916744,
      "grad_norm": 1.1366145610809326,
      "learning_rate": 2.672352553531432e-06,
      "loss": 0.7742,
      "step": 19947
    },
    {
      "epoch": 3.6890912469801154,
      "grad_norm": 0.9123889803886414,
      "learning_rate": 2.668998754860774e-06,
      "loss": 0.8288,
      "step": 19948
    },
    {
      "epoch": 3.6892770860434863,
      "grad_norm": 0.8435447812080383,
      "learning_rate": 2.6656470335533092e-06,
      "loss": 0.6986,
      "step": 19949
    },
    {
      "epoch": 3.6894629251068576,
      "grad_norm": 0.9356799721717834,
      "learning_rate": 2.662297389680568e-06,
      "loss": 0.857,
      "step": 19950
    },
    {
      "epoch": 3.6896487641702285,
      "grad_norm": 0.993013322353363,
      "learning_rate": 2.658949823314061e-06,
      "loss": 0.9632,
      "step": 19951
    },
    {
      "epoch": 3.6898346032336,
      "grad_norm": 1.127468466758728,
      "learning_rate": 2.65560433452523e-06,
      "loss": 0.7315,
      "step": 19952
    },
    {
      "epoch": 3.6900204422969707,
      "grad_norm": 0.842962384223938,
      "learning_rate": 2.6522609233854633e-06,
      "loss": 0.7186,
      "step": 19953
    },
    {
      "epoch": 3.690206281360342,
      "grad_norm": 1.0940351486206055,
      "learning_rate": 2.6489195899661368e-06,
      "loss": 1.2196,
      "step": 19954
    },
    {
      "epoch": 3.690392120423713,
      "grad_norm": 0.8041806221008301,
      "learning_rate": 2.6455803343385486e-06,
      "loss": 0.71,
      "step": 19955
    },
    {
      "epoch": 3.6905779594870842,
      "grad_norm": 0.8925176858901978,
      "learning_rate": 2.642243156573998e-06,
      "loss": 0.7767,
      "step": 19956
    },
    {
      "epoch": 3.6907637985504556,
      "grad_norm": 1.1546437740325928,
      "learning_rate": 2.638908056743694e-06,
      "loss": 0.8426,
      "step": 19957
    },
    {
      "epoch": 3.6909496376138264,
      "grad_norm": 0.921238899230957,
      "learning_rate": 2.6355750349188136e-06,
      "loss": 0.629,
      "step": 19958
    },
    {
      "epoch": 3.6911354766771973,
      "grad_norm": 0.866148054599762,
      "learning_rate": 2.632244091170499e-06,
      "loss": 0.6771,
      "step": 19959
    },
    {
      "epoch": 3.6913213157405687,
      "grad_norm": 0.864345371723175,
      "learning_rate": 2.6289152255698503e-06,
      "loss": 0.6718,
      "step": 19960
    },
    {
      "epoch": 3.69150715480394,
      "grad_norm": 0.9704990983009338,
      "learning_rate": 2.6255884381879093e-06,
      "loss": 0.9381,
      "step": 19961
    },
    {
      "epoch": 3.691692993867311,
      "grad_norm": 0.9350848197937012,
      "learning_rate": 2.6222637290956642e-06,
      "loss": 0.7384,
      "step": 19962
    },
    {
      "epoch": 3.6918788329306818,
      "grad_norm": 0.8249576091766357,
      "learning_rate": 2.618941098364114e-06,
      "loss": 0.6296,
      "step": 19963
    },
    {
      "epoch": 3.692064671994053,
      "grad_norm": 1.0667258501052856,
      "learning_rate": 2.6156205460641346e-06,
      "loss": 0.7731,
      "step": 19964
    },
    {
      "epoch": 3.6922505110574244,
      "grad_norm": 1.0333415269851685,
      "learning_rate": 2.612302072266637e-06,
      "loss": 0.7406,
      "step": 19965
    },
    {
      "epoch": 3.6924363501207953,
      "grad_norm": 1.2386226654052734,
      "learning_rate": 2.6089856770424194e-06,
      "loss": 0.7804,
      "step": 19966
    },
    {
      "epoch": 3.6926221891841666,
      "grad_norm": 0.9404177069664001,
      "learning_rate": 2.6056713604622697e-06,
      "loss": 0.6818,
      "step": 19967
    },
    {
      "epoch": 3.6928080282475375,
      "grad_norm": 0.8880127668380737,
      "learning_rate": 2.6023591225969312e-06,
      "loss": 0.6698,
      "step": 19968
    },
    {
      "epoch": 3.692993867310909,
      "grad_norm": 0.8711450099945068,
      "learning_rate": 2.5990489635171033e-06,
      "loss": 0.9517,
      "step": 19969
    },
    {
      "epoch": 3.6931797063742797,
      "grad_norm": 1.015639305114746,
      "learning_rate": 2.5957408832934183e-06,
      "loss": 0.7335,
      "step": 19970
    },
    {
      "epoch": 3.693365545437651,
      "grad_norm": 0.8139245510101318,
      "learning_rate": 2.592434881996497e-06,
      "loss": 0.6054,
      "step": 19971
    },
    {
      "epoch": 3.6935513845010224,
      "grad_norm": 0.989398181438446,
      "learning_rate": 2.5891309596968948e-06,
      "loss": 1.0038,
      "step": 19972
    },
    {
      "epoch": 3.6937372235643933,
      "grad_norm": 0.9275818467140198,
      "learning_rate": 2.5858291164651325e-06,
      "loss": 0.7925,
      "step": 19973
    },
    {
      "epoch": 3.693923062627764,
      "grad_norm": 0.8881398439407349,
      "learning_rate": 2.5825293523716654e-06,
      "loss": 1.083,
      "step": 19974
    },
    {
      "epoch": 3.6941089016911355,
      "grad_norm": 0.8966253995895386,
      "learning_rate": 2.5792316674869364e-06,
      "loss": 0.6862,
      "step": 19975
    },
    {
      "epoch": 3.694294740754507,
      "grad_norm": 1.009162425994873,
      "learning_rate": 2.5759360618813454e-06,
      "loss": 0.8876,
      "step": 19976
    },
    {
      "epoch": 3.6944805798178777,
      "grad_norm": 1.0653289556503296,
      "learning_rate": 2.5726425356252026e-06,
      "loss": 0.6619,
      "step": 19977
    },
    {
      "epoch": 3.6946664188812486,
      "grad_norm": 0.9822260141372681,
      "learning_rate": 2.5693510887888074e-06,
      "loss": 0.9509,
      "step": 19978
    },
    {
      "epoch": 3.69485225794462,
      "grad_norm": 1.2793655395507812,
      "learning_rate": 2.5660617214424145e-06,
      "loss": 1.1533,
      "step": 19979
    },
    {
      "epoch": 3.6950380970079912,
      "grad_norm": 0.8063891530036926,
      "learning_rate": 2.5627744336562343e-06,
      "loss": 0.6872,
      "step": 19980
    },
    {
      "epoch": 3.695223936071362,
      "grad_norm": 0.9094991087913513,
      "learning_rate": 2.5594892255004334e-06,
      "loss": 0.5784,
      "step": 19981
    },
    {
      "epoch": 3.6954097751347335,
      "grad_norm": 0.9532362818717957,
      "learning_rate": 2.5562060970450996e-06,
      "loss": 0.6367,
      "step": 19982
    },
    {
      "epoch": 3.6955956141981043,
      "grad_norm": 0.9102233052253723,
      "learning_rate": 2.552925048360344e-06,
      "loss": 0.6689,
      "step": 19983
    },
    {
      "epoch": 3.6957814532614757,
      "grad_norm": 0.8929124474525452,
      "learning_rate": 2.549646079516177e-06,
      "loss": 0.8239,
      "step": 19984
    },
    {
      "epoch": 3.6959672923248466,
      "grad_norm": 0.9813836216926575,
      "learning_rate": 2.546369190582576e-06,
      "loss": 0.7966,
      "step": 19985
    },
    {
      "epoch": 3.696153131388218,
      "grad_norm": 0.991256833076477,
      "learning_rate": 2.543094381629496e-06,
      "loss": 0.7203,
      "step": 19986
    },
    {
      "epoch": 3.6963389704515888,
      "grad_norm": 0.861219048500061,
      "learning_rate": 2.539821652726804e-06,
      "loss": 0.5936,
      "step": 19987
    },
    {
      "epoch": 3.69652480951496,
      "grad_norm": 0.8657010197639465,
      "learning_rate": 2.5365510039443873e-06,
      "loss": 0.7177,
      "step": 19988
    },
    {
      "epoch": 3.696710648578331,
      "grad_norm": 0.9564704895019531,
      "learning_rate": 2.5332824353520357e-06,
      "loss": 0.9188,
      "step": 19989
    },
    {
      "epoch": 3.6968964876417023,
      "grad_norm": 0.907672643661499,
      "learning_rate": 2.530015947019504e-06,
      "loss": 0.8229,
      "step": 19990
    },
    {
      "epoch": 3.6970823267050736,
      "grad_norm": 0.9323121309280396,
      "learning_rate": 2.5267515390165254e-06,
      "loss": 0.501,
      "step": 19991
    },
    {
      "epoch": 3.6972681657684445,
      "grad_norm": 0.9874093532562256,
      "learning_rate": 2.523489211412755e-06,
      "loss": 0.9714,
      "step": 19992
    },
    {
      "epoch": 3.6974540048318154,
      "grad_norm": 0.9057678580284119,
      "learning_rate": 2.5202289642778375e-06,
      "loss": 0.8565,
      "step": 19993
    },
    {
      "epoch": 3.6976398438951867,
      "grad_norm": 0.8049215078353882,
      "learning_rate": 2.5169707976813504e-06,
      "loss": 0.6827,
      "step": 19994
    },
    {
      "epoch": 3.697825682958558,
      "grad_norm": 1.047194004058838,
      "learning_rate": 2.513714711692827e-06,
      "loss": 1.1513,
      "step": 19995
    },
    {
      "epoch": 3.698011522021929,
      "grad_norm": 0.8550397157669067,
      "learning_rate": 2.510460706381779e-06,
      "loss": 0.9733,
      "step": 19996
    },
    {
      "epoch": 3.6981973610853003,
      "grad_norm": 0.9241202473640442,
      "learning_rate": 2.5072087818176382e-06,
      "loss": 0.8653,
      "step": 19997
    },
    {
      "epoch": 3.698383200148671,
      "grad_norm": 1.0444940328598022,
      "learning_rate": 2.503958938069839e-06,
      "loss": 0.7391,
      "step": 19998
    },
    {
      "epoch": 3.6985690392120425,
      "grad_norm": 0.9220775961875916,
      "learning_rate": 2.500711175207704e-06,
      "loss": 0.8743,
      "step": 19999
    },
    {
      "epoch": 3.6987548782754134,
      "grad_norm": 0.9342935681343079,
      "learning_rate": 2.497465493300588e-06,
      "loss": 0.7478,
      "step": 20000
    },
    {
      "epoch": 3.6989407173387847,
      "grad_norm": 0.8913558721542358,
      "learning_rate": 2.4942218924177473e-06,
      "loss": 0.8953,
      "step": 20001
    },
    {
      "epoch": 3.6991265564021556,
      "grad_norm": 0.9742237329483032,
      "learning_rate": 2.4909803726284043e-06,
      "loss": 0.8836,
      "step": 20002
    },
    {
      "epoch": 3.699312395465527,
      "grad_norm": 0.8685630559921265,
      "learning_rate": 2.48774093400177e-06,
      "loss": 0.559,
      "step": 20003
    },
    {
      "epoch": 3.699498234528898,
      "grad_norm": 1.0585001707077026,
      "learning_rate": 2.484503576606967e-06,
      "loss": 0.8954,
      "step": 20004
    },
    {
      "epoch": 3.699684073592269,
      "grad_norm": 0.8607193827629089,
      "learning_rate": 2.4812683005130843e-06,
      "loss": 0.7518,
      "step": 20005
    },
    {
      "epoch": 3.6998699126556405,
      "grad_norm": 0.8997200727462769,
      "learning_rate": 2.478035105789178e-06,
      "loss": 0.5457,
      "step": 20006
    },
    {
      "epoch": 3.7000557517190114,
      "grad_norm": 0.9212594032287598,
      "learning_rate": 2.474803992504249e-06,
      "loss": 0.8139,
      "step": 20007
    },
    {
      "epoch": 3.7002415907823822,
      "grad_norm": 0.991801917552948,
      "learning_rate": 2.4715749607272853e-06,
      "loss": 0.7639,
      "step": 20008
    },
    {
      "epoch": 3.7004274298457536,
      "grad_norm": 0.9359737038612366,
      "learning_rate": 2.4683480105271883e-06,
      "loss": 0.8758,
      "step": 20009
    },
    {
      "epoch": 3.700613268909125,
      "grad_norm": 0.8309842348098755,
      "learning_rate": 2.465123141972825e-06,
      "loss": 0.7622,
      "step": 20010
    },
    {
      "epoch": 3.700799107972496,
      "grad_norm": 0.5806534886360168,
      "learning_rate": 2.461900355133029e-06,
      "loss": 0.323,
      "step": 20011
    },
    {
      "epoch": 3.7009849470358667,
      "grad_norm": 1.1498020887374878,
      "learning_rate": 2.4586796500766005e-06,
      "loss": 0.8802,
      "step": 20012
    },
    {
      "epoch": 3.701170786099238,
      "grad_norm": 0.8998548984527588,
      "learning_rate": 2.4554610268722634e-06,
      "loss": 0.8204,
      "step": 20013
    },
    {
      "epoch": 3.7013566251626093,
      "grad_norm": 0.9816725850105286,
      "learning_rate": 2.452244485588706e-06,
      "loss": 0.765,
      "step": 20014
    },
    {
      "epoch": 3.70154246422598,
      "grad_norm": 1.1572517156600952,
      "learning_rate": 2.4490300262946074e-06,
      "loss": 0.9171,
      "step": 20015
    },
    {
      "epoch": 3.7017283032893515,
      "grad_norm": 0.889315128326416,
      "learning_rate": 2.445817649058546e-06,
      "loss": 0.8954,
      "step": 20016
    },
    {
      "epoch": 3.7019141423527224,
      "grad_norm": 1.7878528833389282,
      "learning_rate": 2.4426073539491e-06,
      "loss": 1.1884,
      "step": 20017
    },
    {
      "epoch": 3.7020999814160938,
      "grad_norm": 1.295170545578003,
      "learning_rate": 2.4393991410347928e-06,
      "loss": 0.7628,
      "step": 20018
    },
    {
      "epoch": 3.7022858204794646,
      "grad_norm": 1.06532621383667,
      "learning_rate": 2.4361930103840802e-06,
      "loss": 0.9315,
      "step": 20019
    },
    {
      "epoch": 3.702471659542836,
      "grad_norm": 0.8872401118278503,
      "learning_rate": 2.4329889620654077e-06,
      "loss": 0.6206,
      "step": 20020
    },
    {
      "epoch": 3.7026574986062073,
      "grad_norm": 1.11594820022583,
      "learning_rate": 2.429786996147154e-06,
      "loss": 1.1982,
      "step": 20021
    },
    {
      "epoch": 3.702843337669578,
      "grad_norm": 1.0438435077667236,
      "learning_rate": 2.4265871126976536e-06,
      "loss": 0.9589,
      "step": 20022
    },
    {
      "epoch": 3.703029176732949,
      "grad_norm": 0.9034283757209778,
      "learning_rate": 2.423389311785207e-06,
      "loss": 0.7611,
      "step": 20023
    },
    {
      "epoch": 3.7032150157963204,
      "grad_norm": 1.0138309001922607,
      "learning_rate": 2.4201935934780816e-06,
      "loss": 0.6974,
      "step": 20024
    },
    {
      "epoch": 3.7034008548596917,
      "grad_norm": 1.1490155458450317,
      "learning_rate": 2.4169999578444678e-06,
      "loss": 0.8325,
      "step": 20025
    },
    {
      "epoch": 3.7035866939230626,
      "grad_norm": 0.8348366618156433,
      "learning_rate": 2.413808404952511e-06,
      "loss": 0.7004,
      "step": 20026
    },
    {
      "epoch": 3.7037725329864335,
      "grad_norm": 1.0912022590637207,
      "learning_rate": 2.4106189348703565e-06,
      "loss": 0.6626,
      "step": 20027
    },
    {
      "epoch": 3.703958372049805,
      "grad_norm": 0.8689257502555847,
      "learning_rate": 2.4074315476660836e-06,
      "loss": 0.6285,
      "step": 20028
    },
    {
      "epoch": 3.704144211113176,
      "grad_norm": 1.1265016794204712,
      "learning_rate": 2.404246243407704e-06,
      "loss": 0.6572,
      "step": 20029
    },
    {
      "epoch": 3.704330050176547,
      "grad_norm": 1.0007442235946655,
      "learning_rate": 2.401063022163197e-06,
      "loss": 0.7885,
      "step": 20030
    },
    {
      "epoch": 3.7045158892399184,
      "grad_norm": 0.8556223511695862,
      "learning_rate": 2.39788188400053e-06,
      "loss": 0.6905,
      "step": 20031
    },
    {
      "epoch": 3.7047017283032893,
      "grad_norm": 0.7509297728538513,
      "learning_rate": 2.3947028289875716e-06,
      "loss": 0.6629,
      "step": 20032
    },
    {
      "epoch": 3.7048875673666606,
      "grad_norm": 0.8223177194595337,
      "learning_rate": 2.3915258571921896e-06,
      "loss": 0.5922,
      "step": 20033
    },
    {
      "epoch": 3.7050734064300315,
      "grad_norm": 0.8994002938270569,
      "learning_rate": 2.388350968682185e-06,
      "loss": 0.8186,
      "step": 20034
    },
    {
      "epoch": 3.705259245493403,
      "grad_norm": 0.9853032827377319,
      "learning_rate": 2.3851781635253147e-06,
      "loss": 0.8247,
      "step": 20035
    },
    {
      "epoch": 3.7054450845567737,
      "grad_norm": 0.8897237181663513,
      "learning_rate": 2.3820074417893136e-06,
      "loss": 0.7914,
      "step": 20036
    },
    {
      "epoch": 3.705630923620145,
      "grad_norm": 0.9126240015029907,
      "learning_rate": 2.3788388035418273e-06,
      "loss": 0.6382,
      "step": 20037
    },
    {
      "epoch": 3.705816762683516,
      "grad_norm": 1.0882678031921387,
      "learning_rate": 2.3756722488505135e-06,
      "loss": 0.8311,
      "step": 20038
    },
    {
      "epoch": 3.7060026017468872,
      "grad_norm": 1.1550171375274658,
      "learning_rate": 2.3725077777829395e-06,
      "loss": 0.8201,
      "step": 20039
    },
    {
      "epoch": 3.7061884408102586,
      "grad_norm": 1.1292064189910889,
      "learning_rate": 2.369345390406663e-06,
      "loss": 0.8463,
      "step": 20040
    },
    {
      "epoch": 3.7063742798736294,
      "grad_norm": 0.9061990976333618,
      "learning_rate": 2.3661850867891634e-06,
      "loss": 0.6411,
      "step": 20041
    },
    {
      "epoch": 3.7065601189370003,
      "grad_norm": 1.173477053642273,
      "learning_rate": 2.363026866997886e-06,
      "loss": 0.7937,
      "step": 20042
    },
    {
      "epoch": 3.7067459580003717,
      "grad_norm": 1.0434966087341309,
      "learning_rate": 2.3598707311002555e-06,
      "loss": 0.8296,
      "step": 20043
    },
    {
      "epoch": 3.706931797063743,
      "grad_norm": 0.9530367851257324,
      "learning_rate": 2.3567166791636284e-06,
      "loss": 0.9408,
      "step": 20044
    },
    {
      "epoch": 3.707117636127114,
      "grad_norm": 0.7768343091011047,
      "learning_rate": 2.3535647112553294e-06,
      "loss": 0.8813,
      "step": 20045
    },
    {
      "epoch": 3.707303475190485,
      "grad_norm": 1.1541277170181274,
      "learning_rate": 2.350414827442604e-06,
      "loss": 0.7388,
      "step": 20046
    },
    {
      "epoch": 3.707489314253856,
      "grad_norm": 0.9838178753852844,
      "learning_rate": 2.347267027792721e-06,
      "loss": 0.5767,
      "step": 20047
    },
    {
      "epoch": 3.7076751533172274,
      "grad_norm": 0.8992997407913208,
      "learning_rate": 2.3441213123728266e-06,
      "loss": 0.7926,
      "step": 20048
    },
    {
      "epoch": 3.7078609923805983,
      "grad_norm": 1.0437636375427246,
      "learning_rate": 2.3409776812500894e-06,
      "loss": 1.0169,
      "step": 20049
    },
    {
      "epoch": 3.7080468314439696,
      "grad_norm": 0.8232665657997131,
      "learning_rate": 2.337836134491589e-06,
      "loss": 0.7918,
      "step": 20050
    },
    {
      "epoch": 3.7082326705073405,
      "grad_norm": 1.0069401264190674,
      "learning_rate": 2.3346966721643825e-06,
      "loss": 0.9552,
      "step": 20051
    },
    {
      "epoch": 3.708418509570712,
      "grad_norm": 0.9937763810157776,
      "learning_rate": 2.3315592943354837e-06,
      "loss": 0.8517,
      "step": 20052
    },
    {
      "epoch": 3.7086043486340827,
      "grad_norm": 0.9887884259223938,
      "learning_rate": 2.3284240010718383e-06,
      "loss": 0.8548,
      "step": 20053
    },
    {
      "epoch": 3.708790187697454,
      "grad_norm": 1.0063434839248657,
      "learning_rate": 2.325290792440371e-06,
      "loss": 0.7341,
      "step": 20054
    },
    {
      "epoch": 3.7089760267608254,
      "grad_norm": 0.9228911995887756,
      "learning_rate": 2.3221596685079616e-06,
      "loss": 0.936,
      "step": 20055
    },
    {
      "epoch": 3.7091618658241963,
      "grad_norm": 0.9210600852966309,
      "learning_rate": 2.319030629341423e-06,
      "loss": 0.7458,
      "step": 20056
    },
    {
      "epoch": 3.709347704887567,
      "grad_norm": 0.8243858218193054,
      "learning_rate": 2.315903675007558e-06,
      "loss": 0.9547,
      "step": 20057
    },
    {
      "epoch": 3.7095335439509385,
      "grad_norm": 1.0575931072235107,
      "learning_rate": 2.3127788055730794e-06,
      "loss": 0.9452,
      "step": 20058
    },
    {
      "epoch": 3.70971938301431,
      "grad_norm": 1.033123254776001,
      "learning_rate": 2.309656021104711e-06,
      "loss": 0.7252,
      "step": 20059
    },
    {
      "epoch": 3.7099052220776807,
      "grad_norm": 1.0101604461669922,
      "learning_rate": 2.30653532166909e-06,
      "loss": 0.6904,
      "step": 20060
    },
    {
      "epoch": 3.7100910611410516,
      "grad_norm": 0.787567675113678,
      "learning_rate": 2.3034167073328284e-06,
      "loss": 0.5783,
      "step": 20061
    },
    {
      "epoch": 3.710276900204423,
      "grad_norm": 1.0227032899856567,
      "learning_rate": 2.3003001781624734e-06,
      "loss": 0.7656,
      "step": 20062
    },
    {
      "epoch": 3.7104627392677942,
      "grad_norm": 0.8273500204086304,
      "learning_rate": 2.2971857342245606e-06,
      "loss": 0.7409,
      "step": 20063
    },
    {
      "epoch": 3.710648578331165,
      "grad_norm": 1.0066863298416138,
      "learning_rate": 2.2940733755855592e-06,
      "loss": 0.93,
      "step": 20064
    },
    {
      "epoch": 3.7108344173945365,
      "grad_norm": 0.9400318264961243,
      "learning_rate": 2.2909631023118826e-06,
      "loss": 0.9107,
      "step": 20065
    },
    {
      "epoch": 3.7110202564579073,
      "grad_norm": 1.2339752912521362,
      "learning_rate": 2.2878549144699223e-06,
      "loss": 0.8317,
      "step": 20066
    },
    {
      "epoch": 3.7112060955212787,
      "grad_norm": 0.9648247957229614,
      "learning_rate": 2.2847488121260253e-06,
      "loss": 0.8801,
      "step": 20067
    },
    {
      "epoch": 3.7113919345846496,
      "grad_norm": 0.9962890148162842,
      "learning_rate": 2.2816447953464715e-06,
      "loss": 0.7403,
      "step": 20068
    },
    {
      "epoch": 3.711577773648021,
      "grad_norm": 0.9467740654945374,
      "learning_rate": 2.278542864197519e-06,
      "loss": 0.8098,
      "step": 20069
    },
    {
      "epoch": 3.7117636127113918,
      "grad_norm": 0.9877055287361145,
      "learning_rate": 2.27544301874536e-06,
      "loss": 0.6955,
      "step": 20070
    },
    {
      "epoch": 3.711949451774763,
      "grad_norm": 1.4038006067276,
      "learning_rate": 2.2723452590561856e-06,
      "loss": 0.8247,
      "step": 20071
    },
    {
      "epoch": 3.712135290838134,
      "grad_norm": 1.0121270418167114,
      "learning_rate": 2.2692495851960983e-06,
      "loss": 0.95,
      "step": 20072
    },
    {
      "epoch": 3.7123211299015053,
      "grad_norm": 0.9226126074790955,
      "learning_rate": 2.2661559972311565e-06,
      "loss": 0.8994,
      "step": 20073
    },
    {
      "epoch": 3.7125069689648766,
      "grad_norm": 0.7741632461547852,
      "learning_rate": 2.263064495227396e-06,
      "loss": 0.6216,
      "step": 20074
    },
    {
      "epoch": 3.7126928080282475,
      "grad_norm": 0.8833262324333191,
      "learning_rate": 2.259975079250798e-06,
      "loss": 0.6881,
      "step": 20075
    },
    {
      "epoch": 3.7128786470916184,
      "grad_norm": 1.0069153308868408,
      "learning_rate": 2.256887749367309e-06,
      "loss": 0.751,
      "step": 20076
    },
    {
      "epoch": 3.7130644861549897,
      "grad_norm": 0.9459255337715149,
      "learning_rate": 2.2538025056428214e-06,
      "loss": 0.8701,
      "step": 20077
    },
    {
      "epoch": 3.713250325218361,
      "grad_norm": 0.8940929770469666,
      "learning_rate": 2.250719348143171e-06,
      "loss": 0.9173,
      "step": 20078
    },
    {
      "epoch": 3.713436164281732,
      "grad_norm": 0.834430992603302,
      "learning_rate": 2.247638276934172e-06,
      "loss": 0.6886,
      "step": 20079
    },
    {
      "epoch": 3.7136220033451033,
      "grad_norm": 0.9722082018852234,
      "learning_rate": 2.2445592920815826e-06,
      "loss": 0.9061,
      "step": 20080
    },
    {
      "epoch": 3.713807842408474,
      "grad_norm": 0.9196001887321472,
      "learning_rate": 2.2414823936511287e-06,
      "loss": 0.7362,
      "step": 20081
    },
    {
      "epoch": 3.7139936814718455,
      "grad_norm": 1.0167895555496216,
      "learning_rate": 2.238407581708468e-06,
      "loss": 0.7281,
      "step": 20082
    },
    {
      "epoch": 3.7141795205352164,
      "grad_norm": 1.020185112953186,
      "learning_rate": 2.2353348563192266e-06,
      "loss": 0.7961,
      "step": 20083
    },
    {
      "epoch": 3.7143653595985877,
      "grad_norm": 1.0760143995285034,
      "learning_rate": 2.2322642175490073e-06,
      "loss": 0.9359,
      "step": 20084
    },
    {
      "epoch": 3.7145511986619586,
      "grad_norm": 1.1122297048568726,
      "learning_rate": 2.229195665463324e-06,
      "loss": 0.7231,
      "step": 20085
    },
    {
      "epoch": 3.71473703772533,
      "grad_norm": 0.9687815308570862,
      "learning_rate": 2.226129200127669e-06,
      "loss": 1.0986,
      "step": 20086
    },
    {
      "epoch": 3.714922876788701,
      "grad_norm": 1.0281693935394287,
      "learning_rate": 2.2230648216075125e-06,
      "loss": 0.7868,
      "step": 20087
    },
    {
      "epoch": 3.715108715852072,
      "grad_norm": 0.90626060962677,
      "learning_rate": 2.220002529968246e-06,
      "loss": 0.9467,
      "step": 20088
    },
    {
      "epoch": 3.7152945549154435,
      "grad_norm": 1.0133919715881348,
      "learning_rate": 2.2169423252752175e-06,
      "loss": 0.5898,
      "step": 20089
    },
    {
      "epoch": 3.7154803939788144,
      "grad_norm": 0.8916006088256836,
      "learning_rate": 2.213884207593742e-06,
      "loss": 0.8046,
      "step": 20090
    },
    {
      "epoch": 3.7156662330421852,
      "grad_norm": 0.7658432126045227,
      "learning_rate": 2.2108281769891217e-06,
      "loss": 0.5187,
      "step": 20091
    },
    {
      "epoch": 3.7158520721055566,
      "grad_norm": 0.937510073184967,
      "learning_rate": 2.2077742335265604e-06,
      "loss": 0.8989,
      "step": 20092
    },
    {
      "epoch": 3.716037911168928,
      "grad_norm": 0.9862101674079895,
      "learning_rate": 2.20472237727124e-06,
      "loss": 0.7326,
      "step": 20093
    },
    {
      "epoch": 3.716223750232299,
      "grad_norm": 0.9088574647903442,
      "learning_rate": 2.2016726082882743e-06,
      "loss": 0.8046,
      "step": 20094
    },
    {
      "epoch": 3.71640958929567,
      "grad_norm": 0.9740971326828003,
      "learning_rate": 2.1986249266428006e-06,
      "loss": 0.7832,
      "step": 20095
    },
    {
      "epoch": 3.716595428359041,
      "grad_norm": 0.8323392271995544,
      "learning_rate": 2.1955793323998442e-06,
      "loss": 0.6632,
      "step": 20096
    },
    {
      "epoch": 3.7167812674224123,
      "grad_norm": 0.8543899059295654,
      "learning_rate": 2.192535825624409e-06,
      "loss": 0.8198,
      "step": 20097
    },
    {
      "epoch": 3.716967106485783,
      "grad_norm": 0.9209011793136597,
      "learning_rate": 2.1894944063814314e-06,
      "loss": 0.7359,
      "step": 20098
    },
    {
      "epoch": 3.7171529455491545,
      "grad_norm": 0.9761630296707153,
      "learning_rate": 2.1864550747358713e-06,
      "loss": 0.7975,
      "step": 20099
    },
    {
      "epoch": 3.7173387846125254,
      "grad_norm": 0.9385524392127991,
      "learning_rate": 2.1834178307525543e-06,
      "loss": 0.876,
      "step": 20100
    },
    {
      "epoch": 3.7175246236758968,
      "grad_norm": 1.1285103559494019,
      "learning_rate": 2.1803826744963286e-06,
      "loss": 0.8112,
      "step": 20101
    },
    {
      "epoch": 3.7177104627392676,
      "grad_norm": 0.8879675269126892,
      "learning_rate": 2.1773496060319755e-06,
      "loss": 0.8473,
      "step": 20102
    },
    {
      "epoch": 3.717896301802639,
      "grad_norm": 0.9821565747261047,
      "learning_rate": 2.174318625424221e-06,
      "loss": 0.7657,
      "step": 20103
    },
    {
      "epoch": 3.7180821408660103,
      "grad_norm": 0.8077647089958191,
      "learning_rate": 2.171289732737758e-06,
      "loss": 0.7054,
      "step": 20104
    },
    {
      "epoch": 3.718267979929381,
      "grad_norm": 0.9461015462875366,
      "learning_rate": 2.1682629280372456e-06,
      "loss": 0.8369,
      "step": 20105
    },
    {
      "epoch": 3.718453818992752,
      "grad_norm": 0.8106582760810852,
      "learning_rate": 2.165238211387255e-06,
      "loss": 0.5342,
      "step": 20106
    },
    {
      "epoch": 3.7186396580561234,
      "grad_norm": 0.96129310131073,
      "learning_rate": 2.1622155828523784e-06,
      "loss": 0.9539,
      "step": 20107
    },
    {
      "epoch": 3.7188254971194947,
      "grad_norm": 0.8640586733818054,
      "learning_rate": 2.15919504249712e-06,
      "loss": 0.754,
      "step": 20108
    },
    {
      "epoch": 3.7190113361828656,
      "grad_norm": 1.0741902589797974,
      "learning_rate": 2.1561765903859275e-06,
      "loss": 0.5654,
      "step": 20109
    },
    {
      "epoch": 3.7191971752462365,
      "grad_norm": 0.9164366126060486,
      "learning_rate": 2.1531602265832397e-06,
      "loss": 0.7342,
      "step": 20110
    },
    {
      "epoch": 3.719383014309608,
      "grad_norm": 1.1637158393859863,
      "learning_rate": 2.1501459511534374e-06,
      "loss": 0.9248,
      "step": 20111
    },
    {
      "epoch": 3.719568853372979,
      "grad_norm": 0.9291113018989563,
      "learning_rate": 2.147133764160858e-06,
      "loss": 0.8318,
      "step": 20112
    },
    {
      "epoch": 3.71975469243635,
      "grad_norm": 0.956951916217804,
      "learning_rate": 2.144123665669784e-06,
      "loss": 0.8349,
      "step": 20113
    },
    {
      "epoch": 3.7199405314997214,
      "grad_norm": 0.8551384806632996,
      "learning_rate": 2.1411156557444525e-06,
      "loss": 0.9349,
      "step": 20114
    },
    {
      "epoch": 3.7201263705630923,
      "grad_norm": 0.9560601115226746,
      "learning_rate": 2.1381097344490895e-06,
      "loss": 0.8236,
      "step": 20115
    },
    {
      "epoch": 3.7203122096264636,
      "grad_norm": 0.7943173050880432,
      "learning_rate": 2.1351059018478336e-06,
      "loss": 0.7864,
      "step": 20116
    },
    {
      "epoch": 3.7204980486898345,
      "grad_norm": 0.8827927708625793,
      "learning_rate": 2.1321041580047996e-06,
      "loss": 0.6593,
      "step": 20117
    },
    {
      "epoch": 3.720683887753206,
      "grad_norm": 0.9819035530090332,
      "learning_rate": 2.1291045029840474e-06,
      "loss": 1.1072,
      "step": 20118
    },
    {
      "epoch": 3.7208697268165767,
      "grad_norm": 0.7115758061408997,
      "learning_rate": 2.1261069368496033e-06,
      "loss": 0.5483,
      "step": 20119
    },
    {
      "epoch": 3.721055565879948,
      "grad_norm": 1.027331829071045,
      "learning_rate": 2.123111459665461e-06,
      "loss": 1.0232,
      "step": 20120
    },
    {
      "epoch": 3.721241404943319,
      "grad_norm": 0.8737844228744507,
      "learning_rate": 2.120118071495536e-06,
      "loss": 0.6384,
      "step": 20121
    },
    {
      "epoch": 3.7214272440066902,
      "grad_norm": 0.831642210483551,
      "learning_rate": 2.1171267724037105e-06,
      "loss": 0.6566,
      "step": 20122
    },
    {
      "epoch": 3.7216130830700616,
      "grad_norm": 0.8471463918685913,
      "learning_rate": 2.114137562453855e-06,
      "loss": 0.8067,
      "step": 20123
    },
    {
      "epoch": 3.7217989221334324,
      "grad_norm": 1.0067334175109863,
      "learning_rate": 2.1111504417097525e-06,
      "loss": 0.8335,
      "step": 20124
    },
    {
      "epoch": 3.7219847611968033,
      "grad_norm": 1.0657848119735718,
      "learning_rate": 2.1081654102351635e-06,
      "loss": 0.892,
      "step": 20125
    },
    {
      "epoch": 3.7221706002601747,
      "grad_norm": 0.9801212549209595,
      "learning_rate": 2.1051824680937802e-06,
      "loss": 0.7435,
      "step": 20126
    },
    {
      "epoch": 3.722356439323546,
      "grad_norm": 0.9074841141700745,
      "learning_rate": 2.1022016153492973e-06,
      "loss": 0.6944,
      "step": 20127
    },
    {
      "epoch": 3.722542278386917,
      "grad_norm": 0.9464606642723083,
      "learning_rate": 2.099222852065319e-06,
      "loss": 0.8186,
      "step": 20128
    },
    {
      "epoch": 3.722728117450288,
      "grad_norm": 0.9544642567634583,
      "learning_rate": 2.0962461783054167e-06,
      "loss": 0.8442,
      "step": 20129
    },
    {
      "epoch": 3.722913956513659,
      "grad_norm": 1.5007946491241455,
      "learning_rate": 2.093271594133128e-06,
      "loss": 1.1873,
      "step": 20130
    },
    {
      "epoch": 3.7230997955770304,
      "grad_norm": 1.1086900234222412,
      "learning_rate": 2.090299099611959e-06,
      "loss": 0.9498,
      "step": 20131
    },
    {
      "epoch": 3.7232856346404013,
      "grad_norm": 0.9293392896652222,
      "learning_rate": 2.087328694805324e-06,
      "loss": 0.8043,
      "step": 20132
    },
    {
      "epoch": 3.7234714737037726,
      "grad_norm": 0.9821919202804565,
      "learning_rate": 2.0843603797766287e-06,
      "loss": 0.746,
      "step": 20133
    },
    {
      "epoch": 3.7236573127671435,
      "grad_norm": 0.9458260536193848,
      "learning_rate": 2.0813941545892337e-06,
      "loss": 0.6984,
      "step": 20134
    },
    {
      "epoch": 3.723843151830515,
      "grad_norm": 1.0252138376235962,
      "learning_rate": 2.0784300193064544e-06,
      "loss": 0.8409,
      "step": 20135
    },
    {
      "epoch": 3.7240289908938857,
      "grad_norm": 0.932839035987854,
      "learning_rate": 2.0754679739915406e-06,
      "loss": 0.8454,
      "step": 20136
    },
    {
      "epoch": 3.724214829957257,
      "grad_norm": 0.7695898413658142,
      "learning_rate": 2.0725080187077305e-06,
      "loss": 0.6664,
      "step": 20137
    },
    {
      "epoch": 3.7244006690206284,
      "grad_norm": 1.0405521392822266,
      "learning_rate": 2.0695501535181626e-06,
      "loss": 0.8003,
      "step": 20138
    },
    {
      "epoch": 3.7245865080839993,
      "grad_norm": 1.0741987228393555,
      "learning_rate": 2.066594378486009e-06,
      "loss": 0.8593,
      "step": 20139
    },
    {
      "epoch": 3.72477234714737,
      "grad_norm": 1.0499932765960693,
      "learning_rate": 2.063640693674329e-06,
      "loss": 0.786,
      "step": 20140
    },
    {
      "epoch": 3.7249581862107415,
      "grad_norm": 0.9765279293060303,
      "learning_rate": 2.0606890991461737e-06,
      "loss": 0.8632,
      "step": 20141
    },
    {
      "epoch": 3.725144025274113,
      "grad_norm": 1.08168363571167,
      "learning_rate": 2.057739594964536e-06,
      "loss": 0.9217,
      "step": 20142
    },
    {
      "epoch": 3.7253298643374837,
      "grad_norm": 0.9003187417984009,
      "learning_rate": 2.0547921811923774e-06,
      "loss": 0.7954,
      "step": 20143
    },
    {
      "epoch": 3.7255157034008546,
      "grad_norm": 0.9419955611228943,
      "learning_rate": 2.051846857892603e-06,
      "loss": 0.5936,
      "step": 20144
    },
    {
      "epoch": 3.725701542464226,
      "grad_norm": 0.9181314706802368,
      "learning_rate": 2.0489036251280624e-06,
      "loss": 0.6663,
      "step": 20145
    },
    {
      "epoch": 3.7258873815275972,
      "grad_norm": 0.9209869503974915,
      "learning_rate": 2.045962482961583e-06,
      "loss": 0.6843,
      "step": 20146
    },
    {
      "epoch": 3.726073220590968,
      "grad_norm": 1.0010343790054321,
      "learning_rate": 2.0430234314559482e-06,
      "loss": 0.9073,
      "step": 20147
    },
    {
      "epoch": 3.7262590596543395,
      "grad_norm": 0.9539319276809692,
      "learning_rate": 2.0400864706738744e-06,
      "loss": 0.8202,
      "step": 20148
    },
    {
      "epoch": 3.7264448987177103,
      "grad_norm": 1.0110539197921753,
      "learning_rate": 2.0371516006780443e-06,
      "loss": 1.0652,
      "step": 20149
    },
    {
      "epoch": 3.7266307377810817,
      "grad_norm": 1.076846718788147,
      "learning_rate": 2.0342188215311086e-06,
      "loss": 0.7156,
      "step": 20150
    },
    {
      "epoch": 3.7268165768444526,
      "grad_norm": 0.9414911270141602,
      "learning_rate": 2.03128813329565e-06,
      "loss": 0.8184,
      "step": 20151
    },
    {
      "epoch": 3.727002415907824,
      "grad_norm": 0.985910177230835,
      "learning_rate": 2.02835953603423e-06,
      "loss": 0.7397,
      "step": 20152
    },
    {
      "epoch": 3.727188254971195,
      "grad_norm": 0.8706223368644714,
      "learning_rate": 2.0254330298093425e-06,
      "loss": 0.6884,
      "step": 20153
    },
    {
      "epoch": 3.727374094034566,
      "grad_norm": 0.9221065640449524,
      "learning_rate": 2.0225086146834495e-06,
      "loss": 0.8807,
      "step": 20154
    },
    {
      "epoch": 3.727559933097937,
      "grad_norm": 0.9632253646850586,
      "learning_rate": 2.019586290718989e-06,
      "loss": 0.8141,
      "step": 20155
    },
    {
      "epoch": 3.7277457721613083,
      "grad_norm": 0.9486887454986572,
      "learning_rate": 2.0166660579783113e-06,
      "loss": 0.8536,
      "step": 20156
    },
    {
      "epoch": 3.7279316112246796,
      "grad_norm": 1.1962312459945679,
      "learning_rate": 2.013747916523756e-06,
      "loss": 0.8527,
      "step": 20157
    },
    {
      "epoch": 3.7281174502880505,
      "grad_norm": 0.8917415738105774,
      "learning_rate": 2.010831866417595e-06,
      "loss": 0.9358,
      "step": 20158
    },
    {
      "epoch": 3.7283032893514214,
      "grad_norm": 0.8610925078392029,
      "learning_rate": 2.0079179077220788e-06,
      "loss": 0.8847,
      "step": 20159
    },
    {
      "epoch": 3.7284891284147927,
      "grad_norm": 0.9495860934257507,
      "learning_rate": 2.005006040499391e-06,
      "loss": 0.8034,
      "step": 20160
    },
    {
      "epoch": 3.728674967478164,
      "grad_norm": 1.1175827980041504,
      "learning_rate": 2.0020962648116704e-06,
      "loss": 0.766,
      "step": 20161
    },
    {
      "epoch": 3.728860806541535,
      "grad_norm": 1.092516303062439,
      "learning_rate": 1.999188580721045e-06,
      "loss": 0.88,
      "step": 20162
    },
    {
      "epoch": 3.7290466456049063,
      "grad_norm": 0.9286985993385315,
      "learning_rate": 1.9962829882895662e-06,
      "loss": 0.8497,
      "step": 20163
    },
    {
      "epoch": 3.729232484668277,
      "grad_norm": 1.028407335281372,
      "learning_rate": 1.993379487579239e-06,
      "loss": 0.8222,
      "step": 20164
    },
    {
      "epoch": 3.7294183237316485,
      "grad_norm": 0.7155711650848389,
      "learning_rate": 1.990478078652047e-06,
      "loss": 0.5466,
      "step": 20165
    },
    {
      "epoch": 3.7296041627950194,
      "grad_norm": 1.143075704574585,
      "learning_rate": 1.987578761569897e-06,
      "loss": 0.9321,
      "step": 20166
    },
    {
      "epoch": 3.7297900018583907,
      "grad_norm": 0.830189049243927,
      "learning_rate": 1.9846815363946947e-06,
      "loss": 0.7996,
      "step": 20167
    },
    {
      "epoch": 3.7299758409217616,
      "grad_norm": 0.958895206451416,
      "learning_rate": 1.981786403188268e-06,
      "loss": 1.0407,
      "step": 20168
    },
    {
      "epoch": 3.730161679985133,
      "grad_norm": 0.8601774573326111,
      "learning_rate": 1.978893362012402e-06,
      "loss": 0.9049,
      "step": 20169
    },
    {
      "epoch": 3.730347519048504,
      "grad_norm": 0.9910159707069397,
      "learning_rate": 1.9760024129288344e-06,
      "loss": 0.8518,
      "step": 20170
    },
    {
      "epoch": 3.730533358111875,
      "grad_norm": 0.9023825526237488,
      "learning_rate": 1.973113555999284e-06,
      "loss": 0.77,
      "step": 20171
    },
    {
      "epoch": 3.7307191971752465,
      "grad_norm": 1.6084246635437012,
      "learning_rate": 1.970226791285412e-06,
      "loss": 1.2125,
      "step": 20172
    },
    {
      "epoch": 3.7309050362386174,
      "grad_norm": 0.9333736896514893,
      "learning_rate": 1.9673421188488137e-06,
      "loss": 0.5978,
      "step": 20173
    },
    {
      "epoch": 3.7310908753019882,
      "grad_norm": 1.0360727310180664,
      "learning_rate": 1.9644595387510624e-06,
      "loss": 0.8183,
      "step": 20174
    },
    {
      "epoch": 3.7312767143653596,
      "grad_norm": 0.8903640508651733,
      "learning_rate": 1.9615790510536967e-06,
      "loss": 0.7548,
      "step": 20175
    },
    {
      "epoch": 3.731462553428731,
      "grad_norm": 0.8818002343177795,
      "learning_rate": 1.9587006558181795e-06,
      "loss": 0.9411,
      "step": 20176
    },
    {
      "epoch": 3.731648392492102,
      "grad_norm": 0.8426865339279175,
      "learning_rate": 1.955824353105962e-06,
      "loss": 0.6555,
      "step": 20177
    },
    {
      "epoch": 3.731834231555473,
      "grad_norm": 1.3011345863342285,
      "learning_rate": 1.9529501429784158e-06,
      "loss": 0.7965,
      "step": 20178
    },
    {
      "epoch": 3.732020070618844,
      "grad_norm": 0.9135447144508362,
      "learning_rate": 1.9500780254968932e-06,
      "loss": 0.9401,
      "step": 20179
    },
    {
      "epoch": 3.7322059096822153,
      "grad_norm": 0.8463492393493652,
      "learning_rate": 1.9472080007227e-06,
      "loss": 0.9496,
      "step": 20180
    },
    {
      "epoch": 3.732391748745586,
      "grad_norm": 0.9751183390617371,
      "learning_rate": 1.9443400687170763e-06,
      "loss": 0.8128,
      "step": 20181
    },
    {
      "epoch": 3.7325775878089575,
      "grad_norm": 1.836098551750183,
      "learning_rate": 1.9414742295412515e-06,
      "loss": 1.3495,
      "step": 20182
    },
    {
      "epoch": 3.7327634268723284,
      "grad_norm": 0.9058772325515747,
      "learning_rate": 1.9386104832563867e-06,
      "loss": 0.917,
      "step": 20183
    },
    {
      "epoch": 3.7329492659356998,
      "grad_norm": 0.9225811958312988,
      "learning_rate": 1.9357488299235894e-06,
      "loss": 0.7237,
      "step": 20184
    },
    {
      "epoch": 3.7331351049990706,
      "grad_norm": 1.1890093088150024,
      "learning_rate": 1.9328892696039545e-06,
      "loss": 0.8495,
      "step": 20185
    },
    {
      "epoch": 3.733320944062442,
      "grad_norm": 0.877237856388092,
      "learning_rate": 1.930031802358501e-06,
      "loss": 0.7467,
      "step": 20186
    },
    {
      "epoch": 3.7335067831258133,
      "grad_norm": 0.9402475357055664,
      "learning_rate": 1.927176428248234e-06,
      "loss": 0.8492,
      "step": 20187
    },
    {
      "epoch": 3.733692622189184,
      "grad_norm": 0.8594173789024353,
      "learning_rate": 1.924323147334095e-06,
      "loss": 0.8439,
      "step": 20188
    },
    {
      "epoch": 3.733878461252555,
      "grad_norm": 0.876255214214325,
      "learning_rate": 1.921471959676957e-06,
      "loss": 0.6615,
      "step": 20189
    },
    {
      "epoch": 3.7340643003159264,
      "grad_norm": 0.966371476650238,
      "learning_rate": 1.918622865337694e-06,
      "loss": 0.7981,
      "step": 20190
    },
    {
      "epoch": 3.7340643003159264,
      "eval_loss": 1.0422886610031128,
      "eval_runtime": 23.1336,
      "eval_samples_per_second": 47.204,
      "eval_steps_per_second": 23.602,
      "step": 20190
    },
    {
      "epoch": 3.7342501393792977,
      "grad_norm": 1.0176982879638672,
      "learning_rate": 1.915775864377123e-06,
      "loss": 0.9423,
      "step": 20191
    },
    {
      "epoch": 3.7344359784426686,
      "grad_norm": 1.9772957563400269,
      "learning_rate": 1.912930956855985e-06,
      "loss": 1.09,
      "step": 20192
    },
    {
      "epoch": 3.7346218175060395,
      "grad_norm": 0.8535692095756531,
      "learning_rate": 1.9100881428350094e-06,
      "loss": 0.8229,
      "step": 20193
    },
    {
      "epoch": 3.734807656569411,
      "grad_norm": 0.8185490369796753,
      "learning_rate": 1.9072474223748805e-06,
      "loss": 0.9348,
      "step": 20194
    },
    {
      "epoch": 3.734993495632782,
      "grad_norm": 1.1920720338821411,
      "learning_rate": 1.9044087955362167e-06,
      "loss": 0.8847,
      "step": 20195
    },
    {
      "epoch": 3.735179334696153,
      "grad_norm": 1.0626229047775269,
      "learning_rate": 1.901572262379614e-06,
      "loss": 0.9121,
      "step": 20196
    },
    {
      "epoch": 3.7353651737595244,
      "grad_norm": 1.131238341331482,
      "learning_rate": 1.8987378229656016e-06,
      "loss": 0.76,
      "step": 20197
    },
    {
      "epoch": 3.7355510128228953,
      "grad_norm": 0.8504828214645386,
      "learning_rate": 1.895905477354687e-06,
      "loss": 0.4397,
      "step": 20198
    },
    {
      "epoch": 3.7357368518862666,
      "grad_norm": 0.991796612739563,
      "learning_rate": 1.8930752256073103e-06,
      "loss": 0.8937,
      "step": 20199
    },
    {
      "epoch": 3.7359226909496375,
      "grad_norm": 0.8266373872756958,
      "learning_rate": 1.8902470677839013e-06,
      "loss": 0.7949,
      "step": 20200
    },
    {
      "epoch": 3.736108530013009,
      "grad_norm": 1.1079061031341553,
      "learning_rate": 1.8874210039447782e-06,
      "loss": 0.8597,
      "step": 20201
    },
    {
      "epoch": 3.73629436907638,
      "grad_norm": 0.7702348232269287,
      "learning_rate": 1.8845970341503038e-06,
      "loss": 0.6479,
      "step": 20202
    },
    {
      "epoch": 3.736480208139751,
      "grad_norm": 1.0181301832199097,
      "learning_rate": 1.8817751584607413e-06,
      "loss": 0.5921,
      "step": 20203
    },
    {
      "epoch": 3.736666047203122,
      "grad_norm": 1.0332307815551758,
      "learning_rate": 1.8789553769362978e-06,
      "loss": 0.7162,
      "step": 20204
    },
    {
      "epoch": 3.7368518862664932,
      "grad_norm": 0.8270226120948792,
      "learning_rate": 1.8761376896371697e-06,
      "loss": 0.811,
      "step": 20205
    },
    {
      "epoch": 3.7370377253298646,
      "grad_norm": 1.0913925170898438,
      "learning_rate": 1.8733220966234866e-06,
      "loss": 0.8046,
      "step": 20206
    },
    {
      "epoch": 3.7372235643932354,
      "grad_norm": 0.9492542743682861,
      "learning_rate": 1.870508597955367e-06,
      "loss": 0.8421,
      "step": 20207
    },
    {
      "epoch": 3.7374094034566063,
      "grad_norm": 1.0270322561264038,
      "learning_rate": 1.8676971936928412e-06,
      "loss": 0.8161,
      "step": 20208
    },
    {
      "epoch": 3.7375952425199777,
      "grad_norm": 1.083715796470642,
      "learning_rate": 1.864887883895905e-06,
      "loss": 0.8378,
      "step": 20209
    },
    {
      "epoch": 3.737781081583349,
      "grad_norm": 1.139499545097351,
      "learning_rate": 1.862080668624544e-06,
      "loss": 0.9432,
      "step": 20210
    },
    {
      "epoch": 3.73796692064672,
      "grad_norm": 0.8852711915969849,
      "learning_rate": 1.8592755479386549e-06,
      "loss": 0.731,
      "step": 20211
    },
    {
      "epoch": 3.738152759710091,
      "grad_norm": 0.8319820761680603,
      "learning_rate": 1.8564725218981227e-06,
      "loss": 0.9048,
      "step": 20212
    },
    {
      "epoch": 3.738338598773462,
      "grad_norm": 1.0262010097503662,
      "learning_rate": 1.8536715905627443e-06,
      "loss": 0.8344,
      "step": 20213
    },
    {
      "epoch": 3.7385244378368334,
      "grad_norm": 0.8073480725288391,
      "learning_rate": 1.8508727539923276e-06,
      "loss": 0.8241,
      "step": 20214
    },
    {
      "epoch": 3.7387102769002043,
      "grad_norm": 0.7722681164741516,
      "learning_rate": 1.8480760122466134e-06,
      "loss": 0.7719,
      "step": 20215
    },
    {
      "epoch": 3.7388961159635756,
      "grad_norm": 0.8970304727554321,
      "learning_rate": 1.845281365385265e-06,
      "loss": 0.7599,
      "step": 20216
    },
    {
      "epoch": 3.7390819550269465,
      "grad_norm": 0.8711795210838318,
      "learning_rate": 1.8424888134679574e-06,
      "loss": 0.8864,
      "step": 20217
    },
    {
      "epoch": 3.739267794090318,
      "grad_norm": 1.1783910989761353,
      "learning_rate": 1.8396983565542648e-06,
      "loss": 0.8427,
      "step": 20218
    },
    {
      "epoch": 3.7394536331536887,
      "grad_norm": 0.8600555658340454,
      "learning_rate": 1.8369099947037728e-06,
      "loss": 0.7282,
      "step": 20219
    },
    {
      "epoch": 3.73963947221706,
      "grad_norm": 0.8570142984390259,
      "learning_rate": 1.8341237279759894e-06,
      "loss": 0.7585,
      "step": 20220
    },
    {
      "epoch": 3.7398253112804314,
      "grad_norm": 1.1565258502960205,
      "learning_rate": 1.831339556430356e-06,
      "loss": 0.9635,
      "step": 20221
    },
    {
      "epoch": 3.7400111503438023,
      "grad_norm": 1.0019115209579468,
      "learning_rate": 1.828557480126336e-06,
      "loss": 0.8266,
      "step": 20222
    },
    {
      "epoch": 3.740196989407173,
      "grad_norm": 1.0627496242523193,
      "learning_rate": 1.8257774991232824e-06,
      "loss": 0.8761,
      "step": 20223
    },
    {
      "epoch": 3.7403828284705445,
      "grad_norm": 1.1928831338882446,
      "learning_rate": 1.8229996134805249e-06,
      "loss": 0.9779,
      "step": 20224
    },
    {
      "epoch": 3.740568667533916,
      "grad_norm": 0.8400055766105652,
      "learning_rate": 1.820223823257372e-06,
      "loss": 0.6205,
      "step": 20225
    },
    {
      "epoch": 3.7407545065972867,
      "grad_norm": 0.9616957902908325,
      "learning_rate": 1.8174501285130429e-06,
      "loss": 0.9422,
      "step": 20226
    },
    {
      "epoch": 3.740940345660658,
      "grad_norm": 1.165166974067688,
      "learning_rate": 1.8146785293067681e-06,
      "loss": 0.6762,
      "step": 20227
    },
    {
      "epoch": 3.741126184724029,
      "grad_norm": 0.9141461253166199,
      "learning_rate": 1.811909025697689e-06,
      "loss": 0.7108,
      "step": 20228
    },
    {
      "epoch": 3.7413120237874002,
      "grad_norm": 0.931006133556366,
      "learning_rate": 1.8091416177449029e-06,
      "loss": 0.7751,
      "step": 20229
    },
    {
      "epoch": 3.741497862850771,
      "grad_norm": 0.9470266103744507,
      "learning_rate": 1.8063763055074957e-06,
      "loss": 0.8048,
      "step": 20230
    },
    {
      "epoch": 3.7416837019141425,
      "grad_norm": 1.2754347324371338,
      "learning_rate": 1.8036130890444757e-06,
      "loss": 1.2652,
      "step": 20231
    },
    {
      "epoch": 3.7418695409775133,
      "grad_norm": 0.8617785573005676,
      "learning_rate": 1.8008519684148295e-06,
      "loss": 0.8024,
      "step": 20232
    },
    {
      "epoch": 3.7420553800408847,
      "grad_norm": 1.0762972831726074,
      "learning_rate": 1.7980929436774651e-06,
      "loss": 0.7773,
      "step": 20233
    },
    {
      "epoch": 3.7422412191042556,
      "grad_norm": 1.0489073991775513,
      "learning_rate": 1.7953360148913023e-06,
      "loss": 0.6359,
      "step": 20234
    },
    {
      "epoch": 3.742427058167627,
      "grad_norm": 0.9695454835891724,
      "learning_rate": 1.7925811821151606e-06,
      "loss": 0.6696,
      "step": 20235
    },
    {
      "epoch": 3.742612897230998,
      "grad_norm": 1.1188420057296753,
      "learning_rate": 1.7898284454078484e-06,
      "loss": 0.8205,
      "step": 20236
    },
    {
      "epoch": 3.742798736294369,
      "grad_norm": 0.9165396690368652,
      "learning_rate": 1.7870778048280966e-06,
      "loss": 0.7197,
      "step": 20237
    },
    {
      "epoch": 3.74298457535774,
      "grad_norm": 0.9933660626411438,
      "learning_rate": 1.7843292604346362e-06,
      "loss": 0.7233,
      "step": 20238
    },
    {
      "epoch": 3.7431704144211113,
      "grad_norm": 0.9684987664222717,
      "learning_rate": 1.78158281228612e-06,
      "loss": 0.6795,
      "step": 20239
    },
    {
      "epoch": 3.7433562534844826,
      "grad_norm": 0.9409486651420593,
      "learning_rate": 1.778838460441179e-06,
      "loss": 0.836,
      "step": 20240
    },
    {
      "epoch": 3.7435420925478535,
      "grad_norm": 1.025458812713623,
      "learning_rate": 1.7760962049583663e-06,
      "loss": 0.9465,
      "step": 20241
    },
    {
      "epoch": 3.7437279316112244,
      "grad_norm": 0.9537796974182129,
      "learning_rate": 1.7733560458962351e-06,
      "loss": 0.6594,
      "step": 20242
    },
    {
      "epoch": 3.7439137706745957,
      "grad_norm": 0.884209930896759,
      "learning_rate": 1.7706179833132497e-06,
      "loss": 0.7746,
      "step": 20243
    },
    {
      "epoch": 3.744099609737967,
      "grad_norm": 1.0387382507324219,
      "learning_rate": 1.7678820172678523e-06,
      "loss": 0.8117,
      "step": 20244
    },
    {
      "epoch": 3.744285448801338,
      "grad_norm": 1.0448459386825562,
      "learning_rate": 1.7651481478184296e-06,
      "loss": 0.7244,
      "step": 20245
    },
    {
      "epoch": 3.7444712878647093,
      "grad_norm": 0.8891094326972961,
      "learning_rate": 1.7624163750233568e-06,
      "loss": 0.7074,
      "step": 20246
    },
    {
      "epoch": 3.74465712692808,
      "grad_norm": 0.7286022305488586,
      "learning_rate": 1.7596866989409211e-06,
      "loss": 0.5185,
      "step": 20247
    },
    {
      "epoch": 3.7448429659914515,
      "grad_norm": 1.0601147413253784,
      "learning_rate": 1.7569591196293867e-06,
      "loss": 0.6587,
      "step": 20248
    },
    {
      "epoch": 3.7450288050548224,
      "grad_norm": 0.9401094317436218,
      "learning_rate": 1.7542336371469513e-06,
      "loss": 0.7293,
      "step": 20249
    },
    {
      "epoch": 3.7452146441181937,
      "grad_norm": 1.1351622343063354,
      "learning_rate": 1.7515102515518245e-06,
      "loss": 1.0622,
      "step": 20250
    },
    {
      "epoch": 3.7454004831815646,
      "grad_norm": 0.8532180190086365,
      "learning_rate": 1.7487889629021037e-06,
      "loss": 0.6642,
      "step": 20251
    },
    {
      "epoch": 3.745586322244936,
      "grad_norm": 1.0489792823791504,
      "learning_rate": 1.746069771255876e-06,
      "loss": 0.8137,
      "step": 20252
    },
    {
      "epoch": 3.745772161308307,
      "grad_norm": 0.803139865398407,
      "learning_rate": 1.7433526766711728e-06,
      "loss": 0.6271,
      "step": 20253
    },
    {
      "epoch": 3.745958000371678,
      "grad_norm": 1.0900132656097412,
      "learning_rate": 1.740637679206003e-06,
      "loss": 1.0272,
      "step": 20254
    },
    {
      "epoch": 3.7461438394350495,
      "grad_norm": 0.9377065896987915,
      "learning_rate": 1.7379247789182985e-06,
      "loss": 0.898,
      "step": 20255
    },
    {
      "epoch": 3.7463296784984204,
      "grad_norm": 1.0326143503189087,
      "learning_rate": 1.735213975865968e-06,
      "loss": 0.9932,
      "step": 20256
    },
    {
      "epoch": 3.7465155175617912,
      "grad_norm": 0.9098765850067139,
      "learning_rate": 1.7325052701068545e-06,
      "loss": 0.858,
      "step": 20257
    },
    {
      "epoch": 3.7467013566251626,
      "grad_norm": 0.9849448800086975,
      "learning_rate": 1.7297986616987783e-06,
      "loss": 0.9108,
      "step": 20258
    },
    {
      "epoch": 3.746887195688534,
      "grad_norm": 0.8679027557373047,
      "learning_rate": 1.7270941506995265e-06,
      "loss": 0.721,
      "step": 20259
    },
    {
      "epoch": 3.747073034751905,
      "grad_norm": 0.8865405321121216,
      "learning_rate": 1.7243917371668083e-06,
      "loss": 0.6828,
      "step": 20260
    },
    {
      "epoch": 3.747258873815276,
      "grad_norm": 0.8816723227500916,
      "learning_rate": 1.7216914211582891e-06,
      "loss": 0.9355,
      "step": 20261
    },
    {
      "epoch": 3.747444712878647,
      "grad_norm": 0.8717049956321716,
      "learning_rate": 1.7189932027316225e-06,
      "loss": 0.5854,
      "step": 20262
    },
    {
      "epoch": 3.7476305519420183,
      "grad_norm": 0.9659672379493713,
      "learning_rate": 1.7162970819443846e-06,
      "loss": 0.6806,
      "step": 20263
    },
    {
      "epoch": 3.747816391005389,
      "grad_norm": 0.8695626258850098,
      "learning_rate": 1.7136030588541297e-06,
      "loss": 0.7705,
      "step": 20264
    },
    {
      "epoch": 3.7480022300687605,
      "grad_norm": 0.7861249446868896,
      "learning_rate": 1.7109111335183336e-06,
      "loss": 0.8505,
      "step": 20265
    },
    {
      "epoch": 3.7481880691321314,
      "grad_norm": 0.9285872578620911,
      "learning_rate": 1.7082213059944841e-06,
      "loss": 0.6255,
      "step": 20266
    },
    {
      "epoch": 3.7483739081955028,
      "grad_norm": 1.0697044134140015,
      "learning_rate": 1.7055335763399794e-06,
      "loss": 0.7684,
      "step": 20267
    },
    {
      "epoch": 3.7485597472588736,
      "grad_norm": 0.9197176694869995,
      "learning_rate": 1.7028479446121737e-06,
      "loss": 0.8119,
      "step": 20268
    },
    {
      "epoch": 3.748745586322245,
      "grad_norm": 0.8771321773529053,
      "learning_rate": 1.7001644108683878e-06,
      "loss": 0.6254,
      "step": 20269
    },
    {
      "epoch": 3.7489314253856163,
      "grad_norm": 0.8180269002914429,
      "learning_rate": 1.6974829751659094e-06,
      "loss": 0.6798,
      "step": 20270
    },
    {
      "epoch": 3.749117264448987,
      "grad_norm": 0.9938405156135559,
      "learning_rate": 1.6948036375619702e-06,
      "loss": 0.8099,
      "step": 20271
    },
    {
      "epoch": 3.749303103512358,
      "grad_norm": 0.8783755302429199,
      "learning_rate": 1.6921263981137359e-06,
      "loss": 0.699,
      "step": 20272
    },
    {
      "epoch": 3.7494889425757294,
      "grad_norm": 1.013279914855957,
      "learning_rate": 1.6894512568783716e-06,
      "loss": 0.6969,
      "step": 20273
    },
    {
      "epoch": 3.7496747816391007,
      "grad_norm": 0.8796840310096741,
      "learning_rate": 1.686778213912954e-06,
      "loss": 0.7338,
      "step": 20274
    },
    {
      "epoch": 3.7498606207024716,
      "grad_norm": 0.925635039806366,
      "learning_rate": 1.6841072692745485e-06,
      "loss": 0.7921,
      "step": 20275
    },
    {
      "epoch": 3.750046459765843,
      "grad_norm": 1.2063930034637451,
      "learning_rate": 1.6814384230201541e-06,
      "loss": 0.8636,
      "step": 20276
    },
    {
      "epoch": 3.750232298829214,
      "grad_norm": 1.013020634651184,
      "learning_rate": 1.6787716752067362e-06,
      "loss": 0.7568,
      "step": 20277
    },
    {
      "epoch": 3.750418137892585,
      "grad_norm": 0.9455508589744568,
      "learning_rate": 1.676107025891216e-06,
      "loss": 0.8089,
      "step": 20278
    },
    {
      "epoch": 3.750603976955956,
      "grad_norm": 1.0164023637771606,
      "learning_rate": 1.673444475130448e-06,
      "loss": 0.8108,
      "step": 20279
    },
    {
      "epoch": 3.7507898160193274,
      "grad_norm": 1.0061945915222168,
      "learning_rate": 1.6707840229812866e-06,
      "loss": 0.8351,
      "step": 20280
    },
    {
      "epoch": 3.7509756550826983,
      "grad_norm": 1.016903042793274,
      "learning_rate": 1.6681256695004866e-06,
      "loss": 0.8434,
      "step": 20281
    },
    {
      "epoch": 3.7511614941460696,
      "grad_norm": 0.8196696639060974,
      "learning_rate": 1.6654694147448024e-06,
      "loss": 0.6653,
      "step": 20282
    },
    {
      "epoch": 3.7513473332094405,
      "grad_norm": 1.1199551820755005,
      "learning_rate": 1.6628152587709335e-06,
      "loss": 1.0246,
      "step": 20283
    },
    {
      "epoch": 3.751533172272812,
      "grad_norm": 1.1850839853286743,
      "learning_rate": 1.660163201635523e-06,
      "loss": 0.7926,
      "step": 20284
    },
    {
      "epoch": 3.751719011336183,
      "grad_norm": 0.8495197892189026,
      "learning_rate": 1.657513243395159e-06,
      "loss": 0.7317,
      "step": 20285
    },
    {
      "epoch": 3.751904850399554,
      "grad_norm": 0.982175350189209,
      "learning_rate": 1.6548653841064187e-06,
      "loss": 0.7159,
      "step": 20286
    },
    {
      "epoch": 3.752090689462925,
      "grad_norm": 1.007477879524231,
      "learning_rate": 1.6522196238258126e-06,
      "loss": 0.8783,
      "step": 20287
    },
    {
      "epoch": 3.7522765285262962,
      "grad_norm": 1.0071723461151123,
      "learning_rate": 1.649575962609795e-06,
      "loss": 0.6093,
      "step": 20288
    },
    {
      "epoch": 3.7524623675896676,
      "grad_norm": 0.8891699910163879,
      "learning_rate": 1.646934400514799e-06,
      "loss": 0.789,
      "step": 20289
    },
    {
      "epoch": 3.7526482066530384,
      "grad_norm": 0.9914696216583252,
      "learning_rate": 1.644294937597224e-06,
      "loss": 0.9451,
      "step": 20290
    },
    {
      "epoch": 3.7528340457164093,
      "grad_norm": 0.9179845452308655,
      "learning_rate": 1.64165757391338e-06,
      "loss": 0.6384,
      "step": 20291
    },
    {
      "epoch": 3.7530198847797807,
      "grad_norm": 1.0186740159988403,
      "learning_rate": 1.6390223095195779e-06,
      "loss": 0.7442,
      "step": 20292
    },
    {
      "epoch": 3.753205723843152,
      "grad_norm": 1.0324479341506958,
      "learning_rate": 1.636389144472028e-06,
      "loss": 0.8352,
      "step": 20293
    },
    {
      "epoch": 3.753391562906523,
      "grad_norm": 0.8830369710922241,
      "learning_rate": 1.6337580788269747e-06,
      "loss": 0.7471,
      "step": 20294
    },
    {
      "epoch": 3.753577401969894,
      "grad_norm": 1.063448190689087,
      "learning_rate": 1.6311291126405393e-06,
      "loss": 1.0423,
      "step": 20295
    },
    {
      "epoch": 3.753763241033265,
      "grad_norm": 0.9705498814582825,
      "learning_rate": 1.6285022459688549e-06,
      "loss": 0.9218,
      "step": 20296
    },
    {
      "epoch": 3.7539490800966364,
      "grad_norm": 0.6145190596580505,
      "learning_rate": 1.6258774788679655e-06,
      "loss": 0.3206,
      "step": 20297
    },
    {
      "epoch": 3.7541349191600073,
      "grad_norm": 0.9766961336135864,
      "learning_rate": 1.623254811393915e-06,
      "loss": 0.6823,
      "step": 20298
    },
    {
      "epoch": 3.7543207582233786,
      "grad_norm": 0.9116829633712769,
      "learning_rate": 1.62063424360267e-06,
      "loss": 0.8332,
      "step": 20299
    },
    {
      "epoch": 3.7545065972867495,
      "grad_norm": 0.977838933467865,
      "learning_rate": 1.6180157755501523e-06,
      "loss": 0.9376,
      "step": 20300
    },
    {
      "epoch": 3.754692436350121,
      "grad_norm": 1.1162077188491821,
      "learning_rate": 1.6153994072922506e-06,
      "loss": 0.8784,
      "step": 20301
    },
    {
      "epoch": 3.7548782754134917,
      "grad_norm": 0.8749675154685974,
      "learning_rate": 1.6127851388848315e-06,
      "loss": 0.7572,
      "step": 20302
    },
    {
      "epoch": 3.755064114476863,
      "grad_norm": 0.9837542176246643,
      "learning_rate": 1.610172970383672e-06,
      "loss": 0.7145,
      "step": 20303
    },
    {
      "epoch": 3.7552499535402344,
      "grad_norm": 0.8241393566131592,
      "learning_rate": 1.6075629018445281e-06,
      "loss": 0.9012,
      "step": 20304
    },
    {
      "epoch": 3.7554357926036053,
      "grad_norm": 0.8609997630119324,
      "learning_rate": 1.6049549333230884e-06,
      "loss": 0.6587,
      "step": 20305
    },
    {
      "epoch": 3.755621631666976,
      "grad_norm": 1.0367193222045898,
      "learning_rate": 1.6023490648750528e-06,
      "loss": 0.9347,
      "step": 20306
    },
    {
      "epoch": 3.7558074707303475,
      "grad_norm": 0.9241992235183716,
      "learning_rate": 1.5997452965560101e-06,
      "loss": 0.7351,
      "step": 20307
    },
    {
      "epoch": 3.755993309793719,
      "grad_norm": 0.9562812447547913,
      "learning_rate": 1.597143628421549e-06,
      "loss": 0.7696,
      "step": 20308
    },
    {
      "epoch": 3.7561791488570897,
      "grad_norm": 0.8069552779197693,
      "learning_rate": 1.5945440605271811e-06,
      "loss": 0.5351,
      "step": 20309
    },
    {
      "epoch": 3.756364987920461,
      "grad_norm": 0.9494739174842834,
      "learning_rate": 1.5919465929283949e-06,
      "loss": 0.8539,
      "step": 20310
    },
    {
      "epoch": 3.756550826983832,
      "grad_norm": 0.9122812747955322,
      "learning_rate": 1.5893512256806464e-06,
      "loss": 0.7785,
      "step": 20311
    },
    {
      "epoch": 3.7567366660472032,
      "grad_norm": 0.9202660322189331,
      "learning_rate": 1.5867579588393133e-06,
      "loss": 0.8463,
      "step": 20312
    },
    {
      "epoch": 3.756922505110574,
      "grad_norm": 0.9109393358230591,
      "learning_rate": 1.5841667924597513e-06,
      "loss": 0.8587,
      "step": 20313
    },
    {
      "epoch": 3.7571083441739455,
      "grad_norm": 1.0824145078659058,
      "learning_rate": 1.5815777265972608e-06,
      "loss": 0.816,
      "step": 20314
    },
    {
      "epoch": 3.7572941832373163,
      "grad_norm": 1.070737361907959,
      "learning_rate": 1.5789907613070976e-06,
      "loss": 0.7731,
      "step": 20315
    },
    {
      "epoch": 3.7574800223006877,
      "grad_norm": 1.022955298423767,
      "learning_rate": 1.576405896644484e-06,
      "loss": 0.605,
      "step": 20316
    },
    {
      "epoch": 3.7576658613640586,
      "grad_norm": 1.0162962675094604,
      "learning_rate": 1.5738231326645758e-06,
      "loss": 0.7206,
      "step": 20317
    },
    {
      "epoch": 3.75785170042743,
      "grad_norm": 1.1087743043899536,
      "learning_rate": 1.571242469422507e-06,
      "loss": 0.7641,
      "step": 20318
    },
    {
      "epoch": 3.758037539490801,
      "grad_norm": 1.0301648378372192,
      "learning_rate": 1.5686639069733667e-06,
      "loss": 0.9089,
      "step": 20319
    },
    {
      "epoch": 3.758223378554172,
      "grad_norm": 1.5919631719589233,
      "learning_rate": 1.5660874453721774e-06,
      "loss": 0.8016,
      "step": 20320
    },
    {
      "epoch": 3.758409217617543,
      "grad_norm": 0.8339594602584839,
      "learning_rate": 1.5635130846739176e-06,
      "loss": 0.8136,
      "step": 20321
    },
    {
      "epoch": 3.7585950566809143,
      "grad_norm": 0.7868873476982117,
      "learning_rate": 1.5609408249335544e-06,
      "loss": 0.7063,
      "step": 20322
    },
    {
      "epoch": 3.7587808957442856,
      "grad_norm": 0.9781759977340698,
      "learning_rate": 1.5583706662059993e-06,
      "loss": 0.8121,
      "step": 20323
    },
    {
      "epoch": 3.7589667348076565,
      "grad_norm": 0.9512813091278076,
      "learning_rate": 1.5558026085460753e-06,
      "loss": 0.7352,
      "step": 20324
    },
    {
      "epoch": 3.7591525738710274,
      "grad_norm": 0.8733657598495483,
      "learning_rate": 1.553236652008605e-06,
      "loss": 0.8675,
      "step": 20325
    },
    {
      "epoch": 3.7593384129343987,
      "grad_norm": 1.2183723449707031,
      "learning_rate": 1.550672796648356e-06,
      "loss": 0.8708,
      "step": 20326
    },
    {
      "epoch": 3.75952425199777,
      "grad_norm": 0.8526677489280701,
      "learning_rate": 1.5481110425200618e-06,
      "loss": 0.7549,
      "step": 20327
    },
    {
      "epoch": 3.759710091061141,
      "grad_norm": 0.891410768032074,
      "learning_rate": 1.5455513896783792e-06,
      "loss": 0.6507,
      "step": 20328
    },
    {
      "epoch": 3.7598959301245123,
      "grad_norm": 0.956473708152771,
      "learning_rate": 1.542993838177953e-06,
      "loss": 0.9669,
      "step": 20329
    },
    {
      "epoch": 3.760081769187883,
      "grad_norm": 1.0115660429000854,
      "learning_rate": 1.540438388073362e-06,
      "loss": 0.7748,
      "step": 20330
    },
    {
      "epoch": 3.7602676082512545,
      "grad_norm": 0.9662947654724121,
      "learning_rate": 1.5378850394191402e-06,
      "loss": 1.0672,
      "step": 20331
    },
    {
      "epoch": 3.7604534473146254,
      "grad_norm": 0.8652477860450745,
      "learning_rate": 1.5353337922697996e-06,
      "loss": 0.7552,
      "step": 20332
    },
    {
      "epoch": 3.7606392863779967,
      "grad_norm": 0.8552069664001465,
      "learning_rate": 1.5327846466797856e-06,
      "loss": 0.8663,
      "step": 20333
    },
    {
      "epoch": 3.760825125441368,
      "grad_norm": 0.9188373684883118,
      "learning_rate": 1.5302376027035214e-06,
      "loss": 0.7211,
      "step": 20334
    },
    {
      "epoch": 3.761010964504739,
      "grad_norm": 0.9596465229988098,
      "learning_rate": 1.5276926603953413e-06,
      "loss": 0.7554,
      "step": 20335
    },
    {
      "epoch": 3.76119680356811,
      "grad_norm": 0.9444963932037354,
      "learning_rate": 1.5251498198095793e-06,
      "loss": 0.6637,
      "step": 20336
    },
    {
      "epoch": 3.761382642631481,
      "grad_norm": 0.9106894731521606,
      "learning_rate": 1.5226090810005034e-06,
      "loss": 0.5993,
      "step": 20337
    },
    {
      "epoch": 3.7615684816948525,
      "grad_norm": 0.8916698694229126,
      "learning_rate": 1.520070444022348e-06,
      "loss": 0.7076,
      "step": 20338
    },
    {
      "epoch": 3.7617543207582234,
      "grad_norm": 0.8964123725891113,
      "learning_rate": 1.5175339089292918e-06,
      "loss": 0.7878,
      "step": 20339
    },
    {
      "epoch": 3.7619401598215942,
      "grad_norm": 1.3913319110870361,
      "learning_rate": 1.5149994757754582e-06,
      "loss": 1.0517,
      "step": 20340
    },
    {
      "epoch": 3.7621259988849656,
      "grad_norm": 1.08450186252594,
      "learning_rate": 1.5124671446149596e-06,
      "loss": 1.087,
      "step": 20341
    },
    {
      "epoch": 3.762311837948337,
      "grad_norm": 0.7509164214134216,
      "learning_rate": 1.5099369155018417e-06,
      "loss": 0.6092,
      "step": 20342
    },
    {
      "epoch": 3.762497677011708,
      "grad_norm": 0.96497642993927,
      "learning_rate": 1.5074087884901056e-06,
      "loss": 1.0966,
      "step": 20343
    },
    {
      "epoch": 3.762683516075079,
      "grad_norm": 0.8731896281242371,
      "learning_rate": 1.5048827636337082e-06,
      "loss": 0.8057,
      "step": 20344
    },
    {
      "epoch": 3.76286935513845,
      "grad_norm": 1.0215585231781006,
      "learning_rate": 1.502358840986562e-06,
      "loss": 0.9523,
      "step": 20345
    },
    {
      "epoch": 3.7630551942018213,
      "grad_norm": 0.8914654850959778,
      "learning_rate": 1.4998370206025348e-06,
      "loss": 0.8197,
      "step": 20346
    },
    {
      "epoch": 3.763241033265192,
      "grad_norm": 1.267439603805542,
      "learning_rate": 1.4973173025354615e-06,
      "loss": 1.0239,
      "step": 20347
    },
    {
      "epoch": 3.7634268723285635,
      "grad_norm": 1.2980300188064575,
      "learning_rate": 1.49479968683911e-06,
      "loss": 0.8948,
      "step": 20348
    },
    {
      "epoch": 3.7636127113919344,
      "grad_norm": 0.9254906177520752,
      "learning_rate": 1.4922841735672156e-06,
      "loss": 0.9238,
      "step": 20349
    },
    {
      "epoch": 3.7637985504553058,
      "grad_norm": 0.9976295828819275,
      "learning_rate": 1.489770762773468e-06,
      "loss": 0.8328,
      "step": 20350
    },
    {
      "epoch": 3.7639843895186766,
      "grad_norm": 0.8693231344223022,
      "learning_rate": 1.487259454511525e-06,
      "loss": 0.9526,
      "step": 20351
    },
    {
      "epoch": 3.764170228582048,
      "grad_norm": 1.0530657768249512,
      "learning_rate": 1.4847502488349542e-06,
      "loss": 0.7263,
      "step": 20352
    },
    {
      "epoch": 3.7643560676454193,
      "grad_norm": 0.9318096041679382,
      "learning_rate": 1.482243145797335e-06,
      "loss": 0.9668,
      "step": 20353
    },
    {
      "epoch": 3.76454190670879,
      "grad_norm": 0.851360559463501,
      "learning_rate": 1.479738145452181e-06,
      "loss": 0.8775,
      "step": 20354
    },
    {
      "epoch": 3.764727745772161,
      "grad_norm": 1.3592497110366821,
      "learning_rate": 1.4772352478529483e-06,
      "loss": 1.015,
      "step": 20355
    },
    {
      "epoch": 3.7649135848355324,
      "grad_norm": 1.0484044551849365,
      "learning_rate": 1.4747344530530615e-06,
      "loss": 0.7613,
      "step": 20356
    },
    {
      "epoch": 3.7650994238989037,
      "grad_norm": 1.0023231506347656,
      "learning_rate": 1.472235761105878e-06,
      "loss": 0.8241,
      "step": 20357
    },
    {
      "epoch": 3.7652852629622746,
      "grad_norm": 1.194373369216919,
      "learning_rate": 1.4697391720647435e-06,
      "loss": 0.9966,
      "step": 20358
    },
    {
      "epoch": 3.765471102025646,
      "grad_norm": 0.8711073994636536,
      "learning_rate": 1.4672446859829492e-06,
      "loss": 0.7771,
      "step": 20359
    },
    {
      "epoch": 3.765656941089017,
      "grad_norm": 0.8670382499694824,
      "learning_rate": 1.46475230291373e-06,
      "loss": 0.7617,
      "step": 20360
    },
    {
      "epoch": 3.765842780152388,
      "grad_norm": 0.806544840335846,
      "learning_rate": 1.4622620229102767e-06,
      "loss": 0.6538,
      "step": 20361
    },
    {
      "epoch": 3.766028619215759,
      "grad_norm": 0.8053675889968872,
      "learning_rate": 1.459773846025747e-06,
      "loss": 0.5867,
      "step": 20362
    },
    {
      "epoch": 3.7662144582791304,
      "grad_norm": 0.9925404191017151,
      "learning_rate": 1.4572877723132316e-06,
      "loss": 0.93,
      "step": 20363
    },
    {
      "epoch": 3.7664002973425013,
      "grad_norm": 0.9203683137893677,
      "learning_rate": 1.4548038018258103e-06,
      "loss": 0.7603,
      "step": 20364
    },
    {
      "epoch": 3.7665861364058726,
      "grad_norm": 0.8202987909317017,
      "learning_rate": 1.4523219346164852e-06,
      "loss": 0.5888,
      "step": 20365
    },
    {
      "epoch": 3.7667719754692435,
      "grad_norm": 0.9790657758712769,
      "learning_rate": 1.449842170738236e-06,
      "loss": 0.7736,
      "step": 20366
    },
    {
      "epoch": 3.766957814532615,
      "grad_norm": 1.0166908502578735,
      "learning_rate": 1.4473645102439981e-06,
      "loss": 0.9156,
      "step": 20367
    },
    {
      "epoch": 3.767143653595986,
      "grad_norm": 1.00437331199646,
      "learning_rate": 1.4448889531866295e-06,
      "loss": 0.7816,
      "step": 20368
    },
    {
      "epoch": 3.767329492659357,
      "grad_norm": 0.9228891730308533,
      "learning_rate": 1.4424154996189765e-06,
      "loss": 0.7313,
      "step": 20369
    },
    {
      "epoch": 3.767515331722728,
      "grad_norm": 0.8260902762413025,
      "learning_rate": 1.4399441495938416e-06,
      "loss": 0.6848,
      "step": 20370
    },
    {
      "epoch": 3.7677011707860992,
      "grad_norm": 1.1458709239959717,
      "learning_rate": 1.43747490316396e-06,
      "loss": 0.9087,
      "step": 20371
    },
    {
      "epoch": 3.7678870098494706,
      "grad_norm": 0.9048810601234436,
      "learning_rate": 1.4350077603820345e-06,
      "loss": 0.659,
      "step": 20372
    },
    {
      "epoch": 3.7680728489128414,
      "grad_norm": 0.9446879029273987,
      "learning_rate": 1.4325427213007224e-06,
      "loss": 0.5988,
      "step": 20373
    },
    {
      "epoch": 3.7682586879762123,
      "grad_norm": 1.0057936906814575,
      "learning_rate": 1.4300797859726379e-06,
      "loss": 0.7141,
      "step": 20374
    },
    {
      "epoch": 3.7684445270395837,
      "grad_norm": 1.0417944192886353,
      "learning_rate": 1.4276189544503605e-06,
      "loss": 0.76,
      "step": 20375
    },
    {
      "epoch": 3.768630366102955,
      "grad_norm": 0.9925251603126526,
      "learning_rate": 1.4251602267863929e-06,
      "loss": 0.7363,
      "step": 20376
    },
    {
      "epoch": 3.768816205166326,
      "grad_norm": 0.9868441820144653,
      "learning_rate": 1.4227036030332041e-06,
      "loss": 0.981,
      "step": 20377
    },
    {
      "epoch": 3.769002044229697,
      "grad_norm": 0.9268509745597839,
      "learning_rate": 1.4202490832432635e-06,
      "loss": 0.887,
      "step": 20378
    },
    {
      "epoch": 3.769187883293068,
      "grad_norm": 0.9644826650619507,
      "learning_rate": 1.4177966674689292e-06,
      "loss": 0.7274,
      "step": 20379
    },
    {
      "epoch": 3.7693737223564394,
      "grad_norm": 0.8194114565849304,
      "learning_rate": 1.415346355762537e-06,
      "loss": 0.8135,
      "step": 20380
    },
    {
      "epoch": 3.7695595614198103,
      "grad_norm": 0.9341859221458435,
      "learning_rate": 1.4128981481764115e-06,
      "loss": 0.5956,
      "step": 20381
    },
    {
      "epoch": 3.7697454004831816,
      "grad_norm": 0.9932578802108765,
      "learning_rate": 1.4104520447628e-06,
      "loss": 1.014,
      "step": 20382
    },
    {
      "epoch": 3.769931239546553,
      "grad_norm": 1.0174542665481567,
      "learning_rate": 1.4080080455738942e-06,
      "loss": 0.9121,
      "step": 20383
    },
    {
      "epoch": 3.770117078609924,
      "grad_norm": 0.8680111765861511,
      "learning_rate": 1.4055661506618635e-06,
      "loss": 0.7486,
      "step": 20384
    },
    {
      "epoch": 3.7703029176732947,
      "grad_norm": 1.0944784879684448,
      "learning_rate": 1.4031263600788214e-06,
      "loss": 0.9201,
      "step": 20385
    },
    {
      "epoch": 3.770488756736666,
      "grad_norm": 0.9446232914924622,
      "learning_rate": 1.4006886738768598e-06,
      "loss": 0.6932,
      "step": 20386
    },
    {
      "epoch": 3.7706745958000374,
      "grad_norm": 1.0787503719329834,
      "learning_rate": 1.398253092108004e-06,
      "loss": 0.9325,
      "step": 20387
    },
    {
      "epoch": 3.7708604348634083,
      "grad_norm": 0.9129641652107239,
      "learning_rate": 1.395819614824212e-06,
      "loss": 0.6115,
      "step": 20388
    },
    {
      "epoch": 3.771046273926779,
      "grad_norm": 0.9007181525230408,
      "learning_rate": 1.3933882420774425e-06,
      "loss": 0.8728,
      "step": 20389
    },
    {
      "epoch": 3.7712321129901505,
      "grad_norm": 0.9841013550758362,
      "learning_rate": 1.390958973919587e-06,
      "loss": 0.9041,
      "step": 20390
    },
    {
      "epoch": 3.771417952053522,
      "grad_norm": 0.9039537906646729,
      "learning_rate": 1.3885318104025047e-06,
      "loss": 0.7337,
      "step": 20391
    },
    {
      "epoch": 3.7716037911168927,
      "grad_norm": 1.195389747619629,
      "learning_rate": 1.3861067515779647e-06,
      "loss": 0.6884,
      "step": 20392
    },
    {
      "epoch": 3.771789630180264,
      "grad_norm": 0.9455596208572388,
      "learning_rate": 1.383683797497759e-06,
      "loss": 0.8562,
      "step": 20393
    },
    {
      "epoch": 3.771975469243635,
      "grad_norm": 0.9761845469474792,
      "learning_rate": 1.3812629482135909e-06,
      "loss": 0.7933,
      "step": 20394
    },
    {
      "epoch": 3.7721613083070062,
      "grad_norm": 0.9781766533851624,
      "learning_rate": 1.3788442037771298e-06,
      "loss": 0.9388,
      "step": 20395
    },
    {
      "epoch": 3.772347147370377,
      "grad_norm": 0.9669922590255737,
      "learning_rate": 1.3764275642399904e-06,
      "loss": 0.9848,
      "step": 20396
    },
    {
      "epoch": 3.7725329864337485,
      "grad_norm": 1.0537102222442627,
      "learning_rate": 1.3740130296537646e-06,
      "loss": 0.8952,
      "step": 20397
    },
    {
      "epoch": 3.7727188254971193,
      "grad_norm": 0.9086155891418457,
      "learning_rate": 1.3716006000699888e-06,
      "loss": 0.7,
      "step": 20398
    },
    {
      "epoch": 3.7729046645604907,
      "grad_norm": 1.0693638324737549,
      "learning_rate": 1.3691902755401442e-06,
      "loss": 0.9063,
      "step": 20399
    },
    {
      "epoch": 3.7730905036238616,
      "grad_norm": 0.9978718757629395,
      "learning_rate": 1.3667820561156563e-06,
      "loss": 0.9654,
      "step": 20400
    },
    {
      "epoch": 3.773276342687233,
      "grad_norm": 0.9386802911758423,
      "learning_rate": 1.3643759418479618e-06,
      "loss": 0.7988,
      "step": 20401
    },
    {
      "epoch": 3.773462181750604,
      "grad_norm": 0.9663328528404236,
      "learning_rate": 1.3619719327883974e-06,
      "loss": 0.922,
      "step": 20402
    },
    {
      "epoch": 3.773648020813975,
      "grad_norm": 0.7999980449676514,
      "learning_rate": 1.3595700289882774e-06,
      "loss": 0.6153,
      "step": 20403
    },
    {
      "epoch": 3.773833859877346,
      "grad_norm": 1.1055376529693604,
      "learning_rate": 1.3571702304988388e-06,
      "loss": 0.8979,
      "step": 20404
    },
    {
      "epoch": 3.7740196989407173,
      "grad_norm": 1.031434178352356,
      "learning_rate": 1.3547725373713405e-06,
      "loss": 0.7908,
      "step": 20405
    },
    {
      "epoch": 3.7742055380040886,
      "grad_norm": 0.8867281079292297,
      "learning_rate": 1.3523769496569417e-06,
      "loss": 0.9089,
      "step": 20406
    },
    {
      "epoch": 3.7743913770674595,
      "grad_norm": 0.9022477865219116,
      "learning_rate": 1.349983467406768e-06,
      "loss": 0.7944,
      "step": 20407
    },
    {
      "epoch": 3.774577216130831,
      "grad_norm": 1.0239287614822388,
      "learning_rate": 1.3475920906719008e-06,
      "loss": 1.1125,
      "step": 20408
    },
    {
      "epoch": 3.7747630551942017,
      "grad_norm": 0.8552660346031189,
      "learning_rate": 1.3452028195033883e-06,
      "loss": 0.9971,
      "step": 20409
    },
    {
      "epoch": 3.774948894257573,
      "grad_norm": 0.9268264770507812,
      "learning_rate": 1.342815653952234e-06,
      "loss": 0.7839,
      "step": 20410
    },
    {
      "epoch": 3.775134733320944,
      "grad_norm": 1.1092424392700195,
      "learning_rate": 1.340430594069364e-06,
      "loss": 0.9033,
      "step": 20411
    },
    {
      "epoch": 3.7753205723843153,
      "grad_norm": 0.9117311835289001,
      "learning_rate": 1.338047639905704e-06,
      "loss": 0.7707,
      "step": 20412
    },
    {
      "epoch": 3.775506411447686,
      "grad_norm": 1.024247646331787,
      "learning_rate": 1.3356667915121025e-06,
      "loss": 0.8204,
      "step": 20413
    },
    {
      "epoch": 3.7756922505110575,
      "grad_norm": 0.9615598320960999,
      "learning_rate": 1.333288048939385e-06,
      "loss": 0.7454,
      "step": 20414
    },
    {
      "epoch": 3.7758780895744284,
      "grad_norm": 0.9834153652191162,
      "learning_rate": 1.3309114122383004e-06,
      "loss": 0.8387,
      "step": 20415
    },
    {
      "epoch": 3.7760639286377997,
      "grad_norm": 1.088229775428772,
      "learning_rate": 1.3285368814596078e-06,
      "loss": 0.7928,
      "step": 20416
    },
    {
      "epoch": 3.776249767701171,
      "grad_norm": 1.0894205570220947,
      "learning_rate": 1.3261644566539443e-06,
      "loss": 0.9444,
      "step": 20417
    },
    {
      "epoch": 3.776435606764542,
      "grad_norm": 0.9492201805114746,
      "learning_rate": 1.323794137871992e-06,
      "loss": 0.8596,
      "step": 20418
    },
    {
      "epoch": 3.776621445827913,
      "grad_norm": 1.5879576206207275,
      "learning_rate": 1.32142592516431e-06,
      "loss": 0.8944,
      "step": 20419
    },
    {
      "epoch": 3.776807284891284,
      "grad_norm": 0.9181216359138489,
      "learning_rate": 1.3190598185814474e-06,
      "loss": 0.9116,
      "step": 20420
    },
    {
      "epoch": 3.7769931239546555,
      "grad_norm": 1.0042266845703125,
      "learning_rate": 1.3166958181739186e-06,
      "loss": 1.0265,
      "step": 20421
    },
    {
      "epoch": 3.7771789630180264,
      "grad_norm": 0.8743492960929871,
      "learning_rate": 1.3143339239921727e-06,
      "loss": 0.7216,
      "step": 20422
    },
    {
      "epoch": 3.7773648020813972,
      "grad_norm": 0.9112885594367981,
      "learning_rate": 1.3119741360866134e-06,
      "loss": 0.8286,
      "step": 20423
    },
    {
      "epoch": 3.7775506411447686,
      "grad_norm": 1.1501792669296265,
      "learning_rate": 1.3096164545076007e-06,
      "loss": 0.8404,
      "step": 20424
    },
    {
      "epoch": 3.77773648020814,
      "grad_norm": 0.8658446073532104,
      "learning_rate": 1.307260879305483e-06,
      "loss": 0.6555,
      "step": 20425
    },
    {
      "epoch": 3.777922319271511,
      "grad_norm": 0.9965296387672424,
      "learning_rate": 1.3049074105304982e-06,
      "loss": 0.7141,
      "step": 20426
    },
    {
      "epoch": 3.778108158334882,
      "grad_norm": 1.0920850038528442,
      "learning_rate": 1.302556048232917e-06,
      "loss": 0.79,
      "step": 20427
    },
    {
      "epoch": 3.778293997398253,
      "grad_norm": 0.9294142127037048,
      "learning_rate": 1.3002067924628992e-06,
      "loss": 0.8282,
      "step": 20428
    },
    {
      "epoch": 3.7784798364616243,
      "grad_norm": 0.8633919954299927,
      "learning_rate": 1.2978596432705825e-06,
      "loss": 0.9949,
      "step": 20429
    },
    {
      "epoch": 3.778665675524995,
      "grad_norm": 0.9597409963607788,
      "learning_rate": 1.2955146007060826e-06,
      "loss": 0.739,
      "step": 20430
    },
    {
      "epoch": 3.7788515145883665,
      "grad_norm": 1.1199076175689697,
      "learning_rate": 1.293171664819437e-06,
      "loss": 0.9229,
      "step": 20431
    },
    {
      "epoch": 3.7790373536517374,
      "grad_norm": 1.026311993598938,
      "learning_rate": 1.2908308356606503e-06,
      "loss": 0.9393,
      "step": 20432
    },
    {
      "epoch": 3.7792231927151088,
      "grad_norm": 1.0156198740005493,
      "learning_rate": 1.2884921132796934e-06,
      "loss": 0.8739,
      "step": 20433
    },
    {
      "epoch": 3.7794090317784796,
      "grad_norm": 0.968630313873291,
      "learning_rate": 1.2861554977264712e-06,
      "loss": 0.8528,
      "step": 20434
    },
    {
      "epoch": 3.779594870841851,
      "grad_norm": 1.049005389213562,
      "learning_rate": 1.2838209890508657e-06,
      "loss": 0.6951,
      "step": 20435
    },
    {
      "epoch": 3.7797807099052223,
      "grad_norm": 1.0795906782150269,
      "learning_rate": 1.2814885873026927e-06,
      "loss": 0.8106,
      "step": 20436
    },
    {
      "epoch": 3.779966548968593,
      "grad_norm": 0.8436425924301147,
      "learning_rate": 1.2791582925317348e-06,
      "loss": 0.7104,
      "step": 20437
    },
    {
      "epoch": 3.780152388031964,
      "grad_norm": 0.9369443655014038,
      "learning_rate": 1.2768301047877406e-06,
      "loss": 0.8956,
      "step": 20438
    },
    {
      "epoch": 3.7803382270953354,
      "grad_norm": 1.007063865661621,
      "learning_rate": 1.2745040241203933e-06,
      "loss": 0.9896,
      "step": 20439
    },
    {
      "epoch": 3.7805240661587067,
      "grad_norm": 0.9993696808815002,
      "learning_rate": 1.2721800505793302e-06,
      "loss": 0.8909,
      "step": 20440
    },
    {
      "epoch": 3.7807099052220776,
      "grad_norm": 0.9917848706245422,
      "learning_rate": 1.2698581842141567e-06,
      "loss": 0.6445,
      "step": 20441
    },
    {
      "epoch": 3.780895744285449,
      "grad_norm": 1.1443630456924438,
      "learning_rate": 1.2675384250744437e-06,
      "loss": 0.6763,
      "step": 20442
    },
    {
      "epoch": 3.78108158334882,
      "grad_norm": 0.9800666570663452,
      "learning_rate": 1.2652207732096966e-06,
      "loss": 0.8198,
      "step": 20443
    },
    {
      "epoch": 3.781267422412191,
      "grad_norm": 1.080905556678772,
      "learning_rate": 1.2629052286693533e-06,
      "loss": 0.9398,
      "step": 20444
    },
    {
      "epoch": 3.781453261475562,
      "grad_norm": 0.9579738974571228,
      "learning_rate": 1.2605917915028742e-06,
      "loss": 0.6774,
      "step": 20445
    },
    {
      "epoch": 3.7816391005389334,
      "grad_norm": 1.017539620399475,
      "learning_rate": 1.25828046175962e-06,
      "loss": 1.009,
      "step": 20446
    },
    {
      "epoch": 3.7818249396023043,
      "grad_norm": 0.8087149262428284,
      "learning_rate": 1.2559712394889179e-06,
      "loss": 0.6337,
      "step": 20447
    },
    {
      "epoch": 3.7820107786656756,
      "grad_norm": 0.9919185042381287,
      "learning_rate": 1.2536641247400616e-06,
      "loss": 0.9481,
      "step": 20448
    },
    {
      "epoch": 3.7821966177290465,
      "grad_norm": 0.8533022403717041,
      "learning_rate": 1.2513591175622785e-06,
      "loss": 0.8498,
      "step": 20449
    },
    {
      "epoch": 3.782382456792418,
      "grad_norm": 1.1120437383651733,
      "learning_rate": 1.249056218004785e-06,
      "loss": 0.7406,
      "step": 20450
    },
    {
      "epoch": 3.782568295855789,
      "grad_norm": 0.973164439201355,
      "learning_rate": 1.2467554261167192e-06,
      "loss": 0.8322,
      "step": 20451
    },
    {
      "epoch": 3.78275413491916,
      "grad_norm": 0.8706233501434326,
      "learning_rate": 1.2444567419471976e-06,
      "loss": 0.894,
      "step": 20452
    },
    {
      "epoch": 3.782939973982531,
      "grad_norm": 0.7955963611602783,
      "learning_rate": 1.2421601655452698e-06,
      "loss": 0.78,
      "step": 20453
    },
    {
      "epoch": 3.7831258130459022,
      "grad_norm": 1.0311912298202515,
      "learning_rate": 1.239865696959963e-06,
      "loss": 0.8277,
      "step": 20454
    },
    {
      "epoch": 3.7833116521092736,
      "grad_norm": 0.847622275352478,
      "learning_rate": 1.2375733362402387e-06,
      "loss": 0.763,
      "step": 20455
    },
    {
      "epoch": 3.7834974911726444,
      "grad_norm": 0.9051774740219116,
      "learning_rate": 1.2352830834350238e-06,
      "loss": 0.7094,
      "step": 20456
    },
    {
      "epoch": 3.7836833302360158,
      "grad_norm": 0.8949741125106812,
      "learning_rate": 1.2329949385932127e-06,
      "loss": 0.7451,
      "step": 20457
    },
    {
      "epoch": 3.7838691692993867,
      "grad_norm": 1.043743371963501,
      "learning_rate": 1.2307089017636332e-06,
      "loss": 0.7586,
      "step": 20458
    },
    {
      "epoch": 3.784055008362758,
      "grad_norm": 0.8539413809776306,
      "learning_rate": 1.2284249729950792e-06,
      "loss": 0.9639,
      "step": 20459
    },
    {
      "epoch": 3.784240847426129,
      "grad_norm": 0.8604346513748169,
      "learning_rate": 1.22614315233629e-06,
      "loss": 0.7509,
      "step": 20460
    },
    {
      "epoch": 3.7844266864895,
      "grad_norm": 2.567366361618042,
      "learning_rate": 1.2238634398359706e-06,
      "loss": 1.1025,
      "step": 20461
    },
    {
      "epoch": 3.784612525552871,
      "grad_norm": 0.9127428531646729,
      "learning_rate": 1.2215858355427824e-06,
      "loss": 0.7166,
      "step": 20462
    },
    {
      "epoch": 3.7847983646162424,
      "grad_norm": 0.966698944568634,
      "learning_rate": 1.2193103395053418e-06,
      "loss": 0.7546,
      "step": 20463
    },
    {
      "epoch": 3.7849842036796133,
      "grad_norm": 1.4357216358184814,
      "learning_rate": 1.2170369517721991e-06,
      "loss": 1.1201,
      "step": 20464
    },
    {
      "epoch": 3.7851700427429846,
      "grad_norm": 1.5461227893829346,
      "learning_rate": 1.2147656723918821e-06,
      "loss": 1.4152,
      "step": 20465
    },
    {
      "epoch": 3.785355881806356,
      "grad_norm": 1.0198841094970703,
      "learning_rate": 1.212496501412874e-06,
      "loss": 0.9104,
      "step": 20466
    },
    {
      "epoch": 3.785541720869727,
      "grad_norm": 0.8557313084602356,
      "learning_rate": 1.2102294388836033e-06,
      "loss": 0.8077,
      "step": 20467
    },
    {
      "epoch": 3.7857275599330977,
      "grad_norm": 0.818263053894043,
      "learning_rate": 1.207964484852453e-06,
      "loss": 0.8271,
      "step": 20468
    },
    {
      "epoch": 3.785913398996469,
      "grad_norm": 1.010506272315979,
      "learning_rate": 1.2057016393677623e-06,
      "loss": 0.7292,
      "step": 20469
    },
    {
      "epoch": 3.7860992380598404,
      "grad_norm": 1.0275390148162842,
      "learning_rate": 1.203440902477837e-06,
      "loss": 0.8306,
      "step": 20470
    },
    {
      "epoch": 3.7862850771232113,
      "grad_norm": 0.9307306408882141,
      "learning_rate": 1.2011822742309275e-06,
      "loss": 0.7356,
      "step": 20471
    },
    {
      "epoch": 3.786470916186582,
      "grad_norm": 0.8827458024024963,
      "learning_rate": 1.1989257546752285e-06,
      "loss": 0.879,
      "step": 20472
    },
    {
      "epoch": 3.7866567552499535,
      "grad_norm": 0.9304745197296143,
      "learning_rate": 1.1966713438589128e-06,
      "loss": 0.787,
      "step": 20473
    },
    {
      "epoch": 3.786842594313325,
      "grad_norm": 0.8192884922027588,
      "learning_rate": 1.1944190418301083e-06,
      "loss": 0.5985,
      "step": 20474
    },
    {
      "epoch": 3.7870284333766957,
      "grad_norm": 0.9939797520637512,
      "learning_rate": 1.1921688486368543e-06,
      "loss": 0.9617,
      "step": 20475
    },
    {
      "epoch": 3.787214272440067,
      "grad_norm": 0.9714579582214355,
      "learning_rate": 1.1899207643272014e-06,
      "loss": 0.725,
      "step": 20476
    },
    {
      "epoch": 3.787400111503438,
      "grad_norm": 0.8307166695594788,
      "learning_rate": 1.1876747889491223e-06,
      "loss": 0.6712,
      "step": 20477
    },
    {
      "epoch": 3.7875859505668092,
      "grad_norm": 0.8596906065940857,
      "learning_rate": 1.1854309225505566e-06,
      "loss": 0.7358,
      "step": 20478
    },
    {
      "epoch": 3.78777178963018,
      "grad_norm": 0.865994930267334,
      "learning_rate": 1.1831891651793992e-06,
      "loss": 0.898,
      "step": 20479
    },
    {
      "epoch": 3.7879576286935515,
      "grad_norm": 0.9152920842170715,
      "learning_rate": 1.1809495168834783e-06,
      "loss": 0.7202,
      "step": 20480
    },
    {
      "epoch": 3.7881434677569223,
      "grad_norm": 0.860379159450531,
      "learning_rate": 1.1787119777106225e-06,
      "loss": 0.8855,
      "step": 20481
    },
    {
      "epoch": 3.7883293068202937,
      "grad_norm": 0.9607135057449341,
      "learning_rate": 1.1764765477085827e-06,
      "loss": 0.7073,
      "step": 20482
    },
    {
      "epoch": 3.7885151458836646,
      "grad_norm": 0.8664621710777283,
      "learning_rate": 1.1742432269250536e-06,
      "loss": 0.775,
      "step": 20483
    },
    {
      "epoch": 3.788700984947036,
      "grad_norm": 0.7738723158836365,
      "learning_rate": 1.1720120154077085e-06,
      "loss": 0.6682,
      "step": 20484
    },
    {
      "epoch": 3.788886824010407,
      "grad_norm": 1.108939290046692,
      "learning_rate": 1.169782913204176e-06,
      "loss": 0.7816,
      "step": 20485
    },
    {
      "epoch": 3.789072663073778,
      "grad_norm": 0.9274181723594666,
      "learning_rate": 1.167555920362029e-06,
      "loss": 0.9537,
      "step": 20486
    },
    {
      "epoch": 3.789258502137149,
      "grad_norm": 0.9644546508789062,
      "learning_rate": 1.1653310369287962e-06,
      "loss": 0.9466,
      "step": 20487
    },
    {
      "epoch": 3.7894443412005203,
      "grad_norm": 0.9580609202384949,
      "learning_rate": 1.163108262951962e-06,
      "loss": 0.6057,
      "step": 20488
    },
    {
      "epoch": 3.7896301802638916,
      "grad_norm": 1.0273137092590332,
      "learning_rate": 1.1608875984789659e-06,
      "loss": 0.7508,
      "step": 20489
    },
    {
      "epoch": 3.7898160193272625,
      "grad_norm": 0.8549194931983948,
      "learning_rate": 1.1586690435572256e-06,
      "loss": 0.9798,
      "step": 20490
    },
    {
      "epoch": 3.790001858390634,
      "grad_norm": 1.169147253036499,
      "learning_rate": 1.1564525982340702e-06,
      "loss": 0.831,
      "step": 20491
    },
    {
      "epoch": 3.7901876974540047,
      "grad_norm": 0.8162899613380432,
      "learning_rate": 1.154238262556795e-06,
      "loss": 0.7717,
      "step": 20492
    },
    {
      "epoch": 3.790373536517376,
      "grad_norm": 1.4703549146652222,
      "learning_rate": 1.1520260365726954e-06,
      "loss": 1.0135,
      "step": 20493
    },
    {
      "epoch": 3.790559375580747,
      "grad_norm": 1.3136372566223145,
      "learning_rate": 1.1498159203289672e-06,
      "loss": 0.8336,
      "step": 20494
    },
    {
      "epoch": 3.7907452146441183,
      "grad_norm": 0.9716363549232483,
      "learning_rate": 1.1476079138727836e-06,
      "loss": 0.8506,
      "step": 20495
    },
    {
      "epoch": 3.790931053707489,
      "grad_norm": 1.0611934661865234,
      "learning_rate": 1.1454020172512737e-06,
      "loss": 0.7567,
      "step": 20496
    },
    {
      "epoch": 3.7911168927708605,
      "grad_norm": 0.9886415600776672,
      "learning_rate": 1.1431982305115108e-06,
      "loss": 0.918,
      "step": 20497
    },
    {
      "epoch": 3.7913027318342314,
      "grad_norm": 0.9046659469604492,
      "learning_rate": 1.1409965537005463e-06,
      "loss": 0.829,
      "step": 20498
    },
    {
      "epoch": 3.7914885708976027,
      "grad_norm": 1.3718743324279785,
      "learning_rate": 1.1387969868653537e-06,
      "loss": 0.6962,
      "step": 20499
    },
    {
      "epoch": 3.791674409960974,
      "grad_norm": 1.355650782585144,
      "learning_rate": 1.1365995300528841e-06,
      "loss": 1.0893,
      "step": 20500
    },
    {
      "epoch": 3.791860249024345,
      "grad_norm": 0.8272162675857544,
      "learning_rate": 1.134404183310045e-06,
      "loss": 0.6227,
      "step": 20501
    },
    {
      "epoch": 3.792046088087716,
      "grad_norm": 0.842643141746521,
      "learning_rate": 1.1322109466836871e-06,
      "loss": 0.8684,
      "step": 20502
    },
    {
      "epoch": 3.792231927151087,
      "grad_norm": 0.9802500009536743,
      "learning_rate": 1.130019820220629e-06,
      "loss": 0.8027,
      "step": 20503
    },
    {
      "epoch": 3.7924177662144585,
      "grad_norm": 0.9252774715423584,
      "learning_rate": 1.1278308039676221e-06,
      "loss": 0.7099,
      "step": 20504
    },
    {
      "epoch": 3.7926036052778294,
      "grad_norm": 0.9800246953964233,
      "learning_rate": 1.1256438979713958e-06,
      "loss": 0.7664,
      "step": 20505
    },
    {
      "epoch": 3.7927894443412002,
      "grad_norm": 0.9181430339813232,
      "learning_rate": 1.1234591022786345e-06,
      "loss": 0.8109,
      "step": 20506
    },
    {
      "epoch": 3.7929752834045716,
      "grad_norm": 1.1357393264770508,
      "learning_rate": 1.121276416935957e-06,
      "loss": 0.6259,
      "step": 20507
    },
    {
      "epoch": 3.793161122467943,
      "grad_norm": 1.1219927072525024,
      "learning_rate": 1.1190958419899478e-06,
      "loss": 0.8326,
      "step": 20508
    },
    {
      "epoch": 3.793346961531314,
      "grad_norm": 1.0360108613967896,
      "learning_rate": 1.1169173774871478e-06,
      "loss": 1.0121,
      "step": 20509
    },
    {
      "epoch": 3.793532800594685,
      "grad_norm": 0.9814891219139099,
      "learning_rate": 1.114741023474053e-06,
      "loss": 0.716,
      "step": 20510
    },
    {
      "epoch": 3.793718639658056,
      "grad_norm": 1.0262317657470703,
      "learning_rate": 1.1125667799971262e-06,
      "loss": 0.8805,
      "step": 20511
    },
    {
      "epoch": 3.7939044787214273,
      "grad_norm": 0.9185773134231567,
      "learning_rate": 1.1103946471027637e-06,
      "loss": 0.8587,
      "step": 20512
    },
    {
      "epoch": 3.794090317784798,
      "grad_norm": 0.8239337801933289,
      "learning_rate": 1.1082246248373285e-06,
      "loss": 0.9026,
      "step": 20513
    },
    {
      "epoch": 3.7942761568481695,
      "grad_norm": 1.0945147275924683,
      "learning_rate": 1.1060567132471279e-06,
      "loss": 0.9878,
      "step": 20514
    },
    {
      "epoch": 3.794461995911541,
      "grad_norm": 1.008570909500122,
      "learning_rate": 1.103890912378447e-06,
      "loss": 0.9664,
      "step": 20515
    },
    {
      "epoch": 3.7946478349749118,
      "grad_norm": 0.9568138718605042,
      "learning_rate": 1.101727222277482e-06,
      "loss": 0.8537,
      "step": 20516
    },
    {
      "epoch": 3.7948336740382826,
      "grad_norm": 0.9105019569396973,
      "learning_rate": 1.0995656429904523e-06,
      "loss": 0.8228,
      "step": 20517
    },
    {
      "epoch": 3.795019513101654,
      "grad_norm": 0.9967132210731506,
      "learning_rate": 1.0974061745634646e-06,
      "loss": 1.0734,
      "step": 20518
    },
    {
      "epoch": 3.7952053521650253,
      "grad_norm": 1.0719634294509888,
      "learning_rate": 1.0952488170426268e-06,
      "loss": 0.7706,
      "step": 20519
    },
    {
      "epoch": 3.795391191228396,
      "grad_norm": 0.9113902449607849,
      "learning_rate": 1.0930935704739576e-06,
      "loss": 0.7723,
      "step": 20520
    },
    {
      "epoch": 3.795577030291767,
      "grad_norm": 1.1396875381469727,
      "learning_rate": 1.0909404349034868e-06,
      "loss": 0.7241,
      "step": 20521
    },
    {
      "epoch": 3.7957628693551384,
      "grad_norm": 0.885818600654602,
      "learning_rate": 1.0887894103771668e-06,
      "loss": 0.7088,
      "step": 20522
    },
    {
      "epoch": 3.7959487084185097,
      "grad_norm": 0.8519472479820251,
      "learning_rate": 1.0866404969408938e-06,
      "loss": 0.8594,
      "step": 20523
    },
    {
      "epoch": 3.7961345474818806,
      "grad_norm": 0.8347841501235962,
      "learning_rate": 1.0844936946405203e-06,
      "loss": 0.7616,
      "step": 20524
    },
    {
      "epoch": 3.796320386545252,
      "grad_norm": 1.0317564010620117,
      "learning_rate": 1.0823490035218987e-06,
      "loss": 0.6993,
      "step": 20525
    },
    {
      "epoch": 3.796506225608623,
      "grad_norm": 1.1278623342514038,
      "learning_rate": 1.0802064236307917e-06,
      "loss": 1.0551,
      "step": 20526
    },
    {
      "epoch": 3.796692064671994,
      "grad_norm": 0.937333345413208,
      "learning_rate": 1.0780659550129191e-06,
      "loss": 0.84,
      "step": 20527
    },
    {
      "epoch": 3.796877903735365,
      "grad_norm": 0.841964602470398,
      "learning_rate": 1.075927597713966e-06,
      "loss": 0.7571,
      "step": 20528
    },
    {
      "epoch": 3.7970637427987364,
      "grad_norm": 1.0166205167770386,
      "learning_rate": 1.073791351779596e-06,
      "loss": 0.8675,
      "step": 20529
    },
    {
      "epoch": 3.7972495818621073,
      "grad_norm": 1.0265706777572632,
      "learning_rate": 1.071657217255373e-06,
      "loss": 0.7984,
      "step": 20530
    },
    {
      "epoch": 3.7974354209254786,
      "grad_norm": 0.9770160913467407,
      "learning_rate": 1.069525194186849e-06,
      "loss": 0.7979,
      "step": 20531
    },
    {
      "epoch": 3.7976212599888495,
      "grad_norm": 0.9987168908119202,
      "learning_rate": 1.0673952826195432e-06,
      "loss": 0.9849,
      "step": 20532
    },
    {
      "epoch": 3.797807099052221,
      "grad_norm": 0.8842899799346924,
      "learning_rate": 1.0652674825989196e-06,
      "loss": 0.8466,
      "step": 20533
    },
    {
      "epoch": 3.797992938115592,
      "grad_norm": 0.9251445531845093,
      "learning_rate": 1.0631417941703858e-06,
      "loss": 0.9717,
      "step": 20534
    },
    {
      "epoch": 3.798178777178963,
      "grad_norm": 1.066029667854309,
      "learning_rate": 1.0610182173792948e-06,
      "loss": 0.8857,
      "step": 20535
    },
    {
      "epoch": 3.798364616242334,
      "grad_norm": 0.9665650129318237,
      "learning_rate": 1.058896752270988e-06,
      "loss": 0.9257,
      "step": 20536
    },
    {
      "epoch": 3.7985504553057052,
      "grad_norm": 1.0575613975524902,
      "learning_rate": 1.0567773988907514e-06,
      "loss": 0.8657,
      "step": 20537
    },
    {
      "epoch": 3.7987362943690766,
      "grad_norm": 0.8713177442550659,
      "learning_rate": 1.0546601572837933e-06,
      "loss": 0.8402,
      "step": 20538
    },
    {
      "epoch": 3.7989221334324474,
      "grad_norm": 1.0530487298965454,
      "learning_rate": 1.0525450274953218e-06,
      "loss": 0.6614,
      "step": 20539
    },
    {
      "epoch": 3.7991079724958188,
      "grad_norm": 1.0554367303848267,
      "learning_rate": 1.050432009570479e-06,
      "loss": 0.8807,
      "step": 20540
    },
    {
      "epoch": 3.7992938115591897,
      "grad_norm": 0.8764644861221313,
      "learning_rate": 1.0483211035543504e-06,
      "loss": 0.8133,
      "step": 20541
    },
    {
      "epoch": 3.799479650622561,
      "grad_norm": 1.0848758220672607,
      "learning_rate": 1.0462123094920117e-06,
      "loss": 0.8538,
      "step": 20542
    },
    {
      "epoch": 3.799665489685932,
      "grad_norm": 0.9339497685432434,
      "learning_rate": 1.04410562742846e-06,
      "loss": 0.9361,
      "step": 20543
    },
    {
      "epoch": 3.799851328749303,
      "grad_norm": 1.0878101587295532,
      "learning_rate": 1.0420010574086481e-06,
      "loss": 0.7417,
      "step": 20544
    },
    {
      "epoch": 3.800037167812674,
      "grad_norm": 0.9283502697944641,
      "learning_rate": 1.0398985994775069e-06,
      "loss": 0.8909,
      "step": 20545
    },
    {
      "epoch": 3.8002230068760454,
      "grad_norm": 0.9685944318771362,
      "learning_rate": 1.0377982536799224e-06,
      "loss": 0.8446,
      "step": 20546
    },
    {
      "epoch": 3.8004088459394163,
      "grad_norm": 1.0181951522827148,
      "learning_rate": 1.0357000200606926e-06,
      "loss": 0.7759,
      "step": 20547
    },
    {
      "epoch": 3.8005946850027876,
      "grad_norm": 1.517647624015808,
      "learning_rate": 1.0336038986646258e-06,
      "loss": 1.0891,
      "step": 20548
    },
    {
      "epoch": 3.800780524066159,
      "grad_norm": 0.9938087463378906,
      "learning_rate": 1.0315098895364416e-06,
      "loss": 0.8461,
      "step": 20549
    },
    {
      "epoch": 3.80096636312953,
      "grad_norm": 0.9021065831184387,
      "learning_rate": 1.029417992720849e-06,
      "loss": 0.7976,
      "step": 20550
    },
    {
      "epoch": 3.8011522021929007,
      "grad_norm": 0.8116894364356995,
      "learning_rate": 1.0273282082624901e-06,
      "loss": 0.8055,
      "step": 20551
    },
    {
      "epoch": 3.801338041256272,
      "grad_norm": 0.9933568835258484,
      "learning_rate": 1.0252405362059514e-06,
      "loss": 0.7171,
      "step": 20552
    },
    {
      "epoch": 3.8015238803196434,
      "grad_norm": 0.7740741968154907,
      "learning_rate": 1.0231549765958192e-06,
      "loss": 0.493,
      "step": 20553
    },
    {
      "epoch": 3.8017097193830143,
      "grad_norm": 0.8002797365188599,
      "learning_rate": 1.0210715294765916e-06,
      "loss": 0.8268,
      "step": 20554
    },
    {
      "epoch": 3.801895558446385,
      "grad_norm": 0.9246355891227722,
      "learning_rate": 1.0189901948927438e-06,
      "loss": 0.8769,
      "step": 20555
    },
    {
      "epoch": 3.8020813975097565,
      "grad_norm": 1.1914242506027222,
      "learning_rate": 1.0169109728886849e-06,
      "loss": 0.9132,
      "step": 20556
    },
    {
      "epoch": 3.802267236573128,
      "grad_norm": 1.0889278650283813,
      "learning_rate": 1.0148338635088018e-06,
      "loss": 0.7392,
      "step": 20557
    },
    {
      "epoch": 3.8024530756364987,
      "grad_norm": 1.1419200897216797,
      "learning_rate": 1.0127588667974253e-06,
      "loss": 0.868,
      "step": 20558
    },
    {
      "epoch": 3.80263891469987,
      "grad_norm": 0.9914236068725586,
      "learning_rate": 1.0106859827988424e-06,
      "loss": 0.8623,
      "step": 20559
    },
    {
      "epoch": 3.802824753763241,
      "grad_norm": 1.0282906293869019,
      "learning_rate": 1.0086152115572843e-06,
      "loss": 0.9744,
      "step": 20560
    },
    {
      "epoch": 3.8030105928266122,
      "grad_norm": 1.0758063793182373,
      "learning_rate": 1.0065465531169715e-06,
      "loss": 0.8696,
      "step": 20561
    },
    {
      "epoch": 3.803196431889983,
      "grad_norm": 0.7498318552970886,
      "learning_rate": 1.0044800075220352e-06,
      "loss": 0.6165,
      "step": 20562
    },
    {
      "epoch": 3.8033822709533545,
      "grad_norm": 0.7701108455657959,
      "learning_rate": 1.0024155748165953e-06,
      "loss": 0.8269,
      "step": 20563
    },
    {
      "epoch": 3.803568110016726,
      "grad_norm": 1.0195519924163818,
      "learning_rate": 1.0003532550446949e-06,
      "loss": 0.8114,
      "step": 20564
    },
    {
      "epoch": 3.8037539490800967,
      "grad_norm": 0.8638911247253418,
      "learning_rate": 9.98293048250376e-07,
      "loss": 0.6922,
      "step": 20565
    },
    {
      "epoch": 3.8039397881434676,
      "grad_norm": 1.0542092323303223,
      "learning_rate": 9.96234954477604e-07,
      "loss": 0.7888,
      "step": 20566
    },
    {
      "epoch": 3.804125627206839,
      "grad_norm": 0.9138762950897217,
      "learning_rate": 9.94178973770299e-07,
      "loss": 0.7489,
      "step": 20567
    },
    {
      "epoch": 3.80431146627021,
      "grad_norm": 0.6343844532966614,
      "learning_rate": 9.921251061723257e-07,
      "loss": 0.444,
      "step": 20568
    },
    {
      "epoch": 3.804497305333581,
      "grad_norm": 0.8197211027145386,
      "learning_rate": 9.900733517275495e-07,
      "loss": 0.6423,
      "step": 20569
    },
    {
      "epoch": 3.804683144396952,
      "grad_norm": 0.9333930611610413,
      "learning_rate": 9.880237104797575e-07,
      "loss": 0.7554,
      "step": 20570
    },
    {
      "epoch": 3.8048689834603233,
      "grad_norm": 0.8915857672691345,
      "learning_rate": 9.8597618247267e-07,
      "loss": 0.733,
      "step": 20571
    },
    {
      "epoch": 3.8050548225236946,
      "grad_norm": 0.8643396496772766,
      "learning_rate": 9.839307677500186e-07,
      "loss": 0.8303,
      "step": 20572
    },
    {
      "epoch": 3.8052406615870655,
      "grad_norm": 0.8911091685295105,
      "learning_rate": 9.818874663554357e-07,
      "loss": 0.8417,
      "step": 20573
    },
    {
      "epoch": 3.805426500650437,
      "grad_norm": 0.9695454835891724,
      "learning_rate": 9.798462783325524e-07,
      "loss": 0.8411,
      "step": 20574
    },
    {
      "epoch": 3.8056123397138077,
      "grad_norm": 0.6776549816131592,
      "learning_rate": 9.778072037249342e-07,
      "loss": 0.4747,
      "step": 20575
    },
    {
      "epoch": 3.805798178777179,
      "grad_norm": 0.903198778629303,
      "learning_rate": 9.757702425760685e-07,
      "loss": 0.7752,
      "step": 20576
    },
    {
      "epoch": 3.80598401784055,
      "grad_norm": 1.0293703079223633,
      "learning_rate": 9.737353949294759e-07,
      "loss": 1.0133,
      "step": 20577
    },
    {
      "epoch": 3.8061698569039213,
      "grad_norm": 1.0670793056488037,
      "learning_rate": 9.717026608285552e-07,
      "loss": 1.1066,
      "step": 20578
    },
    {
      "epoch": 3.806355695967292,
      "grad_norm": 1.0591223239898682,
      "learning_rate": 9.696720403167048e-07,
      "loss": 0.8048,
      "step": 20579
    },
    {
      "epoch": 3.8065415350306635,
      "grad_norm": 0.9527950286865234,
      "learning_rate": 9.67643533437257e-07,
      "loss": 0.8519,
      "step": 20580
    },
    {
      "epoch": 3.8067273740940344,
      "grad_norm": 1.242479681968689,
      "learning_rate": 9.656171402335213e-07,
      "loss": 0.6964,
      "step": 20581
    },
    {
      "epoch": 3.8069132131574057,
      "grad_norm": 0.9502848982810974,
      "learning_rate": 9.635928607487298e-07,
      "loss": 0.9544,
      "step": 20582
    },
    {
      "epoch": 3.807099052220777,
      "grad_norm": 0.8187753558158875,
      "learning_rate": 9.615706950260928e-07,
      "loss": 0.7997,
      "step": 20583
    },
    {
      "epoch": 3.807284891284148,
      "grad_norm": 1.0020241737365723,
      "learning_rate": 9.595506431087752e-07,
      "loss": 0.7883,
      "step": 20584
    },
    {
      "epoch": 3.807470730347519,
      "grad_norm": 0.8587465882301331,
      "learning_rate": 9.575327050398875e-07,
      "loss": 0.9517,
      "step": 20585
    },
    {
      "epoch": 3.80765656941089,
      "grad_norm": 0.9860709309577942,
      "learning_rate": 9.55516880862506e-07,
      "loss": 0.769,
      "step": 20586
    },
    {
      "epoch": 3.8078424084742615,
      "grad_norm": 0.9654290080070496,
      "learning_rate": 9.535031706196517e-07,
      "loss": 0.9548,
      "step": 20587
    },
    {
      "epoch": 3.8080282475376324,
      "grad_norm": 0.8824122548103333,
      "learning_rate": 9.514915743542907e-07,
      "loss": 0.9232,
      "step": 20588
    },
    {
      "epoch": 3.8082140866010037,
      "grad_norm": 1.4160360097885132,
      "learning_rate": 9.494820921093772e-07,
      "loss": 0.8105,
      "step": 20589
    },
    {
      "epoch": 3.8083999256643746,
      "grad_norm": 0.8777419924736023,
      "learning_rate": 9.474747239277993e-07,
      "loss": 0.7099,
      "step": 20590
    },
    {
      "epoch": 3.808585764727746,
      "grad_norm": 1.0006147623062134,
      "learning_rate": 9.45469469852378e-07,
      "loss": 0.8877,
      "step": 20591
    },
    {
      "epoch": 3.808771603791117,
      "grad_norm": 1.0253385305404663,
      "learning_rate": 9.434663299259239e-07,
      "loss": 0.7588,
      "step": 20592
    },
    {
      "epoch": 3.808957442854488,
      "grad_norm": 1.6540416479110718,
      "learning_rate": 9.414653041912025e-07,
      "loss": 0.6916,
      "step": 20593
    },
    {
      "epoch": 3.809143281917859,
      "grad_norm": 0.8838813900947571,
      "learning_rate": 9.39466392690913e-07,
      "loss": 0.8394,
      "step": 20594
    },
    {
      "epoch": 3.8093291209812303,
      "grad_norm": 0.9816716909408569,
      "learning_rate": 9.374695954677104e-07,
      "loss": 0.7785,
      "step": 20595
    },
    {
      "epoch": 3.809514960044601,
      "grad_norm": 1.0879133939743042,
      "learning_rate": 9.35474912564227e-07,
      "loss": 0.8887,
      "step": 20596
    },
    {
      "epoch": 3.8097007991079725,
      "grad_norm": 0.8840386271476746,
      "learning_rate": 9.33482344023029e-07,
      "loss": 0.7582,
      "step": 20597
    },
    {
      "epoch": 3.809886638171344,
      "grad_norm": 0.9541440606117249,
      "learning_rate": 9.314918898866487e-07,
      "loss": 0.9832,
      "step": 20598
    },
    {
      "epoch": 3.8100724772347148,
      "grad_norm": 1.02316153049469,
      "learning_rate": 9.295035501975636e-07,
      "loss": 0.862,
      "step": 20599
    },
    {
      "epoch": 3.8102583162980856,
      "grad_norm": 0.8722786903381348,
      "learning_rate": 9.275173249982061e-07,
      "loss": 0.8949,
      "step": 20600
    },
    {
      "epoch": 3.810444155361457,
      "grad_norm": 0.8821304440498352,
      "learning_rate": 9.25533214330987e-07,
      "loss": 0.7753,
      "step": 20601
    },
    {
      "epoch": 3.8106299944248283,
      "grad_norm": 0.8481576442718506,
      "learning_rate": 9.235512182382388e-07,
      "loss": 0.7523,
      "step": 20602
    },
    {
      "epoch": 3.810815833488199,
      "grad_norm": 1.0084933042526245,
      "learning_rate": 9.215713367622613e-07,
      "loss": 0.9815,
      "step": 20603
    },
    {
      "epoch": 3.81100167255157,
      "grad_norm": 0.9282767176628113,
      "learning_rate": 9.195935699453207e-07,
      "loss": 0.716,
      "step": 20604
    },
    {
      "epoch": 3.8111875116149414,
      "grad_norm": 1.095992922782898,
      "learning_rate": 9.176179178296385e-07,
      "loss": 0.7956,
      "step": 20605
    },
    {
      "epoch": 3.8113733506783127,
      "grad_norm": 1.1294467449188232,
      "learning_rate": 9.156443804573588e-07,
      "loss": 0.8014,
      "step": 20606
    },
    {
      "epoch": 3.8115591897416836,
      "grad_norm": 1.0914392471313477,
      "learning_rate": 9.13672957870615e-07,
      "loss": 1.0108,
      "step": 20607
    },
    {
      "epoch": 3.811745028805055,
      "grad_norm": 0.9797545075416565,
      "learning_rate": 9.117036501114839e-07,
      "loss": 0.7969,
      "step": 20608
    },
    {
      "epoch": 3.811930867868426,
      "grad_norm": 0.924508273601532,
      "learning_rate": 9.09736457221999e-07,
      "loss": 0.8286,
      "step": 20609
    },
    {
      "epoch": 3.812116706931797,
      "grad_norm": 1.0149329900741577,
      "learning_rate": 9.077713792441489e-07,
      "loss": 0.7124,
      "step": 20610
    },
    {
      "epoch": 3.812302545995168,
      "grad_norm": 1.2102501392364502,
      "learning_rate": 9.058084162198555e-07,
      "loss": 1.0784,
      "step": 20611
    },
    {
      "epoch": 3.8124883850585394,
      "grad_norm": 0.8796945810317993,
      "learning_rate": 9.038475681910296e-07,
      "loss": 0.7931,
      "step": 20612
    },
    {
      "epoch": 3.8126742241219103,
      "grad_norm": 1.0211964845657349,
      "learning_rate": 9.01888835199538e-07,
      "loss": 0.7759,
      "step": 20613
    },
    {
      "epoch": 3.8128600631852816,
      "grad_norm": 0.9216942191123962,
      "learning_rate": 8.999322172871583e-07,
      "loss": 0.9624,
      "step": 20614
    },
    {
      "epoch": 3.8130459022486525,
      "grad_norm": 1.1108382940292358,
      "learning_rate": 8.97977714495657e-07,
      "loss": 0.9636,
      "step": 20615
    },
    {
      "epoch": 3.813231741312024,
      "grad_norm": 1.3374735116958618,
      "learning_rate": 8.960253268667673e-07,
      "loss": 0.8246,
      "step": 20616
    },
    {
      "epoch": 3.813417580375395,
      "grad_norm": 0.8724303245544434,
      "learning_rate": 8.94075054442145e-07,
      "loss": 0.6753,
      "step": 20617
    },
    {
      "epoch": 3.813603419438766,
      "grad_norm": 0.8837342858314514,
      "learning_rate": 8.921268972634123e-07,
      "loss": 0.8393,
      "step": 20618
    },
    {
      "epoch": 3.813789258502137,
      "grad_norm": 0.996241569519043,
      "learning_rate": 8.90180855372158e-07,
      "loss": 0.8471,
      "step": 20619
    },
    {
      "epoch": 3.8139750975655082,
      "grad_norm": 0.8742111325263977,
      "learning_rate": 8.882369288099157e-07,
      "loss": 1.0219,
      "step": 20620
    },
    {
      "epoch": 3.8141609366288796,
      "grad_norm": 1.0098148584365845,
      "learning_rate": 8.862951176181744e-07,
      "loss": 0.9206,
      "step": 20621
    },
    {
      "epoch": 3.8143467756922504,
      "grad_norm": 0.9189237952232361,
      "learning_rate": 8.843554218383787e-07,
      "loss": 0.8208,
      "step": 20622
    },
    {
      "epoch": 3.8145326147556218,
      "grad_norm": 1.065704345703125,
      "learning_rate": 8.824178415119177e-07,
      "loss": 0.6901,
      "step": 20623
    },
    {
      "epoch": 3.8147184538189927,
      "grad_norm": 0.8512879610061646,
      "learning_rate": 8.804823766801696e-07,
      "loss": 0.5791,
      "step": 20624
    },
    {
      "epoch": 3.814904292882364,
      "grad_norm": 0.9663471579551697,
      "learning_rate": 8.785490273844121e-07,
      "loss": 0.8668,
      "step": 20625
    },
    {
      "epoch": 3.815090131945735,
      "grad_norm": 1.5551700592041016,
      "learning_rate": 8.766177936659458e-07,
      "loss": 0.8712,
      "step": 20626
    },
    {
      "epoch": 3.815275971009106,
      "grad_norm": 1.0716501474380493,
      "learning_rate": 8.746886755659601e-07,
      "loss": 0.8028,
      "step": 20627
    },
    {
      "epoch": 3.815461810072477,
      "grad_norm": 0.8538807034492493,
      "learning_rate": 8.727616731256327e-07,
      "loss": 0.803,
      "step": 20628
    },
    {
      "epoch": 3.8156476491358484,
      "grad_norm": 0.890023410320282,
      "learning_rate": 8.708367863861089e-07,
      "loss": 0.6986,
      "step": 20629
    },
    {
      "epoch": 3.8158334881992193,
      "grad_norm": 0.9053801894187927,
      "learning_rate": 8.689140153884556e-07,
      "loss": 0.7898,
      "step": 20630
    },
    {
      "epoch": 3.8160193272625906,
      "grad_norm": 1.0622429847717285,
      "learning_rate": 8.66993360173718e-07,
      "loss": 0.8973,
      "step": 20631
    },
    {
      "epoch": 3.816205166325962,
      "grad_norm": 0.9308961629867554,
      "learning_rate": 8.650748207828963e-07,
      "loss": 0.8658,
      "step": 20632
    },
    {
      "epoch": 3.816391005389333,
      "grad_norm": 1.0338478088378906,
      "learning_rate": 8.631583972569246e-07,
      "loss": 0.8523,
      "step": 20633
    },
    {
      "epoch": 3.8165768444527037,
      "grad_norm": 0.9793767929077148,
      "learning_rate": 8.612440896367035e-07,
      "loss": 0.7213,
      "step": 20634
    },
    {
      "epoch": 3.816762683516075,
      "grad_norm": 0.7624946236610413,
      "learning_rate": 8.593318979631004e-07,
      "loss": 0.524,
      "step": 20635
    },
    {
      "epoch": 3.8169485225794464,
      "grad_norm": 0.9758871793746948,
      "learning_rate": 8.57421822276927e-07,
      "loss": 0.7867,
      "step": 20636
    },
    {
      "epoch": 3.8171343616428173,
      "grad_norm": 0.9500848054885864,
      "learning_rate": 8.555138626189618e-07,
      "loss": 0.7514,
      "step": 20637
    },
    {
      "epoch": 3.8173202007061886,
      "grad_norm": 0.9387444257736206,
      "learning_rate": 8.536080190299056e-07,
      "loss": 0.7099,
      "step": 20638
    },
    {
      "epoch": 3.8175060397695595,
      "grad_norm": 0.9620884656906128,
      "learning_rate": 8.517042915504481e-07,
      "loss": 0.9058,
      "step": 20639
    },
    {
      "epoch": 3.817691878832931,
      "grad_norm": 0.9171242117881775,
      "learning_rate": 8.498026802212012e-07,
      "loss": 0.7526,
      "step": 20640
    },
    {
      "epoch": 3.8178777178963017,
      "grad_norm": 1.1685776710510254,
      "learning_rate": 8.479031850827879e-07,
      "loss": 0.9633,
      "step": 20641
    },
    {
      "epoch": 3.818063556959673,
      "grad_norm": 0.9995948076248169,
      "learning_rate": 8.460058061757203e-07,
      "loss": 0.7484,
      "step": 20642
    },
    {
      "epoch": 3.818249396023044,
      "grad_norm": 0.8692541122436523,
      "learning_rate": 8.441105435404995e-07,
      "loss": 0.908,
      "step": 20643
    },
    {
      "epoch": 3.8184352350864152,
      "grad_norm": 0.9087664484977722,
      "learning_rate": 8.422173972175818e-07,
      "loss": 0.8699,
      "step": 20644
    },
    {
      "epoch": 3.818621074149786,
      "grad_norm": 0.7954351902008057,
      "learning_rate": 8.403263672473793e-07,
      "loss": 0.6872,
      "step": 20645
    },
    {
      "epoch": 3.8188069132131575,
      "grad_norm": 0.9517496824264526,
      "learning_rate": 8.384374536702377e-07,
      "loss": 0.7928,
      "step": 20646
    },
    {
      "epoch": 3.818992752276529,
      "grad_norm": 0.947588324546814,
      "learning_rate": 8.365506565264913e-07,
      "loss": 0.8637,
      "step": 20647
    },
    {
      "epoch": 3.8191785913398997,
      "grad_norm": 1.1128437519073486,
      "learning_rate": 8.346659758563857e-07,
      "loss": 0.7051,
      "step": 20648
    },
    {
      "epoch": 3.8193644304032706,
      "grad_norm": 1.1500704288482666,
      "learning_rate": 8.327834117001665e-07,
      "loss": 0.9687,
      "step": 20649
    },
    {
      "epoch": 3.819550269466642,
      "grad_norm": 1.0038901567459106,
      "learning_rate": 8.309029640980126e-07,
      "loss": 0.6612,
      "step": 20650
    },
    {
      "epoch": 3.819736108530013,
      "grad_norm": 0.894903302192688,
      "learning_rate": 8.290246330900476e-07,
      "loss": 0.7769,
      "step": 20651
    },
    {
      "epoch": 3.819921947593384,
      "grad_norm": 1.0159218311309814,
      "learning_rate": 8.271484187163725e-07,
      "loss": 0.7865,
      "step": 20652
    },
    {
      "epoch": 3.820107786656755,
      "grad_norm": 0.9636523127555847,
      "learning_rate": 8.252743210170222e-07,
      "loss": 0.5899,
      "step": 20653
    },
    {
      "epoch": 3.8202936257201263,
      "grad_norm": 0.9794251918792725,
      "learning_rate": 8.23402340032009e-07,
      "loss": 0.6804,
      "step": 20654
    },
    {
      "epoch": 3.8204794647834976,
      "grad_norm": 0.9215170741081238,
      "learning_rate": 8.215324758012677e-07,
      "loss": 0.6189,
      "step": 20655
    },
    {
      "epoch": 3.8206653038468685,
      "grad_norm": 0.836532473564148,
      "learning_rate": 8.196647283647218e-07,
      "loss": 0.77,
      "step": 20656
    },
    {
      "epoch": 3.82085114291024,
      "grad_norm": 0.8718463182449341,
      "learning_rate": 8.177990977622396e-07,
      "loss": 0.8012,
      "step": 20657
    },
    {
      "epoch": 3.8210369819736107,
      "grad_norm": 0.9456227421760559,
      "learning_rate": 8.159355840336225e-07,
      "loss": 0.8262,
      "step": 20658
    },
    {
      "epoch": 3.821222821036982,
      "grad_norm": 0.8164328336715698,
      "learning_rate": 8.140741872186608e-07,
      "loss": 0.8653,
      "step": 20659
    },
    {
      "epoch": 3.821408660100353,
      "grad_norm": 1.0025135278701782,
      "learning_rate": 8.122149073570673e-07,
      "loss": 0.7972,
      "step": 20660
    },
    {
      "epoch": 3.8215944991637243,
      "grad_norm": 0.7859447598457336,
      "learning_rate": 8.103577444885435e-07,
      "loss": 0.8528,
      "step": 20661
    },
    {
      "epoch": 3.821780338227095,
      "grad_norm": 0.8448495864868164,
      "learning_rate": 8.085026986527133e-07,
      "loss": 0.6718,
      "step": 20662
    },
    {
      "epoch": 3.8219661772904665,
      "grad_norm": 0.8926912546157837,
      "learning_rate": 8.066497698891673e-07,
      "loss": 0.7745,
      "step": 20663
    },
    {
      "epoch": 3.8221520163538374,
      "grad_norm": 0.9226291179656982,
      "learning_rate": 8.047989582374738e-07,
      "loss": 0.8601,
      "step": 20664
    },
    {
      "epoch": 3.8223378554172087,
      "grad_norm": 0.8001793026924133,
      "learning_rate": 8.029502637371123e-07,
      "loss": 0.7786,
      "step": 20665
    },
    {
      "epoch": 3.82252369448058,
      "grad_norm": 1.0437462329864502,
      "learning_rate": 8.011036864275512e-07,
      "loss": 0.9139,
      "step": 20666
    },
    {
      "epoch": 3.822709533543951,
      "grad_norm": 0.8408721089363098,
      "learning_rate": 7.992592263481924e-07,
      "loss": 0.7911,
      "step": 20667
    },
    {
      "epoch": 3.822895372607322,
      "grad_norm": 0.9707818031311035,
      "learning_rate": 7.974168835384155e-07,
      "loss": 0.8485,
      "step": 20668
    },
    {
      "epoch": 3.823081211670693,
      "grad_norm": 0.9045571684837341,
      "learning_rate": 7.955766580375335e-07,
      "loss": 0.8209,
      "step": 20669
    },
    {
      "epoch": 3.8232670507340645,
      "grad_norm": 0.9762415289878845,
      "learning_rate": 7.937385498848371e-07,
      "loss": 0.8957,
      "step": 20670
    },
    {
      "epoch": 3.8234528897974354,
      "grad_norm": 0.8625732660293579,
      "learning_rate": 7.919025591195284e-07,
      "loss": 0.7141,
      "step": 20671
    },
    {
      "epoch": 3.8236387288608067,
      "grad_norm": 0.9887514114379883,
      "learning_rate": 7.900686857808204e-07,
      "loss": 0.8672,
      "step": 20672
    },
    {
      "epoch": 3.8238245679241776,
      "grad_norm": 1.12395179271698,
      "learning_rate": 7.882369299078485e-07,
      "loss": 0.8746,
      "step": 20673
    },
    {
      "epoch": 3.824010406987549,
      "grad_norm": 0.8853467106819153,
      "learning_rate": 7.864072915397036e-07,
      "loss": 0.9147,
      "step": 20674
    },
    {
      "epoch": 3.82419624605092,
      "grad_norm": 1.0425935983657837,
      "learning_rate": 7.845797707154434e-07,
      "loss": 0.8855,
      "step": 20675
    },
    {
      "epoch": 3.824382085114291,
      "grad_norm": 1.0093079805374146,
      "learning_rate": 7.827543674740589e-07,
      "loss": 0.9353,
      "step": 20676
    },
    {
      "epoch": 3.824567924177662,
      "grad_norm": 1.0275657176971436,
      "learning_rate": 7.80931081854519e-07,
      "loss": 0.902,
      "step": 20677
    },
    {
      "epoch": 3.8247537632410333,
      "grad_norm": 1.0822455883026123,
      "learning_rate": 7.79109913895737e-07,
      "loss": 0.8277,
      "step": 20678
    },
    {
      "epoch": 3.824939602304404,
      "grad_norm": 0.9282769560813904,
      "learning_rate": 7.772908636365927e-07,
      "loss": 0.6831,
      "step": 20679
    },
    {
      "epoch": 3.8251254413677755,
      "grad_norm": 1.0311388969421387,
      "learning_rate": 7.754739311158887e-07,
      "loss": 0.7844,
      "step": 20680
    },
    {
      "epoch": 3.825311280431147,
      "grad_norm": 1.0050128698349,
      "learning_rate": 7.736591163724272e-07,
      "loss": 0.8631,
      "step": 20681
    },
    {
      "epoch": 3.8254971194945178,
      "grad_norm": 0.9121170043945312,
      "learning_rate": 7.718464194449326e-07,
      "loss": 0.8428,
      "step": 20682
    },
    {
      "epoch": 3.8256829585578886,
      "grad_norm": 1.0084474086761475,
      "learning_rate": 7.700358403720964e-07,
      "loss": 0.6792,
      "step": 20683
    },
    {
      "epoch": 3.82586879762126,
      "grad_norm": 1.1749839782714844,
      "learning_rate": 7.682273791925543e-07,
      "loss": 0.7773,
      "step": 20684
    },
    {
      "epoch": 3.8260546366846313,
      "grad_norm": 0.960232138633728,
      "learning_rate": 7.664210359449086e-07,
      "loss": 0.8197,
      "step": 20685
    },
    {
      "epoch": 3.826240475748002,
      "grad_norm": 0.980080783367157,
      "learning_rate": 7.646168106677287e-07,
      "loss": 0.8017,
      "step": 20686
    },
    {
      "epoch": 3.826426314811373,
      "grad_norm": 0.9059486389160156,
      "learning_rate": 7.628147033994837e-07,
      "loss": 0.9325,
      "step": 20687
    },
    {
      "epoch": 3.8266121538747444,
      "grad_norm": 1.0272023677825928,
      "learning_rate": 7.610147141786761e-07,
      "loss": 0.8221,
      "step": 20688
    },
    {
      "epoch": 3.8267979929381157,
      "grad_norm": 0.9386231303215027,
      "learning_rate": 7.592168430437086e-07,
      "loss": 0.7834,
      "step": 20689
    },
    {
      "epoch": 3.8269838320014866,
      "grad_norm": 0.9377012252807617,
      "learning_rate": 7.574210900329503e-07,
      "loss": 0.8522,
      "step": 20690
    },
    {
      "epoch": 3.827169671064858,
      "grad_norm": 0.9702726006507874,
      "learning_rate": 7.556274551847264e-07,
      "loss": 0.8191,
      "step": 20691
    },
    {
      "epoch": 3.827355510128229,
      "grad_norm": 0.8971997499465942,
      "learning_rate": 7.538359385373395e-07,
      "loss": 0.6999,
      "step": 20692
    },
    {
      "epoch": 3.8275413491916,
      "grad_norm": 0.9465587139129639,
      "learning_rate": 7.520465401290033e-07,
      "loss": 0.8327,
      "step": 20693
    },
    {
      "epoch": 3.827727188254971,
      "grad_norm": 1.0489004850387573,
      "learning_rate": 7.502592599979208e-07,
      "loss": 0.6599,
      "step": 20694
    },
    {
      "epoch": 3.8279130273183424,
      "grad_norm": 1.0122950077056885,
      "learning_rate": 7.48474098182228e-07,
      "loss": 0.7954,
      "step": 20695
    },
    {
      "epoch": 3.8280988663817137,
      "grad_norm": 0.8980873227119446,
      "learning_rate": 7.466910547200279e-07,
      "loss": 0.7559,
      "step": 20696
    },
    {
      "epoch": 3.8282847054450846,
      "grad_norm": 0.889599084854126,
      "learning_rate": 7.44910129649401e-07,
      "loss": 0.8781,
      "step": 20697
    },
    {
      "epoch": 3.8284705445084555,
      "grad_norm": 1.1924166679382324,
      "learning_rate": 7.431313230083281e-07,
      "loss": 0.9235,
      "step": 20698
    },
    {
      "epoch": 3.828656383571827,
      "grad_norm": 0.8644866347312927,
      "learning_rate": 7.413546348347789e-07,
      "loss": 0.7679,
      "step": 20699
    },
    {
      "epoch": 3.828842222635198,
      "grad_norm": 1.0182651281356812,
      "learning_rate": 7.395800651666785e-07,
      "loss": 0.9471,
      "step": 20700
    },
    {
      "epoch": 3.829028061698569,
      "grad_norm": 0.887749195098877,
      "learning_rate": 7.378076140419187e-07,
      "loss": 0.8044,
      "step": 20701
    },
    {
      "epoch": 3.82921390076194,
      "grad_norm": 0.9445759654045105,
      "learning_rate": 7.360372814983141e-07,
      "loss": 0.7114,
      "step": 20702
    },
    {
      "epoch": 3.8293997398253112,
      "grad_norm": 1.3595925569534302,
      "learning_rate": 7.342690675736341e-07,
      "loss": 0.8659,
      "step": 20703
    },
    {
      "epoch": 3.8295855788886826,
      "grad_norm": 1.240745186805725,
      "learning_rate": 7.325029723056487e-07,
      "loss": 0.8501,
      "step": 20704
    },
    {
      "epoch": 3.8297714179520534,
      "grad_norm": 1.0776466131210327,
      "learning_rate": 7.307389957320276e-07,
      "loss": 1.0036,
      "step": 20705
    },
    {
      "epoch": 3.8299572570154248,
      "grad_norm": 0.8508837819099426,
      "learning_rate": 7.289771378904408e-07,
      "loss": 0.8033,
      "step": 20706
    },
    {
      "epoch": 3.8301430960787957,
      "grad_norm": 0.8653962016105652,
      "learning_rate": 7.272173988184694e-07,
      "loss": 0.8732,
      "step": 20707
    },
    {
      "epoch": 3.830328935142167,
      "grad_norm": 0.8358247876167297,
      "learning_rate": 7.254597785536943e-07,
      "loss": 0.8679,
      "step": 20708
    },
    {
      "epoch": 3.830514774205538,
      "grad_norm": 0.983256995677948,
      "learning_rate": 7.237042771336078e-07,
      "loss": 1.0119,
      "step": 20709
    },
    {
      "epoch": 3.830700613268909,
      "grad_norm": 0.9095485806465149,
      "learning_rate": 7.219508945956909e-07,
      "loss": 0.7091,
      "step": 20710
    },
    {
      "epoch": 3.83088645233228,
      "grad_norm": 0.7739748954772949,
      "learning_rate": 7.201996309773695e-07,
      "loss": 0.7629,
      "step": 20711
    },
    {
      "epoch": 3.8310722913956514,
      "grad_norm": 0.9428021907806396,
      "learning_rate": 7.184504863160135e-07,
      "loss": 0.7273,
      "step": 20712
    },
    {
      "epoch": 3.8312581304590223,
      "grad_norm": 0.9342936277389526,
      "learning_rate": 7.167034606489598e-07,
      "loss": 0.7602,
      "step": 20713
    },
    {
      "epoch": 3.8314439695223936,
      "grad_norm": 0.8526061177253723,
      "learning_rate": 7.149585540134896e-07,
      "loss": 0.629,
      "step": 20714
    },
    {
      "epoch": 3.831629808585765,
      "grad_norm": 0.9603124260902405,
      "learning_rate": 7.132157664468509e-07,
      "loss": 0.677,
      "step": 20715
    },
    {
      "epoch": 3.831815647649136,
      "grad_norm": 0.8807042837142944,
      "learning_rate": 7.114750979862472e-07,
      "loss": 0.8594,
      "step": 20716
    },
    {
      "epoch": 3.8320014867125067,
      "grad_norm": 1.1333434581756592,
      "learning_rate": 7.097365486688157e-07,
      "loss": 0.8751,
      "step": 20717
    },
    {
      "epoch": 3.832187325775878,
      "grad_norm": 1.110044240951538,
      "learning_rate": 7.080001185316599e-07,
      "loss": 1.0981,
      "step": 20718
    },
    {
      "epoch": 3.8323731648392494,
      "grad_norm": 0.9268306493759155,
      "learning_rate": 7.062658076118611e-07,
      "loss": 0.8715,
      "step": 20719
    },
    {
      "epoch": 3.8325590039026203,
      "grad_norm": 0.9917341470718384,
      "learning_rate": 7.045336159464122e-07,
      "loss": 0.7835,
      "step": 20720
    },
    {
      "epoch": 3.8327448429659916,
      "grad_norm": 0.9318150281906128,
      "learning_rate": 7.028035435723058e-07,
      "loss": 0.7383,
      "step": 20721
    },
    {
      "epoch": 3.8329306820293625,
      "grad_norm": 0.8739350438117981,
      "learning_rate": 7.010755905264565e-07,
      "loss": 0.8592,
      "step": 20722
    },
    {
      "epoch": 3.833116521092734,
      "grad_norm": 0.9437780380249023,
      "learning_rate": 6.993497568457352e-07,
      "loss": 0.9203,
      "step": 20723
    },
    {
      "epoch": 3.8333023601561047,
      "grad_norm": 0.8102313876152039,
      "learning_rate": 6.9762604256699e-07,
      "loss": 0.5722,
      "step": 20724
    },
    {
      "epoch": 3.833488199219476,
      "grad_norm": 0.8980199098587036,
      "learning_rate": 6.959044477270138e-07,
      "loss": 0.7649,
      "step": 20725
    },
    {
      "epoch": 3.833674038282847,
      "grad_norm": 0.9712707996368408,
      "learning_rate": 6.941849723625437e-07,
      "loss": 1.068,
      "step": 20726
    },
    {
      "epoch": 3.8338598773462182,
      "grad_norm": 1.315290093421936,
      "learning_rate": 6.924676165102729e-07,
      "loss": 0.9794,
      "step": 20727
    },
    {
      "epoch": 3.834045716409589,
      "grad_norm": 0.8967644572257996,
      "learning_rate": 6.907523802068716e-07,
      "loss": 0.9972,
      "step": 20728
    },
    {
      "epoch": 3.8342315554729605,
      "grad_norm": 0.8474511504173279,
      "learning_rate": 6.890392634889332e-07,
      "loss": 0.8099,
      "step": 20729
    },
    {
      "epoch": 3.834417394536332,
      "grad_norm": 0.928673505783081,
      "learning_rate": 6.873282663930281e-07,
      "loss": 0.7815,
      "step": 20730
    },
    {
      "epoch": 3.8346032335997027,
      "grad_norm": 0.9721923470497131,
      "learning_rate": 6.856193889556717e-07,
      "loss": 0.6947,
      "step": 20731
    },
    {
      "epoch": 3.8347890726630736,
      "grad_norm": 1.0192464590072632,
      "learning_rate": 6.839126312133459e-07,
      "loss": 0.7105,
      "step": 20732
    },
    {
      "epoch": 3.834974911726445,
      "grad_norm": 0.8009982705116272,
      "learning_rate": 6.822079932024661e-07,
      "loss": 0.7098,
      "step": 20733
    },
    {
      "epoch": 3.835160750789816,
      "grad_norm": 0.9369624257087708,
      "learning_rate": 6.805054749594253e-07,
      "loss": 0.9278,
      "step": 20734
    },
    {
      "epoch": 3.835346589853187,
      "grad_norm": 0.8212448358535767,
      "learning_rate": 6.7880507652055e-07,
      "loss": 0.6242,
      "step": 20735
    },
    {
      "epoch": 3.835532428916558,
      "grad_norm": 0.9031669497489929,
      "learning_rate": 6.771067979221335e-07,
      "loss": 0.7569,
      "step": 20736
    },
    {
      "epoch": 3.8357182679799293,
      "grad_norm": 1.0682936906814575,
      "learning_rate": 6.754106392004467e-07,
      "loss": 0.8424,
      "step": 20737
    },
    {
      "epoch": 3.8359041070433006,
      "grad_norm": 1.0823919773101807,
      "learning_rate": 6.737166003916606e-07,
      "loss": 0.899,
      "step": 20738
    },
    {
      "epoch": 3.8360899461066715,
      "grad_norm": 1.1261800527572632,
      "learning_rate": 6.720246815319354e-07,
      "loss": 0.8051,
      "step": 20739
    },
    {
      "epoch": 3.836275785170043,
      "grad_norm": 0.9055033326148987,
      "learning_rate": 6.703348826573974e-07,
      "loss": 0.8597,
      "step": 20740
    },
    {
      "epoch": 3.8364616242334137,
      "grad_norm": 0.9246683716773987,
      "learning_rate": 6.68647203804107e-07,
      "loss": 0.7764,
      "step": 20741
    },
    {
      "epoch": 3.836647463296785,
      "grad_norm": 0.8555891513824463,
      "learning_rate": 6.669616450080796e-07,
      "loss": 0.7487,
      "step": 20742
    },
    {
      "epoch": 3.836833302360156,
      "grad_norm": 0.8516334295272827,
      "learning_rate": 6.652782063052865e-07,
      "loss": 0.8647,
      "step": 20743
    },
    {
      "epoch": 3.8370191414235273,
      "grad_norm": 0.9311104416847229,
      "learning_rate": 6.635968877316767e-07,
      "loss": 0.6996,
      "step": 20744
    },
    {
      "epoch": 3.8372049804868986,
      "grad_norm": 1.1723272800445557,
      "learning_rate": 6.619176893231215e-07,
      "loss": 0.9617,
      "step": 20745
    },
    {
      "epoch": 3.8373908195502695,
      "grad_norm": 1.2130335569381714,
      "learning_rate": 6.602406111154591e-07,
      "loss": 1.0183,
      "step": 20746
    },
    {
      "epoch": 3.8375766586136404,
      "grad_norm": 0.856591522693634,
      "learning_rate": 6.585656531444828e-07,
      "loss": 0.7832,
      "step": 20747
    },
    {
      "epoch": 3.8377624976770117,
      "grad_norm": 0.9378266334533691,
      "learning_rate": 6.56892815445953e-07,
      "loss": 0.9207,
      "step": 20748
    },
    {
      "epoch": 3.837948336740383,
      "grad_norm": 1.0121698379516602,
      "learning_rate": 6.552220980555635e-07,
      "loss": 0.906,
      "step": 20749
    },
    {
      "epoch": 3.838134175803754,
      "grad_norm": 1.2572370767593384,
      "learning_rate": 6.535535010089744e-07,
      "loss": 0.5664,
      "step": 20750
    },
    {
      "epoch": 3.838320014867125,
      "grad_norm": 0.9440824389457703,
      "learning_rate": 6.518870243418018e-07,
      "loss": 0.7313,
      "step": 20751
    },
    {
      "epoch": 3.838505853930496,
      "grad_norm": 0.7588942050933838,
      "learning_rate": 6.502226680896062e-07,
      "loss": 0.8024,
      "step": 20752
    },
    {
      "epoch": 3.8386916929938675,
      "grad_norm": 1.0178306102752686,
      "learning_rate": 6.485604322879257e-07,
      "loss": 0.8894,
      "step": 20753
    },
    {
      "epoch": 3.8388775320572384,
      "grad_norm": 0.9385819435119629,
      "learning_rate": 6.469003169722321e-07,
      "loss": 0.5797,
      "step": 20754
    },
    {
      "epoch": 3.8390633711206097,
      "grad_norm": 0.9427735805511475,
      "learning_rate": 6.452423221779414e-07,
      "loss": 0.6894,
      "step": 20755
    },
    {
      "epoch": 3.8392492101839806,
      "grad_norm": 0.9153196215629578,
      "learning_rate": 6.435864479404696e-07,
      "loss": 0.6369,
      "step": 20756
    },
    {
      "epoch": 3.839435049247352,
      "grad_norm": 1.2121299505233765,
      "learning_rate": 6.419326942951331e-07,
      "loss": 0.765,
      "step": 20757
    },
    {
      "epoch": 3.839620888310723,
      "grad_norm": 1.0967673063278198,
      "learning_rate": 6.40281061277237e-07,
      "loss": 1.0784,
      "step": 20758
    },
    {
      "epoch": 3.839806727374094,
      "grad_norm": 0.9232576489448547,
      "learning_rate": 6.386315489220418e-07,
      "loss": 0.9331,
      "step": 20759
    },
    {
      "epoch": 3.839992566437465,
      "grad_norm": 0.9867680072784424,
      "learning_rate": 6.369841572647306e-07,
      "loss": 0.7843,
      "step": 20760
    },
    {
      "epoch": 3.8401784055008363,
      "grad_norm": 0.8611258864402771,
      "learning_rate": 6.353388863404974e-07,
      "loss": 0.7441,
      "step": 20761
    },
    {
      "epoch": 3.840364244564207,
      "grad_norm": 0.837388277053833,
      "learning_rate": 6.336957361844254e-07,
      "loss": 0.7768,
      "step": 20762
    },
    {
      "epoch": 3.8405500836275785,
      "grad_norm": 1.0685689449310303,
      "learning_rate": 6.320547068315974e-07,
      "loss": 0.8462,
      "step": 20763
    },
    {
      "epoch": 3.84073592269095,
      "grad_norm": 1.0068045854568481,
      "learning_rate": 6.304157983170411e-07,
      "loss": 0.9256,
      "step": 20764
    },
    {
      "epoch": 3.8409217617543208,
      "grad_norm": 0.9282239675521851,
      "learning_rate": 6.287790106757396e-07,
      "loss": 0.8004,
      "step": 20765
    },
    {
      "epoch": 3.8411076008176916,
      "grad_norm": 1.6104170083999634,
      "learning_rate": 6.271443439426205e-07,
      "loss": 0.8226,
      "step": 20766
    },
    {
      "epoch": 3.841293439881063,
      "grad_norm": 0.8550835847854614,
      "learning_rate": 6.255117981525782e-07,
      "loss": 0.7964,
      "step": 20767
    },
    {
      "epoch": 3.8414792789444343,
      "grad_norm": 1.3087283372879028,
      "learning_rate": 6.238813733404514e-07,
      "loss": 1.0183,
      "step": 20768
    },
    {
      "epoch": 3.841665118007805,
      "grad_norm": 0.9228026270866394,
      "learning_rate": 6.222530695410344e-07,
      "loss": 0.6845,
      "step": 20769
    },
    {
      "epoch": 3.8418509570711765,
      "grad_norm": 1.0402220487594604,
      "learning_rate": 6.206268867890997e-07,
      "loss": 0.8275,
      "step": 20770
    },
    {
      "epoch": 3.8420367961345474,
      "grad_norm": 0.86312335729599,
      "learning_rate": 6.190028251193303e-07,
      "loss": 0.6399,
      "step": 20771
    },
    {
      "epoch": 3.8422226351979187,
      "grad_norm": 0.9186187386512756,
      "learning_rate": 6.173808845663987e-07,
      "loss": 0.6913,
      "step": 20772
    },
    {
      "epoch": 3.8424084742612896,
      "grad_norm": 0.7808519005775452,
      "learning_rate": 6.157610651649326e-07,
      "loss": 0.7809,
      "step": 20773
    },
    {
      "epoch": 3.842594313324661,
      "grad_norm": 0.8238528966903687,
      "learning_rate": 6.141433669494934e-07,
      "loss": 0.6946,
      "step": 20774
    },
    {
      "epoch": 3.842780152388032,
      "grad_norm": 0.8859274983406067,
      "learning_rate": 6.12527789954609e-07,
      "loss": 0.7788,
      "step": 20775
    },
    {
      "epoch": 3.842965991451403,
      "grad_norm": 0.8922582864761353,
      "learning_rate": 6.109143342147627e-07,
      "loss": 0.7511,
      "step": 20776
    },
    {
      "epoch": 3.843151830514774,
      "grad_norm": 0.8180863261222839,
      "learning_rate": 6.09302999764394e-07,
      "loss": 0.8641,
      "step": 20777
    },
    {
      "epoch": 3.8433376695781454,
      "grad_norm": 0.9965726733207703,
      "learning_rate": 6.076937866378974e-07,
      "loss": 0.6953,
      "step": 20778
    },
    {
      "epoch": 3.8435235086415167,
      "grad_norm": 0.9308102130889893,
      "learning_rate": 6.060866948696009e-07,
      "loss": 0.6541,
      "step": 20779
    },
    {
      "epoch": 3.8437093477048876,
      "grad_norm": 0.9346275925636292,
      "learning_rate": 6.044817244938328e-07,
      "loss": 0.9128,
      "step": 20780
    },
    {
      "epoch": 3.8438951867682585,
      "grad_norm": 0.8471858501434326,
      "learning_rate": 6.028788755448211e-07,
      "loss": 1.1086,
      "step": 20781
    },
    {
      "epoch": 3.84408102583163,
      "grad_norm": 0.9544927477836609,
      "learning_rate": 6.012781480567831e-07,
      "loss": 0.9041,
      "step": 20782
    },
    {
      "epoch": 3.844266864895001,
      "grad_norm": 0.8701556324958801,
      "learning_rate": 5.996795420638912e-07,
      "loss": 0.7266,
      "step": 20783
    },
    {
      "epoch": 3.844452703958372,
      "grad_norm": 1.0673178434371948,
      "learning_rate": 5.980830576002738e-07,
      "loss": 0.9692,
      "step": 20784
    },
    {
      "epoch": 3.844638543021743,
      "grad_norm": 1.1484763622283936,
      "learning_rate": 5.964886946999815e-07,
      "loss": 0.9399,
      "step": 20785
    },
    {
      "epoch": 3.8448243820851142,
      "grad_norm": 0.7817178964614868,
      "learning_rate": 5.948964533970647e-07,
      "loss": 0.8361,
      "step": 20786
    },
    {
      "epoch": 3.8450102211484856,
      "grad_norm": 1.080405354499817,
      "learning_rate": 5.933063337254852e-07,
      "loss": 0.7481,
      "step": 20787
    },
    {
      "epoch": 3.8451960602118564,
      "grad_norm": 1.027951955795288,
      "learning_rate": 5.917183357191935e-07,
      "loss": 0.6884,
      "step": 20788
    },
    {
      "epoch": 3.8453818992752278,
      "grad_norm": 0.9318289756774902,
      "learning_rate": 5.90132459412096e-07,
      "loss": 0.8408,
      "step": 20789
    },
    {
      "epoch": 3.8455677383385987,
      "grad_norm": 0.8515191078186035,
      "learning_rate": 5.885487048380101e-07,
      "loss": 0.7485,
      "step": 20790
    },
    {
      "epoch": 3.84575357740197,
      "grad_norm": 0.814934253692627,
      "learning_rate": 5.869670720307641e-07,
      "loss": 0.7777,
      "step": 20791
    },
    {
      "epoch": 3.845939416465341,
      "grad_norm": 0.7820918560028076,
      "learning_rate": 5.853875610241088e-07,
      "loss": 0.6791,
      "step": 20792
    },
    {
      "epoch": 3.846125255528712,
      "grad_norm": 1.0850684642791748,
      "learning_rate": 5.838101718517508e-07,
      "loss": 0.7751,
      "step": 20793
    },
    {
      "epoch": 3.846311094592083,
      "grad_norm": 0.90702885389328,
      "learning_rate": 5.822349045473518e-07,
      "loss": 1.1127,
      "step": 20794
    },
    {
      "epoch": 3.8464969336554544,
      "grad_norm": 0.9162556529045105,
      "learning_rate": 5.806617591445518e-07,
      "loss": 0.8811,
      "step": 20795
    },
    {
      "epoch": 3.8466827727188253,
      "grad_norm": 1.4018508195877075,
      "learning_rate": 5.790907356769127e-07,
      "loss": 1.1379,
      "step": 20796
    },
    {
      "epoch": 3.8468686117821966,
      "grad_norm": 1.215028166770935,
      "learning_rate": 5.775218341779742e-07,
      "loss": 0.9514,
      "step": 20797
    },
    {
      "epoch": 3.847054450845568,
      "grad_norm": 0.8906833529472351,
      "learning_rate": 5.759550546812098e-07,
      "loss": 0.8951,
      "step": 20798
    },
    {
      "epoch": 3.847240289908939,
      "grad_norm": 0.8343894481658936,
      "learning_rate": 5.743903972200592e-07,
      "loss": 0.9802,
      "step": 20799
    },
    {
      "epoch": 3.8474261289723097,
      "grad_norm": 0.9710038304328918,
      "learning_rate": 5.728278618279404e-07,
      "loss": 0.8445,
      "step": 20800
    },
    {
      "epoch": 3.847611968035681,
      "grad_norm": 0.8098582029342651,
      "learning_rate": 5.712674485381819e-07,
      "loss": 0.6027,
      "step": 20801
    },
    {
      "epoch": 3.8477978070990524,
      "grad_norm": 0.9980126023292542,
      "learning_rate": 5.697091573840796e-07,
      "loss": 0.8672,
      "step": 20802
    },
    {
      "epoch": 3.8479836461624233,
      "grad_norm": 0.9283662438392639,
      "learning_rate": 5.681529883989178e-07,
      "loss": 0.6205,
      "step": 20803
    },
    {
      "epoch": 3.8481694852257946,
      "grad_norm": 1.1998335123062134,
      "learning_rate": 5.665989416159035e-07,
      "loss": 0.7646,
      "step": 20804
    },
    {
      "epoch": 3.8483553242891655,
      "grad_norm": 0.8913629055023193,
      "learning_rate": 5.650470170681876e-07,
      "loss": 0.7736,
      "step": 20805
    },
    {
      "epoch": 3.848541163352537,
      "grad_norm": 0.8077051043510437,
      "learning_rate": 5.634972147889217e-07,
      "loss": 0.7823,
      "step": 20806
    },
    {
      "epoch": 3.8487270024159077,
      "grad_norm": 0.9425461888313293,
      "learning_rate": 5.61949534811157e-07,
      "loss": 0.793,
      "step": 20807
    },
    {
      "epoch": 3.848912841479279,
      "grad_norm": 0.9253484606742859,
      "learning_rate": 5.604039771679337e-07,
      "loss": 0.917,
      "step": 20808
    },
    {
      "epoch": 3.84909868054265,
      "grad_norm": 0.9143958687782288,
      "learning_rate": 5.588605418922476e-07,
      "loss": 0.8883,
      "step": 20809
    },
    {
      "epoch": 3.8492845196060212,
      "grad_norm": 0.9746808409690857,
      "learning_rate": 5.573192290170393e-07,
      "loss": 0.9594,
      "step": 20810
    },
    {
      "epoch": 3.849470358669392,
      "grad_norm": 0.9156635403633118,
      "learning_rate": 5.557800385751932e-07,
      "loss": 0.7471,
      "step": 20811
    },
    {
      "epoch": 3.8496561977327635,
      "grad_norm": 0.9907944202423096,
      "learning_rate": 5.542429705995833e-07,
      "loss": 0.9115,
      "step": 20812
    },
    {
      "epoch": 3.849842036796135,
      "grad_norm": 1.0210583209991455,
      "learning_rate": 5.527080251229833e-07,
      "loss": 0.7514,
      "step": 20813
    },
    {
      "epoch": 3.8500278758595057,
      "grad_norm": 0.9511533975601196,
      "learning_rate": 5.51175202178178e-07,
      "loss": 0.8369,
      "step": 20814
    },
    {
      "epoch": 3.8502137149228766,
      "grad_norm": 0.9177201390266418,
      "learning_rate": 5.496445017978857e-07,
      "loss": 0.9094,
      "step": 20815
    },
    {
      "epoch": 3.850399553986248,
      "grad_norm": 0.7297847867012024,
      "learning_rate": 5.481159240147582e-07,
      "loss": 0.5455,
      "step": 20816
    },
    {
      "epoch": 3.850585393049619,
      "grad_norm": 1.1877726316452026,
      "learning_rate": 5.465894688614359e-07,
      "loss": 1.014,
      "step": 20817
    },
    {
      "epoch": 3.85077123211299,
      "grad_norm": 0.9924846887588501,
      "learning_rate": 5.450651363704929e-07,
      "loss": 0.6386,
      "step": 20818
    },
    {
      "epoch": 3.8509570711763614,
      "grad_norm": 0.8977491855621338,
      "learning_rate": 5.435429265744585e-07,
      "loss": 0.749,
      "step": 20819
    },
    {
      "epoch": 3.8511429102397323,
      "grad_norm": 0.9722229838371277,
      "learning_rate": 5.420228395058291e-07,
      "loss": 0.8737,
      "step": 20820
    },
    {
      "epoch": 3.8513287493031036,
      "grad_norm": 1.1354950666427612,
      "learning_rate": 5.405048751970454e-07,
      "loss": 0.8821,
      "step": 20821
    },
    {
      "epoch": 3.8515145883664745,
      "grad_norm": 0.8959169983863831,
      "learning_rate": 5.389890336805037e-07,
      "loss": 1.0039,
      "step": 20822
    },
    {
      "epoch": 3.851700427429846,
      "grad_norm": 0.8557883501052856,
      "learning_rate": 5.374753149885558e-07,
      "loss": 0.7533,
      "step": 20823
    },
    {
      "epoch": 3.8518862664932167,
      "grad_norm": 1.0070464611053467,
      "learning_rate": 5.359637191535205e-07,
      "loss": 0.7772,
      "step": 20824
    },
    {
      "epoch": 3.852072105556588,
      "grad_norm": 0.9542919397354126,
      "learning_rate": 5.344542462076496e-07,
      "loss": 0.6336,
      "step": 20825
    },
    {
      "epoch": 3.852257944619959,
      "grad_norm": 0.988021731376648,
      "learning_rate": 5.329468961831619e-07,
      "loss": 0.9381,
      "step": 20826
    },
    {
      "epoch": 3.8524437836833303,
      "grad_norm": 0.8486770391464233,
      "learning_rate": 5.314416691122204e-07,
      "loss": 0.8043,
      "step": 20827
    },
    {
      "epoch": 3.8526296227467016,
      "grad_norm": 0.847549557685852,
      "learning_rate": 5.299385650269773e-07,
      "loss": 1.0205,
      "step": 20828
    },
    {
      "epoch": 3.8528154618100725,
      "grad_norm": 0.9774998426437378,
      "learning_rate": 5.284375839594957e-07,
      "loss": 0.8089,
      "step": 20829
    },
    {
      "epoch": 3.8530013008734434,
      "grad_norm": 1.0979493856430054,
      "learning_rate": 5.269387259418168e-07,
      "loss": 0.7921,
      "step": 20830
    },
    {
      "epoch": 3.8531871399368147,
      "grad_norm": 0.8957627415657043,
      "learning_rate": 5.25441991005915e-07,
      "loss": 0.801,
      "step": 20831
    },
    {
      "epoch": 3.853372979000186,
      "grad_norm": 0.8519086241722107,
      "learning_rate": 5.239473791837534e-07,
      "loss": 0.6381,
      "step": 20832
    },
    {
      "epoch": 3.853558818063557,
      "grad_norm": 0.9422652721405029,
      "learning_rate": 5.224548905072402e-07,
      "loss": 0.876,
      "step": 20833
    },
    {
      "epoch": 3.853744657126928,
      "grad_norm": 1.0043606758117676,
      "learning_rate": 5.20964525008194e-07,
      "loss": 0.8307,
      "step": 20834
    },
    {
      "epoch": 3.853930496190299,
      "grad_norm": 0.8691344857215881,
      "learning_rate": 5.194762827184673e-07,
      "loss": 0.8215,
      "step": 20835
    },
    {
      "epoch": 3.8541163352536705,
      "grad_norm": 0.9490212202072144,
      "learning_rate": 5.179901636697904e-07,
      "loss": 0.7445,
      "step": 20836
    },
    {
      "epoch": 3.8543021743170414,
      "grad_norm": 1.0769680738449097,
      "learning_rate": 5.165061678939042e-07,
      "loss": 1.0122,
      "step": 20837
    },
    {
      "epoch": 3.8544880133804127,
      "grad_norm": 0.8654699325561523,
      "learning_rate": 5.150242954224727e-07,
      "loss": 0.655,
      "step": 20838
    },
    {
      "epoch": 3.8546738524437836,
      "grad_norm": 0.9956970810890198,
      "learning_rate": 5.135445462871258e-07,
      "loss": 0.8539,
      "step": 20839
    },
    {
      "epoch": 3.854859691507155,
      "grad_norm": 0.9140504002571106,
      "learning_rate": 5.120669205194384e-07,
      "loss": 0.7781,
      "step": 20840
    },
    {
      "epoch": 3.855045530570526,
      "grad_norm": 0.7998533248901367,
      "learning_rate": 5.10591418150963e-07,
      "loss": 0.7059,
      "step": 20841
    },
    {
      "epoch": 3.855231369633897,
      "grad_norm": 1.070513129234314,
      "learning_rate": 5.091180392131745e-07,
      "loss": 0.7954,
      "step": 20842
    },
    {
      "epoch": 3.855417208697268,
      "grad_norm": 1.021838665008545,
      "learning_rate": 5.076467837375254e-07,
      "loss": 0.9054,
      "step": 20843
    },
    {
      "epoch": 3.8556030477606393,
      "grad_norm": 0.8108730316162109,
      "learning_rate": 5.061776517554351e-07,
      "loss": 0.625,
      "step": 20844
    },
    {
      "epoch": 3.85578888682401,
      "grad_norm": 1.0125364065170288,
      "learning_rate": 5.04710643298234e-07,
      "loss": 0.8624,
      "step": 20845
    },
    {
      "epoch": 3.8559747258873815,
      "grad_norm": 1.2414863109588623,
      "learning_rate": 5.032457583972416e-07,
      "loss": 1.0397,
      "step": 20846
    },
    {
      "epoch": 3.856160564950753,
      "grad_norm": 0.8799111247062683,
      "learning_rate": 5.017829970837329e-07,
      "loss": 0.6313,
      "step": 20847
    },
    {
      "epoch": 3.8563464040141238,
      "grad_norm": 1.1502881050109863,
      "learning_rate": 5.003223593889161e-07,
      "loss": 0.801,
      "step": 20848
    },
    {
      "epoch": 3.8565322430774946,
      "grad_norm": 0.8884891271591187,
      "learning_rate": 4.988638453439665e-07,
      "loss": 0.9028,
      "step": 20849
    },
    {
      "epoch": 3.856718082140866,
      "grad_norm": 1.1250226497650146,
      "learning_rate": 4.974074549800256e-07,
      "loss": 0.8561,
      "step": 20850
    },
    {
      "epoch": 3.8569039212042373,
      "grad_norm": 0.9480788111686707,
      "learning_rate": 4.959531883281576e-07,
      "loss": 0.8672,
      "step": 20851
    },
    {
      "epoch": 3.857089760267608,
      "grad_norm": 0.9684773683547974,
      "learning_rate": 4.945010454194265e-07,
      "loss": 0.9812,
      "step": 20852
    },
    {
      "epoch": 3.8572755993309795,
      "grad_norm": 0.8317880034446716,
      "learning_rate": 4.930510262848076e-07,
      "loss": 0.6962,
      "step": 20853
    },
    {
      "epoch": 3.8574614383943504,
      "grad_norm": 1.09212064743042,
      "learning_rate": 4.916031309552538e-07,
      "loss": 0.7375,
      "step": 20854
    },
    {
      "epoch": 3.8576472774577217,
      "grad_norm": 1.0218623876571655,
      "learning_rate": 4.901573594616626e-07,
      "loss": 0.7847,
      "step": 20855
    },
    {
      "epoch": 3.8578331165210926,
      "grad_norm": 1.1442655324935913,
      "learning_rate": 4.887137118348983e-07,
      "loss": 0.7224,
      "step": 20856
    },
    {
      "epoch": 3.858018955584464,
      "grad_norm": 0.874628484249115,
      "learning_rate": 4.872721881057806e-07,
      "loss": 0.7401,
      "step": 20857
    },
    {
      "epoch": 3.858204794647835,
      "grad_norm": 0.837985634803772,
      "learning_rate": 4.858327883050628e-07,
      "loss": 0.656,
      "step": 20858
    },
    {
      "epoch": 3.858390633711206,
      "grad_norm": 0.870708703994751,
      "learning_rate": 4.843955124634647e-07,
      "loss": 0.8925,
      "step": 20859
    },
    {
      "epoch": 3.858576472774577,
      "grad_norm": 1.1414045095443726,
      "learning_rate": 4.829603606116728e-07,
      "loss": 0.6571,
      "step": 20860
    },
    {
      "epoch": 3.8587623118379484,
      "grad_norm": 1.0180896520614624,
      "learning_rate": 4.815273327803182e-07,
      "loss": 0.8944,
      "step": 20861
    },
    {
      "epoch": 3.8589481509013197,
      "grad_norm": 1.1275979280471802,
      "learning_rate": 4.800964289999654e-07,
      "loss": 1.0819,
      "step": 20862
    },
    {
      "epoch": 3.8591339899646906,
      "grad_norm": 0.9097642302513123,
      "learning_rate": 4.786676493011899e-07,
      "loss": 0.7079,
      "step": 20863
    },
    {
      "epoch": 3.8593198290280615,
      "grad_norm": 1.1204291582107544,
      "learning_rate": 4.772409937144673e-07,
      "loss": 0.9284,
      "step": 20864
    },
    {
      "epoch": 3.859505668091433,
      "grad_norm": 1.0713176727294922,
      "learning_rate": 4.758164622702399e-07,
      "loss": 1.0045,
      "step": 20865
    },
    {
      "epoch": 3.859691507154804,
      "grad_norm": 0.8504579663276672,
      "learning_rate": 4.7439405499891674e-07,
      "loss": 0.6306,
      "step": 20866
    },
    {
      "epoch": 3.859877346218175,
      "grad_norm": 1.0859054327011108,
      "learning_rate": 4.729737719308625e-07,
      "loss": 0.9759,
      "step": 20867
    },
    {
      "epoch": 3.860063185281546,
      "grad_norm": 0.9498307108879089,
      "learning_rate": 4.7155561309639716e-07,
      "loss": 0.7607,
      "step": 20868
    },
    {
      "epoch": 3.8602490243449172,
      "grad_norm": 0.882290780544281,
      "learning_rate": 4.7013957852577453e-07,
      "loss": 0.814,
      "step": 20869
    },
    {
      "epoch": 3.8604348634082886,
      "grad_norm": 0.8619256615638733,
      "learning_rate": 4.6872566824922584e-07,
      "loss": 0.8288,
      "step": 20870
    },
    {
      "epoch": 3.8606207024716594,
      "grad_norm": 0.9371104836463928,
      "learning_rate": 4.673138822969159e-07,
      "loss": 0.7566,
      "step": 20871
    },
    {
      "epoch": 3.860806541535031,
      "grad_norm": 0.9974023103713989,
      "learning_rate": 4.659042206989872e-07,
      "loss": 0.9281,
      "step": 20872
    },
    {
      "epoch": 3.8609923805984017,
      "grad_norm": 0.9402217268943787,
      "learning_rate": 4.6449668348553797e-07,
      "loss": 0.779,
      "step": 20873
    },
    {
      "epoch": 3.861178219661773,
      "grad_norm": 0.9545169472694397,
      "learning_rate": 4.6309127068657754e-07,
      "loss": 1.095,
      "step": 20874
    },
    {
      "epoch": 3.861364058725144,
      "grad_norm": 0.9999849200248718,
      "learning_rate": 4.616879823321374e-07,
      "loss": 0.8332,
      "step": 20875
    },
    {
      "epoch": 3.861549897788515,
      "grad_norm": 0.811683177947998,
      "learning_rate": 4.6028681845214915e-07,
      "loss": 0.7105,
      "step": 20876
    },
    {
      "epoch": 3.8617357368518865,
      "grad_norm": 0.9631338119506836,
      "learning_rate": 4.5888777907652227e-07,
      "loss": 0.9943,
      "step": 20877
    },
    {
      "epoch": 3.8619215759152574,
      "grad_norm": 0.9109195470809937,
      "learning_rate": 4.574908642351106e-07,
      "loss": 0.893,
      "step": 20878
    },
    {
      "epoch": 3.8621074149786283,
      "grad_norm": 1.31691575050354,
      "learning_rate": 4.5609607395773466e-07,
      "loss": 1.0398,
      "step": 20879
    },
    {
      "epoch": 3.8622932540419996,
      "grad_norm": 0.8035569787025452,
      "learning_rate": 4.547034082741708e-07,
      "loss": 0.493,
      "step": 20880
    },
    {
      "epoch": 3.862479093105371,
      "grad_norm": 0.8431886434555054,
      "learning_rate": 4.5331286721413956e-07,
      "loss": 1.0258,
      "step": 20881
    },
    {
      "epoch": 3.862664932168742,
      "grad_norm": 0.9853758215904236,
      "learning_rate": 4.519244508073062e-07,
      "loss": 0.9369,
      "step": 20882
    },
    {
      "epoch": 3.8628507712321127,
      "grad_norm": 1.0060194730758667,
      "learning_rate": 4.5053815908331353e-07,
      "loss": 0.9941,
      "step": 20883
    },
    {
      "epoch": 3.863036610295484,
      "grad_norm": 1.0062130689620972,
      "learning_rate": 4.4915399207176026e-07,
      "loss": 0.8644,
      "step": 20884
    },
    {
      "epoch": 3.8632224493588554,
      "grad_norm": 0.9900531768798828,
      "learning_rate": 4.477719498021782e-07,
      "loss": 0.9401,
      "step": 20885
    },
    {
      "epoch": 3.8634082884222263,
      "grad_norm": 0.909517228603363,
      "learning_rate": 4.463920323040549e-07,
      "loss": 0.6103,
      "step": 20886
    },
    {
      "epoch": 3.8635941274855976,
      "grad_norm": 0.970980703830719,
      "learning_rate": 4.450142396068557e-07,
      "loss": 1.0352,
      "step": 20887
    },
    {
      "epoch": 3.8637799665489685,
      "grad_norm": 0.9788309931755066,
      "learning_rate": 4.4363857173999045e-07,
      "loss": 0.8198,
      "step": 20888
    },
    {
      "epoch": 3.86396580561234,
      "grad_norm": 0.9254130125045776,
      "learning_rate": 4.422650287328134e-07,
      "loss": 1.0075,
      "step": 20889
    },
    {
      "epoch": 3.8641516446757107,
      "grad_norm": 0.8469333648681641,
      "learning_rate": 4.408936106146344e-07,
      "loss": 0.8859,
      "step": 20890
    },
    {
      "epoch": 3.864337483739082,
      "grad_norm": 1.055687427520752,
      "learning_rate": 4.395243174147301e-07,
      "loss": 0.8075,
      "step": 20891
    },
    {
      "epoch": 3.864523322802453,
      "grad_norm": 1.0496764183044434,
      "learning_rate": 4.3815714916233263e-07,
      "loss": 0.8729,
      "step": 20892
    },
    {
      "epoch": 3.8647091618658242,
      "grad_norm": 1.0249707698822021,
      "learning_rate": 4.3679210588661866e-07,
      "loss": 1.014,
      "step": 20893
    },
    {
      "epoch": 3.864895000929195,
      "grad_norm": 1.0160168409347534,
      "learning_rate": 4.3542918761672044e-07,
      "loss": 0.6459,
      "step": 20894
    },
    {
      "epoch": 3.8650808399925665,
      "grad_norm": 0.9592434763908386,
      "learning_rate": 4.340683943817148e-07,
      "loss": 1.0442,
      "step": 20895
    },
    {
      "epoch": 3.865266679055938,
      "grad_norm": 0.7899896502494812,
      "learning_rate": 4.327097262106672e-07,
      "loss": 0.7655,
      "step": 20896
    },
    {
      "epoch": 3.8654525181193087,
      "grad_norm": 1.0063457489013672,
      "learning_rate": 4.313531831325657e-07,
      "loss": 0.5508,
      "step": 20897
    },
    {
      "epoch": 3.8656383571826796,
      "grad_norm": 0.874465823173523,
      "learning_rate": 4.2999876517636484e-07,
      "loss": 0.762,
      "step": 20898
    },
    {
      "epoch": 3.865824196246051,
      "grad_norm": 1.07059645652771,
      "learning_rate": 4.286464723709638e-07,
      "loss": 0.7365,
      "step": 20899
    },
    {
      "epoch": 3.866010035309422,
      "grad_norm": 0.804924488067627,
      "learning_rate": 4.272963047452394e-07,
      "loss": 0.7519,
      "step": 20900
    },
    {
      "epoch": 3.866195874372793,
      "grad_norm": 1.277814507484436,
      "learning_rate": 4.25948262328002e-07,
      "loss": 0.5872,
      "step": 20901
    },
    {
      "epoch": 3.8663817134361644,
      "grad_norm": 0.8175092339515686,
      "learning_rate": 4.2460234514801743e-07,
      "loss": 0.6297,
      "step": 20902
    },
    {
      "epoch": 3.8665675524995353,
      "grad_norm": 0.9114618301391602,
      "learning_rate": 4.232585532340183e-07,
      "loss": 0.9771,
      "step": 20903
    },
    {
      "epoch": 3.8667533915629066,
      "grad_norm": 0.8819079399108887,
      "learning_rate": 4.2191688661469275e-07,
      "loss": 0.7006,
      "step": 20904
    },
    {
      "epoch": 3.8669392306262775,
      "grad_norm": 0.9290643930435181,
      "learning_rate": 4.2057734531866235e-07,
      "loss": 0.7299,
      "step": 20905
    },
    {
      "epoch": 3.867125069689649,
      "grad_norm": 0.9308085441589355,
      "learning_rate": 4.192399293745264e-07,
      "loss": 0.9075,
      "step": 20906
    },
    {
      "epoch": 3.8673109087530197,
      "grad_norm": 0.9288163781166077,
      "learning_rate": 4.179046388108177e-07,
      "loss": 0.9751,
      "step": 20907
    },
    {
      "epoch": 3.867496747816391,
      "grad_norm": 0.9171245098114014,
      "learning_rate": 4.1657147365605775e-07,
      "loss": 0.7872,
      "step": 20908
    },
    {
      "epoch": 3.867682586879762,
      "grad_norm": 0.7444671392440796,
      "learning_rate": 4.152404339386795e-07,
      "loss": 0.6115,
      "step": 20909
    },
    {
      "epoch": 3.8678684259431333,
      "grad_norm": 1.1080163717269897,
      "learning_rate": 4.1391151968710464e-07,
      "loss": 0.8745,
      "step": 20910
    },
    {
      "epoch": 3.8680542650065046,
      "grad_norm": 1.0134642124176025,
      "learning_rate": 4.1258473092967706e-07,
      "loss": 0.9199,
      "step": 20911
    },
    {
      "epoch": 3.8682401040698755,
      "grad_norm": 1.1440978050231934,
      "learning_rate": 4.11260067694752e-07,
      "loss": 0.7557,
      "step": 20912
    },
    {
      "epoch": 3.8684259431332464,
      "grad_norm": 0.8744987845420837,
      "learning_rate": 4.0993753001056236e-07,
      "loss": 0.8767,
      "step": 20913
    },
    {
      "epoch": 3.8686117821966177,
      "grad_norm": 1.6089298725128174,
      "learning_rate": 4.086171179053633e-07,
      "loss": 1.2797,
      "step": 20914
    },
    {
      "epoch": 3.868797621259989,
      "grad_norm": 0.9238671064376831,
      "learning_rate": 4.0729883140732115e-07,
      "loss": 0.8659,
      "step": 20915
    },
    {
      "epoch": 3.86898346032336,
      "grad_norm": 0.9194700717926025,
      "learning_rate": 4.059826705445802e-07,
      "loss": 0.7047,
      "step": 20916
    },
    {
      "epoch": 3.869169299386731,
      "grad_norm": 0.7992855906486511,
      "learning_rate": 4.0466863534522893e-07,
      "loss": 0.8166,
      "step": 20917
    },
    {
      "epoch": 3.869355138450102,
      "grad_norm": 1.237243890762329,
      "learning_rate": 4.0335672583731167e-07,
      "loss": 0.8881,
      "step": 20918
    },
    {
      "epoch": 3.8695409775134735,
      "grad_norm": 0.99635249376297,
      "learning_rate": 4.020469420488282e-07,
      "loss": 0.6967,
      "step": 20919
    },
    {
      "epoch": 3.8697268165768444,
      "grad_norm": 1.0364038944244385,
      "learning_rate": 4.0073928400773396e-07,
      "loss": 0.7512,
      "step": 20920
    },
    {
      "epoch": 3.8699126556402157,
      "grad_norm": 0.8522542119026184,
      "learning_rate": 3.9943375174195107e-07,
      "loss": 0.8132,
      "step": 20921
    },
    {
      "epoch": 3.8700984947035866,
      "grad_norm": 0.8687629699707031,
      "learning_rate": 3.981303452793239e-07,
      "loss": 0.98,
      "step": 20922
    },
    {
      "epoch": 3.870284333766958,
      "grad_norm": 0.8941859006881714,
      "learning_rate": 3.968290646476858e-07,
      "loss": 0.6673,
      "step": 20923
    },
    {
      "epoch": 3.870470172830329,
      "grad_norm": 0.852587103843689,
      "learning_rate": 3.9552990987480333e-07,
      "loss": 0.8243,
      "step": 20924
    },
    {
      "epoch": 3.8706560118937,
      "grad_norm": 1.0379388332366943,
      "learning_rate": 3.942328809884099e-07,
      "loss": 0.8923,
      "step": 20925
    },
    {
      "epoch": 3.8708418509570715,
      "grad_norm": 0.9077103734016418,
      "learning_rate": 3.929379780161724e-07,
      "loss": 0.7595,
      "step": 20926
    },
    {
      "epoch": 3.8710276900204423,
      "grad_norm": 0.8946455717086792,
      "learning_rate": 3.916452009857574e-07,
      "loss": 0.7433,
      "step": 20927
    },
    {
      "epoch": 3.871213529083813,
      "grad_norm": 1.0783296823501587,
      "learning_rate": 3.9035454992473184e-07,
      "loss": 0.8262,
      "step": 20928
    },
    {
      "epoch": 3.8713993681471845,
      "grad_norm": 1.0029511451721191,
      "learning_rate": 3.8906602486066256e-07,
      "loss": 0.9491,
      "step": 20929
    },
    {
      "epoch": 3.871585207210556,
      "grad_norm": 1.003530740737915,
      "learning_rate": 3.8777962582102757e-07,
      "loss": 0.7677,
      "step": 20930
    },
    {
      "epoch": 3.8717710462739268,
      "grad_norm": 0.9285503625869751,
      "learning_rate": 3.8649535283329376e-07,
      "loss": 0.9863,
      "step": 20931
    },
    {
      "epoch": 3.8719568853372976,
      "grad_norm": 0.8985755443572998,
      "learning_rate": 3.852132059248836e-07,
      "loss": 0.6707,
      "step": 20932
    },
    {
      "epoch": 3.872142724400669,
      "grad_norm": 0.9956530332565308,
      "learning_rate": 3.83933185123142e-07,
      "loss": 0.8795,
      "step": 20933
    },
    {
      "epoch": 3.8723285634640403,
      "grad_norm": 0.8931170701980591,
      "learning_rate": 3.826552904554026e-07,
      "loss": 0.9117,
      "step": 20934
    },
    {
      "epoch": 3.872514402527411,
      "grad_norm": 0.9893044829368591,
      "learning_rate": 3.8137952194893244e-07,
      "loss": 1.0049,
      "step": 20935
    },
    {
      "epoch": 3.8727002415907825,
      "grad_norm": 0.9650073051452637,
      "learning_rate": 3.8010587963097646e-07,
      "loss": 0.8627,
      "step": 20936
    },
    {
      "epoch": 3.8728860806541534,
      "grad_norm": 0.8656110167503357,
      "learning_rate": 3.788343635286906e-07,
      "loss": 0.6881,
      "step": 20937
    },
    {
      "epoch": 3.8730719197175247,
      "grad_norm": 1.0848356485366821,
      "learning_rate": 3.7756497366923103e-07,
      "loss": 0.9858,
      "step": 20938
    },
    {
      "epoch": 3.8732577587808956,
      "grad_norm": 0.9150021076202393,
      "learning_rate": 3.762977100796983e-07,
      "loss": 0.7277,
      "step": 20939
    },
    {
      "epoch": 3.873443597844267,
      "grad_norm": 1.00950026512146,
      "learning_rate": 3.7503257278712623e-07,
      "loss": 0.857,
      "step": 20940
    },
    {
      "epoch": 3.873629436907638,
      "grad_norm": 1.1124643087387085,
      "learning_rate": 3.737695618185155e-07,
      "loss": 0.7612,
      "step": 20941
    },
    {
      "epoch": 3.873815275971009,
      "grad_norm": 0.854637086391449,
      "learning_rate": 3.7250867720082236e-07,
      "loss": 0.7645,
      "step": 20942
    },
    {
      "epoch": 3.87400111503438,
      "grad_norm": 0.7971043586730957,
      "learning_rate": 3.712499189609808e-07,
      "loss": 0.7094,
      "step": 20943
    },
    {
      "epoch": 3.8741869540977514,
      "grad_norm": 0.8815410137176514,
      "learning_rate": 3.6999328712582495e-07,
      "loss": 0.9448,
      "step": 20944
    },
    {
      "epoch": 3.8743727931611227,
      "grad_norm": 1.0783112049102783,
      "learning_rate": 3.687387817221999e-07,
      "loss": 0.8899,
      "step": 20945
    },
    {
      "epoch": 3.8745586322244936,
      "grad_norm": 0.8870981931686401,
      "learning_rate": 3.674864027768621e-07,
      "loss": 0.5712,
      "step": 20946
    },
    {
      "epoch": 3.8747444712878645,
      "grad_norm": 0.9387726187705994,
      "learning_rate": 3.662361503165679e-07,
      "loss": 0.8228,
      "step": 20947
    },
    {
      "epoch": 3.874930310351236,
      "grad_norm": 1.0043076276779175,
      "learning_rate": 3.6498802436797374e-07,
      "loss": 0.6124,
      "step": 20948
    },
    {
      "epoch": 3.875116149414607,
      "grad_norm": 1.0434253215789795,
      "learning_rate": 3.6374202495772505e-07,
      "loss": 0.8393,
      "step": 20949
    },
    {
      "epoch": 3.875301988477978,
      "grad_norm": 0.9208751916885376,
      "learning_rate": 3.6249815211242265e-07,
      "loss": 0.9704,
      "step": 20950
    },
    {
      "epoch": 3.8754878275413494,
      "grad_norm": 0.7848434448242188,
      "learning_rate": 3.612564058586121e-07,
      "loss": 0.6018,
      "step": 20951
    },
    {
      "epoch": 3.8756736666047202,
      "grad_norm": 0.9232147932052612,
      "learning_rate": 3.6001678622280546e-07,
      "loss": 0.9302,
      "step": 20952
    },
    {
      "epoch": 3.8758595056680916,
      "grad_norm": 0.7875188589096069,
      "learning_rate": 3.587792932314371e-07,
      "loss": 0.6236,
      "step": 20953
    },
    {
      "epoch": 3.8760453447314624,
      "grad_norm": 0.9610685110092163,
      "learning_rate": 3.5754392691094154e-07,
      "loss": 0.9156,
      "step": 20954
    },
    {
      "epoch": 3.876231183794834,
      "grad_norm": 1.028588056564331,
      "learning_rate": 3.5631068728767536e-07,
      "loss": 1.0818,
      "step": 20955
    },
    {
      "epoch": 3.8764170228582047,
      "grad_norm": 0.9451630711555481,
      "learning_rate": 3.5507957438797314e-07,
      "loss": 0.9733,
      "step": 20956
    },
    {
      "epoch": 3.876602861921576,
      "grad_norm": 1.1087616682052612,
      "learning_rate": 3.5385058823809156e-07,
      "loss": 0.7876,
      "step": 20957
    },
    {
      "epoch": 3.876788700984947,
      "grad_norm": 0.74570232629776,
      "learning_rate": 3.526237288642653e-07,
      "loss": 0.5052,
      "step": 20958
    },
    {
      "epoch": 3.876974540048318,
      "grad_norm": 1.2114689350128174,
      "learning_rate": 3.5139899629268445e-07,
      "loss": 0.929,
      "step": 20959
    },
    {
      "epoch": 3.8771603791116895,
      "grad_norm": 0.9840768575668335,
      "learning_rate": 3.501763905494948e-07,
      "loss": 0.7526,
      "step": 20960
    },
    {
      "epoch": 3.8773462181750604,
      "grad_norm": 0.9244621992111206,
      "learning_rate": 3.4895591166077544e-07,
      "loss": 0.8511,
      "step": 20961
    },
    {
      "epoch": 3.8775320572384313,
      "grad_norm": 0.8813620209693909,
      "learning_rate": 3.4773755965258335e-07,
      "loss": 0.8786,
      "step": 20962
    },
    {
      "epoch": 3.8777178963018026,
      "grad_norm": 1.096408486366272,
      "learning_rate": 3.4652133455093106e-07,
      "loss": 0.773,
      "step": 20963
    },
    {
      "epoch": 3.877903735365174,
      "grad_norm": 1.1385411024093628,
      "learning_rate": 3.4530723638177553e-07,
      "loss": 1.0432,
      "step": 20964
    },
    {
      "epoch": 3.878089574428545,
      "grad_norm": 0.8550912141799927,
      "learning_rate": 3.440952651710072e-07,
      "loss": 0.8016,
      "step": 20965
    },
    {
      "epoch": 3.8782754134919157,
      "grad_norm": 0.8417757749557495,
      "learning_rate": 3.4288542094451646e-07,
      "loss": 0.5985,
      "step": 20966
    },
    {
      "epoch": 3.878461252555287,
      "grad_norm": 1.0538462400436401,
      "learning_rate": 3.4167770372811604e-07,
      "loss": 0.9619,
      "step": 20967
    },
    {
      "epoch": 3.8786470916186584,
      "grad_norm": 1.0869489908218384,
      "learning_rate": 3.404721135475852e-07,
      "loss": 0.7688,
      "step": 20968
    },
    {
      "epoch": 3.8788329306820293,
      "grad_norm": 2.1352310180664062,
      "learning_rate": 3.392686504286591e-07,
      "loss": 1.2084,
      "step": 20969
    },
    {
      "epoch": 3.8790187697454006,
      "grad_norm": 1.1426483392715454,
      "learning_rate": 3.380673143970059e-07,
      "loss": 0.8039,
      "step": 20970
    },
    {
      "epoch": 3.8792046088087715,
      "grad_norm": 0.8323079943656921,
      "learning_rate": 3.368681054782941e-07,
      "loss": 0.7903,
      "step": 20971
    },
    {
      "epoch": 3.879390447872143,
      "grad_norm": 0.9386733770370483,
      "learning_rate": 3.35671023698092e-07,
      "loss": 0.9971,
      "step": 20972
    },
    {
      "epoch": 3.8795762869355137,
      "grad_norm": 1.0048978328704834,
      "learning_rate": 3.3447606908196817e-07,
      "loss": 0.9458,
      "step": 20973
    },
    {
      "epoch": 3.879762125998885,
      "grad_norm": 0.8610646724700928,
      "learning_rate": 3.3328324165541327e-07,
      "loss": 0.7504,
      "step": 20974
    },
    {
      "epoch": 3.879947965062256,
      "grad_norm": 0.9040335416793823,
      "learning_rate": 3.320925414438958e-07,
      "loss": 0.7795,
      "step": 20975
    },
    {
      "epoch": 3.8801338041256273,
      "grad_norm": 1.0290958881378174,
      "learning_rate": 3.309039684728288e-07,
      "loss": 0.695,
      "step": 20976
    },
    {
      "epoch": 3.880319643188998,
      "grad_norm": 1.0693012475967407,
      "learning_rate": 3.297175227675808e-07,
      "loss": 0.7905,
      "step": 20977
    },
    {
      "epoch": 3.8805054822523695,
      "grad_norm": 1.0066351890563965,
      "learning_rate": 3.28533204353465e-07,
      "loss": 0.7107,
      "step": 20978
    },
    {
      "epoch": 3.880691321315741,
      "grad_norm": 1.0007990598678589,
      "learning_rate": 3.2735101325577224e-07,
      "loss": 0.5716,
      "step": 20979
    },
    {
      "epoch": 3.8808771603791117,
      "grad_norm": 0.7360565066337585,
      "learning_rate": 3.2617094949971563e-07,
      "loss": 0.5082,
      "step": 20980
    },
    {
      "epoch": 3.8810629994424826,
      "grad_norm": 1.1390726566314697,
      "learning_rate": 3.2499301311050834e-07,
      "loss": 0.9744,
      "step": 20981
    },
    {
      "epoch": 3.881248838505854,
      "grad_norm": 0.8924768567085266,
      "learning_rate": 3.2381720411325256e-07,
      "loss": 0.7307,
      "step": 20982
    },
    {
      "epoch": 3.8814346775692252,
      "grad_norm": 0.9569699764251709,
      "learning_rate": 3.2264352253308373e-07,
      "loss": 0.8021,
      "step": 20983
    },
    {
      "epoch": 3.881620516632596,
      "grad_norm": 0.9881690740585327,
      "learning_rate": 3.2147196839503734e-07,
      "loss": 0.7866,
      "step": 20984
    },
    {
      "epoch": 3.8818063556959674,
      "grad_norm": 0.9270148873329163,
      "learning_rate": 3.203025417241157e-07,
      "loss": 0.8255,
      "step": 20985
    },
    {
      "epoch": 3.8819921947593383,
      "grad_norm": 0.8654059767723083,
      "learning_rate": 3.1913524254526537e-07,
      "loss": 0.9177,
      "step": 20986
    },
    {
      "epoch": 3.8821780338227097,
      "grad_norm": 1.0542529821395874,
      "learning_rate": 3.179700708834332e-07,
      "loss": 0.8161,
      "step": 20987
    },
    {
      "epoch": 3.8823638728860805,
      "grad_norm": 1.0827463865280151,
      "learning_rate": 3.168070267634549e-07,
      "loss": 0.8735,
      "step": 20988
    },
    {
      "epoch": 3.882549711949452,
      "grad_norm": 0.8836554884910583,
      "learning_rate": 3.156461102101771e-07,
      "loss": 0.6402,
      "step": 20989
    },
    {
      "epoch": 3.8827355510128228,
      "grad_norm": 0.9729933142662048,
      "learning_rate": 3.1448732124835787e-07,
      "loss": 0.757,
      "step": 20990
    },
    {
      "epoch": 3.882921390076194,
      "grad_norm": 0.919170081615448,
      "learning_rate": 3.133306599027441e-07,
      "loss": 0.6836,
      "step": 20991
    },
    {
      "epoch": 3.883107229139565,
      "grad_norm": 0.8769204020500183,
      "learning_rate": 3.1217612619802715e-07,
      "loss": 0.6689,
      "step": 20992
    },
    {
      "epoch": 3.8832930682029363,
      "grad_norm": 0.7491322159767151,
      "learning_rate": 3.110237201588206e-07,
      "loss": 0.6114,
      "step": 20993
    },
    {
      "epoch": 3.8834789072663076,
      "grad_norm": 0.9114632606506348,
      "learning_rate": 3.098734418097493e-07,
      "loss": 0.9487,
      "step": 20994
    },
    {
      "epoch": 3.8836647463296785,
      "grad_norm": 0.8882074356079102,
      "learning_rate": 3.087252911753602e-07,
      "loss": 0.8461,
      "step": 20995
    },
    {
      "epoch": 3.8838505853930494,
      "grad_norm": 1.2083011865615845,
      "learning_rate": 3.075792682801448e-07,
      "loss": 0.9698,
      "step": 20996
    },
    {
      "epoch": 3.8840364244564207,
      "grad_norm": 0.9191475510597229,
      "learning_rate": 3.0643537314858364e-07,
      "loss": 0.8748,
      "step": 20997
    },
    {
      "epoch": 3.884222263519792,
      "grad_norm": 1.1168391704559326,
      "learning_rate": 3.0529360580505707e-07,
      "loss": 0.9781,
      "step": 20998
    },
    {
      "epoch": 3.884408102583163,
      "grad_norm": 0.9397374987602234,
      "learning_rate": 3.0415396627396785e-07,
      "loss": 0.7974,
      "step": 20999
    },
    {
      "epoch": 3.8845939416465343,
      "grad_norm": 0.8357664346694946,
      "learning_rate": 3.030164545796299e-07,
      "loss": 0.7544,
      "step": 21000
    },
    {
      "epoch": 3.884779780709905,
      "grad_norm": 0.920737087726593,
      "learning_rate": 3.0188107074632376e-07,
      "loss": 0.9297,
      "step": 21001
    },
    {
      "epoch": 3.8849656197732765,
      "grad_norm": 0.9226137399673462,
      "learning_rate": 3.0074781479826344e-07,
      "loss": 0.9672,
      "step": 21002
    },
    {
      "epoch": 3.8851514588366474,
      "grad_norm": 0.9706295132637024,
      "learning_rate": 2.9961668675965173e-07,
      "loss": 0.9407,
      "step": 21003
    },
    {
      "epoch": 3.8853372979000187,
      "grad_norm": 1.1339836120605469,
      "learning_rate": 2.98487686654636e-07,
      "loss": 0.8755,
      "step": 21004
    },
    {
      "epoch": 3.8855231369633896,
      "grad_norm": 0.961578369140625,
      "learning_rate": 2.9736081450730815e-07,
      "loss": 0.659,
      "step": 21005
    },
    {
      "epoch": 3.885708976026761,
      "grad_norm": 0.9136162996292114,
      "learning_rate": 2.962360703417044e-07,
      "loss": 0.8642,
      "step": 21006
    },
    {
      "epoch": 3.885894815090132,
      "grad_norm": 0.9740614295005798,
      "learning_rate": 2.951134541818501e-07,
      "loss": 0.8162,
      "step": 21007
    },
    {
      "epoch": 3.886080654153503,
      "grad_norm": 1.0303674936294556,
      "learning_rate": 2.939929660517038e-07,
      "loss": 1.0471,
      "step": 21008
    },
    {
      "epoch": 3.8862664932168745,
      "grad_norm": 1.094256043434143,
      "learning_rate": 2.928746059751686e-07,
      "loss": 0.6487,
      "step": 21009
    },
    {
      "epoch": 3.8864523322802453,
      "grad_norm": 0.8360663056373596,
      "learning_rate": 2.9175837397612536e-07,
      "loss": 0.6802,
      "step": 21010
    },
    {
      "epoch": 3.886638171343616,
      "grad_norm": 0.9560714960098267,
      "learning_rate": 2.9064427007838844e-07,
      "loss": 0.7438,
      "step": 21011
    },
    {
      "epoch": 3.8868240104069876,
      "grad_norm": 1.2549717426300049,
      "learning_rate": 2.895322943057499e-07,
      "loss": 0.7281,
      "step": 21012
    },
    {
      "epoch": 3.887009849470359,
      "grad_norm": 0.9083113670349121,
      "learning_rate": 2.8842244668192406e-07,
      "loss": 0.5523,
      "step": 21013
    },
    {
      "epoch": 3.8871956885337298,
      "grad_norm": 1.0238784551620483,
      "learning_rate": 2.8731472723061424e-07,
      "loss": 0.8316,
      "step": 21014
    },
    {
      "epoch": 3.8873815275971006,
      "grad_norm": 0.8489323854446411,
      "learning_rate": 2.8620913597545707e-07,
      "loss": 0.7387,
      "step": 21015
    },
    {
      "epoch": 3.887567366660472,
      "grad_norm": 0.9123976230621338,
      "learning_rate": 2.8510567294005587e-07,
      "loss": 0.8579,
      "step": 21016
    },
    {
      "epoch": 3.8877532057238433,
      "grad_norm": 1.016202688217163,
      "learning_rate": 2.8400433814794734e-07,
      "loss": 0.848,
      "step": 21017
    },
    {
      "epoch": 3.887939044787214,
      "grad_norm": 0.9767157435417175,
      "learning_rate": 2.829051316226461e-07,
      "loss": 0.9816,
      "step": 21018
    },
    {
      "epoch": 3.8881248838505855,
      "grad_norm": 0.9810354113578796,
      "learning_rate": 2.8180805338762217e-07,
      "loss": 0.7172,
      "step": 21019
    },
    {
      "epoch": 3.8883107229139564,
      "grad_norm": 0.8121437430381775,
      "learning_rate": 2.807131034662791e-07,
      "loss": 0.7956,
      "step": 21020
    },
    {
      "epoch": 3.8884965619773277,
      "grad_norm": 1.0200241804122925,
      "learning_rate": 2.7962028188198706e-07,
      "loss": 0.6751,
      "step": 21021
    },
    {
      "epoch": 3.8886824010406986,
      "grad_norm": 1.0330476760864258,
      "learning_rate": 2.785295886580719e-07,
      "loss": 0.9685,
      "step": 21022
    },
    {
      "epoch": 3.88886824010407,
      "grad_norm": 0.8258622288703918,
      "learning_rate": 2.774410238178149e-07,
      "loss": 0.682,
      "step": 21023
    },
    {
      "epoch": 3.889054079167441,
      "grad_norm": 1.0162584781646729,
      "learning_rate": 2.763545873844531e-07,
      "loss": 0.7645,
      "step": 21024
    },
    {
      "epoch": 3.889239918230812,
      "grad_norm": 0.9385042786598206,
      "learning_rate": 2.752702793811679e-07,
      "loss": 0.8204,
      "step": 21025
    },
    {
      "epoch": 3.889425757294183,
      "grad_norm": 0.9334362745285034,
      "learning_rate": 2.741880998310964e-07,
      "loss": 0.8146,
      "step": 21026
    },
    {
      "epoch": 3.8896115963575544,
      "grad_norm": 0.9243485331535339,
      "learning_rate": 2.731080487573534e-07,
      "loss": 0.8734,
      "step": 21027
    },
    {
      "epoch": 3.8897974354209257,
      "grad_norm": 0.8317142724990845,
      "learning_rate": 2.7203012618297607e-07,
      "loss": 0.453,
      "step": 21028
    },
    {
      "epoch": 3.8899832744842966,
      "grad_norm": 0.8083961009979248,
      "learning_rate": 2.709543321309793e-07,
      "loss": 0.51,
      "step": 21029
    },
    {
      "epoch": 3.8901691135476675,
      "grad_norm": 0.8945092558860779,
      "learning_rate": 2.6988066662432253e-07,
      "loss": 0.7041,
      "step": 21030
    },
    {
      "epoch": 3.890354952611039,
      "grad_norm": 0.7138225436210632,
      "learning_rate": 2.6880912968590965e-07,
      "loss": 0.6267,
      "step": 21031
    },
    {
      "epoch": 3.89054079167441,
      "grad_norm": 0.9707046151161194,
      "learning_rate": 2.6773972133862237e-07,
      "loss": 0.8071,
      "step": 21032
    },
    {
      "epoch": 3.890726630737781,
      "grad_norm": 1.011085867881775,
      "learning_rate": 2.6667244160528684e-07,
      "loss": 0.8037,
      "step": 21033
    },
    {
      "epoch": 3.8909124698011524,
      "grad_norm": 0.9872880578041077,
      "learning_rate": 2.656072905086848e-07,
      "loss": 0.9047,
      "step": 21034
    },
    {
      "epoch": 3.8910983088645232,
      "grad_norm": 1.0062687397003174,
      "learning_rate": 2.6454426807153154e-07,
      "loss": 0.8721,
      "step": 21035
    },
    {
      "epoch": 3.8912841479278946,
      "grad_norm": 0.86174476146698,
      "learning_rate": 2.634833743165421e-07,
      "loss": 0.7465,
      "step": 21036
    },
    {
      "epoch": 3.8914699869912655,
      "grad_norm": 0.9085232019424438,
      "learning_rate": 2.624246092663318e-07,
      "loss": 0.7574,
      "step": 21037
    },
    {
      "epoch": 3.891655826054637,
      "grad_norm": 0.9221695065498352,
      "learning_rate": 2.613679729435159e-07,
      "loss": 0.7558,
      "step": 21038
    },
    {
      "epoch": 3.8918416651180077,
      "grad_norm": 0.7637650370597839,
      "learning_rate": 2.6031346537064293e-07,
      "loss": 0.7128,
      "step": 21039
    },
    {
      "epoch": 3.892027504181379,
      "grad_norm": 0.9544199705123901,
      "learning_rate": 2.592610865702172e-07,
      "loss": 0.847,
      "step": 21040
    },
    {
      "epoch": 3.89221334324475,
      "grad_norm": 1.0526127815246582,
      "learning_rate": 2.5821083656470956e-07,
      "loss": 0.8271,
      "step": 21041
    },
    {
      "epoch": 3.892399182308121,
      "grad_norm": 0.9636459946632385,
      "learning_rate": 2.5716271537652437e-07,
      "loss": 0.7335,
      "step": 21042
    },
    {
      "epoch": 3.8925850213714925,
      "grad_norm": 0.8651697635650635,
      "learning_rate": 2.5611672302803257e-07,
      "loss": 0.8061,
      "step": 21043
    },
    {
      "epoch": 3.8927708604348634,
      "grad_norm": 0.9114117622375488,
      "learning_rate": 2.550728595415608e-07,
      "loss": 0.8719,
      "step": 21044
    },
    {
      "epoch": 3.8929566994982343,
      "grad_norm": 0.9638585448265076,
      "learning_rate": 2.540311249393912e-07,
      "loss": 0.8053,
      "step": 21045
    },
    {
      "epoch": 3.8931425385616056,
      "grad_norm": 0.9569137692451477,
      "learning_rate": 2.529915192437504e-07,
      "loss": 0.6136,
      "step": 21046
    },
    {
      "epoch": 3.893328377624977,
      "grad_norm": 0.9131402373313904,
      "learning_rate": 2.5195404247684294e-07,
      "loss": 0.7985,
      "step": 21047
    },
    {
      "epoch": 3.893514216688348,
      "grad_norm": 0.7240068316459656,
      "learning_rate": 2.5091869466079555e-07,
      "loss": 0.5631,
      "step": 21048
    },
    {
      "epoch": 3.8937000557517187,
      "grad_norm": 1.0669825077056885,
      "learning_rate": 2.4988547581772383e-07,
      "loss": 0.8824,
      "step": 21049
    },
    {
      "epoch": 3.89388589481509,
      "grad_norm": 0.9768051505088806,
      "learning_rate": 2.488543859696546e-07,
      "loss": 0.7538,
      "step": 21050
    },
    {
      "epoch": 3.8940717338784614,
      "grad_norm": 0.9522945284843445,
      "learning_rate": 2.478254251386147e-07,
      "loss": 0.777,
      "step": 21051
    },
    {
      "epoch": 3.8942575729418323,
      "grad_norm": 0.8581794500350952,
      "learning_rate": 2.467985933465644e-07,
      "loss": 0.8562,
      "step": 21052
    },
    {
      "epoch": 3.8944434120052036,
      "grad_norm": 1.0297988653182983,
      "learning_rate": 2.457738906153972e-07,
      "loss": 0.8718,
      "step": 21053
    },
    {
      "epoch": 3.8946292510685745,
      "grad_norm": 1.201256275177002,
      "learning_rate": 2.4475131696701794e-07,
      "loss": 0.8034,
      "step": 21054
    },
    {
      "epoch": 3.894815090131946,
      "grad_norm": 0.9601525664329529,
      "learning_rate": 2.4373087242323125e-07,
      "loss": 0.5696,
      "step": 21055
    },
    {
      "epoch": 3.8950009291953167,
      "grad_norm": 0.7941562533378601,
      "learning_rate": 2.4271255700581974e-07,
      "loss": 0.7143,
      "step": 21056
    },
    {
      "epoch": 3.895186768258688,
      "grad_norm": 0.9030526876449585,
      "learning_rate": 2.416963707365105e-07,
      "loss": 0.909,
      "step": 21057
    },
    {
      "epoch": 3.8953726073220594,
      "grad_norm": 0.8962125778198242,
      "learning_rate": 2.4068231363700846e-07,
      "loss": 0.8633,
      "step": 21058
    },
    {
      "epoch": 3.8955584463854303,
      "grad_norm": 0.8150654435157776,
      "learning_rate": 2.3967038572895173e-07,
      "loss": 0.6983,
      "step": 21059
    },
    {
      "epoch": 3.895744285448801,
      "grad_norm": 1.003766417503357,
      "learning_rate": 2.386605870339342e-07,
      "loss": 0.7517,
      "step": 21060
    },
    {
      "epoch": 3.8959301245121725,
      "grad_norm": 0.8307972550392151,
      "learning_rate": 2.376529175735054e-07,
      "loss": 0.8183,
      "step": 21061
    },
    {
      "epoch": 3.896115963575544,
      "grad_norm": 0.9385664463043213,
      "learning_rate": 2.366473773691702e-07,
      "loss": 0.8329,
      "step": 21062
    },
    {
      "epoch": 3.8963018026389147,
      "grad_norm": 1.100223183631897,
      "learning_rate": 2.3564396644240038e-07,
      "loss": 0.6488,
      "step": 21063
    },
    {
      "epoch": 3.8964876417022856,
      "grad_norm": 0.985472559928894,
      "learning_rate": 2.3464268481461216e-07,
      "loss": 0.8551,
      "step": 21064
    },
    {
      "epoch": 3.896673480765657,
      "grad_norm": 1.1585676670074463,
      "learning_rate": 2.3364353250716619e-07,
      "loss": 0.8933,
      "step": 21065
    },
    {
      "epoch": 3.8968593198290282,
      "grad_norm": 0.815650999546051,
      "learning_rate": 2.3264650954140098e-07,
      "loss": 0.7755,
      "step": 21066
    },
    {
      "epoch": 3.897045158892399,
      "grad_norm": 0.8249333500862122,
      "learning_rate": 2.3165161593857733e-07,
      "loss": 0.7678,
      "step": 21067
    },
    {
      "epoch": 3.8972309979557704,
      "grad_norm": 0.9866358637809753,
      "learning_rate": 2.3065885171994483e-07,
      "loss": 0.8614,
      "step": 21068
    },
    {
      "epoch": 3.8974168370191413,
      "grad_norm": 0.8926262259483337,
      "learning_rate": 2.2966821690669771e-07,
      "loss": 0.8931,
      "step": 21069
    },
    {
      "epoch": 3.8976026760825127,
      "grad_norm": 1.156703233718872,
      "learning_rate": 2.2867971151995238e-07,
      "loss": 0.9257,
      "step": 21070
    },
    {
      "epoch": 3.8977885151458835,
      "grad_norm": 1.1967850923538208,
      "learning_rate": 2.2769333558083638e-07,
      "loss": 0.7539,
      "step": 21071
    },
    {
      "epoch": 3.897974354209255,
      "grad_norm": 1.0719119310379028,
      "learning_rate": 2.2670908911038846e-07,
      "loss": 1.0875,
      "step": 21072
    },
    {
      "epoch": 3.8981601932726258,
      "grad_norm": 0.8525536060333252,
      "learning_rate": 2.25726972129614e-07,
      "loss": 0.8303,
      "step": 21073
    },
    {
      "epoch": 3.898346032335997,
      "grad_norm": 0.8941748738288879,
      "learning_rate": 2.2474698465948518e-07,
      "loss": 0.6014,
      "step": 21074
    },
    {
      "epoch": 3.898531871399368,
      "grad_norm": 1.1483436822891235,
      "learning_rate": 2.2376912672090744e-07,
      "loss": 0.8563,
      "step": 21075
    },
    {
      "epoch": 3.8987177104627393,
      "grad_norm": 0.9525496363639832,
      "learning_rate": 2.22793398334753e-07,
      "loss": 0.7296,
      "step": 21076
    },
    {
      "epoch": 3.8989035495261106,
      "grad_norm": 0.948024332523346,
      "learning_rate": 2.2181979952183852e-07,
      "loss": 0.7623,
      "step": 21077
    },
    {
      "epoch": 3.8990893885894815,
      "grad_norm": 0.8140870928764343,
      "learning_rate": 2.2084833030296958e-07,
      "loss": 0.8169,
      "step": 21078
    },
    {
      "epoch": 3.8992752276528524,
      "grad_norm": 1.0897465944290161,
      "learning_rate": 2.1987899069886298e-07,
      "loss": 0.8071,
      "step": 21079
    },
    {
      "epoch": 3.8994610667162237,
      "grad_norm": 0.8725020289421082,
      "learning_rate": 2.1891178073020213e-07,
      "loss": 0.8565,
      "step": 21080
    },
    {
      "epoch": 3.899646905779595,
      "grad_norm": 0.9559839963912964,
      "learning_rate": 2.1794670041764832e-07,
      "loss": 0.8861,
      "step": 21081
    },
    {
      "epoch": 3.899832744842966,
      "grad_norm": 1.0393133163452148,
      "learning_rate": 2.1698374978177392e-07,
      "loss": 0.8598,
      "step": 21082
    },
    {
      "epoch": 3.9000185839063373,
      "grad_norm": 1.010941505432129,
      "learning_rate": 2.1602292884316254e-07,
      "loss": 0.6854,
      "step": 21083
    },
    {
      "epoch": 3.900204422969708,
      "grad_norm": 1.0234802961349487,
      "learning_rate": 2.1506423762229778e-07,
      "loss": 0.7249,
      "step": 21084
    },
    {
      "epoch": 3.9003902620330795,
      "grad_norm": 0.9321912527084351,
      "learning_rate": 2.141076761396521e-07,
      "loss": 0.684,
      "step": 21085
    },
    {
      "epoch": 3.9005761010964504,
      "grad_norm": 0.9883886575698853,
      "learning_rate": 2.1315324441563146e-07,
      "loss": 0.9127,
      "step": 21086
    },
    {
      "epoch": 3.9007619401598217,
      "grad_norm": 0.9180409908294678,
      "learning_rate": 2.122009424706195e-07,
      "loss": 0.787,
      "step": 21087
    },
    {
      "epoch": 3.9009477792231926,
      "grad_norm": 0.9692301750183105,
      "learning_rate": 2.1125077032494445e-07,
      "loss": 0.614,
      "step": 21088
    },
    {
      "epoch": 3.901133618286564,
      "grad_norm": 0.8420822620391846,
      "learning_rate": 2.1030272799886785e-07,
      "loss": 0.7879,
      "step": 21089
    },
    {
      "epoch": 3.901319457349935,
      "grad_norm": 0.8528715372085571,
      "learning_rate": 2.0935681551262908e-07,
      "loss": 1.0001,
      "step": 21090
    },
    {
      "epoch": 3.901505296413306,
      "grad_norm": 1.1028200387954712,
      "learning_rate": 2.0841303288642311e-07,
      "loss": 0.8211,
      "step": 21091
    },
    {
      "epoch": 3.9016911354766775,
      "grad_norm": 0.9070749878883362,
      "learning_rate": 2.0747138014040047e-07,
      "loss": 0.8908,
      "step": 21092
    },
    {
      "epoch": 3.9018769745400483,
      "grad_norm": 0.9431641101837158,
      "learning_rate": 2.065318572946451e-07,
      "loss": 1.0222,
      "step": 21093
    },
    {
      "epoch": 3.9020628136034192,
      "grad_norm": 1.069291591644287,
      "learning_rate": 2.055944643692187e-07,
      "loss": 0.9555,
      "step": 21094
    },
    {
      "epoch": 3.9022486526667906,
      "grad_norm": 1.1835315227508545,
      "learning_rate": 2.0465920138411644e-07,
      "loss": 0.9396,
      "step": 21095
    },
    {
      "epoch": 3.902434491730162,
      "grad_norm": 0.9170207977294922,
      "learning_rate": 2.0372606835931118e-07,
      "loss": 0.8386,
      "step": 21096
    },
    {
      "epoch": 3.9026203307935328,
      "grad_norm": 0.9324058294296265,
      "learning_rate": 2.0279506531472037e-07,
      "loss": 0.873,
      "step": 21097
    },
    {
      "epoch": 3.9028061698569037,
      "grad_norm": 0.9698527455329895,
      "learning_rate": 2.0186619227020586e-07,
      "loss": 0.5926,
      "step": 21098
    },
    {
      "epoch": 3.902992008920275,
      "grad_norm": 0.8722599148750305,
      "learning_rate": 2.0093944924559626e-07,
      "loss": 0.8603,
      "step": 21099
    },
    {
      "epoch": 3.9031778479836463,
      "grad_norm": 0.8694491982460022,
      "learning_rate": 2.0001483626067573e-07,
      "loss": 0.7546,
      "step": 21100
    },
    {
      "epoch": 3.903363687047017,
      "grad_norm": 0.871938943862915,
      "learning_rate": 1.9909235333517295e-07,
      "loss": 0.8239,
      "step": 21101
    },
    {
      "epoch": 3.9035495261103885,
      "grad_norm": 0.8448165655136108,
      "learning_rate": 1.9817200048877217e-07,
      "loss": 0.8066,
      "step": 21102
    },
    {
      "epoch": 3.9037353651737594,
      "grad_norm": 0.9169268012046814,
      "learning_rate": 1.972537777411243e-07,
      "loss": 0.9164,
      "step": 21103
    },
    {
      "epoch": 3.9039212042371307,
      "grad_norm": 0.9491472840309143,
      "learning_rate": 1.9633768511183592e-07,
      "loss": 0.7313,
      "step": 21104
    },
    {
      "epoch": 3.9041070433005016,
      "grad_norm": 1.0520941019058228,
      "learning_rate": 1.9542372262044694e-07,
      "loss": 0.9784,
      "step": 21105
    },
    {
      "epoch": 3.904292882363873,
      "grad_norm": 0.9484320878982544,
      "learning_rate": 1.94511890286464e-07,
      "loss": 0.7181,
      "step": 21106
    },
    {
      "epoch": 3.9044787214272443,
      "grad_norm": 0.94563227891922,
      "learning_rate": 1.9360218812934927e-07,
      "loss": 0.7048,
      "step": 21107
    },
    {
      "epoch": 3.904664560490615,
      "grad_norm": 0.8737128376960754,
      "learning_rate": 1.9269461616852057e-07,
      "loss": 0.7899,
      "step": 21108
    },
    {
      "epoch": 3.904850399553986,
      "grad_norm": 1.0051727294921875,
      "learning_rate": 1.917891744233624e-07,
      "loss": 0.6657,
      "step": 21109
    },
    {
      "epoch": 3.9050362386173574,
      "grad_norm": 1.0031359195709229,
      "learning_rate": 1.9088586291317045e-07,
      "loss": 0.8422,
      "step": 21110
    },
    {
      "epoch": 3.9052220776807287,
      "grad_norm": 0.7909043431282043,
      "learning_rate": 1.8998468165725148e-07,
      "loss": 0.7197,
      "step": 21111
    },
    {
      "epoch": 3.9054079167440996,
      "grad_norm": 1.001685619354248,
      "learning_rate": 1.8908563067482343e-07,
      "loss": 0.9523,
      "step": 21112
    },
    {
      "epoch": 3.9055937558074705,
      "grad_norm": 0.9117027521133423,
      "learning_rate": 1.8818870998508208e-07,
      "loss": 0.7413,
      "step": 21113
    },
    {
      "epoch": 3.905779594870842,
      "grad_norm": 0.8238965272903442,
      "learning_rate": 1.872939196071677e-07,
      "loss": 0.7549,
      "step": 21114
    },
    {
      "epoch": 3.905965433934213,
      "grad_norm": 0.9219626784324646,
      "learning_rate": 1.8640125956017607e-07,
      "loss": 0.8353,
      "step": 21115
    },
    {
      "epoch": 3.906151272997584,
      "grad_norm": 1.1130602359771729,
      "learning_rate": 1.8551072986316974e-07,
      "loss": 0.6794,
      "step": 21116
    },
    {
      "epoch": 3.9063371120609554,
      "grad_norm": 1.287622094154358,
      "learning_rate": 1.8462233053514467e-07,
      "loss": 0.7526,
      "step": 21117
    },
    {
      "epoch": 3.9065229511243262,
      "grad_norm": 0.8811305165290833,
      "learning_rate": 1.837360615950634e-07,
      "loss": 0.8361,
      "step": 21118
    },
    {
      "epoch": 3.9067087901876976,
      "grad_norm": 0.9759013056755066,
      "learning_rate": 1.8285192306184417e-07,
      "loss": 1.0873,
      "step": 21119
    },
    {
      "epoch": 3.9068946292510685,
      "grad_norm": 1.404094934463501,
      "learning_rate": 1.8196991495436077e-07,
      "loss": 0.8197,
      "step": 21120
    },
    {
      "epoch": 3.90708046831444,
      "grad_norm": 0.8916171193122864,
      "learning_rate": 1.8109003729143147e-07,
      "loss": 0.7677,
      "step": 21121
    },
    {
      "epoch": 3.9072663073778107,
      "grad_norm": 0.8397266268730164,
      "learning_rate": 1.8021229009184126e-07,
      "loss": 0.7017,
      "step": 21122
    },
    {
      "epoch": 3.907452146441182,
      "grad_norm": 0.9295265078544617,
      "learning_rate": 1.793366733743307e-07,
      "loss": 0.9705,
      "step": 21123
    },
    {
      "epoch": 3.907637985504553,
      "grad_norm": 1.106117606163025,
      "learning_rate": 1.784631871575626e-07,
      "loss": 0.8659,
      "step": 21124
    },
    {
      "epoch": 3.907823824567924,
      "grad_norm": 0.9938054084777832,
      "learning_rate": 1.7759183146021096e-07,
      "loss": 0.9143,
      "step": 21125
    },
    {
      "epoch": 3.9080096636312955,
      "grad_norm": 1.1148672103881836,
      "learning_rate": 1.767226063008498e-07,
      "loss": 0.8098,
      "step": 21126
    },
    {
      "epoch": 3.9081955026946664,
      "grad_norm": 0.8424129486083984,
      "learning_rate": 1.7585551169805315e-07,
      "loss": 0.8031,
      "step": 21127
    },
    {
      "epoch": 3.9083813417580373,
      "grad_norm": 0.9194619655609131,
      "learning_rate": 1.749905476703062e-07,
      "loss": 0.7698,
      "step": 21128
    },
    {
      "epoch": 3.9085671808214086,
      "grad_norm": 0.9470589756965637,
      "learning_rate": 1.741277142360831e-07,
      "loss": 0.8279,
      "step": 21129
    },
    {
      "epoch": 3.90875301988478,
      "grad_norm": 0.9974692463874817,
      "learning_rate": 1.7326701141379132e-07,
      "loss": 0.7498,
      "step": 21130
    },
    {
      "epoch": 3.908938858948151,
      "grad_norm": 1.03632390499115,
      "learning_rate": 1.7240843922180505e-07,
      "loss": 0.77,
      "step": 21131
    },
    {
      "epoch": 3.909124698011522,
      "grad_norm": 1.0640885829925537,
      "learning_rate": 1.7155199767845408e-07,
      "loss": 0.823,
      "step": 21132
    },
    {
      "epoch": 3.909310537074893,
      "grad_norm": 1.1156837940216064,
      "learning_rate": 1.706976868020127e-07,
      "loss": 0.9895,
      "step": 21133
    },
    {
      "epoch": 3.9094963761382644,
      "grad_norm": 0.7735521793365479,
      "learning_rate": 1.6984550661071073e-07,
      "loss": 0.7006,
      "step": 21134
    },
    {
      "epoch": 3.9096822152016353,
      "grad_norm": 1.045270562171936,
      "learning_rate": 1.6899545712275588e-07,
      "loss": 0.7921,
      "step": 21135
    },
    {
      "epoch": 3.9098680542650066,
      "grad_norm": 0.9739266633987427,
      "learning_rate": 1.6814753835626695e-07,
      "loss": 0.9616,
      "step": 21136
    },
    {
      "epoch": 3.9100538933283775,
      "grad_norm": 0.9834922552108765,
      "learning_rate": 1.673017503293406e-07,
      "loss": 0.9297,
      "step": 21137
    },
    {
      "epoch": 3.910239732391749,
      "grad_norm": 0.9054717421531677,
      "learning_rate": 1.6645809306005124e-07,
      "loss": 0.7545,
      "step": 21138
    },
    {
      "epoch": 3.9104255714551197,
      "grad_norm": 1.0187366008758545,
      "learning_rate": 1.6561656656638447e-07,
      "loss": 0.7447,
      "step": 21139
    },
    {
      "epoch": 3.910611410518491,
      "grad_norm": 0.8211098313331604,
      "learning_rate": 1.6477717086631483e-07,
      "loss": 0.8485,
      "step": 21140
    },
    {
      "epoch": 3.9107972495818624,
      "grad_norm": 1.1753753423690796,
      "learning_rate": 1.6393990597775022e-07,
      "loss": 0.8201,
      "step": 21141
    },
    {
      "epoch": 3.9109830886452333,
      "grad_norm": 0.9633937478065491,
      "learning_rate": 1.6310477191856522e-07,
      "loss": 1.0452,
      "step": 21142
    },
    {
      "epoch": 3.911168927708604,
      "grad_norm": 0.9461058974266052,
      "learning_rate": 1.622717687065789e-07,
      "loss": 0.8729,
      "step": 21143
    },
    {
      "epoch": 3.9113547667719755,
      "grad_norm": 0.911384642124176,
      "learning_rate": 1.6144089635957704e-07,
      "loss": 0.8119,
      "step": 21144
    },
    {
      "epoch": 3.911540605835347,
      "grad_norm": 0.9880189895629883,
      "learning_rate": 1.6061215489527882e-07,
      "loss": 0.7411,
      "step": 21145
    },
    {
      "epoch": 3.9117264448987177,
      "grad_norm": 0.8829910755157471,
      "learning_rate": 1.5978554433139226e-07,
      "loss": 0.9446,
      "step": 21146
    },
    {
      "epoch": 3.9119122839620886,
      "grad_norm": 0.8976996541023254,
      "learning_rate": 1.5896106468555883e-07,
      "loss": 0.7811,
      "step": 21147
    },
    {
      "epoch": 3.91209812302546,
      "grad_norm": 0.8251069784164429,
      "learning_rate": 1.5813871597535334e-07,
      "loss": 0.8636,
      "step": 21148
    },
    {
      "epoch": 3.9122839620888312,
      "grad_norm": 1.216132402420044,
      "learning_rate": 1.5731849821833954e-07,
      "loss": 0.7413,
      "step": 21149
    },
    {
      "epoch": 3.912469801152202,
      "grad_norm": 0.9787434935569763,
      "learning_rate": 1.5650041143203677e-07,
      "loss": 0.6992,
      "step": 21150
    },
    {
      "epoch": 3.9126556402155734,
      "grad_norm": 0.9688008427619934,
      "learning_rate": 1.5568445563388657e-07,
      "loss": 0.7277,
      "step": 21151
    },
    {
      "epoch": 3.9128414792789443,
      "grad_norm": 0.949082612991333,
      "learning_rate": 1.5487063084131946e-07,
      "loss": 0.8283,
      "step": 21152
    },
    {
      "epoch": 3.9130273183423157,
      "grad_norm": 1.3479398488998413,
      "learning_rate": 1.540589370716883e-07,
      "loss": 0.7075,
      "step": 21153
    },
    {
      "epoch": 3.9132131574056865,
      "grad_norm": 0.8778617978096008,
      "learning_rate": 1.5324937434233467e-07,
      "loss": 0.7449,
      "step": 21154
    },
    {
      "epoch": 3.913398996469058,
      "grad_norm": 0.971396267414093,
      "learning_rate": 1.524419426705226e-07,
      "loss": 0.9297,
      "step": 21155
    },
    {
      "epoch": 3.9135848355324288,
      "grad_norm": 0.8594685196876526,
      "learning_rate": 1.5163664207349381e-07,
      "loss": 0.9008,
      "step": 21156
    },
    {
      "epoch": 3.9137706745958,
      "grad_norm": 0.9365720748901367,
      "learning_rate": 1.508334725684346e-07,
      "loss": 1.102,
      "step": 21157
    },
    {
      "epoch": 3.913956513659171,
      "grad_norm": 1.1041709184646606,
      "learning_rate": 1.500324341724979e-07,
      "loss": 0.891,
      "step": 21158
    },
    {
      "epoch": 3.9141423527225423,
      "grad_norm": 0.815019428730011,
      "learning_rate": 1.4923352690275893e-07,
      "loss": 0.7458,
      "step": 21159
    },
    {
      "epoch": 3.9143281917859136,
      "grad_norm": 0.9310709238052368,
      "learning_rate": 1.4843675077628183e-07,
      "loss": 0.6284,
      "step": 21160
    },
    {
      "epoch": 3.9145140308492845,
      "grad_norm": 1.229316234588623,
      "learning_rate": 1.476421058100641e-07,
      "loss": 0.9508,
      "step": 21161
    },
    {
      "epoch": 3.9146998699126554,
      "grad_norm": 0.8601710796356201,
      "learning_rate": 1.4684959202106997e-07,
      "loss": 0.7945,
      "step": 21162
    },
    {
      "epoch": 3.9148857089760267,
      "grad_norm": 1.241491675376892,
      "learning_rate": 1.460592094262303e-07,
      "loss": 1.0845,
      "step": 21163
    },
    {
      "epoch": 3.915071548039398,
      "grad_norm": 1.1255581378936768,
      "learning_rate": 1.452709580423872e-07,
      "loss": 0.9065,
      "step": 21164
    },
    {
      "epoch": 3.915257387102769,
      "grad_norm": 1.0429335832595825,
      "learning_rate": 1.444848378863828e-07,
      "loss": 0.9329,
      "step": 21165
    },
    {
      "epoch": 3.9154432261661403,
      "grad_norm": 1.027580976486206,
      "learning_rate": 1.4370084897499247e-07,
      "loss": 0.8746,
      "step": 21166
    },
    {
      "epoch": 3.915629065229511,
      "grad_norm": 1.0583301782608032,
      "learning_rate": 1.4291899132494736e-07,
      "loss": 0.7915,
      "step": 21167
    },
    {
      "epoch": 3.9158149042928825,
      "grad_norm": 0.9076946973800659,
      "learning_rate": 1.4213926495293407e-07,
      "loss": 0.7156,
      "step": 21168
    },
    {
      "epoch": 3.9160007433562534,
      "grad_norm": 1.085066318511963,
      "learning_rate": 1.4136166987559485e-07,
      "loss": 0.8511,
      "step": 21169
    },
    {
      "epoch": 3.9161865824196247,
      "grad_norm": 0.9009854197502136,
      "learning_rate": 1.4058620610952755e-07,
      "loss": 0.6704,
      "step": 21170
    },
    {
      "epoch": 3.9163724214829956,
      "grad_norm": 0.8545687794685364,
      "learning_rate": 1.398128736712856e-07,
      "loss": 0.7344,
      "step": 21171
    },
    {
      "epoch": 3.916558260546367,
      "grad_norm": 0.9864504337310791,
      "learning_rate": 1.390416725773669e-07,
      "loss": 0.8692,
      "step": 21172
    },
    {
      "epoch": 3.916744099609738,
      "grad_norm": 0.850251317024231,
      "learning_rate": 1.3827260284423604e-07,
      "loss": 0.7236,
      "step": 21173
    },
    {
      "epoch": 3.916929938673109,
      "grad_norm": 0.8641321659088135,
      "learning_rate": 1.3750566448830215e-07,
      "loss": 0.8228,
      "step": 21174
    },
    {
      "epoch": 3.9171157777364805,
      "grad_norm": 0.9118969440460205,
      "learning_rate": 1.3674085752594103e-07,
      "loss": 0.8112,
      "step": 21175
    },
    {
      "epoch": 3.9173016167998513,
      "grad_norm": 0.951363742351532,
      "learning_rate": 1.35978181973484e-07,
      "loss": 0.7726,
      "step": 21176
    },
    {
      "epoch": 3.9174874558632222,
      "grad_norm": 1.010247826576233,
      "learning_rate": 1.3521763784718477e-07,
      "loss": 0.6253,
      "step": 21177
    },
    {
      "epoch": 3.9176732949265936,
      "grad_norm": 1.1614164113998413,
      "learning_rate": 1.3445922516329702e-07,
      "loss": 0.9844,
      "step": 21178
    },
    {
      "epoch": 3.917859133989965,
      "grad_norm": 1.0017623901367188,
      "learning_rate": 1.3370294393799666e-07,
      "loss": 0.7972,
      "step": 21179
    },
    {
      "epoch": 3.9180449730533358,
      "grad_norm": 1.0043132305145264,
      "learning_rate": 1.3294879418743743e-07,
      "loss": 1.0765,
      "step": 21180
    },
    {
      "epoch": 3.918230812116707,
      "grad_norm": 0.9212683439254761,
      "learning_rate": 1.3219677592770652e-07,
      "loss": 0.7776,
      "step": 21181
    },
    {
      "epoch": 3.918416651180078,
      "grad_norm": 0.9996517896652222,
      "learning_rate": 1.314468891748466e-07,
      "loss": 0.9532,
      "step": 21182
    },
    {
      "epoch": 3.9186024902434493,
      "grad_norm": 0.8499897718429565,
      "learning_rate": 1.306991339448782e-07,
      "loss": 0.9038,
      "step": 21183
    },
    {
      "epoch": 3.91878832930682,
      "grad_norm": 0.8690329790115356,
      "learning_rate": 1.2995351025375524e-07,
      "loss": 0.5533,
      "step": 21184
    },
    {
      "epoch": 3.9189741683701915,
      "grad_norm": 0.8721495866775513,
      "learning_rate": 1.292100181173872e-07,
      "loss": 0.9053,
      "step": 21185
    },
    {
      "epoch": 3.9191600074335624,
      "grad_norm": 0.9867119193077087,
      "learning_rate": 1.2846865755165028e-07,
      "loss": 0.8639,
      "step": 21186
    },
    {
      "epoch": 3.9193458464969337,
      "grad_norm": 0.8210247755050659,
      "learning_rate": 1.2772942857235404e-07,
      "loss": 0.7302,
      "step": 21187
    },
    {
      "epoch": 3.9195316855603046,
      "grad_norm": 0.9453384876251221,
      "learning_rate": 1.2699233119529696e-07,
      "loss": 0.7579,
      "step": 21188
    },
    {
      "epoch": 3.919717524623676,
      "grad_norm": 0.8770848512649536,
      "learning_rate": 1.262573654361887e-07,
      "loss": 0.7229,
      "step": 21189
    },
    {
      "epoch": 3.9199033636870473,
      "grad_norm": 1.0798051357269287,
      "learning_rate": 1.2552453131073893e-07,
      "loss": 0.9094,
      "step": 21190
    },
    {
      "epoch": 3.920089202750418,
      "grad_norm": 0.9015201330184937,
      "learning_rate": 1.2479382883456847e-07,
      "loss": 0.7454,
      "step": 21191
    },
    {
      "epoch": 3.920275041813789,
      "grad_norm": 0.8095447421073914,
      "learning_rate": 1.2406525802328706e-07,
      "loss": 0.6548,
      "step": 21192
    },
    {
      "epoch": 3.9204608808771604,
      "grad_norm": 1.061845064163208,
      "learning_rate": 1.233388188924267e-07,
      "loss": 0.9094,
      "step": 21193
    },
    {
      "epoch": 3.9206467199405317,
      "grad_norm": 0.9467784762382507,
      "learning_rate": 1.2261451145751946e-07,
      "loss": 0.6258,
      "step": 21194
    },
    {
      "epoch": 3.9208325590039026,
      "grad_norm": 1.0385093688964844,
      "learning_rate": 1.218923357339974e-07,
      "loss": 0.8663,
      "step": 21195
    },
    {
      "epoch": 3.9210183980672735,
      "grad_norm": 0.909909725189209,
      "learning_rate": 1.2117229173729262e-07,
      "loss": 0.8303,
      "step": 21196
    },
    {
      "epoch": 3.921204237130645,
      "grad_norm": 1.0109440088272095,
      "learning_rate": 1.204543794827595e-07,
      "loss": 0.8107,
      "step": 21197
    },
    {
      "epoch": 3.921390076194016,
      "grad_norm": 0.7961100935935974,
      "learning_rate": 1.1973859898573026e-07,
      "loss": 0.7006,
      "step": 21198
    },
    {
      "epoch": 3.921575915257387,
      "grad_norm": 0.8510643243789673,
      "learning_rate": 1.1902495026148153e-07,
      "loss": 0.7086,
      "step": 21199
    },
    {
      "epoch": 3.9217617543207584,
      "grad_norm": 1.1658729314804077,
      "learning_rate": 1.1831343332524558e-07,
      "loss": 1.0314,
      "step": 21200
    },
    {
      "epoch": 3.9219475933841292,
      "grad_norm": 0.911927342414856,
      "learning_rate": 1.1760404819219916e-07,
      "loss": 0.7124,
      "step": 21201
    },
    {
      "epoch": 3.9221334324475006,
      "grad_norm": 1.0180903673171997,
      "learning_rate": 1.1689679487749683e-07,
      "loss": 0.7781,
      "step": 21202
    },
    {
      "epoch": 3.9223192715108715,
      "grad_norm": 1.9628604650497437,
      "learning_rate": 1.1619167339621539e-07,
      "loss": 0.8688,
      "step": 21203
    },
    {
      "epoch": 3.922505110574243,
      "grad_norm": 0.924095094203949,
      "learning_rate": 1.1548868376342059e-07,
      "loss": 0.9179,
      "step": 21204
    },
    {
      "epoch": 3.9226909496376137,
      "grad_norm": 0.9204443693161011,
      "learning_rate": 1.1478782599411153e-07,
      "loss": 0.9451,
      "step": 21205
    },
    {
      "epoch": 3.922876788700985,
      "grad_norm": 0.9130086302757263,
      "learning_rate": 1.1408910010324292e-07,
      "loss": 0.9045,
      "step": 21206
    },
    {
      "epoch": 3.923062627764356,
      "grad_norm": 0.9768751263618469,
      "learning_rate": 1.1339250610573615e-07,
      "loss": 0.8263,
      "step": 21207
    },
    {
      "epoch": 3.923248466827727,
      "grad_norm": 0.8459489345550537,
      "learning_rate": 1.1269804401645712e-07,
      "loss": 0.658,
      "step": 21208
    },
    {
      "epoch": 3.9234343058910985,
      "grad_norm": 0.8942362666130066,
      "learning_rate": 1.1200571385021618e-07,
      "loss": 0.7145,
      "step": 21209
    },
    {
      "epoch": 3.9236201449544694,
      "grad_norm": 0.8998235464096069,
      "learning_rate": 1.1131551562180153e-07,
      "loss": 0.8589,
      "step": 21210
    },
    {
      "epoch": 3.9238059840178403,
      "grad_norm": 0.9604059457778931,
      "learning_rate": 1.1062744934594582e-07,
      "loss": 0.8735,
      "step": 21211
    },
    {
      "epoch": 3.9239918230812116,
      "grad_norm": 1.559274673461914,
      "learning_rate": 1.099415150373262e-07,
      "loss": 1.1403,
      "step": 21212
    },
    {
      "epoch": 3.924177662144583,
      "grad_norm": 0.9578410387039185,
      "learning_rate": 1.0925771271058649e-07,
      "loss": 0.9836,
      "step": 21213
    },
    {
      "epoch": 3.924363501207954,
      "grad_norm": 0.8517030477523804,
      "learning_rate": 1.0857604238032614e-07,
      "loss": 0.8048,
      "step": 21214
    },
    {
      "epoch": 3.924549340271325,
      "grad_norm": 0.939640998840332,
      "learning_rate": 1.0789650406108909e-07,
      "loss": 0.6859,
      "step": 21215
    },
    {
      "epoch": 3.924735179334696,
      "grad_norm": 1.3704183101654053,
      "learning_rate": 1.0721909776737482e-07,
      "loss": 0.7968,
      "step": 21216
    },
    {
      "epoch": 3.9249210183980674,
      "grad_norm": 0.9239481687545776,
      "learning_rate": 1.0654382351363845e-07,
      "loss": 0.9119,
      "step": 21217
    },
    {
      "epoch": 3.9251068574614383,
      "grad_norm": 1.0346596240997314,
      "learning_rate": 1.0587068131431289e-07,
      "loss": 0.7972,
      "step": 21218
    },
    {
      "epoch": 3.9252926965248096,
      "grad_norm": 0.9344537258148193,
      "learning_rate": 1.0519967118375329e-07,
      "loss": 1.0446,
      "step": 21219
    },
    {
      "epoch": 3.9254785355881805,
      "grad_norm": 0.9868215322494507,
      "learning_rate": 1.0453079313627045e-07,
      "loss": 0.654,
      "step": 21220
    },
    {
      "epoch": 3.925664374651552,
      "grad_norm": 0.851256787776947,
      "learning_rate": 1.0386404718616405e-07,
      "loss": 0.63,
      "step": 21221
    },
    {
      "epoch": 3.9258502137149227,
      "grad_norm": 1.0101230144500732,
      "learning_rate": 1.0319943334763383e-07,
      "loss": 0.7849,
      "step": 21222
    },
    {
      "epoch": 3.926036052778294,
      "grad_norm": 0.9066257476806641,
      "learning_rate": 1.0253695163489063e-07,
      "loss": 0.6061,
      "step": 21223
    },
    {
      "epoch": 3.9262218918416654,
      "grad_norm": 0.8656601905822754,
      "learning_rate": 1.0187660206206761e-07,
      "loss": 0.7616,
      "step": 21224
    },
    {
      "epoch": 3.9264077309050363,
      "grad_norm": 0.7974125146865845,
      "learning_rate": 1.012183846432535e-07,
      "loss": 0.9409,
      "step": 21225
    },
    {
      "epoch": 3.926593569968407,
      "grad_norm": 1.3469806909561157,
      "learning_rate": 1.0056229939250372e-07,
      "loss": 0.7043,
      "step": 21226
    },
    {
      "epoch": 3.9267794090317785,
      "grad_norm": 0.9865778684616089,
      "learning_rate": 9.990834632381818e-08,
      "loss": 0.7802,
      "step": 21227
    },
    {
      "epoch": 3.92696524809515,
      "grad_norm": 1.0248531103134155,
      "learning_rate": 9.925652545114129e-08,
      "loss": 0.772,
      "step": 21228
    },
    {
      "epoch": 3.9271510871585207,
      "grad_norm": 1.012474775314331,
      "learning_rate": 9.860683678840632e-08,
      "loss": 0.7757,
      "step": 21229
    },
    {
      "epoch": 3.9273369262218916,
      "grad_norm": 0.7945632934570312,
      "learning_rate": 9.795928034946889e-08,
      "loss": 0.764,
      "step": 21230
    },
    {
      "epoch": 3.927522765285263,
      "grad_norm": 0.811353862285614,
      "learning_rate": 9.731385614816236e-08,
      "loss": 0.7347,
      "step": 21231
    },
    {
      "epoch": 3.9277086043486342,
      "grad_norm": 1.1800938844680786,
      "learning_rate": 9.667056419824238e-08,
      "loss": 0.8599,
      "step": 21232
    },
    {
      "epoch": 3.927894443412005,
      "grad_norm": 0.8375271558761597,
      "learning_rate": 9.602940451344244e-08,
      "loss": 0.7644,
      "step": 21233
    },
    {
      "epoch": 3.9280802824753764,
      "grad_norm": 0.8486659526824951,
      "learning_rate": 9.539037710745158e-08,
      "loss": 0.7136,
      "step": 21234
    },
    {
      "epoch": 3.9282661215387473,
      "grad_norm": 1.0186020135879517,
      "learning_rate": 9.475348199392553e-08,
      "loss": 0.7333,
      "step": 21235
    },
    {
      "epoch": 3.9284519606021187,
      "grad_norm": 0.9781624674797058,
      "learning_rate": 9.411871918643122e-08,
      "loss": 1.0204,
      "step": 21236
    },
    {
      "epoch": 3.9286377996654895,
      "grad_norm": 0.9855211973190308,
      "learning_rate": 9.348608869852448e-08,
      "loss": 0.7462,
      "step": 21237
    },
    {
      "epoch": 3.928823638728861,
      "grad_norm": 1.0383330583572388,
      "learning_rate": 9.285559054371672e-08,
      "loss": 0.7325,
      "step": 21238
    },
    {
      "epoch": 3.929009477792232,
      "grad_norm": 0.8758466243743896,
      "learning_rate": 9.222722473546386e-08,
      "loss": 0.7156,
      "step": 21239
    },
    {
      "epoch": 3.929195316855603,
      "grad_norm": 0.9373558759689331,
      "learning_rate": 9.160099128716627e-08,
      "loss": 0.8481,
      "step": 21240
    },
    {
      "epoch": 3.929381155918974,
      "grad_norm": 0.9394092559814453,
      "learning_rate": 9.097689021219102e-08,
      "loss": 0.7789,
      "step": 21241
    },
    {
      "epoch": 3.9295669949823453,
      "grad_norm": 0.9431646466255188,
      "learning_rate": 9.035492152387193e-08,
      "loss": 0.8201,
      "step": 21242
    },
    {
      "epoch": 3.9297528340457166,
      "grad_norm": 0.9627437591552734,
      "learning_rate": 8.973508523547613e-08,
      "loss": 0.9604,
      "step": 21243
    },
    {
      "epoch": 3.9299386731090875,
      "grad_norm": 1.0630133152008057,
      "learning_rate": 8.91173813602264e-08,
      "loss": 0.8981,
      "step": 21244
    },
    {
      "epoch": 3.9301245121724584,
      "grad_norm": 1.1388155221939087,
      "learning_rate": 8.850180991131219e-08,
      "loss": 0.8641,
      "step": 21245
    },
    {
      "epoch": 3.9303103512358297,
      "grad_norm": 0.8789702653884888,
      "learning_rate": 8.788837090186742e-08,
      "loss": 0.7783,
      "step": 21246
    },
    {
      "epoch": 3.930496190299201,
      "grad_norm": 1.3608787059783936,
      "learning_rate": 8.727706434499272e-08,
      "loss": 1.8397,
      "step": 21247
    },
    {
      "epoch": 3.930682029362572,
      "grad_norm": 0.7911452054977417,
      "learning_rate": 8.666789025374433e-08,
      "loss": 0.8227,
      "step": 21248
    },
    {
      "epoch": 3.9308678684259433,
      "grad_norm": 1.0240737199783325,
      "learning_rate": 8.606084864110076e-08,
      "loss": 0.9596,
      "step": 21249
    },
    {
      "epoch": 3.931053707489314,
      "grad_norm": 1.040171504020691,
      "learning_rate": 8.545593952002939e-08,
      "loss": 0.9922,
      "step": 21250
    },
    {
      "epoch": 3.9312395465526855,
      "grad_norm": 0.7827027440071106,
      "learning_rate": 8.485316290344214e-08,
      "loss": 0.6822,
      "step": 21251
    },
    {
      "epoch": 3.9314253856160564,
      "grad_norm": 1.0231354236602783,
      "learning_rate": 8.425251880419537e-08,
      "loss": 0.6931,
      "step": 21252
    },
    {
      "epoch": 3.9316112246794277,
      "grad_norm": 1.0050798654556274,
      "learning_rate": 8.365400723512328e-08,
      "loss": 0.8074,
      "step": 21253
    },
    {
      "epoch": 3.9317970637427986,
      "grad_norm": 0.7175681591033936,
      "learning_rate": 8.305762820899343e-08,
      "loss": 0.5561,
      "step": 21254
    },
    {
      "epoch": 3.93198290280617,
      "grad_norm": 1.0384067296981812,
      "learning_rate": 8.246338173852896e-08,
      "loss": 0.8317,
      "step": 21255
    },
    {
      "epoch": 3.932168741869541,
      "grad_norm": 0.9028663039207458,
      "learning_rate": 8.187126783643084e-08,
      "loss": 0.8666,
      "step": 21256
    },
    {
      "epoch": 3.932354580932912,
      "grad_norm": 0.9729875922203064,
      "learning_rate": 8.12812865153112e-08,
      "loss": 0.9616,
      "step": 21257
    },
    {
      "epoch": 3.9325404199962835,
      "grad_norm": 1.0092549324035645,
      "learning_rate": 8.069343778778215e-08,
      "loss": 0.8919,
      "step": 21258
    },
    {
      "epoch": 3.9327262590596543,
      "grad_norm": 0.7683966159820557,
      "learning_rate": 8.010772166637814e-08,
      "loss": 0.6543,
      "step": 21259
    },
    {
      "epoch": 3.9329120981230252,
      "grad_norm": 1.0346338748931885,
      "learning_rate": 7.952413816361138e-08,
      "loss": 0.9382,
      "step": 21260
    },
    {
      "epoch": 3.9330979371863966,
      "grad_norm": 0.9304220080375671,
      "learning_rate": 7.894268729192744e-08,
      "loss": 0.9117,
      "step": 21261
    },
    {
      "epoch": 3.933283776249768,
      "grad_norm": 0.9711853265762329,
      "learning_rate": 7.836336906373865e-08,
      "loss": 0.8058,
      "step": 21262
    },
    {
      "epoch": 3.9334696153131388,
      "grad_norm": 0.8730411529541016,
      "learning_rate": 7.778618349141287e-08,
      "loss": 0.8607,
      "step": 21263
    },
    {
      "epoch": 3.93365545437651,
      "grad_norm": 1.2616333961486816,
      "learning_rate": 7.721113058727359e-08,
      "loss": 0.9632,
      "step": 21264
    },
    {
      "epoch": 3.933841293439881,
      "grad_norm": 1.0677188634872437,
      "learning_rate": 7.663821036358876e-08,
      "loss": 0.9732,
      "step": 21265
    },
    {
      "epoch": 3.9340271325032523,
      "grad_norm": 1.28867769241333,
      "learning_rate": 7.606742283257085e-08,
      "loss": 0.9898,
      "step": 21266
    },
    {
      "epoch": 3.934212971566623,
      "grad_norm": 1.0222501754760742,
      "learning_rate": 7.54987680064323e-08,
      "loss": 1.0614,
      "step": 21267
    },
    {
      "epoch": 3.9343988106299945,
      "grad_norm": 1.1863951683044434,
      "learning_rate": 7.493224589728565e-08,
      "loss": 0.8528,
      "step": 21268
    },
    {
      "epoch": 3.9345846496933654,
      "grad_norm": 0.8715998530387878,
      "learning_rate": 7.436785651724342e-08,
      "loss": 0.8946,
      "step": 21269
    },
    {
      "epoch": 3.9347704887567367,
      "grad_norm": 1.1731165647506714,
      "learning_rate": 7.380559987832935e-08,
      "loss": 0.7065,
      "step": 21270
    },
    {
      "epoch": 3.9349563278201076,
      "grad_norm": 1.1409883499145508,
      "learning_rate": 7.324547599255605e-08,
      "loss": 0.7519,
      "step": 21271
    },
    {
      "epoch": 3.935142166883479,
      "grad_norm": 0.8426750302314758,
      "learning_rate": 7.26874848718806e-08,
      "loss": 0.5539,
      "step": 21272
    },
    {
      "epoch": 3.9353280059468503,
      "grad_norm": 0.8278335928916931,
      "learning_rate": 7.21316265282046e-08,
      "loss": 0.6854,
      "step": 21273
    },
    {
      "epoch": 3.935513845010221,
      "grad_norm": 0.8848078846931458,
      "learning_rate": 7.157790097340745e-08,
      "loss": 0.918,
      "step": 21274
    },
    {
      "epoch": 3.935699684073592,
      "grad_norm": 0.8748083710670471,
      "learning_rate": 7.102630821927969e-08,
      "loss": 0.7788,
      "step": 21275
    },
    {
      "epoch": 3.9358855231369634,
      "grad_norm": 1.6303963661193848,
      "learning_rate": 7.047684827762302e-08,
      "loss": 1.3235,
      "step": 21276
    },
    {
      "epoch": 3.9360713622003347,
      "grad_norm": 0.8565230965614319,
      "learning_rate": 6.992952116013918e-08,
      "loss": 0.6716,
      "step": 21277
    },
    {
      "epoch": 3.9362572012637056,
      "grad_norm": 1.0051913261413574,
      "learning_rate": 6.938432687854102e-08,
      "loss": 0.8032,
      "step": 21278
    },
    {
      "epoch": 3.9364430403270765,
      "grad_norm": 1.0851858854293823,
      "learning_rate": 6.884126544443037e-08,
      "loss": 0.8638,
      "step": 21279
    },
    {
      "epoch": 3.936628879390448,
      "grad_norm": 1.3282707929611206,
      "learning_rate": 6.830033686942017e-08,
      "loss": 0.6202,
      "step": 21280
    },
    {
      "epoch": 3.936814718453819,
      "grad_norm": 1.0067006349563599,
      "learning_rate": 6.776154116504562e-08,
      "loss": 0.9785,
      "step": 21281
    },
    {
      "epoch": 3.93700055751719,
      "grad_norm": 0.9127426147460938,
      "learning_rate": 6.722487834281977e-08,
      "loss": 0.9038,
      "step": 21282
    },
    {
      "epoch": 3.9371863965805614,
      "grad_norm": 1.1649229526519775,
      "learning_rate": 6.669034841417787e-08,
      "loss": 0.946,
      "step": 21283
    },
    {
      "epoch": 3.9373722356439322,
      "grad_norm": 0.8898347020149231,
      "learning_rate": 6.615795139054415e-08,
      "loss": 0.8945,
      "step": 21284
    },
    {
      "epoch": 3.9375580747073036,
      "grad_norm": 0.9791845083236694,
      "learning_rate": 6.562768728327618e-08,
      "loss": 0.6055,
      "step": 21285
    },
    {
      "epoch": 3.9377439137706745,
      "grad_norm": 1.031858205795288,
      "learning_rate": 6.509955610368711e-08,
      "loss": 0.6925,
      "step": 21286
    },
    {
      "epoch": 3.937929752834046,
      "grad_norm": 0.7945656776428223,
      "learning_rate": 6.457355786305685e-08,
      "loss": 0.7017,
      "step": 21287
    },
    {
      "epoch": 3.938115591897417,
      "grad_norm": 0.8898307681083679,
      "learning_rate": 6.404969257259863e-08,
      "loss": 0.757,
      "step": 21288
    },
    {
      "epoch": 3.938301430960788,
      "grad_norm": 0.9096782803535461,
      "learning_rate": 6.35279602435146e-08,
      "loss": 0.7248,
      "step": 21289
    },
    {
      "epoch": 3.938487270024159,
      "grad_norm": 0.9898115396499634,
      "learning_rate": 6.300836088691809e-08,
      "loss": 0.813,
      "step": 21290
    },
    {
      "epoch": 3.93867310908753,
      "grad_norm": 0.9275386929512024,
      "learning_rate": 6.249089451391133e-08,
      "loss": 0.7457,
      "step": 21291
    },
    {
      "epoch": 3.9388589481509015,
      "grad_norm": 1.1459001302719116,
      "learning_rate": 6.197556113554104e-08,
      "loss": 0.8262,
      "step": 21292
    },
    {
      "epoch": 3.9390447872142724,
      "grad_norm": 1.0498716831207275,
      "learning_rate": 6.146236076279843e-08,
      "loss": 0.8605,
      "step": 21293
    },
    {
      "epoch": 3.9392306262776433,
      "grad_norm": 1.0647568702697754,
      "learning_rate": 6.095129340665252e-08,
      "loss": 0.9404,
      "step": 21294
    },
    {
      "epoch": 3.9394164653410146,
      "grad_norm": 0.9955742359161377,
      "learning_rate": 6.044235907798346e-08,
      "loss": 0.8509,
      "step": 21295
    },
    {
      "epoch": 3.939602304404386,
      "grad_norm": 0.9847276210784912,
      "learning_rate": 5.993555778767146e-08,
      "loss": 0.8834,
      "step": 21296
    },
    {
      "epoch": 3.939788143467757,
      "grad_norm": 0.9324856996536255,
      "learning_rate": 5.943088954654119e-08,
      "loss": 0.9322,
      "step": 21297
    },
    {
      "epoch": 3.939973982531128,
      "grad_norm": 1.1173300743103027,
      "learning_rate": 5.8928354365339614e-08,
      "loss": 0.8917,
      "step": 21298
    },
    {
      "epoch": 3.940159821594499,
      "grad_norm": 1.0777596235275269,
      "learning_rate": 5.842795225481368e-08,
      "loss": 0.9903,
      "step": 21299
    },
    {
      "epoch": 3.9403456606578704,
      "grad_norm": 1.0002269744873047,
      "learning_rate": 5.792968322564374e-08,
      "loss": 0.8653,
      "step": 21300
    },
    {
      "epoch": 3.9405314997212413,
      "grad_norm": 0.8822920918464661,
      "learning_rate": 5.7433547288443525e-08,
      "loss": 0.7842,
      "step": 21301
    },
    {
      "epoch": 3.9407173387846126,
      "grad_norm": 1.0127395391464233,
      "learning_rate": 5.6939544453826766e-08,
      "loss": 0.8987,
      "step": 21302
    },
    {
      "epoch": 3.9409031778479835,
      "grad_norm": 0.8914490342140198,
      "learning_rate": 5.644767473231838e-08,
      "loss": 0.6468,
      "step": 21303
    },
    {
      "epoch": 3.941089016911355,
      "grad_norm": 0.851958692073822,
      "learning_rate": 5.595793813443217e-08,
      "loss": 0.6582,
      "step": 21304
    },
    {
      "epoch": 3.9412748559747257,
      "grad_norm": 0.9173169136047363,
      "learning_rate": 5.547033467060425e-08,
      "loss": 0.7409,
      "step": 21305
    },
    {
      "epoch": 3.941460695038097,
      "grad_norm": 0.9577040672302246,
      "learning_rate": 5.49848643512596e-08,
      "loss": 0.8937,
      "step": 21306
    },
    {
      "epoch": 3.9416465341014684,
      "grad_norm": 1.2994239330291748,
      "learning_rate": 5.4501527186745505e-08,
      "loss": 0.6323,
      "step": 21307
    },
    {
      "epoch": 3.9418323731648393,
      "grad_norm": 0.8939706683158875,
      "learning_rate": 5.4020323187375934e-08,
      "loss": 0.7016,
      "step": 21308
    },
    {
      "epoch": 3.94201821222821,
      "grad_norm": 1.0897858142852783,
      "learning_rate": 5.354125236343155e-08,
      "loss": 0.7664,
      "step": 21309
    },
    {
      "epoch": 3.9422040512915815,
      "grad_norm": 1.5078178644180298,
      "learning_rate": 5.3064314725137507e-08,
      "loss": 1.0231,
      "step": 21310
    },
    {
      "epoch": 3.942389890354953,
      "grad_norm": 1.005741834640503,
      "learning_rate": 5.258951028267456e-08,
      "loss": 0.982,
      "step": 21311
    },
    {
      "epoch": 3.9425757294183237,
      "grad_norm": 0.9288942217826843,
      "learning_rate": 5.211683904616793e-08,
      "loss": 0.6563,
      "step": 21312
    },
    {
      "epoch": 3.942761568481695,
      "grad_norm": 0.8981716632843018,
      "learning_rate": 5.164630102570955e-08,
      "loss": 0.8861,
      "step": 21313
    },
    {
      "epoch": 3.942947407545066,
      "grad_norm": 1.05691659450531,
      "learning_rate": 5.117789623134694e-08,
      "loss": 0.8238,
      "step": 21314
    },
    {
      "epoch": 3.9431332466084372,
      "grad_norm": 0.8106885552406311,
      "learning_rate": 5.0711624673061006e-08,
      "loss": 0.776,
      "step": 21315
    },
    {
      "epoch": 3.943319085671808,
      "grad_norm": 0.9093728065490723,
      "learning_rate": 5.024748636082155e-08,
      "loss": 0.8649,
      "step": 21316
    },
    {
      "epoch": 3.9435049247351794,
      "grad_norm": 0.8479557037353516,
      "learning_rate": 4.9785481304531754e-08,
      "loss": 0.7704,
      "step": 21317
    },
    {
      "epoch": 3.9436907637985503,
      "grad_norm": 0.9083126187324524,
      "learning_rate": 4.932560951405041e-08,
      "loss": 0.7381,
      "step": 21318
    },
    {
      "epoch": 3.9438766028619217,
      "grad_norm": 1.0219753980636597,
      "learning_rate": 4.886787099919188e-08,
      "loss": 0.8371,
      "step": 21319
    },
    {
      "epoch": 3.9440624419252925,
      "grad_norm": 1.0285660028457642,
      "learning_rate": 4.841226576972613e-08,
      "loss": 0.7398,
      "step": 21320
    },
    {
      "epoch": 3.944248280988664,
      "grad_norm": 1.084557294845581,
      "learning_rate": 4.795879383536761e-08,
      "loss": 0.7634,
      "step": 21321
    },
    {
      "epoch": 3.944434120052035,
      "grad_norm": 1.0166188478469849,
      "learning_rate": 4.750745520580857e-08,
      "loss": 0.7617,
      "step": 21322
    },
    {
      "epoch": 3.944619959115406,
      "grad_norm": 0.8051891922950745,
      "learning_rate": 4.7058249890685743e-08,
      "loss": 0.7276,
      "step": 21323
    },
    {
      "epoch": 3.944805798178777,
      "grad_norm": 0.8736680150032043,
      "learning_rate": 4.661117789956926e-08,
      "loss": 0.6811,
      "step": 21324
    },
    {
      "epoch": 3.9449916372421483,
      "grad_norm": 0.9255815148353577,
      "learning_rate": 4.616623924201813e-08,
      "loss": 0.9285,
      "step": 21325
    },
    {
      "epoch": 3.9451774763055196,
      "grad_norm": 0.9325020909309387,
      "learning_rate": 4.572343392751366e-08,
      "loss": 0.4965,
      "step": 21326
    },
    {
      "epoch": 3.9453633153688905,
      "grad_norm": 0.7916317582130432,
      "learning_rate": 4.528276196551495e-08,
      "loss": 0.6926,
      "step": 21327
    },
    {
      "epoch": 3.9455491544322614,
      "grad_norm": 1.3394813537597656,
      "learning_rate": 4.48442233654256e-08,
      "loss": 0.7654,
      "step": 21328
    },
    {
      "epoch": 3.9457349934956327,
      "grad_norm": 1.0707465410232544,
      "learning_rate": 4.440781813660477e-08,
      "loss": 0.8737,
      "step": 21329
    },
    {
      "epoch": 3.945920832559004,
      "grad_norm": 0.9279564619064331,
      "learning_rate": 4.397354628837835e-08,
      "loss": 0.6375,
      "step": 21330
    },
    {
      "epoch": 3.946106671622375,
      "grad_norm": 1.1137791872024536,
      "learning_rate": 4.3541407829994496e-08,
      "loss": 0.8231,
      "step": 21331
    },
    {
      "epoch": 3.9462925106857463,
      "grad_norm": 0.9537021517753601,
      "learning_rate": 4.311140277070136e-08,
      "loss": 0.7397,
      "step": 21332
    },
    {
      "epoch": 3.946478349749117,
      "grad_norm": 0.9723032712936401,
      "learning_rate": 4.268353111964718e-08,
      "loss": 0.8534,
      "step": 21333
    },
    {
      "epoch": 3.9466641888124885,
      "grad_norm": 1.1306790113449097,
      "learning_rate": 4.22577928859913e-08,
      "loss": 0.8695,
      "step": 21334
    },
    {
      "epoch": 3.9468500278758594,
      "grad_norm": 1.071981430053711,
      "learning_rate": 4.183418807879313e-08,
      "loss": 0.9349,
      "step": 21335
    },
    {
      "epoch": 3.9470358669392307,
      "grad_norm": 0.960877537727356,
      "learning_rate": 4.141271670712321e-08,
      "loss": 0.8431,
      "step": 21336
    },
    {
      "epoch": 3.9472217060026016,
      "grad_norm": 0.8617332577705383,
      "learning_rate": 4.099337877995213e-08,
      "loss": 0.7918,
      "step": 21337
    },
    {
      "epoch": 3.947407545065973,
      "grad_norm": 0.9909121990203857,
      "learning_rate": 4.0576174306250494e-08,
      "loss": 0.9285,
      "step": 21338
    },
    {
      "epoch": 3.947593384129344,
      "grad_norm": 1.2329769134521484,
      "learning_rate": 4.0161103294911184e-08,
      "loss": 1.0796,
      "step": 21339
    },
    {
      "epoch": 3.947779223192715,
      "grad_norm": 0.8551541566848755,
      "learning_rate": 3.974816575479379e-08,
      "loss": 0.7998,
      "step": 21340
    },
    {
      "epoch": 3.9479650622560865,
      "grad_norm": 0.9515479803085327,
      "learning_rate": 3.933736169471347e-08,
      "loss": 0.6935,
      "step": 21341
    },
    {
      "epoch": 3.9481509013194573,
      "grad_norm": 0.9743008017539978,
      "learning_rate": 3.8928691123429894e-08,
      "loss": 0.732,
      "step": 21342
    },
    {
      "epoch": 3.9483367403828282,
      "grad_norm": 1.077339768409729,
      "learning_rate": 3.852215404969162e-08,
      "loss": 1.0379,
      "step": 21343
    },
    {
      "epoch": 3.9485225794461996,
      "grad_norm": 0.8660666346549988,
      "learning_rate": 3.8117750482136174e-08,
      "loss": 0.7244,
      "step": 21344
    },
    {
      "epoch": 3.948708418509571,
      "grad_norm": 1.029205322265625,
      "learning_rate": 3.77154804294344e-08,
      "loss": 0.9903,
      "step": 21345
    },
    {
      "epoch": 3.9488942575729418,
      "grad_norm": 0.8247634172439575,
      "learning_rate": 3.731534390014613e-08,
      "loss": 0.7297,
      "step": 21346
    },
    {
      "epoch": 3.949080096636313,
      "grad_norm": 1.1168490648269653,
      "learning_rate": 3.691734090280896e-08,
      "loss": 0.8315,
      "step": 21347
    },
    {
      "epoch": 3.949265935699684,
      "grad_norm": 0.9257239103317261,
      "learning_rate": 3.652147144593832e-08,
      "loss": 0.8475,
      "step": 21348
    },
    {
      "epoch": 3.9494517747630553,
      "grad_norm": 0.8547592163085938,
      "learning_rate": 3.612773553797188e-08,
      "loss": 0.7528,
      "step": 21349
    },
    {
      "epoch": 3.949637613826426,
      "grad_norm": 0.9390490651130676,
      "learning_rate": 3.573613318731406e-08,
      "loss": 0.8821,
      "step": 21350
    },
    {
      "epoch": 3.9498234528897975,
      "grad_norm": 0.8484833240509033,
      "learning_rate": 3.534666440232481e-08,
      "loss": 0.7977,
      "step": 21351
    },
    {
      "epoch": 3.9500092919531684,
      "grad_norm": 0.8618779182434082,
      "learning_rate": 3.495932919130862e-08,
      "loss": 0.8179,
      "step": 21352
    },
    {
      "epoch": 3.9501951310165397,
      "grad_norm": 0.8798786401748657,
      "learning_rate": 3.457412756253664e-08,
      "loss": 0.6562,
      "step": 21353
    },
    {
      "epoch": 3.9503809700799106,
      "grad_norm": 0.9681945443153381,
      "learning_rate": 3.4191059524246724e-08,
      "loss": 0.9839,
      "step": 21354
    },
    {
      "epoch": 3.950566809143282,
      "grad_norm": 0.8329968452453613,
      "learning_rate": 3.381012508458792e-08,
      "loss": 0.7537,
      "step": 21355
    },
    {
      "epoch": 3.9507526482066533,
      "grad_norm": 1.0116149187088013,
      "learning_rate": 3.343132425170925e-08,
      "loss": 0.7618,
      "step": 21356
    },
    {
      "epoch": 3.950938487270024,
      "grad_norm": 0.9968962073326111,
      "learning_rate": 3.305465703368205e-08,
      "loss": 0.7926,
      "step": 21357
    },
    {
      "epoch": 3.951124326333395,
      "grad_norm": 0.8955625295639038,
      "learning_rate": 3.268012343855542e-08,
      "loss": 0.9569,
      "step": 21358
    },
    {
      "epoch": 3.9513101653967664,
      "grad_norm": 1.4347474575042725,
      "learning_rate": 3.230772347432298e-08,
      "loss": 1.0598,
      "step": 21359
    },
    {
      "epoch": 3.9514960044601377,
      "grad_norm": 1.2289568185806274,
      "learning_rate": 3.193745714893392e-08,
      "loss": 0.773,
      "step": 21360
    },
    {
      "epoch": 3.9516818435235086,
      "grad_norm": 0.9194411635398865,
      "learning_rate": 3.156932447028194e-08,
      "loss": 0.6479,
      "step": 21361
    },
    {
      "epoch": 3.95186768258688,
      "grad_norm": 0.9544026851654053,
      "learning_rate": 3.1203325446238496e-08,
      "loss": 0.8541,
      "step": 21362
    },
    {
      "epoch": 3.952053521650251,
      "grad_norm": 0.9884023070335388,
      "learning_rate": 3.083946008459737e-08,
      "loss": 0.8302,
      "step": 21363
    },
    {
      "epoch": 3.952239360713622,
      "grad_norm": 1.0158047676086426,
      "learning_rate": 3.0477728393141226e-08,
      "loss": 0.6797,
      "step": 21364
    },
    {
      "epoch": 3.952425199776993,
      "grad_norm": 0.8721694350242615,
      "learning_rate": 3.0118130379575005e-08,
      "loss": 0.7315,
      "step": 21365
    },
    {
      "epoch": 3.9526110388403644,
      "grad_norm": 0.929004967212677,
      "learning_rate": 2.9760666051592555e-08,
      "loss": 0.8917,
      "step": 21366
    },
    {
      "epoch": 3.9527968779037352,
      "grad_norm": 0.931168794631958,
      "learning_rate": 2.940533541681001e-08,
      "loss": 0.9602,
      "step": 21367
    },
    {
      "epoch": 3.9529827169671066,
      "grad_norm": 1.0448548793792725,
      "learning_rate": 2.9052138482810188e-08,
      "loss": 0.8694,
      "step": 21368
    },
    {
      "epoch": 3.9531685560304775,
      "grad_norm": 1.0332188606262207,
      "learning_rate": 2.8701075257131503e-08,
      "loss": 0.7027,
      "step": 21369
    },
    {
      "epoch": 3.953354395093849,
      "grad_norm": 0.9469727873802185,
      "learning_rate": 2.835214574727907e-08,
      "loss": 0.6672,
      "step": 21370
    },
    {
      "epoch": 3.95354023415722,
      "grad_norm": 0.7369052171707153,
      "learning_rate": 2.8005349960691375e-08,
      "loss": 0.5705,
      "step": 21371
    },
    {
      "epoch": 3.953726073220591,
      "grad_norm": 1.1198176145553589,
      "learning_rate": 2.7660687904762506e-08,
      "loss": 0.9384,
      "step": 21372
    },
    {
      "epoch": 3.953911912283962,
      "grad_norm": 0.9621124267578125,
      "learning_rate": 2.7318159586864344e-08,
      "loss": 0.8375,
      "step": 21373
    },
    {
      "epoch": 3.954097751347333,
      "grad_norm": 0.9608624577522278,
      "learning_rate": 2.6977765014302158e-08,
      "loss": 0.6683,
      "step": 21374
    },
    {
      "epoch": 3.9542835904107045,
      "grad_norm": 1.0401973724365234,
      "learning_rate": 2.6639504194325703e-08,
      "loss": 0.9533,
      "step": 21375
    },
    {
      "epoch": 3.9544694294740754,
      "grad_norm": 0.826633632183075,
      "learning_rate": 2.6303377134184738e-08,
      "loss": 0.6486,
      "step": 21376
    },
    {
      "epoch": 3.9546552685374463,
      "grad_norm": 0.971384584903717,
      "learning_rate": 2.5969383841029094e-08,
      "loss": 0.8556,
      "step": 21377
    },
    {
      "epoch": 3.9548411076008176,
      "grad_norm": 1.117881178855896,
      "learning_rate": 2.5637524321986404e-08,
      "loss": 0.8109,
      "step": 21378
    },
    {
      "epoch": 3.955026946664189,
      "grad_norm": 0.9405913352966309,
      "learning_rate": 2.53077985841621e-08,
      "loss": 0.7834,
      "step": 21379
    },
    {
      "epoch": 3.95521278572756,
      "grad_norm": 0.727852463722229,
      "learning_rate": 2.498020663457279e-08,
      "loss": 0.4824,
      "step": 21380
    },
    {
      "epoch": 3.955398624790931,
      "grad_norm": 0.955696702003479,
      "learning_rate": 2.4654748480212876e-08,
      "loss": 0.8243,
      "step": 21381
    },
    {
      "epoch": 3.955584463854302,
      "grad_norm": 0.8906865119934082,
      "learning_rate": 2.4331424128043457e-08,
      "loss": 0.9608,
      "step": 21382
    },
    {
      "epoch": 3.9557703029176734,
      "grad_norm": 1.1015154123306274,
      "learning_rate": 2.4010233584947916e-08,
      "loss": 0.9698,
      "step": 21383
    },
    {
      "epoch": 3.9559561419810443,
      "grad_norm": 0.9718268513679504,
      "learning_rate": 2.3691176857787433e-08,
      "loss": 0.9026,
      "step": 21384
    },
    {
      "epoch": 3.9561419810444156,
      "grad_norm": 0.980043888092041,
      "learning_rate": 2.3374253953367676e-08,
      "loss": 0.8598,
      "step": 21385
    },
    {
      "epoch": 3.9563278201077865,
      "grad_norm": 0.871580958366394,
      "learning_rate": 2.3059464878472104e-08,
      "loss": 0.8815,
      "step": 21386
    },
    {
      "epoch": 3.956513659171158,
      "grad_norm": 0.8850259780883789,
      "learning_rate": 2.2746809639795362e-08,
      "loss": 0.8521,
      "step": 21387
    },
    {
      "epoch": 3.9566994982345287,
      "grad_norm": 1.4275609254837036,
      "learning_rate": 2.2436288244020996e-08,
      "loss": 0.7822,
      "step": 21388
    },
    {
      "epoch": 3.9568853372979,
      "grad_norm": 0.841894805431366,
      "learning_rate": 2.2127900697777036e-08,
      "loss": 0.7784,
      "step": 21389
    },
    {
      "epoch": 3.9570711763612714,
      "grad_norm": 0.91501384973526,
      "learning_rate": 2.1821647007636004e-08,
      "loss": 0.6936,
      "step": 21390
    },
    {
      "epoch": 3.9572570154246423,
      "grad_norm": 1.1249712705612183,
      "learning_rate": 2.1517527180159315e-08,
      "loss": 0.6923,
      "step": 21391
    },
    {
      "epoch": 3.957442854488013,
      "grad_norm": 0.9765527844429016,
      "learning_rate": 2.1215541221808467e-08,
      "loss": 0.9196,
      "step": 21392
    },
    {
      "epoch": 3.9576286935513845,
      "grad_norm": 0.8743729591369629,
      "learning_rate": 2.091568913904496e-08,
      "loss": 0.8915,
      "step": 21393
    },
    {
      "epoch": 3.957814532614756,
      "grad_norm": 1.0052157640457153,
      "learning_rate": 2.0617970938263676e-08,
      "loss": 0.9449,
      "step": 21394
    },
    {
      "epoch": 3.9580003716781267,
      "grad_norm": 0.8857167959213257,
      "learning_rate": 2.0322386625815092e-08,
      "loss": 0.7643,
      "step": 21395
    },
    {
      "epoch": 3.958186210741498,
      "grad_norm": 0.9562453627586365,
      "learning_rate": 2.0028936208027483e-08,
      "loss": 0.834,
      "step": 21396
    },
    {
      "epoch": 3.958372049804869,
      "grad_norm": 0.8608560562133789,
      "learning_rate": 1.9737619691140297e-08,
      "loss": 0.8131,
      "step": 21397
    },
    {
      "epoch": 3.9585578888682402,
      "grad_norm": 0.9653459787368774,
      "learning_rate": 1.9448437081381887e-08,
      "loss": 0.697,
      "step": 21398
    },
    {
      "epoch": 3.958743727931611,
      "grad_norm": 0.9109430909156799,
      "learning_rate": 1.9161388384925093e-08,
      "loss": 0.755,
      "step": 21399
    },
    {
      "epoch": 3.9589295669949824,
      "grad_norm": 0.858791172504425,
      "learning_rate": 1.8876473607887245e-08,
      "loss": 0.6814,
      "step": 21400
    },
    {
      "epoch": 3.9591154060583533,
      "grad_norm": Infinity,
      "learning_rate": 1.8876473607887245e-08,
      "loss": 1.3839,
      "step": 21401
    },
    {
      "epoch": 3.9593012451217247,
      "grad_norm": 1.032059907913208,
      "learning_rate": 1.859369275636347e-08,
      "loss": 0.8256,
      "step": 21402
    },
    {
      "epoch": 3.9594870841850955,
      "grad_norm": 1.0155627727508545,
      "learning_rate": 1.831304583638227e-08,
      "loss": 0.862,
      "step": 21403
    },
    {
      "epoch": 3.959672923248467,
      "grad_norm": 0.857572078704834,
      "learning_rate": 1.8034532853927754e-08,
      "loss": 0.8946,
      "step": 21404
    },
    {
      "epoch": 3.959858762311838,
      "grad_norm": 1.0889736413955688,
      "learning_rate": 1.775815381495072e-08,
      "loss": 0.9409,
      "step": 21405
    },
    {
      "epoch": 3.960044601375209,
      "grad_norm": 0.8809292316436768,
      "learning_rate": 1.7483908725357545e-08,
      "loss": 0.5094,
      "step": 21406
    },
    {
      "epoch": 3.96023044043858,
      "grad_norm": 1.019468903541565,
      "learning_rate": 1.721179759098801e-08,
      "loss": 0.8253,
      "step": 21407
    },
    {
      "epoch": 3.9604162795019513,
      "grad_norm": 0.8753800988197327,
      "learning_rate": 1.694182041765968e-08,
      "loss": 0.8791,
      "step": 21408
    },
    {
      "epoch": 3.9606021185653226,
      "grad_norm": 1.059045433998108,
      "learning_rate": 1.6673977211123515e-08,
      "loss": 0.9221,
      "step": 21409
    },
    {
      "epoch": 3.9607879576286935,
      "grad_norm": 0.9621778130531311,
      "learning_rate": 1.640826797710826e-08,
      "loss": 0.8444,
      "step": 21410
    },
    {
      "epoch": 3.9609737966920644,
      "grad_norm": 0.8941175937652588,
      "learning_rate": 1.6144692721276054e-08,
      "loss": 0.6987,
      "step": 21411
    },
    {
      "epoch": 3.9611596357554357,
      "grad_norm": 0.9373825788497925,
      "learning_rate": 1.5883251449266832e-08,
      "loss": 0.8148,
      "step": 21412
    },
    {
      "epoch": 3.961345474818807,
      "grad_norm": 1.0011820793151855,
      "learning_rate": 1.562394416663171e-08,
      "loss": 0.9052,
      "step": 21413
    },
    {
      "epoch": 3.961531313882178,
      "grad_norm": 0.7819288969039917,
      "learning_rate": 1.53667708789329e-08,
      "loss": 0.6291,
      "step": 21414
    },
    {
      "epoch": 3.9617171529455493,
      "grad_norm": 0.9070609211921692,
      "learning_rate": 1.511173159165491e-08,
      "loss": 0.736,
      "step": 21415
    },
    {
      "epoch": 3.96190299200892,
      "grad_norm": 0.9019717574119568,
      "learning_rate": 1.4858826310226725e-08,
      "loss": 0.7922,
      "step": 21416
    },
    {
      "epoch": 3.9620888310722915,
      "grad_norm": 1.0442172288894653,
      "learning_rate": 1.4608055040066237e-08,
      "loss": 0.844,
      "step": 21417
    },
    {
      "epoch": 3.9622746701356624,
      "grad_norm": 1.027233362197876,
      "learning_rate": 1.4359417786502516e-08,
      "loss": 0.8094,
      "step": 21418
    },
    {
      "epoch": 3.9624605091990337,
      "grad_norm": 0.9201172590255737,
      "learning_rate": 1.4112914554864631e-08,
      "loss": 0.8485,
      "step": 21419
    },
    {
      "epoch": 3.962646348262405,
      "grad_norm": 1.0947834253311157,
      "learning_rate": 1.3868545350403939e-08,
      "loss": 0.8768,
      "step": 21420
    },
    {
      "epoch": 3.962832187325776,
      "grad_norm": 0.980541467666626,
      "learning_rate": 1.3626310178327384e-08,
      "loss": 0.6502,
      "step": 21421
    },
    {
      "epoch": 3.963018026389147,
      "grad_norm": 0.9271747469902039,
      "learning_rate": 1.3386209043819708e-08,
      "loss": 0.5833,
      "step": 21422
    },
    {
      "epoch": 3.963203865452518,
      "grad_norm": 1.063616156578064,
      "learning_rate": 1.314824195199904e-08,
      "loss": 0.8515,
      "step": 21423
    },
    {
      "epoch": 3.9633897045158895,
      "grad_norm": 0.8098671436309814,
      "learning_rate": 1.2912408907950201e-08,
      "loss": 0.7521,
      "step": 21424
    },
    {
      "epoch": 3.9635755435792603,
      "grad_norm": 1.0300740003585815,
      "learning_rate": 1.26787099166914e-08,
      "loss": 0.7567,
      "step": 21425
    },
    {
      "epoch": 3.9637613826426312,
      "grad_norm": 1.0278445482254028,
      "learning_rate": 1.2447144983229742e-08,
      "loss": 0.9015,
      "step": 21426
    },
    {
      "epoch": 3.9639472217060026,
      "grad_norm": 0.8696473240852356,
      "learning_rate": 1.2217714112483514e-08,
      "loss": 0.7775,
      "step": 21427
    },
    {
      "epoch": 3.964133060769374,
      "grad_norm": 0.7928922772407532,
      "learning_rate": 1.1990417309382107e-08,
      "loss": 0.708,
      "step": 21428
    },
    {
      "epoch": 3.9643188998327448,
      "grad_norm": 1.0644580125808716,
      "learning_rate": 1.176525457874389e-08,
      "loss": 0.798,
      "step": 21429
    },
    {
      "epoch": 3.964504738896116,
      "grad_norm": 0.9723425507545471,
      "learning_rate": 1.1542225925398332e-08,
      "loss": 0.86,
      "step": 21430
    },
    {
      "epoch": 3.964690577959487,
      "grad_norm": 0.8694878816604614,
      "learning_rate": 1.1321331354097186e-08,
      "loss": 0.7656,
      "step": 21431
    },
    {
      "epoch": 3.9648764170228583,
      "grad_norm": 1.1160660982131958,
      "learning_rate": 1.1102570869547801e-08,
      "loss": 0.7742,
      "step": 21432
    },
    {
      "epoch": 3.965062256086229,
      "grad_norm": 0.9629189372062683,
      "learning_rate": 1.0885944476435316e-08,
      "loss": 0.9387,
      "step": 21433
    },
    {
      "epoch": 3.9652480951496005,
      "grad_norm": 0.9273543357849121,
      "learning_rate": 1.0671452179367158e-08,
      "loss": 0.7062,
      "step": 21434
    },
    {
      "epoch": 3.9654339342129714,
      "grad_norm": 0.8834670186042786,
      "learning_rate": 1.0459093982928547e-08,
      "loss": 0.7967,
      "step": 21435
    },
    {
      "epoch": 3.9656197732763427,
      "grad_norm": 0.8330079317092896,
      "learning_rate": 1.0248869891660295e-08,
      "loss": 0.8148,
      "step": 21436
    },
    {
      "epoch": 3.9658056123397136,
      "grad_norm": 0.9004113078117371,
      "learning_rate": 1.0040779910036602e-08,
      "loss": 0.8104,
      "step": 21437
    },
    {
      "epoch": 3.965991451403085,
      "grad_norm": 0.9802833199501038,
      "learning_rate": 9.834824042498358e-09,
      "loss": 0.8014,
      "step": 21438
    },
    {
      "epoch": 3.9661772904664563,
      "grad_norm": 0.7356975078582764,
      "learning_rate": 9.631002293442049e-09,
      "loss": 0.6674,
      "step": 21439
    },
    {
      "epoch": 3.966363129529827,
      "grad_norm": 0.8870053887367249,
      "learning_rate": 9.429314667230849e-09,
      "loss": 0.7549,
      "step": 21440
    },
    {
      "epoch": 3.966548968593198,
      "grad_norm": 0.9902865886688232,
      "learning_rate": 9.229761168161321e-09,
      "loss": 0.7859,
      "step": 21441
    },
    {
      "epoch": 3.9667348076565694,
      "grad_norm": 0.9357664585113525,
      "learning_rate": 9.032341800485622e-09,
      "loss": 0.8592,
      "step": 21442
    },
    {
      "epoch": 3.9669206467199407,
      "grad_norm": 0.8407206535339355,
      "learning_rate": 8.837056568422597e-09,
      "loss": 0.7167,
      "step": 21443
    },
    {
      "epoch": 3.9671064857833116,
      "grad_norm": 0.9275605082511902,
      "learning_rate": 8.643905476146686e-09,
      "loss": 0.7188,
      "step": 21444
    },
    {
      "epoch": 3.967292324846683,
      "grad_norm": 0.9496821761131287,
      "learning_rate": 8.452888527765712e-09,
      "loss": 0.8235,
      "step": 21445
    },
    {
      "epoch": 3.967478163910054,
      "grad_norm": 0.7953393459320068,
      "learning_rate": 8.2640057273764e-09,
      "loss": 0.7986,
      "step": 21446
    },
    {
      "epoch": 3.967664002973425,
      "grad_norm": 0.9482079148292542,
      "learning_rate": 8.077257078986655e-09,
      "loss": 0.7884,
      "step": 21447
    },
    {
      "epoch": 3.967849842036796,
      "grad_norm": 0.8741645216941833,
      "learning_rate": 7.892642586604383e-09,
      "loss": 0.8575,
      "step": 21448
    },
    {
      "epoch": 3.9680356811001674,
      "grad_norm": 1.0921684503555298,
      "learning_rate": 7.71016225415977e-09,
      "loss": 0.6118,
      "step": 21449
    },
    {
      "epoch": 3.9682215201635382,
      "grad_norm": 0.8780234456062317,
      "learning_rate": 7.529816085549702e-09,
      "loss": 0.8748,
      "step": 21450
    },
    {
      "epoch": 3.9684073592269096,
      "grad_norm": 0.8375685214996338,
      "learning_rate": 7.351604084615549e-09,
      "loss": 1.0171,
      "step": 21451
    },
    {
      "epoch": 3.9685931982902805,
      "grad_norm": 0.9099856019020081,
      "learning_rate": 7.175526255165377e-09,
      "loss": 0.8002,
      "step": 21452
    },
    {
      "epoch": 3.968779037353652,
      "grad_norm": 0.9492738246917725,
      "learning_rate": 7.001582600962841e-09,
      "loss": 0.9728,
      "step": 21453
    },
    {
      "epoch": 3.968964876417023,
      "grad_norm": 0.9797655940055847,
      "learning_rate": 6.829773125716088e-09,
      "loss": 0.7906,
      "step": 21454
    },
    {
      "epoch": 3.969150715480394,
      "grad_norm": 0.9608103036880493,
      "learning_rate": 6.660097833099954e-09,
      "loss": 0.7766,
      "step": 21455
    },
    {
      "epoch": 3.969336554543765,
      "grad_norm": 1.1397579908370972,
      "learning_rate": 6.492556726722665e-09,
      "loss": 0.9013,
      "step": 21456
    },
    {
      "epoch": 3.969522393607136,
      "grad_norm": 0.8630730509757996,
      "learning_rate": 6.3271498101591386e-09,
      "loss": 0.8758,
      "step": 21457
    },
    {
      "epoch": 3.9697082326705075,
      "grad_norm": 0.9617315530776978,
      "learning_rate": 6.163877086962089e-09,
      "loss": 0.8588,
      "step": 21458
    },
    {
      "epoch": 3.9698940717338784,
      "grad_norm": 0.9304690957069397,
      "learning_rate": 6.002738560584309e-09,
      "loss": 0.7827,
      "step": 21459
    },
    {
      "epoch": 3.9700799107972493,
      "grad_norm": 0.7468497157096863,
      "learning_rate": 5.843734234489695e-09,
      "loss": 0.5286,
      "step": 21460
    },
    {
      "epoch": 3.9702657498606206,
      "grad_norm": 0.99229496717453,
      "learning_rate": 5.686864112064427e-09,
      "loss": 0.8848,
      "step": 21461
    },
    {
      "epoch": 3.970451588923992,
      "grad_norm": 0.9466888308525085,
      "learning_rate": 5.5321281966502765e-09,
      "loss": 0.8736,
      "step": 21462
    },
    {
      "epoch": 3.970637427987363,
      "grad_norm": 1.1540240049362183,
      "learning_rate": 5.3795264915668106e-09,
      "loss": 0.8687,
      "step": 21463
    },
    {
      "epoch": 3.970823267050734,
      "grad_norm": 0.7100964188575745,
      "learning_rate": 5.22905900005588e-09,
      "loss": 0.5729,
      "step": 21464
    },
    {
      "epoch": 3.971009106114105,
      "grad_norm": 1.0300414562225342,
      "learning_rate": 5.08072572532603e-09,
      "loss": 0.801,
      "step": 21465
    },
    {
      "epoch": 3.9711949451774764,
      "grad_norm": 0.9186599254608154,
      "learning_rate": 4.934526670552497e-09,
      "loss": 0.6908,
      "step": 21466
    },
    {
      "epoch": 3.9713807842408473,
      "grad_norm": 1.088610291481018,
      "learning_rate": 4.7904618388550095e-09,
      "loss": 0.8143,
      "step": 21467
    },
    {
      "epoch": 3.9715666233042186,
      "grad_norm": 0.7681308388710022,
      "learning_rate": 4.6485312333088835e-09,
      "loss": 0.7136,
      "step": 21468
    },
    {
      "epoch": 3.97175246236759,
      "grad_norm": 0.9112083911895752,
      "learning_rate": 4.508734856933927e-09,
      "loss": 0.7703,
      "step": 21469
    },
    {
      "epoch": 3.971938301430961,
      "grad_norm": 0.8502170443534851,
      "learning_rate": 4.371072712727742e-09,
      "loss": 0.7863,
      "step": 21470
    },
    {
      "epoch": 3.9721241404943317,
      "grad_norm": 0.8810780644416809,
      "learning_rate": 4.235544803610214e-09,
      "loss": 0.7088,
      "step": 21471
    },
    {
      "epoch": 3.972309979557703,
      "grad_norm": 0.7818952798843384,
      "learning_rate": 4.102151132501231e-09,
      "loss": 0.6869,
      "step": 21472
    },
    {
      "epoch": 3.9724958186210744,
      "grad_norm": 0.9463311433792114,
      "learning_rate": 3.970891702220758e-09,
      "loss": 0.8012,
      "step": 21473
    },
    {
      "epoch": 3.9726816576844453,
      "grad_norm": 1.06798255443573,
      "learning_rate": 3.841766515588763e-09,
      "loss": 0.7741,
      "step": 21474
    },
    {
      "epoch": 3.972867496747816,
      "grad_norm": 0.8491981029510498,
      "learning_rate": 3.7147755753474954e-09,
      "loss": 0.5917,
      "step": 21475
    },
    {
      "epoch": 3.9730533358111875,
      "grad_norm": 0.906574010848999,
      "learning_rate": 3.5899188842170027e-09,
      "loss": 0.5356,
      "step": 21476
    },
    {
      "epoch": 3.973239174874559,
      "grad_norm": 0.9730631113052368,
      "learning_rate": 3.467196444850718e-09,
      "loss": 0.8405,
      "step": 21477
    },
    {
      "epoch": 3.9734250139379297,
      "grad_norm": 0.8137460350990295,
      "learning_rate": 3.346608259890971e-09,
      "loss": 0.7913,
      "step": 21478
    },
    {
      "epoch": 3.973610853001301,
      "grad_norm": 1.1336978673934937,
      "learning_rate": 3.2281543318912756e-09,
      "loss": 0.8974,
      "step": 21479
    },
    {
      "epoch": 3.973796692064672,
      "grad_norm": 1.1150788068771362,
      "learning_rate": 3.1118346633829397e-09,
      "loss": 0.5933,
      "step": 21480
    },
    {
      "epoch": 3.9739825311280432,
      "grad_norm": 1.1757761240005493,
      "learning_rate": 2.9976492568528635e-09,
      "loss": 0.9587,
      "step": 21481
    },
    {
      "epoch": 3.974168370191414,
      "grad_norm": 0.9290224313735962,
      "learning_rate": 2.8855981147324352e-09,
      "loss": 0.8138,
      "step": 21482
    },
    {
      "epoch": 3.9743542092547854,
      "grad_norm": 0.8209988474845886,
      "learning_rate": 2.7756812394197363e-09,
      "loss": 0.7527,
      "step": 21483
    },
    {
      "epoch": 3.9745400483181563,
      "grad_norm": 0.9280112981796265,
      "learning_rate": 2.6678986332573373e-09,
      "loss": 0.8674,
      "step": 21484
    },
    {
      "epoch": 3.9747258873815277,
      "grad_norm": 1.1409305334091187,
      "learning_rate": 2.5622502985434003e-09,
      "loss": 0.8977,
      "step": 21485
    },
    {
      "epoch": 3.9749117264448985,
      "grad_norm": 1.0304864645004272,
      "learning_rate": 2.45873623754278e-09,
      "loss": 0.7986,
      "step": 21486
    },
    {
      "epoch": 3.97509756550827,
      "grad_norm": 0.8403981924057007,
      "learning_rate": 2.3573564524537183e-09,
      "loss": 0.8795,
      "step": 21487
    },
    {
      "epoch": 3.975283404571641,
      "grad_norm": 0.9824759364128113,
      "learning_rate": 2.2581109454411496e-09,
      "loss": 1.0885,
      "step": 21488
    },
    {
      "epoch": 3.975469243635012,
      "grad_norm": 1.0141150951385498,
      "learning_rate": 2.1609997186367024e-09,
      "loss": 1.0805,
      "step": 21489
    },
    {
      "epoch": 3.975655082698383,
      "grad_norm": 0.9527256488800049,
      "learning_rate": 2.0660227740942893e-09,
      "loss": 0.7499,
      "step": 21490
    },
    {
      "epoch": 3.9758409217617543,
      "grad_norm": 0.9705147743225098,
      "learning_rate": 1.9731801138456184e-09,
      "loss": 0.8076,
      "step": 21491
    },
    {
      "epoch": 3.9760267608251256,
      "grad_norm": 1.0493558645248413,
      "learning_rate": 1.882471739889091e-09,
      "loss": 0.6033,
      "step": 21492
    },
    {
      "epoch": 3.9762125998884965,
      "grad_norm": 0.9863125681877136,
      "learning_rate": 1.7938976541342912e-09,
      "loss": 0.9957,
      "step": 21493
    },
    {
      "epoch": 3.976398438951868,
      "grad_norm": 0.9200931191444397,
      "learning_rate": 1.707457858490802e-09,
      "loss": 0.8426,
      "step": 21494
    },
    {
      "epoch": 3.9765842780152387,
      "grad_norm": 1.5785354375839233,
      "learning_rate": 1.6231523548015937e-09,
      "loss": 1.3327,
      "step": 21495
    },
    {
      "epoch": 3.97677011707861,
      "grad_norm": 0.9231945276260376,
      "learning_rate": 1.5409811448541257e-09,
      "loss": 0.9703,
      "step": 21496
    },
    {
      "epoch": 3.976955956141981,
      "grad_norm": 0.8975953459739685,
      "learning_rate": 1.4609442304136523e-09,
      "loss": 0.7697,
      "step": 21497
    },
    {
      "epoch": 3.9771417952053523,
      "grad_norm": 1.0590022802352905,
      "learning_rate": 1.383041613178815e-09,
      "loss": 0.9046,
      "step": 21498
    },
    {
      "epoch": 3.977327634268723,
      "grad_norm": 1.0810893774032593,
      "learning_rate": 1.3072732948260502e-09,
      "loss": 0.7008,
      "step": 21499
    },
    {
      "epoch": 3.9775134733320945,
      "grad_norm": 0.938327431678772,
      "learning_rate": 1.2336392769651816e-09,
      "loss": 0.7577,
      "step": 21500
    },
    {
      "epoch": 3.9776993123954654,
      "grad_norm": 0.9448954463005066,
      "learning_rate": 1.162139561172726e-09,
      "loss": 0.7967,
      "step": 21501
    },
    {
      "epoch": 3.9778851514588367,
      "grad_norm": 1.4106966257095337,
      "learning_rate": 1.0927741489585862e-09,
      "loss": 0.6138,
      "step": 21502
    },
    {
      "epoch": 3.978070990522208,
      "grad_norm": 0.882112979888916,
      "learning_rate": 1.0255430418215638e-09,
      "loss": 0.6142,
      "step": 21503
    },
    {
      "epoch": 3.978256829585579,
      "grad_norm": 1.1243953704833984,
      "learning_rate": 9.604462411827441e-10,
      "loss": 1.0005,
      "step": 21504
    },
    {
      "epoch": 3.97844266864895,
      "grad_norm": 0.925744891166687,
      "learning_rate": 8.974837484410081e-10,
      "loss": 0.9544,
      "step": 21505
    },
    {
      "epoch": 3.978628507712321,
      "grad_norm": 0.8985171318054199,
      "learning_rate": 8.366555649397256e-10,
      "loss": 0.9202,
      "step": 21506
    },
    {
      "epoch": 3.9788143467756925,
      "grad_norm": 0.9336652755737305,
      "learning_rate": 7.779616919667553e-10,
      "loss": 0.7498,
      "step": 21507
    },
    {
      "epoch": 3.9790001858390633,
      "grad_norm": 0.9400238990783691,
      "learning_rate": 7.214021307877517e-10,
      "loss": 0.8356,
      "step": 21508
    },
    {
      "epoch": 3.9791860249024342,
      "grad_norm": 0.9942156076431274,
      "learning_rate": 6.669768826017552e-10,
      "loss": 0.7825,
      "step": 21509
    },
    {
      "epoch": 3.9793718639658056,
      "grad_norm": 1.040473222732544,
      "learning_rate": 6.146859485745004e-10,
      "loss": 0.7813,
      "step": 21510
    },
    {
      "epoch": 3.979557703029177,
      "grad_norm": 1.0206513404846191,
      "learning_rate": 5.645293298273124e-10,
      "loss": 0.8759,
      "step": 21511
    },
    {
      "epoch": 3.9797435420925478,
      "grad_norm": 0.8693331480026245,
      "learning_rate": 5.165070274260053e-10,
      "loss": 0.72,
      "step": 21512
    },
    {
      "epoch": 3.979929381155919,
      "grad_norm": 0.9240720868110657,
      "learning_rate": 4.706190423808821e-10,
      "loss": 0.7878,
      "step": 21513
    },
    {
      "epoch": 3.98011522021929,
      "grad_norm": 0.8996371626853943,
      "learning_rate": 4.268653756911434e-10,
      "loss": 0.851,
      "step": 21514
    },
    {
      "epoch": 3.9803010592826613,
      "grad_norm": 1.18083655834198,
      "learning_rate": 3.852460282893766e-10,
      "loss": 0.8969,
      "step": 21515
    },
    {
      "epoch": 3.980486898346032,
      "grad_norm": 0.8848041296005249,
      "learning_rate": 3.4576100105265796e-10,
      "loss": 0.9113,
      "step": 21516
    },
    {
      "epoch": 3.9806727374094035,
      "grad_norm": 1.0805909633636475,
      "learning_rate": 3.084102948247569e-10,
      "loss": 0.8013,
      "step": 21517
    },
    {
      "epoch": 3.9808585764727744,
      "grad_norm": 1.1475094556808472,
      "learning_rate": 2.731939104161363e-10,
      "loss": 0.7648,
      "step": 21518
    },
    {
      "epoch": 3.9810444155361457,
      "grad_norm": 0.8849775195121765,
      "learning_rate": 2.4011184855954327e-10,
      "loss": 0.6718,
      "step": 21519
    },
    {
      "epoch": 3.9812302545995166,
      "grad_norm": 0.8735383749008179,
      "learning_rate": 2.0916410997662283e-10,
      "loss": 0.683,
      "step": 21520
    },
    {
      "epoch": 3.981416093662888,
      "grad_norm": 0.8057239055633545,
      "learning_rate": 1.803506953335088e-10,
      "loss": 0.7949,
      "step": 21521
    },
    {
      "epoch": 3.9816019327262593,
      "grad_norm": 0.8094781041145325,
      "learning_rate": 1.5367160521861935e-10,
      "loss": 0.6191,
      "step": 21522
    },
    {
      "epoch": 3.98178777178963,
      "grad_norm": 0.9692800045013428,
      "learning_rate": 1.2912684022037269e-10,
      "loss": 0.9201,
      "step": 21523
    },
    {
      "epoch": 3.981973610853001,
      "grad_norm": 1.0223878622055054,
      "learning_rate": 1.0671640086057366e-10,
      "loss": 0.9353,
      "step": 21524
    }
  ],
  "logging_steps": 1,
  "max_steps": 21524,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 5381,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.362292502065971e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}