{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4872267579668159,
  "eval_steps": 500,
  "global_step": 18500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.633658151171978e-05,
      "grad_norm": 1.25602126121521,
      "learning_rate": 4.999868317092442e-05,
      "loss": 2.5915,
      "step": 1
    },
    {
      "epoch": 5.267316302343956e-05,
      "grad_norm": 1.4423277378082275,
      "learning_rate": 4.999736634184883e-05,
      "loss": 1.4082,
      "step": 2
    },
    {
      "epoch": 7.900974453515933e-05,
      "grad_norm": NaN,
      "learning_rate": 4.999736634184883e-05,
      "loss": 2.7673,
      "step": 3
    },
    {
      "epoch": 0.00010534632604687912,
      "grad_norm": 1.9636341333389282,
      "learning_rate": 4.999604951277325e-05,
      "loss": 2.1738,
      "step": 4
    },
    {
      "epoch": 0.0001316829075585989,
      "grad_norm": NaN,
      "learning_rate": 4.999604951277325e-05,
      "loss": 1.7754,
      "step": 5
    },
    {
      "epoch": 0.00015801948907031866,
      "grad_norm": 2.365417242050171,
      "learning_rate": 4.999473268369766e-05,
      "loss": 0.8906,
      "step": 6
    },
    {
      "epoch": 0.00018435607058203845,
      "grad_norm": 3.35546612739563,
      "learning_rate": 4.9993415854622074e-05,
      "loss": 1.3068,
      "step": 7
    },
    {
      "epoch": 0.00021069265209375823,
      "grad_norm": 1.5731785297393799,
      "learning_rate": 4.999209902554648e-05,
      "loss": 1.8704,
      "step": 8
    },
    {
      "epoch": 0.00023702923360547802,
      "grad_norm": NaN,
      "learning_rate": 4.999209902554648e-05,
      "loss": 1.7605,
      "step": 9
    },
    {
      "epoch": 0.0002633658151171978,
      "grad_norm": 1.6602925062179565,
      "learning_rate": 4.99907821964709e-05,
      "loss": 1.4179,
      "step": 10
    },
    {
      "epoch": 0.00028970239662891756,
      "grad_norm": 1.1264127492904663,
      "learning_rate": 4.9989465367395314e-05,
      "loss": 2.0449,
      "step": 11
    },
    {
      "epoch": 0.0003160389781406373,
      "grad_norm": 1.6265169382095337,
      "learning_rate": 4.998814853831973e-05,
      "loss": 1.6321,
      "step": 12
    },
    {
      "epoch": 0.00034237555965235713,
      "grad_norm": 1.510681390762329,
      "learning_rate": 4.9986831709244145e-05,
      "loss": 2.4123,
      "step": 13
    },
    {
      "epoch": 0.0003687121411640769,
      "grad_norm": 1.0148299932479858,
      "learning_rate": 4.9985514880168554e-05,
      "loss": 1.9112,
      "step": 14
    },
    {
      "epoch": 0.0003950487226757967,
      "grad_norm": 1.2252342700958252,
      "learning_rate": 4.9984198051092976e-05,
      "loss": 1.7106,
      "step": 15
    },
    {
      "epoch": 0.00042138530418751646,
      "grad_norm": 2.102924346923828,
      "learning_rate": 4.9982881222017385e-05,
      "loss": 1.2634,
      "step": 16
    },
    {
      "epoch": 0.0004477218856992362,
      "grad_norm": 1.4299705028533936,
      "learning_rate": 4.99815643929418e-05,
      "loss": 2.1772,
      "step": 17
    },
    {
      "epoch": 0.00047405846721095603,
      "grad_norm": 1.1273866891860962,
      "learning_rate": 4.998024756386621e-05,
      "loss": 2.1013,
      "step": 18
    },
    {
      "epoch": 0.0005003950487226758,
      "grad_norm": 2.7761473655700684,
      "learning_rate": 4.9978930734790625e-05,
      "loss": 1.3722,
      "step": 19
    },
    {
      "epoch": 0.0005267316302343956,
      "grad_norm": 4.346610069274902,
      "learning_rate": 4.997761390571504e-05,
      "loss": 2.1313,
      "step": 20
    },
    {
      "epoch": 0.0005530682117461153,
      "grad_norm": 1.7366644144058228,
      "learning_rate": 4.9976297076639456e-05,
      "loss": 2.2213,
      "step": 21
    },
    {
      "epoch": 0.0005794047932578351,
      "grad_norm": 1.5845476388931274,
      "learning_rate": 4.997498024756387e-05,
      "loss": 1.8117,
      "step": 22
    },
    {
      "epoch": 0.0006057413747695549,
      "grad_norm": 4.5487775802612305,
      "learning_rate": 4.997366341848828e-05,
      "loss": 1.1827,
      "step": 23
    },
    {
      "epoch": 0.0006320779562812746,
      "grad_norm": 0.8642082214355469,
      "learning_rate": 4.9972346589412696e-05,
      "loss": 1.695,
      "step": 24
    },
    {
      "epoch": 0.0006584145377929945,
      "grad_norm": 1.0657001733779907,
      "learning_rate": 4.997102976033711e-05,
      "loss": 2.2334,
      "step": 25
    },
    {
      "epoch": 0.0006847511193047143,
      "grad_norm": 1.2822493314743042,
      "learning_rate": 4.996971293126153e-05,
      "loss": 1.9787,
      "step": 26
    },
    {
      "epoch": 0.0007110877008164341,
      "grad_norm": 1.2216416597366333,
      "learning_rate": 4.9968396102185936e-05,
      "loss": 2.0054,
      "step": 27
    },
    {
      "epoch": 0.0007374242823281538,
      "grad_norm": 1.1048316955566406,
      "learning_rate": 4.996707927311035e-05,
      "loss": 1.4634,
      "step": 28
    },
    {
      "epoch": 0.0007637608638398736,
      "grad_norm": 1.488633394241333,
      "learning_rate": 4.996576244403477e-05,
      "loss": 1.6665,
      "step": 29
    },
    {
      "epoch": 0.0007900974453515934,
      "grad_norm": 1.2460627555847168,
      "learning_rate": 4.996444561495918e-05,
      "loss": 1.8692,
      "step": 30
    },
    {
      "epoch": 0.0008164340268633131,
      "grad_norm": 2.1408908367156982,
      "learning_rate": 4.996312878588359e-05,
      "loss": 1.8691,
      "step": 31
    },
    {
      "epoch": 0.0008427706083750329,
      "grad_norm": 1.1469541788101196,
      "learning_rate": 4.996181195680801e-05,
      "loss": 1.669,
      "step": 32
    },
    {
      "epoch": 0.0008691071898867527,
      "grad_norm": 4.090087890625,
      "learning_rate": 4.996049512773242e-05,
      "loss": 1.0411,
      "step": 33
    },
    {
      "epoch": 0.0008954437713984724,
      "grad_norm": 2.047508716583252,
      "learning_rate": 4.995917829865684e-05,
      "loss": 2.0276,
      "step": 34
    },
    {
      "epoch": 0.0009217803529101923,
      "grad_norm": 1.4882947206497192,
      "learning_rate": 4.9957861469581254e-05,
      "loss": 1.3422,
      "step": 35
    },
    {
      "epoch": 0.0009481169344219121,
      "grad_norm": 2.8444130420684814,
      "learning_rate": 4.995654464050566e-05,
      "loss": 1.6189,
      "step": 36
    },
    {
      "epoch": 0.0009744535159336318,
      "grad_norm": 1.9858633279800415,
      "learning_rate": 4.995522781143008e-05,
      "loss": 1.7378,
      "step": 37
    },
    {
      "epoch": 0.0010007900974453516,
      "grad_norm": 2.514693260192871,
      "learning_rate": 4.995391098235449e-05,
      "loss": 0.9876,
      "step": 38
    },
    {
      "epoch": 0.0010271266789570713,
      "grad_norm": 1.6809574365615845,
      "learning_rate": 4.995259415327891e-05,
      "loss": 1.6963,
      "step": 39
    },
    {
      "epoch": 0.0010534632604687912,
      "grad_norm": 1.7100722789764404,
      "learning_rate": 4.995127732420332e-05,
      "loss": 1.9217,
      "step": 40
    },
    {
      "epoch": 0.001079799841980511,
      "grad_norm": 5.513702392578125,
      "learning_rate": 4.9949960495127734e-05,
      "loss": 1.5183,
      "step": 41
    },
    {
      "epoch": 0.0011061364234922306,
      "grad_norm": 2.2067952156066895,
      "learning_rate": 4.994864366605215e-05,
      "loss": 1.5994,
      "step": 42
    },
    {
      "epoch": 0.0011324730050039505,
      "grad_norm": 1.6485905647277832,
      "learning_rate": 4.994732683697656e-05,
      "loss": 1.8023,
      "step": 43
    },
    {
      "epoch": 0.0011588095865156702,
      "grad_norm": 2.0764386653900146,
      "learning_rate": 4.994601000790098e-05,
      "loss": 2.0264,
      "step": 44
    },
    {
      "epoch": 0.00118514616802739,
      "grad_norm": 2.3295562267303467,
      "learning_rate": 4.994469317882539e-05,
      "loss": 1.438,
      "step": 45
    },
    {
      "epoch": 0.0012114827495391099,
      "grad_norm": 1.6137781143188477,
      "learning_rate": 4.9943376349749805e-05,
      "loss": 2.0789,
      "step": 46
    },
    {
      "epoch": 0.0012378193310508296,
      "grad_norm": 3.2873547077178955,
      "learning_rate": 4.9942059520674214e-05,
      "loss": 1.6047,
      "step": 47
    },
    {
      "epoch": 0.0012641559125625493,
      "grad_norm": 1.6551767587661743,
      "learning_rate": 4.9940742691598637e-05,
      "loss": 1.678,
      "step": 48
    },
    {
      "epoch": 0.0012904924940742692,
      "grad_norm": 2.1045918464660645,
      "learning_rate": 4.9939425862523045e-05,
      "loss": 2.122,
      "step": 49
    },
    {
      "epoch": 0.001316829075585989,
      "grad_norm": 1.585898995399475,
      "learning_rate": 4.993810903344746e-05,
      "loss": 1.3005,
      "step": 50
    },
    {
      "epoch": 0.0013431656570977088,
      "grad_norm": 2.065453052520752,
      "learning_rate": 4.9936792204371877e-05,
      "loss": 2.0565,
      "step": 51
    },
    {
      "epoch": 0.0013695022386094285,
      "grad_norm": 2.091850519180298,
      "learning_rate": 4.9935475375296285e-05,
      "loss": 2.1221,
      "step": 52
    },
    {
      "epoch": 0.0013958388201211482,
      "grad_norm": 1.811124563217163,
      "learning_rate": 4.993415854622071e-05,
      "loss": 2.0501,
      "step": 53
    },
    {
      "epoch": 0.0014221754016328682,
      "grad_norm": 2.2109591960906982,
      "learning_rate": 4.9932841717145117e-05,
      "loss": 1.5615,
      "step": 54
    },
    {
      "epoch": 0.0014485119831445879,
      "grad_norm": 1.7171549797058105,
      "learning_rate": 4.993152488806953e-05,
      "loss": 2.0356,
      "step": 55
    },
    {
      "epoch": 0.0014748485646563076,
      "grad_norm": 4.514270305633545,
      "learning_rate": 4.993020805899394e-05,
      "loss": 1.6096,
      "step": 56
    },
    {
      "epoch": 0.0015011851461680275,
      "grad_norm": 4.055466175079346,
      "learning_rate": 4.9928891229918357e-05,
      "loss": 1.5004,
      "step": 57
    },
    {
      "epoch": 0.0015275217276797472,
      "grad_norm": NaN,
      "learning_rate": 4.9928891229918357e-05,
      "loss": 1.6247,
      "step": 58
    },
    {
      "epoch": 0.001553858309191467,
      "grad_norm": 1.5687782764434814,
      "learning_rate": 4.992757440084277e-05,
      "loss": 2.2262,
      "step": 59
    },
    {
      "epoch": 0.0015801948907031868,
      "grad_norm": 1.6692463159561157,
      "learning_rate": 4.992625757176719e-05,
      "loss": 1.7712,
      "step": 60
    },
    {
      "epoch": 0.0016065314722149065,
      "grad_norm": 3.837777853012085,
      "learning_rate": 4.99249407426916e-05,
      "loss": 1.3561,
      "step": 61
    },
    {
      "epoch": 0.0016328680537266262,
      "grad_norm": 4.855952739715576,
      "learning_rate": 4.992362391361601e-05,
      "loss": 0.9074,
      "step": 62
    },
    {
      "epoch": 0.0016592046352383461,
      "grad_norm": 3.2597079277038574,
      "learning_rate": 4.9922307084540435e-05,
      "loss": 2.1736,
      "step": 63
    },
    {
      "epoch": 0.0016855412167500658,
      "grad_norm": 2.073603391647339,
      "learning_rate": 4.9920990255464843e-05,
      "loss": 1.8285,
      "step": 64
    },
    {
      "epoch": 0.0017118777982617856,
      "grad_norm": 1.8987481594085693,
      "learning_rate": 4.991967342638926e-05,
      "loss": 1.7371,
      "step": 65
    },
    {
      "epoch": 0.0017382143797735055,
      "grad_norm": 2.7572085857391357,
      "learning_rate": 4.991835659731367e-05,
      "loss": 1.8387,
      "step": 66
    },
    {
      "epoch": 0.0017645509612852252,
      "grad_norm": 4.392783164978027,
      "learning_rate": 4.9917039768238083e-05,
      "loss": 2.2349,
      "step": 67
    },
    {
      "epoch": 0.0017908875427969449,
      "grad_norm": 3.262601613998413,
      "learning_rate": 4.99157229391625e-05,
      "loss": 1.9045,
      "step": 68
    },
    {
      "epoch": 0.0018172241243086648,
      "grad_norm": 4.613892078399658,
      "learning_rate": 4.9914406110086915e-05,
      "loss": 0.9683,
      "step": 69
    },
    {
      "epoch": 0.0018435607058203845,
      "grad_norm": 3.338005542755127,
      "learning_rate": 4.991308928101133e-05,
      "loss": 1.6835,
      "step": 70
    },
    {
      "epoch": 0.0018698972873321042,
      "grad_norm": 1.9964314699172974,
      "learning_rate": 4.991177245193574e-05,
      "loss": 1.5167,
      "step": 71
    },
    {
      "epoch": 0.0018962338688438241,
      "grad_norm": 2.4896624088287354,
      "learning_rate": 4.9910455622860155e-05,
      "loss": 1.8422,
      "step": 72
    },
    {
      "epoch": 0.0019225704503555438,
      "grad_norm": 6.072330951690674,
      "learning_rate": 4.990913879378457e-05,
      "loss": 1.5565,
      "step": 73
    },
    {
      "epoch": 0.0019489070318672635,
      "grad_norm": 2.1096174716949463,
      "learning_rate": 4.9907821964708986e-05,
      "loss": 1.5544,
      "step": 74
    },
    {
      "epoch": 0.0019752436133789832,
      "grad_norm": 3.130415201187134,
      "learning_rate": 4.9906505135633395e-05,
      "loss": 2.1596,
      "step": 75
    },
    {
      "epoch": 0.002001580194890703,
      "grad_norm": 9.269598007202148,
      "learning_rate": 4.990518830655781e-05,
      "loss": 1.0918,
      "step": 76
    },
    {
      "epoch": 0.002027916776402423,
      "grad_norm": 2.547788381576538,
      "learning_rate": 4.9903871477482226e-05,
      "loss": 1.8652,
      "step": 77
    },
    {
      "epoch": 0.0020542533579141426,
      "grad_norm": 2.4005014896392822,
      "learning_rate": 4.990255464840664e-05,
      "loss": 1.8942,
      "step": 78
    },
    {
      "epoch": 0.0020805899394258625,
      "grad_norm": 1.9952259063720703,
      "learning_rate": 4.990123781933106e-05,
      "loss": 1.9271,
      "step": 79
    },
    {
      "epoch": 0.0021069265209375824,
      "grad_norm": 2.127342939376831,
      "learning_rate": 4.9899920990255466e-05,
      "loss": 1.7014,
      "step": 80
    },
    {
      "epoch": 0.002133263102449302,
      "grad_norm": 1.9199120998382568,
      "learning_rate": 4.989860416117988e-05,
      "loss": 1.9958,
      "step": 81
    },
    {
      "epoch": 0.002159599683961022,
      "grad_norm": 4.213884353637695,
      "learning_rate": 4.98972873321043e-05,
      "loss": 0.6952,
      "step": 82
    },
    {
      "epoch": 0.0021859362654727417,
      "grad_norm": 3.3724217414855957,
      "learning_rate": 4.989597050302871e-05,
      "loss": 1.3471,
      "step": 83
    },
    {
      "epoch": 0.0022122728469844612,
      "grad_norm": 1.9994274377822876,
      "learning_rate": 4.989465367395312e-05,
      "loss": 1.8571,
      "step": 84
    },
    {
      "epoch": 0.002238609428496181,
      "grad_norm": 3.1660397052764893,
      "learning_rate": 4.989333684487754e-05,
      "loss": 1.7784,
      "step": 85
    },
    {
      "epoch": 0.002264946010007901,
      "grad_norm": 2.1498875617980957,
      "learning_rate": 4.9892020015801946e-05,
      "loss": 1.8134,
      "step": 86
    },
    {
      "epoch": 0.0022912825915196206,
      "grad_norm": 3.4811267852783203,
      "learning_rate": 4.989070318672637e-05,
      "loss": 1.863,
      "step": 87
    },
    {
      "epoch": 0.0023176191730313405,
      "grad_norm": 4.082001686096191,
      "learning_rate": 4.988938635765078e-05,
      "loss": 0.6746,
      "step": 88
    },
    {
      "epoch": 0.0023439557545430604,
      "grad_norm": 2.0179851055145264,
      "learning_rate": 4.988806952857519e-05,
      "loss": 1.8533,
      "step": 89
    },
    {
      "epoch": 0.00237029233605478,
      "grad_norm": 2.6662609577178955,
      "learning_rate": 4.988675269949961e-05,
      "loss": 2.4785,
      "step": 90
    },
    {
      "epoch": 0.0023966289175665,
      "grad_norm": 2.4066240787506104,
      "learning_rate": 4.988543587042402e-05,
      "loss": 2.0299,
      "step": 91
    },
    {
      "epoch": 0.0024229654990782197,
      "grad_norm": 2.665332555770874,
      "learning_rate": 4.988411904134844e-05,
      "loss": 1.9629,
      "step": 92
    },
    {
      "epoch": 0.0024493020805899392,
      "grad_norm": 2.670659065246582,
      "learning_rate": 4.988280221227285e-05,
      "loss": 2.6515,
      "step": 93
    },
    {
      "epoch": 0.002475638662101659,
      "grad_norm": 2.028151035308838,
      "learning_rate": 4.9881485383197264e-05,
      "loss": 2.3114,
      "step": 94
    },
    {
      "epoch": 0.002501975243613379,
      "grad_norm": 1.8326789140701294,
      "learning_rate": 4.988016855412167e-05,
      "loss": 2.2585,
      "step": 95
    },
    {
      "epoch": 0.0025283118251250986,
      "grad_norm": 2.9877381324768066,
      "learning_rate": 4.9878851725046095e-05,
      "loss": 0.6073,
      "step": 96
    },
    {
      "epoch": 0.0025546484066368185,
      "grad_norm": 5.428811550140381,
      "learning_rate": 4.9877534895970504e-05,
      "loss": 2.1438,
      "step": 97
    },
    {
      "epoch": 0.0025809849881485384,
      "grad_norm": 6.648031234741211,
      "learning_rate": 4.987621806689492e-05,
      "loss": 2.4835,
      "step": 98
    },
    {
      "epoch": 0.002607321569660258,
      "grad_norm": 6.937345504760742,
      "learning_rate": 4.9874901237819335e-05,
      "loss": 1.4094,
      "step": 99
    },
    {
      "epoch": 0.002633658151171978,
      "grad_norm": 2.093762159347534,
      "learning_rate": 4.9873584408743744e-05,
      "loss": 1.5811,
      "step": 100
    },
    {
      "epoch": 0.0026599947326836977,
      "grad_norm": 5.363945007324219,
      "learning_rate": 4.9872267579668166e-05,
      "loss": 1.472,
      "step": 101
    },
    {
      "epoch": 0.0026863313141954176,
      "grad_norm": 2.478797674179077,
      "learning_rate": 4.9870950750592575e-05,
      "loss": 2.3101,
      "step": 102
    },
    {
      "epoch": 0.002712667895707137,
      "grad_norm": 6.754124641418457,
      "learning_rate": 4.986963392151699e-05,
      "loss": 2.2972,
      "step": 103
    },
    {
      "epoch": 0.002739004477218857,
      "grad_norm": 4.061160087585449,
      "learning_rate": 4.98683170924414e-05,
      "loss": 1.0694,
      "step": 104
    },
    {
      "epoch": 0.002765341058730577,
      "grad_norm": 3.0984303951263428,
      "learning_rate": 4.9867000263365815e-05,
      "loss": 1.7015,
      "step": 105
    },
    {
      "epoch": 0.0027916776402422965,
      "grad_norm": 4.187580108642578,
      "learning_rate": 4.986568343429023e-05,
      "loss": 2.1395,
      "step": 106
    },
    {
      "epoch": 0.0028180142217540164,
      "grad_norm": 3.4013631343841553,
      "learning_rate": 4.9864366605214646e-05,
      "loss": 2.0131,
      "step": 107
    },
    {
      "epoch": 0.0028443508032657363,
      "grad_norm": 2.3839163780212402,
      "learning_rate": 4.986304977613906e-05,
      "loss": 1.7257,
      "step": 108
    },
    {
      "epoch": 0.002870687384777456,
      "grad_norm": 3.7074756622314453,
      "learning_rate": 4.986173294706347e-05,
      "loss": 1.8269,
      "step": 109
    },
    {
      "epoch": 0.0028970239662891757,
      "grad_norm": 3.0593509674072266,
      "learning_rate": 4.9860416117987886e-05,
      "loss": 1.8013,
      "step": 110
    },
    {
      "epoch": 0.0029233605478008956,
      "grad_norm": 4.329958438873291,
      "learning_rate": 4.98590992889123e-05,
      "loss": 2.7713,
      "step": 111
    },
    {
      "epoch": 0.002949697129312615,
      "grad_norm": 2.1304702758789062,
      "learning_rate": 4.985778245983672e-05,
      "loss": 1.8745,
      "step": 112
    },
    {
      "epoch": 0.002976033710824335,
      "grad_norm": 2.2222111225128174,
      "learning_rate": 4.9856465630761126e-05,
      "loss": 1.8628,
      "step": 113
    },
    {
      "epoch": 0.003002370292336055,
      "grad_norm": 2.2765445709228516,
      "learning_rate": 4.985514880168554e-05,
      "loss": 2.5081,
      "step": 114
    },
    {
      "epoch": 0.0030287068738477745,
      "grad_norm": 2.0583174228668213,
      "learning_rate": 4.985383197260996e-05,
      "loss": 1.906,
      "step": 115
    },
    {
      "epoch": 0.0030550434553594944,
      "grad_norm": 1.9065154790878296,
      "learning_rate": 4.985251514353437e-05,
      "loss": 1.6274,
      "step": 116
    },
    {
      "epoch": 0.0030813800368712143,
      "grad_norm": 2.5122358798980713,
      "learning_rate": 4.985119831445879e-05,
      "loss": 2.2167,
      "step": 117
    },
    {
      "epoch": 0.003107716618382934,
      "grad_norm": 2.5497119426727295,
      "learning_rate": 4.98498814853832e-05,
      "loss": 1.5221,
      "step": 118
    },
    {
      "epoch": 0.0031340531998946537,
      "grad_norm": 5.591714382171631,
      "learning_rate": 4.984856465630761e-05,
      "loss": 1.5908,
      "step": 119
    },
    {
      "epoch": 0.0031603897814063736,
      "grad_norm": 2.0516998767852783,
      "learning_rate": 4.984724782723203e-05,
      "loss": 2.1565,
      "step": 120
    },
    {
      "epoch": 0.003186726362918093,
      "grad_norm": 2.954334259033203,
      "learning_rate": 4.9845930998156444e-05,
      "loss": 1.9972,
      "step": 121
    },
    {
      "epoch": 0.003213062944429813,
      "grad_norm": 3.947854518890381,
      "learning_rate": 4.984461416908085e-05,
      "loss": 0.8273,
      "step": 122
    },
    {
      "epoch": 0.003239399525941533,
      "grad_norm": 2.1221768856048584,
      "learning_rate": 4.984329734000527e-05,
      "loss": 2.0991,
      "step": 123
    },
    {
      "epoch": 0.0032657361074532524,
      "grad_norm": 4.618890285491943,
      "learning_rate": 4.9841980510929684e-05,
      "loss": 0.8006,
      "step": 124
    },
    {
      "epoch": 0.0032920726889649724,
      "grad_norm": 2.2099010944366455,
      "learning_rate": 4.98406636818541e-05,
      "loss": 1.8216,
      "step": 125
    },
    {
      "epoch": 0.0033184092704766923,
      "grad_norm": 5.8832292556762695,
      "learning_rate": 4.9839346852778516e-05,
      "loss": 2.2214,
      "step": 126
    },
    {
      "epoch": 0.0033447458519884118,
      "grad_norm": 1.9785484075546265,
      "learning_rate": 4.9838030023702924e-05,
      "loss": 1.4308,
      "step": 127
    },
    {
      "epoch": 0.0033710824335001317,
      "grad_norm": 2.9469151496887207,
      "learning_rate": 4.983671319462734e-05,
      "loss": 2.159,
      "step": 128
    },
    {
      "epoch": 0.0033974190150118516,
      "grad_norm": 3.7625527381896973,
      "learning_rate": 4.9835396365551756e-05,
      "loss": 1.6932,
      "step": 129
    },
    {
      "epoch": 0.003423755596523571,
      "grad_norm": 3.0249016284942627,
      "learning_rate": 4.983407953647617e-05,
      "loss": 1.8042,
      "step": 130
    },
    {
      "epoch": 0.003450092178035291,
      "grad_norm": 2.883127450942993,
      "learning_rate": 4.983276270740058e-05,
      "loss": 2.1843,
      "step": 131
    },
    {
      "epoch": 0.003476428759547011,
      "grad_norm": 3.135927200317383,
      "learning_rate": 4.9831445878324996e-05,
      "loss": 1.9215,
      "step": 132
    },
    {
      "epoch": 0.0035027653410587304,
      "grad_norm": 1.9830724000930786,
      "learning_rate": 4.9830129049249404e-05,
      "loss": 1.5282,
      "step": 133
    },
    {
      "epoch": 0.0035291019225704504,
      "grad_norm": 4.631160259246826,
      "learning_rate": 4.982881222017383e-05,
      "loss": 1.9417,
      "step": 134
    },
    {
      "epoch": 0.0035554385040821703,
      "grad_norm": 2.9444193840026855,
      "learning_rate": 4.9827495391098236e-05,
      "loss": 1.5634,
      "step": 135
    },
    {
      "epoch": 0.0035817750855938898,
      "grad_norm": 5.769080638885498,
      "learning_rate": 4.982617856202265e-05,
      "loss": 1.2814,
      "step": 136
    },
    {
      "epoch": 0.0036081116671056097,
      "grad_norm": 3.2159616947174072,
      "learning_rate": 4.982486173294707e-05,
      "loss": 2.2346,
      "step": 137
    },
    {
      "epoch": 0.0036344482486173296,
      "grad_norm": 3.8492844104766846,
      "learning_rate": 4.9823544903871476e-05,
      "loss": 0.8781,
      "step": 138
    },
    {
      "epoch": 0.003660784830129049,
      "grad_norm": 2.716857433319092,
      "learning_rate": 4.98222280747959e-05,
      "loss": 1.17,
      "step": 139
    },
    {
      "epoch": 0.003687121411640769,
      "grad_norm": 2.482445001602173,
      "learning_rate": 4.982091124572031e-05,
      "loss": 2.1708,
      "step": 140
    },
    {
      "epoch": 0.003713457993152489,
      "grad_norm": 5.520145893096924,
      "learning_rate": 4.981959441664472e-05,
      "loss": 0.9422,
      "step": 141
    },
    {
      "epoch": 0.0037397945746642084,
      "grad_norm": 2.650184392929077,
      "learning_rate": 4.981827758756913e-05,
      "loss": 1.5212,
      "step": 142
    },
    {
      "epoch": 0.0037661311561759283,
      "grad_norm": 3.017289161682129,
      "learning_rate": 4.981696075849355e-05,
      "loss": 2.0877,
      "step": 143
    },
    {
      "epoch": 0.0037924677376876483,
      "grad_norm": 5.6365838050842285,
      "learning_rate": 4.981564392941796e-05,
      "loss": 1.8011,
      "step": 144
    },
    {
      "epoch": 0.0038188043191993678,
      "grad_norm": 1.9732463359832764,
      "learning_rate": 4.981432710034238e-05,
      "loss": 1.7964,
      "step": 145
    },
    {
      "epoch": 0.0038451409007110877,
      "grad_norm": 2.4963321685791016,
      "learning_rate": 4.9813010271266794e-05,
      "loss": 1.8117,
      "step": 146
    },
    {
      "epoch": 0.0038714774822228076,
      "grad_norm": 4.752861499786377,
      "learning_rate": 4.98116934421912e-05,
      "loss": 1.487,
      "step": 147
    },
    {
      "epoch": 0.003897814063734527,
      "grad_norm": 1.9943513870239258,
      "learning_rate": 4.9810376613115625e-05,
      "loss": 1.4832,
      "step": 148
    },
    {
      "epoch": 0.0039241506452462474,
      "grad_norm": 2.6101386547088623,
      "learning_rate": 4.9809059784040034e-05,
      "loss": 1.9399,
      "step": 149
    },
    {
      "epoch": 0.0039504872267579665,
      "grad_norm": 3.150451421737671,
      "learning_rate": 4.980774295496445e-05,
      "loss": 0.5693,
      "step": 150
    },
    {
      "epoch": 0.003976823808269686,
      "grad_norm": 2.680048704147339,
      "learning_rate": 4.980642612588886e-05,
      "loss": 1.704,
      "step": 151
    },
    {
      "epoch": 0.004003160389781406,
      "grad_norm": 3.1196019649505615,
      "learning_rate": 4.9805109296813274e-05,
      "loss": 1.541,
      "step": 152
    },
    {
      "epoch": 0.004029496971293126,
      "grad_norm": 3.6955935955047607,
      "learning_rate": 4.980379246773769e-05,
      "loss": 2.1527,
      "step": 153
    },
    {
      "epoch": 0.004055833552804846,
      "grad_norm": 5.005454063415527,
      "learning_rate": 4.9802475638662105e-05,
      "loss": 1.7925,
      "step": 154
    },
    {
      "epoch": 0.004082170134316566,
      "grad_norm": 2.632319927215576,
      "learning_rate": 4.980115880958652e-05,
      "loss": 2.8897,
      "step": 155
    },
    {
      "epoch": 0.004108506715828285,
      "grad_norm": 4.2010087966918945,
      "learning_rate": 4.979984198051093e-05,
      "loss": 0.841,
      "step": 156
    },
    {
      "epoch": 0.004134843297340005,
      "grad_norm": 2.79358172416687,
      "learning_rate": 4.9798525151435345e-05,
      "loss": 1.6884,
      "step": 157
    },
    {
      "epoch": 0.004161179878851725,
      "grad_norm": 3.154688596725464,
      "learning_rate": 4.979720832235976e-05,
      "loss": 1.0247,
      "step": 158
    },
    {
      "epoch": 0.004187516460363445,
      "grad_norm": 2.1128103733062744,
      "learning_rate": 4.9795891493284176e-05,
      "loss": 1.512,
      "step": 159
    },
    {
      "epoch": 0.004213853041875165,
      "grad_norm": 2.2439041137695312,
      "learning_rate": 4.9794574664208585e-05,
      "loss": 1.8068,
      "step": 160
    },
    {
      "epoch": 0.004240189623386885,
      "grad_norm": 2.2295446395874023,
      "learning_rate": 4.9793257835133e-05,
      "loss": 1.9697,
      "step": 161
    },
    {
      "epoch": 0.004266526204898604,
      "grad_norm": 2.5257863998413086,
      "learning_rate": 4.9791941006057416e-05,
      "loss": 1.8954,
      "step": 162
    },
    {
      "epoch": 0.004292862786410324,
      "grad_norm": 3.872302532196045,
      "learning_rate": 4.979062417698183e-05,
      "loss": 1.0661,
      "step": 163
    },
    {
      "epoch": 0.004319199367922044,
      "grad_norm": 3.818669080734253,
      "learning_rate": 4.978930734790625e-05,
      "loss": 2.4765,
      "step": 164
    },
    {
      "epoch": 0.004345535949433764,
      "grad_norm": 2.772634267807007,
      "learning_rate": 4.9787990518830656e-05,
      "loss": 2.3597,
      "step": 165
    },
    {
      "epoch": 0.0043718725309454835,
      "grad_norm": 2.113095760345459,
      "learning_rate": 4.978667368975507e-05,
      "loss": 1.6092,
      "step": 166
    },
    {
      "epoch": 0.004398209112457203,
      "grad_norm": 2.2079544067382812,
      "learning_rate": 4.978535686067949e-05,
      "loss": 1.3975,
      "step": 167
    },
    {
      "epoch": 0.0044245456939689225,
      "grad_norm": 2.840900421142578,
      "learning_rate": 4.97840400316039e-05,
      "loss": 1.766,
      "step": 168
    },
    {
      "epoch": 0.004450882275480642,
      "grad_norm": 3.2003445625305176,
      "learning_rate": 4.978272320252831e-05,
      "loss": 2.0482,
      "step": 169
    },
    {
      "epoch": 0.004477218856992362,
      "grad_norm": 4.071146488189697,
      "learning_rate": 4.978140637345273e-05,
      "loss": 1.7155,
      "step": 170
    },
    {
      "epoch": 0.004503555438504082,
      "grad_norm": 2.204969644546509,
      "learning_rate": 4.978008954437714e-05,
      "loss": 1.5785,
      "step": 171
    },
    {
      "epoch": 0.004529892020015802,
      "grad_norm": 2.2772178649902344,
      "learning_rate": 4.977877271530156e-05,
      "loss": 2.2593,
      "step": 172
    },
    {
      "epoch": 0.004556228601527522,
      "grad_norm": 3.250837564468384,
      "learning_rate": 4.9777455886225974e-05,
      "loss": 1.6909,
      "step": 173
    },
    {
      "epoch": 0.004582565183039241,
      "grad_norm": 3.1715753078460693,
      "learning_rate": 4.977613905715038e-05,
      "loss": 1.6968,
      "step": 174
    },
    {
      "epoch": 0.004608901764550961,
      "grad_norm": 4.073447227478027,
      "learning_rate": 4.97748222280748e-05,
      "loss": 2.8818,
      "step": 175
    },
    {
      "epoch": 0.004635238346062681,
      "grad_norm": 2.430662155151367,
      "learning_rate": 4.9773505398999214e-05,
      "loss": 1.9849,
      "step": 176
    },
    {
      "epoch": 0.004661574927574401,
      "grad_norm": 2.2059905529022217,
      "learning_rate": 4.977218856992363e-05,
      "loss": 1.9293,
      "step": 177
    },
    {
      "epoch": 0.004687911509086121,
      "grad_norm": 4.93207311630249,
      "learning_rate": 4.977087174084804e-05,
      "loss": 1.2192,
      "step": 178
    },
    {
      "epoch": 0.004714248090597841,
      "grad_norm": 3.036472797393799,
      "learning_rate": 4.9769554911772454e-05,
      "loss": 1.6647,
      "step": 179
    },
    {
      "epoch": 0.00474058467210956,
      "grad_norm": 2.2520482540130615,
      "learning_rate": 4.976823808269687e-05,
      "loss": 1.8166,
      "step": 180
    },
    {
      "epoch": 0.00476692125362128,
      "grad_norm": 5.294158935546875,
      "learning_rate": 4.9766921253621285e-05,
      "loss": 1.2037,
      "step": 181
    },
    {
      "epoch": 0.004793257835133,
      "grad_norm": 3.03607439994812,
      "learning_rate": 4.97656044245457e-05,
      "loss": 1.8333,
      "step": 182
    },
    {
      "epoch": 0.0048195944166447196,
      "grad_norm": 2.839785575866699,
      "learning_rate": 4.976428759547011e-05,
      "loss": 2.0289,
      "step": 183
    },
    {
      "epoch": 0.0048459309981564395,
      "grad_norm": 2.876850128173828,
      "learning_rate": 4.9762970766394525e-05,
      "loss": 1.7722,
      "step": 184
    },
    {
      "epoch": 0.004872267579668159,
      "grad_norm": 3.0148491859436035,
      "learning_rate": 4.9761653937318934e-05,
      "loss": 1.7479,
      "step": 185
    },
    {
      "epoch": 0.0048986041611798785,
      "grad_norm": 3.0036590099334717,
      "learning_rate": 4.9760337108243357e-05,
      "loss": 1.5007,
      "step": 186
    },
    {
      "epoch": 0.004924940742691598,
      "grad_norm": 5.28188943862915,
      "learning_rate": 4.9759020279167765e-05,
      "loss": 0.8105,
      "step": 187
    },
    {
      "epoch": 0.004951277324203318,
      "grad_norm": 2.7707080841064453,
      "learning_rate": 4.975770345009218e-05,
      "loss": 1.8496,
      "step": 188
    },
    {
      "epoch": 0.004977613905715038,
      "grad_norm": 2.996790647506714,
      "learning_rate": 4.975638662101659e-05,
      "loss": 2.58,
      "step": 189
    },
    {
      "epoch": 0.005003950487226758,
      "grad_norm": 2.5800158977508545,
      "learning_rate": 4.9755069791941005e-05,
      "loss": 1.6247,
      "step": 190
    },
    {
      "epoch": 0.005030287068738478,
      "grad_norm": 4.0949554443359375,
      "learning_rate": 4.975375296286542e-05,
      "loss": 1.8341,
      "step": 191
    },
    {
      "epoch": 0.005056623650250197,
      "grad_norm": 5.023255348205566,
      "learning_rate": 4.9752436133789837e-05,
      "loss": 2.1623,
      "step": 192
    },
    {
      "epoch": 0.005082960231761917,
      "grad_norm": 2.2133235931396484,
      "learning_rate": 4.975111930471425e-05,
      "loss": 1.6761,
      "step": 193
    },
    {
      "epoch": 0.005109296813273637,
      "grad_norm": 5.704970836639404,
      "learning_rate": 4.974980247563866e-05,
      "loss": 0.9858,
      "step": 194
    },
    {
      "epoch": 0.005135633394785357,
      "grad_norm": 1.9332650899887085,
      "learning_rate": 4.9748485646563083e-05,
      "loss": 1.2981,
      "step": 195
    },
    {
      "epoch": 0.005161969976297077,
      "grad_norm": 4.054973125457764,
      "learning_rate": 4.974716881748749e-05,
      "loss": 1.9202,
      "step": 196
    },
    {
      "epoch": 0.005188306557808797,
      "grad_norm": 3.442539691925049,
      "learning_rate": 4.974585198841191e-05,
      "loss": 1.8708,
      "step": 197
    },
    {
      "epoch": 0.005214643139320516,
      "grad_norm": 2.023343086242676,
      "learning_rate": 4.974453515933632e-05,
      "loss": 1.9631,
      "step": 198
    },
    {
      "epoch": 0.005240979720832236,
      "grad_norm": 3.7599754333496094,
      "learning_rate": 4.974321833026073e-05,
      "loss": 2.3993,
      "step": 199
    },
    {
      "epoch": 0.005267316302343956,
      "grad_norm": 2.6702961921691895,
      "learning_rate": 4.974190150118515e-05,
      "loss": 1.4371,
      "step": 200
    },
    {
      "epoch": 0.0052936528838556755,
      "grad_norm": 2.936985492706299,
      "learning_rate": 4.9740584672109563e-05,
      "loss": 1.654,
      "step": 201
    },
    {
      "epoch": 0.0053199894653673955,
      "grad_norm": 3.7769246101379395,
      "learning_rate": 4.973926784303398e-05,
      "loss": 2.0605,
      "step": 202
    },
    {
      "epoch": 0.005346326046879115,
      "grad_norm": 4.476908206939697,
      "learning_rate": 4.973795101395839e-05,
      "loss": 1.8792,
      "step": 203
    },
    {
      "epoch": 0.005372662628390835,
      "grad_norm": 2.43996524810791,
      "learning_rate": 4.9736634184882803e-05,
      "loss": 1.5303,
      "step": 204
    },
    {
      "epoch": 0.005398999209902554,
      "grad_norm": 2.9250011444091797,
      "learning_rate": 4.973531735580722e-05,
      "loss": 2.1826,
      "step": 205
    },
    {
      "epoch": 0.005425335791414274,
      "grad_norm": 3.787264585494995,
      "learning_rate": 4.9734000526731635e-05,
      "loss": 1.1572,
      "step": 206
    },
    {
      "epoch": 0.005451672372925994,
      "grad_norm": 2.1952826976776123,
      "learning_rate": 4.9732683697656043e-05,
      "loss": 1.9259,
      "step": 207
    },
    {
      "epoch": 0.005478008954437714,
      "grad_norm": 2.5176422595977783,
      "learning_rate": 4.973136686858046e-05,
      "loss": 1.8724,
      "step": 208
    },
    {
      "epoch": 0.005504345535949434,
      "grad_norm": 3.8239634037017822,
      "learning_rate": 4.9730050039504875e-05,
      "loss": 1.3196,
      "step": 209
    },
    {
      "epoch": 0.005530682117461154,
      "grad_norm": 3.663858652114868,
      "learning_rate": 4.972873321042929e-05,
      "loss": 1.0881,
      "step": 210
    },
    {
      "epoch": 0.005557018698972873,
      "grad_norm": 5.210198879241943,
      "learning_rate": 4.9727416381353706e-05,
      "loss": 1.4144,
      "step": 211
    },
    {
      "epoch": 0.005583355280484593,
      "grad_norm": 5.618483066558838,
      "learning_rate": 4.9726099552278115e-05,
      "loss": 1.3746,
      "step": 212
    },
    {
      "epoch": 0.005609691861996313,
      "grad_norm": 2.5995712280273438,
      "learning_rate": 4.972478272320253e-05,
      "loss": 1.839,
      "step": 213
    },
    {
      "epoch": 0.005636028443508033,
      "grad_norm": 4.0986785888671875,
      "learning_rate": 4.9723465894126946e-05,
      "loss": 2.2452,
      "step": 214
    },
    {
      "epoch": 0.005662365025019753,
      "grad_norm": 7.395411968231201,
      "learning_rate": 4.972214906505136e-05,
      "loss": 2.079,
      "step": 215
    },
    {
      "epoch": 0.005688701606531473,
      "grad_norm": 4.094386100769043,
      "learning_rate": 4.972083223597577e-05,
      "loss": 2.1434,
      "step": 216
    },
    {
      "epoch": 0.005715038188043192,
      "grad_norm": 3.1119894981384277,
      "learning_rate": 4.9719515406900186e-05,
      "loss": 1.4083,
      "step": 217
    },
    {
      "epoch": 0.005741374769554912,
      "grad_norm": 3.389418601989746,
      "learning_rate": 4.97181985778246e-05,
      "loss": 2.3632,
      "step": 218
    },
    {
      "epoch": 0.0057677113510666315,
      "grad_norm": 4.414575099945068,
      "learning_rate": 4.971688174874902e-05,
      "loss": 1.5154,
      "step": 219
    },
    {
      "epoch": 0.0057940479325783514,
      "grad_norm": 2.6311278343200684,
      "learning_rate": 4.971556491967343e-05,
      "loss": 1.9642,
      "step": 220
    },
    {
      "epoch": 0.005820384514090071,
      "grad_norm": 2.435178756713867,
      "learning_rate": 4.971424809059784e-05,
      "loss": 2.0197,
      "step": 221
    },
    {
      "epoch": 0.005846721095601791,
      "grad_norm": 6.44195556640625,
      "learning_rate": 4.971293126152226e-05,
      "loss": 1.4618,
      "step": 222
    },
    {
      "epoch": 0.00587305767711351,
      "grad_norm": 4.706697940826416,
      "learning_rate": 4.9711614432446666e-05,
      "loss": 1.185,
      "step": 223
    },
    {
      "epoch": 0.00589939425862523,
      "grad_norm": 2.524803876876831,
      "learning_rate": 4.971029760337109e-05,
      "loss": 2.1728,
      "step": 224
    },
    {
      "epoch": 0.00592573084013695,
      "grad_norm": 14.731494903564453,
      "learning_rate": 4.97089807742955e-05,
      "loss": 2.0406,
      "step": 225
    },
    {
      "epoch": 0.00595206742164867,
      "grad_norm": 3.8456811904907227,
      "learning_rate": 4.970766394521991e-05,
      "loss": 2.0363,
      "step": 226
    },
    {
      "epoch": 0.00597840400316039,
      "grad_norm": 5.158308982849121,
      "learning_rate": 4.970634711614433e-05,
      "loss": 1.2614,
      "step": 227
    },
    {
      "epoch": 0.00600474058467211,
      "grad_norm": 2.347411632537842,
      "learning_rate": 4.9705030287068744e-05,
      "loss": 1.5423,
      "step": 228
    },
    {
      "epoch": 0.006031077166183829,
      "grad_norm": 7.151821136474609,
      "learning_rate": 4.970371345799316e-05,
      "loss": 0.8542,
      "step": 229
    },
    {
      "epoch": 0.006057413747695549,
      "grad_norm": 6.39750337600708,
      "learning_rate": 4.970239662891757e-05,
      "loss": 0.706,
      "step": 230
    },
    {
      "epoch": 0.006083750329207269,
      "grad_norm": 3.0930991172790527,
      "learning_rate": 4.9701079799841984e-05,
      "loss": 1.631,
      "step": 231
    },
    {
      "epoch": 0.006110086910718989,
      "grad_norm": 3.5407567024230957,
      "learning_rate": 4.969976297076639e-05,
      "loss": 1.6488,
      "step": 232
    },
    {
      "epoch": 0.006136423492230709,
      "grad_norm": 4.05654239654541,
      "learning_rate": 4.9698446141690815e-05,
      "loss": 1.2322,
      "step": 233
    },
    {
      "epoch": 0.006162760073742429,
      "grad_norm": 3.4510419368743896,
      "learning_rate": 4.9697129312615224e-05,
      "loss": 1.9873,
      "step": 234
    },
    {
      "epoch": 0.006189096655254148,
      "grad_norm": 3.6991634368896484,
      "learning_rate": 4.969581248353964e-05,
      "loss": 2.6805,
      "step": 235
    },
    {
      "epoch": 0.006215433236765868,
      "grad_norm": 3.597952127456665,
      "learning_rate": 4.969449565446405e-05,
      "loss": 1.7013,
      "step": 236
    },
    {
      "epoch": 0.0062417698182775875,
      "grad_norm": 4.022460460662842,
      "learning_rate": 4.9693178825388464e-05,
      "loss": 1.2796,
      "step": 237
    },
    {
      "epoch": 0.006268106399789307,
      "grad_norm": 2.287949800491333,
      "learning_rate": 4.969186199631288e-05,
      "loss": 1.2856,
      "step": 238
    },
    {
      "epoch": 0.006294442981301027,
      "grad_norm": 3.069411516189575,
      "learning_rate": 4.9690545167237295e-05,
      "loss": 1.8174,
      "step": 239
    },
    {
      "epoch": 0.006320779562812747,
      "grad_norm": 4.457437038421631,
      "learning_rate": 4.968922833816171e-05,
      "loss": 1.6977,
      "step": 240
    },
    {
      "epoch": 0.006347116144324466,
      "grad_norm": 2.3935649394989014,
      "learning_rate": 4.968791150908612e-05,
      "loss": 2.2043,
      "step": 241
    },
    {
      "epoch": 0.006373452725836186,
      "grad_norm": 2.437356948852539,
      "learning_rate": 4.968659468001054e-05,
      "loss": 1.689,
      "step": 242
    },
    {
      "epoch": 0.006399789307347906,
      "grad_norm": 2.710400104522705,
      "learning_rate": 4.968527785093495e-05,
      "loss": 1.8618,
      "step": 243
    },
    {
      "epoch": 0.006426125888859626,
      "grad_norm": 6.757627487182617,
      "learning_rate": 4.9683961021859366e-05,
      "loss": 1.0029,
      "step": 244
    },
    {
      "epoch": 0.006452462470371346,
      "grad_norm": 3.6147267818450928,
      "learning_rate": 4.9682644192783775e-05,
      "loss": 2.1118,
      "step": 245
    },
    {
      "epoch": 0.006478799051883066,
      "grad_norm": 6.167456150054932,
      "learning_rate": 4.968132736370819e-05,
      "loss": 1.7851,
      "step": 246
    },
    {
      "epoch": 0.006505135633394785,
      "grad_norm": 2.6882426738739014,
      "learning_rate": 4.9680010534632606e-05,
      "loss": 1.7571,
      "step": 247
    },
    {
      "epoch": 0.006531472214906505,
      "grad_norm": 3.0543911457061768,
      "learning_rate": 4.967869370555702e-05,
      "loss": 1.9994,
      "step": 248
    },
    {
      "epoch": 0.006557808796418225,
      "grad_norm": 2.950988531112671,
      "learning_rate": 4.967737687648144e-05,
      "loss": 1.7448,
      "step": 249
    },
    {
      "epoch": 0.006584145377929945,
      "grad_norm": 2.491550922393799,
      "learning_rate": 4.9676060047405846e-05,
      "loss": 1.6253,
      "step": 250
    },
    {
      "epoch": 0.006610481959441665,
      "grad_norm": 2.45288348197937,
      "learning_rate": 4.967474321833026e-05,
      "loss": 1.7883,
      "step": 251
    },
    {
      "epoch": 0.006636818540953385,
      "grad_norm": 2.80439829826355,
      "learning_rate": 4.967342638925468e-05,
      "loss": 1.5507,
      "step": 252
    },
    {
      "epoch": 0.006663155122465104,
      "grad_norm": 3.8137927055358887,
      "learning_rate": 4.967210956017909e-05,
      "loss": 1.7122,
      "step": 253
    },
    {
      "epoch": 0.0066894917039768236,
      "grad_norm": 3.2373127937316895,
      "learning_rate": 4.96707927311035e-05,
      "loss": 0.5187,
      "step": 254
    },
    {
      "epoch": 0.0067158282854885435,
      "grad_norm": 3.006848096847534,
      "learning_rate": 4.966947590202792e-05,
      "loss": 2.3407,
      "step": 255
    },
    {
      "epoch": 0.006742164867000263,
      "grad_norm": 2.6437928676605225,
      "learning_rate": 4.966815907295233e-05,
      "loss": 1.3933,
      "step": 256
    },
    {
      "epoch": 0.006768501448511983,
      "grad_norm": 2.1438775062561035,
      "learning_rate": 4.966684224387675e-05,
      "loss": 1.3451,
      "step": 257
    },
    {
      "epoch": 0.006794838030023703,
      "grad_norm": 3.741879463195801,
      "learning_rate": 4.9665525414801164e-05,
      "loss": 1.9437,
      "step": 258
    },
    {
      "epoch": 0.006821174611535422,
      "grad_norm": 3.7661421298980713,
      "learning_rate": 4.966420858572557e-05,
      "loss": 1.866,
      "step": 259
    },
    {
      "epoch": 0.006847511193047142,
      "grad_norm": 3.1512551307678223,
      "learning_rate": 4.966289175664999e-05,
      "loss": 1.2263,
      "step": 260
    },
    {
      "epoch": 0.006873847774558862,
      "grad_norm": 5.102208137512207,
      "learning_rate": 4.9661574927574404e-05,
      "loss": 1.7225,
      "step": 261
    },
    {
      "epoch": 0.006900184356070582,
      "grad_norm": 2.772230386734009,
      "learning_rate": 4.966025809849882e-05,
      "loss": 2.1089,
      "step": 262
    },
    {
      "epoch": 0.006926520937582302,
      "grad_norm": 4.984233379364014,
      "learning_rate": 4.965894126942323e-05,
      "loss": 0.9502,
      "step": 263
    },
    {
      "epoch": 0.006952857519094022,
      "grad_norm": 3.336632490158081,
      "learning_rate": 4.9657624440347644e-05,
      "loss": 1.5431,
      "step": 264
    },
    {
      "epoch": 0.006979194100605741,
      "grad_norm": 4.93807315826416,
      "learning_rate": 4.965630761127206e-05,
      "loss": 0.5891,
      "step": 265
    },
    {
      "epoch": 0.007005530682117461,
      "grad_norm": 3.4254891872406006,
      "learning_rate": 4.9654990782196476e-05,
      "loss": 2.1058,
      "step": 266
    },
    {
      "epoch": 0.007031867263629181,
      "grad_norm": 2.414081573486328,
      "learning_rate": 4.965367395312089e-05,
      "loss": 1.9151,
      "step": 267
    },
    {
      "epoch": 0.007058203845140901,
      "grad_norm": 4.4041876792907715,
      "learning_rate": 4.96523571240453e-05,
      "loss": 0.6274,
      "step": 268
    },
    {
      "epoch": 0.007084540426652621,
      "grad_norm": 3.4176602363586426,
      "learning_rate": 4.9651040294969716e-05,
      "loss": 1.4568,
      "step": 269
    },
    {
      "epoch": 0.0071108770081643406,
      "grad_norm": 4.991017818450928,
      "learning_rate": 4.9649723465894124e-05,
      "loss": 1.3682,
      "step": 270
    },
    {
      "epoch": 0.0071372135896760605,
      "grad_norm": 2.882906675338745,
      "learning_rate": 4.964840663681855e-05,
      "loss": 1.9827,
      "step": 271
    },
    {
      "epoch": 0.0071635501711877795,
      "grad_norm": 2.41365385055542,
      "learning_rate": 4.9647089807742956e-05,
      "loss": 2.269,
      "step": 272
    },
    {
      "epoch": 0.0071898867526994995,
      "grad_norm": 3.56182599067688,
      "learning_rate": 4.964577297866737e-05,
      "loss": 1.4444,
      "step": 273
    },
    {
      "epoch": 0.007216223334211219,
      "grad_norm": 2.774643659591675,
      "learning_rate": 4.964445614959179e-05,
      "loss": 2.2455,
      "step": 274
    },
    {
      "epoch": 0.007242559915722939,
      "grad_norm": 6.077582836151123,
      "learning_rate": 4.96431393205162e-05,
      "loss": 2.4372,
      "step": 275
    },
    {
      "epoch": 0.007268896497234659,
      "grad_norm": 2.978055477142334,
      "learning_rate": 4.964182249144062e-05,
      "loss": 1.6934,
      "step": 276
    },
    {
      "epoch": 0.007295233078746379,
      "grad_norm": 3.786829710006714,
      "learning_rate": 4.964050566236503e-05,
      "loss": 1.9201,
      "step": 277
    },
    {
      "epoch": 0.007321569660258098,
      "grad_norm": 3.4524993896484375,
      "learning_rate": 4.963918883328944e-05,
      "loss": 1.4841,
      "step": 278
    },
    {
      "epoch": 0.007347906241769818,
      "grad_norm": 4.23499059677124,
      "learning_rate": 4.963787200421385e-05,
      "loss": 1.6539,
      "step": 279
    },
    {
      "epoch": 0.007374242823281538,
      "grad_norm": 2.3960883617401123,
      "learning_rate": 4.9636555175138274e-05,
      "loss": 2.087,
      "step": 280
    },
    {
      "epoch": 0.007400579404793258,
      "grad_norm": 4.268434524536133,
      "learning_rate": 4.963523834606268e-05,
      "loss": 1.4911,
      "step": 281
    },
    {
      "epoch": 0.007426915986304978,
      "grad_norm": 2.243865728378296,
      "learning_rate": 4.96339215169871e-05,
      "loss": 1.5874,
      "step": 282
    },
    {
      "epoch": 0.007453252567816698,
      "grad_norm": 2.7072665691375732,
      "learning_rate": 4.963260468791151e-05,
      "loss": 1.8963,
      "step": 283
    },
    {
      "epoch": 0.007479589149328417,
      "grad_norm": 3.8264856338500977,
      "learning_rate": 4.963128785883592e-05,
      "loss": 0.6007,
      "step": 284
    },
    {
      "epoch": 0.007505925730840137,
      "grad_norm": 3.1545920372009277,
      "learning_rate": 4.9629971029760345e-05,
      "loss": 1.6745,
      "step": 285
    },
    {
      "epoch": 0.007532262312351857,
      "grad_norm": 3.4124999046325684,
      "learning_rate": 4.9628654200684754e-05,
      "loss": 1.4972,
      "step": 286
    },
    {
      "epoch": 0.007558598893863577,
      "grad_norm": 2.751809597015381,
      "learning_rate": 4.962733737160917e-05,
      "loss": 1.8819,
      "step": 287
    },
    {
      "epoch": 0.0075849354753752965,
      "grad_norm": 4.699490070343018,
      "learning_rate": 4.962602054253358e-05,
      "loss": 1.8979,
      "step": 288
    },
    {
      "epoch": 0.0076112720568870165,
      "grad_norm": 2.683763027191162,
      "learning_rate": 4.9624703713457994e-05,
      "loss": 1.862,
      "step": 289
    },
    {
      "epoch": 0.0076376086383987355,
      "grad_norm": 4.744758605957031,
      "learning_rate": 4.962338688438241e-05,
      "loss": 1.603,
      "step": 290
    },
    {
      "epoch": 0.007663945219910455,
      "grad_norm": 3.8711984157562256,
      "learning_rate": 4.9622070055306825e-05,
      "loss": 1.6996,
      "step": 291
    },
    {
      "epoch": 0.007690281801422175,
      "grad_norm": 3.0812060832977295,
      "learning_rate": 4.9620753226231234e-05,
      "loss": 2.4187,
      "step": 292
    },
    {
      "epoch": 0.007716618382933895,
      "grad_norm": 3.9074389934539795,
      "learning_rate": 4.961943639715565e-05,
      "loss": 2.0815,
      "step": 293
    },
    {
      "epoch": 0.007742954964445615,
      "grad_norm": 2.894277811050415,
      "learning_rate": 4.9618119568080065e-05,
      "loss": 2.0652,
      "step": 294
    },
    {
      "epoch": 0.007769291545957335,
      "grad_norm": 3.9091920852661133,
      "learning_rate": 4.961680273900448e-05,
      "loss": 1.7737,
      "step": 295
    },
    {
      "epoch": 0.007795628127469054,
      "grad_norm": 1.9550001621246338,
      "learning_rate": 4.9615485909928896e-05,
      "loss": 1.3876,
      "step": 296
    },
    {
      "epoch": 0.007821964708980774,
      "grad_norm": 3.390186309814453,
      "learning_rate": 4.9614169080853305e-05,
      "loss": 2.3364,
      "step": 297
    },
    {
      "epoch": 0.007848301290492495,
      "grad_norm": 3.1299211978912354,
      "learning_rate": 4.961285225177772e-05,
      "loss": 2.1138,
      "step": 298
    },
    {
      "epoch": 0.007874637872004214,
      "grad_norm": 7.753682613372803,
      "learning_rate": 4.9611535422702136e-05,
      "loss": 0.6388,
      "step": 299
    },
    {
      "epoch": 0.007900974453515933,
      "grad_norm": 3.2066988945007324,
      "learning_rate": 4.961021859362655e-05,
      "loss": 2.3243,
      "step": 300
    },
    {
      "epoch": 0.007927311035027654,
      "grad_norm": 2.5140504837036133,
      "learning_rate": 4.960890176455096e-05,
      "loss": 1.968,
      "step": 301
    },
    {
      "epoch": 0.007953647616539373,
      "grad_norm": 9.608623504638672,
      "learning_rate": 4.9607584935475376e-05,
      "loss": 1.0477,
      "step": 302
    },
    {
      "epoch": 0.007979984198051094,
      "grad_norm": 4.970414638519287,
      "learning_rate": 4.960626810639979e-05,
      "loss": 2.2033,
      "step": 303
    },
    {
      "epoch": 0.008006320779562813,
      "grad_norm": 4.394804000854492,
      "learning_rate": 4.960495127732421e-05,
      "loss": 1.423,
      "step": 304
    },
    {
      "epoch": 0.008032657361074532,
      "grad_norm": 4.038106441497803,
      "learning_rate": 4.960363444824862e-05,
      "loss": 1.6763,
      "step": 305
    },
    {
      "epoch": 0.008058993942586253,
      "grad_norm": 2.361182451248169,
      "learning_rate": 4.960231761917303e-05,
      "loss": 1.8892,
      "step": 306
    },
    {
      "epoch": 0.008085330524097972,
      "grad_norm": 2.8867974281311035,
      "learning_rate": 4.960100079009745e-05,
      "loss": 1.0601,
      "step": 307
    },
    {
      "epoch": 0.008111667105609692,
      "grad_norm": 3.2779343128204346,
      "learning_rate": 4.959968396102186e-05,
      "loss": 1.7609,
      "step": 308
    },
    {
      "epoch": 0.008138003687121411,
      "grad_norm": 3.8762402534484863,
      "learning_rate": 4.959836713194628e-05,
      "loss": 0.6232,
      "step": 309
    },
    {
      "epoch": 0.008164340268633132,
      "grad_norm": 2.778155565261841,
      "learning_rate": 4.959705030287069e-05,
      "loss": 1.9539,
      "step": 310
    },
    {
      "epoch": 0.008190676850144851,
      "grad_norm": 4.9701056480407715,
      "learning_rate": 4.95957334737951e-05,
      "loss": 1.8396,
      "step": 311
    },
    {
      "epoch": 0.00821701343165657,
      "grad_norm": 4.125095844268799,
      "learning_rate": 4.959441664471952e-05,
      "loss": 0.4815,
      "step": 312
    },
    {
      "epoch": 0.008243350013168291,
      "grad_norm": 5.6718292236328125,
      "learning_rate": 4.9593099815643934e-05,
      "loss": 2.0331,
      "step": 313
    },
    {
      "epoch": 0.00826968659468001,
      "grad_norm": 6.395870208740234,
      "learning_rate": 4.959178298656835e-05,
      "loss": 1.3781,
      "step": 314
    },
    {
      "epoch": 0.008296023176191731,
      "grad_norm": 2.874675989151001,
      "learning_rate": 4.959046615749276e-05,
      "loss": 2.1396,
      "step": 315
    },
    {
      "epoch": 0.00832235975770345,
      "grad_norm": 2.8582258224487305,
      "learning_rate": 4.9589149328417174e-05,
      "loss": 1.6284,
      "step": 316
    },
    {
      "epoch": 0.008348696339215169,
      "grad_norm": 3.6091315746307373,
      "learning_rate": 4.958783249934158e-05,
      "loss": 1.5799,
      "step": 317
    },
    {
      "epoch": 0.00837503292072689,
      "grad_norm": 3.2684743404388428,
      "learning_rate": 4.9586515670266005e-05,
      "loss": 1.8535,
      "step": 318
    },
    {
      "epoch": 0.008401369502238609,
      "grad_norm": 2.686674118041992,
      "learning_rate": 4.9585198841190414e-05,
      "loss": 1.7055,
      "step": 319
    },
    {
      "epoch": 0.00842770608375033,
      "grad_norm": 2.14872670173645,
      "learning_rate": 4.958388201211483e-05,
      "loss": 1.6215,
      "step": 320
    },
    {
      "epoch": 0.008454042665262049,
      "grad_norm": 2.932607650756836,
      "learning_rate": 4.9582565183039245e-05,
      "loss": 1.7547,
      "step": 321
    },
    {
      "epoch": 0.00848037924677377,
      "grad_norm": 2.8945400714874268,
      "learning_rate": 4.9581248353963654e-05,
      "loss": 1.7845,
      "step": 322
    },
    {
      "epoch": 0.008506715828285489,
      "grad_norm": 3.505058526992798,
      "learning_rate": 4.957993152488808e-05,
      "loss": 2.6208,
      "step": 323
    },
    {
      "epoch": 0.008533052409797208,
      "grad_norm": 2.4234344959259033,
      "learning_rate": 4.9578614695812485e-05,
      "loss": 2.0702,
      "step": 324
    },
    {
      "epoch": 0.008559388991308928,
      "grad_norm": 2.21173095703125,
      "learning_rate": 4.95772978667369e-05,
      "loss": 1.2613,
      "step": 325
    },
    {
      "epoch": 0.008585725572820647,
      "grad_norm": 2.6429340839385986,
      "learning_rate": 4.957598103766131e-05,
      "loss": 1.7807,
      "step": 326
    },
    {
      "epoch": 0.008612062154332368,
      "grad_norm": 3.20620059967041,
      "learning_rate": 4.957466420858573e-05,
      "loss": 2.1573,
      "step": 327
    },
    {
      "epoch": 0.008638398735844087,
      "grad_norm": 2.7805612087249756,
      "learning_rate": 4.957334737951014e-05,
      "loss": 1.8478,
      "step": 328
    },
    {
      "epoch": 0.008664735317355806,
      "grad_norm": 5.594062328338623,
      "learning_rate": 4.957203055043456e-05,
      "loss": 2.3134,
      "step": 329
    },
    {
      "epoch": 0.008691071898867527,
      "grad_norm": 3.3740992546081543,
      "learning_rate": 4.957071372135897e-05,
      "loss": 1.9,
      "step": 330
    },
    {
      "epoch": 0.008717408480379246,
      "grad_norm": 4.023157596588135,
      "learning_rate": 4.956939689228338e-05,
      "loss": 2.2293,
      "step": 331
    },
    {
      "epoch": 0.008743745061890967,
      "grad_norm": 4.384700775146484,
      "learning_rate": 4.9568080063207803e-05,
      "loss": 2.7289,
      "step": 332
    },
    {
      "epoch": 0.008770081643402686,
      "grad_norm": 3.8390510082244873,
      "learning_rate": 4.956676323413221e-05,
      "loss": 1.9188,
      "step": 333
    },
    {
      "epoch": 0.008796418224914407,
      "grad_norm": 6.378127574920654,
      "learning_rate": 4.956544640505663e-05,
      "loss": 1.3166,
      "step": 334
    },
    {
      "epoch": 0.008822754806426126,
      "grad_norm": 5.20752477645874,
      "learning_rate": 4.956412957598104e-05,
      "loss": 1.0243,
      "step": 335
    },
    {
      "epoch": 0.008849091387937845,
      "grad_norm": 4.862656116485596,
      "learning_rate": 4.956281274690545e-05,
      "loss": 2.7114,
      "step": 336
    },
    {
      "epoch": 0.008875427969449566,
      "grad_norm": 4.407208442687988,
      "learning_rate": 4.956149591782987e-05,
      "loss": 2.0917,
      "step": 337
    },
    {
      "epoch": 0.008901764550961285,
      "grad_norm": 4.253674030303955,
      "learning_rate": 4.9560179088754283e-05,
      "loss": 1.4495,
      "step": 338
    },
    {
      "epoch": 0.008928101132473006,
      "grad_norm": 4.106147766113281,
      "learning_rate": 4.955886225967869e-05,
      "loss": 1.1252,
      "step": 339
    },
    {
      "epoch": 0.008954437713984725,
      "grad_norm": 3.6390883922576904,
      "learning_rate": 4.955754543060311e-05,
      "loss": 2.0238,
      "step": 340
    },
    {
      "epoch": 0.008980774295496445,
      "grad_norm": 2.9469077587127686,
      "learning_rate": 4.9556228601527524e-05,
      "loss": 2.4644,
      "step": 341
    },
    {
      "epoch": 0.009007110877008164,
      "grad_norm": 2.9715445041656494,
      "learning_rate": 4.955491177245194e-05,
      "loss": 1.2653,
      "step": 342
    },
    {
      "epoch": 0.009033447458519884,
      "grad_norm": 2.7931387424468994,
      "learning_rate": 4.9553594943376355e-05,
      "loss": 2.2082,
      "step": 343
    },
    {
      "epoch": 0.009059784040031604,
      "grad_norm": 4.04490327835083,
      "learning_rate": 4.9552278114300764e-05,
      "loss": 1.8727,
      "step": 344
    },
    {
      "epoch": 0.009086120621543323,
      "grad_norm": 2.391226053237915,
      "learning_rate": 4.955096128522518e-05,
      "loss": 1.5419,
      "step": 345
    },
    {
      "epoch": 0.009112457203055044,
      "grad_norm": 4.560610771179199,
      "learning_rate": 4.9549644456149595e-05,
      "loss": 1.168,
      "step": 346
    },
    {
      "epoch": 0.009138793784566763,
      "grad_norm": 2.3792755603790283,
      "learning_rate": 4.954832762707401e-05,
      "loss": 1.6401,
      "step": 347
    },
    {
      "epoch": 0.009165130366078482,
      "grad_norm": 2.6933135986328125,
      "learning_rate": 4.954701079799842e-05,
      "loss": 1.8095,
      "step": 348
    },
    {
      "epoch": 0.009191466947590203,
      "grad_norm": 2.3354787826538086,
      "learning_rate": 4.9545693968922835e-05,
      "loss": 2.004,
      "step": 349
    },
    {
      "epoch": 0.009217803529101922,
      "grad_norm": 4.505835056304932,
      "learning_rate": 4.954437713984725e-05,
      "loss": 1.4959,
      "step": 350
    },
    {
      "epoch": 0.009244140110613643,
      "grad_norm": 9.2621431350708,
      "learning_rate": 4.9543060310771666e-05,
      "loss": 1.2403,
      "step": 351
    },
    {
      "epoch": 0.009270476692125362,
      "grad_norm": 2.8335564136505127,
      "learning_rate": 4.954174348169608e-05,
      "loss": 1.5436,
      "step": 352
    },
    {
      "epoch": 0.009296813273637083,
      "grad_norm": 2.834834337234497,
      "learning_rate": 4.954042665262049e-05,
      "loss": 2.4434,
      "step": 353
    },
    {
      "epoch": 0.009323149855148802,
      "grad_norm": 2.5483241081237793,
      "learning_rate": 4.9539109823544906e-05,
      "loss": 1.8674,
      "step": 354
    },
    {
      "epoch": 0.00934948643666052,
      "grad_norm": 4.1416916847229,
      "learning_rate": 4.9537792994469315e-05,
      "loss": 1.6774,
      "step": 355
    },
    {
      "epoch": 0.009375823018172242,
      "grad_norm": 3.8100359439849854,
      "learning_rate": 4.953647616539374e-05,
      "loss": 1.3106,
      "step": 356
    },
    {
      "epoch": 0.00940215959968396,
      "grad_norm": 3.3069844245910645,
      "learning_rate": 4.9535159336318146e-05,
      "loss": 0.7217,
      "step": 357
    },
    {
      "epoch": 0.009428496181195681,
      "grad_norm": 3.8136091232299805,
      "learning_rate": 4.953384250724256e-05,
      "loss": 0.8796,
      "step": 358
    },
    {
      "epoch": 0.0094548327627074,
      "grad_norm": 3.232954263687134,
      "learning_rate": 4.953252567816698e-05,
      "loss": 2.0419,
      "step": 359
    },
    {
      "epoch": 0.00948116934421912,
      "grad_norm": 4.396567344665527,
      "learning_rate": 4.953120884909139e-05,
      "loss": 1.7555,
      "step": 360
    },
    {
      "epoch": 0.00950750592573084,
      "grad_norm": 3.195117712020874,
      "learning_rate": 4.952989202001581e-05,
      "loss": 2.1575,
      "step": 361
    },
    {
      "epoch": 0.00953384250724256,
      "grad_norm": 4.96207857131958,
      "learning_rate": 4.952857519094022e-05,
      "loss": 1.5802,
      "step": 362
    },
    {
      "epoch": 0.00956017908875428,
      "grad_norm": 3.737401008605957,
      "learning_rate": 4.952725836186463e-05,
      "loss": 1.8285,
      "step": 363
    },
    {
      "epoch": 0.009586515670266,
      "grad_norm": 2.9919257164001465,
      "learning_rate": 4.952594153278904e-05,
      "loss": 2.0902,
      "step": 364
    },
    {
      "epoch": 0.00961285225177772,
      "grad_norm": 5.612448692321777,
      "learning_rate": 4.9524624703713464e-05,
      "loss": 2.0094,
      "step": 365
    },
    {
      "epoch": 0.009639188833289439,
      "grad_norm": 2.9793035984039307,
      "learning_rate": 4.952330787463787e-05,
      "loss": 1.1435,
      "step": 366
    },
    {
      "epoch": 0.009665525414801158,
      "grad_norm": 7.405142307281494,
      "learning_rate": 4.952199104556229e-05,
      "loss": 0.7576,
      "step": 367
    },
    {
      "epoch": 0.009691861996312879,
      "grad_norm": 5.179115295410156,
      "learning_rate": 4.9520674216486704e-05,
      "loss": 1.5322,
      "step": 368
    },
    {
      "epoch": 0.009718198577824598,
      "grad_norm": 3.821916341781616,
      "learning_rate": 4.951935738741111e-05,
      "loss": 1.9987,
      "step": 369
    },
    {
      "epoch": 0.009744535159336319,
      "grad_norm": 3.794161796569824,
      "learning_rate": 4.9518040558335535e-05,
      "loss": 2.6062,
      "step": 370
    },
    {
      "epoch": 0.009770871740848038,
      "grad_norm": 4.306283473968506,
      "learning_rate": 4.9516723729259944e-05,
      "loss": 1.7716,
      "step": 371
    },
    {
      "epoch": 0.009797208322359757,
      "grad_norm": 3.607884645462036,
      "learning_rate": 4.951540690018436e-05,
      "loss": 0.8448,
      "step": 372
    },
    {
      "epoch": 0.009823544903871478,
      "grad_norm": 3.094651222229004,
      "learning_rate": 4.951409007110877e-05,
      "loss": 2.1413,
      "step": 373
    },
    {
      "epoch": 0.009849881485383197,
      "grad_norm": 6.7893829345703125,
      "learning_rate": 4.951277324203319e-05,
      "loss": 1.5486,
      "step": 374
    },
    {
      "epoch": 0.009876218066894918,
      "grad_norm": 3.5733284950256348,
      "learning_rate": 4.95114564129576e-05,
      "loss": 2.0736,
      "step": 375
    },
    {
      "epoch": 0.009902554648406637,
      "grad_norm": 3.4923906326293945,
      "learning_rate": 4.9510139583882015e-05,
      "loss": 1.6066,
      "step": 376
    },
    {
      "epoch": 0.009928891229918357,
      "grad_norm": 2.3045380115509033,
      "learning_rate": 4.950882275480643e-05,
      "loss": 1.8924,
      "step": 377
    },
    {
      "epoch": 0.009955227811430076,
      "grad_norm": 2.6986491680145264,
      "learning_rate": 4.950750592573084e-05,
      "loss": 2.1344,
      "step": 378
    },
    {
      "epoch": 0.009981564392941795,
      "grad_norm": 6.939801216125488,
      "learning_rate": 4.950618909665526e-05,
      "loss": 1.8125,
      "step": 379
    },
    {
      "epoch": 0.010007900974453516,
      "grad_norm": 2.977921485900879,
      "learning_rate": 4.950487226757967e-05,
      "loss": 2.3517,
      "step": 380
    },
    {
      "epoch": 0.010034237555965235,
      "grad_norm": 5.064753532409668,
      "learning_rate": 4.9503555438504086e-05,
      "loss": 1.9185,
      "step": 381
    },
    {
      "epoch": 0.010060574137476956,
      "grad_norm": 3.0815107822418213,
      "learning_rate": 4.9502238609428495e-05,
      "loss": 1.2712,
      "step": 382
    },
    {
      "epoch": 0.010086910718988675,
      "grad_norm": 3.1293399333953857,
      "learning_rate": 4.950092178035291e-05,
      "loss": 1.7528,
      "step": 383
    },
    {
      "epoch": 0.010113247300500394,
      "grad_norm": 2.841268539428711,
      "learning_rate": 4.9499604951277326e-05,
      "loss": 2.5219,
      "step": 384
    },
    {
      "epoch": 0.010139583882012115,
      "grad_norm": 2.7860684394836426,
      "learning_rate": 4.949828812220174e-05,
      "loss": 1.7577,
      "step": 385
    },
    {
      "epoch": 0.010165920463523834,
      "grad_norm": 5.101641654968262,
      "learning_rate": 4.949697129312615e-05,
      "loss": 1.5768,
      "step": 386
    },
    {
      "epoch": 0.010192257045035555,
      "grad_norm": 2.6400580406188965,
      "learning_rate": 4.9495654464050566e-05,
      "loss": 1.6106,
      "step": 387
    },
    {
      "epoch": 0.010218593626547274,
      "grad_norm": 3.29453182220459,
      "learning_rate": 4.949433763497498e-05,
      "loss": 2.3339,
      "step": 388
    },
    {
      "epoch": 0.010244930208058995,
      "grad_norm": 3.2891499996185303,
      "learning_rate": 4.94930208058994e-05,
      "loss": 2.093,
      "step": 389
    },
    {
      "epoch": 0.010271266789570714,
      "grad_norm": 3.197697162628174,
      "learning_rate": 4.949170397682381e-05,
      "loss": 1.8503,
      "step": 390
    },
    {
      "epoch": 0.010297603371082433,
      "grad_norm": 3.106626510620117,
      "learning_rate": 4.949038714774822e-05,
      "loss": 1.5676,
      "step": 391
    },
    {
      "epoch": 0.010323939952594154,
      "grad_norm": 3.074042320251465,
      "learning_rate": 4.948907031867264e-05,
      "loss": 2.0652,
      "step": 392
    },
    {
      "epoch": 0.010350276534105873,
      "grad_norm": 5.854307174682617,
      "learning_rate": 4.948775348959705e-05,
      "loss": 1.3206,
      "step": 393
    },
    {
      "epoch": 0.010376613115617593,
      "grad_norm": 4.576664447784424,
      "learning_rate": 4.948643666052147e-05,
      "loss": 1.5156,
      "step": 394
    },
    {
      "epoch": 0.010402949697129312,
      "grad_norm": 2.5585691928863525,
      "learning_rate": 4.948511983144588e-05,
      "loss": 1.7142,
      "step": 395
    },
    {
      "epoch": 0.010429286278641032,
      "grad_norm": 7.966944217681885,
      "learning_rate": 4.948380300237029e-05,
      "loss": 1.8537,
      "step": 396
    },
    {
      "epoch": 0.010455622860152752,
      "grad_norm": 2.7815499305725098,
      "learning_rate": 4.948248617329471e-05,
      "loss": 2.3155,
      "step": 397
    },
    {
      "epoch": 0.010481959441664471,
      "grad_norm": 2.548046827316284,
      "learning_rate": 4.9481169344219124e-05,
      "loss": 1.4609,
      "step": 398
    },
    {
      "epoch": 0.010508296023176192,
      "grad_norm": 3.1576316356658936,
      "learning_rate": 4.947985251514354e-05,
      "loss": 2.3327,
      "step": 399
    },
    {
      "epoch": 0.010534632604687911,
      "grad_norm": 4.465269088745117,
      "learning_rate": 4.947853568606795e-05,
      "loss": 1.3708,
      "step": 400
    },
    {
      "epoch": 0.010560969186199632,
      "grad_norm": 3.72763729095459,
      "learning_rate": 4.9477218856992364e-05,
      "loss": 2.9096,
      "step": 401
    },
    {
      "epoch": 0.010587305767711351,
      "grad_norm": 2.7007408142089844,
      "learning_rate": 4.947590202791677e-05,
      "loss": 1.7649,
      "step": 402
    },
    {
      "epoch": 0.01061364234922307,
      "grad_norm": 3.6093926429748535,
      "learning_rate": 4.9474585198841196e-05,
      "loss": 2.3405,
      "step": 403
    },
    {
      "epoch": 0.010639978930734791,
      "grad_norm": 3.598163366317749,
      "learning_rate": 4.9473268369765605e-05,
      "loss": 2.168,
      "step": 404
    },
    {
      "epoch": 0.01066631551224651,
      "grad_norm": 6.245233058929443,
      "learning_rate": 4.947195154069002e-05,
      "loss": 2.1714,
      "step": 405
    },
    {
      "epoch": 0.01069265209375823,
      "grad_norm": 3.7399916648864746,
      "learning_rate": 4.9470634711614436e-05,
      "loss": 2.121,
      "step": 406
    },
    {
      "epoch": 0.01071898867526995,
      "grad_norm": 3.036510467529297,
      "learning_rate": 4.946931788253885e-05,
      "loss": 1.5228,
      "step": 407
    },
    {
      "epoch": 0.01074532525678167,
      "grad_norm": 4.524996757507324,
      "learning_rate": 4.946800105346327e-05,
      "loss": 2.098,
      "step": 408
    },
    {
      "epoch": 0.01077166183829339,
      "grad_norm": 3.4973762035369873,
      "learning_rate": 4.9466684224387676e-05,
      "loss": 2.0426,
      "step": 409
    },
    {
      "epoch": 0.010797998419805109,
      "grad_norm": 3.5525729656219482,
      "learning_rate": 4.946536739531209e-05,
      "loss": 1.6071,
      "step": 410
    },
    {
      "epoch": 0.01082433500131683,
      "grad_norm": 8.208961486816406,
      "learning_rate": 4.94640505662365e-05,
      "loss": 1.7934,
      "step": 411
    },
    {
      "epoch": 0.010850671582828549,
      "grad_norm": 3.2176403999328613,
      "learning_rate": 4.946273373716092e-05,
      "loss": 1.9724,
      "step": 412
    },
    {
      "epoch": 0.01087700816434027,
      "grad_norm": 3.4763741493225098,
      "learning_rate": 4.946141690808533e-05,
      "loss": 0.4135,
      "step": 413
    },
    {
      "epoch": 0.010903344745851988,
      "grad_norm": 4.857624053955078,
      "learning_rate": 4.946010007900975e-05,
      "loss": 1.4289,
      "step": 414
    },
    {
      "epoch": 0.010929681327363707,
      "grad_norm": 3.0980494022369385,
      "learning_rate": 4.945878324993416e-05,
      "loss": 1.68,
      "step": 415
    },
    {
      "epoch": 0.010956017908875428,
      "grad_norm": 2.611757755279541,
      "learning_rate": 4.945746642085857e-05,
      "loss": 1.987,
      "step": 416
    },
    {
      "epoch": 0.010982354490387147,
      "grad_norm": 3.18953537940979,
      "learning_rate": 4.9456149591782994e-05,
      "loss": 1.9162,
      "step": 417
    },
    {
      "epoch": 0.011008691071898868,
      "grad_norm": 1.9519106149673462,
      "learning_rate": 4.94548327627074e-05,
      "loss": 1.7891,
      "step": 418
    },
    {
      "epoch": 0.011035027653410587,
      "grad_norm": 2.3198137283325195,
      "learning_rate": 4.945351593363182e-05,
      "loss": 1.6574,
      "step": 419
    },
    {
      "epoch": 0.011061364234922308,
      "grad_norm": 2.448354482650757,
      "learning_rate": 4.945219910455623e-05,
      "loss": 1.4682,
      "step": 420
    },
    {
      "epoch": 0.011087700816434027,
      "grad_norm": 3.6776375770568848,
      "learning_rate": 4.945088227548064e-05,
      "loss": 1.6907,
      "step": 421
    },
    {
      "epoch": 0.011114037397945746,
      "grad_norm": 5.139711380004883,
      "learning_rate": 4.944956544640506e-05,
      "loss": 2.0815,
      "step": 422
    },
    {
      "epoch": 0.011140373979457467,
      "grad_norm": 3.2158284187316895,
      "learning_rate": 4.9448248617329474e-05,
      "loss": 2.2924,
      "step": 423
    },
    {
      "epoch": 0.011166710560969186,
      "grad_norm": 3.731276035308838,
      "learning_rate": 4.944693178825389e-05,
      "loss": 1.5524,
      "step": 424
    },
    {
      "epoch": 0.011193047142480907,
      "grad_norm": 2.7019474506378174,
      "learning_rate": 4.94456149591783e-05,
      "loss": 2.0697,
      "step": 425
    },
    {
      "epoch": 0.011219383723992626,
      "grad_norm": 4.75086784362793,
      "learning_rate": 4.944429813010272e-05,
      "loss": 1.8361,
      "step": 426
    },
    {
      "epoch": 0.011245720305504345,
      "grad_norm": 2.408961057662964,
      "learning_rate": 4.944298130102713e-05,
      "loss": 1.6186,
      "step": 427
    },
    {
      "epoch": 0.011272056887016066,
      "grad_norm": 2.6383657455444336,
      "learning_rate": 4.9441664471951545e-05,
      "loss": 2.0564,
      "step": 428
    },
    {
      "epoch": 0.011298393468527785,
      "grad_norm": 2.8923819065093994,
      "learning_rate": 4.9440347642875954e-05,
      "loss": 1.1521,
      "step": 429
    },
    {
      "epoch": 0.011324730050039505,
      "grad_norm": 4.4877424240112305,
      "learning_rate": 4.943903081380037e-05,
      "loss": 1.5687,
      "step": 430
    },
    {
      "epoch": 0.011351066631551224,
      "grad_norm": 3.236896514892578,
      "learning_rate": 4.9437713984724785e-05,
      "loss": 1.7857,
      "step": 431
    },
    {
      "epoch": 0.011377403213062945,
      "grad_norm": 2.7711963653564453,
      "learning_rate": 4.94363971556492e-05,
      "loss": 2.5365,
      "step": 432
    },
    {
      "epoch": 0.011403739794574664,
      "grad_norm": 2.926581859588623,
      "learning_rate": 4.9435080326573616e-05,
      "loss": 2.3504,
      "step": 433
    },
    {
      "epoch": 0.011430076376086383,
      "grad_norm": 2.491097927093506,
      "learning_rate": 4.9433763497498025e-05,
      "loss": 1.8617,
      "step": 434
    },
    {
      "epoch": 0.011456412957598104,
      "grad_norm": 2.979182243347168,
      "learning_rate": 4.943244666842244e-05,
      "loss": 1.6826,
      "step": 435
    },
    {
      "epoch": 0.011482749539109823,
      "grad_norm": 8.16455364227295,
      "learning_rate": 4.9431129839346856e-05,
      "loss": 1.5242,
      "step": 436
    },
    {
      "epoch": 0.011509086120621544,
      "grad_norm": 3.306053400039673,
      "learning_rate": 4.942981301027127e-05,
      "loss": 2.1168,
      "step": 437
    },
    {
      "epoch": 0.011535422702133263,
      "grad_norm": 2.775508403778076,
      "learning_rate": 4.942849618119568e-05,
      "loss": 2.3886,
      "step": 438
    },
    {
      "epoch": 0.011561759283644982,
      "grad_norm": 3.597965717315674,
      "learning_rate": 4.9427179352120096e-05,
      "loss": 1.5943,
      "step": 439
    },
    {
      "epoch": 0.011588095865156703,
      "grad_norm": 4.4641289710998535,
      "learning_rate": 4.942586252304451e-05,
      "loss": 1.3134,
      "step": 440
    },
    {
      "epoch": 0.011614432446668422,
      "grad_norm": 2.697011709213257,
      "learning_rate": 4.942454569396893e-05,
      "loss": 1.9472,
      "step": 441
    },
    {
      "epoch": 0.011640769028180143,
      "grad_norm": 3.3379714488983154,
      "learning_rate": 4.9423228864893336e-05,
      "loss": 2.0254,
      "step": 442
    },
    {
      "epoch": 0.011667105609691862,
      "grad_norm": 4.314624786376953,
      "learning_rate": 4.942191203581775e-05,
      "loss": 1.5491,
      "step": 443
    },
    {
      "epoch": 0.011693442191203583,
      "grad_norm": 4.256104469299316,
      "learning_rate": 4.942059520674217e-05,
      "loss": 1.7194,
      "step": 444
    },
    {
      "epoch": 0.011719778772715302,
      "grad_norm": 4.3630852699279785,
      "learning_rate": 4.941927837766658e-05,
      "loss": 2.1208,
      "step": 445
    },
    {
      "epoch": 0.01174611535422702,
      "grad_norm": 3.4363515377044678,
      "learning_rate": 4.9417961548591e-05,
      "loss": 1.9275,
      "step": 446
    },
    {
      "epoch": 0.011772451935738741,
      "grad_norm": 2.999098062515259,
      "learning_rate": 4.941664471951541e-05,
      "loss": 1.8565,
      "step": 447
    },
    {
      "epoch": 0.01179878851725046,
      "grad_norm": 4.614634037017822,
      "learning_rate": 4.941532789043982e-05,
      "loss": 1.7001,
      "step": 448
    },
    {
      "epoch": 0.011825125098762181,
      "grad_norm": 3.698587417602539,
      "learning_rate": 4.941401106136423e-05,
      "loss": 0.8016,
      "step": 449
    },
    {
      "epoch": 0.0118514616802739,
      "grad_norm": 6.296628475189209,
      "learning_rate": 4.9412694232288654e-05,
      "loss": 1.8436,
      "step": 450
    },
    {
      "epoch": 0.01187779826178562,
      "grad_norm": 2.57922101020813,
      "learning_rate": 4.941137740321306e-05,
      "loss": 1.5764,
      "step": 451
    },
    {
      "epoch": 0.01190413484329734,
      "grad_norm": 2.7335686683654785,
      "learning_rate": 4.941006057413748e-05,
      "loss": 1.8235,
      "step": 452
    },
    {
      "epoch": 0.01193047142480906,
      "grad_norm": 2.7058944702148438,
      "learning_rate": 4.9408743745061894e-05,
      "loss": 2.3749,
      "step": 453
    },
    {
      "epoch": 0.01195680800632078,
      "grad_norm": 2.696030378341675,
      "learning_rate": 4.94074269159863e-05,
      "loss": 1.5161,
      "step": 454
    },
    {
      "epoch": 0.011983144587832499,
      "grad_norm": 3.3722379207611084,
      "learning_rate": 4.9406110086910725e-05,
      "loss": 1.9797,
      "step": 455
    },
    {
      "epoch": 0.01200948116934422,
      "grad_norm": 3.0010156631469727,
      "learning_rate": 4.9404793257835134e-05,
      "loss": 1.4081,
      "step": 456
    },
    {
      "epoch": 0.012035817750855939,
      "grad_norm": 4.13242769241333,
      "learning_rate": 4.940347642875955e-05,
      "loss": 2.4764,
      "step": 457
    },
    {
      "epoch": 0.012062154332367658,
      "grad_norm": 4.92889928817749,
      "learning_rate": 4.940215959968396e-05,
      "loss": 1.9282,
      "step": 458
    },
    {
      "epoch": 0.012088490913879379,
      "grad_norm": 5.652385711669922,
      "learning_rate": 4.940084277060838e-05,
      "loss": 1.3044,
      "step": 459
    },
    {
      "epoch": 0.012114827495391098,
      "grad_norm": 3.8656020164489746,
      "learning_rate": 4.939952594153279e-05,
      "loss": 1.5397,
      "step": 460
    },
    {
      "epoch": 0.012141164076902819,
      "grad_norm": 2.3113582134246826,
      "learning_rate": 4.9398209112457205e-05,
      "loss": 1.9664,
      "step": 461
    },
    {
      "epoch": 0.012167500658414538,
      "grad_norm": 2.551178216934204,
      "learning_rate": 4.939689228338162e-05,
      "loss": 0.413,
      "step": 462
    },
    {
      "epoch": 0.012193837239926257,
      "grad_norm": 2.828598737716675,
      "learning_rate": 4.939557545430603e-05,
      "loss": 1.9058,
      "step": 463
    },
    {
      "epoch": 0.012220173821437978,
      "grad_norm": 5.021803379058838,
      "learning_rate": 4.939425862523045e-05,
      "loss": 0.8865,
      "step": 464
    },
    {
      "epoch": 0.012246510402949697,
      "grad_norm": 4.49089241027832,
      "learning_rate": 4.939294179615486e-05,
      "loss": 1.7977,
      "step": 465
    },
    {
      "epoch": 0.012272846984461417,
      "grad_norm": 2.908942937850952,
      "learning_rate": 4.939162496707928e-05,
      "loss": 1.941,
      "step": 466
    },
    {
      "epoch": 0.012299183565973136,
      "grad_norm": 3.2470862865448,
      "learning_rate": 4.9390308138003686e-05,
      "loss": 2.221,
      "step": 467
    },
    {
      "epoch": 0.012325520147484857,
      "grad_norm": 3.354400396347046,
      "learning_rate": 4.93889913089281e-05,
      "loss": 1.1864,
      "step": 468
    },
    {
      "epoch": 0.012351856728996576,
      "grad_norm": 2.356088638305664,
      "learning_rate": 4.938767447985252e-05,
      "loss": 1.8456,
      "step": 469
    },
    {
      "epoch": 0.012378193310508295,
      "grad_norm": 2.8001549243927,
      "learning_rate": 4.938635765077693e-05,
      "loss": 1.9366,
      "step": 470
    },
    {
      "epoch": 0.012404529892020016,
      "grad_norm": 5.683740615844727,
      "learning_rate": 4.938504082170135e-05,
      "loss": 0.8475,
      "step": 471
    },
    {
      "epoch": 0.012430866473531735,
      "grad_norm": 3.8350062370300293,
      "learning_rate": 4.938372399262576e-05,
      "loss": 2.0938,
      "step": 472
    },
    {
      "epoch": 0.012457203055043456,
      "grad_norm": 4.309392929077148,
      "learning_rate": 4.938240716355018e-05,
      "loss": 1.1875,
      "step": 473
    },
    {
      "epoch": 0.012483539636555175,
      "grad_norm": 2.567596912384033,
      "learning_rate": 4.938109033447459e-05,
      "loss": 1.8618,
      "step": 474
    },
    {
      "epoch": 0.012509876218066896,
      "grad_norm": 4.895605564117432,
      "learning_rate": 4.9379773505399004e-05,
      "loss": 1.7448,
      "step": 475
    },
    {
      "epoch": 0.012536212799578615,
      "grad_norm": 2.557440757751465,
      "learning_rate": 4.937845667632341e-05,
      "loss": 1.2474,
      "step": 476
    },
    {
      "epoch": 0.012562549381090334,
      "grad_norm": 2.6986734867095947,
      "learning_rate": 4.937713984724783e-05,
      "loss": 1.8171,
      "step": 477
    },
    {
      "epoch": 0.012588885962602055,
      "grad_norm": 3.2066280841827393,
      "learning_rate": 4.9375823018172244e-05,
      "loss": 1.8473,
      "step": 478
    },
    {
      "epoch": 0.012615222544113774,
      "grad_norm": 3.2558040618896484,
      "learning_rate": 4.937450618909666e-05,
      "loss": 1.7312,
      "step": 479
    },
    {
      "epoch": 0.012641559125625495,
      "grad_norm": 2.3930885791778564,
      "learning_rate": 4.9373189360021075e-05,
      "loss": 1.7677,
      "step": 480
    },
    {
      "epoch": 0.012667895707137214,
      "grad_norm": 3.230320453643799,
      "learning_rate": 4.9371872530945484e-05,
      "loss": 1.7896,
      "step": 481
    },
    {
      "epoch": 0.012694232288648933,
      "grad_norm": 4.14604377746582,
      "learning_rate": 4.93705557018699e-05,
      "loss": 1.4854,
      "step": 482
    },
    {
      "epoch": 0.012720568870160653,
      "grad_norm": 3.058246612548828,
      "learning_rate": 4.9369238872794315e-05,
      "loss": 1.3328,
      "step": 483
    },
    {
      "epoch": 0.012746905451672372,
      "grad_norm": 4.122186183929443,
      "learning_rate": 4.936792204371873e-05,
      "loss": 1.7708,
      "step": 484
    },
    {
      "epoch": 0.012773242033184093,
      "grad_norm": 3.2511589527130127,
      "learning_rate": 4.936660521464314e-05,
      "loss": 2.2072,
      "step": 485
    },
    {
      "epoch": 0.012799578614695812,
      "grad_norm": 2.4089932441711426,
      "learning_rate": 4.9365288385567555e-05,
      "loss": 2.111,
      "step": 486
    },
    {
      "epoch": 0.012825915196207533,
      "grad_norm": 4.144657135009766,
      "learning_rate": 4.936397155649197e-05,
      "loss": 2.2266,
      "step": 487
    },
    {
      "epoch": 0.012852251777719252,
      "grad_norm": 3.3680617809295654,
      "learning_rate": 4.9362654727416386e-05,
      "loss": 2.0625,
      "step": 488
    },
    {
      "epoch": 0.012878588359230971,
      "grad_norm": 5.534539699554443,
      "learning_rate": 4.9361337898340795e-05,
      "loss": 1.2348,
      "step": 489
    },
    {
      "epoch": 0.012904924940742692,
      "grad_norm": 11.291756629943848,
      "learning_rate": 4.936002106926521e-05,
      "loss": 2.3942,
      "step": 490
    },
    {
      "epoch": 0.012931261522254411,
      "grad_norm": 10.59220027923584,
      "learning_rate": 4.9358704240189626e-05,
      "loss": 1.3182,
      "step": 491
    },
    {
      "epoch": 0.012957598103766132,
      "grad_norm": 4.226751327514648,
      "learning_rate": 4.935738741111404e-05,
      "loss": 2.6083,
      "step": 492
    },
    {
      "epoch": 0.012983934685277851,
      "grad_norm": 2.6384074687957764,
      "learning_rate": 4.935607058203846e-05,
      "loss": 0.3831,
      "step": 493
    },
    {
      "epoch": 0.01301027126678957,
      "grad_norm": 3.1196553707122803,
      "learning_rate": 4.9354753752962866e-05,
      "loss": 1.7627,
      "step": 494
    },
    {
      "epoch": 0.01303660784830129,
      "grad_norm": 8.276817321777344,
      "learning_rate": 4.935343692388728e-05,
      "loss": 1.3917,
      "step": 495
    },
    {
      "epoch": 0.01306294442981301,
      "grad_norm": 3.178133726119995,
      "learning_rate": 4.935212009481169e-05,
      "loss": 1.6477,
      "step": 496
    },
    {
      "epoch": 0.01308928101132473,
      "grad_norm": 3.6952528953552246,
      "learning_rate": 4.935080326573611e-05,
      "loss": 0.7113,
      "step": 497
    },
    {
      "epoch": 0.01311561759283645,
      "grad_norm": 3.7713797092437744,
      "learning_rate": 4.934948643666052e-05,
      "loss": 2.0973,
      "step": 498
    },
    {
      "epoch": 0.01314195417434817,
      "grad_norm": 3.135864496231079,
      "learning_rate": 4.934816960758494e-05,
      "loss": 1.3988,
      "step": 499
    },
    {
      "epoch": 0.01316829075585989,
      "grad_norm": 3.300065040588379,
      "learning_rate": 4.934685277850935e-05,
      "loss": 1.7863,
      "step": 500
    },
    {
      "epoch": 0.013194627337371609,
      "grad_norm": 2.766282796859741,
      "learning_rate": 4.934553594943376e-05,
      "loss": 1.2625,
      "step": 501
    },
    {
      "epoch": 0.01322096391888333,
      "grad_norm": 7.230177879333496,
      "learning_rate": 4.9344219120358184e-05,
      "loss": 1.3097,
      "step": 502
    },
    {
      "epoch": 0.013247300500395048,
      "grad_norm": 3.0049235820770264,
      "learning_rate": 4.934290229128259e-05,
      "loss": 2.1888,
      "step": 503
    },
    {
      "epoch": 0.01327363708190677,
      "grad_norm": 2.8356151580810547,
      "learning_rate": 4.934158546220701e-05,
      "loss": 1.573,
      "step": 504
    },
    {
      "epoch": 0.013299973663418488,
      "grad_norm": 3.1539864540100098,
      "learning_rate": 4.934026863313142e-05,
      "loss": 0.6297,
      "step": 505
    },
    {
      "epoch": 0.013326310244930207,
      "grad_norm": 6.145493984222412,
      "learning_rate": 4.933895180405584e-05,
      "loss": 1.6996,
      "step": 506
    },
    {
      "epoch": 0.013352646826441928,
      "grad_norm": 2.7827095985412598,
      "learning_rate": 4.933763497498025e-05,
      "loss": 2.1537,
      "step": 507
    },
    {
      "epoch": 0.013378983407953647,
      "grad_norm": 5.041752815246582,
      "learning_rate": 4.9336318145904664e-05,
      "loss": 1.9504,
      "step": 508
    },
    {
      "epoch": 0.013405319989465368,
      "grad_norm": 2.551994562149048,
      "learning_rate": 4.933500131682908e-05,
      "loss": 1.9604,
      "step": 509
    },
    {
      "epoch": 0.013431656570977087,
      "grad_norm": 7.829330921173096,
      "learning_rate": 4.933368448775349e-05,
      "loss": 1.1279,
      "step": 510
    },
    {
      "epoch": 0.013457993152488808,
      "grad_norm": 5.9089789390563965,
      "learning_rate": 4.933236765867791e-05,
      "loss": 1.7507,
      "step": 511
    },
    {
      "epoch": 0.013484329734000527,
      "grad_norm": 2.3757100105285645,
      "learning_rate": 4.933105082960232e-05,
      "loss": 2.0618,
      "step": 512
    },
    {
      "epoch": 0.013510666315512246,
      "grad_norm": 5.670836925506592,
      "learning_rate": 4.9329734000526735e-05,
      "loss": 0.9599,
      "step": 513
    },
    {
      "epoch": 0.013537002897023967,
      "grad_norm": 5.219981670379639,
      "learning_rate": 4.9328417171451144e-05,
      "loss": 1.4321,
      "step": 514
    },
    {
      "epoch": 0.013563339478535686,
      "grad_norm": 2.96686053276062,
      "learning_rate": 4.932710034237556e-05,
      "loss": 1.7634,
      "step": 515
    },
    {
      "epoch": 0.013589676060047406,
      "grad_norm": 3.4565460681915283,
      "learning_rate": 4.9325783513299975e-05,
      "loss": 1.6794,
      "step": 516
    },
    {
      "epoch": 0.013616012641559126,
      "grad_norm": 3.0890915393829346,
      "learning_rate": 4.932446668422439e-05,
      "loss": 1.9017,
      "step": 517
    },
    {
      "epoch": 0.013642349223070845,
      "grad_norm": 3.332601308822632,
      "learning_rate": 4.9323149855148806e-05,
      "loss": 1.7231,
      "step": 518
    },
    {
      "epoch": 0.013668685804582565,
      "grad_norm": 4.089598655700684,
      "learning_rate": 4.9321833026073215e-05,
      "loss": 1.1871,
      "step": 519
    },
    {
      "epoch": 0.013695022386094284,
      "grad_norm": 2.75116229057312,
      "learning_rate": 4.932051619699764e-05,
      "loss": 1.5234,
      "step": 520
    },
    {
      "epoch": 0.013721358967606005,
      "grad_norm": 3.515044927597046,
      "learning_rate": 4.9319199367922046e-05,
      "loss": 1.8234,
      "step": 521
    },
    {
      "epoch": 0.013747695549117724,
      "grad_norm": 5.515471935272217,
      "learning_rate": 4.931788253884646e-05,
      "loss": 1.3104,
      "step": 522
    },
    {
      "epoch": 0.013774032130629445,
      "grad_norm": 2.4399476051330566,
      "learning_rate": 4.931656570977087e-05,
      "loss": 2.2612,
      "step": 523
    },
    {
      "epoch": 0.013800368712141164,
      "grad_norm": 3.2772269248962402,
      "learning_rate": 4.9315248880695287e-05,
      "loss": 2.083,
      "step": 524
    },
    {
      "epoch": 0.013826705293652883,
      "grad_norm": 2.3090224266052246,
      "learning_rate": 4.93139320516197e-05,
      "loss": 2.3116,
      "step": 525
    },
    {
      "epoch": 0.013853041875164604,
      "grad_norm": 2.9195845127105713,
      "learning_rate": 4.931261522254412e-05,
      "loss": 0.931,
      "step": 526
    },
    {
      "epoch": 0.013879378456676323,
      "grad_norm": 3.011740207672119,
      "learning_rate": 4.931129839346853e-05,
      "loss": 2.0923,
      "step": 527
    },
    {
      "epoch": 0.013905715038188044,
      "grad_norm": 2.624511241912842,
      "learning_rate": 4.930998156439294e-05,
      "loss": 1.6522,
      "step": 528
    },
    {
      "epoch": 0.013932051619699763,
      "grad_norm": 6.0027756690979,
      "learning_rate": 4.930866473531736e-05,
      "loss": 1.2117,
      "step": 529
    },
    {
      "epoch": 0.013958388201211482,
      "grad_norm": 3.0832457542419434,
      "learning_rate": 4.930734790624177e-05,
      "loss": 2.8007,
      "step": 530
    },
    {
      "epoch": 0.013984724782723203,
      "grad_norm": 3.3799095153808594,
      "learning_rate": 4.930603107716619e-05,
      "loss": 2.4809,
      "step": 531
    },
    {
      "epoch": 0.014011061364234922,
      "grad_norm": 3.8424103260040283,
      "learning_rate": 4.93047142480906e-05,
      "loss": 1.6528,
      "step": 532
    },
    {
      "epoch": 0.014037397945746643,
      "grad_norm": 9.509858131408691,
      "learning_rate": 4.930339741901501e-05,
      "loss": 2.5631,
      "step": 533
    },
    {
      "epoch": 0.014063734527258362,
      "grad_norm": 4.948150157928467,
      "learning_rate": 4.930208058993943e-05,
      "loss": 1.0504,
      "step": 534
    },
    {
      "epoch": 0.014090071108770082,
      "grad_norm": 3.894061803817749,
      "learning_rate": 4.9300763760863845e-05,
      "loss": 1.9647,
      "step": 535
    },
    {
      "epoch": 0.014116407690281801,
      "grad_norm": 2.8031275272369385,
      "learning_rate": 4.929944693178826e-05,
      "loss": 1.6106,
      "step": 536
    },
    {
      "epoch": 0.01414274427179352,
      "grad_norm": 2.3836934566497803,
      "learning_rate": 4.929813010271267e-05,
      "loss": 2.0547,
      "step": 537
    },
    {
      "epoch": 0.014169080853305241,
      "grad_norm": 4.073704242706299,
      "learning_rate": 4.9296813273637085e-05,
      "loss": 0.9845,
      "step": 538
    },
    {
      "epoch": 0.01419541743481696,
      "grad_norm": 2.5212810039520264,
      "learning_rate": 4.92954964445615e-05,
      "loss": 1.879,
      "step": 539
    },
    {
      "epoch": 0.014221754016328681,
      "grad_norm": 2.55657696723938,
      "learning_rate": 4.9294179615485916e-05,
      "loss": 1.8492,
      "step": 540
    },
    {
      "epoch": 0.0142480905978404,
      "grad_norm": 2.444204092025757,
      "learning_rate": 4.9292862786410325e-05,
      "loss": 1.7488,
      "step": 541
    },
    {
      "epoch": 0.014274427179352121,
      "grad_norm": 2.6518898010253906,
      "learning_rate": 4.929154595733474e-05,
      "loss": 1.8359,
      "step": 542
    },
    {
      "epoch": 0.01430076376086384,
      "grad_norm": 7.798391342163086,
      "learning_rate": 4.929022912825915e-05,
      "loss": 1.2831,
      "step": 543
    },
    {
      "epoch": 0.014327100342375559,
      "grad_norm": 3.268341302871704,
      "learning_rate": 4.928891229918357e-05,
      "loss": 0.7458,
      "step": 544
    },
    {
      "epoch": 0.01435343692388728,
      "grad_norm": 2.8737120628356934,
      "learning_rate": 4.928759547010798e-05,
      "loss": 1.043,
      "step": 545
    },
    {
      "epoch": 0.014379773505398999,
      "grad_norm": 2.81797194480896,
      "learning_rate": 4.9286278641032396e-05,
      "loss": 2.3844,
      "step": 546
    },
    {
      "epoch": 0.01440611008691072,
      "grad_norm": 2.3775641918182373,
      "learning_rate": 4.928496181195681e-05,
      "loss": 2.0042,
      "step": 547
    },
    {
      "epoch": 0.014432446668422439,
      "grad_norm": 2.2898850440979004,
      "learning_rate": 4.928364498288122e-05,
      "loss": 1.5706,
      "step": 548
    },
    {
      "epoch": 0.014458783249934158,
      "grad_norm": 3.3940882682800293,
      "learning_rate": 4.928232815380564e-05,
      "loss": 1.8806,
      "step": 549
    },
    {
      "epoch": 0.014485119831445879,
      "grad_norm": 4.997721195220947,
      "learning_rate": 4.928101132473005e-05,
      "loss": 2.2222,
      "step": 550
    },
    {
      "epoch": 0.014511456412957598,
      "grad_norm": 2.8004939556121826,
      "learning_rate": 4.927969449565447e-05,
      "loss": 1.9449,
      "step": 551
    },
    {
      "epoch": 0.014537792994469318,
      "grad_norm": 3.135756731033325,
      "learning_rate": 4.9278377666578876e-05,
      "loss": 1.8732,
      "step": 552
    },
    {
      "epoch": 0.014564129575981037,
      "grad_norm": 2.623967170715332,
      "learning_rate": 4.92770608375033e-05,
      "loss": 2.3925,
      "step": 553
    },
    {
      "epoch": 0.014590466157492758,
      "grad_norm": 3.6214091777801514,
      "learning_rate": 4.927574400842771e-05,
      "loss": 1.5284,
      "step": 554
    },
    {
      "epoch": 0.014616802739004477,
      "grad_norm": 5.104559898376465,
      "learning_rate": 4.927442717935212e-05,
      "loss": 1.6069,
      "step": 555
    },
    {
      "epoch": 0.014643139320516196,
      "grad_norm": 6.233405113220215,
      "learning_rate": 4.927311035027654e-05,
      "loss": 1.5233,
      "step": 556
    },
    {
      "epoch": 0.014669475902027917,
      "grad_norm": 4.184660911560059,
      "learning_rate": 4.927179352120095e-05,
      "loss": 0.8444,
      "step": 557
    },
    {
      "epoch": 0.014695812483539636,
      "grad_norm": 4.182737827301025,
      "learning_rate": 4.927047669212537e-05,
      "loss": 2.0278,
      "step": 558
    },
    {
      "epoch": 0.014722149065051357,
      "grad_norm": 3.9111337661743164,
      "learning_rate": 4.926915986304978e-05,
      "loss": 1.534,
      "step": 559
    },
    {
      "epoch": 0.014748485646563076,
      "grad_norm": 3.3704192638397217,
      "learning_rate": 4.9267843033974194e-05,
      "loss": 2.1861,
      "step": 560
    },
    {
      "epoch": 0.014774822228074795,
      "grad_norm": 2.4734256267547607,
      "learning_rate": 4.92665262048986e-05,
      "loss": 2.3121,
      "step": 561
    },
    {
      "epoch": 0.014801158809586516,
      "grad_norm": 3.1234734058380127,
      "learning_rate": 4.926520937582302e-05,
      "loss": 1.4346,
      "step": 562
    },
    {
      "epoch": 0.014827495391098235,
      "grad_norm": 5.726578712463379,
      "learning_rate": 4.9263892546747434e-05,
      "loss": 1.6069,
      "step": 563
    },
    {
      "epoch": 0.014853831972609956,
      "grad_norm": 3.6641364097595215,
      "learning_rate": 4.926257571767185e-05,
      "loss": 1.6828,
      "step": 564
    },
    {
      "epoch": 0.014880168554121675,
      "grad_norm": 2.650238037109375,
      "learning_rate": 4.9261258888596265e-05,
      "loss": 1.6859,
      "step": 565
    },
    {
      "epoch": 0.014906505135633396,
      "grad_norm": 3.7750728130340576,
      "learning_rate": 4.9259942059520674e-05,
      "loss": 1.4728,
      "step": 566
    },
    {
      "epoch": 0.014932841717145115,
      "grad_norm": 4.534443378448486,
      "learning_rate": 4.925862523044509e-05,
      "loss": 1.7134,
      "step": 567
    },
    {
      "epoch": 0.014959178298656834,
      "grad_norm": 2.571948528289795,
      "learning_rate": 4.9257308401369505e-05,
      "loss": 0.7603,
      "step": 568
    },
    {
      "epoch": 0.014985514880168554,
      "grad_norm": 3.9538166522979736,
      "learning_rate": 4.925599157229392e-05,
      "loss": 0.983,
      "step": 569
    },
    {
      "epoch": 0.015011851461680274,
      "grad_norm": 3.185822010040283,
      "learning_rate": 4.925467474321833e-05,
      "loss": 2.4089,
      "step": 570
    },
    {
      "epoch": 0.015038188043191994,
      "grad_norm": 3.3399198055267334,
      "learning_rate": 4.9253357914142745e-05,
      "loss": 2.342,
      "step": 571
    },
    {
      "epoch": 0.015064524624703713,
      "grad_norm": 4.784788131713867,
      "learning_rate": 4.925204108506716e-05,
      "loss": 1.4173,
      "step": 572
    },
    {
      "epoch": 0.015090861206215432,
      "grad_norm": 2.2170331478118896,
      "learning_rate": 4.9250724255991576e-05,
      "loss": 2.0383,
      "step": 573
    },
    {
      "epoch": 0.015117197787727153,
      "grad_norm": 4.919084072113037,
      "learning_rate": 4.924940742691599e-05,
      "loss": 1.6311,
      "step": 574
    },
    {
      "epoch": 0.015143534369238872,
      "grad_norm": 3.2493388652801514,
      "learning_rate": 4.92480905978404e-05,
      "loss": 1.2793,
      "step": 575
    },
    {
      "epoch": 0.015169870950750593,
      "grad_norm": 4.027344703674316,
      "learning_rate": 4.9246773768764816e-05,
      "loss": 1.116,
      "step": 576
    },
    {
      "epoch": 0.015196207532262312,
      "grad_norm": 3.5965209007263184,
      "learning_rate": 4.924545693968923e-05,
      "loss": 1.8381,
      "step": 577
    },
    {
      "epoch": 0.015222544113774033,
      "grad_norm": 3.134528636932373,
      "learning_rate": 4.924414011061365e-05,
      "loss": 2.4347,
      "step": 578
    },
    {
      "epoch": 0.015248880695285752,
      "grad_norm": 4.555936336517334,
      "learning_rate": 4.9242823281538056e-05,
      "loss": 1.6732,
      "step": 579
    },
    {
      "epoch": 0.015275217276797471,
      "grad_norm": 3.238778591156006,
      "learning_rate": 4.924150645246247e-05,
      "loss": 1.7157,
      "step": 580
    },
    {
      "epoch": 0.015301553858309192,
      "grad_norm": 3.3418161869049072,
      "learning_rate": 4.924018962338689e-05,
      "loss": 2.3644,
      "step": 581
    },
    {
      "epoch": 0.01532789043982091,
      "grad_norm": 2.421600818634033,
      "learning_rate": 4.92388727943113e-05,
      "loss": 1.5561,
      "step": 582
    },
    {
      "epoch": 0.015354227021332632,
      "grad_norm": 2.887904405593872,
      "learning_rate": 4.923755596523572e-05,
      "loss": 2.036,
      "step": 583
    },
    {
      "epoch": 0.01538056360284435,
      "grad_norm": 5.282592296600342,
      "learning_rate": 4.923623913616013e-05,
      "loss": 0.9187,
      "step": 584
    },
    {
      "epoch": 0.01540690018435607,
      "grad_norm": 5.465512752532959,
      "learning_rate": 4.923492230708454e-05,
      "loss": 1.6847,
      "step": 585
    },
    {
      "epoch": 0.01543323676586779,
      "grad_norm": 3.950766086578369,
      "learning_rate": 4.923360547800896e-05,
      "loss": 1.1451,
      "step": 586
    },
    {
      "epoch": 0.01545957334737951,
      "grad_norm": 3.5993056297302246,
      "learning_rate": 4.9232288648933374e-05,
      "loss": 1.4428,
      "step": 587
    },
    {
      "epoch": 0.01548590992889123,
      "grad_norm": 2.4341535568237305,
      "learning_rate": 4.923097181985778e-05,
      "loss": 2.0307,
      "step": 588
    },
    {
      "epoch": 0.01551224651040295,
      "grad_norm": 5.462024688720703,
      "learning_rate": 4.92296549907822e-05,
      "loss": 2.4178,
      "step": 589
    },
    {
      "epoch": 0.01553858309191467,
      "grad_norm": 2.402330160140991,
      "learning_rate": 4.922833816170661e-05,
      "loss": 1.8485,
      "step": 590
    },
    {
      "epoch": 0.01556491967342639,
      "grad_norm": 2.640392541885376,
      "learning_rate": 4.922702133263103e-05,
      "loss": 1.4291,
      "step": 591
    },
    {
      "epoch": 0.015591256254938108,
      "grad_norm": 2.1738691329956055,
      "learning_rate": 4.922570450355544e-05,
      "loss": 1.9312,
      "step": 592
    },
    {
      "epoch": 0.015617592836449829,
      "grad_norm": 2.7656919956207275,
      "learning_rate": 4.9224387674479854e-05,
      "loss": 1.912,
      "step": 593
    },
    {
      "epoch": 0.015643929417961548,
      "grad_norm": 4.148161888122559,
      "learning_rate": 4.922307084540427e-05,
      "loss": 1.0477,
      "step": 594
    },
    {
      "epoch": 0.015670265999473267,
      "grad_norm": 3.2078702449798584,
      "learning_rate": 4.922175401632868e-05,
      "loss": 2.0759,
      "step": 595
    },
    {
      "epoch": 0.01569660258098499,
      "grad_norm": 4.229988098144531,
      "learning_rate": 4.92204371872531e-05,
      "loss": 1.3012,
      "step": 596
    },
    {
      "epoch": 0.01572293916249671,
      "grad_norm": 9.312508583068848,
      "learning_rate": 4.921912035817751e-05,
      "loss": 1.6295,
      "step": 597
    },
    {
      "epoch": 0.015749275744008428,
      "grad_norm": 4.225595474243164,
      "learning_rate": 4.9217803529101926e-05,
      "loss": 0.9376,
      "step": 598
    },
    {
      "epoch": 0.015775612325520147,
      "grad_norm": 4.413676738739014,
      "learning_rate": 4.9216486700026334e-05,
      "loss": 1.6213,
      "step": 599
    },
    {
      "epoch": 0.015801948907031866,
      "grad_norm": 2.6570732593536377,
      "learning_rate": 4.921516987095075e-05,
      "loss": 0.8403,
      "step": 600
    },
    {
      "epoch": 0.01582828548854359,
      "grad_norm": 2.103667736053467,
      "learning_rate": 4.9213853041875166e-05,
      "loss": 2.0725,
      "step": 601
    },
    {
      "epoch": 0.015854622070055308,
      "grad_norm": 4.095400333404541,
      "learning_rate": 4.921253621279958e-05,
      "loss": 2.4883,
      "step": 602
    },
    {
      "epoch": 0.015880958651567027,
      "grad_norm": 4.9196672439575195,
      "learning_rate": 4.9211219383724e-05,
      "loss": 1.3018,
      "step": 603
    },
    {
      "epoch": 0.015907295233078746,
      "grad_norm": 4.2212653160095215,
      "learning_rate": 4.9209902554648406e-05,
      "loss": 1.6165,
      "step": 604
    },
    {
      "epoch": 0.015933631814590465,
      "grad_norm": 2.3772711753845215,
      "learning_rate": 4.920858572557283e-05,
      "loss": 1.5943,
      "step": 605
    },
    {
      "epoch": 0.015959968396102187,
      "grad_norm": 3.723097085952759,
      "learning_rate": 4.920726889649724e-05,
      "loss": 1.5913,
      "step": 606
    },
    {
      "epoch": 0.015986304977613906,
      "grad_norm": 3.3020520210266113,
      "learning_rate": 4.920595206742165e-05,
      "loss": 1.8639,
      "step": 607
    },
    {
      "epoch": 0.016012641559125625,
      "grad_norm": 3.7389028072357178,
      "learning_rate": 4.920463523834606e-05,
      "loss": 1.9168,
      "step": 608
    },
    {
      "epoch": 0.016038978140637344,
      "grad_norm": 5.393535614013672,
      "learning_rate": 4.920331840927048e-05,
      "loss": 1.916,
      "step": 609
    },
    {
      "epoch": 0.016065314722149063,
      "grad_norm": 6.769744873046875,
      "learning_rate": 4.920200158019489e-05,
      "loss": 1.4624,
      "step": 610
    },
    {
      "epoch": 0.016091651303660786,
      "grad_norm": 5.085109710693359,
      "learning_rate": 4.920068475111931e-05,
      "loss": 1.7557,
      "step": 611
    },
    {
      "epoch": 0.016117987885172505,
      "grad_norm": 2.350942611694336,
      "learning_rate": 4.9199367922043724e-05,
      "loss": 1.8141,
      "step": 612
    },
    {
      "epoch": 0.016144324466684224,
      "grad_norm": 5.252973556518555,
      "learning_rate": 4.919805109296813e-05,
      "loss": 1.4946,
      "step": 613
    },
    {
      "epoch": 0.016170661048195943,
      "grad_norm": 5.195660591125488,
      "learning_rate": 4.919673426389255e-05,
      "loss": 1.6677,
      "step": 614
    },
    {
      "epoch": 0.016196997629707666,
      "grad_norm": 2.2531235218048096,
      "learning_rate": 4.9195417434816964e-05,
      "loss": 1.6768,
      "step": 615
    },
    {
      "epoch": 0.016223334211219385,
      "grad_norm": 2.1095333099365234,
      "learning_rate": 4.919410060574138e-05,
      "loss": 0.4763,
      "step": 616
    },
    {
      "epoch": 0.016249670792731104,
      "grad_norm": 6.1692023277282715,
      "learning_rate": 4.919278377666579e-05,
      "loss": 1.9035,
      "step": 617
    },
    {
      "epoch": 0.016276007374242823,
      "grad_norm": 3.5493061542510986,
      "learning_rate": 4.9191466947590204e-05,
      "loss": 1.9656,
      "step": 618
    },
    {
      "epoch": 0.016302343955754542,
      "grad_norm": 2.4924559593200684,
      "learning_rate": 4.919015011851462e-05,
      "loss": 1.7477,
      "step": 619
    },
    {
      "epoch": 0.016328680537266264,
      "grad_norm": 2.7681593894958496,
      "learning_rate": 4.9188833289439035e-05,
      "loss": 1.2489,
      "step": 620
    },
    {
      "epoch": 0.016355017118777983,
      "grad_norm": 2.5957839488983154,
      "learning_rate": 4.918751646036345e-05,
      "loss": 1.3904,
      "step": 621
    },
    {
      "epoch": 0.016381353700289703,
      "grad_norm": 4.6176228523254395,
      "learning_rate": 4.918619963128786e-05,
      "loss": 1.8987,
      "step": 622
    },
    {
      "epoch": 0.01640769028180142,
      "grad_norm": 2.1113476753234863,
      "learning_rate": 4.9184882802212275e-05,
      "loss": 1.2665,
      "step": 623
    },
    {
      "epoch": 0.01643402686331314,
      "grad_norm": 6.761231422424316,
      "learning_rate": 4.918356597313669e-05,
      "loss": 1.9924,
      "step": 624
    },
    {
      "epoch": 0.016460363444824863,
      "grad_norm": 4.207162857055664,
      "learning_rate": 4.9182249144061106e-05,
      "loss": 1.1325,
      "step": 625
    },
    {
      "epoch": 0.016486700026336582,
      "grad_norm": 5.167304039001465,
      "learning_rate": 4.9180932314985515e-05,
      "loss": 1.551,
      "step": 626
    },
    {
      "epoch": 0.0165130366078483,
      "grad_norm": 2.5618858337402344,
      "learning_rate": 4.917961548590993e-05,
      "loss": 1.9741,
      "step": 627
    },
    {
      "epoch": 0.01653937318936002,
      "grad_norm": 2.5403671264648438,
      "learning_rate": 4.9178298656834346e-05,
      "loss": 2.673,
      "step": 628
    },
    {
      "epoch": 0.01656570977087174,
      "grad_norm": 4.460215091705322,
      "learning_rate": 4.917698182775876e-05,
      "loss": 1.5169,
      "step": 629
    },
    {
      "epoch": 0.016592046352383462,
      "grad_norm": 2.9325146675109863,
      "learning_rate": 4.917566499868318e-05,
      "loss": 0.7922,
      "step": 630
    },
    {
      "epoch": 0.01661838293389518,
      "grad_norm": 3.34602427482605,
      "learning_rate": 4.9174348169607586e-05,
      "loss": 1.6785,
      "step": 631
    },
    {
      "epoch": 0.0166447195154069,
      "grad_norm": 2.8243367671966553,
      "learning_rate": 4.9173031340532e-05,
      "loss": 2.3495,
      "step": 632
    },
    {
      "epoch": 0.01667105609691862,
      "grad_norm": 2.8493475914001465,
      "learning_rate": 4.917171451145641e-05,
      "loss": 1.6853,
      "step": 633
    },
    {
      "epoch": 0.016697392678430338,
      "grad_norm": 4.395769119262695,
      "learning_rate": 4.917039768238083e-05,
      "loss": 0.831,
      "step": 634
    },
    {
      "epoch": 0.01672372925994206,
      "grad_norm": 3.789914846420288,
      "learning_rate": 4.916908085330524e-05,
      "loss": 2.1424,
      "step": 635
    },
    {
      "epoch": 0.01675006584145378,
      "grad_norm": 5.724221706390381,
      "learning_rate": 4.916776402422966e-05,
      "loss": 1.2813,
      "step": 636
    },
    {
      "epoch": 0.0167764024229655,
      "grad_norm": 3.548966884613037,
      "learning_rate": 4.916644719515407e-05,
      "loss": 2.5177,
      "step": 637
    },
    {
      "epoch": 0.016802739004477218,
      "grad_norm": 2.3427090644836426,
      "learning_rate": 4.916513036607849e-05,
      "loss": 1.5009,
      "step": 638
    },
    {
      "epoch": 0.01682907558598894,
      "grad_norm": 2.6451728343963623,
      "learning_rate": 4.9163813537002904e-05,
      "loss": 2.0077,
      "step": 639
    },
    {
      "epoch": 0.01685541216750066,
      "grad_norm": 3.45279860496521,
      "learning_rate": 4.916249670792731e-05,
      "loss": 1.0719,
      "step": 640
    },
    {
      "epoch": 0.01688174874901238,
      "grad_norm": 5.000061988830566,
      "learning_rate": 4.916117987885173e-05,
      "loss": 1.411,
      "step": 641
    },
    {
      "epoch": 0.016908085330524097,
      "grad_norm": 3.6413791179656982,
      "learning_rate": 4.915986304977614e-05,
      "loss": 0.9491,
      "step": 642
    },
    {
      "epoch": 0.016934421912035817,
      "grad_norm": 6.532812595367432,
      "learning_rate": 4.915854622070056e-05,
      "loss": 0.8769,
      "step": 643
    },
    {
      "epoch": 0.01696075849354754,
      "grad_norm": 2.975863218307495,
      "learning_rate": 4.915722939162497e-05,
      "loss": 1.6162,
      "step": 644
    },
    {
      "epoch": 0.016987095075059258,
      "grad_norm": 2.6892645359039307,
      "learning_rate": 4.9155912562549384e-05,
      "loss": 1.9541,
      "step": 645
    },
    {
      "epoch": 0.017013431656570977,
      "grad_norm": 7.905243396759033,
      "learning_rate": 4.915459573347379e-05,
      "loss": 2.6824,
      "step": 646
    },
    {
      "epoch": 0.017039768238082696,
      "grad_norm": 2.5596559047698975,
      "learning_rate": 4.915327890439821e-05,
      "loss": 1.8071,
      "step": 647
    },
    {
      "epoch": 0.017066104819594415,
      "grad_norm": 2.167295455932617,
      "learning_rate": 4.9151962075322624e-05,
      "loss": 1.6388,
      "step": 648
    },
    {
      "epoch": 0.017092441401106138,
      "grad_norm": 4.472445964813232,
      "learning_rate": 4.915064524624704e-05,
      "loss": 0.4865,
      "step": 649
    },
    {
      "epoch": 0.017118777982617857,
      "grad_norm": 2.858304262161255,
      "learning_rate": 4.9149328417171455e-05,
      "loss": 2.0971,
      "step": 650
    },
    {
      "epoch": 0.017145114564129576,
      "grad_norm": 2.73447585105896,
      "learning_rate": 4.9148011588095864e-05,
      "loss": 0.6435,
      "step": 651
    },
    {
      "epoch": 0.017171451145641295,
      "grad_norm": 3.694622039794922,
      "learning_rate": 4.9146694759020287e-05,
      "loss": 1.6498,
      "step": 652
    },
    {
      "epoch": 0.017197787727153014,
      "grad_norm": 2.649141311645508,
      "learning_rate": 4.9145377929944695e-05,
      "loss": 2.7969,
      "step": 653
    },
    {
      "epoch": 0.017224124308664737,
      "grad_norm": 2.7107765674591064,
      "learning_rate": 4.914406110086911e-05,
      "loss": 1.9972,
      "step": 654
    },
    {
      "epoch": 0.017250460890176456,
      "grad_norm": 6.076614856719971,
      "learning_rate": 4.914274427179352e-05,
      "loss": 0.6942,
      "step": 655
    },
    {
      "epoch": 0.017276797471688175,
      "grad_norm": 3.9620161056518555,
      "learning_rate": 4.9141427442717935e-05,
      "loss": 2.1558,
      "step": 656
    },
    {
      "epoch": 0.017303134053199894,
      "grad_norm": 6.982167720794678,
      "learning_rate": 4.914011061364235e-05,
      "loss": 1.713,
      "step": 657
    },
    {
      "epoch": 0.017329470634711613,
      "grad_norm": 6.892648696899414,
      "learning_rate": 4.9138793784566767e-05,
      "loss": 1.5932,
      "step": 658
    },
    {
      "epoch": 0.017355807216223335,
      "grad_norm": 4.177656650543213,
      "learning_rate": 4.913747695549118e-05,
      "loss": 1.1373,
      "step": 659
    },
    {
      "epoch": 0.017382143797735054,
      "grad_norm": 3.7838337421417236,
      "learning_rate": 4.913616012641559e-05,
      "loss": 1.3375,
      "step": 660
    },
    {
      "epoch": 0.017408480379246773,
      "grad_norm": 2.8625714778900146,
      "learning_rate": 4.9134843297340007e-05,
      "loss": 1.4474,
      "step": 661
    },
    {
      "epoch": 0.017434816960758492,
      "grad_norm": 2.763807773590088,
      "learning_rate": 4.913352646826442e-05,
      "loss": 2.0358,
      "step": 662
    },
    {
      "epoch": 0.017461153542270215,
      "grad_norm": 4.135056972503662,
      "learning_rate": 4.913220963918884e-05,
      "loss": 1.5602,
      "step": 663
    },
    {
      "epoch": 0.017487490123781934,
      "grad_norm": 2.3488876819610596,
      "learning_rate": 4.9130892810113247e-05,
      "loss": 1.8612,
      "step": 664
    },
    {
      "epoch": 0.017513826705293653,
      "grad_norm": 3.223292112350464,
      "learning_rate": 4.912957598103766e-05,
      "loss": 1.9454,
      "step": 665
    },
    {
      "epoch": 0.017540163286805372,
      "grad_norm": 9.494622230529785,
      "learning_rate": 4.912825915196208e-05,
      "loss": 0.6274,
      "step": 666
    },
    {
      "epoch": 0.01756649986831709,
      "grad_norm": 2.9268500804901123,
      "learning_rate": 4.912694232288649e-05,
      "loss": 2.4056,
      "step": 667
    },
    {
      "epoch": 0.017592836449828814,
      "grad_norm": 2.4697117805480957,
      "learning_rate": 4.912562549381091e-05,
      "loss": 2.1795,
      "step": 668
    },
    {
      "epoch": 0.017619173031340533,
      "grad_norm": 3.0642313957214355,
      "learning_rate": 4.912430866473532e-05,
      "loss": 1.8113,
      "step": 669
    },
    {
      "epoch": 0.017645509612852252,
      "grad_norm": 2.972248077392578,
      "learning_rate": 4.912299183565973e-05,
      "loss": 1.6875,
      "step": 670
    },
    {
      "epoch": 0.01767184619436397,
      "grad_norm": 2.488313913345337,
      "learning_rate": 4.912167500658415e-05,
      "loss": 2.3337,
      "step": 671
    },
    {
      "epoch": 0.01769818277587569,
      "grad_norm": 3.433907985687256,
      "learning_rate": 4.9120358177508565e-05,
      "loss": 1.2048,
      "step": 672
    },
    {
      "epoch": 0.017724519357387412,
      "grad_norm": 3.9574859142303467,
      "learning_rate": 4.9119041348432973e-05,
      "loss": 1.9056,
      "step": 673
    },
    {
      "epoch": 0.01775085593889913,
      "grad_norm": 2.6303348541259766,
      "learning_rate": 4.911772451935739e-05,
      "loss": 1.4656,
      "step": 674
    },
    {
      "epoch": 0.01777719252041085,
      "grad_norm": 4.3974714279174805,
      "learning_rate": 4.9116407690281805e-05,
      "loss": 2.0426,
      "step": 675
    },
    {
      "epoch": 0.01780352910192257,
      "grad_norm": 3.659027099609375,
      "learning_rate": 4.911509086120622e-05,
      "loss": 2.0193,
      "step": 676
    },
    {
      "epoch": 0.01782986568343429,
      "grad_norm": 3.02225923538208,
      "learning_rate": 4.9113774032130636e-05,
      "loss": 1.5299,
      "step": 677
    },
    {
      "epoch": 0.01785620226494601,
      "grad_norm": 2.1771230697631836,
      "learning_rate": 4.9112457203055045e-05,
      "loss": 2.0673,
      "step": 678
    },
    {
      "epoch": 0.01788253884645773,
      "grad_norm": 3.230940580368042,
      "learning_rate": 4.911114037397946e-05,
      "loss": 1.759,
      "step": 679
    },
    {
      "epoch": 0.01790887542796945,
      "grad_norm": 3.779658794403076,
      "learning_rate": 4.910982354490387e-05,
      "loss": 1.4368,
      "step": 680
    },
    {
      "epoch": 0.01793521200948117,
      "grad_norm": 3.737619638442993,
      "learning_rate": 4.910850671582829e-05,
      "loss": 1.7809,
      "step": 681
    },
    {
      "epoch": 0.01796154859099289,
      "grad_norm": 3.0055973529815674,
      "learning_rate": 4.91071898867527e-05,
      "loss": 1.9685,
      "step": 682
    },
    {
      "epoch": 0.01798788517250461,
      "grad_norm": 4.044579982757568,
      "learning_rate": 4.9105873057677116e-05,
      "loss": 1.2692,
      "step": 683
    },
    {
      "epoch": 0.01801422175401633,
      "grad_norm": 2.2516746520996094,
      "learning_rate": 4.910455622860153e-05,
      "loss": 2.1363,
      "step": 684
    },
    {
      "epoch": 0.018040558335528048,
      "grad_norm": 3.4395387172698975,
      "learning_rate": 4.910323939952595e-05,
      "loss": 1.7178,
      "step": 685
    },
    {
      "epoch": 0.018066894917039767,
      "grad_norm": 2.8364453315734863,
      "learning_rate": 4.910192257045036e-05,
      "loss": 1.5717,
      "step": 686
    },
    {
      "epoch": 0.01809323149855149,
      "grad_norm": 2.7511515617370605,
      "learning_rate": 4.910060574137477e-05,
      "loss": 2.0651,
      "step": 687
    },
    {
      "epoch": 0.01811956808006321,
      "grad_norm": 4.945553779602051,
      "learning_rate": 4.909928891229919e-05,
      "loss": 2.0403,
      "step": 688
    },
    {
      "epoch": 0.018145904661574928,
      "grad_norm": 3.2428159713745117,
      "learning_rate": 4.9097972083223596e-05,
      "loss": 1.2601,
      "step": 689
    },
    {
      "epoch": 0.018172241243086647,
      "grad_norm": 2.8357133865356445,
      "learning_rate": 4.909665525414802e-05,
      "loss": 1.9201,
      "step": 690
    },
    {
      "epoch": 0.018198577824598366,
      "grad_norm": 3.0857584476470947,
      "learning_rate": 4.909533842507243e-05,
      "loss": 2.613,
      "step": 691
    },
    {
      "epoch": 0.01822491440611009,
      "grad_norm": 3.888237953186035,
      "learning_rate": 4.909402159599684e-05,
      "loss": 1.3062,
      "step": 692
    },
    {
      "epoch": 0.018251250987621807,
      "grad_norm": 3.859483003616333,
      "learning_rate": 4.909270476692125e-05,
      "loss": 1.2223,
      "step": 693
    },
    {
      "epoch": 0.018277587569133526,
      "grad_norm": 2.710242986679077,
      "learning_rate": 4.909138793784567e-05,
      "loss": 0.5843,
      "step": 694
    },
    {
      "epoch": 0.018303924150645245,
      "grad_norm": 4.768449783325195,
      "learning_rate": 4.909007110877008e-05,
      "loss": 1.3698,
      "step": 695
    },
    {
      "epoch": 0.018330260732156965,
      "grad_norm": 4.321253299713135,
      "learning_rate": 4.90887542796945e-05,
      "loss": 0.5832,
      "step": 696
    },
    {
      "epoch": 0.018356597313668687,
      "grad_norm": 2.6788434982299805,
      "learning_rate": 4.9087437450618914e-05,
      "loss": 2.3529,
      "step": 697
    },
    {
      "epoch": 0.018382933895180406,
      "grad_norm": 4.308788299560547,
      "learning_rate": 4.908612062154332e-05,
      "loss": 2.8389,
      "step": 698
    },
    {
      "epoch": 0.018409270476692125,
      "grad_norm": 3.0056040287017822,
      "learning_rate": 4.908480379246774e-05,
      "loss": 1.7889,
      "step": 699
    },
    {
      "epoch": 0.018435607058203844,
      "grad_norm": 2.3902392387390137,
      "learning_rate": 4.9083486963392154e-05,
      "loss": 1.9402,
      "step": 700
    },
    {
      "epoch": 0.018461943639715563,
      "grad_norm": 2.8440802097320557,
      "learning_rate": 4.908217013431657e-05,
      "loss": 1.4952,
      "step": 701
    },
    {
      "epoch": 0.018488280221227286,
      "grad_norm": 3.7777483463287354,
      "learning_rate": 4.908085330524098e-05,
      "loss": 0.997,
      "step": 702
    },
    {
      "epoch": 0.018514616802739005,
      "grad_norm": 4.508836269378662,
      "learning_rate": 4.9079536476165394e-05,
      "loss": 1.8724,
      "step": 703
    },
    {
      "epoch": 0.018540953384250724,
      "grad_norm": 4.095557689666748,
      "learning_rate": 4.907821964708981e-05,
      "loss": 1.6751,
      "step": 704
    },
    {
      "epoch": 0.018567289965762443,
      "grad_norm": 3.1924331188201904,
      "learning_rate": 4.9076902818014225e-05,
      "loss": 2.0006,
      "step": 705
    },
    {
      "epoch": 0.018593626547274165,
      "grad_norm": 4.989907264709473,
      "learning_rate": 4.907558598893864e-05,
      "loss": 1.7909,
      "step": 706
    },
    {
      "epoch": 0.018619963128785885,
      "grad_norm": 2.632371187210083,
      "learning_rate": 4.907426915986305e-05,
      "loss": 2.3802,
      "step": 707
    },
    {
      "epoch": 0.018646299710297604,
      "grad_norm": 3.216073513031006,
      "learning_rate": 4.9072952330787465e-05,
      "loss": 1.894,
      "step": 708
    },
    {
      "epoch": 0.018672636291809323,
      "grad_norm": 3.9569344520568848,
      "learning_rate": 4.907163550171188e-05,
      "loss": 1.0289,
      "step": 709
    },
    {
      "epoch": 0.01869897287332104,
      "grad_norm": 2.3757307529449463,
      "learning_rate": 4.9070318672636296e-05,
      "loss": 1.5536,
      "step": 710
    },
    {
      "epoch": 0.018725309454832764,
      "grad_norm": 4.2826714515686035,
      "learning_rate": 4.9069001843560705e-05,
      "loss": 2.2823,
      "step": 711
    },
    {
      "epoch": 0.018751646036344483,
      "grad_norm": 3.4693210124969482,
      "learning_rate": 4.906768501448512e-05,
      "loss": 1.5999,
      "step": 712
    },
    {
      "epoch": 0.018777982617856202,
      "grad_norm": 3.102646589279175,
      "learning_rate": 4.9066368185409536e-05,
      "loss": 0.3172,
      "step": 713
    },
    {
      "epoch": 0.01880431919936792,
      "grad_norm": 4.541385650634766,
      "learning_rate": 4.906505135633395e-05,
      "loss": 1.5117,
      "step": 714
    },
    {
      "epoch": 0.01883065578087964,
      "grad_norm": 2.585900068283081,
      "learning_rate": 4.906373452725837e-05,
      "loss": 1.6201,
      "step": 715
    },
    {
      "epoch": 0.018856992362391363,
      "grad_norm": 2.852032423019409,
      "learning_rate": 4.9062417698182776e-05,
      "loss": 2.2742,
      "step": 716
    },
    {
      "epoch": 0.018883328943903082,
      "grad_norm": 2.8967807292938232,
      "learning_rate": 4.906110086910719e-05,
      "loss": 1.8815,
      "step": 717
    },
    {
      "epoch": 0.0189096655254148,
      "grad_norm": 7.6833038330078125,
      "learning_rate": 4.905978404003161e-05,
      "loss": 2.237,
      "step": 718
    },
    {
      "epoch": 0.01893600210692652,
      "grad_norm": 3.026221990585327,
      "learning_rate": 4.905846721095602e-05,
      "loss": 1.4727,
      "step": 719
    },
    {
      "epoch": 0.01896233868843824,
      "grad_norm": 7.7613725662231445,
      "learning_rate": 4.905715038188043e-05,
      "loss": 1.9785,
      "step": 720
    },
    {
      "epoch": 0.01898867526994996,
      "grad_norm": 23.023826599121094,
      "learning_rate": 4.905583355280485e-05,
      "loss": 1.1939,
      "step": 721
    },
    {
      "epoch": 0.01901501185146168,
      "grad_norm": 3.0799660682678223,
      "learning_rate": 4.905451672372926e-05,
      "loss": 2.0895,
      "step": 722
    },
    {
      "epoch": 0.0190413484329734,
      "grad_norm": 4.517638683319092,
      "learning_rate": 4.905319989465368e-05,
      "loss": 2.0233,
      "step": 723
    },
    {
      "epoch": 0.01906768501448512,
      "grad_norm": 6.13850212097168,
      "learning_rate": 4.9051883065578094e-05,
      "loss": 1.5015,
      "step": 724
    },
    {
      "epoch": 0.019094021595996838,
      "grad_norm": 4.092638969421387,
      "learning_rate": 4.90505662365025e-05,
      "loss": 1.8314,
      "step": 725
    },
    {
      "epoch": 0.01912035817750856,
      "grad_norm": 2.1235930919647217,
      "learning_rate": 4.904924940742692e-05,
      "loss": 2.2646,
      "step": 726
    },
    {
      "epoch": 0.01914669475902028,
      "grad_norm": 2.9502296447753906,
      "learning_rate": 4.904793257835133e-05,
      "loss": 0.6506,
      "step": 727
    },
    {
      "epoch": 0.019173031340532,
      "grad_norm": 3.1551878452301025,
      "learning_rate": 4.904661574927575e-05,
      "loss": 1.7719,
      "step": 728
    },
    {
      "epoch": 0.019199367922043718,
      "grad_norm": 2.6724605560302734,
      "learning_rate": 4.904529892020016e-05,
      "loss": 1.792,
      "step": 729
    },
    {
      "epoch": 0.01922570450355544,
      "grad_norm": 2.949495553970337,
      "learning_rate": 4.9043982091124574e-05,
      "loss": 1.4856,
      "step": 730
    },
    {
      "epoch": 0.01925204108506716,
      "grad_norm": 2.914775848388672,
      "learning_rate": 4.904266526204899e-05,
      "loss": 1.5858,
      "step": 731
    },
    {
      "epoch": 0.019278377666578878,
      "grad_norm": 18.118562698364258,
      "learning_rate": 4.90413484329734e-05,
      "loss": 2.7145,
      "step": 732
    },
    {
      "epoch": 0.019304714248090597,
      "grad_norm": 2.8546435832977295,
      "learning_rate": 4.904003160389782e-05,
      "loss": 1.5072,
      "step": 733
    },
    {
      "epoch": 0.019331050829602316,
      "grad_norm": 2.6304736137390137,
      "learning_rate": 4.903871477482223e-05,
      "loss": 2.0668,
      "step": 734
    },
    {
      "epoch": 0.01935738741111404,
      "grad_norm": 2.861255407333374,
      "learning_rate": 4.9037397945746646e-05,
      "loss": 1.0789,
      "step": 735
    },
    {
      "epoch": 0.019383723992625758,
      "grad_norm": 3.318653106689453,
      "learning_rate": 4.9036081116671054e-05,
      "loss": 1.6155,
      "step": 736
    },
    {
      "epoch": 0.019410060574137477,
      "grad_norm": 2.467630386352539,
      "learning_rate": 4.903476428759548e-05,
      "loss": 1.9833,
      "step": 737
    },
    {
      "epoch": 0.019436397155649196,
      "grad_norm": 2.600247859954834,
      "learning_rate": 4.9033447458519886e-05,
      "loss": 1.5517,
      "step": 738
    },
    {
      "epoch": 0.019462733737160915,
      "grad_norm": 6.090360641479492,
      "learning_rate": 4.90321306294443e-05,
      "loss": 1.1927,
      "step": 739
    },
    {
      "epoch": 0.019489070318672638,
      "grad_norm": 2.7624895572662354,
      "learning_rate": 4.903081380036872e-05,
      "loss": 1.5016,
      "step": 740
    },
    {
      "epoch": 0.019515406900184357,
      "grad_norm": 3.2997429370880127,
      "learning_rate": 4.9029496971293126e-05,
      "loss": 1.8181,
      "step": 741
    },
    {
      "epoch": 0.019541743481696076,
      "grad_norm": 2.4257280826568604,
      "learning_rate": 4.902818014221755e-05,
      "loss": 1.5971,
      "step": 742
    },
    {
      "epoch": 0.019568080063207795,
      "grad_norm": 5.636271953582764,
      "learning_rate": 4.902686331314196e-05,
      "loss": 2.5826,
      "step": 743
    },
    {
      "epoch": 0.019594416644719514,
      "grad_norm": 4.720231056213379,
      "learning_rate": 4.902554648406637e-05,
      "loss": 0.7437,
      "step": 744
    },
    {
      "epoch": 0.019620753226231236,
      "grad_norm": 2.4664206504821777,
      "learning_rate": 4.902422965499078e-05,
      "loss": 2.8211,
      "step": 745
    },
    {
      "epoch": 0.019647089807742955,
      "grad_norm": 3.704176664352417,
      "learning_rate": 4.90229128259152e-05,
      "loss": 1.4692,
      "step": 746
    },
    {
      "epoch": 0.019673426389254674,
      "grad_norm": 3.1931588649749756,
      "learning_rate": 4.902159599683961e-05,
      "loss": 1.7259,
      "step": 747
    },
    {
      "epoch": 0.019699762970766393,
      "grad_norm": 3.2398006916046143,
      "learning_rate": 4.902027916776403e-05,
      "loss": 2.1453,
      "step": 748
    },
    {
      "epoch": 0.019726099552278116,
      "grad_norm": 2.211707830429077,
      "learning_rate": 4.901896233868844e-05,
      "loss": 1.4725,
      "step": 749
    },
    {
      "epoch": 0.019752436133789835,
      "grad_norm": 5.693408966064453,
      "learning_rate": 4.901764550961285e-05,
      "loss": 2.1259,
      "step": 750
    },
    {
      "epoch": 0.019778772715301554,
      "grad_norm": 4.840297698974609,
      "learning_rate": 4.901632868053727e-05,
      "loss": 1.5723,
      "step": 751
    },
    {
      "epoch": 0.019805109296813273,
      "grad_norm": 2.4746127128601074,
      "learning_rate": 4.9015011851461684e-05,
      "loss": 2.0466,
      "step": 752
    },
    {
      "epoch": 0.019831445878324992,
      "grad_norm": 3.711047410964966,
      "learning_rate": 4.90136950223861e-05,
      "loss": 0.8246,
      "step": 753
    },
    {
      "epoch": 0.019857782459836715,
      "grad_norm": 3.0407891273498535,
      "learning_rate": 4.901237819331051e-05,
      "loss": 2.0245,
      "step": 754
    },
    {
      "epoch": 0.019884119041348434,
      "grad_norm": 4.244345188140869,
      "learning_rate": 4.9011061364234924e-05,
      "loss": 1.5863,
      "step": 755
    },
    {
      "epoch": 0.019910455622860153,
      "grad_norm": 3.2105660438537598,
      "learning_rate": 4.900974453515934e-05,
      "loss": 1.7174,
      "step": 756
    },
    {
      "epoch": 0.019936792204371872,
      "grad_norm": 3.018162727355957,
      "learning_rate": 4.9008427706083755e-05,
      "loss": 1.5234,
      "step": 757
    },
    {
      "epoch": 0.01996312878588359,
      "grad_norm": 5.20952033996582,
      "learning_rate": 4.9007110877008164e-05,
      "loss": 1.5424,
      "step": 758
    },
    {
      "epoch": 0.019989465367395313,
      "grad_norm": 4.107244968414307,
      "learning_rate": 4.900579404793258e-05,
      "loss": 2.334,
      "step": 759
    },
    {
      "epoch": 0.020015801948907033,
      "grad_norm": 3.29659104347229,
      "learning_rate": 4.9004477218856995e-05,
      "loss": 1.7371,
      "step": 760
    },
    {
      "epoch": 0.02004213853041875,
      "grad_norm": 2.391526699066162,
      "learning_rate": 4.900316038978141e-05,
      "loss": 1.4182,
      "step": 761
    },
    {
      "epoch": 0.02006847511193047,
      "grad_norm": 3.2081947326660156,
      "learning_rate": 4.9001843560705826e-05,
      "loss": 2.1381,
      "step": 762
    },
    {
      "epoch": 0.02009481169344219,
      "grad_norm": 5.759424209594727,
      "learning_rate": 4.9000526731630235e-05,
      "loss": 1.0129,
      "step": 763
    },
    {
      "epoch": 0.020121148274953912,
      "grad_norm": 2.460491895675659,
      "learning_rate": 4.899920990255465e-05,
      "loss": 2.0876,
      "step": 764
    },
    {
      "epoch": 0.02014748485646563,
      "grad_norm": 2.915313482284546,
      "learning_rate": 4.899789307347906e-05,
      "loss": 2.4913,
      "step": 765
    },
    {
      "epoch": 0.02017382143797735,
      "grad_norm": 2.943474531173706,
      "learning_rate": 4.899657624440348e-05,
      "loss": 1.2272,
      "step": 766
    },
    {
      "epoch": 0.02020015801948907,
      "grad_norm": 3.699199676513672,
      "learning_rate": 4.899525941532789e-05,
      "loss": 0.9989,
      "step": 767
    },
    {
      "epoch": 0.02022649460100079,
      "grad_norm": 2.4398043155670166,
      "learning_rate": 4.8993942586252306e-05,
      "loss": 1.7294,
      "step": 768
    },
    {
      "epoch": 0.02025283118251251,
      "grad_norm": 2.853638172149658,
      "learning_rate": 4.899262575717672e-05,
      "loss": 1.5957,
      "step": 769
    },
    {
      "epoch": 0.02027916776402423,
      "grad_norm": 2.8390755653381348,
      "learning_rate": 4.899130892810114e-05,
      "loss": 1.4267,
      "step": 770
    },
    {
      "epoch": 0.02030550434553595,
      "grad_norm": 2.1481523513793945,
      "learning_rate": 4.898999209902555e-05,
      "loss": 2.0185,
      "step": 771
    },
    {
      "epoch": 0.020331840927047668,
      "grad_norm": 2.8952174186706543,
      "learning_rate": 4.898867526994996e-05,
      "loss": 2.0734,
      "step": 772
    },
    {
      "epoch": 0.02035817750855939,
      "grad_norm": 4.551745891571045,
      "learning_rate": 4.898735844087438e-05,
      "loss": 1.5128,
      "step": 773
    },
    {
      "epoch": 0.02038451409007111,
      "grad_norm": 2.6201367378234863,
      "learning_rate": 4.8986041611798786e-05,
      "loss": 1.4021,
      "step": 774
    },
    {
      "epoch": 0.02041085067158283,
      "grad_norm": 4.821719169616699,
      "learning_rate": 4.898472478272321e-05,
      "loss": 2.5525,
      "step": 775
    },
    {
      "epoch": 0.020437187253094548,
      "grad_norm": 2.4338624477386475,
      "learning_rate": 4.898340795364762e-05,
      "loss": 1.7702,
      "step": 776
    },
    {
      "epoch": 0.020463523834606267,
      "grad_norm": 2.9143950939178467,
      "learning_rate": 4.898209112457203e-05,
      "loss": 1.8072,
      "step": 777
    },
    {
      "epoch": 0.02048986041611799,
      "grad_norm": 4.292383193969727,
      "learning_rate": 4.898077429549645e-05,
      "loss": 2.1108,
      "step": 778
    },
    {
      "epoch": 0.02051619699762971,
      "grad_norm": 3.321052312850952,
      "learning_rate": 4.897945746642086e-05,
      "loss": 1.9431,
      "step": 779
    },
    {
      "epoch": 0.020542533579141427,
      "grad_norm": 3.4524850845336914,
      "learning_rate": 4.897814063734528e-05,
      "loss": 1.8244,
      "step": 780
    },
    {
      "epoch": 0.020568870160653147,
      "grad_norm": 3.016355514526367,
      "learning_rate": 4.897682380826969e-05,
      "loss": 1.8159,
      "step": 781
    },
    {
      "epoch": 0.020595206742164866,
      "grad_norm": 9.822920799255371,
      "learning_rate": 4.8975506979194104e-05,
      "loss": 2.6316,
      "step": 782
    },
    {
      "epoch": 0.020621543323676588,
      "grad_norm": 3.655283212661743,
      "learning_rate": 4.897419015011851e-05,
      "loss": 1.9178,
      "step": 783
    },
    {
      "epoch": 0.020647879905188307,
      "grad_norm": 3.4776971340179443,
      "learning_rate": 4.8972873321042935e-05,
      "loss": 0.8009,
      "step": 784
    },
    {
      "epoch": 0.020674216486700026,
      "grad_norm": 5.085859298706055,
      "learning_rate": 4.8971556491967344e-05,
      "loss": 1.2723,
      "step": 785
    },
    {
      "epoch": 0.020700553068211745,
      "grad_norm": 2.645732879638672,
      "learning_rate": 4.897023966289176e-05,
      "loss": 1.8745,
      "step": 786
    },
    {
      "epoch": 0.020726889649723464,
      "grad_norm": 2.9395639896392822,
      "learning_rate": 4.8968922833816175e-05,
      "loss": 1.4894,
      "step": 787
    },
    {
      "epoch": 0.020753226231235187,
      "grad_norm": 7.118697166442871,
      "learning_rate": 4.8967606004740584e-05,
      "loss": 1.4846,
      "step": 788
    },
    {
      "epoch": 0.020779562812746906,
      "grad_norm": 3.920042037963867,
      "learning_rate": 4.8966289175665007e-05,
      "loss": 1.4626,
      "step": 789
    },
    {
      "epoch": 0.020805899394258625,
      "grad_norm": 4.029362678527832,
      "learning_rate": 4.8964972346589415e-05,
      "loss": 2.1308,
      "step": 790
    },
    {
      "epoch": 0.020832235975770344,
      "grad_norm": 2.5646917819976807,
      "learning_rate": 4.896365551751383e-05,
      "loss": 0.8227,
      "step": 791
    },
    {
      "epoch": 0.020858572557282063,
      "grad_norm": 4.099377155303955,
      "learning_rate": 4.896233868843824e-05,
      "loss": 1.7736,
      "step": 792
    },
    {
      "epoch": 0.020884909138793786,
      "grad_norm": 3.346822500228882,
      "learning_rate": 4.8961021859362655e-05,
      "loss": 2.4753,
      "step": 793
    },
    {
      "epoch": 0.020911245720305505,
      "grad_norm": 3.1177546977996826,
      "learning_rate": 4.895970503028707e-05,
      "loss": 1.2981,
      "step": 794
    },
    {
      "epoch": 0.020937582301817224,
      "grad_norm": 3.4955661296844482,
      "learning_rate": 4.8958388201211487e-05,
      "loss": 1.9598,
      "step": 795
    },
    {
      "epoch": 0.020963918883328943,
      "grad_norm": 3.7434422969818115,
      "learning_rate": 4.8957071372135895e-05,
      "loss": 1.9448,
      "step": 796
    },
    {
      "epoch": 0.020990255464840665,
      "grad_norm": 4.377191543579102,
      "learning_rate": 4.895575454306031e-05,
      "loss": 1.5413,
      "step": 797
    },
    {
      "epoch": 0.021016592046352384,
      "grad_norm": 3.4647278785705566,
      "learning_rate": 4.8954437713984727e-05,
      "loss": 1.4311,
      "step": 798
    },
    {
      "epoch": 0.021042928627864103,
      "grad_norm": 3.394336700439453,
      "learning_rate": 4.895312088490914e-05,
      "loss": 1.3913,
      "step": 799
    },
    {
      "epoch": 0.021069265209375822,
      "grad_norm": 3.9088380336761475,
      "learning_rate": 4.895180405583356e-05,
      "loss": 1.6976,
      "step": 800
    },
    {
      "epoch": 0.02109560179088754,
      "grad_norm": 2.859862804412842,
      "learning_rate": 4.895048722675797e-05,
      "loss": 1.5352,
      "step": 801
    },
    {
      "epoch": 0.021121938372399264,
      "grad_norm": 2.3032243251800537,
      "learning_rate": 4.894917039768238e-05,
      "loss": 2.3919,
      "step": 802
    },
    {
      "epoch": 0.021148274953910983,
      "grad_norm": 4.2017130851745605,
      "learning_rate": 4.89478535686068e-05,
      "loss": 1.2697,
      "step": 803
    },
    {
      "epoch": 0.021174611535422702,
      "grad_norm": 3.5838441848754883,
      "learning_rate": 4.8946536739531213e-05,
      "loss": 1.6274,
      "step": 804
    },
    {
      "epoch": 0.02120094811693442,
      "grad_norm": 3.4935455322265625,
      "learning_rate": 4.894521991045562e-05,
      "loss": 2.0953,
      "step": 805
    },
    {
      "epoch": 0.02122728469844614,
      "grad_norm": 4.362836837768555,
      "learning_rate": 4.894390308138004e-05,
      "loss": 1.5424,
      "step": 806
    },
    {
      "epoch": 0.021253621279957863,
      "grad_norm": 7.931687831878662,
      "learning_rate": 4.8942586252304453e-05,
      "loss": 1.2894,
      "step": 807
    },
    {
      "epoch": 0.021279957861469582,
      "grad_norm": 2.71730375289917,
      "learning_rate": 4.894126942322887e-05,
      "loss": 2.1413,
      "step": 808
    },
    {
      "epoch": 0.0213062944429813,
      "grad_norm": 2.596869707107544,
      "learning_rate": 4.8939952594153285e-05,
      "loss": 1.5955,
      "step": 809
    },
    {
      "epoch": 0.02133263102449302,
      "grad_norm": 7.385660648345947,
      "learning_rate": 4.8938635765077693e-05,
      "loss": 1.9132,
      "step": 810
    },
    {
      "epoch": 0.02135896760600474,
      "grad_norm": 2.570417881011963,
      "learning_rate": 4.893731893600211e-05,
      "loss": 1.4441,
      "step": 811
    },
    {
      "epoch": 0.02138530418751646,
      "grad_norm": 2.4142816066741943,
      "learning_rate": 4.893600210692652e-05,
      "loss": 1.8909,
      "step": 812
    },
    {
      "epoch": 0.02141164076902818,
      "grad_norm": 3.1070990562438965,
      "learning_rate": 4.893468527785094e-05,
      "loss": 1.5658,
      "step": 813
    },
    {
      "epoch": 0.0214379773505399,
      "grad_norm": 2.599963665008545,
      "learning_rate": 4.893336844877535e-05,
      "loss": 1.5471,
      "step": 814
    },
    {
      "epoch": 0.02146431393205162,
      "grad_norm": 5.090512275695801,
      "learning_rate": 4.8932051619699765e-05,
      "loss": 1.8339,
      "step": 815
    },
    {
      "epoch": 0.02149065051356334,
      "grad_norm": 4.4836907386779785,
      "learning_rate": 4.893073479062418e-05,
      "loss": 2.67,
      "step": 816
    },
    {
      "epoch": 0.02151698709507506,
      "grad_norm": 4.416508197784424,
      "learning_rate": 4.8929417961548596e-05,
      "loss": 1.3435,
      "step": 817
    },
    {
      "epoch": 0.02154332367658678,
      "grad_norm": 3.6350066661834717,
      "learning_rate": 4.892810113247301e-05,
      "loss": 2.2413,
      "step": 818
    },
    {
      "epoch": 0.0215696602580985,
      "grad_norm": 4.3631768226623535,
      "learning_rate": 4.892678430339742e-05,
      "loss": 1.1512,
      "step": 819
    },
    {
      "epoch": 0.021595996839610217,
      "grad_norm": 3.968855381011963,
      "learning_rate": 4.8925467474321836e-05,
      "loss": 1.4528,
      "step": 820
    },
    {
      "epoch": 0.02162233342112194,
      "grad_norm": 3.9882712364196777,
      "learning_rate": 4.8924150645246245e-05,
      "loss": 1.3471,
      "step": 821
    },
    {
      "epoch": 0.02164867000263366,
      "grad_norm": 4.810586929321289,
      "learning_rate": 4.892283381617067e-05,
      "loss": 1.8159,
      "step": 822
    },
    {
      "epoch": 0.021675006584145378,
      "grad_norm": 4.532617092132568,
      "learning_rate": 4.8921516987095076e-05,
      "loss": 2.188,
      "step": 823
    },
    {
      "epoch": 0.021701343165657097,
      "grad_norm": 4.6967339515686035,
      "learning_rate": 4.892020015801949e-05,
      "loss": 1.9399,
      "step": 824
    },
    {
      "epoch": 0.021727679747168816,
      "grad_norm": 6.225020408630371,
      "learning_rate": 4.891888332894391e-05,
      "loss": 1.5084,
      "step": 825
    },
    {
      "epoch": 0.02175401632868054,
      "grad_norm": 2.7183237075805664,
      "learning_rate": 4.8917566499868316e-05,
      "loss": 1.7584,
      "step": 826
    },
    {
      "epoch": 0.021780352910192258,
      "grad_norm": 3.1652307510375977,
      "learning_rate": 4.891624967079274e-05,
      "loss": 2.3683,
      "step": 827
    },
    {
      "epoch": 0.021806689491703977,
      "grad_norm": 3.4707798957824707,
      "learning_rate": 4.891493284171715e-05,
      "loss": 2.1044,
      "step": 828
    },
    {
      "epoch": 0.021833026073215696,
      "grad_norm": 5.230329990386963,
      "learning_rate": 4.891361601264156e-05,
      "loss": 2.9414,
      "step": 829
    },
    {
      "epoch": 0.021859362654727415,
      "grad_norm": 3.530433177947998,
      "learning_rate": 4.891229918356597e-05,
      "loss": 1.5021,
      "step": 830
    },
    {
      "epoch": 0.021885699236239137,
      "grad_norm": 2.7317306995391846,
      "learning_rate": 4.8910982354490394e-05,
      "loss": 2.3066,
      "step": 831
    },
    {
      "epoch": 0.021912035817750856,
      "grad_norm": 2.825913667678833,
      "learning_rate": 4.89096655254148e-05,
      "loss": 1.836,
      "step": 832
    },
    {
      "epoch": 0.021938372399262576,
      "grad_norm": 4.459882736206055,
      "learning_rate": 4.890834869633922e-05,
      "loss": 1.3156,
      "step": 833
    },
    {
      "epoch": 0.021964708980774295,
      "grad_norm": 3.0017170906066895,
      "learning_rate": 4.8907031867263634e-05,
      "loss": 1.903,
      "step": 834
    },
    {
      "epoch": 0.021991045562286014,
      "grad_norm": 2.3982746601104736,
      "learning_rate": 4.890571503818804e-05,
      "loss": 2.0684,
      "step": 835
    },
    {
      "epoch": 0.022017382143797736,
      "grad_norm": 2.809511661529541,
      "learning_rate": 4.8904398209112465e-05,
      "loss": 0.8368,
      "step": 836
    },
    {
      "epoch": 0.022043718725309455,
      "grad_norm": 5.6450581550598145,
      "learning_rate": 4.8903081380036874e-05,
      "loss": 1.9314,
      "step": 837
    },
    {
      "epoch": 0.022070055306821174,
      "grad_norm": 3.1256988048553467,
      "learning_rate": 4.890176455096129e-05,
      "loss": 1.7095,
      "step": 838
    },
    {
      "epoch": 0.022096391888332893,
      "grad_norm": 7.690213680267334,
      "learning_rate": 4.89004477218857e-05,
      "loss": 1.7726,
      "step": 839
    },
    {
      "epoch": 0.022122728469844616,
      "grad_norm": 2.41968035697937,
      "learning_rate": 4.8899130892810114e-05,
      "loss": 2.0054,
      "step": 840
    },
    {
      "epoch": 0.022149065051356335,
      "grad_norm": 3.7235939502716064,
      "learning_rate": 4.889781406373453e-05,
      "loss": 1.6424,
      "step": 841
    },
    {
      "epoch": 0.022175401632868054,
      "grad_norm": 4.181987285614014,
      "learning_rate": 4.8896497234658945e-05,
      "loss": 1.497,
      "step": 842
    },
    {
      "epoch": 0.022201738214379773,
      "grad_norm": 4.181909561157227,
      "learning_rate": 4.8895180405583354e-05,
      "loss": 2.0824,
      "step": 843
    },
    {
      "epoch": 0.022228074795891492,
      "grad_norm": 4.448643684387207,
      "learning_rate": 4.889386357650777e-05,
      "loss": 1.926,
      "step": 844
    },
    {
      "epoch": 0.022254411377403215,
      "grad_norm": 2.937082052230835,
      "learning_rate": 4.8892546747432185e-05,
      "loss": 2.2453,
      "step": 845
    },
    {
      "epoch": 0.022280747958914934,
      "grad_norm": 3.1048316955566406,
      "learning_rate": 4.88912299183566e-05,
      "loss": 2.1669,
      "step": 846
    },
    {
      "epoch": 0.022307084540426653,
      "grad_norm": 3.082481861114502,
      "learning_rate": 4.8889913089281016e-05,
      "loss": 1.7621,
      "step": 847
    },
    {
      "epoch": 0.02233342112193837,
      "grad_norm": 3.7697834968566895,
      "learning_rate": 4.8888596260205425e-05,
      "loss": 1.3528,
      "step": 848
    },
    {
      "epoch": 0.02235975770345009,
      "grad_norm": 7.551332473754883,
      "learning_rate": 4.888727943112984e-05,
      "loss": 1.1198,
      "step": 849
    },
    {
      "epoch": 0.022386094284961813,
      "grad_norm": 3.309126615524292,
      "learning_rate": 4.8885962602054256e-05,
      "loss": 2.1807,
      "step": 850
    },
    {
      "epoch": 0.022412430866473532,
      "grad_norm": 4.420648574829102,
      "learning_rate": 4.888464577297867e-05,
      "loss": 1.4397,
      "step": 851
    },
    {
      "epoch": 0.02243876744798525,
      "grad_norm": 3.5129315853118896,
      "learning_rate": 4.888332894390308e-05,
      "loss": 0.6918,
      "step": 852
    },
    {
      "epoch": 0.02246510402949697,
      "grad_norm": 3.1707139015197754,
      "learning_rate": 4.8882012114827496e-05,
      "loss": 0.9939,
      "step": 853
    },
    {
      "epoch": 0.02249144061100869,
      "grad_norm": 3.777784824371338,
      "learning_rate": 4.888069528575191e-05,
      "loss": 2.0356,
      "step": 854
    },
    {
      "epoch": 0.022517777192520412,
      "grad_norm": 5.508716106414795,
      "learning_rate": 4.887937845667633e-05,
      "loss": 2.249,
      "step": 855
    },
    {
      "epoch": 0.02254411377403213,
      "grad_norm": 6.5840864181518555,
      "learning_rate": 4.887806162760074e-05,
      "loss": 1.6953,
      "step": 856
    },
    {
      "epoch": 0.02257045035554385,
      "grad_norm": 4.6469340324401855,
      "learning_rate": 4.887674479852515e-05,
      "loss": 1.1643,
      "step": 857
    },
    {
      "epoch": 0.02259678693705557,
      "grad_norm": 3.110201835632324,
      "learning_rate": 4.887542796944957e-05,
      "loss": 2.6273,
      "step": 858
    },
    {
      "epoch": 0.022623123518567288,
      "grad_norm": 2.776529550552368,
      "learning_rate": 4.8874111140373976e-05,
      "loss": 1.2361,
      "step": 859
    },
    {
      "epoch": 0.02264946010007901,
      "grad_norm": 4.608916759490967,
      "learning_rate": 4.88727943112984e-05,
      "loss": 1.5542,
      "step": 860
    },
    {
      "epoch": 0.02267579668159073,
      "grad_norm": 4.610592842102051,
      "learning_rate": 4.887147748222281e-05,
      "loss": 1.5604,
      "step": 861
    },
    {
      "epoch": 0.02270213326310245,
      "grad_norm": 2.4414761066436768,
      "learning_rate": 4.887016065314722e-05,
      "loss": 1.8716,
      "step": 862
    },
    {
      "epoch": 0.022728469844614168,
      "grad_norm": 4.946755886077881,
      "learning_rate": 4.886884382407164e-05,
      "loss": 1.2964,
      "step": 863
    },
    {
      "epoch": 0.02275480642612589,
      "grad_norm": 3.669976234436035,
      "learning_rate": 4.8867526994996054e-05,
      "loss": 1.6591,
      "step": 864
    },
    {
      "epoch": 0.02278114300763761,
      "grad_norm": 2.9368083477020264,
      "learning_rate": 4.886621016592047e-05,
      "loss": 2.0373,
      "step": 865
    },
    {
      "epoch": 0.02280747958914933,
      "grad_norm": 2.7219672203063965,
      "learning_rate": 4.886489333684488e-05,
      "loss": 1.6246,
      "step": 866
    },
    {
      "epoch": 0.022833816170661048,
      "grad_norm": 5.060786247253418,
      "learning_rate": 4.8863576507769294e-05,
      "loss": 2.0179,
      "step": 867
    },
    {
      "epoch": 0.022860152752172767,
      "grad_norm": 5.938994407653809,
      "learning_rate": 4.88622596786937e-05,
      "loss": 1.6806,
      "step": 868
    },
    {
      "epoch": 0.02288648933368449,
      "grad_norm": 3.05311918258667,
      "learning_rate": 4.8860942849618126e-05,
      "loss": 2.0176,
      "step": 869
    },
    {
      "epoch": 0.022912825915196208,
      "grad_norm": 3.7118985652923584,
      "learning_rate": 4.8859626020542534e-05,
      "loss": 1.1776,
      "step": 870
    },
    {
      "epoch": 0.022939162496707927,
      "grad_norm": 2.4207539558410645,
      "learning_rate": 4.885830919146695e-05,
      "loss": 1.9499,
      "step": 871
    },
    {
      "epoch": 0.022965499078219646,
      "grad_norm": 2.7396440505981445,
      "learning_rate": 4.8856992362391366e-05,
      "loss": 0.7162,
      "step": 872
    },
    {
      "epoch": 0.022991835659731365,
      "grad_norm": 3.2223732471466064,
      "learning_rate": 4.8855675533315774e-05,
      "loss": 1.4836,
      "step": 873
    },
    {
      "epoch": 0.023018172241243088,
      "grad_norm": 3.0279431343078613,
      "learning_rate": 4.88543587042402e-05,
      "loss": 1.8642,
      "step": 874
    },
    {
      "epoch": 0.023044508822754807,
      "grad_norm": 3.1469645500183105,
      "learning_rate": 4.8853041875164606e-05,
      "loss": 1.6819,
      "step": 875
    },
    {
      "epoch": 0.023070845404266526,
      "grad_norm": 4.9359450340271,
      "learning_rate": 4.885172504608902e-05,
      "loss": 1.52,
      "step": 876
    },
    {
      "epoch": 0.023097181985778245,
      "grad_norm": 5.223402500152588,
      "learning_rate": 4.885040821701343e-05,
      "loss": 0.8004,
      "step": 877
    },
    {
      "epoch": 0.023123518567289964,
      "grad_norm": 3.0356979370117188,
      "learning_rate": 4.8849091387937846e-05,
      "loss": 0.2715,
      "step": 878
    },
    {
      "epoch": 0.023149855148801687,
      "grad_norm": 2.9502663612365723,
      "learning_rate": 4.884777455886226e-05,
      "loss": 1.8226,
      "step": 879
    },
    {
      "epoch": 0.023176191730313406,
      "grad_norm": 3.033193349838257,
      "learning_rate": 4.884645772978668e-05,
      "loss": 1.9013,
      "step": 880
    },
    {
      "epoch": 0.023202528311825125,
      "grad_norm": 4.26116418838501,
      "learning_rate": 4.884514090071109e-05,
      "loss": 1.8828,
      "step": 881
    },
    {
      "epoch": 0.023228864893336844,
      "grad_norm": 2.8875274658203125,
      "learning_rate": 4.88438240716355e-05,
      "loss": 1.5716,
      "step": 882
    },
    {
      "epoch": 0.023255201474848566,
      "grad_norm": 4.26268196105957,
      "learning_rate": 4.8842507242559924e-05,
      "loss": 1.6109,
      "step": 883
    },
    {
      "epoch": 0.023281538056360285,
      "grad_norm": 2.3405840396881104,
      "learning_rate": 4.884119041348433e-05,
      "loss": 3.0292,
      "step": 884
    },
    {
      "epoch": 0.023307874637872004,
      "grad_norm": 2.4629173278808594,
      "learning_rate": 4.883987358440875e-05,
      "loss": 1.7301,
      "step": 885
    },
    {
      "epoch": 0.023334211219383724,
      "grad_norm": 2.497011184692383,
      "learning_rate": 4.883855675533316e-05,
      "loss": 2.2863,
      "step": 886
    },
    {
      "epoch": 0.023360547800895443,
      "grad_norm": 2.809114694595337,
      "learning_rate": 4.883723992625757e-05,
      "loss": 1.8682,
      "step": 887
    },
    {
      "epoch": 0.023386884382407165,
      "grad_norm": 2.7719645500183105,
      "learning_rate": 4.883592309718199e-05,
      "loss": 2.3763,
      "step": 888
    },
    {
      "epoch": 0.023413220963918884,
      "grad_norm": 2.340416669845581,
      "learning_rate": 4.8834606268106404e-05,
      "loss": 1.559,
      "step": 889
    },
    {
      "epoch": 0.023439557545430603,
      "grad_norm": 5.768289566040039,
      "learning_rate": 4.883328943903082e-05,
      "loss": 1.3442,
      "step": 890
    },
    {
      "epoch": 0.023465894126942322,
      "grad_norm": 3.9083545207977295,
      "learning_rate": 4.883197260995523e-05,
      "loss": 1.5086,
      "step": 891
    },
    {
      "epoch": 0.02349223070845404,
      "grad_norm": 2.550734043121338,
      "learning_rate": 4.8830655780879644e-05,
      "loss": 1.7015,
      "step": 892
    },
    {
      "epoch": 0.023518567289965764,
      "grad_norm": 2.9778783321380615,
      "learning_rate": 4.882933895180406e-05,
      "loss": 1.2605,
      "step": 893
    },
    {
      "epoch": 0.023544903871477483,
      "grad_norm": 2.945540189743042,
      "learning_rate": 4.8828022122728475e-05,
      "loss": 1.8471,
      "step": 894
    },
    {
      "epoch": 0.023571240452989202,
      "grad_norm": 3.3573033809661865,
      "learning_rate": 4.8826705293652884e-05,
      "loss": 1.4592,
      "step": 895
    },
    {
      "epoch": 0.02359757703450092,
      "grad_norm": 8.201980590820312,
      "learning_rate": 4.88253884645773e-05,
      "loss": 2.0542,
      "step": 896
    },
    {
      "epoch": 0.02362391361601264,
      "grad_norm": 2.681422710418701,
      "learning_rate": 4.8824071635501715e-05,
      "loss": 1.8801,
      "step": 897
    },
    {
      "epoch": 0.023650250197524363,
      "grad_norm": 2.5939226150512695,
      "learning_rate": 4.882275480642613e-05,
      "loss": 2.2178,
      "step": 898
    },
    {
      "epoch": 0.02367658677903608,
      "grad_norm": 4.881478786468506,
      "learning_rate": 4.882143797735054e-05,
      "loss": 2.1732,
      "step": 899
    },
    {
      "epoch": 0.0237029233605478,
      "grad_norm": 3.984933614730835,
      "learning_rate": 4.8820121148274955e-05,
      "loss": 1.7548,
      "step": 900
    },
    {
      "epoch": 0.02372925994205952,
      "grad_norm": 3.310772180557251,
      "learning_rate": 4.881880431919937e-05,
      "loss": 0.8692,
      "step": 901
    },
    {
      "epoch": 0.02375559652357124,
      "grad_norm": 5.751379489898682,
      "learning_rate": 4.8817487490123786e-05,
      "loss": 1.0979,
      "step": 902
    },
    {
      "epoch": 0.02378193310508296,
      "grad_norm": 4.656223297119141,
      "learning_rate": 4.88161706610482e-05,
      "loss": 0.528,
      "step": 903
    },
    {
      "epoch": 0.02380826968659468,
      "grad_norm": 3.370290756225586,
      "learning_rate": 4.881485383197261e-05,
      "loss": 2.2476,
      "step": 904
    },
    {
      "epoch": 0.0238346062681064,
      "grad_norm": 6.551933765411377,
      "learning_rate": 4.8813537002897026e-05,
      "loss": 1.2507,
      "step": 905
    },
    {
      "epoch": 0.02386094284961812,
      "grad_norm": 3.8577675819396973,
      "learning_rate": 4.8812220173821435e-05,
      "loss": 1.8234,
      "step": 906
    },
    {
      "epoch": 0.02388727943112984,
      "grad_norm": 2.82926344871521,
      "learning_rate": 4.881090334474586e-05,
      "loss": 1.7066,
      "step": 907
    },
    {
      "epoch": 0.02391361601264156,
      "grad_norm": 2.4536755084991455,
      "learning_rate": 4.8809586515670266e-05,
      "loss": 1.5696,
      "step": 908
    },
    {
      "epoch": 0.02393995259415328,
      "grad_norm": 3.422820568084717,
      "learning_rate": 4.880826968659468e-05,
      "loss": 0.8928,
      "step": 909
    },
    {
      "epoch": 0.023966289175664998,
      "grad_norm": 2.8297548294067383,
      "learning_rate": 4.88069528575191e-05,
      "loss": 0.3006,
      "step": 910
    },
    {
      "epoch": 0.023992625757176717,
      "grad_norm": 5.321094036102295,
      "learning_rate": 4.8805636028443506e-05,
      "loss": 2.1824,
      "step": 911
    },
    {
      "epoch": 0.02401896233868844,
      "grad_norm": 2.4695184230804443,
      "learning_rate": 4.880431919936793e-05,
      "loss": 1.9922,
      "step": 912
    },
    {
      "epoch": 0.02404529892020016,
      "grad_norm": 4.551111221313477,
      "learning_rate": 4.880300237029234e-05,
      "loss": 2.2532,
      "step": 913
    },
    {
      "epoch": 0.024071635501711878,
      "grad_norm": 3.917362689971924,
      "learning_rate": 4.880168554121675e-05,
      "loss": 1.858,
      "step": 914
    },
    {
      "epoch": 0.024097972083223597,
      "grad_norm": 2.7697012424468994,
      "learning_rate": 4.880036871214116e-05,
      "loss": 1.8184,
      "step": 915
    },
    {
      "epoch": 0.024124308664735316,
      "grad_norm": 2.403337240219116,
      "learning_rate": 4.8799051883065584e-05,
      "loss": 1.4952,
      "step": 916
    },
    {
      "epoch": 0.02415064524624704,
      "grad_norm": 3.2855231761932373,
      "learning_rate": 4.879773505398999e-05,
      "loss": 2.4549,
      "step": 917
    },
    {
      "epoch": 0.024176981827758758,
      "grad_norm": 5.837979316711426,
      "learning_rate": 4.879641822491441e-05,
      "loss": 1.776,
      "step": 918
    },
    {
      "epoch": 0.024203318409270477,
      "grad_norm": 3.956073760986328,
      "learning_rate": 4.8795101395838824e-05,
      "loss": 1.4027,
      "step": 919
    },
    {
      "epoch": 0.024229654990782196,
      "grad_norm": 3.2240607738494873,
      "learning_rate": 4.879378456676323e-05,
      "loss": 2.0036,
      "step": 920
    },
    {
      "epoch": 0.024255991572293915,
      "grad_norm": 3.176830530166626,
      "learning_rate": 4.8792467737687655e-05,
      "loss": 2.0702,
      "step": 921
    },
    {
      "epoch": 0.024282328153805637,
      "grad_norm": 3.58725905418396,
      "learning_rate": 4.8791150908612064e-05,
      "loss": 1.8628,
      "step": 922
    },
    {
      "epoch": 0.024308664735317356,
      "grad_norm": 5.654665946960449,
      "learning_rate": 4.878983407953648e-05,
      "loss": 1.5798,
      "step": 923
    },
    {
      "epoch": 0.024335001316829075,
      "grad_norm": 7.691973686218262,
      "learning_rate": 4.878851725046089e-05,
      "loss": 1.8847,
      "step": 924
    },
    {
      "epoch": 0.024361337898340794,
      "grad_norm": 2.611201524734497,
      "learning_rate": 4.8787200421385304e-05,
      "loss": 2.0848,
      "step": 925
    },
    {
      "epoch": 0.024387674479852513,
      "grad_norm": 2.717235565185547,
      "learning_rate": 4.878588359230972e-05,
      "loss": 1.8219,
      "step": 926
    },
    {
      "epoch": 0.024414011061364236,
      "grad_norm": 3.0322349071502686,
      "learning_rate": 4.8784566763234135e-05,
      "loss": 1.9138,
      "step": 927
    },
    {
      "epoch": 0.024440347642875955,
      "grad_norm": 2.383579730987549,
      "learning_rate": 4.878324993415855e-05,
      "loss": 2.0439,
      "step": 928
    },
    {
      "epoch": 0.024466684224387674,
      "grad_norm": 3.828047752380371,
      "learning_rate": 4.878193310508296e-05,
      "loss": 1.931,
      "step": 929
    },
    {
      "epoch": 0.024493020805899393,
      "grad_norm": 3.5176422595977783,
      "learning_rate": 4.878061627600738e-05,
      "loss": 1.4164,
      "step": 930
    },
    {
      "epoch": 0.024519357387411116,
      "grad_norm": 2.933224678039551,
      "learning_rate": 4.877929944693179e-05,
      "loss": 2.2594,
      "step": 931
    },
    {
      "epoch": 0.024545693968922835,
      "grad_norm": 7.70943546295166,
      "learning_rate": 4.877798261785621e-05,
      "loss": 1.0859,
      "step": 932
    },
    {
      "epoch": 0.024572030550434554,
      "grad_norm": 2.605595111846924,
      "learning_rate": 4.8776665788780615e-05,
      "loss": 1.6807,
      "step": 933
    },
    {
      "epoch": 0.024598367131946273,
      "grad_norm": 3.604996681213379,
      "learning_rate": 4.877534895970503e-05,
      "loss": 1.4888,
      "step": 934
    },
    {
      "epoch": 0.024624703713457992,
      "grad_norm": 2.6662745475769043,
      "learning_rate": 4.877403213062945e-05,
      "loss": 1.8947,
      "step": 935
    },
    {
      "epoch": 0.024651040294969714,
      "grad_norm": 3.2520012855529785,
      "learning_rate": 4.877271530155386e-05,
      "loss": 1.7753,
      "step": 936
    },
    {
      "epoch": 0.024677376876481433,
      "grad_norm": 4.311880111694336,
      "learning_rate": 4.877139847247828e-05,
      "loss": 2.1511,
      "step": 937
    },
    {
      "epoch": 0.024703713457993152,
      "grad_norm": 2.115872859954834,
      "learning_rate": 4.877008164340269e-05,
      "loss": 1.6899,
      "step": 938
    },
    {
      "epoch": 0.02473005003950487,
      "grad_norm": 4.769501209259033,
      "learning_rate": 4.87687648143271e-05,
      "loss": 2.4844,
      "step": 939
    },
    {
      "epoch": 0.02475638662101659,
      "grad_norm": 3.7659108638763428,
      "learning_rate": 4.876744798525152e-05,
      "loss": 2.0844,
      "step": 940
    },
    {
      "epoch": 0.024782723202528313,
      "grad_norm": 2.8463897705078125,
      "learning_rate": 4.8766131156175933e-05,
      "loss": 1.9853,
      "step": 941
    },
    {
      "epoch": 0.024809059784040032,
      "grad_norm": 2.3476593494415283,
      "learning_rate": 4.876481432710034e-05,
      "loss": 2.2732,
      "step": 942
    },
    {
      "epoch": 0.02483539636555175,
      "grad_norm": 3.762220859527588,
      "learning_rate": 4.876349749802476e-05,
      "loss": 1.3429,
      "step": 943
    },
    {
      "epoch": 0.02486173294706347,
      "grad_norm": 3.1902225017547607,
      "learning_rate": 4.876218066894917e-05,
      "loss": 0.9517,
      "step": 944
    },
    {
      "epoch": 0.02488806952857519,
      "grad_norm": 2.5766420364379883,
      "learning_rate": 4.876086383987359e-05,
      "loss": 2.0754,
      "step": 945
    },
    {
      "epoch": 0.024914406110086912,
      "grad_norm": 3.0132415294647217,
      "learning_rate": 4.8759547010798e-05,
      "loss": 1.9503,
      "step": 946
    },
    {
      "epoch": 0.02494074269159863,
      "grad_norm": 4.317477703094482,
      "learning_rate": 4.8758230181722413e-05,
      "loss": 1.7543,
      "step": 947
    },
    {
      "epoch": 0.02496707927311035,
      "grad_norm": 2.949892997741699,
      "learning_rate": 4.875691335264683e-05,
      "loss": 0.9036,
      "step": 948
    },
    {
      "epoch": 0.02499341585462207,
      "grad_norm": 3.8822638988494873,
      "learning_rate": 4.8755596523571245e-05,
      "loss": 0.9061,
      "step": 949
    },
    {
      "epoch": 0.02501975243613379,
      "grad_norm": 3.780977487564087,
      "learning_rate": 4.875427969449566e-05,
      "loss": 1.4815,
      "step": 950
    },
    {
      "epoch": 0.02504608901764551,
      "grad_norm": 2.225395441055298,
      "learning_rate": 4.875296286542007e-05,
      "loss": 1.3013,
      "step": 951
    },
    {
      "epoch": 0.02507242559915723,
      "grad_norm": 3.3168656826019287,
      "learning_rate": 4.8751646036344485e-05,
      "loss": 1.5468,
      "step": 952
    },
    {
      "epoch": 0.02509876218066895,
      "grad_norm": 2.268721580505371,
      "learning_rate": 4.8750329207268894e-05,
      "loss": 1.5199,
      "step": 953
    },
    {
      "epoch": 0.025125098762180668,
      "grad_norm": 2.3924734592437744,
      "learning_rate": 4.8749012378193316e-05,
      "loss": 1.9441,
      "step": 954
    },
    {
      "epoch": 0.02515143534369239,
      "grad_norm": 2.3553285598754883,
      "learning_rate": 4.8747695549117725e-05,
      "loss": 1.9051,
      "step": 955
    },
    {
      "epoch": 0.02517777192520411,
      "grad_norm": 2.352037191390991,
      "learning_rate": 4.874637872004214e-05,
      "loss": 1.7318,
      "step": 956
    },
    {
      "epoch": 0.02520410850671583,
      "grad_norm": 2.3553106784820557,
      "learning_rate": 4.8745061890966556e-05,
      "loss": 1.4107,
      "step": 957
    },
    {
      "epoch": 0.025230445088227547,
      "grad_norm": 4.828738689422607,
      "learning_rate": 4.8743745061890965e-05,
      "loss": 1.0074,
      "step": 958
    },
    {
      "epoch": 0.025256781669739266,
      "grad_norm": 3.7099461555480957,
      "learning_rate": 4.874242823281539e-05,
      "loss": 2.2068,
      "step": 959
    },
    {
      "epoch": 0.02528311825125099,
      "grad_norm": 2.071575164794922,
      "learning_rate": 4.8741111403739796e-05,
      "loss": 1.8892,
      "step": 960
    },
    {
      "epoch": 0.025309454832762708,
      "grad_norm": 6.233997344970703,
      "learning_rate": 4.873979457466421e-05,
      "loss": 1.186,
      "step": 961
    },
    {
      "epoch": 0.025335791414274427,
      "grad_norm": 5.112436294555664,
      "learning_rate": 4.873847774558862e-05,
      "loss": 1.8068,
      "step": 962
    },
    {
      "epoch": 0.025362127995786146,
      "grad_norm": 4.417271614074707,
      "learning_rate": 4.873716091651304e-05,
      "loss": 2.0767,
      "step": 963
    },
    {
      "epoch": 0.025388464577297865,
      "grad_norm": 5.198521137237549,
      "learning_rate": 4.873584408743745e-05,
      "loss": 2.3583,
      "step": 964
    },
    {
      "epoch": 0.025414801158809588,
      "grad_norm": 2.9751198291778564,
      "learning_rate": 4.873452725836187e-05,
      "loss": 1.74,
      "step": 965
    },
    {
      "epoch": 0.025441137740321307,
      "grad_norm": 3.3860409259796143,
      "learning_rate": 4.873321042928628e-05,
      "loss": 1.3247,
      "step": 966
    },
    {
      "epoch": 0.025467474321833026,
      "grad_norm": 4.9931817054748535,
      "learning_rate": 4.873189360021069e-05,
      "loss": 1.3301,
      "step": 967
    },
    {
      "epoch": 0.025493810903344745,
      "grad_norm": 7.048219203948975,
      "learning_rate": 4.8730576771135114e-05,
      "loss": 1.6089,
      "step": 968
    },
    {
      "epoch": 0.025520147484856464,
      "grad_norm": 3.70931339263916,
      "learning_rate": 4.872925994205952e-05,
      "loss": 1.8571,
      "step": 969
    },
    {
      "epoch": 0.025546484066368186,
      "grad_norm": 4.149421691894531,
      "learning_rate": 4.872794311298394e-05,
      "loss": 1.6011,
      "step": 970
    },
    {
      "epoch": 0.025572820647879906,
      "grad_norm": 2.326721429824829,
      "learning_rate": 4.872662628390835e-05,
      "loss": 2.4341,
      "step": 971
    },
    {
      "epoch": 0.025599157229391625,
      "grad_norm": 2.405155897140503,
      "learning_rate": 4.872530945483276e-05,
      "loss": 1.7249,
      "step": 972
    },
    {
      "epoch": 0.025625493810903344,
      "grad_norm": 4.748103141784668,
      "learning_rate": 4.872399262575718e-05,
      "loss": 1.9832,
      "step": 973
    },
    {
      "epoch": 0.025651830392415066,
      "grad_norm": 3.56463360786438,
      "learning_rate": 4.8722675796681594e-05,
      "loss": 1.6517,
      "step": 974
    },
    {
      "epoch": 0.025678166973926785,
      "grad_norm": 3.736293315887451,
      "learning_rate": 4.872135896760601e-05,
      "loss": 0.6747,
      "step": 975
    },
    {
      "epoch": 0.025704503555438504,
      "grad_norm": 2.9828007221221924,
      "learning_rate": 4.872004213853042e-05,
      "loss": 1.765,
      "step": 976
    },
    {
      "epoch": 0.025730840136950223,
      "grad_norm": 8.252352714538574,
      "learning_rate": 4.8718725309454834e-05,
      "loss": 1.6381,
      "step": 977
    },
    {
      "epoch": 0.025757176718461942,
      "grad_norm": 4.802109241485596,
      "learning_rate": 4.871740848037925e-05,
      "loss": 1.2956,
      "step": 978
    },
    {
      "epoch": 0.025783513299973665,
      "grad_norm": 4.678725242614746,
      "learning_rate": 4.8716091651303665e-05,
      "loss": 1.8981,
      "step": 979
    },
    {
      "epoch": 0.025809849881485384,
      "grad_norm": 2.763256072998047,
      "learning_rate": 4.8714774822228074e-05,
      "loss": 0.8677,
      "step": 980
    },
    {
      "epoch": 0.025836186462997103,
      "grad_norm": 3.9678878784179688,
      "learning_rate": 4.871345799315249e-05,
      "loss": 1.1441,
      "step": 981
    },
    {
      "epoch": 0.025862523044508822,
      "grad_norm": 4.467024803161621,
      "learning_rate": 4.8712141164076905e-05,
      "loss": 1.3829,
      "step": 982
    },
    {
      "epoch": 0.02588885962602054,
      "grad_norm": 2.9426639080047607,
      "learning_rate": 4.871082433500132e-05,
      "loss": 1.5174,
      "step": 983
    },
    {
      "epoch": 0.025915196207532264,
      "grad_norm": 2.438995838165283,
      "learning_rate": 4.8709507505925736e-05,
      "loss": 1.7923,
      "step": 984
    },
    {
      "epoch": 0.025941532789043983,
      "grad_norm": 2.7595555782318115,
      "learning_rate": 4.8708190676850145e-05,
      "loss": 2.3838,
      "step": 985
    },
    {
      "epoch": 0.025967869370555702,
      "grad_norm": 3.703192949295044,
      "learning_rate": 4.870687384777456e-05,
      "loss": 1.6882,
      "step": 986
    },
    {
      "epoch": 0.02599420595206742,
      "grad_norm": 2.6279776096343994,
      "learning_rate": 4.8705557018698976e-05,
      "loss": 1.0209,
      "step": 987
    },
    {
      "epoch": 0.02602054253357914,
      "grad_norm": 3.901630163192749,
      "learning_rate": 4.870424018962339e-05,
      "loss": 1.2321,
      "step": 988
    },
    {
      "epoch": 0.026046879115090862,
      "grad_norm": 3.1039302349090576,
      "learning_rate": 4.87029233605478e-05,
      "loss": 0.7191,
      "step": 989
    },
    {
      "epoch": 0.02607321569660258,
      "grad_norm": 5.460488319396973,
      "learning_rate": 4.8701606531472216e-05,
      "loss": 2.5224,
      "step": 990
    },
    {
      "epoch": 0.0260995522781143,
      "grad_norm": 4.716031551361084,
      "learning_rate": 4.870028970239663e-05,
      "loss": 1.3963,
      "step": 991
    },
    {
      "epoch": 0.02612588885962602,
      "grad_norm": 7.198510646820068,
      "learning_rate": 4.869897287332105e-05,
      "loss": 1.13,
      "step": 992
    },
    {
      "epoch": 0.02615222544113774,
      "grad_norm": 3.0804200172424316,
      "learning_rate": 4.869765604424546e-05,
      "loss": 0.5474,
      "step": 993
    },
    {
      "epoch": 0.02617856202264946,
      "grad_norm": 2.0527169704437256,
      "learning_rate": 4.869633921516987e-05,
      "loss": 1.741,
      "step": 994
    },
    {
      "epoch": 0.02620489860416118,
      "grad_norm": 2.9041318893432617,
      "learning_rate": 4.869502238609429e-05,
      "loss": 1.4097,
      "step": 995
    },
    {
      "epoch": 0.0262312351856729,
      "grad_norm": 2.4357457160949707,
      "learning_rate": 4.86937055570187e-05,
      "loss": 1.7655,
      "step": 996
    },
    {
      "epoch": 0.02625757176718462,
      "grad_norm": 2.678795337677002,
      "learning_rate": 4.869238872794312e-05,
      "loss": 1.855,
      "step": 997
    },
    {
      "epoch": 0.02628390834869634,
      "grad_norm": 2.4673001766204834,
      "learning_rate": 4.869107189886753e-05,
      "loss": 1.6506,
      "step": 998
    },
    {
      "epoch": 0.02631024493020806,
      "grad_norm": 2.4799981117248535,
      "learning_rate": 4.868975506979194e-05,
      "loss": 1.8883,
      "step": 999
    },
    {
      "epoch": 0.02633658151171978,
      "grad_norm": 2.46384859085083,
      "learning_rate": 4.868843824071635e-05,
      "loss": 1.5749,
      "step": 1000
    },
    {
      "epoch": 0.026362918093231498,
      "grad_norm": 3.096233606338501,
      "learning_rate": 4.8687121411640774e-05,
      "loss": 1.5581,
      "step": 1001
    },
    {
      "epoch": 0.026389254674743217,
      "grad_norm": 2.272268295288086,
      "learning_rate": 4.868580458256518e-05,
      "loss": 1.3949,
      "step": 1002
    },
    {
      "epoch": 0.02641559125625494,
      "grad_norm": 3.7771518230438232,
      "learning_rate": 4.86844877534896e-05,
      "loss": 1.4578,
      "step": 1003
    },
    {
      "epoch": 0.02644192783776666,
      "grad_norm": 4.052642822265625,
      "learning_rate": 4.8683170924414014e-05,
      "loss": 1.7712,
      "step": 1004
    },
    {
      "epoch": 0.026468264419278378,
      "grad_norm": 2.4844229221343994,
      "learning_rate": 4.868185409533842e-05,
      "loss": 2.2215,
      "step": 1005
    },
    {
      "epoch": 0.026494601000790097,
      "grad_norm": 4.846006870269775,
      "learning_rate": 4.8680537266262846e-05,
      "loss": 0.8832,
      "step": 1006
    },
    {
      "epoch": 0.026520937582301816,
      "grad_norm": 2.821293354034424,
      "learning_rate": 4.8679220437187254e-05,
      "loss": 1.6234,
      "step": 1007
    },
    {
      "epoch": 0.02654727416381354,
      "grad_norm": 2.3138506412506104,
      "learning_rate": 4.867790360811167e-05,
      "loss": 1.7553,
      "step": 1008
    },
    {
      "epoch": 0.026573610745325257,
      "grad_norm": 2.4005398750305176,
      "learning_rate": 4.867658677903608e-05,
      "loss": 1.8875,
      "step": 1009
    },
    {
      "epoch": 0.026599947326836976,
      "grad_norm": 7.436602592468262,
      "learning_rate": 4.8675269949960495e-05,
      "loss": 2.3947,
      "step": 1010
    },
    {
      "epoch": 0.026626283908348695,
      "grad_norm": 3.919088840484619,
      "learning_rate": 4.867395312088491e-05,
      "loss": 1.2254,
      "step": 1011
    },
    {
      "epoch": 0.026652620489860415,
      "grad_norm": 7.609036922454834,
      "learning_rate": 4.8672636291809326e-05,
      "loss": 1.5265,
      "step": 1012
    },
    {
      "epoch": 0.026678957071372137,
      "grad_norm": 2.81717586517334,
      "learning_rate": 4.867131946273374e-05,
      "loss": 1.7644,
      "step": 1013
    },
    {
      "epoch": 0.026705293652883856,
      "grad_norm": 4.284199237823486,
      "learning_rate": 4.867000263365815e-05,
      "loss": 2.0247,
      "step": 1014
    },
    {
      "epoch": 0.026731630234395575,
      "grad_norm": 6.873014450073242,
      "learning_rate": 4.866868580458257e-05,
      "loss": 1.9102,
      "step": 1015
    },
    {
      "epoch": 0.026757966815907294,
      "grad_norm": 3.0735270977020264,
      "learning_rate": 4.866736897550698e-05,
      "loss": 1.7031,
      "step": 1016
    },
    {
      "epoch": 0.026784303397419017,
      "grad_norm": 4.263140678405762,
      "learning_rate": 4.86660521464314e-05,
      "loss": 1.1439,
      "step": 1017
    },
    {
      "epoch": 0.026810639978930736,
      "grad_norm": 2.453805923461914,
      "learning_rate": 4.8664735317355806e-05,
      "loss": 2.4004,
      "step": 1018
    },
    {
      "epoch": 0.026836976560442455,
      "grad_norm": 3.2226572036743164,
      "learning_rate": 4.866341848828022e-05,
      "loss": 1.6197,
      "step": 1019
    },
    {
      "epoch": 0.026863313141954174,
      "grad_norm": 2.62640643119812,
      "learning_rate": 4.866210165920464e-05,
      "loss": 1.7825,
      "step": 1020
    },
    {
      "epoch": 0.026889649723465893,
      "grad_norm": 2.3195648193359375,
      "learning_rate": 4.866078483012905e-05,
      "loss": 1.6573,
      "step": 1021
    },
    {
      "epoch": 0.026915986304977615,
      "grad_norm": 2.2187161445617676,
      "learning_rate": 4.865946800105347e-05,
      "loss": 1.6928,
      "step": 1022
    },
    {
      "epoch": 0.026942322886489335,
      "grad_norm": 2.3962841033935547,
      "learning_rate": 4.865815117197788e-05,
      "loss": 2.4874,
      "step": 1023
    },
    {
      "epoch": 0.026968659468001054,
      "grad_norm": 3.23622465133667,
      "learning_rate": 4.865683434290229e-05,
      "loss": 0.4497,
      "step": 1024
    },
    {
      "epoch": 0.026994996049512773,
      "grad_norm": 2.3920416831970215,
      "learning_rate": 4.865551751382671e-05,
      "loss": 1.6916,
      "step": 1025
    },
    {
      "epoch": 0.02702133263102449,
      "grad_norm": 2.71406626701355,
      "learning_rate": 4.8654200684751124e-05,
      "loss": 1.749,
      "step": 1026
    },
    {
      "epoch": 0.027047669212536214,
      "grad_norm": 2.2783379554748535,
      "learning_rate": 4.865288385567553e-05,
      "loss": 1.7853,
      "step": 1027
    },
    {
      "epoch": 0.027074005794047933,
      "grad_norm": 2.941490650177002,
      "learning_rate": 4.865156702659995e-05,
      "loss": 1.2299,
      "step": 1028
    },
    {
      "epoch": 0.027100342375559652,
      "grad_norm": 3.930819272994995,
      "learning_rate": 4.8650250197524364e-05,
      "loss": 0.9998,
      "step": 1029
    },
    {
      "epoch": 0.02712667895707137,
      "grad_norm": 4.769101619720459,
      "learning_rate": 4.864893336844878e-05,
      "loss": 1.6989,
      "step": 1030
    },
    {
      "epoch": 0.02715301553858309,
      "grad_norm": 3.156647205352783,
      "learning_rate": 4.8647616539373195e-05,
      "loss": 1.8619,
      "step": 1031
    },
    {
      "epoch": 0.027179352120094813,
      "grad_norm": 2.7352776527404785,
      "learning_rate": 4.8646299710297604e-05,
      "loss": 0.7755,
      "step": 1032
    },
    {
      "epoch": 0.027205688701606532,
      "grad_norm": 6.724462032318115,
      "learning_rate": 4.864498288122202e-05,
      "loss": 0.9797,
      "step": 1033
    },
    {
      "epoch": 0.02723202528311825,
      "grad_norm": 3.0356597900390625,
      "learning_rate": 4.8643666052146435e-05,
      "loss": 1.4286,
      "step": 1034
    },
    {
      "epoch": 0.02725836186462997,
      "grad_norm": 4.173969268798828,
      "learning_rate": 4.864234922307085e-05,
      "loss": 1.1673,
      "step": 1035
    },
    {
      "epoch": 0.02728469844614169,
      "grad_norm": 2.5415663719177246,
      "learning_rate": 4.864103239399526e-05,
      "loss": 1.5362,
      "step": 1036
    },
    {
      "epoch": 0.02731103502765341,
      "grad_norm": 3.3292665481567383,
      "learning_rate": 4.8639715564919675e-05,
      "loss": 2.4023,
      "step": 1037
    },
    {
      "epoch": 0.02733737160916513,
      "grad_norm": 3.314760208129883,
      "learning_rate": 4.863839873584409e-05,
      "loss": 1.8555,
      "step": 1038
    },
    {
      "epoch": 0.02736370819067685,
      "grad_norm": 3.2847180366516113,
      "learning_rate": 4.8637081906768506e-05,
      "loss": 0.5167,
      "step": 1039
    },
    {
      "epoch": 0.02739004477218857,
      "grad_norm": 2.7714438438415527,
      "learning_rate": 4.863576507769292e-05,
      "loss": 1.9907,
      "step": 1040
    },
    {
      "epoch": 0.02741638135370029,
      "grad_norm": 2.5430490970611572,
      "learning_rate": 4.863444824861733e-05,
      "loss": 2.2006,
      "step": 1041
    },
    {
      "epoch": 0.02744271793521201,
      "grad_norm": 2.117619514465332,
      "learning_rate": 4.8633131419541746e-05,
      "loss": 1.8754,
      "step": 1042
    },
    {
      "epoch": 0.02746905451672373,
      "grad_norm": 3.6804146766662598,
      "learning_rate": 4.8631814590466155e-05,
      "loss": 1.7422,
      "step": 1043
    },
    {
      "epoch": 0.02749539109823545,
      "grad_norm": 2.070969820022583,
      "learning_rate": 4.863049776139058e-05,
      "loss": 1.903,
      "step": 1044
    },
    {
      "epoch": 0.027521727679747168,
      "grad_norm": 2.4523069858551025,
      "learning_rate": 4.8629180932314986e-05,
      "loss": 1.9201,
      "step": 1045
    },
    {
      "epoch": 0.02754806426125889,
      "grad_norm": 3.284054756164551,
      "learning_rate": 4.86278641032394e-05,
      "loss": 1.6978,
      "step": 1046
    },
    {
      "epoch": 0.02757440084277061,
      "grad_norm": 3.6890952587127686,
      "learning_rate": 4.862654727416381e-05,
      "loss": 1.5353,
      "step": 1047
    },
    {
      "epoch": 0.027600737424282328,
      "grad_norm": 5.518612861633301,
      "learning_rate": 4.862523044508823e-05,
      "loss": 2.5851,
      "step": 1048
    },
    {
      "epoch": 0.027627074005794047,
      "grad_norm": 2.9064016342163086,
      "learning_rate": 4.862391361601264e-05,
      "loss": 1.6828,
      "step": 1049
    },
    {
      "epoch": 0.027653410587305766,
      "grad_norm": 4.6123738288879395,
      "learning_rate": 4.862259678693706e-05,
      "loss": 0.8455,
      "step": 1050
    },
    {
      "epoch": 0.02767974716881749,
      "grad_norm": 9.094766616821289,
      "learning_rate": 4.862127995786147e-05,
      "loss": 2.7074,
      "step": 1051
    },
    {
      "epoch": 0.027706083750329208,
      "grad_norm": 3.067086696624756,
      "learning_rate": 4.861996312878588e-05,
      "loss": 1.8295,
      "step": 1052
    },
    {
      "epoch": 0.027732420331840927,
      "grad_norm": 3.5218935012817383,
      "learning_rate": 4.8618646299710304e-05,
      "loss": 1.7841,
      "step": 1053
    },
    {
      "epoch": 0.027758756913352646,
      "grad_norm": 2.4527480602264404,
      "learning_rate": 4.861732947063471e-05,
      "loss": 1.5477,
      "step": 1054
    },
    {
      "epoch": 0.027785093494864365,
      "grad_norm": 5.567698955535889,
      "learning_rate": 4.861601264155913e-05,
      "loss": 1.481,
      "step": 1055
    },
    {
      "epoch": 0.027811430076376088,
      "grad_norm": 2.672114849090576,
      "learning_rate": 4.861469581248354e-05,
      "loss": 1.8227,
      "step": 1056
    },
    {
      "epoch": 0.027837766657887807,
      "grad_norm": 2.226153612136841,
      "learning_rate": 4.861337898340795e-05,
      "loss": 1.7298,
      "step": 1057
    },
    {
      "epoch": 0.027864103239399526,
      "grad_norm": 2.4429683685302734,
      "learning_rate": 4.861206215433237e-05,
      "loss": 1.373,
      "step": 1058
    },
    {
      "epoch": 0.027890439820911245,
      "grad_norm": 2.6967573165893555,
      "learning_rate": 4.8610745325256784e-05,
      "loss": 1.9188,
      "step": 1059
    },
    {
      "epoch": 0.027916776402422964,
      "grad_norm": 4.453019142150879,
      "learning_rate": 4.86094284961812e-05,
      "loss": 1.6254,
      "step": 1060
    },
    {
      "epoch": 0.027943112983934686,
      "grad_norm": 4.1539626121521,
      "learning_rate": 4.860811166710561e-05,
      "loss": 2.2024,
      "step": 1061
    },
    {
      "epoch": 0.027969449565446405,
      "grad_norm": 4.406345367431641,
      "learning_rate": 4.860679483803003e-05,
      "loss": 1.5992,
      "step": 1062
    },
    {
      "epoch": 0.027995786146958124,
      "grad_norm": 2.6692306995391846,
      "learning_rate": 4.860547800895444e-05,
      "loss": 2.7249,
      "step": 1063
    },
    {
      "epoch": 0.028022122728469843,
      "grad_norm": 5.10105037689209,
      "learning_rate": 4.8604161179878855e-05,
      "loss": 1.7338,
      "step": 1064
    },
    {
      "epoch": 0.028048459309981566,
      "grad_norm": 3.1361806392669678,
      "learning_rate": 4.8602844350803264e-05,
      "loss": 1.9357,
      "step": 1065
    },
    {
      "epoch": 0.028074795891493285,
      "grad_norm": 2.493241310119629,
      "learning_rate": 4.860152752172768e-05,
      "loss": 2.2922,
      "step": 1066
    },
    {
      "epoch": 0.028101132473005004,
      "grad_norm": 3.7938878536224365,
      "learning_rate": 4.8600210692652095e-05,
      "loss": 1.9997,
      "step": 1067
    },
    {
      "epoch": 0.028127469054516723,
      "grad_norm": 2.4619712829589844,
      "learning_rate": 4.859889386357651e-05,
      "loss": 2.0023,
      "step": 1068
    },
    {
      "epoch": 0.028153805636028442,
      "grad_norm": 3.2266292572021484,
      "learning_rate": 4.859757703450093e-05,
      "loss": 1.4799,
      "step": 1069
    },
    {
      "epoch": 0.028180142217540165,
      "grad_norm": 2.6764299869537354,
      "learning_rate": 4.8596260205425335e-05,
      "loss": 2.2143,
      "step": 1070
    },
    {
      "epoch": 0.028206478799051884,
      "grad_norm": 2.3708853721618652,
      "learning_rate": 4.859494337634975e-05,
      "loss": 2.6099,
      "step": 1071
    },
    {
      "epoch": 0.028232815380563603,
      "grad_norm": 4.766274929046631,
      "learning_rate": 4.859362654727417e-05,
      "loss": 1.9012,
      "step": 1072
    },
    {
      "epoch": 0.028259151962075322,
      "grad_norm": 4.662174701690674,
      "learning_rate": 4.859230971819858e-05,
      "loss": 1.8046,
      "step": 1073
    },
    {
      "epoch": 0.02828548854358704,
      "grad_norm": 5.352644443511963,
      "learning_rate": 4.859099288912299e-05,
      "loss": 1.6289,
      "step": 1074
    },
    {
      "epoch": 0.028311825125098763,
      "grad_norm": 3.8625576496124268,
      "learning_rate": 4.858967606004741e-05,
      "loss": 2.3011,
      "step": 1075
    },
    {
      "epoch": 0.028338161706610483,
      "grad_norm": 2.4151413440704346,
      "learning_rate": 4.858835923097182e-05,
      "loss": 1.2742,
      "step": 1076
    },
    {
      "epoch": 0.0283644982881222,
      "grad_norm": 2.240936756134033,
      "learning_rate": 4.858704240189624e-05,
      "loss": 2.1133,
      "step": 1077
    },
    {
      "epoch": 0.02839083486963392,
      "grad_norm": 4.756962299346924,
      "learning_rate": 4.8585725572820654e-05,
      "loss": 2.4781,
      "step": 1078
    },
    {
      "epoch": 0.02841717145114564,
      "grad_norm": 2.3123276233673096,
      "learning_rate": 4.858440874374506e-05,
      "loss": 0.6893,
      "step": 1079
    },
    {
      "epoch": 0.028443508032657362,
      "grad_norm": 2.4670345783233643,
      "learning_rate": 4.858309191466948e-05,
      "loss": 1.8863,
      "step": 1080
    },
    {
      "epoch": 0.02846984461416908,
      "grad_norm": 4.072238445281982,
      "learning_rate": 4.8581775085593894e-05,
      "loss": 1.6139,
      "step": 1081
    },
    {
      "epoch": 0.0284961811956808,
      "grad_norm": 2.351463556289673,
      "learning_rate": 4.858045825651831e-05,
      "loss": 1.8861,
      "step": 1082
    },
    {
      "epoch": 0.02852251777719252,
      "grad_norm": 3.012080430984497,
      "learning_rate": 4.857914142744272e-05,
      "loss": 2.1996,
      "step": 1083
    },
    {
      "epoch": 0.028548854358704242,
      "grad_norm": 2.5738027095794678,
      "learning_rate": 4.8577824598367134e-05,
      "loss": 1.7826,
      "step": 1084
    },
    {
      "epoch": 0.02857519094021596,
      "grad_norm": 5.832880973815918,
      "learning_rate": 4.857650776929155e-05,
      "loss": 1.1905,
      "step": 1085
    },
    {
      "epoch": 0.02860152752172768,
      "grad_norm": 2.4658961296081543,
      "learning_rate": 4.8575190940215965e-05,
      "loss": 1.8326,
      "step": 1086
    },
    {
      "epoch": 0.0286278641032394,
      "grad_norm": 2.3752379417419434,
      "learning_rate": 4.857387411114038e-05,
      "loss": 2.2214,
      "step": 1087
    },
    {
      "epoch": 0.028654200684751118,
      "grad_norm": 2.201660633087158,
      "learning_rate": 4.857255728206479e-05,
      "loss": 2.3061,
      "step": 1088
    },
    {
      "epoch": 0.02868053726626284,
      "grad_norm": 4.142340660095215,
      "learning_rate": 4.8571240452989205e-05,
      "loss": 1.2818,
      "step": 1089
    },
    {
      "epoch": 0.02870687384777456,
      "grad_norm": 2.3261709213256836,
      "learning_rate": 4.8569923623913614e-05,
      "loss": 1.4621,
      "step": 1090
    },
    {
      "epoch": 0.02873321042928628,
      "grad_norm": 3.5733439922332764,
      "learning_rate": 4.8568606794838036e-05,
      "loss": 1.6464,
      "step": 1091
    },
    {
      "epoch": 0.028759547010797998,
      "grad_norm": 3.020409345626831,
      "learning_rate": 4.8567289965762445e-05,
      "loss": 2.7091,
      "step": 1092
    },
    {
      "epoch": 0.028785883592309717,
      "grad_norm": 2.2784829139709473,
      "learning_rate": 4.856597313668686e-05,
      "loss": 1.9396,
      "step": 1093
    },
    {
      "epoch": 0.02881222017382144,
      "grad_norm": 2.57547664642334,
      "learning_rate": 4.8564656307611276e-05,
      "loss": 1.3221,
      "step": 1094
    },
    {
      "epoch": 0.02883855675533316,
      "grad_norm": 5.381950378417969,
      "learning_rate": 4.856333947853569e-05,
      "loss": 1.6726,
      "step": 1095
    },
    {
      "epoch": 0.028864893336844877,
      "grad_norm": 1.9736328125,
      "learning_rate": 4.856202264946011e-05,
      "loss": 1.8708,
      "step": 1096
    },
    {
      "epoch": 0.028891229918356597,
      "grad_norm": 4.961425304412842,
      "learning_rate": 4.8560705820384516e-05,
      "loss": 2.0083,
      "step": 1097
    },
    {
      "epoch": 0.028917566499868316,
      "grad_norm": 3.0586788654327393,
      "learning_rate": 4.855938899130893e-05,
      "loss": 1.7768,
      "step": 1098
    },
    {
      "epoch": 0.028943903081380038,
      "grad_norm": 3.3477911949157715,
      "learning_rate": 4.855807216223334e-05,
      "loss": 1.9127,
      "step": 1099
    },
    {
      "epoch": 0.028970239662891757,
      "grad_norm": 4.469985485076904,
      "learning_rate": 4.855675533315776e-05,
      "loss": 1.5644,
      "step": 1100
    },
    {
      "epoch": 0.028996576244403476,
      "grad_norm": 3.75849986076355,
      "learning_rate": 4.855543850408217e-05,
      "loss": 1.6602,
      "step": 1101
    },
    {
      "epoch": 0.029022912825915195,
      "grad_norm": 2.754058361053467,
      "learning_rate": 4.855412167500659e-05,
      "loss": 1.6427,
      "step": 1102
    },
    {
      "epoch": 0.029049249407426914,
      "grad_norm": 2.0229575634002686,
      "learning_rate": 4.8552804845930996e-05,
      "loss": 1.745,
      "step": 1103
    },
    {
      "epoch": 0.029075585988938637,
      "grad_norm": 3.46441912651062,
      "learning_rate": 4.855148801685541e-05,
      "loss": 2.6647,
      "step": 1104
    },
    {
      "epoch": 0.029101922570450356,
      "grad_norm": 7.25486421585083,
      "learning_rate": 4.855017118777983e-05,
      "loss": 1.9896,
      "step": 1105
    },
    {
      "epoch": 0.029128259151962075,
      "grad_norm": 3.0288054943084717,
      "learning_rate": 4.854885435870424e-05,
      "loss": 2.3027,
      "step": 1106
    },
    {
      "epoch": 0.029154595733473794,
      "grad_norm": 4.595308303833008,
      "learning_rate": 4.854753752962866e-05,
      "loss": 1.6661,
      "step": 1107
    },
    {
      "epoch": 0.029180932314985517,
      "grad_norm": 2.545994997024536,
      "learning_rate": 4.854622070055307e-05,
      "loss": 0.7444,
      "step": 1108
    },
    {
      "epoch": 0.029207268896497236,
      "grad_norm": 2.2626681327819824,
      "learning_rate": 4.854490387147749e-05,
      "loss": 2.0572,
      "step": 1109
    },
    {
      "epoch": 0.029233605478008955,
      "grad_norm": 2.315185070037842,
      "learning_rate": 4.85435870424019e-05,
      "loss": 1.3084,
      "step": 1110
    },
    {
      "epoch": 0.029259942059520674,
      "grad_norm": 3.43516206741333,
      "learning_rate": 4.8542270213326314e-05,
      "loss": 0.4376,
      "step": 1111
    },
    {
      "epoch": 0.029286278641032393,
      "grad_norm": 2.599492073059082,
      "learning_rate": 4.854095338425072e-05,
      "loss": 1.4326,
      "step": 1112
    },
    {
      "epoch": 0.029312615222544115,
      "grad_norm": 2.722194194793701,
      "learning_rate": 4.853963655517514e-05,
      "loss": 1.9376,
      "step": 1113
    },
    {
      "epoch": 0.029338951804055834,
      "grad_norm": 3.498096466064453,
      "learning_rate": 4.8538319726099554e-05,
      "loss": 1.2945,
      "step": 1114
    },
    {
      "epoch": 0.029365288385567553,
      "grad_norm": 2.3955607414245605,
      "learning_rate": 4.853700289702397e-05,
      "loss": 1.42,
      "step": 1115
    },
    {
      "epoch": 0.029391624967079272,
      "grad_norm": 3.262183666229248,
      "learning_rate": 4.8535686067948385e-05,
      "loss": 1.5349,
      "step": 1116
    },
    {
      "epoch": 0.02941796154859099,
      "grad_norm": 4.325577259063721,
      "learning_rate": 4.8534369238872794e-05,
      "loss": 2.2147,
      "step": 1117
    },
    {
      "epoch": 0.029444298130102714,
      "grad_norm": 3.8473992347717285,
      "learning_rate": 4.853305240979721e-05,
      "loss": 2.2715,
      "step": 1118
    },
    {
      "epoch": 0.029470634711614433,
      "grad_norm": 3.3051633834838867,
      "learning_rate": 4.8531735580721625e-05,
      "loss": 1.3309,
      "step": 1119
    },
    {
      "epoch": 0.029496971293126152,
      "grad_norm": 3.3608505725860596,
      "learning_rate": 4.853041875164604e-05,
      "loss": 0.877,
      "step": 1120
    },
    {
      "epoch": 0.02952330787463787,
      "grad_norm": 5.844008922576904,
      "learning_rate": 4.852910192257045e-05,
      "loss": 1.8371,
      "step": 1121
    },
    {
      "epoch": 0.02954964445614959,
      "grad_norm": 10.57100772857666,
      "learning_rate": 4.8527785093494865e-05,
      "loss": 2.5075,
      "step": 1122
    },
    {
      "epoch": 0.029575981037661313,
      "grad_norm": 5.153822898864746,
      "learning_rate": 4.852646826441928e-05,
      "loss": 1.6994,
      "step": 1123
    },
    {
      "epoch": 0.029602317619173032,
      "grad_norm": 3.8108110427856445,
      "learning_rate": 4.8525151435343696e-05,
      "loss": 1.8838,
      "step": 1124
    },
    {
      "epoch": 0.02962865420068475,
      "grad_norm": 3.1336042881011963,
      "learning_rate": 4.852383460626811e-05,
      "loss": 1.9083,
      "step": 1125
    },
    {
      "epoch": 0.02965499078219647,
      "grad_norm": 3.490872383117676,
      "learning_rate": 4.852251777719252e-05,
      "loss": 1.912,
      "step": 1126
    },
    {
      "epoch": 0.02968132736370819,
      "grad_norm": 2.1635420322418213,
      "learning_rate": 4.8521200948116936e-05,
      "loss": 0.6838,
      "step": 1127
    },
    {
      "epoch": 0.02970766394521991,
      "grad_norm": 2.9938266277313232,
      "learning_rate": 4.851988411904135e-05,
      "loss": 2.2292,
      "step": 1128
    },
    {
      "epoch": 0.02973400052673163,
      "grad_norm": 4.7393670082092285,
      "learning_rate": 4.851856728996577e-05,
      "loss": 1.3348,
      "step": 1129
    },
    {
      "epoch": 0.02976033710824335,
      "grad_norm": 2.9201533794403076,
      "learning_rate": 4.8517250460890176e-05,
      "loss": 1.4847,
      "step": 1130
    },
    {
      "epoch": 0.02978667368975507,
      "grad_norm": 5.068320274353027,
      "learning_rate": 4.851593363181459e-05,
      "loss": 1.5796,
      "step": 1131
    },
    {
      "epoch": 0.02981301027126679,
      "grad_norm": 3.0481390953063965,
      "learning_rate": 4.851461680273901e-05,
      "loss": 0.413,
      "step": 1132
    },
    {
      "epoch": 0.02983934685277851,
      "grad_norm": 3.275697708129883,
      "learning_rate": 4.851329997366342e-05,
      "loss": 1.8815,
      "step": 1133
    },
    {
      "epoch": 0.02986568343429023,
      "grad_norm": 2.432833671569824,
      "learning_rate": 4.851198314458784e-05,
      "loss": 2.0992,
      "step": 1134
    },
    {
      "epoch": 0.02989202001580195,
      "grad_norm": 2.92842960357666,
      "learning_rate": 4.851066631551225e-05,
      "loss": 0.7407,
      "step": 1135
    },
    {
      "epoch": 0.029918356597313667,
      "grad_norm": 2.2532742023468018,
      "learning_rate": 4.850934948643666e-05,
      "loss": 2.0612,
      "step": 1136
    },
    {
      "epoch": 0.02994469317882539,
      "grad_norm": 2.632624387741089,
      "learning_rate": 4.850803265736107e-05,
      "loss": 1.5505,
      "step": 1137
    },
    {
      "epoch": 0.02997102976033711,
      "grad_norm": 4.293117523193359,
      "learning_rate": 4.8506715828285494e-05,
      "loss": 1.9591,
      "step": 1138
    },
    {
      "epoch": 0.029997366341848828,
      "grad_norm": 9.684476852416992,
      "learning_rate": 4.85053989992099e-05,
      "loss": 0.8484,
      "step": 1139
    },
    {
      "epoch": 0.030023702923360547,
      "grad_norm": 2.6307811737060547,
      "learning_rate": 4.850408217013432e-05,
      "loss": 1.5203,
      "step": 1140
    },
    {
      "epoch": 0.030050039504872266,
      "grad_norm": 2.824821710586548,
      "learning_rate": 4.8502765341058735e-05,
      "loss": 1.9105,
      "step": 1141
    },
    {
      "epoch": 0.03007637608638399,
      "grad_norm": 2.490757942199707,
      "learning_rate": 4.850144851198315e-05,
      "loss": 2.4798,
      "step": 1142
    },
    {
      "epoch": 0.030102712667895708,
      "grad_norm": 2.523568630218506,
      "learning_rate": 4.8500131682907566e-05,
      "loss": 1.7346,
      "step": 1143
    },
    {
      "epoch": 0.030129049249407427,
      "grad_norm": 2.3697028160095215,
      "learning_rate": 4.8498814853831975e-05,
      "loss": 1.7498,
      "step": 1144
    },
    {
      "epoch": 0.030155385830919146,
      "grad_norm": 2.4945552349090576,
      "learning_rate": 4.849749802475639e-05,
      "loss": 1.8484,
      "step": 1145
    },
    {
      "epoch": 0.030181722412430865,
      "grad_norm": 5.43267297744751,
      "learning_rate": 4.84961811956808e-05,
      "loss": 2.0347,
      "step": 1146
    },
    {
      "epoch": 0.030208058993942587,
      "grad_norm": 3.8459794521331787,
      "learning_rate": 4.849486436660522e-05,
      "loss": 1.9875,
      "step": 1147
    },
    {
      "epoch": 0.030234395575454306,
      "grad_norm": 3.389763593673706,
      "learning_rate": 4.849354753752963e-05,
      "loss": 1.36,
      "step": 1148
    },
    {
      "epoch": 0.030260732156966026,
      "grad_norm": 3.047062873840332,
      "learning_rate": 4.8492230708454046e-05,
      "loss": 2.0154,
      "step": 1149
    },
    {
      "epoch": 0.030287068738477745,
      "grad_norm": 1.9687374830245972,
      "learning_rate": 4.8490913879378455e-05,
      "loss": 2.14,
      "step": 1150
    },
    {
      "epoch": 0.030313405319989467,
      "grad_norm": 5.142848014831543,
      "learning_rate": 4.848959705030287e-05,
      "loss": 1.3017,
      "step": 1151
    },
    {
      "epoch": 0.030339741901501186,
      "grad_norm": 5.077940940856934,
      "learning_rate": 4.8488280221227286e-05,
      "loss": 1.7044,
      "step": 1152
    },
    {
      "epoch": 0.030366078483012905,
      "grad_norm": 2.410503387451172,
      "learning_rate": 4.84869633921517e-05,
      "loss": 1.8168,
      "step": 1153
    },
    {
      "epoch": 0.030392415064524624,
      "grad_norm": 2.8993778228759766,
      "learning_rate": 4.848564656307612e-05,
      "loss": 2.3745,
      "step": 1154
    },
    {
      "epoch": 0.030418751646036343,
      "grad_norm": 2.937490224838257,
      "learning_rate": 4.8484329734000526e-05,
      "loss": 1.3934,
      "step": 1155
    },
    {
      "epoch": 0.030445088227548066,
      "grad_norm": 3.265104055404663,
      "learning_rate": 4.848301290492494e-05,
      "loss": 1.978,
      "step": 1156
    },
    {
      "epoch": 0.030471424809059785,
      "grad_norm": 2.479564905166626,
      "learning_rate": 4.848169607584936e-05,
      "loss": 0.8255,
      "step": 1157
    },
    {
      "epoch": 0.030497761390571504,
      "grad_norm": 2.08847975730896,
      "learning_rate": 4.848037924677377e-05,
      "loss": 1.8818,
      "step": 1158
    },
    {
      "epoch": 0.030524097972083223,
      "grad_norm": 4.553527355194092,
      "learning_rate": 4.847906241769818e-05,
      "loss": 0.8432,
      "step": 1159
    },
    {
      "epoch": 0.030550434553594942,
      "grad_norm": 2.920544385910034,
      "learning_rate": 4.84777455886226e-05,
      "loss": 1.516,
      "step": 1160
    },
    {
      "epoch": 0.030576771135106665,
      "grad_norm": 2.890082359313965,
      "learning_rate": 4.847642875954701e-05,
      "loss": 1.5008,
      "step": 1161
    },
    {
      "epoch": 0.030603107716618384,
      "grad_norm": 6.989933013916016,
      "learning_rate": 4.847511193047143e-05,
      "loss": 0.8749,
      "step": 1162
    },
    {
      "epoch": 0.030629444298130103,
      "grad_norm": 4.9764509201049805,
      "learning_rate": 4.8473795101395844e-05,
      "loss": 1.455,
      "step": 1163
    },
    {
      "epoch": 0.03065578087964182,
      "grad_norm": 4.515769004821777,
      "learning_rate": 4.847247827232025e-05,
      "loss": 1.1419,
      "step": 1164
    },
    {
      "epoch": 0.03068211746115354,
      "grad_norm": 2.244635820388794,
      "learning_rate": 4.847116144324467e-05,
      "loss": 2.5669,
      "step": 1165
    },
    {
      "epoch": 0.030708454042665263,
      "grad_norm": 5.076201915740967,
      "learning_rate": 4.8469844614169084e-05,
      "loss": 1.9085,
      "step": 1166
    },
    {
      "epoch": 0.030734790624176982,
      "grad_norm": 2.9700934886932373,
      "learning_rate": 4.84685277850935e-05,
      "loss": 1.81,
      "step": 1167
    },
    {
      "epoch": 0.0307611272056887,
      "grad_norm": 5.795651435852051,
      "learning_rate": 4.846721095601791e-05,
      "loss": 1.3432,
      "step": 1168
    },
    {
      "epoch": 0.03078746378720042,
      "grad_norm": 2.369987726211548,
      "learning_rate": 4.8465894126942324e-05,
      "loss": 1.9682,
      "step": 1169
    },
    {
      "epoch": 0.03081380036871214,
      "grad_norm": 3.435899496078491,
      "learning_rate": 4.846457729786674e-05,
      "loss": 2.5022,
      "step": 1170
    },
    {
      "epoch": 0.030840136950223862,
      "grad_norm": 3.1760265827178955,
      "learning_rate": 4.8463260468791155e-05,
      "loss": 2.0571,
      "step": 1171
    },
    {
      "epoch": 0.03086647353173558,
      "grad_norm": 2.5435879230499268,
      "learning_rate": 4.846194363971557e-05,
      "loss": 1.8374,
      "step": 1172
    },
    {
      "epoch": 0.0308928101132473,
      "grad_norm": 2.7065792083740234,
      "learning_rate": 4.846062681063998e-05,
      "loss": 2.1015,
      "step": 1173
    },
    {
      "epoch": 0.03091914669475902,
      "grad_norm": 1.9133766889572144,
      "learning_rate": 4.8459309981564395e-05,
      "loss": 1.6722,
      "step": 1174
    },
    {
      "epoch": 0.03094548327627074,
      "grad_norm": 2.596497058868408,
      "learning_rate": 4.845799315248881e-05,
      "loss": 1.9452,
      "step": 1175
    },
    {
      "epoch": 0.03097181985778246,
      "grad_norm": 5.125361442565918,
      "learning_rate": 4.8456676323413226e-05,
      "loss": 1.1212,
      "step": 1176
    },
    {
      "epoch": 0.03099815643929418,
      "grad_norm": 2.5910115242004395,
      "learning_rate": 4.8455359494337635e-05,
      "loss": 1.9922,
      "step": 1177
    },
    {
      "epoch": 0.0310244930208059,
      "grad_norm": 4.699963092803955,
      "learning_rate": 4.845404266526205e-05,
      "loss": 1.7486,
      "step": 1178
    },
    {
      "epoch": 0.031050829602317618,
      "grad_norm": 3.3821754455566406,
      "learning_rate": 4.8452725836186466e-05,
      "loss": 1.8035,
      "step": 1179
    },
    {
      "epoch": 0.03107716618382934,
      "grad_norm": 2.596687078475952,
      "learning_rate": 4.845140900711088e-05,
      "loss": 1.3562,
      "step": 1180
    },
    {
      "epoch": 0.03110350276534106,
      "grad_norm": 3.078029155731201,
      "learning_rate": 4.84500921780353e-05,
      "loss": 1.5201,
      "step": 1181
    },
    {
      "epoch": 0.03112983934685278,
      "grad_norm": 5.427302360534668,
      "learning_rate": 4.8448775348959706e-05,
      "loss": 2.043,
      "step": 1182
    },
    {
      "epoch": 0.031156175928364498,
      "grad_norm": 5.404221534729004,
      "learning_rate": 4.844745851988412e-05,
      "loss": 2.0277,
      "step": 1183
    },
    {
      "epoch": 0.031182512509876217,
      "grad_norm": 2.2742812633514404,
      "learning_rate": 4.844614169080853e-05,
      "loss": 2.0188,
      "step": 1184
    },
    {
      "epoch": 0.03120884909138794,
      "grad_norm": 3.477735996246338,
      "learning_rate": 4.844482486173295e-05,
      "loss": 0.3864,
      "step": 1185
    },
    {
      "epoch": 0.031235185672899658,
      "grad_norm": 2.8333590030670166,
      "learning_rate": 4.844350803265736e-05,
      "loss": 1.3481,
      "step": 1186
    },
    {
      "epoch": 0.03126152225441138,
      "grad_norm": 3.7202811241149902,
      "learning_rate": 4.844219120358178e-05,
      "loss": 1.4413,
      "step": 1187
    },
    {
      "epoch": 0.031287858835923096,
      "grad_norm": 3.836300849914551,
      "learning_rate": 4.844087437450619e-05,
      "loss": 2.4155,
      "step": 1188
    },
    {
      "epoch": 0.03131419541743482,
      "grad_norm": 2.1705095767974854,
      "learning_rate": 4.84395575454306e-05,
      "loss": 1.5203,
      "step": 1189
    },
    {
      "epoch": 0.031340531998946534,
      "grad_norm": 9.435272216796875,
      "learning_rate": 4.8438240716355024e-05,
      "loss": 2.1017,
      "step": 1190
    },
    {
      "epoch": 0.03136686858045826,
      "grad_norm": 2.069197177886963,
      "learning_rate": 4.843692388727943e-05,
      "loss": 1.6487,
      "step": 1191
    },
    {
      "epoch": 0.03139320516196998,
      "grad_norm": 4.131947040557861,
      "learning_rate": 4.843560705820385e-05,
      "loss": 1.4584,
      "step": 1192
    },
    {
      "epoch": 0.031419541743481695,
      "grad_norm": 2.280083417892456,
      "learning_rate": 4.843429022912826e-05,
      "loss": 1.2831,
      "step": 1193
    },
    {
      "epoch": 0.03144587832499342,
      "grad_norm": 2.114161729812622,
      "learning_rate": 4.843297340005268e-05,
      "loss": 1.3618,
      "step": 1194
    },
    {
      "epoch": 0.03147221490650513,
      "grad_norm": 2.3315417766571045,
      "learning_rate": 4.843165657097709e-05,
      "loss": 1.4538,
      "step": 1195
    },
    {
      "epoch": 0.031498551488016856,
      "grad_norm": 2.4717631340026855,
      "learning_rate": 4.8430339741901504e-05,
      "loss": 1.127,
      "step": 1196
    },
    {
      "epoch": 0.03152488806952858,
      "grad_norm": 3.3255231380462646,
      "learning_rate": 4.842902291282592e-05,
      "loss": 0.7941,
      "step": 1197
    },
    {
      "epoch": 0.031551224651040294,
      "grad_norm": 2.67148494720459,
      "learning_rate": 4.842770608375033e-05,
      "loss": 2.1661,
      "step": 1198
    },
    {
      "epoch": 0.031577561232552016,
      "grad_norm": 4.5087714195251465,
      "learning_rate": 4.842638925467475e-05,
      "loss": 2.2192,
      "step": 1199
    },
    {
      "epoch": 0.03160389781406373,
      "grad_norm": 2.4082632064819336,
      "learning_rate": 4.842507242559916e-05,
      "loss": 2.1281,
      "step": 1200
    },
    {
      "epoch": 0.031630234395575454,
      "grad_norm": 2.3811023235321045,
      "learning_rate": 4.8423755596523576e-05,
      "loss": 1.706,
      "step": 1201
    },
    {
      "epoch": 0.03165657097708718,
      "grad_norm": 4.598793029785156,
      "learning_rate": 4.8422438767447984e-05,
      "loss": 1.8155,
      "step": 1202
    },
    {
      "epoch": 0.03168290755859889,
      "grad_norm": 2.2454915046691895,
      "learning_rate": 4.84211219383724e-05,
      "loss": 1.6846,
      "step": 1203
    },
    {
      "epoch": 0.031709244140110615,
      "grad_norm": 6.481629371643066,
      "learning_rate": 4.8419805109296816e-05,
      "loss": 2.5402,
      "step": 1204
    },
    {
      "epoch": 0.03173558072162233,
      "grad_norm": 3.875046730041504,
      "learning_rate": 4.841848828022123e-05,
      "loss": 1.0837,
      "step": 1205
    },
    {
      "epoch": 0.03176191730313405,
      "grad_norm": 2.3516602516174316,
      "learning_rate": 4.841717145114564e-05,
      "loss": 1.891,
      "step": 1206
    },
    {
      "epoch": 0.031788253884645776,
      "grad_norm": 2.502894639968872,
      "learning_rate": 4.8415854622070056e-05,
      "loss": 1.9226,
      "step": 1207
    },
    {
      "epoch": 0.03181459046615749,
      "grad_norm": 2.2583303451538086,
      "learning_rate": 4.841453779299447e-05,
      "loss": 1.9457,
      "step": 1208
    },
    {
      "epoch": 0.031840927047669214,
      "grad_norm": 2.4761016368865967,
      "learning_rate": 4.841322096391889e-05,
      "loss": 1.5265,
      "step": 1209
    },
    {
      "epoch": 0.03186726362918093,
      "grad_norm": 2.4672107696533203,
      "learning_rate": 4.84119041348433e-05,
      "loss": 2.1244,
      "step": 1210
    },
    {
      "epoch": 0.03189360021069265,
      "grad_norm": 3.0499157905578613,
      "learning_rate": 4.841058730576771e-05,
      "loss": 0.3298,
      "step": 1211
    },
    {
      "epoch": 0.031919936792204374,
      "grad_norm": 2.463407516479492,
      "learning_rate": 4.840927047669213e-05,
      "loss": 1.7443,
      "step": 1212
    },
    {
      "epoch": 0.03194627337371609,
      "grad_norm": 5.298512935638428,
      "learning_rate": 4.840795364761654e-05,
      "loss": 1.075,
      "step": 1213
    },
    {
      "epoch": 0.03197260995522781,
      "grad_norm": 2.7082419395446777,
      "learning_rate": 4.840663681854096e-05,
      "loss": 1.8387,
      "step": 1214
    },
    {
      "epoch": 0.03199894653673953,
      "grad_norm": 3.170705556869507,
      "learning_rate": 4.840531998946537e-05,
      "loss": 0.6658,
      "step": 1215
    },
    {
      "epoch": 0.03202528311825125,
      "grad_norm": 3.09491229057312,
      "learning_rate": 4.840400316038978e-05,
      "loss": 1.971,
      "step": 1216
    },
    {
      "epoch": 0.03205161969976297,
      "grad_norm": 3.932175874710083,
      "learning_rate": 4.84026863313142e-05,
      "loss": 1.4447,
      "step": 1217
    },
    {
      "epoch": 0.03207795628127469,
      "grad_norm": 2.247992515563965,
      "learning_rate": 4.8401369502238614e-05,
      "loss": 1.7858,
      "step": 1218
    },
    {
      "epoch": 0.03210429286278641,
      "grad_norm": 2.677718162536621,
      "learning_rate": 4.840005267316303e-05,
      "loss": 1.7789,
      "step": 1219
    },
    {
      "epoch": 0.03213062944429813,
      "grad_norm": 2.337888240814209,
      "learning_rate": 4.839873584408744e-05,
      "loss": 1.9844,
      "step": 1220
    },
    {
      "epoch": 0.03215696602580985,
      "grad_norm": 2.4834091663360596,
      "learning_rate": 4.8397419015011854e-05,
      "loss": 1.7121,
      "step": 1221
    },
    {
      "epoch": 0.03218330260732157,
      "grad_norm": 2.4524142742156982,
      "learning_rate": 4.839610218593626e-05,
      "loss": 2.3675,
      "step": 1222
    },
    {
      "epoch": 0.03220963918883329,
      "grad_norm": 2.556180953979492,
      "learning_rate": 4.8394785356860685e-05,
      "loss": 1.9163,
      "step": 1223
    },
    {
      "epoch": 0.03223597577034501,
      "grad_norm": 5.075287342071533,
      "learning_rate": 4.8393468527785094e-05,
      "loss": 1.3693,
      "step": 1224
    },
    {
      "epoch": 0.032262312351856726,
      "grad_norm": 2.355077028274536,
      "learning_rate": 4.839215169870951e-05,
      "loss": 1.6558,
      "step": 1225
    },
    {
      "epoch": 0.03228864893336845,
      "grad_norm": 2.9545958042144775,
      "learning_rate": 4.8390834869633925e-05,
      "loss": 0.5404,
      "step": 1226
    },
    {
      "epoch": 0.03231498551488017,
      "grad_norm": 2.1071887016296387,
      "learning_rate": 4.838951804055834e-05,
      "loss": 1.8877,
      "step": 1227
    },
    {
      "epoch": 0.032341322096391886,
      "grad_norm": 6.698026180267334,
      "learning_rate": 4.8388201211482756e-05,
      "loss": 1.1915,
      "step": 1228
    },
    {
      "epoch": 0.03236765867790361,
      "grad_norm": 2.412713050842285,
      "learning_rate": 4.8386884382407165e-05,
      "loss": 1.2396,
      "step": 1229
    },
    {
      "epoch": 0.03239399525941533,
      "grad_norm": 2.383208990097046,
      "learning_rate": 4.838556755333158e-05,
      "loss": 1.9404,
      "step": 1230
    },
    {
      "epoch": 0.03242033184092705,
      "grad_norm": 2.873124122619629,
      "learning_rate": 4.838425072425599e-05,
      "loss": 2.3502,
      "step": 1231
    },
    {
      "epoch": 0.03244666842243877,
      "grad_norm": 4.161839962005615,
      "learning_rate": 4.838293389518041e-05,
      "loss": 1.9853,
      "step": 1232
    },
    {
      "epoch": 0.032473005003950485,
      "grad_norm": 2.328953742980957,
      "learning_rate": 4.838161706610482e-05,
      "loss": 2.03,
      "step": 1233
    },
    {
      "epoch": 0.03249934158546221,
      "grad_norm": 3.1227409839630127,
      "learning_rate": 4.8380300237029236e-05,
      "loss": 0.5901,
      "step": 1234
    },
    {
      "epoch": 0.03252567816697393,
      "grad_norm": 2.4840292930603027,
      "learning_rate": 4.837898340795365e-05,
      "loss": 2.0666,
      "step": 1235
    },
    {
      "epoch": 0.032552014748485646,
      "grad_norm": 2.3030710220336914,
      "learning_rate": 4.837766657887806e-05,
      "loss": 1.6181,
      "step": 1236
    },
    {
      "epoch": 0.03257835132999737,
      "grad_norm": 4.652442455291748,
      "learning_rate": 4.837634974980248e-05,
      "loss": 1.3754,
      "step": 1237
    },
    {
      "epoch": 0.032604687911509084,
      "grad_norm": 2.7813150882720947,
      "learning_rate": 4.837503292072689e-05,
      "loss": 2.5307,
      "step": 1238
    },
    {
      "epoch": 0.032631024493020806,
      "grad_norm": 2.065514326095581,
      "learning_rate": 4.837371609165131e-05,
      "loss": 1.521,
      "step": 1239
    },
    {
      "epoch": 0.03265736107453253,
      "grad_norm": 4.542300224304199,
      "learning_rate": 4.8372399262575716e-05,
      "loss": 1.3511,
      "step": 1240
    },
    {
      "epoch": 0.032683697656044244,
      "grad_norm": 3.3543527126312256,
      "learning_rate": 4.837108243350014e-05,
      "loss": 2.1861,
      "step": 1241
    },
    {
      "epoch": 0.03271003423755597,
      "grad_norm": 3.2510035037994385,
      "learning_rate": 4.836976560442455e-05,
      "loss": 1.738,
      "step": 1242
    },
    {
      "epoch": 0.03273637081906768,
      "grad_norm": 3.5997731685638428,
      "learning_rate": 4.836844877534896e-05,
      "loss": 1.1752,
      "step": 1243
    },
    {
      "epoch": 0.032762707400579405,
      "grad_norm": 2.349310874938965,
      "learning_rate": 4.836713194627338e-05,
      "loss": 1.907,
      "step": 1244
    },
    {
      "epoch": 0.03278904398209113,
      "grad_norm": 3.2032058238983154,
      "learning_rate": 4.836581511719779e-05,
      "loss": 1.8664,
      "step": 1245
    },
    {
      "epoch": 0.03281538056360284,
      "grad_norm": 4.305539608001709,
      "learning_rate": 4.836449828812221e-05,
      "loss": 0.6863,
      "step": 1246
    },
    {
      "epoch": 0.032841717145114566,
      "grad_norm": 6.19760799407959,
      "learning_rate": 4.836318145904662e-05,
      "loss": 0.8177,
      "step": 1247
    },
    {
      "epoch": 0.03286805372662628,
      "grad_norm": 3.9096808433532715,
      "learning_rate": 4.8361864629971034e-05,
      "loss": 2.1843,
      "step": 1248
    },
    {
      "epoch": 0.032894390308138004,
      "grad_norm": 3.2951722145080566,
      "learning_rate": 4.836054780089544e-05,
      "loss": 1.9124,
      "step": 1249
    },
    {
      "epoch": 0.032920726889649726,
      "grad_norm": 3.348398208618164,
      "learning_rate": 4.835923097181986e-05,
      "loss": 1.6557,
      "step": 1250
    },
    {
      "epoch": 0.03294706347116144,
      "grad_norm": 2.1526827812194824,
      "learning_rate": 4.8357914142744274e-05,
      "loss": 1.8642,
      "step": 1251
    },
    {
      "epoch": 0.032973400052673164,
      "grad_norm": 6.212658882141113,
      "learning_rate": 4.835659731366869e-05,
      "loss": 1.4845,
      "step": 1252
    },
    {
      "epoch": 0.03299973663418488,
      "grad_norm": 2.499819278717041,
      "learning_rate": 4.83552804845931e-05,
      "loss": 2.1482,
      "step": 1253
    },
    {
      "epoch": 0.0330260732156966,
      "grad_norm": 2.5231502056121826,
      "learning_rate": 4.8353963655517514e-05,
      "loss": 1.3152,
      "step": 1254
    },
    {
      "epoch": 0.033052409797208325,
      "grad_norm": 4.560873985290527,
      "learning_rate": 4.835264682644193e-05,
      "loss": 1.0126,
      "step": 1255
    },
    {
      "epoch": 0.03307874637872004,
      "grad_norm": 6.148919582366943,
      "learning_rate": 4.8351329997366345e-05,
      "loss": 2.2314,
      "step": 1256
    },
    {
      "epoch": 0.03310508296023176,
      "grad_norm": 8.156359672546387,
      "learning_rate": 4.835001316829076e-05,
      "loss": 0.9903,
      "step": 1257
    },
    {
      "epoch": 0.03313141954174348,
      "grad_norm": 2.1040937900543213,
      "learning_rate": 4.834869633921517e-05,
      "loss": 2.2143,
      "step": 1258
    },
    {
      "epoch": 0.0331577561232552,
      "grad_norm": 2.2730515003204346,
      "learning_rate": 4.8347379510139585e-05,
      "loss": 1.8334,
      "step": 1259
    },
    {
      "epoch": 0.033184092704766924,
      "grad_norm": 2.8352062702178955,
      "learning_rate": 4.8346062681064e-05,
      "loss": 1.7796,
      "step": 1260
    },
    {
      "epoch": 0.03321042928627864,
      "grad_norm": 4.116631984710693,
      "learning_rate": 4.8344745851988417e-05,
      "loss": 1.8516,
      "step": 1261
    },
    {
      "epoch": 0.03323676586779036,
      "grad_norm": 2.0427281856536865,
      "learning_rate": 4.8343429022912825e-05,
      "loss": 2.4238,
      "step": 1262
    },
    {
      "epoch": 0.03326310244930208,
      "grad_norm": 3.4862911701202393,
      "learning_rate": 4.834211219383724e-05,
      "loss": 1.7011,
      "step": 1263
    },
    {
      "epoch": 0.0332894390308138,
      "grad_norm": 3.1244916915893555,
      "learning_rate": 4.8340795364761657e-05,
      "loss": 1.1916,
      "step": 1264
    },
    {
      "epoch": 0.03331577561232552,
      "grad_norm": 2.486405849456787,
      "learning_rate": 4.833947853568607e-05,
      "loss": 2.1259,
      "step": 1265
    },
    {
      "epoch": 0.03334211219383724,
      "grad_norm": 6.2503767013549805,
      "learning_rate": 4.833816170661049e-05,
      "loss": 0.8626,
      "step": 1266
    },
    {
      "epoch": 0.03336844877534896,
      "grad_norm": 8.085247993469238,
      "learning_rate": 4.8336844877534897e-05,
      "loss": 1.1699,
      "step": 1267
    },
    {
      "epoch": 0.033394785356860676,
      "grad_norm": 2.33953857421875,
      "learning_rate": 4.833552804845931e-05,
      "loss": 2.0922,
      "step": 1268
    },
    {
      "epoch": 0.0334211219383724,
      "grad_norm": 5.278698921203613,
      "learning_rate": 4.833421121938372e-05,
      "loss": 0.9735,
      "step": 1269
    },
    {
      "epoch": 0.03344745851988412,
      "grad_norm": 3.065584421157837,
      "learning_rate": 4.833289439030814e-05,
      "loss": 1.7886,
      "step": 1270
    },
    {
      "epoch": 0.03347379510139584,
      "grad_norm": 3.733517646789551,
      "learning_rate": 4.833157756123255e-05,
      "loss": 1.2978,
      "step": 1271
    },
    {
      "epoch": 0.03350013168290756,
      "grad_norm": 1.9371833801269531,
      "learning_rate": 4.833026073215697e-05,
      "loss": 1.7297,
      "step": 1272
    },
    {
      "epoch": 0.033526468264419275,
      "grad_norm": 2.1965014934539795,
      "learning_rate": 4.832894390308138e-05,
      "loss": 1.7929,
      "step": 1273
    },
    {
      "epoch": 0.033552804845931,
      "grad_norm": 2.186753511428833,
      "learning_rate": 4.83276270740058e-05,
      "loss": 2.1285,
      "step": 1274
    },
    {
      "epoch": 0.03357914142744272,
      "grad_norm": 2.922542095184326,
      "learning_rate": 4.8326310244930215e-05,
      "loss": 1.6997,
      "step": 1275
    },
    {
      "epoch": 0.033605478008954436,
      "grad_norm": 3.0036470890045166,
      "learning_rate": 4.832499341585462e-05,
      "loss": 1.7925,
      "step": 1276
    },
    {
      "epoch": 0.03363181459046616,
      "grad_norm": 7.521142959594727,
      "learning_rate": 4.832367658677904e-05,
      "loss": 2.1324,
      "step": 1277
    },
    {
      "epoch": 0.03365815117197788,
      "grad_norm": 4.4400739669799805,
      "learning_rate": 4.832235975770345e-05,
      "loss": 1.2237,
      "step": 1278
    },
    {
      "epoch": 0.033684487753489596,
      "grad_norm": 2.646451234817505,
      "learning_rate": 4.832104292862787e-05,
      "loss": 1.3697,
      "step": 1279
    },
    {
      "epoch": 0.03371082433500132,
      "grad_norm": 2.9623894691467285,
      "learning_rate": 4.831972609955228e-05,
      "loss": 2.2205,
      "step": 1280
    },
    {
      "epoch": 0.033737160916513034,
      "grad_norm": 4.186872482299805,
      "learning_rate": 4.8318409270476695e-05,
      "loss": 1.6858,
      "step": 1281
    },
    {
      "epoch": 0.03376349749802476,
      "grad_norm": 3.4387664794921875,
      "learning_rate": 4.831709244140111e-05,
      "loss": 1.0801,
      "step": 1282
    },
    {
      "epoch": 0.03378983407953648,
      "grad_norm": 4.123729705810547,
      "learning_rate": 4.831577561232552e-05,
      "loss": 1.5339,
      "step": 1283
    },
    {
      "epoch": 0.033816170661048195,
      "grad_norm": 2.3596715927124023,
      "learning_rate": 4.831445878324994e-05,
      "loss": 1.7717,
      "step": 1284
    },
    {
      "epoch": 0.03384250724255992,
      "grad_norm": 4.8778276443481445,
      "learning_rate": 4.831314195417435e-05,
      "loss": 1.4071,
      "step": 1285
    },
    {
      "epoch": 0.03386884382407163,
      "grad_norm": 2.4648845195770264,
      "learning_rate": 4.8311825125098766e-05,
      "loss": 2.5404,
      "step": 1286
    },
    {
      "epoch": 0.033895180405583356,
      "grad_norm": 2.552497625350952,
      "learning_rate": 4.8310508296023175e-05,
      "loss": 2.2106,
      "step": 1287
    },
    {
      "epoch": 0.03392151698709508,
      "grad_norm": 8.436152458190918,
      "learning_rate": 4.830919146694759e-05,
      "loss": 1.8232,
      "step": 1288
    },
    {
      "epoch": 0.033947853568606794,
      "grad_norm": 2.4023749828338623,
      "learning_rate": 4.8307874637872006e-05,
      "loss": 1.6959,
      "step": 1289
    },
    {
      "epoch": 0.033974190150118516,
      "grad_norm": 8.067594528198242,
      "learning_rate": 4.830655780879642e-05,
      "loss": 1.52,
      "step": 1290
    },
    {
      "epoch": 0.03400052673163023,
      "grad_norm": 6.440790176391602,
      "learning_rate": 4.830524097972084e-05,
      "loss": 1.4087,
      "step": 1291
    },
    {
      "epoch": 0.034026863313141954,
      "grad_norm": 6.596911430358887,
      "learning_rate": 4.8303924150645246e-05,
      "loss": 1.3801,
      "step": 1292
    },
    {
      "epoch": 0.03405319989465368,
      "grad_norm": 2.277406930923462,
      "learning_rate": 4.830260732156967e-05,
      "loss": 1.5915,
      "step": 1293
    },
    {
      "epoch": 0.03407953647616539,
      "grad_norm": 2.351935863494873,
      "learning_rate": 4.830129049249408e-05,
      "loss": 2.3419,
      "step": 1294
    },
    {
      "epoch": 0.034105873057677115,
      "grad_norm": 3.372990846633911,
      "learning_rate": 4.829997366341849e-05,
      "loss": 0.3246,
      "step": 1295
    },
    {
      "epoch": 0.03413220963918883,
      "grad_norm": 4.0242156982421875,
      "learning_rate": 4.82986568343429e-05,
      "loss": 1.3814,
      "step": 1296
    },
    {
      "epoch": 0.03415854622070055,
      "grad_norm": 2.2410614490509033,
      "learning_rate": 4.829734000526732e-05,
      "loss": 1.9321,
      "step": 1297
    },
    {
      "epoch": 0.034184882802212276,
      "grad_norm": 2.1041512489318848,
      "learning_rate": 4.829602317619173e-05,
      "loss": 1.1096,
      "step": 1298
    },
    {
      "epoch": 0.03421121938372399,
      "grad_norm": 11.442770004272461,
      "learning_rate": 4.829470634711615e-05,
      "loss": 1.6356,
      "step": 1299
    },
    {
      "epoch": 0.034237555965235714,
      "grad_norm": 2.5029757022857666,
      "learning_rate": 4.829338951804056e-05,
      "loss": 2.2546,
      "step": 1300
    },
    {
      "epoch": 0.03426389254674743,
      "grad_norm": 3.0751092433929443,
      "learning_rate": 4.829207268896497e-05,
      "loss": 2.0434,
      "step": 1301
    },
    {
      "epoch": 0.03429022912825915,
      "grad_norm": 3.357970714569092,
      "learning_rate": 4.829075585988939e-05,
      "loss": 1.894,
      "step": 1302
    },
    {
      "epoch": 0.034316565709770874,
      "grad_norm": 2.2058346271514893,
      "learning_rate": 4.8289439030813804e-05,
      "loss": 1.9861,
      "step": 1303
    },
    {
      "epoch": 0.03434290229128259,
      "grad_norm": 3.02363657951355,
      "learning_rate": 4.828812220173822e-05,
      "loss": 0.3299,
      "step": 1304
    },
    {
      "epoch": 0.03436923887279431,
      "grad_norm": 4.684982776641846,
      "learning_rate": 4.828680537266263e-05,
      "loss": 1.0779,
      "step": 1305
    },
    {
      "epoch": 0.03439557545430603,
      "grad_norm": 5.264890193939209,
      "learning_rate": 4.8285488543587044e-05,
      "loss": 0.9333,
      "step": 1306
    },
    {
      "epoch": 0.03442191203581775,
      "grad_norm": 3.341413974761963,
      "learning_rate": 4.828417171451146e-05,
      "loss": 1.3823,
      "step": 1307
    },
    {
      "epoch": 0.03444824861732947,
      "grad_norm": 3.4844167232513428,
      "learning_rate": 4.8282854885435875e-05,
      "loss": 1.4234,
      "step": 1308
    },
    {
      "epoch": 0.03447458519884119,
      "grad_norm": 3.4124138355255127,
      "learning_rate": 4.8281538056360284e-05,
      "loss": 1.9774,
      "step": 1309
    },
    {
      "epoch": 0.03450092178035291,
      "grad_norm": 2.2452385425567627,
      "learning_rate": 4.82802212272847e-05,
      "loss": 2.1438,
      "step": 1310
    },
    {
      "epoch": 0.03452725836186463,
      "grad_norm": 3.668550729751587,
      "learning_rate": 4.8278904398209115e-05,
      "loss": 1.36,
      "step": 1311
    },
    {
      "epoch": 0.03455359494337635,
      "grad_norm": 2.3201591968536377,
      "learning_rate": 4.827758756913353e-05,
      "loss": 1.6325,
      "step": 1312
    },
    {
      "epoch": 0.03457993152488807,
      "grad_norm": 2.956763505935669,
      "learning_rate": 4.8276270740057946e-05,
      "loss": 2.9976,
      "step": 1313
    },
    {
      "epoch": 0.03460626810639979,
      "grad_norm": 2.1899678707122803,
      "learning_rate": 4.8274953910982355e-05,
      "loss": 2.1028,
      "step": 1314
    },
    {
      "epoch": 0.03463260468791151,
      "grad_norm": 3.3504998683929443,
      "learning_rate": 4.827363708190677e-05,
      "loss": 0.732,
      "step": 1315
    },
    {
      "epoch": 0.034658941269423225,
      "grad_norm": 2.741833448410034,
      "learning_rate": 4.827232025283118e-05,
      "loss": 1.6869,
      "step": 1316
    },
    {
      "epoch": 0.03468527785093495,
      "grad_norm": 2.77390193939209,
      "learning_rate": 4.82710034237556e-05,
      "loss": 1.678,
      "step": 1317
    },
    {
      "epoch": 0.03471161443244667,
      "grad_norm": 5.46265983581543,
      "learning_rate": 4.826968659468001e-05,
      "loss": 1.5652,
      "step": 1318
    },
    {
      "epoch": 0.034737951013958386,
      "grad_norm": 6.403779983520508,
      "learning_rate": 4.8268369765604426e-05,
      "loss": 0.9487,
      "step": 1319
    },
    {
      "epoch": 0.03476428759547011,
      "grad_norm": 2.5105652809143066,
      "learning_rate": 4.826705293652884e-05,
      "loss": 2.2179,
      "step": 1320
    },
    {
      "epoch": 0.03479062417698183,
      "grad_norm": 2.5746607780456543,
      "learning_rate": 4.826573610745325e-05,
      "loss": 2.0969,
      "step": 1321
    },
    {
      "epoch": 0.03481696075849355,
      "grad_norm": 4.139590263366699,
      "learning_rate": 4.826441927837767e-05,
      "loss": 1.494,
      "step": 1322
    },
    {
      "epoch": 0.03484329734000527,
      "grad_norm": 2.3788700103759766,
      "learning_rate": 4.826310244930208e-05,
      "loss": 1.5353,
      "step": 1323
    },
    {
      "epoch": 0.034869633921516985,
      "grad_norm": 1.9991793632507324,
      "learning_rate": 4.82617856202265e-05,
      "loss": 1.8536,
      "step": 1324
    },
    {
      "epoch": 0.03489597050302871,
      "grad_norm": 3.0393738746643066,
      "learning_rate": 4.8260468791150906e-05,
      "loss": 0.8268,
      "step": 1325
    },
    {
      "epoch": 0.03492230708454043,
      "grad_norm": 3.374997138977051,
      "learning_rate": 4.825915196207533e-05,
      "loss": 1.8157,
      "step": 1326
    },
    {
      "epoch": 0.034948643666052145,
      "grad_norm": 6.295700550079346,
      "learning_rate": 4.825783513299974e-05,
      "loss": 2.6546,
      "step": 1327
    },
    {
      "epoch": 0.03497498024756387,
      "grad_norm": 3.399216651916504,
      "learning_rate": 4.825651830392415e-05,
      "loss": 1.482,
      "step": 1328
    },
    {
      "epoch": 0.035001316829075584,
      "grad_norm": 2.5268938541412354,
      "learning_rate": 4.825520147484857e-05,
      "loss": 2.2271,
      "step": 1329
    },
    {
      "epoch": 0.035027653410587306,
      "grad_norm": 4.140182971954346,
      "learning_rate": 4.825388464577298e-05,
      "loss": 1.6582,
      "step": 1330
    },
    {
      "epoch": 0.03505398999209903,
      "grad_norm": 2.9830386638641357,
      "learning_rate": 4.82525678166974e-05,
      "loss": 2.0572,
      "step": 1331
    },
    {
      "epoch": 0.035080326573610744,
      "grad_norm": 5.903008460998535,
      "learning_rate": 4.825125098762181e-05,
      "loss": 0.4726,
      "step": 1332
    },
    {
      "epoch": 0.03510666315512247,
      "grad_norm": 4.696227550506592,
      "learning_rate": 4.8249934158546224e-05,
      "loss": 1.269,
      "step": 1333
    },
    {
      "epoch": 0.03513299973663418,
      "grad_norm": 4.289698123931885,
      "learning_rate": 4.824861732947063e-05,
      "loss": 0.7344,
      "step": 1334
    },
    {
      "epoch": 0.035159336318145905,
      "grad_norm": 2.677140235900879,
      "learning_rate": 4.824730050039505e-05,
      "loss": 1.0939,
      "step": 1335
    },
    {
      "epoch": 0.03518567289965763,
      "grad_norm": 2.237262487411499,
      "learning_rate": 4.8245983671319464e-05,
      "loss": 2.1044,
      "step": 1336
    },
    {
      "epoch": 0.03521200948116934,
      "grad_norm": 2.694110155105591,
      "learning_rate": 4.824466684224388e-05,
      "loss": 0.5662,
      "step": 1337
    },
    {
      "epoch": 0.035238346062681065,
      "grad_norm": 2.9993741512298584,
      "learning_rate": 4.8243350013168296e-05,
      "loss": 0.7674,
      "step": 1338
    },
    {
      "epoch": 0.03526468264419278,
      "grad_norm": 1.9640650749206543,
      "learning_rate": 4.8242033184092704e-05,
      "loss": 2.2223,
      "step": 1339
    },
    {
      "epoch": 0.035291019225704504,
      "grad_norm": 2.2978978157043457,
      "learning_rate": 4.824071635501713e-05,
      "loss": 2.1875,
      "step": 1340
    },
    {
      "epoch": 0.035317355807216226,
      "grad_norm": 3.9391028881073,
      "learning_rate": 4.8239399525941536e-05,
      "loss": 1.8525,
      "step": 1341
    },
    {
      "epoch": 0.03534369238872794,
      "grad_norm": 3.5753555297851562,
      "learning_rate": 4.823808269686595e-05,
      "loss": 2.3247,
      "step": 1342
    },
    {
      "epoch": 0.035370028970239664,
      "grad_norm": 3.101198673248291,
      "learning_rate": 4.823676586779036e-05,
      "loss": 2.4757,
      "step": 1343
    },
    {
      "epoch": 0.03539636555175138,
      "grad_norm": 3.337867021560669,
      "learning_rate": 4.8235449038714776e-05,
      "loss": 1.4835,
      "step": 1344
    },
    {
      "epoch": 0.0354227021332631,
      "grad_norm": 2.6907167434692383,
      "learning_rate": 4.823413220963919e-05,
      "loss": 1.7228,
      "step": 1345
    },
    {
      "epoch": 0.035449038714774825,
      "grad_norm": 3.3301050662994385,
      "learning_rate": 4.823281538056361e-05,
      "loss": 1.9144,
      "step": 1346
    },
    {
      "epoch": 0.03547537529628654,
      "grad_norm": 3.553663492202759,
      "learning_rate": 4.823149855148802e-05,
      "loss": 2.2447,
      "step": 1347
    },
    {
      "epoch": 0.03550171187779826,
      "grad_norm": 4.311988353729248,
      "learning_rate": 4.823018172241243e-05,
      "loss": 1.6596,
      "step": 1348
    },
    {
      "epoch": 0.03552804845930998,
      "grad_norm": 3.39910888671875,
      "learning_rate": 4.822886489333685e-05,
      "loss": 1.9699,
      "step": 1349
    },
    {
      "epoch": 0.0355543850408217,
      "grad_norm": 2.2859280109405518,
      "learning_rate": 4.822754806426126e-05,
      "loss": 1.8481,
      "step": 1350
    },
    {
      "epoch": 0.035580721622333424,
      "grad_norm": 4.299882888793945,
      "learning_rate": 4.822623123518568e-05,
      "loss": 1.5176,
      "step": 1351
    },
    {
      "epoch": 0.03560705820384514,
      "grad_norm": 4.792234420776367,
      "learning_rate": 4.822491440611009e-05,
      "loss": 1.6701,
      "step": 1352
    },
    {
      "epoch": 0.03563339478535686,
      "grad_norm": 3.010335683822632,
      "learning_rate": 4.82235975770345e-05,
      "loss": 0.8111,
      "step": 1353
    },
    {
      "epoch": 0.03565973136686858,
      "grad_norm": 2.2211203575134277,
      "learning_rate": 4.822228074795891e-05,
      "loss": 1.9087,
      "step": 1354
    },
    {
      "epoch": 0.0356860679483803,
      "grad_norm": 2.451058864593506,
      "learning_rate": 4.8220963918883334e-05,
      "loss": 1.545,
      "step": 1355
    },
    {
      "epoch": 0.03571240452989202,
      "grad_norm": 2.579103946685791,
      "learning_rate": 4.821964708980774e-05,
      "loss": 1.796,
      "step": 1356
    },
    {
      "epoch": 0.03573874111140374,
      "grad_norm": 4.573684215545654,
      "learning_rate": 4.821833026073216e-05,
      "loss": 1.0103,
      "step": 1357
    },
    {
      "epoch": 0.03576507769291546,
      "grad_norm": 2.26727294921875,
      "learning_rate": 4.8217013431656574e-05,
      "loss": 1.5844,
      "step": 1358
    },
    {
      "epoch": 0.035791414274427176,
      "grad_norm": 3.4333019256591797,
      "learning_rate": 4.821569660258099e-05,
      "loss": 1.7426,
      "step": 1359
    },
    {
      "epoch": 0.0358177508559389,
      "grad_norm": 7.677754878997803,
      "learning_rate": 4.8214379773505405e-05,
      "loss": 1.8325,
      "step": 1360
    },
    {
      "epoch": 0.03584408743745062,
      "grad_norm": 3.0336837768554688,
      "learning_rate": 4.8213062944429814e-05,
      "loss": 2.039,
      "step": 1361
    },
    {
      "epoch": 0.03587042401896234,
      "grad_norm": 3.4892420768737793,
      "learning_rate": 4.821174611535423e-05,
      "loss": 1.5181,
      "step": 1362
    },
    {
      "epoch": 0.03589676060047406,
      "grad_norm": 3.1478965282440186,
      "learning_rate": 4.821042928627864e-05,
      "loss": 1.4686,
      "step": 1363
    },
    {
      "epoch": 0.03592309718198578,
      "grad_norm": 2.5794179439544678,
      "learning_rate": 4.820911245720306e-05,
      "loss": 0.4452,
      "step": 1364
    },
    {
      "epoch": 0.0359494337634975,
      "grad_norm": 2.2024765014648438,
      "learning_rate": 4.820779562812747e-05,
      "loss": 1.4755,
      "step": 1365
    },
    {
      "epoch": 0.03597577034500922,
      "grad_norm": 2.669931173324585,
      "learning_rate": 4.8206478799051885e-05,
      "loss": 1.8739,
      "step": 1366
    },
    {
      "epoch": 0.036002106926520935,
      "grad_norm": 3.913017511367798,
      "learning_rate": 4.82051619699763e-05,
      "loss": 1.9402,
      "step": 1367
    },
    {
      "epoch": 0.03602844350803266,
      "grad_norm": 3.415797472000122,
      "learning_rate": 4.820384514090071e-05,
      "loss": 1.0633,
      "step": 1368
    },
    {
      "epoch": 0.03605478008954438,
      "grad_norm": 3.1644015312194824,
      "learning_rate": 4.820252831182513e-05,
      "loss": 1.5442,
      "step": 1369
    },
    {
      "epoch": 0.036081116671056096,
      "grad_norm": 3.9022562503814697,
      "learning_rate": 4.820121148274954e-05,
      "loss": 1.7415,
      "step": 1370
    },
    {
      "epoch": 0.03610745325256782,
      "grad_norm": 3.3646533489227295,
      "learning_rate": 4.8199894653673956e-05,
      "loss": 2.3507,
      "step": 1371
    },
    {
      "epoch": 0.036133789834079534,
      "grad_norm": 2.9992384910583496,
      "learning_rate": 4.8198577824598365e-05,
      "loss": 1.7051,
      "step": 1372
    },
    {
      "epoch": 0.03616012641559126,
      "grad_norm": 3.279506206512451,
      "learning_rate": 4.819726099552279e-05,
      "loss": 1.307,
      "step": 1373
    },
    {
      "epoch": 0.03618646299710298,
      "grad_norm": 3.8295230865478516,
      "learning_rate": 4.8195944166447196e-05,
      "loss": 1.657,
      "step": 1374
    },
    {
      "epoch": 0.036212799578614695,
      "grad_norm": 4.198893070220947,
      "learning_rate": 4.819462733737161e-05,
      "loss": 0.9144,
      "step": 1375
    },
    {
      "epoch": 0.03623913616012642,
      "grad_norm": 2.8750100135803223,
      "learning_rate": 4.819331050829603e-05,
      "loss": 2.0248,
      "step": 1376
    },
    {
      "epoch": 0.03626547274163813,
      "grad_norm": 2.485234260559082,
      "learning_rate": 4.8191993679220436e-05,
      "loss": 1.6745,
      "step": 1377
    },
    {
      "epoch": 0.036291809323149855,
      "grad_norm": 2.555755376815796,
      "learning_rate": 4.819067685014486e-05,
      "loss": 1.0228,
      "step": 1378
    },
    {
      "epoch": 0.03631814590466158,
      "grad_norm": 3.703505754470825,
      "learning_rate": 4.818936002106927e-05,
      "loss": 0.6707,
      "step": 1379
    },
    {
      "epoch": 0.03634448248617329,
      "grad_norm": 2.726428508758545,
      "learning_rate": 4.818804319199368e-05,
      "loss": 1.5401,
      "step": 1380
    },
    {
      "epoch": 0.036370819067685016,
      "grad_norm": 3.7582430839538574,
      "learning_rate": 4.818672636291809e-05,
      "loss": 1.1392,
      "step": 1381
    },
    {
      "epoch": 0.03639715564919673,
      "grad_norm": 2.7370986938476562,
      "learning_rate": 4.818540953384251e-05,
      "loss": 2.1486,
      "step": 1382
    },
    {
      "epoch": 0.036423492230708454,
      "grad_norm": 3.982513189315796,
      "learning_rate": 4.818409270476692e-05,
      "loss": 2.6714,
      "step": 1383
    },
    {
      "epoch": 0.03644982881222018,
      "grad_norm": 3.420853614807129,
      "learning_rate": 4.818277587569134e-05,
      "loss": 0.4569,
      "step": 1384
    },
    {
      "epoch": 0.03647616539373189,
      "grad_norm": 2.9245407581329346,
      "learning_rate": 4.8181459046615754e-05,
      "loss": 2.3995,
      "step": 1385
    },
    {
      "epoch": 0.036502501975243615,
      "grad_norm": 8.815753936767578,
      "learning_rate": 4.818014221754016e-05,
      "loss": 1.1951,
      "step": 1386
    },
    {
      "epoch": 0.03652883855675533,
      "grad_norm": 5.456750392913818,
      "learning_rate": 4.8178825388464585e-05,
      "loss": 1.5686,
      "step": 1387
    },
    {
      "epoch": 0.03655517513826705,
      "grad_norm": 3.259455919265747,
      "learning_rate": 4.8177508559388994e-05,
      "loss": 1.5534,
      "step": 1388
    },
    {
      "epoch": 0.036581511719778775,
      "grad_norm": 2.732182741165161,
      "learning_rate": 4.817619173031341e-05,
      "loss": 1.7391,
      "step": 1389
    },
    {
      "epoch": 0.03660784830129049,
      "grad_norm": 3.078794479370117,
      "learning_rate": 4.817487490123782e-05,
      "loss": 2.286,
      "step": 1390
    },
    {
      "epoch": 0.03663418488280221,
      "grad_norm": 4.257495880126953,
      "learning_rate": 4.8173558072162234e-05,
      "loss": 1.0608,
      "step": 1391
    },
    {
      "epoch": 0.03666052146431393,
      "grad_norm": 3.027405261993408,
      "learning_rate": 4.817224124308665e-05,
      "loss": 1.176,
      "step": 1392
    },
    {
      "epoch": 0.03668685804582565,
      "grad_norm": 3.272449493408203,
      "learning_rate": 4.8170924414011065e-05,
      "loss": 1.7577,
      "step": 1393
    },
    {
      "epoch": 0.036713194627337374,
      "grad_norm": 7.281032562255859,
      "learning_rate": 4.816960758493548e-05,
      "loss": 2.452,
      "step": 1394
    },
    {
      "epoch": 0.03673953120884909,
      "grad_norm": 2.2292678356170654,
      "learning_rate": 4.816829075585989e-05,
      "loss": 1.9123,
      "step": 1395
    },
    {
      "epoch": 0.03676586779036081,
      "grad_norm": 4.264851093292236,
      "learning_rate": 4.8166973926784305e-05,
      "loss": 0.8762,
      "step": 1396
    },
    {
      "epoch": 0.03679220437187253,
      "grad_norm": 3.4080307483673096,
      "learning_rate": 4.816565709770872e-05,
      "loss": 1.0579,
      "step": 1397
    },
    {
      "epoch": 0.03681854095338425,
      "grad_norm": 2.9511606693267822,
      "learning_rate": 4.8164340268633137e-05,
      "loss": 1.876,
      "step": 1398
    },
    {
      "epoch": 0.03684487753489597,
      "grad_norm": 3.078354597091675,
      "learning_rate": 4.8163023439557545e-05,
      "loss": 2.0897,
      "step": 1399
    },
    {
      "epoch": 0.03687121411640769,
      "grad_norm": 2.9722883701324463,
      "learning_rate": 4.816170661048196e-05,
      "loss": 1.5776,
      "step": 1400
    },
    {
      "epoch": 0.03689755069791941,
      "grad_norm": 4.6948723793029785,
      "learning_rate": 4.816038978140637e-05,
      "loss": 1.19,
      "step": 1401
    },
    {
      "epoch": 0.036923887279431127,
      "grad_norm": 2.6448488235473633,
      "learning_rate": 4.815907295233079e-05,
      "loss": 1.4374,
      "step": 1402
    },
    {
      "epoch": 0.03695022386094285,
      "grad_norm": 2.951441526412964,
      "learning_rate": 4.81577561232552e-05,
      "loss": 2.1182,
      "step": 1403
    },
    {
      "epoch": 0.03697656044245457,
      "grad_norm": 2.4465129375457764,
      "learning_rate": 4.8156439294179617e-05,
      "loss": 1.7781,
      "step": 1404
    },
    {
      "epoch": 0.03700289702396629,
      "grad_norm": 2.9098169803619385,
      "learning_rate": 4.815512246510403e-05,
      "loss": 1.9132,
      "step": 1405
    },
    {
      "epoch": 0.03702923360547801,
      "grad_norm": 2.493217945098877,
      "learning_rate": 4.815380563602845e-05,
      "loss": 2.0592,
      "step": 1406
    },
    {
      "epoch": 0.037055570186989725,
      "grad_norm": 3.7359561920166016,
      "learning_rate": 4.815248880695286e-05,
      "loss": 1.5837,
      "step": 1407
    },
    {
      "epoch": 0.03708190676850145,
      "grad_norm": 2.9856550693511963,
      "learning_rate": 4.815117197787727e-05,
      "loss": 2.203,
      "step": 1408
    },
    {
      "epoch": 0.03710824335001317,
      "grad_norm": 2.2589473724365234,
      "learning_rate": 4.814985514880169e-05,
      "loss": 1.9073,
      "step": 1409
    },
    {
      "epoch": 0.037134579931524886,
      "grad_norm": 3.434154748916626,
      "learning_rate": 4.81485383197261e-05,
      "loss": 1.8538,
      "step": 1410
    },
    {
      "epoch": 0.03716091651303661,
      "grad_norm": 2.994636058807373,
      "learning_rate": 4.814722149065052e-05,
      "loss": 0.6366,
      "step": 1411
    },
    {
      "epoch": 0.03718725309454833,
      "grad_norm": 2.82177734375,
      "learning_rate": 4.814590466157493e-05,
      "loss": 1.4591,
      "step": 1412
    },
    {
      "epoch": 0.037213589676060047,
      "grad_norm": 4.66945219039917,
      "learning_rate": 4.8144587832499343e-05,
      "loss": 2.239,
      "step": 1413
    },
    {
      "epoch": 0.03723992625757177,
      "grad_norm": 3.21661639213562,
      "learning_rate": 4.814327100342376e-05,
      "loss": 1.3781,
      "step": 1414
    },
    {
      "epoch": 0.037266262839083485,
      "grad_norm": 2.4987006187438965,
      "learning_rate": 4.814195417434817e-05,
      "loss": 1.7669,
      "step": 1415
    },
    {
      "epoch": 0.03729259942059521,
      "grad_norm": 2.091668128967285,
      "learning_rate": 4.814063734527259e-05,
      "loss": 1.5082,
      "step": 1416
    },
    {
      "epoch": 0.03731893600210693,
      "grad_norm": 2.086125373840332,
      "learning_rate": 4.8139320516197e-05,
      "loss": 1.6896,
      "step": 1417
    },
    {
      "epoch": 0.037345272583618645,
      "grad_norm": 2.369629144668579,
      "learning_rate": 4.8138003687121415e-05,
      "loss": 2.1499,
      "step": 1418
    },
    {
      "epoch": 0.03737160916513037,
      "grad_norm": 2.1372570991516113,
      "learning_rate": 4.8136686858045823e-05,
      "loss": 1.6387,
      "step": 1419
    },
    {
      "epoch": 0.03739794574664208,
      "grad_norm": 2.225848436355591,
      "learning_rate": 4.8135370028970246e-05,
      "loss": 2.3318,
      "step": 1420
    },
    {
      "epoch": 0.037424282328153806,
      "grad_norm": 4.289445877075195,
      "learning_rate": 4.8134053199894655e-05,
      "loss": 1.773,
      "step": 1421
    },
    {
      "epoch": 0.03745061890966553,
      "grad_norm": 5.045720100402832,
      "learning_rate": 4.813273637081907e-05,
      "loss": 1.1577,
      "step": 1422
    },
    {
      "epoch": 0.037476955491177244,
      "grad_norm": 2.802020311355591,
      "learning_rate": 4.8131419541743486e-05,
      "loss": 1.5965,
      "step": 1423
    },
    {
      "epoch": 0.037503292072688967,
      "grad_norm": 3.742877244949341,
      "learning_rate": 4.8130102712667895e-05,
      "loss": 1.2814,
      "step": 1424
    },
    {
      "epoch": 0.03752962865420068,
      "grad_norm": 4.184816360473633,
      "learning_rate": 4.812878588359232e-05,
      "loss": 1.8437,
      "step": 1425
    },
    {
      "epoch": 0.037555965235712405,
      "grad_norm": 5.02267599105835,
      "learning_rate": 4.8127469054516726e-05,
      "loss": 0.6091,
      "step": 1426
    },
    {
      "epoch": 0.03758230181722413,
      "grad_norm": 5.021767616271973,
      "learning_rate": 4.812615222544114e-05,
      "loss": 0.8637,
      "step": 1427
    },
    {
      "epoch": 0.03760863839873584,
      "grad_norm": 6.322299480438232,
      "learning_rate": 4.812483539636555e-05,
      "loss": 1.9106,
      "step": 1428
    },
    {
      "epoch": 0.037634974980247565,
      "grad_norm": 4.426939487457275,
      "learning_rate": 4.8123518567289966e-05,
      "loss": 0.513,
      "step": 1429
    },
    {
      "epoch": 0.03766131156175928,
      "grad_norm": 3.189406394958496,
      "learning_rate": 4.812220173821438e-05,
      "loss": 2.2815,
      "step": 1430
    },
    {
      "epoch": 0.037687648143271,
      "grad_norm": 4.226385593414307,
      "learning_rate": 4.81208849091388e-05,
      "loss": 1.8733,
      "step": 1431
    },
    {
      "epoch": 0.037713984724782726,
      "grad_norm": 2.647603988647461,
      "learning_rate": 4.811956808006321e-05,
      "loss": 2.4634,
      "step": 1432
    },
    {
      "epoch": 0.03774032130629444,
      "grad_norm": 2.241933584213257,
      "learning_rate": 4.811825125098762e-05,
      "loss": 1.9525,
      "step": 1433
    },
    {
      "epoch": 0.037766657887806164,
      "grad_norm": 3.8084769248962402,
      "learning_rate": 4.811693442191204e-05,
      "loss": 1.5282,
      "step": 1434
    },
    {
      "epoch": 0.03779299446931788,
      "grad_norm": 2.568134307861328,
      "learning_rate": 4.811561759283645e-05,
      "loss": 1.8027,
      "step": 1435
    },
    {
      "epoch": 0.0378193310508296,
      "grad_norm": 1.9084701538085938,
      "learning_rate": 4.811430076376087e-05,
      "loss": 1.7684,
      "step": 1436
    },
    {
      "epoch": 0.037845667632341325,
      "grad_norm": 4.7902679443359375,
      "learning_rate": 4.811298393468528e-05,
      "loss": 1.4196,
      "step": 1437
    },
    {
      "epoch": 0.03787200421385304,
      "grad_norm": 2.587052822113037,
      "learning_rate": 4.811166710560969e-05,
      "loss": 1.7431,
      "step": 1438
    },
    {
      "epoch": 0.03789834079536476,
      "grad_norm": 3.3005499839782715,
      "learning_rate": 4.811035027653411e-05,
      "loss": 1.8728,
      "step": 1439
    },
    {
      "epoch": 0.03792467737687648,
      "grad_norm": 3.298290252685547,
      "learning_rate": 4.8109033447458524e-05,
      "loss": 1.9283,
      "step": 1440
    },
    {
      "epoch": 0.0379510139583882,
      "grad_norm": 2.685931921005249,
      "learning_rate": 4.810771661838294e-05,
      "loss": 1.651,
      "step": 1441
    },
    {
      "epoch": 0.03797735053989992,
      "grad_norm": 1.9227145910263062,
      "learning_rate": 4.810639978930735e-05,
      "loss": 1.8952,
      "step": 1442
    },
    {
      "epoch": 0.03800368712141164,
      "grad_norm": 2.178524971008301,
      "learning_rate": 4.8105082960231764e-05,
      "loss": 1.5836,
      "step": 1443
    },
    {
      "epoch": 0.03803002370292336,
      "grad_norm": 3.530819892883301,
      "learning_rate": 4.810376613115618e-05,
      "loss": 1.6536,
      "step": 1444
    },
    {
      "epoch": 0.03805636028443508,
      "grad_norm": 4.413656234741211,
      "learning_rate": 4.8102449302080595e-05,
      "loss": 1.7439,
      "step": 1445
    },
    {
      "epoch": 0.0380826968659468,
      "grad_norm": 2.618088960647583,
      "learning_rate": 4.8101132473005004e-05,
      "loss": 1.6252,
      "step": 1446
    },
    {
      "epoch": 0.03810903344745852,
      "grad_norm": 6.07904052734375,
      "learning_rate": 4.809981564392942e-05,
      "loss": 2.3006,
      "step": 1447
    },
    {
      "epoch": 0.03813537002897024,
      "grad_norm": 4.130184650421143,
      "learning_rate": 4.8098498814853835e-05,
      "loss": 1.7968,
      "step": 1448
    },
    {
      "epoch": 0.03816170661048196,
      "grad_norm": 2.354788064956665,
      "learning_rate": 4.809718198577825e-05,
      "loss": 1.7768,
      "step": 1449
    },
    {
      "epoch": 0.038188043191993676,
      "grad_norm": 4.8974409103393555,
      "learning_rate": 4.8095865156702666e-05,
      "loss": 2.2463,
      "step": 1450
    },
    {
      "epoch": 0.0382143797735054,
      "grad_norm": 2.6150875091552734,
      "learning_rate": 4.8094548327627075e-05,
      "loss": 1.1499,
      "step": 1451
    },
    {
      "epoch": 0.03824071635501712,
      "grad_norm": 2.972714900970459,
      "learning_rate": 4.809323149855149e-05,
      "loss": 0.6017,
      "step": 1452
    },
    {
      "epoch": 0.038267052936528836,
      "grad_norm": 2.5725221633911133,
      "learning_rate": 4.8091914669475906e-05,
      "loss": 1.7313,
      "step": 1453
    },
    {
      "epoch": 0.03829338951804056,
      "grad_norm": 3.1481668949127197,
      "learning_rate": 4.809059784040032e-05,
      "loss": 2.4714,
      "step": 1454
    },
    {
      "epoch": 0.03831972609955228,
      "grad_norm": 3.2930145263671875,
      "learning_rate": 4.808928101132473e-05,
      "loss": 1.4067,
      "step": 1455
    },
    {
      "epoch": 0.038346062681064,
      "grad_norm": 2.3901965618133545,
      "learning_rate": 4.8087964182249146e-05,
      "loss": 2.0908,
      "step": 1456
    },
    {
      "epoch": 0.03837239926257572,
      "grad_norm": 2.4288535118103027,
      "learning_rate": 4.8086647353173555e-05,
      "loss": 1.7124,
      "step": 1457
    },
    {
      "epoch": 0.038398735844087435,
      "grad_norm": 2.4708545207977295,
      "learning_rate": 4.808533052409798e-05,
      "loss": 1.7156,
      "step": 1458
    },
    {
      "epoch": 0.03842507242559916,
      "grad_norm": 4.000490665435791,
      "learning_rate": 4.8084013695022386e-05,
      "loss": 0.8025,
      "step": 1459
    },
    {
      "epoch": 0.03845140900711088,
      "grad_norm": 2.33880615234375,
      "learning_rate": 4.80826968659468e-05,
      "loss": 2.1641,
      "step": 1460
    },
    {
      "epoch": 0.038477745588622596,
      "grad_norm": 4.5710673332214355,
      "learning_rate": 4.808138003687122e-05,
      "loss": 1.4686,
      "step": 1461
    },
    {
      "epoch": 0.03850408217013432,
      "grad_norm": 7.145897388458252,
      "learning_rate": 4.8080063207795626e-05,
      "loss": 0.7443,
      "step": 1462
    },
    {
      "epoch": 0.038530418751646034,
      "grad_norm": 2.459108352661133,
      "learning_rate": 4.807874637872005e-05,
      "loss": 1.6073,
      "step": 1463
    },
    {
      "epoch": 0.038556755333157756,
      "grad_norm": 3.707882881164551,
      "learning_rate": 4.807742954964446e-05,
      "loss": 1.2412,
      "step": 1464
    },
    {
      "epoch": 0.03858309191466948,
      "grad_norm": 5.278344631195068,
      "learning_rate": 4.807611272056887e-05,
      "loss": 1.1009,
      "step": 1465
    },
    {
      "epoch": 0.038609428496181195,
      "grad_norm": 2.403775930404663,
      "learning_rate": 4.807479589149328e-05,
      "loss": 1.7357,
      "step": 1466
    },
    {
      "epoch": 0.03863576507769292,
      "grad_norm": 4.183655738830566,
      "learning_rate": 4.80734790624177e-05,
      "loss": 1.335,
      "step": 1467
    },
    {
      "epoch": 0.03866210165920463,
      "grad_norm": 2.9094831943511963,
      "learning_rate": 4.807216223334211e-05,
      "loss": 0.8503,
      "step": 1468
    },
    {
      "epoch": 0.038688438240716355,
      "grad_norm": 3.1184732913970947,
      "learning_rate": 4.807084540426653e-05,
      "loss": 0.6333,
      "step": 1469
    },
    {
      "epoch": 0.03871477482222808,
      "grad_norm": 2.4614548683166504,
      "learning_rate": 4.8069528575190944e-05,
      "loss": 1.896,
      "step": 1470
    },
    {
      "epoch": 0.03874111140373979,
      "grad_norm": 2.6103456020355225,
      "learning_rate": 4.806821174611535e-05,
      "loss": 1.2525,
      "step": 1471
    },
    {
      "epoch": 0.038767447985251516,
      "grad_norm": 3.9586451053619385,
      "learning_rate": 4.8066894917039776e-05,
      "loss": 0.6833,
      "step": 1472
    },
    {
      "epoch": 0.03879378456676323,
      "grad_norm": 2.902448892593384,
      "learning_rate": 4.8065578087964184e-05,
      "loss": 1.4491,
      "step": 1473
    },
    {
      "epoch": 0.038820121148274954,
      "grad_norm": 2.5038821697235107,
      "learning_rate": 4.80642612588886e-05,
      "loss": 2.2628,
      "step": 1474
    },
    {
      "epoch": 0.038846457729786676,
      "grad_norm": 3.35122013092041,
      "learning_rate": 4.806294442981301e-05,
      "loss": 2.0252,
      "step": 1475
    },
    {
      "epoch": 0.03887279431129839,
      "grad_norm": 6.653318881988525,
      "learning_rate": 4.8061627600737424e-05,
      "loss": 1.5859,
      "step": 1476
    },
    {
      "epoch": 0.038899130892810115,
      "grad_norm": 2.248802423477173,
      "learning_rate": 4.806031077166184e-05,
      "loss": 1.708,
      "step": 1477
    },
    {
      "epoch": 0.03892546747432183,
      "grad_norm": 4.171624183654785,
      "learning_rate": 4.8058993942586256e-05,
      "loss": 1.8427,
      "step": 1478
    },
    {
      "epoch": 0.03895180405583355,
      "grad_norm": 6.605432987213135,
      "learning_rate": 4.805767711351067e-05,
      "loss": 1.4973,
      "step": 1479
    },
    {
      "epoch": 0.038978140637345275,
      "grad_norm": 4.893253326416016,
      "learning_rate": 4.805636028443508e-05,
      "loss": 1.768,
      "step": 1480
    },
    {
      "epoch": 0.03900447721885699,
      "grad_norm": 2.6387462615966797,
      "learning_rate": 4.8055043455359496e-05,
      "loss": 1.2058,
      "step": 1481
    },
    {
      "epoch": 0.03903081380036871,
      "grad_norm": 2.1615798473358154,
      "learning_rate": 4.805372662628391e-05,
      "loss": 2.7933,
      "step": 1482
    },
    {
      "epoch": 0.03905715038188043,
      "grad_norm": 4.006496906280518,
      "learning_rate": 4.805240979720833e-05,
      "loss": 1.5641,
      "step": 1483
    },
    {
      "epoch": 0.03908348696339215,
      "grad_norm": 3.03674578666687,
      "learning_rate": 4.8051092968132736e-05,
      "loss": 1.5177,
      "step": 1484
    },
    {
      "epoch": 0.039109823544903874,
      "grad_norm": 3.581326723098755,
      "learning_rate": 4.804977613905715e-05,
      "loss": 1.3614,
      "step": 1485
    },
    {
      "epoch": 0.03913616012641559,
      "grad_norm": 2.2554171085357666,
      "learning_rate": 4.804845930998157e-05,
      "loss": 2.1849,
      "step": 1486
    },
    {
      "epoch": 0.03916249670792731,
      "grad_norm": 2.508643388748169,
      "learning_rate": 4.804714248090598e-05,
      "loss": 0.3424,
      "step": 1487
    },
    {
      "epoch": 0.03918883328943903,
      "grad_norm": 2.149822235107422,
      "learning_rate": 4.80458256518304e-05,
      "loss": 1.5553,
      "step": 1488
    },
    {
      "epoch": 0.03921516987095075,
      "grad_norm": 2.9578163623809814,
      "learning_rate": 4.804450882275481e-05,
      "loss": 1.2616,
      "step": 1489
    },
    {
      "epoch": 0.03924150645246247,
      "grad_norm": 3.0891036987304688,
      "learning_rate": 4.804319199367922e-05,
      "loss": 0.5437,
      "step": 1490
    },
    {
      "epoch": 0.03926784303397419,
      "grad_norm": 3.854132652282715,
      "learning_rate": 4.804187516460364e-05,
      "loss": 1.0159,
      "step": 1491
    },
    {
      "epoch": 0.03929417961548591,
      "grad_norm": 2.0185046195983887,
      "learning_rate": 4.8040558335528054e-05,
      "loss": 1.455,
      "step": 1492
    },
    {
      "epoch": 0.039320516196997626,
      "grad_norm": 2.6549127101898193,
      "learning_rate": 4.803924150645246e-05,
      "loss": 1.2516,
      "step": 1493
    },
    {
      "epoch": 0.03934685277850935,
      "grad_norm": 2.0132808685302734,
      "learning_rate": 4.803792467737688e-05,
      "loss": 2.1761,
      "step": 1494
    },
    {
      "epoch": 0.03937318936002107,
      "grad_norm": 3.6217565536499023,
      "learning_rate": 4.8036607848301294e-05,
      "loss": 1.6783,
      "step": 1495
    },
    {
      "epoch": 0.03939952594153279,
      "grad_norm": 2.7061996459960938,
      "learning_rate": 4.803529101922571e-05,
      "loss": 1.9596,
      "step": 1496
    },
    {
      "epoch": 0.03942586252304451,
      "grad_norm": 3.6776225566864014,
      "learning_rate": 4.8033974190150125e-05,
      "loss": 2.2301,
      "step": 1497
    },
    {
      "epoch": 0.03945219910455623,
      "grad_norm": 2.7040247917175293,
      "learning_rate": 4.8032657361074534e-05,
      "loss": 1.9844,
      "step": 1498
    },
    {
      "epoch": 0.03947853568606795,
      "grad_norm": 5.953075885772705,
      "learning_rate": 4.803134053199895e-05,
      "loss": 1.189,
      "step": 1499
    },
    {
      "epoch": 0.03950487226757967,
      "grad_norm": 4.968349933624268,
      "learning_rate": 4.803002370292336e-05,
      "loss": 1.6651,
      "step": 1500
    },
    {
      "epoch": 0.039531208849091386,
      "grad_norm": 2.8154876232147217,
      "learning_rate": 4.802870687384778e-05,
      "loss": 2.0657,
      "step": 1501
    },
    {
      "epoch": 0.03955754543060311,
      "grad_norm": 5.4353203773498535,
      "learning_rate": 4.802739004477219e-05,
      "loss": 1.0015,
      "step": 1502
    },
    {
      "epoch": 0.03958388201211483,
      "grad_norm": 6.000146389007568,
      "learning_rate": 4.8026073215696605e-05,
      "loss": 1.6951,
      "step": 1503
    },
    {
      "epoch": 0.039610218593626546,
      "grad_norm": 3.825589656829834,
      "learning_rate": 4.8024756386621014e-05,
      "loss": 1.8409,
      "step": 1504
    },
    {
      "epoch": 0.03963655517513827,
      "grad_norm": 3.282968044281006,
      "learning_rate": 4.8023439557545436e-05,
      "loss": 0.5316,
      "step": 1505
    },
    {
      "epoch": 0.039662891756649984,
      "grad_norm": 5.311993598937988,
      "learning_rate": 4.8022122728469845e-05,
      "loss": 1.5813,
      "step": 1506
    },
    {
      "epoch": 0.03968922833816171,
      "grad_norm": 2.694779872894287,
      "learning_rate": 4.802080589939426e-05,
      "loss": 2.5022,
      "step": 1507
    },
    {
      "epoch": 0.03971556491967343,
      "grad_norm": 3.5496184825897217,
      "learning_rate": 4.8019489070318676e-05,
      "loss": 0.5585,
      "step": 1508
    },
    {
      "epoch": 0.039741901501185145,
      "grad_norm": 4.04374885559082,
      "learning_rate": 4.8018172241243085e-05,
      "loss": 1.0659,
      "step": 1509
    },
    {
      "epoch": 0.03976823808269687,
      "grad_norm": 3.6027233600616455,
      "learning_rate": 4.801685541216751e-05,
      "loss": 2.686,
      "step": 1510
    },
    {
      "epoch": 0.03979457466420858,
      "grad_norm": 3.1707308292388916,
      "learning_rate": 4.8015538583091916e-05,
      "loss": 1.6032,
      "step": 1511
    },
    {
      "epoch": 0.039820911245720306,
      "grad_norm": 2.2083845138549805,
      "learning_rate": 4.801422175401633e-05,
      "loss": 2.2039,
      "step": 1512
    },
    {
      "epoch": 0.03984724782723203,
      "grad_norm": 2.095097541809082,
      "learning_rate": 4.801290492494074e-05,
      "loss": 2.2174,
      "step": 1513
    },
    {
      "epoch": 0.039873584408743744,
      "grad_norm": 4.769643306732178,
      "learning_rate": 4.8011588095865156e-05,
      "loss": 1.6258,
      "step": 1514
    },
    {
      "epoch": 0.039899920990255466,
      "grad_norm": 2.147710084915161,
      "learning_rate": 4.801027126678957e-05,
      "loss": 1.9945,
      "step": 1515
    },
    {
      "epoch": 0.03992625757176718,
      "grad_norm": 2.94934344291687,
      "learning_rate": 4.800895443771399e-05,
      "loss": 1.8538,
      "step": 1516
    },
    {
      "epoch": 0.039952594153278904,
      "grad_norm": 3.88328218460083,
      "learning_rate": 4.80076376086384e-05,
      "loss": 2.2841,
      "step": 1517
    },
    {
      "epoch": 0.03997893073479063,
      "grad_norm": 2.9859044551849365,
      "learning_rate": 4.800632077956281e-05,
      "loss": 0.4016,
      "step": 1518
    },
    {
      "epoch": 0.04000526731630234,
      "grad_norm": 2.992422342300415,
      "learning_rate": 4.8005003950487234e-05,
      "loss": 0.351,
      "step": 1519
    },
    {
      "epoch": 0.040031603897814065,
      "grad_norm": 4.386923313140869,
      "learning_rate": 4.800368712141164e-05,
      "loss": 1.9155,
      "step": 1520
    },
    {
      "epoch": 0.04005794047932578,
      "grad_norm": 8.80577564239502,
      "learning_rate": 4.800237029233606e-05,
      "loss": 1.6105,
      "step": 1521
    },
    {
      "epoch": 0.0400842770608375,
      "grad_norm": 3.1761035919189453,
      "learning_rate": 4.800105346326047e-05,
      "loss": 1.8264,
      "step": 1522
    },
    {
      "epoch": 0.040110613642349226,
      "grad_norm": 2.6750943660736084,
      "learning_rate": 4.799973663418488e-05,
      "loss": 1.7529,
      "step": 1523
    },
    {
      "epoch": 0.04013695022386094,
      "grad_norm": 2.4789552688598633,
      "learning_rate": 4.79984198051093e-05,
      "loss": 2.0503,
      "step": 1524
    },
    {
      "epoch": 0.040163286805372664,
      "grad_norm": 3.246100664138794,
      "learning_rate": 4.7997102976033714e-05,
      "loss": 0.3434,
      "step": 1525
    },
    {
      "epoch": 0.04018962338688438,
      "grad_norm": 2.3975906372070312,
      "learning_rate": 4.799578614695813e-05,
      "loss": 2.0262,
      "step": 1526
    },
    {
      "epoch": 0.0402159599683961,
      "grad_norm": 3.271538257598877,
      "learning_rate": 4.799446931788254e-05,
      "loss": 1.6912,
      "step": 1527
    },
    {
      "epoch": 0.040242296549907824,
      "grad_norm": 2.627981424331665,
      "learning_rate": 4.7993152488806954e-05,
      "loss": 2.229,
      "step": 1528
    },
    {
      "epoch": 0.04026863313141954,
      "grad_norm": 4.80177640914917,
      "learning_rate": 4.799183565973137e-05,
      "loss": 1.9914,
      "step": 1529
    },
    {
      "epoch": 0.04029496971293126,
      "grad_norm": 4.581860065460205,
      "learning_rate": 4.7990518830655785e-05,
      "loss": 0.5321,
      "step": 1530
    },
    {
      "epoch": 0.04032130629444298,
      "grad_norm": 4.275857448577881,
      "learning_rate": 4.7989202001580194e-05,
      "loss": 1.0698,
      "step": 1531
    },
    {
      "epoch": 0.0403476428759547,
      "grad_norm": 3.4956071376800537,
      "learning_rate": 4.798788517250461e-05,
      "loss": 1.1492,
      "step": 1532
    },
    {
      "epoch": 0.04037397945746642,
      "grad_norm": 3.086482524871826,
      "learning_rate": 4.7986568343429025e-05,
      "loss": 1.7993,
      "step": 1533
    },
    {
      "epoch": 0.04040031603897814,
      "grad_norm": 3.3483619689941406,
      "learning_rate": 4.798525151435344e-05,
      "loss": 1.977,
      "step": 1534
    },
    {
      "epoch": 0.04042665262048986,
      "grad_norm": 2.910607099533081,
      "learning_rate": 4.7983934685277857e-05,
      "loss": 1.6934,
      "step": 1535
    },
    {
      "epoch": 0.04045298920200158,
      "grad_norm": 4.313564777374268,
      "learning_rate": 4.7982617856202265e-05,
      "loss": 1.9829,
      "step": 1536
    },
    {
      "epoch": 0.0404793257835133,
      "grad_norm": 3.657001495361328,
      "learning_rate": 4.798130102712668e-05,
      "loss": 2.2356,
      "step": 1537
    },
    {
      "epoch": 0.04050566236502502,
      "grad_norm": 2.117579460144043,
      "learning_rate": 4.79799841980511e-05,
      "loss": 1.7355,
      "step": 1538
    },
    {
      "epoch": 0.04053199894653674,
      "grad_norm": 2.5381734371185303,
      "learning_rate": 4.797866736897551e-05,
      "loss": 2.1239,
      "step": 1539
    },
    {
      "epoch": 0.04055833552804846,
      "grad_norm": 2.609988212585449,
      "learning_rate": 4.797735053989992e-05,
      "loss": 1.7085,
      "step": 1540
    },
    {
      "epoch": 0.040584672109560176,
      "grad_norm": 2.351916551589966,
      "learning_rate": 4.797603371082434e-05,
      "loss": 1.9941,
      "step": 1541
    },
    {
      "epoch": 0.0406110086910719,
      "grad_norm": 4.297755241394043,
      "learning_rate": 4.797471688174875e-05,
      "loss": 1.2975,
      "step": 1542
    },
    {
      "epoch": 0.04063734527258362,
      "grad_norm": 3.406371593475342,
      "learning_rate": 4.797340005267317e-05,
      "loss": 0.6757,
      "step": 1543
    },
    {
      "epoch": 0.040663681854095336,
      "grad_norm": 2.7242140769958496,
      "learning_rate": 4.7972083223597583e-05,
      "loss": 1.4838,
      "step": 1544
    },
    {
      "epoch": 0.04069001843560706,
      "grad_norm": 5.119045734405518,
      "learning_rate": 4.797076639452199e-05,
      "loss": 1.1646,
      "step": 1545
    },
    {
      "epoch": 0.04071635501711878,
      "grad_norm": 2.1951379776000977,
      "learning_rate": 4.796944956544641e-05,
      "loss": 1.6532,
      "step": 1546
    },
    {
      "epoch": 0.0407426915986305,
      "grad_norm": 2.7918548583984375,
      "learning_rate": 4.796813273637082e-05,
      "loss": 1.8081,
      "step": 1547
    },
    {
      "epoch": 0.04076902818014222,
      "grad_norm": 3.408937931060791,
      "learning_rate": 4.796681590729524e-05,
      "loss": 1.7224,
      "step": 1548
    },
    {
      "epoch": 0.040795364761653935,
      "grad_norm": 13.387161254882812,
      "learning_rate": 4.796549907821965e-05,
      "loss": 1.2584,
      "step": 1549
    },
    {
      "epoch": 0.04082170134316566,
      "grad_norm": 2.8224565982818604,
      "learning_rate": 4.7964182249144063e-05,
      "loss": 2.2906,
      "step": 1550
    },
    {
      "epoch": 0.04084803792467738,
      "grad_norm": 1.834579348564148,
      "learning_rate": 4.796286542006848e-05,
      "loss": 2.7615,
      "step": 1551
    },
    {
      "epoch": 0.040874374506189096,
      "grad_norm": 3.280954599380493,
      "learning_rate": 4.7961548590992895e-05,
      "loss": 1.973,
      "step": 1552
    },
    {
      "epoch": 0.04090071108770082,
      "grad_norm": 2.621882438659668,
      "learning_rate": 4.796023176191731e-05,
      "loss": 2.1953,
      "step": 1553
    },
    {
      "epoch": 0.040927047669212534,
      "grad_norm": 2.110182046890259,
      "learning_rate": 4.795891493284172e-05,
      "loss": 1.7225,
      "step": 1554
    },
    {
      "epoch": 0.040953384250724256,
      "grad_norm": 3.0083208084106445,
      "learning_rate": 4.7957598103766135e-05,
      "loss": 0.3899,
      "step": 1555
    },
    {
      "epoch": 0.04097972083223598,
      "grad_norm": 3.3830409049987793,
      "learning_rate": 4.7956281274690543e-05,
      "loss": 1.9343,
      "step": 1556
    },
    {
      "epoch": 0.041006057413747694,
      "grad_norm": 2.7673492431640625,
      "learning_rate": 4.7954964445614966e-05,
      "loss": 2.0526,
      "step": 1557
    },
    {
      "epoch": 0.04103239399525942,
      "grad_norm": 2.359412431716919,
      "learning_rate": 4.7953647616539375e-05,
      "loss": 1.9374,
      "step": 1558
    },
    {
      "epoch": 0.04105873057677113,
      "grad_norm": 2.9968972206115723,
      "learning_rate": 4.795233078746379e-05,
      "loss": 2.271,
      "step": 1559
    },
    {
      "epoch": 0.041085067158282855,
      "grad_norm": 3.117983341217041,
      "learning_rate": 4.79510139583882e-05,
      "loss": 1.6007,
      "step": 1560
    },
    {
      "epoch": 0.04111140373979458,
      "grad_norm": 5.773384094238281,
      "learning_rate": 4.7949697129312615e-05,
      "loss": 1.176,
      "step": 1561
    },
    {
      "epoch": 0.04113774032130629,
      "grad_norm": 2.4290173053741455,
      "learning_rate": 4.794838030023703e-05,
      "loss": 1.1747,
      "step": 1562
    },
    {
      "epoch": 0.041164076902818016,
      "grad_norm": 3.631800651550293,
      "learning_rate": 4.7947063471161446e-05,
      "loss": 1.7565,
      "step": 1563
    },
    {
      "epoch": 0.04119041348432973,
      "grad_norm": 4.197127342224121,
      "learning_rate": 4.794574664208586e-05,
      "loss": 1.7623,
      "step": 1564
    },
    {
      "epoch": 0.041216750065841454,
      "grad_norm": 3.4556450843811035,
      "learning_rate": 4.794442981301027e-05,
      "loss": 1.9954,
      "step": 1565
    },
    {
      "epoch": 0.041243086647353176,
      "grad_norm": 2.8269834518432617,
      "learning_rate": 4.7943112983934686e-05,
      "loss": 2.2221,
      "step": 1566
    },
    {
      "epoch": 0.04126942322886489,
      "grad_norm": 1.908091425895691,
      "learning_rate": 4.79417961548591e-05,
      "loss": 1.8832,
      "step": 1567
    },
    {
      "epoch": 0.041295759810376614,
      "grad_norm": 4.320079326629639,
      "learning_rate": 4.794047932578352e-05,
      "loss": 1.6265,
      "step": 1568
    },
    {
      "epoch": 0.04132209639188833,
      "grad_norm": 3.9409115314483643,
      "learning_rate": 4.7939162496707926e-05,
      "loss": 1.2814,
      "step": 1569
    },
    {
      "epoch": 0.04134843297340005,
      "grad_norm": 3.6473047733306885,
      "learning_rate": 4.793784566763234e-05,
      "loss": 2.4627,
      "step": 1570
    },
    {
      "epoch": 0.041374769554911775,
      "grad_norm": 2.154637575149536,
      "learning_rate": 4.793652883855676e-05,
      "loss": 0.5109,
      "step": 1571
    },
    {
      "epoch": 0.04140110613642349,
      "grad_norm": 2.0151844024658203,
      "learning_rate": 4.793521200948117e-05,
      "loss": 1.7914,
      "step": 1572
    },
    {
      "epoch": 0.04142744271793521,
      "grad_norm": 3.6110353469848633,
      "learning_rate": 4.793389518040559e-05,
      "loss": 1.1011,
      "step": 1573
    },
    {
      "epoch": 0.04145377929944693,
      "grad_norm": 2.5403449535369873,
      "learning_rate": 4.793257835133e-05,
      "loss": 1.3453,
      "step": 1574
    },
    {
      "epoch": 0.04148011588095865,
      "grad_norm": 3.1963884830474854,
      "learning_rate": 4.793126152225441e-05,
      "loss": 0.7116,
      "step": 1575
    },
    {
      "epoch": 0.041506452462470374,
      "grad_norm": 2.1681923866271973,
      "learning_rate": 4.792994469317883e-05,
      "loss": 1.6642,
      "step": 1576
    },
    {
      "epoch": 0.04153278904398209,
      "grad_norm": 2.9085628986358643,
      "learning_rate": 4.7928627864103244e-05,
      "loss": 1.6519,
      "step": 1577
    },
    {
      "epoch": 0.04155912562549381,
      "grad_norm": 2.453711986541748,
      "learning_rate": 4.792731103502765e-05,
      "loss": 1.5192,
      "step": 1578
    },
    {
      "epoch": 0.04158546220700553,
      "grad_norm": 5.075864791870117,
      "learning_rate": 4.792599420595207e-05,
      "loss": 1.5274,
      "step": 1579
    },
    {
      "epoch": 0.04161179878851725,
      "grad_norm": 3.4300835132598877,
      "learning_rate": 4.7924677376876484e-05,
      "loss": 1.7227,
      "step": 1580
    },
    {
      "epoch": 0.04163813537002897,
      "grad_norm": 2.6349151134490967,
      "learning_rate": 4.79233605478009e-05,
      "loss": 1.8592,
      "step": 1581
    },
    {
      "epoch": 0.04166447195154069,
      "grad_norm": 2.22281813621521,
      "learning_rate": 4.7922043718725315e-05,
      "loss": 0.9179,
      "step": 1582
    },
    {
      "epoch": 0.04169080853305241,
      "grad_norm": 2.7420451641082764,
      "learning_rate": 4.7920726889649724e-05,
      "loss": 2.0062,
      "step": 1583
    },
    {
      "epoch": 0.041717145114564126,
      "grad_norm": 2.932339668273926,
      "learning_rate": 4.791941006057414e-05,
      "loss": 2.4226,
      "step": 1584
    },
    {
      "epoch": 0.04174348169607585,
      "grad_norm": 2.3704586029052734,
      "learning_rate": 4.7918093231498555e-05,
      "loss": 2.0831,
      "step": 1585
    },
    {
      "epoch": 0.04176981827758757,
      "grad_norm": 3.4765636920928955,
      "learning_rate": 4.791677640242297e-05,
      "loss": 1.7034,
      "step": 1586
    },
    {
      "epoch": 0.04179615485909929,
      "grad_norm": 4.112740993499756,
      "learning_rate": 4.791545957334738e-05,
      "loss": 0.7328,
      "step": 1587
    },
    {
      "epoch": 0.04182249144061101,
      "grad_norm": 2.626997232437134,
      "learning_rate": 4.7914142744271795e-05,
      "loss": 2.0492,
      "step": 1588
    },
    {
      "epoch": 0.04184882802212273,
      "grad_norm": 6.23048734664917,
      "learning_rate": 4.791282591519621e-05,
      "loss": 1.3451,
      "step": 1589
    },
    {
      "epoch": 0.04187516460363445,
      "grad_norm": 4.811365604400635,
      "learning_rate": 4.7911509086120626e-05,
      "loss": 1.1983,
      "step": 1590
    },
    {
      "epoch": 0.04190150118514617,
      "grad_norm": 2.2768898010253906,
      "learning_rate": 4.791019225704504e-05,
      "loss": 2.4662,
      "step": 1591
    },
    {
      "epoch": 0.041927837766657886,
      "grad_norm": 2.5405027866363525,
      "learning_rate": 4.790887542796945e-05,
      "loss": 2.5304,
      "step": 1592
    },
    {
      "epoch": 0.04195417434816961,
      "grad_norm": 2.4795734882354736,
      "learning_rate": 4.7907558598893866e-05,
      "loss": 2.1591,
      "step": 1593
    },
    {
      "epoch": 0.04198051092968133,
      "grad_norm": 2.5752766132354736,
      "learning_rate": 4.7906241769818275e-05,
      "loss": 2.2811,
      "step": 1594
    },
    {
      "epoch": 0.042006847511193046,
      "grad_norm": 2.638512134552002,
      "learning_rate": 4.79049249407427e-05,
      "loss": 1.8319,
      "step": 1595
    },
    {
      "epoch": 0.04203318409270477,
      "grad_norm": 3.4777679443359375,
      "learning_rate": 4.7903608111667106e-05,
      "loss": 2.272,
      "step": 1596
    },
    {
      "epoch": 0.042059520674216484,
      "grad_norm": 20.70643424987793,
      "learning_rate": 4.790229128259152e-05,
      "loss": 1.9126,
      "step": 1597
    },
    {
      "epoch": 0.04208585725572821,
      "grad_norm": 3.7559597492218018,
      "learning_rate": 4.790097445351594e-05,
      "loss": 0.4809,
      "step": 1598
    },
    {
      "epoch": 0.04211219383723993,
      "grad_norm": 3.2905445098876953,
      "learning_rate": 4.7899657624440346e-05,
      "loss": 1.1684,
      "step": 1599
    },
    {
      "epoch": 0.042138530418751645,
      "grad_norm": 3.089296817779541,
      "learning_rate": 4.789834079536477e-05,
      "loss": 2.2523,
      "step": 1600
    },
    {
      "epoch": 0.04216486700026337,
      "grad_norm": 2.502190589904785,
      "learning_rate": 4.789702396628918e-05,
      "loss": 1.8332,
      "step": 1601
    },
    {
      "epoch": 0.04219120358177508,
      "grad_norm": 2.6924796104431152,
      "learning_rate": 4.789570713721359e-05,
      "loss": 1.8899,
      "step": 1602
    },
    {
      "epoch": 0.042217540163286806,
      "grad_norm": 8.66147232055664,
      "learning_rate": 4.7894390308138e-05,
      "loss": 1.1298,
      "step": 1603
    },
    {
      "epoch": 0.04224387674479853,
      "grad_norm": 3.2306127548217773,
      "learning_rate": 4.7893073479062424e-05,
      "loss": 2.0096,
      "step": 1604
    },
    {
      "epoch": 0.042270213326310244,
      "grad_norm": 2.53001070022583,
      "learning_rate": 4.789175664998683e-05,
      "loss": 1.6158,
      "step": 1605
    },
    {
      "epoch": 0.042296549907821966,
      "grad_norm": 2.410015344619751,
      "learning_rate": 4.789043982091125e-05,
      "loss": 2.1817,
      "step": 1606
    },
    {
      "epoch": 0.04232288648933368,
      "grad_norm": 3.794081926345825,
      "learning_rate": 4.788912299183566e-05,
      "loss": 1.2384,
      "step": 1607
    },
    {
      "epoch": 0.042349223070845404,
      "grad_norm": 2.8532073497772217,
      "learning_rate": 4.788780616276007e-05,
      "loss": 1.6316,
      "step": 1608
    },
    {
      "epoch": 0.04237555965235713,
      "grad_norm": 3.2820916175842285,
      "learning_rate": 4.788648933368449e-05,
      "loss": 1.4657,
      "step": 1609
    },
    {
      "epoch": 0.04240189623386884,
      "grad_norm": 2.917919397354126,
      "learning_rate": 4.7885172504608904e-05,
      "loss": 1.5911,
      "step": 1610
    },
    {
      "epoch": 0.042428232815380565,
      "grad_norm": 2.8589375019073486,
      "learning_rate": 4.788385567553332e-05,
      "loss": 1.4094,
      "step": 1611
    },
    {
      "epoch": 0.04245456939689228,
      "grad_norm": 2.2336888313293457,
      "learning_rate": 4.788253884645773e-05,
      "loss": 1.4805,
      "step": 1612
    },
    {
      "epoch": 0.042480905978404,
      "grad_norm": 1.9819060564041138,
      "learning_rate": 4.7881222017382144e-05,
      "loss": 1.9223,
      "step": 1613
    },
    {
      "epoch": 0.042507242559915726,
      "grad_norm": 3.8112151622772217,
      "learning_rate": 4.787990518830656e-05,
      "loss": 1.6173,
      "step": 1614
    },
    {
      "epoch": 0.04253357914142744,
      "grad_norm": 4.478504657745361,
      "learning_rate": 4.7878588359230976e-05,
      "loss": 2.2684,
      "step": 1615
    },
    {
      "epoch": 0.042559915722939164,
      "grad_norm": 5.168664932250977,
      "learning_rate": 4.7877271530155384e-05,
      "loss": 0.9267,
      "step": 1616
    },
    {
      "epoch": 0.04258625230445088,
      "grad_norm": 2.1900522708892822,
      "learning_rate": 4.78759547010798e-05,
      "loss": 1.5993,
      "step": 1617
    },
    {
      "epoch": 0.0426125888859626,
      "grad_norm": 2.6770801544189453,
      "learning_rate": 4.7874637872004216e-05,
      "loss": 1.8449,
      "step": 1618
    },
    {
      "epoch": 0.042638925467474324,
      "grad_norm": 3.2104368209838867,
      "learning_rate": 4.787332104292863e-05,
      "loss": 1.2252,
      "step": 1619
    },
    {
      "epoch": 0.04266526204898604,
      "grad_norm": 2.747957706451416,
      "learning_rate": 4.787200421385305e-05,
      "loss": 2.3908,
      "step": 1620
    },
    {
      "epoch": 0.04269159863049776,
      "grad_norm": 6.097873687744141,
      "learning_rate": 4.7870687384777456e-05,
      "loss": 1.9629,
      "step": 1621
    },
    {
      "epoch": 0.04271793521200948,
      "grad_norm": 2.296062469482422,
      "learning_rate": 4.786937055570187e-05,
      "loss": 2.1889,
      "step": 1622
    },
    {
      "epoch": 0.0427442717935212,
      "grad_norm": 3.2836878299713135,
      "learning_rate": 4.786805372662629e-05,
      "loss": 0.7493,
      "step": 1623
    },
    {
      "epoch": 0.04277060837503292,
      "grad_norm": 2.845104455947876,
      "learning_rate": 4.78667368975507e-05,
      "loss": 2.0775,
      "step": 1624
    },
    {
      "epoch": 0.04279694495654464,
      "grad_norm": 2.390986680984497,
      "learning_rate": 4.786542006847511e-05,
      "loss": 1.8683,
      "step": 1625
    },
    {
      "epoch": 0.04282328153805636,
      "grad_norm": 4.470747470855713,
      "learning_rate": 4.786410323939953e-05,
      "loss": 1.2525,
      "step": 1626
    },
    {
      "epoch": 0.04284961811956808,
      "grad_norm": 2.0850131511688232,
      "learning_rate": 4.786278641032394e-05,
      "loss": 2.0149,
      "step": 1627
    },
    {
      "epoch": 0.0428759547010798,
      "grad_norm": 2.494018077850342,
      "learning_rate": 4.786146958124836e-05,
      "loss": 1.3731,
      "step": 1628
    },
    {
      "epoch": 0.04290229128259152,
      "grad_norm": 3.647768974304199,
      "learning_rate": 4.7860152752172774e-05,
      "loss": 1.2123,
      "step": 1629
    },
    {
      "epoch": 0.04292862786410324,
      "grad_norm": 2.676943302154541,
      "learning_rate": 4.785883592309718e-05,
      "loss": 1.5728,
      "step": 1630
    },
    {
      "epoch": 0.04295496444561496,
      "grad_norm": 1.8378205299377441,
      "learning_rate": 4.78575190940216e-05,
      "loss": 0.6929,
      "step": 1631
    },
    {
      "epoch": 0.04298130102712668,
      "grad_norm": 3.5674049854278564,
      "learning_rate": 4.785620226494601e-05,
      "loss": 1.3228,
      "step": 1632
    },
    {
      "epoch": 0.0430076376086384,
      "grad_norm": 2.0570549964904785,
      "learning_rate": 4.785488543587043e-05,
      "loss": 1.7695,
      "step": 1633
    },
    {
      "epoch": 0.04303397419015012,
      "grad_norm": 4.9037370681762695,
      "learning_rate": 4.785356860679484e-05,
      "loss": 1.747,
      "step": 1634
    },
    {
      "epoch": 0.043060310771661836,
      "grad_norm": 3.769916534423828,
      "learning_rate": 4.7852251777719254e-05,
      "loss": 1.6977,
      "step": 1635
    },
    {
      "epoch": 0.04308664735317356,
      "grad_norm": 2.68953800201416,
      "learning_rate": 4.785093494864367e-05,
      "loss": 0.4149,
      "step": 1636
    },
    {
      "epoch": 0.04311298393468528,
      "grad_norm": 4.373473644256592,
      "learning_rate": 4.7849618119568085e-05,
      "loss": 1.1404,
      "step": 1637
    },
    {
      "epoch": 0.043139320516197,
      "grad_norm": 7.387909412384033,
      "learning_rate": 4.78483012904925e-05,
      "loss": 1.8211,
      "step": 1638
    },
    {
      "epoch": 0.04316565709770872,
      "grad_norm": 3.6866443157196045,
      "learning_rate": 4.784698446141691e-05,
      "loss": 1.7778,
      "step": 1639
    },
    {
      "epoch": 0.043191993679220435,
      "grad_norm": 3.591341018676758,
      "learning_rate": 4.7845667632341325e-05,
      "loss": 1.6675,
      "step": 1640
    },
    {
      "epoch": 0.04321833026073216,
      "grad_norm": 3.440986394882202,
      "learning_rate": 4.7844350803265734e-05,
      "loss": 1.5045,
      "step": 1641
    },
    {
      "epoch": 0.04324466684224388,
      "grad_norm": 3.4958741664886475,
      "learning_rate": 4.7843033974190156e-05,
      "loss": 2.2951,
      "step": 1642
    },
    {
      "epoch": 0.043271003423755595,
      "grad_norm": 2.1908135414123535,
      "learning_rate": 4.7841717145114565e-05,
      "loss": 1.3439,
      "step": 1643
    },
    {
      "epoch": 0.04329734000526732,
      "grad_norm": 2.8393232822418213,
      "learning_rate": 4.784040031603898e-05,
      "loss": 1.4972,
      "step": 1644
    },
    {
      "epoch": 0.043323676586779034,
      "grad_norm": 2.38371205329895,
      "learning_rate": 4.7839083486963396e-05,
      "loss": 1.8012,
      "step": 1645
    },
    {
      "epoch": 0.043350013168290756,
      "grad_norm": 2.383183002471924,
      "learning_rate": 4.7837766657887805e-05,
      "loss": 1.8169,
      "step": 1646
    },
    {
      "epoch": 0.04337634974980248,
      "grad_norm": 4.216830730438232,
      "learning_rate": 4.783644982881223e-05,
      "loss": 2.4733,
      "step": 1647
    },
    {
      "epoch": 0.043402686331314194,
      "grad_norm": 3.641187906265259,
      "learning_rate": 4.7835132999736636e-05,
      "loss": 2.337,
      "step": 1648
    },
    {
      "epoch": 0.04342902291282592,
      "grad_norm": 3.450556993484497,
      "learning_rate": 4.783381617066105e-05,
      "loss": 0.6939,
      "step": 1649
    },
    {
      "epoch": 0.04345535949433763,
      "grad_norm": 3.0492610931396484,
      "learning_rate": 4.783249934158546e-05,
      "loss": 0.5371,
      "step": 1650
    },
    {
      "epoch": 0.043481696075849355,
      "grad_norm": 4.064547061920166,
      "learning_rate": 4.783118251250988e-05,
      "loss": 0.7245,
      "step": 1651
    },
    {
      "epoch": 0.04350803265736108,
      "grad_norm": 3.24357271194458,
      "learning_rate": 4.782986568343429e-05,
      "loss": 1.5819,
      "step": 1652
    },
    {
      "epoch": 0.04353436923887279,
      "grad_norm": 2.13814640045166,
      "learning_rate": 4.782854885435871e-05,
      "loss": 1.7686,
      "step": 1653
    },
    {
      "epoch": 0.043560705820384515,
      "grad_norm": 2.536398410797119,
      "learning_rate": 4.782723202528312e-05,
      "loss": 1.4857,
      "step": 1654
    },
    {
      "epoch": 0.04358704240189623,
      "grad_norm": 4.551061630249023,
      "learning_rate": 4.782591519620753e-05,
      "loss": 1.4803,
      "step": 1655
    },
    {
      "epoch": 0.043613378983407954,
      "grad_norm": 2.0763628482818604,
      "learning_rate": 4.7824598367131954e-05,
      "loss": 2.0751,
      "step": 1656
    },
    {
      "epoch": 0.043639715564919676,
      "grad_norm": 1.9243184328079224,
      "learning_rate": 4.782328153805636e-05,
      "loss": 1.5623,
      "step": 1657
    },
    {
      "epoch": 0.04366605214643139,
      "grad_norm": 3.0155081748962402,
      "learning_rate": 4.782196470898078e-05,
      "loss": 2.1019,
      "step": 1658
    },
    {
      "epoch": 0.043692388727943114,
      "grad_norm": 3.214911460876465,
      "learning_rate": 4.782064787990519e-05,
      "loss": 1.6881,
      "step": 1659
    },
    {
      "epoch": 0.04371872530945483,
      "grad_norm": 2.7685279846191406,
      "learning_rate": 4.78193310508296e-05,
      "loss": 2.7239,
      "step": 1660
    },
    {
      "epoch": 0.04374506189096655,
      "grad_norm": 4.537632942199707,
      "learning_rate": 4.781801422175402e-05,
      "loss": 1.7678,
      "step": 1661
    },
    {
      "epoch": 0.043771398472478275,
      "grad_norm": 3.539085626602173,
      "learning_rate": 4.7816697392678434e-05,
      "loss": 1.7179,
      "step": 1662
    },
    {
      "epoch": 0.04379773505398999,
      "grad_norm": 3.8898887634277344,
      "learning_rate": 4.781538056360284e-05,
      "loss": 0.8136,
      "step": 1663
    },
    {
      "epoch": 0.04382407163550171,
      "grad_norm": 6.164101600646973,
      "learning_rate": 4.781406373452726e-05,
      "loss": 2.2738,
      "step": 1664
    },
    {
      "epoch": 0.04385040821701343,
      "grad_norm": 3.5677146911621094,
      "learning_rate": 4.7812746905451674e-05,
      "loss": 1.68,
      "step": 1665
    },
    {
      "epoch": 0.04387674479852515,
      "grad_norm": 2.4857919216156006,
      "learning_rate": 4.781143007637609e-05,
      "loss": 0.4742,
      "step": 1666
    },
    {
      "epoch": 0.043903081380036874,
      "grad_norm": 2.818009376525879,
      "learning_rate": 4.7810113247300505e-05,
      "loss": 1.6659,
      "step": 1667
    },
    {
      "epoch": 0.04392941796154859,
      "grad_norm": 2.3224613666534424,
      "learning_rate": 4.7808796418224914e-05,
      "loss": 1.6821,
      "step": 1668
    },
    {
      "epoch": 0.04395575454306031,
      "grad_norm": 8.352319717407227,
      "learning_rate": 4.780747958914933e-05,
      "loss": 1.7879,
      "step": 1669
    },
    {
      "epoch": 0.04398209112457203,
      "grad_norm": 3.1340174674987793,
      "learning_rate": 4.7806162760073745e-05,
      "loss": 1.6865,
      "step": 1670
    },
    {
      "epoch": 0.04400842770608375,
      "grad_norm": 3.5655996799468994,
      "learning_rate": 4.780484593099816e-05,
      "loss": 1.8297,
      "step": 1671
    },
    {
      "epoch": 0.04403476428759547,
      "grad_norm": 2.270786762237549,
      "learning_rate": 4.780352910192257e-05,
      "loss": 2.0561,
      "step": 1672
    },
    {
      "epoch": 0.04406110086910719,
      "grad_norm": 3.1496076583862305,
      "learning_rate": 4.7802212272846985e-05,
      "loss": 1.2317,
      "step": 1673
    },
    {
      "epoch": 0.04408743745061891,
      "grad_norm": 6.753971576690674,
      "learning_rate": 4.78008954437714e-05,
      "loss": 1.1494,
      "step": 1674
    },
    {
      "epoch": 0.044113774032130626,
      "grad_norm": 3.142165422439575,
      "learning_rate": 4.779957861469582e-05,
      "loss": 1.7838,
      "step": 1675
    },
    {
      "epoch": 0.04414011061364235,
      "grad_norm": 6.98069429397583,
      "learning_rate": 4.779826178562023e-05,
      "loss": 2.4283,
      "step": 1676
    },
    {
      "epoch": 0.04416644719515407,
      "grad_norm": 3.2393054962158203,
      "learning_rate": 4.779694495654464e-05,
      "loss": 1.7799,
      "step": 1677
    },
    {
      "epoch": 0.04419278377666579,
      "grad_norm": 3.4538471698760986,
      "learning_rate": 4.779562812746906e-05,
      "loss": 2.1511,
      "step": 1678
    },
    {
      "epoch": 0.04421912035817751,
      "grad_norm": 2.1494596004486084,
      "learning_rate": 4.7794311298393466e-05,
      "loss": 1.5318,
      "step": 1679
    },
    {
      "epoch": 0.04424545693968923,
      "grad_norm": 3.0062994956970215,
      "learning_rate": 4.779299446931789e-05,
      "loss": 2.2842,
      "step": 1680
    },
    {
      "epoch": 0.04427179352120095,
      "grad_norm": 2.5928919315338135,
      "learning_rate": 4.77916776402423e-05,
      "loss": 2.4077,
      "step": 1681
    },
    {
      "epoch": 0.04429813010271267,
      "grad_norm": 3.13175368309021,
      "learning_rate": 4.779036081116671e-05,
      "loss": 2.0801,
      "step": 1682
    },
    {
      "epoch": 0.044324466684224385,
      "grad_norm": 2.3422467708587646,
      "learning_rate": 4.778904398209113e-05,
      "loss": 2.3598,
      "step": 1683
    },
    {
      "epoch": 0.04435080326573611,
      "grad_norm": 3.7321741580963135,
      "learning_rate": 4.7787727153015543e-05,
      "loss": 1.4985,
      "step": 1684
    },
    {
      "epoch": 0.04437713984724783,
      "grad_norm": 3.111586570739746,
      "learning_rate": 4.778641032393996e-05,
      "loss": 0.7241,
      "step": 1685
    },
    {
      "epoch": 0.044403476428759546,
      "grad_norm": 3.2628390789031982,
      "learning_rate": 4.778509349486437e-05,
      "loss": 0.9041,
      "step": 1686
    },
    {
      "epoch": 0.04442981301027127,
      "grad_norm": 2.896920680999756,
      "learning_rate": 4.7783776665788784e-05,
      "loss": 2.3112,
      "step": 1687
    },
    {
      "epoch": 0.044456149591782984,
      "grad_norm": 2.1524009704589844,
      "learning_rate": 4.778245983671319e-05,
      "loss": 1.6757,
      "step": 1688
    },
    {
      "epoch": 0.04448248617329471,
      "grad_norm": 4.034472465515137,
      "learning_rate": 4.7781143007637615e-05,
      "loss": 1.7498,
      "step": 1689
    },
    {
      "epoch": 0.04450882275480643,
      "grad_norm": 2.1806130409240723,
      "learning_rate": 4.7779826178562024e-05,
      "loss": 1.2612,
      "step": 1690
    },
    {
      "epoch": 0.044535159336318145,
      "grad_norm": 2.5430376529693604,
      "learning_rate": 4.777850934948644e-05,
      "loss": 1.7788,
      "step": 1691
    },
    {
      "epoch": 0.04456149591782987,
      "grad_norm": 2.6140480041503906,
      "learning_rate": 4.7777192520410855e-05,
      "loss": 0.7482,
      "step": 1692
    },
    {
      "epoch": 0.04458783249934158,
      "grad_norm": 3.0158402919769287,
      "learning_rate": 4.7775875691335264e-05,
      "loss": 1.2215,
      "step": 1693
    },
    {
      "epoch": 0.044614169080853305,
      "grad_norm": 1.6862404346466064,
      "learning_rate": 4.7774558862259686e-05,
      "loss": 1.801,
      "step": 1694
    },
    {
      "epoch": 0.04464050566236503,
      "grad_norm": 7.785438537597656,
      "learning_rate": 4.7773242033184095e-05,
      "loss": 0.9266,
      "step": 1695
    },
    {
      "epoch": 0.04466684224387674,
      "grad_norm": 4.126392841339111,
      "learning_rate": 4.777192520410851e-05,
      "loss": 0.8319,
      "step": 1696
    },
    {
      "epoch": 0.044693178825388466,
      "grad_norm": 2.515692949295044,
      "learning_rate": 4.777060837503292e-05,
      "loss": 1.9474,
      "step": 1697
    },
    {
      "epoch": 0.04471951540690018,
      "grad_norm": 3.4862043857574463,
      "learning_rate": 4.776929154595734e-05,
      "loss": 0.9102,
      "step": 1698
    },
    {
      "epoch": 0.044745851988411904,
      "grad_norm": 5.722166061401367,
      "learning_rate": 4.776797471688175e-05,
      "loss": 2.1326,
      "step": 1699
    },
    {
      "epoch": 0.04477218856992363,
      "grad_norm": 2.5878779888153076,
      "learning_rate": 4.7766657887806166e-05,
      "loss": 1.6677,
      "step": 1700
    },
    {
      "epoch": 0.04479852515143534,
      "grad_norm": 3.3507132530212402,
      "learning_rate": 4.776534105873058e-05,
      "loss": 1.5343,
      "step": 1701
    },
    {
      "epoch": 0.044824861732947065,
      "grad_norm": 2.779021978378296,
      "learning_rate": 4.776402422965499e-05,
      "loss": 1.5199,
      "step": 1702
    },
    {
      "epoch": 0.04485119831445878,
      "grad_norm": 3.0135865211486816,
      "learning_rate": 4.776270740057941e-05,
      "loss": 0.6944,
      "step": 1703
    },
    {
      "epoch": 0.0448775348959705,
      "grad_norm": 3.047968626022339,
      "learning_rate": 4.776139057150382e-05,
      "loss": 1.4053,
      "step": 1704
    },
    {
      "epoch": 0.044903871477482225,
      "grad_norm": 3.3515453338623047,
      "learning_rate": 4.776007374242824e-05,
      "loss": 1.9602,
      "step": 1705
    },
    {
      "epoch": 0.04493020805899394,
      "grad_norm": 7.582807540893555,
      "learning_rate": 4.7758756913352646e-05,
      "loss": 1.8265,
      "step": 1706
    },
    {
      "epoch": 0.04495654464050566,
      "grad_norm": 2.6268017292022705,
      "learning_rate": 4.775744008427706e-05,
      "loss": 1.9812,
      "step": 1707
    },
    {
      "epoch": 0.04498288122201738,
      "grad_norm": 2.0065243244171143,
      "learning_rate": 4.775612325520148e-05,
      "loss": 1.7819,
      "step": 1708
    },
    {
      "epoch": 0.0450092178035291,
      "grad_norm": 4.183550834655762,
      "learning_rate": 4.775480642612589e-05,
      "loss": 1.3637,
      "step": 1709
    },
    {
      "epoch": 0.045035554385040824,
      "grad_norm": 6.974421977996826,
      "learning_rate": 4.77534895970503e-05,
      "loss": 1.2507,
      "step": 1710
    },
    {
      "epoch": 0.04506189096655254,
      "grad_norm": 2.9028408527374268,
      "learning_rate": 4.775217276797472e-05,
      "loss": 1.9522,
      "step": 1711
    },
    {
      "epoch": 0.04508822754806426,
      "grad_norm": 2.099782943725586,
      "learning_rate": 4.775085593889913e-05,
      "loss": 1.7498,
      "step": 1712
    },
    {
      "epoch": 0.04511456412957598,
      "grad_norm": 3.4039838314056396,
      "learning_rate": 4.774953910982355e-05,
      "loss": 1.1261,
      "step": 1713
    },
    {
      "epoch": 0.0451409007110877,
      "grad_norm": 3.148505449295044,
      "learning_rate": 4.7748222280747964e-05,
      "loss": 2.0884,
      "step": 1714
    },
    {
      "epoch": 0.04516723729259942,
      "grad_norm": 4.304432392120361,
      "learning_rate": 4.774690545167237e-05,
      "loss": 1.6005,
      "step": 1715
    },
    {
      "epoch": 0.04519357387411114,
      "grad_norm": 2.3728487491607666,
      "learning_rate": 4.774558862259679e-05,
      "loss": 1.8499,
      "step": 1716
    },
    {
      "epoch": 0.04521991045562286,
      "grad_norm": 2.966421604156494,
      "learning_rate": 4.7744271793521204e-05,
      "loss": 1.3817,
      "step": 1717
    },
    {
      "epoch": 0.045246247037134577,
      "grad_norm": 2.327455759048462,
      "learning_rate": 4.774295496444562e-05,
      "loss": 1.7435,
      "step": 1718
    },
    {
      "epoch": 0.0452725836186463,
      "grad_norm": 2.3098363876342773,
      "learning_rate": 4.774163813537003e-05,
      "loss": 1.8852,
      "step": 1719
    },
    {
      "epoch": 0.04529892020015802,
      "grad_norm": 2.8407859802246094,
      "learning_rate": 4.7740321306294444e-05,
      "loss": 2.2121,
      "step": 1720
    },
    {
      "epoch": 0.04532525678166974,
      "grad_norm": 2.419339418411255,
      "learning_rate": 4.773900447721886e-05,
      "loss": 1.8823,
      "step": 1721
    },
    {
      "epoch": 0.04535159336318146,
      "grad_norm": 2.6375315189361572,
      "learning_rate": 4.7737687648143275e-05,
      "loss": 1.795,
      "step": 1722
    },
    {
      "epoch": 0.04537792994469318,
      "grad_norm": 2.089695692062378,
      "learning_rate": 4.773637081906769e-05,
      "loss": 1.5587,
      "step": 1723
    },
    {
      "epoch": 0.0454042665262049,
      "grad_norm": 2.309267044067383,
      "learning_rate": 4.77350539899921e-05,
      "loss": 1.6003,
      "step": 1724
    },
    {
      "epoch": 0.04543060310771662,
      "grad_norm": 2.6133782863616943,
      "learning_rate": 4.7733737160916515e-05,
      "loss": 1.8045,
      "step": 1725
    },
    {
      "epoch": 0.045456939689228336,
      "grad_norm": 2.815037727355957,
      "learning_rate": 4.7732420331840924e-05,
      "loss": 1.9914,
      "step": 1726
    },
    {
      "epoch": 0.04548327627074006,
      "grad_norm": 4.499743461608887,
      "learning_rate": 4.7731103502765346e-05,
      "loss": 1.8553,
      "step": 1727
    },
    {
      "epoch": 0.04550961285225178,
      "grad_norm": 2.416105270385742,
      "learning_rate": 4.7729786673689755e-05,
      "loss": 2.1857,
      "step": 1728
    },
    {
      "epoch": 0.045535949433763497,
      "grad_norm": 2.986762523651123,
      "learning_rate": 4.772846984461417e-05,
      "loss": 1.8378,
      "step": 1729
    },
    {
      "epoch": 0.04556228601527522,
      "grad_norm": 3.844364881515503,
      "learning_rate": 4.7727153015538586e-05,
      "loss": 2.5077,
      "step": 1730
    },
    {
      "epoch": 0.045588622596786935,
      "grad_norm": 2.905626058578491,
      "learning_rate": 4.7725836186463e-05,
      "loss": 1.5804,
      "step": 1731
    },
    {
      "epoch": 0.04561495917829866,
      "grad_norm": 3.7784745693206787,
      "learning_rate": 4.772451935738742e-05,
      "loss": 1.1766,
      "step": 1732
    },
    {
      "epoch": 0.04564129575981038,
      "grad_norm": 2.265690565109253,
      "learning_rate": 4.7723202528311826e-05,
      "loss": 1.9701,
      "step": 1733
    },
    {
      "epoch": 0.045667632341322095,
      "grad_norm": 4.532913684844971,
      "learning_rate": 4.772188569923624e-05,
      "loss": 1.8431,
      "step": 1734
    },
    {
      "epoch": 0.04569396892283382,
      "grad_norm": 1.9257761240005493,
      "learning_rate": 4.772056887016065e-05,
      "loss": 1.9792,
      "step": 1735
    },
    {
      "epoch": 0.04572030550434553,
      "grad_norm": 2.6322133541107178,
      "learning_rate": 4.771925204108507e-05,
      "loss": 1.5659,
      "step": 1736
    },
    {
      "epoch": 0.045746642085857256,
      "grad_norm": 2.7895565032958984,
      "learning_rate": 4.771793521200948e-05,
      "loss": 1.7882,
      "step": 1737
    },
    {
      "epoch": 0.04577297866736898,
      "grad_norm": 2.231600046157837,
      "learning_rate": 4.77166183829339e-05,
      "loss": 1.8305,
      "step": 1738
    },
    {
      "epoch": 0.045799315248880694,
      "grad_norm": 3.5905938148498535,
      "learning_rate": 4.771530155385831e-05,
      "loss": 1.2803,
      "step": 1739
    },
    {
      "epoch": 0.045825651830392417,
      "grad_norm": 1.8099578619003296,
      "learning_rate": 4.771398472478272e-05,
      "loss": 0.4731,
      "step": 1740
    },
    {
      "epoch": 0.04585198841190413,
      "grad_norm": 3.470010280609131,
      "learning_rate": 4.7712667895707144e-05,
      "loss": 2.1015,
      "step": 1741
    },
    {
      "epoch": 0.045878324993415855,
      "grad_norm": 3.340970516204834,
      "learning_rate": 4.771135106663155e-05,
      "loss": 1.8298,
      "step": 1742
    },
    {
      "epoch": 0.04590466157492758,
      "grad_norm": 2.5090110301971436,
      "learning_rate": 4.771003423755597e-05,
      "loss": 2.4194,
      "step": 1743
    },
    {
      "epoch": 0.04593099815643929,
      "grad_norm": 2.6823501586914062,
      "learning_rate": 4.770871740848038e-05,
      "loss": 0.5867,
      "step": 1744
    },
    {
      "epoch": 0.045957334737951015,
      "grad_norm": 2.7022886276245117,
      "learning_rate": 4.770740057940479e-05,
      "loss": 1.8058,
      "step": 1745
    },
    {
      "epoch": 0.04598367131946273,
      "grad_norm": 2.5525381565093994,
      "learning_rate": 4.770608375032921e-05,
      "loss": 2.1268,
      "step": 1746
    },
    {
      "epoch": 0.04601000790097445,
      "grad_norm": 2.399473190307617,
      "learning_rate": 4.7704766921253625e-05,
      "loss": 2.1018,
      "step": 1747
    },
    {
      "epoch": 0.046036344482486176,
      "grad_norm": 3.300828456878662,
      "learning_rate": 4.770345009217804e-05,
      "loss": 1.9747,
      "step": 1748
    },
    {
      "epoch": 0.04606268106399789,
      "grad_norm": 3.060251235961914,
      "learning_rate": 4.770213326310245e-05,
      "loss": 1.6118,
      "step": 1749
    },
    {
      "epoch": 0.046089017645509614,
      "grad_norm": 4.1365227699279785,
      "learning_rate": 4.770081643402687e-05,
      "loss": 1.4013,
      "step": 1750
    },
    {
      "epoch": 0.04611535422702133,
      "grad_norm": 2.2801895141601562,
      "learning_rate": 4.769949960495128e-05,
      "loss": 1.9977,
      "step": 1751
    },
    {
      "epoch": 0.04614169080853305,
      "grad_norm": 2.8356292247772217,
      "learning_rate": 4.7698182775875696e-05,
      "loss": 1.7156,
      "step": 1752
    },
    {
      "epoch": 0.046168027390044775,
      "grad_norm": 4.43156623840332,
      "learning_rate": 4.7696865946800105e-05,
      "loss": 1.5122,
      "step": 1753
    },
    {
      "epoch": 0.04619436397155649,
      "grad_norm": 1.935263752937317,
      "learning_rate": 4.769554911772452e-05,
      "loss": 1.9238,
      "step": 1754
    },
    {
      "epoch": 0.04622070055306821,
      "grad_norm": 2.5804927349090576,
      "learning_rate": 4.7694232288648936e-05,
      "loss": 1.6656,
      "step": 1755
    },
    {
      "epoch": 0.04624703713457993,
      "grad_norm": 2.096917152404785,
      "learning_rate": 4.769291545957335e-05,
      "loss": 2.3052,
      "step": 1756
    },
    {
      "epoch": 0.04627337371609165,
      "grad_norm": 3.051651954650879,
      "learning_rate": 4.769159863049777e-05,
      "loss": 2.1008,
      "step": 1757
    },
    {
      "epoch": 0.04629971029760337,
      "grad_norm": 2.965366840362549,
      "learning_rate": 4.7690281801422176e-05,
      "loss": 1.3892,
      "step": 1758
    },
    {
      "epoch": 0.04632604687911509,
      "grad_norm": 2.560187578201294,
      "learning_rate": 4.768896497234659e-05,
      "loss": 1.3878,
      "step": 1759
    },
    {
      "epoch": 0.04635238346062681,
      "grad_norm": 3.776427984237671,
      "learning_rate": 4.768764814327101e-05,
      "loss": 2.7926,
      "step": 1760
    },
    {
      "epoch": 0.04637872004213853,
      "grad_norm": 3.302241802215576,
      "learning_rate": 4.768633131419542e-05,
      "loss": 1.3911,
      "step": 1761
    },
    {
      "epoch": 0.04640505662365025,
      "grad_norm": 2.7292490005493164,
      "learning_rate": 4.768501448511983e-05,
      "loss": 1.3793,
      "step": 1762
    },
    {
      "epoch": 0.04643139320516197,
      "grad_norm": 5.511817932128906,
      "learning_rate": 4.768369765604425e-05,
      "loss": 1.7398,
      "step": 1763
    },
    {
      "epoch": 0.04645772978667369,
      "grad_norm": 3.4631576538085938,
      "learning_rate": 4.768238082696866e-05,
      "loss": 1.5239,
      "step": 1764
    },
    {
      "epoch": 0.04648406636818541,
      "grad_norm": 1.8677030801773071,
      "learning_rate": 4.768106399789308e-05,
      "loss": 1.3443,
      "step": 1765
    },
    {
      "epoch": 0.04651040294969713,
      "grad_norm": 4.334285736083984,
      "learning_rate": 4.767974716881749e-05,
      "loss": 1.2896,
      "step": 1766
    },
    {
      "epoch": 0.04653673953120885,
      "grad_norm": 2.226372718811035,
      "learning_rate": 4.76784303397419e-05,
      "loss": 1.3587,
      "step": 1767
    },
    {
      "epoch": 0.04656307611272057,
      "grad_norm": 2.619678497314453,
      "learning_rate": 4.767711351066632e-05,
      "loss": 1.237,
      "step": 1768
    },
    {
      "epoch": 0.046589412694232286,
      "grad_norm": 3.8238658905029297,
      "learning_rate": 4.7675796681590734e-05,
      "loss": 1.3064,
      "step": 1769
    },
    {
      "epoch": 0.04661574927574401,
      "grad_norm": 4.150592803955078,
      "learning_rate": 4.767447985251515e-05,
      "loss": 1.5192,
      "step": 1770
    },
    {
      "epoch": 0.04664208585725573,
      "grad_norm": 3.409359931945801,
      "learning_rate": 4.767316302343956e-05,
      "loss": 1.2098,
      "step": 1771
    },
    {
      "epoch": 0.04666842243876745,
      "grad_norm": 2.423478126525879,
      "learning_rate": 4.7671846194363974e-05,
      "loss": 2.569,
      "step": 1772
    },
    {
      "epoch": 0.04669475902027917,
      "grad_norm": 3.326936960220337,
      "learning_rate": 4.767052936528838e-05,
      "loss": 2.1252,
      "step": 1773
    },
    {
      "epoch": 0.046721095601790885,
      "grad_norm": 2.609152317047119,
      "learning_rate": 4.7669212536212805e-05,
      "loss": 1.7812,
      "step": 1774
    },
    {
      "epoch": 0.04674743218330261,
      "grad_norm": 9.348191261291504,
      "learning_rate": 4.7667895707137214e-05,
      "loss": 0.5157,
      "step": 1775
    },
    {
      "epoch": 0.04677376876481433,
      "grad_norm": 3.1011111736297607,
      "learning_rate": 4.766657887806163e-05,
      "loss": 2.4706,
      "step": 1776
    },
    {
      "epoch": 0.046800105346326046,
      "grad_norm": 3.4174981117248535,
      "learning_rate": 4.7665262048986045e-05,
      "loss": 1.6857,
      "step": 1777
    },
    {
      "epoch": 0.04682644192783777,
      "grad_norm": 5.005181789398193,
      "learning_rate": 4.7663945219910454e-05,
      "loss": 1.5254,
      "step": 1778
    },
    {
      "epoch": 0.046852778509349484,
      "grad_norm": 2.7576558589935303,
      "learning_rate": 4.7662628390834876e-05,
      "loss": 1.1868,
      "step": 1779
    },
    {
      "epoch": 0.046879115090861206,
      "grad_norm": 3.931638240814209,
      "learning_rate": 4.7661311561759285e-05,
      "loss": 2.2186,
      "step": 1780
    },
    {
      "epoch": 0.04690545167237293,
      "grad_norm": 1.7449153661727905,
      "learning_rate": 4.76599947326837e-05,
      "loss": 2.1682,
      "step": 1781
    },
    {
      "epoch": 0.046931788253884645,
      "grad_norm": 2.6958789825439453,
      "learning_rate": 4.765867790360811e-05,
      "loss": 2.0851,
      "step": 1782
    },
    {
      "epoch": 0.04695812483539637,
      "grad_norm": 4.813263893127441,
      "learning_rate": 4.765736107453253e-05,
      "loss": 1.0686,
      "step": 1783
    },
    {
      "epoch": 0.04698446141690808,
      "grad_norm": 2.0777738094329834,
      "learning_rate": 4.765604424545694e-05,
      "loss": 2.0219,
      "step": 1784
    },
    {
      "epoch": 0.047010797998419805,
      "grad_norm": 3.966855525970459,
      "learning_rate": 4.7654727416381356e-05,
      "loss": 1.7394,
      "step": 1785
    },
    {
      "epoch": 0.04703713457993153,
      "grad_norm": 2.7082858085632324,
      "learning_rate": 4.765341058730577e-05,
      "loss": 1.7643,
      "step": 1786
    },
    {
      "epoch": 0.04706347116144324,
      "grad_norm": 2.816910982131958,
      "learning_rate": 4.765209375823018e-05,
      "loss": 1.9363,
      "step": 1787
    },
    {
      "epoch": 0.047089807742954966,
      "grad_norm": 2.363173246383667,
      "learning_rate": 4.76507769291546e-05,
      "loss": 1.8564,
      "step": 1788
    },
    {
      "epoch": 0.04711614432446668,
      "grad_norm": 2.4330899715423584,
      "learning_rate": 4.764946010007901e-05,
      "loss": 2.0441,
      "step": 1789
    },
    {
      "epoch": 0.047142480905978404,
      "grad_norm": 2.9492557048797607,
      "learning_rate": 4.764814327100343e-05,
      "loss": 2.008,
      "step": 1790
    },
    {
      "epoch": 0.047168817487490126,
      "grad_norm": 6.210226058959961,
      "learning_rate": 4.7646826441927836e-05,
      "loss": 1.1207,
      "step": 1791
    },
    {
      "epoch": 0.04719515406900184,
      "grad_norm": 2.4410698413848877,
      "learning_rate": 4.764550961285225e-05,
      "loss": 1.7368,
      "step": 1792
    },
    {
      "epoch": 0.047221490650513565,
      "grad_norm": 2.2780191898345947,
      "learning_rate": 4.764419278377667e-05,
      "loss": 1.9566,
      "step": 1793
    },
    {
      "epoch": 0.04724782723202528,
      "grad_norm": 3.4953715801239014,
      "learning_rate": 4.764287595470108e-05,
      "loss": 0.9462,
      "step": 1794
    },
    {
      "epoch": 0.047274163813537,
      "grad_norm": 2.6264867782592773,
      "learning_rate": 4.76415591256255e-05,
      "loss": 1.8387,
      "step": 1795
    },
    {
      "epoch": 0.047300500395048725,
      "grad_norm": 2.1491644382476807,
      "learning_rate": 4.764024229654991e-05,
      "loss": 1.5921,
      "step": 1796
    },
    {
      "epoch": 0.04732683697656044,
      "grad_norm": 3.6468658447265625,
      "learning_rate": 4.763892546747433e-05,
      "loss": 1.2979,
      "step": 1797
    },
    {
      "epoch": 0.04735317355807216,
      "grad_norm": 2.1967475414276123,
      "learning_rate": 4.763760863839874e-05,
      "loss": 1.4175,
      "step": 1798
    },
    {
      "epoch": 0.04737951013958388,
      "grad_norm": 2.6574835777282715,
      "learning_rate": 4.7636291809323154e-05,
      "loss": 2.0071,
      "step": 1799
    },
    {
      "epoch": 0.0474058467210956,
      "grad_norm": 1.8592238426208496,
      "learning_rate": 4.763497498024756e-05,
      "loss": 2.7643,
      "step": 1800
    },
    {
      "epoch": 0.047432183302607324,
      "grad_norm": 3.7676124572753906,
      "learning_rate": 4.763365815117198e-05,
      "loss": 0.4047,
      "step": 1801
    },
    {
      "epoch": 0.04745851988411904,
      "grad_norm": 3.908045768737793,
      "learning_rate": 4.7632341322096394e-05,
      "loss": 1.8999,
      "step": 1802
    },
    {
      "epoch": 0.04748485646563076,
      "grad_norm": 2.474395513534546,
      "learning_rate": 4.763102449302081e-05,
      "loss": 2.4039,
      "step": 1803
    },
    {
      "epoch": 0.04751119304714248,
      "grad_norm": 2.8622822761535645,
      "learning_rate": 4.7629707663945225e-05,
      "loss": 0.5707,
      "step": 1804
    },
    {
      "epoch": 0.0475375296286542,
      "grad_norm": 2.638190507888794,
      "learning_rate": 4.7628390834869634e-05,
      "loss": 1.6038,
      "step": 1805
    },
    {
      "epoch": 0.04756386621016592,
      "grad_norm": 2.4967191219329834,
      "learning_rate": 4.762707400579405e-05,
      "loss": 1.9041,
      "step": 1806
    },
    {
      "epoch": 0.04759020279167764,
      "grad_norm": 3.1558327674865723,
      "learning_rate": 4.7625757176718465e-05,
      "loss": 0.7622,
      "step": 1807
    },
    {
      "epoch": 0.04761653937318936,
      "grad_norm": 3.3613836765289307,
      "learning_rate": 4.762444034764288e-05,
      "loss": 1.114,
      "step": 1808
    },
    {
      "epoch": 0.047642875954701076,
      "grad_norm": 3.9828338623046875,
      "learning_rate": 4.762312351856729e-05,
      "loss": 2.1822,
      "step": 1809
    },
    {
      "epoch": 0.0476692125362128,
      "grad_norm": 4.770702362060547,
      "learning_rate": 4.7621806689491706e-05,
      "loss": 0.9249,
      "step": 1810
    },
    {
      "epoch": 0.04769554911772452,
      "grad_norm": 2.768343448638916,
      "learning_rate": 4.7620489860416114e-05,
      "loss": 1.7084,
      "step": 1811
    },
    {
      "epoch": 0.04772188569923624,
      "grad_norm": 2.334273338317871,
      "learning_rate": 4.761917303134054e-05,
      "loss": 1.744,
      "step": 1812
    },
    {
      "epoch": 0.04774822228074796,
      "grad_norm": 2.6795191764831543,
      "learning_rate": 4.7617856202264946e-05,
      "loss": 1.3036,
      "step": 1813
    },
    {
      "epoch": 0.04777455886225968,
      "grad_norm": 3.734348773956299,
      "learning_rate": 4.761653937318936e-05,
      "loss": 2.1771,
      "step": 1814
    },
    {
      "epoch": 0.0478008954437714,
      "grad_norm": 3.245972156524658,
      "learning_rate": 4.761522254411378e-05,
      "loss": 1.9377,
      "step": 1815
    },
    {
      "epoch": 0.04782723202528312,
      "grad_norm": 2.143364906311035,
      "learning_rate": 4.761390571503819e-05,
      "loss": 2.0788,
      "step": 1816
    },
    {
      "epoch": 0.047853568606794836,
      "grad_norm": 2.610452890396118,
      "learning_rate": 4.761258888596261e-05,
      "loss": 0.83,
      "step": 1817
    },
    {
      "epoch": 0.04787990518830656,
      "grad_norm": 2.242518424987793,
      "learning_rate": 4.761127205688702e-05,
      "loss": 1.9233,
      "step": 1818
    },
    {
      "epoch": 0.04790624176981828,
      "grad_norm": 2.1876494884490967,
      "learning_rate": 4.760995522781143e-05,
      "loss": 2.1151,
      "step": 1819
    },
    {
      "epoch": 0.047932578351329996,
      "grad_norm": 3.099712371826172,
      "learning_rate": 4.760863839873584e-05,
      "loss": 0.5995,
      "step": 1820
    },
    {
      "epoch": 0.04795891493284172,
      "grad_norm": 3.743948459625244,
      "learning_rate": 4.7607321569660264e-05,
      "loss": 2.0935,
      "step": 1821
    },
    {
      "epoch": 0.047985251514353434,
      "grad_norm": 2.697472333908081,
      "learning_rate": 4.760600474058467e-05,
      "loss": 2.4041,
      "step": 1822
    },
    {
      "epoch": 0.04801158809586516,
      "grad_norm": 5.2074785232543945,
      "learning_rate": 4.760468791150909e-05,
      "loss": 1.6589,
      "step": 1823
    },
    {
      "epoch": 0.04803792467737688,
      "grad_norm": 3.097689390182495,
      "learning_rate": 4.7603371082433504e-05,
      "loss": 2.1798,
      "step": 1824
    },
    {
      "epoch": 0.048064261258888595,
      "grad_norm": 2.648519277572632,
      "learning_rate": 4.760205425335791e-05,
      "loss": 1.5074,
      "step": 1825
    },
    {
      "epoch": 0.04809059784040032,
      "grad_norm": 3.4516360759735107,
      "learning_rate": 4.7600737424282335e-05,
      "loss": 1.97,
      "step": 1826
    },
    {
      "epoch": 0.04811693442191203,
      "grad_norm": 3.2515652179718018,
      "learning_rate": 4.7599420595206744e-05,
      "loss": 2.2869,
      "step": 1827
    },
    {
      "epoch": 0.048143271003423756,
      "grad_norm": 2.2592008113861084,
      "learning_rate": 4.759810376613116e-05,
      "loss": 2.0299,
      "step": 1828
    },
    {
      "epoch": 0.04816960758493548,
      "grad_norm": 3.1698997020721436,
      "learning_rate": 4.759678693705557e-05,
      "loss": 1.6406,
      "step": 1829
    },
    {
      "epoch": 0.048195944166447194,
      "grad_norm": 2.9967458248138428,
      "learning_rate": 4.759547010797999e-05,
      "loss": 2.2723,
      "step": 1830
    },
    {
      "epoch": 0.048222280747958916,
      "grad_norm": 2.8788657188415527,
      "learning_rate": 4.75941532789044e-05,
      "loss": 0.3471,
      "step": 1831
    },
    {
      "epoch": 0.04824861732947063,
      "grad_norm": 3.3488869667053223,
      "learning_rate": 4.7592836449828815e-05,
      "loss": 2.5425,
      "step": 1832
    },
    {
      "epoch": 0.048274953910982354,
      "grad_norm": 2.062664031982422,
      "learning_rate": 4.759151962075323e-05,
      "loss": 1.9583,
      "step": 1833
    },
    {
      "epoch": 0.04830129049249408,
      "grad_norm": 2.0781054496765137,
      "learning_rate": 4.759020279167764e-05,
      "loss": 1.4181,
      "step": 1834
    },
    {
      "epoch": 0.04832762707400579,
      "grad_norm": 3.9546961784362793,
      "learning_rate": 4.758888596260206e-05,
      "loss": 1.2198,
      "step": 1835
    },
    {
      "epoch": 0.048353963655517515,
      "grad_norm": 9.188541412353516,
      "learning_rate": 4.758756913352647e-05,
      "loss": 2.2232,
      "step": 1836
    },
    {
      "epoch": 0.04838030023702923,
      "grad_norm": 4.707973003387451,
      "learning_rate": 4.7586252304450886e-05,
      "loss": 1.7777,
      "step": 1837
    },
    {
      "epoch": 0.04840663681854095,
      "grad_norm": 2.385406494140625,
      "learning_rate": 4.7584935475375295e-05,
      "loss": 1.7226,
      "step": 1838
    },
    {
      "epoch": 0.048432973400052676,
      "grad_norm": 2.3620717525482178,
      "learning_rate": 4.758361864629971e-05,
      "loss": 1.9293,
      "step": 1839
    },
    {
      "epoch": 0.04845930998156439,
      "grad_norm": 2.837907075881958,
      "learning_rate": 4.7582301817224126e-05,
      "loss": 0.9535,
      "step": 1840
    },
    {
      "epoch": 0.048485646563076114,
      "grad_norm": 2.6598939895629883,
      "learning_rate": 4.758098498814854e-05,
      "loss": 0.6699,
      "step": 1841
    },
    {
      "epoch": 0.04851198314458783,
      "grad_norm": 2.891472339630127,
      "learning_rate": 4.757966815907296e-05,
      "loss": 1.4541,
      "step": 1842
    },
    {
      "epoch": 0.04853831972609955,
      "grad_norm": 3.0237386226654053,
      "learning_rate": 4.7578351329997366e-05,
      "loss": 2.0308,
      "step": 1843
    },
    {
      "epoch": 0.048564656307611274,
      "grad_norm": 3.1423702239990234,
      "learning_rate": 4.757703450092178e-05,
      "loss": 2.256,
      "step": 1844
    },
    {
      "epoch": 0.04859099288912299,
      "grad_norm": 2.3241753578186035,
      "learning_rate": 4.75757176718462e-05,
      "loss": 2.286,
      "step": 1845
    },
    {
      "epoch": 0.04861732947063471,
      "grad_norm": 3.8985023498535156,
      "learning_rate": 4.757440084277061e-05,
      "loss": 1.4076,
      "step": 1846
    },
    {
      "epoch": 0.04864366605214643,
      "grad_norm": 3.129281997680664,
      "learning_rate": 4.757308401369502e-05,
      "loss": 2.116,
      "step": 1847
    },
    {
      "epoch": 0.04867000263365815,
      "grad_norm": 1.8293781280517578,
      "learning_rate": 4.757176718461944e-05,
      "loss": 1.856,
      "step": 1848
    },
    {
      "epoch": 0.04869633921516987,
      "grad_norm": 3.197397232055664,
      "learning_rate": 4.757045035554385e-05,
      "loss": 0.9616,
      "step": 1849
    },
    {
      "epoch": 0.04872267579668159,
      "grad_norm": 7.709231853485107,
      "learning_rate": 4.756913352646827e-05,
      "loss": 2.0307,
      "step": 1850
    },
    {
      "epoch": 0.04874901237819331,
      "grad_norm": 2.38158917427063,
      "learning_rate": 4.7567816697392684e-05,
      "loss": 2.0916,
      "step": 1851
    },
    {
      "epoch": 0.04877534895970503,
      "grad_norm": 5.8824567794799805,
      "learning_rate": 4.756649986831709e-05,
      "loss": 0.9896,
      "step": 1852
    },
    {
      "epoch": 0.04880168554121675,
      "grad_norm": 2.287475824356079,
      "learning_rate": 4.756518303924151e-05,
      "loss": 1.4894,
      "step": 1853
    },
    {
      "epoch": 0.04882802212272847,
      "grad_norm": 3.0922799110412598,
      "learning_rate": 4.7563866210165924e-05,
      "loss": 2.0622,
      "step": 1854
    },
    {
      "epoch": 0.04885435870424019,
      "grad_norm": 3.4541170597076416,
      "learning_rate": 4.756254938109034e-05,
      "loss": 1.5996,
      "step": 1855
    },
    {
      "epoch": 0.04888069528575191,
      "grad_norm": 3.240755558013916,
      "learning_rate": 4.756123255201475e-05,
      "loss": 1.6617,
      "step": 1856
    },
    {
      "epoch": 0.04890703186726363,
      "grad_norm": 6.699325084686279,
      "learning_rate": 4.7559915722939164e-05,
      "loss": 1.6213,
      "step": 1857
    },
    {
      "epoch": 0.04893336844877535,
      "grad_norm": 2.5335021018981934,
      "learning_rate": 4.755859889386357e-05,
      "loss": 1.7242,
      "step": 1858
    },
    {
      "epoch": 0.04895970503028707,
      "grad_norm": 4.607972621917725,
      "learning_rate": 4.7557282064787995e-05,
      "loss": 1.5549,
      "step": 1859
    },
    {
      "epoch": 0.048986041611798786,
      "grad_norm": 2.0614113807678223,
      "learning_rate": 4.7555965235712404e-05,
      "loss": 1.8949,
      "step": 1860
    },
    {
      "epoch": 0.04901237819331051,
      "grad_norm": 5.062224388122559,
      "learning_rate": 4.755464840663682e-05,
      "loss": 1.2759,
      "step": 1861
    },
    {
      "epoch": 0.04903871477482223,
      "grad_norm": 5.996825218200684,
      "learning_rate": 4.7553331577561235e-05,
      "loss": 2.8841,
      "step": 1862
    },
    {
      "epoch": 0.04906505135633395,
      "grad_norm": 2.899484634399414,
      "learning_rate": 4.755201474848565e-05,
      "loss": 2.3065,
      "step": 1863
    },
    {
      "epoch": 0.04909138793784567,
      "grad_norm": 5.001921653747559,
      "learning_rate": 4.7550697919410066e-05,
      "loss": 2.2021,
      "step": 1864
    },
    {
      "epoch": 0.049117724519357385,
      "grad_norm": 5.116119861602783,
      "learning_rate": 4.7549381090334475e-05,
      "loss": 1.6329,
      "step": 1865
    },
    {
      "epoch": 0.04914406110086911,
      "grad_norm": 3.9319283962249756,
      "learning_rate": 4.754806426125889e-05,
      "loss": 1.4503,
      "step": 1866
    },
    {
      "epoch": 0.04917039768238083,
      "grad_norm": 2.820613384246826,
      "learning_rate": 4.75467474321833e-05,
      "loss": 1.9275,
      "step": 1867
    },
    {
      "epoch": 0.049196734263892546,
      "grad_norm": 2.5694661140441895,
      "learning_rate": 4.754543060310772e-05,
      "loss": 3.0441,
      "step": 1868
    },
    {
      "epoch": 0.04922307084540427,
      "grad_norm": 2.513005018234253,
      "learning_rate": 4.754411377403213e-05,
      "loss": 1.9219,
      "step": 1869
    },
    {
      "epoch": 0.049249407426915984,
      "grad_norm": 3.107499837875366,
      "learning_rate": 4.7542796944956547e-05,
      "loss": 0.9544,
      "step": 1870
    },
    {
      "epoch": 0.049275744008427706,
      "grad_norm": 2.8746564388275146,
      "learning_rate": 4.754148011588096e-05,
      "loss": 1.4186,
      "step": 1871
    },
    {
      "epoch": 0.04930208058993943,
      "grad_norm": 7.512135982513428,
      "learning_rate": 4.754016328680537e-05,
      "loss": 1.6365,
      "step": 1872
    },
    {
      "epoch": 0.049328417171451144,
      "grad_norm": 2.2588987350463867,
      "learning_rate": 4.753884645772979e-05,
      "loss": 1.4779,
      "step": 1873
    },
    {
      "epoch": 0.04935475375296287,
      "grad_norm": 2.7753746509552,
      "learning_rate": 4.75375296286542e-05,
      "loss": 1.6723,
      "step": 1874
    },
    {
      "epoch": 0.04938109033447458,
      "grad_norm": 5.017754554748535,
      "learning_rate": 4.753621279957862e-05,
      "loss": 1.3852,
      "step": 1875
    },
    {
      "epoch": 0.049407426915986305,
      "grad_norm": 2.4810543060302734,
      "learning_rate": 4.7534895970503027e-05,
      "loss": 1.6004,
      "step": 1876
    },
    {
      "epoch": 0.04943376349749803,
      "grad_norm": 5.922138690948486,
      "learning_rate": 4.753357914142744e-05,
      "loss": 2.0072,
      "step": 1877
    },
    {
      "epoch": 0.04946010007900974,
      "grad_norm": 2.1590514183044434,
      "learning_rate": 4.753226231235186e-05,
      "loss": 1.1749,
      "step": 1878
    },
    {
      "epoch": 0.049486436660521466,
      "grad_norm": 2.207684278488159,
      "learning_rate": 4.753094548327627e-05,
      "loss": 2.7352,
      "step": 1879
    },
    {
      "epoch": 0.04951277324203318,
      "grad_norm": 2.755525588989258,
      "learning_rate": 4.752962865420069e-05,
      "loss": 1.8859,
      "step": 1880
    },
    {
      "epoch": 0.049539109823544904,
      "grad_norm": 2.835995674133301,
      "learning_rate": 4.75283118251251e-05,
      "loss": 1.9658,
      "step": 1881
    },
    {
      "epoch": 0.049565446405056626,
      "grad_norm": 2.135390281677246,
      "learning_rate": 4.752699499604952e-05,
      "loss": 2.206,
      "step": 1882
    },
    {
      "epoch": 0.04959178298656834,
      "grad_norm": 2.1092138290405273,
      "learning_rate": 4.752567816697393e-05,
      "loss": 1.7261,
      "step": 1883
    },
    {
      "epoch": 0.049618119568080064,
      "grad_norm": 2.323204517364502,
      "learning_rate": 4.7524361337898345e-05,
      "loss": 1.7719,
      "step": 1884
    },
    {
      "epoch": 0.04964445614959178,
      "grad_norm": 2.8365225791931152,
      "learning_rate": 4.752304450882275e-05,
      "loss": 1.8774,
      "step": 1885
    },
    {
      "epoch": 0.0496707927311035,
      "grad_norm": 2.5356972217559814,
      "learning_rate": 4.752172767974717e-05,
      "loss": 2.5013,
      "step": 1886
    },
    {
      "epoch": 0.049697129312615225,
      "grad_norm": 3.487934112548828,
      "learning_rate": 4.7520410850671585e-05,
      "loss": 1.7786,
      "step": 1887
    },
    {
      "epoch": 0.04972346589412694,
      "grad_norm": 2.3606793880462646,
      "learning_rate": 4.7519094021596e-05,
      "loss": 1.9574,
      "step": 1888
    },
    {
      "epoch": 0.04974980247563866,
      "grad_norm": 2.667154312133789,
      "learning_rate": 4.7517777192520416e-05,
      "loss": 1.5387,
      "step": 1889
    },
    {
      "epoch": 0.04977613905715038,
      "grad_norm": 2.8656506538391113,
      "learning_rate": 4.7516460363444825e-05,
      "loss": 0.3605,
      "step": 1890
    },
    {
      "epoch": 0.0498024756386621,
      "grad_norm": 3.4387459754943848,
      "learning_rate": 4.751514353436924e-05,
      "loss": 1.4456,
      "step": 1891
    },
    {
      "epoch": 0.049828812220173824,
      "grad_norm": 2.3970303535461426,
      "learning_rate": 4.7513826705293656e-05,
      "loss": 2.17,
      "step": 1892
    },
    {
      "epoch": 0.04985514880168554,
      "grad_norm": 1.9291952848434448,
      "learning_rate": 4.751250987621807e-05,
      "loss": 1.9023,
      "step": 1893
    },
    {
      "epoch": 0.04988148538319726,
      "grad_norm": 3.236123561859131,
      "learning_rate": 4.751119304714248e-05,
      "loss": 2.0105,
      "step": 1894
    },
    {
      "epoch": 0.04990782196470898,
      "grad_norm": 2.9148108959198,
      "learning_rate": 4.7509876218066896e-05,
      "loss": 2.5134,
      "step": 1895
    },
    {
      "epoch": 0.0499341585462207,
      "grad_norm": 2.144341230392456,
      "learning_rate": 4.750855938899131e-05,
      "loss": 1.3602,
      "step": 1896
    },
    {
      "epoch": 0.04996049512773242,
      "grad_norm": 2.2980000972747803,
      "learning_rate": 4.750724255991573e-05,
      "loss": 2.0295,
      "step": 1897
    },
    {
      "epoch": 0.04998683170924414,
      "grad_norm": 4.49240255355835,
      "learning_rate": 4.750592573084014e-05,
      "loss": 1.6687,
      "step": 1898
    },
    {
      "epoch": 0.05001316829075586,
      "grad_norm": 6.845585346221924,
      "learning_rate": 4.750460890176455e-05,
      "loss": 1.5149,
      "step": 1899
    },
    {
      "epoch": 0.05003950487226758,
      "grad_norm": 2.1115126609802246,
      "learning_rate": 4.750329207268897e-05,
      "loss": 1.8899,
      "step": 1900
    },
    {
      "epoch": 0.0500658414537793,
      "grad_norm": 3.358717918395996,
      "learning_rate": 4.750197524361338e-05,
      "loss": 1.9376,
      "step": 1901
    },
    {
      "epoch": 0.05009217803529102,
      "grad_norm": 2.899951457977295,
      "learning_rate": 4.75006584145378e-05,
      "loss": 1.5477,
      "step": 1902
    },
    {
      "epoch": 0.05011851461680274,
      "grad_norm": 3.117509126663208,
      "learning_rate": 4.749934158546221e-05,
      "loss": 1.6119,
      "step": 1903
    },
    {
      "epoch": 0.05014485119831446,
      "grad_norm": 2.152562379837036,
      "learning_rate": 4.749802475638662e-05,
      "loss": 1.6742,
      "step": 1904
    },
    {
      "epoch": 0.05017118777982618,
      "grad_norm": 6.079864978790283,
      "learning_rate": 4.749670792731104e-05,
      "loss": 1.015,
      "step": 1905
    },
    {
      "epoch": 0.0501975243613379,
      "grad_norm": 2.2239325046539307,
      "learning_rate": 4.7495391098235454e-05,
      "loss": 1.7363,
      "step": 1906
    },
    {
      "epoch": 0.05022386094284962,
      "grad_norm": 2.863947868347168,
      "learning_rate": 4.749407426915987e-05,
      "loss": 1.8702,
      "step": 1907
    },
    {
      "epoch": 0.050250197524361336,
      "grad_norm": 2.382564067840576,
      "learning_rate": 4.749275744008428e-05,
      "loss": 2.1574,
      "step": 1908
    },
    {
      "epoch": 0.05027653410587306,
      "grad_norm": 4.304083347320557,
      "learning_rate": 4.7491440611008694e-05,
      "loss": 1.6574,
      "step": 1909
    },
    {
      "epoch": 0.05030287068738478,
      "grad_norm": 2.6413674354553223,
      "learning_rate": 4.74901237819331e-05,
      "loss": 1.942,
      "step": 1910
    },
    {
      "epoch": 0.050329207268896496,
      "grad_norm": 4.041344165802002,
      "learning_rate": 4.7488806952857525e-05,
      "loss": 1.8458,
      "step": 1911
    },
    {
      "epoch": 0.05035554385040822,
      "grad_norm": 2.3719329833984375,
      "learning_rate": 4.7487490123781934e-05,
      "loss": 1.7017,
      "step": 1912
    },
    {
      "epoch": 0.050381880431919934,
      "grad_norm": 2.3702516555786133,
      "learning_rate": 4.748617329470635e-05,
      "loss": 1.9608,
      "step": 1913
    },
    {
      "epoch": 0.05040821701343166,
      "grad_norm": 1.9186686277389526,
      "learning_rate": 4.748485646563076e-05,
      "loss": 1.6874,
      "step": 1914
    },
    {
      "epoch": 0.05043455359494338,
      "grad_norm": 5.602931976318359,
      "learning_rate": 4.748353963655518e-05,
      "loss": 1.7823,
      "step": 1915
    },
    {
      "epoch": 0.050460890176455095,
      "grad_norm": 2.4267005920410156,
      "learning_rate": 4.748222280747959e-05,
      "loss": 1.386,
      "step": 1916
    },
    {
      "epoch": 0.05048722675796682,
      "grad_norm": 2.46811580657959,
      "learning_rate": 4.7480905978404005e-05,
      "loss": 1.0042,
      "step": 1917
    },
    {
      "epoch": 0.05051356333947853,
      "grad_norm": 4.005268573760986,
      "learning_rate": 4.747958914932842e-05,
      "loss": 0.7043,
      "step": 1918
    },
    {
      "epoch": 0.050539899920990256,
      "grad_norm": 2.318715810775757,
      "learning_rate": 4.747827232025283e-05,
      "loss": 1.4512,
      "step": 1919
    },
    {
      "epoch": 0.05056623650250198,
      "grad_norm": 2.486562728881836,
      "learning_rate": 4.747695549117725e-05,
      "loss": 1.8038,
      "step": 1920
    },
    {
      "epoch": 0.050592573084013694,
      "grad_norm": 3.297780990600586,
      "learning_rate": 4.747563866210166e-05,
      "loss": 1.6142,
      "step": 1921
    },
    {
      "epoch": 0.050618909665525416,
      "grad_norm": 3.328735589981079,
      "learning_rate": 4.7474321833026076e-05,
      "loss": 2.6925,
      "step": 1922
    },
    {
      "epoch": 0.05064524624703713,
      "grad_norm": 3.1211743354797363,
      "learning_rate": 4.7473005003950485e-05,
      "loss": 2.216,
      "step": 1923
    },
    {
      "epoch": 0.050671582828548854,
      "grad_norm": 3.6800456047058105,
      "learning_rate": 4.74716881748749e-05,
      "loss": 2.0336,
      "step": 1924
    },
    {
      "epoch": 0.05069791941006058,
      "grad_norm": 7.806819915771484,
      "learning_rate": 4.7470371345799316e-05,
      "loss": 2.03,
      "step": 1925
    },
    {
      "epoch": 0.05072425599157229,
      "grad_norm": 2.605405330657959,
      "learning_rate": 4.746905451672373e-05,
      "loss": 1.9685,
      "step": 1926
    },
    {
      "epoch": 0.050750592573084015,
      "grad_norm": 2.262094497680664,
      "learning_rate": 4.746773768764815e-05,
      "loss": 1.8916,
      "step": 1927
    },
    {
      "epoch": 0.05077692915459573,
      "grad_norm": 2.370267391204834,
      "learning_rate": 4.7466420858572556e-05,
      "loss": 1.4207,
      "step": 1928
    },
    {
      "epoch": 0.05080326573610745,
      "grad_norm": 3.779275417327881,
      "learning_rate": 4.746510402949698e-05,
      "loss": 2.0838,
      "step": 1929
    },
    {
      "epoch": 0.050829602317619176,
      "grad_norm": 2.3945186138153076,
      "learning_rate": 4.746378720042139e-05,
      "loss": 1.3614,
      "step": 1930
    },
    {
      "epoch": 0.05085593889913089,
      "grad_norm": 2.312326669692993,
      "learning_rate": 4.74624703713458e-05,
      "loss": 1.7729,
      "step": 1931
    },
    {
      "epoch": 0.050882275480642614,
      "grad_norm": 3.108835458755493,
      "learning_rate": 4.746115354227021e-05,
      "loss": 1.3388,
      "step": 1932
    },
    {
      "epoch": 0.05090861206215433,
      "grad_norm": 3.0566723346710205,
      "learning_rate": 4.745983671319463e-05,
      "loss": 1.6849,
      "step": 1933
    },
    {
      "epoch": 0.05093494864366605,
      "grad_norm": 2.147061347961426,
      "learning_rate": 4.745851988411904e-05,
      "loss": 1.4168,
      "step": 1934
    },
    {
      "epoch": 0.050961285225177774,
      "grad_norm": 2.6942696571350098,
      "learning_rate": 4.745720305504346e-05,
      "loss": 1.9201,
      "step": 1935
    },
    {
      "epoch": 0.05098762180668949,
      "grad_norm": 2.424778461456299,
      "learning_rate": 4.7455886225967874e-05,
      "loss": 1.5221,
      "step": 1936
    },
    {
      "epoch": 0.05101395838820121,
      "grad_norm": 3.8146185874938965,
      "learning_rate": 4.745456939689228e-05,
      "loss": 1.5397,
      "step": 1937
    },
    {
      "epoch": 0.05104029496971293,
      "grad_norm": 2.4015159606933594,
      "learning_rate": 4.74532525678167e-05,
      "loss": 1.7502,
      "step": 1938
    },
    {
      "epoch": 0.05106663155122465,
      "grad_norm": 1.926942229270935,
      "learning_rate": 4.7451935738741114e-05,
      "loss": 1.7521,
      "step": 1939
    },
    {
      "epoch": 0.05109296813273637,
      "grad_norm": 1.8515033721923828,
      "learning_rate": 4.745061890966553e-05,
      "loss": 1.8109,
      "step": 1940
    },
    {
      "epoch": 0.05111930471424809,
      "grad_norm": 2.4684648513793945,
      "learning_rate": 4.744930208058994e-05,
      "loss": 1.9839,
      "step": 1941
    },
    {
      "epoch": 0.05114564129575981,
      "grad_norm": 3.5624656677246094,
      "learning_rate": 4.7447985251514354e-05,
      "loss": 2.0394,
      "step": 1942
    },
    {
      "epoch": 0.05117197787727153,
      "grad_norm": 3.420684576034546,
      "learning_rate": 4.744666842243877e-05,
      "loss": 2.0959,
      "step": 1943
    },
    {
      "epoch": 0.05119831445878325,
      "grad_norm": 1.8533002138137817,
      "learning_rate": 4.7445351593363186e-05,
      "loss": 2.078,
      "step": 1944
    },
    {
      "epoch": 0.05122465104029497,
      "grad_norm": 2.8681881427764893,
      "learning_rate": 4.74440347642876e-05,
      "loss": 0.4574,
      "step": 1945
    },
    {
      "epoch": 0.05125098762180669,
      "grad_norm": 3.17950177192688,
      "learning_rate": 4.744271793521201e-05,
      "loss": 1.1603,
      "step": 1946
    },
    {
      "epoch": 0.05127732420331841,
      "grad_norm": 2.905369758605957,
      "learning_rate": 4.7441401106136426e-05,
      "loss": 1.7011,
      "step": 1947
    },
    {
      "epoch": 0.05130366078483013,
      "grad_norm": 2.4327666759490967,
      "learning_rate": 4.744008427706084e-05,
      "loss": 1.4245,
      "step": 1948
    },
    {
      "epoch": 0.05132999736634185,
      "grad_norm": 2.5221340656280518,
      "learning_rate": 4.743876744798526e-05,
      "loss": 1.6163,
      "step": 1949
    },
    {
      "epoch": 0.05135633394785357,
      "grad_norm": 2.8383095264434814,
      "learning_rate": 4.7437450618909666e-05,
      "loss": 0.8223,
      "step": 1950
    },
    {
      "epoch": 0.051382670529365286,
      "grad_norm": 3.493562698364258,
      "learning_rate": 4.743613378983408e-05,
      "loss": 1.3465,
      "step": 1951
    },
    {
      "epoch": 0.05140900711087701,
      "grad_norm": 3.039860486984253,
      "learning_rate": 4.74348169607585e-05,
      "loss": 1.5341,
      "step": 1952
    },
    {
      "epoch": 0.05143534369238873,
      "grad_norm": 19.08000373840332,
      "learning_rate": 4.743350013168291e-05,
      "loss": 3.6486,
      "step": 1953
    },
    {
      "epoch": 0.05146168027390045,
      "grad_norm": 3.1299734115600586,
      "learning_rate": 4.743218330260733e-05,
      "loss": 1.8175,
      "step": 1954
    },
    {
      "epoch": 0.05148801685541217,
      "grad_norm": 1.8462104797363281,
      "learning_rate": 4.743086647353174e-05,
      "loss": 1.7258,
      "step": 1955
    },
    {
      "epoch": 0.051514353436923885,
      "grad_norm": 2.8475687503814697,
      "learning_rate": 4.742954964445615e-05,
      "loss": 1.458,
      "step": 1956
    },
    {
      "epoch": 0.05154069001843561,
      "grad_norm": 2.0529065132141113,
      "learning_rate": 4.742823281538056e-05,
      "loss": 1.8204,
      "step": 1957
    },
    {
      "epoch": 0.05156702659994733,
      "grad_norm": 2.9291958808898926,
      "learning_rate": 4.7426915986304984e-05,
      "loss": 1.7926,
      "step": 1958
    },
    {
      "epoch": 0.051593363181459045,
      "grad_norm": 2.488250970840454,
      "learning_rate": 4.742559915722939e-05,
      "loss": 2.1954,
      "step": 1959
    },
    {
      "epoch": 0.05161969976297077,
      "grad_norm": 3.042243480682373,
      "learning_rate": 4.742428232815381e-05,
      "loss": 0.3938,
      "step": 1960
    },
    {
      "epoch": 0.051646036344482484,
      "grad_norm": 3.34202241897583,
      "learning_rate": 4.742296549907822e-05,
      "loss": 1.6101,
      "step": 1961
    },
    {
      "epoch": 0.051672372925994206,
      "grad_norm": 2.0528404712677,
      "learning_rate": 4.742164867000264e-05,
      "loss": 2.2328,
      "step": 1962
    },
    {
      "epoch": 0.05169870950750593,
      "grad_norm": 2.2398762702941895,
      "learning_rate": 4.742033184092705e-05,
      "loss": 1.4511,
      "step": 1963
    },
    {
      "epoch": 0.051725046089017644,
      "grad_norm": 2.497408628463745,
      "learning_rate": 4.7419015011851464e-05,
      "loss": 2.3698,
      "step": 1964
    },
    {
      "epoch": 0.05175138267052937,
      "grad_norm": 1.7545833587646484,
      "learning_rate": 4.741769818277588e-05,
      "loss": 2.3217,
      "step": 1965
    },
    {
      "epoch": 0.05177771925204108,
      "grad_norm": 1.993113398551941,
      "learning_rate": 4.741638135370029e-05,
      "loss": 2.019,
      "step": 1966
    },
    {
      "epoch": 0.051804055833552805,
      "grad_norm": 3.2278997898101807,
      "learning_rate": 4.741506452462471e-05,
      "loss": 1.867,
      "step": 1967
    },
    {
      "epoch": 0.05183039241506453,
      "grad_norm": 2.4407131671905518,
      "learning_rate": 4.741374769554912e-05,
      "loss": 1.8683,
      "step": 1968
    },
    {
      "epoch": 0.05185672899657624,
      "grad_norm": 2.2846662998199463,
      "learning_rate": 4.7412430866473535e-05,
      "loss": 2.114,
      "step": 1969
    },
    {
      "epoch": 0.051883065578087965,
      "grad_norm": 2.683457851409912,
      "learning_rate": 4.7411114037397944e-05,
      "loss": 1.9092,
      "step": 1970
    },
    {
      "epoch": 0.05190940215959968,
      "grad_norm": 2.1620802879333496,
      "learning_rate": 4.740979720832236e-05,
      "loss": 1.622,
      "step": 1971
    },
    {
      "epoch": 0.051935738741111404,
      "grad_norm": 2.3451576232910156,
      "learning_rate": 4.7408480379246775e-05,
      "loss": 2.3576,
      "step": 1972
    },
    {
      "epoch": 0.051962075322623126,
      "grad_norm": 2.0542962551116943,
      "learning_rate": 4.740716355017119e-05,
      "loss": 1.7956,
      "step": 1973
    },
    {
      "epoch": 0.05198841190413484,
      "grad_norm": 4.009820938110352,
      "learning_rate": 4.7405846721095606e-05,
      "loss": 0.915,
      "step": 1974
    },
    {
      "epoch": 0.052014748485646564,
      "grad_norm": 2.7401866912841797,
      "learning_rate": 4.7404529892020015e-05,
      "loss": 2.1814,
      "step": 1975
    },
    {
      "epoch": 0.05204108506715828,
      "grad_norm": 2.3345112800598145,
      "learning_rate": 4.740321306294443e-05,
      "loss": 2.1897,
      "step": 1976
    },
    {
      "epoch": 0.05206742164867,
      "grad_norm": 2.5021276473999023,
      "learning_rate": 4.7401896233868846e-05,
      "loss": 0.9051,
      "step": 1977
    },
    {
      "epoch": 0.052093758230181725,
      "grad_norm": 2.211122512817383,
      "learning_rate": 4.740057940479326e-05,
      "loss": 1.7051,
      "step": 1978
    },
    {
      "epoch": 0.05212009481169344,
      "grad_norm": 2.5293128490448,
      "learning_rate": 4.739926257571767e-05,
      "loss": 1.7382,
      "step": 1979
    },
    {
      "epoch": 0.05214643139320516,
      "grad_norm": 3.504066228866577,
      "learning_rate": 4.7397945746642086e-05,
      "loss": 0.5729,
      "step": 1980
    },
    {
      "epoch": 0.05217276797471688,
      "grad_norm": 2.8518624305725098,
      "learning_rate": 4.73966289175665e-05,
      "loss": 0.8239,
      "step": 1981
    },
    {
      "epoch": 0.0521991045562286,
      "grad_norm": 5.398329257965088,
      "learning_rate": 4.739531208849092e-05,
      "loss": 1.1326,
      "step": 1982
    },
    {
      "epoch": 0.052225441137740324,
      "grad_norm": 2.0230722427368164,
      "learning_rate": 4.739399525941533e-05,
      "loss": 1.6693,
      "step": 1983
    },
    {
      "epoch": 0.05225177771925204,
      "grad_norm": 2.5795300006866455,
      "learning_rate": 4.739267843033974e-05,
      "loss": 1.5578,
      "step": 1984
    },
    {
      "epoch": 0.05227811430076376,
      "grad_norm": 2.0967042446136475,
      "learning_rate": 4.739136160126416e-05,
      "loss": 2.1262,
      "step": 1985
    },
    {
      "epoch": 0.05230445088227548,
      "grad_norm": 4.75216007232666,
      "learning_rate": 4.739004477218857e-05,
      "loss": 1.7363,
      "step": 1986
    },
    {
      "epoch": 0.0523307874637872,
      "grad_norm": 1.9590355157852173,
      "learning_rate": 4.738872794311299e-05,
      "loss": 2.3006,
      "step": 1987
    },
    {
      "epoch": 0.05235712404529892,
      "grad_norm": 2.3280680179595947,
      "learning_rate": 4.73874111140374e-05,
      "loss": 0.8837,
      "step": 1988
    },
    {
      "epoch": 0.05238346062681064,
      "grad_norm": 2.8242263793945312,
      "learning_rate": 4.738609428496181e-05,
      "loss": 1.6547,
      "step": 1989
    },
    {
      "epoch": 0.05240979720832236,
      "grad_norm": 3.7524397373199463,
      "learning_rate": 4.738477745588623e-05,
      "loss": 1.3506,
      "step": 1990
    },
    {
      "epoch": 0.05243613378983408,
      "grad_norm": 1.93019437789917,
      "learning_rate": 4.7383460626810644e-05,
      "loss": 1.6627,
      "step": 1991
    },
    {
      "epoch": 0.0524624703713458,
      "grad_norm": 2.211174488067627,
      "learning_rate": 4.738214379773506e-05,
      "loss": 2.0234,
      "step": 1992
    },
    {
      "epoch": 0.05248880695285752,
      "grad_norm": 3.0913209915161133,
      "learning_rate": 4.738082696865947e-05,
      "loss": 2.3623,
      "step": 1993
    },
    {
      "epoch": 0.05251514353436924,
      "grad_norm": 2.308661937713623,
      "learning_rate": 4.7379510139583884e-05,
      "loss": 1.3141,
      "step": 1994
    },
    {
      "epoch": 0.05254148011588096,
      "grad_norm": 2.420914888381958,
      "learning_rate": 4.73781933105083e-05,
      "loss": 1.675,
      "step": 1995
    },
    {
      "epoch": 0.05256781669739268,
      "grad_norm": 2.018430471420288,
      "learning_rate": 4.7376876481432715e-05,
      "loss": 1.2073,
      "step": 1996
    },
    {
      "epoch": 0.0525941532789044,
      "grad_norm": 2.0899672508239746,
      "learning_rate": 4.7375559652357124e-05,
      "loss": 2.0052,
      "step": 1997
    },
    {
      "epoch": 0.05262048986041612,
      "grad_norm": 2.986414909362793,
      "learning_rate": 4.737424282328154e-05,
      "loss": 1.9322,
      "step": 1998
    },
    {
      "epoch": 0.052646826441927835,
      "grad_norm": 2.661581516265869,
      "learning_rate": 4.7372925994205955e-05,
      "loss": 1.9804,
      "step": 1999
    },
    {
      "epoch": 0.05267316302343956,
      "grad_norm": 2.6744041442871094,
      "learning_rate": 4.737160916513037e-05,
      "loss": 0.6653,
      "step": 2000
    },
    {
      "epoch": 0.05269949960495128,
      "grad_norm": 3.131136178970337,
      "learning_rate": 4.7370292336054787e-05,
      "loss": 0.6623,
      "step": 2001
    },
    {
      "epoch": 0.052725836186462996,
      "grad_norm": 2.4772188663482666,
      "learning_rate": 4.7368975506979195e-05,
      "loss": 1.4164,
      "step": 2002
    },
    {
      "epoch": 0.05275217276797472,
      "grad_norm": 3.378445863723755,
      "learning_rate": 4.736765867790361e-05,
      "loss": 1.4511,
      "step": 2003
    },
    {
      "epoch": 0.052778509349486434,
      "grad_norm": 2.7198100090026855,
      "learning_rate": 4.736634184882802e-05,
      "loss": 0.6776,
      "step": 2004
    },
    {
      "epoch": 0.05280484593099816,
      "grad_norm": 2.3264641761779785,
      "learning_rate": 4.736502501975244e-05,
      "loss": 1.6766,
      "step": 2005
    },
    {
      "epoch": 0.05283118251250988,
      "grad_norm": 2.581003189086914,
      "learning_rate": 4.736370819067685e-05,
      "loss": 2.2525,
      "step": 2006
    },
    {
      "epoch": 0.052857519094021595,
      "grad_norm": 2.278221845626831,
      "learning_rate": 4.7362391361601267e-05,
      "loss": 1.8756,
      "step": 2007
    },
    {
      "epoch": 0.05288385567553332,
      "grad_norm": 2.241326093673706,
      "learning_rate": 4.736107453252568e-05,
      "loss": 2.9003,
      "step": 2008
    },
    {
      "epoch": 0.05291019225704503,
      "grad_norm": 2.409233570098877,
      "learning_rate": 4.73597577034501e-05,
      "loss": 1.9883,
      "step": 2009
    },
    {
      "epoch": 0.052936528838556755,
      "grad_norm": 3.8274152278900146,
      "learning_rate": 4.735844087437451e-05,
      "loss": 2.2167,
      "step": 2010
    },
    {
      "epoch": 0.05296286542006848,
      "grad_norm": 6.1324143409729,
      "learning_rate": 4.735712404529892e-05,
      "loss": 0.7901,
      "step": 2011
    },
    {
      "epoch": 0.05298920200158019,
      "grad_norm": 4.438776969909668,
      "learning_rate": 4.735580721622334e-05,
      "loss": 1.5326,
      "step": 2012
    },
    {
      "epoch": 0.053015538583091916,
      "grad_norm": 4.440764904022217,
      "learning_rate": 4.7354490387147747e-05,
      "loss": 0.9479,
      "step": 2013
    },
    {
      "epoch": 0.05304187516460363,
      "grad_norm": 4.195770740509033,
      "learning_rate": 4.735317355807217e-05,
      "loss": 2.1979,
      "step": 2014
    },
    {
      "epoch": 0.053068211746115354,
      "grad_norm": 2.461336851119995,
      "learning_rate": 4.735185672899658e-05,
      "loss": 2.2412,
      "step": 2015
    },
    {
      "epoch": 0.05309454832762708,
      "grad_norm": 2.2120697498321533,
      "learning_rate": 4.735053989992099e-05,
      "loss": 1.4593,
      "step": 2016
    },
    {
      "epoch": 0.05312088490913879,
      "grad_norm": 1.7244751453399658,
      "learning_rate": 4.73492230708454e-05,
      "loss": 2.0892,
      "step": 2017
    },
    {
      "epoch": 0.053147221490650515,
      "grad_norm": 2.106907844543457,
      "learning_rate": 4.734790624176982e-05,
      "loss": 1.7483,
      "step": 2018
    },
    {
      "epoch": 0.05317355807216223,
      "grad_norm": 1.8575800657272339,
      "learning_rate": 4.7346589412694233e-05,
      "loss": 1.9242,
      "step": 2019
    },
    {
      "epoch": 0.05319989465367395,
      "grad_norm": 2.1366355419158936,
      "learning_rate": 4.734527258361865e-05,
      "loss": 1.9495,
      "step": 2020
    },
    {
      "epoch": 0.053226231235185675,
      "grad_norm": 2.7665510177612305,
      "learning_rate": 4.7343955754543065e-05,
      "loss": 1.6444,
      "step": 2021
    },
    {
      "epoch": 0.05325256781669739,
      "grad_norm": 3.159970283508301,
      "learning_rate": 4.7342638925467473e-05,
      "loss": 1.0339,
      "step": 2022
    },
    {
      "epoch": 0.05327890439820911,
      "grad_norm": 4.321499347686768,
      "learning_rate": 4.734132209639189e-05,
      "loss": 0.8682,
      "step": 2023
    },
    {
      "epoch": 0.05330524097972083,
      "grad_norm": 2.1694865226745605,
      "learning_rate": 4.7340005267316305e-05,
      "loss": 1.4608,
      "step": 2024
    },
    {
      "epoch": 0.05333157756123255,
      "grad_norm": 2.515545129776001,
      "learning_rate": 4.733868843824072e-05,
      "loss": 1.7315,
      "step": 2025
    },
    {
      "epoch": 0.053357914142744274,
      "grad_norm": 3.302790880203247,
      "learning_rate": 4.733737160916513e-05,
      "loss": 1.6774,
      "step": 2026
    },
    {
      "epoch": 0.05338425072425599,
      "grad_norm": 2.1500535011291504,
      "learning_rate": 4.7336054780089545e-05,
      "loss": 2.338,
      "step": 2027
    },
    {
      "epoch": 0.05341058730576771,
      "grad_norm": 2.0290355682373047,
      "learning_rate": 4.733473795101396e-05,
      "loss": 1.7395,
      "step": 2028
    },
    {
      "epoch": 0.05343692388727943,
      "grad_norm": 4.227078914642334,
      "learning_rate": 4.7333421121938376e-05,
      "loss": 1.0584,
      "step": 2029
    },
    {
      "epoch": 0.05346326046879115,
      "grad_norm": 2.465289831161499,
      "learning_rate": 4.733210429286279e-05,
      "loss": 2.0544,
      "step": 2030
    },
    {
      "epoch": 0.05348959705030287,
      "grad_norm": 2.5843186378479004,
      "learning_rate": 4.73307874637872e-05,
      "loss": 1.5605,
      "step": 2031
    },
    {
      "epoch": 0.05351593363181459,
      "grad_norm": 2.026824474334717,
      "learning_rate": 4.7329470634711616e-05,
      "loss": 1.2874,
      "step": 2032
    },
    {
      "epoch": 0.05354227021332631,
      "grad_norm": 2.224200963973999,
      "learning_rate": 4.732815380563603e-05,
      "loss": 2.2449,
      "step": 2033
    },
    {
      "epoch": 0.05356860679483803,
      "grad_norm": 2.0412447452545166,
      "learning_rate": 4.732683697656045e-05,
      "loss": 1.8891,
      "step": 2034
    },
    {
      "epoch": 0.05359494337634975,
      "grad_norm": 5.431974411010742,
      "learning_rate": 4.7325520147484856e-05,
      "loss": 0.7076,
      "step": 2035
    },
    {
      "epoch": 0.05362127995786147,
      "grad_norm": 2.5169289112091064,
      "learning_rate": 4.732420331840927e-05,
      "loss": 1.9156,
      "step": 2036
    },
    {
      "epoch": 0.05364761653937319,
      "grad_norm": 4.09473180770874,
      "learning_rate": 4.732288648933369e-05,
      "loss": 2.0082,
      "step": 2037
    },
    {
      "epoch": 0.05367395312088491,
      "grad_norm": 1.8645721673965454,
      "learning_rate": 4.73215696602581e-05,
      "loss": 2.0827,
      "step": 2038
    },
    {
      "epoch": 0.05370028970239663,
      "grad_norm": 1.8461860418319702,
      "learning_rate": 4.732025283118252e-05,
      "loss": 2.6198,
      "step": 2039
    },
    {
      "epoch": 0.05372662628390835,
      "grad_norm": 2.1248743534088135,
      "learning_rate": 4.731893600210693e-05,
      "loss": 2.2577,
      "step": 2040
    },
    {
      "epoch": 0.05375296286542007,
      "grad_norm": 3.238090753555298,
      "learning_rate": 4.731761917303134e-05,
      "loss": 1.6984,
      "step": 2041
    },
    {
      "epoch": 0.053779299446931786,
      "grad_norm": 6.792541980743408,
      "learning_rate": 4.731630234395576e-05,
      "loss": 0.805,
      "step": 2042
    },
    {
      "epoch": 0.05380563602844351,
      "grad_norm": 3.9255001544952393,
      "learning_rate": 4.7314985514880174e-05,
      "loss": 1.999,
      "step": 2043
    },
    {
      "epoch": 0.05383197260995523,
      "grad_norm": 2.77415132522583,
      "learning_rate": 4.731366868580458e-05,
      "loss": 1.7732,
      "step": 2044
    },
    {
      "epoch": 0.053858309191466947,
      "grad_norm": 2.2490503787994385,
      "learning_rate": 4.7312351856729e-05,
      "loss": 1.3486,
      "step": 2045
    },
    {
      "epoch": 0.05388464577297867,
      "grad_norm": 2.318188428878784,
      "learning_rate": 4.7311035027653414e-05,
      "loss": 1.5477,
      "step": 2046
    },
    {
      "epoch": 0.053910982354490385,
      "grad_norm": 2.8028390407562256,
      "learning_rate": 4.730971819857783e-05,
      "loss": 1.3602,
      "step": 2047
    },
    {
      "epoch": 0.05393731893600211,
      "grad_norm": 4.531374454498291,
      "learning_rate": 4.7308401369502245e-05,
      "loss": 0.584,
      "step": 2048
    },
    {
      "epoch": 0.05396365551751383,
      "grad_norm": 2.0498664379119873,
      "learning_rate": 4.7307084540426654e-05,
      "loss": 1.6595,
      "step": 2049
    },
    {
      "epoch": 0.053989992099025545,
      "grad_norm": 1.9256951808929443,
      "learning_rate": 4.730576771135107e-05,
      "loss": 1.7819,
      "step": 2050
    },
    {
      "epoch": 0.05401632868053727,
      "grad_norm": 2.812506675720215,
      "learning_rate": 4.730445088227548e-05,
      "loss": 1.5771,
      "step": 2051
    },
    {
      "epoch": 0.05404266526204898,
      "grad_norm": 4.5180840492248535,
      "learning_rate": 4.73031340531999e-05,
      "loss": 0.9664,
      "step": 2052
    },
    {
      "epoch": 0.054069001843560706,
      "grad_norm": 4.770064353942871,
      "learning_rate": 4.730181722412431e-05,
      "loss": 2.0111,
      "step": 2053
    },
    {
      "epoch": 0.05409533842507243,
      "grad_norm": 2.6490061283111572,
      "learning_rate": 4.7300500395048725e-05,
      "loss": 2.3096,
      "step": 2054
    },
    {
      "epoch": 0.054121675006584144,
      "grad_norm": 1.717597246170044,
      "learning_rate": 4.729918356597314e-05,
      "loss": 1.7607,
      "step": 2055
    },
    {
      "epoch": 0.054148011588095867,
      "grad_norm": 2.0759642124176025,
      "learning_rate": 4.729786673689755e-05,
      "loss": 0.8702,
      "step": 2056
    },
    {
      "epoch": 0.05417434816960758,
      "grad_norm": 3.1635491847991943,
      "learning_rate": 4.729654990782197e-05,
      "loss": 1.9108,
      "step": 2057
    },
    {
      "epoch": 0.054200684751119305,
      "grad_norm": 2.7228267192840576,
      "learning_rate": 4.729523307874638e-05,
      "loss": 2.328,
      "step": 2058
    },
    {
      "epoch": 0.05422702133263103,
      "grad_norm": 3.286054849624634,
      "learning_rate": 4.7293916249670796e-05,
      "loss": 1.3617,
      "step": 2059
    },
    {
      "epoch": 0.05425335791414274,
      "grad_norm": 2.2653841972351074,
      "learning_rate": 4.7292599420595205e-05,
      "loss": 1.937,
      "step": 2060
    },
    {
      "epoch": 0.054279694495654465,
      "grad_norm": 2.1899664402008057,
      "learning_rate": 4.729128259151963e-05,
      "loss": 1.411,
      "step": 2061
    },
    {
      "epoch": 0.05430603107716618,
      "grad_norm": 2.935560941696167,
      "learning_rate": 4.7289965762444036e-05,
      "loss": 1.1944,
      "step": 2062
    },
    {
      "epoch": 0.0543323676586779,
      "grad_norm": 2.700507879257202,
      "learning_rate": 4.728864893336845e-05,
      "loss": 1.7689,
      "step": 2063
    },
    {
      "epoch": 0.054358704240189626,
      "grad_norm": 3.2642228603363037,
      "learning_rate": 4.728733210429286e-05,
      "loss": 0.4289,
      "step": 2064
    },
    {
      "epoch": 0.05438504082170134,
      "grad_norm": 8.299562454223633,
      "learning_rate": 4.7286015275217276e-05,
      "loss": 1.7906,
      "step": 2065
    },
    {
      "epoch": 0.054411377403213064,
      "grad_norm": 4.752680778503418,
      "learning_rate": 4.728469844614169e-05,
      "loss": 1.2543,
      "step": 2066
    },
    {
      "epoch": 0.05443771398472478,
      "grad_norm": 2.2282509803771973,
      "learning_rate": 4.728338161706611e-05,
      "loss": 1.5959,
      "step": 2067
    },
    {
      "epoch": 0.0544640505662365,
      "grad_norm": 2.411721706390381,
      "learning_rate": 4.728206478799052e-05,
      "loss": 2.1507,
      "step": 2068
    },
    {
      "epoch": 0.054490387147748225,
      "grad_norm": 2.560208559036255,
      "learning_rate": 4.728074795891493e-05,
      "loss": 1.7338,
      "step": 2069
    },
    {
      "epoch": 0.05451672372925994,
      "grad_norm": 2.463653564453125,
      "learning_rate": 4.727943112983935e-05,
      "loss": 1.6155,
      "step": 2070
    },
    {
      "epoch": 0.05454306031077166,
      "grad_norm": 3.0527713298797607,
      "learning_rate": 4.727811430076376e-05,
      "loss": 2.5102,
      "step": 2071
    },
    {
      "epoch": 0.05456939689228338,
      "grad_norm": 2.5716476440429688,
      "learning_rate": 4.727679747168818e-05,
      "loss": 1.1894,
      "step": 2072
    },
    {
      "epoch": 0.0545957334737951,
      "grad_norm": 1.9859488010406494,
      "learning_rate": 4.727548064261259e-05,
      "loss": 1.3319,
      "step": 2073
    },
    {
      "epoch": 0.05462207005530682,
      "grad_norm": 3.26849627494812,
      "learning_rate": 4.7274163813537e-05,
      "loss": 0.9321,
      "step": 2074
    },
    {
      "epoch": 0.05464840663681854,
      "grad_norm": 2.9846725463867188,
      "learning_rate": 4.727284698446142e-05,
      "loss": 1.77,
      "step": 2075
    },
    {
      "epoch": 0.05467474321833026,
      "grad_norm": 3.5807576179504395,
      "learning_rate": 4.7271530155385834e-05,
      "loss": 1.042,
      "step": 2076
    },
    {
      "epoch": 0.05470107979984198,
      "grad_norm": 2.4289557933807373,
      "learning_rate": 4.727021332631025e-05,
      "loss": 1.2542,
      "step": 2077
    },
    {
      "epoch": 0.0547274163813537,
      "grad_norm": 2.07059645652771,
      "learning_rate": 4.726889649723466e-05,
      "loss": 1.8483,
      "step": 2078
    },
    {
      "epoch": 0.05475375296286542,
      "grad_norm": 2.3563039302825928,
      "learning_rate": 4.7267579668159074e-05,
      "loss": 1.5422,
      "step": 2079
    },
    {
      "epoch": 0.05478008954437714,
      "grad_norm": 2.043539047241211,
      "learning_rate": 4.726626283908349e-05,
      "loss": 2.4442,
      "step": 2080
    },
    {
      "epoch": 0.05480642612588886,
      "grad_norm": 2.965324640274048,
      "learning_rate": 4.7264946010007906e-05,
      "loss": 0.4117,
      "step": 2081
    },
    {
      "epoch": 0.05483276270740058,
      "grad_norm": 4.158634185791016,
      "learning_rate": 4.7263629180932314e-05,
      "loss": 1.752,
      "step": 2082
    },
    {
      "epoch": 0.0548590992889123,
      "grad_norm": 2.371791124343872,
      "learning_rate": 4.726231235185673e-05,
      "loss": 2.4796,
      "step": 2083
    },
    {
      "epoch": 0.05488543587042402,
      "grad_norm": 4.616940021514893,
      "learning_rate": 4.7260995522781146e-05,
      "loss": 1.6888,
      "step": 2084
    },
    {
      "epoch": 0.054911772451935736,
      "grad_norm": 2.3304145336151123,
      "learning_rate": 4.725967869370556e-05,
      "loss": 0.5402,
      "step": 2085
    },
    {
      "epoch": 0.05493810903344746,
      "grad_norm": 2.3909904956817627,
      "learning_rate": 4.725836186462998e-05,
      "loss": 1.8232,
      "step": 2086
    },
    {
      "epoch": 0.05496444561495918,
      "grad_norm": 2.0792007446289062,
      "learning_rate": 4.7257045035554386e-05,
      "loss": 1.6262,
      "step": 2087
    },
    {
      "epoch": 0.0549907821964709,
      "grad_norm": 2.2291672229766846,
      "learning_rate": 4.72557282064788e-05,
      "loss": 1.4469,
      "step": 2088
    },
    {
      "epoch": 0.05501711877798262,
      "grad_norm": 2.22428560256958,
      "learning_rate": 4.725441137740321e-05,
      "loss": 1.6023,
      "step": 2089
    },
    {
      "epoch": 0.055043455359494335,
      "grad_norm": 2.5402305126190186,
      "learning_rate": 4.725309454832763e-05,
      "loss": 1.342,
      "step": 2090
    },
    {
      "epoch": 0.05506979194100606,
      "grad_norm": 2.6844735145568848,
      "learning_rate": 4.725177771925204e-05,
      "loss": 2.3668,
      "step": 2091
    },
    {
      "epoch": 0.05509612852251778,
      "grad_norm": 1.9301756620407104,
      "learning_rate": 4.725046089017646e-05,
      "loss": 1.5754,
      "step": 2092
    },
    {
      "epoch": 0.055122465104029496,
      "grad_norm": 2.6414220333099365,
      "learning_rate": 4.724914406110087e-05,
      "loss": 0.6245,
      "step": 2093
    },
    {
      "epoch": 0.05514880168554122,
      "grad_norm": 2.5184977054595947,
      "learning_rate": 4.724782723202529e-05,
      "loss": 1.8511,
      "step": 2094
    },
    {
      "epoch": 0.055175138267052934,
      "grad_norm": 2.051305055618286,
      "learning_rate": 4.7246510402949704e-05,
      "loss": 1.6211,
      "step": 2095
    },
    {
      "epoch": 0.055201474848564656,
      "grad_norm": 3.3867344856262207,
      "learning_rate": 4.724519357387411e-05,
      "loss": 1.7756,
      "step": 2096
    },
    {
      "epoch": 0.05522781143007638,
      "grad_norm": 4.963647365570068,
      "learning_rate": 4.724387674479853e-05,
      "loss": 2.0092,
      "step": 2097
    },
    {
      "epoch": 0.055254148011588095,
      "grad_norm": 7.482678413391113,
      "learning_rate": 4.724255991572294e-05,
      "loss": 0.9619,
      "step": 2098
    },
    {
      "epoch": 0.05528048459309982,
      "grad_norm": 2.995453119277954,
      "learning_rate": 4.724124308664736e-05,
      "loss": 2.2678,
      "step": 2099
    },
    {
      "epoch": 0.05530682117461153,
      "grad_norm": 16.927141189575195,
      "learning_rate": 4.723992625757177e-05,
      "loss": 1.1171,
      "step": 2100
    },
    {
      "epoch": 0.055333157756123255,
      "grad_norm": 5.2754621505737305,
      "learning_rate": 4.7238609428496184e-05,
      "loss": 1.3992,
      "step": 2101
    },
    {
      "epoch": 0.05535949433763498,
      "grad_norm": 2.401787281036377,
      "learning_rate": 4.72372925994206e-05,
      "loss": 1.8031,
      "step": 2102
    },
    {
      "epoch": 0.05538583091914669,
      "grad_norm": 2.4323031902313232,
      "learning_rate": 4.723597577034501e-05,
      "loss": 1.491,
      "step": 2103
    },
    {
      "epoch": 0.055412167500658416,
      "grad_norm": 2.329332113265991,
      "learning_rate": 4.723465894126943e-05,
      "loss": 2.2786,
      "step": 2104
    },
    {
      "epoch": 0.05543850408217013,
      "grad_norm": 3.1811001300811768,
      "learning_rate": 4.723334211219384e-05,
      "loss": 1.8405,
      "step": 2105
    },
    {
      "epoch": 0.055464840663681854,
      "grad_norm": 1.9039262533187866,
      "learning_rate": 4.7232025283118255e-05,
      "loss": 1.8215,
      "step": 2106
    },
    {
      "epoch": 0.055491177245193576,
      "grad_norm": 6.489499092102051,
      "learning_rate": 4.7230708454042664e-05,
      "loss": 1.3855,
      "step": 2107
    },
    {
      "epoch": 0.05551751382670529,
      "grad_norm": 1.8143821954727173,
      "learning_rate": 4.7229391624967086e-05,
      "loss": 0.4841,
      "step": 2108
    },
    {
      "epoch": 0.055543850408217015,
      "grad_norm": 3.48903751373291,
      "learning_rate": 4.7228074795891495e-05,
      "loss": 0.7245,
      "step": 2109
    },
    {
      "epoch": 0.05557018698972873,
      "grad_norm": 2.996108293533325,
      "learning_rate": 4.722675796681591e-05,
      "loss": 1.8393,
      "step": 2110
    },
    {
      "epoch": 0.05559652357124045,
      "grad_norm": 3.148120880126953,
      "learning_rate": 4.7225441137740326e-05,
      "loss": 1.9642,
      "step": 2111
    },
    {
      "epoch": 0.055622860152752175,
      "grad_norm": 3.929673433303833,
      "learning_rate": 4.7224124308664735e-05,
      "loss": 0.6346,
      "step": 2112
    },
    {
      "epoch": 0.05564919673426389,
      "grad_norm": 2.003108501434326,
      "learning_rate": 4.722280747958916e-05,
      "loss": 1.8192,
      "step": 2113
    },
    {
      "epoch": 0.05567553331577561,
      "grad_norm": 5.596811771392822,
      "learning_rate": 4.7221490650513566e-05,
      "loss": 1.0959,
      "step": 2114
    },
    {
      "epoch": 0.05570186989728733,
      "grad_norm": 2.6867105960845947,
      "learning_rate": 4.722017382143798e-05,
      "loss": 1.9643,
      "step": 2115
    },
    {
      "epoch": 0.05572820647879905,
      "grad_norm": 2.384495258331299,
      "learning_rate": 4.721885699236239e-05,
      "loss": 2.4373,
      "step": 2116
    },
    {
      "epoch": 0.055754543060310774,
      "grad_norm": 2.1462979316711426,
      "learning_rate": 4.7217540163286806e-05,
      "loss": 1.8085,
      "step": 2117
    },
    {
      "epoch": 0.05578087964182249,
      "grad_norm": 2.061133623123169,
      "learning_rate": 4.721622333421122e-05,
      "loss": 2.0714,
      "step": 2118
    },
    {
      "epoch": 0.05580721622333421,
      "grad_norm": 2.018908739089966,
      "learning_rate": 4.721490650513564e-05,
      "loss": 0.5484,
      "step": 2119
    },
    {
      "epoch": 0.05583355280484593,
      "grad_norm": 5.552015781402588,
      "learning_rate": 4.7213589676060046e-05,
      "loss": 1.4557,
      "step": 2120
    },
    {
      "epoch": 0.05585988938635765,
      "grad_norm": 2.1100313663482666,
      "learning_rate": 4.721227284698446e-05,
      "loss": 2.0805,
      "step": 2121
    },
    {
      "epoch": 0.05588622596786937,
      "grad_norm": 2.795180320739746,
      "learning_rate": 4.721095601790888e-05,
      "loss": 1.7733,
      "step": 2122
    },
    {
      "epoch": 0.05591256254938109,
      "grad_norm": 2.8228659629821777,
      "learning_rate": 4.720963918883329e-05,
      "loss": 1.2784,
      "step": 2123
    },
    {
      "epoch": 0.05593889913089281,
      "grad_norm": 1.9794082641601562,
      "learning_rate": 4.720832235975771e-05,
      "loss": 1.956,
      "step": 2124
    },
    {
      "epoch": 0.05596523571240453,
      "grad_norm": 5.0670976638793945,
      "learning_rate": 4.720700553068212e-05,
      "loss": 1.3835,
      "step": 2125
    },
    {
      "epoch": 0.05599157229391625,
      "grad_norm": 2.25516414642334,
      "learning_rate": 4.720568870160653e-05,
      "loss": 1.8704,
      "step": 2126
    },
    {
      "epoch": 0.05601790887542797,
      "grad_norm": 1.8798249959945679,
      "learning_rate": 4.720437187253095e-05,
      "loss": 1.6445,
      "step": 2127
    },
    {
      "epoch": 0.05604424545693969,
      "grad_norm": 3.3134806156158447,
      "learning_rate": 4.7203055043455364e-05,
      "loss": 1.9723,
      "step": 2128
    },
    {
      "epoch": 0.05607058203845141,
      "grad_norm": 2.083958625793457,
      "learning_rate": 4.720173821437977e-05,
      "loss": 1.4712,
      "step": 2129
    },
    {
      "epoch": 0.05609691861996313,
      "grad_norm": 3.3280749320983887,
      "learning_rate": 4.720042138530419e-05,
      "loss": 1.3424,
      "step": 2130
    },
    {
      "epoch": 0.05612325520147485,
      "grad_norm": 3.359832525253296,
      "learning_rate": 4.7199104556228604e-05,
      "loss": 1.7044,
      "step": 2131
    },
    {
      "epoch": 0.05614959178298657,
      "grad_norm": 2.6637461185455322,
      "learning_rate": 4.719778772715302e-05,
      "loss": 1.3722,
      "step": 2132
    },
    {
      "epoch": 0.056175928364498286,
      "grad_norm": 5.161062240600586,
      "learning_rate": 4.7196470898077435e-05,
      "loss": 1.555,
      "step": 2133
    },
    {
      "epoch": 0.05620226494601001,
      "grad_norm": 2.719771385192871,
      "learning_rate": 4.7195154069001844e-05,
      "loss": 2.2629,
      "step": 2134
    },
    {
      "epoch": 0.05622860152752173,
      "grad_norm": 2.229858875274658,
      "learning_rate": 4.719383723992626e-05,
      "loss": 2.4039,
      "step": 2135
    },
    {
      "epoch": 0.056254938109033446,
      "grad_norm": 2.868281364440918,
      "learning_rate": 4.719252041085067e-05,
      "loss": 1.7693,
      "step": 2136
    },
    {
      "epoch": 0.05628127469054517,
      "grad_norm": 3.25990629196167,
      "learning_rate": 4.719120358177509e-05,
      "loss": 1.8586,
      "step": 2137
    },
    {
      "epoch": 0.056307611272056884,
      "grad_norm": 4.00319242477417,
      "learning_rate": 4.71898867526995e-05,
      "loss": 0.9102,
      "step": 2138
    },
    {
      "epoch": 0.05633394785356861,
      "grad_norm": 3.3860199451446533,
      "learning_rate": 4.7188569923623915e-05,
      "loss": 1.4904,
      "step": 2139
    },
    {
      "epoch": 0.05636028443508033,
      "grad_norm": 3.3914263248443604,
      "learning_rate": 4.718725309454833e-05,
      "loss": 1.7373,
      "step": 2140
    },
    {
      "epoch": 0.056386621016592045,
      "grad_norm": 2.3849058151245117,
      "learning_rate": 4.7185936265472747e-05,
      "loss": 2.0921,
      "step": 2141
    },
    {
      "epoch": 0.05641295759810377,
      "grad_norm": 2.0900299549102783,
      "learning_rate": 4.718461943639716e-05,
      "loss": 2.193,
      "step": 2142
    },
    {
      "epoch": 0.05643929417961548,
      "grad_norm": 2.992194414138794,
      "learning_rate": 4.718330260732157e-05,
      "loss": 1.1533,
      "step": 2143
    },
    {
      "epoch": 0.056465630761127206,
      "grad_norm": 2.0007081031799316,
      "learning_rate": 4.7181985778245987e-05,
      "loss": 1.4728,
      "step": 2144
    },
    {
      "epoch": 0.05649196734263893,
      "grad_norm": 3.823934316635132,
      "learning_rate": 4.7180668949170395e-05,
      "loss": 0.9316,
      "step": 2145
    },
    {
      "epoch": 0.056518303924150644,
      "grad_norm": 2.3916842937469482,
      "learning_rate": 4.717935212009482e-05,
      "loss": 1.9451,
      "step": 2146
    },
    {
      "epoch": 0.056544640505662366,
      "grad_norm": 4.011995792388916,
      "learning_rate": 4.717803529101923e-05,
      "loss": 1.6147,
      "step": 2147
    },
    {
      "epoch": 0.05657097708717408,
      "grad_norm": 2.2616333961486816,
      "learning_rate": 4.717671846194364e-05,
      "loss": 1.6991,
      "step": 2148
    },
    {
      "epoch": 0.056597313668685804,
      "grad_norm": 5.333600044250488,
      "learning_rate": 4.717540163286806e-05,
      "loss": 2.1645,
      "step": 2149
    },
    {
      "epoch": 0.05662365025019753,
      "grad_norm": 4.291018009185791,
      "learning_rate": 4.717408480379247e-05,
      "loss": 2.3565,
      "step": 2150
    },
    {
      "epoch": 0.05664998683170924,
      "grad_norm": 2.2324442863464355,
      "learning_rate": 4.717276797471689e-05,
      "loss": 2.1599,
      "step": 2151
    },
    {
      "epoch": 0.056676323413220965,
      "grad_norm": 3.4782960414886475,
      "learning_rate": 4.71714511456413e-05,
      "loss": 0.9414,
      "step": 2152
    },
    {
      "epoch": 0.05670265999473268,
      "grad_norm": 5.408863067626953,
      "learning_rate": 4.7170134316565713e-05,
      "loss": 1.6579,
      "step": 2153
    },
    {
      "epoch": 0.0567289965762444,
      "grad_norm": 3.2574551105499268,
      "learning_rate": 4.716881748749012e-05,
      "loss": 2.3205,
      "step": 2154
    },
    {
      "epoch": 0.056755333157756126,
      "grad_norm": 2.595188856124878,
      "learning_rate": 4.716750065841454e-05,
      "loss": 0.7521,
      "step": 2155
    },
    {
      "epoch": 0.05678166973926784,
      "grad_norm": 5.192615509033203,
      "learning_rate": 4.7166183829338953e-05,
      "loss": 1.6112,
      "step": 2156
    },
    {
      "epoch": 0.056808006320779564,
      "grad_norm": 2.759761095046997,
      "learning_rate": 4.716486700026337e-05,
      "loss": 1.6369,
      "step": 2157
    },
    {
      "epoch": 0.05683434290229128,
      "grad_norm": 3.5821824073791504,
      "learning_rate": 4.7163550171187785e-05,
      "loss": 2.138,
      "step": 2158
    },
    {
      "epoch": 0.056860679483803,
      "grad_norm": 2.3638978004455566,
      "learning_rate": 4.7162233342112193e-05,
      "loss": 1.6788,
      "step": 2159
    },
    {
      "epoch": 0.056887016065314724,
      "grad_norm": 2.5548527240753174,
      "learning_rate": 4.7160916513036616e-05,
      "loss": 1.9895,
      "step": 2160
    },
    {
      "epoch": 0.05691335264682644,
      "grad_norm": 3.196124792098999,
      "learning_rate": 4.7159599683961025e-05,
      "loss": 1.556,
      "step": 2161
    },
    {
      "epoch": 0.05693968922833816,
      "grad_norm": 2.3067831993103027,
      "learning_rate": 4.715828285488544e-05,
      "loss": 2.3408,
      "step": 2162
    },
    {
      "epoch": 0.05696602580984988,
      "grad_norm": 3.1644434928894043,
      "learning_rate": 4.715696602580985e-05,
      "loss": 1.891,
      "step": 2163
    },
    {
      "epoch": 0.0569923623913616,
      "grad_norm": 2.0824637413024902,
      "learning_rate": 4.7155649196734265e-05,
      "loss": 1.6948,
      "step": 2164
    },
    {
      "epoch": 0.05701869897287332,
      "grad_norm": 4.186317443847656,
      "learning_rate": 4.715433236765868e-05,
      "loss": 0.397,
      "step": 2165
    },
    {
      "epoch": 0.05704503555438504,
      "grad_norm": 3.7899069786071777,
      "learning_rate": 4.7153015538583096e-05,
      "loss": 0.8298,
      "step": 2166
    },
    {
      "epoch": 0.05707137213589676,
      "grad_norm": 4.509664535522461,
      "learning_rate": 4.7151698709507505e-05,
      "loss": 1.593,
      "step": 2167
    },
    {
      "epoch": 0.057097708717408484,
      "grad_norm": 3.0293211936950684,
      "learning_rate": 4.715038188043192e-05,
      "loss": 1.2093,
      "step": 2168
    },
    {
      "epoch": 0.0571240452989202,
      "grad_norm": 3.6300442218780518,
      "learning_rate": 4.7149065051356336e-05,
      "loss": 2.1209,
      "step": 2169
    },
    {
      "epoch": 0.05715038188043192,
      "grad_norm": 1.6079756021499634,
      "learning_rate": 4.714774822228075e-05,
      "loss": 1.7361,
      "step": 2170
    },
    {
      "epoch": 0.05717671846194364,
      "grad_norm": 2.393828868865967,
      "learning_rate": 4.714643139320517e-05,
      "loss": 1.1658,
      "step": 2171
    },
    {
      "epoch": 0.05720305504345536,
      "grad_norm": 2.2417051792144775,
      "learning_rate": 4.7145114564129576e-05,
      "loss": 2.2137,
      "step": 2172
    },
    {
      "epoch": 0.05722939162496708,
      "grad_norm": 3.120640516281128,
      "learning_rate": 4.714379773505399e-05,
      "loss": 1.5996,
      "step": 2173
    },
    {
      "epoch": 0.0572557282064788,
      "grad_norm": 5.414018630981445,
      "learning_rate": 4.714248090597841e-05,
      "loss": 1.99,
      "step": 2174
    },
    {
      "epoch": 0.05728206478799052,
      "grad_norm": 2.18473219871521,
      "learning_rate": 4.714116407690282e-05,
      "loss": 1.4559,
      "step": 2175
    },
    {
      "epoch": 0.057308401369502236,
      "grad_norm": 2.0574967861175537,
      "learning_rate": 4.713984724782723e-05,
      "loss": 1.3193,
      "step": 2176
    },
    {
      "epoch": 0.05733473795101396,
      "grad_norm": 2.525825262069702,
      "learning_rate": 4.713853041875165e-05,
      "loss": 0.5651,
      "step": 2177
    },
    {
      "epoch": 0.05736107453252568,
      "grad_norm": 2.9111571311950684,
      "learning_rate": 4.713721358967606e-05,
      "loss": 1.3808,
      "step": 2178
    },
    {
      "epoch": 0.0573874111140374,
      "grad_norm": 10.607233047485352,
      "learning_rate": 4.713589676060048e-05,
      "loss": 2.1026,
      "step": 2179
    },
    {
      "epoch": 0.05741374769554912,
      "grad_norm": 2.3342490196228027,
      "learning_rate": 4.7134579931524894e-05,
      "loss": 1.5054,
      "step": 2180
    },
    {
      "epoch": 0.057440084277060835,
      "grad_norm": 5.707728862762451,
      "learning_rate": 4.71332631024493e-05,
      "loss": 1.3972,
      "step": 2181
    },
    {
      "epoch": 0.05746642085857256,
      "grad_norm": 2.1899068355560303,
      "learning_rate": 4.713194627337372e-05,
      "loss": 1.7015,
      "step": 2182
    },
    {
      "epoch": 0.05749275744008428,
      "grad_norm": 2.359449863433838,
      "learning_rate": 4.713062944429813e-05,
      "loss": 1.6338,
      "step": 2183
    },
    {
      "epoch": 0.057519094021595996,
      "grad_norm": 1.9393079280853271,
      "learning_rate": 4.712931261522255e-05,
      "loss": 2.5717,
      "step": 2184
    },
    {
      "epoch": 0.05754543060310772,
      "grad_norm": 5.754360675811768,
      "learning_rate": 4.712799578614696e-05,
      "loss": 1.4453,
      "step": 2185
    },
    {
      "epoch": 0.057571767184619434,
      "grad_norm": 7.439517021179199,
      "learning_rate": 4.7126678957071374e-05,
      "loss": 1.2166,
      "step": 2186
    },
    {
      "epoch": 0.057598103766131156,
      "grad_norm": 3.0107805728912354,
      "learning_rate": 4.712536212799579e-05,
      "loss": 2.4133,
      "step": 2187
    },
    {
      "epoch": 0.05762444034764288,
      "grad_norm": 2.320903778076172,
      "learning_rate": 4.71240452989202e-05,
      "loss": 1.4147,
      "step": 2188
    },
    {
      "epoch": 0.057650776929154594,
      "grad_norm": 2.8712406158447266,
      "learning_rate": 4.712272846984462e-05,
      "loss": 1.2879,
      "step": 2189
    },
    {
      "epoch": 0.05767711351066632,
      "grad_norm": 2.044370651245117,
      "learning_rate": 4.712141164076903e-05,
      "loss": 1.1372,
      "step": 2190
    },
    {
      "epoch": 0.05770345009217803,
      "grad_norm": 4.43475341796875,
      "learning_rate": 4.7120094811693445e-05,
      "loss": 0.8103,
      "step": 2191
    },
    {
      "epoch": 0.057729786673689755,
      "grad_norm": 1.9338033199310303,
      "learning_rate": 4.7118777982617854e-05,
      "loss": 1.3003,
      "step": 2192
    },
    {
      "epoch": 0.05775612325520148,
      "grad_norm": 2.5325772762298584,
      "learning_rate": 4.7117461153542276e-05,
      "loss": 1.4003,
      "step": 2193
    },
    {
      "epoch": 0.05778245983671319,
      "grad_norm": 3.6060450077056885,
      "learning_rate": 4.7116144324466685e-05,
      "loss": 1.9385,
      "step": 2194
    },
    {
      "epoch": 0.057808796418224916,
      "grad_norm": 2.596099615097046,
      "learning_rate": 4.71148274953911e-05,
      "loss": 1.5216,
      "step": 2195
    },
    {
      "epoch": 0.05783513299973663,
      "grad_norm": 2.9077229499816895,
      "learning_rate": 4.7113510666315516e-05,
      "loss": 1.9273,
      "step": 2196
    },
    {
      "epoch": 0.057861469581248354,
      "grad_norm": 2.57293701171875,
      "learning_rate": 4.7112193837239925e-05,
      "loss": 1.8177,
      "step": 2197
    },
    {
      "epoch": 0.057887806162760076,
      "grad_norm": 4.707583427429199,
      "learning_rate": 4.711087700816435e-05,
      "loss": 1.2391,
      "step": 2198
    },
    {
      "epoch": 0.05791414274427179,
      "grad_norm": 3.1155073642730713,
      "learning_rate": 4.7109560179088756e-05,
      "loss": 1.5555,
      "step": 2199
    },
    {
      "epoch": 0.057940479325783514,
      "grad_norm": 2.756563425064087,
      "learning_rate": 4.710824335001317e-05,
      "loss": 1.2731,
      "step": 2200
    },
    {
      "epoch": 0.05796681590729523,
      "grad_norm": 3.910980224609375,
      "learning_rate": 4.710692652093758e-05,
      "loss": 1.1219,
      "step": 2201
    },
    {
      "epoch": 0.05799315248880695,
      "grad_norm": 3.0736021995544434,
      "learning_rate": 4.7105609691861996e-05,
      "loss": 2.3444,
      "step": 2202
    },
    {
      "epoch": 0.058019489070318675,
      "grad_norm": 2.265638828277588,
      "learning_rate": 4.710429286278641e-05,
      "loss": 2.2269,
      "step": 2203
    },
    {
      "epoch": 0.05804582565183039,
      "grad_norm": 3.4562597274780273,
      "learning_rate": 4.710297603371083e-05,
      "loss": 1.4524,
      "step": 2204
    },
    {
      "epoch": 0.05807216223334211,
      "grad_norm": 2.9995017051696777,
      "learning_rate": 4.710165920463524e-05,
      "loss": 1.5533,
      "step": 2205
    },
    {
      "epoch": 0.05809849881485383,
      "grad_norm": 4.540975570678711,
      "learning_rate": 4.710034237555965e-05,
      "loss": 1.4094,
      "step": 2206
    },
    {
      "epoch": 0.05812483539636555,
      "grad_norm": 1.7497889995574951,
      "learning_rate": 4.7099025546484074e-05,
      "loss": 1.9038,
      "step": 2207
    },
    {
      "epoch": 0.058151171977877274,
      "grad_norm": 2.1031877994537354,
      "learning_rate": 4.709770871740848e-05,
      "loss": 2.0104,
      "step": 2208
    },
    {
      "epoch": 0.05817750855938899,
      "grad_norm": 2.4106247425079346,
      "learning_rate": 4.70963918883329e-05,
      "loss": 2.0538,
      "step": 2209
    },
    {
      "epoch": 0.05820384514090071,
      "grad_norm": 1.9706648588180542,
      "learning_rate": 4.709507505925731e-05,
      "loss": 1.5769,
      "step": 2210
    },
    {
      "epoch": 0.05823018172241243,
      "grad_norm": 2.4568169116973877,
      "learning_rate": 4.709375823018172e-05,
      "loss": 0.731,
      "step": 2211
    },
    {
      "epoch": 0.05825651830392415,
      "grad_norm": 4.356393814086914,
      "learning_rate": 4.709244140110614e-05,
      "loss": 1.8031,
      "step": 2212
    },
    {
      "epoch": 0.05828285488543587,
      "grad_norm": 1.876869559288025,
      "learning_rate": 4.7091124572030554e-05,
      "loss": 1.6414,
      "step": 2213
    },
    {
      "epoch": 0.05830919146694759,
      "grad_norm": 2.18074631690979,
      "learning_rate": 4.708980774295497e-05,
      "loss": 2.3957,
      "step": 2214
    },
    {
      "epoch": 0.05833552804845931,
      "grad_norm": 2.849780321121216,
      "learning_rate": 4.708849091387938e-05,
      "loss": 2.2857,
      "step": 2215
    },
    {
      "epoch": 0.05836186462997103,
      "grad_norm": 2.6972780227661133,
      "learning_rate": 4.7087174084803794e-05,
      "loss": 2.3248,
      "step": 2216
    },
    {
      "epoch": 0.05838820121148275,
      "grad_norm": 1.7781732082366943,
      "learning_rate": 4.708585725572821e-05,
      "loss": 1.1056,
      "step": 2217
    },
    {
      "epoch": 0.05841453779299447,
      "grad_norm": 2.488926649093628,
      "learning_rate": 4.7084540426652626e-05,
      "loss": 1.7901,
      "step": 2218
    },
    {
      "epoch": 0.05844087437450619,
      "grad_norm": 2.4412338733673096,
      "learning_rate": 4.7083223597577034e-05,
      "loss": 2.3414,
      "step": 2219
    },
    {
      "epoch": 0.05846721095601791,
      "grad_norm": 2.6597626209259033,
      "learning_rate": 4.708190676850145e-05,
      "loss": 1.9127,
      "step": 2220
    },
    {
      "epoch": 0.05849354753752963,
      "grad_norm": 3.2455780506134033,
      "learning_rate": 4.708058993942586e-05,
      "loss": 1.54,
      "step": 2221
    },
    {
      "epoch": 0.05851988411904135,
      "grad_norm": 1.9604851007461548,
      "learning_rate": 4.707927311035028e-05,
      "loss": 2.1296,
      "step": 2222
    },
    {
      "epoch": 0.05854622070055307,
      "grad_norm": 1.9545475244522095,
      "learning_rate": 4.707795628127469e-05,
      "loss": 1.8768,
      "step": 2223
    },
    {
      "epoch": 0.058572557282064786,
      "grad_norm": 4.185596466064453,
      "learning_rate": 4.7076639452199106e-05,
      "loss": 1.928,
      "step": 2224
    },
    {
      "epoch": 0.05859889386357651,
      "grad_norm": 2.5845375061035156,
      "learning_rate": 4.707532262312352e-05,
      "loss": 2.0505,
      "step": 2225
    },
    {
      "epoch": 0.05862523044508823,
      "grad_norm": 2.424807548522949,
      "learning_rate": 4.707400579404794e-05,
      "loss": 1.1109,
      "step": 2226
    },
    {
      "epoch": 0.058651567026599946,
      "grad_norm": 2.557514190673828,
      "learning_rate": 4.707268896497235e-05,
      "loss": 2.0236,
      "step": 2227
    },
    {
      "epoch": 0.05867790360811167,
      "grad_norm": 2.9080758094787598,
      "learning_rate": 4.707137213589676e-05,
      "loss": 1.4031,
      "step": 2228
    },
    {
      "epoch": 0.058704240189623384,
      "grad_norm": 3.306455135345459,
      "learning_rate": 4.707005530682118e-05,
      "loss": 0.3484,
      "step": 2229
    },
    {
      "epoch": 0.05873057677113511,
      "grad_norm": 3.528730630874634,
      "learning_rate": 4.7068738477745586e-05,
      "loss": 1.573,
      "step": 2230
    },
    {
      "epoch": 0.05875691335264683,
      "grad_norm": 2.327526330947876,
      "learning_rate": 4.706742164867001e-05,
      "loss": 1.4864,
      "step": 2231
    },
    {
      "epoch": 0.058783249934158545,
      "grad_norm": 2.211899518966675,
      "learning_rate": 4.706610481959442e-05,
      "loss": 2.1208,
      "step": 2232
    },
    {
      "epoch": 0.05880958651567027,
      "grad_norm": 3.8888766765594482,
      "learning_rate": 4.706478799051883e-05,
      "loss": 1.3574,
      "step": 2233
    },
    {
      "epoch": 0.05883592309718198,
      "grad_norm": 1.8692330121994019,
      "learning_rate": 4.706347116144325e-05,
      "loss": 1.9271,
      "step": 2234
    },
    {
      "epoch": 0.058862259678693706,
      "grad_norm": 3.9838359355926514,
      "learning_rate": 4.706215433236766e-05,
      "loss": 2.0212,
      "step": 2235
    },
    {
      "epoch": 0.05888859626020543,
      "grad_norm": 2.807814598083496,
      "learning_rate": 4.706083750329208e-05,
      "loss": 1.8291,
      "step": 2236
    },
    {
      "epoch": 0.058914932841717144,
      "grad_norm": 2.5180716514587402,
      "learning_rate": 4.705952067421649e-05,
      "loss": 1.3584,
      "step": 2237
    },
    {
      "epoch": 0.058941269423228866,
      "grad_norm": 4.121596336364746,
      "learning_rate": 4.7058203845140904e-05,
      "loss": 2.1318,
      "step": 2238
    },
    {
      "epoch": 0.05896760600474058,
      "grad_norm": 2.9912428855895996,
      "learning_rate": 4.705688701606531e-05,
      "loss": 1.5819,
      "step": 2239
    },
    {
      "epoch": 0.058993942586252304,
      "grad_norm": 2.5563719272613525,
      "learning_rate": 4.7055570186989735e-05,
      "loss": 2.1908,
      "step": 2240
    },
    {
      "epoch": 0.05902027916776403,
      "grad_norm": 2.503976821899414,
      "learning_rate": 4.7054253357914144e-05,
      "loss": 2.4987,
      "step": 2241
    },
    {
      "epoch": 0.05904661574927574,
      "grad_norm": 1.8304951190948486,
      "learning_rate": 4.705293652883856e-05,
      "loss": 1.6273,
      "step": 2242
    },
    {
      "epoch": 0.059072952330787465,
      "grad_norm": 4.101771354675293,
      "learning_rate": 4.7051619699762975e-05,
      "loss": 1.5418,
      "step": 2243
    },
    {
      "epoch": 0.05909928891229918,
      "grad_norm": 2.231008529663086,
      "learning_rate": 4.7050302870687384e-05,
      "loss": 1.3712,
      "step": 2244
    },
    {
      "epoch": 0.0591256254938109,
      "grad_norm": 2.1481590270996094,
      "learning_rate": 4.7048986041611806e-05,
      "loss": 1.4567,
      "step": 2245
    },
    {
      "epoch": 0.059151962075322626,
      "grad_norm": 2.145627975463867,
      "learning_rate": 4.7047669212536215e-05,
      "loss": 2.6714,
      "step": 2246
    },
    {
      "epoch": 0.05917829865683434,
      "grad_norm": 2.559617280960083,
      "learning_rate": 4.704635238346063e-05,
      "loss": 1.5486,
      "step": 2247
    },
    {
      "epoch": 0.059204635238346064,
      "grad_norm": 2.089916467666626,
      "learning_rate": 4.704503555438504e-05,
      "loss": 1.7973,
      "step": 2248
    },
    {
      "epoch": 0.05923097181985778,
      "grad_norm": 1.8132014274597168,
      "learning_rate": 4.7043718725309455e-05,
      "loss": 1.8544,
      "step": 2249
    },
    {
      "epoch": 0.0592573084013695,
      "grad_norm": 2.802475929260254,
      "learning_rate": 4.704240189623387e-05,
      "loss": 1.7482,
      "step": 2250
    },
    {
      "epoch": 0.059283644982881224,
      "grad_norm": 5.057201862335205,
      "learning_rate": 4.7041085067158286e-05,
      "loss": 0.9728,
      "step": 2251
    },
    {
      "epoch": 0.05930998156439294,
      "grad_norm": 2.865506410598755,
      "learning_rate": 4.70397682380827e-05,
      "loss": 2.0518,
      "step": 2252
    },
    {
      "epoch": 0.05933631814590466,
      "grad_norm": 2.898247003555298,
      "learning_rate": 4.703845140900711e-05,
      "loss": 1.0441,
      "step": 2253
    },
    {
      "epoch": 0.05936265472741638,
      "grad_norm": 2.4292194843292236,
      "learning_rate": 4.7037134579931526e-05,
      "loss": 0.4388,
      "step": 2254
    },
    {
      "epoch": 0.0593889913089281,
      "grad_norm": 8.985533714294434,
      "learning_rate": 4.703581775085594e-05,
      "loss": 1.8704,
      "step": 2255
    },
    {
      "epoch": 0.05941532789043982,
      "grad_norm": 2.1336731910705566,
      "learning_rate": 4.703450092178036e-05,
      "loss": 0.6001,
      "step": 2256
    },
    {
      "epoch": 0.05944166447195154,
      "grad_norm": 2.3788089752197266,
      "learning_rate": 4.7033184092704766e-05,
      "loss": 1.8919,
      "step": 2257
    },
    {
      "epoch": 0.05946800105346326,
      "grad_norm": 2.526772975921631,
      "learning_rate": 4.703186726362918e-05,
      "loss": 1.7069,
      "step": 2258
    },
    {
      "epoch": 0.059494337634974984,
      "grad_norm": 2.207334280014038,
      "learning_rate": 4.70305504345536e-05,
      "loss": 1.4619,
      "step": 2259
    },
    {
      "epoch": 0.0595206742164867,
      "grad_norm": 3.2663016319274902,
      "learning_rate": 4.702923360547801e-05,
      "loss": 1.6889,
      "step": 2260
    },
    {
      "epoch": 0.05954701079799842,
      "grad_norm": 3.2809529304504395,
      "learning_rate": 4.702791677640243e-05,
      "loss": 2.2813,
      "step": 2261
    },
    {
      "epoch": 0.05957334737951014,
      "grad_norm": 3.178739547729492,
      "learning_rate": 4.702659994732684e-05,
      "loss": 1.358,
      "step": 2262
    },
    {
      "epoch": 0.05959968396102186,
      "grad_norm": 2.3345863819122314,
      "learning_rate": 4.702528311825125e-05,
      "loss": 2.1037,
      "step": 2263
    },
    {
      "epoch": 0.05962602054253358,
      "grad_norm": 4.884681224822998,
      "learning_rate": 4.702396628917567e-05,
      "loss": 0.7522,
      "step": 2264
    },
    {
      "epoch": 0.0596523571240453,
      "grad_norm": 3.6622402667999268,
      "learning_rate": 4.7022649460100084e-05,
      "loss": 1.4466,
      "step": 2265
    },
    {
      "epoch": 0.05967869370555702,
      "grad_norm": 2.3584275245666504,
      "learning_rate": 4.702133263102449e-05,
      "loss": 1.9784,
      "step": 2266
    },
    {
      "epoch": 0.059705030287068736,
      "grad_norm": 2.529788017272949,
      "learning_rate": 4.702001580194891e-05,
      "loss": 1.7532,
      "step": 2267
    },
    {
      "epoch": 0.05973136686858046,
      "grad_norm": 4.265054225921631,
      "learning_rate": 4.701869897287332e-05,
      "loss": 1.5829,
      "step": 2268
    },
    {
      "epoch": 0.05975770345009218,
      "grad_norm": 1.83254075050354,
      "learning_rate": 4.701738214379774e-05,
      "loss": 1.7224,
      "step": 2269
    },
    {
      "epoch": 0.0597840400316039,
      "grad_norm": 2.8131589889526367,
      "learning_rate": 4.701606531472215e-05,
      "loss": 1.5929,
      "step": 2270
    },
    {
      "epoch": 0.05981037661311562,
      "grad_norm": 3.6532633304595947,
      "learning_rate": 4.7014748485646564e-05,
      "loss": 1.2478,
      "step": 2271
    },
    {
      "epoch": 0.059836713194627335,
      "grad_norm": 2.2759995460510254,
      "learning_rate": 4.701343165657098e-05,
      "loss": 1.5491,
      "step": 2272
    },
    {
      "epoch": 0.05986304977613906,
      "grad_norm": 2.3043057918548584,
      "learning_rate": 4.7012114827495395e-05,
      "loss": 1.8527,
      "step": 2273
    },
    {
      "epoch": 0.05988938635765078,
      "grad_norm": 2.1862733364105225,
      "learning_rate": 4.701079799841981e-05,
      "loss": 1.7428,
      "step": 2274
    },
    {
      "epoch": 0.059915722939162495,
      "grad_norm": 3.3454487323760986,
      "learning_rate": 4.700948116934422e-05,
      "loss": 1.7384,
      "step": 2275
    },
    {
      "epoch": 0.05994205952067422,
      "grad_norm": 3.5950281620025635,
      "learning_rate": 4.7008164340268635e-05,
      "loss": 1.4028,
      "step": 2276
    },
    {
      "epoch": 0.059968396102185934,
      "grad_norm": 4.540416717529297,
      "learning_rate": 4.7006847511193044e-05,
      "loss": 1.705,
      "step": 2277
    },
    {
      "epoch": 0.059994732683697656,
      "grad_norm": 2.2978692054748535,
      "learning_rate": 4.700553068211747e-05,
      "loss": 2.5032,
      "step": 2278
    },
    {
      "epoch": 0.06002106926520938,
      "grad_norm": 1.6918925046920776,
      "learning_rate": 4.7004213853041875e-05,
      "loss": 1.5561,
      "step": 2279
    },
    {
      "epoch": 0.060047405846721094,
      "grad_norm": 2.237168312072754,
      "learning_rate": 4.700289702396629e-05,
      "loss": 1.4813,
      "step": 2280
    },
    {
      "epoch": 0.06007374242823282,
      "grad_norm": 3.1560251712799072,
      "learning_rate": 4.700158019489071e-05,
      "loss": 2.2517,
      "step": 2281
    },
    {
      "epoch": 0.06010007900974453,
      "grad_norm": 3.299541473388672,
      "learning_rate": 4.7000263365815115e-05,
      "loss": 2.9007,
      "step": 2282
    },
    {
      "epoch": 0.060126415591256255,
      "grad_norm": 2.6424081325531006,
      "learning_rate": 4.699894653673954e-05,
      "loss": 1.9763,
      "step": 2283
    },
    {
      "epoch": 0.06015275217276798,
      "grad_norm": 2.60740065574646,
      "learning_rate": 4.699762970766395e-05,
      "loss": 2.1791,
      "step": 2284
    },
    {
      "epoch": 0.06017908875427969,
      "grad_norm": 3.41214656829834,
      "learning_rate": 4.699631287858836e-05,
      "loss": 1.596,
      "step": 2285
    },
    {
      "epoch": 0.060205425335791415,
      "grad_norm": 2.762927532196045,
      "learning_rate": 4.699499604951277e-05,
      "loss": 1.9012,
      "step": 2286
    },
    {
      "epoch": 0.06023176191730313,
      "grad_norm": 2.6899235248565674,
      "learning_rate": 4.699367922043719e-05,
      "loss": 1.2745,
      "step": 2287
    },
    {
      "epoch": 0.060258098498814854,
      "grad_norm": 1.9002881050109863,
      "learning_rate": 4.69923623913616e-05,
      "loss": 1.752,
      "step": 2288
    },
    {
      "epoch": 0.060284435080326576,
      "grad_norm": 4.1252899169921875,
      "learning_rate": 4.699104556228602e-05,
      "loss": 1.3271,
      "step": 2289
    },
    {
      "epoch": 0.06031077166183829,
      "grad_norm": 2.194669723510742,
      "learning_rate": 4.6989728733210433e-05,
      "loss": 1.9517,
      "step": 2290
    },
    {
      "epoch": 0.060337108243350014,
      "grad_norm": 2.555772066116333,
      "learning_rate": 4.698841190413484e-05,
      "loss": 1.4758,
      "step": 2291
    },
    {
      "epoch": 0.06036344482486173,
      "grad_norm": 2.6042699813842773,
      "learning_rate": 4.6987095075059265e-05,
      "loss": 1.7974,
      "step": 2292
    },
    {
      "epoch": 0.06038978140637345,
      "grad_norm": 2.769167184829712,
      "learning_rate": 4.6985778245983673e-05,
      "loss": 1.5591,
      "step": 2293
    },
    {
      "epoch": 0.060416117987885175,
      "grad_norm": 5.088955402374268,
      "learning_rate": 4.698446141690809e-05,
      "loss": 1.0187,
      "step": 2294
    },
    {
      "epoch": 0.06044245456939689,
      "grad_norm": 2.3629088401794434,
      "learning_rate": 4.69831445878325e-05,
      "loss": 1.4315,
      "step": 2295
    },
    {
      "epoch": 0.06046879115090861,
      "grad_norm": 2.3658950328826904,
      "learning_rate": 4.6981827758756914e-05,
      "loss": 1.578,
      "step": 2296
    },
    {
      "epoch": 0.06049512773242033,
      "grad_norm": 1.7849318981170654,
      "learning_rate": 4.698051092968133e-05,
      "loss": 2.0656,
      "step": 2297
    },
    {
      "epoch": 0.06052146431393205,
      "grad_norm": 3.287752866744995,
      "learning_rate": 4.6979194100605745e-05,
      "loss": 1.6277,
      "step": 2298
    },
    {
      "epoch": 0.060547800895443774,
      "grad_norm": 2.5434162616729736,
      "learning_rate": 4.697787727153016e-05,
      "loss": 1.6112,
      "step": 2299
    },
    {
      "epoch": 0.06057413747695549,
      "grad_norm": 3.7894043922424316,
      "learning_rate": 4.697656044245457e-05,
      "loss": 2.4612,
      "step": 2300
    },
    {
      "epoch": 0.06060047405846721,
      "grad_norm": 2.033865451812744,
      "learning_rate": 4.6975243613378985e-05,
      "loss": 1.8314,
      "step": 2301
    },
    {
      "epoch": 0.060626810639978934,
      "grad_norm": 2.481954574584961,
      "learning_rate": 4.69739267843034e-05,
      "loss": 1.2539,
      "step": 2302
    },
    {
      "epoch": 0.06065314722149065,
      "grad_norm": 3.2213282585144043,
      "learning_rate": 4.6972609955227816e-05,
      "loss": 0.858,
      "step": 2303
    },
    {
      "epoch": 0.06067948380300237,
      "grad_norm": 3.107637882232666,
      "learning_rate": 4.6971293126152225e-05,
      "loss": 1.984,
      "step": 2304
    },
    {
      "epoch": 0.06070582038451409,
      "grad_norm": 2.3892407417297363,
      "learning_rate": 4.696997629707664e-05,
      "loss": 1.4793,
      "step": 2305
    },
    {
      "epoch": 0.06073215696602581,
      "grad_norm": 2.00102162361145,
      "learning_rate": 4.6968659468001056e-05,
      "loss": 1.6171,
      "step": 2306
    },
    {
      "epoch": 0.06075849354753753,
      "grad_norm": 1.9140921831130981,
      "learning_rate": 4.696734263892547e-05,
      "loss": 1.5184,
      "step": 2307
    },
    {
      "epoch": 0.06078483012904925,
      "grad_norm": 2.6027801036834717,
      "learning_rate": 4.696602580984989e-05,
      "loss": 1.4163,
      "step": 2308
    },
    {
      "epoch": 0.06081116671056097,
      "grad_norm": 3.8211586475372314,
      "learning_rate": 4.6964708980774296e-05,
      "loss": 1.7563,
      "step": 2309
    },
    {
      "epoch": 0.06083750329207269,
      "grad_norm": 3.426889657974243,
      "learning_rate": 4.696339215169871e-05,
      "loss": 1.4254,
      "step": 2310
    },
    {
      "epoch": 0.06086383987358441,
      "grad_norm": 2.7382829189300537,
      "learning_rate": 4.696207532262313e-05,
      "loss": 1.942,
      "step": 2311
    },
    {
      "epoch": 0.06089017645509613,
      "grad_norm": 4.564896106719971,
      "learning_rate": 4.696075849354754e-05,
      "loss": 1.7901,
      "step": 2312
    },
    {
      "epoch": 0.06091651303660785,
      "grad_norm": 2.068636178970337,
      "learning_rate": 4.695944166447195e-05,
      "loss": 1.6466,
      "step": 2313
    },
    {
      "epoch": 0.06094284961811957,
      "grad_norm": 2.961036443710327,
      "learning_rate": 4.695812483539637e-05,
      "loss": 0.681,
      "step": 2314
    },
    {
      "epoch": 0.060969186199631285,
      "grad_norm": 2.3746185302734375,
      "learning_rate": 4.6956808006320776e-05,
      "loss": 1.4592,
      "step": 2315
    },
    {
      "epoch": 0.06099552278114301,
      "grad_norm": 2.8096494674682617,
      "learning_rate": 4.69554911772452e-05,
      "loss": 1.7898,
      "step": 2316
    },
    {
      "epoch": 0.06102185936265473,
      "grad_norm": 2.3663113117218018,
      "learning_rate": 4.6954174348169614e-05,
      "loss": 1.999,
      "step": 2317
    },
    {
      "epoch": 0.061048195944166446,
      "grad_norm": 2.512408971786499,
      "learning_rate": 4.695285751909402e-05,
      "loss": 2.1202,
      "step": 2318
    },
    {
      "epoch": 0.06107453252567817,
      "grad_norm": 2.44903302192688,
      "learning_rate": 4.695154069001844e-05,
      "loss": 2.0955,
      "step": 2319
    },
    {
      "epoch": 0.061100869107189884,
      "grad_norm": 5.071829319000244,
      "learning_rate": 4.6950223860942854e-05,
      "loss": 1.3134,
      "step": 2320
    },
    {
      "epoch": 0.06112720568870161,
      "grad_norm": 2.7499380111694336,
      "learning_rate": 4.694890703186727e-05,
      "loss": 2.114,
      "step": 2321
    },
    {
      "epoch": 0.06115354227021333,
      "grad_norm": 3.3781466484069824,
      "learning_rate": 4.694759020279168e-05,
      "loss": 1.6296,
      "step": 2322
    },
    {
      "epoch": 0.061179878851725045,
      "grad_norm": 1.9659886360168457,
      "learning_rate": 4.6946273373716094e-05,
      "loss": 2.3236,
      "step": 2323
    },
    {
      "epoch": 0.06120621543323677,
      "grad_norm": 2.6231226921081543,
      "learning_rate": 4.69449565446405e-05,
      "loss": 2.2305,
      "step": 2324
    },
    {
      "epoch": 0.06123255201474848,
      "grad_norm": 2.6958329677581787,
      "learning_rate": 4.6943639715564925e-05,
      "loss": 1.122,
      "step": 2325
    },
    {
      "epoch": 0.061258888596260205,
      "grad_norm": 1.8875097036361694,
      "learning_rate": 4.6942322886489334e-05,
      "loss": 1.9176,
      "step": 2326
    },
    {
      "epoch": 0.06128522517777193,
      "grad_norm": 2.3418688774108887,
      "learning_rate": 4.694100605741375e-05,
      "loss": 1.4749,
      "step": 2327
    },
    {
      "epoch": 0.06131156175928364,
      "grad_norm": 4.613044261932373,
      "learning_rate": 4.6939689228338165e-05,
      "loss": 0.6705,
      "step": 2328
    },
    {
      "epoch": 0.061337898340795366,
      "grad_norm": 2.5555670261383057,
      "learning_rate": 4.6938372399262574e-05,
      "loss": 1.8351,
      "step": 2329
    },
    {
      "epoch": 0.06136423492230708,
      "grad_norm": 2.5072150230407715,
      "learning_rate": 4.6937055570186996e-05,
      "loss": 1.9891,
      "step": 2330
    },
    {
      "epoch": 0.061390571503818804,
      "grad_norm": 2.2406907081604004,
      "learning_rate": 4.6935738741111405e-05,
      "loss": 1.8428,
      "step": 2331
    },
    {
      "epoch": 0.06141690808533053,
      "grad_norm": 2.5359418392181396,
      "learning_rate": 4.693442191203582e-05,
      "loss": 1.0713,
      "step": 2332
    },
    {
      "epoch": 0.06144324466684224,
      "grad_norm": 3.020839214324951,
      "learning_rate": 4.693310508296023e-05,
      "loss": 1.6319,
      "step": 2333
    },
    {
      "epoch": 0.061469581248353965,
      "grad_norm": 2.4607772827148438,
      "learning_rate": 4.6931788253884645e-05,
      "loss": 2.2703,
      "step": 2334
    },
    {
      "epoch": 0.06149591782986568,
      "grad_norm": 2.916142463684082,
      "learning_rate": 4.693047142480906e-05,
      "loss": 1.6846,
      "step": 2335
    },
    {
      "epoch": 0.0615222544113774,
      "grad_norm": 2.408010721206665,
      "learning_rate": 4.6929154595733476e-05,
      "loss": 1.0948,
      "step": 2336
    },
    {
      "epoch": 0.061548590992889125,
      "grad_norm": 2.9166905879974365,
      "learning_rate": 4.692783776665789e-05,
      "loss": 1.607,
      "step": 2337
    },
    {
      "epoch": 0.06157492757440084,
      "grad_norm": 5.828845024108887,
      "learning_rate": 4.69265209375823e-05,
      "loss": 1.6196,
      "step": 2338
    },
    {
      "epoch": 0.06160126415591256,
      "grad_norm": 4.621794700622559,
      "learning_rate": 4.692520410850672e-05,
      "loss": 1.149,
      "step": 2339
    },
    {
      "epoch": 0.06162760073742428,
      "grad_norm": 3.325955390930176,
      "learning_rate": 4.692388727943113e-05,
      "loss": 1.0291,
      "step": 2340
    },
    {
      "epoch": 0.061653937318936,
      "grad_norm": 2.8587429523468018,
      "learning_rate": 4.692257045035555e-05,
      "loss": 2.1007,
      "step": 2341
    },
    {
      "epoch": 0.061680273900447724,
      "grad_norm": 3.480433464050293,
      "learning_rate": 4.6921253621279956e-05,
      "loss": 2.3088,
      "step": 2342
    },
    {
      "epoch": 0.06170661048195944,
      "grad_norm": 1.83161199092865,
      "learning_rate": 4.691993679220437e-05,
      "loss": 1.5615,
      "step": 2343
    },
    {
      "epoch": 0.06173294706347116,
      "grad_norm": 3.0452425479888916,
      "learning_rate": 4.691861996312879e-05,
      "loss": 1.9265,
      "step": 2344
    },
    {
      "epoch": 0.06175928364498288,
      "grad_norm": 3.0918824672698975,
      "learning_rate": 4.69173031340532e-05,
      "loss": 0.9461,
      "step": 2345
    },
    {
      "epoch": 0.0617856202264946,
      "grad_norm": 1.8512251377105713,
      "learning_rate": 4.691598630497762e-05,
      "loss": 1.2078,
      "step": 2346
    },
    {
      "epoch": 0.06181195680800632,
      "grad_norm": 3.598766326904297,
      "learning_rate": 4.691466947590203e-05,
      "loss": 1.4373,
      "step": 2347
    },
    {
      "epoch": 0.06183829338951804,
      "grad_norm": 2.05191707611084,
      "learning_rate": 4.691335264682644e-05,
      "loss": 1.6134,
      "step": 2348
    },
    {
      "epoch": 0.06186462997102976,
      "grad_norm": 2.3048503398895264,
      "learning_rate": 4.691203581775086e-05,
      "loss": 1.6072,
      "step": 2349
    },
    {
      "epoch": 0.06189096655254148,
      "grad_norm": 2.4440760612487793,
      "learning_rate": 4.6910718988675274e-05,
      "loss": 1.5599,
      "step": 2350
    },
    {
      "epoch": 0.0619173031340532,
      "grad_norm": 2.9048287868499756,
      "learning_rate": 4.690940215959968e-05,
      "loss": 2.4745,
      "step": 2351
    },
    {
      "epoch": 0.06194363971556492,
      "grad_norm": 3.846914529800415,
      "learning_rate": 4.69080853305241e-05,
      "loss": 0.8357,
      "step": 2352
    },
    {
      "epoch": 0.06196997629707664,
      "grad_norm": 2.4925453662872314,
      "learning_rate": 4.6906768501448514e-05,
      "loss": 2.2837,
      "step": 2353
    },
    {
      "epoch": 0.06199631287858836,
      "grad_norm": 8.011929512023926,
      "learning_rate": 4.690545167237293e-05,
      "loss": 1.6929,
      "step": 2354
    },
    {
      "epoch": 0.06202264946010008,
      "grad_norm": 2.592222213745117,
      "learning_rate": 4.6904134843297346e-05,
      "loss": 1.5417,
      "step": 2355
    },
    {
      "epoch": 0.0620489860416118,
      "grad_norm": 4.365649700164795,
      "learning_rate": 4.6902818014221755e-05,
      "loss": 1.6431,
      "step": 2356
    },
    {
      "epoch": 0.06207532262312352,
      "grad_norm": 2.6333377361297607,
      "learning_rate": 4.690150118514617e-05,
      "loss": 1.4739,
      "step": 2357
    },
    {
      "epoch": 0.062101659204635236,
      "grad_norm": 2.52740740776062,
      "learning_rate": 4.6900184356070586e-05,
      "loss": 2.8303,
      "step": 2358
    },
    {
      "epoch": 0.06212799578614696,
      "grad_norm": 2.5063893795013428,
      "learning_rate": 4.6898867526995e-05,
      "loss": 1.8748,
      "step": 2359
    },
    {
      "epoch": 0.06215433236765868,
      "grad_norm": 2.152168035507202,
      "learning_rate": 4.689755069791941e-05,
      "loss": 1.6331,
      "step": 2360
    },
    {
      "epoch": 0.062180668949170396,
      "grad_norm": 2.199742317199707,
      "learning_rate": 4.6896233868843826e-05,
      "loss": 2.1367,
      "step": 2361
    },
    {
      "epoch": 0.06220700553068212,
      "grad_norm": 6.430613994598389,
      "learning_rate": 4.689491703976824e-05,
      "loss": 1.0867,
      "step": 2362
    },
    {
      "epoch": 0.062233342112193835,
      "grad_norm": 2.8499956130981445,
      "learning_rate": 4.689360021069266e-05,
      "loss": 1.3986,
      "step": 2363
    },
    {
      "epoch": 0.06225967869370556,
      "grad_norm": 1.7732125520706177,
      "learning_rate": 4.689228338161707e-05,
      "loss": 1.6242,
      "step": 2364
    },
    {
      "epoch": 0.06228601527521728,
      "grad_norm": 2.204768657684326,
      "learning_rate": 4.689096655254148e-05,
      "loss": 2.108,
      "step": 2365
    },
    {
      "epoch": 0.062312351856728995,
      "grad_norm": 3.954988956451416,
      "learning_rate": 4.68896497234659e-05,
      "loss": 1.5833,
      "step": 2366
    },
    {
      "epoch": 0.06233868843824072,
      "grad_norm": 2.904905319213867,
      "learning_rate": 4.6888332894390306e-05,
      "loss": 2.2774,
      "step": 2367
    },
    {
      "epoch": 0.06236502501975243,
      "grad_norm": 1.716963291168213,
      "learning_rate": 4.688701606531473e-05,
      "loss": 1.7282,
      "step": 2368
    },
    {
      "epoch": 0.062391361601264156,
      "grad_norm": 3.9335405826568604,
      "learning_rate": 4.688569923623914e-05,
      "loss": 2.0746,
      "step": 2369
    },
    {
      "epoch": 0.06241769818277588,
      "grad_norm": 2.6345374584198,
      "learning_rate": 4.688438240716355e-05,
      "loss": 1.9933,
      "step": 2370
    },
    {
      "epoch": 0.062444034764287594,
      "grad_norm": 2.059537887573242,
      "learning_rate": 4.688306557808796e-05,
      "loss": 1.6789,
      "step": 2371
    },
    {
      "epoch": 0.062470371345799316,
      "grad_norm": 3.3977279663085938,
      "learning_rate": 4.6881748749012384e-05,
      "loss": 1.5188,
      "step": 2372
    },
    {
      "epoch": 0.06249670792731103,
      "grad_norm": 2.193197011947632,
      "learning_rate": 4.688043191993679e-05,
      "loss": 1.9147,
      "step": 2373
    },
    {
      "epoch": 0.06252304450882276,
      "grad_norm": 2.059917449951172,
      "learning_rate": 4.687911509086121e-05,
      "loss": 2.5271,
      "step": 2374
    },
    {
      "epoch": 0.06254938109033448,
      "grad_norm": 2.6211464405059814,
      "learning_rate": 4.6877798261785624e-05,
      "loss": 1.31,
      "step": 2375
    },
    {
      "epoch": 0.06257571767184619,
      "grad_norm": 5.8871169090271,
      "learning_rate": 4.687648143271003e-05,
      "loss": 0.9713,
      "step": 2376
    },
    {
      "epoch": 0.06260205425335791,
      "grad_norm": 4.014262676239014,
      "learning_rate": 4.6875164603634455e-05,
      "loss": 1.2031,
      "step": 2377
    },
    {
      "epoch": 0.06262839083486964,
      "grad_norm": 2.2473819255828857,
      "learning_rate": 4.6873847774558864e-05,
      "loss": 2.0133,
      "step": 2378
    },
    {
      "epoch": 0.06265472741638135,
      "grad_norm": 5.412081718444824,
      "learning_rate": 4.687253094548328e-05,
      "loss": 1.2997,
      "step": 2379
    },
    {
      "epoch": 0.06268106399789307,
      "grad_norm": 2.1226892471313477,
      "learning_rate": 4.687121411640769e-05,
      "loss": 1.6275,
      "step": 2380
    },
    {
      "epoch": 0.0627074005794048,
      "grad_norm": 3.127661943435669,
      "learning_rate": 4.6869897287332104e-05,
      "loss": 1.4595,
      "step": 2381
    },
    {
      "epoch": 0.06273373716091651,
      "grad_norm": 2.2085540294647217,
      "learning_rate": 4.686858045825652e-05,
      "loss": 1.445,
      "step": 2382
    },
    {
      "epoch": 0.06276007374242823,
      "grad_norm": 2.1729025840759277,
      "learning_rate": 4.6867263629180935e-05,
      "loss": 1.5951,
      "step": 2383
    },
    {
      "epoch": 0.06278641032393996,
      "grad_norm": 1.9573615789413452,
      "learning_rate": 4.686594680010535e-05,
      "loss": 1.9353,
      "step": 2384
    },
    {
      "epoch": 0.06281274690545167,
      "grad_norm": 2.1518776416778564,
      "learning_rate": 4.686462997102976e-05,
      "loss": 2.1263,
      "step": 2385
    },
    {
      "epoch": 0.06283908348696339,
      "grad_norm": 2.1036746501922607,
      "learning_rate": 4.686331314195418e-05,
      "loss": 1.6969,
      "step": 2386
    },
    {
      "epoch": 0.0628654200684751,
      "grad_norm": 2.245342493057251,
      "learning_rate": 4.686199631287859e-05,
      "loss": 1.7591,
      "step": 2387
    },
    {
      "epoch": 0.06289175664998684,
      "grad_norm": 2.0827269554138184,
      "learning_rate": 4.6860679483803006e-05,
      "loss": 0.5328,
      "step": 2388
    },
    {
      "epoch": 0.06291809323149855,
      "grad_norm": 2.7235958576202393,
      "learning_rate": 4.6859362654727415e-05,
      "loss": 1.3614,
      "step": 2389
    },
    {
      "epoch": 0.06294442981301027,
      "grad_norm": 7.018543243408203,
      "learning_rate": 4.685804582565183e-05,
      "loss": 1.3847,
      "step": 2390
    },
    {
      "epoch": 0.062970766394522,
      "grad_norm": 5.000580310821533,
      "learning_rate": 4.6856728996576246e-05,
      "loss": 1.9968,
      "step": 2391
    },
    {
      "epoch": 0.06299710297603371,
      "grad_norm": 5.311912536621094,
      "learning_rate": 4.685541216750066e-05,
      "loss": 1.8817,
      "step": 2392
    },
    {
      "epoch": 0.06302343955754543,
      "grad_norm": 3.7038862705230713,
      "learning_rate": 4.685409533842508e-05,
      "loss": 1.4629,
      "step": 2393
    },
    {
      "epoch": 0.06304977613905716,
      "grad_norm": 2.9747352600097656,
      "learning_rate": 4.6852778509349486e-05,
      "loss": 1.8721,
      "step": 2394
    },
    {
      "epoch": 0.06307611272056887,
      "grad_norm": 1.9817137718200684,
      "learning_rate": 4.68514616802739e-05,
      "loss": 1.8767,
      "step": 2395
    },
    {
      "epoch": 0.06310244930208059,
      "grad_norm": 2.0077948570251465,
      "learning_rate": 4.685014485119832e-05,
      "loss": 2.0026,
      "step": 2396
    },
    {
      "epoch": 0.0631287858835923,
      "grad_norm": 2.7527992725372314,
      "learning_rate": 4.684882802212273e-05,
      "loss": 1.8603,
      "step": 2397
    },
    {
      "epoch": 0.06315512246510403,
      "grad_norm": 4.314809799194336,
      "learning_rate": 4.684751119304714e-05,
      "loss": 0.5443,
      "step": 2398
    },
    {
      "epoch": 0.06318145904661575,
      "grad_norm": 3.1604745388031006,
      "learning_rate": 4.684619436397156e-05,
      "loss": 1.3714,
      "step": 2399
    },
    {
      "epoch": 0.06320779562812746,
      "grad_norm": 2.6723835468292236,
      "learning_rate": 4.684487753489597e-05,
      "loss": 0.4497,
      "step": 2400
    },
    {
      "epoch": 0.0632341322096392,
      "grad_norm": 2.9711673259735107,
      "learning_rate": 4.684356070582039e-05,
      "loss": 1.8005,
      "step": 2401
    },
    {
      "epoch": 0.06326046879115091,
      "grad_norm": 3.8346498012542725,
      "learning_rate": 4.6842243876744804e-05,
      "loss": 1.554,
      "step": 2402
    },
    {
      "epoch": 0.06328680537266262,
      "grad_norm": 2.6312010288238525,
      "learning_rate": 4.684092704766921e-05,
      "loss": 1.906,
      "step": 2403
    },
    {
      "epoch": 0.06331314195417435,
      "grad_norm": 2.83207368850708,
      "learning_rate": 4.683961021859363e-05,
      "loss": 2.0679,
      "step": 2404
    },
    {
      "epoch": 0.06333947853568607,
      "grad_norm": 7.46080207824707,
      "learning_rate": 4.6838293389518044e-05,
      "loss": 1.9119,
      "step": 2405
    },
    {
      "epoch": 0.06336581511719779,
      "grad_norm": 3.579887866973877,
      "learning_rate": 4.683697656044246e-05,
      "loss": 0.4994,
      "step": 2406
    },
    {
      "epoch": 0.0633921516987095,
      "grad_norm": 2.598342180252075,
      "learning_rate": 4.683565973136687e-05,
      "loss": 0.7125,
      "step": 2407
    },
    {
      "epoch": 0.06341848828022123,
      "grad_norm": 4.512175559997559,
      "learning_rate": 4.6834342902291284e-05,
      "loss": 1.9072,
      "step": 2408
    },
    {
      "epoch": 0.06344482486173295,
      "grad_norm": 5.32249641418457,
      "learning_rate": 4.68330260732157e-05,
      "loss": 2.4061,
      "step": 2409
    },
    {
      "epoch": 0.06347116144324466,
      "grad_norm": 3.2500739097595215,
      "learning_rate": 4.6831709244140115e-05,
      "loss": 1.7072,
      "step": 2410
    },
    {
      "epoch": 0.06349749802475639,
      "grad_norm": 1.9146647453308105,
      "learning_rate": 4.683039241506453e-05,
      "loss": 2.4875,
      "step": 2411
    },
    {
      "epoch": 0.0635238346062681,
      "grad_norm": 4.5336198806762695,
      "learning_rate": 4.682907558598894e-05,
      "loss": 0.5381,
      "step": 2412
    },
    {
      "epoch": 0.06355017118777982,
      "grad_norm": 2.675198793411255,
      "learning_rate": 4.6827758756913355e-05,
      "loss": 2.0764,
      "step": 2413
    },
    {
      "epoch": 0.06357650776929155,
      "grad_norm": 2.0779879093170166,
      "learning_rate": 4.6826441927837764e-05,
      "loss": 1.3311,
      "step": 2414
    },
    {
      "epoch": 0.06360284435080327,
      "grad_norm": 8.918055534362793,
      "learning_rate": 4.682512509876219e-05,
      "loss": 1.4941,
      "step": 2415
    },
    {
      "epoch": 0.06362918093231498,
      "grad_norm": 2.119518756866455,
      "learning_rate": 4.6823808269686596e-05,
      "loss": 1.5165,
      "step": 2416
    },
    {
      "epoch": 0.06365551751382671,
      "grad_norm": 4.731598377227783,
      "learning_rate": 4.682249144061101e-05,
      "loss": 1.2126,
      "step": 2417
    },
    {
      "epoch": 0.06368185409533843,
      "grad_norm": 2.4830589294433594,
      "learning_rate": 4.682117461153542e-05,
      "loss": 1.4372,
      "step": 2418
    },
    {
      "epoch": 0.06370819067685014,
      "grad_norm": 1.9942201375961304,
      "learning_rate": 4.681985778245984e-05,
      "loss": 1.8839,
      "step": 2419
    },
    {
      "epoch": 0.06373452725836186,
      "grad_norm": 2.2299985885620117,
      "learning_rate": 4.681854095338425e-05,
      "loss": 1.9067,
      "step": 2420
    },
    {
      "epoch": 0.06376086383987359,
      "grad_norm": 3.289201021194458,
      "learning_rate": 4.681722412430867e-05,
      "loss": 1.9577,
      "step": 2421
    },
    {
      "epoch": 0.0637872004213853,
      "grad_norm": 5.261205196380615,
      "learning_rate": 4.681590729523308e-05,
      "loss": 2.384,
      "step": 2422
    },
    {
      "epoch": 0.06381353700289702,
      "grad_norm": 2.5124638080596924,
      "learning_rate": 4.681459046615749e-05,
      "loss": 1.8541,
      "step": 2423
    },
    {
      "epoch": 0.06383987358440875,
      "grad_norm": 1.994523286819458,
      "learning_rate": 4.6813273637081914e-05,
      "loss": 1.584,
      "step": 2424
    },
    {
      "epoch": 0.06386621016592046,
      "grad_norm": 1.8511320352554321,
      "learning_rate": 4.681195680800632e-05,
      "loss": 1.5149,
      "step": 2425
    },
    {
      "epoch": 0.06389254674743218,
      "grad_norm": 2.3349974155426025,
      "learning_rate": 4.681063997893074e-05,
      "loss": 2.6602,
      "step": 2426
    },
    {
      "epoch": 0.06391888332894391,
      "grad_norm": 3.3603076934814453,
      "learning_rate": 4.680932314985515e-05,
      "loss": 1.4353,
      "step": 2427
    },
    {
      "epoch": 0.06394521991045563,
      "grad_norm": 2.5830607414245605,
      "learning_rate": 4.680800632077956e-05,
      "loss": 0.5467,
      "step": 2428
    },
    {
      "epoch": 0.06397155649196734,
      "grad_norm": 1.9991215467453003,
      "learning_rate": 4.680668949170398e-05,
      "loss": 2.1647,
      "step": 2429
    },
    {
      "epoch": 0.06399789307347906,
      "grad_norm": 4.537223815917969,
      "learning_rate": 4.6805372662628394e-05,
      "loss": 0.7771,
      "step": 2430
    },
    {
      "epoch": 0.06402422965499079,
      "grad_norm": 2.468398094177246,
      "learning_rate": 4.680405583355281e-05,
      "loss": 1.8903,
      "step": 2431
    },
    {
      "epoch": 0.0640505662365025,
      "grad_norm": 2.197237968444824,
      "learning_rate": 4.680273900447722e-05,
      "loss": 2.1243,
      "step": 2432
    },
    {
      "epoch": 0.06407690281801422,
      "grad_norm": 3.1796863079071045,
      "learning_rate": 4.6801422175401634e-05,
      "loss": 0.3634,
      "step": 2433
    },
    {
      "epoch": 0.06410323939952595,
      "grad_norm": 3.5290427207946777,
      "learning_rate": 4.680010534632605e-05,
      "loss": 1.8702,
      "step": 2434
    },
    {
      "epoch": 0.06412957598103766,
      "grad_norm": 2.695802927017212,
      "learning_rate": 4.6798788517250465e-05,
      "loss": 2.3979,
      "step": 2435
    },
    {
      "epoch": 0.06415591256254938,
      "grad_norm": 3.778914213180542,
      "learning_rate": 4.6797471688174874e-05,
      "loss": 2.0953,
      "step": 2436
    },
    {
      "epoch": 0.06418224914406111,
      "grad_norm": 2.339081048965454,
      "learning_rate": 4.679615485909929e-05,
      "loss": 1.6206,
      "step": 2437
    },
    {
      "epoch": 0.06420858572557282,
      "grad_norm": 2.2991528511047363,
      "learning_rate": 4.6794838030023705e-05,
      "loss": 2.1085,
      "step": 2438
    },
    {
      "epoch": 0.06423492230708454,
      "grad_norm": 3.41226863861084,
      "learning_rate": 4.679352120094812e-05,
      "loss": 1.7824,
      "step": 2439
    },
    {
      "epoch": 0.06426125888859625,
      "grad_norm": 3.640415668487549,
      "learning_rate": 4.6792204371872536e-05,
      "loss": 1.7638,
      "step": 2440
    },
    {
      "epoch": 0.06428759547010798,
      "grad_norm": 2.1721041202545166,
      "learning_rate": 4.6790887542796945e-05,
      "loss": 1.9222,
      "step": 2441
    },
    {
      "epoch": 0.0643139320516197,
      "grad_norm": 3.03312611579895,
      "learning_rate": 4.678957071372136e-05,
      "loss": 1.763,
      "step": 2442
    },
    {
      "epoch": 0.06434026863313141,
      "grad_norm": 2.873274803161621,
      "learning_rate": 4.6788253884645776e-05,
      "loss": 1.17,
      "step": 2443
    },
    {
      "epoch": 0.06436660521464314,
      "grad_norm": 1.8585783243179321,
      "learning_rate": 4.678693705557019e-05,
      "loss": 2.0744,
      "step": 2444
    },
    {
      "epoch": 0.06439294179615486,
      "grad_norm": 2.5648512840270996,
      "learning_rate": 4.67856202264946e-05,
      "loss": 1.8095,
      "step": 2445
    },
    {
      "epoch": 0.06441927837766658,
      "grad_norm": 3.6596877574920654,
      "learning_rate": 4.6784303397419016e-05,
      "loss": 1.8555,
      "step": 2446
    },
    {
      "epoch": 0.0644456149591783,
      "grad_norm": 4.499950885772705,
      "learning_rate": 4.678298656834343e-05,
      "loss": 2.3845,
      "step": 2447
    },
    {
      "epoch": 0.06447195154069002,
      "grad_norm": 1.8886054754257202,
      "learning_rate": 4.678166973926785e-05,
      "loss": 1.4694,
      "step": 2448
    },
    {
      "epoch": 0.06449828812220174,
      "grad_norm": 3.0235955715179443,
      "learning_rate": 4.678035291019226e-05,
      "loss": 1.8034,
      "step": 2449
    },
    {
      "epoch": 0.06452462470371345,
      "grad_norm": 4.964101314544678,
      "learning_rate": 4.677903608111667e-05,
      "loss": 1.4596,
      "step": 2450
    },
    {
      "epoch": 0.06455096128522518,
      "grad_norm": 3.664073944091797,
      "learning_rate": 4.677771925204109e-05,
      "loss": 0.5822,
      "step": 2451
    },
    {
      "epoch": 0.0645772978667369,
      "grad_norm": 4.64607572555542,
      "learning_rate": 4.67764024229655e-05,
      "loss": 1.8566,
      "step": 2452
    },
    {
      "epoch": 0.06460363444824861,
      "grad_norm": 2.4599313735961914,
      "learning_rate": 4.677508559388992e-05,
      "loss": 2.019,
      "step": 2453
    },
    {
      "epoch": 0.06462997102976034,
      "grad_norm": 2.585855722427368,
      "learning_rate": 4.677376876481433e-05,
      "loss": 1.8816,
      "step": 2454
    },
    {
      "epoch": 0.06465630761127206,
      "grad_norm": 4.79041862487793,
      "learning_rate": 4.677245193573874e-05,
      "loss": 1.0569,
      "step": 2455
    },
    {
      "epoch": 0.06468264419278377,
      "grad_norm": 7.284574508666992,
      "learning_rate": 4.677113510666316e-05,
      "loss": 1.7071,
      "step": 2456
    },
    {
      "epoch": 0.0647089807742955,
      "grad_norm": 2.1668169498443604,
      "learning_rate": 4.6769818277587574e-05,
      "loss": 1.3356,
      "step": 2457
    },
    {
      "epoch": 0.06473531735580722,
      "grad_norm": 1.8819416761398315,
      "learning_rate": 4.676850144851199e-05,
      "loss": 1.8233,
      "step": 2458
    },
    {
      "epoch": 0.06476165393731893,
      "grad_norm": 3.402418613433838,
      "learning_rate": 4.67671846194364e-05,
      "loss": 1.613,
      "step": 2459
    },
    {
      "epoch": 0.06478799051883066,
      "grad_norm": 2.380220890045166,
      "learning_rate": 4.6765867790360814e-05,
      "loss": 1.8139,
      "step": 2460
    },
    {
      "epoch": 0.06481432710034238,
      "grad_norm": 3.6850509643554688,
      "learning_rate": 4.676455096128522e-05,
      "loss": 1.77,
      "step": 2461
    },
    {
      "epoch": 0.0648406636818541,
      "grad_norm": 2.4299466609954834,
      "learning_rate": 4.6763234132209645e-05,
      "loss": 1.8125,
      "step": 2462
    },
    {
      "epoch": 0.06486700026336581,
      "grad_norm": 1.9561840295791626,
      "learning_rate": 4.6761917303134054e-05,
      "loss": 2.0749,
      "step": 2463
    },
    {
      "epoch": 0.06489333684487754,
      "grad_norm": 4.448574542999268,
      "learning_rate": 4.676060047405847e-05,
      "loss": 1.0838,
      "step": 2464
    },
    {
      "epoch": 0.06491967342638925,
      "grad_norm": 3.413048028945923,
      "learning_rate": 4.6759283644982885e-05,
      "loss": 0.5598,
      "step": 2465
    },
    {
      "epoch": 0.06494601000790097,
      "grad_norm": 2.1648099422454834,
      "learning_rate": 4.6757966815907294e-05,
      "loss": 2.1142,
      "step": 2466
    },
    {
      "epoch": 0.0649723465894127,
      "grad_norm": 4.810667991638184,
      "learning_rate": 4.6756649986831716e-05,
      "loss": 2.2293,
      "step": 2467
    },
    {
      "epoch": 0.06499868317092442,
      "grad_norm": 3.0840086936950684,
      "learning_rate": 4.6755333157756125e-05,
      "loss": 1.5708,
      "step": 2468
    },
    {
      "epoch": 0.06502501975243613,
      "grad_norm": 4.743466377258301,
      "learning_rate": 4.675401632868054e-05,
      "loss": 1.9965,
      "step": 2469
    },
    {
      "epoch": 0.06505135633394786,
      "grad_norm": 2.19016695022583,
      "learning_rate": 4.675269949960495e-05,
      "loss": 2.3954,
      "step": 2470
    },
    {
      "epoch": 0.06507769291545958,
      "grad_norm": 3.0504465103149414,
      "learning_rate": 4.675138267052937e-05,
      "loss": 1.7875,
      "step": 2471
    },
    {
      "epoch": 0.06510402949697129,
      "grad_norm": 2.3120687007904053,
      "learning_rate": 4.675006584145378e-05,
      "loss": 1.3664,
      "step": 2472
    },
    {
      "epoch": 0.065130366078483,
      "grad_norm": 3.733536958694458,
      "learning_rate": 4.6748749012378196e-05,
      "loss": 1.4547,
      "step": 2473
    },
    {
      "epoch": 0.06515670265999474,
      "grad_norm": 2.8666889667510986,
      "learning_rate": 4.6747432183302605e-05,
      "loss": 1.45,
      "step": 2474
    },
    {
      "epoch": 0.06518303924150645,
      "grad_norm": 1.8116925954818726,
      "learning_rate": 4.674611535422702e-05,
      "loss": 1.8247,
      "step": 2475
    },
    {
      "epoch": 0.06520937582301817,
      "grad_norm": 2.427353858947754,
      "learning_rate": 4.6744798525151436e-05,
      "loss": 0.5902,
      "step": 2476
    },
    {
      "epoch": 0.0652357124045299,
      "grad_norm": 3.56376314163208,
      "learning_rate": 4.674348169607585e-05,
      "loss": 1.092,
      "step": 2477
    },
    {
      "epoch": 0.06526204898604161,
      "grad_norm": 4.646085739135742,
      "learning_rate": 4.674216486700027e-05,
      "loss": 1.2398,
      "step": 2478
    },
    {
      "epoch": 0.06528838556755333,
      "grad_norm": 5.491124153137207,
      "learning_rate": 4.6740848037924677e-05,
      "loss": 1.3864,
      "step": 2479
    },
    {
      "epoch": 0.06531472214906506,
      "grad_norm": 3.0339837074279785,
      "learning_rate": 4.673953120884909e-05,
      "loss": 2.0601,
      "step": 2480
    },
    {
      "epoch": 0.06534105873057677,
      "grad_norm": 3.239531993865967,
      "learning_rate": 4.673821437977351e-05,
      "loss": 0.4445,
      "step": 2481
    },
    {
      "epoch": 0.06536739531208849,
      "grad_norm": 2.132255792617798,
      "learning_rate": 4.673689755069792e-05,
      "loss": 1.8743,
      "step": 2482
    },
    {
      "epoch": 0.0653937318936002,
      "grad_norm": 1.9474587440490723,
      "learning_rate": 4.673558072162233e-05,
      "loss": 2.0281,
      "step": 2483
    },
    {
      "epoch": 0.06542006847511193,
      "grad_norm": 2.3764939308166504,
      "learning_rate": 4.673426389254675e-05,
      "loss": 1.3453,
      "step": 2484
    },
    {
      "epoch": 0.06544640505662365,
      "grad_norm": 7.971607208251953,
      "learning_rate": 4.673294706347116e-05,
      "loss": 2.0164,
      "step": 2485
    },
    {
      "epoch": 0.06547274163813536,
      "grad_norm": 4.840243339538574,
      "learning_rate": 4.673163023439558e-05,
      "loss": 1.0984,
      "step": 2486
    },
    {
      "epoch": 0.0654990782196471,
      "grad_norm": 5.740271091461182,
      "learning_rate": 4.6730313405319995e-05,
      "loss": 0.5724,
      "step": 2487
    },
    {
      "epoch": 0.06552541480115881,
      "grad_norm": 2.0132150650024414,
      "learning_rate": 4.67289965762444e-05,
      "loss": 1.471,
      "step": 2488
    },
    {
      "epoch": 0.06555175138267053,
      "grad_norm": 3.446458578109741,
      "learning_rate": 4.672767974716882e-05,
      "loss": 1.7057,
      "step": 2489
    },
    {
      "epoch": 0.06557808796418226,
      "grad_norm": 2.107990264892578,
      "learning_rate": 4.6726362918093235e-05,
      "loss": 1.2397,
      "step": 2490
    },
    {
      "epoch": 0.06560442454569397,
      "grad_norm": 6.227726459503174,
      "learning_rate": 4.672504608901765e-05,
      "loss": 1.8721,
      "step": 2491
    },
    {
      "epoch": 0.06563076112720569,
      "grad_norm": 3.4746124744415283,
      "learning_rate": 4.672372925994206e-05,
      "loss": 2.0732,
      "step": 2492
    },
    {
      "epoch": 0.0656570977087174,
      "grad_norm": 3.332350254058838,
      "learning_rate": 4.6722412430866475e-05,
      "loss": 1.6787,
      "step": 2493
    },
    {
      "epoch": 0.06568343429022913,
      "grad_norm": 2.8294148445129395,
      "learning_rate": 4.672109560179089e-05,
      "loss": 1.6272,
      "step": 2494
    },
    {
      "epoch": 0.06570977087174085,
      "grad_norm": 3.3581392765045166,
      "learning_rate": 4.6719778772715306e-05,
      "loss": 2.2576,
      "step": 2495
    },
    {
      "epoch": 0.06573610745325256,
      "grad_norm": 4.122766494750977,
      "learning_rate": 4.671846194363972e-05,
      "loss": 1.5237,
      "step": 2496
    },
    {
      "epoch": 0.06576244403476429,
      "grad_norm": 2.0318691730499268,
      "learning_rate": 4.671714511456413e-05,
      "loss": 1.8989,
      "step": 2497
    },
    {
      "epoch": 0.06578878061627601,
      "grad_norm": 2.39886474609375,
      "learning_rate": 4.6715828285488546e-05,
      "loss": 1.6113,
      "step": 2498
    },
    {
      "epoch": 0.06581511719778772,
      "grad_norm": 2.074373960494995,
      "learning_rate": 4.6714511456412955e-05,
      "loss": 1.0614,
      "step": 2499
    },
    {
      "epoch": 0.06584145377929945,
      "grad_norm": 1.8505825996398926,
      "learning_rate": 4.671319462733738e-05,
      "loss": 0.5445,
      "step": 2500
    },
    {
      "epoch": 0.06586779036081117,
      "grad_norm": 2.2921855449676514,
      "learning_rate": 4.6711877798261786e-05,
      "loss": 1.5375,
      "step": 2501
    },
    {
      "epoch": 0.06589412694232288,
      "grad_norm": 1.9849905967712402,
      "learning_rate": 4.67105609691862e-05,
      "loss": 1.5585,
      "step": 2502
    },
    {
      "epoch": 0.06592046352383461,
      "grad_norm": 2.228806734085083,
      "learning_rate": 4.670924414011062e-05,
      "loss": 1.8496,
      "step": 2503
    },
    {
      "epoch": 0.06594680010534633,
      "grad_norm": 3.650649070739746,
      "learning_rate": 4.670792731103503e-05,
      "loss": 0.5406,
      "step": 2504
    },
    {
      "epoch": 0.06597313668685804,
      "grad_norm": 2.1462583541870117,
      "learning_rate": 4.670661048195945e-05,
      "loss": 1.9369,
      "step": 2505
    },
    {
      "epoch": 0.06599947326836976,
      "grad_norm": 2.357234239578247,
      "learning_rate": 4.670529365288386e-05,
      "loss": 2.3181,
      "step": 2506
    },
    {
      "epoch": 0.06602580984988149,
      "grad_norm": 3.277623414993286,
      "learning_rate": 4.670397682380827e-05,
      "loss": 2.0376,
      "step": 2507
    },
    {
      "epoch": 0.0660521464313932,
      "grad_norm": 2.1119625568389893,
      "learning_rate": 4.670265999473268e-05,
      "loss": 1.1413,
      "step": 2508
    },
    {
      "epoch": 0.06607848301290492,
      "grad_norm": 5.124713897705078,
      "learning_rate": 4.6701343165657104e-05,
      "loss": 1.632,
      "step": 2509
    },
    {
      "epoch": 0.06610481959441665,
      "grad_norm": 1.9271942377090454,
      "learning_rate": 4.670002633658151e-05,
      "loss": 1.2184,
      "step": 2510
    },
    {
      "epoch": 0.06613115617592837,
      "grad_norm": 2.0628840923309326,
      "learning_rate": 4.669870950750593e-05,
      "loss": 1.9178,
      "step": 2511
    },
    {
      "epoch": 0.06615749275744008,
      "grad_norm": 1.8928604125976562,
      "learning_rate": 4.6697392678430344e-05,
      "loss": 2.0566,
      "step": 2512
    },
    {
      "epoch": 0.06618382933895181,
      "grad_norm": 1.9438021183013916,
      "learning_rate": 4.669607584935475e-05,
      "loss": 2.5018,
      "step": 2513
    },
    {
      "epoch": 0.06621016592046353,
      "grad_norm": 2.5813844203948975,
      "learning_rate": 4.6694759020279175e-05,
      "loss": 2.4215,
      "step": 2514
    },
    {
      "epoch": 0.06623650250197524,
      "grad_norm": 1.6554195880889893,
      "learning_rate": 4.6693442191203584e-05,
      "loss": 2.1316,
      "step": 2515
    },
    {
      "epoch": 0.06626283908348696,
      "grad_norm": 2.4325571060180664,
      "learning_rate": 4.6692125362128e-05,
      "loss": 1.8311,
      "step": 2516
    },
    {
      "epoch": 0.06628917566499869,
      "grad_norm": 4.826909065246582,
      "learning_rate": 4.669080853305241e-05,
      "loss": 1.0181,
      "step": 2517
    },
    {
      "epoch": 0.0663155122465104,
      "grad_norm": 5.109097480773926,
      "learning_rate": 4.668949170397683e-05,
      "loss": 2.0126,
      "step": 2518
    },
    {
      "epoch": 0.06634184882802212,
      "grad_norm": 3.1182119846343994,
      "learning_rate": 4.668817487490124e-05,
      "loss": 1.213,
      "step": 2519
    },
    {
      "epoch": 0.06636818540953385,
      "grad_norm": 1.836631417274475,
      "learning_rate": 4.6686858045825655e-05,
      "loss": 2.2042,
      "step": 2520
    },
    {
      "epoch": 0.06639452199104556,
      "grad_norm": 4.009081840515137,
      "learning_rate": 4.6685541216750064e-05,
      "loss": 2.114,
      "step": 2521
    },
    {
      "epoch": 0.06642085857255728,
      "grad_norm": 6.382277488708496,
      "learning_rate": 4.668422438767448e-05,
      "loss": 2.1955,
      "step": 2522
    },
    {
      "epoch": 0.06644719515406901,
      "grad_norm": 3.5269582271575928,
      "learning_rate": 4.6682907558598895e-05,
      "loss": 0.5381,
      "step": 2523
    },
    {
      "epoch": 0.06647353173558072,
      "grad_norm": 2.0849244594573975,
      "learning_rate": 4.668159072952331e-05,
      "loss": 0.4977,
      "step": 2524
    },
    {
      "epoch": 0.06649986831709244,
      "grad_norm": 1.678020715713501,
      "learning_rate": 4.6680273900447726e-05,
      "loss": 1.4492,
      "step": 2525
    },
    {
      "epoch": 0.06652620489860415,
      "grad_norm": 2.257838249206543,
      "learning_rate": 4.6678957071372135e-05,
      "loss": 1.4787,
      "step": 2526
    },
    {
      "epoch": 0.06655254148011588,
      "grad_norm": 2.6151444911956787,
      "learning_rate": 4.667764024229655e-05,
      "loss": 0.8137,
      "step": 2527
    },
    {
      "epoch": 0.0665788780616276,
      "grad_norm": 2.042978048324585,
      "learning_rate": 4.6676323413220966e-05,
      "loss": 1.5399,
      "step": 2528
    },
    {
      "epoch": 0.06660521464313932,
      "grad_norm": 4.034358978271484,
      "learning_rate": 4.667500658414538e-05,
      "loss": 0.6036,
      "step": 2529
    },
    {
      "epoch": 0.06663155122465104,
      "grad_norm": 4.255539894104004,
      "learning_rate": 4.667368975506979e-05,
      "loss": 1.634,
      "step": 2530
    },
    {
      "epoch": 0.06665788780616276,
      "grad_norm": 2.1822259426116943,
      "learning_rate": 4.6672372925994206e-05,
      "loss": 2.0315,
      "step": 2531
    },
    {
      "epoch": 0.06668422438767448,
      "grad_norm": 4.216916084289551,
      "learning_rate": 4.667105609691862e-05,
      "loss": 1.98,
      "step": 2532
    },
    {
      "epoch": 0.0667105609691862,
      "grad_norm": 2.0080924034118652,
      "learning_rate": 4.666973926784304e-05,
      "loss": 1.5931,
      "step": 2533
    },
    {
      "epoch": 0.06673689755069792,
      "grad_norm": 1.8816441297531128,
      "learning_rate": 4.666842243876745e-05,
      "loss": 1.5876,
      "step": 2534
    },
    {
      "epoch": 0.06676323413220964,
      "grad_norm": 2.8267483711242676,
      "learning_rate": 4.666710560969186e-05,
      "loss": 1.3382,
      "step": 2535
    },
    {
      "epoch": 0.06678957071372135,
      "grad_norm": 3.6327762603759766,
      "learning_rate": 4.666578878061628e-05,
      "loss": 1.9714,
      "step": 2536
    },
    {
      "epoch": 0.06681590729523308,
      "grad_norm": 1.8365795612335205,
      "learning_rate": 4.666447195154069e-05,
      "loss": 1.6223,
      "step": 2537
    },
    {
      "epoch": 0.0668422438767448,
      "grad_norm": 3.036679983139038,
      "learning_rate": 4.666315512246511e-05,
      "loss": 1.7841,
      "step": 2538
    },
    {
      "epoch": 0.06686858045825651,
      "grad_norm": 2.0765795707702637,
      "learning_rate": 4.666183829338952e-05,
      "loss": 2.0547,
      "step": 2539
    },
    {
      "epoch": 0.06689491703976824,
      "grad_norm": 2.1875293254852295,
      "learning_rate": 4.666052146431393e-05,
      "loss": 2.1577,
      "step": 2540
    },
    {
      "epoch": 0.06692125362127996,
      "grad_norm": 2.086848258972168,
      "learning_rate": 4.665920463523835e-05,
      "loss": 2.0517,
      "step": 2541
    },
    {
      "epoch": 0.06694759020279167,
      "grad_norm": 2.2084317207336426,
      "learning_rate": 4.6657887806162764e-05,
      "loss": 2.7073,
      "step": 2542
    },
    {
      "epoch": 0.0669739267843034,
      "grad_norm": 2.1722753047943115,
      "learning_rate": 4.665657097708718e-05,
      "loss": 1.4848,
      "step": 2543
    },
    {
      "epoch": 0.06700026336581512,
      "grad_norm": 2.3003392219543457,
      "learning_rate": 4.665525414801159e-05,
      "loss": 2.2607,
      "step": 2544
    },
    {
      "epoch": 0.06702659994732683,
      "grad_norm": 3.7220823764801025,
      "learning_rate": 4.6653937318936004e-05,
      "loss": 1.4799,
      "step": 2545
    },
    {
      "epoch": 0.06705293652883855,
      "grad_norm": 2.39377498626709,
      "learning_rate": 4.665262048986041e-05,
      "loss": 1.3127,
      "step": 2546
    },
    {
      "epoch": 0.06707927311035028,
      "grad_norm": 2.2810888290405273,
      "learning_rate": 4.6651303660784836e-05,
      "loss": 1.6819,
      "step": 2547
    },
    {
      "epoch": 0.067105609691862,
      "grad_norm": 9.56706714630127,
      "learning_rate": 4.6649986831709244e-05,
      "loss": 1.1164,
      "step": 2548
    },
    {
      "epoch": 0.06713194627337371,
      "grad_norm": 2.0614781379699707,
      "learning_rate": 4.664867000263366e-05,
      "loss": 0.5602,
      "step": 2549
    },
    {
      "epoch": 0.06715828285488544,
      "grad_norm": 3.2973594665527344,
      "learning_rate": 4.6647353173558076e-05,
      "loss": 1.2567,
      "step": 2550
    },
    {
      "epoch": 0.06718461943639716,
      "grad_norm": 1.8161203861236572,
      "learning_rate": 4.664603634448249e-05,
      "loss": 1.7685,
      "step": 2551
    },
    {
      "epoch": 0.06721095601790887,
      "grad_norm": 1.6718690395355225,
      "learning_rate": 4.664471951540691e-05,
      "loss": 1.1297,
      "step": 2552
    },
    {
      "epoch": 0.0672372925994206,
      "grad_norm": 1.7231602668762207,
      "learning_rate": 4.6643402686331316e-05,
      "loss": 0.9104,
      "step": 2553
    },
    {
      "epoch": 0.06726362918093232,
      "grad_norm": 3.418454170227051,
      "learning_rate": 4.664208585725573e-05,
      "loss": 0.9618,
      "step": 2554
    },
    {
      "epoch": 0.06728996576244403,
      "grad_norm": 1.7077583074569702,
      "learning_rate": 4.664076902818014e-05,
      "loss": 1.8292,
      "step": 2555
    },
    {
      "epoch": 0.06731630234395576,
      "grad_norm": 1.8046586513519287,
      "learning_rate": 4.663945219910456e-05,
      "loss": 1.6216,
      "step": 2556
    },
    {
      "epoch": 0.06734263892546748,
      "grad_norm": 3.2594175338745117,
      "learning_rate": 4.663813537002897e-05,
      "loss": 1.4472,
      "step": 2557
    },
    {
      "epoch": 0.06736897550697919,
      "grad_norm": 2.9247069358825684,
      "learning_rate": 4.663681854095339e-05,
      "loss": 2.147,
      "step": 2558
    },
    {
      "epoch": 0.06739531208849091,
      "grad_norm": 2.0607879161834717,
      "learning_rate": 4.66355017118778e-05,
      "loss": 1.6401,
      "step": 2559
    },
    {
      "epoch": 0.06742164867000264,
      "grad_norm": 1.7980475425720215,
      "learning_rate": 4.663418488280221e-05,
      "loss": 1.8501,
      "step": 2560
    },
    {
      "epoch": 0.06744798525151435,
      "grad_norm": 1.9060691595077515,
      "learning_rate": 4.6632868053726634e-05,
      "loss": 1.8517,
      "step": 2561
    },
    {
      "epoch": 0.06747432183302607,
      "grad_norm": 5.874664306640625,
      "learning_rate": 4.663155122465104e-05,
      "loss": 1.4543,
      "step": 2562
    },
    {
      "epoch": 0.0675006584145378,
      "grad_norm": 5.053830623626709,
      "learning_rate": 4.663023439557546e-05,
      "loss": 1.3172,
      "step": 2563
    },
    {
      "epoch": 0.06752699499604951,
      "grad_norm": 1.8805735111236572,
      "learning_rate": 4.662891756649987e-05,
      "loss": 1.6243,
      "step": 2564
    },
    {
      "epoch": 0.06755333157756123,
      "grad_norm": 2.795335292816162,
      "learning_rate": 4.662760073742428e-05,
      "loss": 1.7764,
      "step": 2565
    },
    {
      "epoch": 0.06757966815907296,
      "grad_norm": 2.3032498359680176,
      "learning_rate": 4.66262839083487e-05,
      "loss": 2.2232,
      "step": 2566
    },
    {
      "epoch": 0.06760600474058467,
      "grad_norm": 3.9962282180786133,
      "learning_rate": 4.6624967079273114e-05,
      "loss": 1.4959,
      "step": 2567
    },
    {
      "epoch": 0.06763234132209639,
      "grad_norm": 2.8505661487579346,
      "learning_rate": 4.662365025019753e-05,
      "loss": 1.1346,
      "step": 2568
    },
    {
      "epoch": 0.0676586779036081,
      "grad_norm": 2.709254503250122,
      "learning_rate": 4.662233342112194e-05,
      "loss": 2.3283,
      "step": 2569
    },
    {
      "epoch": 0.06768501448511983,
      "grad_norm": 1.8439650535583496,
      "learning_rate": 4.662101659204636e-05,
      "loss": 1.8535,
      "step": 2570
    },
    {
      "epoch": 0.06771135106663155,
      "grad_norm": 2.210843086242676,
      "learning_rate": 4.661969976297077e-05,
      "loss": 1.4266,
      "step": 2571
    },
    {
      "epoch": 0.06773768764814327,
      "grad_norm": 2.4437777996063232,
      "learning_rate": 4.6618382933895185e-05,
      "loss": 1.8741,
      "step": 2572
    },
    {
      "epoch": 0.067764024229655,
      "grad_norm": 2.7891852855682373,
      "learning_rate": 4.6617066104819594e-05,
      "loss": 1.0567,
      "step": 2573
    },
    {
      "epoch": 0.06779036081116671,
      "grad_norm": 3.1497902870178223,
      "learning_rate": 4.661574927574401e-05,
      "loss": 1.6176,
      "step": 2574
    },
    {
      "epoch": 0.06781669739267843,
      "grad_norm": 1.909374475479126,
      "learning_rate": 4.6614432446668425e-05,
      "loss": 1.6587,
      "step": 2575
    },
    {
      "epoch": 0.06784303397419016,
      "grad_norm": 2.338366985321045,
      "learning_rate": 4.661311561759284e-05,
      "loss": 1.6408,
      "step": 2576
    },
    {
      "epoch": 0.06786937055570187,
      "grad_norm": 3.6487927436828613,
      "learning_rate": 4.661179878851725e-05,
      "loss": 1.5463,
      "step": 2577
    },
    {
      "epoch": 0.06789570713721359,
      "grad_norm": 2.0124223232269287,
      "learning_rate": 4.6610481959441665e-05,
      "loss": 1.5813,
      "step": 2578
    },
    {
      "epoch": 0.0679220437187253,
      "grad_norm": 1.8438575267791748,
      "learning_rate": 4.660916513036608e-05,
      "loss": 0.7456,
      "step": 2579
    },
    {
      "epoch": 0.06794838030023703,
      "grad_norm": 3.658520221710205,
      "learning_rate": 4.6607848301290496e-05,
      "loss": 2.2583,
      "step": 2580
    },
    {
      "epoch": 0.06797471688174875,
      "grad_norm": 2.3209245204925537,
      "learning_rate": 4.660653147221491e-05,
      "loss": 2.1204,
      "step": 2581
    },
    {
      "epoch": 0.06800105346326046,
      "grad_norm": 2.0832419395446777,
      "learning_rate": 4.660521464313932e-05,
      "loss": 1.6178,
      "step": 2582
    },
    {
      "epoch": 0.06802739004477219,
      "grad_norm": 1.811886191368103,
      "learning_rate": 4.6603897814063736e-05,
      "loss": 1.7885,
      "step": 2583
    },
    {
      "epoch": 0.06805372662628391,
      "grad_norm": 1.7254831790924072,
      "learning_rate": 4.660258098498815e-05,
      "loss": 1.3446,
      "step": 2584
    },
    {
      "epoch": 0.06808006320779562,
      "grad_norm": 4.650494575500488,
      "learning_rate": 4.660126415591257e-05,
      "loss": 0.8956,
      "step": 2585
    },
    {
      "epoch": 0.06810639978930735,
      "grad_norm": 2.0037341117858887,
      "learning_rate": 4.6599947326836976e-05,
      "loss": 1.9698,
      "step": 2586
    },
    {
      "epoch": 0.06813273637081907,
      "grad_norm": 3.3112096786499023,
      "learning_rate": 4.659863049776139e-05,
      "loss": 1.4204,
      "step": 2587
    },
    {
      "epoch": 0.06815907295233078,
      "grad_norm": 2.6080074310302734,
      "learning_rate": 4.659731366868581e-05,
      "loss": 0.8499,
      "step": 2588
    },
    {
      "epoch": 0.0681854095338425,
      "grad_norm": 2.7312350273132324,
      "learning_rate": 4.659599683961022e-05,
      "loss": 1.3241,
      "step": 2589
    },
    {
      "epoch": 0.06821174611535423,
      "grad_norm": 1.6900057792663574,
      "learning_rate": 4.659468001053464e-05,
      "loss": 1.97,
      "step": 2590
    },
    {
      "epoch": 0.06823808269686595,
      "grad_norm": 3.0792837142944336,
      "learning_rate": 4.659336318145905e-05,
      "loss": 2.3653,
      "step": 2591
    },
    {
      "epoch": 0.06826441927837766,
      "grad_norm": 4.47657585144043,
      "learning_rate": 4.659204635238346e-05,
      "loss": 1.6617,
      "step": 2592
    },
    {
      "epoch": 0.06829075585988939,
      "grad_norm": 2.885293960571289,
      "learning_rate": 4.659072952330787e-05,
      "loss": 1.5072,
      "step": 2593
    },
    {
      "epoch": 0.0683170924414011,
      "grad_norm": 5.959609031677246,
      "learning_rate": 4.6589412694232294e-05,
      "loss": 1.2388,
      "step": 2594
    },
    {
      "epoch": 0.06834342902291282,
      "grad_norm": 4.799948215484619,
      "learning_rate": 4.65880958651567e-05,
      "loss": 1.683,
      "step": 2595
    },
    {
      "epoch": 0.06836976560442455,
      "grad_norm": 2.2544190883636475,
      "learning_rate": 4.658677903608112e-05,
      "loss": 1.7779,
      "step": 2596
    },
    {
      "epoch": 0.06839610218593627,
      "grad_norm": 1.947739839553833,
      "learning_rate": 4.6585462207005534e-05,
      "loss": 1.8944,
      "step": 2597
    },
    {
      "epoch": 0.06842243876744798,
      "grad_norm": 2.574916362762451,
      "learning_rate": 4.658414537792994e-05,
      "loss": 1.8165,
      "step": 2598
    },
    {
      "epoch": 0.06844877534895971,
      "grad_norm": 3.0508999824523926,
      "learning_rate": 4.6582828548854365e-05,
      "loss": 1.5944,
      "step": 2599
    },
    {
      "epoch": 0.06847511193047143,
      "grad_norm": 5.070839881896973,
      "learning_rate": 4.6581511719778774e-05,
      "loss": 1.2167,
      "step": 2600
    },
    {
      "epoch": 0.06850144851198314,
      "grad_norm": 2.3829562664031982,
      "learning_rate": 4.658019489070319e-05,
      "loss": 0.651,
      "step": 2601
    },
    {
      "epoch": 0.06852778509349486,
      "grad_norm": 4.177533149719238,
      "learning_rate": 4.65788780616276e-05,
      "loss": 2.3382,
      "step": 2602
    },
    {
      "epoch": 0.06855412167500659,
      "grad_norm": 2.7740416526794434,
      "learning_rate": 4.657756123255202e-05,
      "loss": 2.0598,
      "step": 2603
    },
    {
      "epoch": 0.0685804582565183,
      "grad_norm": 3.516300916671753,
      "learning_rate": 4.657624440347643e-05,
      "loss": 0.2323,
      "step": 2604
    },
    {
      "epoch": 0.06860679483803002,
      "grad_norm": 3.4978506565093994,
      "learning_rate": 4.6574927574400845e-05,
      "loss": 2.2463,
      "step": 2605
    },
    {
      "epoch": 0.06863313141954175,
      "grad_norm": 2.4125983715057373,
      "learning_rate": 4.657361074532526e-05,
      "loss": 1.8078,
      "step": 2606
    },
    {
      "epoch": 0.06865946800105346,
      "grad_norm": 2.26993727684021,
      "learning_rate": 4.657229391624967e-05,
      "loss": 1.6311,
      "step": 2607
    },
    {
      "epoch": 0.06868580458256518,
      "grad_norm": 3.9862186908721924,
      "learning_rate": 4.657097708717409e-05,
      "loss": 1.7732,
      "step": 2608
    },
    {
      "epoch": 0.06871214116407691,
      "grad_norm": 2.0518388748168945,
      "learning_rate": 4.65696602580985e-05,
      "loss": 1.4654,
      "step": 2609
    },
    {
      "epoch": 0.06873847774558862,
      "grad_norm": 3.12658953666687,
      "learning_rate": 4.6568343429022917e-05,
      "loss": 0.5777,
      "step": 2610
    },
    {
      "epoch": 0.06876481432710034,
      "grad_norm": 8.929038047790527,
      "learning_rate": 4.6567026599947325e-05,
      "loss": 1.638,
      "step": 2611
    },
    {
      "epoch": 0.06879115090861206,
      "grad_norm": 3.825570583343506,
      "learning_rate": 4.656570977087174e-05,
      "loss": 1.5552,
      "step": 2612
    },
    {
      "epoch": 0.06881748749012379,
      "grad_norm": 2.1644906997680664,
      "learning_rate": 4.6564392941796157e-05,
      "loss": 2.1256,
      "step": 2613
    },
    {
      "epoch": 0.0688438240716355,
      "grad_norm": 2.153041124343872,
      "learning_rate": 4.656307611272057e-05,
      "loss": 1.6542,
      "step": 2614
    },
    {
      "epoch": 0.06887016065314722,
      "grad_norm": 2.481976270675659,
      "learning_rate": 4.656175928364499e-05,
      "loss": 1.1864,
      "step": 2615
    },
    {
      "epoch": 0.06889649723465895,
      "grad_norm": 2.1133615970611572,
      "learning_rate": 4.6560442454569397e-05,
      "loss": 1.4337,
      "step": 2616
    },
    {
      "epoch": 0.06892283381617066,
      "grad_norm": 2.5127944946289062,
      "learning_rate": 4.655912562549382e-05,
      "loss": 1.8533,
      "step": 2617
    },
    {
      "epoch": 0.06894917039768238,
      "grad_norm": 2.4987242221832275,
      "learning_rate": 4.655780879641823e-05,
      "loss": 1.8171,
      "step": 2618
    },
    {
      "epoch": 0.0689755069791941,
      "grad_norm": 2.4039416313171387,
      "learning_rate": 4.655649196734264e-05,
      "loss": 2.0467,
      "step": 2619
    },
    {
      "epoch": 0.06900184356070582,
      "grad_norm": 3.5297069549560547,
      "learning_rate": 4.655517513826705e-05,
      "loss": 1.1857,
      "step": 2620
    },
    {
      "epoch": 0.06902818014221754,
      "grad_norm": 2.375171661376953,
      "learning_rate": 4.655385830919147e-05,
      "loss": 2.0219,
      "step": 2621
    },
    {
      "epoch": 0.06905451672372925,
      "grad_norm": 1.766218662261963,
      "learning_rate": 4.655254148011588e-05,
      "loss": 1.7518,
      "step": 2622
    },
    {
      "epoch": 0.06908085330524098,
      "grad_norm": 3.488804578781128,
      "learning_rate": 4.65512246510403e-05,
      "loss": 1.6097,
      "step": 2623
    },
    {
      "epoch": 0.0691071898867527,
      "grad_norm": 2.894591808319092,
      "learning_rate": 4.654990782196471e-05,
      "loss": 2.0309,
      "step": 2624
    },
    {
      "epoch": 0.06913352646826441,
      "grad_norm": 7.223591327667236,
      "learning_rate": 4.6548590992889123e-05,
      "loss": 0.6836,
      "step": 2625
    },
    {
      "epoch": 0.06915986304977614,
      "grad_norm": 2.455810070037842,
      "learning_rate": 4.654727416381354e-05,
      "loss": 2.0535,
      "step": 2626
    },
    {
      "epoch": 0.06918619963128786,
      "grad_norm": 1.8866004943847656,
      "learning_rate": 4.6545957334737955e-05,
      "loss": 1.6238,
      "step": 2627
    },
    {
      "epoch": 0.06921253621279957,
      "grad_norm": 2.6919102668762207,
      "learning_rate": 4.654464050566237e-05,
      "loss": 1.6533,
      "step": 2628
    },
    {
      "epoch": 0.0692388727943113,
      "grad_norm": 2.480489730834961,
      "learning_rate": 4.654332367658678e-05,
      "loss": 2.1289,
      "step": 2629
    },
    {
      "epoch": 0.06926520937582302,
      "grad_norm": 4.822482585906982,
      "learning_rate": 4.6542006847511195e-05,
      "loss": 1.734,
      "step": 2630
    },
    {
      "epoch": 0.06929154595733474,
      "grad_norm": 4.469937324523926,
      "learning_rate": 4.654069001843561e-05,
      "loss": 1.2581,
      "step": 2631
    },
    {
      "epoch": 0.06931788253884645,
      "grad_norm": 2.7554965019226074,
      "learning_rate": 4.6539373189360026e-05,
      "loss": 2.1281,
      "step": 2632
    },
    {
      "epoch": 0.06934421912035818,
      "grad_norm": 2.2402126789093018,
      "learning_rate": 4.6538056360284435e-05,
      "loss": 1.9077,
      "step": 2633
    },
    {
      "epoch": 0.0693705557018699,
      "grad_norm": 2.3627853393554688,
      "learning_rate": 4.653673953120885e-05,
      "loss": 2.1034,
      "step": 2634
    },
    {
      "epoch": 0.06939689228338161,
      "grad_norm": 2.813047409057617,
      "learning_rate": 4.6535422702133266e-05,
      "loss": 0.725,
      "step": 2635
    },
    {
      "epoch": 0.06942322886489334,
      "grad_norm": 2.6547837257385254,
      "learning_rate": 4.653410587305768e-05,
      "loss": 1.9301,
      "step": 2636
    },
    {
      "epoch": 0.06944956544640506,
      "grad_norm": 3.0400991439819336,
      "learning_rate": 4.65327890439821e-05,
      "loss": 2.0376,
      "step": 2637
    },
    {
      "epoch": 0.06947590202791677,
      "grad_norm": 2.4060683250427246,
      "learning_rate": 4.6531472214906506e-05,
      "loss": 1.9381,
      "step": 2638
    },
    {
      "epoch": 0.0695022386094285,
      "grad_norm": 4.608154773712158,
      "learning_rate": 4.653015538583092e-05,
      "loss": 1.8165,
      "step": 2639
    },
    {
      "epoch": 0.06952857519094022,
      "grad_norm": 2.51985764503479,
      "learning_rate": 4.652883855675533e-05,
      "loss": 0.2848,
      "step": 2640
    },
    {
      "epoch": 0.06955491177245193,
      "grad_norm": 2.6939659118652344,
      "learning_rate": 4.652752172767975e-05,
      "loss": 0.4254,
      "step": 2641
    },
    {
      "epoch": 0.06958124835396366,
      "grad_norm": 3.266824960708618,
      "learning_rate": 4.652620489860416e-05,
      "loss": 2.2181,
      "step": 2642
    },
    {
      "epoch": 0.06960758493547538,
      "grad_norm": 4.517200469970703,
      "learning_rate": 4.652488806952858e-05,
      "loss": 0.746,
      "step": 2643
    },
    {
      "epoch": 0.0696339215169871,
      "grad_norm": 1.7529770135879517,
      "learning_rate": 4.652357124045299e-05,
      "loss": 1.9167,
      "step": 2644
    },
    {
      "epoch": 0.06966025809849881,
      "grad_norm": 2.715451717376709,
      "learning_rate": 4.65222544113774e-05,
      "loss": 1.5291,
      "step": 2645
    },
    {
      "epoch": 0.06968659468001054,
      "grad_norm": 2.0796825885772705,
      "learning_rate": 4.6520937582301824e-05,
      "loss": 1.8641,
      "step": 2646
    },
    {
      "epoch": 0.06971293126152225,
      "grad_norm": 4.067246913909912,
      "learning_rate": 4.651962075322623e-05,
      "loss": 1.1368,
      "step": 2647
    },
    {
      "epoch": 0.06973926784303397,
      "grad_norm": 1.6699625253677368,
      "learning_rate": 4.651830392415065e-05,
      "loss": 1.742,
      "step": 2648
    },
    {
      "epoch": 0.0697656044245457,
      "grad_norm": 2.325200080871582,
      "learning_rate": 4.651698709507506e-05,
      "loss": 2.3476,
      "step": 2649
    },
    {
      "epoch": 0.06979194100605741,
      "grad_norm": 2.1421377658843994,
      "learning_rate": 4.651567026599948e-05,
      "loss": 1.7371,
      "step": 2650
    },
    {
      "epoch": 0.06981827758756913,
      "grad_norm": 2.5462629795074463,
      "learning_rate": 4.651435343692389e-05,
      "loss": 1.7869,
      "step": 2651
    },
    {
      "epoch": 0.06984461416908086,
      "grad_norm": 1.9661940336227417,
      "learning_rate": 4.6513036607848304e-05,
      "loss": 1.966,
      "step": 2652
    },
    {
      "epoch": 0.06987095075059258,
      "grad_norm": 4.2822370529174805,
      "learning_rate": 4.651171977877272e-05,
      "loss": 1.8279,
      "step": 2653
    },
    {
      "epoch": 0.06989728733210429,
      "grad_norm": 1.9207484722137451,
      "learning_rate": 4.651040294969713e-05,
      "loss": 1.3293,
      "step": 2654
    },
    {
      "epoch": 0.069923623913616,
      "grad_norm": 2.5592851638793945,
      "learning_rate": 4.650908612062155e-05,
      "loss": 1.9382,
      "step": 2655
    },
    {
      "epoch": 0.06994996049512774,
      "grad_norm": 2.754074811935425,
      "learning_rate": 4.650776929154596e-05,
      "loss": 2.0085,
      "step": 2656
    },
    {
      "epoch": 0.06997629707663945,
      "grad_norm": 3.550414800643921,
      "learning_rate": 4.6506452462470375e-05,
      "loss": 1.9523,
      "step": 2657
    },
    {
      "epoch": 0.07000263365815117,
      "grad_norm": 2.1844401359558105,
      "learning_rate": 4.6505135633394784e-05,
      "loss": 1.7098,
      "step": 2658
    },
    {
      "epoch": 0.0700289702396629,
      "grad_norm": 3.489062547683716,
      "learning_rate": 4.65038188043192e-05,
      "loss": 1.4202,
      "step": 2659
    },
    {
      "epoch": 0.07005530682117461,
      "grad_norm": 2.4555776119232178,
      "learning_rate": 4.6502501975243615e-05,
      "loss": 2.1021,
      "step": 2660
    },
    {
      "epoch": 0.07008164340268633,
      "grad_norm": 2.9414126873016357,
      "learning_rate": 4.650118514616803e-05,
      "loss": 0.2593,
      "step": 2661
    },
    {
      "epoch": 0.07010797998419806,
      "grad_norm": 2.705049991607666,
      "learning_rate": 4.6499868317092446e-05,
      "loss": 1.1382,
      "step": 2662
    },
    {
      "epoch": 0.07013431656570977,
      "grad_norm": 3.071199655532837,
      "learning_rate": 4.6498551488016855e-05,
      "loss": 0.5684,
      "step": 2663
    },
    {
      "epoch": 0.07016065314722149,
      "grad_norm": 2.3718481063842773,
      "learning_rate": 4.649723465894128e-05,
      "loss": 1.8246,
      "step": 2664
    },
    {
      "epoch": 0.0701869897287332,
      "grad_norm": 3.0356109142303467,
      "learning_rate": 4.6495917829865686e-05,
      "loss": 1.1527,
      "step": 2665
    },
    {
      "epoch": 0.07021332631024493,
      "grad_norm": 1.8045742511749268,
      "learning_rate": 4.64946010007901e-05,
      "loss": 1.7838,
      "step": 2666
    },
    {
      "epoch": 0.07023966289175665,
      "grad_norm": 2.5278451442718506,
      "learning_rate": 4.649328417171451e-05,
      "loss": 1.235,
      "step": 2667
    },
    {
      "epoch": 0.07026599947326836,
      "grad_norm": 3.5595812797546387,
      "learning_rate": 4.6491967342638926e-05,
      "loss": 2.2291,
      "step": 2668
    },
    {
      "epoch": 0.0702923360547801,
      "grad_norm": 2.0275182723999023,
      "learning_rate": 4.649065051356334e-05,
      "loss": 1.5879,
      "step": 2669
    },
    {
      "epoch": 0.07031867263629181,
      "grad_norm": 3.090912342071533,
      "learning_rate": 4.648933368448776e-05,
      "loss": 1.6005,
      "step": 2670
    },
    {
      "epoch": 0.07034500921780353,
      "grad_norm": 1.8806706666946411,
      "learning_rate": 4.648801685541217e-05,
      "loss": 1.8741,
      "step": 2671
    },
    {
      "epoch": 0.07037134579931525,
      "grad_norm": 4.31226110458374,
      "learning_rate": 4.648670002633658e-05,
      "loss": 2.1137,
      "step": 2672
    },
    {
      "epoch": 0.07039768238082697,
      "grad_norm": 1.993150234222412,
      "learning_rate": 4.6485383197261e-05,
      "loss": 2.1648,
      "step": 2673
    },
    {
      "epoch": 0.07042401896233869,
      "grad_norm": 2.3159568309783936,
      "learning_rate": 4.648406636818541e-05,
      "loss": 1.5296,
      "step": 2674
    },
    {
      "epoch": 0.0704503555438504,
      "grad_norm": 4.529603958129883,
      "learning_rate": 4.648274953910983e-05,
      "loss": 1.8003,
      "step": 2675
    },
    {
      "epoch": 0.07047669212536213,
      "grad_norm": 2.405383586883545,
      "learning_rate": 4.648143271003424e-05,
      "loss": 1.782,
      "step": 2676
    },
    {
      "epoch": 0.07050302870687385,
      "grad_norm": 2.07312273979187,
      "learning_rate": 4.648011588095865e-05,
      "loss": 1.6537,
      "step": 2677
    },
    {
      "epoch": 0.07052936528838556,
      "grad_norm": 3.462106227874756,
      "learning_rate": 4.647879905188306e-05,
      "loss": 1.8495,
      "step": 2678
    },
    {
      "epoch": 0.07055570186989729,
      "grad_norm": 3.344675302505493,
      "learning_rate": 4.6477482222807484e-05,
      "loss": 1.4403,
      "step": 2679
    },
    {
      "epoch": 0.07058203845140901,
      "grad_norm": 2.6176297664642334,
      "learning_rate": 4.647616539373189e-05,
      "loss": 2.0217,
      "step": 2680
    },
    {
      "epoch": 0.07060837503292072,
      "grad_norm": 2.351436138153076,
      "learning_rate": 4.647484856465631e-05,
      "loss": 1.8672,
      "step": 2681
    },
    {
      "epoch": 0.07063471161443245,
      "grad_norm": 4.086192607879639,
      "learning_rate": 4.6473531735580724e-05,
      "loss": 1.7592,
      "step": 2682
    },
    {
      "epoch": 0.07066104819594417,
      "grad_norm": 3.0787124633789062,
      "learning_rate": 4.647221490650514e-05,
      "loss": 1.8843,
      "step": 2683
    },
    {
      "epoch": 0.07068738477745588,
      "grad_norm": 2.002134084701538,
      "learning_rate": 4.6470898077429556e-05,
      "loss": 1.7586,
      "step": 2684
    },
    {
      "epoch": 0.07071372135896761,
      "grad_norm": 4.480076313018799,
      "learning_rate": 4.6469581248353964e-05,
      "loss": 1.7915,
      "step": 2685
    },
    {
      "epoch": 0.07074005794047933,
      "grad_norm": 7.935029983520508,
      "learning_rate": 4.646826441927838e-05,
      "loss": 2.3479,
      "step": 2686
    },
    {
      "epoch": 0.07076639452199104,
      "grad_norm": 2.092501640319824,
      "learning_rate": 4.646694759020279e-05,
      "loss": 1.5596,
      "step": 2687
    },
    {
      "epoch": 0.07079273110350276,
      "grad_norm": 2.9171571731567383,
      "learning_rate": 4.646563076112721e-05,
      "loss": 2.1102,
      "step": 2688
    },
    {
      "epoch": 0.07081906768501449,
      "grad_norm": 2.918426036834717,
      "learning_rate": 4.646431393205162e-05,
      "loss": 1.9477,
      "step": 2689
    },
    {
      "epoch": 0.0708454042665262,
      "grad_norm": 2.7173399925231934,
      "learning_rate": 4.6462997102976036e-05,
      "loss": 0.8485,
      "step": 2690
    },
    {
      "epoch": 0.07087174084803792,
      "grad_norm": 2.0940628051757812,
      "learning_rate": 4.646168027390045e-05,
      "loss": 1.8431,
      "step": 2691
    },
    {
      "epoch": 0.07089807742954965,
      "grad_norm": 3.8046703338623047,
      "learning_rate": 4.646036344482486e-05,
      "loss": 1.626,
      "step": 2692
    },
    {
      "epoch": 0.07092441401106137,
      "grad_norm": 3.061880350112915,
      "learning_rate": 4.645904661574928e-05,
      "loss": 1.9849,
      "step": 2693
    },
    {
      "epoch": 0.07095075059257308,
      "grad_norm": 3.367072343826294,
      "learning_rate": 4.645772978667369e-05,
      "loss": 2.0503,
      "step": 2694
    },
    {
      "epoch": 0.07097708717408481,
      "grad_norm": 4.6529645919799805,
      "learning_rate": 4.645641295759811e-05,
      "loss": 1.0493,
      "step": 2695
    },
    {
      "epoch": 0.07100342375559653,
      "grad_norm": 5.167924880981445,
      "learning_rate": 4.6455096128522516e-05,
      "loss": 1.4928,
      "step": 2696
    },
    {
      "epoch": 0.07102976033710824,
      "grad_norm": 3.35506534576416,
      "learning_rate": 4.645377929944694e-05,
      "loss": 1.5682,
      "step": 2697
    },
    {
      "epoch": 0.07105609691861996,
      "grad_norm": 2.238064765930176,
      "learning_rate": 4.645246247037135e-05,
      "loss": 1.8461,
      "step": 2698
    },
    {
      "epoch": 0.07108243350013169,
      "grad_norm": 1.931767463684082,
      "learning_rate": 4.645114564129576e-05,
      "loss": 1.3575,
      "step": 2699
    },
    {
      "epoch": 0.0711087700816434,
      "grad_norm": 3.225132465362549,
      "learning_rate": 4.644982881222018e-05,
      "loss": 2.0281,
      "step": 2700
    },
    {
      "epoch": 0.07113510666315512,
      "grad_norm": 2.8764572143554688,
      "learning_rate": 4.644851198314459e-05,
      "loss": 2.5044,
      "step": 2701
    },
    {
      "epoch": 0.07116144324466685,
      "grad_norm": 1.9680535793304443,
      "learning_rate": 4.644719515406901e-05,
      "loss": 1.55,
      "step": 2702
    },
    {
      "epoch": 0.07118777982617856,
      "grad_norm": 2.0848116874694824,
      "learning_rate": 4.644587832499342e-05,
      "loss": 1.5599,
      "step": 2703
    },
    {
      "epoch": 0.07121411640769028,
      "grad_norm": 4.546876907348633,
      "learning_rate": 4.6444561495917834e-05,
      "loss": 1.7025,
      "step": 2704
    },
    {
      "epoch": 0.07124045298920201,
      "grad_norm": 4.038058280944824,
      "learning_rate": 4.644324466684224e-05,
      "loss": 1.5236,
      "step": 2705
    },
    {
      "epoch": 0.07126678957071372,
      "grad_norm": 4.274616718292236,
      "learning_rate": 4.644192783776666e-05,
      "loss": 1.3206,
      "step": 2706
    },
    {
      "epoch": 0.07129312615222544,
      "grad_norm": 2.615851879119873,
      "learning_rate": 4.6440611008691074e-05,
      "loss": 1.705,
      "step": 2707
    },
    {
      "epoch": 0.07131946273373715,
      "grad_norm": 3.249099016189575,
      "learning_rate": 4.643929417961549e-05,
      "loss": 2.0197,
      "step": 2708
    },
    {
      "epoch": 0.07134579931524888,
      "grad_norm": 5.246429920196533,
      "learning_rate": 4.6437977350539905e-05,
      "loss": 2.0912,
      "step": 2709
    },
    {
      "epoch": 0.0713721358967606,
      "grad_norm": 1.8831321001052856,
      "learning_rate": 4.6436660521464314e-05,
      "loss": 1.8726,
      "step": 2710
    },
    {
      "epoch": 0.07139847247827232,
      "grad_norm": 3.412057399749756,
      "learning_rate": 4.643534369238873e-05,
      "loss": 1.4548,
      "step": 2711
    },
    {
      "epoch": 0.07142480905978404,
      "grad_norm": 1.715803623199463,
      "learning_rate": 4.6434026863313145e-05,
      "loss": 1.7469,
      "step": 2712
    },
    {
      "epoch": 0.07145114564129576,
      "grad_norm": 1.728883147239685,
      "learning_rate": 4.643271003423756e-05,
      "loss": 1.5368,
      "step": 2713
    },
    {
      "epoch": 0.07147748222280748,
      "grad_norm": 3.7357983589172363,
      "learning_rate": 4.643139320516197e-05,
      "loss": 1.9267,
      "step": 2714
    },
    {
      "epoch": 0.0715038188043192,
      "grad_norm": 4.025263786315918,
      "learning_rate": 4.6430076376086385e-05,
      "loss": 1.8416,
      "step": 2715
    },
    {
      "epoch": 0.07153015538583092,
      "grad_norm": 2.205474853515625,
      "learning_rate": 4.64287595470108e-05,
      "loss": 1.3173,
      "step": 2716
    },
    {
      "epoch": 0.07155649196734264,
      "grad_norm": 2.4038355350494385,
      "learning_rate": 4.6427442717935216e-05,
      "loss": 1.8354,
      "step": 2717
    },
    {
      "epoch": 0.07158282854885435,
      "grad_norm": 2.608717918395996,
      "learning_rate": 4.642612588885963e-05,
      "loss": 1.8021,
      "step": 2718
    },
    {
      "epoch": 0.07160916513036608,
      "grad_norm": 1.6370832920074463,
      "learning_rate": 4.642480905978404e-05,
      "loss": 1.5468,
      "step": 2719
    },
    {
      "epoch": 0.0716355017118778,
      "grad_norm": 2.068415641784668,
      "learning_rate": 4.6423492230708456e-05,
      "loss": 1.5858,
      "step": 2720
    },
    {
      "epoch": 0.07166183829338951,
      "grad_norm": 2.628811836242676,
      "learning_rate": 4.642217540163287e-05,
      "loss": 2.2224,
      "step": 2721
    },
    {
      "epoch": 0.07168817487490124,
      "grad_norm": 3.618770122528076,
      "learning_rate": 4.642085857255729e-05,
      "loss": 1.7167,
      "step": 2722
    },
    {
      "epoch": 0.07171451145641296,
      "grad_norm": 4.247048377990723,
      "learning_rate": 4.6419541743481696e-05,
      "loss": 1.224,
      "step": 2723
    },
    {
      "epoch": 0.07174084803792467,
      "grad_norm": 2.9350998401641846,
      "learning_rate": 4.641822491440611e-05,
      "loss": 2.1076,
      "step": 2724
    },
    {
      "epoch": 0.0717671846194364,
      "grad_norm": 3.7061140537261963,
      "learning_rate": 4.641690808533052e-05,
      "loss": 1.7952,
      "step": 2725
    },
    {
      "epoch": 0.07179352120094812,
      "grad_norm": 2.312399387359619,
      "learning_rate": 4.641559125625494e-05,
      "loss": 1.2502,
      "step": 2726
    },
    {
      "epoch": 0.07181985778245983,
      "grad_norm": 2.017063617706299,
      "learning_rate": 4.641427442717935e-05,
      "loss": 1.7595,
      "step": 2727
    },
    {
      "epoch": 0.07184619436397156,
      "grad_norm": 2.7324509620666504,
      "learning_rate": 4.641295759810377e-05,
      "loss": 1.33,
      "step": 2728
    },
    {
      "epoch": 0.07187253094548328,
      "grad_norm": 3.6948230266571045,
      "learning_rate": 4.641164076902818e-05,
      "loss": 2.3593,
      "step": 2729
    },
    {
      "epoch": 0.071898867526995,
      "grad_norm": 1.7741047143936157,
      "learning_rate": 4.64103239399526e-05,
      "loss": 1.5545,
      "step": 2730
    },
    {
      "epoch": 0.07192520410850671,
      "grad_norm": 5.119868278503418,
      "learning_rate": 4.6409007110877014e-05,
      "loss": 1.3571,
      "step": 2731
    },
    {
      "epoch": 0.07195154069001844,
      "grad_norm": 2.158773183822632,
      "learning_rate": 4.640769028180142e-05,
      "loss": 1.581,
      "step": 2732
    },
    {
      "epoch": 0.07197787727153016,
      "grad_norm": 2.9639108180999756,
      "learning_rate": 4.640637345272584e-05,
      "loss": 1.7442,
      "step": 2733
    },
    {
      "epoch": 0.07200421385304187,
      "grad_norm": 4.275511741638184,
      "learning_rate": 4.640505662365025e-05,
      "loss": 1.7204,
      "step": 2734
    },
    {
      "epoch": 0.0720305504345536,
      "grad_norm": 2.5602612495422363,
      "learning_rate": 4.640373979457467e-05,
      "loss": 2.0076,
      "step": 2735
    },
    {
      "epoch": 0.07205688701606532,
      "grad_norm": 2.8891618251800537,
      "learning_rate": 4.640242296549908e-05,
      "loss": 1.8932,
      "step": 2736
    },
    {
      "epoch": 0.07208322359757703,
      "grad_norm": 2.761899948120117,
      "learning_rate": 4.6401106136423494e-05,
      "loss": 1.7741,
      "step": 2737
    },
    {
      "epoch": 0.07210956017908876,
      "grad_norm": 4.940716743469238,
      "learning_rate": 4.639978930734791e-05,
      "loss": 1.1045,
      "step": 2738
    },
    {
      "epoch": 0.07213589676060048,
      "grad_norm": 2.88822340965271,
      "learning_rate": 4.639847247827232e-05,
      "loss": 0.7925,
      "step": 2739
    },
    {
      "epoch": 0.07216223334211219,
      "grad_norm": 2.7456512451171875,
      "learning_rate": 4.639715564919674e-05,
      "loss": 2.0357,
      "step": 2740
    },
    {
      "epoch": 0.07218856992362391,
      "grad_norm": 3.2925772666931152,
      "learning_rate": 4.639583882012115e-05,
      "loss": 1.6522,
      "step": 2741
    },
    {
      "epoch": 0.07221490650513564,
      "grad_norm": 2.1087934970855713,
      "learning_rate": 4.6394521991045565e-05,
      "loss": 1.4956,
      "step": 2742
    },
    {
      "epoch": 0.07224124308664735,
      "grad_norm": 2.858607292175293,
      "learning_rate": 4.6393205161969974e-05,
      "loss": 1.7281,
      "step": 2743
    },
    {
      "epoch": 0.07226757966815907,
      "grad_norm": 4.906359672546387,
      "learning_rate": 4.639188833289439e-05,
      "loss": 1.8161,
      "step": 2744
    },
    {
      "epoch": 0.0722939162496708,
      "grad_norm": 3.8215255737304688,
      "learning_rate": 4.6390571503818805e-05,
      "loss": 1.9722,
      "step": 2745
    },
    {
      "epoch": 0.07232025283118251,
      "grad_norm": 9.646496772766113,
      "learning_rate": 4.638925467474322e-05,
      "loss": 1.0756,
      "step": 2746
    },
    {
      "epoch": 0.07234658941269423,
      "grad_norm": 4.426901340484619,
      "learning_rate": 4.6387937845667637e-05,
      "loss": 0.9286,
      "step": 2747
    },
    {
      "epoch": 0.07237292599420596,
      "grad_norm": 4.849145412445068,
      "learning_rate": 4.6386621016592045e-05,
      "loss": 1.1735,
      "step": 2748
    },
    {
      "epoch": 0.07239926257571767,
      "grad_norm": NaN,
      "learning_rate": 4.6386621016592045e-05,
      "loss": 2.466,
      "step": 2749
    },
    {
      "epoch": 0.07242559915722939,
      "grad_norm": 2.6882951259613037,
      "learning_rate": 4.638530418751647e-05,
      "loss": 2.6353,
      "step": 2750
    },
    {
      "epoch": 0.0724519357387411,
      "grad_norm": 1.9634631872177124,
      "learning_rate": 4.6383987358440877e-05,
      "loss": 1.5359,
      "step": 2751
    },
    {
      "epoch": 0.07247827232025283,
      "grad_norm": 3.8971569538116455,
      "learning_rate": 4.638267052936529e-05,
      "loss": 1.5967,
      "step": 2752
    },
    {
      "epoch": 0.07250460890176455,
      "grad_norm": 4.099257469177246,
      "learning_rate": 4.63813537002897e-05,
      "loss": 1.8788,
      "step": 2753
    },
    {
      "epoch": 0.07253094548327627,
      "grad_norm": 1.8790581226348877,
      "learning_rate": 4.6380036871214117e-05,
      "loss": 2.2948,
      "step": 2754
    },
    {
      "epoch": 0.072557282064788,
      "grad_norm": 2.6746301651000977,
      "learning_rate": 4.637872004213853e-05,
      "loss": 1.6662,
      "step": 2755
    },
    {
      "epoch": 0.07258361864629971,
      "grad_norm": 2.729245185852051,
      "learning_rate": 4.637740321306295e-05,
      "loss": 2.0442,
      "step": 2756
    },
    {
      "epoch": 0.07260995522781143,
      "grad_norm": 1.7360540628433228,
      "learning_rate": 4.6376086383987363e-05,
      "loss": 2.1775,
      "step": 2757
    },
    {
      "epoch": 0.07263629180932316,
      "grad_norm": 1.9921828508377075,
      "learning_rate": 4.637476955491177e-05,
      "loss": 1.5739,
      "step": 2758
    },
    {
      "epoch": 0.07266262839083487,
      "grad_norm": 4.548732757568359,
      "learning_rate": 4.637345272583619e-05,
      "loss": 0.9905,
      "step": 2759
    },
    {
      "epoch": 0.07268896497234659,
      "grad_norm": 2.191924810409546,
      "learning_rate": 4.6372135896760603e-05,
      "loss": 1.5967,
      "step": 2760
    },
    {
      "epoch": 0.0727153015538583,
      "grad_norm": 2.550980806350708,
      "learning_rate": 4.637081906768502e-05,
      "loss": 2.0663,
      "step": 2761
    },
    {
      "epoch": 0.07274163813537003,
      "grad_norm": 2.215832233428955,
      "learning_rate": 4.636950223860943e-05,
      "loss": 1.0186,
      "step": 2762
    },
    {
      "epoch": 0.07276797471688175,
      "grad_norm": 1.8944953680038452,
      "learning_rate": 4.6368185409533843e-05,
      "loss": 1.9387,
      "step": 2763
    },
    {
      "epoch": 0.07279431129839346,
      "grad_norm": 2.056403160095215,
      "learning_rate": 4.636686858045826e-05,
      "loss": 2.1449,
      "step": 2764
    },
    {
      "epoch": 0.07282064787990519,
      "grad_norm": 2.469348192214966,
      "learning_rate": 4.6365551751382675e-05,
      "loss": 1.9617,
      "step": 2765
    },
    {
      "epoch": 0.07284698446141691,
      "grad_norm": 5.035886764526367,
      "learning_rate": 4.636423492230709e-05,
      "loss": 1.4871,
      "step": 2766
    },
    {
      "epoch": 0.07287332104292862,
      "grad_norm": 2.6207799911499023,
      "learning_rate": 4.63629180932315e-05,
      "loss": 1.7257,
      "step": 2767
    },
    {
      "epoch": 0.07289965762444035,
      "grad_norm": 2.999866008758545,
      "learning_rate": 4.6361601264155915e-05,
      "loss": 1.291,
      "step": 2768
    },
    {
      "epoch": 0.07292599420595207,
      "grad_norm": 4.365527629852295,
      "learning_rate": 4.636028443508033e-05,
      "loss": 1.7727,
      "step": 2769
    },
    {
      "epoch": 0.07295233078746378,
      "grad_norm": 1.7593533992767334,
      "learning_rate": 4.6358967606004746e-05,
      "loss": 1.4748,
      "step": 2770
    },
    {
      "epoch": 0.07297866736897551,
      "grad_norm": 4.576137065887451,
      "learning_rate": 4.6357650776929155e-05,
      "loss": 1.8213,
      "step": 2771
    },
    {
      "epoch": 0.07300500395048723,
      "grad_norm": 2.4472999572753906,
      "learning_rate": 4.635633394785357e-05,
      "loss": 2.0751,
      "step": 2772
    },
    {
      "epoch": 0.07303134053199895,
      "grad_norm": 1.9276094436645508,
      "learning_rate": 4.6355017118777986e-05,
      "loss": 1.4565,
      "step": 2773
    },
    {
      "epoch": 0.07305767711351066,
      "grad_norm": 3.5174710750579834,
      "learning_rate": 4.63537002897024e-05,
      "loss": 1.4744,
      "step": 2774
    },
    {
      "epoch": 0.07308401369502239,
      "grad_norm": 3.8854033946990967,
      "learning_rate": 4.635238346062682e-05,
      "loss": 0.8084,
      "step": 2775
    },
    {
      "epoch": 0.0731103502765341,
      "grad_norm": 2.2220587730407715,
      "learning_rate": 4.6351066631551226e-05,
      "loss": 0.8171,
      "step": 2776
    },
    {
      "epoch": 0.07313668685804582,
      "grad_norm": 2.2211291790008545,
      "learning_rate": 4.634974980247564e-05,
      "loss": 1.754,
      "step": 2777
    },
    {
      "epoch": 0.07316302343955755,
      "grad_norm": 3.8703322410583496,
      "learning_rate": 4.634843297340005e-05,
      "loss": 1.7574,
      "step": 2778
    },
    {
      "epoch": 0.07318936002106927,
      "grad_norm": 2.7567102909088135,
      "learning_rate": 4.634711614432447e-05,
      "loss": 0.5344,
      "step": 2779
    },
    {
      "epoch": 0.07321569660258098,
      "grad_norm": 2.733025550842285,
      "learning_rate": 4.634579931524888e-05,
      "loss": 0.7457,
      "step": 2780
    },
    {
      "epoch": 0.07324203318409271,
      "grad_norm": 2.9197328090667725,
      "learning_rate": 4.63444824861733e-05,
      "loss": 1.8035,
      "step": 2781
    },
    {
      "epoch": 0.07326836976560443,
      "grad_norm": 4.82131814956665,
      "learning_rate": 4.6343165657097706e-05,
      "loss": 1.3844,
      "step": 2782
    },
    {
      "epoch": 0.07329470634711614,
      "grad_norm": 2.0215184688568115,
      "learning_rate": 4.634184882802213e-05,
      "loss": 1.7277,
      "step": 2783
    },
    {
      "epoch": 0.07332104292862786,
      "grad_norm": 1.888427495956421,
      "learning_rate": 4.634053199894654e-05,
      "loss": 2.1028,
      "step": 2784
    },
    {
      "epoch": 0.07334737951013959,
      "grad_norm": 2.3030498027801514,
      "learning_rate": 4.633921516987095e-05,
      "loss": 1.8546,
      "step": 2785
    },
    {
      "epoch": 0.0733737160916513,
      "grad_norm": 2.647611379623413,
      "learning_rate": 4.633789834079537e-05,
      "loss": 1.9227,
      "step": 2786
    },
    {
      "epoch": 0.07340005267316302,
      "grad_norm": 2.333958148956299,
      "learning_rate": 4.633658151171978e-05,
      "loss": 1.9955,
      "step": 2787
    },
    {
      "epoch": 0.07342638925467475,
      "grad_norm": 3.4081473350524902,
      "learning_rate": 4.63352646826442e-05,
      "loss": 1.4516,
      "step": 2788
    },
    {
      "epoch": 0.07345272583618646,
      "grad_norm": 1.7533105611801147,
      "learning_rate": 4.633394785356861e-05,
      "loss": 1.5543,
      "step": 2789
    },
    {
      "epoch": 0.07347906241769818,
      "grad_norm": 3.1015400886535645,
      "learning_rate": 4.6332631024493024e-05,
      "loss": 0.3429,
      "step": 2790
    },
    {
      "epoch": 0.07350539899920991,
      "grad_norm": 1.7616078853607178,
      "learning_rate": 4.633131419541743e-05,
      "loss": 2.0438,
      "step": 2791
    },
    {
      "epoch": 0.07353173558072162,
      "grad_norm": 2.3825430870056152,
      "learning_rate": 4.632999736634185e-05,
      "loss": 1.4313,
      "step": 2792
    },
    {
      "epoch": 0.07355807216223334,
      "grad_norm": 3.4073896408081055,
      "learning_rate": 4.6328680537266264e-05,
      "loss": 1.5878,
      "step": 2793
    },
    {
      "epoch": 0.07358440874374506,
      "grad_norm": 2.3406107425689697,
      "learning_rate": 4.632736370819068e-05,
      "loss": 2.2283,
      "step": 2794
    },
    {
      "epoch": 0.07361074532525679,
      "grad_norm": 2.7205867767333984,
      "learning_rate": 4.6326046879115095e-05,
      "loss": 1.7144,
      "step": 2795
    },
    {
      "epoch": 0.0736370819067685,
      "grad_norm": 6.849981784820557,
      "learning_rate": 4.6324730050039504e-05,
      "loss": 1.2354,
      "step": 2796
    },
    {
      "epoch": 0.07366341848828022,
      "grad_norm": 2.2369143962860107,
      "learning_rate": 4.6323413220963926e-05,
      "loss": 1.9458,
      "step": 2797
    },
    {
      "epoch": 0.07368975506979195,
      "grad_norm": 2.0094168186187744,
      "learning_rate": 4.6322096391888335e-05,
      "loss": 1.8407,
      "step": 2798
    },
    {
      "epoch": 0.07371609165130366,
      "grad_norm": 2.328657627105713,
      "learning_rate": 4.632077956281275e-05,
      "loss": 1.1801,
      "step": 2799
    },
    {
      "epoch": 0.07374242823281538,
      "grad_norm": 2.2383835315704346,
      "learning_rate": 4.631946273373716e-05,
      "loss": 1.5971,
      "step": 2800
    },
    {
      "epoch": 0.0737687648143271,
      "grad_norm": 2.6518330574035645,
      "learning_rate": 4.6318145904661575e-05,
      "loss": 1.8028,
      "step": 2801
    },
    {
      "epoch": 0.07379510139583882,
      "grad_norm": 3.24520206451416,
      "learning_rate": 4.631682907558599e-05,
      "loss": 1.3934,
      "step": 2802
    },
    {
      "epoch": 0.07382143797735054,
      "grad_norm": 2.1288161277770996,
      "learning_rate": 4.6315512246510406e-05,
      "loss": 1.6207,
      "step": 2803
    },
    {
      "epoch": 0.07384777455886225,
      "grad_norm": 2.2702548503875732,
      "learning_rate": 4.631419541743482e-05,
      "loss": 1.5697,
      "step": 2804
    },
    {
      "epoch": 0.07387411114037398,
      "grad_norm": 1.9476163387298584,
      "learning_rate": 4.631287858835923e-05,
      "loss": 1.6175,
      "step": 2805
    },
    {
      "epoch": 0.0739004477218857,
      "grad_norm": 2.1629433631896973,
      "learning_rate": 4.6311561759283646e-05,
      "loss": 1.8719,
      "step": 2806
    },
    {
      "epoch": 0.07392678430339741,
      "grad_norm": 2.0354788303375244,
      "learning_rate": 4.631024493020806e-05,
      "loss": 2.6282,
      "step": 2807
    },
    {
      "epoch": 0.07395312088490914,
      "grad_norm": 3.1469314098358154,
      "learning_rate": 4.630892810113248e-05,
      "loss": 2.3025,
      "step": 2808
    },
    {
      "epoch": 0.07397945746642086,
      "grad_norm": 2.7114975452423096,
      "learning_rate": 4.6307611272056886e-05,
      "loss": 2.1133,
      "step": 2809
    },
    {
      "epoch": 0.07400579404793257,
      "grad_norm": 2.010967254638672,
      "learning_rate": 4.63062944429813e-05,
      "loss": 1.8658,
      "step": 2810
    },
    {
      "epoch": 0.0740321306294443,
      "grad_norm": 4.655788421630859,
      "learning_rate": 4.630497761390572e-05,
      "loss": 0.6612,
      "step": 2811
    },
    {
      "epoch": 0.07405846721095602,
      "grad_norm": 2.373643636703491,
      "learning_rate": 4.630366078483013e-05,
      "loss": 1.4071,
      "step": 2812
    },
    {
      "epoch": 0.07408480379246773,
      "grad_norm": 1.9067589044570923,
      "learning_rate": 4.630234395575455e-05,
      "loss": 2.3246,
      "step": 2813
    },
    {
      "epoch": 0.07411114037397945,
      "grad_norm": 1.7788095474243164,
      "learning_rate": 4.630102712667896e-05,
      "loss": 0.5815,
      "step": 2814
    },
    {
      "epoch": 0.07413747695549118,
      "grad_norm": 2.4540064334869385,
      "learning_rate": 4.629971029760337e-05,
      "loss": 2.1468,
      "step": 2815
    },
    {
      "epoch": 0.0741638135370029,
      "grad_norm": 3.333746910095215,
      "learning_rate": 4.629839346852779e-05,
      "loss": 1.8047,
      "step": 2816
    },
    {
      "epoch": 0.07419015011851461,
      "grad_norm": 3.1054818630218506,
      "learning_rate": 4.6297076639452204e-05,
      "loss": 1.43,
      "step": 2817
    },
    {
      "epoch": 0.07421648670002634,
      "grad_norm": 1.972234845161438,
      "learning_rate": 4.629575981037661e-05,
      "loss": 1.26,
      "step": 2818
    },
    {
      "epoch": 0.07424282328153806,
      "grad_norm": 2.5473155975341797,
      "learning_rate": 4.629444298130103e-05,
      "loss": 1.6786,
      "step": 2819
    },
    {
      "epoch": 0.07426915986304977,
      "grad_norm": 2.884748697280884,
      "learning_rate": 4.6293126152225444e-05,
      "loss": 1.4978,
      "step": 2820
    },
    {
      "epoch": 0.0742954964445615,
      "grad_norm": 2.6371817588806152,
      "learning_rate": 4.629180932314986e-05,
      "loss": 1.6876,
      "step": 2821
    },
    {
      "epoch": 0.07432183302607322,
      "grad_norm": 3.1315529346466064,
      "learning_rate": 4.6290492494074276e-05,
      "loss": 2.1545,
      "step": 2822
    },
    {
      "epoch": 0.07434816960758493,
      "grad_norm": 3.07360577583313,
      "learning_rate": 4.6289175664998684e-05,
      "loss": 1.89,
      "step": 2823
    },
    {
      "epoch": 0.07437450618909666,
      "grad_norm": 2.0975546836853027,
      "learning_rate": 4.62878588359231e-05,
      "loss": 0.5298,
      "step": 2824
    },
    {
      "epoch": 0.07440084277060838,
      "grad_norm": 2.6233761310577393,
      "learning_rate": 4.628654200684751e-05,
      "loss": 1.6488,
      "step": 2825
    },
    {
      "epoch": 0.07442717935212009,
      "grad_norm": 1.8994566202163696,
      "learning_rate": 4.628522517777193e-05,
      "loss": 1.9329,
      "step": 2826
    },
    {
      "epoch": 0.07445351593363181,
      "grad_norm": 1.836676836013794,
      "learning_rate": 4.628390834869634e-05,
      "loss": 1.715,
      "step": 2827
    },
    {
      "epoch": 0.07447985251514354,
      "grad_norm": 1.7107295989990234,
      "learning_rate": 4.6282591519620756e-05,
      "loss": 2.1464,
      "step": 2828
    },
    {
      "epoch": 0.07450618909665525,
      "grad_norm": 3.448990821838379,
      "learning_rate": 4.6281274690545164e-05,
      "loss": 1.5147,
      "step": 2829
    },
    {
      "epoch": 0.07453252567816697,
      "grad_norm": 2.2580504417419434,
      "learning_rate": 4.627995786146959e-05,
      "loss": 2.1167,
      "step": 2830
    },
    {
      "epoch": 0.0745588622596787,
      "grad_norm": 2.84763240814209,
      "learning_rate": 4.6278641032393996e-05,
      "loss": 1.2402,
      "step": 2831
    },
    {
      "epoch": 0.07458519884119041,
      "grad_norm": 2.063436508178711,
      "learning_rate": 4.627732420331841e-05,
      "loss": 2.049,
      "step": 2832
    },
    {
      "epoch": 0.07461153542270213,
      "grad_norm": 4.745668411254883,
      "learning_rate": 4.627600737424283e-05,
      "loss": 1.9608,
      "step": 2833
    },
    {
      "epoch": 0.07463787200421386,
      "grad_norm": 3.4372541904449463,
      "learning_rate": 4.6274690545167236e-05,
      "loss": 1.491,
      "step": 2834
    },
    {
      "epoch": 0.07466420858572557,
      "grad_norm": 2.8895318508148193,
      "learning_rate": 4.627337371609166e-05,
      "loss": 1.6019,
      "step": 2835
    },
    {
      "epoch": 0.07469054516723729,
      "grad_norm": 4.08427619934082,
      "learning_rate": 4.627205688701607e-05,
      "loss": 1.1404,
      "step": 2836
    },
    {
      "epoch": 0.074716881748749,
      "grad_norm": 2.6643011569976807,
      "learning_rate": 4.627074005794048e-05,
      "loss": 1.5322,
      "step": 2837
    },
    {
      "epoch": 0.07474321833026074,
      "grad_norm": 9.928352355957031,
      "learning_rate": 4.626942322886489e-05,
      "loss": 1.3646,
      "step": 2838
    },
    {
      "epoch": 0.07476955491177245,
      "grad_norm": 2.4245355129241943,
      "learning_rate": 4.626810639978931e-05,
      "loss": 1.6984,
      "step": 2839
    },
    {
      "epoch": 0.07479589149328417,
      "grad_norm": 3.102386474609375,
      "learning_rate": 4.626678957071372e-05,
      "loss": 0.578,
      "step": 2840
    },
    {
      "epoch": 0.0748222280747959,
      "grad_norm": 4.1624555587768555,
      "learning_rate": 4.626547274163814e-05,
      "loss": 1.5494,
      "step": 2841
    },
    {
      "epoch": 0.07484856465630761,
      "grad_norm": 6.126819610595703,
      "learning_rate": 4.6264155912562554e-05,
      "loss": 2.2051,
      "step": 2842
    },
    {
      "epoch": 0.07487490123781933,
      "grad_norm": 2.342970848083496,
      "learning_rate": 4.626283908348696e-05,
      "loss": 1.99,
      "step": 2843
    },
    {
      "epoch": 0.07490123781933106,
      "grad_norm": 2.877779006958008,
      "learning_rate": 4.626152225441138e-05,
      "loss": 1.5391,
      "step": 2844
    },
    {
      "epoch": 0.07492757440084277,
      "grad_norm": 3.9619438648223877,
      "learning_rate": 4.6260205425335794e-05,
      "loss": 0.6488,
      "step": 2845
    },
    {
      "epoch": 0.07495391098235449,
      "grad_norm": 2.223205327987671,
      "learning_rate": 4.625888859626021e-05,
      "loss": 1.1693,
      "step": 2846
    },
    {
      "epoch": 0.0749802475638662,
      "grad_norm": 3.2557179927825928,
      "learning_rate": 4.625757176718462e-05,
      "loss": 1.8447,
      "step": 2847
    },
    {
      "epoch": 0.07500658414537793,
      "grad_norm": 3.4557149410247803,
      "learning_rate": 4.6256254938109034e-05,
      "loss": 1.9351,
      "step": 2848
    },
    {
      "epoch": 0.07503292072688965,
      "grad_norm": 2.156149387359619,
      "learning_rate": 4.625493810903345e-05,
      "loss": 1.7212,
      "step": 2849
    },
    {
      "epoch": 0.07505925730840136,
      "grad_norm": 3.2540628910064697,
      "learning_rate": 4.6253621279957865e-05,
      "loss": 2.0247,
      "step": 2850
    },
    {
      "epoch": 0.0750855938899131,
      "grad_norm": 4.749756813049316,
      "learning_rate": 4.625230445088228e-05,
      "loss": 1.2521,
      "step": 2851
    },
    {
      "epoch": 0.07511193047142481,
      "grad_norm": 2.281892776489258,
      "learning_rate": 4.625098762180669e-05,
      "loss": 1.8424,
      "step": 2852
    },
    {
      "epoch": 0.07513826705293652,
      "grad_norm": 4.3686957359313965,
      "learning_rate": 4.6249670792731105e-05,
      "loss": 0.5506,
      "step": 2853
    },
    {
      "epoch": 0.07516460363444825,
      "grad_norm": 2.1627280712127686,
      "learning_rate": 4.624835396365552e-05,
      "loss": 2.2248,
      "step": 2854
    },
    {
      "epoch": 0.07519094021595997,
      "grad_norm": 2.4081296920776367,
      "learning_rate": 4.6247037134579936e-05,
      "loss": 1.6437,
      "step": 2855
    },
    {
      "epoch": 0.07521727679747169,
      "grad_norm": 1.8965318202972412,
      "learning_rate": 4.6245720305504345e-05,
      "loss": 1.3987,
      "step": 2856
    },
    {
      "epoch": 0.0752436133789834,
      "grad_norm": 2.997220754623413,
      "learning_rate": 4.624440347642876e-05,
      "loss": 2.5019,
      "step": 2857
    },
    {
      "epoch": 0.07526994996049513,
      "grad_norm": 2.150428533554077,
      "learning_rate": 4.6243086647353176e-05,
      "loss": 1.9793,
      "step": 2858
    },
    {
      "epoch": 0.07529628654200685,
      "grad_norm": 3.462949275970459,
      "learning_rate": 4.624176981827759e-05,
      "loss": 2.0437,
      "step": 2859
    },
    {
      "epoch": 0.07532262312351856,
      "grad_norm": 3.389911651611328,
      "learning_rate": 4.624045298920201e-05,
      "loss": 1.4694,
      "step": 2860
    },
    {
      "epoch": 0.07534895970503029,
      "grad_norm": 2.236562728881836,
      "learning_rate": 4.6239136160126416e-05,
      "loss": 2.0452,
      "step": 2861
    },
    {
      "epoch": 0.075375296286542,
      "grad_norm": 7.113264560699463,
      "learning_rate": 4.623781933105083e-05,
      "loss": 2.1405,
      "step": 2862
    },
    {
      "epoch": 0.07540163286805372,
      "grad_norm": 3.2841694355010986,
      "learning_rate": 4.623650250197525e-05,
      "loss": 1.3068,
      "step": 2863
    },
    {
      "epoch": 0.07542796944956545,
      "grad_norm": 7.079731464385986,
      "learning_rate": 4.623518567289966e-05,
      "loss": 1.5962,
      "step": 2864
    },
    {
      "epoch": 0.07545430603107717,
      "grad_norm": 2.9107322692871094,
      "learning_rate": 4.623386884382407e-05,
      "loss": 1.1992,
      "step": 2865
    },
    {
      "epoch": 0.07548064261258888,
      "grad_norm": 2.7110557556152344,
      "learning_rate": 4.623255201474849e-05,
      "loss": 1.9561,
      "step": 2866
    },
    {
      "epoch": 0.07550697919410061,
      "grad_norm": 2.516211986541748,
      "learning_rate": 4.62312351856729e-05,
      "loss": 2.2882,
      "step": 2867
    },
    {
      "epoch": 0.07553331577561233,
      "grad_norm": 2.0445072650909424,
      "learning_rate": 4.622991835659732e-05,
      "loss": 1.7402,
      "step": 2868
    },
    {
      "epoch": 0.07555965235712404,
      "grad_norm": 5.0942559242248535,
      "learning_rate": 4.6228601527521734e-05,
      "loss": 1.6377,
      "step": 2869
    },
    {
      "epoch": 0.07558598893863576,
      "grad_norm": 4.013828754425049,
      "learning_rate": 4.622728469844614e-05,
      "loss": 1.3201,
      "step": 2870
    },
    {
      "epoch": 0.07561232552014749,
      "grad_norm": 2.3567638397216797,
      "learning_rate": 4.622596786937056e-05,
      "loss": 0.6248,
      "step": 2871
    },
    {
      "epoch": 0.0756386621016592,
      "grad_norm": 4.042812824249268,
      "learning_rate": 4.622465104029497e-05,
      "loss": 1.4718,
      "step": 2872
    },
    {
      "epoch": 0.07566499868317092,
      "grad_norm": 2.1142780780792236,
      "learning_rate": 4.622333421121939e-05,
      "loss": 1.9396,
      "step": 2873
    },
    {
      "epoch": 0.07569133526468265,
      "grad_norm": 2.020040512084961,
      "learning_rate": 4.62220173821438e-05,
      "loss": 1.9907,
      "step": 2874
    },
    {
      "epoch": 0.07571767184619436,
      "grad_norm": 3.9280433654785156,
      "learning_rate": 4.6220700553068214e-05,
      "loss": 0.4253,
      "step": 2875
    },
    {
      "epoch": 0.07574400842770608,
      "grad_norm": 2.530353307723999,
      "learning_rate": 4.621938372399262e-05,
      "loss": 1.4493,
      "step": 2876
    },
    {
      "epoch": 0.07577034500921781,
      "grad_norm": 3.8539600372314453,
      "learning_rate": 4.621806689491704e-05,
      "loss": 1.9191,
      "step": 2877
    },
    {
      "epoch": 0.07579668159072953,
      "grad_norm": 1.726455807685852,
      "learning_rate": 4.621675006584146e-05,
      "loss": 1.4166,
      "step": 2878
    },
    {
      "epoch": 0.07582301817224124,
      "grad_norm": 1.8120595216751099,
      "learning_rate": 4.621543323676587e-05,
      "loss": 0.8819,
      "step": 2879
    },
    {
      "epoch": 0.07584935475375296,
      "grad_norm": 2.1814632415771484,
      "learning_rate": 4.6214116407690285e-05,
      "loss": 0.7063,
      "step": 2880
    },
    {
      "epoch": 0.07587569133526469,
      "grad_norm": 2.7255566120147705,
      "learning_rate": 4.6212799578614694e-05,
      "loss": 1.6888,
      "step": 2881
    },
    {
      "epoch": 0.0759020279167764,
      "grad_norm": 3.285147190093994,
      "learning_rate": 4.6211482749539117e-05,
      "loss": 1.525,
      "step": 2882
    },
    {
      "epoch": 0.07592836449828812,
      "grad_norm": 2.5586330890655518,
      "learning_rate": 4.6210165920463525e-05,
      "loss": 1.696,
      "step": 2883
    },
    {
      "epoch": 0.07595470107979985,
      "grad_norm": 2.305975914001465,
      "learning_rate": 4.620884909138794e-05,
      "loss": 1.3767,
      "step": 2884
    },
    {
      "epoch": 0.07598103766131156,
      "grad_norm": 7.872295379638672,
      "learning_rate": 4.620753226231235e-05,
      "loss": 0.7357,
      "step": 2885
    },
    {
      "epoch": 0.07600737424282328,
      "grad_norm": 1.8039056062698364,
      "learning_rate": 4.6206215433236765e-05,
      "loss": 1.8057,
      "step": 2886
    },
    {
      "epoch": 0.07603371082433501,
      "grad_norm": 3.1082780361175537,
      "learning_rate": 4.620489860416118e-05,
      "loss": 1.4222,
      "step": 2887
    },
    {
      "epoch": 0.07606004740584672,
      "grad_norm": 1.9146350622177124,
      "learning_rate": 4.62035817750856e-05,
      "loss": 1.9234,
      "step": 2888
    },
    {
      "epoch": 0.07608638398735844,
      "grad_norm": 2.7180566787719727,
      "learning_rate": 4.620226494601001e-05,
      "loss": 1.8174,
      "step": 2889
    },
    {
      "epoch": 0.07611272056887015,
      "grad_norm": 2.9868123531341553,
      "learning_rate": 4.620094811693442e-05,
      "loss": 1.9753,
      "step": 2890
    },
    {
      "epoch": 0.07613905715038188,
      "grad_norm": 2.3932347297668457,
      "learning_rate": 4.619963128785884e-05,
      "loss": 1.0506,
      "step": 2891
    },
    {
      "epoch": 0.0761653937318936,
      "grad_norm": 2.998116970062256,
      "learning_rate": 4.619831445878325e-05,
      "loss": 1.4265,
      "step": 2892
    },
    {
      "epoch": 0.07619173031340531,
      "grad_norm": 2.2100486755371094,
      "learning_rate": 4.619699762970767e-05,
      "loss": 1.5486,
      "step": 2893
    },
    {
      "epoch": 0.07621806689491704,
      "grad_norm": 2.5608062744140625,
      "learning_rate": 4.619568080063208e-05,
      "loss": 1.167,
      "step": 2894
    },
    {
      "epoch": 0.07624440347642876,
      "grad_norm": 6.639842510223389,
      "learning_rate": 4.619436397155649e-05,
      "loss": 1.3116,
      "step": 2895
    },
    {
      "epoch": 0.07627074005794048,
      "grad_norm": 2.0278637409210205,
      "learning_rate": 4.619304714248091e-05,
      "loss": 1.5837,
      "step": 2896
    },
    {
      "epoch": 0.0762970766394522,
      "grad_norm": 8.712678909301758,
      "learning_rate": 4.6191730313405323e-05,
      "loss": 1.5904,
      "step": 2897
    },
    {
      "epoch": 0.07632341322096392,
      "grad_norm": 3.8720109462738037,
      "learning_rate": 4.619041348432974e-05,
      "loss": 1.6175,
      "step": 2898
    },
    {
      "epoch": 0.07634974980247564,
      "grad_norm": 4.7755327224731445,
      "learning_rate": 4.618909665525415e-05,
      "loss": 1.3538,
      "step": 2899
    },
    {
      "epoch": 0.07637608638398735,
      "grad_norm": 2.2061166763305664,
      "learning_rate": 4.6187779826178563e-05,
      "loss": 1.4728,
      "step": 2900
    },
    {
      "epoch": 0.07640242296549908,
      "grad_norm": 7.621268272399902,
      "learning_rate": 4.618646299710298e-05,
      "loss": 1.6505,
      "step": 2901
    },
    {
      "epoch": 0.0764287595470108,
      "grad_norm": 3.170424461364746,
      "learning_rate": 4.6185146168027395e-05,
      "loss": 0.6962,
      "step": 2902
    },
    {
      "epoch": 0.07645509612852251,
      "grad_norm": 4.69133996963501,
      "learning_rate": 4.6183829338951804e-05,
      "loss": 1.7371,
      "step": 2903
    },
    {
      "epoch": 0.07648143271003424,
      "grad_norm": 3.802349328994751,
      "learning_rate": 4.618251250987622e-05,
      "loss": 0.4832,
      "step": 2904
    },
    {
      "epoch": 0.07650776929154596,
      "grad_norm": 2.538747787475586,
      "learning_rate": 4.6181195680800635e-05,
      "loss": 1.9829,
      "step": 2905
    },
    {
      "epoch": 0.07653410587305767,
      "grad_norm": 3.03297758102417,
      "learning_rate": 4.617987885172505e-05,
      "loss": 1.972,
      "step": 2906
    },
    {
      "epoch": 0.0765604424545694,
      "grad_norm": 2.6025068759918213,
      "learning_rate": 4.6178562022649466e-05,
      "loss": 1.4485,
      "step": 2907
    },
    {
      "epoch": 0.07658677903608112,
      "grad_norm": 2.474449396133423,
      "learning_rate": 4.6177245193573875e-05,
      "loss": 1.5531,
      "step": 2908
    },
    {
      "epoch": 0.07661311561759283,
      "grad_norm": 2.2864882946014404,
      "learning_rate": 4.617592836449829e-05,
      "loss": 2.2275,
      "step": 2909
    },
    {
      "epoch": 0.07663945219910456,
      "grad_norm": 5.040701866149902,
      "learning_rate": 4.61746115354227e-05,
      "loss": 2.3351,
      "step": 2910
    },
    {
      "epoch": 0.07666578878061628,
      "grad_norm": 2.6368234157562256,
      "learning_rate": 4.617329470634712e-05,
      "loss": 1.7002,
      "step": 2911
    },
    {
      "epoch": 0.076692125362128,
      "grad_norm": 3.585076093673706,
      "learning_rate": 4.617197787727153e-05,
      "loss": 1.1735,
      "step": 2912
    },
    {
      "epoch": 0.07671846194363971,
      "grad_norm": 2.4913384914398193,
      "learning_rate": 4.6170661048195946e-05,
      "loss": 1.3096,
      "step": 2913
    },
    {
      "epoch": 0.07674479852515144,
      "grad_norm": 2.65311336517334,
      "learning_rate": 4.616934421912036e-05,
      "loss": 0.9502,
      "step": 2914
    },
    {
      "epoch": 0.07677113510666315,
      "grad_norm": 2.1741271018981934,
      "learning_rate": 4.616802739004478e-05,
      "loss": 1.5825,
      "step": 2915
    },
    {
      "epoch": 0.07679747168817487,
      "grad_norm": 2.6662893295288086,
      "learning_rate": 4.616671056096919e-05,
      "loss": 1.4234,
      "step": 2916
    },
    {
      "epoch": 0.0768238082696866,
      "grad_norm": 2.2269794940948486,
      "learning_rate": 4.61653937318936e-05,
      "loss": 2.6377,
      "step": 2917
    },
    {
      "epoch": 0.07685014485119832,
      "grad_norm": 3.0671260356903076,
      "learning_rate": 4.616407690281802e-05,
      "loss": 1.3336,
      "step": 2918
    },
    {
      "epoch": 0.07687648143271003,
      "grad_norm": 4.627152442932129,
      "learning_rate": 4.6162760073742426e-05,
      "loss": 1.2017,
      "step": 2919
    },
    {
      "epoch": 0.07690281801422176,
      "grad_norm": 2.6742775440216064,
      "learning_rate": 4.616144324466685e-05,
      "loss": 1.9193,
      "step": 2920
    },
    {
      "epoch": 0.07692915459573348,
      "grad_norm": 4.334056377410889,
      "learning_rate": 4.616012641559126e-05,
      "loss": 1.7124,
      "step": 2921
    },
    {
      "epoch": 0.07695549117724519,
      "grad_norm": 3.9531803131103516,
      "learning_rate": 4.615880958651567e-05,
      "loss": 1.9557,
      "step": 2922
    },
    {
      "epoch": 0.07698182775875691,
      "grad_norm": 1.8117527961730957,
      "learning_rate": 4.615749275744009e-05,
      "loss": 2.0753,
      "step": 2923
    },
    {
      "epoch": 0.07700816434026864,
      "grad_norm": 2.708638906478882,
      "learning_rate": 4.61561759283645e-05,
      "loss": 2.0953,
      "step": 2924
    },
    {
      "epoch": 0.07703450092178035,
      "grad_norm": 4.326902866363525,
      "learning_rate": 4.615485909928892e-05,
      "loss": 2.238,
      "step": 2925
    },
    {
      "epoch": 0.07706083750329207,
      "grad_norm": 2.4174532890319824,
      "learning_rate": 4.615354227021333e-05,
      "loss": 1.7706,
      "step": 2926
    },
    {
      "epoch": 0.0770871740848038,
      "grad_norm": 2.5048410892486572,
      "learning_rate": 4.6152225441137744e-05,
      "loss": 2.3998,
      "step": 2927
    },
    {
      "epoch": 0.07711351066631551,
      "grad_norm": 1.947087049484253,
      "learning_rate": 4.615090861206215e-05,
      "loss": 2.1922,
      "step": 2928
    },
    {
      "epoch": 0.07713984724782723,
      "grad_norm": 2.6374828815460205,
      "learning_rate": 4.6149591782986575e-05,
      "loss": 1.994,
      "step": 2929
    },
    {
      "epoch": 0.07716618382933896,
      "grad_norm": 2.377981424331665,
      "learning_rate": 4.6148274953910984e-05,
      "loss": 1.6372,
      "step": 2930
    },
    {
      "epoch": 0.07719252041085067,
      "grad_norm": 2.6368439197540283,
      "learning_rate": 4.61469581248354e-05,
      "loss": 1.0267,
      "step": 2931
    },
    {
      "epoch": 0.07721885699236239,
      "grad_norm": 2.7755324840545654,
      "learning_rate": 4.614564129575981e-05,
      "loss": 2.0596,
      "step": 2932
    },
    {
      "epoch": 0.0772451935738741,
      "grad_norm": 3.1608474254608154,
      "learning_rate": 4.6144324466684224e-05,
      "loss": 1.355,
      "step": 2933
    },
    {
      "epoch": 0.07727153015538583,
      "grad_norm": 1.9804837703704834,
      "learning_rate": 4.614300763760864e-05,
      "loss": 2.3291,
      "step": 2934
    },
    {
      "epoch": 0.07729786673689755,
      "grad_norm": 1.8469260931015015,
      "learning_rate": 4.6141690808533055e-05,
      "loss": 1.9021,
      "step": 2935
    },
    {
      "epoch": 0.07732420331840927,
      "grad_norm": 2.2760767936706543,
      "learning_rate": 4.614037397945747e-05,
      "loss": 2.091,
      "step": 2936
    },
    {
      "epoch": 0.077350539899921,
      "grad_norm": 2.4643969535827637,
      "learning_rate": 4.613905715038188e-05,
      "loss": 1.9725,
      "step": 2937
    },
    {
      "epoch": 0.07737687648143271,
      "grad_norm": 5.384614944458008,
      "learning_rate": 4.6137740321306295e-05,
      "loss": 1.4552,
      "step": 2938
    },
    {
      "epoch": 0.07740321306294443,
      "grad_norm": 4.860240459442139,
      "learning_rate": 4.613642349223071e-05,
      "loss": 0.9666,
      "step": 2939
    },
    {
      "epoch": 0.07742954964445616,
      "grad_norm": 2.0333573818206787,
      "learning_rate": 4.6135106663155126e-05,
      "loss": 1.0944,
      "step": 2940
    },
    {
      "epoch": 0.07745588622596787,
      "grad_norm": 4.997055530548096,
      "learning_rate": 4.6133789834079535e-05,
      "loss": 1.4165,
      "step": 2941
    },
    {
      "epoch": 0.07748222280747959,
      "grad_norm": 4.055838584899902,
      "learning_rate": 4.613247300500395e-05,
      "loss": 1.3761,
      "step": 2942
    },
    {
      "epoch": 0.0775085593889913,
      "grad_norm": 2.942561149597168,
      "learning_rate": 4.6131156175928366e-05,
      "loss": 1.5633,
      "step": 2943
    },
    {
      "epoch": 0.07753489597050303,
      "grad_norm": 2.882842779159546,
      "learning_rate": 4.612983934685278e-05,
      "loss": 1.5864,
      "step": 2944
    },
    {
      "epoch": 0.07756123255201475,
      "grad_norm": 2.0620086193084717,
      "learning_rate": 4.61285225177772e-05,
      "loss": 1.9958,
      "step": 2945
    },
    {
      "epoch": 0.07758756913352646,
      "grad_norm": 4.141435146331787,
      "learning_rate": 4.6127205688701606e-05,
      "loss": 2.2422,
      "step": 2946
    },
    {
      "epoch": 0.07761390571503819,
      "grad_norm": 1.8920514583587646,
      "learning_rate": 4.612588885962602e-05,
      "loss": 1.7647,
      "step": 2947
    },
    {
      "epoch": 0.07764024229654991,
      "grad_norm": 1.988913655281067,
      "learning_rate": 4.612457203055044e-05,
      "loss": 1.7631,
      "step": 2948
    },
    {
      "epoch": 0.07766657887806162,
      "grad_norm": 11.161272048950195,
      "learning_rate": 4.612325520147485e-05,
      "loss": 1.6981,
      "step": 2949
    },
    {
      "epoch": 0.07769291545957335,
      "grad_norm": 3.5461437702178955,
      "learning_rate": 4.612193837239926e-05,
      "loss": 0.6758,
      "step": 2950
    },
    {
      "epoch": 0.07771925204108507,
      "grad_norm": 2.4620890617370605,
      "learning_rate": 4.612062154332368e-05,
      "loss": 2.3795,
      "step": 2951
    },
    {
      "epoch": 0.07774558862259678,
      "grad_norm": 2.7031571865081787,
      "learning_rate": 4.611930471424809e-05,
      "loss": 2.0075,
      "step": 2952
    },
    {
      "epoch": 0.07777192520410851,
      "grad_norm": 3.0769670009613037,
      "learning_rate": 4.611798788517251e-05,
      "loss": 1.2456,
      "step": 2953
    },
    {
      "epoch": 0.07779826178562023,
      "grad_norm": 7.008205413818359,
      "learning_rate": 4.6116671056096924e-05,
      "loss": 3.1135,
      "step": 2954
    },
    {
      "epoch": 0.07782459836713194,
      "grad_norm": 3.0412566661834717,
      "learning_rate": 4.611535422702133e-05,
      "loss": 1.9332,
      "step": 2955
    },
    {
      "epoch": 0.07785093494864366,
      "grad_norm": 2.492886781692505,
      "learning_rate": 4.611403739794575e-05,
      "loss": 0.3785,
      "step": 2956
    },
    {
      "epoch": 0.07787727153015539,
      "grad_norm": 2.2769527435302734,
      "learning_rate": 4.611272056887016e-05,
      "loss": 1.4016,
      "step": 2957
    },
    {
      "epoch": 0.0779036081116671,
      "grad_norm": 4.312548637390137,
      "learning_rate": 4.611140373979458e-05,
      "loss": 2.0011,
      "step": 2958
    },
    {
      "epoch": 0.07792994469317882,
      "grad_norm": 1.95980966091156,
      "learning_rate": 4.611008691071899e-05,
      "loss": 2.3541,
      "step": 2959
    },
    {
      "epoch": 0.07795628127469055,
      "grad_norm": 3.9868710041046143,
      "learning_rate": 4.6108770081643404e-05,
      "loss": 1.0886,
      "step": 2960
    },
    {
      "epoch": 0.07798261785620227,
      "grad_norm": 1.928377628326416,
      "learning_rate": 4.610745325256782e-05,
      "loss": 1.8441,
      "step": 2961
    },
    {
      "epoch": 0.07800895443771398,
      "grad_norm": 2.732083797454834,
      "learning_rate": 4.6106136423492236e-05,
      "loss": 1.5392,
      "step": 2962
    },
    {
      "epoch": 0.07803529101922571,
      "grad_norm": 5.650703430175781,
      "learning_rate": 4.610481959441665e-05,
      "loss": 0.7904,
      "step": 2963
    },
    {
      "epoch": 0.07806162760073743,
      "grad_norm": 1.938529133796692,
      "learning_rate": 4.610350276534106e-05,
      "loss": 1.845,
      "step": 2964
    },
    {
      "epoch": 0.07808796418224914,
      "grad_norm": 6.214179992675781,
      "learning_rate": 4.6102185936265476e-05,
      "loss": 1.4523,
      "step": 2965
    },
    {
      "epoch": 0.07811430076376086,
      "grad_norm": 4.206273078918457,
      "learning_rate": 4.6100869107189885e-05,
      "loss": 0.8772,
      "step": 2966
    },
    {
      "epoch": 0.07814063734527259,
      "grad_norm": 2.002572536468506,
      "learning_rate": 4.609955227811431e-05,
      "loss": 2.2036,
      "step": 2967
    },
    {
      "epoch": 0.0781669739267843,
      "grad_norm": 2.311776638031006,
      "learning_rate": 4.6098235449038716e-05,
      "loss": 1.8375,
      "step": 2968
    },
    {
      "epoch": 0.07819331050829602,
      "grad_norm": 2.337829828262329,
      "learning_rate": 4.609691861996313e-05,
      "loss": 1.8979,
      "step": 2969
    },
    {
      "epoch": 0.07821964708980775,
      "grad_norm": 3.020416736602783,
      "learning_rate": 4.609560179088755e-05,
      "loss": 0.9351,
      "step": 2970
    },
    {
      "epoch": 0.07824598367131946,
      "grad_norm": 5.188978672027588,
      "learning_rate": 4.6094284961811956e-05,
      "loss": 1.3461,
      "step": 2971
    },
    {
      "epoch": 0.07827232025283118,
      "grad_norm": 1.8756476640701294,
      "learning_rate": 4.609296813273638e-05,
      "loss": 1.492,
      "step": 2972
    },
    {
      "epoch": 0.07829865683434291,
      "grad_norm": 2.01534366607666,
      "learning_rate": 4.609165130366079e-05,
      "loss": 2.0342,
      "step": 2973
    },
    {
      "epoch": 0.07832499341585462,
      "grad_norm": 2.973799467086792,
      "learning_rate": 4.60903344745852e-05,
      "loss": 0.8264,
      "step": 2974
    },
    {
      "epoch": 0.07835132999736634,
      "grad_norm": 2.7090418338775635,
      "learning_rate": 4.608901764550961e-05,
      "loss": 1.9434,
      "step": 2975
    },
    {
      "epoch": 0.07837766657887806,
      "grad_norm": 1.9312663078308105,
      "learning_rate": 4.6087700816434034e-05,
      "loss": 0.5209,
      "step": 2976
    },
    {
      "epoch": 0.07840400316038978,
      "grad_norm": 2.0510969161987305,
      "learning_rate": 4.608638398735844e-05,
      "loss": 2.0458,
      "step": 2977
    },
    {
      "epoch": 0.0784303397419015,
      "grad_norm": 5.430920124053955,
      "learning_rate": 4.608506715828286e-05,
      "loss": 1.4459,
      "step": 2978
    },
    {
      "epoch": 0.07845667632341322,
      "grad_norm": 7.730001449584961,
      "learning_rate": 4.608375032920727e-05,
      "loss": 1.1434,
      "step": 2979
    },
    {
      "epoch": 0.07848301290492495,
      "grad_norm": 3.235725164413452,
      "learning_rate": 4.608243350013168e-05,
      "loss": 1.8174,
      "step": 2980
    },
    {
      "epoch": 0.07850934948643666,
      "grad_norm": 3.830639362335205,
      "learning_rate": 4.60811166710561e-05,
      "loss": 1.165,
      "step": 2981
    },
    {
      "epoch": 0.07853568606794838,
      "grad_norm": 2.1870667934417725,
      "learning_rate": 4.6079799841980514e-05,
      "loss": 1.4659,
      "step": 2982
    },
    {
      "epoch": 0.0785620226494601,
      "grad_norm": 1.709317922592163,
      "learning_rate": 4.607848301290493e-05,
      "loss": 1.7562,
      "step": 2983
    },
    {
      "epoch": 0.07858835923097182,
      "grad_norm": 2.114741086959839,
      "learning_rate": 4.607716618382934e-05,
      "loss": 2.4639,
      "step": 2984
    },
    {
      "epoch": 0.07861469581248354,
      "grad_norm": 3.909815549850464,
      "learning_rate": 4.6075849354753754e-05,
      "loss": 1.3364,
      "step": 2985
    },
    {
      "epoch": 0.07864103239399525,
      "grad_norm": 2.1716513633728027,
      "learning_rate": 4.607453252567817e-05,
      "loss": 2.0291,
      "step": 2986
    },
    {
      "epoch": 0.07866736897550698,
      "grad_norm": 1.996850609779358,
      "learning_rate": 4.6073215696602585e-05,
      "loss": 1.5977,
      "step": 2987
    },
    {
      "epoch": 0.0786937055570187,
      "grad_norm": 2.8697946071624756,
      "learning_rate": 4.6071898867526994e-05,
      "loss": 0.8517,
      "step": 2988
    },
    {
      "epoch": 0.07872004213853041,
      "grad_norm": 2.9075510501861572,
      "learning_rate": 4.607058203845141e-05,
      "loss": 1.3262,
      "step": 2989
    },
    {
      "epoch": 0.07874637872004214,
      "grad_norm": 3.050605535507202,
      "learning_rate": 4.6069265209375825e-05,
      "loss": 0.8237,
      "step": 2990
    },
    {
      "epoch": 0.07877271530155386,
      "grad_norm": 3.7436141967773438,
      "learning_rate": 4.606794838030024e-05,
      "loss": 1.6384,
      "step": 2991
    },
    {
      "epoch": 0.07879905188306557,
      "grad_norm": 4.508068561553955,
      "learning_rate": 4.6066631551224656e-05,
      "loss": 1.618,
      "step": 2992
    },
    {
      "epoch": 0.0788253884645773,
      "grad_norm": 1.782735824584961,
      "learning_rate": 4.6065314722149065e-05,
      "loss": 1.6343,
      "step": 2993
    },
    {
      "epoch": 0.07885172504608902,
      "grad_norm": 1.9519214630126953,
      "learning_rate": 4.606399789307348e-05,
      "loss": 1.9377,
      "step": 2994
    },
    {
      "epoch": 0.07887806162760073,
      "grad_norm": 2.068739414215088,
      "learning_rate": 4.6062681063997896e-05,
      "loss": 1.2672,
      "step": 2995
    },
    {
      "epoch": 0.07890439820911246,
      "grad_norm": 5.3604021072387695,
      "learning_rate": 4.606136423492231e-05,
      "loss": 1.3025,
      "step": 2996
    },
    {
      "epoch": 0.07893073479062418,
      "grad_norm": 1.7631012201309204,
      "learning_rate": 4.606004740584672e-05,
      "loss": 1.6901,
      "step": 2997
    },
    {
      "epoch": 0.0789570713721359,
      "grad_norm": 4.020761966705322,
      "learning_rate": 4.6058730576771136e-05,
      "loss": 1.8172,
      "step": 2998
    },
    {
      "epoch": 0.07898340795364761,
      "grad_norm": 2.0464046001434326,
      "learning_rate": 4.605741374769555e-05,
      "loss": 2.3855,
      "step": 2999
    },
    {
      "epoch": 0.07900974453515934,
      "grad_norm": 2.5864334106445312,
      "learning_rate": 4.605609691861997e-05,
      "loss": 1.9189,
      "step": 3000
    },
    {
      "epoch": 0.07903608111667106,
      "grad_norm": 3.0215535163879395,
      "learning_rate": 4.605478008954438e-05,
      "loss": 0.5917,
      "step": 3001
    },
    {
      "epoch": 0.07906241769818277,
      "grad_norm": 3.3121135234832764,
      "learning_rate": 4.605346326046879e-05,
      "loss": 2.4299,
      "step": 3002
    },
    {
      "epoch": 0.0790887542796945,
      "grad_norm": 2.774287462234497,
      "learning_rate": 4.605214643139321e-05,
      "loss": 2.07,
      "step": 3003
    },
    {
      "epoch": 0.07911509086120622,
      "grad_norm": 8.307010650634766,
      "learning_rate": 4.6050829602317616e-05,
      "loss": 1.2052,
      "step": 3004
    },
    {
      "epoch": 0.07914142744271793,
      "grad_norm": 1.8780688047409058,
      "learning_rate": 4.604951277324204e-05,
      "loss": 1.8467,
      "step": 3005
    },
    {
      "epoch": 0.07916776402422966,
      "grad_norm": 3.9588537216186523,
      "learning_rate": 4.604819594416645e-05,
      "loss": 1.4457,
      "step": 3006
    },
    {
      "epoch": 0.07919410060574138,
      "grad_norm": 1.9787161350250244,
      "learning_rate": 4.604687911509086e-05,
      "loss": 1.7519,
      "step": 3007
    },
    {
      "epoch": 0.07922043718725309,
      "grad_norm": 2.7040998935699463,
      "learning_rate": 4.604556228601528e-05,
      "loss": 1.5914,
      "step": 3008
    },
    {
      "epoch": 0.07924677376876481,
      "grad_norm": 2.356250047683716,
      "learning_rate": 4.6044245456939694e-05,
      "loss": 1.4581,
      "step": 3009
    },
    {
      "epoch": 0.07927311035027654,
      "grad_norm": 7.49885892868042,
      "learning_rate": 4.604292862786411e-05,
      "loss": 2.0232,
      "step": 3010
    },
    {
      "epoch": 0.07929944693178825,
      "grad_norm": 3.5787158012390137,
      "learning_rate": 4.604161179878852e-05,
      "loss": 1.452,
      "step": 3011
    },
    {
      "epoch": 0.07932578351329997,
      "grad_norm": 2.3142282962799072,
      "learning_rate": 4.6040294969712934e-05,
      "loss": 1.9553,
      "step": 3012
    },
    {
      "epoch": 0.0793521200948117,
      "grad_norm": 3.6337833404541016,
      "learning_rate": 4.603897814063734e-05,
      "loss": 1.4466,
      "step": 3013
    },
    {
      "epoch": 0.07937845667632341,
      "grad_norm": 3.2178382873535156,
      "learning_rate": 4.6037661311561765e-05,
      "loss": 2.0248,
      "step": 3014
    },
    {
      "epoch": 0.07940479325783513,
      "grad_norm": 2.0973870754241943,
      "learning_rate": 4.6036344482486174e-05,
      "loss": 2.4394,
      "step": 3015
    },
    {
      "epoch": 0.07943112983934686,
      "grad_norm": 3.0410025119781494,
      "learning_rate": 4.603502765341059e-05,
      "loss": 0.9493,
      "step": 3016
    },
    {
      "epoch": 0.07945746642085857,
      "grad_norm": 2.0005555152893066,
      "learning_rate": 4.6033710824335005e-05,
      "loss": 1.7615,
      "step": 3017
    },
    {
      "epoch": 0.07948380300237029,
      "grad_norm": 2.3067333698272705,
      "learning_rate": 4.6032393995259414e-05,
      "loss": 1.8746,
      "step": 3018
    },
    {
      "epoch": 0.079510139583882,
      "grad_norm": 3.151745557785034,
      "learning_rate": 4.603107716618384e-05,
      "loss": 1.1948,
      "step": 3019
    },
    {
      "epoch": 0.07953647616539374,
      "grad_norm": 2.040066719055176,
      "learning_rate": 4.6029760337108245e-05,
      "loss": 1.383,
      "step": 3020
    },
    {
      "epoch": 0.07956281274690545,
      "grad_norm": 5.638552188873291,
      "learning_rate": 4.602844350803266e-05,
      "loss": 1.2643,
      "step": 3021
    },
    {
      "epoch": 0.07958914932841717,
      "grad_norm": 3.586662769317627,
      "learning_rate": 4.602712667895707e-05,
      "loss": 1.6929,
      "step": 3022
    },
    {
      "epoch": 0.0796154859099289,
      "grad_norm": 2.253549098968506,
      "learning_rate": 4.6025809849881485e-05,
      "loss": 1.6356,
      "step": 3023
    },
    {
      "epoch": 0.07964182249144061,
      "grad_norm": 2.4989919662475586,
      "learning_rate": 4.60244930208059e-05,
      "loss": 1.3934,
      "step": 3024
    },
    {
      "epoch": 0.07966815907295233,
      "grad_norm": 2.0018396377563477,
      "learning_rate": 4.602317619173032e-05,
      "loss": 2.1064,
      "step": 3025
    },
    {
      "epoch": 0.07969449565446406,
      "grad_norm": 3.5105741024017334,
      "learning_rate": 4.602185936265473e-05,
      "loss": 0.7057,
      "step": 3026
    },
    {
      "epoch": 0.07972083223597577,
      "grad_norm": 2.9432575702667236,
      "learning_rate": 4.602054253357914e-05,
      "loss": 2.017,
      "step": 3027
    },
    {
      "epoch": 0.07974716881748749,
      "grad_norm": 2.45501446723938,
      "learning_rate": 4.6019225704503563e-05,
      "loss": 1.8069,
      "step": 3028
    },
    {
      "epoch": 0.0797735053989992,
      "grad_norm": 2.6317880153656006,
      "learning_rate": 4.601790887542797e-05,
      "loss": 1.1185,
      "step": 3029
    },
    {
      "epoch": 0.07979984198051093,
      "grad_norm": 1.8810955286026,
      "learning_rate": 4.601659204635239e-05,
      "loss": 0.757,
      "step": 3030
    },
    {
      "epoch": 0.07982617856202265,
      "grad_norm": 1.8643970489501953,
      "learning_rate": 4.60152752172768e-05,
      "loss": 2.212,
      "step": 3031
    },
    {
      "epoch": 0.07985251514353436,
      "grad_norm": 4.852502822875977,
      "learning_rate": 4.601395838820121e-05,
      "loss": 1.0029,
      "step": 3032
    },
    {
      "epoch": 0.0798788517250461,
      "grad_norm": 2.6382932662963867,
      "learning_rate": 4.601264155912563e-05,
      "loss": 1.3618,
      "step": 3033
    },
    {
      "epoch": 0.07990518830655781,
      "grad_norm": 2.009019136428833,
      "learning_rate": 4.6011324730050044e-05,
      "loss": 1.7227,
      "step": 3034
    },
    {
      "epoch": 0.07993152488806952,
      "grad_norm": 2.0312139987945557,
      "learning_rate": 4.601000790097445e-05,
      "loss": 1.0722,
      "step": 3035
    },
    {
      "epoch": 0.07995786146958125,
      "grad_norm": 2.287750720977783,
      "learning_rate": 4.600869107189887e-05,
      "loss": 2.1844,
      "step": 3036
    },
    {
      "epoch": 0.07998419805109297,
      "grad_norm": 2.407707929611206,
      "learning_rate": 4.6007374242823284e-05,
      "loss": 2.1177,
      "step": 3037
    },
    {
      "epoch": 0.08001053463260469,
      "grad_norm": 4.104620933532715,
      "learning_rate": 4.60060574137477e-05,
      "loss": 1.6166,
      "step": 3038
    },
    {
      "epoch": 0.08003687121411641,
      "grad_norm": 1.9677842855453491,
      "learning_rate": 4.6004740584672115e-05,
      "loss": 2.0887,
      "step": 3039
    },
    {
      "epoch": 0.08006320779562813,
      "grad_norm": 3.8040413856506348,
      "learning_rate": 4.6003423755596524e-05,
      "loss": 1.3351,
      "step": 3040
    },
    {
      "epoch": 0.08008954437713985,
      "grad_norm": 3.152820587158203,
      "learning_rate": 4.600210692652094e-05,
      "loss": 0.5594,
      "step": 3041
    },
    {
      "epoch": 0.08011588095865156,
      "grad_norm": 2.030339241027832,
      "learning_rate": 4.6000790097445355e-05,
      "loss": 1.8194,
      "step": 3042
    },
    {
      "epoch": 0.08014221754016329,
      "grad_norm": 2.9567949771881104,
      "learning_rate": 4.599947326836977e-05,
      "loss": 2.0513,
      "step": 3043
    },
    {
      "epoch": 0.080168554121675,
      "grad_norm": 2.779965400695801,
      "learning_rate": 4.599815643929418e-05,
      "loss": 1.0721,
      "step": 3044
    },
    {
      "epoch": 0.08019489070318672,
      "grad_norm": 3.786208391189575,
      "learning_rate": 4.5996839610218595e-05,
      "loss": 0.5259,
      "step": 3045
    },
    {
      "epoch": 0.08022122728469845,
      "grad_norm": 2.9450812339782715,
      "learning_rate": 4.599552278114301e-05,
      "loss": 0.7078,
      "step": 3046
    },
    {
      "epoch": 0.08024756386621017,
      "grad_norm": 1.8191871643066406,
      "learning_rate": 4.5994205952067426e-05,
      "loss": 1.6206,
      "step": 3047
    },
    {
      "epoch": 0.08027390044772188,
      "grad_norm": 3.709888219833374,
      "learning_rate": 4.599288912299184e-05,
      "loss": 1.5232,
      "step": 3048
    },
    {
      "epoch": 0.08030023702923361,
      "grad_norm": 3.206434726715088,
      "learning_rate": 4.599157229391625e-05,
      "loss": 0.6644,
      "step": 3049
    },
    {
      "epoch": 0.08032657361074533,
      "grad_norm": 8.217453002929688,
      "learning_rate": 4.5990255464840666e-05,
      "loss": 2.3626,
      "step": 3050
    },
    {
      "epoch": 0.08035291019225704,
      "grad_norm": 2.6956565380096436,
      "learning_rate": 4.5988938635765075e-05,
      "loss": 2.2998,
      "step": 3051
    },
    {
      "epoch": 0.08037924677376876,
      "grad_norm": 2.4701461791992188,
      "learning_rate": 4.59876218066895e-05,
      "loss": 1.7273,
      "step": 3052
    },
    {
      "epoch": 0.08040558335528049,
      "grad_norm": 4.820345878601074,
      "learning_rate": 4.5986304977613906e-05,
      "loss": 2.4332,
      "step": 3053
    },
    {
      "epoch": 0.0804319199367922,
      "grad_norm": 3.2415659427642822,
      "learning_rate": 4.598498814853832e-05,
      "loss": 1.9357,
      "step": 3054
    },
    {
      "epoch": 0.08045825651830392,
      "grad_norm": 3.080388307571411,
      "learning_rate": 4.598367131946274e-05,
      "loss": 2.0317,
      "step": 3055
    },
    {
      "epoch": 0.08048459309981565,
      "grad_norm": 2.7453036308288574,
      "learning_rate": 4.5982354490387146e-05,
      "loss": 1.9507,
      "step": 3056
    },
    {
      "epoch": 0.08051092968132736,
      "grad_norm": 1.9261882305145264,
      "learning_rate": 4.598103766131157e-05,
      "loss": 1.6568,
      "step": 3057
    },
    {
      "epoch": 0.08053726626283908,
      "grad_norm": 2.7540392875671387,
      "learning_rate": 4.597972083223598e-05,
      "loss": 1.5972,
      "step": 3058
    },
    {
      "epoch": 0.08056360284435081,
      "grad_norm": 6.094642162322998,
      "learning_rate": 4.597840400316039e-05,
      "loss": 1.7056,
      "step": 3059
    },
    {
      "epoch": 0.08058993942586253,
      "grad_norm": 1.748673915863037,
      "learning_rate": 4.59770871740848e-05,
      "loss": 1.7786,
      "step": 3060
    },
    {
      "epoch": 0.08061627600737424,
      "grad_norm": 2.5155184268951416,
      "learning_rate": 4.5975770345009224e-05,
      "loss": 2.1524,
      "step": 3061
    },
    {
      "epoch": 0.08064261258888596,
      "grad_norm": 5.381051540374756,
      "learning_rate": 4.597445351593363e-05,
      "loss": 1.9852,
      "step": 3062
    },
    {
      "epoch": 0.08066894917039769,
      "grad_norm": 1.9428106546401978,
      "learning_rate": 4.597313668685805e-05,
      "loss": 1.9864,
      "step": 3063
    },
    {
      "epoch": 0.0806952857519094,
      "grad_norm": 3.8232650756835938,
      "learning_rate": 4.5971819857782464e-05,
      "loss": 1.4981,
      "step": 3064
    },
    {
      "epoch": 0.08072162233342112,
      "grad_norm": 3.798069953918457,
      "learning_rate": 4.597050302870687e-05,
      "loss": 1.5659,
      "step": 3065
    },
    {
      "epoch": 0.08074795891493285,
      "grad_norm": 3.831735372543335,
      "learning_rate": 4.5969186199631295e-05,
      "loss": 0.9129,
      "step": 3066
    },
    {
      "epoch": 0.08077429549644456,
      "grad_norm": 2.117058277130127,
      "learning_rate": 4.5967869370555704e-05,
      "loss": 1.9879,
      "step": 3067
    },
    {
      "epoch": 0.08080063207795628,
      "grad_norm": 2.2150096893310547,
      "learning_rate": 4.596655254148012e-05,
      "loss": 2.1295,
      "step": 3068
    },
    {
      "epoch": 0.08082696865946801,
      "grad_norm": 3.386474132537842,
      "learning_rate": 4.596523571240453e-05,
      "loss": 1.8026,
      "step": 3069
    },
    {
      "epoch": 0.08085330524097972,
      "grad_norm": 3.0243096351623535,
      "learning_rate": 4.5963918883328944e-05,
      "loss": 1.653,
      "step": 3070
    },
    {
      "epoch": 0.08087964182249144,
      "grad_norm": 4.91994571685791,
      "learning_rate": 4.596260205425336e-05,
      "loss": 0.8772,
      "step": 3071
    },
    {
      "epoch": 0.08090597840400315,
      "grad_norm": 7.770267963409424,
      "learning_rate": 4.5961285225177775e-05,
      "loss": 1.2967,
      "step": 3072
    },
    {
      "epoch": 0.08093231498551488,
      "grad_norm": 2.2486369609832764,
      "learning_rate": 4.595996839610219e-05,
      "loss": 1.8587,
      "step": 3073
    },
    {
      "epoch": 0.0809586515670266,
      "grad_norm": 2.6649832725524902,
      "learning_rate": 4.59586515670266e-05,
      "loss": 1.575,
      "step": 3074
    },
    {
      "epoch": 0.08098498814853831,
      "grad_norm": 2.645906686782837,
      "learning_rate": 4.595733473795102e-05,
      "loss": 1.1733,
      "step": 3075
    },
    {
      "epoch": 0.08101132473005004,
      "grad_norm": 5.157594680786133,
      "learning_rate": 4.595601790887543e-05,
      "loss": 0.7833,
      "step": 3076
    },
    {
      "epoch": 0.08103766131156176,
      "grad_norm": 2.006969690322876,
      "learning_rate": 4.5954701079799846e-05,
      "loss": 1.4374,
      "step": 3077
    },
    {
      "epoch": 0.08106399789307348,
      "grad_norm": 2.1734561920166016,
      "learning_rate": 4.5953384250724255e-05,
      "loss": 2.2792,
      "step": 3078
    },
    {
      "epoch": 0.0810903344745852,
      "grad_norm": 2.1892194747924805,
      "learning_rate": 4.595206742164867e-05,
      "loss": 1.7573,
      "step": 3079
    },
    {
      "epoch": 0.08111667105609692,
      "grad_norm": 3.7811434268951416,
      "learning_rate": 4.5950750592573086e-05,
      "loss": 2.2416,
      "step": 3080
    },
    {
      "epoch": 0.08114300763760864,
      "grad_norm": 4.700692176818848,
      "learning_rate": 4.59494337634975e-05,
      "loss": 1.8576,
      "step": 3081
    },
    {
      "epoch": 0.08116934421912035,
      "grad_norm": 1.868679165840149,
      "learning_rate": 4.594811693442191e-05,
      "loss": 0.5832,
      "step": 3082
    },
    {
      "epoch": 0.08119568080063208,
      "grad_norm": 2.693378210067749,
      "learning_rate": 4.5946800105346326e-05,
      "loss": 1.9498,
      "step": 3083
    },
    {
      "epoch": 0.0812220173821438,
      "grad_norm": 3.871093511581421,
      "learning_rate": 4.594548327627074e-05,
      "loss": 1.1099,
      "step": 3084
    },
    {
      "epoch": 0.08124835396365551,
      "grad_norm": 6.053171634674072,
      "learning_rate": 4.594416644719516e-05,
      "loss": 1.9812,
      "step": 3085
    },
    {
      "epoch": 0.08127469054516724,
      "grad_norm": 2.1427392959594727,
      "learning_rate": 4.594284961811957e-05,
      "loss": 1.6059,
      "step": 3086
    },
    {
      "epoch": 0.08130102712667896,
      "grad_norm": 2.369745969772339,
      "learning_rate": 4.594153278904398e-05,
      "loss": 0.469,
      "step": 3087
    },
    {
      "epoch": 0.08132736370819067,
      "grad_norm": 2.579380512237549,
      "learning_rate": 4.59402159599684e-05,
      "loss": 0.7767,
      "step": 3088
    },
    {
      "epoch": 0.0813537002897024,
      "grad_norm": 2.1181387901306152,
      "learning_rate": 4.5938899130892807e-05,
      "loss": 1.9556,
      "step": 3089
    },
    {
      "epoch": 0.08138003687121412,
      "grad_norm": 9.613819122314453,
      "learning_rate": 4.593758230181723e-05,
      "loss": 2.4814,
      "step": 3090
    },
    {
      "epoch": 0.08140637345272583,
      "grad_norm": 2.4750685691833496,
      "learning_rate": 4.593626547274164e-05,
      "loss": 2.0728,
      "step": 3091
    },
    {
      "epoch": 0.08143271003423756,
      "grad_norm": 1.8456908464431763,
      "learning_rate": 4.593494864366605e-05,
      "loss": 1.4081,
      "step": 3092
    },
    {
      "epoch": 0.08145904661574928,
      "grad_norm": 2.7805752754211426,
      "learning_rate": 4.593363181459047e-05,
      "loss": 1.8494,
      "step": 3093
    },
    {
      "epoch": 0.081485383197261,
      "grad_norm": 2.2097983360290527,
      "learning_rate": 4.5932314985514885e-05,
      "loss": 1.4948,
      "step": 3094
    },
    {
      "epoch": 0.08151171977877271,
      "grad_norm": 3.662487745285034,
      "learning_rate": 4.59309981564393e-05,
      "loss": 0.5671,
      "step": 3095
    },
    {
      "epoch": 0.08153805636028444,
      "grad_norm": 2.623788356781006,
      "learning_rate": 4.592968132736371e-05,
      "loss": 1.7313,
      "step": 3096
    },
    {
      "epoch": 0.08156439294179615,
      "grad_norm": 1.7780513763427734,
      "learning_rate": 4.5928364498288125e-05,
      "loss": 1.422,
      "step": 3097
    },
    {
      "epoch": 0.08159072952330787,
      "grad_norm": 2.2921080589294434,
      "learning_rate": 4.592704766921253e-05,
      "loss": 2.0359,
      "step": 3098
    },
    {
      "epoch": 0.0816170661048196,
      "grad_norm": 2.3026371002197266,
      "learning_rate": 4.5925730840136956e-05,
      "loss": 1.9653,
      "step": 3099
    },
    {
      "epoch": 0.08164340268633132,
      "grad_norm": 1.8745219707489014,
      "learning_rate": 4.5924414011061365e-05,
      "loss": 1.7081,
      "step": 3100
    },
    {
      "epoch": 0.08166973926784303,
      "grad_norm": 2.1467669010162354,
      "learning_rate": 4.592309718198578e-05,
      "loss": 2.3063,
      "step": 3101
    },
    {
      "epoch": 0.08169607584935476,
      "grad_norm": 6.352311611175537,
      "learning_rate": 4.5921780352910196e-05,
      "loss": 0.8252,
      "step": 3102
    },
    {
      "epoch": 0.08172241243086648,
      "grad_norm": 2.388911724090576,
      "learning_rate": 4.5920463523834605e-05,
      "loss": 1.4659,
      "step": 3103
    },
    {
      "epoch": 0.08174874901237819,
      "grad_norm": 3.9200141429901123,
      "learning_rate": 4.591914669475903e-05,
      "loss": 1.8962,
      "step": 3104
    },
    {
      "epoch": 0.0817750855938899,
      "grad_norm": 2.021963357925415,
      "learning_rate": 4.5917829865683436e-05,
      "loss": 2.1938,
      "step": 3105
    },
    {
      "epoch": 0.08180142217540164,
      "grad_norm": 1.8183528184890747,
      "learning_rate": 4.591651303660785e-05,
      "loss": 1.7039,
      "step": 3106
    },
    {
      "epoch": 0.08182775875691335,
      "grad_norm": 3.588721752166748,
      "learning_rate": 4.591519620753226e-05,
      "loss": 1.4668,
      "step": 3107
    },
    {
      "epoch": 0.08185409533842507,
      "grad_norm": 4.492329120635986,
      "learning_rate": 4.591387937845668e-05,
      "loss": 1.3517,
      "step": 3108
    },
    {
      "epoch": 0.0818804319199368,
      "grad_norm": 2.134694814682007,
      "learning_rate": 4.591256254938109e-05,
      "loss": 0.7819,
      "step": 3109
    },
    {
      "epoch": 0.08190676850144851,
      "grad_norm": 2.926333427429199,
      "learning_rate": 4.591124572030551e-05,
      "loss": 1.5406,
      "step": 3110
    },
    {
      "epoch": 0.08193310508296023,
      "grad_norm": 5.121904373168945,
      "learning_rate": 4.590992889122992e-05,
      "loss": 0.9217,
      "step": 3111
    },
    {
      "epoch": 0.08195944166447196,
      "grad_norm": 1.8765417337417603,
      "learning_rate": 4.590861206215433e-05,
      "loss": 2.0339,
      "step": 3112
    },
    {
      "epoch": 0.08198577824598367,
      "grad_norm": 5.389040470123291,
      "learning_rate": 4.5907295233078754e-05,
      "loss": 1.0556,
      "step": 3113
    },
    {
      "epoch": 0.08201211482749539,
      "grad_norm": 2.2882680892944336,
      "learning_rate": 4.590597840400316e-05,
      "loss": 1.8318,
      "step": 3114
    },
    {
      "epoch": 0.0820384514090071,
      "grad_norm": 2.079467296600342,
      "learning_rate": 4.590466157492758e-05,
      "loss": 1.736,
      "step": 3115
    },
    {
      "epoch": 0.08206478799051883,
      "grad_norm": 1.993194818496704,
      "learning_rate": 4.590334474585199e-05,
      "loss": 2.0398,
      "step": 3116
    },
    {
      "epoch": 0.08209112457203055,
      "grad_norm": 4.376698017120361,
      "learning_rate": 4.59020279167764e-05,
      "loss": 1.1316,
      "step": 3117
    },
    {
      "epoch": 0.08211746115354226,
      "grad_norm": 2.508852243423462,
      "learning_rate": 4.590071108770082e-05,
      "loss": 2.047,
      "step": 3118
    },
    {
      "epoch": 0.082143797735054,
      "grad_norm": 7.158236026763916,
      "learning_rate": 4.5899394258625234e-05,
      "loss": 0.8504,
      "step": 3119
    },
    {
      "epoch": 0.08217013431656571,
      "grad_norm": 3.2473576068878174,
      "learning_rate": 4.589807742954965e-05,
      "loss": 1.3946,
      "step": 3120
    },
    {
      "epoch": 0.08219647089807743,
      "grad_norm": 5.032447814941406,
      "learning_rate": 4.589676060047406e-05,
      "loss": 1.9325,
      "step": 3121
    },
    {
      "epoch": 0.08222280747958916,
      "grad_norm": 3.326627731323242,
      "learning_rate": 4.5895443771398474e-05,
      "loss": 1.6226,
      "step": 3122
    },
    {
      "epoch": 0.08224914406110087,
      "grad_norm": 3.977592945098877,
      "learning_rate": 4.589412694232289e-05,
      "loss": 1.6423,
      "step": 3123
    },
    {
      "epoch": 0.08227548064261259,
      "grad_norm": 2.88101863861084,
      "learning_rate": 4.5892810113247305e-05,
      "loss": 2.1351,
      "step": 3124
    },
    {
      "epoch": 0.0823018172241243,
      "grad_norm": 1.8409802913665771,
      "learning_rate": 4.5891493284171714e-05,
      "loss": 2.2556,
      "step": 3125
    },
    {
      "epoch": 0.08232815380563603,
      "grad_norm": 2.643402338027954,
      "learning_rate": 4.589017645509613e-05,
      "loss": 0.7815,
      "step": 3126
    },
    {
      "epoch": 0.08235449038714775,
      "grad_norm": 6.87968635559082,
      "learning_rate": 4.5888859626020545e-05,
      "loss": 2.7936,
      "step": 3127
    },
    {
      "epoch": 0.08238082696865946,
      "grad_norm": 5.262762069702148,
      "learning_rate": 4.588754279694496e-05,
      "loss": 1.8864,
      "step": 3128
    },
    {
      "epoch": 0.08240716355017119,
      "grad_norm": 2.5659544467926025,
      "learning_rate": 4.5886225967869376e-05,
      "loss": 2.2452,
      "step": 3129
    },
    {
      "epoch": 0.08243350013168291,
      "grad_norm": 1.4848806858062744,
      "learning_rate": 4.5884909138793785e-05,
      "loss": 1.3881,
      "step": 3130
    },
    {
      "epoch": 0.08245983671319462,
      "grad_norm": 3.090043783187866,
      "learning_rate": 4.58835923097182e-05,
      "loss": 0.4795,
      "step": 3131
    },
    {
      "epoch": 0.08248617329470635,
      "grad_norm": 5.4616241455078125,
      "learning_rate": 4.5882275480642616e-05,
      "loss": 1.7237,
      "step": 3132
    },
    {
      "epoch": 0.08251250987621807,
      "grad_norm": 3.349935293197632,
      "learning_rate": 4.588095865156703e-05,
      "loss": 1.4697,
      "step": 3133
    },
    {
      "epoch": 0.08253884645772978,
      "grad_norm": 3.1029186248779297,
      "learning_rate": 4.587964182249144e-05,
      "loss": 1.9954,
      "step": 3134
    },
    {
      "epoch": 0.08256518303924151,
      "grad_norm": 1.8943114280700684,
      "learning_rate": 4.5878324993415856e-05,
      "loss": 1.7665,
      "step": 3135
    },
    {
      "epoch": 0.08259151962075323,
      "grad_norm": 2.9870240688323975,
      "learning_rate": 4.5877008164340265e-05,
      "loss": 1.3939,
      "step": 3136
    },
    {
      "epoch": 0.08261785620226494,
      "grad_norm": 2.2673745155334473,
      "learning_rate": 4.587569133526469e-05,
      "loss": 0.6984,
      "step": 3137
    },
    {
      "epoch": 0.08264419278377666,
      "grad_norm": 3.3405747413635254,
      "learning_rate": 4.5874374506189096e-05,
      "loss": 1.8298,
      "step": 3138
    },
    {
      "epoch": 0.08267052936528839,
      "grad_norm": 3.767385482788086,
      "learning_rate": 4.587305767711351e-05,
      "loss": 2.4307,
      "step": 3139
    },
    {
      "epoch": 0.0826968659468001,
      "grad_norm": 1.9182902574539185,
      "learning_rate": 4.587174084803793e-05,
      "loss": 2.1343,
      "step": 3140
    },
    {
      "epoch": 0.08272320252831182,
      "grad_norm": 3.1401264667510986,
      "learning_rate": 4.587042401896234e-05,
      "loss": 1.8234,
      "step": 3141
    },
    {
      "epoch": 0.08274953910982355,
      "grad_norm": 3.5475192070007324,
      "learning_rate": 4.586910718988676e-05,
      "loss": 1.2326,
      "step": 3142
    },
    {
      "epoch": 0.08277587569133527,
      "grad_norm": 2.9847376346588135,
      "learning_rate": 4.586779036081117e-05,
      "loss": 1.4619,
      "step": 3143
    },
    {
      "epoch": 0.08280221227284698,
      "grad_norm": 2.3415334224700928,
      "learning_rate": 4.586647353173558e-05,
      "loss": 1.7182,
      "step": 3144
    },
    {
      "epoch": 0.08282854885435871,
      "grad_norm": 2.0909996032714844,
      "learning_rate": 4.586515670265999e-05,
      "loss": 1.9498,
      "step": 3145
    },
    {
      "epoch": 0.08285488543587043,
      "grad_norm": 2.3579351902008057,
      "learning_rate": 4.5863839873584414e-05,
      "loss": 0.7382,
      "step": 3146
    },
    {
      "epoch": 0.08288122201738214,
      "grad_norm": 4.94607400894165,
      "learning_rate": 4.586252304450882e-05,
      "loss": 1.8791,
      "step": 3147
    },
    {
      "epoch": 0.08290755859889386,
      "grad_norm": 2.106743812561035,
      "learning_rate": 4.586120621543324e-05,
      "loss": 1.7887,
      "step": 3148
    },
    {
      "epoch": 0.08293389518040559,
      "grad_norm": 1.873420000076294,
      "learning_rate": 4.5859889386357654e-05,
      "loss": 1.9993,
      "step": 3149
    },
    {
      "epoch": 0.0829602317619173,
      "grad_norm": 1.831897497177124,
      "learning_rate": 4.585857255728206e-05,
      "loss": 1.8923,
      "step": 3150
    },
    {
      "epoch": 0.08298656834342902,
      "grad_norm": 3.0289714336395264,
      "learning_rate": 4.5857255728206485e-05,
      "loss": 1.9192,
      "step": 3151
    },
    {
      "epoch": 0.08301290492494075,
      "grad_norm": 2.8784220218658447,
      "learning_rate": 4.5855938899130894e-05,
      "loss": 1.5155,
      "step": 3152
    },
    {
      "epoch": 0.08303924150645246,
      "grad_norm": 6.403001308441162,
      "learning_rate": 4.585462207005531e-05,
      "loss": 1.0225,
      "step": 3153
    },
    {
      "epoch": 0.08306557808796418,
      "grad_norm": 2.122565984725952,
      "learning_rate": 4.585330524097972e-05,
      "loss": 1.9004,
      "step": 3154
    },
    {
      "epoch": 0.08309191466947591,
      "grad_norm": 2.0493125915527344,
      "learning_rate": 4.5851988411904134e-05,
      "loss": 2.023,
      "step": 3155
    },
    {
      "epoch": 0.08311825125098762,
      "grad_norm": 2.776400089263916,
      "learning_rate": 4.585067158282855e-05,
      "loss": 2.266,
      "step": 3156
    },
    {
      "epoch": 0.08314458783249934,
      "grad_norm": 4.151550769805908,
      "learning_rate": 4.5849354753752966e-05,
      "loss": 0.8544,
      "step": 3157
    },
    {
      "epoch": 0.08317092441401105,
      "grad_norm": 1.9945951700210571,
      "learning_rate": 4.584803792467738e-05,
      "loss": 1.3352,
      "step": 3158
    },
    {
      "epoch": 0.08319726099552278,
      "grad_norm": 1.8696441650390625,
      "learning_rate": 4.584672109560179e-05,
      "loss": 1.3425,
      "step": 3159
    },
    {
      "epoch": 0.0832235975770345,
      "grad_norm": 2.921455144882202,
      "learning_rate": 4.584540426652621e-05,
      "loss": 1.5419,
      "step": 3160
    },
    {
      "epoch": 0.08324993415854622,
      "grad_norm": 2.025024652481079,
      "learning_rate": 4.584408743745062e-05,
      "loss": 1.6507,
      "step": 3161
    },
    {
      "epoch": 0.08327627074005794,
      "grad_norm": 2.3824870586395264,
      "learning_rate": 4.584277060837504e-05,
      "loss": 1.9859,
      "step": 3162
    },
    {
      "epoch": 0.08330260732156966,
      "grad_norm": 2.3922998905181885,
      "learning_rate": 4.5841453779299446e-05,
      "loss": 2.1497,
      "step": 3163
    },
    {
      "epoch": 0.08332894390308138,
      "grad_norm": 1.7682502269744873,
      "learning_rate": 4.584013695022386e-05,
      "loss": 1.5649,
      "step": 3164
    },
    {
      "epoch": 0.0833552804845931,
      "grad_norm": 3.6769139766693115,
      "learning_rate": 4.583882012114828e-05,
      "loss": 1.7707,
      "step": 3165
    },
    {
      "epoch": 0.08338161706610482,
      "grad_norm": 2.7153096199035645,
      "learning_rate": 4.583750329207269e-05,
      "loss": 0.948,
      "step": 3166
    },
    {
      "epoch": 0.08340795364761654,
      "grad_norm": 2.6838431358337402,
      "learning_rate": 4.583618646299711e-05,
      "loss": 1.1769,
      "step": 3167
    },
    {
      "epoch": 0.08343429022912825,
      "grad_norm": 3.939176559448242,
      "learning_rate": 4.583486963392152e-05,
      "loss": 1.149,
      "step": 3168
    },
    {
      "epoch": 0.08346062681063998,
      "grad_norm": 4.521262168884277,
      "learning_rate": 4.583355280484593e-05,
      "loss": 0.6673,
      "step": 3169
    },
    {
      "epoch": 0.0834869633921517,
      "grad_norm": 3.738725185394287,
      "learning_rate": 4.583223597577035e-05,
      "loss": 1.2397,
      "step": 3170
    },
    {
      "epoch": 0.08351329997366341,
      "grad_norm": 2.1133363246917725,
      "learning_rate": 4.5830919146694764e-05,
      "loss": 1.4473,
      "step": 3171
    },
    {
      "epoch": 0.08353963655517514,
      "grad_norm": 2.7973620891571045,
      "learning_rate": 4.582960231761917e-05,
      "loss": 2.2941,
      "step": 3172
    },
    {
      "epoch": 0.08356597313668686,
      "grad_norm": 3.1256542205810547,
      "learning_rate": 4.582828548854359e-05,
      "loss": 1.3881,
      "step": 3173
    },
    {
      "epoch": 0.08359230971819857,
      "grad_norm": 3.162682294845581,
      "learning_rate": 4.5826968659468004e-05,
      "loss": 1.7668,
      "step": 3174
    },
    {
      "epoch": 0.0836186462997103,
      "grad_norm": 2.5578365325927734,
      "learning_rate": 4.582565183039242e-05,
      "loss": 1.2669,
      "step": 3175
    },
    {
      "epoch": 0.08364498288122202,
      "grad_norm": 2.274585723876953,
      "learning_rate": 4.5824335001316835e-05,
      "loss": 1.7379,
      "step": 3176
    },
    {
      "epoch": 0.08367131946273373,
      "grad_norm": 2.08831787109375,
      "learning_rate": 4.5823018172241244e-05,
      "loss": 1.5238,
      "step": 3177
    },
    {
      "epoch": 0.08369765604424546,
      "grad_norm": 3.0488479137420654,
      "learning_rate": 4.582170134316566e-05,
      "loss": 1.7891,
      "step": 3178
    },
    {
      "epoch": 0.08372399262575718,
      "grad_norm": 3.5930607318878174,
      "learning_rate": 4.5820384514090075e-05,
      "loss": 1.6324,
      "step": 3179
    },
    {
      "epoch": 0.0837503292072689,
      "grad_norm": 3.64334774017334,
      "learning_rate": 4.581906768501449e-05,
      "loss": 1.7488,
      "step": 3180
    },
    {
      "epoch": 0.08377666578878061,
      "grad_norm": 3.94474458694458,
      "learning_rate": 4.58177508559389e-05,
      "loss": 0.6436,
      "step": 3181
    },
    {
      "epoch": 0.08380300237029234,
      "grad_norm": 3.3415911197662354,
      "learning_rate": 4.5816434026863315e-05,
      "loss": 1.619,
      "step": 3182
    },
    {
      "epoch": 0.08382933895180406,
      "grad_norm": 3.105062484741211,
      "learning_rate": 4.5815117197787724e-05,
      "loss": 1.5413,
      "step": 3183
    },
    {
      "epoch": 0.08385567553331577,
      "grad_norm": 1.6706628799438477,
      "learning_rate": 4.5813800368712146e-05,
      "loss": 1.9191,
      "step": 3184
    },
    {
      "epoch": 0.0838820121148275,
      "grad_norm": 3.660766363143921,
      "learning_rate": 4.5812483539636555e-05,
      "loss": 2.1776,
      "step": 3185
    },
    {
      "epoch": 0.08390834869633922,
      "grad_norm": 1.9701188802719116,
      "learning_rate": 4.581116671056097e-05,
      "loss": 1.7777,
      "step": 3186
    },
    {
      "epoch": 0.08393468527785093,
      "grad_norm": 2.0551421642303467,
      "learning_rate": 4.5809849881485386e-05,
      "loss": 1.9264,
      "step": 3187
    },
    {
      "epoch": 0.08396102185936266,
      "grad_norm": 1.8702586889266968,
      "learning_rate": 4.5808533052409795e-05,
      "loss": 1.8949,
      "step": 3188
    },
    {
      "epoch": 0.08398735844087438,
      "grad_norm": 2.8003616333007812,
      "learning_rate": 4.580721622333422e-05,
      "loss": 2.1091,
      "step": 3189
    },
    {
      "epoch": 0.08401369502238609,
      "grad_norm": 2.0695996284484863,
      "learning_rate": 4.5805899394258626e-05,
      "loss": 1.8386,
      "step": 3190
    },
    {
      "epoch": 0.08404003160389781,
      "grad_norm": 4.105753421783447,
      "learning_rate": 4.580458256518304e-05,
      "loss": 0.9941,
      "step": 3191
    },
    {
      "epoch": 0.08406636818540954,
      "grad_norm": 5.378888130187988,
      "learning_rate": 4.580326573610745e-05,
      "loss": 1.6269,
      "step": 3192
    },
    {
      "epoch": 0.08409270476692125,
      "grad_norm": 2.4382083415985107,
      "learning_rate": 4.580194890703187e-05,
      "loss": 2.5767,
      "step": 3193
    },
    {
      "epoch": 0.08411904134843297,
      "grad_norm": 5.3425116539001465,
      "learning_rate": 4.580063207795628e-05,
      "loss": 1.0474,
      "step": 3194
    },
    {
      "epoch": 0.0841453779299447,
      "grad_norm": 1.9401028156280518,
      "learning_rate": 4.57993152488807e-05,
      "loss": 2.0606,
      "step": 3195
    },
    {
      "epoch": 0.08417171451145641,
      "grad_norm": 3.2426631450653076,
      "learning_rate": 4.579799841980511e-05,
      "loss": 2.38,
      "step": 3196
    },
    {
      "epoch": 0.08419805109296813,
      "grad_norm": 1.6750125885009766,
      "learning_rate": 4.579668159072952e-05,
      "loss": 1.5457,
      "step": 3197
    },
    {
      "epoch": 0.08422438767447986,
      "grad_norm": 1.8950235843658447,
      "learning_rate": 4.5795364761653944e-05,
      "loss": 1.8299,
      "step": 3198
    },
    {
      "epoch": 0.08425072425599157,
      "grad_norm": 2.9908668994903564,
      "learning_rate": 4.579404793257835e-05,
      "loss": 1.1757,
      "step": 3199
    },
    {
      "epoch": 0.08427706083750329,
      "grad_norm": 2.055785655975342,
      "learning_rate": 4.579273110350277e-05,
      "loss": 2.0399,
      "step": 3200
    },
    {
      "epoch": 0.084303397419015,
      "grad_norm": 3.2788565158843994,
      "learning_rate": 4.579141427442718e-05,
      "loss": 2.2281,
      "step": 3201
    },
    {
      "epoch": 0.08432973400052673,
      "grad_norm": 2.5476937294006348,
      "learning_rate": 4.579009744535159e-05,
      "loss": 0.9109,
      "step": 3202
    },
    {
      "epoch": 0.08435607058203845,
      "grad_norm": 3.333198070526123,
      "learning_rate": 4.578878061627601e-05,
      "loss": 2.1412,
      "step": 3203
    },
    {
      "epoch": 0.08438240716355017,
      "grad_norm": 2.6660966873168945,
      "learning_rate": 4.5787463787200424e-05,
      "loss": 0.8025,
      "step": 3204
    },
    {
      "epoch": 0.0844087437450619,
      "grad_norm": 3.7357802391052246,
      "learning_rate": 4.578614695812484e-05,
      "loss": 0.9043,
      "step": 3205
    },
    {
      "epoch": 0.08443508032657361,
      "grad_norm": 2.3206193447113037,
      "learning_rate": 4.578483012904925e-05,
      "loss": 2.3118,
      "step": 3206
    },
    {
      "epoch": 0.08446141690808533,
      "grad_norm": 4.291245937347412,
      "learning_rate": 4.578351329997367e-05,
      "loss": 1.599,
      "step": 3207
    },
    {
      "epoch": 0.08448775348959706,
      "grad_norm": 2.4323413372039795,
      "learning_rate": 4.578219647089808e-05,
      "loss": 2.2438,
      "step": 3208
    },
    {
      "epoch": 0.08451409007110877,
      "grad_norm": 2.935488700866699,
      "learning_rate": 4.5780879641822495e-05,
      "loss": 1.6592,
      "step": 3209
    },
    {
      "epoch": 0.08454042665262049,
      "grad_norm": 2.0307133197784424,
      "learning_rate": 4.5779562812746904e-05,
      "loss": 2.2286,
      "step": 3210
    },
    {
      "epoch": 0.0845667632341322,
      "grad_norm": 2.061506748199463,
      "learning_rate": 4.577824598367132e-05,
      "loss": 1.5909,
      "step": 3211
    },
    {
      "epoch": 0.08459309981564393,
      "grad_norm": 2.495828151702881,
      "learning_rate": 4.5776929154595735e-05,
      "loss": 1.5772,
      "step": 3212
    },
    {
      "epoch": 0.08461943639715565,
      "grad_norm": 1.9573198556900024,
      "learning_rate": 4.577561232552015e-05,
      "loss": 0.6242,
      "step": 3213
    },
    {
      "epoch": 0.08464577297866736,
      "grad_norm": 1.838287353515625,
      "learning_rate": 4.5774295496444567e-05,
      "loss": 1.7578,
      "step": 3214
    },
    {
      "epoch": 0.08467210956017909,
      "grad_norm": 1.61006760597229,
      "learning_rate": 4.5772978667368975e-05,
      "loss": 1.8254,
      "step": 3215
    },
    {
      "epoch": 0.08469844614169081,
      "grad_norm": 1.8644553422927856,
      "learning_rate": 4.577166183829339e-05,
      "loss": 1.7631,
      "step": 3216
    },
    {
      "epoch": 0.08472478272320252,
      "grad_norm": 2.303036689758301,
      "learning_rate": 4.5770345009217807e-05,
      "loss": 2.7301,
      "step": 3217
    },
    {
      "epoch": 0.08475111930471425,
      "grad_norm": 2.0750932693481445,
      "learning_rate": 4.576902818014222e-05,
      "loss": 0.6355,
      "step": 3218
    },
    {
      "epoch": 0.08477745588622597,
      "grad_norm": 3.8102564811706543,
      "learning_rate": 4.576771135106663e-05,
      "loss": 1.4302,
      "step": 3219
    },
    {
      "epoch": 0.08480379246773768,
      "grad_norm": 2.098309278488159,
      "learning_rate": 4.5766394521991047e-05,
      "loss": 1.5723,
      "step": 3220
    },
    {
      "epoch": 0.08483012904924941,
      "grad_norm": 2.583853244781494,
      "learning_rate": 4.576507769291546e-05,
      "loss": 1.5143,
      "step": 3221
    },
    {
      "epoch": 0.08485646563076113,
      "grad_norm": 2.2531728744506836,
      "learning_rate": 4.576376086383988e-05,
      "loss": 2.1432,
      "step": 3222
    },
    {
      "epoch": 0.08488280221227285,
      "grad_norm": 2.871525764465332,
      "learning_rate": 4.576244403476429e-05,
      "loss": 1.8594,
      "step": 3223
    },
    {
      "epoch": 0.08490913879378456,
      "grad_norm": 2.0095560550689697,
      "learning_rate": 4.57611272056887e-05,
      "loss": 1.5931,
      "step": 3224
    },
    {
      "epoch": 0.08493547537529629,
      "grad_norm": 3.202383279800415,
      "learning_rate": 4.575981037661312e-05,
      "loss": 1.0461,
      "step": 3225
    },
    {
      "epoch": 0.084961811956808,
      "grad_norm": 2.3500051498413086,
      "learning_rate": 4.575849354753753e-05,
      "loss": 1.7214,
      "step": 3226
    },
    {
      "epoch": 0.08498814853831972,
      "grad_norm": 3.089770555496216,
      "learning_rate": 4.575717671846195e-05,
      "loss": 0.8427,
      "step": 3227
    },
    {
      "epoch": 0.08501448511983145,
      "grad_norm": 2.141021490097046,
      "learning_rate": 4.575585988938636e-05,
      "loss": 2.2557,
      "step": 3228
    },
    {
      "epoch": 0.08504082170134317,
      "grad_norm": 3.637605667114258,
      "learning_rate": 4.575454306031077e-05,
      "loss": 1.9752,
      "step": 3229
    },
    {
      "epoch": 0.08506715828285488,
      "grad_norm": 2.2627432346343994,
      "learning_rate": 4.575322623123519e-05,
      "loss": 2.217,
      "step": 3230
    },
    {
      "epoch": 0.08509349486436661,
      "grad_norm": 3.7502896785736084,
      "learning_rate": 4.5751909402159605e-05,
      "loss": 1.3013,
      "step": 3231
    },
    {
      "epoch": 0.08511983144587833,
      "grad_norm": 4.817307949066162,
      "learning_rate": 4.575059257308402e-05,
      "loss": 1.3074,
      "step": 3232
    },
    {
      "epoch": 0.08514616802739004,
      "grad_norm": 2.2230911254882812,
      "learning_rate": 4.574927574400843e-05,
      "loss": 0.486,
      "step": 3233
    },
    {
      "epoch": 0.08517250460890176,
      "grad_norm": 2.1989874839782715,
      "learning_rate": 4.5747958914932845e-05,
      "loss": 1.7278,
      "step": 3234
    },
    {
      "epoch": 0.08519884119041349,
      "grad_norm": 2.4527993202209473,
      "learning_rate": 4.5746642085857253e-05,
      "loss": 2.0587,
      "step": 3235
    },
    {
      "epoch": 0.0852251777719252,
      "grad_norm": 4.338960647583008,
      "learning_rate": 4.5745325256781676e-05,
      "loss": 1.4725,
      "step": 3236
    },
    {
      "epoch": 0.08525151435343692,
      "grad_norm": 2.488229751586914,
      "learning_rate": 4.5744008427706085e-05,
      "loss": 1.6375,
      "step": 3237
    },
    {
      "epoch": 0.08527785093494865,
      "grad_norm": 3.5188822746276855,
      "learning_rate": 4.57426915986305e-05,
      "loss": 1.622,
      "step": 3238
    },
    {
      "epoch": 0.08530418751646036,
      "grad_norm": 2.0137503147125244,
      "learning_rate": 4.574137476955491e-05,
      "loss": 1.6166,
      "step": 3239
    },
    {
      "epoch": 0.08533052409797208,
      "grad_norm": 4.743828296661377,
      "learning_rate": 4.574005794047933e-05,
      "loss": 1.155,
      "step": 3240
    },
    {
      "epoch": 0.08535686067948381,
      "grad_norm": 4.959859371185303,
      "learning_rate": 4.573874111140374e-05,
      "loss": 0.4543,
      "step": 3241
    },
    {
      "epoch": 0.08538319726099552,
      "grad_norm": 2.6431691646575928,
      "learning_rate": 4.5737424282328156e-05,
      "loss": 2.4788,
      "step": 3242
    },
    {
      "epoch": 0.08540953384250724,
      "grad_norm": 2.981017827987671,
      "learning_rate": 4.573610745325257e-05,
      "loss": 2.1791,
      "step": 3243
    },
    {
      "epoch": 0.08543587042401896,
      "grad_norm": 2.1606719493865967,
      "learning_rate": 4.573479062417698e-05,
      "loss": 2.3109,
      "step": 3244
    },
    {
      "epoch": 0.08546220700553069,
      "grad_norm": 2.191272258758545,
      "learning_rate": 4.57334737951014e-05,
      "loss": 1.4289,
      "step": 3245
    },
    {
      "epoch": 0.0854885435870424,
      "grad_norm": 1.7807561159133911,
      "learning_rate": 4.573215696602581e-05,
      "loss": 2.6139,
      "step": 3246
    },
    {
      "epoch": 0.08551488016855412,
      "grad_norm": 2.424823760986328,
      "learning_rate": 4.573084013695023e-05,
      "loss": 1.6689,
      "step": 3247
    },
    {
      "epoch": 0.08554121675006585,
      "grad_norm": 1.7393673658370972,
      "learning_rate": 4.5729523307874636e-05,
      "loss": 2.0159,
      "step": 3248
    },
    {
      "epoch": 0.08556755333157756,
      "grad_norm": 1.9312829971313477,
      "learning_rate": 4.572820647879905e-05,
      "loss": 1.7716,
      "step": 3249
    },
    {
      "epoch": 0.08559388991308928,
      "grad_norm": 2.718719482421875,
      "learning_rate": 4.572688964972347e-05,
      "loss": 1.7089,
      "step": 3250
    },
    {
      "epoch": 0.085620226494601,
      "grad_norm": 2.220393419265747,
      "learning_rate": 4.572557282064788e-05,
      "loss": 1.9193,
      "step": 3251
    },
    {
      "epoch": 0.08564656307611272,
      "grad_norm": 2.8280107975006104,
      "learning_rate": 4.57242559915723e-05,
      "loss": 1.7072,
      "step": 3252
    },
    {
      "epoch": 0.08567289965762444,
      "grad_norm": 2.6569924354553223,
      "learning_rate": 4.572293916249671e-05,
      "loss": 2.2634,
      "step": 3253
    },
    {
      "epoch": 0.08569923623913615,
      "grad_norm": 3.409811019897461,
      "learning_rate": 4.572162233342113e-05,
      "loss": 1.4841,
      "step": 3254
    },
    {
      "epoch": 0.08572557282064788,
      "grad_norm": 2.3559229373931885,
      "learning_rate": 4.572030550434554e-05,
      "loss": 0.4952,
      "step": 3255
    },
    {
      "epoch": 0.0857519094021596,
      "grad_norm": 3.11930775642395,
      "learning_rate": 4.5718988675269954e-05,
      "loss": 1.9097,
      "step": 3256
    },
    {
      "epoch": 0.08577824598367131,
      "grad_norm": 3.1896774768829346,
      "learning_rate": 4.571767184619436e-05,
      "loss": 1.6084,
      "step": 3257
    },
    {
      "epoch": 0.08580458256518304,
      "grad_norm": 4.589871406555176,
      "learning_rate": 4.571635501711878e-05,
      "loss": 1.4056,
      "step": 3258
    },
    {
      "epoch": 0.08583091914669476,
      "grad_norm": 4.741261005401611,
      "learning_rate": 4.5715038188043194e-05,
      "loss": 0.9644,
      "step": 3259
    },
    {
      "epoch": 0.08585725572820647,
      "grad_norm": 5.810362815856934,
      "learning_rate": 4.571372135896761e-05,
      "loss": 1.5796,
      "step": 3260
    },
    {
      "epoch": 0.0858835923097182,
      "grad_norm": 1.920261025428772,
      "learning_rate": 4.5712404529892025e-05,
      "loss": 2.0457,
      "step": 3261
    },
    {
      "epoch": 0.08590992889122992,
      "grad_norm": 2.4054150581359863,
      "learning_rate": 4.5711087700816434e-05,
      "loss": 2.0352,
      "step": 3262
    },
    {
      "epoch": 0.08593626547274164,
      "grad_norm": 2.4277303218841553,
      "learning_rate": 4.570977087174085e-05,
      "loss": 2.0703,
      "step": 3263
    },
    {
      "epoch": 0.08596260205425336,
      "grad_norm": 3.1370632648468018,
      "learning_rate": 4.5708454042665265e-05,
      "loss": 1.2243,
      "step": 3264
    },
    {
      "epoch": 0.08598893863576508,
      "grad_norm": 2.9451024532318115,
      "learning_rate": 4.570713721358968e-05,
      "loss": 1.9078,
      "step": 3265
    },
    {
      "epoch": 0.0860152752172768,
      "grad_norm": 2.067363977432251,
      "learning_rate": 4.570582038451409e-05,
      "loss": 2.0494,
      "step": 3266
    },
    {
      "epoch": 0.08604161179878851,
      "grad_norm": 3.4219279289245605,
      "learning_rate": 4.5704503555438505e-05,
      "loss": 2.0001,
      "step": 3267
    },
    {
      "epoch": 0.08606794838030024,
      "grad_norm": 4.3392720222473145,
      "learning_rate": 4.570318672636292e-05,
      "loss": 2.0908,
      "step": 3268
    },
    {
      "epoch": 0.08609428496181196,
      "grad_norm": 1.5715715885162354,
      "learning_rate": 4.5701869897287336e-05,
      "loss": 1.7674,
      "step": 3269
    },
    {
      "epoch": 0.08612062154332367,
      "grad_norm": 6.030750751495361,
      "learning_rate": 4.570055306821175e-05,
      "loss": 1.3608,
      "step": 3270
    },
    {
      "epoch": 0.0861469581248354,
      "grad_norm": 1.8376785516738892,
      "learning_rate": 4.569923623913616e-05,
      "loss": 2.0341,
      "step": 3271
    },
    {
      "epoch": 0.08617329470634712,
      "grad_norm": 2.9195828437805176,
      "learning_rate": 4.5697919410060576e-05,
      "loss": 1.8199,
      "step": 3272
    },
    {
      "epoch": 0.08619963128785883,
      "grad_norm": 3.337942600250244,
      "learning_rate": 4.569660258098499e-05,
      "loss": 1.2327,
      "step": 3273
    },
    {
      "epoch": 0.08622596786937056,
      "grad_norm": 2.0117812156677246,
      "learning_rate": 4.569528575190941e-05,
      "loss": 1.5599,
      "step": 3274
    },
    {
      "epoch": 0.08625230445088228,
      "grad_norm": 2.2164361476898193,
      "learning_rate": 4.5693968922833816e-05,
      "loss": 1.401,
      "step": 3275
    },
    {
      "epoch": 0.086278641032394,
      "grad_norm": 2.879408836364746,
      "learning_rate": 4.569265209375823e-05,
      "loss": 0.7136,
      "step": 3276
    },
    {
      "epoch": 0.08630497761390571,
      "grad_norm": 2.3157923221588135,
      "learning_rate": 4.569133526468265e-05,
      "loss": 0.9512,
      "step": 3277
    },
    {
      "epoch": 0.08633131419541744,
      "grad_norm": 2.668488025665283,
      "learning_rate": 4.569001843560706e-05,
      "loss": 1.8315,
      "step": 3278
    },
    {
      "epoch": 0.08635765077692915,
      "grad_norm": 2.078740358352661,
      "learning_rate": 4.568870160653148e-05,
      "loss": 1.7008,
      "step": 3279
    },
    {
      "epoch": 0.08638398735844087,
      "grad_norm": 2.9175126552581787,
      "learning_rate": 4.568738477745589e-05,
      "loss": 1.534,
      "step": 3280
    },
    {
      "epoch": 0.0864103239399526,
      "grad_norm": 3.69405198097229,
      "learning_rate": 4.56860679483803e-05,
      "loss": 1.7667,
      "step": 3281
    },
    {
      "epoch": 0.08643666052146431,
      "grad_norm": 2.7728805541992188,
      "learning_rate": 4.568475111930471e-05,
      "loss": 1.6557,
      "step": 3282
    },
    {
      "epoch": 0.08646299710297603,
      "grad_norm": 1.509304404258728,
      "learning_rate": 4.5683434290229134e-05,
      "loss": 1.6786,
      "step": 3283
    },
    {
      "epoch": 0.08648933368448776,
      "grad_norm": 5.2924089431762695,
      "learning_rate": 4.568211746115354e-05,
      "loss": 2.3038,
      "step": 3284
    },
    {
      "epoch": 0.08651567026599948,
      "grad_norm": 2.9266762733459473,
      "learning_rate": 4.568080063207796e-05,
      "loss": 1.6544,
      "step": 3285
    },
    {
      "epoch": 0.08654200684751119,
      "grad_norm": 3.1236605644226074,
      "learning_rate": 4.567948380300237e-05,
      "loss": 1.7951,
      "step": 3286
    },
    {
      "epoch": 0.0865683434290229,
      "grad_norm": 2.3273816108703613,
      "learning_rate": 4.567816697392679e-05,
      "loss": 2.18,
      "step": 3287
    },
    {
      "epoch": 0.08659468001053464,
      "grad_norm": 1.8327178955078125,
      "learning_rate": 4.56768501448512e-05,
      "loss": 1.8965,
      "step": 3288
    },
    {
      "epoch": 0.08662101659204635,
      "grad_norm": 2.5005617141723633,
      "learning_rate": 4.5675533315775614e-05,
      "loss": 1.9199,
      "step": 3289
    },
    {
      "epoch": 0.08664735317355807,
      "grad_norm": 4.19956636428833,
      "learning_rate": 4.567421648670003e-05,
      "loss": 2.0134,
      "step": 3290
    },
    {
      "epoch": 0.0866736897550698,
      "grad_norm": 2.499419689178467,
      "learning_rate": 4.567289965762444e-05,
      "loss": 1.8685,
      "step": 3291
    },
    {
      "epoch": 0.08670002633658151,
      "grad_norm": 1.7010895013809204,
      "learning_rate": 4.567158282854886e-05,
      "loss": 1.8863,
      "step": 3292
    },
    {
      "epoch": 0.08672636291809323,
      "grad_norm": 2.4933762550354004,
      "learning_rate": 4.567026599947327e-05,
      "loss": 0.4979,
      "step": 3293
    },
    {
      "epoch": 0.08675269949960496,
      "grad_norm": 1.8119902610778809,
      "learning_rate": 4.5668949170397686e-05,
      "loss": 1.7239,
      "step": 3294
    },
    {
      "epoch": 0.08677903608111667,
      "grad_norm": 3.5810108184814453,
      "learning_rate": 4.5667632341322094e-05,
      "loss": 2.1185,
      "step": 3295
    },
    {
      "epoch": 0.08680537266262839,
      "grad_norm": 2.684758424758911,
      "learning_rate": 4.566631551224651e-05,
      "loss": 1.7439,
      "step": 3296
    },
    {
      "epoch": 0.0868317092441401,
      "grad_norm": 2.041480779647827,
      "learning_rate": 4.5664998683170926e-05,
      "loss": 2.1769,
      "step": 3297
    },
    {
      "epoch": 0.08685804582565183,
      "grad_norm": 2.4490764141082764,
      "learning_rate": 4.566368185409534e-05,
      "loss": 1.5937,
      "step": 3298
    },
    {
      "epoch": 0.08688438240716355,
      "grad_norm": 2.4468672275543213,
      "learning_rate": 4.566236502501976e-05,
      "loss": 1.6006,
      "step": 3299
    },
    {
      "epoch": 0.08691071898867526,
      "grad_norm": 2.040010929107666,
      "learning_rate": 4.5661048195944166e-05,
      "loss": 2.0957,
      "step": 3300
    },
    {
      "epoch": 0.086937055570187,
      "grad_norm": 2.194331169128418,
      "learning_rate": 4.565973136686858e-05,
      "loss": 1.9661,
      "step": 3301
    },
    {
      "epoch": 0.08696339215169871,
      "grad_norm": 1.8147886991500854,
      "learning_rate": 4.5658414537793e-05,
      "loss": 2.434,
      "step": 3302
    },
    {
      "epoch": 0.08698972873321043,
      "grad_norm": 3.349013090133667,
      "learning_rate": 4.565709770871741e-05,
      "loss": 1.229,
      "step": 3303
    },
    {
      "epoch": 0.08701606531472215,
      "grad_norm": 2.3615927696228027,
      "learning_rate": 4.565578087964182e-05,
      "loss": 1.8129,
      "step": 3304
    },
    {
      "epoch": 0.08704240189623387,
      "grad_norm": 2.8906731605529785,
      "learning_rate": 4.565446405056624e-05,
      "loss": 1.3058,
      "step": 3305
    },
    {
      "epoch": 0.08706873847774559,
      "grad_norm": 4.31870698928833,
      "learning_rate": 4.565314722149065e-05,
      "loss": 2.7197,
      "step": 3306
    },
    {
      "epoch": 0.08709507505925732,
      "grad_norm": 6.206526756286621,
      "learning_rate": 4.565183039241507e-05,
      "loss": 0.878,
      "step": 3307
    },
    {
      "epoch": 0.08712141164076903,
      "grad_norm": 5.137819766998291,
      "learning_rate": 4.5650513563339484e-05,
      "loss": 1.5677,
      "step": 3308
    },
    {
      "epoch": 0.08714774822228075,
      "grad_norm": 2.115330219268799,
      "learning_rate": 4.564919673426389e-05,
      "loss": 1.6648,
      "step": 3309
    },
    {
      "epoch": 0.08717408480379246,
      "grad_norm": 4.449472427368164,
      "learning_rate": 4.564787990518831e-05,
      "loss": 1.173,
      "step": 3310
    },
    {
      "epoch": 0.08720042138530419,
      "grad_norm": 3.948798418045044,
      "learning_rate": 4.5646563076112724e-05,
      "loss": 1.4963,
      "step": 3311
    },
    {
      "epoch": 0.08722675796681591,
      "grad_norm": 4.386968612670898,
      "learning_rate": 4.564524624703714e-05,
      "loss": 1.9632,
      "step": 3312
    },
    {
      "epoch": 0.08725309454832762,
      "grad_norm": 2.518096923828125,
      "learning_rate": 4.564392941796155e-05,
      "loss": 1.0785,
      "step": 3313
    },
    {
      "epoch": 0.08727943112983935,
      "grad_norm": 3.4689698219299316,
      "learning_rate": 4.5642612588885964e-05,
      "loss": 2.1707,
      "step": 3314
    },
    {
      "epoch": 0.08730576771135107,
      "grad_norm": 3.268017053604126,
      "learning_rate": 4.564129575981038e-05,
      "loss": 1.1932,
      "step": 3315
    },
    {
      "epoch": 0.08733210429286278,
      "grad_norm": 7.4629807472229,
      "learning_rate": 4.5639978930734795e-05,
      "loss": 2.1586,
      "step": 3316
    },
    {
      "epoch": 0.08735844087437451,
      "grad_norm": 2.441282272338867,
      "learning_rate": 4.563866210165921e-05,
      "loss": 0.4565,
      "step": 3317
    },
    {
      "epoch": 0.08738477745588623,
      "grad_norm": 4.256004810333252,
      "learning_rate": 4.563734527258362e-05,
      "loss": 1.4775,
      "step": 3318
    },
    {
      "epoch": 0.08741111403739794,
      "grad_norm": 1.818750023841858,
      "learning_rate": 4.5636028443508035e-05,
      "loss": 1.7899,
      "step": 3319
    },
    {
      "epoch": 0.08743745061890966,
      "grad_norm": 1.843833565711975,
      "learning_rate": 4.563471161443245e-05,
      "loss": 1.5933,
      "step": 3320
    },
    {
      "epoch": 0.08746378720042139,
      "grad_norm": 2.0348916053771973,
      "learning_rate": 4.5633394785356866e-05,
      "loss": 1.5735,
      "step": 3321
    },
    {
      "epoch": 0.0874901237819331,
      "grad_norm": 2.6325459480285645,
      "learning_rate": 4.5632077956281275e-05,
      "loss": 1.827,
      "step": 3322
    },
    {
      "epoch": 0.08751646036344482,
      "grad_norm": 2.4956932067871094,
      "learning_rate": 4.563076112720569e-05,
      "loss": 2.3305,
      "step": 3323
    },
    {
      "epoch": 0.08754279694495655,
      "grad_norm": 2.065242290496826,
      "learning_rate": 4.5629444298130106e-05,
      "loss": 1.725,
      "step": 3324
    },
    {
      "epoch": 0.08756913352646827,
      "grad_norm": 3.2543764114379883,
      "learning_rate": 4.562812746905452e-05,
      "loss": 0.6695,
      "step": 3325
    },
    {
      "epoch": 0.08759547010797998,
      "grad_norm": 2.7452926635742188,
      "learning_rate": 4.562681063997894e-05,
      "loss": 1.5987,
      "step": 3326
    },
    {
      "epoch": 0.08762180668949171,
      "grad_norm": 2.2238802909851074,
      "learning_rate": 4.5625493810903346e-05,
      "loss": 2.0862,
      "step": 3327
    },
    {
      "epoch": 0.08764814327100343,
      "grad_norm": 2.7854669094085693,
      "learning_rate": 4.562417698182776e-05,
      "loss": 2.3381,
      "step": 3328
    },
    {
      "epoch": 0.08767447985251514,
      "grad_norm": 1.9521734714508057,
      "learning_rate": 4.562286015275217e-05,
      "loss": 1.4309,
      "step": 3329
    },
    {
      "epoch": 0.08770081643402686,
      "grad_norm": 3.3219892978668213,
      "learning_rate": 4.562154332367659e-05,
      "loss": 2.1148,
      "step": 3330
    },
    {
      "epoch": 0.08772715301553859,
      "grad_norm": 2.3819589614868164,
      "learning_rate": 4.5620226494601e-05,
      "loss": 1.559,
      "step": 3331
    },
    {
      "epoch": 0.0877534895970503,
      "grad_norm": 3.2879862785339355,
      "learning_rate": 4.561890966552542e-05,
      "loss": 1.5046,
      "step": 3332
    },
    {
      "epoch": 0.08777982617856202,
      "grad_norm": 3.7280631065368652,
      "learning_rate": 4.561759283644983e-05,
      "loss": 1.404,
      "step": 3333
    },
    {
      "epoch": 0.08780616276007375,
      "grad_norm": 1.9950778484344482,
      "learning_rate": 4.561627600737424e-05,
      "loss": 0.3617,
      "step": 3334
    },
    {
      "epoch": 0.08783249934158546,
      "grad_norm": 1.8072651624679565,
      "learning_rate": 4.5614959178298664e-05,
      "loss": 1.5139,
      "step": 3335
    },
    {
      "epoch": 0.08785883592309718,
      "grad_norm": 1.9130542278289795,
      "learning_rate": 4.561364234922307e-05,
      "loss": 2.3877,
      "step": 3336
    },
    {
      "epoch": 0.08788517250460891,
      "grad_norm": 2.688465118408203,
      "learning_rate": 4.561232552014749e-05,
      "loss": 1.4498,
      "step": 3337
    },
    {
      "epoch": 0.08791150908612062,
      "grad_norm": 1.6436034440994263,
      "learning_rate": 4.56110086910719e-05,
      "loss": 1.5885,
      "step": 3338
    },
    {
      "epoch": 0.08793784566763234,
      "grad_norm": 2.0929412841796875,
      "learning_rate": 4.560969186199632e-05,
      "loss": 1.8179,
      "step": 3339
    },
    {
      "epoch": 0.08796418224914405,
      "grad_norm": 5.781081199645996,
      "learning_rate": 4.560837503292073e-05,
      "loss": 1.0893,
      "step": 3340
    },
    {
      "epoch": 0.08799051883065578,
      "grad_norm": 2.7853565216064453,
      "learning_rate": 4.5607058203845144e-05,
      "loss": 1.7618,
      "step": 3341
    },
    {
      "epoch": 0.0880168554121675,
      "grad_norm": 1.579538106918335,
      "learning_rate": 4.560574137476955e-05,
      "loss": 1.7477,
      "step": 3342
    },
    {
      "epoch": 0.08804319199367922,
      "grad_norm": 3.7131972312927246,
      "learning_rate": 4.560442454569397e-05,
      "loss": 1.6057,
      "step": 3343
    },
    {
      "epoch": 0.08806952857519094,
      "grad_norm": 3.0380332469940186,
      "learning_rate": 4.5603107716618384e-05,
      "loss": 1.5221,
      "step": 3344
    },
    {
      "epoch": 0.08809586515670266,
      "grad_norm": 1.8556195497512817,
      "learning_rate": 4.56017908875428e-05,
      "loss": 1.7945,
      "step": 3345
    },
    {
      "epoch": 0.08812220173821438,
      "grad_norm": 2.110743522644043,
      "learning_rate": 4.5600474058467215e-05,
      "loss": 1.7812,
      "step": 3346
    },
    {
      "epoch": 0.0881485383197261,
      "grad_norm": 1.9272798299789429,
      "learning_rate": 4.5599157229391624e-05,
      "loss": 1.2776,
      "step": 3347
    },
    {
      "epoch": 0.08817487490123782,
      "grad_norm": 1.881285309791565,
      "learning_rate": 4.559784040031604e-05,
      "loss": 2.1988,
      "step": 3348
    },
    {
      "epoch": 0.08820121148274954,
      "grad_norm": 2.1280012130737305,
      "learning_rate": 4.5596523571240455e-05,
      "loss": 1.8454,
      "step": 3349
    },
    {
      "epoch": 0.08822754806426125,
      "grad_norm": 2.944765329360962,
      "learning_rate": 4.559520674216487e-05,
      "loss": 0.4876,
      "step": 3350
    },
    {
      "epoch": 0.08825388464577298,
      "grad_norm": 2.803616523742676,
      "learning_rate": 4.559388991308928e-05,
      "loss": 1.9499,
      "step": 3351
    },
    {
      "epoch": 0.0882802212272847,
      "grad_norm": 2.4479265213012695,
      "learning_rate": 4.5592573084013695e-05,
      "loss": 2.2732,
      "step": 3352
    },
    {
      "epoch": 0.08830655780879641,
      "grad_norm": 3.8088347911834717,
      "learning_rate": 4.559125625493811e-05,
      "loss": 1.345,
      "step": 3353
    },
    {
      "epoch": 0.08833289439030814,
      "grad_norm": 3.1336209774017334,
      "learning_rate": 4.5589939425862527e-05,
      "loss": 1.9881,
      "step": 3354
    },
    {
      "epoch": 0.08835923097181986,
      "grad_norm": 3.227994441986084,
      "learning_rate": 4.558862259678694e-05,
      "loss": 1.7598,
      "step": 3355
    },
    {
      "epoch": 0.08838556755333157,
      "grad_norm": 3.023447275161743,
      "learning_rate": 4.558730576771135e-05,
      "loss": 1.3893,
      "step": 3356
    },
    {
      "epoch": 0.0884119041348433,
      "grad_norm": 2.735240936279297,
      "learning_rate": 4.5585988938635767e-05,
      "loss": 0.5625,
      "step": 3357
    },
    {
      "epoch": 0.08843824071635502,
      "grad_norm": 2.5172595977783203,
      "learning_rate": 4.558467210956018e-05,
      "loss": 1.4498,
      "step": 3358
    },
    {
      "epoch": 0.08846457729786673,
      "grad_norm": 1.993535041809082,
      "learning_rate": 4.55833552804846e-05,
      "loss": 1.9375,
      "step": 3359
    },
    {
      "epoch": 0.08849091387937846,
      "grad_norm": 2.281315803527832,
      "learning_rate": 4.5582038451409007e-05,
      "loss": 0.3755,
      "step": 3360
    },
    {
      "epoch": 0.08851725046089018,
      "grad_norm": 3.085216999053955,
      "learning_rate": 4.558072162233342e-05,
      "loss": 1.9078,
      "step": 3361
    },
    {
      "epoch": 0.0885435870424019,
      "grad_norm": 2.7327311038970947,
      "learning_rate": 4.557940479325784e-05,
      "loss": 1.5368,
      "step": 3362
    },
    {
      "epoch": 0.08856992362391361,
      "grad_norm": 2.1845152378082275,
      "learning_rate": 4.5578087964182253e-05,
      "loss": 1.6946,
      "step": 3363
    },
    {
      "epoch": 0.08859626020542534,
      "grad_norm": 2.8556973934173584,
      "learning_rate": 4.557677113510667e-05,
      "loss": 2.3051,
      "step": 3364
    },
    {
      "epoch": 0.08862259678693706,
      "grad_norm": 2.062838554382324,
      "learning_rate": 4.557545430603108e-05,
      "loss": 1.8454,
      "step": 3365
    },
    {
      "epoch": 0.08864893336844877,
      "grad_norm": 2.3930327892303467,
      "learning_rate": 4.5574137476955493e-05,
      "loss": 2.152,
      "step": 3366
    },
    {
      "epoch": 0.0886752699499605,
      "grad_norm": 2.4107680320739746,
      "learning_rate": 4.55728206478799e-05,
      "loss": 2.0974,
      "step": 3367
    },
    {
      "epoch": 0.08870160653147222,
      "grad_norm": 1.9558051824569702,
      "learning_rate": 4.5571503818804325e-05,
      "loss": 1.4819,
      "step": 3368
    },
    {
      "epoch": 0.08872794311298393,
      "grad_norm": 2.191049575805664,
      "learning_rate": 4.5570186989728733e-05,
      "loss": 1.6372,
      "step": 3369
    },
    {
      "epoch": 0.08875427969449566,
      "grad_norm": 2.010561466217041,
      "learning_rate": 4.556887016065315e-05,
      "loss": 1.9864,
      "step": 3370
    },
    {
      "epoch": 0.08878061627600738,
      "grad_norm": 2.129725217819214,
      "learning_rate": 4.5567553331577565e-05,
      "loss": 1.8083,
      "step": 3371
    },
    {
      "epoch": 0.08880695285751909,
      "grad_norm": 1.9936469793319702,
      "learning_rate": 4.556623650250198e-05,
      "loss": 1.8133,
      "step": 3372
    },
    {
      "epoch": 0.08883328943903081,
      "grad_norm": 3.681715965270996,
      "learning_rate": 4.5564919673426396e-05,
      "loss": 1.5447,
      "step": 3373
    },
    {
      "epoch": 0.08885962602054254,
      "grad_norm": 2.613180637359619,
      "learning_rate": 4.5563602844350805e-05,
      "loss": 1.6233,
      "step": 3374
    },
    {
      "epoch": 0.08888596260205425,
      "grad_norm": 2.479228973388672,
      "learning_rate": 4.556228601527522e-05,
      "loss": 1.3793,
      "step": 3375
    },
    {
      "epoch": 0.08891229918356597,
      "grad_norm": 1.781434416770935,
      "learning_rate": 4.556096918619963e-05,
      "loss": 1.7175,
      "step": 3376
    },
    {
      "epoch": 0.0889386357650777,
      "grad_norm": 2.4801998138427734,
      "learning_rate": 4.555965235712405e-05,
      "loss": 1.5128,
      "step": 3377
    },
    {
      "epoch": 0.08896497234658941,
      "grad_norm": 2.037580728530884,
      "learning_rate": 4.555833552804846e-05,
      "loss": 1.0426,
      "step": 3378
    },
    {
      "epoch": 0.08899130892810113,
      "grad_norm": 1.9650875329971313,
      "learning_rate": 4.5557018698972876e-05,
      "loss": 1.9092,
      "step": 3379
    },
    {
      "epoch": 0.08901764550961286,
      "grad_norm": 2.6709585189819336,
      "learning_rate": 4.555570186989729e-05,
      "loss": 2.0747,
      "step": 3380
    },
    {
      "epoch": 0.08904398209112457,
      "grad_norm": 1.9449385404586792,
      "learning_rate": 4.55543850408217e-05,
      "loss": 1.8836,
      "step": 3381
    },
    {
      "epoch": 0.08907031867263629,
      "grad_norm": 2.2440335750579834,
      "learning_rate": 4.555306821174612e-05,
      "loss": 1.2759,
      "step": 3382
    },
    {
      "epoch": 0.089096655254148,
      "grad_norm": 3.023714542388916,
      "learning_rate": 4.555175138267053e-05,
      "loss": 0.8821,
      "step": 3383
    },
    {
      "epoch": 0.08912299183565973,
      "grad_norm": 3.3295540809631348,
      "learning_rate": 4.555043455359495e-05,
      "loss": 1.4563,
      "step": 3384
    },
    {
      "epoch": 0.08914932841717145,
      "grad_norm": 3.4079713821411133,
      "learning_rate": 4.5549117724519356e-05,
      "loss": 1.2193,
      "step": 3385
    },
    {
      "epoch": 0.08917566499868317,
      "grad_norm": 1.8142025470733643,
      "learning_rate": 4.554780089544378e-05,
      "loss": 1.2313,
      "step": 3386
    },
    {
      "epoch": 0.0892020015801949,
      "grad_norm": 3.524857521057129,
      "learning_rate": 4.554648406636819e-05,
      "loss": 1.9658,
      "step": 3387
    },
    {
      "epoch": 0.08922833816170661,
      "grad_norm": 2.367950439453125,
      "learning_rate": 4.55451672372926e-05,
      "loss": 1.8446,
      "step": 3388
    },
    {
      "epoch": 0.08925467474321833,
      "grad_norm": 2.3109920024871826,
      "learning_rate": 4.554385040821701e-05,
      "loss": 1.0784,
      "step": 3389
    },
    {
      "epoch": 0.08928101132473006,
      "grad_norm": 2.0897305011749268,
      "learning_rate": 4.554253357914143e-05,
      "loss": 1.6419,
      "step": 3390
    },
    {
      "epoch": 0.08930734790624177,
      "grad_norm": 1.8517029285430908,
      "learning_rate": 4.554121675006584e-05,
      "loss": 2.2631,
      "step": 3391
    },
    {
      "epoch": 0.08933368448775349,
      "grad_norm": 2.5218851566314697,
      "learning_rate": 4.553989992099026e-05,
      "loss": 1.8359,
      "step": 3392
    },
    {
      "epoch": 0.0893600210692652,
      "grad_norm": 2.8504812717437744,
      "learning_rate": 4.5538583091914674e-05,
      "loss": 1.9991,
      "step": 3393
    },
    {
      "epoch": 0.08938635765077693,
      "grad_norm": 3.4940335750579834,
      "learning_rate": 4.553726626283908e-05,
      "loss": 1.4142,
      "step": 3394
    },
    {
      "epoch": 0.08941269423228865,
      "grad_norm": 3.1050009727478027,
      "learning_rate": 4.55359494337635e-05,
      "loss": 0.478,
      "step": 3395
    },
    {
      "epoch": 0.08943903081380036,
      "grad_norm": 2.702674388885498,
      "learning_rate": 4.5534632604687914e-05,
      "loss": 1.977,
      "step": 3396
    },
    {
      "epoch": 0.08946536739531209,
      "grad_norm": 3.2361903190612793,
      "learning_rate": 4.553331577561233e-05,
      "loss": 1.8961,
      "step": 3397
    },
    {
      "epoch": 0.08949170397682381,
      "grad_norm": 3.6690638065338135,
      "learning_rate": 4.553199894653674e-05,
      "loss": 2.0012,
      "step": 3398
    },
    {
      "epoch": 0.08951804055833552,
      "grad_norm": 2.043118476867676,
      "learning_rate": 4.5530682117461154e-05,
      "loss": 1.9398,
      "step": 3399
    },
    {
      "epoch": 0.08954437713984725,
      "grad_norm": 2.2660138607025146,
      "learning_rate": 4.552936528838557e-05,
      "loss": 2.2989,
      "step": 3400
    },
    {
      "epoch": 0.08957071372135897,
      "grad_norm": 2.0987002849578857,
      "learning_rate": 4.5528048459309985e-05,
      "loss": 1.5779,
      "step": 3401
    },
    {
      "epoch": 0.08959705030287068,
      "grad_norm": 2.952312707901001,
      "learning_rate": 4.55267316302344e-05,
      "loss": 1.9738,
      "step": 3402
    },
    {
      "epoch": 0.08962338688438241,
      "grad_norm": 2.3099260330200195,
      "learning_rate": 4.552541480115881e-05,
      "loss": 1.7237,
      "step": 3403
    },
    {
      "epoch": 0.08964972346589413,
      "grad_norm": 5.9042134284973145,
      "learning_rate": 4.5524097972083225e-05,
      "loss": 1.4515,
      "step": 3404
    },
    {
      "epoch": 0.08967606004740585,
      "grad_norm": 1.9012696743011475,
      "learning_rate": 4.552278114300764e-05,
      "loss": 1.9434,
      "step": 3405
    },
    {
      "epoch": 0.08970239662891756,
      "grad_norm": 2.4749765396118164,
      "learning_rate": 4.5521464313932056e-05,
      "loss": 1.6463,
      "step": 3406
    },
    {
      "epoch": 0.08972873321042929,
      "grad_norm": 1.9249846935272217,
      "learning_rate": 4.5520147484856465e-05,
      "loss": 2.0145,
      "step": 3407
    },
    {
      "epoch": 0.089755069791941,
      "grad_norm": 2.121868371963501,
      "learning_rate": 4.551883065578088e-05,
      "loss": 0.6021,
      "step": 3408
    },
    {
      "epoch": 0.08978140637345272,
      "grad_norm": 2.3962197303771973,
      "learning_rate": 4.5517513826705296e-05,
      "loss": 2.4078,
      "step": 3409
    },
    {
      "epoch": 0.08980774295496445,
      "grad_norm": 2.1263630390167236,
      "learning_rate": 4.551619699762971e-05,
      "loss": 1.9646,
      "step": 3410
    },
    {
      "epoch": 0.08983407953647617,
      "grad_norm": 3.167188882827759,
      "learning_rate": 4.551488016855413e-05,
      "loss": 1.3661,
      "step": 3411
    },
    {
      "epoch": 0.08986041611798788,
      "grad_norm": 3.5980520248413086,
      "learning_rate": 4.5513563339478536e-05,
      "loss": 1.7937,
      "step": 3412
    },
    {
      "epoch": 0.08988675269949961,
      "grad_norm": 1.9002790451049805,
      "learning_rate": 4.551224651040295e-05,
      "loss": 1.268,
      "step": 3413
    },
    {
      "epoch": 0.08991308928101133,
      "grad_norm": 1.7264384031295776,
      "learning_rate": 4.551092968132736e-05,
      "loss": 2.1844,
      "step": 3414
    },
    {
      "epoch": 0.08993942586252304,
      "grad_norm": 1.6253485679626465,
      "learning_rate": 4.550961285225178e-05,
      "loss": 1.6023,
      "step": 3415
    },
    {
      "epoch": 0.08996576244403476,
      "grad_norm": 2.992448091506958,
      "learning_rate": 4.550829602317619e-05,
      "loss": 2.0443,
      "step": 3416
    },
    {
      "epoch": 0.08999209902554649,
      "grad_norm": 2.267564058303833,
      "learning_rate": 4.550697919410061e-05,
      "loss": 0.4088,
      "step": 3417
    },
    {
      "epoch": 0.0900184356070582,
      "grad_norm": 2.0954949855804443,
      "learning_rate": 4.550566236502502e-05,
      "loss": 1.5922,
      "step": 3418
    },
    {
      "epoch": 0.09004477218856992,
      "grad_norm": 2.935631275177002,
      "learning_rate": 4.550434553594944e-05,
      "loss": 2.4555,
      "step": 3419
    },
    {
      "epoch": 0.09007110877008165,
      "grad_norm": 3.6485068798065186,
      "learning_rate": 4.5503028706873854e-05,
      "loss": 1.3428,
      "step": 3420
    },
    {
      "epoch": 0.09009744535159336,
      "grad_norm": 2.8711252212524414,
      "learning_rate": 4.550171187779826e-05,
      "loss": 2.4802,
      "step": 3421
    },
    {
      "epoch": 0.09012378193310508,
      "grad_norm": 2.427990674972534,
      "learning_rate": 4.550039504872268e-05,
      "loss": 1.9352,
      "step": 3422
    },
    {
      "epoch": 0.09015011851461681,
      "grad_norm": 3.124286413192749,
      "learning_rate": 4.549907821964709e-05,
      "loss": 1.6265,
      "step": 3423
    },
    {
      "epoch": 0.09017645509612852,
      "grad_norm": 3.1010942459106445,
      "learning_rate": 4.549776139057151e-05,
      "loss": 2.1104,
      "step": 3424
    },
    {
      "epoch": 0.09020279167764024,
      "grad_norm": 3.4730076789855957,
      "learning_rate": 4.549644456149592e-05,
      "loss": 1.4576,
      "step": 3425
    },
    {
      "epoch": 0.09022912825915196,
      "grad_norm": 3.8209004402160645,
      "learning_rate": 4.5495127732420334e-05,
      "loss": 1.1925,
      "step": 3426
    },
    {
      "epoch": 0.09025546484066369,
      "grad_norm": 3.584869384765625,
      "learning_rate": 4.549381090334475e-05,
      "loss": 1.4934,
      "step": 3427
    },
    {
      "epoch": 0.0902818014221754,
      "grad_norm": 1.8450781106948853,
      "learning_rate": 4.549249407426916e-05,
      "loss": 1.7063,
      "step": 3428
    },
    {
      "epoch": 0.09030813800368712,
      "grad_norm": 2.419992685317993,
      "learning_rate": 4.549117724519358e-05,
      "loss": 1.5129,
      "step": 3429
    },
    {
      "epoch": 0.09033447458519885,
      "grad_norm": 2.637809991836548,
      "learning_rate": 4.548986041611799e-05,
      "loss": 1.4932,
      "step": 3430
    },
    {
      "epoch": 0.09036081116671056,
      "grad_norm": 6.464300632476807,
      "learning_rate": 4.5488543587042406e-05,
      "loss": 1.943,
      "step": 3431
    },
    {
      "epoch": 0.09038714774822228,
      "grad_norm": 2.107273578643799,
      "learning_rate": 4.5487226757966814e-05,
      "loss": 2.2235,
      "step": 3432
    },
    {
      "epoch": 0.090413484329734,
      "grad_norm": 2.3448657989501953,
      "learning_rate": 4.548590992889123e-05,
      "loss": 1.7162,
      "step": 3433
    },
    {
      "epoch": 0.09043982091124572,
      "grad_norm": 2.627598762512207,
      "learning_rate": 4.5484593099815646e-05,
      "loss": 1.4002,
      "step": 3434
    },
    {
      "epoch": 0.09046615749275744,
      "grad_norm": 2.0682199001312256,
      "learning_rate": 4.548327627074006e-05,
      "loss": 1.5173,
      "step": 3435
    },
    {
      "epoch": 0.09049249407426915,
      "grad_norm": 3.4184250831604004,
      "learning_rate": 4.548195944166447e-05,
      "loss": 1.856,
      "step": 3436
    },
    {
      "epoch": 0.09051883065578088,
      "grad_norm": 4.626522541046143,
      "learning_rate": 4.5480642612588886e-05,
      "loss": 1.2222,
      "step": 3437
    },
    {
      "epoch": 0.0905451672372926,
      "grad_norm": 2.2340171337127686,
      "learning_rate": 4.547932578351331e-05,
      "loss": 2.0774,
      "step": 3438
    },
    {
      "epoch": 0.09057150381880431,
      "grad_norm": 4.535863399505615,
      "learning_rate": 4.547800895443772e-05,
      "loss": 2.6402,
      "step": 3439
    },
    {
      "epoch": 0.09059784040031604,
      "grad_norm": 1.9701533317565918,
      "learning_rate": 4.547669212536213e-05,
      "loss": 2.5163,
      "step": 3440
    },
    {
      "epoch": 0.09062417698182776,
      "grad_norm": 1.865283727645874,
      "learning_rate": 4.547537529628654e-05,
      "loss": 1.7879,
      "step": 3441
    },
    {
      "epoch": 0.09065051356333947,
      "grad_norm": 1.8001735210418701,
      "learning_rate": 4.547405846721096e-05,
      "loss": 1.6645,
      "step": 3442
    },
    {
      "epoch": 0.0906768501448512,
      "grad_norm": 2.2745540142059326,
      "learning_rate": 4.547274163813537e-05,
      "loss": 1.4718,
      "step": 3443
    },
    {
      "epoch": 0.09070318672636292,
      "grad_norm": 5.53325891494751,
      "learning_rate": 4.547142480905979e-05,
      "loss": 1.6999,
      "step": 3444
    },
    {
      "epoch": 0.09072952330787463,
      "grad_norm": 2.608701467514038,
      "learning_rate": 4.54701079799842e-05,
      "loss": 1.7494,
      "step": 3445
    },
    {
      "epoch": 0.09075585988938636,
      "grad_norm": 2.4674465656280518,
      "learning_rate": 4.546879115090861e-05,
      "loss": 1.474,
      "step": 3446
    },
    {
      "epoch": 0.09078219647089808,
      "grad_norm": 1.9420452117919922,
      "learning_rate": 4.546747432183303e-05,
      "loss": 1.978,
      "step": 3447
    },
    {
      "epoch": 0.0908085330524098,
      "grad_norm": 3.726248264312744,
      "learning_rate": 4.5466157492757444e-05,
      "loss": 0.7192,
      "step": 3448
    },
    {
      "epoch": 0.09083486963392151,
      "grad_norm": 3.000051259994507,
      "learning_rate": 4.546484066368186e-05,
      "loss": 1.6535,
      "step": 3449
    },
    {
      "epoch": 0.09086120621543324,
      "grad_norm": 2.382986068725586,
      "learning_rate": 4.546352383460627e-05,
      "loss": 2.1222,
      "step": 3450
    },
    {
      "epoch": 0.09088754279694496,
      "grad_norm": 4.2474775314331055,
      "learning_rate": 4.5462207005530684e-05,
      "loss": 1.8899,
      "step": 3451
    },
    {
      "epoch": 0.09091387937845667,
      "grad_norm": 4.1579670906066895,
      "learning_rate": 4.54608901764551e-05,
      "loss": 1.727,
      "step": 3452
    },
    {
      "epoch": 0.0909402159599684,
      "grad_norm": 1.8609178066253662,
      "learning_rate": 4.5459573347379515e-05,
      "loss": 1.9584,
      "step": 3453
    },
    {
      "epoch": 0.09096655254148012,
      "grad_norm": 2.892361879348755,
      "learning_rate": 4.5458256518303924e-05,
      "loss": 1.4133,
      "step": 3454
    },
    {
      "epoch": 0.09099288912299183,
      "grad_norm": 3.1268770694732666,
      "learning_rate": 4.545693968922834e-05,
      "loss": 1.2705,
      "step": 3455
    },
    {
      "epoch": 0.09101922570450356,
      "grad_norm": 2.5478532314300537,
      "learning_rate": 4.5455622860152755e-05,
      "loss": 2.2414,
      "step": 3456
    },
    {
      "epoch": 0.09104556228601528,
      "grad_norm": 3.5269534587860107,
      "learning_rate": 4.545430603107717e-05,
      "loss": 2.0966,
      "step": 3457
    },
    {
      "epoch": 0.09107189886752699,
      "grad_norm": 3.479748010635376,
      "learning_rate": 4.5452989202001586e-05,
      "loss": 1.0934,
      "step": 3458
    },
    {
      "epoch": 0.09109823544903871,
      "grad_norm": 2.448650598526001,
      "learning_rate": 4.5451672372925995e-05,
      "loss": 1.6114,
      "step": 3459
    },
    {
      "epoch": 0.09112457203055044,
      "grad_norm": 4.655614852905273,
      "learning_rate": 4.545035554385041e-05,
      "loss": 0.7682,
      "step": 3460
    },
    {
      "epoch": 0.09115090861206215,
      "grad_norm": 3.621953248977661,
      "learning_rate": 4.544903871477482e-05,
      "loss": 1.2884,
      "step": 3461
    },
    {
      "epoch": 0.09117724519357387,
      "grad_norm": 1.8357905149459839,
      "learning_rate": 4.544772188569924e-05,
      "loss": 1.9588,
      "step": 3462
    },
    {
      "epoch": 0.0912035817750856,
      "grad_norm": 2.350175619125366,
      "learning_rate": 4.544640505662365e-05,
      "loss": 1.5626,
      "step": 3463
    },
    {
      "epoch": 0.09122991835659731,
      "grad_norm": 2.3754656314849854,
      "learning_rate": 4.5445088227548066e-05,
      "loss": 1.5809,
      "step": 3464
    },
    {
      "epoch": 0.09125625493810903,
      "grad_norm": 3.776057243347168,
      "learning_rate": 4.544377139847248e-05,
      "loss": 1.0369,
      "step": 3465
    },
    {
      "epoch": 0.09128259151962076,
      "grad_norm": 2.334263324737549,
      "learning_rate": 4.544245456939689e-05,
      "loss": 1.731,
      "step": 3466
    },
    {
      "epoch": 0.09130892810113247,
      "grad_norm": 4.7469305992126465,
      "learning_rate": 4.544113774032131e-05,
      "loss": 1.9404,
      "step": 3467
    },
    {
      "epoch": 0.09133526468264419,
      "grad_norm": 3.5409865379333496,
      "learning_rate": 4.543982091124572e-05,
      "loss": 1.7411,
      "step": 3468
    },
    {
      "epoch": 0.0913616012641559,
      "grad_norm": 1.9411232471466064,
      "learning_rate": 4.543850408217014e-05,
      "loss": 1.692,
      "step": 3469
    },
    {
      "epoch": 0.09138793784566764,
      "grad_norm": 1.9997941255569458,
      "learning_rate": 4.5437187253094546e-05,
      "loss": 1.9027,
      "step": 3470
    },
    {
      "epoch": 0.09141427442717935,
      "grad_norm": 4.417593479156494,
      "learning_rate": 4.543587042401897e-05,
      "loss": 2.0089,
      "step": 3471
    },
    {
      "epoch": 0.09144061100869107,
      "grad_norm": 1.9682583808898926,
      "learning_rate": 4.543455359494338e-05,
      "loss": 1.6445,
      "step": 3472
    },
    {
      "epoch": 0.0914669475902028,
      "grad_norm": 2.2500627040863037,
      "learning_rate": 4.543323676586779e-05,
      "loss": 1.7452,
      "step": 3473
    },
    {
      "epoch": 0.09149328417171451,
      "grad_norm": 2.26936411857605,
      "learning_rate": 4.543191993679221e-05,
      "loss": 2.1707,
      "step": 3474
    },
    {
      "epoch": 0.09151962075322623,
      "grad_norm": 4.783372402191162,
      "learning_rate": 4.543060310771662e-05,
      "loss": 2.3268,
      "step": 3475
    },
    {
      "epoch": 0.09154595733473796,
      "grad_norm": 2.153541088104248,
      "learning_rate": 4.542928627864104e-05,
      "loss": 1.7916,
      "step": 3476
    },
    {
      "epoch": 0.09157229391624967,
      "grad_norm": 2.3099043369293213,
      "learning_rate": 4.542796944956545e-05,
      "loss": 2.0975,
      "step": 3477
    },
    {
      "epoch": 0.09159863049776139,
      "grad_norm": 3.0742597579956055,
      "learning_rate": 4.5426652620489864e-05,
      "loss": 1.5007,
      "step": 3478
    },
    {
      "epoch": 0.0916249670792731,
      "grad_norm": 4.193145275115967,
      "learning_rate": 4.542533579141427e-05,
      "loss": 1.5927,
      "step": 3479
    },
    {
      "epoch": 0.09165130366078483,
      "grad_norm": 2.1241064071655273,
      "learning_rate": 4.542401896233869e-05,
      "loss": 2.051,
      "step": 3480
    },
    {
      "epoch": 0.09167764024229655,
      "grad_norm": 2.8343396186828613,
      "learning_rate": 4.5422702133263104e-05,
      "loss": 2.055,
      "step": 3481
    },
    {
      "epoch": 0.09170397682380826,
      "grad_norm": 3.0514485836029053,
      "learning_rate": 4.542138530418752e-05,
      "loss": 2.1494,
      "step": 3482
    },
    {
      "epoch": 0.09173031340532,
      "grad_norm": 4.142845630645752,
      "learning_rate": 4.5420068475111935e-05,
      "loss": 1.5494,
      "step": 3483
    },
    {
      "epoch": 0.09175664998683171,
      "grad_norm": 5.5925774574279785,
      "learning_rate": 4.5418751646036344e-05,
      "loss": 1.7694,
      "step": 3484
    },
    {
      "epoch": 0.09178298656834342,
      "grad_norm": 2.1784701347351074,
      "learning_rate": 4.5417434816960767e-05,
      "loss": 1.7135,
      "step": 3485
    },
    {
      "epoch": 0.09180932314985515,
      "grad_norm": 4.068819999694824,
      "learning_rate": 4.5416117987885175e-05,
      "loss": 1.0339,
      "step": 3486
    },
    {
      "epoch": 0.09183565973136687,
      "grad_norm": 4.306507110595703,
      "learning_rate": 4.541480115880959e-05,
      "loss": 1.8897,
      "step": 3487
    },
    {
      "epoch": 0.09186199631287859,
      "grad_norm": 2.95487904548645,
      "learning_rate": 4.5413484329734e-05,
      "loss": 0.6175,
      "step": 3488
    },
    {
      "epoch": 0.09188833289439031,
      "grad_norm": 1.7955442667007446,
      "learning_rate": 4.5412167500658415e-05,
      "loss": 2.0957,
      "step": 3489
    },
    {
      "epoch": 0.09191466947590203,
      "grad_norm": 1.7751002311706543,
      "learning_rate": 4.541085067158283e-05,
      "loss": 2.3951,
      "step": 3490
    },
    {
      "epoch": 0.09194100605741375,
      "grad_norm": 3.020322799682617,
      "learning_rate": 4.5409533842507247e-05,
      "loss": 1.2985,
      "step": 3491
    },
    {
      "epoch": 0.09196734263892546,
      "grad_norm": 2.5244171619415283,
      "learning_rate": 4.5408217013431655e-05,
      "loss": 0.9907,
      "step": 3492
    },
    {
      "epoch": 0.09199367922043719,
      "grad_norm": 2.9127509593963623,
      "learning_rate": 4.540690018435607e-05,
      "loss": 1.3944,
      "step": 3493
    },
    {
      "epoch": 0.0920200158019489,
      "grad_norm": 4.527369022369385,
      "learning_rate": 4.540558335528049e-05,
      "loss": 1.9289,
      "step": 3494
    },
    {
      "epoch": 0.09204635238346062,
      "grad_norm": 3.5853872299194336,
      "learning_rate": 4.54042665262049e-05,
      "loss": 1.4307,
      "step": 3495
    },
    {
      "epoch": 0.09207268896497235,
      "grad_norm": 4.448419570922852,
      "learning_rate": 4.540294969712932e-05,
      "loss": 1.4803,
      "step": 3496
    },
    {
      "epoch": 0.09209902554648407,
      "grad_norm": 2.9541196823120117,
      "learning_rate": 4.540163286805373e-05,
      "loss": 2.2405,
      "step": 3497
    },
    {
      "epoch": 0.09212536212799578,
      "grad_norm": 5.0814948081970215,
      "learning_rate": 4.540031603897814e-05,
      "loss": 0.8587,
      "step": 3498
    },
    {
      "epoch": 0.09215169870950751,
      "grad_norm": 2.039577007293701,
      "learning_rate": 4.539899920990255e-05,
      "loss": 1.8517,
      "step": 3499
    },
    {
      "epoch": 0.09217803529101923,
      "grad_norm": 3.2318668365478516,
      "learning_rate": 4.5397682380826973e-05,
      "loss": 1.984,
      "step": 3500
    },
    {
      "epoch": 0.09220437187253094,
      "grad_norm": 4.1662163734436035,
      "learning_rate": 4.539636555175138e-05,
      "loss": 2.0723,
      "step": 3501
    },
    {
      "epoch": 0.09223070845404266,
      "grad_norm": 3.0352349281311035,
      "learning_rate": 4.53950487226758e-05,
      "loss": 1.4149,
      "step": 3502
    },
    {
      "epoch": 0.09225704503555439,
      "grad_norm": 3.3878934383392334,
      "learning_rate": 4.5393731893600213e-05,
      "loss": 1.1382,
      "step": 3503
    },
    {
      "epoch": 0.0922833816170661,
      "grad_norm": 3.484917402267456,
      "learning_rate": 4.539241506452463e-05,
      "loss": 2.4563,
      "step": 3504
    },
    {
      "epoch": 0.09230971819857782,
      "grad_norm": 2.1364645957946777,
      "learning_rate": 4.5391098235449045e-05,
      "loss": 1.8152,
      "step": 3505
    },
    {
      "epoch": 0.09233605478008955,
      "grad_norm": 3.4091784954071045,
      "learning_rate": 4.5389781406373453e-05,
      "loss": 2.3247,
      "step": 3506
    },
    {
      "epoch": 0.09236239136160126,
      "grad_norm": 2.75504732131958,
      "learning_rate": 4.538846457729787e-05,
      "loss": 2.0052,
      "step": 3507
    },
    {
      "epoch": 0.09238872794311298,
      "grad_norm": 1.5291812419891357,
      "learning_rate": 4.538714774822228e-05,
      "loss": 1.6434,
      "step": 3508
    },
    {
      "epoch": 0.09241506452462471,
      "grad_norm": 2.5041794776916504,
      "learning_rate": 4.53858309191467e-05,
      "loss": 1.8082,
      "step": 3509
    },
    {
      "epoch": 0.09244140110613643,
      "grad_norm": 4.507685661315918,
      "learning_rate": 4.538451409007111e-05,
      "loss": 1.6125,
      "step": 3510
    },
    {
      "epoch": 0.09246773768764814,
      "grad_norm": 4.7342634201049805,
      "learning_rate": 4.5383197260995525e-05,
      "loss": 1.9749,
      "step": 3511
    },
    {
      "epoch": 0.09249407426915986,
      "grad_norm": 2.9032680988311768,
      "learning_rate": 4.538188043191994e-05,
      "loss": 2.1286,
      "step": 3512
    },
    {
      "epoch": 0.09252041085067159,
      "grad_norm": 3.5357062816619873,
      "learning_rate": 4.538056360284435e-05,
      "loss": 2.0115,
      "step": 3513
    },
    {
      "epoch": 0.0925467474321833,
      "grad_norm": 2.528106927871704,
      "learning_rate": 4.537924677376877e-05,
      "loss": 1.2313,
      "step": 3514
    },
    {
      "epoch": 0.09257308401369502,
      "grad_norm": 1.802107334136963,
      "learning_rate": 4.537792994469318e-05,
      "loss": 1.2219,
      "step": 3515
    },
    {
      "epoch": 0.09259942059520675,
      "grad_norm": 2.346094846725464,
      "learning_rate": 4.5376613115617596e-05,
      "loss": 2.0631,
      "step": 3516
    },
    {
      "epoch": 0.09262575717671846,
      "grad_norm": 3.0261690616607666,
      "learning_rate": 4.5375296286542005e-05,
      "loss": 1.4302,
      "step": 3517
    },
    {
      "epoch": 0.09265209375823018,
      "grad_norm": 2.1139276027679443,
      "learning_rate": 4.537397945746643e-05,
      "loss": 1.4349,
      "step": 3518
    },
    {
      "epoch": 0.09267843033974191,
      "grad_norm": 2.240539312362671,
      "learning_rate": 4.5372662628390836e-05,
      "loss": 1.9774,
      "step": 3519
    },
    {
      "epoch": 0.09270476692125362,
      "grad_norm": 4.797918319702148,
      "learning_rate": 4.537134579931525e-05,
      "loss": 0.8176,
      "step": 3520
    },
    {
      "epoch": 0.09273110350276534,
      "grad_norm": 2.712881326675415,
      "learning_rate": 4.537002897023967e-05,
      "loss": 0.6057,
      "step": 3521
    },
    {
      "epoch": 0.09275744008427705,
      "grad_norm": 2.269455671310425,
      "learning_rate": 4.5368712141164076e-05,
      "loss": 2.0535,
      "step": 3522
    },
    {
      "epoch": 0.09278377666578878,
      "grad_norm": 3.852405548095703,
      "learning_rate": 4.53673953120885e-05,
      "loss": 0.8587,
      "step": 3523
    },
    {
      "epoch": 0.0928101132473005,
      "grad_norm": 3.5605695247650146,
      "learning_rate": 4.536607848301291e-05,
      "loss": 1.5908,
      "step": 3524
    },
    {
      "epoch": 0.09283644982881221,
      "grad_norm": 3.5673930644989014,
      "learning_rate": 4.536476165393732e-05,
      "loss": 2.4376,
      "step": 3525
    },
    {
      "epoch": 0.09286278641032394,
      "grad_norm": 2.3191635608673096,
      "learning_rate": 4.536344482486173e-05,
      "loss": 2.012,
      "step": 3526
    },
    {
      "epoch": 0.09288912299183566,
      "grad_norm": 1.6937735080718994,
      "learning_rate": 4.536212799578615e-05,
      "loss": 1.7562,
      "step": 3527
    },
    {
      "epoch": 0.09291545957334738,
      "grad_norm": 2.5325427055358887,
      "learning_rate": 4.536081116671056e-05,
      "loss": 1.7308,
      "step": 3528
    },
    {
      "epoch": 0.0929417961548591,
      "grad_norm": 1.8451688289642334,
      "learning_rate": 4.535949433763498e-05,
      "loss": 1.8,
      "step": 3529
    },
    {
      "epoch": 0.09296813273637082,
      "grad_norm": 2.1341207027435303,
      "learning_rate": 4.5358177508559394e-05,
      "loss": 1.6055,
      "step": 3530
    },
    {
      "epoch": 0.09299446931788254,
      "grad_norm": 3.2626891136169434,
      "learning_rate": 4.53568606794838e-05,
      "loss": 1.4218,
      "step": 3531
    },
    {
      "epoch": 0.09302080589939427,
      "grad_norm": 2.2403175830841064,
      "learning_rate": 4.5355543850408225e-05,
      "loss": 0.917,
      "step": 3532
    },
    {
      "epoch": 0.09304714248090598,
      "grad_norm": 2.618361234664917,
      "learning_rate": 4.5354227021332634e-05,
      "loss": 0.5938,
      "step": 3533
    },
    {
      "epoch": 0.0930734790624177,
      "grad_norm": 2.2554256916046143,
      "learning_rate": 4.535291019225705e-05,
      "loss": 1.2907,
      "step": 3534
    },
    {
      "epoch": 0.09309981564392941,
      "grad_norm": 1.6092514991760254,
      "learning_rate": 4.535159336318146e-05,
      "loss": 1.7301,
      "step": 3535
    },
    {
      "epoch": 0.09312615222544114,
      "grad_norm": 1.8308345079421997,
      "learning_rate": 4.5350276534105874e-05,
      "loss": 2.04,
      "step": 3536
    },
    {
      "epoch": 0.09315248880695286,
      "grad_norm": 1.8022361993789673,
      "learning_rate": 4.534895970503029e-05,
      "loss": 2.0434,
      "step": 3537
    },
    {
      "epoch": 0.09317882538846457,
      "grad_norm": 2.0214109420776367,
      "learning_rate": 4.5347642875954705e-05,
      "loss": 1.6638,
      "step": 3538
    },
    {
      "epoch": 0.0932051619699763,
      "grad_norm": 2.7885022163391113,
      "learning_rate": 4.5346326046879114e-05,
      "loss": 2.6573,
      "step": 3539
    },
    {
      "epoch": 0.09323149855148802,
      "grad_norm": 4.343166828155518,
      "learning_rate": 4.534500921780353e-05,
      "loss": 1.0507,
      "step": 3540
    },
    {
      "epoch": 0.09325783513299973,
      "grad_norm": 1.6867811679840088,
      "learning_rate": 4.5343692388727945e-05,
      "loss": 1.7796,
      "step": 3541
    },
    {
      "epoch": 0.09328417171451146,
      "grad_norm": 2.3982093334198,
      "learning_rate": 4.534237555965236e-05,
      "loss": 1.7246,
      "step": 3542
    },
    {
      "epoch": 0.09331050829602318,
      "grad_norm": 1.8931176662445068,
      "learning_rate": 4.5341058730576776e-05,
      "loss": 1.804,
      "step": 3543
    },
    {
      "epoch": 0.0933368448775349,
      "grad_norm": 2.561126947402954,
      "learning_rate": 4.5339741901501185e-05,
      "loss": 1.1635,
      "step": 3544
    },
    {
      "epoch": 0.09336318145904661,
      "grad_norm": 3.875622034072876,
      "learning_rate": 4.53384250724256e-05,
      "loss": 0.9318,
      "step": 3545
    },
    {
      "epoch": 0.09338951804055834,
      "grad_norm": 4.270608901977539,
      "learning_rate": 4.533710824335001e-05,
      "loss": 2.0671,
      "step": 3546
    },
    {
      "epoch": 0.09341585462207005,
      "grad_norm": 2.244062662124634,
      "learning_rate": 4.533579141427443e-05,
      "loss": 1.3546,
      "step": 3547
    },
    {
      "epoch": 0.09344219120358177,
      "grad_norm": 2.1929867267608643,
      "learning_rate": 4.533447458519884e-05,
      "loss": 1.8429,
      "step": 3548
    },
    {
      "epoch": 0.0934685277850935,
      "grad_norm": 2.2113425731658936,
      "learning_rate": 4.5333157756123256e-05,
      "loss": 1.9521,
      "step": 3549
    },
    {
      "epoch": 0.09349486436660522,
      "grad_norm": 1.865720510482788,
      "learning_rate": 4.533184092704767e-05,
      "loss": 1.7548,
      "step": 3550
    },
    {
      "epoch": 0.09352120094811693,
      "grad_norm": 4.912597179412842,
      "learning_rate": 4.533052409797209e-05,
      "loss": 1.0805,
      "step": 3551
    },
    {
      "epoch": 0.09354753752962866,
      "grad_norm": 2.5251331329345703,
      "learning_rate": 4.53292072688965e-05,
      "loss": 1.7525,
      "step": 3552
    },
    {
      "epoch": 0.09357387411114038,
      "grad_norm": 1.4777976274490356,
      "learning_rate": 4.532789043982091e-05,
      "loss": 1.5884,
      "step": 3553
    },
    {
      "epoch": 0.09360021069265209,
      "grad_norm": 3.00777006149292,
      "learning_rate": 4.532657361074533e-05,
      "loss": 2.3691,
      "step": 3554
    },
    {
      "epoch": 0.09362654727416381,
      "grad_norm": 7.6721038818359375,
      "learning_rate": 4.5325256781669736e-05,
      "loss": 1.1305,
      "step": 3555
    },
    {
      "epoch": 0.09365288385567554,
      "grad_norm": 1.8230565786361694,
      "learning_rate": 4.532393995259416e-05,
      "loss": 2.1075,
      "step": 3556
    },
    {
      "epoch": 0.09367922043718725,
      "grad_norm": 1.7814339399337769,
      "learning_rate": 4.532262312351857e-05,
      "loss": 2.1857,
      "step": 3557
    },
    {
      "epoch": 0.09370555701869897,
      "grad_norm": 4.5540289878845215,
      "learning_rate": 4.532130629444298e-05,
      "loss": 1.3009,
      "step": 3558
    },
    {
      "epoch": 0.0937318936002107,
      "grad_norm": 1.9780340194702148,
      "learning_rate": 4.53199894653674e-05,
      "loss": 1.6564,
      "step": 3559
    },
    {
      "epoch": 0.09375823018172241,
      "grad_norm": 2.5220141410827637,
      "learning_rate": 4.531867263629181e-05,
      "loss": 1.9551,
      "step": 3560
    },
    {
      "epoch": 0.09378456676323413,
      "grad_norm": 4.293489456176758,
      "learning_rate": 4.531735580721623e-05,
      "loss": 2.0275,
      "step": 3561
    },
    {
      "epoch": 0.09381090334474586,
      "grad_norm": 2.2610857486724854,
      "learning_rate": 4.531603897814064e-05,
      "loss": 1.5956,
      "step": 3562
    },
    {
      "epoch": 0.09383723992625757,
      "grad_norm": 4.735538482666016,
      "learning_rate": 4.5314722149065054e-05,
      "loss": 0.8918,
      "step": 3563
    },
    {
      "epoch": 0.09386357650776929,
      "grad_norm": 3.5435924530029297,
      "learning_rate": 4.531340531998946e-05,
      "loss": 1.6614,
      "step": 3564
    },
    {
      "epoch": 0.093889913089281,
      "grad_norm": 1.715957522392273,
      "learning_rate": 4.5312088490913886e-05,
      "loss": 2.4809,
      "step": 3565
    },
    {
      "epoch": 0.09391624967079273,
      "grad_norm": 4.611833572387695,
      "learning_rate": 4.5310771661838294e-05,
      "loss": 1.4671,
      "step": 3566
    },
    {
      "epoch": 0.09394258625230445,
      "grad_norm": 2.0678489208221436,
      "learning_rate": 4.530945483276271e-05,
      "loss": 0.5457,
      "step": 3567
    },
    {
      "epoch": 0.09396892283381617,
      "grad_norm": 2.0815773010253906,
      "learning_rate": 4.5308138003687126e-05,
      "loss": 1.6149,
      "step": 3568
    },
    {
      "epoch": 0.0939952594153279,
      "grad_norm": 3.303248882293701,
      "learning_rate": 4.5306821174611534e-05,
      "loss": 1.2468,
      "step": 3569
    },
    {
      "epoch": 0.09402159599683961,
      "grad_norm": 2.458906888961792,
      "learning_rate": 4.530550434553596e-05,
      "loss": 1.6612,
      "step": 3570
    },
    {
      "epoch": 0.09404793257835133,
      "grad_norm": 3.835932493209839,
      "learning_rate": 4.5304187516460366e-05,
      "loss": 2.5313,
      "step": 3571
    },
    {
      "epoch": 0.09407426915986306,
      "grad_norm": 1.7421433925628662,
      "learning_rate": 4.530287068738478e-05,
      "loss": 1.5607,
      "step": 3572
    },
    {
      "epoch": 0.09410060574137477,
      "grad_norm": 5.627799987792969,
      "learning_rate": 4.530155385830919e-05,
      "loss": 0.3134,
      "step": 3573
    },
    {
      "epoch": 0.09412694232288649,
      "grad_norm": 2.220440149307251,
      "learning_rate": 4.5300237029233606e-05,
      "loss": 2.4299,
      "step": 3574
    },
    {
      "epoch": 0.09415327890439822,
      "grad_norm": 2.5359511375427246,
      "learning_rate": 4.529892020015802e-05,
      "loss": 2.1116,
      "step": 3575
    },
    {
      "epoch": 0.09417961548590993,
      "grad_norm": 3.1033644676208496,
      "learning_rate": 4.529760337108244e-05,
      "loss": 1.6407,
      "step": 3576
    },
    {
      "epoch": 0.09420595206742165,
      "grad_norm": 1.7653049230575562,
      "learning_rate": 4.529628654200685e-05,
      "loss": 1.629,
      "step": 3577
    },
    {
      "epoch": 0.09423228864893336,
      "grad_norm": 1.9302960634231567,
      "learning_rate": 4.529496971293126e-05,
      "loss": 1.71,
      "step": 3578
    },
    {
      "epoch": 0.09425862523044509,
      "grad_norm": 2.943619966506958,
      "learning_rate": 4.529365288385568e-05,
      "loss": 0.7295,
      "step": 3579
    },
    {
      "epoch": 0.09428496181195681,
      "grad_norm": 2.1683950424194336,
      "learning_rate": 4.529233605478009e-05,
      "loss": 1.6699,
      "step": 3580
    },
    {
      "epoch": 0.09431129839346852,
      "grad_norm": 2.8118507862091064,
      "learning_rate": 4.529101922570451e-05,
      "loss": 1.3244,
      "step": 3581
    },
    {
      "epoch": 0.09433763497498025,
      "grad_norm": 2.927462100982666,
      "learning_rate": 4.528970239662892e-05,
      "loss": 2.366,
      "step": 3582
    },
    {
      "epoch": 0.09436397155649197,
      "grad_norm": 4.5996599197387695,
      "learning_rate": 4.528838556755333e-05,
      "loss": 1.4591,
      "step": 3583
    },
    {
      "epoch": 0.09439030813800368,
      "grad_norm": 2.266023635864258,
      "learning_rate": 4.528706873847775e-05,
      "loss": 2.0292,
      "step": 3584
    },
    {
      "epoch": 0.09441664471951541,
      "grad_norm": 4.423757076263428,
      "learning_rate": 4.5285751909402164e-05,
      "loss": 1.973,
      "step": 3585
    },
    {
      "epoch": 0.09444298130102713,
      "grad_norm": 6.177351951599121,
      "learning_rate": 4.528443508032658e-05,
      "loss": 1.9698,
      "step": 3586
    },
    {
      "epoch": 0.09446931788253884,
      "grad_norm": 3.440419912338257,
      "learning_rate": 4.528311825125099e-05,
      "loss": 1.8221,
      "step": 3587
    },
    {
      "epoch": 0.09449565446405056,
      "grad_norm": 2.492399215698242,
      "learning_rate": 4.5281801422175404e-05,
      "loss": 2.0154,
      "step": 3588
    },
    {
      "epoch": 0.09452199104556229,
      "grad_norm": 2.0901925563812256,
      "learning_rate": 4.528048459309982e-05,
      "loss": 0.9605,
      "step": 3589
    },
    {
      "epoch": 0.094548327627074,
      "grad_norm": 1.8072923421859741,
      "learning_rate": 4.5279167764024235e-05,
      "loss": 1.7053,
      "step": 3590
    },
    {
      "epoch": 0.09457466420858572,
      "grad_norm": 2.0344669818878174,
      "learning_rate": 4.5277850934948644e-05,
      "loss": 1.8751,
      "step": 3591
    },
    {
      "epoch": 0.09460100079009745,
      "grad_norm": 2.647554397583008,
      "learning_rate": 4.527653410587306e-05,
      "loss": 1.9211,
      "step": 3592
    },
    {
      "epoch": 0.09462733737160917,
      "grad_norm": 3.688228130340576,
      "learning_rate": 4.527521727679747e-05,
      "loss": 2.1693,
      "step": 3593
    },
    {
      "epoch": 0.09465367395312088,
      "grad_norm": 2.506467580795288,
      "learning_rate": 4.527390044772189e-05,
      "loss": 1.9605,
      "step": 3594
    },
    {
      "epoch": 0.09468001053463261,
      "grad_norm": 4.199458599090576,
      "learning_rate": 4.52725836186463e-05,
      "loss": 0.9748,
      "step": 3595
    },
    {
      "epoch": 0.09470634711614433,
      "grad_norm": 2.637802839279175,
      "learning_rate": 4.5271266789570715e-05,
      "loss": 0.9159,
      "step": 3596
    },
    {
      "epoch": 0.09473268369765604,
      "grad_norm": 1.9187687635421753,
      "learning_rate": 4.526994996049513e-05,
      "loss": 1.8459,
      "step": 3597
    },
    {
      "epoch": 0.09475902027916776,
      "grad_norm": 2.075516700744629,
      "learning_rate": 4.5268633131419546e-05,
      "loss": 2.3203,
      "step": 3598
    },
    {
      "epoch": 0.09478535686067949,
      "grad_norm": 2.219097375869751,
      "learning_rate": 4.526731630234396e-05,
      "loss": 2.166,
      "step": 3599
    },
    {
      "epoch": 0.0948116934421912,
      "grad_norm": 2.6724448204040527,
      "learning_rate": 4.526599947326837e-05,
      "loss": 1.9467,
      "step": 3600
    },
    {
      "epoch": 0.09483803002370292,
      "grad_norm": 1.9259095191955566,
      "learning_rate": 4.5264682644192786e-05,
      "loss": 1.9302,
      "step": 3601
    },
    {
      "epoch": 0.09486436660521465,
      "grad_norm": 2.850823163986206,
      "learning_rate": 4.5263365815117195e-05,
      "loss": 1.8097,
      "step": 3602
    },
    {
      "epoch": 0.09489070318672636,
      "grad_norm": 3.107180595397949,
      "learning_rate": 4.526204898604162e-05,
      "loss": 1.3213,
      "step": 3603
    },
    {
      "epoch": 0.09491703976823808,
      "grad_norm": 3.6365256309509277,
      "learning_rate": 4.5260732156966026e-05,
      "loss": 1.0158,
      "step": 3604
    },
    {
      "epoch": 0.09494337634974981,
      "grad_norm": 1.9909608364105225,
      "learning_rate": 4.525941532789044e-05,
      "loss": 1.9825,
      "step": 3605
    },
    {
      "epoch": 0.09496971293126152,
      "grad_norm": 2.534580945968628,
      "learning_rate": 4.525809849881486e-05,
      "loss": 2.0879,
      "step": 3606
    },
    {
      "epoch": 0.09499604951277324,
      "grad_norm": 2.238964796066284,
      "learning_rate": 4.5256781669739266e-05,
      "loss": 1.5413,
      "step": 3607
    },
    {
      "epoch": 0.09502238609428496,
      "grad_norm": 1.7729644775390625,
      "learning_rate": 4.525546484066369e-05,
      "loss": 1.5519,
      "step": 3608
    },
    {
      "epoch": 0.09504872267579668,
      "grad_norm": 2.237271547317505,
      "learning_rate": 4.52541480115881e-05,
      "loss": 1.8953,
      "step": 3609
    },
    {
      "epoch": 0.0950750592573084,
      "grad_norm": 1.7214950323104858,
      "learning_rate": 4.525283118251251e-05,
      "loss": 1.4514,
      "step": 3610
    },
    {
      "epoch": 0.09510139583882012,
      "grad_norm": 2.7791130542755127,
      "learning_rate": 4.525151435343692e-05,
      "loss": 1.595,
      "step": 3611
    },
    {
      "epoch": 0.09512773242033185,
      "grad_norm": 5.118468284606934,
      "learning_rate": 4.525019752436134e-05,
      "loss": 1.3965,
      "step": 3612
    },
    {
      "epoch": 0.09515406900184356,
      "grad_norm": 2.1532459259033203,
      "learning_rate": 4.524888069528575e-05,
      "loss": 1.1161,
      "step": 3613
    },
    {
      "epoch": 0.09518040558335528,
      "grad_norm": 2.744793653488159,
      "learning_rate": 4.524756386621017e-05,
      "loss": 0.6077,
      "step": 3614
    },
    {
      "epoch": 0.095206742164867,
      "grad_norm": 3.585491895675659,
      "learning_rate": 4.5246247037134584e-05,
      "loss": 1.3194,
      "step": 3615
    },
    {
      "epoch": 0.09523307874637872,
      "grad_norm": 1.915829062461853,
      "learning_rate": 4.524493020805899e-05,
      "loss": 1.469,
      "step": 3616
    },
    {
      "epoch": 0.09525941532789044,
      "grad_norm": 2.767784595489502,
      "learning_rate": 4.5243613378983415e-05,
      "loss": 2.0906,
      "step": 3617
    },
    {
      "epoch": 0.09528575190940215,
      "grad_norm": 2.7908408641815186,
      "learning_rate": 4.5242296549907824e-05,
      "loss": 2.1139,
      "step": 3618
    },
    {
      "epoch": 0.09531208849091388,
      "grad_norm": 2.8103015422821045,
      "learning_rate": 4.524097972083224e-05,
      "loss": 1.83,
      "step": 3619
    },
    {
      "epoch": 0.0953384250724256,
      "grad_norm": 1.7638869285583496,
      "learning_rate": 4.523966289175665e-05,
      "loss": 0.5245,
      "step": 3620
    },
    {
      "epoch": 0.09536476165393731,
      "grad_norm": 3.2313201427459717,
      "learning_rate": 4.5238346062681064e-05,
      "loss": 1.5791,
      "step": 3621
    },
    {
      "epoch": 0.09539109823544904,
      "grad_norm": 2.9392025470733643,
      "learning_rate": 4.523702923360548e-05,
      "loss": 1.0644,
      "step": 3622
    },
    {
      "epoch": 0.09541743481696076,
      "grad_norm": 2.4455037117004395,
      "learning_rate": 4.5235712404529895e-05,
      "loss": 2.24,
      "step": 3623
    },
    {
      "epoch": 0.09544377139847247,
      "grad_norm": 6.37355899810791,
      "learning_rate": 4.523439557545431e-05,
      "loss": 1.3735,
      "step": 3624
    },
    {
      "epoch": 0.0954701079799842,
      "grad_norm": 6.340606212615967,
      "learning_rate": 4.523307874637872e-05,
      "loss": 0.7241,
      "step": 3625
    },
    {
      "epoch": 0.09549644456149592,
      "grad_norm": 2.154714584350586,
      "learning_rate": 4.5231761917303135e-05,
      "loss": 1.112,
      "step": 3626
    },
    {
      "epoch": 0.09552278114300763,
      "grad_norm": 2.133312940597534,
      "learning_rate": 4.523044508822755e-05,
      "loss": 1.2442,
      "step": 3627
    },
    {
      "epoch": 0.09554911772451936,
      "grad_norm": 3.2518069744110107,
      "learning_rate": 4.522912825915197e-05,
      "loss": 2.0202,
      "step": 3628
    },
    {
      "epoch": 0.09557545430603108,
      "grad_norm": 1.9350520372390747,
      "learning_rate": 4.5227811430076375e-05,
      "loss": 1.9149,
      "step": 3629
    },
    {
      "epoch": 0.0956017908875428,
      "grad_norm": 2.430617332458496,
      "learning_rate": 4.522649460100079e-05,
      "loss": 2.3257,
      "step": 3630
    },
    {
      "epoch": 0.09562812746905451,
      "grad_norm": 2.9433438777923584,
      "learning_rate": 4.522517777192521e-05,
      "loss": 1.324,
      "step": 3631
    },
    {
      "epoch": 0.09565446405056624,
      "grad_norm": 8.794106483459473,
      "learning_rate": 4.522386094284962e-05,
      "loss": 1.8983,
      "step": 3632
    },
    {
      "epoch": 0.09568080063207796,
      "grad_norm": 5.53743314743042,
      "learning_rate": 4.522254411377404e-05,
      "loss": 1.3177,
      "step": 3633
    },
    {
      "epoch": 0.09570713721358967,
      "grad_norm": 3.19970703125,
      "learning_rate": 4.522122728469845e-05,
      "loss": 1.6594,
      "step": 3634
    },
    {
      "epoch": 0.0957334737951014,
      "grad_norm": 2.3984477519989014,
      "learning_rate": 4.521991045562286e-05,
      "loss": 1.6546,
      "step": 3635
    },
    {
      "epoch": 0.09575981037661312,
      "grad_norm": 4.994995594024658,
      "learning_rate": 4.521859362654728e-05,
      "loss": 2.6488,
      "step": 3636
    },
    {
      "epoch": 0.09578614695812483,
      "grad_norm": 3.187831163406372,
      "learning_rate": 4.5217276797471693e-05,
      "loss": 2.6566,
      "step": 3637
    },
    {
      "epoch": 0.09581248353963656,
      "grad_norm": 1.8058795928955078,
      "learning_rate": 4.52159599683961e-05,
      "loss": 1.6738,
      "step": 3638
    },
    {
      "epoch": 0.09583882012114828,
      "grad_norm": 2.445007085800171,
      "learning_rate": 4.521464313932052e-05,
      "loss": 1.3263,
      "step": 3639
    },
    {
      "epoch": 0.09586515670265999,
      "grad_norm": 1.6659449338912964,
      "learning_rate": 4.521332631024493e-05,
      "loss": 1.6478,
      "step": 3640
    },
    {
      "epoch": 0.09589149328417171,
      "grad_norm": 2.046175956726074,
      "learning_rate": 4.521200948116935e-05,
      "loss": 1.7378,
      "step": 3641
    },
    {
      "epoch": 0.09591782986568344,
      "grad_norm": 2.381976366043091,
      "learning_rate": 4.521069265209376e-05,
      "loss": 1.7872,
      "step": 3642
    },
    {
      "epoch": 0.09594416644719515,
      "grad_norm": 3.948195457458496,
      "learning_rate": 4.5209375823018174e-05,
      "loss": 1.816,
      "step": 3643
    },
    {
      "epoch": 0.09597050302870687,
      "grad_norm": 2.0074939727783203,
      "learning_rate": 4.520805899394259e-05,
      "loss": 2.2602,
      "step": 3644
    },
    {
      "epoch": 0.0959968396102186,
      "grad_norm": 2.225593328475952,
      "learning_rate": 4.5206742164867e-05,
      "loss": 2.3454,
      "step": 3645
    },
    {
      "epoch": 0.09602317619173031,
      "grad_norm": 3.638373613357544,
      "learning_rate": 4.520542533579142e-05,
      "loss": 0.9837,
      "step": 3646
    },
    {
      "epoch": 0.09604951277324203,
      "grad_norm": 2.8617241382598877,
      "learning_rate": 4.520410850671583e-05,
      "loss": 0.8882,
      "step": 3647
    },
    {
      "epoch": 0.09607584935475376,
      "grad_norm": 7.497365474700928,
      "learning_rate": 4.5202791677640245e-05,
      "loss": 1.0505,
      "step": 3648
    },
    {
      "epoch": 0.09610218593626547,
      "grad_norm": 2.914956569671631,
      "learning_rate": 4.5201474848564654e-05,
      "loss": 1.8209,
      "step": 3649
    },
    {
      "epoch": 0.09612852251777719,
      "grad_norm": 2.335761308670044,
      "learning_rate": 4.5200158019489076e-05,
      "loss": 1.4838,
      "step": 3650
    },
    {
      "epoch": 0.0961548590992889,
      "grad_norm": 3.184492826461792,
      "learning_rate": 4.5198841190413485e-05,
      "loss": 2.1476,
      "step": 3651
    },
    {
      "epoch": 0.09618119568080064,
      "grad_norm": 3.5369904041290283,
      "learning_rate": 4.51975243613379e-05,
      "loss": 1.3361,
      "step": 3652
    },
    {
      "epoch": 0.09620753226231235,
      "grad_norm": 2.3214669227600098,
      "learning_rate": 4.5196207532262316e-05,
      "loss": 2.0435,
      "step": 3653
    },
    {
      "epoch": 0.09623386884382407,
      "grad_norm": 1.8957983255386353,
      "learning_rate": 4.5194890703186725e-05,
      "loss": 1.9756,
      "step": 3654
    },
    {
      "epoch": 0.0962602054253358,
      "grad_norm": 6.836841106414795,
      "learning_rate": 4.519357387411115e-05,
      "loss": 4.0496,
      "step": 3655
    },
    {
      "epoch": 0.09628654200684751,
      "grad_norm": 8.642605781555176,
      "learning_rate": 4.5192257045035556e-05,
      "loss": 1.8218,
      "step": 3656
    },
    {
      "epoch": 0.09631287858835923,
      "grad_norm": 5.0724334716796875,
      "learning_rate": 4.519094021595997e-05,
      "loss": 2.0119,
      "step": 3657
    },
    {
      "epoch": 0.09633921516987096,
      "grad_norm": 1.6873488426208496,
      "learning_rate": 4.518962338688438e-05,
      "loss": 1.9877,
      "step": 3658
    },
    {
      "epoch": 0.09636555175138267,
      "grad_norm": 2.1610593795776367,
      "learning_rate": 4.5188306557808796e-05,
      "loss": 2.7636,
      "step": 3659
    },
    {
      "epoch": 0.09639188833289439,
      "grad_norm": 3.336202383041382,
      "learning_rate": 4.518698972873321e-05,
      "loss": 1.2608,
      "step": 3660
    },
    {
      "epoch": 0.0964182249144061,
      "grad_norm": 2.1662914752960205,
      "learning_rate": 4.518567289965763e-05,
      "loss": 1.6087,
      "step": 3661
    },
    {
      "epoch": 0.09644456149591783,
      "grad_norm": 1.8431601524353027,
      "learning_rate": 4.518435607058204e-05,
      "loss": 1.5027,
      "step": 3662
    },
    {
      "epoch": 0.09647089807742955,
      "grad_norm": 2.0472123622894287,
      "learning_rate": 4.518303924150645e-05,
      "loss": 2.1419,
      "step": 3663
    },
    {
      "epoch": 0.09649723465894126,
      "grad_norm": 1.5350587368011475,
      "learning_rate": 4.5181722412430874e-05,
      "loss": 1.2688,
      "step": 3664
    },
    {
      "epoch": 0.096523571240453,
      "grad_norm": 2.0831425189971924,
      "learning_rate": 4.518040558335528e-05,
      "loss": 1.7411,
      "step": 3665
    },
    {
      "epoch": 0.09654990782196471,
      "grad_norm": 2.745490789413452,
      "learning_rate": 4.51790887542797e-05,
      "loss": 1.8446,
      "step": 3666
    },
    {
      "epoch": 0.09657624440347642,
      "grad_norm": 2.4591481685638428,
      "learning_rate": 4.517777192520411e-05,
      "loss": 1.3824,
      "step": 3667
    },
    {
      "epoch": 0.09660258098498815,
      "grad_norm": 4.647483825683594,
      "learning_rate": 4.517645509612852e-05,
      "loss": 2.1942,
      "step": 3668
    },
    {
      "epoch": 0.09662891756649987,
      "grad_norm": 2.449812650680542,
      "learning_rate": 4.517513826705294e-05,
      "loss": 1.9773,
      "step": 3669
    },
    {
      "epoch": 0.09665525414801159,
      "grad_norm": 2.2432150840759277,
      "learning_rate": 4.5173821437977354e-05,
      "loss": 1.8984,
      "step": 3670
    },
    {
      "epoch": 0.09668159072952331,
      "grad_norm": 2.4704086780548096,
      "learning_rate": 4.517250460890177e-05,
      "loss": 0.739,
      "step": 3671
    },
    {
      "epoch": 0.09670792731103503,
      "grad_norm": 3.129194498062134,
      "learning_rate": 4.517118777982618e-05,
      "loss": 2.5484,
      "step": 3672
    },
    {
      "epoch": 0.09673426389254675,
      "grad_norm": 2.480876922607422,
      "learning_rate": 4.5169870950750594e-05,
      "loss": 0.6907,
      "step": 3673
    },
    {
      "epoch": 0.09676060047405846,
      "grad_norm": 1.9893954992294312,
      "learning_rate": 4.516855412167501e-05,
      "loss": 2.3519,
      "step": 3674
    },
    {
      "epoch": 0.09678693705557019,
      "grad_norm": 2.693394184112549,
      "learning_rate": 4.5167237292599425e-05,
      "loss": 2.2721,
      "step": 3675
    },
    {
      "epoch": 0.0968132736370819,
      "grad_norm": 1.923799753189087,
      "learning_rate": 4.5165920463523834e-05,
      "loss": 1.2271,
      "step": 3676
    },
    {
      "epoch": 0.09683961021859362,
      "grad_norm": 2.5754921436309814,
      "learning_rate": 4.516460363444825e-05,
      "loss": 2.0139,
      "step": 3677
    },
    {
      "epoch": 0.09686594680010535,
      "grad_norm": 2.0736756324768066,
      "learning_rate": 4.5163286805372665e-05,
      "loss": 1.5492,
      "step": 3678
    },
    {
      "epoch": 0.09689228338161707,
      "grad_norm": 2.9396755695343018,
      "learning_rate": 4.516196997629708e-05,
      "loss": 1.3545,
      "step": 3679
    },
    {
      "epoch": 0.09691861996312878,
      "grad_norm": 2.0697171688079834,
      "learning_rate": 4.5160653147221496e-05,
      "loss": 2.2178,
      "step": 3680
    },
    {
      "epoch": 0.09694495654464051,
      "grad_norm": 2.6935935020446777,
      "learning_rate": 4.5159336318145905e-05,
      "loss": 2.1558,
      "step": 3681
    },
    {
      "epoch": 0.09697129312615223,
      "grad_norm": 2.0553812980651855,
      "learning_rate": 4.515801948907032e-05,
      "loss": 1.7593,
      "step": 3682
    },
    {
      "epoch": 0.09699762970766394,
      "grad_norm": 4.6022419929504395,
      "learning_rate": 4.5156702659994736e-05,
      "loss": 1.3408,
      "step": 3683
    },
    {
      "epoch": 0.09702396628917566,
      "grad_norm": 2.250375747680664,
      "learning_rate": 4.515538583091915e-05,
      "loss": 1.8138,
      "step": 3684
    },
    {
      "epoch": 0.09705030287068739,
      "grad_norm": 3.7652809619903564,
      "learning_rate": 4.515406900184356e-05,
      "loss": 2.105,
      "step": 3685
    },
    {
      "epoch": 0.0970766394521991,
      "grad_norm": 2.4491586685180664,
      "learning_rate": 4.5152752172767976e-05,
      "loss": 2.4553,
      "step": 3686
    },
    {
      "epoch": 0.09710297603371082,
      "grad_norm": 3.584064245223999,
      "learning_rate": 4.515143534369239e-05,
      "loss": 0.5796,
      "step": 3687
    },
    {
      "epoch": 0.09712931261522255,
      "grad_norm": 1.9849361181259155,
      "learning_rate": 4.515011851461681e-05,
      "loss": 1.7573,
      "step": 3688
    },
    {
      "epoch": 0.09715564919673426,
      "grad_norm": 1.837116003036499,
      "learning_rate": 4.514880168554122e-05,
      "loss": 1.8401,
      "step": 3689
    },
    {
      "epoch": 0.09718198577824598,
      "grad_norm": 3.2957603931427,
      "learning_rate": 4.514748485646563e-05,
      "loss": 1.9486,
      "step": 3690
    },
    {
      "epoch": 0.09720832235975771,
      "grad_norm": 1.8041918277740479,
      "learning_rate": 4.514616802739005e-05,
      "loss": 1.36,
      "step": 3691
    },
    {
      "epoch": 0.09723465894126943,
      "grad_norm": 3.6805500984191895,
      "learning_rate": 4.5144851198314456e-05,
      "loss": 1.9729,
      "step": 3692
    },
    {
      "epoch": 0.09726099552278114,
      "grad_norm": 2.446524143218994,
      "learning_rate": 4.514353436923888e-05,
      "loss": 1.4213,
      "step": 3693
    },
    {
      "epoch": 0.09728733210429286,
      "grad_norm": 2.0969126224517822,
      "learning_rate": 4.514221754016329e-05,
      "loss": 1.9055,
      "step": 3694
    },
    {
      "epoch": 0.09731366868580459,
      "grad_norm": 4.652068614959717,
      "learning_rate": 4.51409007110877e-05,
      "loss": 0.7785,
      "step": 3695
    },
    {
      "epoch": 0.0973400052673163,
      "grad_norm": 2.7421815395355225,
      "learning_rate": 4.513958388201211e-05,
      "loss": 1.282,
      "step": 3696
    },
    {
      "epoch": 0.09736634184882802,
      "grad_norm": 4.3839569091796875,
      "learning_rate": 4.5138267052936534e-05,
      "loss": 1.85,
      "step": 3697
    },
    {
      "epoch": 0.09739267843033975,
      "grad_norm": 2.232586145401001,
      "learning_rate": 4.513695022386094e-05,
      "loss": 1.8399,
      "step": 3698
    },
    {
      "epoch": 0.09741901501185146,
      "grad_norm": 2.443162202835083,
      "learning_rate": 4.513563339478536e-05,
      "loss": 2.2778,
      "step": 3699
    },
    {
      "epoch": 0.09744535159336318,
      "grad_norm": 2.664928674697876,
      "learning_rate": 4.5134316565709774e-05,
      "loss": 1.0631,
      "step": 3700
    },
    {
      "epoch": 0.09747168817487491,
      "grad_norm": 1.9210928678512573,
      "learning_rate": 4.513299973663418e-05,
      "loss": 2.0139,
      "step": 3701
    },
    {
      "epoch": 0.09749802475638662,
      "grad_norm": 2.0483391284942627,
      "learning_rate": 4.5131682907558606e-05,
      "loss": 1.6869,
      "step": 3702
    },
    {
      "epoch": 0.09752436133789834,
      "grad_norm": 2.570066213607788,
      "learning_rate": 4.5130366078483015e-05,
      "loss": 1.7603,
      "step": 3703
    },
    {
      "epoch": 0.09755069791941005,
      "grad_norm": 2.8687045574188232,
      "learning_rate": 4.512904924940743e-05,
      "loss": 2.3633,
      "step": 3704
    },
    {
      "epoch": 0.09757703450092178,
      "grad_norm": 2.9738430976867676,
      "learning_rate": 4.512773242033184e-05,
      "loss": 1.4073,
      "step": 3705
    },
    {
      "epoch": 0.0976033710824335,
      "grad_norm": 2.5782880783081055,
      "learning_rate": 4.5126415591256255e-05,
      "loss": 1.9642,
      "step": 3706
    },
    {
      "epoch": 0.09762970766394521,
      "grad_norm": 2.512622117996216,
      "learning_rate": 4.512509876218067e-05,
      "loss": 1.6972,
      "step": 3707
    },
    {
      "epoch": 0.09765604424545694,
      "grad_norm": 4.486298084259033,
      "learning_rate": 4.5123781933105086e-05,
      "loss": 1.5843,
      "step": 3708
    },
    {
      "epoch": 0.09768238082696866,
      "grad_norm": 1.773552417755127,
      "learning_rate": 4.51224651040295e-05,
      "loss": 1.689,
      "step": 3709
    },
    {
      "epoch": 0.09770871740848038,
      "grad_norm": 2.6420490741729736,
      "learning_rate": 4.512114827495391e-05,
      "loss": 1.6237,
      "step": 3710
    },
    {
      "epoch": 0.0977350539899921,
      "grad_norm": 3.664464235305786,
      "learning_rate": 4.5119831445878326e-05,
      "loss": 1.0444,
      "step": 3711
    },
    {
      "epoch": 0.09776139057150382,
      "grad_norm": 1.7949111461639404,
      "learning_rate": 4.511851461680274e-05,
      "loss": 1.9621,
      "step": 3712
    },
    {
      "epoch": 0.09778772715301554,
      "grad_norm": 4.075164794921875,
      "learning_rate": 4.511719778772716e-05,
      "loss": 1.4136,
      "step": 3713
    },
    {
      "epoch": 0.09781406373452727,
      "grad_norm": 2.90339994430542,
      "learning_rate": 4.5115880958651566e-05,
      "loss": 2.0128,
      "step": 3714
    },
    {
      "epoch": 0.09784040031603898,
      "grad_norm": 2.8007373809814453,
      "learning_rate": 4.511456412957598e-05,
      "loss": 0.6457,
      "step": 3715
    },
    {
      "epoch": 0.0978667368975507,
      "grad_norm": 3.6463592052459717,
      "learning_rate": 4.51132473005004e-05,
      "loss": 1.9723,
      "step": 3716
    },
    {
      "epoch": 0.09789307347906241,
      "grad_norm": 2.673933506011963,
      "learning_rate": 4.511193047142481e-05,
      "loss": 2.447,
      "step": 3717
    },
    {
      "epoch": 0.09791941006057414,
      "grad_norm": 1.8435684442520142,
      "learning_rate": 4.511061364234923e-05,
      "loss": 2.3292,
      "step": 3718
    },
    {
      "epoch": 0.09794574664208586,
      "grad_norm": 2.1149821281433105,
      "learning_rate": 4.510929681327364e-05,
      "loss": 1.9996,
      "step": 3719
    },
    {
      "epoch": 0.09797208322359757,
      "grad_norm": 2.6121387481689453,
      "learning_rate": 4.510797998419805e-05,
      "loss": 1.7739,
      "step": 3720
    },
    {
      "epoch": 0.0979984198051093,
      "grad_norm": 1.902207374572754,
      "learning_rate": 4.510666315512247e-05,
      "loss": 1.6011,
      "step": 3721
    },
    {
      "epoch": 0.09802475638662102,
      "grad_norm": 2.0520217418670654,
      "learning_rate": 4.5105346326046884e-05,
      "loss": 2.1635,
      "step": 3722
    },
    {
      "epoch": 0.09805109296813273,
      "grad_norm": 2.4916579723358154,
      "learning_rate": 4.510402949697129e-05,
      "loss": 1.272,
      "step": 3723
    },
    {
      "epoch": 0.09807742954964446,
      "grad_norm": 2.4104409217834473,
      "learning_rate": 4.510271266789571e-05,
      "loss": 1.9387,
      "step": 3724
    },
    {
      "epoch": 0.09810376613115618,
      "grad_norm": 1.992605447769165,
      "learning_rate": 4.5101395838820124e-05,
      "loss": 2.2106,
      "step": 3725
    },
    {
      "epoch": 0.0981301027126679,
      "grad_norm": 2.632085084915161,
      "learning_rate": 4.510007900974454e-05,
      "loss": 1.8529,
      "step": 3726
    },
    {
      "epoch": 0.09815643929417961,
      "grad_norm": 1.9421182870864868,
      "learning_rate": 4.5098762180668955e-05,
      "loss": 2.2319,
      "step": 3727
    },
    {
      "epoch": 0.09818277587569134,
      "grad_norm": 4.704944610595703,
      "learning_rate": 4.5097445351593364e-05,
      "loss": 2.316,
      "step": 3728
    },
    {
      "epoch": 0.09820911245720305,
      "grad_norm": 2.330127477645874,
      "learning_rate": 4.509612852251778e-05,
      "loss": 1.648,
      "step": 3729
    },
    {
      "epoch": 0.09823544903871477,
      "grad_norm": 3.828066110610962,
      "learning_rate": 4.5094811693442195e-05,
      "loss": 0.742,
      "step": 3730
    },
    {
      "epoch": 0.0982617856202265,
      "grad_norm": 5.853951454162598,
      "learning_rate": 4.509349486436661e-05,
      "loss": 1.2046,
      "step": 3731
    },
    {
      "epoch": 0.09828812220173822,
      "grad_norm": 3.631951093673706,
      "learning_rate": 4.509217803529102e-05,
      "loss": 1.0181,
      "step": 3732
    },
    {
      "epoch": 0.09831445878324993,
      "grad_norm": 1.8185824155807495,
      "learning_rate": 4.5090861206215435e-05,
      "loss": 1.5716,
      "step": 3733
    },
    {
      "epoch": 0.09834079536476166,
      "grad_norm": 3.048617362976074,
      "learning_rate": 4.508954437713985e-05,
      "loss": 1.9356,
      "step": 3734
    },
    {
      "epoch": 0.09836713194627338,
      "grad_norm": 2.251716375350952,
      "learning_rate": 4.5088227548064266e-05,
      "loss": 2.5036,
      "step": 3735
    },
    {
      "epoch": 0.09839346852778509,
      "grad_norm": 2.165480852127075,
      "learning_rate": 4.508691071898868e-05,
      "loss": 1.5794,
      "step": 3736
    },
    {
      "epoch": 0.0984198051092968,
      "grad_norm": 1.8818281888961792,
      "learning_rate": 4.508559388991309e-05,
      "loss": 1.8376,
      "step": 3737
    },
    {
      "epoch": 0.09844614169080854,
      "grad_norm": 4.45981502532959,
      "learning_rate": 4.5084277060837506e-05,
      "loss": 1.5519,
      "step": 3738
    },
    {
      "epoch": 0.09847247827232025,
      "grad_norm": 2.865576982498169,
      "learning_rate": 4.5082960231761915e-05,
      "loss": 0.4992,
      "step": 3739
    },
    {
      "epoch": 0.09849881485383197,
      "grad_norm": 2.348257541656494,
      "learning_rate": 4.508164340268634e-05,
      "loss": 2.2336,
      "step": 3740
    },
    {
      "epoch": 0.0985251514353437,
      "grad_norm": 2.0799033641815186,
      "learning_rate": 4.5080326573610746e-05,
      "loss": 2.0153,
      "step": 3741
    },
    {
      "epoch": 0.09855148801685541,
      "grad_norm": 3.990330457687378,
      "learning_rate": 4.507900974453516e-05,
      "loss": 1.3609,
      "step": 3742
    },
    {
      "epoch": 0.09857782459836713,
      "grad_norm": 2.3788907527923584,
      "learning_rate": 4.507769291545957e-05,
      "loss": 1.8452,
      "step": 3743
    },
    {
      "epoch": 0.09860416117987886,
      "grad_norm": 2.714266777038574,
      "learning_rate": 4.5076376086383986e-05,
      "loss": 2.0757,
      "step": 3744
    },
    {
      "epoch": 0.09863049776139057,
      "grad_norm": 2.107374906539917,
      "learning_rate": 4.50750592573084e-05,
      "loss": 1.8191,
      "step": 3745
    },
    {
      "epoch": 0.09865683434290229,
      "grad_norm": 4.436236381530762,
      "learning_rate": 4.507374242823282e-05,
      "loss": 1.948,
      "step": 3746
    },
    {
      "epoch": 0.098683170924414,
      "grad_norm": 2.7902300357818604,
      "learning_rate": 4.507242559915723e-05,
      "loss": 2.0981,
      "step": 3747
    },
    {
      "epoch": 0.09870950750592573,
      "grad_norm": 1.75342857837677,
      "learning_rate": 4.507110877008164e-05,
      "loss": 1.8215,
      "step": 3748
    },
    {
      "epoch": 0.09873584408743745,
      "grad_norm": 3.086707830429077,
      "learning_rate": 4.5069791941006064e-05,
      "loss": 1.8107,
      "step": 3749
    },
    {
      "epoch": 0.09876218066894916,
      "grad_norm": 2.3465688228607178,
      "learning_rate": 4.506847511193047e-05,
      "loss": 1.8832,
      "step": 3750
    },
    {
      "epoch": 0.0987885172504609,
      "grad_norm": 2.120803117752075,
      "learning_rate": 4.506715828285489e-05,
      "loss": 1.3609,
      "step": 3751
    },
    {
      "epoch": 0.09881485383197261,
      "grad_norm": 4.1749348640441895,
      "learning_rate": 4.50658414537793e-05,
      "loss": 1.6664,
      "step": 3752
    },
    {
      "epoch": 0.09884119041348433,
      "grad_norm": 2.644705295562744,
      "learning_rate": 4.506452462470371e-05,
      "loss": 1.3371,
      "step": 3753
    },
    {
      "epoch": 0.09886752699499606,
      "grad_norm": 2.0972976684570312,
      "learning_rate": 4.506320779562813e-05,
      "loss": 1.8616,
      "step": 3754
    },
    {
      "epoch": 0.09889386357650777,
      "grad_norm": 2.455526113510132,
      "learning_rate": 4.5061890966552544e-05,
      "loss": 1.9646,
      "step": 3755
    },
    {
      "epoch": 0.09892020015801949,
      "grad_norm": 1.8271840810775757,
      "learning_rate": 4.506057413747696e-05,
      "loss": 0.5271,
      "step": 3756
    },
    {
      "epoch": 0.09894653673953122,
      "grad_norm": 2.690463066101074,
      "learning_rate": 4.505925730840137e-05,
      "loss": 2.1319,
      "step": 3757
    },
    {
      "epoch": 0.09897287332104293,
      "grad_norm": 2.247929573059082,
      "learning_rate": 4.5057940479325784e-05,
      "loss": 1.5843,
      "step": 3758
    },
    {
      "epoch": 0.09899920990255465,
      "grad_norm": 1.897557020187378,
      "learning_rate": 4.50566236502502e-05,
      "loss": 2.2191,
      "step": 3759
    },
    {
      "epoch": 0.09902554648406636,
      "grad_norm": 1.8032574653625488,
      "learning_rate": 4.5055306821174615e-05,
      "loss": 1.7499,
      "step": 3760
    },
    {
      "epoch": 0.09905188306557809,
      "grad_norm": 2.043447732925415,
      "learning_rate": 4.5053989992099024e-05,
      "loss": 1.8648,
      "step": 3761
    },
    {
      "epoch": 0.09907821964708981,
      "grad_norm": 2.2304294109344482,
      "learning_rate": 4.505267316302344e-05,
      "loss": 0.5406,
      "step": 3762
    },
    {
      "epoch": 0.09910455622860152,
      "grad_norm": 3.9684667587280273,
      "learning_rate": 4.5051356333947856e-05,
      "loss": 2.356,
      "step": 3763
    },
    {
      "epoch": 0.09913089281011325,
      "grad_norm": 2.1880557537078857,
      "learning_rate": 4.505003950487227e-05,
      "loss": 1.0215,
      "step": 3764
    },
    {
      "epoch": 0.09915722939162497,
      "grad_norm": 3.6301941871643066,
      "learning_rate": 4.504872267579669e-05,
      "loss": 1.4861,
      "step": 3765
    },
    {
      "epoch": 0.09918356597313668,
      "grad_norm": 5.493913173675537,
      "learning_rate": 4.5047405846721096e-05,
      "loss": 1.5048,
      "step": 3766
    },
    {
      "epoch": 0.09920990255464841,
      "grad_norm": 1.8711073398590088,
      "learning_rate": 4.504608901764551e-05,
      "loss": 1.7595,
      "step": 3767
    },
    {
      "epoch": 0.09923623913616013,
      "grad_norm": 3.579517126083374,
      "learning_rate": 4.504477218856993e-05,
      "loss": 1.4115,
      "step": 3768
    },
    {
      "epoch": 0.09926257571767184,
      "grad_norm": 2.168915033340454,
      "learning_rate": 4.504345535949434e-05,
      "loss": 1.7171,
      "step": 3769
    },
    {
      "epoch": 0.09928891229918356,
      "grad_norm": 3.1862964630126953,
      "learning_rate": 4.504213853041875e-05,
      "loss": 0.8614,
      "step": 3770
    },
    {
      "epoch": 0.09931524888069529,
      "grad_norm": 3.594576835632324,
      "learning_rate": 4.504082170134317e-05,
      "loss": 1.4088,
      "step": 3771
    },
    {
      "epoch": 0.099341585462207,
      "grad_norm": 3.186267375946045,
      "learning_rate": 4.503950487226758e-05,
      "loss": 1.3349,
      "step": 3772
    },
    {
      "epoch": 0.09936792204371872,
      "grad_norm": 2.0694220066070557,
      "learning_rate": 4.5038188043192e-05,
      "loss": 1.8724,
      "step": 3773
    },
    {
      "epoch": 0.09939425862523045,
      "grad_norm": 2.560370683670044,
      "learning_rate": 4.5036871214116414e-05,
      "loss": 2.3651,
      "step": 3774
    },
    {
      "epoch": 0.09942059520674217,
      "grad_norm": 1.939302682876587,
      "learning_rate": 4.503555438504082e-05,
      "loss": 1.5869,
      "step": 3775
    },
    {
      "epoch": 0.09944693178825388,
      "grad_norm": 2.949557065963745,
      "learning_rate": 4.503423755596524e-05,
      "loss": 2.1721,
      "step": 3776
    },
    {
      "epoch": 0.09947326836976561,
      "grad_norm": 3.7444915771484375,
      "learning_rate": 4.503292072688965e-05,
      "loss": 2.4473,
      "step": 3777
    },
    {
      "epoch": 0.09949960495127733,
      "grad_norm": 2.419501543045044,
      "learning_rate": 4.503160389781407e-05,
      "loss": 1.7612,
      "step": 3778
    },
    {
      "epoch": 0.09952594153278904,
      "grad_norm": 4.271428108215332,
      "learning_rate": 4.503028706873848e-05,
      "loss": 1.4787,
      "step": 3779
    },
    {
      "epoch": 0.09955227811430076,
      "grad_norm": 1.6048818826675415,
      "learning_rate": 4.5028970239662894e-05,
      "loss": 1.8313,
      "step": 3780
    },
    {
      "epoch": 0.09957861469581249,
      "grad_norm": 2.2156386375427246,
      "learning_rate": 4.502765341058731e-05,
      "loss": 0.429,
      "step": 3781
    },
    {
      "epoch": 0.0996049512773242,
      "grad_norm": 5.712406635284424,
      "learning_rate": 4.5026336581511725e-05,
      "loss": 2.3388,
      "step": 3782
    },
    {
      "epoch": 0.09963128785883592,
      "grad_norm": 4.480116367340088,
      "learning_rate": 4.502501975243614e-05,
      "loss": 1.572,
      "step": 3783
    },
    {
      "epoch": 0.09965762444034765,
      "grad_norm": 1.6718860864639282,
      "learning_rate": 4.502370292336055e-05,
      "loss": 1.8966,
      "step": 3784
    },
    {
      "epoch": 0.09968396102185936,
      "grad_norm": 3.7211596965789795,
      "learning_rate": 4.5022386094284965e-05,
      "loss": 1.313,
      "step": 3785
    },
    {
      "epoch": 0.09971029760337108,
      "grad_norm": 4.084120750427246,
      "learning_rate": 4.5021069265209374e-05,
      "loss": 0.7511,
      "step": 3786
    },
    {
      "epoch": 0.09973663418488281,
      "grad_norm": 8.82149887084961,
      "learning_rate": 4.5019752436133796e-05,
      "loss": 1.9127,
      "step": 3787
    },
    {
      "epoch": 0.09976297076639452,
      "grad_norm": 3.8063817024230957,
      "learning_rate": 4.5018435607058205e-05,
      "loss": 0.9581,
      "step": 3788
    },
    {
      "epoch": 0.09978930734790624,
      "grad_norm": 2.2536633014678955,
      "learning_rate": 4.501711877798262e-05,
      "loss": 1.6813,
      "step": 3789
    },
    {
      "epoch": 0.09981564392941795,
      "grad_norm": 3.1599223613739014,
      "learning_rate": 4.5015801948907036e-05,
      "loss": 0.4418,
      "step": 3790
    },
    {
      "epoch": 0.09984198051092968,
      "grad_norm": 1.986271858215332,
      "learning_rate": 4.5014485119831445e-05,
      "loss": 1.4102,
      "step": 3791
    },
    {
      "epoch": 0.0998683170924414,
      "grad_norm": 2.5842885971069336,
      "learning_rate": 4.501316829075587e-05,
      "loss": 2.2324,
      "step": 3792
    },
    {
      "epoch": 0.09989465367395312,
      "grad_norm": 3.0341603755950928,
      "learning_rate": 4.5011851461680276e-05,
      "loss": 1.2503,
      "step": 3793
    },
    {
      "epoch": 0.09992099025546484,
      "grad_norm": 1.8510781526565552,
      "learning_rate": 4.501053463260469e-05,
      "loss": 1.4495,
      "step": 3794
    },
    {
      "epoch": 0.09994732683697656,
      "grad_norm": 2.9165852069854736,
      "learning_rate": 4.50092178035291e-05,
      "loss": 2.8417,
      "step": 3795
    },
    {
      "epoch": 0.09997366341848828,
      "grad_norm": 2.408564567565918,
      "learning_rate": 4.500790097445352e-05,
      "loss": 2.2565,
      "step": 3796
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.188035726547241,
      "learning_rate": 4.500658414537793e-05,
      "loss": 1.5012,
      "step": 3797
    },
    {
      "epoch": 0.10002633658151172,
      "grad_norm": 4.150546550750732,
      "learning_rate": 4.500526731630235e-05,
      "loss": 2.3743,
      "step": 3798
    },
    {
      "epoch": 0.10005267316302344,
      "grad_norm": 3.530560255050659,
      "learning_rate": 4.5003950487226756e-05,
      "loss": 1.2454,
      "step": 3799
    },
    {
      "epoch": 0.10007900974453517,
      "grad_norm": 4.042998313903809,
      "learning_rate": 4.500263365815117e-05,
      "loss": 1.4332,
      "step": 3800
    },
    {
      "epoch": 0.10010534632604688,
      "grad_norm": 2.05966854095459,
      "learning_rate": 4.500131682907559e-05,
      "loss": 2.2651,
      "step": 3801
    },
    {
      "epoch": 0.1001316829075586,
      "grad_norm": 1.9672224521636963,
      "learning_rate": 4.5e-05,
      "loss": 1.5866,
      "step": 3802
    },
    {
      "epoch": 0.10015801948907031,
      "grad_norm": 2.380561351776123,
      "learning_rate": 4.499868317092442e-05,
      "loss": 1.4811,
      "step": 3803
    },
    {
      "epoch": 0.10018435607058204,
      "grad_norm": 2.6420717239379883,
      "learning_rate": 4.499736634184883e-05,
      "loss": 1.7474,
      "step": 3804
    },
    {
      "epoch": 0.10021069265209376,
      "grad_norm": 9.223584175109863,
      "learning_rate": 4.499604951277324e-05,
      "loss": 0.8603,
      "step": 3805
    },
    {
      "epoch": 0.10023702923360547,
      "grad_norm": 2.276423215866089,
      "learning_rate": 4.499473268369766e-05,
      "loss": 1.6951,
      "step": 3806
    },
    {
      "epoch": 0.1002633658151172,
      "grad_norm": 2.397859811782837,
      "learning_rate": 4.4993415854622074e-05,
      "loss": 1.2482,
      "step": 3807
    },
    {
      "epoch": 0.10028970239662892,
      "grad_norm": 3.1488842964172363,
      "learning_rate": 4.499209902554648e-05,
      "loss": 1.3942,
      "step": 3808
    },
    {
      "epoch": 0.10031603897814063,
      "grad_norm": 2.4266960620880127,
      "learning_rate": 4.49907821964709e-05,
      "loss": 2.1123,
      "step": 3809
    },
    {
      "epoch": 0.10034237555965236,
      "grad_norm": 2.485085964202881,
      "learning_rate": 4.4989465367395314e-05,
      "loss": 1.7803,
      "step": 3810
    },
    {
      "epoch": 0.10036871214116408,
      "grad_norm": 3.4337573051452637,
      "learning_rate": 4.498814853831973e-05,
      "loss": 0.5831,
      "step": 3811
    },
    {
      "epoch": 0.1003950487226758,
      "grad_norm": 3.493136405944824,
      "learning_rate": 4.4986831709244145e-05,
      "loss": 1.7557,
      "step": 3812
    },
    {
      "epoch": 0.10042138530418751,
      "grad_norm": 2.224128007888794,
      "learning_rate": 4.4985514880168554e-05,
      "loss": 2.2421,
      "step": 3813
    },
    {
      "epoch": 0.10044772188569924,
      "grad_norm": 1.8008248805999756,
      "learning_rate": 4.498419805109297e-05,
      "loss": 2.1127,
      "step": 3814
    },
    {
      "epoch": 0.10047405846721096,
      "grad_norm": 2.4031283855438232,
      "learning_rate": 4.4982881222017385e-05,
      "loss": 1.8835,
      "step": 3815
    },
    {
      "epoch": 0.10050039504872267,
      "grad_norm": 2.955813407897949,
      "learning_rate": 4.49815643929418e-05,
      "loss": 2.2879,
      "step": 3816
    },
    {
      "epoch": 0.1005267316302344,
      "grad_norm": 3.2041842937469482,
      "learning_rate": 4.498024756386621e-05,
      "loss": 1.4607,
      "step": 3817
    },
    {
      "epoch": 0.10055306821174612,
      "grad_norm": 3.0943050384521484,
      "learning_rate": 4.4978930734790625e-05,
      "loss": 1.6395,
      "step": 3818
    },
    {
      "epoch": 0.10057940479325783,
      "grad_norm": 2.2135558128356934,
      "learning_rate": 4.497761390571504e-05,
      "loss": 1.8168,
      "step": 3819
    },
    {
      "epoch": 0.10060574137476956,
      "grad_norm": 2.5748634338378906,
      "learning_rate": 4.4976297076639456e-05,
      "loss": 1.5128,
      "step": 3820
    },
    {
      "epoch": 0.10063207795628128,
      "grad_norm": 4.733027935028076,
      "learning_rate": 4.497498024756387e-05,
      "loss": 0.8025,
      "step": 3821
    },
    {
      "epoch": 0.10065841453779299,
      "grad_norm": 2.3711979389190674,
      "learning_rate": 4.497366341848828e-05,
      "loss": 2.2044,
      "step": 3822
    },
    {
      "epoch": 0.10068475111930471,
      "grad_norm": 2.8120198249816895,
      "learning_rate": 4.4972346589412697e-05,
      "loss": 1.6934,
      "step": 3823
    },
    {
      "epoch": 0.10071108770081644,
      "grad_norm": 2.93912935256958,
      "learning_rate": 4.4971029760337105e-05,
      "loss": 1.3493,
      "step": 3824
    },
    {
      "epoch": 0.10073742428232815,
      "grad_norm": 2.880110502243042,
      "learning_rate": 4.496971293126153e-05,
      "loss": 1.2028,
      "step": 3825
    },
    {
      "epoch": 0.10076376086383987,
      "grad_norm": 2.13470196723938,
      "learning_rate": 4.4968396102185937e-05,
      "loss": 1.8668,
      "step": 3826
    },
    {
      "epoch": 0.1007900974453516,
      "grad_norm": 3.4758222103118896,
      "learning_rate": 4.496707927311035e-05,
      "loss": 2.0979,
      "step": 3827
    },
    {
      "epoch": 0.10081643402686331,
      "grad_norm": 2.111302375793457,
      "learning_rate": 4.496576244403477e-05,
      "loss": 1.6521,
      "step": 3828
    },
    {
      "epoch": 0.10084277060837503,
      "grad_norm": 2.630613327026367,
      "learning_rate": 4.496444561495918e-05,
      "loss": 1.3449,
      "step": 3829
    },
    {
      "epoch": 0.10086910718988676,
      "grad_norm": 5.345747470855713,
      "learning_rate": 4.49631287858836e-05,
      "loss": 1.1698,
      "step": 3830
    },
    {
      "epoch": 0.10089544377139847,
      "grad_norm": 3.3519644737243652,
      "learning_rate": 4.496181195680801e-05,
      "loss": 0.7235,
      "step": 3831
    },
    {
      "epoch": 0.10092178035291019,
      "grad_norm": 1.7442739009857178,
      "learning_rate": 4.496049512773242e-05,
      "loss": 1.81,
      "step": 3832
    },
    {
      "epoch": 0.1009481169344219,
      "grad_norm": 2.1400156021118164,
      "learning_rate": 4.495917829865683e-05,
      "loss": 1.7726,
      "step": 3833
    },
    {
      "epoch": 0.10097445351593363,
      "grad_norm": 2.0222418308258057,
      "learning_rate": 4.4957861469581255e-05,
      "loss": 2.4957,
      "step": 3834
    },
    {
      "epoch": 0.10100079009744535,
      "grad_norm": 2.494149923324585,
      "learning_rate": 4.495654464050566e-05,
      "loss": 2.0013,
      "step": 3835
    },
    {
      "epoch": 0.10102712667895707,
      "grad_norm": 4.976408958435059,
      "learning_rate": 4.495522781143008e-05,
      "loss": 1.5433,
      "step": 3836
    },
    {
      "epoch": 0.1010534632604688,
      "grad_norm": 1.8104454278945923,
      "learning_rate": 4.4953910982354495e-05,
      "loss": 1.3188,
      "step": 3837
    },
    {
      "epoch": 0.10107979984198051,
      "grad_norm": 1.821811318397522,
      "learning_rate": 4.49525941532789e-05,
      "loss": 1.5422,
      "step": 3838
    },
    {
      "epoch": 0.10110613642349223,
      "grad_norm": 2.410536289215088,
      "learning_rate": 4.4951277324203326e-05,
      "loss": 1.7623,
      "step": 3839
    },
    {
      "epoch": 0.10113247300500396,
      "grad_norm": 4.866095542907715,
      "learning_rate": 4.4949960495127735e-05,
      "loss": 2.2759,
      "step": 3840
    },
    {
      "epoch": 0.10115880958651567,
      "grad_norm": 2.438474655151367,
      "learning_rate": 4.494864366605215e-05,
      "loss": 1.7839,
      "step": 3841
    },
    {
      "epoch": 0.10118514616802739,
      "grad_norm": 2.5334908962249756,
      "learning_rate": 4.494732683697656e-05,
      "loss": 2.0451,
      "step": 3842
    },
    {
      "epoch": 0.10121148274953912,
      "grad_norm": 1.5801764726638794,
      "learning_rate": 4.494601000790098e-05,
      "loss": 1.6087,
      "step": 3843
    },
    {
      "epoch": 0.10123781933105083,
      "grad_norm": 2.1839675903320312,
      "learning_rate": 4.494469317882539e-05,
      "loss": 2.2102,
      "step": 3844
    },
    {
      "epoch": 0.10126415591256255,
      "grad_norm": 3.1293671131134033,
      "learning_rate": 4.4943376349749806e-05,
      "loss": 1.4554,
      "step": 3845
    },
    {
      "epoch": 0.10129049249407426,
      "grad_norm": 2.179328441619873,
      "learning_rate": 4.4942059520674215e-05,
      "loss": 1.833,
      "step": 3846
    },
    {
      "epoch": 0.10131682907558599,
      "grad_norm": 1.850671410560608,
      "learning_rate": 4.494074269159863e-05,
      "loss": 2.6449,
      "step": 3847
    },
    {
      "epoch": 0.10134316565709771,
      "grad_norm": 4.069072723388672,
      "learning_rate": 4.4939425862523046e-05,
      "loss": 1.3082,
      "step": 3848
    },
    {
      "epoch": 0.10136950223860942,
      "grad_norm": 3.0500688552856445,
      "learning_rate": 4.493810903344746e-05,
      "loss": 1.5405,
      "step": 3849
    },
    {
      "epoch": 0.10139583882012115,
      "grad_norm": 2.249415397644043,
      "learning_rate": 4.493679220437188e-05,
      "loss": 2.0155,
      "step": 3850
    },
    {
      "epoch": 0.10142217540163287,
      "grad_norm": 2.2017037868499756,
      "learning_rate": 4.4935475375296286e-05,
      "loss": 1.9497,
      "step": 3851
    },
    {
      "epoch": 0.10144851198314458,
      "grad_norm": 3.5100765228271484,
      "learning_rate": 4.49341585462207e-05,
      "loss": 1.483,
      "step": 3852
    },
    {
      "epoch": 0.10147484856465631,
      "grad_norm": 2.7092349529266357,
      "learning_rate": 4.493284171714512e-05,
      "loss": 2.5334,
      "step": 3853
    },
    {
      "epoch": 0.10150118514616803,
      "grad_norm": 2.613349676132202,
      "learning_rate": 4.493152488806953e-05,
      "loss": 1.9086,
      "step": 3854
    },
    {
      "epoch": 0.10152752172767975,
      "grad_norm": 3.615673780441284,
      "learning_rate": 4.493020805899394e-05,
      "loss": 1.4175,
      "step": 3855
    },
    {
      "epoch": 0.10155385830919146,
      "grad_norm": 2.193631887435913,
      "learning_rate": 4.492889122991836e-05,
      "loss": 1.7111,
      "step": 3856
    },
    {
      "epoch": 0.10158019489070319,
      "grad_norm": 1.9345149993896484,
      "learning_rate": 4.492757440084277e-05,
      "loss": 0.6001,
      "step": 3857
    },
    {
      "epoch": 0.1016065314722149,
      "grad_norm": 1.9012631177902222,
      "learning_rate": 4.492625757176719e-05,
      "loss": 1.5349,
      "step": 3858
    },
    {
      "epoch": 0.10163286805372662,
      "grad_norm": 2.148568630218506,
      "learning_rate": 4.4924940742691604e-05,
      "loss": 2.1311,
      "step": 3859
    },
    {
      "epoch": 0.10165920463523835,
      "grad_norm": 3.65657639503479,
      "learning_rate": 4.492362391361601e-05,
      "loss": 0.6288,
      "step": 3860
    },
    {
      "epoch": 0.10168554121675007,
      "grad_norm": 2.1184465885162354,
      "learning_rate": 4.492230708454043e-05,
      "loss": 2.0239,
      "step": 3861
    },
    {
      "epoch": 0.10171187779826178,
      "grad_norm": 2.5524072647094727,
      "learning_rate": 4.4920990255464844e-05,
      "loss": 1.528,
      "step": 3862
    },
    {
      "epoch": 0.10173821437977351,
      "grad_norm": 2.9298508167266846,
      "learning_rate": 4.491967342638926e-05,
      "loss": 1.4404,
      "step": 3863
    },
    {
      "epoch": 0.10176455096128523,
      "grad_norm": 2.2487053871154785,
      "learning_rate": 4.491835659731367e-05,
      "loss": 0.1919,
      "step": 3864
    },
    {
      "epoch": 0.10179088754279694,
      "grad_norm": 5.0985870361328125,
      "learning_rate": 4.4917039768238084e-05,
      "loss": 0.9194,
      "step": 3865
    },
    {
      "epoch": 0.10181722412430866,
      "grad_norm": 2.198937177658081,
      "learning_rate": 4.49157229391625e-05,
      "loss": 1.7879,
      "step": 3866
    },
    {
      "epoch": 0.10184356070582039,
      "grad_norm": 2.556926727294922,
      "learning_rate": 4.4914406110086915e-05,
      "loss": 1.6204,
      "step": 3867
    },
    {
      "epoch": 0.1018698972873321,
      "grad_norm": 1.9259483814239502,
      "learning_rate": 4.491308928101133e-05,
      "loss": 1.5702,
      "step": 3868
    },
    {
      "epoch": 0.10189623386884382,
      "grad_norm": 2.51617169380188,
      "learning_rate": 4.491177245193574e-05,
      "loss": 1.7788,
      "step": 3869
    },
    {
      "epoch": 0.10192257045035555,
      "grad_norm": 2.9019148349761963,
      "learning_rate": 4.4910455622860155e-05,
      "loss": 1.4209,
      "step": 3870
    },
    {
      "epoch": 0.10194890703186726,
      "grad_norm": 2.2281227111816406,
      "learning_rate": 4.4909138793784564e-05,
      "loss": 2.0085,
      "step": 3871
    },
    {
      "epoch": 0.10197524361337898,
      "grad_norm": 1.8960057497024536,
      "learning_rate": 4.4907821964708986e-05,
      "loss": 1.9725,
      "step": 3872
    },
    {
      "epoch": 0.10200158019489071,
      "grad_norm": 1.6910125017166138,
      "learning_rate": 4.4906505135633395e-05,
      "loss": 2.1222,
      "step": 3873
    },
    {
      "epoch": 0.10202791677640242,
      "grad_norm": 2.12577223777771,
      "learning_rate": 4.490518830655781e-05,
      "loss": 2.1383,
      "step": 3874
    },
    {
      "epoch": 0.10205425335791414,
      "grad_norm": 2.494915246963501,
      "learning_rate": 4.4903871477482226e-05,
      "loss": 1.8938,
      "step": 3875
    },
    {
      "epoch": 0.10208058993942586,
      "grad_norm": 1.9448983669281006,
      "learning_rate": 4.490255464840664e-05,
      "loss": 1.7263,
      "step": 3876
    },
    {
      "epoch": 0.10210692652093759,
      "grad_norm": 2.9931201934814453,
      "learning_rate": 4.490123781933106e-05,
      "loss": 0.954,
      "step": 3877
    },
    {
      "epoch": 0.1021332631024493,
      "grad_norm": 3.022486925125122,
      "learning_rate": 4.4899920990255466e-05,
      "loss": 2.2079,
      "step": 3878
    },
    {
      "epoch": 0.10215959968396102,
      "grad_norm": 4.349597454071045,
      "learning_rate": 4.489860416117988e-05,
      "loss": 1.0891,
      "step": 3879
    },
    {
      "epoch": 0.10218593626547275,
      "grad_norm": 2.0917694568634033,
      "learning_rate": 4.489728733210429e-05,
      "loss": 1.8808,
      "step": 3880
    },
    {
      "epoch": 0.10221227284698446,
      "grad_norm": 2.205721855163574,
      "learning_rate": 4.489597050302871e-05,
      "loss": 1.5352,
      "step": 3881
    },
    {
      "epoch": 0.10223860942849618,
      "grad_norm": 3.1652493476867676,
      "learning_rate": 4.489465367395312e-05,
      "loss": 1.918,
      "step": 3882
    },
    {
      "epoch": 0.1022649460100079,
      "grad_norm": 6.0257792472839355,
      "learning_rate": 4.489333684487754e-05,
      "loss": 2.119,
      "step": 3883
    },
    {
      "epoch": 0.10229128259151962,
      "grad_norm": 2.2120919227600098,
      "learning_rate": 4.489202001580195e-05,
      "loss": 1.2787,
      "step": 3884
    },
    {
      "epoch": 0.10231761917303134,
      "grad_norm": 3.2696492671966553,
      "learning_rate": 4.489070318672636e-05,
      "loss": 0.8269,
      "step": 3885
    },
    {
      "epoch": 0.10234395575454305,
      "grad_norm": 1.874639868736267,
      "learning_rate": 4.4889386357650784e-05,
      "loss": 1.6064,
      "step": 3886
    },
    {
      "epoch": 0.10237029233605478,
      "grad_norm": 1.9932085275650024,
      "learning_rate": 4.488806952857519e-05,
      "loss": 2.0404,
      "step": 3887
    },
    {
      "epoch": 0.1023966289175665,
      "grad_norm": 2.500261068344116,
      "learning_rate": 4.488675269949961e-05,
      "loss": 1.0359,
      "step": 3888
    },
    {
      "epoch": 0.10242296549907821,
      "grad_norm": 2.3774125576019287,
      "learning_rate": 4.488543587042402e-05,
      "loss": 2.052,
      "step": 3889
    },
    {
      "epoch": 0.10244930208058994,
      "grad_norm": 2.0700395107269287,
      "learning_rate": 4.488411904134843e-05,
      "loss": 1.8927,
      "step": 3890
    },
    {
      "epoch": 0.10247563866210166,
      "grad_norm": 1.9385424852371216,
      "learning_rate": 4.488280221227285e-05,
      "loss": 2.1316,
      "step": 3891
    },
    {
      "epoch": 0.10250197524361337,
      "grad_norm": 2.2179131507873535,
      "learning_rate": 4.4881485383197264e-05,
      "loss": 1.596,
      "step": 3892
    },
    {
      "epoch": 0.1025283118251251,
      "grad_norm": 1.9597073793411255,
      "learning_rate": 4.488016855412167e-05,
      "loss": 2.0666,
      "step": 3893
    },
    {
      "epoch": 0.10255464840663682,
      "grad_norm": 3.590493679046631,
      "learning_rate": 4.487885172504609e-05,
      "loss": 0.797,
      "step": 3894
    },
    {
      "epoch": 0.10258098498814854,
      "grad_norm": 3.5431103706359863,
      "learning_rate": 4.487753489597051e-05,
      "loss": 0.9543,
      "step": 3895
    },
    {
      "epoch": 0.10260732156966026,
      "grad_norm": 3.236133098602295,
      "learning_rate": 4.487621806689492e-05,
      "loss": 1.0766,
      "step": 3896
    },
    {
      "epoch": 0.10263365815117198,
      "grad_norm": 5.906928539276123,
      "learning_rate": 4.4874901237819336e-05,
      "loss": 1.1328,
      "step": 3897
    },
    {
      "epoch": 0.1026599947326837,
      "grad_norm": 2.431095600128174,
      "learning_rate": 4.4873584408743744e-05,
      "loss": 1.3497,
      "step": 3898
    },
    {
      "epoch": 0.10268633131419541,
      "grad_norm": 2.8399951457977295,
      "learning_rate": 4.487226757966816e-05,
      "loss": 1.8871,
      "step": 3899
    },
    {
      "epoch": 0.10271266789570714,
      "grad_norm": 2.45115065574646,
      "learning_rate": 4.4870950750592576e-05,
      "loss": 1.7018,
      "step": 3900
    },
    {
      "epoch": 0.10273900447721886,
      "grad_norm": 2.5587565898895264,
      "learning_rate": 4.486963392151699e-05,
      "loss": 1.5922,
      "step": 3901
    },
    {
      "epoch": 0.10276534105873057,
      "grad_norm": 3.362809181213379,
      "learning_rate": 4.48683170924414e-05,
      "loss": 1.3428,
      "step": 3902
    },
    {
      "epoch": 0.1027916776402423,
      "grad_norm": 3.032858371734619,
      "learning_rate": 4.4867000263365816e-05,
      "loss": 1.6904,
      "step": 3903
    },
    {
      "epoch": 0.10281801422175402,
      "grad_norm": 3.6812729835510254,
      "learning_rate": 4.486568343429023e-05,
      "loss": 1.4182,
      "step": 3904
    },
    {
      "epoch": 0.10284435080326573,
      "grad_norm": 2.6831905841827393,
      "learning_rate": 4.486436660521465e-05,
      "loss": 1.6019,
      "step": 3905
    },
    {
      "epoch": 0.10287068738477746,
      "grad_norm": 1.8663004636764526,
      "learning_rate": 4.486304977613906e-05,
      "loss": 2.0442,
      "step": 3906
    },
    {
      "epoch": 0.10289702396628918,
      "grad_norm": 2.102212905883789,
      "learning_rate": 4.486173294706347e-05,
      "loss": 1.786,
      "step": 3907
    },
    {
      "epoch": 0.1029233605478009,
      "grad_norm": 4.404050350189209,
      "learning_rate": 4.486041611798789e-05,
      "loss": 1.5842,
      "step": 3908
    },
    {
      "epoch": 0.10294969712931261,
      "grad_norm": 7.1329755783081055,
      "learning_rate": 4.48590992889123e-05,
      "loss": 2.2315,
      "step": 3909
    },
    {
      "epoch": 0.10297603371082434,
      "grad_norm": 2.8774871826171875,
      "learning_rate": 4.485778245983672e-05,
      "loss": 1.2062,
      "step": 3910
    },
    {
      "epoch": 0.10300237029233605,
      "grad_norm": 1.801954984664917,
      "learning_rate": 4.485646563076113e-05,
      "loss": 1.6084,
      "step": 3911
    },
    {
      "epoch": 0.10302870687384777,
      "grad_norm": 2.9546315670013428,
      "learning_rate": 4.485514880168554e-05,
      "loss": 0.5987,
      "step": 3912
    },
    {
      "epoch": 0.1030550434553595,
      "grad_norm": 1.9714163541793823,
      "learning_rate": 4.485383197260996e-05,
      "loss": 1.8321,
      "step": 3913
    },
    {
      "epoch": 0.10308138003687121,
      "grad_norm": 1.6383371353149414,
      "learning_rate": 4.4852515143534374e-05,
      "loss": 1.4924,
      "step": 3914
    },
    {
      "epoch": 0.10310771661838293,
      "grad_norm": 3.2470390796661377,
      "learning_rate": 4.485119831445879e-05,
      "loss": 0.6634,
      "step": 3915
    },
    {
      "epoch": 0.10313405319989466,
      "grad_norm": 2.703115463256836,
      "learning_rate": 4.48498814853832e-05,
      "loss": 1.5607,
      "step": 3916
    },
    {
      "epoch": 0.10316038978140638,
      "grad_norm": 4.0662994384765625,
      "learning_rate": 4.4848564656307614e-05,
      "loss": 2.0106,
      "step": 3917
    },
    {
      "epoch": 0.10318672636291809,
      "grad_norm": 3.7793779373168945,
      "learning_rate": 4.484724782723202e-05,
      "loss": 2.5227,
      "step": 3918
    },
    {
      "epoch": 0.1032130629444298,
      "grad_norm": 2.2200205326080322,
      "learning_rate": 4.4845930998156445e-05,
      "loss": 1.8839,
      "step": 3919
    },
    {
      "epoch": 0.10323939952594154,
      "grad_norm": 1.6686019897460938,
      "learning_rate": 4.4844614169080854e-05,
      "loss": 1.5918,
      "step": 3920
    },
    {
      "epoch": 0.10326573610745325,
      "grad_norm": 2.117866039276123,
      "learning_rate": 4.484329734000527e-05,
      "loss": 1.644,
      "step": 3921
    },
    {
      "epoch": 0.10329207268896497,
      "grad_norm": 2.207977533340454,
      "learning_rate": 4.4841980510929685e-05,
      "loss": 1.5119,
      "step": 3922
    },
    {
      "epoch": 0.1033184092704767,
      "grad_norm": 2.9472129344940186,
      "learning_rate": 4.4840663681854094e-05,
      "loss": 1.8325,
      "step": 3923
    },
    {
      "epoch": 0.10334474585198841,
      "grad_norm": 2.2142364978790283,
      "learning_rate": 4.4839346852778516e-05,
      "loss": 2.0007,
      "step": 3924
    },
    {
      "epoch": 0.10337108243350013,
      "grad_norm": 1.6783168315887451,
      "learning_rate": 4.4838030023702925e-05,
      "loss": 1.7686,
      "step": 3925
    },
    {
      "epoch": 0.10339741901501186,
      "grad_norm": 1.9474951028823853,
      "learning_rate": 4.483671319462734e-05,
      "loss": 1.6394,
      "step": 3926
    },
    {
      "epoch": 0.10342375559652357,
      "grad_norm": 2.798863172531128,
      "learning_rate": 4.483539636555175e-05,
      "loss": 1.4046,
      "step": 3927
    },
    {
      "epoch": 0.10345009217803529,
      "grad_norm": 4.639814376831055,
      "learning_rate": 4.483407953647617e-05,
      "loss": 0.7852,
      "step": 3928
    },
    {
      "epoch": 0.103476428759547,
      "grad_norm": 1.954859972000122,
      "learning_rate": 4.483276270740058e-05,
      "loss": 1.6035,
      "step": 3929
    },
    {
      "epoch": 0.10350276534105873,
      "grad_norm": 1.8147886991500854,
      "learning_rate": 4.4831445878324996e-05,
      "loss": 1.994,
      "step": 3930
    },
    {
      "epoch": 0.10352910192257045,
      "grad_norm": 3.092109441757202,
      "learning_rate": 4.483012904924941e-05,
      "loss": 1.8483,
      "step": 3931
    },
    {
      "epoch": 0.10355543850408216,
      "grad_norm": 4.166642189025879,
      "learning_rate": 4.482881222017382e-05,
      "loss": 1.3048,
      "step": 3932
    },
    {
      "epoch": 0.1035817750855939,
      "grad_norm": 1.8559051752090454,
      "learning_rate": 4.482749539109824e-05,
      "loss": 1.5706,
      "step": 3933
    },
    {
      "epoch": 0.10360811166710561,
      "grad_norm": 2.076110601425171,
      "learning_rate": 4.482617856202265e-05,
      "loss": 1.1648,
      "step": 3934
    },
    {
      "epoch": 0.10363444824861733,
      "grad_norm": 3.1907925605773926,
      "learning_rate": 4.482486173294707e-05,
      "loss": 1.7439,
      "step": 3935
    },
    {
      "epoch": 0.10366078483012905,
      "grad_norm": 2.254758596420288,
      "learning_rate": 4.4823544903871476e-05,
      "loss": 1.3704,
      "step": 3936
    },
    {
      "epoch": 0.10368712141164077,
      "grad_norm": 2.8785667419433594,
      "learning_rate": 4.482222807479589e-05,
      "loss": 1.7097,
      "step": 3937
    },
    {
      "epoch": 0.10371345799315249,
      "grad_norm": 2.2197372913360596,
      "learning_rate": 4.482091124572031e-05,
      "loss": 1.4248,
      "step": 3938
    },
    {
      "epoch": 0.10373979457466422,
      "grad_norm": 2.4855663776397705,
      "learning_rate": 4.481959441664472e-05,
      "loss": 1.6056,
      "step": 3939
    },
    {
      "epoch": 0.10376613115617593,
      "grad_norm": 3.4125561714172363,
      "learning_rate": 4.481827758756914e-05,
      "loss": 1.2521,
      "step": 3940
    },
    {
      "epoch": 0.10379246773768765,
      "grad_norm": 3.5316920280456543,
      "learning_rate": 4.481696075849355e-05,
      "loss": 1.2078,
      "step": 3941
    },
    {
      "epoch": 0.10381880431919936,
      "grad_norm": 1.9573456048965454,
      "learning_rate": 4.481564392941797e-05,
      "loss": 2.1511,
      "step": 3942
    },
    {
      "epoch": 0.10384514090071109,
      "grad_norm": 2.3475427627563477,
      "learning_rate": 4.481432710034238e-05,
      "loss": 1.44,
      "step": 3943
    },
    {
      "epoch": 0.10387147748222281,
      "grad_norm": 1.5809345245361328,
      "learning_rate": 4.4813010271266794e-05,
      "loss": 1.5978,
      "step": 3944
    },
    {
      "epoch": 0.10389781406373452,
      "grad_norm": 3.8339791297912598,
      "learning_rate": 4.48116934421912e-05,
      "loss": 2.7539,
      "step": 3945
    },
    {
      "epoch": 0.10392415064524625,
      "grad_norm": 1.9423012733459473,
      "learning_rate": 4.481037661311562e-05,
      "loss": 0.798,
      "step": 3946
    },
    {
      "epoch": 0.10395048722675797,
      "grad_norm": 4.6782145500183105,
      "learning_rate": 4.4809059784040034e-05,
      "loss": 1.3771,
      "step": 3947
    },
    {
      "epoch": 0.10397682380826968,
      "grad_norm": 2.536159038543701,
      "learning_rate": 4.480774295496445e-05,
      "loss": 1.1967,
      "step": 3948
    },
    {
      "epoch": 0.10400316038978141,
      "grad_norm": 3.16945481300354,
      "learning_rate": 4.480642612588886e-05,
      "loss": 2.148,
      "step": 3949
    },
    {
      "epoch": 0.10402949697129313,
      "grad_norm": 2.1217691898345947,
      "learning_rate": 4.4805109296813274e-05,
      "loss": 2.1199,
      "step": 3950
    },
    {
      "epoch": 0.10405583355280484,
      "grad_norm": 6.577202320098877,
      "learning_rate": 4.480379246773769e-05,
      "loss": 1.214,
      "step": 3951
    },
    {
      "epoch": 0.10408217013431656,
      "grad_norm": 3.3185997009277344,
      "learning_rate": 4.4802475638662105e-05,
      "loss": 1.468,
      "step": 3952
    },
    {
      "epoch": 0.10410850671582829,
      "grad_norm": 1.7041481733322144,
      "learning_rate": 4.480115880958652e-05,
      "loss": 2.129,
      "step": 3953
    },
    {
      "epoch": 0.10413484329734,
      "grad_norm": 1.6504563093185425,
      "learning_rate": 4.479984198051093e-05,
      "loss": 1.8116,
      "step": 3954
    },
    {
      "epoch": 0.10416117987885172,
      "grad_norm": 1.7092504501342773,
      "learning_rate": 4.4798525151435345e-05,
      "loss": 1.6747,
      "step": 3955
    },
    {
      "epoch": 0.10418751646036345,
      "grad_norm": 4.163072109222412,
      "learning_rate": 4.4797208322359754e-05,
      "loss": 1.1396,
      "step": 3956
    },
    {
      "epoch": 0.10421385304187517,
      "grad_norm": 3.019253730773926,
      "learning_rate": 4.4795891493284177e-05,
      "loss": 0.8102,
      "step": 3957
    },
    {
      "epoch": 0.10424018962338688,
      "grad_norm": 1.8424838781356812,
      "learning_rate": 4.4794574664208585e-05,
      "loss": 2.2282,
      "step": 3958
    },
    {
      "epoch": 0.10426652620489861,
      "grad_norm": 2.632228374481201,
      "learning_rate": 4.4793257835133e-05,
      "loss": 1.8617,
      "step": 3959
    },
    {
      "epoch": 0.10429286278641033,
      "grad_norm": 2.02886700630188,
      "learning_rate": 4.4791941006057417e-05,
      "loss": 1.7877,
      "step": 3960
    },
    {
      "epoch": 0.10431919936792204,
      "grad_norm": 1.9014345407485962,
      "learning_rate": 4.479062417698183e-05,
      "loss": 1.3873,
      "step": 3961
    },
    {
      "epoch": 0.10434553594943376,
      "grad_norm": 2.2207207679748535,
      "learning_rate": 4.478930734790625e-05,
      "loss": 1.9763,
      "step": 3962
    },
    {
      "epoch": 0.10437187253094549,
      "grad_norm": 3.94708514213562,
      "learning_rate": 4.4787990518830657e-05,
      "loss": 1.4662,
      "step": 3963
    },
    {
      "epoch": 0.1043982091124572,
      "grad_norm": 4.338469982147217,
      "learning_rate": 4.478667368975507e-05,
      "loss": 1.4012,
      "step": 3964
    },
    {
      "epoch": 0.10442454569396892,
      "grad_norm": 2.5421175956726074,
      "learning_rate": 4.478535686067948e-05,
      "loss": 1.165,
      "step": 3965
    },
    {
      "epoch": 0.10445088227548065,
      "grad_norm": 5.195154190063477,
      "learning_rate": 4.47840400316039e-05,
      "loss": 0.9609,
      "step": 3966
    },
    {
      "epoch": 0.10447721885699236,
      "grad_norm": 1.7909053564071655,
      "learning_rate": 4.478272320252831e-05,
      "loss": 2.2669,
      "step": 3967
    },
    {
      "epoch": 0.10450355543850408,
      "grad_norm": 2.0777971744537354,
      "learning_rate": 4.478140637345273e-05,
      "loss": 1.4504,
      "step": 3968
    },
    {
      "epoch": 0.10452989202001581,
      "grad_norm": 1.5917402505874634,
      "learning_rate": 4.478008954437714e-05,
      "loss": 1.138,
      "step": 3969
    },
    {
      "epoch": 0.10455622860152752,
      "grad_norm": 6.32620906829834,
      "learning_rate": 4.477877271530155e-05,
      "loss": 1.4478,
      "step": 3970
    },
    {
      "epoch": 0.10458256518303924,
      "grad_norm": 2.0029797554016113,
      "learning_rate": 4.4777455886225975e-05,
      "loss": 2.1641,
      "step": 3971
    },
    {
      "epoch": 0.10460890176455095,
      "grad_norm": 2.4755024909973145,
      "learning_rate": 4.4776139057150383e-05,
      "loss": 0.8509,
      "step": 3972
    },
    {
      "epoch": 0.10463523834606268,
      "grad_norm": 2.942523956298828,
      "learning_rate": 4.47748222280748e-05,
      "loss": 1.3487,
      "step": 3973
    },
    {
      "epoch": 0.1046615749275744,
      "grad_norm": 3.0445621013641357,
      "learning_rate": 4.477350539899921e-05,
      "loss": 1.497,
      "step": 3974
    },
    {
      "epoch": 0.10468791150908612,
      "grad_norm": 4.136894226074219,
      "learning_rate": 4.477218856992363e-05,
      "loss": 0.9682,
      "step": 3975
    },
    {
      "epoch": 0.10471424809059784,
      "grad_norm": 1.9515690803527832,
      "learning_rate": 4.477087174084804e-05,
      "loss": 1.8275,
      "step": 3976
    },
    {
      "epoch": 0.10474058467210956,
      "grad_norm": 2.0134923458099365,
      "learning_rate": 4.4769554911772455e-05,
      "loss": 1.7219,
      "step": 3977
    },
    {
      "epoch": 0.10476692125362128,
      "grad_norm": 2.482132911682129,
      "learning_rate": 4.476823808269687e-05,
      "loss": 2.0579,
      "step": 3978
    },
    {
      "epoch": 0.104793257835133,
      "grad_norm": 4.157577037811279,
      "learning_rate": 4.476692125362128e-05,
      "loss": 2.6826,
      "step": 3979
    },
    {
      "epoch": 0.10481959441664472,
      "grad_norm": 1.9987140893936157,
      "learning_rate": 4.47656044245457e-05,
      "loss": 2.4325,
      "step": 3980
    },
    {
      "epoch": 0.10484593099815644,
      "grad_norm": 5.881584167480469,
      "learning_rate": 4.476428759547011e-05,
      "loss": 1.5879,
      "step": 3981
    },
    {
      "epoch": 0.10487226757966817,
      "grad_norm": 1.9028657674789429,
      "learning_rate": 4.4762970766394526e-05,
      "loss": 0.3889,
      "step": 3982
    },
    {
      "epoch": 0.10489860416117988,
      "grad_norm": 2.433701276779175,
      "learning_rate": 4.4761653937318935e-05,
      "loss": 2.031,
      "step": 3983
    },
    {
      "epoch": 0.1049249407426916,
      "grad_norm": 2.0801424980163574,
      "learning_rate": 4.476033710824335e-05,
      "loss": 1.8975,
      "step": 3984
    },
    {
      "epoch": 0.10495127732420331,
      "grad_norm": 2.2493185997009277,
      "learning_rate": 4.4759020279167766e-05,
      "loss": 1.2808,
      "step": 3985
    },
    {
      "epoch": 0.10497761390571504,
      "grad_norm": 1.7209901809692383,
      "learning_rate": 4.475770345009218e-05,
      "loss": 2.4493,
      "step": 3986
    },
    {
      "epoch": 0.10500395048722676,
      "grad_norm": 4.634757041931152,
      "learning_rate": 4.47563866210166e-05,
      "loss": 1.1222,
      "step": 3987
    },
    {
      "epoch": 0.10503028706873847,
      "grad_norm": 2.6888427734375,
      "learning_rate": 4.4755069791941006e-05,
      "loss": 0.4693,
      "step": 3988
    },
    {
      "epoch": 0.1050566236502502,
      "grad_norm": 3.5422425270080566,
      "learning_rate": 4.475375296286542e-05,
      "loss": 1.7103,
      "step": 3989
    },
    {
      "epoch": 0.10508296023176192,
      "grad_norm": 2.344755172729492,
      "learning_rate": 4.475243613378984e-05,
      "loss": 0.7759,
      "step": 3990
    },
    {
      "epoch": 0.10510929681327363,
      "grad_norm": 2.7901930809020996,
      "learning_rate": 4.475111930471425e-05,
      "loss": 2.1823,
      "step": 3991
    },
    {
      "epoch": 0.10513563339478536,
      "grad_norm": 4.578499794006348,
      "learning_rate": 4.474980247563866e-05,
      "loss": 1.5064,
      "step": 3992
    },
    {
      "epoch": 0.10516196997629708,
      "grad_norm": 2.010589838027954,
      "learning_rate": 4.474848564656308e-05,
      "loss": 1.8707,
      "step": 3993
    },
    {
      "epoch": 0.1051883065578088,
      "grad_norm": 3.7168216705322266,
      "learning_rate": 4.474716881748749e-05,
      "loss": 1.6647,
      "step": 3994
    },
    {
      "epoch": 0.10521464313932051,
      "grad_norm": 2.7797844409942627,
      "learning_rate": 4.474585198841191e-05,
      "loss": 2.1472,
      "step": 3995
    },
    {
      "epoch": 0.10524097972083224,
      "grad_norm": 2.030571222305298,
      "learning_rate": 4.474453515933632e-05,
      "loss": 2.2849,
      "step": 3996
    },
    {
      "epoch": 0.10526731630234396,
      "grad_norm": 1.6757009029388428,
      "learning_rate": 4.474321833026073e-05,
      "loss": 1.708,
      "step": 3997
    },
    {
      "epoch": 0.10529365288385567,
      "grad_norm": 2.0670464038848877,
      "learning_rate": 4.474190150118515e-05,
      "loss": 1.8766,
      "step": 3998
    },
    {
      "epoch": 0.1053199894653674,
      "grad_norm": 3.673553466796875,
      "learning_rate": 4.4740584672109564e-05,
      "loss": 1.6483,
      "step": 3999
    },
    {
      "epoch": 0.10534632604687912,
      "grad_norm": 1.9316781759262085,
      "learning_rate": 4.473926784303398e-05,
      "loss": 1.5283,
      "step": 4000
    },
    {
      "epoch": 0.10537266262839083,
      "grad_norm": 3.0214569568634033,
      "learning_rate": 4.473795101395839e-05,
      "loss": 0.876,
      "step": 4001
    },
    {
      "epoch": 0.10539899920990256,
      "grad_norm": 2.2037527561187744,
      "learning_rate": 4.4736634184882804e-05,
      "loss": 1.4227,
      "step": 4002
    },
    {
      "epoch": 0.10542533579141428,
      "grad_norm": 2.4519338607788086,
      "learning_rate": 4.473531735580721e-05,
      "loss": 1.7417,
      "step": 4003
    },
    {
      "epoch": 0.10545167237292599,
      "grad_norm": 4.407434940338135,
      "learning_rate": 4.4734000526731635e-05,
      "loss": 1.3909,
      "step": 4004
    },
    {
      "epoch": 0.10547800895443771,
      "grad_norm": 2.551903009414673,
      "learning_rate": 4.4732683697656044e-05,
      "loss": 1.966,
      "step": 4005
    },
    {
      "epoch": 0.10550434553594944,
      "grad_norm": 2.2575459480285645,
      "learning_rate": 4.473136686858046e-05,
      "loss": 1.7654,
      "step": 4006
    },
    {
      "epoch": 0.10553068211746115,
      "grad_norm": 2.2231922149658203,
      "learning_rate": 4.4730050039504875e-05,
      "loss": 1.8128,
      "step": 4007
    },
    {
      "epoch": 0.10555701869897287,
      "grad_norm": 2.688328981399536,
      "learning_rate": 4.472873321042929e-05,
      "loss": 1.3064,
      "step": 4008
    },
    {
      "epoch": 0.1055833552804846,
      "grad_norm": 2.364464282989502,
      "learning_rate": 4.4727416381353706e-05,
      "loss": 1.5036,
      "step": 4009
    },
    {
      "epoch": 0.10560969186199631,
      "grad_norm": 3.155242681503296,
      "learning_rate": 4.4726099552278115e-05,
      "loss": 1.2735,
      "step": 4010
    },
    {
      "epoch": 0.10563602844350803,
      "grad_norm": 2.4731411933898926,
      "learning_rate": 4.472478272320253e-05,
      "loss": 2.0021,
      "step": 4011
    },
    {
      "epoch": 0.10566236502501976,
      "grad_norm": 3.477363348007202,
      "learning_rate": 4.472346589412694e-05,
      "loss": 0.7553,
      "step": 4012
    },
    {
      "epoch": 0.10568870160653147,
      "grad_norm": 2.0620481967926025,
      "learning_rate": 4.472214906505136e-05,
      "loss": 1.0688,
      "step": 4013
    },
    {
      "epoch": 0.10571503818804319,
      "grad_norm": 2.5248160362243652,
      "learning_rate": 4.472083223597577e-05,
      "loss": 1.4607,
      "step": 4014
    },
    {
      "epoch": 0.1057413747695549,
      "grad_norm": 1.8622386455535889,
      "learning_rate": 4.4719515406900186e-05,
      "loss": 1.5775,
      "step": 4015
    },
    {
      "epoch": 0.10576771135106663,
      "grad_norm": 1.6190394163131714,
      "learning_rate": 4.47181985778246e-05,
      "loss": 1.2959,
      "step": 4016
    },
    {
      "epoch": 0.10579404793257835,
      "grad_norm": 4.105989933013916,
      "learning_rate": 4.471688174874901e-05,
      "loss": 1.3315,
      "step": 4017
    },
    {
      "epoch": 0.10582038451409007,
      "grad_norm": 2.533867359161377,
      "learning_rate": 4.471556491967343e-05,
      "loss": 1.0959,
      "step": 4018
    },
    {
      "epoch": 0.1058467210956018,
      "grad_norm": 3.2738797664642334,
      "learning_rate": 4.471424809059784e-05,
      "loss": 1.7608,
      "step": 4019
    },
    {
      "epoch": 0.10587305767711351,
      "grad_norm": 3.3484888076782227,
      "learning_rate": 4.471293126152226e-05,
      "loss": 1.952,
      "step": 4020
    },
    {
      "epoch": 0.10589939425862523,
      "grad_norm": 2.184448480606079,
      "learning_rate": 4.4711614432446666e-05,
      "loss": 1.7956,
      "step": 4021
    },
    {
      "epoch": 0.10592573084013696,
      "grad_norm": 2.1217143535614014,
      "learning_rate": 4.471029760337108e-05,
      "loss": 1.5918,
      "step": 4022
    },
    {
      "epoch": 0.10595206742164867,
      "grad_norm": 1.9207357168197632,
      "learning_rate": 4.47089807742955e-05,
      "loss": 1.9232,
      "step": 4023
    },
    {
      "epoch": 0.10597840400316039,
      "grad_norm": 2.3726885318756104,
      "learning_rate": 4.470766394521991e-05,
      "loss": 1.6053,
      "step": 4024
    },
    {
      "epoch": 0.10600474058467212,
      "grad_norm": 2.4885849952697754,
      "learning_rate": 4.470634711614433e-05,
      "loss": 2.5578,
      "step": 4025
    },
    {
      "epoch": 0.10603107716618383,
      "grad_norm": 2.848715305328369,
      "learning_rate": 4.470503028706874e-05,
      "loss": 1.7455,
      "step": 4026
    },
    {
      "epoch": 0.10605741374769555,
      "grad_norm": 3.8315703868865967,
      "learning_rate": 4.470371345799316e-05,
      "loss": 1.6488,
      "step": 4027
    },
    {
      "epoch": 0.10608375032920726,
      "grad_norm": 1.9673184156417847,
      "learning_rate": 4.470239662891757e-05,
      "loss": 0.8827,
      "step": 4028
    },
    {
      "epoch": 0.10611008691071899,
      "grad_norm": 2.9483304023742676,
      "learning_rate": 4.4701079799841984e-05,
      "loss": 1.31,
      "step": 4029
    },
    {
      "epoch": 0.10613642349223071,
      "grad_norm": 1.9678658246994019,
      "learning_rate": 4.469976297076639e-05,
      "loss": 0.3727,
      "step": 4030
    },
    {
      "epoch": 0.10616276007374242,
      "grad_norm": 1.7625017166137695,
      "learning_rate": 4.469844614169081e-05,
      "loss": 1.9109,
      "step": 4031
    },
    {
      "epoch": 0.10618909665525415,
      "grad_norm": 3.294492483139038,
      "learning_rate": 4.4697129312615224e-05,
      "loss": 2.2724,
      "step": 4032
    },
    {
      "epoch": 0.10621543323676587,
      "grad_norm": 1.878441572189331,
      "learning_rate": 4.469581248353964e-05,
      "loss": 2.2334,
      "step": 4033
    },
    {
      "epoch": 0.10624176981827758,
      "grad_norm": 3.995889663696289,
      "learning_rate": 4.4694495654464056e-05,
      "loss": 1.3493,
      "step": 4034
    },
    {
      "epoch": 0.10626810639978931,
      "grad_norm": 3.1018190383911133,
      "learning_rate": 4.4693178825388464e-05,
      "loss": 1.5428,
      "step": 4035
    },
    {
      "epoch": 0.10629444298130103,
      "grad_norm": 3.6806719303131104,
      "learning_rate": 4.469186199631288e-05,
      "loss": 1.6416,
      "step": 4036
    },
    {
      "epoch": 0.10632077956281275,
      "grad_norm": 2.073110580444336,
      "learning_rate": 4.4690545167237296e-05,
      "loss": 1.5843,
      "step": 4037
    },
    {
      "epoch": 0.10634711614432446,
      "grad_norm": 3.1895880699157715,
      "learning_rate": 4.468922833816171e-05,
      "loss": 2.2509,
      "step": 4038
    },
    {
      "epoch": 0.10637345272583619,
      "grad_norm": 1.8896957635879517,
      "learning_rate": 4.468791150908612e-05,
      "loss": 0.6443,
      "step": 4039
    },
    {
      "epoch": 0.1063997893073479,
      "grad_norm": 2.06345534324646,
      "learning_rate": 4.4686594680010536e-05,
      "loss": 1.6476,
      "step": 4040
    },
    {
      "epoch": 0.10642612588885962,
      "grad_norm": 2.168785333633423,
      "learning_rate": 4.468527785093495e-05,
      "loss": 2.1182,
      "step": 4041
    },
    {
      "epoch": 0.10645246247037135,
      "grad_norm": 3.1613264083862305,
      "learning_rate": 4.468396102185937e-05,
      "loss": 1.5156,
      "step": 4042
    },
    {
      "epoch": 0.10647879905188307,
      "grad_norm": 2.9733388423919678,
      "learning_rate": 4.468264419278378e-05,
      "loss": 0.9647,
      "step": 4043
    },
    {
      "epoch": 0.10650513563339478,
      "grad_norm": 3.550116539001465,
      "learning_rate": 4.468132736370819e-05,
      "loss": 1.8665,
      "step": 4044
    },
    {
      "epoch": 0.10653147221490651,
      "grad_norm": 4.879088878631592,
      "learning_rate": 4.468001053463261e-05,
      "loss": 1.4219,
      "step": 4045
    },
    {
      "epoch": 0.10655780879641823,
      "grad_norm": 2.0437891483306885,
      "learning_rate": 4.467869370555702e-05,
      "loss": 1.4834,
      "step": 4046
    },
    {
      "epoch": 0.10658414537792994,
      "grad_norm": 3.944870948791504,
      "learning_rate": 4.467737687648144e-05,
      "loss": 1.9115,
      "step": 4047
    },
    {
      "epoch": 0.10661048195944166,
      "grad_norm": 1.9715152978897095,
      "learning_rate": 4.467606004740585e-05,
      "loss": 1.826,
      "step": 4048
    },
    {
      "epoch": 0.10663681854095339,
      "grad_norm": 2.3294150829315186,
      "learning_rate": 4.467474321833026e-05,
      "loss": 2.1391,
      "step": 4049
    },
    {
      "epoch": 0.1066631551224651,
      "grad_norm": 2.313127279281616,
      "learning_rate": 4.467342638925467e-05,
      "loss": 0.3878,
      "step": 4050
    },
    {
      "epoch": 0.10668949170397682,
      "grad_norm": 5.281714916229248,
      "learning_rate": 4.4672109560179094e-05,
      "loss": 1.6318,
      "step": 4051
    },
    {
      "epoch": 0.10671582828548855,
      "grad_norm": 4.25741720199585,
      "learning_rate": 4.46707927311035e-05,
      "loss": 1.8003,
      "step": 4052
    },
    {
      "epoch": 0.10674216486700026,
      "grad_norm": 1.9785746335983276,
      "learning_rate": 4.466947590202792e-05,
      "loss": 1.6576,
      "step": 4053
    },
    {
      "epoch": 0.10676850144851198,
      "grad_norm": 3.613093614578247,
      "learning_rate": 4.4668159072952334e-05,
      "loss": 1.4468,
      "step": 4054
    },
    {
      "epoch": 0.10679483803002371,
      "grad_norm": 2.0653152465820312,
      "learning_rate": 4.466684224387674e-05,
      "loss": 1.9659,
      "step": 4055
    },
    {
      "epoch": 0.10682117461153542,
      "grad_norm": 1.8395441770553589,
      "learning_rate": 4.4665525414801165e-05,
      "loss": 2.1076,
      "step": 4056
    },
    {
      "epoch": 0.10684751119304714,
      "grad_norm": 1.7062104940414429,
      "learning_rate": 4.4664208585725574e-05,
      "loss": 1.736,
      "step": 4057
    },
    {
      "epoch": 0.10687384777455886,
      "grad_norm": 2.5451226234436035,
      "learning_rate": 4.466289175664999e-05,
      "loss": 1.6051,
      "step": 4058
    },
    {
      "epoch": 0.10690018435607059,
      "grad_norm": 2.8845815658569336,
      "learning_rate": 4.46615749275744e-05,
      "loss": 1.7561,
      "step": 4059
    },
    {
      "epoch": 0.1069265209375823,
      "grad_norm": 4.4622063636779785,
      "learning_rate": 4.466025809849882e-05,
      "loss": 1.5482,
      "step": 4060
    },
    {
      "epoch": 0.10695285751909402,
      "grad_norm": 4.345691680908203,
      "learning_rate": 4.465894126942323e-05,
      "loss": 2.1165,
      "step": 4061
    },
    {
      "epoch": 0.10697919410060575,
      "grad_norm": 3.109825849533081,
      "learning_rate": 4.4657624440347645e-05,
      "loss": 1.601,
      "step": 4062
    },
    {
      "epoch": 0.10700553068211746,
      "grad_norm": 2.058493137359619,
      "learning_rate": 4.465630761127206e-05,
      "loss": 2.0234,
      "step": 4063
    },
    {
      "epoch": 0.10703186726362918,
      "grad_norm": 3.9762372970581055,
      "learning_rate": 4.465499078219647e-05,
      "loss": 1.4851,
      "step": 4064
    },
    {
      "epoch": 0.1070582038451409,
      "grad_norm": 3.4705705642700195,
      "learning_rate": 4.465367395312089e-05,
      "loss": 2.2298,
      "step": 4065
    },
    {
      "epoch": 0.10708454042665262,
      "grad_norm": 2.324753761291504,
      "learning_rate": 4.46523571240453e-05,
      "loss": 2.2071,
      "step": 4066
    },
    {
      "epoch": 0.10711087700816434,
      "grad_norm": 2.5844130516052246,
      "learning_rate": 4.4651040294969716e-05,
      "loss": 1.4767,
      "step": 4067
    },
    {
      "epoch": 0.10713721358967607,
      "grad_norm": 1.953588604927063,
      "learning_rate": 4.4649723465894125e-05,
      "loss": 1.6706,
      "step": 4068
    },
    {
      "epoch": 0.10716355017118778,
      "grad_norm": 2.027574062347412,
      "learning_rate": 4.464840663681854e-05,
      "loss": 1.6953,
      "step": 4069
    },
    {
      "epoch": 0.1071898867526995,
      "grad_norm": 4.431998252868652,
      "learning_rate": 4.4647089807742956e-05,
      "loss": 1.5115,
      "step": 4070
    },
    {
      "epoch": 0.10721622333421121,
      "grad_norm": 3.2604241371154785,
      "learning_rate": 4.464577297866737e-05,
      "loss": 1.9019,
      "step": 4071
    },
    {
      "epoch": 0.10724255991572294,
      "grad_norm": 2.3794023990631104,
      "learning_rate": 4.464445614959179e-05,
      "loss": 1.8177,
      "step": 4072
    },
    {
      "epoch": 0.10726889649723466,
      "grad_norm": 2.0373446941375732,
      "learning_rate": 4.4643139320516196e-05,
      "loss": 1.8285,
      "step": 4073
    },
    {
      "epoch": 0.10729523307874637,
      "grad_norm": 2.27317476272583,
      "learning_rate": 4.464182249144062e-05,
      "loss": 1.8145,
      "step": 4074
    },
    {
      "epoch": 0.1073215696602581,
      "grad_norm": 2.8986411094665527,
      "learning_rate": 4.464050566236503e-05,
      "loss": 2.1543,
      "step": 4075
    },
    {
      "epoch": 0.10734790624176982,
      "grad_norm": 2.6787140369415283,
      "learning_rate": 4.463918883328944e-05,
      "loss": 2.4231,
      "step": 4076
    },
    {
      "epoch": 0.10737424282328153,
      "grad_norm": 1.8390060663223267,
      "learning_rate": 4.463787200421385e-05,
      "loss": 1.5888,
      "step": 4077
    },
    {
      "epoch": 0.10740057940479326,
      "grad_norm": 1.8164650201797485,
      "learning_rate": 4.463655517513827e-05,
      "loss": 2.0236,
      "step": 4078
    },
    {
      "epoch": 0.10742691598630498,
      "grad_norm": 2.007504940032959,
      "learning_rate": 4.463523834606268e-05,
      "loss": 1.6726,
      "step": 4079
    },
    {
      "epoch": 0.1074532525678167,
      "grad_norm": 1.733331322669983,
      "learning_rate": 4.46339215169871e-05,
      "loss": 1.7026,
      "step": 4080
    },
    {
      "epoch": 0.10747958914932841,
      "grad_norm": 4.508018970489502,
      "learning_rate": 4.4632604687911514e-05,
      "loss": 1.7218,
      "step": 4081
    },
    {
      "epoch": 0.10750592573084014,
      "grad_norm": 5.116292476654053,
      "learning_rate": 4.463128785883592e-05,
      "loss": 1.6436,
      "step": 4082
    },
    {
      "epoch": 0.10753226231235186,
      "grad_norm": 1.760777473449707,
      "learning_rate": 4.462997102976034e-05,
      "loss": 1.6767,
      "step": 4083
    },
    {
      "epoch": 0.10755859889386357,
      "grad_norm": 3.2570009231567383,
      "learning_rate": 4.4628654200684754e-05,
      "loss": 1.9036,
      "step": 4084
    },
    {
      "epoch": 0.1075849354753753,
      "grad_norm": 1.937599539756775,
      "learning_rate": 4.462733737160917e-05,
      "loss": 1.5554,
      "step": 4085
    },
    {
      "epoch": 0.10761127205688702,
      "grad_norm": 1.9227243661880493,
      "learning_rate": 4.462602054253358e-05,
      "loss": 0.6297,
      "step": 4086
    },
    {
      "epoch": 0.10763760863839873,
      "grad_norm": 2.8662517070770264,
      "learning_rate": 4.4624703713457994e-05,
      "loss": 1.8503,
      "step": 4087
    },
    {
      "epoch": 0.10766394521991046,
      "grad_norm": 1.8032362461090088,
      "learning_rate": 4.462338688438241e-05,
      "loss": 1.04,
      "step": 4088
    },
    {
      "epoch": 0.10769028180142218,
      "grad_norm": 4.272941589355469,
      "learning_rate": 4.4622070055306825e-05,
      "loss": 1.5103,
      "step": 4089
    },
    {
      "epoch": 0.10771661838293389,
      "grad_norm": 2.3737215995788574,
      "learning_rate": 4.462075322623124e-05,
      "loss": 0.9787,
      "step": 4090
    },
    {
      "epoch": 0.10774295496444561,
      "grad_norm": 2.4204230308532715,
      "learning_rate": 4.461943639715565e-05,
      "loss": 1.4417,
      "step": 4091
    },
    {
      "epoch": 0.10776929154595734,
      "grad_norm": 1.9358290433883667,
      "learning_rate": 4.4618119568080065e-05,
      "loss": 2.0035,
      "step": 4092
    },
    {
      "epoch": 0.10779562812746905,
      "grad_norm": 5.602682590484619,
      "learning_rate": 4.461680273900448e-05,
      "loss": 2.1958,
      "step": 4093
    },
    {
      "epoch": 0.10782196470898077,
      "grad_norm": 3.716036558151245,
      "learning_rate": 4.4615485909928897e-05,
      "loss": 1.3942,
      "step": 4094
    },
    {
      "epoch": 0.1078483012904925,
      "grad_norm": 3.2236311435699463,
      "learning_rate": 4.4614169080853305e-05,
      "loss": 2.1268,
      "step": 4095
    },
    {
      "epoch": 0.10787463787200421,
      "grad_norm": 3.3057029247283936,
      "learning_rate": 4.461285225177772e-05,
      "loss": 2.2917,
      "step": 4096
    },
    {
      "epoch": 0.10790097445351593,
      "grad_norm": 3.5505335330963135,
      "learning_rate": 4.461153542270213e-05,
      "loss": 0.6697,
      "step": 4097
    },
    {
      "epoch": 0.10792731103502766,
      "grad_norm": 2.0468151569366455,
      "learning_rate": 4.461021859362655e-05,
      "loss": 2.5624,
      "step": 4098
    },
    {
      "epoch": 0.10795364761653937,
      "grad_norm": 1.9204453229904175,
      "learning_rate": 4.460890176455096e-05,
      "loss": 1.5151,
      "step": 4099
    },
    {
      "epoch": 0.10797998419805109,
      "grad_norm": 1.622131586074829,
      "learning_rate": 4.460758493547538e-05,
      "loss": 1.4166,
      "step": 4100
    },
    {
      "epoch": 0.1080063207795628,
      "grad_norm": 2.30871319770813,
      "learning_rate": 4.460626810639979e-05,
      "loss": 1.3913,
      "step": 4101
    },
    {
      "epoch": 0.10803265736107454,
      "grad_norm": 3.7625954151153564,
      "learning_rate": 4.46049512773242e-05,
      "loss": 1.3604,
      "step": 4102
    },
    {
      "epoch": 0.10805899394258625,
      "grad_norm": 11.044343948364258,
      "learning_rate": 4.4603634448248623e-05,
      "loss": 1.9603,
      "step": 4103
    },
    {
      "epoch": 0.10808533052409797,
      "grad_norm": 2.669440746307373,
      "learning_rate": 4.460231761917303e-05,
      "loss": 2.1086,
      "step": 4104
    },
    {
      "epoch": 0.1081116671056097,
      "grad_norm": 4.807341575622559,
      "learning_rate": 4.460100079009745e-05,
      "loss": 1.349,
      "step": 4105
    },
    {
      "epoch": 0.10813800368712141,
      "grad_norm": 2.4318532943725586,
      "learning_rate": 4.459968396102186e-05,
      "loss": 2.1921,
      "step": 4106
    },
    {
      "epoch": 0.10816434026863313,
      "grad_norm": 3.815917491912842,
      "learning_rate": 4.459836713194628e-05,
      "loss": 0.2555,
      "step": 4107
    },
    {
      "epoch": 0.10819067685014486,
      "grad_norm": 1.9405546188354492,
      "learning_rate": 4.459705030287069e-05,
      "loss": 2.0096,
      "step": 4108
    },
    {
      "epoch": 0.10821701343165657,
      "grad_norm": 2.8776192665100098,
      "learning_rate": 4.4595733473795103e-05,
      "loss": 1.9219,
      "step": 4109
    },
    {
      "epoch": 0.10824335001316829,
      "grad_norm": 3.1089329719543457,
      "learning_rate": 4.459441664471952e-05,
      "loss": 0.4953,
      "step": 4110
    },
    {
      "epoch": 0.10826968659468002,
      "grad_norm": 2.188464403152466,
      "learning_rate": 4.459309981564393e-05,
      "loss": 1.3146,
      "step": 4111
    },
    {
      "epoch": 0.10829602317619173,
      "grad_norm": 5.935386657714844,
      "learning_rate": 4.459178298656835e-05,
      "loss": 1.643,
      "step": 4112
    },
    {
      "epoch": 0.10832235975770345,
      "grad_norm": 3.1096858978271484,
      "learning_rate": 4.459046615749276e-05,
      "loss": 1.1696,
      "step": 4113
    },
    {
      "epoch": 0.10834869633921516,
      "grad_norm": 1.9883707761764526,
      "learning_rate": 4.4589149328417175e-05,
      "loss": 1.9735,
      "step": 4114
    },
    {
      "epoch": 0.1083750329207269,
      "grad_norm": 2.773411512374878,
      "learning_rate": 4.4587832499341583e-05,
      "loss": 2.0659,
      "step": 4115
    },
    {
      "epoch": 0.10840136950223861,
      "grad_norm": 2.15742826461792,
      "learning_rate": 4.4586515670266e-05,
      "loss": 1.3497,
      "step": 4116
    },
    {
      "epoch": 0.10842770608375032,
      "grad_norm": 1.8726258277893066,
      "learning_rate": 4.4585198841190415e-05,
      "loss": 1.1951,
      "step": 4117
    },
    {
      "epoch": 0.10845404266526205,
      "grad_norm": 2.60878324508667,
      "learning_rate": 4.458388201211483e-05,
      "loss": 0.5625,
      "step": 4118
    },
    {
      "epoch": 0.10848037924677377,
      "grad_norm": 3.1422879695892334,
      "learning_rate": 4.4582565183039246e-05,
      "loss": 1.6653,
      "step": 4119
    },
    {
      "epoch": 0.10850671582828549,
      "grad_norm": 2.1638922691345215,
      "learning_rate": 4.4581248353963655e-05,
      "loss": 1.8287,
      "step": 4120
    },
    {
      "epoch": 0.10853305240979721,
      "grad_norm": 1.718577265739441,
      "learning_rate": 4.457993152488807e-05,
      "loss": 1.8536,
      "step": 4121
    },
    {
      "epoch": 0.10855938899130893,
      "grad_norm": 1.9097579717636108,
      "learning_rate": 4.4578614695812486e-05,
      "loss": 1.5093,
      "step": 4122
    },
    {
      "epoch": 0.10858572557282065,
      "grad_norm": 2.0030362606048584,
      "learning_rate": 4.45772978667369e-05,
      "loss": 2.147,
      "step": 4123
    },
    {
      "epoch": 0.10861206215433236,
      "grad_norm": 3.1014513969421387,
      "learning_rate": 4.457598103766131e-05,
      "loss": 1.6916,
      "step": 4124
    },
    {
      "epoch": 0.10863839873584409,
      "grad_norm": 2.2505898475646973,
      "learning_rate": 4.4574664208585726e-05,
      "loss": 2.3022,
      "step": 4125
    },
    {
      "epoch": 0.1086647353173558,
      "grad_norm": 1.8901978731155396,
      "learning_rate": 4.457334737951014e-05,
      "loss": 1.8867,
      "step": 4126
    },
    {
      "epoch": 0.10869107189886752,
      "grad_norm": 3.16646409034729,
      "learning_rate": 4.457203055043456e-05,
      "loss": 1.7358,
      "step": 4127
    },
    {
      "epoch": 0.10871740848037925,
      "grad_norm": 3.0400588512420654,
      "learning_rate": 4.457071372135897e-05,
      "loss": 1.9286,
      "step": 4128
    },
    {
      "epoch": 0.10874374506189097,
      "grad_norm": 3.8501882553100586,
      "learning_rate": 4.456939689228338e-05,
      "loss": 1.295,
      "step": 4129
    },
    {
      "epoch": 0.10877008164340268,
      "grad_norm": 4.805459022521973,
      "learning_rate": 4.45680800632078e-05,
      "loss": 2.2903,
      "step": 4130
    },
    {
      "epoch": 0.10879641822491441,
      "grad_norm": 2.4454216957092285,
      "learning_rate": 4.456676323413221e-05,
      "loss": 2.1617,
      "step": 4131
    },
    {
      "epoch": 0.10882275480642613,
      "grad_norm": 2.029470920562744,
      "learning_rate": 4.456544640505663e-05,
      "loss": 2.1926,
      "step": 4132
    },
    {
      "epoch": 0.10884909138793784,
      "grad_norm": 2.4619250297546387,
      "learning_rate": 4.456412957598104e-05,
      "loss": 2.8683,
      "step": 4133
    },
    {
      "epoch": 0.10887542796944956,
      "grad_norm": 7.48714542388916,
      "learning_rate": 4.456281274690545e-05,
      "loss": 1.8232,
      "step": 4134
    },
    {
      "epoch": 0.10890176455096129,
      "grad_norm": 1.9956969022750854,
      "learning_rate": 4.456149591782987e-05,
      "loss": 1.5241,
      "step": 4135
    },
    {
      "epoch": 0.108928101132473,
      "grad_norm": 2.5294861793518066,
      "learning_rate": 4.4560179088754284e-05,
      "loss": 1.8114,
      "step": 4136
    },
    {
      "epoch": 0.10895443771398472,
      "grad_norm": 3.4034552574157715,
      "learning_rate": 4.45588622596787e-05,
      "loss": 0.2944,
      "step": 4137
    },
    {
      "epoch": 0.10898077429549645,
      "grad_norm": 2.0800437927246094,
      "learning_rate": 4.455754543060311e-05,
      "loss": 1.628,
      "step": 4138
    },
    {
      "epoch": 0.10900711087700816,
      "grad_norm": 3.241866111755371,
      "learning_rate": 4.4556228601527524e-05,
      "loss": 1.865,
      "step": 4139
    },
    {
      "epoch": 0.10903344745851988,
      "grad_norm": 1.9164265394210815,
      "learning_rate": 4.455491177245194e-05,
      "loss": 1.8292,
      "step": 4140
    },
    {
      "epoch": 0.10905978404003161,
      "grad_norm": 2.2351768016815186,
      "learning_rate": 4.4553594943376355e-05,
      "loss": 0.3051,
      "step": 4141
    },
    {
      "epoch": 0.10908612062154333,
      "grad_norm": 2.382841110229492,
      "learning_rate": 4.4552278114300764e-05,
      "loss": 1.2609,
      "step": 4142
    },
    {
      "epoch": 0.10911245720305504,
      "grad_norm": 2.1350529193878174,
      "learning_rate": 4.455096128522518e-05,
      "loss": 0.786,
      "step": 4143
    },
    {
      "epoch": 0.10913879378456676,
      "grad_norm": 2.8577592372894287,
      "learning_rate": 4.4549644456149595e-05,
      "loss": 1.8832,
      "step": 4144
    },
    {
      "epoch": 0.10916513036607849,
      "grad_norm": 4.506251335144043,
      "learning_rate": 4.454832762707401e-05,
      "loss": 1.184,
      "step": 4145
    },
    {
      "epoch": 0.1091914669475902,
      "grad_norm": 4.11175537109375,
      "learning_rate": 4.4547010797998426e-05,
      "loss": 0.3751,
      "step": 4146
    },
    {
      "epoch": 0.10921780352910192,
      "grad_norm": 1.7342851161956787,
      "learning_rate": 4.4545693968922835e-05,
      "loss": 0.5253,
      "step": 4147
    },
    {
      "epoch": 0.10924414011061365,
      "grad_norm": 4.022400379180908,
      "learning_rate": 4.454437713984725e-05,
      "loss": 1.8998,
      "step": 4148
    },
    {
      "epoch": 0.10927047669212536,
      "grad_norm": 2.1567111015319824,
      "learning_rate": 4.454306031077166e-05,
      "loss": 1.7549,
      "step": 4149
    },
    {
      "epoch": 0.10929681327363708,
      "grad_norm": 3.7303826808929443,
      "learning_rate": 4.454174348169608e-05,
      "loss": 1.4587,
      "step": 4150
    },
    {
      "epoch": 0.10932314985514881,
      "grad_norm": 2.7888081073760986,
      "learning_rate": 4.454042665262049e-05,
      "loss": 1.4682,
      "step": 4151
    },
    {
      "epoch": 0.10934948643666052,
      "grad_norm": 2.8986802101135254,
      "learning_rate": 4.4539109823544906e-05,
      "loss": 0.6415,
      "step": 4152
    },
    {
      "epoch": 0.10937582301817224,
      "grad_norm": 2.200427770614624,
      "learning_rate": 4.4537792994469315e-05,
      "loss": 1.8819,
      "step": 4153
    },
    {
      "epoch": 0.10940215959968395,
      "grad_norm": 3.080820322036743,
      "learning_rate": 4.453647616539374e-05,
      "loss": 1.7558,
      "step": 4154
    },
    {
      "epoch": 0.10942849618119568,
      "grad_norm": 2.1669230461120605,
      "learning_rate": 4.4535159336318146e-05,
      "loss": 2.0016,
      "step": 4155
    },
    {
      "epoch": 0.1094548327627074,
      "grad_norm": 2.913347005844116,
      "learning_rate": 4.453384250724256e-05,
      "loss": 1.5105,
      "step": 4156
    },
    {
      "epoch": 0.10948116934421911,
      "grad_norm": 3.9509127140045166,
      "learning_rate": 4.453252567816698e-05,
      "loss": 2.5394,
      "step": 4157
    },
    {
      "epoch": 0.10950750592573084,
      "grad_norm": 2.0484113693237305,
      "learning_rate": 4.4531208849091386e-05,
      "loss": 1.6148,
      "step": 4158
    },
    {
      "epoch": 0.10953384250724256,
      "grad_norm": 3.5272421836853027,
      "learning_rate": 4.452989202001581e-05,
      "loss": 1.8762,
      "step": 4159
    },
    {
      "epoch": 0.10956017908875428,
      "grad_norm": 2.2493324279785156,
      "learning_rate": 4.452857519094022e-05,
      "loss": 2.4996,
      "step": 4160
    },
    {
      "epoch": 0.109586515670266,
      "grad_norm": 2.2131247520446777,
      "learning_rate": 4.452725836186463e-05,
      "loss": 2.3426,
      "step": 4161
    },
    {
      "epoch": 0.10961285225177772,
      "grad_norm": 1.9247936010360718,
      "learning_rate": 4.452594153278904e-05,
      "loss": 2.1403,
      "step": 4162
    },
    {
      "epoch": 0.10963918883328944,
      "grad_norm": 2.1571877002716064,
      "learning_rate": 4.452462470371346e-05,
      "loss": 1.6365,
      "step": 4163
    },
    {
      "epoch": 0.10966552541480117,
      "grad_norm": 1.528718113899231,
      "learning_rate": 4.452330787463787e-05,
      "loss": 0.2761,
      "step": 4164
    },
    {
      "epoch": 0.10969186199631288,
      "grad_norm": 1.7336452007293701,
      "learning_rate": 4.452199104556229e-05,
      "loss": 2.2681,
      "step": 4165
    },
    {
      "epoch": 0.1097181985778246,
      "grad_norm": 1.8995816707611084,
      "learning_rate": 4.4520674216486704e-05,
      "loss": 1.7106,
      "step": 4166
    },
    {
      "epoch": 0.10974453515933631,
      "grad_norm": 5.518387794494629,
      "learning_rate": 4.451935738741111e-05,
      "loss": 1.471,
      "step": 4167
    },
    {
      "epoch": 0.10977087174084804,
      "grad_norm": 4.71088981628418,
      "learning_rate": 4.451804055833553e-05,
      "loss": 1.0357,
      "step": 4168
    },
    {
      "epoch": 0.10979720832235976,
      "grad_norm": 2.0973432064056396,
      "learning_rate": 4.4516723729259944e-05,
      "loss": 2.1014,
      "step": 4169
    },
    {
      "epoch": 0.10982354490387147,
      "grad_norm": 2.453827142715454,
      "learning_rate": 4.451540690018436e-05,
      "loss": 1.4535,
      "step": 4170
    },
    {
      "epoch": 0.1098498814853832,
      "grad_norm": 1.9312286376953125,
      "learning_rate": 4.451409007110877e-05,
      "loss": 1.7561,
      "step": 4171
    },
    {
      "epoch": 0.10987621806689492,
      "grad_norm": 2.0955183506011963,
      "learning_rate": 4.4512773242033184e-05,
      "loss": 1.7761,
      "step": 4172
    },
    {
      "epoch": 0.10990255464840663,
      "grad_norm": 2.345484972000122,
      "learning_rate": 4.45114564129576e-05,
      "loss": 0.5964,
      "step": 4173
    },
    {
      "epoch": 0.10992889122991836,
      "grad_norm": 2.8912203311920166,
      "learning_rate": 4.4510139583882016e-05,
      "loss": 1.9131,
      "step": 4174
    },
    {
      "epoch": 0.10995522781143008,
      "grad_norm": 2.093498945236206,
      "learning_rate": 4.450882275480643e-05,
      "loss": 1.7598,
      "step": 4175
    },
    {
      "epoch": 0.1099815643929418,
      "grad_norm": 2.3428702354431152,
      "learning_rate": 4.450750592573084e-05,
      "loss": 1.8295,
      "step": 4176
    },
    {
      "epoch": 0.11000790097445351,
      "grad_norm": 2.0588157176971436,
      "learning_rate": 4.4506189096655256e-05,
      "loss": 1.612,
      "step": 4177
    },
    {
      "epoch": 0.11003423755596524,
      "grad_norm": 2.188891887664795,
      "learning_rate": 4.450487226757967e-05,
      "loss": 2.5297,
      "step": 4178
    },
    {
      "epoch": 0.11006057413747695,
      "grad_norm": 4.410722255706787,
      "learning_rate": 4.450355543850409e-05,
      "loss": 1.0905,
      "step": 4179
    },
    {
      "epoch": 0.11008691071898867,
      "grad_norm": 5.933962345123291,
      "learning_rate": 4.4502238609428496e-05,
      "loss": 1.5406,
      "step": 4180
    },
    {
      "epoch": 0.1101132473005004,
      "grad_norm": 2.8442652225494385,
      "learning_rate": 4.450092178035291e-05,
      "loss": 1.8647,
      "step": 4181
    },
    {
      "epoch": 0.11013958388201212,
      "grad_norm": 2.6928162574768066,
      "learning_rate": 4.449960495127733e-05,
      "loss": 1.7226,
      "step": 4182
    },
    {
      "epoch": 0.11016592046352383,
      "grad_norm": 2.676680564880371,
      "learning_rate": 4.449828812220174e-05,
      "loss": 1.7621,
      "step": 4183
    },
    {
      "epoch": 0.11019225704503556,
      "grad_norm": 1.8384419679641724,
      "learning_rate": 4.449697129312616e-05,
      "loss": 1.8842,
      "step": 4184
    },
    {
      "epoch": 0.11021859362654728,
      "grad_norm": 2.114710807800293,
      "learning_rate": 4.449565446405057e-05,
      "loss": 1.2852,
      "step": 4185
    },
    {
      "epoch": 0.11024493020805899,
      "grad_norm": 3.0580577850341797,
      "learning_rate": 4.449433763497498e-05,
      "loss": 1.1934,
      "step": 4186
    },
    {
      "epoch": 0.11027126678957071,
      "grad_norm": 1.8078515529632568,
      "learning_rate": 4.44930208058994e-05,
      "loss": 1.5872,
      "step": 4187
    },
    {
      "epoch": 0.11029760337108244,
      "grad_norm": 2.373305082321167,
      "learning_rate": 4.4491703976823814e-05,
      "loss": 1.7719,
      "step": 4188
    },
    {
      "epoch": 0.11032393995259415,
      "grad_norm": 2.870018482208252,
      "learning_rate": 4.449038714774822e-05,
      "loss": 2.0621,
      "step": 4189
    },
    {
      "epoch": 0.11035027653410587,
      "grad_norm": 2.607174873352051,
      "learning_rate": 4.448907031867264e-05,
      "loss": 1.8878,
      "step": 4190
    },
    {
      "epoch": 0.1103766131156176,
      "grad_norm": 2.303746223449707,
      "learning_rate": 4.4487753489597054e-05,
      "loss": 1.9227,
      "step": 4191
    },
    {
      "epoch": 0.11040294969712931,
      "grad_norm": 2.5911405086517334,
      "learning_rate": 4.448643666052147e-05,
      "loss": 1.0763,
      "step": 4192
    },
    {
      "epoch": 0.11042928627864103,
      "grad_norm": 3.95622181892395,
      "learning_rate": 4.4485119831445885e-05,
      "loss": 1.2832,
      "step": 4193
    },
    {
      "epoch": 0.11045562286015276,
      "grad_norm": 2.520411729812622,
      "learning_rate": 4.4483803002370294e-05,
      "loss": 2.4911,
      "step": 4194
    },
    {
      "epoch": 0.11048195944166447,
      "grad_norm": 3.7951996326446533,
      "learning_rate": 4.448248617329471e-05,
      "loss": 1.7614,
      "step": 4195
    },
    {
      "epoch": 0.11050829602317619,
      "grad_norm": 1.9339969158172607,
      "learning_rate": 4.448116934421912e-05,
      "loss": 1.4992,
      "step": 4196
    },
    {
      "epoch": 0.1105346326046879,
      "grad_norm": 4.729146957397461,
      "learning_rate": 4.447985251514354e-05,
      "loss": 1.4408,
      "step": 4197
    },
    {
      "epoch": 0.11056096918619963,
      "grad_norm": 2.343226671218872,
      "learning_rate": 4.447853568606795e-05,
      "loss": 1.6741,
      "step": 4198
    },
    {
      "epoch": 0.11058730576771135,
      "grad_norm": 1.902189016342163,
      "learning_rate": 4.4477218856992365e-05,
      "loss": 2.0228,
      "step": 4199
    },
    {
      "epoch": 0.11061364234922307,
      "grad_norm": 1.986051321029663,
      "learning_rate": 4.4475902027916774e-05,
      "loss": 2.0894,
      "step": 4200
    },
    {
      "epoch": 0.1106399789307348,
      "grad_norm": 2.330847978591919,
      "learning_rate": 4.447458519884119e-05,
      "loss": 0.2927,
      "step": 4201
    },
    {
      "epoch": 0.11066631551224651,
      "grad_norm": 2.673935651779175,
      "learning_rate": 4.4473268369765605e-05,
      "loss": 1.1512,
      "step": 4202
    },
    {
      "epoch": 0.11069265209375823,
      "grad_norm": 3.250035285949707,
      "learning_rate": 4.447195154069002e-05,
      "loss": 1.3046,
      "step": 4203
    },
    {
      "epoch": 0.11071898867526996,
      "grad_norm": 3.924764394760132,
      "learning_rate": 4.4470634711614436e-05,
      "loss": 2.0056,
      "step": 4204
    },
    {
      "epoch": 0.11074532525678167,
      "grad_norm": 5.089044570922852,
      "learning_rate": 4.4469317882538845e-05,
      "loss": 1.4291,
      "step": 4205
    },
    {
      "epoch": 0.11077166183829339,
      "grad_norm": 2.209367275238037,
      "learning_rate": 4.446800105346327e-05,
      "loss": 2.1428,
      "step": 4206
    },
    {
      "epoch": 0.11079799841980512,
      "grad_norm": 3.346619129180908,
      "learning_rate": 4.4466684224387676e-05,
      "loss": 0.4643,
      "step": 4207
    },
    {
      "epoch": 0.11082433500131683,
      "grad_norm": 4.271258354187012,
      "learning_rate": 4.446536739531209e-05,
      "loss": 1.0904,
      "step": 4208
    },
    {
      "epoch": 0.11085067158282855,
      "grad_norm": 3.2781922817230225,
      "learning_rate": 4.44640505662365e-05,
      "loss": 2.4328,
      "step": 4209
    },
    {
      "epoch": 0.11087700816434026,
      "grad_norm": 1.5749311447143555,
      "learning_rate": 4.4462733737160916e-05,
      "loss": 1.4835,
      "step": 4210
    },
    {
      "epoch": 0.11090334474585199,
      "grad_norm": 1.7231374979019165,
      "learning_rate": 4.446141690808533e-05,
      "loss": 1.6169,
      "step": 4211
    },
    {
      "epoch": 0.11092968132736371,
      "grad_norm": 2.5679404735565186,
      "learning_rate": 4.446010007900975e-05,
      "loss": 1.6938,
      "step": 4212
    },
    {
      "epoch": 0.11095601790887542,
      "grad_norm": 3.925067186355591,
      "learning_rate": 4.445878324993416e-05,
      "loss": 0.9371,
      "step": 4213
    },
    {
      "epoch": 0.11098235449038715,
      "grad_norm": 3.1073853969573975,
      "learning_rate": 4.445746642085857e-05,
      "loss": 2.0418,
      "step": 4214
    },
    {
      "epoch": 0.11100869107189887,
      "grad_norm": 2.345743179321289,
      "learning_rate": 4.445614959178299e-05,
      "loss": 3.329,
      "step": 4215
    },
    {
      "epoch": 0.11103502765341058,
      "grad_norm": 1.7030315399169922,
      "learning_rate": 4.44548327627074e-05,
      "loss": 2.2939,
      "step": 4216
    },
    {
      "epoch": 0.11106136423492231,
      "grad_norm": 2.1973705291748047,
      "learning_rate": 4.445351593363182e-05,
      "loss": 1.5548,
      "step": 4217
    },
    {
      "epoch": 0.11108770081643403,
      "grad_norm": 2.013047695159912,
      "learning_rate": 4.445219910455623e-05,
      "loss": 1.3957,
      "step": 4218
    },
    {
      "epoch": 0.11111403739794574,
      "grad_norm": 1.4847007989883423,
      "learning_rate": 4.445088227548064e-05,
      "loss": 1.7038,
      "step": 4219
    },
    {
      "epoch": 0.11114037397945746,
      "grad_norm": 1.9858585596084595,
      "learning_rate": 4.444956544640506e-05,
      "loss": 0.3813,
      "step": 4220
    },
    {
      "epoch": 0.11116671056096919,
      "grad_norm": 1.7559846639633179,
      "learning_rate": 4.4448248617329474e-05,
      "loss": 1.6474,
      "step": 4221
    },
    {
      "epoch": 0.1111930471424809,
      "grad_norm": 1.98350989818573,
      "learning_rate": 4.444693178825389e-05,
      "loss": 1.4676,
      "step": 4222
    },
    {
      "epoch": 0.11121938372399262,
      "grad_norm": 1.9280142784118652,
      "learning_rate": 4.44456149591783e-05,
      "loss": 2.0786,
      "step": 4223
    },
    {
      "epoch": 0.11124572030550435,
      "grad_norm": 3.110429525375366,
      "learning_rate": 4.4444298130102714e-05,
      "loss": 2.2039,
      "step": 4224
    },
    {
      "epoch": 0.11127205688701607,
      "grad_norm": 8.633528709411621,
      "learning_rate": 4.444298130102713e-05,
      "loss": 1.6533,
      "step": 4225
    },
    {
      "epoch": 0.11129839346852778,
      "grad_norm": 5.191372871398926,
      "learning_rate": 4.4441664471951545e-05,
      "loss": 0.8776,
      "step": 4226
    },
    {
      "epoch": 0.11132473005003951,
      "grad_norm": 2.7773046493530273,
      "learning_rate": 4.4440347642875954e-05,
      "loss": 2.2684,
      "step": 4227
    },
    {
      "epoch": 0.11135106663155123,
      "grad_norm": 1.843032956123352,
      "learning_rate": 4.443903081380037e-05,
      "loss": 1.562,
      "step": 4228
    },
    {
      "epoch": 0.11137740321306294,
      "grad_norm": 3.5764057636260986,
      "learning_rate": 4.4437713984724785e-05,
      "loss": 1.987,
      "step": 4229
    },
    {
      "epoch": 0.11140373979457466,
      "grad_norm": 2.1517395973205566,
      "learning_rate": 4.44363971556492e-05,
      "loss": 1.8344,
      "step": 4230
    },
    {
      "epoch": 0.11143007637608639,
      "grad_norm": 2.2035233974456787,
      "learning_rate": 4.443508032657362e-05,
      "loss": 1.7476,
      "step": 4231
    },
    {
      "epoch": 0.1114564129575981,
      "grad_norm": 1.5685129165649414,
      "learning_rate": 4.4433763497498025e-05,
      "loss": 2.0277,
      "step": 4232
    },
    {
      "epoch": 0.11148274953910982,
      "grad_norm": 3.6586594581604004,
      "learning_rate": 4.443244666842244e-05,
      "loss": 2.1176,
      "step": 4233
    },
    {
      "epoch": 0.11150908612062155,
      "grad_norm": 4.339465618133545,
      "learning_rate": 4.443112983934685e-05,
      "loss": 1.5764,
      "step": 4234
    },
    {
      "epoch": 0.11153542270213326,
      "grad_norm": 5.024068355560303,
      "learning_rate": 4.442981301027127e-05,
      "loss": 1.3535,
      "step": 4235
    },
    {
      "epoch": 0.11156175928364498,
      "grad_norm": 2.071877956390381,
      "learning_rate": 4.442849618119568e-05,
      "loss": 2.2792,
      "step": 4236
    },
    {
      "epoch": 0.11158809586515671,
      "grad_norm": 2.2989859580993652,
      "learning_rate": 4.44271793521201e-05,
      "loss": 2.0244,
      "step": 4237
    },
    {
      "epoch": 0.11161443244666842,
      "grad_norm": 3.2281546592712402,
      "learning_rate": 4.442586252304451e-05,
      "loss": 1.5432,
      "step": 4238
    },
    {
      "epoch": 0.11164076902818014,
      "grad_norm": 2.5454723834991455,
      "learning_rate": 4.442454569396893e-05,
      "loss": 1.3705,
      "step": 4239
    },
    {
      "epoch": 0.11166710560969186,
      "grad_norm": 2.0404036045074463,
      "learning_rate": 4.4423228864893343e-05,
      "loss": 2.1095,
      "step": 4240
    },
    {
      "epoch": 0.11169344219120358,
      "grad_norm": 1.8799490928649902,
      "learning_rate": 4.442191203581775e-05,
      "loss": 2.1393,
      "step": 4241
    },
    {
      "epoch": 0.1117197787727153,
      "grad_norm": 3.2960891723632812,
      "learning_rate": 4.442059520674217e-05,
      "loss": 1.0129,
      "step": 4242
    },
    {
      "epoch": 0.11174611535422702,
      "grad_norm": 2.4646873474121094,
      "learning_rate": 4.441927837766658e-05,
      "loss": 1.8543,
      "step": 4243
    },
    {
      "epoch": 0.11177245193573875,
      "grad_norm": 4.208311557769775,
      "learning_rate": 4.4417961548591e-05,
      "loss": 1.468,
      "step": 4244
    },
    {
      "epoch": 0.11179878851725046,
      "grad_norm": 2.6525988578796387,
      "learning_rate": 4.441664471951541e-05,
      "loss": 1.7819,
      "step": 4245
    },
    {
      "epoch": 0.11182512509876218,
      "grad_norm": 1.6626691818237305,
      "learning_rate": 4.4415327890439823e-05,
      "loss": 1.8913,
      "step": 4246
    },
    {
      "epoch": 0.1118514616802739,
      "grad_norm": 5.3551344871521,
      "learning_rate": 4.441401106136424e-05,
      "loss": 2.6188,
      "step": 4247
    },
    {
      "epoch": 0.11187779826178562,
      "grad_norm": 1.665309190750122,
      "learning_rate": 4.441269423228865e-05,
      "loss": 1.8647,
      "step": 4248
    },
    {
      "epoch": 0.11190413484329734,
      "grad_norm": 3.119785785675049,
      "learning_rate": 4.441137740321307e-05,
      "loss": 2.627,
      "step": 4249
    },
    {
      "epoch": 0.11193047142480907,
      "grad_norm": 2.7680482864379883,
      "learning_rate": 4.441006057413748e-05,
      "loss": 1.6754,
      "step": 4250
    },
    {
      "epoch": 0.11195680800632078,
      "grad_norm": 2.355776786804199,
      "learning_rate": 4.4408743745061895e-05,
      "loss": 1.3855,
      "step": 4251
    },
    {
      "epoch": 0.1119831445878325,
      "grad_norm": 2.3840339183807373,
      "learning_rate": 4.4407426915986304e-05,
      "loss": 1.5975,
      "step": 4252
    },
    {
      "epoch": 0.11200948116934421,
      "grad_norm": 1.9081884622573853,
      "learning_rate": 4.4406110086910726e-05,
      "loss": 2.3249,
      "step": 4253
    },
    {
      "epoch": 0.11203581775085594,
      "grad_norm": 2.7819390296936035,
      "learning_rate": 4.4404793257835135e-05,
      "loss": 0.9198,
      "step": 4254
    },
    {
      "epoch": 0.11206215433236766,
      "grad_norm": 2.959852695465088,
      "learning_rate": 4.440347642875955e-05,
      "loss": 2.663,
      "step": 4255
    },
    {
      "epoch": 0.11208849091387937,
      "grad_norm": 1.9725509881973267,
      "learning_rate": 4.440215959968396e-05,
      "loss": 1.8166,
      "step": 4256
    },
    {
      "epoch": 0.1121148274953911,
      "grad_norm": 2.1061978340148926,
      "learning_rate": 4.4400842770608375e-05,
      "loss": 1.8506,
      "step": 4257
    },
    {
      "epoch": 0.11214116407690282,
      "grad_norm": 4.271763801574707,
      "learning_rate": 4.439952594153279e-05,
      "loss": 2.2017,
      "step": 4258
    },
    {
      "epoch": 0.11216750065841453,
      "grad_norm": 5.5284647941589355,
      "learning_rate": 4.4398209112457206e-05,
      "loss": 1.2421,
      "step": 4259
    },
    {
      "epoch": 0.11219383723992626,
      "grad_norm": 6.562661170959473,
      "learning_rate": 4.439689228338162e-05,
      "loss": 1.4948,
      "step": 4260
    },
    {
      "epoch": 0.11222017382143798,
      "grad_norm": 2.111541748046875,
      "learning_rate": 4.439557545430603e-05,
      "loss": 2.1255,
      "step": 4261
    },
    {
      "epoch": 0.1122465104029497,
      "grad_norm": 1.9101742506027222,
      "learning_rate": 4.4394258625230446e-05,
      "loss": 2.2274,
      "step": 4262
    },
    {
      "epoch": 0.11227284698446141,
      "grad_norm": 3.4336953163146973,
      "learning_rate": 4.439294179615486e-05,
      "loss": 1.6914,
      "step": 4263
    },
    {
      "epoch": 0.11229918356597314,
      "grad_norm": 4.317891597747803,
      "learning_rate": 4.439162496707928e-05,
      "loss": 2.1988,
      "step": 4264
    },
    {
      "epoch": 0.11232552014748486,
      "grad_norm": 2.151965618133545,
      "learning_rate": 4.4390308138003686e-05,
      "loss": 1.9896,
      "step": 4265
    },
    {
      "epoch": 0.11235185672899657,
      "grad_norm": 2.7007954120635986,
      "learning_rate": 4.43889913089281e-05,
      "loss": 1.6312,
      "step": 4266
    },
    {
      "epoch": 0.1123781933105083,
      "grad_norm": 1.9390941858291626,
      "learning_rate": 4.438767447985252e-05,
      "loss": 2.132,
      "step": 4267
    },
    {
      "epoch": 0.11240452989202002,
      "grad_norm": 5.374405860900879,
      "learning_rate": 4.438635765077693e-05,
      "loss": 2.2693,
      "step": 4268
    },
    {
      "epoch": 0.11243086647353173,
      "grad_norm": 2.5718936920166016,
      "learning_rate": 4.438504082170135e-05,
      "loss": 0.6173,
      "step": 4269
    },
    {
      "epoch": 0.11245720305504346,
      "grad_norm": 1.7235485315322876,
      "learning_rate": 4.438372399262576e-05,
      "loss": 1.767,
      "step": 4270
    },
    {
      "epoch": 0.11248353963655518,
      "grad_norm": 2.4974799156188965,
      "learning_rate": 4.438240716355017e-05,
      "loss": 2.0177,
      "step": 4271
    },
    {
      "epoch": 0.11250987621806689,
      "grad_norm": 2.061169147491455,
      "learning_rate": 4.438109033447459e-05,
      "loss": 0.8659,
      "step": 4272
    },
    {
      "epoch": 0.11253621279957861,
      "grad_norm": 3.2096126079559326,
      "learning_rate": 4.4379773505399004e-05,
      "loss": 2.5423,
      "step": 4273
    },
    {
      "epoch": 0.11256254938109034,
      "grad_norm": 2.260287284851074,
      "learning_rate": 4.437845667632341e-05,
      "loss": 1.917,
      "step": 4274
    },
    {
      "epoch": 0.11258888596260205,
      "grad_norm": 2.7628087997436523,
      "learning_rate": 4.437713984724783e-05,
      "loss": 1.6597,
      "step": 4275
    },
    {
      "epoch": 0.11261522254411377,
      "grad_norm": 1.9369984865188599,
      "learning_rate": 4.4375823018172244e-05,
      "loss": 1.8688,
      "step": 4276
    },
    {
      "epoch": 0.1126415591256255,
      "grad_norm": 2.6044199466705322,
      "learning_rate": 4.437450618909666e-05,
      "loss": 1.4238,
      "step": 4277
    },
    {
      "epoch": 0.11266789570713721,
      "grad_norm": 2.716409206390381,
      "learning_rate": 4.4373189360021075e-05,
      "loss": 1.6664,
      "step": 4278
    },
    {
      "epoch": 0.11269423228864893,
      "grad_norm": 1.6078535318374634,
      "learning_rate": 4.4371872530945484e-05,
      "loss": 1.5937,
      "step": 4279
    },
    {
      "epoch": 0.11272056887016066,
      "grad_norm": 3.0117154121398926,
      "learning_rate": 4.43705557018699e-05,
      "loss": 2.0069,
      "step": 4280
    },
    {
      "epoch": 0.11274690545167237,
      "grad_norm": 5.708094120025635,
      "learning_rate": 4.436923887279431e-05,
      "loss": 1.7758,
      "step": 4281
    },
    {
      "epoch": 0.11277324203318409,
      "grad_norm": 3.4249651432037354,
      "learning_rate": 4.436792204371873e-05,
      "loss": 0.6774,
      "step": 4282
    },
    {
      "epoch": 0.1127995786146958,
      "grad_norm": 3.1791722774505615,
      "learning_rate": 4.436660521464314e-05,
      "loss": 1.8117,
      "step": 4283
    },
    {
      "epoch": 0.11282591519620754,
      "grad_norm": 1.8081732988357544,
      "learning_rate": 4.4365288385567555e-05,
      "loss": 2.0789,
      "step": 4284
    },
    {
      "epoch": 0.11285225177771925,
      "grad_norm": 2.9484305381774902,
      "learning_rate": 4.436397155649197e-05,
      "loss": 2.2982,
      "step": 4285
    },
    {
      "epoch": 0.11287858835923097,
      "grad_norm": 2.696687936782837,
      "learning_rate": 4.4362654727416386e-05,
      "loss": 3.3121,
      "step": 4286
    },
    {
      "epoch": 0.1129049249407427,
      "grad_norm": 2.456861734390259,
      "learning_rate": 4.43613378983408e-05,
      "loss": 1.4228,
      "step": 4287
    },
    {
      "epoch": 0.11293126152225441,
      "grad_norm": 3.1820244789123535,
      "learning_rate": 4.436002106926521e-05,
      "loss": 1.8875,
      "step": 4288
    },
    {
      "epoch": 0.11295759810376613,
      "grad_norm": 2.078258752822876,
      "learning_rate": 4.4358704240189626e-05,
      "loss": 1.7224,
      "step": 4289
    },
    {
      "epoch": 0.11298393468527786,
      "grad_norm": 5.4816975593566895,
      "learning_rate": 4.4357387411114035e-05,
      "loss": 1.4001,
      "step": 4290
    },
    {
      "epoch": 0.11301027126678957,
      "grad_norm": 3.8903496265411377,
      "learning_rate": 4.435607058203846e-05,
      "loss": 1.4585,
      "step": 4291
    },
    {
      "epoch": 0.11303660784830129,
      "grad_norm": 2.2217657566070557,
      "learning_rate": 4.4354753752962866e-05,
      "loss": 1.5178,
      "step": 4292
    },
    {
      "epoch": 0.11306294442981302,
      "grad_norm": 5.49448823928833,
      "learning_rate": 4.435343692388728e-05,
      "loss": 1.6709,
      "step": 4293
    },
    {
      "epoch": 0.11308928101132473,
      "grad_norm": 2.8648664951324463,
      "learning_rate": 4.43521200948117e-05,
      "loss": 1.9657,
      "step": 4294
    },
    {
      "epoch": 0.11311561759283645,
      "grad_norm": 1.9732083082199097,
      "learning_rate": 4.4350803265736106e-05,
      "loss": 1.5418,
      "step": 4295
    },
    {
      "epoch": 0.11314195417434816,
      "grad_norm": 2.34871244430542,
      "learning_rate": 4.434948643666053e-05,
      "loss": 0.8984,
      "step": 4296
    },
    {
      "epoch": 0.1131682907558599,
      "grad_norm": 3.1706786155700684,
      "learning_rate": 4.434816960758494e-05,
      "loss": 1.7576,
      "step": 4297
    },
    {
      "epoch": 0.11319462733737161,
      "grad_norm": 1.9729595184326172,
      "learning_rate": 4.434685277850935e-05,
      "loss": 1.951,
      "step": 4298
    },
    {
      "epoch": 0.11322096391888332,
      "grad_norm": 1.9731125831604004,
      "learning_rate": 4.434553594943376e-05,
      "loss": 1.5688,
      "step": 4299
    },
    {
      "epoch": 0.11324730050039505,
      "grad_norm": 2.2782936096191406,
      "learning_rate": 4.434421912035818e-05,
      "loss": 2.0701,
      "step": 4300
    },
    {
      "epoch": 0.11327363708190677,
      "grad_norm": 3.6428115367889404,
      "learning_rate": 4.434290229128259e-05,
      "loss": 1.0274,
      "step": 4301
    },
    {
      "epoch": 0.11329997366341849,
      "grad_norm": 3.1329870223999023,
      "learning_rate": 4.434158546220701e-05,
      "loss": 2.4332,
      "step": 4302
    },
    {
      "epoch": 0.11332631024493021,
      "grad_norm": 1.8293405771255493,
      "learning_rate": 4.434026863313142e-05,
      "loss": 1.6498,
      "step": 4303
    },
    {
      "epoch": 0.11335264682644193,
      "grad_norm": 1.7683696746826172,
      "learning_rate": 4.433895180405583e-05,
      "loss": 1.5913,
      "step": 4304
    },
    {
      "epoch": 0.11337898340795365,
      "grad_norm": 1.813410997390747,
      "learning_rate": 4.433763497498025e-05,
      "loss": 1.578,
      "step": 4305
    },
    {
      "epoch": 0.11340531998946536,
      "grad_norm": 4.987881183624268,
      "learning_rate": 4.4336318145904664e-05,
      "loss": 2.0715,
      "step": 4306
    },
    {
      "epoch": 0.11343165657097709,
      "grad_norm": 3.3109843730926514,
      "learning_rate": 4.433500131682908e-05,
      "loss": 2.0603,
      "step": 4307
    },
    {
      "epoch": 0.1134579931524888,
      "grad_norm": 3.753157138824463,
      "learning_rate": 4.433368448775349e-05,
      "loss": 1.6679,
      "step": 4308
    },
    {
      "epoch": 0.11348432973400052,
      "grad_norm": 3.773768901824951,
      "learning_rate": 4.4332367658677905e-05,
      "loss": 1.7714,
      "step": 4309
    },
    {
      "epoch": 0.11351066631551225,
      "grad_norm": 2.662864923477173,
      "learning_rate": 4.433105082960232e-05,
      "loss": 1.7384,
      "step": 4310
    },
    {
      "epoch": 0.11353700289702397,
      "grad_norm": 2.2981088161468506,
      "learning_rate": 4.4329734000526736e-05,
      "loss": 2.5741,
      "step": 4311
    },
    {
      "epoch": 0.11356333947853568,
      "grad_norm": 12.36007022857666,
      "learning_rate": 4.4328417171451145e-05,
      "loss": 1.2394,
      "step": 4312
    },
    {
      "epoch": 0.11358967606004741,
      "grad_norm": 2.2096548080444336,
      "learning_rate": 4.432710034237556e-05,
      "loss": 1.0605,
      "step": 4313
    },
    {
      "epoch": 0.11361601264155913,
      "grad_norm": 1.8356379270553589,
      "learning_rate": 4.4325783513299976e-05,
      "loss": 1.6582,
      "step": 4314
    },
    {
      "epoch": 0.11364234922307084,
      "grad_norm": 3.6824419498443604,
      "learning_rate": 4.432446668422439e-05,
      "loss": 0.8603,
      "step": 4315
    },
    {
      "epoch": 0.11366868580458256,
      "grad_norm": 2.039644479751587,
      "learning_rate": 4.432314985514881e-05,
      "loss": 1.8762,
      "step": 4316
    },
    {
      "epoch": 0.11369502238609429,
      "grad_norm": 5.607219696044922,
      "learning_rate": 4.4321833026073216e-05,
      "loss": 1.3149,
      "step": 4317
    },
    {
      "epoch": 0.113721358967606,
      "grad_norm": 2.1787233352661133,
      "learning_rate": 4.432051619699763e-05,
      "loss": 2.1058,
      "step": 4318
    },
    {
      "epoch": 0.11374769554911772,
      "grad_norm": 1.7568917274475098,
      "learning_rate": 4.431919936792205e-05,
      "loss": 1.8436,
      "step": 4319
    },
    {
      "epoch": 0.11377403213062945,
      "grad_norm": 3.1339690685272217,
      "learning_rate": 4.431788253884646e-05,
      "loss": 1.667,
      "step": 4320
    },
    {
      "epoch": 0.11380036871214116,
      "grad_norm": 4.216392993927002,
      "learning_rate": 4.431656570977087e-05,
      "loss": 1.0022,
      "step": 4321
    },
    {
      "epoch": 0.11382670529365288,
      "grad_norm": 3.501495122909546,
      "learning_rate": 4.431524888069529e-05,
      "loss": 0.5404,
      "step": 4322
    },
    {
      "epoch": 0.11385304187516461,
      "grad_norm": 2.12831711769104,
      "learning_rate": 4.43139320516197e-05,
      "loss": 1.9329,
      "step": 4323
    },
    {
      "epoch": 0.11387937845667633,
      "grad_norm": 2.5285794734954834,
      "learning_rate": 4.431261522254412e-05,
      "loss": 2.2999,
      "step": 4324
    },
    {
      "epoch": 0.11390571503818804,
      "grad_norm": 3.4024181365966797,
      "learning_rate": 4.4311298393468534e-05,
      "loss": 2.2381,
      "step": 4325
    },
    {
      "epoch": 0.11393205161969976,
      "grad_norm": 3.5027482509613037,
      "learning_rate": 4.430998156439294e-05,
      "loss": 1.9063,
      "step": 4326
    },
    {
      "epoch": 0.11395838820121149,
      "grad_norm": 3.702543258666992,
      "learning_rate": 4.430866473531736e-05,
      "loss": 1.6912,
      "step": 4327
    },
    {
      "epoch": 0.1139847247827232,
      "grad_norm": 3.80104398727417,
      "learning_rate": 4.430734790624177e-05,
      "loss": 1.3014,
      "step": 4328
    },
    {
      "epoch": 0.11401106136423492,
      "grad_norm": 3.2665998935699463,
      "learning_rate": 4.430603107716619e-05,
      "loss": 2.0383,
      "step": 4329
    },
    {
      "epoch": 0.11403739794574665,
      "grad_norm": 2.95882511138916,
      "learning_rate": 4.43047142480906e-05,
      "loss": 1.384,
      "step": 4330
    },
    {
      "epoch": 0.11406373452725836,
      "grad_norm": 2.2256672382354736,
      "learning_rate": 4.4303397419015014e-05,
      "loss": 1.7927,
      "step": 4331
    },
    {
      "epoch": 0.11409007110877008,
      "grad_norm": 2.3855338096618652,
      "learning_rate": 4.430208058993943e-05,
      "loss": 2.0536,
      "step": 4332
    },
    {
      "epoch": 0.11411640769028181,
      "grad_norm": 1.888027548789978,
      "learning_rate": 4.430076376086384e-05,
      "loss": 1.8424,
      "step": 4333
    },
    {
      "epoch": 0.11414274427179352,
      "grad_norm": 6.155549049377441,
      "learning_rate": 4.429944693178826e-05,
      "loss": 0.8036,
      "step": 4334
    },
    {
      "epoch": 0.11416908085330524,
      "grad_norm": 2.398144483566284,
      "learning_rate": 4.429813010271267e-05,
      "loss": 1.5755,
      "step": 4335
    },
    {
      "epoch": 0.11419541743481697,
      "grad_norm": 2.5533294677734375,
      "learning_rate": 4.4296813273637085e-05,
      "loss": 1.4531,
      "step": 4336
    },
    {
      "epoch": 0.11422175401632868,
      "grad_norm": 1.786769986152649,
      "learning_rate": 4.4295496444561494e-05,
      "loss": 2.0445,
      "step": 4337
    },
    {
      "epoch": 0.1142480905978404,
      "grad_norm": 1.889117956161499,
      "learning_rate": 4.4294179615485916e-05,
      "loss": 1.5585,
      "step": 4338
    },
    {
      "epoch": 0.11427442717935211,
      "grad_norm": 5.505272388458252,
      "learning_rate": 4.4292862786410325e-05,
      "loss": 1.442,
      "step": 4339
    },
    {
      "epoch": 0.11430076376086384,
      "grad_norm": 2.247007131576538,
      "learning_rate": 4.429154595733474e-05,
      "loss": 1.902,
      "step": 4340
    },
    {
      "epoch": 0.11432710034237556,
      "grad_norm": 2.9492099285125732,
      "learning_rate": 4.4290229128259156e-05,
      "loss": 2.0655,
      "step": 4341
    },
    {
      "epoch": 0.11435343692388727,
      "grad_norm": 4.861425399780273,
      "learning_rate": 4.4288912299183565e-05,
      "loss": 1.5592,
      "step": 4342
    },
    {
      "epoch": 0.114379773505399,
      "grad_norm": 2.548419952392578,
      "learning_rate": 4.428759547010799e-05,
      "loss": 1.8656,
      "step": 4343
    },
    {
      "epoch": 0.11440611008691072,
      "grad_norm": 3.408047914505005,
      "learning_rate": 4.4286278641032396e-05,
      "loss": 1.0554,
      "step": 4344
    },
    {
      "epoch": 0.11443244666842244,
      "grad_norm": 1.7642710208892822,
      "learning_rate": 4.428496181195681e-05,
      "loss": 1.8416,
      "step": 4345
    },
    {
      "epoch": 0.11445878324993417,
      "grad_norm": 3.3986451625823975,
      "learning_rate": 4.428364498288122e-05,
      "loss": 2.1011,
      "step": 4346
    },
    {
      "epoch": 0.11448511983144588,
      "grad_norm": 2.040459632873535,
      "learning_rate": 4.4282328153805636e-05,
      "loss": 2.2828,
      "step": 4347
    },
    {
      "epoch": 0.1145114564129576,
      "grad_norm": 3.525449275970459,
      "learning_rate": 4.428101132473005e-05,
      "loss": 1.4548,
      "step": 4348
    },
    {
      "epoch": 0.11453779299446931,
      "grad_norm": 2.458203077316284,
      "learning_rate": 4.427969449565447e-05,
      "loss": 2.0462,
      "step": 4349
    },
    {
      "epoch": 0.11456412957598104,
      "grad_norm": 2.7096199989318848,
      "learning_rate": 4.427837766657888e-05,
      "loss": 1.4242,
      "step": 4350
    },
    {
      "epoch": 0.11459046615749276,
      "grad_norm": 2.2402617931365967,
      "learning_rate": 4.427706083750329e-05,
      "loss": 0.9532,
      "step": 4351
    },
    {
      "epoch": 0.11461680273900447,
      "grad_norm": 2.5060954093933105,
      "learning_rate": 4.4275744008427714e-05,
      "loss": 1.7735,
      "step": 4352
    },
    {
      "epoch": 0.1146431393205162,
      "grad_norm": 3.0581908226013184,
      "learning_rate": 4.427442717935212e-05,
      "loss": 1.2129,
      "step": 4353
    },
    {
      "epoch": 0.11466947590202792,
      "grad_norm": 1.8943681716918945,
      "learning_rate": 4.427311035027654e-05,
      "loss": 1.5854,
      "step": 4354
    },
    {
      "epoch": 0.11469581248353963,
      "grad_norm": 1.9518150091171265,
      "learning_rate": 4.427179352120095e-05,
      "loss": 1.3087,
      "step": 4355
    },
    {
      "epoch": 0.11472214906505136,
      "grad_norm": 5.299782752990723,
      "learning_rate": 4.427047669212536e-05,
      "loss": 0.6505,
      "step": 4356
    },
    {
      "epoch": 0.11474848564656308,
      "grad_norm": 5.175239086151123,
      "learning_rate": 4.426915986304978e-05,
      "loss": 2.0356,
      "step": 4357
    },
    {
      "epoch": 0.1147748222280748,
      "grad_norm": 2.3872861862182617,
      "learning_rate": 4.4267843033974194e-05,
      "loss": 1.8172,
      "step": 4358
    },
    {
      "epoch": 0.11480115880958651,
      "grad_norm": 1.7819041013717651,
      "learning_rate": 4.42665262048986e-05,
      "loss": 1.2511,
      "step": 4359
    },
    {
      "epoch": 0.11482749539109824,
      "grad_norm": 2.0215232372283936,
      "learning_rate": 4.426520937582302e-05,
      "loss": 1.1734,
      "step": 4360
    },
    {
      "epoch": 0.11485383197260995,
      "grad_norm": 2.68278431892395,
      "learning_rate": 4.4263892546747434e-05,
      "loss": 1.2245,
      "step": 4361
    },
    {
      "epoch": 0.11488016855412167,
      "grad_norm": 3.159667491912842,
      "learning_rate": 4.426257571767185e-05,
      "loss": 1.6832,
      "step": 4362
    },
    {
      "epoch": 0.1149065051356334,
      "grad_norm": 2.626835584640503,
      "learning_rate": 4.4261258888596265e-05,
      "loss": 1.5326,
      "step": 4363
    },
    {
      "epoch": 0.11493284171714511,
      "grad_norm": 2.1548802852630615,
      "learning_rate": 4.4259942059520674e-05,
      "loss": 1.8406,
      "step": 4364
    },
    {
      "epoch": 0.11495917829865683,
      "grad_norm": 2.756100654602051,
      "learning_rate": 4.425862523044509e-05,
      "loss": 1.5675,
      "step": 4365
    },
    {
      "epoch": 0.11498551488016856,
      "grad_norm": 2.976527214050293,
      "learning_rate": 4.42573084013695e-05,
      "loss": 1.6547,
      "step": 4366
    },
    {
      "epoch": 0.11501185146168028,
      "grad_norm": 2.537472724914551,
      "learning_rate": 4.425599157229392e-05,
      "loss": 1.5997,
      "step": 4367
    },
    {
      "epoch": 0.11503818804319199,
      "grad_norm": 4.277581691741943,
      "learning_rate": 4.425467474321833e-05,
      "loss": 2.3393,
      "step": 4368
    },
    {
      "epoch": 0.1150645246247037,
      "grad_norm": 2.5709848403930664,
      "learning_rate": 4.4253357914142745e-05,
      "loss": 2.4128,
      "step": 4369
    },
    {
      "epoch": 0.11509086120621544,
      "grad_norm": 1.4659626483917236,
      "learning_rate": 4.425204108506716e-05,
      "loss": 0.7346,
      "step": 4370
    },
    {
      "epoch": 0.11511719778772715,
      "grad_norm": 1.9410346746444702,
      "learning_rate": 4.425072425599158e-05,
      "loss": 2.3115,
      "step": 4371
    },
    {
      "epoch": 0.11514353436923887,
      "grad_norm": 3.8114194869995117,
      "learning_rate": 4.424940742691599e-05,
      "loss": 1.9819,
      "step": 4372
    },
    {
      "epoch": 0.1151698709507506,
      "grad_norm": 2.099423885345459,
      "learning_rate": 4.42480905978404e-05,
      "loss": 1.8187,
      "step": 4373
    },
    {
      "epoch": 0.11519620753226231,
      "grad_norm": 5.10236120223999,
      "learning_rate": 4.424677376876482e-05,
      "loss": 1.2605,
      "step": 4374
    },
    {
      "epoch": 0.11522254411377403,
      "grad_norm": 6.026031494140625,
      "learning_rate": 4.4245456939689226e-05,
      "loss": 1.1565,
      "step": 4375
    },
    {
      "epoch": 0.11524888069528576,
      "grad_norm": 3.2395246028900146,
      "learning_rate": 4.424414011061365e-05,
      "loss": 1.5653,
      "step": 4376
    },
    {
      "epoch": 0.11527521727679747,
      "grad_norm": 4.563273906707764,
      "learning_rate": 4.424282328153806e-05,
      "loss": 2.3523,
      "step": 4377
    },
    {
      "epoch": 0.11530155385830919,
      "grad_norm": 1.7929317951202393,
      "learning_rate": 4.424150645246247e-05,
      "loss": 2.1543,
      "step": 4378
    },
    {
      "epoch": 0.11532789043982092,
      "grad_norm": 1.9524186849594116,
      "learning_rate": 4.424018962338689e-05,
      "loss": 1.9696,
      "step": 4379
    },
    {
      "epoch": 0.11535422702133263,
      "grad_norm": 2.2234647274017334,
      "learning_rate": 4.42388727943113e-05,
      "loss": 2.6179,
      "step": 4380
    },
    {
      "epoch": 0.11538056360284435,
      "grad_norm": 2.636587619781494,
      "learning_rate": 4.423755596523572e-05,
      "loss": 2.3227,
      "step": 4381
    },
    {
      "epoch": 0.11540690018435606,
      "grad_norm": 3.5364274978637695,
      "learning_rate": 4.423623913616013e-05,
      "loss": 1.9982,
      "step": 4382
    },
    {
      "epoch": 0.1154332367658678,
      "grad_norm": 1.893670678138733,
      "learning_rate": 4.4234922307084544e-05,
      "loss": 1.6248,
      "step": 4383
    },
    {
      "epoch": 0.11545957334737951,
      "grad_norm": 2.328727960586548,
      "learning_rate": 4.423360547800895e-05,
      "loss": 1.0495,
      "step": 4384
    },
    {
      "epoch": 0.11548590992889123,
      "grad_norm": 2.2435641288757324,
      "learning_rate": 4.4232288648933375e-05,
      "loss": 0.8707,
      "step": 4385
    },
    {
      "epoch": 0.11551224651040295,
      "grad_norm": 3.589625358581543,
      "learning_rate": 4.4230971819857784e-05,
      "loss": 0.4759,
      "step": 4386
    },
    {
      "epoch": 0.11553858309191467,
      "grad_norm": 1.8171758651733398,
      "learning_rate": 4.42296549907822e-05,
      "loss": 1.389,
      "step": 4387
    },
    {
      "epoch": 0.11556491967342639,
      "grad_norm": 3.317521333694458,
      "learning_rate": 4.4228338161706615e-05,
      "loss": 1.5492,
      "step": 4388
    },
    {
      "epoch": 0.11559125625493812,
      "grad_norm": 3.8359012603759766,
      "learning_rate": 4.4227021332631024e-05,
      "loss": 1.8196,
      "step": 4389
    },
    {
      "epoch": 0.11561759283644983,
      "grad_norm": 1.9237251281738281,
      "learning_rate": 4.4225704503555446e-05,
      "loss": 1.737,
      "step": 4390
    },
    {
      "epoch": 0.11564392941796155,
      "grad_norm": 2.886448383331299,
      "learning_rate": 4.4224387674479855e-05,
      "loss": 1.9436,
      "step": 4391
    },
    {
      "epoch": 0.11567026599947326,
      "grad_norm": 4.651662826538086,
      "learning_rate": 4.422307084540427e-05,
      "loss": 1.0072,
      "step": 4392
    },
    {
      "epoch": 0.11569660258098499,
      "grad_norm": 2.4865639209747314,
      "learning_rate": 4.422175401632868e-05,
      "loss": 2.3585,
      "step": 4393
    },
    {
      "epoch": 0.11572293916249671,
      "grad_norm": 1.986808180809021,
      "learning_rate": 4.4220437187253095e-05,
      "loss": 1.8663,
      "step": 4394
    },
    {
      "epoch": 0.11574927574400842,
      "grad_norm": 2.155029058456421,
      "learning_rate": 4.421912035817751e-05,
      "loss": 2.2016,
      "step": 4395
    },
    {
      "epoch": 0.11577561232552015,
      "grad_norm": 2.616759777069092,
      "learning_rate": 4.4217803529101926e-05,
      "loss": 2.0208,
      "step": 4396
    },
    {
      "epoch": 0.11580194890703187,
      "grad_norm": 3.513251543045044,
      "learning_rate": 4.421648670002634e-05,
      "loss": 1.2551,
      "step": 4397
    },
    {
      "epoch": 0.11582828548854358,
      "grad_norm": 1.8409112691879272,
      "learning_rate": 4.421516987095075e-05,
      "loss": 1.3185,
      "step": 4398
    },
    {
      "epoch": 0.11585462207005531,
      "grad_norm": 2.120098352432251,
      "learning_rate": 4.4213853041875166e-05,
      "loss": 2.6171,
      "step": 4399
    },
    {
      "epoch": 0.11588095865156703,
      "grad_norm": 2.0425994396209717,
      "learning_rate": 4.421253621279958e-05,
      "loss": 0.8799,
      "step": 4400
    },
    {
      "epoch": 0.11590729523307874,
      "grad_norm": 3.7647507190704346,
      "learning_rate": 4.4211219383724e-05,
      "loss": 1.5859,
      "step": 4401
    },
    {
      "epoch": 0.11593363181459046,
      "grad_norm": 2.8809452056884766,
      "learning_rate": 4.4209902554648406e-05,
      "loss": 2.4356,
      "step": 4402
    },
    {
      "epoch": 0.11595996839610219,
      "grad_norm": 1.9078770875930786,
      "learning_rate": 4.420858572557282e-05,
      "loss": 1.5714,
      "step": 4403
    },
    {
      "epoch": 0.1159863049776139,
      "grad_norm": 3.08406662940979,
      "learning_rate": 4.420726889649724e-05,
      "loss": 2.1748,
      "step": 4404
    },
    {
      "epoch": 0.11601264155912562,
      "grad_norm": 1.6534942388534546,
      "learning_rate": 4.420595206742165e-05,
      "loss": 1.8346,
      "step": 4405
    },
    {
      "epoch": 0.11603897814063735,
      "grad_norm": 1.8040796518325806,
      "learning_rate": 4.420463523834606e-05,
      "loss": 2.1775,
      "step": 4406
    },
    {
      "epoch": 0.11606531472214907,
      "grad_norm": 3.455887794494629,
      "learning_rate": 4.420331840927048e-05,
      "loss": 0.9073,
      "step": 4407
    },
    {
      "epoch": 0.11609165130366078,
      "grad_norm": 3.195606231689453,
      "learning_rate": 4.420200158019489e-05,
      "loss": 1.8064,
      "step": 4408
    },
    {
      "epoch": 0.11611798788517251,
      "grad_norm": 2.3142263889312744,
      "learning_rate": 4.420068475111931e-05,
      "loss": 1.9647,
      "step": 4409
    },
    {
      "epoch": 0.11614432446668423,
      "grad_norm": 2.7577309608459473,
      "learning_rate": 4.4199367922043724e-05,
      "loss": 2.1959,
      "step": 4410
    },
    {
      "epoch": 0.11617066104819594,
      "grad_norm": 2.356194257736206,
      "learning_rate": 4.419805109296813e-05,
      "loss": 2.4659,
      "step": 4411
    },
    {
      "epoch": 0.11619699762970766,
      "grad_norm": 2.681997537612915,
      "learning_rate": 4.419673426389255e-05,
      "loss": 1.7866,
      "step": 4412
    },
    {
      "epoch": 0.11622333421121939,
      "grad_norm": 2.571408987045288,
      "learning_rate": 4.419541743481696e-05,
      "loss": 1.5353,
      "step": 4413
    },
    {
      "epoch": 0.1162496707927311,
      "grad_norm": 2.0447728633880615,
      "learning_rate": 4.419410060574138e-05,
      "loss": 1.4663,
      "step": 4414
    },
    {
      "epoch": 0.11627600737424282,
      "grad_norm": 3.892343044281006,
      "learning_rate": 4.419278377666579e-05,
      "loss": 1.7325,
      "step": 4415
    },
    {
      "epoch": 0.11630234395575455,
      "grad_norm": 3.722790241241455,
      "learning_rate": 4.4191466947590204e-05,
      "loss": 2.0224,
      "step": 4416
    },
    {
      "epoch": 0.11632868053726626,
      "grad_norm": 3.7777955532073975,
      "learning_rate": 4.419015011851462e-05,
      "loss": 1.4628,
      "step": 4417
    },
    {
      "epoch": 0.11635501711877798,
      "grad_norm": 1.8749061822891235,
      "learning_rate": 4.4188833289439035e-05,
      "loss": 2.4254,
      "step": 4418
    },
    {
      "epoch": 0.11638135370028971,
      "grad_norm": 4.084940433502197,
      "learning_rate": 4.418751646036345e-05,
      "loss": 1.0573,
      "step": 4419
    },
    {
      "epoch": 0.11640769028180142,
      "grad_norm": 1.9295003414154053,
      "learning_rate": 4.418619963128786e-05,
      "loss": 0.38,
      "step": 4420
    },
    {
      "epoch": 0.11643402686331314,
      "grad_norm": 3.350985288619995,
      "learning_rate": 4.4184882802212275e-05,
      "loss": 2.4217,
      "step": 4421
    },
    {
      "epoch": 0.11646036344482485,
      "grad_norm": 3.1467080116271973,
      "learning_rate": 4.4183565973136684e-05,
      "loss": 1.3987,
      "step": 4422
    },
    {
      "epoch": 0.11648670002633658,
      "grad_norm": 2.770132303237915,
      "learning_rate": 4.4182249144061106e-05,
      "loss": 1.2506,
      "step": 4423
    },
    {
      "epoch": 0.1165130366078483,
      "grad_norm": 3.275036334991455,
      "learning_rate": 4.4180932314985515e-05,
      "loss": 1.0688,
      "step": 4424
    },
    {
      "epoch": 0.11653937318936002,
      "grad_norm": 2.3301479816436768,
      "learning_rate": 4.417961548590993e-05,
      "loss": 1.8775,
      "step": 4425
    },
    {
      "epoch": 0.11656570977087174,
      "grad_norm": 1.8609044551849365,
      "learning_rate": 4.4178298656834346e-05,
      "loss": 2.3729,
      "step": 4426
    },
    {
      "epoch": 0.11659204635238346,
      "grad_norm": 2.461369276046753,
      "learning_rate": 4.4176981827758755e-05,
      "loss": 1.6169,
      "step": 4427
    },
    {
      "epoch": 0.11661838293389518,
      "grad_norm": 2.050262212753296,
      "learning_rate": 4.417566499868318e-05,
      "loss": 1.8469,
      "step": 4428
    },
    {
      "epoch": 0.1166447195154069,
      "grad_norm": 3.288084030151367,
      "learning_rate": 4.4174348169607586e-05,
      "loss": 1.7402,
      "step": 4429
    },
    {
      "epoch": 0.11667105609691862,
      "grad_norm": 1.8451225757598877,
      "learning_rate": 4.4173031340532e-05,
      "loss": 1.852,
      "step": 4430
    },
    {
      "epoch": 0.11669739267843034,
      "grad_norm": 2.001999616622925,
      "learning_rate": 4.417171451145641e-05,
      "loss": 0.3579,
      "step": 4431
    },
    {
      "epoch": 0.11672372925994207,
      "grad_norm": 4.541996002197266,
      "learning_rate": 4.4170397682380827e-05,
      "loss": 1.3137,
      "step": 4432
    },
    {
      "epoch": 0.11675006584145378,
      "grad_norm": 1.6650617122650146,
      "learning_rate": 4.416908085330524e-05,
      "loss": 1.6788,
      "step": 4433
    },
    {
      "epoch": 0.1167764024229655,
      "grad_norm": 4.194488525390625,
      "learning_rate": 4.416776402422966e-05,
      "loss": 1.2758,
      "step": 4434
    },
    {
      "epoch": 0.11680273900447721,
      "grad_norm": 2.6946659088134766,
      "learning_rate": 4.416644719515407e-05,
      "loss": 2.0421,
      "step": 4435
    },
    {
      "epoch": 0.11682907558598894,
      "grad_norm": 2.4091713428497314,
      "learning_rate": 4.416513036607848e-05,
      "loss": 2.1079,
      "step": 4436
    },
    {
      "epoch": 0.11685541216750066,
      "grad_norm": 3.072972297668457,
      "learning_rate": 4.4163813537002904e-05,
      "loss": 1.7528,
      "step": 4437
    },
    {
      "epoch": 0.11688174874901237,
      "grad_norm": 4.11431884765625,
      "learning_rate": 4.416249670792731e-05,
      "loss": 2.0374,
      "step": 4438
    },
    {
      "epoch": 0.1169080853305241,
      "grad_norm": 2.783322334289551,
      "learning_rate": 4.416117987885173e-05,
      "loss": 1.8257,
      "step": 4439
    },
    {
      "epoch": 0.11693442191203582,
      "grad_norm": 1.4758005142211914,
      "learning_rate": 4.415986304977614e-05,
      "loss": 1.4166,
      "step": 4440
    },
    {
      "epoch": 0.11696075849354753,
      "grad_norm": 2.110168933868408,
      "learning_rate": 4.415854622070055e-05,
      "loss": 1.9646,
      "step": 4441
    },
    {
      "epoch": 0.11698709507505926,
      "grad_norm": 2.3040671348571777,
      "learning_rate": 4.415722939162497e-05,
      "loss": 2.2283,
      "step": 4442
    },
    {
      "epoch": 0.11701343165657098,
      "grad_norm": 4.929816722869873,
      "learning_rate": 4.4155912562549385e-05,
      "loss": 0.9143,
      "step": 4443
    },
    {
      "epoch": 0.1170397682380827,
      "grad_norm": 1.7468531131744385,
      "learning_rate": 4.41545957334738e-05,
      "loss": 1.6556,
      "step": 4444
    },
    {
      "epoch": 0.11706610481959441,
      "grad_norm": 5.195878028869629,
      "learning_rate": 4.415327890439821e-05,
      "loss": 2.1819,
      "step": 4445
    },
    {
      "epoch": 0.11709244140110614,
      "grad_norm": 2.4151008129119873,
      "learning_rate": 4.4151962075322625e-05,
      "loss": 2.0081,
      "step": 4446
    },
    {
      "epoch": 0.11711877798261786,
      "grad_norm": 2.482985496520996,
      "learning_rate": 4.415064524624704e-05,
      "loss": 1.6568,
      "step": 4447
    },
    {
      "epoch": 0.11714511456412957,
      "grad_norm": 2.796194314956665,
      "learning_rate": 4.4149328417171456e-05,
      "loss": 1.6216,
      "step": 4448
    },
    {
      "epoch": 0.1171714511456413,
      "grad_norm": 2.431485414505005,
      "learning_rate": 4.4148011588095865e-05,
      "loss": 1.9602,
      "step": 4449
    },
    {
      "epoch": 0.11719778772715302,
      "grad_norm": 2.156215190887451,
      "learning_rate": 4.414669475902028e-05,
      "loss": 1.5413,
      "step": 4450
    },
    {
      "epoch": 0.11722412430866473,
      "grad_norm": 5.469089031219482,
      "learning_rate": 4.4145377929944696e-05,
      "loss": 0.6391,
      "step": 4451
    },
    {
      "epoch": 0.11725046089017646,
      "grad_norm": 2.213456392288208,
      "learning_rate": 4.414406110086911e-05,
      "loss": 1.3894,
      "step": 4452
    },
    {
      "epoch": 0.11727679747168818,
      "grad_norm": 3.5617663860321045,
      "learning_rate": 4.414274427179352e-05,
      "loss": 1.3308,
      "step": 4453
    },
    {
      "epoch": 0.11730313405319989,
      "grad_norm": 2.372602701187134,
      "learning_rate": 4.4141427442717936e-05,
      "loss": 1.6788,
      "step": 4454
    },
    {
      "epoch": 0.11732947063471161,
      "grad_norm": 2.067337989807129,
      "learning_rate": 4.414011061364235e-05,
      "loss": 1.0302,
      "step": 4455
    },
    {
      "epoch": 0.11735580721622334,
      "grad_norm": 2.4312102794647217,
      "learning_rate": 4.413879378456677e-05,
      "loss": 1.8871,
      "step": 4456
    },
    {
      "epoch": 0.11738214379773505,
      "grad_norm": 3.186422348022461,
      "learning_rate": 4.413747695549118e-05,
      "loss": 1.5799,
      "step": 4457
    },
    {
      "epoch": 0.11740848037924677,
      "grad_norm": 2.366577625274658,
      "learning_rate": 4.413616012641559e-05,
      "loss": 3.096,
      "step": 4458
    },
    {
      "epoch": 0.1174348169607585,
      "grad_norm": 2.444317102432251,
      "learning_rate": 4.413484329734001e-05,
      "loss": 1.997,
      "step": 4459
    },
    {
      "epoch": 0.11746115354227021,
      "grad_norm": 2.177645206451416,
      "learning_rate": 4.4133526468264416e-05,
      "loss": 1.3416,
      "step": 4460
    },
    {
      "epoch": 0.11748749012378193,
      "grad_norm": 3.835712194442749,
      "learning_rate": 4.413220963918884e-05,
      "loss": 1.3609,
      "step": 4461
    },
    {
      "epoch": 0.11751382670529366,
      "grad_norm": 1.9772382974624634,
      "learning_rate": 4.413089281011325e-05,
      "loss": 2.2611,
      "step": 4462
    },
    {
      "epoch": 0.11754016328680537,
      "grad_norm": 2.3028218746185303,
      "learning_rate": 4.412957598103766e-05,
      "loss": 0.8586,
      "step": 4463
    },
    {
      "epoch": 0.11756649986831709,
      "grad_norm": 2.191162586212158,
      "learning_rate": 4.412825915196208e-05,
      "loss": 2.1453,
      "step": 4464
    },
    {
      "epoch": 0.1175928364498288,
      "grad_norm": 3.3839304447174072,
      "learning_rate": 4.4126942322886494e-05,
      "loss": 1.7196,
      "step": 4465
    },
    {
      "epoch": 0.11761917303134053,
      "grad_norm": 2.2345473766326904,
      "learning_rate": 4.412562549381091e-05,
      "loss": 1.9077,
      "step": 4466
    },
    {
      "epoch": 0.11764550961285225,
      "grad_norm": 5.274810314178467,
      "learning_rate": 4.412430866473532e-05,
      "loss": 2.0036,
      "step": 4467
    },
    {
      "epoch": 0.11767184619436397,
      "grad_norm": 3.1181788444519043,
      "learning_rate": 4.4122991835659734e-05,
      "loss": 1.215,
      "step": 4468
    },
    {
      "epoch": 0.1176981827758757,
      "grad_norm": 1.5179028511047363,
      "learning_rate": 4.412167500658414e-05,
      "loss": 0.4065,
      "step": 4469
    },
    {
      "epoch": 0.11772451935738741,
      "grad_norm": 3.7254276275634766,
      "learning_rate": 4.4120358177508565e-05,
      "loss": 2.488,
      "step": 4470
    },
    {
      "epoch": 0.11775085593889913,
      "grad_norm": 2.8294014930725098,
      "learning_rate": 4.4119041348432974e-05,
      "loss": 0.8961,
      "step": 4471
    },
    {
      "epoch": 0.11777719252041086,
      "grad_norm": 3.7836203575134277,
      "learning_rate": 4.411772451935739e-05,
      "loss": 0.4027,
      "step": 4472
    },
    {
      "epoch": 0.11780352910192257,
      "grad_norm": 3.876450777053833,
      "learning_rate": 4.4116407690281805e-05,
      "loss": 0.9419,
      "step": 4473
    },
    {
      "epoch": 0.11782986568343429,
      "grad_norm": 1.7913403511047363,
      "learning_rate": 4.4115090861206214e-05,
      "loss": 1.9551,
      "step": 4474
    },
    {
      "epoch": 0.11785620226494602,
      "grad_norm": 1.810213565826416,
      "learning_rate": 4.4113774032130636e-05,
      "loss": 2.0296,
      "step": 4475
    },
    {
      "epoch": 0.11788253884645773,
      "grad_norm": 1.8780161142349243,
      "learning_rate": 4.4112457203055045e-05,
      "loss": 1.8583,
      "step": 4476
    },
    {
      "epoch": 0.11790887542796945,
      "grad_norm": 1.73005211353302,
      "learning_rate": 4.411114037397946e-05,
      "loss": 1.7836,
      "step": 4477
    },
    {
      "epoch": 0.11793521200948116,
      "grad_norm": 2.005183458328247,
      "learning_rate": 4.410982354490387e-05,
      "loss": 1.4563,
      "step": 4478
    },
    {
      "epoch": 0.11796154859099289,
      "grad_norm": 2.8055245876312256,
      "learning_rate": 4.4108506715828285e-05,
      "loss": 1.2794,
      "step": 4479
    },
    {
      "epoch": 0.11798788517250461,
      "grad_norm": 3.500542163848877,
      "learning_rate": 4.41071898867527e-05,
      "loss": 1.1338,
      "step": 4480
    },
    {
      "epoch": 0.11801422175401632,
      "grad_norm": 3.309410333633423,
      "learning_rate": 4.4105873057677116e-05,
      "loss": 1.0386,
      "step": 4481
    },
    {
      "epoch": 0.11804055833552805,
      "grad_norm": 1.834859848022461,
      "learning_rate": 4.410455622860153e-05,
      "loss": 2.4931,
      "step": 4482
    },
    {
      "epoch": 0.11806689491703977,
      "grad_norm": 3.1914002895355225,
      "learning_rate": 4.410323939952594e-05,
      "loss": 1.3354,
      "step": 4483
    },
    {
      "epoch": 0.11809323149855148,
      "grad_norm": 3.621843099594116,
      "learning_rate": 4.410192257045036e-05,
      "loss": 1.3452,
      "step": 4484
    },
    {
      "epoch": 0.11811956808006321,
      "grad_norm": 2.77040696144104,
      "learning_rate": 4.410060574137477e-05,
      "loss": 0.4439,
      "step": 4485
    },
    {
      "epoch": 0.11814590466157493,
      "grad_norm": 1.976322054862976,
      "learning_rate": 4.409928891229919e-05,
      "loss": 2.6194,
      "step": 4486
    },
    {
      "epoch": 0.11817224124308665,
      "grad_norm": 1.6605380773544312,
      "learning_rate": 4.4097972083223596e-05,
      "loss": 1.6788,
      "step": 4487
    },
    {
      "epoch": 0.11819857782459836,
      "grad_norm": 1.6394164562225342,
      "learning_rate": 4.409665525414801e-05,
      "loss": 2.0224,
      "step": 4488
    },
    {
      "epoch": 0.11822491440611009,
      "grad_norm": 4.603525161743164,
      "learning_rate": 4.409533842507243e-05,
      "loss": 0.6049,
      "step": 4489
    },
    {
      "epoch": 0.1182512509876218,
      "grad_norm": 2.52878999710083,
      "learning_rate": 4.409402159599684e-05,
      "loss": 1.1091,
      "step": 4490
    },
    {
      "epoch": 0.11827758756913352,
      "grad_norm": 2.8281538486480713,
      "learning_rate": 4.409270476692126e-05,
      "loss": 1.707,
      "step": 4491
    },
    {
      "epoch": 0.11830392415064525,
      "grad_norm": 2.3989648818969727,
      "learning_rate": 4.409138793784567e-05,
      "loss": 1.2226,
      "step": 4492
    },
    {
      "epoch": 0.11833026073215697,
      "grad_norm": 2.66068172454834,
      "learning_rate": 4.409007110877008e-05,
      "loss": 1.646,
      "step": 4493
    },
    {
      "epoch": 0.11835659731366868,
      "grad_norm": 6.114838600158691,
      "learning_rate": 4.40887542796945e-05,
      "loss": 1.6339,
      "step": 4494
    },
    {
      "epoch": 0.11838293389518041,
      "grad_norm": 2.7871336936950684,
      "learning_rate": 4.4087437450618914e-05,
      "loss": 1.2906,
      "step": 4495
    },
    {
      "epoch": 0.11840927047669213,
      "grad_norm": 2.0278446674346924,
      "learning_rate": 4.408612062154332e-05,
      "loss": 1.9115,
      "step": 4496
    },
    {
      "epoch": 0.11843560705820384,
      "grad_norm": 3.571059226989746,
      "learning_rate": 4.408480379246774e-05,
      "loss": 0.5902,
      "step": 4497
    },
    {
      "epoch": 0.11846194363971556,
      "grad_norm": 3.208005666732788,
      "learning_rate": 4.4083486963392154e-05,
      "loss": 1.744,
      "step": 4498
    },
    {
      "epoch": 0.11848828022122729,
      "grad_norm": 3.4264402389526367,
      "learning_rate": 4.408217013431657e-05,
      "loss": 1.8997,
      "step": 4499
    },
    {
      "epoch": 0.118514616802739,
      "grad_norm": 2.0087668895721436,
      "learning_rate": 4.4080853305240986e-05,
      "loss": 0.9058,
      "step": 4500
    },
    {
      "epoch": 0.11854095338425072,
      "grad_norm": 2.763491630554199,
      "learning_rate": 4.4079536476165394e-05,
      "loss": 0.5899,
      "step": 4501
    },
    {
      "epoch": 0.11856728996576245,
      "grad_norm": 1.5228254795074463,
      "learning_rate": 4.407821964708981e-05,
      "loss": 1.2187,
      "step": 4502
    },
    {
      "epoch": 0.11859362654727416,
      "grad_norm": 4.362586975097656,
      "learning_rate": 4.4076902818014226e-05,
      "loss": 0.6471,
      "step": 4503
    },
    {
      "epoch": 0.11861996312878588,
      "grad_norm": 3.0736706256866455,
      "learning_rate": 4.407558598893864e-05,
      "loss": 1.9065,
      "step": 4504
    },
    {
      "epoch": 0.11864629971029761,
      "grad_norm": 2.788661479949951,
      "learning_rate": 4.407426915986305e-05,
      "loss": 1.7559,
      "step": 4505
    },
    {
      "epoch": 0.11867263629180932,
      "grad_norm": 2.1233596801757812,
      "learning_rate": 4.4072952330787466e-05,
      "loss": 1.9864,
      "step": 4506
    },
    {
      "epoch": 0.11869897287332104,
      "grad_norm": 2.975221872329712,
      "learning_rate": 4.4071635501711874e-05,
      "loss": 1.6282,
      "step": 4507
    },
    {
      "epoch": 0.11872530945483276,
      "grad_norm": 1.9074469804763794,
      "learning_rate": 4.40703186726363e-05,
      "loss": 1.7752,
      "step": 4508
    },
    {
      "epoch": 0.11875164603634449,
      "grad_norm": 2.320554733276367,
      "learning_rate": 4.4069001843560706e-05,
      "loss": 1.2108,
      "step": 4509
    },
    {
      "epoch": 0.1187779826178562,
      "grad_norm": 2.093651533126831,
      "learning_rate": 4.406768501448512e-05,
      "loss": 1.5476,
      "step": 4510
    },
    {
      "epoch": 0.11880431919936792,
      "grad_norm": 1.9673291444778442,
      "learning_rate": 4.406636818540954e-05,
      "loss": 1.1547,
      "step": 4511
    },
    {
      "epoch": 0.11883065578087965,
      "grad_norm": 2.612429141998291,
      "learning_rate": 4.4065051356333946e-05,
      "loss": 1.7214,
      "step": 4512
    },
    {
      "epoch": 0.11885699236239136,
      "grad_norm": 2.106727361679077,
      "learning_rate": 4.406373452725837e-05,
      "loss": 1.6774,
      "step": 4513
    },
    {
      "epoch": 0.11888332894390308,
      "grad_norm": 1.991256594657898,
      "learning_rate": 4.406241769818278e-05,
      "loss": 2.1493,
      "step": 4514
    },
    {
      "epoch": 0.1189096655254148,
      "grad_norm": 2.3556244373321533,
      "learning_rate": 4.406110086910719e-05,
      "loss": 2.2043,
      "step": 4515
    },
    {
      "epoch": 0.11893600210692652,
      "grad_norm": 7.154943943023682,
      "learning_rate": 4.40597840400316e-05,
      "loss": 2.2684,
      "step": 4516
    },
    {
      "epoch": 0.11896233868843824,
      "grad_norm": 4.374931335449219,
      "learning_rate": 4.4058467210956024e-05,
      "loss": 0.9902,
      "step": 4517
    },
    {
      "epoch": 0.11898867526994997,
      "grad_norm": 2.3428380489349365,
      "learning_rate": 4.405715038188043e-05,
      "loss": 1.4315,
      "step": 4518
    },
    {
      "epoch": 0.11901501185146168,
      "grad_norm": 2.3284173011779785,
      "learning_rate": 4.405583355280485e-05,
      "loss": 1.6703,
      "step": 4519
    },
    {
      "epoch": 0.1190413484329734,
      "grad_norm": 1.9064738750457764,
      "learning_rate": 4.4054516723729264e-05,
      "loss": 1.3962,
      "step": 4520
    },
    {
      "epoch": 0.11906768501448511,
      "grad_norm": 6.036314010620117,
      "learning_rate": 4.405319989465367e-05,
      "loss": 1.7169,
      "step": 4521
    },
    {
      "epoch": 0.11909402159599684,
      "grad_norm": 2.0272128582000732,
      "learning_rate": 4.4051883065578095e-05,
      "loss": 1.6366,
      "step": 4522
    },
    {
      "epoch": 0.11912035817750856,
      "grad_norm": 2.339869976043701,
      "learning_rate": 4.4050566236502504e-05,
      "loss": 1.4492,
      "step": 4523
    },
    {
      "epoch": 0.11914669475902027,
      "grad_norm": 5.801568508148193,
      "learning_rate": 4.404924940742692e-05,
      "loss": 1.7531,
      "step": 4524
    },
    {
      "epoch": 0.119173031340532,
      "grad_norm": 2.9039807319641113,
      "learning_rate": 4.404793257835133e-05,
      "loss": 1.3748,
      "step": 4525
    },
    {
      "epoch": 0.11919936792204372,
      "grad_norm": 2.5488572120666504,
      "learning_rate": 4.4046615749275744e-05,
      "loss": 1.5856,
      "step": 4526
    },
    {
      "epoch": 0.11922570450355544,
      "grad_norm": 2.6418426036834717,
      "learning_rate": 4.404529892020016e-05,
      "loss": 0.3907,
      "step": 4527
    },
    {
      "epoch": 0.11925204108506716,
      "grad_norm": 3.457848072052002,
      "learning_rate": 4.4043982091124575e-05,
      "loss": 1.5812,
      "step": 4528
    },
    {
      "epoch": 0.11927837766657888,
      "grad_norm": 2.1689720153808594,
      "learning_rate": 4.404266526204899e-05,
      "loss": 2.193,
      "step": 4529
    },
    {
      "epoch": 0.1193047142480906,
      "grad_norm": 1.7394002676010132,
      "learning_rate": 4.40413484329734e-05,
      "loss": 1.5981,
      "step": 4530
    },
    {
      "epoch": 0.11933105082960231,
      "grad_norm": 2.2339563369750977,
      "learning_rate": 4.404003160389782e-05,
      "loss": 2.0593,
      "step": 4531
    },
    {
      "epoch": 0.11935738741111404,
      "grad_norm": 2.624907970428467,
      "learning_rate": 4.403871477482223e-05,
      "loss": 1.7177,
      "step": 4532
    },
    {
      "epoch": 0.11938372399262576,
      "grad_norm": 3.305166482925415,
      "learning_rate": 4.4037397945746646e-05,
      "loss": 1.1525,
      "step": 4533
    },
    {
      "epoch": 0.11941006057413747,
      "grad_norm": 2.7178993225097656,
      "learning_rate": 4.4036081116671055e-05,
      "loss": 2.3004,
      "step": 4534
    },
    {
      "epoch": 0.1194363971556492,
      "grad_norm": 2.7217884063720703,
      "learning_rate": 4.403476428759547e-05,
      "loss": 1.9671,
      "step": 4535
    },
    {
      "epoch": 0.11946273373716092,
      "grad_norm": 5.2226362228393555,
      "learning_rate": 4.4033447458519886e-05,
      "loss": 1.4713,
      "step": 4536
    },
    {
      "epoch": 0.11948907031867263,
      "grad_norm": 2.570767641067505,
      "learning_rate": 4.40321306294443e-05,
      "loss": 1.4283,
      "step": 4537
    },
    {
      "epoch": 0.11951540690018436,
      "grad_norm": 3.9101603031158447,
      "learning_rate": 4.403081380036872e-05,
      "loss": 0.804,
      "step": 4538
    },
    {
      "epoch": 0.11954174348169608,
      "grad_norm": 2.461411237716675,
      "learning_rate": 4.4029496971293126e-05,
      "loss": 0.8779,
      "step": 4539
    },
    {
      "epoch": 0.1195680800632078,
      "grad_norm": 2.4096240997314453,
      "learning_rate": 4.402818014221754e-05,
      "loss": 2.0811,
      "step": 4540
    },
    {
      "epoch": 0.11959441664471951,
      "grad_norm": 5.311163425445557,
      "learning_rate": 4.402686331314196e-05,
      "loss": 2.0921,
      "step": 4541
    },
    {
      "epoch": 0.11962075322623124,
      "grad_norm": 2.9684927463531494,
      "learning_rate": 4.402554648406637e-05,
      "loss": 1.413,
      "step": 4542
    },
    {
      "epoch": 0.11964708980774295,
      "grad_norm": 4.3449225425720215,
      "learning_rate": 4.402422965499078e-05,
      "loss": 1.9694,
      "step": 4543
    },
    {
      "epoch": 0.11967342638925467,
      "grad_norm": 2.3601181507110596,
      "learning_rate": 4.40229128259152e-05,
      "loss": 2.0336,
      "step": 4544
    },
    {
      "epoch": 0.1196997629707664,
      "grad_norm": 2.681947708129883,
      "learning_rate": 4.402159599683961e-05,
      "loss": 1.2407,
      "step": 4545
    },
    {
      "epoch": 0.11972609955227811,
      "grad_norm": 3.324223279953003,
      "learning_rate": 4.402027916776403e-05,
      "loss": 1.5159,
      "step": 4546
    },
    {
      "epoch": 0.11975243613378983,
      "grad_norm": 1.7762402296066284,
      "learning_rate": 4.4018962338688444e-05,
      "loss": 2.0747,
      "step": 4547
    },
    {
      "epoch": 0.11977877271530156,
      "grad_norm": 2.1038661003112793,
      "learning_rate": 4.401764550961285e-05,
      "loss": 1.3491,
      "step": 4548
    },
    {
      "epoch": 0.11980510929681328,
      "grad_norm": 2.5586345195770264,
      "learning_rate": 4.401632868053727e-05,
      "loss": 1.7166,
      "step": 4549
    },
    {
      "epoch": 0.11983144587832499,
      "grad_norm": 2.3258981704711914,
      "learning_rate": 4.4015011851461684e-05,
      "loss": 1.2625,
      "step": 4550
    },
    {
      "epoch": 0.1198577824598367,
      "grad_norm": 2.0945794582366943,
      "learning_rate": 4.40136950223861e-05,
      "loss": 1.7931,
      "step": 4551
    },
    {
      "epoch": 0.11988411904134844,
      "grad_norm": 4.948436260223389,
      "learning_rate": 4.401237819331051e-05,
      "loss": 1.2743,
      "step": 4552
    },
    {
      "epoch": 0.11991045562286015,
      "grad_norm": 2.5348498821258545,
      "learning_rate": 4.4011061364234924e-05,
      "loss": 2.0265,
      "step": 4553
    },
    {
      "epoch": 0.11993679220437187,
      "grad_norm": 2.0693464279174805,
      "learning_rate": 4.400974453515933e-05,
      "loss": 1.8904,
      "step": 4554
    },
    {
      "epoch": 0.1199631287858836,
      "grad_norm": 3.35764217376709,
      "learning_rate": 4.4008427706083755e-05,
      "loss": 1.615,
      "step": 4555
    },
    {
      "epoch": 0.11998946536739531,
      "grad_norm": 1.5819950103759766,
      "learning_rate": 4.4007110877008164e-05,
      "loss": 2.043,
      "step": 4556
    },
    {
      "epoch": 0.12001580194890703,
      "grad_norm": 2.327550172805786,
      "learning_rate": 4.400579404793258e-05,
      "loss": 1.8321,
      "step": 4557
    },
    {
      "epoch": 0.12004213853041876,
      "grad_norm": 1.9448949098587036,
      "learning_rate": 4.4004477218856995e-05,
      "loss": 2.3237,
      "step": 4558
    },
    {
      "epoch": 0.12006847511193047,
      "grad_norm": 1.9863765239715576,
      "learning_rate": 4.4003160389781404e-05,
      "loss": 0.6008,
      "step": 4559
    },
    {
      "epoch": 0.12009481169344219,
      "grad_norm": 2.417666435241699,
      "learning_rate": 4.4001843560705827e-05,
      "loss": 0.5446,
      "step": 4560
    },
    {
      "epoch": 0.12012114827495392,
      "grad_norm": 2.5216031074523926,
      "learning_rate": 4.4000526731630235e-05,
      "loss": 2.6096,
      "step": 4561
    },
    {
      "epoch": 0.12014748485646563,
      "grad_norm": 4.250596523284912,
      "learning_rate": 4.399920990255465e-05,
      "loss": 1.689,
      "step": 4562
    },
    {
      "epoch": 0.12017382143797735,
      "grad_norm": 2.095987319946289,
      "learning_rate": 4.399789307347906e-05,
      "loss": 1.9332,
      "step": 4563
    },
    {
      "epoch": 0.12020015801948906,
      "grad_norm": 3.8956432342529297,
      "learning_rate": 4.399657624440348e-05,
      "loss": 2.0838,
      "step": 4564
    },
    {
      "epoch": 0.1202264946010008,
      "grad_norm": 1.9411897659301758,
      "learning_rate": 4.399525941532789e-05,
      "loss": 2.0952,
      "step": 4565
    },
    {
      "epoch": 0.12025283118251251,
      "grad_norm": 2.8632423877716064,
      "learning_rate": 4.3993942586252307e-05,
      "loss": 2.3953,
      "step": 4566
    },
    {
      "epoch": 0.12027916776402423,
      "grad_norm": 1.9260960817337036,
      "learning_rate": 4.399262575717672e-05,
      "loss": 1.8821,
      "step": 4567
    },
    {
      "epoch": 0.12030550434553595,
      "grad_norm": 2.5925159454345703,
      "learning_rate": 4.399130892810113e-05,
      "loss": 0.6486,
      "step": 4568
    },
    {
      "epoch": 0.12033184092704767,
      "grad_norm": 2.303072452545166,
      "learning_rate": 4.398999209902555e-05,
      "loss": 1.766,
      "step": 4569
    },
    {
      "epoch": 0.12035817750855939,
      "grad_norm": 1.9586025476455688,
      "learning_rate": 4.398867526994996e-05,
      "loss": 1.392,
      "step": 4570
    },
    {
      "epoch": 0.12038451409007112,
      "grad_norm": 4.656185626983643,
      "learning_rate": 4.398735844087438e-05,
      "loss": 1.5239,
      "step": 4571
    },
    {
      "epoch": 0.12041085067158283,
      "grad_norm": 4.9770097732543945,
      "learning_rate": 4.3986041611798787e-05,
      "loss": 1.2803,
      "step": 4572
    },
    {
      "epoch": 0.12043718725309455,
      "grad_norm": 2.0840556621551514,
      "learning_rate": 4.39847247827232e-05,
      "loss": 2.349,
      "step": 4573
    },
    {
      "epoch": 0.12046352383460626,
      "grad_norm": 2.684706211090088,
      "learning_rate": 4.398340795364762e-05,
      "loss": 1.922,
      "step": 4574
    },
    {
      "epoch": 0.12048986041611799,
      "grad_norm": 1.8182473182678223,
      "learning_rate": 4.398209112457203e-05,
      "loss": 2.0707,
      "step": 4575
    },
    {
      "epoch": 0.12051619699762971,
      "grad_norm": 1.966376781463623,
      "learning_rate": 4.398077429549645e-05,
      "loss": 1.8312,
      "step": 4576
    },
    {
      "epoch": 0.12054253357914142,
      "grad_norm": 3.737230062484741,
      "learning_rate": 4.397945746642086e-05,
      "loss": 1.2757,
      "step": 4577
    },
    {
      "epoch": 0.12056887016065315,
      "grad_norm": 2.854975700378418,
      "learning_rate": 4.397814063734527e-05,
      "loss": 1.5723,
      "step": 4578
    },
    {
      "epoch": 0.12059520674216487,
      "grad_norm": 2.072096824645996,
      "learning_rate": 4.397682380826969e-05,
      "loss": 1.4735,
      "step": 4579
    },
    {
      "epoch": 0.12062154332367658,
      "grad_norm": 3.2030415534973145,
      "learning_rate": 4.3975506979194105e-05,
      "loss": 1.8364,
      "step": 4580
    },
    {
      "epoch": 0.12064787990518831,
      "grad_norm": 3.832263946533203,
      "learning_rate": 4.3974190150118513e-05,
      "loss": 0.9231,
      "step": 4581
    },
    {
      "epoch": 0.12067421648670003,
      "grad_norm": 2.062819242477417,
      "learning_rate": 4.397287332104293e-05,
      "loss": 1.1318,
      "step": 4582
    },
    {
      "epoch": 0.12070055306821174,
      "grad_norm": 1.9557278156280518,
      "learning_rate": 4.3971556491967345e-05,
      "loss": 1.9448,
      "step": 4583
    },
    {
      "epoch": 0.12072688964972346,
      "grad_norm": 2.545456886291504,
      "learning_rate": 4.397023966289176e-05,
      "loss": 2.3229,
      "step": 4584
    },
    {
      "epoch": 0.12075322623123519,
      "grad_norm": 2.5751824378967285,
      "learning_rate": 4.3968922833816176e-05,
      "loss": 1.7099,
      "step": 4585
    },
    {
      "epoch": 0.1207795628127469,
      "grad_norm": 2.2981085777282715,
      "learning_rate": 4.3967606004740585e-05,
      "loss": 1.779,
      "step": 4586
    },
    {
      "epoch": 0.12080589939425862,
      "grad_norm": 2.2591090202331543,
      "learning_rate": 4.3966289175665e-05,
      "loss": 1.4552,
      "step": 4587
    },
    {
      "epoch": 0.12083223597577035,
      "grad_norm": 3.642763137817383,
      "learning_rate": 4.3964972346589416e-05,
      "loss": 1.5614,
      "step": 4588
    },
    {
      "epoch": 0.12085857255728207,
      "grad_norm": 2.989940881729126,
      "learning_rate": 4.396365551751383e-05,
      "loss": 1.5025,
      "step": 4589
    },
    {
      "epoch": 0.12088490913879378,
      "grad_norm": 3.334362745285034,
      "learning_rate": 4.396233868843824e-05,
      "loss": 1.6429,
      "step": 4590
    },
    {
      "epoch": 0.12091124572030551,
      "grad_norm": 2.275623083114624,
      "learning_rate": 4.3961021859362656e-05,
      "loss": 1.6776,
      "step": 4591
    },
    {
      "epoch": 0.12093758230181723,
      "grad_norm": 3.960989475250244,
      "learning_rate": 4.395970503028707e-05,
      "loss": 1.333,
      "step": 4592
    },
    {
      "epoch": 0.12096391888332894,
      "grad_norm": 2.0593600273132324,
      "learning_rate": 4.395838820121149e-05,
      "loss": 2.0727,
      "step": 4593
    },
    {
      "epoch": 0.12099025546484066,
      "grad_norm": 2.603482484817505,
      "learning_rate": 4.39570713721359e-05,
      "loss": 1.8435,
      "step": 4594
    },
    {
      "epoch": 0.12101659204635239,
      "grad_norm": 2.5595359802246094,
      "learning_rate": 4.395575454306031e-05,
      "loss": 2.0078,
      "step": 4595
    },
    {
      "epoch": 0.1210429286278641,
      "grad_norm": 2.7202510833740234,
      "learning_rate": 4.395443771398473e-05,
      "loss": 1.2769,
      "step": 4596
    },
    {
      "epoch": 0.12106926520937582,
      "grad_norm": 2.8965611457824707,
      "learning_rate": 4.395312088490914e-05,
      "loss": 0.5525,
      "step": 4597
    },
    {
      "epoch": 0.12109560179088755,
      "grad_norm": 3.5813839435577393,
      "learning_rate": 4.395180405583356e-05,
      "loss": 1.3512,
      "step": 4598
    },
    {
      "epoch": 0.12112193837239926,
      "grad_norm": 2.405693769454956,
      "learning_rate": 4.395048722675797e-05,
      "loss": 1.937,
      "step": 4599
    },
    {
      "epoch": 0.12114827495391098,
      "grad_norm": 2.0294175148010254,
      "learning_rate": 4.394917039768238e-05,
      "loss": 1.4466,
      "step": 4600
    },
    {
      "epoch": 0.12117461153542271,
      "grad_norm": 2.122802972793579,
      "learning_rate": 4.39478535686068e-05,
      "loss": 2.1325,
      "step": 4601
    },
    {
      "epoch": 0.12120094811693442,
      "grad_norm": 3.272130012512207,
      "learning_rate": 4.3946536739531214e-05,
      "loss": 2.0251,
      "step": 4602
    },
    {
      "epoch": 0.12122728469844614,
      "grad_norm": 1.7862468957901,
      "learning_rate": 4.394521991045563e-05,
      "loss": 2.0858,
      "step": 4603
    },
    {
      "epoch": 0.12125362127995787,
      "grad_norm": 1.9192084074020386,
      "learning_rate": 4.394390308138004e-05,
      "loss": 1.0839,
      "step": 4604
    },
    {
      "epoch": 0.12127995786146958,
      "grad_norm": 3.5498342514038086,
      "learning_rate": 4.3942586252304454e-05,
      "loss": 2.2891,
      "step": 4605
    },
    {
      "epoch": 0.1213062944429813,
      "grad_norm": 1.8106166124343872,
      "learning_rate": 4.394126942322886e-05,
      "loss": 2.3355,
      "step": 4606
    },
    {
      "epoch": 0.12133263102449302,
      "grad_norm": 3.650259256362915,
      "learning_rate": 4.3939952594153285e-05,
      "loss": 1.3046,
      "step": 4607
    },
    {
      "epoch": 0.12135896760600474,
      "grad_norm": 2.2454311847686768,
      "learning_rate": 4.3938635765077694e-05,
      "loss": 1.5541,
      "step": 4608
    },
    {
      "epoch": 0.12138530418751646,
      "grad_norm": 1.8960703611373901,
      "learning_rate": 4.393731893600211e-05,
      "loss": 0.7282,
      "step": 4609
    },
    {
      "epoch": 0.12141164076902818,
      "grad_norm": 5.468796730041504,
      "learning_rate": 4.393600210692652e-05,
      "loss": 1.2263,
      "step": 4610
    },
    {
      "epoch": 0.1214379773505399,
      "grad_norm": 2.372835636138916,
      "learning_rate": 4.3934685277850934e-05,
      "loss": 1.4008,
      "step": 4611
    },
    {
      "epoch": 0.12146431393205162,
      "grad_norm": 1.8523073196411133,
      "learning_rate": 4.393336844877535e-05,
      "loss": 0.6759,
      "step": 4612
    },
    {
      "epoch": 0.12149065051356334,
      "grad_norm": 3.4270529747009277,
      "learning_rate": 4.3932051619699765e-05,
      "loss": 2.3706,
      "step": 4613
    },
    {
      "epoch": 0.12151698709507507,
      "grad_norm": 3.632455348968506,
      "learning_rate": 4.393073479062418e-05,
      "loss": 1.1427,
      "step": 4614
    },
    {
      "epoch": 0.12154332367658678,
      "grad_norm": 6.611170768737793,
      "learning_rate": 4.392941796154859e-05,
      "loss": 1.3079,
      "step": 4615
    },
    {
      "epoch": 0.1215696602580985,
      "grad_norm": 3.0112364292144775,
      "learning_rate": 4.392810113247301e-05,
      "loss": 2.2547,
      "step": 4616
    },
    {
      "epoch": 0.12159599683961021,
      "grad_norm": 2.2450125217437744,
      "learning_rate": 4.392678430339742e-05,
      "loss": 1.7717,
      "step": 4617
    },
    {
      "epoch": 0.12162233342112194,
      "grad_norm": 3.6134462356567383,
      "learning_rate": 4.3925467474321836e-05,
      "loss": 2.7018,
      "step": 4618
    },
    {
      "epoch": 0.12164867000263366,
      "grad_norm": 2.246868848800659,
      "learning_rate": 4.3924150645246245e-05,
      "loss": 2.1674,
      "step": 4619
    },
    {
      "epoch": 0.12167500658414537,
      "grad_norm": 1.966476559638977,
      "learning_rate": 4.392283381617066e-05,
      "loss": 1.8422,
      "step": 4620
    },
    {
      "epoch": 0.1217013431656571,
      "grad_norm": 3.2045938968658447,
      "learning_rate": 4.3921516987095076e-05,
      "loss": 1.2722,
      "step": 4621
    },
    {
      "epoch": 0.12172767974716882,
      "grad_norm": 6.169275283813477,
      "learning_rate": 4.392020015801949e-05,
      "loss": 1.1417,
      "step": 4622
    },
    {
      "epoch": 0.12175401632868053,
      "grad_norm": 3.556260347366333,
      "learning_rate": 4.391888332894391e-05,
      "loss": 1.1964,
      "step": 4623
    },
    {
      "epoch": 0.12178035291019226,
      "grad_norm": 2.3873257637023926,
      "learning_rate": 4.3917566499868316e-05,
      "loss": 1.9118,
      "step": 4624
    },
    {
      "epoch": 0.12180668949170398,
      "grad_norm": 2.158400774002075,
      "learning_rate": 4.391624967079273e-05,
      "loss": 1.7923,
      "step": 4625
    },
    {
      "epoch": 0.1218330260732157,
      "grad_norm": 2.245790719985962,
      "learning_rate": 4.391493284171715e-05,
      "loss": 1.4026,
      "step": 4626
    },
    {
      "epoch": 0.12185936265472741,
      "grad_norm": 1.8083654642105103,
      "learning_rate": 4.391361601264156e-05,
      "loss": 1.8232,
      "step": 4627
    },
    {
      "epoch": 0.12188569923623914,
      "grad_norm": 4.552290439605713,
      "learning_rate": 4.391229918356597e-05,
      "loss": 1.6143,
      "step": 4628
    },
    {
      "epoch": 0.12191203581775086,
      "grad_norm": 2.729383945465088,
      "learning_rate": 4.391098235449039e-05,
      "loss": 1.6407,
      "step": 4629
    },
    {
      "epoch": 0.12193837239926257,
      "grad_norm": 2.899068593978882,
      "learning_rate": 4.39096655254148e-05,
      "loss": 2.163,
      "step": 4630
    },
    {
      "epoch": 0.1219647089807743,
      "grad_norm": 1.8001106977462769,
      "learning_rate": 4.390834869633922e-05,
      "loss": 1.5818,
      "step": 4631
    },
    {
      "epoch": 0.12199104556228602,
      "grad_norm": 5.082118988037109,
      "learning_rate": 4.3907031867263634e-05,
      "loss": 1.2649,
      "step": 4632
    },
    {
      "epoch": 0.12201738214379773,
      "grad_norm": 1.9306411743164062,
      "learning_rate": 4.390571503818804e-05,
      "loss": 1.8876,
      "step": 4633
    },
    {
      "epoch": 0.12204371872530946,
      "grad_norm": 4.088321685791016,
      "learning_rate": 4.390439820911246e-05,
      "loss": 0.6823,
      "step": 4634
    },
    {
      "epoch": 0.12207005530682118,
      "grad_norm": 1.8974683284759521,
      "learning_rate": 4.3903081380036874e-05,
      "loss": 0.7718,
      "step": 4635
    },
    {
      "epoch": 0.12209639188833289,
      "grad_norm": 1.7598472833633423,
      "learning_rate": 4.390176455096129e-05,
      "loss": 1.6616,
      "step": 4636
    },
    {
      "epoch": 0.12212272846984461,
      "grad_norm": 2.717180013656616,
      "learning_rate": 4.39004477218857e-05,
      "loss": 2.1474,
      "step": 4637
    },
    {
      "epoch": 0.12214906505135634,
      "grad_norm": 1.8452847003936768,
      "learning_rate": 4.3899130892810114e-05,
      "loss": 0.1924,
      "step": 4638
    },
    {
      "epoch": 0.12217540163286805,
      "grad_norm": 2.279766798019409,
      "learning_rate": 4.389781406373453e-05,
      "loss": 1.6768,
      "step": 4639
    },
    {
      "epoch": 0.12220173821437977,
      "grad_norm": 2.480419874191284,
      "learning_rate": 4.3896497234658946e-05,
      "loss": 1.458,
      "step": 4640
    },
    {
      "epoch": 0.1222280747958915,
      "grad_norm": 1.9092397689819336,
      "learning_rate": 4.389518040558336e-05,
      "loss": 1.0179,
      "step": 4641
    },
    {
      "epoch": 0.12225441137740321,
      "grad_norm": 3.4327282905578613,
      "learning_rate": 4.389386357650777e-05,
      "loss": 1.774,
      "step": 4642
    },
    {
      "epoch": 0.12228074795891493,
      "grad_norm": 3.2832391262054443,
      "learning_rate": 4.3892546747432186e-05,
      "loss": 1.3595,
      "step": 4643
    },
    {
      "epoch": 0.12230708454042666,
      "grad_norm": 2.413118362426758,
      "learning_rate": 4.3891229918356594e-05,
      "loss": 1.3041,
      "step": 4644
    },
    {
      "epoch": 0.12233342112193837,
      "grad_norm": 2.2710001468658447,
      "learning_rate": 4.388991308928102e-05,
      "loss": 1.5093,
      "step": 4645
    },
    {
      "epoch": 0.12235975770345009,
      "grad_norm": 3.111807107925415,
      "learning_rate": 4.3888596260205426e-05,
      "loss": 1.9066,
      "step": 4646
    },
    {
      "epoch": 0.12238609428496182,
      "grad_norm": 1.690782070159912,
      "learning_rate": 4.388727943112984e-05,
      "loss": 1.5699,
      "step": 4647
    },
    {
      "epoch": 0.12241243086647353,
      "grad_norm": 2.160367012023926,
      "learning_rate": 4.388596260205426e-05,
      "loss": 1.564,
      "step": 4648
    },
    {
      "epoch": 0.12243876744798525,
      "grad_norm": 2.432783842086792,
      "learning_rate": 4.388464577297867e-05,
      "loss": 2.3584,
      "step": 4649
    },
    {
      "epoch": 0.12246510402949697,
      "grad_norm": 3.1996405124664307,
      "learning_rate": 4.388332894390309e-05,
      "loss": 1.7356,
      "step": 4650
    },
    {
      "epoch": 0.1224914406110087,
      "grad_norm": 1.723605751991272,
      "learning_rate": 4.38820121148275e-05,
      "loss": 1.0746,
      "step": 4651
    },
    {
      "epoch": 0.12251777719252041,
      "grad_norm": 2.0372090339660645,
      "learning_rate": 4.388069528575191e-05,
      "loss": 1.2094,
      "step": 4652
    },
    {
      "epoch": 0.12254411377403213,
      "grad_norm": 1.6394349336624146,
      "learning_rate": 4.387937845667632e-05,
      "loss": 2.4443,
      "step": 4653
    },
    {
      "epoch": 0.12257045035554386,
      "grad_norm": 4.406085968017578,
      "learning_rate": 4.3878061627600744e-05,
      "loss": 1.4122,
      "step": 4654
    },
    {
      "epoch": 0.12259678693705557,
      "grad_norm": 3.4900963306427,
      "learning_rate": 4.387674479852515e-05,
      "loss": 2.1812,
      "step": 4655
    },
    {
      "epoch": 0.12262312351856729,
      "grad_norm": 2.078904867172241,
      "learning_rate": 4.387542796944957e-05,
      "loss": 1.9236,
      "step": 4656
    },
    {
      "epoch": 0.12264946010007902,
      "grad_norm": 2.413954734802246,
      "learning_rate": 4.387411114037398e-05,
      "loss": 2.0406,
      "step": 4657
    },
    {
      "epoch": 0.12267579668159073,
      "grad_norm": 1.8602428436279297,
      "learning_rate": 4.387279431129839e-05,
      "loss": 1.5564,
      "step": 4658
    },
    {
      "epoch": 0.12270213326310245,
      "grad_norm": 2.4664242267608643,
      "learning_rate": 4.387147748222281e-05,
      "loss": 1.7709,
      "step": 4659
    },
    {
      "epoch": 0.12272846984461416,
      "grad_norm": 2.4134647846221924,
      "learning_rate": 4.3870160653147224e-05,
      "loss": 2.1152,
      "step": 4660
    },
    {
      "epoch": 0.12275480642612589,
      "grad_norm": 2.130995750427246,
      "learning_rate": 4.386884382407164e-05,
      "loss": 1.8566,
      "step": 4661
    },
    {
      "epoch": 0.12278114300763761,
      "grad_norm": 1.8650908470153809,
      "learning_rate": 4.386752699499605e-05,
      "loss": 2.2976,
      "step": 4662
    },
    {
      "epoch": 0.12280747958914932,
      "grad_norm": 2.470064640045166,
      "learning_rate": 4.386621016592047e-05,
      "loss": 0.8065,
      "step": 4663
    },
    {
      "epoch": 0.12283381617066105,
      "grad_norm": 3.8376693725585938,
      "learning_rate": 4.386489333684488e-05,
      "loss": 0.8042,
      "step": 4664
    },
    {
      "epoch": 0.12286015275217277,
      "grad_norm": 1.9035619497299194,
      "learning_rate": 4.3863576507769295e-05,
      "loss": 2.5298,
      "step": 4665
    },
    {
      "epoch": 0.12288648933368448,
      "grad_norm": 2.975315809249878,
      "learning_rate": 4.3862259678693704e-05,
      "loss": 1.054,
      "step": 4666
    },
    {
      "epoch": 0.12291282591519621,
      "grad_norm": 3.355088233947754,
      "learning_rate": 4.386094284961812e-05,
      "loss": 1.1019,
      "step": 4667
    },
    {
      "epoch": 0.12293916249670793,
      "grad_norm": 3.4372174739837646,
      "learning_rate": 4.3859626020542535e-05,
      "loss": 1.185,
      "step": 4668
    },
    {
      "epoch": 0.12296549907821964,
      "grad_norm": 1.6783677339553833,
      "learning_rate": 4.385830919146695e-05,
      "loss": 1.6823,
      "step": 4669
    },
    {
      "epoch": 0.12299183565973136,
      "grad_norm": 1.8151222467422485,
      "learning_rate": 4.3856992362391366e-05,
      "loss": 0.9305,
      "step": 4670
    },
    {
      "epoch": 0.12301817224124309,
      "grad_norm": 2.9164371490478516,
      "learning_rate": 4.3855675533315775e-05,
      "loss": 1.2978,
      "step": 4671
    },
    {
      "epoch": 0.1230445088227548,
      "grad_norm": 1.9816832542419434,
      "learning_rate": 4.385435870424019e-05,
      "loss": 2.2482,
      "step": 4672
    },
    {
      "epoch": 0.12307084540426652,
      "grad_norm": 2.254826307296753,
      "learning_rate": 4.3853041875164606e-05,
      "loss": 1.6485,
      "step": 4673
    },
    {
      "epoch": 0.12309718198577825,
      "grad_norm": 2.6538503170013428,
      "learning_rate": 4.385172504608902e-05,
      "loss": 1.1908,
      "step": 4674
    },
    {
      "epoch": 0.12312351856728997,
      "grad_norm": 2.3163418769836426,
      "learning_rate": 4.385040821701343e-05,
      "loss": 1.822,
      "step": 4675
    },
    {
      "epoch": 0.12314985514880168,
      "grad_norm": 3.323758363723755,
      "learning_rate": 4.3849091387937846e-05,
      "loss": 0.9073,
      "step": 4676
    },
    {
      "epoch": 0.12317619173031341,
      "grad_norm": 2.26656174659729,
      "learning_rate": 4.384777455886226e-05,
      "loss": 1.5782,
      "step": 4677
    },
    {
      "epoch": 0.12320252831182513,
      "grad_norm": 2.4748458862304688,
      "learning_rate": 4.384645772978668e-05,
      "loss": 1.8417,
      "step": 4678
    },
    {
      "epoch": 0.12322886489333684,
      "grad_norm": 6.037837982177734,
      "learning_rate": 4.384514090071109e-05,
      "loss": 1.5944,
      "step": 4679
    },
    {
      "epoch": 0.12325520147484856,
      "grad_norm": 2.3028833866119385,
      "learning_rate": 4.38438240716355e-05,
      "loss": 1.4918,
      "step": 4680
    },
    {
      "epoch": 0.12328153805636029,
      "grad_norm": 2.6609320640563965,
      "learning_rate": 4.384250724255992e-05,
      "loss": 1.5208,
      "step": 4681
    },
    {
      "epoch": 0.123307874637872,
      "grad_norm": 1.6242191791534424,
      "learning_rate": 4.384119041348433e-05,
      "loss": 1.9448,
      "step": 4682
    },
    {
      "epoch": 0.12333421121938372,
      "grad_norm": 1.7447011470794678,
      "learning_rate": 4.383987358440875e-05,
      "loss": 1.5254,
      "step": 4683
    },
    {
      "epoch": 0.12336054780089545,
      "grad_norm": 2.2075204849243164,
      "learning_rate": 4.383855675533316e-05,
      "loss": 2.1902,
      "step": 4684
    },
    {
      "epoch": 0.12338688438240716,
      "grad_norm": 1.8057124614715576,
      "learning_rate": 4.383723992625757e-05,
      "loss": 2.1283,
      "step": 4685
    },
    {
      "epoch": 0.12341322096391888,
      "grad_norm": 4.820918083190918,
      "learning_rate": 4.383592309718199e-05,
      "loss": 1.2956,
      "step": 4686
    },
    {
      "epoch": 0.12343955754543061,
      "grad_norm": 1.7067325115203857,
      "learning_rate": 4.3834606268106404e-05,
      "loss": 1.4117,
      "step": 4687
    },
    {
      "epoch": 0.12346589412694232,
      "grad_norm": 2.3989779949188232,
      "learning_rate": 4.383328943903082e-05,
      "loss": 2.0893,
      "step": 4688
    },
    {
      "epoch": 0.12349223070845404,
      "grad_norm": 2.5785791873931885,
      "learning_rate": 4.383197260995523e-05,
      "loss": 1.694,
      "step": 4689
    },
    {
      "epoch": 0.12351856728996576,
      "grad_norm": 2.5003888607025146,
      "learning_rate": 4.3830655780879644e-05,
      "loss": 1.7038,
      "step": 4690
    },
    {
      "epoch": 0.12354490387147748,
      "grad_norm": 2.5338175296783447,
      "learning_rate": 4.382933895180405e-05,
      "loss": 1.5793,
      "step": 4691
    },
    {
      "epoch": 0.1235712404529892,
      "grad_norm": 3.209324598312378,
      "learning_rate": 4.3828022122728475e-05,
      "loss": 0.8265,
      "step": 4692
    },
    {
      "epoch": 0.12359757703450092,
      "grad_norm": 2.568403959274292,
      "learning_rate": 4.3826705293652884e-05,
      "loss": 1.5413,
      "step": 4693
    },
    {
      "epoch": 0.12362391361601265,
      "grad_norm": 1.9092358350753784,
      "learning_rate": 4.38253884645773e-05,
      "loss": 1.9075,
      "step": 4694
    },
    {
      "epoch": 0.12365025019752436,
      "grad_norm": 1.8922903537750244,
      "learning_rate": 4.3824071635501715e-05,
      "loss": 1.7268,
      "step": 4695
    },
    {
      "epoch": 0.12367658677903608,
      "grad_norm": 1.9539427757263184,
      "learning_rate": 4.382275480642613e-05,
      "loss": 2.9814,
      "step": 4696
    },
    {
      "epoch": 0.1237029233605478,
      "grad_norm": 1.4077868461608887,
      "learning_rate": 4.3821437977350547e-05,
      "loss": 1.6842,
      "step": 4697
    },
    {
      "epoch": 0.12372925994205952,
      "grad_norm": 3.8124887943267822,
      "learning_rate": 4.3820121148274955e-05,
      "loss": 1.8621,
      "step": 4698
    },
    {
      "epoch": 0.12375559652357124,
      "grad_norm": 2.13594913482666,
      "learning_rate": 4.381880431919937e-05,
      "loss": 0.4539,
      "step": 4699
    },
    {
      "epoch": 0.12378193310508297,
      "grad_norm": 2.145198345184326,
      "learning_rate": 4.381748749012378e-05,
      "loss": 1.0364,
      "step": 4700
    },
    {
      "epoch": 0.12380826968659468,
      "grad_norm": 1.9599852561950684,
      "learning_rate": 4.38161706610482e-05,
      "loss": 1.3169,
      "step": 4701
    },
    {
      "epoch": 0.1238346062681064,
      "grad_norm": 3.1412858963012695,
      "learning_rate": 4.381485383197261e-05,
      "loss": 1.6835,
      "step": 4702
    },
    {
      "epoch": 0.12386094284961811,
      "grad_norm": 3.127105712890625,
      "learning_rate": 4.3813537002897027e-05,
      "loss": 1.308,
      "step": 4703
    },
    {
      "epoch": 0.12388727943112984,
      "grad_norm": 3.4913148880004883,
      "learning_rate": 4.381222017382144e-05,
      "loss": 1.629,
      "step": 4704
    },
    {
      "epoch": 0.12391361601264156,
      "grad_norm": 3.8447394371032715,
      "learning_rate": 4.381090334474585e-05,
      "loss": 2.0455,
      "step": 4705
    },
    {
      "epoch": 0.12393995259415327,
      "grad_norm": 1.6441948413848877,
      "learning_rate": 4.380958651567027e-05,
      "loss": 0.2486,
      "step": 4706
    },
    {
      "epoch": 0.123966289175665,
      "grad_norm": 2.0133681297302246,
      "learning_rate": 4.380826968659468e-05,
      "loss": 1.5702,
      "step": 4707
    },
    {
      "epoch": 0.12399262575717672,
      "grad_norm": 2.8952345848083496,
      "learning_rate": 4.38069528575191e-05,
      "loss": 1.6503,
      "step": 4708
    },
    {
      "epoch": 0.12401896233868843,
      "grad_norm": 3.1991989612579346,
      "learning_rate": 4.380563602844351e-05,
      "loss": 1.5906,
      "step": 4709
    },
    {
      "epoch": 0.12404529892020016,
      "grad_norm": 1.8866389989852905,
      "learning_rate": 4.380431919936792e-05,
      "loss": 1.7536,
      "step": 4710
    },
    {
      "epoch": 0.12407163550171188,
      "grad_norm": 2.81256365776062,
      "learning_rate": 4.380300237029234e-05,
      "loss": 1.305,
      "step": 4711
    },
    {
      "epoch": 0.1240979720832236,
      "grad_norm": 4.803744316101074,
      "learning_rate": 4.3801685541216753e-05,
      "loss": 1.7908,
      "step": 4712
    },
    {
      "epoch": 0.12412430866473531,
      "grad_norm": 2.1657397747039795,
      "learning_rate": 4.380036871214116e-05,
      "loss": 1.4083,
      "step": 4713
    },
    {
      "epoch": 0.12415064524624704,
      "grad_norm": 2.398895025253296,
      "learning_rate": 4.379905188306558e-05,
      "loss": 1.4321,
      "step": 4714
    },
    {
      "epoch": 0.12417698182775876,
      "grad_norm": 2.256537914276123,
      "learning_rate": 4.3797735053989993e-05,
      "loss": 1.5301,
      "step": 4715
    },
    {
      "epoch": 0.12420331840927047,
      "grad_norm": 2.276139736175537,
      "learning_rate": 4.379641822491441e-05,
      "loss": 2.0023,
      "step": 4716
    },
    {
      "epoch": 0.1242296549907822,
      "grad_norm": 4.326174259185791,
      "learning_rate": 4.3795101395838825e-05,
      "loss": 1.5146,
      "step": 4717
    },
    {
      "epoch": 0.12425599157229392,
      "grad_norm": 4.015639781951904,
      "learning_rate": 4.3793784566763233e-05,
      "loss": 0.9385,
      "step": 4718
    },
    {
      "epoch": 0.12428232815380563,
      "grad_norm": 2.3562822341918945,
      "learning_rate": 4.379246773768765e-05,
      "loss": 1.251,
      "step": 4719
    },
    {
      "epoch": 0.12430866473531736,
      "grad_norm": 1.6875540018081665,
      "learning_rate": 4.3791150908612065e-05,
      "loss": 2.0256,
      "step": 4720
    },
    {
      "epoch": 0.12433500131682908,
      "grad_norm": 3.528935432434082,
      "learning_rate": 4.378983407953648e-05,
      "loss": 0.6218,
      "step": 4721
    },
    {
      "epoch": 0.12436133789834079,
      "grad_norm": 2.5248193740844727,
      "learning_rate": 4.378851725046089e-05,
      "loss": 1.5558,
      "step": 4722
    },
    {
      "epoch": 0.12438767447985251,
      "grad_norm": 2.8863699436187744,
      "learning_rate": 4.3787200421385305e-05,
      "loss": 2.2871,
      "step": 4723
    },
    {
      "epoch": 0.12441401106136424,
      "grad_norm": 2.0777347087860107,
      "learning_rate": 4.378588359230972e-05,
      "loss": 1.4763,
      "step": 4724
    },
    {
      "epoch": 0.12444034764287595,
      "grad_norm": 4.451682090759277,
      "learning_rate": 4.3784566763234136e-05,
      "loss": 1.5911,
      "step": 4725
    },
    {
      "epoch": 0.12446668422438767,
      "grad_norm": 1.9852718114852905,
      "learning_rate": 4.378324993415855e-05,
      "loss": 2.0354,
      "step": 4726
    },
    {
      "epoch": 0.1244930208058994,
      "grad_norm": 4.1549530029296875,
      "learning_rate": 4.378193310508296e-05,
      "loss": 1.0154,
      "step": 4727
    },
    {
      "epoch": 0.12451935738741111,
      "grad_norm": 2.3205487728118896,
      "learning_rate": 4.3780616276007376e-05,
      "loss": 1.8725,
      "step": 4728
    },
    {
      "epoch": 0.12454569396892283,
      "grad_norm": 1.9256564378738403,
      "learning_rate": 4.377929944693179e-05,
      "loss": 1.7368,
      "step": 4729
    },
    {
      "epoch": 0.12457203055043456,
      "grad_norm": 2.4510157108306885,
      "learning_rate": 4.377798261785621e-05,
      "loss": 1.9596,
      "step": 4730
    },
    {
      "epoch": 0.12459836713194627,
      "grad_norm": 2.789548635482788,
      "learning_rate": 4.3776665788780616e-05,
      "loss": 2.0809,
      "step": 4731
    },
    {
      "epoch": 0.12462470371345799,
      "grad_norm": 2.673419713973999,
      "learning_rate": 4.377534895970503e-05,
      "loss": 3.0775,
      "step": 4732
    },
    {
      "epoch": 0.1246510402949697,
      "grad_norm": 2.358151435852051,
      "learning_rate": 4.377403213062945e-05,
      "loss": 1.8011,
      "step": 4733
    },
    {
      "epoch": 0.12467737687648144,
      "grad_norm": 1.626806378364563,
      "learning_rate": 4.377271530155386e-05,
      "loss": 1.8686,
      "step": 4734
    },
    {
      "epoch": 0.12470371345799315,
      "grad_norm": 3.8930649757385254,
      "learning_rate": 4.377139847247828e-05,
      "loss": 1.2501,
      "step": 4735
    },
    {
      "epoch": 0.12473005003950487,
      "grad_norm": 6.212684154510498,
      "learning_rate": 4.377008164340269e-05,
      "loss": 2.0236,
      "step": 4736
    },
    {
      "epoch": 0.1247563866210166,
      "grad_norm": 3.866426944732666,
      "learning_rate": 4.37687648143271e-05,
      "loss": 1.3742,
      "step": 4737
    },
    {
      "epoch": 0.12478272320252831,
      "grad_norm": 3.705449104309082,
      "learning_rate": 4.376744798525151e-05,
      "loss": 1.393,
      "step": 4738
    },
    {
      "epoch": 0.12480905978404003,
      "grad_norm": 1.7463220357894897,
      "learning_rate": 4.3766131156175934e-05,
      "loss": 1.9235,
      "step": 4739
    },
    {
      "epoch": 0.12483539636555176,
      "grad_norm": 2.7118091583251953,
      "learning_rate": 4.376481432710034e-05,
      "loss": 2.5418,
      "step": 4740
    },
    {
      "epoch": 0.12486173294706347,
      "grad_norm": 3.8343799114227295,
      "learning_rate": 4.376349749802476e-05,
      "loss": 1.697,
      "step": 4741
    },
    {
      "epoch": 0.12488806952857519,
      "grad_norm": 1.4628170728683472,
      "learning_rate": 4.3762180668949174e-05,
      "loss": 2.3846,
      "step": 4742
    },
    {
      "epoch": 0.12491440611008692,
      "grad_norm": 1.8428781032562256,
      "learning_rate": 4.376086383987358e-05,
      "loss": 1.7052,
      "step": 4743
    },
    {
      "epoch": 0.12494074269159863,
      "grad_norm": 3.012197256088257,
      "learning_rate": 4.3759547010798005e-05,
      "loss": 1.3005,
      "step": 4744
    },
    {
      "epoch": 0.12496707927311035,
      "grad_norm": 2.2392208576202393,
      "learning_rate": 4.3758230181722414e-05,
      "loss": 1.1542,
      "step": 4745
    },
    {
      "epoch": 0.12499341585462206,
      "grad_norm": 3.821258306503296,
      "learning_rate": 4.375691335264683e-05,
      "loss": 0.7939,
      "step": 4746
    },
    {
      "epoch": 0.12501975243613378,
      "grad_norm": 1.9974212646484375,
      "learning_rate": 4.375559652357124e-05,
      "loss": 1.8858,
      "step": 4747
    },
    {
      "epoch": 0.12504608901764552,
      "grad_norm": 2.2754218578338623,
      "learning_rate": 4.375427969449566e-05,
      "loss": 1.9591,
      "step": 4748
    },
    {
      "epoch": 0.12507242559915724,
      "grad_norm": 1.5792393684387207,
      "learning_rate": 4.375296286542007e-05,
      "loss": 0.8318,
      "step": 4749
    },
    {
      "epoch": 0.12509876218066895,
      "grad_norm": 2.1220436096191406,
      "learning_rate": 4.3751646036344485e-05,
      "loss": 1.5546,
      "step": 4750
    },
    {
      "epoch": 0.12512509876218067,
      "grad_norm": 3.427337408065796,
      "learning_rate": 4.37503292072689e-05,
      "loss": 1.6584,
      "step": 4751
    },
    {
      "epoch": 0.12515143534369239,
      "grad_norm": 3.002098321914673,
      "learning_rate": 4.374901237819331e-05,
      "loss": 2.036,
      "step": 4752
    },
    {
      "epoch": 0.1251777719252041,
      "grad_norm": 2.1741037368774414,
      "learning_rate": 4.374769554911773e-05,
      "loss": 1.4488,
      "step": 4753
    },
    {
      "epoch": 0.12520410850671582,
      "grad_norm": 2.436464309692383,
      "learning_rate": 4.374637872004214e-05,
      "loss": 1.5231,
      "step": 4754
    },
    {
      "epoch": 0.12523044508822756,
      "grad_norm": 2.3738396167755127,
      "learning_rate": 4.3745061890966556e-05,
      "loss": 1.8681,
      "step": 4755
    },
    {
      "epoch": 0.12525678166973928,
      "grad_norm": 4.001001834869385,
      "learning_rate": 4.3743745061890965e-05,
      "loss": 1.1224,
      "step": 4756
    },
    {
      "epoch": 0.125283118251251,
      "grad_norm": 3.5733444690704346,
      "learning_rate": 4.374242823281538e-05,
      "loss": 1.6748,
      "step": 4757
    },
    {
      "epoch": 0.1253094548327627,
      "grad_norm": 8.686369895935059,
      "learning_rate": 4.3741111403739796e-05,
      "loss": 2.6691,
      "step": 4758
    },
    {
      "epoch": 0.12533579141427442,
      "grad_norm": 2.162923574447632,
      "learning_rate": 4.373979457466421e-05,
      "loss": 1.5718,
      "step": 4759
    },
    {
      "epoch": 0.12536212799578614,
      "grad_norm": 3.8729188442230225,
      "learning_rate": 4.373847774558862e-05,
      "loss": 1.7658,
      "step": 4760
    },
    {
      "epoch": 0.12538846457729785,
      "grad_norm": 2.4855551719665527,
      "learning_rate": 4.3737160916513036e-05,
      "loss": 2.3868,
      "step": 4761
    },
    {
      "epoch": 0.1254148011588096,
      "grad_norm": 1.9619601964950562,
      "learning_rate": 4.373584408743745e-05,
      "loss": 1.2659,
      "step": 4762
    },
    {
      "epoch": 0.1254411377403213,
      "grad_norm": 2.4714925289154053,
      "learning_rate": 4.373452725836187e-05,
      "loss": 1.7941,
      "step": 4763
    },
    {
      "epoch": 0.12546747432183303,
      "grad_norm": 2.151272773742676,
      "learning_rate": 4.373321042928628e-05,
      "loss": 1.8296,
      "step": 4764
    },
    {
      "epoch": 0.12549381090334474,
      "grad_norm": 2.1843862533569336,
      "learning_rate": 4.373189360021069e-05,
      "loss": 1.5575,
      "step": 4765
    },
    {
      "epoch": 0.12552014748485646,
      "grad_norm": 2.729646682739258,
      "learning_rate": 4.373057677113511e-05,
      "loss": 0.8134,
      "step": 4766
    },
    {
      "epoch": 0.12554648406636817,
      "grad_norm": 2.3161377906799316,
      "learning_rate": 4.372925994205952e-05,
      "loss": 1.6068,
      "step": 4767
    },
    {
      "epoch": 0.12557282064787992,
      "grad_norm": 1.9280983209609985,
      "learning_rate": 4.372794311298394e-05,
      "loss": 1.9839,
      "step": 4768
    },
    {
      "epoch": 0.12559915722939163,
      "grad_norm": 1.9818122386932373,
      "learning_rate": 4.372662628390835e-05,
      "loss": 1.8064,
      "step": 4769
    },
    {
      "epoch": 0.12562549381090335,
      "grad_norm": 1.721292495727539,
      "learning_rate": 4.372530945483276e-05,
      "loss": 0.533,
      "step": 4770
    },
    {
      "epoch": 0.12565183039241506,
      "grad_norm": 4.04070520401001,
      "learning_rate": 4.372399262575718e-05,
      "loss": 1.8365,
      "step": 4771
    },
    {
      "epoch": 0.12567816697392678,
      "grad_norm": 4.287169456481934,
      "learning_rate": 4.3722675796681594e-05,
      "loss": 1.5298,
      "step": 4772
    },
    {
      "epoch": 0.1257045035554385,
      "grad_norm": 2.601172685623169,
      "learning_rate": 4.372135896760601e-05,
      "loss": 1.0548,
      "step": 4773
    },
    {
      "epoch": 0.1257308401369502,
      "grad_norm": 2.757880210876465,
      "learning_rate": 4.372004213853042e-05,
      "loss": 1.7561,
      "step": 4774
    },
    {
      "epoch": 0.12575717671846195,
      "grad_norm": 2.030111789703369,
      "learning_rate": 4.3718725309454834e-05,
      "loss": 1.4399,
      "step": 4775
    },
    {
      "epoch": 0.12578351329997367,
      "grad_norm": 2.478980302810669,
      "learning_rate": 4.371740848037925e-05,
      "loss": 2.1174,
      "step": 4776
    },
    {
      "epoch": 0.12580984988148539,
      "grad_norm": 3.160599708557129,
      "learning_rate": 4.3716091651303666e-05,
      "loss": 1.8178,
      "step": 4777
    },
    {
      "epoch": 0.1258361864629971,
      "grad_norm": 1.8637760877609253,
      "learning_rate": 4.3714774822228074e-05,
      "loss": 1.6946,
      "step": 4778
    },
    {
      "epoch": 0.12586252304450882,
      "grad_norm": 2.038792371749878,
      "learning_rate": 4.371345799315249e-05,
      "loss": 1.9369,
      "step": 4779
    },
    {
      "epoch": 0.12588885962602053,
      "grad_norm": 2.1864607334136963,
      "learning_rate": 4.3712141164076906e-05,
      "loss": 1.6541,
      "step": 4780
    },
    {
      "epoch": 0.12591519620753228,
      "grad_norm": 3.3782198429107666,
      "learning_rate": 4.371082433500132e-05,
      "loss": 0.4839,
      "step": 4781
    },
    {
      "epoch": 0.125941532789044,
      "grad_norm": 2.9364256858825684,
      "learning_rate": 4.370950750592574e-05,
      "loss": 2.1456,
      "step": 4782
    },
    {
      "epoch": 0.1259678693705557,
      "grad_norm": 4.76910924911499,
      "learning_rate": 4.3708190676850146e-05,
      "loss": 1.4221,
      "step": 4783
    },
    {
      "epoch": 0.12599420595206742,
      "grad_norm": 1.7972780466079712,
      "learning_rate": 4.370687384777456e-05,
      "loss": 1.9187,
      "step": 4784
    },
    {
      "epoch": 0.12602054253357914,
      "grad_norm": 2.302013397216797,
      "learning_rate": 4.370555701869897e-05,
      "loss": 1.759,
      "step": 4785
    },
    {
      "epoch": 0.12604687911509085,
      "grad_norm": 2.296588659286499,
      "learning_rate": 4.370424018962339e-05,
      "loss": 1.8282,
      "step": 4786
    },
    {
      "epoch": 0.12607321569660257,
      "grad_norm": 2.1458377838134766,
      "learning_rate": 4.37029233605478e-05,
      "loss": 1.2792,
      "step": 4787
    },
    {
      "epoch": 0.1260995522781143,
      "grad_norm": 1.976080060005188,
      "learning_rate": 4.370160653147222e-05,
      "loss": 0.6266,
      "step": 4788
    },
    {
      "epoch": 0.12612588885962603,
      "grad_norm": 1.8807599544525146,
      "learning_rate": 4.370028970239663e-05,
      "loss": 1.45,
      "step": 4789
    },
    {
      "epoch": 0.12615222544113774,
      "grad_norm": 1.6742404699325562,
      "learning_rate": 4.369897287332104e-05,
      "loss": 1.5706,
      "step": 4790
    },
    {
      "epoch": 0.12617856202264946,
      "grad_norm": 2.7705278396606445,
      "learning_rate": 4.3697656044245464e-05,
      "loss": 1.9122,
      "step": 4791
    },
    {
      "epoch": 0.12620489860416118,
      "grad_norm": 2.441680669784546,
      "learning_rate": 4.369633921516987e-05,
      "loss": 1.7503,
      "step": 4792
    },
    {
      "epoch": 0.1262312351856729,
      "grad_norm": 2.1823110580444336,
      "learning_rate": 4.369502238609429e-05,
      "loss": 1.564,
      "step": 4793
    },
    {
      "epoch": 0.1262575717671846,
      "grad_norm": 1.9613033533096313,
      "learning_rate": 4.36937055570187e-05,
      "loss": 1.7863,
      "step": 4794
    },
    {
      "epoch": 0.12628390834869635,
      "grad_norm": 1.6110904216766357,
      "learning_rate": 4.369238872794312e-05,
      "loss": 2.3363,
      "step": 4795
    },
    {
      "epoch": 0.12631024493020807,
      "grad_norm": 2.40889573097229,
      "learning_rate": 4.369107189886753e-05,
      "loss": 1.9754,
      "step": 4796
    },
    {
      "epoch": 0.12633658151171978,
      "grad_norm": 3.6205410957336426,
      "learning_rate": 4.3689755069791944e-05,
      "loss": 1.1272,
      "step": 4797
    },
    {
      "epoch": 0.1263629180932315,
      "grad_norm": 4.322226047515869,
      "learning_rate": 4.368843824071636e-05,
      "loss": 1.5361,
      "step": 4798
    },
    {
      "epoch": 0.1263892546747432,
      "grad_norm": 2.373908758163452,
      "learning_rate": 4.368712141164077e-05,
      "loss": 0.8833,
      "step": 4799
    },
    {
      "epoch": 0.12641559125625493,
      "grad_norm": 3.53767466545105,
      "learning_rate": 4.368580458256519e-05,
      "loss": 1.4994,
      "step": 4800
    },
    {
      "epoch": 0.12644192783776667,
      "grad_norm": 2.071329116821289,
      "learning_rate": 4.36844877534896e-05,
      "loss": 1.8627,
      "step": 4801
    },
    {
      "epoch": 0.1264682644192784,
      "grad_norm": 1.9687137603759766,
      "learning_rate": 4.3683170924414015e-05,
      "loss": 2.0375,
      "step": 4802
    },
    {
      "epoch": 0.1264946010007901,
      "grad_norm": 1.76338529586792,
      "learning_rate": 4.3681854095338424e-05,
      "loss": 1.3896,
      "step": 4803
    },
    {
      "epoch": 0.12652093758230182,
      "grad_norm": 3.2583870887756348,
      "learning_rate": 4.368053726626284e-05,
      "loss": 1.7348,
      "step": 4804
    },
    {
      "epoch": 0.12654727416381353,
      "grad_norm": 2.3921329975128174,
      "learning_rate": 4.3679220437187255e-05,
      "loss": 2.3131,
      "step": 4805
    },
    {
      "epoch": 0.12657361074532525,
      "grad_norm": 3.8275365829467773,
      "learning_rate": 4.367790360811167e-05,
      "loss": 1.2922,
      "step": 4806
    },
    {
      "epoch": 0.12659994732683696,
      "grad_norm": 3.6275076866149902,
      "learning_rate": 4.3676586779036086e-05,
      "loss": 1.9291,
      "step": 4807
    },
    {
      "epoch": 0.1266262839083487,
      "grad_norm": 1.7525030374526978,
      "learning_rate": 4.3675269949960495e-05,
      "loss": 2.0664,
      "step": 4808
    },
    {
      "epoch": 0.12665262048986042,
      "grad_norm": 3.133984088897705,
      "learning_rate": 4.367395312088492e-05,
      "loss": 2.1156,
      "step": 4809
    },
    {
      "epoch": 0.12667895707137214,
      "grad_norm": 2.2137491703033447,
      "learning_rate": 4.3672636291809326e-05,
      "loss": 1.9825,
      "step": 4810
    },
    {
      "epoch": 0.12670529365288385,
      "grad_norm": 4.29949951171875,
      "learning_rate": 4.367131946273374e-05,
      "loss": 1.4226,
      "step": 4811
    },
    {
      "epoch": 0.12673163023439557,
      "grad_norm": 3.194653034210205,
      "learning_rate": 4.367000263365815e-05,
      "loss": 2.1576,
      "step": 4812
    },
    {
      "epoch": 0.12675796681590729,
      "grad_norm": 8.790407180786133,
      "learning_rate": 4.3668685804582566e-05,
      "loss": 0.582,
      "step": 4813
    },
    {
      "epoch": 0.126784303397419,
      "grad_norm": 1.9416428804397583,
      "learning_rate": 4.366736897550698e-05,
      "loss": 1.3912,
      "step": 4814
    },
    {
      "epoch": 0.12681063997893074,
      "grad_norm": 2.547950029373169,
      "learning_rate": 4.36660521464314e-05,
      "loss": 2.3496,
      "step": 4815
    },
    {
      "epoch": 0.12683697656044246,
      "grad_norm": 2.2785916328430176,
      "learning_rate": 4.3664735317355806e-05,
      "loss": 1.5071,
      "step": 4816
    },
    {
      "epoch": 0.12686331314195418,
      "grad_norm": 3.295245409011841,
      "learning_rate": 4.366341848828022e-05,
      "loss": 0.7992,
      "step": 4817
    },
    {
      "epoch": 0.1268896497234659,
      "grad_norm": 3.8526058197021484,
      "learning_rate": 4.366210165920464e-05,
      "loss": 1.9285,
      "step": 4818
    },
    {
      "epoch": 0.1269159863049776,
      "grad_norm": 2.907660484313965,
      "learning_rate": 4.366078483012905e-05,
      "loss": 1.9041,
      "step": 4819
    },
    {
      "epoch": 0.12694232288648932,
      "grad_norm": 3.1727676391601562,
      "learning_rate": 4.365946800105347e-05,
      "loss": 1.0849,
      "step": 4820
    },
    {
      "epoch": 0.12696865946800107,
      "grad_norm": 2.2417054176330566,
      "learning_rate": 4.365815117197788e-05,
      "loss": 1.3252,
      "step": 4821
    },
    {
      "epoch": 0.12699499604951278,
      "grad_norm": 2.5461864471435547,
      "learning_rate": 4.365683434290229e-05,
      "loss": 1.3434,
      "step": 4822
    },
    {
      "epoch": 0.1270213326310245,
      "grad_norm": 3.2222328186035156,
      "learning_rate": 4.36555175138267e-05,
      "loss": 2.0012,
      "step": 4823
    },
    {
      "epoch": 0.1270476692125362,
      "grad_norm": 4.481986999511719,
      "learning_rate": 4.3654200684751124e-05,
      "loss": 1.6714,
      "step": 4824
    },
    {
      "epoch": 0.12707400579404793,
      "grad_norm": 1.9870394468307495,
      "learning_rate": 4.365288385567553e-05,
      "loss": 1.7413,
      "step": 4825
    },
    {
      "epoch": 0.12710034237555964,
      "grad_norm": 2.8645412921905518,
      "learning_rate": 4.365156702659995e-05,
      "loss": 0.7193,
      "step": 4826
    },
    {
      "epoch": 0.12712667895707136,
      "grad_norm": 4.1999406814575195,
      "learning_rate": 4.3650250197524364e-05,
      "loss": 1.0632,
      "step": 4827
    },
    {
      "epoch": 0.1271530155385831,
      "grad_norm": 2.2947449684143066,
      "learning_rate": 4.364893336844878e-05,
      "loss": 2.4117,
      "step": 4828
    },
    {
      "epoch": 0.12717935212009482,
      "grad_norm": 1.658942699432373,
      "learning_rate": 4.3647616539373195e-05,
      "loss": 1.7368,
      "step": 4829
    },
    {
      "epoch": 0.12720568870160653,
      "grad_norm": 3.9009971618652344,
      "learning_rate": 4.3646299710297604e-05,
      "loss": 2.0825,
      "step": 4830
    },
    {
      "epoch": 0.12723202528311825,
      "grad_norm": 1.627124309539795,
      "learning_rate": 4.364498288122202e-05,
      "loss": 1.1915,
      "step": 4831
    },
    {
      "epoch": 0.12725836186462997,
      "grad_norm": 1.7527695894241333,
      "learning_rate": 4.364366605214643e-05,
      "loss": 1.9,
      "step": 4832
    },
    {
      "epoch": 0.12728469844614168,
      "grad_norm": 2.3488717079162598,
      "learning_rate": 4.364234922307085e-05,
      "loss": 1.7931,
      "step": 4833
    },
    {
      "epoch": 0.12731103502765342,
      "grad_norm": 2.6414217948913574,
      "learning_rate": 4.364103239399526e-05,
      "loss": 1.2725,
      "step": 4834
    },
    {
      "epoch": 0.12733737160916514,
      "grad_norm": 2.2473621368408203,
      "learning_rate": 4.3639715564919675e-05,
      "loss": 1.6283,
      "step": 4835
    },
    {
      "epoch": 0.12736370819067686,
      "grad_norm": 5.0474348068237305,
      "learning_rate": 4.363839873584409e-05,
      "loss": 1.8299,
      "step": 4836
    },
    {
      "epoch": 0.12739004477218857,
      "grad_norm": 2.4065027236938477,
      "learning_rate": 4.36370819067685e-05,
      "loss": 1.1844,
      "step": 4837
    },
    {
      "epoch": 0.1274163813537003,
      "grad_norm": 2.514936923980713,
      "learning_rate": 4.363576507769292e-05,
      "loss": 1.6467,
      "step": 4838
    },
    {
      "epoch": 0.127442717935212,
      "grad_norm": 3.0012967586517334,
      "learning_rate": 4.363444824861733e-05,
      "loss": 2.094,
      "step": 4839
    },
    {
      "epoch": 0.12746905451672372,
      "grad_norm": 2.124875783920288,
      "learning_rate": 4.363313141954175e-05,
      "loss": 2.0942,
      "step": 4840
    },
    {
      "epoch": 0.12749539109823546,
      "grad_norm": 2.279796600341797,
      "learning_rate": 4.3631814590466155e-05,
      "loss": 1.5752,
      "step": 4841
    },
    {
      "epoch": 0.12752172767974718,
      "grad_norm": 2.1477503776550293,
      "learning_rate": 4.363049776139058e-05,
      "loss": 2.0256,
      "step": 4842
    },
    {
      "epoch": 0.1275480642612589,
      "grad_norm": 2.6695644855499268,
      "learning_rate": 4.362918093231499e-05,
      "loss": 1.5857,
      "step": 4843
    },
    {
      "epoch": 0.1275744008427706,
      "grad_norm": 2.1043524742126465,
      "learning_rate": 4.36278641032394e-05,
      "loss": 1.5499,
      "step": 4844
    },
    {
      "epoch": 0.12760073742428232,
      "grad_norm": 3.444185972213745,
      "learning_rate": 4.362654727416382e-05,
      "loss": 1.6529,
      "step": 4845
    },
    {
      "epoch": 0.12762707400579404,
      "grad_norm": 3.8504743576049805,
      "learning_rate": 4.362523044508823e-05,
      "loss": 1.5419,
      "step": 4846
    },
    {
      "epoch": 0.12765341058730575,
      "grad_norm": 1.9481192827224731,
      "learning_rate": 4.362391361601265e-05,
      "loss": 1.9174,
      "step": 4847
    },
    {
      "epoch": 0.1276797471688175,
      "grad_norm": 1.85298490524292,
      "learning_rate": 4.362259678693706e-05,
      "loss": 1.0213,
      "step": 4848
    },
    {
      "epoch": 0.1277060837503292,
      "grad_norm": 1.7944222688674927,
      "learning_rate": 4.3621279957861473e-05,
      "loss": 2.5456,
      "step": 4849
    },
    {
      "epoch": 0.12773242033184093,
      "grad_norm": 2.6556293964385986,
      "learning_rate": 4.361996312878588e-05,
      "loss": 2.1569,
      "step": 4850
    },
    {
      "epoch": 0.12775875691335264,
      "grad_norm": 1.8357347249984741,
      "learning_rate": 4.36186462997103e-05,
      "loss": 1.1135,
      "step": 4851
    },
    {
      "epoch": 0.12778509349486436,
      "grad_norm": 2.0435867309570312,
      "learning_rate": 4.3617329470634713e-05,
      "loss": 2.1799,
      "step": 4852
    },
    {
      "epoch": 0.12781143007637608,
      "grad_norm": 2.5294694900512695,
      "learning_rate": 4.361601264155913e-05,
      "loss": 0.7898,
      "step": 4853
    },
    {
      "epoch": 0.12783776665788782,
      "grad_norm": 2.220301389694214,
      "learning_rate": 4.3614695812483545e-05,
      "loss": 1.2596,
      "step": 4854
    },
    {
      "epoch": 0.12786410323939953,
      "grad_norm": 2.7293739318847656,
      "learning_rate": 4.3613378983407953e-05,
      "loss": 1.9187,
      "step": 4855
    },
    {
      "epoch": 0.12789043982091125,
      "grad_norm": 1.6175456047058105,
      "learning_rate": 4.361206215433237e-05,
      "loss": 1.8108,
      "step": 4856
    },
    {
      "epoch": 0.12791677640242297,
      "grad_norm": 2.1558797359466553,
      "learning_rate": 4.3610745325256785e-05,
      "loss": 1.8806,
      "step": 4857
    },
    {
      "epoch": 0.12794311298393468,
      "grad_norm": 2.410466432571411,
      "learning_rate": 4.36094284961812e-05,
      "loss": 2.3897,
      "step": 4858
    },
    {
      "epoch": 0.1279694495654464,
      "grad_norm": 1.6283433437347412,
      "learning_rate": 4.360811166710561e-05,
      "loss": 1.8579,
      "step": 4859
    },
    {
      "epoch": 0.1279957861469581,
      "grad_norm": 1.9884577989578247,
      "learning_rate": 4.3606794838030025e-05,
      "loss": 1.6454,
      "step": 4860
    },
    {
      "epoch": 0.12802212272846986,
      "grad_norm": 2.200735092163086,
      "learning_rate": 4.360547800895444e-05,
      "loss": 1.9783,
      "step": 4861
    },
    {
      "epoch": 0.12804845930998157,
      "grad_norm": 2.014044761657715,
      "learning_rate": 4.3604161179878856e-05,
      "loss": 2.2169,
      "step": 4862
    },
    {
      "epoch": 0.1280747958914933,
      "grad_norm": 1.9855282306671143,
      "learning_rate": 4.3602844350803265e-05,
      "loss": 1.971,
      "step": 4863
    },
    {
      "epoch": 0.128101132473005,
      "grad_norm": 2.1201257705688477,
      "learning_rate": 4.360152752172768e-05,
      "loss": 1.6651,
      "step": 4864
    },
    {
      "epoch": 0.12812746905451672,
      "grad_norm": 3.384716033935547,
      "learning_rate": 4.3600210692652096e-05,
      "loss": 1.2641,
      "step": 4865
    },
    {
      "epoch": 0.12815380563602843,
      "grad_norm": 2.70874285697937,
      "learning_rate": 4.359889386357651e-05,
      "loss": 1.9227,
      "step": 4866
    },
    {
      "epoch": 0.12818014221754015,
      "grad_norm": 2.4259488582611084,
      "learning_rate": 4.359757703450093e-05,
      "loss": 1.4218,
      "step": 4867
    },
    {
      "epoch": 0.1282064787990519,
      "grad_norm": 1.9935314655303955,
      "learning_rate": 4.3596260205425336e-05,
      "loss": 1.9993,
      "step": 4868
    },
    {
      "epoch": 0.1282328153805636,
      "grad_norm": 3.310474395751953,
      "learning_rate": 4.359494337634975e-05,
      "loss": 2.1742,
      "step": 4869
    },
    {
      "epoch": 0.12825915196207532,
      "grad_norm": 4.381847381591797,
      "learning_rate": 4.359362654727416e-05,
      "loss": 1.1536,
      "step": 4870
    },
    {
      "epoch": 0.12828548854358704,
      "grad_norm": 2.7546303272247314,
      "learning_rate": 4.359230971819858e-05,
      "loss": 1.9668,
      "step": 4871
    },
    {
      "epoch": 0.12831182512509876,
      "grad_norm": 4.072642803192139,
      "learning_rate": 4.359099288912299e-05,
      "loss": 1.9433,
      "step": 4872
    },
    {
      "epoch": 0.12833816170661047,
      "grad_norm": 2.637031316757202,
      "learning_rate": 4.358967606004741e-05,
      "loss": 0.8301,
      "step": 4873
    },
    {
      "epoch": 0.12836449828812221,
      "grad_norm": 1.7782573699951172,
      "learning_rate": 4.358835923097182e-05,
      "loss": 2.1999,
      "step": 4874
    },
    {
      "epoch": 0.12839083486963393,
      "grad_norm": 2.4592955112457275,
      "learning_rate": 4.358704240189624e-05,
      "loss": 1.8854,
      "step": 4875
    },
    {
      "epoch": 0.12841717145114565,
      "grad_norm": 2.6669809818267822,
      "learning_rate": 4.3585725572820654e-05,
      "loss": 1.4189,
      "step": 4876
    },
    {
      "epoch": 0.12844350803265736,
      "grad_norm": 2.2287843227386475,
      "learning_rate": 4.358440874374506e-05,
      "loss": 1.8585,
      "step": 4877
    },
    {
      "epoch": 0.12846984461416908,
      "grad_norm": 1.9186570644378662,
      "learning_rate": 4.358309191466948e-05,
      "loss": 1.1951,
      "step": 4878
    },
    {
      "epoch": 0.1284961811956808,
      "grad_norm": 1.6467444896697998,
      "learning_rate": 4.358177508559389e-05,
      "loss": 1.4028,
      "step": 4879
    },
    {
      "epoch": 0.1285225177771925,
      "grad_norm": 2.876457452774048,
      "learning_rate": 4.358045825651831e-05,
      "loss": 2.62,
      "step": 4880
    },
    {
      "epoch": 0.12854885435870425,
      "grad_norm": 1.8239116668701172,
      "learning_rate": 4.357914142744272e-05,
      "loss": 1.6768,
      "step": 4881
    },
    {
      "epoch": 0.12857519094021597,
      "grad_norm": 1.834315538406372,
      "learning_rate": 4.3577824598367134e-05,
      "loss": 1.9142,
      "step": 4882
    },
    {
      "epoch": 0.12860152752172768,
      "grad_norm": 1.9506486654281616,
      "learning_rate": 4.357650776929155e-05,
      "loss": 1.9521,
      "step": 4883
    },
    {
      "epoch": 0.1286278641032394,
      "grad_norm": 2.1320502758026123,
      "learning_rate": 4.357519094021596e-05,
      "loss": 1.4065,
      "step": 4884
    },
    {
      "epoch": 0.1286542006847511,
      "grad_norm": 3.5862576961517334,
      "learning_rate": 4.357387411114038e-05,
      "loss": 1.5959,
      "step": 4885
    },
    {
      "epoch": 0.12868053726626283,
      "grad_norm": 3.104295492172241,
      "learning_rate": 4.357255728206479e-05,
      "loss": 2.0909,
      "step": 4886
    },
    {
      "epoch": 0.12870687384777457,
      "grad_norm": 2.3152544498443604,
      "learning_rate": 4.3571240452989205e-05,
      "loss": 2.2944,
      "step": 4887
    },
    {
      "epoch": 0.1287332104292863,
      "grad_norm": 2.4190573692321777,
      "learning_rate": 4.3569923623913614e-05,
      "loss": 2.707,
      "step": 4888
    },
    {
      "epoch": 0.128759547010798,
      "grad_norm": 2.317089557647705,
      "learning_rate": 4.356860679483803e-05,
      "loss": 1.7596,
      "step": 4889
    },
    {
      "epoch": 0.12878588359230972,
      "grad_norm": 2.1676392555236816,
      "learning_rate": 4.3567289965762445e-05,
      "loss": 1.9278,
      "step": 4890
    },
    {
      "epoch": 0.12881222017382143,
      "grad_norm": 2.047677755355835,
      "learning_rate": 4.356597313668686e-05,
      "loss": 2.1574,
      "step": 4891
    },
    {
      "epoch": 0.12883855675533315,
      "grad_norm": 2.923931360244751,
      "learning_rate": 4.3564656307611276e-05,
      "loss": 1.8902,
      "step": 4892
    },
    {
      "epoch": 0.12886489333684487,
      "grad_norm": 2.223792791366577,
      "learning_rate": 4.3563339478535685e-05,
      "loss": 2.0463,
      "step": 4893
    },
    {
      "epoch": 0.1288912299183566,
      "grad_norm": 3.2405905723571777,
      "learning_rate": 4.356202264946011e-05,
      "loss": 0.458,
      "step": 4894
    },
    {
      "epoch": 0.12891756649986832,
      "grad_norm": 1.754435658454895,
      "learning_rate": 4.3560705820384516e-05,
      "loss": 1.1948,
      "step": 4895
    },
    {
      "epoch": 0.12894390308138004,
      "grad_norm": 1.9974554777145386,
      "learning_rate": 4.355938899130893e-05,
      "loss": 2.2276,
      "step": 4896
    },
    {
      "epoch": 0.12897023966289176,
      "grad_norm": 2.052645206451416,
      "learning_rate": 4.355807216223334e-05,
      "loss": 2.0828,
      "step": 4897
    },
    {
      "epoch": 0.12899657624440347,
      "grad_norm": 2.0300865173339844,
      "learning_rate": 4.3556755333157756e-05,
      "loss": 2.0949,
      "step": 4898
    },
    {
      "epoch": 0.1290229128259152,
      "grad_norm": 5.598296642303467,
      "learning_rate": 4.355543850408217e-05,
      "loss": 1.4868,
      "step": 4899
    },
    {
      "epoch": 0.1290492494074269,
      "grad_norm": 4.022143840789795,
      "learning_rate": 4.355412167500659e-05,
      "loss": 0.7322,
      "step": 4900
    },
    {
      "epoch": 0.12907558598893865,
      "grad_norm": 2.7101197242736816,
      "learning_rate": 4.3552804845931e-05,
      "loss": 1.8003,
      "step": 4901
    },
    {
      "epoch": 0.12910192257045036,
      "grad_norm": 3.0394439697265625,
      "learning_rate": 4.355148801685541e-05,
      "loss": 1.2617,
      "step": 4902
    },
    {
      "epoch": 0.12912825915196208,
      "grad_norm": 3.5920867919921875,
      "learning_rate": 4.355017118777983e-05,
      "loss": 1.0817,
      "step": 4903
    },
    {
      "epoch": 0.1291545957334738,
      "grad_norm": 2.459078311920166,
      "learning_rate": 4.354885435870424e-05,
      "loss": 1.3892,
      "step": 4904
    },
    {
      "epoch": 0.1291809323149855,
      "grad_norm": 3.5525012016296387,
      "learning_rate": 4.354753752962866e-05,
      "loss": 1.0874,
      "step": 4905
    },
    {
      "epoch": 0.12920726889649722,
      "grad_norm": 3.375631093978882,
      "learning_rate": 4.354622070055307e-05,
      "loss": 1.8962,
      "step": 4906
    },
    {
      "epoch": 0.12923360547800897,
      "grad_norm": 2.7625391483306885,
      "learning_rate": 4.354490387147748e-05,
      "loss": 2.3779,
      "step": 4907
    },
    {
      "epoch": 0.12925994205952068,
      "grad_norm": 2.535620927810669,
      "learning_rate": 4.35435870424019e-05,
      "loss": 1.8906,
      "step": 4908
    },
    {
      "epoch": 0.1292862786410324,
      "grad_norm": 3.0496389865875244,
      "learning_rate": 4.3542270213326314e-05,
      "loss": 2.121,
      "step": 4909
    },
    {
      "epoch": 0.12931261522254411,
      "grad_norm": 2.272063970565796,
      "learning_rate": 4.354095338425073e-05,
      "loss": 2.343,
      "step": 4910
    },
    {
      "epoch": 0.12933895180405583,
      "grad_norm": 3.0031349658966064,
      "learning_rate": 4.353963655517514e-05,
      "loss": 1.7176,
      "step": 4911
    },
    {
      "epoch": 0.12936528838556755,
      "grad_norm": 2.950576066970825,
      "learning_rate": 4.3538319726099554e-05,
      "loss": 1.4557,
      "step": 4912
    },
    {
      "epoch": 0.12939162496707926,
      "grad_norm": 2.217730760574341,
      "learning_rate": 4.353700289702397e-05,
      "loss": 1.4137,
      "step": 4913
    },
    {
      "epoch": 0.129417961548591,
      "grad_norm": 1.7896230220794678,
      "learning_rate": 4.3535686067948386e-05,
      "loss": 2.3877,
      "step": 4914
    },
    {
      "epoch": 0.12944429813010272,
      "grad_norm": 2.033858060836792,
      "learning_rate": 4.3534369238872794e-05,
      "loss": 1.6677,
      "step": 4915
    },
    {
      "epoch": 0.12947063471161444,
      "grad_norm": 5.609281063079834,
      "learning_rate": 4.353305240979721e-05,
      "loss": 2.0149,
      "step": 4916
    },
    {
      "epoch": 0.12949697129312615,
      "grad_norm": 1.972954511642456,
      "learning_rate": 4.353173558072162e-05,
      "loss": 0.5295,
      "step": 4917
    },
    {
      "epoch": 0.12952330787463787,
      "grad_norm": 4.894515037536621,
      "learning_rate": 4.353041875164604e-05,
      "loss": 1.9628,
      "step": 4918
    },
    {
      "epoch": 0.12954964445614958,
      "grad_norm": 2.383387804031372,
      "learning_rate": 4.352910192257045e-05,
      "loss": 2.2276,
      "step": 4919
    },
    {
      "epoch": 0.12957598103766133,
      "grad_norm": 4.4952521324157715,
      "learning_rate": 4.3527785093494866e-05,
      "loss": 1.7002,
      "step": 4920
    },
    {
      "epoch": 0.12960231761917304,
      "grad_norm": 4.960112571716309,
      "learning_rate": 4.352646826441928e-05,
      "loss": 1.9125,
      "step": 4921
    },
    {
      "epoch": 0.12962865420068476,
      "grad_norm": 1.7794078588485718,
      "learning_rate": 4.352515143534369e-05,
      "loss": 1.572,
      "step": 4922
    },
    {
      "epoch": 0.12965499078219647,
      "grad_norm": 2.117893695831299,
      "learning_rate": 4.352383460626811e-05,
      "loss": 1.9986,
      "step": 4923
    },
    {
      "epoch": 0.1296813273637082,
      "grad_norm": 1.951404094696045,
      "learning_rate": 4.352251777719252e-05,
      "loss": 1.8263,
      "step": 4924
    },
    {
      "epoch": 0.1297076639452199,
      "grad_norm": 1.9802696704864502,
      "learning_rate": 4.352120094811694e-05,
      "loss": 1.6162,
      "step": 4925
    },
    {
      "epoch": 0.12973400052673162,
      "grad_norm": 1.7484253644943237,
      "learning_rate": 4.3519884119041346e-05,
      "loss": 1.4627,
      "step": 4926
    },
    {
      "epoch": 0.12976033710824336,
      "grad_norm": 2.6771886348724365,
      "learning_rate": 4.351856728996577e-05,
      "loss": 1.9035,
      "step": 4927
    },
    {
      "epoch": 0.12978667368975508,
      "grad_norm": 3.7285354137420654,
      "learning_rate": 4.351725046089018e-05,
      "loss": 1.6032,
      "step": 4928
    },
    {
      "epoch": 0.1298130102712668,
      "grad_norm": 3.075129270553589,
      "learning_rate": 4.351593363181459e-05,
      "loss": 1.7752,
      "step": 4929
    },
    {
      "epoch": 0.1298393468527785,
      "grad_norm": 2.682663679122925,
      "learning_rate": 4.351461680273901e-05,
      "loss": 1.6784,
      "step": 4930
    },
    {
      "epoch": 0.12986568343429022,
      "grad_norm": 3.128638744354248,
      "learning_rate": 4.351329997366342e-05,
      "loss": 1.5732,
      "step": 4931
    },
    {
      "epoch": 0.12989202001580194,
      "grad_norm": 2.231123924255371,
      "learning_rate": 4.351198314458784e-05,
      "loss": 2.0879,
      "step": 4932
    },
    {
      "epoch": 0.12991835659731366,
      "grad_norm": 2.612884283065796,
      "learning_rate": 4.351066631551225e-05,
      "loss": 1.0566,
      "step": 4933
    },
    {
      "epoch": 0.1299446931788254,
      "grad_norm": 3.413738250732422,
      "learning_rate": 4.3509349486436664e-05,
      "loss": 1.9005,
      "step": 4934
    },
    {
      "epoch": 0.12997102976033711,
      "grad_norm": 3.060476779937744,
      "learning_rate": 4.350803265736107e-05,
      "loss": 1.1479,
      "step": 4935
    },
    {
      "epoch": 0.12999736634184883,
      "grad_norm": 1.8835155963897705,
      "learning_rate": 4.350671582828549e-05,
      "loss": 1.5931,
      "step": 4936
    },
    {
      "epoch": 0.13002370292336055,
      "grad_norm": 1.9978944063186646,
      "learning_rate": 4.3505398999209904e-05,
      "loss": 1.6942,
      "step": 4937
    },
    {
      "epoch": 0.13005003950487226,
      "grad_norm": 1.673864483833313,
      "learning_rate": 4.350408217013432e-05,
      "loss": 2.6585,
      "step": 4938
    },
    {
      "epoch": 0.13007637608638398,
      "grad_norm": 2.6167819499969482,
      "learning_rate": 4.3502765341058735e-05,
      "loss": 1.9704,
      "step": 4939
    },
    {
      "epoch": 0.13010271266789572,
      "grad_norm": 1.9489105939865112,
      "learning_rate": 4.3501448511983144e-05,
      "loss": 1.9714,
      "step": 4940
    },
    {
      "epoch": 0.13012904924940744,
      "grad_norm": 2.7575132846832275,
      "learning_rate": 4.3500131682907566e-05,
      "loss": 1.6787,
      "step": 4941
    },
    {
      "epoch": 0.13015538583091915,
      "grad_norm": 2.8675830364227295,
      "learning_rate": 4.3498814853831975e-05,
      "loss": 1.7496,
      "step": 4942
    },
    {
      "epoch": 0.13018172241243087,
      "grad_norm": 3.113664388656616,
      "learning_rate": 4.349749802475639e-05,
      "loss": 1.8952,
      "step": 4943
    },
    {
      "epoch": 0.13020805899394258,
      "grad_norm": 2.416672706604004,
      "learning_rate": 4.34961811956808e-05,
      "loss": 2.2312,
      "step": 4944
    },
    {
      "epoch": 0.1302343955754543,
      "grad_norm": 2.531726121902466,
      "learning_rate": 4.3494864366605215e-05,
      "loss": 1.9506,
      "step": 4945
    },
    {
      "epoch": 0.130260732156966,
      "grad_norm": 2.0439252853393555,
      "learning_rate": 4.349354753752963e-05,
      "loss": 2.4105,
      "step": 4946
    },
    {
      "epoch": 0.13028706873847776,
      "grad_norm": 2.0388312339782715,
      "learning_rate": 4.3492230708454046e-05,
      "loss": 1.8746,
      "step": 4947
    },
    {
      "epoch": 0.13031340531998947,
      "grad_norm": 3.1810717582702637,
      "learning_rate": 4.349091387937846e-05,
      "loss": 0.7253,
      "step": 4948
    },
    {
      "epoch": 0.1303397419015012,
      "grad_norm": 6.146149635314941,
      "learning_rate": 4.348959705030287e-05,
      "loss": 1.1874,
      "step": 4949
    },
    {
      "epoch": 0.1303660784830129,
      "grad_norm": 13.208438873291016,
      "learning_rate": 4.3488280221227286e-05,
      "loss": 2.7628,
      "step": 4950
    },
    {
      "epoch": 0.13039241506452462,
      "grad_norm": 5.336616039276123,
      "learning_rate": 4.34869633921517e-05,
      "loss": 0.8695,
      "step": 4951
    },
    {
      "epoch": 0.13041875164603633,
      "grad_norm": 4.059808731079102,
      "learning_rate": 4.348564656307612e-05,
      "loss": 1.4424,
      "step": 4952
    },
    {
      "epoch": 0.13044508822754805,
      "grad_norm": 3.747255563735962,
      "learning_rate": 4.3484329734000526e-05,
      "loss": 1.6961,
      "step": 4953
    },
    {
      "epoch": 0.1304714248090598,
      "grad_norm": 1.8338106870651245,
      "learning_rate": 4.348301290492494e-05,
      "loss": 1.643,
      "step": 4954
    },
    {
      "epoch": 0.1304977613905715,
      "grad_norm": 3.2388992309570312,
      "learning_rate": 4.348169607584936e-05,
      "loss": 0.992,
      "step": 4955
    },
    {
      "epoch": 0.13052409797208323,
      "grad_norm": 2.0383005142211914,
      "learning_rate": 4.348037924677377e-05,
      "loss": 2.2097,
      "step": 4956
    },
    {
      "epoch": 0.13055043455359494,
      "grad_norm": 1.8579005002975464,
      "learning_rate": 4.347906241769819e-05,
      "loss": 1.7505,
      "step": 4957
    },
    {
      "epoch": 0.13057677113510666,
      "grad_norm": 2.4393959045410156,
      "learning_rate": 4.34777455886226e-05,
      "loss": 1.5814,
      "step": 4958
    },
    {
      "epoch": 0.13060310771661837,
      "grad_norm": 2.926823854446411,
      "learning_rate": 4.347642875954701e-05,
      "loss": 1.4456,
      "step": 4959
    },
    {
      "epoch": 0.13062944429813012,
      "grad_norm": 2.2165017127990723,
      "learning_rate": 4.347511193047143e-05,
      "loss": 1.7174,
      "step": 4960
    },
    {
      "epoch": 0.13065578087964183,
      "grad_norm": 1.8240114450454712,
      "learning_rate": 4.3473795101395844e-05,
      "loss": 1.8518,
      "step": 4961
    },
    {
      "epoch": 0.13068211746115355,
      "grad_norm": 2.0179178714752197,
      "learning_rate": 4.347247827232025e-05,
      "loss": 1.9723,
      "step": 4962
    },
    {
      "epoch": 0.13070845404266526,
      "grad_norm": 3.113919258117676,
      "learning_rate": 4.347116144324467e-05,
      "loss": 0.5512,
      "step": 4963
    },
    {
      "epoch": 0.13073479062417698,
      "grad_norm": 1.8319178819656372,
      "learning_rate": 4.346984461416908e-05,
      "loss": 1.9527,
      "step": 4964
    },
    {
      "epoch": 0.1307611272056887,
      "grad_norm": 2.051953077316284,
      "learning_rate": 4.34685277850935e-05,
      "loss": 1.2838,
      "step": 4965
    },
    {
      "epoch": 0.1307874637872004,
      "grad_norm": 1.9818252325057983,
      "learning_rate": 4.346721095601791e-05,
      "loss": 1.659,
      "step": 4966
    },
    {
      "epoch": 0.13081380036871215,
      "grad_norm": 2.53497576713562,
      "learning_rate": 4.3465894126942324e-05,
      "loss": 1.5803,
      "step": 4967
    },
    {
      "epoch": 0.13084013695022387,
      "grad_norm": 1.9523745775222778,
      "learning_rate": 4.346457729786674e-05,
      "loss": 1.9161,
      "step": 4968
    },
    {
      "epoch": 0.13086647353173558,
      "grad_norm": 3.2774455547332764,
      "learning_rate": 4.346326046879115e-05,
      "loss": 1.5952,
      "step": 4969
    },
    {
      "epoch": 0.1308928101132473,
      "grad_norm": 2.6098408699035645,
      "learning_rate": 4.346194363971557e-05,
      "loss": 1.4197,
      "step": 4970
    },
    {
      "epoch": 0.13091914669475901,
      "grad_norm": 3.690000534057617,
      "learning_rate": 4.346062681063998e-05,
      "loss": 0.8491,
      "step": 4971
    },
    {
      "epoch": 0.13094548327627073,
      "grad_norm": 2.4877395629882812,
      "learning_rate": 4.3459309981564395e-05,
      "loss": 2.0821,
      "step": 4972
    },
    {
      "epoch": 0.13097181985778247,
      "grad_norm": 3.571730613708496,
      "learning_rate": 4.3457993152488804e-05,
      "loss": 2.1172,
      "step": 4973
    },
    {
      "epoch": 0.1309981564392942,
      "grad_norm": 2.088020086288452,
      "learning_rate": 4.345667632341323e-05,
      "loss": 1.5379,
      "step": 4974
    },
    {
      "epoch": 0.1310244930208059,
      "grad_norm": 2.455195903778076,
      "learning_rate": 4.3455359494337635e-05,
      "loss": 2.1159,
      "step": 4975
    },
    {
      "epoch": 0.13105082960231762,
      "grad_norm": 2.3960843086242676,
      "learning_rate": 4.345404266526205e-05,
      "loss": 1.7115,
      "step": 4976
    },
    {
      "epoch": 0.13107716618382934,
      "grad_norm": 1.7375211715698242,
      "learning_rate": 4.345272583618647e-05,
      "loss": 2.0856,
      "step": 4977
    },
    {
      "epoch": 0.13110350276534105,
      "grad_norm": 4.1101226806640625,
      "learning_rate": 4.3451409007110876e-05,
      "loss": 0.9096,
      "step": 4978
    },
    {
      "epoch": 0.13112983934685277,
      "grad_norm": 2.3209168910980225,
      "learning_rate": 4.34500921780353e-05,
      "loss": 2.0009,
      "step": 4979
    },
    {
      "epoch": 0.1311561759283645,
      "grad_norm": 2.01666522026062,
      "learning_rate": 4.344877534895971e-05,
      "loss": 0.5954,
      "step": 4980
    },
    {
      "epoch": 0.13118251250987623,
      "grad_norm": 1.6926815509796143,
      "learning_rate": 4.344745851988412e-05,
      "loss": 1.9913,
      "step": 4981
    },
    {
      "epoch": 0.13120884909138794,
      "grad_norm": 2.902639389038086,
      "learning_rate": 4.344614169080853e-05,
      "loss": 0.8996,
      "step": 4982
    },
    {
      "epoch": 0.13123518567289966,
      "grad_norm": 2.468534231185913,
      "learning_rate": 4.344482486173295e-05,
      "loss": 2.0002,
      "step": 4983
    },
    {
      "epoch": 0.13126152225441137,
      "grad_norm": 1.7324596643447876,
      "learning_rate": 4.344350803265736e-05,
      "loss": 1.7669,
      "step": 4984
    },
    {
      "epoch": 0.1312878588359231,
      "grad_norm": 2.492088794708252,
      "learning_rate": 4.344219120358178e-05,
      "loss": 2.3001,
      "step": 4985
    },
    {
      "epoch": 0.1313141954174348,
      "grad_norm": 2.460096597671509,
      "learning_rate": 4.3440874374506194e-05,
      "loss": 1.9039,
      "step": 4986
    },
    {
      "epoch": 0.13134053199894655,
      "grad_norm": 1.9136123657226562,
      "learning_rate": 4.34395575454306e-05,
      "loss": 1.9006,
      "step": 4987
    },
    {
      "epoch": 0.13136686858045826,
      "grad_norm": 2.5846214294433594,
      "learning_rate": 4.343824071635502e-05,
      "loss": 1.6338,
      "step": 4988
    },
    {
      "epoch": 0.13139320516196998,
      "grad_norm": 2.966732978820801,
      "learning_rate": 4.3436923887279434e-05,
      "loss": 1.6819,
      "step": 4989
    },
    {
      "epoch": 0.1314195417434817,
      "grad_norm": 1.8261289596557617,
      "learning_rate": 4.343560705820385e-05,
      "loss": 1.071,
      "step": 4990
    },
    {
      "epoch": 0.1314458783249934,
      "grad_norm": 3.538123369216919,
      "learning_rate": 4.343429022912826e-05,
      "loss": 1.2945,
      "step": 4991
    },
    {
      "epoch": 0.13147221490650512,
      "grad_norm": 3.3220880031585693,
      "learning_rate": 4.3432973400052674e-05,
      "loss": 1.6983,
      "step": 4992
    },
    {
      "epoch": 0.13149855148801687,
      "grad_norm": 1.8842847347259521,
      "learning_rate": 4.343165657097709e-05,
      "loss": 1.873,
      "step": 4993
    },
    {
      "epoch": 0.13152488806952858,
      "grad_norm": 4.425382614135742,
      "learning_rate": 4.3430339741901505e-05,
      "loss": 1.9297,
      "step": 4994
    },
    {
      "epoch": 0.1315512246510403,
      "grad_norm": 2.080655336380005,
      "learning_rate": 4.342902291282592e-05,
      "loss": 1.8793,
      "step": 4995
    },
    {
      "epoch": 0.13157756123255201,
      "grad_norm": 1.711382508277893,
      "learning_rate": 4.342770608375033e-05,
      "loss": 0.6949,
      "step": 4996
    },
    {
      "epoch": 0.13160389781406373,
      "grad_norm": 2.0423460006713867,
      "learning_rate": 4.3426389254674745e-05,
      "loss": 2.1115,
      "step": 4997
    },
    {
      "epoch": 0.13163023439557545,
      "grad_norm": 6.049909591674805,
      "learning_rate": 4.342507242559916e-05,
      "loss": 1.6203,
      "step": 4998
    },
    {
      "epoch": 0.13165657097708716,
      "grad_norm": 2.2805228233337402,
      "learning_rate": 4.3423755596523576e-05,
      "loss": 1.6269,
      "step": 4999
    },
    {
      "epoch": 0.1316829075585989,
      "grad_norm": 2.6756784915924072,
      "learning_rate": 4.3422438767447985e-05,
      "loss": 2.2693,
      "step": 5000
    },
    {
      "epoch": 0.13170924414011062,
      "grad_norm": 2.5561397075653076,
      "learning_rate": 4.34211219383724e-05,
      "loss": 0.6804,
      "step": 5001
    },
    {
      "epoch": 0.13173558072162234,
      "grad_norm": 3.445418119430542,
      "learning_rate": 4.3419805109296816e-05,
      "loss": 1.6649,
      "step": 5002
    },
    {
      "epoch": 0.13176191730313405,
      "grad_norm": 2.469170093536377,
      "learning_rate": 4.341848828022123e-05,
      "loss": 2.1,
      "step": 5003
    },
    {
      "epoch": 0.13178825388464577,
      "grad_norm": 2.5011355876922607,
      "learning_rate": 4.341717145114565e-05,
      "loss": 1.6431,
      "step": 5004
    },
    {
      "epoch": 0.13181459046615748,
      "grad_norm": 2.8519020080566406,
      "learning_rate": 4.3415854622070056e-05,
      "loss": 1.4282,
      "step": 5005
    },
    {
      "epoch": 0.13184092704766923,
      "grad_norm": 2.6416091918945312,
      "learning_rate": 4.341453779299447e-05,
      "loss": 1.4372,
      "step": 5006
    },
    {
      "epoch": 0.13186726362918094,
      "grad_norm": 2.075183629989624,
      "learning_rate": 4.341322096391889e-05,
      "loss": 1.756,
      "step": 5007
    },
    {
      "epoch": 0.13189360021069266,
      "grad_norm": 3.2140908241271973,
      "learning_rate": 4.34119041348433e-05,
      "loss": 1.4966,
      "step": 5008
    },
    {
      "epoch": 0.13191993679220437,
      "grad_norm": 1.8258230686187744,
      "learning_rate": 4.341058730576771e-05,
      "loss": 1.8709,
      "step": 5009
    },
    {
      "epoch": 0.1319462733737161,
      "grad_norm": 3.9869112968444824,
      "learning_rate": 4.340927047669213e-05,
      "loss": 1.6423,
      "step": 5010
    },
    {
      "epoch": 0.1319726099552278,
      "grad_norm": 3.122354745864868,
      "learning_rate": 4.3407953647616536e-05,
      "loss": 1.8475,
      "step": 5011
    },
    {
      "epoch": 0.13199894653673952,
      "grad_norm": 2.240496873855591,
      "learning_rate": 4.340663681854096e-05,
      "loss": 1.635,
      "step": 5012
    },
    {
      "epoch": 0.13202528311825126,
      "grad_norm": 1.9871795177459717,
      "learning_rate": 4.340531998946537e-05,
      "loss": 1.4934,
      "step": 5013
    },
    {
      "epoch": 0.13205161969976298,
      "grad_norm": 1.8549997806549072,
      "learning_rate": 4.340400316038978e-05,
      "loss": 1.5648,
      "step": 5014
    },
    {
      "epoch": 0.1320779562812747,
      "grad_norm": 2.6075167655944824,
      "learning_rate": 4.34026863313142e-05,
      "loss": 1.5973,
      "step": 5015
    },
    {
      "epoch": 0.1321042928627864,
      "grad_norm": 4.122131824493408,
      "learning_rate": 4.340136950223861e-05,
      "loss": 0.6709,
      "step": 5016
    },
    {
      "epoch": 0.13213062944429813,
      "grad_norm": 2.711064577102661,
      "learning_rate": 4.340005267316303e-05,
      "loss": 1.0608,
      "step": 5017
    },
    {
      "epoch": 0.13215696602580984,
      "grad_norm": 3.5394294261932373,
      "learning_rate": 4.339873584408744e-05,
      "loss": 1.4017,
      "step": 5018
    },
    {
      "epoch": 0.13218330260732156,
      "grad_norm": 2.4709997177124023,
      "learning_rate": 4.3397419015011854e-05,
      "loss": 2.4186,
      "step": 5019
    },
    {
      "epoch": 0.1322096391888333,
      "grad_norm": 2.5101897716522217,
      "learning_rate": 4.339610218593626e-05,
      "loss": 1.653,
      "step": 5020
    },
    {
      "epoch": 0.13223597577034502,
      "grad_norm": 1.7431272268295288,
      "learning_rate": 4.339478535686068e-05,
      "loss": 1.737,
      "step": 5021
    },
    {
      "epoch": 0.13226231235185673,
      "grad_norm": 2.694981575012207,
      "learning_rate": 4.3393468527785094e-05,
      "loss": 1.3452,
      "step": 5022
    },
    {
      "epoch": 0.13228864893336845,
      "grad_norm": 4.8767476081848145,
      "learning_rate": 4.339215169870951e-05,
      "loss": 1.9107,
      "step": 5023
    },
    {
      "epoch": 0.13231498551488016,
      "grad_norm": 1.9706566333770752,
      "learning_rate": 4.3390834869633925e-05,
      "loss": 2.4199,
      "step": 5024
    },
    {
      "epoch": 0.13234132209639188,
      "grad_norm": 3.8935935497283936,
      "learning_rate": 4.3389518040558334e-05,
      "loss": 2.6432,
      "step": 5025
    },
    {
      "epoch": 0.13236765867790362,
      "grad_norm": 2.5822763442993164,
      "learning_rate": 4.3388201211482756e-05,
      "loss": 0.5459,
      "step": 5026
    },
    {
      "epoch": 0.13239399525941534,
      "grad_norm": 1.74319326877594,
      "learning_rate": 4.3386884382407165e-05,
      "loss": 1.7364,
      "step": 5027
    },
    {
      "epoch": 0.13242033184092705,
      "grad_norm": 1.8894635438919067,
      "learning_rate": 4.338556755333158e-05,
      "loss": 0.7984,
      "step": 5028
    },
    {
      "epoch": 0.13244666842243877,
      "grad_norm": 2.569887161254883,
      "learning_rate": 4.338425072425599e-05,
      "loss": 1.5337,
      "step": 5029
    },
    {
      "epoch": 0.13247300500395048,
      "grad_norm": 2.3233461380004883,
      "learning_rate": 4.3382933895180405e-05,
      "loss": 1.4143,
      "step": 5030
    },
    {
      "epoch": 0.1324993415854622,
      "grad_norm": 2.9843146800994873,
      "learning_rate": 4.338161706610482e-05,
      "loss": 1.4322,
      "step": 5031
    },
    {
      "epoch": 0.13252567816697391,
      "grad_norm": 2.2174909114837646,
      "learning_rate": 4.3380300237029236e-05,
      "loss": 1.4158,
      "step": 5032
    },
    {
      "epoch": 0.13255201474848566,
      "grad_norm": 2.8234145641326904,
      "learning_rate": 4.337898340795365e-05,
      "loss": 2.3157,
      "step": 5033
    },
    {
      "epoch": 0.13257835132999737,
      "grad_norm": 3.4443726539611816,
      "learning_rate": 4.337766657887806e-05,
      "loss": 1.3387,
      "step": 5034
    },
    {
      "epoch": 0.1326046879115091,
      "grad_norm": 4.984408855438232,
      "learning_rate": 4.3376349749802476e-05,
      "loss": 0.4735,
      "step": 5035
    },
    {
      "epoch": 0.1326310244930208,
      "grad_norm": 3.2236123085021973,
      "learning_rate": 4.337503292072689e-05,
      "loss": 1.5339,
      "step": 5036
    },
    {
      "epoch": 0.13265736107453252,
      "grad_norm": 2.643332004547119,
      "learning_rate": 4.337371609165131e-05,
      "loss": 0.853,
      "step": 5037
    },
    {
      "epoch": 0.13268369765604424,
      "grad_norm": 4.635191440582275,
      "learning_rate": 4.3372399262575716e-05,
      "loss": 1.551,
      "step": 5038
    },
    {
      "epoch": 0.13271003423755595,
      "grad_norm": 2.006472110748291,
      "learning_rate": 4.337108243350013e-05,
      "loss": 1.6254,
      "step": 5039
    },
    {
      "epoch": 0.1327363708190677,
      "grad_norm": 2.1525752544403076,
      "learning_rate": 4.336976560442455e-05,
      "loss": 2.2399,
      "step": 5040
    },
    {
      "epoch": 0.1327627074005794,
      "grad_norm": 3.3003857135772705,
      "learning_rate": 4.336844877534896e-05,
      "loss": 1.8106,
      "step": 5041
    },
    {
      "epoch": 0.13278904398209113,
      "grad_norm": 5.573011875152588,
      "learning_rate": 4.336713194627338e-05,
      "loss": 1.3819,
      "step": 5042
    },
    {
      "epoch": 0.13281538056360284,
      "grad_norm": 5.78550910949707,
      "learning_rate": 4.336581511719779e-05,
      "loss": 1.3246,
      "step": 5043
    },
    {
      "epoch": 0.13284171714511456,
      "grad_norm": 3.1437580585479736,
      "learning_rate": 4.33644982881222e-05,
      "loss": 1.7696,
      "step": 5044
    },
    {
      "epoch": 0.13286805372662627,
      "grad_norm": 2.3968985080718994,
      "learning_rate": 4.336318145904662e-05,
      "loss": 1.5599,
      "step": 5045
    },
    {
      "epoch": 0.13289439030813802,
      "grad_norm": 2.5823965072631836,
      "learning_rate": 4.3361864629971035e-05,
      "loss": 2.0172,
      "step": 5046
    },
    {
      "epoch": 0.13292072688964973,
      "grad_norm": 1.743168592453003,
      "learning_rate": 4.336054780089544e-05,
      "loss": 2.0607,
      "step": 5047
    },
    {
      "epoch": 0.13294706347116145,
      "grad_norm": 2.6268415451049805,
      "learning_rate": 4.335923097181986e-05,
      "loss": 2.1507,
      "step": 5048
    },
    {
      "epoch": 0.13297340005267316,
      "grad_norm": 2.297821044921875,
      "learning_rate": 4.3357914142744275e-05,
      "loss": 2.2302,
      "step": 5049
    },
    {
      "epoch": 0.13299973663418488,
      "grad_norm": 1.988111972808838,
      "learning_rate": 4.335659731366869e-05,
      "loss": 1.6013,
      "step": 5050
    },
    {
      "epoch": 0.1330260732156966,
      "grad_norm": 4.546943187713623,
      "learning_rate": 4.3355280484593106e-05,
      "loss": 1.2326,
      "step": 5051
    },
    {
      "epoch": 0.1330524097972083,
      "grad_norm": 3.1077394485473633,
      "learning_rate": 4.3353963655517515e-05,
      "loss": 1.5866,
      "step": 5052
    },
    {
      "epoch": 0.13307874637872005,
      "grad_norm": 13.691081047058105,
      "learning_rate": 4.335264682644193e-05,
      "loss": 2.7875,
      "step": 5053
    },
    {
      "epoch": 0.13310508296023177,
      "grad_norm": 2.2888708114624023,
      "learning_rate": 4.335132999736634e-05,
      "loss": 1.2606,
      "step": 5054
    },
    {
      "epoch": 0.13313141954174348,
      "grad_norm": 2.7057507038116455,
      "learning_rate": 4.335001316829076e-05,
      "loss": 1.8512,
      "step": 5055
    },
    {
      "epoch": 0.1331577561232552,
      "grad_norm": 2.7668721675872803,
      "learning_rate": 4.334869633921517e-05,
      "loss": 1.9467,
      "step": 5056
    },
    {
      "epoch": 0.13318409270476692,
      "grad_norm": 2.0387139320373535,
      "learning_rate": 4.3347379510139586e-05,
      "loss": 1.7528,
      "step": 5057
    },
    {
      "epoch": 0.13321042928627863,
      "grad_norm": 2.803766965866089,
      "learning_rate": 4.3346062681064e-05,
      "loss": 2.3653,
      "step": 5058
    },
    {
      "epoch": 0.13323676586779037,
      "grad_norm": 4.716665267944336,
      "learning_rate": 4.334474585198842e-05,
      "loss": 1.2364,
      "step": 5059
    },
    {
      "epoch": 0.1332631024493021,
      "grad_norm": 1.9011110067367554,
      "learning_rate": 4.334342902291283e-05,
      "loss": 1.4576,
      "step": 5060
    },
    {
      "epoch": 0.1332894390308138,
      "grad_norm": 3.019423484802246,
      "learning_rate": 4.334211219383724e-05,
      "loss": 0.7776,
      "step": 5061
    },
    {
      "epoch": 0.13331577561232552,
      "grad_norm": 2.002833127975464,
      "learning_rate": 4.334079536476166e-05,
      "loss": 2.0146,
      "step": 5062
    },
    {
      "epoch": 0.13334211219383724,
      "grad_norm": 5.950443744659424,
      "learning_rate": 4.3339478535686066e-05,
      "loss": 1.0814,
      "step": 5063
    },
    {
      "epoch": 0.13336844877534895,
      "grad_norm": 2.671046733856201,
      "learning_rate": 4.333816170661049e-05,
      "loss": 1.1213,
      "step": 5064
    },
    {
      "epoch": 0.13339478535686067,
      "grad_norm": 2.2727088928222656,
      "learning_rate": 4.33368448775349e-05,
      "loss": 1.684,
      "step": 5065
    },
    {
      "epoch": 0.1334211219383724,
      "grad_norm": 1.7243009805679321,
      "learning_rate": 4.333552804845931e-05,
      "loss": 2.0375,
      "step": 5066
    },
    {
      "epoch": 0.13344745851988413,
      "grad_norm": 4.161171913146973,
      "learning_rate": 4.333421121938372e-05,
      "loss": 1.1078,
      "step": 5067
    },
    {
      "epoch": 0.13347379510139584,
      "grad_norm": 6.115097522735596,
      "learning_rate": 4.333289439030814e-05,
      "loss": 1.3167,
      "step": 5068
    },
    {
      "epoch": 0.13350013168290756,
      "grad_norm": 1.7010360956192017,
      "learning_rate": 4.333157756123255e-05,
      "loss": 1.5568,
      "step": 5069
    },
    {
      "epoch": 0.13352646826441927,
      "grad_norm": 5.253629684448242,
      "learning_rate": 4.333026073215697e-05,
      "loss": 1.6685,
      "step": 5070
    },
    {
      "epoch": 0.133552804845931,
      "grad_norm": 6.189787864685059,
      "learning_rate": 4.3328943903081384e-05,
      "loss": 1.5829,
      "step": 5071
    },
    {
      "epoch": 0.1335791414274427,
      "grad_norm": 3.063844919204712,
      "learning_rate": 4.332762707400579e-05,
      "loss": 2.8246,
      "step": 5072
    },
    {
      "epoch": 0.13360547800895445,
      "grad_norm": 3.2947885990142822,
      "learning_rate": 4.3326310244930215e-05,
      "loss": 0.5063,
      "step": 5073
    },
    {
      "epoch": 0.13363181459046616,
      "grad_norm": 2.2428524494171143,
      "learning_rate": 4.3324993415854624e-05,
      "loss": 1.5747,
      "step": 5074
    },
    {
      "epoch": 0.13365815117197788,
      "grad_norm": 2.1810524463653564,
      "learning_rate": 4.332367658677904e-05,
      "loss": 1.1996,
      "step": 5075
    },
    {
      "epoch": 0.1336844877534896,
      "grad_norm": 3.762620449066162,
      "learning_rate": 4.332235975770345e-05,
      "loss": 1.7491,
      "step": 5076
    },
    {
      "epoch": 0.1337108243350013,
      "grad_norm": 2.241297960281372,
      "learning_rate": 4.3321042928627864e-05,
      "loss": 1.9091,
      "step": 5077
    },
    {
      "epoch": 0.13373716091651303,
      "grad_norm": 2.325817823410034,
      "learning_rate": 4.331972609955228e-05,
      "loss": 2.1807,
      "step": 5078
    },
    {
      "epoch": 0.13376349749802477,
      "grad_norm": 2.515507459640503,
      "learning_rate": 4.3318409270476695e-05,
      "loss": 1.7233,
      "step": 5079
    },
    {
      "epoch": 0.13378983407953648,
      "grad_norm": 2.0451629161834717,
      "learning_rate": 4.331709244140111e-05,
      "loss": 1.8693,
      "step": 5080
    },
    {
      "epoch": 0.1338161706610482,
      "grad_norm": 2.8333163261413574,
      "learning_rate": 4.331577561232552e-05,
      "loss": 0.6471,
      "step": 5081
    },
    {
      "epoch": 0.13384250724255992,
      "grad_norm": 2.7954535484313965,
      "learning_rate": 4.3314458783249935e-05,
      "loss": 0.9213,
      "step": 5082
    },
    {
      "epoch": 0.13386884382407163,
      "grad_norm": 2.4277091026306152,
      "learning_rate": 4.331314195417435e-05,
      "loss": 1.2899,
      "step": 5083
    },
    {
      "epoch": 0.13389518040558335,
      "grad_norm": 2.7773337364196777,
      "learning_rate": 4.3311825125098766e-05,
      "loss": 1.3181,
      "step": 5084
    },
    {
      "epoch": 0.13392151698709506,
      "grad_norm": 2.029878854751587,
      "learning_rate": 4.3310508296023175e-05,
      "loss": 1.6522,
      "step": 5085
    },
    {
      "epoch": 0.1339478535686068,
      "grad_norm": 4.992863178253174,
      "learning_rate": 4.330919146694759e-05,
      "loss": 1.5919,
      "step": 5086
    },
    {
      "epoch": 0.13397419015011852,
      "grad_norm": 3.617291212081909,
      "learning_rate": 4.3307874637872006e-05,
      "loss": 2.0032,
      "step": 5087
    },
    {
      "epoch": 0.13400052673163024,
      "grad_norm": 3.9616904258728027,
      "learning_rate": 4.330655780879642e-05,
      "loss": 2.205,
      "step": 5088
    },
    {
      "epoch": 0.13402686331314195,
      "grad_norm": 2.2840869426727295,
      "learning_rate": 4.330524097972084e-05,
      "loss": 1.9219,
      "step": 5089
    },
    {
      "epoch": 0.13405319989465367,
      "grad_norm": 3.934386730194092,
      "learning_rate": 4.3303924150645246e-05,
      "loss": 1.0264,
      "step": 5090
    },
    {
      "epoch": 0.13407953647616538,
      "grad_norm": 2.615556001663208,
      "learning_rate": 4.330260732156966e-05,
      "loss": 2.6591,
      "step": 5091
    },
    {
      "epoch": 0.1341058730576771,
      "grad_norm": 1.6333255767822266,
      "learning_rate": 4.330129049249408e-05,
      "loss": 1.4891,
      "step": 5092
    },
    {
      "epoch": 0.13413220963918884,
      "grad_norm": 1.5788342952728271,
      "learning_rate": 4.329997366341849e-05,
      "loss": 1.739,
      "step": 5093
    },
    {
      "epoch": 0.13415854622070056,
      "grad_norm": 3.678856611251831,
      "learning_rate": 4.32986568343429e-05,
      "loss": 1.141,
      "step": 5094
    },
    {
      "epoch": 0.13418488280221227,
      "grad_norm": 2.5560591220855713,
      "learning_rate": 4.329734000526732e-05,
      "loss": 2.0155,
      "step": 5095
    },
    {
      "epoch": 0.134211219383724,
      "grad_norm": 2.2019999027252197,
      "learning_rate": 4.329602317619173e-05,
      "loss": 1.6245,
      "step": 5096
    },
    {
      "epoch": 0.1342375559652357,
      "grad_norm": 3.82480525970459,
      "learning_rate": 4.329470634711615e-05,
      "loss": 1.9263,
      "step": 5097
    },
    {
      "epoch": 0.13426389254674742,
      "grad_norm": 2.56531023979187,
      "learning_rate": 4.3293389518040564e-05,
      "loss": 2.2144,
      "step": 5098
    },
    {
      "epoch": 0.13429022912825916,
      "grad_norm": 1.8532615900039673,
      "learning_rate": 4.329207268896497e-05,
      "loss": 1.8402,
      "step": 5099
    },
    {
      "epoch": 0.13431656570977088,
      "grad_norm": 3.629487991333008,
      "learning_rate": 4.329075585988939e-05,
      "loss": 2.5898,
      "step": 5100
    },
    {
      "epoch": 0.1343429022912826,
      "grad_norm": 2.0321056842803955,
      "learning_rate": 4.32894390308138e-05,
      "loss": 1.8867,
      "step": 5101
    },
    {
      "epoch": 0.1343692388727943,
      "grad_norm": 2.923245906829834,
      "learning_rate": 4.328812220173822e-05,
      "loss": 1.4847,
      "step": 5102
    },
    {
      "epoch": 0.13439557545430603,
      "grad_norm": 2.0400636196136475,
      "learning_rate": 4.328680537266263e-05,
      "loss": 2.1398,
      "step": 5103
    },
    {
      "epoch": 0.13442191203581774,
      "grad_norm": 3.8229000568389893,
      "learning_rate": 4.3285488543587044e-05,
      "loss": 0.9901,
      "step": 5104
    },
    {
      "epoch": 0.13444824861732946,
      "grad_norm": 2.7378499507904053,
      "learning_rate": 4.328417171451146e-05,
      "loss": 1.7758,
      "step": 5105
    },
    {
      "epoch": 0.1344745851988412,
      "grad_norm": 1.6413730382919312,
      "learning_rate": 4.3282854885435875e-05,
      "loss": 1.6947,
      "step": 5106
    },
    {
      "epoch": 0.13450092178035292,
      "grad_norm": 2.3563058376312256,
      "learning_rate": 4.328153805636029e-05,
      "loss": 2.247,
      "step": 5107
    },
    {
      "epoch": 0.13452725836186463,
      "grad_norm": 3.0296270847320557,
      "learning_rate": 4.32802212272847e-05,
      "loss": 1.2415,
      "step": 5108
    },
    {
      "epoch": 0.13455359494337635,
      "grad_norm": 1.9270447492599487,
      "learning_rate": 4.3278904398209116e-05,
      "loss": 2.082,
      "step": 5109
    },
    {
      "epoch": 0.13457993152488806,
      "grad_norm": 2.082814931869507,
      "learning_rate": 4.3277587569133524e-05,
      "loss": 0.4957,
      "step": 5110
    },
    {
      "epoch": 0.13460626810639978,
      "grad_norm": 2.044718027114868,
      "learning_rate": 4.327627074005795e-05,
      "loss": 2.3478,
      "step": 5111
    },
    {
      "epoch": 0.13463260468791152,
      "grad_norm": 4.23112678527832,
      "learning_rate": 4.3274953910982356e-05,
      "loss": 1.8147,
      "step": 5112
    },
    {
      "epoch": 0.13465894126942324,
      "grad_norm": 1.7309212684631348,
      "learning_rate": 4.327363708190677e-05,
      "loss": 1.9121,
      "step": 5113
    },
    {
      "epoch": 0.13468527785093495,
      "grad_norm": 1.6054695844650269,
      "learning_rate": 4.327232025283118e-05,
      "loss": 1.9575,
      "step": 5114
    },
    {
      "epoch": 0.13471161443244667,
      "grad_norm": 1.9052006006240845,
      "learning_rate": 4.3271003423755596e-05,
      "loss": 2.5209,
      "step": 5115
    },
    {
      "epoch": 0.13473795101395838,
      "grad_norm": 2.179224967956543,
      "learning_rate": 4.326968659468001e-05,
      "loss": 1.9486,
      "step": 5116
    },
    {
      "epoch": 0.1347642875954701,
      "grad_norm": 2.9472708702087402,
      "learning_rate": 4.326836976560443e-05,
      "loss": 1.3445,
      "step": 5117
    },
    {
      "epoch": 0.13479062417698182,
      "grad_norm": 4.949427127838135,
      "learning_rate": 4.326705293652884e-05,
      "loss": 1.6875,
      "step": 5118
    },
    {
      "epoch": 0.13481696075849356,
      "grad_norm": NaN,
      "learning_rate": 4.326705293652884e-05,
      "loss": 1.1288,
      "step": 5119
    },
    {
      "epoch": 0.13484329734000527,
      "grad_norm": 1.7440154552459717,
      "learning_rate": 4.326573610745325e-05,
      "loss": 1.9201,
      "step": 5120
    },
    {
      "epoch": 0.134869633921517,
      "grad_norm": 2.23038387298584,
      "learning_rate": 4.3264419278377674e-05,
      "loss": 1.0692,
      "step": 5121
    },
    {
      "epoch": 0.1348959705030287,
      "grad_norm": 1.9543406963348389,
      "learning_rate": 4.326310244930208e-05,
      "loss": 1.9731,
      "step": 5122
    },
    {
      "epoch": 0.13492230708454042,
      "grad_norm": 2.2074053287506104,
      "learning_rate": 4.32617856202265e-05,
      "loss": 2.8003,
      "step": 5123
    },
    {
      "epoch": 0.13494864366605214,
      "grad_norm": 1.8646095991134644,
      "learning_rate": 4.326046879115091e-05,
      "loss": 1.4831,
      "step": 5124
    },
    {
      "epoch": 0.13497498024756385,
      "grad_norm": 1.9908664226531982,
      "learning_rate": 4.325915196207532e-05,
      "loss": 1.6991,
      "step": 5125
    },
    {
      "epoch": 0.1350013168290756,
      "grad_norm": 4.11297607421875,
      "learning_rate": 4.325783513299974e-05,
      "loss": 0.7609,
      "step": 5126
    },
    {
      "epoch": 0.1350276534105873,
      "grad_norm": 5.092033386230469,
      "learning_rate": 4.3256518303924154e-05,
      "loss": 1.6576,
      "step": 5127
    },
    {
      "epoch": 0.13505398999209903,
      "grad_norm": 2.7819604873657227,
      "learning_rate": 4.325520147484857e-05,
      "loss": 1.633,
      "step": 5128
    },
    {
      "epoch": 0.13508032657361074,
      "grad_norm": 2.531390905380249,
      "learning_rate": 4.325388464577298e-05,
      "loss": 2.1205,
      "step": 5129
    },
    {
      "epoch": 0.13510666315512246,
      "grad_norm": 2.7674753665924072,
      "learning_rate": 4.3252567816697394e-05,
      "loss": 2.1588,
      "step": 5130
    },
    {
      "epoch": 0.13513299973663417,
      "grad_norm": 4.091637134552002,
      "learning_rate": 4.325125098762181e-05,
      "loss": 2.1967,
      "step": 5131
    },
    {
      "epoch": 0.13515933631814592,
      "grad_norm": 2.6674866676330566,
      "learning_rate": 4.3249934158546225e-05,
      "loss": 2.101,
      "step": 5132
    },
    {
      "epoch": 0.13518567289965763,
      "grad_norm": 2.726515769958496,
      "learning_rate": 4.3248617329470634e-05,
      "loss": 0.5051,
      "step": 5133
    },
    {
      "epoch": 0.13521200948116935,
      "grad_norm": 2.311643362045288,
      "learning_rate": 4.324730050039505e-05,
      "loss": 1.9457,
      "step": 5134
    },
    {
      "epoch": 0.13523834606268106,
      "grad_norm": 3.0122528076171875,
      "learning_rate": 4.3245983671319465e-05,
      "loss": 0.5019,
      "step": 5135
    },
    {
      "epoch": 0.13526468264419278,
      "grad_norm": 2.6952977180480957,
      "learning_rate": 4.324466684224388e-05,
      "loss": 2.2212,
      "step": 5136
    },
    {
      "epoch": 0.1352910192257045,
      "grad_norm": 4.774688243865967,
      "learning_rate": 4.3243350013168296e-05,
      "loss": 1.8913,
      "step": 5137
    },
    {
      "epoch": 0.1353173558072162,
      "grad_norm": 1.9841638803482056,
      "learning_rate": 4.3242033184092705e-05,
      "loss": 2.1792,
      "step": 5138
    },
    {
      "epoch": 0.13534369238872795,
      "grad_norm": 1.6687781810760498,
      "learning_rate": 4.324071635501712e-05,
      "loss": 1.8076,
      "step": 5139
    },
    {
      "epoch": 0.13537002897023967,
      "grad_norm": 3.0525543689727783,
      "learning_rate": 4.3239399525941536e-05,
      "loss": 2.4661,
      "step": 5140
    },
    {
      "epoch": 0.13539636555175139,
      "grad_norm": 1.989721417427063,
      "learning_rate": 4.323808269686595e-05,
      "loss": 1.7802,
      "step": 5141
    },
    {
      "epoch": 0.1354227021332631,
      "grad_norm": 1.4657495021820068,
      "learning_rate": 4.323676586779036e-05,
      "loss": 1.4623,
      "step": 5142
    },
    {
      "epoch": 0.13544903871477482,
      "grad_norm": 2.152785062789917,
      "learning_rate": 4.3235449038714776e-05,
      "loss": 1.9602,
      "step": 5143
    },
    {
      "epoch": 0.13547537529628653,
      "grad_norm": 4.588759422302246,
      "learning_rate": 4.323413220963919e-05,
      "loss": 0.5224,
      "step": 5144
    },
    {
      "epoch": 0.13550171187779828,
      "grad_norm": 2.155844211578369,
      "learning_rate": 4.323281538056361e-05,
      "loss": 2.0237,
      "step": 5145
    },
    {
      "epoch": 0.13552804845931,
      "grad_norm": 2.232266426086426,
      "learning_rate": 4.323149855148802e-05,
      "loss": 1.8489,
      "step": 5146
    },
    {
      "epoch": 0.1355543850408217,
      "grad_norm": 3.1980278491973877,
      "learning_rate": 4.323018172241243e-05,
      "loss": 1.0601,
      "step": 5147
    },
    {
      "epoch": 0.13558072162233342,
      "grad_norm": 2.158320426940918,
      "learning_rate": 4.322886489333685e-05,
      "loss": 1.9878,
      "step": 5148
    },
    {
      "epoch": 0.13560705820384514,
      "grad_norm": 2.39532208442688,
      "learning_rate": 4.3227548064261256e-05,
      "loss": 1.9243,
      "step": 5149
    },
    {
      "epoch": 0.13563339478535685,
      "grad_norm": 5.758508205413818,
      "learning_rate": 4.322623123518568e-05,
      "loss": 1.8366,
      "step": 5150
    },
    {
      "epoch": 0.13565973136686857,
      "grad_norm": 2.9325613975524902,
      "learning_rate": 4.322491440611009e-05,
      "loss": 0.7884,
      "step": 5151
    },
    {
      "epoch": 0.1356860679483803,
      "grad_norm": 2.801708459854126,
      "learning_rate": 4.32235975770345e-05,
      "loss": 0.9392,
      "step": 5152
    },
    {
      "epoch": 0.13571240452989203,
      "grad_norm": 3.685453414916992,
      "learning_rate": 4.322228074795892e-05,
      "loss": 2.3278,
      "step": 5153
    },
    {
      "epoch": 0.13573874111140374,
      "grad_norm": 2.65653920173645,
      "learning_rate": 4.3220963918883334e-05,
      "loss": 2.1537,
      "step": 5154
    },
    {
      "epoch": 0.13576507769291546,
      "grad_norm": 3.002680778503418,
      "learning_rate": 4.321964708980775e-05,
      "loss": 2.2661,
      "step": 5155
    },
    {
      "epoch": 0.13579141427442717,
      "grad_norm": 1.8394556045532227,
      "learning_rate": 4.321833026073216e-05,
      "loss": 2.2173,
      "step": 5156
    },
    {
      "epoch": 0.1358177508559389,
      "grad_norm": 2.311682939529419,
      "learning_rate": 4.3217013431656574e-05,
      "loss": 1.85,
      "step": 5157
    },
    {
      "epoch": 0.1358440874374506,
      "grad_norm": 2.7533247470855713,
      "learning_rate": 4.321569660258098e-05,
      "loss": 0.9464,
      "step": 5158
    },
    {
      "epoch": 0.13587042401896235,
      "grad_norm": 3.7418079376220703,
      "learning_rate": 4.3214379773505405e-05,
      "loss": 1.4527,
      "step": 5159
    },
    {
      "epoch": 0.13589676060047406,
      "grad_norm": 4.358343601226807,
      "learning_rate": 4.3213062944429814e-05,
      "loss": 1.3739,
      "step": 5160
    },
    {
      "epoch": 0.13592309718198578,
      "grad_norm": 4.877420902252197,
      "learning_rate": 4.321174611535423e-05,
      "loss": 2.4356,
      "step": 5161
    },
    {
      "epoch": 0.1359494337634975,
      "grad_norm": 4.410695552825928,
      "learning_rate": 4.3210429286278645e-05,
      "loss": 1.6822,
      "step": 5162
    },
    {
      "epoch": 0.1359757703450092,
      "grad_norm": 8.638838768005371,
      "learning_rate": 4.3209112457203054e-05,
      "loss": 2.0576,
      "step": 5163
    },
    {
      "epoch": 0.13600210692652093,
      "grad_norm": 2.762232780456543,
      "learning_rate": 4.3207795628127476e-05,
      "loss": 1.8403,
      "step": 5164
    },
    {
      "epoch": 0.13602844350803267,
      "grad_norm": 3.232184886932373,
      "learning_rate": 4.3206478799051885e-05,
      "loss": 1.6123,
      "step": 5165
    },
    {
      "epoch": 0.13605478008954439,
      "grad_norm": 3.2808170318603516,
      "learning_rate": 4.32051619699763e-05,
      "loss": 1.0584,
      "step": 5166
    },
    {
      "epoch": 0.1360811166710561,
      "grad_norm": 2.531832456588745,
      "learning_rate": 4.320384514090071e-05,
      "loss": 1.7505,
      "step": 5167
    },
    {
      "epoch": 0.13610745325256782,
      "grad_norm": 3.6374802589416504,
      "learning_rate": 4.3202528311825125e-05,
      "loss": 1.2648,
      "step": 5168
    },
    {
      "epoch": 0.13613378983407953,
      "grad_norm": 3.9847042560577393,
      "learning_rate": 4.320121148274954e-05,
      "loss": 2.2801,
      "step": 5169
    },
    {
      "epoch": 0.13616012641559125,
      "grad_norm": 1.6192246675491333,
      "learning_rate": 4.3199894653673957e-05,
      "loss": 2.1486,
      "step": 5170
    },
    {
      "epoch": 0.13618646299710296,
      "grad_norm": 1.9894036054611206,
      "learning_rate": 4.3198577824598365e-05,
      "loss": 1.7824,
      "step": 5171
    },
    {
      "epoch": 0.1362127995786147,
      "grad_norm": 2.728970766067505,
      "learning_rate": 4.319726099552278e-05,
      "loss": 1.5232,
      "step": 5172
    },
    {
      "epoch": 0.13623913616012642,
      "grad_norm": 1.9817637205123901,
      "learning_rate": 4.3195944166447197e-05,
      "loss": 2.2884,
      "step": 5173
    },
    {
      "epoch": 0.13626547274163814,
      "grad_norm": 1.8956005573272705,
      "learning_rate": 4.319462733737161e-05,
      "loss": 1.9107,
      "step": 5174
    },
    {
      "epoch": 0.13629180932314985,
      "grad_norm": 2.97184419631958,
      "learning_rate": 4.319331050829603e-05,
      "loss": 1.9211,
      "step": 5175
    },
    {
      "epoch": 0.13631814590466157,
      "grad_norm": 2.0654196739196777,
      "learning_rate": 4.3191993679220437e-05,
      "loss": 3.451,
      "step": 5176
    },
    {
      "epoch": 0.13634448248617329,
      "grad_norm": 2.0442097187042236,
      "learning_rate": 4.319067685014485e-05,
      "loss": 2.138,
      "step": 5177
    },
    {
      "epoch": 0.136370819067685,
      "grad_norm": 1.800565481185913,
      "learning_rate": 4.318936002106927e-05,
      "loss": 1.5426,
      "step": 5178
    },
    {
      "epoch": 0.13639715564919674,
      "grad_norm": 1.8814598321914673,
      "learning_rate": 4.318804319199368e-05,
      "loss": 1.6583,
      "step": 5179
    },
    {
      "epoch": 0.13642349223070846,
      "grad_norm": 2.1714470386505127,
      "learning_rate": 4.318672636291809e-05,
      "loss": 2.308,
      "step": 5180
    },
    {
      "epoch": 0.13644982881222018,
      "grad_norm": 2.5671043395996094,
      "learning_rate": 4.318540953384251e-05,
      "loss": 1.5029,
      "step": 5181
    },
    {
      "epoch": 0.1364761653937319,
      "grad_norm": 1.9010472297668457,
      "learning_rate": 4.318409270476692e-05,
      "loss": 1.458,
      "step": 5182
    },
    {
      "epoch": 0.1365025019752436,
      "grad_norm": 2.0927422046661377,
      "learning_rate": 4.318277587569134e-05,
      "loss": 2.095,
      "step": 5183
    },
    {
      "epoch": 0.13652883855675532,
      "grad_norm": 2.9164364337921143,
      "learning_rate": 4.3181459046615755e-05,
      "loss": 1.2047,
      "step": 5184
    },
    {
      "epoch": 0.13655517513826707,
      "grad_norm": 1.4114418029785156,
      "learning_rate": 4.318014221754016e-05,
      "loss": 1.9129,
      "step": 5185
    },
    {
      "epoch": 0.13658151171977878,
      "grad_norm": 2.114582061767578,
      "learning_rate": 4.317882538846458e-05,
      "loss": 1.8152,
      "step": 5186
    },
    {
      "epoch": 0.1366078483012905,
      "grad_norm": 2.897184133529663,
      "learning_rate": 4.3177508559388995e-05,
      "loss": 1.8791,
      "step": 5187
    },
    {
      "epoch": 0.1366341848828022,
      "grad_norm": 3.748249053955078,
      "learning_rate": 4.317619173031341e-05,
      "loss": 1.0289,
      "step": 5188
    },
    {
      "epoch": 0.13666052146431393,
      "grad_norm": 2.620304584503174,
      "learning_rate": 4.317487490123782e-05,
      "loss": 1.1702,
      "step": 5189
    },
    {
      "epoch": 0.13668685804582564,
      "grad_norm": 3.8103697299957275,
      "learning_rate": 4.3173558072162235e-05,
      "loss": 1.7618,
      "step": 5190
    },
    {
      "epoch": 0.13671319462733736,
      "grad_norm": 4.340106964111328,
      "learning_rate": 4.317224124308665e-05,
      "loss": 1.4328,
      "step": 5191
    },
    {
      "epoch": 0.1367395312088491,
      "grad_norm": 2.6316447257995605,
      "learning_rate": 4.3170924414011066e-05,
      "loss": 0.9378,
      "step": 5192
    },
    {
      "epoch": 0.13676586779036082,
      "grad_norm": 2.8905019760131836,
      "learning_rate": 4.316960758493548e-05,
      "loss": 1.5811,
      "step": 5193
    },
    {
      "epoch": 0.13679220437187253,
      "grad_norm": 3.1205954551696777,
      "learning_rate": 4.316829075585989e-05,
      "loss": 2.4785,
      "step": 5194
    },
    {
      "epoch": 0.13681854095338425,
      "grad_norm": 4.276710510253906,
      "learning_rate": 4.3166973926784306e-05,
      "loss": 1.0512,
      "step": 5195
    },
    {
      "epoch": 0.13684487753489596,
      "grad_norm": 1.6796799898147583,
      "learning_rate": 4.3165657097708715e-05,
      "loss": 1.81,
      "step": 5196
    },
    {
      "epoch": 0.13687121411640768,
      "grad_norm": 2.3783931732177734,
      "learning_rate": 4.316434026863314e-05,
      "loss": 1.8936,
      "step": 5197
    },
    {
      "epoch": 0.13689755069791942,
      "grad_norm": 1.7245712280273438,
      "learning_rate": 4.3163023439557546e-05,
      "loss": 1.8404,
      "step": 5198
    },
    {
      "epoch": 0.13692388727943114,
      "grad_norm": 2.432018518447876,
      "learning_rate": 4.316170661048196e-05,
      "loss": 1.7911,
      "step": 5199
    },
    {
      "epoch": 0.13695022386094285,
      "grad_norm": 3.3141655921936035,
      "learning_rate": 4.316038978140638e-05,
      "loss": 0.9153,
      "step": 5200
    },
    {
      "epoch": 0.13697656044245457,
      "grad_norm": 3.472564697265625,
      "learning_rate": 4.3159072952330786e-05,
      "loss": 1.5215,
      "step": 5201
    },
    {
      "epoch": 0.13700289702396629,
      "grad_norm": 2.152303695678711,
      "learning_rate": 4.315775612325521e-05,
      "loss": 1.695,
      "step": 5202
    },
    {
      "epoch": 0.137029233605478,
      "grad_norm": 4.102975845336914,
      "learning_rate": 4.315643929417962e-05,
      "loss": 2.8923,
      "step": 5203
    },
    {
      "epoch": 0.13705557018698972,
      "grad_norm": 1.769879937171936,
      "learning_rate": 4.315512246510403e-05,
      "loss": 2.0335,
      "step": 5204
    },
    {
      "epoch": 0.13708190676850146,
      "grad_norm": 2.633875608444214,
      "learning_rate": 4.315380563602844e-05,
      "loss": 2.0759,
      "step": 5205
    },
    {
      "epoch": 0.13710824335001318,
      "grad_norm": 1.999447226524353,
      "learning_rate": 4.3152488806952864e-05,
      "loss": 1.6448,
      "step": 5206
    },
    {
      "epoch": 0.1371345799315249,
      "grad_norm": 3.656423568725586,
      "learning_rate": 4.315117197787727e-05,
      "loss": 1.7968,
      "step": 5207
    },
    {
      "epoch": 0.1371609165130366,
      "grad_norm": 1.698837161064148,
      "learning_rate": 4.314985514880169e-05,
      "loss": 1.7172,
      "step": 5208
    },
    {
      "epoch": 0.13718725309454832,
      "grad_norm": 2.1964449882507324,
      "learning_rate": 4.3148538319726104e-05,
      "loss": 1.7485,
      "step": 5209
    },
    {
      "epoch": 0.13721358967606004,
      "grad_norm": 6.2827839851379395,
      "learning_rate": 4.314722149065051e-05,
      "loss": 1.9537,
      "step": 5210
    },
    {
      "epoch": 0.13723992625757175,
      "grad_norm": 1.9901573657989502,
      "learning_rate": 4.3145904661574935e-05,
      "loss": 1.5434,
      "step": 5211
    },
    {
      "epoch": 0.1372662628390835,
      "grad_norm": 2.386258125305176,
      "learning_rate": 4.3144587832499344e-05,
      "loss": 1.7594,
      "step": 5212
    },
    {
      "epoch": 0.1372925994205952,
      "grad_norm": 1.9436391592025757,
      "learning_rate": 4.314327100342376e-05,
      "loss": 1.5699,
      "step": 5213
    },
    {
      "epoch": 0.13731893600210693,
      "grad_norm": 1.8265070915222168,
      "learning_rate": 4.314195417434817e-05,
      "loss": 1.7264,
      "step": 5214
    },
    {
      "epoch": 0.13734527258361864,
      "grad_norm": 4.723089218139648,
      "learning_rate": 4.3140637345272584e-05,
      "loss": 2.0678,
      "step": 5215
    },
    {
      "epoch": 0.13737160916513036,
      "grad_norm": 2.4445502758026123,
      "learning_rate": 4.3139320516197e-05,
      "loss": 2.1188,
      "step": 5216
    },
    {
      "epoch": 0.13739794574664208,
      "grad_norm": 4.167450904846191,
      "learning_rate": 4.3138003687121415e-05,
      "loss": 1.5661,
      "step": 5217
    },
    {
      "epoch": 0.13742428232815382,
      "grad_norm": 3.330721616744995,
      "learning_rate": 4.3136686858045824e-05,
      "loss": 1.8171,
      "step": 5218
    },
    {
      "epoch": 0.13745061890966553,
      "grad_norm": 2.3990249633789062,
      "learning_rate": 4.313537002897024e-05,
      "loss": 1.7148,
      "step": 5219
    },
    {
      "epoch": 0.13747695549117725,
      "grad_norm": 2.406870126724243,
      "learning_rate": 4.3134053199894655e-05,
      "loss": 1.3883,
      "step": 5220
    },
    {
      "epoch": 0.13750329207268897,
      "grad_norm": 3.1100730895996094,
      "learning_rate": 4.313273637081907e-05,
      "loss": 1.6218,
      "step": 5221
    },
    {
      "epoch": 0.13752962865420068,
      "grad_norm": 2.6337523460388184,
      "learning_rate": 4.3131419541743486e-05,
      "loss": 2.2444,
      "step": 5222
    },
    {
      "epoch": 0.1375559652357124,
      "grad_norm": 1.8023779392242432,
      "learning_rate": 4.3130102712667895e-05,
      "loss": 1.7551,
      "step": 5223
    },
    {
      "epoch": 0.1375823018172241,
      "grad_norm": 2.707688570022583,
      "learning_rate": 4.312878588359231e-05,
      "loss": 1.931,
      "step": 5224
    },
    {
      "epoch": 0.13760863839873586,
      "grad_norm": 1.6479952335357666,
      "learning_rate": 4.3127469054516726e-05,
      "loss": 1.5038,
      "step": 5225
    },
    {
      "epoch": 0.13763497498024757,
      "grad_norm": 2.9928996562957764,
      "learning_rate": 4.312615222544114e-05,
      "loss": 1.4031,
      "step": 5226
    },
    {
      "epoch": 0.1376613115617593,
      "grad_norm": 2.356449842453003,
      "learning_rate": 4.312483539636555e-05,
      "loss": 1.7151,
      "step": 5227
    },
    {
      "epoch": 0.137687648143271,
      "grad_norm": 5.041073799133301,
      "learning_rate": 4.3123518567289966e-05,
      "loss": 0.9312,
      "step": 5228
    },
    {
      "epoch": 0.13771398472478272,
      "grad_norm": 3.6362922191619873,
      "learning_rate": 4.312220173821438e-05,
      "loss": 0.6874,
      "step": 5229
    },
    {
      "epoch": 0.13774032130629443,
      "grad_norm": 2.334411144256592,
      "learning_rate": 4.31208849091388e-05,
      "loss": 1.7671,
      "step": 5230
    },
    {
      "epoch": 0.13776665788780618,
      "grad_norm": 1.6797144412994385,
      "learning_rate": 4.311956808006321e-05,
      "loss": 1.4012,
      "step": 5231
    },
    {
      "epoch": 0.1377929944693179,
      "grad_norm": 4.525765895843506,
      "learning_rate": 4.311825125098762e-05,
      "loss": 0.6192,
      "step": 5232
    },
    {
      "epoch": 0.1378193310508296,
      "grad_norm": 3.5723278522491455,
      "learning_rate": 4.311693442191204e-05,
      "loss": 1.1591,
      "step": 5233
    },
    {
      "epoch": 0.13784566763234132,
      "grad_norm": 1.8095427751541138,
      "learning_rate": 4.3115617592836446e-05,
      "loss": 2.1363,
      "step": 5234
    },
    {
      "epoch": 0.13787200421385304,
      "grad_norm": 2.4696295261383057,
      "learning_rate": 4.311430076376087e-05,
      "loss": 1.5249,
      "step": 5235
    },
    {
      "epoch": 0.13789834079536475,
      "grad_norm": 1.9217630624771118,
      "learning_rate": 4.311298393468528e-05,
      "loss": 0.9953,
      "step": 5236
    },
    {
      "epoch": 0.13792467737687647,
      "grad_norm": 7.475039482116699,
      "learning_rate": 4.311166710560969e-05,
      "loss": 2.0335,
      "step": 5237
    },
    {
      "epoch": 0.1379510139583882,
      "grad_norm": 2.167692184448242,
      "learning_rate": 4.311035027653411e-05,
      "loss": 1.4621,
      "step": 5238
    },
    {
      "epoch": 0.13797735053989993,
      "grad_norm": 1.872039794921875,
      "learning_rate": 4.3109033447458524e-05,
      "loss": 1.7059,
      "step": 5239
    },
    {
      "epoch": 0.13800368712141164,
      "grad_norm": 1.7853387594223022,
      "learning_rate": 4.310771661838294e-05,
      "loss": 2.2233,
      "step": 5240
    },
    {
      "epoch": 0.13803002370292336,
      "grad_norm": 2.5420360565185547,
      "learning_rate": 4.310639978930735e-05,
      "loss": 1.4443,
      "step": 5241
    },
    {
      "epoch": 0.13805636028443508,
      "grad_norm": 2.502525806427002,
      "learning_rate": 4.3105082960231764e-05,
      "loss": 0.9432,
      "step": 5242
    },
    {
      "epoch": 0.1380826968659468,
      "grad_norm": 2.235863447189331,
      "learning_rate": 4.310376613115617e-05,
      "loss": 1.5561,
      "step": 5243
    },
    {
      "epoch": 0.1381090334474585,
      "grad_norm": 1.5832233428955078,
      "learning_rate": 4.3102449302080596e-05,
      "loss": 1.6918,
      "step": 5244
    },
    {
      "epoch": 0.13813537002897025,
      "grad_norm": 3.3127658367156982,
      "learning_rate": 4.3101132473005004e-05,
      "loss": 2.0622,
      "step": 5245
    },
    {
      "epoch": 0.13816170661048197,
      "grad_norm": 2.0305511951446533,
      "learning_rate": 4.309981564392942e-05,
      "loss": 2.0898,
      "step": 5246
    },
    {
      "epoch": 0.13818804319199368,
      "grad_norm": 1.9707815647125244,
      "learning_rate": 4.3098498814853836e-05,
      "loss": 2.4752,
      "step": 5247
    },
    {
      "epoch": 0.1382143797735054,
      "grad_norm": 2.595893144607544,
      "learning_rate": 4.3097181985778244e-05,
      "loss": 2.2197,
      "step": 5248
    },
    {
      "epoch": 0.1382407163550171,
      "grad_norm": 3.9535906314849854,
      "learning_rate": 4.309586515670267e-05,
      "loss": 0.9025,
      "step": 5249
    },
    {
      "epoch": 0.13826705293652883,
      "grad_norm": 8.726988792419434,
      "learning_rate": 4.3094548327627076e-05,
      "loss": 1.3627,
      "step": 5250
    },
    {
      "epoch": 0.13829338951804057,
      "grad_norm": 2.1553943157196045,
      "learning_rate": 4.309323149855149e-05,
      "loss": 1.641,
      "step": 5251
    },
    {
      "epoch": 0.1383197260995523,
      "grad_norm": 3.6073081493377686,
      "learning_rate": 4.30919146694759e-05,
      "loss": 1.3084,
      "step": 5252
    },
    {
      "epoch": 0.138346062681064,
      "grad_norm": 1.9121544361114502,
      "learning_rate": 4.309059784040032e-05,
      "loss": 1.9856,
      "step": 5253
    },
    {
      "epoch": 0.13837239926257572,
      "grad_norm": 2.1146183013916016,
      "learning_rate": 4.308928101132473e-05,
      "loss": 1.2974,
      "step": 5254
    },
    {
      "epoch": 0.13839873584408743,
      "grad_norm": 2.2476680278778076,
      "learning_rate": 4.308796418224915e-05,
      "loss": 1.7303,
      "step": 5255
    },
    {
      "epoch": 0.13842507242559915,
      "grad_norm": 1.6491483449935913,
      "learning_rate": 4.308664735317356e-05,
      "loss": 1.6044,
      "step": 5256
    },
    {
      "epoch": 0.13845140900711086,
      "grad_norm": 1.7803096771240234,
      "learning_rate": 4.308533052409797e-05,
      "loss": 1.233,
      "step": 5257
    },
    {
      "epoch": 0.1384777455886226,
      "grad_norm": 2.172182321548462,
      "learning_rate": 4.3084013695022394e-05,
      "loss": 2.0498,
      "step": 5258
    },
    {
      "epoch": 0.13850408217013432,
      "grad_norm": 2.2713263034820557,
      "learning_rate": 4.30826968659468e-05,
      "loss": 1.7635,
      "step": 5259
    },
    {
      "epoch": 0.13853041875164604,
      "grad_norm": 1.8058420419692993,
      "learning_rate": 4.308138003687122e-05,
      "loss": 0.3705,
      "step": 5260
    },
    {
      "epoch": 0.13855675533315776,
      "grad_norm": 2.0371696949005127,
      "learning_rate": 4.308006320779563e-05,
      "loss": 1.9953,
      "step": 5261
    },
    {
      "epoch": 0.13858309191466947,
      "grad_norm": 3.878199338912964,
      "learning_rate": 4.307874637872004e-05,
      "loss": 2.118,
      "step": 5262
    },
    {
      "epoch": 0.1386094284961812,
      "grad_norm": 1.545540452003479,
      "learning_rate": 4.307742954964446e-05,
      "loss": 1.3873,
      "step": 5263
    },
    {
      "epoch": 0.1386357650776929,
      "grad_norm": 2.0873782634735107,
      "learning_rate": 4.3076112720568874e-05,
      "loss": 1.6184,
      "step": 5264
    },
    {
      "epoch": 0.13866210165920465,
      "grad_norm": 2.4241604804992676,
      "learning_rate": 4.307479589149329e-05,
      "loss": 2.1249,
      "step": 5265
    },
    {
      "epoch": 0.13868843824071636,
      "grad_norm": 2.53702974319458,
      "learning_rate": 4.30734790624177e-05,
      "loss": 1.8551,
      "step": 5266
    },
    {
      "epoch": 0.13871477482222808,
      "grad_norm": 3.770756721496582,
      "learning_rate": 4.3072162233342114e-05,
      "loss": 1.0779,
      "step": 5267
    },
    {
      "epoch": 0.1387411114037398,
      "grad_norm": 2.1507341861724854,
      "learning_rate": 4.307084540426653e-05,
      "loss": 1.3775,
      "step": 5268
    },
    {
      "epoch": 0.1387674479852515,
      "grad_norm": 1.8591163158416748,
      "learning_rate": 4.3069528575190945e-05,
      "loss": 1.2577,
      "step": 5269
    },
    {
      "epoch": 0.13879378456676322,
      "grad_norm": 1.8454272747039795,
      "learning_rate": 4.3068211746115354e-05,
      "loss": 2.4208,
      "step": 5270
    },
    {
      "epoch": 0.13882012114827497,
      "grad_norm": 3.1865649223327637,
      "learning_rate": 4.306689491703977e-05,
      "loss": 1.126,
      "step": 5271
    },
    {
      "epoch": 0.13884645772978668,
      "grad_norm": 2.2785346508026123,
      "learning_rate": 4.3065578087964185e-05,
      "loss": 1.7934,
      "step": 5272
    },
    {
      "epoch": 0.1388727943112984,
      "grad_norm": 1.8782856464385986,
      "learning_rate": 4.30642612588886e-05,
      "loss": 1.7596,
      "step": 5273
    },
    {
      "epoch": 0.1388991308928101,
      "grad_norm": 2.8826820850372314,
      "learning_rate": 4.306294442981301e-05,
      "loss": 1.2076,
      "step": 5274
    },
    {
      "epoch": 0.13892546747432183,
      "grad_norm": 1.967508316040039,
      "learning_rate": 4.3061627600737425e-05,
      "loss": 1.2397,
      "step": 5275
    },
    {
      "epoch": 0.13895180405583354,
      "grad_norm": 2.5633678436279297,
      "learning_rate": 4.306031077166184e-05,
      "loss": 1.5113,
      "step": 5276
    },
    {
      "epoch": 0.13897814063734526,
      "grad_norm": 2.3238189220428467,
      "learning_rate": 4.3058993942586256e-05,
      "loss": 1.415,
      "step": 5277
    },
    {
      "epoch": 0.139004477218857,
      "grad_norm": 2.784306287765503,
      "learning_rate": 4.305767711351067e-05,
      "loss": 0.7378,
      "step": 5278
    },
    {
      "epoch": 0.13903081380036872,
      "grad_norm": 1.964415192604065,
      "learning_rate": 4.305636028443508e-05,
      "loss": 2.1115,
      "step": 5279
    },
    {
      "epoch": 0.13905715038188043,
      "grad_norm": 2.1276159286499023,
      "learning_rate": 4.3055043455359496e-05,
      "loss": 0.8108,
      "step": 5280
    },
    {
      "epoch": 0.13908348696339215,
      "grad_norm": 2.8581173419952393,
      "learning_rate": 4.3053726626283905e-05,
      "loss": 2.1912,
      "step": 5281
    },
    {
      "epoch": 0.13910982354490387,
      "grad_norm": 2.2042317390441895,
      "learning_rate": 4.305240979720833e-05,
      "loss": 2.4503,
      "step": 5282
    },
    {
      "epoch": 0.13913616012641558,
      "grad_norm": 1.9640851020812988,
      "learning_rate": 4.3051092968132736e-05,
      "loss": 1.9167,
      "step": 5283
    },
    {
      "epoch": 0.13916249670792732,
      "grad_norm": 1.6979402303695679,
      "learning_rate": 4.304977613905715e-05,
      "loss": 1.6881,
      "step": 5284
    },
    {
      "epoch": 0.13918883328943904,
      "grad_norm": 3.2987239360809326,
      "learning_rate": 4.304845930998157e-05,
      "loss": 1.9442,
      "step": 5285
    },
    {
      "epoch": 0.13921516987095076,
      "grad_norm": 2.034170150756836,
      "learning_rate": 4.304714248090598e-05,
      "loss": 1.5473,
      "step": 5286
    },
    {
      "epoch": 0.13924150645246247,
      "grad_norm": 1.9552794694900513,
      "learning_rate": 4.30458256518304e-05,
      "loss": 1.7752,
      "step": 5287
    },
    {
      "epoch": 0.1392678430339742,
      "grad_norm": 3.249704599380493,
      "learning_rate": 4.304450882275481e-05,
      "loss": 1.486,
      "step": 5288
    },
    {
      "epoch": 0.1392941796154859,
      "grad_norm": 2.3470027446746826,
      "learning_rate": 4.304319199367922e-05,
      "loss": 2.0688,
      "step": 5289
    },
    {
      "epoch": 0.13932051619699762,
      "grad_norm": 3.5709376335144043,
      "learning_rate": 4.304187516460363e-05,
      "loss": 2.2988,
      "step": 5290
    },
    {
      "epoch": 0.13934685277850936,
      "grad_norm": 1.8416324853897095,
      "learning_rate": 4.3040558335528054e-05,
      "loss": 1.2856,
      "step": 5291
    },
    {
      "epoch": 0.13937318936002108,
      "grad_norm": 1.8821313381195068,
      "learning_rate": 4.303924150645246e-05,
      "loss": 1.5477,
      "step": 5292
    },
    {
      "epoch": 0.1393995259415328,
      "grad_norm": 2.9171810150146484,
      "learning_rate": 4.303792467737688e-05,
      "loss": 1.3891,
      "step": 5293
    },
    {
      "epoch": 0.1394258625230445,
      "grad_norm": 3.3034346103668213,
      "learning_rate": 4.3036607848301294e-05,
      "loss": 0.7083,
      "step": 5294
    },
    {
      "epoch": 0.13945219910455622,
      "grad_norm": 1.651649832725525,
      "learning_rate": 4.30352910192257e-05,
      "loss": 1.6374,
      "step": 5295
    },
    {
      "epoch": 0.13947853568606794,
      "grad_norm": 2.5994820594787598,
      "learning_rate": 4.3033974190150125e-05,
      "loss": 1.5246,
      "step": 5296
    },
    {
      "epoch": 0.13950487226757965,
      "grad_norm": 2.1049609184265137,
      "learning_rate": 4.3032657361074534e-05,
      "loss": 2.0054,
      "step": 5297
    },
    {
      "epoch": 0.1395312088490914,
      "grad_norm": 1.8851418495178223,
      "learning_rate": 4.303134053199895e-05,
      "loss": 1.5025,
      "step": 5298
    },
    {
      "epoch": 0.1395575454306031,
      "grad_norm": 3.3094029426574707,
      "learning_rate": 4.303002370292336e-05,
      "loss": 1.6932,
      "step": 5299
    },
    {
      "epoch": 0.13958388201211483,
      "grad_norm": 2.852307081222534,
      "learning_rate": 4.3028706873847774e-05,
      "loss": 1.4242,
      "step": 5300
    },
    {
      "epoch": 0.13961021859362654,
      "grad_norm": 1.9864132404327393,
      "learning_rate": 4.302739004477219e-05,
      "loss": 1.6872,
      "step": 5301
    },
    {
      "epoch": 0.13963655517513826,
      "grad_norm": 1.6435332298278809,
      "learning_rate": 4.3026073215696605e-05,
      "loss": 1.5577,
      "step": 5302
    },
    {
      "epoch": 0.13966289175664998,
      "grad_norm": 4.0532026290893555,
      "learning_rate": 4.302475638662102e-05,
      "loss": 0.7639,
      "step": 5303
    },
    {
      "epoch": 0.13968922833816172,
      "grad_norm": 2.8549790382385254,
      "learning_rate": 4.302343955754543e-05,
      "loss": 2.1158,
      "step": 5304
    },
    {
      "epoch": 0.13971556491967344,
      "grad_norm": 2.293412208557129,
      "learning_rate": 4.302212272846985e-05,
      "loss": 1.6977,
      "step": 5305
    },
    {
      "epoch": 0.13974190150118515,
      "grad_norm": 1.766440510749817,
      "learning_rate": 4.302080589939426e-05,
      "loss": 1.617,
      "step": 5306
    },
    {
      "epoch": 0.13976823808269687,
      "grad_norm": 3.213672399520874,
      "learning_rate": 4.3019489070318677e-05,
      "loss": 1.5126,
      "step": 5307
    },
    {
      "epoch": 0.13979457466420858,
      "grad_norm": 3.596798896789551,
      "learning_rate": 4.3018172241243085e-05,
      "loss": 0.8919,
      "step": 5308
    },
    {
      "epoch": 0.1398209112457203,
      "grad_norm": 2.3942911624908447,
      "learning_rate": 4.30168554121675e-05,
      "loss": 0.3629,
      "step": 5309
    },
    {
      "epoch": 0.139847247827232,
      "grad_norm": 3.009469985961914,
      "learning_rate": 4.3015538583091917e-05,
      "loss": 1.8567,
      "step": 5310
    },
    {
      "epoch": 0.13987358440874376,
      "grad_norm": 1.6067448854446411,
      "learning_rate": 4.301422175401633e-05,
      "loss": 1.6066,
      "step": 5311
    },
    {
      "epoch": 0.13989992099025547,
      "grad_norm": 1.8433202505111694,
      "learning_rate": 4.301290492494075e-05,
      "loss": 1.7712,
      "step": 5312
    },
    {
      "epoch": 0.1399262575717672,
      "grad_norm": 2.316406011581421,
      "learning_rate": 4.3011588095865157e-05,
      "loss": 1.5872,
      "step": 5313
    },
    {
      "epoch": 0.1399525941532789,
      "grad_norm": 1.78685462474823,
      "learning_rate": 4.301027126678957e-05,
      "loss": 1.6376,
      "step": 5314
    },
    {
      "epoch": 0.13997893073479062,
      "grad_norm": 3.2384531497955322,
      "learning_rate": 4.300895443771399e-05,
      "loss": 1.6906,
      "step": 5315
    },
    {
      "epoch": 0.14000526731630233,
      "grad_norm": 2.1208434104919434,
      "learning_rate": 4.30076376086384e-05,
      "loss": 1.7709,
      "step": 5316
    },
    {
      "epoch": 0.14003160389781408,
      "grad_norm": 2.973423480987549,
      "learning_rate": 4.300632077956281e-05,
      "loss": 1.4483,
      "step": 5317
    },
    {
      "epoch": 0.1400579404793258,
      "grad_norm": 5.9371137619018555,
      "learning_rate": 4.300500395048723e-05,
      "loss": 1.6957,
      "step": 5318
    },
    {
      "epoch": 0.1400842770608375,
      "grad_norm": 2.1459813117980957,
      "learning_rate": 4.3003687121411643e-05,
      "loss": 2.0978,
      "step": 5319
    },
    {
      "epoch": 0.14011061364234922,
      "grad_norm": 2.6176798343658447,
      "learning_rate": 4.300237029233606e-05,
      "loss": 1.3801,
      "step": 5320
    },
    {
      "epoch": 0.14013695022386094,
      "grad_norm": 2.8769965171813965,
      "learning_rate": 4.300105346326047e-05,
      "loss": 1.9502,
      "step": 5321
    },
    {
      "epoch": 0.14016328680537266,
      "grad_norm": 2.8037261962890625,
      "learning_rate": 4.2999736634184883e-05,
      "loss": 2.1592,
      "step": 5322
    },
    {
      "epoch": 0.14018962338688437,
      "grad_norm": 3.456369161605835,
      "learning_rate": 4.29984198051093e-05,
      "loss": 1.4191,
      "step": 5323
    },
    {
      "epoch": 0.14021595996839611,
      "grad_norm": 1.7822257280349731,
      "learning_rate": 4.2997102976033715e-05,
      "loss": 2.4011,
      "step": 5324
    },
    {
      "epoch": 0.14024229654990783,
      "grad_norm": 2.4780359268188477,
      "learning_rate": 4.299578614695813e-05,
      "loss": 2.3557,
      "step": 5325
    },
    {
      "epoch": 0.14026863313141955,
      "grad_norm": 2.358567714691162,
      "learning_rate": 4.299446931788254e-05,
      "loss": 2.1387,
      "step": 5326
    },
    {
      "epoch": 0.14029496971293126,
      "grad_norm": 2.3652851581573486,
      "learning_rate": 4.2993152488806955e-05,
      "loss": 1.8933,
      "step": 5327
    },
    {
      "epoch": 0.14032130629444298,
      "grad_norm": 1.890342354774475,
      "learning_rate": 4.2991835659731363e-05,
      "loss": 1.6166,
      "step": 5328
    },
    {
      "epoch": 0.1403476428759547,
      "grad_norm": 1.642198920249939,
      "learning_rate": 4.2990518830655786e-05,
      "loss": 1.934,
      "step": 5329
    },
    {
      "epoch": 0.1403739794574664,
      "grad_norm": 2.969290018081665,
      "learning_rate": 4.2989202001580195e-05,
      "loss": 1.0556,
      "step": 5330
    },
    {
      "epoch": 0.14040031603897815,
      "grad_norm": 3.4994609355926514,
      "learning_rate": 4.298788517250461e-05,
      "loss": 1.791,
      "step": 5331
    },
    {
      "epoch": 0.14042665262048987,
      "grad_norm": 2.1280102729797363,
      "learning_rate": 4.2986568343429026e-05,
      "loss": 1.4283,
      "step": 5332
    },
    {
      "epoch": 0.14045298920200158,
      "grad_norm": 2.64162540435791,
      "learning_rate": 4.2985251514353435e-05,
      "loss": 1.8572,
      "step": 5333
    },
    {
      "epoch": 0.1404793257835133,
      "grad_norm": 2.2723042964935303,
      "learning_rate": 4.298393468527786e-05,
      "loss": 1.9276,
      "step": 5334
    },
    {
      "epoch": 0.140505662365025,
      "grad_norm": 2.2977218627929688,
      "learning_rate": 4.2982617856202266e-05,
      "loss": 0.5672,
      "step": 5335
    },
    {
      "epoch": 0.14053199894653673,
      "grad_norm": 2.1390562057495117,
      "learning_rate": 4.298130102712668e-05,
      "loss": 1.6938,
      "step": 5336
    },
    {
      "epoch": 0.14055833552804847,
      "grad_norm": 5.347156524658203,
      "learning_rate": 4.297998419805109e-05,
      "loss": 0.322,
      "step": 5337
    },
    {
      "epoch": 0.1405846721095602,
      "grad_norm": 4.87186861038208,
      "learning_rate": 4.297866736897551e-05,
      "loss": 2.3787,
      "step": 5338
    },
    {
      "epoch": 0.1406110086910719,
      "grad_norm": 1.8002957105636597,
      "learning_rate": 4.297735053989992e-05,
      "loss": 1.7799,
      "step": 5339
    },
    {
      "epoch": 0.14063734527258362,
      "grad_norm": 2.777805805206299,
      "learning_rate": 4.297603371082434e-05,
      "loss": 0.7216,
      "step": 5340
    },
    {
      "epoch": 0.14066368185409533,
      "grad_norm": 4.567877769470215,
      "learning_rate": 4.297471688174875e-05,
      "loss": 1.4741,
      "step": 5341
    },
    {
      "epoch": 0.14069001843560705,
      "grad_norm": 4.0375800132751465,
      "learning_rate": 4.297340005267316e-05,
      "loss": 2.5037,
      "step": 5342
    },
    {
      "epoch": 0.14071635501711877,
      "grad_norm": 3.46451735496521,
      "learning_rate": 4.2972083223597584e-05,
      "loss": 1.5835,
      "step": 5343
    },
    {
      "epoch": 0.1407426915986305,
      "grad_norm": 5.962404727935791,
      "learning_rate": 4.297076639452199e-05,
      "loss": 0.8314,
      "step": 5344
    },
    {
      "epoch": 0.14076902818014222,
      "grad_norm": 1.6963344812393188,
      "learning_rate": 4.296944956544641e-05,
      "loss": 2.2048,
      "step": 5345
    },
    {
      "epoch": 0.14079536476165394,
      "grad_norm": 2.7813546657562256,
      "learning_rate": 4.296813273637082e-05,
      "loss": 1.7117,
      "step": 5346
    },
    {
      "epoch": 0.14082170134316566,
      "grad_norm": 1.8465445041656494,
      "learning_rate": 4.296681590729523e-05,
      "loss": 2.1499,
      "step": 5347
    },
    {
      "epoch": 0.14084803792467737,
      "grad_norm": 1.759569764137268,
      "learning_rate": 4.296549907821965e-05,
      "loss": 2.4608,
      "step": 5348
    },
    {
      "epoch": 0.1408743745061891,
      "grad_norm": 3.9533181190490723,
      "learning_rate": 4.2964182249144064e-05,
      "loss": 0.9339,
      "step": 5349
    },
    {
      "epoch": 0.1409007110877008,
      "grad_norm": 2.5577051639556885,
      "learning_rate": 4.296286542006848e-05,
      "loss": 1.0319,
      "step": 5350
    },
    {
      "epoch": 0.14092704766921255,
      "grad_norm": 1.5836293697357178,
      "learning_rate": 4.296154859099289e-05,
      "loss": 1.7003,
      "step": 5351
    },
    {
      "epoch": 0.14095338425072426,
      "grad_norm": 3.7681639194488525,
      "learning_rate": 4.296023176191731e-05,
      "loss": 1.4407,
      "step": 5352
    },
    {
      "epoch": 0.14097972083223598,
      "grad_norm": 4.495939254760742,
      "learning_rate": 4.295891493284172e-05,
      "loss": 1.2109,
      "step": 5353
    },
    {
      "epoch": 0.1410060574137477,
      "grad_norm": 1.6568726301193237,
      "learning_rate": 4.2957598103766135e-05,
      "loss": 1.8267,
      "step": 5354
    },
    {
      "epoch": 0.1410323939952594,
      "grad_norm": 2.114642858505249,
      "learning_rate": 4.2956281274690544e-05,
      "loss": 2.169,
      "step": 5355
    },
    {
      "epoch": 0.14105873057677112,
      "grad_norm": 3.1813671588897705,
      "learning_rate": 4.295496444561496e-05,
      "loss": 2.3141,
      "step": 5356
    },
    {
      "epoch": 0.14108506715828287,
      "grad_norm": 2.931057929992676,
      "learning_rate": 4.2953647616539375e-05,
      "loss": 1.9705,
      "step": 5357
    },
    {
      "epoch": 0.14111140373979458,
      "grad_norm": 3.206650495529175,
      "learning_rate": 4.295233078746379e-05,
      "loss": 1.6772,
      "step": 5358
    },
    {
      "epoch": 0.1411377403213063,
      "grad_norm": 2.918741464614868,
      "learning_rate": 4.2951013958388206e-05,
      "loss": 1.8946,
      "step": 5359
    },
    {
      "epoch": 0.14116407690281801,
      "grad_norm": 1.9289441108703613,
      "learning_rate": 4.2949697129312615e-05,
      "loss": 2.2393,
      "step": 5360
    },
    {
      "epoch": 0.14119041348432973,
      "grad_norm": 3.684124708175659,
      "learning_rate": 4.294838030023703e-05,
      "loss": 1.6195,
      "step": 5361
    },
    {
      "epoch": 0.14121675006584145,
      "grad_norm": 2.0609798431396484,
      "learning_rate": 4.2947063471161446e-05,
      "loss": 2.0449,
      "step": 5362
    },
    {
      "epoch": 0.14124308664735316,
      "grad_norm": 1.9729681015014648,
      "learning_rate": 4.294574664208586e-05,
      "loss": 1.8299,
      "step": 5363
    },
    {
      "epoch": 0.1412694232288649,
      "grad_norm": 2.7491116523742676,
      "learning_rate": 4.294442981301027e-05,
      "loss": 0.3671,
      "step": 5364
    },
    {
      "epoch": 0.14129575981037662,
      "grad_norm": 3.1971828937530518,
      "learning_rate": 4.2943112983934686e-05,
      "loss": 2.3418,
      "step": 5365
    },
    {
      "epoch": 0.14132209639188834,
      "grad_norm": 2.176037311553955,
      "learning_rate": 4.29417961548591e-05,
      "loss": 1.4641,
      "step": 5366
    },
    {
      "epoch": 0.14134843297340005,
      "grad_norm": 2.338270425796509,
      "learning_rate": 4.294047932578352e-05,
      "loss": 2.6389,
      "step": 5367
    },
    {
      "epoch": 0.14137476955491177,
      "grad_norm": 1.7673707008361816,
      "learning_rate": 4.293916249670793e-05,
      "loss": 1.8622,
      "step": 5368
    },
    {
      "epoch": 0.14140110613642348,
      "grad_norm": 1.4852733612060547,
      "learning_rate": 4.293784566763234e-05,
      "loss": 1.9444,
      "step": 5369
    },
    {
      "epoch": 0.14142744271793523,
      "grad_norm": 6.920309066772461,
      "learning_rate": 4.293652883855676e-05,
      "loss": 1.3008,
      "step": 5370
    },
    {
      "epoch": 0.14145377929944694,
      "grad_norm": 2.7387514114379883,
      "learning_rate": 4.293521200948117e-05,
      "loss": 1.1729,
      "step": 5371
    },
    {
      "epoch": 0.14148011588095866,
      "grad_norm": 3.4094245433807373,
      "learning_rate": 4.293389518040559e-05,
      "loss": 1.4583,
      "step": 5372
    },
    {
      "epoch": 0.14150645246247037,
      "grad_norm": 1.66677725315094,
      "learning_rate": 4.293257835133e-05,
      "loss": 1.1794,
      "step": 5373
    },
    {
      "epoch": 0.1415327890439821,
      "grad_norm": 3.2916173934936523,
      "learning_rate": 4.293126152225441e-05,
      "loss": 1.5495,
      "step": 5374
    },
    {
      "epoch": 0.1415591256254938,
      "grad_norm": 3.422957181930542,
      "learning_rate": 4.292994469317882e-05,
      "loss": 0.886,
      "step": 5375
    },
    {
      "epoch": 0.14158546220700552,
      "grad_norm": 4.083506107330322,
      "learning_rate": 4.2928627864103244e-05,
      "loss": 1.4356,
      "step": 5376
    },
    {
      "epoch": 0.14161179878851726,
      "grad_norm": 3.9313323497772217,
      "learning_rate": 4.292731103502765e-05,
      "loss": 2.0779,
      "step": 5377
    },
    {
      "epoch": 0.14163813537002898,
      "grad_norm": 1.9234659671783447,
      "learning_rate": 4.292599420595207e-05,
      "loss": 1.7068,
      "step": 5378
    },
    {
      "epoch": 0.1416644719515407,
      "grad_norm": 2.0492684841156006,
      "learning_rate": 4.2924677376876484e-05,
      "loss": 1.626,
      "step": 5379
    },
    {
      "epoch": 0.1416908085330524,
      "grad_norm": 1.9004372358322144,
      "learning_rate": 4.292336054780089e-05,
      "loss": 1.7198,
      "step": 5380
    },
    {
      "epoch": 0.14171714511456412,
      "grad_norm": 3.774111032485962,
      "learning_rate": 4.2922043718725316e-05,
      "loss": 1.1441,
      "step": 5381
    },
    {
      "epoch": 0.14174348169607584,
      "grad_norm": 2.471282482147217,
      "learning_rate": 4.2920726889649724e-05,
      "loss": 1.725,
      "step": 5382
    },
    {
      "epoch": 0.14176981827758756,
      "grad_norm": 3.1594924926757812,
      "learning_rate": 4.291941006057414e-05,
      "loss": 0.8389,
      "step": 5383
    },
    {
      "epoch": 0.1417961548590993,
      "grad_norm": 2.274911403656006,
      "learning_rate": 4.291809323149855e-05,
      "loss": 2.0022,
      "step": 5384
    },
    {
      "epoch": 0.14182249144061101,
      "grad_norm": 5.938522815704346,
      "learning_rate": 4.291677640242297e-05,
      "loss": 2.0639,
      "step": 5385
    },
    {
      "epoch": 0.14184882802212273,
      "grad_norm": 6.627666473388672,
      "learning_rate": 4.291545957334738e-05,
      "loss": 1.1636,
      "step": 5386
    },
    {
      "epoch": 0.14187516460363445,
      "grad_norm": 2.3253440856933594,
      "learning_rate": 4.2914142744271796e-05,
      "loss": 2.3221,
      "step": 5387
    },
    {
      "epoch": 0.14190150118514616,
      "grad_norm": 2.9312222003936768,
      "learning_rate": 4.291282591519621e-05,
      "loss": 1.3796,
      "step": 5388
    },
    {
      "epoch": 0.14192783776665788,
      "grad_norm": 1.3603320121765137,
      "learning_rate": 4.291150908612062e-05,
      "loss": 1.706,
      "step": 5389
    },
    {
      "epoch": 0.14195417434816962,
      "grad_norm": 1.9491292238235474,
      "learning_rate": 4.291019225704504e-05,
      "loss": 1.3716,
      "step": 5390
    },
    {
      "epoch": 0.14198051092968134,
      "grad_norm": 1.877495288848877,
      "learning_rate": 4.290887542796945e-05,
      "loss": 2.1832,
      "step": 5391
    },
    {
      "epoch": 0.14200684751119305,
      "grad_norm": 1.968705415725708,
      "learning_rate": 4.290755859889387e-05,
      "loss": 1.839,
      "step": 5392
    },
    {
      "epoch": 0.14203318409270477,
      "grad_norm": 2.91729736328125,
      "learning_rate": 4.2906241769818276e-05,
      "loss": 1.793,
      "step": 5393
    },
    {
      "epoch": 0.14205952067421648,
      "grad_norm": 2.32688570022583,
      "learning_rate": 4.290492494074269e-05,
      "loss": 1.7722,
      "step": 5394
    },
    {
      "epoch": 0.1420858572557282,
      "grad_norm": 3.8894801139831543,
      "learning_rate": 4.290360811166711e-05,
      "loss": 1.7161,
      "step": 5395
    },
    {
      "epoch": 0.14211219383723991,
      "grad_norm": 3.7999372482299805,
      "learning_rate": 4.290229128259152e-05,
      "loss": 2.0999,
      "step": 5396
    },
    {
      "epoch": 0.14213853041875166,
      "grad_norm": 3.7862586975097656,
      "learning_rate": 4.290097445351594e-05,
      "loss": 0.7159,
      "step": 5397
    },
    {
      "epoch": 0.14216486700026337,
      "grad_norm": 1.9424610137939453,
      "learning_rate": 4.289965762444035e-05,
      "loss": 1.9509,
      "step": 5398
    },
    {
      "epoch": 0.1421912035817751,
      "grad_norm": 2.3114311695098877,
      "learning_rate": 4.289834079536477e-05,
      "loss": 1.5204,
      "step": 5399
    },
    {
      "epoch": 0.1422175401632868,
      "grad_norm": 3.0062406063079834,
      "learning_rate": 4.289702396628918e-05,
      "loss": 0.6657,
      "step": 5400
    },
    {
      "epoch": 0.14224387674479852,
      "grad_norm": 2.8383936882019043,
      "learning_rate": 4.2895707137213594e-05,
      "loss": 1.4026,
      "step": 5401
    },
    {
      "epoch": 0.14227021332631024,
      "grad_norm": 1.8685115575790405,
      "learning_rate": 4.2894390308138e-05,
      "loss": 1.484,
      "step": 5402
    },
    {
      "epoch": 0.14229654990782195,
      "grad_norm": 6.029784679412842,
      "learning_rate": 4.289307347906242e-05,
      "loss": 2.1724,
      "step": 5403
    },
    {
      "epoch": 0.1423228864893337,
      "grad_norm": 2.173064947128296,
      "learning_rate": 4.2891756649986834e-05,
      "loss": 1.3124,
      "step": 5404
    },
    {
      "epoch": 0.1423492230708454,
      "grad_norm": 1.7509270906448364,
      "learning_rate": 4.289043982091125e-05,
      "loss": 1.7561,
      "step": 5405
    },
    {
      "epoch": 0.14237555965235713,
      "grad_norm": 5.834602355957031,
      "learning_rate": 4.2889122991835665e-05,
      "loss": 1.899,
      "step": 5406
    },
    {
      "epoch": 0.14240189623386884,
      "grad_norm": 2.7962148189544678,
      "learning_rate": 4.2887806162760074e-05,
      "loss": 1.3563,
      "step": 5407
    },
    {
      "epoch": 0.14242823281538056,
      "grad_norm": 4.1680498123168945,
      "learning_rate": 4.288648933368449e-05,
      "loss": 1.2368,
      "step": 5408
    },
    {
      "epoch": 0.14245456939689227,
      "grad_norm": 1.9682034254074097,
      "learning_rate": 4.2885172504608905e-05,
      "loss": 2.9022,
      "step": 5409
    },
    {
      "epoch": 0.14248090597840402,
      "grad_norm": 1.7193231582641602,
      "learning_rate": 4.288385567553332e-05,
      "loss": 2.0055,
      "step": 5410
    },
    {
      "epoch": 0.14250724255991573,
      "grad_norm": 1.616080641746521,
      "learning_rate": 4.288253884645773e-05,
      "loss": 1.0752,
      "step": 5411
    },
    {
      "epoch": 0.14253357914142745,
      "grad_norm": 3.5982000827789307,
      "learning_rate": 4.2881222017382145e-05,
      "loss": 1.8522,
      "step": 5412
    },
    {
      "epoch": 0.14255991572293916,
      "grad_norm": 1.935608983039856,
      "learning_rate": 4.287990518830656e-05,
      "loss": 2.0533,
      "step": 5413
    },
    {
      "epoch": 0.14258625230445088,
      "grad_norm": 2.089115619659424,
      "learning_rate": 4.2878588359230976e-05,
      "loss": 1.4031,
      "step": 5414
    },
    {
      "epoch": 0.1426125888859626,
      "grad_norm": 1.8780349493026733,
      "learning_rate": 4.287727153015539e-05,
      "loss": 2.3138,
      "step": 5415
    },
    {
      "epoch": 0.1426389254674743,
      "grad_norm": 2.8866679668426514,
      "learning_rate": 4.28759547010798e-05,
      "loss": 2.4278,
      "step": 5416
    },
    {
      "epoch": 0.14266526204898605,
      "grad_norm": 2.3272926807403564,
      "learning_rate": 4.2874637872004216e-05,
      "loss": 0.3782,
      "step": 5417
    },
    {
      "epoch": 0.14269159863049777,
      "grad_norm": 1.9832936525344849,
      "learning_rate": 4.287332104292863e-05,
      "loss": 2.0203,
      "step": 5418
    },
    {
      "epoch": 0.14271793521200948,
      "grad_norm": 8.99078369140625,
      "learning_rate": 4.287200421385305e-05,
      "loss": 1.1857,
      "step": 5419
    },
    {
      "epoch": 0.1427442717935212,
      "grad_norm": 2.8711023330688477,
      "learning_rate": 4.2870687384777456e-05,
      "loss": 1.5664,
      "step": 5420
    },
    {
      "epoch": 0.14277060837503291,
      "grad_norm": 2.628261089324951,
      "learning_rate": 4.286937055570187e-05,
      "loss": 1.4518,
      "step": 5421
    },
    {
      "epoch": 0.14279694495654463,
      "grad_norm": 1.844959020614624,
      "learning_rate": 4.286805372662628e-05,
      "loss": 1.6122,
      "step": 5422
    },
    {
      "epoch": 0.14282328153805637,
      "grad_norm": 2.2086498737335205,
      "learning_rate": 4.28667368975507e-05,
      "loss": 0.5435,
      "step": 5423
    },
    {
      "epoch": 0.1428496181195681,
      "grad_norm": 2.238442897796631,
      "learning_rate": 4.286542006847511e-05,
      "loss": 1.8622,
      "step": 5424
    },
    {
      "epoch": 0.1428759547010798,
      "grad_norm": 2.8718161582946777,
      "learning_rate": 4.286410323939953e-05,
      "loss": 0.86,
      "step": 5425
    },
    {
      "epoch": 0.14290229128259152,
      "grad_norm": 2.1810548305511475,
      "learning_rate": 4.286278641032394e-05,
      "loss": 1.9892,
      "step": 5426
    },
    {
      "epoch": 0.14292862786410324,
      "grad_norm": 2.9680850505828857,
      "learning_rate": 4.286146958124835e-05,
      "loss": 1.5726,
      "step": 5427
    },
    {
      "epoch": 0.14295496444561495,
      "grad_norm": 2.748436450958252,
      "learning_rate": 4.2860152752172774e-05,
      "loss": 0.9523,
      "step": 5428
    },
    {
      "epoch": 0.14298130102712667,
      "grad_norm": 2.198032855987549,
      "learning_rate": 4.285883592309718e-05,
      "loss": 1.8121,
      "step": 5429
    },
    {
      "epoch": 0.1430076376086384,
      "grad_norm": 1.7660022974014282,
      "learning_rate": 4.28575190940216e-05,
      "loss": 1.6698,
      "step": 5430
    },
    {
      "epoch": 0.14303397419015013,
      "grad_norm": 1.730087399482727,
      "learning_rate": 4.285620226494601e-05,
      "loss": 1.7881,
      "step": 5431
    },
    {
      "epoch": 0.14306031077166184,
      "grad_norm": 2.480919599533081,
      "learning_rate": 4.285488543587043e-05,
      "loss": 1.8092,
      "step": 5432
    },
    {
      "epoch": 0.14308664735317356,
      "grad_norm": 1.8338545560836792,
      "learning_rate": 4.285356860679484e-05,
      "loss": 1.4528,
      "step": 5433
    },
    {
      "epoch": 0.14311298393468527,
      "grad_norm": 1.7347301244735718,
      "learning_rate": 4.2852251777719254e-05,
      "loss": 1.6911,
      "step": 5434
    },
    {
      "epoch": 0.143139320516197,
      "grad_norm": 4.22085428237915,
      "learning_rate": 4.285093494864367e-05,
      "loss": 1.1755,
      "step": 5435
    },
    {
      "epoch": 0.1431656570977087,
      "grad_norm": 2.8604445457458496,
      "learning_rate": 4.284961811956808e-05,
      "loss": 1.2737,
      "step": 5436
    },
    {
      "epoch": 0.14319199367922045,
      "grad_norm": 1.9961583614349365,
      "learning_rate": 4.28483012904925e-05,
      "loss": 1.8545,
      "step": 5437
    },
    {
      "epoch": 0.14321833026073216,
      "grad_norm": 4.361468315124512,
      "learning_rate": 4.284698446141691e-05,
      "loss": 2.0559,
      "step": 5438
    },
    {
      "epoch": 0.14324466684224388,
      "grad_norm": 3.103485584259033,
      "learning_rate": 4.2845667632341325e-05,
      "loss": 1.7961,
      "step": 5439
    },
    {
      "epoch": 0.1432710034237556,
      "grad_norm": 5.533717632293701,
      "learning_rate": 4.2844350803265734e-05,
      "loss": 1.6107,
      "step": 5440
    },
    {
      "epoch": 0.1432973400052673,
      "grad_norm": 2.6506903171539307,
      "learning_rate": 4.284303397419015e-05,
      "loss": 1.771,
      "step": 5441
    },
    {
      "epoch": 0.14332367658677903,
      "grad_norm": 2.8953254222869873,
      "learning_rate": 4.2841717145114565e-05,
      "loss": 1.1778,
      "step": 5442
    },
    {
      "epoch": 0.14335001316829077,
      "grad_norm": 1.9882150888442993,
      "learning_rate": 4.284040031603898e-05,
      "loss": 0.4579,
      "step": 5443
    },
    {
      "epoch": 0.14337634974980248,
      "grad_norm": 1.5002952814102173,
      "learning_rate": 4.2839083486963397e-05,
      "loss": 1.9492,
      "step": 5444
    },
    {
      "epoch": 0.1434026863313142,
      "grad_norm": 3.0559091567993164,
      "learning_rate": 4.2837766657887805e-05,
      "loss": 1.5274,
      "step": 5445
    },
    {
      "epoch": 0.14342902291282592,
      "grad_norm": 2.09110689163208,
      "learning_rate": 4.283644982881222e-05,
      "loss": 1.7752,
      "step": 5446
    },
    {
      "epoch": 0.14345535949433763,
      "grad_norm": 3.924508571624756,
      "learning_rate": 4.283513299973664e-05,
      "loss": 1.5881,
      "step": 5447
    },
    {
      "epoch": 0.14348169607584935,
      "grad_norm": 1.88231360912323,
      "learning_rate": 4.283381617066105e-05,
      "loss": 2.0659,
      "step": 5448
    },
    {
      "epoch": 0.14350803265736106,
      "grad_norm": 4.032523155212402,
      "learning_rate": 4.283249934158546e-05,
      "loss": 1.7669,
      "step": 5449
    },
    {
      "epoch": 0.1435343692388728,
      "grad_norm": 2.589891195297241,
      "learning_rate": 4.283118251250988e-05,
      "loss": 1.5581,
      "step": 5450
    },
    {
      "epoch": 0.14356070582038452,
      "grad_norm": 2.022139310836792,
      "learning_rate": 4.282986568343429e-05,
      "loss": 1.7378,
      "step": 5451
    },
    {
      "epoch": 0.14358704240189624,
      "grad_norm": 1.6292555332183838,
      "learning_rate": 4.282854885435871e-05,
      "loss": 2.0916,
      "step": 5452
    },
    {
      "epoch": 0.14361337898340795,
      "grad_norm": 2.1246378421783447,
      "learning_rate": 4.2827232025283123e-05,
      "loss": 1.8504,
      "step": 5453
    },
    {
      "epoch": 0.14363971556491967,
      "grad_norm": 5.700298309326172,
      "learning_rate": 4.282591519620753e-05,
      "loss": 1.1815,
      "step": 5454
    },
    {
      "epoch": 0.14366605214643138,
      "grad_norm": 2.0003139972686768,
      "learning_rate": 4.282459836713195e-05,
      "loss": 1.9312,
      "step": 5455
    },
    {
      "epoch": 0.14369238872794313,
      "grad_norm": 2.5048341751098633,
      "learning_rate": 4.2823281538056363e-05,
      "loss": 1.8645,
      "step": 5456
    },
    {
      "epoch": 0.14371872530945484,
      "grad_norm": 1.7863818407058716,
      "learning_rate": 4.282196470898078e-05,
      "loss": 1.9543,
      "step": 5457
    },
    {
      "epoch": 0.14374506189096656,
      "grad_norm": 3.901688814163208,
      "learning_rate": 4.282064787990519e-05,
      "loss": 1.1884,
      "step": 5458
    },
    {
      "epoch": 0.14377139847247827,
      "grad_norm": 1.991378664970398,
      "learning_rate": 4.2819331050829603e-05,
      "loss": 2.0037,
      "step": 5459
    },
    {
      "epoch": 0.14379773505399,
      "grad_norm": 3.321007013320923,
      "learning_rate": 4.281801422175402e-05,
      "loss": 1.3125,
      "step": 5460
    },
    {
      "epoch": 0.1438240716355017,
      "grad_norm": 2.6633682250976562,
      "learning_rate": 4.2816697392678435e-05,
      "loss": 1.9761,
      "step": 5461
    },
    {
      "epoch": 0.14385040821701342,
      "grad_norm": 1.621220350265503,
      "learning_rate": 4.281538056360285e-05,
      "loss": 0.544,
      "step": 5462
    },
    {
      "epoch": 0.14387674479852516,
      "grad_norm": 3.61556077003479,
      "learning_rate": 4.281406373452726e-05,
      "loss": 2.1541,
      "step": 5463
    },
    {
      "epoch": 0.14390308138003688,
      "grad_norm": 1.7097405195236206,
      "learning_rate": 4.2812746905451675e-05,
      "loss": 1.8777,
      "step": 5464
    },
    {
      "epoch": 0.1439294179615486,
      "grad_norm": 2.161938428878784,
      "learning_rate": 4.281143007637609e-05,
      "loss": 1.9369,
      "step": 5465
    },
    {
      "epoch": 0.1439557545430603,
      "grad_norm": 2.1637978553771973,
      "learning_rate": 4.2810113247300506e-05,
      "loss": 2.1154,
      "step": 5466
    },
    {
      "epoch": 0.14398209112457203,
      "grad_norm": 2.689159393310547,
      "learning_rate": 4.2808796418224915e-05,
      "loss": 1.481,
      "step": 5467
    },
    {
      "epoch": 0.14400842770608374,
      "grad_norm": 2.8505122661590576,
      "learning_rate": 4.280747958914933e-05,
      "loss": 1.4586,
      "step": 5468
    },
    {
      "epoch": 0.14403476428759546,
      "grad_norm": 4.715952396392822,
      "learning_rate": 4.280616276007374e-05,
      "loss": 2.2118,
      "step": 5469
    },
    {
      "epoch": 0.1440611008691072,
      "grad_norm": 5.930228233337402,
      "learning_rate": 4.280484593099816e-05,
      "loss": 1.4922,
      "step": 5470
    },
    {
      "epoch": 0.14408743745061892,
      "grad_norm": 1.986402153968811,
      "learning_rate": 4.280352910192258e-05,
      "loss": 1.3131,
      "step": 5471
    },
    {
      "epoch": 0.14411377403213063,
      "grad_norm": 1.7451683282852173,
      "learning_rate": 4.2802212272846986e-05,
      "loss": 2.1062,
      "step": 5472
    },
    {
      "epoch": 0.14414011061364235,
      "grad_norm": 2.088634729385376,
      "learning_rate": 4.28008954437714e-05,
      "loss": 1.433,
      "step": 5473
    },
    {
      "epoch": 0.14416644719515406,
      "grad_norm": 2.082184076309204,
      "learning_rate": 4.279957861469581e-05,
      "loss": 1.7962,
      "step": 5474
    },
    {
      "epoch": 0.14419278377666578,
      "grad_norm": 1.7730066776275635,
      "learning_rate": 4.279826178562023e-05,
      "loss": 2.0036,
      "step": 5475
    },
    {
      "epoch": 0.14421912035817752,
      "grad_norm": 1.8608057498931885,
      "learning_rate": 4.279694495654464e-05,
      "loss": 2.4359,
      "step": 5476
    },
    {
      "epoch": 0.14424545693968924,
      "grad_norm": 2.328056812286377,
      "learning_rate": 4.279562812746906e-05,
      "loss": 1.7591,
      "step": 5477
    },
    {
      "epoch": 0.14427179352120095,
      "grad_norm": 1.8919919729232788,
      "learning_rate": 4.2794311298393466e-05,
      "loss": 1.5027,
      "step": 5478
    },
    {
      "epoch": 0.14429813010271267,
      "grad_norm": 1.827974796295166,
      "learning_rate": 4.279299446931788e-05,
      "loss": 1.661,
      "step": 5479
    },
    {
      "epoch": 0.14432446668422438,
      "grad_norm": 2.0404622554779053,
      "learning_rate": 4.27916776402423e-05,
      "loss": 0.7236,
      "step": 5480
    },
    {
      "epoch": 0.1443508032657361,
      "grad_norm": 1.8391458988189697,
      "learning_rate": 4.279036081116671e-05,
      "loss": 1.7807,
      "step": 5481
    },
    {
      "epoch": 0.14437713984724782,
      "grad_norm": 2.4015066623687744,
      "learning_rate": 4.278904398209113e-05,
      "loss": 1.6419,
      "step": 5482
    },
    {
      "epoch": 0.14440347642875956,
      "grad_norm": 1.997928261756897,
      "learning_rate": 4.278772715301554e-05,
      "loss": 1.6739,
      "step": 5483
    },
    {
      "epoch": 0.14442981301027127,
      "grad_norm": 2.2104389667510986,
      "learning_rate": 4.278641032393996e-05,
      "loss": 1.9055,
      "step": 5484
    },
    {
      "epoch": 0.144456149591783,
      "grad_norm": 2.20745849609375,
      "learning_rate": 4.278509349486437e-05,
      "loss": 1.9459,
      "step": 5485
    },
    {
      "epoch": 0.1444824861732947,
      "grad_norm": 2.5828356742858887,
      "learning_rate": 4.2783776665788784e-05,
      "loss": 2.2961,
      "step": 5486
    },
    {
      "epoch": 0.14450882275480642,
      "grad_norm": 5.043327808380127,
      "learning_rate": 4.278245983671319e-05,
      "loss": 0.9504,
      "step": 5487
    },
    {
      "epoch": 0.14453515933631814,
      "grad_norm": 2.0591795444488525,
      "learning_rate": 4.278114300763761e-05,
      "loss": 1.4051,
      "step": 5488
    },
    {
      "epoch": 0.14456149591782985,
      "grad_norm": 1.72860586643219,
      "learning_rate": 4.2779826178562024e-05,
      "loss": 1.822,
      "step": 5489
    },
    {
      "epoch": 0.1445878324993416,
      "grad_norm": 2.47076153755188,
      "learning_rate": 4.277850934948644e-05,
      "loss": 1.6271,
      "step": 5490
    },
    {
      "epoch": 0.1446141690808533,
      "grad_norm": 3.8603389263153076,
      "learning_rate": 4.2777192520410855e-05,
      "loss": 2.1521,
      "step": 5491
    },
    {
      "epoch": 0.14464050566236503,
      "grad_norm": 4.7592267990112305,
      "learning_rate": 4.2775875691335264e-05,
      "loss": 1.4681,
      "step": 5492
    },
    {
      "epoch": 0.14466684224387674,
      "grad_norm": 1.618865966796875,
      "learning_rate": 4.277455886225968e-05,
      "loss": 1.3689,
      "step": 5493
    },
    {
      "epoch": 0.14469317882538846,
      "grad_norm": 1.8949971199035645,
      "learning_rate": 4.2773242033184095e-05,
      "loss": 1.7505,
      "step": 5494
    },
    {
      "epoch": 0.14471951540690017,
      "grad_norm": 1.722700834274292,
      "learning_rate": 4.277192520410851e-05,
      "loss": 1.4854,
      "step": 5495
    },
    {
      "epoch": 0.14474585198841192,
      "grad_norm": 2.876819610595703,
      "learning_rate": 4.277060837503292e-05,
      "loss": 2.2436,
      "step": 5496
    },
    {
      "epoch": 0.14477218856992363,
      "grad_norm": 2.3207411766052246,
      "learning_rate": 4.2769291545957335e-05,
      "loss": 1.6843,
      "step": 5497
    },
    {
      "epoch": 0.14479852515143535,
      "grad_norm": 4.414300918579102,
      "learning_rate": 4.276797471688175e-05,
      "loss": 1.5186,
      "step": 5498
    },
    {
      "epoch": 0.14482486173294706,
      "grad_norm": 1.7444515228271484,
      "learning_rate": 4.2766657887806166e-05,
      "loss": 1.8863,
      "step": 5499
    },
    {
      "epoch": 0.14485119831445878,
      "grad_norm": 1.6739468574523926,
      "learning_rate": 4.276534105873058e-05,
      "loss": 1.5695,
      "step": 5500
    },
    {
      "epoch": 0.1448775348959705,
      "grad_norm": 2.244536876678467,
      "learning_rate": 4.276402422965499e-05,
      "loss": 1.7561,
      "step": 5501
    },
    {
      "epoch": 0.1449038714774822,
      "grad_norm": 4.810878753662109,
      "learning_rate": 4.2762707400579406e-05,
      "loss": 1.4633,
      "step": 5502
    },
    {
      "epoch": 0.14493020805899395,
      "grad_norm": 3.660832643508911,
      "learning_rate": 4.276139057150382e-05,
      "loss": 0.5941,
      "step": 5503
    },
    {
      "epoch": 0.14495654464050567,
      "grad_norm": 2.0839686393737793,
      "learning_rate": 4.276007374242824e-05,
      "loss": 1.9839,
      "step": 5504
    },
    {
      "epoch": 0.14498288122201738,
      "grad_norm": 1.932313323020935,
      "learning_rate": 4.2758756913352646e-05,
      "loss": 1.5547,
      "step": 5505
    },
    {
      "epoch": 0.1450092178035291,
      "grad_norm": 3.07942795753479,
      "learning_rate": 4.275744008427706e-05,
      "loss": 1.8634,
      "step": 5506
    },
    {
      "epoch": 0.14503555438504082,
      "grad_norm": 2.2211496829986572,
      "learning_rate": 4.275612325520148e-05,
      "loss": 1.9627,
      "step": 5507
    },
    {
      "epoch": 0.14506189096655253,
      "grad_norm": 1.9119904041290283,
      "learning_rate": 4.275480642612589e-05,
      "loss": 0.308,
      "step": 5508
    },
    {
      "epoch": 0.14508822754806427,
      "grad_norm": 3.349985122680664,
      "learning_rate": 4.275348959705031e-05,
      "loss": 1.6698,
      "step": 5509
    },
    {
      "epoch": 0.145114564129576,
      "grad_norm": 1.9620261192321777,
      "learning_rate": 4.275217276797472e-05,
      "loss": 1.6017,
      "step": 5510
    },
    {
      "epoch": 0.1451409007110877,
      "grad_norm": 1.760911464691162,
      "learning_rate": 4.275085593889913e-05,
      "loss": 2.1849,
      "step": 5511
    },
    {
      "epoch": 0.14516723729259942,
      "grad_norm": 3.4776453971862793,
      "learning_rate": 4.274953910982354e-05,
      "loss": 0.8801,
      "step": 5512
    },
    {
      "epoch": 0.14519357387411114,
      "grad_norm": 2.4966847896575928,
      "learning_rate": 4.2748222280747964e-05,
      "loss": 1.7783,
      "step": 5513
    },
    {
      "epoch": 0.14521991045562285,
      "grad_norm": 1.7693718671798706,
      "learning_rate": 4.274690545167237e-05,
      "loss": 1.3738,
      "step": 5514
    },
    {
      "epoch": 0.14524624703713457,
      "grad_norm": 1.6176201105117798,
      "learning_rate": 4.274558862259679e-05,
      "loss": 1.4028,
      "step": 5515
    },
    {
      "epoch": 0.1452725836186463,
      "grad_norm": 3.2332048416137695,
      "learning_rate": 4.2744271793521204e-05,
      "loss": 1.9422,
      "step": 5516
    },
    {
      "epoch": 0.14529892020015803,
      "grad_norm": 2.100635051727295,
      "learning_rate": 4.274295496444562e-05,
      "loss": 1.7591,
      "step": 5517
    },
    {
      "epoch": 0.14532525678166974,
      "grad_norm": 1.867606520652771,
      "learning_rate": 4.2741638135370036e-05,
      "loss": 1.8669,
      "step": 5518
    },
    {
      "epoch": 0.14535159336318146,
      "grad_norm": 2.3108527660369873,
      "learning_rate": 4.2740321306294444e-05,
      "loss": 1.6326,
      "step": 5519
    },
    {
      "epoch": 0.14537792994469317,
      "grad_norm": 2.2782132625579834,
      "learning_rate": 4.273900447721886e-05,
      "loss": 2.0259,
      "step": 5520
    },
    {
      "epoch": 0.1454042665262049,
      "grad_norm": 2.0683164596557617,
      "learning_rate": 4.273768764814327e-05,
      "loss": 1.8712,
      "step": 5521
    },
    {
      "epoch": 0.1454306031077166,
      "grad_norm": 2.270616292953491,
      "learning_rate": 4.273637081906769e-05,
      "loss": 1.8014,
      "step": 5522
    },
    {
      "epoch": 0.14545693968922835,
      "grad_norm": 1.6381725072860718,
      "learning_rate": 4.27350539899921e-05,
      "loss": 1.6721,
      "step": 5523
    },
    {
      "epoch": 0.14548327627074006,
      "grad_norm": 2.6140260696411133,
      "learning_rate": 4.2733737160916516e-05,
      "loss": 1.869,
      "step": 5524
    },
    {
      "epoch": 0.14550961285225178,
      "grad_norm": 1.9675624370574951,
      "learning_rate": 4.2732420331840924e-05,
      "loss": 1.6683,
      "step": 5525
    },
    {
      "epoch": 0.1455359494337635,
      "grad_norm": 1.9817615747451782,
      "learning_rate": 4.273110350276534e-05,
      "loss": 1.7122,
      "step": 5526
    },
    {
      "epoch": 0.1455622860152752,
      "grad_norm": 2.7572903633117676,
      "learning_rate": 4.2729786673689756e-05,
      "loss": 1.972,
      "step": 5527
    },
    {
      "epoch": 0.14558862259678693,
      "grad_norm": 2.3718225955963135,
      "learning_rate": 4.272846984461417e-05,
      "loss": 1.8248,
      "step": 5528
    },
    {
      "epoch": 0.14561495917829867,
      "grad_norm": 3.4232852458953857,
      "learning_rate": 4.272715301553859e-05,
      "loss": 0.6497,
      "step": 5529
    },
    {
      "epoch": 0.14564129575981039,
      "grad_norm": 3.068422555923462,
      "learning_rate": 4.2725836186462996e-05,
      "loss": 1.3399,
      "step": 5530
    },
    {
      "epoch": 0.1456676323413221,
      "grad_norm": 2.2739012241363525,
      "learning_rate": 4.272451935738742e-05,
      "loss": 2.1571,
      "step": 5531
    },
    {
      "epoch": 0.14569396892283382,
      "grad_norm": 3.3147454261779785,
      "learning_rate": 4.272320252831183e-05,
      "loss": 1.0281,
      "step": 5532
    },
    {
      "epoch": 0.14572030550434553,
      "grad_norm": 2.3716354370117188,
      "learning_rate": 4.272188569923624e-05,
      "loss": 0.434,
      "step": 5533
    },
    {
      "epoch": 0.14574664208585725,
      "grad_norm": 2.1226859092712402,
      "learning_rate": 4.272056887016065e-05,
      "loss": 2.041,
      "step": 5534
    },
    {
      "epoch": 0.14577297866736896,
      "grad_norm": 2.0871102809906006,
      "learning_rate": 4.271925204108507e-05,
      "loss": 1.865,
      "step": 5535
    },
    {
      "epoch": 0.1457993152488807,
      "grad_norm": 1.696796178817749,
      "learning_rate": 4.271793521200948e-05,
      "loss": 1.5224,
      "step": 5536
    },
    {
      "epoch": 0.14582565183039242,
      "grad_norm": 1.9547193050384521,
      "learning_rate": 4.27166183829339e-05,
      "loss": 2.4481,
      "step": 5537
    },
    {
      "epoch": 0.14585198841190414,
      "grad_norm": 2.4103574752807617,
      "learning_rate": 4.2715301553858314e-05,
      "loss": 1.4425,
      "step": 5538
    },
    {
      "epoch": 0.14587832499341585,
      "grad_norm": 3.5615122318267822,
      "learning_rate": 4.271398472478272e-05,
      "loss": 1.5543,
      "step": 5539
    },
    {
      "epoch": 0.14590466157492757,
      "grad_norm": 1.5633794069290161,
      "learning_rate": 4.271266789570714e-05,
      "loss": 1.7641,
      "step": 5540
    },
    {
      "epoch": 0.14593099815643928,
      "grad_norm": 2.4548375606536865,
      "learning_rate": 4.2711351066631554e-05,
      "loss": 1.694,
      "step": 5541
    },
    {
      "epoch": 0.14595733473795103,
      "grad_norm": 2.1368513107299805,
      "learning_rate": 4.271003423755597e-05,
      "loss": 1.979,
      "step": 5542
    },
    {
      "epoch": 0.14598367131946274,
      "grad_norm": 1.925728678703308,
      "learning_rate": 4.270871740848038e-05,
      "loss": 1.9446,
      "step": 5543
    },
    {
      "epoch": 0.14601000790097446,
      "grad_norm": 1.9377928972244263,
      "learning_rate": 4.2707400579404794e-05,
      "loss": 1.7018,
      "step": 5544
    },
    {
      "epoch": 0.14603634448248617,
      "grad_norm": 3.370821475982666,
      "learning_rate": 4.270608375032921e-05,
      "loss": 1.4789,
      "step": 5545
    },
    {
      "epoch": 0.1460626810639979,
      "grad_norm": 4.439446449279785,
      "learning_rate": 4.2704766921253625e-05,
      "loss": 1.4931,
      "step": 5546
    },
    {
      "epoch": 0.1460890176455096,
      "grad_norm": 2.0447187423706055,
      "learning_rate": 4.270345009217804e-05,
      "loss": 2.1148,
      "step": 5547
    },
    {
      "epoch": 0.14611535422702132,
      "grad_norm": 1.9269649982452393,
      "learning_rate": 4.270213326310245e-05,
      "loss": 1.8261,
      "step": 5548
    },
    {
      "epoch": 0.14614169080853306,
      "grad_norm": 3.57525897026062,
      "learning_rate": 4.2700816434026865e-05,
      "loss": 1.8854,
      "step": 5549
    },
    {
      "epoch": 0.14616802739004478,
      "grad_norm": 1.9000951051712036,
      "learning_rate": 4.269949960495128e-05,
      "loss": 1.9626,
      "step": 5550
    },
    {
      "epoch": 0.1461943639715565,
      "grad_norm": 2.2235636711120605,
      "learning_rate": 4.2698182775875696e-05,
      "loss": 1.3551,
      "step": 5551
    },
    {
      "epoch": 0.1462207005530682,
      "grad_norm": 1.8338112831115723,
      "learning_rate": 4.2696865946800105e-05,
      "loss": 1.3472,
      "step": 5552
    },
    {
      "epoch": 0.14624703713457993,
      "grad_norm": 4.085555553436279,
      "learning_rate": 4.269554911772452e-05,
      "loss": 0.946,
      "step": 5553
    },
    {
      "epoch": 0.14627337371609164,
      "grad_norm": 2.6587250232696533,
      "learning_rate": 4.2694232288648936e-05,
      "loss": 1.5805,
      "step": 5554
    },
    {
      "epoch": 0.14629971029760336,
      "grad_norm": 2.6181790828704834,
      "learning_rate": 4.269291545957335e-05,
      "loss": 2.4685,
      "step": 5555
    },
    {
      "epoch": 0.1463260468791151,
      "grad_norm": 3.2557456493377686,
      "learning_rate": 4.269159863049777e-05,
      "loss": 0.7642,
      "step": 5556
    },
    {
      "epoch": 0.14635238346062682,
      "grad_norm": 2.2250311374664307,
      "learning_rate": 4.2690281801422176e-05,
      "loss": 1.9393,
      "step": 5557
    },
    {
      "epoch": 0.14637872004213853,
      "grad_norm": 2.177250623703003,
      "learning_rate": 4.268896497234659e-05,
      "loss": 1.6364,
      "step": 5558
    },
    {
      "epoch": 0.14640505662365025,
      "grad_norm": 2.7469115257263184,
      "learning_rate": 4.2687648143271e-05,
      "loss": 1.5623,
      "step": 5559
    },
    {
      "epoch": 0.14643139320516196,
      "grad_norm": 2.6492576599121094,
      "learning_rate": 4.268633131419542e-05,
      "loss": 1.3835,
      "step": 5560
    },
    {
      "epoch": 0.14645772978667368,
      "grad_norm": 1.836521863937378,
      "learning_rate": 4.268501448511983e-05,
      "loss": 2.1792,
      "step": 5561
    },
    {
      "epoch": 0.14648406636818542,
      "grad_norm": 4.0711750984191895,
      "learning_rate": 4.268369765604425e-05,
      "loss": 2.0186,
      "step": 5562
    },
    {
      "epoch": 0.14651040294969714,
      "grad_norm": 2.392573356628418,
      "learning_rate": 4.268238082696866e-05,
      "loss": 1.6863,
      "step": 5563
    },
    {
      "epoch": 0.14653673953120885,
      "grad_norm": 1.8146653175354004,
      "learning_rate": 4.268106399789308e-05,
      "loss": 1.0686,
      "step": 5564
    },
    {
      "epoch": 0.14656307611272057,
      "grad_norm": 2.204972267150879,
      "learning_rate": 4.2679747168817494e-05,
      "loss": 1.8641,
      "step": 5565
    },
    {
      "epoch": 0.14658941269423229,
      "grad_norm": 1.7449761629104614,
      "learning_rate": 4.26784303397419e-05,
      "loss": 0.4732,
      "step": 5566
    },
    {
      "epoch": 0.146615749275744,
      "grad_norm": 2.22798228263855,
      "learning_rate": 4.267711351066632e-05,
      "loss": 1.9302,
      "step": 5567
    },
    {
      "epoch": 0.14664208585725572,
      "grad_norm": 2.165184259414673,
      "learning_rate": 4.267579668159073e-05,
      "loss": 1.7797,
      "step": 5568
    },
    {
      "epoch": 0.14666842243876746,
      "grad_norm": 1.9731875658035278,
      "learning_rate": 4.267447985251515e-05,
      "loss": 1.5318,
      "step": 5569
    },
    {
      "epoch": 0.14669475902027918,
      "grad_norm": 1.9895118474960327,
      "learning_rate": 4.267316302343956e-05,
      "loss": 0.4572,
      "step": 5570
    },
    {
      "epoch": 0.1467210956017909,
      "grad_norm": 2.1495420932769775,
      "learning_rate": 4.2671846194363974e-05,
      "loss": 1.476,
      "step": 5571
    },
    {
      "epoch": 0.1467474321833026,
      "grad_norm": 2.0674545764923096,
      "learning_rate": 4.267052936528838e-05,
      "loss": 1.4671,
      "step": 5572
    },
    {
      "epoch": 0.14677376876481432,
      "grad_norm": 3.436493158340454,
      "learning_rate": 4.26692125362128e-05,
      "loss": 0.996,
      "step": 5573
    },
    {
      "epoch": 0.14680010534632604,
      "grad_norm": 2.3036608695983887,
      "learning_rate": 4.2667895707137214e-05,
      "loss": 2.6386,
      "step": 5574
    },
    {
      "epoch": 0.14682644192783775,
      "grad_norm": 2.138401508331299,
      "learning_rate": 4.266657887806163e-05,
      "loss": 1.8376,
      "step": 5575
    },
    {
      "epoch": 0.1468527785093495,
      "grad_norm": 5.032169818878174,
      "learning_rate": 4.2665262048986045e-05,
      "loss": 0.9696,
      "step": 5576
    },
    {
      "epoch": 0.1468791150908612,
      "grad_norm": 3.0435707569122314,
      "learning_rate": 4.2663945219910454e-05,
      "loss": 1.9079,
      "step": 5577
    },
    {
      "epoch": 0.14690545167237293,
      "grad_norm": 1.8727219104766846,
      "learning_rate": 4.266262839083487e-05,
      "loss": 2.1838,
      "step": 5578
    },
    {
      "epoch": 0.14693178825388464,
      "grad_norm": 3.9931259155273438,
      "learning_rate": 4.2661311561759285e-05,
      "loss": 1.3218,
      "step": 5579
    },
    {
      "epoch": 0.14695812483539636,
      "grad_norm": 2.8434035778045654,
      "learning_rate": 4.26599947326837e-05,
      "loss": 1.4108,
      "step": 5580
    },
    {
      "epoch": 0.14698446141690807,
      "grad_norm": 2.304365634918213,
      "learning_rate": 4.265867790360811e-05,
      "loss": 2.3515,
      "step": 5581
    },
    {
      "epoch": 0.14701079799841982,
      "grad_norm": 2.368215560913086,
      "learning_rate": 4.2657361074532525e-05,
      "loss": 1.785,
      "step": 5582
    },
    {
      "epoch": 0.14703713457993153,
      "grad_norm": 2.0873093605041504,
      "learning_rate": 4.265604424545694e-05,
      "loss": 1.5827,
      "step": 5583
    },
    {
      "epoch": 0.14706347116144325,
      "grad_norm": 5.458306312561035,
      "learning_rate": 4.265472741638136e-05,
      "loss": 1.1177,
      "step": 5584
    },
    {
      "epoch": 0.14708980774295496,
      "grad_norm": 1.6125884056091309,
      "learning_rate": 4.265341058730577e-05,
      "loss": 1.7194,
      "step": 5585
    },
    {
      "epoch": 0.14711614432446668,
      "grad_norm": 2.083245277404785,
      "learning_rate": 4.265209375823018e-05,
      "loss": 1.6941,
      "step": 5586
    },
    {
      "epoch": 0.1471424809059784,
      "grad_norm": 2.5999622344970703,
      "learning_rate": 4.26507769291546e-05,
      "loss": 1.2682,
      "step": 5587
    },
    {
      "epoch": 0.1471688174874901,
      "grad_norm": 2.612147808074951,
      "learning_rate": 4.264946010007901e-05,
      "loss": 0.5691,
      "step": 5588
    },
    {
      "epoch": 0.14719515406900185,
      "grad_norm": 1.8802284002304077,
      "learning_rate": 4.264814327100343e-05,
      "loss": 0.5906,
      "step": 5589
    },
    {
      "epoch": 0.14722149065051357,
      "grad_norm": 3.8250322341918945,
      "learning_rate": 4.264682644192784e-05,
      "loss": 0.6344,
      "step": 5590
    },
    {
      "epoch": 0.14724782723202529,
      "grad_norm": 2.7785391807556152,
      "learning_rate": 4.264550961285225e-05,
      "loss": 2.0483,
      "step": 5591
    },
    {
      "epoch": 0.147274163813537,
      "grad_norm": 1.826798915863037,
      "learning_rate": 4.264419278377667e-05,
      "loss": 1.8139,
      "step": 5592
    },
    {
      "epoch": 0.14730050039504872,
      "grad_norm": 3.454930543899536,
      "learning_rate": 4.2642875954701083e-05,
      "loss": 0.6545,
      "step": 5593
    },
    {
      "epoch": 0.14732683697656043,
      "grad_norm": 1.6765172481536865,
      "learning_rate": 4.26415591256255e-05,
      "loss": 2.0717,
      "step": 5594
    },
    {
      "epoch": 0.14735317355807218,
      "grad_norm": 2.713090419769287,
      "learning_rate": 4.264024229654991e-05,
      "loss": 2.0588,
      "step": 5595
    },
    {
      "epoch": 0.1473795101395839,
      "grad_norm": 2.3142645359039307,
      "learning_rate": 4.2638925467474324e-05,
      "loss": 1.4029,
      "step": 5596
    },
    {
      "epoch": 0.1474058467210956,
      "grad_norm": 2.2539315223693848,
      "learning_rate": 4.263760863839874e-05,
      "loss": 1.5575,
      "step": 5597
    },
    {
      "epoch": 0.14743218330260732,
      "grad_norm": 1.4994795322418213,
      "learning_rate": 4.2636291809323155e-05,
      "loss": 1.7899,
      "step": 5598
    },
    {
      "epoch": 0.14745851988411904,
      "grad_norm": 2.115285634994507,
      "learning_rate": 4.2634974980247564e-05,
      "loss": 1.8093,
      "step": 5599
    },
    {
      "epoch": 0.14748485646563075,
      "grad_norm": 1.8860194683074951,
      "learning_rate": 4.263365815117198e-05,
      "loss": 1.688,
      "step": 5600
    },
    {
      "epoch": 0.14751119304714247,
      "grad_norm": 3.8553032875061035,
      "learning_rate": 4.2632341322096395e-05,
      "loss": 1.9707,
      "step": 5601
    },
    {
      "epoch": 0.1475375296286542,
      "grad_norm": 2.2329020500183105,
      "learning_rate": 4.263102449302081e-05,
      "loss": 1.5304,
      "step": 5602
    },
    {
      "epoch": 0.14756386621016593,
      "grad_norm": 2.9448134899139404,
      "learning_rate": 4.2629707663945226e-05,
      "loss": 2.0549,
      "step": 5603
    },
    {
      "epoch": 0.14759020279167764,
      "grad_norm": 3.7261550426483154,
      "learning_rate": 4.2628390834869635e-05,
      "loss": 1.5402,
      "step": 5604
    },
    {
      "epoch": 0.14761653937318936,
      "grad_norm": 3.063939332962036,
      "learning_rate": 4.262707400579405e-05,
      "loss": 0.8647,
      "step": 5605
    },
    {
      "epoch": 0.14764287595470107,
      "grad_norm": 1.8527138233184814,
      "learning_rate": 4.262575717671846e-05,
      "loss": 1.7482,
      "step": 5606
    },
    {
      "epoch": 0.1476692125362128,
      "grad_norm": 2.4305081367492676,
      "learning_rate": 4.262444034764288e-05,
      "loss": 1.7115,
      "step": 5607
    },
    {
      "epoch": 0.1476955491177245,
      "grad_norm": 2.2378013134002686,
      "learning_rate": 4.262312351856729e-05,
      "loss": 1.7482,
      "step": 5608
    },
    {
      "epoch": 0.14772188569923625,
      "grad_norm": 2.779639959335327,
      "learning_rate": 4.2621806689491706e-05,
      "loss": 1.1633,
      "step": 5609
    },
    {
      "epoch": 0.14774822228074797,
      "grad_norm": 1.7249113321304321,
      "learning_rate": 4.262048986041612e-05,
      "loss": 1.781,
      "step": 5610
    },
    {
      "epoch": 0.14777455886225968,
      "grad_norm": 2.5951218605041504,
      "learning_rate": 4.261917303134053e-05,
      "loss": 1.9406,
      "step": 5611
    },
    {
      "epoch": 0.1478008954437714,
      "grad_norm": 1.778315782546997,
      "learning_rate": 4.261785620226495e-05,
      "loss": 1.6417,
      "step": 5612
    },
    {
      "epoch": 0.1478272320252831,
      "grad_norm": 4.979842185974121,
      "learning_rate": 4.261653937318936e-05,
      "loss": 1.2926,
      "step": 5613
    },
    {
      "epoch": 0.14785356860679483,
      "grad_norm": 2.3503050804138184,
      "learning_rate": 4.261522254411378e-05,
      "loss": 1.7667,
      "step": 5614
    },
    {
      "epoch": 0.14787990518830657,
      "grad_norm": 1.7864164113998413,
      "learning_rate": 4.2613905715038186e-05,
      "loss": 1.6159,
      "step": 5615
    },
    {
      "epoch": 0.1479062417698183,
      "grad_norm": 1.709181547164917,
      "learning_rate": 4.261258888596261e-05,
      "loss": 1.8267,
      "step": 5616
    },
    {
      "epoch": 0.14793257835133,
      "grad_norm": 3.542180299758911,
      "learning_rate": 4.261127205688702e-05,
      "loss": 0.6127,
      "step": 5617
    },
    {
      "epoch": 0.14795891493284172,
      "grad_norm": 3.4691050052642822,
      "learning_rate": 4.260995522781143e-05,
      "loss": 0.847,
      "step": 5618
    },
    {
      "epoch": 0.14798525151435343,
      "grad_norm": 1.9438574314117432,
      "learning_rate": 4.260863839873585e-05,
      "loss": 1.8687,
      "step": 5619
    },
    {
      "epoch": 0.14801158809586515,
      "grad_norm": 2.5891008377075195,
      "learning_rate": 4.260732156966026e-05,
      "loss": 0.6519,
      "step": 5620
    },
    {
      "epoch": 0.14803792467737686,
      "grad_norm": 4.490084171295166,
      "learning_rate": 4.260600474058468e-05,
      "loss": 0.9922,
      "step": 5621
    },
    {
      "epoch": 0.1480642612588886,
      "grad_norm": 2.307469129562378,
      "learning_rate": 4.260468791150909e-05,
      "loss": 1.4318,
      "step": 5622
    },
    {
      "epoch": 0.14809059784040032,
      "grad_norm": 1.982391595840454,
      "learning_rate": 4.2603371082433504e-05,
      "loss": 2.3119,
      "step": 5623
    },
    {
      "epoch": 0.14811693442191204,
      "grad_norm": 2.176826238632202,
      "learning_rate": 4.260205425335791e-05,
      "loss": 1.662,
      "step": 5624
    },
    {
      "epoch": 0.14814327100342375,
      "grad_norm": 1.7105432748794556,
      "learning_rate": 4.260073742428233e-05,
      "loss": 2.2901,
      "step": 5625
    },
    {
      "epoch": 0.14816960758493547,
      "grad_norm": 2.202418804168701,
      "learning_rate": 4.2599420595206744e-05,
      "loss": 1.379,
      "step": 5626
    },
    {
      "epoch": 0.14819594416644719,
      "grad_norm": 1.9317352771759033,
      "learning_rate": 4.259810376613116e-05,
      "loss": 1.8742,
      "step": 5627
    },
    {
      "epoch": 0.1482222807479589,
      "grad_norm": 3.7710983753204346,
      "learning_rate": 4.259678693705557e-05,
      "loss": 1.067,
      "step": 5628
    },
    {
      "epoch": 0.14824861732947064,
      "grad_norm": 1.6439965963363647,
      "learning_rate": 4.2595470107979984e-05,
      "loss": 1.6551,
      "step": 5629
    },
    {
      "epoch": 0.14827495391098236,
      "grad_norm": 1.9266676902770996,
      "learning_rate": 4.25941532789044e-05,
      "loss": 1.6155,
      "step": 5630
    },
    {
      "epoch": 0.14830129049249408,
      "grad_norm": 3.8426513671875,
      "learning_rate": 4.2592836449828815e-05,
      "loss": 1.3305,
      "step": 5631
    },
    {
      "epoch": 0.1483276270740058,
      "grad_norm": 3.590303421020508,
      "learning_rate": 4.259151962075323e-05,
      "loss": 1.5859,
      "step": 5632
    },
    {
      "epoch": 0.1483539636555175,
      "grad_norm": 1.9086166620254517,
      "learning_rate": 4.259020279167764e-05,
      "loss": 2.2934,
      "step": 5633
    },
    {
      "epoch": 0.14838030023702922,
      "grad_norm": 4.422410011291504,
      "learning_rate": 4.2588885962602055e-05,
      "loss": 0.9766,
      "step": 5634
    },
    {
      "epoch": 0.14840663681854097,
      "grad_norm": 2.8753767013549805,
      "learning_rate": 4.258756913352647e-05,
      "loss": 2.1532,
      "step": 5635
    },
    {
      "epoch": 0.14843297340005268,
      "grad_norm": 3.993929386138916,
      "learning_rate": 4.2586252304450886e-05,
      "loss": 1.3343,
      "step": 5636
    },
    {
      "epoch": 0.1484593099815644,
      "grad_norm": 3.0543391704559326,
      "learning_rate": 4.2584935475375295e-05,
      "loss": 1.6136,
      "step": 5637
    },
    {
      "epoch": 0.1484856465630761,
      "grad_norm": 3.9620182514190674,
      "learning_rate": 4.258361864629971e-05,
      "loss": 2.2358,
      "step": 5638
    },
    {
      "epoch": 0.14851198314458783,
      "grad_norm": 3.4023518562316895,
      "learning_rate": 4.2582301817224126e-05,
      "loss": 2.3857,
      "step": 5639
    },
    {
      "epoch": 0.14853831972609954,
      "grad_norm": 1.8487714529037476,
      "learning_rate": 4.258098498814854e-05,
      "loss": 1.9589,
      "step": 5640
    },
    {
      "epoch": 0.14856465630761126,
      "grad_norm": 2.166978120803833,
      "learning_rate": 4.257966815907296e-05,
      "loss": 1.8964,
      "step": 5641
    },
    {
      "epoch": 0.148590992889123,
      "grad_norm": 1.955157995223999,
      "learning_rate": 4.2578351329997366e-05,
      "loss": 2.2347,
      "step": 5642
    },
    {
      "epoch": 0.14861732947063472,
      "grad_norm": 2.189171314239502,
      "learning_rate": 4.257703450092178e-05,
      "loss": 1.1701,
      "step": 5643
    },
    {
      "epoch": 0.14864366605214643,
      "grad_norm": 3.317697048187256,
      "learning_rate": 4.257571767184619e-05,
      "loss": 1.9047,
      "step": 5644
    },
    {
      "epoch": 0.14867000263365815,
      "grad_norm": 5.230594158172607,
      "learning_rate": 4.257440084277061e-05,
      "loss": 1.1145,
      "step": 5645
    },
    {
      "epoch": 0.14869633921516986,
      "grad_norm": 2.1864380836486816,
      "learning_rate": 4.257308401369502e-05,
      "loss": 1.6484,
      "step": 5646
    },
    {
      "epoch": 0.14872267579668158,
      "grad_norm": 2.72421932220459,
      "learning_rate": 4.257176718461944e-05,
      "loss": 2.4499,
      "step": 5647
    },
    {
      "epoch": 0.14874901237819332,
      "grad_norm": 1.924412727355957,
      "learning_rate": 4.257045035554385e-05,
      "loss": 1.6235,
      "step": 5648
    },
    {
      "epoch": 0.14877534895970504,
      "grad_norm": 2.7526772022247314,
      "learning_rate": 4.256913352646827e-05,
      "loss": 0.8698,
      "step": 5649
    },
    {
      "epoch": 0.14880168554121675,
      "grad_norm": 2.680344581604004,
      "learning_rate": 4.2567816697392684e-05,
      "loss": 1.6177,
      "step": 5650
    },
    {
      "epoch": 0.14882802212272847,
      "grad_norm": 2.1998343467712402,
      "learning_rate": 4.256649986831709e-05,
      "loss": 1.1037,
      "step": 5651
    },
    {
      "epoch": 0.14885435870424019,
      "grad_norm": 4.720434188842773,
      "learning_rate": 4.256518303924151e-05,
      "loss": 1.4442,
      "step": 5652
    },
    {
      "epoch": 0.1488806952857519,
      "grad_norm": 2.881120204925537,
      "learning_rate": 4.256386621016592e-05,
      "loss": 1.6693,
      "step": 5653
    },
    {
      "epoch": 0.14890703186726362,
      "grad_norm": 2.2526192665100098,
      "learning_rate": 4.256254938109034e-05,
      "loss": 0.6066,
      "step": 5654
    },
    {
      "epoch": 0.14893336844877536,
      "grad_norm": 2.795752763748169,
      "learning_rate": 4.256123255201475e-05,
      "loss": 1.5708,
      "step": 5655
    },
    {
      "epoch": 0.14895970503028708,
      "grad_norm": 2.176715135574341,
      "learning_rate": 4.2559915722939165e-05,
      "loss": 1.5439,
      "step": 5656
    },
    {
      "epoch": 0.1489860416117988,
      "grad_norm": 3.603760004043579,
      "learning_rate": 4.255859889386358e-05,
      "loss": 1.6982,
      "step": 5657
    },
    {
      "epoch": 0.1490123781933105,
      "grad_norm": 3.7984838485717773,
      "learning_rate": 4.255728206478799e-05,
      "loss": 0.2778,
      "step": 5658
    },
    {
      "epoch": 0.14903871477482222,
      "grad_norm": 1.7825828790664673,
      "learning_rate": 4.255596523571241e-05,
      "loss": 1.7973,
      "step": 5659
    },
    {
      "epoch": 0.14906505135633394,
      "grad_norm": 2.003811836242676,
      "learning_rate": 4.255464840663682e-05,
      "loss": 1.6944,
      "step": 5660
    },
    {
      "epoch": 0.14909138793784565,
      "grad_norm": 2.4772143363952637,
      "learning_rate": 4.2553331577561236e-05,
      "loss": 1.3405,
      "step": 5661
    },
    {
      "epoch": 0.1491177245193574,
      "grad_norm": 2.669477939605713,
      "learning_rate": 4.2552014748485645e-05,
      "loss": 2.1803,
      "step": 5662
    },
    {
      "epoch": 0.1491440611008691,
      "grad_norm": 2.170348644256592,
      "learning_rate": 4.255069791941007e-05,
      "loss": 1.5437,
      "step": 5663
    },
    {
      "epoch": 0.14917039768238083,
      "grad_norm": 2.1788415908813477,
      "learning_rate": 4.2549381090334476e-05,
      "loss": 1.6009,
      "step": 5664
    },
    {
      "epoch": 0.14919673426389254,
      "grad_norm": 2.217836618423462,
      "learning_rate": 4.254806426125889e-05,
      "loss": 1.526,
      "step": 5665
    },
    {
      "epoch": 0.14922307084540426,
      "grad_norm": 4.839338302612305,
      "learning_rate": 4.254674743218331e-05,
      "loss": 1.0264,
      "step": 5666
    },
    {
      "epoch": 0.14924940742691598,
      "grad_norm": 3.5093142986297607,
      "learning_rate": 4.2545430603107716e-05,
      "loss": 2.0052,
      "step": 5667
    },
    {
      "epoch": 0.14927574400842772,
      "grad_norm": 2.4167253971099854,
      "learning_rate": 4.254411377403214e-05,
      "loss": 1.2477,
      "step": 5668
    },
    {
      "epoch": 0.14930208058993943,
      "grad_norm": 2.1677913665771484,
      "learning_rate": 4.254279694495655e-05,
      "loss": 1.8527,
      "step": 5669
    },
    {
      "epoch": 0.14932841717145115,
      "grad_norm": 4.827413558959961,
      "learning_rate": 4.254148011588096e-05,
      "loss": 0.8257,
      "step": 5670
    },
    {
      "epoch": 0.14935475375296287,
      "grad_norm": 2.6457746028900146,
      "learning_rate": 4.254016328680537e-05,
      "loss": 1.809,
      "step": 5671
    },
    {
      "epoch": 0.14938109033447458,
      "grad_norm": 1.7769595384597778,
      "learning_rate": 4.253884645772979e-05,
      "loss": 2.2455,
      "step": 5672
    },
    {
      "epoch": 0.1494074269159863,
      "grad_norm": 3.718106746673584,
      "learning_rate": 4.25375296286542e-05,
      "loss": 1.3036,
      "step": 5673
    },
    {
      "epoch": 0.149433763497498,
      "grad_norm": 2.424023151397705,
      "learning_rate": 4.253621279957862e-05,
      "loss": 1.1773,
      "step": 5674
    },
    {
      "epoch": 0.14946010007900976,
      "grad_norm": 2.3052265644073486,
      "learning_rate": 4.253489597050303e-05,
      "loss": 1.8745,
      "step": 5675
    },
    {
      "epoch": 0.14948643666052147,
      "grad_norm": 2.0570812225341797,
      "learning_rate": 4.253357914142744e-05,
      "loss": 1.6851,
      "step": 5676
    },
    {
      "epoch": 0.1495127732420332,
      "grad_norm": 4.507837295532227,
      "learning_rate": 4.253226231235186e-05,
      "loss": 1.3516,
      "step": 5677
    },
    {
      "epoch": 0.1495391098235449,
      "grad_norm": 2.8513917922973633,
      "learning_rate": 4.2530945483276274e-05,
      "loss": 1.4759,
      "step": 5678
    },
    {
      "epoch": 0.14956544640505662,
      "grad_norm": 2.387956380844116,
      "learning_rate": 4.252962865420069e-05,
      "loss": 1.612,
      "step": 5679
    },
    {
      "epoch": 0.14959178298656833,
      "grad_norm": 2.281764030456543,
      "learning_rate": 4.25283118251251e-05,
      "loss": 2.5507,
      "step": 5680
    },
    {
      "epoch": 0.14961811956808008,
      "grad_norm": 1.7326940298080444,
      "learning_rate": 4.2526994996049514e-05,
      "loss": 1.345,
      "step": 5681
    },
    {
      "epoch": 0.1496444561495918,
      "grad_norm": 2.286283016204834,
      "learning_rate": 4.252567816697393e-05,
      "loss": 2.2979,
      "step": 5682
    },
    {
      "epoch": 0.1496707927311035,
      "grad_norm": 1.9607174396514893,
      "learning_rate": 4.2524361337898345e-05,
      "loss": 0.7526,
      "step": 5683
    },
    {
      "epoch": 0.14969712931261522,
      "grad_norm": 2.015150785446167,
      "learning_rate": 4.2523044508822754e-05,
      "loss": 1.5515,
      "step": 5684
    },
    {
      "epoch": 0.14972346589412694,
      "grad_norm": 4.143985271453857,
      "learning_rate": 4.252172767974717e-05,
      "loss": 1.5143,
      "step": 5685
    },
    {
      "epoch": 0.14974980247563865,
      "grad_norm": 1.7674788236618042,
      "learning_rate": 4.2520410850671585e-05,
      "loss": 1.7692,
      "step": 5686
    },
    {
      "epoch": 0.14977613905715037,
      "grad_norm": 2.926802158355713,
      "learning_rate": 4.2519094021596e-05,
      "loss": 2.151,
      "step": 5687
    },
    {
      "epoch": 0.1498024756386621,
      "grad_norm": 1.8197352886199951,
      "learning_rate": 4.2517777192520416e-05,
      "loss": 1.7826,
      "step": 5688
    },
    {
      "epoch": 0.14982881222017383,
      "grad_norm": 2.624211072921753,
      "learning_rate": 4.2516460363444825e-05,
      "loss": 1.301,
      "step": 5689
    },
    {
      "epoch": 0.14985514880168554,
      "grad_norm": 1.795958399772644,
      "learning_rate": 4.251514353436924e-05,
      "loss": 2.1483,
      "step": 5690
    },
    {
      "epoch": 0.14988148538319726,
      "grad_norm": 2.796091079711914,
      "learning_rate": 4.251382670529365e-05,
      "loss": 2.1952,
      "step": 5691
    },
    {
      "epoch": 0.14990782196470898,
      "grad_norm": 1.5790303945541382,
      "learning_rate": 4.251250987621807e-05,
      "loss": 1.8803,
      "step": 5692
    },
    {
      "epoch": 0.1499341585462207,
      "grad_norm": 3.784172534942627,
      "learning_rate": 4.251119304714248e-05,
      "loss": 1.1298,
      "step": 5693
    },
    {
      "epoch": 0.1499604951277324,
      "grad_norm": 1.9059382677078247,
      "learning_rate": 4.2509876218066896e-05,
      "loss": 2.4398,
      "step": 5694
    },
    {
      "epoch": 0.14998683170924415,
      "grad_norm": 1.3986990451812744,
      "learning_rate": 4.250855938899131e-05,
      "loss": 0.3382,
      "step": 5695
    },
    {
      "epoch": 0.15001316829075587,
      "grad_norm": 3.4716920852661133,
      "learning_rate": 4.250724255991573e-05,
      "loss": 2.3116,
      "step": 5696
    },
    {
      "epoch": 0.15003950487226758,
      "grad_norm": 2.536043643951416,
      "learning_rate": 4.250592573084014e-05,
      "loss": 1.5416,
      "step": 5697
    },
    {
      "epoch": 0.1500658414537793,
      "grad_norm": 2.361250162124634,
      "learning_rate": 4.250460890176455e-05,
      "loss": 1.4363,
      "step": 5698
    },
    {
      "epoch": 0.150092178035291,
      "grad_norm": 2.0393974781036377,
      "learning_rate": 4.250329207268897e-05,
      "loss": 1.24,
      "step": 5699
    },
    {
      "epoch": 0.15011851461680273,
      "grad_norm": 1.827251672744751,
      "learning_rate": 4.2501975243613376e-05,
      "loss": 1.4273,
      "step": 5700
    },
    {
      "epoch": 0.15014485119831447,
      "grad_norm": 3.612252950668335,
      "learning_rate": 4.25006584145378e-05,
      "loss": 0.5793,
      "step": 5701
    },
    {
      "epoch": 0.1501711877798262,
      "grad_norm": 4.103225231170654,
      "learning_rate": 4.249934158546221e-05,
      "loss": 1.0051,
      "step": 5702
    },
    {
      "epoch": 0.1501975243613379,
      "grad_norm": 3.4101593494415283,
      "learning_rate": 4.249802475638662e-05,
      "loss": 1.4616,
      "step": 5703
    },
    {
      "epoch": 0.15022386094284962,
      "grad_norm": 2.2099180221557617,
      "learning_rate": 4.249670792731104e-05,
      "loss": 0.7844,
      "step": 5704
    },
    {
      "epoch": 0.15025019752436133,
      "grad_norm": 2.0153872966766357,
      "learning_rate": 4.249539109823545e-05,
      "loss": 1.3207,
      "step": 5705
    },
    {
      "epoch": 0.15027653410587305,
      "grad_norm": 2.031503677368164,
      "learning_rate": 4.249407426915987e-05,
      "loss": 1.8856,
      "step": 5706
    },
    {
      "epoch": 0.15030287068738477,
      "grad_norm": 3.351882219314575,
      "learning_rate": 4.249275744008428e-05,
      "loss": 2.2434,
      "step": 5707
    },
    {
      "epoch": 0.1503292072688965,
      "grad_norm": 5.673913955688477,
      "learning_rate": 4.2491440611008694e-05,
      "loss": 1.0375,
      "step": 5708
    },
    {
      "epoch": 0.15035554385040822,
      "grad_norm": 4.268165588378906,
      "learning_rate": 4.24901237819331e-05,
      "loss": 2.4776,
      "step": 5709
    },
    {
      "epoch": 0.15038188043191994,
      "grad_norm": 2.6325466632843018,
      "learning_rate": 4.2488806952857525e-05,
      "loss": 1.8972,
      "step": 5710
    },
    {
      "epoch": 0.15040821701343166,
      "grad_norm": 1.6593612432479858,
      "learning_rate": 4.2487490123781934e-05,
      "loss": 1.4751,
      "step": 5711
    },
    {
      "epoch": 0.15043455359494337,
      "grad_norm": 2.330878973007202,
      "learning_rate": 4.248617329470635e-05,
      "loss": 2.0391,
      "step": 5712
    },
    {
      "epoch": 0.1504608901764551,
      "grad_norm": 1.9831138849258423,
      "learning_rate": 4.2484856465630765e-05,
      "loss": 2.4672,
      "step": 5713
    },
    {
      "epoch": 0.1504872267579668,
      "grad_norm": 3.397312641143799,
      "learning_rate": 4.2483539636555174e-05,
      "loss": 0.7698,
      "step": 5714
    },
    {
      "epoch": 0.15051356333947855,
      "grad_norm": 1.577579379081726,
      "learning_rate": 4.24822228074796e-05,
      "loss": 1.7782,
      "step": 5715
    },
    {
      "epoch": 0.15053989992099026,
      "grad_norm": 1.7842483520507812,
      "learning_rate": 4.2480905978404006e-05,
      "loss": 1.9677,
      "step": 5716
    },
    {
      "epoch": 0.15056623650250198,
      "grad_norm": 3.7436392307281494,
      "learning_rate": 4.247958914932842e-05,
      "loss": 2.0013,
      "step": 5717
    },
    {
      "epoch": 0.1505925730840137,
      "grad_norm": 2.3510167598724365,
      "learning_rate": 4.247827232025283e-05,
      "loss": 1.6836,
      "step": 5718
    },
    {
      "epoch": 0.1506189096655254,
      "grad_norm": 3.6194794178009033,
      "learning_rate": 4.2476955491177246e-05,
      "loss": 0.7397,
      "step": 5719
    },
    {
      "epoch": 0.15064524624703712,
      "grad_norm": 1.8981231451034546,
      "learning_rate": 4.247563866210166e-05,
      "loss": 1.7762,
      "step": 5720
    },
    {
      "epoch": 0.15067158282854887,
      "grad_norm": 2.065647840499878,
      "learning_rate": 4.247432183302608e-05,
      "loss": 1.665,
      "step": 5721
    },
    {
      "epoch": 0.15069791941006058,
      "grad_norm": 2.4365437030792236,
      "learning_rate": 4.247300500395049e-05,
      "loss": 2.4467,
      "step": 5722
    },
    {
      "epoch": 0.1507242559915723,
      "grad_norm": 2.105393409729004,
      "learning_rate": 4.24716881748749e-05,
      "loss": 2.0182,
      "step": 5723
    },
    {
      "epoch": 0.150750592573084,
      "grad_norm": 2.433941125869751,
      "learning_rate": 4.247037134579932e-05,
      "loss": 1.8597,
      "step": 5724
    },
    {
      "epoch": 0.15077692915459573,
      "grad_norm": 2.0535686016082764,
      "learning_rate": 4.246905451672373e-05,
      "loss": 2.237,
      "step": 5725
    },
    {
      "epoch": 0.15080326573610744,
      "grad_norm": 3.9928605556488037,
      "learning_rate": 4.246773768764815e-05,
      "loss": 1.4777,
      "step": 5726
    },
    {
      "epoch": 0.15082960231761916,
      "grad_norm": 3.1019158363342285,
      "learning_rate": 4.246642085857256e-05,
      "loss": 1.9152,
      "step": 5727
    },
    {
      "epoch": 0.1508559388991309,
      "grad_norm": 1.9945987462997437,
      "learning_rate": 4.246510402949697e-05,
      "loss": 2.0096,
      "step": 5728
    },
    {
      "epoch": 0.15088227548064262,
      "grad_norm": 2.5533556938171387,
      "learning_rate": 4.246378720042139e-05,
      "loss": 2.1651,
      "step": 5729
    },
    {
      "epoch": 0.15090861206215433,
      "grad_norm": 5.138617515563965,
      "learning_rate": 4.2462470371345804e-05,
      "loss": 1.3718,
      "step": 5730
    },
    {
      "epoch": 0.15093494864366605,
      "grad_norm": 4.645598888397217,
      "learning_rate": 4.246115354227021e-05,
      "loss": 1.9161,
      "step": 5731
    },
    {
      "epoch": 0.15096128522517777,
      "grad_norm": 1.916338562965393,
      "learning_rate": 4.245983671319463e-05,
      "loss": 1.6664,
      "step": 5732
    },
    {
      "epoch": 0.15098762180668948,
      "grad_norm": 3.737555742263794,
      "learning_rate": 4.2458519884119044e-05,
      "loss": 0.9509,
      "step": 5733
    },
    {
      "epoch": 0.15101395838820122,
      "grad_norm": 3.6648004055023193,
      "learning_rate": 4.245720305504346e-05,
      "loss": 1.4428,
      "step": 5734
    },
    {
      "epoch": 0.15104029496971294,
      "grad_norm": 2.296168327331543,
      "learning_rate": 4.2455886225967875e-05,
      "loss": 1.603,
      "step": 5735
    },
    {
      "epoch": 0.15106663155122466,
      "grad_norm": 2.2506659030914307,
      "learning_rate": 4.2454569396892284e-05,
      "loss": 1.9465,
      "step": 5736
    },
    {
      "epoch": 0.15109296813273637,
      "grad_norm": 2.5359582901000977,
      "learning_rate": 4.24532525678167e-05,
      "loss": 2.2259,
      "step": 5737
    },
    {
      "epoch": 0.1511193047142481,
      "grad_norm": 3.7990384101867676,
      "learning_rate": 4.245193573874111e-05,
      "loss": 1.1931,
      "step": 5738
    },
    {
      "epoch": 0.1511456412957598,
      "grad_norm": 2.727841854095459,
      "learning_rate": 4.245061890966553e-05,
      "loss": 1.8222,
      "step": 5739
    },
    {
      "epoch": 0.15117197787727152,
      "grad_norm": 1.7299180030822754,
      "learning_rate": 4.244930208058994e-05,
      "loss": 1.5055,
      "step": 5740
    },
    {
      "epoch": 0.15119831445878326,
      "grad_norm": 5.285443305969238,
      "learning_rate": 4.2447985251514355e-05,
      "loss": 1.5848,
      "step": 5741
    },
    {
      "epoch": 0.15122465104029498,
      "grad_norm": 1.8614643812179565,
      "learning_rate": 4.244666842243877e-05,
      "loss": 1.8033,
      "step": 5742
    },
    {
      "epoch": 0.1512509876218067,
      "grad_norm": 2.1428117752075195,
      "learning_rate": 4.2445351593363186e-05,
      "loss": 2.1194,
      "step": 5743
    },
    {
      "epoch": 0.1512773242033184,
      "grad_norm": 5.770636081695557,
      "learning_rate": 4.24440347642876e-05,
      "loss": 1.7573,
      "step": 5744
    },
    {
      "epoch": 0.15130366078483012,
      "grad_norm": 3.3526344299316406,
      "learning_rate": 4.244271793521201e-05,
      "loss": 1.0669,
      "step": 5745
    },
    {
      "epoch": 0.15132999736634184,
      "grad_norm": 1.7971385717391968,
      "learning_rate": 4.2441401106136426e-05,
      "loss": 2.1406,
      "step": 5746
    },
    {
      "epoch": 0.15135633394785356,
      "grad_norm": 1.5810035467147827,
      "learning_rate": 4.2440084277060835e-05,
      "loss": 1.6191,
      "step": 5747
    },
    {
      "epoch": 0.1513826705293653,
      "grad_norm": 1.8192449808120728,
      "learning_rate": 4.243876744798526e-05,
      "loss": 1.4455,
      "step": 5748
    },
    {
      "epoch": 0.15140900711087701,
      "grad_norm": 2.4061219692230225,
      "learning_rate": 4.2437450618909666e-05,
      "loss": 2.3767,
      "step": 5749
    },
    {
      "epoch": 0.15143534369238873,
      "grad_norm": 2.6958067417144775,
      "learning_rate": 4.243613378983408e-05,
      "loss": 1.9432,
      "step": 5750
    },
    {
      "epoch": 0.15146168027390045,
      "grad_norm": 2.1837010383605957,
      "learning_rate": 4.24348169607585e-05,
      "loss": 0.4489,
      "step": 5751
    },
    {
      "epoch": 0.15148801685541216,
      "grad_norm": 1.946349859237671,
      "learning_rate": 4.2433500131682906e-05,
      "loss": 2.2124,
      "step": 5752
    },
    {
      "epoch": 0.15151435343692388,
      "grad_norm": 2.974581003189087,
      "learning_rate": 4.243218330260733e-05,
      "loss": 1.3092,
      "step": 5753
    },
    {
      "epoch": 0.15154069001843562,
      "grad_norm": 3.3074655532836914,
      "learning_rate": 4.243086647353174e-05,
      "loss": 1.5098,
      "step": 5754
    },
    {
      "epoch": 0.15156702659994734,
      "grad_norm": 2.089951276779175,
      "learning_rate": 4.242954964445615e-05,
      "loss": 1.58,
      "step": 5755
    },
    {
      "epoch": 0.15159336318145905,
      "grad_norm": 2.130981683731079,
      "learning_rate": 4.242823281538056e-05,
      "loss": 2.154,
      "step": 5756
    },
    {
      "epoch": 0.15161969976297077,
      "grad_norm": 2.550985097885132,
      "learning_rate": 4.242691598630498e-05,
      "loss": 1.6586,
      "step": 5757
    },
    {
      "epoch": 0.15164603634448248,
      "grad_norm": 2.7756130695343018,
      "learning_rate": 4.242559915722939e-05,
      "loss": 1.251,
      "step": 5758
    },
    {
      "epoch": 0.1516723729259942,
      "grad_norm": 3.435023784637451,
      "learning_rate": 4.242428232815381e-05,
      "loss": 1.1601,
      "step": 5759
    },
    {
      "epoch": 0.1516987095075059,
      "grad_norm": 2.8068125247955322,
      "learning_rate": 4.2422965499078224e-05,
      "loss": 1.3543,
      "step": 5760
    },
    {
      "epoch": 0.15172504608901766,
      "grad_norm": 3.4712586402893066,
      "learning_rate": 4.242164867000263e-05,
      "loss": 1.4659,
      "step": 5761
    },
    {
      "epoch": 0.15175138267052937,
      "grad_norm": 1.9302070140838623,
      "learning_rate": 4.2420331840927055e-05,
      "loss": 2.3592,
      "step": 5762
    },
    {
      "epoch": 0.1517777192520411,
      "grad_norm": 3.338070869445801,
      "learning_rate": 4.2419015011851464e-05,
      "loss": 1.0453,
      "step": 5763
    },
    {
      "epoch": 0.1518040558335528,
      "grad_norm": 4.471688270568848,
      "learning_rate": 4.241769818277588e-05,
      "loss": 1.6777,
      "step": 5764
    },
    {
      "epoch": 0.15183039241506452,
      "grad_norm": 3.090397357940674,
      "learning_rate": 4.241638135370029e-05,
      "loss": 0.4641,
      "step": 5765
    },
    {
      "epoch": 0.15185672899657623,
      "grad_norm": 2.181962013244629,
      "learning_rate": 4.2415064524624704e-05,
      "loss": 1.0882,
      "step": 5766
    },
    {
      "epoch": 0.15188306557808798,
      "grad_norm": 2.0378878116607666,
      "learning_rate": 4.241374769554912e-05,
      "loss": 0.949,
      "step": 5767
    },
    {
      "epoch": 0.1519094021595997,
      "grad_norm": 3.855985641479492,
      "learning_rate": 4.2412430866473535e-05,
      "loss": 1.6958,
      "step": 5768
    },
    {
      "epoch": 0.1519357387411114,
      "grad_norm": 2.005575180053711,
      "learning_rate": 4.241111403739795e-05,
      "loss": 1.6069,
      "step": 5769
    },
    {
      "epoch": 0.15196207532262312,
      "grad_norm": 1.6121034622192383,
      "learning_rate": 4.240979720832236e-05,
      "loss": 1.6984,
      "step": 5770
    },
    {
      "epoch": 0.15198841190413484,
      "grad_norm": 1.7583940029144287,
      "learning_rate": 4.2408480379246775e-05,
      "loss": 1.4442,
      "step": 5771
    },
    {
      "epoch": 0.15201474848564656,
      "grad_norm": 1.902397632598877,
      "learning_rate": 4.240716355017119e-05,
      "loss": 1.9341,
      "step": 5772
    },
    {
      "epoch": 0.15204108506715827,
      "grad_norm": 2.898103713989258,
      "learning_rate": 4.2405846721095606e-05,
      "loss": 1.8768,
      "step": 5773
    },
    {
      "epoch": 0.15206742164867001,
      "grad_norm": 1.9971503019332886,
      "learning_rate": 4.2404529892020015e-05,
      "loss": 1.7612,
      "step": 5774
    },
    {
      "epoch": 0.15209375823018173,
      "grad_norm": 4.194947242736816,
      "learning_rate": 4.240321306294443e-05,
      "loss": 1.7472,
      "step": 5775
    },
    {
      "epoch": 0.15212009481169345,
      "grad_norm": 3.3143115043640137,
      "learning_rate": 4.2401896233868846e-05,
      "loss": 0.9654,
      "step": 5776
    },
    {
      "epoch": 0.15214643139320516,
      "grad_norm": 1.7036153078079224,
      "learning_rate": 4.240057940479326e-05,
      "loss": 2.0887,
      "step": 5777
    },
    {
      "epoch": 0.15217276797471688,
      "grad_norm": 2.3918776512145996,
      "learning_rate": 4.239926257571767e-05,
      "loss": 1.7401,
      "step": 5778
    },
    {
      "epoch": 0.1521991045562286,
      "grad_norm": 2.058460235595703,
      "learning_rate": 4.2397945746642087e-05,
      "loss": 1.7126,
      "step": 5779
    },
    {
      "epoch": 0.1522254411377403,
      "grad_norm": 2.071458578109741,
      "learning_rate": 4.23966289175665e-05,
      "loss": 1.5062,
      "step": 5780
    },
    {
      "epoch": 0.15225177771925205,
      "grad_norm": 2.674434185028076,
      "learning_rate": 4.239531208849092e-05,
      "loss": 1.7909,
      "step": 5781
    },
    {
      "epoch": 0.15227811430076377,
      "grad_norm": 5.508162975311279,
      "learning_rate": 4.239399525941533e-05,
      "loss": 1.5076,
      "step": 5782
    },
    {
      "epoch": 0.15230445088227548,
      "grad_norm": 2.6319167613983154,
      "learning_rate": 4.239267843033974e-05,
      "loss": 2.1987,
      "step": 5783
    },
    {
      "epoch": 0.1523307874637872,
      "grad_norm": 1.5744264125823975,
      "learning_rate": 4.239136160126416e-05,
      "loss": 1.3576,
      "step": 5784
    },
    {
      "epoch": 0.15235712404529891,
      "grad_norm": 5.3464508056640625,
      "learning_rate": 4.2390044772188567e-05,
      "loss": 1.8021,
      "step": 5785
    },
    {
      "epoch": 0.15238346062681063,
      "grad_norm": 2.5581536293029785,
      "learning_rate": 4.238872794311299e-05,
      "loss": 2.1716,
      "step": 5786
    },
    {
      "epoch": 0.15240979720832237,
      "grad_norm": 6.438775539398193,
      "learning_rate": 4.23874111140374e-05,
      "loss": 1.9232,
      "step": 5787
    },
    {
      "epoch": 0.1524361337898341,
      "grad_norm": 3.27165150642395,
      "learning_rate": 4.238609428496181e-05,
      "loss": 1.9477,
      "step": 5788
    },
    {
      "epoch": 0.1524624703713458,
      "grad_norm": 2.4896976947784424,
      "learning_rate": 4.238477745588623e-05,
      "loss": 0.6221,
      "step": 5789
    },
    {
      "epoch": 0.15248880695285752,
      "grad_norm": 4.103713512420654,
      "learning_rate": 4.238346062681064e-05,
      "loss": 0.4915,
      "step": 5790
    },
    {
      "epoch": 0.15251514353436924,
      "grad_norm": 2.127819538116455,
      "learning_rate": 4.238214379773506e-05,
      "loss": 1.8545,
      "step": 5791
    },
    {
      "epoch": 0.15254148011588095,
      "grad_norm": 2.1432206630706787,
      "learning_rate": 4.238082696865947e-05,
      "loss": 0.5304,
      "step": 5792
    },
    {
      "epoch": 0.15256781669739267,
      "grad_norm": 2.7395143508911133,
      "learning_rate": 4.2379510139583885e-05,
      "loss": 0.702,
      "step": 5793
    },
    {
      "epoch": 0.1525941532789044,
      "grad_norm": 1.7573314905166626,
      "learning_rate": 4.237819331050829e-05,
      "loss": 1.4624,
      "step": 5794
    },
    {
      "epoch": 0.15262048986041613,
      "grad_norm": 1.7015575170516968,
      "learning_rate": 4.2376876481432716e-05,
      "loss": 1.2979,
      "step": 5795
    },
    {
      "epoch": 0.15264682644192784,
      "grad_norm": 4.185626983642578,
      "learning_rate": 4.2375559652357125e-05,
      "loss": 2.7003,
      "step": 5796
    },
    {
      "epoch": 0.15267316302343956,
      "grad_norm": 1.7594037055969238,
      "learning_rate": 4.237424282328154e-05,
      "loss": 1.8664,
      "step": 5797
    },
    {
      "epoch": 0.15269949960495127,
      "grad_norm": 2.6182191371917725,
      "learning_rate": 4.2372925994205956e-05,
      "loss": 1.7097,
      "step": 5798
    },
    {
      "epoch": 0.152725836186463,
      "grad_norm": 3.1217453479766846,
      "learning_rate": 4.2371609165130365e-05,
      "loss": 0.5211,
      "step": 5799
    },
    {
      "epoch": 0.1527521727679747,
      "grad_norm": 1.9852346181869507,
      "learning_rate": 4.237029233605479e-05,
      "loss": 1.6807,
      "step": 5800
    },
    {
      "epoch": 0.15277850934948645,
      "grad_norm": 1.7992138862609863,
      "learning_rate": 4.2368975506979196e-05,
      "loss": 1.5188,
      "step": 5801
    },
    {
      "epoch": 0.15280484593099816,
      "grad_norm": 3.129620313644409,
      "learning_rate": 4.236765867790361e-05,
      "loss": 1.5125,
      "step": 5802
    },
    {
      "epoch": 0.15283118251250988,
      "grad_norm": 2.4605212211608887,
      "learning_rate": 4.236634184882802e-05,
      "loss": 1.9487,
      "step": 5803
    },
    {
      "epoch": 0.1528575190940216,
      "grad_norm": 2.1186375617980957,
      "learning_rate": 4.2365025019752436e-05,
      "loss": 1.4671,
      "step": 5804
    },
    {
      "epoch": 0.1528838556755333,
      "grad_norm": 2.487518072128296,
      "learning_rate": 4.236370819067685e-05,
      "loss": 1.4194,
      "step": 5805
    },
    {
      "epoch": 0.15291019225704502,
      "grad_norm": 2.1010618209838867,
      "learning_rate": 4.236239136160127e-05,
      "loss": 2.1205,
      "step": 5806
    },
    {
      "epoch": 0.15293652883855677,
      "grad_norm": 3.364722728729248,
      "learning_rate": 4.236107453252568e-05,
      "loss": 1.8298,
      "step": 5807
    },
    {
      "epoch": 0.15296286542006848,
      "grad_norm": 1.6985399723052979,
      "learning_rate": 4.235975770345009e-05,
      "loss": 1.9882,
      "step": 5808
    },
    {
      "epoch": 0.1529892020015802,
      "grad_norm": 5.559665203094482,
      "learning_rate": 4.2358440874374514e-05,
      "loss": 0.9976,
      "step": 5809
    },
    {
      "epoch": 0.15301553858309191,
      "grad_norm": 3.8553013801574707,
      "learning_rate": 4.235712404529892e-05,
      "loss": 0.6162,
      "step": 5810
    },
    {
      "epoch": 0.15304187516460363,
      "grad_norm": 2.2669570446014404,
      "learning_rate": 4.235580721622334e-05,
      "loss": 1.9062,
      "step": 5811
    },
    {
      "epoch": 0.15306821174611535,
      "grad_norm": 3.381948232650757,
      "learning_rate": 4.235449038714775e-05,
      "loss": 2.244,
      "step": 5812
    },
    {
      "epoch": 0.15309454832762706,
      "grad_norm": 5.034691333770752,
      "learning_rate": 4.235317355807216e-05,
      "loss": 1.4453,
      "step": 5813
    },
    {
      "epoch": 0.1531208849091388,
      "grad_norm": 2.2089407444000244,
      "learning_rate": 4.235185672899658e-05,
      "loss": 1.9648,
      "step": 5814
    },
    {
      "epoch": 0.15314722149065052,
      "grad_norm": 1.9475656747817993,
      "learning_rate": 4.2350539899920994e-05,
      "loss": 1.8764,
      "step": 5815
    },
    {
      "epoch": 0.15317355807216224,
      "grad_norm": 1.8945266008377075,
      "learning_rate": 4.234922307084541e-05,
      "loss": 2.6819,
      "step": 5816
    },
    {
      "epoch": 0.15319989465367395,
      "grad_norm": 1.7970775365829468,
      "learning_rate": 4.234790624176982e-05,
      "loss": 0.5971,
      "step": 5817
    },
    {
      "epoch": 0.15322623123518567,
      "grad_norm": 1.876449704170227,
      "learning_rate": 4.2346589412694234e-05,
      "loss": 2.2377,
      "step": 5818
    },
    {
      "epoch": 0.15325256781669738,
      "grad_norm": 3.9584176540374756,
      "learning_rate": 4.234527258361865e-05,
      "loss": 2.1886,
      "step": 5819
    },
    {
      "epoch": 0.15327890439820913,
      "grad_norm": 2.0388834476470947,
      "learning_rate": 4.2343955754543065e-05,
      "loss": 1.7564,
      "step": 5820
    },
    {
      "epoch": 0.15330524097972084,
      "grad_norm": 1.9596577882766724,
      "learning_rate": 4.2342638925467474e-05,
      "loss": 2.2364,
      "step": 5821
    },
    {
      "epoch": 0.15333157756123256,
      "grad_norm": 5.078189373016357,
      "learning_rate": 4.234132209639189e-05,
      "loss": 1.8395,
      "step": 5822
    },
    {
      "epoch": 0.15335791414274427,
      "grad_norm": 2.767908811569214,
      "learning_rate": 4.2340005267316305e-05,
      "loss": 1.4972,
      "step": 5823
    },
    {
      "epoch": 0.153384250724256,
      "grad_norm": 3.0601449012756348,
      "learning_rate": 4.233868843824072e-05,
      "loss": 1.2673,
      "step": 5824
    },
    {
      "epoch": 0.1534105873057677,
      "grad_norm": 1.8390586376190186,
      "learning_rate": 4.2337371609165136e-05,
      "loss": 1.7768,
      "step": 5825
    },
    {
      "epoch": 0.15343692388727942,
      "grad_norm": 1.7578758001327515,
      "learning_rate": 4.2336054780089545e-05,
      "loss": 1.5339,
      "step": 5826
    },
    {
      "epoch": 0.15346326046879116,
      "grad_norm": 2.2950122356414795,
      "learning_rate": 4.233473795101396e-05,
      "loss": 2.3913,
      "step": 5827
    },
    {
      "epoch": 0.15348959705030288,
      "grad_norm": 2.3848166465759277,
      "learning_rate": 4.2333421121938376e-05,
      "loss": 2.1667,
      "step": 5828
    },
    {
      "epoch": 0.1535159336318146,
      "grad_norm": 2.779529333114624,
      "learning_rate": 4.233210429286279e-05,
      "loss": 1.5195,
      "step": 5829
    },
    {
      "epoch": 0.1535422702133263,
      "grad_norm": 2.484830379486084,
      "learning_rate": 4.23307874637872e-05,
      "loss": 1.6426,
      "step": 5830
    },
    {
      "epoch": 0.15356860679483803,
      "grad_norm": 4.905272483825684,
      "learning_rate": 4.2329470634711616e-05,
      "loss": 1.1128,
      "step": 5831
    },
    {
      "epoch": 0.15359494337634974,
      "grad_norm": 1.969044804573059,
      "learning_rate": 4.2328153805636025e-05,
      "loss": 1.4686,
      "step": 5832
    },
    {
      "epoch": 0.15362127995786146,
      "grad_norm": 2.164372205734253,
      "learning_rate": 4.232683697656045e-05,
      "loss": 1.7307,
      "step": 5833
    },
    {
      "epoch": 0.1536476165393732,
      "grad_norm": 2.4948770999908447,
      "learning_rate": 4.2325520147484856e-05,
      "loss": 0.5157,
      "step": 5834
    },
    {
      "epoch": 0.15367395312088492,
      "grad_norm": 4.046682834625244,
      "learning_rate": 4.232420331840927e-05,
      "loss": 1.154,
      "step": 5835
    },
    {
      "epoch": 0.15370028970239663,
      "grad_norm": 3.777179718017578,
      "learning_rate": 4.232288648933369e-05,
      "loss": 1.7118,
      "step": 5836
    },
    {
      "epoch": 0.15372662628390835,
      "grad_norm": 3.2744557857513428,
      "learning_rate": 4.2321569660258096e-05,
      "loss": 1.9194,
      "step": 5837
    },
    {
      "epoch": 0.15375296286542006,
      "grad_norm": 1.9500840902328491,
      "learning_rate": 4.232025283118252e-05,
      "loss": 2.2378,
      "step": 5838
    },
    {
      "epoch": 0.15377929944693178,
      "grad_norm": 2.1920278072357178,
      "learning_rate": 4.231893600210693e-05,
      "loss": 2.0352,
      "step": 5839
    },
    {
      "epoch": 0.15380563602844352,
      "grad_norm": 5.23524808883667,
      "learning_rate": 4.231761917303134e-05,
      "loss": 1.5955,
      "step": 5840
    },
    {
      "epoch": 0.15383197260995524,
      "grad_norm": 6.422600269317627,
      "learning_rate": 4.231630234395575e-05,
      "loss": 1.1283,
      "step": 5841
    },
    {
      "epoch": 0.15385830919146695,
      "grad_norm": 2.6680405139923096,
      "learning_rate": 4.2314985514880174e-05,
      "loss": 1.639,
      "step": 5842
    },
    {
      "epoch": 0.15388464577297867,
      "grad_norm": 1.9572499990463257,
      "learning_rate": 4.231366868580458e-05,
      "loss": 2.2494,
      "step": 5843
    },
    {
      "epoch": 0.15391098235449038,
      "grad_norm": 3.598973512649536,
      "learning_rate": 4.2312351856729e-05,
      "loss": 0.8718,
      "step": 5844
    },
    {
      "epoch": 0.1539373189360021,
      "grad_norm": 4.288572788238525,
      "learning_rate": 4.2311035027653414e-05,
      "loss": 1.7746,
      "step": 5845
    },
    {
      "epoch": 0.15396365551751381,
      "grad_norm": 2.566802501678467,
      "learning_rate": 4.230971819857782e-05,
      "loss": 1.4888,
      "step": 5846
    },
    {
      "epoch": 0.15398999209902556,
      "grad_norm": 2.754087209701538,
      "learning_rate": 4.2308401369502246e-05,
      "loss": 1.6778,
      "step": 5847
    },
    {
      "epoch": 0.15401632868053727,
      "grad_norm": 2.596709728240967,
      "learning_rate": 4.2307084540426654e-05,
      "loss": 1.6571,
      "step": 5848
    },
    {
      "epoch": 0.154042665262049,
      "grad_norm": 1.8524587154388428,
      "learning_rate": 4.230576771135107e-05,
      "loss": 1.7538,
      "step": 5849
    },
    {
      "epoch": 0.1540690018435607,
      "grad_norm": 2.5000038146972656,
      "learning_rate": 4.230445088227548e-05,
      "loss": 2.0407,
      "step": 5850
    },
    {
      "epoch": 0.15409533842507242,
      "grad_norm": 3.3258731365203857,
      "learning_rate": 4.2303134053199894e-05,
      "loss": 1.3404,
      "step": 5851
    },
    {
      "epoch": 0.15412167500658414,
      "grad_norm": 2.608553886413574,
      "learning_rate": 4.230181722412431e-05,
      "loss": 1.535,
      "step": 5852
    },
    {
      "epoch": 0.15414801158809588,
      "grad_norm": 2.5919270515441895,
      "learning_rate": 4.2300500395048726e-05,
      "loss": 0.9003,
      "step": 5853
    },
    {
      "epoch": 0.1541743481696076,
      "grad_norm": 2.164841413497925,
      "learning_rate": 4.229918356597314e-05,
      "loss": 1.7515,
      "step": 5854
    },
    {
      "epoch": 0.1542006847511193,
      "grad_norm": 8.342192649841309,
      "learning_rate": 4.229786673689755e-05,
      "loss": 1.4762,
      "step": 5855
    },
    {
      "epoch": 0.15422702133263103,
      "grad_norm": 2.4404876232147217,
      "learning_rate": 4.2296549907821966e-05,
      "loss": 1.7751,
      "step": 5856
    },
    {
      "epoch": 0.15425335791414274,
      "grad_norm": 1.9116249084472656,
      "learning_rate": 4.229523307874638e-05,
      "loss": 2.2532,
      "step": 5857
    },
    {
      "epoch": 0.15427969449565446,
      "grad_norm": 1.9052069187164307,
      "learning_rate": 4.22939162496708e-05,
      "loss": 1.7236,
      "step": 5858
    },
    {
      "epoch": 0.15430603107716617,
      "grad_norm": 2.3022842407226562,
      "learning_rate": 4.2292599420595206e-05,
      "loss": 2.0284,
      "step": 5859
    },
    {
      "epoch": 0.15433236765867792,
      "grad_norm": 2.4093234539031982,
      "learning_rate": 4.229128259151962e-05,
      "loss": 2.0325,
      "step": 5860
    },
    {
      "epoch": 0.15435870424018963,
      "grad_norm": 4.183977127075195,
      "learning_rate": 4.228996576244404e-05,
      "loss": 1.6332,
      "step": 5861
    },
    {
      "epoch": 0.15438504082170135,
      "grad_norm": 2.769796371459961,
      "learning_rate": 4.228864893336845e-05,
      "loss": 0.5197,
      "step": 5862
    },
    {
      "epoch": 0.15441137740321306,
      "grad_norm": 3.7631874084472656,
      "learning_rate": 4.228733210429287e-05,
      "loss": 1.8485,
      "step": 5863
    },
    {
      "epoch": 0.15443771398472478,
      "grad_norm": 2.9936611652374268,
      "learning_rate": 4.228601527521728e-05,
      "loss": 1.0529,
      "step": 5864
    },
    {
      "epoch": 0.1544640505662365,
      "grad_norm": 1.7187716960906982,
      "learning_rate": 4.228469844614169e-05,
      "loss": 2.0852,
      "step": 5865
    },
    {
      "epoch": 0.1544903871477482,
      "grad_norm": 2.605057716369629,
      "learning_rate": 4.228338161706611e-05,
      "loss": 1.7557,
      "step": 5866
    },
    {
      "epoch": 0.15451672372925995,
      "grad_norm": 2.624366521835327,
      "learning_rate": 4.2282064787990524e-05,
      "loss": 1.6579,
      "step": 5867
    },
    {
      "epoch": 0.15454306031077167,
      "grad_norm": 2.2019760608673096,
      "learning_rate": 4.228074795891493e-05,
      "loss": 1.8905,
      "step": 5868
    },
    {
      "epoch": 0.15456939689228338,
      "grad_norm": 3.4228620529174805,
      "learning_rate": 4.227943112983935e-05,
      "loss": 0.8433,
      "step": 5869
    },
    {
      "epoch": 0.1545957334737951,
      "grad_norm": 2.1239914894104004,
      "learning_rate": 4.2278114300763764e-05,
      "loss": 1.8914,
      "step": 5870
    },
    {
      "epoch": 0.15462207005530682,
      "grad_norm": 2.1944406032562256,
      "learning_rate": 4.227679747168818e-05,
      "loss": 2.1417,
      "step": 5871
    },
    {
      "epoch": 0.15464840663681853,
      "grad_norm": 1.8497647047042847,
      "learning_rate": 4.2275480642612595e-05,
      "loss": 1.9463,
      "step": 5872
    },
    {
      "epoch": 0.15467474321833027,
      "grad_norm": 2.543242931365967,
      "learning_rate": 4.2274163813537004e-05,
      "loss": 2.1293,
      "step": 5873
    },
    {
      "epoch": 0.154701079799842,
      "grad_norm": 1.7186356782913208,
      "learning_rate": 4.227284698446142e-05,
      "loss": 1.4707,
      "step": 5874
    },
    {
      "epoch": 0.1547274163813537,
      "grad_norm": 2.7276201248168945,
      "learning_rate": 4.2271530155385835e-05,
      "loss": 1.3838,
      "step": 5875
    },
    {
      "epoch": 0.15475375296286542,
      "grad_norm": 1.9710484743118286,
      "learning_rate": 4.227021332631025e-05,
      "loss": 1.7793,
      "step": 5876
    },
    {
      "epoch": 0.15478008954437714,
      "grad_norm": 2.6219520568847656,
      "learning_rate": 4.226889649723466e-05,
      "loss": 1.9442,
      "step": 5877
    },
    {
      "epoch": 0.15480642612588885,
      "grad_norm": 2.085421562194824,
      "learning_rate": 4.2267579668159075e-05,
      "loss": 1.7946,
      "step": 5878
    },
    {
      "epoch": 0.15483276270740057,
      "grad_norm": 6.2612738609313965,
      "learning_rate": 4.2266262839083484e-05,
      "loss": 2.296,
      "step": 5879
    },
    {
      "epoch": 0.1548590992889123,
      "grad_norm": 3.3673183917999268,
      "learning_rate": 4.2264946010007906e-05,
      "loss": 1.9816,
      "step": 5880
    },
    {
      "epoch": 0.15488543587042403,
      "grad_norm": 2.5657103061676025,
      "learning_rate": 4.2263629180932315e-05,
      "loss": 1.9609,
      "step": 5881
    },
    {
      "epoch": 0.15491177245193574,
      "grad_norm": 1.882688283920288,
      "learning_rate": 4.226231235185673e-05,
      "loss": 1.8117,
      "step": 5882
    },
    {
      "epoch": 0.15493810903344746,
      "grad_norm": 1.9992015361785889,
      "learning_rate": 4.2260995522781146e-05,
      "loss": 0.8731,
      "step": 5883
    },
    {
      "epoch": 0.15496444561495917,
      "grad_norm": 3.999999523162842,
      "learning_rate": 4.2259678693705555e-05,
      "loss": 1.2519,
      "step": 5884
    },
    {
      "epoch": 0.1549907821964709,
      "grad_norm": 2.3278517723083496,
      "learning_rate": 4.225836186462998e-05,
      "loss": 1.2304,
      "step": 5885
    },
    {
      "epoch": 0.1550171187779826,
      "grad_norm": 3.1681950092315674,
      "learning_rate": 4.2257045035554386e-05,
      "loss": 2.021,
      "step": 5886
    },
    {
      "epoch": 0.15504345535949435,
      "grad_norm": 2.4006595611572266,
      "learning_rate": 4.22557282064788e-05,
      "loss": 1.8774,
      "step": 5887
    },
    {
      "epoch": 0.15506979194100606,
      "grad_norm": 4.515176773071289,
      "learning_rate": 4.225441137740321e-05,
      "loss": 1.3982,
      "step": 5888
    },
    {
      "epoch": 0.15509612852251778,
      "grad_norm": 2.631305456161499,
      "learning_rate": 4.2253094548327626e-05,
      "loss": 1.4711,
      "step": 5889
    },
    {
      "epoch": 0.1551224651040295,
      "grad_norm": 1.4728326797485352,
      "learning_rate": 4.225177771925204e-05,
      "loss": 1.8721,
      "step": 5890
    },
    {
      "epoch": 0.1551488016855412,
      "grad_norm": 1.9591310024261475,
      "learning_rate": 4.225046089017646e-05,
      "loss": 1.8931,
      "step": 5891
    },
    {
      "epoch": 0.15517513826705293,
      "grad_norm": 2.040419816970825,
      "learning_rate": 4.224914406110087e-05,
      "loss": 1.7898,
      "step": 5892
    },
    {
      "epoch": 0.15520147484856467,
      "grad_norm": 1.7353781461715698,
      "learning_rate": 4.224782723202528e-05,
      "loss": 2.1022,
      "step": 5893
    },
    {
      "epoch": 0.15522781143007638,
      "grad_norm": 1.6009823083877563,
      "learning_rate": 4.2246510402949704e-05,
      "loss": 1.6196,
      "step": 5894
    },
    {
      "epoch": 0.1552541480115881,
      "grad_norm": 4.091841220855713,
      "learning_rate": 4.224519357387411e-05,
      "loss": 1.0809,
      "step": 5895
    },
    {
      "epoch": 0.15528048459309982,
      "grad_norm": 1.8655463457107544,
      "learning_rate": 4.224387674479853e-05,
      "loss": 2.0149,
      "step": 5896
    },
    {
      "epoch": 0.15530682117461153,
      "grad_norm": 2.3955373764038086,
      "learning_rate": 4.224255991572294e-05,
      "loss": 2.4355,
      "step": 5897
    },
    {
      "epoch": 0.15533315775612325,
      "grad_norm": 4.653458595275879,
      "learning_rate": 4.224124308664735e-05,
      "loss": 1.4602,
      "step": 5898
    },
    {
      "epoch": 0.15535949433763496,
      "grad_norm": 3.4966752529144287,
      "learning_rate": 4.223992625757177e-05,
      "loss": 1.204,
      "step": 5899
    },
    {
      "epoch": 0.1553858309191467,
      "grad_norm": 4.484646320343018,
      "learning_rate": 4.2238609428496184e-05,
      "loss": 1.6046,
      "step": 5900
    },
    {
      "epoch": 0.15541216750065842,
      "grad_norm": 2.649949550628662,
      "learning_rate": 4.22372925994206e-05,
      "loss": 1.954,
      "step": 5901
    },
    {
      "epoch": 0.15543850408217014,
      "grad_norm": 1.8868272304534912,
      "learning_rate": 4.223597577034501e-05,
      "loss": 1.4536,
      "step": 5902
    },
    {
      "epoch": 0.15546484066368185,
      "grad_norm": 2.0686264038085938,
      "learning_rate": 4.2234658941269424e-05,
      "loss": 1.6357,
      "step": 5903
    },
    {
      "epoch": 0.15549117724519357,
      "grad_norm": 1.8744795322418213,
      "learning_rate": 4.223334211219384e-05,
      "loss": 1.544,
      "step": 5904
    },
    {
      "epoch": 0.15551751382670528,
      "grad_norm": 2.567208766937256,
      "learning_rate": 4.2232025283118255e-05,
      "loss": 1.4843,
      "step": 5905
    },
    {
      "epoch": 0.15554385040821703,
      "grad_norm": 3.946901559829712,
      "learning_rate": 4.2230708454042664e-05,
      "loss": 1.6515,
      "step": 5906
    },
    {
      "epoch": 0.15557018698972874,
      "grad_norm": 1.8149755001068115,
      "learning_rate": 4.222939162496708e-05,
      "loss": 2.1445,
      "step": 5907
    },
    {
      "epoch": 0.15559652357124046,
      "grad_norm": 2.1219542026519775,
      "learning_rate": 4.2228074795891495e-05,
      "loss": 2.3332,
      "step": 5908
    },
    {
      "epoch": 0.15562286015275217,
      "grad_norm": 1.72514808177948,
      "learning_rate": 4.222675796681591e-05,
      "loss": 1.3152,
      "step": 5909
    },
    {
      "epoch": 0.1556491967342639,
      "grad_norm": 1.9629729986190796,
      "learning_rate": 4.2225441137740327e-05,
      "loss": 1.8084,
      "step": 5910
    },
    {
      "epoch": 0.1556755333157756,
      "grad_norm": 2.301128387451172,
      "learning_rate": 4.2224124308664735e-05,
      "loss": 1.7519,
      "step": 5911
    },
    {
      "epoch": 0.15570186989728732,
      "grad_norm": 2.1626784801483154,
      "learning_rate": 4.222280747958915e-05,
      "loss": 2.0657,
      "step": 5912
    },
    {
      "epoch": 0.15572820647879906,
      "grad_norm": 2.2094016075134277,
      "learning_rate": 4.2221490650513567e-05,
      "loss": 1.8229,
      "step": 5913
    },
    {
      "epoch": 0.15575454306031078,
      "grad_norm": 2.475167751312256,
      "learning_rate": 4.222017382143798e-05,
      "loss": 1.6667,
      "step": 5914
    },
    {
      "epoch": 0.1557808796418225,
      "grad_norm": 3.6198947429656982,
      "learning_rate": 4.221885699236239e-05,
      "loss": 1.192,
      "step": 5915
    },
    {
      "epoch": 0.1558072162233342,
      "grad_norm": 1.987687349319458,
      "learning_rate": 4.2217540163286807e-05,
      "loss": 2.4872,
      "step": 5916
    },
    {
      "epoch": 0.15583355280484593,
      "grad_norm": 2.642976760864258,
      "learning_rate": 4.221622333421122e-05,
      "loss": 1.0824,
      "step": 5917
    },
    {
      "epoch": 0.15585988938635764,
      "grad_norm": 1.8391703367233276,
      "learning_rate": 4.221490650513564e-05,
      "loss": 0.5032,
      "step": 5918
    },
    {
      "epoch": 0.15588622596786936,
      "grad_norm": 2.433856725692749,
      "learning_rate": 4.221358967606005e-05,
      "loss": 2.2421,
      "step": 5919
    },
    {
      "epoch": 0.1559125625493811,
      "grad_norm": 2.7764840126037598,
      "learning_rate": 4.221227284698446e-05,
      "loss": 1.4468,
      "step": 5920
    },
    {
      "epoch": 0.15593889913089282,
      "grad_norm": 1.781196117401123,
      "learning_rate": 4.221095601790888e-05,
      "loss": 1.862,
      "step": 5921
    },
    {
      "epoch": 0.15596523571240453,
      "grad_norm": 3.049417734146118,
      "learning_rate": 4.2209639188833287e-05,
      "loss": 1.8256,
      "step": 5922
    },
    {
      "epoch": 0.15599157229391625,
      "grad_norm": 4.471404075622559,
      "learning_rate": 4.220832235975771e-05,
      "loss": 2.0908,
      "step": 5923
    },
    {
      "epoch": 0.15601790887542796,
      "grad_norm": 2.175081491470337,
      "learning_rate": 4.220700553068212e-05,
      "loss": 1.7903,
      "step": 5924
    },
    {
      "epoch": 0.15604424545693968,
      "grad_norm": 1.913242220878601,
      "learning_rate": 4.2205688701606533e-05,
      "loss": 1.7829,
      "step": 5925
    },
    {
      "epoch": 0.15607058203845142,
      "grad_norm": 3.270598888397217,
      "learning_rate": 4.220437187253095e-05,
      "loss": 1.3993,
      "step": 5926
    },
    {
      "epoch": 0.15609691861996314,
      "grad_norm": 3.6847281455993652,
      "learning_rate": 4.2203055043455365e-05,
      "loss": 1.1214,
      "step": 5927
    },
    {
      "epoch": 0.15612325520147485,
      "grad_norm": 5.7359442710876465,
      "learning_rate": 4.220173821437978e-05,
      "loss": 1.7749,
      "step": 5928
    },
    {
      "epoch": 0.15614959178298657,
      "grad_norm": 2.10076642036438,
      "learning_rate": 4.220042138530419e-05,
      "loss": 2.0794,
      "step": 5929
    },
    {
      "epoch": 0.15617592836449828,
      "grad_norm": 2.211111545562744,
      "learning_rate": 4.2199104556228605e-05,
      "loss": 1.8982,
      "step": 5930
    },
    {
      "epoch": 0.15620226494601,
      "grad_norm": 2.931358814239502,
      "learning_rate": 4.2197787727153013e-05,
      "loss": 1.2573,
      "step": 5931
    },
    {
      "epoch": 0.15622860152752172,
      "grad_norm": 2.6709563732147217,
      "learning_rate": 4.2196470898077436e-05,
      "loss": 1.8708,
      "step": 5932
    },
    {
      "epoch": 0.15625493810903346,
      "grad_norm": 3.6660375595092773,
      "learning_rate": 4.2195154069001845e-05,
      "loss": 0.463,
      "step": 5933
    },
    {
      "epoch": 0.15628127469054517,
      "grad_norm": 2.1038217544555664,
      "learning_rate": 4.219383723992626e-05,
      "loss": 2.1543,
      "step": 5934
    },
    {
      "epoch": 0.1563076112720569,
      "grad_norm": 1.6519107818603516,
      "learning_rate": 4.219252041085067e-05,
      "loss": 2.2121,
      "step": 5935
    },
    {
      "epoch": 0.1563339478535686,
      "grad_norm": 2.7920644283294678,
      "learning_rate": 4.2191203581775085e-05,
      "loss": 0.483,
      "step": 5936
    },
    {
      "epoch": 0.15636028443508032,
      "grad_norm": 2.220864772796631,
      "learning_rate": 4.21898867526995e-05,
      "loss": 1.8855,
      "step": 5937
    },
    {
      "epoch": 0.15638662101659204,
      "grad_norm": 2.7000720500946045,
      "learning_rate": 4.2188569923623916e-05,
      "loss": 0.5763,
      "step": 5938
    },
    {
      "epoch": 0.15641295759810375,
      "grad_norm": 1.735314130783081,
      "learning_rate": 4.218725309454833e-05,
      "loss": 1.363,
      "step": 5939
    },
    {
      "epoch": 0.1564392941796155,
      "grad_norm": 3.029871940612793,
      "learning_rate": 4.218593626547274e-05,
      "loss": 1.6139,
      "step": 5940
    },
    {
      "epoch": 0.1564656307611272,
      "grad_norm": 1.7269423007965088,
      "learning_rate": 4.218461943639716e-05,
      "loss": 1.9701,
      "step": 5941
    },
    {
      "epoch": 0.15649196734263893,
      "grad_norm": 2.508410692214966,
      "learning_rate": 4.218330260732157e-05,
      "loss": 1.7556,
      "step": 5942
    },
    {
      "epoch": 0.15651830392415064,
      "grad_norm": 6.6043314933776855,
      "learning_rate": 4.218198577824599e-05,
      "loss": 2.4914,
      "step": 5943
    },
    {
      "epoch": 0.15654464050566236,
      "grad_norm": 2.9760141372680664,
      "learning_rate": 4.2180668949170396e-05,
      "loss": 2.1465,
      "step": 5944
    },
    {
      "epoch": 0.15657097708717407,
      "grad_norm": 1.6758267879486084,
      "learning_rate": 4.217935212009481e-05,
      "loss": 1.1808,
      "step": 5945
    },
    {
      "epoch": 0.15659731366868582,
      "grad_norm": 2.32725191116333,
      "learning_rate": 4.217803529101923e-05,
      "loss": 2.24,
      "step": 5946
    },
    {
      "epoch": 0.15662365025019753,
      "grad_norm": 2.053637981414795,
      "learning_rate": 4.217671846194364e-05,
      "loss": 1.7966,
      "step": 5947
    },
    {
      "epoch": 0.15664998683170925,
      "grad_norm": 1.6508933305740356,
      "learning_rate": 4.217540163286806e-05,
      "loss": 2.4334,
      "step": 5948
    },
    {
      "epoch": 0.15667632341322096,
      "grad_norm": 2.3459978103637695,
      "learning_rate": 4.217408480379247e-05,
      "loss": 2.1246,
      "step": 5949
    },
    {
      "epoch": 0.15670265999473268,
      "grad_norm": 2.6824984550476074,
      "learning_rate": 4.217276797471688e-05,
      "loss": 1.7024,
      "step": 5950
    },
    {
      "epoch": 0.1567289965762444,
      "grad_norm": 2.0082266330718994,
      "learning_rate": 4.21714511456413e-05,
      "loss": 1.6473,
      "step": 5951
    },
    {
      "epoch": 0.1567553331577561,
      "grad_norm": 2.1118314266204834,
      "learning_rate": 4.2170134316565714e-05,
      "loss": 1.8222,
      "step": 5952
    },
    {
      "epoch": 0.15678166973926785,
      "grad_norm": 2.196316719055176,
      "learning_rate": 4.216881748749012e-05,
      "loss": 1.1248,
      "step": 5953
    },
    {
      "epoch": 0.15680800632077957,
      "grad_norm": 3.4021639823913574,
      "learning_rate": 4.216750065841454e-05,
      "loss": 1.7389,
      "step": 5954
    },
    {
      "epoch": 0.15683434290229128,
      "grad_norm": 1.839551329612732,
      "learning_rate": 4.2166183829338954e-05,
      "loss": 1.8474,
      "step": 5955
    },
    {
      "epoch": 0.156860679483803,
      "grad_norm": 1.7327178716659546,
      "learning_rate": 4.216486700026337e-05,
      "loss": 1.9408,
      "step": 5956
    },
    {
      "epoch": 0.15688701606531472,
      "grad_norm": 3.7114171981811523,
      "learning_rate": 4.2163550171187785e-05,
      "loss": 2.2037,
      "step": 5957
    },
    {
      "epoch": 0.15691335264682643,
      "grad_norm": 2.627505302429199,
      "learning_rate": 4.2162233342112194e-05,
      "loss": 1.0022,
      "step": 5958
    },
    {
      "epoch": 0.15693968922833818,
      "grad_norm": 2.1350152492523193,
      "learning_rate": 4.216091651303661e-05,
      "loss": 0.3743,
      "step": 5959
    },
    {
      "epoch": 0.1569660258098499,
      "grad_norm": 3.300657033920288,
      "learning_rate": 4.2159599683961025e-05,
      "loss": 1.3048,
      "step": 5960
    },
    {
      "epoch": 0.1569923623913616,
      "grad_norm": 2.4122374057769775,
      "learning_rate": 4.215828285488544e-05,
      "loss": 2.2151,
      "step": 5961
    },
    {
      "epoch": 0.15701869897287332,
      "grad_norm": 3.083698272705078,
      "learning_rate": 4.215696602580985e-05,
      "loss": 1.3165,
      "step": 5962
    },
    {
      "epoch": 0.15704503555438504,
      "grad_norm": 2.317965269088745,
      "learning_rate": 4.2155649196734265e-05,
      "loss": 0.5157,
      "step": 5963
    },
    {
      "epoch": 0.15707137213589675,
      "grad_norm": 2.3910367488861084,
      "learning_rate": 4.215433236765868e-05,
      "loss": 2.3669,
      "step": 5964
    },
    {
      "epoch": 0.15709770871740847,
      "grad_norm": 1.941091537475586,
      "learning_rate": 4.2153015538583096e-05,
      "loss": 2.5886,
      "step": 5965
    },
    {
      "epoch": 0.1571240452989202,
      "grad_norm": 2.7749321460723877,
      "learning_rate": 4.215169870950751e-05,
      "loss": 1.7127,
      "step": 5966
    },
    {
      "epoch": 0.15715038188043193,
      "grad_norm": 5.011218547821045,
      "learning_rate": 4.215038188043192e-05,
      "loss": 1.9305,
      "step": 5967
    },
    {
      "epoch": 0.15717671846194364,
      "grad_norm": 2.4140255451202393,
      "learning_rate": 4.2149065051356336e-05,
      "loss": 2.2752,
      "step": 5968
    },
    {
      "epoch": 0.15720305504345536,
      "grad_norm": 3.319852113723755,
      "learning_rate": 4.2147748222280745e-05,
      "loss": 1.2568,
      "step": 5969
    },
    {
      "epoch": 0.15722939162496707,
      "grad_norm": 3.5438194274902344,
      "learning_rate": 4.214643139320517e-05,
      "loss": 1.2369,
      "step": 5970
    },
    {
      "epoch": 0.1572557282064788,
      "grad_norm": 6.879681587219238,
      "learning_rate": 4.2145114564129576e-05,
      "loss": 2.2988,
      "step": 5971
    },
    {
      "epoch": 0.1572820647879905,
      "grad_norm": 2.6652133464813232,
      "learning_rate": 4.214379773505399e-05,
      "loss": 1.8378,
      "step": 5972
    },
    {
      "epoch": 0.15730840136950225,
      "grad_norm": 2.0418589115142822,
      "learning_rate": 4.214248090597841e-05,
      "loss": 2.3504,
      "step": 5973
    },
    {
      "epoch": 0.15733473795101396,
      "grad_norm": 2.0920841693878174,
      "learning_rate": 4.214116407690282e-05,
      "loss": 2.1954,
      "step": 5974
    },
    {
      "epoch": 0.15736107453252568,
      "grad_norm": 6.576579570770264,
      "learning_rate": 4.213984724782724e-05,
      "loss": 2.1881,
      "step": 5975
    },
    {
      "epoch": 0.1573874111140374,
      "grad_norm": 2.9959959983825684,
      "learning_rate": 4.213853041875165e-05,
      "loss": 1.5807,
      "step": 5976
    },
    {
      "epoch": 0.1574137476955491,
      "grad_norm": 1.959325909614563,
      "learning_rate": 4.213721358967606e-05,
      "loss": 1.3493,
      "step": 5977
    },
    {
      "epoch": 0.15744008427706083,
      "grad_norm": 9.182555198669434,
      "learning_rate": 4.213589676060047e-05,
      "loss": 2.2217,
      "step": 5978
    },
    {
      "epoch": 0.15746642085857257,
      "grad_norm": 2.128164529800415,
      "learning_rate": 4.2134579931524894e-05,
      "loss": 1.9559,
      "step": 5979
    },
    {
      "epoch": 0.15749275744008429,
      "grad_norm": 1.984562635421753,
      "learning_rate": 4.21332631024493e-05,
      "loss": 2.0752,
      "step": 5980
    },
    {
      "epoch": 0.157519094021596,
      "grad_norm": 2.0138955116271973,
      "learning_rate": 4.213194627337372e-05,
      "loss": 1.9993,
      "step": 5981
    },
    {
      "epoch": 0.15754543060310772,
      "grad_norm": 2.8540875911712646,
      "learning_rate": 4.213062944429813e-05,
      "loss": 2.5846,
      "step": 5982
    },
    {
      "epoch": 0.15757176718461943,
      "grad_norm": 4.09743595123291,
      "learning_rate": 4.212931261522254e-05,
      "loss": 1.5782,
      "step": 5983
    },
    {
      "epoch": 0.15759810376613115,
      "grad_norm": 2.2861154079437256,
      "learning_rate": 4.212799578614696e-05,
      "loss": 1.4951,
      "step": 5984
    },
    {
      "epoch": 0.15762444034764286,
      "grad_norm": 3.8065881729125977,
      "learning_rate": 4.2126678957071374e-05,
      "loss": 1.2832,
      "step": 5985
    },
    {
      "epoch": 0.1576507769291546,
      "grad_norm": 2.0236055850982666,
      "learning_rate": 4.212536212799579e-05,
      "loss": 1.5231,
      "step": 5986
    },
    {
      "epoch": 0.15767711351066632,
      "grad_norm": 2.220747947692871,
      "learning_rate": 4.21240452989202e-05,
      "loss": 0.7072,
      "step": 5987
    },
    {
      "epoch": 0.15770345009217804,
      "grad_norm": 3.2718474864959717,
      "learning_rate": 4.212272846984462e-05,
      "loss": 1.3849,
      "step": 5988
    },
    {
      "epoch": 0.15772978667368975,
      "grad_norm": 1.6934703588485718,
      "learning_rate": 4.212141164076903e-05,
      "loss": 1.7882,
      "step": 5989
    },
    {
      "epoch": 0.15775612325520147,
      "grad_norm": 1.4751981496810913,
      "learning_rate": 4.2120094811693446e-05,
      "loss": 0.4046,
      "step": 5990
    },
    {
      "epoch": 0.15778245983671318,
      "grad_norm": 1.5996410846710205,
      "learning_rate": 4.2118777982617854e-05,
      "loss": 2.0365,
      "step": 5991
    },
    {
      "epoch": 0.15780879641822493,
      "grad_norm": 2.111327648162842,
      "learning_rate": 4.211746115354227e-05,
      "loss": 0.646,
      "step": 5992
    },
    {
      "epoch": 0.15783513299973664,
      "grad_norm": 2.0970349311828613,
      "learning_rate": 4.2116144324466686e-05,
      "loss": 1.7066,
      "step": 5993
    },
    {
      "epoch": 0.15786146958124836,
      "grad_norm": 3.1678991317749023,
      "learning_rate": 4.21148274953911e-05,
      "loss": 1.5934,
      "step": 5994
    },
    {
      "epoch": 0.15788780616276007,
      "grad_norm": 2.683861255645752,
      "learning_rate": 4.211351066631552e-05,
      "loss": 1.8981,
      "step": 5995
    },
    {
      "epoch": 0.1579141427442718,
      "grad_norm": 3.6035783290863037,
      "learning_rate": 4.2112193837239926e-05,
      "loss": 2.1261,
      "step": 5996
    },
    {
      "epoch": 0.1579404793257835,
      "grad_norm": 2.172053813934326,
      "learning_rate": 4.211087700816434e-05,
      "loss": 0.4462,
      "step": 5997
    },
    {
      "epoch": 0.15796681590729522,
      "grad_norm": 1.8016635179519653,
      "learning_rate": 4.210956017908876e-05,
      "loss": 1.7273,
      "step": 5998
    },
    {
      "epoch": 0.15799315248880696,
      "grad_norm": 5.404572486877441,
      "learning_rate": 4.210824335001317e-05,
      "loss": 2.403,
      "step": 5999
    },
    {
      "epoch": 0.15801948907031868,
      "grad_norm": 4.310553550720215,
      "learning_rate": 4.210692652093758e-05,
      "loss": 1.1839,
      "step": 6000
    },
    {
      "epoch": 0.1580458256518304,
      "grad_norm": 1.9727861881256104,
      "learning_rate": 4.2105609691862e-05,
      "loss": 1.3579,
      "step": 6001
    },
    {
      "epoch": 0.1580721622333421,
      "grad_norm": 2.3161890506744385,
      "learning_rate": 4.210429286278641e-05,
      "loss": 1.7293,
      "step": 6002
    },
    {
      "epoch": 0.15809849881485383,
      "grad_norm": 3.0780434608459473,
      "learning_rate": 4.210297603371083e-05,
      "loss": 1.6469,
      "step": 6003
    },
    {
      "epoch": 0.15812483539636554,
      "grad_norm": 2.0891878604888916,
      "learning_rate": 4.2101659204635244e-05,
      "loss": 1.943,
      "step": 6004
    },
    {
      "epoch": 0.15815117197787726,
      "grad_norm": 3.240722894668579,
      "learning_rate": 4.210034237555965e-05,
      "loss": 1.1894,
      "step": 6005
    },
    {
      "epoch": 0.158177508559389,
      "grad_norm": 2.5613348484039307,
      "learning_rate": 4.209902554648407e-05,
      "loss": 1.5481,
      "step": 6006
    },
    {
      "epoch": 0.15820384514090072,
      "grad_norm": 1.823609709739685,
      "learning_rate": 4.2097708717408484e-05,
      "loss": 1.7977,
      "step": 6007
    },
    {
      "epoch": 0.15823018172241243,
      "grad_norm": 2.269915819168091,
      "learning_rate": 4.20963918883329e-05,
      "loss": 2.1684,
      "step": 6008
    },
    {
      "epoch": 0.15825651830392415,
      "grad_norm": 2.3334107398986816,
      "learning_rate": 4.209507505925731e-05,
      "loss": 3.0635,
      "step": 6009
    },
    {
      "epoch": 0.15828285488543586,
      "grad_norm": 2.455613136291504,
      "learning_rate": 4.2093758230181724e-05,
      "loss": 0.5861,
      "step": 6010
    },
    {
      "epoch": 0.15830919146694758,
      "grad_norm": 2.6880509853363037,
      "learning_rate": 4.209244140110614e-05,
      "loss": 1.3298,
      "step": 6011
    },
    {
      "epoch": 0.15833552804845932,
      "grad_norm": 2.403714179992676,
      "learning_rate": 4.2091124572030555e-05,
      "loss": 1.9459,
      "step": 6012
    },
    {
      "epoch": 0.15836186462997104,
      "grad_norm": 1.4829332828521729,
      "learning_rate": 4.208980774295497e-05,
      "loss": 0.4841,
      "step": 6013
    },
    {
      "epoch": 0.15838820121148275,
      "grad_norm": 1.6144860982894897,
      "learning_rate": 4.208849091387938e-05,
      "loss": 1.8183,
      "step": 6014
    },
    {
      "epoch": 0.15841453779299447,
      "grad_norm": 2.2351977825164795,
      "learning_rate": 4.2087174084803795e-05,
      "loss": 2.0403,
      "step": 6015
    },
    {
      "epoch": 0.15844087437450619,
      "grad_norm": 2.477430820465088,
      "learning_rate": 4.2085857255728204e-05,
      "loss": 0.9024,
      "step": 6016
    },
    {
      "epoch": 0.1584672109560179,
      "grad_norm": 2.137519359588623,
      "learning_rate": 4.2084540426652626e-05,
      "loss": 1.5758,
      "step": 6017
    },
    {
      "epoch": 0.15849354753752962,
      "grad_norm": 4.021152496337891,
      "learning_rate": 4.2083223597577035e-05,
      "loss": 2.4805,
      "step": 6018
    },
    {
      "epoch": 0.15851988411904136,
      "grad_norm": 2.429715871810913,
      "learning_rate": 4.208190676850145e-05,
      "loss": 0.5486,
      "step": 6019
    },
    {
      "epoch": 0.15854622070055308,
      "grad_norm": 1.847489356994629,
      "learning_rate": 4.2080589939425866e-05,
      "loss": 2.4164,
      "step": 6020
    },
    {
      "epoch": 0.1585725572820648,
      "grad_norm": 2.0357372760772705,
      "learning_rate": 4.207927311035028e-05,
      "loss": 1.968,
      "step": 6021
    },
    {
      "epoch": 0.1585988938635765,
      "grad_norm": 2.8058173656463623,
      "learning_rate": 4.20779562812747e-05,
      "loss": 1.7579,
      "step": 6022
    },
    {
      "epoch": 0.15862523044508822,
      "grad_norm": 1.9289313554763794,
      "learning_rate": 4.2076639452199106e-05,
      "loss": 1.7415,
      "step": 6023
    },
    {
      "epoch": 0.15865156702659994,
      "grad_norm": 1.9205622673034668,
      "learning_rate": 4.207532262312352e-05,
      "loss": 1.9149,
      "step": 6024
    },
    {
      "epoch": 0.15867790360811165,
      "grad_norm": 5.0248637199401855,
      "learning_rate": 4.207400579404793e-05,
      "loss": 1.9054,
      "step": 6025
    },
    {
      "epoch": 0.1587042401896234,
      "grad_norm": 3.2937707901000977,
      "learning_rate": 4.207268896497235e-05,
      "loss": 1.3608,
      "step": 6026
    },
    {
      "epoch": 0.1587305767711351,
      "grad_norm": 2.269594192504883,
      "learning_rate": 4.207137213589676e-05,
      "loss": 0.7694,
      "step": 6027
    },
    {
      "epoch": 0.15875691335264683,
      "grad_norm": 3.008793592453003,
      "learning_rate": 4.207005530682118e-05,
      "loss": 1.2093,
      "step": 6028
    },
    {
      "epoch": 0.15878324993415854,
      "grad_norm": 2.551845073699951,
      "learning_rate": 4.2068738477745586e-05,
      "loss": 1.7906,
      "step": 6029
    },
    {
      "epoch": 0.15880958651567026,
      "grad_norm": 2.498796224594116,
      "learning_rate": 4.206742164867e-05,
      "loss": 1.7394,
      "step": 6030
    },
    {
      "epoch": 0.15883592309718197,
      "grad_norm": 2.4335715770721436,
      "learning_rate": 4.2066104819594424e-05,
      "loss": 1.7267,
      "step": 6031
    },
    {
      "epoch": 0.15886225967869372,
      "grad_norm": 1.9502872228622437,
      "learning_rate": 4.206478799051883e-05,
      "loss": 2.2899,
      "step": 6032
    },
    {
      "epoch": 0.15888859626020543,
      "grad_norm": 2.2342894077301025,
      "learning_rate": 4.206347116144325e-05,
      "loss": 1.6091,
      "step": 6033
    },
    {
      "epoch": 0.15891493284171715,
      "grad_norm": 1.7040371894836426,
      "learning_rate": 4.206215433236766e-05,
      "loss": 1.3489,
      "step": 6034
    },
    {
      "epoch": 0.15894126942322886,
      "grad_norm": 2.3223648071289062,
      "learning_rate": 4.206083750329207e-05,
      "loss": 1.6438,
      "step": 6035
    },
    {
      "epoch": 0.15896760600474058,
      "grad_norm": 1.715370535850525,
      "learning_rate": 4.205952067421649e-05,
      "loss": 1.6805,
      "step": 6036
    },
    {
      "epoch": 0.1589939425862523,
      "grad_norm": 3.320110559463501,
      "learning_rate": 4.2058203845140904e-05,
      "loss": 1.7764,
      "step": 6037
    },
    {
      "epoch": 0.159020279167764,
      "grad_norm": 3.9556894302368164,
      "learning_rate": 4.205688701606531e-05,
      "loss": 1.7075,
      "step": 6038
    },
    {
      "epoch": 0.15904661574927575,
      "grad_norm": 1.596814751625061,
      "learning_rate": 4.205557018698973e-05,
      "loss": 1.8104,
      "step": 6039
    },
    {
      "epoch": 0.15907295233078747,
      "grad_norm": 2.765883684158325,
      "learning_rate": 4.2054253357914144e-05,
      "loss": 1.5613,
      "step": 6040
    },
    {
      "epoch": 0.15909928891229919,
      "grad_norm": 1.7439045906066895,
      "learning_rate": 4.205293652883856e-05,
      "loss": 2.2978,
      "step": 6041
    },
    {
      "epoch": 0.1591256254938109,
      "grad_norm": 2.8818860054016113,
      "learning_rate": 4.2051619699762975e-05,
      "loss": 2.4009,
      "step": 6042
    },
    {
      "epoch": 0.15915196207532262,
      "grad_norm": 2.9042367935180664,
      "learning_rate": 4.2050302870687384e-05,
      "loss": 1.0585,
      "step": 6043
    },
    {
      "epoch": 0.15917829865683433,
      "grad_norm": 1.871683120727539,
      "learning_rate": 4.20489860416118e-05,
      "loss": 1.6521,
      "step": 6044
    },
    {
      "epoch": 0.15920463523834608,
      "grad_norm": 4.8750433921813965,
      "learning_rate": 4.2047669212536215e-05,
      "loss": 0.8817,
      "step": 6045
    },
    {
      "epoch": 0.1592309718198578,
      "grad_norm": 1.8055793046951294,
      "learning_rate": 4.204635238346063e-05,
      "loss": 1.9334,
      "step": 6046
    },
    {
      "epoch": 0.1592573084013695,
      "grad_norm": 2.7020907402038574,
      "learning_rate": 4.204503555438504e-05,
      "loss": 0.5869,
      "step": 6047
    },
    {
      "epoch": 0.15928364498288122,
      "grad_norm": 1.832258939743042,
      "learning_rate": 4.2043718725309455e-05,
      "loss": 1.9233,
      "step": 6048
    },
    {
      "epoch": 0.15930998156439294,
      "grad_norm": 2.2810006141662598,
      "learning_rate": 4.204240189623387e-05,
      "loss": 1.8354,
      "step": 6049
    },
    {
      "epoch": 0.15933631814590465,
      "grad_norm": 1.8633068799972534,
      "learning_rate": 4.2041085067158287e-05,
      "loss": 1.589,
      "step": 6050
    },
    {
      "epoch": 0.15936265472741637,
      "grad_norm": 2.8353729248046875,
      "learning_rate": 4.20397682380827e-05,
      "loss": 1.6054,
      "step": 6051
    },
    {
      "epoch": 0.1593889913089281,
      "grad_norm": 1.9458913803100586,
      "learning_rate": 4.203845140900711e-05,
      "loss": 1.167,
      "step": 6052
    },
    {
      "epoch": 0.15941532789043983,
      "grad_norm": 1.5614548921585083,
      "learning_rate": 4.2037134579931527e-05,
      "loss": 1.0224,
      "step": 6053
    },
    {
      "epoch": 0.15944166447195154,
      "grad_norm": 2.363529682159424,
      "learning_rate": 4.203581775085594e-05,
      "loss": 2.1562,
      "step": 6054
    },
    {
      "epoch": 0.15946800105346326,
      "grad_norm": 1.907061219215393,
      "learning_rate": 4.203450092178036e-05,
      "loss": 2.2413,
      "step": 6055
    },
    {
      "epoch": 0.15949433763497498,
      "grad_norm": 5.0372819900512695,
      "learning_rate": 4.203318409270477e-05,
      "loss": 0.5852,
      "step": 6056
    },
    {
      "epoch": 0.1595206742164867,
      "grad_norm": 1.8625068664550781,
      "learning_rate": 4.203186726362918e-05,
      "loss": 1.1972,
      "step": 6057
    },
    {
      "epoch": 0.1595470107979984,
      "grad_norm": 2.006800413131714,
      "learning_rate": 4.20305504345536e-05,
      "loss": 1.2923,
      "step": 6058
    },
    {
      "epoch": 0.15957334737951015,
      "grad_norm": 1.7793184518814087,
      "learning_rate": 4.2029233605478013e-05,
      "loss": 1.1352,
      "step": 6059
    },
    {
      "epoch": 0.15959968396102187,
      "grad_norm": 3.3056442737579346,
      "learning_rate": 4.202791677640243e-05,
      "loss": 1.5788,
      "step": 6060
    },
    {
      "epoch": 0.15962602054253358,
      "grad_norm": 1.7238984107971191,
      "learning_rate": 4.202659994732684e-05,
      "loss": 2.3818,
      "step": 6061
    },
    {
      "epoch": 0.1596523571240453,
      "grad_norm": 3.204756498336792,
      "learning_rate": 4.2025283118251253e-05,
      "loss": 2.0978,
      "step": 6062
    },
    {
      "epoch": 0.159678693705557,
      "grad_norm": 2.7013306617736816,
      "learning_rate": 4.202396628917566e-05,
      "loss": 1.4563,
      "step": 6063
    },
    {
      "epoch": 0.15970503028706873,
      "grad_norm": 1.7570176124572754,
      "learning_rate": 4.2022649460100085e-05,
      "loss": 1.7673,
      "step": 6064
    },
    {
      "epoch": 0.15973136686858047,
      "grad_norm": 1.803415060043335,
      "learning_rate": 4.2021332631024493e-05,
      "loss": 1.7306,
      "step": 6065
    },
    {
      "epoch": 0.1597577034500922,
      "grad_norm": 2.3842899799346924,
      "learning_rate": 4.202001580194891e-05,
      "loss": 2.4828,
      "step": 6066
    },
    {
      "epoch": 0.1597840400316039,
      "grad_norm": 3.046964645385742,
      "learning_rate": 4.2018698972873325e-05,
      "loss": 1.9093,
      "step": 6067
    },
    {
      "epoch": 0.15981037661311562,
      "grad_norm": 2.353818416595459,
      "learning_rate": 4.2017382143797733e-05,
      "loss": 1.5821,
      "step": 6068
    },
    {
      "epoch": 0.15983671319462733,
      "grad_norm": 2.210801839828491,
      "learning_rate": 4.2016065314722156e-05,
      "loss": 0.9711,
      "step": 6069
    },
    {
      "epoch": 0.15986304977613905,
      "grad_norm": 3.385026454925537,
      "learning_rate": 4.2014748485646565e-05,
      "loss": 1.688,
      "step": 6070
    },
    {
      "epoch": 0.15988938635765076,
      "grad_norm": 3.45007586479187,
      "learning_rate": 4.201343165657098e-05,
      "loss": 2.0668,
      "step": 6071
    },
    {
      "epoch": 0.1599157229391625,
      "grad_norm": 3.061713933944702,
      "learning_rate": 4.201211482749539e-05,
      "loss": 1.9086,
      "step": 6072
    },
    {
      "epoch": 0.15994205952067422,
      "grad_norm": 2.809657096862793,
      "learning_rate": 4.201079799841981e-05,
      "loss": 1.4217,
      "step": 6073
    },
    {
      "epoch": 0.15996839610218594,
      "grad_norm": 2.6209864616394043,
      "learning_rate": 4.200948116934422e-05,
      "loss": 1.914,
      "step": 6074
    },
    {
      "epoch": 0.15999473268369765,
      "grad_norm": 3.004488468170166,
      "learning_rate": 4.2008164340268636e-05,
      "loss": 2.0692,
      "step": 6075
    },
    {
      "epoch": 0.16002106926520937,
      "grad_norm": 3.5518791675567627,
      "learning_rate": 4.200684751119305e-05,
      "loss": 1.2734,
      "step": 6076
    },
    {
      "epoch": 0.16004740584672109,
      "grad_norm": 2.149366855621338,
      "learning_rate": 4.200553068211746e-05,
      "loss": 1.0786,
      "step": 6077
    },
    {
      "epoch": 0.16007374242823283,
      "grad_norm": 2.207019090652466,
      "learning_rate": 4.200421385304188e-05,
      "loss": 1.9736,
      "step": 6078
    },
    {
      "epoch": 0.16010007900974454,
      "grad_norm": 3.112743377685547,
      "learning_rate": 4.200289702396629e-05,
      "loss": 2.1226,
      "step": 6079
    },
    {
      "epoch": 0.16012641559125626,
      "grad_norm": 2.382636785507202,
      "learning_rate": 4.200158019489071e-05,
      "loss": 1.677,
      "step": 6080
    },
    {
      "epoch": 0.16015275217276798,
      "grad_norm": 2.8617513179779053,
      "learning_rate": 4.2000263365815116e-05,
      "loss": 1.5793,
      "step": 6081
    },
    {
      "epoch": 0.1601790887542797,
      "grad_norm": 2.2968790531158447,
      "learning_rate": 4.199894653673953e-05,
      "loss": 2.1024,
      "step": 6082
    },
    {
      "epoch": 0.1602054253357914,
      "grad_norm": 2.184506416320801,
      "learning_rate": 4.199762970766395e-05,
      "loss": 1.9684,
      "step": 6083
    },
    {
      "epoch": 0.16023176191730312,
      "grad_norm": 4.663570404052734,
      "learning_rate": 4.199631287858836e-05,
      "loss": 1.7537,
      "step": 6084
    },
    {
      "epoch": 0.16025809849881487,
      "grad_norm": 2.0986099243164062,
      "learning_rate": 4.199499604951277e-05,
      "loss": 1.8167,
      "step": 6085
    },
    {
      "epoch": 0.16028443508032658,
      "grad_norm": 3.0223231315612793,
      "learning_rate": 4.199367922043719e-05,
      "loss": 1.5036,
      "step": 6086
    },
    {
      "epoch": 0.1603107716618383,
      "grad_norm": 5.098803520202637,
      "learning_rate": 4.19923623913616e-05,
      "loss": 1.5529,
      "step": 6087
    },
    {
      "epoch": 0.16033710824335,
      "grad_norm": 3.4167065620422363,
      "learning_rate": 4.199104556228602e-05,
      "loss": 1.4869,
      "step": 6088
    },
    {
      "epoch": 0.16036344482486173,
      "grad_norm": 1.9871636629104614,
      "learning_rate": 4.1989728733210434e-05,
      "loss": 1.6781,
      "step": 6089
    },
    {
      "epoch": 0.16038978140637344,
      "grad_norm": 1.926769495010376,
      "learning_rate": 4.198841190413484e-05,
      "loss": 1.6861,
      "step": 6090
    },
    {
      "epoch": 0.16041611798788516,
      "grad_norm": 3.153432607650757,
      "learning_rate": 4.198709507505926e-05,
      "loss": 1.4504,
      "step": 6091
    },
    {
      "epoch": 0.1604424545693969,
      "grad_norm": 4.245177268981934,
      "learning_rate": 4.1985778245983674e-05,
      "loss": 1.6688,
      "step": 6092
    },
    {
      "epoch": 0.16046879115090862,
      "grad_norm": 6.108002185821533,
      "learning_rate": 4.198446141690809e-05,
      "loss": 2.126,
      "step": 6093
    },
    {
      "epoch": 0.16049512773242033,
      "grad_norm": 2.3100204467773438,
      "learning_rate": 4.19831445878325e-05,
      "loss": 1.6948,
      "step": 6094
    },
    {
      "epoch": 0.16052146431393205,
      "grad_norm": 2.5932717323303223,
      "learning_rate": 4.1981827758756914e-05,
      "loss": 1.6972,
      "step": 6095
    },
    {
      "epoch": 0.16054780089544377,
      "grad_norm": 2.8772497177124023,
      "learning_rate": 4.198051092968133e-05,
      "loss": 0.9477,
      "step": 6096
    },
    {
      "epoch": 0.16057413747695548,
      "grad_norm": 1.6577174663543701,
      "learning_rate": 4.1979194100605745e-05,
      "loss": 1.4815,
      "step": 6097
    },
    {
      "epoch": 0.16060047405846722,
      "grad_norm": 1.90794038772583,
      "learning_rate": 4.197787727153016e-05,
      "loss": 1.7716,
      "step": 6098
    },
    {
      "epoch": 0.16062681063997894,
      "grad_norm": 3.2309482097625732,
      "learning_rate": 4.197656044245457e-05,
      "loss": 0.4605,
      "step": 6099
    },
    {
      "epoch": 0.16065314722149066,
      "grad_norm": 2.1344552040100098,
      "learning_rate": 4.1975243613378985e-05,
      "loss": 1.7736,
      "step": 6100
    },
    {
      "epoch": 0.16067948380300237,
      "grad_norm": 2.5901072025299072,
      "learning_rate": 4.1973926784303394e-05,
      "loss": 1.629,
      "step": 6101
    },
    {
      "epoch": 0.1607058203845141,
      "grad_norm": 1.4747397899627686,
      "learning_rate": 4.1972609955227816e-05,
      "loss": 1.7209,
      "step": 6102
    },
    {
      "epoch": 0.1607321569660258,
      "grad_norm": 3.776444435119629,
      "learning_rate": 4.1971293126152225e-05,
      "loss": 1.5893,
      "step": 6103
    },
    {
      "epoch": 0.16075849354753752,
      "grad_norm": 2.1120872497558594,
      "learning_rate": 4.196997629707664e-05,
      "loss": 1.2256,
      "step": 6104
    },
    {
      "epoch": 0.16078483012904926,
      "grad_norm": 1.907181739807129,
      "learning_rate": 4.1968659468001056e-05,
      "loss": 1.0122,
      "step": 6105
    },
    {
      "epoch": 0.16081116671056098,
      "grad_norm": 2.6752939224243164,
      "learning_rate": 4.196734263892547e-05,
      "loss": 1.5295,
      "step": 6106
    },
    {
      "epoch": 0.1608375032920727,
      "grad_norm": 2.014364004135132,
      "learning_rate": 4.196602580984989e-05,
      "loss": 2.0183,
      "step": 6107
    },
    {
      "epoch": 0.1608638398735844,
      "grad_norm": 3.470395565032959,
      "learning_rate": 4.1964708980774296e-05,
      "loss": 1.6397,
      "step": 6108
    },
    {
      "epoch": 0.16089017645509612,
      "grad_norm": 1.9015464782714844,
      "learning_rate": 4.196339215169871e-05,
      "loss": 1.7849,
      "step": 6109
    },
    {
      "epoch": 0.16091651303660784,
      "grad_norm": 2.7722132205963135,
      "learning_rate": 4.196207532262312e-05,
      "loss": 2.2164,
      "step": 6110
    },
    {
      "epoch": 0.16094284961811955,
      "grad_norm": 3.117102861404419,
      "learning_rate": 4.196075849354754e-05,
      "loss": 0.886,
      "step": 6111
    },
    {
      "epoch": 0.1609691861996313,
      "grad_norm": 1.9750547409057617,
      "learning_rate": 4.195944166447195e-05,
      "loss": 2.0688,
      "step": 6112
    },
    {
      "epoch": 0.160995522781143,
      "grad_norm": 2.2968084812164307,
      "learning_rate": 4.195812483539637e-05,
      "loss": 1.6109,
      "step": 6113
    },
    {
      "epoch": 0.16102185936265473,
      "grad_norm": 3.515033483505249,
      "learning_rate": 4.195680800632078e-05,
      "loss": 1.7049,
      "step": 6114
    },
    {
      "epoch": 0.16104819594416644,
      "grad_norm": 2.277261972427368,
      "learning_rate": 4.195549117724519e-05,
      "loss": 2.1311,
      "step": 6115
    },
    {
      "epoch": 0.16107453252567816,
      "grad_norm": 1.7919727563858032,
      "learning_rate": 4.1954174348169614e-05,
      "loss": 1.7363,
      "step": 6116
    },
    {
      "epoch": 0.16110086910718988,
      "grad_norm": 3.9228217601776123,
      "learning_rate": 4.195285751909402e-05,
      "loss": 1.7039,
      "step": 6117
    },
    {
      "epoch": 0.16112720568870162,
      "grad_norm": 3.793929100036621,
      "learning_rate": 4.195154069001844e-05,
      "loss": 1.812,
      "step": 6118
    },
    {
      "epoch": 0.16115354227021333,
      "grad_norm": 3.7437984943389893,
      "learning_rate": 4.195022386094285e-05,
      "loss": 1.0529,
      "step": 6119
    },
    {
      "epoch": 0.16117987885172505,
      "grad_norm": 3.8403408527374268,
      "learning_rate": 4.194890703186727e-05,
      "loss": 1.6427,
      "step": 6120
    },
    {
      "epoch": 0.16120621543323677,
      "grad_norm": 3.307833194732666,
      "learning_rate": 4.194759020279168e-05,
      "loss": 1.6098,
      "step": 6121
    },
    {
      "epoch": 0.16123255201474848,
      "grad_norm": 2.962394952774048,
      "learning_rate": 4.1946273373716094e-05,
      "loss": 2.1496,
      "step": 6122
    },
    {
      "epoch": 0.1612588885962602,
      "grad_norm": 1.7113622426986694,
      "learning_rate": 4.194495654464051e-05,
      "loss": 1.3856,
      "step": 6123
    },
    {
      "epoch": 0.1612852251777719,
      "grad_norm": 5.497935771942139,
      "learning_rate": 4.194363971556492e-05,
      "loss": 1.1016,
      "step": 6124
    },
    {
      "epoch": 0.16131156175928366,
      "grad_norm": 5.836162567138672,
      "learning_rate": 4.194232288648934e-05,
      "loss": 1.5436,
      "step": 6125
    },
    {
      "epoch": 0.16133789834079537,
      "grad_norm": 2.716874122619629,
      "learning_rate": 4.194100605741375e-05,
      "loss": 2.0618,
      "step": 6126
    },
    {
      "epoch": 0.1613642349223071,
      "grad_norm": 3.397531270980835,
      "learning_rate": 4.1939689228338166e-05,
      "loss": 1.6688,
      "step": 6127
    },
    {
      "epoch": 0.1613905715038188,
      "grad_norm": 1.9136908054351807,
      "learning_rate": 4.1938372399262574e-05,
      "loss": 1.5963,
      "step": 6128
    },
    {
      "epoch": 0.16141690808533052,
      "grad_norm": 1.7691987752914429,
      "learning_rate": 4.193705557018699e-05,
      "loss": 0.4547,
      "step": 6129
    },
    {
      "epoch": 0.16144324466684223,
      "grad_norm": 1.9893584251403809,
      "learning_rate": 4.1935738741111406e-05,
      "loss": 1.6099,
      "step": 6130
    },
    {
      "epoch": 0.16146958124835398,
      "grad_norm": 1.7568279504776,
      "learning_rate": 4.193442191203582e-05,
      "loss": 2.1532,
      "step": 6131
    },
    {
      "epoch": 0.1614959178298657,
      "grad_norm": 2.622365713119507,
      "learning_rate": 4.193310508296023e-05,
      "loss": 2.0063,
      "step": 6132
    },
    {
      "epoch": 0.1615222544113774,
      "grad_norm": 2.176316738128662,
      "learning_rate": 4.1931788253884646e-05,
      "loss": 1.6139,
      "step": 6133
    },
    {
      "epoch": 0.16154859099288912,
      "grad_norm": 1.9413889646530151,
      "learning_rate": 4.193047142480906e-05,
      "loss": 2.0126,
      "step": 6134
    },
    {
      "epoch": 0.16157492757440084,
      "grad_norm": 3.032728672027588,
      "learning_rate": 4.192915459573348e-05,
      "loss": 2.1629,
      "step": 6135
    },
    {
      "epoch": 0.16160126415591256,
      "grad_norm": 4.204956531524658,
      "learning_rate": 4.192783776665789e-05,
      "loss": 1.4779,
      "step": 6136
    },
    {
      "epoch": 0.16162760073742427,
      "grad_norm": 2.328699827194214,
      "learning_rate": 4.19265209375823e-05,
      "loss": 1.7515,
      "step": 6137
    },
    {
      "epoch": 0.16165393731893601,
      "grad_norm": 2.612292528152466,
      "learning_rate": 4.192520410850672e-05,
      "loss": 1.3,
      "step": 6138
    },
    {
      "epoch": 0.16168027390044773,
      "grad_norm": 2.522700309753418,
      "learning_rate": 4.192388727943113e-05,
      "loss": 2.0042,
      "step": 6139
    },
    {
      "epoch": 0.16170661048195945,
      "grad_norm": 2.2790493965148926,
      "learning_rate": 4.192257045035555e-05,
      "loss": 2.5038,
      "step": 6140
    },
    {
      "epoch": 0.16173294706347116,
      "grad_norm": 2.1907012462615967,
      "learning_rate": 4.192125362127996e-05,
      "loss": 1.3398,
      "step": 6141
    },
    {
      "epoch": 0.16175928364498288,
      "grad_norm": 1.9467926025390625,
      "learning_rate": 4.191993679220437e-05,
      "loss": 1.9665,
      "step": 6142
    },
    {
      "epoch": 0.1617856202264946,
      "grad_norm": 2.0438179969787598,
      "learning_rate": 4.191861996312879e-05,
      "loss": 1.4199,
      "step": 6143
    },
    {
      "epoch": 0.1618119568080063,
      "grad_norm": 1.715132474899292,
      "learning_rate": 4.1917303134053204e-05,
      "loss": 1.4549,
      "step": 6144
    },
    {
      "epoch": 0.16183829338951805,
      "grad_norm": 1.9151490926742554,
      "learning_rate": 4.191598630497762e-05,
      "loss": 2.2842,
      "step": 6145
    },
    {
      "epoch": 0.16186462997102977,
      "grad_norm": 1.6929700374603271,
      "learning_rate": 4.191466947590203e-05,
      "loss": 1.7138,
      "step": 6146
    },
    {
      "epoch": 0.16189096655254148,
      "grad_norm": 1.975162148475647,
      "learning_rate": 4.1913352646826444e-05,
      "loss": 1.6597,
      "step": 6147
    },
    {
      "epoch": 0.1619173031340532,
      "grad_norm": 4.588531017303467,
      "learning_rate": 4.191203581775085e-05,
      "loss": 1.9448,
      "step": 6148
    },
    {
      "epoch": 0.1619436397155649,
      "grad_norm": 2.539626121520996,
      "learning_rate": 4.1910718988675275e-05,
      "loss": 2.2849,
      "step": 6149
    },
    {
      "epoch": 0.16196997629707663,
      "grad_norm": 1.9681073427200317,
      "learning_rate": 4.1909402159599684e-05,
      "loss": 1.4366,
      "step": 6150
    },
    {
      "epoch": 0.16199631287858837,
      "grad_norm": 4.842199802398682,
      "learning_rate": 4.19080853305241e-05,
      "loss": 1.5735,
      "step": 6151
    },
    {
      "epoch": 0.1620226494601001,
      "grad_norm": 1.9533764123916626,
      "learning_rate": 4.1906768501448515e-05,
      "loss": 1.6544,
      "step": 6152
    },
    {
      "epoch": 0.1620489860416118,
      "grad_norm": 2.1794426441192627,
      "learning_rate": 4.190545167237293e-05,
      "loss": 0.6113,
      "step": 6153
    },
    {
      "epoch": 0.16207532262312352,
      "grad_norm": 3.911369562149048,
      "learning_rate": 4.1904134843297346e-05,
      "loss": 1.2914,
      "step": 6154
    },
    {
      "epoch": 0.16210165920463523,
      "grad_norm": 2.940295696258545,
      "learning_rate": 4.1902818014221755e-05,
      "loss": 1.7576,
      "step": 6155
    },
    {
      "epoch": 0.16212799578614695,
      "grad_norm": 1.6716759204864502,
      "learning_rate": 4.190150118514617e-05,
      "loss": 1.9524,
      "step": 6156
    },
    {
      "epoch": 0.16215433236765867,
      "grad_norm": 1.8164016008377075,
      "learning_rate": 4.190018435607058e-05,
      "loss": 1.902,
      "step": 6157
    },
    {
      "epoch": 0.1621806689491704,
      "grad_norm": 3.107964038848877,
      "learning_rate": 4.1898867526995e-05,
      "loss": 2.0422,
      "step": 6158
    },
    {
      "epoch": 0.16220700553068212,
      "grad_norm": 2.4870853424072266,
      "learning_rate": 4.189755069791941e-05,
      "loss": 0.3071,
      "step": 6159
    },
    {
      "epoch": 0.16223334211219384,
      "grad_norm": 4.143788814544678,
      "learning_rate": 4.1896233868843826e-05,
      "loss": 0.534,
      "step": 6160
    },
    {
      "epoch": 0.16225967869370556,
      "grad_norm": 2.9861960411071777,
      "learning_rate": 4.189491703976824e-05,
      "loss": 1.9037,
      "step": 6161
    },
    {
      "epoch": 0.16228601527521727,
      "grad_norm": 4.6565327644348145,
      "learning_rate": 4.189360021069265e-05,
      "loss": 0.4978,
      "step": 6162
    },
    {
      "epoch": 0.162312351856729,
      "grad_norm": 2.0895419120788574,
      "learning_rate": 4.189228338161707e-05,
      "loss": 1.9403,
      "step": 6163
    },
    {
      "epoch": 0.1623386884382407,
      "grad_norm": 4.719089984893799,
      "learning_rate": 4.189096655254148e-05,
      "loss": 1.3816,
      "step": 6164
    },
    {
      "epoch": 0.16236502501975245,
      "grad_norm": 2.3824193477630615,
      "learning_rate": 4.18896497234659e-05,
      "loss": 0.949,
      "step": 6165
    },
    {
      "epoch": 0.16239136160126416,
      "grad_norm": 1.7241528034210205,
      "learning_rate": 4.1888332894390306e-05,
      "loss": 2.2248,
      "step": 6166
    },
    {
      "epoch": 0.16241769818277588,
      "grad_norm": 4.268033027648926,
      "learning_rate": 4.188701606531472e-05,
      "loss": 1.3197,
      "step": 6167
    },
    {
      "epoch": 0.1624440347642876,
      "grad_norm": 2.030693292617798,
      "learning_rate": 4.188569923623914e-05,
      "loss": 1.9805,
      "step": 6168
    },
    {
      "epoch": 0.1624703713457993,
      "grad_norm": 2.5701515674591064,
      "learning_rate": 4.188438240716355e-05,
      "loss": 1.8422,
      "step": 6169
    },
    {
      "epoch": 0.16249670792731102,
      "grad_norm": 2.068448543548584,
      "learning_rate": 4.188306557808797e-05,
      "loss": 2.0175,
      "step": 6170
    },
    {
      "epoch": 0.16252304450882277,
      "grad_norm": 3.963059663772583,
      "learning_rate": 4.188174874901238e-05,
      "loss": 1.6335,
      "step": 6171
    },
    {
      "epoch": 0.16254938109033448,
      "grad_norm": 1.912456750869751,
      "learning_rate": 4.18804319199368e-05,
      "loss": 1.8062,
      "step": 6172
    },
    {
      "epoch": 0.1625757176718462,
      "grad_norm": 2.8689017295837402,
      "learning_rate": 4.187911509086121e-05,
      "loss": 1.5806,
      "step": 6173
    },
    {
      "epoch": 0.1626020542533579,
      "grad_norm": 3.1292359828948975,
      "learning_rate": 4.1877798261785624e-05,
      "loss": 1.7629,
      "step": 6174
    },
    {
      "epoch": 0.16262839083486963,
      "grad_norm": 2.039140462875366,
      "learning_rate": 4.187648143271003e-05,
      "loss": 1.7487,
      "step": 6175
    },
    {
      "epoch": 0.16265472741638135,
      "grad_norm": 3.5004191398620605,
      "learning_rate": 4.187516460363445e-05,
      "loss": 1.0709,
      "step": 6176
    },
    {
      "epoch": 0.16268106399789306,
      "grad_norm": 2.3860397338867188,
      "learning_rate": 4.1873847774558864e-05,
      "loss": 1.8962,
      "step": 6177
    },
    {
      "epoch": 0.1627074005794048,
      "grad_norm": 2.4505045413970947,
      "learning_rate": 4.187253094548328e-05,
      "loss": 1.6988,
      "step": 6178
    },
    {
      "epoch": 0.16273373716091652,
      "grad_norm": 2.4118874073028564,
      "learning_rate": 4.1871214116407695e-05,
      "loss": 2.0337,
      "step": 6179
    },
    {
      "epoch": 0.16276007374242824,
      "grad_norm": 2.1553492546081543,
      "learning_rate": 4.1869897287332104e-05,
      "loss": 1.7035,
      "step": 6180
    },
    {
      "epoch": 0.16278641032393995,
      "grad_norm": 6.59297513961792,
      "learning_rate": 4.186858045825652e-05,
      "loss": 1.7664,
      "step": 6181
    },
    {
      "epoch": 0.16281274690545167,
      "grad_norm": 3.220435619354248,
      "learning_rate": 4.1867263629180935e-05,
      "loss": 1.2137,
      "step": 6182
    },
    {
      "epoch": 0.16283908348696338,
      "grad_norm": 3.285412073135376,
      "learning_rate": 4.186594680010535e-05,
      "loss": 2.2588,
      "step": 6183
    },
    {
      "epoch": 0.16286542006847513,
      "grad_norm": 2.169034481048584,
      "learning_rate": 4.186462997102976e-05,
      "loss": 1.836,
      "step": 6184
    },
    {
      "epoch": 0.16289175664998684,
      "grad_norm": 12.569430351257324,
      "learning_rate": 4.1863313141954175e-05,
      "loss": 1.3789,
      "step": 6185
    },
    {
      "epoch": 0.16291809323149856,
      "grad_norm": 4.0849080085754395,
      "learning_rate": 4.186199631287859e-05,
      "loss": 1.4266,
      "step": 6186
    },
    {
      "epoch": 0.16294442981301027,
      "grad_norm": 1.7760883569717407,
      "learning_rate": 4.186067948380301e-05,
      "loss": 1.8275,
      "step": 6187
    },
    {
      "epoch": 0.162970766394522,
      "grad_norm": 3.1768648624420166,
      "learning_rate": 4.1859362654727415e-05,
      "loss": 1.6196,
      "step": 6188
    },
    {
      "epoch": 0.1629971029760337,
      "grad_norm": 1.7993084192276,
      "learning_rate": 4.185804582565183e-05,
      "loss": 1.5371,
      "step": 6189
    },
    {
      "epoch": 0.16302343955754542,
      "grad_norm": 2.5547139644622803,
      "learning_rate": 4.185672899657625e-05,
      "loss": 2.1122,
      "step": 6190
    },
    {
      "epoch": 0.16304977613905716,
      "grad_norm": 2.568890333175659,
      "learning_rate": 4.185541216750066e-05,
      "loss": 1.3982,
      "step": 6191
    },
    {
      "epoch": 0.16307611272056888,
      "grad_norm": 1.9501646757125854,
      "learning_rate": 4.185409533842508e-05,
      "loss": 2.7147,
      "step": 6192
    },
    {
      "epoch": 0.1631024493020806,
      "grad_norm": 2.309138536453247,
      "learning_rate": 4.185277850934949e-05,
      "loss": 1.1929,
      "step": 6193
    },
    {
      "epoch": 0.1631287858835923,
      "grad_norm": 3.3782432079315186,
      "learning_rate": 4.18514616802739e-05,
      "loss": 1.6183,
      "step": 6194
    },
    {
      "epoch": 0.16315512246510402,
      "grad_norm": 1.8713680505752563,
      "learning_rate": 4.185014485119831e-05,
      "loss": 1.2001,
      "step": 6195
    },
    {
      "epoch": 0.16318145904661574,
      "grad_norm": 9.779690742492676,
      "learning_rate": 4.1848828022122733e-05,
      "loss": 1.5312,
      "step": 6196
    },
    {
      "epoch": 0.16320779562812746,
      "grad_norm": 2.7976479530334473,
      "learning_rate": 4.184751119304714e-05,
      "loss": 1.6511,
      "step": 6197
    },
    {
      "epoch": 0.1632341322096392,
      "grad_norm": 2.537200927734375,
      "learning_rate": 4.184619436397156e-05,
      "loss": 1.8137,
      "step": 6198
    },
    {
      "epoch": 0.16326046879115091,
      "grad_norm": 1.6719151735305786,
      "learning_rate": 4.1844877534895973e-05,
      "loss": 1.2775,
      "step": 6199
    },
    {
      "epoch": 0.16328680537266263,
      "grad_norm": 1.9692857265472412,
      "learning_rate": 4.184356070582038e-05,
      "loss": 2.1277,
      "step": 6200
    },
    {
      "epoch": 0.16331314195417435,
      "grad_norm": 1.7134268283843994,
      "learning_rate": 4.1842243876744805e-05,
      "loss": 2.4531,
      "step": 6201
    },
    {
      "epoch": 0.16333947853568606,
      "grad_norm": 1.583983063697815,
      "learning_rate": 4.1840927047669214e-05,
      "loss": 1.805,
      "step": 6202
    },
    {
      "epoch": 0.16336581511719778,
      "grad_norm": 3.674165725708008,
      "learning_rate": 4.183961021859363e-05,
      "loss": 0.8692,
      "step": 6203
    },
    {
      "epoch": 0.16339215169870952,
      "grad_norm": 5.6180620193481445,
      "learning_rate": 4.183829338951804e-05,
      "loss": 2.1638,
      "step": 6204
    },
    {
      "epoch": 0.16341848828022124,
      "grad_norm": 2.1328909397125244,
      "learning_rate": 4.183697656044246e-05,
      "loss": 1.9604,
      "step": 6205
    },
    {
      "epoch": 0.16344482486173295,
      "grad_norm": 2.927257537841797,
      "learning_rate": 4.183565973136687e-05,
      "loss": 1.3435,
      "step": 6206
    },
    {
      "epoch": 0.16347116144324467,
      "grad_norm": 1.9658448696136475,
      "learning_rate": 4.1834342902291285e-05,
      "loss": 1.912,
      "step": 6207
    },
    {
      "epoch": 0.16349749802475638,
      "grad_norm": 2.342245101928711,
      "learning_rate": 4.18330260732157e-05,
      "loss": 1.5374,
      "step": 6208
    },
    {
      "epoch": 0.1635238346062681,
      "grad_norm": 7.295834541320801,
      "learning_rate": 4.183170924414011e-05,
      "loss": 1.7454,
      "step": 6209
    },
    {
      "epoch": 0.1635501711877798,
      "grad_norm": 3.025780200958252,
      "learning_rate": 4.183039241506453e-05,
      "loss": 0.7368,
      "step": 6210
    },
    {
      "epoch": 0.16357650776929156,
      "grad_norm": 1.6089125871658325,
      "learning_rate": 4.182907558598894e-05,
      "loss": 1.0085,
      "step": 6211
    },
    {
      "epoch": 0.16360284435080327,
      "grad_norm": 1.572649359703064,
      "learning_rate": 4.1827758756913356e-05,
      "loss": 1.7478,
      "step": 6212
    },
    {
      "epoch": 0.163629180932315,
      "grad_norm": 2.767467737197876,
      "learning_rate": 4.1826441927837765e-05,
      "loss": 1.7427,
      "step": 6213
    },
    {
      "epoch": 0.1636555175138267,
      "grad_norm": 1.5712791681289673,
      "learning_rate": 4.182512509876218e-05,
      "loss": 0.3449,
      "step": 6214
    },
    {
      "epoch": 0.16368185409533842,
      "grad_norm": 2.7520124912261963,
      "learning_rate": 4.1823808269686596e-05,
      "loss": 0.5863,
      "step": 6215
    },
    {
      "epoch": 0.16370819067685013,
      "grad_norm": 2.7478747367858887,
      "learning_rate": 4.182249144061101e-05,
      "loss": 1.4459,
      "step": 6216
    },
    {
      "epoch": 0.16373452725836188,
      "grad_norm": 4.27979850769043,
      "learning_rate": 4.182117461153543e-05,
      "loss": 3.5372,
      "step": 6217
    },
    {
      "epoch": 0.1637608638398736,
      "grad_norm": 1.6363543272018433,
      "learning_rate": 4.1819857782459836e-05,
      "loss": 2.1614,
      "step": 6218
    },
    {
      "epoch": 0.1637872004213853,
      "grad_norm": 2.104304552078247,
      "learning_rate": 4.181854095338426e-05,
      "loss": 1.7553,
      "step": 6219
    },
    {
      "epoch": 0.16381353700289703,
      "grad_norm": 1.8212826251983643,
      "learning_rate": 4.181722412430867e-05,
      "loss": 1.9183,
      "step": 6220
    },
    {
      "epoch": 0.16383987358440874,
      "grad_norm": 4.419536113739014,
      "learning_rate": 4.181590729523308e-05,
      "loss": 1.2457,
      "step": 6221
    },
    {
      "epoch": 0.16386621016592046,
      "grad_norm": 3.3643383979797363,
      "learning_rate": 4.181459046615749e-05,
      "loss": 2.2115,
      "step": 6222
    },
    {
      "epoch": 0.16389254674743217,
      "grad_norm": 3.484835624694824,
      "learning_rate": 4.181327363708191e-05,
      "loss": 2.1387,
      "step": 6223
    },
    {
      "epoch": 0.16391888332894392,
      "grad_norm": 1.7184138298034668,
      "learning_rate": 4.181195680800632e-05,
      "loss": 1.5123,
      "step": 6224
    },
    {
      "epoch": 0.16394521991045563,
      "grad_norm": 1.5833100080490112,
      "learning_rate": 4.181063997893074e-05,
      "loss": 1.7886,
      "step": 6225
    },
    {
      "epoch": 0.16397155649196735,
      "grad_norm": 2.703047752380371,
      "learning_rate": 4.1809323149855154e-05,
      "loss": 1.4228,
      "step": 6226
    },
    {
      "epoch": 0.16399789307347906,
      "grad_norm": 2.10036563873291,
      "learning_rate": 4.180800632077956e-05,
      "loss": 2.0066,
      "step": 6227
    },
    {
      "epoch": 0.16402422965499078,
      "grad_norm": 2.0519776344299316,
      "learning_rate": 4.180668949170398e-05,
      "loss": 2.0121,
      "step": 6228
    },
    {
      "epoch": 0.1640505662365025,
      "grad_norm": 2.7119178771972656,
      "learning_rate": 4.1805372662628394e-05,
      "loss": 1.6072,
      "step": 6229
    },
    {
      "epoch": 0.1640769028180142,
      "grad_norm": 2.0890233516693115,
      "learning_rate": 4.180405583355281e-05,
      "loss": 1.1275,
      "step": 6230
    },
    {
      "epoch": 0.16410323939952595,
      "grad_norm": 3.6675961017608643,
      "learning_rate": 4.180273900447722e-05,
      "loss": 1.4267,
      "step": 6231
    },
    {
      "epoch": 0.16412957598103767,
      "grad_norm": 1.7845052480697632,
      "learning_rate": 4.1801422175401634e-05,
      "loss": 2.0269,
      "step": 6232
    },
    {
      "epoch": 0.16415591256254938,
      "grad_norm": 3.4579153060913086,
      "learning_rate": 4.180010534632604e-05,
      "loss": 2.2869,
      "step": 6233
    },
    {
      "epoch": 0.1641822491440611,
      "grad_norm": 2.1843247413635254,
      "learning_rate": 4.1798788517250465e-05,
      "loss": 2.0065,
      "step": 6234
    },
    {
      "epoch": 0.16420858572557281,
      "grad_norm": 1.7261897325515747,
      "learning_rate": 4.1797471688174874e-05,
      "loss": 0.9492,
      "step": 6235
    },
    {
      "epoch": 0.16423492230708453,
      "grad_norm": 5.073116302490234,
      "learning_rate": 4.179615485909929e-05,
      "loss": 0.5242,
      "step": 6236
    },
    {
      "epoch": 0.16426125888859627,
      "grad_norm": 4.026960849761963,
      "learning_rate": 4.1794838030023705e-05,
      "loss": 0.598,
      "step": 6237
    },
    {
      "epoch": 0.164287595470108,
      "grad_norm": 2.2366933822631836,
      "learning_rate": 4.179352120094812e-05,
      "loss": 1.9456,
      "step": 6238
    },
    {
      "epoch": 0.1643139320516197,
      "grad_norm": 1.8947323560714722,
      "learning_rate": 4.1792204371872536e-05,
      "loss": 1.6567,
      "step": 6239
    },
    {
      "epoch": 0.16434026863313142,
      "grad_norm": 1.7033538818359375,
      "learning_rate": 4.1790887542796945e-05,
      "loss": 1.9291,
      "step": 6240
    },
    {
      "epoch": 0.16436660521464314,
      "grad_norm": 1.8342688083648682,
      "learning_rate": 4.178957071372136e-05,
      "loss": 1.8968,
      "step": 6241
    },
    {
      "epoch": 0.16439294179615485,
      "grad_norm": 2.2728381156921387,
      "learning_rate": 4.178825388464577e-05,
      "loss": 1.5177,
      "step": 6242
    },
    {
      "epoch": 0.16441927837766657,
      "grad_norm": 2.142536163330078,
      "learning_rate": 4.178693705557019e-05,
      "loss": 1.4954,
      "step": 6243
    },
    {
      "epoch": 0.1644456149591783,
      "grad_norm": 1.87212073802948,
      "learning_rate": 4.17856202264946e-05,
      "loss": 0.2432,
      "step": 6244
    },
    {
      "epoch": 0.16447195154069003,
      "grad_norm": 1.8364509344100952,
      "learning_rate": 4.1784303397419016e-05,
      "loss": 1.779,
      "step": 6245
    },
    {
      "epoch": 0.16449828812220174,
      "grad_norm": 2.9865927696228027,
      "learning_rate": 4.178298656834343e-05,
      "loss": 1.2728,
      "step": 6246
    },
    {
      "epoch": 0.16452462470371346,
      "grad_norm": 1.7972159385681152,
      "learning_rate": 4.178166973926784e-05,
      "loss": 2.1082,
      "step": 6247
    },
    {
      "epoch": 0.16455096128522517,
      "grad_norm": 3.559039831161499,
      "learning_rate": 4.178035291019226e-05,
      "loss": 1.7195,
      "step": 6248
    },
    {
      "epoch": 0.1645772978667369,
      "grad_norm": 1.6659404039382935,
      "learning_rate": 4.177903608111667e-05,
      "loss": 1.3815,
      "step": 6249
    },
    {
      "epoch": 0.1646036344482486,
      "grad_norm": 2.0792975425720215,
      "learning_rate": 4.177771925204109e-05,
      "loss": 1.7125,
      "step": 6250
    },
    {
      "epoch": 0.16462997102976035,
      "grad_norm": 2.2669241428375244,
      "learning_rate": 4.1776402422965496e-05,
      "loss": 1.4363,
      "step": 6251
    },
    {
      "epoch": 0.16465630761127206,
      "grad_norm": 2.20127534866333,
      "learning_rate": 4.177508559388992e-05,
      "loss": 1.7931,
      "step": 6252
    },
    {
      "epoch": 0.16468264419278378,
      "grad_norm": 2.8577163219451904,
      "learning_rate": 4.177376876481433e-05,
      "loss": 1.613,
      "step": 6253
    },
    {
      "epoch": 0.1647089807742955,
      "grad_norm": 2.5715982913970947,
      "learning_rate": 4.177245193573874e-05,
      "loss": 1.5152,
      "step": 6254
    },
    {
      "epoch": 0.1647353173558072,
      "grad_norm": 3.654378890991211,
      "learning_rate": 4.177113510666316e-05,
      "loss": 1.9085,
      "step": 6255
    },
    {
      "epoch": 0.16476165393731892,
      "grad_norm": 1.9232184886932373,
      "learning_rate": 4.176981827758757e-05,
      "loss": 1.8253,
      "step": 6256
    },
    {
      "epoch": 0.16478799051883067,
      "grad_norm": 1.6215038299560547,
      "learning_rate": 4.176850144851199e-05,
      "loss": 2.5152,
      "step": 6257
    },
    {
      "epoch": 0.16481432710034238,
      "grad_norm": 2.3463377952575684,
      "learning_rate": 4.17671846194364e-05,
      "loss": 0.9331,
      "step": 6258
    },
    {
      "epoch": 0.1648406636818541,
      "grad_norm": 1.8195308446884155,
      "learning_rate": 4.1765867790360814e-05,
      "loss": 2.0912,
      "step": 6259
    },
    {
      "epoch": 0.16486700026336581,
      "grad_norm": 2.8469274044036865,
      "learning_rate": 4.176455096128522e-05,
      "loss": 1.3566,
      "step": 6260
    },
    {
      "epoch": 0.16489333684487753,
      "grad_norm": 2.154100179672241,
      "learning_rate": 4.176323413220964e-05,
      "loss": 2.2292,
      "step": 6261
    },
    {
      "epoch": 0.16491967342638925,
      "grad_norm": 1.9302618503570557,
      "learning_rate": 4.1761917303134054e-05,
      "loss": 1.8194,
      "step": 6262
    },
    {
      "epoch": 0.16494601000790096,
      "grad_norm": 2.492159843444824,
      "learning_rate": 4.176060047405847e-05,
      "loss": 1.3806,
      "step": 6263
    },
    {
      "epoch": 0.1649723465894127,
      "grad_norm": 4.924745559692383,
      "learning_rate": 4.1759283644982886e-05,
      "loss": 1.7607,
      "step": 6264
    },
    {
      "epoch": 0.16499868317092442,
      "grad_norm": 1.7426139116287231,
      "learning_rate": 4.1757966815907295e-05,
      "loss": 1.577,
      "step": 6265
    },
    {
      "epoch": 0.16502501975243614,
      "grad_norm": 2.427046775817871,
      "learning_rate": 4.175664998683171e-05,
      "loss": 2.5492,
      "step": 6266
    },
    {
      "epoch": 0.16505135633394785,
      "grad_norm": 1.8286429643630981,
      "learning_rate": 4.1755333157756126e-05,
      "loss": 2.2028,
      "step": 6267
    },
    {
      "epoch": 0.16507769291545957,
      "grad_norm": 2.536653757095337,
      "learning_rate": 4.175401632868054e-05,
      "loss": 0.9838,
      "step": 6268
    },
    {
      "epoch": 0.16510402949697128,
      "grad_norm": 2.0288827419281006,
      "learning_rate": 4.175269949960495e-05,
      "loss": 2.4019,
      "step": 6269
    },
    {
      "epoch": 0.16513036607848303,
      "grad_norm": 1.8791232109069824,
      "learning_rate": 4.1751382670529366e-05,
      "loss": 2.0111,
      "step": 6270
    },
    {
      "epoch": 0.16515670265999474,
      "grad_norm": 2.0149269104003906,
      "learning_rate": 4.175006584145378e-05,
      "loss": 1.8392,
      "step": 6271
    },
    {
      "epoch": 0.16518303924150646,
      "grad_norm": 1.7191165685653687,
      "learning_rate": 4.17487490123782e-05,
      "loss": 2.2062,
      "step": 6272
    },
    {
      "epoch": 0.16520937582301817,
      "grad_norm": 1.7985368967056274,
      "learning_rate": 4.174743218330261e-05,
      "loss": 1.9149,
      "step": 6273
    },
    {
      "epoch": 0.1652357124045299,
      "grad_norm": 2.5112264156341553,
      "learning_rate": 4.174611535422702e-05,
      "loss": 1.9655,
      "step": 6274
    },
    {
      "epoch": 0.1652620489860416,
      "grad_norm": 5.3558878898620605,
      "learning_rate": 4.174479852515144e-05,
      "loss": 1.9623,
      "step": 6275
    },
    {
      "epoch": 0.16528838556755332,
      "grad_norm": 3.8078792095184326,
      "learning_rate": 4.174348169607585e-05,
      "loss": 0.6285,
      "step": 6276
    },
    {
      "epoch": 0.16531472214906506,
      "grad_norm": 4.321228504180908,
      "learning_rate": 4.174216486700027e-05,
      "loss": 0.8884,
      "step": 6277
    },
    {
      "epoch": 0.16534105873057678,
      "grad_norm": 2.3482091426849365,
      "learning_rate": 4.174084803792468e-05,
      "loss": 2.2635,
      "step": 6278
    },
    {
      "epoch": 0.1653673953120885,
      "grad_norm": 2.1774532794952393,
      "learning_rate": 4.173953120884909e-05,
      "loss": 2.1163,
      "step": 6279
    },
    {
      "epoch": 0.1653937318936002,
      "grad_norm": 2.4843313694000244,
      "learning_rate": 4.173821437977351e-05,
      "loss": 1.7013,
      "step": 6280
    },
    {
      "epoch": 0.16542006847511193,
      "grad_norm": 1.821319341659546,
      "learning_rate": 4.1736897550697924e-05,
      "loss": 2.3493,
      "step": 6281
    },
    {
      "epoch": 0.16544640505662364,
      "grad_norm": 2.395052194595337,
      "learning_rate": 4.173558072162234e-05,
      "loss": 1.9074,
      "step": 6282
    },
    {
      "epoch": 0.16547274163813536,
      "grad_norm": 1.82941472530365,
      "learning_rate": 4.173426389254675e-05,
      "loss": 2.0776,
      "step": 6283
    },
    {
      "epoch": 0.1654990782196471,
      "grad_norm": 2.0111989974975586,
      "learning_rate": 4.1732947063471164e-05,
      "loss": 1.2991,
      "step": 6284
    },
    {
      "epoch": 0.16552541480115882,
      "grad_norm": 2.0000054836273193,
      "learning_rate": 4.173163023439558e-05,
      "loss": 1.8515,
      "step": 6285
    },
    {
      "epoch": 0.16555175138267053,
      "grad_norm": 2.083451986312866,
      "learning_rate": 4.1730313405319995e-05,
      "loss": 1.7277,
      "step": 6286
    },
    {
      "epoch": 0.16557808796418225,
      "grad_norm": 1.7810637950897217,
      "learning_rate": 4.1728996576244404e-05,
      "loss": 1.7744,
      "step": 6287
    },
    {
      "epoch": 0.16560442454569396,
      "grad_norm": 2.8143627643585205,
      "learning_rate": 4.172767974716882e-05,
      "loss": 0.6735,
      "step": 6288
    },
    {
      "epoch": 0.16563076112720568,
      "grad_norm": 3.857470750808716,
      "learning_rate": 4.172636291809323e-05,
      "loss": 1.1304,
      "step": 6289
    },
    {
      "epoch": 0.16565709770871742,
      "grad_norm": 2.2950751781463623,
      "learning_rate": 4.172504608901765e-05,
      "loss": 0.4887,
      "step": 6290
    },
    {
      "epoch": 0.16568343429022914,
      "grad_norm": 6.222325325012207,
      "learning_rate": 4.172372925994206e-05,
      "loss": 1.5784,
      "step": 6291
    },
    {
      "epoch": 0.16570977087174085,
      "grad_norm": 2.973029375076294,
      "learning_rate": 4.1722412430866475e-05,
      "loss": 1.1543,
      "step": 6292
    },
    {
      "epoch": 0.16573610745325257,
      "grad_norm": 2.169588565826416,
      "learning_rate": 4.172109560179089e-05,
      "loss": 2.4109,
      "step": 6293
    },
    {
      "epoch": 0.16576244403476428,
      "grad_norm": 1.5478515625,
      "learning_rate": 4.17197787727153e-05,
      "loss": 1.7889,
      "step": 6294
    },
    {
      "epoch": 0.165788780616276,
      "grad_norm": 3.2532997131347656,
      "learning_rate": 4.171846194363972e-05,
      "loss": 0.5807,
      "step": 6295
    },
    {
      "epoch": 0.16581511719778771,
      "grad_norm": 4.012513637542725,
      "learning_rate": 4.171714511456413e-05,
      "loss": 0.7988,
      "step": 6296
    },
    {
      "epoch": 0.16584145377929946,
      "grad_norm": 2.912229537963867,
      "learning_rate": 4.1715828285488546e-05,
      "loss": 0.8654,
      "step": 6297
    },
    {
      "epoch": 0.16586779036081117,
      "grad_norm": 2.3448755741119385,
      "learning_rate": 4.1714511456412955e-05,
      "loss": 1.5029,
      "step": 6298
    },
    {
      "epoch": 0.1658941269423229,
      "grad_norm": 2.821657180786133,
      "learning_rate": 4.171319462733738e-05,
      "loss": 1.6197,
      "step": 6299
    },
    {
      "epoch": 0.1659204635238346,
      "grad_norm": 3.6999006271362305,
      "learning_rate": 4.1711877798261786e-05,
      "loss": 1.5829,
      "step": 6300
    },
    {
      "epoch": 0.16594680010534632,
      "grad_norm": 1.7603660821914673,
      "learning_rate": 4.17105609691862e-05,
      "loss": 1.8102,
      "step": 6301
    },
    {
      "epoch": 0.16597313668685804,
      "grad_norm": 1.6552979946136475,
      "learning_rate": 4.170924414011062e-05,
      "loss": 2.0413,
      "step": 6302
    },
    {
      "epoch": 0.16599947326836978,
      "grad_norm": 4.111746311187744,
      "learning_rate": 4.1707927311035026e-05,
      "loss": 1.0443,
      "step": 6303
    },
    {
      "epoch": 0.1660258098498815,
      "grad_norm": 6.605031967163086,
      "learning_rate": 4.170661048195945e-05,
      "loss": 0.6922,
      "step": 6304
    },
    {
      "epoch": 0.1660521464313932,
      "grad_norm": 2.572258472442627,
      "learning_rate": 4.170529365288386e-05,
      "loss": 1.4043,
      "step": 6305
    },
    {
      "epoch": 0.16607848301290493,
      "grad_norm": 2.5222651958465576,
      "learning_rate": 4.170397682380827e-05,
      "loss": 1.6599,
      "step": 6306
    },
    {
      "epoch": 0.16610481959441664,
      "grad_norm": 2.052239418029785,
      "learning_rate": 4.170265999473268e-05,
      "loss": 1.5124,
      "step": 6307
    },
    {
      "epoch": 0.16613115617592836,
      "grad_norm": 2.027329444885254,
      "learning_rate": 4.17013431656571e-05,
      "loss": 1.193,
      "step": 6308
    },
    {
      "epoch": 0.16615749275744007,
      "grad_norm": 3.2304461002349854,
      "learning_rate": 4.170002633658151e-05,
      "loss": 1.5395,
      "step": 6309
    },
    {
      "epoch": 0.16618382933895182,
      "grad_norm": 3.3135313987731934,
      "learning_rate": 4.169870950750593e-05,
      "loss": 1.7107,
      "step": 6310
    },
    {
      "epoch": 0.16621016592046353,
      "grad_norm": 1.9948111772537231,
      "learning_rate": 4.1697392678430344e-05,
      "loss": 1.3134,
      "step": 6311
    },
    {
      "epoch": 0.16623650250197525,
      "grad_norm": 2.704598903656006,
      "learning_rate": 4.169607584935475e-05,
      "loss": 0.681,
      "step": 6312
    },
    {
      "epoch": 0.16626283908348696,
      "grad_norm": 2.8368823528289795,
      "learning_rate": 4.169475902027917e-05,
      "loss": 1.1584,
      "step": 6313
    },
    {
      "epoch": 0.16628917566499868,
      "grad_norm": 2.1248085498809814,
      "learning_rate": 4.1693442191203584e-05,
      "loss": 1.5046,
      "step": 6314
    },
    {
      "epoch": 0.1663155122465104,
      "grad_norm": 3.1605188846588135,
      "learning_rate": 4.1692125362128e-05,
      "loss": 0.5833,
      "step": 6315
    },
    {
      "epoch": 0.1663418488280221,
      "grad_norm": 2.2313647270202637,
      "learning_rate": 4.169080853305241e-05,
      "loss": 1.135,
      "step": 6316
    },
    {
      "epoch": 0.16636818540953385,
      "grad_norm": 2.5924830436706543,
      "learning_rate": 4.1689491703976824e-05,
      "loss": 1.5192,
      "step": 6317
    },
    {
      "epoch": 0.16639452199104557,
      "grad_norm": 4.161013126373291,
      "learning_rate": 4.168817487490124e-05,
      "loss": 2.529,
      "step": 6318
    },
    {
      "epoch": 0.16642085857255728,
      "grad_norm": 2.3811326026916504,
      "learning_rate": 4.1686858045825655e-05,
      "loss": 2.3211,
      "step": 6319
    },
    {
      "epoch": 0.166447195154069,
      "grad_norm": 3.1531453132629395,
      "learning_rate": 4.168554121675007e-05,
      "loss": 1.3661,
      "step": 6320
    },
    {
      "epoch": 0.16647353173558072,
      "grad_norm": 2.6829094886779785,
      "learning_rate": 4.168422438767448e-05,
      "loss": 1.6922,
      "step": 6321
    },
    {
      "epoch": 0.16649986831709243,
      "grad_norm": 2.0825557708740234,
      "learning_rate": 4.1682907558598895e-05,
      "loss": 1.755,
      "step": 6322
    },
    {
      "epoch": 0.16652620489860417,
      "grad_norm": 1.9286105632781982,
      "learning_rate": 4.168159072952331e-05,
      "loss": 1.583,
      "step": 6323
    },
    {
      "epoch": 0.1665525414801159,
      "grad_norm": 3.6238157749176025,
      "learning_rate": 4.168027390044773e-05,
      "loss": 1.9119,
      "step": 6324
    },
    {
      "epoch": 0.1665788780616276,
      "grad_norm": 1.9939560890197754,
      "learning_rate": 4.1678957071372136e-05,
      "loss": 0.7373,
      "step": 6325
    },
    {
      "epoch": 0.16660521464313932,
      "grad_norm": 2.046525716781616,
      "learning_rate": 4.167764024229655e-05,
      "loss": 1.7242,
      "step": 6326
    },
    {
      "epoch": 0.16663155122465104,
      "grad_norm": 3.0568177700042725,
      "learning_rate": 4.167632341322097e-05,
      "loss": 2.2206,
      "step": 6327
    },
    {
      "epoch": 0.16665788780616275,
      "grad_norm": 3.57637095451355,
      "learning_rate": 4.167500658414538e-05,
      "loss": 1.6992,
      "step": 6328
    },
    {
      "epoch": 0.16668422438767447,
      "grad_norm": 1.8986932039260864,
      "learning_rate": 4.16736897550698e-05,
      "loss": 1.8675,
      "step": 6329
    },
    {
      "epoch": 0.1667105609691862,
      "grad_norm": 1.8276662826538086,
      "learning_rate": 4.167237292599421e-05,
      "loss": 1.5032,
      "step": 6330
    },
    {
      "epoch": 0.16673689755069793,
      "grad_norm": 2.2076597213745117,
      "learning_rate": 4.167105609691862e-05,
      "loss": 1.5994,
      "step": 6331
    },
    {
      "epoch": 0.16676323413220964,
      "grad_norm": 2.1224892139434814,
      "learning_rate": 4.166973926784304e-05,
      "loss": 2.0706,
      "step": 6332
    },
    {
      "epoch": 0.16678957071372136,
      "grad_norm": 3.413240671157837,
      "learning_rate": 4.1668422438767454e-05,
      "loss": 0.8709,
      "step": 6333
    },
    {
      "epoch": 0.16681590729523307,
      "grad_norm": 1.8400017023086548,
      "learning_rate": 4.166710560969186e-05,
      "loss": 1.4148,
      "step": 6334
    },
    {
      "epoch": 0.1668422438767448,
      "grad_norm": 1.775442361831665,
      "learning_rate": 4.166578878061628e-05,
      "loss": 1.7614,
      "step": 6335
    },
    {
      "epoch": 0.1668685804582565,
      "grad_norm": 1.7555932998657227,
      "learning_rate": 4.166447195154069e-05,
      "loss": 1.9382,
      "step": 6336
    },
    {
      "epoch": 0.16689491703976825,
      "grad_norm": 3.6088674068450928,
      "learning_rate": 4.166315512246511e-05,
      "loss": 1.4609,
      "step": 6337
    },
    {
      "epoch": 0.16692125362127996,
      "grad_norm": 2.4535605907440186,
      "learning_rate": 4.166183829338952e-05,
      "loss": 1.9765,
      "step": 6338
    },
    {
      "epoch": 0.16694759020279168,
      "grad_norm": 2.234781503677368,
      "learning_rate": 4.1660521464313934e-05,
      "loss": 2.0701,
      "step": 6339
    },
    {
      "epoch": 0.1669739267843034,
      "grad_norm": 2.3924405574798584,
      "learning_rate": 4.165920463523835e-05,
      "loss": 1.7917,
      "step": 6340
    },
    {
      "epoch": 0.1670002633658151,
      "grad_norm": 2.171189785003662,
      "learning_rate": 4.165788780616276e-05,
      "loss": 1.9543,
      "step": 6341
    },
    {
      "epoch": 0.16702659994732683,
      "grad_norm": 2.223950147628784,
      "learning_rate": 4.165657097708718e-05,
      "loss": 1.3898,
      "step": 6342
    },
    {
      "epoch": 0.16705293652883857,
      "grad_norm": 1.7114344835281372,
      "learning_rate": 4.165525414801159e-05,
      "loss": 1.0963,
      "step": 6343
    },
    {
      "epoch": 0.16707927311035028,
      "grad_norm": 2.054962635040283,
      "learning_rate": 4.1653937318936005e-05,
      "loss": 1.5417,
      "step": 6344
    },
    {
      "epoch": 0.167105609691862,
      "grad_norm": 1.5714259147644043,
      "learning_rate": 4.1652620489860414e-05,
      "loss": 1.8038,
      "step": 6345
    },
    {
      "epoch": 0.16713194627337372,
      "grad_norm": 3.3276352882385254,
      "learning_rate": 4.165130366078483e-05,
      "loss": 1.6062,
      "step": 6346
    },
    {
      "epoch": 0.16715828285488543,
      "grad_norm": 2.267390727996826,
      "learning_rate": 4.1649986831709245e-05,
      "loss": 1.6488,
      "step": 6347
    },
    {
      "epoch": 0.16718461943639715,
      "grad_norm": 3.921799898147583,
      "learning_rate": 4.164867000263366e-05,
      "loss": 1.8439,
      "step": 6348
    },
    {
      "epoch": 0.16721095601790886,
      "grad_norm": 1.8757134675979614,
      "learning_rate": 4.1647353173558076e-05,
      "loss": 1.5935,
      "step": 6349
    },
    {
      "epoch": 0.1672372925994206,
      "grad_norm": 2.898759365081787,
      "learning_rate": 4.1646036344482485e-05,
      "loss": 1.9947,
      "step": 6350
    },
    {
      "epoch": 0.16726362918093232,
      "grad_norm": 1.6849979162216187,
      "learning_rate": 4.164471951540691e-05,
      "loss": 1.5262,
      "step": 6351
    },
    {
      "epoch": 0.16728996576244404,
      "grad_norm": 2.1206817626953125,
      "learning_rate": 4.1643402686331316e-05,
      "loss": 1.5047,
      "step": 6352
    },
    {
      "epoch": 0.16731630234395575,
      "grad_norm": 2.3731672763824463,
      "learning_rate": 4.164208585725573e-05,
      "loss": 0.4956,
      "step": 6353
    },
    {
      "epoch": 0.16734263892546747,
      "grad_norm": 2.3113131523132324,
      "learning_rate": 4.164076902818014e-05,
      "loss": 1.8525,
      "step": 6354
    },
    {
      "epoch": 0.16736897550697918,
      "grad_norm": 1.9033572673797607,
      "learning_rate": 4.1639452199104556e-05,
      "loss": 2.2655,
      "step": 6355
    },
    {
      "epoch": 0.16739531208849093,
      "grad_norm": 9.518451690673828,
      "learning_rate": 4.163813537002897e-05,
      "loss": 2.1358,
      "step": 6356
    },
    {
      "epoch": 0.16742164867000264,
      "grad_norm": 2.6317458152770996,
      "learning_rate": 4.163681854095339e-05,
      "loss": 1.5336,
      "step": 6357
    },
    {
      "epoch": 0.16744798525151436,
      "grad_norm": 2.8073017597198486,
      "learning_rate": 4.16355017118778e-05,
      "loss": 2.2666,
      "step": 6358
    },
    {
      "epoch": 0.16747432183302607,
      "grad_norm": 2.4643218517303467,
      "learning_rate": 4.163418488280221e-05,
      "loss": 1.7047,
      "step": 6359
    },
    {
      "epoch": 0.1675006584145378,
      "grad_norm": 4.039646625518799,
      "learning_rate": 4.163286805372663e-05,
      "loss": 0.8799,
      "step": 6360
    },
    {
      "epoch": 0.1675269949960495,
      "grad_norm": 5.677095413208008,
      "learning_rate": 4.163155122465104e-05,
      "loss": 2.553,
      "step": 6361
    },
    {
      "epoch": 0.16755333157756122,
      "grad_norm": 1.8579610586166382,
      "learning_rate": 4.163023439557546e-05,
      "loss": 1.3293,
      "step": 6362
    },
    {
      "epoch": 0.16757966815907296,
      "grad_norm": 3.0890026092529297,
      "learning_rate": 4.162891756649987e-05,
      "loss": 1.5316,
      "step": 6363
    },
    {
      "epoch": 0.16760600474058468,
      "grad_norm": 2.2251076698303223,
      "learning_rate": 4.162760073742428e-05,
      "loss": 1.4928,
      "step": 6364
    },
    {
      "epoch": 0.1676323413220964,
      "grad_norm": 1.745855450630188,
      "learning_rate": 4.16262839083487e-05,
      "loss": 1.8408,
      "step": 6365
    },
    {
      "epoch": 0.1676586779036081,
      "grad_norm": 2.259474277496338,
      "learning_rate": 4.1624967079273114e-05,
      "loss": 1.9137,
      "step": 6366
    },
    {
      "epoch": 0.16768501448511983,
      "grad_norm": 2.1383891105651855,
      "learning_rate": 4.162365025019753e-05,
      "loss": 2.0164,
      "step": 6367
    },
    {
      "epoch": 0.16771135106663154,
      "grad_norm": 3.731020927429199,
      "learning_rate": 4.162233342112194e-05,
      "loss": 0.6502,
      "step": 6368
    },
    {
      "epoch": 0.16773768764814326,
      "grad_norm": 6.15726900100708,
      "learning_rate": 4.1621016592046354e-05,
      "loss": 1.616,
      "step": 6369
    },
    {
      "epoch": 0.167764024229655,
      "grad_norm": 1.8864214420318604,
      "learning_rate": 4.161969976297077e-05,
      "loss": 1.562,
      "step": 6370
    },
    {
      "epoch": 0.16779036081116672,
      "grad_norm": 1.9158926010131836,
      "learning_rate": 4.1618382933895185e-05,
      "loss": 1.7563,
      "step": 6371
    },
    {
      "epoch": 0.16781669739267843,
      "grad_norm": 2.8187687397003174,
      "learning_rate": 4.1617066104819594e-05,
      "loss": 1.344,
      "step": 6372
    },
    {
      "epoch": 0.16784303397419015,
      "grad_norm": 1.8098351955413818,
      "learning_rate": 4.161574927574401e-05,
      "loss": 1.4523,
      "step": 6373
    },
    {
      "epoch": 0.16786937055570186,
      "grad_norm": 2.0661325454711914,
      "learning_rate": 4.1614432446668425e-05,
      "loss": 2.0314,
      "step": 6374
    },
    {
      "epoch": 0.16789570713721358,
      "grad_norm": 4.071370601654053,
      "learning_rate": 4.161311561759284e-05,
      "loss": 1.4825,
      "step": 6375
    },
    {
      "epoch": 0.16792204371872532,
      "grad_norm": 4.503345012664795,
      "learning_rate": 4.1611798788517256e-05,
      "loss": 0.865,
      "step": 6376
    },
    {
      "epoch": 0.16794838030023704,
      "grad_norm": 2.2279651165008545,
      "learning_rate": 4.1610481959441665e-05,
      "loss": 1.8657,
      "step": 6377
    },
    {
      "epoch": 0.16797471688174875,
      "grad_norm": 5.581786632537842,
      "learning_rate": 4.160916513036608e-05,
      "loss": 1.9477,
      "step": 6378
    },
    {
      "epoch": 0.16800105346326047,
      "grad_norm": 2.2516820430755615,
      "learning_rate": 4.160784830129049e-05,
      "loss": 1.1118,
      "step": 6379
    },
    {
      "epoch": 0.16802739004477218,
      "grad_norm": 2.087665319442749,
      "learning_rate": 4.160653147221491e-05,
      "loss": 1.9211,
      "step": 6380
    },
    {
      "epoch": 0.1680537266262839,
      "grad_norm": 3.2751336097717285,
      "learning_rate": 4.160521464313932e-05,
      "loss": 2.434,
      "step": 6381
    },
    {
      "epoch": 0.16808006320779562,
      "grad_norm": 2.2820301055908203,
      "learning_rate": 4.1603897814063736e-05,
      "loss": 1.8353,
      "step": 6382
    },
    {
      "epoch": 0.16810639978930736,
      "grad_norm": 1.8434832096099854,
      "learning_rate": 4.160258098498815e-05,
      "loss": 2.0345,
      "step": 6383
    },
    {
      "epoch": 0.16813273637081907,
      "grad_norm": 1.7959675788879395,
      "learning_rate": 4.160126415591257e-05,
      "loss": 1.8559,
      "step": 6384
    },
    {
      "epoch": 0.1681590729523308,
      "grad_norm": 3.790647506713867,
      "learning_rate": 4.159994732683698e-05,
      "loss": 0.675,
      "step": 6385
    },
    {
      "epoch": 0.1681854095338425,
      "grad_norm": 3.7308340072631836,
      "learning_rate": 4.159863049776139e-05,
      "loss": 1.0194,
      "step": 6386
    },
    {
      "epoch": 0.16821174611535422,
      "grad_norm": 3.0776009559631348,
      "learning_rate": 4.159731366868581e-05,
      "loss": 1.2393,
      "step": 6387
    },
    {
      "epoch": 0.16823808269686594,
      "grad_norm": 2.9640960693359375,
      "learning_rate": 4.1595996839610217e-05,
      "loss": 1.4725,
      "step": 6388
    },
    {
      "epoch": 0.16826441927837768,
      "grad_norm": 3.3266842365264893,
      "learning_rate": 4.159468001053464e-05,
      "loss": 1.0055,
      "step": 6389
    },
    {
      "epoch": 0.1682907558598894,
      "grad_norm": 3.820871353149414,
      "learning_rate": 4.159336318145905e-05,
      "loss": 0.6642,
      "step": 6390
    },
    {
      "epoch": 0.1683170924414011,
      "grad_norm": 2.974294900894165,
      "learning_rate": 4.159204635238346e-05,
      "loss": 1.9128,
      "step": 6391
    },
    {
      "epoch": 0.16834342902291283,
      "grad_norm": 2.456472635269165,
      "learning_rate": 4.159072952330787e-05,
      "loss": 1.8701,
      "step": 6392
    },
    {
      "epoch": 0.16836976560442454,
      "grad_norm": 3.471158027648926,
      "learning_rate": 4.158941269423229e-05,
      "loss": 1.1492,
      "step": 6393
    },
    {
      "epoch": 0.16839610218593626,
      "grad_norm": 3.248304605484009,
      "learning_rate": 4.15880958651567e-05,
      "loss": 1.9342,
      "step": 6394
    },
    {
      "epoch": 0.16842243876744797,
      "grad_norm": 5.455419540405273,
      "learning_rate": 4.158677903608112e-05,
      "loss": 1.2831,
      "step": 6395
    },
    {
      "epoch": 0.16844877534895972,
      "grad_norm": 3.4281656742095947,
      "learning_rate": 4.1585462207005535e-05,
      "loss": 1.116,
      "step": 6396
    },
    {
      "epoch": 0.16847511193047143,
      "grad_norm": 2.4252750873565674,
      "learning_rate": 4.158414537792994e-05,
      "loss": 1.4518,
      "step": 6397
    },
    {
      "epoch": 0.16850144851198315,
      "grad_norm": 1.7467241287231445,
      "learning_rate": 4.1582828548854366e-05,
      "loss": 1.6271,
      "step": 6398
    },
    {
      "epoch": 0.16852778509349486,
      "grad_norm": 5.699279308319092,
      "learning_rate": 4.1581511719778775e-05,
      "loss": 0.9435,
      "step": 6399
    },
    {
      "epoch": 0.16855412167500658,
      "grad_norm": 3.9772253036499023,
      "learning_rate": 4.158019489070319e-05,
      "loss": 0.774,
      "step": 6400
    },
    {
      "epoch": 0.1685804582565183,
      "grad_norm": 2.4051127433776855,
      "learning_rate": 4.15788780616276e-05,
      "loss": 1.7705,
      "step": 6401
    },
    {
      "epoch": 0.16860679483803,
      "grad_norm": 1.9040956497192383,
      "learning_rate": 4.1577561232552015e-05,
      "loss": 1.6628,
      "step": 6402
    },
    {
      "epoch": 0.16863313141954175,
      "grad_norm": 1.9643168449401855,
      "learning_rate": 4.157624440347643e-05,
      "loss": 1.1949,
      "step": 6403
    },
    {
      "epoch": 0.16865946800105347,
      "grad_norm": 3.734225273132324,
      "learning_rate": 4.1574927574400846e-05,
      "loss": 1.8115,
      "step": 6404
    },
    {
      "epoch": 0.16868580458256519,
      "grad_norm": 1.7562239170074463,
      "learning_rate": 4.157361074532526e-05,
      "loss": 1.5727,
      "step": 6405
    },
    {
      "epoch": 0.1687121411640769,
      "grad_norm": 3.9833977222442627,
      "learning_rate": 4.157229391624967e-05,
      "loss": 2.0763,
      "step": 6406
    },
    {
      "epoch": 0.16873847774558862,
      "grad_norm": 1.9149258136749268,
      "learning_rate": 4.1570977087174086e-05,
      "loss": 2.1516,
      "step": 6407
    },
    {
      "epoch": 0.16876481432710033,
      "grad_norm": 4.471598148345947,
      "learning_rate": 4.15696602580985e-05,
      "loss": 1.8673,
      "step": 6408
    },
    {
      "epoch": 0.16879115090861208,
      "grad_norm": 4.501389980316162,
      "learning_rate": 4.156834342902292e-05,
      "loss": 1.4924,
      "step": 6409
    },
    {
      "epoch": 0.1688174874901238,
      "grad_norm": 2.703948736190796,
      "learning_rate": 4.1567026599947326e-05,
      "loss": 1.4937,
      "step": 6410
    },
    {
      "epoch": 0.1688438240716355,
      "grad_norm": 3.3694334030151367,
      "learning_rate": 4.156570977087174e-05,
      "loss": 0.8934,
      "step": 6411
    },
    {
      "epoch": 0.16887016065314722,
      "grad_norm": 2.8854258060455322,
      "learning_rate": 4.156439294179616e-05,
      "loss": 0.887,
      "step": 6412
    },
    {
      "epoch": 0.16889649723465894,
      "grad_norm": 2.223555326461792,
      "learning_rate": 4.156307611272057e-05,
      "loss": 2.0356,
      "step": 6413
    },
    {
      "epoch": 0.16892283381617065,
      "grad_norm": 2.5965170860290527,
      "learning_rate": 4.156175928364499e-05,
      "loss": 0.6074,
      "step": 6414
    },
    {
      "epoch": 0.16894917039768237,
      "grad_norm": 2.111039876937866,
      "learning_rate": 4.15604424545694e-05,
      "loss": 1.847,
      "step": 6415
    },
    {
      "epoch": 0.1689755069791941,
      "grad_norm": 10.21255874633789,
      "learning_rate": 4.155912562549381e-05,
      "loss": 1.2639,
      "step": 6416
    },
    {
      "epoch": 0.16900184356070583,
      "grad_norm": 2.9755373001098633,
      "learning_rate": 4.155780879641823e-05,
      "loss": 0.5679,
      "step": 6417
    },
    {
      "epoch": 0.16902818014221754,
      "grad_norm": 1.7586227655410767,
      "learning_rate": 4.1556491967342644e-05,
      "loss": 1.7276,
      "step": 6418
    },
    {
      "epoch": 0.16905451672372926,
      "grad_norm": 3.6247305870056152,
      "learning_rate": 4.155517513826705e-05,
      "loss": 1.3592,
      "step": 6419
    },
    {
      "epoch": 0.16908085330524097,
      "grad_norm": 2.6267502307891846,
      "learning_rate": 4.155385830919147e-05,
      "loss": 1.3719,
      "step": 6420
    },
    {
      "epoch": 0.1691071898867527,
      "grad_norm": 2.071226119995117,
      "learning_rate": 4.1552541480115884e-05,
      "loss": 1.5613,
      "step": 6421
    },
    {
      "epoch": 0.1691335264682644,
      "grad_norm": 8.02920150756836,
      "learning_rate": 4.15512246510403e-05,
      "loss": 1.6523,
      "step": 6422
    },
    {
      "epoch": 0.16915986304977615,
      "grad_norm": 1.599113941192627,
      "learning_rate": 4.1549907821964715e-05,
      "loss": 2.5116,
      "step": 6423
    },
    {
      "epoch": 0.16918619963128786,
      "grad_norm": 1.7589728832244873,
      "learning_rate": 4.1548590992889124e-05,
      "loss": 0.9217,
      "step": 6424
    },
    {
      "epoch": 0.16921253621279958,
      "grad_norm": 2.3213589191436768,
      "learning_rate": 4.154727416381354e-05,
      "loss": 1.948,
      "step": 6425
    },
    {
      "epoch": 0.1692388727943113,
      "grad_norm": 1.698765754699707,
      "learning_rate": 4.154595733473795e-05,
      "loss": 1.4775,
      "step": 6426
    },
    {
      "epoch": 0.169265209375823,
      "grad_norm": 3.4839954376220703,
      "learning_rate": 4.154464050566237e-05,
      "loss": 1.6487,
      "step": 6427
    },
    {
      "epoch": 0.16929154595733473,
      "grad_norm": 3.0693438053131104,
      "learning_rate": 4.154332367658678e-05,
      "loss": 2.2047,
      "step": 6428
    },
    {
      "epoch": 0.16931788253884647,
      "grad_norm": 2.398834705352783,
      "learning_rate": 4.1542006847511195e-05,
      "loss": 2.2971,
      "step": 6429
    },
    {
      "epoch": 0.16934421912035819,
      "grad_norm": 3.881516456604004,
      "learning_rate": 4.154069001843561e-05,
      "loss": 1.8524,
      "step": 6430
    },
    {
      "epoch": 0.1693705557018699,
      "grad_norm": 2.030421257019043,
      "learning_rate": 4.1539373189360026e-05,
      "loss": 1.8384,
      "step": 6431
    },
    {
      "epoch": 0.16939689228338162,
      "grad_norm": 2.8169984817504883,
      "learning_rate": 4.153805636028444e-05,
      "loss": 1.3217,
      "step": 6432
    },
    {
      "epoch": 0.16942322886489333,
      "grad_norm": 3.0963947772979736,
      "learning_rate": 4.153673953120885e-05,
      "loss": 1.5044,
      "step": 6433
    },
    {
      "epoch": 0.16944956544640505,
      "grad_norm": 2.2183096408843994,
      "learning_rate": 4.1535422702133266e-05,
      "loss": 2.0749,
      "step": 6434
    },
    {
      "epoch": 0.16947590202791676,
      "grad_norm": 2.6742281913757324,
      "learning_rate": 4.1534105873057675e-05,
      "loss": 2.1226,
      "step": 6435
    },
    {
      "epoch": 0.1695022386094285,
      "grad_norm": 2.041149139404297,
      "learning_rate": 4.15327890439821e-05,
      "loss": 1.8089,
      "step": 6436
    },
    {
      "epoch": 0.16952857519094022,
      "grad_norm": 1.8499716520309448,
      "learning_rate": 4.1531472214906506e-05,
      "loss": 1.1947,
      "step": 6437
    },
    {
      "epoch": 0.16955491177245194,
      "grad_norm": 2.663862466812134,
      "learning_rate": 4.153015538583092e-05,
      "loss": 2.1878,
      "step": 6438
    },
    {
      "epoch": 0.16958124835396365,
      "grad_norm": 2.132267475128174,
      "learning_rate": 4.152883855675533e-05,
      "loss": 2.2935,
      "step": 6439
    },
    {
      "epoch": 0.16960758493547537,
      "grad_norm": 1.706586241722107,
      "learning_rate": 4.1527521727679746e-05,
      "loss": 0.457,
      "step": 6440
    },
    {
      "epoch": 0.16963392151698709,
      "grad_norm": 2.2798023223876953,
      "learning_rate": 4.152620489860416e-05,
      "loss": 1.5746,
      "step": 6441
    },
    {
      "epoch": 0.16966025809849883,
      "grad_norm": 2.7831649780273438,
      "learning_rate": 4.152488806952858e-05,
      "loss": 2.3509,
      "step": 6442
    },
    {
      "epoch": 0.16968659468001054,
      "grad_norm": 3.353847026824951,
      "learning_rate": 4.152357124045299e-05,
      "loss": 1.2742,
      "step": 6443
    },
    {
      "epoch": 0.16971293126152226,
      "grad_norm": 2.7306783199310303,
      "learning_rate": 4.15222544113774e-05,
      "loss": 2.2132,
      "step": 6444
    },
    {
      "epoch": 0.16973926784303398,
      "grad_norm": 3.453615427017212,
      "learning_rate": 4.152093758230182e-05,
      "loss": 2.1731,
      "step": 6445
    },
    {
      "epoch": 0.1697656044245457,
      "grad_norm": 2.8907649517059326,
      "learning_rate": 4.151962075322623e-05,
      "loss": 1.985,
      "step": 6446
    },
    {
      "epoch": 0.1697919410060574,
      "grad_norm": 3.216059923171997,
      "learning_rate": 4.151830392415065e-05,
      "loss": 1.7245,
      "step": 6447
    },
    {
      "epoch": 0.16981827758756912,
      "grad_norm": 2.4432458877563477,
      "learning_rate": 4.151698709507506e-05,
      "loss": 2.4146,
      "step": 6448
    },
    {
      "epoch": 0.16984461416908087,
      "grad_norm": 2.241258382797241,
      "learning_rate": 4.151567026599947e-05,
      "loss": 1.1708,
      "step": 6449
    },
    {
      "epoch": 0.16987095075059258,
      "grad_norm": 3.880747079849243,
      "learning_rate": 4.151435343692389e-05,
      "loss": 2.7322,
      "step": 6450
    },
    {
      "epoch": 0.1698972873321043,
      "grad_norm": 5.091732025146484,
      "learning_rate": 4.1513036607848304e-05,
      "loss": 1.1156,
      "step": 6451
    },
    {
      "epoch": 0.169923623913616,
      "grad_norm": 2.030904769897461,
      "learning_rate": 4.151171977877272e-05,
      "loss": 2.0242,
      "step": 6452
    },
    {
      "epoch": 0.16994996049512773,
      "grad_norm": 3.2502996921539307,
      "learning_rate": 4.151040294969713e-05,
      "loss": 1.492,
      "step": 6453
    },
    {
      "epoch": 0.16997629707663944,
      "grad_norm": 1.8628132343292236,
      "learning_rate": 4.1509086120621544e-05,
      "loss": 1.8729,
      "step": 6454
    },
    {
      "epoch": 0.17000263365815116,
      "grad_norm": 5.081038475036621,
      "learning_rate": 4.150776929154596e-05,
      "loss": 1.171,
      "step": 6455
    },
    {
      "epoch": 0.1700289702396629,
      "grad_norm": 2.8787131309509277,
      "learning_rate": 4.1506452462470376e-05,
      "loss": 1.8752,
      "step": 6456
    },
    {
      "epoch": 0.17005530682117462,
      "grad_norm": 4.341348171234131,
      "learning_rate": 4.1505135633394784e-05,
      "loss": 0.863,
      "step": 6457
    },
    {
      "epoch": 0.17008164340268633,
      "grad_norm": 2.048917293548584,
      "learning_rate": 4.15038188043192e-05,
      "loss": 1.969,
      "step": 6458
    },
    {
      "epoch": 0.17010797998419805,
      "grad_norm": 5.052336692810059,
      "learning_rate": 4.1502501975243616e-05,
      "loss": 0.8067,
      "step": 6459
    },
    {
      "epoch": 0.17013431656570976,
      "grad_norm": 1.709985375404358,
      "learning_rate": 4.150118514616803e-05,
      "loss": 1.4318,
      "step": 6460
    },
    {
      "epoch": 0.17016065314722148,
      "grad_norm": 3.442023992538452,
      "learning_rate": 4.149986831709245e-05,
      "loss": 1.7595,
      "step": 6461
    },
    {
      "epoch": 0.17018698972873322,
      "grad_norm": 2.279386520385742,
      "learning_rate": 4.1498551488016856e-05,
      "loss": 1.5026,
      "step": 6462
    },
    {
      "epoch": 0.17021332631024494,
      "grad_norm": 3.368828058242798,
      "learning_rate": 4.149723465894127e-05,
      "loss": 1.5382,
      "step": 6463
    },
    {
      "epoch": 0.17023966289175665,
      "grad_norm": 2.1874027252197266,
      "learning_rate": 4.149591782986569e-05,
      "loss": 1.3267,
      "step": 6464
    },
    {
      "epoch": 0.17026599947326837,
      "grad_norm": 2.9538815021514893,
      "learning_rate": 4.14946010007901e-05,
      "loss": 1.3876,
      "step": 6465
    },
    {
      "epoch": 0.17029233605478009,
      "grad_norm": 2.819532632827759,
      "learning_rate": 4.149328417171451e-05,
      "loss": 1.0644,
      "step": 6466
    },
    {
      "epoch": 0.1703186726362918,
      "grad_norm": 1.8892769813537598,
      "learning_rate": 4.149196734263893e-05,
      "loss": 1.9414,
      "step": 6467
    },
    {
      "epoch": 0.17034500921780352,
      "grad_norm": 2.458207607269287,
      "learning_rate": 4.149065051356334e-05,
      "loss": 1.2594,
      "step": 6468
    },
    {
      "epoch": 0.17037134579931526,
      "grad_norm": 1.8219809532165527,
      "learning_rate": 4.148933368448776e-05,
      "loss": 1.4863,
      "step": 6469
    },
    {
      "epoch": 0.17039768238082698,
      "grad_norm": 2.603959083557129,
      "learning_rate": 4.1488016855412174e-05,
      "loss": 2.0385,
      "step": 6470
    },
    {
      "epoch": 0.1704240189623387,
      "grad_norm": 4.030280590057373,
      "learning_rate": 4.148670002633658e-05,
      "loss": 0.72,
      "step": 6471
    },
    {
      "epoch": 0.1704503555438504,
      "grad_norm": 4.136963844299316,
      "learning_rate": 4.1485383197261e-05,
      "loss": 1.5288,
      "step": 6472
    },
    {
      "epoch": 0.17047669212536212,
      "grad_norm": 4.837243556976318,
      "learning_rate": 4.148406636818541e-05,
      "loss": 1.5353,
      "step": 6473
    },
    {
      "epoch": 0.17050302870687384,
      "grad_norm": 2.52107310295105,
      "learning_rate": 4.148274953910983e-05,
      "loss": 0.4875,
      "step": 6474
    },
    {
      "epoch": 0.17052936528838555,
      "grad_norm": 1.8930702209472656,
      "learning_rate": 4.148143271003424e-05,
      "loss": 1.8783,
      "step": 6475
    },
    {
      "epoch": 0.1705557018698973,
      "grad_norm": 1.9251892566680908,
      "learning_rate": 4.1480115880958654e-05,
      "loss": 2.0516,
      "step": 6476
    },
    {
      "epoch": 0.170582038451409,
      "grad_norm": 3.5314857959747314,
      "learning_rate": 4.147879905188307e-05,
      "loss": 1.2225,
      "step": 6477
    },
    {
      "epoch": 0.17060837503292073,
      "grad_norm": 2.236572027206421,
      "learning_rate": 4.147748222280748e-05,
      "loss": 2.0008,
      "step": 6478
    },
    {
      "epoch": 0.17063471161443244,
      "grad_norm": 2.191998243331909,
      "learning_rate": 4.14761653937319e-05,
      "loss": 1.4144,
      "step": 6479
    },
    {
      "epoch": 0.17066104819594416,
      "grad_norm": 2.1222879886627197,
      "learning_rate": 4.147484856465631e-05,
      "loss": 1.988,
      "step": 6480
    },
    {
      "epoch": 0.17068738477745588,
      "grad_norm": 2.2267048358917236,
      "learning_rate": 4.1473531735580725e-05,
      "loss": 1.4148,
      "step": 6481
    },
    {
      "epoch": 0.17071372135896762,
      "grad_norm": 1.9699732065200806,
      "learning_rate": 4.1472214906505134e-05,
      "loss": 1.7027,
      "step": 6482
    },
    {
      "epoch": 0.17074005794047933,
      "grad_norm": 2.0228424072265625,
      "learning_rate": 4.1470898077429556e-05,
      "loss": 2.1025,
      "step": 6483
    },
    {
      "epoch": 0.17076639452199105,
      "grad_norm": 2.1892802715301514,
      "learning_rate": 4.1469581248353965e-05,
      "loss": 1.9244,
      "step": 6484
    },
    {
      "epoch": 0.17079273110350277,
      "grad_norm": 2.0006086826324463,
      "learning_rate": 4.146826441927838e-05,
      "loss": 1.5034,
      "step": 6485
    },
    {
      "epoch": 0.17081906768501448,
      "grad_norm": 1.6685675382614136,
      "learning_rate": 4.146694759020279e-05,
      "loss": 2.0658,
      "step": 6486
    },
    {
      "epoch": 0.1708454042665262,
      "grad_norm": 1.7031437158584595,
      "learning_rate": 4.1465630761127205e-05,
      "loss": 1.7038,
      "step": 6487
    },
    {
      "epoch": 0.1708717408480379,
      "grad_norm": 2.0737340450286865,
      "learning_rate": 4.146431393205163e-05,
      "loss": 0.7668,
      "step": 6488
    },
    {
      "epoch": 0.17089807742954966,
      "grad_norm": 3.201235294342041,
      "learning_rate": 4.1462997102976036e-05,
      "loss": 0.6086,
      "step": 6489
    },
    {
      "epoch": 0.17092441401106137,
      "grad_norm": 2.792245864868164,
      "learning_rate": 4.146168027390045e-05,
      "loss": 1.6417,
      "step": 6490
    },
    {
      "epoch": 0.1709507505925731,
      "grad_norm": 1.8106646537780762,
      "learning_rate": 4.146036344482486e-05,
      "loss": 1.9466,
      "step": 6491
    },
    {
      "epoch": 0.1709770871740848,
      "grad_norm": 2.1337807178497314,
      "learning_rate": 4.1459046615749276e-05,
      "loss": 1.3608,
      "step": 6492
    },
    {
      "epoch": 0.17100342375559652,
      "grad_norm": 1.7259540557861328,
      "learning_rate": 4.145772978667369e-05,
      "loss": 1.649,
      "step": 6493
    },
    {
      "epoch": 0.17102976033710823,
      "grad_norm": 2.1733360290527344,
      "learning_rate": 4.145641295759811e-05,
      "loss": 1.8353,
      "step": 6494
    },
    {
      "epoch": 0.17105609691861998,
      "grad_norm": 1.8094542026519775,
      "learning_rate": 4.1455096128522516e-05,
      "loss": 1.7019,
      "step": 6495
    },
    {
      "epoch": 0.1710824335001317,
      "grad_norm": 2.2861244678497314,
      "learning_rate": 4.145377929944693e-05,
      "loss": 2.0309,
      "step": 6496
    },
    {
      "epoch": 0.1711087700816434,
      "grad_norm": 4.071788787841797,
      "learning_rate": 4.145246247037135e-05,
      "loss": 1.4797,
      "step": 6497
    },
    {
      "epoch": 0.17113510666315512,
      "grad_norm": 3.26546311378479,
      "learning_rate": 4.145114564129576e-05,
      "loss": 1.1261,
      "step": 6498
    },
    {
      "epoch": 0.17116144324466684,
      "grad_norm": 8.301631927490234,
      "learning_rate": 4.144982881222018e-05,
      "loss": 1.6048,
      "step": 6499
    },
    {
      "epoch": 0.17118777982617855,
      "grad_norm": 1.9324514865875244,
      "learning_rate": 4.144851198314459e-05,
      "loss": 0.3971,
      "step": 6500
    },
    {
      "epoch": 0.17121411640769027,
      "grad_norm": 3.5214147567749023,
      "learning_rate": 4.1447195154069e-05,
      "loss": 1.6008,
      "step": 6501
    },
    {
      "epoch": 0.171240452989202,
      "grad_norm": 2.0774364471435547,
      "learning_rate": 4.144587832499342e-05,
      "loss": 1.9798,
      "step": 6502
    },
    {
      "epoch": 0.17126678957071373,
      "grad_norm": 1.9394217729568481,
      "learning_rate": 4.1444561495917834e-05,
      "loss": 1.6655,
      "step": 6503
    },
    {
      "epoch": 0.17129312615222544,
      "grad_norm": 3.2418298721313477,
      "learning_rate": 4.144324466684224e-05,
      "loss": 1.675,
      "step": 6504
    },
    {
      "epoch": 0.17131946273373716,
      "grad_norm": 1.9306080341339111,
      "learning_rate": 4.144192783776666e-05,
      "loss": 1.8287,
      "step": 6505
    },
    {
      "epoch": 0.17134579931524888,
      "grad_norm": 1.3904544115066528,
      "learning_rate": 4.1440611008691074e-05,
      "loss": 2.1646,
      "step": 6506
    },
    {
      "epoch": 0.1713721358967606,
      "grad_norm": 1.7397165298461914,
      "learning_rate": 4.143929417961549e-05,
      "loss": 1.3778,
      "step": 6507
    },
    {
      "epoch": 0.1713984724782723,
      "grad_norm": 1.8746224641799927,
      "learning_rate": 4.1437977350539905e-05,
      "loss": 0.8404,
      "step": 6508
    },
    {
      "epoch": 0.17142480905978405,
      "grad_norm": 6.493264675140381,
      "learning_rate": 4.1436660521464314e-05,
      "loss": 1.4011,
      "step": 6509
    },
    {
      "epoch": 0.17145114564129577,
      "grad_norm": 2.357227087020874,
      "learning_rate": 4.143534369238873e-05,
      "loss": 1.1308,
      "step": 6510
    },
    {
      "epoch": 0.17147748222280748,
      "grad_norm": 1.5707148313522339,
      "learning_rate": 4.143402686331314e-05,
      "loss": 1.1768,
      "step": 6511
    },
    {
      "epoch": 0.1715038188043192,
      "grad_norm": 6.075654029846191,
      "learning_rate": 4.143271003423756e-05,
      "loss": 1.5775,
      "step": 6512
    },
    {
      "epoch": 0.1715301553858309,
      "grad_norm": 3.0365631580352783,
      "learning_rate": 4.143139320516197e-05,
      "loss": 1.7591,
      "step": 6513
    },
    {
      "epoch": 0.17155649196734263,
      "grad_norm": 2.6698756217956543,
      "learning_rate": 4.1430076376086385e-05,
      "loss": 1.5061,
      "step": 6514
    },
    {
      "epoch": 0.17158282854885437,
      "grad_norm": 2.3414206504821777,
      "learning_rate": 4.14287595470108e-05,
      "loss": 1.892,
      "step": 6515
    },
    {
      "epoch": 0.1716091651303661,
      "grad_norm": 3.2102322578430176,
      "learning_rate": 4.1427442717935217e-05,
      "loss": 0.8145,
      "step": 6516
    },
    {
      "epoch": 0.1716355017118778,
      "grad_norm": 4.5180864334106445,
      "learning_rate": 4.142612588885963e-05,
      "loss": 1.5787,
      "step": 6517
    },
    {
      "epoch": 0.17166183829338952,
      "grad_norm": 2.3173553943634033,
      "learning_rate": 4.142480905978404e-05,
      "loss": 1.7247,
      "step": 6518
    },
    {
      "epoch": 0.17168817487490123,
      "grad_norm": 5.040765285491943,
      "learning_rate": 4.1423492230708457e-05,
      "loss": 0.9599,
      "step": 6519
    },
    {
      "epoch": 0.17171451145641295,
      "grad_norm": 2.1651523113250732,
      "learning_rate": 4.1422175401632865e-05,
      "loss": 2.3724,
      "step": 6520
    },
    {
      "epoch": 0.17174084803792466,
      "grad_norm": 2.7365846633911133,
      "learning_rate": 4.142085857255729e-05,
      "loss": 1.509,
      "step": 6521
    },
    {
      "epoch": 0.1717671846194364,
      "grad_norm": 1.9443204402923584,
      "learning_rate": 4.1419541743481697e-05,
      "loss": 2.3737,
      "step": 6522
    },
    {
      "epoch": 0.17179352120094812,
      "grad_norm": 1.792307734489441,
      "learning_rate": 4.141822491440611e-05,
      "loss": 1.9221,
      "step": 6523
    },
    {
      "epoch": 0.17181985778245984,
      "grad_norm": 3.0441348552703857,
      "learning_rate": 4.141690808533053e-05,
      "loss": 1.3485,
      "step": 6524
    },
    {
      "epoch": 0.17184619436397156,
      "grad_norm": 1.9249099493026733,
      "learning_rate": 4.1415591256254937e-05,
      "loss": 0.8577,
      "step": 6525
    },
    {
      "epoch": 0.17187253094548327,
      "grad_norm": 2.668057680130005,
      "learning_rate": 4.141427442717936e-05,
      "loss": 0.6554,
      "step": 6526
    },
    {
      "epoch": 0.171898867526995,
      "grad_norm": 1.8003125190734863,
      "learning_rate": 4.141295759810377e-05,
      "loss": 1.5838,
      "step": 6527
    },
    {
      "epoch": 0.17192520410850673,
      "grad_norm": 2.3780336380004883,
      "learning_rate": 4.141164076902818e-05,
      "loss": 1.9685,
      "step": 6528
    },
    {
      "epoch": 0.17195154069001845,
      "grad_norm": 2.542705774307251,
      "learning_rate": 4.141032393995259e-05,
      "loss": 2.4901,
      "step": 6529
    },
    {
      "epoch": 0.17197787727153016,
      "grad_norm": 2.2208974361419678,
      "learning_rate": 4.1409007110877015e-05,
      "loss": 2.031,
      "step": 6530
    },
    {
      "epoch": 0.17200421385304188,
      "grad_norm": 3.307168483734131,
      "learning_rate": 4.140769028180142e-05,
      "loss": 1.452,
      "step": 6531
    },
    {
      "epoch": 0.1720305504345536,
      "grad_norm": 5.970710277557373,
      "learning_rate": 4.140637345272584e-05,
      "loss": 1.1322,
      "step": 6532
    },
    {
      "epoch": 0.1720568870160653,
      "grad_norm": 1.7820929288864136,
      "learning_rate": 4.1405056623650255e-05,
      "loss": 2.0271,
      "step": 6533
    },
    {
      "epoch": 0.17208322359757702,
      "grad_norm": 2.1190316677093506,
      "learning_rate": 4.1403739794574663e-05,
      "loss": 2.0157,
      "step": 6534
    },
    {
      "epoch": 0.17210956017908877,
      "grad_norm": 2.2790586948394775,
      "learning_rate": 4.1402422965499086e-05,
      "loss": 1.4108,
      "step": 6535
    },
    {
      "epoch": 0.17213589676060048,
      "grad_norm": 2.4451963901519775,
      "learning_rate": 4.1401106136423495e-05,
      "loss": 1.9138,
      "step": 6536
    },
    {
      "epoch": 0.1721622333421122,
      "grad_norm": 2.708577871322632,
      "learning_rate": 4.139978930734791e-05,
      "loss": 0.4258,
      "step": 6537
    },
    {
      "epoch": 0.1721885699236239,
      "grad_norm": 3.1696577072143555,
      "learning_rate": 4.139847247827232e-05,
      "loss": 1.4663,
      "step": 6538
    },
    {
      "epoch": 0.17221490650513563,
      "grad_norm": 1.8144752979278564,
      "learning_rate": 4.1397155649196735e-05,
      "loss": 1.7598,
      "step": 6539
    },
    {
      "epoch": 0.17224124308664734,
      "grad_norm": 2.4526026248931885,
      "learning_rate": 4.139583882012115e-05,
      "loss": 1.702,
      "step": 6540
    },
    {
      "epoch": 0.17226757966815906,
      "grad_norm": 1.9719762802124023,
      "learning_rate": 4.1394521991045566e-05,
      "loss": 1.6701,
      "step": 6541
    },
    {
      "epoch": 0.1722939162496708,
      "grad_norm": 3.4558773040771484,
      "learning_rate": 4.1393205161969975e-05,
      "loss": 1.3219,
      "step": 6542
    },
    {
      "epoch": 0.17232025283118252,
      "grad_norm": 3.586822032928467,
      "learning_rate": 4.139188833289439e-05,
      "loss": 1.8346,
      "step": 6543
    },
    {
      "epoch": 0.17234658941269423,
      "grad_norm": 2.3900208473205566,
      "learning_rate": 4.1390571503818806e-05,
      "loss": 1.7368,
      "step": 6544
    },
    {
      "epoch": 0.17237292599420595,
      "grad_norm": 3.041321277618408,
      "learning_rate": 4.138925467474322e-05,
      "loss": 1.2313,
      "step": 6545
    },
    {
      "epoch": 0.17239926257571767,
      "grad_norm": 1.6797538995742798,
      "learning_rate": 4.138793784566764e-05,
      "loss": 1.6437,
      "step": 6546
    },
    {
      "epoch": 0.17242559915722938,
      "grad_norm": 1.9204726219177246,
      "learning_rate": 4.1386621016592046e-05,
      "loss": 1.6408,
      "step": 6547
    },
    {
      "epoch": 0.17245193573874112,
      "grad_norm": 4.22277307510376,
      "learning_rate": 4.138530418751646e-05,
      "loss": 1.3494,
      "step": 6548
    },
    {
      "epoch": 0.17247827232025284,
      "grad_norm": 2.0165631771087646,
      "learning_rate": 4.138398735844088e-05,
      "loss": 1.788,
      "step": 6549
    },
    {
      "epoch": 0.17250460890176456,
      "grad_norm": 2.410510778427124,
      "learning_rate": 4.138267052936529e-05,
      "loss": 2.1595,
      "step": 6550
    },
    {
      "epoch": 0.17253094548327627,
      "grad_norm": 4.09640645980835,
      "learning_rate": 4.13813537002897e-05,
      "loss": 1.2022,
      "step": 6551
    },
    {
      "epoch": 0.172557282064788,
      "grad_norm": 1.8408640623092651,
      "learning_rate": 4.138003687121412e-05,
      "loss": 2.0705,
      "step": 6552
    },
    {
      "epoch": 0.1725836186462997,
      "grad_norm": 2.3870368003845215,
      "learning_rate": 4.137872004213853e-05,
      "loss": 1.8245,
      "step": 6553
    },
    {
      "epoch": 0.17260995522781142,
      "grad_norm": 2.1747100353240967,
      "learning_rate": 4.137740321306295e-05,
      "loss": 2.0954,
      "step": 6554
    },
    {
      "epoch": 0.17263629180932316,
      "grad_norm": 1.8999983072280884,
      "learning_rate": 4.1376086383987364e-05,
      "loss": 2.2851,
      "step": 6555
    },
    {
      "epoch": 0.17266262839083488,
      "grad_norm": 2.8208136558532715,
      "learning_rate": 4.137476955491177e-05,
      "loss": 1.5715,
      "step": 6556
    },
    {
      "epoch": 0.1726889649723466,
      "grad_norm": 3.2907090187072754,
      "learning_rate": 4.137345272583619e-05,
      "loss": 1.801,
      "step": 6557
    },
    {
      "epoch": 0.1727153015538583,
      "grad_norm": 2.177830696105957,
      "learning_rate": 4.13721358967606e-05,
      "loss": 2.0442,
      "step": 6558
    },
    {
      "epoch": 0.17274163813537002,
      "grad_norm": 4.38843297958374,
      "learning_rate": 4.137081906768502e-05,
      "loss": 2.3097,
      "step": 6559
    },
    {
      "epoch": 0.17276797471688174,
      "grad_norm": 3.4381134510040283,
      "learning_rate": 4.136950223860943e-05,
      "loss": 2.0296,
      "step": 6560
    },
    {
      "epoch": 0.17279431129839345,
      "grad_norm": 2.064281940460205,
      "learning_rate": 4.1368185409533844e-05,
      "loss": 1.7947,
      "step": 6561
    },
    {
      "epoch": 0.1728206478799052,
      "grad_norm": 1.9783134460449219,
      "learning_rate": 4.136686858045826e-05,
      "loss": 1.8043,
      "step": 6562
    },
    {
      "epoch": 0.1728469844614169,
      "grad_norm": 2.5466768741607666,
      "learning_rate": 4.1365551751382675e-05,
      "loss": 1.9159,
      "step": 6563
    },
    {
      "epoch": 0.17287332104292863,
      "grad_norm": 2.6055939197540283,
      "learning_rate": 4.136423492230709e-05,
      "loss": 2.2595,
      "step": 6564
    },
    {
      "epoch": 0.17289965762444034,
      "grad_norm": 1.7846686840057373,
      "learning_rate": 4.13629180932315e-05,
      "loss": 1.3956,
      "step": 6565
    },
    {
      "epoch": 0.17292599420595206,
      "grad_norm": 2.5492637157440186,
      "learning_rate": 4.1361601264155915e-05,
      "loss": 1.8095,
      "step": 6566
    },
    {
      "epoch": 0.17295233078746378,
      "grad_norm": 2.6915223598480225,
      "learning_rate": 4.1360284435080324e-05,
      "loss": 1.8663,
      "step": 6567
    },
    {
      "epoch": 0.17297866736897552,
      "grad_norm": 5.118632793426514,
      "learning_rate": 4.1358967606004746e-05,
      "loss": 1.8523,
      "step": 6568
    },
    {
      "epoch": 0.17300500395048724,
      "grad_norm": 1.6934292316436768,
      "learning_rate": 4.1357650776929155e-05,
      "loss": 1.4678,
      "step": 6569
    },
    {
      "epoch": 0.17303134053199895,
      "grad_norm": 4.224479675292969,
      "learning_rate": 4.135633394785357e-05,
      "loss": 2.2489,
      "step": 6570
    },
    {
      "epoch": 0.17305767711351067,
      "grad_norm": 3.1312708854675293,
      "learning_rate": 4.1355017118777986e-05,
      "loss": 1.3973,
      "step": 6571
    },
    {
      "epoch": 0.17308401369502238,
      "grad_norm": 2.5284790992736816,
      "learning_rate": 4.1353700289702395e-05,
      "loss": 1.6139,
      "step": 6572
    },
    {
      "epoch": 0.1731103502765341,
      "grad_norm": 2.208031177520752,
      "learning_rate": 4.135238346062682e-05,
      "loss": 1.7548,
      "step": 6573
    },
    {
      "epoch": 0.1731366868580458,
      "grad_norm": 1.7932910919189453,
      "learning_rate": 4.1351066631551226e-05,
      "loss": 1.7549,
      "step": 6574
    },
    {
      "epoch": 0.17316302343955756,
      "grad_norm": 4.175819396972656,
      "learning_rate": 4.134974980247564e-05,
      "loss": 1.5917,
      "step": 6575
    },
    {
      "epoch": 0.17318936002106927,
      "grad_norm": 2.4755589962005615,
      "learning_rate": 4.134843297340005e-05,
      "loss": 1.7424,
      "step": 6576
    },
    {
      "epoch": 0.173215696602581,
      "grad_norm": 1.7478114366531372,
      "learning_rate": 4.1347116144324466e-05,
      "loss": 1.5782,
      "step": 6577
    },
    {
      "epoch": 0.1732420331840927,
      "grad_norm": 3.353172540664673,
      "learning_rate": 4.134579931524888e-05,
      "loss": 0.3634,
      "step": 6578
    },
    {
      "epoch": 0.17326836976560442,
      "grad_norm": 3.7557127475738525,
      "learning_rate": 4.13444824861733e-05,
      "loss": 1.5429,
      "step": 6579
    },
    {
      "epoch": 0.17329470634711613,
      "grad_norm": 1.7842997312545776,
      "learning_rate": 4.134316565709771e-05,
      "loss": 1.5903,
      "step": 6580
    },
    {
      "epoch": 0.17332104292862788,
      "grad_norm": 4.350416660308838,
      "learning_rate": 4.134184882802212e-05,
      "loss": 1.0064,
      "step": 6581
    },
    {
      "epoch": 0.1733473795101396,
      "grad_norm": 3.2437937259674072,
      "learning_rate": 4.1340531998946544e-05,
      "loss": 0.9464,
      "step": 6582
    },
    {
      "epoch": 0.1733737160916513,
      "grad_norm": 2.259084701538086,
      "learning_rate": 4.133921516987095e-05,
      "loss": 2.1163,
      "step": 6583
    },
    {
      "epoch": 0.17340005267316302,
      "grad_norm": 3.8716816902160645,
      "learning_rate": 4.133789834079537e-05,
      "loss": 2.0451,
      "step": 6584
    },
    {
      "epoch": 0.17342638925467474,
      "grad_norm": 2.8280696868896484,
      "learning_rate": 4.133658151171978e-05,
      "loss": 1.5448,
      "step": 6585
    },
    {
      "epoch": 0.17345272583618646,
      "grad_norm": 2.231928825378418,
      "learning_rate": 4.133526468264419e-05,
      "loss": 2.0026,
      "step": 6586
    },
    {
      "epoch": 0.17347906241769817,
      "grad_norm": 2.909186840057373,
      "learning_rate": 4.133394785356861e-05,
      "loss": 0.635,
      "step": 6587
    },
    {
      "epoch": 0.17350539899920991,
      "grad_norm": 1.8089038133621216,
      "learning_rate": 4.1332631024493024e-05,
      "loss": 1.9228,
      "step": 6588
    },
    {
      "epoch": 0.17353173558072163,
      "grad_norm": 2.0774645805358887,
      "learning_rate": 4.133131419541743e-05,
      "loss": 2.1094,
      "step": 6589
    },
    {
      "epoch": 0.17355807216223335,
      "grad_norm": 1.901111364364624,
      "learning_rate": 4.132999736634185e-05,
      "loss": 1.8667,
      "step": 6590
    },
    {
      "epoch": 0.17358440874374506,
      "grad_norm": 3.020867347717285,
      "learning_rate": 4.1328680537266264e-05,
      "loss": 1.7473,
      "step": 6591
    },
    {
      "epoch": 0.17361074532525678,
      "grad_norm": 2.1292335987091064,
      "learning_rate": 4.132736370819068e-05,
      "loss": 1.9234,
      "step": 6592
    },
    {
      "epoch": 0.1736370819067685,
      "grad_norm": 2.397891044616699,
      "learning_rate": 4.1326046879115096e-05,
      "loss": 1.343,
      "step": 6593
    },
    {
      "epoch": 0.1736634184882802,
      "grad_norm": 2.3489630222320557,
      "learning_rate": 4.1324730050039504e-05,
      "loss": 1.9326,
      "step": 6594
    },
    {
      "epoch": 0.17368975506979195,
      "grad_norm": 2.04750394821167,
      "learning_rate": 4.132341322096392e-05,
      "loss": 2.1108,
      "step": 6595
    },
    {
      "epoch": 0.17371609165130367,
      "grad_norm": 1.9201856851577759,
      "learning_rate": 4.1322096391888336e-05,
      "loss": 1.6795,
      "step": 6596
    },
    {
      "epoch": 0.17374242823281538,
      "grad_norm": 2.228086471557617,
      "learning_rate": 4.132077956281275e-05,
      "loss": 2.0303,
      "step": 6597
    },
    {
      "epoch": 0.1737687648143271,
      "grad_norm": 1.8679587841033936,
      "learning_rate": 4.131946273373716e-05,
      "loss": 1.8791,
      "step": 6598
    },
    {
      "epoch": 0.1737951013958388,
      "grad_norm": 2.128915548324585,
      "learning_rate": 4.1318145904661576e-05,
      "loss": 1.9337,
      "step": 6599
    },
    {
      "epoch": 0.17382143797735053,
      "grad_norm": 1.869482398033142,
      "learning_rate": 4.131682907558599e-05,
      "loss": 1.7211,
      "step": 6600
    },
    {
      "epoch": 0.17384777455886227,
      "grad_norm": 1.9034199714660645,
      "learning_rate": 4.131551224651041e-05,
      "loss": 1.9604,
      "step": 6601
    },
    {
      "epoch": 0.173874111140374,
      "grad_norm": 1.6794320344924927,
      "learning_rate": 4.131419541743482e-05,
      "loss": 1.3221,
      "step": 6602
    },
    {
      "epoch": 0.1739004477218857,
      "grad_norm": 2.7474470138549805,
      "learning_rate": 4.131287858835923e-05,
      "loss": 0.6641,
      "step": 6603
    },
    {
      "epoch": 0.17392678430339742,
      "grad_norm": 2.337796688079834,
      "learning_rate": 4.131156175928365e-05,
      "loss": 2.3751,
      "step": 6604
    },
    {
      "epoch": 0.17395312088490913,
      "grad_norm": 1.8859007358551025,
      "learning_rate": 4.1310244930208056e-05,
      "loss": 1.7392,
      "step": 6605
    },
    {
      "epoch": 0.17397945746642085,
      "grad_norm": 3.564038038253784,
      "learning_rate": 4.130892810113248e-05,
      "loss": 1.3112,
      "step": 6606
    },
    {
      "epoch": 0.17400579404793257,
      "grad_norm": 2.746978521347046,
      "learning_rate": 4.130761127205689e-05,
      "loss": 2.0967,
      "step": 6607
    },
    {
      "epoch": 0.1740321306294443,
      "grad_norm": 4.363900661468506,
      "learning_rate": 4.13062944429813e-05,
      "loss": 1.4503,
      "step": 6608
    },
    {
      "epoch": 0.17405846721095602,
      "grad_norm": 1.7683770656585693,
      "learning_rate": 4.130497761390572e-05,
      "loss": 1.378,
      "step": 6609
    },
    {
      "epoch": 0.17408480379246774,
      "grad_norm": 2.616419792175293,
      "learning_rate": 4.1303660784830134e-05,
      "loss": 1.0869,
      "step": 6610
    },
    {
      "epoch": 0.17411114037397946,
      "grad_norm": 5.405492305755615,
      "learning_rate": 4.130234395575455e-05,
      "loss": 1.441,
      "step": 6611
    },
    {
      "epoch": 0.17413747695549117,
      "grad_norm": 2.149705171585083,
      "learning_rate": 4.130102712667896e-05,
      "loss": 2.3707,
      "step": 6612
    },
    {
      "epoch": 0.1741638135370029,
      "grad_norm": 1.776121973991394,
      "learning_rate": 4.1299710297603374e-05,
      "loss": 1.8169,
      "step": 6613
    },
    {
      "epoch": 0.17419015011851463,
      "grad_norm": 1.921848177909851,
      "learning_rate": 4.129839346852778e-05,
      "loss": 1.9691,
      "step": 6614
    },
    {
      "epoch": 0.17421648670002635,
      "grad_norm": 2.028383255004883,
      "learning_rate": 4.1297076639452205e-05,
      "loss": 1.8353,
      "step": 6615
    },
    {
      "epoch": 0.17424282328153806,
      "grad_norm": 1.543782114982605,
      "learning_rate": 4.1295759810376614e-05,
      "loss": 1.6592,
      "step": 6616
    },
    {
      "epoch": 0.17426915986304978,
      "grad_norm": 1.9178073406219482,
      "learning_rate": 4.129444298130103e-05,
      "loss": 1.6874,
      "step": 6617
    },
    {
      "epoch": 0.1742954964445615,
      "grad_norm": 1.7270618677139282,
      "learning_rate": 4.1293126152225445e-05,
      "loss": 1.9808,
      "step": 6618
    },
    {
      "epoch": 0.1743218330260732,
      "grad_norm": 3.44000244140625,
      "learning_rate": 4.1291809323149854e-05,
      "loss": 2.0983,
      "step": 6619
    },
    {
      "epoch": 0.17434816960758492,
      "grad_norm": 1.9246654510498047,
      "learning_rate": 4.1290492494074276e-05,
      "loss": 2.5746,
      "step": 6620
    },
    {
      "epoch": 0.17437450618909667,
      "grad_norm": 2.8694772720336914,
      "learning_rate": 4.1289175664998685e-05,
      "loss": 1.7488,
      "step": 6621
    },
    {
      "epoch": 0.17440084277060838,
      "grad_norm": 2.084517478942871,
      "learning_rate": 4.12878588359231e-05,
      "loss": 1.0779,
      "step": 6622
    },
    {
      "epoch": 0.1744271793521201,
      "grad_norm": 2.8672194480895996,
      "learning_rate": 4.128654200684751e-05,
      "loss": 1.8264,
      "step": 6623
    },
    {
      "epoch": 0.17445351593363181,
      "grad_norm": 1.9042001962661743,
      "learning_rate": 4.1285225177771925e-05,
      "loss": 1.7287,
      "step": 6624
    },
    {
      "epoch": 0.17447985251514353,
      "grad_norm": 2.259018659591675,
      "learning_rate": 4.128390834869634e-05,
      "loss": 2.4543,
      "step": 6625
    },
    {
      "epoch": 0.17450618909665525,
      "grad_norm": 2.1348445415496826,
      "learning_rate": 4.1282591519620756e-05,
      "loss": 1.7519,
      "step": 6626
    },
    {
      "epoch": 0.17453252567816696,
      "grad_norm": 2.405287027359009,
      "learning_rate": 4.128127469054517e-05,
      "loss": 1.6668,
      "step": 6627
    },
    {
      "epoch": 0.1745588622596787,
      "grad_norm": 4.773782730102539,
      "learning_rate": 4.127995786146958e-05,
      "loss": 0.9413,
      "step": 6628
    },
    {
      "epoch": 0.17458519884119042,
      "grad_norm": 1.9451075792312622,
      "learning_rate": 4.1278641032394e-05,
      "loss": 2.0911,
      "step": 6629
    },
    {
      "epoch": 0.17461153542270214,
      "grad_norm": 1.5977376699447632,
      "learning_rate": 4.127732420331841e-05,
      "loss": 1.5029,
      "step": 6630
    },
    {
      "epoch": 0.17463787200421385,
      "grad_norm": 3.224968433380127,
      "learning_rate": 4.127600737424283e-05,
      "loss": 1.4132,
      "step": 6631
    },
    {
      "epoch": 0.17466420858572557,
      "grad_norm": 3.925855875015259,
      "learning_rate": 4.1274690545167236e-05,
      "loss": 2.0179,
      "step": 6632
    },
    {
      "epoch": 0.17469054516723728,
      "grad_norm": 2.111485719680786,
      "learning_rate": 4.127337371609165e-05,
      "loss": 1.4731,
      "step": 6633
    },
    {
      "epoch": 0.17471688174874903,
      "grad_norm": 3.393571615219116,
      "learning_rate": 4.127205688701607e-05,
      "loss": 2.2736,
      "step": 6634
    },
    {
      "epoch": 0.17474321833026074,
      "grad_norm": 2.0212717056274414,
      "learning_rate": 4.127074005794048e-05,
      "loss": 2.0275,
      "step": 6635
    },
    {
      "epoch": 0.17476955491177246,
      "grad_norm": 1.6354917287826538,
      "learning_rate": 4.12694232288649e-05,
      "loss": 1.5985,
      "step": 6636
    },
    {
      "epoch": 0.17479589149328417,
      "grad_norm": 1.7444981336593628,
      "learning_rate": 4.126810639978931e-05,
      "loss": 1.7306,
      "step": 6637
    },
    {
      "epoch": 0.1748222280747959,
      "grad_norm": 2.5249311923980713,
      "learning_rate": 4.126678957071372e-05,
      "loss": 0.8634,
      "step": 6638
    },
    {
      "epoch": 0.1748485646563076,
      "grad_norm": 2.2393085956573486,
      "learning_rate": 4.126547274163814e-05,
      "loss": 1.5959,
      "step": 6639
    },
    {
      "epoch": 0.17487490123781932,
      "grad_norm": 2.3244261741638184,
      "learning_rate": 4.1264155912562554e-05,
      "loss": 2.3055,
      "step": 6640
    },
    {
      "epoch": 0.17490123781933106,
      "grad_norm": 2.5192677974700928,
      "learning_rate": 4.126283908348696e-05,
      "loss": 1.9003,
      "step": 6641
    },
    {
      "epoch": 0.17492757440084278,
      "grad_norm": 2.2412896156311035,
      "learning_rate": 4.126152225441138e-05,
      "loss": 1.8259,
      "step": 6642
    },
    {
      "epoch": 0.1749539109823545,
      "grad_norm": 2.9820196628570557,
      "learning_rate": 4.1260205425335794e-05,
      "loss": 0.8889,
      "step": 6643
    },
    {
      "epoch": 0.1749802475638662,
      "grad_norm": 5.168096542358398,
      "learning_rate": 4.125888859626021e-05,
      "loss": 2.1388,
      "step": 6644
    },
    {
      "epoch": 0.17500658414537792,
      "grad_norm": 1.8245837688446045,
      "learning_rate": 4.125757176718462e-05,
      "loss": 1.6325,
      "step": 6645
    },
    {
      "epoch": 0.17503292072688964,
      "grad_norm": 2.578749418258667,
      "learning_rate": 4.1256254938109034e-05,
      "loss": 1.847,
      "step": 6646
    },
    {
      "epoch": 0.17505925730840136,
      "grad_norm": 3.981011390686035,
      "learning_rate": 4.125493810903345e-05,
      "loss": 1.277,
      "step": 6647
    },
    {
      "epoch": 0.1750855938899131,
      "grad_norm": 2.3530519008636475,
      "learning_rate": 4.1253621279957865e-05,
      "loss": 1.7096,
      "step": 6648
    },
    {
      "epoch": 0.17511193047142481,
      "grad_norm": 1.8894442319869995,
      "learning_rate": 4.125230445088228e-05,
      "loss": 1.5303,
      "step": 6649
    },
    {
      "epoch": 0.17513826705293653,
      "grad_norm": 3.984992742538452,
      "learning_rate": 4.125098762180669e-05,
      "loss": 0.7781,
      "step": 6650
    },
    {
      "epoch": 0.17516460363444825,
      "grad_norm": 1.9461286067962646,
      "learning_rate": 4.1249670792731105e-05,
      "loss": 1.6609,
      "step": 6651
    },
    {
      "epoch": 0.17519094021595996,
      "grad_norm": 1.8759702444076538,
      "learning_rate": 4.1248353963655514e-05,
      "loss": 1.8718,
      "step": 6652
    },
    {
      "epoch": 0.17521727679747168,
      "grad_norm": 2.6989259719848633,
      "learning_rate": 4.1247037134579937e-05,
      "loss": 0.3704,
      "step": 6653
    },
    {
      "epoch": 0.17524361337898342,
      "grad_norm": 2.7729644775390625,
      "learning_rate": 4.1245720305504345e-05,
      "loss": 1.6543,
      "step": 6654
    },
    {
      "epoch": 0.17526994996049514,
      "grad_norm": 2.0761306285858154,
      "learning_rate": 4.124440347642876e-05,
      "loss": 1.8262,
      "step": 6655
    },
    {
      "epoch": 0.17529628654200685,
      "grad_norm": 1.3968690633773804,
      "learning_rate": 4.1243086647353177e-05,
      "loss": 1.8761,
      "step": 6656
    },
    {
      "epoch": 0.17532262312351857,
      "grad_norm": 3.094503164291382,
      "learning_rate": 4.1241769818277585e-05,
      "loss": 2.2206,
      "step": 6657
    },
    {
      "epoch": 0.17534895970503028,
      "grad_norm": 1.6809885501861572,
      "learning_rate": 4.124045298920201e-05,
      "loss": 1.7296,
      "step": 6658
    },
    {
      "epoch": 0.175375296286542,
      "grad_norm": 2.3383312225341797,
      "learning_rate": 4.1239136160126417e-05,
      "loss": 1.004,
      "step": 6659
    },
    {
      "epoch": 0.17540163286805371,
      "grad_norm": 1.7880600690841675,
      "learning_rate": 4.123781933105083e-05,
      "loss": 2.1348,
      "step": 6660
    },
    {
      "epoch": 0.17542796944956546,
      "grad_norm": 2.8334717750549316,
      "learning_rate": 4.123650250197524e-05,
      "loss": 1.7766,
      "step": 6661
    },
    {
      "epoch": 0.17545430603107717,
      "grad_norm": 2.4014248847961426,
      "learning_rate": 4.1235185672899663e-05,
      "loss": 1.8362,
      "step": 6662
    },
    {
      "epoch": 0.1754806426125889,
      "grad_norm": 1.9157991409301758,
      "learning_rate": 4.123386884382407e-05,
      "loss": 1.6226,
      "step": 6663
    },
    {
      "epoch": 0.1755069791941006,
      "grad_norm": 2.2434890270233154,
      "learning_rate": 4.123255201474849e-05,
      "loss": 2.0024,
      "step": 6664
    },
    {
      "epoch": 0.17553331577561232,
      "grad_norm": 2.862133264541626,
      "learning_rate": 4.1231235185672903e-05,
      "loss": 1.9178,
      "step": 6665
    },
    {
      "epoch": 0.17555965235712404,
      "grad_norm": 1.7017308473587036,
      "learning_rate": 4.122991835659731e-05,
      "loss": 1.5304,
      "step": 6666
    },
    {
      "epoch": 0.17558598893863578,
      "grad_norm": 3.603447437286377,
      "learning_rate": 4.1228601527521735e-05,
      "loss": 1.0318,
      "step": 6667
    },
    {
      "epoch": 0.1756123255201475,
      "grad_norm": 2.1857197284698486,
      "learning_rate": 4.1227284698446143e-05,
      "loss": 0.9257,
      "step": 6668
    },
    {
      "epoch": 0.1756386621016592,
      "grad_norm": 3.1623499393463135,
      "learning_rate": 4.122596786937056e-05,
      "loss": 1.9929,
      "step": 6669
    },
    {
      "epoch": 0.17566499868317093,
      "grad_norm": 1.7891215085983276,
      "learning_rate": 4.122465104029497e-05,
      "loss": 1.5044,
      "step": 6670
    },
    {
      "epoch": 0.17569133526468264,
      "grad_norm": 3.255054473876953,
      "learning_rate": 4.1223334211219383e-05,
      "loss": 1.9984,
      "step": 6671
    },
    {
      "epoch": 0.17571767184619436,
      "grad_norm": 1.884595513343811,
      "learning_rate": 4.12220173821438e-05,
      "loss": 1.8561,
      "step": 6672
    },
    {
      "epoch": 0.17574400842770607,
      "grad_norm": 2.8022518157958984,
      "learning_rate": 4.1220700553068215e-05,
      "loss": 1.6945,
      "step": 6673
    },
    {
      "epoch": 0.17577034500921782,
      "grad_norm": 2.192901611328125,
      "learning_rate": 4.121938372399263e-05,
      "loss": 2.0344,
      "step": 6674
    },
    {
      "epoch": 0.17579668159072953,
      "grad_norm": 2.163236379623413,
      "learning_rate": 4.121806689491704e-05,
      "loss": 1.7897,
      "step": 6675
    },
    {
      "epoch": 0.17582301817224125,
      "grad_norm": 1.5392736196517944,
      "learning_rate": 4.121675006584146e-05,
      "loss": 1.4971,
      "step": 6676
    },
    {
      "epoch": 0.17584935475375296,
      "grad_norm": 2.7338321208953857,
      "learning_rate": 4.121543323676587e-05,
      "loss": 1.2812,
      "step": 6677
    },
    {
      "epoch": 0.17587569133526468,
      "grad_norm": 2.262636184692383,
      "learning_rate": 4.1214116407690286e-05,
      "loss": 1.882,
      "step": 6678
    },
    {
      "epoch": 0.1759020279167764,
      "grad_norm": 4.032125473022461,
      "learning_rate": 4.1212799578614695e-05,
      "loss": 1.7786,
      "step": 6679
    },
    {
      "epoch": 0.1759283644982881,
      "grad_norm": 1.9584550857543945,
      "learning_rate": 4.121148274953911e-05,
      "loss": 1.105,
      "step": 6680
    },
    {
      "epoch": 0.17595470107979985,
      "grad_norm": 1.9815263748168945,
      "learning_rate": 4.1210165920463526e-05,
      "loss": 1.1418,
      "step": 6681
    },
    {
      "epoch": 0.17598103766131157,
      "grad_norm": 2.129667282104492,
      "learning_rate": 4.120884909138794e-05,
      "loss": 2.0092,
      "step": 6682
    },
    {
      "epoch": 0.17600737424282328,
      "grad_norm": 2.683131694793701,
      "learning_rate": 4.120753226231236e-05,
      "loss": 1.5317,
      "step": 6683
    },
    {
      "epoch": 0.176033710824335,
      "grad_norm": 2.292882204055786,
      "learning_rate": 4.1206215433236766e-05,
      "loss": 1.6714,
      "step": 6684
    },
    {
      "epoch": 0.17606004740584671,
      "grad_norm": 3.0176849365234375,
      "learning_rate": 4.120489860416118e-05,
      "loss": 1.8929,
      "step": 6685
    },
    {
      "epoch": 0.17608638398735843,
      "grad_norm": 2.804069757461548,
      "learning_rate": 4.12035817750856e-05,
      "loss": 1.4643,
      "step": 6686
    },
    {
      "epoch": 0.17611272056887017,
      "grad_norm": 1.796481728553772,
      "learning_rate": 4.120226494601001e-05,
      "loss": 1.7622,
      "step": 6687
    },
    {
      "epoch": 0.1761390571503819,
      "grad_norm": 1.8195126056671143,
      "learning_rate": 4.120094811693442e-05,
      "loss": 1.483,
      "step": 6688
    },
    {
      "epoch": 0.1761653937318936,
      "grad_norm": 3.9995951652526855,
      "learning_rate": 4.119963128785884e-05,
      "loss": 1.4732,
      "step": 6689
    },
    {
      "epoch": 0.17619173031340532,
      "grad_norm": 1.863304615020752,
      "learning_rate": 4.1198314458783246e-05,
      "loss": 1.7272,
      "step": 6690
    },
    {
      "epoch": 0.17621806689491704,
      "grad_norm": 2.2495763301849365,
      "learning_rate": 4.119699762970767e-05,
      "loss": 2.0331,
      "step": 6691
    },
    {
      "epoch": 0.17624440347642875,
      "grad_norm": 2.5434064865112305,
      "learning_rate": 4.119568080063208e-05,
      "loss": 1.7875,
      "step": 6692
    },
    {
      "epoch": 0.17627074005794047,
      "grad_norm": 2.723493814468384,
      "learning_rate": 4.119436397155649e-05,
      "loss": 1.4843,
      "step": 6693
    },
    {
      "epoch": 0.1762970766394522,
      "grad_norm": 1.5455756187438965,
      "learning_rate": 4.119304714248091e-05,
      "loss": 1.9669,
      "step": 6694
    },
    {
      "epoch": 0.17632341322096393,
      "grad_norm": 1.621565818786621,
      "learning_rate": 4.1191730313405324e-05,
      "loss": 1.7554,
      "step": 6695
    },
    {
      "epoch": 0.17634974980247564,
      "grad_norm": 2.664480209350586,
      "learning_rate": 4.119041348432974e-05,
      "loss": 1.3594,
      "step": 6696
    },
    {
      "epoch": 0.17637608638398736,
      "grad_norm": 3.1807620525360107,
      "learning_rate": 4.118909665525415e-05,
      "loss": 1.749,
      "step": 6697
    },
    {
      "epoch": 0.17640242296549907,
      "grad_norm": 1.6837384700775146,
      "learning_rate": 4.1187779826178564e-05,
      "loss": 2.1722,
      "step": 6698
    },
    {
      "epoch": 0.1764287595470108,
      "grad_norm": 2.1074440479278564,
      "learning_rate": 4.118646299710297e-05,
      "loss": 1.2512,
      "step": 6699
    },
    {
      "epoch": 0.1764550961285225,
      "grad_norm": 3.323525905609131,
      "learning_rate": 4.1185146168027395e-05,
      "loss": 1.4003,
      "step": 6700
    },
    {
      "epoch": 0.17648143271003425,
      "grad_norm": 2.9400622844696045,
      "learning_rate": 4.1183829338951804e-05,
      "loss": 1.7325,
      "step": 6701
    },
    {
      "epoch": 0.17650776929154596,
      "grad_norm": 3.638178586959839,
      "learning_rate": 4.118251250987622e-05,
      "loss": 1.1509,
      "step": 6702
    },
    {
      "epoch": 0.17653410587305768,
      "grad_norm": 1.7637969255447388,
      "learning_rate": 4.1181195680800635e-05,
      "loss": 2.327,
      "step": 6703
    },
    {
      "epoch": 0.1765604424545694,
      "grad_norm": 3.6068859100341797,
      "learning_rate": 4.1179878851725044e-05,
      "loss": 0.7452,
      "step": 6704
    },
    {
      "epoch": 0.1765867790360811,
      "grad_norm": 3.0534024238586426,
      "learning_rate": 4.1178562022649466e-05,
      "loss": 2.3666,
      "step": 6705
    },
    {
      "epoch": 0.17661311561759283,
      "grad_norm": 1.7166447639465332,
      "learning_rate": 4.1177245193573875e-05,
      "loss": 1.686,
      "step": 6706
    },
    {
      "epoch": 0.17663945219910457,
      "grad_norm": 2.3784408569335938,
      "learning_rate": 4.117592836449829e-05,
      "loss": 1.6467,
      "step": 6707
    },
    {
      "epoch": 0.17666578878061628,
      "grad_norm": 4.3146820068359375,
      "learning_rate": 4.11746115354227e-05,
      "loss": 1.5173,
      "step": 6708
    },
    {
      "epoch": 0.176692125362128,
      "grad_norm": 1.9582109451293945,
      "learning_rate": 4.117329470634712e-05,
      "loss": 1.898,
      "step": 6709
    },
    {
      "epoch": 0.17671846194363972,
      "grad_norm": 1.8350038528442383,
      "learning_rate": 4.117197787727153e-05,
      "loss": 1.7136,
      "step": 6710
    },
    {
      "epoch": 0.17674479852515143,
      "grad_norm": 1.874682903289795,
      "learning_rate": 4.1170661048195946e-05,
      "loss": 1.9957,
      "step": 6711
    },
    {
      "epoch": 0.17677113510666315,
      "grad_norm": 3.145806074142456,
      "learning_rate": 4.116934421912036e-05,
      "loss": 0.7783,
      "step": 6712
    },
    {
      "epoch": 0.17679747168817486,
      "grad_norm": 3.466364622116089,
      "learning_rate": 4.116802739004477e-05,
      "loss": 1.9053,
      "step": 6713
    },
    {
      "epoch": 0.1768238082696866,
      "grad_norm": 2.290138006210327,
      "learning_rate": 4.116671056096919e-05,
      "loss": 0.4749,
      "step": 6714
    },
    {
      "epoch": 0.17685014485119832,
      "grad_norm": 4.1517767906188965,
      "learning_rate": 4.11653937318936e-05,
      "loss": 2.096,
      "step": 6715
    },
    {
      "epoch": 0.17687648143271004,
      "grad_norm": 1.7356162071228027,
      "learning_rate": 4.116407690281802e-05,
      "loss": 2.6511,
      "step": 6716
    },
    {
      "epoch": 0.17690281801422175,
      "grad_norm": 3.1292381286621094,
      "learning_rate": 4.1162760073742426e-05,
      "loss": 1.3981,
      "step": 6717
    },
    {
      "epoch": 0.17692915459573347,
      "grad_norm": 3.1809940338134766,
      "learning_rate": 4.116144324466684e-05,
      "loss": 1.2252,
      "step": 6718
    },
    {
      "epoch": 0.17695549117724518,
      "grad_norm": 2.8673181533813477,
      "learning_rate": 4.116012641559126e-05,
      "loss": 1.6321,
      "step": 6719
    },
    {
      "epoch": 0.17698182775875693,
      "grad_norm": 4.243430137634277,
      "learning_rate": 4.115880958651567e-05,
      "loss": 1.9734,
      "step": 6720
    },
    {
      "epoch": 0.17700816434026864,
      "grad_norm": 3.3914794921875,
      "learning_rate": 4.115749275744009e-05,
      "loss": 2.1519,
      "step": 6721
    },
    {
      "epoch": 0.17703450092178036,
      "grad_norm": 4.168887138366699,
      "learning_rate": 4.11561759283645e-05,
      "loss": 0.6254,
      "step": 6722
    },
    {
      "epoch": 0.17706083750329207,
      "grad_norm": 1.6878275871276855,
      "learning_rate": 4.115485909928891e-05,
      "loss": 1.9985,
      "step": 6723
    },
    {
      "epoch": 0.1770871740848038,
      "grad_norm": 2.0527000427246094,
      "learning_rate": 4.115354227021333e-05,
      "loss": 1.8863,
      "step": 6724
    },
    {
      "epoch": 0.1771135106663155,
      "grad_norm": 1.7715539932250977,
      "learning_rate": 4.1152225441137744e-05,
      "loss": 2.1905,
      "step": 6725
    },
    {
      "epoch": 0.17713984724782722,
      "grad_norm": 2.5337605476379395,
      "learning_rate": 4.115090861206215e-05,
      "loss": 2.1008,
      "step": 6726
    },
    {
      "epoch": 0.17716618382933896,
      "grad_norm": 2.2231197357177734,
      "learning_rate": 4.114959178298657e-05,
      "loss": 1.4889,
      "step": 6727
    },
    {
      "epoch": 0.17719252041085068,
      "grad_norm": 2.2031188011169434,
      "learning_rate": 4.1148274953910984e-05,
      "loss": 2.1719,
      "step": 6728
    },
    {
      "epoch": 0.1772188569923624,
      "grad_norm": 2.0154097080230713,
      "learning_rate": 4.11469581248354e-05,
      "loss": 1.4547,
      "step": 6729
    },
    {
      "epoch": 0.1772451935738741,
      "grad_norm": 1.9133888483047485,
      "learning_rate": 4.1145641295759816e-05,
      "loss": 1.7201,
      "step": 6730
    },
    {
      "epoch": 0.17727153015538583,
      "grad_norm": 2.965532064437866,
      "learning_rate": 4.1144324466684224e-05,
      "loss": 1.8718,
      "step": 6731
    },
    {
      "epoch": 0.17729786673689754,
      "grad_norm": 5.10041618347168,
      "learning_rate": 4.114300763760864e-05,
      "loss": 1.0623,
      "step": 6732
    },
    {
      "epoch": 0.17732420331840926,
      "grad_norm": 2.9344160556793213,
      "learning_rate": 4.1141690808533056e-05,
      "loss": 2.0449,
      "step": 6733
    },
    {
      "epoch": 0.177350539899921,
      "grad_norm": 2.2921764850616455,
      "learning_rate": 4.114037397945747e-05,
      "loss": 1.422,
      "step": 6734
    },
    {
      "epoch": 0.17737687648143272,
      "grad_norm": 4.644622325897217,
      "learning_rate": 4.113905715038188e-05,
      "loss": 2.4622,
      "step": 6735
    },
    {
      "epoch": 0.17740321306294443,
      "grad_norm": 1.6904956102371216,
      "learning_rate": 4.1137740321306296e-05,
      "loss": 1.9558,
      "step": 6736
    },
    {
      "epoch": 0.17742954964445615,
      "grad_norm": 2.1949121952056885,
      "learning_rate": 4.113642349223071e-05,
      "loss": 1.4928,
      "step": 6737
    },
    {
      "epoch": 0.17745588622596786,
      "grad_norm": 2.7083094120025635,
      "learning_rate": 4.113510666315513e-05,
      "loss": 1.7994,
      "step": 6738
    },
    {
      "epoch": 0.17748222280747958,
      "grad_norm": 2.8347067832946777,
      "learning_rate": 4.113378983407954e-05,
      "loss": 1.7722,
      "step": 6739
    },
    {
      "epoch": 0.17750855938899132,
      "grad_norm": 1.8640238046646118,
      "learning_rate": 4.113247300500395e-05,
      "loss": 1.6394,
      "step": 6740
    },
    {
      "epoch": 0.17753489597050304,
      "grad_norm": 4.407964706420898,
      "learning_rate": 4.113115617592837e-05,
      "loss": 1.5077,
      "step": 6741
    },
    {
      "epoch": 0.17756123255201475,
      "grad_norm": 2.251040458679199,
      "learning_rate": 4.112983934685278e-05,
      "loss": 2.3118,
      "step": 6742
    },
    {
      "epoch": 0.17758756913352647,
      "grad_norm": 3.1646571159362793,
      "learning_rate": 4.11285225177772e-05,
      "loss": 1.354,
      "step": 6743
    },
    {
      "epoch": 0.17761390571503818,
      "grad_norm": 2.2493324279785156,
      "learning_rate": 4.112720568870161e-05,
      "loss": 2.8635,
      "step": 6744
    },
    {
      "epoch": 0.1776402422965499,
      "grad_norm": 2.4689788818359375,
      "learning_rate": 4.112588885962602e-05,
      "loss": 0.6435,
      "step": 6745
    },
    {
      "epoch": 0.17766657887806162,
      "grad_norm": 1.9772800207138062,
      "learning_rate": 4.112457203055043e-05,
      "loss": 1.9938,
      "step": 6746
    },
    {
      "epoch": 0.17769291545957336,
      "grad_norm": 2.447547674179077,
      "learning_rate": 4.1123255201474854e-05,
      "loss": 1.2604,
      "step": 6747
    },
    {
      "epoch": 0.17771925204108507,
      "grad_norm": 2.0492193698883057,
      "learning_rate": 4.112193837239926e-05,
      "loss": 2.1748,
      "step": 6748
    },
    {
      "epoch": 0.1777455886225968,
      "grad_norm": 2.1823806762695312,
      "learning_rate": 4.112062154332368e-05,
      "loss": 1.7454,
      "step": 6749
    },
    {
      "epoch": 0.1777719252041085,
      "grad_norm": 2.7477235794067383,
      "learning_rate": 4.1119304714248094e-05,
      "loss": 1.4454,
      "step": 6750
    },
    {
      "epoch": 0.17779826178562022,
      "grad_norm": 3.9620254039764404,
      "learning_rate": 4.11179878851725e-05,
      "loss": 1.5756,
      "step": 6751
    },
    {
      "epoch": 0.17782459836713194,
      "grad_norm": 1.6999497413635254,
      "learning_rate": 4.1116671056096925e-05,
      "loss": 1.5814,
      "step": 6752
    },
    {
      "epoch": 0.17785093494864368,
      "grad_norm": 2.3639976978302,
      "learning_rate": 4.1115354227021334e-05,
      "loss": 1.5001,
      "step": 6753
    },
    {
      "epoch": 0.1778772715301554,
      "grad_norm": 1.8791390657424927,
      "learning_rate": 4.111403739794575e-05,
      "loss": 1.9452,
      "step": 6754
    },
    {
      "epoch": 0.1779036081116671,
      "grad_norm": 4.2245073318481445,
      "learning_rate": 4.111272056887016e-05,
      "loss": 1.6914,
      "step": 6755
    },
    {
      "epoch": 0.17792994469317883,
      "grad_norm": 2.0726189613342285,
      "learning_rate": 4.1111403739794574e-05,
      "loss": 1.47,
      "step": 6756
    },
    {
      "epoch": 0.17795628127469054,
      "grad_norm": 3.5650253295898438,
      "learning_rate": 4.111008691071899e-05,
      "loss": 1.9222,
      "step": 6757
    },
    {
      "epoch": 0.17798261785620226,
      "grad_norm": 2.593458890914917,
      "learning_rate": 4.1108770081643405e-05,
      "loss": 1.6603,
      "step": 6758
    },
    {
      "epoch": 0.17800895443771397,
      "grad_norm": 2.0861594676971436,
      "learning_rate": 4.110745325256782e-05,
      "loss": 1.9041,
      "step": 6759
    },
    {
      "epoch": 0.17803529101922572,
      "grad_norm": 3.4253945350646973,
      "learning_rate": 4.110613642349223e-05,
      "loss": 1.5267,
      "step": 6760
    },
    {
      "epoch": 0.17806162760073743,
      "grad_norm": 2.0704896450042725,
      "learning_rate": 4.110481959441665e-05,
      "loss": 1.9616,
      "step": 6761
    },
    {
      "epoch": 0.17808796418224915,
      "grad_norm": 2.439908504486084,
      "learning_rate": 4.110350276534106e-05,
      "loss": 0.9929,
      "step": 6762
    },
    {
      "epoch": 0.17811430076376086,
      "grad_norm": 3.2816014289855957,
      "learning_rate": 4.1102185936265476e-05,
      "loss": 1.6153,
      "step": 6763
    },
    {
      "epoch": 0.17814063734527258,
      "grad_norm": 2.4091029167175293,
      "learning_rate": 4.1100869107189885e-05,
      "loss": 1.6636,
      "step": 6764
    },
    {
      "epoch": 0.1781669739267843,
      "grad_norm": 1.7577128410339355,
      "learning_rate": 4.10995522781143e-05,
      "loss": 1.8762,
      "step": 6765
    },
    {
      "epoch": 0.178193310508296,
      "grad_norm": 1.443149209022522,
      "learning_rate": 4.1098235449038716e-05,
      "loss": 1.746,
      "step": 6766
    },
    {
      "epoch": 0.17821964708980775,
      "grad_norm": 4.384613990783691,
      "learning_rate": 4.109691861996313e-05,
      "loss": 1.8637,
      "step": 6767
    },
    {
      "epoch": 0.17824598367131947,
      "grad_norm": 3.624572277069092,
      "learning_rate": 4.109560179088755e-05,
      "loss": 0.8891,
      "step": 6768
    },
    {
      "epoch": 0.17827232025283118,
      "grad_norm": 1.9998635053634644,
      "learning_rate": 4.1094284961811956e-05,
      "loss": 2.0141,
      "step": 6769
    },
    {
      "epoch": 0.1782986568343429,
      "grad_norm": 2.157447099685669,
      "learning_rate": 4.109296813273637e-05,
      "loss": 1.6568,
      "step": 6770
    },
    {
      "epoch": 0.17832499341585462,
      "grad_norm": 2.3195464611053467,
      "learning_rate": 4.109165130366079e-05,
      "loss": 1.3678,
      "step": 6771
    },
    {
      "epoch": 0.17835132999736633,
      "grad_norm": 4.908475399017334,
      "learning_rate": 4.10903344745852e-05,
      "loss": 1.6389,
      "step": 6772
    },
    {
      "epoch": 0.17837766657887807,
      "grad_norm": 2.4665000438690186,
      "learning_rate": 4.108901764550961e-05,
      "loss": 1.854,
      "step": 6773
    },
    {
      "epoch": 0.1784040031603898,
      "grad_norm": 4.117772102355957,
      "learning_rate": 4.108770081643403e-05,
      "loss": 1.8473,
      "step": 6774
    },
    {
      "epoch": 0.1784303397419015,
      "grad_norm": 2.614429235458374,
      "learning_rate": 4.108638398735844e-05,
      "loss": 1.6584,
      "step": 6775
    },
    {
      "epoch": 0.17845667632341322,
      "grad_norm": 4.220699787139893,
      "learning_rate": 4.108506715828286e-05,
      "loss": 2.3101,
      "step": 6776
    },
    {
      "epoch": 0.17848301290492494,
      "grad_norm": 5.078653812408447,
      "learning_rate": 4.1083750329207274e-05,
      "loss": 1.9481,
      "step": 6777
    },
    {
      "epoch": 0.17850934948643665,
      "grad_norm": 2.256284713745117,
      "learning_rate": 4.108243350013168e-05,
      "loss": 1.8684,
      "step": 6778
    },
    {
      "epoch": 0.17853568606794837,
      "grad_norm": 1.7752957344055176,
      "learning_rate": 4.10811166710561e-05,
      "loss": 1.3696,
      "step": 6779
    },
    {
      "epoch": 0.1785620226494601,
      "grad_norm": 1.5159751176834106,
      "learning_rate": 4.1079799841980514e-05,
      "loss": 1.4898,
      "step": 6780
    },
    {
      "epoch": 0.17858835923097183,
      "grad_norm": 2.9525413513183594,
      "learning_rate": 4.107848301290493e-05,
      "loss": 1.5798,
      "step": 6781
    },
    {
      "epoch": 0.17861469581248354,
      "grad_norm": 3.0003139972686768,
      "learning_rate": 4.107716618382934e-05,
      "loss": 1.2852,
      "step": 6782
    },
    {
      "epoch": 0.17864103239399526,
      "grad_norm": 2.001039505004883,
      "learning_rate": 4.1075849354753754e-05,
      "loss": 0.5788,
      "step": 6783
    },
    {
      "epoch": 0.17866736897550697,
      "grad_norm": 2.3685619831085205,
      "learning_rate": 4.107453252567817e-05,
      "loss": 1.7855,
      "step": 6784
    },
    {
      "epoch": 0.1786937055570187,
      "grad_norm": 1.8843557834625244,
      "learning_rate": 4.1073215696602585e-05,
      "loss": 1.5882,
      "step": 6785
    },
    {
      "epoch": 0.1787200421385304,
      "grad_norm": 1.9563547372817993,
      "learning_rate": 4.1071898867527e-05,
      "loss": 2.5971,
      "step": 6786
    },
    {
      "epoch": 0.17874637872004215,
      "grad_norm": 2.013260841369629,
      "learning_rate": 4.107058203845141e-05,
      "loss": 1.8646,
      "step": 6787
    },
    {
      "epoch": 0.17877271530155386,
      "grad_norm": 3.0720396041870117,
      "learning_rate": 4.1069265209375825e-05,
      "loss": 1.9464,
      "step": 6788
    },
    {
      "epoch": 0.17879905188306558,
      "grad_norm": 2.703814744949341,
      "learning_rate": 4.1067948380300234e-05,
      "loss": 1.3062,
      "step": 6789
    },
    {
      "epoch": 0.1788253884645773,
      "grad_norm": 1.8566234111785889,
      "learning_rate": 4.1066631551224657e-05,
      "loss": 1.7328,
      "step": 6790
    },
    {
      "epoch": 0.178851725046089,
      "grad_norm": 1.9944411516189575,
      "learning_rate": 4.1065314722149065e-05,
      "loss": 1.4583,
      "step": 6791
    },
    {
      "epoch": 0.17887806162760073,
      "grad_norm": 2.6147239208221436,
      "learning_rate": 4.106399789307348e-05,
      "loss": 1.3663,
      "step": 6792
    },
    {
      "epoch": 0.17890439820911247,
      "grad_norm": 2.9063587188720703,
      "learning_rate": 4.106268106399789e-05,
      "loss": 1.6115,
      "step": 6793
    },
    {
      "epoch": 0.17893073479062419,
      "grad_norm": 4.7841339111328125,
      "learning_rate": 4.106136423492231e-05,
      "loss": 2.4867,
      "step": 6794
    },
    {
      "epoch": 0.1789570713721359,
      "grad_norm": 3.1169638633728027,
      "learning_rate": 4.106004740584672e-05,
      "loss": 1.2408,
      "step": 6795
    },
    {
      "epoch": 0.17898340795364762,
      "grad_norm": 2.0345919132232666,
      "learning_rate": 4.105873057677114e-05,
      "loss": 0.9196,
      "step": 6796
    },
    {
      "epoch": 0.17900974453515933,
      "grad_norm": 2.243893623352051,
      "learning_rate": 4.105741374769555e-05,
      "loss": 1.7777,
      "step": 6797
    },
    {
      "epoch": 0.17903608111667105,
      "grad_norm": 1.65420663356781,
      "learning_rate": 4.105609691861996e-05,
      "loss": 2.1784,
      "step": 6798
    },
    {
      "epoch": 0.17906241769818276,
      "grad_norm": 3.017810344696045,
      "learning_rate": 4.1054780089544383e-05,
      "loss": 1.3427,
      "step": 6799
    },
    {
      "epoch": 0.1790887542796945,
      "grad_norm": 2.7063002586364746,
      "learning_rate": 4.105346326046879e-05,
      "loss": 1.8331,
      "step": 6800
    },
    {
      "epoch": 0.17911509086120622,
      "grad_norm": 2.237499952316284,
      "learning_rate": 4.105214643139321e-05,
      "loss": 1.9497,
      "step": 6801
    },
    {
      "epoch": 0.17914142744271794,
      "grad_norm": 2.082740545272827,
      "learning_rate": 4.105082960231762e-05,
      "loss": 1.4827,
      "step": 6802
    },
    {
      "epoch": 0.17916776402422965,
      "grad_norm": 2.144864559173584,
      "learning_rate": 4.104951277324203e-05,
      "loss": 1.8197,
      "step": 6803
    },
    {
      "epoch": 0.17919410060574137,
      "grad_norm": 2.2020671367645264,
      "learning_rate": 4.104819594416645e-05,
      "loss": 2.1981,
      "step": 6804
    },
    {
      "epoch": 0.17922043718725308,
      "grad_norm": 2.2199628353118896,
      "learning_rate": 4.1046879115090863e-05,
      "loss": 1.7829,
      "step": 6805
    },
    {
      "epoch": 0.17924677376876483,
      "grad_norm": 2.0566678047180176,
      "learning_rate": 4.104556228601528e-05,
      "loss": 1.9086,
      "step": 6806
    },
    {
      "epoch": 0.17927311035027654,
      "grad_norm": 1.9605798721313477,
      "learning_rate": 4.104424545693969e-05,
      "loss": 2.248,
      "step": 6807
    },
    {
      "epoch": 0.17929944693178826,
      "grad_norm": 1.7758398056030273,
      "learning_rate": 4.104292862786411e-05,
      "loss": 2.2688,
      "step": 6808
    },
    {
      "epoch": 0.17932578351329997,
      "grad_norm": 3.7575817108154297,
      "learning_rate": 4.104161179878852e-05,
      "loss": 2.3091,
      "step": 6809
    },
    {
      "epoch": 0.1793521200948117,
      "grad_norm": 4.644641399383545,
      "learning_rate": 4.1040294969712935e-05,
      "loss": 1.3487,
      "step": 6810
    },
    {
      "epoch": 0.1793784566763234,
      "grad_norm": 1.894126057624817,
      "learning_rate": 4.1038978140637344e-05,
      "loss": 1.6621,
      "step": 6811
    },
    {
      "epoch": 0.17940479325783512,
      "grad_norm": 2.8985049724578857,
      "learning_rate": 4.103766131156176e-05,
      "loss": 1.1934,
      "step": 6812
    },
    {
      "epoch": 0.17943112983934686,
      "grad_norm": 1.7191641330718994,
      "learning_rate": 4.1036344482486175e-05,
      "loss": 1.6747,
      "step": 6813
    },
    {
      "epoch": 0.17945746642085858,
      "grad_norm": 1.8535321950912476,
      "learning_rate": 4.103502765341059e-05,
      "loss": 2.0091,
      "step": 6814
    },
    {
      "epoch": 0.1794838030023703,
      "grad_norm": 2.8188588619232178,
      "learning_rate": 4.1033710824335006e-05,
      "loss": 1.3508,
      "step": 6815
    },
    {
      "epoch": 0.179510139583882,
      "grad_norm": 2.948801040649414,
      "learning_rate": 4.1032393995259415e-05,
      "loss": 0.8262,
      "step": 6816
    },
    {
      "epoch": 0.17953647616539373,
      "grad_norm": 3.9432289600372314,
      "learning_rate": 4.103107716618383e-05,
      "loss": 1.2866,
      "step": 6817
    },
    {
      "epoch": 0.17956281274690544,
      "grad_norm": 2.3603694438934326,
      "learning_rate": 4.1029760337108246e-05,
      "loss": 2.5574,
      "step": 6818
    },
    {
      "epoch": 0.17958914932841716,
      "grad_norm": 3.932704448699951,
      "learning_rate": 4.102844350803266e-05,
      "loss": 1.8841,
      "step": 6819
    },
    {
      "epoch": 0.1796154859099289,
      "grad_norm": 1.7816461324691772,
      "learning_rate": 4.102712667895707e-05,
      "loss": 2.3823,
      "step": 6820
    },
    {
      "epoch": 0.17964182249144062,
      "grad_norm": 1.6622763872146606,
      "learning_rate": 4.1025809849881486e-05,
      "loss": 1.8831,
      "step": 6821
    },
    {
      "epoch": 0.17966815907295233,
      "grad_norm": 2.4232850074768066,
      "learning_rate": 4.10244930208059e-05,
      "loss": 1.3076,
      "step": 6822
    },
    {
      "epoch": 0.17969449565446405,
      "grad_norm": 3.414458990097046,
      "learning_rate": 4.102317619173032e-05,
      "loss": 1.8218,
      "step": 6823
    },
    {
      "epoch": 0.17972083223597576,
      "grad_norm": 1.437122106552124,
      "learning_rate": 4.102185936265473e-05,
      "loss": 1.7803,
      "step": 6824
    },
    {
      "epoch": 0.17974716881748748,
      "grad_norm": 1.7008097171783447,
      "learning_rate": 4.102054253357914e-05,
      "loss": 1.7539,
      "step": 6825
    },
    {
      "epoch": 0.17977350539899922,
      "grad_norm": 6.7796950340271,
      "learning_rate": 4.101922570450356e-05,
      "loss": 1.5306,
      "step": 6826
    },
    {
      "epoch": 0.17979984198051094,
      "grad_norm": 2.5783803462982178,
      "learning_rate": 4.101790887542797e-05,
      "loss": 1.862,
      "step": 6827
    },
    {
      "epoch": 0.17982617856202265,
      "grad_norm": 1.9919512271881104,
      "learning_rate": 4.101659204635239e-05,
      "loss": 1.4089,
      "step": 6828
    },
    {
      "epoch": 0.17985251514353437,
      "grad_norm": 1.8648899793624878,
      "learning_rate": 4.10152752172768e-05,
      "loss": 0.4449,
      "step": 6829
    },
    {
      "epoch": 0.17987885172504608,
      "grad_norm": 3.0019474029541016,
      "learning_rate": 4.101395838820121e-05,
      "loss": 1.8668,
      "step": 6830
    },
    {
      "epoch": 0.1799051883065578,
      "grad_norm": 2.0431602001190186,
      "learning_rate": 4.101264155912563e-05,
      "loss": 0.9655,
      "step": 6831
    },
    {
      "epoch": 0.17993152488806952,
      "grad_norm": 1.9050644636154175,
      "learning_rate": 4.1011324730050044e-05,
      "loss": 2.3416,
      "step": 6832
    },
    {
      "epoch": 0.17995786146958126,
      "grad_norm": 1.5316270589828491,
      "learning_rate": 4.101000790097446e-05,
      "loss": 2.0118,
      "step": 6833
    },
    {
      "epoch": 0.17998419805109298,
      "grad_norm": 2.86852765083313,
      "learning_rate": 4.100869107189887e-05,
      "loss": 2.6354,
      "step": 6834
    },
    {
      "epoch": 0.1800105346326047,
      "grad_norm": 2.293701171875,
      "learning_rate": 4.1007374242823284e-05,
      "loss": 1.7266,
      "step": 6835
    },
    {
      "epoch": 0.1800368712141164,
      "grad_norm": 1.7296762466430664,
      "learning_rate": 4.100605741374769e-05,
      "loss": 1.4435,
      "step": 6836
    },
    {
      "epoch": 0.18006320779562812,
      "grad_norm": 1.9229036569595337,
      "learning_rate": 4.1004740584672115e-05,
      "loss": 2.377,
      "step": 6837
    },
    {
      "epoch": 0.18008954437713984,
      "grad_norm": 2.495021104812622,
      "learning_rate": 4.1003423755596524e-05,
      "loss": 2.3578,
      "step": 6838
    },
    {
      "epoch": 0.18011588095865158,
      "grad_norm": 2.3712573051452637,
      "learning_rate": 4.100210692652094e-05,
      "loss": 1.1256,
      "step": 6839
    },
    {
      "epoch": 0.1801422175401633,
      "grad_norm": 1.759839415550232,
      "learning_rate": 4.1000790097445355e-05,
      "loss": 2.0299,
      "step": 6840
    },
    {
      "epoch": 0.180168554121675,
      "grad_norm": 5.818070411682129,
      "learning_rate": 4.099947326836977e-05,
      "loss": 2.027,
      "step": 6841
    },
    {
      "epoch": 0.18019489070318673,
      "grad_norm": 1.5625042915344238,
      "learning_rate": 4.0998156439294186e-05,
      "loss": 1.9021,
      "step": 6842
    },
    {
      "epoch": 0.18022122728469844,
      "grad_norm": 1.7875632047653198,
      "learning_rate": 4.0996839610218595e-05,
      "loss": 1.9985,
      "step": 6843
    },
    {
      "epoch": 0.18024756386621016,
      "grad_norm": 6.783160209655762,
      "learning_rate": 4.099552278114301e-05,
      "loss": 0.8677,
      "step": 6844
    },
    {
      "epoch": 0.18027390044772187,
      "grad_norm": 2.866274356842041,
      "learning_rate": 4.099420595206742e-05,
      "loss": 2.0127,
      "step": 6845
    },
    {
      "epoch": 0.18030023702923362,
      "grad_norm": 2.130305767059326,
      "learning_rate": 4.099288912299184e-05,
      "loss": 2.098,
      "step": 6846
    },
    {
      "epoch": 0.18032657361074533,
      "grad_norm": 1.7019449472427368,
      "learning_rate": 4.099157229391625e-05,
      "loss": 1.7701,
      "step": 6847
    },
    {
      "epoch": 0.18035291019225705,
      "grad_norm": 3.656829595565796,
      "learning_rate": 4.0990255464840666e-05,
      "loss": 1.7441,
      "step": 6848
    },
    {
      "epoch": 0.18037924677376876,
      "grad_norm": 3.3117520809173584,
      "learning_rate": 4.0988938635765075e-05,
      "loss": 0.6602,
      "step": 6849
    },
    {
      "epoch": 0.18040558335528048,
      "grad_norm": 3.709299325942993,
      "learning_rate": 4.098762180668949e-05,
      "loss": 0.6624,
      "step": 6850
    },
    {
      "epoch": 0.1804319199367922,
      "grad_norm": 1.9445096254348755,
      "learning_rate": 4.0986304977613906e-05,
      "loss": 1.696,
      "step": 6851
    },
    {
      "epoch": 0.1804582565183039,
      "grad_norm": 1.9931341409683228,
      "learning_rate": 4.098498814853832e-05,
      "loss": 1.5443,
      "step": 6852
    },
    {
      "epoch": 0.18048459309981565,
      "grad_norm": 3.580620765686035,
      "learning_rate": 4.098367131946274e-05,
      "loss": 1.9921,
      "step": 6853
    },
    {
      "epoch": 0.18051092968132737,
      "grad_norm": 2.829792022705078,
      "learning_rate": 4.0982354490387146e-05,
      "loss": 2.0868,
      "step": 6854
    },
    {
      "epoch": 0.18053726626283909,
      "grad_norm": 1.9883331060409546,
      "learning_rate": 4.098103766131156e-05,
      "loss": 1.8577,
      "step": 6855
    },
    {
      "epoch": 0.1805636028443508,
      "grad_norm": 2.3210837841033936,
      "learning_rate": 4.097972083223598e-05,
      "loss": 1.7687,
      "step": 6856
    },
    {
      "epoch": 0.18058993942586252,
      "grad_norm": 1.977243185043335,
      "learning_rate": 4.097840400316039e-05,
      "loss": 1.9435,
      "step": 6857
    },
    {
      "epoch": 0.18061627600737423,
      "grad_norm": 2.4407622814178467,
      "learning_rate": 4.09770871740848e-05,
      "loss": 2.161,
      "step": 6858
    },
    {
      "epoch": 0.18064261258888598,
      "grad_norm": 1.699407696723938,
      "learning_rate": 4.097577034500922e-05,
      "loss": 1.2125,
      "step": 6859
    },
    {
      "epoch": 0.1806689491703977,
      "grad_norm": 1.610184669494629,
      "learning_rate": 4.097445351593363e-05,
      "loss": 1.8904,
      "step": 6860
    },
    {
      "epoch": 0.1806952857519094,
      "grad_norm": 1.7996221780776978,
      "learning_rate": 4.097313668685805e-05,
      "loss": 1.5021,
      "step": 6861
    },
    {
      "epoch": 0.18072162233342112,
      "grad_norm": 2.915076732635498,
      "learning_rate": 4.0971819857782464e-05,
      "loss": 2.0225,
      "step": 6862
    },
    {
      "epoch": 0.18074795891493284,
      "grad_norm": 1.4015133380889893,
      "learning_rate": 4.097050302870687e-05,
      "loss": 1.8607,
      "step": 6863
    },
    {
      "epoch": 0.18077429549644455,
      "grad_norm": 2.6054396629333496,
      "learning_rate": 4.096918619963129e-05,
      "loss": 0.9116,
      "step": 6864
    },
    {
      "epoch": 0.18080063207795627,
      "grad_norm": 1.9150731563568115,
      "learning_rate": 4.0967869370555704e-05,
      "loss": 1.885,
      "step": 6865
    },
    {
      "epoch": 0.180826968659468,
      "grad_norm": 2.529771327972412,
      "learning_rate": 4.096655254148012e-05,
      "loss": 1.9681,
      "step": 6866
    },
    {
      "epoch": 0.18085330524097973,
      "grad_norm": 2.0468051433563232,
      "learning_rate": 4.096523571240453e-05,
      "loss": 1.6722,
      "step": 6867
    },
    {
      "epoch": 0.18087964182249144,
      "grad_norm": 1.804702639579773,
      "learning_rate": 4.0963918883328944e-05,
      "loss": 1.9368,
      "step": 6868
    },
    {
      "epoch": 0.18090597840400316,
      "grad_norm": 1.5351810455322266,
      "learning_rate": 4.096260205425336e-05,
      "loss": 2.055,
      "step": 6869
    },
    {
      "epoch": 0.18093231498551487,
      "grad_norm": 1.6544311046600342,
      "learning_rate": 4.0961285225177776e-05,
      "loss": 1.7408,
      "step": 6870
    },
    {
      "epoch": 0.1809586515670266,
      "grad_norm": 2.6199283599853516,
      "learning_rate": 4.095996839610219e-05,
      "loss": 1.8655,
      "step": 6871
    },
    {
      "epoch": 0.1809849881485383,
      "grad_norm": 1.7403453588485718,
      "learning_rate": 4.09586515670266e-05,
      "loss": 2.0365,
      "step": 6872
    },
    {
      "epoch": 0.18101132473005005,
      "grad_norm": 2.176819086074829,
      "learning_rate": 4.0957334737951016e-05,
      "loss": 0.8515,
      "step": 6873
    },
    {
      "epoch": 0.18103766131156176,
      "grad_norm": 1.74526846408844,
      "learning_rate": 4.095601790887543e-05,
      "loss": 1.6314,
      "step": 6874
    },
    {
      "epoch": 0.18106399789307348,
      "grad_norm": 2.622490406036377,
      "learning_rate": 4.095470107979985e-05,
      "loss": 2.0877,
      "step": 6875
    },
    {
      "epoch": 0.1810903344745852,
      "grad_norm": 2.2714836597442627,
      "learning_rate": 4.0953384250724256e-05,
      "loss": 2.3627,
      "step": 6876
    },
    {
      "epoch": 0.1811166710560969,
      "grad_norm": 2.257474422454834,
      "learning_rate": 4.095206742164867e-05,
      "loss": 2.0761,
      "step": 6877
    },
    {
      "epoch": 0.18114300763760863,
      "grad_norm": 1.8922864198684692,
      "learning_rate": 4.095075059257309e-05,
      "loss": 2.2427,
      "step": 6878
    },
    {
      "epoch": 0.18116934421912037,
      "grad_norm": 3.556945323944092,
      "learning_rate": 4.09494337634975e-05,
      "loss": 1.3572,
      "step": 6879
    },
    {
      "epoch": 0.1811956808006321,
      "grad_norm": 3.8870959281921387,
      "learning_rate": 4.094811693442192e-05,
      "loss": 0.7372,
      "step": 6880
    },
    {
      "epoch": 0.1812220173821438,
      "grad_norm": 2.645507335662842,
      "learning_rate": 4.094680010534633e-05,
      "loss": 1.5219,
      "step": 6881
    },
    {
      "epoch": 0.18124835396365552,
      "grad_norm": 2.594209909439087,
      "learning_rate": 4.094548327627074e-05,
      "loss": 1.1077,
      "step": 6882
    },
    {
      "epoch": 0.18127469054516723,
      "grad_norm": 2.21266508102417,
      "learning_rate": 4.094416644719515e-05,
      "loss": 1.7842,
      "step": 6883
    },
    {
      "epoch": 0.18130102712667895,
      "grad_norm": 3.9246649742126465,
      "learning_rate": 4.0942849618119574e-05,
      "loss": 0.9665,
      "step": 6884
    },
    {
      "epoch": 0.18132736370819066,
      "grad_norm": 2.080536365509033,
      "learning_rate": 4.094153278904398e-05,
      "loss": 2.0872,
      "step": 6885
    },
    {
      "epoch": 0.1813537002897024,
      "grad_norm": 2.383018970489502,
      "learning_rate": 4.09402159599684e-05,
      "loss": 1.2582,
      "step": 6886
    },
    {
      "epoch": 0.18138003687121412,
      "grad_norm": 2.284088373184204,
      "learning_rate": 4.0938899130892814e-05,
      "loss": 2.1995,
      "step": 6887
    },
    {
      "epoch": 0.18140637345272584,
      "grad_norm": 2.257924795150757,
      "learning_rate": 4.093758230181722e-05,
      "loss": 1.7252,
      "step": 6888
    },
    {
      "epoch": 0.18143271003423755,
      "grad_norm": 2.044666290283203,
      "learning_rate": 4.0936265472741645e-05,
      "loss": 1.9257,
      "step": 6889
    },
    {
      "epoch": 0.18145904661574927,
      "grad_norm": 3.2120275497436523,
      "learning_rate": 4.0934948643666054e-05,
      "loss": 1.5806,
      "step": 6890
    },
    {
      "epoch": 0.18148538319726099,
      "grad_norm": 2.44638729095459,
      "learning_rate": 4.093363181459047e-05,
      "loss": 2.1068,
      "step": 6891
    },
    {
      "epoch": 0.18151171977877273,
      "grad_norm": 2.268062114715576,
      "learning_rate": 4.093231498551488e-05,
      "loss": 1.4517,
      "step": 6892
    },
    {
      "epoch": 0.18153805636028444,
      "grad_norm": 1.687217354774475,
      "learning_rate": 4.09309981564393e-05,
      "loss": 1.7188,
      "step": 6893
    },
    {
      "epoch": 0.18156439294179616,
      "grad_norm": 2.019071340560913,
      "learning_rate": 4.092968132736371e-05,
      "loss": 2.3121,
      "step": 6894
    },
    {
      "epoch": 0.18159072952330788,
      "grad_norm": 3.3610973358154297,
      "learning_rate": 4.0928364498288125e-05,
      "loss": 1.707,
      "step": 6895
    },
    {
      "epoch": 0.1816170661048196,
      "grad_norm": 2.2466182708740234,
      "learning_rate": 4.0927047669212534e-05,
      "loss": 1.1798,
      "step": 6896
    },
    {
      "epoch": 0.1816434026863313,
      "grad_norm": 1.7998205423355103,
      "learning_rate": 4.092573084013695e-05,
      "loss": 1.4341,
      "step": 6897
    },
    {
      "epoch": 0.18166973926784302,
      "grad_norm": 3.8328516483306885,
      "learning_rate": 4.0924414011061365e-05,
      "loss": 1.0946,
      "step": 6898
    },
    {
      "epoch": 0.18169607584935477,
      "grad_norm": 3.4311389923095703,
      "learning_rate": 4.092309718198578e-05,
      "loss": 0.8161,
      "step": 6899
    },
    {
      "epoch": 0.18172241243086648,
      "grad_norm": 5.416226387023926,
      "learning_rate": 4.0921780352910196e-05,
      "loss": 2.0961,
      "step": 6900
    },
    {
      "epoch": 0.1817487490123782,
      "grad_norm": 3.2724146842956543,
      "learning_rate": 4.0920463523834605e-05,
      "loss": 0.7855,
      "step": 6901
    },
    {
      "epoch": 0.1817750855938899,
      "grad_norm": 2.7790586948394775,
      "learning_rate": 4.091914669475902e-05,
      "loss": 0.9176,
      "step": 6902
    },
    {
      "epoch": 0.18180142217540163,
      "grad_norm": 1.815251111984253,
      "learning_rate": 4.0917829865683436e-05,
      "loss": 1.9013,
      "step": 6903
    },
    {
      "epoch": 0.18182775875691334,
      "grad_norm": 2.084998369216919,
      "learning_rate": 4.091651303660785e-05,
      "loss": 1.6303,
      "step": 6904
    },
    {
      "epoch": 0.18185409533842506,
      "grad_norm": 2.2930824756622314,
      "learning_rate": 4.091519620753226e-05,
      "loss": 1.759,
      "step": 6905
    },
    {
      "epoch": 0.1818804319199368,
      "grad_norm": 3.24057936668396,
      "learning_rate": 4.0913879378456676e-05,
      "loss": 1.1276,
      "step": 6906
    },
    {
      "epoch": 0.18190676850144852,
      "grad_norm": 2.000854969024658,
      "learning_rate": 4.091256254938109e-05,
      "loss": 1.8307,
      "step": 6907
    },
    {
      "epoch": 0.18193310508296023,
      "grad_norm": 2.2538535594940186,
      "learning_rate": 4.091124572030551e-05,
      "loss": 1.882,
      "step": 6908
    },
    {
      "epoch": 0.18195944166447195,
      "grad_norm": 2.1586172580718994,
      "learning_rate": 4.090992889122992e-05,
      "loss": 1.7585,
      "step": 6909
    },
    {
      "epoch": 0.18198577824598366,
      "grad_norm": 6.436657905578613,
      "learning_rate": 4.090861206215433e-05,
      "loss": 0.8941,
      "step": 6910
    },
    {
      "epoch": 0.18201211482749538,
      "grad_norm": 2.8224778175354004,
      "learning_rate": 4.090729523307875e-05,
      "loss": 1.6596,
      "step": 6911
    },
    {
      "epoch": 0.18203845140900712,
      "grad_norm": 1.6180421113967896,
      "learning_rate": 4.090597840400316e-05,
      "loss": 1.8923,
      "step": 6912
    },
    {
      "epoch": 0.18206478799051884,
      "grad_norm": 4.265376567840576,
      "learning_rate": 4.090466157492758e-05,
      "loss": 1.3665,
      "step": 6913
    },
    {
      "epoch": 0.18209112457203055,
      "grad_norm": 3.642742872238159,
      "learning_rate": 4.090334474585199e-05,
      "loss": 1.4445,
      "step": 6914
    },
    {
      "epoch": 0.18211746115354227,
      "grad_norm": 1.9837453365325928,
      "learning_rate": 4.09020279167764e-05,
      "loss": 1.9368,
      "step": 6915
    },
    {
      "epoch": 0.18214379773505399,
      "grad_norm": 1.541934609413147,
      "learning_rate": 4.090071108770082e-05,
      "loss": 1.5826,
      "step": 6916
    },
    {
      "epoch": 0.1821701343165657,
      "grad_norm": 1.923579216003418,
      "learning_rate": 4.0899394258625234e-05,
      "loss": 0.3904,
      "step": 6917
    },
    {
      "epoch": 0.18219647089807742,
      "grad_norm": 8.596114158630371,
      "learning_rate": 4.089807742954965e-05,
      "loss": 1.4123,
      "step": 6918
    },
    {
      "epoch": 0.18222280747958916,
      "grad_norm": 2.7523770332336426,
      "learning_rate": 4.089676060047406e-05,
      "loss": 0.5204,
      "step": 6919
    },
    {
      "epoch": 0.18224914406110088,
      "grad_norm": 4.1075758934021,
      "learning_rate": 4.0895443771398474e-05,
      "loss": 1.5215,
      "step": 6920
    },
    {
      "epoch": 0.1822754806426126,
      "grad_norm": 2.7514095306396484,
      "learning_rate": 4.089412694232289e-05,
      "loss": 0.6175,
      "step": 6921
    },
    {
      "epoch": 0.1823018172241243,
      "grad_norm": 1.769418478012085,
      "learning_rate": 4.0892810113247305e-05,
      "loss": 1.6328,
      "step": 6922
    },
    {
      "epoch": 0.18232815380563602,
      "grad_norm": 1.9333324432373047,
      "learning_rate": 4.0891493284171714e-05,
      "loss": 1.6509,
      "step": 6923
    },
    {
      "epoch": 0.18235449038714774,
      "grad_norm": 2.719660520553589,
      "learning_rate": 4.089017645509613e-05,
      "loss": 1.228,
      "step": 6924
    },
    {
      "epoch": 0.18238082696865948,
      "grad_norm": 2.9050986766815186,
      "learning_rate": 4.0888859626020545e-05,
      "loss": 1.1917,
      "step": 6925
    },
    {
      "epoch": 0.1824071635501712,
      "grad_norm": 3.8585193157196045,
      "learning_rate": 4.088754279694496e-05,
      "loss": 1.911,
      "step": 6926
    },
    {
      "epoch": 0.1824335001316829,
      "grad_norm": 2.263455390930176,
      "learning_rate": 4.088622596786938e-05,
      "loss": 0.5623,
      "step": 6927
    },
    {
      "epoch": 0.18245983671319463,
      "grad_norm": 1.7645083665847778,
      "learning_rate": 4.0884909138793785e-05,
      "loss": 2.1578,
      "step": 6928
    },
    {
      "epoch": 0.18248617329470634,
      "grad_norm": 1.7438981533050537,
      "learning_rate": 4.08835923097182e-05,
      "loss": 1.2436,
      "step": 6929
    },
    {
      "epoch": 0.18251250987621806,
      "grad_norm": 2.469877243041992,
      "learning_rate": 4.088227548064261e-05,
      "loss": 0.9466,
      "step": 6930
    },
    {
      "epoch": 0.18253884645772978,
      "grad_norm": 4.381165504455566,
      "learning_rate": 4.088095865156703e-05,
      "loss": 0.8198,
      "step": 6931
    },
    {
      "epoch": 0.18256518303924152,
      "grad_norm": 2.3040764331817627,
      "learning_rate": 4.087964182249144e-05,
      "loss": 1.5509,
      "step": 6932
    },
    {
      "epoch": 0.18259151962075323,
      "grad_norm": 2.239445447921753,
      "learning_rate": 4.087832499341586e-05,
      "loss": 2.2845,
      "step": 6933
    },
    {
      "epoch": 0.18261785620226495,
      "grad_norm": 1.7413301467895508,
      "learning_rate": 4.087700816434027e-05,
      "loss": 1.9569,
      "step": 6934
    },
    {
      "epoch": 0.18264419278377667,
      "grad_norm": 3.0984885692596436,
      "learning_rate": 4.087569133526468e-05,
      "loss": 1.7785,
      "step": 6935
    },
    {
      "epoch": 0.18267052936528838,
      "grad_norm": 2.176159381866455,
      "learning_rate": 4.0874374506189103e-05,
      "loss": 1.3328,
      "step": 6936
    },
    {
      "epoch": 0.1826968659468001,
      "grad_norm": 2.961353063583374,
      "learning_rate": 4.087305767711351e-05,
      "loss": 1.5821,
      "step": 6937
    },
    {
      "epoch": 0.1827232025283118,
      "grad_norm": 2.226867914199829,
      "learning_rate": 4.087174084803793e-05,
      "loss": 1.7769,
      "step": 6938
    },
    {
      "epoch": 0.18274953910982356,
      "grad_norm": 1.9836151599884033,
      "learning_rate": 4.087042401896234e-05,
      "loss": 1.9426,
      "step": 6939
    },
    {
      "epoch": 0.18277587569133527,
      "grad_norm": 2.933189630508423,
      "learning_rate": 4.086910718988676e-05,
      "loss": 1.5772,
      "step": 6940
    },
    {
      "epoch": 0.182802212272847,
      "grad_norm": 3.351119041442871,
      "learning_rate": 4.086779036081117e-05,
      "loss": 1.6856,
      "step": 6941
    },
    {
      "epoch": 0.1828285488543587,
      "grad_norm": 1.6739122867584229,
      "learning_rate": 4.0866473531735584e-05,
      "loss": 1.589,
      "step": 6942
    },
    {
      "epoch": 0.18285488543587042,
      "grad_norm": 2.4155874252319336,
      "learning_rate": 4.086515670266e-05,
      "loss": 1.7412,
      "step": 6943
    },
    {
      "epoch": 0.18288122201738213,
      "grad_norm": 4.7902398109436035,
      "learning_rate": 4.086383987358441e-05,
      "loss": 1.905,
      "step": 6944
    },
    {
      "epoch": 0.18290755859889388,
      "grad_norm": 3.0446271896362305,
      "learning_rate": 4.086252304450883e-05,
      "loss": 1.8619,
      "step": 6945
    },
    {
      "epoch": 0.1829338951804056,
      "grad_norm": 3.9771223068237305,
      "learning_rate": 4.086120621543324e-05,
      "loss": 1.6439,
      "step": 6946
    },
    {
      "epoch": 0.1829602317619173,
      "grad_norm": 2.1452763080596924,
      "learning_rate": 4.0859889386357655e-05,
      "loss": 1.3718,
      "step": 6947
    },
    {
      "epoch": 0.18298656834342902,
      "grad_norm": 2.008652448654175,
      "learning_rate": 4.0858572557282064e-05,
      "loss": 1.7454,
      "step": 6948
    },
    {
      "epoch": 0.18301290492494074,
      "grad_norm": 1.8076740503311157,
      "learning_rate": 4.085725572820648e-05,
      "loss": 2.0562,
      "step": 6949
    },
    {
      "epoch": 0.18303924150645245,
      "grad_norm": 1.9337666034698486,
      "learning_rate": 4.0855938899130895e-05,
      "loss": 1.8595,
      "step": 6950
    },
    {
      "epoch": 0.18306557808796417,
      "grad_norm": 1.9141703844070435,
      "learning_rate": 4.085462207005531e-05,
      "loss": 1.742,
      "step": 6951
    },
    {
      "epoch": 0.1830919146694759,
      "grad_norm": 2.366124153137207,
      "learning_rate": 4.085330524097972e-05,
      "loss": 1.3431,
      "step": 6952
    },
    {
      "epoch": 0.18311825125098763,
      "grad_norm": 1.792555332183838,
      "learning_rate": 4.0851988411904135e-05,
      "loss": 1.6651,
      "step": 6953
    },
    {
      "epoch": 0.18314458783249934,
      "grad_norm": 2.1594085693359375,
      "learning_rate": 4.085067158282855e-05,
      "loss": 2.1063,
      "step": 6954
    },
    {
      "epoch": 0.18317092441401106,
      "grad_norm": 3.452294111251831,
      "learning_rate": 4.0849354753752966e-05,
      "loss": 2.1025,
      "step": 6955
    },
    {
      "epoch": 0.18319726099552278,
      "grad_norm": 2.360137701034546,
      "learning_rate": 4.084803792467738e-05,
      "loss": 2.1452,
      "step": 6956
    },
    {
      "epoch": 0.1832235975770345,
      "grad_norm": 1.6228358745574951,
      "learning_rate": 4.084672109560179e-05,
      "loss": 2.2041,
      "step": 6957
    },
    {
      "epoch": 0.1832499341585462,
      "grad_norm": 2.022235155105591,
      "learning_rate": 4.0845404266526206e-05,
      "loss": 1.5295,
      "step": 6958
    },
    {
      "epoch": 0.18327627074005795,
      "grad_norm": 2.2516980171203613,
      "learning_rate": 4.084408743745062e-05,
      "loss": 1.9317,
      "step": 6959
    },
    {
      "epoch": 0.18330260732156967,
      "grad_norm": 1.9469623565673828,
      "learning_rate": 4.084277060837504e-05,
      "loss": 1.7795,
      "step": 6960
    },
    {
      "epoch": 0.18332894390308138,
      "grad_norm": 2.279630422592163,
      "learning_rate": 4.0841453779299446e-05,
      "loss": 1.7714,
      "step": 6961
    },
    {
      "epoch": 0.1833552804845931,
      "grad_norm": 2.4443295001983643,
      "learning_rate": 4.084013695022386e-05,
      "loss": 1.9419,
      "step": 6962
    },
    {
      "epoch": 0.1833816170661048,
      "grad_norm": 1.7123942375183105,
      "learning_rate": 4.083882012114828e-05,
      "loss": 1.8043,
      "step": 6963
    },
    {
      "epoch": 0.18340795364761653,
      "grad_norm": 4.375567436218262,
      "learning_rate": 4.083750329207269e-05,
      "loss": 0.9541,
      "step": 6964
    },
    {
      "epoch": 0.18343429022912827,
      "grad_norm": 4.862067699432373,
      "learning_rate": 4.083618646299711e-05,
      "loss": 1.048,
      "step": 6965
    },
    {
      "epoch": 0.18346062681064,
      "grad_norm": 2.5221095085144043,
      "learning_rate": 4.083486963392152e-05,
      "loss": 1.0404,
      "step": 6966
    },
    {
      "epoch": 0.1834869633921517,
      "grad_norm": 5.673100471496582,
      "learning_rate": 4.083355280484593e-05,
      "loss": 1.7307,
      "step": 6967
    },
    {
      "epoch": 0.18351329997366342,
      "grad_norm": 3.7432143688201904,
      "learning_rate": 4.083223597577034e-05,
      "loss": 1.4658,
      "step": 6968
    },
    {
      "epoch": 0.18353963655517513,
      "grad_norm": 1.9975262880325317,
      "learning_rate": 4.0830919146694764e-05,
      "loss": 1.8376,
      "step": 6969
    },
    {
      "epoch": 0.18356597313668685,
      "grad_norm": 3.3046092987060547,
      "learning_rate": 4.082960231761917e-05,
      "loss": 0.9348,
      "step": 6970
    },
    {
      "epoch": 0.18359230971819857,
      "grad_norm": 1.9514634609222412,
      "learning_rate": 4.082828548854359e-05,
      "loss": 1.5875,
      "step": 6971
    },
    {
      "epoch": 0.1836186462997103,
      "grad_norm": 3.79278826713562,
      "learning_rate": 4.0826968659468004e-05,
      "loss": 1.9848,
      "step": 6972
    },
    {
      "epoch": 0.18364498288122202,
      "grad_norm": 3.3770925998687744,
      "learning_rate": 4.082565183039242e-05,
      "loss": 1.7688,
      "step": 6973
    },
    {
      "epoch": 0.18367131946273374,
      "grad_norm": 3.727830171585083,
      "learning_rate": 4.0824335001316835e-05,
      "loss": 1.7535,
      "step": 6974
    },
    {
      "epoch": 0.18369765604424546,
      "grad_norm": 2.100135326385498,
      "learning_rate": 4.0823018172241244e-05,
      "loss": 2.1544,
      "step": 6975
    },
    {
      "epoch": 0.18372399262575717,
      "grad_norm": 2.4139041900634766,
      "learning_rate": 4.082170134316566e-05,
      "loss": 1.4866,
      "step": 6976
    },
    {
      "epoch": 0.1837503292072689,
      "grad_norm": 1.6062692403793335,
      "learning_rate": 4.082038451409007e-05,
      "loss": 1.7361,
      "step": 6977
    },
    {
      "epoch": 0.18377666578878063,
      "grad_norm": 2.123335361480713,
      "learning_rate": 4.081906768501449e-05,
      "loss": 1.6579,
      "step": 6978
    },
    {
      "epoch": 0.18380300237029235,
      "grad_norm": 2.155022382736206,
      "learning_rate": 4.08177508559389e-05,
      "loss": 2.8441,
      "step": 6979
    },
    {
      "epoch": 0.18382933895180406,
      "grad_norm": 4.40614652633667,
      "learning_rate": 4.0816434026863315e-05,
      "loss": 1.5018,
      "step": 6980
    },
    {
      "epoch": 0.18385567553331578,
      "grad_norm": 2.3432910442352295,
      "learning_rate": 4.081511719778773e-05,
      "loss": 1.9595,
      "step": 6981
    },
    {
      "epoch": 0.1838820121148275,
      "grad_norm": 2.8105359077453613,
      "learning_rate": 4.081380036871214e-05,
      "loss": 1.8624,
      "step": 6982
    },
    {
      "epoch": 0.1839083486963392,
      "grad_norm": 8.302210807800293,
      "learning_rate": 4.081248353963656e-05,
      "loss": 1.3675,
      "step": 6983
    },
    {
      "epoch": 0.18393468527785092,
      "grad_norm": 2.0491950511932373,
      "learning_rate": 4.081116671056097e-05,
      "loss": 2.1136,
      "step": 6984
    },
    {
      "epoch": 0.18396102185936267,
      "grad_norm": 4.874288558959961,
      "learning_rate": 4.0809849881485386e-05,
      "loss": 1.4397,
      "step": 6985
    },
    {
      "epoch": 0.18398735844087438,
      "grad_norm": 2.5536129474639893,
      "learning_rate": 4.0808533052409795e-05,
      "loss": 1.7797,
      "step": 6986
    },
    {
      "epoch": 0.1840136950223861,
      "grad_norm": 1.8840323686599731,
      "learning_rate": 4.080721622333422e-05,
      "loss": 2.4845,
      "step": 6987
    },
    {
      "epoch": 0.1840400316038978,
      "grad_norm": 2.0048911571502686,
      "learning_rate": 4.0805899394258626e-05,
      "loss": 1.5175,
      "step": 6988
    },
    {
      "epoch": 0.18406636818540953,
      "grad_norm": 2.539696455001831,
      "learning_rate": 4.080458256518304e-05,
      "loss": 1.9548,
      "step": 6989
    },
    {
      "epoch": 0.18409270476692124,
      "grad_norm": 1.951242446899414,
      "learning_rate": 4.080326573610746e-05,
      "loss": 1.3304,
      "step": 6990
    },
    {
      "epoch": 0.18411904134843296,
      "grad_norm": 4.876642227172852,
      "learning_rate": 4.0801948907031866e-05,
      "loss": 1.3783,
      "step": 6991
    },
    {
      "epoch": 0.1841453779299447,
      "grad_norm": 2.077089548110962,
      "learning_rate": 4.080063207795629e-05,
      "loss": 1.0857,
      "step": 6992
    },
    {
      "epoch": 0.18417171451145642,
      "grad_norm": 1.7162292003631592,
      "learning_rate": 4.07993152488807e-05,
      "loss": 1.4171,
      "step": 6993
    },
    {
      "epoch": 0.18419805109296813,
      "grad_norm": 2.0253751277923584,
      "learning_rate": 4.079799841980511e-05,
      "loss": 1.9105,
      "step": 6994
    },
    {
      "epoch": 0.18422438767447985,
      "grad_norm": 3.2778470516204834,
      "learning_rate": 4.079668159072952e-05,
      "loss": 1.6272,
      "step": 6995
    },
    {
      "epoch": 0.18425072425599157,
      "grad_norm": 2.9602034091949463,
      "learning_rate": 4.079536476165394e-05,
      "loss": 1.7009,
      "step": 6996
    },
    {
      "epoch": 0.18427706083750328,
      "grad_norm": 3.074713706970215,
      "learning_rate": 4.079404793257835e-05,
      "loss": 2.0371,
      "step": 6997
    },
    {
      "epoch": 0.18430339741901502,
      "grad_norm": 2.6605823040008545,
      "learning_rate": 4.079273110350277e-05,
      "loss": 2.234,
      "step": 6998
    },
    {
      "epoch": 0.18432973400052674,
      "grad_norm": 2.15189790725708,
      "learning_rate": 4.079141427442718e-05,
      "loss": 2.2861,
      "step": 6999
    },
    {
      "epoch": 0.18435607058203846,
      "grad_norm": 2.8215811252593994,
      "learning_rate": 4.079009744535159e-05,
      "loss": 2.0102,
      "step": 7000
    },
    {
      "epoch": 0.18438240716355017,
      "grad_norm": 1.8273508548736572,
      "learning_rate": 4.078878061627601e-05,
      "loss": 1.8234,
      "step": 7001
    },
    {
      "epoch": 0.1844087437450619,
      "grad_norm": 1.7942824363708496,
      "learning_rate": 4.0787463787200425e-05,
      "loss": 1.76,
      "step": 7002
    },
    {
      "epoch": 0.1844350803265736,
      "grad_norm": 2.399251937866211,
      "learning_rate": 4.078614695812484e-05,
      "loss": 2.5975,
      "step": 7003
    },
    {
      "epoch": 0.18446141690808532,
      "grad_norm": 2.5136725902557373,
      "learning_rate": 4.078483012904925e-05,
      "loss": 2.2768,
      "step": 7004
    },
    {
      "epoch": 0.18448775348959706,
      "grad_norm": 1.8775962591171265,
      "learning_rate": 4.0783513299973665e-05,
      "loss": 1.6603,
      "step": 7005
    },
    {
      "epoch": 0.18451409007110878,
      "grad_norm": 2.997371196746826,
      "learning_rate": 4.078219647089808e-05,
      "loss": 1.7046,
      "step": 7006
    },
    {
      "epoch": 0.1845404266526205,
      "grad_norm": 2.1706979274749756,
      "learning_rate": 4.0780879641822496e-05,
      "loss": 1.9254,
      "step": 7007
    },
    {
      "epoch": 0.1845667632341322,
      "grad_norm": 2.203439712524414,
      "learning_rate": 4.0779562812746905e-05,
      "loss": 2.3046,
      "step": 7008
    },
    {
      "epoch": 0.18459309981564392,
      "grad_norm": 2.1381664276123047,
      "learning_rate": 4.077824598367132e-05,
      "loss": 1.7123,
      "step": 7009
    },
    {
      "epoch": 0.18461943639715564,
      "grad_norm": 1.6301268339157104,
      "learning_rate": 4.0776929154595736e-05,
      "loss": 2.0264,
      "step": 7010
    },
    {
      "epoch": 0.18464577297866736,
      "grad_norm": 1.6790343523025513,
      "learning_rate": 4.077561232552015e-05,
      "loss": 1.3366,
      "step": 7011
    },
    {
      "epoch": 0.1846721095601791,
      "grad_norm": 1.6542704105377197,
      "learning_rate": 4.077429549644457e-05,
      "loss": 1.8495,
      "step": 7012
    },
    {
      "epoch": 0.18469844614169081,
      "grad_norm": 2.584717273712158,
      "learning_rate": 4.0772978667368976e-05,
      "loss": 1.6227,
      "step": 7013
    },
    {
      "epoch": 0.18472478272320253,
      "grad_norm": 2.4365644454956055,
      "learning_rate": 4.077166183829339e-05,
      "loss": 1.7179,
      "step": 7014
    },
    {
      "epoch": 0.18475111930471425,
      "grad_norm": 1.7049157619476318,
      "learning_rate": 4.07703450092178e-05,
      "loss": 1.6792,
      "step": 7015
    },
    {
      "epoch": 0.18477745588622596,
      "grad_norm": 2.7974448204040527,
      "learning_rate": 4.076902818014222e-05,
      "loss": 1.6315,
      "step": 7016
    },
    {
      "epoch": 0.18480379246773768,
      "grad_norm": 4.127683162689209,
      "learning_rate": 4.076771135106663e-05,
      "loss": 1.3024,
      "step": 7017
    },
    {
      "epoch": 0.18483012904924942,
      "grad_norm": 2.4587247371673584,
      "learning_rate": 4.076639452199105e-05,
      "loss": 1.4815,
      "step": 7018
    },
    {
      "epoch": 0.18485646563076114,
      "grad_norm": 1.9659017324447632,
      "learning_rate": 4.076507769291546e-05,
      "loss": 1.781,
      "step": 7019
    },
    {
      "epoch": 0.18488280221227285,
      "grad_norm": 1.780299425125122,
      "learning_rate": 4.076376086383988e-05,
      "loss": 1.739,
      "step": 7020
    },
    {
      "epoch": 0.18490913879378457,
      "grad_norm": 4.755344867706299,
      "learning_rate": 4.0762444034764294e-05,
      "loss": 0.8956,
      "step": 7021
    },
    {
      "epoch": 0.18493547537529628,
      "grad_norm": 1.8263736963272095,
      "learning_rate": 4.07611272056887e-05,
      "loss": 1.7942,
      "step": 7022
    },
    {
      "epoch": 0.184961811956808,
      "grad_norm": 2.6218724250793457,
      "learning_rate": 4.075981037661312e-05,
      "loss": 1.5062,
      "step": 7023
    },
    {
      "epoch": 0.1849881485383197,
      "grad_norm": 1.7783833742141724,
      "learning_rate": 4.075849354753753e-05,
      "loss": 1.6431,
      "step": 7024
    },
    {
      "epoch": 0.18501448511983146,
      "grad_norm": 2.4616410732269287,
      "learning_rate": 4.075717671846195e-05,
      "loss": 1.6206,
      "step": 7025
    },
    {
      "epoch": 0.18504082170134317,
      "grad_norm": 2.2635910511016846,
      "learning_rate": 4.075585988938636e-05,
      "loss": 1.5831,
      "step": 7026
    },
    {
      "epoch": 0.1850671582828549,
      "grad_norm": 2.142585277557373,
      "learning_rate": 4.0754543060310774e-05,
      "loss": 2.6223,
      "step": 7027
    },
    {
      "epoch": 0.1850934948643666,
      "grad_norm": 2.5691113471984863,
      "learning_rate": 4.075322623123519e-05,
      "loss": 2.0103,
      "step": 7028
    },
    {
      "epoch": 0.18511983144587832,
      "grad_norm": 1.8876320123672485,
      "learning_rate": 4.07519094021596e-05,
      "loss": 2.0573,
      "step": 7029
    },
    {
      "epoch": 0.18514616802739003,
      "grad_norm": 3.4261367321014404,
      "learning_rate": 4.075059257308402e-05,
      "loss": 1.104,
      "step": 7030
    },
    {
      "epoch": 0.18517250460890178,
      "grad_norm": 2.8919003009796143,
      "learning_rate": 4.074927574400843e-05,
      "loss": 2.3249,
      "step": 7031
    },
    {
      "epoch": 0.1851988411904135,
      "grad_norm": 2.020681619644165,
      "learning_rate": 4.0747958914932845e-05,
      "loss": 1.8441,
      "step": 7032
    },
    {
      "epoch": 0.1852251777719252,
      "grad_norm": 3.334242343902588,
      "learning_rate": 4.0746642085857254e-05,
      "loss": 1.6714,
      "step": 7033
    },
    {
      "epoch": 0.18525151435343692,
      "grad_norm": 2.288557529449463,
      "learning_rate": 4.074532525678167e-05,
      "loss": 1.7518,
      "step": 7034
    },
    {
      "epoch": 0.18527785093494864,
      "grad_norm": 2.5938003063201904,
      "learning_rate": 4.0744008427706085e-05,
      "loss": 0.2539,
      "step": 7035
    },
    {
      "epoch": 0.18530418751646036,
      "grad_norm": 2.171165704727173,
      "learning_rate": 4.07426915986305e-05,
      "loss": 1.9121,
      "step": 7036
    },
    {
      "epoch": 0.18533052409797207,
      "grad_norm": 2.739981174468994,
      "learning_rate": 4.0741374769554916e-05,
      "loss": 0.791,
      "step": 7037
    },
    {
      "epoch": 0.18535686067948381,
      "grad_norm": 2.657623291015625,
      "learning_rate": 4.0740057940479325e-05,
      "loss": 2.4257,
      "step": 7038
    },
    {
      "epoch": 0.18538319726099553,
      "grad_norm": 1.6616113185882568,
      "learning_rate": 4.073874111140375e-05,
      "loss": 1.9823,
      "step": 7039
    },
    {
      "epoch": 0.18540953384250725,
      "grad_norm": 4.093944072723389,
      "learning_rate": 4.0737424282328156e-05,
      "loss": 2.1575,
      "step": 7040
    },
    {
      "epoch": 0.18543587042401896,
      "grad_norm": 4.127021789550781,
      "learning_rate": 4.073610745325257e-05,
      "loss": 1.1244,
      "step": 7041
    },
    {
      "epoch": 0.18546220700553068,
      "grad_norm": 1.7074508666992188,
      "learning_rate": 4.073479062417698e-05,
      "loss": 0.444,
      "step": 7042
    },
    {
      "epoch": 0.1854885435870424,
      "grad_norm": 1.7890299558639526,
      "learning_rate": 4.0733473795101396e-05,
      "loss": 2.3078,
      "step": 7043
    },
    {
      "epoch": 0.1855148801685541,
      "grad_norm": 1.825681447982788,
      "learning_rate": 4.073215696602581e-05,
      "loss": 2.0246,
      "step": 7044
    },
    {
      "epoch": 0.18554121675006585,
      "grad_norm": 2.5056962966918945,
      "learning_rate": 4.073084013695023e-05,
      "loss": 0.8313,
      "step": 7045
    },
    {
      "epoch": 0.18556755333157757,
      "grad_norm": 1.8695944547653198,
      "learning_rate": 4.0729523307874636e-05,
      "loss": 1.9256,
      "step": 7046
    },
    {
      "epoch": 0.18559388991308928,
      "grad_norm": 2.4649107456207275,
      "learning_rate": 4.072820647879905e-05,
      "loss": 1.9475,
      "step": 7047
    },
    {
      "epoch": 0.185620226494601,
      "grad_norm": 6.007030487060547,
      "learning_rate": 4.072688964972347e-05,
      "loss": 2.2265,
      "step": 7048
    },
    {
      "epoch": 0.18564656307611271,
      "grad_norm": 2.180332899093628,
      "learning_rate": 4.072557282064788e-05,
      "loss": 2.2295,
      "step": 7049
    },
    {
      "epoch": 0.18567289965762443,
      "grad_norm": 2.95991849899292,
      "learning_rate": 4.07242559915723e-05,
      "loss": 1.4116,
      "step": 7050
    },
    {
      "epoch": 0.18569923623913617,
      "grad_norm": 1.8640722036361694,
      "learning_rate": 4.072293916249671e-05,
      "loss": 1.7615,
      "step": 7051
    },
    {
      "epoch": 0.1857255728206479,
      "grad_norm": 1.9431788921356201,
      "learning_rate": 4.072162233342112e-05,
      "loss": 2.1496,
      "step": 7052
    },
    {
      "epoch": 0.1857519094021596,
      "grad_norm": 3.0725719928741455,
      "learning_rate": 4.072030550434554e-05,
      "loss": 2.0362,
      "step": 7053
    },
    {
      "epoch": 0.18577824598367132,
      "grad_norm": 1.9439278841018677,
      "learning_rate": 4.0718988675269954e-05,
      "loss": 2.0799,
      "step": 7054
    },
    {
      "epoch": 0.18580458256518304,
      "grad_norm": 4.480274200439453,
      "learning_rate": 4.071767184619436e-05,
      "loss": 1.5139,
      "step": 7055
    },
    {
      "epoch": 0.18583091914669475,
      "grad_norm": 1.6987828016281128,
      "learning_rate": 4.071635501711878e-05,
      "loss": 1.7111,
      "step": 7056
    },
    {
      "epoch": 0.18585725572820647,
      "grad_norm": 2.4827072620391846,
      "learning_rate": 4.0715038188043194e-05,
      "loss": 1.8266,
      "step": 7057
    },
    {
      "epoch": 0.1858835923097182,
      "grad_norm": 2.5901682376861572,
      "learning_rate": 4.071372135896761e-05,
      "loss": 2.0084,
      "step": 7058
    },
    {
      "epoch": 0.18590992889122993,
      "grad_norm": 2.8255746364593506,
      "learning_rate": 4.0712404529892025e-05,
      "loss": 1.9873,
      "step": 7059
    },
    {
      "epoch": 0.18593626547274164,
      "grad_norm": 1.98200261592865,
      "learning_rate": 4.0711087700816434e-05,
      "loss": 2.0188,
      "step": 7060
    },
    {
      "epoch": 0.18596260205425336,
      "grad_norm": 2.1183969974517822,
      "learning_rate": 4.070977087174085e-05,
      "loss": 1.9547,
      "step": 7061
    },
    {
      "epoch": 0.18598893863576507,
      "grad_norm": 2.6942882537841797,
      "learning_rate": 4.070845404266526e-05,
      "loss": 2.0844,
      "step": 7062
    },
    {
      "epoch": 0.1860152752172768,
      "grad_norm": 2.1109118461608887,
      "learning_rate": 4.070713721358968e-05,
      "loss": 1.7476,
      "step": 7063
    },
    {
      "epoch": 0.18604161179878853,
      "grad_norm": 1.9203718900680542,
      "learning_rate": 4.070582038451409e-05,
      "loss": 1.9132,
      "step": 7064
    },
    {
      "epoch": 0.18606794838030025,
      "grad_norm": 2.0590131282806396,
      "learning_rate": 4.0704503555438506e-05,
      "loss": 1.8812,
      "step": 7065
    },
    {
      "epoch": 0.18609428496181196,
      "grad_norm": 2.406639575958252,
      "learning_rate": 4.070318672636292e-05,
      "loss": 1.9138,
      "step": 7066
    },
    {
      "epoch": 0.18612062154332368,
      "grad_norm": 1.6308205127716064,
      "learning_rate": 4.070186989728733e-05,
      "loss": 1.5473,
      "step": 7067
    },
    {
      "epoch": 0.1861469581248354,
      "grad_norm": 2.4697418212890625,
      "learning_rate": 4.070055306821175e-05,
      "loss": 1.7348,
      "step": 7068
    },
    {
      "epoch": 0.1861732947063471,
      "grad_norm": 1.9978829622268677,
      "learning_rate": 4.069923623913616e-05,
      "loss": 0.1428,
      "step": 7069
    },
    {
      "epoch": 0.18619963128785882,
      "grad_norm": 2.2762584686279297,
      "learning_rate": 4.069791941006058e-05,
      "loss": 1.7284,
      "step": 7070
    },
    {
      "epoch": 0.18622596786937057,
      "grad_norm": 1.59394109249115,
      "learning_rate": 4.0696602580984986e-05,
      "loss": 1.7665,
      "step": 7071
    },
    {
      "epoch": 0.18625230445088228,
      "grad_norm": 3.349914789199829,
      "learning_rate": 4.069528575190941e-05,
      "loss": 1.4147,
      "step": 7072
    },
    {
      "epoch": 0.186278641032394,
      "grad_norm": 2.417703151702881,
      "learning_rate": 4.069396892283382e-05,
      "loss": 1.507,
      "step": 7073
    },
    {
      "epoch": 0.18630497761390571,
      "grad_norm": 1.7976462841033936,
      "learning_rate": 4.069265209375823e-05,
      "loss": 1.2664,
      "step": 7074
    },
    {
      "epoch": 0.18633131419541743,
      "grad_norm": 2.6340603828430176,
      "learning_rate": 4.069133526468265e-05,
      "loss": 2.2531,
      "step": 7075
    },
    {
      "epoch": 0.18635765077692915,
      "grad_norm": 1.8017858266830444,
      "learning_rate": 4.069001843560706e-05,
      "loss": 1.568,
      "step": 7076
    },
    {
      "epoch": 0.18638398735844086,
      "grad_norm": 1.8842220306396484,
      "learning_rate": 4.068870160653148e-05,
      "loss": 1.9075,
      "step": 7077
    },
    {
      "epoch": 0.1864103239399526,
      "grad_norm": 1.8736804723739624,
      "learning_rate": 4.068738477745589e-05,
      "loss": 2.0031,
      "step": 7078
    },
    {
      "epoch": 0.18643666052146432,
      "grad_norm": 2.1913700103759766,
      "learning_rate": 4.0686067948380304e-05,
      "loss": 2.2514,
      "step": 7079
    },
    {
      "epoch": 0.18646299710297604,
      "grad_norm": 2.041095018386841,
      "learning_rate": 4.068475111930471e-05,
      "loss": 1.2421,
      "step": 7080
    },
    {
      "epoch": 0.18648933368448775,
      "grad_norm": 5.4691948890686035,
      "learning_rate": 4.068343429022913e-05,
      "loss": 0.9759,
      "step": 7081
    },
    {
      "epoch": 0.18651567026599947,
      "grad_norm": 2.5302422046661377,
      "learning_rate": 4.0682117461153544e-05,
      "loss": 1.8126,
      "step": 7082
    },
    {
      "epoch": 0.18654200684751118,
      "grad_norm": 2.04630446434021,
      "learning_rate": 4.068080063207796e-05,
      "loss": 1.8427,
      "step": 7083
    },
    {
      "epoch": 0.18656834342902293,
      "grad_norm": 2.717851161956787,
      "learning_rate": 4.0679483803002375e-05,
      "loss": 1.6675,
      "step": 7084
    },
    {
      "epoch": 0.18659468001053464,
      "grad_norm": 1.954132318496704,
      "learning_rate": 4.0678166973926784e-05,
      "loss": 1.7673,
      "step": 7085
    },
    {
      "epoch": 0.18662101659204636,
      "grad_norm": 2.1075642108917236,
      "learning_rate": 4.0676850144851206e-05,
      "loss": 2.1142,
      "step": 7086
    },
    {
      "epoch": 0.18664735317355807,
      "grad_norm": 5.034909725189209,
      "learning_rate": 4.0675533315775615e-05,
      "loss": 1.1115,
      "step": 7087
    },
    {
      "epoch": 0.1866736897550698,
      "grad_norm": 2.039327383041382,
      "learning_rate": 4.067421648670003e-05,
      "loss": 0.7899,
      "step": 7088
    },
    {
      "epoch": 0.1867000263365815,
      "grad_norm": 2.3611059188842773,
      "learning_rate": 4.067289965762444e-05,
      "loss": 1.8259,
      "step": 7089
    },
    {
      "epoch": 0.18672636291809322,
      "grad_norm": 1.709442138671875,
      "learning_rate": 4.0671582828548855e-05,
      "loss": 1.6856,
      "step": 7090
    },
    {
      "epoch": 0.18675269949960496,
      "grad_norm": 3.9820659160614014,
      "learning_rate": 4.067026599947327e-05,
      "loss": 1.5713,
      "step": 7091
    },
    {
      "epoch": 0.18677903608111668,
      "grad_norm": 2.329197406768799,
      "learning_rate": 4.0668949170397686e-05,
      "loss": 1.7154,
      "step": 7092
    },
    {
      "epoch": 0.1868053726626284,
      "grad_norm": 1.8741422891616821,
      "learning_rate": 4.06676323413221e-05,
      "loss": 1.8546,
      "step": 7093
    },
    {
      "epoch": 0.1868317092441401,
      "grad_norm": 2.687278985977173,
      "learning_rate": 4.066631551224651e-05,
      "loss": 1.5314,
      "step": 7094
    },
    {
      "epoch": 0.18685804582565183,
      "grad_norm": 4.32405424118042,
      "learning_rate": 4.0664998683170926e-05,
      "loss": 1.6117,
      "step": 7095
    },
    {
      "epoch": 0.18688438240716354,
      "grad_norm": 2.0460257530212402,
      "learning_rate": 4.066368185409534e-05,
      "loss": 2.1934,
      "step": 7096
    },
    {
      "epoch": 0.18691071898867526,
      "grad_norm": 1.694000482559204,
      "learning_rate": 4.066236502501976e-05,
      "loss": 1.7865,
      "step": 7097
    },
    {
      "epoch": 0.186937055570187,
      "grad_norm": 5.954615116119385,
      "learning_rate": 4.0661048195944166e-05,
      "loss": 2.3845,
      "step": 7098
    },
    {
      "epoch": 0.18696339215169872,
      "grad_norm": 1.9868425130844116,
      "learning_rate": 4.065973136686858e-05,
      "loss": 2.08,
      "step": 7099
    },
    {
      "epoch": 0.18698972873321043,
      "grad_norm": 2.7583935260772705,
      "learning_rate": 4.065841453779299e-05,
      "loss": 2.2414,
      "step": 7100
    },
    {
      "epoch": 0.18701606531472215,
      "grad_norm": 2.2052414417266846,
      "learning_rate": 4.065709770871741e-05,
      "loss": 1.7992,
      "step": 7101
    },
    {
      "epoch": 0.18704240189623386,
      "grad_norm": 3.4391391277313232,
      "learning_rate": 4.065578087964182e-05,
      "loss": 1.6053,
      "step": 7102
    },
    {
      "epoch": 0.18706873847774558,
      "grad_norm": 3.3688368797302246,
      "learning_rate": 4.065446405056624e-05,
      "loss": 1.7838,
      "step": 7103
    },
    {
      "epoch": 0.18709507505925732,
      "grad_norm": 5.613596439361572,
      "learning_rate": 4.065314722149065e-05,
      "loss": 1.586,
      "step": 7104
    },
    {
      "epoch": 0.18712141164076904,
      "grad_norm": 4.961678504943848,
      "learning_rate": 4.065183039241507e-05,
      "loss": 1.9074,
      "step": 7105
    },
    {
      "epoch": 0.18714774822228075,
      "grad_norm": 2.3860888481140137,
      "learning_rate": 4.0650513563339484e-05,
      "loss": 1.4551,
      "step": 7106
    },
    {
      "epoch": 0.18717408480379247,
      "grad_norm": 2.942716360092163,
      "learning_rate": 4.064919673426389e-05,
      "loss": 1.299,
      "step": 7107
    },
    {
      "epoch": 0.18720042138530418,
      "grad_norm": 1.6908869743347168,
      "learning_rate": 4.064787990518831e-05,
      "loss": 2.1328,
      "step": 7108
    },
    {
      "epoch": 0.1872267579668159,
      "grad_norm": 2.9016764163970947,
      "learning_rate": 4.064656307611272e-05,
      "loss": 0.5772,
      "step": 7109
    },
    {
      "epoch": 0.18725309454832761,
      "grad_norm": 2.7314696311950684,
      "learning_rate": 4.064524624703714e-05,
      "loss": 1.3018,
      "step": 7110
    },
    {
      "epoch": 0.18727943112983936,
      "grad_norm": 2.7531847953796387,
      "learning_rate": 4.064392941796155e-05,
      "loss": 2.3253,
      "step": 7111
    },
    {
      "epoch": 0.18730576771135107,
      "grad_norm": 1.993249535560608,
      "learning_rate": 4.0642612588885964e-05,
      "loss": 1.3962,
      "step": 7112
    },
    {
      "epoch": 0.1873321042928628,
      "grad_norm": 2.0955049991607666,
      "learning_rate": 4.064129575981038e-05,
      "loss": 1.7677,
      "step": 7113
    },
    {
      "epoch": 0.1873584408743745,
      "grad_norm": 2.7482011318206787,
      "learning_rate": 4.063997893073479e-05,
      "loss": 0.7222,
      "step": 7114
    },
    {
      "epoch": 0.18738477745588622,
      "grad_norm": 1.6753559112548828,
      "learning_rate": 4.063866210165921e-05,
      "loss": 1.9886,
      "step": 7115
    },
    {
      "epoch": 0.18741111403739794,
      "grad_norm": 1.7713512182235718,
      "learning_rate": 4.063734527258362e-05,
      "loss": 1.494,
      "step": 7116
    },
    {
      "epoch": 0.18743745061890968,
      "grad_norm": 2.061821460723877,
      "learning_rate": 4.0636028443508035e-05,
      "loss": 2.0898,
      "step": 7117
    },
    {
      "epoch": 0.1874637872004214,
      "grad_norm": 1.9059206247329712,
      "learning_rate": 4.0634711614432444e-05,
      "loss": 1.7705,
      "step": 7118
    },
    {
      "epoch": 0.1874901237819331,
      "grad_norm": 2.3998019695281982,
      "learning_rate": 4.0633394785356866e-05,
      "loss": 1.7339,
      "step": 7119
    },
    {
      "epoch": 0.18751646036344483,
      "grad_norm": 1.636765480041504,
      "learning_rate": 4.0632077956281275e-05,
      "loss": 2.3229,
      "step": 7120
    },
    {
      "epoch": 0.18754279694495654,
      "grad_norm": 1.693140983581543,
      "learning_rate": 4.063076112720569e-05,
      "loss": 1.5128,
      "step": 7121
    },
    {
      "epoch": 0.18756913352646826,
      "grad_norm": 3.616184949874878,
      "learning_rate": 4.0629444298130107e-05,
      "loss": 1.3153,
      "step": 7122
    },
    {
      "epoch": 0.18759547010797997,
      "grad_norm": 1.8281915187835693,
      "learning_rate": 4.0628127469054515e-05,
      "loss": 1.9864,
      "step": 7123
    },
    {
      "epoch": 0.18762180668949172,
      "grad_norm": 2.0416295528411865,
      "learning_rate": 4.062681063997894e-05,
      "loss": 2.1649,
      "step": 7124
    },
    {
      "epoch": 0.18764814327100343,
      "grad_norm": 3.8911514282226562,
      "learning_rate": 4.0625493810903347e-05,
      "loss": 1.2026,
      "step": 7125
    },
    {
      "epoch": 0.18767447985251515,
      "grad_norm": 1.6992837190628052,
      "learning_rate": 4.062417698182776e-05,
      "loss": 1.51,
      "step": 7126
    },
    {
      "epoch": 0.18770081643402686,
      "grad_norm": 2.820889711380005,
      "learning_rate": 4.062286015275217e-05,
      "loss": 1.893,
      "step": 7127
    },
    {
      "epoch": 0.18772715301553858,
      "grad_norm": 2.89487361907959,
      "learning_rate": 4.0621543323676587e-05,
      "loss": 0.7218,
      "step": 7128
    },
    {
      "epoch": 0.1877534895970503,
      "grad_norm": 1.7519015073776245,
      "learning_rate": 4.0620226494601e-05,
      "loss": 2.0041,
      "step": 7129
    },
    {
      "epoch": 0.187779826178562,
      "grad_norm": 6.424541473388672,
      "learning_rate": 4.061890966552542e-05,
      "loss": 1.0124,
      "step": 7130
    },
    {
      "epoch": 0.18780616276007375,
      "grad_norm": 2.0475144386291504,
      "learning_rate": 4.061759283644983e-05,
      "loss": 1.2348,
      "step": 7131
    },
    {
      "epoch": 0.18783249934158547,
      "grad_norm": 2.3664028644561768,
      "learning_rate": 4.061627600737424e-05,
      "loss": 1.5428,
      "step": 7132
    },
    {
      "epoch": 0.18785883592309718,
      "grad_norm": 2.0390541553497314,
      "learning_rate": 4.061495917829866e-05,
      "loss": 2.107,
      "step": 7133
    },
    {
      "epoch": 0.1878851725046089,
      "grad_norm": 5.205618381500244,
      "learning_rate": 4.061364234922307e-05,
      "loss": 0.8842,
      "step": 7134
    },
    {
      "epoch": 0.18791150908612061,
      "grad_norm": 4.260695457458496,
      "learning_rate": 4.061232552014749e-05,
      "loss": 1.6472,
      "step": 7135
    },
    {
      "epoch": 0.18793784566763233,
      "grad_norm": 3.7273504734039307,
      "learning_rate": 4.06110086910719e-05,
      "loss": 0.8915,
      "step": 7136
    },
    {
      "epoch": 0.18796418224914407,
      "grad_norm": 4.018039226531982,
      "learning_rate": 4.060969186199631e-05,
      "loss": 1.9645,
      "step": 7137
    },
    {
      "epoch": 0.1879905188306558,
      "grad_norm": 2.592669725418091,
      "learning_rate": 4.060837503292073e-05,
      "loss": 2.8759,
      "step": 7138
    },
    {
      "epoch": 0.1880168554121675,
      "grad_norm": 1.8754229545593262,
      "learning_rate": 4.0607058203845145e-05,
      "loss": 1.6851,
      "step": 7139
    },
    {
      "epoch": 0.18804319199367922,
      "grad_norm": 2.291823148727417,
      "learning_rate": 4.060574137476956e-05,
      "loss": 2.2798,
      "step": 7140
    },
    {
      "epoch": 0.18806952857519094,
      "grad_norm": 2.3167099952697754,
      "learning_rate": 4.060442454569397e-05,
      "loss": 1.614,
      "step": 7141
    },
    {
      "epoch": 0.18809586515670265,
      "grad_norm": 2.762929677963257,
      "learning_rate": 4.0603107716618385e-05,
      "loss": 1.2255,
      "step": 7142
    },
    {
      "epoch": 0.18812220173821437,
      "grad_norm": 1.8997184038162231,
      "learning_rate": 4.06017908875428e-05,
      "loss": 1.819,
      "step": 7143
    },
    {
      "epoch": 0.1881485383197261,
      "grad_norm": 2.7060487270355225,
      "learning_rate": 4.0600474058467216e-05,
      "loss": 1.522,
      "step": 7144
    },
    {
      "epoch": 0.18817487490123783,
      "grad_norm": 1.8368016481399536,
      "learning_rate": 4.0599157229391625e-05,
      "loss": 1.687,
      "step": 7145
    },
    {
      "epoch": 0.18820121148274954,
      "grad_norm": 2.0138535499572754,
      "learning_rate": 4.059784040031604e-05,
      "loss": 1.899,
      "step": 7146
    },
    {
      "epoch": 0.18822754806426126,
      "grad_norm": 2.190800189971924,
      "learning_rate": 4.059652357124045e-05,
      "loss": 1.8069,
      "step": 7147
    },
    {
      "epoch": 0.18825388464577297,
      "grad_norm": 3.160052537918091,
      "learning_rate": 4.059520674216487e-05,
      "loss": 1.5584,
      "step": 7148
    },
    {
      "epoch": 0.1882802212272847,
      "grad_norm": 2.6142807006835938,
      "learning_rate": 4.059388991308928e-05,
      "loss": 1.8108,
      "step": 7149
    },
    {
      "epoch": 0.18830655780879643,
      "grad_norm": 2.1586642265319824,
      "learning_rate": 4.0592573084013696e-05,
      "loss": 1.8957,
      "step": 7150
    },
    {
      "epoch": 0.18833289439030815,
      "grad_norm": 3.6322038173675537,
      "learning_rate": 4.059125625493811e-05,
      "loss": 1.4507,
      "step": 7151
    },
    {
      "epoch": 0.18835923097181986,
      "grad_norm": 2.8608357906341553,
      "learning_rate": 4.058993942586253e-05,
      "loss": 0.673,
      "step": 7152
    },
    {
      "epoch": 0.18838556755333158,
      "grad_norm": 2.3458855152130127,
      "learning_rate": 4.058862259678694e-05,
      "loss": 2.0301,
      "step": 7153
    },
    {
      "epoch": 0.1884119041348433,
      "grad_norm": 3.176971912384033,
      "learning_rate": 4.058730576771135e-05,
      "loss": 2.2794,
      "step": 7154
    },
    {
      "epoch": 0.188438240716355,
      "grad_norm": 4.816822052001953,
      "learning_rate": 4.058598893863577e-05,
      "loss": 0.9821,
      "step": 7155
    },
    {
      "epoch": 0.18846457729786673,
      "grad_norm": 2.430255651473999,
      "learning_rate": 4.0584672109560176e-05,
      "loss": 0.7312,
      "step": 7156
    },
    {
      "epoch": 0.18849091387937847,
      "grad_norm": 1.8843653202056885,
      "learning_rate": 4.05833552804846e-05,
      "loss": 1.3415,
      "step": 7157
    },
    {
      "epoch": 0.18851725046089018,
      "grad_norm": 3.2687995433807373,
      "learning_rate": 4.058203845140901e-05,
      "loss": 2.3528,
      "step": 7158
    },
    {
      "epoch": 0.1885435870424019,
      "grad_norm": 2.6169800758361816,
      "learning_rate": 4.058072162233342e-05,
      "loss": 1.8997,
      "step": 7159
    },
    {
      "epoch": 0.18856992362391362,
      "grad_norm": 2.1954383850097656,
      "learning_rate": 4.057940479325784e-05,
      "loss": 1.7589,
      "step": 7160
    },
    {
      "epoch": 0.18859626020542533,
      "grad_norm": 2.1346802711486816,
      "learning_rate": 4.057808796418225e-05,
      "loss": 1.878,
      "step": 7161
    },
    {
      "epoch": 0.18862259678693705,
      "grad_norm": 1.7980924844741821,
      "learning_rate": 4.057677113510667e-05,
      "loss": 1.7207,
      "step": 7162
    },
    {
      "epoch": 0.18864893336844876,
      "grad_norm": 2.881147861480713,
      "learning_rate": 4.057545430603108e-05,
      "loss": 0.8091,
      "step": 7163
    },
    {
      "epoch": 0.1886752699499605,
      "grad_norm": 5.948313236236572,
      "learning_rate": 4.0574137476955494e-05,
      "loss": 2.1255,
      "step": 7164
    },
    {
      "epoch": 0.18870160653147222,
      "grad_norm": 2.595654010772705,
      "learning_rate": 4.05728206478799e-05,
      "loss": 0.9837,
      "step": 7165
    },
    {
      "epoch": 0.18872794311298394,
      "grad_norm": 6.243841648101807,
      "learning_rate": 4.057150381880432e-05,
      "loss": 1.5669,
      "step": 7166
    },
    {
      "epoch": 0.18875427969449565,
      "grad_norm": 3.450916290283203,
      "learning_rate": 4.0570186989728734e-05,
      "loss": 2.7154,
      "step": 7167
    },
    {
      "epoch": 0.18878061627600737,
      "grad_norm": 3.38445782661438,
      "learning_rate": 4.056887016065315e-05,
      "loss": 1.58,
      "step": 7168
    },
    {
      "epoch": 0.18880695285751908,
      "grad_norm": 1.9969775676727295,
      "learning_rate": 4.0567553331577565e-05,
      "loss": 1.4763,
      "step": 7169
    },
    {
      "epoch": 0.18883328943903083,
      "grad_norm": 1.6472141742706299,
      "learning_rate": 4.0566236502501974e-05,
      "loss": 1.631,
      "step": 7170
    },
    {
      "epoch": 0.18885962602054254,
      "grad_norm": 2.0674567222595215,
      "learning_rate": 4.0564919673426396e-05,
      "loss": 1.8451,
      "step": 7171
    },
    {
      "epoch": 0.18888596260205426,
      "grad_norm": 2.0916426181793213,
      "learning_rate": 4.0563602844350805e-05,
      "loss": 1.3798,
      "step": 7172
    },
    {
      "epoch": 0.18891229918356597,
      "grad_norm": 3.7419512271881104,
      "learning_rate": 4.056228601527522e-05,
      "loss": 2.24,
      "step": 7173
    },
    {
      "epoch": 0.1889386357650777,
      "grad_norm": 4.419593334197998,
      "learning_rate": 4.056096918619963e-05,
      "loss": 0.9466,
      "step": 7174
    },
    {
      "epoch": 0.1889649723465894,
      "grad_norm": 1.7070976495742798,
      "learning_rate": 4.0559652357124045e-05,
      "loss": 1.5835,
      "step": 7175
    },
    {
      "epoch": 0.18899130892810112,
      "grad_norm": 1.7554688453674316,
      "learning_rate": 4.055833552804846e-05,
      "loss": 2.2235,
      "step": 7176
    },
    {
      "epoch": 0.18901764550961286,
      "grad_norm": 1.7871851921081543,
      "learning_rate": 4.0557018698972876e-05,
      "loss": 1.5603,
      "step": 7177
    },
    {
      "epoch": 0.18904398209112458,
      "grad_norm": 2.1068508625030518,
      "learning_rate": 4.055570186989729e-05,
      "loss": 1.6217,
      "step": 7178
    },
    {
      "epoch": 0.1890703186726363,
      "grad_norm": 5.742961883544922,
      "learning_rate": 4.05543850408217e-05,
      "loss": 1.5055,
      "step": 7179
    },
    {
      "epoch": 0.189096655254148,
      "grad_norm": 1.645997405052185,
      "learning_rate": 4.0553068211746116e-05,
      "loss": 1.6799,
      "step": 7180
    },
    {
      "epoch": 0.18912299183565973,
      "grad_norm": 2.6304750442504883,
      "learning_rate": 4.055175138267053e-05,
      "loss": 1.5918,
      "step": 7181
    },
    {
      "epoch": 0.18914932841717144,
      "grad_norm": 1.7568247318267822,
      "learning_rate": 4.055043455359495e-05,
      "loss": 0.6936,
      "step": 7182
    },
    {
      "epoch": 0.18917566499868316,
      "grad_norm": 1.9679300785064697,
      "learning_rate": 4.0549117724519356e-05,
      "loss": 1.8451,
      "step": 7183
    },
    {
      "epoch": 0.1892020015801949,
      "grad_norm": 1.9350433349609375,
      "learning_rate": 4.054780089544377e-05,
      "loss": 1.8593,
      "step": 7184
    },
    {
      "epoch": 0.18922833816170662,
      "grad_norm": 1.7362843751907349,
      "learning_rate": 4.054648406636819e-05,
      "loss": 1.858,
      "step": 7185
    },
    {
      "epoch": 0.18925467474321833,
      "grad_norm": 1.7821425199508667,
      "learning_rate": 4.05451672372926e-05,
      "loss": 0.4698,
      "step": 7186
    },
    {
      "epoch": 0.18928101132473005,
      "grad_norm": 2.8015902042388916,
      "learning_rate": 4.054385040821702e-05,
      "loss": 1.702,
      "step": 7187
    },
    {
      "epoch": 0.18930734790624176,
      "grad_norm": 1.5574791431427002,
      "learning_rate": 4.054253357914143e-05,
      "loss": 2.206,
      "step": 7188
    },
    {
      "epoch": 0.18933368448775348,
      "grad_norm": 1.8306288719177246,
      "learning_rate": 4.054121675006584e-05,
      "loss": 1.5303,
      "step": 7189
    },
    {
      "epoch": 0.18936002106926522,
      "grad_norm": 2.4754111766815186,
      "learning_rate": 4.053989992099026e-05,
      "loss": 2.0363,
      "step": 7190
    },
    {
      "epoch": 0.18938635765077694,
      "grad_norm": 1.332369327545166,
      "learning_rate": 4.0538583091914674e-05,
      "loss": 0.481,
      "step": 7191
    },
    {
      "epoch": 0.18941269423228865,
      "grad_norm": 1.8940297365188599,
      "learning_rate": 4.053726626283908e-05,
      "loss": 2.0337,
      "step": 7192
    },
    {
      "epoch": 0.18943903081380037,
      "grad_norm": 4.1673583984375,
      "learning_rate": 4.05359494337635e-05,
      "loss": 2.2207,
      "step": 7193
    },
    {
      "epoch": 0.18946536739531208,
      "grad_norm": 2.8621959686279297,
      "learning_rate": 4.0534632604687914e-05,
      "loss": 1.5712,
      "step": 7194
    },
    {
      "epoch": 0.1894917039768238,
      "grad_norm": 2.3763270378112793,
      "learning_rate": 4.053331577561233e-05,
      "loss": 1.1065,
      "step": 7195
    },
    {
      "epoch": 0.18951804055833552,
      "grad_norm": 2.1661956310272217,
      "learning_rate": 4.0531998946536746e-05,
      "loss": 2.0365,
      "step": 7196
    },
    {
      "epoch": 0.18954437713984726,
      "grad_norm": 2.1663999557495117,
      "learning_rate": 4.0530682117461154e-05,
      "loss": 0.4715,
      "step": 7197
    },
    {
      "epoch": 0.18957071372135897,
      "grad_norm": 2.5032851696014404,
      "learning_rate": 4.052936528838557e-05,
      "loss": 1.6401,
      "step": 7198
    },
    {
      "epoch": 0.1895970503028707,
      "grad_norm": 5.326064109802246,
      "learning_rate": 4.052804845930998e-05,
      "loss": 2.4489,
      "step": 7199
    },
    {
      "epoch": 0.1896233868843824,
      "grad_norm": 1.6689244508743286,
      "learning_rate": 4.05267316302344e-05,
      "loss": 2.0198,
      "step": 7200
    },
    {
      "epoch": 0.18964972346589412,
      "grad_norm": 4.380797386169434,
      "learning_rate": 4.052541480115881e-05,
      "loss": 2.0752,
      "step": 7201
    },
    {
      "epoch": 0.18967606004740584,
      "grad_norm": 2.0880212783813477,
      "learning_rate": 4.0524097972083226e-05,
      "loss": 1.7595,
      "step": 7202
    },
    {
      "epoch": 0.18970239662891758,
      "grad_norm": 2.943312883377075,
      "learning_rate": 4.0522781143007634e-05,
      "loss": 1.2728,
      "step": 7203
    },
    {
      "epoch": 0.1897287332104293,
      "grad_norm": 2.3224871158599854,
      "learning_rate": 4.052146431393206e-05,
      "loss": 2.5447,
      "step": 7204
    },
    {
      "epoch": 0.189755069791941,
      "grad_norm": 5.911948204040527,
      "learning_rate": 4.0520147484856466e-05,
      "loss": 1.5255,
      "step": 7205
    },
    {
      "epoch": 0.18978140637345273,
      "grad_norm": 1.636635661125183,
      "learning_rate": 4.051883065578088e-05,
      "loss": 1.6019,
      "step": 7206
    },
    {
      "epoch": 0.18980774295496444,
      "grad_norm": 4.187703609466553,
      "learning_rate": 4.05175138267053e-05,
      "loss": 2.4874,
      "step": 7207
    },
    {
      "epoch": 0.18983407953647616,
      "grad_norm": 1.7526487112045288,
      "learning_rate": 4.0516196997629706e-05,
      "loss": 1.5253,
      "step": 7208
    },
    {
      "epoch": 0.18986041611798787,
      "grad_norm": 2.7439773082733154,
      "learning_rate": 4.051488016855413e-05,
      "loss": 0.5517,
      "step": 7209
    },
    {
      "epoch": 0.18988675269949962,
      "grad_norm": 1.7597113847732544,
      "learning_rate": 4.051356333947854e-05,
      "loss": 1.7332,
      "step": 7210
    },
    {
      "epoch": 0.18991308928101133,
      "grad_norm": 6.962954521179199,
      "learning_rate": 4.051224651040295e-05,
      "loss": 1.8243,
      "step": 7211
    },
    {
      "epoch": 0.18993942586252305,
      "grad_norm": 2.2786426544189453,
      "learning_rate": 4.051092968132736e-05,
      "loss": 2.069,
      "step": 7212
    },
    {
      "epoch": 0.18996576244403476,
      "grad_norm": 3.446714162826538,
      "learning_rate": 4.050961285225178e-05,
      "loss": 0.7096,
      "step": 7213
    },
    {
      "epoch": 0.18999209902554648,
      "grad_norm": 1.9481279850006104,
      "learning_rate": 4.050829602317619e-05,
      "loss": 1.5411,
      "step": 7214
    },
    {
      "epoch": 0.1900184356070582,
      "grad_norm": 3.3399834632873535,
      "learning_rate": 4.050697919410061e-05,
      "loss": 0.9776,
      "step": 7215
    },
    {
      "epoch": 0.1900447721885699,
      "grad_norm": 1.6336928606033325,
      "learning_rate": 4.0505662365025024e-05,
      "loss": 0.7945,
      "step": 7216
    },
    {
      "epoch": 0.19007110877008165,
      "grad_norm": 2.0583527088165283,
      "learning_rate": 4.050434553594943e-05,
      "loss": 1.8646,
      "step": 7217
    },
    {
      "epoch": 0.19009744535159337,
      "grad_norm": 4.298874855041504,
      "learning_rate": 4.0503028706873855e-05,
      "loss": 1.1679,
      "step": 7218
    },
    {
      "epoch": 0.19012378193310508,
      "grad_norm": 1.989362120628357,
      "learning_rate": 4.0501711877798264e-05,
      "loss": 1.9351,
      "step": 7219
    },
    {
      "epoch": 0.1901501185146168,
      "grad_norm": 1.8836759328842163,
      "learning_rate": 4.050039504872268e-05,
      "loss": 2.3136,
      "step": 7220
    },
    {
      "epoch": 0.19017645509612852,
      "grad_norm": 2.2447330951690674,
      "learning_rate": 4.049907821964709e-05,
      "loss": 0.2346,
      "step": 7221
    },
    {
      "epoch": 0.19020279167764023,
      "grad_norm": 1.6300921440124512,
      "learning_rate": 4.0497761390571504e-05,
      "loss": 2.6292,
      "step": 7222
    },
    {
      "epoch": 0.19022912825915197,
      "grad_norm": 1.7325414419174194,
      "learning_rate": 4.049644456149592e-05,
      "loss": 1.7103,
      "step": 7223
    },
    {
      "epoch": 0.1902554648406637,
      "grad_norm": 2.182236909866333,
      "learning_rate": 4.0495127732420335e-05,
      "loss": 1.5909,
      "step": 7224
    },
    {
      "epoch": 0.1902818014221754,
      "grad_norm": 2.5211610794067383,
      "learning_rate": 4.049381090334475e-05,
      "loss": 1.4177,
      "step": 7225
    },
    {
      "epoch": 0.19030813800368712,
      "grad_norm": 1.613010287284851,
      "learning_rate": 4.049249407426916e-05,
      "loss": 1.4293,
      "step": 7226
    },
    {
      "epoch": 0.19033447458519884,
      "grad_norm": 1.730679988861084,
      "learning_rate": 4.0491177245193575e-05,
      "loss": 1.6866,
      "step": 7227
    },
    {
      "epoch": 0.19036081116671055,
      "grad_norm": 1.7852935791015625,
      "learning_rate": 4.048986041611799e-05,
      "loss": 2.0999,
      "step": 7228
    },
    {
      "epoch": 0.19038714774822227,
      "grad_norm": 2.4702770709991455,
      "learning_rate": 4.0488543587042406e-05,
      "loss": 1.6319,
      "step": 7229
    },
    {
      "epoch": 0.190413484329734,
      "grad_norm": 1.6914503574371338,
      "learning_rate": 4.0487226757966815e-05,
      "loss": 1.3973,
      "step": 7230
    },
    {
      "epoch": 0.19043982091124573,
      "grad_norm": 4.83398962020874,
      "learning_rate": 4.048590992889123e-05,
      "loss": 1.7376,
      "step": 7231
    },
    {
      "epoch": 0.19046615749275744,
      "grad_norm": 4.740577220916748,
      "learning_rate": 4.0484593099815646e-05,
      "loss": 1.3972,
      "step": 7232
    },
    {
      "epoch": 0.19049249407426916,
      "grad_norm": 2.465036153793335,
      "learning_rate": 4.048327627074006e-05,
      "loss": 1.5403,
      "step": 7233
    },
    {
      "epoch": 0.19051883065578087,
      "grad_norm": 1.5254331827163696,
      "learning_rate": 4.048195944166448e-05,
      "loss": 1.8449,
      "step": 7234
    },
    {
      "epoch": 0.1905451672372926,
      "grad_norm": 2.33536434173584,
      "learning_rate": 4.0480642612588886e-05,
      "loss": 1.7078,
      "step": 7235
    },
    {
      "epoch": 0.1905715038188043,
      "grad_norm": 2.0740175247192383,
      "learning_rate": 4.04793257835133e-05,
      "loss": 2.3895,
      "step": 7236
    },
    {
      "epoch": 0.19059784040031605,
      "grad_norm": 1.8939977884292603,
      "learning_rate": 4.047800895443772e-05,
      "loss": 1.2493,
      "step": 7237
    },
    {
      "epoch": 0.19062417698182776,
      "grad_norm": 3.422680616378784,
      "learning_rate": 4.047669212536213e-05,
      "loss": 1.5391,
      "step": 7238
    },
    {
      "epoch": 0.19065051356333948,
      "grad_norm": 3.1775338649749756,
      "learning_rate": 4.047537529628654e-05,
      "loss": 1.6707,
      "step": 7239
    },
    {
      "epoch": 0.1906768501448512,
      "grad_norm": 3.0954630374908447,
      "learning_rate": 4.047405846721096e-05,
      "loss": 1.8821,
      "step": 7240
    },
    {
      "epoch": 0.1907031867263629,
      "grad_norm": 2.1109166145324707,
      "learning_rate": 4.047274163813537e-05,
      "loss": 2.3079,
      "step": 7241
    },
    {
      "epoch": 0.19072952330787463,
      "grad_norm": 3.317548990249634,
      "learning_rate": 4.047142480905979e-05,
      "loss": 2.4559,
      "step": 7242
    },
    {
      "epoch": 0.19075585988938637,
      "grad_norm": 2.191545248031616,
      "learning_rate": 4.0470107979984204e-05,
      "loss": 1.9968,
      "step": 7243
    },
    {
      "epoch": 0.19078219647089809,
      "grad_norm": 4.669031620025635,
      "learning_rate": 4.046879115090861e-05,
      "loss": 0.7549,
      "step": 7244
    },
    {
      "epoch": 0.1908085330524098,
      "grad_norm": 1.9186196327209473,
      "learning_rate": 4.046747432183303e-05,
      "loss": 1.6467,
      "step": 7245
    },
    {
      "epoch": 0.19083486963392152,
      "grad_norm": 4.040922164916992,
      "learning_rate": 4.046615749275744e-05,
      "loss": 0.5172,
      "step": 7246
    },
    {
      "epoch": 0.19086120621543323,
      "grad_norm": 2.6314992904663086,
      "learning_rate": 4.046484066368186e-05,
      "loss": 1.5515,
      "step": 7247
    },
    {
      "epoch": 0.19088754279694495,
      "grad_norm": 1.7427303791046143,
      "learning_rate": 4.046352383460627e-05,
      "loss": 0.3602,
      "step": 7248
    },
    {
      "epoch": 0.19091387937845666,
      "grad_norm": 3.3643410205841064,
      "learning_rate": 4.0462207005530684e-05,
      "loss": 1.5196,
      "step": 7249
    },
    {
      "epoch": 0.1909402159599684,
      "grad_norm": 2.2444002628326416,
      "learning_rate": 4.046089017645509e-05,
      "loss": 1.8196,
      "step": 7250
    },
    {
      "epoch": 0.19096655254148012,
      "grad_norm": 3.979214668273926,
      "learning_rate": 4.0459573347379515e-05,
      "loss": 2.3359,
      "step": 7251
    },
    {
      "epoch": 0.19099288912299184,
      "grad_norm": 3.34232497215271,
      "learning_rate": 4.0458256518303924e-05,
      "loss": 1.5687,
      "step": 7252
    },
    {
      "epoch": 0.19101922570450355,
      "grad_norm": 1.6711429357528687,
      "learning_rate": 4.045693968922834e-05,
      "loss": 1.9792,
      "step": 7253
    },
    {
      "epoch": 0.19104556228601527,
      "grad_norm": 4.695145606994629,
      "learning_rate": 4.0455622860152755e-05,
      "loss": 1.0408,
      "step": 7254
    },
    {
      "epoch": 0.19107189886752698,
      "grad_norm": 3.7791454792022705,
      "learning_rate": 4.0454306031077164e-05,
      "loss": 0.7948,
      "step": 7255
    },
    {
      "epoch": 0.19109823544903873,
      "grad_norm": 2.1113526821136475,
      "learning_rate": 4.0452989202001587e-05,
      "loss": 1.7528,
      "step": 7256
    },
    {
      "epoch": 0.19112457203055044,
      "grad_norm": 3.21728253364563,
      "learning_rate": 4.0451672372925995e-05,
      "loss": 1.6714,
      "step": 7257
    },
    {
      "epoch": 0.19115090861206216,
      "grad_norm": 1.7481330633163452,
      "learning_rate": 4.045035554385041e-05,
      "loss": 1.9234,
      "step": 7258
    },
    {
      "epoch": 0.19117724519357387,
      "grad_norm": 1.6544904708862305,
      "learning_rate": 4.044903871477482e-05,
      "loss": 1.7649,
      "step": 7259
    },
    {
      "epoch": 0.1912035817750856,
      "grad_norm": 1.567003846168518,
      "learning_rate": 4.0447721885699235e-05,
      "loss": 1.6215,
      "step": 7260
    },
    {
      "epoch": 0.1912299183565973,
      "grad_norm": 1.9183847904205322,
      "learning_rate": 4.044640505662365e-05,
      "loss": 1.7934,
      "step": 7261
    },
    {
      "epoch": 0.19125625493810902,
      "grad_norm": 2.2221667766571045,
      "learning_rate": 4.0445088227548067e-05,
      "loss": 1.9812,
      "step": 7262
    },
    {
      "epoch": 0.19128259151962076,
      "grad_norm": 4.729203701019287,
      "learning_rate": 4.044377139847248e-05,
      "loss": 0.9816,
      "step": 7263
    },
    {
      "epoch": 0.19130892810113248,
      "grad_norm": 1.99595046043396,
      "learning_rate": 4.044245456939689e-05,
      "loss": 2.0178,
      "step": 7264
    },
    {
      "epoch": 0.1913352646826442,
      "grad_norm": 2.347024440765381,
      "learning_rate": 4.044113774032131e-05,
      "loss": 1.3836,
      "step": 7265
    },
    {
      "epoch": 0.1913616012641559,
      "grad_norm": 1.84151029586792,
      "learning_rate": 4.043982091124572e-05,
      "loss": 2.1222,
      "step": 7266
    },
    {
      "epoch": 0.19138793784566763,
      "grad_norm": 1.9768502712249756,
      "learning_rate": 4.043850408217014e-05,
      "loss": 2.3555,
      "step": 7267
    },
    {
      "epoch": 0.19141427442717934,
      "grad_norm": 1.647308349609375,
      "learning_rate": 4.0437187253094547e-05,
      "loss": 1.7452,
      "step": 7268
    },
    {
      "epoch": 0.19144061100869106,
      "grad_norm": 1.953292965888977,
      "learning_rate": 4.043587042401896e-05,
      "loss": 2.0305,
      "step": 7269
    },
    {
      "epoch": 0.1914669475902028,
      "grad_norm": 2.776155471801758,
      "learning_rate": 4.043455359494338e-05,
      "loss": 1.7882,
      "step": 7270
    },
    {
      "epoch": 0.19149328417171452,
      "grad_norm": 5.300607681274414,
      "learning_rate": 4.0433236765867793e-05,
      "loss": 1.7522,
      "step": 7271
    },
    {
      "epoch": 0.19151962075322623,
      "grad_norm": 1.906984567642212,
      "learning_rate": 4.043191993679221e-05,
      "loss": 1.6374,
      "step": 7272
    },
    {
      "epoch": 0.19154595733473795,
      "grad_norm": 1.7142235040664673,
      "learning_rate": 4.043060310771662e-05,
      "loss": 1.8931,
      "step": 7273
    },
    {
      "epoch": 0.19157229391624966,
      "grad_norm": 2.2480897903442383,
      "learning_rate": 4.0429286278641033e-05,
      "loss": 1.9641,
      "step": 7274
    },
    {
      "epoch": 0.19159863049776138,
      "grad_norm": 1.8112406730651855,
      "learning_rate": 4.042796944956545e-05,
      "loss": 1.6118,
      "step": 7275
    },
    {
      "epoch": 0.19162496707927312,
      "grad_norm": 1.7845630645751953,
      "learning_rate": 4.0426652620489865e-05,
      "loss": 1.8619,
      "step": 7276
    },
    {
      "epoch": 0.19165130366078484,
      "grad_norm": 2.2230772972106934,
      "learning_rate": 4.0425335791414273e-05,
      "loss": 2.0625,
      "step": 7277
    },
    {
      "epoch": 0.19167764024229655,
      "grad_norm": 2.6940886974334717,
      "learning_rate": 4.042401896233869e-05,
      "loss": 1.5254,
      "step": 7278
    },
    {
      "epoch": 0.19170397682380827,
      "grad_norm": 1.3871856927871704,
      "learning_rate": 4.0422702133263105e-05,
      "loss": 1.9714,
      "step": 7279
    },
    {
      "epoch": 0.19173031340531999,
      "grad_norm": 3.610910177230835,
      "learning_rate": 4.042138530418752e-05,
      "loss": 2.1208,
      "step": 7280
    },
    {
      "epoch": 0.1917566499868317,
      "grad_norm": 1.9261505603790283,
      "learning_rate": 4.0420068475111936e-05,
      "loss": 1.9128,
      "step": 7281
    },
    {
      "epoch": 0.19178298656834342,
      "grad_norm": 1.932108998298645,
      "learning_rate": 4.0418751646036345e-05,
      "loss": 1.8876,
      "step": 7282
    },
    {
      "epoch": 0.19180932314985516,
      "grad_norm": 2.5684380531311035,
      "learning_rate": 4.041743481696076e-05,
      "loss": 1.8817,
      "step": 7283
    },
    {
      "epoch": 0.19183565973136688,
      "grad_norm": 1.7948907613754272,
      "learning_rate": 4.0416117987885176e-05,
      "loss": 2.3632,
      "step": 7284
    },
    {
      "epoch": 0.1918619963128786,
      "grad_norm": 1.8713473081588745,
      "learning_rate": 4.041480115880959e-05,
      "loss": 1.821,
      "step": 7285
    },
    {
      "epoch": 0.1918883328943903,
      "grad_norm": 2.3834478855133057,
      "learning_rate": 4.0413484329734e-05,
      "loss": 2.336,
      "step": 7286
    },
    {
      "epoch": 0.19191466947590202,
      "grad_norm": 2.312260389328003,
      "learning_rate": 4.0412167500658416e-05,
      "loss": 1.5203,
      "step": 7287
    },
    {
      "epoch": 0.19194100605741374,
      "grad_norm": 2.2332422733306885,
      "learning_rate": 4.041085067158283e-05,
      "loss": 2.0257,
      "step": 7288
    },
    {
      "epoch": 0.19196734263892548,
      "grad_norm": 3.375622272491455,
      "learning_rate": 4.040953384250725e-05,
      "loss": 1.1201,
      "step": 7289
    },
    {
      "epoch": 0.1919936792204372,
      "grad_norm": 2.6326589584350586,
      "learning_rate": 4.040821701343166e-05,
      "loss": 1.9488,
      "step": 7290
    },
    {
      "epoch": 0.1920200158019489,
      "grad_norm": 3.281137228012085,
      "learning_rate": 4.040690018435607e-05,
      "loss": 2.6588,
      "step": 7291
    },
    {
      "epoch": 0.19204635238346063,
      "grad_norm": 3.1981658935546875,
      "learning_rate": 4.040558335528049e-05,
      "loss": 1.3096,
      "step": 7292
    },
    {
      "epoch": 0.19207268896497234,
      "grad_norm": 2.4739367961883545,
      "learning_rate": 4.0404266526204896e-05,
      "loss": 1.3927,
      "step": 7293
    },
    {
      "epoch": 0.19209902554648406,
      "grad_norm": 4.966895580291748,
      "learning_rate": 4.040294969712932e-05,
      "loss": 1.7816,
      "step": 7294
    },
    {
      "epoch": 0.19212536212799577,
      "grad_norm": 3.845078229904175,
      "learning_rate": 4.040163286805373e-05,
      "loss": 1.0441,
      "step": 7295
    },
    {
      "epoch": 0.19215169870950752,
      "grad_norm": 2.0101237297058105,
      "learning_rate": 4.040031603897814e-05,
      "loss": 1.9505,
      "step": 7296
    },
    {
      "epoch": 0.19217803529101923,
      "grad_norm": 1.7244296073913574,
      "learning_rate": 4.039899920990256e-05,
      "loss": 1.1956,
      "step": 7297
    },
    {
      "epoch": 0.19220437187253095,
      "grad_norm": 1.756265640258789,
      "learning_rate": 4.0397682380826974e-05,
      "loss": 2.237,
      "step": 7298
    },
    {
      "epoch": 0.19223070845404266,
      "grad_norm": 2.3653173446655273,
      "learning_rate": 4.039636555175139e-05,
      "loss": 1.6345,
      "step": 7299
    },
    {
      "epoch": 0.19225704503555438,
      "grad_norm": 6.648596286773682,
      "learning_rate": 4.03950487226758e-05,
      "loss": 2.5191,
      "step": 7300
    },
    {
      "epoch": 0.1922833816170661,
      "grad_norm": 2.2856199741363525,
      "learning_rate": 4.0393731893600214e-05,
      "loss": 2.0633,
      "step": 7301
    },
    {
      "epoch": 0.1923097181985778,
      "grad_norm": 3.177938938140869,
      "learning_rate": 4.039241506452462e-05,
      "loss": 1.6583,
      "step": 7302
    },
    {
      "epoch": 0.19233605478008955,
      "grad_norm": 1.6299958229064941,
      "learning_rate": 4.0391098235449045e-05,
      "loss": 1.2568,
      "step": 7303
    },
    {
      "epoch": 0.19236239136160127,
      "grad_norm": 1.894034504890442,
      "learning_rate": 4.0389781406373454e-05,
      "loss": 2.1901,
      "step": 7304
    },
    {
      "epoch": 0.19238872794311299,
      "grad_norm": 2.6950366497039795,
      "learning_rate": 4.038846457729787e-05,
      "loss": 0.8633,
      "step": 7305
    },
    {
      "epoch": 0.1924150645246247,
      "grad_norm": 3.5823185443878174,
      "learning_rate": 4.038714774822228e-05,
      "loss": 1.4209,
      "step": 7306
    },
    {
      "epoch": 0.19244140110613642,
      "grad_norm": 1.8233287334442139,
      "learning_rate": 4.0385830919146694e-05,
      "loss": 1.8241,
      "step": 7307
    },
    {
      "epoch": 0.19246773768764813,
      "grad_norm": 6.177656173706055,
      "learning_rate": 4.038451409007111e-05,
      "loss": 1.9694,
      "step": 7308
    },
    {
      "epoch": 0.19249407426915988,
      "grad_norm": 1.7469977140426636,
      "learning_rate": 4.0383197260995525e-05,
      "loss": 2.02,
      "step": 7309
    },
    {
      "epoch": 0.1925204108506716,
      "grad_norm": 1.8026920557022095,
      "learning_rate": 4.038188043191994e-05,
      "loss": 2.1229,
      "step": 7310
    },
    {
      "epoch": 0.1925467474321833,
      "grad_norm": 2.9835667610168457,
      "learning_rate": 4.038056360284435e-05,
      "loss": 1.9499,
      "step": 7311
    },
    {
      "epoch": 0.19257308401369502,
      "grad_norm": 2.2449615001678467,
      "learning_rate": 4.0379246773768765e-05,
      "loss": 1.8997,
      "step": 7312
    },
    {
      "epoch": 0.19259942059520674,
      "grad_norm": 1.4395685195922852,
      "learning_rate": 4.037792994469318e-05,
      "loss": 1.742,
      "step": 7313
    },
    {
      "epoch": 0.19262575717671845,
      "grad_norm": 1.9676905870437622,
      "learning_rate": 4.0376613115617596e-05,
      "loss": 2.0727,
      "step": 7314
    },
    {
      "epoch": 0.19265209375823017,
      "grad_norm": 1.7489992380142212,
      "learning_rate": 4.0375296286542005e-05,
      "loss": 1.4162,
      "step": 7315
    },
    {
      "epoch": 0.1926784303397419,
      "grad_norm": 2.629103660583496,
      "learning_rate": 4.037397945746642e-05,
      "loss": 2.1102,
      "step": 7316
    },
    {
      "epoch": 0.19270476692125363,
      "grad_norm": 2.397937059402466,
      "learning_rate": 4.0372662628390836e-05,
      "loss": 1.4276,
      "step": 7317
    },
    {
      "epoch": 0.19273110350276534,
      "grad_norm": 1.9332218170166016,
      "learning_rate": 4.037134579931525e-05,
      "loss": 2.0168,
      "step": 7318
    },
    {
      "epoch": 0.19275744008427706,
      "grad_norm": 2.219738483428955,
      "learning_rate": 4.037002897023967e-05,
      "loss": 1.9417,
      "step": 7319
    },
    {
      "epoch": 0.19278377666578878,
      "grad_norm": 2.0492563247680664,
      "learning_rate": 4.0368712141164076e-05,
      "loss": 1.4014,
      "step": 7320
    },
    {
      "epoch": 0.1928101132473005,
      "grad_norm": 3.0343124866485596,
      "learning_rate": 4.036739531208849e-05,
      "loss": 1.321,
      "step": 7321
    },
    {
      "epoch": 0.1928364498288122,
      "grad_norm": 3.1353862285614014,
      "learning_rate": 4.036607848301291e-05,
      "loss": 0.5889,
      "step": 7322
    },
    {
      "epoch": 0.19286278641032395,
      "grad_norm": 2.5549745559692383,
      "learning_rate": 4.036476165393732e-05,
      "loss": 1.016,
      "step": 7323
    },
    {
      "epoch": 0.19288912299183567,
      "grad_norm": 3.035165786743164,
      "learning_rate": 4.036344482486173e-05,
      "loss": 1.497,
      "step": 7324
    },
    {
      "epoch": 0.19291545957334738,
      "grad_norm": 2.4611964225769043,
      "learning_rate": 4.036212799578615e-05,
      "loss": 1.3419,
      "step": 7325
    },
    {
      "epoch": 0.1929417961548591,
      "grad_norm": 2.7616629600524902,
      "learning_rate": 4.036081116671056e-05,
      "loss": 1.9925,
      "step": 7326
    },
    {
      "epoch": 0.1929681327363708,
      "grad_norm": 4.698431491851807,
      "learning_rate": 4.035949433763498e-05,
      "loss": 0.7497,
      "step": 7327
    },
    {
      "epoch": 0.19299446931788253,
      "grad_norm": 1.5474358797073364,
      "learning_rate": 4.0358177508559394e-05,
      "loss": 1.6545,
      "step": 7328
    },
    {
      "epoch": 0.19302080589939427,
      "grad_norm": 4.562552452087402,
      "learning_rate": 4.03568606794838e-05,
      "loss": 1.0649,
      "step": 7329
    },
    {
      "epoch": 0.193047142480906,
      "grad_norm": 3.056469202041626,
      "learning_rate": 4.035554385040822e-05,
      "loss": 1.6468,
      "step": 7330
    },
    {
      "epoch": 0.1930734790624177,
      "grad_norm": 1.775338888168335,
      "learning_rate": 4.0354227021332634e-05,
      "loss": 2.0318,
      "step": 7331
    },
    {
      "epoch": 0.19309981564392942,
      "grad_norm": 3.844957113265991,
      "learning_rate": 4.035291019225705e-05,
      "loss": 1.6792,
      "step": 7332
    },
    {
      "epoch": 0.19312615222544113,
      "grad_norm": 2.6300363540649414,
      "learning_rate": 4.035159336318146e-05,
      "loss": 1.4449,
      "step": 7333
    },
    {
      "epoch": 0.19315248880695285,
      "grad_norm": 2.6600770950317383,
      "learning_rate": 4.0350276534105874e-05,
      "loss": 1.8157,
      "step": 7334
    },
    {
      "epoch": 0.19317882538846456,
      "grad_norm": 1.9569789171218872,
      "learning_rate": 4.034895970503029e-05,
      "loss": 1.7563,
      "step": 7335
    },
    {
      "epoch": 0.1932051619699763,
      "grad_norm": 2.6759989261627197,
      "learning_rate": 4.0347642875954706e-05,
      "loss": 1.6504,
      "step": 7336
    },
    {
      "epoch": 0.19323149855148802,
      "grad_norm": 2.748774766921997,
      "learning_rate": 4.034632604687912e-05,
      "loss": 1.4448,
      "step": 7337
    },
    {
      "epoch": 0.19325783513299974,
      "grad_norm": 3.0889992713928223,
      "learning_rate": 4.034500921780353e-05,
      "loss": 0.995,
      "step": 7338
    },
    {
      "epoch": 0.19328417171451145,
      "grad_norm": 2.6113176345825195,
      "learning_rate": 4.0343692388727946e-05,
      "loss": 1.7881,
      "step": 7339
    },
    {
      "epoch": 0.19331050829602317,
      "grad_norm": 3.723393440246582,
      "learning_rate": 4.0342375559652354e-05,
      "loss": 0.6865,
      "step": 7340
    },
    {
      "epoch": 0.19333684487753489,
      "grad_norm": 1.691326379776001,
      "learning_rate": 4.034105873057678e-05,
      "loss": 1.8833,
      "step": 7341
    },
    {
      "epoch": 0.19336318145904663,
      "grad_norm": 1.5007450580596924,
      "learning_rate": 4.0339741901501186e-05,
      "loss": 0.3572,
      "step": 7342
    },
    {
      "epoch": 0.19338951804055834,
      "grad_norm": 5.650027275085449,
      "learning_rate": 4.03384250724256e-05,
      "loss": 0.983,
      "step": 7343
    },
    {
      "epoch": 0.19341585462207006,
      "grad_norm": 2.8245296478271484,
      "learning_rate": 4.033710824335002e-05,
      "loss": 1.5689,
      "step": 7344
    },
    {
      "epoch": 0.19344219120358178,
      "grad_norm": 2.153578042984009,
      "learning_rate": 4.0335791414274426e-05,
      "loss": 1.3835,
      "step": 7345
    },
    {
      "epoch": 0.1934685277850935,
      "grad_norm": 3.891183853149414,
      "learning_rate": 4.033447458519885e-05,
      "loss": 1.5072,
      "step": 7346
    },
    {
      "epoch": 0.1934948643666052,
      "grad_norm": 2.4269511699676514,
      "learning_rate": 4.033315775612326e-05,
      "loss": 1.9788,
      "step": 7347
    },
    {
      "epoch": 0.19352120094811692,
      "grad_norm": 2.1383001804351807,
      "learning_rate": 4.033184092704767e-05,
      "loss": 1.8104,
      "step": 7348
    },
    {
      "epoch": 0.19354753752962867,
      "grad_norm": 1.9490697383880615,
      "learning_rate": 4.033052409797208e-05,
      "loss": 1.573,
      "step": 7349
    },
    {
      "epoch": 0.19357387411114038,
      "grad_norm": 2.5149269104003906,
      "learning_rate": 4.0329207268896504e-05,
      "loss": 2.5045,
      "step": 7350
    },
    {
      "epoch": 0.1936002106926521,
      "grad_norm": 1.8030511140823364,
      "learning_rate": 4.032789043982091e-05,
      "loss": 1.9566,
      "step": 7351
    },
    {
      "epoch": 0.1936265472741638,
      "grad_norm": 2.608902931213379,
      "learning_rate": 4.032657361074533e-05,
      "loss": 0.9981,
      "step": 7352
    },
    {
      "epoch": 0.19365288385567553,
      "grad_norm": 1.9074355363845825,
      "learning_rate": 4.032525678166974e-05,
      "loss": 1.7424,
      "step": 7353
    },
    {
      "epoch": 0.19367922043718724,
      "grad_norm": 2.6567094326019287,
      "learning_rate": 4.032393995259415e-05,
      "loss": 2.2168,
      "step": 7354
    },
    {
      "epoch": 0.19370555701869896,
      "grad_norm": 1.9052550792694092,
      "learning_rate": 4.032262312351857e-05,
      "loss": 1.5677,
      "step": 7355
    },
    {
      "epoch": 0.1937318936002107,
      "grad_norm": 1.7533397674560547,
      "learning_rate": 4.0321306294442984e-05,
      "loss": 1.8813,
      "step": 7356
    },
    {
      "epoch": 0.19375823018172242,
      "grad_norm": 1.938494324684143,
      "learning_rate": 4.03199894653674e-05,
      "loss": 2.1152,
      "step": 7357
    },
    {
      "epoch": 0.19378456676323413,
      "grad_norm": 2.4668755531311035,
      "learning_rate": 4.031867263629181e-05,
      "loss": 1.6126,
      "step": 7358
    },
    {
      "epoch": 0.19381090334474585,
      "grad_norm": 2.6709344387054443,
      "learning_rate": 4.0317355807216224e-05,
      "loss": 2.4992,
      "step": 7359
    },
    {
      "epoch": 0.19383723992625757,
      "grad_norm": 3.1036911010742188,
      "learning_rate": 4.031603897814064e-05,
      "loss": 2.1746,
      "step": 7360
    },
    {
      "epoch": 0.19386357650776928,
      "grad_norm": 7.545497417449951,
      "learning_rate": 4.0314722149065055e-05,
      "loss": 1.3802,
      "step": 7361
    },
    {
      "epoch": 0.19388991308928102,
      "grad_norm": 2.495110511779785,
      "learning_rate": 4.0313405319989464e-05,
      "loss": 1.4182,
      "step": 7362
    },
    {
      "epoch": 0.19391624967079274,
      "grad_norm": 2.202874183654785,
      "learning_rate": 4.031208849091388e-05,
      "loss": 1.9828,
      "step": 7363
    },
    {
      "epoch": 0.19394258625230446,
      "grad_norm": 1.6390376091003418,
      "learning_rate": 4.0310771661838295e-05,
      "loss": 2.7577,
      "step": 7364
    },
    {
      "epoch": 0.19396892283381617,
      "grad_norm": 2.361163377761841,
      "learning_rate": 4.030945483276271e-05,
      "loss": 2.1492,
      "step": 7365
    },
    {
      "epoch": 0.1939952594153279,
      "grad_norm": 3.4130513668060303,
      "learning_rate": 4.0308138003687126e-05,
      "loss": 0.7838,
      "step": 7366
    },
    {
      "epoch": 0.1940215959968396,
      "grad_norm": 2.5923759937286377,
      "learning_rate": 4.0306821174611535e-05,
      "loss": 2.2555,
      "step": 7367
    },
    {
      "epoch": 0.19404793257835132,
      "grad_norm": 2.3452064990997314,
      "learning_rate": 4.030550434553595e-05,
      "loss": 1.9365,
      "step": 7368
    },
    {
      "epoch": 0.19407426915986306,
      "grad_norm": 2.118030548095703,
      "learning_rate": 4.0304187516460366e-05,
      "loss": 1.3735,
      "step": 7369
    },
    {
      "epoch": 0.19410060574137478,
      "grad_norm": 1.9401692152023315,
      "learning_rate": 4.030287068738478e-05,
      "loss": 1.5864,
      "step": 7370
    },
    {
      "epoch": 0.1941269423228865,
      "grad_norm": 2.1829240322113037,
      "learning_rate": 4.030155385830919e-05,
      "loss": 1.7901,
      "step": 7371
    },
    {
      "epoch": 0.1941532789043982,
      "grad_norm": 1.7431983947753906,
      "learning_rate": 4.0300237029233606e-05,
      "loss": 1.653,
      "step": 7372
    },
    {
      "epoch": 0.19417961548590992,
      "grad_norm": 2.2366416454315186,
      "learning_rate": 4.029892020015802e-05,
      "loss": 1.4286,
      "step": 7373
    },
    {
      "epoch": 0.19420595206742164,
      "grad_norm": 4.20750617980957,
      "learning_rate": 4.029760337108244e-05,
      "loss": 1.6322,
      "step": 7374
    },
    {
      "epoch": 0.19423228864893338,
      "grad_norm": 1.7967501878738403,
      "learning_rate": 4.029628654200685e-05,
      "loss": 1.71,
      "step": 7375
    },
    {
      "epoch": 0.1942586252304451,
      "grad_norm": 1.6806516647338867,
      "learning_rate": 4.029496971293126e-05,
      "loss": 1.9395,
      "step": 7376
    },
    {
      "epoch": 0.1942849618119568,
      "grad_norm": 1.7641276121139526,
      "learning_rate": 4.029365288385568e-05,
      "loss": 1.8902,
      "step": 7377
    },
    {
      "epoch": 0.19431129839346853,
      "grad_norm": 4.335816860198975,
      "learning_rate": 4.0292336054780086e-05,
      "loss": 1.4072,
      "step": 7378
    },
    {
      "epoch": 0.19433763497498024,
      "grad_norm": 2.0037953853607178,
      "learning_rate": 4.029101922570451e-05,
      "loss": 1.7427,
      "step": 7379
    },
    {
      "epoch": 0.19436397155649196,
      "grad_norm": 4.062325954437256,
      "learning_rate": 4.028970239662892e-05,
      "loss": 1.2865,
      "step": 7380
    },
    {
      "epoch": 0.19439030813800368,
      "grad_norm": 4.657859802246094,
      "learning_rate": 4.028838556755333e-05,
      "loss": 1.484,
      "step": 7381
    },
    {
      "epoch": 0.19441664471951542,
      "grad_norm": 2.2243223190307617,
      "learning_rate": 4.028706873847775e-05,
      "loss": 1.552,
      "step": 7382
    },
    {
      "epoch": 0.19444298130102713,
      "grad_norm": 2.2360358238220215,
      "learning_rate": 4.0285751909402164e-05,
      "loss": 1.6119,
      "step": 7383
    },
    {
      "epoch": 0.19446931788253885,
      "grad_norm": 4.571104526519775,
      "learning_rate": 4.028443508032658e-05,
      "loss": 1.6022,
      "step": 7384
    },
    {
      "epoch": 0.19449565446405057,
      "grad_norm": 2.8746490478515625,
      "learning_rate": 4.028311825125099e-05,
      "loss": 1.3261,
      "step": 7385
    },
    {
      "epoch": 0.19452199104556228,
      "grad_norm": 2.1421756744384766,
      "learning_rate": 4.0281801422175404e-05,
      "loss": 1.5621,
      "step": 7386
    },
    {
      "epoch": 0.194548327627074,
      "grad_norm": 1.843734622001648,
      "learning_rate": 4.028048459309981e-05,
      "loss": 1.5633,
      "step": 7387
    },
    {
      "epoch": 0.1945746642085857,
      "grad_norm": 4.400885581970215,
      "learning_rate": 4.0279167764024235e-05,
      "loss": 1.0452,
      "step": 7388
    },
    {
      "epoch": 0.19460100079009746,
      "grad_norm": 2.5814390182495117,
      "learning_rate": 4.0277850934948644e-05,
      "loss": 1.5494,
      "step": 7389
    },
    {
      "epoch": 0.19462733737160917,
      "grad_norm": 2.7782809734344482,
      "learning_rate": 4.027653410587306e-05,
      "loss": 2.3513,
      "step": 7390
    },
    {
      "epoch": 0.1946536739531209,
      "grad_norm": 2.638397216796875,
      "learning_rate": 4.0275217276797475e-05,
      "loss": 0.6205,
      "step": 7391
    },
    {
      "epoch": 0.1946800105346326,
      "grad_norm": 1.8589435815811157,
      "learning_rate": 4.0273900447721884e-05,
      "loss": 1.2768,
      "step": 7392
    },
    {
      "epoch": 0.19470634711614432,
      "grad_norm": 1.9502047300338745,
      "learning_rate": 4.0272583618646307e-05,
      "loss": 2.1278,
      "step": 7393
    },
    {
      "epoch": 0.19473268369765603,
      "grad_norm": 1.6248949766159058,
      "learning_rate": 4.0271266789570715e-05,
      "loss": 1.6731,
      "step": 7394
    },
    {
      "epoch": 0.19475902027916778,
      "grad_norm": 1.7083396911621094,
      "learning_rate": 4.026994996049513e-05,
      "loss": 1.6324,
      "step": 7395
    },
    {
      "epoch": 0.1947853568606795,
      "grad_norm": 2.254757881164551,
      "learning_rate": 4.026863313141954e-05,
      "loss": 2.2385,
      "step": 7396
    },
    {
      "epoch": 0.1948116934421912,
      "grad_norm": 2.203688144683838,
      "learning_rate": 4.026731630234396e-05,
      "loss": 1.8163,
      "step": 7397
    },
    {
      "epoch": 0.19483803002370292,
      "grad_norm": 1.5879720449447632,
      "learning_rate": 4.026599947326837e-05,
      "loss": 1.5672,
      "step": 7398
    },
    {
      "epoch": 0.19486436660521464,
      "grad_norm": 4.591485023498535,
      "learning_rate": 4.026468264419279e-05,
      "loss": 2.2126,
      "step": 7399
    },
    {
      "epoch": 0.19489070318672636,
      "grad_norm": 3.664841890335083,
      "learning_rate": 4.02633658151172e-05,
      "loss": 0.6928,
      "step": 7400
    },
    {
      "epoch": 0.19491703976823807,
      "grad_norm": 3.8438456058502197,
      "learning_rate": 4.026204898604161e-05,
      "loss": 1.3409,
      "step": 7401
    },
    {
      "epoch": 0.19494337634974981,
      "grad_norm": 1.617113709449768,
      "learning_rate": 4.0260732156966033e-05,
      "loss": 1.7714,
      "step": 7402
    },
    {
      "epoch": 0.19496971293126153,
      "grad_norm": 4.050449848175049,
      "learning_rate": 4.025941532789044e-05,
      "loss": 1.5317,
      "step": 7403
    },
    {
      "epoch": 0.19499604951277325,
      "grad_norm": 4.882868766784668,
      "learning_rate": 4.025809849881486e-05,
      "loss": 1.1551,
      "step": 7404
    },
    {
      "epoch": 0.19502238609428496,
      "grad_norm": 2.0997579097747803,
      "learning_rate": 4.025678166973927e-05,
      "loss": 0.6815,
      "step": 7405
    },
    {
      "epoch": 0.19504872267579668,
      "grad_norm": 2.240664482116699,
      "learning_rate": 4.025546484066368e-05,
      "loss": 2.0105,
      "step": 7406
    },
    {
      "epoch": 0.1950750592573084,
      "grad_norm": 2.197084426879883,
      "learning_rate": 4.02541480115881e-05,
      "loss": 1.0667,
      "step": 7407
    },
    {
      "epoch": 0.1951013958388201,
      "grad_norm": 4.773092746734619,
      "learning_rate": 4.0252831182512513e-05,
      "loss": 2.2518,
      "step": 7408
    },
    {
      "epoch": 0.19512773242033185,
      "grad_norm": 4.768918991088867,
      "learning_rate": 4.025151435343692e-05,
      "loss": 1.3712,
      "step": 7409
    },
    {
      "epoch": 0.19515406900184357,
      "grad_norm": 2.93772029876709,
      "learning_rate": 4.025019752436134e-05,
      "loss": 2.4711,
      "step": 7410
    },
    {
      "epoch": 0.19518040558335528,
      "grad_norm": 1.7906513214111328,
      "learning_rate": 4.0248880695285753e-05,
      "loss": 1.4001,
      "step": 7411
    },
    {
      "epoch": 0.195206742164867,
      "grad_norm": 3.617037296295166,
      "learning_rate": 4.024756386621017e-05,
      "loss": 1.9739,
      "step": 7412
    },
    {
      "epoch": 0.1952330787463787,
      "grad_norm": 1.9617981910705566,
      "learning_rate": 4.0246247037134585e-05,
      "loss": 0.5342,
      "step": 7413
    },
    {
      "epoch": 0.19525941532789043,
      "grad_norm": 3.30856990814209,
      "learning_rate": 4.0244930208058993e-05,
      "loss": 1.9601,
      "step": 7414
    },
    {
      "epoch": 0.19528575190940217,
      "grad_norm": 4.260293960571289,
      "learning_rate": 4.024361337898341e-05,
      "loss": 1.744,
      "step": 7415
    },
    {
      "epoch": 0.1953120884909139,
      "grad_norm": 1.68104088306427,
      "learning_rate": 4.0242296549907825e-05,
      "loss": 1.4665,
      "step": 7416
    },
    {
      "epoch": 0.1953384250724256,
      "grad_norm": 2.505425214767456,
      "learning_rate": 4.024097972083224e-05,
      "loss": 1.1959,
      "step": 7417
    },
    {
      "epoch": 0.19536476165393732,
      "grad_norm": 7.117185115814209,
      "learning_rate": 4.023966289175665e-05,
      "loss": 1.3316,
      "step": 7418
    },
    {
      "epoch": 0.19539109823544903,
      "grad_norm": 1.8270654678344727,
      "learning_rate": 4.0238346062681065e-05,
      "loss": 1.7066,
      "step": 7419
    },
    {
      "epoch": 0.19541743481696075,
      "grad_norm": 1.9719904661178589,
      "learning_rate": 4.023702923360548e-05,
      "loss": 2.1703,
      "step": 7420
    },
    {
      "epoch": 0.19544377139847247,
      "grad_norm": 3.1282644271850586,
      "learning_rate": 4.0235712404529896e-05,
      "loss": 1.3494,
      "step": 7421
    },
    {
      "epoch": 0.1954701079799842,
      "grad_norm": 4.2231340408325195,
      "learning_rate": 4.023439557545431e-05,
      "loss": 1.9168,
      "step": 7422
    },
    {
      "epoch": 0.19549644456149592,
      "grad_norm": 3.291358709335327,
      "learning_rate": 4.023307874637872e-05,
      "loss": 1.4818,
      "step": 7423
    },
    {
      "epoch": 0.19552278114300764,
      "grad_norm": 3.567249298095703,
      "learning_rate": 4.0231761917303136e-05,
      "loss": 1.9085,
      "step": 7424
    },
    {
      "epoch": 0.19554911772451936,
      "grad_norm": 3.560485601425171,
      "learning_rate": 4.0230445088227545e-05,
      "loss": 1.199,
      "step": 7425
    },
    {
      "epoch": 0.19557545430603107,
      "grad_norm": 2.3450729846954346,
      "learning_rate": 4.022912825915197e-05,
      "loss": 1.7006,
      "step": 7426
    },
    {
      "epoch": 0.1956017908875428,
      "grad_norm": 3.816150665283203,
      "learning_rate": 4.0227811430076376e-05,
      "loss": 1.1221,
      "step": 7427
    },
    {
      "epoch": 0.19562812746905453,
      "grad_norm": 2.1040356159210205,
      "learning_rate": 4.022649460100079e-05,
      "loss": 1.8347,
      "step": 7428
    },
    {
      "epoch": 0.19565446405056625,
      "grad_norm": 1.7644938230514526,
      "learning_rate": 4.022517777192521e-05,
      "loss": 2.7302,
      "step": 7429
    },
    {
      "epoch": 0.19568080063207796,
      "grad_norm": 2.0935416221618652,
      "learning_rate": 4.022386094284962e-05,
      "loss": 2.4944,
      "step": 7430
    },
    {
      "epoch": 0.19570713721358968,
      "grad_norm": 6.663039684295654,
      "learning_rate": 4.022254411377404e-05,
      "loss": 2.3264,
      "step": 7431
    },
    {
      "epoch": 0.1957334737951014,
      "grad_norm": 4.065506935119629,
      "learning_rate": 4.022122728469845e-05,
      "loss": 1.9679,
      "step": 7432
    },
    {
      "epoch": 0.1957598103766131,
      "grad_norm": 1.5186145305633545,
      "learning_rate": 4.021991045562286e-05,
      "loss": 1.8782,
      "step": 7433
    },
    {
      "epoch": 0.19578614695812482,
      "grad_norm": 2.1323468685150146,
      "learning_rate": 4.021859362654727e-05,
      "loss": 1.5221,
      "step": 7434
    },
    {
      "epoch": 0.19581248353963657,
      "grad_norm": 2.432612419128418,
      "learning_rate": 4.0217276797471694e-05,
      "loss": 1.9125,
      "step": 7435
    },
    {
      "epoch": 0.19583882012114828,
      "grad_norm": 5.779027938842773,
      "learning_rate": 4.02159599683961e-05,
      "loss": 1.1144,
      "step": 7436
    },
    {
      "epoch": 0.19586515670266,
      "grad_norm": 1.8980718851089478,
      "learning_rate": 4.021464313932052e-05,
      "loss": 1.9799,
      "step": 7437
    },
    {
      "epoch": 0.1958914932841717,
      "grad_norm": 4.200015068054199,
      "learning_rate": 4.0213326310244934e-05,
      "loss": 1.3378,
      "step": 7438
    },
    {
      "epoch": 0.19591782986568343,
      "grad_norm": 8.51160717010498,
      "learning_rate": 4.021200948116934e-05,
      "loss": 1.6298,
      "step": 7439
    },
    {
      "epoch": 0.19594416644719514,
      "grad_norm": 1.7395036220550537,
      "learning_rate": 4.0210692652093765e-05,
      "loss": 2.0194,
      "step": 7440
    },
    {
      "epoch": 0.19597050302870686,
      "grad_norm": 1.6162654161453247,
      "learning_rate": 4.0209375823018174e-05,
      "loss": 1.6401,
      "step": 7441
    },
    {
      "epoch": 0.1959968396102186,
      "grad_norm": 1.832571029663086,
      "learning_rate": 4.020805899394259e-05,
      "loss": 2.0474,
      "step": 7442
    },
    {
      "epoch": 0.19602317619173032,
      "grad_norm": 2.5724527835845947,
      "learning_rate": 4.0206742164867e-05,
      "loss": 1.201,
      "step": 7443
    },
    {
      "epoch": 0.19604951277324204,
      "grad_norm": 2.5892958641052246,
      "learning_rate": 4.0205425335791414e-05,
      "loss": 1.9161,
      "step": 7444
    },
    {
      "epoch": 0.19607584935475375,
      "grad_norm": 2.293693780899048,
      "learning_rate": 4.020410850671583e-05,
      "loss": 0.3135,
      "step": 7445
    },
    {
      "epoch": 0.19610218593626547,
      "grad_norm": 1.8297778367996216,
      "learning_rate": 4.0202791677640245e-05,
      "loss": 1.5035,
      "step": 7446
    },
    {
      "epoch": 0.19612852251777718,
      "grad_norm": 2.4194862842559814,
      "learning_rate": 4.020147484856466e-05,
      "loss": 0.5079,
      "step": 7447
    },
    {
      "epoch": 0.19615485909928893,
      "grad_norm": 1.869683027267456,
      "learning_rate": 4.020015801948907e-05,
      "loss": 1.7675,
      "step": 7448
    },
    {
      "epoch": 0.19618119568080064,
      "grad_norm": 2.2527916431427,
      "learning_rate": 4.019884119041349e-05,
      "loss": 1.8835,
      "step": 7449
    },
    {
      "epoch": 0.19620753226231236,
      "grad_norm": 2.494058132171631,
      "learning_rate": 4.01975243613379e-05,
      "loss": 1.8796,
      "step": 7450
    },
    {
      "epoch": 0.19623386884382407,
      "grad_norm": 2.0544040203094482,
      "learning_rate": 4.0196207532262316e-05,
      "loss": 2.1759,
      "step": 7451
    },
    {
      "epoch": 0.1962602054253358,
      "grad_norm": 3.9496214389801025,
      "learning_rate": 4.0194890703186725e-05,
      "loss": 1.4214,
      "step": 7452
    },
    {
      "epoch": 0.1962865420068475,
      "grad_norm": 1.7092251777648926,
      "learning_rate": 4.019357387411114e-05,
      "loss": 0.5373,
      "step": 7453
    },
    {
      "epoch": 0.19631287858835922,
      "grad_norm": 2.4389095306396484,
      "learning_rate": 4.0192257045035556e-05,
      "loss": 1.4949,
      "step": 7454
    },
    {
      "epoch": 0.19633921516987096,
      "grad_norm": 1.9976189136505127,
      "learning_rate": 4.019094021595997e-05,
      "loss": 1.69,
      "step": 7455
    },
    {
      "epoch": 0.19636555175138268,
      "grad_norm": 1.8615301847457886,
      "learning_rate": 4.018962338688438e-05,
      "loss": 1.7446,
      "step": 7456
    },
    {
      "epoch": 0.1963918883328944,
      "grad_norm": 2.5110418796539307,
      "learning_rate": 4.0188306557808796e-05,
      "loss": 2.0329,
      "step": 7457
    },
    {
      "epoch": 0.1964182249144061,
      "grad_norm": 2.4457669258117676,
      "learning_rate": 4.018698972873321e-05,
      "loss": 1.9296,
      "step": 7458
    },
    {
      "epoch": 0.19644456149591782,
      "grad_norm": 2.5045831203460693,
      "learning_rate": 4.018567289965763e-05,
      "loss": 1.1637,
      "step": 7459
    },
    {
      "epoch": 0.19647089807742954,
      "grad_norm": 1.568500280380249,
      "learning_rate": 4.018435607058204e-05,
      "loss": 2.1627,
      "step": 7460
    },
    {
      "epoch": 0.19649723465894128,
      "grad_norm": 1.9266376495361328,
      "learning_rate": 4.018303924150645e-05,
      "loss": 2.2608,
      "step": 7461
    },
    {
      "epoch": 0.196523571240453,
      "grad_norm": 1.9290324449539185,
      "learning_rate": 4.018172241243087e-05,
      "loss": 1.8411,
      "step": 7462
    },
    {
      "epoch": 0.19654990782196471,
      "grad_norm": 4.751797676086426,
      "learning_rate": 4.018040558335528e-05,
      "loss": 2.4956,
      "step": 7463
    },
    {
      "epoch": 0.19657624440347643,
      "grad_norm": 2.8945770263671875,
      "learning_rate": 4.01790887542797e-05,
      "loss": 1.6411,
      "step": 7464
    },
    {
      "epoch": 0.19660258098498815,
      "grad_norm": 2.223255157470703,
      "learning_rate": 4.017777192520411e-05,
      "loss": 2.4477,
      "step": 7465
    },
    {
      "epoch": 0.19662891756649986,
      "grad_norm": 2.0027096271514893,
      "learning_rate": 4.017645509612852e-05,
      "loss": 1.3819,
      "step": 7466
    },
    {
      "epoch": 0.19665525414801158,
      "grad_norm": 2.9613234996795654,
      "learning_rate": 4.017513826705294e-05,
      "loss": 0.9618,
      "step": 7467
    },
    {
      "epoch": 0.19668159072952332,
      "grad_norm": 2.3709774017333984,
      "learning_rate": 4.0173821437977354e-05,
      "loss": 0.9803,
      "step": 7468
    },
    {
      "epoch": 0.19670792731103504,
      "grad_norm": 4.378414154052734,
      "learning_rate": 4.017250460890177e-05,
      "loss": 1.1307,
      "step": 7469
    },
    {
      "epoch": 0.19673426389254675,
      "grad_norm": 1.6801087856292725,
      "learning_rate": 4.017118777982618e-05,
      "loss": 1.7372,
      "step": 7470
    },
    {
      "epoch": 0.19676060047405847,
      "grad_norm": 5.7648539543151855,
      "learning_rate": 4.0169870950750594e-05,
      "loss": 1.9151,
      "step": 7471
    },
    {
      "epoch": 0.19678693705557018,
      "grad_norm": 4.914088249206543,
      "learning_rate": 4.0168554121675e-05,
      "loss": 1.2514,
      "step": 7472
    },
    {
      "epoch": 0.1968132736370819,
      "grad_norm": 2.4787395000457764,
      "learning_rate": 4.0167237292599426e-05,
      "loss": 1.3829,
      "step": 7473
    },
    {
      "epoch": 0.1968396102185936,
      "grad_norm": 3.0546462535858154,
      "learning_rate": 4.0165920463523834e-05,
      "loss": 2.0835,
      "step": 7474
    },
    {
      "epoch": 0.19686594680010536,
      "grad_norm": 1.8343952894210815,
      "learning_rate": 4.016460363444825e-05,
      "loss": 2.1522,
      "step": 7475
    },
    {
      "epoch": 0.19689228338161707,
      "grad_norm": 1.6831419467926025,
      "learning_rate": 4.0163286805372666e-05,
      "loss": 1.9958,
      "step": 7476
    },
    {
      "epoch": 0.1969186199631288,
      "grad_norm": 3.2217142581939697,
      "learning_rate": 4.0161969976297074e-05,
      "loss": 1.5006,
      "step": 7477
    },
    {
      "epoch": 0.1969449565446405,
      "grad_norm": 3.7038283348083496,
      "learning_rate": 4.01606531472215e-05,
      "loss": 0.662,
      "step": 7478
    },
    {
      "epoch": 0.19697129312615222,
      "grad_norm": 1.6220823526382446,
      "learning_rate": 4.0159336318145906e-05,
      "loss": 2.3064,
      "step": 7479
    },
    {
      "epoch": 0.19699762970766393,
      "grad_norm": 3.2559425830841064,
      "learning_rate": 4.015801948907032e-05,
      "loss": 1.7083,
      "step": 7480
    },
    {
      "epoch": 0.19702396628917568,
      "grad_norm": 3.1635656356811523,
      "learning_rate": 4.015670265999473e-05,
      "loss": 1.559,
      "step": 7481
    },
    {
      "epoch": 0.1970503028706874,
      "grad_norm": 1.976241111755371,
      "learning_rate": 4.015538583091915e-05,
      "loss": 1.5995,
      "step": 7482
    },
    {
      "epoch": 0.1970766394521991,
      "grad_norm": NaN,
      "learning_rate": 4.015538583091915e-05,
      "loss": 2.5233,
      "step": 7483
    },
    {
      "epoch": 0.19710297603371082,
      "grad_norm": 3.5263054370880127,
      "learning_rate": 4.015406900184356e-05,
      "loss": 1.7435,
      "step": 7484
    },
    {
      "epoch": 0.19712931261522254,
      "grad_norm": 5.218058109283447,
      "learning_rate": 4.015275217276798e-05,
      "loss": 0.985,
      "step": 7485
    },
    {
      "epoch": 0.19715564919673426,
      "grad_norm": 4.067135810852051,
      "learning_rate": 4.015143534369239e-05,
      "loss": 1.7492,
      "step": 7486
    },
    {
      "epoch": 0.19718198577824597,
      "grad_norm": 2.8353230953216553,
      "learning_rate": 4.01501185146168e-05,
      "loss": 0.9261,
      "step": 7487
    },
    {
      "epoch": 0.19720832235975772,
      "grad_norm": 2.1453819274902344,
      "learning_rate": 4.0148801685541224e-05,
      "loss": 1.7639,
      "step": 7488
    },
    {
      "epoch": 0.19723465894126943,
      "grad_norm": 2.8541440963745117,
      "learning_rate": 4.014748485646563e-05,
      "loss": 1.1455,
      "step": 7489
    },
    {
      "epoch": 0.19726099552278115,
      "grad_norm": 2.0490052700042725,
      "learning_rate": 4.014616802739005e-05,
      "loss": 2.527,
      "step": 7490
    },
    {
      "epoch": 0.19728733210429286,
      "grad_norm": 1.9368599653244019,
      "learning_rate": 4.014485119831446e-05,
      "loss": 2.3578,
      "step": 7491
    },
    {
      "epoch": 0.19731366868580458,
      "grad_norm": 2.8101859092712402,
      "learning_rate": 4.014353436923887e-05,
      "loss": 1.3551,
      "step": 7492
    },
    {
      "epoch": 0.1973400052673163,
      "grad_norm": 3.3241987228393555,
      "learning_rate": 4.014221754016329e-05,
      "loss": 0.9608,
      "step": 7493
    },
    {
      "epoch": 0.197366341848828,
      "grad_norm": 1.8203599452972412,
      "learning_rate": 4.0140900711087704e-05,
      "loss": 2.0779,
      "step": 7494
    },
    {
      "epoch": 0.19739267843033975,
      "grad_norm": 2.1838648319244385,
      "learning_rate": 4.013958388201212e-05,
      "loss": 1.6494,
      "step": 7495
    },
    {
      "epoch": 0.19741901501185147,
      "grad_norm": 2.1721134185791016,
      "learning_rate": 4.013826705293653e-05,
      "loss": 1.6743,
      "step": 7496
    },
    {
      "epoch": 0.19744535159336318,
      "grad_norm": 1.8995773792266846,
      "learning_rate": 4.013695022386095e-05,
      "loss": 1.5751,
      "step": 7497
    },
    {
      "epoch": 0.1974716881748749,
      "grad_norm": 2.529229164123535,
      "learning_rate": 4.013563339478536e-05,
      "loss": 2.0341,
      "step": 7498
    },
    {
      "epoch": 0.19749802475638661,
      "grad_norm": 7.55031681060791,
      "learning_rate": 4.0134316565709775e-05,
      "loss": 1.9435,
      "step": 7499
    },
    {
      "epoch": 0.19752436133789833,
      "grad_norm": 7.909158229827881,
      "learning_rate": 4.0132999736634184e-05,
      "loss": 1.9976,
      "step": 7500
    },
    {
      "epoch": 0.19755069791941007,
      "grad_norm": 2.531377077102661,
      "learning_rate": 4.01316829075586e-05,
      "loss": 1.0616,
      "step": 7501
    },
    {
      "epoch": 0.1975770345009218,
      "grad_norm": 3.3430471420288086,
      "learning_rate": 4.0130366078483015e-05,
      "loss": 1.5957,
      "step": 7502
    },
    {
      "epoch": 0.1976033710824335,
      "grad_norm": 2.2487452030181885,
      "learning_rate": 4.012904924940743e-05,
      "loss": 1.7767,
      "step": 7503
    },
    {
      "epoch": 0.19762970766394522,
      "grad_norm": 1.7168707847595215,
      "learning_rate": 4.0127732420331846e-05,
      "loss": 1.7068,
      "step": 7504
    },
    {
      "epoch": 0.19765604424545694,
      "grad_norm": 2.2454960346221924,
      "learning_rate": 4.0126415591256255e-05,
      "loss": 1.6414,
      "step": 7505
    },
    {
      "epoch": 0.19768238082696865,
      "grad_norm": 3.3308098316192627,
      "learning_rate": 4.012509876218067e-05,
      "loss": 1.9613,
      "step": 7506
    },
    {
      "epoch": 0.19770871740848037,
      "grad_norm": 1.918514370918274,
      "learning_rate": 4.0123781933105086e-05,
      "loss": 2.301,
      "step": 7507
    },
    {
      "epoch": 0.1977350539899921,
      "grad_norm": 1.9155902862548828,
      "learning_rate": 4.01224651040295e-05,
      "loss": 1.7896,
      "step": 7508
    },
    {
      "epoch": 0.19776139057150383,
      "grad_norm": 1.9534783363342285,
      "learning_rate": 4.012114827495391e-05,
      "loss": 1.7568,
      "step": 7509
    },
    {
      "epoch": 0.19778772715301554,
      "grad_norm": 1.777553915977478,
      "learning_rate": 4.0119831445878326e-05,
      "loss": 0.297,
      "step": 7510
    },
    {
      "epoch": 0.19781406373452726,
      "grad_norm": 2.4240622520446777,
      "learning_rate": 4.011851461680274e-05,
      "loss": 1.4927,
      "step": 7511
    },
    {
      "epoch": 0.19784040031603897,
      "grad_norm": 9.307279586791992,
      "learning_rate": 4.011719778772716e-05,
      "loss": 1.2356,
      "step": 7512
    },
    {
      "epoch": 0.1978667368975507,
      "grad_norm": 2.1926536560058594,
      "learning_rate": 4.0115880958651566e-05,
      "loss": 1.9209,
      "step": 7513
    },
    {
      "epoch": 0.19789307347906243,
      "grad_norm": 4.81790018081665,
      "learning_rate": 4.011456412957598e-05,
      "loss": 1.8867,
      "step": 7514
    },
    {
      "epoch": 0.19791941006057415,
      "grad_norm": 1.858687162399292,
      "learning_rate": 4.01132473005004e-05,
      "loss": 2.5035,
      "step": 7515
    },
    {
      "epoch": 0.19794574664208586,
      "grad_norm": 1.717914342880249,
      "learning_rate": 4.011193047142481e-05,
      "loss": 1.411,
      "step": 7516
    },
    {
      "epoch": 0.19797208322359758,
      "grad_norm": 1.7054836750030518,
      "learning_rate": 4.011061364234923e-05,
      "loss": 1.8476,
      "step": 7517
    },
    {
      "epoch": 0.1979984198051093,
      "grad_norm": 3.8859050273895264,
      "learning_rate": 4.010929681327364e-05,
      "loss": 0.9733,
      "step": 7518
    },
    {
      "epoch": 0.198024756386621,
      "grad_norm": 2.1310760974884033,
      "learning_rate": 4.010797998419805e-05,
      "loss": 1.5537,
      "step": 7519
    },
    {
      "epoch": 0.19805109296813272,
      "grad_norm": 4.281438827514648,
      "learning_rate": 4.010666315512246e-05,
      "loss": 1.502,
      "step": 7520
    },
    {
      "epoch": 0.19807742954964447,
      "grad_norm": 2.4652278423309326,
      "learning_rate": 4.0105346326046884e-05,
      "loss": 1.9364,
      "step": 7521
    },
    {
      "epoch": 0.19810376613115618,
      "grad_norm": 1.8965740203857422,
      "learning_rate": 4.010402949697129e-05,
      "loss": 1.3641,
      "step": 7522
    },
    {
      "epoch": 0.1981301027126679,
      "grad_norm": 4.128808498382568,
      "learning_rate": 4.010271266789571e-05,
      "loss": 1.1411,
      "step": 7523
    },
    {
      "epoch": 0.19815643929417961,
      "grad_norm": 1.7077562808990479,
      "learning_rate": 4.0101395838820124e-05,
      "loss": 1.8364,
      "step": 7524
    },
    {
      "epoch": 0.19818277587569133,
      "grad_norm": 4.297139644622803,
      "learning_rate": 4.010007900974453e-05,
      "loss": 1.7296,
      "step": 7525
    },
    {
      "epoch": 0.19820911245720305,
      "grad_norm": 2.6248910427093506,
      "learning_rate": 4.0098762180668955e-05,
      "loss": 1.7391,
      "step": 7526
    },
    {
      "epoch": 0.19823544903871476,
      "grad_norm": 1.6063523292541504,
      "learning_rate": 4.0097445351593364e-05,
      "loss": 0.7346,
      "step": 7527
    },
    {
      "epoch": 0.1982617856202265,
      "grad_norm": 2.1467947959899902,
      "learning_rate": 4.009612852251778e-05,
      "loss": 1.827,
      "step": 7528
    },
    {
      "epoch": 0.19828812220173822,
      "grad_norm": 1.945920467376709,
      "learning_rate": 4.009481169344219e-05,
      "loss": 1.9262,
      "step": 7529
    },
    {
      "epoch": 0.19831445878324994,
      "grad_norm": 1.5966516733169556,
      "learning_rate": 4.009349486436661e-05,
      "loss": 1.6245,
      "step": 7530
    },
    {
      "epoch": 0.19834079536476165,
      "grad_norm": 3.683802604675293,
      "learning_rate": 4.009217803529102e-05,
      "loss": 1.8138,
      "step": 7531
    },
    {
      "epoch": 0.19836713194627337,
      "grad_norm": 4.670269012451172,
      "learning_rate": 4.0090861206215435e-05,
      "loss": 1.1864,
      "step": 7532
    },
    {
      "epoch": 0.19839346852778508,
      "grad_norm": 2.1269919872283936,
      "learning_rate": 4.008954437713985e-05,
      "loss": 1.9501,
      "step": 7533
    },
    {
      "epoch": 0.19841980510929683,
      "grad_norm": 2.673978328704834,
      "learning_rate": 4.008822754806426e-05,
      "loss": 1.3698,
      "step": 7534
    },
    {
      "epoch": 0.19844614169080854,
      "grad_norm": 3.6344099044799805,
      "learning_rate": 4.008691071898868e-05,
      "loss": 0.2323,
      "step": 7535
    },
    {
      "epoch": 0.19847247827232026,
      "grad_norm": 3.196615219116211,
      "learning_rate": 4.008559388991309e-05,
      "loss": 1.5375,
      "step": 7536
    },
    {
      "epoch": 0.19849881485383197,
      "grad_norm": 1.5741772651672363,
      "learning_rate": 4.008427706083751e-05,
      "loss": 1.9346,
      "step": 7537
    },
    {
      "epoch": 0.1985251514353437,
      "grad_norm": 1.7590272426605225,
      "learning_rate": 4.0082960231761915e-05,
      "loss": 1.8614,
      "step": 7538
    },
    {
      "epoch": 0.1985514880168554,
      "grad_norm": 1.7377195358276367,
      "learning_rate": 4.008164340268633e-05,
      "loss": 0.7918,
      "step": 7539
    },
    {
      "epoch": 0.19857782459836712,
      "grad_norm": 3.493013858795166,
      "learning_rate": 4.008032657361075e-05,
      "loss": 1.7633,
      "step": 7540
    },
    {
      "epoch": 0.19860416117987886,
      "grad_norm": 2.4719555377960205,
      "learning_rate": 4.007900974453516e-05,
      "loss": 0.4619,
      "step": 7541
    },
    {
      "epoch": 0.19863049776139058,
      "grad_norm": 1.7563979625701904,
      "learning_rate": 4.007769291545958e-05,
      "loss": 1.9729,
      "step": 7542
    },
    {
      "epoch": 0.1986568343429023,
      "grad_norm": 5.096905708312988,
      "learning_rate": 4.007637608638399e-05,
      "loss": 0.8012,
      "step": 7543
    },
    {
      "epoch": 0.198683170924414,
      "grad_norm": 2.1098737716674805,
      "learning_rate": 4.007505925730841e-05,
      "loss": 0.9153,
      "step": 7544
    },
    {
      "epoch": 0.19870950750592573,
      "grad_norm": 1.6241297721862793,
      "learning_rate": 4.007374242823282e-05,
      "loss": 1.1565,
      "step": 7545
    },
    {
      "epoch": 0.19873584408743744,
      "grad_norm": 2.1424388885498047,
      "learning_rate": 4.0072425599157233e-05,
      "loss": 0.8252,
      "step": 7546
    },
    {
      "epoch": 0.19876218066894916,
      "grad_norm": 2.1336257457733154,
      "learning_rate": 4.007110877008164e-05,
      "loss": 1.5883,
      "step": 7547
    },
    {
      "epoch": 0.1987885172504609,
      "grad_norm": 1.9175602197647095,
      "learning_rate": 4.006979194100606e-05,
      "loss": 1.7617,
      "step": 7548
    },
    {
      "epoch": 0.19881485383197262,
      "grad_norm": 2.0442330837249756,
      "learning_rate": 4.0068475111930474e-05,
      "loss": 2.5846,
      "step": 7549
    },
    {
      "epoch": 0.19884119041348433,
      "grad_norm": 5.907866954803467,
      "learning_rate": 4.006715828285489e-05,
      "loss": 0.3052,
      "step": 7550
    },
    {
      "epoch": 0.19886752699499605,
      "grad_norm": 2.0236098766326904,
      "learning_rate": 4.0065841453779305e-05,
      "loss": 1.7809,
      "step": 7551
    },
    {
      "epoch": 0.19889386357650776,
      "grad_norm": 1.7944226264953613,
      "learning_rate": 4.0064524624703714e-05,
      "loss": 1.6861,
      "step": 7552
    },
    {
      "epoch": 0.19892020015801948,
      "grad_norm": 2.323636054992676,
      "learning_rate": 4.006320779562813e-05,
      "loss": 1.1197,
      "step": 7553
    },
    {
      "epoch": 0.19894653673953122,
      "grad_norm": 3.2054295539855957,
      "learning_rate": 4.0061890966552545e-05,
      "loss": 1.1433,
      "step": 7554
    },
    {
      "epoch": 0.19897287332104294,
      "grad_norm": 2.9711010456085205,
      "learning_rate": 4.006057413747696e-05,
      "loss": 0.8065,
      "step": 7555
    },
    {
      "epoch": 0.19899920990255465,
      "grad_norm": 1.9916502237319946,
      "learning_rate": 4.005925730840137e-05,
      "loss": 1.8929,
      "step": 7556
    },
    {
      "epoch": 0.19902554648406637,
      "grad_norm": 3.20845365524292,
      "learning_rate": 4.0057940479325785e-05,
      "loss": 0.5068,
      "step": 7557
    },
    {
      "epoch": 0.19905188306557808,
      "grad_norm": 2.848965644836426,
      "learning_rate": 4.0056623650250194e-05,
      "loss": 2.1222,
      "step": 7558
    },
    {
      "epoch": 0.1990782196470898,
      "grad_norm": 2.531967878341675,
      "learning_rate": 4.0055306821174616e-05,
      "loss": 1.8606,
      "step": 7559
    },
    {
      "epoch": 0.19910455622860151,
      "grad_norm": 2.262629270553589,
      "learning_rate": 4.0053989992099025e-05,
      "loss": 1.6576,
      "step": 7560
    },
    {
      "epoch": 0.19913089281011326,
      "grad_norm": 3.4171698093414307,
      "learning_rate": 4.005267316302344e-05,
      "loss": 1.3773,
      "step": 7561
    },
    {
      "epoch": 0.19915722939162497,
      "grad_norm": 7.297581195831299,
      "learning_rate": 4.0051356333947856e-05,
      "loss": 1.4364,
      "step": 7562
    },
    {
      "epoch": 0.1991835659731367,
      "grad_norm": 2.33355975151062,
      "learning_rate": 4.005003950487227e-05,
      "loss": 1.3524,
      "step": 7563
    },
    {
      "epoch": 0.1992099025546484,
      "grad_norm": 3.3187313079833984,
      "learning_rate": 4.004872267579669e-05,
      "loss": 1.369,
      "step": 7564
    },
    {
      "epoch": 0.19923623913616012,
      "grad_norm": 3.691459894180298,
      "learning_rate": 4.0047405846721096e-05,
      "loss": 1.2118,
      "step": 7565
    },
    {
      "epoch": 0.19926257571767184,
      "grad_norm": 1.7060962915420532,
      "learning_rate": 4.004608901764551e-05,
      "loss": 1.9935,
      "step": 7566
    },
    {
      "epoch": 0.19928891229918358,
      "grad_norm": 6.282314300537109,
      "learning_rate": 4.004477218856992e-05,
      "loss": 1.4531,
      "step": 7567
    },
    {
      "epoch": 0.1993152488806953,
      "grad_norm": 2.377469778060913,
      "learning_rate": 4.004345535949434e-05,
      "loss": 2.0445,
      "step": 7568
    },
    {
      "epoch": 0.199341585462207,
      "grad_norm": 2.4734559059143066,
      "learning_rate": 4.004213853041875e-05,
      "loss": 1.8534,
      "step": 7569
    },
    {
      "epoch": 0.19936792204371873,
      "grad_norm": 2.94069504737854,
      "learning_rate": 4.004082170134317e-05,
      "loss": 2.2641,
      "step": 7570
    },
    {
      "epoch": 0.19939425862523044,
      "grad_norm": 2.2741403579711914,
      "learning_rate": 4.003950487226758e-05,
      "loss": 2.3334,
      "step": 7571
    },
    {
      "epoch": 0.19942059520674216,
      "grad_norm": 2.0813348293304443,
      "learning_rate": 4.003818804319199e-05,
      "loss": 1.7502,
      "step": 7572
    },
    {
      "epoch": 0.19944693178825387,
      "grad_norm": 1.5913808345794678,
      "learning_rate": 4.0036871214116414e-05,
      "loss": 1.7071,
      "step": 7573
    },
    {
      "epoch": 0.19947326836976562,
      "grad_norm": 2.7190074920654297,
      "learning_rate": 4.003555438504082e-05,
      "loss": 2.4468,
      "step": 7574
    },
    {
      "epoch": 0.19949960495127733,
      "grad_norm": 1.7631863355636597,
      "learning_rate": 4.003423755596524e-05,
      "loss": 1.2718,
      "step": 7575
    },
    {
      "epoch": 0.19952594153278905,
      "grad_norm": 1.6568305492401123,
      "learning_rate": 4.003292072688965e-05,
      "loss": 1.812,
      "step": 7576
    },
    {
      "epoch": 0.19955227811430076,
      "grad_norm": 2.389723777770996,
      "learning_rate": 4.003160389781407e-05,
      "loss": 1.7941,
      "step": 7577
    },
    {
      "epoch": 0.19957861469581248,
      "grad_norm": 3.206573963165283,
      "learning_rate": 4.003028706873848e-05,
      "loss": 1.3594,
      "step": 7578
    },
    {
      "epoch": 0.1996049512773242,
      "grad_norm": 2.3536691665649414,
      "learning_rate": 4.0028970239662894e-05,
      "loss": 1.8424,
      "step": 7579
    },
    {
      "epoch": 0.1996312878588359,
      "grad_norm": 2.3670692443847656,
      "learning_rate": 4.002765341058731e-05,
      "loss": 1.9065,
      "step": 7580
    },
    {
      "epoch": 0.19965762444034765,
      "grad_norm": 3.001194953918457,
      "learning_rate": 4.002633658151172e-05,
      "loss": 1.7191,
      "step": 7581
    },
    {
      "epoch": 0.19968396102185937,
      "grad_norm": 2.1872239112854004,
      "learning_rate": 4.002501975243614e-05,
      "loss": 2.6511,
      "step": 7582
    },
    {
      "epoch": 0.19971029760337108,
      "grad_norm": 1.8038225173950195,
      "learning_rate": 4.002370292336055e-05,
      "loss": 2.0818,
      "step": 7583
    },
    {
      "epoch": 0.1997366341848828,
      "grad_norm": 2.3791491985321045,
      "learning_rate": 4.0022386094284965e-05,
      "loss": 1.5658,
      "step": 7584
    },
    {
      "epoch": 0.19976297076639452,
      "grad_norm": 2.512356996536255,
      "learning_rate": 4.0021069265209374e-05,
      "loss": 0.212,
      "step": 7585
    },
    {
      "epoch": 0.19978930734790623,
      "grad_norm": 2.272510051727295,
      "learning_rate": 4.001975243613379e-05,
      "loss": 1.697,
      "step": 7586
    },
    {
      "epoch": 0.19981564392941797,
      "grad_norm": 1.9434016942977905,
      "learning_rate": 4.0018435607058205e-05,
      "loss": 2.2373,
      "step": 7587
    },
    {
      "epoch": 0.1998419805109297,
      "grad_norm": 1.732528567314148,
      "learning_rate": 4.001711877798262e-05,
      "loss": 1.8638,
      "step": 7588
    },
    {
      "epoch": 0.1998683170924414,
      "grad_norm": 3.1505141258239746,
      "learning_rate": 4.0015801948907036e-05,
      "loss": 2.2199,
      "step": 7589
    },
    {
      "epoch": 0.19989465367395312,
      "grad_norm": 1.6942317485809326,
      "learning_rate": 4.0014485119831445e-05,
      "loss": 1.828,
      "step": 7590
    },
    {
      "epoch": 0.19992099025546484,
      "grad_norm": 1.985947847366333,
      "learning_rate": 4.001316829075586e-05,
      "loss": 1.9926,
      "step": 7591
    },
    {
      "epoch": 0.19994732683697655,
      "grad_norm": 2.9298174381256104,
      "learning_rate": 4.0011851461680276e-05,
      "loss": 1.153,
      "step": 7592
    },
    {
      "epoch": 0.19997366341848827,
      "grad_norm": 1.9928040504455566,
      "learning_rate": 4.001053463260469e-05,
      "loss": 1.9263,
      "step": 7593
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.906722068786621,
      "learning_rate": 4.00092178035291e-05,
      "loss": 2.216,
      "step": 7594
    },
    {
      "epoch": 0.20002633658151173,
      "grad_norm": 2.406498432159424,
      "learning_rate": 4.0007900974453516e-05,
      "loss": 1.336,
      "step": 7595
    },
    {
      "epoch": 0.20005267316302344,
      "grad_norm": 4.151043891906738,
      "learning_rate": 4.000658414537793e-05,
      "loss": 0.9129,
      "step": 7596
    },
    {
      "epoch": 0.20007900974453516,
      "grad_norm": 2.00247859954834,
      "learning_rate": 4.000526731630235e-05,
      "loss": 1.4707,
      "step": 7597
    },
    {
      "epoch": 0.20010534632604687,
      "grad_norm": 3.530992031097412,
      "learning_rate": 4.000395048722676e-05,
      "loss": 0.7884,
      "step": 7598
    },
    {
      "epoch": 0.2001316829075586,
      "grad_norm": 1.859633445739746,
      "learning_rate": 4.000263365815117e-05,
      "loss": 1.3441,
      "step": 7599
    },
    {
      "epoch": 0.20015801948907033,
      "grad_norm": 2.1075634956359863,
      "learning_rate": 4.000131682907559e-05,
      "loss": 1.915,
      "step": 7600
    },
    {
      "epoch": 0.20018435607058205,
      "grad_norm": 1.7875258922576904,
      "learning_rate": 4e-05,
      "loss": 2.1025,
      "step": 7601
    },
    {
      "epoch": 0.20021069265209376,
      "grad_norm": 2.694178581237793,
      "learning_rate": 3.999868317092442e-05,
      "loss": 2.1537,
      "step": 7602
    },
    {
      "epoch": 0.20023702923360548,
      "grad_norm": 1.946129322052002,
      "learning_rate": 3.999736634184883e-05,
      "loss": 1.8532,
      "step": 7603
    },
    {
      "epoch": 0.2002633658151172,
      "grad_norm": 2.714689254760742,
      "learning_rate": 3.999604951277324e-05,
      "loss": 1.7262,
      "step": 7604
    },
    {
      "epoch": 0.2002897023966289,
      "grad_norm": 2.034921169281006,
      "learning_rate": 3.999473268369765e-05,
      "loss": 2.065,
      "step": 7605
    },
    {
      "epoch": 0.20031603897814063,
      "grad_norm": 2.1698575019836426,
      "learning_rate": 3.9993415854622074e-05,
      "loss": 1.7837,
      "step": 7606
    },
    {
      "epoch": 0.20034237555965237,
      "grad_norm": 1.7990353107452393,
      "learning_rate": 3.999209902554648e-05,
      "loss": 1.7209,
      "step": 7607
    },
    {
      "epoch": 0.20036871214116408,
      "grad_norm": 3.398479461669922,
      "learning_rate": 3.99907821964709e-05,
      "loss": 1.4449,
      "step": 7608
    },
    {
      "epoch": 0.2003950487226758,
      "grad_norm": 1.9705686569213867,
      "learning_rate": 3.9989465367395315e-05,
      "loss": 1.5597,
      "step": 7609
    },
    {
      "epoch": 0.20042138530418752,
      "grad_norm": 5.314675807952881,
      "learning_rate": 3.998814853831973e-05,
      "loss": 1.1518,
      "step": 7610
    },
    {
      "epoch": 0.20044772188569923,
      "grad_norm": 4.319832801818848,
      "learning_rate": 3.9986831709244146e-05,
      "loss": 2.132,
      "step": 7611
    },
    {
      "epoch": 0.20047405846721095,
      "grad_norm": 1.8207014799118042,
      "learning_rate": 3.9985514880168555e-05,
      "loss": 1.807,
      "step": 7612
    },
    {
      "epoch": 0.20050039504872266,
      "grad_norm": 2.74749493598938,
      "learning_rate": 3.998419805109297e-05,
      "loss": 2.2016,
      "step": 7613
    },
    {
      "epoch": 0.2005267316302344,
      "grad_norm": 2.00268816947937,
      "learning_rate": 3.998288122201738e-05,
      "loss": 1.9807,
      "step": 7614
    },
    {
      "epoch": 0.20055306821174612,
      "grad_norm": 1.9017468690872192,
      "learning_rate": 3.99815643929418e-05,
      "loss": 1.917,
      "step": 7615
    },
    {
      "epoch": 0.20057940479325784,
      "grad_norm": 1.6493947505950928,
      "learning_rate": 3.998024756386621e-05,
      "loss": 1.3457,
      "step": 7616
    },
    {
      "epoch": 0.20060574137476955,
      "grad_norm": 1.6002165079116821,
      "learning_rate": 3.9978930734790626e-05,
      "loss": 2.2266,
      "step": 7617
    },
    {
      "epoch": 0.20063207795628127,
      "grad_norm": 3.303194522857666,
      "learning_rate": 3.997761390571504e-05,
      "loss": 2.2041,
      "step": 7618
    },
    {
      "epoch": 0.20065841453779298,
      "grad_norm": 3.5344648361206055,
      "learning_rate": 3.997629707663945e-05,
      "loss": 1.554,
      "step": 7619
    },
    {
      "epoch": 0.20068475111930473,
      "grad_norm": 4.115609169006348,
      "learning_rate": 3.997498024756387e-05,
      "loss": 2.4147,
      "step": 7620
    },
    {
      "epoch": 0.20071108770081644,
      "grad_norm": 2.985434055328369,
      "learning_rate": 3.997366341848828e-05,
      "loss": 0.844,
      "step": 7621
    },
    {
      "epoch": 0.20073742428232816,
      "grad_norm": 1.9391499757766724,
      "learning_rate": 3.99723465894127e-05,
      "loss": 1.9651,
      "step": 7622
    },
    {
      "epoch": 0.20076376086383987,
      "grad_norm": 3.0145184993743896,
      "learning_rate": 3.9971029760337106e-05,
      "loss": 1.1077,
      "step": 7623
    },
    {
      "epoch": 0.2007900974453516,
      "grad_norm": 2.0803065299987793,
      "learning_rate": 3.996971293126152e-05,
      "loss": 2.4723,
      "step": 7624
    },
    {
      "epoch": 0.2008164340268633,
      "grad_norm": 3.519578456878662,
      "learning_rate": 3.996839610218594e-05,
      "loss": 1.2927,
      "step": 7625
    },
    {
      "epoch": 0.20084277060837502,
      "grad_norm": 2.8513505458831787,
      "learning_rate": 3.996707927311035e-05,
      "loss": 0.4603,
      "step": 7626
    },
    {
      "epoch": 0.20086910718988676,
      "grad_norm": 1.9014256000518799,
      "learning_rate": 3.996576244403477e-05,
      "loss": 1.816,
      "step": 7627
    },
    {
      "epoch": 0.20089544377139848,
      "grad_norm": 1.4538400173187256,
      "learning_rate": 3.996444561495918e-05,
      "loss": 1.4964,
      "step": 7628
    },
    {
      "epoch": 0.2009217803529102,
      "grad_norm": 1.9181309938430786,
      "learning_rate": 3.99631287858836e-05,
      "loss": 1.6319,
      "step": 7629
    },
    {
      "epoch": 0.2009481169344219,
      "grad_norm": 2.9189956188201904,
      "learning_rate": 3.996181195680801e-05,
      "loss": 2.1589,
      "step": 7630
    },
    {
      "epoch": 0.20097445351593363,
      "grad_norm": 2.1787703037261963,
      "learning_rate": 3.9960495127732424e-05,
      "loss": 1.8131,
      "step": 7631
    },
    {
      "epoch": 0.20100079009744534,
      "grad_norm": 3.29821515083313,
      "learning_rate": 3.995917829865683e-05,
      "loss": 1.8631,
      "step": 7632
    },
    {
      "epoch": 0.20102712667895706,
      "grad_norm": 2.814399480819702,
      "learning_rate": 3.995786146958125e-05,
      "loss": 0.7122,
      "step": 7633
    },
    {
      "epoch": 0.2010534632604688,
      "grad_norm": 2.0861523151397705,
      "learning_rate": 3.9956544640505664e-05,
      "loss": 2.0949,
      "step": 7634
    },
    {
      "epoch": 0.20107979984198052,
      "grad_norm": 2.0584568977355957,
      "learning_rate": 3.995522781143008e-05,
      "loss": 1.3592,
      "step": 7635
    },
    {
      "epoch": 0.20110613642349223,
      "grad_norm": 2.8228025436401367,
      "learning_rate": 3.9953910982354495e-05,
      "loss": 1.5879,
      "step": 7636
    },
    {
      "epoch": 0.20113247300500395,
      "grad_norm": 3.7035892009735107,
      "learning_rate": 3.9952594153278904e-05,
      "loss": 1.6861,
      "step": 7637
    },
    {
      "epoch": 0.20115880958651566,
      "grad_norm": 2.04569149017334,
      "learning_rate": 3.995127732420332e-05,
      "loss": 0.3637,
      "step": 7638
    },
    {
      "epoch": 0.20118514616802738,
      "grad_norm": 2.5820789337158203,
      "learning_rate": 3.9949960495127735e-05,
      "loss": 1.9954,
      "step": 7639
    },
    {
      "epoch": 0.20121148274953912,
      "grad_norm": 2.9092183113098145,
      "learning_rate": 3.994864366605215e-05,
      "loss": 2.0456,
      "step": 7640
    },
    {
      "epoch": 0.20123781933105084,
      "grad_norm": 2.382683038711548,
      "learning_rate": 3.994732683697656e-05,
      "loss": 0.9198,
      "step": 7641
    },
    {
      "epoch": 0.20126415591256255,
      "grad_norm": 2.3362839221954346,
      "learning_rate": 3.9946010007900975e-05,
      "loss": 2.4442,
      "step": 7642
    },
    {
      "epoch": 0.20129049249407427,
      "grad_norm": 2.193953275680542,
      "learning_rate": 3.994469317882539e-05,
      "loss": 0.476,
      "step": 7643
    },
    {
      "epoch": 0.20131682907558598,
      "grad_norm": 2.4420206546783447,
      "learning_rate": 3.9943376349749806e-05,
      "loss": 1.7504,
      "step": 7644
    },
    {
      "epoch": 0.2013431656570977,
      "grad_norm": 2.1141293048858643,
      "learning_rate": 3.994205952067422e-05,
      "loss": 1.151,
      "step": 7645
    },
    {
      "epoch": 0.20136950223860942,
      "grad_norm": 5.707983016967773,
      "learning_rate": 3.994074269159863e-05,
      "loss": 1.8366,
      "step": 7646
    },
    {
      "epoch": 0.20139583882012116,
      "grad_norm": 3.67509126663208,
      "learning_rate": 3.9939425862523046e-05,
      "loss": 1.8506,
      "step": 7647
    },
    {
      "epoch": 0.20142217540163287,
      "grad_norm": 2.4796736240386963,
      "learning_rate": 3.993810903344746e-05,
      "loss": 1.7999,
      "step": 7648
    },
    {
      "epoch": 0.2014485119831446,
      "grad_norm": 1.6437618732452393,
      "learning_rate": 3.993679220437188e-05,
      "loss": 2.5569,
      "step": 7649
    },
    {
      "epoch": 0.2014748485646563,
      "grad_norm": 2.386502265930176,
      "learning_rate": 3.9935475375296286e-05,
      "loss": 1.4422,
      "step": 7650
    },
    {
      "epoch": 0.20150118514616802,
      "grad_norm": 1.9960157871246338,
      "learning_rate": 3.99341585462207e-05,
      "loss": 2.1284,
      "step": 7651
    },
    {
      "epoch": 0.20152752172767974,
      "grad_norm": 2.392915725708008,
      "learning_rate": 3.993284171714512e-05,
      "loss": 1.9481,
      "step": 7652
    },
    {
      "epoch": 0.20155385830919148,
      "grad_norm": 2.5523152351379395,
      "learning_rate": 3.993152488806953e-05,
      "loss": 2.005,
      "step": 7653
    },
    {
      "epoch": 0.2015801948907032,
      "grad_norm": 1.7755165100097656,
      "learning_rate": 3.993020805899395e-05,
      "loss": 2.0937,
      "step": 7654
    },
    {
      "epoch": 0.2016065314722149,
      "grad_norm": 1.5651800632476807,
      "learning_rate": 3.992889122991836e-05,
      "loss": 2.0953,
      "step": 7655
    },
    {
      "epoch": 0.20163286805372663,
      "grad_norm": 1.7528629302978516,
      "learning_rate": 3.992757440084277e-05,
      "loss": 2.5557,
      "step": 7656
    },
    {
      "epoch": 0.20165920463523834,
      "grad_norm": 2.3901150226593018,
      "learning_rate": 3.992625757176718e-05,
      "loss": 2.2694,
      "step": 7657
    },
    {
      "epoch": 0.20168554121675006,
      "grad_norm": 2.3250856399536133,
      "learning_rate": 3.9924940742691604e-05,
      "loss": 1.598,
      "step": 7658
    },
    {
      "epoch": 0.20171187779826177,
      "grad_norm": 1.883156657218933,
      "learning_rate": 3.992362391361601e-05,
      "loss": 1.973,
      "step": 7659
    },
    {
      "epoch": 0.20173821437977352,
      "grad_norm": 1.984703540802002,
      "learning_rate": 3.992230708454043e-05,
      "loss": 1.3489,
      "step": 7660
    },
    {
      "epoch": 0.20176455096128523,
      "grad_norm": 4.213223457336426,
      "learning_rate": 3.992099025546484e-05,
      "loss": 1.4312,
      "step": 7661
    },
    {
      "epoch": 0.20179088754279695,
      "grad_norm": 1.8153281211853027,
      "learning_rate": 3.991967342638926e-05,
      "loss": 1.92,
      "step": 7662
    },
    {
      "epoch": 0.20181722412430866,
      "grad_norm": 2.0963289737701416,
      "learning_rate": 3.991835659731367e-05,
      "loss": 1.8755,
      "step": 7663
    },
    {
      "epoch": 0.20184356070582038,
      "grad_norm": 2.7666232585906982,
      "learning_rate": 3.9917039768238084e-05,
      "loss": 1.8736,
      "step": 7664
    },
    {
      "epoch": 0.2018698972873321,
      "grad_norm": 4.68382453918457,
      "learning_rate": 3.99157229391625e-05,
      "loss": 1.4773,
      "step": 7665
    },
    {
      "epoch": 0.2018962338688438,
      "grad_norm": 2.283851146697998,
      "learning_rate": 3.991440611008691e-05,
      "loss": 1.8943,
      "step": 7666
    },
    {
      "epoch": 0.20192257045035555,
      "grad_norm": 2.262950897216797,
      "learning_rate": 3.991308928101133e-05,
      "loss": 1.973,
      "step": 7667
    },
    {
      "epoch": 0.20194890703186727,
      "grad_norm": 1.9184603691101074,
      "learning_rate": 3.991177245193574e-05,
      "loss": 1.4776,
      "step": 7668
    },
    {
      "epoch": 0.20197524361337899,
      "grad_norm": 1.4635335206985474,
      "learning_rate": 3.9910455622860155e-05,
      "loss": 1.4053,
      "step": 7669
    },
    {
      "epoch": 0.2020015801948907,
      "grad_norm": 1.6465815305709839,
      "learning_rate": 3.9909138793784564e-05,
      "loss": 1.8378,
      "step": 7670
    },
    {
      "epoch": 0.20202791677640242,
      "grad_norm": 1.9513689279556274,
      "learning_rate": 3.990782196470898e-05,
      "loss": 1.1274,
      "step": 7671
    },
    {
      "epoch": 0.20205425335791413,
      "grad_norm": 4.086761474609375,
      "learning_rate": 3.9906505135633396e-05,
      "loss": 1.8083,
      "step": 7672
    },
    {
      "epoch": 0.20208058993942588,
      "grad_norm": 1.6377832889556885,
      "learning_rate": 3.990518830655781e-05,
      "loss": 1.4912,
      "step": 7673
    },
    {
      "epoch": 0.2021069265209376,
      "grad_norm": 1.731050729751587,
      "learning_rate": 3.990387147748223e-05,
      "loss": 1.8024,
      "step": 7674
    },
    {
      "epoch": 0.2021332631024493,
      "grad_norm": 3.2154223918914795,
      "learning_rate": 3.9902554648406636e-05,
      "loss": 1.5255,
      "step": 7675
    },
    {
      "epoch": 0.20215959968396102,
      "grad_norm": 2.671917200088501,
      "learning_rate": 3.990123781933106e-05,
      "loss": 1.1043,
      "step": 7676
    },
    {
      "epoch": 0.20218593626547274,
      "grad_norm": 3.1091525554656982,
      "learning_rate": 3.989992099025547e-05,
      "loss": 1.2368,
      "step": 7677
    },
    {
      "epoch": 0.20221227284698445,
      "grad_norm": 1.976817011833191,
      "learning_rate": 3.989860416117988e-05,
      "loss": 1.6252,
      "step": 7678
    },
    {
      "epoch": 0.20223860942849617,
      "grad_norm": 1.9005625247955322,
      "learning_rate": 3.989728733210429e-05,
      "loss": 1.9913,
      "step": 7679
    },
    {
      "epoch": 0.2022649460100079,
      "grad_norm": 2.9008219242095947,
      "learning_rate": 3.989597050302871e-05,
      "loss": 2.0195,
      "step": 7680
    },
    {
      "epoch": 0.20229128259151963,
      "grad_norm": 2.651066541671753,
      "learning_rate": 3.989465367395312e-05,
      "loss": 1.1277,
      "step": 7681
    },
    {
      "epoch": 0.20231761917303134,
      "grad_norm": 2.88639760017395,
      "learning_rate": 3.989333684487754e-05,
      "loss": 1.2632,
      "step": 7682
    },
    {
      "epoch": 0.20234395575454306,
      "grad_norm": 1.7234835624694824,
      "learning_rate": 3.9892020015801954e-05,
      "loss": 1.3508,
      "step": 7683
    },
    {
      "epoch": 0.20237029233605477,
      "grad_norm": 3.25190806388855,
      "learning_rate": 3.989070318672636e-05,
      "loss": 1.5945,
      "step": 7684
    },
    {
      "epoch": 0.2023966289175665,
      "grad_norm": 3.527798652648926,
      "learning_rate": 3.988938635765078e-05,
      "loss": 0.7888,
      "step": 7685
    },
    {
      "epoch": 0.20242296549907823,
      "grad_norm": 2.441800117492676,
      "learning_rate": 3.9888069528575194e-05,
      "loss": 0.9202,
      "step": 7686
    },
    {
      "epoch": 0.20244930208058995,
      "grad_norm": 2.2201015949249268,
      "learning_rate": 3.988675269949961e-05,
      "loss": 1.5023,
      "step": 7687
    },
    {
      "epoch": 0.20247563866210166,
      "grad_norm": 2.8298521041870117,
      "learning_rate": 3.988543587042402e-05,
      "loss": 2.0057,
      "step": 7688
    },
    {
      "epoch": 0.20250197524361338,
      "grad_norm": 4.628076076507568,
      "learning_rate": 3.9884119041348434e-05,
      "loss": 1.1235,
      "step": 7689
    },
    {
      "epoch": 0.2025283118251251,
      "grad_norm": 1.9111008644104004,
      "learning_rate": 3.988280221227285e-05,
      "loss": 1.3718,
      "step": 7690
    },
    {
      "epoch": 0.2025546484066368,
      "grad_norm": 1.8705377578735352,
      "learning_rate": 3.9881485383197265e-05,
      "loss": 1.6203,
      "step": 7691
    },
    {
      "epoch": 0.20258098498814853,
      "grad_norm": 1.9395408630371094,
      "learning_rate": 3.988016855412168e-05,
      "loss": 1.9583,
      "step": 7692
    },
    {
      "epoch": 0.20260732156966027,
      "grad_norm": 1.6386895179748535,
      "learning_rate": 3.987885172504609e-05,
      "loss": 1.5762,
      "step": 7693
    },
    {
      "epoch": 0.20263365815117199,
      "grad_norm": 1.8084867000579834,
      "learning_rate": 3.9877534895970505e-05,
      "loss": 1.6403,
      "step": 7694
    },
    {
      "epoch": 0.2026599947326837,
      "grad_norm": 1.8858562707901,
      "learning_rate": 3.987621806689492e-05,
      "loss": 1.6637,
      "step": 7695
    },
    {
      "epoch": 0.20268633131419542,
      "grad_norm": 2.83836030960083,
      "learning_rate": 3.9874901237819336e-05,
      "loss": 2.3678,
      "step": 7696
    },
    {
      "epoch": 0.20271266789570713,
      "grad_norm": 3.136502504348755,
      "learning_rate": 3.9873584408743745e-05,
      "loss": 1.5297,
      "step": 7697
    },
    {
      "epoch": 0.20273900447721885,
      "grad_norm": 2.024453639984131,
      "learning_rate": 3.987226757966816e-05,
      "loss": 1.895,
      "step": 7698
    },
    {
      "epoch": 0.20276534105873056,
      "grad_norm": 2.590121030807495,
      "learning_rate": 3.9870950750592576e-05,
      "loss": 1.9034,
      "step": 7699
    },
    {
      "epoch": 0.2027916776402423,
      "grad_norm": 1.5360984802246094,
      "learning_rate": 3.986963392151699e-05,
      "loss": 1.6613,
      "step": 7700
    },
    {
      "epoch": 0.20281801422175402,
      "grad_norm": 1.9278558492660522,
      "learning_rate": 3.986831709244141e-05,
      "loss": 1.6669,
      "step": 7701
    },
    {
      "epoch": 0.20284435080326574,
      "grad_norm": 2.4258224964141846,
      "learning_rate": 3.9867000263365816e-05,
      "loss": 1.9488,
      "step": 7702
    },
    {
      "epoch": 0.20287068738477745,
      "grad_norm": 2.2309935092926025,
      "learning_rate": 3.986568343429023e-05,
      "loss": 0.4837,
      "step": 7703
    },
    {
      "epoch": 0.20289702396628917,
      "grad_norm": 4.881872177124023,
      "learning_rate": 3.986436660521464e-05,
      "loss": 1.7573,
      "step": 7704
    },
    {
      "epoch": 0.20292336054780089,
      "grad_norm": 3.214388132095337,
      "learning_rate": 3.986304977613906e-05,
      "loss": 2.0867,
      "step": 7705
    },
    {
      "epoch": 0.20294969712931263,
      "grad_norm": 1.8366135358810425,
      "learning_rate": 3.986173294706347e-05,
      "loss": 1.807,
      "step": 7706
    },
    {
      "epoch": 0.20297603371082434,
      "grad_norm": 1.7630726099014282,
      "learning_rate": 3.986041611798789e-05,
      "loss": 1.189,
      "step": 7707
    },
    {
      "epoch": 0.20300237029233606,
      "grad_norm": 2.373157024383545,
      "learning_rate": 3.9859099288912296e-05,
      "loss": 1.0657,
      "step": 7708
    },
    {
      "epoch": 0.20302870687384778,
      "grad_norm": 2.849299430847168,
      "learning_rate": 3.985778245983672e-05,
      "loss": 1.8557,
      "step": 7709
    },
    {
      "epoch": 0.2030550434553595,
      "grad_norm": 1.7904765605926514,
      "learning_rate": 3.985646563076113e-05,
      "loss": 1.7583,
      "step": 7710
    },
    {
      "epoch": 0.2030813800368712,
      "grad_norm": 3.3243374824523926,
      "learning_rate": 3.985514880168554e-05,
      "loss": 1.4315,
      "step": 7711
    },
    {
      "epoch": 0.20310771661838292,
      "grad_norm": 2.1492202281951904,
      "learning_rate": 3.985383197260996e-05,
      "loss": 1.9562,
      "step": 7712
    },
    {
      "epoch": 0.20313405319989467,
      "grad_norm": 1.5599113702774048,
      "learning_rate": 3.985251514353437e-05,
      "loss": 1.7551,
      "step": 7713
    },
    {
      "epoch": 0.20316038978140638,
      "grad_norm": 1.7659612894058228,
      "learning_rate": 3.985119831445879e-05,
      "loss": 2.1388,
      "step": 7714
    },
    {
      "epoch": 0.2031867263629181,
      "grad_norm": 1.6247694492340088,
      "learning_rate": 3.98498814853832e-05,
      "loss": 1.5643,
      "step": 7715
    },
    {
      "epoch": 0.2032130629444298,
      "grad_norm": 3.0842278003692627,
      "learning_rate": 3.9848564656307614e-05,
      "loss": 0.759,
      "step": 7716
    },
    {
      "epoch": 0.20323939952594153,
      "grad_norm": 2.41835355758667,
      "learning_rate": 3.984724782723202e-05,
      "loss": 1.8493,
      "step": 7717
    },
    {
      "epoch": 0.20326573610745324,
      "grad_norm": 3.290679454803467,
      "learning_rate": 3.984593099815644e-05,
      "loss": 1.9082,
      "step": 7718
    },
    {
      "epoch": 0.20329207268896496,
      "grad_norm": 4.662559986114502,
      "learning_rate": 3.9844614169080854e-05,
      "loss": 1.4023,
      "step": 7719
    },
    {
      "epoch": 0.2033184092704767,
      "grad_norm": 3.381883144378662,
      "learning_rate": 3.984329734000527e-05,
      "loss": 1.71,
      "step": 7720
    },
    {
      "epoch": 0.20334474585198842,
      "grad_norm": 1.6710078716278076,
      "learning_rate": 3.9841980510929685e-05,
      "loss": 1.7233,
      "step": 7721
    },
    {
      "epoch": 0.20337108243350013,
      "grad_norm": 2.7248570919036865,
      "learning_rate": 3.9840663681854094e-05,
      "loss": 2.2979,
      "step": 7722
    },
    {
      "epoch": 0.20339741901501185,
      "grad_norm": 1.873510479927063,
      "learning_rate": 3.983934685277851e-05,
      "loss": 1.6392,
      "step": 7723
    },
    {
      "epoch": 0.20342375559652356,
      "grad_norm": 2.1086153984069824,
      "learning_rate": 3.9838030023702925e-05,
      "loss": 1.5673,
      "step": 7724
    },
    {
      "epoch": 0.20345009217803528,
      "grad_norm": 4.5790581703186035,
      "learning_rate": 3.983671319462734e-05,
      "loss": 1.4436,
      "step": 7725
    },
    {
      "epoch": 0.20347642875954702,
      "grad_norm": 1.9357166290283203,
      "learning_rate": 3.983539636555175e-05,
      "loss": 1.6495,
      "step": 7726
    },
    {
      "epoch": 0.20350276534105874,
      "grad_norm": 1.9797827005386353,
      "learning_rate": 3.9834079536476165e-05,
      "loss": 1.9115,
      "step": 7727
    },
    {
      "epoch": 0.20352910192257045,
      "grad_norm": 3.078951120376587,
      "learning_rate": 3.983276270740058e-05,
      "loss": 2.25,
      "step": 7728
    },
    {
      "epoch": 0.20355543850408217,
      "grad_norm": 2.1000711917877197,
      "learning_rate": 3.9831445878324996e-05,
      "loss": 1.2215,
      "step": 7729
    },
    {
      "epoch": 0.20358177508559389,
      "grad_norm": 2.479434013366699,
      "learning_rate": 3.983012904924941e-05,
      "loss": 1.5245,
      "step": 7730
    },
    {
      "epoch": 0.2036081116671056,
      "grad_norm": 3.928635835647583,
      "learning_rate": 3.982881222017382e-05,
      "loss": 1.3264,
      "step": 7731
    },
    {
      "epoch": 0.20363444824861732,
      "grad_norm": 2.1148130893707275,
      "learning_rate": 3.9827495391098237e-05,
      "loss": 1.7587,
      "step": 7732
    },
    {
      "epoch": 0.20366078483012906,
      "grad_norm": 2.1748244762420654,
      "learning_rate": 3.982617856202265e-05,
      "loss": 2.2599,
      "step": 7733
    },
    {
      "epoch": 0.20368712141164078,
      "grad_norm": 4.344339847564697,
      "learning_rate": 3.982486173294707e-05,
      "loss": 0.4548,
      "step": 7734
    },
    {
      "epoch": 0.2037134579931525,
      "grad_norm": 1.735168218612671,
      "learning_rate": 3.9823544903871477e-05,
      "loss": 2.2722,
      "step": 7735
    },
    {
      "epoch": 0.2037397945746642,
      "grad_norm": 1.856628179550171,
      "learning_rate": 3.982222807479589e-05,
      "loss": 2.2704,
      "step": 7736
    },
    {
      "epoch": 0.20376613115617592,
      "grad_norm": 1.8732333183288574,
      "learning_rate": 3.982091124572031e-05,
      "loss": 2.3249,
      "step": 7737
    },
    {
      "epoch": 0.20379246773768764,
      "grad_norm": 3.1309595108032227,
      "learning_rate": 3.981959441664472e-05,
      "loss": 1.7805,
      "step": 7738
    },
    {
      "epoch": 0.20381880431919938,
      "grad_norm": 3.8597004413604736,
      "learning_rate": 3.981827758756914e-05,
      "loss": 1.185,
      "step": 7739
    },
    {
      "epoch": 0.2038451409007111,
      "grad_norm": 1.5584781169891357,
      "learning_rate": 3.981696075849355e-05,
      "loss": 1.5417,
      "step": 7740
    },
    {
      "epoch": 0.2038714774822228,
      "grad_norm": 2.627574920654297,
      "learning_rate": 3.981564392941796e-05,
      "loss": 0.9549,
      "step": 7741
    },
    {
      "epoch": 0.20389781406373453,
      "grad_norm": 1.9498339891433716,
      "learning_rate": 3.981432710034238e-05,
      "loss": 1.8448,
      "step": 7742
    },
    {
      "epoch": 0.20392415064524624,
      "grad_norm": 2.078964948654175,
      "learning_rate": 3.9813010271266795e-05,
      "loss": 0.96,
      "step": 7743
    },
    {
      "epoch": 0.20395048722675796,
      "grad_norm": 2.1001758575439453,
      "learning_rate": 3.98116934421912e-05,
      "loss": 1.2231,
      "step": 7744
    },
    {
      "epoch": 0.20397682380826967,
      "grad_norm": 1.8574081659317017,
      "learning_rate": 3.981037661311562e-05,
      "loss": 2.3228,
      "step": 7745
    },
    {
      "epoch": 0.20400316038978142,
      "grad_norm": 3.739337205886841,
      "learning_rate": 3.9809059784040035e-05,
      "loss": 1.4,
      "step": 7746
    },
    {
      "epoch": 0.20402949697129313,
      "grad_norm": 3.632378578186035,
      "learning_rate": 3.980774295496445e-05,
      "loss": 0.3844,
      "step": 7747
    },
    {
      "epoch": 0.20405583355280485,
      "grad_norm": 2.4780163764953613,
      "learning_rate": 3.9806426125888866e-05,
      "loss": 1.9745,
      "step": 7748
    },
    {
      "epoch": 0.20408217013431657,
      "grad_norm": 3.5137109756469727,
      "learning_rate": 3.9805109296813275e-05,
      "loss": 1.6529,
      "step": 7749
    },
    {
      "epoch": 0.20410850671582828,
      "grad_norm": 3.2006072998046875,
      "learning_rate": 3.980379246773769e-05,
      "loss": 1.1761,
      "step": 7750
    },
    {
      "epoch": 0.20413484329734,
      "grad_norm": 1.951158046722412,
      "learning_rate": 3.98024756386621e-05,
      "loss": 0.6224,
      "step": 7751
    },
    {
      "epoch": 0.2041611798788517,
      "grad_norm": 1.973855972290039,
      "learning_rate": 3.980115880958652e-05,
      "loss": 1.6062,
      "step": 7752
    },
    {
      "epoch": 0.20418751646036346,
      "grad_norm": 2.5205283164978027,
      "learning_rate": 3.979984198051093e-05,
      "loss": 2.3939,
      "step": 7753
    },
    {
      "epoch": 0.20421385304187517,
      "grad_norm": 1.919022560119629,
      "learning_rate": 3.9798525151435346e-05,
      "loss": 1.5804,
      "step": 7754
    },
    {
      "epoch": 0.2042401896233869,
      "grad_norm": 3.5701918601989746,
      "learning_rate": 3.979720832235976e-05,
      "loss": 1.8725,
      "step": 7755
    },
    {
      "epoch": 0.2042665262048986,
      "grad_norm": 2.765096664428711,
      "learning_rate": 3.979589149328417e-05,
      "loss": 1.3745,
      "step": 7756
    },
    {
      "epoch": 0.20429286278641032,
      "grad_norm": 2.1180624961853027,
      "learning_rate": 3.979457466420859e-05,
      "loss": 1.9313,
      "step": 7757
    },
    {
      "epoch": 0.20431919936792203,
      "grad_norm": 2.988429307937622,
      "learning_rate": 3.9793257835133e-05,
      "loss": 1.2353,
      "step": 7758
    },
    {
      "epoch": 0.20434553594943378,
      "grad_norm": 3.308535099029541,
      "learning_rate": 3.979194100605742e-05,
      "loss": 1.2714,
      "step": 7759
    },
    {
      "epoch": 0.2043718725309455,
      "grad_norm": 2.4223952293395996,
      "learning_rate": 3.9790624176981826e-05,
      "loss": 2.1543,
      "step": 7760
    },
    {
      "epoch": 0.2043982091124572,
      "grad_norm": 2.9478893280029297,
      "learning_rate": 3.978930734790625e-05,
      "loss": 1.9599,
      "step": 7761
    },
    {
      "epoch": 0.20442454569396892,
      "grad_norm": 2.621234893798828,
      "learning_rate": 3.978799051883066e-05,
      "loss": 2.0541,
      "step": 7762
    },
    {
      "epoch": 0.20445088227548064,
      "grad_norm": 1.7383151054382324,
      "learning_rate": 3.978667368975507e-05,
      "loss": 1.8054,
      "step": 7763
    },
    {
      "epoch": 0.20447721885699235,
      "grad_norm": 1.9030951261520386,
      "learning_rate": 3.978535686067948e-05,
      "loss": 1.2585,
      "step": 7764
    },
    {
      "epoch": 0.20450355543850407,
      "grad_norm": 2.5771563053131104,
      "learning_rate": 3.97840400316039e-05,
      "loss": 0.3584,
      "step": 7765
    },
    {
      "epoch": 0.2045298920200158,
      "grad_norm": 2.9761757850646973,
      "learning_rate": 3.978272320252831e-05,
      "loss": 1.765,
      "step": 7766
    },
    {
      "epoch": 0.20455622860152753,
      "grad_norm": 2.0030739307403564,
      "learning_rate": 3.978140637345273e-05,
      "loss": 1.6391,
      "step": 7767
    },
    {
      "epoch": 0.20458256518303924,
      "grad_norm": 2.9255735874176025,
      "learning_rate": 3.9780089544377144e-05,
      "loss": 1.7632,
      "step": 7768
    },
    {
      "epoch": 0.20460890176455096,
      "grad_norm": 2.131049394607544,
      "learning_rate": 3.977877271530155e-05,
      "loss": 0.7539,
      "step": 7769
    },
    {
      "epoch": 0.20463523834606268,
      "grad_norm": 3.041555643081665,
      "learning_rate": 3.977745588622597e-05,
      "loss": 2.6063,
      "step": 7770
    },
    {
      "epoch": 0.2046615749275744,
      "grad_norm": 2.188349723815918,
      "learning_rate": 3.9776139057150384e-05,
      "loss": 1.5763,
      "step": 7771
    },
    {
      "epoch": 0.2046879115090861,
      "grad_norm": 1.6239274740219116,
      "learning_rate": 3.97748222280748e-05,
      "loss": 1.3054,
      "step": 7772
    },
    {
      "epoch": 0.20471424809059785,
      "grad_norm": 1.9158140420913696,
      "learning_rate": 3.977350539899921e-05,
      "loss": 1.7708,
      "step": 7773
    },
    {
      "epoch": 0.20474058467210957,
      "grad_norm": 2.2672629356384277,
      "learning_rate": 3.9772188569923624e-05,
      "loss": 1.8807,
      "step": 7774
    },
    {
      "epoch": 0.20476692125362128,
      "grad_norm": 1.7786533832550049,
      "learning_rate": 3.977087174084804e-05,
      "loss": 1.8365,
      "step": 7775
    },
    {
      "epoch": 0.204793257835133,
      "grad_norm": 3.6295759677886963,
      "learning_rate": 3.9769554911772455e-05,
      "loss": 1.4215,
      "step": 7776
    },
    {
      "epoch": 0.2048195944166447,
      "grad_norm": 1.8746525049209595,
      "learning_rate": 3.976823808269687e-05,
      "loss": 1.9446,
      "step": 7777
    },
    {
      "epoch": 0.20484593099815643,
      "grad_norm": 3.5790648460388184,
      "learning_rate": 3.976692125362128e-05,
      "loss": 1.2644,
      "step": 7778
    },
    {
      "epoch": 0.20487226757966817,
      "grad_norm": 4.819107532501221,
      "learning_rate": 3.9765604424545695e-05,
      "loss": 2.1637,
      "step": 7779
    },
    {
      "epoch": 0.2048986041611799,
      "grad_norm": 4.649570465087891,
      "learning_rate": 3.976428759547011e-05,
      "loss": 1.3861,
      "step": 7780
    },
    {
      "epoch": 0.2049249407426916,
      "grad_norm": 2.5128061771392822,
      "learning_rate": 3.9762970766394526e-05,
      "loss": 2.654,
      "step": 7781
    },
    {
      "epoch": 0.20495127732420332,
      "grad_norm": 2.1249327659606934,
      "learning_rate": 3.9761653937318935e-05,
      "loss": 1.5575,
      "step": 7782
    },
    {
      "epoch": 0.20497761390571503,
      "grad_norm": 3.7322750091552734,
      "learning_rate": 3.976033710824335e-05,
      "loss": 1.4085,
      "step": 7783
    },
    {
      "epoch": 0.20500395048722675,
      "grad_norm": 3.530750274658203,
      "learning_rate": 3.9759020279167766e-05,
      "loss": 2.4919,
      "step": 7784
    },
    {
      "epoch": 0.20503028706873846,
      "grad_norm": 3.203949213027954,
      "learning_rate": 3.975770345009218e-05,
      "loss": 1.3842,
      "step": 7785
    },
    {
      "epoch": 0.2050566236502502,
      "grad_norm": 3.26556134223938,
      "learning_rate": 3.97563866210166e-05,
      "loss": 0.3793,
      "step": 7786
    },
    {
      "epoch": 0.20508296023176192,
      "grad_norm": 2.6151974201202393,
      "learning_rate": 3.9755069791941006e-05,
      "loss": 2.2486,
      "step": 7787
    },
    {
      "epoch": 0.20510929681327364,
      "grad_norm": 1.9042385816574097,
      "learning_rate": 3.975375296286542e-05,
      "loss": 1.3336,
      "step": 7788
    },
    {
      "epoch": 0.20513563339478535,
      "grad_norm": 1.940236210823059,
      "learning_rate": 3.975243613378983e-05,
      "loss": 1.7221,
      "step": 7789
    },
    {
      "epoch": 0.20516196997629707,
      "grad_norm": 2.0851359367370605,
      "learning_rate": 3.975111930471425e-05,
      "loss": 1.5509,
      "step": 7790
    },
    {
      "epoch": 0.2051883065578088,
      "grad_norm": 1.7379056215286255,
      "learning_rate": 3.974980247563866e-05,
      "loss": 1.3931,
      "step": 7791
    },
    {
      "epoch": 0.20521464313932053,
      "grad_norm": 4.5370097160339355,
      "learning_rate": 3.974848564656308e-05,
      "loss": 1.0969,
      "step": 7792
    },
    {
      "epoch": 0.20524097972083225,
      "grad_norm": 1.7076642513275146,
      "learning_rate": 3.974716881748749e-05,
      "loss": 1.7391,
      "step": 7793
    },
    {
      "epoch": 0.20526731630234396,
      "grad_norm": 1.7618438005447388,
      "learning_rate": 3.974585198841191e-05,
      "loss": 1.5138,
      "step": 7794
    },
    {
      "epoch": 0.20529365288385568,
      "grad_norm": 2.485245704650879,
      "learning_rate": 3.9744535159336324e-05,
      "loss": 1.6382,
      "step": 7795
    },
    {
      "epoch": 0.2053199894653674,
      "grad_norm": 1.801835060119629,
      "learning_rate": 3.974321833026073e-05,
      "loss": 1.5347,
      "step": 7796
    },
    {
      "epoch": 0.2053463260468791,
      "grad_norm": 2.680452585220337,
      "learning_rate": 3.974190150118515e-05,
      "loss": 0.6489,
      "step": 7797
    },
    {
      "epoch": 0.20537266262839082,
      "grad_norm": 3.282606363296509,
      "learning_rate": 3.974058467210956e-05,
      "loss": 1.2582,
      "step": 7798
    },
    {
      "epoch": 0.20539899920990257,
      "grad_norm": 4.64738655090332,
      "learning_rate": 3.973926784303398e-05,
      "loss": 2.0028,
      "step": 7799
    },
    {
      "epoch": 0.20542533579141428,
      "grad_norm": 2.5218048095703125,
      "learning_rate": 3.973795101395839e-05,
      "loss": 2.6483,
      "step": 7800
    },
    {
      "epoch": 0.205451672372926,
      "grad_norm": 3.784498929977417,
      "learning_rate": 3.9736634184882804e-05,
      "loss": 1.6081,
      "step": 7801
    },
    {
      "epoch": 0.2054780089544377,
      "grad_norm": 2.962722063064575,
      "learning_rate": 3.973531735580722e-05,
      "loss": 1.9634,
      "step": 7802
    },
    {
      "epoch": 0.20550434553594943,
      "grad_norm": 2.387471914291382,
      "learning_rate": 3.973400052673163e-05,
      "loss": 2.4667,
      "step": 7803
    },
    {
      "epoch": 0.20553068211746114,
      "grad_norm": 1.9236146211624146,
      "learning_rate": 3.973268369765605e-05,
      "loss": 2.0,
      "step": 7804
    },
    {
      "epoch": 0.20555701869897286,
      "grad_norm": 3.2074873447418213,
      "learning_rate": 3.973136686858046e-05,
      "loss": 1.2928,
      "step": 7805
    },
    {
      "epoch": 0.2055833552804846,
      "grad_norm": 3.463780403137207,
      "learning_rate": 3.9730050039504876e-05,
      "loss": 1.3878,
      "step": 7806
    },
    {
      "epoch": 0.20560969186199632,
      "grad_norm": 3.3297958374023438,
      "learning_rate": 3.9728733210429284e-05,
      "loss": 0.9296,
      "step": 7807
    },
    {
      "epoch": 0.20563602844350803,
      "grad_norm": 4.504313945770264,
      "learning_rate": 3.972741638135371e-05,
      "loss": 2.1668,
      "step": 7808
    },
    {
      "epoch": 0.20566236502501975,
      "grad_norm": 2.5805702209472656,
      "learning_rate": 3.9726099552278116e-05,
      "loss": 1.8362,
      "step": 7809
    },
    {
      "epoch": 0.20568870160653147,
      "grad_norm": 1.6617090702056885,
      "learning_rate": 3.972478272320253e-05,
      "loss": 2.0187,
      "step": 7810
    },
    {
      "epoch": 0.20571503818804318,
      "grad_norm": 1.7884222269058228,
      "learning_rate": 3.972346589412694e-05,
      "loss": 1.6353,
      "step": 7811
    },
    {
      "epoch": 0.20574137476955492,
      "grad_norm": 3.2227025032043457,
      "learning_rate": 3.9722149065051356e-05,
      "loss": 1.6051,
      "step": 7812
    },
    {
      "epoch": 0.20576771135106664,
      "grad_norm": 2.211688995361328,
      "learning_rate": 3.972083223597577e-05,
      "loss": 1.137,
      "step": 7813
    },
    {
      "epoch": 0.20579404793257836,
      "grad_norm": 1.8780994415283203,
      "learning_rate": 3.971951540690019e-05,
      "loss": 2.3147,
      "step": 7814
    },
    {
      "epoch": 0.20582038451409007,
      "grad_norm": 2.0965094566345215,
      "learning_rate": 3.97181985778246e-05,
      "loss": 2.4826,
      "step": 7815
    },
    {
      "epoch": 0.2058467210956018,
      "grad_norm": 2.779721260070801,
      "learning_rate": 3.971688174874901e-05,
      "loss": 1.6571,
      "step": 7816
    },
    {
      "epoch": 0.2058730576771135,
      "grad_norm": 1.896346092224121,
      "learning_rate": 3.971556491967343e-05,
      "loss": 1.7357,
      "step": 7817
    },
    {
      "epoch": 0.20589939425862522,
      "grad_norm": 2.545363664627075,
      "learning_rate": 3.971424809059784e-05,
      "loss": 1.9361,
      "step": 7818
    },
    {
      "epoch": 0.20592573084013696,
      "grad_norm": 1.7324621677398682,
      "learning_rate": 3.971293126152226e-05,
      "loss": 1.6786,
      "step": 7819
    },
    {
      "epoch": 0.20595206742164868,
      "grad_norm": 2.820807933807373,
      "learning_rate": 3.971161443244667e-05,
      "loss": 1.6343,
      "step": 7820
    },
    {
      "epoch": 0.2059784040031604,
      "grad_norm": 4.268813133239746,
      "learning_rate": 3.971029760337108e-05,
      "loss": 0.8925,
      "step": 7821
    },
    {
      "epoch": 0.2060047405846721,
      "grad_norm": 1.4191045761108398,
      "learning_rate": 3.97089807742955e-05,
      "loss": 1.7947,
      "step": 7822
    },
    {
      "epoch": 0.20603107716618382,
      "grad_norm": 4.610223293304443,
      "learning_rate": 3.9707663945219914e-05,
      "loss": 1.7113,
      "step": 7823
    },
    {
      "epoch": 0.20605741374769554,
      "grad_norm": 6.960455417633057,
      "learning_rate": 3.970634711614433e-05,
      "loss": 1.22,
      "step": 7824
    },
    {
      "epoch": 0.20608375032920728,
      "grad_norm": 1.6945736408233643,
      "learning_rate": 3.970503028706874e-05,
      "loss": 2.01,
      "step": 7825
    },
    {
      "epoch": 0.206110086910719,
      "grad_norm": 1.7801034450531006,
      "learning_rate": 3.9703713457993154e-05,
      "loss": 1.7731,
      "step": 7826
    },
    {
      "epoch": 0.2061364234922307,
      "grad_norm": 1.8213963508605957,
      "learning_rate": 3.970239662891757e-05,
      "loss": 2.5041,
      "step": 7827
    },
    {
      "epoch": 0.20616276007374243,
      "grad_norm": 1.783178687095642,
      "learning_rate": 3.9701079799841985e-05,
      "loss": 2.6439,
      "step": 7828
    },
    {
      "epoch": 0.20618909665525414,
      "grad_norm": 1.5821774005889893,
      "learning_rate": 3.9699762970766394e-05,
      "loss": 1.7259,
      "step": 7829
    },
    {
      "epoch": 0.20621543323676586,
      "grad_norm": 1.9780511856079102,
      "learning_rate": 3.969844614169081e-05,
      "loss": 1.6204,
      "step": 7830
    },
    {
      "epoch": 0.20624176981827758,
      "grad_norm": 2.2103583812713623,
      "learning_rate": 3.9697129312615225e-05,
      "loss": 1.5859,
      "step": 7831
    },
    {
      "epoch": 0.20626810639978932,
      "grad_norm": 3.5597517490386963,
      "learning_rate": 3.969581248353964e-05,
      "loss": 1.6507,
      "step": 7832
    },
    {
      "epoch": 0.20629444298130103,
      "grad_norm": 1.7971757650375366,
      "learning_rate": 3.9694495654464056e-05,
      "loss": 1.845,
      "step": 7833
    },
    {
      "epoch": 0.20632077956281275,
      "grad_norm": 2.1414246559143066,
      "learning_rate": 3.9693178825388465e-05,
      "loss": 2.5907,
      "step": 7834
    },
    {
      "epoch": 0.20634711614432447,
      "grad_norm": 1.8185389041900635,
      "learning_rate": 3.969186199631288e-05,
      "loss": 1.9206,
      "step": 7835
    },
    {
      "epoch": 0.20637345272583618,
      "grad_norm": 2.236732006072998,
      "learning_rate": 3.969054516723729e-05,
      "loss": 2.1211,
      "step": 7836
    },
    {
      "epoch": 0.2063997893073479,
      "grad_norm": 2.784217357635498,
      "learning_rate": 3.968922833816171e-05,
      "loss": 2.3413,
      "step": 7837
    },
    {
      "epoch": 0.2064261258888596,
      "grad_norm": 1.8819929361343384,
      "learning_rate": 3.968791150908612e-05,
      "loss": 2.4825,
      "step": 7838
    },
    {
      "epoch": 0.20645246247037136,
      "grad_norm": 2.5416150093078613,
      "learning_rate": 3.9686594680010536e-05,
      "loss": 2.0612,
      "step": 7839
    },
    {
      "epoch": 0.20647879905188307,
      "grad_norm": 2.7699124813079834,
      "learning_rate": 3.968527785093495e-05,
      "loss": 1.5596,
      "step": 7840
    },
    {
      "epoch": 0.2065051356333948,
      "grad_norm": 2.565701484680176,
      "learning_rate": 3.968396102185937e-05,
      "loss": 1.4777,
      "step": 7841
    },
    {
      "epoch": 0.2065314722149065,
      "grad_norm": 2.101482629776001,
      "learning_rate": 3.968264419278378e-05,
      "loss": 1.5001,
      "step": 7842
    },
    {
      "epoch": 0.20655780879641822,
      "grad_norm": 3.0457894802093506,
      "learning_rate": 3.968132736370819e-05,
      "loss": 1.7086,
      "step": 7843
    },
    {
      "epoch": 0.20658414537792993,
      "grad_norm": 1.9946099519729614,
      "learning_rate": 3.968001053463261e-05,
      "loss": 2.1921,
      "step": 7844
    },
    {
      "epoch": 0.20661048195944168,
      "grad_norm": 5.115975379943848,
      "learning_rate": 3.9678693705557016e-05,
      "loss": 1.0979,
      "step": 7845
    },
    {
      "epoch": 0.2066368185409534,
      "grad_norm": 1.4720896482467651,
      "learning_rate": 3.967737687648144e-05,
      "loss": 1.9008,
      "step": 7846
    },
    {
      "epoch": 0.2066631551224651,
      "grad_norm": 2.379565715789795,
      "learning_rate": 3.967606004740585e-05,
      "loss": 1.3649,
      "step": 7847
    },
    {
      "epoch": 0.20668949170397682,
      "grad_norm": 3.408848762512207,
      "learning_rate": 3.967474321833026e-05,
      "loss": 1.704,
      "step": 7848
    },
    {
      "epoch": 0.20671582828548854,
      "grad_norm": 2.993016004562378,
      "learning_rate": 3.967342638925468e-05,
      "loss": 2.2501,
      "step": 7849
    },
    {
      "epoch": 0.20674216486700026,
      "grad_norm": 2.8244590759277344,
      "learning_rate": 3.967210956017909e-05,
      "loss": 2.5329,
      "step": 7850
    },
    {
      "epoch": 0.20676850144851197,
      "grad_norm": 2.7097952365875244,
      "learning_rate": 3.967079273110351e-05,
      "loss": 0.5743,
      "step": 7851
    },
    {
      "epoch": 0.20679483803002371,
      "grad_norm": 2.438704252243042,
      "learning_rate": 3.966947590202792e-05,
      "loss": 1.8576,
      "step": 7852
    },
    {
      "epoch": 0.20682117461153543,
      "grad_norm": 2.6458921432495117,
      "learning_rate": 3.9668159072952334e-05,
      "loss": 1.5375,
      "step": 7853
    },
    {
      "epoch": 0.20684751119304715,
      "grad_norm": 3.281848430633545,
      "learning_rate": 3.966684224387674e-05,
      "loss": 1.2269,
      "step": 7854
    },
    {
      "epoch": 0.20687384777455886,
      "grad_norm": 4.1413798332214355,
      "learning_rate": 3.9665525414801165e-05,
      "loss": 0.9242,
      "step": 7855
    },
    {
      "epoch": 0.20690018435607058,
      "grad_norm": 3.585641622543335,
      "learning_rate": 3.9664208585725574e-05,
      "loss": 1.3356,
      "step": 7856
    },
    {
      "epoch": 0.2069265209375823,
      "grad_norm": 2.10442852973938,
      "learning_rate": 3.966289175664999e-05,
      "loss": 2.0747,
      "step": 7857
    },
    {
      "epoch": 0.206952857519094,
      "grad_norm": 2.297427177429199,
      "learning_rate": 3.9661574927574405e-05,
      "loss": 1.764,
      "step": 7858
    },
    {
      "epoch": 0.20697919410060575,
      "grad_norm": 1.7722662687301636,
      "learning_rate": 3.9660258098498814e-05,
      "loss": 2.133,
      "step": 7859
    },
    {
      "epoch": 0.20700553068211747,
      "grad_norm": 1.921962857246399,
      "learning_rate": 3.9658941269423237e-05,
      "loss": 2.1509,
      "step": 7860
    },
    {
      "epoch": 0.20703186726362918,
      "grad_norm": 1.724496603012085,
      "learning_rate": 3.9657624440347645e-05,
      "loss": 0.4232,
      "step": 7861
    },
    {
      "epoch": 0.2070582038451409,
      "grad_norm": 1.533440113067627,
      "learning_rate": 3.965630761127206e-05,
      "loss": 1.4872,
      "step": 7862
    },
    {
      "epoch": 0.2070845404266526,
      "grad_norm": 1.7583832740783691,
      "learning_rate": 3.965499078219647e-05,
      "loss": 1.4856,
      "step": 7863
    },
    {
      "epoch": 0.20711087700816433,
      "grad_norm": 2.3645436763763428,
      "learning_rate": 3.9653673953120885e-05,
      "loss": 2.5735,
      "step": 7864
    },
    {
      "epoch": 0.20713721358967607,
      "grad_norm": 2.202500581741333,
      "learning_rate": 3.96523571240453e-05,
      "loss": 1.5335,
      "step": 7865
    },
    {
      "epoch": 0.2071635501711878,
      "grad_norm": 3.426295757293701,
      "learning_rate": 3.9651040294969717e-05,
      "loss": 1.2126,
      "step": 7866
    },
    {
      "epoch": 0.2071898867526995,
      "grad_norm": 2.2901694774627686,
      "learning_rate": 3.9649723465894125e-05,
      "loss": 1.9396,
      "step": 7867
    },
    {
      "epoch": 0.20721622333421122,
      "grad_norm": 1.7368117570877075,
      "learning_rate": 3.964840663681854e-05,
      "loss": 2.2809,
      "step": 7868
    },
    {
      "epoch": 0.20724255991572293,
      "grad_norm": 2.8945674896240234,
      "learning_rate": 3.9647089807742957e-05,
      "loss": 1.7752,
      "step": 7869
    },
    {
      "epoch": 0.20726889649723465,
      "grad_norm": 3.279308557510376,
      "learning_rate": 3.964577297866737e-05,
      "loss": 1.0625,
      "step": 7870
    },
    {
      "epoch": 0.20729523307874637,
      "grad_norm": 6.087759971618652,
      "learning_rate": 3.964445614959179e-05,
      "loss": 2.4217,
      "step": 7871
    },
    {
      "epoch": 0.2073215696602581,
      "grad_norm": 1.9429460763931274,
      "learning_rate": 3.9643139320516197e-05,
      "loss": 1.2704,
      "step": 7872
    },
    {
      "epoch": 0.20734790624176982,
      "grad_norm": 1.9781205654144287,
      "learning_rate": 3.964182249144061e-05,
      "loss": 2.1478,
      "step": 7873
    },
    {
      "epoch": 0.20737424282328154,
      "grad_norm": 2.5948643684387207,
      "learning_rate": 3.964050566236503e-05,
      "loss": 0.6268,
      "step": 7874
    },
    {
      "epoch": 0.20740057940479326,
      "grad_norm": 2.821392059326172,
      "learning_rate": 3.963918883328944e-05,
      "loss": 1.141,
      "step": 7875
    },
    {
      "epoch": 0.20742691598630497,
      "grad_norm": 1.7201229333877563,
      "learning_rate": 3.963787200421385e-05,
      "loss": 0.2368,
      "step": 7876
    },
    {
      "epoch": 0.2074532525678167,
      "grad_norm": 2.5808374881744385,
      "learning_rate": 3.963655517513827e-05,
      "loss": 1.6653,
      "step": 7877
    },
    {
      "epoch": 0.20747958914932843,
      "grad_norm": 1.9911696910858154,
      "learning_rate": 3.963523834606268e-05,
      "loss": 1.6758,
      "step": 7878
    },
    {
      "epoch": 0.20750592573084015,
      "grad_norm": 2.5546717643737793,
      "learning_rate": 3.96339215169871e-05,
      "loss": 2.1234,
      "step": 7879
    },
    {
      "epoch": 0.20753226231235186,
      "grad_norm": 3.1540091037750244,
      "learning_rate": 3.9632604687911515e-05,
      "loss": 0.481,
      "step": 7880
    },
    {
      "epoch": 0.20755859889386358,
      "grad_norm": 3.5949947834014893,
      "learning_rate": 3.9631287858835923e-05,
      "loss": 1.4144,
      "step": 7881
    },
    {
      "epoch": 0.2075849354753753,
      "grad_norm": 1.531070351600647,
      "learning_rate": 3.962997102976034e-05,
      "loss": 0.3953,
      "step": 7882
    },
    {
      "epoch": 0.207611272056887,
      "grad_norm": 1.7296377420425415,
      "learning_rate": 3.962865420068475e-05,
      "loss": 1.5199,
      "step": 7883
    },
    {
      "epoch": 0.20763760863839872,
      "grad_norm": 3.0197765827178955,
      "learning_rate": 3.962733737160917e-05,
      "loss": 0.3168,
      "step": 7884
    },
    {
      "epoch": 0.20766394521991047,
      "grad_norm": 1.5494749546051025,
      "learning_rate": 3.962602054253358e-05,
      "loss": 1.657,
      "step": 7885
    },
    {
      "epoch": 0.20769028180142218,
      "grad_norm": 8.258563041687012,
      "learning_rate": 3.9624703713457995e-05,
      "loss": 1.8136,
      "step": 7886
    },
    {
      "epoch": 0.2077166183829339,
      "grad_norm": 3.0122323036193848,
      "learning_rate": 3.962338688438241e-05,
      "loss": 0.8083,
      "step": 7887
    },
    {
      "epoch": 0.20774295496444561,
      "grad_norm": 4.37794303894043,
      "learning_rate": 3.9622070055306826e-05,
      "loss": 2.3482,
      "step": 7888
    },
    {
      "epoch": 0.20776929154595733,
      "grad_norm": 1.7530187368392944,
      "learning_rate": 3.962075322623124e-05,
      "loss": 1.936,
      "step": 7889
    },
    {
      "epoch": 0.20779562812746905,
      "grad_norm": 2.242640972137451,
      "learning_rate": 3.961943639715565e-05,
      "loss": 0.9809,
      "step": 7890
    },
    {
      "epoch": 0.20782196470898076,
      "grad_norm": 3.2633426189422607,
      "learning_rate": 3.9618119568080066e-05,
      "loss": 1.6336,
      "step": 7891
    },
    {
      "epoch": 0.2078483012904925,
      "grad_norm": 2.079007148742676,
      "learning_rate": 3.9616802739004475e-05,
      "loss": 1.8929,
      "step": 7892
    },
    {
      "epoch": 0.20787463787200422,
      "grad_norm": 2.1454122066497803,
      "learning_rate": 3.96154859099289e-05,
      "loss": 1.1304,
      "step": 7893
    },
    {
      "epoch": 0.20790097445351594,
      "grad_norm": 3.15075421333313,
      "learning_rate": 3.9614169080853306e-05,
      "loss": 1.4001,
      "step": 7894
    },
    {
      "epoch": 0.20792731103502765,
      "grad_norm": 3.404799222946167,
      "learning_rate": 3.961285225177772e-05,
      "loss": 0.5493,
      "step": 7895
    },
    {
      "epoch": 0.20795364761653937,
      "grad_norm": 3.996750831604004,
      "learning_rate": 3.961153542270214e-05,
      "loss": 0.5825,
      "step": 7896
    },
    {
      "epoch": 0.20797998419805108,
      "grad_norm": 2.167515277862549,
      "learning_rate": 3.9610218593626546e-05,
      "loss": 1.8508,
      "step": 7897
    },
    {
      "epoch": 0.20800632077956283,
      "grad_norm": 2.107673168182373,
      "learning_rate": 3.960890176455097e-05,
      "loss": 0.6475,
      "step": 7898
    },
    {
      "epoch": 0.20803265736107454,
      "grad_norm": 2.8799149990081787,
      "learning_rate": 3.960758493547538e-05,
      "loss": 1.3423,
      "step": 7899
    },
    {
      "epoch": 0.20805899394258626,
      "grad_norm": 3.097374439239502,
      "learning_rate": 3.960626810639979e-05,
      "loss": 1.8045,
      "step": 7900
    },
    {
      "epoch": 0.20808533052409797,
      "grad_norm": 3.866426944732666,
      "learning_rate": 3.96049512773242e-05,
      "loss": 2.3912,
      "step": 7901
    },
    {
      "epoch": 0.2081116671056097,
      "grad_norm": 1.7553191184997559,
      "learning_rate": 3.960363444824862e-05,
      "loss": 1.6756,
      "step": 7902
    },
    {
      "epoch": 0.2081380036871214,
      "grad_norm": 6.046514511108398,
      "learning_rate": 3.960231761917303e-05,
      "loss": 1.7576,
      "step": 7903
    },
    {
      "epoch": 0.20816434026863312,
      "grad_norm": 3.3796393871307373,
      "learning_rate": 3.960100079009745e-05,
      "loss": 2.0495,
      "step": 7904
    },
    {
      "epoch": 0.20819067685014486,
      "grad_norm": 2.272343158721924,
      "learning_rate": 3.9599683961021864e-05,
      "loss": 1.525,
      "step": 7905
    },
    {
      "epoch": 0.20821701343165658,
      "grad_norm": 1.5682498216629028,
      "learning_rate": 3.959836713194627e-05,
      "loss": 1.3021,
      "step": 7906
    },
    {
      "epoch": 0.2082433500131683,
      "grad_norm": 3.398550033569336,
      "learning_rate": 3.9597050302870695e-05,
      "loss": 1.5152,
      "step": 7907
    },
    {
      "epoch": 0.20826968659468,
      "grad_norm": 2.3274986743927,
      "learning_rate": 3.9595733473795104e-05,
      "loss": 1.2908,
      "step": 7908
    },
    {
      "epoch": 0.20829602317619172,
      "grad_norm": 3.44852876663208,
      "learning_rate": 3.959441664471952e-05,
      "loss": 1.4496,
      "step": 7909
    },
    {
      "epoch": 0.20832235975770344,
      "grad_norm": 2.179391622543335,
      "learning_rate": 3.959309981564393e-05,
      "loss": 1.7392,
      "step": 7910
    },
    {
      "epoch": 0.20834869633921518,
      "grad_norm": 1.858309268951416,
      "learning_rate": 3.9591782986568344e-05,
      "loss": 1.9066,
      "step": 7911
    },
    {
      "epoch": 0.2083750329207269,
      "grad_norm": 2.2308707237243652,
      "learning_rate": 3.959046615749276e-05,
      "loss": 2.0192,
      "step": 7912
    },
    {
      "epoch": 0.20840136950223861,
      "grad_norm": 3.1797797679901123,
      "learning_rate": 3.9589149328417175e-05,
      "loss": 1.8226,
      "step": 7913
    },
    {
      "epoch": 0.20842770608375033,
      "grad_norm": 2.7949371337890625,
      "learning_rate": 3.9587832499341584e-05,
      "loss": 1.8544,
      "step": 7914
    },
    {
      "epoch": 0.20845404266526205,
      "grad_norm": 4.285569190979004,
      "learning_rate": 3.9586515670266e-05,
      "loss": 1.4802,
      "step": 7915
    },
    {
      "epoch": 0.20848037924677376,
      "grad_norm": 1.9868905544281006,
      "learning_rate": 3.9585198841190415e-05,
      "loss": 1.9951,
      "step": 7916
    },
    {
      "epoch": 0.20850671582828548,
      "grad_norm": 2.244162082672119,
      "learning_rate": 3.958388201211483e-05,
      "loss": 1.7914,
      "step": 7917
    },
    {
      "epoch": 0.20853305240979722,
      "grad_norm": 4.924079418182373,
      "learning_rate": 3.9582565183039246e-05,
      "loss": 1.8412,
      "step": 7918
    },
    {
      "epoch": 0.20855938899130894,
      "grad_norm": 2.948178291320801,
      "learning_rate": 3.9581248353963655e-05,
      "loss": 1.5166,
      "step": 7919
    },
    {
      "epoch": 0.20858572557282065,
      "grad_norm": 2.1109559535980225,
      "learning_rate": 3.957993152488807e-05,
      "loss": 1.6516,
      "step": 7920
    },
    {
      "epoch": 0.20861206215433237,
      "grad_norm": 3.0030221939086914,
      "learning_rate": 3.9578614695812486e-05,
      "loss": 1.0304,
      "step": 7921
    },
    {
      "epoch": 0.20863839873584408,
      "grad_norm": 2.160588026046753,
      "learning_rate": 3.95772978667369e-05,
      "loss": 0.8921,
      "step": 7922
    },
    {
      "epoch": 0.2086647353173558,
      "grad_norm": 2.5613348484039307,
      "learning_rate": 3.957598103766131e-05,
      "loss": 1.8314,
      "step": 7923
    },
    {
      "epoch": 0.20869107189886751,
      "grad_norm": 2.6622626781463623,
      "learning_rate": 3.9574664208585726e-05,
      "loss": 2.0809,
      "step": 7924
    },
    {
      "epoch": 0.20871740848037926,
      "grad_norm": 5.511320114135742,
      "learning_rate": 3.957334737951014e-05,
      "loss": 1.8326,
      "step": 7925
    },
    {
      "epoch": 0.20874374506189097,
      "grad_norm": 1.4537285566329956,
      "learning_rate": 3.957203055043456e-05,
      "loss": 1.9966,
      "step": 7926
    },
    {
      "epoch": 0.2087700816434027,
      "grad_norm": 2.2338857650756836,
      "learning_rate": 3.957071372135897e-05,
      "loss": 1.8528,
      "step": 7927
    },
    {
      "epoch": 0.2087964182249144,
      "grad_norm": 2.480236768722534,
      "learning_rate": 3.956939689228338e-05,
      "loss": 0.5629,
      "step": 7928
    },
    {
      "epoch": 0.20882275480642612,
      "grad_norm": 1.8498308658599854,
      "learning_rate": 3.95680800632078e-05,
      "loss": 1.9333,
      "step": 7929
    },
    {
      "epoch": 0.20884909138793784,
      "grad_norm": 5.344048023223877,
      "learning_rate": 3.9566763234132206e-05,
      "loss": 1.9673,
      "step": 7930
    },
    {
      "epoch": 0.20887542796944958,
      "grad_norm": 3.506174325942993,
      "learning_rate": 3.956544640505663e-05,
      "loss": 1.2029,
      "step": 7931
    },
    {
      "epoch": 0.2089017645509613,
      "grad_norm": 2.1873860359191895,
      "learning_rate": 3.956412957598104e-05,
      "loss": 1.3081,
      "step": 7932
    },
    {
      "epoch": 0.208928101132473,
      "grad_norm": 5.196494102478027,
      "learning_rate": 3.956281274690545e-05,
      "loss": 2.702,
      "step": 7933
    },
    {
      "epoch": 0.20895443771398473,
      "grad_norm": 2.246225595474243,
      "learning_rate": 3.956149591782987e-05,
      "loss": 1.9787,
      "step": 7934
    },
    {
      "epoch": 0.20898077429549644,
      "grad_norm": 4.003600597381592,
      "learning_rate": 3.956017908875428e-05,
      "loss": 1.4045,
      "step": 7935
    },
    {
      "epoch": 0.20900711087700816,
      "grad_norm": 2.243312120437622,
      "learning_rate": 3.95588622596787e-05,
      "loss": 1.6769,
      "step": 7936
    },
    {
      "epoch": 0.20903344745851987,
      "grad_norm": 3.09871244430542,
      "learning_rate": 3.955754543060311e-05,
      "loss": 1.3345,
      "step": 7937
    },
    {
      "epoch": 0.20905978404003162,
      "grad_norm": 1.5548056364059448,
      "learning_rate": 3.9556228601527524e-05,
      "loss": 1.7791,
      "step": 7938
    },
    {
      "epoch": 0.20908612062154333,
      "grad_norm": 1.826427698135376,
      "learning_rate": 3.955491177245193e-05,
      "loss": 1.4339,
      "step": 7939
    },
    {
      "epoch": 0.20911245720305505,
      "grad_norm": 2.251462459564209,
      "learning_rate": 3.9553594943376356e-05,
      "loss": 2.0785,
      "step": 7940
    },
    {
      "epoch": 0.20913879378456676,
      "grad_norm": 2.6527514457702637,
      "learning_rate": 3.9552278114300764e-05,
      "loss": 1.6428,
      "step": 7941
    },
    {
      "epoch": 0.20916513036607848,
      "grad_norm": 1.7091227769851685,
      "learning_rate": 3.955096128522518e-05,
      "loss": 1.4601,
      "step": 7942
    },
    {
      "epoch": 0.2091914669475902,
      "grad_norm": 2.1505069732666016,
      "learning_rate": 3.9549644456149596e-05,
      "loss": 1.8869,
      "step": 7943
    },
    {
      "epoch": 0.2092178035291019,
      "grad_norm": 2.5253753662109375,
      "learning_rate": 3.9548327627074004e-05,
      "loss": 1.8705,
      "step": 7944
    },
    {
      "epoch": 0.20924414011061365,
      "grad_norm": 2.7928519248962402,
      "learning_rate": 3.954701079799843e-05,
      "loss": 1.5282,
      "step": 7945
    },
    {
      "epoch": 0.20927047669212537,
      "grad_norm": 1.557199239730835,
      "learning_rate": 3.9545693968922836e-05,
      "loss": 1.8086,
      "step": 7946
    },
    {
      "epoch": 0.20929681327363708,
      "grad_norm": 2.3240275382995605,
      "learning_rate": 3.954437713984725e-05,
      "loss": 1.7553,
      "step": 7947
    },
    {
      "epoch": 0.2093231498551488,
      "grad_norm": 1.8232619762420654,
      "learning_rate": 3.954306031077166e-05,
      "loss": 1.5949,
      "step": 7948
    },
    {
      "epoch": 0.20934948643666051,
      "grad_norm": 1.6279993057250977,
      "learning_rate": 3.9541743481696076e-05,
      "loss": 1.0965,
      "step": 7949
    },
    {
      "epoch": 0.20937582301817223,
      "grad_norm": 1.9425480365753174,
      "learning_rate": 3.954042665262049e-05,
      "loss": 1.7829,
      "step": 7950
    },
    {
      "epoch": 0.20940215959968397,
      "grad_norm": 1.9660980701446533,
      "learning_rate": 3.953910982354491e-05,
      "loss": 1.6255,
      "step": 7951
    },
    {
      "epoch": 0.2094284961811957,
      "grad_norm": 2.075363874435425,
      "learning_rate": 3.953779299446932e-05,
      "loss": 2.031,
      "step": 7952
    },
    {
      "epoch": 0.2094548327627074,
      "grad_norm": 2.5722994804382324,
      "learning_rate": 3.953647616539373e-05,
      "loss": 2.08,
      "step": 7953
    },
    {
      "epoch": 0.20948116934421912,
      "grad_norm": 1.8071092367172241,
      "learning_rate": 3.9535159336318154e-05,
      "loss": 2.0004,
      "step": 7954
    },
    {
      "epoch": 0.20950750592573084,
      "grad_norm": 6.527217388153076,
      "learning_rate": 3.953384250724256e-05,
      "loss": 2.1184,
      "step": 7955
    },
    {
      "epoch": 0.20953384250724255,
      "grad_norm": 5.131868839263916,
      "learning_rate": 3.953252567816698e-05,
      "loss": 2.0516,
      "step": 7956
    },
    {
      "epoch": 0.20956017908875427,
      "grad_norm": 2.3576133251190186,
      "learning_rate": 3.953120884909139e-05,
      "loss": 1.3078,
      "step": 7957
    },
    {
      "epoch": 0.209586515670266,
      "grad_norm": 3.3976054191589355,
      "learning_rate": 3.95298920200158e-05,
      "loss": 1.2705,
      "step": 7958
    },
    {
      "epoch": 0.20961285225177773,
      "grad_norm": 3.6096582412719727,
      "learning_rate": 3.952857519094022e-05,
      "loss": 1.1667,
      "step": 7959
    },
    {
      "epoch": 0.20963918883328944,
      "grad_norm": 3.5529160499572754,
      "learning_rate": 3.9527258361864634e-05,
      "loss": 1.494,
      "step": 7960
    },
    {
      "epoch": 0.20966552541480116,
      "grad_norm": 4.715931415557861,
      "learning_rate": 3.952594153278905e-05,
      "loss": 1.4576,
      "step": 7961
    },
    {
      "epoch": 0.20969186199631287,
      "grad_norm": 1.8498784303665161,
      "learning_rate": 3.952462470371346e-05,
      "loss": 1.6296,
      "step": 7962
    },
    {
      "epoch": 0.2097181985778246,
      "grad_norm": 2.778883457183838,
      "learning_rate": 3.9523307874637874e-05,
      "loss": 1.9152,
      "step": 7963
    },
    {
      "epoch": 0.20974453515933633,
      "grad_norm": 2.1728055477142334,
      "learning_rate": 3.952199104556229e-05,
      "loss": 1.5361,
      "step": 7964
    },
    {
      "epoch": 0.20977087174084805,
      "grad_norm": 2.0484607219696045,
      "learning_rate": 3.9520674216486705e-05,
      "loss": 1.5986,
      "step": 7965
    },
    {
      "epoch": 0.20979720832235976,
      "grad_norm": 1.9635207653045654,
      "learning_rate": 3.9519357387411114e-05,
      "loss": 2.329,
      "step": 7966
    },
    {
      "epoch": 0.20982354490387148,
      "grad_norm": 2.995798110961914,
      "learning_rate": 3.951804055833553e-05,
      "loss": 1.4193,
      "step": 7967
    },
    {
      "epoch": 0.2098498814853832,
      "grad_norm": 2.4451513290405273,
      "learning_rate": 3.951672372925994e-05,
      "loss": 1.6387,
      "step": 7968
    },
    {
      "epoch": 0.2098762180668949,
      "grad_norm": 3.6962287425994873,
      "learning_rate": 3.951540690018436e-05,
      "loss": 0.9997,
      "step": 7969
    },
    {
      "epoch": 0.20990255464840663,
      "grad_norm": 1.6026966571807861,
      "learning_rate": 3.951409007110877e-05,
      "loss": 1.3702,
      "step": 7970
    },
    {
      "epoch": 0.20992889122991837,
      "grad_norm": 3.512328863143921,
      "learning_rate": 3.9512773242033185e-05,
      "loss": 0.7588,
      "step": 7971
    },
    {
      "epoch": 0.20995522781143008,
      "grad_norm": 3.359097480773926,
      "learning_rate": 3.95114564129576e-05,
      "loss": 1.204,
      "step": 7972
    },
    {
      "epoch": 0.2099815643929418,
      "grad_norm": 1.5736089944839478,
      "learning_rate": 3.9510139583882016e-05,
      "loss": 1.8898,
      "step": 7973
    },
    {
      "epoch": 0.21000790097445352,
      "grad_norm": 2.251445770263672,
      "learning_rate": 3.950882275480643e-05,
      "loss": 1.6537,
      "step": 7974
    },
    {
      "epoch": 0.21003423755596523,
      "grad_norm": 1.9817070960998535,
      "learning_rate": 3.950750592573084e-05,
      "loss": 2.0569,
      "step": 7975
    },
    {
      "epoch": 0.21006057413747695,
      "grad_norm": 1.7193737030029297,
      "learning_rate": 3.9506189096655256e-05,
      "loss": 2.2624,
      "step": 7976
    },
    {
      "epoch": 0.21008691071898866,
      "grad_norm": 4.927498817443848,
      "learning_rate": 3.9504872267579665e-05,
      "loss": 2.5041,
      "step": 7977
    },
    {
      "epoch": 0.2101132473005004,
      "grad_norm": 3.848907232284546,
      "learning_rate": 3.950355543850409e-05,
      "loss": 1.2478,
      "step": 7978
    },
    {
      "epoch": 0.21013958388201212,
      "grad_norm": 4.467748641967773,
      "learning_rate": 3.9502238609428496e-05,
      "loss": 0.8114,
      "step": 7979
    },
    {
      "epoch": 0.21016592046352384,
      "grad_norm": 1.7255222797393799,
      "learning_rate": 3.950092178035291e-05,
      "loss": 1.8872,
      "step": 7980
    },
    {
      "epoch": 0.21019225704503555,
      "grad_norm": 1.6148625612258911,
      "learning_rate": 3.949960495127733e-05,
      "loss": 2.1682,
      "step": 7981
    },
    {
      "epoch": 0.21021859362654727,
      "grad_norm": 2.7554187774658203,
      "learning_rate": 3.9498288122201736e-05,
      "loss": 0.5923,
      "step": 7982
    },
    {
      "epoch": 0.21024493020805898,
      "grad_norm": 3.0917000770568848,
      "learning_rate": 3.949697129312616e-05,
      "loss": 2.0324,
      "step": 7983
    },
    {
      "epoch": 0.21027126678957073,
      "grad_norm": 2.207585573196411,
      "learning_rate": 3.949565446405057e-05,
      "loss": 2.1554,
      "step": 7984
    },
    {
      "epoch": 0.21029760337108244,
      "grad_norm": 2.8638224601745605,
      "learning_rate": 3.949433763497498e-05,
      "loss": 2.3346,
      "step": 7985
    },
    {
      "epoch": 0.21032393995259416,
      "grad_norm": 2.264763832092285,
      "learning_rate": 3.949302080589939e-05,
      "loss": 0.598,
      "step": 7986
    },
    {
      "epoch": 0.21035027653410587,
      "grad_norm": 2.041144609451294,
      "learning_rate": 3.9491703976823814e-05,
      "loss": 1.193,
      "step": 7987
    },
    {
      "epoch": 0.2103766131156176,
      "grad_norm": 1.8995825052261353,
      "learning_rate": 3.949038714774822e-05,
      "loss": 1.4939,
      "step": 7988
    },
    {
      "epoch": 0.2104029496971293,
      "grad_norm": 2.838090181350708,
      "learning_rate": 3.948907031867264e-05,
      "loss": 1.4816,
      "step": 7989
    },
    {
      "epoch": 0.21042928627864102,
      "grad_norm": 1.7127673625946045,
      "learning_rate": 3.9487753489597054e-05,
      "loss": 2.0585,
      "step": 7990
    },
    {
      "epoch": 0.21045562286015276,
      "grad_norm": 3.825827121734619,
      "learning_rate": 3.948643666052146e-05,
      "loss": 1.0772,
      "step": 7991
    },
    {
      "epoch": 0.21048195944166448,
      "grad_norm": 3.1475493907928467,
      "learning_rate": 3.9485119831445885e-05,
      "loss": 0.6566,
      "step": 7992
    },
    {
      "epoch": 0.2105082960231762,
      "grad_norm": 3.4131553173065186,
      "learning_rate": 3.9483803002370294e-05,
      "loss": 1.3837,
      "step": 7993
    },
    {
      "epoch": 0.2105346326046879,
      "grad_norm": 2.033433437347412,
      "learning_rate": 3.948248617329471e-05,
      "loss": 2.53,
      "step": 7994
    },
    {
      "epoch": 0.21056096918619963,
      "grad_norm": 2.1205668449401855,
      "learning_rate": 3.948116934421912e-05,
      "loss": 0.4346,
      "step": 7995
    },
    {
      "epoch": 0.21058730576771134,
      "grad_norm": 4.05999755859375,
      "learning_rate": 3.9479852515143534e-05,
      "loss": 1.8416,
      "step": 7996
    },
    {
      "epoch": 0.21061364234922308,
      "grad_norm": 2.101485252380371,
      "learning_rate": 3.947853568606795e-05,
      "loss": 1.964,
      "step": 7997
    },
    {
      "epoch": 0.2106399789307348,
      "grad_norm": 3.6859042644500732,
      "learning_rate": 3.9477218856992365e-05,
      "loss": 1.0435,
      "step": 7998
    },
    {
      "epoch": 0.21066631551224652,
      "grad_norm": 2.1933062076568604,
      "learning_rate": 3.947590202791678e-05,
      "loss": 1.4206,
      "step": 7999
    },
    {
      "epoch": 0.21069265209375823,
      "grad_norm": 2.146085500717163,
      "learning_rate": 3.947458519884119e-05,
      "loss": 1.8722,
      "step": 8000
    },
    {
      "epoch": 0.21071898867526995,
      "grad_norm": 2.975377082824707,
      "learning_rate": 3.9473268369765605e-05,
      "loss": 1.5781,
      "step": 8001
    },
    {
      "epoch": 0.21074532525678166,
      "grad_norm": 2.0121936798095703,
      "learning_rate": 3.947195154069002e-05,
      "loss": 1.9954,
      "step": 8002
    },
    {
      "epoch": 0.21077166183829338,
      "grad_norm": 2.756566286087036,
      "learning_rate": 3.9470634711614437e-05,
      "loss": 1.8285,
      "step": 8003
    },
    {
      "epoch": 0.21079799841980512,
      "grad_norm": 4.1593017578125,
      "learning_rate": 3.9469317882538845e-05,
      "loss": 0.464,
      "step": 8004
    },
    {
      "epoch": 0.21082433500131684,
      "grad_norm": 1.49372398853302,
      "learning_rate": 3.946800105346326e-05,
      "loss": 2.0069,
      "step": 8005
    },
    {
      "epoch": 0.21085067158282855,
      "grad_norm": 5.051287651062012,
      "learning_rate": 3.9466684224387677e-05,
      "loss": 0.9049,
      "step": 8006
    },
    {
      "epoch": 0.21087700816434027,
      "grad_norm": 6.765331268310547,
      "learning_rate": 3.946536739531209e-05,
      "loss": 1.9278,
      "step": 8007
    },
    {
      "epoch": 0.21090334474585198,
      "grad_norm": 1.9707120656967163,
      "learning_rate": 3.946405056623651e-05,
      "loss": 1.7449,
      "step": 8008
    },
    {
      "epoch": 0.2109296813273637,
      "grad_norm": 2.769061803817749,
      "learning_rate": 3.946273373716092e-05,
      "loss": 1.6184,
      "step": 8009
    },
    {
      "epoch": 0.21095601790887542,
      "grad_norm": 2.0725924968719482,
      "learning_rate": 3.946141690808533e-05,
      "loss": 1.8683,
      "step": 8010
    },
    {
      "epoch": 0.21098235449038716,
      "grad_norm": 1.9264018535614014,
      "learning_rate": 3.946010007900975e-05,
      "loss": 2.1224,
      "step": 8011
    },
    {
      "epoch": 0.21100869107189887,
      "grad_norm": 1.8104604482650757,
      "learning_rate": 3.9458783249934163e-05,
      "loss": 1.8013,
      "step": 8012
    },
    {
      "epoch": 0.2110350276534106,
      "grad_norm": 2.7635090351104736,
      "learning_rate": 3.945746642085857e-05,
      "loss": 1.6869,
      "step": 8013
    },
    {
      "epoch": 0.2110613642349223,
      "grad_norm": 4.2592387199401855,
      "learning_rate": 3.945614959178299e-05,
      "loss": 2.6423,
      "step": 8014
    },
    {
      "epoch": 0.21108770081643402,
      "grad_norm": 2.7908198833465576,
      "learning_rate": 3.94548327627074e-05,
      "loss": 1.6301,
      "step": 8015
    },
    {
      "epoch": 0.21111403739794574,
      "grad_norm": 3.105816602706909,
      "learning_rate": 3.945351593363182e-05,
      "loss": 2.0555,
      "step": 8016
    },
    {
      "epoch": 0.21114037397945748,
      "grad_norm": 1.9376375675201416,
      "learning_rate": 3.945219910455623e-05,
      "loss": 2.1331,
      "step": 8017
    },
    {
      "epoch": 0.2111667105609692,
      "grad_norm": 3.2177765369415283,
      "learning_rate": 3.9450882275480643e-05,
      "loss": 2.3851,
      "step": 8018
    },
    {
      "epoch": 0.2111930471424809,
      "grad_norm": 3.5235183238983154,
      "learning_rate": 3.944956544640506e-05,
      "loss": 1.1531,
      "step": 8019
    },
    {
      "epoch": 0.21121938372399263,
      "grad_norm": 1.6920013427734375,
      "learning_rate": 3.9448248617329475e-05,
      "loss": 1.6535,
      "step": 8020
    },
    {
      "epoch": 0.21124572030550434,
      "grad_norm": 1.7701884508132935,
      "learning_rate": 3.944693178825389e-05,
      "loss": 2.3294,
      "step": 8021
    },
    {
      "epoch": 0.21127205688701606,
      "grad_norm": 3.3554046154022217,
      "learning_rate": 3.94456149591783e-05,
      "loss": 1.0569,
      "step": 8022
    },
    {
      "epoch": 0.21129839346852777,
      "grad_norm": 1.471840739250183,
      "learning_rate": 3.9444298130102715e-05,
      "loss": 2.2983,
      "step": 8023
    },
    {
      "epoch": 0.21132473005003952,
      "grad_norm": 1.7614887952804565,
      "learning_rate": 3.9442981301027123e-05,
      "loss": 2.1428,
      "step": 8024
    },
    {
      "epoch": 0.21135106663155123,
      "grad_norm": 1.8810094594955444,
      "learning_rate": 3.9441664471951546e-05,
      "loss": 2.6995,
      "step": 8025
    },
    {
      "epoch": 0.21137740321306295,
      "grad_norm": 1.9735652208328247,
      "learning_rate": 3.9440347642875955e-05,
      "loss": 1.5631,
      "step": 8026
    },
    {
      "epoch": 0.21140373979457466,
      "grad_norm": 2.613091230392456,
      "learning_rate": 3.943903081380037e-05,
      "loss": 1.3443,
      "step": 8027
    },
    {
      "epoch": 0.21143007637608638,
      "grad_norm": 1.8289783000946045,
      "learning_rate": 3.9437713984724786e-05,
      "loss": 1.7056,
      "step": 8028
    },
    {
      "epoch": 0.2114564129575981,
      "grad_norm": 2.3008124828338623,
      "learning_rate": 3.9436397155649195e-05,
      "loss": 1.8188,
      "step": 8029
    },
    {
      "epoch": 0.2114827495391098,
      "grad_norm": 1.3884129524230957,
      "learning_rate": 3.943508032657362e-05,
      "loss": 1.3644,
      "step": 8030
    },
    {
      "epoch": 0.21150908612062155,
      "grad_norm": 2.093122959136963,
      "learning_rate": 3.9433763497498026e-05,
      "loss": 1.4573,
      "step": 8031
    },
    {
      "epoch": 0.21153542270213327,
      "grad_norm": 1.619066596031189,
      "learning_rate": 3.943244666842244e-05,
      "loss": 1.9759,
      "step": 8032
    },
    {
      "epoch": 0.21156175928364498,
      "grad_norm": 3.6559643745422363,
      "learning_rate": 3.943112983934685e-05,
      "loss": 1.308,
      "step": 8033
    },
    {
      "epoch": 0.2115880958651567,
      "grad_norm": 2.3009626865386963,
      "learning_rate": 3.9429813010271266e-05,
      "loss": 1.958,
      "step": 8034
    },
    {
      "epoch": 0.21161443244666842,
      "grad_norm": 1.6011055707931519,
      "learning_rate": 3.942849618119568e-05,
      "loss": 1.6162,
      "step": 8035
    },
    {
      "epoch": 0.21164076902818013,
      "grad_norm": 2.6110005378723145,
      "learning_rate": 3.94271793521201e-05,
      "loss": 1.6828,
      "step": 8036
    },
    {
      "epoch": 0.21166710560969187,
      "grad_norm": 2.4418509006500244,
      "learning_rate": 3.942586252304451e-05,
      "loss": 2.273,
      "step": 8037
    },
    {
      "epoch": 0.2116934421912036,
      "grad_norm": 1.865779995918274,
      "learning_rate": 3.942454569396892e-05,
      "loss": 1.5067,
      "step": 8038
    },
    {
      "epoch": 0.2117197787727153,
      "grad_norm": 2.686727285385132,
      "learning_rate": 3.9423228864893344e-05,
      "loss": 1.6904,
      "step": 8039
    },
    {
      "epoch": 0.21174611535422702,
      "grad_norm": 1.7998002767562866,
      "learning_rate": 3.942191203581775e-05,
      "loss": 1.8184,
      "step": 8040
    },
    {
      "epoch": 0.21177245193573874,
      "grad_norm": 5.24328088760376,
      "learning_rate": 3.942059520674217e-05,
      "loss": 1.3782,
      "step": 8041
    },
    {
      "epoch": 0.21179878851725045,
      "grad_norm": 2.5756797790527344,
      "learning_rate": 3.941927837766658e-05,
      "loss": 2.2707,
      "step": 8042
    },
    {
      "epoch": 0.21182512509876217,
      "grad_norm": 1.9890908002853394,
      "learning_rate": 3.941796154859099e-05,
      "loss": 0.9836,
      "step": 8043
    },
    {
      "epoch": 0.2118514616802739,
      "grad_norm": 2.926546096801758,
      "learning_rate": 3.941664471951541e-05,
      "loss": 2.0829,
      "step": 8044
    },
    {
      "epoch": 0.21187779826178563,
      "grad_norm": 3.032869815826416,
      "learning_rate": 3.9415327890439824e-05,
      "loss": 0.4195,
      "step": 8045
    },
    {
      "epoch": 0.21190413484329734,
      "grad_norm": 2.888871908187866,
      "learning_rate": 3.941401106136424e-05,
      "loss": 1.5956,
      "step": 8046
    },
    {
      "epoch": 0.21193047142480906,
      "grad_norm": 3.109415292739868,
      "learning_rate": 3.941269423228865e-05,
      "loss": 1.4676,
      "step": 8047
    },
    {
      "epoch": 0.21195680800632077,
      "grad_norm": 1.9078865051269531,
      "learning_rate": 3.9411377403213064e-05,
      "loss": 1.9799,
      "step": 8048
    },
    {
      "epoch": 0.2119831445878325,
      "grad_norm": 2.3028550148010254,
      "learning_rate": 3.941006057413748e-05,
      "loss": 1.7985,
      "step": 8049
    },
    {
      "epoch": 0.21200948116934423,
      "grad_norm": 4.7780280113220215,
      "learning_rate": 3.9408743745061895e-05,
      "loss": 0.7895,
      "step": 8050
    },
    {
      "epoch": 0.21203581775085595,
      "grad_norm": 1.716330885887146,
      "learning_rate": 3.9407426915986304e-05,
      "loss": 1.8136,
      "step": 8051
    },
    {
      "epoch": 0.21206215433236766,
      "grad_norm": 2.3866307735443115,
      "learning_rate": 3.940611008691072e-05,
      "loss": 1.8914,
      "step": 8052
    },
    {
      "epoch": 0.21208849091387938,
      "grad_norm": 2.0796773433685303,
      "learning_rate": 3.9404793257835135e-05,
      "loss": 1.6821,
      "step": 8053
    },
    {
      "epoch": 0.2121148274953911,
      "grad_norm": 3.524477005004883,
      "learning_rate": 3.940347642875955e-05,
      "loss": 0.962,
      "step": 8054
    },
    {
      "epoch": 0.2121411640769028,
      "grad_norm": 2.139411449432373,
      "learning_rate": 3.9402159599683966e-05,
      "loss": 0.2723,
      "step": 8055
    },
    {
      "epoch": 0.21216750065841453,
      "grad_norm": 1.8068301677703857,
      "learning_rate": 3.9400842770608375e-05,
      "loss": 1.7596,
      "step": 8056
    },
    {
      "epoch": 0.21219383723992627,
      "grad_norm": 1.8532053232192993,
      "learning_rate": 3.939952594153279e-05,
      "loss": 1.6732,
      "step": 8057
    },
    {
      "epoch": 0.21222017382143799,
      "grad_norm": 2.1167242527008057,
      "learning_rate": 3.9398209112457206e-05,
      "loss": 1.9052,
      "step": 8058
    },
    {
      "epoch": 0.2122465104029497,
      "grad_norm": 2.0105865001678467,
      "learning_rate": 3.939689228338162e-05,
      "loss": 2.0846,
      "step": 8059
    },
    {
      "epoch": 0.21227284698446142,
      "grad_norm": 2.3318231105804443,
      "learning_rate": 3.939557545430603e-05,
      "loss": 1.567,
      "step": 8060
    },
    {
      "epoch": 0.21229918356597313,
      "grad_norm": 4.269582271575928,
      "learning_rate": 3.9394258625230446e-05,
      "loss": 1.2244,
      "step": 8061
    },
    {
      "epoch": 0.21232552014748485,
      "grad_norm": 3.0847957134246826,
      "learning_rate": 3.9392941796154855e-05,
      "loss": 1.5797,
      "step": 8062
    },
    {
      "epoch": 0.21235185672899656,
      "grad_norm": 2.812221050262451,
      "learning_rate": 3.939162496707928e-05,
      "loss": 2.1353,
      "step": 8063
    },
    {
      "epoch": 0.2123781933105083,
      "grad_norm": 2.608167886734009,
      "learning_rate": 3.939030813800369e-05,
      "loss": 2.238,
      "step": 8064
    },
    {
      "epoch": 0.21240452989202002,
      "grad_norm": 2.6146254539489746,
      "learning_rate": 3.93889913089281e-05,
      "loss": 1.3269,
      "step": 8065
    },
    {
      "epoch": 0.21243086647353174,
      "grad_norm": 3.309941053390503,
      "learning_rate": 3.938767447985252e-05,
      "loss": 0.8602,
      "step": 8066
    },
    {
      "epoch": 0.21245720305504345,
      "grad_norm": 4.590884208679199,
      "learning_rate": 3.9386357650776926e-05,
      "loss": 1.4454,
      "step": 8067
    },
    {
      "epoch": 0.21248353963655517,
      "grad_norm": 1.7693549394607544,
      "learning_rate": 3.938504082170135e-05,
      "loss": 1.9132,
      "step": 8068
    },
    {
      "epoch": 0.21250987621806688,
      "grad_norm": 3.7004294395446777,
      "learning_rate": 3.938372399262576e-05,
      "loss": 1.0606,
      "step": 8069
    },
    {
      "epoch": 0.21253621279957863,
      "grad_norm": 2.025315761566162,
      "learning_rate": 3.938240716355017e-05,
      "loss": 0.9544,
      "step": 8070
    },
    {
      "epoch": 0.21256254938109034,
      "grad_norm": 1.7604368925094604,
      "learning_rate": 3.938109033447458e-05,
      "loss": 2.0598,
      "step": 8071
    },
    {
      "epoch": 0.21258888596260206,
      "grad_norm": 1.6236159801483154,
      "learning_rate": 3.9379773505399004e-05,
      "loss": 1.9443,
      "step": 8072
    },
    {
      "epoch": 0.21261522254411377,
      "grad_norm": 1.57721745967865,
      "learning_rate": 3.937845667632341e-05,
      "loss": 1.1893,
      "step": 8073
    },
    {
      "epoch": 0.2126415591256255,
      "grad_norm": 1.517897129058838,
      "learning_rate": 3.937713984724783e-05,
      "loss": 1.8691,
      "step": 8074
    },
    {
      "epoch": 0.2126678957071372,
      "grad_norm": 1.708858847618103,
      "learning_rate": 3.9375823018172244e-05,
      "loss": 1.4678,
      "step": 8075
    },
    {
      "epoch": 0.21269423228864892,
      "grad_norm": 3.3207497596740723,
      "learning_rate": 3.937450618909665e-05,
      "loss": 1.1941,
      "step": 8076
    },
    {
      "epoch": 0.21272056887016066,
      "grad_norm": 2.11263370513916,
      "learning_rate": 3.9373189360021076e-05,
      "loss": 0.6194,
      "step": 8077
    },
    {
      "epoch": 0.21274690545167238,
      "grad_norm": 2.0888655185699463,
      "learning_rate": 3.9371872530945484e-05,
      "loss": 1.9023,
      "step": 8078
    },
    {
      "epoch": 0.2127732420331841,
      "grad_norm": 3.2067151069641113,
      "learning_rate": 3.93705557018699e-05,
      "loss": 1.5088,
      "step": 8079
    },
    {
      "epoch": 0.2127995786146958,
      "grad_norm": 3.494105577468872,
      "learning_rate": 3.936923887279431e-05,
      "loss": 1.0594,
      "step": 8080
    },
    {
      "epoch": 0.21282591519620753,
      "grad_norm": 1.7514755725860596,
      "learning_rate": 3.9367922043718724e-05,
      "loss": 1.3005,
      "step": 8081
    },
    {
      "epoch": 0.21285225177771924,
      "grad_norm": 2.5412983894348145,
      "learning_rate": 3.936660521464314e-05,
      "loss": 2.2427,
      "step": 8082
    },
    {
      "epoch": 0.21287858835923096,
      "grad_norm": 3.1526293754577637,
      "learning_rate": 3.9365288385567556e-05,
      "loss": 0.5093,
      "step": 8083
    },
    {
      "epoch": 0.2129049249407427,
      "grad_norm": 1.6861717700958252,
      "learning_rate": 3.936397155649197e-05,
      "loss": 2.0373,
      "step": 8084
    },
    {
      "epoch": 0.21293126152225442,
      "grad_norm": 2.658473014831543,
      "learning_rate": 3.936265472741638e-05,
      "loss": 0.8306,
      "step": 8085
    },
    {
      "epoch": 0.21295759810376613,
      "grad_norm": 4.400993824005127,
      "learning_rate": 3.93613378983408e-05,
      "loss": 0.7366,
      "step": 8086
    },
    {
      "epoch": 0.21298393468527785,
      "grad_norm": 3.7876157760620117,
      "learning_rate": 3.936002106926521e-05,
      "loss": 0.705,
      "step": 8087
    },
    {
      "epoch": 0.21301027126678956,
      "grad_norm": 2.0813534259796143,
      "learning_rate": 3.935870424018963e-05,
      "loss": 0.7664,
      "step": 8088
    },
    {
      "epoch": 0.21303660784830128,
      "grad_norm": 3.0450165271759033,
      "learning_rate": 3.9357387411114036e-05,
      "loss": 1.3153,
      "step": 8089
    },
    {
      "epoch": 0.21306294442981302,
      "grad_norm": 2.5086708068847656,
      "learning_rate": 3.935607058203845e-05,
      "loss": 1.8139,
      "step": 8090
    },
    {
      "epoch": 0.21308928101132474,
      "grad_norm": 1.7112269401550293,
      "learning_rate": 3.935475375296287e-05,
      "loss": 1.5665,
      "step": 8091
    },
    {
      "epoch": 0.21311561759283645,
      "grad_norm": 1.685372233390808,
      "learning_rate": 3.935343692388728e-05,
      "loss": 1.7403,
      "step": 8092
    },
    {
      "epoch": 0.21314195417434817,
      "grad_norm": 3.176758289337158,
      "learning_rate": 3.93521200948117e-05,
      "loss": 1.1468,
      "step": 8093
    },
    {
      "epoch": 0.21316829075585988,
      "grad_norm": 3.393217086791992,
      "learning_rate": 3.935080326573611e-05,
      "loss": 1.772,
      "step": 8094
    },
    {
      "epoch": 0.2131946273373716,
      "grad_norm": 1.8444277048110962,
      "learning_rate": 3.934948643666052e-05,
      "loss": 2.1946,
      "step": 8095
    },
    {
      "epoch": 0.21322096391888332,
      "grad_norm": 3.4809162616729736,
      "learning_rate": 3.934816960758494e-05,
      "loss": 2.087,
      "step": 8096
    },
    {
      "epoch": 0.21324730050039506,
      "grad_norm": 1.779625415802002,
      "learning_rate": 3.9346852778509354e-05,
      "loss": 2.1212,
      "step": 8097
    },
    {
      "epoch": 0.21327363708190678,
      "grad_norm": 4.503268241882324,
      "learning_rate": 3.934553594943376e-05,
      "loss": 2.1692,
      "step": 8098
    },
    {
      "epoch": 0.2132999736634185,
      "grad_norm": 1.8279377222061157,
      "learning_rate": 3.934421912035818e-05,
      "loss": 1.3458,
      "step": 8099
    },
    {
      "epoch": 0.2133263102449302,
      "grad_norm": 2.2709789276123047,
      "learning_rate": 3.9342902291282594e-05,
      "loss": 1.43,
      "step": 8100
    },
    {
      "epoch": 0.21335264682644192,
      "grad_norm": 4.984404563903809,
      "learning_rate": 3.934158546220701e-05,
      "loss": 0.9168,
      "step": 8101
    },
    {
      "epoch": 0.21337898340795364,
      "grad_norm": 3.854466438293457,
      "learning_rate": 3.9340268633131425e-05,
      "loss": 1.6919,
      "step": 8102
    },
    {
      "epoch": 0.21340531998946538,
      "grad_norm": 2.469802141189575,
      "learning_rate": 3.9338951804055834e-05,
      "loss": 1.4695,
      "step": 8103
    },
    {
      "epoch": 0.2134316565709771,
      "grad_norm": 2.463916778564453,
      "learning_rate": 3.933763497498025e-05,
      "loss": 1.8117,
      "step": 8104
    },
    {
      "epoch": 0.2134579931524888,
      "grad_norm": 2.3769354820251465,
      "learning_rate": 3.9336318145904665e-05,
      "loss": 2.4983,
      "step": 8105
    },
    {
      "epoch": 0.21348432973400053,
      "grad_norm": 1.6960322856903076,
      "learning_rate": 3.933500131682908e-05,
      "loss": 1.7101,
      "step": 8106
    },
    {
      "epoch": 0.21351066631551224,
      "grad_norm": 2.715395450592041,
      "learning_rate": 3.933368448775349e-05,
      "loss": 1.855,
      "step": 8107
    },
    {
      "epoch": 0.21353700289702396,
      "grad_norm": 1.7114248275756836,
      "learning_rate": 3.9332367658677905e-05,
      "loss": 1.4534,
      "step": 8108
    },
    {
      "epoch": 0.21356333947853567,
      "grad_norm": 1.9994984865188599,
      "learning_rate": 3.933105082960232e-05,
      "loss": 2.0607,
      "step": 8109
    },
    {
      "epoch": 0.21358967606004742,
      "grad_norm": 2.062744379043579,
      "learning_rate": 3.9329734000526736e-05,
      "loss": 1.4558,
      "step": 8110
    },
    {
      "epoch": 0.21361601264155913,
      "grad_norm": 3.768462657928467,
      "learning_rate": 3.932841717145115e-05,
      "loss": 0.8915,
      "step": 8111
    },
    {
      "epoch": 0.21364234922307085,
      "grad_norm": 2.3410520553588867,
      "learning_rate": 3.932710034237556e-05,
      "loss": 0.4157,
      "step": 8112
    },
    {
      "epoch": 0.21366868580458256,
      "grad_norm": 1.735868215560913,
      "learning_rate": 3.9325783513299976e-05,
      "loss": 1.4064,
      "step": 8113
    },
    {
      "epoch": 0.21369502238609428,
      "grad_norm": 1.9412875175476074,
      "learning_rate": 3.9324466684224385e-05,
      "loss": 1.5671,
      "step": 8114
    },
    {
      "epoch": 0.213721358967606,
      "grad_norm": 1.9730874300003052,
      "learning_rate": 3.932314985514881e-05,
      "loss": 1.279,
      "step": 8115
    },
    {
      "epoch": 0.2137476955491177,
      "grad_norm": 3.5073163509368896,
      "learning_rate": 3.9321833026073216e-05,
      "loss": 1.6784,
      "step": 8116
    },
    {
      "epoch": 0.21377403213062945,
      "grad_norm": 2.093677520751953,
      "learning_rate": 3.932051619699763e-05,
      "loss": 0.8411,
      "step": 8117
    },
    {
      "epoch": 0.21380036871214117,
      "grad_norm": 1.8346515893936157,
      "learning_rate": 3.931919936792204e-05,
      "loss": 2.2679,
      "step": 8118
    },
    {
      "epoch": 0.21382670529365289,
      "grad_norm": 2.296050786972046,
      "learning_rate": 3.931788253884646e-05,
      "loss": 1.7204,
      "step": 8119
    },
    {
      "epoch": 0.2138530418751646,
      "grad_norm": 1.9171382188796997,
      "learning_rate": 3.931656570977087e-05,
      "loss": 1.5225,
      "step": 8120
    },
    {
      "epoch": 0.21387937845667632,
      "grad_norm": 2.800374746322632,
      "learning_rate": 3.931524888069529e-05,
      "loss": 1.4449,
      "step": 8121
    },
    {
      "epoch": 0.21390571503818803,
      "grad_norm": 3.167890787124634,
      "learning_rate": 3.93139320516197e-05,
      "loss": 2.0883,
      "step": 8122
    },
    {
      "epoch": 0.21393205161969978,
      "grad_norm": 2.944833993911743,
      "learning_rate": 3.931261522254411e-05,
      "loss": 2.0241,
      "step": 8123
    },
    {
      "epoch": 0.2139583882012115,
      "grad_norm": 2.202903985977173,
      "learning_rate": 3.9311298393468534e-05,
      "loss": 2.2159,
      "step": 8124
    },
    {
      "epoch": 0.2139847247827232,
      "grad_norm": 1.5188966989517212,
      "learning_rate": 3.930998156439294e-05,
      "loss": 1.3158,
      "step": 8125
    },
    {
      "epoch": 0.21401106136423492,
      "grad_norm": 3.3034634590148926,
      "learning_rate": 3.930866473531736e-05,
      "loss": 1.3759,
      "step": 8126
    },
    {
      "epoch": 0.21403739794574664,
      "grad_norm": 3.7458913326263428,
      "learning_rate": 3.930734790624177e-05,
      "loss": 1.1541,
      "step": 8127
    },
    {
      "epoch": 0.21406373452725835,
      "grad_norm": 2.579270362854004,
      "learning_rate": 3.930603107716618e-05,
      "loss": 1.4533,
      "step": 8128
    },
    {
      "epoch": 0.21409007110877007,
      "grad_norm": 2.903691530227661,
      "learning_rate": 3.93047142480906e-05,
      "loss": 2.4269,
      "step": 8129
    },
    {
      "epoch": 0.2141164076902818,
      "grad_norm": 2.269531488418579,
      "learning_rate": 3.9303397419015014e-05,
      "loss": 1.958,
      "step": 8130
    },
    {
      "epoch": 0.21414274427179353,
      "grad_norm": 2.460794448852539,
      "learning_rate": 3.930208058993943e-05,
      "loss": 1.6406,
      "step": 8131
    },
    {
      "epoch": 0.21416908085330524,
      "grad_norm": 2.968259811401367,
      "learning_rate": 3.930076376086384e-05,
      "loss": 0.8086,
      "step": 8132
    },
    {
      "epoch": 0.21419541743481696,
      "grad_norm": 2.827868700027466,
      "learning_rate": 3.929944693178826e-05,
      "loss": 1.6039,
      "step": 8133
    },
    {
      "epoch": 0.21422175401632867,
      "grad_norm": 2.632143497467041,
      "learning_rate": 3.929813010271267e-05,
      "loss": 1.8037,
      "step": 8134
    },
    {
      "epoch": 0.2142480905978404,
      "grad_norm": 3.924879789352417,
      "learning_rate": 3.9296813273637085e-05,
      "loss": 1.2124,
      "step": 8135
    },
    {
      "epoch": 0.21427442717935213,
      "grad_norm": 2.677705764770508,
      "learning_rate": 3.9295496444561494e-05,
      "loss": 1.4815,
      "step": 8136
    },
    {
      "epoch": 0.21430076376086385,
      "grad_norm": 2.3418219089508057,
      "learning_rate": 3.929417961548591e-05,
      "loss": 1.3942,
      "step": 8137
    },
    {
      "epoch": 0.21432710034237556,
      "grad_norm": 1.8835002183914185,
      "learning_rate": 3.9292862786410325e-05,
      "loss": 1.9455,
      "step": 8138
    },
    {
      "epoch": 0.21435343692388728,
      "grad_norm": 1.9340204000473022,
      "learning_rate": 3.929154595733474e-05,
      "loss": 1.9129,
      "step": 8139
    },
    {
      "epoch": 0.214379773505399,
      "grad_norm": 1.8426212072372437,
      "learning_rate": 3.929022912825916e-05,
      "loss": 1.9917,
      "step": 8140
    },
    {
      "epoch": 0.2144061100869107,
      "grad_norm": 1.7304800748825073,
      "learning_rate": 3.9288912299183565e-05,
      "loss": 1.6529,
      "step": 8141
    },
    {
      "epoch": 0.21443244666842243,
      "grad_norm": 3.842853307723999,
      "learning_rate": 3.928759547010798e-05,
      "loss": 1.7706,
      "step": 8142
    },
    {
      "epoch": 0.21445878324993417,
      "grad_norm": 2.678192138671875,
      "learning_rate": 3.92862786410324e-05,
      "loss": 1.1649,
      "step": 8143
    },
    {
      "epoch": 0.2144851198314459,
      "grad_norm": 2.3395910263061523,
      "learning_rate": 3.928496181195681e-05,
      "loss": 2.067,
      "step": 8144
    },
    {
      "epoch": 0.2145114564129576,
      "grad_norm": 2.5694284439086914,
      "learning_rate": 3.928364498288122e-05,
      "loss": 2.1264,
      "step": 8145
    },
    {
      "epoch": 0.21453779299446932,
      "grad_norm": 1.923811912536621,
      "learning_rate": 3.928232815380564e-05,
      "loss": 1.6467,
      "step": 8146
    },
    {
      "epoch": 0.21456412957598103,
      "grad_norm": 1.795712947845459,
      "learning_rate": 3.928101132473005e-05,
      "loss": 2.1434,
      "step": 8147
    },
    {
      "epoch": 0.21459046615749275,
      "grad_norm": 3.4010791778564453,
      "learning_rate": 3.927969449565447e-05,
      "loss": 1.1761,
      "step": 8148
    },
    {
      "epoch": 0.21461680273900446,
      "grad_norm": 2.623258590698242,
      "learning_rate": 3.9278377666578883e-05,
      "loss": 2.1505,
      "step": 8149
    },
    {
      "epoch": 0.2146431393205162,
      "grad_norm": 1.5902289152145386,
      "learning_rate": 3.927706083750329e-05,
      "loss": 1.9774,
      "step": 8150
    },
    {
      "epoch": 0.21466947590202792,
      "grad_norm": 1.7851754426956177,
      "learning_rate": 3.927574400842771e-05,
      "loss": 1.9715,
      "step": 8151
    },
    {
      "epoch": 0.21469581248353964,
      "grad_norm": 2.2035019397735596,
      "learning_rate": 3.9274427179352123e-05,
      "loss": 1.2878,
      "step": 8152
    },
    {
      "epoch": 0.21472214906505135,
      "grad_norm": 2.0112054347991943,
      "learning_rate": 3.927311035027654e-05,
      "loss": 1.5258,
      "step": 8153
    },
    {
      "epoch": 0.21474848564656307,
      "grad_norm": 5.362493515014648,
      "learning_rate": 3.927179352120095e-05,
      "loss": 1.764,
      "step": 8154
    },
    {
      "epoch": 0.21477482222807479,
      "grad_norm": 3.442628860473633,
      "learning_rate": 3.9270476692125363e-05,
      "loss": 2.2062,
      "step": 8155
    },
    {
      "epoch": 0.21480115880958653,
      "grad_norm": 2.052248477935791,
      "learning_rate": 3.926915986304978e-05,
      "loss": 2.1628,
      "step": 8156
    },
    {
      "epoch": 0.21482749539109824,
      "grad_norm": 1.6592828035354614,
      "learning_rate": 3.9267843033974195e-05,
      "loss": 1.7517,
      "step": 8157
    },
    {
      "epoch": 0.21485383197260996,
      "grad_norm": 1.9546493291854858,
      "learning_rate": 3.926652620489861e-05,
      "loss": 1.1069,
      "step": 8158
    },
    {
      "epoch": 0.21488016855412168,
      "grad_norm": 1.9161245822906494,
      "learning_rate": 3.926520937582302e-05,
      "loss": 1.6825,
      "step": 8159
    },
    {
      "epoch": 0.2149065051356334,
      "grad_norm": 1.6454392671585083,
      "learning_rate": 3.9263892546747435e-05,
      "loss": 1.9439,
      "step": 8160
    },
    {
      "epoch": 0.2149328417171451,
      "grad_norm": 3.8970978260040283,
      "learning_rate": 3.9262575717671844e-05,
      "loss": 1.0736,
      "step": 8161
    },
    {
      "epoch": 0.21495917829865682,
      "grad_norm": 1.952965259552002,
      "learning_rate": 3.9261258888596266e-05,
      "loss": 1.2462,
      "step": 8162
    },
    {
      "epoch": 0.21498551488016857,
      "grad_norm": 2.411151885986328,
      "learning_rate": 3.9259942059520675e-05,
      "loss": 1.919,
      "step": 8163
    },
    {
      "epoch": 0.21501185146168028,
      "grad_norm": 4.017030715942383,
      "learning_rate": 3.925862523044509e-05,
      "loss": 0.3093,
      "step": 8164
    },
    {
      "epoch": 0.215038188043192,
      "grad_norm": 1.7976256608963013,
      "learning_rate": 3.92573084013695e-05,
      "loss": 1.2207,
      "step": 8165
    },
    {
      "epoch": 0.2150645246247037,
      "grad_norm": 2.222109317779541,
      "learning_rate": 3.925599157229392e-05,
      "loss": 1.636,
      "step": 8166
    },
    {
      "epoch": 0.21509086120621543,
      "grad_norm": 2.583864688873291,
      "learning_rate": 3.925467474321833e-05,
      "loss": 0.7542,
      "step": 8167
    },
    {
      "epoch": 0.21511719778772714,
      "grad_norm": 1.7864383459091187,
      "learning_rate": 3.9253357914142746e-05,
      "loss": 1.9142,
      "step": 8168
    },
    {
      "epoch": 0.21514353436923886,
      "grad_norm": 1.547110915184021,
      "learning_rate": 3.925204108506716e-05,
      "loss": 2.2719,
      "step": 8169
    },
    {
      "epoch": 0.2151698709507506,
      "grad_norm": 4.676783561706543,
      "learning_rate": 3.925072425599157e-05,
      "loss": 1.7861,
      "step": 8170
    },
    {
      "epoch": 0.21519620753226232,
      "grad_norm": 3.0446035861968994,
      "learning_rate": 3.924940742691599e-05,
      "loss": 2.1101,
      "step": 8171
    },
    {
      "epoch": 0.21522254411377403,
      "grad_norm": 2.4542737007141113,
      "learning_rate": 3.92480905978404e-05,
      "loss": 1.2738,
      "step": 8172
    },
    {
      "epoch": 0.21524888069528575,
      "grad_norm": 2.255436658859253,
      "learning_rate": 3.924677376876482e-05,
      "loss": 0.5037,
      "step": 8173
    },
    {
      "epoch": 0.21527521727679746,
      "grad_norm": 4.145355701446533,
      "learning_rate": 3.9245456939689226e-05,
      "loss": 1.8512,
      "step": 8174
    },
    {
      "epoch": 0.21530155385830918,
      "grad_norm": 3.1572439670562744,
      "learning_rate": 3.924414011061364e-05,
      "loss": 2.0275,
      "step": 8175
    },
    {
      "epoch": 0.21532789043982092,
      "grad_norm": 4.270723819732666,
      "learning_rate": 3.924282328153806e-05,
      "loss": 1.8923,
      "step": 8176
    },
    {
      "epoch": 0.21535422702133264,
      "grad_norm": 3.3614985942840576,
      "learning_rate": 3.924150645246247e-05,
      "loss": 1.5709,
      "step": 8177
    },
    {
      "epoch": 0.21538056360284435,
      "grad_norm": 1.5842660665512085,
      "learning_rate": 3.924018962338689e-05,
      "loss": 1.5191,
      "step": 8178
    },
    {
      "epoch": 0.21540690018435607,
      "grad_norm": 2.3136115074157715,
      "learning_rate": 3.92388727943113e-05,
      "loss": 1.4973,
      "step": 8179
    },
    {
      "epoch": 0.21543323676586779,
      "grad_norm": 3.8716657161712646,
      "learning_rate": 3.923755596523571e-05,
      "loss": 1.9589,
      "step": 8180
    },
    {
      "epoch": 0.2154595733473795,
      "grad_norm": 2.2407655715942383,
      "learning_rate": 3.923623913616013e-05,
      "loss": 1.5226,
      "step": 8181
    },
    {
      "epoch": 0.21548590992889122,
      "grad_norm": 2.9864284992218018,
      "learning_rate": 3.9234922307084544e-05,
      "loss": 0.9025,
      "step": 8182
    },
    {
      "epoch": 0.21551224651040296,
      "grad_norm": 6.218202114105225,
      "learning_rate": 3.923360547800895e-05,
      "loss": 2.3504,
      "step": 8183
    },
    {
      "epoch": 0.21553858309191468,
      "grad_norm": 3.4184114933013916,
      "learning_rate": 3.923228864893337e-05,
      "loss": 1.9084,
      "step": 8184
    },
    {
      "epoch": 0.2155649196734264,
      "grad_norm": 3.4305033683776855,
      "learning_rate": 3.9230971819857784e-05,
      "loss": 1.4184,
      "step": 8185
    },
    {
      "epoch": 0.2155912562549381,
      "grad_norm": 3.723510503768921,
      "learning_rate": 3.92296549907822e-05,
      "loss": 2.2996,
      "step": 8186
    },
    {
      "epoch": 0.21561759283644982,
      "grad_norm": 2.1249070167541504,
      "learning_rate": 3.9228338161706615e-05,
      "loss": 1.625,
      "step": 8187
    },
    {
      "epoch": 0.21564392941796154,
      "grad_norm": 2.4870362281799316,
      "learning_rate": 3.9227021332631024e-05,
      "loss": 0.6118,
      "step": 8188
    },
    {
      "epoch": 0.21567026599947328,
      "grad_norm": 2.8394012451171875,
      "learning_rate": 3.922570450355544e-05,
      "loss": 1.3732,
      "step": 8189
    },
    {
      "epoch": 0.215696602580985,
      "grad_norm": 2.797839403152466,
      "learning_rate": 3.9224387674479855e-05,
      "loss": 0.9737,
      "step": 8190
    },
    {
      "epoch": 0.2157229391624967,
      "grad_norm": 1.4362404346466064,
      "learning_rate": 3.922307084540427e-05,
      "loss": 0.3778,
      "step": 8191
    },
    {
      "epoch": 0.21574927574400843,
      "grad_norm": 2.275557041168213,
      "learning_rate": 3.922175401632868e-05,
      "loss": 2.0854,
      "step": 8192
    },
    {
      "epoch": 0.21577561232552014,
      "grad_norm": 1.8016016483306885,
      "learning_rate": 3.9220437187253095e-05,
      "loss": 1.6024,
      "step": 8193
    },
    {
      "epoch": 0.21580194890703186,
      "grad_norm": 4.864360332489014,
      "learning_rate": 3.921912035817751e-05,
      "loss": 1.9912,
      "step": 8194
    },
    {
      "epoch": 0.21582828548854358,
      "grad_norm": 3.0966172218322754,
      "learning_rate": 3.9217803529101926e-05,
      "loss": 1.7076,
      "step": 8195
    },
    {
      "epoch": 0.21585462207005532,
      "grad_norm": 2.613614559173584,
      "learning_rate": 3.921648670002634e-05,
      "loss": 1.8714,
      "step": 8196
    },
    {
      "epoch": 0.21588095865156703,
      "grad_norm": 1.6378625631332397,
      "learning_rate": 3.921516987095075e-05,
      "loss": 1.7882,
      "step": 8197
    },
    {
      "epoch": 0.21590729523307875,
      "grad_norm": 1.9100937843322754,
      "learning_rate": 3.9213853041875166e-05,
      "loss": 1.4412,
      "step": 8198
    },
    {
      "epoch": 0.21593363181459047,
      "grad_norm": 3.474773406982422,
      "learning_rate": 3.921253621279958e-05,
      "loss": 0.6582,
      "step": 8199
    },
    {
      "epoch": 0.21595996839610218,
      "grad_norm": 4.38875150680542,
      "learning_rate": 3.9211219383724e-05,
      "loss": 1.4912,
      "step": 8200
    },
    {
      "epoch": 0.2159863049776139,
      "grad_norm": 1.802634596824646,
      "learning_rate": 3.9209902554648406e-05,
      "loss": 1.9476,
      "step": 8201
    },
    {
      "epoch": 0.2160126415591256,
      "grad_norm": 4.4473676681518555,
      "learning_rate": 3.920858572557282e-05,
      "loss": 0.8807,
      "step": 8202
    },
    {
      "epoch": 0.21603897814063736,
      "grad_norm": 2.7201600074768066,
      "learning_rate": 3.920726889649724e-05,
      "loss": 2.1349,
      "step": 8203
    },
    {
      "epoch": 0.21606531472214907,
      "grad_norm": 1.725846529006958,
      "learning_rate": 3.920595206742165e-05,
      "loss": 2.3868,
      "step": 8204
    },
    {
      "epoch": 0.2160916513036608,
      "grad_norm": 1.6200461387634277,
      "learning_rate": 3.920463523834607e-05,
      "loss": 1.7024,
      "step": 8205
    },
    {
      "epoch": 0.2161179878851725,
      "grad_norm": 3.9600892066955566,
      "learning_rate": 3.920331840927048e-05,
      "loss": 1.2288,
      "step": 8206
    },
    {
      "epoch": 0.21614432446668422,
      "grad_norm": 3.557732105255127,
      "learning_rate": 3.920200158019489e-05,
      "loss": 0.3516,
      "step": 8207
    },
    {
      "epoch": 0.21617066104819593,
      "grad_norm": 1.9008049964904785,
      "learning_rate": 3.92006847511193e-05,
      "loss": 1.6975,
      "step": 8208
    },
    {
      "epoch": 0.21619699762970768,
      "grad_norm": 2.6943721771240234,
      "learning_rate": 3.9199367922043724e-05,
      "loss": 0.8698,
      "step": 8209
    },
    {
      "epoch": 0.2162233342112194,
      "grad_norm": 4.952938556671143,
      "learning_rate": 3.919805109296813e-05,
      "loss": 1.1114,
      "step": 8210
    },
    {
      "epoch": 0.2162496707927311,
      "grad_norm": 1.764625072479248,
      "learning_rate": 3.919673426389255e-05,
      "loss": 1.96,
      "step": 8211
    },
    {
      "epoch": 0.21627600737424282,
      "grad_norm": 2.2616498470306396,
      "learning_rate": 3.9195417434816964e-05,
      "loss": 1.4512,
      "step": 8212
    },
    {
      "epoch": 0.21630234395575454,
      "grad_norm": 3.3392961025238037,
      "learning_rate": 3.919410060574137e-05,
      "loss": 1.8557,
      "step": 8213
    },
    {
      "epoch": 0.21632868053726625,
      "grad_norm": 2.444782018661499,
      "learning_rate": 3.9192783776665796e-05,
      "loss": 2.4076,
      "step": 8214
    },
    {
      "epoch": 0.21635501711877797,
      "grad_norm": 1.9985350370407104,
      "learning_rate": 3.9191466947590204e-05,
      "loss": 1.8849,
      "step": 8215
    },
    {
      "epoch": 0.2163813537002897,
      "grad_norm": 4.963436126708984,
      "learning_rate": 3.919015011851462e-05,
      "loss": 1.7917,
      "step": 8216
    },
    {
      "epoch": 0.21640769028180143,
      "grad_norm": 4.0998101234436035,
      "learning_rate": 3.918883328943903e-05,
      "loss": 1.798,
      "step": 8217
    },
    {
      "epoch": 0.21643402686331314,
      "grad_norm": 1.7398664951324463,
      "learning_rate": 3.918751646036345e-05,
      "loss": 1.3166,
      "step": 8218
    },
    {
      "epoch": 0.21646036344482486,
      "grad_norm": 3.3675451278686523,
      "learning_rate": 3.918619963128786e-05,
      "loss": 1.3902,
      "step": 8219
    },
    {
      "epoch": 0.21648670002633658,
      "grad_norm": 2.7753496170043945,
      "learning_rate": 3.9184882802212276e-05,
      "loss": 1.4967,
      "step": 8220
    },
    {
      "epoch": 0.2165130366078483,
      "grad_norm": 1.997812032699585,
      "learning_rate": 3.9183565973136685e-05,
      "loss": 1.9883,
      "step": 8221
    },
    {
      "epoch": 0.21653937318936003,
      "grad_norm": 3.634472131729126,
      "learning_rate": 3.91822491440611e-05,
      "loss": 1.9869,
      "step": 8222
    },
    {
      "epoch": 0.21656570977087175,
      "grad_norm": 1.9734104871749878,
      "learning_rate": 3.9180932314985516e-05,
      "loss": 1.7723,
      "step": 8223
    },
    {
      "epoch": 0.21659204635238347,
      "grad_norm": 1.8444863557815552,
      "learning_rate": 3.917961548590993e-05,
      "loss": 1.7217,
      "step": 8224
    },
    {
      "epoch": 0.21661838293389518,
      "grad_norm": 5.061600685119629,
      "learning_rate": 3.917829865683435e-05,
      "loss": 1.4748,
      "step": 8225
    },
    {
      "epoch": 0.2166447195154069,
      "grad_norm": 3.497251272201538,
      "learning_rate": 3.9176981827758756e-05,
      "loss": 0.6821,
      "step": 8226
    },
    {
      "epoch": 0.2166710560969186,
      "grad_norm": 3.130624294281006,
      "learning_rate": 3.917566499868317e-05,
      "loss": 0.8868,
      "step": 8227
    },
    {
      "epoch": 0.21669739267843033,
      "grad_norm": 1.7078262567520142,
      "learning_rate": 3.917434816960759e-05,
      "loss": 1.4787,
      "step": 8228
    },
    {
      "epoch": 0.21672372925994207,
      "grad_norm": 2.04734468460083,
      "learning_rate": 3.9173031340532e-05,
      "loss": 0.2967,
      "step": 8229
    },
    {
      "epoch": 0.2167500658414538,
      "grad_norm": 3.667281150817871,
      "learning_rate": 3.917171451145641e-05,
      "loss": 1.3739,
      "step": 8230
    },
    {
      "epoch": 0.2167764024229655,
      "grad_norm": 2.5657331943511963,
      "learning_rate": 3.917039768238083e-05,
      "loss": 1.4296,
      "step": 8231
    },
    {
      "epoch": 0.21680273900447722,
      "grad_norm": 4.715991497039795,
      "learning_rate": 3.916908085330524e-05,
      "loss": 1.5589,
      "step": 8232
    },
    {
      "epoch": 0.21682907558598893,
      "grad_norm": 2.082846164703369,
      "learning_rate": 3.916776402422966e-05,
      "loss": 1.7433,
      "step": 8233
    },
    {
      "epoch": 0.21685541216750065,
      "grad_norm": 4.797408103942871,
      "learning_rate": 3.9166447195154074e-05,
      "loss": 1.0144,
      "step": 8234
    },
    {
      "epoch": 0.21688174874901237,
      "grad_norm": 2.381758451461792,
      "learning_rate": 3.916513036607848e-05,
      "loss": 1.5672,
      "step": 8235
    },
    {
      "epoch": 0.2169080853305241,
      "grad_norm": 1.805557131767273,
      "learning_rate": 3.91638135370029e-05,
      "loss": 1.6703,
      "step": 8236
    },
    {
      "epoch": 0.21693442191203582,
      "grad_norm": 1.748612403869629,
      "learning_rate": 3.9162496707927314e-05,
      "loss": 1.5996,
      "step": 8237
    },
    {
      "epoch": 0.21696075849354754,
      "grad_norm": 5.534695625305176,
      "learning_rate": 3.916117987885173e-05,
      "loss": 1.5567,
      "step": 8238
    },
    {
      "epoch": 0.21698709507505926,
      "grad_norm": 6.371755123138428,
      "learning_rate": 3.915986304977614e-05,
      "loss": 1.2917,
      "step": 8239
    },
    {
      "epoch": 0.21701343165657097,
      "grad_norm": 2.238318681716919,
      "learning_rate": 3.9158546220700554e-05,
      "loss": 0.422,
      "step": 8240
    },
    {
      "epoch": 0.2170397682380827,
      "grad_norm": 2.0366439819335938,
      "learning_rate": 3.915722939162497e-05,
      "loss": 1.8634,
      "step": 8241
    },
    {
      "epoch": 0.21706610481959443,
      "grad_norm": 2.263731002807617,
      "learning_rate": 3.9155912562549385e-05,
      "loss": 1.8592,
      "step": 8242
    },
    {
      "epoch": 0.21709244140110615,
      "grad_norm": 2.038421869277954,
      "learning_rate": 3.91545957334738e-05,
      "loss": 2.0266,
      "step": 8243
    },
    {
      "epoch": 0.21711877798261786,
      "grad_norm": 1.932921290397644,
      "learning_rate": 3.915327890439821e-05,
      "loss": 1.6988,
      "step": 8244
    },
    {
      "epoch": 0.21714511456412958,
      "grad_norm": 1.6675976514816284,
      "learning_rate": 3.9151962075322625e-05,
      "loss": 2.0794,
      "step": 8245
    },
    {
      "epoch": 0.2171714511456413,
      "grad_norm": 1.9884326457977295,
      "learning_rate": 3.9150645246247034e-05,
      "loss": 1.9962,
      "step": 8246
    },
    {
      "epoch": 0.217197787727153,
      "grad_norm": 2.5663902759552,
      "learning_rate": 3.9149328417171456e-05,
      "loss": 1.8525,
      "step": 8247
    },
    {
      "epoch": 0.21722412430866472,
      "grad_norm": 2.4816417694091797,
      "learning_rate": 3.9148011588095865e-05,
      "loss": 1.9044,
      "step": 8248
    },
    {
      "epoch": 0.21725046089017647,
      "grad_norm": 2.0489895343780518,
      "learning_rate": 3.914669475902028e-05,
      "loss": 1.9175,
      "step": 8249
    },
    {
      "epoch": 0.21727679747168818,
      "grad_norm": 3.5776524543762207,
      "learning_rate": 3.9145377929944696e-05,
      "loss": 1.3965,
      "step": 8250
    },
    {
      "epoch": 0.2173031340531999,
      "grad_norm": 2.6768641471862793,
      "learning_rate": 3.914406110086911e-05,
      "loss": 1.0229,
      "step": 8251
    },
    {
      "epoch": 0.2173294706347116,
      "grad_norm": 7.188946723937988,
      "learning_rate": 3.914274427179353e-05,
      "loss": 2.6083,
      "step": 8252
    },
    {
      "epoch": 0.21735580721622333,
      "grad_norm": 1.8915756940841675,
      "learning_rate": 3.9141427442717936e-05,
      "loss": 2.0729,
      "step": 8253
    },
    {
      "epoch": 0.21738214379773504,
      "grad_norm": 3.2362358570098877,
      "learning_rate": 3.914011061364235e-05,
      "loss": 1.3204,
      "step": 8254
    },
    {
      "epoch": 0.21740848037924676,
      "grad_norm": 2.287172555923462,
      "learning_rate": 3.913879378456676e-05,
      "loss": 2.1033,
      "step": 8255
    },
    {
      "epoch": 0.2174348169607585,
      "grad_norm": 1.8927334547042847,
      "learning_rate": 3.913747695549118e-05,
      "loss": 2.0713,
      "step": 8256
    },
    {
      "epoch": 0.21746115354227022,
      "grad_norm": 2.418806314468384,
      "learning_rate": 3.913616012641559e-05,
      "loss": 1.1095,
      "step": 8257
    },
    {
      "epoch": 0.21748749012378193,
      "grad_norm": 3.6046488285064697,
      "learning_rate": 3.913484329734001e-05,
      "loss": 0.8562,
      "step": 8258
    },
    {
      "epoch": 0.21751382670529365,
      "grad_norm": 1.9790934324264526,
      "learning_rate": 3.913352646826442e-05,
      "loss": 2.047,
      "step": 8259
    },
    {
      "epoch": 0.21754016328680537,
      "grad_norm": 2.254774332046509,
      "learning_rate": 3.913220963918883e-05,
      "loss": 1.5768,
      "step": 8260
    },
    {
      "epoch": 0.21756649986831708,
      "grad_norm": 3.8717594146728516,
      "learning_rate": 3.9130892810113254e-05,
      "loss": 1.3509,
      "step": 8261
    },
    {
      "epoch": 0.21759283644982882,
      "grad_norm": 1.9925005435943604,
      "learning_rate": 3.912957598103766e-05,
      "loss": 0.4052,
      "step": 8262
    },
    {
      "epoch": 0.21761917303134054,
      "grad_norm": 1.6390680074691772,
      "learning_rate": 3.912825915196208e-05,
      "loss": 1.64,
      "step": 8263
    },
    {
      "epoch": 0.21764550961285226,
      "grad_norm": 2.836636781692505,
      "learning_rate": 3.912694232288649e-05,
      "loss": 1.927,
      "step": 8264
    },
    {
      "epoch": 0.21767184619436397,
      "grad_norm": 1.683029294013977,
      "learning_rate": 3.912562549381091e-05,
      "loss": 1.6537,
      "step": 8265
    },
    {
      "epoch": 0.2176981827758757,
      "grad_norm": 3.8068337440490723,
      "learning_rate": 3.912430866473532e-05,
      "loss": 1.6014,
      "step": 8266
    },
    {
      "epoch": 0.2177245193573874,
      "grad_norm": 3.7345430850982666,
      "learning_rate": 3.9122991835659734e-05,
      "loss": 1.5089,
      "step": 8267
    },
    {
      "epoch": 0.21775085593889912,
      "grad_norm": 2.826350688934326,
      "learning_rate": 3.912167500658414e-05,
      "loss": 2.228,
      "step": 8268
    },
    {
      "epoch": 0.21777719252041086,
      "grad_norm": 1.8692281246185303,
      "learning_rate": 3.912035817750856e-05,
      "loss": 1.5176,
      "step": 8269
    },
    {
      "epoch": 0.21780352910192258,
      "grad_norm": 1.8478991985321045,
      "learning_rate": 3.9119041348432974e-05,
      "loss": 1.6985,
      "step": 8270
    },
    {
      "epoch": 0.2178298656834343,
      "grad_norm": 2.1795806884765625,
      "learning_rate": 3.911772451935739e-05,
      "loss": 2.0444,
      "step": 8271
    },
    {
      "epoch": 0.217856202264946,
      "grad_norm": 1.4706774950027466,
      "learning_rate": 3.9116407690281805e-05,
      "loss": 1.9017,
      "step": 8272
    },
    {
      "epoch": 0.21788253884645772,
      "grad_norm": 2.129595994949341,
      "learning_rate": 3.9115090861206214e-05,
      "loss": 1.246,
      "step": 8273
    },
    {
      "epoch": 0.21790887542796944,
      "grad_norm": 2.0203781127929688,
      "learning_rate": 3.911377403213063e-05,
      "loss": 1.4599,
      "step": 8274
    },
    {
      "epoch": 0.21793521200948118,
      "grad_norm": 1.4200843572616577,
      "learning_rate": 3.9112457203055045e-05,
      "loss": 1.4933,
      "step": 8275
    },
    {
      "epoch": 0.2179615485909929,
      "grad_norm": 2.4956321716308594,
      "learning_rate": 3.911114037397946e-05,
      "loss": 2.3448,
      "step": 8276
    },
    {
      "epoch": 0.21798788517250461,
      "grad_norm": 3.0443646907806396,
      "learning_rate": 3.910982354490387e-05,
      "loss": 2.114,
      "step": 8277
    },
    {
      "epoch": 0.21801422175401633,
      "grad_norm": 1.8393988609313965,
      "learning_rate": 3.9108506715828286e-05,
      "loss": 2.6578,
      "step": 8278
    },
    {
      "epoch": 0.21804055833552805,
      "grad_norm": 3.3731274604797363,
      "learning_rate": 3.91071898867527e-05,
      "loss": 2.4777,
      "step": 8279
    },
    {
      "epoch": 0.21806689491703976,
      "grad_norm": 2.2712483406066895,
      "learning_rate": 3.910587305767712e-05,
      "loss": 1.9861,
      "step": 8280
    },
    {
      "epoch": 0.21809323149855148,
      "grad_norm": 2.6521594524383545,
      "learning_rate": 3.910455622860153e-05,
      "loss": 1.853,
      "step": 8281
    },
    {
      "epoch": 0.21811956808006322,
      "grad_norm": 3.0817089080810547,
      "learning_rate": 3.910323939952594e-05,
      "loss": 0.445,
      "step": 8282
    },
    {
      "epoch": 0.21814590466157494,
      "grad_norm": 1.8929076194763184,
      "learning_rate": 3.910192257045036e-05,
      "loss": 1.9595,
      "step": 8283
    },
    {
      "epoch": 0.21817224124308665,
      "grad_norm": 5.388201713562012,
      "learning_rate": 3.910060574137477e-05,
      "loss": 2.1167,
      "step": 8284
    },
    {
      "epoch": 0.21819857782459837,
      "grad_norm": 1.7223021984100342,
      "learning_rate": 3.909928891229919e-05,
      "loss": 1.076,
      "step": 8285
    },
    {
      "epoch": 0.21822491440611008,
      "grad_norm": 3.4151270389556885,
      "learning_rate": 3.90979720832236e-05,
      "loss": 2.0,
      "step": 8286
    },
    {
      "epoch": 0.2182512509876218,
      "grad_norm": 2.442265510559082,
      "learning_rate": 3.909665525414801e-05,
      "loss": 1.4741,
      "step": 8287
    },
    {
      "epoch": 0.2182775875691335,
      "grad_norm": 2.2576863765716553,
      "learning_rate": 3.909533842507243e-05,
      "loss": 2.2601,
      "step": 8288
    },
    {
      "epoch": 0.21830392415064526,
      "grad_norm": 2.8746497631073,
      "learning_rate": 3.9094021595996844e-05,
      "loss": 0.8524,
      "step": 8289
    },
    {
      "epoch": 0.21833026073215697,
      "grad_norm": 2.574685573577881,
      "learning_rate": 3.909270476692126e-05,
      "loss": 1.7461,
      "step": 8290
    },
    {
      "epoch": 0.2183565973136687,
      "grad_norm": 1.8106601238250732,
      "learning_rate": 3.909138793784567e-05,
      "loss": 2.0349,
      "step": 8291
    },
    {
      "epoch": 0.2183829338951804,
      "grad_norm": 1.9987387657165527,
      "learning_rate": 3.9090071108770084e-05,
      "loss": 2.3239,
      "step": 8292
    },
    {
      "epoch": 0.21840927047669212,
      "grad_norm": 5.2676801681518555,
      "learning_rate": 3.908875427969449e-05,
      "loss": 1.3818,
      "step": 8293
    },
    {
      "epoch": 0.21843560705820383,
      "grad_norm": 3.1931509971618652,
      "learning_rate": 3.9087437450618915e-05,
      "loss": 1.4199,
      "step": 8294
    },
    {
      "epoch": 0.21846194363971558,
      "grad_norm": 4.456040382385254,
      "learning_rate": 3.9086120621543324e-05,
      "loss": 2.0105,
      "step": 8295
    },
    {
      "epoch": 0.2184882802212273,
      "grad_norm": 2.5653867721557617,
      "learning_rate": 3.908480379246774e-05,
      "loss": 1.6507,
      "step": 8296
    },
    {
      "epoch": 0.218514616802739,
      "grad_norm": 5.016998767852783,
      "learning_rate": 3.9083486963392155e-05,
      "loss": 1.202,
      "step": 8297
    },
    {
      "epoch": 0.21854095338425072,
      "grad_norm": 1.8235292434692383,
      "learning_rate": 3.908217013431657e-05,
      "loss": 0.6868,
      "step": 8298
    },
    {
      "epoch": 0.21856728996576244,
      "grad_norm": 2.589324951171875,
      "learning_rate": 3.9080853305240986e-05,
      "loss": 1.5412,
      "step": 8299
    },
    {
      "epoch": 0.21859362654727416,
      "grad_norm": 1.9480090141296387,
      "learning_rate": 3.9079536476165395e-05,
      "loss": 1.5285,
      "step": 8300
    },
    {
      "epoch": 0.21861996312878587,
      "grad_norm": 3.0515410900115967,
      "learning_rate": 3.907821964708981e-05,
      "loss": 1.7817,
      "step": 8301
    },
    {
      "epoch": 0.21864629971029761,
      "grad_norm": 2.4323890209198,
      "learning_rate": 3.907690281801422e-05,
      "loss": 1.974,
      "step": 8302
    },
    {
      "epoch": 0.21867263629180933,
      "grad_norm": 2.8883697986602783,
      "learning_rate": 3.907558598893864e-05,
      "loss": 1.6774,
      "step": 8303
    },
    {
      "epoch": 0.21869897287332105,
      "grad_norm": 3.349668264389038,
      "learning_rate": 3.907426915986305e-05,
      "loss": 1.3268,
      "step": 8304
    },
    {
      "epoch": 0.21872530945483276,
      "grad_norm": 2.9275786876678467,
      "learning_rate": 3.9072952330787466e-05,
      "loss": 2.3916,
      "step": 8305
    },
    {
      "epoch": 0.21875164603634448,
      "grad_norm": 2.2405154705047607,
      "learning_rate": 3.907163550171188e-05,
      "loss": 1.4918,
      "step": 8306
    },
    {
      "epoch": 0.2187779826178562,
      "grad_norm": 2.03804087638855,
      "learning_rate": 3.907031867263629e-05,
      "loss": 1.431,
      "step": 8307
    },
    {
      "epoch": 0.2188043191993679,
      "grad_norm": 2.5511977672576904,
      "learning_rate": 3.906900184356071e-05,
      "loss": 1.6714,
      "step": 8308
    },
    {
      "epoch": 0.21883065578087965,
      "grad_norm": 1.6598297357559204,
      "learning_rate": 3.906768501448512e-05,
      "loss": 1.7592,
      "step": 8309
    },
    {
      "epoch": 0.21885699236239137,
      "grad_norm": 2.0941429138183594,
      "learning_rate": 3.906636818540954e-05,
      "loss": 1.936,
      "step": 8310
    },
    {
      "epoch": 0.21888332894390308,
      "grad_norm": 2.101958990097046,
      "learning_rate": 3.9065051356333946e-05,
      "loss": 2.4416,
      "step": 8311
    },
    {
      "epoch": 0.2189096655254148,
      "grad_norm": 2.879706859588623,
      "learning_rate": 3.906373452725836e-05,
      "loss": 1.453,
      "step": 8312
    },
    {
      "epoch": 0.2189360021069265,
      "grad_norm": 5.836144924163818,
      "learning_rate": 3.906241769818278e-05,
      "loss": 0.6958,
      "step": 8313
    },
    {
      "epoch": 0.21896233868843823,
      "grad_norm": 2.725416660308838,
      "learning_rate": 3.906110086910719e-05,
      "loss": 2.085,
      "step": 8314
    },
    {
      "epoch": 0.21898867526994997,
      "grad_norm": 4.340921878814697,
      "learning_rate": 3.905978404003161e-05,
      "loss": 1.7496,
      "step": 8315
    },
    {
      "epoch": 0.2190150118514617,
      "grad_norm": 2.4558308124542236,
      "learning_rate": 3.905846721095602e-05,
      "loss": 1.9737,
      "step": 8316
    },
    {
      "epoch": 0.2190413484329734,
      "grad_norm": 2.857482671737671,
      "learning_rate": 3.905715038188044e-05,
      "loss": 1.9177,
      "step": 8317
    },
    {
      "epoch": 0.21906768501448512,
      "grad_norm": 2.1477527618408203,
      "learning_rate": 3.905583355280485e-05,
      "loss": 1.4247,
      "step": 8318
    },
    {
      "epoch": 0.21909402159599684,
      "grad_norm": 3.0266008377075195,
      "learning_rate": 3.9054516723729264e-05,
      "loss": 1.608,
      "step": 8319
    },
    {
      "epoch": 0.21912035817750855,
      "grad_norm": 7.984438896179199,
      "learning_rate": 3.905319989465367e-05,
      "loss": 1.6013,
      "step": 8320
    },
    {
      "epoch": 0.21914669475902027,
      "grad_norm": 3.805899143218994,
      "learning_rate": 3.905188306557809e-05,
      "loss": 0.1864,
      "step": 8321
    },
    {
      "epoch": 0.219173031340532,
      "grad_norm": 2.510054588317871,
      "learning_rate": 3.9050566236502504e-05,
      "loss": 1.7109,
      "step": 8322
    },
    {
      "epoch": 0.21919936792204373,
      "grad_norm": 2.0700831413269043,
      "learning_rate": 3.904924940742692e-05,
      "loss": 1.0419,
      "step": 8323
    },
    {
      "epoch": 0.21922570450355544,
      "grad_norm": 2.5530920028686523,
      "learning_rate": 3.904793257835133e-05,
      "loss": 1.6875,
      "step": 8324
    },
    {
      "epoch": 0.21925204108506716,
      "grad_norm": 2.2969350814819336,
      "learning_rate": 3.9046615749275744e-05,
      "loss": 1.2661,
      "step": 8325
    },
    {
      "epoch": 0.21927837766657887,
      "grad_norm": 3.7111880779266357,
      "learning_rate": 3.904529892020016e-05,
      "loss": 0.6292,
      "step": 8326
    },
    {
      "epoch": 0.2193047142480906,
      "grad_norm": 5.629997730255127,
      "learning_rate": 3.9043982091124575e-05,
      "loss": 1.0741,
      "step": 8327
    },
    {
      "epoch": 0.21933105082960233,
      "grad_norm": 1.9615687131881714,
      "learning_rate": 3.904266526204899e-05,
      "loss": 1.7914,
      "step": 8328
    },
    {
      "epoch": 0.21935738741111405,
      "grad_norm": 2.299870491027832,
      "learning_rate": 3.90413484329734e-05,
      "loss": 1.5477,
      "step": 8329
    },
    {
      "epoch": 0.21938372399262576,
      "grad_norm": 2.5918002128601074,
      "learning_rate": 3.9040031603897815e-05,
      "loss": 2.2188,
      "step": 8330
    },
    {
      "epoch": 0.21941006057413748,
      "grad_norm": 2.9231417179107666,
      "learning_rate": 3.903871477482223e-05,
      "loss": 1.2888,
      "step": 8331
    },
    {
      "epoch": 0.2194363971556492,
      "grad_norm": 2.1001057624816895,
      "learning_rate": 3.9037397945746646e-05,
      "loss": 1.4899,
      "step": 8332
    },
    {
      "epoch": 0.2194627337371609,
      "grad_norm": 1.4785897731781006,
      "learning_rate": 3.9036081116671055e-05,
      "loss": 2.1613,
      "step": 8333
    },
    {
      "epoch": 0.21948907031867262,
      "grad_norm": 1.7580170631408691,
      "learning_rate": 3.903476428759547e-05,
      "loss": 1.9895,
      "step": 8334
    },
    {
      "epoch": 0.21951540690018437,
      "grad_norm": 3.9516351222991943,
      "learning_rate": 3.9033447458519886e-05,
      "loss": 1.3631,
      "step": 8335
    },
    {
      "epoch": 0.21954174348169608,
      "grad_norm": 2.0070955753326416,
      "learning_rate": 3.90321306294443e-05,
      "loss": 2.1751,
      "step": 8336
    },
    {
      "epoch": 0.2195680800632078,
      "grad_norm": 1.9762344360351562,
      "learning_rate": 3.903081380036872e-05,
      "loss": 1.8103,
      "step": 8337
    },
    {
      "epoch": 0.21959441664471951,
      "grad_norm": 2.8037970066070557,
      "learning_rate": 3.9029496971293126e-05,
      "loss": 2.0264,
      "step": 8338
    },
    {
      "epoch": 0.21962075322623123,
      "grad_norm": 1.8411526679992676,
      "learning_rate": 3.902818014221754e-05,
      "loss": 1.6505,
      "step": 8339
    },
    {
      "epoch": 0.21964708980774295,
      "grad_norm": 1.9780889749526978,
      "learning_rate": 3.902686331314195e-05,
      "loss": 1.597,
      "step": 8340
    },
    {
      "epoch": 0.21967342638925466,
      "grad_norm": 2.954498052597046,
      "learning_rate": 3.902554648406637e-05,
      "loss": 1.4721,
      "step": 8341
    },
    {
      "epoch": 0.2196997629707664,
      "grad_norm": 2.501983404159546,
      "learning_rate": 3.902422965499078e-05,
      "loss": 0.8074,
      "step": 8342
    },
    {
      "epoch": 0.21972609955227812,
      "grad_norm": 1.7199556827545166,
      "learning_rate": 3.90229128259152e-05,
      "loss": 2.0235,
      "step": 8343
    },
    {
      "epoch": 0.21975243613378984,
      "grad_norm": 2.8954038619995117,
      "learning_rate": 3.902159599683961e-05,
      "loss": 1.326,
      "step": 8344
    },
    {
      "epoch": 0.21977877271530155,
      "grad_norm": 4.732748031616211,
      "learning_rate": 3.902027916776402e-05,
      "loss": 1.4896,
      "step": 8345
    },
    {
      "epoch": 0.21980510929681327,
      "grad_norm": 3.54403018951416,
      "learning_rate": 3.9018962338688445e-05,
      "loss": 1.3631,
      "step": 8346
    },
    {
      "epoch": 0.21983144587832498,
      "grad_norm": 1.9328362941741943,
      "learning_rate": 3.901764550961285e-05,
      "loss": 1.6679,
      "step": 8347
    },
    {
      "epoch": 0.21985778245983673,
      "grad_norm": 1.5179873704910278,
      "learning_rate": 3.901632868053727e-05,
      "loss": 1.3858,
      "step": 8348
    },
    {
      "epoch": 0.21988411904134844,
      "grad_norm": 1.7189298868179321,
      "learning_rate": 3.901501185146168e-05,
      "loss": 1.5546,
      "step": 8349
    },
    {
      "epoch": 0.21991045562286016,
      "grad_norm": 1.4995951652526855,
      "learning_rate": 3.90136950223861e-05,
      "loss": 1.4979,
      "step": 8350
    },
    {
      "epoch": 0.21993679220437187,
      "grad_norm": 5.189961910247803,
      "learning_rate": 3.901237819331051e-05,
      "loss": 2.0618,
      "step": 8351
    },
    {
      "epoch": 0.2199631287858836,
      "grad_norm": 1.784239411354065,
      "learning_rate": 3.9011061364234925e-05,
      "loss": 2.2262,
      "step": 8352
    },
    {
      "epoch": 0.2199894653673953,
      "grad_norm": 1.5889970064163208,
      "learning_rate": 3.900974453515934e-05,
      "loss": 1.281,
      "step": 8353
    },
    {
      "epoch": 0.22001580194890702,
      "grad_norm": 2.6087942123413086,
      "learning_rate": 3.900842770608375e-05,
      "loss": 1.6153,
      "step": 8354
    },
    {
      "epoch": 0.22004213853041876,
      "grad_norm": 1.5636043548583984,
      "learning_rate": 3.900711087700817e-05,
      "loss": 1.7055,
      "step": 8355
    },
    {
      "epoch": 0.22006847511193048,
      "grad_norm": 2.0564708709716797,
      "learning_rate": 3.900579404793258e-05,
      "loss": 1.8526,
      "step": 8356
    },
    {
      "epoch": 0.2200948116934422,
      "grad_norm": 2.7637977600097656,
      "learning_rate": 3.9004477218856996e-05,
      "loss": 1.9168,
      "step": 8357
    },
    {
      "epoch": 0.2201211482749539,
      "grad_norm": 1.9751542806625366,
      "learning_rate": 3.9003160389781405e-05,
      "loss": 0.7221,
      "step": 8358
    },
    {
      "epoch": 0.22014748485646563,
      "grad_norm": 2.228431224822998,
      "learning_rate": 3.900184356070582e-05,
      "loss": 2.3343,
      "step": 8359
    },
    {
      "epoch": 0.22017382143797734,
      "grad_norm": 4.339964866638184,
      "learning_rate": 3.9000526731630236e-05,
      "loss": 1.872,
      "step": 8360
    },
    {
      "epoch": 0.22020015801948908,
      "grad_norm": 2.463357448577881,
      "learning_rate": 3.899920990255465e-05,
      "loss": 1.0659,
      "step": 8361
    },
    {
      "epoch": 0.2202264946010008,
      "grad_norm": 4.519904136657715,
      "learning_rate": 3.899789307347907e-05,
      "loss": 1.8219,
      "step": 8362
    },
    {
      "epoch": 0.22025283118251252,
      "grad_norm": 3.036019802093506,
      "learning_rate": 3.8996576244403476e-05,
      "loss": 1.8034,
      "step": 8363
    },
    {
      "epoch": 0.22027916776402423,
      "grad_norm": 3.402838945388794,
      "learning_rate": 3.89952594153279e-05,
      "loss": 1.917,
      "step": 8364
    },
    {
      "epoch": 0.22030550434553595,
      "grad_norm": 2.916759490966797,
      "learning_rate": 3.899394258625231e-05,
      "loss": 1.6904,
      "step": 8365
    },
    {
      "epoch": 0.22033184092704766,
      "grad_norm": 1.6138890981674194,
      "learning_rate": 3.899262575717672e-05,
      "loss": 1.7718,
      "step": 8366
    },
    {
      "epoch": 0.22035817750855938,
      "grad_norm": 5.818990230560303,
      "learning_rate": 3.899130892810113e-05,
      "loss": 1.5933,
      "step": 8367
    },
    {
      "epoch": 0.22038451409007112,
      "grad_norm": 2.0275235176086426,
      "learning_rate": 3.898999209902555e-05,
      "loss": 1.7,
      "step": 8368
    },
    {
      "epoch": 0.22041085067158284,
      "grad_norm": 2.8279902935028076,
      "learning_rate": 3.898867526994996e-05,
      "loss": 1.817,
      "step": 8369
    },
    {
      "epoch": 0.22043718725309455,
      "grad_norm": 2.10720157623291,
      "learning_rate": 3.898735844087438e-05,
      "loss": 1.232,
      "step": 8370
    },
    {
      "epoch": 0.22046352383460627,
      "grad_norm": 2.1118297576904297,
      "learning_rate": 3.898604161179879e-05,
      "loss": 1.6767,
      "step": 8371
    },
    {
      "epoch": 0.22048986041611798,
      "grad_norm": 3.26546049118042,
      "learning_rate": 3.89847247827232e-05,
      "loss": 1.4664,
      "step": 8372
    },
    {
      "epoch": 0.2205161969976297,
      "grad_norm": 1.9262683391571045,
      "learning_rate": 3.898340795364762e-05,
      "loss": 0.7762,
      "step": 8373
    },
    {
      "epoch": 0.22054253357914141,
      "grad_norm": 3.0245628356933594,
      "learning_rate": 3.8982091124572034e-05,
      "loss": 1.4516,
      "step": 8374
    },
    {
      "epoch": 0.22056887016065316,
      "grad_norm": 1.8919014930725098,
      "learning_rate": 3.898077429549645e-05,
      "loss": 0.6358,
      "step": 8375
    },
    {
      "epoch": 0.22059520674216487,
      "grad_norm": 3.9048259258270264,
      "learning_rate": 3.897945746642086e-05,
      "loss": 0.7696,
      "step": 8376
    },
    {
      "epoch": 0.2206215433236766,
      "grad_norm": 2.440701961517334,
      "learning_rate": 3.8978140637345274e-05,
      "loss": 0.4367,
      "step": 8377
    },
    {
      "epoch": 0.2206478799051883,
      "grad_norm": 2.389115571975708,
      "learning_rate": 3.897682380826968e-05,
      "loss": 1.5765,
      "step": 8378
    },
    {
      "epoch": 0.22067421648670002,
      "grad_norm": 2.5260136127471924,
      "learning_rate": 3.8975506979194105e-05,
      "loss": 1.9794,
      "step": 8379
    },
    {
      "epoch": 0.22070055306821174,
      "grad_norm": 2.118802785873413,
      "learning_rate": 3.8974190150118514e-05,
      "loss": 1.6562,
      "step": 8380
    },
    {
      "epoch": 0.22072688964972348,
      "grad_norm": 2.1320691108703613,
      "learning_rate": 3.897287332104293e-05,
      "loss": 1.4441,
      "step": 8381
    },
    {
      "epoch": 0.2207532262312352,
      "grad_norm": 1.8724281787872314,
      "learning_rate": 3.8971556491967345e-05,
      "loss": 2.1027,
      "step": 8382
    },
    {
      "epoch": 0.2207795628127469,
      "grad_norm": 1.9332189559936523,
      "learning_rate": 3.897023966289176e-05,
      "loss": 2.0203,
      "step": 8383
    },
    {
      "epoch": 0.22080589939425863,
      "grad_norm": 1.377746343612671,
      "learning_rate": 3.8968922833816176e-05,
      "loss": 0.9216,
      "step": 8384
    },
    {
      "epoch": 0.22083223597577034,
      "grad_norm": 1.7230174541473389,
      "learning_rate": 3.8967606004740585e-05,
      "loss": 1.5728,
      "step": 8385
    },
    {
      "epoch": 0.22085857255728206,
      "grad_norm": 3.3916029930114746,
      "learning_rate": 3.8966289175665e-05,
      "loss": 1.4796,
      "step": 8386
    },
    {
      "epoch": 0.22088490913879377,
      "grad_norm": 1.917340636253357,
      "learning_rate": 3.896497234658941e-05,
      "loss": 1.8327,
      "step": 8387
    },
    {
      "epoch": 0.22091124572030552,
      "grad_norm": 1.668746829032898,
      "learning_rate": 3.896365551751383e-05,
      "loss": 2.0111,
      "step": 8388
    },
    {
      "epoch": 0.22093758230181723,
      "grad_norm": 1.511021614074707,
      "learning_rate": 3.896233868843824e-05,
      "loss": 1.7179,
      "step": 8389
    },
    {
      "epoch": 0.22096391888332895,
      "grad_norm": 2.1119842529296875,
      "learning_rate": 3.8961021859362656e-05,
      "loss": 2.7975,
      "step": 8390
    },
    {
      "epoch": 0.22099025546484066,
      "grad_norm": 1.987365961074829,
      "learning_rate": 3.895970503028707e-05,
      "loss": 1.3163,
      "step": 8391
    },
    {
      "epoch": 0.22101659204635238,
      "grad_norm": 1.7661840915679932,
      "learning_rate": 3.895838820121148e-05,
      "loss": 2.0107,
      "step": 8392
    },
    {
      "epoch": 0.2210429286278641,
      "grad_norm": 1.6379493474960327,
      "learning_rate": 3.89570713721359e-05,
      "loss": 1.8292,
      "step": 8393
    },
    {
      "epoch": 0.2210692652093758,
      "grad_norm": 4.291928768157959,
      "learning_rate": 3.895575454306031e-05,
      "loss": 1.6041,
      "step": 8394
    },
    {
      "epoch": 0.22109560179088755,
      "grad_norm": 2.297240972518921,
      "learning_rate": 3.895443771398473e-05,
      "loss": 1.5278,
      "step": 8395
    },
    {
      "epoch": 0.22112193837239927,
      "grad_norm": 1.8344429731369019,
      "learning_rate": 3.8953120884909136e-05,
      "loss": 1.9794,
      "step": 8396
    },
    {
      "epoch": 0.22114827495391098,
      "grad_norm": 3.064854860305786,
      "learning_rate": 3.895180405583356e-05,
      "loss": 2.056,
      "step": 8397
    },
    {
      "epoch": 0.2211746115354227,
      "grad_norm": 1.4964947700500488,
      "learning_rate": 3.895048722675797e-05,
      "loss": 2.024,
      "step": 8398
    },
    {
      "epoch": 0.22120094811693441,
      "grad_norm": 1.6163251399993896,
      "learning_rate": 3.894917039768238e-05,
      "loss": 1.2892,
      "step": 8399
    },
    {
      "epoch": 0.22122728469844613,
      "grad_norm": 2.7338666915893555,
      "learning_rate": 3.89478535686068e-05,
      "loss": 2.1958,
      "step": 8400
    },
    {
      "epoch": 0.22125362127995787,
      "grad_norm": 1.7635051012039185,
      "learning_rate": 3.894653673953121e-05,
      "loss": 2.1056,
      "step": 8401
    },
    {
      "epoch": 0.2212799578614696,
      "grad_norm": 2.8649826049804688,
      "learning_rate": 3.894521991045563e-05,
      "loss": 1.349,
      "step": 8402
    },
    {
      "epoch": 0.2213062944429813,
      "grad_norm": 5.887941837310791,
      "learning_rate": 3.894390308138004e-05,
      "loss": 1.3976,
      "step": 8403
    },
    {
      "epoch": 0.22133263102449302,
      "grad_norm": 4.322699069976807,
      "learning_rate": 3.8942586252304454e-05,
      "loss": 1.8634,
      "step": 8404
    },
    {
      "epoch": 0.22135896760600474,
      "grad_norm": 1.8168007135391235,
      "learning_rate": 3.894126942322886e-05,
      "loss": 1.2342,
      "step": 8405
    },
    {
      "epoch": 0.22138530418751645,
      "grad_norm": 2.0172464847564697,
      "learning_rate": 3.893995259415328e-05,
      "loss": 1.347,
      "step": 8406
    },
    {
      "epoch": 0.22141164076902817,
      "grad_norm": 1.9244657754898071,
      "learning_rate": 3.8938635765077694e-05,
      "loss": 1.7271,
      "step": 8407
    },
    {
      "epoch": 0.2214379773505399,
      "grad_norm": 1.7695342302322388,
      "learning_rate": 3.893731893600211e-05,
      "loss": 2.49,
      "step": 8408
    },
    {
      "epoch": 0.22146431393205163,
      "grad_norm": 2.9757330417633057,
      "learning_rate": 3.8936002106926526e-05,
      "loss": 1.6874,
      "step": 8409
    },
    {
      "epoch": 0.22149065051356334,
      "grad_norm": 1.7494467496871948,
      "learning_rate": 3.8934685277850934e-05,
      "loss": 2.6485,
      "step": 8410
    },
    {
      "epoch": 0.22151698709507506,
      "grad_norm": 2.0662641525268555,
      "learning_rate": 3.893336844877535e-05,
      "loss": 1.6856,
      "step": 8411
    },
    {
      "epoch": 0.22154332367658677,
      "grad_norm": 2.0488367080688477,
      "learning_rate": 3.8932051619699766e-05,
      "loss": 2.3371,
      "step": 8412
    },
    {
      "epoch": 0.2215696602580985,
      "grad_norm": 2.8280792236328125,
      "learning_rate": 3.893073479062418e-05,
      "loss": 2.217,
      "step": 8413
    },
    {
      "epoch": 0.22159599683961023,
      "grad_norm": 4.707918643951416,
      "learning_rate": 3.892941796154859e-05,
      "loss": 1.7576,
      "step": 8414
    },
    {
      "epoch": 0.22162233342112195,
      "grad_norm": 1.8562642335891724,
      "learning_rate": 3.8928101132473006e-05,
      "loss": 2.2244,
      "step": 8415
    },
    {
      "epoch": 0.22164867000263366,
      "grad_norm": 2.3995871543884277,
      "learning_rate": 3.892678430339742e-05,
      "loss": 1.3258,
      "step": 8416
    },
    {
      "epoch": 0.22167500658414538,
      "grad_norm": 2.1873204708099365,
      "learning_rate": 3.892546747432184e-05,
      "loss": 1.9068,
      "step": 8417
    },
    {
      "epoch": 0.2217013431656571,
      "grad_norm": 1.8489776849746704,
      "learning_rate": 3.892415064524625e-05,
      "loss": 1.674,
      "step": 8418
    },
    {
      "epoch": 0.2217276797471688,
      "grad_norm": 4.264925479888916,
      "learning_rate": 3.892283381617066e-05,
      "loss": 0.8233,
      "step": 8419
    },
    {
      "epoch": 0.22175401632868053,
      "grad_norm": 1.5581228733062744,
      "learning_rate": 3.892151698709508e-05,
      "loss": 2.2477,
      "step": 8420
    },
    {
      "epoch": 0.22178035291019227,
      "grad_norm": 2.964524269104004,
      "learning_rate": 3.892020015801949e-05,
      "loss": 1.5558,
      "step": 8421
    },
    {
      "epoch": 0.22180668949170398,
      "grad_norm": 2.144294023513794,
      "learning_rate": 3.891888332894391e-05,
      "loss": 1.9218,
      "step": 8422
    },
    {
      "epoch": 0.2218330260732157,
      "grad_norm": 3.3459372520446777,
      "learning_rate": 3.891756649986832e-05,
      "loss": 1.964,
      "step": 8423
    },
    {
      "epoch": 0.22185936265472742,
      "grad_norm": 2.1398229598999023,
      "learning_rate": 3.891624967079273e-05,
      "loss": 0.6664,
      "step": 8424
    },
    {
      "epoch": 0.22188569923623913,
      "grad_norm": 4.393989562988281,
      "learning_rate": 3.891493284171714e-05,
      "loss": 1.631,
      "step": 8425
    },
    {
      "epoch": 0.22191203581775085,
      "grad_norm": 2.6729469299316406,
      "learning_rate": 3.8913616012641564e-05,
      "loss": 2.2048,
      "step": 8426
    },
    {
      "epoch": 0.22193837239926256,
      "grad_norm": 2.602407693862915,
      "learning_rate": 3.891229918356597e-05,
      "loss": 1.3373,
      "step": 8427
    },
    {
      "epoch": 0.2219647089807743,
      "grad_norm": 5.29533052444458,
      "learning_rate": 3.891098235449039e-05,
      "loss": 1.7386,
      "step": 8428
    },
    {
      "epoch": 0.22199104556228602,
      "grad_norm": 3.888007402420044,
      "learning_rate": 3.8909665525414804e-05,
      "loss": 0.9082,
      "step": 8429
    },
    {
      "epoch": 0.22201738214379774,
      "grad_norm": 1.585597038269043,
      "learning_rate": 3.890834869633922e-05,
      "loss": 1.7358,
      "step": 8430
    },
    {
      "epoch": 0.22204371872530945,
      "grad_norm": 1.3620662689208984,
      "learning_rate": 3.8907031867263635e-05,
      "loss": 1.7565,
      "step": 8431
    },
    {
      "epoch": 0.22207005530682117,
      "grad_norm": 1.9046671390533447,
      "learning_rate": 3.8905715038188044e-05,
      "loss": 1.4714,
      "step": 8432
    },
    {
      "epoch": 0.22209639188833288,
      "grad_norm": 1.5443295240402222,
      "learning_rate": 3.890439820911246e-05,
      "loss": 2.1616,
      "step": 8433
    },
    {
      "epoch": 0.22212272846984463,
      "grad_norm": 1.9211442470550537,
      "learning_rate": 3.890308138003687e-05,
      "loss": 1.8194,
      "step": 8434
    },
    {
      "epoch": 0.22214906505135634,
      "grad_norm": 3.39355731010437,
      "learning_rate": 3.890176455096129e-05,
      "loss": 1.2593,
      "step": 8435
    },
    {
      "epoch": 0.22217540163286806,
      "grad_norm": 2.3450522422790527,
      "learning_rate": 3.89004477218857e-05,
      "loss": 1.9045,
      "step": 8436
    },
    {
      "epoch": 0.22220173821437977,
      "grad_norm": 1.8722589015960693,
      "learning_rate": 3.8899130892810115e-05,
      "loss": 1.7518,
      "step": 8437
    },
    {
      "epoch": 0.2222280747958915,
      "grad_norm": 4.835537433624268,
      "learning_rate": 3.889781406373453e-05,
      "loss": 2.1612,
      "step": 8438
    },
    {
      "epoch": 0.2222544113774032,
      "grad_norm": 3.000201463699341,
      "learning_rate": 3.889649723465894e-05,
      "loss": 2.1411,
      "step": 8439
    },
    {
      "epoch": 0.22228074795891492,
      "grad_norm": 1.4438332319259644,
      "learning_rate": 3.889518040558336e-05,
      "loss": 1.7441,
      "step": 8440
    },
    {
      "epoch": 0.22230708454042666,
      "grad_norm": 2.154694080352783,
      "learning_rate": 3.889386357650777e-05,
      "loss": 2.3642,
      "step": 8441
    },
    {
      "epoch": 0.22233342112193838,
      "grad_norm": 1.914270281791687,
      "learning_rate": 3.8892546747432186e-05,
      "loss": 1.8811,
      "step": 8442
    },
    {
      "epoch": 0.2223597577034501,
      "grad_norm": 2.5694048404693604,
      "learning_rate": 3.8891229918356595e-05,
      "loss": 1.3285,
      "step": 8443
    },
    {
      "epoch": 0.2223860942849618,
      "grad_norm": 2.4416675567626953,
      "learning_rate": 3.888991308928102e-05,
      "loss": 1.613,
      "step": 8444
    },
    {
      "epoch": 0.22241243086647353,
      "grad_norm": 1.5371558666229248,
      "learning_rate": 3.8888596260205426e-05,
      "loss": 1.4086,
      "step": 8445
    },
    {
      "epoch": 0.22243876744798524,
      "grad_norm": 1.6044889688491821,
      "learning_rate": 3.888727943112984e-05,
      "loss": 1.6043,
      "step": 8446
    },
    {
      "epoch": 0.22246510402949699,
      "grad_norm": 2.9604430198669434,
      "learning_rate": 3.888596260205426e-05,
      "loss": 2.1392,
      "step": 8447
    },
    {
      "epoch": 0.2224914406110087,
      "grad_norm": 2.8567779064178467,
      "learning_rate": 3.8884645772978666e-05,
      "loss": 2.2244,
      "step": 8448
    },
    {
      "epoch": 0.22251777719252042,
      "grad_norm": 2.0942060947418213,
      "learning_rate": 3.888332894390309e-05,
      "loss": 1.9639,
      "step": 8449
    },
    {
      "epoch": 0.22254411377403213,
      "grad_norm": 2.455822706222534,
      "learning_rate": 3.88820121148275e-05,
      "loss": 1.7601,
      "step": 8450
    },
    {
      "epoch": 0.22257045035554385,
      "grad_norm": 3.185560941696167,
      "learning_rate": 3.888069528575191e-05,
      "loss": 1.8894,
      "step": 8451
    },
    {
      "epoch": 0.22259678693705556,
      "grad_norm": 2.9746787548065186,
      "learning_rate": 3.887937845667632e-05,
      "loss": 2.138,
      "step": 8452
    },
    {
      "epoch": 0.22262312351856728,
      "grad_norm": 2.757542610168457,
      "learning_rate": 3.887806162760074e-05,
      "loss": 1.5617,
      "step": 8453
    },
    {
      "epoch": 0.22264946010007902,
      "grad_norm": 3.7855117321014404,
      "learning_rate": 3.887674479852515e-05,
      "loss": 1.6345,
      "step": 8454
    },
    {
      "epoch": 0.22267579668159074,
      "grad_norm": 1.7267675399780273,
      "learning_rate": 3.887542796944957e-05,
      "loss": 1.8609,
      "step": 8455
    },
    {
      "epoch": 0.22270213326310245,
      "grad_norm": 1.4913760423660278,
      "learning_rate": 3.8874111140373984e-05,
      "loss": 1.6093,
      "step": 8456
    },
    {
      "epoch": 0.22272846984461417,
      "grad_norm": 3.7315099239349365,
      "learning_rate": 3.887279431129839e-05,
      "loss": 1.4499,
      "step": 8457
    },
    {
      "epoch": 0.22275480642612588,
      "grad_norm": 2.175516366958618,
      "learning_rate": 3.887147748222281e-05,
      "loss": 1.322,
      "step": 8458
    },
    {
      "epoch": 0.2227811430076376,
      "grad_norm": 3.672293186187744,
      "learning_rate": 3.8870160653147224e-05,
      "loss": 1.78,
      "step": 8459
    },
    {
      "epoch": 0.22280747958914932,
      "grad_norm": 3.357968807220459,
      "learning_rate": 3.886884382407164e-05,
      "loss": 1.387,
      "step": 8460
    },
    {
      "epoch": 0.22283381617066106,
      "grad_norm": 1.8350592851638794,
      "learning_rate": 3.886752699499605e-05,
      "loss": 1.474,
      "step": 8461
    },
    {
      "epoch": 0.22286015275217277,
      "grad_norm": 2.6834311485290527,
      "learning_rate": 3.8866210165920464e-05,
      "loss": 1.1394,
      "step": 8462
    },
    {
      "epoch": 0.2228864893336845,
      "grad_norm": 3.189242124557495,
      "learning_rate": 3.886489333684488e-05,
      "loss": 1.065,
      "step": 8463
    },
    {
      "epoch": 0.2229128259151962,
      "grad_norm": 2.0183005332946777,
      "learning_rate": 3.8863576507769295e-05,
      "loss": 1.9932,
      "step": 8464
    },
    {
      "epoch": 0.22293916249670792,
      "grad_norm": 3.5083160400390625,
      "learning_rate": 3.886225967869371e-05,
      "loss": 2.7922,
      "step": 8465
    },
    {
      "epoch": 0.22296549907821964,
      "grad_norm": 2.4788997173309326,
      "learning_rate": 3.886094284961812e-05,
      "loss": 1.4043,
      "step": 8466
    },
    {
      "epoch": 0.22299183565973138,
      "grad_norm": 2.37233829498291,
      "learning_rate": 3.8859626020542535e-05,
      "loss": 0.5077,
      "step": 8467
    },
    {
      "epoch": 0.2230181722412431,
      "grad_norm": 2.2201685905456543,
      "learning_rate": 3.885830919146695e-05,
      "loss": 1.2089,
      "step": 8468
    },
    {
      "epoch": 0.2230445088227548,
      "grad_norm": 2.6045634746551514,
      "learning_rate": 3.8856992362391367e-05,
      "loss": 1.7408,
      "step": 8469
    },
    {
      "epoch": 0.22307084540426653,
      "grad_norm": 1.5373564958572388,
      "learning_rate": 3.8855675533315775e-05,
      "loss": 2.3959,
      "step": 8470
    },
    {
      "epoch": 0.22309718198577824,
      "grad_norm": 1.8168038129806519,
      "learning_rate": 3.885435870424019e-05,
      "loss": 1.8689,
      "step": 8471
    },
    {
      "epoch": 0.22312351856728996,
      "grad_norm": 6.167013168334961,
      "learning_rate": 3.88530418751646e-05,
      "loss": 2.291,
      "step": 8472
    },
    {
      "epoch": 0.22314985514880167,
      "grad_norm": 2.320302963256836,
      "learning_rate": 3.885172504608902e-05,
      "loss": 1.5089,
      "step": 8473
    },
    {
      "epoch": 0.22317619173031342,
      "grad_norm": 2.7399489879608154,
      "learning_rate": 3.885040821701343e-05,
      "loss": 1.7874,
      "step": 8474
    },
    {
      "epoch": 0.22320252831182513,
      "grad_norm": 1.4955161809921265,
      "learning_rate": 3.8849091387937847e-05,
      "loss": 1.4719,
      "step": 8475
    },
    {
      "epoch": 0.22322886489333685,
      "grad_norm": 3.9908576011657715,
      "learning_rate": 3.884777455886226e-05,
      "loss": 1.4037,
      "step": 8476
    },
    {
      "epoch": 0.22325520147484856,
      "grad_norm": 2.6037309169769287,
      "learning_rate": 3.884645772978668e-05,
      "loss": 0.7771,
      "step": 8477
    },
    {
      "epoch": 0.22328153805636028,
      "grad_norm": 1.9858044385910034,
      "learning_rate": 3.884514090071109e-05,
      "loss": 2.0154,
      "step": 8478
    },
    {
      "epoch": 0.223307874637872,
      "grad_norm": 1.781132459640503,
      "learning_rate": 3.88438240716355e-05,
      "loss": 1.6182,
      "step": 8479
    },
    {
      "epoch": 0.2233342112193837,
      "grad_norm": 1.9039111137390137,
      "learning_rate": 3.884250724255992e-05,
      "loss": 2.3106,
      "step": 8480
    },
    {
      "epoch": 0.22336054780089545,
      "grad_norm": 2.432525396347046,
      "learning_rate": 3.8841190413484327e-05,
      "loss": 1.3043,
      "step": 8481
    },
    {
      "epoch": 0.22338688438240717,
      "grad_norm": 1.993685007095337,
      "learning_rate": 3.883987358440875e-05,
      "loss": 1.7583,
      "step": 8482
    },
    {
      "epoch": 0.22341322096391888,
      "grad_norm": 2.3282101154327393,
      "learning_rate": 3.883855675533316e-05,
      "loss": 1.4714,
      "step": 8483
    },
    {
      "epoch": 0.2234395575454306,
      "grad_norm": 3.4791910648345947,
      "learning_rate": 3.883723992625757e-05,
      "loss": 1.6807,
      "step": 8484
    },
    {
      "epoch": 0.22346589412694232,
      "grad_norm": 2.2094662189483643,
      "learning_rate": 3.883592309718199e-05,
      "loss": 2.3535,
      "step": 8485
    },
    {
      "epoch": 0.22349223070845403,
      "grad_norm": 1.927895188331604,
      "learning_rate": 3.88346062681064e-05,
      "loss": 1.6366,
      "step": 8486
    },
    {
      "epoch": 0.22351856728996577,
      "grad_norm": 2.2030980587005615,
      "learning_rate": 3.883328943903082e-05,
      "loss": 1.6209,
      "step": 8487
    },
    {
      "epoch": 0.2235449038714775,
      "grad_norm": 1.9624310731887817,
      "learning_rate": 3.883197260995523e-05,
      "loss": 2.0201,
      "step": 8488
    },
    {
      "epoch": 0.2235712404529892,
      "grad_norm": 1.8407219648361206,
      "learning_rate": 3.8830655780879645e-05,
      "loss": 1.8291,
      "step": 8489
    },
    {
      "epoch": 0.22359757703450092,
      "grad_norm": 3.4186699390411377,
      "learning_rate": 3.8829338951804053e-05,
      "loss": 1.4813,
      "step": 8490
    },
    {
      "epoch": 0.22362391361601264,
      "grad_norm": 2.8706843852996826,
      "learning_rate": 3.882802212272847e-05,
      "loss": 1.9512,
      "step": 8491
    },
    {
      "epoch": 0.22365025019752435,
      "grad_norm": 2.0186493396759033,
      "learning_rate": 3.8826705293652885e-05,
      "loss": 1.708,
      "step": 8492
    },
    {
      "epoch": 0.22367658677903607,
      "grad_norm": 2.8753669261932373,
      "learning_rate": 3.88253884645773e-05,
      "loss": 0.2223,
      "step": 8493
    },
    {
      "epoch": 0.2237029233605478,
      "grad_norm": 3.4306418895721436,
      "learning_rate": 3.8824071635501716e-05,
      "loss": 1.1844,
      "step": 8494
    },
    {
      "epoch": 0.22372925994205953,
      "grad_norm": 2.8431928157806396,
      "learning_rate": 3.8822754806426125e-05,
      "loss": 0.7607,
      "step": 8495
    },
    {
      "epoch": 0.22375559652357124,
      "grad_norm": 2.5781729221343994,
      "learning_rate": 3.882143797735055e-05,
      "loss": 1.8525,
      "step": 8496
    },
    {
      "epoch": 0.22378193310508296,
      "grad_norm": 6.239348411560059,
      "learning_rate": 3.8820121148274956e-05,
      "loss": 1.2585,
      "step": 8497
    },
    {
      "epoch": 0.22380826968659467,
      "grad_norm": 1.6619477272033691,
      "learning_rate": 3.881880431919937e-05,
      "loss": 2.1441,
      "step": 8498
    },
    {
      "epoch": 0.2238346062681064,
      "grad_norm": 3.3440001010894775,
      "learning_rate": 3.881748749012378e-05,
      "loss": 2.2624,
      "step": 8499
    },
    {
      "epoch": 0.22386094284961813,
      "grad_norm": 2.327094793319702,
      "learning_rate": 3.8816170661048196e-05,
      "loss": 2.2825,
      "step": 8500
    },
    {
      "epoch": 0.22388727943112985,
      "grad_norm": 2.0841317176818848,
      "learning_rate": 3.881485383197261e-05,
      "loss": 2.0829,
      "step": 8501
    },
    {
      "epoch": 0.22391361601264156,
      "grad_norm": 1.7484424114227295,
      "learning_rate": 3.881353700289703e-05,
      "loss": 0.6641,
      "step": 8502
    },
    {
      "epoch": 0.22393995259415328,
      "grad_norm": 1.9922446012496948,
      "learning_rate": 3.881222017382144e-05,
      "loss": 2.0024,
      "step": 8503
    },
    {
      "epoch": 0.223966289175665,
      "grad_norm": 2.3496694564819336,
      "learning_rate": 3.881090334474585e-05,
      "loss": 1.3751,
      "step": 8504
    },
    {
      "epoch": 0.2239926257571767,
      "grad_norm": 3.416658401489258,
      "learning_rate": 3.880958651567027e-05,
      "loss": 1.425,
      "step": 8505
    },
    {
      "epoch": 0.22401896233868843,
      "grad_norm": 2.39328670501709,
      "learning_rate": 3.880826968659468e-05,
      "loss": 1.3231,
      "step": 8506
    },
    {
      "epoch": 0.22404529892020017,
      "grad_norm": 2.422624111175537,
      "learning_rate": 3.88069528575191e-05,
      "loss": 1.6258,
      "step": 8507
    },
    {
      "epoch": 0.22407163550171189,
      "grad_norm": 4.1461334228515625,
      "learning_rate": 3.880563602844351e-05,
      "loss": 1.278,
      "step": 8508
    },
    {
      "epoch": 0.2240979720832236,
      "grad_norm": 2.1527082920074463,
      "learning_rate": 3.880431919936792e-05,
      "loss": 1.9502,
      "step": 8509
    },
    {
      "epoch": 0.22412430866473532,
      "grad_norm": 3.2467570304870605,
      "learning_rate": 3.880300237029234e-05,
      "loss": 1.6126,
      "step": 8510
    },
    {
      "epoch": 0.22415064524624703,
      "grad_norm": 2.087463617324829,
      "learning_rate": 3.8801685541216754e-05,
      "loss": 2.4824,
      "step": 8511
    },
    {
      "epoch": 0.22417698182775875,
      "grad_norm": 1.4734315872192383,
      "learning_rate": 3.880036871214117e-05,
      "loss": 1.6918,
      "step": 8512
    },
    {
      "epoch": 0.22420331840927046,
      "grad_norm": 2.37542986869812,
      "learning_rate": 3.879905188306558e-05,
      "loss": 1.5313,
      "step": 8513
    },
    {
      "epoch": 0.2242296549907822,
      "grad_norm": 1.6531273126602173,
      "learning_rate": 3.8797735053989994e-05,
      "loss": 0.7821,
      "step": 8514
    },
    {
      "epoch": 0.22425599157229392,
      "grad_norm": 2.556692361831665,
      "learning_rate": 3.879641822491441e-05,
      "loss": 1.7999,
      "step": 8515
    },
    {
      "epoch": 0.22428232815380564,
      "grad_norm": 1.7149287462234497,
      "learning_rate": 3.8795101395838825e-05,
      "loss": 1.41,
      "step": 8516
    },
    {
      "epoch": 0.22430866473531735,
      "grad_norm": 1.8761473894119263,
      "learning_rate": 3.8793784566763234e-05,
      "loss": 1.8582,
      "step": 8517
    },
    {
      "epoch": 0.22433500131682907,
      "grad_norm": 2.844177722930908,
      "learning_rate": 3.879246773768765e-05,
      "loss": 1.6425,
      "step": 8518
    },
    {
      "epoch": 0.22436133789834078,
      "grad_norm": 4.278332233428955,
      "learning_rate": 3.879115090861206e-05,
      "loss": 1.5334,
      "step": 8519
    },
    {
      "epoch": 0.22438767447985253,
      "grad_norm": 4.926701545715332,
      "learning_rate": 3.878983407953648e-05,
      "loss": 1.7066,
      "step": 8520
    },
    {
      "epoch": 0.22441401106136424,
      "grad_norm": 2.12789249420166,
      "learning_rate": 3.8788517250460896e-05,
      "loss": 1.9511,
      "step": 8521
    },
    {
      "epoch": 0.22444034764287596,
      "grad_norm": 3.5499444007873535,
      "learning_rate": 3.8787200421385305e-05,
      "loss": 0.4712,
      "step": 8522
    },
    {
      "epoch": 0.22446668422438767,
      "grad_norm": 2.7263238430023193,
      "learning_rate": 3.878588359230972e-05,
      "loss": 1.6893,
      "step": 8523
    },
    {
      "epoch": 0.2244930208058994,
      "grad_norm": 1.685333251953125,
      "learning_rate": 3.878456676323413e-05,
      "loss": 1.1083,
      "step": 8524
    },
    {
      "epoch": 0.2245193573874111,
      "grad_norm": 2.1979739665985107,
      "learning_rate": 3.878324993415855e-05,
      "loss": 1.131,
      "step": 8525
    },
    {
      "epoch": 0.22454569396892282,
      "grad_norm": 1.5049335956573486,
      "learning_rate": 3.878193310508296e-05,
      "loss": 1.7592,
      "step": 8526
    },
    {
      "epoch": 0.22457203055043456,
      "grad_norm": 1.8547112941741943,
      "learning_rate": 3.8780616276007376e-05,
      "loss": 1.7506,
      "step": 8527
    },
    {
      "epoch": 0.22459836713194628,
      "grad_norm": 1.5974327325820923,
      "learning_rate": 3.8779299446931785e-05,
      "loss": 1.9607,
      "step": 8528
    },
    {
      "epoch": 0.224624703713458,
      "grad_norm": 5.663325786590576,
      "learning_rate": 3.877798261785621e-05,
      "loss": 1.395,
      "step": 8529
    },
    {
      "epoch": 0.2246510402949697,
      "grad_norm": 2.4018030166625977,
      "learning_rate": 3.8776665788780616e-05,
      "loss": 1.5803,
      "step": 8530
    },
    {
      "epoch": 0.22467737687648143,
      "grad_norm": 2.81136417388916,
      "learning_rate": 3.877534895970503e-05,
      "loss": 1.9475,
      "step": 8531
    },
    {
      "epoch": 0.22470371345799314,
      "grad_norm": 8.86347484588623,
      "learning_rate": 3.877403213062945e-05,
      "loss": 2.1431,
      "step": 8532
    },
    {
      "epoch": 0.22473005003950489,
      "grad_norm": 4.538589954376221,
      "learning_rate": 3.8772715301553856e-05,
      "loss": 1.9601,
      "step": 8533
    },
    {
      "epoch": 0.2247563866210166,
      "grad_norm": 3.4627835750579834,
      "learning_rate": 3.877139847247828e-05,
      "loss": 2.2073,
      "step": 8534
    },
    {
      "epoch": 0.22478272320252832,
      "grad_norm": 1.9426106214523315,
      "learning_rate": 3.877008164340269e-05,
      "loss": 2.5875,
      "step": 8535
    },
    {
      "epoch": 0.22480905978404003,
      "grad_norm": 3.228882312774658,
      "learning_rate": 3.87687648143271e-05,
      "loss": 1.3448,
      "step": 8536
    },
    {
      "epoch": 0.22483539636555175,
      "grad_norm": 1.9041908979415894,
      "learning_rate": 3.876744798525151e-05,
      "loss": 1.8954,
      "step": 8537
    },
    {
      "epoch": 0.22486173294706346,
      "grad_norm": 2.3903872966766357,
      "learning_rate": 3.876613115617593e-05,
      "loss": 2.2403,
      "step": 8538
    },
    {
      "epoch": 0.22488806952857518,
      "grad_norm": 3.6657254695892334,
      "learning_rate": 3.876481432710034e-05,
      "loss": 2.3583,
      "step": 8539
    },
    {
      "epoch": 0.22491440611008692,
      "grad_norm": 2.3356516361236572,
      "learning_rate": 3.876349749802476e-05,
      "loss": 2.2189,
      "step": 8540
    },
    {
      "epoch": 0.22494074269159864,
      "grad_norm": 1.4591888189315796,
      "learning_rate": 3.8762180668949174e-05,
      "loss": 0.7034,
      "step": 8541
    },
    {
      "epoch": 0.22496707927311035,
      "grad_norm": 1.8052279949188232,
      "learning_rate": 3.876086383987358e-05,
      "loss": 0.9604,
      "step": 8542
    },
    {
      "epoch": 0.22499341585462207,
      "grad_norm": 2.158219814300537,
      "learning_rate": 3.8759547010798006e-05,
      "loss": 1.6281,
      "step": 8543
    },
    {
      "epoch": 0.22501975243613379,
      "grad_norm": 2.6899774074554443,
      "learning_rate": 3.8758230181722414e-05,
      "loss": 2.2567,
      "step": 8544
    },
    {
      "epoch": 0.2250460890176455,
      "grad_norm": 2.3808324337005615,
      "learning_rate": 3.875691335264683e-05,
      "loss": 1.771,
      "step": 8545
    },
    {
      "epoch": 0.22507242559915722,
      "grad_norm": 1.6924216747283936,
      "learning_rate": 3.875559652357124e-05,
      "loss": 0.634,
      "step": 8546
    },
    {
      "epoch": 0.22509876218066896,
      "grad_norm": 1.7309690713882446,
      "learning_rate": 3.8754279694495654e-05,
      "loss": 2.135,
      "step": 8547
    },
    {
      "epoch": 0.22512509876218068,
      "grad_norm": 1.836403489112854,
      "learning_rate": 3.875296286542007e-05,
      "loss": 1.7224,
      "step": 8548
    },
    {
      "epoch": 0.2251514353436924,
      "grad_norm": 1.7984205484390259,
      "learning_rate": 3.8751646036344486e-05,
      "loss": 1.8589,
      "step": 8549
    },
    {
      "epoch": 0.2251777719252041,
      "grad_norm": 2.807633876800537,
      "learning_rate": 3.87503292072689e-05,
      "loss": 2.1739,
      "step": 8550
    },
    {
      "epoch": 0.22520410850671582,
      "grad_norm": 2.8705146312713623,
      "learning_rate": 3.874901237819331e-05,
      "loss": 2.244,
      "step": 8551
    },
    {
      "epoch": 0.22523044508822754,
      "grad_norm": 2.9991888999938965,
      "learning_rate": 3.8747695549117726e-05,
      "loss": 2.9163,
      "step": 8552
    },
    {
      "epoch": 0.22525678166973928,
      "grad_norm": 1.6255139112472534,
      "learning_rate": 3.874637872004214e-05,
      "loss": 1.8206,
      "step": 8553
    },
    {
      "epoch": 0.225283118251251,
      "grad_norm": 1.4978634119033813,
      "learning_rate": 3.874506189096656e-05,
      "loss": 0.3102,
      "step": 8554
    },
    {
      "epoch": 0.2253094548327627,
      "grad_norm": 4.614594459533691,
      "learning_rate": 3.8743745061890966e-05,
      "loss": 1.5802,
      "step": 8555
    },
    {
      "epoch": 0.22533579141427443,
      "grad_norm": 1.7774407863616943,
      "learning_rate": 3.874242823281538e-05,
      "loss": 2.0975,
      "step": 8556
    },
    {
      "epoch": 0.22536212799578614,
      "grad_norm": 1.6214098930358887,
      "learning_rate": 3.87411114037398e-05,
      "loss": 1.718,
      "step": 8557
    },
    {
      "epoch": 0.22538846457729786,
      "grad_norm": 3.455246686935425,
      "learning_rate": 3.873979457466421e-05,
      "loss": 1.6606,
      "step": 8558
    },
    {
      "epoch": 0.22541480115880957,
      "grad_norm": 3.538031578063965,
      "learning_rate": 3.873847774558863e-05,
      "loss": 1.8543,
      "step": 8559
    },
    {
      "epoch": 0.22544113774032132,
      "grad_norm": 1.7402997016906738,
      "learning_rate": 3.873716091651304e-05,
      "loss": 1.7954,
      "step": 8560
    },
    {
      "epoch": 0.22546747432183303,
      "grad_norm": 2.8732120990753174,
      "learning_rate": 3.873584408743745e-05,
      "loss": 2.2023,
      "step": 8561
    },
    {
      "epoch": 0.22549381090334475,
      "grad_norm": 2.5810511112213135,
      "learning_rate": 3.873452725836187e-05,
      "loss": 1.1899,
      "step": 8562
    },
    {
      "epoch": 0.22552014748485646,
      "grad_norm": 5.389581680297852,
      "learning_rate": 3.8733210429286284e-05,
      "loss": 2.722,
      "step": 8563
    },
    {
      "epoch": 0.22554648406636818,
      "grad_norm": 2.085892677307129,
      "learning_rate": 3.873189360021069e-05,
      "loss": 1.722,
      "step": 8564
    },
    {
      "epoch": 0.2255728206478799,
      "grad_norm": 2.844398021697998,
      "learning_rate": 3.873057677113511e-05,
      "loss": 1.2333,
      "step": 8565
    },
    {
      "epoch": 0.2255991572293916,
      "grad_norm": 2.088977336883545,
      "learning_rate": 3.8729259942059524e-05,
      "loss": 1.0623,
      "step": 8566
    },
    {
      "epoch": 0.22562549381090335,
      "grad_norm": 4.0470757484436035,
      "learning_rate": 3.872794311298394e-05,
      "loss": 0.6148,
      "step": 8567
    },
    {
      "epoch": 0.22565183039241507,
      "grad_norm": 1.453710913658142,
      "learning_rate": 3.8726626283908355e-05,
      "loss": 2.4347,
      "step": 8568
    },
    {
      "epoch": 0.22567816697392679,
      "grad_norm": 1.9698339700698853,
      "learning_rate": 3.8725309454832764e-05,
      "loss": 2.1407,
      "step": 8569
    },
    {
      "epoch": 0.2257045035554385,
      "grad_norm": 2.1381795406341553,
      "learning_rate": 3.872399262575718e-05,
      "loss": 1.4423,
      "step": 8570
    },
    {
      "epoch": 0.22573084013695022,
      "grad_norm": 4.505320072174072,
      "learning_rate": 3.872267579668159e-05,
      "loss": 1.1319,
      "step": 8571
    },
    {
      "epoch": 0.22575717671846193,
      "grad_norm": 2.297389268875122,
      "learning_rate": 3.872135896760601e-05,
      "loss": 1.8162,
      "step": 8572
    },
    {
      "epoch": 0.22578351329997368,
      "grad_norm": 4.679892063140869,
      "learning_rate": 3.872004213853042e-05,
      "loss": 1.24,
      "step": 8573
    },
    {
      "epoch": 0.2258098498814854,
      "grad_norm": 2.213726043701172,
      "learning_rate": 3.8718725309454835e-05,
      "loss": 2.1396,
      "step": 8574
    },
    {
      "epoch": 0.2258361864629971,
      "grad_norm": 1.8329153060913086,
      "learning_rate": 3.8717408480379244e-05,
      "loss": 2.0536,
      "step": 8575
    },
    {
      "epoch": 0.22586252304450882,
      "grad_norm": 3.7163002490997314,
      "learning_rate": 3.8716091651303666e-05,
      "loss": 1.6194,
      "step": 8576
    },
    {
      "epoch": 0.22588885962602054,
      "grad_norm": 2.213752269744873,
      "learning_rate": 3.8714774822228075e-05,
      "loss": 2.2845,
      "step": 8577
    },
    {
      "epoch": 0.22591519620753225,
      "grad_norm": 2.993347644805908,
      "learning_rate": 3.871345799315249e-05,
      "loss": 1.2136,
      "step": 8578
    },
    {
      "epoch": 0.22594153278904397,
      "grad_norm": 2.7208147048950195,
      "learning_rate": 3.8712141164076906e-05,
      "loss": 1.3,
      "step": 8579
    },
    {
      "epoch": 0.2259678693705557,
      "grad_norm": 3.664532423019409,
      "learning_rate": 3.8710824335001315e-05,
      "loss": 1.4404,
      "step": 8580
    },
    {
      "epoch": 0.22599420595206743,
      "grad_norm": 3.191732168197632,
      "learning_rate": 3.870950750592574e-05,
      "loss": 1.2009,
      "step": 8581
    },
    {
      "epoch": 0.22602054253357914,
      "grad_norm": 2.4372830390930176,
      "learning_rate": 3.8708190676850146e-05,
      "loss": 1.9618,
      "step": 8582
    },
    {
      "epoch": 0.22604687911509086,
      "grad_norm": 1.9489986896514893,
      "learning_rate": 3.870687384777456e-05,
      "loss": 1.8435,
      "step": 8583
    },
    {
      "epoch": 0.22607321569660258,
      "grad_norm": 2.5722455978393555,
      "learning_rate": 3.870555701869897e-05,
      "loss": 0.7802,
      "step": 8584
    },
    {
      "epoch": 0.2260995522781143,
      "grad_norm": 2.0546438694000244,
      "learning_rate": 3.8704240189623386e-05,
      "loss": 1.7552,
      "step": 8585
    },
    {
      "epoch": 0.22612588885962603,
      "grad_norm": 1.6043113470077515,
      "learning_rate": 3.87029233605478e-05,
      "loss": 1.2752,
      "step": 8586
    },
    {
      "epoch": 0.22615222544113775,
      "grad_norm": 1.9670912027359009,
      "learning_rate": 3.870160653147222e-05,
      "loss": 1.8038,
      "step": 8587
    },
    {
      "epoch": 0.22617856202264947,
      "grad_norm": 4.275622367858887,
      "learning_rate": 3.870028970239663e-05,
      "loss": 0.8754,
      "step": 8588
    },
    {
      "epoch": 0.22620489860416118,
      "grad_norm": 2.9276010990142822,
      "learning_rate": 3.869897287332104e-05,
      "loss": 1.949,
      "step": 8589
    },
    {
      "epoch": 0.2262312351856729,
      "grad_norm": 1.6066828966140747,
      "learning_rate": 3.869765604424546e-05,
      "loss": 2.0154,
      "step": 8590
    },
    {
      "epoch": 0.2262575717671846,
      "grad_norm": 3.8568646907806396,
      "learning_rate": 3.869633921516987e-05,
      "loss": 1.5891,
      "step": 8591
    },
    {
      "epoch": 0.22628390834869633,
      "grad_norm": 2.780968427658081,
      "learning_rate": 3.869502238609429e-05,
      "loss": 1.7727,
      "step": 8592
    },
    {
      "epoch": 0.22631024493020807,
      "grad_norm": 3.259148359298706,
      "learning_rate": 3.86937055570187e-05,
      "loss": 2.8154,
      "step": 8593
    },
    {
      "epoch": 0.2263365815117198,
      "grad_norm": 2.9252030849456787,
      "learning_rate": 3.869238872794311e-05,
      "loss": 1.9262,
      "step": 8594
    },
    {
      "epoch": 0.2263629180932315,
      "grad_norm": 3.966649293899536,
      "learning_rate": 3.869107189886753e-05,
      "loss": 1.4662,
      "step": 8595
    },
    {
      "epoch": 0.22638925467474322,
      "grad_norm": 2.670992612838745,
      "learning_rate": 3.8689755069791944e-05,
      "loss": 0.5004,
      "step": 8596
    },
    {
      "epoch": 0.22641559125625493,
      "grad_norm": 2.465890884399414,
      "learning_rate": 3.868843824071636e-05,
      "loss": 1.4256,
      "step": 8597
    },
    {
      "epoch": 0.22644192783776665,
      "grad_norm": 3.0254628658294678,
      "learning_rate": 3.868712141164077e-05,
      "loss": 1.8366,
      "step": 8598
    },
    {
      "epoch": 0.22646826441927836,
      "grad_norm": 4.357585430145264,
      "learning_rate": 3.8685804582565184e-05,
      "loss": 1.8118,
      "step": 8599
    },
    {
      "epoch": 0.2264946010007901,
      "grad_norm": 2.545929193496704,
      "learning_rate": 3.86844877534896e-05,
      "loss": 1.2503,
      "step": 8600
    },
    {
      "epoch": 0.22652093758230182,
      "grad_norm": 1.5786738395690918,
      "learning_rate": 3.8683170924414015e-05,
      "loss": 2.1923,
      "step": 8601
    },
    {
      "epoch": 0.22654727416381354,
      "grad_norm": 1.7433059215545654,
      "learning_rate": 3.8681854095338424e-05,
      "loss": 2.0981,
      "step": 8602
    },
    {
      "epoch": 0.22657361074532525,
      "grad_norm": 2.3527612686157227,
      "learning_rate": 3.868053726626284e-05,
      "loss": 1.7351,
      "step": 8603
    },
    {
      "epoch": 0.22659994732683697,
      "grad_norm": 6.146824359893799,
      "learning_rate": 3.8679220437187255e-05,
      "loss": 1.8199,
      "step": 8604
    },
    {
      "epoch": 0.22662628390834869,
      "grad_norm": 1.9697847366333008,
      "learning_rate": 3.867790360811167e-05,
      "loss": 1.2718,
      "step": 8605
    },
    {
      "epoch": 0.22665262048986043,
      "grad_norm": 2.5407605171203613,
      "learning_rate": 3.8676586779036087e-05,
      "loss": 2.0836,
      "step": 8606
    },
    {
      "epoch": 0.22667895707137214,
      "grad_norm": 2.063901662826538,
      "learning_rate": 3.8675269949960495e-05,
      "loss": 1.6355,
      "step": 8607
    },
    {
      "epoch": 0.22670529365288386,
      "grad_norm": 2.405883312225342,
      "learning_rate": 3.867395312088491e-05,
      "loss": 0.6567,
      "step": 8608
    },
    {
      "epoch": 0.22673163023439558,
      "grad_norm": 2.1857759952545166,
      "learning_rate": 3.8672636291809327e-05,
      "loss": 1.3335,
      "step": 8609
    },
    {
      "epoch": 0.2267579668159073,
      "grad_norm": 3.159747362136841,
      "learning_rate": 3.867131946273374e-05,
      "loss": 1.0167,
      "step": 8610
    },
    {
      "epoch": 0.226784303397419,
      "grad_norm": 1.9634112119674683,
      "learning_rate": 3.867000263365815e-05,
      "loss": 2.5962,
      "step": 8611
    },
    {
      "epoch": 0.22681063997893072,
      "grad_norm": 1.862107753753662,
      "learning_rate": 3.8668685804582567e-05,
      "loss": 1.3851,
      "step": 8612
    },
    {
      "epoch": 0.22683697656044247,
      "grad_norm": 2.9085402488708496,
      "learning_rate": 3.866736897550698e-05,
      "loss": 0.8244,
      "step": 8613
    },
    {
      "epoch": 0.22686331314195418,
      "grad_norm": 1.575588583946228,
      "learning_rate": 3.86660521464314e-05,
      "loss": 1.224,
      "step": 8614
    },
    {
      "epoch": 0.2268896497234659,
      "grad_norm": 2.716517686843872,
      "learning_rate": 3.866473531735581e-05,
      "loss": 2.0725,
      "step": 8615
    },
    {
      "epoch": 0.2269159863049776,
      "grad_norm": 1.6224349737167358,
      "learning_rate": 3.866341848828022e-05,
      "loss": 1.5208,
      "step": 8616
    },
    {
      "epoch": 0.22694232288648933,
      "grad_norm": 2.7150113582611084,
      "learning_rate": 3.866210165920464e-05,
      "loss": 2.0537,
      "step": 8617
    },
    {
      "epoch": 0.22696865946800104,
      "grad_norm": 3.4237709045410156,
      "learning_rate": 3.866078483012905e-05,
      "loss": 1.7736,
      "step": 8618
    },
    {
      "epoch": 0.22699499604951276,
      "grad_norm": 1.5503649711608887,
      "learning_rate": 3.865946800105347e-05,
      "loss": 1.6825,
      "step": 8619
    },
    {
      "epoch": 0.2270213326310245,
      "grad_norm": 3.7680604457855225,
      "learning_rate": 3.865815117197788e-05,
      "loss": 1.5244,
      "step": 8620
    },
    {
      "epoch": 0.22704766921253622,
      "grad_norm": 2.461878538131714,
      "learning_rate": 3.8656834342902293e-05,
      "loss": 1.8361,
      "step": 8621
    },
    {
      "epoch": 0.22707400579404793,
      "grad_norm": 2.5986108779907227,
      "learning_rate": 3.86555175138267e-05,
      "loss": 1.7973,
      "step": 8622
    },
    {
      "epoch": 0.22710034237555965,
      "grad_norm": 1.5644572973251343,
      "learning_rate": 3.865420068475112e-05,
      "loss": 0.3525,
      "step": 8623
    },
    {
      "epoch": 0.22712667895707137,
      "grad_norm": 4.417360305786133,
      "learning_rate": 3.865288385567554e-05,
      "loss": 1.1063,
      "step": 8624
    },
    {
      "epoch": 0.22715301553858308,
      "grad_norm": 1.637245535850525,
      "learning_rate": 3.865156702659995e-05,
      "loss": 1.7159,
      "step": 8625
    },
    {
      "epoch": 0.22717935212009482,
      "grad_norm": 3.72438907623291,
      "learning_rate": 3.8650250197524365e-05,
      "loss": 1.7055,
      "step": 8626
    },
    {
      "epoch": 0.22720568870160654,
      "grad_norm": 2.523892641067505,
      "learning_rate": 3.8648933368448773e-05,
      "loss": 1.6392,
      "step": 8627
    },
    {
      "epoch": 0.22723202528311826,
      "grad_norm": 4.000173091888428,
      "learning_rate": 3.8647616539373196e-05,
      "loss": 1.2853,
      "step": 8628
    },
    {
      "epoch": 0.22725836186462997,
      "grad_norm": 2.9876821041107178,
      "learning_rate": 3.8646299710297605e-05,
      "loss": 2.1374,
      "step": 8629
    },
    {
      "epoch": 0.2272846984461417,
      "grad_norm": 2.0255048274993896,
      "learning_rate": 3.864498288122202e-05,
      "loss": 1.5352,
      "step": 8630
    },
    {
      "epoch": 0.2273110350276534,
      "grad_norm": 2.19376802444458,
      "learning_rate": 3.864366605214643e-05,
      "loss": 1.8505,
      "step": 8631
    },
    {
      "epoch": 0.22733737160916512,
      "grad_norm": 3.0949249267578125,
      "learning_rate": 3.8642349223070845e-05,
      "loss": 1.5474,
      "step": 8632
    },
    {
      "epoch": 0.22736370819067686,
      "grad_norm": 2.001195192337036,
      "learning_rate": 3.864103239399526e-05,
      "loss": 1.9927,
      "step": 8633
    },
    {
      "epoch": 0.22739004477218858,
      "grad_norm": 2.0876452922821045,
      "learning_rate": 3.8639715564919676e-05,
      "loss": 1.933,
      "step": 8634
    },
    {
      "epoch": 0.2274163813537003,
      "grad_norm": 1.8517053127288818,
      "learning_rate": 3.863839873584409e-05,
      "loss": 2.1866,
      "step": 8635
    },
    {
      "epoch": 0.227442717935212,
      "grad_norm": 3.2314717769622803,
      "learning_rate": 3.86370819067685e-05,
      "loss": 1.1993,
      "step": 8636
    },
    {
      "epoch": 0.22746905451672372,
      "grad_norm": 1.7439364194869995,
      "learning_rate": 3.8635765077692916e-05,
      "loss": 2.1373,
      "step": 8637
    },
    {
      "epoch": 0.22749539109823544,
      "grad_norm": 1.895079255104065,
      "learning_rate": 3.863444824861733e-05,
      "loss": 1.349,
      "step": 8638
    },
    {
      "epoch": 0.22752172767974718,
      "grad_norm": 3.603242874145508,
      "learning_rate": 3.863313141954175e-05,
      "loss": 1.0467,
      "step": 8639
    },
    {
      "epoch": 0.2275480642612589,
      "grad_norm": 1.9533138275146484,
      "learning_rate": 3.8631814590466156e-05,
      "loss": 1.6837,
      "step": 8640
    },
    {
      "epoch": 0.2275744008427706,
      "grad_norm": 1.6531928777694702,
      "learning_rate": 3.863049776139057e-05,
      "loss": 1.247,
      "step": 8641
    },
    {
      "epoch": 0.22760073742428233,
      "grad_norm": 2.8849799633026123,
      "learning_rate": 3.862918093231499e-05,
      "loss": 0.5407,
      "step": 8642
    },
    {
      "epoch": 0.22762707400579404,
      "grad_norm": 1.795291543006897,
      "learning_rate": 3.86278641032394e-05,
      "loss": 1.8623,
      "step": 8643
    },
    {
      "epoch": 0.22765341058730576,
      "grad_norm": 1.7819745540618896,
      "learning_rate": 3.862654727416382e-05,
      "loss": 1.6354,
      "step": 8644
    },
    {
      "epoch": 0.22767974716881748,
      "grad_norm": 2.4156219959259033,
      "learning_rate": 3.862523044508823e-05,
      "loss": 1.0683,
      "step": 8645
    },
    {
      "epoch": 0.22770608375032922,
      "grad_norm": 2.533712148666382,
      "learning_rate": 3.862391361601264e-05,
      "loss": 1.6396,
      "step": 8646
    },
    {
      "epoch": 0.22773242033184093,
      "grad_norm": 2.627272367477417,
      "learning_rate": 3.862259678693706e-05,
      "loss": 2.0669,
      "step": 8647
    },
    {
      "epoch": 0.22775875691335265,
      "grad_norm": 1.8279216289520264,
      "learning_rate": 3.8621279957861474e-05,
      "loss": 1.713,
      "step": 8648
    },
    {
      "epoch": 0.22778509349486437,
      "grad_norm": 3.0978190898895264,
      "learning_rate": 3.861996312878588e-05,
      "loss": 0.9317,
      "step": 8649
    },
    {
      "epoch": 0.22781143007637608,
      "grad_norm": 2.731104612350464,
      "learning_rate": 3.86186462997103e-05,
      "loss": 0.7556,
      "step": 8650
    },
    {
      "epoch": 0.2278377666578878,
      "grad_norm": 2.8023698329925537,
      "learning_rate": 3.8617329470634714e-05,
      "loss": 1.6117,
      "step": 8651
    },
    {
      "epoch": 0.2278641032393995,
      "grad_norm": 4.751079082489014,
      "learning_rate": 3.861601264155913e-05,
      "loss": 0.7975,
      "step": 8652
    },
    {
      "epoch": 0.22789043982091126,
      "grad_norm": 4.326756000518799,
      "learning_rate": 3.8614695812483545e-05,
      "loss": 0.9873,
      "step": 8653
    },
    {
      "epoch": 0.22791677640242297,
      "grad_norm": 2.7647929191589355,
      "learning_rate": 3.8613378983407954e-05,
      "loss": 1.5003,
      "step": 8654
    },
    {
      "epoch": 0.2279431129839347,
      "grad_norm": 2.266779899597168,
      "learning_rate": 3.861206215433237e-05,
      "loss": 1.2917,
      "step": 8655
    },
    {
      "epoch": 0.2279694495654464,
      "grad_norm": 1.672104001045227,
      "learning_rate": 3.861074532525678e-05,
      "loss": 2.2904,
      "step": 8656
    },
    {
      "epoch": 0.22799578614695812,
      "grad_norm": 9.445061683654785,
      "learning_rate": 3.86094284961812e-05,
      "loss": 1.6486,
      "step": 8657
    },
    {
      "epoch": 0.22802212272846983,
      "grad_norm": 2.426698923110962,
      "learning_rate": 3.860811166710561e-05,
      "loss": 1.8155,
      "step": 8658
    },
    {
      "epoch": 0.22804845930998158,
      "grad_norm": 2.0746910572052,
      "learning_rate": 3.8606794838030025e-05,
      "loss": 1.2942,
      "step": 8659
    },
    {
      "epoch": 0.2280747958914933,
      "grad_norm": 3.771878242492676,
      "learning_rate": 3.860547800895444e-05,
      "loss": 1.7471,
      "step": 8660
    },
    {
      "epoch": 0.228101132473005,
      "grad_norm": 2.3991341590881348,
      "learning_rate": 3.8604161179878856e-05,
      "loss": 2.0276,
      "step": 8661
    },
    {
      "epoch": 0.22812746905451672,
      "grad_norm": 3.5623087882995605,
      "learning_rate": 3.860284435080327e-05,
      "loss": 1.838,
      "step": 8662
    },
    {
      "epoch": 0.22815380563602844,
      "grad_norm": 2.001025676727295,
      "learning_rate": 3.860152752172768e-05,
      "loss": 2.0591,
      "step": 8663
    },
    {
      "epoch": 0.22818014221754016,
      "grad_norm": 1.6273208856582642,
      "learning_rate": 3.8600210692652096e-05,
      "loss": 1.4582,
      "step": 8664
    },
    {
      "epoch": 0.22820647879905187,
      "grad_norm": 2.0918726921081543,
      "learning_rate": 3.8598893863576505e-05,
      "loss": 1.5005,
      "step": 8665
    },
    {
      "epoch": 0.22823281538056361,
      "grad_norm": 4.101078033447266,
      "learning_rate": 3.859757703450093e-05,
      "loss": 1.5328,
      "step": 8666
    },
    {
      "epoch": 0.22825915196207533,
      "grad_norm": 2.197524070739746,
      "learning_rate": 3.8596260205425336e-05,
      "loss": 1.294,
      "step": 8667
    },
    {
      "epoch": 0.22828548854358705,
      "grad_norm": 1.9406253099441528,
      "learning_rate": 3.859494337634975e-05,
      "loss": 2.0219,
      "step": 8668
    },
    {
      "epoch": 0.22831182512509876,
      "grad_norm": 3.7249929904937744,
      "learning_rate": 3.859362654727417e-05,
      "loss": 2.7619,
      "step": 8669
    },
    {
      "epoch": 0.22833816170661048,
      "grad_norm": 2.5667500495910645,
      "learning_rate": 3.8592309718198576e-05,
      "loss": 1.7927,
      "step": 8670
    },
    {
      "epoch": 0.2283644982881222,
      "grad_norm": 1.8960399627685547,
      "learning_rate": 3.8590992889123e-05,
      "loss": 2.1342,
      "step": 8671
    },
    {
      "epoch": 0.22839083486963394,
      "grad_norm": 2.6433939933776855,
      "learning_rate": 3.858967606004741e-05,
      "loss": 1.6063,
      "step": 8672
    },
    {
      "epoch": 0.22841717145114565,
      "grad_norm": 3.4651176929473877,
      "learning_rate": 3.858835923097182e-05,
      "loss": 1.0563,
      "step": 8673
    },
    {
      "epoch": 0.22844350803265737,
      "grad_norm": 2.608092784881592,
      "learning_rate": 3.858704240189623e-05,
      "loss": 2.0434,
      "step": 8674
    },
    {
      "epoch": 0.22846984461416908,
      "grad_norm": 2.3010904788970947,
      "learning_rate": 3.8585725572820654e-05,
      "loss": 1.5795,
      "step": 8675
    },
    {
      "epoch": 0.2284961811956808,
      "grad_norm": 1.8976138830184937,
      "learning_rate": 3.858440874374506e-05,
      "loss": 1.5403,
      "step": 8676
    },
    {
      "epoch": 0.2285225177771925,
      "grad_norm": 2.553696632385254,
      "learning_rate": 3.858309191466948e-05,
      "loss": 1.9057,
      "step": 8677
    },
    {
      "epoch": 0.22854885435870423,
      "grad_norm": 2.6526734828948975,
      "learning_rate": 3.858177508559389e-05,
      "loss": 1.5814,
      "step": 8678
    },
    {
      "epoch": 0.22857519094021597,
      "grad_norm": 2.6710147857666016,
      "learning_rate": 3.85804582565183e-05,
      "loss": 1.928,
      "step": 8679
    },
    {
      "epoch": 0.2286015275217277,
      "grad_norm": 1.5780129432678223,
      "learning_rate": 3.857914142744272e-05,
      "loss": 1.3791,
      "step": 8680
    },
    {
      "epoch": 0.2286278641032394,
      "grad_norm": 1.8756920099258423,
      "learning_rate": 3.8577824598367134e-05,
      "loss": 1.6023,
      "step": 8681
    },
    {
      "epoch": 0.22865420068475112,
      "grad_norm": 2.1717419624328613,
      "learning_rate": 3.857650776929155e-05,
      "loss": 2.0012,
      "step": 8682
    },
    {
      "epoch": 0.22868053726626283,
      "grad_norm": 2.16422176361084,
      "learning_rate": 3.857519094021596e-05,
      "loss": 2.1048,
      "step": 8683
    },
    {
      "epoch": 0.22870687384777455,
      "grad_norm": 1.9747705459594727,
      "learning_rate": 3.8573874111140374e-05,
      "loss": 1.5198,
      "step": 8684
    },
    {
      "epoch": 0.22873321042928627,
      "grad_norm": 1.9603595733642578,
      "learning_rate": 3.857255728206479e-05,
      "loss": 1.9746,
      "step": 8685
    },
    {
      "epoch": 0.228759547010798,
      "grad_norm": 1.5197569131851196,
      "learning_rate": 3.8571240452989206e-05,
      "loss": 2.3079,
      "step": 8686
    },
    {
      "epoch": 0.22878588359230972,
      "grad_norm": 7.051483631134033,
      "learning_rate": 3.8569923623913614e-05,
      "loss": 1.4505,
      "step": 8687
    },
    {
      "epoch": 0.22881222017382144,
      "grad_norm": 1.56491219997406,
      "learning_rate": 3.856860679483803e-05,
      "loss": 1.7867,
      "step": 8688
    },
    {
      "epoch": 0.22883855675533316,
      "grad_norm": 2.5304079055786133,
      "learning_rate": 3.8567289965762446e-05,
      "loss": 0.6964,
      "step": 8689
    },
    {
      "epoch": 0.22886489333684487,
      "grad_norm": 2.102848529815674,
      "learning_rate": 3.856597313668686e-05,
      "loss": 1.6336,
      "step": 8690
    },
    {
      "epoch": 0.2288912299183566,
      "grad_norm": 3.069793224334717,
      "learning_rate": 3.856465630761128e-05,
      "loss": 1.8046,
      "step": 8691
    },
    {
      "epoch": 0.22891756649986833,
      "grad_norm": 3.1155624389648438,
      "learning_rate": 3.8563339478535686e-05,
      "loss": 1.6283,
      "step": 8692
    },
    {
      "epoch": 0.22894390308138005,
      "grad_norm": 1.3181477785110474,
      "learning_rate": 3.85620226494601e-05,
      "loss": 0.8697,
      "step": 8693
    },
    {
      "epoch": 0.22897023966289176,
      "grad_norm": 2.799595355987549,
      "learning_rate": 3.856070582038452e-05,
      "loss": 1.8261,
      "step": 8694
    },
    {
      "epoch": 0.22899657624440348,
      "grad_norm": 2.9487504959106445,
      "learning_rate": 3.855938899130893e-05,
      "loss": 2.2233,
      "step": 8695
    },
    {
      "epoch": 0.2290229128259152,
      "grad_norm": 1.8640750646591187,
      "learning_rate": 3.855807216223334e-05,
      "loss": 1.7792,
      "step": 8696
    },
    {
      "epoch": 0.2290492494074269,
      "grad_norm": 1.5914180278778076,
      "learning_rate": 3.855675533315776e-05,
      "loss": 2.6861,
      "step": 8697
    },
    {
      "epoch": 0.22907558598893862,
      "grad_norm": 2.3961830139160156,
      "learning_rate": 3.855543850408217e-05,
      "loss": 2.3185,
      "step": 8698
    },
    {
      "epoch": 0.22910192257045037,
      "grad_norm": 2.0692360401153564,
      "learning_rate": 3.855412167500659e-05,
      "loss": 1.9847,
      "step": 8699
    },
    {
      "epoch": 0.22912825915196208,
      "grad_norm": 3.728013277053833,
      "learning_rate": 3.8552804845931004e-05,
      "loss": 2.3875,
      "step": 8700
    },
    {
      "epoch": 0.2291545957334738,
      "grad_norm": 3.812608003616333,
      "learning_rate": 3.855148801685541e-05,
      "loss": 0.7028,
      "step": 8701
    },
    {
      "epoch": 0.2291809323149855,
      "grad_norm": 1.8863166570663452,
      "learning_rate": 3.855017118777983e-05,
      "loss": 2.0994,
      "step": 8702
    },
    {
      "epoch": 0.22920726889649723,
      "grad_norm": 1.8751435279846191,
      "learning_rate": 3.854885435870424e-05,
      "loss": 1.5438,
      "step": 8703
    },
    {
      "epoch": 0.22923360547800894,
      "grad_norm": 2.2387068271636963,
      "learning_rate": 3.854753752962866e-05,
      "loss": 1.6201,
      "step": 8704
    },
    {
      "epoch": 0.22925994205952066,
      "grad_norm": 2.119868755340576,
      "learning_rate": 3.854622070055307e-05,
      "loss": 1.9271,
      "step": 8705
    },
    {
      "epoch": 0.2292862786410324,
      "grad_norm": 2.0415561199188232,
      "learning_rate": 3.8544903871477484e-05,
      "loss": 1.8325,
      "step": 8706
    },
    {
      "epoch": 0.22931261522254412,
      "grad_norm": 2.3197097778320312,
      "learning_rate": 3.85435870424019e-05,
      "loss": 2.1155,
      "step": 8707
    },
    {
      "epoch": 0.22933895180405584,
      "grad_norm": 1.7115092277526855,
      "learning_rate": 3.8542270213326315e-05,
      "loss": 0.441,
      "step": 8708
    },
    {
      "epoch": 0.22936528838556755,
      "grad_norm": 1.941003441810608,
      "learning_rate": 3.854095338425073e-05,
      "loss": 1.8441,
      "step": 8709
    },
    {
      "epoch": 0.22939162496707927,
      "grad_norm": 2.673499345779419,
      "learning_rate": 3.853963655517514e-05,
      "loss": 0.7984,
      "step": 8710
    },
    {
      "epoch": 0.22941796154859098,
      "grad_norm": 1.9680812358856201,
      "learning_rate": 3.8538319726099555e-05,
      "loss": 1.8498,
      "step": 8711
    },
    {
      "epoch": 0.22944429813010273,
      "grad_norm": 2.821601152420044,
      "learning_rate": 3.8537002897023964e-05,
      "loss": 1.9629,
      "step": 8712
    },
    {
      "epoch": 0.22947063471161444,
      "grad_norm": 3.7915849685668945,
      "learning_rate": 3.8535686067948386e-05,
      "loss": 1.6068,
      "step": 8713
    },
    {
      "epoch": 0.22949697129312616,
      "grad_norm": 2.261084794998169,
      "learning_rate": 3.8534369238872795e-05,
      "loss": 1.568,
      "step": 8714
    },
    {
      "epoch": 0.22952330787463787,
      "grad_norm": 1.4744764566421509,
      "learning_rate": 3.853305240979721e-05,
      "loss": 1.5126,
      "step": 8715
    },
    {
      "epoch": 0.2295496444561496,
      "grad_norm": 2.1644668579101562,
      "learning_rate": 3.8531735580721626e-05,
      "loss": 1.5953,
      "step": 8716
    },
    {
      "epoch": 0.2295759810376613,
      "grad_norm": 2.81292462348938,
      "learning_rate": 3.8530418751646035e-05,
      "loss": 1.2637,
      "step": 8717
    },
    {
      "epoch": 0.22960231761917302,
      "grad_norm": 2.1572868824005127,
      "learning_rate": 3.852910192257046e-05,
      "loss": 1.9149,
      "step": 8718
    },
    {
      "epoch": 0.22962865420068476,
      "grad_norm": 2.567028522491455,
      "learning_rate": 3.8527785093494866e-05,
      "loss": 1.6274,
      "step": 8719
    },
    {
      "epoch": 0.22965499078219648,
      "grad_norm": 1.6461268663406372,
      "learning_rate": 3.852646826441928e-05,
      "loss": 1.7598,
      "step": 8720
    },
    {
      "epoch": 0.2296813273637082,
      "grad_norm": 1.711926817893982,
      "learning_rate": 3.852515143534369e-05,
      "loss": 1.0451,
      "step": 8721
    },
    {
      "epoch": 0.2297076639452199,
      "grad_norm": 2.110971689224243,
      "learning_rate": 3.8523834606268106e-05,
      "loss": 1.6453,
      "step": 8722
    },
    {
      "epoch": 0.22973400052673162,
      "grad_norm": 2.7735655307769775,
      "learning_rate": 3.852251777719252e-05,
      "loss": 1.7858,
      "step": 8723
    },
    {
      "epoch": 0.22976033710824334,
      "grad_norm": 1.9985178709030151,
      "learning_rate": 3.852120094811694e-05,
      "loss": 1.7008,
      "step": 8724
    },
    {
      "epoch": 0.22978667368975508,
      "grad_norm": 1.734743595123291,
      "learning_rate": 3.8519884119041346e-05,
      "loss": 0.6781,
      "step": 8725
    },
    {
      "epoch": 0.2298130102712668,
      "grad_norm": 2.0480659008026123,
      "learning_rate": 3.851856728996576e-05,
      "loss": 1.1985,
      "step": 8726
    },
    {
      "epoch": 0.22983934685277851,
      "grad_norm": 2.515310049057007,
      "learning_rate": 3.851725046089018e-05,
      "loss": 2.2086,
      "step": 8727
    },
    {
      "epoch": 0.22986568343429023,
      "grad_norm": 1.9793624877929688,
      "learning_rate": 3.851593363181459e-05,
      "loss": 1.4715,
      "step": 8728
    },
    {
      "epoch": 0.22989202001580195,
      "grad_norm": 1.7837162017822266,
      "learning_rate": 3.851461680273901e-05,
      "loss": 2.1454,
      "step": 8729
    },
    {
      "epoch": 0.22991835659731366,
      "grad_norm": 1.9805967807769775,
      "learning_rate": 3.851329997366342e-05,
      "loss": 1.6898,
      "step": 8730
    },
    {
      "epoch": 0.22994469317882538,
      "grad_norm": 2.5281834602355957,
      "learning_rate": 3.851198314458783e-05,
      "loss": 1.4195,
      "step": 8731
    },
    {
      "epoch": 0.22997102976033712,
      "grad_norm": 4.5295820236206055,
      "learning_rate": 3.851066631551225e-05,
      "loss": 1.5477,
      "step": 8732
    },
    {
      "epoch": 0.22999736634184884,
      "grad_norm": 5.615579605102539,
      "learning_rate": 3.8509349486436664e-05,
      "loss": 1.9838,
      "step": 8733
    },
    {
      "epoch": 0.23002370292336055,
      "grad_norm": 3.848259687423706,
      "learning_rate": 3.850803265736107e-05,
      "loss": 1.6755,
      "step": 8734
    },
    {
      "epoch": 0.23005003950487227,
      "grad_norm": 1.982571005821228,
      "learning_rate": 3.850671582828549e-05,
      "loss": 2.6243,
      "step": 8735
    },
    {
      "epoch": 0.23007637608638398,
      "grad_norm": 1.8228106498718262,
      "learning_rate": 3.8505398999209904e-05,
      "loss": 1.8012,
      "step": 8736
    },
    {
      "epoch": 0.2301027126678957,
      "grad_norm": 2.055089235305786,
      "learning_rate": 3.850408217013432e-05,
      "loss": 2.1505,
      "step": 8737
    },
    {
      "epoch": 0.2301290492494074,
      "grad_norm": 2.0300891399383545,
      "learning_rate": 3.8502765341058735e-05,
      "loss": 0.6957,
      "step": 8738
    },
    {
      "epoch": 0.23015538583091916,
      "grad_norm": 4.362094402313232,
      "learning_rate": 3.8501448511983144e-05,
      "loss": 1.9208,
      "step": 8739
    },
    {
      "epoch": 0.23018172241243087,
      "grad_norm": 3.210151195526123,
      "learning_rate": 3.850013168290756e-05,
      "loss": 1.6491,
      "step": 8740
    },
    {
      "epoch": 0.2302080589939426,
      "grad_norm": 1.9036731719970703,
      "learning_rate": 3.8498814853831975e-05,
      "loss": 1.7272,
      "step": 8741
    },
    {
      "epoch": 0.2302343955754543,
      "grad_norm": 4.6128082275390625,
      "learning_rate": 3.849749802475639e-05,
      "loss": 1.7663,
      "step": 8742
    },
    {
      "epoch": 0.23026073215696602,
      "grad_norm": 3.624314308166504,
      "learning_rate": 3.84961811956808e-05,
      "loss": 1.9732,
      "step": 8743
    },
    {
      "epoch": 0.23028706873847773,
      "grad_norm": 1.6349352598190308,
      "learning_rate": 3.8494864366605215e-05,
      "loss": 1.2109,
      "step": 8744
    },
    {
      "epoch": 0.23031340531998948,
      "grad_norm": 3.401796817779541,
      "learning_rate": 3.849354753752963e-05,
      "loss": 0.9124,
      "step": 8745
    },
    {
      "epoch": 0.2303397419015012,
      "grad_norm": 1.9930672645568848,
      "learning_rate": 3.849223070845405e-05,
      "loss": 1.1927,
      "step": 8746
    },
    {
      "epoch": 0.2303660784830129,
      "grad_norm": 4.248035430908203,
      "learning_rate": 3.849091387937846e-05,
      "loss": 0.7357,
      "step": 8747
    },
    {
      "epoch": 0.23039241506452462,
      "grad_norm": 2.0215535163879395,
      "learning_rate": 3.848959705030287e-05,
      "loss": 1.3645,
      "step": 8748
    },
    {
      "epoch": 0.23041875164603634,
      "grad_norm": 2.5462646484375,
      "learning_rate": 3.848828022122729e-05,
      "loss": 1.1863,
      "step": 8749
    },
    {
      "epoch": 0.23044508822754806,
      "grad_norm": 2.440798282623291,
      "learning_rate": 3.8486963392151695e-05,
      "loss": 1.8985,
      "step": 8750
    },
    {
      "epoch": 0.23047142480905977,
      "grad_norm": 2.9896252155303955,
      "learning_rate": 3.848564656307612e-05,
      "loss": 1.5054,
      "step": 8751
    },
    {
      "epoch": 0.23049776139057152,
      "grad_norm": 1.8389194011688232,
      "learning_rate": 3.848432973400053e-05,
      "loss": 1.977,
      "step": 8752
    },
    {
      "epoch": 0.23052409797208323,
      "grad_norm": 1.918176531791687,
      "learning_rate": 3.848301290492494e-05,
      "loss": 1.731,
      "step": 8753
    },
    {
      "epoch": 0.23055043455359495,
      "grad_norm": 2.1689364910125732,
      "learning_rate": 3.848169607584936e-05,
      "loss": 2.4061,
      "step": 8754
    },
    {
      "epoch": 0.23057677113510666,
      "grad_norm": 2.393303632736206,
      "learning_rate": 3.8480379246773773e-05,
      "loss": 1.6514,
      "step": 8755
    },
    {
      "epoch": 0.23060310771661838,
      "grad_norm": 2.7109014987945557,
      "learning_rate": 3.847906241769819e-05,
      "loss": 1.9559,
      "step": 8756
    },
    {
      "epoch": 0.2306294442981301,
      "grad_norm": 3.7777557373046875,
      "learning_rate": 3.84777455886226e-05,
      "loss": 1.4828,
      "step": 8757
    },
    {
      "epoch": 0.23065578087964184,
      "grad_norm": 1.9608242511749268,
      "learning_rate": 3.8476428759547013e-05,
      "loss": 1.4569,
      "step": 8758
    },
    {
      "epoch": 0.23068211746115355,
      "grad_norm": 1.7105075120925903,
      "learning_rate": 3.847511193047142e-05,
      "loss": 2.0292,
      "step": 8759
    },
    {
      "epoch": 0.23070845404266527,
      "grad_norm": 1.7458996772766113,
      "learning_rate": 3.8473795101395845e-05,
      "loss": 1.7096,
      "step": 8760
    },
    {
      "epoch": 0.23073479062417698,
      "grad_norm": 1.5461111068725586,
      "learning_rate": 3.8472478272320253e-05,
      "loss": 1.824,
      "step": 8761
    },
    {
      "epoch": 0.2307611272056887,
      "grad_norm": 1.8432605266571045,
      "learning_rate": 3.847116144324467e-05,
      "loss": 2.0229,
      "step": 8762
    },
    {
      "epoch": 0.23078746378720041,
      "grad_norm": 1.6574029922485352,
      "learning_rate": 3.8469844614169085e-05,
      "loss": 1.8793,
      "step": 8763
    },
    {
      "epoch": 0.23081380036871213,
      "grad_norm": 2.9303579330444336,
      "learning_rate": 3.8468527785093493e-05,
      "loss": 0.7993,
      "step": 8764
    },
    {
      "epoch": 0.23084013695022387,
      "grad_norm": 2.927445650100708,
      "learning_rate": 3.8467210956017916e-05,
      "loss": 2.1572,
      "step": 8765
    },
    {
      "epoch": 0.2308664735317356,
      "grad_norm": 2.6877167224884033,
      "learning_rate": 3.8465894126942325e-05,
      "loss": 0.5014,
      "step": 8766
    },
    {
      "epoch": 0.2308928101132473,
      "grad_norm": 1.4402517080307007,
      "learning_rate": 3.846457729786674e-05,
      "loss": 1.0874,
      "step": 8767
    },
    {
      "epoch": 0.23091914669475902,
      "grad_norm": 1.9190280437469482,
      "learning_rate": 3.846326046879115e-05,
      "loss": 1.78,
      "step": 8768
    },
    {
      "epoch": 0.23094548327627074,
      "grad_norm": 6.018390655517578,
      "learning_rate": 3.8461943639715565e-05,
      "loss": 1.4677,
      "step": 8769
    },
    {
      "epoch": 0.23097181985778245,
      "grad_norm": 1.4859380722045898,
      "learning_rate": 3.846062681063998e-05,
      "loss": 2.048,
      "step": 8770
    },
    {
      "epoch": 0.23099815643929417,
      "grad_norm": 1.430658221244812,
      "learning_rate": 3.8459309981564396e-05,
      "loss": 1.9816,
      "step": 8771
    },
    {
      "epoch": 0.2310244930208059,
      "grad_norm": 7.866768836975098,
      "learning_rate": 3.845799315248881e-05,
      "loss": 2.1579,
      "step": 8772
    },
    {
      "epoch": 0.23105082960231763,
      "grad_norm": 2.8055260181427,
      "learning_rate": 3.845667632341322e-05,
      "loss": 1.7819,
      "step": 8773
    },
    {
      "epoch": 0.23107716618382934,
      "grad_norm": 3.5743625164031982,
      "learning_rate": 3.845535949433764e-05,
      "loss": 1.326,
      "step": 8774
    },
    {
      "epoch": 0.23110350276534106,
      "grad_norm": 2.8899219036102295,
      "learning_rate": 3.845404266526205e-05,
      "loss": 0.8887,
      "step": 8775
    },
    {
      "epoch": 0.23112983934685277,
      "grad_norm": 2.2441341876983643,
      "learning_rate": 3.845272583618647e-05,
      "loss": 1.3165,
      "step": 8776
    },
    {
      "epoch": 0.2311561759283645,
      "grad_norm": 2.4311869144439697,
      "learning_rate": 3.8451409007110876e-05,
      "loss": 0.7973,
      "step": 8777
    },
    {
      "epoch": 0.23118251250987623,
      "grad_norm": 2.458587884902954,
      "learning_rate": 3.845009217803529e-05,
      "loss": 1.4253,
      "step": 8778
    },
    {
      "epoch": 0.23120884909138795,
      "grad_norm": 2.057854413986206,
      "learning_rate": 3.844877534895971e-05,
      "loss": 2.2464,
      "step": 8779
    },
    {
      "epoch": 0.23123518567289966,
      "grad_norm": 3.1313319206237793,
      "learning_rate": 3.844745851988412e-05,
      "loss": 1.7106,
      "step": 8780
    },
    {
      "epoch": 0.23126152225441138,
      "grad_norm": 2.7365825176239014,
      "learning_rate": 3.844614169080853e-05,
      "loss": 2.2137,
      "step": 8781
    },
    {
      "epoch": 0.2312878588359231,
      "grad_norm": 3.1917386054992676,
      "learning_rate": 3.844482486173295e-05,
      "loss": 2.1663,
      "step": 8782
    },
    {
      "epoch": 0.2313141954174348,
      "grad_norm": 4.121448993682861,
      "learning_rate": 3.844350803265736e-05,
      "loss": 0.5522,
      "step": 8783
    },
    {
      "epoch": 0.23134053199894652,
      "grad_norm": 4.569949626922607,
      "learning_rate": 3.844219120358178e-05,
      "loss": 1.788,
      "step": 8784
    },
    {
      "epoch": 0.23136686858045827,
      "grad_norm": 2.647071123123169,
      "learning_rate": 3.8440874374506194e-05,
      "loss": 2.181,
      "step": 8785
    },
    {
      "epoch": 0.23139320516196998,
      "grad_norm": 2.7681360244750977,
      "learning_rate": 3.84395575454306e-05,
      "loss": 1.3345,
      "step": 8786
    },
    {
      "epoch": 0.2314195417434817,
      "grad_norm": 3.2671942710876465,
      "learning_rate": 3.843824071635502e-05,
      "loss": 2.0853,
      "step": 8787
    },
    {
      "epoch": 0.23144587832499341,
      "grad_norm": 2.92760968208313,
      "learning_rate": 3.8436923887279434e-05,
      "loss": 0.8017,
      "step": 8788
    },
    {
      "epoch": 0.23147221490650513,
      "grad_norm": 2.052534580230713,
      "learning_rate": 3.843560705820385e-05,
      "loss": 0.7354,
      "step": 8789
    },
    {
      "epoch": 0.23149855148801685,
      "grad_norm": 2.1374247074127197,
      "learning_rate": 3.843429022912826e-05,
      "loss": 0.7031,
      "step": 8790
    },
    {
      "epoch": 0.23152488806952856,
      "grad_norm": 1.9190794229507446,
      "learning_rate": 3.8432973400052674e-05,
      "loss": 0.8202,
      "step": 8791
    },
    {
      "epoch": 0.2315512246510403,
      "grad_norm": 3.549086332321167,
      "learning_rate": 3.843165657097709e-05,
      "loss": 0.6385,
      "step": 8792
    },
    {
      "epoch": 0.23157756123255202,
      "grad_norm": 2.4704179763793945,
      "learning_rate": 3.8430339741901505e-05,
      "loss": 2.4266,
      "step": 8793
    },
    {
      "epoch": 0.23160389781406374,
      "grad_norm": 1.7736200094223022,
      "learning_rate": 3.842902291282592e-05,
      "loss": 1.9201,
      "step": 8794
    },
    {
      "epoch": 0.23163023439557545,
      "grad_norm": 4.950268745422363,
      "learning_rate": 3.842770608375033e-05,
      "loss": 1.6361,
      "step": 8795
    },
    {
      "epoch": 0.23165657097708717,
      "grad_norm": 2.5951664447784424,
      "learning_rate": 3.8426389254674745e-05,
      "loss": 1.0932,
      "step": 8796
    },
    {
      "epoch": 0.23168290755859888,
      "grad_norm": 2.1899523735046387,
      "learning_rate": 3.8425072425599154e-05,
      "loss": 1.1663,
      "step": 8797
    },
    {
      "epoch": 0.23170924414011063,
      "grad_norm": 2.146493911743164,
      "learning_rate": 3.8423755596523576e-05,
      "loss": 1.5842,
      "step": 8798
    },
    {
      "epoch": 0.23173558072162234,
      "grad_norm": 1.532976746559143,
      "learning_rate": 3.8422438767447985e-05,
      "loss": 0.4931,
      "step": 8799
    },
    {
      "epoch": 0.23176191730313406,
      "grad_norm": 2.102846145629883,
      "learning_rate": 3.84211219383724e-05,
      "loss": 1.738,
      "step": 8800
    },
    {
      "epoch": 0.23178825388464577,
      "grad_norm": 6.328239917755127,
      "learning_rate": 3.8419805109296816e-05,
      "loss": 2.3549,
      "step": 8801
    },
    {
      "epoch": 0.2318145904661575,
      "grad_norm": 3.091710329055786,
      "learning_rate": 3.8418488280221225e-05,
      "loss": 1.3087,
      "step": 8802
    },
    {
      "epoch": 0.2318409270476692,
      "grad_norm": 2.3866326808929443,
      "learning_rate": 3.841717145114565e-05,
      "loss": 1.8067,
      "step": 8803
    },
    {
      "epoch": 0.23186726362918092,
      "grad_norm": 4.606212139129639,
      "learning_rate": 3.8415854622070056e-05,
      "loss": 2.209,
      "step": 8804
    },
    {
      "epoch": 0.23189360021069266,
      "grad_norm": 4.543942928314209,
      "learning_rate": 3.841453779299447e-05,
      "loss": 1.5549,
      "step": 8805
    },
    {
      "epoch": 0.23191993679220438,
      "grad_norm": 3.7005951404571533,
      "learning_rate": 3.841322096391888e-05,
      "loss": 1.0762,
      "step": 8806
    },
    {
      "epoch": 0.2319462733737161,
      "grad_norm": 3.2195284366607666,
      "learning_rate": 3.84119041348433e-05,
      "loss": 1.2201,
      "step": 8807
    },
    {
      "epoch": 0.2319726099552278,
      "grad_norm": 2.577655553817749,
      "learning_rate": 3.841058730576771e-05,
      "loss": 1.4712,
      "step": 8808
    },
    {
      "epoch": 0.23199894653673953,
      "grad_norm": 2.9648633003234863,
      "learning_rate": 3.840927047669213e-05,
      "loss": 1.5106,
      "step": 8809
    },
    {
      "epoch": 0.23202528311825124,
      "grad_norm": 1.9411506652832031,
      "learning_rate": 3.840795364761654e-05,
      "loss": 1.7886,
      "step": 8810
    },
    {
      "epoch": 0.23205161969976298,
      "grad_norm": 2.274669647216797,
      "learning_rate": 3.840663681854095e-05,
      "loss": 1.5786,
      "step": 8811
    },
    {
      "epoch": 0.2320779562812747,
      "grad_norm": 3.650491237640381,
      "learning_rate": 3.8405319989465374e-05,
      "loss": 2.1217,
      "step": 8812
    },
    {
      "epoch": 0.23210429286278642,
      "grad_norm": 1.8162002563476562,
      "learning_rate": 3.840400316038978e-05,
      "loss": 1.9206,
      "step": 8813
    },
    {
      "epoch": 0.23213062944429813,
      "grad_norm": 3.3469510078430176,
      "learning_rate": 3.84026863313142e-05,
      "loss": 1.7331,
      "step": 8814
    },
    {
      "epoch": 0.23215696602580985,
      "grad_norm": 7.383838653564453,
      "learning_rate": 3.840136950223861e-05,
      "loss": 1.7761,
      "step": 8815
    },
    {
      "epoch": 0.23218330260732156,
      "grad_norm": 3.184968948364258,
      "learning_rate": 3.840005267316302e-05,
      "loss": 1.1129,
      "step": 8816
    },
    {
      "epoch": 0.23220963918883328,
      "grad_norm": 2.9226291179656982,
      "learning_rate": 3.839873584408744e-05,
      "loss": 1.6157,
      "step": 8817
    },
    {
      "epoch": 0.23223597577034502,
      "grad_norm": 1.9389110803604126,
      "learning_rate": 3.8397419015011854e-05,
      "loss": 2.3295,
      "step": 8818
    },
    {
      "epoch": 0.23226231235185674,
      "grad_norm": 2.777867078781128,
      "learning_rate": 3.839610218593627e-05,
      "loss": 2.1738,
      "step": 8819
    },
    {
      "epoch": 0.23228864893336845,
      "grad_norm": 2.0441317558288574,
      "learning_rate": 3.839478535686068e-05,
      "loss": 0.8182,
      "step": 8820
    },
    {
      "epoch": 0.23231498551488017,
      "grad_norm": 3.5684680938720703,
      "learning_rate": 3.83934685277851e-05,
      "loss": 2.0045,
      "step": 8821
    },
    {
      "epoch": 0.23234132209639188,
      "grad_norm": 2.815356492996216,
      "learning_rate": 3.839215169870951e-05,
      "loss": 1.831,
      "step": 8822
    },
    {
      "epoch": 0.2323676586779036,
      "grad_norm": 3.127657651901245,
      "learning_rate": 3.8390834869633926e-05,
      "loss": 0.5086,
      "step": 8823
    },
    {
      "epoch": 0.23239399525941531,
      "grad_norm": 1.9306000471115112,
      "learning_rate": 3.8389518040558334e-05,
      "loss": 1.7544,
      "step": 8824
    },
    {
      "epoch": 0.23242033184092706,
      "grad_norm": 1.8755202293395996,
      "learning_rate": 3.838820121148275e-05,
      "loss": 1.9621,
      "step": 8825
    },
    {
      "epoch": 0.23244666842243877,
      "grad_norm": 2.0463740825653076,
      "learning_rate": 3.8386884382407166e-05,
      "loss": 1.5315,
      "step": 8826
    },
    {
      "epoch": 0.2324730050039505,
      "grad_norm": 1.8066118955612183,
      "learning_rate": 3.838556755333158e-05,
      "loss": 2.4106,
      "step": 8827
    },
    {
      "epoch": 0.2324993415854622,
      "grad_norm": 2.7764689922332764,
      "learning_rate": 3.838425072425599e-05,
      "loss": 0.9899,
      "step": 8828
    },
    {
      "epoch": 0.23252567816697392,
      "grad_norm": 1.6359795331954956,
      "learning_rate": 3.8382933895180406e-05,
      "loss": 1.9047,
      "step": 8829
    },
    {
      "epoch": 0.23255201474848564,
      "grad_norm": 4.946788787841797,
      "learning_rate": 3.838161706610482e-05,
      "loss": 0.9244,
      "step": 8830
    },
    {
      "epoch": 0.23257835132999738,
      "grad_norm": 2.9737300872802734,
      "learning_rate": 3.838030023702924e-05,
      "loss": 1.648,
      "step": 8831
    },
    {
      "epoch": 0.2326046879115091,
      "grad_norm": 2.2024760246276855,
      "learning_rate": 3.837898340795365e-05,
      "loss": 0.5833,
      "step": 8832
    },
    {
      "epoch": 0.2326310244930208,
      "grad_norm": 5.991247653961182,
      "learning_rate": 3.837766657887806e-05,
      "loss": 1.0519,
      "step": 8833
    },
    {
      "epoch": 0.23265736107453253,
      "grad_norm": 1.4719481468200684,
      "learning_rate": 3.837634974980248e-05,
      "loss": 1.001,
      "step": 8834
    },
    {
      "epoch": 0.23268369765604424,
      "grad_norm": 1.9669474363327026,
      "learning_rate": 3.8375032920726886e-05,
      "loss": 2.4314,
      "step": 8835
    },
    {
      "epoch": 0.23271003423755596,
      "grad_norm": 2.3021914958953857,
      "learning_rate": 3.837371609165131e-05,
      "loss": 1.7301,
      "step": 8836
    },
    {
      "epoch": 0.23273637081906767,
      "grad_norm": 5.052045822143555,
      "learning_rate": 3.837239926257572e-05,
      "loss": 1.4613,
      "step": 8837
    },
    {
      "epoch": 0.23276270740057942,
      "grad_norm": 2.0064573287963867,
      "learning_rate": 3.837108243350013e-05,
      "loss": 1.8329,
      "step": 8838
    },
    {
      "epoch": 0.23278904398209113,
      "grad_norm": 1.8239521980285645,
      "learning_rate": 3.836976560442455e-05,
      "loss": 1.5283,
      "step": 8839
    },
    {
      "epoch": 0.23281538056360285,
      "grad_norm": 2.086679220199585,
      "learning_rate": 3.8368448775348964e-05,
      "loss": 2.1112,
      "step": 8840
    },
    {
      "epoch": 0.23284171714511456,
      "grad_norm": 1.819541573524475,
      "learning_rate": 3.836713194627338e-05,
      "loss": 2.1888,
      "step": 8841
    },
    {
      "epoch": 0.23286805372662628,
      "grad_norm": 2.143944025039673,
      "learning_rate": 3.836581511719779e-05,
      "loss": 2.3739,
      "step": 8842
    },
    {
      "epoch": 0.232894390308138,
      "grad_norm": 2.3751306533813477,
      "learning_rate": 3.8364498288122204e-05,
      "loss": 1.367,
      "step": 8843
    },
    {
      "epoch": 0.2329207268896497,
      "grad_norm": 1.697487473487854,
      "learning_rate": 3.836318145904661e-05,
      "loss": 1.6871,
      "step": 8844
    },
    {
      "epoch": 0.23294706347116145,
      "grad_norm": 1.8505195379257202,
      "learning_rate": 3.8361864629971035e-05,
      "loss": 1.4418,
      "step": 8845
    },
    {
      "epoch": 0.23297340005267317,
      "grad_norm": 2.0387566089630127,
      "learning_rate": 3.8360547800895444e-05,
      "loss": 1.7394,
      "step": 8846
    },
    {
      "epoch": 0.23299973663418488,
      "grad_norm": 2.9100518226623535,
      "learning_rate": 3.835923097181986e-05,
      "loss": 1.4584,
      "step": 8847
    },
    {
      "epoch": 0.2330260732156966,
      "grad_norm": 4.852202415466309,
      "learning_rate": 3.8357914142744275e-05,
      "loss": 2.2022,
      "step": 8848
    },
    {
      "epoch": 0.23305240979720832,
      "grad_norm": 3.094252109527588,
      "learning_rate": 3.8356597313668684e-05,
      "loss": 1.3535,
      "step": 8849
    },
    {
      "epoch": 0.23307874637872003,
      "grad_norm": 1.699992299079895,
      "learning_rate": 3.8355280484593106e-05,
      "loss": 1.9854,
      "step": 8850
    },
    {
      "epoch": 0.23310508296023177,
      "grad_norm": 3.062504768371582,
      "learning_rate": 3.8353963655517515e-05,
      "loss": 0.6641,
      "step": 8851
    },
    {
      "epoch": 0.2331314195417435,
      "grad_norm": 2.3903961181640625,
      "learning_rate": 3.835264682644193e-05,
      "loss": 1.7965,
      "step": 8852
    },
    {
      "epoch": 0.2331577561232552,
      "grad_norm": 2.542442560195923,
      "learning_rate": 3.835132999736634e-05,
      "loss": 2.3973,
      "step": 8853
    },
    {
      "epoch": 0.23318409270476692,
      "grad_norm": 1.528419852256775,
      "learning_rate": 3.835001316829076e-05,
      "loss": 1.7073,
      "step": 8854
    },
    {
      "epoch": 0.23321042928627864,
      "grad_norm": 2.7661101818084717,
      "learning_rate": 3.834869633921517e-05,
      "loss": 2.2516,
      "step": 8855
    },
    {
      "epoch": 0.23323676586779035,
      "grad_norm": 4.172338962554932,
      "learning_rate": 3.8347379510139586e-05,
      "loss": 1.1142,
      "step": 8856
    },
    {
      "epoch": 0.23326310244930207,
      "grad_norm": 3.0811731815338135,
      "learning_rate": 3.8346062681064e-05,
      "loss": 1.9807,
      "step": 8857
    },
    {
      "epoch": 0.2332894390308138,
      "grad_norm": 1.539762020111084,
      "learning_rate": 3.834474585198841e-05,
      "loss": 1.7075,
      "step": 8858
    },
    {
      "epoch": 0.23331577561232553,
      "grad_norm": 2.513007879257202,
      "learning_rate": 3.834342902291283e-05,
      "loss": 0.744,
      "step": 8859
    },
    {
      "epoch": 0.23334211219383724,
      "grad_norm": 1.6982077360153198,
      "learning_rate": 3.834211219383724e-05,
      "loss": 1.8962,
      "step": 8860
    },
    {
      "epoch": 0.23336844877534896,
      "grad_norm": 4.641372203826904,
      "learning_rate": 3.834079536476166e-05,
      "loss": 1.4002,
      "step": 8861
    },
    {
      "epoch": 0.23339478535686067,
      "grad_norm": 1.8269281387329102,
      "learning_rate": 3.8339478535686066e-05,
      "loss": 1.7396,
      "step": 8862
    },
    {
      "epoch": 0.2334211219383724,
      "grad_norm": 2.211444139480591,
      "learning_rate": 3.833816170661048e-05,
      "loss": 1.7778,
      "step": 8863
    },
    {
      "epoch": 0.23344745851988413,
      "grad_norm": 1.9364101886749268,
      "learning_rate": 3.83368448775349e-05,
      "loss": 0.6063,
      "step": 8864
    },
    {
      "epoch": 0.23347379510139585,
      "grad_norm": 2.820455312728882,
      "learning_rate": 3.833552804845931e-05,
      "loss": 2.2119,
      "step": 8865
    },
    {
      "epoch": 0.23350013168290756,
      "grad_norm": 1.9101132154464722,
      "learning_rate": 3.833421121938373e-05,
      "loss": 0.9698,
      "step": 8866
    },
    {
      "epoch": 0.23352646826441928,
      "grad_norm": 1.4153937101364136,
      "learning_rate": 3.833289439030814e-05,
      "loss": 2.0112,
      "step": 8867
    },
    {
      "epoch": 0.233552804845931,
      "grad_norm": 1.9030933380126953,
      "learning_rate": 3.833157756123255e-05,
      "loss": 1.5937,
      "step": 8868
    },
    {
      "epoch": 0.2335791414274427,
      "grad_norm": 1.5382078886032104,
      "learning_rate": 3.833026073215697e-05,
      "loss": 1.6992,
      "step": 8869
    },
    {
      "epoch": 0.23360547800895443,
      "grad_norm": 1.5609915256500244,
      "learning_rate": 3.8328943903081384e-05,
      "loss": 1.6782,
      "step": 8870
    },
    {
      "epoch": 0.23363181459046617,
      "grad_norm": 3.965820789337158,
      "learning_rate": 3.832762707400579e-05,
      "loss": 1.6336,
      "step": 8871
    },
    {
      "epoch": 0.23365815117197788,
      "grad_norm": 3.165330648422241,
      "learning_rate": 3.832631024493021e-05,
      "loss": 1.5335,
      "step": 8872
    },
    {
      "epoch": 0.2336844877534896,
      "grad_norm": 2.4327380657196045,
      "learning_rate": 3.8324993415854624e-05,
      "loss": 1.2543,
      "step": 8873
    },
    {
      "epoch": 0.23371082433500132,
      "grad_norm": 2.084284543991089,
      "learning_rate": 3.832367658677904e-05,
      "loss": 2.1076,
      "step": 8874
    },
    {
      "epoch": 0.23373716091651303,
      "grad_norm": 1.7423182725906372,
      "learning_rate": 3.8322359757703455e-05,
      "loss": 2.0306,
      "step": 8875
    },
    {
      "epoch": 0.23376349749802475,
      "grad_norm": 2.029839515686035,
      "learning_rate": 3.8321042928627864e-05,
      "loss": 0.688,
      "step": 8876
    },
    {
      "epoch": 0.23378983407953646,
      "grad_norm": 1.9252088069915771,
      "learning_rate": 3.831972609955228e-05,
      "loss": 0.8262,
      "step": 8877
    },
    {
      "epoch": 0.2338161706610482,
      "grad_norm": 3.151832342147827,
      "learning_rate": 3.8318409270476695e-05,
      "loss": 1.7748,
      "step": 8878
    },
    {
      "epoch": 0.23384250724255992,
      "grad_norm": 2.617858409881592,
      "learning_rate": 3.831709244140111e-05,
      "loss": 2.2719,
      "step": 8879
    },
    {
      "epoch": 0.23386884382407164,
      "grad_norm": 2.3172693252563477,
      "learning_rate": 3.831577561232552e-05,
      "loss": 1.9707,
      "step": 8880
    },
    {
      "epoch": 0.23389518040558335,
      "grad_norm": 4.9011921882629395,
      "learning_rate": 3.8314458783249935e-05,
      "loss": 1.8664,
      "step": 8881
    },
    {
      "epoch": 0.23392151698709507,
      "grad_norm": 2.3205227851867676,
      "learning_rate": 3.8313141954174344e-05,
      "loss": 0.7283,
      "step": 8882
    },
    {
      "epoch": 0.23394785356860678,
      "grad_norm": 2.58687424659729,
      "learning_rate": 3.831182512509877e-05,
      "loss": 1.4471,
      "step": 8883
    },
    {
      "epoch": 0.23397419015011853,
      "grad_norm": 4.4518938064575195,
      "learning_rate": 3.8310508296023175e-05,
      "loss": 1.2096,
      "step": 8884
    },
    {
      "epoch": 0.23400052673163024,
      "grad_norm": 4.318954944610596,
      "learning_rate": 3.830919146694759e-05,
      "loss": 1.1394,
      "step": 8885
    },
    {
      "epoch": 0.23402686331314196,
      "grad_norm": 1.8930089473724365,
      "learning_rate": 3.830787463787201e-05,
      "loss": 1.5532,
      "step": 8886
    },
    {
      "epoch": 0.23405319989465367,
      "grad_norm": 1.6042455434799194,
      "learning_rate": 3.830655780879642e-05,
      "loss": 2.1812,
      "step": 8887
    },
    {
      "epoch": 0.2340795364761654,
      "grad_norm": 1.786065697669983,
      "learning_rate": 3.830524097972084e-05,
      "loss": 2.0149,
      "step": 8888
    },
    {
      "epoch": 0.2341058730576771,
      "grad_norm": 3.1769027709960938,
      "learning_rate": 3.830392415064525e-05,
      "loss": 2.0626,
      "step": 8889
    },
    {
      "epoch": 0.23413220963918882,
      "grad_norm": 3.0187506675720215,
      "learning_rate": 3.830260732156966e-05,
      "loss": 1.7158,
      "step": 8890
    },
    {
      "epoch": 0.23415854622070056,
      "grad_norm": 2.120159387588501,
      "learning_rate": 3.830129049249407e-05,
      "loss": 0.3464,
      "step": 8891
    },
    {
      "epoch": 0.23418488280221228,
      "grad_norm": 4.124617576599121,
      "learning_rate": 3.8299973663418493e-05,
      "loss": 0.9884,
      "step": 8892
    },
    {
      "epoch": 0.234211219383724,
      "grad_norm": 1.733627438545227,
      "learning_rate": 3.82986568343429e-05,
      "loss": 0.7134,
      "step": 8893
    },
    {
      "epoch": 0.2342375559652357,
      "grad_norm": 2.665647268295288,
      "learning_rate": 3.829734000526732e-05,
      "loss": 1.8067,
      "step": 8894
    },
    {
      "epoch": 0.23426389254674743,
      "grad_norm": 1.5547395944595337,
      "learning_rate": 3.8296023176191734e-05,
      "loss": 0.5754,
      "step": 8895
    },
    {
      "epoch": 0.23429022912825914,
      "grad_norm": 1.867587685585022,
      "learning_rate": 3.829470634711614e-05,
      "loss": 2.0367,
      "step": 8896
    },
    {
      "epoch": 0.23431656570977089,
      "grad_norm": 3.2139203548431396,
      "learning_rate": 3.8293389518040565e-05,
      "loss": 0.7397,
      "step": 8897
    },
    {
      "epoch": 0.2343429022912826,
      "grad_norm": 1.5913513898849487,
      "learning_rate": 3.8292072688964974e-05,
      "loss": 0.8447,
      "step": 8898
    },
    {
      "epoch": 0.23436923887279432,
      "grad_norm": 1.602441668510437,
      "learning_rate": 3.829075585988939e-05,
      "loss": 1.5207,
      "step": 8899
    },
    {
      "epoch": 0.23439557545430603,
      "grad_norm": 1.5813241004943848,
      "learning_rate": 3.82894390308138e-05,
      "loss": 1.8338,
      "step": 8900
    },
    {
      "epoch": 0.23442191203581775,
      "grad_norm": 2.0932424068450928,
      "learning_rate": 3.8288122201738214e-05,
      "loss": 2.3471,
      "step": 8901
    },
    {
      "epoch": 0.23444824861732946,
      "grad_norm": 2.052216053009033,
      "learning_rate": 3.828680537266263e-05,
      "loss": 2.3069,
      "step": 8902
    },
    {
      "epoch": 0.23447458519884118,
      "grad_norm": 3.160872220993042,
      "learning_rate": 3.8285488543587045e-05,
      "loss": 1.7844,
      "step": 8903
    },
    {
      "epoch": 0.23450092178035292,
      "grad_norm": 3.706263780593872,
      "learning_rate": 3.828417171451146e-05,
      "loss": 1.4202,
      "step": 8904
    },
    {
      "epoch": 0.23452725836186464,
      "grad_norm": 2.0213115215301514,
      "learning_rate": 3.828285488543587e-05,
      "loss": 2.0583,
      "step": 8905
    },
    {
      "epoch": 0.23455359494337635,
      "grad_norm": 2.1955199241638184,
      "learning_rate": 3.828153805636029e-05,
      "loss": 1.889,
      "step": 8906
    },
    {
      "epoch": 0.23457993152488807,
      "grad_norm": 1.8465187549591064,
      "learning_rate": 3.82802212272847e-05,
      "loss": 1.3723,
      "step": 8907
    },
    {
      "epoch": 0.23460626810639978,
      "grad_norm": 1.5337226390838623,
      "learning_rate": 3.8278904398209116e-05,
      "loss": 1.718,
      "step": 8908
    },
    {
      "epoch": 0.2346326046879115,
      "grad_norm": 2.3587608337402344,
      "learning_rate": 3.8277587569133525e-05,
      "loss": 2.153,
      "step": 8909
    },
    {
      "epoch": 0.23465894126942322,
      "grad_norm": 3.854475736618042,
      "learning_rate": 3.827627074005794e-05,
      "loss": 1.5479,
      "step": 8910
    },
    {
      "epoch": 0.23468527785093496,
      "grad_norm": 2.820453643798828,
      "learning_rate": 3.8274953910982356e-05,
      "loss": 0.7081,
      "step": 8911
    },
    {
      "epoch": 0.23471161443244667,
      "grad_norm": 1.948861002922058,
      "learning_rate": 3.827363708190677e-05,
      "loss": 2.0023,
      "step": 8912
    },
    {
      "epoch": 0.2347379510139584,
      "grad_norm": 2.404877185821533,
      "learning_rate": 3.827232025283119e-05,
      "loss": 2.1299,
      "step": 8913
    },
    {
      "epoch": 0.2347642875954701,
      "grad_norm": 2.2956411838531494,
      "learning_rate": 3.8271003423755596e-05,
      "loss": 1.4998,
      "step": 8914
    },
    {
      "epoch": 0.23479062417698182,
      "grad_norm": 3.2377097606658936,
      "learning_rate": 3.826968659468001e-05,
      "loss": 1.4843,
      "step": 8915
    },
    {
      "epoch": 0.23481696075849354,
      "grad_norm": 2.416879653930664,
      "learning_rate": 3.826836976560443e-05,
      "loss": 1.8884,
      "step": 8916
    },
    {
      "epoch": 0.23484329734000528,
      "grad_norm": 3.0116682052612305,
      "learning_rate": 3.826705293652884e-05,
      "loss": 2.5791,
      "step": 8917
    },
    {
      "epoch": 0.234869633921517,
      "grad_norm": 4.6304612159729,
      "learning_rate": 3.826573610745325e-05,
      "loss": 1.0188,
      "step": 8918
    },
    {
      "epoch": 0.2348959705030287,
      "grad_norm": 3.1109516620635986,
      "learning_rate": 3.826441927837767e-05,
      "loss": 2.2889,
      "step": 8919
    },
    {
      "epoch": 0.23492230708454043,
      "grad_norm": 2.2072911262512207,
      "learning_rate": 3.826310244930208e-05,
      "loss": 2.4439,
      "step": 8920
    },
    {
      "epoch": 0.23494864366605214,
      "grad_norm": 3.9602625370025635,
      "learning_rate": 3.82617856202265e-05,
      "loss": 1.5081,
      "step": 8921
    },
    {
      "epoch": 0.23497498024756386,
      "grad_norm": 2.291158437728882,
      "learning_rate": 3.8260468791150914e-05,
      "loss": 1.6707,
      "step": 8922
    },
    {
      "epoch": 0.23500131682907557,
      "grad_norm": 4.118983745574951,
      "learning_rate": 3.825915196207532e-05,
      "loss": 2.3479,
      "step": 8923
    },
    {
      "epoch": 0.23502765341058732,
      "grad_norm": 1.745460867881775,
      "learning_rate": 3.825783513299974e-05,
      "loss": 1.7726,
      "step": 8924
    },
    {
      "epoch": 0.23505398999209903,
      "grad_norm": 1.618269920349121,
      "learning_rate": 3.8256518303924154e-05,
      "loss": 1.6318,
      "step": 8925
    },
    {
      "epoch": 0.23508032657361075,
      "grad_norm": 4.956629276275635,
      "learning_rate": 3.825520147484857e-05,
      "loss": 2.3063,
      "step": 8926
    },
    {
      "epoch": 0.23510666315512246,
      "grad_norm": 3.23354434967041,
      "learning_rate": 3.825388464577298e-05,
      "loss": 1.2218,
      "step": 8927
    },
    {
      "epoch": 0.23513299973663418,
      "grad_norm": 1.6146546602249146,
      "learning_rate": 3.8252567816697394e-05,
      "loss": 1.593,
      "step": 8928
    },
    {
      "epoch": 0.2351593363181459,
      "grad_norm": 1.7741047143936157,
      "learning_rate": 3.82512509876218e-05,
      "loss": 1.7241,
      "step": 8929
    },
    {
      "epoch": 0.2351856728996576,
      "grad_norm": 2.6417601108551025,
      "learning_rate": 3.8249934158546225e-05,
      "loss": 0.5566,
      "step": 8930
    },
    {
      "epoch": 0.23521200948116935,
      "grad_norm": 1.7497538328170776,
      "learning_rate": 3.8248617329470634e-05,
      "loss": 1.2505,
      "step": 8931
    },
    {
      "epoch": 0.23523834606268107,
      "grad_norm": 6.498108386993408,
      "learning_rate": 3.824730050039505e-05,
      "loss": 1.7914,
      "step": 8932
    },
    {
      "epoch": 0.23526468264419279,
      "grad_norm": 3.6201765537261963,
      "learning_rate": 3.8245983671319465e-05,
      "loss": 1.905,
      "step": 8933
    },
    {
      "epoch": 0.2352910192257045,
      "grad_norm": 1.7686960697174072,
      "learning_rate": 3.8244666842243874e-05,
      "loss": 2.0151,
      "step": 8934
    },
    {
      "epoch": 0.23531735580721622,
      "grad_norm": 2.550863027572632,
      "learning_rate": 3.8243350013168296e-05,
      "loss": 1.9397,
      "step": 8935
    },
    {
      "epoch": 0.23534369238872793,
      "grad_norm": 2.5576789379119873,
      "learning_rate": 3.8242033184092705e-05,
      "loss": 0.429,
      "step": 8936
    },
    {
      "epoch": 0.23537002897023968,
      "grad_norm": 2.9477245807647705,
      "learning_rate": 3.824071635501712e-05,
      "loss": 1.4103,
      "step": 8937
    },
    {
      "epoch": 0.2353963655517514,
      "grad_norm": 4.0079498291015625,
      "learning_rate": 3.823939952594153e-05,
      "loss": 1.0337,
      "step": 8938
    },
    {
      "epoch": 0.2354227021332631,
      "grad_norm": 1.8713847398757935,
      "learning_rate": 3.823808269686595e-05,
      "loss": 2.0018,
      "step": 8939
    },
    {
      "epoch": 0.23544903871477482,
      "grad_norm": 1.8662177324295044,
      "learning_rate": 3.823676586779036e-05,
      "loss": 1.5008,
      "step": 8940
    },
    {
      "epoch": 0.23547537529628654,
      "grad_norm": 3.5579030513763428,
      "learning_rate": 3.8235449038714776e-05,
      "loss": 1.049,
      "step": 8941
    },
    {
      "epoch": 0.23550171187779825,
      "grad_norm": 1.962793231010437,
      "learning_rate": 3.823413220963919e-05,
      "loss": 1.3536,
      "step": 8942
    },
    {
      "epoch": 0.23552804845930997,
      "grad_norm": 3.0141546726226807,
      "learning_rate": 3.82328153805636e-05,
      "loss": 1.6883,
      "step": 8943
    },
    {
      "epoch": 0.2355543850408217,
      "grad_norm": 2.788313388824463,
      "learning_rate": 3.823149855148802e-05,
      "loss": 1.6545,
      "step": 8944
    },
    {
      "epoch": 0.23558072162233343,
      "grad_norm": 2.075899600982666,
      "learning_rate": 3.823018172241243e-05,
      "loss": 1.3079,
      "step": 8945
    },
    {
      "epoch": 0.23560705820384514,
      "grad_norm": 5.5302734375,
      "learning_rate": 3.822886489333685e-05,
      "loss": 1.0914,
      "step": 8946
    },
    {
      "epoch": 0.23563339478535686,
      "grad_norm": 5.843254566192627,
      "learning_rate": 3.8227548064261257e-05,
      "loss": 1.6165,
      "step": 8947
    },
    {
      "epoch": 0.23565973136686857,
      "grad_norm": 2.9635374546051025,
      "learning_rate": 3.822623123518567e-05,
      "loss": 1.4004,
      "step": 8948
    },
    {
      "epoch": 0.2356860679483803,
      "grad_norm": 2.420072317123413,
      "learning_rate": 3.822491440611009e-05,
      "loss": 1.0397,
      "step": 8949
    },
    {
      "epoch": 0.23571240452989203,
      "grad_norm": 2.560356855392456,
      "learning_rate": 3.82235975770345e-05,
      "loss": 1.0354,
      "step": 8950
    },
    {
      "epoch": 0.23573874111140375,
      "grad_norm": 3.587472438812256,
      "learning_rate": 3.822228074795892e-05,
      "loss": 0.5169,
      "step": 8951
    },
    {
      "epoch": 0.23576507769291546,
      "grad_norm": 3.9375946521759033,
      "learning_rate": 3.822096391888333e-05,
      "loss": 2.734,
      "step": 8952
    },
    {
      "epoch": 0.23579141427442718,
      "grad_norm": 2.2739546298980713,
      "learning_rate": 3.821964708980775e-05,
      "loss": 1.3848,
      "step": 8953
    },
    {
      "epoch": 0.2358177508559389,
      "grad_norm": 2.0366899967193604,
      "learning_rate": 3.821833026073216e-05,
      "loss": 1.4427,
      "step": 8954
    },
    {
      "epoch": 0.2358440874374506,
      "grad_norm": 1.8842791318893433,
      "learning_rate": 3.8217013431656575e-05,
      "loss": 1.8446,
      "step": 8955
    },
    {
      "epoch": 0.23587042401896233,
      "grad_norm": 1.8873313665390015,
      "learning_rate": 3.821569660258098e-05,
      "loss": 1.6234,
      "step": 8956
    },
    {
      "epoch": 0.23589676060047407,
      "grad_norm": 2.7381744384765625,
      "learning_rate": 3.82143797735054e-05,
      "loss": 0.8085,
      "step": 8957
    },
    {
      "epoch": 0.23592309718198579,
      "grad_norm": 3.809654712677002,
      "learning_rate": 3.8213062944429815e-05,
      "loss": 0.6761,
      "step": 8958
    },
    {
      "epoch": 0.2359494337634975,
      "grad_norm": 2.1754937171936035,
      "learning_rate": 3.821174611535423e-05,
      "loss": 1.44,
      "step": 8959
    },
    {
      "epoch": 0.23597577034500922,
      "grad_norm": 3.9207212924957275,
      "learning_rate": 3.8210429286278646e-05,
      "loss": 0.7976,
      "step": 8960
    },
    {
      "epoch": 0.23600210692652093,
      "grad_norm": 3.327483892440796,
      "learning_rate": 3.8209112457203055e-05,
      "loss": 0.4344,
      "step": 8961
    },
    {
      "epoch": 0.23602844350803265,
      "grad_norm": 3.8254613876342773,
      "learning_rate": 3.820779562812747e-05,
      "loss": 1.1983,
      "step": 8962
    },
    {
      "epoch": 0.23605478008954436,
      "grad_norm": 1.8120373487472534,
      "learning_rate": 3.8206478799051886e-05,
      "loss": 1.789,
      "step": 8963
    },
    {
      "epoch": 0.2360811166710561,
      "grad_norm": 2.8455874919891357,
      "learning_rate": 3.82051619699763e-05,
      "loss": 1.2715,
      "step": 8964
    },
    {
      "epoch": 0.23610745325256782,
      "grad_norm": 2.352724552154541,
      "learning_rate": 3.820384514090071e-05,
      "loss": 1.5288,
      "step": 8965
    },
    {
      "epoch": 0.23613378983407954,
      "grad_norm": 3.0580079555511475,
      "learning_rate": 3.8202528311825126e-05,
      "loss": 2.1386,
      "step": 8966
    },
    {
      "epoch": 0.23616012641559125,
      "grad_norm": 3.5651791095733643,
      "learning_rate": 3.820121148274954e-05,
      "loss": 2.6831,
      "step": 8967
    },
    {
      "epoch": 0.23618646299710297,
      "grad_norm": 3.2723565101623535,
      "learning_rate": 3.819989465367396e-05,
      "loss": 1.5994,
      "step": 8968
    },
    {
      "epoch": 0.23621279957861469,
      "grad_norm": 1.4684089422225952,
      "learning_rate": 3.819857782459837e-05,
      "loss": 2.1179,
      "step": 8969
    },
    {
      "epoch": 0.23623913616012643,
      "grad_norm": 1.7993519306182861,
      "learning_rate": 3.819726099552278e-05,
      "loss": 1.7187,
      "step": 8970
    },
    {
      "epoch": 0.23626547274163814,
      "grad_norm": 2.4115402698516846,
      "learning_rate": 3.81959441664472e-05,
      "loss": 1.4155,
      "step": 8971
    },
    {
      "epoch": 0.23629180932314986,
      "grad_norm": 1.740483045578003,
      "learning_rate": 3.819462733737161e-05,
      "loss": 1.9074,
      "step": 8972
    },
    {
      "epoch": 0.23631814590466158,
      "grad_norm": 1.5675009489059448,
      "learning_rate": 3.819331050829603e-05,
      "loss": 1.7608,
      "step": 8973
    },
    {
      "epoch": 0.2363444824861733,
      "grad_norm": 2.571057081222534,
      "learning_rate": 3.819199367922044e-05,
      "loss": 1.5261,
      "step": 8974
    },
    {
      "epoch": 0.236370819067685,
      "grad_norm": 3.5776636600494385,
      "learning_rate": 3.819067685014485e-05,
      "loss": 0.8671,
      "step": 8975
    },
    {
      "epoch": 0.23639715564919672,
      "grad_norm": 2.386263132095337,
      "learning_rate": 3.818936002106927e-05,
      "loss": 2.2032,
      "step": 8976
    },
    {
      "epoch": 0.23642349223070847,
      "grad_norm": 1.9783543348312378,
      "learning_rate": 3.8188043191993684e-05,
      "loss": 1.4508,
      "step": 8977
    },
    {
      "epoch": 0.23644982881222018,
      "grad_norm": 1.8393594026565552,
      "learning_rate": 3.81867263629181e-05,
      "loss": 0.7998,
      "step": 8978
    },
    {
      "epoch": 0.2364761653937319,
      "grad_norm": 2.0811891555786133,
      "learning_rate": 3.818540953384251e-05,
      "loss": 1.763,
      "step": 8979
    },
    {
      "epoch": 0.2365025019752436,
      "grad_norm": 3.1697185039520264,
      "learning_rate": 3.8184092704766924e-05,
      "loss": 0.9507,
      "step": 8980
    },
    {
      "epoch": 0.23652883855675533,
      "grad_norm": 2.105729579925537,
      "learning_rate": 3.818277587569133e-05,
      "loss": 1.6966,
      "step": 8981
    },
    {
      "epoch": 0.23655517513826704,
      "grad_norm": 3.137927770614624,
      "learning_rate": 3.8181459046615755e-05,
      "loss": 0.5815,
      "step": 8982
    },
    {
      "epoch": 0.2365815117197788,
      "grad_norm": 1.87382173538208,
      "learning_rate": 3.8180142217540164e-05,
      "loss": 1.9441,
      "step": 8983
    },
    {
      "epoch": 0.2366078483012905,
      "grad_norm": 1.5987637042999268,
      "learning_rate": 3.817882538846458e-05,
      "loss": 1.7637,
      "step": 8984
    },
    {
      "epoch": 0.23663418488280222,
      "grad_norm": 1.9208414554595947,
      "learning_rate": 3.817750855938899e-05,
      "loss": 1.9402,
      "step": 8985
    },
    {
      "epoch": 0.23666052146431393,
      "grad_norm": 1.628235936164856,
      "learning_rate": 3.817619173031341e-05,
      "loss": 0.525,
      "step": 8986
    },
    {
      "epoch": 0.23668685804582565,
      "grad_norm": 2.460887908935547,
      "learning_rate": 3.817487490123782e-05,
      "loss": 2.2777,
      "step": 8987
    },
    {
      "epoch": 0.23671319462733736,
      "grad_norm": 1.8031535148620605,
      "learning_rate": 3.8173558072162235e-05,
      "loss": 1.734,
      "step": 8988
    },
    {
      "epoch": 0.23673953120884908,
      "grad_norm": 2.9295225143432617,
      "learning_rate": 3.817224124308665e-05,
      "loss": 2.0813,
      "step": 8989
    },
    {
      "epoch": 0.23676586779036082,
      "grad_norm": 8.491218566894531,
      "learning_rate": 3.817092441401106e-05,
      "loss": 1.5099,
      "step": 8990
    },
    {
      "epoch": 0.23679220437187254,
      "grad_norm": 6.878724575042725,
      "learning_rate": 3.816960758493548e-05,
      "loss": 2.0645,
      "step": 8991
    },
    {
      "epoch": 0.23681854095338425,
      "grad_norm": 3.4732770919799805,
      "learning_rate": 3.816829075585989e-05,
      "loss": 1.4976,
      "step": 8992
    },
    {
      "epoch": 0.23684487753489597,
      "grad_norm": 3.7131786346435547,
      "learning_rate": 3.8166973926784306e-05,
      "loss": 2.8987,
      "step": 8993
    },
    {
      "epoch": 0.23687121411640769,
      "grad_norm": 1.7151122093200684,
      "learning_rate": 3.8165657097708715e-05,
      "loss": 0.5245,
      "step": 8994
    },
    {
      "epoch": 0.2368975506979194,
      "grad_norm": 3.2155251502990723,
      "learning_rate": 3.816434026863313e-05,
      "loss": 1.6148,
      "step": 8995
    },
    {
      "epoch": 0.23692388727943112,
      "grad_norm": 1.8275508880615234,
      "learning_rate": 3.8163023439557546e-05,
      "loss": 1.5996,
      "step": 8996
    },
    {
      "epoch": 0.23695022386094286,
      "grad_norm": 1.5089374780654907,
      "learning_rate": 3.816170661048196e-05,
      "loss": 1.6117,
      "step": 8997
    },
    {
      "epoch": 0.23697656044245458,
      "grad_norm": 1.5572309494018555,
      "learning_rate": 3.816038978140638e-05,
      "loss": 2.0907,
      "step": 8998
    },
    {
      "epoch": 0.2370028970239663,
      "grad_norm": 4.3987555503845215,
      "learning_rate": 3.8159072952330786e-05,
      "loss": 2.0462,
      "step": 8999
    },
    {
      "epoch": 0.237029233605478,
      "grad_norm": 3.0734729766845703,
      "learning_rate": 3.81577561232552e-05,
      "loss": 2.6626,
      "step": 9000
    },
    {
      "epoch": 0.23705557018698972,
      "grad_norm": 1.7914612293243408,
      "learning_rate": 3.815643929417962e-05,
      "loss": 1.7885,
      "step": 9001
    },
    {
      "epoch": 0.23708190676850144,
      "grad_norm": 1.5557209253311157,
      "learning_rate": 3.815512246510403e-05,
      "loss": 1.4949,
      "step": 9002
    },
    {
      "epoch": 0.23710824335001318,
      "grad_norm": 2.067309856414795,
      "learning_rate": 3.815380563602844e-05,
      "loss": 2.0708,
      "step": 9003
    },
    {
      "epoch": 0.2371345799315249,
      "grad_norm": 2.3373732566833496,
      "learning_rate": 3.815248880695286e-05,
      "loss": 2.0256,
      "step": 9004
    },
    {
      "epoch": 0.2371609165130366,
      "grad_norm": 4.035815238952637,
      "learning_rate": 3.815117197787727e-05,
      "loss": 1.4594,
      "step": 9005
    },
    {
      "epoch": 0.23718725309454833,
      "grad_norm": 1.917300820350647,
      "learning_rate": 3.814985514880169e-05,
      "loss": 0.3119,
      "step": 9006
    },
    {
      "epoch": 0.23721358967606004,
      "grad_norm": 1.7235420942306519,
      "learning_rate": 3.8148538319726104e-05,
      "loss": 1.7185,
      "step": 9007
    },
    {
      "epoch": 0.23723992625757176,
      "grad_norm": 4.063268184661865,
      "learning_rate": 3.814722149065051e-05,
      "loss": 1.0906,
      "step": 9008
    },
    {
      "epoch": 0.23726626283908347,
      "grad_norm": 2.1704537868499756,
      "learning_rate": 3.814590466157493e-05,
      "loss": 1.584,
      "step": 9009
    },
    {
      "epoch": 0.23729259942059522,
      "grad_norm": 1.6469006538391113,
      "learning_rate": 3.8144587832499344e-05,
      "loss": 1.5391,
      "step": 9010
    },
    {
      "epoch": 0.23731893600210693,
      "grad_norm": 2.907432794570923,
      "learning_rate": 3.814327100342376e-05,
      "loss": 1.5396,
      "step": 9011
    },
    {
      "epoch": 0.23734527258361865,
      "grad_norm": 3.2695088386535645,
      "learning_rate": 3.814195417434817e-05,
      "loss": 1.2955,
      "step": 9012
    },
    {
      "epoch": 0.23737160916513037,
      "grad_norm": 5.305758953094482,
      "learning_rate": 3.8140637345272584e-05,
      "loss": 1.2816,
      "step": 9013
    },
    {
      "epoch": 0.23739794574664208,
      "grad_norm": 3.164250373840332,
      "learning_rate": 3.8139320516197e-05,
      "loss": 1.4329,
      "step": 9014
    },
    {
      "epoch": 0.2374242823281538,
      "grad_norm": 2.0373001098632812,
      "learning_rate": 3.8138003687121416e-05,
      "loss": 1.4984,
      "step": 9015
    },
    {
      "epoch": 0.2374506189096655,
      "grad_norm": 2.999027729034424,
      "learning_rate": 3.813668685804583e-05,
      "loss": 1.3634,
      "step": 9016
    },
    {
      "epoch": 0.23747695549117726,
      "grad_norm": 2.648401975631714,
      "learning_rate": 3.813537002897024e-05,
      "loss": 1.6166,
      "step": 9017
    },
    {
      "epoch": 0.23750329207268897,
      "grad_norm": 1.7565820217132568,
      "learning_rate": 3.8134053199894656e-05,
      "loss": 1.6777,
      "step": 9018
    },
    {
      "epoch": 0.2375296286542007,
      "grad_norm": 2.587958335876465,
      "learning_rate": 3.813273637081907e-05,
      "loss": 0.6778,
      "step": 9019
    },
    {
      "epoch": 0.2375559652357124,
      "grad_norm": 2.393542766571045,
      "learning_rate": 3.813141954174349e-05,
      "loss": 1.7176,
      "step": 9020
    },
    {
      "epoch": 0.23758230181722412,
      "grad_norm": 4.26826810836792,
      "learning_rate": 3.8130102712667896e-05,
      "loss": 1.0747,
      "step": 9021
    },
    {
      "epoch": 0.23760863839873583,
      "grad_norm": 1.5846703052520752,
      "learning_rate": 3.812878588359231e-05,
      "loss": 1.8606,
      "step": 9022
    },
    {
      "epoch": 0.23763497498024758,
      "grad_norm": 2.3236591815948486,
      "learning_rate": 3.812746905451673e-05,
      "loss": 0.696,
      "step": 9023
    },
    {
      "epoch": 0.2376613115617593,
      "grad_norm": 2.3510401248931885,
      "learning_rate": 3.812615222544114e-05,
      "loss": 1.832,
      "step": 9024
    },
    {
      "epoch": 0.237687648143271,
      "grad_norm": 2.3777177333831787,
      "learning_rate": 3.812483539636556e-05,
      "loss": 2.0763,
      "step": 9025
    },
    {
      "epoch": 0.23771398472478272,
      "grad_norm": 5.4839301109313965,
      "learning_rate": 3.812351856728997e-05,
      "loss": 2.0837,
      "step": 9026
    },
    {
      "epoch": 0.23774032130629444,
      "grad_norm": 2.2859816551208496,
      "learning_rate": 3.812220173821438e-05,
      "loss": 1.4909,
      "step": 9027
    },
    {
      "epoch": 0.23776665788780615,
      "grad_norm": 1.9150323867797852,
      "learning_rate": 3.812088490913879e-05,
      "loss": 1.8763,
      "step": 9028
    },
    {
      "epoch": 0.23779299446931787,
      "grad_norm": 3.0372586250305176,
      "learning_rate": 3.8119568080063214e-05,
      "loss": 1.7709,
      "step": 9029
    },
    {
      "epoch": 0.2378193310508296,
      "grad_norm": 3.4950637817382812,
      "learning_rate": 3.811825125098762e-05,
      "loss": 1.5326,
      "step": 9030
    },
    {
      "epoch": 0.23784566763234133,
      "grad_norm": 2.4495930671691895,
      "learning_rate": 3.811693442191204e-05,
      "loss": 1.5065,
      "step": 9031
    },
    {
      "epoch": 0.23787200421385304,
      "grad_norm": 1.6274694204330444,
      "learning_rate": 3.811561759283645e-05,
      "loss": 1.608,
      "step": 9032
    },
    {
      "epoch": 0.23789834079536476,
      "grad_norm": 3.8316566944122314,
      "learning_rate": 3.811430076376086e-05,
      "loss": 1.9117,
      "step": 9033
    },
    {
      "epoch": 0.23792467737687648,
      "grad_norm": 2.6330342292785645,
      "learning_rate": 3.811298393468528e-05,
      "loss": 1.8568,
      "step": 9034
    },
    {
      "epoch": 0.2379510139583882,
      "grad_norm": 1.8126193284988403,
      "learning_rate": 3.8111667105609694e-05,
      "loss": 2.1386,
      "step": 9035
    },
    {
      "epoch": 0.23797735053989993,
      "grad_norm": 2.749074697494507,
      "learning_rate": 3.811035027653411e-05,
      "loss": 0.6549,
      "step": 9036
    },
    {
      "epoch": 0.23800368712141165,
      "grad_norm": 2.4599876403808594,
      "learning_rate": 3.810903344745852e-05,
      "loss": 1.6619,
      "step": 9037
    },
    {
      "epoch": 0.23803002370292337,
      "grad_norm": 2.7067179679870605,
      "learning_rate": 3.810771661838294e-05,
      "loss": 1.4006,
      "step": 9038
    },
    {
      "epoch": 0.23805636028443508,
      "grad_norm": 1.9752647876739502,
      "learning_rate": 3.810639978930735e-05,
      "loss": 0.3052,
      "step": 9039
    },
    {
      "epoch": 0.2380826968659468,
      "grad_norm": 1.9348949193954468,
      "learning_rate": 3.8105082960231765e-05,
      "loss": 2.0829,
      "step": 9040
    },
    {
      "epoch": 0.2381090334474585,
      "grad_norm": 11.67044448852539,
      "learning_rate": 3.8103766131156174e-05,
      "loss": 1.9523,
      "step": 9041
    },
    {
      "epoch": 0.23813537002897023,
      "grad_norm": 1.4356908798217773,
      "learning_rate": 3.810244930208059e-05,
      "loss": 0.6888,
      "step": 9042
    },
    {
      "epoch": 0.23816170661048197,
      "grad_norm": 3.7504501342773438,
      "learning_rate": 3.8101132473005005e-05,
      "loss": 1.1703,
      "step": 9043
    },
    {
      "epoch": 0.2381880431919937,
      "grad_norm": 3.1898488998413086,
      "learning_rate": 3.809981564392942e-05,
      "loss": 2.3678,
      "step": 9044
    },
    {
      "epoch": 0.2382143797735054,
      "grad_norm": 3.4472553730010986,
      "learning_rate": 3.8098498814853836e-05,
      "loss": 1.8964,
      "step": 9045
    },
    {
      "epoch": 0.23824071635501712,
      "grad_norm": 1.5964696407318115,
      "learning_rate": 3.8097181985778245e-05,
      "loss": 2.2231,
      "step": 9046
    },
    {
      "epoch": 0.23826705293652883,
      "grad_norm": 1.8006901741027832,
      "learning_rate": 3.809586515670266e-05,
      "loss": 1.6885,
      "step": 9047
    },
    {
      "epoch": 0.23829338951804055,
      "grad_norm": 2.045583963394165,
      "learning_rate": 3.8094548327627076e-05,
      "loss": 1.2975,
      "step": 9048
    },
    {
      "epoch": 0.23831972609955226,
      "grad_norm": 2.854395866394043,
      "learning_rate": 3.809323149855149e-05,
      "loss": 1.8723,
      "step": 9049
    },
    {
      "epoch": 0.238346062681064,
      "grad_norm": 1.7403329610824585,
      "learning_rate": 3.80919146694759e-05,
      "loss": 1.3867,
      "step": 9050
    },
    {
      "epoch": 0.23837239926257572,
      "grad_norm": 1.701168417930603,
      "learning_rate": 3.8090597840400316e-05,
      "loss": 2.1915,
      "step": 9051
    },
    {
      "epoch": 0.23839873584408744,
      "grad_norm": 1.7723575830459595,
      "learning_rate": 3.808928101132473e-05,
      "loss": 1.3636,
      "step": 9052
    },
    {
      "epoch": 0.23842507242559915,
      "grad_norm": 2.355898380279541,
      "learning_rate": 3.808796418224915e-05,
      "loss": 1.6345,
      "step": 9053
    },
    {
      "epoch": 0.23845140900711087,
      "grad_norm": 2.2978508472442627,
      "learning_rate": 3.808664735317356e-05,
      "loss": 1.7493,
      "step": 9054
    },
    {
      "epoch": 0.23847774558862259,
      "grad_norm": 3.296550750732422,
      "learning_rate": 3.808533052409797e-05,
      "loss": 1.6683,
      "step": 9055
    },
    {
      "epoch": 0.23850408217013433,
      "grad_norm": 2.169276475906372,
      "learning_rate": 3.808401369502239e-05,
      "loss": 1.7903,
      "step": 9056
    },
    {
      "epoch": 0.23853041875164605,
      "grad_norm": 1.586709976196289,
      "learning_rate": 3.80826968659468e-05,
      "loss": 1.8071,
      "step": 9057
    },
    {
      "epoch": 0.23855675533315776,
      "grad_norm": 2.4321131706237793,
      "learning_rate": 3.808138003687122e-05,
      "loss": 1.2764,
      "step": 9058
    },
    {
      "epoch": 0.23858309191466948,
      "grad_norm": 4.890425682067871,
      "learning_rate": 3.808006320779563e-05,
      "loss": 1.5407,
      "step": 9059
    },
    {
      "epoch": 0.2386094284961812,
      "grad_norm": 1.9176816940307617,
      "learning_rate": 3.807874637872004e-05,
      "loss": 0.7178,
      "step": 9060
    },
    {
      "epoch": 0.2386357650776929,
      "grad_norm": 1.8805367946624756,
      "learning_rate": 3.807742954964446e-05,
      "loss": 1.3999,
      "step": 9061
    },
    {
      "epoch": 0.23866210165920462,
      "grad_norm": 2.8186519145965576,
      "learning_rate": 3.8076112720568874e-05,
      "loss": 2.0395,
      "step": 9062
    },
    {
      "epoch": 0.23868843824071637,
      "grad_norm": 2.288559913635254,
      "learning_rate": 3.807479589149329e-05,
      "loss": 1.4923,
      "step": 9063
    },
    {
      "epoch": 0.23871477482222808,
      "grad_norm": 4.442562580108643,
      "learning_rate": 3.80734790624177e-05,
      "loss": 0.6598,
      "step": 9064
    },
    {
      "epoch": 0.2387411114037398,
      "grad_norm": 5.586864948272705,
      "learning_rate": 3.8072162233342114e-05,
      "loss": 1.0739,
      "step": 9065
    },
    {
      "epoch": 0.2387674479852515,
      "grad_norm": 2.1113388538360596,
      "learning_rate": 3.807084540426653e-05,
      "loss": 1.4191,
      "step": 9066
    },
    {
      "epoch": 0.23879378456676323,
      "grad_norm": 2.2378933429718018,
      "learning_rate": 3.8069528575190945e-05,
      "loss": 2.125,
      "step": 9067
    },
    {
      "epoch": 0.23882012114827494,
      "grad_norm": 2.3209145069122314,
      "learning_rate": 3.8068211746115354e-05,
      "loss": 1.3652,
      "step": 9068
    },
    {
      "epoch": 0.2388464577297867,
      "grad_norm": 1.9900540113449097,
      "learning_rate": 3.806689491703977e-05,
      "loss": 2.5439,
      "step": 9069
    },
    {
      "epoch": 0.2388727943112984,
      "grad_norm": 2.5610764026641846,
      "learning_rate": 3.8065578087964185e-05,
      "loss": 0.5374,
      "step": 9070
    },
    {
      "epoch": 0.23889913089281012,
      "grad_norm": 2.2821385860443115,
      "learning_rate": 3.80642612588886e-05,
      "loss": 1.256,
      "step": 9071
    },
    {
      "epoch": 0.23892546747432183,
      "grad_norm": 3.5710813999176025,
      "learning_rate": 3.8062944429813016e-05,
      "loss": 0.9633,
      "step": 9072
    },
    {
      "epoch": 0.23895180405583355,
      "grad_norm": 1.5225094556808472,
      "learning_rate": 3.8061627600737425e-05,
      "loss": 0.3832,
      "step": 9073
    },
    {
      "epoch": 0.23897814063734527,
      "grad_norm": 3.0547895431518555,
      "learning_rate": 3.806031077166184e-05,
      "loss": 1.1273,
      "step": 9074
    },
    {
      "epoch": 0.23900447721885698,
      "grad_norm": 2.001650333404541,
      "learning_rate": 3.805899394258625e-05,
      "loss": 1.6785,
      "step": 9075
    },
    {
      "epoch": 0.23903081380036872,
      "grad_norm": 3.8463213443756104,
      "learning_rate": 3.805767711351067e-05,
      "loss": 1.4484,
      "step": 9076
    },
    {
      "epoch": 0.23905715038188044,
      "grad_norm": 1.5718679428100586,
      "learning_rate": 3.805636028443508e-05,
      "loss": 1.4768,
      "step": 9077
    },
    {
      "epoch": 0.23908348696339216,
      "grad_norm": 2.100315809249878,
      "learning_rate": 3.8055043455359497e-05,
      "loss": 1.626,
      "step": 9078
    },
    {
      "epoch": 0.23910982354490387,
      "grad_norm": 3.065241575241089,
      "learning_rate": 3.8053726626283905e-05,
      "loss": 1.6768,
      "step": 9079
    },
    {
      "epoch": 0.2391361601264156,
      "grad_norm": 6.284860134124756,
      "learning_rate": 3.805240979720832e-05,
      "loss": 1.3801,
      "step": 9080
    },
    {
      "epoch": 0.2391624967079273,
      "grad_norm": 2.563067674636841,
      "learning_rate": 3.805109296813274e-05,
      "loss": 2.074,
      "step": 9081
    },
    {
      "epoch": 0.23918883328943902,
      "grad_norm": 2.1605827808380127,
      "learning_rate": 3.804977613905715e-05,
      "loss": 2.5389,
      "step": 9082
    },
    {
      "epoch": 0.23921516987095076,
      "grad_norm": 3.8266963958740234,
      "learning_rate": 3.804845930998157e-05,
      "loss": 1.9169,
      "step": 9083
    },
    {
      "epoch": 0.23924150645246248,
      "grad_norm": 4.3211188316345215,
      "learning_rate": 3.8047142480905977e-05,
      "loss": 1.9045,
      "step": 9084
    },
    {
      "epoch": 0.2392678430339742,
      "grad_norm": 2.3694863319396973,
      "learning_rate": 3.80458256518304e-05,
      "loss": 1.9605,
      "step": 9085
    },
    {
      "epoch": 0.2392941796154859,
      "grad_norm": 1.8015602827072144,
      "learning_rate": 3.804450882275481e-05,
      "loss": 1.8454,
      "step": 9086
    },
    {
      "epoch": 0.23932051619699762,
      "grad_norm": 2.8384194374084473,
      "learning_rate": 3.804319199367922e-05,
      "loss": 1.9281,
      "step": 9087
    },
    {
      "epoch": 0.23934685277850934,
      "grad_norm": 2.843345880508423,
      "learning_rate": 3.804187516460363e-05,
      "loss": 2.0674,
      "step": 9088
    },
    {
      "epoch": 0.23937318936002108,
      "grad_norm": 1.6297242641448975,
      "learning_rate": 3.804055833552805e-05,
      "loss": 2.4755,
      "step": 9089
    },
    {
      "epoch": 0.2393995259415328,
      "grad_norm": 3.193549633026123,
      "learning_rate": 3.803924150645246e-05,
      "loss": 1.8073,
      "step": 9090
    },
    {
      "epoch": 0.2394258625230445,
      "grad_norm": 3.4964730739593506,
      "learning_rate": 3.803792467737688e-05,
      "loss": 1.4154,
      "step": 9091
    },
    {
      "epoch": 0.23945219910455623,
      "grad_norm": 3.969574451446533,
      "learning_rate": 3.8036607848301295e-05,
      "loss": 0.9161,
      "step": 9092
    },
    {
      "epoch": 0.23947853568606794,
      "grad_norm": 1.7516297101974487,
      "learning_rate": 3.80352910192257e-05,
      "loss": 1.5023,
      "step": 9093
    },
    {
      "epoch": 0.23950487226757966,
      "grad_norm": 2.2985336780548096,
      "learning_rate": 3.803397419015012e-05,
      "loss": 2.0499,
      "step": 9094
    },
    {
      "epoch": 0.23953120884909138,
      "grad_norm": 2.7875607013702393,
      "learning_rate": 3.8032657361074535e-05,
      "loss": 1.8492,
      "step": 9095
    },
    {
      "epoch": 0.23955754543060312,
      "grad_norm": 1.9965405464172363,
      "learning_rate": 3.803134053199895e-05,
      "loss": 1.7533,
      "step": 9096
    },
    {
      "epoch": 0.23958388201211483,
      "grad_norm": 2.5655736923217773,
      "learning_rate": 3.803002370292336e-05,
      "loss": 1.4348,
      "step": 9097
    },
    {
      "epoch": 0.23961021859362655,
      "grad_norm": 2.612562656402588,
      "learning_rate": 3.8028706873847775e-05,
      "loss": 2.1023,
      "step": 9098
    },
    {
      "epoch": 0.23963655517513827,
      "grad_norm": 4.2543511390686035,
      "learning_rate": 3.802739004477219e-05,
      "loss": 1.4321,
      "step": 9099
    },
    {
      "epoch": 0.23966289175664998,
      "grad_norm": 2.7995800971984863,
      "learning_rate": 3.8026073215696606e-05,
      "loss": 1.1688,
      "step": 9100
    },
    {
      "epoch": 0.2396892283381617,
      "grad_norm": 3.423882484436035,
      "learning_rate": 3.802475638662102e-05,
      "loss": 2.5094,
      "step": 9101
    },
    {
      "epoch": 0.2397155649196734,
      "grad_norm": 3.353536605834961,
      "learning_rate": 3.802343955754543e-05,
      "loss": 1.7535,
      "step": 9102
    },
    {
      "epoch": 0.23974190150118516,
      "grad_norm": 1.7034701108932495,
      "learning_rate": 3.8022122728469846e-05,
      "loss": 0.8672,
      "step": 9103
    },
    {
      "epoch": 0.23976823808269687,
      "grad_norm": 3.1750571727752686,
      "learning_rate": 3.802080589939426e-05,
      "loss": 1.1865,
      "step": 9104
    },
    {
      "epoch": 0.2397945746642086,
      "grad_norm": 1.6175776720046997,
      "learning_rate": 3.801948907031868e-05,
      "loss": 1.6612,
      "step": 9105
    },
    {
      "epoch": 0.2398209112457203,
      "grad_norm": 3.0557985305786133,
      "learning_rate": 3.8018172241243086e-05,
      "loss": 0.9933,
      "step": 9106
    },
    {
      "epoch": 0.23984724782723202,
      "grad_norm": 2.8511931896209717,
      "learning_rate": 3.80168554121675e-05,
      "loss": 2.2315,
      "step": 9107
    },
    {
      "epoch": 0.23987358440874373,
      "grad_norm": 2.606637954711914,
      "learning_rate": 3.801553858309192e-05,
      "loss": 1.4886,
      "step": 9108
    },
    {
      "epoch": 0.23989992099025548,
      "grad_norm": 2.238206624984741,
      "learning_rate": 3.801422175401633e-05,
      "loss": 1.8346,
      "step": 9109
    },
    {
      "epoch": 0.2399262575717672,
      "grad_norm": 2.12542724609375,
      "learning_rate": 3.801290492494075e-05,
      "loss": 1.8932,
      "step": 9110
    },
    {
      "epoch": 0.2399525941532789,
      "grad_norm": 2.5104942321777344,
      "learning_rate": 3.801158809586516e-05,
      "loss": 1.3672,
      "step": 9111
    },
    {
      "epoch": 0.23997893073479062,
      "grad_norm": 2.029500961303711,
      "learning_rate": 3.801027126678957e-05,
      "loss": 2.0438,
      "step": 9112
    },
    {
      "epoch": 0.24000526731630234,
      "grad_norm": 2.992079019546509,
      "learning_rate": 3.800895443771398e-05,
      "loss": 1.8737,
      "step": 9113
    },
    {
      "epoch": 0.24003160389781406,
      "grad_norm": 3.4098570346832275,
      "learning_rate": 3.8007637608638404e-05,
      "loss": 1.3028,
      "step": 9114
    },
    {
      "epoch": 0.24005794047932577,
      "grad_norm": 5.301133632659912,
      "learning_rate": 3.800632077956281e-05,
      "loss": 1.7078,
      "step": 9115
    },
    {
      "epoch": 0.24008427706083751,
      "grad_norm": 1.9357210397720337,
      "learning_rate": 3.800500395048723e-05,
      "loss": 2.0063,
      "step": 9116
    },
    {
      "epoch": 0.24011061364234923,
      "grad_norm": 2.0835912227630615,
      "learning_rate": 3.8003687121411644e-05,
      "loss": 2.0374,
      "step": 9117
    },
    {
      "epoch": 0.24013695022386095,
      "grad_norm": 1.8628278970718384,
      "learning_rate": 3.800237029233606e-05,
      "loss": 1.9094,
      "step": 9118
    },
    {
      "epoch": 0.24016328680537266,
      "grad_norm": 1.7916760444641113,
      "learning_rate": 3.8001053463260475e-05,
      "loss": 1.6695,
      "step": 9119
    },
    {
      "epoch": 0.24018962338688438,
      "grad_norm": 2.5060653686523438,
      "learning_rate": 3.7999736634184884e-05,
      "loss": 0.8932,
      "step": 9120
    },
    {
      "epoch": 0.2402159599683961,
      "grad_norm": 1.9011561870574951,
      "learning_rate": 3.79984198051093e-05,
      "loss": 1.7803,
      "step": 9121
    },
    {
      "epoch": 0.24024229654990784,
      "grad_norm": 3.5035581588745117,
      "learning_rate": 3.799710297603371e-05,
      "loss": 0.4986,
      "step": 9122
    },
    {
      "epoch": 0.24026863313141955,
      "grad_norm": 4.221654415130615,
      "learning_rate": 3.799578614695813e-05,
      "loss": 1.03,
      "step": 9123
    },
    {
      "epoch": 0.24029496971293127,
      "grad_norm": 7.845032691955566,
      "learning_rate": 3.799446931788254e-05,
      "loss": 1.5312,
      "step": 9124
    },
    {
      "epoch": 0.24032130629444298,
      "grad_norm": 2.5314698219299316,
      "learning_rate": 3.7993152488806955e-05,
      "loss": 1.739,
      "step": 9125
    },
    {
      "epoch": 0.2403476428759547,
      "grad_norm": 1.9726166725158691,
      "learning_rate": 3.799183565973137e-05,
      "loss": 1.8529,
      "step": 9126
    },
    {
      "epoch": 0.2403739794574664,
      "grad_norm": 3.5129637718200684,
      "learning_rate": 3.799051883065578e-05,
      "loss": 1.5295,
      "step": 9127
    },
    {
      "epoch": 0.24040031603897813,
      "grad_norm": 2.7426881790161133,
      "learning_rate": 3.79892020015802e-05,
      "loss": 1.7939,
      "step": 9128
    },
    {
      "epoch": 0.24042665262048987,
      "grad_norm": 1.4880263805389404,
      "learning_rate": 3.798788517250461e-05,
      "loss": 1.3835,
      "step": 9129
    },
    {
      "epoch": 0.2404529892020016,
      "grad_norm": 2.7338085174560547,
      "learning_rate": 3.7986568343429026e-05,
      "loss": 0.5393,
      "step": 9130
    },
    {
      "epoch": 0.2404793257835133,
      "grad_norm": 1.945019245147705,
      "learning_rate": 3.7985251514353435e-05,
      "loss": 0.5236,
      "step": 9131
    },
    {
      "epoch": 0.24050566236502502,
      "grad_norm": 4.0086493492126465,
      "learning_rate": 3.798393468527786e-05,
      "loss": 2.0813,
      "step": 9132
    },
    {
      "epoch": 0.24053199894653673,
      "grad_norm": 2.984492301940918,
      "learning_rate": 3.7982617856202266e-05,
      "loss": 1.8398,
      "step": 9133
    },
    {
      "epoch": 0.24055833552804845,
      "grad_norm": 3.3578622341156006,
      "learning_rate": 3.798130102712668e-05,
      "loss": 0.9848,
      "step": 9134
    },
    {
      "epoch": 0.24058467210956017,
      "grad_norm": 2.6167993545532227,
      "learning_rate": 3.797998419805109e-05,
      "loss": 1.6624,
      "step": 9135
    },
    {
      "epoch": 0.2406110086910719,
      "grad_norm": 2.0487067699432373,
      "learning_rate": 3.7978667368975506e-05,
      "loss": 2.4291,
      "step": 9136
    },
    {
      "epoch": 0.24063734527258362,
      "grad_norm": 3.0543909072875977,
      "learning_rate": 3.797735053989992e-05,
      "loss": 1.7561,
      "step": 9137
    },
    {
      "epoch": 0.24066368185409534,
      "grad_norm": 5.187302112579346,
      "learning_rate": 3.797603371082434e-05,
      "loss": 1.6032,
      "step": 9138
    },
    {
      "epoch": 0.24069001843560706,
      "grad_norm": 1.904990553855896,
      "learning_rate": 3.797471688174875e-05,
      "loss": 1.9227,
      "step": 9139
    },
    {
      "epoch": 0.24071635501711877,
      "grad_norm": 1.8882147073745728,
      "learning_rate": 3.797340005267316e-05,
      "loss": 1.5641,
      "step": 9140
    },
    {
      "epoch": 0.2407426915986305,
      "grad_norm": 2.557469606399536,
      "learning_rate": 3.797208322359758e-05,
      "loss": 1.4592,
      "step": 9141
    },
    {
      "epoch": 0.24076902818014223,
      "grad_norm": 1.8143415451049805,
      "learning_rate": 3.797076639452199e-05,
      "loss": 1.685,
      "step": 9142
    },
    {
      "epoch": 0.24079536476165395,
      "grad_norm": 3.132756471633911,
      "learning_rate": 3.796944956544641e-05,
      "loss": 1.4478,
      "step": 9143
    },
    {
      "epoch": 0.24082170134316566,
      "grad_norm": 2.269442319869995,
      "learning_rate": 3.796813273637082e-05,
      "loss": 1.9974,
      "step": 9144
    },
    {
      "epoch": 0.24084803792467738,
      "grad_norm": 3.5727124214172363,
      "learning_rate": 3.796681590729523e-05,
      "loss": 0.6317,
      "step": 9145
    },
    {
      "epoch": 0.2408743745061891,
      "grad_norm": 3.152127504348755,
      "learning_rate": 3.796549907821965e-05,
      "loss": 1.6475,
      "step": 9146
    },
    {
      "epoch": 0.2409007110877008,
      "grad_norm": 2.6046741008758545,
      "learning_rate": 3.7964182249144064e-05,
      "loss": 1.1536,
      "step": 9147
    },
    {
      "epoch": 0.24092704766921252,
      "grad_norm": 1.647168517112732,
      "learning_rate": 3.796286542006848e-05,
      "loss": 1.7544,
      "step": 9148
    },
    {
      "epoch": 0.24095338425072427,
      "grad_norm": 3.2681217193603516,
      "learning_rate": 3.796154859099289e-05,
      "loss": 1.6074,
      "step": 9149
    },
    {
      "epoch": 0.24097972083223598,
      "grad_norm": 2.1700446605682373,
      "learning_rate": 3.7960231761917304e-05,
      "loss": 2.002,
      "step": 9150
    },
    {
      "epoch": 0.2410060574137477,
      "grad_norm": 3.1137824058532715,
      "learning_rate": 3.795891493284172e-05,
      "loss": 1.5384,
      "step": 9151
    },
    {
      "epoch": 0.24103239399525941,
      "grad_norm": 1.6794297695159912,
      "learning_rate": 3.7957598103766136e-05,
      "loss": 1.6418,
      "step": 9152
    },
    {
      "epoch": 0.24105873057677113,
      "grad_norm": 2.084293842315674,
      "learning_rate": 3.7956281274690544e-05,
      "loss": 0.8927,
      "step": 9153
    },
    {
      "epoch": 0.24108506715828285,
      "grad_norm": 1.805039644241333,
      "learning_rate": 3.795496444561496e-05,
      "loss": 1.6548,
      "step": 9154
    },
    {
      "epoch": 0.24111140373979456,
      "grad_norm": 1.7283471822738647,
      "learning_rate": 3.7953647616539376e-05,
      "loss": 1.5744,
      "step": 9155
    },
    {
      "epoch": 0.2411377403213063,
      "grad_norm": 1.9662374258041382,
      "learning_rate": 3.795233078746379e-05,
      "loss": 1.859,
      "step": 9156
    },
    {
      "epoch": 0.24116407690281802,
      "grad_norm": 2.372647523880005,
      "learning_rate": 3.795101395838821e-05,
      "loss": 1.816,
      "step": 9157
    },
    {
      "epoch": 0.24119041348432974,
      "grad_norm": 2.753685712814331,
      "learning_rate": 3.7949697129312616e-05,
      "loss": 1.5058,
      "step": 9158
    },
    {
      "epoch": 0.24121675006584145,
      "grad_norm": 1.8998364210128784,
      "learning_rate": 3.794838030023703e-05,
      "loss": 2.2558,
      "step": 9159
    },
    {
      "epoch": 0.24124308664735317,
      "grad_norm": 2.019319534301758,
      "learning_rate": 3.794706347116144e-05,
      "loss": 0.751,
      "step": 9160
    },
    {
      "epoch": 0.24126942322886488,
      "grad_norm": 1.8629621267318726,
      "learning_rate": 3.794574664208586e-05,
      "loss": 1.9928,
      "step": 9161
    },
    {
      "epoch": 0.24129575981037663,
      "grad_norm": 3.576493501663208,
      "learning_rate": 3.794442981301027e-05,
      "loss": 1.7328,
      "step": 9162
    },
    {
      "epoch": 0.24132209639188834,
      "grad_norm": 2.0093233585357666,
      "learning_rate": 3.794311298393469e-05,
      "loss": 1.6453,
      "step": 9163
    },
    {
      "epoch": 0.24134843297340006,
      "grad_norm": 1.61049222946167,
      "learning_rate": 3.79417961548591e-05,
      "loss": 0.5151,
      "step": 9164
    },
    {
      "epoch": 0.24137476955491177,
      "grad_norm": 4.014994144439697,
      "learning_rate": 3.794047932578352e-05,
      "loss": 0.8762,
      "step": 9165
    },
    {
      "epoch": 0.2414011061364235,
      "grad_norm": 1.7349263429641724,
      "learning_rate": 3.7939162496707934e-05,
      "loss": 1.9185,
      "step": 9166
    },
    {
      "epoch": 0.2414274427179352,
      "grad_norm": 2.0635013580322266,
      "learning_rate": 3.793784566763234e-05,
      "loss": 2.1196,
      "step": 9167
    },
    {
      "epoch": 0.24145377929944692,
      "grad_norm": 3.3577659130096436,
      "learning_rate": 3.793652883855676e-05,
      "loss": 1.9061,
      "step": 9168
    },
    {
      "epoch": 0.24148011588095866,
      "grad_norm": 7.884344100952148,
      "learning_rate": 3.793521200948117e-05,
      "loss": 1.7619,
      "step": 9169
    },
    {
      "epoch": 0.24150645246247038,
      "grad_norm": 3.6228582859039307,
      "learning_rate": 3.793389518040559e-05,
      "loss": 1.1273,
      "step": 9170
    },
    {
      "epoch": 0.2415327890439821,
      "grad_norm": 3.201294422149658,
      "learning_rate": 3.793257835133e-05,
      "loss": 2.2822,
      "step": 9171
    },
    {
      "epoch": 0.2415591256254938,
      "grad_norm": 2.4404683113098145,
      "learning_rate": 3.7931261522254414e-05,
      "loss": 1.8809,
      "step": 9172
    },
    {
      "epoch": 0.24158546220700552,
      "grad_norm": 1.6834548711776733,
      "learning_rate": 3.792994469317883e-05,
      "loss": 1.7893,
      "step": 9173
    },
    {
      "epoch": 0.24161179878851724,
      "grad_norm": 3.217729330062866,
      "learning_rate": 3.792862786410324e-05,
      "loss": 1.4802,
      "step": 9174
    },
    {
      "epoch": 0.24163813537002898,
      "grad_norm": 1.4206888675689697,
      "learning_rate": 3.792731103502766e-05,
      "loss": 0.2576,
      "step": 9175
    },
    {
      "epoch": 0.2416644719515407,
      "grad_norm": 1.8686909675598145,
      "learning_rate": 3.792599420595207e-05,
      "loss": 1.7188,
      "step": 9176
    },
    {
      "epoch": 0.24169080853305241,
      "grad_norm": 4.568958759307861,
      "learning_rate": 3.7924677376876485e-05,
      "loss": 1.3826,
      "step": 9177
    },
    {
      "epoch": 0.24171714511456413,
      "grad_norm": 2.1568028926849365,
      "learning_rate": 3.7923360547800894e-05,
      "loss": 1.4661,
      "step": 9178
    },
    {
      "epoch": 0.24174348169607585,
      "grad_norm": 2.121312379837036,
      "learning_rate": 3.792204371872531e-05,
      "loss": 1.9278,
      "step": 9179
    },
    {
      "epoch": 0.24176981827758756,
      "grad_norm": 2.118227243423462,
      "learning_rate": 3.7920726889649725e-05,
      "loss": 1.8126,
      "step": 9180
    },
    {
      "epoch": 0.24179615485909928,
      "grad_norm": 2.5075924396514893,
      "learning_rate": 3.791941006057414e-05,
      "loss": 0.182,
      "step": 9181
    },
    {
      "epoch": 0.24182249144061102,
      "grad_norm": 3.305026054382324,
      "learning_rate": 3.791809323149855e-05,
      "loss": 1.9408,
      "step": 9182
    },
    {
      "epoch": 0.24184882802212274,
      "grad_norm": 1.909716248512268,
      "learning_rate": 3.7916776402422965e-05,
      "loss": 1.8562,
      "step": 9183
    },
    {
      "epoch": 0.24187516460363445,
      "grad_norm": 2.0505053997039795,
      "learning_rate": 3.791545957334739e-05,
      "loss": 1.9769,
      "step": 9184
    },
    {
      "epoch": 0.24190150118514617,
      "grad_norm": 1.7961729764938354,
      "learning_rate": 3.7914142744271796e-05,
      "loss": 0.5204,
      "step": 9185
    },
    {
      "epoch": 0.24192783776665788,
      "grad_norm": 2.0554816722869873,
      "learning_rate": 3.791282591519621e-05,
      "loss": 1.4629,
      "step": 9186
    },
    {
      "epoch": 0.2419541743481696,
      "grad_norm": 2.257201910018921,
      "learning_rate": 3.791150908612062e-05,
      "loss": 1.7419,
      "step": 9187
    },
    {
      "epoch": 0.24198051092968131,
      "grad_norm": 1.7493021488189697,
      "learning_rate": 3.7910192257045036e-05,
      "loss": 1.8869,
      "step": 9188
    },
    {
      "epoch": 0.24200684751119306,
      "grad_norm": 1.585355281829834,
      "learning_rate": 3.790887542796945e-05,
      "loss": 2.3149,
      "step": 9189
    },
    {
      "epoch": 0.24203318409270477,
      "grad_norm": 2.6028037071228027,
      "learning_rate": 3.790755859889387e-05,
      "loss": 1.4708,
      "step": 9190
    },
    {
      "epoch": 0.2420595206742165,
      "grad_norm": 2.316624402999878,
      "learning_rate": 3.7906241769818276e-05,
      "loss": 1.5252,
      "step": 9191
    },
    {
      "epoch": 0.2420858572557282,
      "grad_norm": 3.5405890941619873,
      "learning_rate": 3.790492494074269e-05,
      "loss": 0.8296,
      "step": 9192
    },
    {
      "epoch": 0.24211219383723992,
      "grad_norm": 1.8235653638839722,
      "learning_rate": 3.790360811166711e-05,
      "loss": 2.1218,
      "step": 9193
    },
    {
      "epoch": 0.24213853041875164,
      "grad_norm": 1.585139274597168,
      "learning_rate": 3.790229128259152e-05,
      "loss": 2.1037,
      "step": 9194
    },
    {
      "epoch": 0.24216486700026338,
      "grad_norm": 2.2083747386932373,
      "learning_rate": 3.790097445351594e-05,
      "loss": 0.8379,
      "step": 9195
    },
    {
      "epoch": 0.2421912035817751,
      "grad_norm": 1.646804690361023,
      "learning_rate": 3.789965762444035e-05,
      "loss": 1.8582,
      "step": 9196
    },
    {
      "epoch": 0.2422175401632868,
      "grad_norm": 4.820605278015137,
      "learning_rate": 3.789834079536476e-05,
      "loss": 1.6665,
      "step": 9197
    },
    {
      "epoch": 0.24224387674479853,
      "grad_norm": 4.446354866027832,
      "learning_rate": 3.789702396628918e-05,
      "loss": 1.4955,
      "step": 9198
    },
    {
      "epoch": 0.24227021332631024,
      "grad_norm": 1.4681748151779175,
      "learning_rate": 3.7895707137213594e-05,
      "loss": 2.2789,
      "step": 9199
    },
    {
      "epoch": 0.24229654990782196,
      "grad_norm": 2.9111170768737793,
      "learning_rate": 3.7894390308138e-05,
      "loss": 2.0965,
      "step": 9200
    },
    {
      "epoch": 0.24232288648933367,
      "grad_norm": 3.720651865005493,
      "learning_rate": 3.789307347906242e-05,
      "loss": 2.1935,
      "step": 9201
    },
    {
      "epoch": 0.24234922307084542,
      "grad_norm": 2.03873872756958,
      "learning_rate": 3.7891756649986834e-05,
      "loss": 1.2195,
      "step": 9202
    },
    {
      "epoch": 0.24237555965235713,
      "grad_norm": 2.1937954425811768,
      "learning_rate": 3.789043982091125e-05,
      "loss": 2.0024,
      "step": 9203
    },
    {
      "epoch": 0.24240189623386885,
      "grad_norm": 1.8905243873596191,
      "learning_rate": 3.7889122991835665e-05,
      "loss": 1.5583,
      "step": 9204
    },
    {
      "epoch": 0.24242823281538056,
      "grad_norm": 1.5541455745697021,
      "learning_rate": 3.7887806162760074e-05,
      "loss": 2.2141,
      "step": 9205
    },
    {
      "epoch": 0.24245456939689228,
      "grad_norm": 3.704789400100708,
      "learning_rate": 3.788648933368449e-05,
      "loss": 0.8928,
      "step": 9206
    },
    {
      "epoch": 0.242480905978404,
      "grad_norm": 2.060554027557373,
      "learning_rate": 3.78851725046089e-05,
      "loss": 2.0226,
      "step": 9207
    },
    {
      "epoch": 0.24250724255991574,
      "grad_norm": 1.6431756019592285,
      "learning_rate": 3.788385567553332e-05,
      "loss": 2.1302,
      "step": 9208
    },
    {
      "epoch": 0.24253357914142745,
      "grad_norm": 3.4376893043518066,
      "learning_rate": 3.788253884645773e-05,
      "loss": 1.0592,
      "step": 9209
    },
    {
      "epoch": 0.24255991572293917,
      "grad_norm": 2.362325668334961,
      "learning_rate": 3.7881222017382145e-05,
      "loss": 1.1576,
      "step": 9210
    },
    {
      "epoch": 0.24258625230445088,
      "grad_norm": 2.5028932094573975,
      "learning_rate": 3.787990518830656e-05,
      "loss": 1.3836,
      "step": 9211
    },
    {
      "epoch": 0.2426125888859626,
      "grad_norm": 1.8605539798736572,
      "learning_rate": 3.787858835923097e-05,
      "loss": 2.4236,
      "step": 9212
    },
    {
      "epoch": 0.24263892546747431,
      "grad_norm": 4.813013076782227,
      "learning_rate": 3.787727153015539e-05,
      "loss": 1.0566,
      "step": 9213
    },
    {
      "epoch": 0.24266526204898603,
      "grad_norm": 2.4445302486419678,
      "learning_rate": 3.78759547010798e-05,
      "loss": 0.6525,
      "step": 9214
    },
    {
      "epoch": 0.24269159863049777,
      "grad_norm": 2.031694173812866,
      "learning_rate": 3.7874637872004217e-05,
      "loss": 1.6449,
      "step": 9215
    },
    {
      "epoch": 0.2427179352120095,
      "grad_norm": 1.2822418212890625,
      "learning_rate": 3.7873321042928625e-05,
      "loss": 1.7172,
      "step": 9216
    },
    {
      "epoch": 0.2427442717935212,
      "grad_norm": 2.2313852310180664,
      "learning_rate": 3.787200421385305e-05,
      "loss": 2.4019,
      "step": 9217
    },
    {
      "epoch": 0.24277060837503292,
      "grad_norm": 3.127763271331787,
      "learning_rate": 3.7870687384777457e-05,
      "loss": 1.0407,
      "step": 9218
    },
    {
      "epoch": 0.24279694495654464,
      "grad_norm": 2.701584815979004,
      "learning_rate": 3.786937055570187e-05,
      "loss": 1.8679,
      "step": 9219
    },
    {
      "epoch": 0.24282328153805635,
      "grad_norm": 2.3775875568389893,
      "learning_rate": 3.786805372662629e-05,
      "loss": 1.9762,
      "step": 9220
    },
    {
      "epoch": 0.24284961811956807,
      "grad_norm": 1.9174168109893799,
      "learning_rate": 3.7866736897550697e-05,
      "loss": 1.9555,
      "step": 9221
    },
    {
      "epoch": 0.2428759547010798,
      "grad_norm": 5.04189395904541,
      "learning_rate": 3.786542006847512e-05,
      "loss": 0.5557,
      "step": 9222
    },
    {
      "epoch": 0.24290229128259153,
      "grad_norm": 3.3648109436035156,
      "learning_rate": 3.786410323939953e-05,
      "loss": 1.2738,
      "step": 9223
    },
    {
      "epoch": 0.24292862786410324,
      "grad_norm": 3.704732894897461,
      "learning_rate": 3.7862786410323943e-05,
      "loss": 1.0398,
      "step": 9224
    },
    {
      "epoch": 0.24295496444561496,
      "grad_norm": 3.191092014312744,
      "learning_rate": 3.786146958124835e-05,
      "loss": 1.7896,
      "step": 9225
    },
    {
      "epoch": 0.24298130102712667,
      "grad_norm": 2.2654383182525635,
      "learning_rate": 3.786015275217277e-05,
      "loss": 2.048,
      "step": 9226
    },
    {
      "epoch": 0.2430076376086384,
      "grad_norm": 2.8884716033935547,
      "learning_rate": 3.7858835923097183e-05,
      "loss": 1.2261,
      "step": 9227
    },
    {
      "epoch": 0.24303397419015013,
      "grad_norm": 4.790567874908447,
      "learning_rate": 3.78575190940216e-05,
      "loss": 0.7408,
      "step": 9228
    },
    {
      "epoch": 0.24306031077166185,
      "grad_norm": 1.757062554359436,
      "learning_rate": 3.7856202264946015e-05,
      "loss": 2.1357,
      "step": 9229
    },
    {
      "epoch": 0.24308664735317356,
      "grad_norm": 2.4384217262268066,
      "learning_rate": 3.7854885435870423e-05,
      "loss": 1.7885,
      "step": 9230
    },
    {
      "epoch": 0.24311298393468528,
      "grad_norm": 1.7342662811279297,
      "learning_rate": 3.7853568606794846e-05,
      "loss": 1.6452,
      "step": 9231
    },
    {
      "epoch": 0.243139320516197,
      "grad_norm": 1.8292129039764404,
      "learning_rate": 3.7852251777719255e-05,
      "loss": 1.9006,
      "step": 9232
    },
    {
      "epoch": 0.2431656570977087,
      "grad_norm": 7.914848327636719,
      "learning_rate": 3.785093494864367e-05,
      "loss": 1.334,
      "step": 9233
    },
    {
      "epoch": 0.24319199367922043,
      "grad_norm": 2.721095085144043,
      "learning_rate": 3.784961811956808e-05,
      "loss": 0.6011,
      "step": 9234
    },
    {
      "epoch": 0.24321833026073217,
      "grad_norm": 2.089242458343506,
      "learning_rate": 3.7848301290492495e-05,
      "loss": 2.5129,
      "step": 9235
    },
    {
      "epoch": 0.24324466684224388,
      "grad_norm": 3.5150718688964844,
      "learning_rate": 3.784698446141691e-05,
      "loss": 0.7179,
      "step": 9236
    },
    {
      "epoch": 0.2432710034237556,
      "grad_norm": 1.794505000114441,
      "learning_rate": 3.7845667632341326e-05,
      "loss": 2.337,
      "step": 9237
    },
    {
      "epoch": 0.24329734000526732,
      "grad_norm": 2.0927414894104004,
      "learning_rate": 3.7844350803265735e-05,
      "loss": 1.1237,
      "step": 9238
    },
    {
      "epoch": 0.24332367658677903,
      "grad_norm": 1.8351795673370361,
      "learning_rate": 3.784303397419015e-05,
      "loss": 1.9189,
      "step": 9239
    },
    {
      "epoch": 0.24335001316829075,
      "grad_norm": 2.752796173095703,
      "learning_rate": 3.7841717145114566e-05,
      "loss": 2.0942,
      "step": 9240
    },
    {
      "epoch": 0.24337634974980246,
      "grad_norm": 3.8616297245025635,
      "learning_rate": 3.784040031603898e-05,
      "loss": 1.3218,
      "step": 9241
    },
    {
      "epoch": 0.2434026863313142,
      "grad_norm": 1.4050439596176147,
      "learning_rate": 3.78390834869634e-05,
      "loss": 1.6661,
      "step": 9242
    },
    {
      "epoch": 0.24342902291282592,
      "grad_norm": 6.63370943069458,
      "learning_rate": 3.7837766657887806e-05,
      "loss": 3.2083,
      "step": 9243
    },
    {
      "epoch": 0.24345535949433764,
      "grad_norm": 2.123335361480713,
      "learning_rate": 3.783644982881222e-05,
      "loss": 2.1397,
      "step": 9244
    },
    {
      "epoch": 0.24348169607584935,
      "grad_norm": 2.123399257659912,
      "learning_rate": 3.783513299973663e-05,
      "loss": 1.8308,
      "step": 9245
    },
    {
      "epoch": 0.24350803265736107,
      "grad_norm": 5.068130970001221,
      "learning_rate": 3.783381617066105e-05,
      "loss": 2.5819,
      "step": 9246
    },
    {
      "epoch": 0.24353436923887278,
      "grad_norm": 2.0146961212158203,
      "learning_rate": 3.783249934158546e-05,
      "loss": 0.486,
      "step": 9247
    },
    {
      "epoch": 0.24356070582038453,
      "grad_norm": 1.8296869993209839,
      "learning_rate": 3.783118251250988e-05,
      "loss": 1.52,
      "step": 9248
    },
    {
      "epoch": 0.24358704240189624,
      "grad_norm": 2.267693281173706,
      "learning_rate": 3.782986568343429e-05,
      "loss": 1.995,
      "step": 9249
    },
    {
      "epoch": 0.24361337898340796,
      "grad_norm": 2.441950798034668,
      "learning_rate": 3.782854885435871e-05,
      "loss": 1.3178,
      "step": 9250
    },
    {
      "epoch": 0.24363971556491967,
      "grad_norm": 1.6176522970199585,
      "learning_rate": 3.7827232025283124e-05,
      "loss": 1.6522,
      "step": 9251
    },
    {
      "epoch": 0.2436660521464314,
      "grad_norm": 2.1549603939056396,
      "learning_rate": 3.782591519620753e-05,
      "loss": 1.8736,
      "step": 9252
    },
    {
      "epoch": 0.2436923887279431,
      "grad_norm": 1.7482891082763672,
      "learning_rate": 3.782459836713195e-05,
      "loss": 1.644,
      "step": 9253
    },
    {
      "epoch": 0.24371872530945482,
      "grad_norm": 2.006420135498047,
      "learning_rate": 3.782328153805636e-05,
      "loss": 1.7266,
      "step": 9254
    },
    {
      "epoch": 0.24374506189096656,
      "grad_norm": 1.6657357215881348,
      "learning_rate": 3.782196470898078e-05,
      "loss": 1.2061,
      "step": 9255
    },
    {
      "epoch": 0.24377139847247828,
      "grad_norm": 1.837896704673767,
      "learning_rate": 3.782064787990519e-05,
      "loss": 1.5926,
      "step": 9256
    },
    {
      "epoch": 0.24379773505399,
      "grad_norm": 2.1787383556365967,
      "learning_rate": 3.7819331050829604e-05,
      "loss": 2.0773,
      "step": 9257
    },
    {
      "epoch": 0.2438240716355017,
      "grad_norm": 1.453203558921814,
      "learning_rate": 3.781801422175402e-05,
      "loss": 2.2414,
      "step": 9258
    },
    {
      "epoch": 0.24385040821701343,
      "grad_norm": 3.399627447128296,
      "learning_rate": 3.781669739267843e-05,
      "loss": 1.6381,
      "step": 9259
    },
    {
      "epoch": 0.24387674479852514,
      "grad_norm": 3.2974464893341064,
      "learning_rate": 3.781538056360285e-05,
      "loss": 1.0649,
      "step": 9260
    },
    {
      "epoch": 0.24390308138003688,
      "grad_norm": 2.1594533920288086,
      "learning_rate": 3.781406373452726e-05,
      "loss": 0.9662,
      "step": 9261
    },
    {
      "epoch": 0.2439294179615486,
      "grad_norm": 1.795055627822876,
      "learning_rate": 3.7812746905451675e-05,
      "loss": 1.9675,
      "step": 9262
    },
    {
      "epoch": 0.24395575454306032,
      "grad_norm": 1.754931092262268,
      "learning_rate": 3.7811430076376084e-05,
      "loss": 2.0979,
      "step": 9263
    },
    {
      "epoch": 0.24398209112457203,
      "grad_norm": 2.881612777709961,
      "learning_rate": 3.7810113247300506e-05,
      "loss": 1.9986,
      "step": 9264
    },
    {
      "epoch": 0.24400842770608375,
      "grad_norm": 2.1526458263397217,
      "learning_rate": 3.7808796418224915e-05,
      "loss": 2.1643,
      "step": 9265
    },
    {
      "epoch": 0.24403476428759546,
      "grad_norm": 2.0660085678100586,
      "learning_rate": 3.780747958914933e-05,
      "loss": 1.6235,
      "step": 9266
    },
    {
      "epoch": 0.24406110086910718,
      "grad_norm": 2.211167335510254,
      "learning_rate": 3.7806162760073746e-05,
      "loss": 2.0696,
      "step": 9267
    },
    {
      "epoch": 0.24408743745061892,
      "grad_norm": 3.424025297164917,
      "learning_rate": 3.7804845930998155e-05,
      "loss": 1.6922,
      "step": 9268
    },
    {
      "epoch": 0.24411377403213064,
      "grad_norm": 2.800313949584961,
      "learning_rate": 3.780352910192258e-05,
      "loss": 1.008,
      "step": 9269
    },
    {
      "epoch": 0.24414011061364235,
      "grad_norm": 2.6103975772857666,
      "learning_rate": 3.7802212272846986e-05,
      "loss": 2.2026,
      "step": 9270
    },
    {
      "epoch": 0.24416644719515407,
      "grad_norm": 1.4573440551757812,
      "learning_rate": 3.78008954437714e-05,
      "loss": 1.6872,
      "step": 9271
    },
    {
      "epoch": 0.24419278377666578,
      "grad_norm": 2.1108124256134033,
      "learning_rate": 3.779957861469581e-05,
      "loss": 1.2656,
      "step": 9272
    },
    {
      "epoch": 0.2442191203581775,
      "grad_norm": 1.9290087223052979,
      "learning_rate": 3.7798261785620226e-05,
      "loss": 0.4959,
      "step": 9273
    },
    {
      "epoch": 0.24424545693968922,
      "grad_norm": 1.559777021408081,
      "learning_rate": 3.779694495654464e-05,
      "loss": 1.5429,
      "step": 9274
    },
    {
      "epoch": 0.24427179352120096,
      "grad_norm": 2.850970983505249,
      "learning_rate": 3.779562812746906e-05,
      "loss": 1.7653,
      "step": 9275
    },
    {
      "epoch": 0.24429813010271267,
      "grad_norm": 1.8330388069152832,
      "learning_rate": 3.779431129839347e-05,
      "loss": 1.538,
      "step": 9276
    },
    {
      "epoch": 0.2443244666842244,
      "grad_norm": 2.406566619873047,
      "learning_rate": 3.779299446931788e-05,
      "loss": 1.8792,
      "step": 9277
    },
    {
      "epoch": 0.2443508032657361,
      "grad_norm": 1.6691175699234009,
      "learning_rate": 3.77916776402423e-05,
      "loss": 1.5142,
      "step": 9278
    },
    {
      "epoch": 0.24437713984724782,
      "grad_norm": 1.8278650045394897,
      "learning_rate": 3.779036081116671e-05,
      "loss": 1.9484,
      "step": 9279
    },
    {
      "epoch": 0.24440347642875954,
      "grad_norm": 1.9198180437088013,
      "learning_rate": 3.778904398209113e-05,
      "loss": 1.4031,
      "step": 9280
    },
    {
      "epoch": 0.24442981301027128,
      "grad_norm": 2.0104076862335205,
      "learning_rate": 3.778772715301554e-05,
      "loss": 1.44,
      "step": 9281
    },
    {
      "epoch": 0.244456149591783,
      "grad_norm": 1.4412245750427246,
      "learning_rate": 3.778641032393995e-05,
      "loss": 1.9661,
      "step": 9282
    },
    {
      "epoch": 0.2444824861732947,
      "grad_norm": 1.3967386484146118,
      "learning_rate": 3.778509349486437e-05,
      "loss": 1.5204,
      "step": 9283
    },
    {
      "epoch": 0.24450882275480643,
      "grad_norm": 3.0562007427215576,
      "learning_rate": 3.7783776665788784e-05,
      "loss": 0.5748,
      "step": 9284
    },
    {
      "epoch": 0.24453515933631814,
      "grad_norm": 2.1729013919830322,
      "learning_rate": 3.778245983671319e-05,
      "loss": 2.5757,
      "step": 9285
    },
    {
      "epoch": 0.24456149591782986,
      "grad_norm": 1.6921613216400146,
      "learning_rate": 3.778114300763761e-05,
      "loss": 1.3119,
      "step": 9286
    },
    {
      "epoch": 0.24458783249934157,
      "grad_norm": 1.8305718898773193,
      "learning_rate": 3.7779826178562024e-05,
      "loss": 1.7485,
      "step": 9287
    },
    {
      "epoch": 0.24461416908085332,
      "grad_norm": 3.1347134113311768,
      "learning_rate": 3.777850934948644e-05,
      "loss": 1.2699,
      "step": 9288
    },
    {
      "epoch": 0.24464050566236503,
      "grad_norm": 3.9561259746551514,
      "learning_rate": 3.7777192520410856e-05,
      "loss": 1.2056,
      "step": 9289
    },
    {
      "epoch": 0.24466684224387675,
      "grad_norm": 2.354530096054077,
      "learning_rate": 3.7775875691335264e-05,
      "loss": 2.2964,
      "step": 9290
    },
    {
      "epoch": 0.24469317882538846,
      "grad_norm": 3.5120456218719482,
      "learning_rate": 3.777455886225968e-05,
      "loss": 0.8609,
      "step": 9291
    },
    {
      "epoch": 0.24471951540690018,
      "grad_norm": 1.886204719543457,
      "learning_rate": 3.777324203318409e-05,
      "loss": 1.966,
      "step": 9292
    },
    {
      "epoch": 0.2447458519884119,
      "grad_norm": 2.966050148010254,
      "learning_rate": 3.777192520410851e-05,
      "loss": 1.5666,
      "step": 9293
    },
    {
      "epoch": 0.24477218856992364,
      "grad_norm": 1.7321619987487793,
      "learning_rate": 3.777060837503292e-05,
      "loss": 2.0653,
      "step": 9294
    },
    {
      "epoch": 0.24479852515143535,
      "grad_norm": 1.6937493085861206,
      "learning_rate": 3.7769291545957336e-05,
      "loss": 1.6134,
      "step": 9295
    },
    {
      "epoch": 0.24482486173294707,
      "grad_norm": 2.7914743423461914,
      "learning_rate": 3.776797471688175e-05,
      "loss": 0.3406,
      "step": 9296
    },
    {
      "epoch": 0.24485119831445878,
      "grad_norm": 2.6603357791900635,
      "learning_rate": 3.776665788780617e-05,
      "loss": 0.7411,
      "step": 9297
    },
    {
      "epoch": 0.2448775348959705,
      "grad_norm": 1.9323912858963013,
      "learning_rate": 3.776534105873058e-05,
      "loss": 2.189,
      "step": 9298
    },
    {
      "epoch": 0.24490387147748222,
      "grad_norm": 2.30798602104187,
      "learning_rate": 3.776402422965499e-05,
      "loss": 1.8301,
      "step": 9299
    },
    {
      "epoch": 0.24493020805899393,
      "grad_norm": 3.0702922344207764,
      "learning_rate": 3.776270740057941e-05,
      "loss": 0.7602,
      "step": 9300
    },
    {
      "epoch": 0.24495654464050567,
      "grad_norm": 2.0642573833465576,
      "learning_rate": 3.7761390571503816e-05,
      "loss": 2.2123,
      "step": 9301
    },
    {
      "epoch": 0.2449828812220174,
      "grad_norm": 1.8380906581878662,
      "learning_rate": 3.776007374242824e-05,
      "loss": 1.4186,
      "step": 9302
    },
    {
      "epoch": 0.2450092178035291,
      "grad_norm": 1.7286680936813354,
      "learning_rate": 3.775875691335265e-05,
      "loss": 1.8775,
      "step": 9303
    },
    {
      "epoch": 0.24503555438504082,
      "grad_norm": 3.878495454788208,
      "learning_rate": 3.775744008427706e-05,
      "loss": 2.6115,
      "step": 9304
    },
    {
      "epoch": 0.24506189096655254,
      "grad_norm": 3.5993738174438477,
      "learning_rate": 3.775612325520148e-05,
      "loss": 0.9903,
      "step": 9305
    },
    {
      "epoch": 0.24508822754806425,
      "grad_norm": 3.8528695106506348,
      "learning_rate": 3.775480642612589e-05,
      "loss": 1.0769,
      "step": 9306
    },
    {
      "epoch": 0.24511456412957597,
      "grad_norm": 1.7006994485855103,
      "learning_rate": 3.775348959705031e-05,
      "loss": 1.5106,
      "step": 9307
    },
    {
      "epoch": 0.2451409007110877,
      "grad_norm": 2.118520736694336,
      "learning_rate": 3.775217276797472e-05,
      "loss": 1.5786,
      "step": 9308
    },
    {
      "epoch": 0.24516723729259943,
      "grad_norm": 1.5905060768127441,
      "learning_rate": 3.7750855938899134e-05,
      "loss": 2.0823,
      "step": 9309
    },
    {
      "epoch": 0.24519357387411114,
      "grad_norm": 2.3758137226104736,
      "learning_rate": 3.774953910982354e-05,
      "loss": 1.7364,
      "step": 9310
    },
    {
      "epoch": 0.24521991045562286,
      "grad_norm": 1.9911389350891113,
      "learning_rate": 3.774822228074796e-05,
      "loss": 1.4854,
      "step": 9311
    },
    {
      "epoch": 0.24524624703713457,
      "grad_norm": 3.24302339553833,
      "learning_rate": 3.7746905451672374e-05,
      "loss": 2.0166,
      "step": 9312
    },
    {
      "epoch": 0.2452725836186463,
      "grad_norm": 1.7609572410583496,
      "learning_rate": 3.774558862259679e-05,
      "loss": 2.0834,
      "step": 9313
    },
    {
      "epoch": 0.24529892020015803,
      "grad_norm": 2.0814785957336426,
      "learning_rate": 3.7744271793521205e-05,
      "loss": 2.4209,
      "step": 9314
    },
    {
      "epoch": 0.24532525678166975,
      "grad_norm": 1.5561857223510742,
      "learning_rate": 3.7742954964445614e-05,
      "loss": 1.3433,
      "step": 9315
    },
    {
      "epoch": 0.24535159336318146,
      "grad_norm": 1.8540364503860474,
      "learning_rate": 3.7741638135370036e-05,
      "loss": 1.7081,
      "step": 9316
    },
    {
      "epoch": 0.24537792994469318,
      "grad_norm": 1.9510375261306763,
      "learning_rate": 3.7740321306294445e-05,
      "loss": 1.9614,
      "step": 9317
    },
    {
      "epoch": 0.2454042665262049,
      "grad_norm": 3.4084575176239014,
      "learning_rate": 3.773900447721886e-05,
      "loss": 1.8789,
      "step": 9318
    },
    {
      "epoch": 0.2454306031077166,
      "grad_norm": 2.46370792388916,
      "learning_rate": 3.773768764814327e-05,
      "loss": 1.4233,
      "step": 9319
    },
    {
      "epoch": 0.24545693968922833,
      "grad_norm": 1.9882310628890991,
      "learning_rate": 3.7736370819067685e-05,
      "loss": 1.8152,
      "step": 9320
    },
    {
      "epoch": 0.24548327627074007,
      "grad_norm": 2.6884074211120605,
      "learning_rate": 3.77350539899921e-05,
      "loss": 0.9255,
      "step": 9321
    },
    {
      "epoch": 0.24550961285225179,
      "grad_norm": 1.8098918199539185,
      "learning_rate": 3.7733737160916516e-05,
      "loss": 2.2152,
      "step": 9322
    },
    {
      "epoch": 0.2455359494337635,
      "grad_norm": 2.385812997817993,
      "learning_rate": 3.773242033184093e-05,
      "loss": 2.6968,
      "step": 9323
    },
    {
      "epoch": 0.24556228601527522,
      "grad_norm": 1.5554205179214478,
      "learning_rate": 3.773110350276534e-05,
      "loss": 1.531,
      "step": 9324
    },
    {
      "epoch": 0.24558862259678693,
      "grad_norm": 3.7166237831115723,
      "learning_rate": 3.7729786673689756e-05,
      "loss": 1.86,
      "step": 9325
    },
    {
      "epoch": 0.24561495917829865,
      "grad_norm": 1.9936190843582153,
      "learning_rate": 3.772846984461417e-05,
      "loss": 1.6313,
      "step": 9326
    },
    {
      "epoch": 0.24564129575981036,
      "grad_norm": 2.7776761054992676,
      "learning_rate": 3.772715301553859e-05,
      "loss": 1.0065,
      "step": 9327
    },
    {
      "epoch": 0.2456676323413221,
      "grad_norm": 1.987105369567871,
      "learning_rate": 3.7725836186462996e-05,
      "loss": 1.8912,
      "step": 9328
    },
    {
      "epoch": 0.24569396892283382,
      "grad_norm": 2.3901798725128174,
      "learning_rate": 3.772451935738741e-05,
      "loss": 1.5387,
      "step": 9329
    },
    {
      "epoch": 0.24572030550434554,
      "grad_norm": 2.1114540100097656,
      "learning_rate": 3.772320252831183e-05,
      "loss": 0.4655,
      "step": 9330
    },
    {
      "epoch": 0.24574664208585725,
      "grad_norm": 2.5467770099639893,
      "learning_rate": 3.772188569923624e-05,
      "loss": 0.6255,
      "step": 9331
    },
    {
      "epoch": 0.24577297866736897,
      "grad_norm": 3.1845436096191406,
      "learning_rate": 3.772056887016066e-05,
      "loss": 1.7654,
      "step": 9332
    },
    {
      "epoch": 0.24579931524888068,
      "grad_norm": 3.4342520236968994,
      "learning_rate": 3.771925204108507e-05,
      "loss": 1.8115,
      "step": 9333
    },
    {
      "epoch": 0.24582565183039243,
      "grad_norm": 2.281667470932007,
      "learning_rate": 3.771793521200948e-05,
      "loss": 2.0688,
      "step": 9334
    },
    {
      "epoch": 0.24585198841190414,
      "grad_norm": 2.334261894226074,
      "learning_rate": 3.77166183829339e-05,
      "loss": 1.7786,
      "step": 9335
    },
    {
      "epoch": 0.24587832499341586,
      "grad_norm": 1.9963724613189697,
      "learning_rate": 3.7715301553858314e-05,
      "loss": 1.7811,
      "step": 9336
    },
    {
      "epoch": 0.24590466157492757,
      "grad_norm": 4.233668327331543,
      "learning_rate": 3.771398472478272e-05,
      "loss": 0.9255,
      "step": 9337
    },
    {
      "epoch": 0.2459309981564393,
      "grad_norm": 3.008275270462036,
      "learning_rate": 3.771266789570714e-05,
      "loss": 1.4952,
      "step": 9338
    },
    {
      "epoch": 0.245957334737951,
      "grad_norm": 3.359005928039551,
      "learning_rate": 3.771135106663155e-05,
      "loss": 1.4314,
      "step": 9339
    },
    {
      "epoch": 0.24598367131946272,
      "grad_norm": 2.820333242416382,
      "learning_rate": 3.771003423755597e-05,
      "loss": 1.3702,
      "step": 9340
    },
    {
      "epoch": 0.24601000790097446,
      "grad_norm": 1.56570303440094,
      "learning_rate": 3.770871740848038e-05,
      "loss": 2.0693,
      "step": 9341
    },
    {
      "epoch": 0.24603634448248618,
      "grad_norm": 2.0925402641296387,
      "learning_rate": 3.7707400579404794e-05,
      "loss": 1.7175,
      "step": 9342
    },
    {
      "epoch": 0.2460626810639979,
      "grad_norm": 2.139431953430176,
      "learning_rate": 3.770608375032921e-05,
      "loss": 2.2563,
      "step": 9343
    },
    {
      "epoch": 0.2460890176455096,
      "grad_norm": 2.698939323425293,
      "learning_rate": 3.770476692125362e-05,
      "loss": 1.1589,
      "step": 9344
    },
    {
      "epoch": 0.24611535422702133,
      "grad_norm": 2.2857675552368164,
      "learning_rate": 3.770345009217804e-05,
      "loss": 2.3414,
      "step": 9345
    },
    {
      "epoch": 0.24614169080853304,
      "grad_norm": 2.204475164413452,
      "learning_rate": 3.770213326310245e-05,
      "loss": 1.7499,
      "step": 9346
    },
    {
      "epoch": 0.24616802739004479,
      "grad_norm": 1.9875195026397705,
      "learning_rate": 3.7700816434026865e-05,
      "loss": 2.0305,
      "step": 9347
    },
    {
      "epoch": 0.2461943639715565,
      "grad_norm": 1.8272719383239746,
      "learning_rate": 3.7699499604951274e-05,
      "loss": 2.3956,
      "step": 9348
    },
    {
      "epoch": 0.24622070055306822,
      "grad_norm": 2.7291312217712402,
      "learning_rate": 3.7698182775875697e-05,
      "loss": 0.7569,
      "step": 9349
    },
    {
      "epoch": 0.24624703713457993,
      "grad_norm": 1.694484829902649,
      "learning_rate": 3.7696865946800105e-05,
      "loss": 1.3337,
      "step": 9350
    },
    {
      "epoch": 0.24627337371609165,
      "grad_norm": 1.9520087242126465,
      "learning_rate": 3.769554911772452e-05,
      "loss": 1.8412,
      "step": 9351
    },
    {
      "epoch": 0.24629971029760336,
      "grad_norm": 1.5728037357330322,
      "learning_rate": 3.7694232288648937e-05,
      "loss": 1.6443,
      "step": 9352
    },
    {
      "epoch": 0.24632604687911508,
      "grad_norm": 2.5613174438476562,
      "learning_rate": 3.7692915459573345e-05,
      "loss": 2.3595,
      "step": 9353
    },
    {
      "epoch": 0.24635238346062682,
      "grad_norm": 2.300185203552246,
      "learning_rate": 3.769159863049777e-05,
      "loss": 1.8731,
      "step": 9354
    },
    {
      "epoch": 0.24637872004213854,
      "grad_norm": 1.8807735443115234,
      "learning_rate": 3.769028180142218e-05,
      "loss": 1.4396,
      "step": 9355
    },
    {
      "epoch": 0.24640505662365025,
      "grad_norm": 1.9244358539581299,
      "learning_rate": 3.768896497234659e-05,
      "loss": 1.6437,
      "step": 9356
    },
    {
      "epoch": 0.24643139320516197,
      "grad_norm": 1.8906482458114624,
      "learning_rate": 3.7687648143271e-05,
      "loss": 1.3937,
      "step": 9357
    },
    {
      "epoch": 0.24645772978667368,
      "grad_norm": 1.834443211555481,
      "learning_rate": 3.768633131419542e-05,
      "loss": 2.2964,
      "step": 9358
    },
    {
      "epoch": 0.2464840663681854,
      "grad_norm": 1.5918519496917725,
      "learning_rate": 3.768501448511983e-05,
      "loss": 1.6159,
      "step": 9359
    },
    {
      "epoch": 0.24651040294969712,
      "grad_norm": 3.436182737350464,
      "learning_rate": 3.768369765604425e-05,
      "loss": 1.4593,
      "step": 9360
    },
    {
      "epoch": 0.24653673953120886,
      "grad_norm": 2.331357479095459,
      "learning_rate": 3.7682380826968663e-05,
      "loss": 1.7932,
      "step": 9361
    },
    {
      "epoch": 0.24656307611272058,
      "grad_norm": 3.6024861335754395,
      "learning_rate": 3.768106399789307e-05,
      "loss": 1.0316,
      "step": 9362
    },
    {
      "epoch": 0.2465894126942323,
      "grad_norm": 3.5238559246063232,
      "learning_rate": 3.7679747168817495e-05,
      "loss": 2.0608,
      "step": 9363
    },
    {
      "epoch": 0.246615749275744,
      "grad_norm": 1.7826991081237793,
      "learning_rate": 3.7678430339741903e-05,
      "loss": 1.5082,
      "step": 9364
    },
    {
      "epoch": 0.24664208585725572,
      "grad_norm": 1.9705603122711182,
      "learning_rate": 3.767711351066632e-05,
      "loss": 2.4939,
      "step": 9365
    },
    {
      "epoch": 0.24666842243876744,
      "grad_norm": 2.5866169929504395,
      "learning_rate": 3.767579668159073e-05,
      "loss": 0.7636,
      "step": 9366
    },
    {
      "epoch": 0.24669475902027918,
      "grad_norm": 1.7340705394744873,
      "learning_rate": 3.7674479852515143e-05,
      "loss": 2.1591,
      "step": 9367
    },
    {
      "epoch": 0.2467210956017909,
      "grad_norm": 4.0733232498168945,
      "learning_rate": 3.767316302343956e-05,
      "loss": 0.7121,
      "step": 9368
    },
    {
      "epoch": 0.2467474321833026,
      "grad_norm": 2.3712146282196045,
      "learning_rate": 3.7671846194363975e-05,
      "loss": 1.928,
      "step": 9369
    },
    {
      "epoch": 0.24677376876481433,
      "grad_norm": 3.0647618770599365,
      "learning_rate": 3.767052936528839e-05,
      "loss": 1.9315,
      "step": 9370
    },
    {
      "epoch": 0.24680010534632604,
      "grad_norm": 1.8452814817428589,
      "learning_rate": 3.76692125362128e-05,
      "loss": 1.6652,
      "step": 9371
    },
    {
      "epoch": 0.24682644192783776,
      "grad_norm": 2.4918529987335205,
      "learning_rate": 3.7667895707137215e-05,
      "loss": 2.1092,
      "step": 9372
    },
    {
      "epoch": 0.24685277850934947,
      "grad_norm": 2.1917600631713867,
      "learning_rate": 3.766657887806163e-05,
      "loss": 1.9167,
      "step": 9373
    },
    {
      "epoch": 0.24687911509086122,
      "grad_norm": 2.591287136077881,
      "learning_rate": 3.7665262048986046e-05,
      "loss": 1.2811,
      "step": 9374
    },
    {
      "epoch": 0.24690545167237293,
      "grad_norm": 1.6492140293121338,
      "learning_rate": 3.7663945219910455e-05,
      "loss": 1.9454,
      "step": 9375
    },
    {
      "epoch": 0.24693178825388465,
      "grad_norm": 2.2528035640716553,
      "learning_rate": 3.766262839083487e-05,
      "loss": 1.9057,
      "step": 9376
    },
    {
      "epoch": 0.24695812483539636,
      "grad_norm": 2.1465160846710205,
      "learning_rate": 3.7661311561759286e-05,
      "loss": 1.4572,
      "step": 9377
    },
    {
      "epoch": 0.24698446141690808,
      "grad_norm": 1.577000617980957,
      "learning_rate": 3.76599947326837e-05,
      "loss": 1.4408,
      "step": 9378
    },
    {
      "epoch": 0.2470107979984198,
      "grad_norm": 2.171461343765259,
      "learning_rate": 3.765867790360812e-05,
      "loss": 2.0906,
      "step": 9379
    },
    {
      "epoch": 0.2470371345799315,
      "grad_norm": 2.246291399002075,
      "learning_rate": 3.7657361074532526e-05,
      "loss": 1.9148,
      "step": 9380
    },
    {
      "epoch": 0.24706347116144325,
      "grad_norm": 4.294859886169434,
      "learning_rate": 3.765604424545694e-05,
      "loss": 2.1422,
      "step": 9381
    },
    {
      "epoch": 0.24708980774295497,
      "grad_norm": 2.15989089012146,
      "learning_rate": 3.765472741638136e-05,
      "loss": 1.3629,
      "step": 9382
    },
    {
      "epoch": 0.24711614432446669,
      "grad_norm": 2.1761250495910645,
      "learning_rate": 3.765341058730577e-05,
      "loss": 1.8728,
      "step": 9383
    },
    {
      "epoch": 0.2471424809059784,
      "grad_norm": 3.8124423027038574,
      "learning_rate": 3.765209375823018e-05,
      "loss": 1.8839,
      "step": 9384
    },
    {
      "epoch": 0.24716881748749012,
      "grad_norm": 1.5608952045440674,
      "learning_rate": 3.76507769291546e-05,
      "loss": 1.9388,
      "step": 9385
    },
    {
      "epoch": 0.24719515406900183,
      "grad_norm": 2.71919584274292,
      "learning_rate": 3.7649460100079006e-05,
      "loss": 1.3599,
      "step": 9386
    },
    {
      "epoch": 0.24722149065051358,
      "grad_norm": 4.256839752197266,
      "learning_rate": 3.764814327100343e-05,
      "loss": 2.4145,
      "step": 9387
    },
    {
      "epoch": 0.2472478272320253,
      "grad_norm": 1.5942931175231934,
      "learning_rate": 3.764682644192784e-05,
      "loss": 1.742,
      "step": 9388
    },
    {
      "epoch": 0.247274163813537,
      "grad_norm": 2.4733927249908447,
      "learning_rate": 3.764550961285225e-05,
      "loss": 2.5946,
      "step": 9389
    },
    {
      "epoch": 0.24730050039504872,
      "grad_norm": 4.844635486602783,
      "learning_rate": 3.764419278377667e-05,
      "loss": 2.207,
      "step": 9390
    },
    {
      "epoch": 0.24732683697656044,
      "grad_norm": 1.9164692163467407,
      "learning_rate": 3.764287595470108e-05,
      "loss": 0.1823,
      "step": 9391
    },
    {
      "epoch": 0.24735317355807215,
      "grad_norm": 2.549802541732788,
      "learning_rate": 3.76415591256255e-05,
      "loss": 1.7088,
      "step": 9392
    },
    {
      "epoch": 0.24737951013958387,
      "grad_norm": 3.6922531127929688,
      "learning_rate": 3.764024229654991e-05,
      "loss": 1.4142,
      "step": 9393
    },
    {
      "epoch": 0.2474058467210956,
      "grad_norm": 2.3875904083251953,
      "learning_rate": 3.7638925467474324e-05,
      "loss": 2.0409,
      "step": 9394
    },
    {
      "epoch": 0.24743218330260733,
      "grad_norm": 3.4685134887695312,
      "learning_rate": 3.763760863839873e-05,
      "loss": 0.9303,
      "step": 9395
    },
    {
      "epoch": 0.24745851988411904,
      "grad_norm": 1.6364049911499023,
      "learning_rate": 3.7636291809323155e-05,
      "loss": 1.8949,
      "step": 9396
    },
    {
      "epoch": 0.24748485646563076,
      "grad_norm": 1.799635648727417,
      "learning_rate": 3.7634974980247564e-05,
      "loss": 1.9883,
      "step": 9397
    },
    {
      "epoch": 0.24751119304714247,
      "grad_norm": 2.8206584453582764,
      "learning_rate": 3.763365815117198e-05,
      "loss": 0.9572,
      "step": 9398
    },
    {
      "epoch": 0.2475375296286542,
      "grad_norm": 1.947456955909729,
      "learning_rate": 3.7632341322096395e-05,
      "loss": 1.4944,
      "step": 9399
    },
    {
      "epoch": 0.24756386621016593,
      "grad_norm": 1.634972095489502,
      "learning_rate": 3.7631024493020804e-05,
      "loss": 1.7038,
      "step": 9400
    },
    {
      "epoch": 0.24759020279167765,
      "grad_norm": 2.0548524856567383,
      "learning_rate": 3.7629707663945226e-05,
      "loss": 1.1977,
      "step": 9401
    },
    {
      "epoch": 0.24761653937318936,
      "grad_norm": 3.009627103805542,
      "learning_rate": 3.7628390834869635e-05,
      "loss": 1.3542,
      "step": 9402
    },
    {
      "epoch": 0.24764287595470108,
      "grad_norm": 2.9424219131469727,
      "learning_rate": 3.762707400579405e-05,
      "loss": 2.4282,
      "step": 9403
    },
    {
      "epoch": 0.2476692125362128,
      "grad_norm": 2.1092593669891357,
      "learning_rate": 3.762575717671846e-05,
      "loss": 1.0205,
      "step": 9404
    },
    {
      "epoch": 0.2476955491177245,
      "grad_norm": 2.342179536819458,
      "learning_rate": 3.7624440347642875e-05,
      "loss": 2.0314,
      "step": 9405
    },
    {
      "epoch": 0.24772188569923623,
      "grad_norm": 6.2738118171691895,
      "learning_rate": 3.762312351856729e-05,
      "loss": 1.4511,
      "step": 9406
    },
    {
      "epoch": 0.24774822228074797,
      "grad_norm": 3.217378854751587,
      "learning_rate": 3.7621806689491706e-05,
      "loss": 2.4851,
      "step": 9407
    },
    {
      "epoch": 0.2477745588622597,
      "grad_norm": 2.0456595420837402,
      "learning_rate": 3.762048986041612e-05,
      "loss": 1.6556,
      "step": 9408
    },
    {
      "epoch": 0.2478008954437714,
      "grad_norm": 2.476691961288452,
      "learning_rate": 3.761917303134053e-05,
      "loss": 1.9022,
      "step": 9409
    },
    {
      "epoch": 0.24782723202528312,
      "grad_norm": 1.6512302160263062,
      "learning_rate": 3.761785620226495e-05,
      "loss": 1.5487,
      "step": 9410
    },
    {
      "epoch": 0.24785356860679483,
      "grad_norm": 4.01853609085083,
      "learning_rate": 3.761653937318936e-05,
      "loss": 1.6983,
      "step": 9411
    },
    {
      "epoch": 0.24787990518830655,
      "grad_norm": 2.401923656463623,
      "learning_rate": 3.761522254411378e-05,
      "loss": 1.3442,
      "step": 9412
    },
    {
      "epoch": 0.24790624176981826,
      "grad_norm": 1.6326779127120972,
      "learning_rate": 3.7613905715038186e-05,
      "loss": 2.4835,
      "step": 9413
    },
    {
      "epoch": 0.24793257835133,
      "grad_norm": 1.719183325767517,
      "learning_rate": 3.76125888859626e-05,
      "loss": 2.3692,
      "step": 9414
    },
    {
      "epoch": 0.24795891493284172,
      "grad_norm": 3.0368611812591553,
      "learning_rate": 3.761127205688702e-05,
      "loss": 1.5506,
      "step": 9415
    },
    {
      "epoch": 0.24798525151435344,
      "grad_norm": 1.6918164491653442,
      "learning_rate": 3.760995522781143e-05,
      "loss": 1.6433,
      "step": 9416
    },
    {
      "epoch": 0.24801158809586515,
      "grad_norm": 3.1472530364990234,
      "learning_rate": 3.760863839873585e-05,
      "loss": 1.3656,
      "step": 9417
    },
    {
      "epoch": 0.24803792467737687,
      "grad_norm": 2.1720311641693115,
      "learning_rate": 3.760732156966026e-05,
      "loss": 1.9507,
      "step": 9418
    },
    {
      "epoch": 0.24806426125888859,
      "grad_norm": 2.212719440460205,
      "learning_rate": 3.760600474058467e-05,
      "loss": 2.0337,
      "step": 9419
    },
    {
      "epoch": 0.24809059784040033,
      "grad_norm": 3.5032010078430176,
      "learning_rate": 3.760468791150909e-05,
      "loss": 1.3772,
      "step": 9420
    },
    {
      "epoch": 0.24811693442191204,
      "grad_norm": 2.362844467163086,
      "learning_rate": 3.7603371082433504e-05,
      "loss": 1.7551,
      "step": 9421
    },
    {
      "epoch": 0.24814327100342376,
      "grad_norm": 3.3651692867279053,
      "learning_rate": 3.760205425335791e-05,
      "loss": 1.0103,
      "step": 9422
    },
    {
      "epoch": 0.24816960758493548,
      "grad_norm": 4.137308597564697,
      "learning_rate": 3.760073742428233e-05,
      "loss": 1.4831,
      "step": 9423
    },
    {
      "epoch": 0.2481959441664472,
      "grad_norm": 2.5527591705322266,
      "learning_rate": 3.7599420595206744e-05,
      "loss": 1.2846,
      "step": 9424
    },
    {
      "epoch": 0.2482222807479589,
      "grad_norm": 1.8308279514312744,
      "learning_rate": 3.759810376613116e-05,
      "loss": 2.2615,
      "step": 9425
    },
    {
      "epoch": 0.24824861732947062,
      "grad_norm": 3.7004990577697754,
      "learning_rate": 3.7596786937055576e-05,
      "loss": 1.0325,
      "step": 9426
    },
    {
      "epoch": 0.24827495391098237,
      "grad_norm": 2.8788669109344482,
      "learning_rate": 3.7595470107979984e-05,
      "loss": 0.7264,
      "step": 9427
    },
    {
      "epoch": 0.24830129049249408,
      "grad_norm": 1.7954556941986084,
      "learning_rate": 3.75941532789044e-05,
      "loss": 0.7045,
      "step": 9428
    },
    {
      "epoch": 0.2483276270740058,
      "grad_norm": 2.058328628540039,
      "learning_rate": 3.7592836449828816e-05,
      "loss": 1.9194,
      "step": 9429
    },
    {
      "epoch": 0.2483539636555175,
      "grad_norm": 2.114301919937134,
      "learning_rate": 3.759151962075323e-05,
      "loss": 1.9982,
      "step": 9430
    },
    {
      "epoch": 0.24838030023702923,
      "grad_norm": 3.6203978061676025,
      "learning_rate": 3.759020279167764e-05,
      "loss": 0.8585,
      "step": 9431
    },
    {
      "epoch": 0.24840663681854094,
      "grad_norm": 2.9751482009887695,
      "learning_rate": 3.7588885962602056e-05,
      "loss": 1.8626,
      "step": 9432
    },
    {
      "epoch": 0.2484329734000527,
      "grad_norm": 1.9818267822265625,
      "learning_rate": 3.758756913352647e-05,
      "loss": 2.6741,
      "step": 9433
    },
    {
      "epoch": 0.2484593099815644,
      "grad_norm": 1.5844560861587524,
      "learning_rate": 3.758625230445089e-05,
      "loss": 1.3837,
      "step": 9434
    },
    {
      "epoch": 0.24848564656307612,
      "grad_norm": 1.6429064273834229,
      "learning_rate": 3.75849354753753e-05,
      "loss": 1.4993,
      "step": 9435
    },
    {
      "epoch": 0.24851198314458783,
      "grad_norm": 1.9596023559570312,
      "learning_rate": 3.758361864629971e-05,
      "loss": 1.4399,
      "step": 9436
    },
    {
      "epoch": 0.24853831972609955,
      "grad_norm": 3.07328462600708,
      "learning_rate": 3.758230181722413e-05,
      "loss": 1.4072,
      "step": 9437
    },
    {
      "epoch": 0.24856465630761126,
      "grad_norm": 2.0965077877044678,
      "learning_rate": 3.7580984988148536e-05,
      "loss": 1.9421,
      "step": 9438
    },
    {
      "epoch": 0.24859099288912298,
      "grad_norm": 3.9975225925445557,
      "learning_rate": 3.757966815907296e-05,
      "loss": 2.1747,
      "step": 9439
    },
    {
      "epoch": 0.24861732947063472,
      "grad_norm": 2.217804193496704,
      "learning_rate": 3.757835132999737e-05,
      "loss": 1.4465,
      "step": 9440
    },
    {
      "epoch": 0.24864366605214644,
      "grad_norm": 2.4609575271606445,
      "learning_rate": 3.757703450092178e-05,
      "loss": 0.7346,
      "step": 9441
    },
    {
      "epoch": 0.24867000263365815,
      "grad_norm": 3.542363166809082,
      "learning_rate": 3.757571767184619e-05,
      "loss": 1.7552,
      "step": 9442
    },
    {
      "epoch": 0.24869633921516987,
      "grad_norm": 1.7418606281280518,
      "learning_rate": 3.7574400842770614e-05,
      "loss": 1.722,
      "step": 9443
    },
    {
      "epoch": 0.24872267579668159,
      "grad_norm": 2.213932752609253,
      "learning_rate": 3.757308401369502e-05,
      "loss": 1.5757,
      "step": 9444
    },
    {
      "epoch": 0.2487490123781933,
      "grad_norm": 3.2281250953674316,
      "learning_rate": 3.757176718461944e-05,
      "loss": 1.6851,
      "step": 9445
    },
    {
      "epoch": 0.24877534895970502,
      "grad_norm": 4.151186943054199,
      "learning_rate": 3.7570450355543854e-05,
      "loss": 1.1654,
      "step": 9446
    },
    {
      "epoch": 0.24880168554121676,
      "grad_norm": 4.256567001342773,
      "learning_rate": 3.756913352646826e-05,
      "loss": 1.1562,
      "step": 9447
    },
    {
      "epoch": 0.24882802212272848,
      "grad_norm": 2.671936273574829,
      "learning_rate": 3.7567816697392685e-05,
      "loss": 1.7515,
      "step": 9448
    },
    {
      "epoch": 0.2488543587042402,
      "grad_norm": 2.370760202407837,
      "learning_rate": 3.7566499868317094e-05,
      "loss": 2.0849,
      "step": 9449
    },
    {
      "epoch": 0.2488806952857519,
      "grad_norm": 2.531735420227051,
      "learning_rate": 3.756518303924151e-05,
      "loss": 1.6962,
      "step": 9450
    },
    {
      "epoch": 0.24890703186726362,
      "grad_norm": 1.6980692148208618,
      "learning_rate": 3.756386621016592e-05,
      "loss": 1.6919,
      "step": 9451
    },
    {
      "epoch": 0.24893336844877534,
      "grad_norm": 1.8127738237380981,
      "learning_rate": 3.7562549381090334e-05,
      "loss": 1.5787,
      "step": 9452
    },
    {
      "epoch": 0.24895970503028708,
      "grad_norm": 1.7807490825653076,
      "learning_rate": 3.756123255201475e-05,
      "loss": 1.33,
      "step": 9453
    },
    {
      "epoch": 0.2489860416117988,
      "grad_norm": 2.3240952491760254,
      "learning_rate": 3.7559915722939165e-05,
      "loss": 1.7208,
      "step": 9454
    },
    {
      "epoch": 0.2490123781933105,
      "grad_norm": 1.6140258312225342,
      "learning_rate": 3.755859889386358e-05,
      "loss": 1.3656,
      "step": 9455
    },
    {
      "epoch": 0.24903871477482223,
      "grad_norm": 3.2601892948150635,
      "learning_rate": 3.755728206478799e-05,
      "loss": 1.7914,
      "step": 9456
    },
    {
      "epoch": 0.24906505135633394,
      "grad_norm": 1.826035737991333,
      "learning_rate": 3.7555965235712405e-05,
      "loss": 1.5036,
      "step": 9457
    },
    {
      "epoch": 0.24909138793784566,
      "grad_norm": 1.7450965642929077,
      "learning_rate": 3.755464840663682e-05,
      "loss": 2.0786,
      "step": 9458
    },
    {
      "epoch": 0.24911772451935738,
      "grad_norm": 3.048736572265625,
      "learning_rate": 3.7553331577561236e-05,
      "loss": 2.1121,
      "step": 9459
    },
    {
      "epoch": 0.24914406110086912,
      "grad_norm": 2.082838773727417,
      "learning_rate": 3.7552014748485645e-05,
      "loss": 1.9109,
      "step": 9460
    },
    {
      "epoch": 0.24917039768238083,
      "grad_norm": 2.270730495452881,
      "learning_rate": 3.755069791941006e-05,
      "loss": 1.7991,
      "step": 9461
    },
    {
      "epoch": 0.24919673426389255,
      "grad_norm": 3.197336196899414,
      "learning_rate": 3.7549381090334476e-05,
      "loss": 1.6597,
      "step": 9462
    },
    {
      "epoch": 0.24922307084540427,
      "grad_norm": 1.5741242170333862,
      "learning_rate": 3.754806426125889e-05,
      "loss": 0.2953,
      "step": 9463
    },
    {
      "epoch": 0.24924940742691598,
      "grad_norm": 2.0129241943359375,
      "learning_rate": 3.754674743218331e-05,
      "loss": 2.0393,
      "step": 9464
    },
    {
      "epoch": 0.2492757440084277,
      "grad_norm": 3.4816834926605225,
      "learning_rate": 3.7545430603107716e-05,
      "loss": 1.2914,
      "step": 9465
    },
    {
      "epoch": 0.2493020805899394,
      "grad_norm": 3.7552454471588135,
      "learning_rate": 3.754411377403213e-05,
      "loss": 1.5201,
      "step": 9466
    },
    {
      "epoch": 0.24932841717145116,
      "grad_norm": 2.8539235591888428,
      "learning_rate": 3.754279694495655e-05,
      "loss": 1.7869,
      "step": 9467
    },
    {
      "epoch": 0.24935475375296287,
      "grad_norm": 2.0584511756896973,
      "learning_rate": 3.754148011588096e-05,
      "loss": 1.1935,
      "step": 9468
    },
    {
      "epoch": 0.2493810903344746,
      "grad_norm": 2.09688663482666,
      "learning_rate": 3.754016328680537e-05,
      "loss": 1.6549,
      "step": 9469
    },
    {
      "epoch": 0.2494074269159863,
      "grad_norm": 2.091367244720459,
      "learning_rate": 3.753884645772979e-05,
      "loss": 1.7702,
      "step": 9470
    },
    {
      "epoch": 0.24943376349749802,
      "grad_norm": 5.105950832366943,
      "learning_rate": 3.75375296286542e-05,
      "loss": 2.0232,
      "step": 9471
    },
    {
      "epoch": 0.24946010007900973,
      "grad_norm": 2.230776309967041,
      "learning_rate": 3.753621279957862e-05,
      "loss": 1.4869,
      "step": 9472
    },
    {
      "epoch": 0.24948643666052148,
      "grad_norm": 1.6570568084716797,
      "learning_rate": 3.7534895970503034e-05,
      "loss": 1.9204,
      "step": 9473
    },
    {
      "epoch": 0.2495127732420332,
      "grad_norm": 3.0584867000579834,
      "learning_rate": 3.753357914142744e-05,
      "loss": 1.9227,
      "step": 9474
    },
    {
      "epoch": 0.2495391098235449,
      "grad_norm": 4.87863302230835,
      "learning_rate": 3.753226231235186e-05,
      "loss": 1.4147,
      "step": 9475
    },
    {
      "epoch": 0.24956544640505662,
      "grad_norm": 1.9131696224212646,
      "learning_rate": 3.7530945483276274e-05,
      "loss": 0.7294,
      "step": 9476
    },
    {
      "epoch": 0.24959178298656834,
      "grad_norm": 1.7261526584625244,
      "learning_rate": 3.752962865420069e-05,
      "loss": 2.2689,
      "step": 9477
    },
    {
      "epoch": 0.24961811956808005,
      "grad_norm": 1.9611601829528809,
      "learning_rate": 3.75283118251251e-05,
      "loss": 1.9983,
      "step": 9478
    },
    {
      "epoch": 0.24964445614959177,
      "grad_norm": 1.7864912748336792,
      "learning_rate": 3.7526994996049514e-05,
      "loss": 1.8251,
      "step": 9479
    },
    {
      "epoch": 0.2496707927311035,
      "grad_norm": 2.0357465744018555,
      "learning_rate": 3.752567816697393e-05,
      "loss": 1.3445,
      "step": 9480
    },
    {
      "epoch": 0.24969712931261523,
      "grad_norm": 2.4738476276397705,
      "learning_rate": 3.7524361337898345e-05,
      "loss": 2.1068,
      "step": 9481
    },
    {
      "epoch": 0.24972346589412694,
      "grad_norm": 2.471388339996338,
      "learning_rate": 3.752304450882276e-05,
      "loss": 1.8538,
      "step": 9482
    },
    {
      "epoch": 0.24974980247563866,
      "grad_norm": 2.5244197845458984,
      "learning_rate": 3.752172767974717e-05,
      "loss": 0.9427,
      "step": 9483
    },
    {
      "epoch": 0.24977613905715038,
      "grad_norm": 2.4803152084350586,
      "learning_rate": 3.7520410850671585e-05,
      "loss": 1.8732,
      "step": 9484
    },
    {
      "epoch": 0.2498024756386621,
      "grad_norm": 1.862506628036499,
      "learning_rate": 3.7519094021595994e-05,
      "loss": 1.8955,
      "step": 9485
    },
    {
      "epoch": 0.24982881222017383,
      "grad_norm": 3.2220823764801025,
      "learning_rate": 3.751777719252042e-05,
      "loss": 0.7518,
      "step": 9486
    },
    {
      "epoch": 0.24985514880168555,
      "grad_norm": 1.572978138923645,
      "learning_rate": 3.7516460363444825e-05,
      "loss": 1.4945,
      "step": 9487
    },
    {
      "epoch": 0.24988148538319727,
      "grad_norm": 2.6795876026153564,
      "learning_rate": 3.751514353436924e-05,
      "loss": 1.8778,
      "step": 9488
    },
    {
      "epoch": 0.24990782196470898,
      "grad_norm": 3.7850356101989746,
      "learning_rate": 3.751382670529365e-05,
      "loss": 2.5065,
      "step": 9489
    },
    {
      "epoch": 0.2499341585462207,
      "grad_norm": 1.7539210319519043,
      "learning_rate": 3.7512509876218065e-05,
      "loss": 2.0139,
      "step": 9490
    },
    {
      "epoch": 0.2499604951277324,
      "grad_norm": 6.198868751525879,
      "learning_rate": 3.751119304714248e-05,
      "loss": 1.4639,
      "step": 9491
    },
    {
      "epoch": 0.24998683170924413,
      "grad_norm": 1.6415857076644897,
      "learning_rate": 3.75098762180669e-05,
      "loss": 2.0316,
      "step": 9492
    },
    {
      "epoch": 0.25001316829075587,
      "grad_norm": 4.67836856842041,
      "learning_rate": 3.750855938899131e-05,
      "loss": 0.8169,
      "step": 9493
    },
    {
      "epoch": 0.25003950487226756,
      "grad_norm": 1.7568237781524658,
      "learning_rate": 3.750724255991572e-05,
      "loss": 0.6639,
      "step": 9494
    },
    {
      "epoch": 0.2500658414537793,
      "grad_norm": 3.358640432357788,
      "learning_rate": 3.7505925730840143e-05,
      "loss": 1.4922,
      "step": 9495
    },
    {
      "epoch": 0.25009217803529105,
      "grad_norm": 2.1151347160339355,
      "learning_rate": 3.750460890176455e-05,
      "loss": 1.4312,
      "step": 9496
    },
    {
      "epoch": 0.25011851461680273,
      "grad_norm": 3.34977126121521,
      "learning_rate": 3.750329207268897e-05,
      "loss": 1.6327,
      "step": 9497
    },
    {
      "epoch": 0.2501448511983145,
      "grad_norm": 4.54311466217041,
      "learning_rate": 3.750197524361338e-05,
      "loss": 1.1835,
      "step": 9498
    },
    {
      "epoch": 0.25017118777982617,
      "grad_norm": 1.5967661142349243,
      "learning_rate": 3.750065841453779e-05,
      "loss": 1.7695,
      "step": 9499
    },
    {
      "epoch": 0.2501975243613379,
      "grad_norm": 2.2285587787628174,
      "learning_rate": 3.749934158546221e-05,
      "loss": 1.961,
      "step": 9500
    },
    {
      "epoch": 0.2502238609428496,
      "grad_norm": 3.322103500366211,
      "learning_rate": 3.7498024756386623e-05,
      "loss": 1.1913,
      "step": 9501
    },
    {
      "epoch": 0.25025019752436134,
      "grad_norm": 2.247872829437256,
      "learning_rate": 3.749670792731104e-05,
      "loss": 1.9805,
      "step": 9502
    },
    {
      "epoch": 0.2502765341058731,
      "grad_norm": 2.0715060234069824,
      "learning_rate": 3.749539109823545e-05,
      "loss": 1.7396,
      "step": 9503
    },
    {
      "epoch": 0.25030287068738477,
      "grad_norm": 2.527590036392212,
      "learning_rate": 3.7494074269159864e-05,
      "loss": 1.3825,
      "step": 9504
    },
    {
      "epoch": 0.2503292072688965,
      "grad_norm": 2.449042797088623,
      "learning_rate": 3.749275744008428e-05,
      "loss": 1.204,
      "step": 9505
    },
    {
      "epoch": 0.2503555438504082,
      "grad_norm": 2.8567423820495605,
      "learning_rate": 3.7491440611008695e-05,
      "loss": 0.7857,
      "step": 9506
    },
    {
      "epoch": 0.25038188043191995,
      "grad_norm": 1.9735890626907349,
      "learning_rate": 3.7490123781933104e-05,
      "loss": 2.0595,
      "step": 9507
    },
    {
      "epoch": 0.25040821701343163,
      "grad_norm": 2.7401175498962402,
      "learning_rate": 3.748880695285752e-05,
      "loss": 1.4095,
      "step": 9508
    },
    {
      "epoch": 0.2504345535949434,
      "grad_norm": 1.7880487442016602,
      "learning_rate": 3.7487490123781935e-05,
      "loss": 0.6231,
      "step": 9509
    },
    {
      "epoch": 0.2504608901764551,
      "grad_norm": 2.08772873878479,
      "learning_rate": 3.748617329470635e-05,
      "loss": 1.8496,
      "step": 9510
    },
    {
      "epoch": 0.2504872267579668,
      "grad_norm": 2.7481024265289307,
      "learning_rate": 3.7484856465630766e-05,
      "loss": 1.524,
      "step": 9511
    },
    {
      "epoch": 0.25051356333947855,
      "grad_norm": 2.2460877895355225,
      "learning_rate": 3.7483539636555175e-05,
      "loss": 1.8958,
      "step": 9512
    },
    {
      "epoch": 0.25053989992099024,
      "grad_norm": 2.1388838291168213,
      "learning_rate": 3.748222280747959e-05,
      "loss": 1.9058,
      "step": 9513
    },
    {
      "epoch": 0.250566236502502,
      "grad_norm": 2.4700698852539062,
      "learning_rate": 3.7480905978404006e-05,
      "loss": 2.0323,
      "step": 9514
    },
    {
      "epoch": 0.25059257308401367,
      "grad_norm": 1.7383674383163452,
      "learning_rate": 3.747958914932842e-05,
      "loss": 1.5088,
      "step": 9515
    },
    {
      "epoch": 0.2506189096655254,
      "grad_norm": 1.8023300170898438,
      "learning_rate": 3.747827232025283e-05,
      "loss": 1.7283,
      "step": 9516
    },
    {
      "epoch": 0.25064524624703716,
      "grad_norm": 2.087075710296631,
      "learning_rate": 3.7476955491177246e-05,
      "loss": 0.9805,
      "step": 9517
    },
    {
      "epoch": 0.25067158282854884,
      "grad_norm": 6.363293170928955,
      "learning_rate": 3.747563866210166e-05,
      "loss": 2.4237,
      "step": 9518
    },
    {
      "epoch": 0.2506979194100606,
      "grad_norm": 1.7582508325576782,
      "learning_rate": 3.747432183302608e-05,
      "loss": 1.9084,
      "step": 9519
    },
    {
      "epoch": 0.2507242559915723,
      "grad_norm": 1.5304358005523682,
      "learning_rate": 3.747300500395049e-05,
      "loss": 1.7855,
      "step": 9520
    },
    {
      "epoch": 0.250750592573084,
      "grad_norm": 1.6910958290100098,
      "learning_rate": 3.74716881748749e-05,
      "loss": 1.6028,
      "step": 9521
    },
    {
      "epoch": 0.2507769291545957,
      "grad_norm": 2.206517219543457,
      "learning_rate": 3.747037134579932e-05,
      "loss": 1.2626,
      "step": 9522
    },
    {
      "epoch": 0.25080326573610745,
      "grad_norm": 2.0970842838287354,
      "learning_rate": 3.7469054516723726e-05,
      "loss": 1.7463,
      "step": 9523
    },
    {
      "epoch": 0.2508296023176192,
      "grad_norm": 3.262721300125122,
      "learning_rate": 3.746773768764815e-05,
      "loss": 1.4458,
      "step": 9524
    },
    {
      "epoch": 0.2508559388991309,
      "grad_norm": 2.794424057006836,
      "learning_rate": 3.746642085857256e-05,
      "loss": 0.8474,
      "step": 9525
    },
    {
      "epoch": 0.2508822754806426,
      "grad_norm": 1.6718446016311646,
      "learning_rate": 3.746510402949697e-05,
      "loss": 1.8155,
      "step": 9526
    },
    {
      "epoch": 0.2509086120621543,
      "grad_norm": 2.027240753173828,
      "learning_rate": 3.746378720042139e-05,
      "loss": 2.0762,
      "step": 9527
    },
    {
      "epoch": 0.25093494864366606,
      "grad_norm": 3.090517520904541,
      "learning_rate": 3.7462470371345804e-05,
      "loss": 2.488,
      "step": 9528
    },
    {
      "epoch": 0.2509612852251778,
      "grad_norm": 2.1532089710235596,
      "learning_rate": 3.746115354227022e-05,
      "loss": 2.2541,
      "step": 9529
    },
    {
      "epoch": 0.2509876218066895,
      "grad_norm": 1.409540057182312,
      "learning_rate": 3.745983671319463e-05,
      "loss": 1.1668,
      "step": 9530
    },
    {
      "epoch": 0.25101395838820123,
      "grad_norm": 1.402719259262085,
      "learning_rate": 3.7458519884119044e-05,
      "loss": 1.6927,
      "step": 9531
    },
    {
      "epoch": 0.2510402949697129,
      "grad_norm": 1.6526787281036377,
      "learning_rate": 3.745720305504345e-05,
      "loss": 2.0521,
      "step": 9532
    },
    {
      "epoch": 0.25106663155122466,
      "grad_norm": 1.7547125816345215,
      "learning_rate": 3.7455886225967875e-05,
      "loss": 1.0086,
      "step": 9533
    },
    {
      "epoch": 0.25109296813273635,
      "grad_norm": 1.8315263986587524,
      "learning_rate": 3.7454569396892284e-05,
      "loss": 1.2294,
      "step": 9534
    },
    {
      "epoch": 0.2511193047142481,
      "grad_norm": 2.541095018386841,
      "learning_rate": 3.74532525678167e-05,
      "loss": 0.7088,
      "step": 9535
    },
    {
      "epoch": 0.25114564129575984,
      "grad_norm": 3.2906458377838135,
      "learning_rate": 3.7451935738741115e-05,
      "loss": 1.787,
      "step": 9536
    },
    {
      "epoch": 0.2511719778772715,
      "grad_norm": 3.915858030319214,
      "learning_rate": 3.7450618909665524e-05,
      "loss": 1.5705,
      "step": 9537
    },
    {
      "epoch": 0.25119831445878327,
      "grad_norm": 2.6768863201141357,
      "learning_rate": 3.7449302080589946e-05,
      "loss": 1.7437,
      "step": 9538
    },
    {
      "epoch": 0.25122465104029496,
      "grad_norm": 1.751901626586914,
      "learning_rate": 3.7447985251514355e-05,
      "loss": 1.4041,
      "step": 9539
    },
    {
      "epoch": 0.2512509876218067,
      "grad_norm": 1.7146458625793457,
      "learning_rate": 3.744666842243877e-05,
      "loss": 1.5067,
      "step": 9540
    },
    {
      "epoch": 0.2512773242033184,
      "grad_norm": 2.255998373031616,
      "learning_rate": 3.744535159336318e-05,
      "loss": 2.035,
      "step": 9541
    },
    {
      "epoch": 0.25130366078483013,
      "grad_norm": 2.358640193939209,
      "learning_rate": 3.74440347642876e-05,
      "loss": 1.7264,
      "step": 9542
    },
    {
      "epoch": 0.2513299973663419,
      "grad_norm": 2.8867604732513428,
      "learning_rate": 3.744271793521201e-05,
      "loss": 1.3457,
      "step": 9543
    },
    {
      "epoch": 0.25135633394785356,
      "grad_norm": 3.680968999862671,
      "learning_rate": 3.7441401106136426e-05,
      "loss": 1.1702,
      "step": 9544
    },
    {
      "epoch": 0.2513826705293653,
      "grad_norm": 1.8887977600097656,
      "learning_rate": 3.7440084277060835e-05,
      "loss": 1.3718,
      "step": 9545
    },
    {
      "epoch": 0.251409007110877,
      "grad_norm": 1.5648424625396729,
      "learning_rate": 3.743876744798525e-05,
      "loss": 1.6228,
      "step": 9546
    },
    {
      "epoch": 0.25143534369238874,
      "grad_norm": 3.817568778991699,
      "learning_rate": 3.7437450618909666e-05,
      "loss": 1.0193,
      "step": 9547
    },
    {
      "epoch": 0.2514616802739004,
      "grad_norm": 1.48291015625,
      "learning_rate": 3.743613378983408e-05,
      "loss": 1.6568,
      "step": 9548
    },
    {
      "epoch": 0.25148801685541217,
      "grad_norm": 1.7690905332565308,
      "learning_rate": 3.74348169607585e-05,
      "loss": 1.854,
      "step": 9549
    },
    {
      "epoch": 0.2515143534369239,
      "grad_norm": 2.000364303588867,
      "learning_rate": 3.7433500131682906e-05,
      "loss": 1.9702,
      "step": 9550
    },
    {
      "epoch": 0.2515406900184356,
      "grad_norm": 3.544632911682129,
      "learning_rate": 3.743218330260732e-05,
      "loss": 1.2276,
      "step": 9551
    },
    {
      "epoch": 0.25156702659994734,
      "grad_norm": 1.84476637840271,
      "learning_rate": 3.743086647353174e-05,
      "loss": 1.7679,
      "step": 9552
    },
    {
      "epoch": 0.25159336318145903,
      "grad_norm": 2.341996908187866,
      "learning_rate": 3.742954964445615e-05,
      "loss": 1.7968,
      "step": 9553
    },
    {
      "epoch": 0.25161969976297077,
      "grad_norm": 2.355987548828125,
      "learning_rate": 3.742823281538056e-05,
      "loss": 2.08,
      "step": 9554
    },
    {
      "epoch": 0.25164603634448246,
      "grad_norm": 1.776052713394165,
      "learning_rate": 3.742691598630498e-05,
      "loss": 2.1826,
      "step": 9555
    },
    {
      "epoch": 0.2516723729259942,
      "grad_norm": 2.5412628650665283,
      "learning_rate": 3.742559915722939e-05,
      "loss": 1.868,
      "step": 9556
    },
    {
      "epoch": 0.25169870950750595,
      "grad_norm": 1.9412487745285034,
      "learning_rate": 3.742428232815381e-05,
      "loss": 1.9086,
      "step": 9557
    },
    {
      "epoch": 0.25172504608901763,
      "grad_norm": 2.5037190914154053,
      "learning_rate": 3.7422965499078224e-05,
      "loss": 1.6867,
      "step": 9558
    },
    {
      "epoch": 0.2517513826705294,
      "grad_norm": 9.2487154006958,
      "learning_rate": 3.742164867000263e-05,
      "loss": 2.3271,
      "step": 9559
    },
    {
      "epoch": 0.25177771925204107,
      "grad_norm": 3.0891542434692383,
      "learning_rate": 3.742033184092705e-05,
      "loss": 1.0216,
      "step": 9560
    },
    {
      "epoch": 0.2518040558335528,
      "grad_norm": 2.2402870655059814,
      "learning_rate": 3.7419015011851464e-05,
      "loss": 1.182,
      "step": 9561
    },
    {
      "epoch": 0.25183039241506455,
      "grad_norm": 2.1816024780273438,
      "learning_rate": 3.741769818277588e-05,
      "loss": 1.3848,
      "step": 9562
    },
    {
      "epoch": 0.25185672899657624,
      "grad_norm": 6.280076026916504,
      "learning_rate": 3.741638135370029e-05,
      "loss": 2.5219,
      "step": 9563
    },
    {
      "epoch": 0.251883065578088,
      "grad_norm": 2.479802131652832,
      "learning_rate": 3.7415064524624705e-05,
      "loss": 1.175,
      "step": 9564
    },
    {
      "epoch": 0.25190940215959967,
      "grad_norm": 2.814124345779419,
      "learning_rate": 3.741374769554912e-05,
      "loss": 1.729,
      "step": 9565
    },
    {
      "epoch": 0.2519357387411114,
      "grad_norm": 2.2586817741394043,
      "learning_rate": 3.7412430866473536e-05,
      "loss": 1.9833,
      "step": 9566
    },
    {
      "epoch": 0.2519620753226231,
      "grad_norm": 2.9755196571350098,
      "learning_rate": 3.741111403739795e-05,
      "loss": 1.2308,
      "step": 9567
    },
    {
      "epoch": 0.25198841190413485,
      "grad_norm": 2.1252360343933105,
      "learning_rate": 3.740979720832236e-05,
      "loss": 2.1942,
      "step": 9568
    },
    {
      "epoch": 0.2520147484856466,
      "grad_norm": 2.2151808738708496,
      "learning_rate": 3.7408480379246776e-05,
      "loss": 1.3213,
      "step": 9569
    },
    {
      "epoch": 0.2520410850671583,
      "grad_norm": 1.918562889099121,
      "learning_rate": 3.7407163550171185e-05,
      "loss": 0.3259,
      "step": 9570
    },
    {
      "epoch": 0.25206742164867,
      "grad_norm": 4.135371208190918,
      "learning_rate": 3.740584672109561e-05,
      "loss": 1.8212,
      "step": 9571
    },
    {
      "epoch": 0.2520937582301817,
      "grad_norm": 1.9333744049072266,
      "learning_rate": 3.7404529892020016e-05,
      "loss": 1.7043,
      "step": 9572
    },
    {
      "epoch": 0.25212009481169345,
      "grad_norm": 1.8396936655044556,
      "learning_rate": 3.740321306294443e-05,
      "loss": 2.163,
      "step": 9573
    },
    {
      "epoch": 0.25214643139320514,
      "grad_norm": 1.914660096168518,
      "learning_rate": 3.740189623386885e-05,
      "loss": 1.4764,
      "step": 9574
    },
    {
      "epoch": 0.2521727679747169,
      "grad_norm": 1.869932770729065,
      "learning_rate": 3.740057940479326e-05,
      "loss": 2.1826,
      "step": 9575
    },
    {
      "epoch": 0.2521991045562286,
      "grad_norm": 1.817205548286438,
      "learning_rate": 3.739926257571768e-05,
      "loss": 1.437,
      "step": 9576
    },
    {
      "epoch": 0.2522254411377403,
      "grad_norm": 2.358226776123047,
      "learning_rate": 3.739794574664209e-05,
      "loss": 2.1461,
      "step": 9577
    },
    {
      "epoch": 0.25225177771925206,
      "grad_norm": 3.138655424118042,
      "learning_rate": 3.73966289175665e-05,
      "loss": 1.4892,
      "step": 9578
    },
    {
      "epoch": 0.25227811430076374,
      "grad_norm": 1.862527847290039,
      "learning_rate": 3.739531208849091e-05,
      "loss": 0.8819,
      "step": 9579
    },
    {
      "epoch": 0.2523044508822755,
      "grad_norm": 2.721653461456299,
      "learning_rate": 3.7393995259415334e-05,
      "loss": 2.4808,
      "step": 9580
    },
    {
      "epoch": 0.2523307874637872,
      "grad_norm": 2.8426930904388428,
      "learning_rate": 3.739267843033974e-05,
      "loss": 1.6447,
      "step": 9581
    },
    {
      "epoch": 0.2523571240452989,
      "grad_norm": 2.238645553588867,
      "learning_rate": 3.739136160126416e-05,
      "loss": 1.9963,
      "step": 9582
    },
    {
      "epoch": 0.25238346062681066,
      "grad_norm": 1.9386051893234253,
      "learning_rate": 3.7390044772188574e-05,
      "loss": 1.9467,
      "step": 9583
    },
    {
      "epoch": 0.25240979720832235,
      "grad_norm": 1.8041932582855225,
      "learning_rate": 3.738872794311298e-05,
      "loss": 1.802,
      "step": 9584
    },
    {
      "epoch": 0.2524361337898341,
      "grad_norm": 2.4384052753448486,
      "learning_rate": 3.7387411114037405e-05,
      "loss": 1.6788,
      "step": 9585
    },
    {
      "epoch": 0.2524624703713458,
      "grad_norm": 1.6500581502914429,
      "learning_rate": 3.7386094284961814e-05,
      "loss": 1.7741,
      "step": 9586
    },
    {
      "epoch": 0.2524888069528575,
      "grad_norm": 3.3525314331054688,
      "learning_rate": 3.738477745588623e-05,
      "loss": 1.2178,
      "step": 9587
    },
    {
      "epoch": 0.2525151435343692,
      "grad_norm": 2.4203553199768066,
      "learning_rate": 3.738346062681064e-05,
      "loss": 1.2048,
      "step": 9588
    },
    {
      "epoch": 0.25254148011588096,
      "grad_norm": 1.5846211910247803,
      "learning_rate": 3.7382143797735054e-05,
      "loss": 0.3042,
      "step": 9589
    },
    {
      "epoch": 0.2525678166973927,
      "grad_norm": 2.4709599018096924,
      "learning_rate": 3.738082696865947e-05,
      "loss": 1.898,
      "step": 9590
    },
    {
      "epoch": 0.2525941532789044,
      "grad_norm": 1.626758098602295,
      "learning_rate": 3.7379510139583885e-05,
      "loss": 1.7603,
      "step": 9591
    },
    {
      "epoch": 0.25262048986041613,
      "grad_norm": 3.4821956157684326,
      "learning_rate": 3.7378193310508294e-05,
      "loss": 1.5059,
      "step": 9592
    },
    {
      "epoch": 0.2526468264419278,
      "grad_norm": 2.0917601585388184,
      "learning_rate": 3.737687648143271e-05,
      "loss": 1.7907,
      "step": 9593
    },
    {
      "epoch": 0.25267316302343956,
      "grad_norm": 2.327014207839966,
      "learning_rate": 3.7375559652357125e-05,
      "loss": 2.3821,
      "step": 9594
    },
    {
      "epoch": 0.25269949960495125,
      "grad_norm": 2.840378999710083,
      "learning_rate": 3.737424282328154e-05,
      "loss": 2.8007,
      "step": 9595
    },
    {
      "epoch": 0.252725836186463,
      "grad_norm": 1.7067112922668457,
      "learning_rate": 3.7372925994205956e-05,
      "loss": 1.6312,
      "step": 9596
    },
    {
      "epoch": 0.25275217276797474,
      "grad_norm": 1.8548580408096313,
      "learning_rate": 3.7371609165130365e-05,
      "loss": 1.5074,
      "step": 9597
    },
    {
      "epoch": 0.2527785093494864,
      "grad_norm": 2.1230478286743164,
      "learning_rate": 3.737029233605478e-05,
      "loss": 1.5026,
      "step": 9598
    },
    {
      "epoch": 0.25280484593099817,
      "grad_norm": 4.288404941558838,
      "learning_rate": 3.7368975506979196e-05,
      "loss": 0.8427,
      "step": 9599
    },
    {
      "epoch": 0.25283118251250986,
      "grad_norm": 2.0781381130218506,
      "learning_rate": 3.736765867790361e-05,
      "loss": 1.5195,
      "step": 9600
    },
    {
      "epoch": 0.2528575190940216,
      "grad_norm": 1.8319816589355469,
      "learning_rate": 3.736634184882802e-05,
      "loss": 1.9692,
      "step": 9601
    },
    {
      "epoch": 0.25288385567553334,
      "grad_norm": 3.4707813262939453,
      "learning_rate": 3.7365025019752436e-05,
      "loss": 1.1606,
      "step": 9602
    },
    {
      "epoch": 0.25291019225704503,
      "grad_norm": 5.787014007568359,
      "learning_rate": 3.736370819067685e-05,
      "loss": 1.603,
      "step": 9603
    },
    {
      "epoch": 0.2529365288385568,
      "grad_norm": 2.4565348625183105,
      "learning_rate": 3.736239136160127e-05,
      "loss": 1.6115,
      "step": 9604
    },
    {
      "epoch": 0.25296286542006846,
      "grad_norm": 2.6019861698150635,
      "learning_rate": 3.736107453252568e-05,
      "loss": 0.6824,
      "step": 9605
    },
    {
      "epoch": 0.2529892020015802,
      "grad_norm": 1.8150345087051392,
      "learning_rate": 3.735975770345009e-05,
      "loss": 2.243,
      "step": 9606
    },
    {
      "epoch": 0.2530155385830919,
      "grad_norm": 5.225513458251953,
      "learning_rate": 3.735844087437451e-05,
      "loss": 1.6182,
      "step": 9607
    },
    {
      "epoch": 0.25304187516460364,
      "grad_norm": 2.257249116897583,
      "learning_rate": 3.735712404529892e-05,
      "loss": 1.953,
      "step": 9608
    },
    {
      "epoch": 0.2530682117461154,
      "grad_norm": 4.514675140380859,
      "learning_rate": 3.735580721622334e-05,
      "loss": 1.3993,
      "step": 9609
    },
    {
      "epoch": 0.25309454832762707,
      "grad_norm": 2.362719774246216,
      "learning_rate": 3.735449038714775e-05,
      "loss": 0.6299,
      "step": 9610
    },
    {
      "epoch": 0.2531208849091388,
      "grad_norm": 2.76245379447937,
      "learning_rate": 3.735317355807216e-05,
      "loss": 2.4718,
      "step": 9611
    },
    {
      "epoch": 0.2531472214906505,
      "grad_norm": 2.9407243728637695,
      "learning_rate": 3.735185672899658e-05,
      "loss": 1.8316,
      "step": 9612
    },
    {
      "epoch": 0.25317355807216224,
      "grad_norm": 6.190566539764404,
      "learning_rate": 3.7350539899920994e-05,
      "loss": 2.0521,
      "step": 9613
    },
    {
      "epoch": 0.25319989465367393,
      "grad_norm": 1.9451617002487183,
      "learning_rate": 3.734922307084541e-05,
      "loss": 1.8535,
      "step": 9614
    },
    {
      "epoch": 0.2532262312351857,
      "grad_norm": 1.9362940788269043,
      "learning_rate": 3.734790624176982e-05,
      "loss": 1.7743,
      "step": 9615
    },
    {
      "epoch": 0.2532525678166974,
      "grad_norm": 3.7016282081604004,
      "learning_rate": 3.7346589412694234e-05,
      "loss": 0.7037,
      "step": 9616
    },
    {
      "epoch": 0.2532789043982091,
      "grad_norm": 2.2456562519073486,
      "learning_rate": 3.734527258361864e-05,
      "loss": 1.8348,
      "step": 9617
    },
    {
      "epoch": 0.25330524097972085,
      "grad_norm": 1.6477816104888916,
      "learning_rate": 3.7343955754543065e-05,
      "loss": 1.7671,
      "step": 9618
    },
    {
      "epoch": 0.25333157756123253,
      "grad_norm": 1.6278709173202515,
      "learning_rate": 3.7342638925467474e-05,
      "loss": 1.5653,
      "step": 9619
    },
    {
      "epoch": 0.2533579141427443,
      "grad_norm": 2.5155160427093506,
      "learning_rate": 3.734132209639189e-05,
      "loss": 1.5245,
      "step": 9620
    },
    {
      "epoch": 0.25338425072425597,
      "grad_norm": 2.0082337856292725,
      "learning_rate": 3.7340005267316305e-05,
      "loss": 1.4992,
      "step": 9621
    },
    {
      "epoch": 0.2534105873057677,
      "grad_norm": 6.262524127960205,
      "learning_rate": 3.7338688438240714e-05,
      "loss": 1.6302,
      "step": 9622
    },
    {
      "epoch": 0.25343692388727945,
      "grad_norm": 3.1318280696868896,
      "learning_rate": 3.733737160916514e-05,
      "loss": 1.4374,
      "step": 9623
    },
    {
      "epoch": 0.25346326046879114,
      "grad_norm": 1.822563648223877,
      "learning_rate": 3.7336054780089546e-05,
      "loss": 1.2263,
      "step": 9624
    },
    {
      "epoch": 0.2534895970503029,
      "grad_norm": 2.0121610164642334,
      "learning_rate": 3.733473795101396e-05,
      "loss": 1.7846,
      "step": 9625
    },
    {
      "epoch": 0.25351593363181457,
      "grad_norm": 1.7135684490203857,
      "learning_rate": 3.733342112193837e-05,
      "loss": 2.3712,
      "step": 9626
    },
    {
      "epoch": 0.2535422702133263,
      "grad_norm": 1.620875358581543,
      "learning_rate": 3.733210429286279e-05,
      "loss": 1.4597,
      "step": 9627
    },
    {
      "epoch": 0.253568606794838,
      "grad_norm": 2.601332187652588,
      "learning_rate": 3.73307874637872e-05,
      "loss": 2.0782,
      "step": 9628
    },
    {
      "epoch": 0.25359494337634975,
      "grad_norm": 2.2117395401000977,
      "learning_rate": 3.732947063471162e-05,
      "loss": 2.0553,
      "step": 9629
    },
    {
      "epoch": 0.2536212799578615,
      "grad_norm": 1.6906989812850952,
      "learning_rate": 3.732815380563603e-05,
      "loss": 0.544,
      "step": 9630
    },
    {
      "epoch": 0.2536476165393732,
      "grad_norm": 3.0786561965942383,
      "learning_rate": 3.732683697656044e-05,
      "loss": 1.9324,
      "step": 9631
    },
    {
      "epoch": 0.2536739531208849,
      "grad_norm": 1.5395272970199585,
      "learning_rate": 3.7325520147484864e-05,
      "loss": 1.6829,
      "step": 9632
    },
    {
      "epoch": 0.2537002897023966,
      "grad_norm": 1.7805149555206299,
      "learning_rate": 3.732420331840927e-05,
      "loss": 1.6129,
      "step": 9633
    },
    {
      "epoch": 0.25372662628390835,
      "grad_norm": 2.0277090072631836,
      "learning_rate": 3.732288648933369e-05,
      "loss": 2.0794,
      "step": 9634
    },
    {
      "epoch": 0.2537529628654201,
      "grad_norm": 2.461071014404297,
      "learning_rate": 3.73215696602581e-05,
      "loss": 1.5116,
      "step": 9635
    },
    {
      "epoch": 0.2537792994469318,
      "grad_norm": 1.6589722633361816,
      "learning_rate": 3.732025283118251e-05,
      "loss": 1.9653,
      "step": 9636
    },
    {
      "epoch": 0.2538056360284435,
      "grad_norm": 4.968806266784668,
      "learning_rate": 3.731893600210693e-05,
      "loss": 2.5758,
      "step": 9637
    },
    {
      "epoch": 0.2538319726099552,
      "grad_norm": 5.118461608886719,
      "learning_rate": 3.7317619173031344e-05,
      "loss": 0.8312,
      "step": 9638
    },
    {
      "epoch": 0.25385830919146696,
      "grad_norm": 1.9708750247955322,
      "learning_rate": 3.731630234395575e-05,
      "loss": 1.8017,
      "step": 9639
    },
    {
      "epoch": 0.25388464577297865,
      "grad_norm": 1.9421560764312744,
      "learning_rate": 3.731498551488017e-05,
      "loss": 1.6465,
      "step": 9640
    },
    {
      "epoch": 0.2539109823544904,
      "grad_norm": 1.9702420234680176,
      "learning_rate": 3.731366868580459e-05,
      "loss": 0.9592,
      "step": 9641
    },
    {
      "epoch": 0.25393731893600213,
      "grad_norm": 1.6837055683135986,
      "learning_rate": 3.7312351856729e-05,
      "loss": 1.5235,
      "step": 9642
    },
    {
      "epoch": 0.2539636555175138,
      "grad_norm": 2.189059019088745,
      "learning_rate": 3.7311035027653415e-05,
      "loss": 1.7711,
      "step": 9643
    },
    {
      "epoch": 0.25398999209902556,
      "grad_norm": 2.7619576454162598,
      "learning_rate": 3.7309718198577824e-05,
      "loss": 2.23,
      "step": 9644
    },
    {
      "epoch": 0.25401632868053725,
      "grad_norm": 2.131391763687134,
      "learning_rate": 3.730840136950224e-05,
      "loss": 2.1333,
      "step": 9645
    },
    {
      "epoch": 0.254042665262049,
      "grad_norm": 4.068904399871826,
      "learning_rate": 3.7307084540426655e-05,
      "loss": 0.6892,
      "step": 9646
    },
    {
      "epoch": 0.2540690018435607,
      "grad_norm": 2.9153926372528076,
      "learning_rate": 3.730576771135107e-05,
      "loss": 1.6668,
      "step": 9647
    },
    {
      "epoch": 0.2540953384250724,
      "grad_norm": 3.7939934730529785,
      "learning_rate": 3.730445088227548e-05,
      "loss": 1.4415,
      "step": 9648
    },
    {
      "epoch": 0.25412167500658417,
      "grad_norm": 1.8799753189086914,
      "learning_rate": 3.7303134053199895e-05,
      "loss": 0.4689,
      "step": 9649
    },
    {
      "epoch": 0.25414801158809586,
      "grad_norm": 1.6384416818618774,
      "learning_rate": 3.730181722412431e-05,
      "loss": 1.6233,
      "step": 9650
    },
    {
      "epoch": 0.2541743481696076,
      "grad_norm": 4.40449857711792,
      "learning_rate": 3.7300500395048726e-05,
      "loss": 1.0271,
      "step": 9651
    },
    {
      "epoch": 0.2542006847511193,
      "grad_norm": 1.7981282472610474,
      "learning_rate": 3.729918356597314e-05,
      "loss": 2.2554,
      "step": 9652
    },
    {
      "epoch": 0.25422702133263103,
      "grad_norm": 3.553616523742676,
      "learning_rate": 3.729786673689755e-05,
      "loss": 1.516,
      "step": 9653
    },
    {
      "epoch": 0.2542533579141427,
      "grad_norm": 2.7820286750793457,
      "learning_rate": 3.7296549907821966e-05,
      "loss": 1.3957,
      "step": 9654
    },
    {
      "epoch": 0.25427969449565446,
      "grad_norm": 2.124394178390503,
      "learning_rate": 3.729523307874638e-05,
      "loss": 1.0345,
      "step": 9655
    },
    {
      "epoch": 0.2543060310771662,
      "grad_norm": 2.05969500541687,
      "learning_rate": 3.72939162496708e-05,
      "loss": 1.8036,
      "step": 9656
    },
    {
      "epoch": 0.2543323676586779,
      "grad_norm": 4.607055187225342,
      "learning_rate": 3.7292599420595206e-05,
      "loss": 1.8117,
      "step": 9657
    },
    {
      "epoch": 0.25435870424018964,
      "grad_norm": 2.608032464981079,
      "learning_rate": 3.729128259151962e-05,
      "loss": 1.6529,
      "step": 9658
    },
    {
      "epoch": 0.2543850408217013,
      "grad_norm": 1.6836975812911987,
      "learning_rate": 3.728996576244404e-05,
      "loss": 1.7342,
      "step": 9659
    },
    {
      "epoch": 0.25441137740321307,
      "grad_norm": 2.188162088394165,
      "learning_rate": 3.728864893336845e-05,
      "loss": 1.5757,
      "step": 9660
    },
    {
      "epoch": 0.25443771398472476,
      "grad_norm": 2.5714704990386963,
      "learning_rate": 3.728733210429287e-05,
      "loss": 1.5288,
      "step": 9661
    },
    {
      "epoch": 0.2544640505662365,
      "grad_norm": 1.8239662647247314,
      "learning_rate": 3.728601527521728e-05,
      "loss": 1.7063,
      "step": 9662
    },
    {
      "epoch": 0.25449038714774824,
      "grad_norm": 2.715965986251831,
      "learning_rate": 3.728469844614169e-05,
      "loss": 1.1891,
      "step": 9663
    },
    {
      "epoch": 0.25451672372925993,
      "grad_norm": 2.429906129837036,
      "learning_rate": 3.72833816170661e-05,
      "loss": 0.7726,
      "step": 9664
    },
    {
      "epoch": 0.2545430603107717,
      "grad_norm": 1.5424458980560303,
      "learning_rate": 3.7282064787990524e-05,
      "loss": 1.3551,
      "step": 9665
    },
    {
      "epoch": 0.25456939689228336,
      "grad_norm": 2.020559549331665,
      "learning_rate": 3.728074795891493e-05,
      "loss": 1.5823,
      "step": 9666
    },
    {
      "epoch": 0.2545957334737951,
      "grad_norm": 1.7699886560440063,
      "learning_rate": 3.727943112983935e-05,
      "loss": 2.1829,
      "step": 9667
    },
    {
      "epoch": 0.25462207005530685,
      "grad_norm": 2.178121566772461,
      "learning_rate": 3.7278114300763764e-05,
      "loss": 1.5704,
      "step": 9668
    },
    {
      "epoch": 0.25464840663681854,
      "grad_norm": 6.326632976531982,
      "learning_rate": 3.727679747168817e-05,
      "loss": 0.6792,
      "step": 9669
    },
    {
      "epoch": 0.2546747432183303,
      "grad_norm": 2.2320096492767334,
      "learning_rate": 3.7275480642612595e-05,
      "loss": 1.7751,
      "step": 9670
    },
    {
      "epoch": 0.25470107979984197,
      "grad_norm": 1.9952192306518555,
      "learning_rate": 3.7274163813537004e-05,
      "loss": 1.7288,
      "step": 9671
    },
    {
      "epoch": 0.2547274163813537,
      "grad_norm": 2.917086362838745,
      "learning_rate": 3.727284698446142e-05,
      "loss": 1.3647,
      "step": 9672
    },
    {
      "epoch": 0.2547537529628654,
      "grad_norm": 1.6897929906845093,
      "learning_rate": 3.727153015538583e-05,
      "loss": 1.6752,
      "step": 9673
    },
    {
      "epoch": 0.25478008954437714,
      "grad_norm": 2.030916452407837,
      "learning_rate": 3.727021332631025e-05,
      "loss": 2.0869,
      "step": 9674
    },
    {
      "epoch": 0.2548064261258889,
      "grad_norm": 2.204450845718384,
      "learning_rate": 3.726889649723466e-05,
      "loss": 2.0832,
      "step": 9675
    },
    {
      "epoch": 0.2548327627074006,
      "grad_norm": 2.0757219791412354,
      "learning_rate": 3.7267579668159075e-05,
      "loss": 2.0643,
      "step": 9676
    },
    {
      "epoch": 0.2548590992889123,
      "grad_norm": 1.6636179685592651,
      "learning_rate": 3.726626283908349e-05,
      "loss": 2.0463,
      "step": 9677
    },
    {
      "epoch": 0.254885435870424,
      "grad_norm": 1.8233202695846558,
      "learning_rate": 3.72649460100079e-05,
      "loss": 2.1082,
      "step": 9678
    },
    {
      "epoch": 0.25491177245193575,
      "grad_norm": 3.6059629917144775,
      "learning_rate": 3.726362918093232e-05,
      "loss": 1.4597,
      "step": 9679
    },
    {
      "epoch": 0.25493810903344744,
      "grad_norm": 1.9130117893218994,
      "learning_rate": 3.726231235185673e-05,
      "loss": 1.839,
      "step": 9680
    },
    {
      "epoch": 0.2549644456149592,
      "grad_norm": 4.750220775604248,
      "learning_rate": 3.7260995522781146e-05,
      "loss": 0.6948,
      "step": 9681
    },
    {
      "epoch": 0.2549907821964709,
      "grad_norm": 4.482402801513672,
      "learning_rate": 3.7259678693705555e-05,
      "loss": 1.2409,
      "step": 9682
    },
    {
      "epoch": 0.2550171187779826,
      "grad_norm": 2.10050630569458,
      "learning_rate": 3.725836186462997e-05,
      "loss": 1.5929,
      "step": 9683
    },
    {
      "epoch": 0.25504345535949435,
      "grad_norm": 3.0294337272644043,
      "learning_rate": 3.7257045035554387e-05,
      "loss": 1.4726,
      "step": 9684
    },
    {
      "epoch": 0.25506979194100604,
      "grad_norm": 1.498418927192688,
      "learning_rate": 3.72557282064788e-05,
      "loss": 1.5774,
      "step": 9685
    },
    {
      "epoch": 0.2550961285225178,
      "grad_norm": 1.7631802558898926,
      "learning_rate": 3.725441137740322e-05,
      "loss": 2.1942,
      "step": 9686
    },
    {
      "epoch": 0.25512246510402947,
      "grad_norm": 1.6969068050384521,
      "learning_rate": 3.7253094548327627e-05,
      "loss": 2.3295,
      "step": 9687
    },
    {
      "epoch": 0.2551488016855412,
      "grad_norm": 1.7383668422698975,
      "learning_rate": 3.725177771925205e-05,
      "loss": 2.0288,
      "step": 9688
    },
    {
      "epoch": 0.25517513826705296,
      "grad_norm": 1.8575865030288696,
      "learning_rate": 3.725046089017646e-05,
      "loss": 1.8065,
      "step": 9689
    },
    {
      "epoch": 0.25520147484856465,
      "grad_norm": 2.979454278945923,
      "learning_rate": 3.724914406110087e-05,
      "loss": 1.1739,
      "step": 9690
    },
    {
      "epoch": 0.2552278114300764,
      "grad_norm": 1.554207682609558,
      "learning_rate": 3.724782723202528e-05,
      "loss": 1.743,
      "step": 9691
    },
    {
      "epoch": 0.2552541480115881,
      "grad_norm": 1.430321455001831,
      "learning_rate": 3.72465104029497e-05,
      "loss": 1.4131,
      "step": 9692
    },
    {
      "epoch": 0.2552804845930998,
      "grad_norm": 3.4111132621765137,
      "learning_rate": 3.724519357387411e-05,
      "loss": 1.4037,
      "step": 9693
    },
    {
      "epoch": 0.2553068211746115,
      "grad_norm": 5.329657554626465,
      "learning_rate": 3.724387674479853e-05,
      "loss": 0.9999,
      "step": 9694
    },
    {
      "epoch": 0.25533315775612325,
      "grad_norm": 5.334298133850098,
      "learning_rate": 3.724255991572294e-05,
      "loss": 1.6436,
      "step": 9695
    },
    {
      "epoch": 0.255359494337635,
      "grad_norm": 2.733109951019287,
      "learning_rate": 3.724124308664735e-05,
      "loss": 1.4872,
      "step": 9696
    },
    {
      "epoch": 0.2553858309191467,
      "grad_norm": 3.851780652999878,
      "learning_rate": 3.723992625757177e-05,
      "loss": 1.9459,
      "step": 9697
    },
    {
      "epoch": 0.2554121675006584,
      "grad_norm": 1.7495603561401367,
      "learning_rate": 3.7238609428496185e-05,
      "loss": 1.9351,
      "step": 9698
    },
    {
      "epoch": 0.2554385040821701,
      "grad_norm": 3.068164587020874,
      "learning_rate": 3.72372925994206e-05,
      "loss": 1.1152,
      "step": 9699
    },
    {
      "epoch": 0.25546484066368186,
      "grad_norm": 1.4387151002883911,
      "learning_rate": 3.723597577034501e-05,
      "loss": 2.1803,
      "step": 9700
    },
    {
      "epoch": 0.2554911772451936,
      "grad_norm": 1.874611735343933,
      "learning_rate": 3.7234658941269425e-05,
      "loss": 2.0498,
      "step": 9701
    },
    {
      "epoch": 0.2555175138267053,
      "grad_norm": 2.6465258598327637,
      "learning_rate": 3.723334211219383e-05,
      "loss": 1.3539,
      "step": 9702
    },
    {
      "epoch": 0.25554385040821703,
      "grad_norm": 1.9960482120513916,
      "learning_rate": 3.7232025283118256e-05,
      "loss": 0.4908,
      "step": 9703
    },
    {
      "epoch": 0.2555701869897287,
      "grad_norm": 1.5477030277252197,
      "learning_rate": 3.7230708454042665e-05,
      "loss": 2.0217,
      "step": 9704
    },
    {
      "epoch": 0.25559652357124046,
      "grad_norm": 2.119279146194458,
      "learning_rate": 3.722939162496708e-05,
      "loss": 0.4998,
      "step": 9705
    },
    {
      "epoch": 0.25562286015275215,
      "grad_norm": 3.4473531246185303,
      "learning_rate": 3.7228074795891496e-05,
      "loss": 1.502,
      "step": 9706
    },
    {
      "epoch": 0.2556491967342639,
      "grad_norm": 1.8358161449432373,
      "learning_rate": 3.722675796681591e-05,
      "loss": 0.7645,
      "step": 9707
    },
    {
      "epoch": 0.25567553331577564,
      "grad_norm": 1.765196442604065,
      "learning_rate": 3.722544113774033e-05,
      "loss": 1.8553,
      "step": 9708
    },
    {
      "epoch": 0.2557018698972873,
      "grad_norm": 2.1057639122009277,
      "learning_rate": 3.7224124308664736e-05,
      "loss": 1.1446,
      "step": 9709
    },
    {
      "epoch": 0.25572820647879907,
      "grad_norm": 2.1482160091400146,
      "learning_rate": 3.722280747958915e-05,
      "loss": 1.4923,
      "step": 9710
    },
    {
      "epoch": 0.25575454306031076,
      "grad_norm": 4.367096900939941,
      "learning_rate": 3.722149065051356e-05,
      "loss": 0.7062,
      "step": 9711
    },
    {
      "epoch": 0.2557808796418225,
      "grad_norm": 2.0393383502960205,
      "learning_rate": 3.722017382143798e-05,
      "loss": 1.6342,
      "step": 9712
    },
    {
      "epoch": 0.2558072162233342,
      "grad_norm": 1.8395330905914307,
      "learning_rate": 3.721885699236239e-05,
      "loss": 1.5871,
      "step": 9713
    },
    {
      "epoch": 0.25583355280484593,
      "grad_norm": 1.9447764158248901,
      "learning_rate": 3.721754016328681e-05,
      "loss": 1.8653,
      "step": 9714
    },
    {
      "epoch": 0.2558598893863577,
      "grad_norm": 2.2410926818847656,
      "learning_rate": 3.721622333421122e-05,
      "loss": 0.4755,
      "step": 9715
    },
    {
      "epoch": 0.25588622596786936,
      "grad_norm": 2.066312074661255,
      "learning_rate": 3.721490650513563e-05,
      "loss": 1.0124,
      "step": 9716
    },
    {
      "epoch": 0.2559125625493811,
      "grad_norm": 2.3067197799682617,
      "learning_rate": 3.7213589676060054e-05,
      "loss": 1.1726,
      "step": 9717
    },
    {
      "epoch": 0.2559388991308928,
      "grad_norm": 2.2440619468688965,
      "learning_rate": 3.721227284698446e-05,
      "loss": 2.5235,
      "step": 9718
    },
    {
      "epoch": 0.25596523571240454,
      "grad_norm": 2.3411786556243896,
      "learning_rate": 3.721095601790888e-05,
      "loss": 2.0143,
      "step": 9719
    },
    {
      "epoch": 0.2559915722939162,
      "grad_norm": 1.8204071521759033,
      "learning_rate": 3.720963918883329e-05,
      "loss": 1.6909,
      "step": 9720
    },
    {
      "epoch": 0.25601790887542797,
      "grad_norm": 2.0097641944885254,
      "learning_rate": 3.720832235975771e-05,
      "loss": 1.0509,
      "step": 9721
    },
    {
      "epoch": 0.2560442454569397,
      "grad_norm": 1.885451316833496,
      "learning_rate": 3.720700553068212e-05,
      "loss": 1.2706,
      "step": 9722
    },
    {
      "epoch": 0.2560705820384514,
      "grad_norm": 2.780299425125122,
      "learning_rate": 3.7205688701606534e-05,
      "loss": 1.2881,
      "step": 9723
    },
    {
      "epoch": 0.25609691861996314,
      "grad_norm": 4.1458420753479,
      "learning_rate": 3.720437187253095e-05,
      "loss": 0.8259,
      "step": 9724
    },
    {
      "epoch": 0.25612325520147483,
      "grad_norm": 1.5310566425323486,
      "learning_rate": 3.720305504345536e-05,
      "loss": 2.2724,
      "step": 9725
    },
    {
      "epoch": 0.2561495917829866,
      "grad_norm": 1.6194044351577759,
      "learning_rate": 3.720173821437978e-05,
      "loss": 1.6572,
      "step": 9726
    },
    {
      "epoch": 0.25617592836449826,
      "grad_norm": 3.186223030090332,
      "learning_rate": 3.720042138530419e-05,
      "loss": 0.6834,
      "step": 9727
    },
    {
      "epoch": 0.25620226494601,
      "grad_norm": 2.9181206226348877,
      "learning_rate": 3.7199104556228605e-05,
      "loss": 1.6049,
      "step": 9728
    },
    {
      "epoch": 0.25622860152752175,
      "grad_norm": 2.823756694793701,
      "learning_rate": 3.7197787727153014e-05,
      "loss": 0.7815,
      "step": 9729
    },
    {
      "epoch": 0.25625493810903344,
      "grad_norm": 1.807732105255127,
      "learning_rate": 3.719647089807743e-05,
      "loss": 1.7068,
      "step": 9730
    },
    {
      "epoch": 0.2562812746905452,
      "grad_norm": 2.2641496658325195,
      "learning_rate": 3.7195154069001845e-05,
      "loss": 1.9368,
      "step": 9731
    },
    {
      "epoch": 0.25630761127205687,
      "grad_norm": 3.063370943069458,
      "learning_rate": 3.719383723992626e-05,
      "loss": 1.5704,
      "step": 9732
    },
    {
      "epoch": 0.2563339478535686,
      "grad_norm": 1.7358959913253784,
      "learning_rate": 3.7192520410850676e-05,
      "loss": 2.2386,
      "step": 9733
    },
    {
      "epoch": 0.2563602844350803,
      "grad_norm": 1.8289241790771484,
      "learning_rate": 3.7191203581775085e-05,
      "loss": 2.0111,
      "step": 9734
    },
    {
      "epoch": 0.25638662101659204,
      "grad_norm": 3.409494161605835,
      "learning_rate": 3.71898867526995e-05,
      "loss": 1.0603,
      "step": 9735
    },
    {
      "epoch": 0.2564129575981038,
      "grad_norm": 1.4900327920913696,
      "learning_rate": 3.7188569923623916e-05,
      "loss": 1.4456,
      "step": 9736
    },
    {
      "epoch": 0.2564392941796155,
      "grad_norm": 3.1553726196289062,
      "learning_rate": 3.718725309454833e-05,
      "loss": 2.2029,
      "step": 9737
    },
    {
      "epoch": 0.2564656307611272,
      "grad_norm": 2.7145864963531494,
      "learning_rate": 3.718593626547274e-05,
      "loss": 0.7665,
      "step": 9738
    },
    {
      "epoch": 0.2564919673426389,
      "grad_norm": 2.8754632472991943,
      "learning_rate": 3.7184619436397156e-05,
      "loss": 0.9922,
      "step": 9739
    },
    {
      "epoch": 0.25651830392415065,
      "grad_norm": 1.6334422826766968,
      "learning_rate": 3.718330260732157e-05,
      "loss": 2.2661,
      "step": 9740
    },
    {
      "epoch": 0.2565446405056624,
      "grad_norm": 1.6821001768112183,
      "learning_rate": 3.718198577824599e-05,
      "loss": 2.0174,
      "step": 9741
    },
    {
      "epoch": 0.2565709770871741,
      "grad_norm": 2.5980312824249268,
      "learning_rate": 3.7180668949170396e-05,
      "loss": 1.797,
      "step": 9742
    },
    {
      "epoch": 0.2565973136686858,
      "grad_norm": 1.8130223751068115,
      "learning_rate": 3.717935212009481e-05,
      "loss": 1.6366,
      "step": 9743
    },
    {
      "epoch": 0.2566236502501975,
      "grad_norm": 2.012077569961548,
      "learning_rate": 3.717803529101923e-05,
      "loss": 1.8363,
      "step": 9744
    },
    {
      "epoch": 0.25664998683170925,
      "grad_norm": 2.2008514404296875,
      "learning_rate": 3.717671846194364e-05,
      "loss": 1.6171,
      "step": 9745
    },
    {
      "epoch": 0.25667632341322094,
      "grad_norm": 3.118563413619995,
      "learning_rate": 3.717540163286806e-05,
      "loss": 1.5961,
      "step": 9746
    },
    {
      "epoch": 0.2567026599947327,
      "grad_norm": 2.0176212787628174,
      "learning_rate": 3.717408480379247e-05,
      "loss": 1.6506,
      "step": 9747
    },
    {
      "epoch": 0.25672899657624443,
      "grad_norm": 3.3909292221069336,
      "learning_rate": 3.717276797471688e-05,
      "loss": 1.8123,
      "step": 9748
    },
    {
      "epoch": 0.2567553331577561,
      "grad_norm": 2.441117286682129,
      "learning_rate": 3.717145114564129e-05,
      "loss": 1.4192,
      "step": 9749
    },
    {
      "epoch": 0.25678166973926786,
      "grad_norm": 5.388453006744385,
      "learning_rate": 3.7170134316565714e-05,
      "loss": 1.8883,
      "step": 9750
    },
    {
      "epoch": 0.25680800632077955,
      "grad_norm": 2.9193570613861084,
      "learning_rate": 3.716881748749012e-05,
      "loss": 1.8319,
      "step": 9751
    },
    {
      "epoch": 0.2568343429022913,
      "grad_norm": 2.458540678024292,
      "learning_rate": 3.716750065841454e-05,
      "loss": 1.5926,
      "step": 9752
    },
    {
      "epoch": 0.256860679483803,
      "grad_norm": 2.311155080795288,
      "learning_rate": 3.7166183829338954e-05,
      "loss": 1.7621,
      "step": 9753
    },
    {
      "epoch": 0.2568870160653147,
      "grad_norm": 1.8875707387924194,
      "learning_rate": 3.716486700026337e-05,
      "loss": 1.7791,
      "step": 9754
    },
    {
      "epoch": 0.25691335264682646,
      "grad_norm": 3.3996996879577637,
      "learning_rate": 3.7163550171187786e-05,
      "loss": 2.1299,
      "step": 9755
    },
    {
      "epoch": 0.25693968922833815,
      "grad_norm": 2.3091018199920654,
      "learning_rate": 3.7162233342112194e-05,
      "loss": 1.4769,
      "step": 9756
    },
    {
      "epoch": 0.2569660258098499,
      "grad_norm": 4.962271690368652,
      "learning_rate": 3.716091651303661e-05,
      "loss": 1.5402,
      "step": 9757
    },
    {
      "epoch": 0.2569923623913616,
      "grad_norm": 3.1826913356781006,
      "learning_rate": 3.715959968396102e-05,
      "loss": 1.719,
      "step": 9758
    },
    {
      "epoch": 0.2570186989728733,
      "grad_norm": 1.8783457279205322,
      "learning_rate": 3.715828285488544e-05,
      "loss": 2.2721,
      "step": 9759
    },
    {
      "epoch": 0.257045035554385,
      "grad_norm": 6.030035495758057,
      "learning_rate": 3.715696602580985e-05,
      "loss": 1.4771,
      "step": 9760
    },
    {
      "epoch": 0.25707137213589676,
      "grad_norm": 1.5319576263427734,
      "learning_rate": 3.7155649196734266e-05,
      "loss": 1.1791,
      "step": 9761
    },
    {
      "epoch": 0.2570977087174085,
      "grad_norm": 2.60465407371521,
      "learning_rate": 3.715433236765868e-05,
      "loss": 1.9611,
      "step": 9762
    },
    {
      "epoch": 0.2571240452989202,
      "grad_norm": 3.9286725521087646,
      "learning_rate": 3.715301553858309e-05,
      "loss": 2.2252,
      "step": 9763
    },
    {
      "epoch": 0.25715038188043193,
      "grad_norm": 2.477926731109619,
      "learning_rate": 3.715169870950751e-05,
      "loss": 1.5371,
      "step": 9764
    },
    {
      "epoch": 0.2571767184619436,
      "grad_norm": 1.6478830575942993,
      "learning_rate": 3.715038188043192e-05,
      "loss": 1.765,
      "step": 9765
    },
    {
      "epoch": 0.25720305504345536,
      "grad_norm": 2.1479403972625732,
      "learning_rate": 3.714906505135634e-05,
      "loss": 0.5111,
      "step": 9766
    },
    {
      "epoch": 0.25722939162496705,
      "grad_norm": 4.6129326820373535,
      "learning_rate": 3.7147748222280746e-05,
      "loss": 0.6595,
      "step": 9767
    },
    {
      "epoch": 0.2572557282064788,
      "grad_norm": 1.7507762908935547,
      "learning_rate": 3.714643139320516e-05,
      "loss": 1.698,
      "step": 9768
    },
    {
      "epoch": 0.25728206478799054,
      "grad_norm": 1.6026253700256348,
      "learning_rate": 3.714511456412958e-05,
      "loss": 1.8894,
      "step": 9769
    },
    {
      "epoch": 0.2573084013695022,
      "grad_norm": 1.9273667335510254,
      "learning_rate": 3.714379773505399e-05,
      "loss": 1.9654,
      "step": 9770
    },
    {
      "epoch": 0.25733473795101397,
      "grad_norm": 2.0849626064300537,
      "learning_rate": 3.714248090597841e-05,
      "loss": 2.2426,
      "step": 9771
    },
    {
      "epoch": 0.25736107453252566,
      "grad_norm": 2.39615535736084,
      "learning_rate": 3.714116407690282e-05,
      "loss": 1.7556,
      "step": 9772
    },
    {
      "epoch": 0.2573874111140374,
      "grad_norm": 4.067844390869141,
      "learning_rate": 3.713984724782724e-05,
      "loss": 1.9581,
      "step": 9773
    },
    {
      "epoch": 0.25741374769554914,
      "grad_norm": 1.7253034114837646,
      "learning_rate": 3.713853041875165e-05,
      "loss": 2.5465,
      "step": 9774
    },
    {
      "epoch": 0.25744008427706083,
      "grad_norm": 2.9143927097320557,
      "learning_rate": 3.7137213589676064e-05,
      "loss": 2.1546,
      "step": 9775
    },
    {
      "epoch": 0.2574664208585726,
      "grad_norm": 3.3653507232666016,
      "learning_rate": 3.713589676060047e-05,
      "loss": 1.3208,
      "step": 9776
    },
    {
      "epoch": 0.25749275744008426,
      "grad_norm": 2.5108065605163574,
      "learning_rate": 3.713457993152489e-05,
      "loss": 1.6654,
      "step": 9777
    },
    {
      "epoch": 0.257519094021596,
      "grad_norm": 2.457394599914551,
      "learning_rate": 3.7133263102449304e-05,
      "loss": 1.9365,
      "step": 9778
    },
    {
      "epoch": 0.2575454306031077,
      "grad_norm": 1.4360820055007935,
      "learning_rate": 3.713194627337372e-05,
      "loss": 2.0449,
      "step": 9779
    },
    {
      "epoch": 0.25757176718461944,
      "grad_norm": 1.6703145503997803,
      "learning_rate": 3.7130629444298135e-05,
      "loss": 2.0669,
      "step": 9780
    },
    {
      "epoch": 0.2575981037661312,
      "grad_norm": 3.4239883422851562,
      "learning_rate": 3.7129312615222544e-05,
      "loss": 1.7407,
      "step": 9781
    },
    {
      "epoch": 0.25762444034764287,
      "grad_norm": 1.9549875259399414,
      "learning_rate": 3.712799578614696e-05,
      "loss": 1.3951,
      "step": 9782
    },
    {
      "epoch": 0.2576507769291546,
      "grad_norm": 1.9713608026504517,
      "learning_rate": 3.7126678957071375e-05,
      "loss": 1.8081,
      "step": 9783
    },
    {
      "epoch": 0.2576771135106663,
      "grad_norm": 2.0435633659362793,
      "learning_rate": 3.712536212799579e-05,
      "loss": 2.6383,
      "step": 9784
    },
    {
      "epoch": 0.25770345009217804,
      "grad_norm": 1.9134286642074585,
      "learning_rate": 3.71240452989202e-05,
      "loss": 1.7348,
      "step": 9785
    },
    {
      "epoch": 0.25772978667368973,
      "grad_norm": 3.18192195892334,
      "learning_rate": 3.7122728469844615e-05,
      "loss": 1.7457,
      "step": 9786
    },
    {
      "epoch": 0.2577561232552015,
      "grad_norm": 2.622532844543457,
      "learning_rate": 3.712141164076903e-05,
      "loss": 1.011,
      "step": 9787
    },
    {
      "epoch": 0.2577824598367132,
      "grad_norm": 2.440190076828003,
      "learning_rate": 3.7120094811693446e-05,
      "loss": 1.1379,
      "step": 9788
    },
    {
      "epoch": 0.2578087964182249,
      "grad_norm": 2.6533925533294678,
      "learning_rate": 3.711877798261786e-05,
      "loss": 1.9678,
      "step": 9789
    },
    {
      "epoch": 0.25783513299973665,
      "grad_norm": 2.3086330890655518,
      "learning_rate": 3.711746115354227e-05,
      "loss": 3.0058,
      "step": 9790
    },
    {
      "epoch": 0.25786146958124834,
      "grad_norm": 1.8761622905731201,
      "learning_rate": 3.7116144324466686e-05,
      "loss": 2.3342,
      "step": 9791
    },
    {
      "epoch": 0.2578878061627601,
      "grad_norm": 2.0247251987457275,
      "learning_rate": 3.71148274953911e-05,
      "loss": 0.8121,
      "step": 9792
    },
    {
      "epoch": 0.25791414274427177,
      "grad_norm": 3.9546287059783936,
      "learning_rate": 3.711351066631552e-05,
      "loss": 1.0587,
      "step": 9793
    },
    {
      "epoch": 0.2579404793257835,
      "grad_norm": 3.985302209854126,
      "learning_rate": 3.7112193837239926e-05,
      "loss": 1.2702,
      "step": 9794
    },
    {
      "epoch": 0.25796681590729525,
      "grad_norm": 3.0532419681549072,
      "learning_rate": 3.711087700816434e-05,
      "loss": 2.1767,
      "step": 9795
    },
    {
      "epoch": 0.25799315248880694,
      "grad_norm": 2.437147378921509,
      "learning_rate": 3.710956017908875e-05,
      "loss": 2.0605,
      "step": 9796
    },
    {
      "epoch": 0.2580194890703187,
      "grad_norm": 2.489968776702881,
      "learning_rate": 3.710824335001317e-05,
      "loss": 1.9573,
      "step": 9797
    },
    {
      "epoch": 0.2580458256518304,
      "grad_norm": 2.065798759460449,
      "learning_rate": 3.710692652093758e-05,
      "loss": 1.7157,
      "step": 9798
    },
    {
      "epoch": 0.2580721622333421,
      "grad_norm": 3.5356953144073486,
      "learning_rate": 3.7105609691862e-05,
      "loss": 2.182,
      "step": 9799
    },
    {
      "epoch": 0.2580984988148538,
      "grad_norm": 1.7834603786468506,
      "learning_rate": 3.710429286278641e-05,
      "loss": 1.7254,
      "step": 9800
    },
    {
      "epoch": 0.25812483539636555,
      "grad_norm": 1.8092585802078247,
      "learning_rate": 3.710297603371082e-05,
      "loss": 1.4736,
      "step": 9801
    },
    {
      "epoch": 0.2581511719778773,
      "grad_norm": 4.423865795135498,
      "learning_rate": 3.7101659204635244e-05,
      "loss": 1.285,
      "step": 9802
    },
    {
      "epoch": 0.258177508559389,
      "grad_norm": 2.351874589920044,
      "learning_rate": 3.710034237555965e-05,
      "loss": 1.6583,
      "step": 9803
    },
    {
      "epoch": 0.2582038451409007,
      "grad_norm": 3.509718418121338,
      "learning_rate": 3.709902554648407e-05,
      "loss": 1.4889,
      "step": 9804
    },
    {
      "epoch": 0.2582301817224124,
      "grad_norm": 2.1435463428497314,
      "learning_rate": 3.709770871740848e-05,
      "loss": 1.9123,
      "step": 9805
    },
    {
      "epoch": 0.25825651830392415,
      "grad_norm": 1.7862725257873535,
      "learning_rate": 3.70963918883329e-05,
      "loss": 1.6919,
      "step": 9806
    },
    {
      "epoch": 0.2582828548854359,
      "grad_norm": 3.4659645557403564,
      "learning_rate": 3.709507505925731e-05,
      "loss": 1.1544,
      "step": 9807
    },
    {
      "epoch": 0.2583091914669476,
      "grad_norm": 1.6393003463745117,
      "learning_rate": 3.7093758230181724e-05,
      "loss": 1.9384,
      "step": 9808
    },
    {
      "epoch": 0.25833552804845933,
      "grad_norm": 2.61045503616333,
      "learning_rate": 3.709244140110614e-05,
      "loss": 1.5775,
      "step": 9809
    },
    {
      "epoch": 0.258361864629971,
      "grad_norm": 2.4012460708618164,
      "learning_rate": 3.709112457203055e-05,
      "loss": 1.7477,
      "step": 9810
    },
    {
      "epoch": 0.25838820121148276,
      "grad_norm": 2.9234864711761475,
      "learning_rate": 3.708980774295497e-05,
      "loss": 2.2103,
      "step": 9811
    },
    {
      "epoch": 0.25841453779299445,
      "grad_norm": 2.4255216121673584,
      "learning_rate": 3.708849091387938e-05,
      "loss": 1.7509,
      "step": 9812
    },
    {
      "epoch": 0.2584408743745062,
      "grad_norm": 3.6548266410827637,
      "learning_rate": 3.7087174084803795e-05,
      "loss": 1.0274,
      "step": 9813
    },
    {
      "epoch": 0.25846721095601793,
      "grad_norm": 4.084821701049805,
      "learning_rate": 3.7085857255728204e-05,
      "loss": 1.5399,
      "step": 9814
    },
    {
      "epoch": 0.2584935475375296,
      "grad_norm": 1.5028916597366333,
      "learning_rate": 3.708454042665262e-05,
      "loss": 1.7955,
      "step": 9815
    },
    {
      "epoch": 0.25851988411904137,
      "grad_norm": 2.013051986694336,
      "learning_rate": 3.7083223597577035e-05,
      "loss": 1.3864,
      "step": 9816
    },
    {
      "epoch": 0.25854622070055305,
      "grad_norm": 4.284978866577148,
      "learning_rate": 3.708190676850145e-05,
      "loss": 1.2436,
      "step": 9817
    },
    {
      "epoch": 0.2585725572820648,
      "grad_norm": 2.6555631160736084,
      "learning_rate": 3.7080589939425867e-05,
      "loss": 1.7561,
      "step": 9818
    },
    {
      "epoch": 0.2585988938635765,
      "grad_norm": 2.107290267944336,
      "learning_rate": 3.7079273110350275e-05,
      "loss": 1.793,
      "step": 9819
    },
    {
      "epoch": 0.25862523044508823,
      "grad_norm": 2.438192129135132,
      "learning_rate": 3.70779562812747e-05,
      "loss": 1.6924,
      "step": 9820
    },
    {
      "epoch": 0.25865156702659997,
      "grad_norm": 1.8615882396697998,
      "learning_rate": 3.7076639452199107e-05,
      "loss": 1.4206,
      "step": 9821
    },
    {
      "epoch": 0.25867790360811166,
      "grad_norm": 2.988931179046631,
      "learning_rate": 3.707532262312352e-05,
      "loss": 1.9875,
      "step": 9822
    },
    {
      "epoch": 0.2587042401896234,
      "grad_norm": 2.202460765838623,
      "learning_rate": 3.707400579404793e-05,
      "loss": 1.4759,
      "step": 9823
    },
    {
      "epoch": 0.2587305767711351,
      "grad_norm": 2.3205885887145996,
      "learning_rate": 3.7072688964972347e-05,
      "loss": 1.7492,
      "step": 9824
    },
    {
      "epoch": 0.25875691335264683,
      "grad_norm": 3.144115447998047,
      "learning_rate": 3.707137213589676e-05,
      "loss": 2.4225,
      "step": 9825
    },
    {
      "epoch": 0.2587832499341585,
      "grad_norm": 1.8301434516906738,
      "learning_rate": 3.707005530682118e-05,
      "loss": 1.0252,
      "step": 9826
    },
    {
      "epoch": 0.25880958651567026,
      "grad_norm": 1.796115756034851,
      "learning_rate": 3.706873847774559e-05,
      "loss": 1.8776,
      "step": 9827
    },
    {
      "epoch": 0.258835923097182,
      "grad_norm": 2.254316806793213,
      "learning_rate": 3.706742164867e-05,
      "loss": 2.3335,
      "step": 9828
    },
    {
      "epoch": 0.2588622596786937,
      "grad_norm": 2.632594347000122,
      "learning_rate": 3.706610481959442e-05,
      "loss": 2.092,
      "step": 9829
    },
    {
      "epoch": 0.25888859626020544,
      "grad_norm": 1.641855239868164,
      "learning_rate": 3.706478799051883e-05,
      "loss": 1.7947,
      "step": 9830
    },
    {
      "epoch": 0.2589149328417171,
      "grad_norm": 1.7753918170928955,
      "learning_rate": 3.706347116144325e-05,
      "loss": 1.8587,
      "step": 9831
    },
    {
      "epoch": 0.25894126942322887,
      "grad_norm": 2.247600793838501,
      "learning_rate": 3.706215433236766e-05,
      "loss": 2.5807,
      "step": 9832
    },
    {
      "epoch": 0.25896760600474056,
      "grad_norm": 2.4937613010406494,
      "learning_rate": 3.7060837503292073e-05,
      "loss": 0.4096,
      "step": 9833
    },
    {
      "epoch": 0.2589939425862523,
      "grad_norm": 2.8936948776245117,
      "learning_rate": 3.705952067421649e-05,
      "loss": 1.5429,
      "step": 9834
    },
    {
      "epoch": 0.25902027916776404,
      "grad_norm": 1.5941886901855469,
      "learning_rate": 3.7058203845140905e-05,
      "loss": 2.0619,
      "step": 9835
    },
    {
      "epoch": 0.25904661574927573,
      "grad_norm": 1.6545512676239014,
      "learning_rate": 3.705688701606532e-05,
      "loss": 0.6942,
      "step": 9836
    },
    {
      "epoch": 0.2590729523307875,
      "grad_norm": 1.8232911825180054,
      "learning_rate": 3.705557018698973e-05,
      "loss": 2.2493,
      "step": 9837
    },
    {
      "epoch": 0.25909928891229916,
      "grad_norm": 2.0688045024871826,
      "learning_rate": 3.7054253357914145e-05,
      "loss": 1.8822,
      "step": 9838
    },
    {
      "epoch": 0.2591256254938109,
      "grad_norm": 1.9047807455062866,
      "learning_rate": 3.705293652883856e-05,
      "loss": 1.6492,
      "step": 9839
    },
    {
      "epoch": 0.25915196207532265,
      "grad_norm": 1.8137409687042236,
      "learning_rate": 3.7051619699762976e-05,
      "loss": 1.9918,
      "step": 9840
    },
    {
      "epoch": 0.25917829865683434,
      "grad_norm": 2.571833848953247,
      "learning_rate": 3.7050302870687385e-05,
      "loss": 0.7357,
      "step": 9841
    },
    {
      "epoch": 0.2592046352383461,
      "grad_norm": 1.9742419719696045,
      "learning_rate": 3.70489860416118e-05,
      "loss": 2.6158,
      "step": 9842
    },
    {
      "epoch": 0.25923097181985777,
      "grad_norm": 1.8618266582489014,
      "learning_rate": 3.704766921253621e-05,
      "loss": 1.9506,
      "step": 9843
    },
    {
      "epoch": 0.2592573084013695,
      "grad_norm": 1.8881477117538452,
      "learning_rate": 3.704635238346063e-05,
      "loss": 1.622,
      "step": 9844
    },
    {
      "epoch": 0.2592836449828812,
      "grad_norm": 3.0669729709625244,
      "learning_rate": 3.704503555438504e-05,
      "loss": 2.5703,
      "step": 9845
    },
    {
      "epoch": 0.25930998156439294,
      "grad_norm": 2.1940395832061768,
      "learning_rate": 3.7043718725309456e-05,
      "loss": 0.7745,
      "step": 9846
    },
    {
      "epoch": 0.2593363181459047,
      "grad_norm": 1.9180564880371094,
      "learning_rate": 3.704240189623387e-05,
      "loss": 1.6913,
      "step": 9847
    },
    {
      "epoch": 0.2593626547274164,
      "grad_norm": 1.9764330387115479,
      "learning_rate": 3.704108506715828e-05,
      "loss": 1.8624,
      "step": 9848
    },
    {
      "epoch": 0.2593889913089281,
      "grad_norm": 2.088878631591797,
      "learning_rate": 3.70397682380827e-05,
      "loss": 1.8017,
      "step": 9849
    },
    {
      "epoch": 0.2594153278904398,
      "grad_norm": 1.4600499868392944,
      "learning_rate": 3.703845140900711e-05,
      "loss": 1.5805,
      "step": 9850
    },
    {
      "epoch": 0.25944166447195155,
      "grad_norm": 3.7565665245056152,
      "learning_rate": 3.703713457993153e-05,
      "loss": 2.0024,
      "step": 9851
    },
    {
      "epoch": 0.25946800105346324,
      "grad_norm": 3.5762574672698975,
      "learning_rate": 3.7035817750855936e-05,
      "loss": 1.1081,
      "step": 9852
    },
    {
      "epoch": 0.259494337634975,
      "grad_norm": 2.9315707683563232,
      "learning_rate": 3.703450092178036e-05,
      "loss": 2.3125,
      "step": 9853
    },
    {
      "epoch": 0.2595206742164867,
      "grad_norm": 2.9471585750579834,
      "learning_rate": 3.703318409270477e-05,
      "loss": 1.5932,
      "step": 9854
    },
    {
      "epoch": 0.2595470107979984,
      "grad_norm": 3.957176685333252,
      "learning_rate": 3.703186726362918e-05,
      "loss": 0.9993,
      "step": 9855
    },
    {
      "epoch": 0.25957334737951016,
      "grad_norm": 1.6149044036865234,
      "learning_rate": 3.70305504345536e-05,
      "loss": 1.3921,
      "step": 9856
    },
    {
      "epoch": 0.25959968396102184,
      "grad_norm": 2.1875369548797607,
      "learning_rate": 3.702923360547801e-05,
      "loss": 1.2329,
      "step": 9857
    },
    {
      "epoch": 0.2596260205425336,
      "grad_norm": 6.689393520355225,
      "learning_rate": 3.702791677640243e-05,
      "loss": 2.0037,
      "step": 9858
    },
    {
      "epoch": 0.2596523571240453,
      "grad_norm": 2.014817476272583,
      "learning_rate": 3.702659994732684e-05,
      "loss": 1.9978,
      "step": 9859
    },
    {
      "epoch": 0.259678693705557,
      "grad_norm": 3.453049659729004,
      "learning_rate": 3.7025283118251254e-05,
      "loss": 1.3455,
      "step": 9860
    },
    {
      "epoch": 0.25970503028706876,
      "grad_norm": 2.203376531600952,
      "learning_rate": 3.702396628917566e-05,
      "loss": 1.6716,
      "step": 9861
    },
    {
      "epoch": 0.25973136686858045,
      "grad_norm": 1.8605821132659912,
      "learning_rate": 3.702264946010008e-05,
      "loss": 2.3472,
      "step": 9862
    },
    {
      "epoch": 0.2597577034500922,
      "grad_norm": 3.7724006175994873,
      "learning_rate": 3.7021332631024494e-05,
      "loss": 1.4838,
      "step": 9863
    },
    {
      "epoch": 0.2597840400316039,
      "grad_norm": 2.3422861099243164,
      "learning_rate": 3.702001580194891e-05,
      "loss": 1.6953,
      "step": 9864
    },
    {
      "epoch": 0.2598103766131156,
      "grad_norm": 2.079458475112915,
      "learning_rate": 3.7018698972873325e-05,
      "loss": 1.4839,
      "step": 9865
    },
    {
      "epoch": 0.2598367131946273,
      "grad_norm": 1.7584192752838135,
      "learning_rate": 3.7017382143797734e-05,
      "loss": 1.9271,
      "step": 9866
    },
    {
      "epoch": 0.25986304977613905,
      "grad_norm": 1.5001235008239746,
      "learning_rate": 3.701606531472215e-05,
      "loss": 1.8783,
      "step": 9867
    },
    {
      "epoch": 0.2598893863576508,
      "grad_norm": 1.7222158908843994,
      "learning_rate": 3.7014748485646565e-05,
      "loss": 1.7547,
      "step": 9868
    },
    {
      "epoch": 0.2599157229391625,
      "grad_norm": 2.3868236541748047,
      "learning_rate": 3.701343165657098e-05,
      "loss": 1.7547,
      "step": 9869
    },
    {
      "epoch": 0.25994205952067423,
      "grad_norm": 1.9237710237503052,
      "learning_rate": 3.701211482749539e-05,
      "loss": 1.8095,
      "step": 9870
    },
    {
      "epoch": 0.2599683961021859,
      "grad_norm": 2.7741787433624268,
      "learning_rate": 3.7010797998419805e-05,
      "loss": 0.8023,
      "step": 9871
    },
    {
      "epoch": 0.25999473268369766,
      "grad_norm": 3.004370927810669,
      "learning_rate": 3.700948116934422e-05,
      "loss": 1.7289,
      "step": 9872
    },
    {
      "epoch": 0.2600210692652094,
      "grad_norm": 2.0840609073638916,
      "learning_rate": 3.7008164340268636e-05,
      "loss": 2.2154,
      "step": 9873
    },
    {
      "epoch": 0.2600474058467211,
      "grad_norm": 1.8691085577011108,
      "learning_rate": 3.700684751119305e-05,
      "loss": 1.9526,
      "step": 9874
    },
    {
      "epoch": 0.26007374242823283,
      "grad_norm": 4.453634738922119,
      "learning_rate": 3.700553068211746e-05,
      "loss": 1.2187,
      "step": 9875
    },
    {
      "epoch": 0.2601000790097445,
      "grad_norm": 1.4608912467956543,
      "learning_rate": 3.7004213853041876e-05,
      "loss": 1.92,
      "step": 9876
    },
    {
      "epoch": 0.26012641559125627,
      "grad_norm": 1.9461586475372314,
      "learning_rate": 3.700289702396629e-05,
      "loss": 1.1693,
      "step": 9877
    },
    {
      "epoch": 0.26015275217276795,
      "grad_norm": 2.532621145248413,
      "learning_rate": 3.700158019489071e-05,
      "loss": 2.1008,
      "step": 9878
    },
    {
      "epoch": 0.2601790887542797,
      "grad_norm": 2.151224374771118,
      "learning_rate": 3.7000263365815116e-05,
      "loss": 0.9723,
      "step": 9879
    },
    {
      "epoch": 0.26020542533579144,
      "grad_norm": 3.0210020542144775,
      "learning_rate": 3.699894653673953e-05,
      "loss": 1.7721,
      "step": 9880
    },
    {
      "epoch": 0.26023176191730313,
      "grad_norm": 2.443286418914795,
      "learning_rate": 3.699762970766395e-05,
      "loss": 1.8425,
      "step": 9881
    },
    {
      "epoch": 0.26025809849881487,
      "grad_norm": 6.711511611938477,
      "learning_rate": 3.699631287858836e-05,
      "loss": 1.5152,
      "step": 9882
    },
    {
      "epoch": 0.26028443508032656,
      "grad_norm": 2.5607032775878906,
      "learning_rate": 3.699499604951278e-05,
      "loss": 0.8706,
      "step": 9883
    },
    {
      "epoch": 0.2603107716618383,
      "grad_norm": 1.7296491861343384,
      "learning_rate": 3.699367922043719e-05,
      "loss": 2.0387,
      "step": 9884
    },
    {
      "epoch": 0.26033710824335,
      "grad_norm": 2.640047311782837,
      "learning_rate": 3.69923623913616e-05,
      "loss": 1.0993,
      "step": 9885
    },
    {
      "epoch": 0.26036344482486173,
      "grad_norm": 2.0113070011138916,
      "learning_rate": 3.699104556228602e-05,
      "loss": 1.2701,
      "step": 9886
    },
    {
      "epoch": 0.2603897814063735,
      "grad_norm": 2.4416544437408447,
      "learning_rate": 3.6989728733210434e-05,
      "loss": 0.801,
      "step": 9887
    },
    {
      "epoch": 0.26041611798788517,
      "grad_norm": 2.1388752460479736,
      "learning_rate": 3.698841190413484e-05,
      "loss": 1.9399,
      "step": 9888
    },
    {
      "epoch": 0.2604424545693969,
      "grad_norm": 1.9972481727600098,
      "learning_rate": 3.698709507505926e-05,
      "loss": 2.1499,
      "step": 9889
    },
    {
      "epoch": 0.2604687911509086,
      "grad_norm": 1.8717926740646362,
      "learning_rate": 3.6985778245983674e-05,
      "loss": 1.7889,
      "step": 9890
    },
    {
      "epoch": 0.26049512773242034,
      "grad_norm": 1.902442455291748,
      "learning_rate": 3.698446141690809e-05,
      "loss": 2.1941,
      "step": 9891
    },
    {
      "epoch": 0.260521464313932,
      "grad_norm": 2.221508741378784,
      "learning_rate": 3.6983144587832506e-05,
      "loss": 1.5553,
      "step": 9892
    },
    {
      "epoch": 0.26054780089544377,
      "grad_norm": 3.134248971939087,
      "learning_rate": 3.6981827758756914e-05,
      "loss": 1.6244,
      "step": 9893
    },
    {
      "epoch": 0.2605741374769555,
      "grad_norm": 3.5815107822418213,
      "learning_rate": 3.698051092968133e-05,
      "loss": 1.7655,
      "step": 9894
    },
    {
      "epoch": 0.2606004740584672,
      "grad_norm": 1.7182793617248535,
      "learning_rate": 3.697919410060574e-05,
      "loss": 1.4358,
      "step": 9895
    },
    {
      "epoch": 0.26062681063997895,
      "grad_norm": 1.9965176582336426,
      "learning_rate": 3.697787727153016e-05,
      "loss": 1.9596,
      "step": 9896
    },
    {
      "epoch": 0.26065314722149063,
      "grad_norm": 1.9375555515289307,
      "learning_rate": 3.697656044245457e-05,
      "loss": 1.4071,
      "step": 9897
    },
    {
      "epoch": 0.2606794838030024,
      "grad_norm": 4.147969722747803,
      "learning_rate": 3.6975243613378986e-05,
      "loss": 0.9175,
      "step": 9898
    },
    {
      "epoch": 0.26070582038451406,
      "grad_norm": 1.8712466955184937,
      "learning_rate": 3.6973926784303394e-05,
      "loss": 2.114,
      "step": 9899
    },
    {
      "epoch": 0.2607321569660258,
      "grad_norm": 1.628216028213501,
      "learning_rate": 3.697260995522781e-05,
      "loss": 1.8131,
      "step": 9900
    },
    {
      "epoch": 0.26075849354753755,
      "grad_norm": 1.7184875011444092,
      "learning_rate": 3.6971293126152226e-05,
      "loss": 1.756,
      "step": 9901
    },
    {
      "epoch": 0.26078483012904924,
      "grad_norm": 2.2704854011535645,
      "learning_rate": 3.696997629707664e-05,
      "loss": 0.5708,
      "step": 9902
    },
    {
      "epoch": 0.260811166710561,
      "grad_norm": 3.0642383098602295,
      "learning_rate": 3.696865946800106e-05,
      "loss": 3.1794,
      "step": 9903
    },
    {
      "epoch": 0.26083750329207267,
      "grad_norm": 2.0807080268859863,
      "learning_rate": 3.6967342638925466e-05,
      "loss": 2.1727,
      "step": 9904
    },
    {
      "epoch": 0.2608638398735844,
      "grad_norm": 1.9923043251037598,
      "learning_rate": 3.696602580984989e-05,
      "loss": 1.7169,
      "step": 9905
    },
    {
      "epoch": 0.2608901764550961,
      "grad_norm": 4.368671417236328,
      "learning_rate": 3.69647089807743e-05,
      "loss": 1.3595,
      "step": 9906
    },
    {
      "epoch": 0.26091651303660784,
      "grad_norm": 1.7777026891708374,
      "learning_rate": 3.696339215169871e-05,
      "loss": 1.8276,
      "step": 9907
    },
    {
      "epoch": 0.2609428496181196,
      "grad_norm": 2.1066837310791016,
      "learning_rate": 3.696207532262312e-05,
      "loss": 1.9016,
      "step": 9908
    },
    {
      "epoch": 0.2609691861996313,
      "grad_norm": 5.21611213684082,
      "learning_rate": 3.696075849354754e-05,
      "loss": 1.5036,
      "step": 9909
    },
    {
      "epoch": 0.260995522781143,
      "grad_norm": 2.433588743209839,
      "learning_rate": 3.695944166447195e-05,
      "loss": 1.7911,
      "step": 9910
    },
    {
      "epoch": 0.2610218593626547,
      "grad_norm": 3.4112401008605957,
      "learning_rate": 3.695812483539637e-05,
      "loss": 1.7093,
      "step": 9911
    },
    {
      "epoch": 0.26104819594416645,
      "grad_norm": 2.103821039199829,
      "learning_rate": 3.6956808006320784e-05,
      "loss": 1.9149,
      "step": 9912
    },
    {
      "epoch": 0.2610745325256782,
      "grad_norm": 2.3162713050842285,
      "learning_rate": 3.695549117724519e-05,
      "loss": 1.6869,
      "step": 9913
    },
    {
      "epoch": 0.2611008691071899,
      "grad_norm": 2.0363857746124268,
      "learning_rate": 3.695417434816961e-05,
      "loss": 1.6803,
      "step": 9914
    },
    {
      "epoch": 0.2611272056887016,
      "grad_norm": 3.4832189083099365,
      "learning_rate": 3.6952857519094024e-05,
      "loss": 1.8491,
      "step": 9915
    },
    {
      "epoch": 0.2611535422702133,
      "grad_norm": 3.0477230548858643,
      "learning_rate": 3.695154069001844e-05,
      "loss": 1.2986,
      "step": 9916
    },
    {
      "epoch": 0.26117987885172506,
      "grad_norm": 1.7535345554351807,
      "learning_rate": 3.695022386094285e-05,
      "loss": 2.2733,
      "step": 9917
    },
    {
      "epoch": 0.26120621543323674,
      "grad_norm": 1.6236014366149902,
      "learning_rate": 3.6948907031867264e-05,
      "loss": 2.0663,
      "step": 9918
    },
    {
      "epoch": 0.2612325520147485,
      "grad_norm": 1.8621058464050293,
      "learning_rate": 3.694759020279168e-05,
      "loss": 1.5392,
      "step": 9919
    },
    {
      "epoch": 0.26125888859626023,
      "grad_norm": 1.4390201568603516,
      "learning_rate": 3.6946273373716095e-05,
      "loss": 1.9462,
      "step": 9920
    },
    {
      "epoch": 0.2612852251777719,
      "grad_norm": 3.6695492267608643,
      "learning_rate": 3.694495654464051e-05,
      "loss": 1.5147,
      "step": 9921
    },
    {
      "epoch": 0.26131156175928366,
      "grad_norm": 2.468250274658203,
      "learning_rate": 3.694363971556492e-05,
      "loss": 1.2001,
      "step": 9922
    },
    {
      "epoch": 0.26133789834079535,
      "grad_norm": 2.3285114765167236,
      "learning_rate": 3.6942322886489335e-05,
      "loss": 2.0223,
      "step": 9923
    },
    {
      "epoch": 0.2613642349223071,
      "grad_norm": 1.681993007659912,
      "learning_rate": 3.694100605741375e-05,
      "loss": 1.7715,
      "step": 9924
    },
    {
      "epoch": 0.2613905715038188,
      "grad_norm": 1.577872395515442,
      "learning_rate": 3.6939689228338166e-05,
      "loss": 1.9831,
      "step": 9925
    },
    {
      "epoch": 0.2614169080853305,
      "grad_norm": 3.7698237895965576,
      "learning_rate": 3.6938372399262575e-05,
      "loss": 1.3589,
      "step": 9926
    },
    {
      "epoch": 0.26144324466684227,
      "grad_norm": 2.0850603580474854,
      "learning_rate": 3.693705557018699e-05,
      "loss": 1.534,
      "step": 9927
    },
    {
      "epoch": 0.26146958124835395,
      "grad_norm": 3.415149211883545,
      "learning_rate": 3.6935738741111406e-05,
      "loss": 1.3738,
      "step": 9928
    },
    {
      "epoch": 0.2614959178298657,
      "grad_norm": 2.7064318656921387,
      "learning_rate": 3.693442191203582e-05,
      "loss": 1.9847,
      "step": 9929
    },
    {
      "epoch": 0.2615222544113774,
      "grad_norm": 2.607041597366333,
      "learning_rate": 3.693310508296024e-05,
      "loss": 1.8455,
      "step": 9930
    },
    {
      "epoch": 0.26154859099288913,
      "grad_norm": 2.6907150745391846,
      "learning_rate": 3.6931788253884646e-05,
      "loss": 0.9775,
      "step": 9931
    },
    {
      "epoch": 0.2615749275744008,
      "grad_norm": 1.8635156154632568,
      "learning_rate": 3.693047142480906e-05,
      "loss": 1.5347,
      "step": 9932
    },
    {
      "epoch": 0.26160126415591256,
      "grad_norm": 2.091691732406616,
      "learning_rate": 3.692915459573347e-05,
      "loss": 2.0928,
      "step": 9933
    },
    {
      "epoch": 0.2616276007374243,
      "grad_norm": 2.8933329582214355,
      "learning_rate": 3.692783776665789e-05,
      "loss": 1.2597,
      "step": 9934
    },
    {
      "epoch": 0.261653937318936,
      "grad_norm": 1.788756012916565,
      "learning_rate": 3.69265209375823e-05,
      "loss": 1.6322,
      "step": 9935
    },
    {
      "epoch": 0.26168027390044774,
      "grad_norm": 5.010241508483887,
      "learning_rate": 3.692520410850672e-05,
      "loss": 1.2798,
      "step": 9936
    },
    {
      "epoch": 0.2617066104819594,
      "grad_norm": 2.166435480117798,
      "learning_rate": 3.692388727943113e-05,
      "loss": 2.1485,
      "step": 9937
    },
    {
      "epoch": 0.26173294706347117,
      "grad_norm": 1.651205062866211,
      "learning_rate": 3.692257045035555e-05,
      "loss": 1.5987,
      "step": 9938
    },
    {
      "epoch": 0.26175928364498285,
      "grad_norm": 2.5741209983825684,
      "learning_rate": 3.6921253621279964e-05,
      "loss": 1.7725,
      "step": 9939
    },
    {
      "epoch": 0.2617856202264946,
      "grad_norm": 1.998192310333252,
      "learning_rate": 3.691993679220437e-05,
      "loss": 2.0674,
      "step": 9940
    },
    {
      "epoch": 0.26181195680800634,
      "grad_norm": 2.596903085708618,
      "learning_rate": 3.691861996312879e-05,
      "loss": 2.3742,
      "step": 9941
    },
    {
      "epoch": 0.26183829338951803,
      "grad_norm": 2.1205623149871826,
      "learning_rate": 3.69173031340532e-05,
      "loss": 0.4241,
      "step": 9942
    },
    {
      "epoch": 0.26186462997102977,
      "grad_norm": 1.6825841665267944,
      "learning_rate": 3.691598630497762e-05,
      "loss": 1.4807,
      "step": 9943
    },
    {
      "epoch": 0.26189096655254146,
      "grad_norm": 1.6058809757232666,
      "learning_rate": 3.691466947590203e-05,
      "loss": 1.7782,
      "step": 9944
    },
    {
      "epoch": 0.2619173031340532,
      "grad_norm": 3.9677579402923584,
      "learning_rate": 3.6913352646826444e-05,
      "loss": 1.2704,
      "step": 9945
    },
    {
      "epoch": 0.26194363971556495,
      "grad_norm": 2.0853161811828613,
      "learning_rate": 3.691203581775085e-05,
      "loss": 2.0454,
      "step": 9946
    },
    {
      "epoch": 0.26196997629707663,
      "grad_norm": 2.0083141326904297,
      "learning_rate": 3.691071898867527e-05,
      "loss": 0.4336,
      "step": 9947
    },
    {
      "epoch": 0.2619963128785884,
      "grad_norm": 2.3993773460388184,
      "learning_rate": 3.6909402159599684e-05,
      "loss": 1.3382,
      "step": 9948
    },
    {
      "epoch": 0.26202264946010007,
      "grad_norm": 5.320145606994629,
      "learning_rate": 3.69080853305241e-05,
      "loss": 1.0417,
      "step": 9949
    },
    {
      "epoch": 0.2620489860416118,
      "grad_norm": 3.865410327911377,
      "learning_rate": 3.6906768501448515e-05,
      "loss": 0.7176,
      "step": 9950
    },
    {
      "epoch": 0.2620753226231235,
      "grad_norm": 4.403519153594971,
      "learning_rate": 3.6905451672372924e-05,
      "loss": 1.0294,
      "step": 9951
    },
    {
      "epoch": 0.26210165920463524,
      "grad_norm": 1.9168848991394043,
      "learning_rate": 3.6904134843297347e-05,
      "loss": 1.6702,
      "step": 9952
    },
    {
      "epoch": 0.262127995786147,
      "grad_norm": 1.7266138792037964,
      "learning_rate": 3.6902818014221755e-05,
      "loss": 2.1554,
      "step": 9953
    },
    {
      "epoch": 0.26215433236765867,
      "grad_norm": 3.792888641357422,
      "learning_rate": 3.690150118514617e-05,
      "loss": 0.8645,
      "step": 9954
    },
    {
      "epoch": 0.2621806689491704,
      "grad_norm": 2.359764337539673,
      "learning_rate": 3.690018435607058e-05,
      "loss": 1.6779,
      "step": 9955
    },
    {
      "epoch": 0.2622070055306821,
      "grad_norm": 2.330033302307129,
      "learning_rate": 3.6898867526994995e-05,
      "loss": 0.4269,
      "step": 9956
    },
    {
      "epoch": 0.26223334211219385,
      "grad_norm": 2.988826036453247,
      "learning_rate": 3.689755069791941e-05,
      "loss": 1.5425,
      "step": 9957
    },
    {
      "epoch": 0.26225967869370553,
      "grad_norm": 2.079436779022217,
      "learning_rate": 3.6896233868843827e-05,
      "loss": 1.6766,
      "step": 9958
    },
    {
      "epoch": 0.2622860152752173,
      "grad_norm": 2.129084348678589,
      "learning_rate": 3.689491703976824e-05,
      "loss": 1.7665,
      "step": 9959
    },
    {
      "epoch": 0.262312351856729,
      "grad_norm": 2.360966920852661,
      "learning_rate": 3.689360021069265e-05,
      "loss": 1.7158,
      "step": 9960
    },
    {
      "epoch": 0.2623386884382407,
      "grad_norm": 1.8067656755447388,
      "learning_rate": 3.689228338161707e-05,
      "loss": 1.4482,
      "step": 9961
    },
    {
      "epoch": 0.26236502501975245,
      "grad_norm": 1.6358373165130615,
      "learning_rate": 3.689096655254148e-05,
      "loss": 2.3643,
      "step": 9962
    },
    {
      "epoch": 0.26239136160126414,
      "grad_norm": 5.185770034790039,
      "learning_rate": 3.68896497234659e-05,
      "loss": 1.3845,
      "step": 9963
    },
    {
      "epoch": 0.2624176981827759,
      "grad_norm": 2.6043448448181152,
      "learning_rate": 3.688833289439031e-05,
      "loss": 1.6341,
      "step": 9964
    },
    {
      "epoch": 0.26244403476428757,
      "grad_norm": 2.543100595474243,
      "learning_rate": 3.688701606531472e-05,
      "loss": 2.2213,
      "step": 9965
    },
    {
      "epoch": 0.2624703713457993,
      "grad_norm": 1.7823302745819092,
      "learning_rate": 3.688569923623914e-05,
      "loss": 0.9165,
      "step": 9966
    },
    {
      "epoch": 0.26249670792731106,
      "grad_norm": 1.6086454391479492,
      "learning_rate": 3.6884382407163553e-05,
      "loss": 1.8282,
      "step": 9967
    },
    {
      "epoch": 0.26252304450882274,
      "grad_norm": 1.7740914821624756,
      "learning_rate": 3.688306557808797e-05,
      "loss": 1.5867,
      "step": 9968
    },
    {
      "epoch": 0.2625493810903345,
      "grad_norm": 2.1103787422180176,
      "learning_rate": 3.688174874901238e-05,
      "loss": 1.982,
      "step": 9969
    },
    {
      "epoch": 0.2625757176718462,
      "grad_norm": 1.7448289394378662,
      "learning_rate": 3.6880431919936793e-05,
      "loss": 1.6368,
      "step": 9970
    },
    {
      "epoch": 0.2626020542533579,
      "grad_norm": 3.1486148834228516,
      "learning_rate": 3.687911509086121e-05,
      "loss": 1.4089,
      "step": 9971
    },
    {
      "epoch": 0.2626283908348696,
      "grad_norm": 2.2418668270111084,
      "learning_rate": 3.6877798261785625e-05,
      "loss": 1.7981,
      "step": 9972
    },
    {
      "epoch": 0.26265472741638135,
      "grad_norm": 1.9854317903518677,
      "learning_rate": 3.6876481432710033e-05,
      "loss": 1.9868,
      "step": 9973
    },
    {
      "epoch": 0.2626810639978931,
      "grad_norm": 2.909822463989258,
      "learning_rate": 3.687516460363445e-05,
      "loss": 1.4789,
      "step": 9974
    },
    {
      "epoch": 0.2627074005794048,
      "grad_norm": 1.796769142150879,
      "learning_rate": 3.6873847774558865e-05,
      "loss": 2.0325,
      "step": 9975
    },
    {
      "epoch": 0.2627337371609165,
      "grad_norm": 1.8284469842910767,
      "learning_rate": 3.687253094548328e-05,
      "loss": 1.7165,
      "step": 9976
    },
    {
      "epoch": 0.2627600737424282,
      "grad_norm": 1.6428061723709106,
      "learning_rate": 3.6871214116407696e-05,
      "loss": 1.6384,
      "step": 9977
    },
    {
      "epoch": 0.26278641032393996,
      "grad_norm": 1.7576966285705566,
      "learning_rate": 3.6869897287332105e-05,
      "loss": 0.9424,
      "step": 9978
    },
    {
      "epoch": 0.2628127469054517,
      "grad_norm": 2.577503204345703,
      "learning_rate": 3.686858045825652e-05,
      "loss": 2.1824,
      "step": 9979
    },
    {
      "epoch": 0.2628390834869634,
      "grad_norm": 1.9705294370651245,
      "learning_rate": 3.686726362918093e-05,
      "loss": 0.4674,
      "step": 9980
    },
    {
      "epoch": 0.26286542006847513,
      "grad_norm": 1.730053186416626,
      "learning_rate": 3.686594680010535e-05,
      "loss": 1.8857,
      "step": 9981
    },
    {
      "epoch": 0.2628917566499868,
      "grad_norm": 6.366098403930664,
      "learning_rate": 3.686462997102976e-05,
      "loss": 1.4864,
      "step": 9982
    },
    {
      "epoch": 0.26291809323149856,
      "grad_norm": 2.1583688259124756,
      "learning_rate": 3.6863313141954176e-05,
      "loss": 1.4697,
      "step": 9983
    },
    {
      "epoch": 0.26294442981301025,
      "grad_norm": 1.7725675106048584,
      "learning_rate": 3.686199631287859e-05,
      "loss": 2.431,
      "step": 9984
    },
    {
      "epoch": 0.262970766394522,
      "grad_norm": 5.0243072509765625,
      "learning_rate": 3.686067948380301e-05,
      "loss": 1.1403,
      "step": 9985
    },
    {
      "epoch": 0.26299710297603374,
      "grad_norm": 3.0046603679656982,
      "learning_rate": 3.685936265472742e-05,
      "loss": 0.8956,
      "step": 9986
    },
    {
      "epoch": 0.2630234395575454,
      "grad_norm": 3.7581191062927246,
      "learning_rate": 3.685804582565183e-05,
      "loss": 1.2484,
      "step": 9987
    },
    {
      "epoch": 0.26304977613905717,
      "grad_norm": 3.3439056873321533,
      "learning_rate": 3.685672899657625e-05,
      "loss": 1.5957,
      "step": 9988
    },
    {
      "epoch": 0.26307611272056886,
      "grad_norm": 5.87441349029541,
      "learning_rate": 3.6855412167500656e-05,
      "loss": 1.4431,
      "step": 9989
    },
    {
      "epoch": 0.2631024493020806,
      "grad_norm": 1.8784219026565552,
      "learning_rate": 3.685409533842508e-05,
      "loss": 1.949,
      "step": 9990
    },
    {
      "epoch": 0.2631287858835923,
      "grad_norm": 2.4301981925964355,
      "learning_rate": 3.685277850934949e-05,
      "loss": 2.1704,
      "step": 9991
    },
    {
      "epoch": 0.26315512246510403,
      "grad_norm": 7.405130386352539,
      "learning_rate": 3.68514616802739e-05,
      "loss": 1.0526,
      "step": 9992
    },
    {
      "epoch": 0.2631814590466158,
      "grad_norm": 3.237091064453125,
      "learning_rate": 3.685014485119832e-05,
      "loss": 1.5404,
      "step": 9993
    },
    {
      "epoch": 0.26320779562812746,
      "grad_norm": 3.4121577739715576,
      "learning_rate": 3.684882802212273e-05,
      "loss": 2.0441,
      "step": 9994
    },
    {
      "epoch": 0.2632341322096392,
      "grad_norm": 2.61460018157959,
      "learning_rate": 3.684751119304715e-05,
      "loss": 2.4158,
      "step": 9995
    },
    {
      "epoch": 0.2632604687911509,
      "grad_norm": 1.840182900428772,
      "learning_rate": 3.684619436397156e-05,
      "loss": 1.8731,
      "step": 9996
    },
    {
      "epoch": 0.26328680537266264,
      "grad_norm": 2.847264051437378,
      "learning_rate": 3.6844877534895974e-05,
      "loss": 1.2876,
      "step": 9997
    },
    {
      "epoch": 0.2633131419541743,
      "grad_norm": 3.600921869277954,
      "learning_rate": 3.684356070582038e-05,
      "loss": 2.1099,
      "step": 9998
    },
    {
      "epoch": 0.26333947853568607,
      "grad_norm": 1.7776530981063843,
      "learning_rate": 3.6842243876744805e-05,
      "loss": 1.6777,
      "step": 9999
    },
    {
      "epoch": 0.2633658151171978,
      "grad_norm": 2.220614194869995,
      "learning_rate": 3.6840927047669214e-05,
      "loss": 1.7253,
      "step": 10000
    },
    {
      "epoch": 0.2633921516987095,
      "grad_norm": 2.796560525894165,
      "learning_rate": 3.683961021859363e-05,
      "loss": 1.5488,
      "step": 10001
    },
    {
      "epoch": 0.26341848828022124,
      "grad_norm": 2.200914144515991,
      "learning_rate": 3.683829338951804e-05,
      "loss": 2.0682,
      "step": 10002
    },
    {
      "epoch": 0.26344482486173293,
      "grad_norm": 2.2607386112213135,
      "learning_rate": 3.6836976560442454e-05,
      "loss": 1.8996,
      "step": 10003
    },
    {
      "epoch": 0.2634711614432447,
      "grad_norm": 1.764689564704895,
      "learning_rate": 3.683565973136687e-05,
      "loss": 2.3482,
      "step": 10004
    },
    {
      "epoch": 0.26349749802475636,
      "grad_norm": 2.6437742710113525,
      "learning_rate": 3.6834342902291285e-05,
      "loss": 1.8611,
      "step": 10005
    },
    {
      "epoch": 0.2635238346062681,
      "grad_norm": 3.04620099067688,
      "learning_rate": 3.68330260732157e-05,
      "loss": 1.2221,
      "step": 10006
    },
    {
      "epoch": 0.26355017118777985,
      "grad_norm": 2.4913136959075928,
      "learning_rate": 3.683170924414011e-05,
      "loss": 1.737,
      "step": 10007
    },
    {
      "epoch": 0.26357650776929153,
      "grad_norm": 2.2860565185546875,
      "learning_rate": 3.6830392415064525e-05,
      "loss": 1.5518,
      "step": 10008
    },
    {
      "epoch": 0.2636028443508033,
      "grad_norm": 2.7356841564178467,
      "learning_rate": 3.682907558598894e-05,
      "loss": 2.9802,
      "step": 10009
    },
    {
      "epoch": 0.26362918093231497,
      "grad_norm": 2.5983242988586426,
      "learning_rate": 3.6827758756913356e-05,
      "loss": 2.0625,
      "step": 10010
    },
    {
      "epoch": 0.2636555175138267,
      "grad_norm": 2.2959084510803223,
      "learning_rate": 3.6826441927837765e-05,
      "loss": 2.1491,
      "step": 10011
    },
    {
      "epoch": 0.26368185409533845,
      "grad_norm": 1.8262606859207153,
      "learning_rate": 3.682512509876218e-05,
      "loss": 1.7723,
      "step": 10012
    },
    {
      "epoch": 0.26370819067685014,
      "grad_norm": 2.623422145843506,
      "learning_rate": 3.6823808269686596e-05,
      "loss": 1.3331,
      "step": 10013
    },
    {
      "epoch": 0.2637345272583619,
      "grad_norm": 1.9574443101882935,
      "learning_rate": 3.682249144061101e-05,
      "loss": 1.6323,
      "step": 10014
    },
    {
      "epoch": 0.26376086383987357,
      "grad_norm": 5.081628799438477,
      "learning_rate": 3.682117461153543e-05,
      "loss": 1.25,
      "step": 10015
    },
    {
      "epoch": 0.2637872004213853,
      "grad_norm": 4.234491348266602,
      "learning_rate": 3.6819857782459836e-05,
      "loss": 1.0458,
      "step": 10016
    },
    {
      "epoch": 0.263813537002897,
      "grad_norm": 1.6453907489776611,
      "learning_rate": 3.681854095338425e-05,
      "loss": 1.5962,
      "step": 10017
    },
    {
      "epoch": 0.26383987358440875,
      "grad_norm": 2.469261884689331,
      "learning_rate": 3.681722412430867e-05,
      "loss": 1.741,
      "step": 10018
    },
    {
      "epoch": 0.2638662101659205,
      "grad_norm": 4.05128812789917,
      "learning_rate": 3.681590729523308e-05,
      "loss": 0.7735,
      "step": 10019
    },
    {
      "epoch": 0.2638925467474322,
      "grad_norm": 2.9753005504608154,
      "learning_rate": 3.681459046615749e-05,
      "loss": 0.9617,
      "step": 10020
    },
    {
      "epoch": 0.2639188833289439,
      "grad_norm": 1.9553937911987305,
      "learning_rate": 3.681327363708191e-05,
      "loss": 1.8002,
      "step": 10021
    },
    {
      "epoch": 0.2639452199104556,
      "grad_norm": 1.9491946697235107,
      "learning_rate": 3.681195680800632e-05,
      "loss": 1.2068,
      "step": 10022
    },
    {
      "epoch": 0.26397155649196735,
      "grad_norm": 2.2591712474823,
      "learning_rate": 3.681063997893074e-05,
      "loss": 1.7033,
      "step": 10023
    },
    {
      "epoch": 0.26399789307347904,
      "grad_norm": 2.861011028289795,
      "learning_rate": 3.6809323149855154e-05,
      "loss": 1.6876,
      "step": 10024
    },
    {
      "epoch": 0.2640242296549908,
      "grad_norm": 1.7646710872650146,
      "learning_rate": 3.680800632077956e-05,
      "loss": 2.0335,
      "step": 10025
    },
    {
      "epoch": 0.2640505662365025,
      "grad_norm": 2.0951666831970215,
      "learning_rate": 3.680668949170398e-05,
      "loss": 2.422,
      "step": 10026
    },
    {
      "epoch": 0.2640769028180142,
      "grad_norm": 2.4043684005737305,
      "learning_rate": 3.680537266262839e-05,
      "loss": 1.8476,
      "step": 10027
    },
    {
      "epoch": 0.26410323939952596,
      "grad_norm": 2.605814218521118,
      "learning_rate": 3.680405583355281e-05,
      "loss": 1.9494,
      "step": 10028
    },
    {
      "epoch": 0.26412957598103765,
      "grad_norm": 2.503455400466919,
      "learning_rate": 3.680273900447722e-05,
      "loss": 0.2176,
      "step": 10029
    },
    {
      "epoch": 0.2641559125625494,
      "grad_norm": 1.784052848815918,
      "learning_rate": 3.6801422175401634e-05,
      "loss": 1.5564,
      "step": 10030
    },
    {
      "epoch": 0.2641822491440611,
      "grad_norm": 1.4798592329025269,
      "learning_rate": 3.680010534632605e-05,
      "loss": 1.5415,
      "step": 10031
    },
    {
      "epoch": 0.2642085857255728,
      "grad_norm": 2.9827847480773926,
      "learning_rate": 3.6798788517250466e-05,
      "loss": 1.5892,
      "step": 10032
    },
    {
      "epoch": 0.26423492230708456,
      "grad_norm": 1.920401930809021,
      "learning_rate": 3.679747168817488e-05,
      "loss": 2.3112,
      "step": 10033
    },
    {
      "epoch": 0.26426125888859625,
      "grad_norm": 3.290771961212158,
      "learning_rate": 3.679615485909929e-05,
      "loss": 1.7108,
      "step": 10034
    },
    {
      "epoch": 0.264287595470108,
      "grad_norm": 2.365751266479492,
      "learning_rate": 3.6794838030023706e-05,
      "loss": 1.9305,
      "step": 10035
    },
    {
      "epoch": 0.2643139320516197,
      "grad_norm": 2.2461624145507812,
      "learning_rate": 3.6793521200948114e-05,
      "loss": 2.036,
      "step": 10036
    },
    {
      "epoch": 0.2643402686331314,
      "grad_norm": 2.9866960048675537,
      "learning_rate": 3.679220437187254e-05,
      "loss": 1.0285,
      "step": 10037
    },
    {
      "epoch": 0.2643666052146431,
      "grad_norm": 3.6891326904296875,
      "learning_rate": 3.6790887542796946e-05,
      "loss": 2.0806,
      "step": 10038
    },
    {
      "epoch": 0.26439294179615486,
      "grad_norm": 2.289691925048828,
      "learning_rate": 3.678957071372136e-05,
      "loss": 0.6964,
      "step": 10039
    },
    {
      "epoch": 0.2644192783776666,
      "grad_norm": 5.166062831878662,
      "learning_rate": 3.678825388464578e-05,
      "loss": 0.8084,
      "step": 10040
    },
    {
      "epoch": 0.2644456149591783,
      "grad_norm": 2.6089706420898438,
      "learning_rate": 3.6786937055570186e-05,
      "loss": 2.2339,
      "step": 10041
    },
    {
      "epoch": 0.26447195154069003,
      "grad_norm": 2.1294283866882324,
      "learning_rate": 3.678562022649461e-05,
      "loss": 2.0517,
      "step": 10042
    },
    {
      "epoch": 0.2644982881222017,
      "grad_norm": 1.9592230319976807,
      "learning_rate": 3.678430339741902e-05,
      "loss": 1.9269,
      "step": 10043
    },
    {
      "epoch": 0.26452462470371346,
      "grad_norm": 1.92788565158844,
      "learning_rate": 3.678298656834343e-05,
      "loss": 1.7954,
      "step": 10044
    },
    {
      "epoch": 0.26455096128522515,
      "grad_norm": 3.0028562545776367,
      "learning_rate": 3.678166973926784e-05,
      "loss": 1.7689,
      "step": 10045
    },
    {
      "epoch": 0.2645772978667369,
      "grad_norm": 4.070221424102783,
      "learning_rate": 3.678035291019226e-05,
      "loss": 1.4069,
      "step": 10046
    },
    {
      "epoch": 0.26460363444824864,
      "grad_norm": 1.8371425867080688,
      "learning_rate": 3.677903608111667e-05,
      "loss": 2.8702,
      "step": 10047
    },
    {
      "epoch": 0.2646299710297603,
      "grad_norm": 1.7458940744400024,
      "learning_rate": 3.677771925204109e-05,
      "loss": 2.056,
      "step": 10048
    },
    {
      "epoch": 0.26465630761127207,
      "grad_norm": 9.552285194396973,
      "learning_rate": 3.67764024229655e-05,
      "loss": 2.6065,
      "step": 10049
    },
    {
      "epoch": 0.26468264419278376,
      "grad_norm": 1.8528327941894531,
      "learning_rate": 3.677508559388991e-05,
      "loss": 1.7531,
      "step": 10050
    },
    {
      "epoch": 0.2647089807742955,
      "grad_norm": 2.379115581512451,
      "learning_rate": 3.677376876481433e-05,
      "loss": 2.5959,
      "step": 10051
    },
    {
      "epoch": 0.26473531735580724,
      "grad_norm": 2.3762495517730713,
      "learning_rate": 3.6772451935738744e-05,
      "loss": 1.7078,
      "step": 10052
    },
    {
      "epoch": 0.26476165393731893,
      "grad_norm": 2.559075355529785,
      "learning_rate": 3.677113510666316e-05,
      "loss": 2.0787,
      "step": 10053
    },
    {
      "epoch": 0.2647879905188307,
      "grad_norm": 1.5611319541931152,
      "learning_rate": 3.676981827758757e-05,
      "loss": 1.6679,
      "step": 10054
    },
    {
      "epoch": 0.26481432710034236,
      "grad_norm": 2.640212059020996,
      "learning_rate": 3.6768501448511984e-05,
      "loss": 1.3131,
      "step": 10055
    },
    {
      "epoch": 0.2648406636818541,
      "grad_norm": 2.510906219482422,
      "learning_rate": 3.67671846194364e-05,
      "loss": 0.6519,
      "step": 10056
    },
    {
      "epoch": 0.2648670002633658,
      "grad_norm": 2.74783992767334,
      "learning_rate": 3.6765867790360815e-05,
      "loss": 2.6449,
      "step": 10057
    },
    {
      "epoch": 0.26489333684487754,
      "grad_norm": 1.6115412712097168,
      "learning_rate": 3.6764550961285224e-05,
      "loss": 1.4667,
      "step": 10058
    },
    {
      "epoch": 0.2649196734263893,
      "grad_norm": 2.5120208263397217,
      "learning_rate": 3.676323413220964e-05,
      "loss": 0.5017,
      "step": 10059
    },
    {
      "epoch": 0.26494601000790097,
      "grad_norm": 4.157475471496582,
      "learning_rate": 3.6761917303134055e-05,
      "loss": 2.4723,
      "step": 10060
    },
    {
      "epoch": 0.2649723465894127,
      "grad_norm": 4.212089538574219,
      "learning_rate": 3.676060047405847e-05,
      "loss": 1.3918,
      "step": 10061
    },
    {
      "epoch": 0.2649986831709244,
      "grad_norm": 1.997915506362915,
      "learning_rate": 3.6759283644982886e-05,
      "loss": 1.2492,
      "step": 10062
    },
    {
      "epoch": 0.26502501975243614,
      "grad_norm": 2.1331331729888916,
      "learning_rate": 3.6757966815907295e-05,
      "loss": 2.1265,
      "step": 10063
    },
    {
      "epoch": 0.26505135633394783,
      "grad_norm": 2.222640037536621,
      "learning_rate": 3.675664998683171e-05,
      "loss": 1.0801,
      "step": 10064
    },
    {
      "epoch": 0.2650776929154596,
      "grad_norm": 2.3524866104125977,
      "learning_rate": 3.6755333157756126e-05,
      "loss": 2.6966,
      "step": 10065
    },
    {
      "epoch": 0.2651040294969713,
      "grad_norm": 2.105419158935547,
      "learning_rate": 3.675401632868054e-05,
      "loss": 1.1735,
      "step": 10066
    },
    {
      "epoch": 0.265130366078483,
      "grad_norm": 3.2529942989349365,
      "learning_rate": 3.675269949960495e-05,
      "loss": 1.5403,
      "step": 10067
    },
    {
      "epoch": 0.26515670265999475,
      "grad_norm": 1.942391037940979,
      "learning_rate": 3.6751382670529366e-05,
      "loss": 1.7382,
      "step": 10068
    },
    {
      "epoch": 0.26518303924150644,
      "grad_norm": 3.163064479827881,
      "learning_rate": 3.675006584145378e-05,
      "loss": 1.8874,
      "step": 10069
    },
    {
      "epoch": 0.2652093758230182,
      "grad_norm": 2.4406042098999023,
      "learning_rate": 3.67487490123782e-05,
      "loss": 1.7741,
      "step": 10070
    },
    {
      "epoch": 0.26523571240452987,
      "grad_norm": 1.6882086992263794,
      "learning_rate": 3.674743218330261e-05,
      "loss": 1.0368,
      "step": 10071
    },
    {
      "epoch": 0.2652620489860416,
      "grad_norm": 1.9586982727050781,
      "learning_rate": 3.674611535422702e-05,
      "loss": 2.183,
      "step": 10072
    },
    {
      "epoch": 0.26528838556755335,
      "grad_norm": 1.9157731533050537,
      "learning_rate": 3.674479852515144e-05,
      "loss": 1.4521,
      "step": 10073
    },
    {
      "epoch": 0.26531472214906504,
      "grad_norm": 1.4000648260116577,
      "learning_rate": 3.6743481696075846e-05,
      "loss": 1.6729,
      "step": 10074
    },
    {
      "epoch": 0.2653410587305768,
      "grad_norm": 2.051919937133789,
      "learning_rate": 3.674216486700027e-05,
      "loss": 2.5286,
      "step": 10075
    },
    {
      "epoch": 0.26536739531208847,
      "grad_norm": 2.4728801250457764,
      "learning_rate": 3.674084803792468e-05,
      "loss": 1.3984,
      "step": 10076
    },
    {
      "epoch": 0.2653937318936002,
      "grad_norm": 1.9722304344177246,
      "learning_rate": 3.673953120884909e-05,
      "loss": 1.5037,
      "step": 10077
    },
    {
      "epoch": 0.2654200684751119,
      "grad_norm": 2.188192367553711,
      "learning_rate": 3.673821437977351e-05,
      "loss": 2.309,
      "step": 10078
    },
    {
      "epoch": 0.26544640505662365,
      "grad_norm": 5.090904712677002,
      "learning_rate": 3.673689755069792e-05,
      "loss": 0.7312,
      "step": 10079
    },
    {
      "epoch": 0.2654727416381354,
      "grad_norm": 3.69519305229187,
      "learning_rate": 3.673558072162234e-05,
      "loss": 0.7831,
      "step": 10080
    },
    {
      "epoch": 0.2654990782196471,
      "grad_norm": 2.031735897064209,
      "learning_rate": 3.673426389254675e-05,
      "loss": 1.2649,
      "step": 10081
    },
    {
      "epoch": 0.2655254148011588,
      "grad_norm": 2.5361666679382324,
      "learning_rate": 3.6732947063471164e-05,
      "loss": 1.8613,
      "step": 10082
    },
    {
      "epoch": 0.2655517513826705,
      "grad_norm": 2.226508378982544,
      "learning_rate": 3.673163023439557e-05,
      "loss": 0.9733,
      "step": 10083
    },
    {
      "epoch": 0.26557808796418225,
      "grad_norm": 3.9375691413879395,
      "learning_rate": 3.6730313405319995e-05,
      "loss": 0.7214,
      "step": 10084
    },
    {
      "epoch": 0.265604424545694,
      "grad_norm": 1.6749688386917114,
      "learning_rate": 3.6728996576244404e-05,
      "loss": 1.9015,
      "step": 10085
    },
    {
      "epoch": 0.2656307611272057,
      "grad_norm": 4.377026557922363,
      "learning_rate": 3.672767974716882e-05,
      "loss": 2.2142,
      "step": 10086
    },
    {
      "epoch": 0.2656570977087174,
      "grad_norm": 3.526095151901245,
      "learning_rate": 3.6726362918093235e-05,
      "loss": 2.3948,
      "step": 10087
    },
    {
      "epoch": 0.2656834342902291,
      "grad_norm": 4.275936603546143,
      "learning_rate": 3.6725046089017644e-05,
      "loss": 1.2027,
      "step": 10088
    },
    {
      "epoch": 0.26570977087174086,
      "grad_norm": 2.2248125076293945,
      "learning_rate": 3.6723729259942067e-05,
      "loss": 1.8308,
      "step": 10089
    },
    {
      "epoch": 0.26573610745325255,
      "grad_norm": 2.8518710136413574,
      "learning_rate": 3.6722412430866475e-05,
      "loss": 1.6113,
      "step": 10090
    },
    {
      "epoch": 0.2657624440347643,
      "grad_norm": 2.782762050628662,
      "learning_rate": 3.672109560179089e-05,
      "loss": 1.4446,
      "step": 10091
    },
    {
      "epoch": 0.26578878061627603,
      "grad_norm": 2.3260200023651123,
      "learning_rate": 3.67197787727153e-05,
      "loss": 1.5937,
      "step": 10092
    },
    {
      "epoch": 0.2658151171977877,
      "grad_norm": 2.2114033699035645,
      "learning_rate": 3.6718461943639715e-05,
      "loss": 2.2396,
      "step": 10093
    },
    {
      "epoch": 0.26584145377929946,
      "grad_norm": 1.9129087924957275,
      "learning_rate": 3.671714511456413e-05,
      "loss": 1.4507,
      "step": 10094
    },
    {
      "epoch": 0.26586779036081115,
      "grad_norm": 1.6789244413375854,
      "learning_rate": 3.671582828548855e-05,
      "loss": 1.6753,
      "step": 10095
    },
    {
      "epoch": 0.2658941269423229,
      "grad_norm": 1.5938291549682617,
      "learning_rate": 3.671451145641296e-05,
      "loss": 1.8683,
      "step": 10096
    },
    {
      "epoch": 0.2659204635238346,
      "grad_norm": 4.494757652282715,
      "learning_rate": 3.671319462733737e-05,
      "loss": 1.4146,
      "step": 10097
    },
    {
      "epoch": 0.2659468001053463,
      "grad_norm": 1.9412486553192139,
      "learning_rate": 3.6711877798261793e-05,
      "loss": 1.3101,
      "step": 10098
    },
    {
      "epoch": 0.26597313668685807,
      "grad_norm": 8.31187915802002,
      "learning_rate": 3.67105609691862e-05,
      "loss": 2.1754,
      "step": 10099
    },
    {
      "epoch": 0.26599947326836976,
      "grad_norm": 2.717020273208618,
      "learning_rate": 3.670924414011062e-05,
      "loss": 1.5488,
      "step": 10100
    },
    {
      "epoch": 0.2660258098498815,
      "grad_norm": 1.9534482955932617,
      "learning_rate": 3.670792731103503e-05,
      "loss": 1.6723,
      "step": 10101
    },
    {
      "epoch": 0.2660521464313932,
      "grad_norm": 4.684081077575684,
      "learning_rate": 3.670661048195944e-05,
      "loss": 1.4647,
      "step": 10102
    },
    {
      "epoch": 0.26607848301290493,
      "grad_norm": 2.3786094188690186,
      "learning_rate": 3.670529365288386e-05,
      "loss": 1.2828,
      "step": 10103
    },
    {
      "epoch": 0.2661048195944166,
      "grad_norm": 3.372189521789551,
      "learning_rate": 3.6703976823808273e-05,
      "loss": 0.991,
      "step": 10104
    },
    {
      "epoch": 0.26613115617592836,
      "grad_norm": 2.048107862472534,
      "learning_rate": 3.670265999473268e-05,
      "loss": 2.0632,
      "step": 10105
    },
    {
      "epoch": 0.2661574927574401,
      "grad_norm": 1.9173202514648438,
      "learning_rate": 3.67013431656571e-05,
      "loss": 1.7183,
      "step": 10106
    },
    {
      "epoch": 0.2661838293389518,
      "grad_norm": 2.5450634956359863,
      "learning_rate": 3.6700026336581513e-05,
      "loss": 1.9082,
      "step": 10107
    },
    {
      "epoch": 0.26621016592046354,
      "grad_norm": 2.0234475135803223,
      "learning_rate": 3.669870950750593e-05,
      "loss": 1.7987,
      "step": 10108
    },
    {
      "epoch": 0.2662365025019752,
      "grad_norm": 4.430846214294434,
      "learning_rate": 3.6697392678430345e-05,
      "loss": 1.1304,
      "step": 10109
    },
    {
      "epoch": 0.26626283908348697,
      "grad_norm": 2.909736156463623,
      "learning_rate": 3.6696075849354754e-05,
      "loss": 1.7499,
      "step": 10110
    },
    {
      "epoch": 0.26628917566499866,
      "grad_norm": 1.8816019296646118,
      "learning_rate": 3.669475902027917e-05,
      "loss": 0.3428,
      "step": 10111
    },
    {
      "epoch": 0.2663155122465104,
      "grad_norm": 4.846202373504639,
      "learning_rate": 3.669344219120358e-05,
      "loss": 0.7626,
      "step": 10112
    },
    {
      "epoch": 0.26634184882802214,
      "grad_norm": 1.4240412712097168,
      "learning_rate": 3.6692125362128e-05,
      "loss": 1.7212,
      "step": 10113
    },
    {
      "epoch": 0.26636818540953383,
      "grad_norm": 1.6693060398101807,
      "learning_rate": 3.669080853305241e-05,
      "loss": 1.2432,
      "step": 10114
    },
    {
      "epoch": 0.2663945219910456,
      "grad_norm": 2.1589486598968506,
      "learning_rate": 3.6689491703976825e-05,
      "loss": 0.5404,
      "step": 10115
    },
    {
      "epoch": 0.26642085857255726,
      "grad_norm": 2.379560947418213,
      "learning_rate": 3.668817487490124e-05,
      "loss": 2.0994,
      "step": 10116
    },
    {
      "epoch": 0.266447195154069,
      "grad_norm": 2.8508524894714355,
      "learning_rate": 3.6686858045825656e-05,
      "loss": 1.7082,
      "step": 10117
    },
    {
      "epoch": 0.26647353173558075,
      "grad_norm": 2.458432912826538,
      "learning_rate": 3.668554121675007e-05,
      "loss": 2.1248,
      "step": 10118
    },
    {
      "epoch": 0.26649986831709244,
      "grad_norm": 3.4483277797698975,
      "learning_rate": 3.668422438767448e-05,
      "loss": 1.3698,
      "step": 10119
    },
    {
      "epoch": 0.2665262048986042,
      "grad_norm": 1.8100333213806152,
      "learning_rate": 3.6682907558598896e-05,
      "loss": 2.3127,
      "step": 10120
    },
    {
      "epoch": 0.26655254148011587,
      "grad_norm": 2.026071548461914,
      "learning_rate": 3.6681590729523305e-05,
      "loss": 1.992,
      "step": 10121
    },
    {
      "epoch": 0.2665788780616276,
      "grad_norm": 2.3289945125579834,
      "learning_rate": 3.668027390044773e-05,
      "loss": 1.6703,
      "step": 10122
    },
    {
      "epoch": 0.2666052146431393,
      "grad_norm": 1.7287365198135376,
      "learning_rate": 3.6678957071372136e-05,
      "loss": 2.3727,
      "step": 10123
    },
    {
      "epoch": 0.26663155122465104,
      "grad_norm": 1.7984932661056519,
      "learning_rate": 3.667764024229655e-05,
      "loss": 1.7627,
      "step": 10124
    },
    {
      "epoch": 0.2666578878061628,
      "grad_norm": 1.8620436191558838,
      "learning_rate": 3.667632341322097e-05,
      "loss": 2.183,
      "step": 10125
    },
    {
      "epoch": 0.2666842243876745,
      "grad_norm": 1.9804375171661377,
      "learning_rate": 3.6675006584145376e-05,
      "loss": 1.1698,
      "step": 10126
    },
    {
      "epoch": 0.2667105609691862,
      "grad_norm": 3.017799139022827,
      "learning_rate": 3.66736897550698e-05,
      "loss": 0.9595,
      "step": 10127
    },
    {
      "epoch": 0.2667368975506979,
      "grad_norm": 2.9736342430114746,
      "learning_rate": 3.667237292599421e-05,
      "loss": 1.3165,
      "step": 10128
    },
    {
      "epoch": 0.26676323413220965,
      "grad_norm": 1.9643712043762207,
      "learning_rate": 3.667105609691862e-05,
      "loss": 0.4521,
      "step": 10129
    },
    {
      "epoch": 0.26678957071372134,
      "grad_norm": 6.517406463623047,
      "learning_rate": 3.666973926784303e-05,
      "loss": 1.2143,
      "step": 10130
    },
    {
      "epoch": 0.2668159072952331,
      "grad_norm": 1.7456799745559692,
      "learning_rate": 3.6668422438767454e-05,
      "loss": 2.1619,
      "step": 10131
    },
    {
      "epoch": 0.2668422438767448,
      "grad_norm": 2.0272860527038574,
      "learning_rate": 3.666710560969186e-05,
      "loss": 1.5795,
      "step": 10132
    },
    {
      "epoch": 0.2668685804582565,
      "grad_norm": 2.6654109954833984,
      "learning_rate": 3.666578878061628e-05,
      "loss": 1.6556,
      "step": 10133
    },
    {
      "epoch": 0.26689491703976825,
      "grad_norm": 3.296703338623047,
      "learning_rate": 3.6664471951540694e-05,
      "loss": 0.6678,
      "step": 10134
    },
    {
      "epoch": 0.26692125362127994,
      "grad_norm": 1.7900242805480957,
      "learning_rate": 3.66631551224651e-05,
      "loss": 1.8282,
      "step": 10135
    },
    {
      "epoch": 0.2669475902027917,
      "grad_norm": 1.3976291418075562,
      "learning_rate": 3.6661838293389525e-05,
      "loss": 1.3654,
      "step": 10136
    },
    {
      "epoch": 0.2669739267843034,
      "grad_norm": 2.2098259925842285,
      "learning_rate": 3.6660521464313934e-05,
      "loss": 0.9748,
      "step": 10137
    },
    {
      "epoch": 0.2670002633658151,
      "grad_norm": 3.5597519874572754,
      "learning_rate": 3.665920463523835e-05,
      "loss": 0.7587,
      "step": 10138
    },
    {
      "epoch": 0.26702659994732686,
      "grad_norm": 1.924920916557312,
      "learning_rate": 3.665788780616276e-05,
      "loss": 1.4954,
      "step": 10139
    },
    {
      "epoch": 0.26705293652883855,
      "grad_norm": 2.8917338848114014,
      "learning_rate": 3.6656570977087174e-05,
      "loss": 1.2898,
      "step": 10140
    },
    {
      "epoch": 0.2670792731103503,
      "grad_norm": 2.3523623943328857,
      "learning_rate": 3.665525414801159e-05,
      "loss": 1.8597,
      "step": 10141
    },
    {
      "epoch": 0.267105609691862,
      "grad_norm": 1.6046857833862305,
      "learning_rate": 3.6653937318936005e-05,
      "loss": 1.63,
      "step": 10142
    },
    {
      "epoch": 0.2671319462733737,
      "grad_norm": 3.940185546875,
      "learning_rate": 3.665262048986042e-05,
      "loss": 1.4117,
      "step": 10143
    },
    {
      "epoch": 0.2671582828548854,
      "grad_norm": 2.744216203689575,
      "learning_rate": 3.665130366078483e-05,
      "loss": 1.0407,
      "step": 10144
    },
    {
      "epoch": 0.26718461943639715,
      "grad_norm": 1.977405309677124,
      "learning_rate": 3.6649986831709245e-05,
      "loss": 1.4877,
      "step": 10145
    },
    {
      "epoch": 0.2672109560179089,
      "grad_norm": 1.750759243965149,
      "learning_rate": 3.664867000263366e-05,
      "loss": 1.9072,
      "step": 10146
    },
    {
      "epoch": 0.2672372925994206,
      "grad_norm": 1.7603579759597778,
      "learning_rate": 3.6647353173558076e-05,
      "loss": 1.9183,
      "step": 10147
    },
    {
      "epoch": 0.2672636291809323,
      "grad_norm": 1.503250241279602,
      "learning_rate": 3.6646036344482485e-05,
      "loss": 1.9238,
      "step": 10148
    },
    {
      "epoch": 0.267289965762444,
      "grad_norm": 3.797058343887329,
      "learning_rate": 3.66447195154069e-05,
      "loss": 0.977,
      "step": 10149
    },
    {
      "epoch": 0.26731630234395576,
      "grad_norm": 2.008352756500244,
      "learning_rate": 3.6643402686331316e-05,
      "loss": 1.9535,
      "step": 10150
    },
    {
      "epoch": 0.2673426389254675,
      "grad_norm": 2.1281659603118896,
      "learning_rate": 3.664208585725573e-05,
      "loss": 2.8871,
      "step": 10151
    },
    {
      "epoch": 0.2673689755069792,
      "grad_norm": 2.482161521911621,
      "learning_rate": 3.664076902818014e-05,
      "loss": 1.6571,
      "step": 10152
    },
    {
      "epoch": 0.26739531208849093,
      "grad_norm": 2.4182276725769043,
      "learning_rate": 3.6639452199104556e-05,
      "loss": 0.7212,
      "step": 10153
    },
    {
      "epoch": 0.2674216486700026,
      "grad_norm": 2.4475228786468506,
      "learning_rate": 3.663813537002897e-05,
      "loss": 2.244,
      "step": 10154
    },
    {
      "epoch": 0.26744798525151436,
      "grad_norm": 2.8437564373016357,
      "learning_rate": 3.663681854095339e-05,
      "loss": 2.1432,
      "step": 10155
    },
    {
      "epoch": 0.26747432183302605,
      "grad_norm": 1.8764352798461914,
      "learning_rate": 3.66355017118778e-05,
      "loss": 2.2595,
      "step": 10156
    },
    {
      "epoch": 0.2675006584145378,
      "grad_norm": 1.4289652109146118,
      "learning_rate": 3.663418488280221e-05,
      "loss": 0.1806,
      "step": 10157
    },
    {
      "epoch": 0.26752699499604954,
      "grad_norm": 2.258349657058716,
      "learning_rate": 3.663286805372663e-05,
      "loss": 1.8163,
      "step": 10158
    },
    {
      "epoch": 0.2675533315775612,
      "grad_norm": 1.6824095249176025,
      "learning_rate": 3.6631551224651036e-05,
      "loss": 1.5649,
      "step": 10159
    },
    {
      "epoch": 0.26757966815907297,
      "grad_norm": 2.620978355407715,
      "learning_rate": 3.663023439557546e-05,
      "loss": 2.3284,
      "step": 10160
    },
    {
      "epoch": 0.26760600474058466,
      "grad_norm": 1.7957783937454224,
      "learning_rate": 3.662891756649987e-05,
      "loss": 2.1384,
      "step": 10161
    },
    {
      "epoch": 0.2676323413220964,
      "grad_norm": 2.219515800476074,
      "learning_rate": 3.662760073742428e-05,
      "loss": 1.3335,
      "step": 10162
    },
    {
      "epoch": 0.2676586779036081,
      "grad_norm": 1.5960590839385986,
      "learning_rate": 3.66262839083487e-05,
      "loss": 1.7403,
      "step": 10163
    },
    {
      "epoch": 0.26768501448511983,
      "grad_norm": 2.2094027996063232,
      "learning_rate": 3.6624967079273114e-05,
      "loss": 1.9251,
      "step": 10164
    },
    {
      "epoch": 0.2677113510666316,
      "grad_norm": 1.57603120803833,
      "learning_rate": 3.662365025019753e-05,
      "loss": 2.1808,
      "step": 10165
    },
    {
      "epoch": 0.26773768764814326,
      "grad_norm": 2.3066632747650146,
      "learning_rate": 3.662233342112194e-05,
      "loss": 1.9552,
      "step": 10166
    },
    {
      "epoch": 0.267764024229655,
      "grad_norm": 1.8146836757659912,
      "learning_rate": 3.6621016592046354e-05,
      "loss": 2.3886,
      "step": 10167
    },
    {
      "epoch": 0.2677903608111667,
      "grad_norm": 2.3190245628356934,
      "learning_rate": 3.661969976297076e-05,
      "loss": 2.0307,
      "step": 10168
    },
    {
      "epoch": 0.26781669739267844,
      "grad_norm": 2.938628673553467,
      "learning_rate": 3.6618382933895186e-05,
      "loss": 1.5953,
      "step": 10169
    },
    {
      "epoch": 0.2678430339741901,
      "grad_norm": 2.054081439971924,
      "learning_rate": 3.6617066104819594e-05,
      "loss": 1.9068,
      "step": 10170
    },
    {
      "epoch": 0.26786937055570187,
      "grad_norm": 1.9563339948654175,
      "learning_rate": 3.661574927574401e-05,
      "loss": 1.6163,
      "step": 10171
    },
    {
      "epoch": 0.2678957071372136,
      "grad_norm": 3.8861870765686035,
      "learning_rate": 3.6614432446668426e-05,
      "loss": 0.9697,
      "step": 10172
    },
    {
      "epoch": 0.2679220437187253,
      "grad_norm": 2.57987642288208,
      "learning_rate": 3.6613115617592835e-05,
      "loss": 2.7066,
      "step": 10173
    },
    {
      "epoch": 0.26794838030023704,
      "grad_norm": 2.2081856727600098,
      "learning_rate": 3.661179878851726e-05,
      "loss": 1.7714,
      "step": 10174
    },
    {
      "epoch": 0.26797471688174873,
      "grad_norm": 2.464604616165161,
      "learning_rate": 3.6610481959441666e-05,
      "loss": 0.6583,
      "step": 10175
    },
    {
      "epoch": 0.2680010534632605,
      "grad_norm": 1.5755434036254883,
      "learning_rate": 3.660916513036608e-05,
      "loss": 0.5501,
      "step": 10176
    },
    {
      "epoch": 0.26802739004477216,
      "grad_norm": 1.4400713443756104,
      "learning_rate": 3.660784830129049e-05,
      "loss": 1.4302,
      "step": 10177
    },
    {
      "epoch": 0.2680537266262839,
      "grad_norm": 2.777074098587036,
      "learning_rate": 3.6606531472214906e-05,
      "loss": 1.2679,
      "step": 10178
    },
    {
      "epoch": 0.26808006320779565,
      "grad_norm": 2.064556837081909,
      "learning_rate": 3.660521464313932e-05,
      "loss": 2.2791,
      "step": 10179
    },
    {
      "epoch": 0.26810639978930734,
      "grad_norm": 6.590086460113525,
      "learning_rate": 3.660389781406374e-05,
      "loss": 1.3977,
      "step": 10180
    },
    {
      "epoch": 0.2681327363708191,
      "grad_norm": 2.8759360313415527,
      "learning_rate": 3.660258098498815e-05,
      "loss": 1.6432,
      "step": 10181
    },
    {
      "epoch": 0.26815907295233077,
      "grad_norm": 2.0994532108306885,
      "learning_rate": 3.660126415591256e-05,
      "loss": 1.9579,
      "step": 10182
    },
    {
      "epoch": 0.2681854095338425,
      "grad_norm": 2.383129596710205,
      "learning_rate": 3.6599947326836984e-05,
      "loss": 1.7171,
      "step": 10183
    },
    {
      "epoch": 0.2682117461153542,
      "grad_norm": 3.364213466644287,
      "learning_rate": 3.659863049776139e-05,
      "loss": 0.6375,
      "step": 10184
    },
    {
      "epoch": 0.26823808269686594,
      "grad_norm": 4.401752948760986,
      "learning_rate": 3.659731366868581e-05,
      "loss": 2.4599,
      "step": 10185
    },
    {
      "epoch": 0.2682644192783777,
      "grad_norm": 2.8946151733398438,
      "learning_rate": 3.659599683961022e-05,
      "loss": 2.0729,
      "step": 10186
    },
    {
      "epoch": 0.2682907558598894,
      "grad_norm": 1.7243560552597046,
      "learning_rate": 3.659468001053463e-05,
      "loss": 1.4998,
      "step": 10187
    },
    {
      "epoch": 0.2683170924414011,
      "grad_norm": 1.4112474918365479,
      "learning_rate": 3.659336318145905e-05,
      "loss": 1.7683,
      "step": 10188
    },
    {
      "epoch": 0.2683434290229128,
      "grad_norm": 1.7918068170547485,
      "learning_rate": 3.6592046352383464e-05,
      "loss": 2.0734,
      "step": 10189
    },
    {
      "epoch": 0.26836976560442455,
      "grad_norm": 1.8346103429794312,
      "learning_rate": 3.659072952330788e-05,
      "loss": 1.7252,
      "step": 10190
    },
    {
      "epoch": 0.2683961021859363,
      "grad_norm": 1.9054176807403564,
      "learning_rate": 3.658941269423229e-05,
      "loss": 2.2812,
      "step": 10191
    },
    {
      "epoch": 0.268422438767448,
      "grad_norm": 3.0302562713623047,
      "learning_rate": 3.6588095865156704e-05,
      "loss": 1.2557,
      "step": 10192
    },
    {
      "epoch": 0.2684487753489597,
      "grad_norm": 1.806452989578247,
      "learning_rate": 3.658677903608112e-05,
      "loss": 1.9694,
      "step": 10193
    },
    {
      "epoch": 0.2684751119304714,
      "grad_norm": 1.979327917098999,
      "learning_rate": 3.6585462207005535e-05,
      "loss": 1.3223,
      "step": 10194
    },
    {
      "epoch": 0.26850144851198315,
      "grad_norm": 2.0761165618896484,
      "learning_rate": 3.6584145377929944e-05,
      "loss": 2.0764,
      "step": 10195
    },
    {
      "epoch": 0.26852778509349484,
      "grad_norm": 3.167067527770996,
      "learning_rate": 3.658282854885436e-05,
      "loss": 2.0625,
      "step": 10196
    },
    {
      "epoch": 0.2685541216750066,
      "grad_norm": 1.328862190246582,
      "learning_rate": 3.6581511719778775e-05,
      "loss": 0.3699,
      "step": 10197
    },
    {
      "epoch": 0.26858045825651833,
      "grad_norm": 2.5438637733459473,
      "learning_rate": 3.658019489070319e-05,
      "loss": 1.339,
      "step": 10198
    },
    {
      "epoch": 0.26860679483803,
      "grad_norm": 2.776309013366699,
      "learning_rate": 3.65788780616276e-05,
      "loss": 1.8084,
      "step": 10199
    },
    {
      "epoch": 0.26863313141954176,
      "grad_norm": 1.5442655086517334,
      "learning_rate": 3.6577561232552015e-05,
      "loss": 1.6825,
      "step": 10200
    },
    {
      "epoch": 0.26865946800105345,
      "grad_norm": 1.637276291847229,
      "learning_rate": 3.657624440347643e-05,
      "loss": 1.7556,
      "step": 10201
    },
    {
      "epoch": 0.2686858045825652,
      "grad_norm": 1.8612775802612305,
      "learning_rate": 3.6574927574400846e-05,
      "loss": 1.9085,
      "step": 10202
    },
    {
      "epoch": 0.2687121411640769,
      "grad_norm": 1.9783895015716553,
      "learning_rate": 3.657361074532526e-05,
      "loss": 0.5545,
      "step": 10203
    },
    {
      "epoch": 0.2687384777455886,
      "grad_norm": 1.7216050624847412,
      "learning_rate": 3.657229391624967e-05,
      "loss": 2.0621,
      "step": 10204
    },
    {
      "epoch": 0.26876481432710037,
      "grad_norm": 2.0586533546447754,
      "learning_rate": 3.6570977087174086e-05,
      "loss": 1.7193,
      "step": 10205
    },
    {
      "epoch": 0.26879115090861205,
      "grad_norm": 3.382467031478882,
      "learning_rate": 3.6569660258098495e-05,
      "loss": 1.0798,
      "step": 10206
    },
    {
      "epoch": 0.2688174874901238,
      "grad_norm": 1.9821090698242188,
      "learning_rate": 3.656834342902292e-05,
      "loss": 1.2546,
      "step": 10207
    },
    {
      "epoch": 0.2688438240716355,
      "grad_norm": 3.441862106323242,
      "learning_rate": 3.6567026599947326e-05,
      "loss": 1.7819,
      "step": 10208
    },
    {
      "epoch": 0.2688701606531472,
      "grad_norm": 1.9196189641952515,
      "learning_rate": 3.656570977087174e-05,
      "loss": 1.7627,
      "step": 10209
    },
    {
      "epoch": 0.2688964972346589,
      "grad_norm": 2.1342666149139404,
      "learning_rate": 3.656439294179616e-05,
      "loss": 1.842,
      "step": 10210
    },
    {
      "epoch": 0.26892283381617066,
      "grad_norm": 4.082106590270996,
      "learning_rate": 3.6563076112720566e-05,
      "loss": 1.4424,
      "step": 10211
    },
    {
      "epoch": 0.2689491703976824,
      "grad_norm": 1.8392866849899292,
      "learning_rate": 3.656175928364499e-05,
      "loss": 1.7364,
      "step": 10212
    },
    {
      "epoch": 0.2689755069791941,
      "grad_norm": 2.4836320877075195,
      "learning_rate": 3.65604424545694e-05,
      "loss": 1.3697,
      "step": 10213
    },
    {
      "epoch": 0.26900184356070583,
      "grad_norm": 2.3564136028289795,
      "learning_rate": 3.655912562549381e-05,
      "loss": 2.1738,
      "step": 10214
    },
    {
      "epoch": 0.2690281801422175,
      "grad_norm": 1.7465589046478271,
      "learning_rate": 3.655780879641822e-05,
      "loss": 1.643,
      "step": 10215
    },
    {
      "epoch": 0.26905451672372926,
      "grad_norm": 3.9097602367401123,
      "learning_rate": 3.6556491967342644e-05,
      "loss": 1.5084,
      "step": 10216
    },
    {
      "epoch": 0.26908085330524095,
      "grad_norm": 3.0084879398345947,
      "learning_rate": 3.655517513826705e-05,
      "loss": 1.6082,
      "step": 10217
    },
    {
      "epoch": 0.2691071898867527,
      "grad_norm": 1.3666701316833496,
      "learning_rate": 3.655385830919147e-05,
      "loss": 1.8943,
      "step": 10218
    },
    {
      "epoch": 0.26913352646826444,
      "grad_norm": 1.8530930280685425,
      "learning_rate": 3.6552541480115884e-05,
      "loss": 1.7454,
      "step": 10219
    },
    {
      "epoch": 0.2691598630497761,
      "grad_norm": 2.229616165161133,
      "learning_rate": 3.655122465104029e-05,
      "loss": 1.8255,
      "step": 10220
    },
    {
      "epoch": 0.26918619963128787,
      "grad_norm": 3.463870048522949,
      "learning_rate": 3.6549907821964715e-05,
      "loss": 1.612,
      "step": 10221
    },
    {
      "epoch": 0.26921253621279956,
      "grad_norm": 3.262535572052002,
      "learning_rate": 3.6548590992889124e-05,
      "loss": 1.2556,
      "step": 10222
    },
    {
      "epoch": 0.2692388727943113,
      "grad_norm": 1.6866883039474487,
      "learning_rate": 3.654727416381354e-05,
      "loss": 1.618,
      "step": 10223
    },
    {
      "epoch": 0.26926520937582304,
      "grad_norm": 2.051145076751709,
      "learning_rate": 3.654595733473795e-05,
      "loss": 1.3316,
      "step": 10224
    },
    {
      "epoch": 0.26929154595733473,
      "grad_norm": 1.9978785514831543,
      "learning_rate": 3.6544640505662364e-05,
      "loss": 1.4455,
      "step": 10225
    },
    {
      "epoch": 0.2693178825388465,
      "grad_norm": 4.349775314331055,
      "learning_rate": 3.654332367658678e-05,
      "loss": 1.0462,
      "step": 10226
    },
    {
      "epoch": 0.26934421912035816,
      "grad_norm": 2.0261640548706055,
      "learning_rate": 3.6542006847511195e-05,
      "loss": 1.5558,
      "step": 10227
    },
    {
      "epoch": 0.2693705557018699,
      "grad_norm": 4.261337757110596,
      "learning_rate": 3.654069001843561e-05,
      "loss": 2.2034,
      "step": 10228
    },
    {
      "epoch": 0.2693968922833816,
      "grad_norm": 2.3293251991271973,
      "learning_rate": 3.653937318936002e-05,
      "loss": 0.9925,
      "step": 10229
    },
    {
      "epoch": 0.26942322886489334,
      "grad_norm": 2.065385580062866,
      "learning_rate": 3.653805636028444e-05,
      "loss": 1.5747,
      "step": 10230
    },
    {
      "epoch": 0.2694495654464051,
      "grad_norm": 1.7524702548980713,
      "learning_rate": 3.653673953120885e-05,
      "loss": 1.8194,
      "step": 10231
    },
    {
      "epoch": 0.26947590202791677,
      "grad_norm": 2.126185894012451,
      "learning_rate": 3.653542270213327e-05,
      "loss": 1.3905,
      "step": 10232
    },
    {
      "epoch": 0.2695022386094285,
      "grad_norm": 1.7913614511489868,
      "learning_rate": 3.6534105873057676e-05,
      "loss": 1.561,
      "step": 10233
    },
    {
      "epoch": 0.2695285751909402,
      "grad_norm": 4.287265777587891,
      "learning_rate": 3.653278904398209e-05,
      "loss": 1.7555,
      "step": 10234
    },
    {
      "epoch": 0.26955491177245194,
      "grad_norm": 2.717137336730957,
      "learning_rate": 3.653147221490651e-05,
      "loss": 2.051,
      "step": 10235
    },
    {
      "epoch": 0.26958124835396363,
      "grad_norm": 2.6772007942199707,
      "learning_rate": 3.653015538583092e-05,
      "loss": 1.4684,
      "step": 10236
    },
    {
      "epoch": 0.2696075849354754,
      "grad_norm": 3.531998872756958,
      "learning_rate": 3.652883855675534e-05,
      "loss": 1.3407,
      "step": 10237
    },
    {
      "epoch": 0.2696339215169871,
      "grad_norm": 2.233687400817871,
      "learning_rate": 3.652752172767975e-05,
      "loss": 1.4213,
      "step": 10238
    },
    {
      "epoch": 0.2696602580984988,
      "grad_norm": 1.4807181358337402,
      "learning_rate": 3.652620489860416e-05,
      "loss": 1.6249,
      "step": 10239
    },
    {
      "epoch": 0.26968659468001055,
      "grad_norm": 2.244846820831299,
      "learning_rate": 3.652488806952858e-05,
      "loss": 0.6817,
      "step": 10240
    },
    {
      "epoch": 0.26971293126152224,
      "grad_norm": 1.7698038816452026,
      "learning_rate": 3.6523571240452994e-05,
      "loss": 1.9588,
      "step": 10241
    },
    {
      "epoch": 0.269739267843034,
      "grad_norm": 5.134593486785889,
      "learning_rate": 3.65222544113774e-05,
      "loss": 1.7539,
      "step": 10242
    },
    {
      "epoch": 0.26976560442454567,
      "grad_norm": 2.8440356254577637,
      "learning_rate": 3.652093758230182e-05,
      "loss": 2.2596,
      "step": 10243
    },
    {
      "epoch": 0.2697919410060574,
      "grad_norm": 2.1762497425079346,
      "learning_rate": 3.6519620753226234e-05,
      "loss": 1.9777,
      "step": 10244
    },
    {
      "epoch": 0.26981827758756916,
      "grad_norm": 3.0282039642333984,
      "learning_rate": 3.651830392415065e-05,
      "loss": 1.4478,
      "step": 10245
    },
    {
      "epoch": 0.26984461416908084,
      "grad_norm": 1.7023380994796753,
      "learning_rate": 3.6516987095075065e-05,
      "loss": 1.8943,
      "step": 10246
    },
    {
      "epoch": 0.2698709507505926,
      "grad_norm": 1.9163802862167358,
      "learning_rate": 3.6515670265999474e-05,
      "loss": 1.6845,
      "step": 10247
    },
    {
      "epoch": 0.2698972873321043,
      "grad_norm": 1.7483388185501099,
      "learning_rate": 3.651435343692389e-05,
      "loss": 2.1314,
      "step": 10248
    },
    {
      "epoch": 0.269923623913616,
      "grad_norm": 1.972560167312622,
      "learning_rate": 3.6513036607848305e-05,
      "loss": 2.1892,
      "step": 10249
    },
    {
      "epoch": 0.2699499604951277,
      "grad_norm": 1.8761812448501587,
      "learning_rate": 3.651171977877272e-05,
      "loss": 1.6801,
      "step": 10250
    },
    {
      "epoch": 0.26997629707663945,
      "grad_norm": 1.4988170862197876,
      "learning_rate": 3.651040294969713e-05,
      "loss": 1.7104,
      "step": 10251
    },
    {
      "epoch": 0.2700026336581512,
      "grad_norm": 2.0416862964630127,
      "learning_rate": 3.6509086120621545e-05,
      "loss": 1.9879,
      "step": 10252
    },
    {
      "epoch": 0.2700289702396629,
      "grad_norm": 2.491778612136841,
      "learning_rate": 3.6507769291545954e-05,
      "loss": 1.3525,
      "step": 10253
    },
    {
      "epoch": 0.2700553068211746,
      "grad_norm": 2.2661430835723877,
      "learning_rate": 3.6506452462470376e-05,
      "loss": 1.8682,
      "step": 10254
    },
    {
      "epoch": 0.2700816434026863,
      "grad_norm": 2.217207431793213,
      "learning_rate": 3.6505135633394785e-05,
      "loss": 1.878,
      "step": 10255
    },
    {
      "epoch": 0.27010797998419805,
      "grad_norm": 1.5754505395889282,
      "learning_rate": 3.65038188043192e-05,
      "loss": 1.9861,
      "step": 10256
    },
    {
      "epoch": 0.2701343165657098,
      "grad_norm": 3.6967532634735107,
      "learning_rate": 3.6502501975243616e-05,
      "loss": 1.2931,
      "step": 10257
    },
    {
      "epoch": 0.2701606531472215,
      "grad_norm": 2.2997963428497314,
      "learning_rate": 3.6501185146168025e-05,
      "loss": 1.3636,
      "step": 10258
    },
    {
      "epoch": 0.27018698972873323,
      "grad_norm": 1.778559684753418,
      "learning_rate": 3.649986831709245e-05,
      "loss": 1.8104,
      "step": 10259
    },
    {
      "epoch": 0.2702133263102449,
      "grad_norm": 2.6709375381469727,
      "learning_rate": 3.6498551488016856e-05,
      "loss": 2.0059,
      "step": 10260
    },
    {
      "epoch": 0.27023966289175666,
      "grad_norm": 2.425659656524658,
      "learning_rate": 3.649723465894127e-05,
      "loss": 0.6497,
      "step": 10261
    },
    {
      "epoch": 0.27026599947326835,
      "grad_norm": 2.145305871963501,
      "learning_rate": 3.649591782986568e-05,
      "loss": 2.8096,
      "step": 10262
    },
    {
      "epoch": 0.2702923360547801,
      "grad_norm": 1.9830477237701416,
      "learning_rate": 3.64946010007901e-05,
      "loss": 1.8591,
      "step": 10263
    },
    {
      "epoch": 0.27031867263629183,
      "grad_norm": 1.9636954069137573,
      "learning_rate": 3.649328417171451e-05,
      "loss": 1.9282,
      "step": 10264
    },
    {
      "epoch": 0.2703450092178035,
      "grad_norm": 2.2643651962280273,
      "learning_rate": 3.649196734263893e-05,
      "loss": 2.0527,
      "step": 10265
    },
    {
      "epoch": 0.27037134579931527,
      "grad_norm": 1.8713598251342773,
      "learning_rate": 3.649065051356334e-05,
      "loss": 2.5977,
      "step": 10266
    },
    {
      "epoch": 0.27039768238082695,
      "grad_norm": 1.8056583404541016,
      "learning_rate": 3.648933368448775e-05,
      "loss": 1.5701,
      "step": 10267
    },
    {
      "epoch": 0.2704240189623387,
      "grad_norm": 2.695420503616333,
      "learning_rate": 3.6488016855412174e-05,
      "loss": 1.2163,
      "step": 10268
    },
    {
      "epoch": 0.2704503555438504,
      "grad_norm": 4.3372931480407715,
      "learning_rate": 3.648670002633658e-05,
      "loss": 1.5018,
      "step": 10269
    },
    {
      "epoch": 0.27047669212536213,
      "grad_norm": 4.943918704986572,
      "learning_rate": 3.6485383197261e-05,
      "loss": 1.6296,
      "step": 10270
    },
    {
      "epoch": 0.27050302870687387,
      "grad_norm": 1.8845360279083252,
      "learning_rate": 3.648406636818541e-05,
      "loss": 2.0695,
      "step": 10271
    },
    {
      "epoch": 0.27052936528838556,
      "grad_norm": 4.754239559173584,
      "learning_rate": 3.648274953910982e-05,
      "loss": 0.8524,
      "step": 10272
    },
    {
      "epoch": 0.2705557018698973,
      "grad_norm": 1.9632288217544556,
      "learning_rate": 3.648143271003424e-05,
      "loss": 2.0543,
      "step": 10273
    },
    {
      "epoch": 0.270582038451409,
      "grad_norm": 1.8391855955123901,
      "learning_rate": 3.6480115880958654e-05,
      "loss": 1.5461,
      "step": 10274
    },
    {
      "epoch": 0.27060837503292073,
      "grad_norm": 4.078208923339844,
      "learning_rate": 3.647879905188307e-05,
      "loss": 0.8907,
      "step": 10275
    },
    {
      "epoch": 0.2706347116144324,
      "grad_norm": 2.712761640548706,
      "learning_rate": 3.647748222280748e-05,
      "loss": 2.1295,
      "step": 10276
    },
    {
      "epoch": 0.27066104819594416,
      "grad_norm": 2.2774975299835205,
      "learning_rate": 3.64761653937319e-05,
      "loss": 1.0989,
      "step": 10277
    },
    {
      "epoch": 0.2706873847774559,
      "grad_norm": 2.394390106201172,
      "learning_rate": 3.647484856465631e-05,
      "loss": 1.8499,
      "step": 10278
    },
    {
      "epoch": 0.2707137213589676,
      "grad_norm": 1.5043600797653198,
      "learning_rate": 3.6473531735580725e-05,
      "loss": 0.4539,
      "step": 10279
    },
    {
      "epoch": 0.27074005794047934,
      "grad_norm": 2.093470335006714,
      "learning_rate": 3.6472214906505134e-05,
      "loss": 0.4458,
      "step": 10280
    },
    {
      "epoch": 0.270766394521991,
      "grad_norm": 1.9578824043273926,
      "learning_rate": 3.647089807742955e-05,
      "loss": 1.9961,
      "step": 10281
    },
    {
      "epoch": 0.27079273110350277,
      "grad_norm": 2.1867403984069824,
      "learning_rate": 3.6469581248353965e-05,
      "loss": 1.7453,
      "step": 10282
    },
    {
      "epoch": 0.27081906768501446,
      "grad_norm": 1.919355034828186,
      "learning_rate": 3.646826441927838e-05,
      "loss": 1.4913,
      "step": 10283
    },
    {
      "epoch": 0.2708454042665262,
      "grad_norm": 2.5337464809417725,
      "learning_rate": 3.6466947590202796e-05,
      "loss": 1.8079,
      "step": 10284
    },
    {
      "epoch": 0.27087174084803795,
      "grad_norm": 2.3828213214874268,
      "learning_rate": 3.6465630761127205e-05,
      "loss": 2.2451,
      "step": 10285
    },
    {
      "epoch": 0.27089807742954963,
      "grad_norm": 2.907728433609009,
      "learning_rate": 3.646431393205162e-05,
      "loss": 1.1044,
      "step": 10286
    },
    {
      "epoch": 0.2709244140110614,
      "grad_norm": 1.8900034427642822,
      "learning_rate": 3.6462997102976036e-05,
      "loss": 1.6356,
      "step": 10287
    },
    {
      "epoch": 0.27095075059257306,
      "grad_norm": 2.902811288833618,
      "learning_rate": 3.646168027390045e-05,
      "loss": 0.5479,
      "step": 10288
    },
    {
      "epoch": 0.2709770871740848,
      "grad_norm": 2.088362455368042,
      "learning_rate": 3.646036344482486e-05,
      "loss": 2.0231,
      "step": 10289
    },
    {
      "epoch": 0.27100342375559655,
      "grad_norm": 1.5732823610305786,
      "learning_rate": 3.6459046615749276e-05,
      "loss": 1.4694,
      "step": 10290
    },
    {
      "epoch": 0.27102976033710824,
      "grad_norm": 2.044257640838623,
      "learning_rate": 3.645772978667369e-05,
      "loss": 1.2716,
      "step": 10291
    },
    {
      "epoch": 0.27105609691862,
      "grad_norm": 1.5156606435775757,
      "learning_rate": 3.645641295759811e-05,
      "loss": 2.5358,
      "step": 10292
    },
    {
      "epoch": 0.27108243350013167,
      "grad_norm": 2.078439474105835,
      "learning_rate": 3.645509612852252e-05,
      "loss": 1.5971,
      "step": 10293
    },
    {
      "epoch": 0.2711087700816434,
      "grad_norm": 2.194434642791748,
      "learning_rate": 3.645377929944693e-05,
      "loss": 1.6092,
      "step": 10294
    },
    {
      "epoch": 0.2711351066631551,
      "grad_norm": 3.3006484508514404,
      "learning_rate": 3.645246247037135e-05,
      "loss": 1.2254,
      "step": 10295
    },
    {
      "epoch": 0.27116144324466684,
      "grad_norm": 2.5422470569610596,
      "learning_rate": 3.645114564129576e-05,
      "loss": 1.8761,
      "step": 10296
    },
    {
      "epoch": 0.2711877798261786,
      "grad_norm": 2.4962961673736572,
      "learning_rate": 3.644982881222018e-05,
      "loss": 0.685,
      "step": 10297
    },
    {
      "epoch": 0.2712141164076903,
      "grad_norm": 2.2493808269500732,
      "learning_rate": 3.644851198314459e-05,
      "loss": 1.89,
      "step": 10298
    },
    {
      "epoch": 0.271240452989202,
      "grad_norm": 1.9526785612106323,
      "learning_rate": 3.6447195154069e-05,
      "loss": 2.5459,
      "step": 10299
    },
    {
      "epoch": 0.2712667895707137,
      "grad_norm": 1.5954350233078003,
      "learning_rate": 3.644587832499341e-05,
      "loss": 2.182,
      "step": 10300
    },
    {
      "epoch": 0.27129312615222545,
      "grad_norm": 2.677779197692871,
      "learning_rate": 3.6444561495917835e-05,
      "loss": 0.5432,
      "step": 10301
    },
    {
      "epoch": 0.27131946273373714,
      "grad_norm": 1.6161595582962036,
      "learning_rate": 3.644324466684224e-05,
      "loss": 0.3551,
      "step": 10302
    },
    {
      "epoch": 0.2713457993152489,
      "grad_norm": 2.9487502574920654,
      "learning_rate": 3.644192783776666e-05,
      "loss": 1.8901,
      "step": 10303
    },
    {
      "epoch": 0.2713721358967606,
      "grad_norm": 1.8856022357940674,
      "learning_rate": 3.6440611008691075e-05,
      "loss": 2.214,
      "step": 10304
    },
    {
      "epoch": 0.2713984724782723,
      "grad_norm": 2.3007776737213135,
      "learning_rate": 3.643929417961548e-05,
      "loss": 2.2138,
      "step": 10305
    },
    {
      "epoch": 0.27142480905978406,
      "grad_norm": 1.6632148027420044,
      "learning_rate": 3.6437977350539906e-05,
      "loss": 2.1418,
      "step": 10306
    },
    {
      "epoch": 0.27145114564129574,
      "grad_norm": 1.6475247144699097,
      "learning_rate": 3.6436660521464315e-05,
      "loss": 1.8145,
      "step": 10307
    },
    {
      "epoch": 0.2714774822228075,
      "grad_norm": 3.1381053924560547,
      "learning_rate": 3.643534369238873e-05,
      "loss": 0.7929,
      "step": 10308
    },
    {
      "epoch": 0.2715038188043192,
      "grad_norm": 1.8969988822937012,
      "learning_rate": 3.643402686331314e-05,
      "loss": 0.7786,
      "step": 10309
    },
    {
      "epoch": 0.2715301553858309,
      "grad_norm": 1.9219800233840942,
      "learning_rate": 3.643271003423756e-05,
      "loss": 1.5643,
      "step": 10310
    },
    {
      "epoch": 0.27155649196734266,
      "grad_norm": 2.1371917724609375,
      "learning_rate": 3.643139320516197e-05,
      "loss": 0.5427,
      "step": 10311
    },
    {
      "epoch": 0.27158282854885435,
      "grad_norm": 3.16784405708313,
      "learning_rate": 3.6430076376086386e-05,
      "loss": 1.7544,
      "step": 10312
    },
    {
      "epoch": 0.2716091651303661,
      "grad_norm": 2.1667733192443848,
      "learning_rate": 3.64287595470108e-05,
      "loss": 1.9134,
      "step": 10313
    },
    {
      "epoch": 0.2716355017118778,
      "grad_norm": 2.247311592102051,
      "learning_rate": 3.642744271793521e-05,
      "loss": 1.8834,
      "step": 10314
    },
    {
      "epoch": 0.2716618382933895,
      "grad_norm": 2.0783884525299072,
      "learning_rate": 3.642612588885963e-05,
      "loss": 1.6994,
      "step": 10315
    },
    {
      "epoch": 0.2716881748749012,
      "grad_norm": 1.5845547914505005,
      "learning_rate": 3.642480905978404e-05,
      "loss": 1.692,
      "step": 10316
    },
    {
      "epoch": 0.27171451145641295,
      "grad_norm": 2.274195909500122,
      "learning_rate": 3.642349223070846e-05,
      "loss": 1.1233,
      "step": 10317
    },
    {
      "epoch": 0.2717408480379247,
      "grad_norm": 2.503321647644043,
      "learning_rate": 3.6422175401632866e-05,
      "loss": 1.7993,
      "step": 10318
    },
    {
      "epoch": 0.2717671846194364,
      "grad_norm": 6.4886980056762695,
      "learning_rate": 3.642085857255728e-05,
      "loss": 1.2929,
      "step": 10319
    },
    {
      "epoch": 0.27179352120094813,
      "grad_norm": 1.6680946350097656,
      "learning_rate": 3.64195417434817e-05,
      "loss": 1.0955,
      "step": 10320
    },
    {
      "epoch": 0.2718198577824598,
      "grad_norm": 4.5090789794921875,
      "learning_rate": 3.641822491440611e-05,
      "loss": 0.7915,
      "step": 10321
    },
    {
      "epoch": 0.27184619436397156,
      "grad_norm": 2.5213704109191895,
      "learning_rate": 3.641690808533053e-05,
      "loss": 2.5157,
      "step": 10322
    },
    {
      "epoch": 0.2718725309454833,
      "grad_norm": 1.5279197692871094,
      "learning_rate": 3.641559125625494e-05,
      "loss": 2.0742,
      "step": 10323
    },
    {
      "epoch": 0.271898867526995,
      "grad_norm": 2.4287710189819336,
      "learning_rate": 3.641427442717935e-05,
      "loss": 2.5385,
      "step": 10324
    },
    {
      "epoch": 0.27192520410850674,
      "grad_norm": 1.6410257816314697,
      "learning_rate": 3.641295759810377e-05,
      "loss": 1.9969,
      "step": 10325
    },
    {
      "epoch": 0.2719515406900184,
      "grad_norm": 1.8392621278762817,
      "learning_rate": 3.6411640769028184e-05,
      "loss": 0.6193,
      "step": 10326
    },
    {
      "epoch": 0.27197787727153017,
      "grad_norm": 5.369431495666504,
      "learning_rate": 3.641032393995259e-05,
      "loss": 1.2819,
      "step": 10327
    },
    {
      "epoch": 0.27200421385304185,
      "grad_norm": 2.5196642875671387,
      "learning_rate": 3.640900711087701e-05,
      "loss": 1.9757,
      "step": 10328
    },
    {
      "epoch": 0.2720305504345536,
      "grad_norm": 2.9400951862335205,
      "learning_rate": 3.6407690281801424e-05,
      "loss": 1.8495,
      "step": 10329
    },
    {
      "epoch": 0.27205688701606534,
      "grad_norm": 2.3373031616210938,
      "learning_rate": 3.640637345272584e-05,
      "loss": 1.5409,
      "step": 10330
    },
    {
      "epoch": 0.27208322359757703,
      "grad_norm": 3.901477813720703,
      "learning_rate": 3.6405056623650255e-05,
      "loss": 1.9812,
      "step": 10331
    },
    {
      "epoch": 0.27210956017908877,
      "grad_norm": 1.6351262331008911,
      "learning_rate": 3.6403739794574664e-05,
      "loss": 1.3154,
      "step": 10332
    },
    {
      "epoch": 0.27213589676060046,
      "grad_norm": 2.3115203380584717,
      "learning_rate": 3.640242296549908e-05,
      "loss": 1.4164,
      "step": 10333
    },
    {
      "epoch": 0.2721622333421122,
      "grad_norm": 2.477191209793091,
      "learning_rate": 3.6401106136423495e-05,
      "loss": 0.87,
      "step": 10334
    },
    {
      "epoch": 0.2721885699236239,
      "grad_norm": 2.5607640743255615,
      "learning_rate": 3.639978930734791e-05,
      "loss": 1.7121,
      "step": 10335
    },
    {
      "epoch": 0.27221490650513563,
      "grad_norm": 2.4744653701782227,
      "learning_rate": 3.639847247827232e-05,
      "loss": 1.4515,
      "step": 10336
    },
    {
      "epoch": 0.2722412430866474,
      "grad_norm": 1.6541136503219604,
      "learning_rate": 3.6397155649196735e-05,
      "loss": 2.0111,
      "step": 10337
    },
    {
      "epoch": 0.27226757966815907,
      "grad_norm": 2.601785898208618,
      "learning_rate": 3.639583882012115e-05,
      "loss": 0.8639,
      "step": 10338
    },
    {
      "epoch": 0.2722939162496708,
      "grad_norm": 2.812194585800171,
      "learning_rate": 3.6394521991045566e-05,
      "loss": 1.4026,
      "step": 10339
    },
    {
      "epoch": 0.2723202528311825,
      "grad_norm": 2.294760227203369,
      "learning_rate": 3.639320516196998e-05,
      "loss": 1.4016,
      "step": 10340
    },
    {
      "epoch": 0.27234658941269424,
      "grad_norm": 1.8534947633743286,
      "learning_rate": 3.639188833289439e-05,
      "loss": 1.8126,
      "step": 10341
    },
    {
      "epoch": 0.27237292599420593,
      "grad_norm": 2.672706365585327,
      "learning_rate": 3.6390571503818806e-05,
      "loss": 2.0264,
      "step": 10342
    },
    {
      "epoch": 0.27239926257571767,
      "grad_norm": 3.424736738204956,
      "learning_rate": 3.638925467474322e-05,
      "loss": 0.6968,
      "step": 10343
    },
    {
      "epoch": 0.2724255991572294,
      "grad_norm": 2.4977211952209473,
      "learning_rate": 3.638793784566764e-05,
      "loss": 1.8002,
      "step": 10344
    },
    {
      "epoch": 0.2724519357387411,
      "grad_norm": 2.5923004150390625,
      "learning_rate": 3.6386621016592046e-05,
      "loss": 1.0333,
      "step": 10345
    },
    {
      "epoch": 0.27247827232025285,
      "grad_norm": 2.979314088821411,
      "learning_rate": 3.638530418751646e-05,
      "loss": 1.5285,
      "step": 10346
    },
    {
      "epoch": 0.27250460890176453,
      "grad_norm": 2.8747994899749756,
      "learning_rate": 3.638398735844088e-05,
      "loss": 1.6198,
      "step": 10347
    },
    {
      "epoch": 0.2725309454832763,
      "grad_norm": 2.0636353492736816,
      "learning_rate": 3.638267052936529e-05,
      "loss": 1.6727,
      "step": 10348
    },
    {
      "epoch": 0.27255728206478796,
      "grad_norm": 4.8102216720581055,
      "learning_rate": 3.638135370028971e-05,
      "loss": 1.1448,
      "step": 10349
    },
    {
      "epoch": 0.2725836186462997,
      "grad_norm": 1.7907400131225586,
      "learning_rate": 3.638003687121412e-05,
      "loss": 1.7525,
      "step": 10350
    },
    {
      "epoch": 0.27260995522781145,
      "grad_norm": 2.293001413345337,
      "learning_rate": 3.637872004213853e-05,
      "loss": 1.6077,
      "step": 10351
    },
    {
      "epoch": 0.27263629180932314,
      "grad_norm": 2.3834128379821777,
      "learning_rate": 3.637740321306294e-05,
      "loss": 1.5715,
      "step": 10352
    },
    {
      "epoch": 0.2726626283908349,
      "grad_norm": 1.9715545177459717,
      "learning_rate": 3.6376086383987364e-05,
      "loss": 1.3433,
      "step": 10353
    },
    {
      "epoch": 0.27268896497234657,
      "grad_norm": 2.0868146419525146,
      "learning_rate": 3.637476955491177e-05,
      "loss": 2.0491,
      "step": 10354
    },
    {
      "epoch": 0.2727153015538583,
      "grad_norm": 2.4551172256469727,
      "learning_rate": 3.637345272583619e-05,
      "loss": 0.9652,
      "step": 10355
    },
    {
      "epoch": 0.27274163813537,
      "grad_norm": 1.6090114116668701,
      "learning_rate": 3.63721358967606e-05,
      "loss": 1.5764,
      "step": 10356
    },
    {
      "epoch": 0.27276797471688174,
      "grad_norm": 2.094609260559082,
      "learning_rate": 3.637081906768501e-05,
      "loss": 0.3454,
      "step": 10357
    },
    {
      "epoch": 0.2727943112983935,
      "grad_norm": 3.1312239170074463,
      "learning_rate": 3.636950223860943e-05,
      "loss": 2.2632,
      "step": 10358
    },
    {
      "epoch": 0.2728206478799052,
      "grad_norm": 2.2778000831604004,
      "learning_rate": 3.6368185409533844e-05,
      "loss": 0.7167,
      "step": 10359
    },
    {
      "epoch": 0.2728469844614169,
      "grad_norm": 2.317643165588379,
      "learning_rate": 3.636686858045826e-05,
      "loss": 1.7029,
      "step": 10360
    },
    {
      "epoch": 0.2728733210429286,
      "grad_norm": 1.8217017650604248,
      "learning_rate": 3.636555175138267e-05,
      "loss": 1.6405,
      "step": 10361
    },
    {
      "epoch": 0.27289965762444035,
      "grad_norm": 2.6209917068481445,
      "learning_rate": 3.636423492230709e-05,
      "loss": 1.9569,
      "step": 10362
    },
    {
      "epoch": 0.2729259942059521,
      "grad_norm": 1.8491734266281128,
      "learning_rate": 3.63629180932315e-05,
      "loss": 1.6601,
      "step": 10363
    },
    {
      "epoch": 0.2729523307874638,
      "grad_norm": 5.69320821762085,
      "learning_rate": 3.6361601264155916e-05,
      "loss": 1.5579,
      "step": 10364
    },
    {
      "epoch": 0.2729786673689755,
      "grad_norm": 2.1832668781280518,
      "learning_rate": 3.6360284435080324e-05,
      "loss": 0.5198,
      "step": 10365
    },
    {
      "epoch": 0.2730050039504872,
      "grad_norm": 3.8296897411346436,
      "learning_rate": 3.635896760600474e-05,
      "loss": 0.942,
      "step": 10366
    },
    {
      "epoch": 0.27303134053199896,
      "grad_norm": 2.4550304412841797,
      "learning_rate": 3.6357650776929156e-05,
      "loss": 2.419,
      "step": 10367
    },
    {
      "epoch": 0.27305767711351064,
      "grad_norm": 2.2154898643493652,
      "learning_rate": 3.635633394785357e-05,
      "loss": 1.8225,
      "step": 10368
    },
    {
      "epoch": 0.2730840136950224,
      "grad_norm": 2.387238025665283,
      "learning_rate": 3.635501711877799e-05,
      "loss": 1.633,
      "step": 10369
    },
    {
      "epoch": 0.27311035027653413,
      "grad_norm": 1.885642647743225,
      "learning_rate": 3.6353700289702396e-05,
      "loss": 1.8477,
      "step": 10370
    },
    {
      "epoch": 0.2731366868580458,
      "grad_norm": 2.8946776390075684,
      "learning_rate": 3.635238346062681e-05,
      "loss": 1.5476,
      "step": 10371
    },
    {
      "epoch": 0.27316302343955756,
      "grad_norm": 3.3582749366760254,
      "learning_rate": 3.635106663155123e-05,
      "loss": 0.4198,
      "step": 10372
    },
    {
      "epoch": 0.27318936002106925,
      "grad_norm": 2.1126949787139893,
      "learning_rate": 3.634974980247564e-05,
      "loss": 2.3508,
      "step": 10373
    },
    {
      "epoch": 0.273215696602581,
      "grad_norm": 1.748124599456787,
      "learning_rate": 3.634843297340005e-05,
      "loss": 1.9212,
      "step": 10374
    },
    {
      "epoch": 0.2732420331840927,
      "grad_norm": 4.46715784072876,
      "learning_rate": 3.634711614432447e-05,
      "loss": 1.4425,
      "step": 10375
    },
    {
      "epoch": 0.2732683697656044,
      "grad_norm": 2.0045342445373535,
      "learning_rate": 3.634579931524888e-05,
      "loss": 1.6954,
      "step": 10376
    },
    {
      "epoch": 0.27329470634711617,
      "grad_norm": 2.654986619949341,
      "learning_rate": 3.63444824861733e-05,
      "loss": 1.2639,
      "step": 10377
    },
    {
      "epoch": 0.27332104292862786,
      "grad_norm": 1.6185369491577148,
      "learning_rate": 3.6343165657097714e-05,
      "loss": 2.3115,
      "step": 10378
    },
    {
      "epoch": 0.2733473795101396,
      "grad_norm": 3.293684959411621,
      "learning_rate": 3.634184882802212e-05,
      "loss": 1.452,
      "step": 10379
    },
    {
      "epoch": 0.2733737160916513,
      "grad_norm": 2.2743654251098633,
      "learning_rate": 3.634053199894654e-05,
      "loss": 1.8677,
      "step": 10380
    },
    {
      "epoch": 0.27340005267316303,
      "grad_norm": 1.528550386428833,
      "learning_rate": 3.6339215169870954e-05,
      "loss": 1.9407,
      "step": 10381
    },
    {
      "epoch": 0.2734263892546747,
      "grad_norm": 4.636200904846191,
      "learning_rate": 3.633789834079537e-05,
      "loss": 0.4386,
      "step": 10382
    },
    {
      "epoch": 0.27345272583618646,
      "grad_norm": 2.5900821685791016,
      "learning_rate": 3.633658151171978e-05,
      "loss": 1.9323,
      "step": 10383
    },
    {
      "epoch": 0.2734790624176982,
      "grad_norm": 1.6953577995300293,
      "learning_rate": 3.6335264682644194e-05,
      "loss": 1.9221,
      "step": 10384
    },
    {
      "epoch": 0.2735053989992099,
      "grad_norm": 1.9355050325393677,
      "learning_rate": 3.633394785356861e-05,
      "loss": 2.1167,
      "step": 10385
    },
    {
      "epoch": 0.27353173558072164,
      "grad_norm": 2.9548747539520264,
      "learning_rate": 3.6332631024493025e-05,
      "loss": 1.0196,
      "step": 10386
    },
    {
      "epoch": 0.2735580721622333,
      "grad_norm": 1.5879417657852173,
      "learning_rate": 3.633131419541744e-05,
      "loss": 1.5018,
      "step": 10387
    },
    {
      "epoch": 0.27358440874374507,
      "grad_norm": 3.798142194747925,
      "learning_rate": 3.632999736634185e-05,
      "loss": 1.1752,
      "step": 10388
    },
    {
      "epoch": 0.27361074532525675,
      "grad_norm": 2.2538506984710693,
      "learning_rate": 3.6328680537266265e-05,
      "loss": 2.4424,
      "step": 10389
    },
    {
      "epoch": 0.2736370819067685,
      "grad_norm": 2.5515694618225098,
      "learning_rate": 3.6327363708190674e-05,
      "loss": 0.5356,
      "step": 10390
    },
    {
      "epoch": 0.27366341848828024,
      "grad_norm": 1.8959494829177856,
      "learning_rate": 3.6326046879115096e-05,
      "loss": 1.8924,
      "step": 10391
    },
    {
      "epoch": 0.27368975506979193,
      "grad_norm": 4.369131088256836,
      "learning_rate": 3.6324730050039505e-05,
      "loss": 1.2374,
      "step": 10392
    },
    {
      "epoch": 0.2737160916513037,
      "grad_norm": 3.6731436252593994,
      "learning_rate": 3.632341322096392e-05,
      "loss": 1.5548,
      "step": 10393
    },
    {
      "epoch": 0.27374242823281536,
      "grad_norm": 2.345093250274658,
      "learning_rate": 3.6322096391888336e-05,
      "loss": 1.6417,
      "step": 10394
    },
    {
      "epoch": 0.2737687648143271,
      "grad_norm": 2.260816812515259,
      "learning_rate": 3.632077956281275e-05,
      "loss": 1.9425,
      "step": 10395
    },
    {
      "epoch": 0.27379510139583885,
      "grad_norm": 4.049269199371338,
      "learning_rate": 3.631946273373717e-05,
      "loss": 2.0169,
      "step": 10396
    },
    {
      "epoch": 0.27382143797735053,
      "grad_norm": 1.8645570278167725,
      "learning_rate": 3.6318145904661576e-05,
      "loss": 1.8006,
      "step": 10397
    },
    {
      "epoch": 0.2738477745588623,
      "grad_norm": 2.959125518798828,
      "learning_rate": 3.631682907558599e-05,
      "loss": 1.4961,
      "step": 10398
    },
    {
      "epoch": 0.27387411114037397,
      "grad_norm": 2.605584144592285,
      "learning_rate": 3.63155122465104e-05,
      "loss": 0.6815,
      "step": 10399
    },
    {
      "epoch": 0.2739004477218857,
      "grad_norm": 3.761110782623291,
      "learning_rate": 3.631419541743482e-05,
      "loss": 1.1962,
      "step": 10400
    },
    {
      "epoch": 0.2739267843033974,
      "grad_norm": 2.673065185546875,
      "learning_rate": 3.631287858835923e-05,
      "loss": 1.8779,
      "step": 10401
    },
    {
      "epoch": 0.27395312088490914,
      "grad_norm": 1.6883829832077026,
      "learning_rate": 3.631156175928365e-05,
      "loss": 2.304,
      "step": 10402
    },
    {
      "epoch": 0.2739794574664209,
      "grad_norm": 1.82489013671875,
      "learning_rate": 3.6310244930208056e-05,
      "loss": 2.2771,
      "step": 10403
    },
    {
      "epoch": 0.27400579404793257,
      "grad_norm": 2.5863540172576904,
      "learning_rate": 3.630892810113247e-05,
      "loss": 1.9069,
      "step": 10404
    },
    {
      "epoch": 0.2740321306294443,
      "grad_norm": 1.6233669519424438,
      "learning_rate": 3.630761127205689e-05,
      "loss": 1.8689,
      "step": 10405
    },
    {
      "epoch": 0.274058467210956,
      "grad_norm": 1.4105255603790283,
      "learning_rate": 3.63062944429813e-05,
      "loss": 2.0748,
      "step": 10406
    },
    {
      "epoch": 0.27408480379246775,
      "grad_norm": 3.1554150581359863,
      "learning_rate": 3.630497761390572e-05,
      "loss": 2.2471,
      "step": 10407
    },
    {
      "epoch": 0.27411114037397943,
      "grad_norm": 2.0766801834106445,
      "learning_rate": 3.630366078483013e-05,
      "loss": 1.8686,
      "step": 10408
    },
    {
      "epoch": 0.2741374769554912,
      "grad_norm": 3.3039138317108154,
      "learning_rate": 3.630234395575455e-05,
      "loss": 1.3837,
      "step": 10409
    },
    {
      "epoch": 0.2741638135370029,
      "grad_norm": 1.975628137588501,
      "learning_rate": 3.630102712667896e-05,
      "loss": 1.6537,
      "step": 10410
    },
    {
      "epoch": 0.2741901501185146,
      "grad_norm": 1.6447898149490356,
      "learning_rate": 3.6299710297603374e-05,
      "loss": 1.7791,
      "step": 10411
    },
    {
      "epoch": 0.27421648670002635,
      "grad_norm": 1.8509248495101929,
      "learning_rate": 3.629839346852778e-05,
      "loss": 1.5306,
      "step": 10412
    },
    {
      "epoch": 0.27424282328153804,
      "grad_norm": 3.179069757461548,
      "learning_rate": 3.62970766394522e-05,
      "loss": 1.0179,
      "step": 10413
    },
    {
      "epoch": 0.2742691598630498,
      "grad_norm": 2.821903944015503,
      "learning_rate": 3.6295759810376614e-05,
      "loss": 1.6608,
      "step": 10414
    },
    {
      "epoch": 0.27429549644456147,
      "grad_norm": 1.7800081968307495,
      "learning_rate": 3.629444298130103e-05,
      "loss": 1.3441,
      "step": 10415
    },
    {
      "epoch": 0.2743218330260732,
      "grad_norm": 2.4287467002868652,
      "learning_rate": 3.6293126152225445e-05,
      "loss": 1.1843,
      "step": 10416
    },
    {
      "epoch": 0.27434816960758496,
      "grad_norm": 1.6981650590896606,
      "learning_rate": 3.6291809323149854e-05,
      "loss": 1.6747,
      "step": 10417
    },
    {
      "epoch": 0.27437450618909665,
      "grad_norm": 2.2687954902648926,
      "learning_rate": 3.629049249407427e-05,
      "loss": 1.5328,
      "step": 10418
    },
    {
      "epoch": 0.2744008427706084,
      "grad_norm": 4.247220516204834,
      "learning_rate": 3.6289175664998685e-05,
      "loss": 1.1176,
      "step": 10419
    },
    {
      "epoch": 0.2744271793521201,
      "grad_norm": 3.4652798175811768,
      "learning_rate": 3.62878588359231e-05,
      "loss": 0.9128,
      "step": 10420
    },
    {
      "epoch": 0.2744535159336318,
      "grad_norm": 1.9754316806793213,
      "learning_rate": 3.628654200684751e-05,
      "loss": 0.6499,
      "step": 10421
    },
    {
      "epoch": 0.2744798525151435,
      "grad_norm": 2.5423145294189453,
      "learning_rate": 3.6285225177771925e-05,
      "loss": 1.5877,
      "step": 10422
    },
    {
      "epoch": 0.27450618909665525,
      "grad_norm": 2.2928049564361572,
      "learning_rate": 3.628390834869634e-05,
      "loss": 1.7479,
      "step": 10423
    },
    {
      "epoch": 0.274532525678167,
      "grad_norm": 1.8844287395477295,
      "learning_rate": 3.6282591519620757e-05,
      "loss": 2.2141,
      "step": 10424
    },
    {
      "epoch": 0.2745588622596787,
      "grad_norm": 1.8125797510147095,
      "learning_rate": 3.628127469054517e-05,
      "loss": 0.4839,
      "step": 10425
    },
    {
      "epoch": 0.2745851988411904,
      "grad_norm": 1.764673113822937,
      "learning_rate": 3.627995786146958e-05,
      "loss": 1.9221,
      "step": 10426
    },
    {
      "epoch": 0.2746115354227021,
      "grad_norm": 2.157508134841919,
      "learning_rate": 3.6278641032393997e-05,
      "loss": 1.8939,
      "step": 10427
    },
    {
      "epoch": 0.27463787200421386,
      "grad_norm": 3.0758440494537354,
      "learning_rate": 3.627732420331841e-05,
      "loss": 1.078,
      "step": 10428
    },
    {
      "epoch": 0.2746642085857256,
      "grad_norm": 1.9144856929779053,
      "learning_rate": 3.627600737424283e-05,
      "loss": 2.5218,
      "step": 10429
    },
    {
      "epoch": 0.2746905451672373,
      "grad_norm": 2.0303256511688232,
      "learning_rate": 3.6274690545167237e-05,
      "loss": 1.6567,
      "step": 10430
    },
    {
      "epoch": 0.27471688174874903,
      "grad_norm": 2.235045909881592,
      "learning_rate": 3.627337371609165e-05,
      "loss": 1.4615,
      "step": 10431
    },
    {
      "epoch": 0.2747432183302607,
      "grad_norm": 1.6566983461380005,
      "learning_rate": 3.627205688701607e-05,
      "loss": 0.9702,
      "step": 10432
    },
    {
      "epoch": 0.27476955491177246,
      "grad_norm": 2.245187759399414,
      "learning_rate": 3.627074005794048e-05,
      "loss": 1.8433,
      "step": 10433
    },
    {
      "epoch": 0.27479589149328415,
      "grad_norm": 1.9110386371612549,
      "learning_rate": 3.62694232288649e-05,
      "loss": 2.14,
      "step": 10434
    },
    {
      "epoch": 0.2748222280747959,
      "grad_norm": 2.1642751693725586,
      "learning_rate": 3.626810639978931e-05,
      "loss": 2.1116,
      "step": 10435
    },
    {
      "epoch": 0.27484856465630764,
      "grad_norm": 2.8699982166290283,
      "learning_rate": 3.626678957071372e-05,
      "loss": 1.4268,
      "step": 10436
    },
    {
      "epoch": 0.2748749012378193,
      "grad_norm": 1.994243860244751,
      "learning_rate": 3.626547274163813e-05,
      "loss": 1.571,
      "step": 10437
    },
    {
      "epoch": 0.27490123781933107,
      "grad_norm": 1.787546992301941,
      "learning_rate": 3.6264155912562555e-05,
      "loss": 1.9877,
      "step": 10438
    },
    {
      "epoch": 0.27492757440084276,
      "grad_norm": 3.318443536758423,
      "learning_rate": 3.626283908348696e-05,
      "loss": 0.7593,
      "step": 10439
    },
    {
      "epoch": 0.2749539109823545,
      "grad_norm": 2.943997859954834,
      "learning_rate": 3.626152225441138e-05,
      "loss": 1.3951,
      "step": 10440
    },
    {
      "epoch": 0.2749802475638662,
      "grad_norm": 5.653837203979492,
      "learning_rate": 3.6260205425335795e-05,
      "loss": 1.4294,
      "step": 10441
    },
    {
      "epoch": 0.27500658414537793,
      "grad_norm": 1.8154033422470093,
      "learning_rate": 3.625888859626021e-05,
      "loss": 1.8327,
      "step": 10442
    },
    {
      "epoch": 0.2750329207268897,
      "grad_norm": 2.034886598587036,
      "learning_rate": 3.6257571767184626e-05,
      "loss": 2.5932,
      "step": 10443
    },
    {
      "epoch": 0.27505925730840136,
      "grad_norm": 1.9043903350830078,
      "learning_rate": 3.6256254938109035e-05,
      "loss": 1.3506,
      "step": 10444
    },
    {
      "epoch": 0.2750855938899131,
      "grad_norm": 1.687850832939148,
      "learning_rate": 3.625493810903345e-05,
      "loss": 2.2564,
      "step": 10445
    },
    {
      "epoch": 0.2751119304714248,
      "grad_norm": 1.6645928621292114,
      "learning_rate": 3.625362127995786e-05,
      "loss": 0.9856,
      "step": 10446
    },
    {
      "epoch": 0.27513826705293654,
      "grad_norm": 1.674831509590149,
      "learning_rate": 3.625230445088228e-05,
      "loss": 2.3101,
      "step": 10447
    },
    {
      "epoch": 0.2751646036344482,
      "grad_norm": 1.8255071640014648,
      "learning_rate": 3.625098762180669e-05,
      "loss": 2.1872,
      "step": 10448
    },
    {
      "epoch": 0.27519094021595997,
      "grad_norm": 1.874585509300232,
      "learning_rate": 3.6249670792731106e-05,
      "loss": 1.8736,
      "step": 10449
    },
    {
      "epoch": 0.2752172767974717,
      "grad_norm": 2.40224027633667,
      "learning_rate": 3.624835396365552e-05,
      "loss": 1.2941,
      "step": 10450
    },
    {
      "epoch": 0.2752436133789834,
      "grad_norm": 6.703125,
      "learning_rate": 3.624703713457993e-05,
      "loss": 1.2435,
      "step": 10451
    },
    {
      "epoch": 0.27526994996049514,
      "grad_norm": 4.762519836425781,
      "learning_rate": 3.624572030550435e-05,
      "loss": 1.3299,
      "step": 10452
    },
    {
      "epoch": 0.27529628654200683,
      "grad_norm": 2.813340902328491,
      "learning_rate": 3.624440347642876e-05,
      "loss": 1.7707,
      "step": 10453
    },
    {
      "epoch": 0.2753226231235186,
      "grad_norm": 2.2296957969665527,
      "learning_rate": 3.624308664735318e-05,
      "loss": 1.7691,
      "step": 10454
    },
    {
      "epoch": 0.27534895970503026,
      "grad_norm": 3.706146001815796,
      "learning_rate": 3.6241769818277586e-05,
      "loss": 2.2148,
      "step": 10455
    },
    {
      "epoch": 0.275375296286542,
      "grad_norm": 1.4004148244857788,
      "learning_rate": 3.6240452989202e-05,
      "loss": 1.8777,
      "step": 10456
    },
    {
      "epoch": 0.27540163286805375,
      "grad_norm": 2.524167537689209,
      "learning_rate": 3.623913616012642e-05,
      "loss": 2.1514,
      "step": 10457
    },
    {
      "epoch": 0.27542796944956544,
      "grad_norm": 2.8889214992523193,
      "learning_rate": 3.623781933105083e-05,
      "loss": 1.1664,
      "step": 10458
    },
    {
      "epoch": 0.2754543060310772,
      "grad_norm": 1.6348000764846802,
      "learning_rate": 3.623650250197524e-05,
      "loss": 1.3697,
      "step": 10459
    },
    {
      "epoch": 0.27548064261258887,
      "grad_norm": 1.8383204936981201,
      "learning_rate": 3.623518567289966e-05,
      "loss": 1.7364,
      "step": 10460
    },
    {
      "epoch": 0.2755069791941006,
      "grad_norm": 1.8291422128677368,
      "learning_rate": 3.623386884382407e-05,
      "loss": 1.5751,
      "step": 10461
    },
    {
      "epoch": 0.27553331577561235,
      "grad_norm": 5.297059059143066,
      "learning_rate": 3.623255201474849e-05,
      "loss": 1.4437,
      "step": 10462
    },
    {
      "epoch": 0.27555965235712404,
      "grad_norm": 2.6070845127105713,
      "learning_rate": 3.6231235185672904e-05,
      "loss": 1.3462,
      "step": 10463
    },
    {
      "epoch": 0.2755859889386358,
      "grad_norm": 3.5472006797790527,
      "learning_rate": 3.622991835659731e-05,
      "loss": 0.9255,
      "step": 10464
    },
    {
      "epoch": 0.27561232552014747,
      "grad_norm": 2.4515955448150635,
      "learning_rate": 3.622860152752173e-05,
      "loss": 1.8441,
      "step": 10465
    },
    {
      "epoch": 0.2756386621016592,
      "grad_norm": 4.607582092285156,
      "learning_rate": 3.6227284698446144e-05,
      "loss": 1.3286,
      "step": 10466
    },
    {
      "epoch": 0.2756649986831709,
      "grad_norm": 1.7889039516448975,
      "learning_rate": 3.622596786937056e-05,
      "loss": 1.6902,
      "step": 10467
    },
    {
      "epoch": 0.27569133526468265,
      "grad_norm": 2.2339816093444824,
      "learning_rate": 3.622465104029497e-05,
      "loss": 1.9056,
      "step": 10468
    },
    {
      "epoch": 0.2757176718461944,
      "grad_norm": 1.987520456314087,
      "learning_rate": 3.6223334211219384e-05,
      "loss": 1.8437,
      "step": 10469
    },
    {
      "epoch": 0.2757440084277061,
      "grad_norm": 2.2537505626678467,
      "learning_rate": 3.62220173821438e-05,
      "loss": 1.652,
      "step": 10470
    },
    {
      "epoch": 0.2757703450092178,
      "grad_norm": 3.435530424118042,
      "learning_rate": 3.6220700553068215e-05,
      "loss": 0.672,
      "step": 10471
    },
    {
      "epoch": 0.2757966815907295,
      "grad_norm": 3.3695995807647705,
      "learning_rate": 3.621938372399263e-05,
      "loss": 1.9393,
      "step": 10472
    },
    {
      "epoch": 0.27582301817224125,
      "grad_norm": 3.0079569816589355,
      "learning_rate": 3.621806689491704e-05,
      "loss": 2.1282,
      "step": 10473
    },
    {
      "epoch": 0.27584935475375294,
      "grad_norm": 1.5797780752182007,
      "learning_rate": 3.6216750065841455e-05,
      "loss": 1.6722,
      "step": 10474
    },
    {
      "epoch": 0.2758756913352647,
      "grad_norm": 2.2939703464508057,
      "learning_rate": 3.621543323676587e-05,
      "loss": 1.8333,
      "step": 10475
    },
    {
      "epoch": 0.2759020279167764,
      "grad_norm": 1.8671458959579468,
      "learning_rate": 3.6214116407690286e-05,
      "loss": 1.8496,
      "step": 10476
    },
    {
      "epoch": 0.2759283644982881,
      "grad_norm": 2.9520442485809326,
      "learning_rate": 3.6212799578614695e-05,
      "loss": 1.837,
      "step": 10477
    },
    {
      "epoch": 0.27595470107979986,
      "grad_norm": 4.0135817527771,
      "learning_rate": 3.621148274953911e-05,
      "loss": 0.6895,
      "step": 10478
    },
    {
      "epoch": 0.27598103766131155,
      "grad_norm": 1.4256529808044434,
      "learning_rate": 3.6210165920463526e-05,
      "loss": 0.9646,
      "step": 10479
    },
    {
      "epoch": 0.2760073742428233,
      "grad_norm": 1.755150556564331,
      "learning_rate": 3.620884909138794e-05,
      "loss": 0.6211,
      "step": 10480
    },
    {
      "epoch": 0.276033710824335,
      "grad_norm": 3.8370206356048584,
      "learning_rate": 3.620753226231236e-05,
      "loss": 1.8618,
      "step": 10481
    },
    {
      "epoch": 0.2760600474058467,
      "grad_norm": 2.5078601837158203,
      "learning_rate": 3.6206215433236766e-05,
      "loss": 1.7979,
      "step": 10482
    },
    {
      "epoch": 0.27608638398735846,
      "grad_norm": 1.6675509214401245,
      "learning_rate": 3.620489860416118e-05,
      "loss": 1.9262,
      "step": 10483
    },
    {
      "epoch": 0.27611272056887015,
      "grad_norm": 3.0252974033355713,
      "learning_rate": 3.620358177508559e-05,
      "loss": 1.7034,
      "step": 10484
    },
    {
      "epoch": 0.2761390571503819,
      "grad_norm": 1.6886072158813477,
      "learning_rate": 3.620226494601001e-05,
      "loss": 1.5362,
      "step": 10485
    },
    {
      "epoch": 0.2761653937318936,
      "grad_norm": 3.0428547859191895,
      "learning_rate": 3.620094811693442e-05,
      "loss": 1.7356,
      "step": 10486
    },
    {
      "epoch": 0.2761917303134053,
      "grad_norm": 1.7481582164764404,
      "learning_rate": 3.619963128785884e-05,
      "loss": 2.1848,
      "step": 10487
    },
    {
      "epoch": 0.276218066894917,
      "grad_norm": 2.8296101093292236,
      "learning_rate": 3.619831445878325e-05,
      "loss": 2.1159,
      "step": 10488
    },
    {
      "epoch": 0.27624440347642876,
      "grad_norm": 2.57230544090271,
      "learning_rate": 3.619699762970766e-05,
      "loss": 0.7766,
      "step": 10489
    },
    {
      "epoch": 0.2762707400579405,
      "grad_norm": 1.5827943086624146,
      "learning_rate": 3.6195680800632084e-05,
      "loss": 1.4311,
      "step": 10490
    },
    {
      "epoch": 0.2762970766394522,
      "grad_norm": 2.5470101833343506,
      "learning_rate": 3.619436397155649e-05,
      "loss": 0.6682,
      "step": 10491
    },
    {
      "epoch": 0.27632341322096393,
      "grad_norm": 1.8786563873291016,
      "learning_rate": 3.619304714248091e-05,
      "loss": 1.3432,
      "step": 10492
    },
    {
      "epoch": 0.2763497498024756,
      "grad_norm": 1.6635288000106812,
      "learning_rate": 3.619173031340532e-05,
      "loss": 1.8164,
      "step": 10493
    },
    {
      "epoch": 0.27637608638398736,
      "grad_norm": 3.3679616451263428,
      "learning_rate": 3.619041348432974e-05,
      "loss": 1.8374,
      "step": 10494
    },
    {
      "epoch": 0.27640242296549905,
      "grad_norm": 1.6528295278549194,
      "learning_rate": 3.618909665525415e-05,
      "loss": 1.8446,
      "step": 10495
    },
    {
      "epoch": 0.2764287595470108,
      "grad_norm": 2.233024835586548,
      "learning_rate": 3.6187779826178564e-05,
      "loss": 1.4532,
      "step": 10496
    },
    {
      "epoch": 0.27645509612852254,
      "grad_norm": 2.285409450531006,
      "learning_rate": 3.618646299710298e-05,
      "loss": 0.3484,
      "step": 10497
    },
    {
      "epoch": 0.2764814327100342,
      "grad_norm": 2.079171895980835,
      "learning_rate": 3.618514616802739e-05,
      "loss": 2.2279,
      "step": 10498
    },
    {
      "epoch": 0.27650776929154597,
      "grad_norm": 1.615371584892273,
      "learning_rate": 3.618382933895181e-05,
      "loss": 1.7252,
      "step": 10499
    },
    {
      "epoch": 0.27653410587305766,
      "grad_norm": 2.8617053031921387,
      "learning_rate": 3.618251250987622e-05,
      "loss": 1.9224,
      "step": 10500
    },
    {
      "epoch": 0.2765604424545694,
      "grad_norm": 2.6724143028259277,
      "learning_rate": 3.6181195680800636e-05,
      "loss": 1.9771,
      "step": 10501
    },
    {
      "epoch": 0.27658677903608114,
      "grad_norm": 2.234105110168457,
      "learning_rate": 3.6179878851725044e-05,
      "loss": 2.2792,
      "step": 10502
    },
    {
      "epoch": 0.27661311561759283,
      "grad_norm": 2.2478859424591064,
      "learning_rate": 3.617856202264946e-05,
      "loss": 2.338,
      "step": 10503
    },
    {
      "epoch": 0.2766394521991046,
      "grad_norm": 3.4104225635528564,
      "learning_rate": 3.6177245193573876e-05,
      "loss": 1.8354,
      "step": 10504
    },
    {
      "epoch": 0.27666578878061626,
      "grad_norm": 1.7714762687683105,
      "learning_rate": 3.617592836449829e-05,
      "loss": 1.8856,
      "step": 10505
    },
    {
      "epoch": 0.276692125362128,
      "grad_norm": 3.2018375396728516,
      "learning_rate": 3.61746115354227e-05,
      "loss": 0.7587,
      "step": 10506
    },
    {
      "epoch": 0.2767184619436397,
      "grad_norm": 2.6129369735717773,
      "learning_rate": 3.6173294706347116e-05,
      "loss": 1.5104,
      "step": 10507
    },
    {
      "epoch": 0.27674479852515144,
      "grad_norm": 1.854946494102478,
      "learning_rate": 3.617197787727153e-05,
      "loss": 1.8627,
      "step": 10508
    },
    {
      "epoch": 0.2767711351066632,
      "grad_norm": 1.9550607204437256,
      "learning_rate": 3.617066104819595e-05,
      "loss": 1.4634,
      "step": 10509
    },
    {
      "epoch": 0.27679747168817487,
      "grad_norm": 1.94491446018219,
      "learning_rate": 3.616934421912036e-05,
      "loss": 1.342,
      "step": 10510
    },
    {
      "epoch": 0.2768238082696866,
      "grad_norm": 2.0520408153533936,
      "learning_rate": 3.616802739004477e-05,
      "loss": 0.7579,
      "step": 10511
    },
    {
      "epoch": 0.2768501448511983,
      "grad_norm": 1.580329418182373,
      "learning_rate": 3.616671056096919e-05,
      "loss": 2.1161,
      "step": 10512
    },
    {
      "epoch": 0.27687648143271004,
      "grad_norm": 9.892873764038086,
      "learning_rate": 3.61653937318936e-05,
      "loss": 1.7196,
      "step": 10513
    },
    {
      "epoch": 0.27690281801422173,
      "grad_norm": 2.776784658432007,
      "learning_rate": 3.616407690281802e-05,
      "loss": 1.2163,
      "step": 10514
    },
    {
      "epoch": 0.2769291545957335,
      "grad_norm": 2.1026298999786377,
      "learning_rate": 3.616276007374243e-05,
      "loss": 0.7488,
      "step": 10515
    },
    {
      "epoch": 0.2769554911772452,
      "grad_norm": 3.374598741531372,
      "learning_rate": 3.616144324466684e-05,
      "loss": 1.1478,
      "step": 10516
    },
    {
      "epoch": 0.2769818277587569,
      "grad_norm": 2.0458824634552,
      "learning_rate": 3.616012641559126e-05,
      "loss": 0.8614,
      "step": 10517
    },
    {
      "epoch": 0.27700816434026865,
      "grad_norm": 2.6714377403259277,
      "learning_rate": 3.6158809586515674e-05,
      "loss": 0.7945,
      "step": 10518
    },
    {
      "epoch": 0.27703450092178034,
      "grad_norm": 3.6527814865112305,
      "learning_rate": 3.615749275744009e-05,
      "loss": 1.4755,
      "step": 10519
    },
    {
      "epoch": 0.2770608375032921,
      "grad_norm": 1.7627272605895996,
      "learning_rate": 3.61561759283645e-05,
      "loss": 1.8425,
      "step": 10520
    },
    {
      "epoch": 0.27708717408480377,
      "grad_norm": 2.1502063274383545,
      "learning_rate": 3.6154859099288914e-05,
      "loss": 1.6817,
      "step": 10521
    },
    {
      "epoch": 0.2771135106663155,
      "grad_norm": 3.362943172454834,
      "learning_rate": 3.615354227021332e-05,
      "loss": 1.3013,
      "step": 10522
    },
    {
      "epoch": 0.27713984724782725,
      "grad_norm": 2.4210638999938965,
      "learning_rate": 3.6152225441137745e-05,
      "loss": 1.6636,
      "step": 10523
    },
    {
      "epoch": 0.27716618382933894,
      "grad_norm": 2.529226064682007,
      "learning_rate": 3.6150908612062154e-05,
      "loss": 0.6731,
      "step": 10524
    },
    {
      "epoch": 0.2771925204108507,
      "grad_norm": 3.0127105712890625,
      "learning_rate": 3.614959178298657e-05,
      "loss": 1.0777,
      "step": 10525
    },
    {
      "epoch": 0.2772188569923624,
      "grad_norm": 3.4717442989349365,
      "learning_rate": 3.6148274953910985e-05,
      "loss": 1.0245,
      "step": 10526
    },
    {
      "epoch": 0.2772451935738741,
      "grad_norm": 3.9940237998962402,
      "learning_rate": 3.61469581248354e-05,
      "loss": 1.9003,
      "step": 10527
    },
    {
      "epoch": 0.2772715301553858,
      "grad_norm": 5.008260250091553,
      "learning_rate": 3.6145641295759816e-05,
      "loss": 1.633,
      "step": 10528
    },
    {
      "epoch": 0.27729786673689755,
      "grad_norm": 2.702284336090088,
      "learning_rate": 3.6144324466684225e-05,
      "loss": 1.9457,
      "step": 10529
    },
    {
      "epoch": 0.2773242033184093,
      "grad_norm": 1.7166951894760132,
      "learning_rate": 3.614300763760864e-05,
      "loss": 2.2659,
      "step": 10530
    },
    {
      "epoch": 0.277350539899921,
      "grad_norm": 1.9460575580596924,
      "learning_rate": 3.614169080853305e-05,
      "loss": 1.7085,
      "step": 10531
    },
    {
      "epoch": 0.2773768764814327,
      "grad_norm": 4.915445327758789,
      "learning_rate": 3.614037397945747e-05,
      "loss": 2.1869,
      "step": 10532
    },
    {
      "epoch": 0.2774032130629444,
      "grad_norm": 1.7240326404571533,
      "learning_rate": 3.613905715038188e-05,
      "loss": 1.8895,
      "step": 10533
    },
    {
      "epoch": 0.27742954964445615,
      "grad_norm": 2.6020450592041016,
      "learning_rate": 3.6137740321306296e-05,
      "loss": 1.6892,
      "step": 10534
    },
    {
      "epoch": 0.2774558862259679,
      "grad_norm": 3.6185905933380127,
      "learning_rate": 3.613642349223071e-05,
      "loss": 1.4694,
      "step": 10535
    },
    {
      "epoch": 0.2774822228074796,
      "grad_norm": 1.8450874090194702,
      "learning_rate": 3.613510666315512e-05,
      "loss": 1.9159,
      "step": 10536
    },
    {
      "epoch": 0.2775085593889913,
      "grad_norm": 2.4880099296569824,
      "learning_rate": 3.613378983407954e-05,
      "loss": 2.2643,
      "step": 10537
    },
    {
      "epoch": 0.277534895970503,
      "grad_norm": 1.7124220132827759,
      "learning_rate": 3.613247300500395e-05,
      "loss": 1.8497,
      "step": 10538
    },
    {
      "epoch": 0.27756123255201476,
      "grad_norm": 1.8947703838348389,
      "learning_rate": 3.613115617592837e-05,
      "loss": 1.4309,
      "step": 10539
    },
    {
      "epoch": 0.27758756913352645,
      "grad_norm": 1.5899467468261719,
      "learning_rate": 3.6129839346852776e-05,
      "loss": 1.5447,
      "step": 10540
    },
    {
      "epoch": 0.2776139057150382,
      "grad_norm": 1.6295113563537598,
      "learning_rate": 3.61285225177772e-05,
      "loss": 2.3948,
      "step": 10541
    },
    {
      "epoch": 0.27764024229654993,
      "grad_norm": 2.512416362762451,
      "learning_rate": 3.612720568870161e-05,
      "loss": 2.3674,
      "step": 10542
    },
    {
      "epoch": 0.2776665788780616,
      "grad_norm": 2.4679908752441406,
      "learning_rate": 3.612588885962602e-05,
      "loss": 1.2602,
      "step": 10543
    },
    {
      "epoch": 0.27769291545957336,
      "grad_norm": 1.6646476984024048,
      "learning_rate": 3.612457203055044e-05,
      "loss": 1.6468,
      "step": 10544
    },
    {
      "epoch": 0.27771925204108505,
      "grad_norm": 4.851602077484131,
      "learning_rate": 3.612325520147485e-05,
      "loss": 1.2795,
      "step": 10545
    },
    {
      "epoch": 0.2777455886225968,
      "grad_norm": 2.18587327003479,
      "learning_rate": 3.612193837239927e-05,
      "loss": 1.4798,
      "step": 10546
    },
    {
      "epoch": 0.2777719252041085,
      "grad_norm": 2.0642988681793213,
      "learning_rate": 3.612062154332368e-05,
      "loss": 1.6178,
      "step": 10547
    },
    {
      "epoch": 0.2777982617856202,
      "grad_norm": 1.6430569887161255,
      "learning_rate": 3.6119304714248094e-05,
      "loss": 2.2333,
      "step": 10548
    },
    {
      "epoch": 0.27782459836713197,
      "grad_norm": 2.343280076980591,
      "learning_rate": 3.61179878851725e-05,
      "loss": 1.7574,
      "step": 10549
    },
    {
      "epoch": 0.27785093494864366,
      "grad_norm": 2.0710694789886475,
      "learning_rate": 3.611667105609692e-05,
      "loss": 2.4,
      "step": 10550
    },
    {
      "epoch": 0.2778772715301554,
      "grad_norm": 2.84663987159729,
      "learning_rate": 3.6115354227021334e-05,
      "loss": 0.4437,
      "step": 10551
    },
    {
      "epoch": 0.2779036081116671,
      "grad_norm": 2.0948305130004883,
      "learning_rate": 3.611403739794575e-05,
      "loss": 1.502,
      "step": 10552
    },
    {
      "epoch": 0.27792994469317883,
      "grad_norm": 4.115591049194336,
      "learning_rate": 3.6112720568870165e-05,
      "loss": 1.327,
      "step": 10553
    },
    {
      "epoch": 0.2779562812746905,
      "grad_norm": 2.811091184616089,
      "learning_rate": 3.6111403739794574e-05,
      "loss": 1.2381,
      "step": 10554
    },
    {
      "epoch": 0.27798261785620226,
      "grad_norm": 2.0517990589141846,
      "learning_rate": 3.611008691071899e-05,
      "loss": 1.5499,
      "step": 10555
    },
    {
      "epoch": 0.278008954437714,
      "grad_norm": 1.8026422262191772,
      "learning_rate": 3.6108770081643405e-05,
      "loss": 1.6637,
      "step": 10556
    },
    {
      "epoch": 0.2780352910192257,
      "grad_norm": 2.157827138900757,
      "learning_rate": 3.610745325256782e-05,
      "loss": 1.8581,
      "step": 10557
    },
    {
      "epoch": 0.27806162760073744,
      "grad_norm": 1.9836013317108154,
      "learning_rate": 3.610613642349223e-05,
      "loss": 1.9057,
      "step": 10558
    },
    {
      "epoch": 0.2780879641822491,
      "grad_norm": 1.6347705125808716,
      "learning_rate": 3.6104819594416645e-05,
      "loss": 1.8962,
      "step": 10559
    },
    {
      "epoch": 0.27811430076376087,
      "grad_norm": 2.470343589782715,
      "learning_rate": 3.610350276534106e-05,
      "loss": 1.8694,
      "step": 10560
    },
    {
      "epoch": 0.27814063734527256,
      "grad_norm": 1.6528599262237549,
      "learning_rate": 3.6102185936265477e-05,
      "loss": 2.3773,
      "step": 10561
    },
    {
      "epoch": 0.2781669739267843,
      "grad_norm": 1.9485795497894287,
      "learning_rate": 3.6100869107189885e-05,
      "loss": 1.8891,
      "step": 10562
    },
    {
      "epoch": 0.27819331050829604,
      "grad_norm": 3.1197283267974854,
      "learning_rate": 3.60995522781143e-05,
      "loss": 1.4587,
      "step": 10563
    },
    {
      "epoch": 0.27821964708980773,
      "grad_norm": 1.660825252532959,
      "learning_rate": 3.6098235449038717e-05,
      "loss": 1.7337,
      "step": 10564
    },
    {
      "epoch": 0.2782459836713195,
      "grad_norm": 5.408761501312256,
      "learning_rate": 3.609691861996313e-05,
      "loss": 1.3651,
      "step": 10565
    },
    {
      "epoch": 0.27827232025283116,
      "grad_norm": 2.2877254486083984,
      "learning_rate": 3.609560179088755e-05,
      "loss": 0.9732,
      "step": 10566
    },
    {
      "epoch": 0.2782986568343429,
      "grad_norm": 2.5487403869628906,
      "learning_rate": 3.6094284961811957e-05,
      "loss": 1.9014,
      "step": 10567
    },
    {
      "epoch": 0.27832499341585465,
      "grad_norm": 2.2451066970825195,
      "learning_rate": 3.609296813273637e-05,
      "loss": 1.8042,
      "step": 10568
    },
    {
      "epoch": 0.27835132999736634,
      "grad_norm": 2.410404920578003,
      "learning_rate": 3.609165130366078e-05,
      "loss": 1.6118,
      "step": 10569
    },
    {
      "epoch": 0.2783776665788781,
      "grad_norm": 1.6133015155792236,
      "learning_rate": 3.6090334474585203e-05,
      "loss": 0.547,
      "step": 10570
    },
    {
      "epoch": 0.27840400316038977,
      "grad_norm": 3.749532699584961,
      "learning_rate": 3.608901764550961e-05,
      "loss": 0.6882,
      "step": 10571
    },
    {
      "epoch": 0.2784303397419015,
      "grad_norm": 3.7434544563293457,
      "learning_rate": 3.608770081643403e-05,
      "loss": 1.4323,
      "step": 10572
    },
    {
      "epoch": 0.2784566763234132,
      "grad_norm": 1.835457444190979,
      "learning_rate": 3.6086383987358443e-05,
      "loss": 1.9915,
      "step": 10573
    },
    {
      "epoch": 0.27848301290492494,
      "grad_norm": 3.5884530544281006,
      "learning_rate": 3.608506715828286e-05,
      "loss": 1.5782,
      "step": 10574
    },
    {
      "epoch": 0.2785093494864367,
      "grad_norm": 1.5033602714538574,
      "learning_rate": 3.6083750329207275e-05,
      "loss": 0.2876,
      "step": 10575
    },
    {
      "epoch": 0.2785356860679484,
      "grad_norm": 1.9221035242080688,
      "learning_rate": 3.6082433500131683e-05,
      "loss": 2.6736,
      "step": 10576
    },
    {
      "epoch": 0.2785620226494601,
      "grad_norm": 3.4220848083496094,
      "learning_rate": 3.60811166710561e-05,
      "loss": 1.6724,
      "step": 10577
    },
    {
      "epoch": 0.2785883592309718,
      "grad_norm": 4.6325578689575195,
      "learning_rate": 3.607979984198051e-05,
      "loss": 1.0641,
      "step": 10578
    },
    {
      "epoch": 0.27861469581248355,
      "grad_norm": 2.0484626293182373,
      "learning_rate": 3.607848301290493e-05,
      "loss": 1.754,
      "step": 10579
    },
    {
      "epoch": 0.27864103239399524,
      "grad_norm": 1.7742486000061035,
      "learning_rate": 3.607716618382934e-05,
      "loss": 1.252,
      "step": 10580
    },
    {
      "epoch": 0.278667368975507,
      "grad_norm": 3.1879234313964844,
      "learning_rate": 3.6075849354753755e-05,
      "loss": 1.6456,
      "step": 10581
    },
    {
      "epoch": 0.2786937055570187,
      "grad_norm": 1.864965558052063,
      "learning_rate": 3.607453252567817e-05,
      "loss": 2.0505,
      "step": 10582
    },
    {
      "epoch": 0.2787200421385304,
      "grad_norm": 1.4988855123519897,
      "learning_rate": 3.607321569660258e-05,
      "loss": 1.4652,
      "step": 10583
    },
    {
      "epoch": 0.27874637872004215,
      "grad_norm": 1.9227646589279175,
      "learning_rate": 3.6071898867527e-05,
      "loss": 1.1357,
      "step": 10584
    },
    {
      "epoch": 0.27877271530155384,
      "grad_norm": 2.266540050506592,
      "learning_rate": 3.607058203845141e-05,
      "loss": 0.6786,
      "step": 10585
    },
    {
      "epoch": 0.2787990518830656,
      "grad_norm": 2.252044200897217,
      "learning_rate": 3.6069265209375826e-05,
      "loss": 1.9835,
      "step": 10586
    },
    {
      "epoch": 0.2788253884645773,
      "grad_norm": 1.7922183275222778,
      "learning_rate": 3.6067948380300235e-05,
      "loss": 1.6849,
      "step": 10587
    },
    {
      "epoch": 0.278851725046089,
      "grad_norm": 2.043391466140747,
      "learning_rate": 3.606663155122466e-05,
      "loss": 1.8441,
      "step": 10588
    },
    {
      "epoch": 0.27887806162760076,
      "grad_norm": 1.9932903051376343,
      "learning_rate": 3.6065314722149066e-05,
      "loss": 1.1982,
      "step": 10589
    },
    {
      "epoch": 0.27890439820911245,
      "grad_norm": 3.703014373779297,
      "learning_rate": 3.606399789307348e-05,
      "loss": 0.6268,
      "step": 10590
    },
    {
      "epoch": 0.2789307347906242,
      "grad_norm": 3.817866325378418,
      "learning_rate": 3.60626810639979e-05,
      "loss": 2.0868,
      "step": 10591
    },
    {
      "epoch": 0.2789570713721359,
      "grad_norm": 2.0294957160949707,
      "learning_rate": 3.6061364234922306e-05,
      "loss": 2.2436,
      "step": 10592
    },
    {
      "epoch": 0.2789834079536476,
      "grad_norm": 1.622313380241394,
      "learning_rate": 3.606004740584673e-05,
      "loss": 0.5224,
      "step": 10593
    },
    {
      "epoch": 0.2790097445351593,
      "grad_norm": 1.5825412273406982,
      "learning_rate": 3.605873057677114e-05,
      "loss": 2.2051,
      "step": 10594
    },
    {
      "epoch": 0.27903608111667105,
      "grad_norm": 1.7760380506515503,
      "learning_rate": 3.605741374769555e-05,
      "loss": 1.7089,
      "step": 10595
    },
    {
      "epoch": 0.2790624176981828,
      "grad_norm": 6.593071937561035,
      "learning_rate": 3.605609691861996e-05,
      "loss": 1.1579,
      "step": 10596
    },
    {
      "epoch": 0.2790887542796945,
      "grad_norm": 2.3868892192840576,
      "learning_rate": 3.605478008954438e-05,
      "loss": 1.9262,
      "step": 10597
    },
    {
      "epoch": 0.2791150908612062,
      "grad_norm": 1.863279938697815,
      "learning_rate": 3.605346326046879e-05,
      "loss": 1.3981,
      "step": 10598
    },
    {
      "epoch": 0.2791414274427179,
      "grad_norm": 2.025031805038452,
      "learning_rate": 3.605214643139321e-05,
      "loss": 2.6291,
      "step": 10599
    },
    {
      "epoch": 0.27916776402422966,
      "grad_norm": 2.297893762588501,
      "learning_rate": 3.6050829602317624e-05,
      "loss": 1.5302,
      "step": 10600
    },
    {
      "epoch": 0.2791941006057414,
      "grad_norm": 3.0263586044311523,
      "learning_rate": 3.604951277324203e-05,
      "loss": 1.314,
      "step": 10601
    },
    {
      "epoch": 0.2792204371872531,
      "grad_norm": 1.8723514080047607,
      "learning_rate": 3.604819594416645e-05,
      "loss": 1.707,
      "step": 10602
    },
    {
      "epoch": 0.27924677376876483,
      "grad_norm": 4.323929786682129,
      "learning_rate": 3.6046879115090864e-05,
      "loss": 1.8483,
      "step": 10603
    },
    {
      "epoch": 0.2792731103502765,
      "grad_norm": 4.868648052215576,
      "learning_rate": 3.604556228601528e-05,
      "loss": 1.5361,
      "step": 10604
    },
    {
      "epoch": 0.27929944693178826,
      "grad_norm": 1.6330366134643555,
      "learning_rate": 3.604424545693969e-05,
      "loss": 1.955,
      "step": 10605
    },
    {
      "epoch": 0.27932578351329995,
      "grad_norm": 1.8268793821334839,
      "learning_rate": 3.6042928627864104e-05,
      "loss": 2.5499,
      "step": 10606
    },
    {
      "epoch": 0.2793521200948117,
      "grad_norm": 1.7831532955169678,
      "learning_rate": 3.604161179878852e-05,
      "loss": 1.7541,
      "step": 10607
    },
    {
      "epoch": 0.27937845667632344,
      "grad_norm": 1.713757038116455,
      "learning_rate": 3.6040294969712935e-05,
      "loss": 1.8245,
      "step": 10608
    },
    {
      "epoch": 0.2794047932578351,
      "grad_norm": 3.6432838439941406,
      "learning_rate": 3.6038978140637344e-05,
      "loss": 1.1547,
      "step": 10609
    },
    {
      "epoch": 0.27943112983934687,
      "grad_norm": 2.1600749492645264,
      "learning_rate": 3.603766131156176e-05,
      "loss": 1.5969,
      "step": 10610
    },
    {
      "epoch": 0.27945746642085856,
      "grad_norm": 2.707930564880371,
      "learning_rate": 3.6036344482486175e-05,
      "loss": 2.5188,
      "step": 10611
    },
    {
      "epoch": 0.2794838030023703,
      "grad_norm": 2.9858036041259766,
      "learning_rate": 3.603502765341059e-05,
      "loss": 1.1827,
      "step": 10612
    },
    {
      "epoch": 0.279510139583882,
      "grad_norm": 1.812178373336792,
      "learning_rate": 3.6033710824335006e-05,
      "loss": 1.9033,
      "step": 10613
    },
    {
      "epoch": 0.27953647616539373,
      "grad_norm": 1.4573904275894165,
      "learning_rate": 3.6032393995259415e-05,
      "loss": 1.8184,
      "step": 10614
    },
    {
      "epoch": 0.2795628127469055,
      "grad_norm": 2.8986144065856934,
      "learning_rate": 3.603107716618383e-05,
      "loss": 1.4632,
      "step": 10615
    },
    {
      "epoch": 0.27958914932841716,
      "grad_norm": 3.6511080265045166,
      "learning_rate": 3.602976033710824e-05,
      "loss": 0.9638,
      "step": 10616
    },
    {
      "epoch": 0.2796154859099289,
      "grad_norm": 1.5710798501968384,
      "learning_rate": 3.602844350803266e-05,
      "loss": 2.1146,
      "step": 10617
    },
    {
      "epoch": 0.2796418224914406,
      "grad_norm": 4.432260513305664,
      "learning_rate": 3.602712667895707e-05,
      "loss": 1.0949,
      "step": 10618
    },
    {
      "epoch": 0.27966815907295234,
      "grad_norm": 2.828469753265381,
      "learning_rate": 3.6025809849881486e-05,
      "loss": 1.5586,
      "step": 10619
    },
    {
      "epoch": 0.279694495654464,
      "grad_norm": 1.6373722553253174,
      "learning_rate": 3.60244930208059e-05,
      "loss": 1.9336,
      "step": 10620
    },
    {
      "epoch": 0.27972083223597577,
      "grad_norm": 2.660008430480957,
      "learning_rate": 3.602317619173032e-05,
      "loss": 2.4598,
      "step": 10621
    },
    {
      "epoch": 0.2797471688174875,
      "grad_norm": 2.2494709491729736,
      "learning_rate": 3.602185936265473e-05,
      "loss": 1.9613,
      "step": 10622
    },
    {
      "epoch": 0.2797735053989992,
      "grad_norm": 3.685098886489868,
      "learning_rate": 3.602054253357914e-05,
      "loss": 1.0617,
      "step": 10623
    },
    {
      "epoch": 0.27979984198051094,
      "grad_norm": 1.8566356897354126,
      "learning_rate": 3.601922570450356e-05,
      "loss": 1.7472,
      "step": 10624
    },
    {
      "epoch": 0.27982617856202263,
      "grad_norm": 1.5898544788360596,
      "learning_rate": 3.6017908875427966e-05,
      "loss": 1.6229,
      "step": 10625
    },
    {
      "epoch": 0.2798525151435344,
      "grad_norm": 3.930208921432495,
      "learning_rate": 3.601659204635239e-05,
      "loss": 1.2664,
      "step": 10626
    },
    {
      "epoch": 0.27987885172504606,
      "grad_norm": 2.7435531616210938,
      "learning_rate": 3.60152752172768e-05,
      "loss": 0.605,
      "step": 10627
    },
    {
      "epoch": 0.2799051883065578,
      "grad_norm": 1.8299710750579834,
      "learning_rate": 3.601395838820121e-05,
      "loss": 1.8009,
      "step": 10628
    },
    {
      "epoch": 0.27993152488806955,
      "grad_norm": 4.164587497711182,
      "learning_rate": 3.601264155912563e-05,
      "loss": 1.8989,
      "step": 10629
    },
    {
      "epoch": 0.27995786146958124,
      "grad_norm": 3.663940668106079,
      "learning_rate": 3.601132473005004e-05,
      "loss": 2.3288,
      "step": 10630
    },
    {
      "epoch": 0.279984198051093,
      "grad_norm": 1.454330325126648,
      "learning_rate": 3.601000790097446e-05,
      "loss": 1.584,
      "step": 10631
    },
    {
      "epoch": 0.28001053463260467,
      "grad_norm": 2.1807925701141357,
      "learning_rate": 3.600869107189887e-05,
      "loss": 2.5548,
      "step": 10632
    },
    {
      "epoch": 0.2800368712141164,
      "grad_norm": 1.6666271686553955,
      "learning_rate": 3.6007374242823284e-05,
      "loss": 1.5791,
      "step": 10633
    },
    {
      "epoch": 0.28006320779562816,
      "grad_norm": 2.1405491828918457,
      "learning_rate": 3.600605741374769e-05,
      "loss": 1.87,
      "step": 10634
    },
    {
      "epoch": 0.28008954437713984,
      "grad_norm": 2.001499652862549,
      "learning_rate": 3.600474058467211e-05,
      "loss": 1.5485,
      "step": 10635
    },
    {
      "epoch": 0.2801158809586516,
      "grad_norm": 3.800605535507202,
      "learning_rate": 3.6003423755596524e-05,
      "loss": 1.2924,
      "step": 10636
    },
    {
      "epoch": 0.2801422175401633,
      "grad_norm": 1.8137459754943848,
      "learning_rate": 3.600210692652094e-05,
      "loss": 1.8019,
      "step": 10637
    },
    {
      "epoch": 0.280168554121675,
      "grad_norm": 4.043032646179199,
      "learning_rate": 3.6000790097445356e-05,
      "loss": 1.1812,
      "step": 10638
    },
    {
      "epoch": 0.2801948907031867,
      "grad_norm": 2.0625429153442383,
      "learning_rate": 3.5999473268369764e-05,
      "loss": 0.9757,
      "step": 10639
    },
    {
      "epoch": 0.28022122728469845,
      "grad_norm": 2.455791711807251,
      "learning_rate": 3.599815643929419e-05,
      "loss": 0.8719,
      "step": 10640
    },
    {
      "epoch": 0.2802475638662102,
      "grad_norm": 2.3219807147979736,
      "learning_rate": 3.5996839610218596e-05,
      "loss": 0.5165,
      "step": 10641
    },
    {
      "epoch": 0.2802739004477219,
      "grad_norm": 1.8683165311813354,
      "learning_rate": 3.599552278114301e-05,
      "loss": 0.33,
      "step": 10642
    },
    {
      "epoch": 0.2803002370292336,
      "grad_norm": 1.7563034296035767,
      "learning_rate": 3.599420595206742e-05,
      "loss": 1.75,
      "step": 10643
    },
    {
      "epoch": 0.2803265736107453,
      "grad_norm": 3.0634682178497314,
      "learning_rate": 3.5992889122991836e-05,
      "loss": 1.8466,
      "step": 10644
    },
    {
      "epoch": 0.28035291019225705,
      "grad_norm": 1.5442073345184326,
      "learning_rate": 3.599157229391625e-05,
      "loss": 2.156,
      "step": 10645
    },
    {
      "epoch": 0.28037924677376874,
      "grad_norm": 1.9861361980438232,
      "learning_rate": 3.599025546484067e-05,
      "loss": 1.5096,
      "step": 10646
    },
    {
      "epoch": 0.2804055833552805,
      "grad_norm": 3.358288288116455,
      "learning_rate": 3.598893863576508e-05,
      "loss": 1.0369,
      "step": 10647
    },
    {
      "epoch": 0.28043191993679223,
      "grad_norm": 2.8081891536712646,
      "learning_rate": 3.598762180668949e-05,
      "loss": 1.0704,
      "step": 10648
    },
    {
      "epoch": 0.2804582565183039,
      "grad_norm": 1.6818207502365112,
      "learning_rate": 3.598630497761391e-05,
      "loss": 2.0621,
      "step": 10649
    },
    {
      "epoch": 0.28048459309981566,
      "grad_norm": 2.4079747200012207,
      "learning_rate": 3.598498814853832e-05,
      "loss": 1.6631,
      "step": 10650
    },
    {
      "epoch": 0.28051092968132735,
      "grad_norm": 3.6856985092163086,
      "learning_rate": 3.598367131946274e-05,
      "loss": 2.029,
      "step": 10651
    },
    {
      "epoch": 0.2805372662628391,
      "grad_norm": 2.1202104091644287,
      "learning_rate": 3.598235449038715e-05,
      "loss": 1.9108,
      "step": 10652
    },
    {
      "epoch": 0.2805636028443508,
      "grad_norm": 7.725017547607422,
      "learning_rate": 3.598103766131156e-05,
      "loss": 1.2323,
      "step": 10653
    },
    {
      "epoch": 0.2805899394258625,
      "grad_norm": 1.7582473754882812,
      "learning_rate": 3.597972083223598e-05,
      "loss": 1.3198,
      "step": 10654
    },
    {
      "epoch": 0.28061627600737427,
      "grad_norm": 2.8983075618743896,
      "learning_rate": 3.5978404003160394e-05,
      "loss": 1.9464,
      "step": 10655
    },
    {
      "epoch": 0.28064261258888595,
      "grad_norm": 1.5993150472640991,
      "learning_rate": 3.597708717408481e-05,
      "loss": 2.208,
      "step": 10656
    },
    {
      "epoch": 0.2806689491703977,
      "grad_norm": 1.9493886232376099,
      "learning_rate": 3.597577034500922e-05,
      "loss": 1.7803,
      "step": 10657
    },
    {
      "epoch": 0.2806952857519094,
      "grad_norm": 1.8038997650146484,
      "learning_rate": 3.5974453515933634e-05,
      "loss": 2.0923,
      "step": 10658
    },
    {
      "epoch": 0.28072162233342113,
      "grad_norm": 3.5540947914123535,
      "learning_rate": 3.597313668685805e-05,
      "loss": 0.386,
      "step": 10659
    },
    {
      "epoch": 0.2807479589149328,
      "grad_norm": 2.0793118476867676,
      "learning_rate": 3.5971819857782465e-05,
      "loss": 1.6538,
      "step": 10660
    },
    {
      "epoch": 0.28077429549644456,
      "grad_norm": 1.7483510971069336,
      "learning_rate": 3.5970503028706874e-05,
      "loss": 1.9041,
      "step": 10661
    },
    {
      "epoch": 0.2808006320779563,
      "grad_norm": 6.435232639312744,
      "learning_rate": 3.596918619963129e-05,
      "loss": 1.3221,
      "step": 10662
    },
    {
      "epoch": 0.280826968659468,
      "grad_norm": 2.415607213973999,
      "learning_rate": 3.59678693705557e-05,
      "loss": 1.6288,
      "step": 10663
    },
    {
      "epoch": 0.28085330524097973,
      "grad_norm": 1.8533782958984375,
      "learning_rate": 3.596655254148012e-05,
      "loss": 1.5603,
      "step": 10664
    },
    {
      "epoch": 0.2808796418224914,
      "grad_norm": 3.2497153282165527,
      "learning_rate": 3.596523571240453e-05,
      "loss": 1.237,
      "step": 10665
    },
    {
      "epoch": 0.28090597840400316,
      "grad_norm": 1.4536837339401245,
      "learning_rate": 3.5963918883328945e-05,
      "loss": 2.5447,
      "step": 10666
    },
    {
      "epoch": 0.28093231498551485,
      "grad_norm": 1.774584412574768,
      "learning_rate": 3.596260205425336e-05,
      "loss": 1.4216,
      "step": 10667
    },
    {
      "epoch": 0.2809586515670266,
      "grad_norm": 2.871354103088379,
      "learning_rate": 3.596128522517777e-05,
      "loss": 1.5716,
      "step": 10668
    },
    {
      "epoch": 0.28098498814853834,
      "grad_norm": 1.5408275127410889,
      "learning_rate": 3.595996839610219e-05,
      "loss": 2.125,
      "step": 10669
    },
    {
      "epoch": 0.28101132473005,
      "grad_norm": 2.006460189819336,
      "learning_rate": 3.59586515670266e-05,
      "loss": 1.6612,
      "step": 10670
    },
    {
      "epoch": 0.28103766131156177,
      "grad_norm": 3.255465030670166,
      "learning_rate": 3.5957334737951016e-05,
      "loss": 1.4083,
      "step": 10671
    },
    {
      "epoch": 0.28106399789307346,
      "grad_norm": 4.274167537689209,
      "learning_rate": 3.5956017908875425e-05,
      "loss": 1.2642,
      "step": 10672
    },
    {
      "epoch": 0.2810903344745852,
      "grad_norm": 2.426962375640869,
      "learning_rate": 3.595470107979985e-05,
      "loss": 1.4912,
      "step": 10673
    },
    {
      "epoch": 0.28111667105609695,
      "grad_norm": 1.8696622848510742,
      "learning_rate": 3.5953384250724256e-05,
      "loss": 1.2084,
      "step": 10674
    },
    {
      "epoch": 0.28114300763760863,
      "grad_norm": 4.604039192199707,
      "learning_rate": 3.595206742164867e-05,
      "loss": 0.9946,
      "step": 10675
    },
    {
      "epoch": 0.2811693442191204,
      "grad_norm": 5.191089153289795,
      "learning_rate": 3.595075059257309e-05,
      "loss": 2.1446,
      "step": 10676
    },
    {
      "epoch": 0.28119568080063206,
      "grad_norm": 2.00488018989563,
      "learning_rate": 3.5949433763497496e-05,
      "loss": 1.9144,
      "step": 10677
    },
    {
      "epoch": 0.2812220173821438,
      "grad_norm": 2.6417410373687744,
      "learning_rate": 3.594811693442192e-05,
      "loss": 1.2435,
      "step": 10678
    },
    {
      "epoch": 0.2812483539636555,
      "grad_norm": 3.796337842941284,
      "learning_rate": 3.594680010534633e-05,
      "loss": 0.8128,
      "step": 10679
    },
    {
      "epoch": 0.28127469054516724,
      "grad_norm": 2.1222522258758545,
      "learning_rate": 3.594548327627074e-05,
      "loss": 2.1263,
      "step": 10680
    },
    {
      "epoch": 0.281301027126679,
      "grad_norm": 6.180479526519775,
      "learning_rate": 3.594416644719515e-05,
      "loss": 1.6803,
      "step": 10681
    },
    {
      "epoch": 0.28132736370819067,
      "grad_norm": 4.049415588378906,
      "learning_rate": 3.594284961811957e-05,
      "loss": 1.141,
      "step": 10682
    },
    {
      "epoch": 0.2813537002897024,
      "grad_norm": 1.708604335784912,
      "learning_rate": 3.594153278904398e-05,
      "loss": 2.5271,
      "step": 10683
    },
    {
      "epoch": 0.2813800368712141,
      "grad_norm": 2.8715932369232178,
      "learning_rate": 3.59402159599684e-05,
      "loss": 1.3198,
      "step": 10684
    },
    {
      "epoch": 0.28140637345272584,
      "grad_norm": 3.1334052085876465,
      "learning_rate": 3.5938899130892814e-05,
      "loss": 0.2673,
      "step": 10685
    },
    {
      "epoch": 0.28143271003423753,
      "grad_norm": 3.3107383251190186,
      "learning_rate": 3.593758230181722e-05,
      "loss": 1.4493,
      "step": 10686
    },
    {
      "epoch": 0.2814590466157493,
      "grad_norm": 1.8343831300735474,
      "learning_rate": 3.5936265472741645e-05,
      "loss": 2.1448,
      "step": 10687
    },
    {
      "epoch": 0.281485383197261,
      "grad_norm": 1.6674515008926392,
      "learning_rate": 3.5934948643666054e-05,
      "loss": 1.8204,
      "step": 10688
    },
    {
      "epoch": 0.2815117197787727,
      "grad_norm": 2.22741436958313,
      "learning_rate": 3.593363181459047e-05,
      "loss": 1.5256,
      "step": 10689
    },
    {
      "epoch": 0.28153805636028445,
      "grad_norm": 2.178539991378784,
      "learning_rate": 3.593231498551488e-05,
      "loss": 1.7154,
      "step": 10690
    },
    {
      "epoch": 0.28156439294179614,
      "grad_norm": 2.9959981441497803,
      "learning_rate": 3.5930998156439294e-05,
      "loss": 1.8,
      "step": 10691
    },
    {
      "epoch": 0.2815907295233079,
      "grad_norm": 1.6000972986221313,
      "learning_rate": 3.592968132736371e-05,
      "loss": 1.5372,
      "step": 10692
    },
    {
      "epoch": 0.28161706610481957,
      "grad_norm": 1.7989879846572876,
      "learning_rate": 3.5928364498288125e-05,
      "loss": 2.565,
      "step": 10693
    },
    {
      "epoch": 0.2816434026863313,
      "grad_norm": 1.635472059249878,
      "learning_rate": 3.592704766921254e-05,
      "loss": 1.303,
      "step": 10694
    },
    {
      "epoch": 0.28166973926784306,
      "grad_norm": 2.0246100425720215,
      "learning_rate": 3.592573084013695e-05,
      "loss": 1.5681,
      "step": 10695
    },
    {
      "epoch": 0.28169607584935474,
      "grad_norm": 1.5443994998931885,
      "learning_rate": 3.5924414011061365e-05,
      "loss": 2.0599,
      "step": 10696
    },
    {
      "epoch": 0.2817224124308665,
      "grad_norm": 3.6381704807281494,
      "learning_rate": 3.592309718198578e-05,
      "loss": 0.5802,
      "step": 10697
    },
    {
      "epoch": 0.2817487490123782,
      "grad_norm": 2.2468931674957275,
      "learning_rate": 3.59217803529102e-05,
      "loss": 2.0681,
      "step": 10698
    },
    {
      "epoch": 0.2817750855938899,
      "grad_norm": 1.9977346658706665,
      "learning_rate": 3.5920463523834605e-05,
      "loss": 2.037,
      "step": 10699
    },
    {
      "epoch": 0.2818014221754016,
      "grad_norm": 1.7926998138427734,
      "learning_rate": 3.591914669475902e-05,
      "loss": 1.299,
      "step": 10700
    },
    {
      "epoch": 0.28182775875691335,
      "grad_norm": 1.8555229902267456,
      "learning_rate": 3.591782986568344e-05,
      "loss": 1.7055,
      "step": 10701
    },
    {
      "epoch": 0.2818540953384251,
      "grad_norm": 2.0828359127044678,
      "learning_rate": 3.591651303660785e-05,
      "loss": 1.5137,
      "step": 10702
    },
    {
      "epoch": 0.2818804319199368,
      "grad_norm": 2.4519834518432617,
      "learning_rate": 3.591519620753227e-05,
      "loss": 1.5815,
      "step": 10703
    },
    {
      "epoch": 0.2819067685014485,
      "grad_norm": 3.7882378101348877,
      "learning_rate": 3.591387937845668e-05,
      "loss": 2.279,
      "step": 10704
    },
    {
      "epoch": 0.2819331050829602,
      "grad_norm": 3.561875581741333,
      "learning_rate": 3.591256254938109e-05,
      "loss": 0.529,
      "step": 10705
    },
    {
      "epoch": 0.28195944166447195,
      "grad_norm": 2.8583076000213623,
      "learning_rate": 3.591124572030551e-05,
      "loss": 1.5987,
      "step": 10706
    },
    {
      "epoch": 0.2819857782459837,
      "grad_norm": 2.1120409965515137,
      "learning_rate": 3.5909928891229923e-05,
      "loss": 2.0727,
      "step": 10707
    },
    {
      "epoch": 0.2820121148274954,
      "grad_norm": 2.837987184524536,
      "learning_rate": 3.590861206215433e-05,
      "loss": 1.2557,
      "step": 10708
    },
    {
      "epoch": 0.28203845140900713,
      "grad_norm": 1.9793260097503662,
      "learning_rate": 3.590729523307875e-05,
      "loss": 2.0804,
      "step": 10709
    },
    {
      "epoch": 0.2820647879905188,
      "grad_norm": 2.5520782470703125,
      "learning_rate": 3.590597840400316e-05,
      "loss": 1.4869,
      "step": 10710
    },
    {
      "epoch": 0.28209112457203056,
      "grad_norm": 1.3926054239273071,
      "learning_rate": 3.590466157492758e-05,
      "loss": 1.2704,
      "step": 10711
    },
    {
      "epoch": 0.28211746115354225,
      "grad_norm": 2.701925277709961,
      "learning_rate": 3.590334474585199e-05,
      "loss": 1.4179,
      "step": 10712
    },
    {
      "epoch": 0.282143797735054,
      "grad_norm": 2.1796493530273438,
      "learning_rate": 3.5902027916776403e-05,
      "loss": 1.6388,
      "step": 10713
    },
    {
      "epoch": 0.28217013431656573,
      "grad_norm": 3.534756660461426,
      "learning_rate": 3.590071108770082e-05,
      "loss": 0.7621,
      "step": 10714
    },
    {
      "epoch": 0.2821964708980774,
      "grad_norm": 3.0433273315429688,
      "learning_rate": 3.589939425862523e-05,
      "loss": 1.814,
      "step": 10715
    },
    {
      "epoch": 0.28222280747958917,
      "grad_norm": 1.9669334888458252,
      "learning_rate": 3.589807742954965e-05,
      "loss": 2.0966,
      "step": 10716
    },
    {
      "epoch": 0.28224914406110085,
      "grad_norm": 1.8439295291900635,
      "learning_rate": 3.589676060047406e-05,
      "loss": 2.0059,
      "step": 10717
    },
    {
      "epoch": 0.2822754806426126,
      "grad_norm": 1.5100206136703491,
      "learning_rate": 3.5895443771398475e-05,
      "loss": 1.4924,
      "step": 10718
    },
    {
      "epoch": 0.2823018172241243,
      "grad_norm": 2.1831393241882324,
      "learning_rate": 3.5894126942322884e-05,
      "loss": 1.3613,
      "step": 10719
    },
    {
      "epoch": 0.28232815380563603,
      "grad_norm": 2.080354928970337,
      "learning_rate": 3.5892810113247306e-05,
      "loss": 1.6664,
      "step": 10720
    },
    {
      "epoch": 0.28235449038714777,
      "grad_norm": 4.941184997558594,
      "learning_rate": 3.5891493284171715e-05,
      "loss": 1.6204,
      "step": 10721
    },
    {
      "epoch": 0.28238082696865946,
      "grad_norm": 2.2045881748199463,
      "learning_rate": 3.589017645509613e-05,
      "loss": 1.1392,
      "step": 10722
    },
    {
      "epoch": 0.2824071635501712,
      "grad_norm": 3.418728828430176,
      "learning_rate": 3.5888859626020546e-05,
      "loss": 1.6709,
      "step": 10723
    },
    {
      "epoch": 0.2824335001316829,
      "grad_norm": 1.8697012662887573,
      "learning_rate": 3.5887542796944955e-05,
      "loss": 2.127,
      "step": 10724
    },
    {
      "epoch": 0.28245983671319463,
      "grad_norm": 3.483039140701294,
      "learning_rate": 3.588622596786938e-05,
      "loss": 2.0922,
      "step": 10725
    },
    {
      "epoch": 0.2824861732947063,
      "grad_norm": 2.179659128189087,
      "learning_rate": 3.5884909138793786e-05,
      "loss": 1.1217,
      "step": 10726
    },
    {
      "epoch": 0.28251250987621807,
      "grad_norm": 1.5021837949752808,
      "learning_rate": 3.58835923097182e-05,
      "loss": 1.3667,
      "step": 10727
    },
    {
      "epoch": 0.2825388464577298,
      "grad_norm": 1.7014353275299072,
      "learning_rate": 3.588227548064261e-05,
      "loss": 1.4668,
      "step": 10728
    },
    {
      "epoch": 0.2825651830392415,
      "grad_norm": 1.534012794494629,
      "learning_rate": 3.5880958651567026e-05,
      "loss": 1.7634,
      "step": 10729
    },
    {
      "epoch": 0.28259151962075324,
      "grad_norm": 1.7933578491210938,
      "learning_rate": 3.587964182249144e-05,
      "loss": 0.7792,
      "step": 10730
    },
    {
      "epoch": 0.2826178562022649,
      "grad_norm": 3.4233410358428955,
      "learning_rate": 3.587832499341586e-05,
      "loss": 0.8375,
      "step": 10731
    },
    {
      "epoch": 0.28264419278377667,
      "grad_norm": 3.7914533615112305,
      "learning_rate": 3.587700816434027e-05,
      "loss": 1.1995,
      "step": 10732
    },
    {
      "epoch": 0.28267052936528836,
      "grad_norm": 2.1551201343536377,
      "learning_rate": 3.587569133526468e-05,
      "loss": 1.1777,
      "step": 10733
    },
    {
      "epoch": 0.2826968659468001,
      "grad_norm": 3.0795228481292725,
      "learning_rate": 3.58743745061891e-05,
      "loss": 1.6719,
      "step": 10734
    },
    {
      "epoch": 0.28272320252831185,
      "grad_norm": 1.899985909461975,
      "learning_rate": 3.587305767711351e-05,
      "loss": 2.0629,
      "step": 10735
    },
    {
      "epoch": 0.28274953910982353,
      "grad_norm": 2.0340893268585205,
      "learning_rate": 3.587174084803793e-05,
      "loss": 1.375,
      "step": 10736
    },
    {
      "epoch": 0.2827758756913353,
      "grad_norm": 1.7357654571533203,
      "learning_rate": 3.587042401896234e-05,
      "loss": 1.704,
      "step": 10737
    },
    {
      "epoch": 0.28280221227284696,
      "grad_norm": 3.342672348022461,
      "learning_rate": 3.586910718988675e-05,
      "loss": 1.2899,
      "step": 10738
    },
    {
      "epoch": 0.2828285488543587,
      "grad_norm": 3.6308014392852783,
      "learning_rate": 3.586779036081117e-05,
      "loss": 1.6223,
      "step": 10739
    },
    {
      "epoch": 0.28285488543587045,
      "grad_norm": 3.73903226852417,
      "learning_rate": 3.5866473531735584e-05,
      "loss": 1.5266,
      "step": 10740
    },
    {
      "epoch": 0.28288122201738214,
      "grad_norm": 1.6772918701171875,
      "learning_rate": 3.586515670266e-05,
      "loss": 1.6978,
      "step": 10741
    },
    {
      "epoch": 0.2829075585988939,
      "grad_norm": 1.7821944952011108,
      "learning_rate": 3.586383987358441e-05,
      "loss": 1.883,
      "step": 10742
    },
    {
      "epoch": 0.28293389518040557,
      "grad_norm": 4.070680618286133,
      "learning_rate": 3.5862523044508824e-05,
      "loss": 1.7298,
      "step": 10743
    },
    {
      "epoch": 0.2829602317619173,
      "grad_norm": 2.752892255783081,
      "learning_rate": 3.586120621543324e-05,
      "loss": 1.6368,
      "step": 10744
    },
    {
      "epoch": 0.282986568343429,
      "grad_norm": 2.187781572341919,
      "learning_rate": 3.5859889386357655e-05,
      "loss": 1.8481,
      "step": 10745
    },
    {
      "epoch": 0.28301290492494074,
      "grad_norm": 1.964159607887268,
      "learning_rate": 3.5858572557282064e-05,
      "loss": 1.5193,
      "step": 10746
    },
    {
      "epoch": 0.2830392415064525,
      "grad_norm": 2.1101324558258057,
      "learning_rate": 3.585725572820648e-05,
      "loss": 2.3428,
      "step": 10747
    },
    {
      "epoch": 0.2830655780879642,
      "grad_norm": 1.8688033819198608,
      "learning_rate": 3.5855938899130895e-05,
      "loss": 2.0471,
      "step": 10748
    },
    {
      "epoch": 0.2830919146694759,
      "grad_norm": 2.8875467777252197,
      "learning_rate": 3.585462207005531e-05,
      "loss": 1.9323,
      "step": 10749
    },
    {
      "epoch": 0.2831182512509876,
      "grad_norm": 1.721235990524292,
      "learning_rate": 3.5853305240979726e-05,
      "loss": 2.2385,
      "step": 10750
    },
    {
      "epoch": 0.28314458783249935,
      "grad_norm": 2.5938611030578613,
      "learning_rate": 3.5851988411904135e-05,
      "loss": 1.7843,
      "step": 10751
    },
    {
      "epoch": 0.28317092441401104,
      "grad_norm": 1.7995103597640991,
      "learning_rate": 3.585067158282855e-05,
      "loss": 1.9036,
      "step": 10752
    },
    {
      "epoch": 0.2831972609955228,
      "grad_norm": 3.8045501708984375,
      "learning_rate": 3.5849354753752966e-05,
      "loss": 2.0679,
      "step": 10753
    },
    {
      "epoch": 0.2832235975770345,
      "grad_norm": 2.0957834720611572,
      "learning_rate": 3.584803792467738e-05,
      "loss": 1.7508,
      "step": 10754
    },
    {
      "epoch": 0.2832499341585462,
      "grad_norm": 4.18516206741333,
      "learning_rate": 3.584672109560179e-05,
      "loss": 1.6536,
      "step": 10755
    },
    {
      "epoch": 0.28327627074005796,
      "grad_norm": 2.9924893379211426,
      "learning_rate": 3.5845404266526206e-05,
      "loss": 1.7985,
      "step": 10756
    },
    {
      "epoch": 0.28330260732156964,
      "grad_norm": 2.581116199493408,
      "learning_rate": 3.5844087437450615e-05,
      "loss": 1.4849,
      "step": 10757
    },
    {
      "epoch": 0.2833289439030814,
      "grad_norm": 2.794194221496582,
      "learning_rate": 3.584277060837504e-05,
      "loss": 1.2438,
      "step": 10758
    },
    {
      "epoch": 0.2833552804845931,
      "grad_norm": 3.0024542808532715,
      "learning_rate": 3.5841453779299446e-05,
      "loss": 1.3339,
      "step": 10759
    },
    {
      "epoch": 0.2833816170661048,
      "grad_norm": 2.0632169246673584,
      "learning_rate": 3.584013695022386e-05,
      "loss": 1.1334,
      "step": 10760
    },
    {
      "epoch": 0.28340795364761656,
      "grad_norm": 3.399566888809204,
      "learning_rate": 3.583882012114828e-05,
      "loss": 1.8846,
      "step": 10761
    },
    {
      "epoch": 0.28343429022912825,
      "grad_norm": 2.2777318954467773,
      "learning_rate": 3.5837503292072686e-05,
      "loss": 1.5551,
      "step": 10762
    },
    {
      "epoch": 0.28346062681064,
      "grad_norm": 1.8050422668457031,
      "learning_rate": 3.583618646299711e-05,
      "loss": 1.6948,
      "step": 10763
    },
    {
      "epoch": 0.2834869633921517,
      "grad_norm": 1.8940041065216064,
      "learning_rate": 3.583486963392152e-05,
      "loss": 1.2282,
      "step": 10764
    },
    {
      "epoch": 0.2835132999736634,
      "grad_norm": 2.728278398513794,
      "learning_rate": 3.583355280484593e-05,
      "loss": 1.2962,
      "step": 10765
    },
    {
      "epoch": 0.2835396365551751,
      "grad_norm": 4.125481605529785,
      "learning_rate": 3.583223597577034e-05,
      "loss": 2.2674,
      "step": 10766
    },
    {
      "epoch": 0.28356597313668686,
      "grad_norm": 3.0636467933654785,
      "learning_rate": 3.583091914669476e-05,
      "loss": 1.7247,
      "step": 10767
    },
    {
      "epoch": 0.2835923097181986,
      "grad_norm": 3.9943528175354004,
      "learning_rate": 3.582960231761917e-05,
      "loss": 1.2233,
      "step": 10768
    },
    {
      "epoch": 0.2836186462997103,
      "grad_norm": 5.200692653656006,
      "learning_rate": 3.582828548854359e-05,
      "loss": 1.1838,
      "step": 10769
    },
    {
      "epoch": 0.28364498288122203,
      "grad_norm": 6.776669025421143,
      "learning_rate": 3.5826968659468004e-05,
      "loss": 1.1258,
      "step": 10770
    },
    {
      "epoch": 0.2836713194627337,
      "grad_norm": 1.9097925424575806,
      "learning_rate": 3.582565183039241e-05,
      "loss": 1.449,
      "step": 10771
    },
    {
      "epoch": 0.28369765604424546,
      "grad_norm": 2.9563980102539062,
      "learning_rate": 3.5824335001316836e-05,
      "loss": 1.4752,
      "step": 10772
    },
    {
      "epoch": 0.2837239926257572,
      "grad_norm": 3.4084765911102295,
      "learning_rate": 3.5823018172241244e-05,
      "loss": 0.757,
      "step": 10773
    },
    {
      "epoch": 0.2837503292072689,
      "grad_norm": 2.4827938079833984,
      "learning_rate": 3.582170134316566e-05,
      "loss": 1.5668,
      "step": 10774
    },
    {
      "epoch": 0.28377666578878064,
      "grad_norm": 1.687890648841858,
      "learning_rate": 3.582038451409007e-05,
      "loss": 1.8184,
      "step": 10775
    },
    {
      "epoch": 0.2838030023702923,
      "grad_norm": 1.8881187438964844,
      "learning_rate": 3.5819067685014484e-05,
      "loss": 2.5232,
      "step": 10776
    },
    {
      "epoch": 0.28382933895180407,
      "grad_norm": 3.3049566745758057,
      "learning_rate": 3.58177508559389e-05,
      "loss": 1.678,
      "step": 10777
    },
    {
      "epoch": 0.28385567553331575,
      "grad_norm": 3.8362557888031006,
      "learning_rate": 3.5816434026863316e-05,
      "loss": 1.7891,
      "step": 10778
    },
    {
      "epoch": 0.2838820121148275,
      "grad_norm": 1.5089341402053833,
      "learning_rate": 3.581511719778773e-05,
      "loss": 2.3159,
      "step": 10779
    },
    {
      "epoch": 0.28390834869633924,
      "grad_norm": 1.741356611251831,
      "learning_rate": 3.581380036871214e-05,
      "loss": 1.8636,
      "step": 10780
    },
    {
      "epoch": 0.28393468527785093,
      "grad_norm": 1.5450695753097534,
      "learning_rate": 3.5812483539636556e-05,
      "loss": 0.3157,
      "step": 10781
    },
    {
      "epoch": 0.2839610218593627,
      "grad_norm": 1.453835129737854,
      "learning_rate": 3.581116671056097e-05,
      "loss": 2.0476,
      "step": 10782
    },
    {
      "epoch": 0.28398735844087436,
      "grad_norm": 3.3188846111297607,
      "learning_rate": 3.580984988148539e-05,
      "loss": 0.5036,
      "step": 10783
    },
    {
      "epoch": 0.2840136950223861,
      "grad_norm": 1.9738532304763794,
      "learning_rate": 3.5808533052409796e-05,
      "loss": 1.659,
      "step": 10784
    },
    {
      "epoch": 0.2840400316038978,
      "grad_norm": 1.987406611442566,
      "learning_rate": 3.580721622333421e-05,
      "loss": 1.9322,
      "step": 10785
    },
    {
      "epoch": 0.28406636818540953,
      "grad_norm": 3.094905138015747,
      "learning_rate": 3.580589939425863e-05,
      "loss": 1.2558,
      "step": 10786
    },
    {
      "epoch": 0.2840927047669213,
      "grad_norm": 2.3643505573272705,
      "learning_rate": 3.580458256518304e-05,
      "loss": 1.6724,
      "step": 10787
    },
    {
      "epoch": 0.28411904134843297,
      "grad_norm": 2.3726208209991455,
      "learning_rate": 3.580326573610746e-05,
      "loss": 1.7988,
      "step": 10788
    },
    {
      "epoch": 0.2841453779299447,
      "grad_norm": 2.0558271408081055,
      "learning_rate": 3.580194890703187e-05,
      "loss": 2.0569,
      "step": 10789
    },
    {
      "epoch": 0.2841717145114564,
      "grad_norm": 3.745953321456909,
      "learning_rate": 3.580063207795628e-05,
      "loss": 1.3069,
      "step": 10790
    },
    {
      "epoch": 0.28419805109296814,
      "grad_norm": 1.679540753364563,
      "learning_rate": 3.57993152488807e-05,
      "loss": 1.4655,
      "step": 10791
    },
    {
      "epoch": 0.28422438767447983,
      "grad_norm": 1.8753336668014526,
      "learning_rate": 3.5797998419805114e-05,
      "loss": 2.1195,
      "step": 10792
    },
    {
      "epoch": 0.28425072425599157,
      "grad_norm": 1.9696885347366333,
      "learning_rate": 3.579668159072952e-05,
      "loss": 1.4623,
      "step": 10793
    },
    {
      "epoch": 0.2842770608375033,
      "grad_norm": 11.313634872436523,
      "learning_rate": 3.579536476165394e-05,
      "loss": 2.0706,
      "step": 10794
    },
    {
      "epoch": 0.284303397419015,
      "grad_norm": 1.9421197175979614,
      "learning_rate": 3.5794047932578354e-05,
      "loss": 1.9291,
      "step": 10795
    },
    {
      "epoch": 0.28432973400052675,
      "grad_norm": 2.6556382179260254,
      "learning_rate": 3.579273110350277e-05,
      "loss": 0.9597,
      "step": 10796
    },
    {
      "epoch": 0.28435607058203843,
      "grad_norm": 2.0371716022491455,
      "learning_rate": 3.5791414274427185e-05,
      "loss": 1.846,
      "step": 10797
    },
    {
      "epoch": 0.2843824071635502,
      "grad_norm": 2.046513319015503,
      "learning_rate": 3.5790097445351594e-05,
      "loss": 1.5943,
      "step": 10798
    },
    {
      "epoch": 0.28440874374506186,
      "grad_norm": 1.6495250463485718,
      "learning_rate": 3.578878061627601e-05,
      "loss": 1.9194,
      "step": 10799
    },
    {
      "epoch": 0.2844350803265736,
      "grad_norm": 1.8406317234039307,
      "learning_rate": 3.578746378720042e-05,
      "loss": 1.9922,
      "step": 10800
    },
    {
      "epoch": 0.28446141690808535,
      "grad_norm": 3.196396589279175,
      "learning_rate": 3.578614695812484e-05,
      "loss": 2.1663,
      "step": 10801
    },
    {
      "epoch": 0.28448775348959704,
      "grad_norm": 2.3981986045837402,
      "learning_rate": 3.578483012904925e-05,
      "loss": 1.365,
      "step": 10802
    },
    {
      "epoch": 0.2845140900711088,
      "grad_norm": 1.6877681016921997,
      "learning_rate": 3.5783513299973665e-05,
      "loss": 1.7587,
      "step": 10803
    },
    {
      "epoch": 0.28454042665262047,
      "grad_norm": 1.5932350158691406,
      "learning_rate": 3.578219647089808e-05,
      "loss": 1.6025,
      "step": 10804
    },
    {
      "epoch": 0.2845667632341322,
      "grad_norm": 2.3335299491882324,
      "learning_rate": 3.5780879641822496e-05,
      "loss": 1.7822,
      "step": 10805
    },
    {
      "epoch": 0.2845930998156439,
      "grad_norm": 1.9196819067001343,
      "learning_rate": 3.577956281274691e-05,
      "loss": 1.8837,
      "step": 10806
    },
    {
      "epoch": 0.28461943639715565,
      "grad_norm": 1.398080825805664,
      "learning_rate": 3.577824598367132e-05,
      "loss": 1.9654,
      "step": 10807
    },
    {
      "epoch": 0.2846457729786674,
      "grad_norm": 5.5888261795043945,
      "learning_rate": 3.5776929154595736e-05,
      "loss": 1.0024,
      "step": 10808
    },
    {
      "epoch": 0.2846721095601791,
      "grad_norm": 1.979221224784851,
      "learning_rate": 3.5775612325520145e-05,
      "loss": 1.6808,
      "step": 10809
    },
    {
      "epoch": 0.2846984461416908,
      "grad_norm": 3.1664185523986816,
      "learning_rate": 3.577429549644457e-05,
      "loss": 1.4544,
      "step": 10810
    },
    {
      "epoch": 0.2847247827232025,
      "grad_norm": 2.1926774978637695,
      "learning_rate": 3.5772978667368976e-05,
      "loss": 1.1609,
      "step": 10811
    },
    {
      "epoch": 0.28475111930471425,
      "grad_norm": 1.938710331916809,
      "learning_rate": 3.577166183829339e-05,
      "loss": 0.4026,
      "step": 10812
    },
    {
      "epoch": 0.284777455886226,
      "grad_norm": 1.8479552268981934,
      "learning_rate": 3.57703450092178e-05,
      "loss": 1.0639,
      "step": 10813
    },
    {
      "epoch": 0.2848037924677377,
      "grad_norm": 4.910101413726807,
      "learning_rate": 3.5769028180142216e-05,
      "loss": 1.3656,
      "step": 10814
    },
    {
      "epoch": 0.2848301290492494,
      "grad_norm": 1.845170021057129,
      "learning_rate": 3.576771135106663e-05,
      "loss": 1.8049,
      "step": 10815
    },
    {
      "epoch": 0.2848564656307611,
      "grad_norm": 2.0903091430664062,
      "learning_rate": 3.576639452199105e-05,
      "loss": 2.0961,
      "step": 10816
    },
    {
      "epoch": 0.28488280221227286,
      "grad_norm": 4.3012003898620605,
      "learning_rate": 3.576507769291546e-05,
      "loss": 1.1164,
      "step": 10817
    },
    {
      "epoch": 0.28490913879378454,
      "grad_norm": 1.6772291660308838,
      "learning_rate": 3.576376086383987e-05,
      "loss": 1.9686,
      "step": 10818
    },
    {
      "epoch": 0.2849354753752963,
      "grad_norm": 1.6850790977478027,
      "learning_rate": 3.5762444034764294e-05,
      "loss": 1.7021,
      "step": 10819
    },
    {
      "epoch": 0.28496181195680803,
      "grad_norm": 1.6811139583587646,
      "learning_rate": 3.57611272056887e-05,
      "loss": 2.0023,
      "step": 10820
    },
    {
      "epoch": 0.2849881485383197,
      "grad_norm": 2.713277816772461,
      "learning_rate": 3.575981037661312e-05,
      "loss": 1.8991,
      "step": 10821
    },
    {
      "epoch": 0.28501448511983146,
      "grad_norm": 2.5310380458831787,
      "learning_rate": 3.575849354753753e-05,
      "loss": 1.5155,
      "step": 10822
    },
    {
      "epoch": 0.28504082170134315,
      "grad_norm": 4.494997024536133,
      "learning_rate": 3.575717671846194e-05,
      "loss": 1.2162,
      "step": 10823
    },
    {
      "epoch": 0.2850671582828549,
      "grad_norm": 2.0674829483032227,
      "learning_rate": 3.575585988938636e-05,
      "loss": 2.3371,
      "step": 10824
    },
    {
      "epoch": 0.2850934948643666,
      "grad_norm": 3.2719314098358154,
      "learning_rate": 3.5754543060310774e-05,
      "loss": 0.9209,
      "step": 10825
    },
    {
      "epoch": 0.2851198314458783,
      "grad_norm": 1.9471609592437744,
      "learning_rate": 3.575322623123519e-05,
      "loss": 2.8147,
      "step": 10826
    },
    {
      "epoch": 0.28514616802739007,
      "grad_norm": 2.836864709854126,
      "learning_rate": 3.57519094021596e-05,
      "loss": 1.7648,
      "step": 10827
    },
    {
      "epoch": 0.28517250460890176,
      "grad_norm": 2.6008498668670654,
      "learning_rate": 3.5750592573084014e-05,
      "loss": 1.9024,
      "step": 10828
    },
    {
      "epoch": 0.2851988411904135,
      "grad_norm": 2.5339577198028564,
      "learning_rate": 3.574927574400843e-05,
      "loss": 1.5394,
      "step": 10829
    },
    {
      "epoch": 0.2852251777719252,
      "grad_norm": 1.392446517944336,
      "learning_rate": 3.5747958914932845e-05,
      "loss": 1.4745,
      "step": 10830
    },
    {
      "epoch": 0.28525151435343693,
      "grad_norm": 2.5190680027008057,
      "learning_rate": 3.5746642085857254e-05,
      "loss": 1.4114,
      "step": 10831
    },
    {
      "epoch": 0.2852778509349486,
      "grad_norm": 3.5672900676727295,
      "learning_rate": 3.574532525678167e-05,
      "loss": 2.1517,
      "step": 10832
    },
    {
      "epoch": 0.28530418751646036,
      "grad_norm": 1.529719352722168,
      "learning_rate": 3.5744008427706085e-05,
      "loss": 1.8108,
      "step": 10833
    },
    {
      "epoch": 0.2853305240979721,
      "grad_norm": 3.915332555770874,
      "learning_rate": 3.57426915986305e-05,
      "loss": 1.2517,
      "step": 10834
    },
    {
      "epoch": 0.2853568606794838,
      "grad_norm": 2.5521485805511475,
      "learning_rate": 3.574137476955492e-05,
      "loss": 1.9467,
      "step": 10835
    },
    {
      "epoch": 0.28538319726099554,
      "grad_norm": 1.6555213928222656,
      "learning_rate": 3.5740057940479325e-05,
      "loss": 2.1273,
      "step": 10836
    },
    {
      "epoch": 0.2854095338425072,
      "grad_norm": 2.305870294570923,
      "learning_rate": 3.573874111140374e-05,
      "loss": 1.8347,
      "step": 10837
    },
    {
      "epoch": 0.28543587042401897,
      "grad_norm": 2.5898029804229736,
      "learning_rate": 3.573742428232816e-05,
      "loss": 1.5236,
      "step": 10838
    },
    {
      "epoch": 0.28546220700553065,
      "grad_norm": 1.9200682640075684,
      "learning_rate": 3.573610745325257e-05,
      "loss": 2.0033,
      "step": 10839
    },
    {
      "epoch": 0.2854885435870424,
      "grad_norm": 3.0526697635650635,
      "learning_rate": 3.573479062417698e-05,
      "loss": 2.4709,
      "step": 10840
    },
    {
      "epoch": 0.28551488016855414,
      "grad_norm": 1.7224936485290527,
      "learning_rate": 3.57334737951014e-05,
      "loss": 1.8446,
      "step": 10841
    },
    {
      "epoch": 0.28554121675006583,
      "grad_norm": 1.5152957439422607,
      "learning_rate": 3.573215696602581e-05,
      "loss": 1.792,
      "step": 10842
    },
    {
      "epoch": 0.2855675533315776,
      "grad_norm": 4.488541603088379,
      "learning_rate": 3.573084013695023e-05,
      "loss": 1.5066,
      "step": 10843
    },
    {
      "epoch": 0.28559388991308926,
      "grad_norm": 2.974428415298462,
      "learning_rate": 3.5729523307874643e-05,
      "loss": 1.7532,
      "step": 10844
    },
    {
      "epoch": 0.285620226494601,
      "grad_norm": 2.302385091781616,
      "learning_rate": 3.572820647879905e-05,
      "loss": 1.5413,
      "step": 10845
    },
    {
      "epoch": 0.28564656307611275,
      "grad_norm": 1.6653081178665161,
      "learning_rate": 3.572688964972347e-05,
      "loss": 1.825,
      "step": 10846
    },
    {
      "epoch": 0.28567289965762444,
      "grad_norm": 1.374388575553894,
      "learning_rate": 3.572557282064788e-05,
      "loss": 0.7493,
      "step": 10847
    },
    {
      "epoch": 0.2856992362391362,
      "grad_norm": 2.031874418258667,
      "learning_rate": 3.57242559915723e-05,
      "loss": 1.3097,
      "step": 10848
    },
    {
      "epoch": 0.28572557282064787,
      "grad_norm": 4.8944621086120605,
      "learning_rate": 3.572293916249671e-05,
      "loss": 2.3464,
      "step": 10849
    },
    {
      "epoch": 0.2857519094021596,
      "grad_norm": 2.4388692378997803,
      "learning_rate": 3.5721622333421124e-05,
      "loss": 1.7899,
      "step": 10850
    },
    {
      "epoch": 0.2857782459836713,
      "grad_norm": 3.541611909866333,
      "learning_rate": 3.572030550434554e-05,
      "loss": 0.405,
      "step": 10851
    },
    {
      "epoch": 0.28580458256518304,
      "grad_norm": 1.9741238355636597,
      "learning_rate": 3.5718988675269955e-05,
      "loss": 1.5981,
      "step": 10852
    },
    {
      "epoch": 0.2858309191466948,
      "grad_norm": 1.4928537607192993,
      "learning_rate": 3.571767184619437e-05,
      "loss": 0.6627,
      "step": 10853
    },
    {
      "epoch": 0.28585725572820647,
      "grad_norm": 2.9098763465881348,
      "learning_rate": 3.571635501711878e-05,
      "loss": 1.8585,
      "step": 10854
    },
    {
      "epoch": 0.2858835923097182,
      "grad_norm": 1.4445680379867554,
      "learning_rate": 3.5715038188043195e-05,
      "loss": 1.4236,
      "step": 10855
    },
    {
      "epoch": 0.2859099288912299,
      "grad_norm": 1.6740747690200806,
      "learning_rate": 3.5713721358967604e-05,
      "loss": 2.0825,
      "step": 10856
    },
    {
      "epoch": 0.28593626547274165,
      "grad_norm": 2.372706174850464,
      "learning_rate": 3.5712404529892026e-05,
      "loss": 1.7807,
      "step": 10857
    },
    {
      "epoch": 0.28596260205425333,
      "grad_norm": 1.9517550468444824,
      "learning_rate": 3.5711087700816435e-05,
      "loss": 0.7374,
      "step": 10858
    },
    {
      "epoch": 0.2859889386357651,
      "grad_norm": 1.688175916671753,
      "learning_rate": 3.570977087174085e-05,
      "loss": 2.0738,
      "step": 10859
    },
    {
      "epoch": 0.2860152752172768,
      "grad_norm": 4.115051746368408,
      "learning_rate": 3.570845404266526e-05,
      "loss": 0.9316,
      "step": 10860
    },
    {
      "epoch": 0.2860416117987885,
      "grad_norm": 2.710395574569702,
      "learning_rate": 3.5707137213589675e-05,
      "loss": 1.8353,
      "step": 10861
    },
    {
      "epoch": 0.28606794838030025,
      "grad_norm": 2.231415033340454,
      "learning_rate": 3.570582038451409e-05,
      "loss": 2.2586,
      "step": 10862
    },
    {
      "epoch": 0.28609428496181194,
      "grad_norm": 2.0620667934417725,
      "learning_rate": 3.5704503555438506e-05,
      "loss": 1.6772,
      "step": 10863
    },
    {
      "epoch": 0.2861206215433237,
      "grad_norm": 2.5980162620544434,
      "learning_rate": 3.570318672636292e-05,
      "loss": 1.7098,
      "step": 10864
    },
    {
      "epoch": 0.28614695812483537,
      "grad_norm": 1.8276180028915405,
      "learning_rate": 3.570186989728733e-05,
      "loss": 1.6537,
      "step": 10865
    },
    {
      "epoch": 0.2861732947063471,
      "grad_norm": 1.5904067754745483,
      "learning_rate": 3.5700553068211746e-05,
      "loss": 2.1786,
      "step": 10866
    },
    {
      "epoch": 0.28619963128785886,
      "grad_norm": 2.0989725589752197,
      "learning_rate": 3.569923623913616e-05,
      "loss": 1.2591,
      "step": 10867
    },
    {
      "epoch": 0.28622596786937055,
      "grad_norm": 2.1903600692749023,
      "learning_rate": 3.569791941006058e-05,
      "loss": 0.3586,
      "step": 10868
    },
    {
      "epoch": 0.2862523044508823,
      "grad_norm": 2.014195680618286,
      "learning_rate": 3.5696602580984986e-05,
      "loss": 1.6001,
      "step": 10869
    },
    {
      "epoch": 0.286278641032394,
      "grad_norm": 2.124044418334961,
      "learning_rate": 3.56952857519094e-05,
      "loss": 2.0115,
      "step": 10870
    },
    {
      "epoch": 0.2863049776139057,
      "grad_norm": 2.830047369003296,
      "learning_rate": 3.569396892283382e-05,
      "loss": 1.9418,
      "step": 10871
    },
    {
      "epoch": 0.2863313141954174,
      "grad_norm": 2.600907564163208,
      "learning_rate": 3.569265209375823e-05,
      "loss": 1.5323,
      "step": 10872
    },
    {
      "epoch": 0.28635765077692915,
      "grad_norm": 3.5687639713287354,
      "learning_rate": 3.569133526468265e-05,
      "loss": 1.0994,
      "step": 10873
    },
    {
      "epoch": 0.2863839873584409,
      "grad_norm": 1.9190971851348877,
      "learning_rate": 3.569001843560706e-05,
      "loss": 2.4625,
      "step": 10874
    },
    {
      "epoch": 0.2864103239399526,
      "grad_norm": 1.8622959852218628,
      "learning_rate": 3.568870160653147e-05,
      "loss": 2.3653,
      "step": 10875
    },
    {
      "epoch": 0.2864366605214643,
      "grad_norm": 1.8052875995635986,
      "learning_rate": 3.568738477745589e-05,
      "loss": 1.2829,
      "step": 10876
    },
    {
      "epoch": 0.286462997102976,
      "grad_norm": 1.6404985189437866,
      "learning_rate": 3.5686067948380304e-05,
      "loss": 1.8094,
      "step": 10877
    },
    {
      "epoch": 0.28648933368448776,
      "grad_norm": 4.276154518127441,
      "learning_rate": 3.568475111930471e-05,
      "loss": 1.5444,
      "step": 10878
    },
    {
      "epoch": 0.2865156702659995,
      "grad_norm": 2.249417543411255,
      "learning_rate": 3.568343429022913e-05,
      "loss": 1.2385,
      "step": 10879
    },
    {
      "epoch": 0.2865420068475112,
      "grad_norm": 2.065884590148926,
      "learning_rate": 3.5682117461153544e-05,
      "loss": 0.924,
      "step": 10880
    },
    {
      "epoch": 0.28656834342902293,
      "grad_norm": 3.236016273498535,
      "learning_rate": 3.568080063207796e-05,
      "loss": 2.198,
      "step": 10881
    },
    {
      "epoch": 0.2865946800105346,
      "grad_norm": 5.809149742126465,
      "learning_rate": 3.5679483803002375e-05,
      "loss": 1.4607,
      "step": 10882
    },
    {
      "epoch": 0.28662101659204636,
      "grad_norm": 4.300374984741211,
      "learning_rate": 3.5678166973926784e-05,
      "loss": 0.4194,
      "step": 10883
    },
    {
      "epoch": 0.28664735317355805,
      "grad_norm": 1.977638602256775,
      "learning_rate": 3.56768501448512e-05,
      "loss": 2.344,
      "step": 10884
    },
    {
      "epoch": 0.2866736897550698,
      "grad_norm": 4.997640609741211,
      "learning_rate": 3.5675533315775615e-05,
      "loss": 1.664,
      "step": 10885
    },
    {
      "epoch": 0.28670002633658154,
      "grad_norm": 3.3185009956359863,
      "learning_rate": 3.567421648670003e-05,
      "loss": 2.1228,
      "step": 10886
    },
    {
      "epoch": 0.2867263629180932,
      "grad_norm": 1.818388819694519,
      "learning_rate": 3.567289965762444e-05,
      "loss": 2.1683,
      "step": 10887
    },
    {
      "epoch": 0.28675269949960497,
      "grad_norm": 1.9575581550598145,
      "learning_rate": 3.5671582828548855e-05,
      "loss": 1.1518,
      "step": 10888
    },
    {
      "epoch": 0.28677903608111666,
      "grad_norm": 3.258960008621216,
      "learning_rate": 3.567026599947327e-05,
      "loss": 0.9538,
      "step": 10889
    },
    {
      "epoch": 0.2868053726626284,
      "grad_norm": 2.2418830394744873,
      "learning_rate": 3.5668949170397686e-05,
      "loss": 1.5618,
      "step": 10890
    },
    {
      "epoch": 0.2868317092441401,
      "grad_norm": 2.235569715499878,
      "learning_rate": 3.56676323413221e-05,
      "loss": 1.2931,
      "step": 10891
    },
    {
      "epoch": 0.28685804582565183,
      "grad_norm": 6.3410749435424805,
      "learning_rate": 3.566631551224651e-05,
      "loss": 1.3158,
      "step": 10892
    },
    {
      "epoch": 0.2868843824071636,
      "grad_norm": 3.698495864868164,
      "learning_rate": 3.5664998683170926e-05,
      "loss": 1.4903,
      "step": 10893
    },
    {
      "epoch": 0.28691071898867526,
      "grad_norm": 1.6409622430801392,
      "learning_rate": 3.5663681854095335e-05,
      "loss": 1.9041,
      "step": 10894
    },
    {
      "epoch": 0.286937055570187,
      "grad_norm": 2.1133944988250732,
      "learning_rate": 3.566236502501976e-05,
      "loss": 1.7888,
      "step": 10895
    },
    {
      "epoch": 0.2869633921516987,
      "grad_norm": 1.7504150867462158,
      "learning_rate": 3.5661048195944166e-05,
      "loss": 1.6241,
      "step": 10896
    },
    {
      "epoch": 0.28698972873321044,
      "grad_norm": 3.844424247741699,
      "learning_rate": 3.565973136686858e-05,
      "loss": 1.9478,
      "step": 10897
    },
    {
      "epoch": 0.2870160653147221,
      "grad_norm": 2.768083333969116,
      "learning_rate": 3.5658414537793e-05,
      "loss": 1.5144,
      "step": 10898
    },
    {
      "epoch": 0.28704240189623387,
      "grad_norm": 1.9580880403518677,
      "learning_rate": 3.565709770871741e-05,
      "loss": 1.4769,
      "step": 10899
    },
    {
      "epoch": 0.2870687384777456,
      "grad_norm": 2.030640125274658,
      "learning_rate": 3.565578087964183e-05,
      "loss": 0.4653,
      "step": 10900
    },
    {
      "epoch": 0.2870950750592573,
      "grad_norm": 4.788317680358887,
      "learning_rate": 3.565446405056624e-05,
      "loss": 1.5075,
      "step": 10901
    },
    {
      "epoch": 0.28712141164076904,
      "grad_norm": 4.070150375366211,
      "learning_rate": 3.565314722149065e-05,
      "loss": 1.6207,
      "step": 10902
    },
    {
      "epoch": 0.28714774822228073,
      "grad_norm": 2.3336422443389893,
      "learning_rate": 3.565183039241506e-05,
      "loss": 0.6337,
      "step": 10903
    },
    {
      "epoch": 0.2871740848037925,
      "grad_norm": 2.0194108486175537,
      "learning_rate": 3.5650513563339484e-05,
      "loss": 2.1127,
      "step": 10904
    },
    {
      "epoch": 0.28720042138530416,
      "grad_norm": 2.1101112365722656,
      "learning_rate": 3.564919673426389e-05,
      "loss": 2.5143,
      "step": 10905
    },
    {
      "epoch": 0.2872267579668159,
      "grad_norm": 1.6519547700881958,
      "learning_rate": 3.564787990518831e-05,
      "loss": 0.6376,
      "step": 10906
    },
    {
      "epoch": 0.28725309454832765,
      "grad_norm": 2.7773334980010986,
      "learning_rate": 3.5646563076112725e-05,
      "loss": 1.6727,
      "step": 10907
    },
    {
      "epoch": 0.28727943112983934,
      "grad_norm": 4.362887859344482,
      "learning_rate": 3.564524624703713e-05,
      "loss": 1.2566,
      "step": 10908
    },
    {
      "epoch": 0.2873057677113511,
      "grad_norm": 1.9870001077651978,
      "learning_rate": 3.5643929417961556e-05,
      "loss": 2.2809,
      "step": 10909
    },
    {
      "epoch": 0.28733210429286277,
      "grad_norm": 2.3629238605499268,
      "learning_rate": 3.5642612588885965e-05,
      "loss": 1.1103,
      "step": 10910
    },
    {
      "epoch": 0.2873584408743745,
      "grad_norm": 1.5843186378479004,
      "learning_rate": 3.564129575981038e-05,
      "loss": 1.4126,
      "step": 10911
    },
    {
      "epoch": 0.28738477745588625,
      "grad_norm": 2.0120108127593994,
      "learning_rate": 3.563997893073479e-05,
      "loss": 1.0796,
      "step": 10912
    },
    {
      "epoch": 0.28741111403739794,
      "grad_norm": 2.035494089126587,
      "learning_rate": 3.5638662101659205e-05,
      "loss": 1.4534,
      "step": 10913
    },
    {
      "epoch": 0.2874374506189097,
      "grad_norm": 1.856073021888733,
      "learning_rate": 3.563734527258362e-05,
      "loss": 1.6312,
      "step": 10914
    },
    {
      "epoch": 0.2874637872004214,
      "grad_norm": 3.239377737045288,
      "learning_rate": 3.5636028443508036e-05,
      "loss": 1.181,
      "step": 10915
    },
    {
      "epoch": 0.2874901237819331,
      "grad_norm": 1.8262280225753784,
      "learning_rate": 3.5634711614432445e-05,
      "loss": 1.5209,
      "step": 10916
    },
    {
      "epoch": 0.2875164603634448,
      "grad_norm": 1.726887822151184,
      "learning_rate": 3.563339478535686e-05,
      "loss": 2.0076,
      "step": 10917
    },
    {
      "epoch": 0.28754279694495655,
      "grad_norm": 2.1747591495513916,
      "learning_rate": 3.5632077956281276e-05,
      "loss": 2.0095,
      "step": 10918
    },
    {
      "epoch": 0.2875691335264683,
      "grad_norm": 2.274733543395996,
      "learning_rate": 3.563076112720569e-05,
      "loss": 1.4245,
      "step": 10919
    },
    {
      "epoch": 0.28759547010798,
      "grad_norm": 2.9830405712127686,
      "learning_rate": 3.562944429813011e-05,
      "loss": 1.0625,
      "step": 10920
    },
    {
      "epoch": 0.2876218066894917,
      "grad_norm": 3.5884392261505127,
      "learning_rate": 3.5628127469054516e-05,
      "loss": 1.4427,
      "step": 10921
    },
    {
      "epoch": 0.2876481432710034,
      "grad_norm": 3.065458297729492,
      "learning_rate": 3.562681063997893e-05,
      "loss": 2.4932,
      "step": 10922
    },
    {
      "epoch": 0.28767447985251515,
      "grad_norm": 4.81726598739624,
      "learning_rate": 3.562549381090335e-05,
      "loss": 1.1851,
      "step": 10923
    },
    {
      "epoch": 0.28770081643402684,
      "grad_norm": 3.7395832538604736,
      "learning_rate": 3.562417698182776e-05,
      "loss": 0.9141,
      "step": 10924
    },
    {
      "epoch": 0.2877271530155386,
      "grad_norm": 1.4980483055114746,
      "learning_rate": 3.562286015275217e-05,
      "loss": 1.8753,
      "step": 10925
    },
    {
      "epoch": 0.2877534895970503,
      "grad_norm": 3.8506574630737305,
      "learning_rate": 3.562154332367659e-05,
      "loss": 0.9344,
      "step": 10926
    },
    {
      "epoch": 0.287779826178562,
      "grad_norm": 2.8683130741119385,
      "learning_rate": 3.5620226494601e-05,
      "loss": 1.1648,
      "step": 10927
    },
    {
      "epoch": 0.28780616276007376,
      "grad_norm": 2.4282519817352295,
      "learning_rate": 3.561890966552542e-05,
      "loss": 1.1954,
      "step": 10928
    },
    {
      "epoch": 0.28783249934158545,
      "grad_norm": 3.0133583545684814,
      "learning_rate": 3.5617592836449834e-05,
      "loss": 1.7148,
      "step": 10929
    },
    {
      "epoch": 0.2878588359230972,
      "grad_norm": 4.448825359344482,
      "learning_rate": 3.561627600737424e-05,
      "loss": 0.9899,
      "step": 10930
    },
    {
      "epoch": 0.2878851725046089,
      "grad_norm": 4.573001861572266,
      "learning_rate": 3.561495917829866e-05,
      "loss": 1.8964,
      "step": 10931
    },
    {
      "epoch": 0.2879115090861206,
      "grad_norm": 1.802765130996704,
      "learning_rate": 3.5613642349223074e-05,
      "loss": 1.4969,
      "step": 10932
    },
    {
      "epoch": 0.28793784566763236,
      "grad_norm": 2.5864810943603516,
      "learning_rate": 3.561232552014749e-05,
      "loss": 0.7321,
      "step": 10933
    },
    {
      "epoch": 0.28796418224914405,
      "grad_norm": 3.4646875858306885,
      "learning_rate": 3.56110086910719e-05,
      "loss": 1.1532,
      "step": 10934
    },
    {
      "epoch": 0.2879905188306558,
      "grad_norm": 1.8657139539718628,
      "learning_rate": 3.5609691861996314e-05,
      "loss": 2.0382,
      "step": 10935
    },
    {
      "epoch": 0.2880168554121675,
      "grad_norm": 2.163668155670166,
      "learning_rate": 3.560837503292073e-05,
      "loss": 1.452,
      "step": 10936
    },
    {
      "epoch": 0.2880431919936792,
      "grad_norm": 1.600176215171814,
      "learning_rate": 3.5607058203845145e-05,
      "loss": 1.9442,
      "step": 10937
    },
    {
      "epoch": 0.2880695285751909,
      "grad_norm": 3.5224242210388184,
      "learning_rate": 3.560574137476956e-05,
      "loss": 1.7202,
      "step": 10938
    },
    {
      "epoch": 0.28809586515670266,
      "grad_norm": 3.597471237182617,
      "learning_rate": 3.560442454569397e-05,
      "loss": 1.6232,
      "step": 10939
    },
    {
      "epoch": 0.2881222017382144,
      "grad_norm": 3.762521266937256,
      "learning_rate": 3.5603107716618385e-05,
      "loss": 1.1984,
      "step": 10940
    },
    {
      "epoch": 0.2881485383197261,
      "grad_norm": 4.7362589836120605,
      "learning_rate": 3.5601790887542794e-05,
      "loss": 0.9751,
      "step": 10941
    },
    {
      "epoch": 0.28817487490123783,
      "grad_norm": 2.2351064682006836,
      "learning_rate": 3.5600474058467216e-05,
      "loss": 2.4552,
      "step": 10942
    },
    {
      "epoch": 0.2882012114827495,
      "grad_norm": 3.0252811908721924,
      "learning_rate": 3.5599157229391625e-05,
      "loss": 1.0504,
      "step": 10943
    },
    {
      "epoch": 0.28822754806426126,
      "grad_norm": 2.5757875442504883,
      "learning_rate": 3.559784040031604e-05,
      "loss": 1.4499,
      "step": 10944
    },
    {
      "epoch": 0.288253884645773,
      "grad_norm": 2.449920892715454,
      "learning_rate": 3.5596523571240456e-05,
      "loss": 1.7631,
      "step": 10945
    },
    {
      "epoch": 0.2882802212272847,
      "grad_norm": 2.6157121658325195,
      "learning_rate": 3.5595206742164865e-05,
      "loss": 0.9906,
      "step": 10946
    },
    {
      "epoch": 0.28830655780879644,
      "grad_norm": 2.005641460418701,
      "learning_rate": 3.559388991308929e-05,
      "loss": 1.501,
      "step": 10947
    },
    {
      "epoch": 0.2883328943903081,
      "grad_norm": 2.2654950618743896,
      "learning_rate": 3.5592573084013696e-05,
      "loss": 1.2623,
      "step": 10948
    },
    {
      "epoch": 0.28835923097181987,
      "grad_norm": 3.0298023223876953,
      "learning_rate": 3.559125625493811e-05,
      "loss": 0.9493,
      "step": 10949
    },
    {
      "epoch": 0.28838556755333156,
      "grad_norm": 2.7169744968414307,
      "learning_rate": 3.558993942586252e-05,
      "loss": 1.6912,
      "step": 10950
    },
    {
      "epoch": 0.2884119041348433,
      "grad_norm": 5.633264064788818,
      "learning_rate": 3.558862259678694e-05,
      "loss": 1.1934,
      "step": 10951
    },
    {
      "epoch": 0.28843824071635504,
      "grad_norm": 1.7106952667236328,
      "learning_rate": 3.558730576771135e-05,
      "loss": 1.1364,
      "step": 10952
    },
    {
      "epoch": 0.28846457729786673,
      "grad_norm": 1.6821469068527222,
      "learning_rate": 3.558598893863577e-05,
      "loss": 2.2684,
      "step": 10953
    },
    {
      "epoch": 0.2884909138793785,
      "grad_norm": 5.259078502655029,
      "learning_rate": 3.558467210956018e-05,
      "loss": 1.0155,
      "step": 10954
    },
    {
      "epoch": 0.28851725046089016,
      "grad_norm": 1.5545750856399536,
      "learning_rate": 3.558335528048459e-05,
      "loss": 1.2766,
      "step": 10955
    },
    {
      "epoch": 0.2885435870424019,
      "grad_norm": 1.6793551445007324,
      "learning_rate": 3.5582038451409014e-05,
      "loss": 2.3053,
      "step": 10956
    },
    {
      "epoch": 0.2885699236239136,
      "grad_norm": 2.053560256958008,
      "learning_rate": 3.558072162233342e-05,
      "loss": 1.2424,
      "step": 10957
    },
    {
      "epoch": 0.28859626020542534,
      "grad_norm": 1.6040992736816406,
      "learning_rate": 3.557940479325784e-05,
      "loss": 1.8083,
      "step": 10958
    },
    {
      "epoch": 0.2886225967869371,
      "grad_norm": 2.1622772216796875,
      "learning_rate": 3.557808796418225e-05,
      "loss": 2.4501,
      "step": 10959
    },
    {
      "epoch": 0.28864893336844877,
      "grad_norm": 2.141171932220459,
      "learning_rate": 3.557677113510666e-05,
      "loss": 1.5421,
      "step": 10960
    },
    {
      "epoch": 0.2886752699499605,
      "grad_norm": 3.0063464641571045,
      "learning_rate": 3.557545430603108e-05,
      "loss": 1.9009,
      "step": 10961
    },
    {
      "epoch": 0.2887016065314722,
      "grad_norm": 2.160053253173828,
      "learning_rate": 3.5574137476955494e-05,
      "loss": 0.685,
      "step": 10962
    },
    {
      "epoch": 0.28872794311298394,
      "grad_norm": 1.8533016443252563,
      "learning_rate": 3.55728206478799e-05,
      "loss": 1.9313,
      "step": 10963
    },
    {
      "epoch": 0.28875427969449563,
      "grad_norm": 2.5866363048553467,
      "learning_rate": 3.557150381880432e-05,
      "loss": 1.9037,
      "step": 10964
    },
    {
      "epoch": 0.2887806162760074,
      "grad_norm": 1.7641345262527466,
      "learning_rate": 3.5570186989728734e-05,
      "loss": 0.2565,
      "step": 10965
    },
    {
      "epoch": 0.2888069528575191,
      "grad_norm": 3.6756820678710938,
      "learning_rate": 3.556887016065315e-05,
      "loss": 1.713,
      "step": 10966
    },
    {
      "epoch": 0.2888332894390308,
      "grad_norm": 2.8689334392547607,
      "learning_rate": 3.5567553331577565e-05,
      "loss": 0.7,
      "step": 10967
    },
    {
      "epoch": 0.28885962602054255,
      "grad_norm": 2.9871444702148438,
      "learning_rate": 3.5566236502501974e-05,
      "loss": 1.5818,
      "step": 10968
    },
    {
      "epoch": 0.28888596260205424,
      "grad_norm": 1.6511298418045044,
      "learning_rate": 3.556491967342639e-05,
      "loss": 2.0819,
      "step": 10969
    },
    {
      "epoch": 0.288912299183566,
      "grad_norm": 2.019268274307251,
      "learning_rate": 3.5563602844350806e-05,
      "loss": 1.8302,
      "step": 10970
    },
    {
      "epoch": 0.28893863576507767,
      "grad_norm": 2.2264962196350098,
      "learning_rate": 3.556228601527522e-05,
      "loss": 1.5024,
      "step": 10971
    },
    {
      "epoch": 0.2889649723465894,
      "grad_norm": 1.9779112339019775,
      "learning_rate": 3.556096918619963e-05,
      "loss": 2.0727,
      "step": 10972
    },
    {
      "epoch": 0.28899130892810115,
      "grad_norm": 2.1315879821777344,
      "learning_rate": 3.5559652357124046e-05,
      "loss": 2.098,
      "step": 10973
    },
    {
      "epoch": 0.28901764550961284,
      "grad_norm": 3.3385040760040283,
      "learning_rate": 3.555833552804846e-05,
      "loss": 1.0683,
      "step": 10974
    },
    {
      "epoch": 0.2890439820911246,
      "grad_norm": 2.2172985076904297,
      "learning_rate": 3.555701869897288e-05,
      "loss": 1.993,
      "step": 10975
    },
    {
      "epoch": 0.2890703186726363,
      "grad_norm": 2.3556089401245117,
      "learning_rate": 3.555570186989729e-05,
      "loss": 2.4113,
      "step": 10976
    },
    {
      "epoch": 0.289096655254148,
      "grad_norm": 1.7501848936080933,
      "learning_rate": 3.55543850408217e-05,
      "loss": 1.1491,
      "step": 10977
    },
    {
      "epoch": 0.2891229918356597,
      "grad_norm": 2.6829991340637207,
      "learning_rate": 3.555306821174612e-05,
      "loss": 1.0855,
      "step": 10978
    },
    {
      "epoch": 0.28914932841717145,
      "grad_norm": 2.8630950450897217,
      "learning_rate": 3.5551751382670526e-05,
      "loss": 2.0312,
      "step": 10979
    },
    {
      "epoch": 0.2891756649986832,
      "grad_norm": 2.9446446895599365,
      "learning_rate": 3.555043455359495e-05,
      "loss": 1.4823,
      "step": 10980
    },
    {
      "epoch": 0.2892020015801949,
      "grad_norm": 2.8820769786834717,
      "learning_rate": 3.554911772451936e-05,
      "loss": 0.581,
      "step": 10981
    },
    {
      "epoch": 0.2892283381617066,
      "grad_norm": 2.2731316089630127,
      "learning_rate": 3.554780089544377e-05,
      "loss": 1.4354,
      "step": 10982
    },
    {
      "epoch": 0.2892546747432183,
      "grad_norm": 2.1192312240600586,
      "learning_rate": 3.554648406636819e-05,
      "loss": 1.2348,
      "step": 10983
    },
    {
      "epoch": 0.28928101132473005,
      "grad_norm": 2.931610345840454,
      "learning_rate": 3.5545167237292604e-05,
      "loss": 1.7275,
      "step": 10984
    },
    {
      "epoch": 0.2893073479062418,
      "grad_norm": 1.4986823797225952,
      "learning_rate": 3.554385040821702e-05,
      "loss": 1.4245,
      "step": 10985
    },
    {
      "epoch": 0.2893336844877535,
      "grad_norm": 3.0312817096710205,
      "learning_rate": 3.554253357914143e-05,
      "loss": 1.4358,
      "step": 10986
    },
    {
      "epoch": 0.2893600210692652,
      "grad_norm": 1.7814475297927856,
      "learning_rate": 3.5541216750065844e-05,
      "loss": 1.8522,
      "step": 10987
    },
    {
      "epoch": 0.2893863576507769,
      "grad_norm": 1.8796284198760986,
      "learning_rate": 3.553989992099025e-05,
      "loss": 0.4629,
      "step": 10988
    },
    {
      "epoch": 0.28941269423228866,
      "grad_norm": 5.068589687347412,
      "learning_rate": 3.5538583091914675e-05,
      "loss": 0.8754,
      "step": 10989
    },
    {
      "epoch": 0.28943903081380035,
      "grad_norm": 5.044485569000244,
      "learning_rate": 3.5537266262839084e-05,
      "loss": 1.4031,
      "step": 10990
    },
    {
      "epoch": 0.2894653673953121,
      "grad_norm": 3.1948459148406982,
      "learning_rate": 3.55359494337635e-05,
      "loss": 1.2049,
      "step": 10991
    },
    {
      "epoch": 0.28949170397682383,
      "grad_norm": 5.329652786254883,
      "learning_rate": 3.5534632604687915e-05,
      "loss": 2.9056,
      "step": 10992
    },
    {
      "epoch": 0.2895180405583355,
      "grad_norm": 1.6377066373825073,
      "learning_rate": 3.5533315775612324e-05,
      "loss": 1.7653,
      "step": 10993
    },
    {
      "epoch": 0.28954437713984726,
      "grad_norm": 3.317782163619995,
      "learning_rate": 3.5531998946536746e-05,
      "loss": 0.3472,
      "step": 10994
    },
    {
      "epoch": 0.28957071372135895,
      "grad_norm": 1.922645092010498,
      "learning_rate": 3.5530682117461155e-05,
      "loss": 1.3858,
      "step": 10995
    },
    {
      "epoch": 0.2895970503028707,
      "grad_norm": 4.561474800109863,
      "learning_rate": 3.552936528838557e-05,
      "loss": 0.8801,
      "step": 10996
    },
    {
      "epoch": 0.2896233868843824,
      "grad_norm": 2.3178837299346924,
      "learning_rate": 3.552804845930998e-05,
      "loss": 2.066,
      "step": 10997
    },
    {
      "epoch": 0.2896497234658941,
      "grad_norm": 2.91257905960083,
      "learning_rate": 3.55267316302344e-05,
      "loss": 1.4585,
      "step": 10998
    },
    {
      "epoch": 0.28967606004740587,
      "grad_norm": 2.5721254348754883,
      "learning_rate": 3.552541480115881e-05,
      "loss": 2.8255,
      "step": 10999
    },
    {
      "epoch": 0.28970239662891756,
      "grad_norm": 4.280350685119629,
      "learning_rate": 3.5524097972083226e-05,
      "loss": 1.5415,
      "step": 11000
    },
    {
      "epoch": 0.2897287332104293,
      "grad_norm": 2.1255195140838623,
      "learning_rate": 3.552278114300764e-05,
      "loss": 1.6239,
      "step": 11001
    },
    {
      "epoch": 0.289755069791941,
      "grad_norm": 1.4986419677734375,
      "learning_rate": 3.552146431393205e-05,
      "loss": 1.1223,
      "step": 11002
    },
    {
      "epoch": 0.28978140637345273,
      "grad_norm": 1.7992783784866333,
      "learning_rate": 3.552014748485647e-05,
      "loss": 2.0119,
      "step": 11003
    },
    {
      "epoch": 0.2898077429549644,
      "grad_norm": 2.8025548458099365,
      "learning_rate": 3.551883065578088e-05,
      "loss": 1.935,
      "step": 11004
    },
    {
      "epoch": 0.28983407953647616,
      "grad_norm": 4.0590410232543945,
      "learning_rate": 3.55175138267053e-05,
      "loss": 1.0777,
      "step": 11005
    },
    {
      "epoch": 0.2898604161179879,
      "grad_norm": 3.287367105484009,
      "learning_rate": 3.5516196997629706e-05,
      "loss": 1.6116,
      "step": 11006
    },
    {
      "epoch": 0.2898867526994996,
      "grad_norm": 2.9920408725738525,
      "learning_rate": 3.551488016855412e-05,
      "loss": 0.858,
      "step": 11007
    },
    {
      "epoch": 0.28991308928101134,
      "grad_norm": 1.7735719680786133,
      "learning_rate": 3.551356333947854e-05,
      "loss": 1.7671,
      "step": 11008
    },
    {
      "epoch": 0.289939425862523,
      "grad_norm": 1.7227829694747925,
      "learning_rate": 3.551224651040295e-05,
      "loss": 1.6007,
      "step": 11009
    },
    {
      "epoch": 0.28996576244403477,
      "grad_norm": 1.7703499794006348,
      "learning_rate": 3.551092968132737e-05,
      "loss": 1.7369,
      "step": 11010
    },
    {
      "epoch": 0.28999209902554646,
      "grad_norm": 2.5169200897216797,
      "learning_rate": 3.550961285225178e-05,
      "loss": 1.6055,
      "step": 11011
    },
    {
      "epoch": 0.2900184356070582,
      "grad_norm": 2.7595176696777344,
      "learning_rate": 3.550829602317619e-05,
      "loss": 1.7946,
      "step": 11012
    },
    {
      "epoch": 0.29004477218856994,
      "grad_norm": 2.930305004119873,
      "learning_rate": 3.550697919410061e-05,
      "loss": 2.2925,
      "step": 11013
    },
    {
      "epoch": 0.29007110877008163,
      "grad_norm": 1.7983318567276,
      "learning_rate": 3.5505662365025024e-05,
      "loss": 0.4155,
      "step": 11014
    },
    {
      "epoch": 0.2900974453515934,
      "grad_norm": 3.9221348762512207,
      "learning_rate": 3.550434553594943e-05,
      "loss": 1.0817,
      "step": 11015
    },
    {
      "epoch": 0.29012378193310506,
      "grad_norm": 2.2868871688842773,
      "learning_rate": 3.550302870687385e-05,
      "loss": 2.1247,
      "step": 11016
    },
    {
      "epoch": 0.2901501185146168,
      "grad_norm": 7.007227897644043,
      "learning_rate": 3.5501711877798264e-05,
      "loss": 1.0384,
      "step": 11017
    },
    {
      "epoch": 0.29017645509612855,
      "grad_norm": 2.3141703605651855,
      "learning_rate": 3.550039504872268e-05,
      "loss": 1.6293,
      "step": 11018
    },
    {
      "epoch": 0.29020279167764024,
      "grad_norm": 3.3522896766662598,
      "learning_rate": 3.549907821964709e-05,
      "loss": 1.8453,
      "step": 11019
    },
    {
      "epoch": 0.290229128259152,
      "grad_norm": 3.965550422668457,
      "learning_rate": 3.5497761390571504e-05,
      "loss": 1.1338,
      "step": 11020
    },
    {
      "epoch": 0.29025546484066367,
      "grad_norm": 6.852443218231201,
      "learning_rate": 3.549644456149592e-05,
      "loss": 0.908,
      "step": 11021
    },
    {
      "epoch": 0.2902818014221754,
      "grad_norm": 3.539928436279297,
      "learning_rate": 3.5495127732420335e-05,
      "loss": 1.3154,
      "step": 11022
    },
    {
      "epoch": 0.2903081380036871,
      "grad_norm": 3.4951250553131104,
      "learning_rate": 3.549381090334475e-05,
      "loss": 1.2941,
      "step": 11023
    },
    {
      "epoch": 0.29033447458519884,
      "grad_norm": 2.771512269973755,
      "learning_rate": 3.549249407426916e-05,
      "loss": 0.613,
      "step": 11024
    },
    {
      "epoch": 0.2903608111667106,
      "grad_norm": 2.6112723350524902,
      "learning_rate": 3.5491177245193575e-05,
      "loss": 2.3057,
      "step": 11025
    },
    {
      "epoch": 0.2903871477482223,
      "grad_norm": 2.004936695098877,
      "learning_rate": 3.5489860416117984e-05,
      "loss": 1.7175,
      "step": 11026
    },
    {
      "epoch": 0.290413484329734,
      "grad_norm": 1.655478835105896,
      "learning_rate": 3.5488543587042406e-05,
      "loss": 1.7248,
      "step": 11027
    },
    {
      "epoch": 0.2904398209112457,
      "grad_norm": 2.637791633605957,
      "learning_rate": 3.5487226757966815e-05,
      "loss": 1.5018,
      "step": 11028
    },
    {
      "epoch": 0.29046615749275745,
      "grad_norm": 1.5167251825332642,
      "learning_rate": 3.548590992889123e-05,
      "loss": 2.2552,
      "step": 11029
    },
    {
      "epoch": 0.29049249407426914,
      "grad_norm": 2.9432373046875,
      "learning_rate": 3.5484593099815647e-05,
      "loss": 1.998,
      "step": 11030
    },
    {
      "epoch": 0.2905188306557809,
      "grad_norm": 2.922551393508911,
      "learning_rate": 3.548327627074006e-05,
      "loss": 1.5541,
      "step": 11031
    },
    {
      "epoch": 0.2905451672372926,
      "grad_norm": 6.568065643310547,
      "learning_rate": 3.548195944166448e-05,
      "loss": 1.1314,
      "step": 11032
    },
    {
      "epoch": 0.2905715038188043,
      "grad_norm": 2.2814013957977295,
      "learning_rate": 3.5480642612588887e-05,
      "loss": 2.2513,
      "step": 11033
    },
    {
      "epoch": 0.29059784040031605,
      "grad_norm": 2.166842222213745,
      "learning_rate": 3.54793257835133e-05,
      "loss": 1.9773,
      "step": 11034
    },
    {
      "epoch": 0.29062417698182774,
      "grad_norm": 1.8020366430282593,
      "learning_rate": 3.547800895443771e-05,
      "loss": 2.4285,
      "step": 11035
    },
    {
      "epoch": 0.2906505135633395,
      "grad_norm": 1.7066317796707153,
      "learning_rate": 3.547669212536213e-05,
      "loss": 2.1236,
      "step": 11036
    },
    {
      "epoch": 0.2906768501448512,
      "grad_norm": 2.9069323539733887,
      "learning_rate": 3.547537529628654e-05,
      "loss": 2.0676,
      "step": 11037
    },
    {
      "epoch": 0.2907031867263629,
      "grad_norm": 1.9107900857925415,
      "learning_rate": 3.547405846721096e-05,
      "loss": 1.7973,
      "step": 11038
    },
    {
      "epoch": 0.29072952330787466,
      "grad_norm": 2.988912343978882,
      "learning_rate": 3.547274163813537e-05,
      "loss": 0.4131,
      "step": 11039
    },
    {
      "epoch": 0.29075585988938635,
      "grad_norm": 1.850653052330017,
      "learning_rate": 3.547142480905978e-05,
      "loss": 1.9137,
      "step": 11040
    },
    {
      "epoch": 0.2907821964708981,
      "grad_norm": 1.9713221788406372,
      "learning_rate": 3.5470107979984205e-05,
      "loss": 2.2683,
      "step": 11041
    },
    {
      "epoch": 0.2908085330524098,
      "grad_norm": 2.0246012210845947,
      "learning_rate": 3.546879115090861e-05,
      "loss": 2.0447,
      "step": 11042
    },
    {
      "epoch": 0.2908348696339215,
      "grad_norm": 4.18183708190918,
      "learning_rate": 3.546747432183303e-05,
      "loss": 1.5515,
      "step": 11043
    },
    {
      "epoch": 0.2908612062154332,
      "grad_norm": 3.0166642665863037,
      "learning_rate": 3.546615749275744e-05,
      "loss": 1.4106,
      "step": 11044
    },
    {
      "epoch": 0.29088754279694495,
      "grad_norm": 2.317770481109619,
      "learning_rate": 3.546484066368185e-05,
      "loss": 0.3382,
      "step": 11045
    },
    {
      "epoch": 0.2909138793784567,
      "grad_norm": 2.7255659103393555,
      "learning_rate": 3.546352383460627e-05,
      "loss": 0.4388,
      "step": 11046
    },
    {
      "epoch": 0.2909402159599684,
      "grad_norm": 1.7485480308532715,
      "learning_rate": 3.5462207005530685e-05,
      "loss": 1.8356,
      "step": 11047
    },
    {
      "epoch": 0.29096655254148013,
      "grad_norm": 3.4362640380859375,
      "learning_rate": 3.54608901764551e-05,
      "loss": 1.097,
      "step": 11048
    },
    {
      "epoch": 0.2909928891229918,
      "grad_norm": 2.0705902576446533,
      "learning_rate": 3.545957334737951e-05,
      "loss": 2.1396,
      "step": 11049
    },
    {
      "epoch": 0.29101922570450356,
      "grad_norm": 2.3216209411621094,
      "learning_rate": 3.545825651830393e-05,
      "loss": 1.3643,
      "step": 11050
    },
    {
      "epoch": 0.2910455622860153,
      "grad_norm": 2.6257174015045166,
      "learning_rate": 3.545693968922834e-05,
      "loss": 1.7138,
      "step": 11051
    },
    {
      "epoch": 0.291071898867527,
      "grad_norm": 2.5664923191070557,
      "learning_rate": 3.5455622860152756e-05,
      "loss": 1.2035,
      "step": 11052
    },
    {
      "epoch": 0.29109823544903873,
      "grad_norm": 2.6241204738616943,
      "learning_rate": 3.5454306031077165e-05,
      "loss": 2.0248,
      "step": 11053
    },
    {
      "epoch": 0.2911245720305504,
      "grad_norm": 2.7061100006103516,
      "learning_rate": 3.545298920200158e-05,
      "loss": 1.7668,
      "step": 11054
    },
    {
      "epoch": 0.29115090861206216,
      "grad_norm": 2.0117859840393066,
      "learning_rate": 3.5451672372925996e-05,
      "loss": 1.9369,
      "step": 11055
    },
    {
      "epoch": 0.29117724519357385,
      "grad_norm": 3.36904239654541,
      "learning_rate": 3.545035554385041e-05,
      "loss": 0.7704,
      "step": 11056
    },
    {
      "epoch": 0.2912035817750856,
      "grad_norm": 2.7240214347839355,
      "learning_rate": 3.544903871477483e-05,
      "loss": 1.8645,
      "step": 11057
    },
    {
      "epoch": 0.29122991835659734,
      "grad_norm": 3.428537607192993,
      "learning_rate": 3.5447721885699236e-05,
      "loss": 0.3247,
      "step": 11058
    },
    {
      "epoch": 0.291256254938109,
      "grad_norm": 2.109004497528076,
      "learning_rate": 3.544640505662365e-05,
      "loss": 2.179,
      "step": 11059
    },
    {
      "epoch": 0.29128259151962077,
      "grad_norm": 2.1439788341522217,
      "learning_rate": 3.544508822754807e-05,
      "loss": 1.7094,
      "step": 11060
    },
    {
      "epoch": 0.29130892810113246,
      "grad_norm": 3.4488956928253174,
      "learning_rate": 3.544377139847248e-05,
      "loss": 2.4145,
      "step": 11061
    },
    {
      "epoch": 0.2913352646826442,
      "grad_norm": 5.650046348571777,
      "learning_rate": 3.544245456939689e-05,
      "loss": 1.8215,
      "step": 11062
    },
    {
      "epoch": 0.2913616012641559,
      "grad_norm": 1.7302626371383667,
      "learning_rate": 3.544113774032131e-05,
      "loss": 1.3273,
      "step": 11063
    },
    {
      "epoch": 0.29138793784566763,
      "grad_norm": 3.058371067047119,
      "learning_rate": 3.543982091124572e-05,
      "loss": 2.2555,
      "step": 11064
    },
    {
      "epoch": 0.2914142744271794,
      "grad_norm": 5.631766319274902,
      "learning_rate": 3.543850408217014e-05,
      "loss": 1.1217,
      "step": 11065
    },
    {
      "epoch": 0.29144061100869106,
      "grad_norm": 3.0049452781677246,
      "learning_rate": 3.543718725309455e-05,
      "loss": 1.5222,
      "step": 11066
    },
    {
      "epoch": 0.2914669475902028,
      "grad_norm": 2.9334356784820557,
      "learning_rate": 3.543587042401896e-05,
      "loss": 0.6939,
      "step": 11067
    },
    {
      "epoch": 0.2914932841717145,
      "grad_norm": 2.0152101516723633,
      "learning_rate": 3.543455359494338e-05,
      "loss": 2.1507,
      "step": 11068
    },
    {
      "epoch": 0.29151962075322624,
      "grad_norm": 2.8079895973205566,
      "learning_rate": 3.5433236765867794e-05,
      "loss": 1.7229,
      "step": 11069
    },
    {
      "epoch": 0.2915459573347379,
      "grad_norm": 5.658259391784668,
      "learning_rate": 3.543191993679221e-05,
      "loss": 1.9062,
      "step": 11070
    },
    {
      "epoch": 0.29157229391624967,
      "grad_norm": 3.0796990394592285,
      "learning_rate": 3.543060310771662e-05,
      "loss": 1.8644,
      "step": 11071
    },
    {
      "epoch": 0.2915986304977614,
      "grad_norm": 1.983187198638916,
      "learning_rate": 3.5429286278641034e-05,
      "loss": 0.4625,
      "step": 11072
    },
    {
      "epoch": 0.2916249670792731,
      "grad_norm": 1.7298763990402222,
      "learning_rate": 3.542796944956544e-05,
      "loss": 2.4248,
      "step": 11073
    },
    {
      "epoch": 0.29165130366078484,
      "grad_norm": 5.213994979858398,
      "learning_rate": 3.5426652620489865e-05,
      "loss": 1.5353,
      "step": 11074
    },
    {
      "epoch": 0.29167764024229653,
      "grad_norm": 2.2726612091064453,
      "learning_rate": 3.5425335791414274e-05,
      "loss": 1.3574,
      "step": 11075
    },
    {
      "epoch": 0.2917039768238083,
      "grad_norm": 1.9792426824569702,
      "learning_rate": 3.542401896233869e-05,
      "loss": 1.6951,
      "step": 11076
    },
    {
      "epoch": 0.29173031340531996,
      "grad_norm": 2.554072141647339,
      "learning_rate": 3.5422702133263105e-05,
      "loss": 1.8157,
      "step": 11077
    },
    {
      "epoch": 0.2917566499868317,
      "grad_norm": 1.6620303392410278,
      "learning_rate": 3.5421385304187514e-05,
      "loss": 2.1404,
      "step": 11078
    },
    {
      "epoch": 0.29178298656834345,
      "grad_norm": 2.2123641967773438,
      "learning_rate": 3.5420068475111936e-05,
      "loss": 1.5199,
      "step": 11079
    },
    {
      "epoch": 0.29180932314985514,
      "grad_norm": 1.6105784177780151,
      "learning_rate": 3.5418751646036345e-05,
      "loss": 2.1827,
      "step": 11080
    },
    {
      "epoch": 0.2918356597313669,
      "grad_norm": 2.6802256107330322,
      "learning_rate": 3.541743481696076e-05,
      "loss": 1.2632,
      "step": 11081
    },
    {
      "epoch": 0.29186199631287857,
      "grad_norm": 2.152794361114502,
      "learning_rate": 3.541611798788517e-05,
      "loss": 0.2989,
      "step": 11082
    },
    {
      "epoch": 0.2918883328943903,
      "grad_norm": 2.596158504486084,
      "learning_rate": 3.541480115880959e-05,
      "loss": 1.2878,
      "step": 11083
    },
    {
      "epoch": 0.29191466947590206,
      "grad_norm": 3.5145933628082275,
      "learning_rate": 3.5413484329734e-05,
      "loss": 1.6906,
      "step": 11084
    },
    {
      "epoch": 0.29194100605741374,
      "grad_norm": 1.7567402124404907,
      "learning_rate": 3.5412167500658416e-05,
      "loss": 1.4828,
      "step": 11085
    },
    {
      "epoch": 0.2919673426389255,
      "grad_norm": 3.0997812747955322,
      "learning_rate": 3.541085067158283e-05,
      "loss": 0.6779,
      "step": 11086
    },
    {
      "epoch": 0.2919936792204372,
      "grad_norm": 1.9743634462356567,
      "learning_rate": 3.540953384250724e-05,
      "loss": 2.1128,
      "step": 11087
    },
    {
      "epoch": 0.2920200158019489,
      "grad_norm": 7.771797180175781,
      "learning_rate": 3.540821701343166e-05,
      "loss": 1.4228,
      "step": 11088
    },
    {
      "epoch": 0.2920463523834606,
      "grad_norm": 1.853283405303955,
      "learning_rate": 3.540690018435607e-05,
      "loss": 1.5985,
      "step": 11089
    },
    {
      "epoch": 0.29207268896497235,
      "grad_norm": 2.664031744003296,
      "learning_rate": 3.540558335528049e-05,
      "loss": 1.308,
      "step": 11090
    },
    {
      "epoch": 0.2920990255464841,
      "grad_norm": 6.335636615753174,
      "learning_rate": 3.5404266526204896e-05,
      "loss": 2.1488,
      "step": 11091
    },
    {
      "epoch": 0.2921253621279958,
      "grad_norm": 1.7049521207809448,
      "learning_rate": 3.540294969712931e-05,
      "loss": 1.9945,
      "step": 11092
    },
    {
      "epoch": 0.2921516987095075,
      "grad_norm": 3.685929775238037,
      "learning_rate": 3.540163286805373e-05,
      "loss": 0.7921,
      "step": 11093
    },
    {
      "epoch": 0.2921780352910192,
      "grad_norm": 1.8595293760299683,
      "learning_rate": 3.540031603897814e-05,
      "loss": 0.6721,
      "step": 11094
    },
    {
      "epoch": 0.29220437187253095,
      "grad_norm": 3.9109089374542236,
      "learning_rate": 3.539899920990256e-05,
      "loss": 1.0977,
      "step": 11095
    },
    {
      "epoch": 0.29223070845404264,
      "grad_norm": 2.1876718997955322,
      "learning_rate": 3.539768238082697e-05,
      "loss": 1.4682,
      "step": 11096
    },
    {
      "epoch": 0.2922570450355544,
      "grad_norm": 3.5205602645874023,
      "learning_rate": 3.539636555175139e-05,
      "loss": 1.332,
      "step": 11097
    },
    {
      "epoch": 0.29228338161706613,
      "grad_norm": 2.544851064682007,
      "learning_rate": 3.53950487226758e-05,
      "loss": 2.1016,
      "step": 11098
    },
    {
      "epoch": 0.2923097181985778,
      "grad_norm": 2.9030494689941406,
      "learning_rate": 3.5393731893600214e-05,
      "loss": 1.1756,
      "step": 11099
    },
    {
      "epoch": 0.29233605478008956,
      "grad_norm": 2.0888679027557373,
      "learning_rate": 3.539241506452462e-05,
      "loss": 1.5785,
      "step": 11100
    },
    {
      "epoch": 0.29236239136160125,
      "grad_norm": 2.51741361618042,
      "learning_rate": 3.539109823544904e-05,
      "loss": 1.7483,
      "step": 11101
    },
    {
      "epoch": 0.292388727943113,
      "grad_norm": 2.9861435890197754,
      "learning_rate": 3.5389781406373454e-05,
      "loss": 1.9696,
      "step": 11102
    },
    {
      "epoch": 0.2924150645246247,
      "grad_norm": 1.9171050786972046,
      "learning_rate": 3.538846457729787e-05,
      "loss": 1.9575,
      "step": 11103
    },
    {
      "epoch": 0.2924414011061364,
      "grad_norm": 1.6457089185714722,
      "learning_rate": 3.5387147748222286e-05,
      "loss": 1.7112,
      "step": 11104
    },
    {
      "epoch": 0.29246773768764817,
      "grad_norm": 2.636373996734619,
      "learning_rate": 3.5385830919146694e-05,
      "loss": 1.5134,
      "step": 11105
    },
    {
      "epoch": 0.29249407426915985,
      "grad_norm": 2.703622579574585,
      "learning_rate": 3.538451409007111e-05,
      "loss": 1.1645,
      "step": 11106
    },
    {
      "epoch": 0.2925204108506716,
      "grad_norm": 3.1340646743774414,
      "learning_rate": 3.5383197260995526e-05,
      "loss": 1.7501,
      "step": 11107
    },
    {
      "epoch": 0.2925467474321833,
      "grad_norm": 5.145357608795166,
      "learning_rate": 3.538188043191994e-05,
      "loss": 1.4914,
      "step": 11108
    },
    {
      "epoch": 0.29257308401369503,
      "grad_norm": 9.81971263885498,
      "learning_rate": 3.538056360284435e-05,
      "loss": 2.3546,
      "step": 11109
    },
    {
      "epoch": 0.2925994205952067,
      "grad_norm": 3.489922046661377,
      "learning_rate": 3.5379246773768766e-05,
      "loss": 1.9695,
      "step": 11110
    },
    {
      "epoch": 0.29262575717671846,
      "grad_norm": 2.2756972312927246,
      "learning_rate": 3.5377929944693174e-05,
      "loss": 2.0647,
      "step": 11111
    },
    {
      "epoch": 0.2926520937582302,
      "grad_norm": 4.781240940093994,
      "learning_rate": 3.53766131156176e-05,
      "loss": 2.4186,
      "step": 11112
    },
    {
      "epoch": 0.2926784303397419,
      "grad_norm": 1.6885876655578613,
      "learning_rate": 3.537529628654201e-05,
      "loss": 2.1346,
      "step": 11113
    },
    {
      "epoch": 0.29270476692125363,
      "grad_norm": 5.172010898590088,
      "learning_rate": 3.537397945746642e-05,
      "loss": 2.4652,
      "step": 11114
    },
    {
      "epoch": 0.2927311035027653,
      "grad_norm": 1.7945427894592285,
      "learning_rate": 3.537266262839084e-05,
      "loss": 2.0324,
      "step": 11115
    },
    {
      "epoch": 0.29275744008427707,
      "grad_norm": 2.261831283569336,
      "learning_rate": 3.537134579931525e-05,
      "loss": 2.0637,
      "step": 11116
    },
    {
      "epoch": 0.29278377666578875,
      "grad_norm": 1.7589129209518433,
      "learning_rate": 3.537002897023967e-05,
      "loss": 2.0809,
      "step": 11117
    },
    {
      "epoch": 0.2928101132473005,
      "grad_norm": 3.1029694080352783,
      "learning_rate": 3.536871214116408e-05,
      "loss": 1.7428,
      "step": 11118
    },
    {
      "epoch": 0.29283644982881224,
      "grad_norm": 3.1686432361602783,
      "learning_rate": 3.536739531208849e-05,
      "loss": 0.6834,
      "step": 11119
    },
    {
      "epoch": 0.2928627864103239,
      "grad_norm": 10.900420188903809,
      "learning_rate": 3.53660784830129e-05,
      "loss": 1.2351,
      "step": 11120
    },
    {
      "epoch": 0.29288912299183567,
      "grad_norm": 2.672062635421753,
      "learning_rate": 3.5364761653937324e-05,
      "loss": 1.7464,
      "step": 11121
    },
    {
      "epoch": 0.29291545957334736,
      "grad_norm": 2.002274751663208,
      "learning_rate": 3.536344482486173e-05,
      "loss": 0.914,
      "step": 11122
    },
    {
      "epoch": 0.2929417961548591,
      "grad_norm": 1.8139182329177856,
      "learning_rate": 3.536212799578615e-05,
      "loss": 1.5855,
      "step": 11123
    },
    {
      "epoch": 0.29296813273637085,
      "grad_norm": 1.4487301111221313,
      "learning_rate": 3.5360811166710564e-05,
      "loss": 1.7363,
      "step": 11124
    },
    {
      "epoch": 0.29299446931788253,
      "grad_norm": 3.9074110984802246,
      "learning_rate": 3.535949433763497e-05,
      "loss": 2.0718,
      "step": 11125
    },
    {
      "epoch": 0.2930208058993943,
      "grad_norm": 1.735677719116211,
      "learning_rate": 3.5358177508559395e-05,
      "loss": 1.5249,
      "step": 11126
    },
    {
      "epoch": 0.29304714248090596,
      "grad_norm": 2.083416700363159,
      "learning_rate": 3.5356860679483804e-05,
      "loss": 1.5842,
      "step": 11127
    },
    {
      "epoch": 0.2930734790624177,
      "grad_norm": 2.062920331954956,
      "learning_rate": 3.535554385040822e-05,
      "loss": 1.8368,
      "step": 11128
    },
    {
      "epoch": 0.2930998156439294,
      "grad_norm": 3.423672914505005,
      "learning_rate": 3.535422702133263e-05,
      "loss": 2.1068,
      "step": 11129
    },
    {
      "epoch": 0.29312615222544114,
      "grad_norm": 2.391085624694824,
      "learning_rate": 3.535291019225705e-05,
      "loss": 2.1505,
      "step": 11130
    },
    {
      "epoch": 0.2931524888069529,
      "grad_norm": 1.9471317529678345,
      "learning_rate": 3.535159336318146e-05,
      "loss": 1.7257,
      "step": 11131
    },
    {
      "epoch": 0.29317882538846457,
      "grad_norm": 1.749572515487671,
      "learning_rate": 3.5350276534105875e-05,
      "loss": 1.8632,
      "step": 11132
    },
    {
      "epoch": 0.2932051619699763,
      "grad_norm": 1.7597603797912598,
      "learning_rate": 3.534895970503029e-05,
      "loss": 1.4161,
      "step": 11133
    },
    {
      "epoch": 0.293231498551488,
      "grad_norm": 3.6342952251434326,
      "learning_rate": 3.53476428759547e-05,
      "loss": 2.1602,
      "step": 11134
    },
    {
      "epoch": 0.29325783513299974,
      "grad_norm": 3.2049062252044678,
      "learning_rate": 3.534632604687912e-05,
      "loss": 2.4722,
      "step": 11135
    },
    {
      "epoch": 0.29328417171451143,
      "grad_norm": 3.002648115158081,
      "learning_rate": 3.534500921780353e-05,
      "loss": 1.0007,
      "step": 11136
    },
    {
      "epoch": 0.2933105082960232,
      "grad_norm": 2.1872189044952393,
      "learning_rate": 3.5343692388727946e-05,
      "loss": 1.8473,
      "step": 11137
    },
    {
      "epoch": 0.2933368448775349,
      "grad_norm": 2.237682342529297,
      "learning_rate": 3.5342375559652355e-05,
      "loss": 1.7228,
      "step": 11138
    },
    {
      "epoch": 0.2933631814590466,
      "grad_norm": 2.02057147026062,
      "learning_rate": 3.534105873057677e-05,
      "loss": 1.8477,
      "step": 11139
    },
    {
      "epoch": 0.29338951804055835,
      "grad_norm": 1.8187834024429321,
      "learning_rate": 3.5339741901501186e-05,
      "loss": 2.1298,
      "step": 11140
    },
    {
      "epoch": 0.29341585462207004,
      "grad_norm": 1.865159273147583,
      "learning_rate": 3.53384250724256e-05,
      "loss": 1.5547,
      "step": 11141
    },
    {
      "epoch": 0.2934421912035818,
      "grad_norm": 2.249774932861328,
      "learning_rate": 3.533710824335002e-05,
      "loss": 1.7296,
      "step": 11142
    },
    {
      "epoch": 0.29346852778509347,
      "grad_norm": 2.927061080932617,
      "learning_rate": 3.5335791414274426e-05,
      "loss": 1.8314,
      "step": 11143
    },
    {
      "epoch": 0.2934948643666052,
      "grad_norm": 1.8893762826919556,
      "learning_rate": 3.533447458519884e-05,
      "loss": 1.5575,
      "step": 11144
    },
    {
      "epoch": 0.29352120094811696,
      "grad_norm": 6.156041622161865,
      "learning_rate": 3.533315775612326e-05,
      "loss": 1.0783,
      "step": 11145
    },
    {
      "epoch": 0.29354753752962864,
      "grad_norm": 2.1454193592071533,
      "learning_rate": 3.533184092704767e-05,
      "loss": 1.9329,
      "step": 11146
    },
    {
      "epoch": 0.2935738741111404,
      "grad_norm": 1.8228569030761719,
      "learning_rate": 3.533052409797208e-05,
      "loss": 1.8157,
      "step": 11147
    },
    {
      "epoch": 0.2936002106926521,
      "grad_norm": 2.6528375148773193,
      "learning_rate": 3.53292072688965e-05,
      "loss": 1.8013,
      "step": 11148
    },
    {
      "epoch": 0.2936265472741638,
      "grad_norm": 2.364283800125122,
      "learning_rate": 3.532789043982091e-05,
      "loss": 1.6277,
      "step": 11149
    },
    {
      "epoch": 0.2936528838556755,
      "grad_norm": 2.68813157081604,
      "learning_rate": 3.532657361074533e-05,
      "loss": 1.704,
      "step": 11150
    },
    {
      "epoch": 0.29367922043718725,
      "grad_norm": 3.6759097576141357,
      "learning_rate": 3.5325256781669744e-05,
      "loss": 1.4004,
      "step": 11151
    },
    {
      "epoch": 0.293705557018699,
      "grad_norm": 1.7254774570465088,
      "learning_rate": 3.532393995259415e-05,
      "loss": 2.0627,
      "step": 11152
    },
    {
      "epoch": 0.2937318936002107,
      "grad_norm": 5.245553493499756,
      "learning_rate": 3.532262312351857e-05,
      "loss": 1.4824,
      "step": 11153
    },
    {
      "epoch": 0.2937582301817224,
      "grad_norm": 4.06008243560791,
      "learning_rate": 3.5321306294442984e-05,
      "loss": 1.9692,
      "step": 11154
    },
    {
      "epoch": 0.2937845667632341,
      "grad_norm": 2.40374755859375,
      "learning_rate": 3.53199894653674e-05,
      "loss": 2.3672,
      "step": 11155
    },
    {
      "epoch": 0.29381090334474586,
      "grad_norm": 3.136652708053589,
      "learning_rate": 3.531867263629181e-05,
      "loss": 1.3923,
      "step": 11156
    },
    {
      "epoch": 0.2938372399262576,
      "grad_norm": 2.8856682777404785,
      "learning_rate": 3.5317355807216224e-05,
      "loss": 1.6868,
      "step": 11157
    },
    {
      "epoch": 0.2938635765077693,
      "grad_norm": 3.481116771697998,
      "learning_rate": 3.531603897814064e-05,
      "loss": 2.2314,
      "step": 11158
    },
    {
      "epoch": 0.29388991308928103,
      "grad_norm": 2.4352495670318604,
      "learning_rate": 3.5314722149065055e-05,
      "loss": 1.2976,
      "step": 11159
    },
    {
      "epoch": 0.2939162496707927,
      "grad_norm": 1.915453314781189,
      "learning_rate": 3.531340531998947e-05,
      "loss": 1.7207,
      "step": 11160
    },
    {
      "epoch": 0.29394258625230446,
      "grad_norm": 3.60813045501709,
      "learning_rate": 3.531208849091388e-05,
      "loss": 2.2216,
      "step": 11161
    },
    {
      "epoch": 0.29396892283381615,
      "grad_norm": 3.326063871383667,
      "learning_rate": 3.5310771661838295e-05,
      "loss": 1.7641,
      "step": 11162
    },
    {
      "epoch": 0.2939952594153279,
      "grad_norm": 1.6828333139419556,
      "learning_rate": 3.530945483276271e-05,
      "loss": 2.0016,
      "step": 11163
    },
    {
      "epoch": 0.29402159599683964,
      "grad_norm": 2.586540937423706,
      "learning_rate": 3.5308138003687127e-05,
      "loss": 1.7729,
      "step": 11164
    },
    {
      "epoch": 0.2940479325783513,
      "grad_norm": 2.4852166175842285,
      "learning_rate": 3.5306821174611535e-05,
      "loss": 1.3307,
      "step": 11165
    },
    {
      "epoch": 0.29407426915986307,
      "grad_norm": 5.29385232925415,
      "learning_rate": 3.530550434553595e-05,
      "loss": 1.0516,
      "step": 11166
    },
    {
      "epoch": 0.29410060574137475,
      "grad_norm": 2.363166570663452,
      "learning_rate": 3.530418751646036e-05,
      "loss": 1.0703,
      "step": 11167
    },
    {
      "epoch": 0.2941269423228865,
      "grad_norm": 1.8245933055877686,
      "learning_rate": 3.530287068738478e-05,
      "loss": 2.0348,
      "step": 11168
    },
    {
      "epoch": 0.2941532789043982,
      "grad_norm": 2.159219264984131,
      "learning_rate": 3.530155385830919e-05,
      "loss": 1.4938,
      "step": 11169
    },
    {
      "epoch": 0.29417961548590993,
      "grad_norm": 2.0778892040252686,
      "learning_rate": 3.5300237029233607e-05,
      "loss": 0.7858,
      "step": 11170
    },
    {
      "epoch": 0.29420595206742167,
      "grad_norm": 1.5472893714904785,
      "learning_rate": 3.529892020015802e-05,
      "loss": 2.0901,
      "step": 11171
    },
    {
      "epoch": 0.29423228864893336,
      "grad_norm": 3.383136510848999,
      "learning_rate": 3.529760337108243e-05,
      "loss": 1.3084,
      "step": 11172
    },
    {
      "epoch": 0.2942586252304451,
      "grad_norm": 3.1431174278259277,
      "learning_rate": 3.529628654200685e-05,
      "loss": 1.3236,
      "step": 11173
    },
    {
      "epoch": 0.2942849618119568,
      "grad_norm": 1.9847277402877808,
      "learning_rate": 3.529496971293126e-05,
      "loss": 2.0408,
      "step": 11174
    },
    {
      "epoch": 0.29431129839346853,
      "grad_norm": 1.7892119884490967,
      "learning_rate": 3.529365288385568e-05,
      "loss": 2.1783,
      "step": 11175
    },
    {
      "epoch": 0.2943376349749802,
      "grad_norm": 2.5990209579467773,
      "learning_rate": 3.5292336054780087e-05,
      "loss": 2.0511,
      "step": 11176
    },
    {
      "epoch": 0.29436397155649197,
      "grad_norm": 1.5941860675811768,
      "learning_rate": 3.52910192257045e-05,
      "loss": 2.1411,
      "step": 11177
    },
    {
      "epoch": 0.2943903081380037,
      "grad_norm": 2.296689987182617,
      "learning_rate": 3.528970239662892e-05,
      "loss": 0.7347,
      "step": 11178
    },
    {
      "epoch": 0.2944166447195154,
      "grad_norm": 1.9193092584609985,
      "learning_rate": 3.5288385567553333e-05,
      "loss": 1.777,
      "step": 11179
    },
    {
      "epoch": 0.29444298130102714,
      "grad_norm": 1.9996116161346436,
      "learning_rate": 3.528706873847775e-05,
      "loss": 1.6853,
      "step": 11180
    },
    {
      "epoch": 0.29446931788253883,
      "grad_norm": 3.902935266494751,
      "learning_rate": 3.528575190940216e-05,
      "loss": 1.706,
      "step": 11181
    },
    {
      "epoch": 0.29449565446405057,
      "grad_norm": 2.9622862339019775,
      "learning_rate": 3.528443508032658e-05,
      "loss": 1.5944,
      "step": 11182
    },
    {
      "epoch": 0.29452199104556226,
      "grad_norm": 2.138038396835327,
      "learning_rate": 3.528311825125099e-05,
      "loss": 1.861,
      "step": 11183
    },
    {
      "epoch": 0.294548327627074,
      "grad_norm": 1.7757973670959473,
      "learning_rate": 3.5281801422175405e-05,
      "loss": 2.1649,
      "step": 11184
    },
    {
      "epoch": 0.29457466420858575,
      "grad_norm": 2.3491737842559814,
      "learning_rate": 3.5280484593099813e-05,
      "loss": 1.5502,
      "step": 11185
    },
    {
      "epoch": 0.29460100079009743,
      "grad_norm": 2.606473207473755,
      "learning_rate": 3.527916776402423e-05,
      "loss": 2.1967,
      "step": 11186
    },
    {
      "epoch": 0.2946273373716092,
      "grad_norm": 1.9916232824325562,
      "learning_rate": 3.5277850934948645e-05,
      "loss": 1.6064,
      "step": 11187
    },
    {
      "epoch": 0.29465367395312086,
      "grad_norm": 4.016951084136963,
      "learning_rate": 3.527653410587306e-05,
      "loss": 1.3203,
      "step": 11188
    },
    {
      "epoch": 0.2946800105346326,
      "grad_norm": 2.234654426574707,
      "learning_rate": 3.5275217276797476e-05,
      "loss": 1.8732,
      "step": 11189
    },
    {
      "epoch": 0.29470634711614435,
      "grad_norm": 3.873143196105957,
      "learning_rate": 3.5273900447721885e-05,
      "loss": 1.3686,
      "step": 11190
    },
    {
      "epoch": 0.29473268369765604,
      "grad_norm": 1.75237238407135,
      "learning_rate": 3.52725836186463e-05,
      "loss": 1.7468,
      "step": 11191
    },
    {
      "epoch": 0.2947590202791678,
      "grad_norm": 1.793145775794983,
      "learning_rate": 3.5271266789570716e-05,
      "loss": 1.8607,
      "step": 11192
    },
    {
      "epoch": 0.29478535686067947,
      "grad_norm": 4.081301689147949,
      "learning_rate": 3.526994996049513e-05,
      "loss": 1.1937,
      "step": 11193
    },
    {
      "epoch": 0.2948116934421912,
      "grad_norm": 2.46317982673645,
      "learning_rate": 3.526863313141954e-05,
      "loss": 1.3486,
      "step": 11194
    },
    {
      "epoch": 0.2948380300237029,
      "grad_norm": 2.060487747192383,
      "learning_rate": 3.5267316302343956e-05,
      "loss": 1.6392,
      "step": 11195
    },
    {
      "epoch": 0.29486436660521465,
      "grad_norm": 2.104280471801758,
      "learning_rate": 3.526599947326837e-05,
      "loss": 2.046,
      "step": 11196
    },
    {
      "epoch": 0.2948907031867264,
      "grad_norm": 1.8553714752197266,
      "learning_rate": 3.526468264419279e-05,
      "loss": 1.4616,
      "step": 11197
    },
    {
      "epoch": 0.2949170397682381,
      "grad_norm": 2.332540988922119,
      "learning_rate": 3.52633658151172e-05,
      "loss": 2.0226,
      "step": 11198
    },
    {
      "epoch": 0.2949433763497498,
      "grad_norm": 2.6931471824645996,
      "learning_rate": 3.526204898604161e-05,
      "loss": 1.6319,
      "step": 11199
    },
    {
      "epoch": 0.2949697129312615,
      "grad_norm": 1.7424328327178955,
      "learning_rate": 3.526073215696603e-05,
      "loss": 1.5522,
      "step": 11200
    },
    {
      "epoch": 0.29499604951277325,
      "grad_norm": 3.1145408153533936,
      "learning_rate": 3.525941532789044e-05,
      "loss": 2.0246,
      "step": 11201
    },
    {
      "epoch": 0.29502238609428494,
      "grad_norm": 1.8936015367507935,
      "learning_rate": 3.525809849881486e-05,
      "loss": 2.3262,
      "step": 11202
    },
    {
      "epoch": 0.2950487226757967,
      "grad_norm": 1.7812623977661133,
      "learning_rate": 3.525678166973927e-05,
      "loss": 1.5768,
      "step": 11203
    },
    {
      "epoch": 0.2950750592573084,
      "grad_norm": 2.1300299167633057,
      "learning_rate": 3.525546484066368e-05,
      "loss": 2.3458,
      "step": 11204
    },
    {
      "epoch": 0.2951013958388201,
      "grad_norm": 1.9849748611450195,
      "learning_rate": 3.52541480115881e-05,
      "loss": 2.2947,
      "step": 11205
    },
    {
      "epoch": 0.29512773242033186,
      "grad_norm": 3.6287360191345215,
      "learning_rate": 3.5252831182512514e-05,
      "loss": 1.2135,
      "step": 11206
    },
    {
      "epoch": 0.29515406900184354,
      "grad_norm": 2.8776333332061768,
      "learning_rate": 3.525151435343693e-05,
      "loss": 2.1561,
      "step": 11207
    },
    {
      "epoch": 0.2951804055833553,
      "grad_norm": 3.6309022903442383,
      "learning_rate": 3.525019752436134e-05,
      "loss": 1.3714,
      "step": 11208
    },
    {
      "epoch": 0.295206742164867,
      "grad_norm": 3.14775013923645,
      "learning_rate": 3.5248880695285754e-05,
      "loss": 1.3941,
      "step": 11209
    },
    {
      "epoch": 0.2952330787463787,
      "grad_norm": 3.3579788208007812,
      "learning_rate": 3.524756386621017e-05,
      "loss": 1.4462,
      "step": 11210
    },
    {
      "epoch": 0.29525941532789046,
      "grad_norm": 2.4440853595733643,
      "learning_rate": 3.5246247037134585e-05,
      "loss": 1.6734,
      "step": 11211
    },
    {
      "epoch": 0.29528575190940215,
      "grad_norm": 2.0921802520751953,
      "learning_rate": 3.5244930208058994e-05,
      "loss": 1.2276,
      "step": 11212
    },
    {
      "epoch": 0.2953120884909139,
      "grad_norm": 1.534582495689392,
      "learning_rate": 3.524361337898341e-05,
      "loss": 1.446,
      "step": 11213
    },
    {
      "epoch": 0.2953384250724256,
      "grad_norm": 4.487147808074951,
      "learning_rate": 3.524229654990782e-05,
      "loss": 0.9311,
      "step": 11214
    },
    {
      "epoch": 0.2953647616539373,
      "grad_norm": 1.7657101154327393,
      "learning_rate": 3.524097972083224e-05,
      "loss": 2.3719,
      "step": 11215
    },
    {
      "epoch": 0.295391098235449,
      "grad_norm": 2.163151741027832,
      "learning_rate": 3.5239662891756656e-05,
      "loss": 1.6507,
      "step": 11216
    },
    {
      "epoch": 0.29541743481696076,
      "grad_norm": 1.5511177778244019,
      "learning_rate": 3.5238346062681065e-05,
      "loss": 1.7734,
      "step": 11217
    },
    {
      "epoch": 0.2954437713984725,
      "grad_norm": 1.5488502979278564,
      "learning_rate": 3.523702923360548e-05,
      "loss": 0.7871,
      "step": 11218
    },
    {
      "epoch": 0.2954701079799842,
      "grad_norm": 8.044669151306152,
      "learning_rate": 3.523571240452989e-05,
      "loss": 0.8054,
      "step": 11219
    },
    {
      "epoch": 0.29549644456149593,
      "grad_norm": 1.952410101890564,
      "learning_rate": 3.523439557545431e-05,
      "loss": 1.7429,
      "step": 11220
    },
    {
      "epoch": 0.2955227811430076,
      "grad_norm": 3.4446659088134766,
      "learning_rate": 3.523307874637872e-05,
      "loss": 0.7818,
      "step": 11221
    },
    {
      "epoch": 0.29554911772451936,
      "grad_norm": 2.1014182567596436,
      "learning_rate": 3.5231761917303136e-05,
      "loss": 1.0132,
      "step": 11222
    },
    {
      "epoch": 0.2955754543060311,
      "grad_norm": 2.375545024871826,
      "learning_rate": 3.5230445088227545e-05,
      "loss": 2.0499,
      "step": 11223
    },
    {
      "epoch": 0.2956017908875428,
      "grad_norm": 1.7391877174377441,
      "learning_rate": 3.522912825915196e-05,
      "loss": 1.7864,
      "step": 11224
    },
    {
      "epoch": 0.29562812746905454,
      "grad_norm": 2.1989901065826416,
      "learning_rate": 3.5227811430076376e-05,
      "loss": 1.6099,
      "step": 11225
    },
    {
      "epoch": 0.2956544640505662,
      "grad_norm": 3.3491978645324707,
      "learning_rate": 3.522649460100079e-05,
      "loss": 2.3066,
      "step": 11226
    },
    {
      "epoch": 0.29568080063207797,
      "grad_norm": 2.848043441772461,
      "learning_rate": 3.522517777192521e-05,
      "loss": 0.337,
      "step": 11227
    },
    {
      "epoch": 0.29570713721358965,
      "grad_norm": 1.8994667530059814,
      "learning_rate": 3.5223860942849616e-05,
      "loss": 1.3694,
      "step": 11228
    },
    {
      "epoch": 0.2957334737951014,
      "grad_norm": 1.9781851768493652,
      "learning_rate": 3.522254411377404e-05,
      "loss": 1.9247,
      "step": 11229
    },
    {
      "epoch": 0.29575981037661314,
      "grad_norm": 2.438587188720703,
      "learning_rate": 3.522122728469845e-05,
      "loss": 1.4936,
      "step": 11230
    },
    {
      "epoch": 0.29578614695812483,
      "grad_norm": 4.01224422454834,
      "learning_rate": 3.521991045562286e-05,
      "loss": 0.8521,
      "step": 11231
    },
    {
      "epoch": 0.2958124835396366,
      "grad_norm": 3.047328233718872,
      "learning_rate": 3.521859362654727e-05,
      "loss": 2.0677,
      "step": 11232
    },
    {
      "epoch": 0.29583882012114826,
      "grad_norm": 2.571894645690918,
      "learning_rate": 3.521727679747169e-05,
      "loss": 0.4427,
      "step": 11233
    },
    {
      "epoch": 0.29586515670266,
      "grad_norm": 5.571075916290283,
      "learning_rate": 3.52159599683961e-05,
      "loss": 1.6175,
      "step": 11234
    },
    {
      "epoch": 0.2958914932841717,
      "grad_norm": 2.1518359184265137,
      "learning_rate": 3.521464313932052e-05,
      "loss": 1.9484,
      "step": 11235
    },
    {
      "epoch": 0.29591782986568343,
      "grad_norm": 1.8952051401138306,
      "learning_rate": 3.5213326310244934e-05,
      "loss": 1.6916,
      "step": 11236
    },
    {
      "epoch": 0.2959441664471952,
      "grad_norm": 1.8660380840301514,
      "learning_rate": 3.521200948116934e-05,
      "loss": 1.939,
      "step": 11237
    },
    {
      "epoch": 0.29597050302870687,
      "grad_norm": 2.507702350616455,
      "learning_rate": 3.521069265209376e-05,
      "loss": 0.5641,
      "step": 11238
    },
    {
      "epoch": 0.2959968396102186,
      "grad_norm": 3.8901422023773193,
      "learning_rate": 3.5209375823018174e-05,
      "loss": 2.1301,
      "step": 11239
    },
    {
      "epoch": 0.2960231761917303,
      "grad_norm": 1.9510704278945923,
      "learning_rate": 3.520805899394259e-05,
      "loss": 1.4203,
      "step": 11240
    },
    {
      "epoch": 0.29604951277324204,
      "grad_norm": 3.083888292312622,
      "learning_rate": 3.5206742164867e-05,
      "loss": 0.9464,
      "step": 11241
    },
    {
      "epoch": 0.29607584935475373,
      "grad_norm": 1.8966132402420044,
      "learning_rate": 3.5205425335791414e-05,
      "loss": 1.3417,
      "step": 11242
    },
    {
      "epoch": 0.29610218593626547,
      "grad_norm": 2.2648446559906006,
      "learning_rate": 3.520410850671583e-05,
      "loss": 1.8079,
      "step": 11243
    },
    {
      "epoch": 0.2961285225177772,
      "grad_norm": 1.9381766319274902,
      "learning_rate": 3.5202791677640246e-05,
      "loss": 2.2784,
      "step": 11244
    },
    {
      "epoch": 0.2961548590992889,
      "grad_norm": 2.3600974082946777,
      "learning_rate": 3.520147484856466e-05,
      "loss": 1.5315,
      "step": 11245
    },
    {
      "epoch": 0.29618119568080065,
      "grad_norm": 4.122308731079102,
      "learning_rate": 3.520015801948907e-05,
      "loss": 0.6987,
      "step": 11246
    },
    {
      "epoch": 0.29620753226231233,
      "grad_norm": 2.9262075424194336,
      "learning_rate": 3.5198841190413486e-05,
      "loss": 2.2166,
      "step": 11247
    },
    {
      "epoch": 0.2962338688438241,
      "grad_norm": 1.4721442461013794,
      "learning_rate": 3.51975243613379e-05,
      "loss": 0.9561,
      "step": 11248
    },
    {
      "epoch": 0.29626020542533577,
      "grad_norm": 1.8601629734039307,
      "learning_rate": 3.519620753226232e-05,
      "loss": 1.3197,
      "step": 11249
    },
    {
      "epoch": 0.2962865420068475,
      "grad_norm": 2.136586904525757,
      "learning_rate": 3.5194890703186726e-05,
      "loss": 2.053,
      "step": 11250
    },
    {
      "epoch": 0.29631287858835925,
      "grad_norm": 1.799536108970642,
      "learning_rate": 3.519357387411114e-05,
      "loss": 1.5529,
      "step": 11251
    },
    {
      "epoch": 0.29633921516987094,
      "grad_norm": 1.7599679231643677,
      "learning_rate": 3.519225704503556e-05,
      "loss": 1.5343,
      "step": 11252
    },
    {
      "epoch": 0.2963655517513827,
      "grad_norm": 1.6139698028564453,
      "learning_rate": 3.519094021595997e-05,
      "loss": 1.8092,
      "step": 11253
    },
    {
      "epoch": 0.29639188833289437,
      "grad_norm": 2.0157313346862793,
      "learning_rate": 3.518962338688439e-05,
      "loss": 1.721,
      "step": 11254
    },
    {
      "epoch": 0.2964182249144061,
      "grad_norm": 2.6079649925231934,
      "learning_rate": 3.51883065578088e-05,
      "loss": 2.1226,
      "step": 11255
    },
    {
      "epoch": 0.2964445614959178,
      "grad_norm": 1.6632548570632935,
      "learning_rate": 3.518698972873321e-05,
      "loss": 1.8434,
      "step": 11256
    },
    {
      "epoch": 0.29647089807742955,
      "grad_norm": 5.257874011993408,
      "learning_rate": 3.518567289965762e-05,
      "loss": 1.849,
      "step": 11257
    },
    {
      "epoch": 0.2964972346589413,
      "grad_norm": 3.5983164310455322,
      "learning_rate": 3.5184356070582044e-05,
      "loss": 1.6561,
      "step": 11258
    },
    {
      "epoch": 0.296523571240453,
      "grad_norm": 2.2889139652252197,
      "learning_rate": 3.518303924150645e-05,
      "loss": 2.0746,
      "step": 11259
    },
    {
      "epoch": 0.2965499078219647,
      "grad_norm": 1.7331256866455078,
      "learning_rate": 3.518172241243087e-05,
      "loss": 1.9603,
      "step": 11260
    },
    {
      "epoch": 0.2965762444034764,
      "grad_norm": 3.4799764156341553,
      "learning_rate": 3.5180405583355284e-05,
      "loss": 1.679,
      "step": 11261
    },
    {
      "epoch": 0.29660258098498815,
      "grad_norm": 1.8756428956985474,
      "learning_rate": 3.51790887542797e-05,
      "loss": 1.7471,
      "step": 11262
    },
    {
      "epoch": 0.2966289175664999,
      "grad_norm": 1.4509050846099854,
      "learning_rate": 3.5177771925204115e-05,
      "loss": 1.6872,
      "step": 11263
    },
    {
      "epoch": 0.2966552541480116,
      "grad_norm": 2.148665428161621,
      "learning_rate": 3.5176455096128524e-05,
      "loss": 1.1716,
      "step": 11264
    },
    {
      "epoch": 0.2966815907295233,
      "grad_norm": 2.5871989727020264,
      "learning_rate": 3.517513826705294e-05,
      "loss": 0.4398,
      "step": 11265
    },
    {
      "epoch": 0.296707927311035,
      "grad_norm": 3.0054564476013184,
      "learning_rate": 3.517382143797735e-05,
      "loss": 2.1549,
      "step": 11266
    },
    {
      "epoch": 0.29673426389254676,
      "grad_norm": 2.008307933807373,
      "learning_rate": 3.517250460890177e-05,
      "loss": 1.6396,
      "step": 11267
    },
    {
      "epoch": 0.29676060047405844,
      "grad_norm": 1.8543498516082764,
      "learning_rate": 3.517118777982618e-05,
      "loss": 1.9873,
      "step": 11268
    },
    {
      "epoch": 0.2967869370555702,
      "grad_norm": 1.8072952032089233,
      "learning_rate": 3.5169870950750595e-05,
      "loss": 1.7686,
      "step": 11269
    },
    {
      "epoch": 0.29681327363708193,
      "grad_norm": 3.243941307067871,
      "learning_rate": 3.5168554121675004e-05,
      "loss": 2.1289,
      "step": 11270
    },
    {
      "epoch": 0.2968396102185936,
      "grad_norm": 7.306673049926758,
      "learning_rate": 3.516723729259942e-05,
      "loss": 2.2979,
      "step": 11271
    },
    {
      "epoch": 0.29686594680010536,
      "grad_norm": 3.147047996520996,
      "learning_rate": 3.5165920463523835e-05,
      "loss": 1.3544,
      "step": 11272
    },
    {
      "epoch": 0.29689228338161705,
      "grad_norm": 1.3698394298553467,
      "learning_rate": 3.516460363444825e-05,
      "loss": 1.8454,
      "step": 11273
    },
    {
      "epoch": 0.2969186199631288,
      "grad_norm": 2.0408740043640137,
      "learning_rate": 3.5163286805372666e-05,
      "loss": 1.8977,
      "step": 11274
    },
    {
      "epoch": 0.2969449565446405,
      "grad_norm": 2.604503631591797,
      "learning_rate": 3.5161969976297075e-05,
      "loss": 1.2487,
      "step": 11275
    },
    {
      "epoch": 0.2969712931261522,
      "grad_norm": 1.969042181968689,
      "learning_rate": 3.51606531472215e-05,
      "loss": 1.9016,
      "step": 11276
    },
    {
      "epoch": 0.29699762970766397,
      "grad_norm": 1.6118701696395874,
      "learning_rate": 3.5159336318145906e-05,
      "loss": 1.5509,
      "step": 11277
    },
    {
      "epoch": 0.29702396628917566,
      "grad_norm": 4.527522563934326,
      "learning_rate": 3.515801948907032e-05,
      "loss": 1.781,
      "step": 11278
    },
    {
      "epoch": 0.2970503028706874,
      "grad_norm": 1.9608347415924072,
      "learning_rate": 3.515670265999473e-05,
      "loss": 1.8426,
      "step": 11279
    },
    {
      "epoch": 0.2970766394521991,
      "grad_norm": 4.401960849761963,
      "learning_rate": 3.5155385830919146e-05,
      "loss": 2.2969,
      "step": 11280
    },
    {
      "epoch": 0.29710297603371083,
      "grad_norm": 2.0636041164398193,
      "learning_rate": 3.515406900184356e-05,
      "loss": 2.0198,
      "step": 11281
    },
    {
      "epoch": 0.2971293126152225,
      "grad_norm": 2.2625300884246826,
      "learning_rate": 3.515275217276798e-05,
      "loss": 2.2881,
      "step": 11282
    },
    {
      "epoch": 0.29715564919673426,
      "grad_norm": 1.8739567995071411,
      "learning_rate": 3.515143534369239e-05,
      "loss": 1.84,
      "step": 11283
    },
    {
      "epoch": 0.297181985778246,
      "grad_norm": 1.908544659614563,
      "learning_rate": 3.51501185146168e-05,
      "loss": 1.9209,
      "step": 11284
    },
    {
      "epoch": 0.2972083223597577,
      "grad_norm": 3.1405060291290283,
      "learning_rate": 3.514880168554122e-05,
      "loss": 0.639,
      "step": 11285
    },
    {
      "epoch": 0.29723465894126944,
      "grad_norm": 2.2767162322998047,
      "learning_rate": 3.514748485646563e-05,
      "loss": 1.676,
      "step": 11286
    },
    {
      "epoch": 0.2972609955227811,
      "grad_norm": 1.9440308809280396,
      "learning_rate": 3.514616802739005e-05,
      "loss": 1.5822,
      "step": 11287
    },
    {
      "epoch": 0.29728733210429287,
      "grad_norm": 1.9416247606277466,
      "learning_rate": 3.514485119831446e-05,
      "loss": 0.8407,
      "step": 11288
    },
    {
      "epoch": 0.29731366868580456,
      "grad_norm": 3.2492847442626953,
      "learning_rate": 3.514353436923887e-05,
      "loss": 2.1815,
      "step": 11289
    },
    {
      "epoch": 0.2973400052673163,
      "grad_norm": 4.336751937866211,
      "learning_rate": 3.514221754016329e-05,
      "loss": 1.885,
      "step": 11290
    },
    {
      "epoch": 0.29736634184882804,
      "grad_norm": 6.096614360809326,
      "learning_rate": 3.5140900711087704e-05,
      "loss": 1.5798,
      "step": 11291
    },
    {
      "epoch": 0.29739267843033973,
      "grad_norm": 2.607179641723633,
      "learning_rate": 3.513958388201212e-05,
      "loss": 2.0135,
      "step": 11292
    },
    {
      "epoch": 0.2974190150118515,
      "grad_norm": 3.0693907737731934,
      "learning_rate": 3.513826705293653e-05,
      "loss": 2.3704,
      "step": 11293
    },
    {
      "epoch": 0.29744535159336316,
      "grad_norm": 1.8100945949554443,
      "learning_rate": 3.5136950223860944e-05,
      "loss": 1.7142,
      "step": 11294
    },
    {
      "epoch": 0.2974716881748749,
      "grad_norm": 1.720001459121704,
      "learning_rate": 3.513563339478536e-05,
      "loss": 1.6544,
      "step": 11295
    },
    {
      "epoch": 0.29749802475638665,
      "grad_norm": 2.7185206413269043,
      "learning_rate": 3.5134316565709775e-05,
      "loss": 1.7472,
      "step": 11296
    },
    {
      "epoch": 0.29752436133789834,
      "grad_norm": 2.925856590270996,
      "learning_rate": 3.5132999736634184e-05,
      "loss": 0.6453,
      "step": 11297
    },
    {
      "epoch": 0.2975506979194101,
      "grad_norm": 3.8374879360198975,
      "learning_rate": 3.51316829075586e-05,
      "loss": 1.6675,
      "step": 11298
    },
    {
      "epoch": 0.29757703450092177,
      "grad_norm": 1.8744838237762451,
      "learning_rate": 3.5130366078483015e-05,
      "loss": 1.0709,
      "step": 11299
    },
    {
      "epoch": 0.2976033710824335,
      "grad_norm": 3.5693914890289307,
      "learning_rate": 3.512904924940743e-05,
      "loss": 1.7182,
      "step": 11300
    },
    {
      "epoch": 0.2976297076639452,
      "grad_norm": 2.1889984607696533,
      "learning_rate": 3.5127732420331847e-05,
      "loss": 0.6038,
      "step": 11301
    },
    {
      "epoch": 0.29765604424545694,
      "grad_norm": 1.7235954999923706,
      "learning_rate": 3.5126415591256255e-05,
      "loss": 1.4082,
      "step": 11302
    },
    {
      "epoch": 0.2976823808269687,
      "grad_norm": 2.602802038192749,
      "learning_rate": 3.512509876218067e-05,
      "loss": 1.8461,
      "step": 11303
    },
    {
      "epoch": 0.29770871740848037,
      "grad_norm": 3.2257328033447266,
      "learning_rate": 3.512378193310508e-05,
      "loss": 0.7786,
      "step": 11304
    },
    {
      "epoch": 0.2977350539899921,
      "grad_norm": 6.53745174407959,
      "learning_rate": 3.51224651040295e-05,
      "loss": 1.0738,
      "step": 11305
    },
    {
      "epoch": 0.2977613905715038,
      "grad_norm": 2.5591909885406494,
      "learning_rate": 3.512114827495391e-05,
      "loss": 1.5284,
      "step": 11306
    },
    {
      "epoch": 0.29778772715301555,
      "grad_norm": 3.120382070541382,
      "learning_rate": 3.511983144587833e-05,
      "loss": 0.5894,
      "step": 11307
    },
    {
      "epoch": 0.29781406373452723,
      "grad_norm": 1.8904626369476318,
      "learning_rate": 3.511851461680274e-05,
      "loss": 1.4343,
      "step": 11308
    },
    {
      "epoch": 0.297840400316039,
      "grad_norm": 2.502664566040039,
      "learning_rate": 3.511719778772716e-05,
      "loss": 0.897,
      "step": 11309
    },
    {
      "epoch": 0.2978667368975507,
      "grad_norm": 3.501429557800293,
      "learning_rate": 3.5115880958651573e-05,
      "loss": 0.6699,
      "step": 11310
    },
    {
      "epoch": 0.2978930734790624,
      "grad_norm": 2.1692235469818115,
      "learning_rate": 3.511456412957598e-05,
      "loss": 1.7393,
      "step": 11311
    },
    {
      "epoch": 0.29791941006057415,
      "grad_norm": 2.408330202102661,
      "learning_rate": 3.51132473005004e-05,
      "loss": 1.9903,
      "step": 11312
    },
    {
      "epoch": 0.29794574664208584,
      "grad_norm": 1.7968454360961914,
      "learning_rate": 3.511193047142481e-05,
      "loss": 1.9403,
      "step": 11313
    },
    {
      "epoch": 0.2979720832235976,
      "grad_norm": 1.9358646869659424,
      "learning_rate": 3.511061364234923e-05,
      "loss": 2.1813,
      "step": 11314
    },
    {
      "epoch": 0.29799841980510927,
      "grad_norm": 2.53141450881958,
      "learning_rate": 3.510929681327364e-05,
      "loss": 1.9958,
      "step": 11315
    },
    {
      "epoch": 0.298024756386621,
      "grad_norm": 1.9014556407928467,
      "learning_rate": 3.5107979984198053e-05,
      "loss": 1.1513,
      "step": 11316
    },
    {
      "epoch": 0.29805109296813276,
      "grad_norm": 2.8176589012145996,
      "learning_rate": 3.510666315512246e-05,
      "loss": 1.8123,
      "step": 11317
    },
    {
      "epoch": 0.29807742954964445,
      "grad_norm": 8.192499160766602,
      "learning_rate": 3.510534632604688e-05,
      "loss": 2.3802,
      "step": 11318
    },
    {
      "epoch": 0.2981037661311562,
      "grad_norm": 1.729529857635498,
      "learning_rate": 3.5104029496971293e-05,
      "loss": 1.5345,
      "step": 11319
    },
    {
      "epoch": 0.2981301027126679,
      "grad_norm": 2.0037455558776855,
      "learning_rate": 3.510271266789571e-05,
      "loss": 1.1211,
      "step": 11320
    },
    {
      "epoch": 0.2981564392941796,
      "grad_norm": 2.028219223022461,
      "learning_rate": 3.5101395838820125e-05,
      "loss": 1.751,
      "step": 11321
    },
    {
      "epoch": 0.2981827758756913,
      "grad_norm": 7.575572967529297,
      "learning_rate": 3.5100079009744533e-05,
      "loss": 1.3426,
      "step": 11322
    },
    {
      "epoch": 0.29820911245720305,
      "grad_norm": 3.681105852127075,
      "learning_rate": 3.509876218066895e-05,
      "loss": 1.0907,
      "step": 11323
    },
    {
      "epoch": 0.2982354490387148,
      "grad_norm": 2.6946728229522705,
      "learning_rate": 3.5097445351593365e-05,
      "loss": 1.1082,
      "step": 11324
    },
    {
      "epoch": 0.2982617856202265,
      "grad_norm": 3.061182737350464,
      "learning_rate": 3.509612852251778e-05,
      "loss": 0.9191,
      "step": 11325
    },
    {
      "epoch": 0.2982881222017382,
      "grad_norm": 2.9638516902923584,
      "learning_rate": 3.509481169344219e-05,
      "loss": 1.4096,
      "step": 11326
    },
    {
      "epoch": 0.2983144587832499,
      "grad_norm": 1.582367181777954,
      "learning_rate": 3.5093494864366605e-05,
      "loss": 2.2101,
      "step": 11327
    },
    {
      "epoch": 0.29834079536476166,
      "grad_norm": 3.923387289047241,
      "learning_rate": 3.509217803529102e-05,
      "loss": 1.1766,
      "step": 11328
    },
    {
      "epoch": 0.2983671319462734,
      "grad_norm": 1.9604004621505737,
      "learning_rate": 3.5090861206215436e-05,
      "loss": 1.9074,
      "step": 11329
    },
    {
      "epoch": 0.2983934685277851,
      "grad_norm": 2.111985206604004,
      "learning_rate": 3.508954437713985e-05,
      "loss": 2.2323,
      "step": 11330
    },
    {
      "epoch": 0.29841980510929683,
      "grad_norm": 1.5968613624572754,
      "learning_rate": 3.508822754806426e-05,
      "loss": 1.7975,
      "step": 11331
    },
    {
      "epoch": 0.2984461416908085,
      "grad_norm": 3.5455071926116943,
      "learning_rate": 3.5086910718988676e-05,
      "loss": 1.7912,
      "step": 11332
    },
    {
      "epoch": 0.29847247827232026,
      "grad_norm": 2.2526659965515137,
      "learning_rate": 3.508559388991309e-05,
      "loss": 1.9117,
      "step": 11333
    },
    {
      "epoch": 0.29849881485383195,
      "grad_norm": 1.5859909057617188,
      "learning_rate": 3.508427706083751e-05,
      "loss": 1.2353,
      "step": 11334
    },
    {
      "epoch": 0.2985251514353437,
      "grad_norm": 1.991771936416626,
      "learning_rate": 3.5082960231761916e-05,
      "loss": 0.5767,
      "step": 11335
    },
    {
      "epoch": 0.29855148801685544,
      "grad_norm": 1.495442271232605,
      "learning_rate": 3.508164340268633e-05,
      "loss": 2.0683,
      "step": 11336
    },
    {
      "epoch": 0.2985778245983671,
      "grad_norm": 2.049715995788574,
      "learning_rate": 3.508032657361075e-05,
      "loss": 1.6001,
      "step": 11337
    },
    {
      "epoch": 0.29860416117987887,
      "grad_norm": 1.6948968172073364,
      "learning_rate": 3.507900974453516e-05,
      "loss": 0.6071,
      "step": 11338
    },
    {
      "epoch": 0.29863049776139056,
      "grad_norm": 2.6026017665863037,
      "learning_rate": 3.507769291545958e-05,
      "loss": 2.0654,
      "step": 11339
    },
    {
      "epoch": 0.2986568343429023,
      "grad_norm": 2.0262014865875244,
      "learning_rate": 3.507637608638399e-05,
      "loss": 2.1178,
      "step": 11340
    },
    {
      "epoch": 0.298683170924414,
      "grad_norm": 1.9327361583709717,
      "learning_rate": 3.50750592573084e-05,
      "loss": 2.4644,
      "step": 11341
    },
    {
      "epoch": 0.29870950750592573,
      "grad_norm": 4.835659980773926,
      "learning_rate": 3.507374242823282e-05,
      "loss": 0.7732,
      "step": 11342
    },
    {
      "epoch": 0.2987358440874375,
      "grad_norm": 2.3807318210601807,
      "learning_rate": 3.5072425599157234e-05,
      "loss": 1.6574,
      "step": 11343
    },
    {
      "epoch": 0.29876218066894916,
      "grad_norm": 1.720497727394104,
      "learning_rate": 3.507110877008164e-05,
      "loss": 2.2618,
      "step": 11344
    },
    {
      "epoch": 0.2987885172504609,
      "grad_norm": 2.229938268661499,
      "learning_rate": 3.506979194100606e-05,
      "loss": 1.7065,
      "step": 11345
    },
    {
      "epoch": 0.2988148538319726,
      "grad_norm": 1.668868899345398,
      "learning_rate": 3.5068475111930474e-05,
      "loss": 2.1841,
      "step": 11346
    },
    {
      "epoch": 0.29884119041348434,
      "grad_norm": 1.6168829202651978,
      "learning_rate": 3.506715828285489e-05,
      "loss": 1.8651,
      "step": 11347
    },
    {
      "epoch": 0.298867526994996,
      "grad_norm": 2.2111268043518066,
      "learning_rate": 3.5065841453779305e-05,
      "loss": 2.1177,
      "step": 11348
    },
    {
      "epoch": 0.29889386357650777,
      "grad_norm": 3.8437047004699707,
      "learning_rate": 3.5064524624703714e-05,
      "loss": 1.2729,
      "step": 11349
    },
    {
      "epoch": 0.2989202001580195,
      "grad_norm": 2.1236634254455566,
      "learning_rate": 3.506320779562813e-05,
      "loss": 1.7316,
      "step": 11350
    },
    {
      "epoch": 0.2989465367395312,
      "grad_norm": 3.524595022201538,
      "learning_rate": 3.506189096655254e-05,
      "loss": 1.5795,
      "step": 11351
    },
    {
      "epoch": 0.29897287332104294,
      "grad_norm": 3.1641995906829834,
      "learning_rate": 3.506057413747696e-05,
      "loss": 1.5867,
      "step": 11352
    },
    {
      "epoch": 0.29899920990255463,
      "grad_norm": 3.083679676055908,
      "learning_rate": 3.505925730840137e-05,
      "loss": 1.0551,
      "step": 11353
    },
    {
      "epoch": 0.2990255464840664,
      "grad_norm": 2.1841988563537598,
      "learning_rate": 3.5057940479325785e-05,
      "loss": 1.7922,
      "step": 11354
    },
    {
      "epoch": 0.29905188306557806,
      "grad_norm": 2.1207447052001953,
      "learning_rate": 3.50566236502502e-05,
      "loss": 1.8548,
      "step": 11355
    },
    {
      "epoch": 0.2990782196470898,
      "grad_norm": 2.3573505878448486,
      "learning_rate": 3.505530682117461e-05,
      "loss": 1.5995,
      "step": 11356
    },
    {
      "epoch": 0.29910455622860155,
      "grad_norm": 3.5008678436279297,
      "learning_rate": 3.505398999209903e-05,
      "loss": 1.7812,
      "step": 11357
    },
    {
      "epoch": 0.29913089281011324,
      "grad_norm": 3.3185739517211914,
      "learning_rate": 3.505267316302344e-05,
      "loss": 1.3111,
      "step": 11358
    },
    {
      "epoch": 0.299157229391625,
      "grad_norm": 2.8683111667633057,
      "learning_rate": 3.5051356333947856e-05,
      "loss": 0.6612,
      "step": 11359
    },
    {
      "epoch": 0.29918356597313667,
      "grad_norm": 2.66521954536438,
      "learning_rate": 3.5050039504872265e-05,
      "loss": 1.528,
      "step": 11360
    },
    {
      "epoch": 0.2992099025546484,
      "grad_norm": 2.139737367630005,
      "learning_rate": 3.504872267579669e-05,
      "loss": 2.2746,
      "step": 11361
    },
    {
      "epoch": 0.29923623913616015,
      "grad_norm": 1.8905680179595947,
      "learning_rate": 3.5047405846721096e-05,
      "loss": 1.9795,
      "step": 11362
    },
    {
      "epoch": 0.29926257571767184,
      "grad_norm": 1.9279958009719849,
      "learning_rate": 3.504608901764551e-05,
      "loss": 1.773,
      "step": 11363
    },
    {
      "epoch": 0.2992889122991836,
      "grad_norm": 1.6923837661743164,
      "learning_rate": 3.504477218856993e-05,
      "loss": 1.5279,
      "step": 11364
    },
    {
      "epoch": 0.2993152488806953,
      "grad_norm": 3.3340327739715576,
      "learning_rate": 3.5043455359494336e-05,
      "loss": 1.377,
      "step": 11365
    },
    {
      "epoch": 0.299341585462207,
      "grad_norm": 1.6432462930679321,
      "learning_rate": 3.504213853041876e-05,
      "loss": 1.3982,
      "step": 11366
    },
    {
      "epoch": 0.2993679220437187,
      "grad_norm": 3.1528308391571045,
      "learning_rate": 3.504082170134317e-05,
      "loss": 2.1196,
      "step": 11367
    },
    {
      "epoch": 0.29939425862523045,
      "grad_norm": 3.964564085006714,
      "learning_rate": 3.503950487226758e-05,
      "loss": 1.2747,
      "step": 11368
    },
    {
      "epoch": 0.2994205952067422,
      "grad_norm": 1.9572641849517822,
      "learning_rate": 3.503818804319199e-05,
      "loss": 1.5417,
      "step": 11369
    },
    {
      "epoch": 0.2994469317882539,
      "grad_norm": 3.664025068283081,
      "learning_rate": 3.503687121411641e-05,
      "loss": 1.3096,
      "step": 11370
    },
    {
      "epoch": 0.2994732683697656,
      "grad_norm": 2.319117546081543,
      "learning_rate": 3.503555438504082e-05,
      "loss": 1.617,
      "step": 11371
    },
    {
      "epoch": 0.2994996049512773,
      "grad_norm": 2.221092462539673,
      "learning_rate": 3.503423755596524e-05,
      "loss": 1.9934,
      "step": 11372
    },
    {
      "epoch": 0.29952594153278905,
      "grad_norm": 2.0143775939941406,
      "learning_rate": 3.503292072688965e-05,
      "loss": 1.7052,
      "step": 11373
    },
    {
      "epoch": 0.29955227811430074,
      "grad_norm": 2.1504523754119873,
      "learning_rate": 3.503160389781406e-05,
      "loss": 1.4524,
      "step": 11374
    },
    {
      "epoch": 0.2995786146958125,
      "grad_norm": 2.070394992828369,
      "learning_rate": 3.503028706873848e-05,
      "loss": 1.5879,
      "step": 11375
    },
    {
      "epoch": 0.2996049512773242,
      "grad_norm": 2.0374763011932373,
      "learning_rate": 3.5028970239662894e-05,
      "loss": 1.7961,
      "step": 11376
    },
    {
      "epoch": 0.2996312878588359,
      "grad_norm": 4.845815181732178,
      "learning_rate": 3.502765341058731e-05,
      "loss": 1.8492,
      "step": 11377
    },
    {
      "epoch": 0.29965762444034766,
      "grad_norm": 2.439401865005493,
      "learning_rate": 3.502633658151172e-05,
      "loss": 1.8908,
      "step": 11378
    },
    {
      "epoch": 0.29968396102185935,
      "grad_norm": 2.191620111465454,
      "learning_rate": 3.5025019752436134e-05,
      "loss": 1.7195,
      "step": 11379
    },
    {
      "epoch": 0.2997102976033711,
      "grad_norm": 1.9873735904693604,
      "learning_rate": 3.502370292336055e-05,
      "loss": 1.6176,
      "step": 11380
    },
    {
      "epoch": 0.2997366341848828,
      "grad_norm": 2.8773157596588135,
      "learning_rate": 3.5022386094284966e-05,
      "loss": 2.1566,
      "step": 11381
    },
    {
      "epoch": 0.2997629707663945,
      "grad_norm": 3.963364601135254,
      "learning_rate": 3.5021069265209374e-05,
      "loss": 0.8538,
      "step": 11382
    },
    {
      "epoch": 0.29978930734790626,
      "grad_norm": 3.418531656265259,
      "learning_rate": 3.501975243613379e-05,
      "loss": 0.6466,
      "step": 11383
    },
    {
      "epoch": 0.29981564392941795,
      "grad_norm": 1.8965864181518555,
      "learning_rate": 3.5018435607058206e-05,
      "loss": 1.0929,
      "step": 11384
    },
    {
      "epoch": 0.2998419805109297,
      "grad_norm": 3.509310722351074,
      "learning_rate": 3.501711877798262e-05,
      "loss": 1.364,
      "step": 11385
    },
    {
      "epoch": 0.2998683170924414,
      "grad_norm": 1.5227187871932983,
      "learning_rate": 3.501580194890704e-05,
      "loss": 1.4799,
      "step": 11386
    },
    {
      "epoch": 0.2998946536739531,
      "grad_norm": 1.956417202949524,
      "learning_rate": 3.5014485119831446e-05,
      "loss": 1.8624,
      "step": 11387
    },
    {
      "epoch": 0.2999209902554648,
      "grad_norm": 1.7138806581497192,
      "learning_rate": 3.501316829075586e-05,
      "loss": 2.2537,
      "step": 11388
    },
    {
      "epoch": 0.29994732683697656,
      "grad_norm": 1.9416273832321167,
      "learning_rate": 3.501185146168027e-05,
      "loss": 0.4365,
      "step": 11389
    },
    {
      "epoch": 0.2999736634184883,
      "grad_norm": 3.6158647537231445,
      "learning_rate": 3.501053463260469e-05,
      "loss": 1.436,
      "step": 11390
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3787429332733154,
      "learning_rate": 3.50092178035291e-05,
      "loss": 1.4393,
      "step": 11391
    },
    {
      "epoch": 0.30002633658151173,
      "grad_norm": 1.7575633525848389,
      "learning_rate": 3.500790097445352e-05,
      "loss": 2.5512,
      "step": 11392
    },
    {
      "epoch": 0.3000526731630234,
      "grad_norm": 2.107300281524658,
      "learning_rate": 3.500658414537793e-05,
      "loss": 2.4838,
      "step": 11393
    },
    {
      "epoch": 0.30007900974453516,
      "grad_norm": 1.9221919775009155,
      "learning_rate": 3.500526731630235e-05,
      "loss": 2.3892,
      "step": 11394
    },
    {
      "epoch": 0.3001053463260469,
      "grad_norm": 2.0634987354278564,
      "learning_rate": 3.5003950487226764e-05,
      "loss": 2.0733,
      "step": 11395
    },
    {
      "epoch": 0.3001316829075586,
      "grad_norm": 2.9131155014038086,
      "learning_rate": 3.500263365815117e-05,
      "loss": 1.36,
      "step": 11396
    },
    {
      "epoch": 0.30015801948907034,
      "grad_norm": 1.9818575382232666,
      "learning_rate": 3.500131682907559e-05,
      "loss": 1.6081,
      "step": 11397
    },
    {
      "epoch": 0.300184356070582,
      "grad_norm": 1.4716944694519043,
      "learning_rate": 3.5e-05,
      "loss": 1.8672,
      "step": 11398
    },
    {
      "epoch": 0.30021069265209377,
      "grad_norm": 2.087327241897583,
      "learning_rate": 3.499868317092442e-05,
      "loss": 2.099,
      "step": 11399
    },
    {
      "epoch": 0.30023702923360546,
      "grad_norm": 1.72110116481781,
      "learning_rate": 3.499736634184883e-05,
      "loss": 0.6496,
      "step": 11400
    },
    {
      "epoch": 0.3002633658151172,
      "grad_norm": 1.8490465879440308,
      "learning_rate": 3.4996049512773244e-05,
      "loss": 1.8095,
      "step": 11401
    },
    {
      "epoch": 0.30028970239662894,
      "grad_norm": 1.335369348526001,
      "learning_rate": 3.499473268369766e-05,
      "loss": 1.6903,
      "step": 11402
    },
    {
      "epoch": 0.30031603897814063,
      "grad_norm": 1.898796558380127,
      "learning_rate": 3.499341585462207e-05,
      "loss": 1.4037,
      "step": 11403
    },
    {
      "epoch": 0.3003423755596524,
      "grad_norm": 2.4721274375915527,
      "learning_rate": 3.499209902554649e-05,
      "loss": 0.3906,
      "step": 11404
    },
    {
      "epoch": 0.30036871214116406,
      "grad_norm": 4.560434341430664,
      "learning_rate": 3.49907821964709e-05,
      "loss": 1.9955,
      "step": 11405
    },
    {
      "epoch": 0.3003950487226758,
      "grad_norm": 1.8851643800735474,
      "learning_rate": 3.4989465367395315e-05,
      "loss": 1.9015,
      "step": 11406
    },
    {
      "epoch": 0.3004213853041875,
      "grad_norm": 3.1878209114074707,
      "learning_rate": 3.4988148538319724e-05,
      "loss": 1.7438,
      "step": 11407
    },
    {
      "epoch": 0.30044772188569924,
      "grad_norm": 2.026601791381836,
      "learning_rate": 3.4986831709244146e-05,
      "loss": 1.744,
      "step": 11408
    },
    {
      "epoch": 0.300474058467211,
      "grad_norm": 2.912034273147583,
      "learning_rate": 3.4985514880168555e-05,
      "loss": 1.7488,
      "step": 11409
    },
    {
      "epoch": 0.30050039504872267,
      "grad_norm": 2.195408821105957,
      "learning_rate": 3.498419805109297e-05,
      "loss": 1.8744,
      "step": 11410
    },
    {
      "epoch": 0.3005267316302344,
      "grad_norm": 1.757171630859375,
      "learning_rate": 3.4982881222017386e-05,
      "loss": 1.4281,
      "step": 11411
    },
    {
      "epoch": 0.3005530682117461,
      "grad_norm": 2.126519203186035,
      "learning_rate": 3.4981564392941795e-05,
      "loss": 2.2833,
      "step": 11412
    },
    {
      "epoch": 0.30057940479325784,
      "grad_norm": 3.181601047515869,
      "learning_rate": 3.498024756386622e-05,
      "loss": 1.4416,
      "step": 11413
    },
    {
      "epoch": 0.30060574137476953,
      "grad_norm": 3.573237419128418,
      "learning_rate": 3.4978930734790626e-05,
      "loss": 1.0448,
      "step": 11414
    },
    {
      "epoch": 0.3006320779562813,
      "grad_norm": 1.7598329782485962,
      "learning_rate": 3.497761390571504e-05,
      "loss": 1.4158,
      "step": 11415
    },
    {
      "epoch": 0.300658414537793,
      "grad_norm": 3.2319602966308594,
      "learning_rate": 3.497629707663945e-05,
      "loss": 0.6546,
      "step": 11416
    },
    {
      "epoch": 0.3006847511193047,
      "grad_norm": 2.052393674850464,
      "learning_rate": 3.4974980247563866e-05,
      "loss": 0.2748,
      "step": 11417
    },
    {
      "epoch": 0.30071108770081645,
      "grad_norm": 1.855228304862976,
      "learning_rate": 3.497366341848828e-05,
      "loss": 0.4704,
      "step": 11418
    },
    {
      "epoch": 0.30073742428232814,
      "grad_norm": 1.6941291093826294,
      "learning_rate": 3.49723465894127e-05,
      "loss": 1.2878,
      "step": 11419
    },
    {
      "epoch": 0.3007637608638399,
      "grad_norm": 2.1435251235961914,
      "learning_rate": 3.4971029760337106e-05,
      "loss": 1.9211,
      "step": 11420
    },
    {
      "epoch": 0.30079009744535157,
      "grad_norm": 4.7399210929870605,
      "learning_rate": 3.496971293126152e-05,
      "loss": 1.5094,
      "step": 11421
    },
    {
      "epoch": 0.3008164340268633,
      "grad_norm": 1.5705182552337646,
      "learning_rate": 3.496839610218594e-05,
      "loss": 2.2813,
      "step": 11422
    },
    {
      "epoch": 0.30084277060837505,
      "grad_norm": 3.903655529022217,
      "learning_rate": 3.496707927311035e-05,
      "loss": 2.0785,
      "step": 11423
    },
    {
      "epoch": 0.30086910718988674,
      "grad_norm": 2.515242576599121,
      "learning_rate": 3.496576244403477e-05,
      "loss": 1.3157,
      "step": 11424
    },
    {
      "epoch": 0.3008954437713985,
      "grad_norm": 3.384575843811035,
      "learning_rate": 3.496444561495918e-05,
      "loss": 1.4308,
      "step": 11425
    },
    {
      "epoch": 0.3009217803529102,
      "grad_norm": 1.9864667654037476,
      "learning_rate": 3.496312878588359e-05,
      "loss": 1.7673,
      "step": 11426
    },
    {
      "epoch": 0.3009481169344219,
      "grad_norm": 2.24808669090271,
      "learning_rate": 3.496181195680801e-05,
      "loss": 1.6765,
      "step": 11427
    },
    {
      "epoch": 0.3009744535159336,
      "grad_norm": 2.157534122467041,
      "learning_rate": 3.4960495127732424e-05,
      "loss": 1.6835,
      "step": 11428
    },
    {
      "epoch": 0.30100079009744535,
      "grad_norm": 1.9540029764175415,
      "learning_rate": 3.495917829865683e-05,
      "loss": 1.9228,
      "step": 11429
    },
    {
      "epoch": 0.3010271266789571,
      "grad_norm": 1.7175195217132568,
      "learning_rate": 3.495786146958125e-05,
      "loss": 1.1776,
      "step": 11430
    },
    {
      "epoch": 0.3010534632604688,
      "grad_norm": 1.51752769947052,
      "learning_rate": 3.4956544640505664e-05,
      "loss": 0.4166,
      "step": 11431
    },
    {
      "epoch": 0.3010797998419805,
      "grad_norm": 2.9118707180023193,
      "learning_rate": 3.495522781143008e-05,
      "loss": 1.5472,
      "step": 11432
    },
    {
      "epoch": 0.3011061364234922,
      "grad_norm": 1.7199506759643555,
      "learning_rate": 3.4953910982354495e-05,
      "loss": 1.9804,
      "step": 11433
    },
    {
      "epoch": 0.30113247300500395,
      "grad_norm": 1.5731666088104248,
      "learning_rate": 3.4952594153278904e-05,
      "loss": 1.6395,
      "step": 11434
    },
    {
      "epoch": 0.3011588095865157,
      "grad_norm": 3.308396816253662,
      "learning_rate": 3.495127732420332e-05,
      "loss": 1.7894,
      "step": 11435
    },
    {
      "epoch": 0.3011851461680274,
      "grad_norm": 1.9530525207519531,
      "learning_rate": 3.494996049512773e-05,
      "loss": 1.9274,
      "step": 11436
    },
    {
      "epoch": 0.30121148274953913,
      "grad_norm": 4.370994567871094,
      "learning_rate": 3.494864366605215e-05,
      "loss": 0.9156,
      "step": 11437
    },
    {
      "epoch": 0.3012378193310508,
      "grad_norm": 3.1108012199401855,
      "learning_rate": 3.494732683697656e-05,
      "loss": 2.279,
      "step": 11438
    },
    {
      "epoch": 0.30126415591256256,
      "grad_norm": 2.022583246231079,
      "learning_rate": 3.4946010007900975e-05,
      "loss": 2.1374,
      "step": 11439
    },
    {
      "epoch": 0.30129049249407425,
      "grad_norm": 2.1375811100006104,
      "learning_rate": 3.494469317882539e-05,
      "loss": 1.7166,
      "step": 11440
    },
    {
      "epoch": 0.301316829075586,
      "grad_norm": 4.401586055755615,
      "learning_rate": 3.494337634974981e-05,
      "loss": 1.4899,
      "step": 11441
    },
    {
      "epoch": 0.30134316565709773,
      "grad_norm": 4.158323764801025,
      "learning_rate": 3.494205952067422e-05,
      "loss": 1.1617,
      "step": 11442
    },
    {
      "epoch": 0.3013695022386094,
      "grad_norm": 4.1364850997924805,
      "learning_rate": 3.494074269159863e-05,
      "loss": 1.7601,
      "step": 11443
    },
    {
      "epoch": 0.30139583882012116,
      "grad_norm": 2.1257357597351074,
      "learning_rate": 3.493942586252305e-05,
      "loss": 2.1702,
      "step": 11444
    },
    {
      "epoch": 0.30142217540163285,
      "grad_norm": 3.1927878856658936,
      "learning_rate": 3.4938109033447455e-05,
      "loss": 1.2641,
      "step": 11445
    },
    {
      "epoch": 0.3014485119831446,
      "grad_norm": 1.7009096145629883,
      "learning_rate": 3.493679220437188e-05,
      "loss": 1.7328,
      "step": 11446
    },
    {
      "epoch": 0.3014748485646563,
      "grad_norm": 6.142141342163086,
      "learning_rate": 3.493547537529629e-05,
      "loss": 1.3355,
      "step": 11447
    },
    {
      "epoch": 0.301501185146168,
      "grad_norm": 2.5633113384246826,
      "learning_rate": 3.49341585462207e-05,
      "loss": 1.9544,
      "step": 11448
    },
    {
      "epoch": 0.30152752172767977,
      "grad_norm": 1.8000307083129883,
      "learning_rate": 3.493284171714512e-05,
      "loss": 1.628,
      "step": 11449
    },
    {
      "epoch": 0.30155385830919146,
      "grad_norm": 3.10040020942688,
      "learning_rate": 3.493152488806953e-05,
      "loss": 0.6859,
      "step": 11450
    },
    {
      "epoch": 0.3015801948907032,
      "grad_norm": 2.016465425491333,
      "learning_rate": 3.493020805899395e-05,
      "loss": 0.3359,
      "step": 11451
    },
    {
      "epoch": 0.3016065314722149,
      "grad_norm": 2.095041036605835,
      "learning_rate": 3.492889122991836e-05,
      "loss": 1.3384,
      "step": 11452
    },
    {
      "epoch": 0.30163286805372663,
      "grad_norm": 2.615236520767212,
      "learning_rate": 3.4927574400842773e-05,
      "loss": 2.3088,
      "step": 11453
    },
    {
      "epoch": 0.3016592046352383,
      "grad_norm": 4.078410625457764,
      "learning_rate": 3.492625757176718e-05,
      "loss": 1.2441,
      "step": 11454
    },
    {
      "epoch": 0.30168554121675006,
      "grad_norm": 1.917397141456604,
      "learning_rate": 3.49249407426916e-05,
      "loss": 2.6897,
      "step": 11455
    },
    {
      "epoch": 0.3017118777982618,
      "grad_norm": 2.176980972290039,
      "learning_rate": 3.4923623913616014e-05,
      "loss": 1.821,
      "step": 11456
    },
    {
      "epoch": 0.3017382143797735,
      "grad_norm": 2.618560314178467,
      "learning_rate": 3.492230708454043e-05,
      "loss": 0.8655,
      "step": 11457
    },
    {
      "epoch": 0.30176455096128524,
      "grad_norm": 1.825404167175293,
      "learning_rate": 3.4920990255464845e-05,
      "loss": 2.4378,
      "step": 11458
    },
    {
      "epoch": 0.3017908875427969,
      "grad_norm": 2.453624725341797,
      "learning_rate": 3.4919673426389254e-05,
      "loss": 2.4098,
      "step": 11459
    },
    {
      "epoch": 0.30181722412430867,
      "grad_norm": 3.267610549926758,
      "learning_rate": 3.4918356597313676e-05,
      "loss": 1.7119,
      "step": 11460
    },
    {
      "epoch": 0.30184356070582036,
      "grad_norm": 1.6376019716262817,
      "learning_rate": 3.4917039768238085e-05,
      "loss": 0.8118,
      "step": 11461
    },
    {
      "epoch": 0.3018698972873321,
      "grad_norm": 2.9108848571777344,
      "learning_rate": 3.49157229391625e-05,
      "loss": 2.3266,
      "step": 11462
    },
    {
      "epoch": 0.30189623386884384,
      "grad_norm": 2.1124589443206787,
      "learning_rate": 3.491440611008691e-05,
      "loss": 1.2451,
      "step": 11463
    },
    {
      "epoch": 0.30192257045035553,
      "grad_norm": 1.5624676942825317,
      "learning_rate": 3.4913089281011325e-05,
      "loss": 1.7329,
      "step": 11464
    },
    {
      "epoch": 0.3019489070318673,
      "grad_norm": 2.3640170097351074,
      "learning_rate": 3.491177245193574e-05,
      "loss": 2.6786,
      "step": 11465
    },
    {
      "epoch": 0.30197524361337896,
      "grad_norm": 3.3089945316314697,
      "learning_rate": 3.4910455622860156e-05,
      "loss": 0.4188,
      "step": 11466
    },
    {
      "epoch": 0.3020015801948907,
      "grad_norm": 1.6686619520187378,
      "learning_rate": 3.490913879378457e-05,
      "loss": 1.5781,
      "step": 11467
    },
    {
      "epoch": 0.30202791677640245,
      "grad_norm": 1.8771644830703735,
      "learning_rate": 3.490782196470898e-05,
      "loss": 1.8105,
      "step": 11468
    },
    {
      "epoch": 0.30205425335791414,
      "grad_norm": 2.1818089485168457,
      "learning_rate": 3.4906505135633396e-05,
      "loss": 1.2544,
      "step": 11469
    },
    {
      "epoch": 0.3020805899394259,
      "grad_norm": 2.2085797786712646,
      "learning_rate": 3.490518830655781e-05,
      "loss": 2.5588,
      "step": 11470
    },
    {
      "epoch": 0.30210692652093757,
      "grad_norm": 3.314614772796631,
      "learning_rate": 3.490387147748223e-05,
      "loss": 2.4428,
      "step": 11471
    },
    {
      "epoch": 0.3021332631024493,
      "grad_norm": 1.92331063747406,
      "learning_rate": 3.4902554648406636e-05,
      "loss": 0.7612,
      "step": 11472
    },
    {
      "epoch": 0.302159599683961,
      "grad_norm": 2.9464011192321777,
      "learning_rate": 3.490123781933105e-05,
      "loss": 1.8368,
      "step": 11473
    },
    {
      "epoch": 0.30218593626547274,
      "grad_norm": 2.567378520965576,
      "learning_rate": 3.489992099025547e-05,
      "loss": 0.7124,
      "step": 11474
    },
    {
      "epoch": 0.3022122728469845,
      "grad_norm": 2.5734474658966064,
      "learning_rate": 3.489860416117988e-05,
      "loss": 0.7,
      "step": 11475
    },
    {
      "epoch": 0.3022386094284962,
      "grad_norm": 2.0721397399902344,
      "learning_rate": 3.489728733210429e-05,
      "loss": 1.7553,
      "step": 11476
    },
    {
      "epoch": 0.3022649460100079,
      "grad_norm": 3.214118003845215,
      "learning_rate": 3.489597050302871e-05,
      "loss": 1.8072,
      "step": 11477
    },
    {
      "epoch": 0.3022912825915196,
      "grad_norm": 1.5494662523269653,
      "learning_rate": 3.489465367395312e-05,
      "loss": 1.7095,
      "step": 11478
    },
    {
      "epoch": 0.30231761917303135,
      "grad_norm": 3.223907709121704,
      "learning_rate": 3.489333684487754e-05,
      "loss": 1.3393,
      "step": 11479
    },
    {
      "epoch": 0.30234395575454304,
      "grad_norm": 5.31834602355957,
      "learning_rate": 3.4892020015801954e-05,
      "loss": 1.1932,
      "step": 11480
    },
    {
      "epoch": 0.3023702923360548,
      "grad_norm": 1.904130220413208,
      "learning_rate": 3.489070318672636e-05,
      "loss": 1.8768,
      "step": 11481
    },
    {
      "epoch": 0.3023966289175665,
      "grad_norm": 2.699662208557129,
      "learning_rate": 3.488938635765078e-05,
      "loss": 1.8886,
      "step": 11482
    },
    {
      "epoch": 0.3024229654990782,
      "grad_norm": 1.994221806526184,
      "learning_rate": 3.488806952857519e-05,
      "loss": 2.1696,
      "step": 11483
    },
    {
      "epoch": 0.30244930208058995,
      "grad_norm": 2.02368426322937,
      "learning_rate": 3.488675269949961e-05,
      "loss": 1.2955,
      "step": 11484
    },
    {
      "epoch": 0.30247563866210164,
      "grad_norm": 3.2579987049102783,
      "learning_rate": 3.488543587042402e-05,
      "loss": 0.6286,
      "step": 11485
    },
    {
      "epoch": 0.3025019752436134,
      "grad_norm": 2.1361560821533203,
      "learning_rate": 3.4884119041348434e-05,
      "loss": 1.3654,
      "step": 11486
    },
    {
      "epoch": 0.3025283118251251,
      "grad_norm": 1.8205980062484741,
      "learning_rate": 3.488280221227285e-05,
      "loss": 1.7312,
      "step": 11487
    },
    {
      "epoch": 0.3025546484066368,
      "grad_norm": 1.6786627769470215,
      "learning_rate": 3.488148538319726e-05,
      "loss": 1.5492,
      "step": 11488
    },
    {
      "epoch": 0.30258098498814856,
      "grad_norm": 1.9154386520385742,
      "learning_rate": 3.488016855412168e-05,
      "loss": 1.5429,
      "step": 11489
    },
    {
      "epoch": 0.30260732156966025,
      "grad_norm": 1.7850189208984375,
      "learning_rate": 3.487885172504609e-05,
      "loss": 1.4754,
      "step": 11490
    },
    {
      "epoch": 0.302633658151172,
      "grad_norm": 2.176323652267456,
      "learning_rate": 3.4877534895970505e-05,
      "loss": 2.0591,
      "step": 11491
    },
    {
      "epoch": 0.3026599947326837,
      "grad_norm": 2.3130524158477783,
      "learning_rate": 3.4876218066894914e-05,
      "loss": 2.0783,
      "step": 11492
    },
    {
      "epoch": 0.3026863313141954,
      "grad_norm": 3.135699510574341,
      "learning_rate": 3.4874901237819336e-05,
      "loss": 0.4751,
      "step": 11493
    },
    {
      "epoch": 0.3027126678957071,
      "grad_norm": 3.852264881134033,
      "learning_rate": 3.4873584408743745e-05,
      "loss": 1.597,
      "step": 11494
    },
    {
      "epoch": 0.30273900447721885,
      "grad_norm": 2.207695245742798,
      "learning_rate": 3.487226757966816e-05,
      "loss": 1.9708,
      "step": 11495
    },
    {
      "epoch": 0.3027653410587306,
      "grad_norm": 3.5587363243103027,
      "learning_rate": 3.4870950750592576e-05,
      "loss": 1.8565,
      "step": 11496
    },
    {
      "epoch": 0.3027916776402423,
      "grad_norm": 4.040370464324951,
      "learning_rate": 3.4869633921516985e-05,
      "loss": 1.093,
      "step": 11497
    },
    {
      "epoch": 0.30281801422175403,
      "grad_norm": 1.7107130289077759,
      "learning_rate": 3.486831709244141e-05,
      "loss": 1.1887,
      "step": 11498
    },
    {
      "epoch": 0.3028443508032657,
      "grad_norm": 1.9254714250564575,
      "learning_rate": 3.4867000263365816e-05,
      "loss": 0.4478,
      "step": 11499
    },
    {
      "epoch": 0.30287068738477746,
      "grad_norm": 3.078185558319092,
      "learning_rate": 3.486568343429023e-05,
      "loss": 1.2184,
      "step": 11500
    },
    {
      "epoch": 0.3028970239662892,
      "grad_norm": 2.670498847961426,
      "learning_rate": 3.486436660521464e-05,
      "loss": 1.3868,
      "step": 11501
    },
    {
      "epoch": 0.3029233605478009,
      "grad_norm": 2.282266139984131,
      "learning_rate": 3.4863049776139056e-05,
      "loss": 1.7738,
      "step": 11502
    },
    {
      "epoch": 0.30294969712931263,
      "grad_norm": 2.90488600730896,
      "learning_rate": 3.486173294706347e-05,
      "loss": 1.4311,
      "step": 11503
    },
    {
      "epoch": 0.3029760337108243,
      "grad_norm": 1.6077455282211304,
      "learning_rate": 3.486041611798789e-05,
      "loss": 0.583,
      "step": 11504
    },
    {
      "epoch": 0.30300237029233607,
      "grad_norm": 5.220090866088867,
      "learning_rate": 3.48590992889123e-05,
      "loss": 1.1909,
      "step": 11505
    },
    {
      "epoch": 0.30302870687384775,
      "grad_norm": 1.885409951210022,
      "learning_rate": 3.485778245983671e-05,
      "loss": 2.0389,
      "step": 11506
    },
    {
      "epoch": 0.3030550434553595,
      "grad_norm": 2.345433473587036,
      "learning_rate": 3.4856465630761134e-05,
      "loss": 1.8258,
      "step": 11507
    },
    {
      "epoch": 0.30308138003687124,
      "grad_norm": 1.9141958951950073,
      "learning_rate": 3.485514880168554e-05,
      "loss": 1.9422,
      "step": 11508
    },
    {
      "epoch": 0.3031077166183829,
      "grad_norm": 2.8689074516296387,
      "learning_rate": 3.485383197260996e-05,
      "loss": 1.1462,
      "step": 11509
    },
    {
      "epoch": 0.30313405319989467,
      "grad_norm": 3.328822612762451,
      "learning_rate": 3.485251514353437e-05,
      "loss": 2.5067,
      "step": 11510
    },
    {
      "epoch": 0.30316038978140636,
      "grad_norm": 2.0886874198913574,
      "learning_rate": 3.485119831445878e-05,
      "loss": 1.7875,
      "step": 11511
    },
    {
      "epoch": 0.3031867263629181,
      "grad_norm": 2.780932664871216,
      "learning_rate": 3.48498814853832e-05,
      "loss": 2.3524,
      "step": 11512
    },
    {
      "epoch": 0.3032130629444298,
      "grad_norm": 5.195681095123291,
      "learning_rate": 3.4848564656307614e-05,
      "loss": 1.1057,
      "step": 11513
    },
    {
      "epoch": 0.30323939952594153,
      "grad_norm": 2.395005464553833,
      "learning_rate": 3.484724782723203e-05,
      "loss": 1.874,
      "step": 11514
    },
    {
      "epoch": 0.3032657361074533,
      "grad_norm": 2.122614622116089,
      "learning_rate": 3.484593099815644e-05,
      "loss": 1.6429,
      "step": 11515
    },
    {
      "epoch": 0.30329207268896496,
      "grad_norm": 1.6674736738204956,
      "learning_rate": 3.4844614169080855e-05,
      "loss": 1.9288,
      "step": 11516
    },
    {
      "epoch": 0.3033184092704767,
      "grad_norm": 2.5792770385742188,
      "learning_rate": 3.484329734000527e-05,
      "loss": 1.6325,
      "step": 11517
    },
    {
      "epoch": 0.3033447458519884,
      "grad_norm": 3.0751078128814697,
      "learning_rate": 3.4841980510929686e-05,
      "loss": 0.8267,
      "step": 11518
    },
    {
      "epoch": 0.30337108243350014,
      "grad_norm": 2.639277219772339,
      "learning_rate": 3.4840663681854095e-05,
      "loss": 1.8914,
      "step": 11519
    },
    {
      "epoch": 0.3033974190150118,
      "grad_norm": 1.6220495700836182,
      "learning_rate": 3.483934685277851e-05,
      "loss": 2.0251,
      "step": 11520
    },
    {
      "epoch": 0.30342375559652357,
      "grad_norm": 1.768712043762207,
      "learning_rate": 3.4838030023702926e-05,
      "loss": 1.5173,
      "step": 11521
    },
    {
      "epoch": 0.3034500921780353,
      "grad_norm": 2.242807626724243,
      "learning_rate": 3.483671319462734e-05,
      "loss": 1.3552,
      "step": 11522
    },
    {
      "epoch": 0.303476428759547,
      "grad_norm": 3.2393174171447754,
      "learning_rate": 3.483539636555175e-05,
      "loss": 1.5405,
      "step": 11523
    },
    {
      "epoch": 0.30350276534105874,
      "grad_norm": 2.5753369331359863,
      "learning_rate": 3.4834079536476166e-05,
      "loss": 1.8293,
      "step": 11524
    },
    {
      "epoch": 0.30352910192257043,
      "grad_norm": 1.4630845785140991,
      "learning_rate": 3.483276270740058e-05,
      "loss": 2.2216,
      "step": 11525
    },
    {
      "epoch": 0.3035554385040822,
      "grad_norm": 2.8406453132629395,
      "learning_rate": 3.4831445878325e-05,
      "loss": 1.1713,
      "step": 11526
    },
    {
      "epoch": 0.30358177508559386,
      "grad_norm": 2.0083794593811035,
      "learning_rate": 3.483012904924941e-05,
      "loss": 1.4846,
      "step": 11527
    },
    {
      "epoch": 0.3036081116671056,
      "grad_norm": 2.4162118434906006,
      "learning_rate": 3.482881222017382e-05,
      "loss": 2.3402,
      "step": 11528
    },
    {
      "epoch": 0.30363444824861735,
      "grad_norm": 1.5709762573242188,
      "learning_rate": 3.482749539109824e-05,
      "loss": 2.131,
      "step": 11529
    },
    {
      "epoch": 0.30366078483012904,
      "grad_norm": 4.539622783660889,
      "learning_rate": 3.4826178562022646e-05,
      "loss": 1.4055,
      "step": 11530
    },
    {
      "epoch": 0.3036871214116408,
      "grad_norm": 3.038649559020996,
      "learning_rate": 3.482486173294707e-05,
      "loss": 1.8247,
      "step": 11531
    },
    {
      "epoch": 0.30371345799315247,
      "grad_norm": 3.3281667232513428,
      "learning_rate": 3.482354490387148e-05,
      "loss": 1.1562,
      "step": 11532
    },
    {
      "epoch": 0.3037397945746642,
      "grad_norm": 3.070957660675049,
      "learning_rate": 3.482222807479589e-05,
      "loss": 1.8494,
      "step": 11533
    },
    {
      "epoch": 0.30376613115617596,
      "grad_norm": 1.99111807346344,
      "learning_rate": 3.482091124572031e-05,
      "loss": 1.7051,
      "step": 11534
    },
    {
      "epoch": 0.30379246773768764,
      "grad_norm": 2.608184337615967,
      "learning_rate": 3.481959441664472e-05,
      "loss": 1.2814,
      "step": 11535
    },
    {
      "epoch": 0.3038188043191994,
      "grad_norm": 3.271303653717041,
      "learning_rate": 3.481827758756914e-05,
      "loss": 1.2071,
      "step": 11536
    },
    {
      "epoch": 0.3038451409007111,
      "grad_norm": 3.1633152961730957,
      "learning_rate": 3.481696075849355e-05,
      "loss": 1.5983,
      "step": 11537
    },
    {
      "epoch": 0.3038714774822228,
      "grad_norm": 2.2482879161834717,
      "learning_rate": 3.4815643929417964e-05,
      "loss": 1.8539,
      "step": 11538
    },
    {
      "epoch": 0.3038978140637345,
      "grad_norm": 3.257370948791504,
      "learning_rate": 3.481432710034237e-05,
      "loss": 0.2941,
      "step": 11539
    },
    {
      "epoch": 0.30392415064524625,
      "grad_norm": 3.31890606880188,
      "learning_rate": 3.4813010271266795e-05,
      "loss": 1.6001,
      "step": 11540
    },
    {
      "epoch": 0.303950487226758,
      "grad_norm": 1.7338519096374512,
      "learning_rate": 3.4811693442191204e-05,
      "loss": 1.3351,
      "step": 11541
    },
    {
      "epoch": 0.3039768238082697,
      "grad_norm": 1.7897071838378906,
      "learning_rate": 3.481037661311562e-05,
      "loss": 1.5936,
      "step": 11542
    },
    {
      "epoch": 0.3040031603897814,
      "grad_norm": 2.6925737857818604,
      "learning_rate": 3.4809059784040035e-05,
      "loss": 1.4094,
      "step": 11543
    },
    {
      "epoch": 0.3040294969712931,
      "grad_norm": 2.764449119567871,
      "learning_rate": 3.4807742954964444e-05,
      "loss": 0.5718,
      "step": 11544
    },
    {
      "epoch": 0.30405583355280486,
      "grad_norm": 2.2823641300201416,
      "learning_rate": 3.4806426125888866e-05,
      "loss": 1.9221,
      "step": 11545
    },
    {
      "epoch": 0.30408217013431654,
      "grad_norm": 1.7916207313537598,
      "learning_rate": 3.4805109296813275e-05,
      "loss": 2.0903,
      "step": 11546
    },
    {
      "epoch": 0.3041085067158283,
      "grad_norm": 2.2723875045776367,
      "learning_rate": 3.480379246773769e-05,
      "loss": 1.1967,
      "step": 11547
    },
    {
      "epoch": 0.30413484329734003,
      "grad_norm": 3.0824718475341797,
      "learning_rate": 3.48024756386621e-05,
      "loss": 1.0949,
      "step": 11548
    },
    {
      "epoch": 0.3041611798788517,
      "grad_norm": 2.6962482929229736,
      "learning_rate": 3.4801158809586515e-05,
      "loss": 2.256,
      "step": 11549
    },
    {
      "epoch": 0.30418751646036346,
      "grad_norm": 1.5778225660324097,
      "learning_rate": 3.479984198051093e-05,
      "loss": 1.3628,
      "step": 11550
    },
    {
      "epoch": 0.30421385304187515,
      "grad_norm": 1.6850782632827759,
      "learning_rate": 3.4798525151435346e-05,
      "loss": 0.9326,
      "step": 11551
    },
    {
      "epoch": 0.3042401896233869,
      "grad_norm": 3.4709932804107666,
      "learning_rate": 3.479720832235976e-05,
      "loss": 1.3931,
      "step": 11552
    },
    {
      "epoch": 0.3042665262048986,
      "grad_norm": 3.33630108833313,
      "learning_rate": 3.479589149328417e-05,
      "loss": 1.1745,
      "step": 11553
    },
    {
      "epoch": 0.3042928627864103,
      "grad_norm": 2.5795185565948486,
      "learning_rate": 3.479457466420859e-05,
      "loss": 1.827,
      "step": 11554
    },
    {
      "epoch": 0.30431919936792207,
      "grad_norm": 2.360293388366699,
      "learning_rate": 3.4793257835133e-05,
      "loss": 2.0577,
      "step": 11555
    },
    {
      "epoch": 0.30434553594943375,
      "grad_norm": 1.9041166305541992,
      "learning_rate": 3.479194100605742e-05,
      "loss": 1.77,
      "step": 11556
    },
    {
      "epoch": 0.3043718725309455,
      "grad_norm": 4.251124858856201,
      "learning_rate": 3.4790624176981826e-05,
      "loss": 0.6899,
      "step": 11557
    },
    {
      "epoch": 0.3043982091124572,
      "grad_norm": 4.271191596984863,
      "learning_rate": 3.478930734790624e-05,
      "loss": 0.4294,
      "step": 11558
    },
    {
      "epoch": 0.30442454569396893,
      "grad_norm": 2.47845721244812,
      "learning_rate": 3.478799051883066e-05,
      "loss": 0.8647,
      "step": 11559
    },
    {
      "epoch": 0.3044508822754806,
      "grad_norm": 2.1287505626678467,
      "learning_rate": 3.478667368975507e-05,
      "loss": 0.6819,
      "step": 11560
    },
    {
      "epoch": 0.30447721885699236,
      "grad_norm": 2.419715642929077,
      "learning_rate": 3.478535686067949e-05,
      "loss": 2.5048,
      "step": 11561
    },
    {
      "epoch": 0.3045035554385041,
      "grad_norm": 2.0600006580352783,
      "learning_rate": 3.47840400316039e-05,
      "loss": 1.7711,
      "step": 11562
    },
    {
      "epoch": 0.3045298920200158,
      "grad_norm": 1.9151105880737305,
      "learning_rate": 3.478272320252831e-05,
      "loss": 1.4715,
      "step": 11563
    },
    {
      "epoch": 0.30455622860152753,
      "grad_norm": 3.447761297225952,
      "learning_rate": 3.478140637345273e-05,
      "loss": 1.1135,
      "step": 11564
    },
    {
      "epoch": 0.3045825651830392,
      "grad_norm": 2.368546962738037,
      "learning_rate": 3.4780089544377144e-05,
      "loss": 2.203,
      "step": 11565
    },
    {
      "epoch": 0.30460890176455097,
      "grad_norm": 1.8937956094741821,
      "learning_rate": 3.477877271530155e-05,
      "loss": 1.6095,
      "step": 11566
    },
    {
      "epoch": 0.30463523834606265,
      "grad_norm": 1.9551299810409546,
      "learning_rate": 3.477745588622597e-05,
      "loss": 1.8678,
      "step": 11567
    },
    {
      "epoch": 0.3046615749275744,
      "grad_norm": 2.2647898197174072,
      "learning_rate": 3.4776139057150384e-05,
      "loss": 2.5686,
      "step": 11568
    },
    {
      "epoch": 0.30468791150908614,
      "grad_norm": 4.3162102699279785,
      "learning_rate": 3.47748222280748e-05,
      "loss": 1.5623,
      "step": 11569
    },
    {
      "epoch": 0.30471424809059783,
      "grad_norm": 3.346144437789917,
      "learning_rate": 3.4773505398999215e-05,
      "loss": 1.1588,
      "step": 11570
    },
    {
      "epoch": 0.30474058467210957,
      "grad_norm": 3.9300930500030518,
      "learning_rate": 3.4772188569923624e-05,
      "loss": 0.7208,
      "step": 11571
    },
    {
      "epoch": 0.30476692125362126,
      "grad_norm": 1.9246171712875366,
      "learning_rate": 3.477087174084804e-05,
      "loss": 1.6903,
      "step": 11572
    },
    {
      "epoch": 0.304793257835133,
      "grad_norm": 3.4693009853363037,
      "learning_rate": 3.4769554911772455e-05,
      "loss": 1.4383,
      "step": 11573
    },
    {
      "epoch": 0.30481959441664475,
      "grad_norm": 2.248051404953003,
      "learning_rate": 3.476823808269687e-05,
      "loss": 1.1415,
      "step": 11574
    },
    {
      "epoch": 0.30484593099815643,
      "grad_norm": 4.649141311645508,
      "learning_rate": 3.476692125362128e-05,
      "loss": 1.0032,
      "step": 11575
    },
    {
      "epoch": 0.3048722675796682,
      "grad_norm": 4.173797130584717,
      "learning_rate": 3.4765604424545696e-05,
      "loss": 1.318,
      "step": 11576
    },
    {
      "epoch": 0.30489860416117986,
      "grad_norm": 1.853887677192688,
      "learning_rate": 3.4764287595470104e-05,
      "loss": 2.9107,
      "step": 11577
    },
    {
      "epoch": 0.3049249407426916,
      "grad_norm": 3.66133189201355,
      "learning_rate": 3.476297076639453e-05,
      "loss": 2.4301,
      "step": 11578
    },
    {
      "epoch": 0.3049512773242033,
      "grad_norm": 2.828031301498413,
      "learning_rate": 3.4761653937318936e-05,
      "loss": 1.98,
      "step": 11579
    },
    {
      "epoch": 0.30497761390571504,
      "grad_norm": 2.647550344467163,
      "learning_rate": 3.476033710824335e-05,
      "loss": 1.2626,
      "step": 11580
    },
    {
      "epoch": 0.3050039504872268,
      "grad_norm": 3.1275625228881836,
      "learning_rate": 3.475902027916777e-05,
      "loss": 1.2381,
      "step": 11581
    },
    {
      "epoch": 0.30503028706873847,
      "grad_norm": 2.6373543739318848,
      "learning_rate": 3.4757703450092176e-05,
      "loss": 0.8409,
      "step": 11582
    },
    {
      "epoch": 0.3050566236502502,
      "grad_norm": 2.427901029586792,
      "learning_rate": 3.47563866210166e-05,
      "loss": 1.2942,
      "step": 11583
    },
    {
      "epoch": 0.3050829602317619,
      "grad_norm": 2.0694851875305176,
      "learning_rate": 3.475506979194101e-05,
      "loss": 2.3386,
      "step": 11584
    },
    {
      "epoch": 0.30510929681327364,
      "grad_norm": 5.559150695800781,
      "learning_rate": 3.475375296286542e-05,
      "loss": 1.0728,
      "step": 11585
    },
    {
      "epoch": 0.30513563339478533,
      "grad_norm": 1.7264204025268555,
      "learning_rate": 3.475243613378983e-05,
      "loss": 1.8654,
      "step": 11586
    },
    {
      "epoch": 0.3051619699762971,
      "grad_norm": 1.5841379165649414,
      "learning_rate": 3.4751119304714254e-05,
      "loss": 2.2088,
      "step": 11587
    },
    {
      "epoch": 0.3051883065578088,
      "grad_norm": 5.1678032875061035,
      "learning_rate": 3.474980247563866e-05,
      "loss": 1.5628,
      "step": 11588
    },
    {
      "epoch": 0.3052146431393205,
      "grad_norm": 2.632446765899658,
      "learning_rate": 3.474848564656308e-05,
      "loss": 2.0084,
      "step": 11589
    },
    {
      "epoch": 0.30524097972083225,
      "grad_norm": 3.017375946044922,
      "learning_rate": 3.4747168817487494e-05,
      "loss": 1.7709,
      "step": 11590
    },
    {
      "epoch": 0.30526731630234394,
      "grad_norm": 2.235577344894409,
      "learning_rate": 3.47458519884119e-05,
      "loss": 1.1135,
      "step": 11591
    },
    {
      "epoch": 0.3052936528838557,
      "grad_norm": 1.7186636924743652,
      "learning_rate": 3.4744535159336325e-05,
      "loss": 0.7838,
      "step": 11592
    },
    {
      "epoch": 0.30531998946536737,
      "grad_norm": 1.8294529914855957,
      "learning_rate": 3.4743218330260734e-05,
      "loss": 0.4327,
      "step": 11593
    },
    {
      "epoch": 0.3053463260468791,
      "grad_norm": 4.340508460998535,
      "learning_rate": 3.474190150118515e-05,
      "loss": 0.7785,
      "step": 11594
    },
    {
      "epoch": 0.30537266262839086,
      "grad_norm": 2.3846845626831055,
      "learning_rate": 3.474058467210956e-05,
      "loss": 1.774,
      "step": 11595
    },
    {
      "epoch": 0.30539899920990254,
      "grad_norm": 2.962155342102051,
      "learning_rate": 3.4739267843033974e-05,
      "loss": 1.6014,
      "step": 11596
    },
    {
      "epoch": 0.3054253357914143,
      "grad_norm": 1.905815839767456,
      "learning_rate": 3.473795101395839e-05,
      "loss": 1.3828,
      "step": 11597
    },
    {
      "epoch": 0.305451672372926,
      "grad_norm": 2.779789686203003,
      "learning_rate": 3.4736634184882805e-05,
      "loss": 0.5156,
      "step": 11598
    },
    {
      "epoch": 0.3054780089544377,
      "grad_norm": 1.7144635915756226,
      "learning_rate": 3.473531735580722e-05,
      "loss": 1.6641,
      "step": 11599
    },
    {
      "epoch": 0.3055043455359494,
      "grad_norm": 2.321122169494629,
      "learning_rate": 3.473400052673163e-05,
      "loss": 1.8675,
      "step": 11600
    },
    {
      "epoch": 0.30553068211746115,
      "grad_norm": 1.7349375486373901,
      "learning_rate": 3.4732683697656045e-05,
      "loss": 1.6778,
      "step": 11601
    },
    {
      "epoch": 0.3055570186989729,
      "grad_norm": 3.526191473007202,
      "learning_rate": 3.473136686858046e-05,
      "loss": 1.0013,
      "step": 11602
    },
    {
      "epoch": 0.3055833552804846,
      "grad_norm": 3.1799354553222656,
      "learning_rate": 3.4730050039504876e-05,
      "loss": 1.6148,
      "step": 11603
    },
    {
      "epoch": 0.3056096918619963,
      "grad_norm": 5.7747368812561035,
      "learning_rate": 3.4728733210429285e-05,
      "loss": 1.5668,
      "step": 11604
    },
    {
      "epoch": 0.305636028443508,
      "grad_norm": 4.5626349449157715,
      "learning_rate": 3.47274163813537e-05,
      "loss": 1.7021,
      "step": 11605
    },
    {
      "epoch": 0.30566236502501976,
      "grad_norm": 3.611142635345459,
      "learning_rate": 3.4726099552278116e-05,
      "loss": 1.0385,
      "step": 11606
    },
    {
      "epoch": 0.3056887016065315,
      "grad_norm": 2.088200092315674,
      "learning_rate": 3.472478272320253e-05,
      "loss": 2.1357,
      "step": 11607
    },
    {
      "epoch": 0.3057150381880432,
      "grad_norm": 1.9994417428970337,
      "learning_rate": 3.472346589412695e-05,
      "loss": 0.4828,
      "step": 11608
    },
    {
      "epoch": 0.30574137476955493,
      "grad_norm": 2.1813161373138428,
      "learning_rate": 3.4722149065051356e-05,
      "loss": 2.2199,
      "step": 11609
    },
    {
      "epoch": 0.3057677113510666,
      "grad_norm": 2.301156520843506,
      "learning_rate": 3.472083223597577e-05,
      "loss": 0.5075,
      "step": 11610
    },
    {
      "epoch": 0.30579404793257836,
      "grad_norm": 3.0289649963378906,
      "learning_rate": 3.471951540690019e-05,
      "loss": 1.7263,
      "step": 11611
    },
    {
      "epoch": 0.30582038451409005,
      "grad_norm": 2.1083133220672607,
      "learning_rate": 3.47181985778246e-05,
      "loss": 2.0567,
      "step": 11612
    },
    {
      "epoch": 0.3058467210956018,
      "grad_norm": 3.003511905670166,
      "learning_rate": 3.471688174874901e-05,
      "loss": 1.9288,
      "step": 11613
    },
    {
      "epoch": 0.30587305767711354,
      "grad_norm": 2.906557083129883,
      "learning_rate": 3.471556491967343e-05,
      "loss": 1.253,
      "step": 11614
    },
    {
      "epoch": 0.3058993942586252,
      "grad_norm": 2.2326910495758057,
      "learning_rate": 3.471424809059784e-05,
      "loss": 2.158,
      "step": 11615
    },
    {
      "epoch": 0.30592573084013697,
      "grad_norm": 2.362025260925293,
      "learning_rate": 3.471293126152226e-05,
      "loss": 1.4417,
      "step": 11616
    },
    {
      "epoch": 0.30595206742164865,
      "grad_norm": 6.059136867523193,
      "learning_rate": 3.4711614432446674e-05,
      "loss": 0.779,
      "step": 11617
    },
    {
      "epoch": 0.3059784040031604,
      "grad_norm": 1.6216179132461548,
      "learning_rate": 3.471029760337108e-05,
      "loss": 0.7152,
      "step": 11618
    },
    {
      "epoch": 0.3060047405846721,
      "grad_norm": 1.8468326330184937,
      "learning_rate": 3.47089807742955e-05,
      "loss": 1.939,
      "step": 11619
    },
    {
      "epoch": 0.30603107716618383,
      "grad_norm": 2.2648494243621826,
      "learning_rate": 3.4707663945219914e-05,
      "loss": 0.7205,
      "step": 11620
    },
    {
      "epoch": 0.3060574137476956,
      "grad_norm": 2.741974115371704,
      "learning_rate": 3.470634711614433e-05,
      "loss": 0.9894,
      "step": 11621
    },
    {
      "epoch": 0.30608375032920726,
      "grad_norm": 1.7147393226623535,
      "learning_rate": 3.470503028706874e-05,
      "loss": 1.8034,
      "step": 11622
    },
    {
      "epoch": 0.306110086910719,
      "grad_norm": 1.8979274034500122,
      "learning_rate": 3.4703713457993154e-05,
      "loss": 2.1463,
      "step": 11623
    },
    {
      "epoch": 0.3061364234922307,
      "grad_norm": 7.656367301940918,
      "learning_rate": 3.470239662891756e-05,
      "loss": 2.3868,
      "step": 11624
    },
    {
      "epoch": 0.30616276007374243,
      "grad_norm": 1.894753336906433,
      "learning_rate": 3.4701079799841985e-05,
      "loss": 1.416,
      "step": 11625
    },
    {
      "epoch": 0.3061890966552541,
      "grad_norm": 2.048351526260376,
      "learning_rate": 3.4699762970766394e-05,
      "loss": 1.7095,
      "step": 11626
    },
    {
      "epoch": 0.30621543323676587,
      "grad_norm": 3.1628544330596924,
      "learning_rate": 3.469844614169081e-05,
      "loss": 1.5224,
      "step": 11627
    },
    {
      "epoch": 0.3062417698182776,
      "grad_norm": 1.853131890296936,
      "learning_rate": 3.4697129312615225e-05,
      "loss": 1.2695,
      "step": 11628
    },
    {
      "epoch": 0.3062681063997893,
      "grad_norm": 3.5873470306396484,
      "learning_rate": 3.4695812483539634e-05,
      "loss": 2.2628,
      "step": 11629
    },
    {
      "epoch": 0.30629444298130104,
      "grad_norm": 2.3380956649780273,
      "learning_rate": 3.4694495654464056e-05,
      "loss": 2.4333,
      "step": 11630
    },
    {
      "epoch": 0.30632077956281273,
      "grad_norm": 2.702810049057007,
      "learning_rate": 3.4693178825388465e-05,
      "loss": 1.7219,
      "step": 11631
    },
    {
      "epoch": 0.30634711614432447,
      "grad_norm": 1.5835646390914917,
      "learning_rate": 3.469186199631288e-05,
      "loss": 1.5005,
      "step": 11632
    },
    {
      "epoch": 0.30637345272583616,
      "grad_norm": 3.7774085998535156,
      "learning_rate": 3.469054516723729e-05,
      "loss": 1.76,
      "step": 11633
    },
    {
      "epoch": 0.3063997893073479,
      "grad_norm": 4.889803886413574,
      "learning_rate": 3.4689228338161705e-05,
      "loss": 0.8578,
      "step": 11634
    },
    {
      "epoch": 0.30642612588885965,
      "grad_norm": 2.427330732345581,
      "learning_rate": 3.468791150908612e-05,
      "loss": 1.3588,
      "step": 11635
    },
    {
      "epoch": 0.30645246247037133,
      "grad_norm": 1.8620880842208862,
      "learning_rate": 3.4686594680010536e-05,
      "loss": 1.6097,
      "step": 11636
    },
    {
      "epoch": 0.3064787990518831,
      "grad_norm": 1.6033748388290405,
      "learning_rate": 3.468527785093495e-05,
      "loss": 1.4988,
      "step": 11637
    },
    {
      "epoch": 0.30650513563339477,
      "grad_norm": 3.518514633178711,
      "learning_rate": 3.468396102185936e-05,
      "loss": 1.7217,
      "step": 11638
    },
    {
      "epoch": 0.3065314722149065,
      "grad_norm": 1.8962196111679077,
      "learning_rate": 3.468264419278378e-05,
      "loss": 1.7625,
      "step": 11639
    },
    {
      "epoch": 0.30655780879641825,
      "grad_norm": 2.0571370124816895,
      "learning_rate": 3.468132736370819e-05,
      "loss": 2.5235,
      "step": 11640
    },
    {
      "epoch": 0.30658414537792994,
      "grad_norm": 1.7881693840026855,
      "learning_rate": 3.468001053463261e-05,
      "loss": 1.7532,
      "step": 11641
    },
    {
      "epoch": 0.3066104819594417,
      "grad_norm": 7.386948108673096,
      "learning_rate": 3.4678693705557017e-05,
      "loss": 1.8551,
      "step": 11642
    },
    {
      "epoch": 0.30663681854095337,
      "grad_norm": 1.7662712335586548,
      "learning_rate": 3.467737687648143e-05,
      "loss": 2.0711,
      "step": 11643
    },
    {
      "epoch": 0.3066631551224651,
      "grad_norm": 1.8915596008300781,
      "learning_rate": 3.467606004740585e-05,
      "loss": 1.4657,
      "step": 11644
    },
    {
      "epoch": 0.3066894917039768,
      "grad_norm": 4.346502780914307,
      "learning_rate": 3.467474321833026e-05,
      "loss": 1.7115,
      "step": 11645
    },
    {
      "epoch": 0.30671582828548855,
      "grad_norm": 4.3162031173706055,
      "learning_rate": 3.467342638925468e-05,
      "loss": 1.4795,
      "step": 11646
    },
    {
      "epoch": 0.3067421648670003,
      "grad_norm": 3.1514456272125244,
      "learning_rate": 3.467210956017909e-05,
      "loss": 1.6011,
      "step": 11647
    },
    {
      "epoch": 0.306768501448512,
      "grad_norm": 1.5485625267028809,
      "learning_rate": 3.46707927311035e-05,
      "loss": 1.8156,
      "step": 11648
    },
    {
      "epoch": 0.3067948380300237,
      "grad_norm": 1.799734115600586,
      "learning_rate": 3.466947590202792e-05,
      "loss": 1.9392,
      "step": 11649
    },
    {
      "epoch": 0.3068211746115354,
      "grad_norm": 3.8530821800231934,
      "learning_rate": 3.4668159072952335e-05,
      "loss": 1.6469,
      "step": 11650
    },
    {
      "epoch": 0.30684751119304715,
      "grad_norm": 1.5377084016799927,
      "learning_rate": 3.466684224387674e-05,
      "loss": 1.6315,
      "step": 11651
    },
    {
      "epoch": 0.30687384777455884,
      "grad_norm": 2.5824930667877197,
      "learning_rate": 3.466552541480116e-05,
      "loss": 1.3066,
      "step": 11652
    },
    {
      "epoch": 0.3069001843560706,
      "grad_norm": 1.6987619400024414,
      "learning_rate": 3.4664208585725575e-05,
      "loss": 1.6432,
      "step": 11653
    },
    {
      "epoch": 0.3069265209375823,
      "grad_norm": 1.9125723838806152,
      "learning_rate": 3.466289175664999e-05,
      "loss": 1.8698,
      "step": 11654
    },
    {
      "epoch": 0.306952857519094,
      "grad_norm": 2.4199650287628174,
      "learning_rate": 3.4661574927574406e-05,
      "loss": 1.7335,
      "step": 11655
    },
    {
      "epoch": 0.30697919410060576,
      "grad_norm": 1.516898512840271,
      "learning_rate": 3.4660258098498815e-05,
      "loss": 1.7069,
      "step": 11656
    },
    {
      "epoch": 0.30700553068211744,
      "grad_norm": 2.0510690212249756,
      "learning_rate": 3.465894126942323e-05,
      "loss": 1.6581,
      "step": 11657
    },
    {
      "epoch": 0.3070318672636292,
      "grad_norm": 2.3611223697662354,
      "learning_rate": 3.4657624440347646e-05,
      "loss": 0.9748,
      "step": 11658
    },
    {
      "epoch": 0.3070582038451409,
      "grad_norm": 2.9073562622070312,
      "learning_rate": 3.465630761127206e-05,
      "loss": 1.6822,
      "step": 11659
    },
    {
      "epoch": 0.3070845404266526,
      "grad_norm": 2.41892409324646,
      "learning_rate": 3.465499078219647e-05,
      "loss": 0.5274,
      "step": 11660
    },
    {
      "epoch": 0.30711087700816436,
      "grad_norm": 2.74349045753479,
      "learning_rate": 3.4653673953120886e-05,
      "loss": 1.6235,
      "step": 11661
    },
    {
      "epoch": 0.30713721358967605,
      "grad_norm": 2.5569379329681396,
      "learning_rate": 3.46523571240453e-05,
      "loss": 1.648,
      "step": 11662
    },
    {
      "epoch": 0.3071635501711878,
      "grad_norm": 2.395657539367676,
      "learning_rate": 3.465104029496972e-05,
      "loss": 1.8136,
      "step": 11663
    },
    {
      "epoch": 0.3071898867526995,
      "grad_norm": 3.543126106262207,
      "learning_rate": 3.464972346589413e-05,
      "loss": 0.8016,
      "step": 11664
    },
    {
      "epoch": 0.3072162233342112,
      "grad_norm": 1.9433414936065674,
      "learning_rate": 3.464840663681854e-05,
      "loss": 2.0382,
      "step": 11665
    },
    {
      "epoch": 0.3072425599157229,
      "grad_norm": 4.567250728607178,
      "learning_rate": 3.464708980774296e-05,
      "loss": 1.3142,
      "step": 11666
    },
    {
      "epoch": 0.30726889649723466,
      "grad_norm": 2.566758871078491,
      "learning_rate": 3.4645772978667366e-05,
      "loss": 2.0811,
      "step": 11667
    },
    {
      "epoch": 0.3072952330787464,
      "grad_norm": 2.4919593334198,
      "learning_rate": 3.464445614959179e-05,
      "loss": 1.4652,
      "step": 11668
    },
    {
      "epoch": 0.3073215696602581,
      "grad_norm": 1.6978918313980103,
      "learning_rate": 3.46431393205162e-05,
      "loss": 2.0075,
      "step": 11669
    },
    {
      "epoch": 0.30734790624176983,
      "grad_norm": 3.3036160469055176,
      "learning_rate": 3.464182249144061e-05,
      "loss": 2.373,
      "step": 11670
    },
    {
      "epoch": 0.3073742428232815,
      "grad_norm": 3.3027424812316895,
      "learning_rate": 3.464050566236502e-05,
      "loss": 1.166,
      "step": 11671
    },
    {
      "epoch": 0.30740057940479326,
      "grad_norm": 1.8507441282272339,
      "learning_rate": 3.4639188833289444e-05,
      "loss": 1.8131,
      "step": 11672
    },
    {
      "epoch": 0.307426915986305,
      "grad_norm": 3.1358394622802734,
      "learning_rate": 3.463787200421386e-05,
      "loss": 1.8126,
      "step": 11673
    },
    {
      "epoch": 0.3074532525678167,
      "grad_norm": 3.907683849334717,
      "learning_rate": 3.463655517513827e-05,
      "loss": 1.7927,
      "step": 11674
    },
    {
      "epoch": 0.30747958914932844,
      "grad_norm": 2.0035784244537354,
      "learning_rate": 3.4635238346062684e-05,
      "loss": 1.9221,
      "step": 11675
    },
    {
      "epoch": 0.3075059257308401,
      "grad_norm": 1.9177042245864868,
      "learning_rate": 3.463392151698709e-05,
      "loss": 1.5778,
      "step": 11676
    },
    {
      "epoch": 0.30753226231235187,
      "grad_norm": 3.768989086151123,
      "learning_rate": 3.4632604687911515e-05,
      "loss": 0.9479,
      "step": 11677
    },
    {
      "epoch": 0.30755859889386356,
      "grad_norm": 3.2747578620910645,
      "learning_rate": 3.4631287858835924e-05,
      "loss": 1.4946,
      "step": 11678
    },
    {
      "epoch": 0.3075849354753753,
      "grad_norm": 1.6375569105148315,
      "learning_rate": 3.462997102976034e-05,
      "loss": 2.1016,
      "step": 11679
    },
    {
      "epoch": 0.30761127205688704,
      "grad_norm": 2.23073410987854,
      "learning_rate": 3.462865420068475e-05,
      "loss": 1.6804,
      "step": 11680
    },
    {
      "epoch": 0.30763760863839873,
      "grad_norm": 2.1874592304229736,
      "learning_rate": 3.4627337371609164e-05,
      "loss": 0.4643,
      "step": 11681
    },
    {
      "epoch": 0.3076639452199105,
      "grad_norm": 1.587695837020874,
      "learning_rate": 3.462602054253358e-05,
      "loss": 1.3702,
      "step": 11682
    },
    {
      "epoch": 0.30769028180142216,
      "grad_norm": 3.9395804405212402,
      "learning_rate": 3.4624703713457995e-05,
      "loss": 1.0728,
      "step": 11683
    },
    {
      "epoch": 0.3077166183829339,
      "grad_norm": 1.6745870113372803,
      "learning_rate": 3.462338688438241e-05,
      "loss": 0.7248,
      "step": 11684
    },
    {
      "epoch": 0.3077429549644456,
      "grad_norm": 2.6220033168792725,
      "learning_rate": 3.462207005530682e-05,
      "loss": 1.2804,
      "step": 11685
    },
    {
      "epoch": 0.30776929154595734,
      "grad_norm": 1.884389042854309,
      "learning_rate": 3.462075322623124e-05,
      "loss": 2.0532,
      "step": 11686
    },
    {
      "epoch": 0.3077956281274691,
      "grad_norm": 3.220917224884033,
      "learning_rate": 3.461943639715565e-05,
      "loss": 1.6703,
      "step": 11687
    },
    {
      "epoch": 0.30782196470898077,
      "grad_norm": 1.9681662321090698,
      "learning_rate": 3.4618119568080066e-05,
      "loss": 1.8496,
      "step": 11688
    },
    {
      "epoch": 0.3078483012904925,
      "grad_norm": 1.8144863843917847,
      "learning_rate": 3.4616802739004475e-05,
      "loss": 1.5183,
      "step": 11689
    },
    {
      "epoch": 0.3078746378720042,
      "grad_norm": 2.708662509918213,
      "learning_rate": 3.461548590992889e-05,
      "loss": 2.3922,
      "step": 11690
    },
    {
      "epoch": 0.30790097445351594,
      "grad_norm": 3.191736936569214,
      "learning_rate": 3.4614169080853306e-05,
      "loss": 1.3397,
      "step": 11691
    },
    {
      "epoch": 0.30792731103502763,
      "grad_norm": 1.8826956748962402,
      "learning_rate": 3.461285225177772e-05,
      "loss": 1.8414,
      "step": 11692
    },
    {
      "epoch": 0.30795364761653937,
      "grad_norm": 3.622114658355713,
      "learning_rate": 3.461153542270214e-05,
      "loss": 1.5846,
      "step": 11693
    },
    {
      "epoch": 0.3079799841980511,
      "grad_norm": 1.8765538930892944,
      "learning_rate": 3.4610218593626546e-05,
      "loss": 1.5439,
      "step": 11694
    },
    {
      "epoch": 0.3080063207795628,
      "grad_norm": 1.8673611879348755,
      "learning_rate": 3.460890176455096e-05,
      "loss": 2.4222,
      "step": 11695
    },
    {
      "epoch": 0.30803265736107455,
      "grad_norm": 1.5874181985855103,
      "learning_rate": 3.460758493547538e-05,
      "loss": 1.9228,
      "step": 11696
    },
    {
      "epoch": 0.30805899394258623,
      "grad_norm": 2.460047483444214,
      "learning_rate": 3.460626810639979e-05,
      "loss": 1.2548,
      "step": 11697
    },
    {
      "epoch": 0.308085330524098,
      "grad_norm": 2.4794962406158447,
      "learning_rate": 3.46049512773242e-05,
      "loss": 1.5382,
      "step": 11698
    },
    {
      "epoch": 0.30811166710560967,
      "grad_norm": 2.0717644691467285,
      "learning_rate": 3.460363444824862e-05,
      "loss": 1.8282,
      "step": 11699
    },
    {
      "epoch": 0.3081380036871214,
      "grad_norm": 2.1402463912963867,
      "learning_rate": 3.460231761917303e-05,
      "loss": 1.457,
      "step": 11700
    },
    {
      "epoch": 0.30816434026863315,
      "grad_norm": 2.433180570602417,
      "learning_rate": 3.460100079009745e-05,
      "loss": 1.6213,
      "step": 11701
    },
    {
      "epoch": 0.30819067685014484,
      "grad_norm": 2.736400842666626,
      "learning_rate": 3.4599683961021864e-05,
      "loss": 0.968,
      "step": 11702
    },
    {
      "epoch": 0.3082170134316566,
      "grad_norm": 2.3250274658203125,
      "learning_rate": 3.459836713194627e-05,
      "loss": 1.9955,
      "step": 11703
    },
    {
      "epoch": 0.30824335001316827,
      "grad_norm": 1.7911103963851929,
      "learning_rate": 3.459705030287069e-05,
      "loss": 1.6743,
      "step": 11704
    },
    {
      "epoch": 0.30826968659468,
      "grad_norm": 2.7280426025390625,
      "learning_rate": 3.4595733473795104e-05,
      "loss": 1.3697,
      "step": 11705
    },
    {
      "epoch": 0.30829602317619176,
      "grad_norm": 3.187476873397827,
      "learning_rate": 3.459441664471952e-05,
      "loss": 0.7183,
      "step": 11706
    },
    {
      "epoch": 0.30832235975770345,
      "grad_norm": 2.6914868354797363,
      "learning_rate": 3.459309981564393e-05,
      "loss": 1.1978,
      "step": 11707
    },
    {
      "epoch": 0.3083486963392152,
      "grad_norm": 3.1083109378814697,
      "learning_rate": 3.4591782986568344e-05,
      "loss": 1.9836,
      "step": 11708
    },
    {
      "epoch": 0.3083750329207269,
      "grad_norm": 2.3045692443847656,
      "learning_rate": 3.459046615749276e-05,
      "loss": 1.3302,
      "step": 11709
    },
    {
      "epoch": 0.3084013695022386,
      "grad_norm": 3.2504324913024902,
      "learning_rate": 3.4589149328417176e-05,
      "loss": 1.9067,
      "step": 11710
    },
    {
      "epoch": 0.3084277060837503,
      "grad_norm": 1.6680676937103271,
      "learning_rate": 3.458783249934159e-05,
      "loss": 2.6122,
      "step": 11711
    },
    {
      "epoch": 0.30845404266526205,
      "grad_norm": 1.7473156452178955,
      "learning_rate": 3.4586515670266e-05,
      "loss": 2.3533,
      "step": 11712
    },
    {
      "epoch": 0.3084803792467738,
      "grad_norm": 3.5321526527404785,
      "learning_rate": 3.4585198841190416e-05,
      "loss": 1.2481,
      "step": 11713
    },
    {
      "epoch": 0.3085067158282855,
      "grad_norm": 2.50329852104187,
      "learning_rate": 3.4583882012114824e-05,
      "loss": 1.6297,
      "step": 11714
    },
    {
      "epoch": 0.3085330524097972,
      "grad_norm": 2.940936803817749,
      "learning_rate": 3.458256518303925e-05,
      "loss": 1.9327,
      "step": 11715
    },
    {
      "epoch": 0.3085593889913089,
      "grad_norm": 3.2888314723968506,
      "learning_rate": 3.4581248353963656e-05,
      "loss": 0.7863,
      "step": 11716
    },
    {
      "epoch": 0.30858572557282066,
      "grad_norm": 2.1054883003234863,
      "learning_rate": 3.457993152488807e-05,
      "loss": 2.4678,
      "step": 11717
    },
    {
      "epoch": 0.30861206215433235,
      "grad_norm": 3.3386595249176025,
      "learning_rate": 3.457861469581249e-05,
      "loss": 0.3001,
      "step": 11718
    },
    {
      "epoch": 0.3086383987358441,
      "grad_norm": 2.0013539791107178,
      "learning_rate": 3.45772978667369e-05,
      "loss": 2.0131,
      "step": 11719
    },
    {
      "epoch": 0.30866473531735583,
      "grad_norm": 2.938347339630127,
      "learning_rate": 3.457598103766132e-05,
      "loss": 0.8582,
      "step": 11720
    },
    {
      "epoch": 0.3086910718988675,
      "grad_norm": 1.8072912693023682,
      "learning_rate": 3.457466420858573e-05,
      "loss": 0.5579,
      "step": 11721
    },
    {
      "epoch": 0.30871740848037926,
      "grad_norm": 1.6535812616348267,
      "learning_rate": 3.457334737951014e-05,
      "loss": 2.0589,
      "step": 11722
    },
    {
      "epoch": 0.30874374506189095,
      "grad_norm": 2.8924241065979004,
      "learning_rate": 3.457203055043455e-05,
      "loss": 1.8899,
      "step": 11723
    },
    {
      "epoch": 0.3087700816434027,
      "grad_norm": 3.0967254638671875,
      "learning_rate": 3.4570713721358974e-05,
      "loss": 1.9702,
      "step": 11724
    },
    {
      "epoch": 0.3087964182249144,
      "grad_norm": 1.8100147247314453,
      "learning_rate": 3.456939689228338e-05,
      "loss": 1.3615,
      "step": 11725
    },
    {
      "epoch": 0.3088227548064261,
      "grad_norm": 3.3585965633392334,
      "learning_rate": 3.45680800632078e-05,
      "loss": 1.3715,
      "step": 11726
    },
    {
      "epoch": 0.30884909138793787,
      "grad_norm": 1.8501466512680054,
      "learning_rate": 3.456676323413221e-05,
      "loss": 1.6813,
      "step": 11727
    },
    {
      "epoch": 0.30887542796944956,
      "grad_norm": 1.6948282718658447,
      "learning_rate": 3.456544640505662e-05,
      "loss": 1.5062,
      "step": 11728
    },
    {
      "epoch": 0.3089017645509613,
      "grad_norm": 3.161344289779663,
      "learning_rate": 3.456412957598104e-05,
      "loss": 1.7149,
      "step": 11729
    },
    {
      "epoch": 0.308928101132473,
      "grad_norm": 2.444887638092041,
      "learning_rate": 3.4562812746905454e-05,
      "loss": 1.9189,
      "step": 11730
    },
    {
      "epoch": 0.30895443771398473,
      "grad_norm": 1.7201979160308838,
      "learning_rate": 3.456149591782987e-05,
      "loss": 2.4303,
      "step": 11731
    },
    {
      "epoch": 0.3089807742954964,
      "grad_norm": 1.5144646167755127,
      "learning_rate": 3.456017908875428e-05,
      "loss": 1.898,
      "step": 11732
    },
    {
      "epoch": 0.30900711087700816,
      "grad_norm": 2.360762596130371,
      "learning_rate": 3.4558862259678694e-05,
      "loss": 2.447,
      "step": 11733
    },
    {
      "epoch": 0.3090334474585199,
      "grad_norm": 1.8019884824752808,
      "learning_rate": 3.455754543060311e-05,
      "loss": 2.1675,
      "step": 11734
    },
    {
      "epoch": 0.3090597840400316,
      "grad_norm": 2.088752269744873,
      "learning_rate": 3.4556228601527525e-05,
      "loss": 1.6127,
      "step": 11735
    },
    {
      "epoch": 0.30908612062154334,
      "grad_norm": NaN,
      "learning_rate": 3.4556228601527525e-05,
      "loss": 1.6811,
      "step": 11736
    },
    {
      "epoch": 0.309112457203055,
      "grad_norm": 1.9876090288162231,
      "learning_rate": 3.4554911772451934e-05,
      "loss": 1.4072,
      "step": 11737
    },
    {
      "epoch": 0.30913879378456677,
      "grad_norm": 1.7965960502624512,
      "learning_rate": 3.455359494337635e-05,
      "loss": 1.9114,
      "step": 11738
    },
    {
      "epoch": 0.30916513036607846,
      "grad_norm": 8.034951210021973,
      "learning_rate": 3.4552278114300765e-05,
      "loss": 1.8327,
      "step": 11739
    },
    {
      "epoch": 0.3091914669475902,
      "grad_norm": 1.7249491214752197,
      "learning_rate": 3.455096128522518e-05,
      "loss": 1.6869,
      "step": 11740
    },
    {
      "epoch": 0.30921780352910194,
      "grad_norm": 3.2184062004089355,
      "learning_rate": 3.4549644456149596e-05,
      "loss": 1.1492,
      "step": 11741
    },
    {
      "epoch": 0.30924414011061363,
      "grad_norm": 1.6110557317733765,
      "learning_rate": 3.4548327627074005e-05,
      "loss": 1.6734,
      "step": 11742
    },
    {
      "epoch": 0.3092704766921254,
      "grad_norm": 2.355536460876465,
      "learning_rate": 3.454701079799842e-05,
      "loss": 1.1599,
      "step": 11743
    },
    {
      "epoch": 0.30929681327363706,
      "grad_norm": 1.6065218448638916,
      "learning_rate": 3.4545693968922836e-05,
      "loss": 2.1427,
      "step": 11744
    },
    {
      "epoch": 0.3093231498551488,
      "grad_norm": 2.143044948577881,
      "learning_rate": 3.454437713984725e-05,
      "loss": 1.647,
      "step": 11745
    },
    {
      "epoch": 0.30934948643666055,
      "grad_norm": 2.4601407051086426,
      "learning_rate": 3.454306031077166e-05,
      "loss": 1.7834,
      "step": 11746
    },
    {
      "epoch": 0.30937582301817224,
      "grad_norm": 2.9811441898345947,
      "learning_rate": 3.4541743481696076e-05,
      "loss": 1.9454,
      "step": 11747
    },
    {
      "epoch": 0.309402159599684,
      "grad_norm": 1.672289490699768,
      "learning_rate": 3.454042665262049e-05,
      "loss": 2.2854,
      "step": 11748
    },
    {
      "epoch": 0.30942849618119567,
      "grad_norm": 4.177724838256836,
      "learning_rate": 3.453910982354491e-05,
      "loss": 2.0358,
      "step": 11749
    },
    {
      "epoch": 0.3094548327627074,
      "grad_norm": 1.5616108179092407,
      "learning_rate": 3.453779299446932e-05,
      "loss": 2.1984,
      "step": 11750
    },
    {
      "epoch": 0.3094811693442191,
      "grad_norm": 3.5323145389556885,
      "learning_rate": 3.453647616539373e-05,
      "loss": 0.4883,
      "step": 11751
    },
    {
      "epoch": 0.30950750592573084,
      "grad_norm": 2.3145110607147217,
      "learning_rate": 3.453515933631815e-05,
      "loss": 0.9525,
      "step": 11752
    },
    {
      "epoch": 0.3095338425072426,
      "grad_norm": 1.8923053741455078,
      "learning_rate": 3.453384250724256e-05,
      "loss": 0.687,
      "step": 11753
    },
    {
      "epoch": 0.3095601790887543,
      "grad_norm": 2.0831992626190186,
      "learning_rate": 3.453252567816698e-05,
      "loss": 1.9662,
      "step": 11754
    },
    {
      "epoch": 0.309586515670266,
      "grad_norm": 3.045464515686035,
      "learning_rate": 3.453120884909139e-05,
      "loss": 1.7909,
      "step": 11755
    },
    {
      "epoch": 0.3096128522517777,
      "grad_norm": 2.918975591659546,
      "learning_rate": 3.45298920200158e-05,
      "loss": 0.8279,
      "step": 11756
    },
    {
      "epoch": 0.30963918883328945,
      "grad_norm": 6.363572120666504,
      "learning_rate": 3.452857519094022e-05,
      "loss": 2.8167,
      "step": 11757
    },
    {
      "epoch": 0.30966552541480113,
      "grad_norm": 2.8602516651153564,
      "learning_rate": 3.4527258361864634e-05,
      "loss": 0.9209,
      "step": 11758
    },
    {
      "epoch": 0.3096918619963129,
      "grad_norm": 2.3194499015808105,
      "learning_rate": 3.452594153278905e-05,
      "loss": 1.5712,
      "step": 11759
    },
    {
      "epoch": 0.3097181985778246,
      "grad_norm": 1.7898188829421997,
      "learning_rate": 3.452462470371346e-05,
      "loss": 1.7326,
      "step": 11760
    },
    {
      "epoch": 0.3097445351593363,
      "grad_norm": 2.186042308807373,
      "learning_rate": 3.4523307874637874e-05,
      "loss": 2.3141,
      "step": 11761
    },
    {
      "epoch": 0.30977087174084805,
      "grad_norm": 2.2598307132720947,
      "learning_rate": 3.452199104556228e-05,
      "loss": 1.4974,
      "step": 11762
    },
    {
      "epoch": 0.30979720832235974,
      "grad_norm": 2.0029029846191406,
      "learning_rate": 3.4520674216486705e-05,
      "loss": 1.0925,
      "step": 11763
    },
    {
      "epoch": 0.3098235449038715,
      "grad_norm": 2.267841339111328,
      "learning_rate": 3.4519357387411114e-05,
      "loss": 2.3699,
      "step": 11764
    },
    {
      "epoch": 0.30984988148538317,
      "grad_norm": 1.9859501123428345,
      "learning_rate": 3.451804055833553e-05,
      "loss": 2.1064,
      "step": 11765
    },
    {
      "epoch": 0.3098762180668949,
      "grad_norm": 1.9182748794555664,
      "learning_rate": 3.4516723729259945e-05,
      "loss": 2.3219,
      "step": 11766
    },
    {
      "epoch": 0.30990255464840666,
      "grad_norm": 1.7138694524765015,
      "learning_rate": 3.4515406900184354e-05,
      "loss": 1.3656,
      "step": 11767
    },
    {
      "epoch": 0.30992889122991835,
      "grad_norm": 2.314241886138916,
      "learning_rate": 3.4514090071108777e-05,
      "loss": 1.4219,
      "step": 11768
    },
    {
      "epoch": 0.3099552278114301,
      "grad_norm": 4.386459827423096,
      "learning_rate": 3.4512773242033185e-05,
      "loss": 1.7324,
      "step": 11769
    },
    {
      "epoch": 0.3099815643929418,
      "grad_norm": 1.7235097885131836,
      "learning_rate": 3.45114564129576e-05,
      "loss": 1.9367,
      "step": 11770
    },
    {
      "epoch": 0.3100079009744535,
      "grad_norm": 3.6874759197235107,
      "learning_rate": 3.451013958388201e-05,
      "loss": 1.3602,
      "step": 11771
    },
    {
      "epoch": 0.3100342375559652,
      "grad_norm": 2.5705721378326416,
      "learning_rate": 3.450882275480643e-05,
      "loss": 1.8409,
      "step": 11772
    },
    {
      "epoch": 0.31006057413747695,
      "grad_norm": 1.8111506700515747,
      "learning_rate": 3.450750592573084e-05,
      "loss": 1.9093,
      "step": 11773
    },
    {
      "epoch": 0.3100869107189887,
      "grad_norm": 2.2576937675476074,
      "learning_rate": 3.4506189096655257e-05,
      "loss": 1.5432,
      "step": 11774
    },
    {
      "epoch": 0.3101132473005004,
      "grad_norm": 1.5894970893859863,
      "learning_rate": 3.4504872267579665e-05,
      "loss": 1.3978,
      "step": 11775
    },
    {
      "epoch": 0.3101395838820121,
      "grad_norm": 2.517726182937622,
      "learning_rate": 3.450355543850408e-05,
      "loss": 2.1917,
      "step": 11776
    },
    {
      "epoch": 0.3101659204635238,
      "grad_norm": 1.5919756889343262,
      "learning_rate": 3.4502238609428497e-05,
      "loss": 2.3847,
      "step": 11777
    },
    {
      "epoch": 0.31019225704503556,
      "grad_norm": 2.426539659500122,
      "learning_rate": 3.450092178035291e-05,
      "loss": 1.6726,
      "step": 11778
    },
    {
      "epoch": 0.3102185936265473,
      "grad_norm": 3.7654507160186768,
      "learning_rate": 3.449960495127733e-05,
      "loss": 1.026,
      "step": 11779
    },
    {
      "epoch": 0.310244930208059,
      "grad_norm": 1.7406479120254517,
      "learning_rate": 3.4498288122201737e-05,
      "loss": 2.3115,
      "step": 11780
    },
    {
      "epoch": 0.31027126678957073,
      "grad_norm": 3.847716808319092,
      "learning_rate": 3.449697129312615e-05,
      "loss": 1.7675,
      "step": 11781
    },
    {
      "epoch": 0.3102976033710824,
      "grad_norm": 2.8625152111053467,
      "learning_rate": 3.449565446405057e-05,
      "loss": 1.4775,
      "step": 11782
    },
    {
      "epoch": 0.31032393995259416,
      "grad_norm": 1.706165075302124,
      "learning_rate": 3.449433763497498e-05,
      "loss": 1.8195,
      "step": 11783
    },
    {
      "epoch": 0.31035027653410585,
      "grad_norm": 3.5676729679107666,
      "learning_rate": 3.449302080589939e-05,
      "loss": 1.9714,
      "step": 11784
    },
    {
      "epoch": 0.3103766131156176,
      "grad_norm": 1.5900192260742188,
      "learning_rate": 3.449170397682381e-05,
      "loss": 1.1815,
      "step": 11785
    },
    {
      "epoch": 0.31040294969712934,
      "grad_norm": 2.0018367767333984,
      "learning_rate": 3.449038714774822e-05,
      "loss": 1.9062,
      "step": 11786
    },
    {
      "epoch": 0.310429286278641,
      "grad_norm": 2.294731855392456,
      "learning_rate": 3.448907031867264e-05,
      "loss": 1.867,
      "step": 11787
    },
    {
      "epoch": 0.31045562286015277,
      "grad_norm": 2.2968533039093018,
      "learning_rate": 3.4487753489597055e-05,
      "loss": 0.3297,
      "step": 11788
    },
    {
      "epoch": 0.31048195944166446,
      "grad_norm": 2.0658628940582275,
      "learning_rate": 3.4486436660521463e-05,
      "loss": 1.1412,
      "step": 11789
    },
    {
      "epoch": 0.3105082960231762,
      "grad_norm": 2.0782594680786133,
      "learning_rate": 3.448511983144588e-05,
      "loss": 1.4775,
      "step": 11790
    },
    {
      "epoch": 0.3105346326046879,
      "grad_norm": 3.3005943298339844,
      "learning_rate": 3.4483803002370295e-05,
      "loss": 0.987,
      "step": 11791
    },
    {
      "epoch": 0.31056096918619963,
      "grad_norm": 1.907914638519287,
      "learning_rate": 3.448248617329471e-05,
      "loss": 2.0493,
      "step": 11792
    },
    {
      "epoch": 0.3105873057677114,
      "grad_norm": 2.1023778915405273,
      "learning_rate": 3.448116934421912e-05,
      "loss": 2.0876,
      "step": 11793
    },
    {
      "epoch": 0.31061364234922306,
      "grad_norm": 3.929502487182617,
      "learning_rate": 3.4479852515143535e-05,
      "loss": 2.0217,
      "step": 11794
    },
    {
      "epoch": 0.3106399789307348,
      "grad_norm": 1.5751134157180786,
      "learning_rate": 3.447853568606795e-05,
      "loss": 1.2519,
      "step": 11795
    },
    {
      "epoch": 0.3106663155122465,
      "grad_norm": 3.5574424266815186,
      "learning_rate": 3.4477218856992366e-05,
      "loss": 1.86,
      "step": 11796
    },
    {
      "epoch": 0.31069265209375824,
      "grad_norm": 6.201610565185547,
      "learning_rate": 3.447590202791678e-05,
      "loss": 0.9862,
      "step": 11797
    },
    {
      "epoch": 0.3107189886752699,
      "grad_norm": 1.842467188835144,
      "learning_rate": 3.447458519884119e-05,
      "loss": 2.1066,
      "step": 11798
    },
    {
      "epoch": 0.31074532525678167,
      "grad_norm": 1.8609633445739746,
      "learning_rate": 3.4473268369765606e-05,
      "loss": 1.3968,
      "step": 11799
    },
    {
      "epoch": 0.3107716618382934,
      "grad_norm": 3.8730084896087646,
      "learning_rate": 3.447195154069002e-05,
      "loss": 1.5647,
      "step": 11800
    },
    {
      "epoch": 0.3107979984198051,
      "grad_norm": 2.2613658905029297,
      "learning_rate": 3.447063471161444e-05,
      "loss": 1.6678,
      "step": 11801
    },
    {
      "epoch": 0.31082433500131684,
      "grad_norm": 1.7192591428756714,
      "learning_rate": 3.4469317882538846e-05,
      "loss": 1.9186,
      "step": 11802
    },
    {
      "epoch": 0.31085067158282853,
      "grad_norm": 1.6857656240463257,
      "learning_rate": 3.446800105346326e-05,
      "loss": 1.2556,
      "step": 11803
    },
    {
      "epoch": 0.3108770081643403,
      "grad_norm": 1.9221400022506714,
      "learning_rate": 3.446668422438768e-05,
      "loss": 2.2573,
      "step": 11804
    },
    {
      "epoch": 0.31090334474585196,
      "grad_norm": 2.819298505783081,
      "learning_rate": 3.446536739531209e-05,
      "loss": 1.0863,
      "step": 11805
    },
    {
      "epoch": 0.3109296813273637,
      "grad_norm": 2.186988115310669,
      "learning_rate": 3.446405056623651e-05,
      "loss": 2.1734,
      "step": 11806
    },
    {
      "epoch": 0.31095601790887545,
      "grad_norm": 2.2983505725860596,
      "learning_rate": 3.446273373716092e-05,
      "loss": 0.8218,
      "step": 11807
    },
    {
      "epoch": 0.31098235449038714,
      "grad_norm": 2.208716630935669,
      "learning_rate": 3.446141690808533e-05,
      "loss": 0.2124,
      "step": 11808
    },
    {
      "epoch": 0.3110086910718989,
      "grad_norm": 1.9391090869903564,
      "learning_rate": 3.446010007900974e-05,
      "loss": 2.4383,
      "step": 11809
    },
    {
      "epoch": 0.31103502765341057,
      "grad_norm": 2.9998319149017334,
      "learning_rate": 3.4458783249934164e-05,
      "loss": 1.4809,
      "step": 11810
    },
    {
      "epoch": 0.3110613642349223,
      "grad_norm": 2.185720682144165,
      "learning_rate": 3.445746642085857e-05,
      "loss": 1.7887,
      "step": 11811
    },
    {
      "epoch": 0.31108770081643405,
      "grad_norm": 3.1644480228424072,
      "learning_rate": 3.445614959178299e-05,
      "loss": 1.8198,
      "step": 11812
    },
    {
      "epoch": 0.31111403739794574,
      "grad_norm": 1.8772892951965332,
      "learning_rate": 3.4454832762707404e-05,
      "loss": 1.6651,
      "step": 11813
    },
    {
      "epoch": 0.3111403739794575,
      "grad_norm": 1.7570931911468506,
      "learning_rate": 3.445351593363181e-05,
      "loss": 1.8823,
      "step": 11814
    },
    {
      "epoch": 0.3111667105609692,
      "grad_norm": 3.7592968940734863,
      "learning_rate": 3.4452199104556235e-05,
      "loss": 1.3793,
      "step": 11815
    },
    {
      "epoch": 0.3111930471424809,
      "grad_norm": 3.3698315620422363,
      "learning_rate": 3.4450882275480644e-05,
      "loss": 0.8628,
      "step": 11816
    },
    {
      "epoch": 0.3112193837239926,
      "grad_norm": 1.9161858558654785,
      "learning_rate": 3.444956544640506e-05,
      "loss": 1.4254,
      "step": 11817
    },
    {
      "epoch": 0.31124572030550435,
      "grad_norm": 2.1362297534942627,
      "learning_rate": 3.444824861732947e-05,
      "loss": 1.7526,
      "step": 11818
    },
    {
      "epoch": 0.3112720568870161,
      "grad_norm": 2.8416190147399902,
      "learning_rate": 3.444693178825389e-05,
      "loss": 0.9944,
      "step": 11819
    },
    {
      "epoch": 0.3112983934685278,
      "grad_norm": 3.3196158409118652,
      "learning_rate": 3.44456149591783e-05,
      "loss": 1.3767,
      "step": 11820
    },
    {
      "epoch": 0.3113247300500395,
      "grad_norm": 2.473939895629883,
      "learning_rate": 3.4444298130102715e-05,
      "loss": 1.2947,
      "step": 11821
    },
    {
      "epoch": 0.3113510666315512,
      "grad_norm": 3.148097276687622,
      "learning_rate": 3.444298130102713e-05,
      "loss": 1.1591,
      "step": 11822
    },
    {
      "epoch": 0.31137740321306295,
      "grad_norm": 1.6426162719726562,
      "learning_rate": 3.444166447195154e-05,
      "loss": 1.7937,
      "step": 11823
    },
    {
      "epoch": 0.31140373979457464,
      "grad_norm": 1.8858553171157837,
      "learning_rate": 3.444034764287596e-05,
      "loss": 1.8173,
      "step": 11824
    },
    {
      "epoch": 0.3114300763760864,
      "grad_norm": 2.10609769821167,
      "learning_rate": 3.443903081380037e-05,
      "loss": 0.9571,
      "step": 11825
    },
    {
      "epoch": 0.3114564129575981,
      "grad_norm": 2.3406460285186768,
      "learning_rate": 3.4437713984724786e-05,
      "loss": 1.3258,
      "step": 11826
    },
    {
      "epoch": 0.3114827495391098,
      "grad_norm": 1.8892433643341064,
      "learning_rate": 3.4436397155649195e-05,
      "loss": 1.8035,
      "step": 11827
    },
    {
      "epoch": 0.31150908612062156,
      "grad_norm": 5.53434419631958,
      "learning_rate": 3.443508032657361e-05,
      "loss": 1.5868,
      "step": 11828
    },
    {
      "epoch": 0.31153542270213325,
      "grad_norm": 3.3683269023895264,
      "learning_rate": 3.4433763497498026e-05,
      "loss": 1.6449,
      "step": 11829
    },
    {
      "epoch": 0.311561759283645,
      "grad_norm": 2.940948963165283,
      "learning_rate": 3.443244666842244e-05,
      "loss": 0.8559,
      "step": 11830
    },
    {
      "epoch": 0.3115880958651567,
      "grad_norm": 4.314951419830322,
      "learning_rate": 3.443112983934685e-05,
      "loss": 2.3859,
      "step": 11831
    },
    {
      "epoch": 0.3116144324466684,
      "grad_norm": 3.335306167602539,
      "learning_rate": 3.4429813010271266e-05,
      "loss": 1.2916,
      "step": 11832
    },
    {
      "epoch": 0.31164076902818016,
      "grad_norm": 2.280775308609009,
      "learning_rate": 3.442849618119568e-05,
      "loss": 2.2159,
      "step": 11833
    },
    {
      "epoch": 0.31166710560969185,
      "grad_norm": 1.8198018074035645,
      "learning_rate": 3.44271793521201e-05,
      "loss": 1.974,
      "step": 11834
    },
    {
      "epoch": 0.3116934421912036,
      "grad_norm": 3.4338302612304688,
      "learning_rate": 3.442586252304451e-05,
      "loss": 1.9795,
      "step": 11835
    },
    {
      "epoch": 0.3117197787727153,
      "grad_norm": 2.032616138458252,
      "learning_rate": 3.442454569396892e-05,
      "loss": 1.5163,
      "step": 11836
    },
    {
      "epoch": 0.311746115354227,
      "grad_norm": 2.127558946609497,
      "learning_rate": 3.442322886489334e-05,
      "loss": 1.5309,
      "step": 11837
    },
    {
      "epoch": 0.3117724519357387,
      "grad_norm": 4.492937088012695,
      "learning_rate": 3.442191203581775e-05,
      "loss": 1.8677,
      "step": 11838
    },
    {
      "epoch": 0.31179878851725046,
      "grad_norm": 3.4926528930664062,
      "learning_rate": 3.442059520674217e-05,
      "loss": 1.7471,
      "step": 11839
    },
    {
      "epoch": 0.3118251250987622,
      "grad_norm": 2.8821794986724854,
      "learning_rate": 3.441927837766658e-05,
      "loss": 1.5013,
      "step": 11840
    },
    {
      "epoch": 0.3118514616802739,
      "grad_norm": 2.3584909439086914,
      "learning_rate": 3.441796154859099e-05,
      "loss": 1.8278,
      "step": 11841
    },
    {
      "epoch": 0.31187779826178563,
      "grad_norm": 2.28352427482605,
      "learning_rate": 3.441664471951541e-05,
      "loss": 1.4235,
      "step": 11842
    },
    {
      "epoch": 0.3119041348432973,
      "grad_norm": 2.3072593212127686,
      "learning_rate": 3.4415327890439824e-05,
      "loss": 2.0197,
      "step": 11843
    },
    {
      "epoch": 0.31193047142480906,
      "grad_norm": 3.4011669158935547,
      "learning_rate": 3.441401106136424e-05,
      "loss": 1.3139,
      "step": 11844
    },
    {
      "epoch": 0.3119568080063208,
      "grad_norm": 3.5458455085754395,
      "learning_rate": 3.441269423228865e-05,
      "loss": 1.8922,
      "step": 11845
    },
    {
      "epoch": 0.3119831445878325,
      "grad_norm": 2.366835355758667,
      "learning_rate": 3.4411377403213064e-05,
      "loss": 1.7695,
      "step": 11846
    },
    {
      "epoch": 0.31200948116934424,
      "grad_norm": 2.1246747970581055,
      "learning_rate": 3.441006057413747e-05,
      "loss": 1.7576,
      "step": 11847
    },
    {
      "epoch": 0.3120358177508559,
      "grad_norm": 5.915126323699951,
      "learning_rate": 3.4408743745061896e-05,
      "loss": 2.3039,
      "step": 11848
    },
    {
      "epoch": 0.31206215433236767,
      "grad_norm": 2.065683364868164,
      "learning_rate": 3.4407426915986304e-05,
      "loss": 1.4285,
      "step": 11849
    },
    {
      "epoch": 0.31208849091387936,
      "grad_norm": 2.0415496826171875,
      "learning_rate": 3.440611008691072e-05,
      "loss": 1.6201,
      "step": 11850
    },
    {
      "epoch": 0.3121148274953911,
      "grad_norm": 3.2185983657836914,
      "learning_rate": 3.4404793257835136e-05,
      "loss": 0.7069,
      "step": 11851
    },
    {
      "epoch": 0.31214116407690284,
      "grad_norm": 2.1158273220062256,
      "learning_rate": 3.440347642875955e-05,
      "loss": 1.7578,
      "step": 11852
    },
    {
      "epoch": 0.31216750065841453,
      "grad_norm": 1.7559586763381958,
      "learning_rate": 3.440215959968397e-05,
      "loss": 1.6289,
      "step": 11853
    },
    {
      "epoch": 0.3121938372399263,
      "grad_norm": 3.821509838104248,
      "learning_rate": 3.4400842770608376e-05,
      "loss": 1.9091,
      "step": 11854
    },
    {
      "epoch": 0.31222017382143796,
      "grad_norm": 5.022719860076904,
      "learning_rate": 3.439952594153279e-05,
      "loss": 1.1796,
      "step": 11855
    },
    {
      "epoch": 0.3122465104029497,
      "grad_norm": 2.2521557807922363,
      "learning_rate": 3.43982091124572e-05,
      "loss": 1.6953,
      "step": 11856
    },
    {
      "epoch": 0.3122728469844614,
      "grad_norm": 3.792581081390381,
      "learning_rate": 3.439689228338162e-05,
      "loss": 1.7778,
      "step": 11857
    },
    {
      "epoch": 0.31229918356597314,
      "grad_norm": 1.7502951622009277,
      "learning_rate": 3.439557545430603e-05,
      "loss": 1.9176,
      "step": 11858
    },
    {
      "epoch": 0.3123255201474849,
      "grad_norm": 1.73971426486969,
      "learning_rate": 3.439425862523045e-05,
      "loss": 1.8883,
      "step": 11859
    },
    {
      "epoch": 0.31235185672899657,
      "grad_norm": 2.5772743225097656,
      "learning_rate": 3.439294179615486e-05,
      "loss": 1.3508,
      "step": 11860
    },
    {
      "epoch": 0.3123781933105083,
      "grad_norm": 2.179870128631592,
      "learning_rate": 3.439162496707927e-05,
      "loss": 1.5847,
      "step": 11861
    },
    {
      "epoch": 0.31240452989202,
      "grad_norm": 2.0155560970306396,
      "learning_rate": 3.4390308138003694e-05,
      "loss": 1.4594,
      "step": 11862
    },
    {
      "epoch": 0.31243086647353174,
      "grad_norm": 3.4546289443969727,
      "learning_rate": 3.43889913089281e-05,
      "loss": 2.5174,
      "step": 11863
    },
    {
      "epoch": 0.31245720305504343,
      "grad_norm": 1.9563692808151245,
      "learning_rate": 3.438767447985252e-05,
      "loss": 1.3438,
      "step": 11864
    },
    {
      "epoch": 0.3124835396365552,
      "grad_norm": 1.6062841415405273,
      "learning_rate": 3.438635765077693e-05,
      "loss": 1.6144,
      "step": 11865
    },
    {
      "epoch": 0.3125098762180669,
      "grad_norm": 1.5504190921783447,
      "learning_rate": 3.438504082170135e-05,
      "loss": 1.5992,
      "step": 11866
    },
    {
      "epoch": 0.3125362127995786,
      "grad_norm": 2.0044147968292236,
      "learning_rate": 3.438372399262576e-05,
      "loss": 1.8003,
      "step": 11867
    },
    {
      "epoch": 0.31256254938109035,
      "grad_norm": 2.6076581478118896,
      "learning_rate": 3.4382407163550174e-05,
      "loss": 2.0047,
      "step": 11868
    },
    {
      "epoch": 0.31258888596260204,
      "grad_norm": 1.7118685245513916,
      "learning_rate": 3.438109033447459e-05,
      "loss": 1.7316,
      "step": 11869
    },
    {
      "epoch": 0.3126152225441138,
      "grad_norm": 2.617780923843384,
      "learning_rate": 3.4379773505399e-05,
      "loss": 1.9247,
      "step": 11870
    },
    {
      "epoch": 0.31264155912562547,
      "grad_norm": 1.7714011669158936,
      "learning_rate": 3.437845667632342e-05,
      "loss": 2.5514,
      "step": 11871
    },
    {
      "epoch": 0.3126678957071372,
      "grad_norm": 1.7230901718139648,
      "learning_rate": 3.437713984724783e-05,
      "loss": 2.4104,
      "step": 11872
    },
    {
      "epoch": 0.31269423228864895,
      "grad_norm": 2.9638564586639404,
      "learning_rate": 3.4375823018172245e-05,
      "loss": 2.4147,
      "step": 11873
    },
    {
      "epoch": 0.31272056887016064,
      "grad_norm": 4.746386528015137,
      "learning_rate": 3.4374506189096654e-05,
      "loss": 2.0206,
      "step": 11874
    },
    {
      "epoch": 0.3127469054516724,
      "grad_norm": 1.94407057762146,
      "learning_rate": 3.437318936002107e-05,
      "loss": 2.5493,
      "step": 11875
    },
    {
      "epoch": 0.3127732420331841,
      "grad_norm": 1.7084060907363892,
      "learning_rate": 3.4371872530945485e-05,
      "loss": 1.3561,
      "step": 11876
    },
    {
      "epoch": 0.3127995786146958,
      "grad_norm": 2.9552388191223145,
      "learning_rate": 3.43705557018699e-05,
      "loss": 1.8076,
      "step": 11877
    },
    {
      "epoch": 0.3128259151962075,
      "grad_norm": 5.004602432250977,
      "learning_rate": 3.436923887279431e-05,
      "loss": 0.7048,
      "step": 11878
    },
    {
      "epoch": 0.31285225177771925,
      "grad_norm": 4.178142070770264,
      "learning_rate": 3.4367922043718725e-05,
      "loss": 0.4902,
      "step": 11879
    },
    {
      "epoch": 0.312878588359231,
      "grad_norm": 3.9015309810638428,
      "learning_rate": 3.436660521464314e-05,
      "loss": 0.5962,
      "step": 11880
    },
    {
      "epoch": 0.3129049249407427,
      "grad_norm": 2.054222345352173,
      "learning_rate": 3.4365288385567556e-05,
      "loss": 1.7808,
      "step": 11881
    },
    {
      "epoch": 0.3129312615222544,
      "grad_norm": 4.6398725509643555,
      "learning_rate": 3.436397155649197e-05,
      "loss": 2.1166,
      "step": 11882
    },
    {
      "epoch": 0.3129575981037661,
      "grad_norm": 2.2833259105682373,
      "learning_rate": 3.436265472741638e-05,
      "loss": 2.2202,
      "step": 11883
    },
    {
      "epoch": 0.31298393468527785,
      "grad_norm": 1.8960294723510742,
      "learning_rate": 3.4361337898340796e-05,
      "loss": 1.922,
      "step": 11884
    },
    {
      "epoch": 0.3130102712667896,
      "grad_norm": 2.1866676807403564,
      "learning_rate": 3.436002106926521e-05,
      "loss": 1.8204,
      "step": 11885
    },
    {
      "epoch": 0.3130366078483013,
      "grad_norm": 3.9158682823181152,
      "learning_rate": 3.435870424018963e-05,
      "loss": 0.6397,
      "step": 11886
    },
    {
      "epoch": 0.31306294442981303,
      "grad_norm": 1.6878708600997925,
      "learning_rate": 3.4357387411114036e-05,
      "loss": 1.9255,
      "step": 11887
    },
    {
      "epoch": 0.3130892810113247,
      "grad_norm": 1.8021827936172485,
      "learning_rate": 3.435607058203845e-05,
      "loss": 2.3026,
      "step": 11888
    },
    {
      "epoch": 0.31311561759283646,
      "grad_norm": 2.618227243423462,
      "learning_rate": 3.435475375296287e-05,
      "loss": 1.9953,
      "step": 11889
    },
    {
      "epoch": 0.31314195417434815,
      "grad_norm": 3.2791929244995117,
      "learning_rate": 3.435343692388728e-05,
      "loss": 1.9422,
      "step": 11890
    },
    {
      "epoch": 0.3131682907558599,
      "grad_norm": 2.137345790863037,
      "learning_rate": 3.43521200948117e-05,
      "loss": 2.328,
      "step": 11891
    },
    {
      "epoch": 0.31319462733737163,
      "grad_norm": 3.2285099029541016,
      "learning_rate": 3.435080326573611e-05,
      "loss": 1.289,
      "step": 11892
    },
    {
      "epoch": 0.3132209639188833,
      "grad_norm": 1.742781639099121,
      "learning_rate": 3.434948643666052e-05,
      "loss": 2.2633,
      "step": 11893
    },
    {
      "epoch": 0.31324730050039507,
      "grad_norm": 1.6923943758010864,
      "learning_rate": 3.434816960758493e-05,
      "loss": 2.1714,
      "step": 11894
    },
    {
      "epoch": 0.31327363708190675,
      "grad_norm": 3.1474180221557617,
      "learning_rate": 3.4346852778509354e-05,
      "loss": 1.8829,
      "step": 11895
    },
    {
      "epoch": 0.3132999736634185,
      "grad_norm": 7.885355472564697,
      "learning_rate": 3.434553594943376e-05,
      "loss": 2.1793,
      "step": 11896
    },
    {
      "epoch": 0.3133263102449302,
      "grad_norm": 3.4797866344451904,
      "learning_rate": 3.434421912035818e-05,
      "loss": 0.8664,
      "step": 11897
    },
    {
      "epoch": 0.3133526468264419,
      "grad_norm": 2.1376543045043945,
      "learning_rate": 3.4342902291282594e-05,
      "loss": 1.5543,
      "step": 11898
    },
    {
      "epoch": 0.31337898340795367,
      "grad_norm": 1.9626778364181519,
      "learning_rate": 3.434158546220701e-05,
      "loss": 1.6245,
      "step": 11899
    },
    {
      "epoch": 0.31340531998946536,
      "grad_norm": 1.8312442302703857,
      "learning_rate": 3.4340268633131425e-05,
      "loss": 1.9471,
      "step": 11900
    },
    {
      "epoch": 0.3134316565709771,
      "grad_norm": 1.6652500629425049,
      "learning_rate": 3.4338951804055834e-05,
      "loss": 2.1866,
      "step": 11901
    },
    {
      "epoch": 0.3134579931524888,
      "grad_norm": 2.36041259765625,
      "learning_rate": 3.433763497498025e-05,
      "loss": 1.9075,
      "step": 11902
    },
    {
      "epoch": 0.31348432973400053,
      "grad_norm": 2.087564706802368,
      "learning_rate": 3.433631814590466e-05,
      "loss": 1.5581,
      "step": 11903
    },
    {
      "epoch": 0.3135106663155122,
      "grad_norm": 2.3994157314300537,
      "learning_rate": 3.433500131682908e-05,
      "loss": 2.1008,
      "step": 11904
    },
    {
      "epoch": 0.31353700289702396,
      "grad_norm": 2.7171549797058105,
      "learning_rate": 3.433368448775349e-05,
      "loss": 2.3056,
      "step": 11905
    },
    {
      "epoch": 0.3135633394785357,
      "grad_norm": 3.2091329097747803,
      "learning_rate": 3.4332367658677905e-05,
      "loss": 1.7718,
      "step": 11906
    },
    {
      "epoch": 0.3135896760600474,
      "grad_norm": 2.0868232250213623,
      "learning_rate": 3.433105082960232e-05,
      "loss": 1.5704,
      "step": 11907
    },
    {
      "epoch": 0.31361601264155914,
      "grad_norm": 1.8223464488983154,
      "learning_rate": 3.432973400052673e-05,
      "loss": 1.6108,
      "step": 11908
    },
    {
      "epoch": 0.3136423492230708,
      "grad_norm": 1.64521324634552,
      "learning_rate": 3.432841717145115e-05,
      "loss": 1.6907,
      "step": 11909
    },
    {
      "epoch": 0.31366868580458257,
      "grad_norm": 3.0415732860565186,
      "learning_rate": 3.432710034237556e-05,
      "loss": 0.929,
      "step": 11910
    },
    {
      "epoch": 0.31369502238609426,
      "grad_norm": 4.400512218475342,
      "learning_rate": 3.4325783513299977e-05,
      "loss": 1.7874,
      "step": 11911
    },
    {
      "epoch": 0.313721358967606,
      "grad_norm": 1.7861953973770142,
      "learning_rate": 3.4324466684224385e-05,
      "loss": 2.0303,
      "step": 11912
    },
    {
      "epoch": 0.31374769554911774,
      "grad_norm": 1.600196361541748,
      "learning_rate": 3.43231498551488e-05,
      "loss": 1.6419,
      "step": 11913
    },
    {
      "epoch": 0.31377403213062943,
      "grad_norm": 2.017338991165161,
      "learning_rate": 3.4321833026073217e-05,
      "loss": 2.0561,
      "step": 11914
    },
    {
      "epoch": 0.3138003687121412,
      "grad_norm": 1.924303412437439,
      "learning_rate": 3.432051619699763e-05,
      "loss": 2.0066,
      "step": 11915
    },
    {
      "epoch": 0.31382670529365286,
      "grad_norm": 1.8211301565170288,
      "learning_rate": 3.431919936792205e-05,
      "loss": 0.4485,
      "step": 11916
    },
    {
      "epoch": 0.3138530418751646,
      "grad_norm": 2.6942644119262695,
      "learning_rate": 3.431788253884646e-05,
      "loss": 1.2617,
      "step": 11917
    },
    {
      "epoch": 0.31387937845667635,
      "grad_norm": 2.10634708404541,
      "learning_rate": 3.431656570977088e-05,
      "loss": 1.9003,
      "step": 11918
    },
    {
      "epoch": 0.31390571503818804,
      "grad_norm": 2.9917962551116943,
      "learning_rate": 3.431524888069529e-05,
      "loss": 0.3855,
      "step": 11919
    },
    {
      "epoch": 0.3139320516196998,
      "grad_norm": 2.0381088256835938,
      "learning_rate": 3.4313932051619703e-05,
      "loss": 1.8381,
      "step": 11920
    },
    {
      "epoch": 0.31395838820121147,
      "grad_norm": 2.292893648147583,
      "learning_rate": 3.431261522254411e-05,
      "loss": 2.196,
      "step": 11921
    },
    {
      "epoch": 0.3139847247827232,
      "grad_norm": 3.7332489490509033,
      "learning_rate": 3.431129839346853e-05,
      "loss": 1.8974,
      "step": 11922
    },
    {
      "epoch": 0.3140110613642349,
      "grad_norm": 1.586669921875,
      "learning_rate": 3.4309981564392943e-05,
      "loss": 1.6117,
      "step": 11923
    },
    {
      "epoch": 0.31403739794574664,
      "grad_norm": 3.8536012172698975,
      "learning_rate": 3.430866473531736e-05,
      "loss": 1.6309,
      "step": 11924
    },
    {
      "epoch": 0.3140637345272584,
      "grad_norm": 2.281076669692993,
      "learning_rate": 3.4307347906241775e-05,
      "loss": 1.7698,
      "step": 11925
    },
    {
      "epoch": 0.3140900711087701,
      "grad_norm": 3.24711537361145,
      "learning_rate": 3.4306031077166183e-05,
      "loss": 2.044,
      "step": 11926
    },
    {
      "epoch": 0.3141164076902818,
      "grad_norm": 2.8330602645874023,
      "learning_rate": 3.43047142480906e-05,
      "loss": 2.0034,
      "step": 11927
    },
    {
      "epoch": 0.3141427442717935,
      "grad_norm": 2.4326963424682617,
      "learning_rate": 3.4303397419015015e-05,
      "loss": 0.4674,
      "step": 11928
    },
    {
      "epoch": 0.31416908085330525,
      "grad_norm": 1.8757840394973755,
      "learning_rate": 3.430208058993943e-05,
      "loss": 1.8203,
      "step": 11929
    },
    {
      "epoch": 0.31419541743481694,
      "grad_norm": 3.904343605041504,
      "learning_rate": 3.430076376086384e-05,
      "loss": 1.2149,
      "step": 11930
    },
    {
      "epoch": 0.3142217540163287,
      "grad_norm": 2.2272608280181885,
      "learning_rate": 3.4299446931788255e-05,
      "loss": 1.298,
      "step": 11931
    },
    {
      "epoch": 0.3142480905978404,
      "grad_norm": 2.268555164337158,
      "learning_rate": 3.429813010271267e-05,
      "loss": 2.1082,
      "step": 11932
    },
    {
      "epoch": 0.3142744271793521,
      "grad_norm": 2.5162017345428467,
      "learning_rate": 3.4296813273637086e-05,
      "loss": 0.5128,
      "step": 11933
    },
    {
      "epoch": 0.31430076376086385,
      "grad_norm": 1.671085000038147,
      "learning_rate": 3.4295496444561495e-05,
      "loss": 2.1305,
      "step": 11934
    },
    {
      "epoch": 0.31432710034237554,
      "grad_norm": 2.587003231048584,
      "learning_rate": 3.429417961548591e-05,
      "loss": 2.2135,
      "step": 11935
    },
    {
      "epoch": 0.3143534369238873,
      "grad_norm": 3.799715757369995,
      "learning_rate": 3.4292862786410326e-05,
      "loss": 1.9254,
      "step": 11936
    },
    {
      "epoch": 0.314379773505399,
      "grad_norm": 2.330026388168335,
      "learning_rate": 3.429154595733474e-05,
      "loss": 2.338,
      "step": 11937
    },
    {
      "epoch": 0.3144061100869107,
      "grad_norm": 2.73518705368042,
      "learning_rate": 3.429022912825916e-05,
      "loss": 1.9821,
      "step": 11938
    },
    {
      "epoch": 0.31443244666842246,
      "grad_norm": 4.204094409942627,
      "learning_rate": 3.4288912299183566e-05,
      "loss": 1.0299,
      "step": 11939
    },
    {
      "epoch": 0.31445878324993415,
      "grad_norm": 2.7655587196350098,
      "learning_rate": 3.428759547010798e-05,
      "loss": 1.3402,
      "step": 11940
    },
    {
      "epoch": 0.3144851198314459,
      "grad_norm": 1.9417668581008911,
      "learning_rate": 3.428627864103239e-05,
      "loss": 1.8288,
      "step": 11941
    },
    {
      "epoch": 0.3145114564129576,
      "grad_norm": 1.6638494729995728,
      "learning_rate": 3.428496181195681e-05,
      "loss": 1.2098,
      "step": 11942
    },
    {
      "epoch": 0.3145377929944693,
      "grad_norm": 4.269766330718994,
      "learning_rate": 3.428364498288122e-05,
      "loss": 1.5998,
      "step": 11943
    },
    {
      "epoch": 0.314564129575981,
      "grad_norm": 1.4448368549346924,
      "learning_rate": 3.428232815380564e-05,
      "loss": 1.6042,
      "step": 11944
    },
    {
      "epoch": 0.31459046615749275,
      "grad_norm": 2.146402597427368,
      "learning_rate": 3.428101132473005e-05,
      "loss": 1.9682,
      "step": 11945
    },
    {
      "epoch": 0.3146168027390045,
      "grad_norm": 3.751169443130493,
      "learning_rate": 3.427969449565446e-05,
      "loss": 1.2187,
      "step": 11946
    },
    {
      "epoch": 0.3146431393205162,
      "grad_norm": 3.6288695335388184,
      "learning_rate": 3.4278377666578884e-05,
      "loss": 1.1246,
      "step": 11947
    },
    {
      "epoch": 0.31466947590202793,
      "grad_norm": 2.1228652000427246,
      "learning_rate": 3.427706083750329e-05,
      "loss": 2.1966,
      "step": 11948
    },
    {
      "epoch": 0.3146958124835396,
      "grad_norm": 2.092362642288208,
      "learning_rate": 3.427574400842771e-05,
      "loss": 1.7767,
      "step": 11949
    },
    {
      "epoch": 0.31472214906505136,
      "grad_norm": 4.747443199157715,
      "learning_rate": 3.427442717935212e-05,
      "loss": 1.3056,
      "step": 11950
    },
    {
      "epoch": 0.3147484856465631,
      "grad_norm": 2.3654873371124268,
      "learning_rate": 3.427311035027654e-05,
      "loss": 1.5675,
      "step": 11951
    },
    {
      "epoch": 0.3147748222280748,
      "grad_norm": 1.8938640356063843,
      "learning_rate": 3.427179352120095e-05,
      "loss": 1.919,
      "step": 11952
    },
    {
      "epoch": 0.31480115880958653,
      "grad_norm": 1.851574420928955,
      "learning_rate": 3.4270476692125364e-05,
      "loss": 2.2255,
      "step": 11953
    },
    {
      "epoch": 0.3148274953910982,
      "grad_norm": 2.2016735076904297,
      "learning_rate": 3.426915986304978e-05,
      "loss": 1.7719,
      "step": 11954
    },
    {
      "epoch": 0.31485383197260997,
      "grad_norm": 2.4434311389923096,
      "learning_rate": 3.426784303397419e-05,
      "loss": 0.8764,
      "step": 11955
    },
    {
      "epoch": 0.31488016855412165,
      "grad_norm": 2.07375431060791,
      "learning_rate": 3.426652620489861e-05,
      "loss": 1.7581,
      "step": 11956
    },
    {
      "epoch": 0.3149065051356334,
      "grad_norm": 3.103745698928833,
      "learning_rate": 3.426520937582302e-05,
      "loss": 1.5745,
      "step": 11957
    },
    {
      "epoch": 0.31493284171714514,
      "grad_norm": 4.127291202545166,
      "learning_rate": 3.4263892546747435e-05,
      "loss": 1.4292,
      "step": 11958
    },
    {
      "epoch": 0.31495917829865683,
      "grad_norm": 2.5967373847961426,
      "learning_rate": 3.4262575717671844e-05,
      "loss": 1.8639,
      "step": 11959
    },
    {
      "epoch": 0.31498551488016857,
      "grad_norm": 3.1381006240844727,
      "learning_rate": 3.426125888859626e-05,
      "loss": 1.4107,
      "step": 11960
    },
    {
      "epoch": 0.31501185146168026,
      "grad_norm": 3.3036344051361084,
      "learning_rate": 3.4259942059520675e-05,
      "loss": 0.7293,
      "step": 11961
    },
    {
      "epoch": 0.315038188043192,
      "grad_norm": 1.9857488870620728,
      "learning_rate": 3.425862523044509e-05,
      "loss": 0.3787,
      "step": 11962
    },
    {
      "epoch": 0.3150645246247037,
      "grad_norm": 3.5623269081115723,
      "learning_rate": 3.4257308401369506e-05,
      "loss": 1.3363,
      "step": 11963
    },
    {
      "epoch": 0.31509086120621543,
      "grad_norm": 1.5893114805221558,
      "learning_rate": 3.4255991572293915e-05,
      "loss": 1.3695,
      "step": 11964
    },
    {
      "epoch": 0.3151171977877272,
      "grad_norm": 1.6632758378982544,
      "learning_rate": 3.425467474321834e-05,
      "loss": 1.9521,
      "step": 11965
    },
    {
      "epoch": 0.31514353436923886,
      "grad_norm": 1.9022972583770752,
      "learning_rate": 3.4253357914142746e-05,
      "loss": 2.2308,
      "step": 11966
    },
    {
      "epoch": 0.3151698709507506,
      "grad_norm": 1.949738621711731,
      "learning_rate": 3.425204108506716e-05,
      "loss": 2.175,
      "step": 11967
    },
    {
      "epoch": 0.3151962075322623,
      "grad_norm": 1.9077039957046509,
      "learning_rate": 3.425072425599157e-05,
      "loss": 1.5597,
      "step": 11968
    },
    {
      "epoch": 0.31522254411377404,
      "grad_norm": 1.7058870792388916,
      "learning_rate": 3.4249407426915986e-05,
      "loss": 1.5192,
      "step": 11969
    },
    {
      "epoch": 0.3152488806952857,
      "grad_norm": 2.6986324787139893,
      "learning_rate": 3.42480905978404e-05,
      "loss": 0.3859,
      "step": 11970
    },
    {
      "epoch": 0.31527521727679747,
      "grad_norm": 1.5514413118362427,
      "learning_rate": 3.424677376876482e-05,
      "loss": 1.5685,
      "step": 11971
    },
    {
      "epoch": 0.3153015538583092,
      "grad_norm": 2.907836437225342,
      "learning_rate": 3.424545693968923e-05,
      "loss": 2.2402,
      "step": 11972
    },
    {
      "epoch": 0.3153278904398209,
      "grad_norm": 2.425527572631836,
      "learning_rate": 3.424414011061364e-05,
      "loss": 1.0907,
      "step": 11973
    },
    {
      "epoch": 0.31535422702133264,
      "grad_norm": 1.9716839790344238,
      "learning_rate": 3.424282328153806e-05,
      "loss": 0.5059,
      "step": 11974
    },
    {
      "epoch": 0.31538056360284433,
      "grad_norm": 2.9178848266601562,
      "learning_rate": 3.424150645246247e-05,
      "loss": 0.5166,
      "step": 11975
    },
    {
      "epoch": 0.3154069001843561,
      "grad_norm": 1.8250172138214111,
      "learning_rate": 3.424018962338689e-05,
      "loss": 1.7656,
      "step": 11976
    },
    {
      "epoch": 0.31543323676586776,
      "grad_norm": 4.121735095977783,
      "learning_rate": 3.42388727943113e-05,
      "loss": 1.3751,
      "step": 11977
    },
    {
      "epoch": 0.3154595733473795,
      "grad_norm": 2.6907474994659424,
      "learning_rate": 3.423755596523571e-05,
      "loss": 1.9716,
      "step": 11978
    },
    {
      "epoch": 0.31548590992889125,
      "grad_norm": 3.6392195224761963,
      "learning_rate": 3.423623913616012e-05,
      "loss": 2.0938,
      "step": 11979
    },
    {
      "epoch": 0.31551224651040294,
      "grad_norm": 2.5202419757843018,
      "learning_rate": 3.4234922307084544e-05,
      "loss": 1.3766,
      "step": 11980
    },
    {
      "epoch": 0.3155385830919147,
      "grad_norm": 5.536961078643799,
      "learning_rate": 3.423360547800895e-05,
      "loss": 1.0434,
      "step": 11981
    },
    {
      "epoch": 0.31556491967342637,
      "grad_norm": 2.9225029945373535,
      "learning_rate": 3.423228864893337e-05,
      "loss": 0.6428,
      "step": 11982
    },
    {
      "epoch": 0.3155912562549381,
      "grad_norm": 1.761869192123413,
      "learning_rate": 3.4230971819857784e-05,
      "loss": 2.2562,
      "step": 11983
    },
    {
      "epoch": 0.31561759283644986,
      "grad_norm": 1.7251423597335815,
      "learning_rate": 3.42296549907822e-05,
      "loss": 1.9982,
      "step": 11984
    },
    {
      "epoch": 0.31564392941796154,
      "grad_norm": 1.9662408828735352,
      "learning_rate": 3.4228338161706616e-05,
      "loss": 1.976,
      "step": 11985
    },
    {
      "epoch": 0.3156702659994733,
      "grad_norm": 2.432746171951294,
      "learning_rate": 3.4227021332631024e-05,
      "loss": 1.0983,
      "step": 11986
    },
    {
      "epoch": 0.315696602580985,
      "grad_norm": 4.456958293914795,
      "learning_rate": 3.422570450355544e-05,
      "loss": 1.2475,
      "step": 11987
    },
    {
      "epoch": 0.3157229391624967,
      "grad_norm": 2.2281575202941895,
      "learning_rate": 3.422438767447985e-05,
      "loss": 1.646,
      "step": 11988
    },
    {
      "epoch": 0.3157492757440084,
      "grad_norm": 3.2517008781433105,
      "learning_rate": 3.422307084540427e-05,
      "loss": 0.9704,
      "step": 11989
    },
    {
      "epoch": 0.31577561232552015,
      "grad_norm": 3.9694466590881348,
      "learning_rate": 3.422175401632868e-05,
      "loss": 2.6364,
      "step": 11990
    },
    {
      "epoch": 0.3158019489070319,
      "grad_norm": 2.0546627044677734,
      "learning_rate": 3.4220437187253096e-05,
      "loss": 1.3628,
      "step": 11991
    },
    {
      "epoch": 0.3158282854885436,
      "grad_norm": 2.669036388397217,
      "learning_rate": 3.421912035817751e-05,
      "loss": 1.8031,
      "step": 11992
    },
    {
      "epoch": 0.3158546220700553,
      "grad_norm": 1.6669118404388428,
      "learning_rate": 3.421780352910192e-05,
      "loss": 2.0616,
      "step": 11993
    },
    {
      "epoch": 0.315880958651567,
      "grad_norm": 2.085125207901001,
      "learning_rate": 3.421648670002634e-05,
      "loss": 1.5328,
      "step": 11994
    },
    {
      "epoch": 0.31590729523307876,
      "grad_norm": 3.967073917388916,
      "learning_rate": 3.421516987095075e-05,
      "loss": 1.2698,
      "step": 11995
    },
    {
      "epoch": 0.31593363181459044,
      "grad_norm": 3.3713738918304443,
      "learning_rate": 3.421385304187517e-05,
      "loss": 1.4386,
      "step": 11996
    },
    {
      "epoch": 0.3159599683961022,
      "grad_norm": 2.2872700691223145,
      "learning_rate": 3.4212536212799576e-05,
      "loss": 1.8076,
      "step": 11997
    },
    {
      "epoch": 0.31598630497761393,
      "grad_norm": 3.812849283218384,
      "learning_rate": 3.4211219383724e-05,
      "loss": 1.7878,
      "step": 11998
    },
    {
      "epoch": 0.3160126415591256,
      "grad_norm": 1.8914098739624023,
      "learning_rate": 3.420990255464841e-05,
      "loss": 1.7793,
      "step": 11999
    },
    {
      "epoch": 0.31603897814063736,
      "grad_norm": 2.5909225940704346,
      "learning_rate": 3.420858572557282e-05,
      "loss": 1.3282,
      "step": 12000
    },
    {
      "epoch": 0.31606531472214905,
      "grad_norm": 2.4622552394866943,
      "learning_rate": 3.420726889649724e-05,
      "loss": 1.7412,
      "step": 12001
    },
    {
      "epoch": 0.3160916513036608,
      "grad_norm": 2.063779354095459,
      "learning_rate": 3.420595206742165e-05,
      "loss": 1.5135,
      "step": 12002
    },
    {
      "epoch": 0.3161179878851725,
      "grad_norm": 2.2328691482543945,
      "learning_rate": 3.420463523834607e-05,
      "loss": 0.8068,
      "step": 12003
    },
    {
      "epoch": 0.3161443244666842,
      "grad_norm": 3.361863613128662,
      "learning_rate": 3.420331840927048e-05,
      "loss": 0.6206,
      "step": 12004
    },
    {
      "epoch": 0.31617066104819597,
      "grad_norm": 3.3704581260681152,
      "learning_rate": 3.4202001580194894e-05,
      "loss": 1.3984,
      "step": 12005
    },
    {
      "epoch": 0.31619699762970765,
      "grad_norm": 2.329801082611084,
      "learning_rate": 3.42006847511193e-05,
      "loss": 0.8442,
      "step": 12006
    },
    {
      "epoch": 0.3162233342112194,
      "grad_norm": 1.9171693325042725,
      "learning_rate": 3.419936792204372e-05,
      "loss": 1.5431,
      "step": 12007
    },
    {
      "epoch": 0.3162496707927311,
      "grad_norm": 1.9286022186279297,
      "learning_rate": 3.4198051092968134e-05,
      "loss": 1.7693,
      "step": 12008
    },
    {
      "epoch": 0.31627600737424283,
      "grad_norm": 3.3286213874816895,
      "learning_rate": 3.419673426389255e-05,
      "loss": 0.8107,
      "step": 12009
    },
    {
      "epoch": 0.3163023439557545,
      "grad_norm": 1.7902685403823853,
      "learning_rate": 3.4195417434816965e-05,
      "loss": 1.6694,
      "step": 12010
    },
    {
      "epoch": 0.31632868053726626,
      "grad_norm": 3.36136794090271,
      "learning_rate": 3.4194100605741374e-05,
      "loss": 1.6546,
      "step": 12011
    },
    {
      "epoch": 0.316355017118778,
      "grad_norm": 1.8728545904159546,
      "learning_rate": 3.419278377666579e-05,
      "loss": 1.9547,
      "step": 12012
    },
    {
      "epoch": 0.3163813537002897,
      "grad_norm": 2.6027796268463135,
      "learning_rate": 3.4191466947590205e-05,
      "loss": 1.3122,
      "step": 12013
    },
    {
      "epoch": 0.31640769028180143,
      "grad_norm": 1.5420255661010742,
      "learning_rate": 3.419015011851462e-05,
      "loss": 2.2219,
      "step": 12014
    },
    {
      "epoch": 0.3164340268633131,
      "grad_norm": 1.9534305334091187,
      "learning_rate": 3.418883328943903e-05,
      "loss": 1.7044,
      "step": 12015
    },
    {
      "epoch": 0.31646036344482487,
      "grad_norm": 4.399682521820068,
      "learning_rate": 3.4187516460363445e-05,
      "loss": 1.5365,
      "step": 12016
    },
    {
      "epoch": 0.3164867000263366,
      "grad_norm": 2.8296594619750977,
      "learning_rate": 3.418619963128786e-05,
      "loss": 0.8142,
      "step": 12017
    },
    {
      "epoch": 0.3165130366078483,
      "grad_norm": 1.6229559183120728,
      "learning_rate": 3.4184882802212276e-05,
      "loss": 1.8636,
      "step": 12018
    },
    {
      "epoch": 0.31653937318936004,
      "grad_norm": 2.1545701026916504,
      "learning_rate": 3.418356597313669e-05,
      "loss": 1.4344,
      "step": 12019
    },
    {
      "epoch": 0.31656570977087173,
      "grad_norm": 1.4987447261810303,
      "learning_rate": 3.41822491440611e-05,
      "loss": 0.4292,
      "step": 12020
    },
    {
      "epoch": 0.31659204635238347,
      "grad_norm": 1.6686434745788574,
      "learning_rate": 3.4180932314985516e-05,
      "loss": 1.278,
      "step": 12021
    },
    {
      "epoch": 0.31661838293389516,
      "grad_norm": 2.217627763748169,
      "learning_rate": 3.417961548590993e-05,
      "loss": 2.4288,
      "step": 12022
    },
    {
      "epoch": 0.3166447195154069,
      "grad_norm": 1.6432602405548096,
      "learning_rate": 3.417829865683435e-05,
      "loss": 1.5443,
      "step": 12023
    },
    {
      "epoch": 0.31667105609691865,
      "grad_norm": 2.990502119064331,
      "learning_rate": 3.4176981827758756e-05,
      "loss": 1.5581,
      "step": 12024
    },
    {
      "epoch": 0.31669739267843033,
      "grad_norm": 2.026674509048462,
      "learning_rate": 3.417566499868317e-05,
      "loss": 1.8248,
      "step": 12025
    },
    {
      "epoch": 0.3167237292599421,
      "grad_norm": 1.969262957572937,
      "learning_rate": 3.417434816960759e-05,
      "loss": 1.7076,
      "step": 12026
    },
    {
      "epoch": 0.31675006584145377,
      "grad_norm": 2.062495470046997,
      "learning_rate": 3.4173031340532e-05,
      "loss": 2.3502,
      "step": 12027
    },
    {
      "epoch": 0.3167764024229655,
      "grad_norm": 1.755882740020752,
      "learning_rate": 3.417171451145642e-05,
      "loss": 1.6885,
      "step": 12028
    },
    {
      "epoch": 0.3168027390044772,
      "grad_norm": 3.3108365535736084,
      "learning_rate": 3.417039768238083e-05,
      "loss": 0.4137,
      "step": 12029
    },
    {
      "epoch": 0.31682907558598894,
      "grad_norm": 6.628294944763184,
      "learning_rate": 3.416908085330524e-05,
      "loss": 1.0136,
      "step": 12030
    },
    {
      "epoch": 0.3168554121675007,
      "grad_norm": 1.9180766344070435,
      "learning_rate": 3.416776402422966e-05,
      "loss": 1.8115,
      "step": 12031
    },
    {
      "epoch": 0.31688174874901237,
      "grad_norm": 3.267291307449341,
      "learning_rate": 3.4166447195154074e-05,
      "loss": 1.4986,
      "step": 12032
    },
    {
      "epoch": 0.3169080853305241,
      "grad_norm": 2.7749853134155273,
      "learning_rate": 3.416513036607848e-05,
      "loss": 1.1238,
      "step": 12033
    },
    {
      "epoch": 0.3169344219120358,
      "grad_norm": 1.5342379808425903,
      "learning_rate": 3.41638135370029e-05,
      "loss": 1.2737,
      "step": 12034
    },
    {
      "epoch": 0.31696075849354755,
      "grad_norm": 2.0451016426086426,
      "learning_rate": 3.416249670792731e-05,
      "loss": 1.9886,
      "step": 12035
    },
    {
      "epoch": 0.31698709507505923,
      "grad_norm": 2.403731107711792,
      "learning_rate": 3.416117987885173e-05,
      "loss": 1.4137,
      "step": 12036
    },
    {
      "epoch": 0.317013431656571,
      "grad_norm": 1.9101914167404175,
      "learning_rate": 3.415986304977614e-05,
      "loss": 1.4577,
      "step": 12037
    },
    {
      "epoch": 0.3170397682380827,
      "grad_norm": 2.213183641433716,
      "learning_rate": 3.4158546220700554e-05,
      "loss": 1.4521,
      "step": 12038
    },
    {
      "epoch": 0.3170661048195944,
      "grad_norm": 3.8314709663391113,
      "learning_rate": 3.415722939162497e-05,
      "loss": 2.6511,
      "step": 12039
    },
    {
      "epoch": 0.31709244140110615,
      "grad_norm": 3.5145654678344727,
      "learning_rate": 3.415591256254938e-05,
      "loss": 1.7947,
      "step": 12040
    },
    {
      "epoch": 0.31711877798261784,
      "grad_norm": 1.8209129571914673,
      "learning_rate": 3.41545957334738e-05,
      "loss": 1.7092,
      "step": 12041
    },
    {
      "epoch": 0.3171451145641296,
      "grad_norm": 2.524423837661743,
      "learning_rate": 3.415327890439821e-05,
      "loss": 1.6544,
      "step": 12042
    },
    {
      "epoch": 0.31717145114564127,
      "grad_norm": 1.9139906167984009,
      "learning_rate": 3.4151962075322625e-05,
      "loss": 1.6744,
      "step": 12043
    },
    {
      "epoch": 0.317197787727153,
      "grad_norm": 2.0966320037841797,
      "learning_rate": 3.4150645246247034e-05,
      "loss": 1.7232,
      "step": 12044
    },
    {
      "epoch": 0.31722412430866476,
      "grad_norm": 2.1557092666625977,
      "learning_rate": 3.414932841717145e-05,
      "loss": 1.8999,
      "step": 12045
    },
    {
      "epoch": 0.31725046089017644,
      "grad_norm": 4.261433124542236,
      "learning_rate": 3.4148011588095865e-05,
      "loss": 1.0017,
      "step": 12046
    },
    {
      "epoch": 0.3172767974716882,
      "grad_norm": 3.653890609741211,
      "learning_rate": 3.414669475902028e-05,
      "loss": 1.642,
      "step": 12047
    },
    {
      "epoch": 0.3173031340531999,
      "grad_norm": 1.5433707237243652,
      "learning_rate": 3.41453779299447e-05,
      "loss": 1.1799,
      "step": 12048
    },
    {
      "epoch": 0.3173294706347116,
      "grad_norm": 2.0163955688476562,
      "learning_rate": 3.4144061100869105e-05,
      "loss": 1.7505,
      "step": 12049
    },
    {
      "epoch": 0.3173558072162233,
      "grad_norm": 2.1403865814208984,
      "learning_rate": 3.414274427179353e-05,
      "loss": 0.3914,
      "step": 12050
    },
    {
      "epoch": 0.31738214379773505,
      "grad_norm": 3.932182550430298,
      "learning_rate": 3.414142744271794e-05,
      "loss": 1.4982,
      "step": 12051
    },
    {
      "epoch": 0.3174084803792468,
      "grad_norm": 3.1841015815734863,
      "learning_rate": 3.414011061364235e-05,
      "loss": 1.6272,
      "step": 12052
    },
    {
      "epoch": 0.3174348169607585,
      "grad_norm": 1.9604413509368896,
      "learning_rate": 3.413879378456676e-05,
      "loss": 1.4913,
      "step": 12053
    },
    {
      "epoch": 0.3174611535422702,
      "grad_norm": 3.753220319747925,
      "learning_rate": 3.413747695549118e-05,
      "loss": 2.1964,
      "step": 12054
    },
    {
      "epoch": 0.3174874901237819,
      "grad_norm": 2.9770264625549316,
      "learning_rate": 3.413616012641559e-05,
      "loss": 1.4059,
      "step": 12055
    },
    {
      "epoch": 0.31751382670529366,
      "grad_norm": 2.0103933811187744,
      "learning_rate": 3.413484329734001e-05,
      "loss": 1.8425,
      "step": 12056
    },
    {
      "epoch": 0.3175401632868054,
      "grad_norm": 2.5481927394866943,
      "learning_rate": 3.4133526468264423e-05,
      "loss": 1.5227,
      "step": 12057
    },
    {
      "epoch": 0.3175664998683171,
      "grad_norm": 1.9941326379776,
      "learning_rate": 3.413220963918883e-05,
      "loss": 0.549,
      "step": 12058
    },
    {
      "epoch": 0.31759283644982883,
      "grad_norm": 2.102574586868286,
      "learning_rate": 3.413089281011325e-05,
      "loss": 1.8916,
      "step": 12059
    },
    {
      "epoch": 0.3176191730313405,
      "grad_norm": 2.3135499954223633,
      "learning_rate": 3.4129575981037663e-05,
      "loss": 2.2591,
      "step": 12060
    },
    {
      "epoch": 0.31764550961285226,
      "grad_norm": 2.7947845458984375,
      "learning_rate": 3.412825915196208e-05,
      "loss": 1.9145,
      "step": 12061
    },
    {
      "epoch": 0.31767184619436395,
      "grad_norm": 2.105344295501709,
      "learning_rate": 3.412694232288649e-05,
      "loss": 1.4037,
      "step": 12062
    },
    {
      "epoch": 0.3176981827758757,
      "grad_norm": 1.7814490795135498,
      "learning_rate": 3.4125625493810903e-05,
      "loss": 1.9385,
      "step": 12063
    },
    {
      "epoch": 0.31772451935738744,
      "grad_norm": 3.6411960124969482,
      "learning_rate": 3.412430866473532e-05,
      "loss": 1.0843,
      "step": 12064
    },
    {
      "epoch": 0.3177508559388991,
      "grad_norm": 2.3417065143585205,
      "learning_rate": 3.4122991835659735e-05,
      "loss": 1.4668,
      "step": 12065
    },
    {
      "epoch": 0.31777719252041087,
      "grad_norm": 1.9404869079589844,
      "learning_rate": 3.412167500658415e-05,
      "loss": 1.8116,
      "step": 12066
    },
    {
      "epoch": 0.31780352910192256,
      "grad_norm": 1.628248929977417,
      "learning_rate": 3.412035817750856e-05,
      "loss": 0.3777,
      "step": 12067
    },
    {
      "epoch": 0.3178298656834343,
      "grad_norm": 3.219590663909912,
      "learning_rate": 3.4119041348432975e-05,
      "loss": 1.612,
      "step": 12068
    },
    {
      "epoch": 0.317856202264946,
      "grad_norm": 1.9707688093185425,
      "learning_rate": 3.411772451935739e-05,
      "loss": 1.6333,
      "step": 12069
    },
    {
      "epoch": 0.31788253884645773,
      "grad_norm": 3.203376293182373,
      "learning_rate": 3.4116407690281806e-05,
      "loss": 1.9735,
      "step": 12070
    },
    {
      "epoch": 0.3179088754279695,
      "grad_norm": 3.539792060852051,
      "learning_rate": 3.4115090861206215e-05,
      "loss": 0.2714,
      "step": 12071
    },
    {
      "epoch": 0.31793521200948116,
      "grad_norm": 2.378357410430908,
      "learning_rate": 3.411377403213063e-05,
      "loss": 1.9282,
      "step": 12072
    },
    {
      "epoch": 0.3179615485909929,
      "grad_norm": 2.246793746948242,
      "learning_rate": 3.4112457203055046e-05,
      "loss": 1.6198,
      "step": 12073
    },
    {
      "epoch": 0.3179878851725046,
      "grad_norm": 4.417208194732666,
      "learning_rate": 3.411114037397946e-05,
      "loss": 1.5774,
      "step": 12074
    },
    {
      "epoch": 0.31801422175401634,
      "grad_norm": 2.098876714706421,
      "learning_rate": 3.410982354490388e-05,
      "loss": 2.4549,
      "step": 12075
    },
    {
      "epoch": 0.318040558335528,
      "grad_norm": 2.2911365032196045,
      "learning_rate": 3.4108506715828286e-05,
      "loss": 2.2156,
      "step": 12076
    },
    {
      "epoch": 0.31806689491703977,
      "grad_norm": 1.9358553886413574,
      "learning_rate": 3.41071898867527e-05,
      "loss": 1.3755,
      "step": 12077
    },
    {
      "epoch": 0.3180932314985515,
      "grad_norm": 4.161898612976074,
      "learning_rate": 3.410587305767711e-05,
      "loss": 1.6894,
      "step": 12078
    },
    {
      "epoch": 0.3181195680800632,
      "grad_norm": 2.434128999710083,
      "learning_rate": 3.410455622860153e-05,
      "loss": 1.7256,
      "step": 12079
    },
    {
      "epoch": 0.31814590466157494,
      "grad_norm": 1.7992404699325562,
      "learning_rate": 3.410323939952594e-05,
      "loss": 1.8533,
      "step": 12080
    },
    {
      "epoch": 0.31817224124308663,
      "grad_norm": 3.1368579864501953,
      "learning_rate": 3.410192257045036e-05,
      "loss": 0.865,
      "step": 12081
    },
    {
      "epoch": 0.31819857782459837,
      "grad_norm": 2.074144124984741,
      "learning_rate": 3.4100605741374766e-05,
      "loss": 2.236,
      "step": 12082
    },
    {
      "epoch": 0.31822491440611006,
      "grad_norm": 2.1697885990142822,
      "learning_rate": 3.409928891229919e-05,
      "loss": 1.3337,
      "step": 12083
    },
    {
      "epoch": 0.3182512509876218,
      "grad_norm": 4.024163246154785,
      "learning_rate": 3.40979720832236e-05,
      "loss": 0.4982,
      "step": 12084
    },
    {
      "epoch": 0.31827758756913355,
      "grad_norm": 1.7225251197814941,
      "learning_rate": 3.409665525414801e-05,
      "loss": 1.8736,
      "step": 12085
    },
    {
      "epoch": 0.31830392415064523,
      "grad_norm": 1.5807503461837769,
      "learning_rate": 3.409533842507243e-05,
      "loss": 1.2598,
      "step": 12086
    },
    {
      "epoch": 0.318330260732157,
      "grad_norm": 2.1448254585266113,
      "learning_rate": 3.409402159599684e-05,
      "loss": 1.9726,
      "step": 12087
    },
    {
      "epoch": 0.31835659731366867,
      "grad_norm": 2.3013803958892822,
      "learning_rate": 3.409270476692126e-05,
      "loss": 1.5291,
      "step": 12088
    },
    {
      "epoch": 0.3183829338951804,
      "grad_norm": 2.0223681926727295,
      "learning_rate": 3.409138793784567e-05,
      "loss": 1.9953,
      "step": 12089
    },
    {
      "epoch": 0.31840927047669215,
      "grad_norm": 2.174805164337158,
      "learning_rate": 3.4090071108770084e-05,
      "loss": 1.5407,
      "step": 12090
    },
    {
      "epoch": 0.31843560705820384,
      "grad_norm": 3.7747116088867188,
      "learning_rate": 3.408875427969449e-05,
      "loss": 0.6589,
      "step": 12091
    },
    {
      "epoch": 0.3184619436397156,
      "grad_norm": 3.278367042541504,
      "learning_rate": 3.408743745061891e-05,
      "loss": 1.9589,
      "step": 12092
    },
    {
      "epoch": 0.31848828022122727,
      "grad_norm": 2.9114749431610107,
      "learning_rate": 3.4086120621543324e-05,
      "loss": 2.3347,
      "step": 12093
    },
    {
      "epoch": 0.318514616802739,
      "grad_norm": 1.9916917085647583,
      "learning_rate": 3.408480379246774e-05,
      "loss": 1.8849,
      "step": 12094
    },
    {
      "epoch": 0.3185409533842507,
      "grad_norm": 2.3800346851348877,
      "learning_rate": 3.4083486963392155e-05,
      "loss": 2.4327,
      "step": 12095
    },
    {
      "epoch": 0.31856728996576245,
      "grad_norm": 2.8650009632110596,
      "learning_rate": 3.4082170134316564e-05,
      "loss": 2.0588,
      "step": 12096
    },
    {
      "epoch": 0.3185936265472742,
      "grad_norm": 1.5561290979385376,
      "learning_rate": 3.4080853305240986e-05,
      "loss": 1.5646,
      "step": 12097
    },
    {
      "epoch": 0.3186199631287859,
      "grad_norm": 2.254415512084961,
      "learning_rate": 3.4079536476165395e-05,
      "loss": 1.4994,
      "step": 12098
    },
    {
      "epoch": 0.3186462997102976,
      "grad_norm": 1.9972496032714844,
      "learning_rate": 3.407821964708981e-05,
      "loss": 1.8567,
      "step": 12099
    },
    {
      "epoch": 0.3186726362918093,
      "grad_norm": 7.733311653137207,
      "learning_rate": 3.407690281801422e-05,
      "loss": 2.2662,
      "step": 12100
    },
    {
      "epoch": 0.31869897287332105,
      "grad_norm": 2.2288615703582764,
      "learning_rate": 3.4075585988938635e-05,
      "loss": 1.9325,
      "step": 12101
    },
    {
      "epoch": 0.31872530945483274,
      "grad_norm": 3.8718278408050537,
      "learning_rate": 3.407426915986305e-05,
      "loss": 1.1788,
      "step": 12102
    },
    {
      "epoch": 0.3187516460363445,
      "grad_norm": 2.4555227756500244,
      "learning_rate": 3.4072952330787466e-05,
      "loss": 1.8223,
      "step": 12103
    },
    {
      "epoch": 0.3187779826178562,
      "grad_norm": 1.5836929082870483,
      "learning_rate": 3.407163550171188e-05,
      "loss": 1.6193,
      "step": 12104
    },
    {
      "epoch": 0.3188043191993679,
      "grad_norm": 3.5536606311798096,
      "learning_rate": 3.407031867263629e-05,
      "loss": 0.9933,
      "step": 12105
    },
    {
      "epoch": 0.31883065578087966,
      "grad_norm": 2.322253942489624,
      "learning_rate": 3.4069001843560706e-05,
      "loss": 0.5696,
      "step": 12106
    },
    {
      "epoch": 0.31885699236239134,
      "grad_norm": 2.164285182952881,
      "learning_rate": 3.406768501448512e-05,
      "loss": 2.3208,
      "step": 12107
    },
    {
      "epoch": 0.3188833289439031,
      "grad_norm": 1.9893524646759033,
      "learning_rate": 3.406636818540954e-05,
      "loss": 1.8024,
      "step": 12108
    },
    {
      "epoch": 0.3189096655254148,
      "grad_norm": 2.137017011642456,
      "learning_rate": 3.4065051356333946e-05,
      "loss": 1.5274,
      "step": 12109
    },
    {
      "epoch": 0.3189360021069265,
      "grad_norm": 1.9572263956069946,
      "learning_rate": 3.406373452725836e-05,
      "loss": 0.9218,
      "step": 12110
    },
    {
      "epoch": 0.31896233868843826,
      "grad_norm": 1.8778127431869507,
      "learning_rate": 3.406241769818278e-05,
      "loss": 1.9975,
      "step": 12111
    },
    {
      "epoch": 0.31898867526994995,
      "grad_norm": 2.643911361694336,
      "learning_rate": 3.406110086910719e-05,
      "loss": 1.7684,
      "step": 12112
    },
    {
      "epoch": 0.3190150118514617,
      "grad_norm": 1.903601884841919,
      "learning_rate": 3.405978404003161e-05,
      "loss": 1.8392,
      "step": 12113
    },
    {
      "epoch": 0.3190413484329734,
      "grad_norm": 1.7770432233810425,
      "learning_rate": 3.405846721095602e-05,
      "loss": 0.8937,
      "step": 12114
    },
    {
      "epoch": 0.3190676850144851,
      "grad_norm": 4.653463363647461,
      "learning_rate": 3.405715038188043e-05,
      "loss": 1.2333,
      "step": 12115
    },
    {
      "epoch": 0.3190940215959968,
      "grad_norm": 2.2294225692749023,
      "learning_rate": 3.405583355280485e-05,
      "loss": 1.3843,
      "step": 12116
    },
    {
      "epoch": 0.31912035817750856,
      "grad_norm": 1.8459888696670532,
      "learning_rate": 3.4054516723729264e-05,
      "loss": 0.6468,
      "step": 12117
    },
    {
      "epoch": 0.3191466947590203,
      "grad_norm": 2.8435428142547607,
      "learning_rate": 3.405319989465367e-05,
      "loss": 1.4508,
      "step": 12118
    },
    {
      "epoch": 0.319173031340532,
      "grad_norm": 1.8968877792358398,
      "learning_rate": 3.405188306557809e-05,
      "loss": 1.6528,
      "step": 12119
    },
    {
      "epoch": 0.31919936792204373,
      "grad_norm": 2.4548451900482178,
      "learning_rate": 3.4050566236502504e-05,
      "loss": 1.4825,
      "step": 12120
    },
    {
      "epoch": 0.3192257045035554,
      "grad_norm": 1.5156699419021606,
      "learning_rate": 3.404924940742692e-05,
      "loss": 1.7052,
      "step": 12121
    },
    {
      "epoch": 0.31925204108506716,
      "grad_norm": 3.163053512573242,
      "learning_rate": 3.4047932578351336e-05,
      "loss": 1.839,
      "step": 12122
    },
    {
      "epoch": 0.3192783776665789,
      "grad_norm": 1.900566577911377,
      "learning_rate": 3.4046615749275744e-05,
      "loss": 1.8445,
      "step": 12123
    },
    {
      "epoch": 0.3193047142480906,
      "grad_norm": 2.2076849937438965,
      "learning_rate": 3.404529892020016e-05,
      "loss": 1.4097,
      "step": 12124
    },
    {
      "epoch": 0.31933105082960234,
      "grad_norm": 1.9737322330474854,
      "learning_rate": 3.404398209112457e-05,
      "loss": 1.607,
      "step": 12125
    },
    {
      "epoch": 0.319357387411114,
      "grad_norm": 3.2566566467285156,
      "learning_rate": 3.404266526204899e-05,
      "loss": 2.1212,
      "step": 12126
    },
    {
      "epoch": 0.31938372399262577,
      "grad_norm": 2.9648027420043945,
      "learning_rate": 3.40413484329734e-05,
      "loss": 1.6006,
      "step": 12127
    },
    {
      "epoch": 0.31941006057413746,
      "grad_norm": 1.5236337184906006,
      "learning_rate": 3.4040031603897816e-05,
      "loss": 1.8463,
      "step": 12128
    },
    {
      "epoch": 0.3194363971556492,
      "grad_norm": 2.531479835510254,
      "learning_rate": 3.403871477482223e-05,
      "loss": 1.1724,
      "step": 12129
    },
    {
      "epoch": 0.31946273373716094,
      "grad_norm": 4.001087188720703,
      "learning_rate": 3.403739794574665e-05,
      "loss": 1.6758,
      "step": 12130
    },
    {
      "epoch": 0.31948907031867263,
      "grad_norm": 2.445056915283203,
      "learning_rate": 3.403608111667106e-05,
      "loss": 2.7615,
      "step": 12131
    },
    {
      "epoch": 0.3195154069001844,
      "grad_norm": 2.243518829345703,
      "learning_rate": 3.403476428759547e-05,
      "loss": 1.4695,
      "step": 12132
    },
    {
      "epoch": 0.31954174348169606,
      "grad_norm": 1.8329102993011475,
      "learning_rate": 3.403344745851989e-05,
      "loss": 1.9909,
      "step": 12133
    },
    {
      "epoch": 0.3195680800632078,
      "grad_norm": 1.9460046291351318,
      "learning_rate": 3.4032130629444296e-05,
      "loss": 0.6969,
      "step": 12134
    },
    {
      "epoch": 0.3195944166447195,
      "grad_norm": 1.7553805112838745,
      "learning_rate": 3.403081380036872e-05,
      "loss": 2.4984,
      "step": 12135
    },
    {
      "epoch": 0.31962075322623124,
      "grad_norm": 2.204303741455078,
      "learning_rate": 3.402949697129313e-05,
      "loss": 2.247,
      "step": 12136
    },
    {
      "epoch": 0.319647089807743,
      "grad_norm": 2.5163116455078125,
      "learning_rate": 3.402818014221754e-05,
      "loss": 1.3254,
      "step": 12137
    },
    {
      "epoch": 0.31967342638925467,
      "grad_norm": 2.1387217044830322,
      "learning_rate": 3.402686331314195e-05,
      "loss": 1.8045,
      "step": 12138
    },
    {
      "epoch": 0.3196997629707664,
      "grad_norm": 2.000959873199463,
      "learning_rate": 3.402554648406637e-05,
      "loss": 1.3126,
      "step": 12139
    },
    {
      "epoch": 0.3197260995522781,
      "grad_norm": 2.282318592071533,
      "learning_rate": 3.402422965499078e-05,
      "loss": 1.1203,
      "step": 12140
    },
    {
      "epoch": 0.31975243613378984,
      "grad_norm": 2.2452168464660645,
      "learning_rate": 3.40229128259152e-05,
      "loss": 2.0484,
      "step": 12141
    },
    {
      "epoch": 0.31977877271530153,
      "grad_norm": 1.4107005596160889,
      "learning_rate": 3.4021595996839614e-05,
      "loss": 2.1388,
      "step": 12142
    },
    {
      "epoch": 0.3198051092968133,
      "grad_norm": 4.617241382598877,
      "learning_rate": 3.402027916776402e-05,
      "loss": 1.0133,
      "step": 12143
    },
    {
      "epoch": 0.319831445878325,
      "grad_norm": 2.887406826019287,
      "learning_rate": 3.4018962338688445e-05,
      "loss": 1.8506,
      "step": 12144
    },
    {
      "epoch": 0.3198577824598367,
      "grad_norm": 2.859163522720337,
      "learning_rate": 3.4017645509612854e-05,
      "loss": 2.4007,
      "step": 12145
    },
    {
      "epoch": 0.31988411904134845,
      "grad_norm": 2.2629973888397217,
      "learning_rate": 3.401632868053727e-05,
      "loss": 1.2927,
      "step": 12146
    },
    {
      "epoch": 0.31991045562286013,
      "grad_norm": 2.6468231678009033,
      "learning_rate": 3.401501185146168e-05,
      "loss": 0.6667,
      "step": 12147
    },
    {
      "epoch": 0.3199367922043719,
      "grad_norm": 2.2851922512054443,
      "learning_rate": 3.4013695022386094e-05,
      "loss": 2.0294,
      "step": 12148
    },
    {
      "epoch": 0.31996312878588357,
      "grad_norm": 3.090280771255493,
      "learning_rate": 3.401237819331051e-05,
      "loss": 1.3353,
      "step": 12149
    },
    {
      "epoch": 0.3199894653673953,
      "grad_norm": 2.6776058673858643,
      "learning_rate": 3.4011061364234925e-05,
      "loss": 2.5322,
      "step": 12150
    },
    {
      "epoch": 0.32001580194890705,
      "grad_norm": 1.7979923486709595,
      "learning_rate": 3.400974453515934e-05,
      "loss": 1.8427,
      "step": 12151
    },
    {
      "epoch": 0.32004213853041874,
      "grad_norm": 3.174229860305786,
      "learning_rate": 3.400842770608375e-05,
      "loss": 0.7254,
      "step": 12152
    },
    {
      "epoch": 0.3200684751119305,
      "grad_norm": 2.1720619201660156,
      "learning_rate": 3.4007110877008165e-05,
      "loss": 2.8202,
      "step": 12153
    },
    {
      "epoch": 0.32009481169344217,
      "grad_norm": 3.5753750801086426,
      "learning_rate": 3.400579404793258e-05,
      "loss": 1.0511,
      "step": 12154
    },
    {
      "epoch": 0.3201211482749539,
      "grad_norm": 2.3471946716308594,
      "learning_rate": 3.4004477218856996e-05,
      "loss": 1.0405,
      "step": 12155
    },
    {
      "epoch": 0.32014748485646566,
      "grad_norm": 2.4017467498779297,
      "learning_rate": 3.4003160389781405e-05,
      "loss": 2.0965,
      "step": 12156
    },
    {
      "epoch": 0.32017382143797735,
      "grad_norm": 2.4334700107574463,
      "learning_rate": 3.400184356070582e-05,
      "loss": 0.5958,
      "step": 12157
    },
    {
      "epoch": 0.3202001580194891,
      "grad_norm": 2.3893940448760986,
      "learning_rate": 3.4000526731630236e-05,
      "loss": 0.8703,
      "step": 12158
    },
    {
      "epoch": 0.3202264946010008,
      "grad_norm": 2.327883005142212,
      "learning_rate": 3.399920990255465e-05,
      "loss": 1.3811,
      "step": 12159
    },
    {
      "epoch": 0.3202528311825125,
      "grad_norm": 2.3480374813079834,
      "learning_rate": 3.399789307347907e-05,
      "loss": 1.6976,
      "step": 12160
    },
    {
      "epoch": 0.3202791677640242,
      "grad_norm": 1.9788191318511963,
      "learning_rate": 3.3996576244403476e-05,
      "loss": 1.4821,
      "step": 12161
    },
    {
      "epoch": 0.32030550434553595,
      "grad_norm": 2.0377748012542725,
      "learning_rate": 3.399525941532789e-05,
      "loss": 0.6055,
      "step": 12162
    },
    {
      "epoch": 0.3203318409270477,
      "grad_norm": 3.566192865371704,
      "learning_rate": 3.399394258625231e-05,
      "loss": 1.2174,
      "step": 12163
    },
    {
      "epoch": 0.3203581775085594,
      "grad_norm": 5.043156147003174,
      "learning_rate": 3.399262575717672e-05,
      "loss": 2.0138,
      "step": 12164
    },
    {
      "epoch": 0.3203845140900711,
      "grad_norm": 1.7911367416381836,
      "learning_rate": 3.399130892810113e-05,
      "loss": 2.0229,
      "step": 12165
    },
    {
      "epoch": 0.3204108506715828,
      "grad_norm": 3.9641623497009277,
      "learning_rate": 3.398999209902555e-05,
      "loss": 1.1658,
      "step": 12166
    },
    {
      "epoch": 0.32043718725309456,
      "grad_norm": 1.9728283882141113,
      "learning_rate": 3.398867526994996e-05,
      "loss": 1.6782,
      "step": 12167
    },
    {
      "epoch": 0.32046352383460625,
      "grad_norm": 1.463543176651001,
      "learning_rate": 3.398735844087438e-05,
      "loss": 1.6103,
      "step": 12168
    },
    {
      "epoch": 0.320489860416118,
      "grad_norm": 3.910576343536377,
      "learning_rate": 3.3986041611798794e-05,
      "loss": 1.4771,
      "step": 12169
    },
    {
      "epoch": 0.32051619699762973,
      "grad_norm": 1.9064232110977173,
      "learning_rate": 3.39847247827232e-05,
      "loss": 1.6162,
      "step": 12170
    },
    {
      "epoch": 0.3205425335791414,
      "grad_norm": 2.864616632461548,
      "learning_rate": 3.398340795364762e-05,
      "loss": 2.2952,
      "step": 12171
    },
    {
      "epoch": 0.32056887016065316,
      "grad_norm": 3.8245649337768555,
      "learning_rate": 3.398209112457203e-05,
      "loss": 2.1377,
      "step": 12172
    },
    {
      "epoch": 0.32059520674216485,
      "grad_norm": 3.4233615398406982,
      "learning_rate": 3.398077429549645e-05,
      "loss": 2.2099,
      "step": 12173
    },
    {
      "epoch": 0.3206215433236766,
      "grad_norm": 1.4633548259735107,
      "learning_rate": 3.397945746642086e-05,
      "loss": 1.6924,
      "step": 12174
    },
    {
      "epoch": 0.3206478799051883,
      "grad_norm": 4.341440677642822,
      "learning_rate": 3.3978140637345274e-05,
      "loss": 1.1626,
      "step": 12175
    },
    {
      "epoch": 0.3206742164867,
      "grad_norm": 4.72980260848999,
      "learning_rate": 3.397682380826969e-05,
      "loss": 1.0856,
      "step": 12176
    },
    {
      "epoch": 0.32070055306821177,
      "grad_norm": 1.5803526639938354,
      "learning_rate": 3.3975506979194105e-05,
      "loss": 2.1048,
      "step": 12177
    },
    {
      "epoch": 0.32072688964972346,
      "grad_norm": 2.1748430728912354,
      "learning_rate": 3.397419015011852e-05,
      "loss": 1.7495,
      "step": 12178
    },
    {
      "epoch": 0.3207532262312352,
      "grad_norm": 1.723378300666809,
      "learning_rate": 3.397287332104293e-05,
      "loss": 2.1792,
      "step": 12179
    },
    {
      "epoch": 0.3207795628127469,
      "grad_norm": 2.184858560562134,
      "learning_rate": 3.3971556491967345e-05,
      "loss": 1.9408,
      "step": 12180
    },
    {
      "epoch": 0.32080589939425863,
      "grad_norm": 1.7641090154647827,
      "learning_rate": 3.3970239662891754e-05,
      "loss": 2.3994,
      "step": 12181
    },
    {
      "epoch": 0.3208322359757703,
      "grad_norm": 2.993551254272461,
      "learning_rate": 3.396892283381618e-05,
      "loss": 1.1394,
      "step": 12182
    },
    {
      "epoch": 0.32085857255728206,
      "grad_norm": 1.6945013999938965,
      "learning_rate": 3.3967606004740585e-05,
      "loss": 1.6169,
      "step": 12183
    },
    {
      "epoch": 0.3208849091387938,
      "grad_norm": 1.5391260385513306,
      "learning_rate": 3.3966289175665e-05,
      "loss": 2.2243,
      "step": 12184
    },
    {
      "epoch": 0.3209112457203055,
      "grad_norm": 4.776205539703369,
      "learning_rate": 3.396497234658941e-05,
      "loss": 0.7905,
      "step": 12185
    },
    {
      "epoch": 0.32093758230181724,
      "grad_norm": 1.5860819816589355,
      "learning_rate": 3.3963655517513826e-05,
      "loss": 1.9358,
      "step": 12186
    },
    {
      "epoch": 0.3209639188833289,
      "grad_norm": 1.9016315937042236,
      "learning_rate": 3.396233868843824e-05,
      "loss": 1.6194,
      "step": 12187
    },
    {
      "epoch": 0.32099025546484067,
      "grad_norm": 1.9745495319366455,
      "learning_rate": 3.396102185936266e-05,
      "loss": 2.0536,
      "step": 12188
    },
    {
      "epoch": 0.32101659204635236,
      "grad_norm": 2.164306402206421,
      "learning_rate": 3.395970503028707e-05,
      "loss": 2.4298,
      "step": 12189
    },
    {
      "epoch": 0.3210429286278641,
      "grad_norm": 2.338559627532959,
      "learning_rate": 3.395838820121148e-05,
      "loss": 0.7754,
      "step": 12190
    },
    {
      "epoch": 0.32106926520937584,
      "grad_norm": 1.7641993761062622,
      "learning_rate": 3.39570713721359e-05,
      "loss": 1.7563,
      "step": 12191
    },
    {
      "epoch": 0.32109560179088753,
      "grad_norm": 2.2831690311431885,
      "learning_rate": 3.395575454306031e-05,
      "loss": 1.8925,
      "step": 12192
    },
    {
      "epoch": 0.3211219383723993,
      "grad_norm": 2.13026762008667,
      "learning_rate": 3.395443771398473e-05,
      "loss": 1.8684,
      "step": 12193
    },
    {
      "epoch": 0.32114827495391096,
      "grad_norm": 2.231293201446533,
      "learning_rate": 3.395312088490914e-05,
      "loss": 0.8055,
      "step": 12194
    },
    {
      "epoch": 0.3211746115354227,
      "grad_norm": 1.9471639394760132,
      "learning_rate": 3.395180405583355e-05,
      "loss": 1.7911,
      "step": 12195
    },
    {
      "epoch": 0.32120094811693445,
      "grad_norm": 1.8353394269943237,
      "learning_rate": 3.395048722675797e-05,
      "loss": 2.17,
      "step": 12196
    },
    {
      "epoch": 0.32122728469844614,
      "grad_norm": 2.2292640209198,
      "learning_rate": 3.3949170397682384e-05,
      "loss": 0.5162,
      "step": 12197
    },
    {
      "epoch": 0.3212536212799579,
      "grad_norm": 1.4795243740081787,
      "learning_rate": 3.39478535686068e-05,
      "loss": 1.4076,
      "step": 12198
    },
    {
      "epoch": 0.32127995786146957,
      "grad_norm": 3.0761191844940186,
      "learning_rate": 3.394653673953121e-05,
      "loss": 1.4909,
      "step": 12199
    },
    {
      "epoch": 0.3213062944429813,
      "grad_norm": 1.548272967338562,
      "learning_rate": 3.3945219910455624e-05,
      "loss": 1.3293,
      "step": 12200
    },
    {
      "epoch": 0.321332631024493,
      "grad_norm": 1.8872591257095337,
      "learning_rate": 3.394390308138004e-05,
      "loss": 1.068,
      "step": 12201
    },
    {
      "epoch": 0.32135896760600474,
      "grad_norm": 1.4373579025268555,
      "learning_rate": 3.3942586252304455e-05,
      "loss": 1.4413,
      "step": 12202
    },
    {
      "epoch": 0.3213853041875165,
      "grad_norm": 4.177943706512451,
      "learning_rate": 3.3941269423228864e-05,
      "loss": 2.1391,
      "step": 12203
    },
    {
      "epoch": 0.3214116407690282,
      "grad_norm": 1.7268850803375244,
      "learning_rate": 3.393995259415328e-05,
      "loss": 1.5692,
      "step": 12204
    },
    {
      "epoch": 0.3214379773505399,
      "grad_norm": 1.9290614128112793,
      "learning_rate": 3.3938635765077695e-05,
      "loss": 1.8553,
      "step": 12205
    },
    {
      "epoch": 0.3214643139320516,
      "grad_norm": 2.2076988220214844,
      "learning_rate": 3.393731893600211e-05,
      "loss": 1.4893,
      "step": 12206
    },
    {
      "epoch": 0.32149065051356335,
      "grad_norm": 3.3436062335968018,
      "learning_rate": 3.3936002106926526e-05,
      "loss": 1.4433,
      "step": 12207
    },
    {
      "epoch": 0.32151698709507504,
      "grad_norm": 1.653065800666809,
      "learning_rate": 3.3934685277850935e-05,
      "loss": 0.6151,
      "step": 12208
    },
    {
      "epoch": 0.3215433236765868,
      "grad_norm": 3.446596145629883,
      "learning_rate": 3.393336844877535e-05,
      "loss": 2.0906,
      "step": 12209
    },
    {
      "epoch": 0.3215696602580985,
      "grad_norm": 2.523874521255493,
      "learning_rate": 3.3932051619699766e-05,
      "loss": 1.2767,
      "step": 12210
    },
    {
      "epoch": 0.3215959968396102,
      "grad_norm": 2.230555772781372,
      "learning_rate": 3.393073479062418e-05,
      "loss": 2.0091,
      "step": 12211
    },
    {
      "epoch": 0.32162233342112195,
      "grad_norm": 3.2698090076446533,
      "learning_rate": 3.392941796154859e-05,
      "loss": 1.847,
      "step": 12212
    },
    {
      "epoch": 0.32164867000263364,
      "grad_norm": 2.6900603771209717,
      "learning_rate": 3.3928101132473006e-05,
      "loss": 1.9483,
      "step": 12213
    },
    {
      "epoch": 0.3216750065841454,
      "grad_norm": 2.081861734390259,
      "learning_rate": 3.392678430339742e-05,
      "loss": 0.761,
      "step": 12214
    },
    {
      "epoch": 0.32170134316565707,
      "grad_norm": 1.7848742008209229,
      "learning_rate": 3.392546747432184e-05,
      "loss": 0.7695,
      "step": 12215
    },
    {
      "epoch": 0.3217276797471688,
      "grad_norm": 2.8648509979248047,
      "learning_rate": 3.392415064524625e-05,
      "loss": 2.0384,
      "step": 12216
    },
    {
      "epoch": 0.32175401632868056,
      "grad_norm": 8.558164596557617,
      "learning_rate": 3.392283381617066e-05,
      "loss": 1.4706,
      "step": 12217
    },
    {
      "epoch": 0.32178035291019225,
      "grad_norm": 1.7810420989990234,
      "learning_rate": 3.392151698709508e-05,
      "loss": 1.9489,
      "step": 12218
    },
    {
      "epoch": 0.321806689491704,
      "grad_norm": 3.3097095489501953,
      "learning_rate": 3.3920200158019486e-05,
      "loss": 1.6992,
      "step": 12219
    },
    {
      "epoch": 0.3218330260732157,
      "grad_norm": 3.6095333099365234,
      "learning_rate": 3.391888332894391e-05,
      "loss": 0.9989,
      "step": 12220
    },
    {
      "epoch": 0.3218593626547274,
      "grad_norm": 2.969943046569824,
      "learning_rate": 3.391756649986832e-05,
      "loss": 1.9503,
      "step": 12221
    },
    {
      "epoch": 0.3218856992362391,
      "grad_norm": 1.9665263891220093,
      "learning_rate": 3.391624967079273e-05,
      "loss": 1.8913,
      "step": 12222
    },
    {
      "epoch": 0.32191203581775085,
      "grad_norm": 4.242003917694092,
      "learning_rate": 3.391493284171715e-05,
      "loss": 1.1605,
      "step": 12223
    },
    {
      "epoch": 0.3219383723992626,
      "grad_norm": 1.7454184293746948,
      "learning_rate": 3.391361601264156e-05,
      "loss": 1.9505,
      "step": 12224
    },
    {
      "epoch": 0.3219647089807743,
      "grad_norm": 2.203702211380005,
      "learning_rate": 3.391229918356598e-05,
      "loss": 1.9243,
      "step": 12225
    },
    {
      "epoch": 0.321991045562286,
      "grad_norm": 5.079842567443848,
      "learning_rate": 3.391098235449039e-05,
      "loss": 1.8182,
      "step": 12226
    },
    {
      "epoch": 0.3220173821437977,
      "grad_norm": 2.0906500816345215,
      "learning_rate": 3.3909665525414804e-05,
      "loss": 2.1544,
      "step": 12227
    },
    {
      "epoch": 0.32204371872530946,
      "grad_norm": 3.0349137783050537,
      "learning_rate": 3.390834869633921e-05,
      "loss": 1.5002,
      "step": 12228
    },
    {
      "epoch": 0.3220700553068212,
      "grad_norm": 2.0170960426330566,
      "learning_rate": 3.3907031867263635e-05,
      "loss": 1.4028,
      "step": 12229
    },
    {
      "epoch": 0.3220963918883329,
      "grad_norm": 1.68658447265625,
      "learning_rate": 3.3905715038188044e-05,
      "loss": 0.5158,
      "step": 12230
    },
    {
      "epoch": 0.32212272846984463,
      "grad_norm": 4.443417549133301,
      "learning_rate": 3.390439820911246e-05,
      "loss": 1.3005,
      "step": 12231
    },
    {
      "epoch": 0.3221490650513563,
      "grad_norm": 2.4294207096099854,
      "learning_rate": 3.390308138003687e-05,
      "loss": 0.3338,
      "step": 12232
    },
    {
      "epoch": 0.32217540163286806,
      "grad_norm": 2.0338082313537598,
      "learning_rate": 3.3901764550961284e-05,
      "loss": 1.6438,
      "step": 12233
    },
    {
      "epoch": 0.32220173821437975,
      "grad_norm": 2.186739206314087,
      "learning_rate": 3.3900447721885706e-05,
      "loss": 2.1927,
      "step": 12234
    },
    {
      "epoch": 0.3222280747958915,
      "grad_norm": 1.724195957183838,
      "learning_rate": 3.3899130892810115e-05,
      "loss": 1.8484,
      "step": 12235
    },
    {
      "epoch": 0.32225441137740324,
      "grad_norm": 1.8746455907821655,
      "learning_rate": 3.389781406373453e-05,
      "loss": 1.6226,
      "step": 12236
    },
    {
      "epoch": 0.3222807479589149,
      "grad_norm": 2.949380397796631,
      "learning_rate": 3.389649723465894e-05,
      "loss": 1.676,
      "step": 12237
    },
    {
      "epoch": 0.32230708454042667,
      "grad_norm": 2.543724775314331,
      "learning_rate": 3.3895180405583355e-05,
      "loss": 0.8616,
      "step": 12238
    },
    {
      "epoch": 0.32233342112193836,
      "grad_norm": 2.8032801151275635,
      "learning_rate": 3.389386357650777e-05,
      "loss": 1.4834,
      "step": 12239
    },
    {
      "epoch": 0.3223597577034501,
      "grad_norm": 1.9766201972961426,
      "learning_rate": 3.3892546747432186e-05,
      "loss": 2.1883,
      "step": 12240
    },
    {
      "epoch": 0.3223860942849618,
      "grad_norm": 1.6406527757644653,
      "learning_rate": 3.3891229918356595e-05,
      "loss": 2.2855,
      "step": 12241
    },
    {
      "epoch": 0.32241243086647353,
      "grad_norm": 1.7102583646774292,
      "learning_rate": 3.388991308928101e-05,
      "loss": 1.3,
      "step": 12242
    },
    {
      "epoch": 0.3224387674479853,
      "grad_norm": 2.1417951583862305,
      "learning_rate": 3.3888596260205426e-05,
      "loss": 1.8918,
      "step": 12243
    },
    {
      "epoch": 0.32246510402949696,
      "grad_norm": 1.9551180601119995,
      "learning_rate": 3.388727943112984e-05,
      "loss": 1.672,
      "step": 12244
    },
    {
      "epoch": 0.3224914406110087,
      "grad_norm": 2.5408904552459717,
      "learning_rate": 3.388596260205426e-05,
      "loss": 1.7678,
      "step": 12245
    },
    {
      "epoch": 0.3225177771925204,
      "grad_norm": 2.758303165435791,
      "learning_rate": 3.3884645772978667e-05,
      "loss": 1.9766,
      "step": 12246
    },
    {
      "epoch": 0.32254411377403214,
      "grad_norm": 3.7941858768463135,
      "learning_rate": 3.388332894390308e-05,
      "loss": 1.5693,
      "step": 12247
    },
    {
      "epoch": 0.3225704503555438,
      "grad_norm": 2.059492826461792,
      "learning_rate": 3.38820121148275e-05,
      "loss": 1.3157,
      "step": 12248
    },
    {
      "epoch": 0.32259678693705557,
      "grad_norm": 1.7243353128433228,
      "learning_rate": 3.388069528575191e-05,
      "loss": 1.6336,
      "step": 12249
    },
    {
      "epoch": 0.3226231235185673,
      "grad_norm": 2.2909388542175293,
      "learning_rate": 3.387937845667632e-05,
      "loss": 2.0044,
      "step": 12250
    },
    {
      "epoch": 0.322649460100079,
      "grad_norm": 3.018120050430298,
      "learning_rate": 3.387806162760074e-05,
      "loss": 1.943,
      "step": 12251
    },
    {
      "epoch": 0.32267579668159074,
      "grad_norm": 1.9069595336914062,
      "learning_rate": 3.387674479852515e-05,
      "loss": 1.9083,
      "step": 12252
    },
    {
      "epoch": 0.32270213326310243,
      "grad_norm": 1.864722490310669,
      "learning_rate": 3.387542796944957e-05,
      "loss": 1.5301,
      "step": 12253
    },
    {
      "epoch": 0.3227284698446142,
      "grad_norm": 1.5729719400405884,
      "learning_rate": 3.3874111140373985e-05,
      "loss": 0.2737,
      "step": 12254
    },
    {
      "epoch": 0.32275480642612586,
      "grad_norm": 1.5862400531768799,
      "learning_rate": 3.387279431129839e-05,
      "loss": 1.8645,
      "step": 12255
    },
    {
      "epoch": 0.3227811430076376,
      "grad_norm": 3.648317813873291,
      "learning_rate": 3.387147748222281e-05,
      "loss": 1.9888,
      "step": 12256
    },
    {
      "epoch": 0.32280747958914935,
      "grad_norm": 2.1134192943573,
      "learning_rate": 3.387016065314722e-05,
      "loss": 1.7019,
      "step": 12257
    },
    {
      "epoch": 0.32283381617066104,
      "grad_norm": 1.546311378479004,
      "learning_rate": 3.386884382407164e-05,
      "loss": 2.0524,
      "step": 12258
    },
    {
      "epoch": 0.3228601527521728,
      "grad_norm": 3.0315825939178467,
      "learning_rate": 3.386752699499605e-05,
      "loss": 1.0399,
      "step": 12259
    },
    {
      "epoch": 0.32288648933368447,
      "grad_norm": 3.750859260559082,
      "learning_rate": 3.3866210165920465e-05,
      "loss": 1.6299,
      "step": 12260
    },
    {
      "epoch": 0.3229128259151962,
      "grad_norm": 1.4505350589752197,
      "learning_rate": 3.386489333684488e-05,
      "loss": 2.1018,
      "step": 12261
    },
    {
      "epoch": 0.32293916249670795,
      "grad_norm": 2.239183187484741,
      "learning_rate": 3.3863576507769296e-05,
      "loss": 2.2628,
      "step": 12262
    },
    {
      "epoch": 0.32296549907821964,
      "grad_norm": 2.9473445415496826,
      "learning_rate": 3.386225967869371e-05,
      "loss": 1.3276,
      "step": 12263
    },
    {
      "epoch": 0.3229918356597314,
      "grad_norm": 2.2948169708251953,
      "learning_rate": 3.386094284961812e-05,
      "loss": 2.0559,
      "step": 12264
    },
    {
      "epoch": 0.3230181722412431,
      "grad_norm": 4.848300933837891,
      "learning_rate": 3.3859626020542536e-05,
      "loss": 0.8816,
      "step": 12265
    },
    {
      "epoch": 0.3230445088227548,
      "grad_norm": 1.973707675933838,
      "learning_rate": 3.3858309191466945e-05,
      "loss": 1.7563,
      "step": 12266
    },
    {
      "epoch": 0.3230708454042665,
      "grad_norm": 1.7574723958969116,
      "learning_rate": 3.385699236239137e-05,
      "loss": 1.714,
      "step": 12267
    },
    {
      "epoch": 0.32309718198577825,
      "grad_norm": 2.4838573932647705,
      "learning_rate": 3.3855675533315776e-05,
      "loss": 1.6102,
      "step": 12268
    },
    {
      "epoch": 0.32312351856729,
      "grad_norm": 1.9730017185211182,
      "learning_rate": 3.385435870424019e-05,
      "loss": 1.2705,
      "step": 12269
    },
    {
      "epoch": 0.3231498551488017,
      "grad_norm": 2.1752588748931885,
      "learning_rate": 3.385304187516461e-05,
      "loss": 1.6559,
      "step": 12270
    },
    {
      "epoch": 0.3231761917303134,
      "grad_norm": 2.166586399078369,
      "learning_rate": 3.3851725046089016e-05,
      "loss": 1.5732,
      "step": 12271
    },
    {
      "epoch": 0.3232025283118251,
      "grad_norm": 2.059168815612793,
      "learning_rate": 3.385040821701344e-05,
      "loss": 2.0538,
      "step": 12272
    },
    {
      "epoch": 0.32322886489333685,
      "grad_norm": 2.9020776748657227,
      "learning_rate": 3.384909138793785e-05,
      "loss": 1.8277,
      "step": 12273
    },
    {
      "epoch": 0.32325520147484854,
      "grad_norm": 1.9672114849090576,
      "learning_rate": 3.384777455886226e-05,
      "loss": 2.0089,
      "step": 12274
    },
    {
      "epoch": 0.3232815380563603,
      "grad_norm": 3.504380941390991,
      "learning_rate": 3.384645772978667e-05,
      "loss": 1.1946,
      "step": 12275
    },
    {
      "epoch": 0.32330787463787203,
      "grad_norm": 3.9589881896972656,
      "learning_rate": 3.3845140900711094e-05,
      "loss": 0.9693,
      "step": 12276
    },
    {
      "epoch": 0.3233342112193837,
      "grad_norm": 3.597632884979248,
      "learning_rate": 3.38438240716355e-05,
      "loss": 1.7431,
      "step": 12277
    },
    {
      "epoch": 0.32336054780089546,
      "grad_norm": 2.0312702655792236,
      "learning_rate": 3.384250724255992e-05,
      "loss": 1.4212,
      "step": 12278
    },
    {
      "epoch": 0.32338688438240715,
      "grad_norm": 1.6893987655639648,
      "learning_rate": 3.3841190413484334e-05,
      "loss": 1.8204,
      "step": 12279
    },
    {
      "epoch": 0.3234132209639189,
      "grad_norm": 2.1441118717193604,
      "learning_rate": 3.383987358440874e-05,
      "loss": 1.303,
      "step": 12280
    },
    {
      "epoch": 0.3234395575454306,
      "grad_norm": 1.5230028629302979,
      "learning_rate": 3.3838556755333165e-05,
      "loss": 1.7299,
      "step": 12281
    },
    {
      "epoch": 0.3234658941269423,
      "grad_norm": 2.374310255050659,
      "learning_rate": 3.3837239926257574e-05,
      "loss": 1.6029,
      "step": 12282
    },
    {
      "epoch": 0.32349223070845406,
      "grad_norm": 2.3975625038146973,
      "learning_rate": 3.383592309718199e-05,
      "loss": 2.222,
      "step": 12283
    },
    {
      "epoch": 0.32351856728996575,
      "grad_norm": 5.678077220916748,
      "learning_rate": 3.38346062681064e-05,
      "loss": 1.4825,
      "step": 12284
    },
    {
      "epoch": 0.3235449038714775,
      "grad_norm": 3.0202133655548096,
      "learning_rate": 3.3833289439030814e-05,
      "loss": 2.6372,
      "step": 12285
    },
    {
      "epoch": 0.3235712404529892,
      "grad_norm": 4.9063239097595215,
      "learning_rate": 3.383197260995523e-05,
      "loss": 1.6143,
      "step": 12286
    },
    {
      "epoch": 0.3235975770345009,
      "grad_norm": 1.5931549072265625,
      "learning_rate": 3.3830655780879645e-05,
      "loss": 1.6286,
      "step": 12287
    },
    {
      "epoch": 0.3236239136160126,
      "grad_norm": 5.184411525726318,
      "learning_rate": 3.3829338951804054e-05,
      "loss": 1.4513,
      "step": 12288
    },
    {
      "epoch": 0.32365025019752436,
      "grad_norm": 2.0110678672790527,
      "learning_rate": 3.382802212272847e-05,
      "loss": 1.8901,
      "step": 12289
    },
    {
      "epoch": 0.3236765867790361,
      "grad_norm": 2.3237550258636475,
      "learning_rate": 3.3826705293652885e-05,
      "loss": 1.7888,
      "step": 12290
    },
    {
      "epoch": 0.3237029233605478,
      "grad_norm": 2.883796215057373,
      "learning_rate": 3.38253884645773e-05,
      "loss": 2.0913,
      "step": 12291
    },
    {
      "epoch": 0.32372925994205953,
      "grad_norm": 2.086744785308838,
      "learning_rate": 3.3824071635501716e-05,
      "loss": 1.5644,
      "step": 12292
    },
    {
      "epoch": 0.3237555965235712,
      "grad_norm": 1.946567416191101,
      "learning_rate": 3.3822754806426125e-05,
      "loss": 2.0751,
      "step": 12293
    },
    {
      "epoch": 0.32378193310508296,
      "grad_norm": 2.670642137527466,
      "learning_rate": 3.382143797735054e-05,
      "loss": 1.8773,
      "step": 12294
    },
    {
      "epoch": 0.3238082696865947,
      "grad_norm": 1.9451168775558472,
      "learning_rate": 3.3820121148274956e-05,
      "loss": 2.8557,
      "step": 12295
    },
    {
      "epoch": 0.3238346062681064,
      "grad_norm": 1.9541621208190918,
      "learning_rate": 3.381880431919937e-05,
      "loss": 1.9104,
      "step": 12296
    },
    {
      "epoch": 0.32386094284961814,
      "grad_norm": 1.5282248258590698,
      "learning_rate": 3.381748749012378e-05,
      "loss": 0.558,
      "step": 12297
    },
    {
      "epoch": 0.3238872794311298,
      "grad_norm": 1.6153168678283691,
      "learning_rate": 3.3816170661048196e-05,
      "loss": 1.8507,
      "step": 12298
    },
    {
      "epoch": 0.32391361601264157,
      "grad_norm": 2.093909502029419,
      "learning_rate": 3.381485383197261e-05,
      "loss": 0.7208,
      "step": 12299
    },
    {
      "epoch": 0.32393995259415326,
      "grad_norm": 2.114145517349243,
      "learning_rate": 3.381353700289703e-05,
      "loss": 1.6762,
      "step": 12300
    },
    {
      "epoch": 0.323966289175665,
      "grad_norm": 3.022909164428711,
      "learning_rate": 3.381222017382144e-05,
      "loss": 1.3887,
      "step": 12301
    },
    {
      "epoch": 0.32399262575717674,
      "grad_norm": 1.9545685052871704,
      "learning_rate": 3.381090334474585e-05,
      "loss": 1.7816,
      "step": 12302
    },
    {
      "epoch": 0.32401896233868843,
      "grad_norm": 2.194612741470337,
      "learning_rate": 3.380958651567027e-05,
      "loss": 2.0078,
      "step": 12303
    },
    {
      "epoch": 0.3240452989202002,
      "grad_norm": 2.803537368774414,
      "learning_rate": 3.3808269686594676e-05,
      "loss": 1.0695,
      "step": 12304
    },
    {
      "epoch": 0.32407163550171186,
      "grad_norm": 2.8283658027648926,
      "learning_rate": 3.38069528575191e-05,
      "loss": 1.2953,
      "step": 12305
    },
    {
      "epoch": 0.3240979720832236,
      "grad_norm": 2.443341016769409,
      "learning_rate": 3.380563602844351e-05,
      "loss": 1.5769,
      "step": 12306
    },
    {
      "epoch": 0.3241243086647353,
      "grad_norm": 2.552309513092041,
      "learning_rate": 3.380431919936792e-05,
      "loss": 1.56,
      "step": 12307
    },
    {
      "epoch": 0.32415064524624704,
      "grad_norm": 1.6677271127700806,
      "learning_rate": 3.380300237029234e-05,
      "loss": 1.9866,
      "step": 12308
    },
    {
      "epoch": 0.3241769818277588,
      "grad_norm": 4.023653030395508,
      "learning_rate": 3.3801685541216754e-05,
      "loss": 1.5099,
      "step": 12309
    },
    {
      "epoch": 0.32420331840927047,
      "grad_norm": 1.7729556560516357,
      "learning_rate": 3.380036871214117e-05,
      "loss": 1.9922,
      "step": 12310
    },
    {
      "epoch": 0.3242296549907822,
      "grad_norm": 1.8447456359863281,
      "learning_rate": 3.379905188306558e-05,
      "loss": 1.8368,
      "step": 12311
    },
    {
      "epoch": 0.3242559915722939,
      "grad_norm": 1.5462827682495117,
      "learning_rate": 3.3797735053989994e-05,
      "loss": 1.727,
      "step": 12312
    },
    {
      "epoch": 0.32428232815380564,
      "grad_norm": 1.7580639123916626,
      "learning_rate": 3.37964182249144e-05,
      "loss": 2.4199,
      "step": 12313
    },
    {
      "epoch": 0.32430866473531733,
      "grad_norm": 2.0613253116607666,
      "learning_rate": 3.3795101395838826e-05,
      "loss": 1.3943,
      "step": 12314
    },
    {
      "epoch": 0.3243350013168291,
      "grad_norm": 2.076853036880493,
      "learning_rate": 3.3793784566763234e-05,
      "loss": 1.8899,
      "step": 12315
    },
    {
      "epoch": 0.3243613378983408,
      "grad_norm": 2.2645184993743896,
      "learning_rate": 3.379246773768765e-05,
      "loss": 2.3585,
      "step": 12316
    },
    {
      "epoch": 0.3243876744798525,
      "grad_norm": 1.675258994102478,
      "learning_rate": 3.3791150908612066e-05,
      "loss": 2.2064,
      "step": 12317
    },
    {
      "epoch": 0.32441401106136425,
      "grad_norm": 3.983132839202881,
      "learning_rate": 3.3789834079536474e-05,
      "loss": 1.1352,
      "step": 12318
    },
    {
      "epoch": 0.32444034764287594,
      "grad_norm": 2.144740581512451,
      "learning_rate": 3.37885172504609e-05,
      "loss": 1.8053,
      "step": 12319
    },
    {
      "epoch": 0.3244666842243877,
      "grad_norm": 1.438317894935608,
      "learning_rate": 3.3787200421385306e-05,
      "loss": 1.9145,
      "step": 12320
    },
    {
      "epoch": 0.32449302080589937,
      "grad_norm": 1.8782727718353271,
      "learning_rate": 3.378588359230972e-05,
      "loss": 1.996,
      "step": 12321
    },
    {
      "epoch": 0.3245193573874111,
      "grad_norm": 2.22731876373291,
      "learning_rate": 3.378456676323413e-05,
      "loss": 1.0348,
      "step": 12322
    },
    {
      "epoch": 0.32454569396892285,
      "grad_norm": 3.2880945205688477,
      "learning_rate": 3.3783249934158546e-05,
      "loss": 1.9547,
      "step": 12323
    },
    {
      "epoch": 0.32457203055043454,
      "grad_norm": 2.9851460456848145,
      "learning_rate": 3.378193310508296e-05,
      "loss": 1.8034,
      "step": 12324
    },
    {
      "epoch": 0.3245983671319463,
      "grad_norm": 5.274731636047363,
      "learning_rate": 3.378061627600738e-05,
      "loss": 1.7607,
      "step": 12325
    },
    {
      "epoch": 0.324624703713458,
      "grad_norm": 1.842307209968567,
      "learning_rate": 3.377929944693179e-05,
      "loss": 1.4218,
      "step": 12326
    },
    {
      "epoch": 0.3246510402949697,
      "grad_norm": 2.4799489974975586,
      "learning_rate": 3.37779826178562e-05,
      "loss": 1.7938,
      "step": 12327
    },
    {
      "epoch": 0.3246773768764814,
      "grad_norm": 2.7611265182495117,
      "learning_rate": 3.3776665788780624e-05,
      "loss": 1.3737,
      "step": 12328
    },
    {
      "epoch": 0.32470371345799315,
      "grad_norm": 1.8372855186462402,
      "learning_rate": 3.377534895970503e-05,
      "loss": 1.2665,
      "step": 12329
    },
    {
      "epoch": 0.3247300500395049,
      "grad_norm": 1.8387396335601807,
      "learning_rate": 3.377403213062945e-05,
      "loss": 2.0953,
      "step": 12330
    },
    {
      "epoch": 0.3247563866210166,
      "grad_norm": 2.6778085231781006,
      "learning_rate": 3.377271530155386e-05,
      "loss": 1.7535,
      "step": 12331
    },
    {
      "epoch": 0.3247827232025283,
      "grad_norm": 1.7880159616470337,
      "learning_rate": 3.377139847247827e-05,
      "loss": 1.8497,
      "step": 12332
    },
    {
      "epoch": 0.32480905978404,
      "grad_norm": 9.613354682922363,
      "learning_rate": 3.377008164340269e-05,
      "loss": 1.4307,
      "step": 12333
    },
    {
      "epoch": 0.32483539636555175,
      "grad_norm": 2.215589761734009,
      "learning_rate": 3.3768764814327104e-05,
      "loss": 1.9135,
      "step": 12334
    },
    {
      "epoch": 0.3248617329470635,
      "grad_norm": 2.5530459880828857,
      "learning_rate": 3.376744798525151e-05,
      "loss": 1.6609,
      "step": 12335
    },
    {
      "epoch": 0.3248880695285752,
      "grad_norm": 4.462213039398193,
      "learning_rate": 3.376613115617593e-05,
      "loss": 1.2865,
      "step": 12336
    },
    {
      "epoch": 0.32491440611008693,
      "grad_norm": 1.934516191482544,
      "learning_rate": 3.3764814327100344e-05,
      "loss": 1.1295,
      "step": 12337
    },
    {
      "epoch": 0.3249407426915986,
      "grad_norm": 2.748586654663086,
      "learning_rate": 3.376349749802476e-05,
      "loss": 1.414,
      "step": 12338
    },
    {
      "epoch": 0.32496707927311036,
      "grad_norm": 1.7129416465759277,
      "learning_rate": 3.3762180668949175e-05,
      "loss": 1.7636,
      "step": 12339
    },
    {
      "epoch": 0.32499341585462205,
      "grad_norm": 2.5688493251800537,
      "learning_rate": 3.3760863839873584e-05,
      "loss": 2.1608,
      "step": 12340
    },
    {
      "epoch": 0.3250197524361338,
      "grad_norm": 2.878887414932251,
      "learning_rate": 3.3759547010798e-05,
      "loss": 1.1483,
      "step": 12341
    },
    {
      "epoch": 0.32504608901764553,
      "grad_norm": 3.2131569385528564,
      "learning_rate": 3.3758230181722415e-05,
      "loss": 2.0589,
      "step": 12342
    },
    {
      "epoch": 0.3250724255991572,
      "grad_norm": 1.6457082033157349,
      "learning_rate": 3.375691335264683e-05,
      "loss": 1.6136,
      "step": 12343
    },
    {
      "epoch": 0.32509876218066897,
      "grad_norm": 2.799499034881592,
      "learning_rate": 3.375559652357124e-05,
      "loss": 1.3579,
      "step": 12344
    },
    {
      "epoch": 0.32512509876218065,
      "grad_norm": 1.765594482421875,
      "learning_rate": 3.3754279694495655e-05,
      "loss": 0.9123,
      "step": 12345
    },
    {
      "epoch": 0.3251514353436924,
      "grad_norm": 3.843245506286621,
      "learning_rate": 3.375296286542007e-05,
      "loss": 1.3026,
      "step": 12346
    },
    {
      "epoch": 0.3251777719252041,
      "grad_norm": 2.574646234512329,
      "learning_rate": 3.3751646036344486e-05,
      "loss": 1.84,
      "step": 12347
    },
    {
      "epoch": 0.3252041085067158,
      "grad_norm": 2.4024720191955566,
      "learning_rate": 3.37503292072689e-05,
      "loss": 2.2107,
      "step": 12348
    },
    {
      "epoch": 0.32523044508822757,
      "grad_norm": 1.7035294771194458,
      "learning_rate": 3.374901237819331e-05,
      "loss": 1.3948,
      "step": 12349
    },
    {
      "epoch": 0.32525678166973926,
      "grad_norm": 2.154402017593384,
      "learning_rate": 3.3747695549117726e-05,
      "loss": 2.227,
      "step": 12350
    },
    {
      "epoch": 0.325283118251251,
      "grad_norm": 2.6465022563934326,
      "learning_rate": 3.3746378720042135e-05,
      "loss": 1.0617,
      "step": 12351
    },
    {
      "epoch": 0.3253094548327627,
      "grad_norm": 1.896052598953247,
      "learning_rate": 3.374506189096656e-05,
      "loss": 1.5736,
      "step": 12352
    },
    {
      "epoch": 0.32533579141427443,
      "grad_norm": 1.764878273010254,
      "learning_rate": 3.3743745061890966e-05,
      "loss": 1.9243,
      "step": 12353
    },
    {
      "epoch": 0.3253621279957861,
      "grad_norm": 1.9510146379470825,
      "learning_rate": 3.374242823281538e-05,
      "loss": 1.4257,
      "step": 12354
    },
    {
      "epoch": 0.32538846457729786,
      "grad_norm": 2.0460031032562256,
      "learning_rate": 3.37411114037398e-05,
      "loss": 1.4265,
      "step": 12355
    },
    {
      "epoch": 0.3254148011588096,
      "grad_norm": 1.5084552764892578,
      "learning_rate": 3.3739794574664206e-05,
      "loss": 1.3017,
      "step": 12356
    },
    {
      "epoch": 0.3254411377403213,
      "grad_norm": 2.178532600402832,
      "learning_rate": 3.373847774558863e-05,
      "loss": 2.0115,
      "step": 12357
    },
    {
      "epoch": 0.32546747432183304,
      "grad_norm": 2.2687771320343018,
      "learning_rate": 3.373716091651304e-05,
      "loss": 0.5165,
      "step": 12358
    },
    {
      "epoch": 0.3254938109033447,
      "grad_norm": 1.6928813457489014,
      "learning_rate": 3.373584408743745e-05,
      "loss": 0.3237,
      "step": 12359
    },
    {
      "epoch": 0.32552014748485647,
      "grad_norm": 2.705045223236084,
      "learning_rate": 3.373452725836186e-05,
      "loss": 1.531,
      "step": 12360
    },
    {
      "epoch": 0.32554648406636816,
      "grad_norm": 3.2414040565490723,
      "learning_rate": 3.3733210429286284e-05,
      "loss": 1.9866,
      "step": 12361
    },
    {
      "epoch": 0.3255728206478799,
      "grad_norm": 2.9641237258911133,
      "learning_rate": 3.373189360021069e-05,
      "loss": 0.8239,
      "step": 12362
    },
    {
      "epoch": 0.32559915722939164,
      "grad_norm": 2.5464887619018555,
      "learning_rate": 3.373057677113511e-05,
      "loss": 1.5524,
      "step": 12363
    },
    {
      "epoch": 0.32562549381090333,
      "grad_norm": 1.8479719161987305,
      "learning_rate": 3.3729259942059524e-05,
      "loss": 1.626,
      "step": 12364
    },
    {
      "epoch": 0.3256518303924151,
      "grad_norm": 1.687949538230896,
      "learning_rate": 3.372794311298393e-05,
      "loss": 1.6007,
      "step": 12365
    },
    {
      "epoch": 0.32567816697392676,
      "grad_norm": 1.8811193704605103,
      "learning_rate": 3.3726626283908355e-05,
      "loss": 1.191,
      "step": 12366
    },
    {
      "epoch": 0.3257045035554385,
      "grad_norm": 2.518146514892578,
      "learning_rate": 3.3725309454832764e-05,
      "loss": 1.7585,
      "step": 12367
    },
    {
      "epoch": 0.32573084013695025,
      "grad_norm": 2.4291203022003174,
      "learning_rate": 3.372399262575718e-05,
      "loss": 1.6238,
      "step": 12368
    },
    {
      "epoch": 0.32575717671846194,
      "grad_norm": 2.936344861984253,
      "learning_rate": 3.372267579668159e-05,
      "loss": 1.6533,
      "step": 12369
    },
    {
      "epoch": 0.3257835132999737,
      "grad_norm": 2.00290846824646,
      "learning_rate": 3.3721358967606004e-05,
      "loss": 1.9891,
      "step": 12370
    },
    {
      "epoch": 0.32580984988148537,
      "grad_norm": 1.9928607940673828,
      "learning_rate": 3.372004213853042e-05,
      "loss": 1.72,
      "step": 12371
    },
    {
      "epoch": 0.3258361864629971,
      "grad_norm": 1.8765567541122437,
      "learning_rate": 3.3718725309454835e-05,
      "loss": 1.6647,
      "step": 12372
    },
    {
      "epoch": 0.3258625230445088,
      "grad_norm": 1.8526932001113892,
      "learning_rate": 3.371740848037925e-05,
      "loss": 1.9192,
      "step": 12373
    },
    {
      "epoch": 0.32588885962602054,
      "grad_norm": 2.6315598487854004,
      "learning_rate": 3.371609165130366e-05,
      "loss": 1.6672,
      "step": 12374
    },
    {
      "epoch": 0.3259151962075323,
      "grad_norm": 2.2314915657043457,
      "learning_rate": 3.371477482222808e-05,
      "loss": 1.392,
      "step": 12375
    },
    {
      "epoch": 0.325941532789044,
      "grad_norm": 1.7097437381744385,
      "learning_rate": 3.371345799315249e-05,
      "loss": 1.8734,
      "step": 12376
    },
    {
      "epoch": 0.3259678693705557,
      "grad_norm": 3.08762264251709,
      "learning_rate": 3.3712141164076907e-05,
      "loss": 2.1931,
      "step": 12377
    },
    {
      "epoch": 0.3259942059520674,
      "grad_norm": 3.167793035507202,
      "learning_rate": 3.3710824335001315e-05,
      "loss": 2.2833,
      "step": 12378
    },
    {
      "epoch": 0.32602054253357915,
      "grad_norm": 2.146003484725952,
      "learning_rate": 3.370950750592573e-05,
      "loss": 1.7115,
      "step": 12379
    },
    {
      "epoch": 0.32604687911509084,
      "grad_norm": 2.3912789821624756,
      "learning_rate": 3.3708190676850147e-05,
      "loss": 1.6485,
      "step": 12380
    },
    {
      "epoch": 0.3260732156966026,
      "grad_norm": 2.0407657623291016,
      "learning_rate": 3.370687384777456e-05,
      "loss": 1.1698,
      "step": 12381
    },
    {
      "epoch": 0.3260995522781143,
      "grad_norm": 1.600154161453247,
      "learning_rate": 3.370555701869898e-05,
      "loss": 1.6273,
      "step": 12382
    },
    {
      "epoch": 0.326125888859626,
      "grad_norm": 1.9794992208480835,
      "learning_rate": 3.3704240189623387e-05,
      "loss": 1.8946,
      "step": 12383
    },
    {
      "epoch": 0.32615222544113776,
      "grad_norm": 2.033773899078369,
      "learning_rate": 3.37029233605478e-05,
      "loss": 1.9258,
      "step": 12384
    },
    {
      "epoch": 0.32617856202264944,
      "grad_norm": 2.7881083488464355,
      "learning_rate": 3.370160653147222e-05,
      "loss": 1.7056,
      "step": 12385
    },
    {
      "epoch": 0.3262048986041612,
      "grad_norm": 2.3700952529907227,
      "learning_rate": 3.370028970239663e-05,
      "loss": 2.0705,
      "step": 12386
    },
    {
      "epoch": 0.3262312351856729,
      "grad_norm": 2.2574398517608643,
      "learning_rate": 3.369897287332104e-05,
      "loss": 1.6727,
      "step": 12387
    },
    {
      "epoch": 0.3262575717671846,
      "grad_norm": 1.82600736618042,
      "learning_rate": 3.369765604424546e-05,
      "loss": 1.4281,
      "step": 12388
    },
    {
      "epoch": 0.32628390834869636,
      "grad_norm": 1.817922830581665,
      "learning_rate": 3.3696339215169867e-05,
      "loss": 2.0027,
      "step": 12389
    },
    {
      "epoch": 0.32631024493020805,
      "grad_norm": 3.3472657203674316,
      "learning_rate": 3.369502238609429e-05,
      "loss": 0.8953,
      "step": 12390
    },
    {
      "epoch": 0.3263365815117198,
      "grad_norm": 3.2613322734832764,
      "learning_rate": 3.36937055570187e-05,
      "loss": 1.3292,
      "step": 12391
    },
    {
      "epoch": 0.3263629180932315,
      "grad_norm": 3.29406476020813,
      "learning_rate": 3.369238872794311e-05,
      "loss": 0.7488,
      "step": 12392
    },
    {
      "epoch": 0.3263892546747432,
      "grad_norm": 1.9919109344482422,
      "learning_rate": 3.369107189886753e-05,
      "loss": 1.9706,
      "step": 12393
    },
    {
      "epoch": 0.3264155912562549,
      "grad_norm": 2.754307508468628,
      "learning_rate": 3.3689755069791945e-05,
      "loss": 1.9736,
      "step": 12394
    },
    {
      "epoch": 0.32644192783776665,
      "grad_norm": 1.8472378253936768,
      "learning_rate": 3.368843824071636e-05,
      "loss": 1.8427,
      "step": 12395
    },
    {
      "epoch": 0.3264682644192784,
      "grad_norm": 6.103476047515869,
      "learning_rate": 3.368712141164077e-05,
      "loss": 1.8262,
      "step": 12396
    },
    {
      "epoch": 0.3264946010007901,
      "grad_norm": 1.5487394332885742,
      "learning_rate": 3.3685804582565185e-05,
      "loss": 2.1407,
      "step": 12397
    },
    {
      "epoch": 0.32652093758230183,
      "grad_norm": 1.9438989162445068,
      "learning_rate": 3.3684487753489593e-05,
      "loss": 1.1456,
      "step": 12398
    },
    {
      "epoch": 0.3265472741638135,
      "grad_norm": 2.440084457397461,
      "learning_rate": 3.3683170924414016e-05,
      "loss": 1.3676,
      "step": 12399
    },
    {
      "epoch": 0.32657361074532526,
      "grad_norm": 1.7957576513290405,
      "learning_rate": 3.3681854095338425e-05,
      "loss": 2.1561,
      "step": 12400
    },
    {
      "epoch": 0.326599947326837,
      "grad_norm": 7.39895486831665,
      "learning_rate": 3.368053726626284e-05,
      "loss": 1.3212,
      "step": 12401
    },
    {
      "epoch": 0.3266262839083487,
      "grad_norm": 3.1179237365722656,
      "learning_rate": 3.3679220437187256e-05,
      "loss": 1.9624,
      "step": 12402
    },
    {
      "epoch": 0.32665262048986043,
      "grad_norm": 1.9140419960021973,
      "learning_rate": 3.3677903608111665e-05,
      "loss": 0.7599,
      "step": 12403
    },
    {
      "epoch": 0.3266789570713721,
      "grad_norm": 2.248711347579956,
      "learning_rate": 3.367658677903609e-05,
      "loss": 0.7234,
      "step": 12404
    },
    {
      "epoch": 0.32670529365288387,
      "grad_norm": 3.0314104557037354,
      "learning_rate": 3.3675269949960496e-05,
      "loss": 0.3289,
      "step": 12405
    },
    {
      "epoch": 0.32673163023439555,
      "grad_norm": 2.7231507301330566,
      "learning_rate": 3.367395312088491e-05,
      "loss": 1.764,
      "step": 12406
    },
    {
      "epoch": 0.3267579668159073,
      "grad_norm": 2.1077165603637695,
      "learning_rate": 3.367263629180932e-05,
      "loss": 1.4074,
      "step": 12407
    },
    {
      "epoch": 0.32678430339741904,
      "grad_norm": 3.67710542678833,
      "learning_rate": 3.367131946273374e-05,
      "loss": 1.6472,
      "step": 12408
    },
    {
      "epoch": 0.32681063997893073,
      "grad_norm": 2.667994499206543,
      "learning_rate": 3.367000263365815e-05,
      "loss": 1.3321,
      "step": 12409
    },
    {
      "epoch": 0.32683697656044247,
      "grad_norm": 4.413286209106445,
      "learning_rate": 3.366868580458257e-05,
      "loss": 1.6661,
      "step": 12410
    },
    {
      "epoch": 0.32686331314195416,
      "grad_norm": 3.7115375995635986,
      "learning_rate": 3.366736897550698e-05,
      "loss": 2.1462,
      "step": 12411
    },
    {
      "epoch": 0.3268896497234659,
      "grad_norm": 1.8407247066497803,
      "learning_rate": 3.366605214643139e-05,
      "loss": 1.9583,
      "step": 12412
    },
    {
      "epoch": 0.3269159863049776,
      "grad_norm": 2.2176520824432373,
      "learning_rate": 3.3664735317355814e-05,
      "loss": 1.7093,
      "step": 12413
    },
    {
      "epoch": 0.32694232288648933,
      "grad_norm": 2.8115150928497314,
      "learning_rate": 3.366341848828022e-05,
      "loss": 1.3918,
      "step": 12414
    },
    {
      "epoch": 0.3269686594680011,
      "grad_norm": 4.907742977142334,
      "learning_rate": 3.366210165920464e-05,
      "loss": 1.931,
      "step": 12415
    },
    {
      "epoch": 0.32699499604951276,
      "grad_norm": 3.043381929397583,
      "learning_rate": 3.366078483012905e-05,
      "loss": 1.5289,
      "step": 12416
    },
    {
      "epoch": 0.3270213326310245,
      "grad_norm": 1.6548579931259155,
      "learning_rate": 3.365946800105346e-05,
      "loss": 1.6375,
      "step": 12417
    },
    {
      "epoch": 0.3270476692125362,
      "grad_norm": 1.8676625490188599,
      "learning_rate": 3.365815117197788e-05,
      "loss": 1.711,
      "step": 12418
    },
    {
      "epoch": 0.32707400579404794,
      "grad_norm": 2.144073247909546,
      "learning_rate": 3.3656834342902294e-05,
      "loss": 1.778,
      "step": 12419
    },
    {
      "epoch": 0.3271003423755596,
      "grad_norm": 1.729630708694458,
      "learning_rate": 3.365551751382671e-05,
      "loss": 1.9528,
      "step": 12420
    },
    {
      "epoch": 0.32712667895707137,
      "grad_norm": 3.874634027481079,
      "learning_rate": 3.365420068475112e-05,
      "loss": 1.094,
      "step": 12421
    },
    {
      "epoch": 0.3271530155385831,
      "grad_norm": 1.9382306337356567,
      "learning_rate": 3.365288385567554e-05,
      "loss": 1.9941,
      "step": 12422
    },
    {
      "epoch": 0.3271793521200948,
      "grad_norm": 1.4532071352005005,
      "learning_rate": 3.365156702659995e-05,
      "loss": 1.6583,
      "step": 12423
    },
    {
      "epoch": 0.32720568870160655,
      "grad_norm": 2.770235300064087,
      "learning_rate": 3.3650250197524365e-05,
      "loss": 2.3272,
      "step": 12424
    },
    {
      "epoch": 0.32723202528311823,
      "grad_norm": 4.797533988952637,
      "learning_rate": 3.3648933368448774e-05,
      "loss": 1.2025,
      "step": 12425
    },
    {
      "epoch": 0.32725836186463,
      "grad_norm": 1.8548167943954468,
      "learning_rate": 3.364761653937319e-05,
      "loss": 2.0565,
      "step": 12426
    },
    {
      "epoch": 0.32728469844614166,
      "grad_norm": 1.925743579864502,
      "learning_rate": 3.3646299710297605e-05,
      "loss": 2.0501,
      "step": 12427
    },
    {
      "epoch": 0.3273110350276534,
      "grad_norm": 1.7694215774536133,
      "learning_rate": 3.364498288122202e-05,
      "loss": 1.8351,
      "step": 12428
    },
    {
      "epoch": 0.32733737160916515,
      "grad_norm": 2.0810256004333496,
      "learning_rate": 3.3643666052146436e-05,
      "loss": 1.4491,
      "step": 12429
    },
    {
      "epoch": 0.32736370819067684,
      "grad_norm": 1.7464072704315186,
      "learning_rate": 3.3642349223070845e-05,
      "loss": 2.1283,
      "step": 12430
    },
    {
      "epoch": 0.3273900447721886,
      "grad_norm": 2.4301838874816895,
      "learning_rate": 3.364103239399526e-05,
      "loss": 2.4133,
      "step": 12431
    },
    {
      "epoch": 0.32741638135370027,
      "grad_norm": 3.1562137603759766,
      "learning_rate": 3.3639715564919676e-05,
      "loss": 1.9187,
      "step": 12432
    },
    {
      "epoch": 0.327442717935212,
      "grad_norm": 2.2507240772247314,
      "learning_rate": 3.363839873584409e-05,
      "loss": 1.6733,
      "step": 12433
    },
    {
      "epoch": 0.32746905451672376,
      "grad_norm": 4.765883922576904,
      "learning_rate": 3.36370819067685e-05,
      "loss": 1.9338,
      "step": 12434
    },
    {
      "epoch": 0.32749539109823544,
      "grad_norm": 2.0596113204956055,
      "learning_rate": 3.3635765077692916e-05,
      "loss": 2.1882,
      "step": 12435
    },
    {
      "epoch": 0.3275217276797472,
      "grad_norm": 1.7240318059921265,
      "learning_rate": 3.3634448248617325e-05,
      "loss": 1.4548,
      "step": 12436
    },
    {
      "epoch": 0.3275480642612589,
      "grad_norm": 3.7175097465515137,
      "learning_rate": 3.363313141954175e-05,
      "loss": 1.1278,
      "step": 12437
    },
    {
      "epoch": 0.3275744008427706,
      "grad_norm": 1.616626262664795,
      "learning_rate": 3.3631814590466156e-05,
      "loss": 1.6143,
      "step": 12438
    },
    {
      "epoch": 0.3276007374242823,
      "grad_norm": 2.5229551792144775,
      "learning_rate": 3.363049776139057e-05,
      "loss": 2.9027,
      "step": 12439
    },
    {
      "epoch": 0.32762707400579405,
      "grad_norm": 1.7705267667770386,
      "learning_rate": 3.362918093231499e-05,
      "loss": 1.5911,
      "step": 12440
    },
    {
      "epoch": 0.3276534105873058,
      "grad_norm": 2.192378044128418,
      "learning_rate": 3.36278641032394e-05,
      "loss": 2.8516,
      "step": 12441
    },
    {
      "epoch": 0.3276797471688175,
      "grad_norm": 3.0754435062408447,
      "learning_rate": 3.362654727416382e-05,
      "loss": 1.01,
      "step": 12442
    },
    {
      "epoch": 0.3277060837503292,
      "grad_norm": 2.4538142681121826,
      "learning_rate": 3.362523044508823e-05,
      "loss": 1.5274,
      "step": 12443
    },
    {
      "epoch": 0.3277324203318409,
      "grad_norm": 3.1316940784454346,
      "learning_rate": 3.362391361601264e-05,
      "loss": 1.5736,
      "step": 12444
    },
    {
      "epoch": 0.32775875691335266,
      "grad_norm": 2.8405120372772217,
      "learning_rate": 3.362259678693705e-05,
      "loss": 1.0018,
      "step": 12445
    },
    {
      "epoch": 0.32778509349486434,
      "grad_norm": 1.8347320556640625,
      "learning_rate": 3.3621279957861474e-05,
      "loss": 1.2207,
      "step": 12446
    },
    {
      "epoch": 0.3278114300763761,
      "grad_norm": 5.300283432006836,
      "learning_rate": 3.361996312878588e-05,
      "loss": 2.0342,
      "step": 12447
    },
    {
      "epoch": 0.32783776665788783,
      "grad_norm": 2.5827488899230957,
      "learning_rate": 3.36186462997103e-05,
      "loss": 2.1086,
      "step": 12448
    },
    {
      "epoch": 0.3278641032393995,
      "grad_norm": 3.9513463973999023,
      "learning_rate": 3.3617329470634714e-05,
      "loss": 1.9389,
      "step": 12449
    },
    {
      "epoch": 0.32789043982091126,
      "grad_norm": 3.1131837368011475,
      "learning_rate": 3.361601264155912e-05,
      "loss": 1.8678,
      "step": 12450
    },
    {
      "epoch": 0.32791677640242295,
      "grad_norm": 6.7721848487854,
      "learning_rate": 3.3614695812483546e-05,
      "loss": 1.4235,
      "step": 12451
    },
    {
      "epoch": 0.3279431129839347,
      "grad_norm": 2.0119690895080566,
      "learning_rate": 3.3613378983407954e-05,
      "loss": 1.7649,
      "step": 12452
    },
    {
      "epoch": 0.3279694495654464,
      "grad_norm": 1.8033760786056519,
      "learning_rate": 3.361206215433237e-05,
      "loss": 1.6084,
      "step": 12453
    },
    {
      "epoch": 0.3279957861469581,
      "grad_norm": 1.9612101316452026,
      "learning_rate": 3.361074532525678e-05,
      "loss": 1.449,
      "step": 12454
    },
    {
      "epoch": 0.32802212272846987,
      "grad_norm": 4.113655090332031,
      "learning_rate": 3.36094284961812e-05,
      "loss": 1.3775,
      "step": 12455
    },
    {
      "epoch": 0.32804845930998155,
      "grad_norm": 1.8795216083526611,
      "learning_rate": 3.360811166710561e-05,
      "loss": 1.908,
      "step": 12456
    },
    {
      "epoch": 0.3280747958914933,
      "grad_norm": 2.2812535762786865,
      "learning_rate": 3.3606794838030026e-05,
      "loss": 2.4844,
      "step": 12457
    },
    {
      "epoch": 0.328101132473005,
      "grad_norm": 2.2446975708007812,
      "learning_rate": 3.360547800895444e-05,
      "loss": 0.8677,
      "step": 12458
    },
    {
      "epoch": 0.32812746905451673,
      "grad_norm": 2.650233268737793,
      "learning_rate": 3.360416117987885e-05,
      "loss": 1.5683,
      "step": 12459
    },
    {
      "epoch": 0.3281538056360284,
      "grad_norm": 1.758612036705017,
      "learning_rate": 3.360284435080327e-05,
      "loss": 1.8904,
      "step": 12460
    },
    {
      "epoch": 0.32818014221754016,
      "grad_norm": 2.0016775131225586,
      "learning_rate": 3.360152752172768e-05,
      "loss": 1.5744,
      "step": 12461
    },
    {
      "epoch": 0.3282064787990519,
      "grad_norm": 2.1309597492218018,
      "learning_rate": 3.36002106926521e-05,
      "loss": 1.4482,
      "step": 12462
    },
    {
      "epoch": 0.3282328153805636,
      "grad_norm": 1.9692991971969604,
      "learning_rate": 3.3598893863576506e-05,
      "loss": 1.775,
      "step": 12463
    },
    {
      "epoch": 0.32825915196207534,
      "grad_norm": 2.5524942874908447,
      "learning_rate": 3.359757703450092e-05,
      "loss": 1.7822,
      "step": 12464
    },
    {
      "epoch": 0.328285488543587,
      "grad_norm": 1.9405978918075562,
      "learning_rate": 3.359626020542534e-05,
      "loss": 1.4305,
      "step": 12465
    },
    {
      "epoch": 0.32831182512509877,
      "grad_norm": 1.656632423400879,
      "learning_rate": 3.359494337634975e-05,
      "loss": 1.082,
      "step": 12466
    },
    {
      "epoch": 0.3283381617066105,
      "grad_norm": 3.533543825149536,
      "learning_rate": 3.359362654727417e-05,
      "loss": 2.1386,
      "step": 12467
    },
    {
      "epoch": 0.3283644982881222,
      "grad_norm": 2.101546049118042,
      "learning_rate": 3.359230971819858e-05,
      "loss": 1.5001,
      "step": 12468
    },
    {
      "epoch": 0.32839083486963394,
      "grad_norm": 2.174426555633545,
      "learning_rate": 3.359099288912299e-05,
      "loss": 1.789,
      "step": 12469
    },
    {
      "epoch": 0.32841717145114563,
      "grad_norm": 7.761478424072266,
      "learning_rate": 3.358967606004741e-05,
      "loss": 1.0197,
      "step": 12470
    },
    {
      "epoch": 0.32844350803265737,
      "grad_norm": 2.63983154296875,
      "learning_rate": 3.3588359230971824e-05,
      "loss": 2.0716,
      "step": 12471
    },
    {
      "epoch": 0.32846984461416906,
      "grad_norm": 2.8965344429016113,
      "learning_rate": 3.358704240189623e-05,
      "loss": 1.2944,
      "step": 12472
    },
    {
      "epoch": 0.3284961811956808,
      "grad_norm": 5.850398540496826,
      "learning_rate": 3.358572557282065e-05,
      "loss": 1.8968,
      "step": 12473
    },
    {
      "epoch": 0.32852251777719255,
      "grad_norm": 2.1706478595733643,
      "learning_rate": 3.3584408743745064e-05,
      "loss": 2.2624,
      "step": 12474
    },
    {
      "epoch": 0.32854885435870423,
      "grad_norm": 2.218282461166382,
      "learning_rate": 3.358309191466948e-05,
      "loss": 1.8951,
      "step": 12475
    },
    {
      "epoch": 0.328575190940216,
      "grad_norm": 2.908212661743164,
      "learning_rate": 3.3581775085593895e-05,
      "loss": 1.4327,
      "step": 12476
    },
    {
      "epoch": 0.32860152752172767,
      "grad_norm": 1.7072484493255615,
      "learning_rate": 3.3580458256518304e-05,
      "loss": 1.9014,
      "step": 12477
    },
    {
      "epoch": 0.3286278641032394,
      "grad_norm": 1.9307185411453247,
      "learning_rate": 3.357914142744272e-05,
      "loss": 0.7581,
      "step": 12478
    },
    {
      "epoch": 0.3286542006847511,
      "grad_norm": 1.5613802671432495,
      "learning_rate": 3.3577824598367135e-05,
      "loss": 1.4153,
      "step": 12479
    },
    {
      "epoch": 0.32868053726626284,
      "grad_norm": 1.8219326734542847,
      "learning_rate": 3.357650776929155e-05,
      "loss": 1.937,
      "step": 12480
    },
    {
      "epoch": 0.3287068738477746,
      "grad_norm": 2.997546672821045,
      "learning_rate": 3.357519094021596e-05,
      "loss": 1.003,
      "step": 12481
    },
    {
      "epoch": 0.32873321042928627,
      "grad_norm": 1.6718472242355347,
      "learning_rate": 3.3573874111140375e-05,
      "loss": 0.8359,
      "step": 12482
    },
    {
      "epoch": 0.328759547010798,
      "grad_norm": 1.7890629768371582,
      "learning_rate": 3.357255728206479e-05,
      "loss": 0.386,
      "step": 12483
    },
    {
      "epoch": 0.3287858835923097,
      "grad_norm": 1.9585483074188232,
      "learning_rate": 3.3571240452989206e-05,
      "loss": 1.9386,
      "step": 12484
    },
    {
      "epoch": 0.32881222017382145,
      "grad_norm": 1.6718919277191162,
      "learning_rate": 3.356992362391362e-05,
      "loss": 1.7794,
      "step": 12485
    },
    {
      "epoch": 0.32883855675533313,
      "grad_norm": 2.9827446937561035,
      "learning_rate": 3.356860679483803e-05,
      "loss": 1.6119,
      "step": 12486
    },
    {
      "epoch": 0.3288648933368449,
      "grad_norm": 1.8520004749298096,
      "learning_rate": 3.3567289965762446e-05,
      "loss": 1.3068,
      "step": 12487
    },
    {
      "epoch": 0.3288912299183566,
      "grad_norm": 1.957453966140747,
      "learning_rate": 3.356597313668686e-05,
      "loss": 2.1151,
      "step": 12488
    },
    {
      "epoch": 0.3289175664998683,
      "grad_norm": 2.7240259647369385,
      "learning_rate": 3.356465630761128e-05,
      "loss": 1.6568,
      "step": 12489
    },
    {
      "epoch": 0.32894390308138005,
      "grad_norm": 1.7372244596481323,
      "learning_rate": 3.3563339478535686e-05,
      "loss": 1.5207,
      "step": 12490
    },
    {
      "epoch": 0.32897023966289174,
      "grad_norm": 1.7322607040405273,
      "learning_rate": 3.35620226494601e-05,
      "loss": 2.1428,
      "step": 12491
    },
    {
      "epoch": 0.3289965762444035,
      "grad_norm": 1.8577086925506592,
      "learning_rate": 3.356070582038451e-05,
      "loss": 1.2703,
      "step": 12492
    },
    {
      "epoch": 0.32902291282591517,
      "grad_norm": 1.6735259294509888,
      "learning_rate": 3.355938899130893e-05,
      "loss": 1.4099,
      "step": 12493
    },
    {
      "epoch": 0.3290492494074269,
      "grad_norm": 3.109797954559326,
      "learning_rate": 3.355807216223334e-05,
      "loss": 1.8465,
      "step": 12494
    },
    {
      "epoch": 0.32907558598893866,
      "grad_norm": 1.9641128778457642,
      "learning_rate": 3.355675533315776e-05,
      "loss": 1.2254,
      "step": 12495
    },
    {
      "epoch": 0.32910192257045034,
      "grad_norm": 3.574244737625122,
      "learning_rate": 3.355543850408217e-05,
      "loss": 1.7763,
      "step": 12496
    },
    {
      "epoch": 0.3291282591519621,
      "grad_norm": 1.7841097116470337,
      "learning_rate": 3.355412167500658e-05,
      "loss": 1.7523,
      "step": 12497
    },
    {
      "epoch": 0.3291545957334738,
      "grad_norm": 2.0619144439697266,
      "learning_rate": 3.3552804845931004e-05,
      "loss": 1.8517,
      "step": 12498
    },
    {
      "epoch": 0.3291809323149855,
      "grad_norm": 2.235921621322632,
      "learning_rate": 3.355148801685541e-05,
      "loss": 1.2338,
      "step": 12499
    },
    {
      "epoch": 0.3292072688964972,
      "grad_norm": 4.301616191864014,
      "learning_rate": 3.355017118777983e-05,
      "loss": 1.2537,
      "step": 12500
    },
    {
      "epoch": 0.32923360547800895,
      "grad_norm": 3.540292739868164,
      "learning_rate": 3.354885435870424e-05,
      "loss": 2.1768,
      "step": 12501
    },
    {
      "epoch": 0.3292599420595207,
      "grad_norm": 1.493011713027954,
      "learning_rate": 3.354753752962865e-05,
      "loss": 0.552,
      "step": 12502
    },
    {
      "epoch": 0.3292862786410324,
      "grad_norm": 4.220021724700928,
      "learning_rate": 3.354622070055307e-05,
      "loss": 1.2234,
      "step": 12503
    },
    {
      "epoch": 0.3293126152225441,
      "grad_norm": 1.9487255811691284,
      "learning_rate": 3.3544903871477484e-05,
      "loss": 1.1783,
      "step": 12504
    },
    {
      "epoch": 0.3293389518040558,
      "grad_norm": 1.8226747512817383,
      "learning_rate": 3.35435870424019e-05,
      "loss": 1.3528,
      "step": 12505
    },
    {
      "epoch": 0.32936528838556756,
      "grad_norm": 1.8626196384429932,
      "learning_rate": 3.354227021332631e-05,
      "loss": 2.0449,
      "step": 12506
    },
    {
      "epoch": 0.3293916249670793,
      "grad_norm": 1.6120344400405884,
      "learning_rate": 3.354095338425073e-05,
      "loss": 1.9424,
      "step": 12507
    },
    {
      "epoch": 0.329417961548591,
      "grad_norm": 3.8325703144073486,
      "learning_rate": 3.353963655517514e-05,
      "loss": 1.5654,
      "step": 12508
    },
    {
      "epoch": 0.32944429813010273,
      "grad_norm": 1.9195082187652588,
      "learning_rate": 3.3538319726099555e-05,
      "loss": 2.3664,
      "step": 12509
    },
    {
      "epoch": 0.3294706347116144,
      "grad_norm": 1.8883514404296875,
      "learning_rate": 3.3537002897023964e-05,
      "loss": 1.9992,
      "step": 12510
    },
    {
      "epoch": 0.32949697129312616,
      "grad_norm": 3.385572671890259,
      "learning_rate": 3.353568606794838e-05,
      "loss": 1.2949,
      "step": 12511
    },
    {
      "epoch": 0.32952330787463785,
      "grad_norm": 2.470580816268921,
      "learning_rate": 3.3534369238872795e-05,
      "loss": 2.3398,
      "step": 12512
    },
    {
      "epoch": 0.3295496444561496,
      "grad_norm": 1.830094337463379,
      "learning_rate": 3.353305240979721e-05,
      "loss": 1.8316,
      "step": 12513
    },
    {
      "epoch": 0.32957598103766134,
      "grad_norm": 3.9431777000427246,
      "learning_rate": 3.3531735580721627e-05,
      "loss": 1.1087,
      "step": 12514
    },
    {
      "epoch": 0.329602317619173,
      "grad_norm": 1.9579212665557861,
      "learning_rate": 3.3530418751646035e-05,
      "loss": 1.177,
      "step": 12515
    },
    {
      "epoch": 0.32962865420068477,
      "grad_norm": 1.8958979845046997,
      "learning_rate": 3.352910192257045e-05,
      "loss": 0.5646,
      "step": 12516
    },
    {
      "epoch": 0.32965499078219646,
      "grad_norm": 2.0430240631103516,
      "learning_rate": 3.3527785093494867e-05,
      "loss": 1.7158,
      "step": 12517
    },
    {
      "epoch": 0.3296813273637082,
      "grad_norm": 4.056602954864502,
      "learning_rate": 3.352646826441928e-05,
      "loss": 1.1088,
      "step": 12518
    },
    {
      "epoch": 0.3297076639452199,
      "grad_norm": 2.469667434692383,
      "learning_rate": 3.352515143534369e-05,
      "loss": 1.8398,
      "step": 12519
    },
    {
      "epoch": 0.32973400052673163,
      "grad_norm": 4.454177379608154,
      "learning_rate": 3.3523834606268107e-05,
      "loss": 2.3589,
      "step": 12520
    },
    {
      "epoch": 0.3297603371082434,
      "grad_norm": 4.095808029174805,
      "learning_rate": 3.352251777719252e-05,
      "loss": 1.4604,
      "step": 12521
    },
    {
      "epoch": 0.32978667368975506,
      "grad_norm": 1.9336916208267212,
      "learning_rate": 3.352120094811694e-05,
      "loss": 1.6101,
      "step": 12522
    },
    {
      "epoch": 0.3298130102712668,
      "grad_norm": 2.4520530700683594,
      "learning_rate": 3.3519884119041353e-05,
      "loss": 2.602,
      "step": 12523
    },
    {
      "epoch": 0.3298393468527785,
      "grad_norm": 2.2900748252868652,
      "learning_rate": 3.351856728996576e-05,
      "loss": 1.3786,
      "step": 12524
    },
    {
      "epoch": 0.32986568343429024,
      "grad_norm": 2.832521438598633,
      "learning_rate": 3.351725046089018e-05,
      "loss": 1.1313,
      "step": 12525
    },
    {
      "epoch": 0.3298920200158019,
      "grad_norm": 1.9307314157485962,
      "learning_rate": 3.3515933631814593e-05,
      "loss": 0.5179,
      "step": 12526
    },
    {
      "epoch": 0.32991835659731367,
      "grad_norm": 1.689806580543518,
      "learning_rate": 3.351461680273901e-05,
      "loss": 1.8735,
      "step": 12527
    },
    {
      "epoch": 0.3299446931788254,
      "grad_norm": 1.8709040880203247,
      "learning_rate": 3.351329997366342e-05,
      "loss": 2.2646,
      "step": 12528
    },
    {
      "epoch": 0.3299710297603371,
      "grad_norm": 4.464748859405518,
      "learning_rate": 3.3511983144587833e-05,
      "loss": 1.6649,
      "step": 12529
    },
    {
      "epoch": 0.32999736634184884,
      "grad_norm": 5.140900611877441,
      "learning_rate": 3.351066631551225e-05,
      "loss": 1.8141,
      "step": 12530
    },
    {
      "epoch": 0.33002370292336053,
      "grad_norm": 2.42061710357666,
      "learning_rate": 3.3509349486436665e-05,
      "loss": 1.9346,
      "step": 12531
    },
    {
      "epoch": 0.3300500395048723,
      "grad_norm": 2.0868144035339355,
      "learning_rate": 3.350803265736108e-05,
      "loss": 2.2933,
      "step": 12532
    },
    {
      "epoch": 0.33007637608638396,
      "grad_norm": 1.780080795288086,
      "learning_rate": 3.350671582828549e-05,
      "loss": 2.1702,
      "step": 12533
    },
    {
      "epoch": 0.3301027126678957,
      "grad_norm": 1.5264103412628174,
      "learning_rate": 3.3505398999209905e-05,
      "loss": 1.8848,
      "step": 12534
    },
    {
      "epoch": 0.33012904924940745,
      "grad_norm": 1.7465996742248535,
      "learning_rate": 3.3504082170134313e-05,
      "loss": 2.0102,
      "step": 12535
    },
    {
      "epoch": 0.33015538583091913,
      "grad_norm": 1.7347049713134766,
      "learning_rate": 3.3502765341058736e-05,
      "loss": 2.2163,
      "step": 12536
    },
    {
      "epoch": 0.3301817224124309,
      "grad_norm": 6.342837810516357,
      "learning_rate": 3.3501448511983145e-05,
      "loss": 1.8344,
      "step": 12537
    },
    {
      "epoch": 0.33020805899394257,
      "grad_norm": 3.3314433097839355,
      "learning_rate": 3.350013168290756e-05,
      "loss": 1.1838,
      "step": 12538
    },
    {
      "epoch": 0.3302343955754543,
      "grad_norm": 1.7712961435317993,
      "learning_rate": 3.349881485383197e-05,
      "loss": 1.7312,
      "step": 12539
    },
    {
      "epoch": 0.33026073215696605,
      "grad_norm": 2.7138397693634033,
      "learning_rate": 3.349749802475639e-05,
      "loss": 2.2097,
      "step": 12540
    },
    {
      "epoch": 0.33028706873847774,
      "grad_norm": 1.8526287078857422,
      "learning_rate": 3.34961811956808e-05,
      "loss": 1.7366,
      "step": 12541
    },
    {
      "epoch": 0.3303134053199895,
      "grad_norm": 2.1880836486816406,
      "learning_rate": 3.3494864366605216e-05,
      "loss": 1.5198,
      "step": 12542
    },
    {
      "epoch": 0.33033974190150117,
      "grad_norm": 2.667555809020996,
      "learning_rate": 3.349354753752963e-05,
      "loss": 1.75,
      "step": 12543
    },
    {
      "epoch": 0.3303660784830129,
      "grad_norm": 3.640123128890991,
      "learning_rate": 3.349223070845404e-05,
      "loss": 2.2227,
      "step": 12544
    },
    {
      "epoch": 0.3303924150645246,
      "grad_norm": 3.579524517059326,
      "learning_rate": 3.349091387937846e-05,
      "loss": 1.6633,
      "step": 12545
    },
    {
      "epoch": 0.33041875164603635,
      "grad_norm": 2.2585041522979736,
      "learning_rate": 3.348959705030287e-05,
      "loss": 0.4713,
      "step": 12546
    },
    {
      "epoch": 0.3304450882275481,
      "grad_norm": 2.007910966873169,
      "learning_rate": 3.348828022122729e-05,
      "loss": 2.0308,
      "step": 12547
    },
    {
      "epoch": 0.3304714248090598,
      "grad_norm": 2.683708906173706,
      "learning_rate": 3.3486963392151696e-05,
      "loss": 1.2735,
      "step": 12548
    },
    {
      "epoch": 0.3304977613905715,
      "grad_norm": 2.4560959339141846,
      "learning_rate": 3.348564656307611e-05,
      "loss": 0.7222,
      "step": 12549
    },
    {
      "epoch": 0.3305240979720832,
      "grad_norm": 4.675262451171875,
      "learning_rate": 3.348432973400053e-05,
      "loss": 1.3754,
      "step": 12550
    },
    {
      "epoch": 0.33055043455359495,
      "grad_norm": 1.6472126245498657,
      "learning_rate": 3.348301290492494e-05,
      "loss": 1.6764,
      "step": 12551
    },
    {
      "epoch": 0.33057677113510664,
      "grad_norm": 4.038247108459473,
      "learning_rate": 3.348169607584936e-05,
      "loss": 1.5739,
      "step": 12552
    },
    {
      "epoch": 0.3306031077166184,
      "grad_norm": 2.0009679794311523,
      "learning_rate": 3.348037924677377e-05,
      "loss": 2.3286,
      "step": 12553
    },
    {
      "epoch": 0.3306294442981301,
      "grad_norm": 2.8953588008880615,
      "learning_rate": 3.347906241769819e-05,
      "loss": 1.573,
      "step": 12554
    },
    {
      "epoch": 0.3306557808796418,
      "grad_norm": 4.2981977462768555,
      "learning_rate": 3.34777455886226e-05,
      "loss": 1.5264,
      "step": 12555
    },
    {
      "epoch": 0.33068211746115356,
      "grad_norm": 1.6999446153640747,
      "learning_rate": 3.3476428759547014e-05,
      "loss": 0.4752,
      "step": 12556
    },
    {
      "epoch": 0.33070845404266525,
      "grad_norm": 1.5039056539535522,
      "learning_rate": 3.347511193047142e-05,
      "loss": 1.8579,
      "step": 12557
    },
    {
      "epoch": 0.330734790624177,
      "grad_norm": 1.9936712980270386,
      "learning_rate": 3.347379510139584e-05,
      "loss": 2.0047,
      "step": 12558
    },
    {
      "epoch": 0.3307611272056887,
      "grad_norm": 2.062723398208618,
      "learning_rate": 3.3472478272320254e-05,
      "loss": 2.0916,
      "step": 12559
    },
    {
      "epoch": 0.3307874637872004,
      "grad_norm": 2.9814114570617676,
      "learning_rate": 3.347116144324467e-05,
      "loss": 1.3651,
      "step": 12560
    },
    {
      "epoch": 0.33081380036871216,
      "grad_norm": 1.6131727695465088,
      "learning_rate": 3.3469844614169085e-05,
      "loss": 2.3453,
      "step": 12561
    },
    {
      "epoch": 0.33084013695022385,
      "grad_norm": 1.433284044265747,
      "learning_rate": 3.3468527785093494e-05,
      "loss": 2.0605,
      "step": 12562
    },
    {
      "epoch": 0.3308664735317356,
      "grad_norm": 1.5372118949890137,
      "learning_rate": 3.346721095601791e-05,
      "loss": 1.7518,
      "step": 12563
    },
    {
      "epoch": 0.3308928101132473,
      "grad_norm": 3.2360894680023193,
      "learning_rate": 3.3465894126942325e-05,
      "loss": 1.3049,
      "step": 12564
    },
    {
      "epoch": 0.330919146694759,
      "grad_norm": 1.6522250175476074,
      "learning_rate": 3.346457729786674e-05,
      "loss": 1.4266,
      "step": 12565
    },
    {
      "epoch": 0.3309454832762707,
      "grad_norm": 1.8428375720977783,
      "learning_rate": 3.346326046879115e-05,
      "loss": 2.0878,
      "step": 12566
    },
    {
      "epoch": 0.33097181985778246,
      "grad_norm": 2.827117681503296,
      "learning_rate": 3.3461943639715565e-05,
      "loss": 1.0205,
      "step": 12567
    },
    {
      "epoch": 0.3309981564392942,
      "grad_norm": 1.3901176452636719,
      "learning_rate": 3.346062681063998e-05,
      "loss": 1.7708,
      "step": 12568
    },
    {
      "epoch": 0.3310244930208059,
      "grad_norm": 1.8480952978134155,
      "learning_rate": 3.3459309981564396e-05,
      "loss": 1.5716,
      "step": 12569
    },
    {
      "epoch": 0.33105082960231763,
      "grad_norm": 3.0842599868774414,
      "learning_rate": 3.345799315248881e-05,
      "loss": 0.6009,
      "step": 12570
    },
    {
      "epoch": 0.3310771661838293,
      "grad_norm": 3.3772425651550293,
      "learning_rate": 3.345667632341322e-05,
      "loss": 1.7679,
      "step": 12571
    },
    {
      "epoch": 0.33110350276534106,
      "grad_norm": 2.100574016571045,
      "learning_rate": 3.3455359494337636e-05,
      "loss": 1.5663,
      "step": 12572
    },
    {
      "epoch": 0.3311298393468528,
      "grad_norm": 1.4141217470169067,
      "learning_rate": 3.345404266526205e-05,
      "loss": 1.4185,
      "step": 12573
    },
    {
      "epoch": 0.3311561759283645,
      "grad_norm": 1.649835467338562,
      "learning_rate": 3.345272583618647e-05,
      "loss": 2.0746,
      "step": 12574
    },
    {
      "epoch": 0.33118251250987624,
      "grad_norm": 1.8346282243728638,
      "learning_rate": 3.3451409007110876e-05,
      "loss": 1.5286,
      "step": 12575
    },
    {
      "epoch": 0.3312088490913879,
      "grad_norm": 1.6538931131362915,
      "learning_rate": 3.345009217803529e-05,
      "loss": 2.0253,
      "step": 12576
    },
    {
      "epoch": 0.33123518567289967,
      "grad_norm": 2.7198596000671387,
      "learning_rate": 3.344877534895971e-05,
      "loss": 1.4781,
      "step": 12577
    },
    {
      "epoch": 0.33126152225441136,
      "grad_norm": 2.6430201530456543,
      "learning_rate": 3.344745851988412e-05,
      "loss": 1.7649,
      "step": 12578
    },
    {
      "epoch": 0.3312878588359231,
      "grad_norm": 2.3806519508361816,
      "learning_rate": 3.344614169080854e-05,
      "loss": 1.5382,
      "step": 12579
    },
    {
      "epoch": 0.33131419541743484,
      "grad_norm": 1.9562166929244995,
      "learning_rate": 3.344482486173295e-05,
      "loss": 1.769,
      "step": 12580
    },
    {
      "epoch": 0.33134053199894653,
      "grad_norm": 2.0368337631225586,
      "learning_rate": 3.344350803265736e-05,
      "loss": 1.2347,
      "step": 12581
    },
    {
      "epoch": 0.3313668685804583,
      "grad_norm": 1.9362303018569946,
      "learning_rate": 3.344219120358177e-05,
      "loss": 1.8617,
      "step": 12582
    },
    {
      "epoch": 0.33139320516196996,
      "grad_norm": 2.699650764465332,
      "learning_rate": 3.3440874374506194e-05,
      "loss": 1.3428,
      "step": 12583
    },
    {
      "epoch": 0.3314195417434817,
      "grad_norm": 1.5381124019622803,
      "learning_rate": 3.34395575454306e-05,
      "loss": 2.0631,
      "step": 12584
    },
    {
      "epoch": 0.3314458783249934,
      "grad_norm": 2.6243417263031006,
      "learning_rate": 3.343824071635502e-05,
      "loss": 1.845,
      "step": 12585
    },
    {
      "epoch": 0.33147221490650514,
      "grad_norm": 2.563018798828125,
      "learning_rate": 3.3436923887279434e-05,
      "loss": 1.5551,
      "step": 12586
    },
    {
      "epoch": 0.3314985514880169,
      "grad_norm": 1.8089511394500732,
      "learning_rate": 3.343560705820385e-05,
      "loss": 2.6125,
      "step": 12587
    },
    {
      "epoch": 0.33152488806952857,
      "grad_norm": 1.701754093170166,
      "learning_rate": 3.3434290229128266e-05,
      "loss": 1.2412,
      "step": 12588
    },
    {
      "epoch": 0.3315512246510403,
      "grad_norm": 1.690680742263794,
      "learning_rate": 3.3432973400052674e-05,
      "loss": 0.5706,
      "step": 12589
    },
    {
      "epoch": 0.331577561232552,
      "grad_norm": 3.051246404647827,
      "learning_rate": 3.343165657097709e-05,
      "loss": 1.0539,
      "step": 12590
    },
    {
      "epoch": 0.33160389781406374,
      "grad_norm": 5.108091354370117,
      "learning_rate": 3.34303397419015e-05,
      "loss": 0.8685,
      "step": 12591
    },
    {
      "epoch": 0.33163023439557543,
      "grad_norm": 1.8559662103652954,
      "learning_rate": 3.342902291282592e-05,
      "loss": 2.1733,
      "step": 12592
    },
    {
      "epoch": 0.3316565709770872,
      "grad_norm": 1.5760260820388794,
      "learning_rate": 3.342770608375033e-05,
      "loss": 2.0728,
      "step": 12593
    },
    {
      "epoch": 0.3316829075585989,
      "grad_norm": 1.580976128578186,
      "learning_rate": 3.3426389254674746e-05,
      "loss": 1.4601,
      "step": 12594
    },
    {
      "epoch": 0.3317092441401106,
      "grad_norm": 1.7760361433029175,
      "learning_rate": 3.3425072425599154e-05,
      "loss": 0.6615,
      "step": 12595
    },
    {
      "epoch": 0.33173558072162235,
      "grad_norm": 1.6805483102798462,
      "learning_rate": 3.342375559652357e-05,
      "loss": 1.7424,
      "step": 12596
    },
    {
      "epoch": 0.33176191730313404,
      "grad_norm": 2.085524082183838,
      "learning_rate": 3.3422438767447986e-05,
      "loss": 1.8532,
      "step": 12597
    },
    {
      "epoch": 0.3317882538846458,
      "grad_norm": 1.7246835231781006,
      "learning_rate": 3.34211219383724e-05,
      "loss": 1.7175,
      "step": 12598
    },
    {
      "epoch": 0.33181459046615747,
      "grad_norm": 2.244180679321289,
      "learning_rate": 3.341980510929682e-05,
      "loss": 0.5694,
      "step": 12599
    },
    {
      "epoch": 0.3318409270476692,
      "grad_norm": 5.295844554901123,
      "learning_rate": 3.3418488280221226e-05,
      "loss": 2.1871,
      "step": 12600
    },
    {
      "epoch": 0.33186726362918095,
      "grad_norm": 1.8949546813964844,
      "learning_rate": 3.341717145114564e-05,
      "loss": 1.3858,
      "step": 12601
    },
    {
      "epoch": 0.33189360021069264,
      "grad_norm": 1.990843415260315,
      "learning_rate": 3.341585462207006e-05,
      "loss": 1.1522,
      "step": 12602
    },
    {
      "epoch": 0.3319199367922044,
      "grad_norm": NaN,
      "learning_rate": 3.341585462207006e-05,
      "loss": 2.7428,
      "step": 12603
    },
    {
      "epoch": 0.33194627337371607,
      "grad_norm": 3.1444976329803467,
      "learning_rate": 3.341453779299447e-05,
      "loss": 0.9988,
      "step": 12604
    },
    {
      "epoch": 0.3319726099552278,
      "grad_norm": 2.7017078399658203,
      "learning_rate": 3.341322096391888e-05,
      "loss": 1.7436,
      "step": 12605
    },
    {
      "epoch": 0.33199894653673956,
      "grad_norm": 1.9363417625427246,
      "learning_rate": 3.34119041348433e-05,
      "loss": 1.3717,
      "step": 12606
    },
    {
      "epoch": 0.33202528311825125,
      "grad_norm": 2.046461820602417,
      "learning_rate": 3.341058730576771e-05,
      "loss": 1.6626,
      "step": 12607
    },
    {
      "epoch": 0.332051619699763,
      "grad_norm": 2.164912462234497,
      "learning_rate": 3.340927047669213e-05,
      "loss": 1.0039,
      "step": 12608
    },
    {
      "epoch": 0.3320779562812747,
      "grad_norm": 1.5574190616607666,
      "learning_rate": 3.3407953647616544e-05,
      "loss": 2.043,
      "step": 12609
    },
    {
      "epoch": 0.3321042928627864,
      "grad_norm": 2.184595823287964,
      "learning_rate": 3.340663681854095e-05,
      "loss": 2.1146,
      "step": 12610
    },
    {
      "epoch": 0.3321306294442981,
      "grad_norm": 4.399532794952393,
      "learning_rate": 3.340531998946537e-05,
      "loss": 0.994,
      "step": 12611
    },
    {
      "epoch": 0.33215696602580985,
      "grad_norm": 2.892519950866699,
      "learning_rate": 3.3404003160389784e-05,
      "loss": 1.4885,
      "step": 12612
    },
    {
      "epoch": 0.3321833026073216,
      "grad_norm": 2.8012919425964355,
      "learning_rate": 3.34026863313142e-05,
      "loss": 1.4831,
      "step": 12613
    },
    {
      "epoch": 0.3322096391888333,
      "grad_norm": 2.3918423652648926,
      "learning_rate": 3.340136950223861e-05,
      "loss": 1.512,
      "step": 12614
    },
    {
      "epoch": 0.332235975770345,
      "grad_norm": 2.1756582260131836,
      "learning_rate": 3.3400052673163024e-05,
      "loss": 0.7221,
      "step": 12615
    },
    {
      "epoch": 0.3322623123518567,
      "grad_norm": 1.7098534107208252,
      "learning_rate": 3.339873584408744e-05,
      "loss": 2.1871,
      "step": 12616
    },
    {
      "epoch": 0.33228864893336846,
      "grad_norm": 2.1105356216430664,
      "learning_rate": 3.3397419015011855e-05,
      "loss": 2.1751,
      "step": 12617
    },
    {
      "epoch": 0.33231498551488015,
      "grad_norm": 1.6752023696899414,
      "learning_rate": 3.339610218593627e-05,
      "loss": 1.6417,
      "step": 12618
    },
    {
      "epoch": 0.3323413220963919,
      "grad_norm": 3.664382219314575,
      "learning_rate": 3.339478535686068e-05,
      "loss": 1.3972,
      "step": 12619
    },
    {
      "epoch": 0.33236765867790363,
      "grad_norm": 1.8466389179229736,
      "learning_rate": 3.3393468527785095e-05,
      "loss": 1.7655,
      "step": 12620
    },
    {
      "epoch": 0.3323939952594153,
      "grad_norm": 2.00946307182312,
      "learning_rate": 3.339215169870951e-05,
      "loss": 2.0125,
      "step": 12621
    },
    {
      "epoch": 0.33242033184092706,
      "grad_norm": 2.6447982788085938,
      "learning_rate": 3.3390834869633926e-05,
      "loss": 0.6782,
      "step": 12622
    },
    {
      "epoch": 0.33244666842243875,
      "grad_norm": 2.420147180557251,
      "learning_rate": 3.3389518040558335e-05,
      "loss": 1.4998,
      "step": 12623
    },
    {
      "epoch": 0.3324730050039505,
      "grad_norm": 3.5737648010253906,
      "learning_rate": 3.338820121148275e-05,
      "loss": 2.6596,
      "step": 12624
    },
    {
      "epoch": 0.3324993415854622,
      "grad_norm": 3.727470874786377,
      "learning_rate": 3.3386884382407166e-05,
      "loss": 1.3438,
      "step": 12625
    },
    {
      "epoch": 0.3325256781669739,
      "grad_norm": 1.6507397890090942,
      "learning_rate": 3.338556755333158e-05,
      "loss": 0.3542,
      "step": 12626
    },
    {
      "epoch": 0.33255201474848567,
      "grad_norm": 1.7301124334335327,
      "learning_rate": 3.3384250724256e-05,
      "loss": 2.1822,
      "step": 12627
    },
    {
      "epoch": 0.33257835132999736,
      "grad_norm": 3.38923716545105,
      "learning_rate": 3.3382933895180406e-05,
      "loss": 0.9428,
      "step": 12628
    },
    {
      "epoch": 0.3326046879115091,
      "grad_norm": 3.6878015995025635,
      "learning_rate": 3.338161706610482e-05,
      "loss": 1.372,
      "step": 12629
    },
    {
      "epoch": 0.3326310244930208,
      "grad_norm": 1.808009147644043,
      "learning_rate": 3.338030023702923e-05,
      "loss": 1.4246,
      "step": 12630
    },
    {
      "epoch": 0.33265736107453253,
      "grad_norm": 1.6601111888885498,
      "learning_rate": 3.337898340795365e-05,
      "loss": 2.0032,
      "step": 12631
    },
    {
      "epoch": 0.3326836976560442,
      "grad_norm": 1.7675055265426636,
      "learning_rate": 3.337766657887806e-05,
      "loss": 1.874,
      "step": 12632
    },
    {
      "epoch": 0.33271003423755596,
      "grad_norm": 1.8963303565979004,
      "learning_rate": 3.337634974980248e-05,
      "loss": 1.7249,
      "step": 12633
    },
    {
      "epoch": 0.3327363708190677,
      "grad_norm": 2.998598337173462,
      "learning_rate": 3.337503292072689e-05,
      "loss": 1.7926,
      "step": 12634
    },
    {
      "epoch": 0.3327627074005794,
      "grad_norm": 2.936476945877075,
      "learning_rate": 3.33737160916513e-05,
      "loss": 1.5406,
      "step": 12635
    },
    {
      "epoch": 0.33278904398209114,
      "grad_norm": 2.736581325531006,
      "learning_rate": 3.3372399262575724e-05,
      "loss": 1.3174,
      "step": 12636
    },
    {
      "epoch": 0.3328153805636028,
      "grad_norm": 3.6287879943847656,
      "learning_rate": 3.337108243350013e-05,
      "loss": 1.6855,
      "step": 12637
    },
    {
      "epoch": 0.33284171714511457,
      "grad_norm": 1.944210171699524,
      "learning_rate": 3.336976560442455e-05,
      "loss": 1.9133,
      "step": 12638
    },
    {
      "epoch": 0.33286805372662626,
      "grad_norm": 1.5569289922714233,
      "learning_rate": 3.336844877534896e-05,
      "loss": 1.8468,
      "step": 12639
    },
    {
      "epoch": 0.332894390308138,
      "grad_norm": 1.9692211151123047,
      "learning_rate": 3.336713194627338e-05,
      "loss": 1.8621,
      "step": 12640
    },
    {
      "epoch": 0.33292072688964974,
      "grad_norm": 4.389628887176514,
      "learning_rate": 3.336581511719779e-05,
      "loss": 0.577,
      "step": 12641
    },
    {
      "epoch": 0.33294706347116143,
      "grad_norm": 2.817185163497925,
      "learning_rate": 3.3364498288122204e-05,
      "loss": 1.8293,
      "step": 12642
    },
    {
      "epoch": 0.3329734000526732,
      "grad_norm": 2.639174222946167,
      "learning_rate": 3.336318145904661e-05,
      "loss": 1.4784,
      "step": 12643
    },
    {
      "epoch": 0.33299973663418486,
      "grad_norm": 2.9981637001037598,
      "learning_rate": 3.336186462997103e-05,
      "loss": 2.342,
      "step": 12644
    },
    {
      "epoch": 0.3330260732156966,
      "grad_norm": 1.6784089803695679,
      "learning_rate": 3.3360547800895444e-05,
      "loss": 1.8669,
      "step": 12645
    },
    {
      "epoch": 0.33305240979720835,
      "grad_norm": 1.9349316358566284,
      "learning_rate": 3.335923097181986e-05,
      "loss": 1.801,
      "step": 12646
    },
    {
      "epoch": 0.33307874637872004,
      "grad_norm": 1.6820112466812134,
      "learning_rate": 3.3357914142744275e-05,
      "loss": 1.9513,
      "step": 12647
    },
    {
      "epoch": 0.3331050829602318,
      "grad_norm": 2.608950614929199,
      "learning_rate": 3.3356597313668684e-05,
      "loss": 1.0613,
      "step": 12648
    },
    {
      "epoch": 0.33313141954174347,
      "grad_norm": 2.393916606903076,
      "learning_rate": 3.33552804845931e-05,
      "loss": 1.0479,
      "step": 12649
    },
    {
      "epoch": 0.3331577561232552,
      "grad_norm": 2.3476123809814453,
      "learning_rate": 3.3353963655517515e-05,
      "loss": 2.2433,
      "step": 12650
    },
    {
      "epoch": 0.3331840927047669,
      "grad_norm": 1.54328191280365,
      "learning_rate": 3.335264682644193e-05,
      "loss": 2.1673,
      "step": 12651
    },
    {
      "epoch": 0.33321042928627864,
      "grad_norm": 1.5280935764312744,
      "learning_rate": 3.335132999736634e-05,
      "loss": 1.7752,
      "step": 12652
    },
    {
      "epoch": 0.3332367658677904,
      "grad_norm": 2.336001396179199,
      "learning_rate": 3.3350013168290755e-05,
      "loss": 1.8214,
      "step": 12653
    },
    {
      "epoch": 0.3332631024493021,
      "grad_norm": 2.140209436416626,
      "learning_rate": 3.334869633921517e-05,
      "loss": 1.7811,
      "step": 12654
    },
    {
      "epoch": 0.3332894390308138,
      "grad_norm": 2.073690891265869,
      "learning_rate": 3.334737951013959e-05,
      "loss": 2.1738,
      "step": 12655
    },
    {
      "epoch": 0.3333157756123255,
      "grad_norm": 3.458463430404663,
      "learning_rate": 3.3346062681064e-05,
      "loss": 1.7545,
      "step": 12656
    },
    {
      "epoch": 0.33334211219383725,
      "grad_norm": 2.605024814605713,
      "learning_rate": 3.334474585198841e-05,
      "loss": 1.2952,
      "step": 12657
    },
    {
      "epoch": 0.33336844877534894,
      "grad_norm": 2.0049028396606445,
      "learning_rate": 3.334342902291283e-05,
      "loss": 1.8393,
      "step": 12658
    },
    {
      "epoch": 0.3333947853568607,
      "grad_norm": 2.1910781860351562,
      "learning_rate": 3.334211219383724e-05,
      "loss": 1.6451,
      "step": 12659
    },
    {
      "epoch": 0.3334211219383724,
      "grad_norm": 1.8754687309265137,
      "learning_rate": 3.334079536476166e-05,
      "loss": 2.0493,
      "step": 12660
    },
    {
      "epoch": 0.3334474585198841,
      "grad_norm": 1.9820432662963867,
      "learning_rate": 3.333947853568607e-05,
      "loss": 1.6025,
      "step": 12661
    },
    {
      "epoch": 0.33347379510139585,
      "grad_norm": 1.8536455631256104,
      "learning_rate": 3.333816170661048e-05,
      "loss": 1.8685,
      "step": 12662
    },
    {
      "epoch": 0.33350013168290754,
      "grad_norm": 1.7496771812438965,
      "learning_rate": 3.33368448775349e-05,
      "loss": 1.4563,
      "step": 12663
    },
    {
      "epoch": 0.3335264682644193,
      "grad_norm": 1.5448086261749268,
      "learning_rate": 3.3335528048459313e-05,
      "loss": 2.2682,
      "step": 12664
    },
    {
      "epoch": 0.333552804845931,
      "grad_norm": 1.804081916809082,
      "learning_rate": 3.333421121938373e-05,
      "loss": 1.9859,
      "step": 12665
    },
    {
      "epoch": 0.3335791414274427,
      "grad_norm": 2.6542768478393555,
      "learning_rate": 3.333289439030814e-05,
      "loss": 2.5179,
      "step": 12666
    },
    {
      "epoch": 0.33360547800895446,
      "grad_norm": 3.9531445503234863,
      "learning_rate": 3.3331577561232553e-05,
      "loss": 1.6019,
      "step": 12667
    },
    {
      "epoch": 0.33363181459046615,
      "grad_norm": 2.3091623783111572,
      "learning_rate": 3.333026073215696e-05,
      "loss": 1.3007,
      "step": 12668
    },
    {
      "epoch": 0.3336581511719779,
      "grad_norm": 2.5285756587982178,
      "learning_rate": 3.3328943903081385e-05,
      "loss": 2.961,
      "step": 12669
    },
    {
      "epoch": 0.3336844877534896,
      "grad_norm": 2.592228889465332,
      "learning_rate": 3.3327627074005793e-05,
      "loss": 1.9922,
      "step": 12670
    },
    {
      "epoch": 0.3337108243350013,
      "grad_norm": 1.453342080116272,
      "learning_rate": 3.332631024493021e-05,
      "loss": 1.1381,
      "step": 12671
    },
    {
      "epoch": 0.333737160916513,
      "grad_norm": 4.903116703033447,
      "learning_rate": 3.3324993415854625e-05,
      "loss": 1.6986,
      "step": 12672
    },
    {
      "epoch": 0.33376349749802475,
      "grad_norm": 2.249630928039551,
      "learning_rate": 3.332367658677904e-05,
      "loss": 1.9945,
      "step": 12673
    },
    {
      "epoch": 0.3337898340795365,
      "grad_norm": 3.2702884674072266,
      "learning_rate": 3.3322359757703456e-05,
      "loss": 2.5891,
      "step": 12674
    },
    {
      "epoch": 0.3338161706610482,
      "grad_norm": 2.378082752227783,
      "learning_rate": 3.3321042928627865e-05,
      "loss": 2.0394,
      "step": 12675
    },
    {
      "epoch": 0.3338425072425599,
      "grad_norm": 1.7774821519851685,
      "learning_rate": 3.331972609955228e-05,
      "loss": 0.6657,
      "step": 12676
    },
    {
      "epoch": 0.3338688438240716,
      "grad_norm": 1.9021217823028564,
      "learning_rate": 3.331840927047669e-05,
      "loss": 1.6685,
      "step": 12677
    },
    {
      "epoch": 0.33389518040558336,
      "grad_norm": 1.69480562210083,
      "learning_rate": 3.331709244140111e-05,
      "loss": 0.2684,
      "step": 12678
    },
    {
      "epoch": 0.3339215169870951,
      "grad_norm": 1.8669898509979248,
      "learning_rate": 3.331577561232552e-05,
      "loss": 0.7269,
      "step": 12679
    },
    {
      "epoch": 0.3339478535686068,
      "grad_norm": 1.9938433170318604,
      "learning_rate": 3.3314458783249936e-05,
      "loss": 1.1318,
      "step": 12680
    },
    {
      "epoch": 0.33397419015011853,
      "grad_norm": 1.5870530605316162,
      "learning_rate": 3.331314195417435e-05,
      "loss": 1.9258,
      "step": 12681
    },
    {
      "epoch": 0.3340005267316302,
      "grad_norm": 1.573981523513794,
      "learning_rate": 3.331182512509876e-05,
      "loss": 2.4933,
      "step": 12682
    },
    {
      "epoch": 0.33402686331314196,
      "grad_norm": 2.3521769046783447,
      "learning_rate": 3.331050829602318e-05,
      "loss": 0.5474,
      "step": 12683
    },
    {
      "epoch": 0.33405319989465365,
      "grad_norm": 3.06538462638855,
      "learning_rate": 3.330919146694759e-05,
      "loss": 2.3508,
      "step": 12684
    },
    {
      "epoch": 0.3340795364761654,
      "grad_norm": 2.0984601974487305,
      "learning_rate": 3.330787463787201e-05,
      "loss": 2.1414,
      "step": 12685
    },
    {
      "epoch": 0.33410587305767714,
      "grad_norm": 2.98837947845459,
      "learning_rate": 3.3306557808796416e-05,
      "loss": 1.2802,
      "step": 12686
    },
    {
      "epoch": 0.3341322096391888,
      "grad_norm": 2.247215747833252,
      "learning_rate": 3.330524097972084e-05,
      "loss": 1.7084,
      "step": 12687
    },
    {
      "epoch": 0.33415854622070057,
      "grad_norm": 3.222853899002075,
      "learning_rate": 3.330392415064525e-05,
      "loss": 1.945,
      "step": 12688
    },
    {
      "epoch": 0.33418488280221226,
      "grad_norm": 1.94883394241333,
      "learning_rate": 3.330260732156966e-05,
      "loss": 1.546,
      "step": 12689
    },
    {
      "epoch": 0.334211219383724,
      "grad_norm": 3.241529941558838,
      "learning_rate": 3.330129049249408e-05,
      "loss": 1.9507,
      "step": 12690
    },
    {
      "epoch": 0.3342375559652357,
      "grad_norm": 2.0178792476654053,
      "learning_rate": 3.329997366341849e-05,
      "loss": 1.9945,
      "step": 12691
    },
    {
      "epoch": 0.33426389254674743,
      "grad_norm": 3.6998403072357178,
      "learning_rate": 3.329865683434291e-05,
      "loss": 1.37,
      "step": 12692
    },
    {
      "epoch": 0.3342902291282592,
      "grad_norm": 2.0556843280792236,
      "learning_rate": 3.329734000526732e-05,
      "loss": 2.6426,
      "step": 12693
    },
    {
      "epoch": 0.33431656570977086,
      "grad_norm": 1.4851425886154175,
      "learning_rate": 3.3296023176191734e-05,
      "loss": 1.4306,
      "step": 12694
    },
    {
      "epoch": 0.3343429022912826,
      "grad_norm": 1.6809499263763428,
      "learning_rate": 3.329470634711614e-05,
      "loss": 1.8683,
      "step": 12695
    },
    {
      "epoch": 0.3343692388727943,
      "grad_norm": 2.6594858169555664,
      "learning_rate": 3.329338951804056e-05,
      "loss": 0.3499,
      "step": 12696
    },
    {
      "epoch": 0.33439557545430604,
      "grad_norm": 2.2146317958831787,
      "learning_rate": 3.3292072688964974e-05,
      "loss": 2.2095,
      "step": 12697
    },
    {
      "epoch": 0.3344219120358177,
      "grad_norm": 2.6948370933532715,
      "learning_rate": 3.329075585988939e-05,
      "loss": 1.5676,
      "step": 12698
    },
    {
      "epoch": 0.33444824861732947,
      "grad_norm": 2.5437676906585693,
      "learning_rate": 3.32894390308138e-05,
      "loss": 2.3722,
      "step": 12699
    },
    {
      "epoch": 0.3344745851988412,
      "grad_norm": 1.865189790725708,
      "learning_rate": 3.3288122201738214e-05,
      "loss": 0.5159,
      "step": 12700
    },
    {
      "epoch": 0.3345009217803529,
      "grad_norm": 1.8187694549560547,
      "learning_rate": 3.328680537266263e-05,
      "loss": 1.9236,
      "step": 12701
    },
    {
      "epoch": 0.33452725836186464,
      "grad_norm": 1.5275969505310059,
      "learning_rate": 3.3285488543587045e-05,
      "loss": 1.9205,
      "step": 12702
    },
    {
      "epoch": 0.33455359494337633,
      "grad_norm": 1.937986135482788,
      "learning_rate": 3.328417171451146e-05,
      "loss": 0.8848,
      "step": 12703
    },
    {
      "epoch": 0.3345799315248881,
      "grad_norm": 3.11193585395813,
      "learning_rate": 3.328285488543587e-05,
      "loss": 1.464,
      "step": 12704
    },
    {
      "epoch": 0.33460626810639976,
      "grad_norm": 2.118119716644287,
      "learning_rate": 3.3281538056360285e-05,
      "loss": 2.4367,
      "step": 12705
    },
    {
      "epoch": 0.3346326046879115,
      "grad_norm": 1.848755121231079,
      "learning_rate": 3.32802212272847e-05,
      "loss": 1.8896,
      "step": 12706
    },
    {
      "epoch": 0.33465894126942325,
      "grad_norm": 1.2457447052001953,
      "learning_rate": 3.3278904398209116e-05,
      "loss": 1.2816,
      "step": 12707
    },
    {
      "epoch": 0.33468527785093494,
      "grad_norm": 2.7649874687194824,
      "learning_rate": 3.3277587569133525e-05,
      "loss": 1.7536,
      "step": 12708
    },
    {
      "epoch": 0.3347116144324467,
      "grad_norm": 2.5353243350982666,
      "learning_rate": 3.327627074005794e-05,
      "loss": 0.3203,
      "step": 12709
    },
    {
      "epoch": 0.33473795101395837,
      "grad_norm": 1.5999507904052734,
      "learning_rate": 3.3274953910982356e-05,
      "loss": 1.6913,
      "step": 12710
    },
    {
      "epoch": 0.3347642875954701,
      "grad_norm": 2.764409303665161,
      "learning_rate": 3.327363708190677e-05,
      "loss": 1.5894,
      "step": 12711
    },
    {
      "epoch": 0.33479062417698185,
      "grad_norm": 2.7095680236816406,
      "learning_rate": 3.327232025283119e-05,
      "loss": 0.4185,
      "step": 12712
    },
    {
      "epoch": 0.33481696075849354,
      "grad_norm": 2.2557263374328613,
      "learning_rate": 3.3271003423755596e-05,
      "loss": 0.5209,
      "step": 12713
    },
    {
      "epoch": 0.3348432973400053,
      "grad_norm": 2.720989942550659,
      "learning_rate": 3.326968659468001e-05,
      "loss": 0.6882,
      "step": 12714
    },
    {
      "epoch": 0.334869633921517,
      "grad_norm": 1.8580718040466309,
      "learning_rate": 3.326836976560442e-05,
      "loss": 1.4242,
      "step": 12715
    },
    {
      "epoch": 0.3348959705030287,
      "grad_norm": 2.1879920959472656,
      "learning_rate": 3.326705293652884e-05,
      "loss": 0.566,
      "step": 12716
    },
    {
      "epoch": 0.3349223070845404,
      "grad_norm": 1.8712615966796875,
      "learning_rate": 3.326573610745325e-05,
      "loss": 1.7497,
      "step": 12717
    },
    {
      "epoch": 0.33494864366605215,
      "grad_norm": 3.7341501712799072,
      "learning_rate": 3.326441927837767e-05,
      "loss": 0.9295,
      "step": 12718
    },
    {
      "epoch": 0.3349749802475639,
      "grad_norm": 2.814065933227539,
      "learning_rate": 3.326310244930208e-05,
      "loss": 1.5899,
      "step": 12719
    },
    {
      "epoch": 0.3350013168290756,
      "grad_norm": 1.93490469455719,
      "learning_rate": 3.32617856202265e-05,
      "loss": 2.1805,
      "step": 12720
    },
    {
      "epoch": 0.3350276534105873,
      "grad_norm": 2.3104476928710938,
      "learning_rate": 3.3260468791150914e-05,
      "loss": 1.876,
      "step": 12721
    },
    {
      "epoch": 0.335053989992099,
      "grad_norm": 2.130648136138916,
      "learning_rate": 3.325915196207532e-05,
      "loss": 1.8649,
      "step": 12722
    },
    {
      "epoch": 0.33508032657361075,
      "grad_norm": 1.7028783559799194,
      "learning_rate": 3.325783513299974e-05,
      "loss": 2.3934,
      "step": 12723
    },
    {
      "epoch": 0.33510666315512244,
      "grad_norm": 1.3325281143188477,
      "learning_rate": 3.325651830392415e-05,
      "loss": 0.2996,
      "step": 12724
    },
    {
      "epoch": 0.3351329997366342,
      "grad_norm": 2.4049007892608643,
      "learning_rate": 3.325520147484857e-05,
      "loss": 2.0552,
      "step": 12725
    },
    {
      "epoch": 0.33515933631814593,
      "grad_norm": 3.0834882259368896,
      "learning_rate": 3.325388464577298e-05,
      "loss": 2.4281,
      "step": 12726
    },
    {
      "epoch": 0.3351856728996576,
      "grad_norm": 1.9483907222747803,
      "learning_rate": 3.3252567816697394e-05,
      "loss": 1.7043,
      "step": 12727
    },
    {
      "epoch": 0.33521200948116936,
      "grad_norm": 1.6540095806121826,
      "learning_rate": 3.325125098762181e-05,
      "loss": 1.7751,
      "step": 12728
    },
    {
      "epoch": 0.33523834606268105,
      "grad_norm": 2.1088151931762695,
      "learning_rate": 3.324993415854622e-05,
      "loss": 1.3019,
      "step": 12729
    },
    {
      "epoch": 0.3352646826441928,
      "grad_norm": 3.1819138526916504,
      "learning_rate": 3.324861732947064e-05,
      "loss": 1.6342,
      "step": 12730
    },
    {
      "epoch": 0.3352910192257045,
      "grad_norm": 1.47464919090271,
      "learning_rate": 3.324730050039505e-05,
      "loss": 0.8946,
      "step": 12731
    },
    {
      "epoch": 0.3353173558072162,
      "grad_norm": 2.3514928817749023,
      "learning_rate": 3.3245983671319466e-05,
      "loss": 1.2741,
      "step": 12732
    },
    {
      "epoch": 0.33534369238872797,
      "grad_norm": 1.9602787494659424,
      "learning_rate": 3.3244666842243874e-05,
      "loss": 1.7248,
      "step": 12733
    },
    {
      "epoch": 0.33537002897023965,
      "grad_norm": 3.0614047050476074,
      "learning_rate": 3.32433500131683e-05,
      "loss": 1.7016,
      "step": 12734
    },
    {
      "epoch": 0.3353963655517514,
      "grad_norm": 2.4778575897216797,
      "learning_rate": 3.3242033184092706e-05,
      "loss": 1.4643,
      "step": 12735
    },
    {
      "epoch": 0.3354227021332631,
      "grad_norm": 2.4843504428863525,
      "learning_rate": 3.324071635501712e-05,
      "loss": 1.8201,
      "step": 12736
    },
    {
      "epoch": 0.3354490387147748,
      "grad_norm": 3.3799288272857666,
      "learning_rate": 3.323939952594154e-05,
      "loss": 1.8345,
      "step": 12737
    },
    {
      "epoch": 0.3354753752962865,
      "grad_norm": 3.314253330230713,
      "learning_rate": 3.3238082696865946e-05,
      "loss": 0.7715,
      "step": 12738
    },
    {
      "epoch": 0.33550171187779826,
      "grad_norm": 2.94140887260437,
      "learning_rate": 3.323676586779037e-05,
      "loss": 1.6291,
      "step": 12739
    },
    {
      "epoch": 0.33552804845931,
      "grad_norm": 1.8520805835723877,
      "learning_rate": 3.323544903871478e-05,
      "loss": 1.6528,
      "step": 12740
    },
    {
      "epoch": 0.3355543850408217,
      "grad_norm": 2.0638961791992188,
      "learning_rate": 3.323413220963919e-05,
      "loss": 1.5233,
      "step": 12741
    },
    {
      "epoch": 0.33558072162233343,
      "grad_norm": 3.3796169757843018,
      "learning_rate": 3.32328153805636e-05,
      "loss": 1.8255,
      "step": 12742
    },
    {
      "epoch": 0.3356070582038451,
      "grad_norm": 2.9376111030578613,
      "learning_rate": 3.323149855148802e-05,
      "loss": 1.4259,
      "step": 12743
    },
    {
      "epoch": 0.33563339478535686,
      "grad_norm": 2.060811758041382,
      "learning_rate": 3.323018172241243e-05,
      "loss": 1.9522,
      "step": 12744
    },
    {
      "epoch": 0.3356597313668686,
      "grad_norm": 2.9156436920166016,
      "learning_rate": 3.322886489333685e-05,
      "loss": 0.3796,
      "step": 12745
    },
    {
      "epoch": 0.3356860679483803,
      "grad_norm": 1.841123104095459,
      "learning_rate": 3.322754806426126e-05,
      "loss": 1.648,
      "step": 12746
    },
    {
      "epoch": 0.33571240452989204,
      "grad_norm": 3.8256890773773193,
      "learning_rate": 3.322623123518567e-05,
      "loss": 1.8899,
      "step": 12747
    },
    {
      "epoch": 0.3357387411114037,
      "grad_norm": 2.747293472290039,
      "learning_rate": 3.322491440611009e-05,
      "loss": 1.8156,
      "step": 12748
    },
    {
      "epoch": 0.33576507769291547,
      "grad_norm": 2.571321964263916,
      "learning_rate": 3.3223597577034504e-05,
      "loss": 2.0543,
      "step": 12749
    },
    {
      "epoch": 0.33579141427442716,
      "grad_norm": 3.315554618835449,
      "learning_rate": 3.322228074795892e-05,
      "loss": 0.8903,
      "step": 12750
    },
    {
      "epoch": 0.3358177508559389,
      "grad_norm": 1.8580601215362549,
      "learning_rate": 3.322096391888333e-05,
      "loss": 2.3449,
      "step": 12751
    },
    {
      "epoch": 0.33584408743745064,
      "grad_norm": 1.6815662384033203,
      "learning_rate": 3.3219647089807744e-05,
      "loss": 1.8054,
      "step": 12752
    },
    {
      "epoch": 0.33587042401896233,
      "grad_norm": 1.826073169708252,
      "learning_rate": 3.321833026073216e-05,
      "loss": 1.8914,
      "step": 12753
    },
    {
      "epoch": 0.3358967606004741,
      "grad_norm": 4.068779468536377,
      "learning_rate": 3.3217013431656575e-05,
      "loss": 1.1452,
      "step": 12754
    },
    {
      "epoch": 0.33592309718198576,
      "grad_norm": 2.070976495742798,
      "learning_rate": 3.3215696602580984e-05,
      "loss": 2.1186,
      "step": 12755
    },
    {
      "epoch": 0.3359494337634975,
      "grad_norm": 2.1486213207244873,
      "learning_rate": 3.32143797735054e-05,
      "loss": 2.2322,
      "step": 12756
    },
    {
      "epoch": 0.3359757703450092,
      "grad_norm": 1.9530872106552124,
      "learning_rate": 3.3213062944429815e-05,
      "loss": 1.9658,
      "step": 12757
    },
    {
      "epoch": 0.33600210692652094,
      "grad_norm": 2.8226709365844727,
      "learning_rate": 3.321174611535423e-05,
      "loss": 0.8045,
      "step": 12758
    },
    {
      "epoch": 0.3360284435080327,
      "grad_norm": 2.085674524307251,
      "learning_rate": 3.3210429286278646e-05,
      "loss": 1.0863,
      "step": 12759
    },
    {
      "epoch": 0.33605478008954437,
      "grad_norm": 2.4771549701690674,
      "learning_rate": 3.3209112457203055e-05,
      "loss": 0.2493,
      "step": 12760
    },
    {
      "epoch": 0.3360811166710561,
      "grad_norm": 1.5374163389205933,
      "learning_rate": 3.320779562812747e-05,
      "loss": 1.4568,
      "step": 12761
    },
    {
      "epoch": 0.3361074532525678,
      "grad_norm": 3.3248648643493652,
      "learning_rate": 3.320647879905188e-05,
      "loss": 0.7358,
      "step": 12762
    },
    {
      "epoch": 0.33613378983407954,
      "grad_norm": 2.836918830871582,
      "learning_rate": 3.32051619699763e-05,
      "loss": 0.4637,
      "step": 12763
    },
    {
      "epoch": 0.33616012641559123,
      "grad_norm": 1.866237759590149,
      "learning_rate": 3.320384514090071e-05,
      "loss": 1.9141,
      "step": 12764
    },
    {
      "epoch": 0.336186462997103,
      "grad_norm": 1.6651567220687866,
      "learning_rate": 3.3202528311825126e-05,
      "loss": 2.2732,
      "step": 12765
    },
    {
      "epoch": 0.3362127995786147,
      "grad_norm": 2.9306986331939697,
      "learning_rate": 3.320121148274954e-05,
      "loss": 0.9799,
      "step": 12766
    },
    {
      "epoch": 0.3362391361601264,
      "grad_norm": 2.177807569503784,
      "learning_rate": 3.319989465367396e-05,
      "loss": 1.3731,
      "step": 12767
    },
    {
      "epoch": 0.33626547274163815,
      "grad_norm": 3.618196964263916,
      "learning_rate": 3.319857782459837e-05,
      "loss": 0.7953,
      "step": 12768
    },
    {
      "epoch": 0.33629180932314984,
      "grad_norm": 1.4932246208190918,
      "learning_rate": 3.319726099552278e-05,
      "loss": 1.4672,
      "step": 12769
    },
    {
      "epoch": 0.3363181459046616,
      "grad_norm": 4.007235527038574,
      "learning_rate": 3.31959441664472e-05,
      "loss": 1.9454,
      "step": 12770
    },
    {
      "epoch": 0.33634448248617327,
      "grad_norm": 1.438978910446167,
      "learning_rate": 3.3194627337371606e-05,
      "loss": 1.199,
      "step": 12771
    },
    {
      "epoch": 0.336370819067685,
      "grad_norm": 3.1330626010894775,
      "learning_rate": 3.319331050829603e-05,
      "loss": 0.8972,
      "step": 12772
    },
    {
      "epoch": 0.33639715564919676,
      "grad_norm": 2.868407964706421,
      "learning_rate": 3.319199367922044e-05,
      "loss": 1.3196,
      "step": 12773
    },
    {
      "epoch": 0.33642349223070844,
      "grad_norm": 1.7732014656066895,
      "learning_rate": 3.319067685014485e-05,
      "loss": 1.654,
      "step": 12774
    },
    {
      "epoch": 0.3364498288122202,
      "grad_norm": 2.6505112648010254,
      "learning_rate": 3.318936002106927e-05,
      "loss": 1.5214,
      "step": 12775
    },
    {
      "epoch": 0.3364761653937319,
      "grad_norm": 2.1598734855651855,
      "learning_rate": 3.318804319199368e-05,
      "loss": 1.652,
      "step": 12776
    },
    {
      "epoch": 0.3365025019752436,
      "grad_norm": 1.454651117324829,
      "learning_rate": 3.31867263629181e-05,
      "loss": 2.1821,
      "step": 12777
    },
    {
      "epoch": 0.33652883855675536,
      "grad_norm": 1.8590773344039917,
      "learning_rate": 3.318540953384251e-05,
      "loss": 0.7731,
      "step": 12778
    },
    {
      "epoch": 0.33655517513826705,
      "grad_norm": 2.3035006523132324,
      "learning_rate": 3.3184092704766924e-05,
      "loss": 1.4454,
      "step": 12779
    },
    {
      "epoch": 0.3365815117197788,
      "grad_norm": 2.9017720222473145,
      "learning_rate": 3.318277587569133e-05,
      "loss": 1.3172,
      "step": 12780
    },
    {
      "epoch": 0.3366078483012905,
      "grad_norm": 2.5573019981384277,
      "learning_rate": 3.318145904661575e-05,
      "loss": 2.1342,
      "step": 12781
    },
    {
      "epoch": 0.3366341848828022,
      "grad_norm": 5.44240665435791,
      "learning_rate": 3.3180142217540164e-05,
      "loss": 1.5762,
      "step": 12782
    },
    {
      "epoch": 0.3366605214643139,
      "grad_norm": 2.197305679321289,
      "learning_rate": 3.317882538846458e-05,
      "loss": 1.4593,
      "step": 12783
    },
    {
      "epoch": 0.33668685804582565,
      "grad_norm": 1.8757997751235962,
      "learning_rate": 3.3177508559388995e-05,
      "loss": 2.5188,
      "step": 12784
    },
    {
      "epoch": 0.3367131946273374,
      "grad_norm": 2.2119293212890625,
      "learning_rate": 3.3176191730313404e-05,
      "loss": 1.7756,
      "step": 12785
    },
    {
      "epoch": 0.3367395312088491,
      "grad_norm": 2.237448215484619,
      "learning_rate": 3.317487490123783e-05,
      "loss": 2.3963,
      "step": 12786
    },
    {
      "epoch": 0.33676586779036083,
      "grad_norm": 1.9436910152435303,
      "learning_rate": 3.3173558072162235e-05,
      "loss": 1.5204,
      "step": 12787
    },
    {
      "epoch": 0.3367922043718725,
      "grad_norm": 2.5121445655822754,
      "learning_rate": 3.317224124308665e-05,
      "loss": 2.4555,
      "step": 12788
    },
    {
      "epoch": 0.33681854095338426,
      "grad_norm": 3.137997627258301,
      "learning_rate": 3.317092441401106e-05,
      "loss": 1.0162,
      "step": 12789
    },
    {
      "epoch": 0.33684487753489595,
      "grad_norm": 2.0975558757781982,
      "learning_rate": 3.3169607584935475e-05,
      "loss": 1.5044,
      "step": 12790
    },
    {
      "epoch": 0.3368712141164077,
      "grad_norm": 3.707554817199707,
      "learning_rate": 3.316829075585989e-05,
      "loss": 1.8194,
      "step": 12791
    },
    {
      "epoch": 0.33689755069791943,
      "grad_norm": 2.3909502029418945,
      "learning_rate": 3.316697392678431e-05,
      "loss": 0.722,
      "step": 12792
    },
    {
      "epoch": 0.3369238872794311,
      "grad_norm": 3.9681878089904785,
      "learning_rate": 3.3165657097708715e-05,
      "loss": 1.4723,
      "step": 12793
    },
    {
      "epoch": 0.33695022386094287,
      "grad_norm": 3.120326519012451,
      "learning_rate": 3.316434026863313e-05,
      "loss": 1.7973,
      "step": 12794
    },
    {
      "epoch": 0.33697656044245455,
      "grad_norm": 2.3361856937408447,
      "learning_rate": 3.316302343955755e-05,
      "loss": 1.8561,
      "step": 12795
    },
    {
      "epoch": 0.3370028970239663,
      "grad_norm": 1.982590913772583,
      "learning_rate": 3.316170661048196e-05,
      "loss": 1.5747,
      "step": 12796
    },
    {
      "epoch": 0.337029233605478,
      "grad_norm": 2.360163927078247,
      "learning_rate": 3.316038978140638e-05,
      "loss": 1.8484,
      "step": 12797
    },
    {
      "epoch": 0.33705557018698973,
      "grad_norm": 1.897743582725525,
      "learning_rate": 3.315907295233079e-05,
      "loss": 1.6949,
      "step": 12798
    },
    {
      "epoch": 0.33708190676850147,
      "grad_norm": 2.4473464488983154,
      "learning_rate": 3.31577561232552e-05,
      "loss": 1.4171,
      "step": 12799
    },
    {
      "epoch": 0.33710824335001316,
      "grad_norm": 2.8679211139678955,
      "learning_rate": 3.315643929417962e-05,
      "loss": 0.4834,
      "step": 12800
    },
    {
      "epoch": 0.3371345799315249,
      "grad_norm": 3.1247572898864746,
      "learning_rate": 3.3155122465104033e-05,
      "loss": 0.6327,
      "step": 12801
    },
    {
      "epoch": 0.3371609165130366,
      "grad_norm": 1.7025243043899536,
      "learning_rate": 3.315380563602844e-05,
      "loss": 1.6993,
      "step": 12802
    },
    {
      "epoch": 0.33718725309454833,
      "grad_norm": 2.5768539905548096,
      "learning_rate": 3.315248880695286e-05,
      "loss": 1.9954,
      "step": 12803
    },
    {
      "epoch": 0.33721358967606,
      "grad_norm": 1.8112659454345703,
      "learning_rate": 3.3151171977877274e-05,
      "loss": 0.6745,
      "step": 12804
    },
    {
      "epoch": 0.33723992625757176,
      "grad_norm": 2.1577470302581787,
      "learning_rate": 3.314985514880169e-05,
      "loss": 1.6175,
      "step": 12805
    },
    {
      "epoch": 0.3372662628390835,
      "grad_norm": 4.229220390319824,
      "learning_rate": 3.3148538319726105e-05,
      "loss": 0.9116,
      "step": 12806
    },
    {
      "epoch": 0.3372925994205952,
      "grad_norm": 1.7240034341812134,
      "learning_rate": 3.3147221490650514e-05,
      "loss": 1.8792,
      "step": 12807
    },
    {
      "epoch": 0.33731893600210694,
      "grad_norm": 2.0040340423583984,
      "learning_rate": 3.314590466157493e-05,
      "loss": 1.8932,
      "step": 12808
    },
    {
      "epoch": 0.3373452725836186,
      "grad_norm": 1.5207792520523071,
      "learning_rate": 3.314458783249934e-05,
      "loss": 1.4416,
      "step": 12809
    },
    {
      "epoch": 0.33737160916513037,
      "grad_norm": 2.0664150714874268,
      "learning_rate": 3.314327100342376e-05,
      "loss": 1.7887,
      "step": 12810
    },
    {
      "epoch": 0.33739794574664206,
      "grad_norm": 1.918363332748413,
      "learning_rate": 3.314195417434817e-05,
      "loss": 2.1563,
      "step": 12811
    },
    {
      "epoch": 0.3374242823281538,
      "grad_norm": 1.9444149732589722,
      "learning_rate": 3.3140637345272585e-05,
      "loss": 1.494,
      "step": 12812
    },
    {
      "epoch": 0.33745061890966555,
      "grad_norm": 2.98043155670166,
      "learning_rate": 3.3139320516197e-05,
      "loss": 2.9948,
      "step": 12813
    },
    {
      "epoch": 0.33747695549117723,
      "grad_norm": 4.101751804351807,
      "learning_rate": 3.313800368712141e-05,
      "loss": 2.0274,
      "step": 12814
    },
    {
      "epoch": 0.337503292072689,
      "grad_norm": 3.0297694206237793,
      "learning_rate": 3.313668685804583e-05,
      "loss": 0.8011,
      "step": 12815
    },
    {
      "epoch": 0.33752962865420066,
      "grad_norm": 1.9750598669052124,
      "learning_rate": 3.313537002897024e-05,
      "loss": 1.5386,
      "step": 12816
    },
    {
      "epoch": 0.3375559652357124,
      "grad_norm": 1.8156180381774902,
      "learning_rate": 3.3134053199894656e-05,
      "loss": 1.6884,
      "step": 12817
    },
    {
      "epoch": 0.33758230181722415,
      "grad_norm": 1.7295427322387695,
      "learning_rate": 3.3132736370819065e-05,
      "loss": 1.3958,
      "step": 12818
    },
    {
      "epoch": 0.33760863839873584,
      "grad_norm": 3.0939669609069824,
      "learning_rate": 3.313141954174349e-05,
      "loss": 1.1396,
      "step": 12819
    },
    {
      "epoch": 0.3376349749802476,
      "grad_norm": 2.344482421875,
      "learning_rate": 3.3130102712667896e-05,
      "loss": 2.5968,
      "step": 12820
    },
    {
      "epoch": 0.33766131156175927,
      "grad_norm": 1.9602240324020386,
      "learning_rate": 3.312878588359231e-05,
      "loss": 1.378,
      "step": 12821
    },
    {
      "epoch": 0.337687648143271,
      "grad_norm": 2.8866045475006104,
      "learning_rate": 3.312746905451673e-05,
      "loss": 0.6235,
      "step": 12822
    },
    {
      "epoch": 0.3377139847247827,
      "grad_norm": 2.1267950534820557,
      "learning_rate": 3.3126152225441136e-05,
      "loss": 1.7626,
      "step": 12823
    },
    {
      "epoch": 0.33774032130629444,
      "grad_norm": 2.3368029594421387,
      "learning_rate": 3.312483539636556e-05,
      "loss": 1.2872,
      "step": 12824
    },
    {
      "epoch": 0.3377666578878062,
      "grad_norm": 2.0095088481903076,
      "learning_rate": 3.312351856728997e-05,
      "loss": 1.5743,
      "step": 12825
    },
    {
      "epoch": 0.3377929944693179,
      "grad_norm": 2.897247552871704,
      "learning_rate": 3.312220173821438e-05,
      "loss": 0.8913,
      "step": 12826
    },
    {
      "epoch": 0.3378193310508296,
      "grad_norm": 1.7282084226608276,
      "learning_rate": 3.312088490913879e-05,
      "loss": 1.7159,
      "step": 12827
    },
    {
      "epoch": 0.3378456676323413,
      "grad_norm": 2.563190460205078,
      "learning_rate": 3.311956808006321e-05,
      "loss": 1.0071,
      "step": 12828
    },
    {
      "epoch": 0.33787200421385305,
      "grad_norm": 3.6421115398406982,
      "learning_rate": 3.311825125098762e-05,
      "loss": 0.8396,
      "step": 12829
    },
    {
      "epoch": 0.33789834079536474,
      "grad_norm": 2.631117820739746,
      "learning_rate": 3.311693442191204e-05,
      "loss": 1.3745,
      "step": 12830
    },
    {
      "epoch": 0.3379246773768765,
      "grad_norm": 2.5396296977996826,
      "learning_rate": 3.3115617592836454e-05,
      "loss": 2.5701,
      "step": 12831
    },
    {
      "epoch": 0.3379510139583882,
      "grad_norm": 1.7202781438827515,
      "learning_rate": 3.311430076376086e-05,
      "loss": 0.6105,
      "step": 12832
    },
    {
      "epoch": 0.3379773505398999,
      "grad_norm": 3.588066816329956,
      "learning_rate": 3.3112983934685285e-05,
      "loss": 2.2954,
      "step": 12833
    },
    {
      "epoch": 0.33800368712141166,
      "grad_norm": 3.7804176807403564,
      "learning_rate": 3.3111667105609694e-05,
      "loss": 1.9091,
      "step": 12834
    },
    {
      "epoch": 0.33803002370292334,
      "grad_norm": 2.6650660037994385,
      "learning_rate": 3.311035027653411e-05,
      "loss": 2.4609,
      "step": 12835
    },
    {
      "epoch": 0.3380563602844351,
      "grad_norm": 1.6852717399597168,
      "learning_rate": 3.310903344745852e-05,
      "loss": 1.9774,
      "step": 12836
    },
    {
      "epoch": 0.3380826968659468,
      "grad_norm": 4.781483173370361,
      "learning_rate": 3.3107716618382934e-05,
      "loss": 3.3361,
      "step": 12837
    },
    {
      "epoch": 0.3381090334474585,
      "grad_norm": 2.265838146209717,
      "learning_rate": 3.310639978930735e-05,
      "loss": 1.8377,
      "step": 12838
    },
    {
      "epoch": 0.33813537002897026,
      "grad_norm": 3.260784864425659,
      "learning_rate": 3.3105082960231765e-05,
      "loss": 1.0938,
      "step": 12839
    },
    {
      "epoch": 0.33816170661048195,
      "grad_norm": 2.5565221309661865,
      "learning_rate": 3.310376613115618e-05,
      "loss": 1.9097,
      "step": 12840
    },
    {
      "epoch": 0.3381880431919937,
      "grad_norm": 1.9164501428604126,
      "learning_rate": 3.310244930208059e-05,
      "loss": 0.5112,
      "step": 12841
    },
    {
      "epoch": 0.3382143797735054,
      "grad_norm": 2.4099740982055664,
      "learning_rate": 3.3101132473005005e-05,
      "loss": 2.0523,
      "step": 12842
    },
    {
      "epoch": 0.3382407163550171,
      "grad_norm": 2.934231758117676,
      "learning_rate": 3.309981564392942e-05,
      "loss": 0.776,
      "step": 12843
    },
    {
      "epoch": 0.3382670529365288,
      "grad_norm": 2.5678024291992188,
      "learning_rate": 3.3098498814853836e-05,
      "loss": 2.0246,
      "step": 12844
    },
    {
      "epoch": 0.33829338951804055,
      "grad_norm": 2.3322393894195557,
      "learning_rate": 3.3097181985778245e-05,
      "loss": 1.0182,
      "step": 12845
    },
    {
      "epoch": 0.3383197260995523,
      "grad_norm": 4.788005352020264,
      "learning_rate": 3.309586515670266e-05,
      "loss": 1.0697,
      "step": 12846
    },
    {
      "epoch": 0.338346062681064,
      "grad_norm": 1.8777328729629517,
      "learning_rate": 3.309454832762707e-05,
      "loss": 1.9729,
      "step": 12847
    },
    {
      "epoch": 0.33837239926257573,
      "grad_norm": 2.1383659839630127,
      "learning_rate": 3.309323149855149e-05,
      "loss": 0.5294,
      "step": 12848
    },
    {
      "epoch": 0.3383987358440874,
      "grad_norm": 2.4611804485321045,
      "learning_rate": 3.30919146694759e-05,
      "loss": 2.306,
      "step": 12849
    },
    {
      "epoch": 0.33842507242559916,
      "grad_norm": 2.3517186641693115,
      "learning_rate": 3.3090597840400316e-05,
      "loss": 1.616,
      "step": 12850
    },
    {
      "epoch": 0.3384514090071109,
      "grad_norm": 1.985917568206787,
      "learning_rate": 3.308928101132473e-05,
      "loss": 2.3605,
      "step": 12851
    },
    {
      "epoch": 0.3384777455886226,
      "grad_norm": 1.65116286277771,
      "learning_rate": 3.308796418224915e-05,
      "loss": 2.601,
      "step": 12852
    },
    {
      "epoch": 0.33850408217013433,
      "grad_norm": 2.4410359859466553,
      "learning_rate": 3.308664735317356e-05,
      "loss": 2.4911,
      "step": 12853
    },
    {
      "epoch": 0.338530418751646,
      "grad_norm": 2.6291885375976562,
      "learning_rate": 3.308533052409797e-05,
      "loss": 1.7102,
      "step": 12854
    },
    {
      "epoch": 0.33855675533315777,
      "grad_norm": 2.732045888900757,
      "learning_rate": 3.308401369502239e-05,
      "loss": 0.7609,
      "step": 12855
    },
    {
      "epoch": 0.33858309191466945,
      "grad_norm": 3.9982266426086426,
      "learning_rate": 3.3082696865946797e-05,
      "loss": 0.9718,
      "step": 12856
    },
    {
      "epoch": 0.3386094284961812,
      "grad_norm": 1.8279403448104858,
      "learning_rate": 3.308138003687122e-05,
      "loss": 1.5876,
      "step": 12857
    },
    {
      "epoch": 0.33863576507769294,
      "grad_norm": 4.195092678070068,
      "learning_rate": 3.308006320779563e-05,
      "loss": 1.9992,
      "step": 12858
    },
    {
      "epoch": 0.33866210165920463,
      "grad_norm": 2.385598659515381,
      "learning_rate": 3.307874637872004e-05,
      "loss": 0.604,
      "step": 12859
    },
    {
      "epoch": 0.33868843824071637,
      "grad_norm": 2.4315998554229736,
      "learning_rate": 3.307742954964446e-05,
      "loss": 1.0191,
      "step": 12860
    },
    {
      "epoch": 0.33871477482222806,
      "grad_norm": 1.9641562700271606,
      "learning_rate": 3.307611272056887e-05,
      "loss": 1.3106,
      "step": 12861
    },
    {
      "epoch": 0.3387411114037398,
      "grad_norm": 2.0438058376312256,
      "learning_rate": 3.307479589149329e-05,
      "loss": 1.4516,
      "step": 12862
    },
    {
      "epoch": 0.3387674479852515,
      "grad_norm": 2.5008726119995117,
      "learning_rate": 3.30734790624177e-05,
      "loss": 0.714,
      "step": 12863
    },
    {
      "epoch": 0.33879378456676323,
      "grad_norm": 1.8285330533981323,
      "learning_rate": 3.3072162233342115e-05,
      "loss": 1.5515,
      "step": 12864
    },
    {
      "epoch": 0.338820121148275,
      "grad_norm": 3.1687991619110107,
      "learning_rate": 3.307084540426652e-05,
      "loss": 1.5192,
      "step": 12865
    },
    {
      "epoch": 0.33884645772978667,
      "grad_norm": 4.436827182769775,
      "learning_rate": 3.3069528575190946e-05,
      "loss": 1.5977,
      "step": 12866
    },
    {
      "epoch": 0.3388727943112984,
      "grad_norm": 2.423713445663452,
      "learning_rate": 3.3068211746115355e-05,
      "loss": 1.837,
      "step": 12867
    },
    {
      "epoch": 0.3388991308928101,
      "grad_norm": 2.350836992263794,
      "learning_rate": 3.306689491703977e-05,
      "loss": 2.1863,
      "step": 12868
    },
    {
      "epoch": 0.33892546747432184,
      "grad_norm": 1.7617946863174438,
      "learning_rate": 3.3065578087964186e-05,
      "loss": 1.1963,
      "step": 12869
    },
    {
      "epoch": 0.3389518040558335,
      "grad_norm": 1.6627624034881592,
      "learning_rate": 3.3064261258888595e-05,
      "loss": 1.8442,
      "step": 12870
    },
    {
      "epoch": 0.33897814063734527,
      "grad_norm": 3.7690541744232178,
      "learning_rate": 3.306294442981302e-05,
      "loss": 1.418,
      "step": 12871
    },
    {
      "epoch": 0.339004477218857,
      "grad_norm": 1.5701606273651123,
      "learning_rate": 3.3061627600737426e-05,
      "loss": 1.0721,
      "step": 12872
    },
    {
      "epoch": 0.3390308138003687,
      "grad_norm": 1.9976402521133423,
      "learning_rate": 3.306031077166184e-05,
      "loss": 0.2003,
      "step": 12873
    },
    {
      "epoch": 0.33905715038188045,
      "grad_norm": 2.4352684020996094,
      "learning_rate": 3.305899394258625e-05,
      "loss": 2.0669,
      "step": 12874
    },
    {
      "epoch": 0.33908348696339213,
      "grad_norm": 1.9757521152496338,
      "learning_rate": 3.3057677113510666e-05,
      "loss": 1.5744,
      "step": 12875
    },
    {
      "epoch": 0.3391098235449039,
      "grad_norm": 1.7175980806350708,
      "learning_rate": 3.305636028443508e-05,
      "loss": 2.5,
      "step": 12876
    },
    {
      "epoch": 0.33913616012641556,
      "grad_norm": 1.584818959236145,
      "learning_rate": 3.30550434553595e-05,
      "loss": 0.3951,
      "step": 12877
    },
    {
      "epoch": 0.3391624967079273,
      "grad_norm": 2.2938072681427,
      "learning_rate": 3.305372662628391e-05,
      "loss": 2.0871,
      "step": 12878
    },
    {
      "epoch": 0.33918883328943905,
      "grad_norm": 1.6901754140853882,
      "learning_rate": 3.305240979720832e-05,
      "loss": 1.5823,
      "step": 12879
    },
    {
      "epoch": 0.33921516987095074,
      "grad_norm": 1.4461736679077148,
      "learning_rate": 3.305109296813274e-05,
      "loss": 1.4325,
      "step": 12880
    },
    {
      "epoch": 0.3392415064524625,
      "grad_norm": 3.1961100101470947,
      "learning_rate": 3.304977613905715e-05,
      "loss": 1.4385,
      "step": 12881
    },
    {
      "epoch": 0.33926784303397417,
      "grad_norm": 1.6787663698196411,
      "learning_rate": 3.304845930998157e-05,
      "loss": 1.052,
      "step": 12882
    },
    {
      "epoch": 0.3392941796154859,
      "grad_norm": 3.1201679706573486,
      "learning_rate": 3.304714248090598e-05,
      "loss": 2.0348,
      "step": 12883
    },
    {
      "epoch": 0.33932051619699766,
      "grad_norm": 1.6835204362869263,
      "learning_rate": 3.304582565183039e-05,
      "loss": 1.5206,
      "step": 12884
    },
    {
      "epoch": 0.33934685277850934,
      "grad_norm": 2.4557645320892334,
      "learning_rate": 3.304450882275481e-05,
      "loss": 0.5015,
      "step": 12885
    },
    {
      "epoch": 0.3393731893600211,
      "grad_norm": 4.528027534484863,
      "learning_rate": 3.3043191993679224e-05,
      "loss": 1.3313,
      "step": 12886
    },
    {
      "epoch": 0.3393995259415328,
      "grad_norm": 2.5903267860412598,
      "learning_rate": 3.304187516460364e-05,
      "loss": 1.3941,
      "step": 12887
    },
    {
      "epoch": 0.3394258625230445,
      "grad_norm": 3.5780093669891357,
      "learning_rate": 3.304055833552805e-05,
      "loss": 0.9176,
      "step": 12888
    },
    {
      "epoch": 0.3394521991045562,
      "grad_norm": 1.750388741493225,
      "learning_rate": 3.3039241506452464e-05,
      "loss": 2.0672,
      "step": 12889
    },
    {
      "epoch": 0.33947853568606795,
      "grad_norm": 2.0618300437927246,
      "learning_rate": 3.303792467737688e-05,
      "loss": 2.1176,
      "step": 12890
    },
    {
      "epoch": 0.3395048722675797,
      "grad_norm": 5.245141983032227,
      "learning_rate": 3.3036607848301295e-05,
      "loss": 0.8398,
      "step": 12891
    },
    {
      "epoch": 0.3395312088490914,
      "grad_norm": 4.576908588409424,
      "learning_rate": 3.3035291019225704e-05,
      "loss": 1.6884,
      "step": 12892
    },
    {
      "epoch": 0.3395575454306031,
      "grad_norm": 2.666538953781128,
      "learning_rate": 3.303397419015012e-05,
      "loss": 0.3919,
      "step": 12893
    },
    {
      "epoch": 0.3395838820121148,
      "grad_norm": 2.1481711864471436,
      "learning_rate": 3.303265736107453e-05,
      "loss": 2.3544,
      "step": 12894
    },
    {
      "epoch": 0.33961021859362656,
      "grad_norm": 1.6148473024368286,
      "learning_rate": 3.303134053199895e-05,
      "loss": 2.0484,
      "step": 12895
    },
    {
      "epoch": 0.33963655517513824,
      "grad_norm": 4.5628862380981445,
      "learning_rate": 3.303002370292336e-05,
      "loss": 1.7455,
      "step": 12896
    },
    {
      "epoch": 0.33966289175665,
      "grad_norm": 4.540353775024414,
      "learning_rate": 3.3028706873847775e-05,
      "loss": 0.854,
      "step": 12897
    },
    {
      "epoch": 0.33968922833816173,
      "grad_norm": 2.170725107192993,
      "learning_rate": 3.302739004477219e-05,
      "loss": 1.9693,
      "step": 12898
    },
    {
      "epoch": 0.3397155649196734,
      "grad_norm": 1.7289457321166992,
      "learning_rate": 3.3026073215696606e-05,
      "loss": 1.6487,
      "step": 12899
    },
    {
      "epoch": 0.33974190150118516,
      "grad_norm": 3.3022546768188477,
      "learning_rate": 3.302475638662102e-05,
      "loss": 0.3773,
      "step": 12900
    },
    {
      "epoch": 0.33976823808269685,
      "grad_norm": 1.7960892915725708,
      "learning_rate": 3.302343955754543e-05,
      "loss": 1.6454,
      "step": 12901
    },
    {
      "epoch": 0.3397945746642086,
      "grad_norm": 2.1280105113983154,
      "learning_rate": 3.3022122728469846e-05,
      "loss": 0.6364,
      "step": 12902
    },
    {
      "epoch": 0.3398209112457203,
      "grad_norm": 5.191806316375732,
      "learning_rate": 3.3020805899394255e-05,
      "loss": 2.316,
      "step": 12903
    },
    {
      "epoch": 0.339847247827232,
      "grad_norm": 5.268148422241211,
      "learning_rate": 3.301948907031868e-05,
      "loss": 2.0061,
      "step": 12904
    },
    {
      "epoch": 0.33987358440874377,
      "grad_norm": 3.1319453716278076,
      "learning_rate": 3.3018172241243086e-05,
      "loss": 1.4343,
      "step": 12905
    },
    {
      "epoch": 0.33989992099025546,
      "grad_norm": 2.618544578552246,
      "learning_rate": 3.30168554121675e-05,
      "loss": 2.0187,
      "step": 12906
    },
    {
      "epoch": 0.3399262575717672,
      "grad_norm": 3.03944730758667,
      "learning_rate": 3.301553858309192e-05,
      "loss": 1.7853,
      "step": 12907
    },
    {
      "epoch": 0.3399525941532789,
      "grad_norm": 1.9733080863952637,
      "learning_rate": 3.3014221754016326e-05,
      "loss": 1.9372,
      "step": 12908
    },
    {
      "epoch": 0.33997893073479063,
      "grad_norm": 1.8079891204833984,
      "learning_rate": 3.301290492494075e-05,
      "loss": 2.2751,
      "step": 12909
    },
    {
      "epoch": 0.3400052673163023,
      "grad_norm": 2.9763898849487305,
      "learning_rate": 3.301158809586516e-05,
      "loss": 2.1679,
      "step": 12910
    },
    {
      "epoch": 0.34003160389781406,
      "grad_norm": 3.666808605194092,
      "learning_rate": 3.301027126678957e-05,
      "loss": 1.2256,
      "step": 12911
    },
    {
      "epoch": 0.3400579404793258,
      "grad_norm": 4.210350513458252,
      "learning_rate": 3.300895443771398e-05,
      "loss": 1.3845,
      "step": 12912
    },
    {
      "epoch": 0.3400842770608375,
      "grad_norm": 5.638673305511475,
      "learning_rate": 3.30076376086384e-05,
      "loss": 2.2161,
      "step": 12913
    },
    {
      "epoch": 0.34011061364234924,
      "grad_norm": 1.4734793901443481,
      "learning_rate": 3.300632077956281e-05,
      "loss": 1.9166,
      "step": 12914
    },
    {
      "epoch": 0.3401369502238609,
      "grad_norm": 3.058187246322632,
      "learning_rate": 3.300500395048723e-05,
      "loss": 1.6315,
      "step": 12915
    },
    {
      "epoch": 0.34016328680537267,
      "grad_norm": 2.6017966270446777,
      "learning_rate": 3.3003687121411644e-05,
      "loss": 1.0321,
      "step": 12916
    },
    {
      "epoch": 0.3401896233868844,
      "grad_norm": 2.0909831523895264,
      "learning_rate": 3.300237029233605e-05,
      "loss": 0.3767,
      "step": 12917
    },
    {
      "epoch": 0.3402159599683961,
      "grad_norm": 3.3884739875793457,
      "learning_rate": 3.3001053463260475e-05,
      "loss": 0.8194,
      "step": 12918
    },
    {
      "epoch": 0.34024229654990784,
      "grad_norm": 3.3136556148529053,
      "learning_rate": 3.2999736634184884e-05,
      "loss": 1.5631,
      "step": 12919
    },
    {
      "epoch": 0.34026863313141953,
      "grad_norm": 1.5209213495254517,
      "learning_rate": 3.29984198051093e-05,
      "loss": 1.633,
      "step": 12920
    },
    {
      "epoch": 0.3402949697129313,
      "grad_norm": 1.856662392616272,
      "learning_rate": 3.299710297603371e-05,
      "loss": 2.6551,
      "step": 12921
    },
    {
      "epoch": 0.34032130629444296,
      "grad_norm": 3.5940425395965576,
      "learning_rate": 3.2995786146958124e-05,
      "loss": 1.5755,
      "step": 12922
    },
    {
      "epoch": 0.3403476428759547,
      "grad_norm": 1.6247870922088623,
      "learning_rate": 3.299446931788254e-05,
      "loss": 1.5426,
      "step": 12923
    },
    {
      "epoch": 0.34037397945746645,
      "grad_norm": 2.991795539855957,
      "learning_rate": 3.2993152488806956e-05,
      "loss": 1.6865,
      "step": 12924
    },
    {
      "epoch": 0.34040031603897813,
      "grad_norm": 2.3574752807617188,
      "learning_rate": 3.299183565973137e-05,
      "loss": 1.4322,
      "step": 12925
    },
    {
      "epoch": 0.3404266526204899,
      "grad_norm": 2.4562690258026123,
      "learning_rate": 3.299051883065578e-05,
      "loss": 2.1818,
      "step": 12926
    },
    {
      "epoch": 0.34045298920200157,
      "grad_norm": 1.8044371604919434,
      "learning_rate": 3.2989202001580196e-05,
      "loss": 1.235,
      "step": 12927
    },
    {
      "epoch": 0.3404793257835133,
      "grad_norm": 1.6219490766525269,
      "learning_rate": 3.298788517250461e-05,
      "loss": 1.9105,
      "step": 12928
    },
    {
      "epoch": 0.340505662365025,
      "grad_norm": 1.761940360069275,
      "learning_rate": 3.298656834342903e-05,
      "loss": 1.2498,
      "step": 12929
    },
    {
      "epoch": 0.34053199894653674,
      "grad_norm": 1.5885579586029053,
      "learning_rate": 3.2985251514353436e-05,
      "loss": 1.6395,
      "step": 12930
    },
    {
      "epoch": 0.3405583355280485,
      "grad_norm": 2.0496907234191895,
      "learning_rate": 3.298393468527785e-05,
      "loss": 2.1072,
      "step": 12931
    },
    {
      "epoch": 0.34058467210956017,
      "grad_norm": 2.022467851638794,
      "learning_rate": 3.298261785620227e-05,
      "loss": 1.5395,
      "step": 12932
    },
    {
      "epoch": 0.3406110086910719,
      "grad_norm": 3.4611968994140625,
      "learning_rate": 3.298130102712668e-05,
      "loss": 1.7434,
      "step": 12933
    },
    {
      "epoch": 0.3406373452725836,
      "grad_norm": 2.2028777599334717,
      "learning_rate": 3.29799841980511e-05,
      "loss": 2.047,
      "step": 12934
    },
    {
      "epoch": 0.34066368185409535,
      "grad_norm": 2.633039712905884,
      "learning_rate": 3.297866736897551e-05,
      "loss": 0.4042,
      "step": 12935
    },
    {
      "epoch": 0.34069001843560703,
      "grad_norm": 2.170102596282959,
      "learning_rate": 3.297735053989992e-05,
      "loss": 1.5421,
      "step": 12936
    },
    {
      "epoch": 0.3407163550171188,
      "grad_norm": 1.722725749015808,
      "learning_rate": 3.297603371082434e-05,
      "loss": 1.4871,
      "step": 12937
    },
    {
      "epoch": 0.3407426915986305,
      "grad_norm": 2.009075880050659,
      "learning_rate": 3.2974716881748754e-05,
      "loss": 1.5152,
      "step": 12938
    },
    {
      "epoch": 0.3407690281801422,
      "grad_norm": 2.443164825439453,
      "learning_rate": 3.297340005267316e-05,
      "loss": 2.1061,
      "step": 12939
    },
    {
      "epoch": 0.34079536476165395,
      "grad_norm": 1.4557195901870728,
      "learning_rate": 3.297208322359758e-05,
      "loss": 1.889,
      "step": 12940
    },
    {
      "epoch": 0.34082170134316564,
      "grad_norm": 2.3596839904785156,
      "learning_rate": 3.2970766394521994e-05,
      "loss": 2.5858,
      "step": 12941
    },
    {
      "epoch": 0.3408480379246774,
      "grad_norm": 2.858344793319702,
      "learning_rate": 3.296944956544641e-05,
      "loss": 1.8894,
      "step": 12942
    },
    {
      "epoch": 0.34087437450618907,
      "grad_norm": 2.448979377746582,
      "learning_rate": 3.2968132736370825e-05,
      "loss": 1.77,
      "step": 12943
    },
    {
      "epoch": 0.3409007110877008,
      "grad_norm": 2.9496917724609375,
      "learning_rate": 3.2966815907295234e-05,
      "loss": 2.45,
      "step": 12944
    },
    {
      "epoch": 0.34092704766921256,
      "grad_norm": 1.8532429933547974,
      "learning_rate": 3.296549907821965e-05,
      "loss": 1.4375,
      "step": 12945
    },
    {
      "epoch": 0.34095338425072425,
      "grad_norm": 2.1792612075805664,
      "learning_rate": 3.296418224914406e-05,
      "loss": 2.1459,
      "step": 12946
    },
    {
      "epoch": 0.340979720832236,
      "grad_norm": 2.0391368865966797,
      "learning_rate": 3.296286542006848e-05,
      "loss": 2.8481,
      "step": 12947
    },
    {
      "epoch": 0.3410060574137477,
      "grad_norm": 2.3514368534088135,
      "learning_rate": 3.296154859099289e-05,
      "loss": 2.0533,
      "step": 12948
    },
    {
      "epoch": 0.3410323939952594,
      "grad_norm": 4.717340469360352,
      "learning_rate": 3.2960231761917305e-05,
      "loss": 2.1282,
      "step": 12949
    },
    {
      "epoch": 0.3410587305767711,
      "grad_norm": 1.5506004095077515,
      "learning_rate": 3.2958914932841714e-05,
      "loss": 0.3485,
      "step": 12950
    },
    {
      "epoch": 0.34108506715828285,
      "grad_norm": 5.972497463226318,
      "learning_rate": 3.2957598103766136e-05,
      "loss": 1.8649,
      "step": 12951
    },
    {
      "epoch": 0.3411114037397946,
      "grad_norm": 1.9077333211898804,
      "learning_rate": 3.2956281274690545e-05,
      "loss": 1.9687,
      "step": 12952
    },
    {
      "epoch": 0.3411377403213063,
      "grad_norm": 2.513373851776123,
      "learning_rate": 3.295496444561496e-05,
      "loss": 0.8289,
      "step": 12953
    },
    {
      "epoch": 0.341164076902818,
      "grad_norm": 2.042729616165161,
      "learning_rate": 3.2953647616539376e-05,
      "loss": 1.48,
      "step": 12954
    },
    {
      "epoch": 0.3411904134843297,
      "grad_norm": 1.7733979225158691,
      "learning_rate": 3.2952330787463785e-05,
      "loss": 1.4375,
      "step": 12955
    },
    {
      "epoch": 0.34121675006584146,
      "grad_norm": 2.3418757915496826,
      "learning_rate": 3.295101395838821e-05,
      "loss": 1.6959,
      "step": 12956
    },
    {
      "epoch": 0.3412430866473532,
      "grad_norm": 2.2886176109313965,
      "learning_rate": 3.2949697129312616e-05,
      "loss": 1.5131,
      "step": 12957
    },
    {
      "epoch": 0.3412694232288649,
      "grad_norm": 2.2938649654388428,
      "learning_rate": 3.294838030023703e-05,
      "loss": 1.4211,
      "step": 12958
    },
    {
      "epoch": 0.34129575981037663,
      "grad_norm": 1.7915611267089844,
      "learning_rate": 3.294706347116144e-05,
      "loss": 1.9845,
      "step": 12959
    },
    {
      "epoch": 0.3413220963918883,
      "grad_norm": 1.5601946115493774,
      "learning_rate": 3.2945746642085856e-05,
      "loss": 2.1526,
      "step": 12960
    },
    {
      "epoch": 0.34134843297340006,
      "grad_norm": 4.8775715827941895,
      "learning_rate": 3.294442981301027e-05,
      "loss": 1.3841,
      "step": 12961
    },
    {
      "epoch": 0.34137476955491175,
      "grad_norm": 4.573395729064941,
      "learning_rate": 3.294311298393469e-05,
      "loss": 1.3154,
      "step": 12962
    },
    {
      "epoch": 0.3414011061364235,
      "grad_norm": 1.9860544204711914,
      "learning_rate": 3.29417961548591e-05,
      "loss": 1.0699,
      "step": 12963
    },
    {
      "epoch": 0.34142744271793524,
      "grad_norm": 2.873884677886963,
      "learning_rate": 3.294047932578351e-05,
      "loss": 1.3907,
      "step": 12964
    },
    {
      "epoch": 0.3414537792994469,
      "grad_norm": 3.7345573902130127,
      "learning_rate": 3.2939162496707934e-05,
      "loss": 2.4449,
      "step": 12965
    },
    {
      "epoch": 0.34148011588095867,
      "grad_norm": 4.145462989807129,
      "learning_rate": 3.293784566763234e-05,
      "loss": 1.8259,
      "step": 12966
    },
    {
      "epoch": 0.34150645246247036,
      "grad_norm": 1.9677534103393555,
      "learning_rate": 3.293652883855676e-05,
      "loss": 1.6432,
      "step": 12967
    },
    {
      "epoch": 0.3415327890439821,
      "grad_norm": 1.9247239828109741,
      "learning_rate": 3.293521200948117e-05,
      "loss": 2.0941,
      "step": 12968
    },
    {
      "epoch": 0.3415591256254938,
      "grad_norm": 2.2761850357055664,
      "learning_rate": 3.293389518040558e-05,
      "loss": 2.1237,
      "step": 12969
    },
    {
      "epoch": 0.34158546220700553,
      "grad_norm": 4.019107341766357,
      "learning_rate": 3.293257835133e-05,
      "loss": 1.6666,
      "step": 12970
    },
    {
      "epoch": 0.3416117987885173,
      "grad_norm": 3.608370304107666,
      "learning_rate": 3.2931261522254414e-05,
      "loss": 2.2006,
      "step": 12971
    },
    {
      "epoch": 0.34163813537002896,
      "grad_norm": 2.49223256111145,
      "learning_rate": 3.292994469317883e-05,
      "loss": 2.1709,
      "step": 12972
    },
    {
      "epoch": 0.3416644719515407,
      "grad_norm": 2.9999496936798096,
      "learning_rate": 3.292862786410324e-05,
      "loss": 1.9288,
      "step": 12973
    },
    {
      "epoch": 0.3416908085330524,
      "grad_norm": 2.16973876953125,
      "learning_rate": 3.2927311035027654e-05,
      "loss": 1.7177,
      "step": 12974
    },
    {
      "epoch": 0.34171714511456414,
      "grad_norm": 1.6832624673843384,
      "learning_rate": 3.292599420595207e-05,
      "loss": 1.9217,
      "step": 12975
    },
    {
      "epoch": 0.3417434816960758,
      "grad_norm": 2.6180667877197266,
      "learning_rate": 3.2924677376876485e-05,
      "loss": 1.8813,
      "step": 12976
    },
    {
      "epoch": 0.34176981827758757,
      "grad_norm": 3.1550745964050293,
      "learning_rate": 3.2923360547800894e-05,
      "loss": 1.1645,
      "step": 12977
    },
    {
      "epoch": 0.3417961548590993,
      "grad_norm": 1.835385799407959,
      "learning_rate": 3.292204371872531e-05,
      "loss": 1.2515,
      "step": 12978
    },
    {
      "epoch": 0.341822491440611,
      "grad_norm": 3.1698458194732666,
      "learning_rate": 3.2920726889649725e-05,
      "loss": 1.7964,
      "step": 12979
    },
    {
      "epoch": 0.34184882802212274,
      "grad_norm": 1.6796510219573975,
      "learning_rate": 3.291941006057414e-05,
      "loss": 1.0998,
      "step": 12980
    },
    {
      "epoch": 0.34187516460363443,
      "grad_norm": 3.3053088188171387,
      "learning_rate": 3.2918093231498556e-05,
      "loss": 0.6363,
      "step": 12981
    },
    {
      "epoch": 0.3419015011851462,
      "grad_norm": 3.5661163330078125,
      "learning_rate": 3.2916776402422965e-05,
      "loss": 1.1938,
      "step": 12982
    },
    {
      "epoch": 0.34192783776665786,
      "grad_norm": 1.6379643678665161,
      "learning_rate": 3.291545957334738e-05,
      "loss": 1.4223,
      "step": 12983
    },
    {
      "epoch": 0.3419541743481696,
      "grad_norm": 1.9944275617599487,
      "learning_rate": 3.2914142744271797e-05,
      "loss": 1.5809,
      "step": 12984
    },
    {
      "epoch": 0.34198051092968135,
      "grad_norm": 4.442276477813721,
      "learning_rate": 3.291282591519621e-05,
      "loss": 1.8037,
      "step": 12985
    },
    {
      "epoch": 0.34200684751119304,
      "grad_norm": 1.8749949932098389,
      "learning_rate": 3.291150908612062e-05,
      "loss": 1.6713,
      "step": 12986
    },
    {
      "epoch": 0.3420331840927048,
      "grad_norm": 4.132696628570557,
      "learning_rate": 3.2910192257045037e-05,
      "loss": 1.8504,
      "step": 12987
    },
    {
      "epoch": 0.34205952067421647,
      "grad_norm": 4.653345108032227,
      "learning_rate": 3.290887542796945e-05,
      "loss": 1.1259,
      "step": 12988
    },
    {
      "epoch": 0.3420858572557282,
      "grad_norm": 2.014456272125244,
      "learning_rate": 3.290755859889387e-05,
      "loss": 1.4112,
      "step": 12989
    },
    {
      "epoch": 0.34211219383723995,
      "grad_norm": 2.1527702808380127,
      "learning_rate": 3.290624176981828e-05,
      "loss": 1.6946,
      "step": 12990
    },
    {
      "epoch": 0.34213853041875164,
      "grad_norm": 2.077350616455078,
      "learning_rate": 3.290492494074269e-05,
      "loss": 1.5587,
      "step": 12991
    },
    {
      "epoch": 0.3421648670002634,
      "grad_norm": 3.211123466491699,
      "learning_rate": 3.290360811166711e-05,
      "loss": 2.392,
      "step": 12992
    },
    {
      "epoch": 0.34219120358177507,
      "grad_norm": 1.8776613473892212,
      "learning_rate": 3.2902291282591517e-05,
      "loss": 1.6357,
      "step": 12993
    },
    {
      "epoch": 0.3422175401632868,
      "grad_norm": 1.6870720386505127,
      "learning_rate": 3.290097445351594e-05,
      "loss": 0.5568,
      "step": 12994
    },
    {
      "epoch": 0.3422438767447985,
      "grad_norm": 2.017930030822754,
      "learning_rate": 3.289965762444035e-05,
      "loss": 1.4112,
      "step": 12995
    },
    {
      "epoch": 0.34227021332631025,
      "grad_norm": 4.67117977142334,
      "learning_rate": 3.289834079536476e-05,
      "loss": 1.3809,
      "step": 12996
    },
    {
      "epoch": 0.342296549907822,
      "grad_norm": 3.131514549255371,
      "learning_rate": 3.289702396628917e-05,
      "loss": 2.1468,
      "step": 12997
    },
    {
      "epoch": 0.3423228864893337,
      "grad_norm": 1.802449107170105,
      "learning_rate": 3.2895707137213595e-05,
      "loss": 2.4584,
      "step": 12998
    },
    {
      "epoch": 0.3423492230708454,
      "grad_norm": 2.2986903190612793,
      "learning_rate": 3.2894390308138e-05,
      "loss": 2.6267,
      "step": 12999
    },
    {
      "epoch": 0.3423755596523571,
      "grad_norm": 1.7511192560195923,
      "learning_rate": 3.289307347906242e-05,
      "loss": 2.051,
      "step": 13000
    },
    {
      "epoch": 0.34240189623386885,
      "grad_norm": 1.713340401649475,
      "learning_rate": 3.2891756649986835e-05,
      "loss": 2.172,
      "step": 13001
    },
    {
      "epoch": 0.34242823281538054,
      "grad_norm": 3.6396396160125732,
      "learning_rate": 3.289043982091124e-05,
      "loss": 1.2792,
      "step": 13002
    },
    {
      "epoch": 0.3424545693968923,
      "grad_norm": 1.5632052421569824,
      "learning_rate": 3.2889122991835666e-05,
      "loss": 1.8273,
      "step": 13003
    },
    {
      "epoch": 0.342480905978404,
      "grad_norm": 1.641810417175293,
      "learning_rate": 3.2887806162760075e-05,
      "loss": 1.4613,
      "step": 13004
    },
    {
      "epoch": 0.3425072425599157,
      "grad_norm": 1.4302617311477661,
      "learning_rate": 3.288648933368449e-05,
      "loss": 1.393,
      "step": 13005
    },
    {
      "epoch": 0.34253357914142746,
      "grad_norm": 1.902970314025879,
      "learning_rate": 3.28851725046089e-05,
      "loss": 1.6965,
      "step": 13006
    },
    {
      "epoch": 0.34255991572293915,
      "grad_norm": 1.6893277168273926,
      "learning_rate": 3.2883855675533315e-05,
      "loss": 2.3678,
      "step": 13007
    },
    {
      "epoch": 0.3425862523044509,
      "grad_norm": 1.9441275596618652,
      "learning_rate": 3.288253884645773e-05,
      "loss": 0.4138,
      "step": 13008
    },
    {
      "epoch": 0.3426125888859626,
      "grad_norm": 3.6069068908691406,
      "learning_rate": 3.2881222017382146e-05,
      "loss": 1.3208,
      "step": 13009
    },
    {
      "epoch": 0.3426389254674743,
      "grad_norm": 4.586077690124512,
      "learning_rate": 3.287990518830656e-05,
      "loss": 1.0181,
      "step": 13010
    },
    {
      "epoch": 0.34266526204898606,
      "grad_norm": 4.205984115600586,
      "learning_rate": 3.287858835923097e-05,
      "loss": 2.0801,
      "step": 13011
    },
    {
      "epoch": 0.34269159863049775,
      "grad_norm": 1.795978307723999,
      "learning_rate": 3.2877271530155386e-05,
      "loss": 1.7916,
      "step": 13012
    },
    {
      "epoch": 0.3427179352120095,
      "grad_norm": 2.2747647762298584,
      "learning_rate": 3.28759547010798e-05,
      "loss": 0.5522,
      "step": 13013
    },
    {
      "epoch": 0.3427442717935212,
      "grad_norm": 2.388296604156494,
      "learning_rate": 3.287463787200422e-05,
      "loss": 2.6569,
      "step": 13014
    },
    {
      "epoch": 0.3427706083750329,
      "grad_norm": 3.4760310649871826,
      "learning_rate": 3.2873321042928626e-05,
      "loss": 1.2178,
      "step": 13015
    },
    {
      "epoch": 0.3427969449565446,
      "grad_norm": 2.434558868408203,
      "learning_rate": 3.287200421385304e-05,
      "loss": 1.6785,
      "step": 13016
    },
    {
      "epoch": 0.34282328153805636,
      "grad_norm": 1.4341470003128052,
      "learning_rate": 3.287068738477746e-05,
      "loss": 1.9739,
      "step": 13017
    },
    {
      "epoch": 0.3428496181195681,
      "grad_norm": 1.779070496559143,
      "learning_rate": 3.286937055570187e-05,
      "loss": 2.2729,
      "step": 13018
    },
    {
      "epoch": 0.3428759547010798,
      "grad_norm": 1.7738845348358154,
      "learning_rate": 3.286805372662629e-05,
      "loss": 2.2236,
      "step": 13019
    },
    {
      "epoch": 0.34290229128259153,
      "grad_norm": 1.6878130435943604,
      "learning_rate": 3.28667368975507e-05,
      "loss": 1.8886,
      "step": 13020
    },
    {
      "epoch": 0.3429286278641032,
      "grad_norm": 2.979458808898926,
      "learning_rate": 3.286542006847511e-05,
      "loss": 1.8621,
      "step": 13021
    },
    {
      "epoch": 0.34295496444561496,
      "grad_norm": 2.4134817123413086,
      "learning_rate": 3.286410323939953e-05,
      "loss": 1.8015,
      "step": 13022
    },
    {
      "epoch": 0.3429813010271267,
      "grad_norm": 1.909686803817749,
      "learning_rate": 3.2862786410323944e-05,
      "loss": 1.6881,
      "step": 13023
    },
    {
      "epoch": 0.3430076376086384,
      "grad_norm": 2.2279653549194336,
      "learning_rate": 3.286146958124835e-05,
      "loss": 2.0882,
      "step": 13024
    },
    {
      "epoch": 0.34303397419015014,
      "grad_norm": 2.4747092723846436,
      "learning_rate": 3.286015275217277e-05,
      "loss": 1.9472,
      "step": 13025
    },
    {
      "epoch": 0.3430603107716618,
      "grad_norm": 1.6904834508895874,
      "learning_rate": 3.2858835923097184e-05,
      "loss": 2.2355,
      "step": 13026
    },
    {
      "epoch": 0.34308664735317357,
      "grad_norm": 1.9961947202682495,
      "learning_rate": 3.28575190940216e-05,
      "loss": 2.1338,
      "step": 13027
    },
    {
      "epoch": 0.34311298393468526,
      "grad_norm": 3.1862409114837646,
      "learning_rate": 3.2856202264946015e-05,
      "loss": 1.42,
      "step": 13028
    },
    {
      "epoch": 0.343139320516197,
      "grad_norm": 2.469804286956787,
      "learning_rate": 3.2854885435870424e-05,
      "loss": 2.1905,
      "step": 13029
    },
    {
      "epoch": 0.34316565709770874,
      "grad_norm": 1.7831059694290161,
      "learning_rate": 3.285356860679484e-05,
      "loss": 1.8759,
      "step": 13030
    },
    {
      "epoch": 0.34319199367922043,
      "grad_norm": 3.725809335708618,
      "learning_rate": 3.2852251777719255e-05,
      "loss": 2.3442,
      "step": 13031
    },
    {
      "epoch": 0.3432183302607322,
      "grad_norm": 1.7289234399795532,
      "learning_rate": 3.285093494864367e-05,
      "loss": 0.5565,
      "step": 13032
    },
    {
      "epoch": 0.34324466684224386,
      "grad_norm": 1.9075827598571777,
      "learning_rate": 3.284961811956808e-05,
      "loss": 1.2642,
      "step": 13033
    },
    {
      "epoch": 0.3432710034237556,
      "grad_norm": 2.122783899307251,
      "learning_rate": 3.2848301290492495e-05,
      "loss": 1.82,
      "step": 13034
    },
    {
      "epoch": 0.3432973400052673,
      "grad_norm": 2.0404770374298096,
      "learning_rate": 3.284698446141691e-05,
      "loss": 1.8464,
      "step": 13035
    },
    {
      "epoch": 0.34332367658677904,
      "grad_norm": 5.143828868865967,
      "learning_rate": 3.2845667632341326e-05,
      "loss": 1.7895,
      "step": 13036
    },
    {
      "epoch": 0.3433500131682908,
      "grad_norm": 2.072608470916748,
      "learning_rate": 3.284435080326574e-05,
      "loss": 1.6936,
      "step": 13037
    },
    {
      "epoch": 0.34337634974980247,
      "grad_norm": 2.0791525840759277,
      "learning_rate": 3.284303397419015e-05,
      "loss": 1.8769,
      "step": 13038
    },
    {
      "epoch": 0.3434026863313142,
      "grad_norm": 1.8462088108062744,
      "learning_rate": 3.2841717145114566e-05,
      "loss": 1.6844,
      "step": 13039
    },
    {
      "epoch": 0.3434290229128259,
      "grad_norm": 2.5627951622009277,
      "learning_rate": 3.2840400316038975e-05,
      "loss": 2.3559,
      "step": 13040
    },
    {
      "epoch": 0.34345535949433764,
      "grad_norm": 2.5939695835113525,
      "learning_rate": 3.28390834869634e-05,
      "loss": 1.325,
      "step": 13041
    },
    {
      "epoch": 0.34348169607584933,
      "grad_norm": 2.1293795108795166,
      "learning_rate": 3.2837766657887806e-05,
      "loss": 1.7575,
      "step": 13042
    },
    {
      "epoch": 0.3435080326573611,
      "grad_norm": 2.22837233543396,
      "learning_rate": 3.283644982881222e-05,
      "loss": 1.5616,
      "step": 13043
    },
    {
      "epoch": 0.3435343692388728,
      "grad_norm": 2.1316654682159424,
      "learning_rate": 3.283513299973664e-05,
      "loss": 1.1524,
      "step": 13044
    },
    {
      "epoch": 0.3435607058203845,
      "grad_norm": 3.323835849761963,
      "learning_rate": 3.283381617066105e-05,
      "loss": 1.7337,
      "step": 13045
    },
    {
      "epoch": 0.34358704240189625,
      "grad_norm": 2.016371488571167,
      "learning_rate": 3.283249934158547e-05,
      "loss": 1.7525,
      "step": 13046
    },
    {
      "epoch": 0.34361337898340794,
      "grad_norm": 1.7369242906570435,
      "learning_rate": 3.283118251250988e-05,
      "loss": 1.5029,
      "step": 13047
    },
    {
      "epoch": 0.3436397155649197,
      "grad_norm": 3.1583664417266846,
      "learning_rate": 3.282986568343429e-05,
      "loss": 1.9056,
      "step": 13048
    },
    {
      "epoch": 0.34366605214643137,
      "grad_norm": 1.6618739366531372,
      "learning_rate": 3.28285488543587e-05,
      "loss": 2.0452,
      "step": 13049
    },
    {
      "epoch": 0.3436923887279431,
      "grad_norm": 3.042189598083496,
      "learning_rate": 3.2827232025283124e-05,
      "loss": 1.2496,
      "step": 13050
    },
    {
      "epoch": 0.34371872530945485,
      "grad_norm": 1.8863723278045654,
      "learning_rate": 3.282591519620753e-05,
      "loss": 1.9603,
      "step": 13051
    },
    {
      "epoch": 0.34374506189096654,
      "grad_norm": 4.606906890869141,
      "learning_rate": 3.282459836713195e-05,
      "loss": 2.0634,
      "step": 13052
    },
    {
      "epoch": 0.3437713984724783,
      "grad_norm": 2.1368930339813232,
      "learning_rate": 3.282328153805636e-05,
      "loss": 2.0722,
      "step": 13053
    },
    {
      "epoch": 0.34379773505399,
      "grad_norm": 4.334184169769287,
      "learning_rate": 3.282196470898077e-05,
      "loss": 2.0015,
      "step": 13054
    },
    {
      "epoch": 0.3438240716355017,
      "grad_norm": 4.264140605926514,
      "learning_rate": 3.282064787990519e-05,
      "loss": 1.1041,
      "step": 13055
    },
    {
      "epoch": 0.34385040821701346,
      "grad_norm": 1.9202227592468262,
      "learning_rate": 3.2819331050829604e-05,
      "loss": 1.6563,
      "step": 13056
    },
    {
      "epoch": 0.34387674479852515,
      "grad_norm": 1.792994737625122,
      "learning_rate": 3.281801422175402e-05,
      "loss": 1.588,
      "step": 13057
    },
    {
      "epoch": 0.3439030813800369,
      "grad_norm": 1.7406229972839355,
      "learning_rate": 3.281669739267843e-05,
      "loss": 2.0877,
      "step": 13058
    },
    {
      "epoch": 0.3439294179615486,
      "grad_norm": 2.530524253845215,
      "learning_rate": 3.2815380563602844e-05,
      "loss": 1.5121,
      "step": 13059
    },
    {
      "epoch": 0.3439557545430603,
      "grad_norm": 1.6033681631088257,
      "learning_rate": 3.281406373452726e-05,
      "loss": 1.7437,
      "step": 13060
    },
    {
      "epoch": 0.343982091124572,
      "grad_norm": 3.5794901847839355,
      "learning_rate": 3.2812746905451676e-05,
      "loss": 1.6998,
      "step": 13061
    },
    {
      "epoch": 0.34400842770608375,
      "grad_norm": 1.7232699394226074,
      "learning_rate": 3.2811430076376084e-05,
      "loss": 1.8842,
      "step": 13062
    },
    {
      "epoch": 0.3440347642875955,
      "grad_norm": 2.4835448265075684,
      "learning_rate": 3.28101132473005e-05,
      "loss": 2.1897,
      "step": 13063
    },
    {
      "epoch": 0.3440611008691072,
      "grad_norm": 4.697007179260254,
      "learning_rate": 3.2808796418224916e-05,
      "loss": 1.6079,
      "step": 13064
    },
    {
      "epoch": 0.3440874374506189,
      "grad_norm": 1.7171131372451782,
      "learning_rate": 3.280747958914933e-05,
      "loss": 2.024,
      "step": 13065
    },
    {
      "epoch": 0.3441137740321306,
      "grad_norm": 4.3096022605896,
      "learning_rate": 3.280616276007375e-05,
      "loss": 1.5266,
      "step": 13066
    },
    {
      "epoch": 0.34414011061364236,
      "grad_norm": 3.385385751724243,
      "learning_rate": 3.2804845930998156e-05,
      "loss": 0.9625,
      "step": 13067
    },
    {
      "epoch": 0.34416644719515405,
      "grad_norm": 1.8404408693313599,
      "learning_rate": 3.280352910192257e-05,
      "loss": 1.7764,
      "step": 13068
    },
    {
      "epoch": 0.3441927837766658,
      "grad_norm": 1.7703425884246826,
      "learning_rate": 3.280221227284699e-05,
      "loss": 1.8662,
      "step": 13069
    },
    {
      "epoch": 0.34421912035817753,
      "grad_norm": 2.0454976558685303,
      "learning_rate": 3.28008954437714e-05,
      "loss": 2.2252,
      "step": 13070
    },
    {
      "epoch": 0.3442454569396892,
      "grad_norm": 2.457709312438965,
      "learning_rate": 3.279957861469581e-05,
      "loss": 2.1304,
      "step": 13071
    },
    {
      "epoch": 0.34427179352120096,
      "grad_norm": 1.8219228982925415,
      "learning_rate": 3.279826178562023e-05,
      "loss": 1.4442,
      "step": 13072
    },
    {
      "epoch": 0.34429813010271265,
      "grad_norm": 1.6668879985809326,
      "learning_rate": 3.279694495654464e-05,
      "loss": 2.169,
      "step": 13073
    },
    {
      "epoch": 0.3443244666842244,
      "grad_norm": 1.4907307624816895,
      "learning_rate": 3.279562812746906e-05,
      "loss": 1.8519,
      "step": 13074
    },
    {
      "epoch": 0.3443508032657361,
      "grad_norm": 1.9547961950302124,
      "learning_rate": 3.2794311298393474e-05,
      "loss": 1.5461,
      "step": 13075
    },
    {
      "epoch": 0.3443771398472478,
      "grad_norm": 2.242504596710205,
      "learning_rate": 3.279299446931788e-05,
      "loss": 2.4091,
      "step": 13076
    },
    {
      "epoch": 0.34440347642875957,
      "grad_norm": 4.072727203369141,
      "learning_rate": 3.27916776402423e-05,
      "loss": 2.1891,
      "step": 13077
    },
    {
      "epoch": 0.34442981301027126,
      "grad_norm": 1.9525781869888306,
      "learning_rate": 3.2790360811166714e-05,
      "loss": 1.9789,
      "step": 13078
    },
    {
      "epoch": 0.344456149591783,
      "grad_norm": 1.893464207649231,
      "learning_rate": 3.278904398209113e-05,
      "loss": 0.7665,
      "step": 13079
    },
    {
      "epoch": 0.3444824861732947,
      "grad_norm": 3.1126601696014404,
      "learning_rate": 3.278772715301554e-05,
      "loss": 1.3905,
      "step": 13080
    },
    {
      "epoch": 0.34450882275480643,
      "grad_norm": 1.7459403276443481,
      "learning_rate": 3.2786410323939954e-05,
      "loss": 1.6161,
      "step": 13081
    },
    {
      "epoch": 0.3445351593363181,
      "grad_norm": 1.9930229187011719,
      "learning_rate": 3.278509349486437e-05,
      "loss": 1.4954,
      "step": 13082
    },
    {
      "epoch": 0.34456149591782986,
      "grad_norm": 1.9269921779632568,
      "learning_rate": 3.2783776665788785e-05,
      "loss": 1.9001,
      "step": 13083
    },
    {
      "epoch": 0.3445878324993416,
      "grad_norm": 3.482120990753174,
      "learning_rate": 3.27824598367132e-05,
      "loss": 1.5318,
      "step": 13084
    },
    {
      "epoch": 0.3446141690808533,
      "grad_norm": 1.725644588470459,
      "learning_rate": 3.278114300763761e-05,
      "loss": 1.8467,
      "step": 13085
    },
    {
      "epoch": 0.34464050566236504,
      "grad_norm": 1.6535245180130005,
      "learning_rate": 3.2779826178562025e-05,
      "loss": 2.6959,
      "step": 13086
    },
    {
      "epoch": 0.3446668422438767,
      "grad_norm": 3.432035446166992,
      "learning_rate": 3.2778509349486434e-05,
      "loss": 1.0838,
      "step": 13087
    },
    {
      "epoch": 0.34469317882538847,
      "grad_norm": 2.724940061569214,
      "learning_rate": 3.2777192520410856e-05,
      "loss": 2.7897,
      "step": 13088
    },
    {
      "epoch": 0.3447195154069002,
      "grad_norm": 2.3714377880096436,
      "learning_rate": 3.2775875691335265e-05,
      "loss": 1.0903,
      "step": 13089
    },
    {
      "epoch": 0.3447458519884119,
      "grad_norm": 1.5452818870544434,
      "learning_rate": 3.277455886225968e-05,
      "loss": 2.132,
      "step": 13090
    },
    {
      "epoch": 0.34477218856992364,
      "grad_norm": 1.769982099533081,
      "learning_rate": 3.2773242033184096e-05,
      "loss": 1.8803,
      "step": 13091
    },
    {
      "epoch": 0.34479852515143533,
      "grad_norm": 2.5638909339904785,
      "learning_rate": 3.2771925204108505e-05,
      "loss": 1.83,
      "step": 13092
    },
    {
      "epoch": 0.3448248617329471,
      "grad_norm": 1.5546481609344482,
      "learning_rate": 3.277060837503293e-05,
      "loss": 1.5173,
      "step": 13093
    },
    {
      "epoch": 0.34485119831445876,
      "grad_norm": 6.6082305908203125,
      "learning_rate": 3.2769291545957336e-05,
      "loss": 1.6051,
      "step": 13094
    },
    {
      "epoch": 0.3448775348959705,
      "grad_norm": 2.1726670265197754,
      "learning_rate": 3.276797471688175e-05,
      "loss": 1.6503,
      "step": 13095
    },
    {
      "epoch": 0.34490387147748225,
      "grad_norm": 2.1224615573883057,
      "learning_rate": 3.276665788780616e-05,
      "loss": 1.6912,
      "step": 13096
    },
    {
      "epoch": 0.34493020805899394,
      "grad_norm": 4.337820529937744,
      "learning_rate": 3.276534105873058e-05,
      "loss": 1.7426,
      "step": 13097
    },
    {
      "epoch": 0.3449565446405057,
      "grad_norm": 3.636662244796753,
      "learning_rate": 3.276402422965499e-05,
      "loss": 1.1134,
      "step": 13098
    },
    {
      "epoch": 0.34498288122201737,
      "grad_norm": 2.017798900604248,
      "learning_rate": 3.276270740057941e-05,
      "loss": 1.544,
      "step": 13099
    },
    {
      "epoch": 0.3450092178035291,
      "grad_norm": 2.245283365249634,
      "learning_rate": 3.2761390571503816e-05,
      "loss": 0.43,
      "step": 13100
    },
    {
      "epoch": 0.3450355543850408,
      "grad_norm": 1.9564303159713745,
      "learning_rate": 3.276007374242823e-05,
      "loss": 2.0878,
      "step": 13101
    },
    {
      "epoch": 0.34506189096655254,
      "grad_norm": 1.8316985368728638,
      "learning_rate": 3.275875691335265e-05,
      "loss": 1.8981,
      "step": 13102
    },
    {
      "epoch": 0.3450882275480643,
      "grad_norm": 2.384197950363159,
      "learning_rate": 3.275744008427706e-05,
      "loss": 2.1968,
      "step": 13103
    },
    {
      "epoch": 0.345114564129576,
      "grad_norm": 1.8945741653442383,
      "learning_rate": 3.275612325520148e-05,
      "loss": 0.467,
      "step": 13104
    },
    {
      "epoch": 0.3451409007110877,
      "grad_norm": 4.219059467315674,
      "learning_rate": 3.275480642612589e-05,
      "loss": 1.385,
      "step": 13105
    },
    {
      "epoch": 0.3451672372925994,
      "grad_norm": 1.6160128116607666,
      "learning_rate": 3.27534895970503e-05,
      "loss": 1.9738,
      "step": 13106
    },
    {
      "epoch": 0.34519357387411115,
      "grad_norm": 2.2300801277160645,
      "learning_rate": 3.275217276797472e-05,
      "loss": 1.9553,
      "step": 13107
    },
    {
      "epoch": 0.34521991045562284,
      "grad_norm": 2.761461019515991,
      "learning_rate": 3.2750855938899134e-05,
      "loss": 1.8339,
      "step": 13108
    },
    {
      "epoch": 0.3452462470371346,
      "grad_norm": 1.5213525295257568,
      "learning_rate": 3.274953910982354e-05,
      "loss": 1.2781,
      "step": 13109
    },
    {
      "epoch": 0.3452725836186463,
      "grad_norm": 4.686323165893555,
      "learning_rate": 3.274822228074796e-05,
      "loss": 1.2551,
      "step": 13110
    },
    {
      "epoch": 0.345298920200158,
      "grad_norm": 1.545095682144165,
      "learning_rate": 3.2746905451672374e-05,
      "loss": 1.1122,
      "step": 13111
    },
    {
      "epoch": 0.34532525678166975,
      "grad_norm": 2.2858619689941406,
      "learning_rate": 3.274558862259679e-05,
      "loss": 1.7444,
      "step": 13112
    },
    {
      "epoch": 0.34535159336318144,
      "grad_norm": 5.111452102661133,
      "learning_rate": 3.2744271793521205e-05,
      "loss": 1.7006,
      "step": 13113
    },
    {
      "epoch": 0.3453779299446932,
      "grad_norm": 1.594421625137329,
      "learning_rate": 3.2742954964445614e-05,
      "loss": 1.4181,
      "step": 13114
    },
    {
      "epoch": 0.3454042665262049,
      "grad_norm": 1.8380615711212158,
      "learning_rate": 3.274163813537003e-05,
      "loss": 1.8114,
      "step": 13115
    },
    {
      "epoch": 0.3454306031077166,
      "grad_norm": 2.3619723320007324,
      "learning_rate": 3.2740321306294445e-05,
      "loss": 1.2619,
      "step": 13116
    },
    {
      "epoch": 0.34545693968922836,
      "grad_norm": 2.5805907249450684,
      "learning_rate": 3.273900447721886e-05,
      "loss": 1.5314,
      "step": 13117
    },
    {
      "epoch": 0.34548327627074005,
      "grad_norm": 2.029003858566284,
      "learning_rate": 3.273768764814327e-05,
      "loss": 1.6718,
      "step": 13118
    },
    {
      "epoch": 0.3455096128522518,
      "grad_norm": 3.640721321105957,
      "learning_rate": 3.2736370819067685e-05,
      "loss": 0.8545,
      "step": 13119
    },
    {
      "epoch": 0.3455359494337635,
      "grad_norm": 2.7304906845092773,
      "learning_rate": 3.27350539899921e-05,
      "loss": 1.6673,
      "step": 13120
    },
    {
      "epoch": 0.3455622860152752,
      "grad_norm": 2.343400001525879,
      "learning_rate": 3.2733737160916517e-05,
      "loss": 1.9146,
      "step": 13121
    },
    {
      "epoch": 0.3455886225967869,
      "grad_norm": 3.9495620727539062,
      "learning_rate": 3.273242033184093e-05,
      "loss": 0.7853,
      "step": 13122
    },
    {
      "epoch": 0.34561495917829865,
      "grad_norm": 2.723336935043335,
      "learning_rate": 3.273110350276534e-05,
      "loss": 2.1949,
      "step": 13123
    },
    {
      "epoch": 0.3456412957598104,
      "grad_norm": 2.9893717765808105,
      "learning_rate": 3.2729786673689757e-05,
      "loss": 0.6294,
      "step": 13124
    },
    {
      "epoch": 0.3456676323413221,
      "grad_norm": 3.2053024768829346,
      "learning_rate": 3.2728469844614165e-05,
      "loss": 1.748,
      "step": 13125
    },
    {
      "epoch": 0.3456939689228338,
      "grad_norm": 2.148850917816162,
      "learning_rate": 3.272715301553859e-05,
      "loss": 1.8449,
      "step": 13126
    },
    {
      "epoch": 0.3457203055043455,
      "grad_norm": 1.597330093383789,
      "learning_rate": 3.2725836186462997e-05,
      "loss": 2.1249,
      "step": 13127
    },
    {
      "epoch": 0.34574664208585726,
      "grad_norm": 2.352307081222534,
      "learning_rate": 3.272451935738741e-05,
      "loss": 1.9722,
      "step": 13128
    },
    {
      "epoch": 0.345772978667369,
      "grad_norm": 1.9829521179199219,
      "learning_rate": 3.272320252831183e-05,
      "loss": 1.6498,
      "step": 13129
    },
    {
      "epoch": 0.3457993152488807,
      "grad_norm": 2.9146358966827393,
      "learning_rate": 3.272188569923624e-05,
      "loss": 1.5053,
      "step": 13130
    },
    {
      "epoch": 0.34582565183039243,
      "grad_norm": 2.101445436477661,
      "learning_rate": 3.272056887016066e-05,
      "loss": 1.8205,
      "step": 13131
    },
    {
      "epoch": 0.3458519884119041,
      "grad_norm": 2.021803617477417,
      "learning_rate": 3.271925204108507e-05,
      "loss": 1.5883,
      "step": 13132
    },
    {
      "epoch": 0.34587832499341586,
      "grad_norm": 2.4753849506378174,
      "learning_rate": 3.2717935212009483e-05,
      "loss": 1.9541,
      "step": 13133
    },
    {
      "epoch": 0.34590466157492755,
      "grad_norm": 2.9785354137420654,
      "learning_rate": 3.271661838293389e-05,
      "loss": 1.5607,
      "step": 13134
    },
    {
      "epoch": 0.3459309981564393,
      "grad_norm": 2.653076410293579,
      "learning_rate": 3.2715301553858315e-05,
      "loss": 0.7222,
      "step": 13135
    },
    {
      "epoch": 0.34595733473795104,
      "grad_norm": 4.669103622436523,
      "learning_rate": 3.2713984724782723e-05,
      "loss": 0.6319,
      "step": 13136
    },
    {
      "epoch": 0.3459836713194627,
      "grad_norm": 2.5796475410461426,
      "learning_rate": 3.271266789570714e-05,
      "loss": 1.6803,
      "step": 13137
    },
    {
      "epoch": 0.34601000790097447,
      "grad_norm": 1.9838327169418335,
      "learning_rate": 3.2711351066631555e-05,
      "loss": 2.0145,
      "step": 13138
    },
    {
      "epoch": 0.34603634448248616,
      "grad_norm": 2.317049503326416,
      "learning_rate": 3.2710034237555963e-05,
      "loss": 1.7199,
      "step": 13139
    },
    {
      "epoch": 0.3460626810639979,
      "grad_norm": 1.638168215751648,
      "learning_rate": 3.2708717408480386e-05,
      "loss": 1.2904,
      "step": 13140
    },
    {
      "epoch": 0.3460890176455096,
      "grad_norm": 1.910266399383545,
      "learning_rate": 3.2707400579404795e-05,
      "loss": 0.8098,
      "step": 13141
    },
    {
      "epoch": 0.34611535422702133,
      "grad_norm": 2.1312108039855957,
      "learning_rate": 3.270608375032921e-05,
      "loss": 1.7403,
      "step": 13142
    },
    {
      "epoch": 0.3461416908085331,
      "grad_norm": 2.5649542808532715,
      "learning_rate": 3.270476692125362e-05,
      "loss": 1.4819,
      "step": 13143
    },
    {
      "epoch": 0.34616802739004476,
      "grad_norm": 3.8648250102996826,
      "learning_rate": 3.270345009217804e-05,
      "loss": 1.5417,
      "step": 13144
    },
    {
      "epoch": 0.3461943639715565,
      "grad_norm": 3.1407723426818848,
      "learning_rate": 3.270213326310245e-05,
      "loss": 1.0769,
      "step": 13145
    },
    {
      "epoch": 0.3462207005530682,
      "grad_norm": 3.0411038398742676,
      "learning_rate": 3.2700816434026866e-05,
      "loss": 1.1421,
      "step": 13146
    },
    {
      "epoch": 0.34624703713457994,
      "grad_norm": 3.575984239578247,
      "learning_rate": 3.269949960495128e-05,
      "loss": 2.2352,
      "step": 13147
    },
    {
      "epoch": 0.3462733737160916,
      "grad_norm": 2.0217156410217285,
      "learning_rate": 3.269818277587569e-05,
      "loss": 0.9092,
      "step": 13148
    },
    {
      "epoch": 0.34629971029760337,
      "grad_norm": 1.5097116231918335,
      "learning_rate": 3.269686594680011e-05,
      "loss": 1.5321,
      "step": 13149
    },
    {
      "epoch": 0.3463260468791151,
      "grad_norm": 2.2257752418518066,
      "learning_rate": 3.269554911772452e-05,
      "loss": 2.223,
      "step": 13150
    },
    {
      "epoch": 0.3463523834606268,
      "grad_norm": 1.7590159177780151,
      "learning_rate": 3.269423228864894e-05,
      "loss": 0.8483,
      "step": 13151
    },
    {
      "epoch": 0.34637872004213854,
      "grad_norm": 2.7122015953063965,
      "learning_rate": 3.2692915459573346e-05,
      "loss": 1.8246,
      "step": 13152
    },
    {
      "epoch": 0.34640505662365023,
      "grad_norm": 4.413259983062744,
      "learning_rate": 3.269159863049776e-05,
      "loss": 1.0559,
      "step": 13153
    },
    {
      "epoch": 0.346431393205162,
      "grad_norm": 5.453820705413818,
      "learning_rate": 3.269028180142218e-05,
      "loss": 1.1155,
      "step": 13154
    },
    {
      "epoch": 0.34645772978667366,
      "grad_norm": 3.453749656677246,
      "learning_rate": 3.268896497234659e-05,
      "loss": 2.0343,
      "step": 13155
    },
    {
      "epoch": 0.3464840663681854,
      "grad_norm": 1.9833474159240723,
      "learning_rate": 3.2687648143271e-05,
      "loss": 0.4347,
      "step": 13156
    },
    {
      "epoch": 0.34651040294969715,
      "grad_norm": 2.256859540939331,
      "learning_rate": 3.268633131419542e-05,
      "loss": 0.6473,
      "step": 13157
    },
    {
      "epoch": 0.34653673953120884,
      "grad_norm": 1.7505801916122437,
      "learning_rate": 3.268501448511983e-05,
      "loss": 1.6198,
      "step": 13158
    },
    {
      "epoch": 0.3465630761127206,
      "grad_norm": 1.9900546073913574,
      "learning_rate": 3.268369765604425e-05,
      "loss": 1.9354,
      "step": 13159
    },
    {
      "epoch": 0.34658941269423227,
      "grad_norm": 2.2289633750915527,
      "learning_rate": 3.2682380826968664e-05,
      "loss": 1.5528,
      "step": 13160
    },
    {
      "epoch": 0.346615749275744,
      "grad_norm": 1.8603487014770508,
      "learning_rate": 3.268106399789307e-05,
      "loss": 2.3699,
      "step": 13161
    },
    {
      "epoch": 0.34664208585725576,
      "grad_norm": 3.21364164352417,
      "learning_rate": 3.267974716881749e-05,
      "loss": 1.2129,
      "step": 13162
    },
    {
      "epoch": 0.34666842243876744,
      "grad_norm": 3.918067455291748,
      "learning_rate": 3.2678430339741904e-05,
      "loss": 1.1659,
      "step": 13163
    },
    {
      "epoch": 0.3466947590202792,
      "grad_norm": 1.5441951751708984,
      "learning_rate": 3.267711351066632e-05,
      "loss": 1.3983,
      "step": 13164
    },
    {
      "epoch": 0.3467210956017909,
      "grad_norm": 2.0096254348754883,
      "learning_rate": 3.267579668159073e-05,
      "loss": 0.4344,
      "step": 13165
    },
    {
      "epoch": 0.3467474321833026,
      "grad_norm": 2.4570610523223877,
      "learning_rate": 3.2674479852515144e-05,
      "loss": 1.6607,
      "step": 13166
    },
    {
      "epoch": 0.3467737687648143,
      "grad_norm": 2.5036914348602295,
      "learning_rate": 3.267316302343956e-05,
      "loss": 0.5084,
      "step": 13167
    },
    {
      "epoch": 0.34680010534632605,
      "grad_norm": 3.401029348373413,
      "learning_rate": 3.2671846194363975e-05,
      "loss": 2.0058,
      "step": 13168
    },
    {
      "epoch": 0.3468264419278378,
      "grad_norm": 1.9968043565750122,
      "learning_rate": 3.267052936528839e-05,
      "loss": 2.4615,
      "step": 13169
    },
    {
      "epoch": 0.3468527785093495,
      "grad_norm": 1.5256257057189941,
      "learning_rate": 3.26692125362128e-05,
      "loss": 1.7771,
      "step": 13170
    },
    {
      "epoch": 0.3468791150908612,
      "grad_norm": 1.498011827468872,
      "learning_rate": 3.2667895707137215e-05,
      "loss": 1.7604,
      "step": 13171
    },
    {
      "epoch": 0.3469054516723729,
      "grad_norm": 1.7540767192840576,
      "learning_rate": 3.2666578878061624e-05,
      "loss": 1.9506,
      "step": 13172
    },
    {
      "epoch": 0.34693178825388465,
      "grad_norm": 4.443070411682129,
      "learning_rate": 3.2665262048986046e-05,
      "loss": 1.3594,
      "step": 13173
    },
    {
      "epoch": 0.34695812483539634,
      "grad_norm": 1.8055349588394165,
      "learning_rate": 3.2663945219910455e-05,
      "loss": 1.6866,
      "step": 13174
    },
    {
      "epoch": 0.3469844614169081,
      "grad_norm": 3.2847204208374023,
      "learning_rate": 3.266262839083487e-05,
      "loss": 0.4526,
      "step": 13175
    },
    {
      "epoch": 0.34701079799841983,
      "grad_norm": 1.8152081966400146,
      "learning_rate": 3.2661311561759286e-05,
      "loss": 2.342,
      "step": 13176
    },
    {
      "epoch": 0.3470371345799315,
      "grad_norm": 4.156172275543213,
      "learning_rate": 3.26599947326837e-05,
      "loss": 1.5455,
      "step": 13177
    },
    {
      "epoch": 0.34706347116144326,
      "grad_norm": 3.141244888305664,
      "learning_rate": 3.265867790360812e-05,
      "loss": 2.1393,
      "step": 13178
    },
    {
      "epoch": 0.34708980774295495,
      "grad_norm": 1.998428463935852,
      "learning_rate": 3.2657361074532526e-05,
      "loss": 0.3482,
      "step": 13179
    },
    {
      "epoch": 0.3471161443244667,
      "grad_norm": 1.8900747299194336,
      "learning_rate": 3.265604424545694e-05,
      "loss": 2.0578,
      "step": 13180
    },
    {
      "epoch": 0.3471424809059784,
      "grad_norm": 1.4871665239334106,
      "learning_rate": 3.265472741638135e-05,
      "loss": 0.9511,
      "step": 13181
    },
    {
      "epoch": 0.3471688174874901,
      "grad_norm": 1.6186096668243408,
      "learning_rate": 3.265341058730577e-05,
      "loss": 2.3809,
      "step": 13182
    },
    {
      "epoch": 0.34719515406900187,
      "grad_norm": 2.7033965587615967,
      "learning_rate": 3.265209375823018e-05,
      "loss": 1.4537,
      "step": 13183
    },
    {
      "epoch": 0.34722149065051355,
      "grad_norm": 2.3884329795837402,
      "learning_rate": 3.26507769291546e-05,
      "loss": 2.3727,
      "step": 13184
    },
    {
      "epoch": 0.3472478272320253,
      "grad_norm": 2.401170492172241,
      "learning_rate": 3.264946010007901e-05,
      "loss": 2.3979,
      "step": 13185
    },
    {
      "epoch": 0.347274163813537,
      "grad_norm": 1.9125306606292725,
      "learning_rate": 3.264814327100342e-05,
      "loss": 1.0401,
      "step": 13186
    },
    {
      "epoch": 0.34730050039504873,
      "grad_norm": 3.3054792881011963,
      "learning_rate": 3.2646826441927844e-05,
      "loss": 1.7221,
      "step": 13187
    },
    {
      "epoch": 0.3473268369765604,
      "grad_norm": 1.8396989107131958,
      "learning_rate": 3.264550961285225e-05,
      "loss": 2.0625,
      "step": 13188
    },
    {
      "epoch": 0.34735317355807216,
      "grad_norm": 3.386190891265869,
      "learning_rate": 3.264419278377667e-05,
      "loss": 1.1385,
      "step": 13189
    },
    {
      "epoch": 0.3473795101395839,
      "grad_norm": 2.0150368213653564,
      "learning_rate": 3.264287595470108e-05,
      "loss": 1.9855,
      "step": 13190
    },
    {
      "epoch": 0.3474058467210956,
      "grad_norm": 4.199228286743164,
      "learning_rate": 3.264155912562549e-05,
      "loss": 0.7868,
      "step": 13191
    },
    {
      "epoch": 0.34743218330260733,
      "grad_norm": 1.7155773639678955,
      "learning_rate": 3.264024229654991e-05,
      "loss": 2.4113,
      "step": 13192
    },
    {
      "epoch": 0.347458519884119,
      "grad_norm": 4.085933208465576,
      "learning_rate": 3.2638925467474324e-05,
      "loss": 1.4366,
      "step": 13193
    },
    {
      "epoch": 0.34748485646563076,
      "grad_norm": 1.725295901298523,
      "learning_rate": 3.263760863839874e-05,
      "loss": 1.8667,
      "step": 13194
    },
    {
      "epoch": 0.3475111930471425,
      "grad_norm": 2.126094341278076,
      "learning_rate": 3.263629180932315e-05,
      "loss": 1.7663,
      "step": 13195
    },
    {
      "epoch": 0.3475375296286542,
      "grad_norm": 2.895192861557007,
      "learning_rate": 3.263497498024757e-05,
      "loss": 1.2456,
      "step": 13196
    },
    {
      "epoch": 0.34756386621016594,
      "grad_norm": 4.887364387512207,
      "learning_rate": 3.263365815117198e-05,
      "loss": 2.3179,
      "step": 13197
    },
    {
      "epoch": 0.3475902027916776,
      "grad_norm": 2.686774253845215,
      "learning_rate": 3.2632341322096396e-05,
      "loss": 1.737,
      "step": 13198
    },
    {
      "epoch": 0.34761653937318937,
      "grad_norm": 1.8099535703659058,
      "learning_rate": 3.2631024493020804e-05,
      "loss": 1.4543,
      "step": 13199
    },
    {
      "epoch": 0.34764287595470106,
      "grad_norm": 2.709017038345337,
      "learning_rate": 3.262970766394522e-05,
      "loss": 1.5653,
      "step": 13200
    },
    {
      "epoch": 0.3476692125362128,
      "grad_norm": 2.230703592300415,
      "learning_rate": 3.2628390834869636e-05,
      "loss": 1.7813,
      "step": 13201
    },
    {
      "epoch": 0.34769554911772454,
      "grad_norm": 2.315885543823242,
      "learning_rate": 3.262707400579405e-05,
      "loss": 1.67,
      "step": 13202
    },
    {
      "epoch": 0.34772188569923623,
      "grad_norm": 2.4082462787628174,
      "learning_rate": 3.262575717671846e-05,
      "loss": 1.6513,
      "step": 13203
    },
    {
      "epoch": 0.347748222280748,
      "grad_norm": 7.442232131958008,
      "learning_rate": 3.2624440347642876e-05,
      "loss": 1.8802,
      "step": 13204
    },
    {
      "epoch": 0.34777455886225966,
      "grad_norm": 1.9536820650100708,
      "learning_rate": 3.262312351856729e-05,
      "loss": 2.2042,
      "step": 13205
    },
    {
      "epoch": 0.3478008954437714,
      "grad_norm": 3.228766679763794,
      "learning_rate": 3.262180668949171e-05,
      "loss": 1.863,
      "step": 13206
    },
    {
      "epoch": 0.3478272320252831,
      "grad_norm": 1.9316649436950684,
      "learning_rate": 3.262048986041612e-05,
      "loss": 2.1574,
      "step": 13207
    },
    {
      "epoch": 0.34785356860679484,
      "grad_norm": 1.660380244255066,
      "learning_rate": 3.261917303134053e-05,
      "loss": 1.8578,
      "step": 13208
    },
    {
      "epoch": 0.3478799051883066,
      "grad_norm": 1.4779479503631592,
      "learning_rate": 3.261785620226495e-05,
      "loss": 0.3396,
      "step": 13209
    },
    {
      "epoch": 0.34790624176981827,
      "grad_norm": 2.8942437171936035,
      "learning_rate": 3.261653937318936e-05,
      "loss": 0.5271,
      "step": 13210
    },
    {
      "epoch": 0.34793257835133,
      "grad_norm": 2.0104780197143555,
      "learning_rate": 3.261522254411378e-05,
      "loss": 2.1426,
      "step": 13211
    },
    {
      "epoch": 0.3479589149328417,
      "grad_norm": 1.9063493013381958,
      "learning_rate": 3.261390571503819e-05,
      "loss": 0.3851,
      "step": 13212
    },
    {
      "epoch": 0.34798525151435344,
      "grad_norm": 2.0151333808898926,
      "learning_rate": 3.26125888859626e-05,
      "loss": 1.5941,
      "step": 13213
    },
    {
      "epoch": 0.34801158809586513,
      "grad_norm": 1.9707542657852173,
      "learning_rate": 3.261127205688702e-05,
      "loss": 1.3774,
      "step": 13214
    },
    {
      "epoch": 0.3480379246773769,
      "grad_norm": 2.3949368000030518,
      "learning_rate": 3.2609955227811434e-05,
      "loss": 1.7263,
      "step": 13215
    },
    {
      "epoch": 0.3480642612588886,
      "grad_norm": 1.9590846300125122,
      "learning_rate": 3.260863839873585e-05,
      "loss": 1.8277,
      "step": 13216
    },
    {
      "epoch": 0.3480905978404003,
      "grad_norm": 2.1931588649749756,
      "learning_rate": 3.260732156966026e-05,
      "loss": 1.124,
      "step": 13217
    },
    {
      "epoch": 0.34811693442191205,
      "grad_norm": 2.7168567180633545,
      "learning_rate": 3.2606004740584674e-05,
      "loss": 1.2126,
      "step": 13218
    },
    {
      "epoch": 0.34814327100342374,
      "grad_norm": 2.025538206100464,
      "learning_rate": 3.260468791150908e-05,
      "loss": 1.4572,
      "step": 13219
    },
    {
      "epoch": 0.3481696075849355,
      "grad_norm": 3.5675971508026123,
      "learning_rate": 3.2603371082433505e-05,
      "loss": 1.3672,
      "step": 13220
    },
    {
      "epoch": 0.34819594416644717,
      "grad_norm": 1.7756645679473877,
      "learning_rate": 3.2602054253357914e-05,
      "loss": 1.5728,
      "step": 13221
    },
    {
      "epoch": 0.3482222807479589,
      "grad_norm": 1.990281581878662,
      "learning_rate": 3.260073742428233e-05,
      "loss": 1.4883,
      "step": 13222
    },
    {
      "epoch": 0.34824861732947066,
      "grad_norm": 1.7624545097351074,
      "learning_rate": 3.2599420595206745e-05,
      "loss": 1.85,
      "step": 13223
    },
    {
      "epoch": 0.34827495391098234,
      "grad_norm": 3.2606406211853027,
      "learning_rate": 3.2598103766131154e-05,
      "loss": 1.872,
      "step": 13224
    },
    {
      "epoch": 0.3483012904924941,
      "grad_norm": 3.6914162635803223,
      "learning_rate": 3.2596786937055576e-05,
      "loss": 1.2663,
      "step": 13225
    },
    {
      "epoch": 0.3483276270740058,
      "grad_norm": 2.4079458713531494,
      "learning_rate": 3.2595470107979985e-05,
      "loss": 1.5693,
      "step": 13226
    },
    {
      "epoch": 0.3483539636555175,
      "grad_norm": 3.0475223064422607,
      "learning_rate": 3.25941532789044e-05,
      "loss": 2.0814,
      "step": 13227
    },
    {
      "epoch": 0.34838030023702926,
      "grad_norm": 2.0033042430877686,
      "learning_rate": 3.259283644982881e-05,
      "loss": 1.902,
      "step": 13228
    },
    {
      "epoch": 0.34840663681854095,
      "grad_norm": 2.124267101287842,
      "learning_rate": 3.259151962075323e-05,
      "loss": 0.7724,
      "step": 13229
    },
    {
      "epoch": 0.3484329734000527,
      "grad_norm": 3.014657974243164,
      "learning_rate": 3.259020279167764e-05,
      "loss": 1.4377,
      "step": 13230
    },
    {
      "epoch": 0.3484593099815644,
      "grad_norm": 2.3430721759796143,
      "learning_rate": 3.2588885962602056e-05,
      "loss": 1.7488,
      "step": 13231
    },
    {
      "epoch": 0.3484856465630761,
      "grad_norm": 1.9165147542953491,
      "learning_rate": 3.258756913352647e-05,
      "loss": 2.018,
      "step": 13232
    },
    {
      "epoch": 0.3485119831445878,
      "grad_norm": 2.3940341472625732,
      "learning_rate": 3.258625230445088e-05,
      "loss": 1.2182,
      "step": 13233
    },
    {
      "epoch": 0.34853831972609955,
      "grad_norm": 2.1990294456481934,
      "learning_rate": 3.25849354753753e-05,
      "loss": 1.9036,
      "step": 13234
    },
    {
      "epoch": 0.3485646563076113,
      "grad_norm": 2.697700262069702,
      "learning_rate": 3.258361864629971e-05,
      "loss": 1.5049,
      "step": 13235
    },
    {
      "epoch": 0.348590992889123,
      "grad_norm": 3.199383497238159,
      "learning_rate": 3.258230181722413e-05,
      "loss": 1.9841,
      "step": 13236
    },
    {
      "epoch": 0.34861732947063473,
      "grad_norm": 2.8425440788269043,
      "learning_rate": 3.2580984988148536e-05,
      "loss": 2.4673,
      "step": 13237
    },
    {
      "epoch": 0.3486436660521464,
      "grad_norm": 2.4211385250091553,
      "learning_rate": 3.257966815907295e-05,
      "loss": 1.9615,
      "step": 13238
    },
    {
      "epoch": 0.34867000263365816,
      "grad_norm": 3.0204195976257324,
      "learning_rate": 3.257835132999737e-05,
      "loss": 0.7215,
      "step": 13239
    },
    {
      "epoch": 0.34869633921516985,
      "grad_norm": 1.7180991172790527,
      "learning_rate": 3.257703450092178e-05,
      "loss": 2.0885,
      "step": 13240
    },
    {
      "epoch": 0.3487226757966816,
      "grad_norm": 9.043501853942871,
      "learning_rate": 3.25757176718462e-05,
      "loss": 2.3929,
      "step": 13241
    },
    {
      "epoch": 0.34874901237819333,
      "grad_norm": 1.855078101158142,
      "learning_rate": 3.257440084277061e-05,
      "loss": 1.5952,
      "step": 13242
    },
    {
      "epoch": 0.348775348959705,
      "grad_norm": 2.160182237625122,
      "learning_rate": 3.257308401369503e-05,
      "loss": 0.4405,
      "step": 13243
    },
    {
      "epoch": 0.34880168554121677,
      "grad_norm": 1.6421520709991455,
      "learning_rate": 3.257176718461944e-05,
      "loss": 1.5546,
      "step": 13244
    },
    {
      "epoch": 0.34882802212272845,
      "grad_norm": 3.1143510341644287,
      "learning_rate": 3.2570450355543854e-05,
      "loss": 1.4497,
      "step": 13245
    },
    {
      "epoch": 0.3488543587042402,
      "grad_norm": 3.018958330154419,
      "learning_rate": 3.256913352646826e-05,
      "loss": 1.6898,
      "step": 13246
    },
    {
      "epoch": 0.3488806952857519,
      "grad_norm": 5.543989658355713,
      "learning_rate": 3.256781669739268e-05,
      "loss": 0.9067,
      "step": 13247
    },
    {
      "epoch": 0.34890703186726363,
      "grad_norm": 1.92924964427948,
      "learning_rate": 3.2566499868317094e-05,
      "loss": 1.7132,
      "step": 13248
    },
    {
      "epoch": 0.34893336844877537,
      "grad_norm": 1.6289355754852295,
      "learning_rate": 3.256518303924151e-05,
      "loss": 1.461,
      "step": 13249
    },
    {
      "epoch": 0.34895970503028706,
      "grad_norm": 1.9468270540237427,
      "learning_rate": 3.2563866210165925e-05,
      "loss": 1.6564,
      "step": 13250
    },
    {
      "epoch": 0.3489860416117988,
      "grad_norm": 1.8982024192810059,
      "learning_rate": 3.2562549381090334e-05,
      "loss": 1.5647,
      "step": 13251
    },
    {
      "epoch": 0.3490123781933105,
      "grad_norm": 3.0467891693115234,
      "learning_rate": 3.256123255201475e-05,
      "loss": 1.7748,
      "step": 13252
    },
    {
      "epoch": 0.34903871477482223,
      "grad_norm": 1.782957911491394,
      "learning_rate": 3.2559915722939165e-05,
      "loss": 2.0367,
      "step": 13253
    },
    {
      "epoch": 0.3490650513563339,
      "grad_norm": 10.435483932495117,
      "learning_rate": 3.255859889386358e-05,
      "loss": 3.0346,
      "step": 13254
    },
    {
      "epoch": 0.34909138793784567,
      "grad_norm": 2.1778833866119385,
      "learning_rate": 3.255728206478799e-05,
      "loss": 1.4659,
      "step": 13255
    },
    {
      "epoch": 0.3491177245193574,
      "grad_norm": 2.911656141281128,
      "learning_rate": 3.2555965235712405e-05,
      "loss": 0.8559,
      "step": 13256
    },
    {
      "epoch": 0.3491440611008691,
      "grad_norm": 1.9022785425186157,
      "learning_rate": 3.2554648406636814e-05,
      "loss": 1.4912,
      "step": 13257
    },
    {
      "epoch": 0.34917039768238084,
      "grad_norm": 2.2376620769500732,
      "learning_rate": 3.2553331577561237e-05,
      "loss": 2.1282,
      "step": 13258
    },
    {
      "epoch": 0.3491967342638925,
      "grad_norm": 2.574781894683838,
      "learning_rate": 3.2552014748485645e-05,
      "loss": 1.3146,
      "step": 13259
    },
    {
      "epoch": 0.34922307084540427,
      "grad_norm": 2.566833019256592,
      "learning_rate": 3.255069791941006e-05,
      "loss": 1.9038,
      "step": 13260
    },
    {
      "epoch": 0.34924940742691596,
      "grad_norm": 3.688613176345825,
      "learning_rate": 3.254938109033448e-05,
      "loss": 1.5275,
      "step": 13261
    },
    {
      "epoch": 0.3492757440084277,
      "grad_norm": 2.2053143978118896,
      "learning_rate": 3.254806426125889e-05,
      "loss": 1.5929,
      "step": 13262
    },
    {
      "epoch": 0.34930208058993945,
      "grad_norm": 1.8571240901947021,
      "learning_rate": 3.254674743218331e-05,
      "loss": 1.4883,
      "step": 13263
    },
    {
      "epoch": 0.34932841717145113,
      "grad_norm": 1.7839723825454712,
      "learning_rate": 3.254543060310772e-05,
      "loss": 0.7762,
      "step": 13264
    },
    {
      "epoch": 0.3493547537529629,
      "grad_norm": 1.915321707725525,
      "learning_rate": 3.254411377403213e-05,
      "loss": 2.0091,
      "step": 13265
    },
    {
      "epoch": 0.34938109033447456,
      "grad_norm": 1.7027623653411865,
      "learning_rate": 3.254279694495654e-05,
      "loss": 1.6055,
      "step": 13266
    },
    {
      "epoch": 0.3494074269159863,
      "grad_norm": 1.6140773296356201,
      "learning_rate": 3.2541480115880963e-05,
      "loss": 1.0359,
      "step": 13267
    },
    {
      "epoch": 0.34943376349749805,
      "grad_norm": 2.3103713989257812,
      "learning_rate": 3.254016328680537e-05,
      "loss": 1.8346,
      "step": 13268
    },
    {
      "epoch": 0.34946010007900974,
      "grad_norm": 2.0320658683776855,
      "learning_rate": 3.253884645772979e-05,
      "loss": 1.7831,
      "step": 13269
    },
    {
      "epoch": 0.3494864366605215,
      "grad_norm": 1.7528281211853027,
      "learning_rate": 3.2537529628654203e-05,
      "loss": 1.5211,
      "step": 13270
    },
    {
      "epoch": 0.34951277324203317,
      "grad_norm": 4.402647018432617,
      "learning_rate": 3.253621279957861e-05,
      "loss": 2.6433,
      "step": 13271
    },
    {
      "epoch": 0.3495391098235449,
      "grad_norm": 5.450698375701904,
      "learning_rate": 3.2534895970503035e-05,
      "loss": 1.4586,
      "step": 13272
    },
    {
      "epoch": 0.3495654464050566,
      "grad_norm": 1.3967461585998535,
      "learning_rate": 3.2533579141427443e-05,
      "loss": 1.3089,
      "step": 13273
    },
    {
      "epoch": 0.34959178298656834,
      "grad_norm": 2.8539626598358154,
      "learning_rate": 3.253226231235186e-05,
      "loss": 1.6664,
      "step": 13274
    },
    {
      "epoch": 0.3496181195680801,
      "grad_norm": 2.060029983520508,
      "learning_rate": 3.253094548327627e-05,
      "loss": 0.2958,
      "step": 13275
    },
    {
      "epoch": 0.3496444561495918,
      "grad_norm": 5.360213279724121,
      "learning_rate": 3.252962865420069e-05,
      "loss": 1.3407,
      "step": 13276
    },
    {
      "epoch": 0.3496707927311035,
      "grad_norm": 1.8764805793762207,
      "learning_rate": 3.25283118251251e-05,
      "loss": 2.253,
      "step": 13277
    },
    {
      "epoch": 0.3496971293126152,
      "grad_norm": 2.882615089416504,
      "learning_rate": 3.2526994996049515e-05,
      "loss": 2.2276,
      "step": 13278
    },
    {
      "epoch": 0.34972346589412695,
      "grad_norm": 2.2269115447998047,
      "learning_rate": 3.252567816697393e-05,
      "loss": 2.0075,
      "step": 13279
    },
    {
      "epoch": 0.34974980247563864,
      "grad_norm": 2.1476926803588867,
      "learning_rate": 3.252436133789834e-05,
      "loss": 0.8737,
      "step": 13280
    },
    {
      "epoch": 0.3497761390571504,
      "grad_norm": 2.3212385177612305,
      "learning_rate": 3.252304450882276e-05,
      "loss": 1.7872,
      "step": 13281
    },
    {
      "epoch": 0.3498024756386621,
      "grad_norm": 1.9313105344772339,
      "learning_rate": 3.252172767974717e-05,
      "loss": 2.3175,
      "step": 13282
    },
    {
      "epoch": 0.3498288122201738,
      "grad_norm": 5.24793815612793,
      "learning_rate": 3.2520410850671586e-05,
      "loss": 1.6308,
      "step": 13283
    },
    {
      "epoch": 0.34985514880168556,
      "grad_norm": 5.203807830810547,
      "learning_rate": 3.2519094021595995e-05,
      "loss": 1.5655,
      "step": 13284
    },
    {
      "epoch": 0.34988148538319724,
      "grad_norm": 1.610917329788208,
      "learning_rate": 3.251777719252041e-05,
      "loss": 1.7317,
      "step": 13285
    },
    {
      "epoch": 0.349907821964709,
      "grad_norm": 3.397705316543579,
      "learning_rate": 3.2516460363444826e-05,
      "loss": 0.3327,
      "step": 13286
    },
    {
      "epoch": 0.3499341585462207,
      "grad_norm": 2.4374136924743652,
      "learning_rate": 3.251514353436924e-05,
      "loss": 2.3435,
      "step": 13287
    },
    {
      "epoch": 0.3499604951277324,
      "grad_norm": 1.9728240966796875,
      "learning_rate": 3.251382670529366e-05,
      "loss": 1.6025,
      "step": 13288
    },
    {
      "epoch": 0.34998683170924416,
      "grad_norm": 1.6892491579055786,
      "learning_rate": 3.2512509876218066e-05,
      "loss": 2.0315,
      "step": 13289
    },
    {
      "epoch": 0.35001316829075585,
      "grad_norm": 3.261169195175171,
      "learning_rate": 3.251119304714248e-05,
      "loss": 2.3634,
      "step": 13290
    },
    {
      "epoch": 0.3500395048722676,
      "grad_norm": 2.568793296813965,
      "learning_rate": 3.25098762180669e-05,
      "loss": 1.3906,
      "step": 13291
    },
    {
      "epoch": 0.3500658414537793,
      "grad_norm": 2.0497589111328125,
      "learning_rate": 3.250855938899131e-05,
      "loss": 2.0864,
      "step": 13292
    },
    {
      "epoch": 0.350092178035291,
      "grad_norm": 2.225813627243042,
      "learning_rate": 3.250724255991572e-05,
      "loss": 0.2089,
      "step": 13293
    },
    {
      "epoch": 0.3501185146168027,
      "grad_norm": 2.0244300365448,
      "learning_rate": 3.250592573084014e-05,
      "loss": 1.4985,
      "step": 13294
    },
    {
      "epoch": 0.35014485119831446,
      "grad_norm": 2.8434624671936035,
      "learning_rate": 3.250460890176455e-05,
      "loss": 0.8789,
      "step": 13295
    },
    {
      "epoch": 0.3501711877798262,
      "grad_norm": 1.7693883180618286,
      "learning_rate": 3.250329207268897e-05,
      "loss": 2.1361,
      "step": 13296
    },
    {
      "epoch": 0.3501975243613379,
      "grad_norm": 2.94120454788208,
      "learning_rate": 3.2501975243613384e-05,
      "loss": 2.1523,
      "step": 13297
    },
    {
      "epoch": 0.35022386094284963,
      "grad_norm": 2.399862766265869,
      "learning_rate": 3.250065841453779e-05,
      "loss": 1.3294,
      "step": 13298
    },
    {
      "epoch": 0.3502501975243613,
      "grad_norm": 2.7700185775756836,
      "learning_rate": 3.249934158546221e-05,
      "loss": 1.2198,
      "step": 13299
    },
    {
      "epoch": 0.35027653410587306,
      "grad_norm": 2.1279079914093018,
      "learning_rate": 3.2498024756386624e-05,
      "loss": 1.683,
      "step": 13300
    },
    {
      "epoch": 0.3503028706873848,
      "grad_norm": 3.119107723236084,
      "learning_rate": 3.249670792731104e-05,
      "loss": 0.5042,
      "step": 13301
    },
    {
      "epoch": 0.3503292072688965,
      "grad_norm": 3.5494678020477295,
      "learning_rate": 3.249539109823545e-05,
      "loss": 1.184,
      "step": 13302
    },
    {
      "epoch": 0.35035554385040824,
      "grad_norm": 2.842054843902588,
      "learning_rate": 3.2494074269159864e-05,
      "loss": 1.5409,
      "step": 13303
    },
    {
      "epoch": 0.3503818804319199,
      "grad_norm": 1.8012070655822754,
      "learning_rate": 3.249275744008427e-05,
      "loss": 2.1819,
      "step": 13304
    },
    {
      "epoch": 0.35040821701343167,
      "grad_norm": 3.4269027709960938,
      "learning_rate": 3.2491440611008695e-05,
      "loss": 1.6984,
      "step": 13305
    },
    {
      "epoch": 0.35043455359494335,
      "grad_norm": 1.6999813318252563,
      "learning_rate": 3.2490123781933104e-05,
      "loss": 1.9599,
      "step": 13306
    },
    {
      "epoch": 0.3504608901764551,
      "grad_norm": 2.2925491333007812,
      "learning_rate": 3.248880695285752e-05,
      "loss": 2.3996,
      "step": 13307
    },
    {
      "epoch": 0.35048722675796684,
      "grad_norm": 2.1514554023742676,
      "learning_rate": 3.2487490123781935e-05,
      "loss": 1.7487,
      "step": 13308
    },
    {
      "epoch": 0.35051356333947853,
      "grad_norm": 3.124458074569702,
      "learning_rate": 3.248617329470635e-05,
      "loss": 1.66,
      "step": 13309
    },
    {
      "epoch": 0.35053989992099027,
      "grad_norm": 2.718674659729004,
      "learning_rate": 3.2484856465630766e-05,
      "loss": 1.3637,
      "step": 13310
    },
    {
      "epoch": 0.35056623650250196,
      "grad_norm": 4.2831196784973145,
      "learning_rate": 3.2483539636555175e-05,
      "loss": 1.4355,
      "step": 13311
    },
    {
      "epoch": 0.3505925730840137,
      "grad_norm": 2.0692555904388428,
      "learning_rate": 3.248222280747959e-05,
      "loss": 2.0315,
      "step": 13312
    },
    {
      "epoch": 0.3506189096655254,
      "grad_norm": 3.712114095687866,
      "learning_rate": 3.2480905978404e-05,
      "loss": 1.8928,
      "step": 13313
    },
    {
      "epoch": 0.35064524624703713,
      "grad_norm": 1.9922786951065063,
      "learning_rate": 3.247958914932842e-05,
      "loss": 1.3376,
      "step": 13314
    },
    {
      "epoch": 0.3506715828285489,
      "grad_norm": 3.8773305416107178,
      "learning_rate": 3.247827232025283e-05,
      "loss": 1.6715,
      "step": 13315
    },
    {
      "epoch": 0.35069791941006057,
      "grad_norm": 3.6027517318725586,
      "learning_rate": 3.2476955491177246e-05,
      "loss": 1.9208,
      "step": 13316
    },
    {
      "epoch": 0.3507242559915723,
      "grad_norm": 3.8815882205963135,
      "learning_rate": 3.247563866210166e-05,
      "loss": 1.7572,
      "step": 13317
    },
    {
      "epoch": 0.350750592573084,
      "grad_norm": 2.51364803314209,
      "learning_rate": 3.247432183302607e-05,
      "loss": 1.6966,
      "step": 13318
    },
    {
      "epoch": 0.35077692915459574,
      "grad_norm": 1.8374290466308594,
      "learning_rate": 3.247300500395049e-05,
      "loss": 1.0813,
      "step": 13319
    },
    {
      "epoch": 0.35080326573610743,
      "grad_norm": 2.029656171798706,
      "learning_rate": 3.24716881748749e-05,
      "loss": 2.0684,
      "step": 13320
    },
    {
      "epoch": 0.35082960231761917,
      "grad_norm": 1.9576358795166016,
      "learning_rate": 3.247037134579932e-05,
      "loss": 1.8653,
      "step": 13321
    },
    {
      "epoch": 0.3508559388991309,
      "grad_norm": 6.42474889755249,
      "learning_rate": 3.2469054516723726e-05,
      "loss": 0.8853,
      "step": 13322
    },
    {
      "epoch": 0.3508822754806426,
      "grad_norm": 1.848995327949524,
      "learning_rate": 3.246773768764814e-05,
      "loss": 1.4848,
      "step": 13323
    },
    {
      "epoch": 0.35090861206215435,
      "grad_norm": 1.9281425476074219,
      "learning_rate": 3.246642085857256e-05,
      "loss": 1.9347,
      "step": 13324
    },
    {
      "epoch": 0.35093494864366603,
      "grad_norm": 4.627933502197266,
      "learning_rate": 3.246510402949697e-05,
      "loss": 2.0178,
      "step": 13325
    },
    {
      "epoch": 0.3509612852251778,
      "grad_norm": 3.9753177165985107,
      "learning_rate": 3.246378720042139e-05,
      "loss": 2.6702,
      "step": 13326
    },
    {
      "epoch": 0.35098762180668946,
      "grad_norm": 2.228876829147339,
      "learning_rate": 3.24624703713458e-05,
      "loss": 0.9303,
      "step": 13327
    },
    {
      "epoch": 0.3510139583882012,
      "grad_norm": 4.0302629470825195,
      "learning_rate": 3.246115354227022e-05,
      "loss": 1.5857,
      "step": 13328
    },
    {
      "epoch": 0.35104029496971295,
      "grad_norm": 6.6096510887146,
      "learning_rate": 3.245983671319463e-05,
      "loss": 1.6103,
      "step": 13329
    },
    {
      "epoch": 0.35106663155122464,
      "grad_norm": 2.4697418212890625,
      "learning_rate": 3.2458519884119044e-05,
      "loss": 2.5915,
      "step": 13330
    },
    {
      "epoch": 0.3510929681327364,
      "grad_norm": 2.167027711868286,
      "learning_rate": 3.245720305504345e-05,
      "loss": 0.9266,
      "step": 13331
    },
    {
      "epoch": 0.35111930471424807,
      "grad_norm": 2.2277779579162598,
      "learning_rate": 3.245588622596787e-05,
      "loss": 0.7662,
      "step": 13332
    },
    {
      "epoch": 0.3511456412957598,
      "grad_norm": 5.894300937652588,
      "learning_rate": 3.2454569396892284e-05,
      "loss": 3.318,
      "step": 13333
    },
    {
      "epoch": 0.35117197787727156,
      "grad_norm": 1.594135046005249,
      "learning_rate": 3.24532525678167e-05,
      "loss": 1.6331,
      "step": 13334
    },
    {
      "epoch": 0.35119831445878325,
      "grad_norm": 2.1567468643188477,
      "learning_rate": 3.2451935738741116e-05,
      "loss": 1.8917,
      "step": 13335
    },
    {
      "epoch": 0.351224651040295,
      "grad_norm": 1.9755929708480835,
      "learning_rate": 3.2450618909665524e-05,
      "loss": 1.5999,
      "step": 13336
    },
    {
      "epoch": 0.3512509876218067,
      "grad_norm": 4.384850025177002,
      "learning_rate": 3.244930208058994e-05,
      "loss": 1.1528,
      "step": 13337
    },
    {
      "epoch": 0.3512773242033184,
      "grad_norm": 2.446580648422241,
      "learning_rate": 3.2447985251514356e-05,
      "loss": 1.6648,
      "step": 13338
    },
    {
      "epoch": 0.3513036607848301,
      "grad_norm": 3.4179165363311768,
      "learning_rate": 3.244666842243877e-05,
      "loss": 0.9763,
      "step": 13339
    },
    {
      "epoch": 0.35132999736634185,
      "grad_norm": 2.2655434608459473,
      "learning_rate": 3.244535159336318e-05,
      "loss": 1.1891,
      "step": 13340
    },
    {
      "epoch": 0.3513563339478536,
      "grad_norm": 1.9768743515014648,
      "learning_rate": 3.2444034764287596e-05,
      "loss": 2.0416,
      "step": 13341
    },
    {
      "epoch": 0.3513826705293653,
      "grad_norm": 2.30531644821167,
      "learning_rate": 3.244271793521201e-05,
      "loss": 2.5406,
      "step": 13342
    },
    {
      "epoch": 0.351409007110877,
      "grad_norm": 2.2749857902526855,
      "learning_rate": 3.244140110613643e-05,
      "loss": 0.9018,
      "step": 13343
    },
    {
      "epoch": 0.3514353436923887,
      "grad_norm": 1.62916100025177,
      "learning_rate": 3.244008427706084e-05,
      "loss": 1.388,
      "step": 13344
    },
    {
      "epoch": 0.35146168027390046,
      "grad_norm": 2.0928359031677246,
      "learning_rate": 3.243876744798525e-05,
      "loss": 2.0932,
      "step": 13345
    },
    {
      "epoch": 0.35148801685541214,
      "grad_norm": 3.6840946674346924,
      "learning_rate": 3.243745061890967e-05,
      "loss": 1.6484,
      "step": 13346
    },
    {
      "epoch": 0.3515143534369239,
      "grad_norm": 3.200409173965454,
      "learning_rate": 3.243613378983408e-05,
      "loss": 1.8889,
      "step": 13347
    },
    {
      "epoch": 0.35154069001843563,
      "grad_norm": 2.4166016578674316,
      "learning_rate": 3.24348169607585e-05,
      "loss": 1.3953,
      "step": 13348
    },
    {
      "epoch": 0.3515670265999473,
      "grad_norm": 2.2247650623321533,
      "learning_rate": 3.243350013168291e-05,
      "loss": 1.7617,
      "step": 13349
    },
    {
      "epoch": 0.35159336318145906,
      "grad_norm": 1.8270072937011719,
      "learning_rate": 3.243218330260732e-05,
      "loss": 1.7973,
      "step": 13350
    },
    {
      "epoch": 0.35161969976297075,
      "grad_norm": 2.3578460216522217,
      "learning_rate": 3.243086647353173e-05,
      "loss": 1.7287,
      "step": 13351
    },
    {
      "epoch": 0.3516460363444825,
      "grad_norm": 2.202893018722534,
      "learning_rate": 3.2429549644456154e-05,
      "loss": 1.8195,
      "step": 13352
    },
    {
      "epoch": 0.3516723729259942,
      "grad_norm": 2.8357491493225098,
      "learning_rate": 3.242823281538056e-05,
      "loss": 1.8782,
      "step": 13353
    },
    {
      "epoch": 0.3516987095075059,
      "grad_norm": 1.8396583795547485,
      "learning_rate": 3.242691598630498e-05,
      "loss": 2.3974,
      "step": 13354
    },
    {
      "epoch": 0.35172504608901767,
      "grad_norm": 2.1672394275665283,
      "learning_rate": 3.2425599157229394e-05,
      "loss": 1.7462,
      "step": 13355
    },
    {
      "epoch": 0.35175138267052936,
      "grad_norm": 2.935796022415161,
      "learning_rate": 3.242428232815381e-05,
      "loss": 1.8066,
      "step": 13356
    },
    {
      "epoch": 0.3517777192520411,
      "grad_norm": 2.407139301300049,
      "learning_rate": 3.2422965499078225e-05,
      "loss": 1.7109,
      "step": 13357
    },
    {
      "epoch": 0.3518040558335528,
      "grad_norm": 1.8942136764526367,
      "learning_rate": 3.2421648670002634e-05,
      "loss": 1.4957,
      "step": 13358
    },
    {
      "epoch": 0.35183039241506453,
      "grad_norm": 1.977776288986206,
      "learning_rate": 3.242033184092705e-05,
      "loss": 1.5155,
      "step": 13359
    },
    {
      "epoch": 0.3518567289965762,
      "grad_norm": 4.4377570152282715,
      "learning_rate": 3.241901501185146e-05,
      "loss": 1.1665,
      "step": 13360
    },
    {
      "epoch": 0.35188306557808796,
      "grad_norm": 3.2606704235076904,
      "learning_rate": 3.241769818277588e-05,
      "loss": 1.2178,
      "step": 13361
    },
    {
      "epoch": 0.3519094021595997,
      "grad_norm": 2.151022434234619,
      "learning_rate": 3.241638135370029e-05,
      "loss": 1.5879,
      "step": 13362
    },
    {
      "epoch": 0.3519357387411114,
      "grad_norm": 2.694943428039551,
      "learning_rate": 3.2415064524624705e-05,
      "loss": 0.9485,
      "step": 13363
    },
    {
      "epoch": 0.35196207532262314,
      "grad_norm": 1.9729725122451782,
      "learning_rate": 3.241374769554912e-05,
      "loss": 2.0231,
      "step": 13364
    },
    {
      "epoch": 0.3519884119041348,
      "grad_norm": 4.62639856338501,
      "learning_rate": 3.241243086647353e-05,
      "loss": 1.5547,
      "step": 13365
    },
    {
      "epoch": 0.35201474848564657,
      "grad_norm": 1.7988792657852173,
      "learning_rate": 3.241111403739795e-05,
      "loss": 0.3154,
      "step": 13366
    },
    {
      "epoch": 0.3520410850671583,
      "grad_norm": 3.6884405612945557,
      "learning_rate": 3.240979720832236e-05,
      "loss": 1.5824,
      "step": 13367
    },
    {
      "epoch": 0.35206742164867,
      "grad_norm": 1.5911521911621094,
      "learning_rate": 3.2408480379246776e-05,
      "loss": 0.899,
      "step": 13368
    },
    {
      "epoch": 0.35209375823018174,
      "grad_norm": 1.6808979511260986,
      "learning_rate": 3.2407163550171185e-05,
      "loss": 1.6173,
      "step": 13369
    },
    {
      "epoch": 0.35212009481169343,
      "grad_norm": 2.386752128601074,
      "learning_rate": 3.24058467210956e-05,
      "loss": 0.4426,
      "step": 13370
    },
    {
      "epoch": 0.3521464313932052,
      "grad_norm": 3.679684638977051,
      "learning_rate": 3.2404529892020016e-05,
      "loss": 1.5691,
      "step": 13371
    },
    {
      "epoch": 0.35217276797471686,
      "grad_norm": 2.225318670272827,
      "learning_rate": 3.240321306294443e-05,
      "loss": 1.3965,
      "step": 13372
    },
    {
      "epoch": 0.3521991045562286,
      "grad_norm": 1.9934961795806885,
      "learning_rate": 3.240189623386885e-05,
      "loss": 1.4888,
      "step": 13373
    },
    {
      "epoch": 0.35222544113774035,
      "grad_norm": 1.7122739553451538,
      "learning_rate": 3.2400579404793256e-05,
      "loss": 1.6821,
      "step": 13374
    },
    {
      "epoch": 0.35225177771925203,
      "grad_norm": 2.938758134841919,
      "learning_rate": 3.239926257571768e-05,
      "loss": 1.1511,
      "step": 13375
    },
    {
      "epoch": 0.3522781143007638,
      "grad_norm": 2.356916666030884,
      "learning_rate": 3.239794574664209e-05,
      "loss": 2.2471,
      "step": 13376
    },
    {
      "epoch": 0.35230445088227547,
      "grad_norm": 1.8385820388793945,
      "learning_rate": 3.23966289175665e-05,
      "loss": 1.7183,
      "step": 13377
    },
    {
      "epoch": 0.3523307874637872,
      "grad_norm": 1.902010202407837,
      "learning_rate": 3.239531208849091e-05,
      "loss": 1.8149,
      "step": 13378
    },
    {
      "epoch": 0.3523571240452989,
      "grad_norm": 2.9832451343536377,
      "learning_rate": 3.239399525941533e-05,
      "loss": 1.2653,
      "step": 13379
    },
    {
      "epoch": 0.35238346062681064,
      "grad_norm": 1.6726583242416382,
      "learning_rate": 3.239267843033974e-05,
      "loss": 1.8035,
      "step": 13380
    },
    {
      "epoch": 0.3524097972083224,
      "grad_norm": 2.710875988006592,
      "learning_rate": 3.239136160126416e-05,
      "loss": 1.2807,
      "step": 13381
    },
    {
      "epoch": 0.35243613378983407,
      "grad_norm": 2.777984380722046,
      "learning_rate": 3.2390044772188574e-05,
      "loss": 1.5081,
      "step": 13382
    },
    {
      "epoch": 0.3524624703713458,
      "grad_norm": 2.4720728397369385,
      "learning_rate": 3.238872794311298e-05,
      "loss": 1.9083,
      "step": 13383
    },
    {
      "epoch": 0.3524888069528575,
      "grad_norm": 1.59549081325531,
      "learning_rate": 3.23874111140374e-05,
      "loss": 0.672,
      "step": 13384
    },
    {
      "epoch": 0.35251514353436925,
      "grad_norm": 2.0250182151794434,
      "learning_rate": 3.2386094284961814e-05,
      "loss": 1.736,
      "step": 13385
    },
    {
      "epoch": 0.35254148011588093,
      "grad_norm": 1.8447422981262207,
      "learning_rate": 3.238477745588623e-05,
      "loss": 1.2377,
      "step": 13386
    },
    {
      "epoch": 0.3525678166973927,
      "grad_norm": 2.266187906265259,
      "learning_rate": 3.238346062681064e-05,
      "loss": 2.0352,
      "step": 13387
    },
    {
      "epoch": 0.3525941532789044,
      "grad_norm": 3.264073133468628,
      "learning_rate": 3.2382143797735054e-05,
      "loss": 1.4058,
      "step": 13388
    },
    {
      "epoch": 0.3526204898604161,
      "grad_norm": 3.0453848838806152,
      "learning_rate": 3.238082696865947e-05,
      "loss": 1.6684,
      "step": 13389
    },
    {
      "epoch": 0.35264682644192785,
      "grad_norm": 1.4842275381088257,
      "learning_rate": 3.2379510139583885e-05,
      "loss": 1.7708,
      "step": 13390
    },
    {
      "epoch": 0.35267316302343954,
      "grad_norm": 2.915254592895508,
      "learning_rate": 3.23781933105083e-05,
      "loss": 0.908,
      "step": 13391
    },
    {
      "epoch": 0.3526994996049513,
      "grad_norm": 1.4451037645339966,
      "learning_rate": 3.237687648143271e-05,
      "loss": 1.8519,
      "step": 13392
    },
    {
      "epoch": 0.35272583618646297,
      "grad_norm": 1.965159296989441,
      "learning_rate": 3.2375559652357125e-05,
      "loss": 1.0986,
      "step": 13393
    },
    {
      "epoch": 0.3527521727679747,
      "grad_norm": 2.003401041030884,
      "learning_rate": 3.237424282328154e-05,
      "loss": 2.0036,
      "step": 13394
    },
    {
      "epoch": 0.35277850934948646,
      "grad_norm": 2.3906521797180176,
      "learning_rate": 3.237292599420596e-05,
      "loss": 1.1264,
      "step": 13395
    },
    {
      "epoch": 0.35280484593099815,
      "grad_norm": 2.345811367034912,
      "learning_rate": 3.2371609165130365e-05,
      "loss": 1.8041,
      "step": 13396
    },
    {
      "epoch": 0.3528311825125099,
      "grad_norm": 4.376338958740234,
      "learning_rate": 3.237029233605478e-05,
      "loss": 1.7419,
      "step": 13397
    },
    {
      "epoch": 0.3528575190940216,
      "grad_norm": 1.6952074766159058,
      "learning_rate": 3.23689755069792e-05,
      "loss": 1.8855,
      "step": 13398
    },
    {
      "epoch": 0.3528838556755333,
      "grad_norm": 2.1415939331054688,
      "learning_rate": 3.236765867790361e-05,
      "loss": 2.4641,
      "step": 13399
    },
    {
      "epoch": 0.352910192257045,
      "grad_norm": 4.881741523742676,
      "learning_rate": 3.236634184882803e-05,
      "loss": 1.8107,
      "step": 13400
    },
    {
      "epoch": 0.35293652883855675,
      "grad_norm": 1.3757063150405884,
      "learning_rate": 3.236502501975244e-05,
      "loss": 1.6012,
      "step": 13401
    },
    {
      "epoch": 0.3529628654200685,
      "grad_norm": 2.28267502784729,
      "learning_rate": 3.236370819067685e-05,
      "loss": 2.0571,
      "step": 13402
    },
    {
      "epoch": 0.3529892020015802,
      "grad_norm": 2.4455177783966064,
      "learning_rate": 3.236239136160126e-05,
      "loss": 2.2408,
      "step": 13403
    },
    {
      "epoch": 0.3530155385830919,
      "grad_norm": 3.009761333465576,
      "learning_rate": 3.2361074532525683e-05,
      "loss": 1.973,
      "step": 13404
    },
    {
      "epoch": 0.3530418751646036,
      "grad_norm": 1.7002278566360474,
      "learning_rate": 3.235975770345009e-05,
      "loss": 1.9107,
      "step": 13405
    },
    {
      "epoch": 0.35306821174611536,
      "grad_norm": 2.9433672428131104,
      "learning_rate": 3.235844087437451e-05,
      "loss": 1.2209,
      "step": 13406
    },
    {
      "epoch": 0.3530945483276271,
      "grad_norm": 3.264383554458618,
      "learning_rate": 3.235712404529892e-05,
      "loss": 1.9347,
      "step": 13407
    },
    {
      "epoch": 0.3531208849091388,
      "grad_norm": 2.463010311126709,
      "learning_rate": 3.235580721622334e-05,
      "loss": 1.6167,
      "step": 13408
    },
    {
      "epoch": 0.35314722149065053,
      "grad_norm": 4.239632606506348,
      "learning_rate": 3.235449038714775e-05,
      "loss": 1.0254,
      "step": 13409
    },
    {
      "epoch": 0.3531735580721622,
      "grad_norm": 2.923922300338745,
      "learning_rate": 3.2353173558072164e-05,
      "loss": 0.6323,
      "step": 13410
    },
    {
      "epoch": 0.35319989465367396,
      "grad_norm": 2.098973035812378,
      "learning_rate": 3.235185672899658e-05,
      "loss": 1.5639,
      "step": 13411
    },
    {
      "epoch": 0.35322623123518565,
      "grad_norm": 3.2833895683288574,
      "learning_rate": 3.235053989992099e-05,
      "loss": 1.7127,
      "step": 13412
    },
    {
      "epoch": 0.3532525678166974,
      "grad_norm": 5.73862886428833,
      "learning_rate": 3.234922307084541e-05,
      "loss": 1.5556,
      "step": 13413
    },
    {
      "epoch": 0.35327890439820914,
      "grad_norm": 2.7117037773132324,
      "learning_rate": 3.234790624176982e-05,
      "loss": 1.4845,
      "step": 13414
    },
    {
      "epoch": 0.3533052409797208,
      "grad_norm": 1.8839472532272339,
      "learning_rate": 3.2346589412694235e-05,
      "loss": 1.406,
      "step": 13415
    },
    {
      "epoch": 0.35333157756123257,
      "grad_norm": 3.145193338394165,
      "learning_rate": 3.2345272583618644e-05,
      "loss": 1.8111,
      "step": 13416
    },
    {
      "epoch": 0.35335791414274426,
      "grad_norm": 2.4471664428710938,
      "learning_rate": 3.234395575454306e-05,
      "loss": 2.0764,
      "step": 13417
    },
    {
      "epoch": 0.353384250724256,
      "grad_norm": 1.9531912803649902,
      "learning_rate": 3.2342638925467475e-05,
      "loss": 1.7331,
      "step": 13418
    },
    {
      "epoch": 0.3534105873057677,
      "grad_norm": 2.8200695514678955,
      "learning_rate": 3.234132209639189e-05,
      "loss": 1.9612,
      "step": 13419
    },
    {
      "epoch": 0.35343692388727943,
      "grad_norm": 2.060945749282837,
      "learning_rate": 3.2340005267316306e-05,
      "loss": 1.7328,
      "step": 13420
    },
    {
      "epoch": 0.3534632604687912,
      "grad_norm": 5.0167341232299805,
      "learning_rate": 3.2338688438240715e-05,
      "loss": 1.8001,
      "step": 13421
    },
    {
      "epoch": 0.35348959705030286,
      "grad_norm": 2.379786729812622,
      "learning_rate": 3.233737160916514e-05,
      "loss": 1.9806,
      "step": 13422
    },
    {
      "epoch": 0.3535159336318146,
      "grad_norm": 1.8490166664123535,
      "learning_rate": 3.2336054780089546e-05,
      "loss": 1.4596,
      "step": 13423
    },
    {
      "epoch": 0.3535422702133263,
      "grad_norm": 1.505871057510376,
      "learning_rate": 3.233473795101396e-05,
      "loss": 1.5093,
      "step": 13424
    },
    {
      "epoch": 0.35356860679483804,
      "grad_norm": 1.8197870254516602,
      "learning_rate": 3.233342112193837e-05,
      "loss": 1.9049,
      "step": 13425
    },
    {
      "epoch": 0.3535949433763497,
      "grad_norm": 2.3923511505126953,
      "learning_rate": 3.2332104292862786e-05,
      "loss": 0.9344,
      "step": 13426
    },
    {
      "epoch": 0.35362127995786147,
      "grad_norm": 1.7724586725234985,
      "learning_rate": 3.23307874637872e-05,
      "loss": 2.396,
      "step": 13427
    },
    {
      "epoch": 0.3536476165393732,
      "grad_norm": 1.9437782764434814,
      "learning_rate": 3.232947063471162e-05,
      "loss": 2.1122,
      "step": 13428
    },
    {
      "epoch": 0.3536739531208849,
      "grad_norm": 3.4790189266204834,
      "learning_rate": 3.232815380563603e-05,
      "loss": 0.9541,
      "step": 13429
    },
    {
      "epoch": 0.35370028970239664,
      "grad_norm": 1.8331210613250732,
      "learning_rate": 3.232683697656044e-05,
      "loss": 1.9549,
      "step": 13430
    },
    {
      "epoch": 0.35372662628390833,
      "grad_norm": 1.7642755508422852,
      "learning_rate": 3.232552014748486e-05,
      "loss": 1.6784,
      "step": 13431
    },
    {
      "epoch": 0.3537529628654201,
      "grad_norm": 1.832125186920166,
      "learning_rate": 3.232420331840927e-05,
      "loss": 0.401,
      "step": 13432
    },
    {
      "epoch": 0.35377929944693176,
      "grad_norm": 3.800145149230957,
      "learning_rate": 3.232288648933369e-05,
      "loss": 1.004,
      "step": 13433
    },
    {
      "epoch": 0.3538056360284435,
      "grad_norm": 3.660468578338623,
      "learning_rate": 3.23215696602581e-05,
      "loss": 0.9384,
      "step": 13434
    },
    {
      "epoch": 0.35383197260995525,
      "grad_norm": 2.0878841876983643,
      "learning_rate": 3.232025283118251e-05,
      "loss": 1.5391,
      "step": 13435
    },
    {
      "epoch": 0.35385830919146694,
      "grad_norm": 1.6763434410095215,
      "learning_rate": 3.231893600210693e-05,
      "loss": 1.56,
      "step": 13436
    },
    {
      "epoch": 0.3538846457729787,
      "grad_norm": 3.052415609359741,
      "learning_rate": 3.2317619173031344e-05,
      "loss": 1.2428,
      "step": 13437
    },
    {
      "epoch": 0.35391098235449037,
      "grad_norm": 5.878122806549072,
      "learning_rate": 3.231630234395576e-05,
      "loss": 1.4001,
      "step": 13438
    },
    {
      "epoch": 0.3539373189360021,
      "grad_norm": 2.6253037452697754,
      "learning_rate": 3.231498551488017e-05,
      "loss": 1.9999,
      "step": 13439
    },
    {
      "epoch": 0.35396365551751385,
      "grad_norm": 4.330442905426025,
      "learning_rate": 3.2313668685804584e-05,
      "loss": 1.801,
      "step": 13440
    },
    {
      "epoch": 0.35398999209902554,
      "grad_norm": 1.8195282220840454,
      "learning_rate": 3.2312351856729e-05,
      "loss": 1.954,
      "step": 13441
    },
    {
      "epoch": 0.3540163286805373,
      "grad_norm": 2.3423333168029785,
      "learning_rate": 3.2311035027653415e-05,
      "loss": 1.455,
      "step": 13442
    },
    {
      "epoch": 0.354042665262049,
      "grad_norm": 1.7573970556259155,
      "learning_rate": 3.2309718198577824e-05,
      "loss": 2.0738,
      "step": 13443
    },
    {
      "epoch": 0.3540690018435607,
      "grad_norm": 2.465963840484619,
      "learning_rate": 3.230840136950224e-05,
      "loss": 0.6359,
      "step": 13444
    },
    {
      "epoch": 0.3540953384250724,
      "grad_norm": 2.3648228645324707,
      "learning_rate": 3.2307084540426655e-05,
      "loss": 1.4879,
      "step": 13445
    },
    {
      "epoch": 0.35412167500658415,
      "grad_norm": 1.4378904104232788,
      "learning_rate": 3.230576771135107e-05,
      "loss": 1.4814,
      "step": 13446
    },
    {
      "epoch": 0.3541480115880959,
      "grad_norm": 2.2449519634246826,
      "learning_rate": 3.2304450882275486e-05,
      "loss": 1.6078,
      "step": 13447
    },
    {
      "epoch": 0.3541743481696076,
      "grad_norm": 1.7108471393585205,
      "learning_rate": 3.2303134053199895e-05,
      "loss": 0.6002,
      "step": 13448
    },
    {
      "epoch": 0.3542006847511193,
      "grad_norm": 1.500854253768921,
      "learning_rate": 3.230181722412431e-05,
      "loss": 2.1894,
      "step": 13449
    },
    {
      "epoch": 0.354227021332631,
      "grad_norm": 1.9202920198440552,
      "learning_rate": 3.230050039504872e-05,
      "loss": 1.1398,
      "step": 13450
    },
    {
      "epoch": 0.35425335791414275,
      "grad_norm": 2.657386064529419,
      "learning_rate": 3.229918356597314e-05,
      "loss": 1.8775,
      "step": 13451
    },
    {
      "epoch": 0.35427969449565444,
      "grad_norm": 1.9549400806427002,
      "learning_rate": 3.229786673689755e-05,
      "loss": 1.6835,
      "step": 13452
    },
    {
      "epoch": 0.3543060310771662,
      "grad_norm": 2.1215004920959473,
      "learning_rate": 3.2296549907821966e-05,
      "loss": 1.6879,
      "step": 13453
    },
    {
      "epoch": 0.3543323676586779,
      "grad_norm": 2.3544883728027344,
      "learning_rate": 3.2295233078746375e-05,
      "loss": 1.9081,
      "step": 13454
    },
    {
      "epoch": 0.3543587042401896,
      "grad_norm": 2.0533502101898193,
      "learning_rate": 3.22939162496708e-05,
      "loss": 2.0292,
      "step": 13455
    },
    {
      "epoch": 0.35438504082170136,
      "grad_norm": 3.74015212059021,
      "learning_rate": 3.2292599420595206e-05,
      "loss": 2.3287,
      "step": 13456
    },
    {
      "epoch": 0.35441137740321305,
      "grad_norm": 4.024369716644287,
      "learning_rate": 3.229128259151962e-05,
      "loss": 1.5144,
      "step": 13457
    },
    {
      "epoch": 0.3544377139847248,
      "grad_norm": 1.64836585521698,
      "learning_rate": 3.228996576244404e-05,
      "loss": 2.3711,
      "step": 13458
    },
    {
      "epoch": 0.3544640505662365,
      "grad_norm": 2.312168836593628,
      "learning_rate": 3.2288648933368446e-05,
      "loss": 1.7675,
      "step": 13459
    },
    {
      "epoch": 0.3544903871477482,
      "grad_norm": 2.635197401046753,
      "learning_rate": 3.228733210429287e-05,
      "loss": 1.2045,
      "step": 13460
    },
    {
      "epoch": 0.35451672372925996,
      "grad_norm": 5.81008768081665,
      "learning_rate": 3.228601527521728e-05,
      "loss": 1.6074,
      "step": 13461
    },
    {
      "epoch": 0.35454306031077165,
      "grad_norm": 2.237865447998047,
      "learning_rate": 3.228469844614169e-05,
      "loss": 1.7721,
      "step": 13462
    },
    {
      "epoch": 0.3545693968922834,
      "grad_norm": 2.20149564743042,
      "learning_rate": 3.22833816170661e-05,
      "loss": 1.8338,
      "step": 13463
    },
    {
      "epoch": 0.3545957334737951,
      "grad_norm": 3.2823715209960938,
      "learning_rate": 3.228206478799052e-05,
      "loss": 1.0056,
      "step": 13464
    },
    {
      "epoch": 0.3546220700553068,
      "grad_norm": 4.911948204040527,
      "learning_rate": 3.228074795891493e-05,
      "loss": 1.4288,
      "step": 13465
    },
    {
      "epoch": 0.3546484066368185,
      "grad_norm": 2.457892656326294,
      "learning_rate": 3.227943112983935e-05,
      "loss": 1.6921,
      "step": 13466
    },
    {
      "epoch": 0.35467474321833026,
      "grad_norm": 1.9747319221496582,
      "learning_rate": 3.2278114300763764e-05,
      "loss": 1.6734,
      "step": 13467
    },
    {
      "epoch": 0.354701079799842,
      "grad_norm": 1.8965075016021729,
      "learning_rate": 3.227679747168817e-05,
      "loss": 1.1413,
      "step": 13468
    },
    {
      "epoch": 0.3547274163813537,
      "grad_norm": 2.6162514686584473,
      "learning_rate": 3.227548064261259e-05,
      "loss": 1.8204,
      "step": 13469
    },
    {
      "epoch": 0.35475375296286543,
      "grad_norm": 1.559047818183899,
      "learning_rate": 3.2274163813537004e-05,
      "loss": 1.8301,
      "step": 13470
    },
    {
      "epoch": 0.3547800895443771,
      "grad_norm": 2.322930097579956,
      "learning_rate": 3.227284698446142e-05,
      "loss": 0.8137,
      "step": 13471
    },
    {
      "epoch": 0.35480642612588886,
      "grad_norm": 2.4971282482147217,
      "learning_rate": 3.227153015538583e-05,
      "loss": 2.2915,
      "step": 13472
    },
    {
      "epoch": 0.3548327627074006,
      "grad_norm": 1.7717419862747192,
      "learning_rate": 3.2270213326310245e-05,
      "loss": 1.28,
      "step": 13473
    },
    {
      "epoch": 0.3548590992889123,
      "grad_norm": 4.11823034286499,
      "learning_rate": 3.226889649723466e-05,
      "loss": 2.0423,
      "step": 13474
    },
    {
      "epoch": 0.35488543587042404,
      "grad_norm": 1.5041719675064087,
      "learning_rate": 3.2267579668159076e-05,
      "loss": 2.2459,
      "step": 13475
    },
    {
      "epoch": 0.3549117724519357,
      "grad_norm": 2.930955648422241,
      "learning_rate": 3.226626283908349e-05,
      "loss": 2.0029,
      "step": 13476
    },
    {
      "epoch": 0.35493810903344747,
      "grad_norm": 10.966865539550781,
      "learning_rate": 3.22649460100079e-05,
      "loss": 2.6103,
      "step": 13477
    },
    {
      "epoch": 0.35496444561495916,
      "grad_norm": 1.669223666191101,
      "learning_rate": 3.2263629180932316e-05,
      "loss": 2.2197,
      "step": 13478
    },
    {
      "epoch": 0.3549907821964709,
      "grad_norm": 1.4289343357086182,
      "learning_rate": 3.226231235185673e-05,
      "loss": 1.3207,
      "step": 13479
    },
    {
      "epoch": 0.35501711877798264,
      "grad_norm": 2.8188369274139404,
      "learning_rate": 3.226099552278115e-05,
      "loss": 2.0416,
      "step": 13480
    },
    {
      "epoch": 0.35504345535949433,
      "grad_norm": 2.122403144836426,
      "learning_rate": 3.2259678693705556e-05,
      "loss": 0.5008,
      "step": 13481
    },
    {
      "epoch": 0.3550697919410061,
      "grad_norm": 1.9776302576065063,
      "learning_rate": 3.225836186462997e-05,
      "loss": 0.3941,
      "step": 13482
    },
    {
      "epoch": 0.35509612852251776,
      "grad_norm": 2.0202600955963135,
      "learning_rate": 3.225704503555439e-05,
      "loss": 1.5958,
      "step": 13483
    },
    {
      "epoch": 0.3551224651040295,
      "grad_norm": 2.663630485534668,
      "learning_rate": 3.22557282064788e-05,
      "loss": 1.3616,
      "step": 13484
    },
    {
      "epoch": 0.3551488016855412,
      "grad_norm": 2.960670232772827,
      "learning_rate": 3.225441137740322e-05,
      "loss": 1.6209,
      "step": 13485
    },
    {
      "epoch": 0.35517513826705294,
      "grad_norm": 2.3142905235290527,
      "learning_rate": 3.225309454832763e-05,
      "loss": 2.1586,
      "step": 13486
    },
    {
      "epoch": 0.3552014748485647,
      "grad_norm": 1.773917555809021,
      "learning_rate": 3.225177771925204e-05,
      "loss": 2.2527,
      "step": 13487
    },
    {
      "epoch": 0.35522781143007637,
      "grad_norm": 2.314061164855957,
      "learning_rate": 3.225046089017646e-05,
      "loss": 1.4456,
      "step": 13488
    },
    {
      "epoch": 0.3552541480115881,
      "grad_norm": 1.6005955934524536,
      "learning_rate": 3.2249144061100874e-05,
      "loss": 1.9263,
      "step": 13489
    },
    {
      "epoch": 0.3552804845930998,
      "grad_norm": 2.636345863342285,
      "learning_rate": 3.224782723202528e-05,
      "loss": 1.9443,
      "step": 13490
    },
    {
      "epoch": 0.35530682117461154,
      "grad_norm": 2.1058764457702637,
      "learning_rate": 3.22465104029497e-05,
      "loss": 1.413,
      "step": 13491
    },
    {
      "epoch": 0.35533315775612323,
      "grad_norm": 5.849967002868652,
      "learning_rate": 3.2245193573874114e-05,
      "loss": 1.8406,
      "step": 13492
    },
    {
      "epoch": 0.355359494337635,
      "grad_norm": 2.232043981552124,
      "learning_rate": 3.224387674479853e-05,
      "loss": 1.6915,
      "step": 13493
    },
    {
      "epoch": 0.3553858309191467,
      "grad_norm": 2.2010385990142822,
      "learning_rate": 3.2242559915722945e-05,
      "loss": 1.5191,
      "step": 13494
    },
    {
      "epoch": 0.3554121675006584,
      "grad_norm": 2.755608558654785,
      "learning_rate": 3.2241243086647354e-05,
      "loss": 1.1885,
      "step": 13495
    },
    {
      "epoch": 0.35543850408217015,
      "grad_norm": 2.041734457015991,
      "learning_rate": 3.223992625757177e-05,
      "loss": 1.5228,
      "step": 13496
    },
    {
      "epoch": 0.35546484066368184,
      "grad_norm": 1.7045328617095947,
      "learning_rate": 3.223860942849618e-05,
      "loss": 1.7614,
      "step": 13497
    },
    {
      "epoch": 0.3554911772451936,
      "grad_norm": 3.127450704574585,
      "learning_rate": 3.22372925994206e-05,
      "loss": 2.3074,
      "step": 13498
    },
    {
      "epoch": 0.35551751382670527,
      "grad_norm": 2.6444132328033447,
      "learning_rate": 3.223597577034501e-05,
      "loss": 1.9232,
      "step": 13499
    },
    {
      "epoch": 0.355543850408217,
      "grad_norm": 6.129952430725098,
      "learning_rate": 3.2234658941269425e-05,
      "loss": 1.7753,
      "step": 13500
    },
    {
      "epoch": 0.35557018698972875,
      "grad_norm": 4.689990997314453,
      "learning_rate": 3.223334211219384e-05,
      "loss": 1.7396,
      "step": 13501
    },
    {
      "epoch": 0.35559652357124044,
      "grad_norm": 2.0375030040740967,
      "learning_rate": 3.223202528311825e-05,
      "loss": 1.8252,
      "step": 13502
    },
    {
      "epoch": 0.3556228601527522,
      "grad_norm": 2.710348129272461,
      "learning_rate": 3.223070845404267e-05,
      "loss": 1.4784,
      "step": 13503
    },
    {
      "epoch": 0.3556491967342639,
      "grad_norm": 2.265514373779297,
      "learning_rate": 3.222939162496708e-05,
      "loss": 1.9078,
      "step": 13504
    },
    {
      "epoch": 0.3556755333157756,
      "grad_norm": 3.0866858959198,
      "learning_rate": 3.2228074795891496e-05,
      "loss": 0.9979,
      "step": 13505
    },
    {
      "epoch": 0.35570186989728736,
      "grad_norm": 2.3354218006134033,
      "learning_rate": 3.2226757966815905e-05,
      "loss": 2.048,
      "step": 13506
    },
    {
      "epoch": 0.35572820647879905,
      "grad_norm": 2.2681877613067627,
      "learning_rate": 3.222544113774033e-05,
      "loss": 1.1664,
      "step": 13507
    },
    {
      "epoch": 0.3557545430603108,
      "grad_norm": 1.4610751867294312,
      "learning_rate": 3.2224124308664736e-05,
      "loss": 1.7744,
      "step": 13508
    },
    {
      "epoch": 0.3557808796418225,
      "grad_norm": 2.241502285003662,
      "learning_rate": 3.222280747958915e-05,
      "loss": 1.7951,
      "step": 13509
    },
    {
      "epoch": 0.3558072162233342,
      "grad_norm": 2.4376156330108643,
      "learning_rate": 3.222149065051356e-05,
      "loss": 1.3366,
      "step": 13510
    },
    {
      "epoch": 0.3558335528048459,
      "grad_norm": 1.910470962524414,
      "learning_rate": 3.2220173821437976e-05,
      "loss": 1.5958,
      "step": 13511
    },
    {
      "epoch": 0.35585988938635765,
      "grad_norm": 1.63845956325531,
      "learning_rate": 3.221885699236239e-05,
      "loss": 1.2993,
      "step": 13512
    },
    {
      "epoch": 0.3558862259678694,
      "grad_norm": 1.8612556457519531,
      "learning_rate": 3.221754016328681e-05,
      "loss": 0.7126,
      "step": 13513
    },
    {
      "epoch": 0.3559125625493811,
      "grad_norm": 2.365784168243408,
      "learning_rate": 3.221622333421122e-05,
      "loss": 1.0563,
      "step": 13514
    },
    {
      "epoch": 0.3559388991308928,
      "grad_norm": 2.4441957473754883,
      "learning_rate": 3.221490650513563e-05,
      "loss": 1.5295,
      "step": 13515
    },
    {
      "epoch": 0.3559652357124045,
      "grad_norm": 1.5101608037948608,
      "learning_rate": 3.221358967606005e-05,
      "loss": 1.4965,
      "step": 13516
    },
    {
      "epoch": 0.35599157229391626,
      "grad_norm": 2.502692699432373,
      "learning_rate": 3.221227284698446e-05,
      "loss": 1.4537,
      "step": 13517
    },
    {
      "epoch": 0.35601790887542795,
      "grad_norm": 3.574384927749634,
      "learning_rate": 3.221095601790888e-05,
      "loss": 1.475,
      "step": 13518
    },
    {
      "epoch": 0.3560442454569397,
      "grad_norm": 1.7567169666290283,
      "learning_rate": 3.220963918883329e-05,
      "loss": 1.9212,
      "step": 13519
    },
    {
      "epoch": 0.35607058203845143,
      "grad_norm": 4.268300533294678,
      "learning_rate": 3.22083223597577e-05,
      "loss": 1.5846,
      "step": 13520
    },
    {
      "epoch": 0.3560969186199631,
      "grad_norm": 1.6878823041915894,
      "learning_rate": 3.220700553068212e-05,
      "loss": 1.8979,
      "step": 13521
    },
    {
      "epoch": 0.35612325520147486,
      "grad_norm": 1.9981175661087036,
      "learning_rate": 3.2205688701606534e-05,
      "loss": 1.15,
      "step": 13522
    },
    {
      "epoch": 0.35614959178298655,
      "grad_norm": 2.7628042697906494,
      "learning_rate": 3.220437187253095e-05,
      "loss": 0.8069,
      "step": 13523
    },
    {
      "epoch": 0.3561759283644983,
      "grad_norm": 3.2204749584198,
      "learning_rate": 3.220305504345536e-05,
      "loss": 1.7214,
      "step": 13524
    },
    {
      "epoch": 0.35620226494601,
      "grad_norm": 1.607557773590088,
      "learning_rate": 3.2201738214379774e-05,
      "loss": 1.8681,
      "step": 13525
    },
    {
      "epoch": 0.3562286015275217,
      "grad_norm": 3.5377190113067627,
      "learning_rate": 3.220042138530419e-05,
      "loss": 2.5758,
      "step": 13526
    },
    {
      "epoch": 0.35625493810903347,
      "grad_norm": 2.097262382507324,
      "learning_rate": 3.2199104556228605e-05,
      "loss": 1.076,
      "step": 13527
    },
    {
      "epoch": 0.35628127469054516,
      "grad_norm": 3.9482004642486572,
      "learning_rate": 3.2197787727153014e-05,
      "loss": 1.8825,
      "step": 13528
    },
    {
      "epoch": 0.3563076112720569,
      "grad_norm": 2.0969722270965576,
      "learning_rate": 3.219647089807743e-05,
      "loss": 1.5867,
      "step": 13529
    },
    {
      "epoch": 0.3563339478535686,
      "grad_norm": 1.8546539545059204,
      "learning_rate": 3.2195154069001845e-05,
      "loss": 0.797,
      "step": 13530
    },
    {
      "epoch": 0.35636028443508033,
      "grad_norm": 1.6689138412475586,
      "learning_rate": 3.219383723992626e-05,
      "loss": 1.7193,
      "step": 13531
    },
    {
      "epoch": 0.356386621016592,
      "grad_norm": 1.855383276939392,
      "learning_rate": 3.219252041085068e-05,
      "loss": 2.3078,
      "step": 13532
    },
    {
      "epoch": 0.35641295759810376,
      "grad_norm": 1.8257544040679932,
      "learning_rate": 3.2191203581775086e-05,
      "loss": 1.8635,
      "step": 13533
    },
    {
      "epoch": 0.3564392941796155,
      "grad_norm": 2.4283852577209473,
      "learning_rate": 3.21898867526995e-05,
      "loss": 2.4823,
      "step": 13534
    },
    {
      "epoch": 0.3564656307611272,
      "grad_norm": 1.6496526002883911,
      "learning_rate": 3.218856992362391e-05,
      "loss": 1.5239,
      "step": 13535
    },
    {
      "epoch": 0.35649196734263894,
      "grad_norm": 1.8665874004364014,
      "learning_rate": 3.218725309454833e-05,
      "loss": 2.0893,
      "step": 13536
    },
    {
      "epoch": 0.3565183039241506,
      "grad_norm": 2.8739356994628906,
      "learning_rate": 3.218593626547274e-05,
      "loss": 1.8981,
      "step": 13537
    },
    {
      "epoch": 0.35654464050566237,
      "grad_norm": 2.7292630672454834,
      "learning_rate": 3.218461943639716e-05,
      "loss": 1.0954,
      "step": 13538
    },
    {
      "epoch": 0.3565709770871741,
      "grad_norm": 3.0815908908843994,
      "learning_rate": 3.218330260732157e-05,
      "loss": 1.4115,
      "step": 13539
    },
    {
      "epoch": 0.3565973136686858,
      "grad_norm": 1.9336557388305664,
      "learning_rate": 3.218198577824599e-05,
      "loss": 2.1785,
      "step": 13540
    },
    {
      "epoch": 0.35662365025019754,
      "grad_norm": 1.9636908769607544,
      "learning_rate": 3.2180668949170404e-05,
      "loss": 1.5667,
      "step": 13541
    },
    {
      "epoch": 0.35664998683170923,
      "grad_norm": 1.776019811630249,
      "learning_rate": 3.217935212009481e-05,
      "loss": 1.7945,
      "step": 13542
    },
    {
      "epoch": 0.356676323413221,
      "grad_norm": 1.9169420003890991,
      "learning_rate": 3.217803529101923e-05,
      "loss": 2.4759,
      "step": 13543
    },
    {
      "epoch": 0.35670265999473266,
      "grad_norm": 1.8773916959762573,
      "learning_rate": 3.217671846194364e-05,
      "loss": 1.7936,
      "step": 13544
    },
    {
      "epoch": 0.3567289965762444,
      "grad_norm": 1.571679949760437,
      "learning_rate": 3.217540163286806e-05,
      "loss": 1.4563,
      "step": 13545
    },
    {
      "epoch": 0.35675533315775615,
      "grad_norm": 1.9549353122711182,
      "learning_rate": 3.217408480379247e-05,
      "loss": 2.6193,
      "step": 13546
    },
    {
      "epoch": 0.35678166973926784,
      "grad_norm": 1.5232690572738647,
      "learning_rate": 3.2172767974716884e-05,
      "loss": 1.4649,
      "step": 13547
    },
    {
      "epoch": 0.3568080063207796,
      "grad_norm": 2.7277729511260986,
      "learning_rate": 3.21714511456413e-05,
      "loss": 2.143,
      "step": 13548
    },
    {
      "epoch": 0.35683434290229127,
      "grad_norm": 1.6067876815795898,
      "learning_rate": 3.217013431656571e-05,
      "loss": 1.8885,
      "step": 13549
    },
    {
      "epoch": 0.356860679483803,
      "grad_norm": 1.8268053531646729,
      "learning_rate": 3.216881748749013e-05,
      "loss": 0.7139,
      "step": 13550
    },
    {
      "epoch": 0.3568870160653147,
      "grad_norm": 2.4645025730133057,
      "learning_rate": 3.216750065841454e-05,
      "loss": 1.6641,
      "step": 13551
    },
    {
      "epoch": 0.35691335264682644,
      "grad_norm": 2.83951473236084,
      "learning_rate": 3.2166183829338955e-05,
      "loss": 1.5956,
      "step": 13552
    },
    {
      "epoch": 0.3569396892283382,
      "grad_norm": 2.5665435791015625,
      "learning_rate": 3.2164867000263364e-05,
      "loss": 0.8501,
      "step": 13553
    },
    {
      "epoch": 0.3569660258098499,
      "grad_norm": 2.968637466430664,
      "learning_rate": 3.2163550171187786e-05,
      "loss": 0.7681,
      "step": 13554
    },
    {
      "epoch": 0.3569923623913616,
      "grad_norm": 2.0003342628479004,
      "learning_rate": 3.2162233342112195e-05,
      "loss": 1.7505,
      "step": 13555
    },
    {
      "epoch": 0.3570186989728733,
      "grad_norm": 3.052992343902588,
      "learning_rate": 3.216091651303661e-05,
      "loss": 0.6985,
      "step": 13556
    },
    {
      "epoch": 0.35704503555438505,
      "grad_norm": 7.121495723724365,
      "learning_rate": 3.215959968396102e-05,
      "loss": 1.5725,
      "step": 13557
    },
    {
      "epoch": 0.35707137213589674,
      "grad_norm": 4.353183269500732,
      "learning_rate": 3.2158282854885435e-05,
      "loss": 0.9106,
      "step": 13558
    },
    {
      "epoch": 0.3570977087174085,
      "grad_norm": 3.0545105934143066,
      "learning_rate": 3.215696602580985e-05,
      "loss": 1.7404,
      "step": 13559
    },
    {
      "epoch": 0.3571240452989202,
      "grad_norm": 3.831552505493164,
      "learning_rate": 3.2155649196734266e-05,
      "loss": 1.7135,
      "step": 13560
    },
    {
      "epoch": 0.3571503818804319,
      "grad_norm": 3.9955251216888428,
      "learning_rate": 3.215433236765868e-05,
      "loss": 1.7484,
      "step": 13561
    },
    {
      "epoch": 0.35717671846194365,
      "grad_norm": 2.0099432468414307,
      "learning_rate": 3.215301553858309e-05,
      "loss": 1.9631,
      "step": 13562
    },
    {
      "epoch": 0.35720305504345534,
      "grad_norm": 2.879105806350708,
      "learning_rate": 3.2151698709507506e-05,
      "loss": 1.8359,
      "step": 13563
    },
    {
      "epoch": 0.3572293916249671,
      "grad_norm": 2.9097330570220947,
      "learning_rate": 3.215038188043192e-05,
      "loss": 0.4684,
      "step": 13564
    },
    {
      "epoch": 0.3572557282064788,
      "grad_norm": 2.0938796997070312,
      "learning_rate": 3.214906505135634e-05,
      "loss": 1.5081,
      "step": 13565
    },
    {
      "epoch": 0.3572820647879905,
      "grad_norm": 2.119171619415283,
      "learning_rate": 3.2147748222280746e-05,
      "loss": 1.6377,
      "step": 13566
    },
    {
      "epoch": 0.35730840136950226,
      "grad_norm": 1.9012625217437744,
      "learning_rate": 3.214643139320516e-05,
      "loss": 1.5652,
      "step": 13567
    },
    {
      "epoch": 0.35733473795101395,
      "grad_norm": 4.844491004943848,
      "learning_rate": 3.214511456412958e-05,
      "loss": 1.0606,
      "step": 13568
    },
    {
      "epoch": 0.3573610745325257,
      "grad_norm": 2.415804862976074,
      "learning_rate": 3.214379773505399e-05,
      "loss": 1.7434,
      "step": 13569
    },
    {
      "epoch": 0.3573874111140374,
      "grad_norm": 2.3577330112457275,
      "learning_rate": 3.214248090597841e-05,
      "loss": 0.9716,
      "step": 13570
    },
    {
      "epoch": 0.3574137476955491,
      "grad_norm": 2.1850931644439697,
      "learning_rate": 3.214116407690282e-05,
      "loss": 2.2315,
      "step": 13571
    },
    {
      "epoch": 0.3574400842770608,
      "grad_norm": 1.663676142692566,
      "learning_rate": 3.213984724782723e-05,
      "loss": 2.1599,
      "step": 13572
    },
    {
      "epoch": 0.35746642085857255,
      "grad_norm": 2.945284366607666,
      "learning_rate": 3.213853041875165e-05,
      "loss": 2.4107,
      "step": 13573
    },
    {
      "epoch": 0.3574927574400843,
      "grad_norm": 3.5954642295837402,
      "learning_rate": 3.2137213589676064e-05,
      "loss": 1.3837,
      "step": 13574
    },
    {
      "epoch": 0.357519094021596,
      "grad_norm": 1.7816054821014404,
      "learning_rate": 3.213589676060047e-05,
      "loss": 1.7309,
      "step": 13575
    },
    {
      "epoch": 0.35754543060310773,
      "grad_norm": 1.815504789352417,
      "learning_rate": 3.213457993152489e-05,
      "loss": 1.7403,
      "step": 13576
    },
    {
      "epoch": 0.3575717671846194,
      "grad_norm": 4.519070148468018,
      "learning_rate": 3.2133263102449304e-05,
      "loss": 1.2178,
      "step": 13577
    },
    {
      "epoch": 0.35759810376613116,
      "grad_norm": 1.7884514331817627,
      "learning_rate": 3.213194627337372e-05,
      "loss": 1.8143,
      "step": 13578
    },
    {
      "epoch": 0.3576244403476429,
      "grad_norm": 2.0682077407836914,
      "learning_rate": 3.2130629444298135e-05,
      "loss": 1.939,
      "step": 13579
    },
    {
      "epoch": 0.3576507769291546,
      "grad_norm": 1.8512648344039917,
      "learning_rate": 3.2129312615222544e-05,
      "loss": 1.9758,
      "step": 13580
    },
    {
      "epoch": 0.35767711351066633,
      "grad_norm": 1.7880021333694458,
      "learning_rate": 3.212799578614696e-05,
      "loss": 1.6864,
      "step": 13581
    },
    {
      "epoch": 0.357703450092178,
      "grad_norm": 2.221271514892578,
      "learning_rate": 3.212667895707137e-05,
      "loss": 2.2582,
      "step": 13582
    },
    {
      "epoch": 0.35772978667368976,
      "grad_norm": 2.561495065689087,
      "learning_rate": 3.212536212799579e-05,
      "loss": 1.4875,
      "step": 13583
    },
    {
      "epoch": 0.35775612325520145,
      "grad_norm": 3.4658031463623047,
      "learning_rate": 3.21240452989202e-05,
      "loss": 0.9863,
      "step": 13584
    },
    {
      "epoch": 0.3577824598367132,
      "grad_norm": 1.603253960609436,
      "learning_rate": 3.2122728469844615e-05,
      "loss": 2.396,
      "step": 13585
    },
    {
      "epoch": 0.35780879641822494,
      "grad_norm": 2.470778465270996,
      "learning_rate": 3.212141164076903e-05,
      "loss": 1.7577,
      "step": 13586
    },
    {
      "epoch": 0.3578351329997366,
      "grad_norm": 1.4860738515853882,
      "learning_rate": 3.2120094811693446e-05,
      "loss": 2.0019,
      "step": 13587
    },
    {
      "epoch": 0.35786146958124837,
      "grad_norm": 4.009979724884033,
      "learning_rate": 3.211877798261786e-05,
      "loss": 1.8871,
      "step": 13588
    },
    {
      "epoch": 0.35788780616276006,
      "grad_norm": 3.2963204383850098,
      "learning_rate": 3.211746115354227e-05,
      "loss": 1.2485,
      "step": 13589
    },
    {
      "epoch": 0.3579141427442718,
      "grad_norm": 1.6084405183792114,
      "learning_rate": 3.2116144324466686e-05,
      "loss": 1.7061,
      "step": 13590
    },
    {
      "epoch": 0.3579404793257835,
      "grad_norm": 2.6024277210235596,
      "learning_rate": 3.2114827495391095e-05,
      "loss": 1.8364,
      "step": 13591
    },
    {
      "epoch": 0.35796681590729523,
      "grad_norm": 4.801141262054443,
      "learning_rate": 3.211351066631552e-05,
      "loss": 1.0169,
      "step": 13592
    },
    {
      "epoch": 0.357993152488807,
      "grad_norm": 2.9791507720947266,
      "learning_rate": 3.2112193837239927e-05,
      "loss": 2.1186,
      "step": 13593
    },
    {
      "epoch": 0.35801948907031866,
      "grad_norm": 1.6529208421707153,
      "learning_rate": 3.211087700816434e-05,
      "loss": 1.6764,
      "step": 13594
    },
    {
      "epoch": 0.3580458256518304,
      "grad_norm": 1.5181199312210083,
      "learning_rate": 3.210956017908876e-05,
      "loss": 1.9035,
      "step": 13595
    },
    {
      "epoch": 0.3580721622333421,
      "grad_norm": 1.9680874347686768,
      "learning_rate": 3.2108243350013167e-05,
      "loss": 2.2813,
      "step": 13596
    },
    {
      "epoch": 0.35809849881485384,
      "grad_norm": 2.8364412784576416,
      "learning_rate": 3.210692652093759e-05,
      "loss": 1.4889,
      "step": 13597
    },
    {
      "epoch": 0.3581248353963655,
      "grad_norm": 2.6851561069488525,
      "learning_rate": 3.2105609691862e-05,
      "loss": 1.1029,
      "step": 13598
    },
    {
      "epoch": 0.35815117197787727,
      "grad_norm": 3.910747766494751,
      "learning_rate": 3.210429286278641e-05,
      "loss": 1.1004,
      "step": 13599
    },
    {
      "epoch": 0.358177508559389,
      "grad_norm": 1.3464211225509644,
      "learning_rate": 3.210297603371082e-05,
      "loss": 1.922,
      "step": 13600
    },
    {
      "epoch": 0.3582038451409007,
      "grad_norm": 3.04995059967041,
      "learning_rate": 3.210165920463524e-05,
      "loss": 1.5378,
      "step": 13601
    },
    {
      "epoch": 0.35823018172241244,
      "grad_norm": 2.3936808109283447,
      "learning_rate": 3.210034237555965e-05,
      "loss": 1.7822,
      "step": 13602
    },
    {
      "epoch": 0.35825651830392413,
      "grad_norm": 3.0493557453155518,
      "learning_rate": 3.209902554648407e-05,
      "loss": 1.9069,
      "step": 13603
    },
    {
      "epoch": 0.3582828548854359,
      "grad_norm": 3.2228777408599854,
      "learning_rate": 3.2097708717408485e-05,
      "loss": 1.3135,
      "step": 13604
    },
    {
      "epoch": 0.35830919146694756,
      "grad_norm": 2.2920727729797363,
      "learning_rate": 3.209639188833289e-05,
      "loss": 1.9463,
      "step": 13605
    },
    {
      "epoch": 0.3583355280484593,
      "grad_norm": 1.8838752508163452,
      "learning_rate": 3.2095075059257316e-05,
      "loss": 1.9288,
      "step": 13606
    },
    {
      "epoch": 0.35836186462997105,
      "grad_norm": 1.6796777248382568,
      "learning_rate": 3.2093758230181725e-05,
      "loss": 1.7367,
      "step": 13607
    },
    {
      "epoch": 0.35838820121148274,
      "grad_norm": 4.060592174530029,
      "learning_rate": 3.209244140110614e-05,
      "loss": 1.1152,
      "step": 13608
    },
    {
      "epoch": 0.3584145377929945,
      "grad_norm": 1.508970022201538,
      "learning_rate": 3.209112457203055e-05,
      "loss": 1.6006,
      "step": 13609
    },
    {
      "epoch": 0.35844087437450617,
      "grad_norm": 1.9285054206848145,
      "learning_rate": 3.2089807742954965e-05,
      "loss": 1.9426,
      "step": 13610
    },
    {
      "epoch": 0.3584672109560179,
      "grad_norm": 1.9635648727416992,
      "learning_rate": 3.208849091387938e-05,
      "loss": 0.401,
      "step": 13611
    },
    {
      "epoch": 0.35849354753752966,
      "grad_norm": 2.7901957035064697,
      "learning_rate": 3.2087174084803796e-05,
      "loss": 1.241,
      "step": 13612
    },
    {
      "epoch": 0.35851988411904134,
      "grad_norm": 1.3184195756912231,
      "learning_rate": 3.2085857255728205e-05,
      "loss": 0.6769,
      "step": 13613
    },
    {
      "epoch": 0.3585462207005531,
      "grad_norm": 1.9425921440124512,
      "learning_rate": 3.208454042665262e-05,
      "loss": 1.1361,
      "step": 13614
    },
    {
      "epoch": 0.3585725572820648,
      "grad_norm": 2.5613203048706055,
      "learning_rate": 3.2083223597577036e-05,
      "loss": 1.5979,
      "step": 13615
    },
    {
      "epoch": 0.3585988938635765,
      "grad_norm": 4.758947372436523,
      "learning_rate": 3.208190676850145e-05,
      "loss": 1.0036,
      "step": 13616
    },
    {
      "epoch": 0.3586252304450882,
      "grad_norm": 1.8957648277282715,
      "learning_rate": 3.208058993942587e-05,
      "loss": 2.7816,
      "step": 13617
    },
    {
      "epoch": 0.35865156702659995,
      "grad_norm": 2.223864793777466,
      "learning_rate": 3.2079273110350276e-05,
      "loss": 0.4141,
      "step": 13618
    },
    {
      "epoch": 0.3586779036081117,
      "grad_norm": 1.6040529012680054,
      "learning_rate": 3.207795628127469e-05,
      "loss": 1.8158,
      "step": 13619
    },
    {
      "epoch": 0.3587042401896234,
      "grad_norm": 2.693751335144043,
      "learning_rate": 3.207663945219911e-05,
      "loss": 1.2111,
      "step": 13620
    },
    {
      "epoch": 0.3587305767711351,
      "grad_norm": 2.041635513305664,
      "learning_rate": 3.207532262312352e-05,
      "loss": 1.3927,
      "step": 13621
    },
    {
      "epoch": 0.3587569133526468,
      "grad_norm": 2.149848699569702,
      "learning_rate": 3.207400579404793e-05,
      "loss": 1.3231,
      "step": 13622
    },
    {
      "epoch": 0.35878324993415855,
      "grad_norm": 1.643959641456604,
      "learning_rate": 3.207268896497235e-05,
      "loss": 2.0312,
      "step": 13623
    },
    {
      "epoch": 0.35880958651567024,
      "grad_norm": 2.539884090423584,
      "learning_rate": 3.207137213589676e-05,
      "loss": 2.755,
      "step": 13624
    },
    {
      "epoch": 0.358835923097182,
      "grad_norm": 2.047617197036743,
      "learning_rate": 3.207005530682118e-05,
      "loss": 1.6491,
      "step": 13625
    },
    {
      "epoch": 0.35886225967869373,
      "grad_norm": 2.015315532684326,
      "learning_rate": 3.2068738477745594e-05,
      "loss": 1.6363,
      "step": 13626
    },
    {
      "epoch": 0.3588885962602054,
      "grad_norm": 1.962422490119934,
      "learning_rate": 3.206742164867e-05,
      "loss": 2.2409,
      "step": 13627
    },
    {
      "epoch": 0.35891493284171716,
      "grad_norm": 3.1431314945220947,
      "learning_rate": 3.206610481959442e-05,
      "loss": 1.6786,
      "step": 13628
    },
    {
      "epoch": 0.35894126942322885,
      "grad_norm": 3.141082763671875,
      "learning_rate": 3.206478799051883e-05,
      "loss": 1.4124,
      "step": 13629
    },
    {
      "epoch": 0.3589676060047406,
      "grad_norm": 1.9669657945632935,
      "learning_rate": 3.206347116144325e-05,
      "loss": 1.5652,
      "step": 13630
    },
    {
      "epoch": 0.3589939425862523,
      "grad_norm": 3.1610116958618164,
      "learning_rate": 3.206215433236766e-05,
      "loss": 1.2665,
      "step": 13631
    },
    {
      "epoch": 0.359020279167764,
      "grad_norm": 2.7507312297821045,
      "learning_rate": 3.2060837503292074e-05,
      "loss": 1.9826,
      "step": 13632
    },
    {
      "epoch": 0.35904661574927577,
      "grad_norm": 1.7981044054031372,
      "learning_rate": 3.205952067421649e-05,
      "loss": 2.1147,
      "step": 13633
    },
    {
      "epoch": 0.35907295233078745,
      "grad_norm": 1.6805517673492432,
      "learning_rate": 3.20582038451409e-05,
      "loss": 1.9729,
      "step": 13634
    },
    {
      "epoch": 0.3590992889122992,
      "grad_norm": 1.812591791152954,
      "learning_rate": 3.205688701606532e-05,
      "loss": 2.0497,
      "step": 13635
    },
    {
      "epoch": 0.3591256254938109,
      "grad_norm": 2.0936026573181152,
      "learning_rate": 3.205557018698973e-05,
      "loss": 1.9917,
      "step": 13636
    },
    {
      "epoch": 0.35915196207532263,
      "grad_norm": 3.4688844680786133,
      "learning_rate": 3.2054253357914145e-05,
      "loss": 1.5249,
      "step": 13637
    },
    {
      "epoch": 0.3591782986568343,
      "grad_norm": 2.9845259189605713,
      "learning_rate": 3.2052936528838554e-05,
      "loss": 1.0429,
      "step": 13638
    },
    {
      "epoch": 0.35920463523834606,
      "grad_norm": 3.8587541580200195,
      "learning_rate": 3.2051619699762976e-05,
      "loss": 2.0806,
      "step": 13639
    },
    {
      "epoch": 0.3592309718198578,
      "grad_norm": 2.1985764503479004,
      "learning_rate": 3.2050302870687385e-05,
      "loss": 1.6956,
      "step": 13640
    },
    {
      "epoch": 0.3592573084013695,
      "grad_norm": 2.889796495437622,
      "learning_rate": 3.20489860416118e-05,
      "loss": 0.9912,
      "step": 13641
    },
    {
      "epoch": 0.35928364498288123,
      "grad_norm": 2.5522637367248535,
      "learning_rate": 3.2047669212536216e-05,
      "loss": 2.0458,
      "step": 13642
    },
    {
      "epoch": 0.3593099815643929,
      "grad_norm": 1.8815504312515259,
      "learning_rate": 3.2046352383460625e-05,
      "loss": 1.8488,
      "step": 13643
    },
    {
      "epoch": 0.35933631814590467,
      "grad_norm": 2.516996383666992,
      "learning_rate": 3.204503555438505e-05,
      "loss": 1.8765,
      "step": 13644
    },
    {
      "epoch": 0.3593626547274164,
      "grad_norm": 4.308852672576904,
      "learning_rate": 3.2043718725309456e-05,
      "loss": 1.0197,
      "step": 13645
    },
    {
      "epoch": 0.3593889913089281,
      "grad_norm": 3.4652295112609863,
      "learning_rate": 3.204240189623387e-05,
      "loss": 0.9293,
      "step": 13646
    },
    {
      "epoch": 0.35941532789043984,
      "grad_norm": 1.6348474025726318,
      "learning_rate": 3.204108506715828e-05,
      "loss": 1.4954,
      "step": 13647
    },
    {
      "epoch": 0.3594416644719515,
      "grad_norm": 2.7688302993774414,
      "learning_rate": 3.2039768238082696e-05,
      "loss": 1.2846,
      "step": 13648
    },
    {
      "epoch": 0.35946800105346327,
      "grad_norm": 2.0089244842529297,
      "learning_rate": 3.203845140900711e-05,
      "loss": 1.5039,
      "step": 13649
    },
    {
      "epoch": 0.35949433763497496,
      "grad_norm": 2.0078160762786865,
      "learning_rate": 3.203713457993153e-05,
      "loss": 2.3831,
      "step": 13650
    },
    {
      "epoch": 0.3595206742164867,
      "grad_norm": 1.9557090997695923,
      "learning_rate": 3.203581775085594e-05,
      "loss": 1.8255,
      "step": 13651
    },
    {
      "epoch": 0.35954701079799845,
      "grad_norm": 2.9324398040771484,
      "learning_rate": 3.203450092178035e-05,
      "loss": 2.2562,
      "step": 13652
    },
    {
      "epoch": 0.35957334737951013,
      "grad_norm": 2.6382675170898438,
      "learning_rate": 3.2033184092704774e-05,
      "loss": 2.0875,
      "step": 13653
    },
    {
      "epoch": 0.3595996839610219,
      "grad_norm": 1.9812148809432983,
      "learning_rate": 3.203186726362918e-05,
      "loss": 1.0656,
      "step": 13654
    },
    {
      "epoch": 0.35962602054253356,
      "grad_norm": 2.2188010215759277,
      "learning_rate": 3.20305504345536e-05,
      "loss": 1.8069,
      "step": 13655
    },
    {
      "epoch": 0.3596523571240453,
      "grad_norm": 3.9850947856903076,
      "learning_rate": 3.202923360547801e-05,
      "loss": 1.0791,
      "step": 13656
    },
    {
      "epoch": 0.359678693705557,
      "grad_norm": 2.224342107772827,
      "learning_rate": 3.202791677640242e-05,
      "loss": 1.6742,
      "step": 13657
    },
    {
      "epoch": 0.35970503028706874,
      "grad_norm": 3.0013134479522705,
      "learning_rate": 3.202659994732684e-05,
      "loss": 0.882,
      "step": 13658
    },
    {
      "epoch": 0.3597313668685805,
      "grad_norm": 2.805025100708008,
      "learning_rate": 3.2025283118251254e-05,
      "loss": 0.9205,
      "step": 13659
    },
    {
      "epoch": 0.35975770345009217,
      "grad_norm": 1.639492154121399,
      "learning_rate": 3.202396628917566e-05,
      "loss": 1.8974,
      "step": 13660
    },
    {
      "epoch": 0.3597840400316039,
      "grad_norm": 1.6123700141906738,
      "learning_rate": 3.202264946010008e-05,
      "loss": 1.8357,
      "step": 13661
    },
    {
      "epoch": 0.3598103766131156,
      "grad_norm": 3.004641532897949,
      "learning_rate": 3.2021332631024494e-05,
      "loss": 0.6059,
      "step": 13662
    },
    {
      "epoch": 0.35983671319462734,
      "grad_norm": 1.9932104349136353,
      "learning_rate": 3.202001580194891e-05,
      "loss": 1.8332,
      "step": 13663
    },
    {
      "epoch": 0.35986304977613903,
      "grad_norm": 2.7735795974731445,
      "learning_rate": 3.2018698972873326e-05,
      "loss": 2.023,
      "step": 13664
    },
    {
      "epoch": 0.3598893863576508,
      "grad_norm": 1.7488255500793457,
      "learning_rate": 3.2017382143797734e-05,
      "loss": 1.8532,
      "step": 13665
    },
    {
      "epoch": 0.3599157229391625,
      "grad_norm": 2.5629940032958984,
      "learning_rate": 3.201606531472215e-05,
      "loss": 2.1842,
      "step": 13666
    },
    {
      "epoch": 0.3599420595206742,
      "grad_norm": 1.6515700817108154,
      "learning_rate": 3.2014748485646566e-05,
      "loss": 1.3936,
      "step": 13667
    },
    {
      "epoch": 0.35996839610218595,
      "grad_norm": 2.527289390563965,
      "learning_rate": 3.201343165657098e-05,
      "loss": 1.6333,
      "step": 13668
    },
    {
      "epoch": 0.35999473268369764,
      "grad_norm": 1.7834123373031616,
      "learning_rate": 3.201211482749539e-05,
      "loss": 1.8042,
      "step": 13669
    },
    {
      "epoch": 0.3600210692652094,
      "grad_norm": 3.378293752670288,
      "learning_rate": 3.2010797998419806e-05,
      "loss": 1.5895,
      "step": 13670
    },
    {
      "epoch": 0.36004740584672107,
      "grad_norm": 1.6988415718078613,
      "learning_rate": 3.200948116934422e-05,
      "loss": 1.6087,
      "step": 13671
    },
    {
      "epoch": 0.3600737424282328,
      "grad_norm": 2.0479817390441895,
      "learning_rate": 3.200816434026864e-05,
      "loss": 1.2008,
      "step": 13672
    },
    {
      "epoch": 0.36010007900974456,
      "grad_norm": 1.7988804578781128,
      "learning_rate": 3.200684751119305e-05,
      "loss": 1.5911,
      "step": 13673
    },
    {
      "epoch": 0.36012641559125624,
      "grad_norm": 1.8863316774368286,
      "learning_rate": 3.200553068211746e-05,
      "loss": 2.048,
      "step": 13674
    },
    {
      "epoch": 0.360152752172768,
      "grad_norm": 4.116148948669434,
      "learning_rate": 3.200421385304188e-05,
      "loss": 1.6954,
      "step": 13675
    },
    {
      "epoch": 0.3601790887542797,
      "grad_norm": 2.3305835723876953,
      "learning_rate": 3.2002897023966286e-05,
      "loss": 2.4382,
      "step": 13676
    },
    {
      "epoch": 0.3602054253357914,
      "grad_norm": 2.539477586746216,
      "learning_rate": 3.200158019489071e-05,
      "loss": 1.5695,
      "step": 13677
    },
    {
      "epoch": 0.36023176191730316,
      "grad_norm": 2.0611891746520996,
      "learning_rate": 3.200026336581512e-05,
      "loss": 1.953,
      "step": 13678
    },
    {
      "epoch": 0.36025809849881485,
      "grad_norm": 1.764420747756958,
      "learning_rate": 3.199894653673953e-05,
      "loss": 1.1791,
      "step": 13679
    },
    {
      "epoch": 0.3602844350803266,
      "grad_norm": 2.050950050354004,
      "learning_rate": 3.199762970766395e-05,
      "loss": 1.9048,
      "step": 13680
    },
    {
      "epoch": 0.3603107716618383,
      "grad_norm": 3.1922240257263184,
      "learning_rate": 3.199631287858836e-05,
      "loss": 1.1164,
      "step": 13681
    },
    {
      "epoch": 0.36033710824335,
      "grad_norm": 2.410733461380005,
      "learning_rate": 3.199499604951278e-05,
      "loss": 1.9025,
      "step": 13682
    },
    {
      "epoch": 0.3603634448248617,
      "grad_norm": 2.043968915939331,
      "learning_rate": 3.199367922043719e-05,
      "loss": 2.3087,
      "step": 13683
    },
    {
      "epoch": 0.36038978140637346,
      "grad_norm": 2.3882358074188232,
      "learning_rate": 3.1992362391361604e-05,
      "loss": 1.0314,
      "step": 13684
    },
    {
      "epoch": 0.3604161179878852,
      "grad_norm": 1.6741105318069458,
      "learning_rate": 3.199104556228601e-05,
      "loss": 1.1299,
      "step": 13685
    },
    {
      "epoch": 0.3604424545693969,
      "grad_norm": 1.5031324625015259,
      "learning_rate": 3.1989728733210435e-05,
      "loss": 2.1062,
      "step": 13686
    },
    {
      "epoch": 0.36046879115090863,
      "grad_norm": 2.597062349319458,
      "learning_rate": 3.1988411904134844e-05,
      "loss": 1.4521,
      "step": 13687
    },
    {
      "epoch": 0.3604951277324203,
      "grad_norm": 3.2964096069335938,
      "learning_rate": 3.198709507505926e-05,
      "loss": 1.5763,
      "step": 13688
    },
    {
      "epoch": 0.36052146431393206,
      "grad_norm": 2.20664644241333,
      "learning_rate": 3.1985778245983675e-05,
      "loss": 1.9681,
      "step": 13689
    },
    {
      "epoch": 0.36054780089544375,
      "grad_norm": 2.8661561012268066,
      "learning_rate": 3.1984461416908084e-05,
      "loss": 1.8483,
      "step": 13690
    },
    {
      "epoch": 0.3605741374769555,
      "grad_norm": 2.307616710662842,
      "learning_rate": 3.1983144587832506e-05,
      "loss": 1.8336,
      "step": 13691
    },
    {
      "epoch": 0.36060047405846724,
      "grad_norm": 1.659900426864624,
      "learning_rate": 3.1981827758756915e-05,
      "loss": 1.9063,
      "step": 13692
    },
    {
      "epoch": 0.3606268106399789,
      "grad_norm": 7.053070545196533,
      "learning_rate": 3.198051092968133e-05,
      "loss": 1.7106,
      "step": 13693
    },
    {
      "epoch": 0.36065314722149067,
      "grad_norm": 4.508133411407471,
      "learning_rate": 3.197919410060574e-05,
      "loss": 1.4931,
      "step": 13694
    },
    {
      "epoch": 0.36067948380300235,
      "grad_norm": 2.002432107925415,
      "learning_rate": 3.1977877271530155e-05,
      "loss": 1.9358,
      "step": 13695
    },
    {
      "epoch": 0.3607058203845141,
      "grad_norm": 2.658022880554199,
      "learning_rate": 3.197656044245457e-05,
      "loss": 0.9224,
      "step": 13696
    },
    {
      "epoch": 0.3607321569660258,
      "grad_norm": 2.70473575592041,
      "learning_rate": 3.1975243613378986e-05,
      "loss": 1.2302,
      "step": 13697
    },
    {
      "epoch": 0.36075849354753753,
      "grad_norm": 3.511639356613159,
      "learning_rate": 3.19739267843034e-05,
      "loss": 2.2309,
      "step": 13698
    },
    {
      "epoch": 0.36078483012904927,
      "grad_norm": 2.263486385345459,
      "learning_rate": 3.197260995522781e-05,
      "loss": 2.0399,
      "step": 13699
    },
    {
      "epoch": 0.36081116671056096,
      "grad_norm": 1.5785053968429565,
      "learning_rate": 3.197129312615223e-05,
      "loss": 1.9869,
      "step": 13700
    },
    {
      "epoch": 0.3608375032920727,
      "grad_norm": 1.7243232727050781,
      "learning_rate": 3.196997629707664e-05,
      "loss": 1.446,
      "step": 13701
    },
    {
      "epoch": 0.3608638398735844,
      "grad_norm": 2.390009880065918,
      "learning_rate": 3.196865946800106e-05,
      "loss": 1.6143,
      "step": 13702
    },
    {
      "epoch": 0.36089017645509613,
      "grad_norm": 3.2779555320739746,
      "learning_rate": 3.1967342638925466e-05,
      "loss": 0.884,
      "step": 13703
    },
    {
      "epoch": 0.3609165130366078,
      "grad_norm": 4.38844633102417,
      "learning_rate": 3.196602580984988e-05,
      "loss": 0.9997,
      "step": 13704
    },
    {
      "epoch": 0.36094284961811957,
      "grad_norm": 1.8490869998931885,
      "learning_rate": 3.19647089807743e-05,
      "loss": 1.1045,
      "step": 13705
    },
    {
      "epoch": 0.3609691861996313,
      "grad_norm": 2.193153142929077,
      "learning_rate": 3.196339215169871e-05,
      "loss": 1.3679,
      "step": 13706
    },
    {
      "epoch": 0.360995522781143,
      "grad_norm": 2.7777693271636963,
      "learning_rate": 3.196207532262313e-05,
      "loss": 1.8684,
      "step": 13707
    },
    {
      "epoch": 0.36102185936265474,
      "grad_norm": 1.6836715936660767,
      "learning_rate": 3.196075849354754e-05,
      "loss": 1.6561,
      "step": 13708
    },
    {
      "epoch": 0.36104819594416643,
      "grad_norm": 1.59881591796875,
      "learning_rate": 3.195944166447195e-05,
      "loss": 2.1252,
      "step": 13709
    },
    {
      "epoch": 0.36107453252567817,
      "grad_norm": 2.737924098968506,
      "learning_rate": 3.195812483539637e-05,
      "loss": 2.0593,
      "step": 13710
    },
    {
      "epoch": 0.36110086910718986,
      "grad_norm": 2.5390844345092773,
      "learning_rate": 3.1956808006320784e-05,
      "loss": 1.9723,
      "step": 13711
    },
    {
      "epoch": 0.3611272056887016,
      "grad_norm": 1.9286346435546875,
      "learning_rate": 3.195549117724519e-05,
      "loss": 2.3703,
      "step": 13712
    },
    {
      "epoch": 0.36115354227021335,
      "grad_norm": 2.622424840927124,
      "learning_rate": 3.195417434816961e-05,
      "loss": 0.4147,
      "step": 13713
    },
    {
      "epoch": 0.36117987885172503,
      "grad_norm": 1.4867925643920898,
      "learning_rate": 3.195285751909402e-05,
      "loss": 2.0865,
      "step": 13714
    },
    {
      "epoch": 0.3612062154332368,
      "grad_norm": 2.314683437347412,
      "learning_rate": 3.195154069001844e-05,
      "loss": 1.625,
      "step": 13715
    },
    {
      "epoch": 0.36123255201474846,
      "grad_norm": 1.9138222932815552,
      "learning_rate": 3.195022386094285e-05,
      "loss": 1.8405,
      "step": 13716
    },
    {
      "epoch": 0.3612588885962602,
      "grad_norm": 1.9646731615066528,
      "learning_rate": 3.1948907031867264e-05,
      "loss": 1.9014,
      "step": 13717
    },
    {
      "epoch": 0.36128522517777195,
      "grad_norm": 2.100433588027954,
      "learning_rate": 3.194759020279168e-05,
      "loss": 1.491,
      "step": 13718
    },
    {
      "epoch": 0.36131156175928364,
      "grad_norm": 1.6655570268630981,
      "learning_rate": 3.1946273373716095e-05,
      "loss": 2.4489,
      "step": 13719
    },
    {
      "epoch": 0.3613378983407954,
      "grad_norm": 2.423673391342163,
      "learning_rate": 3.194495654464051e-05,
      "loss": 1.9112,
      "step": 13720
    },
    {
      "epoch": 0.36136423492230707,
      "grad_norm": 1.5658369064331055,
      "learning_rate": 3.194363971556492e-05,
      "loss": 1.5957,
      "step": 13721
    },
    {
      "epoch": 0.3613905715038188,
      "grad_norm": 4.259023666381836,
      "learning_rate": 3.1942322886489335e-05,
      "loss": 1.4321,
      "step": 13722
    },
    {
      "epoch": 0.3614169080853305,
      "grad_norm": 1.8690143823623657,
      "learning_rate": 3.1941006057413744e-05,
      "loss": 2.0743,
      "step": 13723
    },
    {
      "epoch": 0.36144324466684224,
      "grad_norm": 3.2913100719451904,
      "learning_rate": 3.1939689228338167e-05,
      "loss": 1.6028,
      "step": 13724
    },
    {
      "epoch": 0.361469581248354,
      "grad_norm": 4.696711540222168,
      "learning_rate": 3.1938372399262575e-05,
      "loss": 1.1349,
      "step": 13725
    },
    {
      "epoch": 0.3614959178298657,
      "grad_norm": 1.7497495412826538,
      "learning_rate": 3.193705557018699e-05,
      "loss": 2.0273,
      "step": 13726
    },
    {
      "epoch": 0.3615222544113774,
      "grad_norm": 1.4944347143173218,
      "learning_rate": 3.1935738741111407e-05,
      "loss": 1.1767,
      "step": 13727
    },
    {
      "epoch": 0.3615485909928891,
      "grad_norm": 2.4703528881073,
      "learning_rate": 3.1934421912035815e-05,
      "loss": 1.9198,
      "step": 13728
    },
    {
      "epoch": 0.36157492757440085,
      "grad_norm": 1.8481627702713013,
      "learning_rate": 3.193310508296024e-05,
      "loss": 1.1765,
      "step": 13729
    },
    {
      "epoch": 0.36160126415591254,
      "grad_norm": 5.01578950881958,
      "learning_rate": 3.1931788253884647e-05,
      "loss": 2.1003,
      "step": 13730
    },
    {
      "epoch": 0.3616276007374243,
      "grad_norm": 2.382566452026367,
      "learning_rate": 3.193047142480906e-05,
      "loss": 2.1148,
      "step": 13731
    },
    {
      "epoch": 0.361653937318936,
      "grad_norm": 2.1275060176849365,
      "learning_rate": 3.192915459573347e-05,
      "loss": 2.4598,
      "step": 13732
    },
    {
      "epoch": 0.3616802739004477,
      "grad_norm": 5.664269924163818,
      "learning_rate": 3.192783776665789e-05,
      "loss": 1.2847,
      "step": 13733
    },
    {
      "epoch": 0.36170661048195946,
      "grad_norm": 1.9295722246170044,
      "learning_rate": 3.19265209375823e-05,
      "loss": 2.077,
      "step": 13734
    },
    {
      "epoch": 0.36173294706347114,
      "grad_norm": 4.477755546569824,
      "learning_rate": 3.192520410850672e-05,
      "loss": 1.0247,
      "step": 13735
    },
    {
      "epoch": 0.3617592836449829,
      "grad_norm": 4.621309757232666,
      "learning_rate": 3.192388727943113e-05,
      "loss": 1.1349,
      "step": 13736
    },
    {
      "epoch": 0.3617856202264946,
      "grad_norm": 1.9026362895965576,
      "learning_rate": 3.192257045035554e-05,
      "loss": 0.9632,
      "step": 13737
    },
    {
      "epoch": 0.3618119568080063,
      "grad_norm": 2.698049783706665,
      "learning_rate": 3.1921253621279965e-05,
      "loss": 1.5551,
      "step": 13738
    },
    {
      "epoch": 0.36183829338951806,
      "grad_norm": 1.847318410873413,
      "learning_rate": 3.191993679220437e-05,
      "loss": 1.8988,
      "step": 13739
    },
    {
      "epoch": 0.36186462997102975,
      "grad_norm": 3.2921340465545654,
      "learning_rate": 3.191861996312879e-05,
      "loss": 1.9521,
      "step": 13740
    },
    {
      "epoch": 0.3618909665525415,
      "grad_norm": 1.75319242477417,
      "learning_rate": 3.19173031340532e-05,
      "loss": 1.7666,
      "step": 13741
    },
    {
      "epoch": 0.3619173031340532,
      "grad_norm": 2.2795064449310303,
      "learning_rate": 3.1915986304977613e-05,
      "loss": 2.241,
      "step": 13742
    },
    {
      "epoch": 0.3619436397155649,
      "grad_norm": 3.087273120880127,
      "learning_rate": 3.191466947590203e-05,
      "loss": 1.6454,
      "step": 13743
    },
    {
      "epoch": 0.3619699762970766,
      "grad_norm": 2.4293100833892822,
      "learning_rate": 3.1913352646826445e-05,
      "loss": 1.8657,
      "step": 13744
    },
    {
      "epoch": 0.36199631287858836,
      "grad_norm": 2.0871572494506836,
      "learning_rate": 3.191203581775086e-05,
      "loss": 0.8201,
      "step": 13745
    },
    {
      "epoch": 0.3620226494601001,
      "grad_norm": 3.9278769493103027,
      "learning_rate": 3.191071898867527e-05,
      "loss": 1.5707,
      "step": 13746
    },
    {
      "epoch": 0.3620489860416118,
      "grad_norm": 2.2865798473358154,
      "learning_rate": 3.1909402159599685e-05,
      "loss": 1.307,
      "step": 13747
    },
    {
      "epoch": 0.36207532262312353,
      "grad_norm": 2.4687952995300293,
      "learning_rate": 3.19080853305241e-05,
      "loss": 1.5658,
      "step": 13748
    },
    {
      "epoch": 0.3621016592046352,
      "grad_norm": 1.641245722770691,
      "learning_rate": 3.1906768501448516e-05,
      "loss": 1.3165,
      "step": 13749
    },
    {
      "epoch": 0.36212799578614696,
      "grad_norm": 3.333493947982788,
      "learning_rate": 3.1905451672372925e-05,
      "loss": 1.6897,
      "step": 13750
    },
    {
      "epoch": 0.3621543323676587,
      "grad_norm": 2.264084577560425,
      "learning_rate": 3.190413484329734e-05,
      "loss": 2.0074,
      "step": 13751
    },
    {
      "epoch": 0.3621806689491704,
      "grad_norm": 2.8487136363983154,
      "learning_rate": 3.1902818014221756e-05,
      "loss": 1.7925,
      "step": 13752
    },
    {
      "epoch": 0.36220700553068214,
      "grad_norm": 1.8497778177261353,
      "learning_rate": 3.190150118514617e-05,
      "loss": 1.4377,
      "step": 13753
    },
    {
      "epoch": 0.3622333421121938,
      "grad_norm": 2.043375253677368,
      "learning_rate": 3.190018435607059e-05,
      "loss": 1.0672,
      "step": 13754
    },
    {
      "epoch": 0.36225967869370557,
      "grad_norm": 2.831580877304077,
      "learning_rate": 3.1898867526994996e-05,
      "loss": 1.8116,
      "step": 13755
    },
    {
      "epoch": 0.36228601527521725,
      "grad_norm": 2.812997579574585,
      "learning_rate": 3.189755069791941e-05,
      "loss": 2.0191,
      "step": 13756
    },
    {
      "epoch": 0.362312351856729,
      "grad_norm": 1.557414174079895,
      "learning_rate": 3.189623386884383e-05,
      "loss": 1.3359,
      "step": 13757
    },
    {
      "epoch": 0.36233868843824074,
      "grad_norm": 1.910851001739502,
      "learning_rate": 3.189491703976824e-05,
      "loss": 2.1032,
      "step": 13758
    },
    {
      "epoch": 0.36236502501975243,
      "grad_norm": 2.2622439861297607,
      "learning_rate": 3.189360021069265e-05,
      "loss": 1.8948,
      "step": 13759
    },
    {
      "epoch": 0.3623913616012642,
      "grad_norm": 2.6514997482299805,
      "learning_rate": 3.189228338161707e-05,
      "loss": 2.0692,
      "step": 13760
    },
    {
      "epoch": 0.36241769818277586,
      "grad_norm": 2.974579334259033,
      "learning_rate": 3.1890966552541476e-05,
      "loss": 2.5212,
      "step": 13761
    },
    {
      "epoch": 0.3624440347642876,
      "grad_norm": 2.500119924545288,
      "learning_rate": 3.18896497234659e-05,
      "loss": 2.3911,
      "step": 13762
    },
    {
      "epoch": 0.3624703713457993,
      "grad_norm": 1.5414624214172363,
      "learning_rate": 3.188833289439031e-05,
      "loss": 1.7751,
      "step": 13763
    },
    {
      "epoch": 0.36249670792731103,
      "grad_norm": 2.443995475769043,
      "learning_rate": 3.188701606531472e-05,
      "loss": 1.7468,
      "step": 13764
    },
    {
      "epoch": 0.3625230445088228,
      "grad_norm": 2.0128374099731445,
      "learning_rate": 3.188569923623914e-05,
      "loss": 2.0167,
      "step": 13765
    },
    {
      "epoch": 0.36254938109033447,
      "grad_norm": 3.2428340911865234,
      "learning_rate": 3.1884382407163554e-05,
      "loss": 1.6361,
      "step": 13766
    },
    {
      "epoch": 0.3625757176718462,
      "grad_norm": 3.4356157779693604,
      "learning_rate": 3.188306557808797e-05,
      "loss": 1.7618,
      "step": 13767
    },
    {
      "epoch": 0.3626020542533579,
      "grad_norm": 1.9007205963134766,
      "learning_rate": 3.188174874901238e-05,
      "loss": 1.5445,
      "step": 13768
    },
    {
      "epoch": 0.36262839083486964,
      "grad_norm": 2.2970147132873535,
      "learning_rate": 3.1880431919936794e-05,
      "loss": 1.9541,
      "step": 13769
    },
    {
      "epoch": 0.36265472741638133,
      "grad_norm": 2.8699581623077393,
      "learning_rate": 3.18791150908612e-05,
      "loss": 1.303,
      "step": 13770
    },
    {
      "epoch": 0.36268106399789307,
      "grad_norm": 2.1957647800445557,
      "learning_rate": 3.1877798261785625e-05,
      "loss": 2.1676,
      "step": 13771
    },
    {
      "epoch": 0.3627074005794048,
      "grad_norm": 3.043992757797241,
      "learning_rate": 3.1876481432710034e-05,
      "loss": 0.6439,
      "step": 13772
    },
    {
      "epoch": 0.3627337371609165,
      "grad_norm": 1.519282579421997,
      "learning_rate": 3.187516460363445e-05,
      "loss": 1.997,
      "step": 13773
    },
    {
      "epoch": 0.36276007374242825,
      "grad_norm": 2.8921141624450684,
      "learning_rate": 3.1873847774558865e-05,
      "loss": 1.475,
      "step": 13774
    },
    {
      "epoch": 0.36278641032393993,
      "grad_norm": 3.218425750732422,
      "learning_rate": 3.1872530945483274e-05,
      "loss": 1.6625,
      "step": 13775
    },
    {
      "epoch": 0.3628127469054517,
      "grad_norm": 1.689947485923767,
      "learning_rate": 3.1871214116407696e-05,
      "loss": 1.5918,
      "step": 13776
    },
    {
      "epoch": 0.36283908348696337,
      "grad_norm": 2.2229950428009033,
      "learning_rate": 3.1869897287332105e-05,
      "loss": 2.3097,
      "step": 13777
    },
    {
      "epoch": 0.3628654200684751,
      "grad_norm": 2.1253154277801514,
      "learning_rate": 3.186858045825652e-05,
      "loss": 1.8892,
      "step": 13778
    },
    {
      "epoch": 0.36289175664998685,
      "grad_norm": 1.8596086502075195,
      "learning_rate": 3.186726362918093e-05,
      "loss": 2.1572,
      "step": 13779
    },
    {
      "epoch": 0.36291809323149854,
      "grad_norm": 2.1153604984283447,
      "learning_rate": 3.1865946800105345e-05,
      "loss": 1.6767,
      "step": 13780
    },
    {
      "epoch": 0.3629444298130103,
      "grad_norm": 2.6510214805603027,
      "learning_rate": 3.186462997102976e-05,
      "loss": 1.7224,
      "step": 13781
    },
    {
      "epoch": 0.36297076639452197,
      "grad_norm": 4.937135696411133,
      "learning_rate": 3.1863313141954176e-05,
      "loss": 2.2747,
      "step": 13782
    },
    {
      "epoch": 0.3629971029760337,
      "grad_norm": 2.1835336685180664,
      "learning_rate": 3.186199631287859e-05,
      "loss": 1.5629,
      "step": 13783
    },
    {
      "epoch": 0.36302343955754546,
      "grad_norm": 2.072180986404419,
      "learning_rate": 3.1860679483803e-05,
      "loss": 0.8332,
      "step": 13784
    },
    {
      "epoch": 0.36304977613905715,
      "grad_norm": 1.637162685394287,
      "learning_rate": 3.185936265472742e-05,
      "loss": 1.308,
      "step": 13785
    },
    {
      "epoch": 0.3630761127205689,
      "grad_norm": 2.446707248687744,
      "learning_rate": 3.185804582565183e-05,
      "loss": 1.0632,
      "step": 13786
    },
    {
      "epoch": 0.3631024493020806,
      "grad_norm": 2.045194625854492,
      "learning_rate": 3.185672899657625e-05,
      "loss": 1.8064,
      "step": 13787
    },
    {
      "epoch": 0.3631287858835923,
      "grad_norm": 2.5239710807800293,
      "learning_rate": 3.1855412167500656e-05,
      "loss": 2.092,
      "step": 13788
    },
    {
      "epoch": 0.363155122465104,
      "grad_norm": 1.823291540145874,
      "learning_rate": 3.185409533842507e-05,
      "loss": 0.7776,
      "step": 13789
    },
    {
      "epoch": 0.36318145904661575,
      "grad_norm": 1.759790301322937,
      "learning_rate": 3.185277850934949e-05,
      "loss": 0.7001,
      "step": 13790
    },
    {
      "epoch": 0.3632077956281275,
      "grad_norm": 2.4322996139526367,
      "learning_rate": 3.18514616802739e-05,
      "loss": 2.4051,
      "step": 13791
    },
    {
      "epoch": 0.3632341322096392,
      "grad_norm": 1.4365615844726562,
      "learning_rate": 3.185014485119832e-05,
      "loss": 1.5694,
      "step": 13792
    },
    {
      "epoch": 0.3632604687911509,
      "grad_norm": 1.5395278930664062,
      "learning_rate": 3.184882802212273e-05,
      "loss": 1.7923,
      "step": 13793
    },
    {
      "epoch": 0.3632868053726626,
      "grad_norm": 1.8849338293075562,
      "learning_rate": 3.184751119304714e-05,
      "loss": 1.4222,
      "step": 13794
    },
    {
      "epoch": 0.36331314195417436,
      "grad_norm": 2.9102373123168945,
      "learning_rate": 3.184619436397156e-05,
      "loss": 1.2308,
      "step": 13795
    },
    {
      "epoch": 0.36333947853568604,
      "grad_norm": 2.196582794189453,
      "learning_rate": 3.1844877534895974e-05,
      "loss": 2.0528,
      "step": 13796
    },
    {
      "epoch": 0.3633658151171978,
      "grad_norm": 1.4518133401870728,
      "learning_rate": 3.184356070582038e-05,
      "loss": 0.5009,
      "step": 13797
    },
    {
      "epoch": 0.36339215169870953,
      "grad_norm": 2.4499549865722656,
      "learning_rate": 3.18422438767448e-05,
      "loss": 1.2751,
      "step": 13798
    },
    {
      "epoch": 0.3634184882802212,
      "grad_norm": 4.62742280960083,
      "learning_rate": 3.1840927047669214e-05,
      "loss": 1.2049,
      "step": 13799
    },
    {
      "epoch": 0.36344482486173296,
      "grad_norm": 3.095268487930298,
      "learning_rate": 3.183961021859363e-05,
      "loss": 1.7698,
      "step": 13800
    },
    {
      "epoch": 0.36347116144324465,
      "grad_norm": 2.9536473751068115,
      "learning_rate": 3.1838293389518046e-05,
      "loss": 1.6317,
      "step": 13801
    },
    {
      "epoch": 0.3634974980247564,
      "grad_norm": 1.8991554975509644,
      "learning_rate": 3.1836976560442454e-05,
      "loss": 1.1643,
      "step": 13802
    },
    {
      "epoch": 0.3635238346062681,
      "grad_norm": 2.9806320667266846,
      "learning_rate": 3.183565973136687e-05,
      "loss": 1.5999,
      "step": 13803
    },
    {
      "epoch": 0.3635501711877798,
      "grad_norm": 2.817274570465088,
      "learning_rate": 3.1834342902291286e-05,
      "loss": 1.756,
      "step": 13804
    },
    {
      "epoch": 0.36357650776929157,
      "grad_norm": 2.6712772846221924,
      "learning_rate": 3.18330260732157e-05,
      "loss": 1.7977,
      "step": 13805
    },
    {
      "epoch": 0.36360284435080326,
      "grad_norm": 2.101950168609619,
      "learning_rate": 3.183170924414011e-05,
      "loss": 1.6977,
      "step": 13806
    },
    {
      "epoch": 0.363629180932315,
      "grad_norm": 3.1772356033325195,
      "learning_rate": 3.1830392415064526e-05,
      "loss": 0.657,
      "step": 13807
    },
    {
      "epoch": 0.3636555175138267,
      "grad_norm": 3.3101799488067627,
      "learning_rate": 3.1829075585988934e-05,
      "loss": 0.3071,
      "step": 13808
    },
    {
      "epoch": 0.36368185409533843,
      "grad_norm": 3.0180869102478027,
      "learning_rate": 3.182775875691336e-05,
      "loss": 0.721,
      "step": 13809
    },
    {
      "epoch": 0.3637081906768501,
      "grad_norm": 3.193493366241455,
      "learning_rate": 3.182644192783777e-05,
      "loss": 1.6397,
      "step": 13810
    },
    {
      "epoch": 0.36373452725836186,
      "grad_norm": 2.2841835021972656,
      "learning_rate": 3.182512509876218e-05,
      "loss": 0.2804,
      "step": 13811
    },
    {
      "epoch": 0.3637608638398736,
      "grad_norm": 2.378143548965454,
      "learning_rate": 3.18238082696866e-05,
      "loss": 1.7643,
      "step": 13812
    },
    {
      "epoch": 0.3637872004213853,
      "grad_norm": 3.1494781970977783,
      "learning_rate": 3.1822491440611006e-05,
      "loss": 1.851,
      "step": 13813
    },
    {
      "epoch": 0.36381353700289704,
      "grad_norm": 2.591458797454834,
      "learning_rate": 3.182117461153543e-05,
      "loss": 0.8916,
      "step": 13814
    },
    {
      "epoch": 0.3638398735844087,
      "grad_norm": 2.2837886810302734,
      "learning_rate": 3.181985778245984e-05,
      "loss": 1.909,
      "step": 13815
    },
    {
      "epoch": 0.36386621016592047,
      "grad_norm": 6.3013529777526855,
      "learning_rate": 3.181854095338425e-05,
      "loss": 0.7834,
      "step": 13816
    },
    {
      "epoch": 0.3638925467474322,
      "grad_norm": 1.827169418334961,
      "learning_rate": 3.181722412430866e-05,
      "loss": 1.7862,
      "step": 13817
    },
    {
      "epoch": 0.3639188833289439,
      "grad_norm": 1.9870465993881226,
      "learning_rate": 3.1815907295233084e-05,
      "loss": 1.679,
      "step": 13818
    },
    {
      "epoch": 0.36394521991045564,
      "grad_norm": 2.318375825881958,
      "learning_rate": 3.181459046615749e-05,
      "loss": 2.2049,
      "step": 13819
    },
    {
      "epoch": 0.36397155649196733,
      "grad_norm": 3.828956127166748,
      "learning_rate": 3.181327363708191e-05,
      "loss": 1.2161,
      "step": 13820
    },
    {
      "epoch": 0.3639978930734791,
      "grad_norm": 3.1623432636260986,
      "learning_rate": 3.1811956808006324e-05,
      "loss": 1.1136,
      "step": 13821
    },
    {
      "epoch": 0.36402422965499076,
      "grad_norm": 3.3095269203186035,
      "learning_rate": 3.181063997893073e-05,
      "loss": 1.496,
      "step": 13822
    },
    {
      "epoch": 0.3640505662365025,
      "grad_norm": 1.5347200632095337,
      "learning_rate": 3.1809323149855155e-05,
      "loss": 1.5613,
      "step": 13823
    },
    {
      "epoch": 0.36407690281801425,
      "grad_norm": 1.721107840538025,
      "learning_rate": 3.1808006320779564e-05,
      "loss": 2.2651,
      "step": 13824
    },
    {
      "epoch": 0.36410323939952594,
      "grad_norm": 1.8439258337020874,
      "learning_rate": 3.180668949170398e-05,
      "loss": 2.05,
      "step": 13825
    },
    {
      "epoch": 0.3641295759810377,
      "grad_norm": 1.7087953090667725,
      "learning_rate": 3.180537266262839e-05,
      "loss": 2.0225,
      "step": 13826
    },
    {
      "epoch": 0.36415591256254937,
      "grad_norm": 1.8066130876541138,
      "learning_rate": 3.1804055833552804e-05,
      "loss": 1.496,
      "step": 13827
    },
    {
      "epoch": 0.3641822491440611,
      "grad_norm": 1.5990849733352661,
      "learning_rate": 3.180273900447722e-05,
      "loss": 1.9284,
      "step": 13828
    },
    {
      "epoch": 0.3642085857255728,
      "grad_norm": 3.6670408248901367,
      "learning_rate": 3.1801422175401635e-05,
      "loss": 1.157,
      "step": 13829
    },
    {
      "epoch": 0.36423492230708454,
      "grad_norm": 3.810096502304077,
      "learning_rate": 3.180010534632605e-05,
      "loss": 1.1495,
      "step": 13830
    },
    {
      "epoch": 0.3642612588885963,
      "grad_norm": 2.1517443656921387,
      "learning_rate": 3.179878851725046e-05,
      "loss": 1.5242,
      "step": 13831
    },
    {
      "epoch": 0.36428759547010797,
      "grad_norm": 1.8491172790527344,
      "learning_rate": 3.179747168817488e-05,
      "loss": 1.7226,
      "step": 13832
    },
    {
      "epoch": 0.3643139320516197,
      "grad_norm": 2.996687412261963,
      "learning_rate": 3.179615485909929e-05,
      "loss": 0.4152,
      "step": 13833
    },
    {
      "epoch": 0.3643402686331314,
      "grad_norm": 2.764782667160034,
      "learning_rate": 3.1794838030023706e-05,
      "loss": 1.2242,
      "step": 13834
    },
    {
      "epoch": 0.36436660521464315,
      "grad_norm": 1.8812557458877563,
      "learning_rate": 3.1793521200948115e-05,
      "loss": 2.0346,
      "step": 13835
    },
    {
      "epoch": 0.36439294179615483,
      "grad_norm": 2.4403786659240723,
      "learning_rate": 3.179220437187253e-05,
      "loss": 2.2777,
      "step": 13836
    },
    {
      "epoch": 0.3644192783776666,
      "grad_norm": 2.702369213104248,
      "learning_rate": 3.1790887542796946e-05,
      "loss": 0.9321,
      "step": 13837
    },
    {
      "epoch": 0.3644456149591783,
      "grad_norm": 2.129138946533203,
      "learning_rate": 3.178957071372136e-05,
      "loss": 2.5341,
      "step": 13838
    },
    {
      "epoch": 0.36447195154069,
      "grad_norm": 1.6012060642242432,
      "learning_rate": 3.178825388464578e-05,
      "loss": 1.3708,
      "step": 13839
    },
    {
      "epoch": 0.36449828812220175,
      "grad_norm": 1.7756980657577515,
      "learning_rate": 3.1786937055570186e-05,
      "loss": 2.2441,
      "step": 13840
    },
    {
      "epoch": 0.36452462470371344,
      "grad_norm": 3.4371607303619385,
      "learning_rate": 3.17856202264946e-05,
      "loss": 0.7192,
      "step": 13841
    },
    {
      "epoch": 0.3645509612852252,
      "grad_norm": 1.8897078037261963,
      "learning_rate": 3.178430339741902e-05,
      "loss": 1.9421,
      "step": 13842
    },
    {
      "epoch": 0.36457729786673687,
      "grad_norm": 2.871610164642334,
      "learning_rate": 3.178298656834343e-05,
      "loss": 1.1143,
      "step": 13843
    },
    {
      "epoch": 0.3646036344482486,
      "grad_norm": 2.6649508476257324,
      "learning_rate": 3.178166973926784e-05,
      "loss": 1.7913,
      "step": 13844
    },
    {
      "epoch": 0.36462997102976036,
      "grad_norm": 1.9537975788116455,
      "learning_rate": 3.178035291019226e-05,
      "loss": 1.6434,
      "step": 13845
    },
    {
      "epoch": 0.36465630761127205,
      "grad_norm": 1.7834559679031372,
      "learning_rate": 3.177903608111667e-05,
      "loss": 1.6543,
      "step": 13846
    },
    {
      "epoch": 0.3646826441927838,
      "grad_norm": 2.697648048400879,
      "learning_rate": 3.177771925204109e-05,
      "loss": 1.8356,
      "step": 13847
    },
    {
      "epoch": 0.3647089807742955,
      "grad_norm": 1.8176823854446411,
      "learning_rate": 3.1776402422965504e-05,
      "loss": 1.0223,
      "step": 13848
    },
    {
      "epoch": 0.3647353173558072,
      "grad_norm": 5.114029407501221,
      "learning_rate": 3.177508559388991e-05,
      "loss": 1.688,
      "step": 13849
    },
    {
      "epoch": 0.36476165393731896,
      "grad_norm": 3.140770196914673,
      "learning_rate": 3.177376876481433e-05,
      "loss": 1.5448,
      "step": 13850
    },
    {
      "epoch": 0.36478799051883065,
      "grad_norm": 3.648015022277832,
      "learning_rate": 3.1772451935738744e-05,
      "loss": 1.0719,
      "step": 13851
    },
    {
      "epoch": 0.3648143271003424,
      "grad_norm": 2.382293939590454,
      "learning_rate": 3.177113510666316e-05,
      "loss": 1.9129,
      "step": 13852
    },
    {
      "epoch": 0.3648406636818541,
      "grad_norm": 1.719102144241333,
      "learning_rate": 3.176981827758757e-05,
      "loss": 2.1479,
      "step": 13853
    },
    {
      "epoch": 0.3648670002633658,
      "grad_norm": 2.2585225105285645,
      "learning_rate": 3.1768501448511984e-05,
      "loss": 1.5737,
      "step": 13854
    },
    {
      "epoch": 0.3648933368448775,
      "grad_norm": 2.392477512359619,
      "learning_rate": 3.17671846194364e-05,
      "loss": 0.1898,
      "step": 13855
    },
    {
      "epoch": 0.36491967342638926,
      "grad_norm": 2.503675699234009,
      "learning_rate": 3.1765867790360815e-05,
      "loss": 1.4292,
      "step": 13856
    },
    {
      "epoch": 0.364946010007901,
      "grad_norm": 2.0193629264831543,
      "learning_rate": 3.176455096128523e-05,
      "loss": 1.087,
      "step": 13857
    },
    {
      "epoch": 0.3649723465894127,
      "grad_norm": 2.2090227603912354,
      "learning_rate": 3.176323413220964e-05,
      "loss": 1.7748,
      "step": 13858
    },
    {
      "epoch": 0.36499868317092443,
      "grad_norm": 2.4805684089660645,
      "learning_rate": 3.1761917303134055e-05,
      "loss": 0.3666,
      "step": 13859
    },
    {
      "epoch": 0.3650250197524361,
      "grad_norm": 3.4558446407318115,
      "learning_rate": 3.1760600474058464e-05,
      "loss": 1.5241,
      "step": 13860
    },
    {
      "epoch": 0.36505135633394786,
      "grad_norm": 1.7901235818862915,
      "learning_rate": 3.1759283644982887e-05,
      "loss": 1.3702,
      "step": 13861
    },
    {
      "epoch": 0.36507769291545955,
      "grad_norm": 2.0470352172851562,
      "learning_rate": 3.1757966815907295e-05,
      "loss": 1.6389,
      "step": 13862
    },
    {
      "epoch": 0.3651040294969713,
      "grad_norm": 2.186436653137207,
      "learning_rate": 3.175664998683171e-05,
      "loss": 2.1418,
      "step": 13863
    },
    {
      "epoch": 0.36513036607848304,
      "grad_norm": 2.1997463703155518,
      "learning_rate": 3.175533315775612e-05,
      "loss": 1.5805,
      "step": 13864
    },
    {
      "epoch": 0.3651567026599947,
      "grad_norm": 1.9361793994903564,
      "learning_rate": 3.175401632868054e-05,
      "loss": 0.3786,
      "step": 13865
    },
    {
      "epoch": 0.36518303924150647,
      "grad_norm": 2.4012434482574463,
      "learning_rate": 3.175269949960495e-05,
      "loss": 2.1721,
      "step": 13866
    },
    {
      "epoch": 0.36520937582301816,
      "grad_norm": 2.103909969329834,
      "learning_rate": 3.1751382670529367e-05,
      "loss": 0.6738,
      "step": 13867
    },
    {
      "epoch": 0.3652357124045299,
      "grad_norm": 2.1778693199157715,
      "learning_rate": 3.175006584145378e-05,
      "loss": 1.7869,
      "step": 13868
    },
    {
      "epoch": 0.3652620489860416,
      "grad_norm": 2.101128578186035,
      "learning_rate": 3.174874901237819e-05,
      "loss": 1.5545,
      "step": 13869
    },
    {
      "epoch": 0.36528838556755333,
      "grad_norm": 1.7771625518798828,
      "learning_rate": 3.1747432183302613e-05,
      "loss": 2.0729,
      "step": 13870
    },
    {
      "epoch": 0.3653147221490651,
      "grad_norm": 2.4678893089294434,
      "learning_rate": 3.174611535422702e-05,
      "loss": 1.8233,
      "step": 13871
    },
    {
      "epoch": 0.36534105873057676,
      "grad_norm": 3.338731288909912,
      "learning_rate": 3.174479852515144e-05,
      "loss": 1.7099,
      "step": 13872
    },
    {
      "epoch": 0.3653673953120885,
      "grad_norm": 4.617074489593506,
      "learning_rate": 3.174348169607585e-05,
      "loss": 1.5664,
      "step": 13873
    },
    {
      "epoch": 0.3653937318936002,
      "grad_norm": 1.9998185634613037,
      "learning_rate": 3.174216486700026e-05,
      "loss": 1.8124,
      "step": 13874
    },
    {
      "epoch": 0.36542006847511194,
      "grad_norm": 3.106323719024658,
      "learning_rate": 3.174084803792468e-05,
      "loss": 2.6014,
      "step": 13875
    },
    {
      "epoch": 0.3654464050566236,
      "grad_norm": 2.9282407760620117,
      "learning_rate": 3.1739531208849093e-05,
      "loss": 1.2614,
      "step": 13876
    },
    {
      "epoch": 0.36547274163813537,
      "grad_norm": 1.9555573463439941,
      "learning_rate": 3.173821437977351e-05,
      "loss": 0.6516,
      "step": 13877
    },
    {
      "epoch": 0.3654990782196471,
      "grad_norm": 2.4293127059936523,
      "learning_rate": 3.173689755069792e-05,
      "loss": 1.6645,
      "step": 13878
    },
    {
      "epoch": 0.3655254148011588,
      "grad_norm": 2.7621352672576904,
      "learning_rate": 3.1735580721622333e-05,
      "loss": 1.2575,
      "step": 13879
    },
    {
      "epoch": 0.36555175138267054,
      "grad_norm": 1.6219511032104492,
      "learning_rate": 3.173426389254675e-05,
      "loss": 1.6224,
      "step": 13880
    },
    {
      "epoch": 0.36557808796418223,
      "grad_norm": 3.178426742553711,
      "learning_rate": 3.1732947063471165e-05,
      "loss": 1.7361,
      "step": 13881
    },
    {
      "epoch": 0.365604424545694,
      "grad_norm": 2.1224637031555176,
      "learning_rate": 3.1731630234395573e-05,
      "loss": 1.6094,
      "step": 13882
    },
    {
      "epoch": 0.36563076112720566,
      "grad_norm": 2.019261598587036,
      "learning_rate": 3.173031340531999e-05,
      "loss": 1.9523,
      "step": 13883
    },
    {
      "epoch": 0.3656570977087174,
      "grad_norm": 1.9840797185897827,
      "learning_rate": 3.1728996576244405e-05,
      "loss": 1.7547,
      "step": 13884
    },
    {
      "epoch": 0.36568343429022915,
      "grad_norm": 4.115976810455322,
      "learning_rate": 3.172767974716882e-05,
      "loss": 1.8783,
      "step": 13885
    },
    {
      "epoch": 0.36570977087174084,
      "grad_norm": 7.090014934539795,
      "learning_rate": 3.1726362918093236e-05,
      "loss": 1.7113,
      "step": 13886
    },
    {
      "epoch": 0.3657361074532526,
      "grad_norm": 1.7239238023757935,
      "learning_rate": 3.1725046089017645e-05,
      "loss": 1.7662,
      "step": 13887
    },
    {
      "epoch": 0.36576244403476427,
      "grad_norm": 1.8261460065841675,
      "learning_rate": 3.172372925994206e-05,
      "loss": 1.9016,
      "step": 13888
    },
    {
      "epoch": 0.365788780616276,
      "grad_norm": 1.575415015220642,
      "learning_rate": 3.1722412430866476e-05,
      "loss": 1.8926,
      "step": 13889
    },
    {
      "epoch": 0.36581511719778775,
      "grad_norm": 1.7461912631988525,
      "learning_rate": 3.172109560179089e-05,
      "loss": 1.7464,
      "step": 13890
    },
    {
      "epoch": 0.36584145377929944,
      "grad_norm": 2.8639485836029053,
      "learning_rate": 3.17197787727153e-05,
      "loss": 0.9562,
      "step": 13891
    },
    {
      "epoch": 0.3658677903608112,
      "grad_norm": 2.717808723449707,
      "learning_rate": 3.1718461943639716e-05,
      "loss": 1.9312,
      "step": 13892
    },
    {
      "epoch": 0.3658941269423229,
      "grad_norm": 2.0594468116760254,
      "learning_rate": 3.171714511456413e-05,
      "loss": 1.3841,
      "step": 13893
    },
    {
      "epoch": 0.3659204635238346,
      "grad_norm": 2.767845869064331,
      "learning_rate": 3.171582828548855e-05,
      "loss": 2.4951,
      "step": 13894
    },
    {
      "epoch": 0.3659468001053463,
      "grad_norm": 3.5502774715423584,
      "learning_rate": 3.171451145641296e-05,
      "loss": 1.9461,
      "step": 13895
    },
    {
      "epoch": 0.36597313668685805,
      "grad_norm": 4.4372172355651855,
      "learning_rate": 3.171319462733737e-05,
      "loss": 0.8682,
      "step": 13896
    },
    {
      "epoch": 0.3659994732683698,
      "grad_norm": 1.8434040546417236,
      "learning_rate": 3.171187779826179e-05,
      "loss": 2.3746,
      "step": 13897
    },
    {
      "epoch": 0.3660258098498815,
      "grad_norm": 2.7245044708251953,
      "learning_rate": 3.17105609691862e-05,
      "loss": 1.8081,
      "step": 13898
    },
    {
      "epoch": 0.3660521464313932,
      "grad_norm": 2.771296739578247,
      "learning_rate": 3.170924414011062e-05,
      "loss": 2.1925,
      "step": 13899
    },
    {
      "epoch": 0.3660784830129049,
      "grad_norm": 6.998247146606445,
      "learning_rate": 3.170792731103503e-05,
      "loss": 0.7724,
      "step": 13900
    },
    {
      "epoch": 0.36610481959441665,
      "grad_norm": 3.2889821529388428,
      "learning_rate": 3.170661048195944e-05,
      "loss": 0.7153,
      "step": 13901
    },
    {
      "epoch": 0.36613115617592834,
      "grad_norm": 4.045352458953857,
      "learning_rate": 3.170529365288386e-05,
      "loss": 0.8026,
      "step": 13902
    },
    {
      "epoch": 0.3661574927574401,
      "grad_norm": 1.4328993558883667,
      "learning_rate": 3.1703976823808274e-05,
      "loss": 1.3598,
      "step": 13903
    },
    {
      "epoch": 0.3661838293389518,
      "grad_norm": 1.7620875835418701,
      "learning_rate": 3.170265999473269e-05,
      "loss": 2.3268,
      "step": 13904
    },
    {
      "epoch": 0.3662101659204635,
      "grad_norm": 1.6862494945526123,
      "learning_rate": 3.17013431656571e-05,
      "loss": 2.4352,
      "step": 13905
    },
    {
      "epoch": 0.36623650250197526,
      "grad_norm": 1.809762716293335,
      "learning_rate": 3.1700026336581514e-05,
      "loss": 1.7069,
      "step": 13906
    },
    {
      "epoch": 0.36626283908348695,
      "grad_norm": 5.644355297088623,
      "learning_rate": 3.169870950750592e-05,
      "loss": 1.3505,
      "step": 13907
    },
    {
      "epoch": 0.3662891756649987,
      "grad_norm": 2.797412157058716,
      "learning_rate": 3.1697392678430345e-05,
      "loss": 1.4805,
      "step": 13908
    },
    {
      "epoch": 0.3663155122465104,
      "grad_norm": 2.3762357234954834,
      "learning_rate": 3.1696075849354754e-05,
      "loss": 1.7948,
      "step": 13909
    },
    {
      "epoch": 0.3663418488280221,
      "grad_norm": 2.9792027473449707,
      "learning_rate": 3.169475902027917e-05,
      "loss": 1.215,
      "step": 13910
    },
    {
      "epoch": 0.36636818540953386,
      "grad_norm": 1.773647665977478,
      "learning_rate": 3.169344219120358e-05,
      "loss": 1.4595,
      "step": 13911
    },
    {
      "epoch": 0.36639452199104555,
      "grad_norm": 1.992507815361023,
      "learning_rate": 3.1692125362127994e-05,
      "loss": 1.9775,
      "step": 13912
    },
    {
      "epoch": 0.3664208585725573,
      "grad_norm": 1.9619947671890259,
      "learning_rate": 3.169080853305241e-05,
      "loss": 1.7646,
      "step": 13913
    },
    {
      "epoch": 0.366447195154069,
      "grad_norm": 2.018383741378784,
      "learning_rate": 3.1689491703976825e-05,
      "loss": 1.9394,
      "step": 13914
    },
    {
      "epoch": 0.3664735317355807,
      "grad_norm": 1.5302658081054688,
      "learning_rate": 3.168817487490124e-05,
      "loss": 2.0235,
      "step": 13915
    },
    {
      "epoch": 0.3664998683170924,
      "grad_norm": 2.3637197017669678,
      "learning_rate": 3.168685804582565e-05,
      "loss": 2.003,
      "step": 13916
    },
    {
      "epoch": 0.36652620489860416,
      "grad_norm": 1.772335171699524,
      "learning_rate": 3.168554121675007e-05,
      "loss": 1.6057,
      "step": 13917
    },
    {
      "epoch": 0.3665525414801159,
      "grad_norm": 2.13639235496521,
      "learning_rate": 3.168422438767448e-05,
      "loss": 1.4412,
      "step": 13918
    },
    {
      "epoch": 0.3665788780616276,
      "grad_norm": 2.634377956390381,
      "learning_rate": 3.1682907558598896e-05,
      "loss": 1.2497,
      "step": 13919
    },
    {
      "epoch": 0.36660521464313933,
      "grad_norm": 2.1368846893310547,
      "learning_rate": 3.1681590729523305e-05,
      "loss": 1.9891,
      "step": 13920
    },
    {
      "epoch": 0.366631551224651,
      "grad_norm": 3.9307925701141357,
      "learning_rate": 3.168027390044772e-05,
      "loss": 1.1267,
      "step": 13921
    },
    {
      "epoch": 0.36665788780616276,
      "grad_norm": 1.6822090148925781,
      "learning_rate": 3.1678957071372136e-05,
      "loss": 1.7764,
      "step": 13922
    },
    {
      "epoch": 0.3666842243876745,
      "grad_norm": 1.7323521375656128,
      "learning_rate": 3.167764024229655e-05,
      "loss": 1.0156,
      "step": 13923
    },
    {
      "epoch": 0.3667105609691862,
      "grad_norm": 2.6515190601348877,
      "learning_rate": 3.167632341322097e-05,
      "loss": 1.5725,
      "step": 13924
    },
    {
      "epoch": 0.36673689755069794,
      "grad_norm": 1.7710896730422974,
      "learning_rate": 3.1675006584145376e-05,
      "loss": 2.4908,
      "step": 13925
    },
    {
      "epoch": 0.3667632341322096,
      "grad_norm": 2.7734415531158447,
      "learning_rate": 3.167368975506979e-05,
      "loss": 2.0622,
      "step": 13926
    },
    {
      "epoch": 0.36678957071372137,
      "grad_norm": 1.8482822179794312,
      "learning_rate": 3.167237292599421e-05,
      "loss": 1.7745,
      "step": 13927
    },
    {
      "epoch": 0.36681590729523306,
      "grad_norm": 1.8044371604919434,
      "learning_rate": 3.167105609691862e-05,
      "loss": 1.8633,
      "step": 13928
    },
    {
      "epoch": 0.3668422438767448,
      "grad_norm": 2.230562686920166,
      "learning_rate": 3.166973926784303e-05,
      "loss": 1.9962,
      "step": 13929
    },
    {
      "epoch": 0.36686858045825654,
      "grad_norm": 2.4081809520721436,
      "learning_rate": 3.166842243876745e-05,
      "loss": 2.0626,
      "step": 13930
    },
    {
      "epoch": 0.36689491703976823,
      "grad_norm": 1.549582600593567,
      "learning_rate": 3.166710560969186e-05,
      "loss": 1.8033,
      "step": 13931
    },
    {
      "epoch": 0.36692125362128,
      "grad_norm": 2.1413769721984863,
      "learning_rate": 3.166578878061628e-05,
      "loss": 1.5805,
      "step": 13932
    },
    {
      "epoch": 0.36694759020279166,
      "grad_norm": 1.5038267374038696,
      "learning_rate": 3.1664471951540694e-05,
      "loss": 2.4161,
      "step": 13933
    },
    {
      "epoch": 0.3669739267843034,
      "grad_norm": 1.8666999340057373,
      "learning_rate": 3.16631551224651e-05,
      "loss": 2.2109,
      "step": 13934
    },
    {
      "epoch": 0.3670002633658151,
      "grad_norm": 3.083068370819092,
      "learning_rate": 3.166183829338952e-05,
      "loss": 2.0459,
      "step": 13935
    },
    {
      "epoch": 0.36702659994732684,
      "grad_norm": 2.6817305088043213,
      "learning_rate": 3.1660521464313934e-05,
      "loss": 1.7322,
      "step": 13936
    },
    {
      "epoch": 0.3670529365288386,
      "grad_norm": 2.3508315086364746,
      "learning_rate": 3.165920463523835e-05,
      "loss": 1.6017,
      "step": 13937
    },
    {
      "epoch": 0.36707927311035027,
      "grad_norm": 3.2340803146362305,
      "learning_rate": 3.165788780616276e-05,
      "loss": 1.3548,
      "step": 13938
    },
    {
      "epoch": 0.367105609691862,
      "grad_norm": 1.6249994039535522,
      "learning_rate": 3.1656570977087174e-05,
      "loss": 0.6447,
      "step": 13939
    },
    {
      "epoch": 0.3671319462733737,
      "grad_norm": 1.6927450895309448,
      "learning_rate": 3.165525414801159e-05,
      "loss": 1.7647,
      "step": 13940
    },
    {
      "epoch": 0.36715828285488544,
      "grad_norm": 1.6416312456130981,
      "learning_rate": 3.1653937318936006e-05,
      "loss": 1.3798,
      "step": 13941
    },
    {
      "epoch": 0.36718461943639713,
      "grad_norm": 4.144277572631836,
      "learning_rate": 3.165262048986042e-05,
      "loss": 1.6596,
      "step": 13942
    },
    {
      "epoch": 0.3672109560179089,
      "grad_norm": 2.075324535369873,
      "learning_rate": 3.165130366078483e-05,
      "loss": 1.4787,
      "step": 13943
    },
    {
      "epoch": 0.3672372925994206,
      "grad_norm": 1.62732994556427,
      "learning_rate": 3.1649986831709246e-05,
      "loss": 1.5477,
      "step": 13944
    },
    {
      "epoch": 0.3672636291809323,
      "grad_norm": 3.860360860824585,
      "learning_rate": 3.164867000263366e-05,
      "loss": 1.0535,
      "step": 13945
    },
    {
      "epoch": 0.36728996576244405,
      "grad_norm": 1.8674042224884033,
      "learning_rate": 3.164735317355808e-05,
      "loss": 1.7931,
      "step": 13946
    },
    {
      "epoch": 0.36731630234395574,
      "grad_norm": 2.8013572692871094,
      "learning_rate": 3.1646036344482486e-05,
      "loss": 1.2016,
      "step": 13947
    },
    {
      "epoch": 0.3673426389254675,
      "grad_norm": 3.1399917602539062,
      "learning_rate": 3.16447195154069e-05,
      "loss": 0.5751,
      "step": 13948
    },
    {
      "epoch": 0.36736897550697917,
      "grad_norm": 2.061265230178833,
      "learning_rate": 3.164340268633132e-05,
      "loss": 2.5703,
      "step": 13949
    },
    {
      "epoch": 0.3673953120884909,
      "grad_norm": 2.405780792236328,
      "learning_rate": 3.164208585725573e-05,
      "loss": 1.323,
      "step": 13950
    },
    {
      "epoch": 0.36742164867000265,
      "grad_norm": 2.8864221572875977,
      "learning_rate": 3.164076902818015e-05,
      "loss": 0.8512,
      "step": 13951
    },
    {
      "epoch": 0.36744798525151434,
      "grad_norm": 1.5359828472137451,
      "learning_rate": 3.163945219910456e-05,
      "loss": 1.984,
      "step": 13952
    },
    {
      "epoch": 0.3674743218330261,
      "grad_norm": 3.3540399074554443,
      "learning_rate": 3.163813537002897e-05,
      "loss": 1.1482,
      "step": 13953
    },
    {
      "epoch": 0.3675006584145378,
      "grad_norm": 1.594154953956604,
      "learning_rate": 3.163681854095338e-05,
      "loss": 2.3009,
      "step": 13954
    },
    {
      "epoch": 0.3675269949960495,
      "grad_norm": 1.3619701862335205,
      "learning_rate": 3.1635501711877804e-05,
      "loss": 0.5792,
      "step": 13955
    },
    {
      "epoch": 0.36755333157756126,
      "grad_norm": 2.1329445838928223,
      "learning_rate": 3.163418488280221e-05,
      "loss": 2.2423,
      "step": 13956
    },
    {
      "epoch": 0.36757966815907295,
      "grad_norm": 1.7652164697647095,
      "learning_rate": 3.163286805372663e-05,
      "loss": 1.7559,
      "step": 13957
    },
    {
      "epoch": 0.3676060047405847,
      "grad_norm": 2.0448415279388428,
      "learning_rate": 3.1631551224651044e-05,
      "loss": 2.2698,
      "step": 13958
    },
    {
      "epoch": 0.3676323413220964,
      "grad_norm": 2.5216546058654785,
      "learning_rate": 3.163023439557545e-05,
      "loss": 1.8614,
      "step": 13959
    },
    {
      "epoch": 0.3676586779036081,
      "grad_norm": 1.7806518077850342,
      "learning_rate": 3.1628917566499875e-05,
      "loss": 1.4737,
      "step": 13960
    },
    {
      "epoch": 0.3676850144851198,
      "grad_norm": 2.049075126647949,
      "learning_rate": 3.1627600737424284e-05,
      "loss": 1.6401,
      "step": 13961
    },
    {
      "epoch": 0.36771135106663155,
      "grad_norm": 2.9491817951202393,
      "learning_rate": 3.16262839083487e-05,
      "loss": 0.9926,
      "step": 13962
    },
    {
      "epoch": 0.3677376876481433,
      "grad_norm": 1.6878855228424072,
      "learning_rate": 3.162496707927311e-05,
      "loss": 1.8869,
      "step": 13963
    },
    {
      "epoch": 0.367764024229655,
      "grad_norm": 1.6858166456222534,
      "learning_rate": 3.162365025019753e-05,
      "loss": 1.8512,
      "step": 13964
    },
    {
      "epoch": 0.3677903608111667,
      "grad_norm": 5.557626247406006,
      "learning_rate": 3.162233342112194e-05,
      "loss": 1.187,
      "step": 13965
    },
    {
      "epoch": 0.3678166973926784,
      "grad_norm": 1.8709737062454224,
      "learning_rate": 3.1621016592046355e-05,
      "loss": 2.032,
      "step": 13966
    },
    {
      "epoch": 0.36784303397419016,
      "grad_norm": 2.255427837371826,
      "learning_rate": 3.1619699762970764e-05,
      "loss": 1.6211,
      "step": 13967
    },
    {
      "epoch": 0.36786937055570185,
      "grad_norm": 3.316115140914917,
      "learning_rate": 3.161838293389518e-05,
      "loss": 1.7174,
      "step": 13968
    },
    {
      "epoch": 0.3678957071372136,
      "grad_norm": 3.3950183391571045,
      "learning_rate": 3.1617066104819595e-05,
      "loss": 1.3451,
      "step": 13969
    },
    {
      "epoch": 0.36792204371872533,
      "grad_norm": 1.9111573696136475,
      "learning_rate": 3.161574927574401e-05,
      "loss": 1.8871,
      "step": 13970
    },
    {
      "epoch": 0.367948380300237,
      "grad_norm": 2.1103687286376953,
      "learning_rate": 3.1614432446668426e-05,
      "loss": 2.2399,
      "step": 13971
    },
    {
      "epoch": 0.36797471688174876,
      "grad_norm": 4.592798233032227,
      "learning_rate": 3.1613115617592835e-05,
      "loss": 2.2421,
      "step": 13972
    },
    {
      "epoch": 0.36800105346326045,
      "grad_norm": 1.7692029476165771,
      "learning_rate": 3.161179878851725e-05,
      "loss": 1.669,
      "step": 13973
    },
    {
      "epoch": 0.3680273900447722,
      "grad_norm": 1.8893780708312988,
      "learning_rate": 3.1610481959441666e-05,
      "loss": 2.2138,
      "step": 13974
    },
    {
      "epoch": 0.3680537266262839,
      "grad_norm": 1.958081841468811,
      "learning_rate": 3.160916513036608e-05,
      "loss": 1.8246,
      "step": 13975
    },
    {
      "epoch": 0.3680800632077956,
      "grad_norm": 3.833266019821167,
      "learning_rate": 3.160784830129049e-05,
      "loss": 2.0845,
      "step": 13976
    },
    {
      "epoch": 0.36810639978930737,
      "grad_norm": 2.422652006149292,
      "learning_rate": 3.1606531472214906e-05,
      "loss": 0.7655,
      "step": 13977
    },
    {
      "epoch": 0.36813273637081906,
      "grad_norm": 2.3722052574157715,
      "learning_rate": 3.160521464313932e-05,
      "loss": 2.0574,
      "step": 13978
    },
    {
      "epoch": 0.3681590729523308,
      "grad_norm": 2.0069072246551514,
      "learning_rate": 3.160389781406374e-05,
      "loss": 1.3412,
      "step": 13979
    },
    {
      "epoch": 0.3681854095338425,
      "grad_norm": 2.0078208446502686,
      "learning_rate": 3.160258098498815e-05,
      "loss": 2.1492,
      "step": 13980
    },
    {
      "epoch": 0.36821174611535423,
      "grad_norm": 2.508742094039917,
      "learning_rate": 3.160126415591256e-05,
      "loss": 1.8643,
      "step": 13981
    },
    {
      "epoch": 0.3682380826968659,
      "grad_norm": 2.8134686946868896,
      "learning_rate": 3.159994732683698e-05,
      "loss": 0.5469,
      "step": 13982
    },
    {
      "epoch": 0.36826441927837766,
      "grad_norm": 3.2314374446868896,
      "learning_rate": 3.159863049776139e-05,
      "loss": 0.4694,
      "step": 13983
    },
    {
      "epoch": 0.3682907558598894,
      "grad_norm": 5.888422966003418,
      "learning_rate": 3.159731366868581e-05,
      "loss": 1.07,
      "step": 13984
    },
    {
      "epoch": 0.3683170924414011,
      "grad_norm": 2.3816802501678467,
      "learning_rate": 3.159599683961022e-05,
      "loss": 1.8833,
      "step": 13985
    },
    {
      "epoch": 0.36834342902291284,
      "grad_norm": 5.047017574310303,
      "learning_rate": 3.159468001053463e-05,
      "loss": 0.8706,
      "step": 13986
    },
    {
      "epoch": 0.3683697656044245,
      "grad_norm": 2.0239899158477783,
      "learning_rate": 3.159336318145905e-05,
      "loss": 2.0173,
      "step": 13987
    },
    {
      "epoch": 0.36839610218593627,
      "grad_norm": 3.18420672416687,
      "learning_rate": 3.1592046352383464e-05,
      "loss": 1.8623,
      "step": 13988
    },
    {
      "epoch": 0.368422438767448,
      "grad_norm": 2.366476535797119,
      "learning_rate": 3.159072952330788e-05,
      "loss": 1.5839,
      "step": 13989
    },
    {
      "epoch": 0.3684487753489597,
      "grad_norm": 2.115718364715576,
      "learning_rate": 3.158941269423229e-05,
      "loss": 1.6292,
      "step": 13990
    },
    {
      "epoch": 0.36847511193047144,
      "grad_norm": 2.5027735233306885,
      "learning_rate": 3.1588095865156704e-05,
      "loss": 2.0801,
      "step": 13991
    },
    {
      "epoch": 0.36850144851198313,
      "grad_norm": 1.529545783996582,
      "learning_rate": 3.158677903608111e-05,
      "loss": 1.6444,
      "step": 13992
    },
    {
      "epoch": 0.3685277850934949,
      "grad_norm": 10.095187187194824,
      "learning_rate": 3.1585462207005535e-05,
      "loss": 2.4624,
      "step": 13993
    },
    {
      "epoch": 0.36855412167500656,
      "grad_norm": 2.531205177307129,
      "learning_rate": 3.1584145377929944e-05,
      "loss": 1.7764,
      "step": 13994
    },
    {
      "epoch": 0.3685804582565183,
      "grad_norm": 1.7562317848205566,
      "learning_rate": 3.158282854885436e-05,
      "loss": 1.4229,
      "step": 13995
    },
    {
      "epoch": 0.36860679483803005,
      "grad_norm": 2.2997307777404785,
      "learning_rate": 3.1581511719778775e-05,
      "loss": 1.3205,
      "step": 13996
    },
    {
      "epoch": 0.36863313141954174,
      "grad_norm": 2.063079595565796,
      "learning_rate": 3.158019489070319e-05,
      "loss": 0.5504,
      "step": 13997
    },
    {
      "epoch": 0.3686594680010535,
      "grad_norm": 2.4551782608032227,
      "learning_rate": 3.157887806162761e-05,
      "loss": 2.3043,
      "step": 13998
    },
    {
      "epoch": 0.36868580458256517,
      "grad_norm": 3.2041988372802734,
      "learning_rate": 3.1577561232552015e-05,
      "loss": 1.3218,
      "step": 13999
    },
    {
      "epoch": 0.3687121411640769,
      "grad_norm": 2.6684558391571045,
      "learning_rate": 3.157624440347643e-05,
      "loss": 2.4471,
      "step": 14000
    },
    {
      "epoch": 0.3687384777455886,
      "grad_norm": 3.5551936626434326,
      "learning_rate": 3.157492757440084e-05,
      "loss": 1.2455,
      "step": 14001
    },
    {
      "epoch": 0.36876481432710034,
      "grad_norm": 1.9673200845718384,
      "learning_rate": 3.157361074532526e-05,
      "loss": 1.1001,
      "step": 14002
    },
    {
      "epoch": 0.3687911509086121,
      "grad_norm": 1.8047468662261963,
      "learning_rate": 3.157229391624967e-05,
      "loss": 2.0843,
      "step": 14003
    },
    {
      "epoch": 0.3688174874901238,
      "grad_norm": 2.695791006088257,
      "learning_rate": 3.157097708717409e-05,
      "loss": 1.5977,
      "step": 14004
    },
    {
      "epoch": 0.3688438240716355,
      "grad_norm": 3.1595258712768555,
      "learning_rate": 3.15696602580985e-05,
      "loss": 1.2335,
      "step": 14005
    },
    {
      "epoch": 0.3688701606531472,
      "grad_norm": 2.4013636112213135,
      "learning_rate": 3.156834342902291e-05,
      "loss": 1.2647,
      "step": 14006
    },
    {
      "epoch": 0.36889649723465895,
      "grad_norm": 1.7196820974349976,
      "learning_rate": 3.1567026599947333e-05,
      "loss": 0.7567,
      "step": 14007
    },
    {
      "epoch": 0.36892283381617064,
      "grad_norm": 2.1799941062927246,
      "learning_rate": 3.156570977087174e-05,
      "loss": 1.4471,
      "step": 14008
    },
    {
      "epoch": 0.3689491703976824,
      "grad_norm": 3.406611442565918,
      "learning_rate": 3.156439294179616e-05,
      "loss": 1.7441,
      "step": 14009
    },
    {
      "epoch": 0.3689755069791941,
      "grad_norm": 4.917551517486572,
      "learning_rate": 3.156307611272057e-05,
      "loss": 1.8835,
      "step": 14010
    },
    {
      "epoch": 0.3690018435607058,
      "grad_norm": 2.311849594116211,
      "learning_rate": 3.156175928364499e-05,
      "loss": 1.934,
      "step": 14011
    },
    {
      "epoch": 0.36902818014221755,
      "grad_norm": 1.8395415544509888,
      "learning_rate": 3.15604424545694e-05,
      "loss": 0.4708,
      "step": 14012
    },
    {
      "epoch": 0.36905451672372924,
      "grad_norm": 1.7007797956466675,
      "learning_rate": 3.1559125625493813e-05,
      "loss": 0.9208,
      "step": 14013
    },
    {
      "epoch": 0.369080853305241,
      "grad_norm": 2.0950088500976562,
      "learning_rate": 3.155780879641822e-05,
      "loss": 0.9732,
      "step": 14014
    },
    {
      "epoch": 0.3691071898867527,
      "grad_norm": 2.068774938583374,
      "learning_rate": 3.155649196734264e-05,
      "loss": 1.4884,
      "step": 14015
    },
    {
      "epoch": 0.3691335264682644,
      "grad_norm": 1.7766845226287842,
      "learning_rate": 3.1555175138267053e-05,
      "loss": 1.9346,
      "step": 14016
    },
    {
      "epoch": 0.36915986304977616,
      "grad_norm": 1.7502129077911377,
      "learning_rate": 3.155385830919147e-05,
      "loss": 1.8656,
      "step": 14017
    },
    {
      "epoch": 0.36918619963128785,
      "grad_norm": 1.852126955986023,
      "learning_rate": 3.1552541480115885e-05,
      "loss": 1.5744,
      "step": 14018
    },
    {
      "epoch": 0.3692125362127996,
      "grad_norm": 2.4089267253875732,
      "learning_rate": 3.1551224651040294e-05,
      "loss": 1.6667,
      "step": 14019
    },
    {
      "epoch": 0.3692388727943113,
      "grad_norm": 4.818479061126709,
      "learning_rate": 3.154990782196471e-05,
      "loss": 1.986,
      "step": 14020
    },
    {
      "epoch": 0.369265209375823,
      "grad_norm": 1.6262494325637817,
      "learning_rate": 3.1548590992889125e-05,
      "loss": 1.9149,
      "step": 14021
    },
    {
      "epoch": 0.3692915459573347,
      "grad_norm": 2.3452014923095703,
      "learning_rate": 3.154727416381354e-05,
      "loss": 2.1754,
      "step": 14022
    },
    {
      "epoch": 0.36931788253884645,
      "grad_norm": 2.2143008708953857,
      "learning_rate": 3.154595733473795e-05,
      "loss": 0.5868,
      "step": 14023
    },
    {
      "epoch": 0.3693442191203582,
      "grad_norm": 2.0937740802764893,
      "learning_rate": 3.1544640505662365e-05,
      "loss": 2.1145,
      "step": 14024
    },
    {
      "epoch": 0.3693705557018699,
      "grad_norm": 1.6205493211746216,
      "learning_rate": 3.154332367658678e-05,
      "loss": 1.9693,
      "step": 14025
    },
    {
      "epoch": 0.36939689228338163,
      "grad_norm": 4.79160213470459,
      "learning_rate": 3.1542006847511196e-05,
      "loss": 2.1585,
      "step": 14026
    },
    {
      "epoch": 0.3694232288648933,
      "grad_norm": 2.5807693004608154,
      "learning_rate": 3.154069001843561e-05,
      "loss": 1.9397,
      "step": 14027
    },
    {
      "epoch": 0.36944956544640506,
      "grad_norm": 1.9860213994979858,
      "learning_rate": 3.153937318936002e-05,
      "loss": 1.8714,
      "step": 14028
    },
    {
      "epoch": 0.3694759020279168,
      "grad_norm": 2.9638314247131348,
      "learning_rate": 3.1538056360284436e-05,
      "loss": 2.0282,
      "step": 14029
    },
    {
      "epoch": 0.3695022386094285,
      "grad_norm": 2.4194157123565674,
      "learning_rate": 3.153673953120885e-05,
      "loss": 1.5762,
      "step": 14030
    },
    {
      "epoch": 0.36952857519094023,
      "grad_norm": 2.640869140625,
      "learning_rate": 3.153542270213327e-05,
      "loss": 1.4509,
      "step": 14031
    },
    {
      "epoch": 0.3695549117724519,
      "grad_norm": 2.0341551303863525,
      "learning_rate": 3.1534105873057676e-05,
      "loss": 2.2935,
      "step": 14032
    },
    {
      "epoch": 0.36958124835396367,
      "grad_norm": 1.718632698059082,
      "learning_rate": 3.153278904398209e-05,
      "loss": 1.522,
      "step": 14033
    },
    {
      "epoch": 0.36960758493547535,
      "grad_norm": 1.8101212978363037,
      "learning_rate": 3.153147221490651e-05,
      "loss": 2.5663,
      "step": 14034
    },
    {
      "epoch": 0.3696339215169871,
      "grad_norm": 2.603178024291992,
      "learning_rate": 3.153015538583092e-05,
      "loss": 1.6246,
      "step": 14035
    },
    {
      "epoch": 0.36966025809849884,
      "grad_norm": 1.9641749858856201,
      "learning_rate": 3.152883855675534e-05,
      "loss": 1.8041,
      "step": 14036
    },
    {
      "epoch": 0.3696865946800105,
      "grad_norm": 1.7965713739395142,
      "learning_rate": 3.152752172767975e-05,
      "loss": 1.8808,
      "step": 14037
    },
    {
      "epoch": 0.36971293126152227,
      "grad_norm": 2.3773248195648193,
      "learning_rate": 3.152620489860416e-05,
      "loss": 1.9717,
      "step": 14038
    },
    {
      "epoch": 0.36973926784303396,
      "grad_norm": 2.828329086303711,
      "learning_rate": 3.152488806952857e-05,
      "loss": 1.5003,
      "step": 14039
    },
    {
      "epoch": 0.3697656044245457,
      "grad_norm": 2.3443846702575684,
      "learning_rate": 3.1523571240452994e-05,
      "loss": 2.7071,
      "step": 14040
    },
    {
      "epoch": 0.3697919410060574,
      "grad_norm": 1.8259594440460205,
      "learning_rate": 3.15222544113774e-05,
      "loss": 2.0529,
      "step": 14041
    },
    {
      "epoch": 0.36981827758756913,
      "grad_norm": 1.7357100248336792,
      "learning_rate": 3.152093758230182e-05,
      "loss": 1.8714,
      "step": 14042
    },
    {
      "epoch": 0.3698446141690809,
      "grad_norm": 2.4874861240386963,
      "learning_rate": 3.1519620753226234e-05,
      "loss": 2.1107,
      "step": 14043
    },
    {
      "epoch": 0.36987095075059256,
      "grad_norm": 2.044776439666748,
      "learning_rate": 3.151830392415065e-05,
      "loss": 1.4348,
      "step": 14044
    },
    {
      "epoch": 0.3698972873321043,
      "grad_norm": 2.6963422298431396,
      "learning_rate": 3.1516987095075065e-05,
      "loss": 1.9687,
      "step": 14045
    },
    {
      "epoch": 0.369923623913616,
      "grad_norm": 4.4860405921936035,
      "learning_rate": 3.1515670265999474e-05,
      "loss": 0.7362,
      "step": 14046
    },
    {
      "epoch": 0.36994996049512774,
      "grad_norm": 2.948392391204834,
      "learning_rate": 3.151435343692389e-05,
      "loss": 1.5328,
      "step": 14047
    },
    {
      "epoch": 0.3699762970766394,
      "grad_norm": 2.040557622909546,
      "learning_rate": 3.15130366078483e-05,
      "loss": 1.6299,
      "step": 14048
    },
    {
      "epoch": 0.37000263365815117,
      "grad_norm": 2.2876884937286377,
      "learning_rate": 3.151171977877272e-05,
      "loss": 1.6243,
      "step": 14049
    },
    {
      "epoch": 0.3700289702396629,
      "grad_norm": 4.5751471519470215,
      "learning_rate": 3.151040294969713e-05,
      "loss": 1.913,
      "step": 14050
    },
    {
      "epoch": 0.3700553068211746,
      "grad_norm": 1.823683261871338,
      "learning_rate": 3.1509086120621545e-05,
      "loss": 1.8026,
      "step": 14051
    },
    {
      "epoch": 0.37008164340268634,
      "grad_norm": 2.302485466003418,
      "learning_rate": 3.150776929154596e-05,
      "loss": 2.0231,
      "step": 14052
    },
    {
      "epoch": 0.37010797998419803,
      "grad_norm": 2.1640784740448,
      "learning_rate": 3.150645246247037e-05,
      "loss": 1.938,
      "step": 14053
    },
    {
      "epoch": 0.3701343165657098,
      "grad_norm": 2.714353561401367,
      "learning_rate": 3.150513563339479e-05,
      "loss": 0.9813,
      "step": 14054
    },
    {
      "epoch": 0.37016065314722146,
      "grad_norm": 2.1586687564849854,
      "learning_rate": 3.15038188043192e-05,
      "loss": 2.364,
      "step": 14055
    },
    {
      "epoch": 0.3701869897287332,
      "grad_norm": 2.4510347843170166,
      "learning_rate": 3.1502501975243616e-05,
      "loss": 0.5145,
      "step": 14056
    },
    {
      "epoch": 0.37021332631024495,
      "grad_norm": 1.7453266382217407,
      "learning_rate": 3.1501185146168025e-05,
      "loss": 1.3365,
      "step": 14057
    },
    {
      "epoch": 0.37023966289175664,
      "grad_norm": 2.4171650409698486,
      "learning_rate": 3.149986831709244e-05,
      "loss": 1.6939,
      "step": 14058
    },
    {
      "epoch": 0.3702659994732684,
      "grad_norm": 1.585357666015625,
      "learning_rate": 3.1498551488016856e-05,
      "loss": 1.4889,
      "step": 14059
    },
    {
      "epoch": 0.37029233605478007,
      "grad_norm": 1.7167292833328247,
      "learning_rate": 3.149723465894127e-05,
      "loss": 1.9869,
      "step": 14060
    },
    {
      "epoch": 0.3703186726362918,
      "grad_norm": 1.7951264381408691,
      "learning_rate": 3.149591782986569e-05,
      "loss": 1.6581,
      "step": 14061
    },
    {
      "epoch": 0.37034500921780356,
      "grad_norm": 4.58056640625,
      "learning_rate": 3.1494601000790096e-05,
      "loss": 1.4034,
      "step": 14062
    },
    {
      "epoch": 0.37037134579931524,
      "grad_norm": 2.6444268226623535,
      "learning_rate": 3.149328417171452e-05,
      "loss": 1.0226,
      "step": 14063
    },
    {
      "epoch": 0.370397682380827,
      "grad_norm": 2.370619773864746,
      "learning_rate": 3.149196734263893e-05,
      "loss": 1.6543,
      "step": 14064
    },
    {
      "epoch": 0.3704240189623387,
      "grad_norm": 1.7518035173416138,
      "learning_rate": 3.149065051356334e-05,
      "loss": 1.5299,
      "step": 14065
    },
    {
      "epoch": 0.3704503555438504,
      "grad_norm": 2.046924114227295,
      "learning_rate": 3.148933368448775e-05,
      "loss": 2.1612,
      "step": 14066
    },
    {
      "epoch": 0.3704766921253621,
      "grad_norm": 3.127081871032715,
      "learning_rate": 3.148801685541217e-05,
      "loss": 2.0245,
      "step": 14067
    },
    {
      "epoch": 0.37050302870687385,
      "grad_norm": 3.3367018699645996,
      "learning_rate": 3.148670002633658e-05,
      "loss": 1.4806,
      "step": 14068
    },
    {
      "epoch": 0.3705293652883856,
      "grad_norm": 3.3937251567840576,
      "learning_rate": 3.1485383197261e-05,
      "loss": 1.8675,
      "step": 14069
    },
    {
      "epoch": 0.3705557018698973,
      "grad_norm": 2.0157434940338135,
      "learning_rate": 3.148406636818541e-05,
      "loss": 2.1464,
      "step": 14070
    },
    {
      "epoch": 0.370582038451409,
      "grad_norm": 1.835428237915039,
      "learning_rate": 3.148274953910982e-05,
      "loss": 1.8917,
      "step": 14071
    },
    {
      "epoch": 0.3706083750329207,
      "grad_norm": 2.2136576175689697,
      "learning_rate": 3.148143271003424e-05,
      "loss": 2.4428,
      "step": 14072
    },
    {
      "epoch": 0.37063471161443245,
      "grad_norm": 2.241074800491333,
      "learning_rate": 3.1480115880958654e-05,
      "loss": 0.9379,
      "step": 14073
    },
    {
      "epoch": 0.37066104819594414,
      "grad_norm": 2.1240148544311523,
      "learning_rate": 3.147879905188307e-05,
      "loss": 1.8163,
      "step": 14074
    },
    {
      "epoch": 0.3706873847774559,
      "grad_norm": 2.615171194076538,
      "learning_rate": 3.147748222280748e-05,
      "loss": 1.5766,
      "step": 14075
    },
    {
      "epoch": 0.37071372135896763,
      "grad_norm": 2.40596866607666,
      "learning_rate": 3.1476165393731894e-05,
      "loss": 1.6367,
      "step": 14076
    },
    {
      "epoch": 0.3707400579404793,
      "grad_norm": 1.585900068283081,
      "learning_rate": 3.147484856465631e-05,
      "loss": 1.8951,
      "step": 14077
    },
    {
      "epoch": 0.37076639452199106,
      "grad_norm": 1.8931550979614258,
      "learning_rate": 3.1473531735580726e-05,
      "loss": 2.3826,
      "step": 14078
    },
    {
      "epoch": 0.37079273110350275,
      "grad_norm": 1.6468175649642944,
      "learning_rate": 3.1472214906505135e-05,
      "loss": 2.4275,
      "step": 14079
    },
    {
      "epoch": 0.3708190676850145,
      "grad_norm": 2.1604626178741455,
      "learning_rate": 3.147089807742955e-05,
      "loss": 2.1381,
      "step": 14080
    },
    {
      "epoch": 0.3708454042665262,
      "grad_norm": 2.925337553024292,
      "learning_rate": 3.1469581248353966e-05,
      "loss": 1.1897,
      "step": 14081
    },
    {
      "epoch": 0.3708717408480379,
      "grad_norm": 4.752488136291504,
      "learning_rate": 3.146826441927838e-05,
      "loss": 1.2945,
      "step": 14082
    },
    {
      "epoch": 0.37089807742954967,
      "grad_norm": 1.8100332021713257,
      "learning_rate": 3.14669475902028e-05,
      "loss": 1.8183,
      "step": 14083
    },
    {
      "epoch": 0.37092441401106135,
      "grad_norm": 1.6442036628723145,
      "learning_rate": 3.1465630761127206e-05,
      "loss": 1.7176,
      "step": 14084
    },
    {
      "epoch": 0.3709507505925731,
      "grad_norm": 2.836866617202759,
      "learning_rate": 3.146431393205162e-05,
      "loss": 0.6367,
      "step": 14085
    },
    {
      "epoch": 0.3709770871740848,
      "grad_norm": 2.276623249053955,
      "learning_rate": 3.146299710297603e-05,
      "loss": 2.0926,
      "step": 14086
    },
    {
      "epoch": 0.37100342375559653,
      "grad_norm": 4.055276870727539,
      "learning_rate": 3.146168027390045e-05,
      "loss": 1.5615,
      "step": 14087
    },
    {
      "epoch": 0.3710297603371082,
      "grad_norm": 5.498299598693848,
      "learning_rate": 3.146036344482486e-05,
      "loss": 0.8085,
      "step": 14088
    },
    {
      "epoch": 0.37105609691861996,
      "grad_norm": 4.85291862487793,
      "learning_rate": 3.145904661574928e-05,
      "loss": 0.8505,
      "step": 14089
    },
    {
      "epoch": 0.3710824335001317,
      "grad_norm": 4.257732391357422,
      "learning_rate": 3.145772978667369e-05,
      "loss": 1.4856,
      "step": 14090
    },
    {
      "epoch": 0.3711087700816434,
      "grad_norm": 3.3910834789276123,
      "learning_rate": 3.14564129575981e-05,
      "loss": 0.9633,
      "step": 14091
    },
    {
      "epoch": 0.37113510666315513,
      "grad_norm": 3.682894468307495,
      "learning_rate": 3.1455096128522524e-05,
      "loss": 1.108,
      "step": 14092
    },
    {
      "epoch": 0.3711614432446668,
      "grad_norm": 1.7492684125900269,
      "learning_rate": 3.145377929944693e-05,
      "loss": 1.7462,
      "step": 14093
    },
    {
      "epoch": 0.37118777982617857,
      "grad_norm": 4.625382900238037,
      "learning_rate": 3.145246247037135e-05,
      "loss": 1.1887,
      "step": 14094
    },
    {
      "epoch": 0.3712141164076903,
      "grad_norm": 3.399789333343506,
      "learning_rate": 3.145114564129576e-05,
      "loss": 1.3714,
      "step": 14095
    },
    {
      "epoch": 0.371240452989202,
      "grad_norm": 1.87528395652771,
      "learning_rate": 3.144982881222018e-05,
      "loss": 2.0846,
      "step": 14096
    },
    {
      "epoch": 0.37126678957071374,
      "grad_norm": 2.0241079330444336,
      "learning_rate": 3.144851198314459e-05,
      "loss": 1.4942,
      "step": 14097
    },
    {
      "epoch": 0.37129312615222543,
      "grad_norm": 4.415994167327881,
      "learning_rate": 3.1447195154069004e-05,
      "loss": 0.8733,
      "step": 14098
    },
    {
      "epoch": 0.37131946273373717,
      "grad_norm": 3.184434413909912,
      "learning_rate": 3.144587832499342e-05,
      "loss": 1.9379,
      "step": 14099
    },
    {
      "epoch": 0.37134579931524886,
      "grad_norm": 2.7417070865631104,
      "learning_rate": 3.144456149591783e-05,
      "loss": 0.6104,
      "step": 14100
    },
    {
      "epoch": 0.3713721358967606,
      "grad_norm": 2.1382856369018555,
      "learning_rate": 3.144324466684225e-05,
      "loss": 1.6291,
      "step": 14101
    },
    {
      "epoch": 0.37139847247827235,
      "grad_norm": 2.849663257598877,
      "learning_rate": 3.144192783776666e-05,
      "loss": 2.1041,
      "step": 14102
    },
    {
      "epoch": 0.37142480905978403,
      "grad_norm": 1.925470232963562,
      "learning_rate": 3.1440611008691075e-05,
      "loss": 1.441,
      "step": 14103
    },
    {
      "epoch": 0.3714511456412958,
      "grad_norm": 1.9539369344711304,
      "learning_rate": 3.1439294179615484e-05,
      "loss": 1.4344,
      "step": 14104
    },
    {
      "epoch": 0.37147748222280746,
      "grad_norm": 1.9117008447647095,
      "learning_rate": 3.14379773505399e-05,
      "loss": 1.6653,
      "step": 14105
    },
    {
      "epoch": 0.3715038188043192,
      "grad_norm": 2.064760446548462,
      "learning_rate": 3.1436660521464315e-05,
      "loss": 1.9968,
      "step": 14106
    },
    {
      "epoch": 0.3715301553858309,
      "grad_norm": 2.7670607566833496,
      "learning_rate": 3.143534369238873e-05,
      "loss": 1.5052,
      "step": 14107
    },
    {
      "epoch": 0.37155649196734264,
      "grad_norm": 1.9472836256027222,
      "learning_rate": 3.1434026863313146e-05,
      "loss": 0.2269,
      "step": 14108
    },
    {
      "epoch": 0.3715828285488544,
      "grad_norm": 4.877762794494629,
      "learning_rate": 3.1432710034237555e-05,
      "loss": 1.0154,
      "step": 14109
    },
    {
      "epoch": 0.37160916513036607,
      "grad_norm": 6.371516227722168,
      "learning_rate": 3.143139320516198e-05,
      "loss": 0.9444,
      "step": 14110
    },
    {
      "epoch": 0.3716355017118778,
      "grad_norm": 1.8459303379058838,
      "learning_rate": 3.1430076376086386e-05,
      "loss": 1.5265,
      "step": 14111
    },
    {
      "epoch": 0.3716618382933895,
      "grad_norm": 2.2648868560791016,
      "learning_rate": 3.14287595470108e-05,
      "loss": 1.8607,
      "step": 14112
    },
    {
      "epoch": 0.37168817487490124,
      "grad_norm": 2.055403232574463,
      "learning_rate": 3.142744271793521e-05,
      "loss": 2.0137,
      "step": 14113
    },
    {
      "epoch": 0.37171451145641293,
      "grad_norm": 1.556557297706604,
      "learning_rate": 3.1426125888859626e-05,
      "loss": 1.997,
      "step": 14114
    },
    {
      "epoch": 0.3717408480379247,
      "grad_norm": 1.945652723312378,
      "learning_rate": 3.142480905978404e-05,
      "loss": 1.4773,
      "step": 14115
    },
    {
      "epoch": 0.3717671846194364,
      "grad_norm": 2.3292694091796875,
      "learning_rate": 3.142349223070846e-05,
      "loss": 1.7397,
      "step": 14116
    },
    {
      "epoch": 0.3717935212009481,
      "grad_norm": 2.0406198501586914,
      "learning_rate": 3.1422175401632866e-05,
      "loss": 1.3014,
      "step": 14117
    },
    {
      "epoch": 0.37181985778245985,
      "grad_norm": 2.925997734069824,
      "learning_rate": 3.142085857255728e-05,
      "loss": 1.5552,
      "step": 14118
    },
    {
      "epoch": 0.37184619436397154,
      "grad_norm": 1.776259422302246,
      "learning_rate": 3.14195417434817e-05,
      "loss": 1.7039,
      "step": 14119
    },
    {
      "epoch": 0.3718725309454833,
      "grad_norm": 1.7593868970870972,
      "learning_rate": 3.141822491440611e-05,
      "loss": 1.7532,
      "step": 14120
    },
    {
      "epoch": 0.37189886752699497,
      "grad_norm": 2.4789812564849854,
      "learning_rate": 3.141690808533053e-05,
      "loss": 0.8185,
      "step": 14121
    },
    {
      "epoch": 0.3719252041085067,
      "grad_norm": 1.977687954902649,
      "learning_rate": 3.141559125625494e-05,
      "loss": 1.1145,
      "step": 14122
    },
    {
      "epoch": 0.37195154069001846,
      "grad_norm": 2.5104732513427734,
      "learning_rate": 3.141427442717935e-05,
      "loss": 2.3535,
      "step": 14123
    },
    {
      "epoch": 0.37197787727153014,
      "grad_norm": 1.8159688711166382,
      "learning_rate": 3.141295759810376e-05,
      "loss": 1.5078,
      "step": 14124
    },
    {
      "epoch": 0.3720042138530419,
      "grad_norm": 7.478771686553955,
      "learning_rate": 3.1411640769028184e-05,
      "loss": 1.4843,
      "step": 14125
    },
    {
      "epoch": 0.3720305504345536,
      "grad_norm": 2.509643793106079,
      "learning_rate": 3.141032393995259e-05,
      "loss": 1.6703,
      "step": 14126
    },
    {
      "epoch": 0.3720568870160653,
      "grad_norm": 3.9190351963043213,
      "learning_rate": 3.140900711087701e-05,
      "loss": 1.6153,
      "step": 14127
    },
    {
      "epoch": 0.37208322359757706,
      "grad_norm": 2.1300253868103027,
      "learning_rate": 3.1407690281801424e-05,
      "loss": 2.2397,
      "step": 14128
    },
    {
      "epoch": 0.37210956017908875,
      "grad_norm": 1.976865530014038,
      "learning_rate": 3.140637345272584e-05,
      "loss": 1.4254,
      "step": 14129
    },
    {
      "epoch": 0.3721358967606005,
      "grad_norm": 3.5399374961853027,
      "learning_rate": 3.1405056623650255e-05,
      "loss": 1.1319,
      "step": 14130
    },
    {
      "epoch": 0.3721622333421122,
      "grad_norm": 1.8781288862228394,
      "learning_rate": 3.1403739794574664e-05,
      "loss": 2.0124,
      "step": 14131
    },
    {
      "epoch": 0.3721885699236239,
      "grad_norm": 3.8063457012176514,
      "learning_rate": 3.140242296549908e-05,
      "loss": 1.9692,
      "step": 14132
    },
    {
      "epoch": 0.3722149065051356,
      "grad_norm": 1.6347147226333618,
      "learning_rate": 3.140110613642349e-05,
      "loss": 2.3463,
      "step": 14133
    },
    {
      "epoch": 0.37224124308664736,
      "grad_norm": 2.5241732597351074,
      "learning_rate": 3.139978930734791e-05,
      "loss": 2.0984,
      "step": 14134
    },
    {
      "epoch": 0.3722675796681591,
      "grad_norm": 2.893841505050659,
      "learning_rate": 3.139847247827232e-05,
      "loss": 2.226,
      "step": 14135
    },
    {
      "epoch": 0.3722939162496708,
      "grad_norm": 1.7463747262954712,
      "learning_rate": 3.1397155649196735e-05,
      "loss": 2.2076,
      "step": 14136
    },
    {
      "epoch": 0.37232025283118253,
      "grad_norm": 1.6108602285385132,
      "learning_rate": 3.139583882012115e-05,
      "loss": 1.8188,
      "step": 14137
    },
    {
      "epoch": 0.3723465894126942,
      "grad_norm": 2.745497465133667,
      "learning_rate": 3.139452199104556e-05,
      "loss": 2.2784,
      "step": 14138
    },
    {
      "epoch": 0.37237292599420596,
      "grad_norm": 2.3145902156829834,
      "learning_rate": 3.139320516196998e-05,
      "loss": 1.0686,
      "step": 14139
    },
    {
      "epoch": 0.37239926257571765,
      "grad_norm": 1.4394451379776,
      "learning_rate": 3.139188833289439e-05,
      "loss": 0.5054,
      "step": 14140
    },
    {
      "epoch": 0.3724255991572294,
      "grad_norm": 1.9597985744476318,
      "learning_rate": 3.139057150381881e-05,
      "loss": 1.9649,
      "step": 14141
    },
    {
      "epoch": 0.37245193573874114,
      "grad_norm": 5.424899578094482,
      "learning_rate": 3.1389254674743216e-05,
      "loss": 0.8165,
      "step": 14142
    },
    {
      "epoch": 0.3724782723202528,
      "grad_norm": 1.6741294860839844,
      "learning_rate": 3.138793784566764e-05,
      "loss": 2.3714,
      "step": 14143
    },
    {
      "epoch": 0.37250460890176457,
      "grad_norm": 2.9758431911468506,
      "learning_rate": 3.138662101659205e-05,
      "loss": 1.5172,
      "step": 14144
    },
    {
      "epoch": 0.37253094548327625,
      "grad_norm": 2.6219727993011475,
      "learning_rate": 3.138530418751646e-05,
      "loss": 2.1054,
      "step": 14145
    },
    {
      "epoch": 0.372557282064788,
      "grad_norm": 2.9232587814331055,
      "learning_rate": 3.138398735844088e-05,
      "loss": 1.5619,
      "step": 14146
    },
    {
      "epoch": 0.3725836186462997,
      "grad_norm": 3.4238758087158203,
      "learning_rate": 3.138267052936529e-05,
      "loss": 0.7913,
      "step": 14147
    },
    {
      "epoch": 0.37260995522781143,
      "grad_norm": 2.1738319396972656,
      "learning_rate": 3.138135370028971e-05,
      "loss": 1.6217,
      "step": 14148
    },
    {
      "epoch": 0.3726362918093232,
      "grad_norm": 2.092710494995117,
      "learning_rate": 3.138003687121412e-05,
      "loss": 0.7267,
      "step": 14149
    },
    {
      "epoch": 0.37266262839083486,
      "grad_norm": 2.406601667404175,
      "learning_rate": 3.1378720042138534e-05,
      "loss": 1.948,
      "step": 14150
    },
    {
      "epoch": 0.3726889649723466,
      "grad_norm": 4.631277084350586,
      "learning_rate": 3.137740321306294e-05,
      "loss": 2.0091,
      "step": 14151
    },
    {
      "epoch": 0.3727153015538583,
      "grad_norm": 2.006218910217285,
      "learning_rate": 3.137608638398736e-05,
      "loss": 1.6647,
      "step": 14152
    },
    {
      "epoch": 0.37274163813537003,
      "grad_norm": 3.4320740699768066,
      "learning_rate": 3.1374769554911774e-05,
      "loss": 1.5823,
      "step": 14153
    },
    {
      "epoch": 0.3727679747168817,
      "grad_norm": 1.6155775785446167,
      "learning_rate": 3.137345272583619e-05,
      "loss": 1.6534,
      "step": 14154
    },
    {
      "epoch": 0.37279431129839347,
      "grad_norm": 3.2763047218322754,
      "learning_rate": 3.1372135896760605e-05,
      "loss": 1.8737,
      "step": 14155
    },
    {
      "epoch": 0.3728206478799052,
      "grad_norm": 1.7518818378448486,
      "learning_rate": 3.1370819067685014e-05,
      "loss": 2.08,
      "step": 14156
    },
    {
      "epoch": 0.3728469844614169,
      "grad_norm": 1.9589077234268188,
      "learning_rate": 3.136950223860943e-05,
      "loss": 1.1261,
      "step": 14157
    },
    {
      "epoch": 0.37287332104292864,
      "grad_norm": 1.6220275163650513,
      "learning_rate": 3.1368185409533845e-05,
      "loss": 0.6425,
      "step": 14158
    },
    {
      "epoch": 0.37289965762444033,
      "grad_norm": 1.9339220523834229,
      "learning_rate": 3.136686858045826e-05,
      "loss": 1.9084,
      "step": 14159
    },
    {
      "epoch": 0.37292599420595207,
      "grad_norm": 5.558414459228516,
      "learning_rate": 3.136555175138267e-05,
      "loss": 1.541,
      "step": 14160
    },
    {
      "epoch": 0.3729523307874638,
      "grad_norm": 1.734666347503662,
      "learning_rate": 3.1364234922307085e-05,
      "loss": 1.9991,
      "step": 14161
    },
    {
      "epoch": 0.3729786673689755,
      "grad_norm": 1.5355716943740845,
      "learning_rate": 3.13629180932315e-05,
      "loss": 2.1111,
      "step": 14162
    },
    {
      "epoch": 0.37300500395048725,
      "grad_norm": 2.3051373958587646,
      "learning_rate": 3.1361601264155916e-05,
      "loss": 1.4456,
      "step": 14163
    },
    {
      "epoch": 0.37303134053199893,
      "grad_norm": 5.503693103790283,
      "learning_rate": 3.136028443508033e-05,
      "loss": 1.0033,
      "step": 14164
    },
    {
      "epoch": 0.3730576771135107,
      "grad_norm": 1.771206259727478,
      "learning_rate": 3.135896760600474e-05,
      "loss": 1.7851,
      "step": 14165
    },
    {
      "epoch": 0.37308401369502237,
      "grad_norm": 1.6473339796066284,
      "learning_rate": 3.1357650776929156e-05,
      "loss": 1.6213,
      "step": 14166
    },
    {
      "epoch": 0.3731103502765341,
      "grad_norm": 2.366232395172119,
      "learning_rate": 3.135633394785357e-05,
      "loss": 1.5638,
      "step": 14167
    },
    {
      "epoch": 0.37313668685804585,
      "grad_norm": 1.6880323886871338,
      "learning_rate": 3.135501711877799e-05,
      "loss": 0.6219,
      "step": 14168
    },
    {
      "epoch": 0.37316302343955754,
      "grad_norm": 1.7996622323989868,
      "learning_rate": 3.1353700289702396e-05,
      "loss": 1.8728,
      "step": 14169
    },
    {
      "epoch": 0.3731893600210693,
      "grad_norm": 2.6654815673828125,
      "learning_rate": 3.135238346062681e-05,
      "loss": 0.6643,
      "step": 14170
    },
    {
      "epoch": 0.37321569660258097,
      "grad_norm": 2.1555862426757812,
      "learning_rate": 3.135106663155122e-05,
      "loss": 1.3774,
      "step": 14171
    },
    {
      "epoch": 0.3732420331840927,
      "grad_norm": 1.8232494592666626,
      "learning_rate": 3.134974980247564e-05,
      "loss": 1.8661,
      "step": 14172
    },
    {
      "epoch": 0.3732683697656044,
      "grad_norm": 2.0101449489593506,
      "learning_rate": 3.134843297340005e-05,
      "loss": 1.9914,
      "step": 14173
    },
    {
      "epoch": 0.37329470634711615,
      "grad_norm": 1.7984455823898315,
      "learning_rate": 3.134711614432447e-05,
      "loss": 2.402,
      "step": 14174
    },
    {
      "epoch": 0.3733210429286279,
      "grad_norm": 2.6387932300567627,
      "learning_rate": 3.134579931524888e-05,
      "loss": 1.3151,
      "step": 14175
    },
    {
      "epoch": 0.3733473795101396,
      "grad_norm": 2.278677225112915,
      "learning_rate": 3.13444824861733e-05,
      "loss": 0.8002,
      "step": 14176
    },
    {
      "epoch": 0.3733737160916513,
      "grad_norm": 2.45867919921875,
      "learning_rate": 3.1343165657097714e-05,
      "loss": 2.1641,
      "step": 14177
    },
    {
      "epoch": 0.373400052673163,
      "grad_norm": 2.5741069316864014,
      "learning_rate": 3.134184882802212e-05,
      "loss": 1.6112,
      "step": 14178
    },
    {
      "epoch": 0.37342638925467475,
      "grad_norm": 3.022954225540161,
      "learning_rate": 3.134053199894654e-05,
      "loss": 1.7018,
      "step": 14179
    },
    {
      "epoch": 0.37345272583618644,
      "grad_norm": 2.2265570163726807,
      "learning_rate": 3.133921516987095e-05,
      "loss": 1.3314,
      "step": 14180
    },
    {
      "epoch": 0.3734790624176982,
      "grad_norm": 6.127312660217285,
      "learning_rate": 3.133789834079537e-05,
      "loss": 1.6663,
      "step": 14181
    },
    {
      "epoch": 0.3735053989992099,
      "grad_norm": 2.0783584117889404,
      "learning_rate": 3.133658151171978e-05,
      "loss": 1.6706,
      "step": 14182
    },
    {
      "epoch": 0.3735317355807216,
      "grad_norm": 2.471951723098755,
      "learning_rate": 3.1335264682644194e-05,
      "loss": 1.0596,
      "step": 14183
    },
    {
      "epoch": 0.37355807216223336,
      "grad_norm": 1.9284427165985107,
      "learning_rate": 3.133394785356861e-05,
      "loss": 2.7822,
      "step": 14184
    },
    {
      "epoch": 0.37358440874374504,
      "grad_norm": 4.185712814331055,
      "learning_rate": 3.133263102449302e-05,
      "loss": 1.5907,
      "step": 14185
    },
    {
      "epoch": 0.3736107453252568,
      "grad_norm": 1.985437035560608,
      "learning_rate": 3.133131419541744e-05,
      "loss": 1.951,
      "step": 14186
    },
    {
      "epoch": 0.3736370819067685,
      "grad_norm": 3.8078482151031494,
      "learning_rate": 3.132999736634185e-05,
      "loss": 1.2052,
      "step": 14187
    },
    {
      "epoch": 0.3736634184882802,
      "grad_norm": 7.1121296882629395,
      "learning_rate": 3.1328680537266265e-05,
      "loss": 1.4383,
      "step": 14188
    },
    {
      "epoch": 0.37368975506979196,
      "grad_norm": 2.2999610900878906,
      "learning_rate": 3.1327363708190674e-05,
      "loss": 1.0848,
      "step": 14189
    },
    {
      "epoch": 0.37371609165130365,
      "grad_norm": 1.4526771306991577,
      "learning_rate": 3.132604687911509e-05,
      "loss": 1.7914,
      "step": 14190
    },
    {
      "epoch": 0.3737424282328154,
      "grad_norm": 3.920475721359253,
      "learning_rate": 3.1324730050039505e-05,
      "loss": 0.5756,
      "step": 14191
    },
    {
      "epoch": 0.3737687648143271,
      "grad_norm": 2.5304274559020996,
      "learning_rate": 3.132341322096392e-05,
      "loss": 1.6966,
      "step": 14192
    },
    {
      "epoch": 0.3737951013958388,
      "grad_norm": 2.783161163330078,
      "learning_rate": 3.1322096391888336e-05,
      "loss": 1.561,
      "step": 14193
    },
    {
      "epoch": 0.3738214379773505,
      "grad_norm": 1.5175174474716187,
      "learning_rate": 3.1320779562812745e-05,
      "loss": 2.2743,
      "step": 14194
    },
    {
      "epoch": 0.37384777455886226,
      "grad_norm": 3.7950727939605713,
      "learning_rate": 3.131946273373717e-05,
      "loss": 1.6094,
      "step": 14195
    },
    {
      "epoch": 0.373874111140374,
      "grad_norm": 1.7602031230926514,
      "learning_rate": 3.1318145904661576e-05,
      "loss": 1.6025,
      "step": 14196
    },
    {
      "epoch": 0.3739004477218857,
      "grad_norm": 2.615161657333374,
      "learning_rate": 3.131682907558599e-05,
      "loss": 1.7188,
      "step": 14197
    },
    {
      "epoch": 0.37392678430339743,
      "grad_norm": 2.873991012573242,
      "learning_rate": 3.13155122465104e-05,
      "loss": 1.7009,
      "step": 14198
    },
    {
      "epoch": 0.3739531208849091,
      "grad_norm": 2.0137271881103516,
      "learning_rate": 3.1314195417434816e-05,
      "loss": 1.8465,
      "step": 14199
    },
    {
      "epoch": 0.37397945746642086,
      "grad_norm": 4.88822603225708,
      "learning_rate": 3.131287858835923e-05,
      "loss": 1.0312,
      "step": 14200
    },
    {
      "epoch": 0.3740057940479326,
      "grad_norm": 2.652528762817383,
      "learning_rate": 3.131156175928365e-05,
      "loss": 2.0592,
      "step": 14201
    },
    {
      "epoch": 0.3740321306294443,
      "grad_norm": 2.9017488956451416,
      "learning_rate": 3.131024493020806e-05,
      "loss": 1.9195,
      "step": 14202
    },
    {
      "epoch": 0.37405846721095604,
      "grad_norm": 3.1204323768615723,
      "learning_rate": 3.130892810113247e-05,
      "loss": 1.662,
      "step": 14203
    },
    {
      "epoch": 0.3740848037924677,
      "grad_norm": 1.5588823556900024,
      "learning_rate": 3.130761127205689e-05,
      "loss": 1.8854,
      "step": 14204
    },
    {
      "epoch": 0.37411114037397947,
      "grad_norm": 2.6608121395111084,
      "learning_rate": 3.13062944429813e-05,
      "loss": 1.3078,
      "step": 14205
    },
    {
      "epoch": 0.37413747695549116,
      "grad_norm": 2.2102999687194824,
      "learning_rate": 3.130497761390572e-05,
      "loss": 2.0173,
      "step": 14206
    },
    {
      "epoch": 0.3741638135370029,
      "grad_norm": 1.4377378225326538,
      "learning_rate": 3.130366078483013e-05,
      "loss": 0.3523,
      "step": 14207
    },
    {
      "epoch": 0.37419015011851464,
      "grad_norm": 1.919411063194275,
      "learning_rate": 3.130234395575454e-05,
      "loss": 1.6903,
      "step": 14208
    },
    {
      "epoch": 0.37421648670002633,
      "grad_norm": 2.787741184234619,
      "learning_rate": 3.130102712667896e-05,
      "loss": 1.117,
      "step": 14209
    },
    {
      "epoch": 0.3742428232815381,
      "grad_norm": 2.6549298763275146,
      "learning_rate": 3.1299710297603375e-05,
      "loss": 1.3847,
      "step": 14210
    },
    {
      "epoch": 0.37426915986304976,
      "grad_norm": 1.8287451267242432,
      "learning_rate": 3.129839346852779e-05,
      "loss": 2.3063,
      "step": 14211
    },
    {
      "epoch": 0.3742954964445615,
      "grad_norm": 1.8640003204345703,
      "learning_rate": 3.12970766394522e-05,
      "loss": 1.1505,
      "step": 14212
    },
    {
      "epoch": 0.3743218330260732,
      "grad_norm": 4.248157024383545,
      "learning_rate": 3.1295759810376615e-05,
      "loss": 2.1364,
      "step": 14213
    },
    {
      "epoch": 0.37434816960758494,
      "grad_norm": 3.034585475921631,
      "learning_rate": 3.129444298130103e-05,
      "loss": 1.7178,
      "step": 14214
    },
    {
      "epoch": 0.3743745061890967,
      "grad_norm": 2.0356380939483643,
      "learning_rate": 3.1293126152225446e-05,
      "loss": 1.6205,
      "step": 14215
    },
    {
      "epoch": 0.37440084277060837,
      "grad_norm": 2.47143292427063,
      "learning_rate": 3.1291809323149855e-05,
      "loss": 1.8065,
      "step": 14216
    },
    {
      "epoch": 0.3744271793521201,
      "grad_norm": 3.6035900115966797,
      "learning_rate": 3.129049249407427e-05,
      "loss": 1.685,
      "step": 14217
    },
    {
      "epoch": 0.3744535159336318,
      "grad_norm": 1.8034420013427734,
      "learning_rate": 3.128917566499868e-05,
      "loss": 1.755,
      "step": 14218
    },
    {
      "epoch": 0.37447985251514354,
      "grad_norm": 2.6940577030181885,
      "learning_rate": 3.12878588359231e-05,
      "loss": 2.3524,
      "step": 14219
    },
    {
      "epoch": 0.37450618909665523,
      "grad_norm": 1.714198112487793,
      "learning_rate": 3.128654200684751e-05,
      "loss": 2.4048,
      "step": 14220
    },
    {
      "epoch": 0.37453252567816697,
      "grad_norm": 2.2402350902557373,
      "learning_rate": 3.1285225177771926e-05,
      "loss": 2.2052,
      "step": 14221
    },
    {
      "epoch": 0.3745588622596787,
      "grad_norm": 1.5270017385482788,
      "learning_rate": 3.128390834869634e-05,
      "loss": 1.6884,
      "step": 14222
    },
    {
      "epoch": 0.3745851988411904,
      "grad_norm": 1.75799560546875,
      "learning_rate": 3.128259151962075e-05,
      "loss": 2.4456,
      "step": 14223
    },
    {
      "epoch": 0.37461153542270215,
      "grad_norm": 1.701562523841858,
      "learning_rate": 3.128127469054517e-05,
      "loss": 1.8755,
      "step": 14224
    },
    {
      "epoch": 0.37463787200421383,
      "grad_norm": 1.492091417312622,
      "learning_rate": 3.127995786146958e-05,
      "loss": 1.8,
      "step": 14225
    },
    {
      "epoch": 0.3746642085857256,
      "grad_norm": 5.475642681121826,
      "learning_rate": 3.1278641032394e-05,
      "loss": 1.9598,
      "step": 14226
    },
    {
      "epoch": 0.37469054516723727,
      "grad_norm": 1.8512247800827026,
      "learning_rate": 3.1277324203318406e-05,
      "loss": 0.8004,
      "step": 14227
    },
    {
      "epoch": 0.374716881748749,
      "grad_norm": 1.6653473377227783,
      "learning_rate": 3.127600737424283e-05,
      "loss": 1.8888,
      "step": 14228
    },
    {
      "epoch": 0.37474321833026075,
      "grad_norm": 2.3999147415161133,
      "learning_rate": 3.127469054516724e-05,
      "loss": 0.4719,
      "step": 14229
    },
    {
      "epoch": 0.37476955491177244,
      "grad_norm": 1.8173974752426147,
      "learning_rate": 3.127337371609165e-05,
      "loss": 1.2811,
      "step": 14230
    },
    {
      "epoch": 0.3747958914932842,
      "grad_norm": 2.0996522903442383,
      "learning_rate": 3.127205688701607e-05,
      "loss": 1.2555,
      "step": 14231
    },
    {
      "epoch": 0.37482222807479587,
      "grad_norm": 3.778411626815796,
      "learning_rate": 3.127074005794048e-05,
      "loss": 1.3732,
      "step": 14232
    },
    {
      "epoch": 0.3748485646563076,
      "grad_norm": 1.517289161682129,
      "learning_rate": 3.12694232288649e-05,
      "loss": 1.9646,
      "step": 14233
    },
    {
      "epoch": 0.37487490123781936,
      "grad_norm": 1.2089574337005615,
      "learning_rate": 3.126810639978931e-05,
      "loss": 1.5684,
      "step": 14234
    },
    {
      "epoch": 0.37490123781933105,
      "grad_norm": 2.1433467864990234,
      "learning_rate": 3.1266789570713724e-05,
      "loss": 1.7222,
      "step": 14235
    },
    {
      "epoch": 0.3749275744008428,
      "grad_norm": 3.5721564292907715,
      "learning_rate": 3.126547274163813e-05,
      "loss": 1.775,
      "step": 14236
    },
    {
      "epoch": 0.3749539109823545,
      "grad_norm": 2.8540072441101074,
      "learning_rate": 3.126415591256255e-05,
      "loss": 1.4602,
      "step": 14237
    },
    {
      "epoch": 0.3749802475638662,
      "grad_norm": 2.129418134689331,
      "learning_rate": 3.1262839083486964e-05,
      "loss": 1.9492,
      "step": 14238
    },
    {
      "epoch": 0.3750065841453779,
      "grad_norm": 2.413728952407837,
      "learning_rate": 3.126152225441138e-05,
      "loss": 1.7976,
      "step": 14239
    },
    {
      "epoch": 0.37503292072688965,
      "grad_norm": 1.9544322490692139,
      "learning_rate": 3.1260205425335795e-05,
      "loss": 1.621,
      "step": 14240
    },
    {
      "epoch": 0.3750592573084014,
      "grad_norm": 2.624230146408081,
      "learning_rate": 3.1258888596260204e-05,
      "loss": 0.5389,
      "step": 14241
    },
    {
      "epoch": 0.3750855938899131,
      "grad_norm": 3.2988929748535156,
      "learning_rate": 3.1257571767184626e-05,
      "loss": 1.164,
      "step": 14242
    },
    {
      "epoch": 0.3751119304714248,
      "grad_norm": 4.124659538269043,
      "learning_rate": 3.1256254938109035e-05,
      "loss": 1.8363,
      "step": 14243
    },
    {
      "epoch": 0.3751382670529365,
      "grad_norm": 2.81253719329834,
      "learning_rate": 3.125493810903345e-05,
      "loss": 0.4286,
      "step": 14244
    },
    {
      "epoch": 0.37516460363444826,
      "grad_norm": 1.5272924900054932,
      "learning_rate": 3.125362127995786e-05,
      "loss": 1.2971,
      "step": 14245
    },
    {
      "epoch": 0.37519094021595994,
      "grad_norm": 1.772472858428955,
      "learning_rate": 3.1252304450882275e-05,
      "loss": 2.2284,
      "step": 14246
    },
    {
      "epoch": 0.3752172767974717,
      "grad_norm": 1.7253715991973877,
      "learning_rate": 3.125098762180669e-05,
      "loss": 1.6207,
      "step": 14247
    },
    {
      "epoch": 0.37524361337898343,
      "grad_norm": 3.1210827827453613,
      "learning_rate": 3.1249670792731106e-05,
      "loss": 1.4943,
      "step": 14248
    },
    {
      "epoch": 0.3752699499604951,
      "grad_norm": 1.8335567712783813,
      "learning_rate": 3.124835396365552e-05,
      "loss": 1.5876,
      "step": 14249
    },
    {
      "epoch": 0.37529628654200686,
      "grad_norm": 2.278203010559082,
      "learning_rate": 3.124703713457993e-05,
      "loss": 2.2487,
      "step": 14250
    },
    {
      "epoch": 0.37532262312351855,
      "grad_norm": 2.0389113426208496,
      "learning_rate": 3.1245720305504346e-05,
      "loss": 2.1409,
      "step": 14251
    },
    {
      "epoch": 0.3753489597050303,
      "grad_norm": 2.7280356884002686,
      "learning_rate": 3.124440347642876e-05,
      "loss": 1.7056,
      "step": 14252
    },
    {
      "epoch": 0.375375296286542,
      "grad_norm": 2.6063644886016846,
      "learning_rate": 3.124308664735318e-05,
      "loss": 1.0552,
      "step": 14253
    },
    {
      "epoch": 0.3754016328680537,
      "grad_norm": 2.29797101020813,
      "learning_rate": 3.1241769818277586e-05,
      "loss": 2.0279,
      "step": 14254
    },
    {
      "epoch": 0.37542796944956547,
      "grad_norm": 2.8565499782562256,
      "learning_rate": 3.1240452989202e-05,
      "loss": 1.7775,
      "step": 14255
    },
    {
      "epoch": 0.37545430603107716,
      "grad_norm": 1.9371744394302368,
      "learning_rate": 3.123913616012642e-05,
      "loss": 0.6295,
      "step": 14256
    },
    {
      "epoch": 0.3754806426125889,
      "grad_norm": 2.2951388359069824,
      "learning_rate": 3.123781933105083e-05,
      "loss": 1.7932,
      "step": 14257
    },
    {
      "epoch": 0.3755069791941006,
      "grad_norm": 1.7291057109832764,
      "learning_rate": 3.123650250197525e-05,
      "loss": 1.6975,
      "step": 14258
    },
    {
      "epoch": 0.37553331577561233,
      "grad_norm": 3.547032594680786,
      "learning_rate": 3.123518567289966e-05,
      "loss": 1.7393,
      "step": 14259
    },
    {
      "epoch": 0.375559652357124,
      "grad_norm": 3.6985456943511963,
      "learning_rate": 3.123386884382407e-05,
      "loss": 0.9514,
      "step": 14260
    },
    {
      "epoch": 0.37558598893863576,
      "grad_norm": 3.034712791442871,
      "learning_rate": 3.123255201474849e-05,
      "loss": 1.9821,
      "step": 14261
    },
    {
      "epoch": 0.3756123255201475,
      "grad_norm": 2.120142698287964,
      "learning_rate": 3.1231235185672904e-05,
      "loss": 1.5462,
      "step": 14262
    },
    {
      "epoch": 0.3756386621016592,
      "grad_norm": 1.5356390476226807,
      "learning_rate": 3.122991835659731e-05,
      "loss": 1.8039,
      "step": 14263
    },
    {
      "epoch": 0.37566499868317094,
      "grad_norm": 2.625803232192993,
      "learning_rate": 3.122860152752173e-05,
      "loss": 1.663,
      "step": 14264
    },
    {
      "epoch": 0.3756913352646826,
      "grad_norm": 2.216063976287842,
      "learning_rate": 3.122728469844614e-05,
      "loss": 0.4178,
      "step": 14265
    },
    {
      "epoch": 0.37571767184619437,
      "grad_norm": 1.937024474143982,
      "learning_rate": 3.122596786937056e-05,
      "loss": 1.9679,
      "step": 14266
    },
    {
      "epoch": 0.3757440084277061,
      "grad_norm": 2.2436394691467285,
      "learning_rate": 3.1224651040294975e-05,
      "loss": 1.7544,
      "step": 14267
    },
    {
      "epoch": 0.3757703450092178,
      "grad_norm": 3.0948965549468994,
      "learning_rate": 3.1223334211219384e-05,
      "loss": 0.8748,
      "step": 14268
    },
    {
      "epoch": 0.37579668159072954,
      "grad_norm": 3.4630696773529053,
      "learning_rate": 3.12220173821438e-05,
      "loss": 1.6059,
      "step": 14269
    },
    {
      "epoch": 0.37582301817224123,
      "grad_norm": 2.200232982635498,
      "learning_rate": 3.122070055306821e-05,
      "loss": 1.9663,
      "step": 14270
    },
    {
      "epoch": 0.375849354753753,
      "grad_norm": 1.6948637962341309,
      "learning_rate": 3.121938372399263e-05,
      "loss": 1.7678,
      "step": 14271
    },
    {
      "epoch": 0.37587569133526466,
      "grad_norm": 2.5107574462890625,
      "learning_rate": 3.121806689491704e-05,
      "loss": 1.7259,
      "step": 14272
    },
    {
      "epoch": 0.3759020279167764,
      "grad_norm": 1.7776470184326172,
      "learning_rate": 3.1216750065841456e-05,
      "loss": 2.071,
      "step": 14273
    },
    {
      "epoch": 0.37592836449828815,
      "grad_norm": 3.5129220485687256,
      "learning_rate": 3.1215433236765864e-05,
      "loss": 2.2855,
      "step": 14274
    },
    {
      "epoch": 0.37595470107979984,
      "grad_norm": 2.658547878265381,
      "learning_rate": 3.121411640769029e-05,
      "loss": 1.6774,
      "step": 14275
    },
    {
      "epoch": 0.3759810376613116,
      "grad_norm": 2.8447399139404297,
      "learning_rate": 3.1212799578614696e-05,
      "loss": 1.5263,
      "step": 14276
    },
    {
      "epoch": 0.37600737424282327,
      "grad_norm": 2.125356435775757,
      "learning_rate": 3.121148274953911e-05,
      "loss": 1.8265,
      "step": 14277
    },
    {
      "epoch": 0.376033710824335,
      "grad_norm": 1.9518109560012817,
      "learning_rate": 3.121016592046353e-05,
      "loss": 1.8052,
      "step": 14278
    },
    {
      "epoch": 0.3760600474058467,
      "grad_norm": 3.5467631816864014,
      "learning_rate": 3.1208849091387936e-05,
      "loss": 0.8193,
      "step": 14279
    },
    {
      "epoch": 0.37608638398735844,
      "grad_norm": 1.7447534799575806,
      "learning_rate": 3.120753226231236e-05,
      "loss": 1.9299,
      "step": 14280
    },
    {
      "epoch": 0.3761127205688702,
      "grad_norm": 4.877032279968262,
      "learning_rate": 3.120621543323677e-05,
      "loss": 1.4584,
      "step": 14281
    },
    {
      "epoch": 0.3761390571503819,
      "grad_norm": 1.6478792428970337,
      "learning_rate": 3.120489860416118e-05,
      "loss": 1.7871,
      "step": 14282
    },
    {
      "epoch": 0.3761653937318936,
      "grad_norm": 4.124168872833252,
      "learning_rate": 3.120358177508559e-05,
      "loss": 1.8913,
      "step": 14283
    },
    {
      "epoch": 0.3761917303134053,
      "grad_norm": 1.914576530456543,
      "learning_rate": 3.120226494601001e-05,
      "loss": 1.765,
      "step": 14284
    },
    {
      "epoch": 0.37621806689491705,
      "grad_norm": 2.2469708919525146,
      "learning_rate": 3.120094811693442e-05,
      "loss": 1.6705,
      "step": 14285
    },
    {
      "epoch": 0.37624440347642873,
      "grad_norm": 2.927720308303833,
      "learning_rate": 3.119963128785884e-05,
      "loss": 0.6566,
      "step": 14286
    },
    {
      "epoch": 0.3762707400579405,
      "grad_norm": 2.779019355773926,
      "learning_rate": 3.1198314458783254e-05,
      "loss": 2.1069,
      "step": 14287
    },
    {
      "epoch": 0.3762970766394522,
      "grad_norm": 4.50929069519043,
      "learning_rate": 3.119699762970766e-05,
      "loss": 1.3856,
      "step": 14288
    },
    {
      "epoch": 0.3763234132209639,
      "grad_norm": 2.9545562267303467,
      "learning_rate": 3.1195680800632085e-05,
      "loss": 1.5467,
      "step": 14289
    },
    {
      "epoch": 0.37634974980247565,
      "grad_norm": 1.5907179117202759,
      "learning_rate": 3.1194363971556494e-05,
      "loss": 1.7183,
      "step": 14290
    },
    {
      "epoch": 0.37637608638398734,
      "grad_norm": 2.221656560897827,
      "learning_rate": 3.119304714248091e-05,
      "loss": 1.9421,
      "step": 14291
    },
    {
      "epoch": 0.3764024229654991,
      "grad_norm": 1.7442340850830078,
      "learning_rate": 3.119173031340532e-05,
      "loss": 1.963,
      "step": 14292
    },
    {
      "epoch": 0.37642875954701077,
      "grad_norm": 2.589993715286255,
      "learning_rate": 3.1190413484329734e-05,
      "loss": 1.234,
      "step": 14293
    },
    {
      "epoch": 0.3764550961285225,
      "grad_norm": 1.6990188360214233,
      "learning_rate": 3.118909665525415e-05,
      "loss": 1.8883,
      "step": 14294
    },
    {
      "epoch": 0.37648143271003426,
      "grad_norm": 2.229484796524048,
      "learning_rate": 3.1187779826178565e-05,
      "loss": 1.3622,
      "step": 14295
    },
    {
      "epoch": 0.37650776929154595,
      "grad_norm": 2.9781570434570312,
      "learning_rate": 3.118646299710298e-05,
      "loss": 1.5697,
      "step": 14296
    },
    {
      "epoch": 0.3765341058730577,
      "grad_norm": 2.28656268119812,
      "learning_rate": 3.118514616802739e-05,
      "loss": 1.4697,
      "step": 14297
    },
    {
      "epoch": 0.3765604424545694,
      "grad_norm": 2.038545608520508,
      "learning_rate": 3.1183829338951805e-05,
      "loss": 2.3618,
      "step": 14298
    },
    {
      "epoch": 0.3765867790360811,
      "grad_norm": 2.4138267040252686,
      "learning_rate": 3.118251250987622e-05,
      "loss": 1.2343,
      "step": 14299
    },
    {
      "epoch": 0.37661311561759286,
      "grad_norm": 2.261301040649414,
      "learning_rate": 3.1181195680800636e-05,
      "loss": 1.678,
      "step": 14300
    },
    {
      "epoch": 0.37663945219910455,
      "grad_norm": 2.581404685974121,
      "learning_rate": 3.1179878851725045e-05,
      "loss": 0.6939,
      "step": 14301
    },
    {
      "epoch": 0.3766657887806163,
      "grad_norm": 2.3161263465881348,
      "learning_rate": 3.117856202264946e-05,
      "loss": 1.53,
      "step": 14302
    },
    {
      "epoch": 0.376692125362128,
      "grad_norm": 4.823411464691162,
      "learning_rate": 3.1177245193573876e-05,
      "loss": 1.2247,
      "step": 14303
    },
    {
      "epoch": 0.3767184619436397,
      "grad_norm": 3.043858528137207,
      "learning_rate": 3.117592836449829e-05,
      "loss": 1.5187,
      "step": 14304
    },
    {
      "epoch": 0.3767447985251514,
      "grad_norm": 3.514554738998413,
      "learning_rate": 3.117461153542271e-05,
      "loss": 1.9775,
      "step": 14305
    },
    {
      "epoch": 0.37677113510666316,
      "grad_norm": 6.378372669219971,
      "learning_rate": 3.1173294706347116e-05,
      "loss": 1.3842,
      "step": 14306
    },
    {
      "epoch": 0.3767974716881749,
      "grad_norm": 1.59242844581604,
      "learning_rate": 3.117197787727153e-05,
      "loss": 1.8225,
      "step": 14307
    },
    {
      "epoch": 0.3768238082696866,
      "grad_norm": 3.1263887882232666,
      "learning_rate": 3.117066104819595e-05,
      "loss": 0.9102,
      "step": 14308
    },
    {
      "epoch": 0.37685014485119833,
      "grad_norm": 2.5310606956481934,
      "learning_rate": 3.116934421912036e-05,
      "loss": 1.5848,
      "step": 14309
    },
    {
      "epoch": 0.37687648143271,
      "grad_norm": 2.244361162185669,
      "learning_rate": 3.116802739004477e-05,
      "loss": 2.0851,
      "step": 14310
    },
    {
      "epoch": 0.37690281801422176,
      "grad_norm": 6.106942653656006,
      "learning_rate": 3.116671056096919e-05,
      "loss": 1.9758,
      "step": 14311
    },
    {
      "epoch": 0.37692915459573345,
      "grad_norm": 3.1216237545013428,
      "learning_rate": 3.11653937318936e-05,
      "loss": 1.7616,
      "step": 14312
    },
    {
      "epoch": 0.3769554911772452,
      "grad_norm": 1.9634357690811157,
      "learning_rate": 3.116407690281802e-05,
      "loss": 1.4297,
      "step": 14313
    },
    {
      "epoch": 0.37698182775875694,
      "grad_norm": 2.4202892780303955,
      "learning_rate": 3.1162760073742434e-05,
      "loss": 1.1619,
      "step": 14314
    },
    {
      "epoch": 0.3770081643402686,
      "grad_norm": 5.1971821784973145,
      "learning_rate": 3.116144324466684e-05,
      "loss": 1.3166,
      "step": 14315
    },
    {
      "epoch": 0.37703450092178037,
      "grad_norm": 4.500625133514404,
      "learning_rate": 3.116012641559126e-05,
      "loss": 1.5447,
      "step": 14316
    },
    {
      "epoch": 0.37706083750329206,
      "grad_norm": 2.278379440307617,
      "learning_rate": 3.115880958651567e-05,
      "loss": 1.3583,
      "step": 14317
    },
    {
      "epoch": 0.3770871740848038,
      "grad_norm": 1.9895977973937988,
      "learning_rate": 3.115749275744009e-05,
      "loss": 2.1477,
      "step": 14318
    },
    {
      "epoch": 0.3771135106663155,
      "grad_norm": 2.3393542766571045,
      "learning_rate": 3.11561759283645e-05,
      "loss": 1.4648,
      "step": 14319
    },
    {
      "epoch": 0.37713984724782723,
      "grad_norm": 2.84989595413208,
      "learning_rate": 3.1154859099288914e-05,
      "loss": 1.08,
      "step": 14320
    },
    {
      "epoch": 0.377166183829339,
      "grad_norm": 2.0685625076293945,
      "learning_rate": 3.115354227021332e-05,
      "loss": 2.2206,
      "step": 14321
    },
    {
      "epoch": 0.37719252041085066,
      "grad_norm": 2.290341377258301,
      "learning_rate": 3.1152225441137745e-05,
      "loss": 1.2715,
      "step": 14322
    },
    {
      "epoch": 0.3772188569923624,
      "grad_norm": 2.8589868545532227,
      "learning_rate": 3.1150908612062154e-05,
      "loss": 1.4608,
      "step": 14323
    },
    {
      "epoch": 0.3772451935738741,
      "grad_norm": 3.002072334289551,
      "learning_rate": 3.114959178298657e-05,
      "loss": 1.0854,
      "step": 14324
    },
    {
      "epoch": 0.37727153015538584,
      "grad_norm": 1.778875470161438,
      "learning_rate": 3.1148274953910985e-05,
      "loss": 1.919,
      "step": 14325
    },
    {
      "epoch": 0.3772978667368975,
      "grad_norm": 3.146024465560913,
      "learning_rate": 3.1146958124835394e-05,
      "loss": 1.6025,
      "step": 14326
    },
    {
      "epoch": 0.37732420331840927,
      "grad_norm": 2.0379104614257812,
      "learning_rate": 3.1145641295759816e-05,
      "loss": 1.8401,
      "step": 14327
    },
    {
      "epoch": 0.377350539899921,
      "grad_norm": 5.192755222320557,
      "learning_rate": 3.1144324466684225e-05,
      "loss": 2.2354,
      "step": 14328
    },
    {
      "epoch": 0.3773768764814327,
      "grad_norm": 3.909512758255005,
      "learning_rate": 3.114300763760864e-05,
      "loss": 1.8898,
      "step": 14329
    },
    {
      "epoch": 0.37740321306294444,
      "grad_norm": 3.0050933361053467,
      "learning_rate": 3.114169080853305e-05,
      "loss": 1.3986,
      "step": 14330
    },
    {
      "epoch": 0.37742954964445613,
      "grad_norm": 4.337152004241943,
      "learning_rate": 3.1140373979457465e-05,
      "loss": 1.3896,
      "step": 14331
    },
    {
      "epoch": 0.3774558862259679,
      "grad_norm": 3.046328067779541,
      "learning_rate": 3.113905715038188e-05,
      "loss": 1.6536,
      "step": 14332
    },
    {
      "epoch": 0.37748222280747956,
      "grad_norm": 1.9460607767105103,
      "learning_rate": 3.1137740321306297e-05,
      "loss": 0.8839,
      "step": 14333
    },
    {
      "epoch": 0.3775085593889913,
      "grad_norm": 1.8497570753097534,
      "learning_rate": 3.113642349223071e-05,
      "loss": 0.5827,
      "step": 14334
    },
    {
      "epoch": 0.37753489597050305,
      "grad_norm": 2.7690281867980957,
      "learning_rate": 3.113510666315512e-05,
      "loss": 2.3619,
      "step": 14335
    },
    {
      "epoch": 0.37756123255201474,
      "grad_norm": 3.2679927349090576,
      "learning_rate": 3.1133789834079537e-05,
      "loss": 2.3432,
      "step": 14336
    },
    {
      "epoch": 0.3775875691335265,
      "grad_norm": 2.757723808288574,
      "learning_rate": 3.113247300500395e-05,
      "loss": 1.5453,
      "step": 14337
    },
    {
      "epoch": 0.37761390571503817,
      "grad_norm": 1.9668294191360474,
      "learning_rate": 3.113115617592837e-05,
      "loss": 1.3436,
      "step": 14338
    },
    {
      "epoch": 0.3776402422965499,
      "grad_norm": 3.4913876056671143,
      "learning_rate": 3.1129839346852777e-05,
      "loss": 1.083,
      "step": 14339
    },
    {
      "epoch": 0.37766657887806165,
      "grad_norm": 1.5953645706176758,
      "learning_rate": 3.112852251777719e-05,
      "loss": 2.4503,
      "step": 14340
    },
    {
      "epoch": 0.37769291545957334,
      "grad_norm": 2.0951671600341797,
      "learning_rate": 3.112720568870161e-05,
      "loss": 2.0702,
      "step": 14341
    },
    {
      "epoch": 0.3777192520410851,
      "grad_norm": 2.076826333999634,
      "learning_rate": 3.112588885962602e-05,
      "loss": 1.8927,
      "step": 14342
    },
    {
      "epoch": 0.3777455886225968,
      "grad_norm": 3.5407419204711914,
      "learning_rate": 3.112457203055044e-05,
      "loss": 1.0983,
      "step": 14343
    },
    {
      "epoch": 0.3777719252041085,
      "grad_norm": 2.486903667449951,
      "learning_rate": 3.112325520147485e-05,
      "loss": 1.7075,
      "step": 14344
    },
    {
      "epoch": 0.3777982617856202,
      "grad_norm": 3.300886392593384,
      "learning_rate": 3.112193837239926e-05,
      "loss": 1.1176,
      "step": 14345
    },
    {
      "epoch": 0.37782459836713195,
      "grad_norm": 3.4425394535064697,
      "learning_rate": 3.112062154332368e-05,
      "loss": 2.2989,
      "step": 14346
    },
    {
      "epoch": 0.3778509349486437,
      "grad_norm": 1.7959916591644287,
      "learning_rate": 3.1119304714248095e-05,
      "loss": 1.9963,
      "step": 14347
    },
    {
      "epoch": 0.3778772715301554,
      "grad_norm": 5.633258819580078,
      "learning_rate": 3.11179878851725e-05,
      "loss": 1.5917,
      "step": 14348
    },
    {
      "epoch": 0.3779036081116671,
      "grad_norm": 2.0160927772521973,
      "learning_rate": 3.111667105609692e-05,
      "loss": 1.7902,
      "step": 14349
    },
    {
      "epoch": 0.3779299446931788,
      "grad_norm": 5.685546398162842,
      "learning_rate": 3.1115354227021335e-05,
      "loss": 1.2652,
      "step": 14350
    },
    {
      "epoch": 0.37795628127469055,
      "grad_norm": 5.127215385437012,
      "learning_rate": 3.111403739794575e-05,
      "loss": 1.3402,
      "step": 14351
    },
    {
      "epoch": 0.37798261785620224,
      "grad_norm": 1.8516260385513306,
      "learning_rate": 3.1112720568870166e-05,
      "loss": 1.6599,
      "step": 14352
    },
    {
      "epoch": 0.378008954437714,
      "grad_norm": 1.600586175918579,
      "learning_rate": 3.1111403739794575e-05,
      "loss": 1.6299,
      "step": 14353
    },
    {
      "epoch": 0.3780352910192257,
      "grad_norm": 3.3044614791870117,
      "learning_rate": 3.111008691071899e-05,
      "loss": 0.5645,
      "step": 14354
    },
    {
      "epoch": 0.3780616276007374,
      "grad_norm": 2.66239333152771,
      "learning_rate": 3.1108770081643406e-05,
      "loss": 0.9753,
      "step": 14355
    },
    {
      "epoch": 0.37808796418224916,
      "grad_norm": 2.1020607948303223,
      "learning_rate": 3.110745325256782e-05,
      "loss": 1.621,
      "step": 14356
    },
    {
      "epoch": 0.37811430076376085,
      "grad_norm": 3.535782814025879,
      "learning_rate": 3.110613642349223e-05,
      "loss": 1.623,
      "step": 14357
    },
    {
      "epoch": 0.3781406373452726,
      "grad_norm": 1.6616016626358032,
      "learning_rate": 3.1104819594416646e-05,
      "loss": 1.938,
      "step": 14358
    },
    {
      "epoch": 0.3781669739267843,
      "grad_norm": 2.5869176387786865,
      "learning_rate": 3.110350276534106e-05,
      "loss": 1.134,
      "step": 14359
    },
    {
      "epoch": 0.378193310508296,
      "grad_norm": 2.189210891723633,
      "learning_rate": 3.110218593626548e-05,
      "loss": 1.5386,
      "step": 14360
    },
    {
      "epoch": 0.37821964708980776,
      "grad_norm": 1.8407204151153564,
      "learning_rate": 3.110086910718989e-05,
      "loss": 1.9766,
      "step": 14361
    },
    {
      "epoch": 0.37824598367131945,
      "grad_norm": 2.904205799102783,
      "learning_rate": 3.10995522781143e-05,
      "loss": 1.7497,
      "step": 14362
    },
    {
      "epoch": 0.3782723202528312,
      "grad_norm": 2.136925220489502,
      "learning_rate": 3.109823544903872e-05,
      "loss": 1.4866,
      "step": 14363
    },
    {
      "epoch": 0.3782986568343429,
      "grad_norm": 3.7563652992248535,
      "learning_rate": 3.1096918619963126e-05,
      "loss": 2.0095,
      "step": 14364
    },
    {
      "epoch": 0.3783249934158546,
      "grad_norm": 1.619258999824524,
      "learning_rate": 3.109560179088755e-05,
      "loss": 1.599,
      "step": 14365
    },
    {
      "epoch": 0.3783513299973663,
      "grad_norm": 1.876483678817749,
      "learning_rate": 3.109428496181196e-05,
      "loss": 2.3886,
      "step": 14366
    },
    {
      "epoch": 0.37837766657887806,
      "grad_norm": 4.6059956550598145,
      "learning_rate": 3.109296813273637e-05,
      "loss": 1.3271,
      "step": 14367
    },
    {
      "epoch": 0.3784040031603898,
      "grad_norm": 2.8964126110076904,
      "learning_rate": 3.109165130366078e-05,
      "loss": 2.1777,
      "step": 14368
    },
    {
      "epoch": 0.3784303397419015,
      "grad_norm": 1.7897956371307373,
      "learning_rate": 3.10903344745852e-05,
      "loss": 1.8077,
      "step": 14369
    },
    {
      "epoch": 0.37845667632341323,
      "grad_norm": 2.401308298110962,
      "learning_rate": 3.108901764550961e-05,
      "loss": 1.1056,
      "step": 14370
    },
    {
      "epoch": 0.3784830129049249,
      "grad_norm": 3.4855611324310303,
      "learning_rate": 3.108770081643403e-05,
      "loss": 1.0769,
      "step": 14371
    },
    {
      "epoch": 0.37850934948643666,
      "grad_norm": 1.5533421039581299,
      "learning_rate": 3.1086383987358444e-05,
      "loss": 1.8009,
      "step": 14372
    },
    {
      "epoch": 0.3785356860679484,
      "grad_norm": 2.0275983810424805,
      "learning_rate": 3.108506715828285e-05,
      "loss": 1.8764,
      "step": 14373
    },
    {
      "epoch": 0.3785620226494601,
      "grad_norm": 3.17368483543396,
      "learning_rate": 3.1083750329207275e-05,
      "loss": 1.5531,
      "step": 14374
    },
    {
      "epoch": 0.37858835923097184,
      "grad_norm": 1.7626057863235474,
      "learning_rate": 3.1082433500131684e-05,
      "loss": 1.8141,
      "step": 14375
    },
    {
      "epoch": 0.3786146958124835,
      "grad_norm": 2.963604211807251,
      "learning_rate": 3.10811166710561e-05,
      "loss": 2.0978,
      "step": 14376
    },
    {
      "epoch": 0.37864103239399527,
      "grad_norm": 1.7826277017593384,
      "learning_rate": 3.107979984198051e-05,
      "loss": 2.2907,
      "step": 14377
    },
    {
      "epoch": 0.37866736897550696,
      "grad_norm": 3.0602974891662598,
      "learning_rate": 3.1078483012904924e-05,
      "loss": 1.7572,
      "step": 14378
    },
    {
      "epoch": 0.3786937055570187,
      "grad_norm": 2.30256986618042,
      "learning_rate": 3.107716618382934e-05,
      "loss": 1.7281,
      "step": 14379
    },
    {
      "epoch": 0.37872004213853044,
      "grad_norm": 4.591120719909668,
      "learning_rate": 3.1075849354753755e-05,
      "loss": 1.9756,
      "step": 14380
    },
    {
      "epoch": 0.37874637872004213,
      "grad_norm": 1.236663818359375,
      "learning_rate": 3.107453252567817e-05,
      "loss": 0.272,
      "step": 14381
    },
    {
      "epoch": 0.3787727153015539,
      "grad_norm": 3.8244235515594482,
      "learning_rate": 3.107321569660258e-05,
      "loss": 1.4714,
      "step": 14382
    },
    {
      "epoch": 0.37879905188306556,
      "grad_norm": 1.9342507123947144,
      "learning_rate": 3.1071898867526995e-05,
      "loss": 2.2386,
      "step": 14383
    },
    {
      "epoch": 0.3788253884645773,
      "grad_norm": 2.342454433441162,
      "learning_rate": 3.107058203845141e-05,
      "loss": 1.7896,
      "step": 14384
    },
    {
      "epoch": 0.378851725046089,
      "grad_norm": 2.1913812160491943,
      "learning_rate": 3.1069265209375826e-05,
      "loss": 1.7999,
      "step": 14385
    },
    {
      "epoch": 0.37887806162760074,
      "grad_norm": 2.001167058944702,
      "learning_rate": 3.1067948380300235e-05,
      "loss": 0.3985,
      "step": 14386
    },
    {
      "epoch": 0.3789043982091125,
      "grad_norm": 2.160733938217163,
      "learning_rate": 3.106663155122465e-05,
      "loss": 1.6194,
      "step": 14387
    },
    {
      "epoch": 0.37893073479062417,
      "grad_norm": 4.565048694610596,
      "learning_rate": 3.1065314722149066e-05,
      "loss": 1.1386,
      "step": 14388
    },
    {
      "epoch": 0.3789570713721359,
      "grad_norm": 2.2453153133392334,
      "learning_rate": 3.106399789307348e-05,
      "loss": 1.8115,
      "step": 14389
    },
    {
      "epoch": 0.3789834079536476,
      "grad_norm": 1.8231923580169678,
      "learning_rate": 3.10626810639979e-05,
      "loss": 1.8439,
      "step": 14390
    },
    {
      "epoch": 0.37900974453515934,
      "grad_norm": 1.9987602233886719,
      "learning_rate": 3.1061364234922306e-05,
      "loss": 0.4152,
      "step": 14391
    },
    {
      "epoch": 0.37903608111667103,
      "grad_norm": 1.613195538520813,
      "learning_rate": 3.106004740584672e-05,
      "loss": 1.7044,
      "step": 14392
    },
    {
      "epoch": 0.3790624176981828,
      "grad_norm": 2.655754327774048,
      "learning_rate": 3.105873057677114e-05,
      "loss": 1.3212,
      "step": 14393
    },
    {
      "epoch": 0.3790887542796945,
      "grad_norm": 1.6318649053573608,
      "learning_rate": 3.105741374769555e-05,
      "loss": 1.1494,
      "step": 14394
    },
    {
      "epoch": 0.3791150908612062,
      "grad_norm": 1.8283952474594116,
      "learning_rate": 3.105609691861996e-05,
      "loss": 1.6822,
      "step": 14395
    },
    {
      "epoch": 0.37914142744271795,
      "grad_norm": 1.5157963037490845,
      "learning_rate": 3.105478008954438e-05,
      "loss": 2.0481,
      "step": 14396
    },
    {
      "epoch": 0.37916776402422964,
      "grad_norm": 3.3179919719696045,
      "learning_rate": 3.105346326046879e-05,
      "loss": 0.7524,
      "step": 14397
    },
    {
      "epoch": 0.3791941006057414,
      "grad_norm": 2.909348964691162,
      "learning_rate": 3.105214643139321e-05,
      "loss": 0.9953,
      "step": 14398
    },
    {
      "epoch": 0.37922043718725307,
      "grad_norm": 2.466460943222046,
      "learning_rate": 3.1050829602317624e-05,
      "loss": 1.6126,
      "step": 14399
    },
    {
      "epoch": 0.3792467737687648,
      "grad_norm": 2.0761566162109375,
      "learning_rate": 3.104951277324203e-05,
      "loss": 1.0647,
      "step": 14400
    },
    {
      "epoch": 0.37927311035027655,
      "grad_norm": 4.008663177490234,
      "learning_rate": 3.104819594416645e-05,
      "loss": 1.4261,
      "step": 14401
    },
    {
      "epoch": 0.37929944693178824,
      "grad_norm": 2.338958740234375,
      "learning_rate": 3.104687911509086e-05,
      "loss": 2.3638,
      "step": 14402
    },
    {
      "epoch": 0.3793257835133,
      "grad_norm": 2.0169806480407715,
      "learning_rate": 3.104556228601528e-05,
      "loss": 2.0514,
      "step": 14403
    },
    {
      "epoch": 0.3793521200948117,
      "grad_norm": 2.0204029083251953,
      "learning_rate": 3.104424545693969e-05,
      "loss": 2.676,
      "step": 14404
    },
    {
      "epoch": 0.3793784566763234,
      "grad_norm": 3.0137548446655273,
      "learning_rate": 3.1042928627864104e-05,
      "loss": 1.4679,
      "step": 14405
    },
    {
      "epoch": 0.37940479325783516,
      "grad_norm": 1.8132253885269165,
      "learning_rate": 3.104161179878852e-05,
      "loss": 1.8828,
      "step": 14406
    },
    {
      "epoch": 0.37943112983934685,
      "grad_norm": 1.7056853771209717,
      "learning_rate": 3.1040294969712936e-05,
      "loss": 1.9261,
      "step": 14407
    },
    {
      "epoch": 0.3794574664208586,
      "grad_norm": 2.4139211177825928,
      "learning_rate": 3.103897814063735e-05,
      "loss": 2.2971,
      "step": 14408
    },
    {
      "epoch": 0.3794838030023703,
      "grad_norm": 2.2264411449432373,
      "learning_rate": 3.103766131156176e-05,
      "loss": 2.0354,
      "step": 14409
    },
    {
      "epoch": 0.379510139583882,
      "grad_norm": 5.448110580444336,
      "learning_rate": 3.1036344482486176e-05,
      "loss": 1.9249,
      "step": 14410
    },
    {
      "epoch": 0.3795364761653937,
      "grad_norm": 1.6863667964935303,
      "learning_rate": 3.1035027653410584e-05,
      "loss": 2.158,
      "step": 14411
    },
    {
      "epoch": 0.37956281274690545,
      "grad_norm": 2.6604042053222656,
      "learning_rate": 3.103371082433501e-05,
      "loss": 1.4999,
      "step": 14412
    },
    {
      "epoch": 0.3795891493284172,
      "grad_norm": 3.0340991020202637,
      "learning_rate": 3.1032393995259416e-05,
      "loss": 1.2997,
      "step": 14413
    },
    {
      "epoch": 0.3796154859099289,
      "grad_norm": 3.5236711502075195,
      "learning_rate": 3.103107716618383e-05,
      "loss": 0.4889,
      "step": 14414
    },
    {
      "epoch": 0.37964182249144063,
      "grad_norm": 2.264178514480591,
      "learning_rate": 3.102976033710825e-05,
      "loss": 1.979,
      "step": 14415
    },
    {
      "epoch": 0.3796681590729523,
      "grad_norm": 3.9308578968048096,
      "learning_rate": 3.1028443508032656e-05,
      "loss": 1.9012,
      "step": 14416
    },
    {
      "epoch": 0.37969449565446406,
      "grad_norm": 2.6817424297332764,
      "learning_rate": 3.102712667895708e-05,
      "loss": 2.2367,
      "step": 14417
    },
    {
      "epoch": 0.37972083223597575,
      "grad_norm": 3.8696348667144775,
      "learning_rate": 3.102580984988149e-05,
      "loss": 1.5145,
      "step": 14418
    },
    {
      "epoch": 0.3797471688174875,
      "grad_norm": 1.4997214078903198,
      "learning_rate": 3.10244930208059e-05,
      "loss": 2.2239,
      "step": 14419
    },
    {
      "epoch": 0.37977350539899923,
      "grad_norm": 3.1607394218444824,
      "learning_rate": 3.102317619173031e-05,
      "loss": 1.5356,
      "step": 14420
    },
    {
      "epoch": 0.3797998419805109,
      "grad_norm": 1.7745493650436401,
      "learning_rate": 3.1021859362654734e-05,
      "loss": 1.7118,
      "step": 14421
    },
    {
      "epoch": 0.37982617856202266,
      "grad_norm": 2.050464630126953,
      "learning_rate": 3.102054253357914e-05,
      "loss": 1.5729,
      "step": 14422
    },
    {
      "epoch": 0.37985251514353435,
      "grad_norm": 3.483598470687866,
      "learning_rate": 3.101922570450356e-05,
      "loss": 1.9902,
      "step": 14423
    },
    {
      "epoch": 0.3798788517250461,
      "grad_norm": 3.1781907081604004,
      "learning_rate": 3.101790887542797e-05,
      "loss": 2.0966,
      "step": 14424
    },
    {
      "epoch": 0.3799051883065578,
      "grad_norm": 2.0831942558288574,
      "learning_rate": 3.101659204635238e-05,
      "loss": 2.1314,
      "step": 14425
    },
    {
      "epoch": 0.3799315248880695,
      "grad_norm": 2.6782853603363037,
      "learning_rate": 3.10152752172768e-05,
      "loss": 1.7379,
      "step": 14426
    },
    {
      "epoch": 0.37995786146958127,
      "grad_norm": 4.894845485687256,
      "learning_rate": 3.1013958388201214e-05,
      "loss": 1.8873,
      "step": 14427
    },
    {
      "epoch": 0.37998419805109296,
      "grad_norm": 1.956839919090271,
      "learning_rate": 3.101264155912563e-05,
      "loss": 1.4455,
      "step": 14428
    },
    {
      "epoch": 0.3800105346326047,
      "grad_norm": 4.308040142059326,
      "learning_rate": 3.101132473005004e-05,
      "loss": 2.4415,
      "step": 14429
    },
    {
      "epoch": 0.3800368712141164,
      "grad_norm": 2.0065255165100098,
      "learning_rate": 3.1010007900974454e-05,
      "loss": 1.9858,
      "step": 14430
    },
    {
      "epoch": 0.38006320779562813,
      "grad_norm": 2.4225807189941406,
      "learning_rate": 3.100869107189887e-05,
      "loss": 1.9079,
      "step": 14431
    },
    {
      "epoch": 0.3800895443771398,
      "grad_norm": 3.417452335357666,
      "learning_rate": 3.1007374242823285e-05,
      "loss": 1.0014,
      "step": 14432
    },
    {
      "epoch": 0.38011588095865156,
      "grad_norm": 2.403873920440674,
      "learning_rate": 3.1006057413747694e-05,
      "loss": 0.5602,
      "step": 14433
    },
    {
      "epoch": 0.3801422175401633,
      "grad_norm": 1.9683886766433716,
      "learning_rate": 3.100474058467211e-05,
      "loss": 1.8379,
      "step": 14434
    },
    {
      "epoch": 0.380168554121675,
      "grad_norm": 1.9898672103881836,
      "learning_rate": 3.1003423755596525e-05,
      "loss": 1.2432,
      "step": 14435
    },
    {
      "epoch": 0.38019489070318674,
      "grad_norm": 2.5677411556243896,
      "learning_rate": 3.100210692652094e-05,
      "loss": 1.6397,
      "step": 14436
    },
    {
      "epoch": 0.3802212272846984,
      "grad_norm": 4.034700393676758,
      "learning_rate": 3.1000790097445356e-05,
      "loss": 1.8715,
      "step": 14437
    },
    {
      "epoch": 0.38024756386621017,
      "grad_norm": 1.8698471784591675,
      "learning_rate": 3.0999473268369765e-05,
      "loss": 2.2374,
      "step": 14438
    },
    {
      "epoch": 0.3802739004477219,
      "grad_norm": 2.496015787124634,
      "learning_rate": 3.099815643929418e-05,
      "loss": 1.8703,
      "step": 14439
    },
    {
      "epoch": 0.3803002370292336,
      "grad_norm": 2.114288568496704,
      "learning_rate": 3.0996839610218596e-05,
      "loss": 1.7906,
      "step": 14440
    },
    {
      "epoch": 0.38032657361074534,
      "grad_norm": 1.9222286939620972,
      "learning_rate": 3.099552278114301e-05,
      "loss": 2.1159,
      "step": 14441
    },
    {
      "epoch": 0.38035291019225703,
      "grad_norm": 2.053709030151367,
      "learning_rate": 3.099420595206742e-05,
      "loss": 2.1609,
      "step": 14442
    },
    {
      "epoch": 0.3803792467737688,
      "grad_norm": 1.8872405290603638,
      "learning_rate": 3.0992889122991836e-05,
      "loss": 1.2362,
      "step": 14443
    },
    {
      "epoch": 0.38040558335528046,
      "grad_norm": 2.4812612533569336,
      "learning_rate": 3.099157229391625e-05,
      "loss": 2.1593,
      "step": 14444
    },
    {
      "epoch": 0.3804319199367922,
      "grad_norm": 3.93999981880188,
      "learning_rate": 3.099025546484067e-05,
      "loss": 1.0586,
      "step": 14445
    },
    {
      "epoch": 0.38045825651830395,
      "grad_norm": 1.6276928186416626,
      "learning_rate": 3.098893863576508e-05,
      "loss": 1.3704,
      "step": 14446
    },
    {
      "epoch": 0.38048459309981564,
      "grad_norm": 3.938692331314087,
      "learning_rate": 3.098762180668949e-05,
      "loss": 1.4437,
      "step": 14447
    },
    {
      "epoch": 0.3805109296813274,
      "grad_norm": 1.9843577146530151,
      "learning_rate": 3.098630497761391e-05,
      "loss": 1.7513,
      "step": 14448
    },
    {
      "epoch": 0.38053726626283907,
      "grad_norm": 2.6324470043182373,
      "learning_rate": 3.0984988148538316e-05,
      "loss": 1.0439,
      "step": 14449
    },
    {
      "epoch": 0.3805636028443508,
      "grad_norm": 4.131014347076416,
      "learning_rate": 3.098367131946274e-05,
      "loss": 1.7271,
      "step": 14450
    },
    {
      "epoch": 0.3805899394258625,
      "grad_norm": 1.9271693229675293,
      "learning_rate": 3.098235449038715e-05,
      "loss": 2.5178,
      "step": 14451
    },
    {
      "epoch": 0.38061627600737424,
      "grad_norm": 1.9114900827407837,
      "learning_rate": 3.098103766131156e-05,
      "loss": 2.3694,
      "step": 14452
    },
    {
      "epoch": 0.380642612588886,
      "grad_norm": 2.3361005783081055,
      "learning_rate": 3.097972083223598e-05,
      "loss": 2.1915,
      "step": 14453
    },
    {
      "epoch": 0.3806689491703977,
      "grad_norm": 2.273688554763794,
      "learning_rate": 3.0978404003160394e-05,
      "loss": 0.9563,
      "step": 14454
    },
    {
      "epoch": 0.3806952857519094,
      "grad_norm": 1.5529550313949585,
      "learning_rate": 3.097708717408481e-05,
      "loss": 1.7367,
      "step": 14455
    },
    {
      "epoch": 0.3807216223334211,
      "grad_norm": 2.8191428184509277,
      "learning_rate": 3.097577034500922e-05,
      "loss": 1.2118,
      "step": 14456
    },
    {
      "epoch": 0.38074795891493285,
      "grad_norm": 1.668314814567566,
      "learning_rate": 3.0974453515933634e-05,
      "loss": 1.6507,
      "step": 14457
    },
    {
      "epoch": 0.38077429549644454,
      "grad_norm": 2.7240145206451416,
      "learning_rate": 3.097313668685804e-05,
      "loss": 1.5051,
      "step": 14458
    },
    {
      "epoch": 0.3808006320779563,
      "grad_norm": 3.0239665508270264,
      "learning_rate": 3.0971819857782465e-05,
      "loss": 1.5282,
      "step": 14459
    },
    {
      "epoch": 0.380826968659468,
      "grad_norm": 2.5275745391845703,
      "learning_rate": 3.0970503028706874e-05,
      "loss": 1.6858,
      "step": 14460
    },
    {
      "epoch": 0.3808533052409797,
      "grad_norm": 2.205512762069702,
      "learning_rate": 3.096918619963129e-05,
      "loss": 1.484,
      "step": 14461
    },
    {
      "epoch": 0.38087964182249145,
      "grad_norm": 4.681382179260254,
      "learning_rate": 3.0967869370555705e-05,
      "loss": 1.3491,
      "step": 14462
    },
    {
      "epoch": 0.38090597840400314,
      "grad_norm": 2.1899831295013428,
      "learning_rate": 3.0966552541480114e-05,
      "loss": 1.7056,
      "step": 14463
    },
    {
      "epoch": 0.3809323149855149,
      "grad_norm": 9.964446067810059,
      "learning_rate": 3.0965235712404537e-05,
      "loss": 1.6069,
      "step": 14464
    },
    {
      "epoch": 0.3809586515670266,
      "grad_norm": 1.5020250082015991,
      "learning_rate": 3.0963918883328945e-05,
      "loss": 1.6391,
      "step": 14465
    },
    {
      "epoch": 0.3809849881485383,
      "grad_norm": 1.814849615097046,
      "learning_rate": 3.096260205425336e-05,
      "loss": 2.075,
      "step": 14466
    },
    {
      "epoch": 0.38101132473005006,
      "grad_norm": 2.910660743713379,
      "learning_rate": 3.096128522517777e-05,
      "loss": 0.4715,
      "step": 14467
    },
    {
      "epoch": 0.38103766131156175,
      "grad_norm": 2.3053295612335205,
      "learning_rate": 3.0959968396102185e-05,
      "loss": 1.8975,
      "step": 14468
    },
    {
      "epoch": 0.3810639978930735,
      "grad_norm": 3.6511478424072266,
      "learning_rate": 3.09586515670266e-05,
      "loss": 0.8469,
      "step": 14469
    },
    {
      "epoch": 0.3810903344745852,
      "grad_norm": 5.193423748016357,
      "learning_rate": 3.0957334737951017e-05,
      "loss": 2.223,
      "step": 14470
    },
    {
      "epoch": 0.3811166710560969,
      "grad_norm": 2.920355796813965,
      "learning_rate": 3.0956017908875425e-05,
      "loss": 1.4857,
      "step": 14471
    },
    {
      "epoch": 0.3811430076376086,
      "grad_norm": 2.167603015899658,
      "learning_rate": 3.095470107979984e-05,
      "loss": 1.2876,
      "step": 14472
    },
    {
      "epoch": 0.38116934421912035,
      "grad_norm": 2.236863851547241,
      "learning_rate": 3.0953384250724257e-05,
      "loss": 2.157,
      "step": 14473
    },
    {
      "epoch": 0.3811956808006321,
      "grad_norm": 1.8141050338745117,
      "learning_rate": 3.095206742164867e-05,
      "loss": 2.1034,
      "step": 14474
    },
    {
      "epoch": 0.3812220173821438,
      "grad_norm": 2.4759411811828613,
      "learning_rate": 3.095075059257309e-05,
      "loss": 1.2742,
      "step": 14475
    },
    {
      "epoch": 0.38124835396365553,
      "grad_norm": 2.4107723236083984,
      "learning_rate": 3.0949433763497497e-05,
      "loss": 1.7625,
      "step": 14476
    },
    {
      "epoch": 0.3812746905451672,
      "grad_norm": 2.3482730388641357,
      "learning_rate": 3.094811693442191e-05,
      "loss": 0.7068,
      "step": 14477
    },
    {
      "epoch": 0.38130102712667896,
      "grad_norm": 2.733046770095825,
      "learning_rate": 3.094680010534633e-05,
      "loss": 2.1464,
      "step": 14478
    },
    {
      "epoch": 0.3813273637081907,
      "grad_norm": 1.857745885848999,
      "learning_rate": 3.0945483276270743e-05,
      "loss": 1.406,
      "step": 14479
    },
    {
      "epoch": 0.3813537002897024,
      "grad_norm": 1.3975458145141602,
      "learning_rate": 3.094416644719515e-05,
      "loss": 2.0023,
      "step": 14480
    },
    {
      "epoch": 0.38138003687121413,
      "grad_norm": 3.0195488929748535,
      "learning_rate": 3.094284961811957e-05,
      "loss": 0.9969,
      "step": 14481
    },
    {
      "epoch": 0.3814063734527258,
      "grad_norm": 2.890155076980591,
      "learning_rate": 3.0941532789043983e-05,
      "loss": 0.9167,
      "step": 14482
    },
    {
      "epoch": 0.38143271003423757,
      "grad_norm": 2.8606345653533936,
      "learning_rate": 3.09402159599684e-05,
      "loss": 1.456,
      "step": 14483
    },
    {
      "epoch": 0.38145904661574925,
      "grad_norm": 1.6089520454406738,
      "learning_rate": 3.0938899130892815e-05,
      "loss": 2.4486,
      "step": 14484
    },
    {
      "epoch": 0.381485383197261,
      "grad_norm": 2.385244846343994,
      "learning_rate": 3.0937582301817223e-05,
      "loss": 0.2576,
      "step": 14485
    },
    {
      "epoch": 0.38151171977877274,
      "grad_norm": 2.513944625854492,
      "learning_rate": 3.093626547274164e-05,
      "loss": 1.2196,
      "step": 14486
    },
    {
      "epoch": 0.3815380563602844,
      "grad_norm": 4.3541364669799805,
      "learning_rate": 3.0934948643666055e-05,
      "loss": 2.4994,
      "step": 14487
    },
    {
      "epoch": 0.38156439294179617,
      "grad_norm": 2.828737258911133,
      "learning_rate": 3.093363181459047e-05,
      "loss": 1.1445,
      "step": 14488
    },
    {
      "epoch": 0.38159072952330786,
      "grad_norm": 1.66840660572052,
      "learning_rate": 3.093231498551488e-05,
      "loss": 1.6256,
      "step": 14489
    },
    {
      "epoch": 0.3816170661048196,
      "grad_norm": 5.299621105194092,
      "learning_rate": 3.0930998156439295e-05,
      "loss": 2.0904,
      "step": 14490
    },
    {
      "epoch": 0.3816434026863313,
      "grad_norm": 1.7058268785476685,
      "learning_rate": 3.092968132736371e-05,
      "loss": 1.6761,
      "step": 14491
    },
    {
      "epoch": 0.38166973926784303,
      "grad_norm": 2.0077342987060547,
      "learning_rate": 3.0928364498288126e-05,
      "loss": 2.0041,
      "step": 14492
    },
    {
      "epoch": 0.3816960758493548,
      "grad_norm": 1.6694526672363281,
      "learning_rate": 3.092704766921254e-05,
      "loss": 2.3546,
      "step": 14493
    },
    {
      "epoch": 0.38172241243086646,
      "grad_norm": 2.921759605407715,
      "learning_rate": 3.092573084013695e-05,
      "loss": 1.4905,
      "step": 14494
    },
    {
      "epoch": 0.3817487490123782,
      "grad_norm": 1.6434262990951538,
      "learning_rate": 3.0924414011061366e-05,
      "loss": 1.5862,
      "step": 14495
    },
    {
      "epoch": 0.3817750855938899,
      "grad_norm": 4.454122543334961,
      "learning_rate": 3.0923097181985775e-05,
      "loss": 1.3625,
      "step": 14496
    },
    {
      "epoch": 0.38180142217540164,
      "grad_norm": 1.860121726989746,
      "learning_rate": 3.09217803529102e-05,
      "loss": 0.3075,
      "step": 14497
    },
    {
      "epoch": 0.3818277587569133,
      "grad_norm": 1.612991213798523,
      "learning_rate": 3.0920463523834606e-05,
      "loss": 1.9426,
      "step": 14498
    },
    {
      "epoch": 0.38185409533842507,
      "grad_norm": 2.6928908824920654,
      "learning_rate": 3.091914669475902e-05,
      "loss": 1.6539,
      "step": 14499
    },
    {
      "epoch": 0.3818804319199368,
      "grad_norm": 1.951027274131775,
      "learning_rate": 3.091782986568344e-05,
      "loss": 1.8509,
      "step": 14500
    },
    {
      "epoch": 0.3819067685014485,
      "grad_norm": 5.054205894470215,
      "learning_rate": 3.0916513036607846e-05,
      "loss": 0.9297,
      "step": 14501
    },
    {
      "epoch": 0.38193310508296024,
      "grad_norm": 2.7478649616241455,
      "learning_rate": 3.091519620753227e-05,
      "loss": 2.3899,
      "step": 14502
    },
    {
      "epoch": 0.38195944166447193,
      "grad_norm": 2.4717133045196533,
      "learning_rate": 3.091387937845668e-05,
      "loss": 1.8558,
      "step": 14503
    },
    {
      "epoch": 0.3819857782459837,
      "grad_norm": 2.6459312438964844,
      "learning_rate": 3.091256254938109e-05,
      "loss": 1.8517,
      "step": 14504
    },
    {
      "epoch": 0.38201211482749536,
      "grad_norm": 2.268254041671753,
      "learning_rate": 3.09112457203055e-05,
      "loss": 2.0417,
      "step": 14505
    },
    {
      "epoch": 0.3820384514090071,
      "grad_norm": 2.6595206260681152,
      "learning_rate": 3.0909928891229924e-05,
      "loss": 0.627,
      "step": 14506
    },
    {
      "epoch": 0.38206478799051885,
      "grad_norm": 1.9094496965408325,
      "learning_rate": 3.090861206215433e-05,
      "loss": 1.8996,
      "step": 14507
    },
    {
      "epoch": 0.38209112457203054,
      "grad_norm": 3.0797224044799805,
      "learning_rate": 3.090729523307875e-05,
      "loss": 2.0929,
      "step": 14508
    },
    {
      "epoch": 0.3821174611535423,
      "grad_norm": 2.0068109035491943,
      "learning_rate": 3.0905978404003164e-05,
      "loss": 0.379,
      "step": 14509
    },
    {
      "epoch": 0.38214379773505397,
      "grad_norm": 5.691693305969238,
      "learning_rate": 3.090466157492757e-05,
      "loss": 2.0054,
      "step": 14510
    },
    {
      "epoch": 0.3821701343165657,
      "grad_norm": 2.9892454147338867,
      "learning_rate": 3.0903344745851995e-05,
      "loss": 0.8063,
      "step": 14511
    },
    {
      "epoch": 0.38219647089807746,
      "grad_norm": 2.283273220062256,
      "learning_rate": 3.0902027916776404e-05,
      "loss": 0.6373,
      "step": 14512
    },
    {
      "epoch": 0.38222280747958914,
      "grad_norm": 1.8923076391220093,
      "learning_rate": 3.090071108770082e-05,
      "loss": 1.8576,
      "step": 14513
    },
    {
      "epoch": 0.3822491440611009,
      "grad_norm": 1.8745142221450806,
      "learning_rate": 3.089939425862523e-05,
      "loss": 1.583,
      "step": 14514
    },
    {
      "epoch": 0.3822754806426126,
      "grad_norm": 1.8707177639007568,
      "learning_rate": 3.0898077429549644e-05,
      "loss": 2.426,
      "step": 14515
    },
    {
      "epoch": 0.3823018172241243,
      "grad_norm": 4.815286636352539,
      "learning_rate": 3.089676060047406e-05,
      "loss": 1.4576,
      "step": 14516
    },
    {
      "epoch": 0.382328153805636,
      "grad_norm": 1.6012474298477173,
      "learning_rate": 3.0895443771398475e-05,
      "loss": 2.2897,
      "step": 14517
    },
    {
      "epoch": 0.38235449038714775,
      "grad_norm": 2.140408515930176,
      "learning_rate": 3.089412694232289e-05,
      "loss": 2.4381,
      "step": 14518
    },
    {
      "epoch": 0.3823808269686595,
      "grad_norm": 2.8224799633026123,
      "learning_rate": 3.08928101132473e-05,
      "loss": 1.1844,
      "step": 14519
    },
    {
      "epoch": 0.3824071635501712,
      "grad_norm": 1.8939311504364014,
      "learning_rate": 3.089149328417172e-05,
      "loss": 1.8609,
      "step": 14520
    },
    {
      "epoch": 0.3824335001316829,
      "grad_norm": 1.5587173700332642,
      "learning_rate": 3.089017645509613e-05,
      "loss": 1.1591,
      "step": 14521
    },
    {
      "epoch": 0.3824598367131946,
      "grad_norm": 1.8770238161087036,
      "learning_rate": 3.0888859626020546e-05,
      "loss": 2.3402,
      "step": 14522
    },
    {
      "epoch": 0.38248617329470636,
      "grad_norm": 2.0927324295043945,
      "learning_rate": 3.0887542796944955e-05,
      "loss": 1.9234,
      "step": 14523
    },
    {
      "epoch": 0.38251250987621804,
      "grad_norm": 3.052306890487671,
      "learning_rate": 3.088622596786937e-05,
      "loss": 1.9123,
      "step": 14524
    },
    {
      "epoch": 0.3825388464577298,
      "grad_norm": 1.836662769317627,
      "learning_rate": 3.0884909138793786e-05,
      "loss": 1.8598,
      "step": 14525
    },
    {
      "epoch": 0.38256518303924153,
      "grad_norm": 1.8895927667617798,
      "learning_rate": 3.08835923097182e-05,
      "loss": 1.899,
      "step": 14526
    },
    {
      "epoch": 0.3825915196207532,
      "grad_norm": 1.8681902885437012,
      "learning_rate": 3.088227548064261e-05,
      "loss": 1.6546,
      "step": 14527
    },
    {
      "epoch": 0.38261785620226496,
      "grad_norm": 5.182409286499023,
      "learning_rate": 3.0880958651567026e-05,
      "loss": 1.2162,
      "step": 14528
    },
    {
      "epoch": 0.38264419278377665,
      "grad_norm": 1.9442241191864014,
      "learning_rate": 3.087964182249144e-05,
      "loss": 1.8905,
      "step": 14529
    },
    {
      "epoch": 0.3826705293652884,
      "grad_norm": 2.077543020248413,
      "learning_rate": 3.087832499341586e-05,
      "loss": 2.0442,
      "step": 14530
    },
    {
      "epoch": 0.3826968659468001,
      "grad_norm": 1.8882946968078613,
      "learning_rate": 3.087700816434027e-05,
      "loss": 1.992,
      "step": 14531
    },
    {
      "epoch": 0.3827232025283118,
      "grad_norm": 1.9640634059906006,
      "learning_rate": 3.087569133526468e-05,
      "loss": 1.8207,
      "step": 14532
    },
    {
      "epoch": 0.38274953910982357,
      "grad_norm": 1.5785528421401978,
      "learning_rate": 3.08743745061891e-05,
      "loss": 2.1099,
      "step": 14533
    },
    {
      "epoch": 0.38277587569133525,
      "grad_norm": 2.320997714996338,
      "learning_rate": 3.0873057677113506e-05,
      "loss": 1.9014,
      "step": 14534
    },
    {
      "epoch": 0.382802212272847,
      "grad_norm": 4.660298824310303,
      "learning_rate": 3.087174084803793e-05,
      "loss": 0.8468,
      "step": 14535
    },
    {
      "epoch": 0.3828285488543587,
      "grad_norm": 3.4066030979156494,
      "learning_rate": 3.087042401896234e-05,
      "loss": 2.2176,
      "step": 14536
    },
    {
      "epoch": 0.38285488543587043,
      "grad_norm": 2.178804874420166,
      "learning_rate": 3.086910718988675e-05,
      "loss": 2.3646,
      "step": 14537
    },
    {
      "epoch": 0.3828812220173821,
      "grad_norm": 3.9844250679016113,
      "learning_rate": 3.086779036081117e-05,
      "loss": 1.2117,
      "step": 14538
    },
    {
      "epoch": 0.38290755859889386,
      "grad_norm": 1.9921941757202148,
      "learning_rate": 3.0866473531735584e-05,
      "loss": 2.7739,
      "step": 14539
    },
    {
      "epoch": 0.3829338951804056,
      "grad_norm": 1.6976536512374878,
      "learning_rate": 3.086515670266e-05,
      "loss": 1.8513,
      "step": 14540
    },
    {
      "epoch": 0.3829602317619173,
      "grad_norm": 1.7839652299880981,
      "learning_rate": 3.086383987358441e-05,
      "loss": 1.8555,
      "step": 14541
    },
    {
      "epoch": 0.38298656834342903,
      "grad_norm": 1.7729908227920532,
      "learning_rate": 3.0862523044508824e-05,
      "loss": 1.3811,
      "step": 14542
    },
    {
      "epoch": 0.3830129049249407,
      "grad_norm": 1.4872736930847168,
      "learning_rate": 3.086120621543323e-05,
      "loss": 1.8093,
      "step": 14543
    },
    {
      "epoch": 0.38303924150645247,
      "grad_norm": 2.047466278076172,
      "learning_rate": 3.0859889386357656e-05,
      "loss": 1.0088,
      "step": 14544
    },
    {
      "epoch": 0.3830655780879642,
      "grad_norm": 1.874760389328003,
      "learning_rate": 3.0858572557282064e-05,
      "loss": 1.8577,
      "step": 14545
    },
    {
      "epoch": 0.3830919146694759,
      "grad_norm": 2.512969970703125,
      "learning_rate": 3.085725572820648e-05,
      "loss": 1.8125,
      "step": 14546
    },
    {
      "epoch": 0.38311825125098764,
      "grad_norm": 1.4420921802520752,
      "learning_rate": 3.0855938899130896e-05,
      "loss": 1.9235,
      "step": 14547
    },
    {
      "epoch": 0.38314458783249933,
      "grad_norm": 3.0532658100128174,
      "learning_rate": 3.0854622070055304e-05,
      "loss": 0.887,
      "step": 14548
    },
    {
      "epoch": 0.38317092441401107,
      "grad_norm": 3.250452756881714,
      "learning_rate": 3.085330524097973e-05,
      "loss": 0.9619,
      "step": 14549
    },
    {
      "epoch": 0.38319726099552276,
      "grad_norm": 3.1377077102661133,
      "learning_rate": 3.0851988411904136e-05,
      "loss": 1.3583,
      "step": 14550
    },
    {
      "epoch": 0.3832235975770345,
      "grad_norm": 2.734541416168213,
      "learning_rate": 3.085067158282855e-05,
      "loss": 0.7901,
      "step": 14551
    },
    {
      "epoch": 0.38324993415854625,
      "grad_norm": 2.777575731277466,
      "learning_rate": 3.084935475375296e-05,
      "loss": 1.5313,
      "step": 14552
    },
    {
      "epoch": 0.38327627074005793,
      "grad_norm": 3.8294436931610107,
      "learning_rate": 3.084803792467738e-05,
      "loss": 1.7996,
      "step": 14553
    },
    {
      "epoch": 0.3833026073215697,
      "grad_norm": 2.3994832038879395,
      "learning_rate": 3.084672109560179e-05,
      "loss": 2.3727,
      "step": 14554
    },
    {
      "epoch": 0.38332894390308137,
      "grad_norm": 1.6705495119094849,
      "learning_rate": 3.084540426652621e-05,
      "loss": 1.6402,
      "step": 14555
    },
    {
      "epoch": 0.3833552804845931,
      "grad_norm": 3.5592119693756104,
      "learning_rate": 3.084408743745062e-05,
      "loss": 0.967,
      "step": 14556
    },
    {
      "epoch": 0.3833816170661048,
      "grad_norm": 1.7447296380996704,
      "learning_rate": 3.084277060837503e-05,
      "loss": 1.5127,
      "step": 14557
    },
    {
      "epoch": 0.38340795364761654,
      "grad_norm": 2.0983502864837646,
      "learning_rate": 3.0841453779299454e-05,
      "loss": 2.4396,
      "step": 14558
    },
    {
      "epoch": 0.3834342902291283,
      "grad_norm": 1.859695553779602,
      "learning_rate": 3.084013695022386e-05,
      "loss": 2.2114,
      "step": 14559
    },
    {
      "epoch": 0.38346062681063997,
      "grad_norm": 1.955888032913208,
      "learning_rate": 3.083882012114828e-05,
      "loss": 1.519,
      "step": 14560
    },
    {
      "epoch": 0.3834869633921517,
      "grad_norm": 4.932207107543945,
      "learning_rate": 3.083750329207269e-05,
      "loss": 1.5375,
      "step": 14561
    },
    {
      "epoch": 0.3835132999736634,
      "grad_norm": 3.0014209747314453,
      "learning_rate": 3.08361864629971e-05,
      "loss": 1.5491,
      "step": 14562
    },
    {
      "epoch": 0.38353963655517515,
      "grad_norm": 1.3994399309158325,
      "learning_rate": 3.083486963392152e-05,
      "loss": 1.6835,
      "step": 14563
    },
    {
      "epoch": 0.38356597313668683,
      "grad_norm": 2.186720371246338,
      "learning_rate": 3.0833552804845934e-05,
      "loss": 1.7745,
      "step": 14564
    },
    {
      "epoch": 0.3835923097181986,
      "grad_norm": 1.9380356073379517,
      "learning_rate": 3.083223597577035e-05,
      "loss": 1.8975,
      "step": 14565
    },
    {
      "epoch": 0.3836186462997103,
      "grad_norm": 3.60066819190979,
      "learning_rate": 3.083091914669476e-05,
      "loss": 1.1451,
      "step": 14566
    },
    {
      "epoch": 0.383644982881222,
      "grad_norm": 1.6131770610809326,
      "learning_rate": 3.082960231761918e-05,
      "loss": 1.4917,
      "step": 14567
    },
    {
      "epoch": 0.38367131946273375,
      "grad_norm": 2.484441041946411,
      "learning_rate": 3.082828548854359e-05,
      "loss": 0.8565,
      "step": 14568
    },
    {
      "epoch": 0.38369765604424544,
      "grad_norm": 1.8261693716049194,
      "learning_rate": 3.0826968659468005e-05,
      "loss": 1.2457,
      "step": 14569
    },
    {
      "epoch": 0.3837239926257572,
      "grad_norm": 2.0360586643218994,
      "learning_rate": 3.0825651830392414e-05,
      "loss": 1.4147,
      "step": 14570
    },
    {
      "epoch": 0.38375032920726887,
      "grad_norm": 4.179394245147705,
      "learning_rate": 3.082433500131683e-05,
      "loss": 1.6205,
      "step": 14571
    },
    {
      "epoch": 0.3837766657887806,
      "grad_norm": 3.0943753719329834,
      "learning_rate": 3.0823018172241245e-05,
      "loss": 1.467,
      "step": 14572
    },
    {
      "epoch": 0.38380300237029236,
      "grad_norm": 2.677536725997925,
      "learning_rate": 3.082170134316566e-05,
      "loss": 1.583,
      "step": 14573
    },
    {
      "epoch": 0.38382933895180404,
      "grad_norm": 2.0822668075561523,
      "learning_rate": 3.082038451409007e-05,
      "loss": 1.9005,
      "step": 14574
    },
    {
      "epoch": 0.3838556755333158,
      "grad_norm": 6.708086967468262,
      "learning_rate": 3.0819067685014485e-05,
      "loss": 1.8824,
      "step": 14575
    },
    {
      "epoch": 0.3838820121148275,
      "grad_norm": 2.0805656909942627,
      "learning_rate": 3.08177508559389e-05,
      "loss": 1.5773,
      "step": 14576
    },
    {
      "epoch": 0.3839083486963392,
      "grad_norm": 2.1125094890594482,
      "learning_rate": 3.0816434026863316e-05,
      "loss": 1.4737,
      "step": 14577
    },
    {
      "epoch": 0.38393468527785096,
      "grad_norm": 1.682759404182434,
      "learning_rate": 3.081511719778773e-05,
      "loss": 1.7282,
      "step": 14578
    },
    {
      "epoch": 0.38396102185936265,
      "grad_norm": 2.025913715362549,
      "learning_rate": 3.081380036871214e-05,
      "loss": 1.4737,
      "step": 14579
    },
    {
      "epoch": 0.3839873584408744,
      "grad_norm": 4.8360795974731445,
      "learning_rate": 3.0812483539636556e-05,
      "loss": 1.2927,
      "step": 14580
    },
    {
      "epoch": 0.3840136950223861,
      "grad_norm": 1.5795741081237793,
      "learning_rate": 3.0811166710560965e-05,
      "loss": 1.8777,
      "step": 14581
    },
    {
      "epoch": 0.3840400316038978,
      "grad_norm": 3.2506656646728516,
      "learning_rate": 3.080984988148539e-05,
      "loss": 0.5757,
      "step": 14582
    },
    {
      "epoch": 0.3840663681854095,
      "grad_norm": 1.514147162437439,
      "learning_rate": 3.0808533052409796e-05,
      "loss": 2.1856,
      "step": 14583
    },
    {
      "epoch": 0.38409270476692126,
      "grad_norm": 2.3269248008728027,
      "learning_rate": 3.080721622333421e-05,
      "loss": 1.1516,
      "step": 14584
    },
    {
      "epoch": 0.384119041348433,
      "grad_norm": 3.3791117668151855,
      "learning_rate": 3.080589939425863e-05,
      "loss": 1.2936,
      "step": 14585
    },
    {
      "epoch": 0.3841453779299447,
      "grad_norm": 2.661522626876831,
      "learning_rate": 3.080458256518304e-05,
      "loss": 2.1369,
      "step": 14586
    },
    {
      "epoch": 0.38417171451145643,
      "grad_norm": 2.0250821113586426,
      "learning_rate": 3.080326573610746e-05,
      "loss": 1.6623,
      "step": 14587
    },
    {
      "epoch": 0.3841980510929681,
      "grad_norm": 3.6315219402313232,
      "learning_rate": 3.080194890703187e-05,
      "loss": 0.7954,
      "step": 14588
    },
    {
      "epoch": 0.38422438767447986,
      "grad_norm": 3.226318597793579,
      "learning_rate": 3.080063207795628e-05,
      "loss": 1.4729,
      "step": 14589
    },
    {
      "epoch": 0.38425072425599155,
      "grad_norm": 2.603069543838501,
      "learning_rate": 3.079931524888069e-05,
      "loss": 1.193,
      "step": 14590
    },
    {
      "epoch": 0.3842770608375033,
      "grad_norm": 1.8471455574035645,
      "learning_rate": 3.0797998419805114e-05,
      "loss": 1.6258,
      "step": 14591
    },
    {
      "epoch": 0.38430339741901504,
      "grad_norm": 3.5210866928100586,
      "learning_rate": 3.079668159072952e-05,
      "loss": 2.4627,
      "step": 14592
    },
    {
      "epoch": 0.3843297340005267,
      "grad_norm": 3.6657567024230957,
      "learning_rate": 3.079536476165394e-05,
      "loss": 1.4888,
      "step": 14593
    },
    {
      "epoch": 0.38435607058203847,
      "grad_norm": 1.986794352531433,
      "learning_rate": 3.0794047932578354e-05,
      "loss": 1.442,
      "step": 14594
    },
    {
      "epoch": 0.38438240716355015,
      "grad_norm": 1.9501502513885498,
      "learning_rate": 3.079273110350276e-05,
      "loss": 2.0969,
      "step": 14595
    },
    {
      "epoch": 0.3844087437450619,
      "grad_norm": 2.129408121109009,
      "learning_rate": 3.0791414274427185e-05,
      "loss": 0.4437,
      "step": 14596
    },
    {
      "epoch": 0.3844350803265736,
      "grad_norm": 1.6300894021987915,
      "learning_rate": 3.0790097445351594e-05,
      "loss": 0.6411,
      "step": 14597
    },
    {
      "epoch": 0.38446141690808533,
      "grad_norm": 1.8381468057632446,
      "learning_rate": 3.078878061627601e-05,
      "loss": 1.7314,
      "step": 14598
    },
    {
      "epoch": 0.3844877534895971,
      "grad_norm": 1.5415160655975342,
      "learning_rate": 3.078746378720042e-05,
      "loss": 1.8967,
      "step": 14599
    },
    {
      "epoch": 0.38451409007110876,
      "grad_norm": 3.338806629180908,
      "learning_rate": 3.078614695812484e-05,
      "loss": 1.8835,
      "step": 14600
    },
    {
      "epoch": 0.3845404266526205,
      "grad_norm": 3.8954193592071533,
      "learning_rate": 3.078483012904925e-05,
      "loss": 1.7467,
      "step": 14601
    },
    {
      "epoch": 0.3845667632341322,
      "grad_norm": 4.459294319152832,
      "learning_rate": 3.0783513299973665e-05,
      "loss": 0.528,
      "step": 14602
    },
    {
      "epoch": 0.38459309981564394,
      "grad_norm": 1.9710445404052734,
      "learning_rate": 3.078219647089808e-05,
      "loss": 2.3077,
      "step": 14603
    },
    {
      "epoch": 0.3846194363971556,
      "grad_norm": 3.190796375274658,
      "learning_rate": 3.078087964182249e-05,
      "loss": 1.3843,
      "step": 14604
    },
    {
      "epoch": 0.38464577297866737,
      "grad_norm": 1.8104718923568726,
      "learning_rate": 3.077956281274691e-05,
      "loss": 1.4553,
      "step": 14605
    },
    {
      "epoch": 0.3846721095601791,
      "grad_norm": 4.741323947906494,
      "learning_rate": 3.077824598367132e-05,
      "loss": 1.6188,
      "step": 14606
    },
    {
      "epoch": 0.3846984461416908,
      "grad_norm": 1.5914028882980347,
      "learning_rate": 3.077692915459574e-05,
      "loss": 1.8913,
      "step": 14607
    },
    {
      "epoch": 0.38472478272320254,
      "grad_norm": 3.255990982055664,
      "learning_rate": 3.0775612325520145e-05,
      "loss": 0.7662,
      "step": 14608
    },
    {
      "epoch": 0.38475111930471423,
      "grad_norm": 2.700408697128296,
      "learning_rate": 3.077429549644456e-05,
      "loss": 2.073,
      "step": 14609
    },
    {
      "epoch": 0.38477745588622597,
      "grad_norm": 2.3465452194213867,
      "learning_rate": 3.077297866736898e-05,
      "loss": 1.8965,
      "step": 14610
    },
    {
      "epoch": 0.3848037924677377,
      "grad_norm": 2.544355869293213,
      "learning_rate": 3.077166183829339e-05,
      "loss": 1.035,
      "step": 14611
    },
    {
      "epoch": 0.3848301290492494,
      "grad_norm": 3.2025418281555176,
      "learning_rate": 3.077034500921781e-05,
      "loss": 1.8649,
      "step": 14612
    },
    {
      "epoch": 0.38485646563076115,
      "grad_norm": 5.358971118927002,
      "learning_rate": 3.076902818014222e-05,
      "loss": 1.8094,
      "step": 14613
    },
    {
      "epoch": 0.38488280221227283,
      "grad_norm": 1.4875799417495728,
      "learning_rate": 3.076771135106663e-05,
      "loss": 1.9678,
      "step": 14614
    },
    {
      "epoch": 0.3849091387937846,
      "grad_norm": 2.4168639183044434,
      "learning_rate": 3.076639452199105e-05,
      "loss": 2.0317,
      "step": 14615
    },
    {
      "epoch": 0.38493547537529627,
      "grad_norm": 1.462188482284546,
      "learning_rate": 3.0765077692915463e-05,
      "loss": 2.0147,
      "step": 14616
    },
    {
      "epoch": 0.384961811956808,
      "grad_norm": 9.157552719116211,
      "learning_rate": 3.076376086383987e-05,
      "loss": 2.6409,
      "step": 14617
    },
    {
      "epoch": 0.38498814853831975,
      "grad_norm": 2.000145196914673,
      "learning_rate": 3.076244403476429e-05,
      "loss": 2.059,
      "step": 14618
    },
    {
      "epoch": 0.38501448511983144,
      "grad_norm": 4.788470268249512,
      "learning_rate": 3.0761127205688703e-05,
      "loss": 1.9909,
      "step": 14619
    },
    {
      "epoch": 0.3850408217013432,
      "grad_norm": 3.421393871307373,
      "learning_rate": 3.075981037661312e-05,
      "loss": 1.3119,
      "step": 14620
    },
    {
      "epoch": 0.38506715828285487,
      "grad_norm": 2.9175899028778076,
      "learning_rate": 3.0758493547537535e-05,
      "loss": 0.2579,
      "step": 14621
    },
    {
      "epoch": 0.3850934948643666,
      "grad_norm": 2.8873727321624756,
      "learning_rate": 3.0757176718461943e-05,
      "loss": 1.5352,
      "step": 14622
    },
    {
      "epoch": 0.3851198314458783,
      "grad_norm": 3.437680959701538,
      "learning_rate": 3.075585988938636e-05,
      "loss": 1.3921,
      "step": 14623
    },
    {
      "epoch": 0.38514616802739005,
      "grad_norm": 2.003678321838379,
      "learning_rate": 3.0754543060310775e-05,
      "loss": 1.202,
      "step": 14624
    },
    {
      "epoch": 0.3851725046089018,
      "grad_norm": 1.838108777999878,
      "learning_rate": 3.075322623123519e-05,
      "loss": 1.346,
      "step": 14625
    },
    {
      "epoch": 0.3851988411904135,
      "grad_norm": 2.86262845993042,
      "learning_rate": 3.07519094021596e-05,
      "loss": 2.4319,
      "step": 14626
    },
    {
      "epoch": 0.3852251777719252,
      "grad_norm": 3.324017286300659,
      "learning_rate": 3.0750592573084015e-05,
      "loss": 1.9097,
      "step": 14627
    },
    {
      "epoch": 0.3852515143534369,
      "grad_norm": 1.6928248405456543,
      "learning_rate": 3.0749275744008424e-05,
      "loss": 1.9146,
      "step": 14628
    },
    {
      "epoch": 0.38527785093494865,
      "grad_norm": 2.3610126972198486,
      "learning_rate": 3.0747958914932846e-05,
      "loss": 1.0951,
      "step": 14629
    },
    {
      "epoch": 0.38530418751646034,
      "grad_norm": 1.782149076461792,
      "learning_rate": 3.0746642085857255e-05,
      "loss": 2.0689,
      "step": 14630
    },
    {
      "epoch": 0.3853305240979721,
      "grad_norm": 2.182767152786255,
      "learning_rate": 3.074532525678167e-05,
      "loss": 0.5372,
      "step": 14631
    },
    {
      "epoch": 0.3853568606794838,
      "grad_norm": 3.282320976257324,
      "learning_rate": 3.0744008427706086e-05,
      "loss": 1.6094,
      "step": 14632
    },
    {
      "epoch": 0.3853831972609955,
      "grad_norm": 1.773176908493042,
      "learning_rate": 3.07426915986305e-05,
      "loss": 1.3609,
      "step": 14633
    },
    {
      "epoch": 0.38540953384250726,
      "grad_norm": 1.8026783466339111,
      "learning_rate": 3.074137476955492e-05,
      "loss": 1.7881,
      "step": 14634
    },
    {
      "epoch": 0.38543587042401894,
      "grad_norm": 2.116224765777588,
      "learning_rate": 3.0740057940479326e-05,
      "loss": 1.9769,
      "step": 14635
    },
    {
      "epoch": 0.3854622070055307,
      "grad_norm": 1.8278933763504028,
      "learning_rate": 3.073874111140374e-05,
      "loss": 1.6444,
      "step": 14636
    },
    {
      "epoch": 0.3854885435870424,
      "grad_norm": 8.668768882751465,
      "learning_rate": 3.073742428232815e-05,
      "loss": 1.471,
      "step": 14637
    },
    {
      "epoch": 0.3855148801685541,
      "grad_norm": 2.8513600826263428,
      "learning_rate": 3.073610745325257e-05,
      "loss": 1.4664,
      "step": 14638
    },
    {
      "epoch": 0.38554121675006586,
      "grad_norm": 1.5551142692565918,
      "learning_rate": 3.073479062417698e-05,
      "loss": 1.7675,
      "step": 14639
    },
    {
      "epoch": 0.38556755333157755,
      "grad_norm": 1.9914963245391846,
      "learning_rate": 3.07334737951014e-05,
      "loss": 0.3447,
      "step": 14640
    },
    {
      "epoch": 0.3855938899130893,
      "grad_norm": 4.118869304656982,
      "learning_rate": 3.073215696602581e-05,
      "loss": 1.2941,
      "step": 14641
    },
    {
      "epoch": 0.385620226494601,
      "grad_norm": 2.391664981842041,
      "learning_rate": 3.073084013695022e-05,
      "loss": 0.751,
      "step": 14642
    },
    {
      "epoch": 0.3856465630761127,
      "grad_norm": 1.451113224029541,
      "learning_rate": 3.0729523307874644e-05,
      "loss": 1.949,
      "step": 14643
    },
    {
      "epoch": 0.3856728996576244,
      "grad_norm": 15.994695663452148,
      "learning_rate": 3.072820647879905e-05,
      "loss": 2.4012,
      "step": 14644
    },
    {
      "epoch": 0.38569923623913616,
      "grad_norm": 1.9516600370407104,
      "learning_rate": 3.072688964972347e-05,
      "loss": 1.9063,
      "step": 14645
    },
    {
      "epoch": 0.3857255728206479,
      "grad_norm": 2.363690137863159,
      "learning_rate": 3.072557282064788e-05,
      "loss": 1.4092,
      "step": 14646
    },
    {
      "epoch": 0.3857519094021596,
      "grad_norm": 4.139781475067139,
      "learning_rate": 3.072425599157229e-05,
      "loss": 2.207,
      "step": 14647
    },
    {
      "epoch": 0.38577824598367133,
      "grad_norm": 2.0568952560424805,
      "learning_rate": 3.072293916249671e-05,
      "loss": 1.8473,
      "step": 14648
    },
    {
      "epoch": 0.385804582565183,
      "grad_norm": 2.984442710876465,
      "learning_rate": 3.0721622333421124e-05,
      "loss": 1.2596,
      "step": 14649
    },
    {
      "epoch": 0.38583091914669476,
      "grad_norm": 1.5389206409454346,
      "learning_rate": 3.072030550434554e-05,
      "loss": 1.9694,
      "step": 14650
    },
    {
      "epoch": 0.3858572557282065,
      "grad_norm": 1.6544585227966309,
      "learning_rate": 3.071898867526995e-05,
      "loss": 1.6766,
      "step": 14651
    },
    {
      "epoch": 0.3858835923097182,
      "grad_norm": 1.9473893642425537,
      "learning_rate": 3.071767184619437e-05,
      "loss": 1.7157,
      "step": 14652
    },
    {
      "epoch": 0.38590992889122994,
      "grad_norm": 2.6641438007354736,
      "learning_rate": 3.071635501711878e-05,
      "loss": 0.8589,
      "step": 14653
    },
    {
      "epoch": 0.3859362654727416,
      "grad_norm": 3.6428356170654297,
      "learning_rate": 3.0715038188043195e-05,
      "loss": 1.7024,
      "step": 14654
    },
    {
      "epoch": 0.38596260205425337,
      "grad_norm": 2.059915781021118,
      "learning_rate": 3.0713721358967604e-05,
      "loss": 1.828,
      "step": 14655
    },
    {
      "epoch": 0.38598893863576506,
      "grad_norm": 5.337553977966309,
      "learning_rate": 3.071240452989202e-05,
      "loss": 0.5662,
      "step": 14656
    },
    {
      "epoch": 0.3860152752172768,
      "grad_norm": 1.7763347625732422,
      "learning_rate": 3.0711087700816435e-05,
      "loss": 2.4173,
      "step": 14657
    },
    {
      "epoch": 0.38604161179878854,
      "grad_norm": 2.045865535736084,
      "learning_rate": 3.070977087174085e-05,
      "loss": 1.8914,
      "step": 14658
    },
    {
      "epoch": 0.38606794838030023,
      "grad_norm": 5.311978816986084,
      "learning_rate": 3.0708454042665266e-05,
      "loss": 1.5053,
      "step": 14659
    },
    {
      "epoch": 0.386094284961812,
      "grad_norm": 1.73963463306427,
      "learning_rate": 3.0707137213589675e-05,
      "loss": 1.9052,
      "step": 14660
    },
    {
      "epoch": 0.38612062154332366,
      "grad_norm": 1.6706151962280273,
      "learning_rate": 3.070582038451409e-05,
      "loss": 1.9272,
      "step": 14661
    },
    {
      "epoch": 0.3861469581248354,
      "grad_norm": 2.0047805309295654,
      "learning_rate": 3.0704503555438506e-05,
      "loss": 1.7219,
      "step": 14662
    },
    {
      "epoch": 0.3861732947063471,
      "grad_norm": 5.656368732452393,
      "learning_rate": 3.070318672636292e-05,
      "loss": 2.0545,
      "step": 14663
    },
    {
      "epoch": 0.38619963128785884,
      "grad_norm": 1.8844788074493408,
      "learning_rate": 3.070186989728733e-05,
      "loss": 1.5804,
      "step": 14664
    },
    {
      "epoch": 0.3862259678693706,
      "grad_norm": 2.44380259513855,
      "learning_rate": 3.0700553068211746e-05,
      "loss": 1.6355,
      "step": 14665
    },
    {
      "epoch": 0.38625230445088227,
      "grad_norm": 1.6666203737258911,
      "learning_rate": 3.069923623913616e-05,
      "loss": 1.7636,
      "step": 14666
    },
    {
      "epoch": 0.386278641032394,
      "grad_norm": 1.6123305559158325,
      "learning_rate": 3.069791941006058e-05,
      "loss": 1.9662,
      "step": 14667
    },
    {
      "epoch": 0.3863049776139057,
      "grad_norm": 1.7656770944595337,
      "learning_rate": 3.069660258098499e-05,
      "loss": 1.5156,
      "step": 14668
    },
    {
      "epoch": 0.38633131419541744,
      "grad_norm": 2.0412185192108154,
      "learning_rate": 3.06952857519094e-05,
      "loss": 1.8337,
      "step": 14669
    },
    {
      "epoch": 0.38635765077692913,
      "grad_norm": 3.4634346961975098,
      "learning_rate": 3.069396892283382e-05,
      "loss": 1.1345,
      "step": 14670
    },
    {
      "epoch": 0.3863839873584409,
      "grad_norm": 1.418215036392212,
      "learning_rate": 3.069265209375823e-05,
      "loss": 1.6277,
      "step": 14671
    },
    {
      "epoch": 0.3864103239399526,
      "grad_norm": 2.4495434761047363,
      "learning_rate": 3.069133526468265e-05,
      "loss": 1.3871,
      "step": 14672
    },
    {
      "epoch": 0.3864366605214643,
      "grad_norm": 2.523446798324585,
      "learning_rate": 3.069001843560706e-05,
      "loss": 1.9659,
      "step": 14673
    },
    {
      "epoch": 0.38646299710297605,
      "grad_norm": 1.8715912103652954,
      "learning_rate": 3.068870160653147e-05,
      "loss": 1.4866,
      "step": 14674
    },
    {
      "epoch": 0.38648933368448773,
      "grad_norm": 4.594829559326172,
      "learning_rate": 3.068738477745588e-05,
      "loss": 0.8213,
      "step": 14675
    },
    {
      "epoch": 0.3865156702659995,
      "grad_norm": 1.8268094062805176,
      "learning_rate": 3.0686067948380304e-05,
      "loss": 2.1443,
      "step": 14676
    },
    {
      "epoch": 0.38654200684751117,
      "grad_norm": 2.2711901664733887,
      "learning_rate": 3.068475111930471e-05,
      "loss": 2.1928,
      "step": 14677
    },
    {
      "epoch": 0.3865683434290229,
      "grad_norm": 1.5995267629623413,
      "learning_rate": 3.068343429022913e-05,
      "loss": 2.1466,
      "step": 14678
    },
    {
      "epoch": 0.38659468001053465,
      "grad_norm": 2.0233495235443115,
      "learning_rate": 3.0682117461153544e-05,
      "loss": 1.6078,
      "step": 14679
    },
    {
      "epoch": 0.38662101659204634,
      "grad_norm": 4.105635166168213,
      "learning_rate": 3.068080063207795e-05,
      "loss": 1.9882,
      "step": 14680
    },
    {
      "epoch": 0.3866473531735581,
      "grad_norm": 2.1287147998809814,
      "learning_rate": 3.0679483803002376e-05,
      "loss": 0.8567,
      "step": 14681
    },
    {
      "epoch": 0.38667368975506977,
      "grad_norm": 2.2934086322784424,
      "learning_rate": 3.0678166973926784e-05,
      "loss": 2.3544,
      "step": 14682
    },
    {
      "epoch": 0.3867000263365815,
      "grad_norm": 2.210073471069336,
      "learning_rate": 3.06768501448512e-05,
      "loss": 2.3108,
      "step": 14683
    },
    {
      "epoch": 0.38672636291809326,
      "grad_norm": 2.6420516967773438,
      "learning_rate": 3.067553331577561e-05,
      "loss": 1.2743,
      "step": 14684
    },
    {
      "epoch": 0.38675269949960495,
      "grad_norm": 6.093904972076416,
      "learning_rate": 3.067421648670003e-05,
      "loss": 0.8936,
      "step": 14685
    },
    {
      "epoch": 0.3867790360811167,
      "grad_norm": 1.5850657224655151,
      "learning_rate": 3.067289965762444e-05,
      "loss": 1.8752,
      "step": 14686
    },
    {
      "epoch": 0.3868053726626284,
      "grad_norm": 1.843186616897583,
      "learning_rate": 3.0671582828548856e-05,
      "loss": 2.3141,
      "step": 14687
    },
    {
      "epoch": 0.3868317092441401,
      "grad_norm": 2.4175901412963867,
      "learning_rate": 3.067026599947327e-05,
      "loss": 1.6592,
      "step": 14688
    },
    {
      "epoch": 0.3868580458256518,
      "grad_norm": 2.634650468826294,
      "learning_rate": 3.066894917039768e-05,
      "loss": 1.691,
      "step": 14689
    },
    {
      "epoch": 0.38688438240716355,
      "grad_norm": 2.40167236328125,
      "learning_rate": 3.06676323413221e-05,
      "loss": 1.4214,
      "step": 14690
    },
    {
      "epoch": 0.3869107189886753,
      "grad_norm": 2.1508870124816895,
      "learning_rate": 3.066631551224651e-05,
      "loss": 1.5924,
      "step": 14691
    },
    {
      "epoch": 0.386937055570187,
      "grad_norm": 2.137068033218384,
      "learning_rate": 3.066499868317093e-05,
      "loss": 1.3394,
      "step": 14692
    },
    {
      "epoch": 0.3869633921516987,
      "grad_norm": 4.683769226074219,
      "learning_rate": 3.0663681854095336e-05,
      "loss": 0.8157,
      "step": 14693
    },
    {
      "epoch": 0.3869897287332104,
      "grad_norm": 2.0051162242889404,
      "learning_rate": 3.066236502501975e-05,
      "loss": 0.3501,
      "step": 14694
    },
    {
      "epoch": 0.38701606531472216,
      "grad_norm": 1.565938115119934,
      "learning_rate": 3.066104819594417e-05,
      "loss": 1.746,
      "step": 14695
    },
    {
      "epoch": 0.38704240189623385,
      "grad_norm": 2.1133971214294434,
      "learning_rate": 3.065973136686858e-05,
      "loss": 2.0262,
      "step": 14696
    },
    {
      "epoch": 0.3870687384777456,
      "grad_norm": 1.8275271654129028,
      "learning_rate": 3.0658414537793e-05,
      "loss": 2.2772,
      "step": 14697
    },
    {
      "epoch": 0.38709507505925733,
      "grad_norm": 2.5442302227020264,
      "learning_rate": 3.065709770871741e-05,
      "loss": 0.4184,
      "step": 14698
    },
    {
      "epoch": 0.387121411640769,
      "grad_norm": 2.144209146499634,
      "learning_rate": 3.065578087964183e-05,
      "loss": 1.6374,
      "step": 14699
    },
    {
      "epoch": 0.38714774822228076,
      "grad_norm": 2.5194780826568604,
      "learning_rate": 3.065446405056624e-05,
      "loss": 1.5709,
      "step": 14700
    },
    {
      "epoch": 0.38717408480379245,
      "grad_norm": 1.77311110496521,
      "learning_rate": 3.0653147221490654e-05,
      "loss": 2.0474,
      "step": 14701
    },
    {
      "epoch": 0.3872004213853042,
      "grad_norm": 1.478083848953247,
      "learning_rate": 3.065183039241506e-05,
      "loss": 1.4352,
      "step": 14702
    },
    {
      "epoch": 0.3872267579668159,
      "grad_norm": 2.535492420196533,
      "learning_rate": 3.065051356333948e-05,
      "loss": 2.0801,
      "step": 14703
    },
    {
      "epoch": 0.3872530945483276,
      "grad_norm": 2.155217170715332,
      "learning_rate": 3.0649196734263894e-05,
      "loss": 1.8583,
      "step": 14704
    },
    {
      "epoch": 0.38727943112983937,
      "grad_norm": 2.1672637462615967,
      "learning_rate": 3.064787990518831e-05,
      "loss": 2.0878,
      "step": 14705
    },
    {
      "epoch": 0.38730576771135106,
      "grad_norm": 3.0483036041259766,
      "learning_rate": 3.0646563076112725e-05,
      "loss": 2.4231,
      "step": 14706
    },
    {
      "epoch": 0.3873321042928628,
      "grad_norm": 3.048377275466919,
      "learning_rate": 3.0645246247037134e-05,
      "loss": 2.7498,
      "step": 14707
    },
    {
      "epoch": 0.3873584408743745,
      "grad_norm": 2.419497489929199,
      "learning_rate": 3.064392941796155e-05,
      "loss": 1.8527,
      "step": 14708
    },
    {
      "epoch": 0.38738477745588623,
      "grad_norm": 2.4085893630981445,
      "learning_rate": 3.0642612588885965e-05,
      "loss": 1.7436,
      "step": 14709
    },
    {
      "epoch": 0.3874111140373979,
      "grad_norm": 2.110053777694702,
      "learning_rate": 3.064129575981038e-05,
      "loss": 1.0266,
      "step": 14710
    },
    {
      "epoch": 0.38743745061890966,
      "grad_norm": 4.174890995025635,
      "learning_rate": 3.063997893073479e-05,
      "loss": 2.5177,
      "step": 14711
    },
    {
      "epoch": 0.3874637872004214,
      "grad_norm": 2.3036508560180664,
      "learning_rate": 3.0638662101659205e-05,
      "loss": 2.0052,
      "step": 14712
    },
    {
      "epoch": 0.3874901237819331,
      "grad_norm": 3.0439138412475586,
      "learning_rate": 3.063734527258362e-05,
      "loss": 0.8506,
      "step": 14713
    },
    {
      "epoch": 0.38751646036344484,
      "grad_norm": 3.9835851192474365,
      "learning_rate": 3.0636028443508036e-05,
      "loss": 1.057,
      "step": 14714
    },
    {
      "epoch": 0.3875427969449565,
      "grad_norm": 2.618096351623535,
      "learning_rate": 3.063471161443245e-05,
      "loss": 2.0236,
      "step": 14715
    },
    {
      "epoch": 0.38756913352646827,
      "grad_norm": 1.820068120956421,
      "learning_rate": 3.063339478535686e-05,
      "loss": 1.9378,
      "step": 14716
    },
    {
      "epoch": 0.38759547010798,
      "grad_norm": 2.012728214263916,
      "learning_rate": 3.0632077956281276e-05,
      "loss": 1.9251,
      "step": 14717
    },
    {
      "epoch": 0.3876218066894917,
      "grad_norm": 5.958434104919434,
      "learning_rate": 3.063076112720569e-05,
      "loss": 2.9743,
      "step": 14718
    },
    {
      "epoch": 0.38764814327100344,
      "grad_norm": 2.6244795322418213,
      "learning_rate": 3.062944429813011e-05,
      "loss": 2.0101,
      "step": 14719
    },
    {
      "epoch": 0.38767447985251513,
      "grad_norm": 2.020388126373291,
      "learning_rate": 3.0628127469054516e-05,
      "loss": 0.8601,
      "step": 14720
    },
    {
      "epoch": 0.3877008164340269,
      "grad_norm": 1.7658718824386597,
      "learning_rate": 3.062681063997893e-05,
      "loss": 1.5219,
      "step": 14721
    },
    {
      "epoch": 0.38772715301553856,
      "grad_norm": 1.4189436435699463,
      "learning_rate": 3.062549381090335e-05,
      "loss": 1.7126,
      "step": 14722
    },
    {
      "epoch": 0.3877534895970503,
      "grad_norm": 3.4577560424804688,
      "learning_rate": 3.062417698182776e-05,
      "loss": 1.8513,
      "step": 14723
    },
    {
      "epoch": 0.38777982617856205,
      "grad_norm": 1.5386041402816772,
      "learning_rate": 3.062286015275218e-05,
      "loss": 2.1029,
      "step": 14724
    },
    {
      "epoch": 0.38780616276007374,
      "grad_norm": 3.1291961669921875,
      "learning_rate": 3.062154332367659e-05,
      "loss": 1.3294,
      "step": 14725
    },
    {
      "epoch": 0.3878324993415855,
      "grad_norm": 3.991101026535034,
      "learning_rate": 3.0620226494601e-05,
      "loss": 1.2592,
      "step": 14726
    },
    {
      "epoch": 0.38785883592309717,
      "grad_norm": 2.179367780685425,
      "learning_rate": 3.061890966552541e-05,
      "loss": 1.814,
      "step": 14727
    },
    {
      "epoch": 0.3878851725046089,
      "grad_norm": 3.0196750164031982,
      "learning_rate": 3.0617592836449834e-05,
      "loss": 2.2316,
      "step": 14728
    },
    {
      "epoch": 0.3879115090861206,
      "grad_norm": 1.6353254318237305,
      "learning_rate": 3.061627600737424e-05,
      "loss": 0.4928,
      "step": 14729
    },
    {
      "epoch": 0.38793784566763234,
      "grad_norm": 4.2615156173706055,
      "learning_rate": 3.061495917829866e-05,
      "loss": 1.2137,
      "step": 14730
    },
    {
      "epoch": 0.3879641822491441,
      "grad_norm": 1.837337613105774,
      "learning_rate": 3.061364234922307e-05,
      "loss": 2.0321,
      "step": 14731
    },
    {
      "epoch": 0.3879905188306558,
      "grad_norm": 4.390244483947754,
      "learning_rate": 3.061232552014749e-05,
      "loss": 0.8569,
      "step": 14732
    },
    {
      "epoch": 0.3880168554121675,
      "grad_norm": 1.7401536703109741,
      "learning_rate": 3.06110086910719e-05,
      "loss": 1.1982,
      "step": 14733
    },
    {
      "epoch": 0.3880431919936792,
      "grad_norm": 2.0399770736694336,
      "learning_rate": 3.0609691861996314e-05,
      "loss": 1.6896,
      "step": 14734
    },
    {
      "epoch": 0.38806952857519095,
      "grad_norm": 2.6990976333618164,
      "learning_rate": 3.060837503292073e-05,
      "loss": 1.8925,
      "step": 14735
    },
    {
      "epoch": 0.38809586515670264,
      "grad_norm": 3.387763261795044,
      "learning_rate": 3.060705820384514e-05,
      "loss": 0.8911,
      "step": 14736
    },
    {
      "epoch": 0.3881222017382144,
      "grad_norm": 2.366509199142456,
      "learning_rate": 3.060574137476956e-05,
      "loss": 2.0934,
      "step": 14737
    },
    {
      "epoch": 0.3881485383197261,
      "grad_norm": 4.434013843536377,
      "learning_rate": 3.060442454569397e-05,
      "loss": 1.4463,
      "step": 14738
    },
    {
      "epoch": 0.3881748749012378,
      "grad_norm": 3.942121982574463,
      "learning_rate": 3.0603107716618385e-05,
      "loss": 1.9234,
      "step": 14739
    },
    {
      "epoch": 0.38820121148274955,
      "grad_norm": 3.9252727031707764,
      "learning_rate": 3.0601790887542794e-05,
      "loss": 1.659,
      "step": 14740
    },
    {
      "epoch": 0.38822754806426124,
      "grad_norm": 2.316638231277466,
      "learning_rate": 3.060047405846721e-05,
      "loss": 2.1054,
      "step": 14741
    },
    {
      "epoch": 0.388253884645773,
      "grad_norm": 1.949794054031372,
      "learning_rate": 3.0599157229391625e-05,
      "loss": 0.58,
      "step": 14742
    },
    {
      "epoch": 0.38828022122728467,
      "grad_norm": 2.1228957176208496,
      "learning_rate": 3.059784040031604e-05,
      "loss": 1.2641,
      "step": 14743
    },
    {
      "epoch": 0.3883065578087964,
      "grad_norm": 1.845453143119812,
      "learning_rate": 3.059652357124046e-05,
      "loss": 0.651,
      "step": 14744
    },
    {
      "epoch": 0.38833289439030816,
      "grad_norm": 1.449431300163269,
      "learning_rate": 3.0595206742164865e-05,
      "loss": 1.8533,
      "step": 14745
    },
    {
      "epoch": 0.38835923097181985,
      "grad_norm": 2.3409101963043213,
      "learning_rate": 3.059388991308928e-05,
      "loss": 1.4373,
      "step": 14746
    },
    {
      "epoch": 0.3883855675533316,
      "grad_norm": 4.517127513885498,
      "learning_rate": 3.05925730840137e-05,
      "loss": 0.7689,
      "step": 14747
    },
    {
      "epoch": 0.3884119041348433,
      "grad_norm": 1.7706965208053589,
      "learning_rate": 3.059125625493811e-05,
      "loss": 1.3555,
      "step": 14748
    },
    {
      "epoch": 0.388438240716355,
      "grad_norm": 1.6963951587677002,
      "learning_rate": 3.058993942586252e-05,
      "loss": 2.0346,
      "step": 14749
    },
    {
      "epoch": 0.38846457729786676,
      "grad_norm": 3.086503028869629,
      "learning_rate": 3.058862259678694e-05,
      "loss": 1.6292,
      "step": 14750
    },
    {
      "epoch": 0.38849091387937845,
      "grad_norm": 1.985912799835205,
      "learning_rate": 3.058730576771135e-05,
      "loss": 1.9139,
      "step": 14751
    },
    {
      "epoch": 0.3885172504608902,
      "grad_norm": 1.7167713642120361,
      "learning_rate": 3.058598893863577e-05,
      "loss": 1.7078,
      "step": 14752
    },
    {
      "epoch": 0.3885435870424019,
      "grad_norm": 2.294015645980835,
      "learning_rate": 3.0584672109560183e-05,
      "loss": 1.6863,
      "step": 14753
    },
    {
      "epoch": 0.3885699236239136,
      "grad_norm": 3.438015937805176,
      "learning_rate": 3.058335528048459e-05,
      "loss": 0.9469,
      "step": 14754
    },
    {
      "epoch": 0.3885962602054253,
      "grad_norm": 2.79647159576416,
      "learning_rate": 3.058203845140901e-05,
      "loss": 0.4692,
      "step": 14755
    },
    {
      "epoch": 0.38862259678693706,
      "grad_norm": 1.8933472633361816,
      "learning_rate": 3.0580721622333424e-05,
      "loss": 2.0658,
      "step": 14756
    },
    {
      "epoch": 0.3886489333684488,
      "grad_norm": 1.668656349182129,
      "learning_rate": 3.057940479325784e-05,
      "loss": 2.131,
      "step": 14757
    },
    {
      "epoch": 0.3886752699499605,
      "grad_norm": 1.7183003425598145,
      "learning_rate": 3.057808796418225e-05,
      "loss": 1.7702,
      "step": 14758
    },
    {
      "epoch": 0.38870160653147223,
      "grad_norm": 3.017056465148926,
      "learning_rate": 3.0576771135106664e-05,
      "loss": 0.9699,
      "step": 14759
    },
    {
      "epoch": 0.3887279431129839,
      "grad_norm": 1.6128453016281128,
      "learning_rate": 3.057545430603108e-05,
      "loss": 2.1362,
      "step": 14760
    },
    {
      "epoch": 0.38875427969449566,
      "grad_norm": 3.795217514038086,
      "learning_rate": 3.0574137476955495e-05,
      "loss": 0.8622,
      "step": 14761
    },
    {
      "epoch": 0.38878061627600735,
      "grad_norm": 1.7457852363586426,
      "learning_rate": 3.057282064787991e-05,
      "loss": 2.3109,
      "step": 14762
    },
    {
      "epoch": 0.3888069528575191,
      "grad_norm": 2.2657079696655273,
      "learning_rate": 3.057150381880432e-05,
      "loss": 1.2325,
      "step": 14763
    },
    {
      "epoch": 0.38883328943903084,
      "grad_norm": 1.9281193017959595,
      "learning_rate": 3.0570186989728735e-05,
      "loss": 0.4385,
      "step": 14764
    },
    {
      "epoch": 0.3888596260205425,
      "grad_norm": 1.8128288984298706,
      "learning_rate": 3.056887016065315e-05,
      "loss": 1.7317,
      "step": 14765
    },
    {
      "epoch": 0.38888596260205427,
      "grad_norm": 1.6248143911361694,
      "learning_rate": 3.0567553331577566e-05,
      "loss": 1.7334,
      "step": 14766
    },
    {
      "epoch": 0.38891229918356596,
      "grad_norm": 2.2053070068359375,
      "learning_rate": 3.0566236502501975e-05,
      "loss": 1.6227,
      "step": 14767
    },
    {
      "epoch": 0.3889386357650777,
      "grad_norm": 2.0217785835266113,
      "learning_rate": 3.056491967342639e-05,
      "loss": 1.4445,
      "step": 14768
    },
    {
      "epoch": 0.3889649723465894,
      "grad_norm": 1.6188602447509766,
      "learning_rate": 3.0563602844350806e-05,
      "loss": 1.5988,
      "step": 14769
    },
    {
      "epoch": 0.38899130892810113,
      "grad_norm": 3.8315913677215576,
      "learning_rate": 3.056228601527522e-05,
      "loss": 1.4201,
      "step": 14770
    },
    {
      "epoch": 0.3890176455096129,
      "grad_norm": 3.125159502029419,
      "learning_rate": 3.056096918619964e-05,
      "loss": 1.4993,
      "step": 14771
    },
    {
      "epoch": 0.38904398209112456,
      "grad_norm": 1.9284790754318237,
      "learning_rate": 3.0559652357124046e-05,
      "loss": 0.6335,
      "step": 14772
    },
    {
      "epoch": 0.3890703186726363,
      "grad_norm": 2.823977470397949,
      "learning_rate": 3.055833552804846e-05,
      "loss": 1.5023,
      "step": 14773
    },
    {
      "epoch": 0.389096655254148,
      "grad_norm": 2.3714962005615234,
      "learning_rate": 3.055701869897287e-05,
      "loss": 2.1538,
      "step": 14774
    },
    {
      "epoch": 0.38912299183565974,
      "grad_norm": 2.0893714427948,
      "learning_rate": 3.055570186989729e-05,
      "loss": 2.2786,
      "step": 14775
    },
    {
      "epoch": 0.3891493284171714,
      "grad_norm": 7.809403896331787,
      "learning_rate": 3.05543850408217e-05,
      "loss": 1.475,
      "step": 14776
    },
    {
      "epoch": 0.38917566499868317,
      "grad_norm": 1.7811617851257324,
      "learning_rate": 3.055306821174612e-05,
      "loss": 2.0439,
      "step": 14777
    },
    {
      "epoch": 0.3892020015801949,
      "grad_norm": 1.877899408340454,
      "learning_rate": 3.0551751382670526e-05,
      "loss": 1.6747,
      "step": 14778
    },
    {
      "epoch": 0.3892283381617066,
      "grad_norm": 1.5328963994979858,
      "learning_rate": 3.055043455359494e-05,
      "loss": 1.8227,
      "step": 14779
    },
    {
      "epoch": 0.38925467474321834,
      "grad_norm": 2.7405691146850586,
      "learning_rate": 3.054911772451936e-05,
      "loss": 1.8776,
      "step": 14780
    },
    {
      "epoch": 0.38928101132473003,
      "grad_norm": 2.2172694206237793,
      "learning_rate": 3.054780089544377e-05,
      "loss": 2.0139,
      "step": 14781
    },
    {
      "epoch": 0.3893073479062418,
      "grad_norm": 1.8459181785583496,
      "learning_rate": 3.054648406636819e-05,
      "loss": 1.8849,
      "step": 14782
    },
    {
      "epoch": 0.38933368448775346,
      "grad_norm": 1.8791639804840088,
      "learning_rate": 3.05451672372926e-05,
      "loss": 1.8156,
      "step": 14783
    },
    {
      "epoch": 0.3893600210692652,
      "grad_norm": 1.9491006135940552,
      "learning_rate": 3.054385040821702e-05,
      "loss": 1.8969,
      "step": 14784
    },
    {
      "epoch": 0.38938635765077695,
      "grad_norm": 3.5040998458862305,
      "learning_rate": 3.054253357914143e-05,
      "loss": 2.5006,
      "step": 14785
    },
    {
      "epoch": 0.38941269423228864,
      "grad_norm": 1.7740840911865234,
      "learning_rate": 3.0541216750065844e-05,
      "loss": 1.8072,
      "step": 14786
    },
    {
      "epoch": 0.3894390308138004,
      "grad_norm": 3.1526856422424316,
      "learning_rate": 3.053989992099025e-05,
      "loss": 1.7881,
      "step": 14787
    },
    {
      "epoch": 0.38946536739531207,
      "grad_norm": 1.7720575332641602,
      "learning_rate": 3.053858309191467e-05,
      "loss": 2.1246,
      "step": 14788
    },
    {
      "epoch": 0.3894917039768238,
      "grad_norm": 1.658571720123291,
      "learning_rate": 3.0537266262839084e-05,
      "loss": 2.1139,
      "step": 14789
    },
    {
      "epoch": 0.38951804055833555,
      "grad_norm": 3.5377626419067383,
      "learning_rate": 3.05359494337635e-05,
      "loss": 1.5869,
      "step": 14790
    },
    {
      "epoch": 0.38954437713984724,
      "grad_norm": 2.2462644577026367,
      "learning_rate": 3.0534632604687915e-05,
      "loss": 1.8108,
      "step": 14791
    },
    {
      "epoch": 0.389570713721359,
      "grad_norm": 1.6197788715362549,
      "learning_rate": 3.0533315775612324e-05,
      "loss": 0.6414,
      "step": 14792
    },
    {
      "epoch": 0.3895970503028707,
      "grad_norm": 2.375152111053467,
      "learning_rate": 3.053199894653674e-05,
      "loss": 1.95,
      "step": 14793
    },
    {
      "epoch": 0.3896233868843824,
      "grad_norm": 2.354613780975342,
      "learning_rate": 3.0530682117461155e-05,
      "loss": 0.9079,
      "step": 14794
    },
    {
      "epoch": 0.3896497234658941,
      "grad_norm": 2.530278444290161,
      "learning_rate": 3.052936528838557e-05,
      "loss": 1.6069,
      "step": 14795
    },
    {
      "epoch": 0.38967606004740585,
      "grad_norm": 3.9799392223358154,
      "learning_rate": 3.052804845930998e-05,
      "loss": 1.6279,
      "step": 14796
    },
    {
      "epoch": 0.3897023966289176,
      "grad_norm": 3.936377763748169,
      "learning_rate": 3.0526731630234395e-05,
      "loss": 2.1348,
      "step": 14797
    },
    {
      "epoch": 0.3897287332104293,
      "grad_norm": 2.7203400135040283,
      "learning_rate": 3.052541480115881e-05,
      "loss": 0.5619,
      "step": 14798
    },
    {
      "epoch": 0.389755069791941,
      "grad_norm": 1.6430703401565552,
      "learning_rate": 3.0524097972083226e-05,
      "loss": 1.5575,
      "step": 14799
    },
    {
      "epoch": 0.3897814063734527,
      "grad_norm": 3.247898817062378,
      "learning_rate": 3.052278114300764e-05,
      "loss": 1.8223,
      "step": 14800
    },
    {
      "epoch": 0.38980774295496445,
      "grad_norm": 1.4750473499298096,
      "learning_rate": 3.052146431393205e-05,
      "loss": 1.5932,
      "step": 14801
    },
    {
      "epoch": 0.38983407953647614,
      "grad_norm": 3.719090223312378,
      "learning_rate": 3.0520147484856466e-05,
      "loss": 2.2278,
      "step": 14802
    },
    {
      "epoch": 0.3898604161179879,
      "grad_norm": 3.1308326721191406,
      "learning_rate": 3.051883065578088e-05,
      "loss": 1.1906,
      "step": 14803
    },
    {
      "epoch": 0.38988675269949963,
      "grad_norm": 3.238124132156372,
      "learning_rate": 3.0517513826705298e-05,
      "loss": 2.3122,
      "step": 14804
    },
    {
      "epoch": 0.3899130892810113,
      "grad_norm": 2.135883331298828,
      "learning_rate": 3.051619699762971e-05,
      "loss": 1.5543,
      "step": 14805
    },
    {
      "epoch": 0.38993942586252306,
      "grad_norm": 2.100149154663086,
      "learning_rate": 3.0514880168554122e-05,
      "loss": 2.0551,
      "step": 14806
    },
    {
      "epoch": 0.38996576244403475,
      "grad_norm": 2.451986789703369,
      "learning_rate": 3.0513563339478534e-05,
      "loss": 0.6453,
      "step": 14807
    },
    {
      "epoch": 0.3899920990255465,
      "grad_norm": 1.7605277299880981,
      "learning_rate": 3.0512246510402953e-05,
      "loss": 1.8757,
      "step": 14808
    },
    {
      "epoch": 0.3900184356070582,
      "grad_norm": 1.8690348863601685,
      "learning_rate": 3.0510929681327365e-05,
      "loss": 1.336,
      "step": 14809
    },
    {
      "epoch": 0.3900447721885699,
      "grad_norm": 1.969211459159851,
      "learning_rate": 3.0509612852251778e-05,
      "loss": 1.9297,
      "step": 14810
    },
    {
      "epoch": 0.39007110877008166,
      "grad_norm": 1.8377444744110107,
      "learning_rate": 3.0508296023176193e-05,
      "loss": 2.1067,
      "step": 14811
    },
    {
      "epoch": 0.39009744535159335,
      "grad_norm": 2.1249094009399414,
      "learning_rate": 3.0506979194100606e-05,
      "loss": 1.1892,
      "step": 14812
    },
    {
      "epoch": 0.3901237819331051,
      "grad_norm": 5.820160388946533,
      "learning_rate": 3.0505662365025024e-05,
      "loss": 0.9323,
      "step": 14813
    },
    {
      "epoch": 0.3901501185146168,
      "grad_norm": 1.8319264650344849,
      "learning_rate": 3.0504345535949437e-05,
      "loss": 1.7942,
      "step": 14814
    },
    {
      "epoch": 0.3901764550961285,
      "grad_norm": 2.809297800064087,
      "learning_rate": 3.050302870687385e-05,
      "loss": 1.0859,
      "step": 14815
    },
    {
      "epoch": 0.3902027916776402,
      "grad_norm": 2.9805855751037598,
      "learning_rate": 3.050171187779826e-05,
      "loss": 1.1923,
      "step": 14816
    },
    {
      "epoch": 0.39022912825915196,
      "grad_norm": 3.4213922023773193,
      "learning_rate": 3.050039504872268e-05,
      "loss": 2.4029,
      "step": 14817
    },
    {
      "epoch": 0.3902554648406637,
      "grad_norm": 1.3115689754486084,
      "learning_rate": 3.0499078219647092e-05,
      "loss": 1.4285,
      "step": 14818
    },
    {
      "epoch": 0.3902818014221754,
      "grad_norm": 1.8398888111114502,
      "learning_rate": 3.0497761390571505e-05,
      "loss": 1.9925,
      "step": 14819
    },
    {
      "epoch": 0.39030813800368713,
      "grad_norm": 1.6225249767303467,
      "learning_rate": 3.0496444561495917e-05,
      "loss": 1.5331,
      "step": 14820
    },
    {
      "epoch": 0.3903344745851988,
      "grad_norm": 3.7366654872894287,
      "learning_rate": 3.0495127732420332e-05,
      "loss": 1.7678,
      "step": 14821
    },
    {
      "epoch": 0.39036081116671056,
      "grad_norm": 1.8978986740112305,
      "learning_rate": 3.0493810903344748e-05,
      "loss": 0.4165,
      "step": 14822
    },
    {
      "epoch": 0.3903871477482223,
      "grad_norm": 1.9186005592346191,
      "learning_rate": 3.0492494074269164e-05,
      "loss": 1.4529,
      "step": 14823
    },
    {
      "epoch": 0.390413484329734,
      "grad_norm": 2.3490982055664062,
      "learning_rate": 3.0491177245193576e-05,
      "loss": 1.8218,
      "step": 14824
    },
    {
      "epoch": 0.39043982091124574,
      "grad_norm": 1.5955188274383545,
      "learning_rate": 3.0489860416117988e-05,
      "loss": 1.6101,
      "step": 14825
    },
    {
      "epoch": 0.3904661574927574,
      "grad_norm": 1.7607296705245972,
      "learning_rate": 3.04885435870424e-05,
      "loss": 1.7532,
      "step": 14826
    },
    {
      "epoch": 0.39049249407426917,
      "grad_norm": 2.687554359436035,
      "learning_rate": 3.048722675796682e-05,
      "loss": 1.0286,
      "step": 14827
    },
    {
      "epoch": 0.39051883065578086,
      "grad_norm": 2.9266109466552734,
      "learning_rate": 3.048590992889123e-05,
      "loss": 1.7798,
      "step": 14828
    },
    {
      "epoch": 0.3905451672372926,
      "grad_norm": 3.858023166656494,
      "learning_rate": 3.0484593099815644e-05,
      "loss": 1.9827,
      "step": 14829
    },
    {
      "epoch": 0.39057150381880434,
      "grad_norm": 10.170050621032715,
      "learning_rate": 3.048327627074006e-05,
      "loss": 1.5157,
      "step": 14830
    },
    {
      "epoch": 0.39059784040031603,
      "grad_norm": 1.9377825260162354,
      "learning_rate": 3.0481959441664475e-05,
      "loss": 2.1592,
      "step": 14831
    },
    {
      "epoch": 0.3906241769818278,
      "grad_norm": 1.8588459491729736,
      "learning_rate": 3.048064261258889e-05,
      "loss": 1.4449,
      "step": 14832
    },
    {
      "epoch": 0.39065051356333946,
      "grad_norm": 2.999317169189453,
      "learning_rate": 3.0479325783513303e-05,
      "loss": 0.9911,
      "step": 14833
    },
    {
      "epoch": 0.3906768501448512,
      "grad_norm": 2.076369047164917,
      "learning_rate": 3.0478008954437715e-05,
      "loss": 2.4455,
      "step": 14834
    },
    {
      "epoch": 0.3907031867263629,
      "grad_norm": 1.6744062900543213,
      "learning_rate": 3.0476692125362127e-05,
      "loss": 0.3062,
      "step": 14835
    },
    {
      "epoch": 0.39072952330787464,
      "grad_norm": 1.8977676630020142,
      "learning_rate": 3.0475375296286546e-05,
      "loss": 1.889,
      "step": 14836
    },
    {
      "epoch": 0.3907558598893864,
      "grad_norm": 1.4191335439682007,
      "learning_rate": 3.0474058467210958e-05,
      "loss": 1.7728,
      "step": 14837
    },
    {
      "epoch": 0.39078219647089807,
      "grad_norm": 2.384071111679077,
      "learning_rate": 3.047274163813537e-05,
      "loss": 1.6112,
      "step": 14838
    },
    {
      "epoch": 0.3908085330524098,
      "grad_norm": 1.572990894317627,
      "learning_rate": 3.0471424809059783e-05,
      "loss": 1.8101,
      "step": 14839
    },
    {
      "epoch": 0.3908348696339215,
      "grad_norm": 2.649071455001831,
      "learning_rate": 3.0470107979984198e-05,
      "loss": 1.7474,
      "step": 14840
    },
    {
      "epoch": 0.39086120621543324,
      "grad_norm": 1.543198823928833,
      "learning_rate": 3.0468791150908614e-05,
      "loss": 2.2882,
      "step": 14841
    },
    {
      "epoch": 0.39088754279694493,
      "grad_norm": 1.572638988494873,
      "learning_rate": 3.046747432183303e-05,
      "loss": 1.6301,
      "step": 14842
    },
    {
      "epoch": 0.3909138793784567,
      "grad_norm": 3.6765494346618652,
      "learning_rate": 3.046615749275744e-05,
      "loss": 1.4851,
      "step": 14843
    },
    {
      "epoch": 0.3909402159599684,
      "grad_norm": 1.8132128715515137,
      "learning_rate": 3.0464840663681854e-05,
      "loss": 1.281,
      "step": 14844
    },
    {
      "epoch": 0.3909665525414801,
      "grad_norm": 2.5002686977386475,
      "learning_rate": 3.0463523834606266e-05,
      "loss": 1.3173,
      "step": 14845
    },
    {
      "epoch": 0.39099288912299185,
      "grad_norm": 2.05584716796875,
      "learning_rate": 3.0462207005530685e-05,
      "loss": 0.381,
      "step": 14846
    },
    {
      "epoch": 0.39101922570450354,
      "grad_norm": 2.1467530727386475,
      "learning_rate": 3.0460890176455097e-05,
      "loss": 1.4935,
      "step": 14847
    },
    {
      "epoch": 0.3910455622860153,
      "grad_norm": 1.642984390258789,
      "learning_rate": 3.045957334737951e-05,
      "loss": 1.7472,
      "step": 14848
    },
    {
      "epoch": 0.39107189886752697,
      "grad_norm": 4.430726528167725,
      "learning_rate": 3.0458256518303925e-05,
      "loss": 1.0849,
      "step": 14849
    },
    {
      "epoch": 0.3910982354490387,
      "grad_norm": 2.195782423019409,
      "learning_rate": 3.045693968922834e-05,
      "loss": 2.0548,
      "step": 14850
    },
    {
      "epoch": 0.39112457203055045,
      "grad_norm": 2.064549207687378,
      "learning_rate": 3.0455622860152756e-05,
      "loss": 1.1847,
      "step": 14851
    },
    {
      "epoch": 0.39115090861206214,
      "grad_norm": 1.8034393787384033,
      "learning_rate": 3.045430603107717e-05,
      "loss": 1.5991,
      "step": 14852
    },
    {
      "epoch": 0.3911772451935739,
      "grad_norm": 1.5642762184143066,
      "learning_rate": 3.045298920200158e-05,
      "loss": 1.4413,
      "step": 14853
    },
    {
      "epoch": 0.3912035817750856,
      "grad_norm": 2.640472412109375,
      "learning_rate": 3.0451672372925993e-05,
      "loss": 0.8813,
      "step": 14854
    },
    {
      "epoch": 0.3912299183565973,
      "grad_norm": 2.2595884799957275,
      "learning_rate": 3.0450355543850412e-05,
      "loss": 2.0476,
      "step": 14855
    },
    {
      "epoch": 0.39125625493810906,
      "grad_norm": 2.333843469619751,
      "learning_rate": 3.0449038714774824e-05,
      "loss": 1.2486,
      "step": 14856
    },
    {
      "epoch": 0.39128259151962075,
      "grad_norm": 2.0612878799438477,
      "learning_rate": 3.0447721885699236e-05,
      "loss": 1.7296,
      "step": 14857
    },
    {
      "epoch": 0.3913089281011325,
      "grad_norm": 1.7651320695877075,
      "learning_rate": 3.0446405056623652e-05,
      "loss": 2.4206,
      "step": 14858
    },
    {
      "epoch": 0.3913352646826442,
      "grad_norm": 2.996703624725342,
      "learning_rate": 3.0445088227548064e-05,
      "loss": 0.7184,
      "step": 14859
    },
    {
      "epoch": 0.3913616012641559,
      "grad_norm": 3.8319647312164307,
      "learning_rate": 3.0443771398472483e-05,
      "loss": 1.1606,
      "step": 14860
    },
    {
      "epoch": 0.3913879378456676,
      "grad_norm": 2.659003257751465,
      "learning_rate": 3.0442454569396895e-05,
      "loss": 1.4834,
      "step": 14861
    },
    {
      "epoch": 0.39141427442717935,
      "grad_norm": 2.2486412525177,
      "learning_rate": 3.0441137740321307e-05,
      "loss": 2.1412,
      "step": 14862
    },
    {
      "epoch": 0.3914406110086911,
      "grad_norm": 1.9141794443130493,
      "learning_rate": 3.043982091124572e-05,
      "loss": 1.6169,
      "step": 14863
    },
    {
      "epoch": 0.3914669475902028,
      "grad_norm": 2.2230496406555176,
      "learning_rate": 3.043850408217014e-05,
      "loss": 2.0887,
      "step": 14864
    },
    {
      "epoch": 0.39149328417171453,
      "grad_norm": 2.0029618740081787,
      "learning_rate": 3.043718725309455e-05,
      "loss": 2.2636,
      "step": 14865
    },
    {
      "epoch": 0.3915196207532262,
      "grad_norm": 1.8449727296829224,
      "learning_rate": 3.0435870424018963e-05,
      "loss": 1.9363,
      "step": 14866
    },
    {
      "epoch": 0.39154595733473796,
      "grad_norm": 1.8683706521987915,
      "learning_rate": 3.0434553594943375e-05,
      "loss": 1.2098,
      "step": 14867
    },
    {
      "epoch": 0.39157229391624965,
      "grad_norm": 1.562099814414978,
      "learning_rate": 3.043323676586779e-05,
      "loss": 0.3352,
      "step": 14868
    },
    {
      "epoch": 0.3915986304977614,
      "grad_norm": 1.8673365116119385,
      "learning_rate": 3.0431919936792206e-05,
      "loss": 1.7897,
      "step": 14869
    },
    {
      "epoch": 0.39162496707927313,
      "grad_norm": 2.4109718799591064,
      "learning_rate": 3.0430603107716622e-05,
      "loss": 1.804,
      "step": 14870
    },
    {
      "epoch": 0.3916513036607848,
      "grad_norm": 1.8080004453659058,
      "learning_rate": 3.0429286278641034e-05,
      "loss": 2.0486,
      "step": 14871
    },
    {
      "epoch": 0.39167764024229657,
      "grad_norm": 1.4065428972244263,
      "learning_rate": 3.0427969449565446e-05,
      "loss": 1.2804,
      "step": 14872
    },
    {
      "epoch": 0.39170397682380825,
      "grad_norm": 1.6550333499908447,
      "learning_rate": 3.042665262048986e-05,
      "loss": 1.8038,
      "step": 14873
    },
    {
      "epoch": 0.39173031340532,
      "grad_norm": 2.656360149383545,
      "learning_rate": 3.0425335791414278e-05,
      "loss": 1.1885,
      "step": 14874
    },
    {
      "epoch": 0.3917566499868317,
      "grad_norm": 3.0260536670684814,
      "learning_rate": 3.042401896233869e-05,
      "loss": 0.8293,
      "step": 14875
    },
    {
      "epoch": 0.3917829865683434,
      "grad_norm": 2.475206136703491,
      "learning_rate": 3.0422702133263102e-05,
      "loss": 1.7176,
      "step": 14876
    },
    {
      "epoch": 0.39180932314985517,
      "grad_norm": 2.055607318878174,
      "learning_rate": 3.0421385304187518e-05,
      "loss": 1.9977,
      "step": 14877
    },
    {
      "epoch": 0.39183565973136686,
      "grad_norm": 3.346900463104248,
      "learning_rate": 3.0420068475111933e-05,
      "loss": 2.0267,
      "step": 14878
    },
    {
      "epoch": 0.3918619963128786,
      "grad_norm": 4.251241207122803,
      "learning_rate": 3.041875164603635e-05,
      "loss": 0.9735,
      "step": 14879
    },
    {
      "epoch": 0.3918883328943903,
      "grad_norm": 1.5219513177871704,
      "learning_rate": 3.041743481696076e-05,
      "loss": 1.7822,
      "step": 14880
    },
    {
      "epoch": 0.39191466947590203,
      "grad_norm": 1.882404088973999,
      "learning_rate": 3.0416117987885173e-05,
      "loss": 1.2649,
      "step": 14881
    },
    {
      "epoch": 0.3919410060574137,
      "grad_norm": 1.8339011669158936,
      "learning_rate": 3.0414801158809586e-05,
      "loss": 1.3799,
      "step": 14882
    },
    {
      "epoch": 0.39196734263892546,
      "grad_norm": 2.841479539871216,
      "learning_rate": 3.0413484329734005e-05,
      "loss": 1.3098,
      "step": 14883
    },
    {
      "epoch": 0.3919936792204372,
      "grad_norm": 4.686696529388428,
      "learning_rate": 3.0412167500658417e-05,
      "loss": 2.2694,
      "step": 14884
    },
    {
      "epoch": 0.3920200158019489,
      "grad_norm": 1.785043716430664,
      "learning_rate": 3.041085067158283e-05,
      "loss": 1.6564,
      "step": 14885
    },
    {
      "epoch": 0.39204635238346064,
      "grad_norm": 2.304102897644043,
      "learning_rate": 3.0409533842507245e-05,
      "loss": 0.3795,
      "step": 14886
    },
    {
      "epoch": 0.3920726889649723,
      "grad_norm": 2.1050198078155518,
      "learning_rate": 3.0408217013431657e-05,
      "loss": 1.7311,
      "step": 14887
    },
    {
      "epoch": 0.39209902554648407,
      "grad_norm": 3.097066879272461,
      "learning_rate": 3.0406900184356076e-05,
      "loss": 2.1014,
      "step": 14888
    },
    {
      "epoch": 0.3921253621279958,
      "grad_norm": 2.9771621227264404,
      "learning_rate": 3.0405583355280488e-05,
      "loss": 1.3121,
      "step": 14889
    },
    {
      "epoch": 0.3921516987095075,
      "grad_norm": 4.402443885803223,
      "learning_rate": 3.04042665262049e-05,
      "loss": 0.8534,
      "step": 14890
    },
    {
      "epoch": 0.39217803529101924,
      "grad_norm": 2.157930374145508,
      "learning_rate": 3.0402949697129312e-05,
      "loss": 1.8071,
      "step": 14891
    },
    {
      "epoch": 0.39220437187253093,
      "grad_norm": 2.9447126388549805,
      "learning_rate": 3.0401632868053725e-05,
      "loss": 1.3017,
      "step": 14892
    },
    {
      "epoch": 0.3922307084540427,
      "grad_norm": 2.579279899597168,
      "learning_rate": 3.0400316038978144e-05,
      "loss": 1.7322,
      "step": 14893
    },
    {
      "epoch": 0.39225704503555436,
      "grad_norm": 2.387075185775757,
      "learning_rate": 3.0398999209902556e-05,
      "loss": 1.056,
      "step": 14894
    },
    {
      "epoch": 0.3922833816170661,
      "grad_norm": 1.9575387239456177,
      "learning_rate": 3.0397682380826968e-05,
      "loss": 1.7838,
      "step": 14895
    },
    {
      "epoch": 0.39230971819857785,
      "grad_norm": 2.322054147720337,
      "learning_rate": 3.0396365551751384e-05,
      "loss": 1.4233,
      "step": 14896
    },
    {
      "epoch": 0.39233605478008954,
      "grad_norm": 1.781912922859192,
      "learning_rate": 3.03950487226758e-05,
      "loss": 1.4996,
      "step": 14897
    },
    {
      "epoch": 0.3923623913616013,
      "grad_norm": 2.9463624954223633,
      "learning_rate": 3.0393731893600215e-05,
      "loss": 1.6873,
      "step": 14898
    },
    {
      "epoch": 0.39238872794311297,
      "grad_norm": 1.6454142332077026,
      "learning_rate": 3.0392415064524627e-05,
      "loss": 1.287,
      "step": 14899
    },
    {
      "epoch": 0.3924150645246247,
      "grad_norm": 1.539923906326294,
      "learning_rate": 3.039109823544904e-05,
      "loss": 1.605,
      "step": 14900
    },
    {
      "epoch": 0.3924414011061364,
      "grad_norm": 2.622748613357544,
      "learning_rate": 3.038978140637345e-05,
      "loss": 1.644,
      "step": 14901
    },
    {
      "epoch": 0.39246773768764814,
      "grad_norm": 2.081028699874878,
      "learning_rate": 3.038846457729787e-05,
      "loss": 2.0464,
      "step": 14902
    },
    {
      "epoch": 0.3924940742691599,
      "grad_norm": 1.8871220350265503,
      "learning_rate": 3.0387147748222283e-05,
      "loss": 1.8112,
      "step": 14903
    },
    {
      "epoch": 0.3925204108506716,
      "grad_norm": 1.806481957435608,
      "learning_rate": 3.0385830919146695e-05,
      "loss": 0.3939,
      "step": 14904
    },
    {
      "epoch": 0.3925467474321833,
      "grad_norm": 1.8234916925430298,
      "learning_rate": 3.038451409007111e-05,
      "loss": 1.6051,
      "step": 14905
    },
    {
      "epoch": 0.392573084013695,
      "grad_norm": 5.302242279052734,
      "learning_rate": 3.0383197260995523e-05,
      "loss": 1.1637,
      "step": 14906
    },
    {
      "epoch": 0.39259942059520675,
      "grad_norm": 2.1670749187469482,
      "learning_rate": 3.038188043191994e-05,
      "loss": 1.913,
      "step": 14907
    },
    {
      "epoch": 0.39262575717671844,
      "grad_norm": 2.4837429523468018,
      "learning_rate": 3.0380563602844354e-05,
      "loss": 1.5496,
      "step": 14908
    },
    {
      "epoch": 0.3926520937582302,
      "grad_norm": 5.978394508361816,
      "learning_rate": 3.0379246773768766e-05,
      "loss": 1.972,
      "step": 14909
    },
    {
      "epoch": 0.3926784303397419,
      "grad_norm": 2.4605159759521484,
      "learning_rate": 3.0377929944693178e-05,
      "loss": 2.3206,
      "step": 14910
    },
    {
      "epoch": 0.3927047669212536,
      "grad_norm": 2.2506847381591797,
      "learning_rate": 3.0376613115617597e-05,
      "loss": 1.839,
      "step": 14911
    },
    {
      "epoch": 0.39273110350276536,
      "grad_norm": 1.8196157217025757,
      "learning_rate": 3.037529628654201e-05,
      "loss": 2.1418,
      "step": 14912
    },
    {
      "epoch": 0.39275744008427704,
      "grad_norm": 2.0755486488342285,
      "learning_rate": 3.037397945746642e-05,
      "loss": 1.9227,
      "step": 14913
    },
    {
      "epoch": 0.3927837766657888,
      "grad_norm": 2.607734441757202,
      "learning_rate": 3.0372662628390837e-05,
      "loss": 2.2429,
      "step": 14914
    },
    {
      "epoch": 0.3928101132473005,
      "grad_norm": 4.638509750366211,
      "learning_rate": 3.037134579931525e-05,
      "loss": 0.7759,
      "step": 14915
    },
    {
      "epoch": 0.3928364498288122,
      "grad_norm": 2.1411778926849365,
      "learning_rate": 3.037002897023967e-05,
      "loss": 1.6683,
      "step": 14916
    },
    {
      "epoch": 0.39286278641032396,
      "grad_norm": 2.4409782886505127,
      "learning_rate": 3.036871214116408e-05,
      "loss": 1.956,
      "step": 14917
    },
    {
      "epoch": 0.39288912299183565,
      "grad_norm": 1.7738896608352661,
      "learning_rate": 3.0367395312088493e-05,
      "loss": 2.1934,
      "step": 14918
    },
    {
      "epoch": 0.3929154595733474,
      "grad_norm": 2.4149856567382812,
      "learning_rate": 3.0366078483012905e-05,
      "loss": 1.902,
      "step": 14919
    },
    {
      "epoch": 0.3929417961548591,
      "grad_norm": 3.9030544757843018,
      "learning_rate": 3.0364761653937317e-05,
      "loss": 1.3667,
      "step": 14920
    },
    {
      "epoch": 0.3929681327363708,
      "grad_norm": 2.000345230102539,
      "learning_rate": 3.0363444824861736e-05,
      "loss": 2.022,
      "step": 14921
    },
    {
      "epoch": 0.39299446931788257,
      "grad_norm": 3.048539400100708,
      "learning_rate": 3.036212799578615e-05,
      "loss": 2.7481,
      "step": 14922
    },
    {
      "epoch": 0.39302080589939425,
      "grad_norm": 2.1777024269104004,
      "learning_rate": 3.036081116671056e-05,
      "loss": 1.5604,
      "step": 14923
    },
    {
      "epoch": 0.393047142480906,
      "grad_norm": 1.943151831626892,
      "learning_rate": 3.0359494337634976e-05,
      "loss": 2.3059,
      "step": 14924
    },
    {
      "epoch": 0.3930734790624177,
      "grad_norm": 2.3496744632720947,
      "learning_rate": 3.035817750855939e-05,
      "loss": 2.3355,
      "step": 14925
    },
    {
      "epoch": 0.39309981564392943,
      "grad_norm": 1.7827156782150269,
      "learning_rate": 3.0356860679483807e-05,
      "loss": 1.8794,
      "step": 14926
    },
    {
      "epoch": 0.3931261522254411,
      "grad_norm": 1.7456151247024536,
      "learning_rate": 3.035554385040822e-05,
      "loss": 0.5393,
      "step": 14927
    },
    {
      "epoch": 0.39315248880695286,
      "grad_norm": 1.9638248682022095,
      "learning_rate": 3.0354227021332632e-05,
      "loss": 0.2196,
      "step": 14928
    },
    {
      "epoch": 0.3931788253884646,
      "grad_norm": 2.987661123275757,
      "learning_rate": 3.0352910192257044e-05,
      "loss": 1.4612,
      "step": 14929
    },
    {
      "epoch": 0.3932051619699763,
      "grad_norm": 1.900553584098816,
      "learning_rate": 3.0351593363181463e-05,
      "loss": 1.0964,
      "step": 14930
    },
    {
      "epoch": 0.39323149855148803,
      "grad_norm": 2.4086766242980957,
      "learning_rate": 3.0350276534105875e-05,
      "loss": 1.6261,
      "step": 14931
    },
    {
      "epoch": 0.3932578351329997,
      "grad_norm": 3.8307127952575684,
      "learning_rate": 3.0348959705030287e-05,
      "loss": 0.9342,
      "step": 14932
    },
    {
      "epoch": 0.39328417171451147,
      "grad_norm": 2.4297211170196533,
      "learning_rate": 3.0347642875954703e-05,
      "loss": 1.6449,
      "step": 14933
    },
    {
      "epoch": 0.39331050829602315,
      "grad_norm": 2.250925302505493,
      "learning_rate": 3.0346326046879115e-05,
      "loss": 1.6995,
      "step": 14934
    },
    {
      "epoch": 0.3933368448775349,
      "grad_norm": 1.9429324865341187,
      "learning_rate": 3.0345009217803534e-05,
      "loss": 1.7846,
      "step": 14935
    },
    {
      "epoch": 0.39336318145904664,
      "grad_norm": 2.4213273525238037,
      "learning_rate": 3.0343692388727946e-05,
      "loss": 0.9225,
      "step": 14936
    },
    {
      "epoch": 0.39338951804055833,
      "grad_norm": 1.8437020778656006,
      "learning_rate": 3.034237555965236e-05,
      "loss": 2.2285,
      "step": 14937
    },
    {
      "epoch": 0.39341585462207007,
      "grad_norm": 2.1062188148498535,
      "learning_rate": 3.034105873057677e-05,
      "loss": 1.8727,
      "step": 14938
    },
    {
      "epoch": 0.39344219120358176,
      "grad_norm": 2.6376893520355225,
      "learning_rate": 3.0339741901501183e-05,
      "loss": 2.5793,
      "step": 14939
    },
    {
      "epoch": 0.3934685277850935,
      "grad_norm": 2.2039520740509033,
      "learning_rate": 3.0338425072425602e-05,
      "loss": 1.2864,
      "step": 14940
    },
    {
      "epoch": 0.3934948643666052,
      "grad_norm": 1.7054497003555298,
      "learning_rate": 3.0337108243350014e-05,
      "loss": 2.1886,
      "step": 14941
    },
    {
      "epoch": 0.39352120094811693,
      "grad_norm": 1.5571074485778809,
      "learning_rate": 3.0335791414274427e-05,
      "loss": 1.6261,
      "step": 14942
    },
    {
      "epoch": 0.3935475375296287,
      "grad_norm": 2.9445605278015137,
      "learning_rate": 3.0334474585198842e-05,
      "loss": 1.5774,
      "step": 14943
    },
    {
      "epoch": 0.39357387411114036,
      "grad_norm": 2.143604040145874,
      "learning_rate": 3.0333157756123258e-05,
      "loss": 1.8837,
      "step": 14944
    },
    {
      "epoch": 0.3936002106926521,
      "grad_norm": 3.8407857418060303,
      "learning_rate": 3.0331840927047673e-05,
      "loss": 1.7972,
      "step": 14945
    },
    {
      "epoch": 0.3936265472741638,
      "grad_norm": 1.8203668594360352,
      "learning_rate": 3.0330524097972086e-05,
      "loss": 2.0145,
      "step": 14946
    },
    {
      "epoch": 0.39365288385567554,
      "grad_norm": 4.857835292816162,
      "learning_rate": 3.0329207268896498e-05,
      "loss": 0.9698,
      "step": 14947
    },
    {
      "epoch": 0.3936792204371872,
      "grad_norm": 1.586806058883667,
      "learning_rate": 3.032789043982091e-05,
      "loss": 2.0293,
      "step": 14948
    },
    {
      "epoch": 0.39370555701869897,
      "grad_norm": 2.2219021320343018,
      "learning_rate": 3.032657361074533e-05,
      "loss": 1.4939,
      "step": 14949
    },
    {
      "epoch": 0.3937318936002107,
      "grad_norm": 2.316866636276245,
      "learning_rate": 3.032525678166974e-05,
      "loss": 0.7091,
      "step": 14950
    },
    {
      "epoch": 0.3937582301817224,
      "grad_norm": 1.5804301500320435,
      "learning_rate": 3.0323939952594153e-05,
      "loss": 1.2736,
      "step": 14951
    },
    {
      "epoch": 0.39378456676323415,
      "grad_norm": 1.731168508529663,
      "learning_rate": 3.032262312351857e-05,
      "loss": 1.6398,
      "step": 14952
    },
    {
      "epoch": 0.39381090334474583,
      "grad_norm": 1.931602954864502,
      "learning_rate": 3.032130629444298e-05,
      "loss": 1.4843,
      "step": 14953
    },
    {
      "epoch": 0.3938372399262576,
      "grad_norm": 3.5113279819488525,
      "learning_rate": 3.03199894653674e-05,
      "loss": 1.0426,
      "step": 14954
    },
    {
      "epoch": 0.39386357650776926,
      "grad_norm": 2.936785936355591,
      "learning_rate": 3.0318672636291812e-05,
      "loss": 0.9301,
      "step": 14955
    },
    {
      "epoch": 0.393889913089281,
      "grad_norm": 2.2373158931732178,
      "learning_rate": 3.0317355807216225e-05,
      "loss": 2.2058,
      "step": 14956
    },
    {
      "epoch": 0.39391624967079275,
      "grad_norm": 4.877676010131836,
      "learning_rate": 3.0316038978140637e-05,
      "loss": 1.7309,
      "step": 14957
    },
    {
      "epoch": 0.39394258625230444,
      "grad_norm": 2.009469985961914,
      "learning_rate": 3.031472214906505e-05,
      "loss": 0.236,
      "step": 14958
    },
    {
      "epoch": 0.3939689228338162,
      "grad_norm": 2.7458291053771973,
      "learning_rate": 3.0313405319989468e-05,
      "loss": 1.7673,
      "step": 14959
    },
    {
      "epoch": 0.39399525941532787,
      "grad_norm": 2.7475838661193848,
      "learning_rate": 3.031208849091388e-05,
      "loss": 1.792,
      "step": 14960
    },
    {
      "epoch": 0.3940215959968396,
      "grad_norm": 3.2309162616729736,
      "learning_rate": 3.0310771661838296e-05,
      "loss": 1.621,
      "step": 14961
    },
    {
      "epoch": 0.39404793257835136,
      "grad_norm": 1.9680989980697632,
      "learning_rate": 3.0309454832762708e-05,
      "loss": 1.4533,
      "step": 14962
    },
    {
      "epoch": 0.39407426915986304,
      "grad_norm": 2.2345597743988037,
      "learning_rate": 3.0308138003687127e-05,
      "loss": 1.6085,
      "step": 14963
    },
    {
      "epoch": 0.3941006057413748,
      "grad_norm": 1.9043540954589844,
      "learning_rate": 3.030682117461154e-05,
      "loss": 2.0718,
      "step": 14964
    },
    {
      "epoch": 0.3941269423228865,
      "grad_norm": 3.382739305496216,
      "learning_rate": 3.030550434553595e-05,
      "loss": 1.3709,
      "step": 14965
    },
    {
      "epoch": 0.3941532789043982,
      "grad_norm": 2.239262580871582,
      "learning_rate": 3.0304187516460364e-05,
      "loss": 0.3446,
      "step": 14966
    },
    {
      "epoch": 0.3941796154859099,
      "grad_norm": 2.3586490154266357,
      "learning_rate": 3.0302870687384776e-05,
      "loss": 1.7833,
      "step": 14967
    },
    {
      "epoch": 0.39420595206742165,
      "grad_norm": 1.9577274322509766,
      "learning_rate": 3.0301553858309195e-05,
      "loss": 1.5152,
      "step": 14968
    },
    {
      "epoch": 0.3942322886489334,
      "grad_norm": 4.146500587463379,
      "learning_rate": 3.0300237029233607e-05,
      "loss": 0.8889,
      "step": 14969
    },
    {
      "epoch": 0.3942586252304451,
      "grad_norm": 1.9502582550048828,
      "learning_rate": 3.029892020015802e-05,
      "loss": 1.7861,
      "step": 14970
    },
    {
      "epoch": 0.3942849618119568,
      "grad_norm": 2.187791585922241,
      "learning_rate": 3.0297603371082435e-05,
      "loss": 1.2702,
      "step": 14971
    },
    {
      "epoch": 0.3943112983934685,
      "grad_norm": 1.8412171602249146,
      "learning_rate": 3.0296286542006847e-05,
      "loss": 1.6453,
      "step": 14972
    },
    {
      "epoch": 0.39433763497498026,
      "grad_norm": 3.715301990509033,
      "learning_rate": 3.0294969712931266e-05,
      "loss": 2.5333,
      "step": 14973
    },
    {
      "epoch": 0.39436397155649194,
      "grad_norm": 2.007598638534546,
      "learning_rate": 3.0293652883855678e-05,
      "loss": 1.884,
      "step": 14974
    },
    {
      "epoch": 0.3943903081380037,
      "grad_norm": 2.7604873180389404,
      "learning_rate": 3.029233605478009e-05,
      "loss": 1.9422,
      "step": 14975
    },
    {
      "epoch": 0.39441664471951543,
      "grad_norm": 1.9046781063079834,
      "learning_rate": 3.0291019225704503e-05,
      "loss": 1.399,
      "step": 14976
    },
    {
      "epoch": 0.3944429813010271,
      "grad_norm": 4.05242395401001,
      "learning_rate": 3.028970239662892e-05,
      "loss": 1.2327,
      "step": 14977
    },
    {
      "epoch": 0.39446931788253886,
      "grad_norm": 2.39886212348938,
      "learning_rate": 3.0288385567553334e-05,
      "loss": 1.4573,
      "step": 14978
    },
    {
      "epoch": 0.39449565446405055,
      "grad_norm": 1.5059698820114136,
      "learning_rate": 3.0287068738477746e-05,
      "loss": 1.9572,
      "step": 14979
    },
    {
      "epoch": 0.3945219910455623,
      "grad_norm": 3.4010074138641357,
      "learning_rate": 3.028575190940216e-05,
      "loss": 0.7581,
      "step": 14980
    },
    {
      "epoch": 0.394548327627074,
      "grad_norm": 2.100843667984009,
      "learning_rate": 3.0284435080326574e-05,
      "loss": 1.8349,
      "step": 14981
    },
    {
      "epoch": 0.3945746642085857,
      "grad_norm": 3.752262592315674,
      "learning_rate": 3.0283118251250993e-05,
      "loss": 1.5736,
      "step": 14982
    },
    {
      "epoch": 0.39460100079009747,
      "grad_norm": 1.9618964195251465,
      "learning_rate": 3.0281801422175405e-05,
      "loss": 1.0923,
      "step": 14983
    },
    {
      "epoch": 0.39462733737160915,
      "grad_norm": 3.5801913738250732,
      "learning_rate": 3.0280484593099817e-05,
      "loss": 1.3707,
      "step": 14984
    },
    {
      "epoch": 0.3946536739531209,
      "grad_norm": 1.8059853315353394,
      "learning_rate": 3.027916776402423e-05,
      "loss": 2.5138,
      "step": 14985
    },
    {
      "epoch": 0.3946800105346326,
      "grad_norm": 1.8930373191833496,
      "learning_rate": 3.027785093494864e-05,
      "loss": 2.2534,
      "step": 14986
    },
    {
      "epoch": 0.39470634711614433,
      "grad_norm": 2.008183240890503,
      "learning_rate": 3.027653410587306e-05,
      "loss": 2.3735,
      "step": 14987
    },
    {
      "epoch": 0.394732683697656,
      "grad_norm": 1.706168293952942,
      "learning_rate": 3.0275217276797473e-05,
      "loss": 1.5547,
      "step": 14988
    },
    {
      "epoch": 0.39475902027916776,
      "grad_norm": 1.617963433265686,
      "learning_rate": 3.027390044772189e-05,
      "loss": 2.0334,
      "step": 14989
    },
    {
      "epoch": 0.3947853568606795,
      "grad_norm": 3.651312828063965,
      "learning_rate": 3.02725836186463e-05,
      "loss": 1.6696,
      "step": 14990
    },
    {
      "epoch": 0.3948116934421912,
      "grad_norm": 1.7130506038665771,
      "learning_rate": 3.0271266789570713e-05,
      "loss": 1.9091,
      "step": 14991
    },
    {
      "epoch": 0.39483803002370294,
      "grad_norm": 2.075068712234497,
      "learning_rate": 3.0269949960495132e-05,
      "loss": 1.3314,
      "step": 14992
    },
    {
      "epoch": 0.3948643666052146,
      "grad_norm": 3.142453193664551,
      "learning_rate": 3.0268633131419544e-05,
      "loss": 1.3264,
      "step": 14993
    },
    {
      "epoch": 0.39489070318672637,
      "grad_norm": 5.37628173828125,
      "learning_rate": 3.0267316302343956e-05,
      "loss": 1.1177,
      "step": 14994
    },
    {
      "epoch": 0.3949170397682381,
      "grad_norm": 2.0898795127868652,
      "learning_rate": 3.026599947326837e-05,
      "loss": 2.521,
      "step": 14995
    },
    {
      "epoch": 0.3949433763497498,
      "grad_norm": 2.007396936416626,
      "learning_rate": 3.0264682644192787e-05,
      "loss": 1.9554,
      "step": 14996
    },
    {
      "epoch": 0.39496971293126154,
      "grad_norm": 2.0360639095306396,
      "learning_rate": 3.02633658151172e-05,
      "loss": 1.5276,
      "step": 14997
    },
    {
      "epoch": 0.39499604951277323,
      "grad_norm": 2.260202646255493,
      "learning_rate": 3.0262048986041612e-05,
      "loss": 0.6219,
      "step": 14998
    },
    {
      "epoch": 0.39502238609428497,
      "grad_norm": 1.609933853149414,
      "learning_rate": 3.0260732156966028e-05,
      "loss": 2.3078,
      "step": 14999
    },
    {
      "epoch": 0.39504872267579666,
      "grad_norm": 2.0231568813323975,
      "learning_rate": 3.025941532789044e-05,
      "loss": 1.7696,
      "step": 15000
    },
    {
      "epoch": 0.3950750592573084,
      "grad_norm": 1.7712076902389526,
      "learning_rate": 3.025809849881486e-05,
      "loss": 2.223,
      "step": 15001
    },
    {
      "epoch": 0.39510139583882015,
      "grad_norm": 1.5881341695785522,
      "learning_rate": 3.025678166973927e-05,
      "loss": 1.6882,
      "step": 15002
    },
    {
      "epoch": 0.39512773242033183,
      "grad_norm": 4.1025238037109375,
      "learning_rate": 3.0255464840663683e-05,
      "loss": 1.678,
      "step": 15003
    },
    {
      "epoch": 0.3951540690018436,
      "grad_norm": 3.62084698677063,
      "learning_rate": 3.0254148011588095e-05,
      "loss": 1.2356,
      "step": 15004
    },
    {
      "epoch": 0.39518040558335527,
      "grad_norm": 2.124835968017578,
      "learning_rate": 3.0252831182512508e-05,
      "loss": 0.7734,
      "step": 15005
    },
    {
      "epoch": 0.395206742164867,
      "grad_norm": 2.9498844146728516,
      "learning_rate": 3.0251514353436927e-05,
      "loss": 2.7692,
      "step": 15006
    },
    {
      "epoch": 0.3952330787463787,
      "grad_norm": 1.5863209962844849,
      "learning_rate": 3.025019752436134e-05,
      "loss": 2.1596,
      "step": 15007
    },
    {
      "epoch": 0.39525941532789044,
      "grad_norm": 1.664631962776184,
      "learning_rate": 3.0248880695285754e-05,
      "loss": 1.7198,
      "step": 15008
    },
    {
      "epoch": 0.3952857519094022,
      "grad_norm": 6.359514236450195,
      "learning_rate": 3.0247563866210167e-05,
      "loss": 1.261,
      "step": 15009
    },
    {
      "epoch": 0.39531208849091387,
      "grad_norm": 2.363149404525757,
      "learning_rate": 3.0246247037134586e-05,
      "loss": 1.6975,
      "step": 15010
    },
    {
      "epoch": 0.3953384250724256,
      "grad_norm": 2.035966634750366,
      "learning_rate": 3.0244930208058998e-05,
      "loss": 2.056,
      "step": 15011
    },
    {
      "epoch": 0.3953647616539373,
      "grad_norm": 1.7164771556854248,
      "learning_rate": 3.024361337898341e-05,
      "loss": 2.0896,
      "step": 15012
    },
    {
      "epoch": 0.39539109823544905,
      "grad_norm": 1.7573095560073853,
      "learning_rate": 3.0242296549907822e-05,
      "loss": 1.5428,
      "step": 15013
    },
    {
      "epoch": 0.39541743481696073,
      "grad_norm": 1.6729694604873657,
      "learning_rate": 3.0240979720832234e-05,
      "loss": 1.9735,
      "step": 15014
    },
    {
      "epoch": 0.3954437713984725,
      "grad_norm": 3.3652641773223877,
      "learning_rate": 3.0239662891756653e-05,
      "loss": 1.4089,
      "step": 15015
    },
    {
      "epoch": 0.3954701079799842,
      "grad_norm": 2.047421455383301,
      "learning_rate": 3.0238346062681066e-05,
      "loss": 2.4291,
      "step": 15016
    },
    {
      "epoch": 0.3954964445614959,
      "grad_norm": 3.5864269733428955,
      "learning_rate": 3.0237029233605478e-05,
      "loss": 1.7278,
      "step": 15017
    },
    {
      "epoch": 0.39552278114300765,
      "grad_norm": 2.2813830375671387,
      "learning_rate": 3.0235712404529893e-05,
      "loss": 1.7869,
      "step": 15018
    },
    {
      "epoch": 0.39554911772451934,
      "grad_norm": 2.0391955375671387,
      "learning_rate": 3.0234395575454306e-05,
      "loss": 1.834,
      "step": 15019
    },
    {
      "epoch": 0.3955754543060311,
      "grad_norm": 3.667698860168457,
      "learning_rate": 3.0233078746378725e-05,
      "loss": 1.115,
      "step": 15020
    },
    {
      "epoch": 0.39560179088754277,
      "grad_norm": 2.070315361022949,
      "learning_rate": 3.0231761917303137e-05,
      "loss": 1.9976,
      "step": 15021
    },
    {
      "epoch": 0.3956281274690545,
      "grad_norm": 2.282550573348999,
      "learning_rate": 3.023044508822755e-05,
      "loss": 2.3074,
      "step": 15022
    },
    {
      "epoch": 0.39565446405056626,
      "grad_norm": 4.889825820922852,
      "learning_rate": 3.022912825915196e-05,
      "loss": 1.904,
      "step": 15023
    },
    {
      "epoch": 0.39568080063207794,
      "grad_norm": 2.5398616790771484,
      "learning_rate": 3.0227811430076373e-05,
      "loss": 2.1748,
      "step": 15024
    },
    {
      "epoch": 0.3957071372135897,
      "grad_norm": 1.7243691682815552,
      "learning_rate": 3.0226494601000792e-05,
      "loss": 1.7195,
      "step": 15025
    },
    {
      "epoch": 0.3957334737951014,
      "grad_norm": 2.2395265102386475,
      "learning_rate": 3.0225177771925205e-05,
      "loss": 1.3513,
      "step": 15026
    },
    {
      "epoch": 0.3957598103766131,
      "grad_norm": 2.8299758434295654,
      "learning_rate": 3.022386094284962e-05,
      "loss": 2.7042,
      "step": 15027
    },
    {
      "epoch": 0.39578614695812486,
      "grad_norm": 2.2655575275421143,
      "learning_rate": 3.0222544113774032e-05,
      "loss": 1.5202,
      "step": 15028
    },
    {
      "epoch": 0.39581248353963655,
      "grad_norm": 6.257394790649414,
      "learning_rate": 3.022122728469845e-05,
      "loss": 1.5074,
      "step": 15029
    },
    {
      "epoch": 0.3958388201211483,
      "grad_norm": 1.8992619514465332,
      "learning_rate": 3.0219910455622864e-05,
      "loss": 1.6583,
      "step": 15030
    },
    {
      "epoch": 0.39586515670266,
      "grad_norm": 1.9327623844146729,
      "learning_rate": 3.0218593626547276e-05,
      "loss": 1.5987,
      "step": 15031
    },
    {
      "epoch": 0.3958914932841717,
      "grad_norm": 2.0884854793548584,
      "learning_rate": 3.0217276797471688e-05,
      "loss": 0.4415,
      "step": 15032
    },
    {
      "epoch": 0.3959178298656834,
      "grad_norm": 1.670845627784729,
      "learning_rate": 3.02159599683961e-05,
      "loss": 1.833,
      "step": 15033
    },
    {
      "epoch": 0.39594416644719516,
      "grad_norm": 1.6577280759811401,
      "learning_rate": 3.021464313932052e-05,
      "loss": 1.8602,
      "step": 15034
    },
    {
      "epoch": 0.3959705030287069,
      "grad_norm": 2.7833259105682373,
      "learning_rate": 3.021332631024493e-05,
      "loss": 1.9647,
      "step": 15035
    },
    {
      "epoch": 0.3959968396102186,
      "grad_norm": 4.874198913574219,
      "learning_rate": 3.0212009481169347e-05,
      "loss": 1.8049,
      "step": 15036
    },
    {
      "epoch": 0.39602317619173033,
      "grad_norm": 3.5404245853424072,
      "learning_rate": 3.021069265209376e-05,
      "loss": 1.9333,
      "step": 15037
    },
    {
      "epoch": 0.396049512773242,
      "grad_norm": 1.8737950325012207,
      "learning_rate": 3.020937582301817e-05,
      "loss": 2.0376,
      "step": 15038
    },
    {
      "epoch": 0.39607584935475376,
      "grad_norm": 5.297119140625,
      "learning_rate": 3.020805899394259e-05,
      "loss": 2.4164,
      "step": 15039
    },
    {
      "epoch": 0.39610218593626545,
      "grad_norm": 4.721755027770996,
      "learning_rate": 3.0206742164867003e-05,
      "loss": 1.1681,
      "step": 15040
    },
    {
      "epoch": 0.3961285225177772,
      "grad_norm": 1.5547116994857788,
      "learning_rate": 3.0205425335791415e-05,
      "loss": 1.5787,
      "step": 15041
    },
    {
      "epoch": 0.39615485909928894,
      "grad_norm": 2.7580041885375977,
      "learning_rate": 3.0204108506715827e-05,
      "loss": 0.4349,
      "step": 15042
    },
    {
      "epoch": 0.3961811956808006,
      "grad_norm": 2.6049814224243164,
      "learning_rate": 3.0202791677640246e-05,
      "loss": 1.7179,
      "step": 15043
    },
    {
      "epoch": 0.39620753226231237,
      "grad_norm": 1.8934935331344604,
      "learning_rate": 3.0201474848564658e-05,
      "loss": 2.4449,
      "step": 15044
    },
    {
      "epoch": 0.39623386884382406,
      "grad_norm": 2.2490322589874268,
      "learning_rate": 3.020015801948907e-05,
      "loss": 1.9849,
      "step": 15045
    },
    {
      "epoch": 0.3962602054253358,
      "grad_norm": 1.8754563331604004,
      "learning_rate": 3.0198841190413486e-05,
      "loss": 1.9317,
      "step": 15046
    },
    {
      "epoch": 0.3962865420068475,
      "grad_norm": 1.4597707986831665,
      "learning_rate": 3.0197524361337898e-05,
      "loss": 1.562,
      "step": 15047
    },
    {
      "epoch": 0.39631287858835923,
      "grad_norm": 2.1651532649993896,
      "learning_rate": 3.0196207532262317e-05,
      "loss": 0.6542,
      "step": 15048
    },
    {
      "epoch": 0.396339215169871,
      "grad_norm": 2.0441808700561523,
      "learning_rate": 3.019489070318673e-05,
      "loss": 1.7482,
      "step": 15049
    },
    {
      "epoch": 0.39636555175138266,
      "grad_norm": 6.155063629150391,
      "learning_rate": 3.019357387411114e-05,
      "loss": 2.167,
      "step": 15050
    },
    {
      "epoch": 0.3963918883328944,
      "grad_norm": 2.0099036693573,
      "learning_rate": 3.0192257045035554e-05,
      "loss": 0.6813,
      "step": 15051
    },
    {
      "epoch": 0.3964182249144061,
      "grad_norm": 4.4810895919799805,
      "learning_rate": 3.0190940215959966e-05,
      "loss": 2.5688,
      "step": 15052
    },
    {
      "epoch": 0.39644456149591784,
      "grad_norm": 2.7615575790405273,
      "learning_rate": 3.0189623386884385e-05,
      "loss": 2.0952,
      "step": 15053
    },
    {
      "epoch": 0.3964708980774295,
      "grad_norm": 1.6449549198150635,
      "learning_rate": 3.0188306557808797e-05,
      "loss": 1.878,
      "step": 15054
    },
    {
      "epoch": 0.39649723465894127,
      "grad_norm": 2.340190887451172,
      "learning_rate": 3.0186989728733213e-05,
      "loss": 1.6676,
      "step": 15055
    },
    {
      "epoch": 0.396523571240453,
      "grad_norm": 3.026587724685669,
      "learning_rate": 3.0185672899657625e-05,
      "loss": 2.116,
      "step": 15056
    },
    {
      "epoch": 0.3965499078219647,
      "grad_norm": 1.80243718624115,
      "learning_rate": 3.0184356070582037e-05,
      "loss": 2.058,
      "step": 15057
    },
    {
      "epoch": 0.39657624440347644,
      "grad_norm": 4.1269707679748535,
      "learning_rate": 3.0183039241506456e-05,
      "loss": 2.1053,
      "step": 15058
    },
    {
      "epoch": 0.39660258098498813,
      "grad_norm": 1.9416321516036987,
      "learning_rate": 3.018172241243087e-05,
      "loss": 1.3026,
      "step": 15059
    },
    {
      "epoch": 0.3966289175664999,
      "grad_norm": 1.2719355821609497,
      "learning_rate": 3.018040558335528e-05,
      "loss": 1.7994,
      "step": 15060
    },
    {
      "epoch": 0.3966552541480116,
      "grad_norm": 2.6693902015686035,
      "learning_rate": 3.0179088754279693e-05,
      "loss": 2.3408,
      "step": 15061
    },
    {
      "epoch": 0.3966815907295233,
      "grad_norm": 2.63718843460083,
      "learning_rate": 3.0177771925204112e-05,
      "loss": 1.6426,
      "step": 15062
    },
    {
      "epoch": 0.39670792731103505,
      "grad_norm": 1.9943994283676147,
      "learning_rate": 3.0176455096128524e-05,
      "loss": 1.999,
      "step": 15063
    },
    {
      "epoch": 0.39673426389254673,
      "grad_norm": 1.724923014640808,
      "learning_rate": 3.017513826705294e-05,
      "loss": 1.8849,
      "step": 15064
    },
    {
      "epoch": 0.3967606004740585,
      "grad_norm": 1.839485764503479,
      "learning_rate": 3.0173821437977352e-05,
      "loss": 1.6697,
      "step": 15065
    },
    {
      "epoch": 0.39678693705557017,
      "grad_norm": 2.3077034950256348,
      "learning_rate": 3.0172504608901764e-05,
      "loss": 1.6201,
      "step": 15066
    },
    {
      "epoch": 0.3968132736370819,
      "grad_norm": 2.369795083999634,
      "learning_rate": 3.0171187779826183e-05,
      "loss": 2.334,
      "step": 15067
    },
    {
      "epoch": 0.39683961021859365,
      "grad_norm": 2.1524362564086914,
      "learning_rate": 3.0169870950750595e-05,
      "loss": 1.6187,
      "step": 15068
    },
    {
      "epoch": 0.39686594680010534,
      "grad_norm": 1.5614216327667236,
      "learning_rate": 3.0168554121675008e-05,
      "loss": 2.3111,
      "step": 15069
    },
    {
      "epoch": 0.3968922833816171,
      "grad_norm": 1.5796303749084473,
      "learning_rate": 3.016723729259942e-05,
      "loss": 2.2566,
      "step": 15070
    },
    {
      "epoch": 0.39691861996312877,
      "grad_norm": 2.016678810119629,
      "learning_rate": 3.0165920463523832e-05,
      "loss": 2.0844,
      "step": 15071
    },
    {
      "epoch": 0.3969449565446405,
      "grad_norm": 2.135490894317627,
      "learning_rate": 3.016460363444825e-05,
      "loss": 1.9233,
      "step": 15072
    },
    {
      "epoch": 0.3969712931261522,
      "grad_norm": 2.128314971923828,
      "learning_rate": 3.0163286805372663e-05,
      "loss": 1.8996,
      "step": 15073
    },
    {
      "epoch": 0.39699762970766395,
      "grad_norm": 4.088359832763672,
      "learning_rate": 3.016196997629708e-05,
      "loss": 2.0225,
      "step": 15074
    },
    {
      "epoch": 0.3970239662891757,
      "grad_norm": 1.775320291519165,
      "learning_rate": 3.016065314722149e-05,
      "loss": 1.7487,
      "step": 15075
    },
    {
      "epoch": 0.3970503028706874,
      "grad_norm": 1.635765552520752,
      "learning_rate": 3.015933631814591e-05,
      "loss": 1.8751,
      "step": 15076
    },
    {
      "epoch": 0.3970766394521991,
      "grad_norm": 3.182035446166992,
      "learning_rate": 3.0158019489070322e-05,
      "loss": 2.136,
      "step": 15077
    },
    {
      "epoch": 0.3971029760337108,
      "grad_norm": 1.9329135417938232,
      "learning_rate": 3.0156702659994734e-05,
      "loss": 1.6874,
      "step": 15078
    },
    {
      "epoch": 0.39712931261522255,
      "grad_norm": 3.233552932739258,
      "learning_rate": 3.0155385830919147e-05,
      "loss": 1.498,
      "step": 15079
    },
    {
      "epoch": 0.39715564919673424,
      "grad_norm": 3.0069446563720703,
      "learning_rate": 3.015406900184356e-05,
      "loss": 0.4941,
      "step": 15080
    },
    {
      "epoch": 0.397181985778246,
      "grad_norm": 2.362168550491333,
      "learning_rate": 3.0152752172767978e-05,
      "loss": 0.9619,
      "step": 15081
    },
    {
      "epoch": 0.3972083223597577,
      "grad_norm": 2.007009983062744,
      "learning_rate": 3.015143534369239e-05,
      "loss": 2.4339,
      "step": 15082
    },
    {
      "epoch": 0.3972346589412694,
      "grad_norm": 1.9347221851348877,
      "learning_rate": 3.0150118514616806e-05,
      "loss": 2.0082,
      "step": 15083
    },
    {
      "epoch": 0.39726099552278116,
      "grad_norm": 2.677497625350952,
      "learning_rate": 3.0148801685541218e-05,
      "loss": 1.3221,
      "step": 15084
    },
    {
      "epoch": 0.39728733210429285,
      "grad_norm": 2.126732349395752,
      "learning_rate": 3.014748485646563e-05,
      "loss": 2.0514,
      "step": 15085
    },
    {
      "epoch": 0.3973136686858046,
      "grad_norm": 2.031367063522339,
      "learning_rate": 3.014616802739005e-05,
      "loss": 1.9752,
      "step": 15086
    },
    {
      "epoch": 0.3973400052673163,
      "grad_norm": 1.534645676612854,
      "learning_rate": 3.014485119831446e-05,
      "loss": 1.7627,
      "step": 15087
    },
    {
      "epoch": 0.397366341848828,
      "grad_norm": 1.8645728826522827,
      "learning_rate": 3.0143534369238873e-05,
      "loss": 0.52,
      "step": 15088
    },
    {
      "epoch": 0.39739267843033976,
      "grad_norm": 2.380117416381836,
      "learning_rate": 3.0142217540163286e-05,
      "loss": 2.1512,
      "step": 15089
    },
    {
      "epoch": 0.39741901501185145,
      "grad_norm": 3.748344898223877,
      "learning_rate": 3.01409007110877e-05,
      "loss": 0.59,
      "step": 15090
    },
    {
      "epoch": 0.3974453515933632,
      "grad_norm": 1.996188759803772,
      "learning_rate": 3.0139583882012117e-05,
      "loss": 1.8976,
      "step": 15091
    },
    {
      "epoch": 0.3974716881748749,
      "grad_norm": 2.647844076156616,
      "learning_rate": 3.0138267052936532e-05,
      "loss": 2.2353,
      "step": 15092
    },
    {
      "epoch": 0.3974980247563866,
      "grad_norm": 2.239309310913086,
      "learning_rate": 3.0136950223860945e-05,
      "loss": 2.1724,
      "step": 15093
    },
    {
      "epoch": 0.3975243613378983,
      "grad_norm": 1.5903981924057007,
      "learning_rate": 3.0135633394785357e-05,
      "loss": 1.6906,
      "step": 15094
    },
    {
      "epoch": 0.39755069791941006,
      "grad_norm": 1.4769396781921387,
      "learning_rate": 3.0134316565709776e-05,
      "loss": 2.3228,
      "step": 15095
    },
    {
      "epoch": 0.3975770345009218,
      "grad_norm": 1.7581170797348022,
      "learning_rate": 3.0132999736634188e-05,
      "loss": 0.5144,
      "step": 15096
    },
    {
      "epoch": 0.3976033710824335,
      "grad_norm": 2.6350605487823486,
      "learning_rate": 3.01316829075586e-05,
      "loss": 1.6155,
      "step": 15097
    },
    {
      "epoch": 0.39762970766394523,
      "grad_norm": 1.9088119268417358,
      "learning_rate": 3.0130366078483012e-05,
      "loss": 1.7295,
      "step": 15098
    },
    {
      "epoch": 0.3976560442454569,
      "grad_norm": 1.4263521432876587,
      "learning_rate": 3.0129049249407425e-05,
      "loss": 2.1658,
      "step": 15099
    },
    {
      "epoch": 0.39768238082696866,
      "grad_norm": 1.822358250617981,
      "learning_rate": 3.0127732420331844e-05,
      "loss": 1.9254,
      "step": 15100
    },
    {
      "epoch": 0.3977087174084804,
      "grad_norm": 1.7165281772613525,
      "learning_rate": 3.0126415591256256e-05,
      "loss": 1.6772,
      "step": 15101
    },
    {
      "epoch": 0.3977350539899921,
      "grad_norm": 2.9805707931518555,
      "learning_rate": 3.012509876218067e-05,
      "loss": 1.379,
      "step": 15102
    },
    {
      "epoch": 0.39776139057150384,
      "grad_norm": 2.0595970153808594,
      "learning_rate": 3.0123781933105084e-05,
      "loss": 1.0463,
      "step": 15103
    },
    {
      "epoch": 0.3977877271530155,
      "grad_norm": 2.1131811141967773,
      "learning_rate": 3.0122465104029496e-05,
      "loss": 2.3488,
      "step": 15104
    },
    {
      "epoch": 0.39781406373452727,
      "grad_norm": 1.6708382368087769,
      "learning_rate": 3.0121148274953915e-05,
      "loss": 1.5547,
      "step": 15105
    },
    {
      "epoch": 0.39784040031603896,
      "grad_norm": 2.2232890129089355,
      "learning_rate": 3.0119831445878327e-05,
      "loss": 1.5653,
      "step": 15106
    },
    {
      "epoch": 0.3978667368975507,
      "grad_norm": 2.054849624633789,
      "learning_rate": 3.011851461680274e-05,
      "loss": 1.8846,
      "step": 15107
    },
    {
      "epoch": 0.39789307347906244,
      "grad_norm": 1.518528699874878,
      "learning_rate": 3.011719778772715e-05,
      "loss": 1.6737,
      "step": 15108
    },
    {
      "epoch": 0.39791941006057413,
      "grad_norm": 2.1234230995178223,
      "learning_rate": 3.011588095865157e-05,
      "loss": 1.5898,
      "step": 15109
    },
    {
      "epoch": 0.3979457466420859,
      "grad_norm": 2.1404612064361572,
      "learning_rate": 3.0114564129575983e-05,
      "loss": 2.6745,
      "step": 15110
    },
    {
      "epoch": 0.39797208322359756,
      "grad_norm": 1.6770737171173096,
      "learning_rate": 3.0113247300500398e-05,
      "loss": 2.4682,
      "step": 15111
    },
    {
      "epoch": 0.3979984198051093,
      "grad_norm": 2.5243773460388184,
      "learning_rate": 3.011193047142481e-05,
      "loss": 1.7645,
      "step": 15112
    },
    {
      "epoch": 0.398024756386621,
      "grad_norm": 2.369124174118042,
      "learning_rate": 3.0110613642349223e-05,
      "loss": 1.8384,
      "step": 15113
    },
    {
      "epoch": 0.39805109296813274,
      "grad_norm": 5.346293926239014,
      "learning_rate": 3.010929681327364e-05,
      "loss": 1.7805,
      "step": 15114
    },
    {
      "epoch": 0.3980774295496445,
      "grad_norm": 2.931126356124878,
      "learning_rate": 3.0107979984198054e-05,
      "loss": 1.0728,
      "step": 15115
    },
    {
      "epoch": 0.39810376613115617,
      "grad_norm": 1.9691942930221558,
      "learning_rate": 3.0106663155122466e-05,
      "loss": 1.8304,
      "step": 15116
    },
    {
      "epoch": 0.3981301027126679,
      "grad_norm": 3.2896289825439453,
      "learning_rate": 3.0105346326046878e-05,
      "loss": 1.493,
      "step": 15117
    },
    {
      "epoch": 0.3981564392941796,
      "grad_norm": 1.655034065246582,
      "learning_rate": 3.010402949697129e-05,
      "loss": 1.7281,
      "step": 15118
    },
    {
      "epoch": 0.39818277587569134,
      "grad_norm": 2.4506442546844482,
      "learning_rate": 3.010271266789571e-05,
      "loss": 1.8252,
      "step": 15119
    },
    {
      "epoch": 0.39820911245720303,
      "grad_norm": 1.6349669694900513,
      "learning_rate": 3.0101395838820122e-05,
      "loss": 1.5295,
      "step": 15120
    },
    {
      "epoch": 0.3982354490387148,
      "grad_norm": 3.4391379356384277,
      "learning_rate": 3.0100079009744537e-05,
      "loss": 2.0839,
      "step": 15121
    },
    {
      "epoch": 0.3982617856202265,
      "grad_norm": 2.0003373622894287,
      "learning_rate": 3.009876218066895e-05,
      "loss": 1.4843,
      "step": 15122
    },
    {
      "epoch": 0.3982881222017382,
      "grad_norm": 2.7001326084136963,
      "learning_rate": 3.0097445351593362e-05,
      "loss": 1.944,
      "step": 15123
    },
    {
      "epoch": 0.39831445878324995,
      "grad_norm": 1.8603417873382568,
      "learning_rate": 3.009612852251778e-05,
      "loss": 1.5126,
      "step": 15124
    },
    {
      "epoch": 0.39834079536476164,
      "grad_norm": 1.9072836637496948,
      "learning_rate": 3.0094811693442193e-05,
      "loss": 2.3098,
      "step": 15125
    },
    {
      "epoch": 0.3983671319462734,
      "grad_norm": 2.9438912868499756,
      "learning_rate": 3.0093494864366605e-05,
      "loss": 1.7904,
      "step": 15126
    },
    {
      "epoch": 0.39839346852778507,
      "grad_norm": 2.894235134124756,
      "learning_rate": 3.0092178035291017e-05,
      "loss": 1.5091,
      "step": 15127
    },
    {
      "epoch": 0.3984198051092968,
      "grad_norm": 1.6695466041564941,
      "learning_rate": 3.0090861206215436e-05,
      "loss": 1.8956,
      "step": 15128
    },
    {
      "epoch": 0.39844614169080855,
      "grad_norm": 2.9755256175994873,
      "learning_rate": 3.008954437713985e-05,
      "loss": 1.5015,
      "step": 15129
    },
    {
      "epoch": 0.39847247827232024,
      "grad_norm": 1.456181526184082,
      "learning_rate": 3.0088227548064264e-05,
      "loss": 1.9403,
      "step": 15130
    },
    {
      "epoch": 0.398498814853832,
      "grad_norm": 1.6136459112167358,
      "learning_rate": 3.0086910718988676e-05,
      "loss": 2.0042,
      "step": 15131
    },
    {
      "epoch": 0.39852515143534367,
      "grad_norm": 2.339348077774048,
      "learning_rate": 3.008559388991309e-05,
      "loss": 1.5742,
      "step": 15132
    },
    {
      "epoch": 0.3985514880168554,
      "grad_norm": 4.0757365226745605,
      "learning_rate": 3.0084277060837508e-05,
      "loss": 1.9648,
      "step": 15133
    },
    {
      "epoch": 0.39857782459836716,
      "grad_norm": 2.3704676628112793,
      "learning_rate": 3.008296023176192e-05,
      "loss": 0.6897,
      "step": 15134
    },
    {
      "epoch": 0.39860416117987885,
      "grad_norm": 4.269392967224121,
      "learning_rate": 3.0081643402686332e-05,
      "loss": 2.3461,
      "step": 15135
    },
    {
      "epoch": 0.3986304977613906,
      "grad_norm": 3.692760944366455,
      "learning_rate": 3.0080326573610744e-05,
      "loss": 0.4298,
      "step": 15136
    },
    {
      "epoch": 0.3986568343429023,
      "grad_norm": 3.4072606563568115,
      "learning_rate": 3.007900974453516e-05,
      "loss": 0.8908,
      "step": 15137
    },
    {
      "epoch": 0.398683170924414,
      "grad_norm": 1.6600042581558228,
      "learning_rate": 3.0077692915459575e-05,
      "loss": 0.7987,
      "step": 15138
    },
    {
      "epoch": 0.3987095075059257,
      "grad_norm": 2.570112466812134,
      "learning_rate": 3.007637608638399e-05,
      "loss": 1.4689,
      "step": 15139
    },
    {
      "epoch": 0.39873584408743745,
      "grad_norm": 2.197693109512329,
      "learning_rate": 3.0075059257308403e-05,
      "loss": 1.5503,
      "step": 15140
    },
    {
      "epoch": 0.3987621806689492,
      "grad_norm": 1.4908580780029297,
      "learning_rate": 3.0073742428232815e-05,
      "loss": 1.8281,
      "step": 15141
    },
    {
      "epoch": 0.3987885172504609,
      "grad_norm": 4.311302185058594,
      "learning_rate": 3.0072425599157234e-05,
      "loss": 2.1156,
      "step": 15142
    },
    {
      "epoch": 0.3988148538319726,
      "grad_norm": 1.8114593029022217,
      "learning_rate": 3.0071108770081647e-05,
      "loss": 2.2262,
      "step": 15143
    },
    {
      "epoch": 0.3988411904134843,
      "grad_norm": 2.8582775592803955,
      "learning_rate": 3.006979194100606e-05,
      "loss": 1.0121,
      "step": 15144
    },
    {
      "epoch": 0.39886752699499606,
      "grad_norm": 1.7252779006958008,
      "learning_rate": 3.006847511193047e-05,
      "loss": 1.7096,
      "step": 15145
    },
    {
      "epoch": 0.39889386357650775,
      "grad_norm": 2.47562575340271,
      "learning_rate": 3.0067158282854883e-05,
      "loss": 1.1531,
      "step": 15146
    },
    {
      "epoch": 0.3989202001580195,
      "grad_norm": 2.3413383960723877,
      "learning_rate": 3.0065841453779302e-05,
      "loss": 2.1053,
      "step": 15147
    },
    {
      "epoch": 0.39894653673953123,
      "grad_norm": 2.820797920227051,
      "learning_rate": 3.0064524624703714e-05,
      "loss": 2.2294,
      "step": 15148
    },
    {
      "epoch": 0.3989728733210429,
      "grad_norm": 2.303964138031006,
      "learning_rate": 3.006320779562813e-05,
      "loss": 1.807,
      "step": 15149
    },
    {
      "epoch": 0.39899920990255466,
      "grad_norm": 2.3117685317993164,
      "learning_rate": 3.0061890966552542e-05,
      "loss": 2.1772,
      "step": 15150
    },
    {
      "epoch": 0.39902554648406635,
      "grad_norm": 2.523735284805298,
      "learning_rate": 3.0060574137476954e-05,
      "loss": 1.8462,
      "step": 15151
    },
    {
      "epoch": 0.3990518830655781,
      "grad_norm": 1.6964685916900635,
      "learning_rate": 3.0059257308401373e-05,
      "loss": 1.3635,
      "step": 15152
    },
    {
      "epoch": 0.3990782196470898,
      "grad_norm": 3.270782709121704,
      "learning_rate": 3.0057940479325786e-05,
      "loss": 1.6391,
      "step": 15153
    },
    {
      "epoch": 0.3991045562286015,
      "grad_norm": 1.6087555885314941,
      "learning_rate": 3.0056623650250198e-05,
      "loss": 1.6876,
      "step": 15154
    },
    {
      "epoch": 0.39913089281011327,
      "grad_norm": 2.850135326385498,
      "learning_rate": 3.005530682117461e-05,
      "loss": 1.0185,
      "step": 15155
    },
    {
      "epoch": 0.39915722939162496,
      "grad_norm": 2.8274800777435303,
      "learning_rate": 3.0053989992099026e-05,
      "loss": 0.7069,
      "step": 15156
    },
    {
      "epoch": 0.3991835659731367,
      "grad_norm": 2.156588315963745,
      "learning_rate": 3.005267316302344e-05,
      "loss": 0.6161,
      "step": 15157
    },
    {
      "epoch": 0.3992099025546484,
      "grad_norm": 3.547437906265259,
      "learning_rate": 3.0051356333947857e-05,
      "loss": 1.4996,
      "step": 15158
    },
    {
      "epoch": 0.39923623913616013,
      "grad_norm": 2.4358303546905518,
      "learning_rate": 3.005003950487227e-05,
      "loss": 1.5792,
      "step": 15159
    },
    {
      "epoch": 0.3992625757176718,
      "grad_norm": 1.5521892309188843,
      "learning_rate": 3.004872267579668e-05,
      "loss": 1.9927,
      "step": 15160
    },
    {
      "epoch": 0.39928891229918356,
      "grad_norm": 1.8601088523864746,
      "learning_rate": 3.00474058467211e-05,
      "loss": 1.7021,
      "step": 15161
    },
    {
      "epoch": 0.3993152488806953,
      "grad_norm": 1.4800333976745605,
      "learning_rate": 3.0046089017645512e-05,
      "loss": 1.7235,
      "step": 15162
    },
    {
      "epoch": 0.399341585462207,
      "grad_norm": 3.543912649154663,
      "learning_rate": 3.0044772188569925e-05,
      "loss": 1.613,
      "step": 15163
    },
    {
      "epoch": 0.39936792204371874,
      "grad_norm": 1.5966562032699585,
      "learning_rate": 3.0043455359494337e-05,
      "loss": 2.065,
      "step": 15164
    },
    {
      "epoch": 0.3993942586252304,
      "grad_norm": 1.6676326990127563,
      "learning_rate": 3.0042138530418752e-05,
      "loss": 1.5868,
      "step": 15165
    },
    {
      "epoch": 0.39942059520674217,
      "grad_norm": 2.480081558227539,
      "learning_rate": 3.0040821701343168e-05,
      "loss": 2.6109,
      "step": 15166
    },
    {
      "epoch": 0.3994469317882539,
      "grad_norm": 1.7020961046218872,
      "learning_rate": 3.0039504872267584e-05,
      "loss": 1.5241,
      "step": 15167
    },
    {
      "epoch": 0.3994732683697656,
      "grad_norm": 1.8415707349777222,
      "learning_rate": 3.0038188043191996e-05,
      "loss": 1.4608,
      "step": 15168
    },
    {
      "epoch": 0.39949960495127734,
      "grad_norm": 1.535257339477539,
      "learning_rate": 3.0036871214116408e-05,
      "loss": 1.8111,
      "step": 15169
    },
    {
      "epoch": 0.39952594153278903,
      "grad_norm": 4.163630485534668,
      "learning_rate": 3.003555438504082e-05,
      "loss": 1.5191,
      "step": 15170
    },
    {
      "epoch": 0.3995522781143008,
      "grad_norm": 1.6537508964538574,
      "learning_rate": 3.003423755596524e-05,
      "loss": 1.6542,
      "step": 15171
    },
    {
      "epoch": 0.39957861469581246,
      "grad_norm": 1.6200356483459473,
      "learning_rate": 3.003292072688965e-05,
      "loss": 1.5739,
      "step": 15172
    },
    {
      "epoch": 0.3996049512773242,
      "grad_norm": 1.6518728733062744,
      "learning_rate": 3.0031603897814064e-05,
      "loss": 1.218,
      "step": 15173
    },
    {
      "epoch": 0.39963128785883595,
      "grad_norm": 2.3867006301879883,
      "learning_rate": 3.0030287068738476e-05,
      "loss": 1.5534,
      "step": 15174
    },
    {
      "epoch": 0.39965762444034764,
      "grad_norm": 3.7688419818878174,
      "learning_rate": 3.0028970239662895e-05,
      "loss": 1.92,
      "step": 15175
    },
    {
      "epoch": 0.3996839610218594,
      "grad_norm": 4.719405651092529,
      "learning_rate": 3.0027653410587307e-05,
      "loss": 1.874,
      "step": 15176
    },
    {
      "epoch": 0.39971029760337107,
      "grad_norm": 3.285053014755249,
      "learning_rate": 3.0026336581511723e-05,
      "loss": 1.3879,
      "step": 15177
    },
    {
      "epoch": 0.3997366341848828,
      "grad_norm": 2.066946029663086,
      "learning_rate": 3.0025019752436135e-05,
      "loss": 1.4685,
      "step": 15178
    },
    {
      "epoch": 0.3997629707663945,
      "grad_norm": 1.6314629316329956,
      "learning_rate": 3.0023702923360547e-05,
      "loss": 1.3335,
      "step": 15179
    },
    {
      "epoch": 0.39978930734790624,
      "grad_norm": 2.5768818855285645,
      "learning_rate": 3.0022386094284966e-05,
      "loss": 2.0223,
      "step": 15180
    },
    {
      "epoch": 0.399815643929418,
      "grad_norm": 1.4182546138763428,
      "learning_rate": 3.0021069265209378e-05,
      "loss": 2.1824,
      "step": 15181
    },
    {
      "epoch": 0.3998419805109297,
      "grad_norm": 1.720131754875183,
      "learning_rate": 3.001975243613379e-05,
      "loss": 1.7485,
      "step": 15182
    },
    {
      "epoch": 0.3998683170924414,
      "grad_norm": 2.2177376747131348,
      "learning_rate": 3.0018435607058203e-05,
      "loss": 1.6336,
      "step": 15183
    },
    {
      "epoch": 0.3998946536739531,
      "grad_norm": 1.6714279651641846,
      "learning_rate": 3.001711877798262e-05,
      "loss": 2.0179,
      "step": 15184
    },
    {
      "epoch": 0.39992099025546485,
      "grad_norm": 2.0650711059570312,
      "learning_rate": 3.0015801948907034e-05,
      "loss": 1.8289,
      "step": 15185
    },
    {
      "epoch": 0.39994732683697654,
      "grad_norm": 1.6439447402954102,
      "learning_rate": 3.001448511983145e-05,
      "loss": 1.9966,
      "step": 15186
    },
    {
      "epoch": 0.3999736634184883,
      "grad_norm": 2.4915354251861572,
      "learning_rate": 3.0013168290755862e-05,
      "loss": 1.5785,
      "step": 15187
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.703201413154602,
      "learning_rate": 3.0011851461680274e-05,
      "loss": 0.9085,
      "step": 15188
    },
    {
      "epoch": 0.4000263365815117,
      "grad_norm": 2.088538646697998,
      "learning_rate": 3.0010534632604693e-05,
      "loss": 1.737,
      "step": 15189
    },
    {
      "epoch": 0.40005267316302345,
      "grad_norm": 1.7217639684677124,
      "learning_rate": 3.0009217803529105e-05,
      "loss": 1.2694,
      "step": 15190
    },
    {
      "epoch": 0.40007900974453514,
      "grad_norm": 3.157841205596924,
      "learning_rate": 3.0007900974453517e-05,
      "loss": 0.91,
      "step": 15191
    },
    {
      "epoch": 0.4001053463260469,
      "grad_norm": 2.1383683681488037,
      "learning_rate": 3.000658414537793e-05,
      "loss": 2.0331,
      "step": 15192
    },
    {
      "epoch": 0.4001316829075586,
      "grad_norm": 2.416210174560547,
      "learning_rate": 3.0005267316302342e-05,
      "loss": 1.6528,
      "step": 15193
    },
    {
      "epoch": 0.4001580194890703,
      "grad_norm": 2.38584041595459,
      "learning_rate": 3.000395048722676e-05,
      "loss": 1.9925,
      "step": 15194
    },
    {
      "epoch": 0.40018435607058206,
      "grad_norm": 5.951260566711426,
      "learning_rate": 3.0002633658151176e-05,
      "loss": 2.1103,
      "step": 15195
    },
    {
      "epoch": 0.40021069265209375,
      "grad_norm": 2.055037021636963,
      "learning_rate": 3.000131682907559e-05,
      "loss": 1.4111,
      "step": 15196
    },
    {
      "epoch": 0.4002370292336055,
      "grad_norm": 1.6284364461898804,
      "learning_rate": 3e-05,
      "loss": 2.157,
      "step": 15197
    },
    {
      "epoch": 0.4002633658151172,
      "grad_norm": 1.3630836009979248,
      "learning_rate": 2.9998683170924413e-05,
      "loss": 1.8156,
      "step": 15198
    },
    {
      "epoch": 0.4002897023966289,
      "grad_norm": 1.895262360572815,
      "learning_rate": 2.9997366341848832e-05,
      "loss": 1.7106,
      "step": 15199
    },
    {
      "epoch": 0.40031603897814066,
      "grad_norm": 1.7643390893936157,
      "learning_rate": 2.9996049512773244e-05,
      "loss": 2.2111,
      "step": 15200
    },
    {
      "epoch": 0.40034237555965235,
      "grad_norm": 2.426508665084839,
      "learning_rate": 2.9994732683697656e-05,
      "loss": 0.8327,
      "step": 15201
    },
    {
      "epoch": 0.4003687121411641,
      "grad_norm": 2.0043630599975586,
      "learning_rate": 2.999341585462207e-05,
      "loss": 1.8773,
      "step": 15202
    },
    {
      "epoch": 0.4003950487226758,
      "grad_norm": 3.117605447769165,
      "learning_rate": 2.9992099025546484e-05,
      "loss": 0.627,
      "step": 15203
    },
    {
      "epoch": 0.4004213853041875,
      "grad_norm": 2.4989943504333496,
      "learning_rate": 2.99907821964709e-05,
      "loss": 1.8619,
      "step": 15204
    },
    {
      "epoch": 0.4004477218856992,
      "grad_norm": 3.1154608726501465,
      "learning_rate": 2.9989465367395315e-05,
      "loss": 0.9862,
      "step": 15205
    },
    {
      "epoch": 0.40047405846721096,
      "grad_norm": 1.5244344472885132,
      "learning_rate": 2.9988148538319728e-05,
      "loss": 2.279,
      "step": 15206
    },
    {
      "epoch": 0.4005003950487227,
      "grad_norm": 1.8436567783355713,
      "learning_rate": 2.998683170924414e-05,
      "loss": 1.3582,
      "step": 15207
    },
    {
      "epoch": 0.4005267316302344,
      "grad_norm": 2.4069817066192627,
      "learning_rate": 2.998551488016856e-05,
      "loss": 1.9272,
      "step": 15208
    },
    {
      "epoch": 0.40055306821174613,
      "grad_norm": 3.6580328941345215,
      "learning_rate": 2.998419805109297e-05,
      "loss": 1.7492,
      "step": 15209
    },
    {
      "epoch": 0.4005794047932578,
      "grad_norm": 2.241776943206787,
      "learning_rate": 2.9982881222017383e-05,
      "loss": 1.7254,
      "step": 15210
    },
    {
      "epoch": 0.40060574137476956,
      "grad_norm": 2.118487596511841,
      "learning_rate": 2.9981564392941795e-05,
      "loss": 1.2622,
      "step": 15211
    },
    {
      "epoch": 0.40063207795628125,
      "grad_norm": 1.6129636764526367,
      "learning_rate": 2.998024756386621e-05,
      "loss": 1.9641,
      "step": 15212
    },
    {
      "epoch": 0.400658414537793,
      "grad_norm": 2.9925856590270996,
      "learning_rate": 2.9978930734790627e-05,
      "loss": 1.3286,
      "step": 15213
    },
    {
      "epoch": 0.40068475111930474,
      "grad_norm": 2.6571152210235596,
      "learning_rate": 2.9977613905715042e-05,
      "loss": 2.051,
      "step": 15214
    },
    {
      "epoch": 0.4007110877008164,
      "grad_norm": 2.9185569286346436,
      "learning_rate": 2.9976297076639454e-05,
      "loss": 0.8716,
      "step": 15215
    },
    {
      "epoch": 0.40073742428232817,
      "grad_norm": 2.085442304611206,
      "learning_rate": 2.9974980247563867e-05,
      "loss": 1.7261,
      "step": 15216
    },
    {
      "epoch": 0.40076376086383986,
      "grad_norm": 1.9198839664459229,
      "learning_rate": 2.997366341848828e-05,
      "loss": 2.0768,
      "step": 15217
    },
    {
      "epoch": 0.4007900974453516,
      "grad_norm": 3.3180854320526123,
      "learning_rate": 2.9972346589412698e-05,
      "loss": 1.3378,
      "step": 15218
    },
    {
      "epoch": 0.4008164340268633,
      "grad_norm": 1.9435937404632568,
      "learning_rate": 2.997102976033711e-05,
      "loss": 2.0091,
      "step": 15219
    },
    {
      "epoch": 0.40084277060837503,
      "grad_norm": 1.6458818912506104,
      "learning_rate": 2.9969712931261522e-05,
      "loss": 1.8368,
      "step": 15220
    },
    {
      "epoch": 0.4008691071898868,
      "grad_norm": 3.5103559494018555,
      "learning_rate": 2.9968396102185934e-05,
      "loss": 0.669,
      "step": 15221
    },
    {
      "epoch": 0.40089544377139846,
      "grad_norm": 2.4205689430236816,
      "learning_rate": 2.9967079273110353e-05,
      "loss": 1.717,
      "step": 15222
    },
    {
      "epoch": 0.4009217803529102,
      "grad_norm": 4.01300573348999,
      "learning_rate": 2.9965762444034766e-05,
      "loss": 1.4283,
      "step": 15223
    },
    {
      "epoch": 0.4009481169344219,
      "grad_norm": 5.249772548675537,
      "learning_rate": 2.996444561495918e-05,
      "loss": 0.4765,
      "step": 15224
    },
    {
      "epoch": 0.40097445351593364,
      "grad_norm": 2.009230375289917,
      "learning_rate": 2.9963128785883593e-05,
      "loss": 2.1892,
      "step": 15225
    },
    {
      "epoch": 0.4010007900974453,
      "grad_norm": 1.9879140853881836,
      "learning_rate": 2.9961811956808006e-05,
      "loss": 1.4495,
      "step": 15226
    },
    {
      "epoch": 0.40102712667895707,
      "grad_norm": 1.5879912376403809,
      "learning_rate": 2.9960495127732425e-05,
      "loss": 1.3655,
      "step": 15227
    },
    {
      "epoch": 0.4010534632604688,
      "grad_norm": 3.1911466121673584,
      "learning_rate": 2.9959178298656837e-05,
      "loss": 1.2413,
      "step": 15228
    },
    {
      "epoch": 0.4010797998419805,
      "grad_norm": 1.970922827720642,
      "learning_rate": 2.995786146958125e-05,
      "loss": 0.3081,
      "step": 15229
    },
    {
      "epoch": 0.40110613642349224,
      "grad_norm": 2.2098400592803955,
      "learning_rate": 2.995654464050566e-05,
      "loss": 1.5208,
      "step": 15230
    },
    {
      "epoch": 0.40113247300500393,
      "grad_norm": 5.250458717346191,
      "learning_rate": 2.9955227811430077e-05,
      "loss": 2.4769,
      "step": 15231
    },
    {
      "epoch": 0.4011588095865157,
      "grad_norm": 2.2800164222717285,
      "learning_rate": 2.9953910982354492e-05,
      "loss": 1.538,
      "step": 15232
    },
    {
      "epoch": 0.4011851461680274,
      "grad_norm": 3.3152554035186768,
      "learning_rate": 2.9952594153278908e-05,
      "loss": 1.314,
      "step": 15233
    },
    {
      "epoch": 0.4012114827495391,
      "grad_norm": 2.258779764175415,
      "learning_rate": 2.995127732420332e-05,
      "loss": 1.8959,
      "step": 15234
    },
    {
      "epoch": 0.40123781933105085,
      "grad_norm": 2.482424259185791,
      "learning_rate": 2.9949960495127732e-05,
      "loss": 1.1497,
      "step": 15235
    },
    {
      "epoch": 0.40126415591256254,
      "grad_norm": 8.307866096496582,
      "learning_rate": 2.9948643666052145e-05,
      "loss": 1.9556,
      "step": 15236
    },
    {
      "epoch": 0.4012904924940743,
      "grad_norm": 2.1465039253234863,
      "learning_rate": 2.9947326836976564e-05,
      "loss": 1.9607,
      "step": 15237
    },
    {
      "epoch": 0.40131682907558597,
      "grad_norm": 2.1289169788360596,
      "learning_rate": 2.9946010007900976e-05,
      "loss": 1.322,
      "step": 15238
    },
    {
      "epoch": 0.4013431656570977,
      "grad_norm": 2.1380739212036133,
      "learning_rate": 2.9944693178825388e-05,
      "loss": 2.5748,
      "step": 15239
    },
    {
      "epoch": 0.40136950223860945,
      "grad_norm": 2.718348264694214,
      "learning_rate": 2.9943376349749804e-05,
      "loss": 1.8643,
      "step": 15240
    },
    {
      "epoch": 0.40139583882012114,
      "grad_norm": 1.6842036247253418,
      "learning_rate": 2.994205952067422e-05,
      "loss": 1.9707,
      "step": 15241
    },
    {
      "epoch": 0.4014221754016329,
      "grad_norm": 1.3637287616729736,
      "learning_rate": 2.9940742691598635e-05,
      "loss": 1.8458,
      "step": 15242
    },
    {
      "epoch": 0.4014485119831446,
      "grad_norm": 1.6709080934524536,
      "learning_rate": 2.9939425862523047e-05,
      "loss": 1.5773,
      "step": 15243
    },
    {
      "epoch": 0.4014748485646563,
      "grad_norm": 1.8406020402908325,
      "learning_rate": 2.993810903344746e-05,
      "loss": 1.7483,
      "step": 15244
    },
    {
      "epoch": 0.401501185146168,
      "grad_norm": 2.2115960121154785,
      "learning_rate": 2.993679220437187e-05,
      "loss": 1.9933,
      "step": 15245
    },
    {
      "epoch": 0.40152752172767975,
      "grad_norm": 2.0160882472991943,
      "learning_rate": 2.993547537529629e-05,
      "loss": 2.0755,
      "step": 15246
    },
    {
      "epoch": 0.4015538583091915,
      "grad_norm": 1.9304190874099731,
      "learning_rate": 2.9934158546220703e-05,
      "loss": 2.1465,
      "step": 15247
    },
    {
      "epoch": 0.4015801948907032,
      "grad_norm": 1.6468080282211304,
      "learning_rate": 2.9932841717145115e-05,
      "loss": 2.1391,
      "step": 15248
    },
    {
      "epoch": 0.4016065314722149,
      "grad_norm": 2.487684488296509,
      "learning_rate": 2.9931524888069527e-05,
      "loss": 1.9947,
      "step": 15249
    },
    {
      "epoch": 0.4016328680537266,
      "grad_norm": 2.044767141342163,
      "learning_rate": 2.9930208058993943e-05,
      "loss": 2.3249,
      "step": 15250
    },
    {
      "epoch": 0.40165920463523835,
      "grad_norm": 2.7380313873291016,
      "learning_rate": 2.992889122991836e-05,
      "loss": 1.1538,
      "step": 15251
    },
    {
      "epoch": 0.40168554121675004,
      "grad_norm": 2.301825523376465,
      "learning_rate": 2.9927574400842774e-05,
      "loss": 2.036,
      "step": 15252
    },
    {
      "epoch": 0.4017118777982618,
      "grad_norm": 1.6190699338912964,
      "learning_rate": 2.9926257571767186e-05,
      "loss": 0.4114,
      "step": 15253
    },
    {
      "epoch": 0.40173821437977353,
      "grad_norm": 2.370206594467163,
      "learning_rate": 2.99249407426916e-05,
      "loss": 1.871,
      "step": 15254
    },
    {
      "epoch": 0.4017645509612852,
      "grad_norm": 2.315791130065918,
      "learning_rate": 2.9923623913616017e-05,
      "loss": 2.0115,
      "step": 15255
    },
    {
      "epoch": 0.40179088754279696,
      "grad_norm": 2.0392849445343018,
      "learning_rate": 2.992230708454043e-05,
      "loss": 1.7318,
      "step": 15256
    },
    {
      "epoch": 0.40181722412430865,
      "grad_norm": 1.7041723728179932,
      "learning_rate": 2.9920990255464842e-05,
      "loss": 1.7035,
      "step": 15257
    },
    {
      "epoch": 0.4018435607058204,
      "grad_norm": 2.8666014671325684,
      "learning_rate": 2.9919673426389254e-05,
      "loss": 1.525,
      "step": 15258
    },
    {
      "epoch": 0.4018698972873321,
      "grad_norm": 1.810825228691101,
      "learning_rate": 2.991835659731367e-05,
      "loss": 1.3962,
      "step": 15259
    },
    {
      "epoch": 0.4018962338688438,
      "grad_norm": 1.681111216545105,
      "learning_rate": 2.9917039768238085e-05,
      "loss": 1.0429,
      "step": 15260
    },
    {
      "epoch": 0.40192257045035557,
      "grad_norm": 1.6370668411254883,
      "learning_rate": 2.99157229391625e-05,
      "loss": 0.6586,
      "step": 15261
    },
    {
      "epoch": 0.40194890703186725,
      "grad_norm": 1.371125340461731,
      "learning_rate": 2.9914406110086913e-05,
      "loss": 1.4529,
      "step": 15262
    },
    {
      "epoch": 0.401975243613379,
      "grad_norm": 1.7869479656219482,
      "learning_rate": 2.9913089281011325e-05,
      "loss": 1.4347,
      "step": 15263
    },
    {
      "epoch": 0.4020015801948907,
      "grad_norm": 4.511172771453857,
      "learning_rate": 2.9911772451935737e-05,
      "loss": 1.6509,
      "step": 15264
    },
    {
      "epoch": 0.4020279167764024,
      "grad_norm": 1.900025725364685,
      "learning_rate": 2.9910455622860156e-05,
      "loss": 1.9286,
      "step": 15265
    },
    {
      "epoch": 0.4020542533579141,
      "grad_norm": 1.5888254642486572,
      "learning_rate": 2.990913879378457e-05,
      "loss": 1.3138,
      "step": 15266
    },
    {
      "epoch": 0.40208058993942586,
      "grad_norm": 1.9465323686599731,
      "learning_rate": 2.990782196470898e-05,
      "loss": 0.7556,
      "step": 15267
    },
    {
      "epoch": 0.4021069265209376,
      "grad_norm": 6.241235733032227,
      "learning_rate": 2.9906505135633396e-05,
      "loss": 1.1109,
      "step": 15268
    },
    {
      "epoch": 0.4021332631024493,
      "grad_norm": 1.9666777849197388,
      "learning_rate": 2.990518830655781e-05,
      "loss": 2.4749,
      "step": 15269
    },
    {
      "epoch": 0.40215959968396103,
      "grad_norm": 2.5248916149139404,
      "learning_rate": 2.9903871477482228e-05,
      "loss": 2.0952,
      "step": 15270
    },
    {
      "epoch": 0.4021859362654727,
      "grad_norm": 1.5898040533065796,
      "learning_rate": 2.990255464840664e-05,
      "loss": 2.0229,
      "step": 15271
    },
    {
      "epoch": 0.40221227284698446,
      "grad_norm": 1.7938041687011719,
      "learning_rate": 2.9901237819331052e-05,
      "loss": 1.8544,
      "step": 15272
    },
    {
      "epoch": 0.4022386094284962,
      "grad_norm": 1.9576348066329956,
      "learning_rate": 2.9899920990255464e-05,
      "loss": 2.3689,
      "step": 15273
    },
    {
      "epoch": 0.4022649460100079,
      "grad_norm": 2.5532422065734863,
      "learning_rate": 2.9898604161179883e-05,
      "loss": 1.3425,
      "step": 15274
    },
    {
      "epoch": 0.40229128259151964,
      "grad_norm": 2.293853282928467,
      "learning_rate": 2.9897287332104295e-05,
      "loss": 0.9007,
      "step": 15275
    },
    {
      "epoch": 0.4023176191730313,
      "grad_norm": 3.9450156688690186,
      "learning_rate": 2.9895970503028708e-05,
      "loss": 1.0792,
      "step": 15276
    },
    {
      "epoch": 0.40234395575454307,
      "grad_norm": 5.532327175140381,
      "learning_rate": 2.989465367395312e-05,
      "loss": 1.6955,
      "step": 15277
    },
    {
      "epoch": 0.40237029233605476,
      "grad_norm": 1.4562660455703735,
      "learning_rate": 2.9893336844877535e-05,
      "loss": 2.1516,
      "step": 15278
    },
    {
      "epoch": 0.4023966289175665,
      "grad_norm": 2.0622482299804688,
      "learning_rate": 2.989202001580195e-05,
      "loss": 1.4577,
      "step": 15279
    },
    {
      "epoch": 0.40242296549907824,
      "grad_norm": 1.4985934495925903,
      "learning_rate": 2.9890703186726367e-05,
      "loss": 0.6312,
      "step": 15280
    },
    {
      "epoch": 0.40244930208058993,
      "grad_norm": 2.648171901702881,
      "learning_rate": 2.988938635765078e-05,
      "loss": 2.422,
      "step": 15281
    },
    {
      "epoch": 0.4024756386621017,
      "grad_norm": 1.9110594987869263,
      "learning_rate": 2.988806952857519e-05,
      "loss": 1.6282,
      "step": 15282
    },
    {
      "epoch": 0.40250197524361336,
      "grad_norm": 1.717250943183899,
      "learning_rate": 2.9886752699499603e-05,
      "loss": 1.9295,
      "step": 15283
    },
    {
      "epoch": 0.4025283118251251,
      "grad_norm": 2.044684648513794,
      "learning_rate": 2.9885435870424022e-05,
      "loss": 1.5773,
      "step": 15284
    },
    {
      "epoch": 0.4025546484066368,
      "grad_norm": 2.2607014179229736,
      "learning_rate": 2.9884119041348434e-05,
      "loss": 2.1883,
      "step": 15285
    },
    {
      "epoch": 0.40258098498814854,
      "grad_norm": 7.618353843688965,
      "learning_rate": 2.9882802212272847e-05,
      "loss": 1.602,
      "step": 15286
    },
    {
      "epoch": 0.4026073215696603,
      "grad_norm": 1.863555908203125,
      "learning_rate": 2.9881485383197262e-05,
      "loss": 2.0446,
      "step": 15287
    },
    {
      "epoch": 0.40263365815117197,
      "grad_norm": 3.020829200744629,
      "learning_rate": 2.9880168554121678e-05,
      "loss": 1.3902,
      "step": 15288
    },
    {
      "epoch": 0.4026599947326837,
      "grad_norm": 1.7568062543869019,
      "learning_rate": 2.9878851725046093e-05,
      "loss": 2.0562,
      "step": 15289
    },
    {
      "epoch": 0.4026863313141954,
      "grad_norm": 2.7996366024017334,
      "learning_rate": 2.9877534895970506e-05,
      "loss": 0.5481,
      "step": 15290
    },
    {
      "epoch": 0.40271266789570714,
      "grad_norm": 3.4967260360717773,
      "learning_rate": 2.9876218066894918e-05,
      "loss": 1.2692,
      "step": 15291
    },
    {
      "epoch": 0.40273900447721883,
      "grad_norm": 1.4886181354522705,
      "learning_rate": 2.987490123781933e-05,
      "loss": 1.8797,
      "step": 15292
    },
    {
      "epoch": 0.4027653410587306,
      "grad_norm": 3.722468137741089,
      "learning_rate": 2.987358440874375e-05,
      "loss": 0.983,
      "step": 15293
    },
    {
      "epoch": 0.4027916776402423,
      "grad_norm": 2.472588539123535,
      "learning_rate": 2.987226757966816e-05,
      "loss": 1.3117,
      "step": 15294
    },
    {
      "epoch": 0.402818014221754,
      "grad_norm": 5.674534320831299,
      "learning_rate": 2.9870950750592573e-05,
      "loss": 1.8303,
      "step": 15295
    },
    {
      "epoch": 0.40284435080326575,
      "grad_norm": 2.336442708969116,
      "learning_rate": 2.9869633921516986e-05,
      "loss": 2.1713,
      "step": 15296
    },
    {
      "epoch": 0.40287068738477744,
      "grad_norm": 3.656493663787842,
      "learning_rate": 2.98683170924414e-05,
      "loss": 1.4337,
      "step": 15297
    },
    {
      "epoch": 0.4028970239662892,
      "grad_norm": 1.4978317022323608,
      "learning_rate": 2.9867000263365817e-05,
      "loss": 1.9913,
      "step": 15298
    },
    {
      "epoch": 0.40292336054780087,
      "grad_norm": 2.005683183670044,
      "learning_rate": 2.9865683434290232e-05,
      "loss": 1.8837,
      "step": 15299
    },
    {
      "epoch": 0.4029496971293126,
      "grad_norm": 2.178419351577759,
      "learning_rate": 2.9864366605214645e-05,
      "loss": 0.7118,
      "step": 15300
    },
    {
      "epoch": 0.40297603371082436,
      "grad_norm": 2.4660890102386475,
      "learning_rate": 2.9863049776139057e-05,
      "loss": 1.3228,
      "step": 15301
    },
    {
      "epoch": 0.40300237029233604,
      "grad_norm": 2.8309640884399414,
      "learning_rate": 2.986173294706347e-05,
      "loss": 1.9071,
      "step": 15302
    },
    {
      "epoch": 0.4030287068738478,
      "grad_norm": 2.135223388671875,
      "learning_rate": 2.9860416117987888e-05,
      "loss": 1.7802,
      "step": 15303
    },
    {
      "epoch": 0.4030550434553595,
      "grad_norm": 2.9628942012786865,
      "learning_rate": 2.98590992889123e-05,
      "loss": 1.1917,
      "step": 15304
    },
    {
      "epoch": 0.4030813800368712,
      "grad_norm": 2.924274206161499,
      "learning_rate": 2.9857782459836713e-05,
      "loss": 0.712,
      "step": 15305
    },
    {
      "epoch": 0.40310771661838296,
      "grad_norm": 1.6790590286254883,
      "learning_rate": 2.9856465630761128e-05,
      "loss": 1.4436,
      "step": 15306
    },
    {
      "epoch": 0.40313405319989465,
      "grad_norm": 1.6263816356658936,
      "learning_rate": 2.9855148801685544e-05,
      "loss": 1.7542,
      "step": 15307
    },
    {
      "epoch": 0.4031603897814064,
      "grad_norm": 1.7824804782867432,
      "learning_rate": 2.985383197260996e-05,
      "loss": 2.2186,
      "step": 15308
    },
    {
      "epoch": 0.4031867263629181,
      "grad_norm": 1.734452724456787,
      "learning_rate": 2.985251514353437e-05,
      "loss": 0.43,
      "step": 15309
    },
    {
      "epoch": 0.4032130629444298,
      "grad_norm": 2.0390615463256836,
      "learning_rate": 2.9851198314458784e-05,
      "loss": 1.274,
      "step": 15310
    },
    {
      "epoch": 0.4032393995259415,
      "grad_norm": 2.4935436248779297,
      "learning_rate": 2.9849881485383196e-05,
      "loss": 0.32,
      "step": 15311
    },
    {
      "epoch": 0.40326573610745325,
      "grad_norm": 2.8859407901763916,
      "learning_rate": 2.9848564656307615e-05,
      "loss": 1.7137,
      "step": 15312
    },
    {
      "epoch": 0.403292072688965,
      "grad_norm": 2.611424446105957,
      "learning_rate": 2.9847247827232027e-05,
      "loss": 1.5125,
      "step": 15313
    },
    {
      "epoch": 0.4033184092704767,
      "grad_norm": 1.930966854095459,
      "learning_rate": 2.984593099815644e-05,
      "loss": 1.6543,
      "step": 15314
    },
    {
      "epoch": 0.40334474585198843,
      "grad_norm": 2.146080255508423,
      "learning_rate": 2.9844614169080855e-05,
      "loss": 1.751,
      "step": 15315
    },
    {
      "epoch": 0.4033710824335001,
      "grad_norm": 2.155388355255127,
      "learning_rate": 2.9843297340005267e-05,
      "loss": 1.1706,
      "step": 15316
    },
    {
      "epoch": 0.40339741901501186,
      "grad_norm": 3.0469207763671875,
      "learning_rate": 2.9841980510929686e-05,
      "loss": 2.4527,
      "step": 15317
    },
    {
      "epoch": 0.40342375559652355,
      "grad_norm": 2.508324146270752,
      "learning_rate": 2.98406636818541e-05,
      "loss": 2.2702,
      "step": 15318
    },
    {
      "epoch": 0.4034500921780353,
      "grad_norm": 2.140326976776123,
      "learning_rate": 2.983934685277851e-05,
      "loss": 1.7395,
      "step": 15319
    },
    {
      "epoch": 0.40347642875954703,
      "grad_norm": 2.279519557952881,
      "learning_rate": 2.9838030023702923e-05,
      "loss": 1.9951,
      "step": 15320
    },
    {
      "epoch": 0.4035027653410587,
      "grad_norm": 2.0406579971313477,
      "learning_rate": 2.9836713194627342e-05,
      "loss": 2.4102,
      "step": 15321
    },
    {
      "epoch": 0.40352910192257047,
      "grad_norm": 1.907092571258545,
      "learning_rate": 2.9835396365551754e-05,
      "loss": 1.8873,
      "step": 15322
    },
    {
      "epoch": 0.40355543850408215,
      "grad_norm": 2.1900689601898193,
      "learning_rate": 2.9834079536476166e-05,
      "loss": 1.9562,
      "step": 15323
    },
    {
      "epoch": 0.4035817750855939,
      "grad_norm": 1.8613706827163696,
      "learning_rate": 2.983276270740058e-05,
      "loss": 1.9495,
      "step": 15324
    },
    {
      "epoch": 0.4036081116671056,
      "grad_norm": 1.395447015762329,
      "learning_rate": 2.9831445878324994e-05,
      "loss": 1.6595,
      "step": 15325
    },
    {
      "epoch": 0.40363444824861733,
      "grad_norm": 1.4495803117752075,
      "learning_rate": 2.983012904924941e-05,
      "loss": 1.687,
      "step": 15326
    },
    {
      "epoch": 0.40366078483012907,
      "grad_norm": 3.6437270641326904,
      "learning_rate": 2.9828812220173825e-05,
      "loss": 1.4631,
      "step": 15327
    },
    {
      "epoch": 0.40368712141164076,
      "grad_norm": 1.7573702335357666,
      "learning_rate": 2.9827495391098237e-05,
      "loss": 1.9854,
      "step": 15328
    },
    {
      "epoch": 0.4037134579931525,
      "grad_norm": NaN,
      "learning_rate": 2.9827495391098237e-05,
      "loss": 2.0548,
      "step": 15329
    },
    {
      "epoch": 0.4037397945746642,
      "grad_norm": 2.279595136642456,
      "learning_rate": 2.982617856202265e-05,
      "loss": 1.5082,
      "step": 15330
    },
    {
      "epoch": 0.40376613115617593,
      "grad_norm": 1.6415857076644897,
      "learning_rate": 2.9824861732947062e-05,
      "loss": 1.8043,
      "step": 15331
    },
    {
      "epoch": 0.4037924677376876,
      "grad_norm": 3.3703339099884033,
      "learning_rate": 2.982354490387148e-05,
      "loss": 1.8585,
      "step": 15332
    },
    {
      "epoch": 0.40381880431919936,
      "grad_norm": 2.0637922286987305,
      "learning_rate": 2.9822228074795893e-05,
      "loss": 1.8432,
      "step": 15333
    },
    {
      "epoch": 0.4038451409007111,
      "grad_norm": 3.3325417041778564,
      "learning_rate": 2.9820911245720305e-05,
      "loss": 1.4587,
      "step": 15334
    },
    {
      "epoch": 0.4038714774822228,
      "grad_norm": 1.8212943077087402,
      "learning_rate": 2.981959441664472e-05,
      "loss": 2.1724,
      "step": 15335
    },
    {
      "epoch": 0.40389781406373454,
      "grad_norm": 2.6107821464538574,
      "learning_rate": 2.9818277587569133e-05,
      "loss": 2.114,
      "step": 15336
    },
    {
      "epoch": 0.4039241506452462,
      "grad_norm": 2.176142454147339,
      "learning_rate": 2.9816960758493552e-05,
      "loss": 1.7516,
      "step": 15337
    },
    {
      "epoch": 0.40395048722675797,
      "grad_norm": 3.1308257579803467,
      "learning_rate": 2.9815643929417964e-05,
      "loss": 1.9061,
      "step": 15338
    },
    {
      "epoch": 0.4039768238082697,
      "grad_norm": 1.6352956295013428,
      "learning_rate": 2.9814327100342376e-05,
      "loss": 1.4167,
      "step": 15339
    },
    {
      "epoch": 0.4040031603897814,
      "grad_norm": 2.701352834701538,
      "learning_rate": 2.981301027126679e-05,
      "loss": 1.2969,
      "step": 15340
    },
    {
      "epoch": 0.40402949697129314,
      "grad_norm": 1.825621485710144,
      "learning_rate": 2.9811693442191208e-05,
      "loss": 2.1002,
      "step": 15341
    },
    {
      "epoch": 0.40405583355280483,
      "grad_norm": 1.8138121366500854,
      "learning_rate": 2.981037661311562e-05,
      "loss": 0.303,
      "step": 15342
    },
    {
      "epoch": 0.4040821701343166,
      "grad_norm": 5.9109272956848145,
      "learning_rate": 2.9809059784040032e-05,
      "loss": 1.5597,
      "step": 15343
    },
    {
      "epoch": 0.40410850671582826,
      "grad_norm": 3.0749809741973877,
      "learning_rate": 2.9807742954964448e-05,
      "loss": 1.9889,
      "step": 15344
    },
    {
      "epoch": 0.40413484329734,
      "grad_norm": 3.8184762001037598,
      "learning_rate": 2.980642612588886e-05,
      "loss": 1.1444,
      "step": 15345
    },
    {
      "epoch": 0.40416117987885175,
      "grad_norm": 1.7192103862762451,
      "learning_rate": 2.980510929681328e-05,
      "loss": 2.154,
      "step": 15346
    },
    {
      "epoch": 0.40418751646036344,
      "grad_norm": 3.9199676513671875,
      "learning_rate": 2.980379246773769e-05,
      "loss": 1.6701,
      "step": 15347
    },
    {
      "epoch": 0.4042138530418752,
      "grad_norm": 3.170886516571045,
      "learning_rate": 2.9802475638662103e-05,
      "loss": 2.2398,
      "step": 15348
    },
    {
      "epoch": 0.40424018962338687,
      "grad_norm": 2.3458900451660156,
      "learning_rate": 2.9801158809586515e-05,
      "loss": 1.8996,
      "step": 15349
    },
    {
      "epoch": 0.4042665262048986,
      "grad_norm": 3.924548387527466,
      "learning_rate": 2.9799841980510928e-05,
      "loss": 0.5476,
      "step": 15350
    },
    {
      "epoch": 0.4042928627864103,
      "grad_norm": 1.8035708665847778,
      "learning_rate": 2.9798525151435347e-05,
      "loss": 2.1935,
      "step": 15351
    },
    {
      "epoch": 0.40431919936792204,
      "grad_norm": 1.8139768838882446,
      "learning_rate": 2.979720832235976e-05,
      "loss": 1.7111,
      "step": 15352
    },
    {
      "epoch": 0.4043455359494338,
      "grad_norm": 1.618435025215149,
      "learning_rate": 2.979589149328417e-05,
      "loss": 1.5709,
      "step": 15353
    },
    {
      "epoch": 0.4043718725309455,
      "grad_norm": 3.874896764755249,
      "learning_rate": 2.9794574664208587e-05,
      "loss": 2.0537,
      "step": 15354
    },
    {
      "epoch": 0.4043982091124572,
      "grad_norm": 1.975533366203308,
      "learning_rate": 2.9793257835133002e-05,
      "loss": 2.1022,
      "step": 15355
    },
    {
      "epoch": 0.4044245456939689,
      "grad_norm": 3.2438900470733643,
      "learning_rate": 2.9791941006057418e-05,
      "loss": 0.9999,
      "step": 15356
    },
    {
      "epoch": 0.40445088227548065,
      "grad_norm": 2.1629087924957275,
      "learning_rate": 2.979062417698183e-05,
      "loss": 1.9705,
      "step": 15357
    },
    {
      "epoch": 0.40447721885699234,
      "grad_norm": 2.615725040435791,
      "learning_rate": 2.9789307347906242e-05,
      "loss": 1.5595,
      "step": 15358
    },
    {
      "epoch": 0.4045035554385041,
      "grad_norm": 1.3952604532241821,
      "learning_rate": 2.9787990518830654e-05,
      "loss": 1.7454,
      "step": 15359
    },
    {
      "epoch": 0.4045298920200158,
      "grad_norm": 1.5876686573028564,
      "learning_rate": 2.9786673689755073e-05,
      "loss": 1.5213,
      "step": 15360
    },
    {
      "epoch": 0.4045562286015275,
      "grad_norm": 2.065218448638916,
      "learning_rate": 2.9785356860679486e-05,
      "loss": 1.4468,
      "step": 15361
    },
    {
      "epoch": 0.40458256518303926,
      "grad_norm": 2.1356821060180664,
      "learning_rate": 2.9784040031603898e-05,
      "loss": 2.152,
      "step": 15362
    },
    {
      "epoch": 0.40460890176455094,
      "grad_norm": 3.850576162338257,
      "learning_rate": 2.9782723202528313e-05,
      "loss": 1.0064,
      "step": 15363
    },
    {
      "epoch": 0.4046352383460627,
      "grad_norm": 2.0404551029205322,
      "learning_rate": 2.9781406373452726e-05,
      "loss": 2.4963,
      "step": 15364
    },
    {
      "epoch": 0.4046615749275744,
      "grad_norm": 2.5017809867858887,
      "learning_rate": 2.9780089544377145e-05,
      "loss": 1.1651,
      "step": 15365
    },
    {
      "epoch": 0.4046879115090861,
      "grad_norm": 2.846797466278076,
      "learning_rate": 2.9778772715301557e-05,
      "loss": 1.5911,
      "step": 15366
    },
    {
      "epoch": 0.40471424809059786,
      "grad_norm": 3.607956647872925,
      "learning_rate": 2.977745588622597e-05,
      "loss": 1.0438,
      "step": 15367
    },
    {
      "epoch": 0.40474058467210955,
      "grad_norm": 3.209007501602173,
      "learning_rate": 2.977613905715038e-05,
      "loss": 1.7777,
      "step": 15368
    },
    {
      "epoch": 0.4047669212536213,
      "grad_norm": 3.2647480964660645,
      "learning_rate": 2.9774822228074794e-05,
      "loss": 0.52,
      "step": 15369
    },
    {
      "epoch": 0.404793257835133,
      "grad_norm": 1.669777274131775,
      "learning_rate": 2.9773505398999213e-05,
      "loss": 2.3838,
      "step": 15370
    },
    {
      "epoch": 0.4048195944166447,
      "grad_norm": 2.823716878890991,
      "learning_rate": 2.9772188569923625e-05,
      "loss": 1.998,
      "step": 15371
    },
    {
      "epoch": 0.40484593099815647,
      "grad_norm": 2.0487589836120605,
      "learning_rate": 2.977087174084804e-05,
      "loss": 1.6007,
      "step": 15372
    },
    {
      "epoch": 0.40487226757966815,
      "grad_norm": 6.423557281494141,
      "learning_rate": 2.9769554911772453e-05,
      "loss": 1.6826,
      "step": 15373
    },
    {
      "epoch": 0.4048986041611799,
      "grad_norm": 2.6650333404541016,
      "learning_rate": 2.976823808269687e-05,
      "loss": 1.6367,
      "step": 15374
    },
    {
      "epoch": 0.4049249407426916,
      "grad_norm": 2.885716438293457,
      "learning_rate": 2.9766921253621284e-05,
      "loss": 1.8185,
      "step": 15375
    },
    {
      "epoch": 0.40495127732420333,
      "grad_norm": 3.1580519676208496,
      "learning_rate": 2.9765604424545696e-05,
      "loss": 1.6559,
      "step": 15376
    },
    {
      "epoch": 0.404977613905715,
      "grad_norm": 1.8138922452926636,
      "learning_rate": 2.9764287595470108e-05,
      "loss": 1.6343,
      "step": 15377
    },
    {
      "epoch": 0.40500395048722676,
      "grad_norm": 1.733119010925293,
      "learning_rate": 2.976297076639452e-05,
      "loss": 1.851,
      "step": 15378
    },
    {
      "epoch": 0.4050302870687385,
      "grad_norm": 5.5334367752075195,
      "learning_rate": 2.976165393731894e-05,
      "loss": 2.0338,
      "step": 15379
    },
    {
      "epoch": 0.4050566236502502,
      "grad_norm": 2.062154531478882,
      "learning_rate": 2.976033710824335e-05,
      "loss": 1.5095,
      "step": 15380
    },
    {
      "epoch": 0.40508296023176193,
      "grad_norm": 2.687317132949829,
      "learning_rate": 2.9759020279167764e-05,
      "loss": 0.8981,
      "step": 15381
    },
    {
      "epoch": 0.4051092968132736,
      "grad_norm": 2.1819286346435547,
      "learning_rate": 2.975770345009218e-05,
      "loss": 1.4309,
      "step": 15382
    },
    {
      "epoch": 0.40513563339478537,
      "grad_norm": 1.8535792827606201,
      "learning_rate": 2.975638662101659e-05,
      "loss": 1.2911,
      "step": 15383
    },
    {
      "epoch": 0.40516196997629705,
      "grad_norm": 3.983138084411621,
      "learning_rate": 2.975506979194101e-05,
      "loss": 1.5396,
      "step": 15384
    },
    {
      "epoch": 0.4051883065578088,
      "grad_norm": 3.4726104736328125,
      "learning_rate": 2.9753752962865423e-05,
      "loss": 1.1441,
      "step": 15385
    },
    {
      "epoch": 0.40521464313932054,
      "grad_norm": 1.7042008638381958,
      "learning_rate": 2.9752436133789835e-05,
      "loss": 1.8604,
      "step": 15386
    },
    {
      "epoch": 0.40524097972083223,
      "grad_norm": 3.164184331893921,
      "learning_rate": 2.9751119304714247e-05,
      "loss": 1.6527,
      "step": 15387
    },
    {
      "epoch": 0.40526731630234397,
      "grad_norm": 1.9780757427215576,
      "learning_rate": 2.9749802475638666e-05,
      "loss": 1.3688,
      "step": 15388
    },
    {
      "epoch": 0.40529365288385566,
      "grad_norm": 2.3368895053863525,
      "learning_rate": 2.974848564656308e-05,
      "loss": 0.4079,
      "step": 15389
    },
    {
      "epoch": 0.4053199894653674,
      "grad_norm": 1.8770625591278076,
      "learning_rate": 2.974716881748749e-05,
      "loss": 0.9074,
      "step": 15390
    },
    {
      "epoch": 0.4053463260468791,
      "grad_norm": 2.0360267162323,
      "learning_rate": 2.9745851988411906e-05,
      "loss": 1.4624,
      "step": 15391
    },
    {
      "epoch": 0.40537266262839083,
      "grad_norm": 4.387003421783447,
      "learning_rate": 2.974453515933632e-05,
      "loss": 1.8201,
      "step": 15392
    },
    {
      "epoch": 0.4053989992099026,
      "grad_norm": 3.6480557918548584,
      "learning_rate": 2.9743218330260737e-05,
      "loss": 2.5745,
      "step": 15393
    },
    {
      "epoch": 0.40542533579141427,
      "grad_norm": 3.384028673171997,
      "learning_rate": 2.974190150118515e-05,
      "loss": 1.6655,
      "step": 15394
    },
    {
      "epoch": 0.405451672372926,
      "grad_norm": 2.7470521926879883,
      "learning_rate": 2.9740584672109562e-05,
      "loss": 1.9622,
      "step": 15395
    },
    {
      "epoch": 0.4054780089544377,
      "grad_norm": 2.243062734603882,
      "learning_rate": 2.9739267843033974e-05,
      "loss": 1.7063,
      "step": 15396
    },
    {
      "epoch": 0.40550434553594944,
      "grad_norm": 4.0135602951049805,
      "learning_rate": 2.9737951013958386e-05,
      "loss": 1.247,
      "step": 15397
    },
    {
      "epoch": 0.4055306821174611,
      "grad_norm": 1.9372515678405762,
      "learning_rate": 2.9736634184882805e-05,
      "loss": 1.7156,
      "step": 15398
    },
    {
      "epoch": 0.40555701869897287,
      "grad_norm": 1.9279767274856567,
      "learning_rate": 2.9735317355807217e-05,
      "loss": 1.7418,
      "step": 15399
    },
    {
      "epoch": 0.4055833552804846,
      "grad_norm": 2.5324831008911133,
      "learning_rate": 2.973400052673163e-05,
      "loss": 1.3552,
      "step": 15400
    },
    {
      "epoch": 0.4056096918619963,
      "grad_norm": 3.1103415489196777,
      "learning_rate": 2.9732683697656045e-05,
      "loss": 1.5672,
      "step": 15401
    },
    {
      "epoch": 0.40563602844350805,
      "grad_norm": 1.7157319784164429,
      "learning_rate": 2.9731366868580457e-05,
      "loss": 2.0582,
      "step": 15402
    },
    {
      "epoch": 0.40566236502501973,
      "grad_norm": 1.8707804679870605,
      "learning_rate": 2.9730050039504876e-05,
      "loss": 2.1754,
      "step": 15403
    },
    {
      "epoch": 0.4056887016065315,
      "grad_norm": 3.8068370819091797,
      "learning_rate": 2.972873321042929e-05,
      "loss": 1.8343,
      "step": 15404
    },
    {
      "epoch": 0.40571503818804316,
      "grad_norm": 5.422422409057617,
      "learning_rate": 2.97274163813537e-05,
      "loss": 1.6686,
      "step": 15405
    },
    {
      "epoch": 0.4057413747695549,
      "grad_norm": 5.505545616149902,
      "learning_rate": 2.9726099552278113e-05,
      "loss": 2.467,
      "step": 15406
    },
    {
      "epoch": 0.40576771135106665,
      "grad_norm": 1.986251711845398,
      "learning_rate": 2.9724782723202532e-05,
      "loss": 2.0649,
      "step": 15407
    },
    {
      "epoch": 0.40579404793257834,
      "grad_norm": 2.211836338043213,
      "learning_rate": 2.9723465894126944e-05,
      "loss": 0.5509,
      "step": 15408
    },
    {
      "epoch": 0.4058203845140901,
      "grad_norm": 1.5220776796340942,
      "learning_rate": 2.9722149065051356e-05,
      "loss": 1.2357,
      "step": 15409
    },
    {
      "epoch": 0.40584672109560177,
      "grad_norm": 3.980742931365967,
      "learning_rate": 2.9720832235975772e-05,
      "loss": 1.2308,
      "step": 15410
    },
    {
      "epoch": 0.4058730576771135,
      "grad_norm": 2.5117199420928955,
      "learning_rate": 2.9719515406900184e-05,
      "loss": 2.5167,
      "step": 15411
    },
    {
      "epoch": 0.40589939425862526,
      "grad_norm": 3.5928966999053955,
      "learning_rate": 2.9718198577824603e-05,
      "loss": 1.5631,
      "step": 15412
    },
    {
      "epoch": 0.40592573084013694,
      "grad_norm": 3.316068649291992,
      "learning_rate": 2.9716881748749015e-05,
      "loss": 2.32,
      "step": 15413
    },
    {
      "epoch": 0.4059520674216487,
      "grad_norm": 3.653067111968994,
      "learning_rate": 2.9715564919673428e-05,
      "loss": 1.0957,
      "step": 15414
    },
    {
      "epoch": 0.4059784040031604,
      "grad_norm": 1.9494534730911255,
      "learning_rate": 2.971424809059784e-05,
      "loss": 1.9805,
      "step": 15415
    },
    {
      "epoch": 0.4060047405846721,
      "grad_norm": 2.3916592597961426,
      "learning_rate": 2.9712931261522252e-05,
      "loss": 0.4726,
      "step": 15416
    },
    {
      "epoch": 0.4060310771661838,
      "grad_norm": 1.8614389896392822,
      "learning_rate": 2.971161443244667e-05,
      "loss": 2.1196,
      "step": 15417
    },
    {
      "epoch": 0.40605741374769555,
      "grad_norm": 2.3933017253875732,
      "learning_rate": 2.9710297603371083e-05,
      "loss": 1.8667,
      "step": 15418
    },
    {
      "epoch": 0.4060837503292073,
      "grad_norm": 2.385232448577881,
      "learning_rate": 2.97089807742955e-05,
      "loss": 1.5811,
      "step": 15419
    },
    {
      "epoch": 0.406110086910719,
      "grad_norm": 3.0019991397857666,
      "learning_rate": 2.970766394521991e-05,
      "loss": 1.588,
      "step": 15420
    },
    {
      "epoch": 0.4061364234922307,
      "grad_norm": 2.4925224781036377,
      "learning_rate": 2.970634711614433e-05,
      "loss": 0.9186,
      "step": 15421
    },
    {
      "epoch": 0.4061627600737424,
      "grad_norm": 3.316859483718872,
      "learning_rate": 2.9705030287068742e-05,
      "loss": 1.296,
      "step": 15422
    },
    {
      "epoch": 0.40618909665525416,
      "grad_norm": 2.2114908695220947,
      "learning_rate": 2.9703713457993154e-05,
      "loss": 2.0354,
      "step": 15423
    },
    {
      "epoch": 0.40621543323676584,
      "grad_norm": 4.095826148986816,
      "learning_rate": 2.9702396628917567e-05,
      "loss": 1.5299,
      "step": 15424
    },
    {
      "epoch": 0.4062417698182776,
      "grad_norm": 1.9702955484390259,
      "learning_rate": 2.970107979984198e-05,
      "loss": 1.6186,
      "step": 15425
    },
    {
      "epoch": 0.40626810639978933,
      "grad_norm": 2.061579704284668,
      "learning_rate": 2.9699762970766398e-05,
      "loss": 1.4586,
      "step": 15426
    },
    {
      "epoch": 0.406294442981301,
      "grad_norm": 1.6376806497573853,
      "learning_rate": 2.969844614169081e-05,
      "loss": 1.5134,
      "step": 15427
    },
    {
      "epoch": 0.40632077956281276,
      "grad_norm": 2.303356885910034,
      "learning_rate": 2.9697129312615222e-05,
      "loss": 0.8613,
      "step": 15428
    },
    {
      "epoch": 0.40634711614432445,
      "grad_norm": 1.7938517332077026,
      "learning_rate": 2.9695812483539638e-05,
      "loss": 1.7145,
      "step": 15429
    },
    {
      "epoch": 0.4063734527258362,
      "grad_norm": 1.947532057762146,
      "learning_rate": 2.969449565446405e-05,
      "loss": 1.7949,
      "step": 15430
    },
    {
      "epoch": 0.4063997893073479,
      "grad_norm": 1.6662968397140503,
      "learning_rate": 2.969317882538847e-05,
      "loss": 1.8401,
      "step": 15431
    },
    {
      "epoch": 0.4064261258888596,
      "grad_norm": 4.729133129119873,
      "learning_rate": 2.969186199631288e-05,
      "loss": 0.8145,
      "step": 15432
    },
    {
      "epoch": 0.40645246247037137,
      "grad_norm": 5.384486198425293,
      "learning_rate": 2.9690545167237294e-05,
      "loss": 2.2228,
      "step": 15433
    },
    {
      "epoch": 0.40647879905188306,
      "grad_norm": 1.9077547788619995,
      "learning_rate": 2.9689228338161706e-05,
      "loss": 1.8364,
      "step": 15434
    },
    {
      "epoch": 0.4065051356333948,
      "grad_norm": 2.904686689376831,
      "learning_rate": 2.9687911509086118e-05,
      "loss": 1.8479,
      "step": 15435
    },
    {
      "epoch": 0.4065314722149065,
      "grad_norm": 2.0099730491638184,
      "learning_rate": 2.9686594680010537e-05,
      "loss": 1.8528,
      "step": 15436
    },
    {
      "epoch": 0.40655780879641823,
      "grad_norm": 5.712948322296143,
      "learning_rate": 2.968527785093495e-05,
      "loss": 0.7206,
      "step": 15437
    },
    {
      "epoch": 0.4065841453779299,
      "grad_norm": 3.6986382007598877,
      "learning_rate": 2.9683961021859365e-05,
      "loss": 2.4564,
      "step": 15438
    },
    {
      "epoch": 0.40661048195944166,
      "grad_norm": 3.9545769691467285,
      "learning_rate": 2.9682644192783777e-05,
      "loss": 1.4015,
      "step": 15439
    },
    {
      "epoch": 0.4066368185409534,
      "grad_norm": 3.0367674827575684,
      "learning_rate": 2.9681327363708196e-05,
      "loss": 1.123,
      "step": 15440
    },
    {
      "epoch": 0.4066631551224651,
      "grad_norm": 2.2434582710266113,
      "learning_rate": 2.9680010534632608e-05,
      "loss": 1.2418,
      "step": 15441
    },
    {
      "epoch": 0.40668949170397684,
      "grad_norm": 2.2279560565948486,
      "learning_rate": 2.967869370555702e-05,
      "loss": 1.505,
      "step": 15442
    },
    {
      "epoch": 0.4067158282854885,
      "grad_norm": 1.5036253929138184,
      "learning_rate": 2.9677376876481433e-05,
      "loss": 1.5769,
      "step": 15443
    },
    {
      "epoch": 0.40674216486700027,
      "grad_norm": 2.491455316543579,
      "learning_rate": 2.9676060047405845e-05,
      "loss": 1.724,
      "step": 15444
    },
    {
      "epoch": 0.406768501448512,
      "grad_norm": 3.2721242904663086,
      "learning_rate": 2.9674743218330264e-05,
      "loss": 1.6572,
      "step": 15445
    },
    {
      "epoch": 0.4067948380300237,
      "grad_norm": 2.012260913848877,
      "learning_rate": 2.9673426389254676e-05,
      "loss": 2.1404,
      "step": 15446
    },
    {
      "epoch": 0.40682117461153544,
      "grad_norm": 1.5121233463287354,
      "learning_rate": 2.967210956017909e-05,
      "loss": 1.7112,
      "step": 15447
    },
    {
      "epoch": 0.40684751119304713,
      "grad_norm": 1.834301233291626,
      "learning_rate": 2.9670792731103504e-05,
      "loss": 1.4894,
      "step": 15448
    },
    {
      "epoch": 0.40687384777455887,
      "grad_norm": 2.0125436782836914,
      "learning_rate": 2.9669475902027916e-05,
      "loss": 1.7273,
      "step": 15449
    },
    {
      "epoch": 0.40690018435607056,
      "grad_norm": 6.842875003814697,
      "learning_rate": 2.9668159072952335e-05,
      "loss": 1.8526,
      "step": 15450
    },
    {
      "epoch": 0.4069265209375823,
      "grad_norm": 2.128511428833008,
      "learning_rate": 2.9666842243876747e-05,
      "loss": 1.5089,
      "step": 15451
    },
    {
      "epoch": 0.40695285751909405,
      "grad_norm": 1.5033659934997559,
      "learning_rate": 2.966552541480116e-05,
      "loss": 2.066,
      "step": 15452
    },
    {
      "epoch": 0.40697919410060573,
      "grad_norm": 3.0011868476867676,
      "learning_rate": 2.966420858572557e-05,
      "loss": 0.6518,
      "step": 15453
    },
    {
      "epoch": 0.4070055306821175,
      "grad_norm": 2.5068671703338623,
      "learning_rate": 2.966289175664999e-05,
      "loss": 1.5624,
      "step": 15454
    },
    {
      "epoch": 0.40703186726362917,
      "grad_norm": 3.2575442790985107,
      "learning_rate": 2.9661574927574403e-05,
      "loss": 0.2517,
      "step": 15455
    },
    {
      "epoch": 0.4070582038451409,
      "grad_norm": 1.6077076196670532,
      "learning_rate": 2.9660258098498815e-05,
      "loss": 1.7122,
      "step": 15456
    },
    {
      "epoch": 0.4070845404266526,
      "grad_norm": 1.9932819604873657,
      "learning_rate": 2.965894126942323e-05,
      "loss": 1.2649,
      "step": 15457
    },
    {
      "epoch": 0.40711087700816434,
      "grad_norm": 3.8360519409179688,
      "learning_rate": 2.9657624440347643e-05,
      "loss": 2.2119,
      "step": 15458
    },
    {
      "epoch": 0.4071372135896761,
      "grad_norm": 1.9107203483581543,
      "learning_rate": 2.9656307611272062e-05,
      "loss": 1.8703,
      "step": 15459
    },
    {
      "epoch": 0.40716355017118777,
      "grad_norm": 1.6759370565414429,
      "learning_rate": 2.9654990782196474e-05,
      "loss": 1.6466,
      "step": 15460
    },
    {
      "epoch": 0.4071898867526995,
      "grad_norm": 3.092898368835449,
      "learning_rate": 2.9653673953120886e-05,
      "loss": 1.3679,
      "step": 15461
    },
    {
      "epoch": 0.4072162233342112,
      "grad_norm": 4.308140277862549,
      "learning_rate": 2.96523571240453e-05,
      "loss": 1.6723,
      "step": 15462
    },
    {
      "epoch": 0.40724255991572295,
      "grad_norm": 1.9461251497268677,
      "learning_rate": 2.965104029496971e-05,
      "loss": 1.3311,
      "step": 15463
    },
    {
      "epoch": 0.40726889649723463,
      "grad_norm": 3.026602029800415,
      "learning_rate": 2.964972346589413e-05,
      "loss": 1.4665,
      "step": 15464
    },
    {
      "epoch": 0.4072952330787464,
      "grad_norm": 1.8129791021347046,
      "learning_rate": 2.9648406636818542e-05,
      "loss": 1.6168,
      "step": 15465
    },
    {
      "epoch": 0.4073215696602581,
      "grad_norm": 1.740459680557251,
      "learning_rate": 2.9647089807742957e-05,
      "loss": 2.5732,
      "step": 15466
    },
    {
      "epoch": 0.4073479062417698,
      "grad_norm": 3.010624647140503,
      "learning_rate": 2.964577297866737e-05,
      "loss": 2.3867,
      "step": 15467
    },
    {
      "epoch": 0.40737424282328155,
      "grad_norm": 3.168281316757202,
      "learning_rate": 2.9644456149591782e-05,
      "loss": 1.5576,
      "step": 15468
    },
    {
      "epoch": 0.40740057940479324,
      "grad_norm": 3.2274458408355713,
      "learning_rate": 2.96431393205162e-05,
      "loss": 1.5266,
      "step": 15469
    },
    {
      "epoch": 0.407426915986305,
      "grad_norm": 3.892594337463379,
      "learning_rate": 2.9641822491440613e-05,
      "loss": 2.3756,
      "step": 15470
    },
    {
      "epoch": 0.40745325256781667,
      "grad_norm": 2.5363903045654297,
      "learning_rate": 2.9640505662365025e-05,
      "loss": 1.7264,
      "step": 15471
    },
    {
      "epoch": 0.4074795891493284,
      "grad_norm": 3.7932369709014893,
      "learning_rate": 2.9639188833289437e-05,
      "loss": 0.8553,
      "step": 15472
    },
    {
      "epoch": 0.40750592573084016,
      "grad_norm": 2.0251471996307373,
      "learning_rate": 2.9637872004213856e-05,
      "loss": 2.0359,
      "step": 15473
    },
    {
      "epoch": 0.40753226231235185,
      "grad_norm": 2.834667444229126,
      "learning_rate": 2.963655517513827e-05,
      "loss": 1.2173,
      "step": 15474
    },
    {
      "epoch": 0.4075585988938636,
      "grad_norm": 1.8057821989059448,
      "learning_rate": 2.9635238346062684e-05,
      "loss": 1.39,
      "step": 15475
    },
    {
      "epoch": 0.4075849354753753,
      "grad_norm": 3.6677236557006836,
      "learning_rate": 2.9633921516987096e-05,
      "loss": 1.0169,
      "step": 15476
    },
    {
      "epoch": 0.407611272056887,
      "grad_norm": 2.107739210128784,
      "learning_rate": 2.963260468791151e-05,
      "loss": 2.0358,
      "step": 15477
    },
    {
      "epoch": 0.40763760863839876,
      "grad_norm": 1.481797695159912,
      "learning_rate": 2.9631287858835928e-05,
      "loss": 2.0033,
      "step": 15478
    },
    {
      "epoch": 0.40766394521991045,
      "grad_norm": 4.4800801277160645,
      "learning_rate": 2.962997102976034e-05,
      "loss": 1.8054,
      "step": 15479
    },
    {
      "epoch": 0.4076902818014222,
      "grad_norm": 2.457227945327759,
      "learning_rate": 2.9628654200684752e-05,
      "loss": 1.0475,
      "step": 15480
    },
    {
      "epoch": 0.4077166183829339,
      "grad_norm": 2.21356463432312,
      "learning_rate": 2.9627337371609164e-05,
      "loss": 0.8472,
      "step": 15481
    },
    {
      "epoch": 0.4077429549644456,
      "grad_norm": 1.7571008205413818,
      "learning_rate": 2.9626020542533577e-05,
      "loss": 1.7823,
      "step": 15482
    },
    {
      "epoch": 0.4077692915459573,
      "grad_norm": 1.961984634399414,
      "learning_rate": 2.9624703713457995e-05,
      "loss": 1.7197,
      "step": 15483
    },
    {
      "epoch": 0.40779562812746906,
      "grad_norm": 2.202336311340332,
      "learning_rate": 2.9623386884382408e-05,
      "loss": 2.2676,
      "step": 15484
    },
    {
      "epoch": 0.4078219647089808,
      "grad_norm": 1.9808951616287231,
      "learning_rate": 2.9622070055306823e-05,
      "loss": 1.9398,
      "step": 15485
    },
    {
      "epoch": 0.4078483012904925,
      "grad_norm": 2.01914119720459,
      "learning_rate": 2.9620753226231236e-05,
      "loss": 1.3534,
      "step": 15486
    },
    {
      "epoch": 0.40787463787200423,
      "grad_norm": 2.289134979248047,
      "learning_rate": 2.9619436397155654e-05,
      "loss": 1.6258,
      "step": 15487
    },
    {
      "epoch": 0.4079009744535159,
      "grad_norm": 3.446135997772217,
      "learning_rate": 2.9618119568080067e-05,
      "loss": 1.4653,
      "step": 15488
    },
    {
      "epoch": 0.40792731103502766,
      "grad_norm": 2.072317123413086,
      "learning_rate": 2.961680273900448e-05,
      "loss": 0.4528,
      "step": 15489
    },
    {
      "epoch": 0.40795364761653935,
      "grad_norm": 1.6260031461715698,
      "learning_rate": 2.961548590992889e-05,
      "loss": 1.7824,
      "step": 15490
    },
    {
      "epoch": 0.4079799841980511,
      "grad_norm": 2.3289427757263184,
      "learning_rate": 2.9614169080853303e-05,
      "loss": 0.9303,
      "step": 15491
    },
    {
      "epoch": 0.40800632077956284,
      "grad_norm": 2.5439000129699707,
      "learning_rate": 2.9612852251777722e-05,
      "loss": 1.4794,
      "step": 15492
    },
    {
      "epoch": 0.4080326573610745,
      "grad_norm": 2.2047746181488037,
      "learning_rate": 2.9611535422702135e-05,
      "loss": 2.3047,
      "step": 15493
    },
    {
      "epoch": 0.40805899394258627,
      "grad_norm": 2.7033281326293945,
      "learning_rate": 2.961021859362655e-05,
      "loss": 1.8353,
      "step": 15494
    },
    {
      "epoch": 0.40808533052409796,
      "grad_norm": 1.8041285276412964,
      "learning_rate": 2.9608901764550962e-05,
      "loss": 1.716,
      "step": 15495
    },
    {
      "epoch": 0.4081116671056097,
      "grad_norm": 2.187530040740967,
      "learning_rate": 2.9607584935475375e-05,
      "loss": 0.8654,
      "step": 15496
    },
    {
      "epoch": 0.4081380036871214,
      "grad_norm": 2.350358009338379,
      "learning_rate": 2.9606268106399794e-05,
      "loss": 1.6203,
      "step": 15497
    },
    {
      "epoch": 0.40816434026863313,
      "grad_norm": 3.5552783012390137,
      "learning_rate": 2.9604951277324206e-05,
      "loss": 0.6763,
      "step": 15498
    },
    {
      "epoch": 0.4081906768501449,
      "grad_norm": 1.7227190732955933,
      "learning_rate": 2.9603634448248618e-05,
      "loss": 1.3624,
      "step": 15499
    },
    {
      "epoch": 0.40821701343165656,
      "grad_norm": 2.537081241607666,
      "learning_rate": 2.960231761917303e-05,
      "loss": 1.8798,
      "step": 15500
    },
    {
      "epoch": 0.4082433500131683,
      "grad_norm": 3.4115798473358154,
      "learning_rate": 2.960100079009745e-05,
      "loss": 1.4744,
      "step": 15501
    },
    {
      "epoch": 0.40826968659468,
      "grad_norm": 1.8979332447052002,
      "learning_rate": 2.959968396102186e-05,
      "loss": 2.2764,
      "step": 15502
    },
    {
      "epoch": 0.40829602317619174,
      "grad_norm": 4.006608963012695,
      "learning_rate": 2.9598367131946274e-05,
      "loss": 0.831,
      "step": 15503
    },
    {
      "epoch": 0.4083223597577034,
      "grad_norm": 3.032528877258301,
      "learning_rate": 2.959705030287069e-05,
      "loss": 0.4143,
      "step": 15504
    },
    {
      "epoch": 0.40834869633921517,
      "grad_norm": 3.0231807231903076,
      "learning_rate": 2.95957334737951e-05,
      "loss": 2.2178,
      "step": 15505
    },
    {
      "epoch": 0.4083750329207269,
      "grad_norm": 1.7320244312286377,
      "learning_rate": 2.959441664471952e-05,
      "loss": 2.0389,
      "step": 15506
    },
    {
      "epoch": 0.4084013695022386,
      "grad_norm": 1.7165251970291138,
      "learning_rate": 2.9593099815643933e-05,
      "loss": 1.8923,
      "step": 15507
    },
    {
      "epoch": 0.40842770608375034,
      "grad_norm": 2.2482924461364746,
      "learning_rate": 2.9591782986568345e-05,
      "loss": 1.9368,
      "step": 15508
    },
    {
      "epoch": 0.40845404266526203,
      "grad_norm": 3.1780142784118652,
      "learning_rate": 2.9590466157492757e-05,
      "loss": 1.6005,
      "step": 15509
    },
    {
      "epoch": 0.4084803792467738,
      "grad_norm": 1.8063474893569946,
      "learning_rate": 2.958914932841717e-05,
      "loss": 1.7215,
      "step": 15510
    },
    {
      "epoch": 0.4085067158282855,
      "grad_norm": 1.6821879148483276,
      "learning_rate": 2.9587832499341588e-05,
      "loss": 2.305,
      "step": 15511
    },
    {
      "epoch": 0.4085330524097972,
      "grad_norm": 1.7118793725967407,
      "learning_rate": 2.9586515670266e-05,
      "loss": 2.0801,
      "step": 15512
    },
    {
      "epoch": 0.40855938899130895,
      "grad_norm": 1.8230926990509033,
      "learning_rate": 2.9585198841190416e-05,
      "loss": 1.7009,
      "step": 15513
    },
    {
      "epoch": 0.40858572557282063,
      "grad_norm": 2.3839879035949707,
      "learning_rate": 2.9583882012114828e-05,
      "loss": 1.3916,
      "step": 15514
    },
    {
      "epoch": 0.4086120621543324,
      "grad_norm": 2.482487440109253,
      "learning_rate": 2.958256518303924e-05,
      "loss": 0.8833,
      "step": 15515
    },
    {
      "epoch": 0.40863839873584407,
      "grad_norm": 3.2751338481903076,
      "learning_rate": 2.958124835396366e-05,
      "loss": 0.968,
      "step": 15516
    },
    {
      "epoch": 0.4086647353173558,
      "grad_norm": 4.714774131774902,
      "learning_rate": 2.957993152488807e-05,
      "loss": 1.4814,
      "step": 15517
    },
    {
      "epoch": 0.40869107189886755,
      "grad_norm": 1.7100814580917358,
      "learning_rate": 2.9578614695812484e-05,
      "loss": 2.2582,
      "step": 15518
    },
    {
      "epoch": 0.40871740848037924,
      "grad_norm": 2.099839448928833,
      "learning_rate": 2.9577297866736896e-05,
      "loss": 1.3471,
      "step": 15519
    },
    {
      "epoch": 0.408743745061891,
      "grad_norm": 1.9254851341247559,
      "learning_rate": 2.9575981037661315e-05,
      "loss": 1.886,
      "step": 15520
    },
    {
      "epoch": 0.40877008164340267,
      "grad_norm": 2.3890559673309326,
      "learning_rate": 2.9574664208585727e-05,
      "loss": 1.6631,
      "step": 15521
    },
    {
      "epoch": 0.4087964182249144,
      "grad_norm": 1.8621312379837036,
      "learning_rate": 2.9573347379510143e-05,
      "loss": 2.6855,
      "step": 15522
    },
    {
      "epoch": 0.4088227548064261,
      "grad_norm": 2.55586314201355,
      "learning_rate": 2.9572030550434555e-05,
      "loss": 2.027,
      "step": 15523
    },
    {
      "epoch": 0.40884909138793785,
      "grad_norm": 1.8009172677993774,
      "learning_rate": 2.9570713721358967e-05,
      "loss": 1.7208,
      "step": 15524
    },
    {
      "epoch": 0.4088754279694496,
      "grad_norm": 1.7379064559936523,
      "learning_rate": 2.9569396892283386e-05,
      "loss": 1.5554,
      "step": 15525
    },
    {
      "epoch": 0.4089017645509613,
      "grad_norm": 3.0741071701049805,
      "learning_rate": 2.95680800632078e-05,
      "loss": 0.4173,
      "step": 15526
    },
    {
      "epoch": 0.408928101132473,
      "grad_norm": 3.096268892288208,
      "learning_rate": 2.956676323413221e-05,
      "loss": 1.8136,
      "step": 15527
    },
    {
      "epoch": 0.4089544377139847,
      "grad_norm": 2.805260419845581,
      "learning_rate": 2.9565446405056623e-05,
      "loss": 1.7647,
      "step": 15528
    },
    {
      "epoch": 0.40898077429549645,
      "grad_norm": 4.796494960784912,
      "learning_rate": 2.9564129575981035e-05,
      "loss": 1.2489,
      "step": 15529
    },
    {
      "epoch": 0.40900711087700814,
      "grad_norm": 1.892318606376648,
      "learning_rate": 2.9562812746905454e-05,
      "loss": 1.9592,
      "step": 15530
    },
    {
      "epoch": 0.4090334474585199,
      "grad_norm": 3.3813490867614746,
      "learning_rate": 2.9561495917829866e-05,
      "loss": 1.3326,
      "step": 15531
    },
    {
      "epoch": 0.4090597840400316,
      "grad_norm": 1.989195466041565,
      "learning_rate": 2.9560179088754282e-05,
      "loss": 2.2213,
      "step": 15532
    },
    {
      "epoch": 0.4090861206215433,
      "grad_norm": 2.7464845180511475,
      "learning_rate": 2.9558862259678694e-05,
      "loss": 1.5868,
      "step": 15533
    },
    {
      "epoch": 0.40911245720305506,
      "grad_norm": 2.1392910480499268,
      "learning_rate": 2.9557545430603113e-05,
      "loss": 1.3225,
      "step": 15534
    },
    {
      "epoch": 0.40913879378456675,
      "grad_norm": 3.0204877853393555,
      "learning_rate": 2.9556228601527525e-05,
      "loss": 1.3558,
      "step": 15535
    },
    {
      "epoch": 0.4091651303660785,
      "grad_norm": 2.2563998699188232,
      "learning_rate": 2.9554911772451937e-05,
      "loss": 0.7977,
      "step": 15536
    },
    {
      "epoch": 0.4091914669475902,
      "grad_norm": 2.4553513526916504,
      "learning_rate": 2.955359494337635e-05,
      "loss": 1.7936,
      "step": 15537
    },
    {
      "epoch": 0.4092178035291019,
      "grad_norm": 3.082031011581421,
      "learning_rate": 2.9552278114300762e-05,
      "loss": 1.9524,
      "step": 15538
    },
    {
      "epoch": 0.40924414011061366,
      "grad_norm": 3.5475902557373047,
      "learning_rate": 2.955096128522518e-05,
      "loss": 1.4063,
      "step": 15539
    },
    {
      "epoch": 0.40927047669212535,
      "grad_norm": 2.4492595195770264,
      "learning_rate": 2.9549644456149593e-05,
      "loss": 1.8252,
      "step": 15540
    },
    {
      "epoch": 0.4092968132736371,
      "grad_norm": 2.0702593326568604,
      "learning_rate": 2.954832762707401e-05,
      "loss": 1.8981,
      "step": 15541
    },
    {
      "epoch": 0.4093231498551488,
      "grad_norm": 1.7347869873046875,
      "learning_rate": 2.954701079799842e-05,
      "loss": 1.5274,
      "step": 15542
    },
    {
      "epoch": 0.4093494864366605,
      "grad_norm": 2.504793643951416,
      "learning_rate": 2.9545693968922833e-05,
      "loss": 1.814,
      "step": 15543
    },
    {
      "epoch": 0.4093758230181722,
      "grad_norm": 1.5416631698608398,
      "learning_rate": 2.9544377139847252e-05,
      "loss": 0.965,
      "step": 15544
    },
    {
      "epoch": 0.40940215959968396,
      "grad_norm": 2.92168927192688,
      "learning_rate": 2.9543060310771664e-05,
      "loss": 1.941,
      "step": 15545
    },
    {
      "epoch": 0.4094284961811957,
      "grad_norm": 1.651750087738037,
      "learning_rate": 2.9541743481696077e-05,
      "loss": 2.0861,
      "step": 15546
    },
    {
      "epoch": 0.4094548327627074,
      "grad_norm": 3.253331184387207,
      "learning_rate": 2.954042665262049e-05,
      "loss": 1.1731,
      "step": 15547
    },
    {
      "epoch": 0.40948116934421913,
      "grad_norm": 3.992542028427124,
      "learning_rate": 2.9539109823544904e-05,
      "loss": 1.5168,
      "step": 15548
    },
    {
      "epoch": 0.4095075059257308,
      "grad_norm": 3.426276683807373,
      "learning_rate": 2.953779299446932e-05,
      "loss": 1.5996,
      "step": 15549
    },
    {
      "epoch": 0.40953384250724256,
      "grad_norm": 1.898941993713379,
      "learning_rate": 2.9536476165393736e-05,
      "loss": 2.5171,
      "step": 15550
    },
    {
      "epoch": 0.4095601790887543,
      "grad_norm": 1.886821985244751,
      "learning_rate": 2.9535159336318148e-05,
      "loss": 1.3618,
      "step": 15551
    },
    {
      "epoch": 0.409586515670266,
      "grad_norm": 1.95891273021698,
      "learning_rate": 2.953384250724256e-05,
      "loss": 2.0994,
      "step": 15552
    },
    {
      "epoch": 0.40961285225177774,
      "grad_norm": 4.050502777099609,
      "learning_rate": 2.953252567816698e-05,
      "loss": 1.8797,
      "step": 15553
    },
    {
      "epoch": 0.4096391888332894,
      "grad_norm": 2.9661693572998047,
      "learning_rate": 2.953120884909139e-05,
      "loss": 1.9262,
      "step": 15554
    },
    {
      "epoch": 0.40966552541480117,
      "grad_norm": 1.9914149045944214,
      "learning_rate": 2.9529892020015803e-05,
      "loss": 2.1554,
      "step": 15555
    },
    {
      "epoch": 0.40969186199631286,
      "grad_norm": 2.4739737510681152,
      "learning_rate": 2.9528575190940216e-05,
      "loss": 1.2371,
      "step": 15556
    },
    {
      "epoch": 0.4097181985778246,
      "grad_norm": 1.9299970865249634,
      "learning_rate": 2.9527258361864628e-05,
      "loss": 0.926,
      "step": 15557
    },
    {
      "epoch": 0.40974453515933634,
      "grad_norm": 3.3257062435150146,
      "learning_rate": 2.9525941532789047e-05,
      "loss": 1.6199,
      "step": 15558
    },
    {
      "epoch": 0.40977087174084803,
      "grad_norm": 1.7143402099609375,
      "learning_rate": 2.952462470371346e-05,
      "loss": 2.1645,
      "step": 15559
    },
    {
      "epoch": 0.4097972083223598,
      "grad_norm": 1.9123141765594482,
      "learning_rate": 2.9523307874637875e-05,
      "loss": 1.7718,
      "step": 15560
    },
    {
      "epoch": 0.40982354490387146,
      "grad_norm": 1.8397691249847412,
      "learning_rate": 2.9521991045562287e-05,
      "loss": 1.7463,
      "step": 15561
    },
    {
      "epoch": 0.4098498814853832,
      "grad_norm": 3.843637466430664,
      "learning_rate": 2.95206742164867e-05,
      "loss": 1.6918,
      "step": 15562
    },
    {
      "epoch": 0.4098762180668949,
      "grad_norm": 2.0925049781799316,
      "learning_rate": 2.9519357387411118e-05,
      "loss": 1.312,
      "step": 15563
    },
    {
      "epoch": 0.40990255464840664,
      "grad_norm": 2.4778192043304443,
      "learning_rate": 2.951804055833553e-05,
      "loss": 1.1761,
      "step": 15564
    },
    {
      "epoch": 0.4099288912299184,
      "grad_norm": 1.7483197450637817,
      "learning_rate": 2.9516723729259942e-05,
      "loss": 2.0501,
      "step": 15565
    },
    {
      "epoch": 0.40995522781143007,
      "grad_norm": 2.340784788131714,
      "learning_rate": 2.9515406900184355e-05,
      "loss": 0.9956,
      "step": 15566
    },
    {
      "epoch": 0.4099815643929418,
      "grad_norm": 2.448150396347046,
      "learning_rate": 2.9514090071108774e-05,
      "loss": 2.1957,
      "step": 15567
    },
    {
      "epoch": 0.4100079009744535,
      "grad_norm": 2.7702066898345947,
      "learning_rate": 2.9512773242033186e-05,
      "loss": 1.0067,
      "step": 15568
    },
    {
      "epoch": 0.41003423755596524,
      "grad_norm": 1.5721662044525146,
      "learning_rate": 2.95114564129576e-05,
      "loss": 1.856,
      "step": 15569
    },
    {
      "epoch": 0.41006057413747693,
      "grad_norm": 2.173760175704956,
      "learning_rate": 2.9510139583882014e-05,
      "loss": 0.4897,
      "step": 15570
    },
    {
      "epoch": 0.4100869107189887,
      "grad_norm": 2.960336923599243,
      "learning_rate": 2.9508822754806426e-05,
      "loss": 1.3055,
      "step": 15571
    },
    {
      "epoch": 0.4101132473005004,
      "grad_norm": 1.64013671875,
      "learning_rate": 2.9507505925730845e-05,
      "loss": 1.875,
      "step": 15572
    },
    {
      "epoch": 0.4101395838820121,
      "grad_norm": 2.414738893508911,
      "learning_rate": 2.9506189096655257e-05,
      "loss": 2.0077,
      "step": 15573
    },
    {
      "epoch": 0.41016592046352385,
      "grad_norm": 3.0318429470062256,
      "learning_rate": 2.950487226757967e-05,
      "loss": 1.9638,
      "step": 15574
    },
    {
      "epoch": 0.41019225704503554,
      "grad_norm": 3.7558023929595947,
      "learning_rate": 2.950355543850408e-05,
      "loss": 1.4986,
      "step": 15575
    },
    {
      "epoch": 0.4102185936265473,
      "grad_norm": 2.0034916400909424,
      "learning_rate": 2.9502238609428494e-05,
      "loss": 1.9444,
      "step": 15576
    },
    {
      "epoch": 0.41024493020805897,
      "grad_norm": 1.6488306522369385,
      "learning_rate": 2.9500921780352913e-05,
      "loss": 2.0599,
      "step": 15577
    },
    {
      "epoch": 0.4102712667895707,
      "grad_norm": 1.8521102666854858,
      "learning_rate": 2.9499604951277325e-05,
      "loss": 1.2668,
      "step": 15578
    },
    {
      "epoch": 0.41029760337108245,
      "grad_norm": 1.8435574769973755,
      "learning_rate": 2.949828812220174e-05,
      "loss": 1.0013,
      "step": 15579
    },
    {
      "epoch": 0.41032393995259414,
      "grad_norm": 5.426388740539551,
      "learning_rate": 2.9496971293126153e-05,
      "loss": 1.9879,
      "step": 15580
    },
    {
      "epoch": 0.4103502765341059,
      "grad_norm": 2.0824577808380127,
      "learning_rate": 2.9495654464050565e-05,
      "loss": 1.8558,
      "step": 15581
    },
    {
      "epoch": 0.4103766131156176,
      "grad_norm": 1.792357325553894,
      "learning_rate": 2.9494337634974984e-05,
      "loss": 1.7638,
      "step": 15582
    },
    {
      "epoch": 0.4104029496971293,
      "grad_norm": 3.8681271076202393,
      "learning_rate": 2.9493020805899396e-05,
      "loss": 1.761,
      "step": 15583
    },
    {
      "epoch": 0.41042928627864106,
      "grad_norm": 1.743836760520935,
      "learning_rate": 2.9491703976823808e-05,
      "loss": 1.5564,
      "step": 15584
    },
    {
      "epoch": 0.41045562286015275,
      "grad_norm": 1.7832916975021362,
      "learning_rate": 2.949038714774822e-05,
      "loss": 1.8188,
      "step": 15585
    },
    {
      "epoch": 0.4104819594416645,
      "grad_norm": 2.4880542755126953,
      "learning_rate": 2.948907031867264e-05,
      "loss": 1.5534,
      "step": 15586
    },
    {
      "epoch": 0.4105082960231762,
      "grad_norm": 4.8710618019104,
      "learning_rate": 2.948775348959705e-05,
      "loss": 1.84,
      "step": 15587
    },
    {
      "epoch": 0.4105346326046879,
      "grad_norm": 1.872733235359192,
      "learning_rate": 2.9486436660521467e-05,
      "loss": 1.9138,
      "step": 15588
    },
    {
      "epoch": 0.4105609691861996,
      "grad_norm": 1.8661011457443237,
      "learning_rate": 2.948511983144588e-05,
      "loss": 1.635,
      "step": 15589
    },
    {
      "epoch": 0.41058730576771135,
      "grad_norm": 1.962339997291565,
      "learning_rate": 2.948380300237029e-05,
      "loss": 1.3617,
      "step": 15590
    },
    {
      "epoch": 0.4106136423492231,
      "grad_norm": 3.3088715076446533,
      "learning_rate": 2.948248617329471e-05,
      "loss": 1.6787,
      "step": 15591
    },
    {
      "epoch": 0.4106399789307348,
      "grad_norm": 2.4555041790008545,
      "learning_rate": 2.9481169344219123e-05,
      "loss": 1.7043,
      "step": 15592
    },
    {
      "epoch": 0.4106663155122465,
      "grad_norm": 1.9453097581863403,
      "learning_rate": 2.9479852515143535e-05,
      "loss": 1.475,
      "step": 15593
    },
    {
      "epoch": 0.4106926520937582,
      "grad_norm": 2.8604366779327393,
      "learning_rate": 2.9478535686067947e-05,
      "loss": 1.1677,
      "step": 15594
    },
    {
      "epoch": 0.41071898867526996,
      "grad_norm": 3.6796457767486572,
      "learning_rate": 2.9477218856992363e-05,
      "loss": 1.6573,
      "step": 15595
    },
    {
      "epoch": 0.41074532525678165,
      "grad_norm": 2.751086473464966,
      "learning_rate": 2.947590202791678e-05,
      "loss": 2.5069,
      "step": 15596
    },
    {
      "epoch": 0.4107716618382934,
      "grad_norm": 2.846808433532715,
      "learning_rate": 2.9474585198841194e-05,
      "loss": 0.495,
      "step": 15597
    },
    {
      "epoch": 0.41079799841980513,
      "grad_norm": 2.100917339324951,
      "learning_rate": 2.9473268369765606e-05,
      "loss": 1.4297,
      "step": 15598
    },
    {
      "epoch": 0.4108243350013168,
      "grad_norm": 2.561904191970825,
      "learning_rate": 2.947195154069002e-05,
      "loss": 1.4738,
      "step": 15599
    },
    {
      "epoch": 0.41085067158282856,
      "grad_norm": 3.0994951725006104,
      "learning_rate": 2.9470634711614437e-05,
      "loss": 0.3795,
      "step": 15600
    },
    {
      "epoch": 0.41087700816434025,
      "grad_norm": 2.137200117111206,
      "learning_rate": 2.946931788253885e-05,
      "loss": 2.4627,
      "step": 15601
    },
    {
      "epoch": 0.410903344745852,
      "grad_norm": 1.5876655578613281,
      "learning_rate": 2.9468001053463262e-05,
      "loss": 1.7906,
      "step": 15602
    },
    {
      "epoch": 0.4109296813273637,
      "grad_norm": 2.9217071533203125,
      "learning_rate": 2.9466684224387674e-05,
      "loss": 2.1725,
      "step": 15603
    },
    {
      "epoch": 0.4109560179088754,
      "grad_norm": 1.8206844329833984,
      "learning_rate": 2.9465367395312086e-05,
      "loss": 1.6498,
      "step": 15604
    },
    {
      "epoch": 0.41098235449038717,
      "grad_norm": 2.162393093109131,
      "learning_rate": 2.9464050566236505e-05,
      "loss": 2.1612,
      "step": 15605
    },
    {
      "epoch": 0.41100869107189886,
      "grad_norm": 2.7006442546844482,
      "learning_rate": 2.9462733737160917e-05,
      "loss": 1.3873,
      "step": 15606
    },
    {
      "epoch": 0.4110350276534106,
      "grad_norm": 1.8123433589935303,
      "learning_rate": 2.9461416908085333e-05,
      "loss": 2.2301,
      "step": 15607
    },
    {
      "epoch": 0.4110613642349223,
      "grad_norm": 1.8404465913772583,
      "learning_rate": 2.9460100079009745e-05,
      "loss": 1.8474,
      "step": 15608
    },
    {
      "epoch": 0.41108770081643403,
      "grad_norm": 4.121599197387695,
      "learning_rate": 2.9458783249934158e-05,
      "loss": 2.6606,
      "step": 15609
    },
    {
      "epoch": 0.4111140373979457,
      "grad_norm": 1.860534429550171,
      "learning_rate": 2.9457466420858576e-05,
      "loss": 2.0835,
      "step": 15610
    },
    {
      "epoch": 0.41114037397945746,
      "grad_norm": 1.944881796836853,
      "learning_rate": 2.945614959178299e-05,
      "loss": 1.8828,
      "step": 15611
    },
    {
      "epoch": 0.4111667105609692,
      "grad_norm": 1.9096068143844604,
      "learning_rate": 2.94548327627074e-05,
      "loss": 0.4872,
      "step": 15612
    },
    {
      "epoch": 0.4111930471424809,
      "grad_norm": 1.6398636102676392,
      "learning_rate": 2.9453515933631813e-05,
      "loss": 1.9815,
      "step": 15613
    },
    {
      "epoch": 0.41121938372399264,
      "grad_norm": 1.7676432132720947,
      "learning_rate": 2.945219910455623e-05,
      "loss": 1.3109,
      "step": 15614
    },
    {
      "epoch": 0.4112457203055043,
      "grad_norm": 5.615110874176025,
      "learning_rate": 2.9450882275480644e-05,
      "loss": 1.7162,
      "step": 15615
    },
    {
      "epoch": 0.41127205688701607,
      "grad_norm": 5.8755316734313965,
      "learning_rate": 2.944956544640506e-05,
      "loss": 2.2305,
      "step": 15616
    },
    {
      "epoch": 0.4112983934685278,
      "grad_norm": 3.4218411445617676,
      "learning_rate": 2.9448248617329472e-05,
      "loss": 1.588,
      "step": 15617
    },
    {
      "epoch": 0.4113247300500395,
      "grad_norm": 2.338719129562378,
      "learning_rate": 2.9446931788253884e-05,
      "loss": 2.0573,
      "step": 15618
    },
    {
      "epoch": 0.41135106663155124,
      "grad_norm": 3.0660626888275146,
      "learning_rate": 2.9445614959178303e-05,
      "loss": 2.1837,
      "step": 15619
    },
    {
      "epoch": 0.41137740321306293,
      "grad_norm": 2.0762314796447754,
      "learning_rate": 2.9444298130102716e-05,
      "loss": 1.0922,
      "step": 15620
    },
    {
      "epoch": 0.4114037397945747,
      "grad_norm": 1.964686393737793,
      "learning_rate": 2.9442981301027128e-05,
      "loss": 1.9216,
      "step": 15621
    },
    {
      "epoch": 0.41143007637608636,
      "grad_norm": 2.870673179626465,
      "learning_rate": 2.944166447195154e-05,
      "loss": 1.2022,
      "step": 15622
    },
    {
      "epoch": 0.4114564129575981,
      "grad_norm": 2.181758165359497,
      "learning_rate": 2.9440347642875956e-05,
      "loss": 1.9927,
      "step": 15623
    },
    {
      "epoch": 0.41148274953910985,
      "grad_norm": 1.7397756576538086,
      "learning_rate": 2.943903081380037e-05,
      "loss": 1.7732,
      "step": 15624
    },
    {
      "epoch": 0.41150908612062154,
      "grad_norm": 4.026196002960205,
      "learning_rate": 2.9437713984724787e-05,
      "loss": 1.7795,
      "step": 15625
    },
    {
      "epoch": 0.4115354227021333,
      "grad_norm": 2.311596393585205,
      "learning_rate": 2.94363971556492e-05,
      "loss": 1.8337,
      "step": 15626
    },
    {
      "epoch": 0.41156175928364497,
      "grad_norm": 2.982847213745117,
      "learning_rate": 2.943508032657361e-05,
      "loss": 0.5568,
      "step": 15627
    },
    {
      "epoch": 0.4115880958651567,
      "grad_norm": 4.708019256591797,
      "learning_rate": 2.9433763497498023e-05,
      "loss": 1.2951,
      "step": 15628
    },
    {
      "epoch": 0.4116144324466684,
      "grad_norm": 2.341606616973877,
      "learning_rate": 2.9432446668422442e-05,
      "loss": 1.5093,
      "step": 15629
    },
    {
      "epoch": 0.41164076902818014,
      "grad_norm": 2.5291550159454346,
      "learning_rate": 2.9431129839346855e-05,
      "loss": 1.6805,
      "step": 15630
    },
    {
      "epoch": 0.4116671056096919,
      "grad_norm": 2.195976734161377,
      "learning_rate": 2.9429813010271267e-05,
      "loss": 1.7182,
      "step": 15631
    },
    {
      "epoch": 0.4116934421912036,
      "grad_norm": 2.999455451965332,
      "learning_rate": 2.942849618119568e-05,
      "loss": 1.9971,
      "step": 15632
    },
    {
      "epoch": 0.4117197787727153,
      "grad_norm": 1.6503539085388184,
      "learning_rate": 2.9427179352120098e-05,
      "loss": 1.8061,
      "step": 15633
    },
    {
      "epoch": 0.411746115354227,
      "grad_norm": 1.6644328832626343,
      "learning_rate": 2.942586252304451e-05,
      "loss": 2.4462,
      "step": 15634
    },
    {
      "epoch": 0.41177245193573875,
      "grad_norm": 1.9927966594696045,
      "learning_rate": 2.9424545693968926e-05,
      "loss": 1.8709,
      "step": 15635
    },
    {
      "epoch": 0.41179878851725044,
      "grad_norm": 1.9618141651153564,
      "learning_rate": 2.9423228864893338e-05,
      "loss": 1.4128,
      "step": 15636
    },
    {
      "epoch": 0.4118251250987622,
      "grad_norm": 1.914323329925537,
      "learning_rate": 2.942191203581775e-05,
      "loss": 2.2699,
      "step": 15637
    },
    {
      "epoch": 0.4118514616802739,
      "grad_norm": 1.8678549528121948,
      "learning_rate": 2.942059520674217e-05,
      "loss": 1.647,
      "step": 15638
    },
    {
      "epoch": 0.4118777982617856,
      "grad_norm": 2.154546022415161,
      "learning_rate": 2.941927837766658e-05,
      "loss": 1.4129,
      "step": 15639
    },
    {
      "epoch": 0.41190413484329735,
      "grad_norm": 1.3817179203033447,
      "learning_rate": 2.9417961548590994e-05,
      "loss": 1.8555,
      "step": 15640
    },
    {
      "epoch": 0.41193047142480904,
      "grad_norm": 2.392235517501831,
      "learning_rate": 2.9416644719515406e-05,
      "loss": 1.6919,
      "step": 15641
    },
    {
      "epoch": 0.4119568080063208,
      "grad_norm": 2.5953352451324463,
      "learning_rate": 2.941532789043982e-05,
      "loss": 1.9096,
      "step": 15642
    },
    {
      "epoch": 0.4119831445878325,
      "grad_norm": 1.8519266843795776,
      "learning_rate": 2.9414011061364237e-05,
      "loss": 0.5479,
      "step": 15643
    },
    {
      "epoch": 0.4120094811693442,
      "grad_norm": 2.8138234615325928,
      "learning_rate": 2.9412694232288653e-05,
      "loss": 0.9956,
      "step": 15644
    },
    {
      "epoch": 0.41203581775085596,
      "grad_norm": 2.1778721809387207,
      "learning_rate": 2.9411377403213065e-05,
      "loss": 0.7772,
      "step": 15645
    },
    {
      "epoch": 0.41206215433236765,
      "grad_norm": 2.140077829360962,
      "learning_rate": 2.9410060574137477e-05,
      "loss": 1.8864,
      "step": 15646
    },
    {
      "epoch": 0.4120884909138794,
      "grad_norm": 6.711975574493408,
      "learning_rate": 2.940874374506189e-05,
      "loss": 0.829,
      "step": 15647
    },
    {
      "epoch": 0.4121148274953911,
      "grad_norm": 1.9497487545013428,
      "learning_rate": 2.9407426915986308e-05,
      "loss": 1.9392,
      "step": 15648
    },
    {
      "epoch": 0.4121411640769028,
      "grad_norm": 3.6461007595062256,
      "learning_rate": 2.940611008691072e-05,
      "loss": 1.3397,
      "step": 15649
    },
    {
      "epoch": 0.41216750065841457,
      "grad_norm": 2.270423412322998,
      "learning_rate": 2.9404793257835133e-05,
      "loss": 1.6914,
      "step": 15650
    },
    {
      "epoch": 0.41219383723992625,
      "grad_norm": 2.1901330947875977,
      "learning_rate": 2.9403476428759548e-05,
      "loss": 0.9298,
      "step": 15651
    },
    {
      "epoch": 0.412220173821438,
      "grad_norm": 1.6930476427078247,
      "learning_rate": 2.9402159599683964e-05,
      "loss": 1.15,
      "step": 15652
    },
    {
      "epoch": 0.4122465104029497,
      "grad_norm": 11.654403686523438,
      "learning_rate": 2.940084277060838e-05,
      "loss": 1.1725,
      "step": 15653
    },
    {
      "epoch": 0.4122728469844614,
      "grad_norm": 1.8939121961593628,
      "learning_rate": 2.939952594153279e-05,
      "loss": 2.4951,
      "step": 15654
    },
    {
      "epoch": 0.4122991835659731,
      "grad_norm": 1.7732517719268799,
      "learning_rate": 2.9398209112457204e-05,
      "loss": 1.036,
      "step": 15655
    },
    {
      "epoch": 0.41232552014748486,
      "grad_norm": 2.204071521759033,
      "learning_rate": 2.9396892283381616e-05,
      "loss": 2.1034,
      "step": 15656
    },
    {
      "epoch": 0.4123518567289966,
      "grad_norm": 2.244224786758423,
      "learning_rate": 2.9395575454306035e-05,
      "loss": 2.2298,
      "step": 15657
    },
    {
      "epoch": 0.4123781933105083,
      "grad_norm": 1.5890324115753174,
      "learning_rate": 2.9394258625230447e-05,
      "loss": 1.909,
      "step": 15658
    },
    {
      "epoch": 0.41240452989202003,
      "grad_norm": 2.022317409515381,
      "learning_rate": 2.939294179615486e-05,
      "loss": 1.4609,
      "step": 15659
    },
    {
      "epoch": 0.4124308664735317,
      "grad_norm": 1.8920706510543823,
      "learning_rate": 2.939162496707927e-05,
      "loss": 1.3667,
      "step": 15660
    },
    {
      "epoch": 0.41245720305504346,
      "grad_norm": 2.809635639190674,
      "learning_rate": 2.9390308138003687e-05,
      "loss": 1.5838,
      "step": 15661
    },
    {
      "epoch": 0.41248353963655515,
      "grad_norm": 2.007632255554199,
      "learning_rate": 2.9388991308928103e-05,
      "loss": 0.4959,
      "step": 15662
    },
    {
      "epoch": 0.4125098762180669,
      "grad_norm": 3.5080292224884033,
      "learning_rate": 2.938767447985252e-05,
      "loss": 0.9457,
      "step": 15663
    },
    {
      "epoch": 0.41253621279957864,
      "grad_norm": 2.179311513900757,
      "learning_rate": 2.938635765077693e-05,
      "loss": 2.4743,
      "step": 15664
    },
    {
      "epoch": 0.4125625493810903,
      "grad_norm": 1.788968563079834,
      "learning_rate": 2.9385040821701343e-05,
      "loss": 1.6155,
      "step": 15665
    },
    {
      "epoch": 0.41258888596260207,
      "grad_norm": 2.640054702758789,
      "learning_rate": 2.9383723992625762e-05,
      "loss": 1.6905,
      "step": 15666
    },
    {
      "epoch": 0.41261522254411376,
      "grad_norm": 3.193744421005249,
      "learning_rate": 2.9382407163550174e-05,
      "loss": 1.9119,
      "step": 15667
    },
    {
      "epoch": 0.4126415591256255,
      "grad_norm": 2.9051668643951416,
      "learning_rate": 2.9381090334474586e-05,
      "loss": 1.6336,
      "step": 15668
    },
    {
      "epoch": 0.4126678957071372,
      "grad_norm": 3.2675421237945557,
      "learning_rate": 2.9379773505399e-05,
      "loss": 0.9673,
      "step": 15669
    },
    {
      "epoch": 0.41269423228864893,
      "grad_norm": 1.8139410018920898,
      "learning_rate": 2.9378456676323414e-05,
      "loss": 1.6944,
      "step": 15670
    },
    {
      "epoch": 0.4127205688701607,
      "grad_norm": 2.856947898864746,
      "learning_rate": 2.937713984724783e-05,
      "loss": 0.563,
      "step": 15671
    },
    {
      "epoch": 0.41274690545167236,
      "grad_norm": 2.128417491912842,
      "learning_rate": 2.9375823018172245e-05,
      "loss": 1.8091,
      "step": 15672
    },
    {
      "epoch": 0.4127732420331841,
      "grad_norm": 1.8805828094482422,
      "learning_rate": 2.9374506189096658e-05,
      "loss": 1.9038,
      "step": 15673
    },
    {
      "epoch": 0.4127995786146958,
      "grad_norm": 1.591185212135315,
      "learning_rate": 2.937318936002107e-05,
      "loss": 1.6888,
      "step": 15674
    },
    {
      "epoch": 0.41282591519620754,
      "grad_norm": 3.423896312713623,
      "learning_rate": 2.9371872530945482e-05,
      "loss": 1.0556,
      "step": 15675
    },
    {
      "epoch": 0.4128522517777192,
      "grad_norm": 1.8265759944915771,
      "learning_rate": 2.93705557018699e-05,
      "loss": 1.3261,
      "step": 15676
    },
    {
      "epoch": 0.41287858835923097,
      "grad_norm": 3.0720016956329346,
      "learning_rate": 2.9369238872794313e-05,
      "loss": 0.9389,
      "step": 15677
    },
    {
      "epoch": 0.4129049249407427,
      "grad_norm": 1.8436627388000488,
      "learning_rate": 2.9367922043718725e-05,
      "loss": 1.9235,
      "step": 15678
    },
    {
      "epoch": 0.4129312615222544,
      "grad_norm": 2.1809499263763428,
      "learning_rate": 2.9366605214643138e-05,
      "loss": 2.3201,
      "step": 15679
    },
    {
      "epoch": 0.41295759810376614,
      "grad_norm": 5.844761848449707,
      "learning_rate": 2.9365288385567553e-05,
      "loss": 1.2053,
      "step": 15680
    },
    {
      "epoch": 0.41298393468527783,
      "grad_norm": 1.7083134651184082,
      "learning_rate": 2.936397155649197e-05,
      "loss": 1.8222,
      "step": 15681
    },
    {
      "epoch": 0.4130102712667896,
      "grad_norm": 4.175252914428711,
      "learning_rate": 2.9362654727416384e-05,
      "loss": 1.3668,
      "step": 15682
    },
    {
      "epoch": 0.4130366078483013,
      "grad_norm": 1.6955143213272095,
      "learning_rate": 2.9361337898340797e-05,
      "loss": 1.9421,
      "step": 15683
    },
    {
      "epoch": 0.413062944429813,
      "grad_norm": 3.372965097427368,
      "learning_rate": 2.936002106926521e-05,
      "loss": 1.31,
      "step": 15684
    },
    {
      "epoch": 0.41308928101132475,
      "grad_norm": 1.7077680826187134,
      "learning_rate": 2.9358704240189628e-05,
      "loss": 0.4251,
      "step": 15685
    },
    {
      "epoch": 0.41311561759283644,
      "grad_norm": 1.6981302499771118,
      "learning_rate": 2.935738741111404e-05,
      "loss": 1.9722,
      "step": 15686
    },
    {
      "epoch": 0.4131419541743482,
      "grad_norm": 1.8662424087524414,
      "learning_rate": 2.9356070582038452e-05,
      "loss": 1.7829,
      "step": 15687
    },
    {
      "epoch": 0.41316829075585987,
      "grad_norm": 2.0773918628692627,
      "learning_rate": 2.9354753752962864e-05,
      "loss": 1.0619,
      "step": 15688
    },
    {
      "epoch": 0.4131946273373716,
      "grad_norm": 2.776118755340576,
      "learning_rate": 2.935343692388728e-05,
      "loss": 0.2652,
      "step": 15689
    },
    {
      "epoch": 0.41322096391888335,
      "grad_norm": 3.3183960914611816,
      "learning_rate": 2.9352120094811696e-05,
      "loss": 1.7356,
      "step": 15690
    },
    {
      "epoch": 0.41324730050039504,
      "grad_norm": 1.9203873872756958,
      "learning_rate": 2.935080326573611e-05,
      "loss": 1.8409,
      "step": 15691
    },
    {
      "epoch": 0.4132736370819068,
      "grad_norm": 4.229273796081543,
      "learning_rate": 2.9349486436660523e-05,
      "loss": 1.3657,
      "step": 15692
    },
    {
      "epoch": 0.4132999736634185,
      "grad_norm": 2.3400189876556396,
      "learning_rate": 2.9348169607584936e-05,
      "loss": 2.1804,
      "step": 15693
    },
    {
      "epoch": 0.4133263102449302,
      "grad_norm": 2.9474289417266846,
      "learning_rate": 2.9346852778509348e-05,
      "loss": 1.4853,
      "step": 15694
    },
    {
      "epoch": 0.4133526468264419,
      "grad_norm": 1.7531671524047852,
      "learning_rate": 2.9345535949433767e-05,
      "loss": 2.0666,
      "step": 15695
    },
    {
      "epoch": 0.41337898340795365,
      "grad_norm": 1.5099817514419556,
      "learning_rate": 2.934421912035818e-05,
      "loss": 2.0596,
      "step": 15696
    },
    {
      "epoch": 0.4134053199894654,
      "grad_norm": 1.5720902681350708,
      "learning_rate": 2.934290229128259e-05,
      "loss": 0.6829,
      "step": 15697
    },
    {
      "epoch": 0.4134316565709771,
      "grad_norm": 4.438876152038574,
      "learning_rate": 2.9341585462207007e-05,
      "loss": 1.9974,
      "step": 15698
    },
    {
      "epoch": 0.4134579931524888,
      "grad_norm": 1.7951011657714844,
      "learning_rate": 2.9340268633131422e-05,
      "loss": 1.7568,
      "step": 15699
    },
    {
      "epoch": 0.4134843297340005,
      "grad_norm": 4.371419906616211,
      "learning_rate": 2.9338951804055838e-05,
      "loss": 1.6832,
      "step": 15700
    },
    {
      "epoch": 0.41351066631551225,
      "grad_norm": 1.3973851203918457,
      "learning_rate": 2.933763497498025e-05,
      "loss": 2.046,
      "step": 15701
    },
    {
      "epoch": 0.41353700289702394,
      "grad_norm": 2.2307183742523193,
      "learning_rate": 2.9336318145904662e-05,
      "loss": 1.6017,
      "step": 15702
    },
    {
      "epoch": 0.4135633394785357,
      "grad_norm": 1.8933690786361694,
      "learning_rate": 2.9335001316829075e-05,
      "loss": 1.7058,
      "step": 15703
    },
    {
      "epoch": 0.41358967606004743,
      "grad_norm": 1.9415560960769653,
      "learning_rate": 2.9333684487753494e-05,
      "loss": 1.6413,
      "step": 15704
    },
    {
      "epoch": 0.4136160126415591,
      "grad_norm": 1.902345895767212,
      "learning_rate": 2.9332367658677906e-05,
      "loss": 1.4989,
      "step": 15705
    },
    {
      "epoch": 0.41364234922307086,
      "grad_norm": 4.692122936248779,
      "learning_rate": 2.9331050829602318e-05,
      "loss": 0.7606,
      "step": 15706
    },
    {
      "epoch": 0.41366868580458255,
      "grad_norm": 2.022602081298828,
      "learning_rate": 2.932973400052673e-05,
      "loss": 1.405,
      "step": 15707
    },
    {
      "epoch": 0.4136950223860943,
      "grad_norm": 2.992035388946533,
      "learning_rate": 2.9328417171451146e-05,
      "loss": 0.4388,
      "step": 15708
    },
    {
      "epoch": 0.413721358967606,
      "grad_norm": 2.544341802597046,
      "learning_rate": 2.932710034237556e-05,
      "loss": 1.5207,
      "step": 15709
    },
    {
      "epoch": 0.4137476955491177,
      "grad_norm": 1.8995567560195923,
      "learning_rate": 2.9325783513299977e-05,
      "loss": 0.8497,
      "step": 15710
    },
    {
      "epoch": 0.41377403213062947,
      "grad_norm": 1.6597929000854492,
      "learning_rate": 2.932446668422439e-05,
      "loss": 1.4809,
      "step": 15711
    },
    {
      "epoch": 0.41380036871214115,
      "grad_norm": 3.636852502822876,
      "learning_rate": 2.93231498551488e-05,
      "loss": 1.3056,
      "step": 15712
    },
    {
      "epoch": 0.4138267052936529,
      "grad_norm": 1.5252666473388672,
      "learning_rate": 2.9321833026073214e-05,
      "loss": 1.4693,
      "step": 15713
    },
    {
      "epoch": 0.4138530418751646,
      "grad_norm": 1.7964345216751099,
      "learning_rate": 2.9320516196997633e-05,
      "loss": 1.7382,
      "step": 15714
    },
    {
      "epoch": 0.41387937845667633,
      "grad_norm": 1.7553869485855103,
      "learning_rate": 2.9319199367922045e-05,
      "loss": 1.8619,
      "step": 15715
    },
    {
      "epoch": 0.413905715038188,
      "grad_norm": 2.1346728801727295,
      "learning_rate": 2.9317882538846457e-05,
      "loss": 1.9323,
      "step": 15716
    },
    {
      "epoch": 0.41393205161969976,
      "grad_norm": 1.9322916269302368,
      "learning_rate": 2.9316565709770873e-05,
      "loss": 2.2971,
      "step": 15717
    },
    {
      "epoch": 0.4139583882012115,
      "grad_norm": 3.9815216064453125,
      "learning_rate": 2.9315248880695288e-05,
      "loss": 1.924,
      "step": 15718
    },
    {
      "epoch": 0.4139847247827232,
      "grad_norm": 4.49163818359375,
      "learning_rate": 2.9313932051619704e-05,
      "loss": 1.0998,
      "step": 15719
    },
    {
      "epoch": 0.41401106136423493,
      "grad_norm": 3.1728100776672363,
      "learning_rate": 2.9312615222544116e-05,
      "loss": 0.986,
      "step": 15720
    },
    {
      "epoch": 0.4140373979457466,
      "grad_norm": 1.8438752889633179,
      "learning_rate": 2.9311298393468528e-05,
      "loss": 2.7322,
      "step": 15721
    },
    {
      "epoch": 0.41406373452725836,
      "grad_norm": 2.5315842628479004,
      "learning_rate": 2.930998156439294e-05,
      "loss": 1.8991,
      "step": 15722
    },
    {
      "epoch": 0.4140900711087701,
      "grad_norm": 1.876872181892395,
      "learning_rate": 2.930866473531736e-05,
      "loss": 0.6388,
      "step": 15723
    },
    {
      "epoch": 0.4141164076902818,
      "grad_norm": 2.0644257068634033,
      "learning_rate": 2.930734790624177e-05,
      "loss": 1.7264,
      "step": 15724
    },
    {
      "epoch": 0.41414274427179354,
      "grad_norm": 2.4856345653533936,
      "learning_rate": 2.9306031077166184e-05,
      "loss": 1.5433,
      "step": 15725
    },
    {
      "epoch": 0.4141690808533052,
      "grad_norm": 1.8194998502731323,
      "learning_rate": 2.93047142480906e-05,
      "loss": 1.6386,
      "step": 15726
    },
    {
      "epoch": 0.41419541743481697,
      "grad_norm": 1.9625462293624878,
      "learning_rate": 2.9303397419015012e-05,
      "loss": 1.8224,
      "step": 15727
    },
    {
      "epoch": 0.41422175401632866,
      "grad_norm": 1.771073818206787,
      "learning_rate": 2.930208058993943e-05,
      "loss": 1.9254,
      "step": 15728
    },
    {
      "epoch": 0.4142480905978404,
      "grad_norm": 1.9150766134262085,
      "learning_rate": 2.9300763760863843e-05,
      "loss": 1.7029,
      "step": 15729
    },
    {
      "epoch": 0.41427442717935214,
      "grad_norm": 3.26265025138855,
      "learning_rate": 2.9299446931788255e-05,
      "loss": 1.9199,
      "step": 15730
    },
    {
      "epoch": 0.41430076376086383,
      "grad_norm": 2.0969038009643555,
      "learning_rate": 2.9298130102712667e-05,
      "loss": 2.1066,
      "step": 15731
    },
    {
      "epoch": 0.4143271003423756,
      "grad_norm": 1.9390043020248413,
      "learning_rate": 2.9296813273637086e-05,
      "loss": 1.9867,
      "step": 15732
    },
    {
      "epoch": 0.41435343692388726,
      "grad_norm": 2.376207113265991,
      "learning_rate": 2.92954964445615e-05,
      "loss": 1.8355,
      "step": 15733
    },
    {
      "epoch": 0.414379773505399,
      "grad_norm": 1.623523235321045,
      "learning_rate": 2.929417961548591e-05,
      "loss": 1.8488,
      "step": 15734
    },
    {
      "epoch": 0.4144061100869107,
      "grad_norm": 3.2486753463745117,
      "learning_rate": 2.9292862786410323e-05,
      "loss": 1.2581,
      "step": 15735
    },
    {
      "epoch": 0.41443244666842244,
      "grad_norm": 3.6002862453460693,
      "learning_rate": 2.929154595733474e-05,
      "loss": 0.5613,
      "step": 15736
    },
    {
      "epoch": 0.4144587832499342,
      "grad_norm": 2.0421414375305176,
      "learning_rate": 2.9290229128259154e-05,
      "loss": 0.2003,
      "step": 15737
    },
    {
      "epoch": 0.41448511983144587,
      "grad_norm": 4.687199115753174,
      "learning_rate": 2.928891229918357e-05,
      "loss": 1.2453,
      "step": 15738
    },
    {
      "epoch": 0.4145114564129576,
      "grad_norm": 3.138654947280884,
      "learning_rate": 2.9287595470107982e-05,
      "loss": 2.3656,
      "step": 15739
    },
    {
      "epoch": 0.4145377929944693,
      "grad_norm": 1.958830714225769,
      "learning_rate": 2.9286278641032394e-05,
      "loss": 1.2919,
      "step": 15740
    },
    {
      "epoch": 0.41456412957598104,
      "grad_norm": 1.6532645225524902,
      "learning_rate": 2.9284961811956806e-05,
      "loss": 1.6098,
      "step": 15741
    },
    {
      "epoch": 0.41459046615749273,
      "grad_norm": 2.0887038707733154,
      "learning_rate": 2.9283644982881225e-05,
      "loss": 1.334,
      "step": 15742
    },
    {
      "epoch": 0.4146168027390045,
      "grad_norm": 2.085057020187378,
      "learning_rate": 2.9282328153805638e-05,
      "loss": 1.8236,
      "step": 15743
    },
    {
      "epoch": 0.4146431393205162,
      "grad_norm": 2.1905174255371094,
      "learning_rate": 2.928101132473005e-05,
      "loss": 1.5245,
      "step": 15744
    },
    {
      "epoch": 0.4146694759020279,
      "grad_norm": 3.5042781829833984,
      "learning_rate": 2.9279694495654465e-05,
      "loss": 1.4656,
      "step": 15745
    },
    {
      "epoch": 0.41469581248353965,
      "grad_norm": 3.223958730697632,
      "learning_rate": 2.9278377666578878e-05,
      "loss": 0.6077,
      "step": 15746
    },
    {
      "epoch": 0.41472214906505134,
      "grad_norm": 1.6542402505874634,
      "learning_rate": 2.9277060837503297e-05,
      "loss": 1.5184,
      "step": 15747
    },
    {
      "epoch": 0.4147484856465631,
      "grad_norm": 3.485166072845459,
      "learning_rate": 2.927574400842771e-05,
      "loss": 1.7476,
      "step": 15748
    },
    {
      "epoch": 0.41477482222807477,
      "grad_norm": 1.248049259185791,
      "learning_rate": 2.927442717935212e-05,
      "loss": 1.5584,
      "step": 15749
    },
    {
      "epoch": 0.4148011588095865,
      "grad_norm": 1.49611496925354,
      "learning_rate": 2.9273110350276533e-05,
      "loss": 2.0072,
      "step": 15750
    },
    {
      "epoch": 0.41482749539109826,
      "grad_norm": 1.6200965642929077,
      "learning_rate": 2.9271793521200952e-05,
      "loss": 1.9917,
      "step": 15751
    },
    {
      "epoch": 0.41485383197260994,
      "grad_norm": 2.377087116241455,
      "learning_rate": 2.9270476692125364e-05,
      "loss": 1.5305,
      "step": 15752
    },
    {
      "epoch": 0.4148801685541217,
      "grad_norm": 1.693340539932251,
      "learning_rate": 2.9269159863049777e-05,
      "loss": 1.9082,
      "step": 15753
    },
    {
      "epoch": 0.4149065051356334,
      "grad_norm": 1.8340027332305908,
      "learning_rate": 2.926784303397419e-05,
      "loss": 1.5705,
      "step": 15754
    },
    {
      "epoch": 0.4149328417171451,
      "grad_norm": 2.9371960163116455,
      "learning_rate": 2.9266526204898604e-05,
      "loss": 1.625,
      "step": 15755
    },
    {
      "epoch": 0.41495917829865686,
      "grad_norm": 3.9172139167785645,
      "learning_rate": 2.9265209375823023e-05,
      "loss": 0.7591,
      "step": 15756
    },
    {
      "epoch": 0.41498551488016855,
      "grad_norm": 6.552670478820801,
      "learning_rate": 2.9263892546747436e-05,
      "loss": 1.898,
      "step": 15757
    },
    {
      "epoch": 0.4150118514616803,
      "grad_norm": 3.311701774597168,
      "learning_rate": 2.9262575717671848e-05,
      "loss": 2.1974,
      "step": 15758
    },
    {
      "epoch": 0.415038188043192,
      "grad_norm": 1.7526640892028809,
      "learning_rate": 2.926125888859626e-05,
      "loss": 2.0597,
      "step": 15759
    },
    {
      "epoch": 0.4150645246247037,
      "grad_norm": 4.0637359619140625,
      "learning_rate": 2.9259942059520672e-05,
      "loss": 1.9377,
      "step": 15760
    },
    {
      "epoch": 0.4150908612062154,
      "grad_norm": 3.394953966140747,
      "learning_rate": 2.925862523044509e-05,
      "loss": 2.076,
      "step": 15761
    },
    {
      "epoch": 0.41511719778772715,
      "grad_norm": 1.8149725198745728,
      "learning_rate": 2.9257308401369503e-05,
      "loss": 1.8685,
      "step": 15762
    },
    {
      "epoch": 0.4151435343692389,
      "grad_norm": 2.1194307804107666,
      "learning_rate": 2.9255991572293916e-05,
      "loss": 0.6746,
      "step": 15763
    },
    {
      "epoch": 0.4151698709507506,
      "grad_norm": 1.7504782676696777,
      "learning_rate": 2.925467474321833e-05,
      "loss": 1.8717,
      "step": 15764
    },
    {
      "epoch": 0.41519620753226233,
      "grad_norm": 5.995121002197266,
      "learning_rate": 2.9253357914142747e-05,
      "loss": 1.748,
      "step": 15765
    },
    {
      "epoch": 0.415222544113774,
      "grad_norm": 1.5490652322769165,
      "learning_rate": 2.9252041085067162e-05,
      "loss": 0.5824,
      "step": 15766
    },
    {
      "epoch": 0.41524888069528576,
      "grad_norm": 1.5723910331726074,
      "learning_rate": 2.9250724255991575e-05,
      "loss": 1.5109,
      "step": 15767
    },
    {
      "epoch": 0.41527521727679745,
      "grad_norm": 2.016627550125122,
      "learning_rate": 2.9249407426915987e-05,
      "loss": 1.8315,
      "step": 15768
    },
    {
      "epoch": 0.4153015538583092,
      "grad_norm": 1.6356393098831177,
      "learning_rate": 2.92480905978404e-05,
      "loss": 1.7194,
      "step": 15769
    },
    {
      "epoch": 0.41532789043982093,
      "grad_norm": 3.4606828689575195,
      "learning_rate": 2.9246773768764818e-05,
      "loss": 0.6875,
      "step": 15770
    },
    {
      "epoch": 0.4153542270213326,
      "grad_norm": 2.5112810134887695,
      "learning_rate": 2.924545693968923e-05,
      "loss": 1.7626,
      "step": 15771
    },
    {
      "epoch": 0.41538056360284437,
      "grad_norm": 1.9759786128997803,
      "learning_rate": 2.9244140110613642e-05,
      "loss": 1.3572,
      "step": 15772
    },
    {
      "epoch": 0.41540690018435605,
      "grad_norm": 5.558615207672119,
      "learning_rate": 2.9242823281538058e-05,
      "loss": 0.9321,
      "step": 15773
    },
    {
      "epoch": 0.4154332367658678,
      "grad_norm": 2.4439620971679688,
      "learning_rate": 2.924150645246247e-05,
      "loss": 0.9004,
      "step": 15774
    },
    {
      "epoch": 0.4154595733473795,
      "grad_norm": 2.812452793121338,
      "learning_rate": 2.924018962338689e-05,
      "loss": 1.4295,
      "step": 15775
    },
    {
      "epoch": 0.41548590992889123,
      "grad_norm": 1.741997241973877,
      "learning_rate": 2.92388727943113e-05,
      "loss": 2.7799,
      "step": 15776
    },
    {
      "epoch": 0.41551224651040297,
      "grad_norm": 1.874658226966858,
      "learning_rate": 2.9237555965235714e-05,
      "loss": 2.321,
      "step": 15777
    },
    {
      "epoch": 0.41553858309191466,
      "grad_norm": 2.0538578033447266,
      "learning_rate": 2.9236239136160126e-05,
      "loss": 1.8635,
      "step": 15778
    },
    {
      "epoch": 0.4155649196734264,
      "grad_norm": 1.9786607027053833,
      "learning_rate": 2.9234922307084538e-05,
      "loss": 2.4563,
      "step": 15779
    },
    {
      "epoch": 0.4155912562549381,
      "grad_norm": 2.311415433883667,
      "learning_rate": 2.9233605478008957e-05,
      "loss": 1.7576,
      "step": 15780
    },
    {
      "epoch": 0.41561759283644983,
      "grad_norm": 3.0113344192504883,
      "learning_rate": 2.923228864893337e-05,
      "loss": 0.5366,
      "step": 15781
    },
    {
      "epoch": 0.4156439294179615,
      "grad_norm": 2.2133822441101074,
      "learning_rate": 2.923097181985778e-05,
      "loss": 1.0131,
      "step": 15782
    },
    {
      "epoch": 0.41567026599947327,
      "grad_norm": 1.736491322517395,
      "learning_rate": 2.9229654990782197e-05,
      "loss": 2.6036,
      "step": 15783
    },
    {
      "epoch": 0.415696602580985,
      "grad_norm": 1.9607834815979004,
      "learning_rate": 2.9228338161706613e-05,
      "loss": 1.8037,
      "step": 15784
    },
    {
      "epoch": 0.4157229391624967,
      "grad_norm": 2.7674288749694824,
      "learning_rate": 2.9227021332631028e-05,
      "loss": 2.2169,
      "step": 15785
    },
    {
      "epoch": 0.41574927574400844,
      "grad_norm": 1.9476398229599,
      "learning_rate": 2.922570450355544e-05,
      "loss": 2.0498,
      "step": 15786
    },
    {
      "epoch": 0.4157756123255201,
      "grad_norm": 7.260297775268555,
      "learning_rate": 2.9224387674479853e-05,
      "loss": 0.6484,
      "step": 15787
    },
    {
      "epoch": 0.41580194890703187,
      "grad_norm": 1.9710294008255005,
      "learning_rate": 2.9223070845404265e-05,
      "loss": 1.9031,
      "step": 15788
    },
    {
      "epoch": 0.4158282854885436,
      "grad_norm": 1.526004433631897,
      "learning_rate": 2.9221754016328684e-05,
      "loss": 2.0321,
      "step": 15789
    },
    {
      "epoch": 0.4158546220700553,
      "grad_norm": 2.4040145874023438,
      "learning_rate": 2.9220437187253096e-05,
      "loss": 1.5207,
      "step": 15790
    },
    {
      "epoch": 0.41588095865156705,
      "grad_norm": 2.398714780807495,
      "learning_rate": 2.921912035817751e-05,
      "loss": 1.8171,
      "step": 15791
    },
    {
      "epoch": 0.41590729523307873,
      "grad_norm": 3.9659485816955566,
      "learning_rate": 2.9217803529101924e-05,
      "loss": 1.5815,
      "step": 15792
    },
    {
      "epoch": 0.4159336318145905,
      "grad_norm": 2.435270071029663,
      "learning_rate": 2.9216486700026336e-05,
      "loss": 2.3045,
      "step": 15793
    },
    {
      "epoch": 0.41595996839610216,
      "grad_norm": 2.325011968612671,
      "learning_rate": 2.9215169870950755e-05,
      "loss": 1.3099,
      "step": 15794
    },
    {
      "epoch": 0.4159863049776139,
      "grad_norm": 3.9564249515533447,
      "learning_rate": 2.9213853041875167e-05,
      "loss": 1.6155,
      "step": 15795
    },
    {
      "epoch": 0.41601264155912565,
      "grad_norm": 3.003162384033203,
      "learning_rate": 2.921253621279958e-05,
      "loss": 2.0027,
      "step": 15796
    },
    {
      "epoch": 0.41603897814063734,
      "grad_norm": 1.8829834461212158,
      "learning_rate": 2.9211219383723992e-05,
      "loss": 1.4139,
      "step": 15797
    },
    {
      "epoch": 0.4160653147221491,
      "grad_norm": 1.7265219688415527,
      "learning_rate": 2.920990255464841e-05,
      "loss": 1.6493,
      "step": 15798
    },
    {
      "epoch": 0.41609165130366077,
      "grad_norm": 1.6604843139648438,
      "learning_rate": 2.9208585725572823e-05,
      "loss": 2.0548,
      "step": 15799
    },
    {
      "epoch": 0.4161179878851725,
      "grad_norm": 2.3974146842956543,
      "learning_rate": 2.9207268896497235e-05,
      "loss": 1.7182,
      "step": 15800
    },
    {
      "epoch": 0.4161443244666842,
      "grad_norm": 1.6551018953323364,
      "learning_rate": 2.920595206742165e-05,
      "loss": 1.833,
      "step": 15801
    },
    {
      "epoch": 0.41617066104819594,
      "grad_norm": 2.1091370582580566,
      "learning_rate": 2.9204635238346063e-05,
      "loss": 1.6866,
      "step": 15802
    },
    {
      "epoch": 0.4161969976297077,
      "grad_norm": 1.6421785354614258,
      "learning_rate": 2.9203318409270482e-05,
      "loss": 1.813,
      "step": 15803
    },
    {
      "epoch": 0.4162233342112194,
      "grad_norm": 4.921753406524658,
      "learning_rate": 2.9202001580194894e-05,
      "loss": 1.7102,
      "step": 15804
    },
    {
      "epoch": 0.4162496707927311,
      "grad_norm": 2.5916521549224854,
      "learning_rate": 2.9200684751119306e-05,
      "loss": 1.1856,
      "step": 15805
    },
    {
      "epoch": 0.4162760073742428,
      "grad_norm": 2.8312504291534424,
      "learning_rate": 2.919936792204372e-05,
      "loss": 2.4651,
      "step": 15806
    },
    {
      "epoch": 0.41630234395575455,
      "grad_norm": 3.3246586322784424,
      "learning_rate": 2.919805109296813e-05,
      "loss": 1.7152,
      "step": 15807
    },
    {
      "epoch": 0.41632868053726624,
      "grad_norm": 1.5243314504623413,
      "learning_rate": 2.919673426389255e-05,
      "loss": 1.6843,
      "step": 15808
    },
    {
      "epoch": 0.416355017118778,
      "grad_norm": 4.375945091247559,
      "learning_rate": 2.9195417434816962e-05,
      "loss": 1.1469,
      "step": 15809
    },
    {
      "epoch": 0.4163813537002897,
      "grad_norm": 2.180367946624756,
      "learning_rate": 2.9194100605741374e-05,
      "loss": 2.4722,
      "step": 15810
    },
    {
      "epoch": 0.4164076902818014,
      "grad_norm": 1.7941190004348755,
      "learning_rate": 2.919278377666579e-05,
      "loss": 1.7475,
      "step": 15811
    },
    {
      "epoch": 0.41643402686331316,
      "grad_norm": 2.0264933109283447,
      "learning_rate": 2.9191466947590205e-05,
      "loss": 1.8101,
      "step": 15812
    },
    {
      "epoch": 0.41646036344482484,
      "grad_norm": 1.70437753200531,
      "learning_rate": 2.919015011851462e-05,
      "loss": 0.8151,
      "step": 15813
    },
    {
      "epoch": 0.4164867000263366,
      "grad_norm": 2.530991554260254,
      "learning_rate": 2.9188833289439033e-05,
      "loss": 1.4086,
      "step": 15814
    },
    {
      "epoch": 0.4165130366078483,
      "grad_norm": 1.8729264736175537,
      "learning_rate": 2.9187516460363445e-05,
      "loss": 1.7716,
      "step": 15815
    },
    {
      "epoch": 0.41653937318936,
      "grad_norm": 4.4749979972839355,
      "learning_rate": 2.9186199631287858e-05,
      "loss": 1.7745,
      "step": 15816
    },
    {
      "epoch": 0.41656570977087176,
      "grad_norm": 4.986862659454346,
      "learning_rate": 2.9184882802212277e-05,
      "loss": 2.198,
      "step": 15817
    },
    {
      "epoch": 0.41659204635238345,
      "grad_norm": 1.4708622694015503,
      "learning_rate": 2.918356597313669e-05,
      "loss": 1.823,
      "step": 15818
    },
    {
      "epoch": 0.4166183829338952,
      "grad_norm": 3.1022653579711914,
      "learning_rate": 2.91822491440611e-05,
      "loss": 0.8135,
      "step": 15819
    },
    {
      "epoch": 0.4166447195154069,
      "grad_norm": 1.5363943576812744,
      "learning_rate": 2.9180932314985517e-05,
      "loss": 2.1016,
      "step": 15820
    },
    {
      "epoch": 0.4166710560969186,
      "grad_norm": 1.6736706495285034,
      "learning_rate": 2.917961548590993e-05,
      "loss": 1.2078,
      "step": 15821
    },
    {
      "epoch": 0.41669739267843037,
      "grad_norm": 1.6720129251480103,
      "learning_rate": 2.9178298656834348e-05,
      "loss": 1.6384,
      "step": 15822
    },
    {
      "epoch": 0.41672372925994206,
      "grad_norm": 1.7238825559616089,
      "learning_rate": 2.917698182775876e-05,
      "loss": 1.902,
      "step": 15823
    },
    {
      "epoch": 0.4167500658414538,
      "grad_norm": 1.7199745178222656,
      "learning_rate": 2.9175664998683172e-05,
      "loss": 1.4587,
      "step": 15824
    },
    {
      "epoch": 0.4167764024229655,
      "grad_norm": 2.393589496612549,
      "learning_rate": 2.9174348169607584e-05,
      "loss": 1.1613,
      "step": 15825
    },
    {
      "epoch": 0.41680273900447723,
      "grad_norm": 3.2851781845092773,
      "learning_rate": 2.9173031340531997e-05,
      "loss": 1.8116,
      "step": 15826
    },
    {
      "epoch": 0.4168290755859889,
      "grad_norm": 3.0794565677642822,
      "learning_rate": 2.9171714511456416e-05,
      "loss": 2.1656,
      "step": 15827
    },
    {
      "epoch": 0.41685541216750066,
      "grad_norm": 3.5296437740325928,
      "learning_rate": 2.9170397682380828e-05,
      "loss": 1.5944,
      "step": 15828
    },
    {
      "epoch": 0.4168817487490124,
      "grad_norm": 1.4852328300476074,
      "learning_rate": 2.9169080853305243e-05,
      "loss": 1.9289,
      "step": 15829
    },
    {
      "epoch": 0.4169080853305241,
      "grad_norm": 3.157398223876953,
      "learning_rate": 2.9167764024229656e-05,
      "loss": 1.7469,
      "step": 15830
    },
    {
      "epoch": 0.41693442191203584,
      "grad_norm": 2.660172462463379,
      "learning_rate": 2.9166447195154075e-05,
      "loss": 0.5563,
      "step": 15831
    },
    {
      "epoch": 0.4169607584935475,
      "grad_norm": 1.7150225639343262,
      "learning_rate": 2.9165130366078487e-05,
      "loss": 2.2012,
      "step": 15832
    },
    {
      "epoch": 0.41698709507505927,
      "grad_norm": 2.8092753887176514,
      "learning_rate": 2.91638135370029e-05,
      "loss": 0.4589,
      "step": 15833
    },
    {
      "epoch": 0.41701343165657095,
      "grad_norm": 2.415296792984009,
      "learning_rate": 2.916249670792731e-05,
      "loss": 1.1623,
      "step": 15834
    },
    {
      "epoch": 0.4170397682380827,
      "grad_norm": 1.9134904146194458,
      "learning_rate": 2.9161179878851723e-05,
      "loss": 1.1477,
      "step": 15835
    },
    {
      "epoch": 0.41706610481959444,
      "grad_norm": 1.5199739933013916,
      "learning_rate": 2.9159863049776142e-05,
      "loss": 2.1967,
      "step": 15836
    },
    {
      "epoch": 0.41709244140110613,
      "grad_norm": 4.269620895385742,
      "learning_rate": 2.9158546220700555e-05,
      "loss": 0.9638,
      "step": 15837
    },
    {
      "epoch": 0.41711877798261787,
      "grad_norm": 5.702027320861816,
      "learning_rate": 2.9157229391624967e-05,
      "loss": 1.925,
      "step": 15838
    },
    {
      "epoch": 0.41714511456412956,
      "grad_norm": 2.487232208251953,
      "learning_rate": 2.9155912562549382e-05,
      "loss": 2.1036,
      "step": 15839
    },
    {
      "epoch": 0.4171714511456413,
      "grad_norm": 3.8568320274353027,
      "learning_rate": 2.9154595733473795e-05,
      "loss": 1.7937,
      "step": 15840
    },
    {
      "epoch": 0.417197787727153,
      "grad_norm": 1.774970531463623,
      "learning_rate": 2.9153278904398214e-05,
      "loss": 1.6955,
      "step": 15841
    },
    {
      "epoch": 0.41722412430866473,
      "grad_norm": 2.2013163566589355,
      "learning_rate": 2.9151962075322626e-05,
      "loss": 1.6738,
      "step": 15842
    },
    {
      "epoch": 0.4172504608901765,
      "grad_norm": 3.0009524822235107,
      "learning_rate": 2.9150645246247038e-05,
      "loss": 1.8865,
      "step": 15843
    },
    {
      "epoch": 0.41727679747168817,
      "grad_norm": 2.259378671646118,
      "learning_rate": 2.914932841717145e-05,
      "loss": 1.9571,
      "step": 15844
    },
    {
      "epoch": 0.4173031340531999,
      "grad_norm": 1.8683756589889526,
      "learning_rate": 2.914801158809587e-05,
      "loss": 1.7056,
      "step": 15845
    },
    {
      "epoch": 0.4173294706347116,
      "grad_norm": 2.313812017440796,
      "learning_rate": 2.914669475902028e-05,
      "loss": 2.0922,
      "step": 15846
    },
    {
      "epoch": 0.41735580721622334,
      "grad_norm": 6.496887683868408,
      "learning_rate": 2.9145377929944694e-05,
      "loss": 2.0727,
      "step": 15847
    },
    {
      "epoch": 0.41738214379773503,
      "grad_norm": 2.928194284439087,
      "learning_rate": 2.914406110086911e-05,
      "loss": 1.9944,
      "step": 15848
    },
    {
      "epoch": 0.41740848037924677,
      "grad_norm": 3.5459048748016357,
      "learning_rate": 2.914274427179352e-05,
      "loss": 1.1235,
      "step": 15849
    },
    {
      "epoch": 0.4174348169607585,
      "grad_norm": 3.09625506401062,
      "learning_rate": 2.914142744271794e-05,
      "loss": 1.624,
      "step": 15850
    },
    {
      "epoch": 0.4174611535422702,
      "grad_norm": 1.8869256973266602,
      "learning_rate": 2.9140110613642353e-05,
      "loss": 1.0171,
      "step": 15851
    },
    {
      "epoch": 0.41748749012378195,
      "grad_norm": 2.104810953140259,
      "learning_rate": 2.9138793784566765e-05,
      "loss": 2.0962,
      "step": 15852
    },
    {
      "epoch": 0.41751382670529363,
      "grad_norm": 2.244263172149658,
      "learning_rate": 2.9137476955491177e-05,
      "loss": 1.9973,
      "step": 15853
    },
    {
      "epoch": 0.4175401632868054,
      "grad_norm": 4.747417449951172,
      "learning_rate": 2.913616012641559e-05,
      "loss": 0.9527,
      "step": 15854
    },
    {
      "epoch": 0.41756649986831706,
      "grad_norm": 2.2802634239196777,
      "learning_rate": 2.9134843297340008e-05,
      "loss": 0.8045,
      "step": 15855
    },
    {
      "epoch": 0.4175928364498288,
      "grad_norm": 2.1140217781066895,
      "learning_rate": 2.913352646826442e-05,
      "loss": 2.0167,
      "step": 15856
    },
    {
      "epoch": 0.41761917303134055,
      "grad_norm": 1.5266413688659668,
      "learning_rate": 2.9132209639188833e-05,
      "loss": 1.2593,
      "step": 15857
    },
    {
      "epoch": 0.41764550961285224,
      "grad_norm": 4.241495609283447,
      "learning_rate": 2.913089281011325e-05,
      "loss": 2.0269,
      "step": 15858
    },
    {
      "epoch": 0.417671846194364,
      "grad_norm": 1.7115955352783203,
      "learning_rate": 2.912957598103766e-05,
      "loss": 0.4918,
      "step": 15859
    },
    {
      "epoch": 0.41769818277587567,
      "grad_norm": 1.7254637479782104,
      "learning_rate": 2.912825915196208e-05,
      "loss": 2.0225,
      "step": 15860
    },
    {
      "epoch": 0.4177245193573874,
      "grad_norm": 3.146047592163086,
      "learning_rate": 2.9126942322886492e-05,
      "loss": 1.7603,
      "step": 15861
    },
    {
      "epoch": 0.41775085593889916,
      "grad_norm": 1.4908065795898438,
      "learning_rate": 2.9125625493810904e-05,
      "loss": 1.612,
      "step": 15862
    },
    {
      "epoch": 0.41777719252041084,
      "grad_norm": 2.681586503982544,
      "learning_rate": 2.9124308664735316e-05,
      "loss": 1.4214,
      "step": 15863
    },
    {
      "epoch": 0.4178035291019226,
      "grad_norm": 16.2877197265625,
      "learning_rate": 2.9122991835659735e-05,
      "loss": 1.8115,
      "step": 15864
    },
    {
      "epoch": 0.4178298656834343,
      "grad_norm": 1.5722683668136597,
      "learning_rate": 2.9121675006584147e-05,
      "loss": 2.499,
      "step": 15865
    },
    {
      "epoch": 0.417856202264946,
      "grad_norm": 1.9022072553634644,
      "learning_rate": 2.912035817750856e-05,
      "loss": 1.5993,
      "step": 15866
    },
    {
      "epoch": 0.4178825388464577,
      "grad_norm": 1.8968604803085327,
      "learning_rate": 2.9119041348432975e-05,
      "loss": 2.1083,
      "step": 15867
    },
    {
      "epoch": 0.41790887542796945,
      "grad_norm": 1.8811109066009521,
      "learning_rate": 2.9117724519357387e-05,
      "loss": 2.2591,
      "step": 15868
    },
    {
      "epoch": 0.4179352120094812,
      "grad_norm": 2.29788875579834,
      "learning_rate": 2.9116407690281806e-05,
      "loss": 1.5296,
      "step": 15869
    },
    {
      "epoch": 0.4179615485909929,
      "grad_norm": 3.653179168701172,
      "learning_rate": 2.911509086120622e-05,
      "loss": 0.9831,
      "step": 15870
    },
    {
      "epoch": 0.4179878851725046,
      "grad_norm": 1.916192650794983,
      "learning_rate": 2.911377403213063e-05,
      "loss": 2.2055,
      "step": 15871
    },
    {
      "epoch": 0.4180142217540163,
      "grad_norm": 3.860635757446289,
      "learning_rate": 2.9112457203055043e-05,
      "loss": 0.6394,
      "step": 15872
    },
    {
      "epoch": 0.41804055833552806,
      "grad_norm": 1.938744306564331,
      "learning_rate": 2.9111140373979455e-05,
      "loss": 1.2032,
      "step": 15873
    },
    {
      "epoch": 0.41806689491703974,
      "grad_norm": 1.6979886293411255,
      "learning_rate": 2.9109823544903874e-05,
      "loss": 0.2738,
      "step": 15874
    },
    {
      "epoch": 0.4180932314985515,
      "grad_norm": 2.175947666168213,
      "learning_rate": 2.9108506715828286e-05,
      "loss": 1.4278,
      "step": 15875
    },
    {
      "epoch": 0.41811956808006323,
      "grad_norm": 5.7649054527282715,
      "learning_rate": 2.9107189886752702e-05,
      "loss": 0.9207,
      "step": 15876
    },
    {
      "epoch": 0.4181459046615749,
      "grad_norm": 3.200817823410034,
      "learning_rate": 2.9105873057677114e-05,
      "loss": 1.3369,
      "step": 15877
    },
    {
      "epoch": 0.41817224124308666,
      "grad_norm": 2.9281327724456787,
      "learning_rate": 2.9104556228601533e-05,
      "loss": 1.7935,
      "step": 15878
    },
    {
      "epoch": 0.41819857782459835,
      "grad_norm": 1.5523358583450317,
      "learning_rate": 2.9103239399525945e-05,
      "loss": 1.7339,
      "step": 15879
    },
    {
      "epoch": 0.4182249144061101,
      "grad_norm": 3.345917224884033,
      "learning_rate": 2.9101922570450358e-05,
      "loss": 1.3455,
      "step": 15880
    },
    {
      "epoch": 0.4182512509876218,
      "grad_norm": 1.7881011962890625,
      "learning_rate": 2.910060574137477e-05,
      "loss": 1.7529,
      "step": 15881
    },
    {
      "epoch": 0.4182775875691335,
      "grad_norm": 1.555479884147644,
      "learning_rate": 2.9099288912299182e-05,
      "loss": 1.7742,
      "step": 15882
    },
    {
      "epoch": 0.41830392415064527,
      "grad_norm": 2.2716665267944336,
      "learning_rate": 2.90979720832236e-05,
      "loss": 1.3526,
      "step": 15883
    },
    {
      "epoch": 0.41833026073215696,
      "grad_norm": 1.9988768100738525,
      "learning_rate": 2.9096655254148013e-05,
      "loss": 2.3314,
      "step": 15884
    },
    {
      "epoch": 0.4183565973136687,
      "grad_norm": 4.702402114868164,
      "learning_rate": 2.9095338425072425e-05,
      "loss": 1.5109,
      "step": 15885
    },
    {
      "epoch": 0.4183829338951804,
      "grad_norm": 1.7483413219451904,
      "learning_rate": 2.909402159599684e-05,
      "loss": 1.9926,
      "step": 15886
    },
    {
      "epoch": 0.41840927047669213,
      "grad_norm": 3.561209201812744,
      "learning_rate": 2.9092704766921253e-05,
      "loss": 1.2674,
      "step": 15887
    },
    {
      "epoch": 0.4184356070582038,
      "grad_norm": 2.1175496578216553,
      "learning_rate": 2.9091387937845672e-05,
      "loss": 1.7887,
      "step": 15888
    },
    {
      "epoch": 0.41846194363971556,
      "grad_norm": 1.6008661985397339,
      "learning_rate": 2.9090071108770084e-05,
      "loss": 1.8637,
      "step": 15889
    },
    {
      "epoch": 0.4184882802212273,
      "grad_norm": 1.8572592735290527,
      "learning_rate": 2.9088754279694497e-05,
      "loss": 1.3288,
      "step": 15890
    },
    {
      "epoch": 0.418514616802739,
      "grad_norm": 2.8444910049438477,
      "learning_rate": 2.908743745061891e-05,
      "loss": 0.7816,
      "step": 15891
    },
    {
      "epoch": 0.41854095338425074,
      "grad_norm": 2.1428418159484863,
      "learning_rate": 2.908612062154332e-05,
      "loss": 0.7519,
      "step": 15892
    },
    {
      "epoch": 0.4185672899657624,
      "grad_norm": 1.5301878452301025,
      "learning_rate": 2.908480379246774e-05,
      "loss": 2.3541,
      "step": 15893
    },
    {
      "epoch": 0.41859362654727417,
      "grad_norm": 2.8264079093933105,
      "learning_rate": 2.9083486963392152e-05,
      "loss": 2.1651,
      "step": 15894
    },
    {
      "epoch": 0.4186199631287859,
      "grad_norm": 1.492467999458313,
      "learning_rate": 2.9082170134316568e-05,
      "loss": 1.3263,
      "step": 15895
    },
    {
      "epoch": 0.4186462997102976,
      "grad_norm": 2.664212226867676,
      "learning_rate": 2.908085330524098e-05,
      "loss": 1.4047,
      "step": 15896
    },
    {
      "epoch": 0.41867263629180934,
      "grad_norm": 2.399510145187378,
      "learning_rate": 2.90795364761654e-05,
      "loss": 2.0924,
      "step": 15897
    },
    {
      "epoch": 0.41869897287332103,
      "grad_norm": 1.7267956733703613,
      "learning_rate": 2.907821964708981e-05,
      "loss": 1.6726,
      "step": 15898
    },
    {
      "epoch": 0.4187253094548328,
      "grad_norm": 1.7243093252182007,
      "learning_rate": 2.9076902818014223e-05,
      "loss": 1.0425,
      "step": 15899
    },
    {
      "epoch": 0.41875164603634446,
      "grad_norm": 1.4584072828292847,
      "learning_rate": 2.9075585988938636e-05,
      "loss": 1.5449,
      "step": 15900
    },
    {
      "epoch": 0.4187779826178562,
      "grad_norm": 3.6513116359710693,
      "learning_rate": 2.9074269159863048e-05,
      "loss": 1.6765,
      "step": 15901
    },
    {
      "epoch": 0.41880431919936795,
      "grad_norm": 5.887742042541504,
      "learning_rate": 2.9072952330787467e-05,
      "loss": 1.8525,
      "step": 15902
    },
    {
      "epoch": 0.41883065578087963,
      "grad_norm": 1.7512121200561523,
      "learning_rate": 2.907163550171188e-05,
      "loss": 1.2852,
      "step": 15903
    },
    {
      "epoch": 0.4188569923623914,
      "grad_norm": 4.285510540008545,
      "learning_rate": 2.9070318672636295e-05,
      "loss": 1.7264,
      "step": 15904
    },
    {
      "epoch": 0.41888332894390307,
      "grad_norm": 1.7619471549987793,
      "learning_rate": 2.9069001843560707e-05,
      "loss": 1.3431,
      "step": 15905
    },
    {
      "epoch": 0.4189096655254148,
      "grad_norm": 2.8740603923797607,
      "learning_rate": 2.906768501448512e-05,
      "loss": 1.9017,
      "step": 15906
    },
    {
      "epoch": 0.4189360021069265,
      "grad_norm": 3.6611809730529785,
      "learning_rate": 2.9066368185409538e-05,
      "loss": 5.5193,
      "step": 15907
    },
    {
      "epoch": 0.41896233868843824,
      "grad_norm": 2.0516936779022217,
      "learning_rate": 2.906505135633395e-05,
      "loss": 1.9473,
      "step": 15908
    },
    {
      "epoch": 0.41898867526995,
      "grad_norm": 5.081444263458252,
      "learning_rate": 2.9063734527258362e-05,
      "loss": 2.0338,
      "step": 15909
    },
    {
      "epoch": 0.41901501185146167,
      "grad_norm": 2.4108309745788574,
      "learning_rate": 2.9062417698182775e-05,
      "loss": 1.1406,
      "step": 15910
    },
    {
      "epoch": 0.4190413484329734,
      "grad_norm": 1.5972779989242554,
      "learning_rate": 2.9061100869107194e-05,
      "loss": 1.1847,
      "step": 15911
    },
    {
      "epoch": 0.4190676850144851,
      "grad_norm": 1.9686408042907715,
      "learning_rate": 2.9059784040031606e-05,
      "loss": 2.0174,
      "step": 15912
    },
    {
      "epoch": 0.41909402159599685,
      "grad_norm": 1.8970401287078857,
      "learning_rate": 2.9058467210956018e-05,
      "loss": 2.7467,
      "step": 15913
    },
    {
      "epoch": 0.41912035817750853,
      "grad_norm": 3.6564159393310547,
      "learning_rate": 2.9057150381880434e-05,
      "loss": 1.0873,
      "step": 15914
    },
    {
      "epoch": 0.4191466947590203,
      "grad_norm": 2.1340556144714355,
      "learning_rate": 2.9055833552804846e-05,
      "loss": 1.7619,
      "step": 15915
    },
    {
      "epoch": 0.419173031340532,
      "grad_norm": 1.7491053342819214,
      "learning_rate": 2.9054516723729265e-05,
      "loss": 1.8354,
      "step": 15916
    },
    {
      "epoch": 0.4191993679220437,
      "grad_norm": 2.103529453277588,
      "learning_rate": 2.9053199894653677e-05,
      "loss": 2.0311,
      "step": 15917
    },
    {
      "epoch": 0.41922570450355545,
      "grad_norm": 1.4501547813415527,
      "learning_rate": 2.905188306557809e-05,
      "loss": 0.7853,
      "step": 15918
    },
    {
      "epoch": 0.41925204108506714,
      "grad_norm": 1.9094527959823608,
      "learning_rate": 2.90505662365025e-05,
      "loss": 2.3446,
      "step": 15919
    },
    {
      "epoch": 0.4192783776665789,
      "grad_norm": 2.823381185531616,
      "learning_rate": 2.9049249407426914e-05,
      "loss": 1.0354,
      "step": 15920
    },
    {
      "epoch": 0.41930471424809057,
      "grad_norm": 1.5064691305160522,
      "learning_rate": 2.9047932578351333e-05,
      "loss": 1.7708,
      "step": 15921
    },
    {
      "epoch": 0.4193310508296023,
      "grad_norm": 2.477412700653076,
      "learning_rate": 2.9046615749275745e-05,
      "loss": 2.3426,
      "step": 15922
    },
    {
      "epoch": 0.41935738741111406,
      "grad_norm": 2.4033079147338867,
      "learning_rate": 2.904529892020016e-05,
      "loss": 1.602,
      "step": 15923
    },
    {
      "epoch": 0.41938372399262575,
      "grad_norm": 2.9287612438201904,
      "learning_rate": 2.9043982091124573e-05,
      "loss": 1.1466,
      "step": 15924
    },
    {
      "epoch": 0.4194100605741375,
      "grad_norm": 2.5729892253875732,
      "learning_rate": 2.9042665262048985e-05,
      "loss": 0.8843,
      "step": 15925
    },
    {
      "epoch": 0.4194363971556492,
      "grad_norm": 2.1236963272094727,
      "learning_rate": 2.9041348432973404e-05,
      "loss": 2.0105,
      "step": 15926
    },
    {
      "epoch": 0.4194627337371609,
      "grad_norm": 1.672542691230774,
      "learning_rate": 2.9040031603897816e-05,
      "loss": 1.7941,
      "step": 15927
    },
    {
      "epoch": 0.41948907031867266,
      "grad_norm": 2.8274424076080322,
      "learning_rate": 2.903871477482223e-05,
      "loss": 0.7584,
      "step": 15928
    },
    {
      "epoch": 0.41951540690018435,
      "grad_norm": 1.7160110473632812,
      "learning_rate": 2.903739794574664e-05,
      "loss": 1.8366,
      "step": 15929
    },
    {
      "epoch": 0.4195417434816961,
      "grad_norm": 1.9629780054092407,
      "learning_rate": 2.903608111667106e-05,
      "loss": 1.5624,
      "step": 15930
    },
    {
      "epoch": 0.4195680800632078,
      "grad_norm": 1.6561269760131836,
      "learning_rate": 2.9034764287595472e-05,
      "loss": 0.4414,
      "step": 15931
    },
    {
      "epoch": 0.4195944166447195,
      "grad_norm": 2.1763134002685547,
      "learning_rate": 2.9033447458519887e-05,
      "loss": 1.8046,
      "step": 15932
    },
    {
      "epoch": 0.4196207532262312,
      "grad_norm": 2.108588218688965,
      "learning_rate": 2.90321306294443e-05,
      "loss": 1.073,
      "step": 15933
    },
    {
      "epoch": 0.41964708980774296,
      "grad_norm": 1.8754918575286865,
      "learning_rate": 2.9030813800368712e-05,
      "loss": 1.8888,
      "step": 15934
    },
    {
      "epoch": 0.4196734263892547,
      "grad_norm": 3.2108142375946045,
      "learning_rate": 2.902949697129313e-05,
      "loss": 0.8351,
      "step": 15935
    },
    {
      "epoch": 0.4196997629707664,
      "grad_norm": 2.4658565521240234,
      "learning_rate": 2.9028180142217543e-05,
      "loss": 1.3003,
      "step": 15936
    },
    {
      "epoch": 0.41972609955227813,
      "grad_norm": 2.130405902862549,
      "learning_rate": 2.9026863313141955e-05,
      "loss": 1.7646,
      "step": 15937
    },
    {
      "epoch": 0.4197524361337898,
      "grad_norm": 1.5278515815734863,
      "learning_rate": 2.9025546484066367e-05,
      "loss": 1.7577,
      "step": 15938
    },
    {
      "epoch": 0.41977877271530156,
      "grad_norm": 1.652101755142212,
      "learning_rate": 2.902422965499078e-05,
      "loss": 2.0732,
      "step": 15939
    },
    {
      "epoch": 0.41980510929681325,
      "grad_norm": 3.1307928562164307,
      "learning_rate": 2.90229128259152e-05,
      "loss": 1.7772,
      "step": 15940
    },
    {
      "epoch": 0.419831445878325,
      "grad_norm": 2.3606410026550293,
      "learning_rate": 2.902159599683961e-05,
      "loss": 1.221,
      "step": 15941
    },
    {
      "epoch": 0.41985778245983674,
      "grad_norm": 1.6596097946166992,
      "learning_rate": 2.9020279167764026e-05,
      "loss": 2.1541,
      "step": 15942
    },
    {
      "epoch": 0.4198841190413484,
      "grad_norm": 2.633572816848755,
      "learning_rate": 2.901896233868844e-05,
      "loss": 1.5241,
      "step": 15943
    },
    {
      "epoch": 0.41991045562286017,
      "grad_norm": 2.3885717391967773,
      "learning_rate": 2.9017645509612858e-05,
      "loss": 0.4491,
      "step": 15944
    },
    {
      "epoch": 0.41993679220437186,
      "grad_norm": 2.233095645904541,
      "learning_rate": 2.901632868053727e-05,
      "loss": 0.9087,
      "step": 15945
    },
    {
      "epoch": 0.4199631287858836,
      "grad_norm": 1.7320566177368164,
      "learning_rate": 2.9015011851461682e-05,
      "loss": 1.9692,
      "step": 15946
    },
    {
      "epoch": 0.4199894653673953,
      "grad_norm": 2.7305448055267334,
      "learning_rate": 2.9013695022386094e-05,
      "loss": 1.0588,
      "step": 15947
    },
    {
      "epoch": 0.42001580194890703,
      "grad_norm": 1.9532499313354492,
      "learning_rate": 2.9012378193310506e-05,
      "loss": 1.2022,
      "step": 15948
    },
    {
      "epoch": 0.4200421385304188,
      "grad_norm": 2.070375919342041,
      "learning_rate": 2.9011061364234925e-05,
      "loss": 2.1529,
      "step": 15949
    },
    {
      "epoch": 0.42006847511193046,
      "grad_norm": 1.5699832439422607,
      "learning_rate": 2.9009744535159338e-05,
      "loss": 1.9472,
      "step": 15950
    },
    {
      "epoch": 0.4200948116934422,
      "grad_norm": 2.310711622238159,
      "learning_rate": 2.9008427706083753e-05,
      "loss": 1.9255,
      "step": 15951
    },
    {
      "epoch": 0.4201211482749539,
      "grad_norm": 2.3300163745880127,
      "learning_rate": 2.9007110877008165e-05,
      "loss": 1.9764,
      "step": 15952
    },
    {
      "epoch": 0.42014748485646564,
      "grad_norm": 1.8245055675506592,
      "learning_rate": 2.9005794047932578e-05,
      "loss": 1.8168,
      "step": 15953
    },
    {
      "epoch": 0.4201738214379773,
      "grad_norm": 3.797804117202759,
      "learning_rate": 2.9004477218856997e-05,
      "loss": 2.3182,
      "step": 15954
    },
    {
      "epoch": 0.42020015801948907,
      "grad_norm": 2.3537936210632324,
      "learning_rate": 2.900316038978141e-05,
      "loss": 2.1785,
      "step": 15955
    },
    {
      "epoch": 0.4202264946010008,
      "grad_norm": 2.155189275741577,
      "learning_rate": 2.900184356070582e-05,
      "loss": 2.2098,
      "step": 15956
    },
    {
      "epoch": 0.4202528311825125,
      "grad_norm": 1.8503748178482056,
      "learning_rate": 2.9000526731630233e-05,
      "loss": 1.9656,
      "step": 15957
    },
    {
      "epoch": 0.42027916776402424,
      "grad_norm": 2.7719359397888184,
      "learning_rate": 2.8999209902554645e-05,
      "loss": 2.2013,
      "step": 15958
    },
    {
      "epoch": 0.42030550434553593,
      "grad_norm": 2.182180404663086,
      "learning_rate": 2.8997893073479064e-05,
      "loss": 1.5532,
      "step": 15959
    },
    {
      "epoch": 0.4203318409270477,
      "grad_norm": 2.1619744300842285,
      "learning_rate": 2.8996576244403477e-05,
      "loss": 1.9065,
      "step": 15960
    },
    {
      "epoch": 0.4203581775085594,
      "grad_norm": 1.6861029863357544,
      "learning_rate": 2.8995259415327892e-05,
      "loss": 1.3705,
      "step": 15961
    },
    {
      "epoch": 0.4203845140900711,
      "grad_norm": 2.342731475830078,
      "learning_rate": 2.8993942586252304e-05,
      "loss": 2.328,
      "step": 15962
    },
    {
      "epoch": 0.42041085067158285,
      "grad_norm": 3.1964547634124756,
      "learning_rate": 2.8992625757176723e-05,
      "loss": 1.206,
      "step": 15963
    },
    {
      "epoch": 0.42043718725309454,
      "grad_norm": 1.734544277191162,
      "learning_rate": 2.8991308928101136e-05,
      "loss": 1.6301,
      "step": 15964
    },
    {
      "epoch": 0.4204635238346063,
      "grad_norm": 1.902092695236206,
      "learning_rate": 2.8989992099025548e-05,
      "loss": 1.8637,
      "step": 15965
    },
    {
      "epoch": 0.42048986041611797,
      "grad_norm": 2.298835515975952,
      "learning_rate": 2.898867526994996e-05,
      "loss": 1.769,
      "step": 15966
    },
    {
      "epoch": 0.4205161969976297,
      "grad_norm": 2.136662483215332,
      "learning_rate": 2.8987358440874372e-05,
      "loss": 1.8443,
      "step": 15967
    },
    {
      "epoch": 0.42054253357914145,
      "grad_norm": 3.5213842391967773,
      "learning_rate": 2.898604161179879e-05,
      "loss": 2.3779,
      "step": 15968
    },
    {
      "epoch": 0.42056887016065314,
      "grad_norm": 1.706467628479004,
      "learning_rate": 2.8984724782723203e-05,
      "loss": 1.424,
      "step": 15969
    },
    {
      "epoch": 0.4205952067421649,
      "grad_norm": 1.7673289775848389,
      "learning_rate": 2.898340795364762e-05,
      "loss": 1.9519,
      "step": 15970
    },
    {
      "epoch": 0.42062154332367657,
      "grad_norm": 4.114656448364258,
      "learning_rate": 2.898209112457203e-05,
      "loss": 0.9185,
      "step": 15971
    },
    {
      "epoch": 0.4206478799051883,
      "grad_norm": 1.8710373640060425,
      "learning_rate": 2.8980774295496444e-05,
      "loss": 1.8882,
      "step": 15972
    },
    {
      "epoch": 0.4206742164867,
      "grad_norm": 1.4987995624542236,
      "learning_rate": 2.8979457466420862e-05,
      "loss": 1.5713,
      "step": 15973
    },
    {
      "epoch": 0.42070055306821175,
      "grad_norm": 1.9544272422790527,
      "learning_rate": 2.8978140637345275e-05,
      "loss": 2.1226,
      "step": 15974
    },
    {
      "epoch": 0.4207268896497235,
      "grad_norm": 1.9978083372116089,
      "learning_rate": 2.8976823808269687e-05,
      "loss": 1.9512,
      "step": 15975
    },
    {
      "epoch": 0.4207532262312352,
      "grad_norm": 3.6790876388549805,
      "learning_rate": 2.89755069791941e-05,
      "loss": 1.3376,
      "step": 15976
    },
    {
      "epoch": 0.4207795628127469,
      "grad_norm": 1.8221933841705322,
      "learning_rate": 2.8974190150118518e-05,
      "loss": 1.1866,
      "step": 15977
    },
    {
      "epoch": 0.4208058993942586,
      "grad_norm": 1.7291162014007568,
      "learning_rate": 2.897287332104293e-05,
      "loss": 2.202,
      "step": 15978
    },
    {
      "epoch": 0.42083223597577035,
      "grad_norm": 2.848832845687866,
      "learning_rate": 2.8971556491967346e-05,
      "loss": 1.5461,
      "step": 15979
    },
    {
      "epoch": 0.42085857255728204,
      "grad_norm": 2.547539472579956,
      "learning_rate": 2.8970239662891758e-05,
      "loss": 1.8744,
      "step": 15980
    },
    {
      "epoch": 0.4208849091387938,
      "grad_norm": 2.420252799987793,
      "learning_rate": 2.896892283381617e-05,
      "loss": 2.2973,
      "step": 15981
    },
    {
      "epoch": 0.4209112457203055,
      "grad_norm": 2.7987349033355713,
      "learning_rate": 2.896760600474059e-05,
      "loss": 1.9818,
      "step": 15982
    },
    {
      "epoch": 0.4209375823018172,
      "grad_norm": 4.058008670806885,
      "learning_rate": 2.8966289175665e-05,
      "loss": 1.3098,
      "step": 15983
    },
    {
      "epoch": 0.42096391888332896,
      "grad_norm": 1.5876502990722656,
      "learning_rate": 2.8964972346589414e-05,
      "loss": 1.5784,
      "step": 15984
    },
    {
      "epoch": 0.42099025546484065,
      "grad_norm": 2.0043179988861084,
      "learning_rate": 2.8963655517513826e-05,
      "loss": 1.8581,
      "step": 15985
    },
    {
      "epoch": 0.4210165920463524,
      "grad_norm": 3.14562726020813,
      "learning_rate": 2.8962338688438238e-05,
      "loss": 1.9681,
      "step": 15986
    },
    {
      "epoch": 0.4210429286278641,
      "grad_norm": 1.9450507164001465,
      "learning_rate": 2.8961021859362657e-05,
      "loss": 1.1009,
      "step": 15987
    },
    {
      "epoch": 0.4210692652093758,
      "grad_norm": 2.42622709274292,
      "learning_rate": 2.895970503028707e-05,
      "loss": 1.4945,
      "step": 15988
    },
    {
      "epoch": 0.42109560179088756,
      "grad_norm": 1.9436426162719727,
      "learning_rate": 2.8958388201211485e-05,
      "loss": 1.044,
      "step": 15989
    },
    {
      "epoch": 0.42112193837239925,
      "grad_norm": 1.7001558542251587,
      "learning_rate": 2.8957071372135897e-05,
      "loss": 1.9212,
      "step": 15990
    },
    {
      "epoch": 0.421148274953911,
      "grad_norm": 3.6431925296783447,
      "learning_rate": 2.895575454306031e-05,
      "loss": 1.3559,
      "step": 15991
    },
    {
      "epoch": 0.4211746115354227,
      "grad_norm": 1.6292749643325806,
      "learning_rate": 2.895443771398473e-05,
      "loss": 1.3776,
      "step": 15992
    },
    {
      "epoch": 0.4212009481169344,
      "grad_norm": 4.4837965965271,
      "learning_rate": 2.895312088490914e-05,
      "loss": 2.5701,
      "step": 15993
    },
    {
      "epoch": 0.42122728469844617,
      "grad_norm": 1.6011847257614136,
      "learning_rate": 2.8951804055833553e-05,
      "loss": 1.9054,
      "step": 15994
    },
    {
      "epoch": 0.42125362127995786,
      "grad_norm": 2.1386990547180176,
      "learning_rate": 2.8950487226757965e-05,
      "loss": 1.4471,
      "step": 15995
    },
    {
      "epoch": 0.4212799578614696,
      "grad_norm": 2.749276638031006,
      "learning_rate": 2.8949170397682384e-05,
      "loss": 1.9002,
      "step": 15996
    },
    {
      "epoch": 0.4213062944429813,
      "grad_norm": 3.0917043685913086,
      "learning_rate": 2.8947853568606796e-05,
      "loss": 1.9788,
      "step": 15997
    },
    {
      "epoch": 0.42133263102449303,
      "grad_norm": 2.294778347015381,
      "learning_rate": 2.8946536739531212e-05,
      "loss": 0.8067,
      "step": 15998
    },
    {
      "epoch": 0.4213589676060047,
      "grad_norm": 1.9548379182815552,
      "learning_rate": 2.8945219910455624e-05,
      "loss": 1.6664,
      "step": 15999
    },
    {
      "epoch": 0.42138530418751646,
      "grad_norm": 1.8013393878936768,
      "learning_rate": 2.8943903081380036e-05,
      "loss": 2.5083,
      "step": 16000
    },
    {
      "epoch": 0.4214116407690282,
      "grad_norm": 4.092260360717773,
      "learning_rate": 2.8942586252304455e-05,
      "loss": 0.9086,
      "step": 16001
    },
    {
      "epoch": 0.4214379773505399,
      "grad_norm": 1.7577232122421265,
      "learning_rate": 2.8941269423228867e-05,
      "loss": 1.8162,
      "step": 16002
    },
    {
      "epoch": 0.42146431393205164,
      "grad_norm": 1.989029884338379,
      "learning_rate": 2.893995259415328e-05,
      "loss": 1.9404,
      "step": 16003
    },
    {
      "epoch": 0.4214906505135633,
      "grad_norm": 3.9387192726135254,
      "learning_rate": 2.8938635765077692e-05,
      "loss": 1.6626,
      "step": 16004
    },
    {
      "epoch": 0.42151698709507507,
      "grad_norm": 1.6529161930084229,
      "learning_rate": 2.8937318936002107e-05,
      "loss": 1.1396,
      "step": 16005
    },
    {
      "epoch": 0.42154332367658676,
      "grad_norm": 3.0110344886779785,
      "learning_rate": 2.8936002106926523e-05,
      "loss": 1.5878,
      "step": 16006
    },
    {
      "epoch": 0.4215696602580985,
      "grad_norm": 4.271462440490723,
      "learning_rate": 2.893468527785094e-05,
      "loss": 0.7398,
      "step": 16007
    },
    {
      "epoch": 0.42159599683961024,
      "grad_norm": 3.1086785793304443,
      "learning_rate": 2.893336844877535e-05,
      "loss": 1.9452,
      "step": 16008
    },
    {
      "epoch": 0.42162233342112193,
      "grad_norm": 1.503722071647644,
      "learning_rate": 2.8932051619699763e-05,
      "loss": 1.4847,
      "step": 16009
    },
    {
      "epoch": 0.4216486700026337,
      "grad_norm": 1.560887098312378,
      "learning_rate": 2.8930734790624182e-05,
      "loss": 1.6213,
      "step": 16010
    },
    {
      "epoch": 0.42167500658414536,
      "grad_norm": 1.760580062866211,
      "learning_rate": 2.8929417961548594e-05,
      "loss": 1.5621,
      "step": 16011
    },
    {
      "epoch": 0.4217013431656571,
      "grad_norm": 1.9348901510238647,
      "learning_rate": 2.8928101132473006e-05,
      "loss": 2.2423,
      "step": 16012
    },
    {
      "epoch": 0.4217276797471688,
      "grad_norm": 1.8959280252456665,
      "learning_rate": 2.892678430339742e-05,
      "loss": 2.3895,
      "step": 16013
    },
    {
      "epoch": 0.42175401632868054,
      "grad_norm": 2.430936813354492,
      "learning_rate": 2.892546747432183e-05,
      "loss": 1.2636,
      "step": 16014
    },
    {
      "epoch": 0.4217803529101923,
      "grad_norm": 1.6049647331237793,
      "learning_rate": 2.892415064524625e-05,
      "loss": 2.0687,
      "step": 16015
    },
    {
      "epoch": 0.42180668949170397,
      "grad_norm": 2.852264165878296,
      "learning_rate": 2.8922833816170662e-05,
      "loss": 0.3198,
      "step": 16016
    },
    {
      "epoch": 0.4218330260732157,
      "grad_norm": 1.8353729248046875,
      "learning_rate": 2.8921516987095078e-05,
      "loss": 1.6445,
      "step": 16017
    },
    {
      "epoch": 0.4218593626547274,
      "grad_norm": 2.0958211421966553,
      "learning_rate": 2.892020015801949e-05,
      "loss": 1.9426,
      "step": 16018
    },
    {
      "epoch": 0.42188569923623914,
      "grad_norm": 3.931833028793335,
      "learning_rate": 2.8918883328943902e-05,
      "loss": 2.7907,
      "step": 16019
    },
    {
      "epoch": 0.42191203581775083,
      "grad_norm": 3.747877836227417,
      "learning_rate": 2.891756649986832e-05,
      "loss": 1.6726,
      "step": 16020
    },
    {
      "epoch": 0.4219383723992626,
      "grad_norm": 3.309068202972412,
      "learning_rate": 2.8916249670792733e-05,
      "loss": 1.791,
      "step": 16021
    },
    {
      "epoch": 0.4219647089807743,
      "grad_norm": 3.228571653366089,
      "learning_rate": 2.8914932841717145e-05,
      "loss": 1.802,
      "step": 16022
    },
    {
      "epoch": 0.421991045562286,
      "grad_norm": 1.9346814155578613,
      "learning_rate": 2.8913616012641558e-05,
      "loss": 1.787,
      "step": 16023
    },
    {
      "epoch": 0.42201738214379775,
      "grad_norm": 1.7779711484909058,
      "learning_rate": 2.8912299183565973e-05,
      "loss": 1.111,
      "step": 16024
    },
    {
      "epoch": 0.42204371872530944,
      "grad_norm": 1.7615411281585693,
      "learning_rate": 2.891098235449039e-05,
      "loss": 1.4725,
      "step": 16025
    },
    {
      "epoch": 0.4220700553068212,
      "grad_norm": 2.5310657024383545,
      "learning_rate": 2.8909665525414804e-05,
      "loss": 1.7039,
      "step": 16026
    },
    {
      "epoch": 0.42209639188833287,
      "grad_norm": 1.7241129875183105,
      "learning_rate": 2.8908348696339217e-05,
      "loss": 1.5877,
      "step": 16027
    },
    {
      "epoch": 0.4221227284698446,
      "grad_norm": 1.4583377838134766,
      "learning_rate": 2.890703186726363e-05,
      "loss": 1.7874,
      "step": 16028
    },
    {
      "epoch": 0.42214906505135635,
      "grad_norm": 3.046745777130127,
      "learning_rate": 2.8905715038188048e-05,
      "loss": 0.8007,
      "step": 16029
    },
    {
      "epoch": 0.42217540163286804,
      "grad_norm": 1.6729148626327515,
      "learning_rate": 2.890439820911246e-05,
      "loss": 1.5125,
      "step": 16030
    },
    {
      "epoch": 0.4222017382143798,
      "grad_norm": 1.5012695789337158,
      "learning_rate": 2.8903081380036872e-05,
      "loss": 1.5603,
      "step": 16031
    },
    {
      "epoch": 0.4222280747958915,
      "grad_norm": 3.8798723220825195,
      "learning_rate": 2.8901764550961284e-05,
      "loss": 1.0792,
      "step": 16032
    },
    {
      "epoch": 0.4222544113774032,
      "grad_norm": 2.6700475215911865,
      "learning_rate": 2.8900447721885697e-05,
      "loss": 1.1422,
      "step": 16033
    },
    {
      "epoch": 0.42228074795891496,
      "grad_norm": 2.944796562194824,
      "learning_rate": 2.8899130892810116e-05,
      "loss": 1.8004,
      "step": 16034
    },
    {
      "epoch": 0.42230708454042665,
      "grad_norm": 1.6480005979537964,
      "learning_rate": 2.8897814063734528e-05,
      "loss": 1.8199,
      "step": 16035
    },
    {
      "epoch": 0.4223334211219384,
      "grad_norm": 2.051287889480591,
      "learning_rate": 2.8896497234658944e-05,
      "loss": 1.6361,
      "step": 16036
    },
    {
      "epoch": 0.4223597577034501,
      "grad_norm": 2.1824262142181396,
      "learning_rate": 2.8895180405583356e-05,
      "loss": 2.0993,
      "step": 16037
    },
    {
      "epoch": 0.4223860942849618,
      "grad_norm": 2.374279499053955,
      "learning_rate": 2.8893863576507768e-05,
      "loss": 2.4941,
      "step": 16038
    },
    {
      "epoch": 0.4224124308664735,
      "grad_norm": 1.8703571557998657,
      "learning_rate": 2.8892546747432187e-05,
      "loss": 1.6973,
      "step": 16039
    },
    {
      "epoch": 0.42243876744798525,
      "grad_norm": 3.0656161308288574,
      "learning_rate": 2.88912299183566e-05,
      "loss": 0.9282,
      "step": 16040
    },
    {
      "epoch": 0.422465104029497,
      "grad_norm": 2.0200493335723877,
      "learning_rate": 2.888991308928101e-05,
      "loss": 1.5263,
      "step": 16041
    },
    {
      "epoch": 0.4224914406110087,
      "grad_norm": 2.0137667655944824,
      "learning_rate": 2.8888596260205424e-05,
      "loss": 1.8077,
      "step": 16042
    },
    {
      "epoch": 0.4225177771925204,
      "grad_norm": 1.68495774269104,
      "learning_rate": 2.8887279431129843e-05,
      "loss": 1.6741,
      "step": 16043
    },
    {
      "epoch": 0.4225441137740321,
      "grad_norm": 2.9733004570007324,
      "learning_rate": 2.8885962602054255e-05,
      "loss": 1.4297,
      "step": 16044
    },
    {
      "epoch": 0.42257045035554386,
      "grad_norm": 1.8878774642944336,
      "learning_rate": 2.888464577297867e-05,
      "loss": 0.8646,
      "step": 16045
    },
    {
      "epoch": 0.42259678693705555,
      "grad_norm": 1.8320461511611938,
      "learning_rate": 2.8883328943903083e-05,
      "loss": 2.0844,
      "step": 16046
    },
    {
      "epoch": 0.4226231235185673,
      "grad_norm": 2.896723985671997,
      "learning_rate": 2.8882012114827495e-05,
      "loss": 1.6323,
      "step": 16047
    },
    {
      "epoch": 0.42264946010007903,
      "grad_norm": 2.671372413635254,
      "learning_rate": 2.8880695285751914e-05,
      "loss": 2.2406,
      "step": 16048
    },
    {
      "epoch": 0.4226757966815907,
      "grad_norm": 2.8025848865509033,
      "learning_rate": 2.8879378456676326e-05,
      "loss": 1.0264,
      "step": 16049
    },
    {
      "epoch": 0.42270213326310246,
      "grad_norm": 1.8795288801193237,
      "learning_rate": 2.8878061627600738e-05,
      "loss": 1.9231,
      "step": 16050
    },
    {
      "epoch": 0.42272846984461415,
      "grad_norm": 2.6844730377197266,
      "learning_rate": 2.887674479852515e-05,
      "loss": 1.764,
      "step": 16051
    },
    {
      "epoch": 0.4227548064261259,
      "grad_norm": 2.2637476921081543,
      "learning_rate": 2.8875427969449566e-05,
      "loss": 2.0575,
      "step": 16052
    },
    {
      "epoch": 0.4227811430076376,
      "grad_norm": 1.7152243852615356,
      "learning_rate": 2.887411114037398e-05,
      "loss": 1.7721,
      "step": 16053
    },
    {
      "epoch": 0.4228074795891493,
      "grad_norm": 1.7404292821884155,
      "learning_rate": 2.8872794311298397e-05,
      "loss": 1.5301,
      "step": 16054
    },
    {
      "epoch": 0.42283381617066107,
      "grad_norm": 2.5333354473114014,
      "learning_rate": 2.887147748222281e-05,
      "loss": 2.0883,
      "step": 16055
    },
    {
      "epoch": 0.42286015275217276,
      "grad_norm": 1.995476484298706,
      "learning_rate": 2.887016065314722e-05,
      "loss": 0.4145,
      "step": 16056
    },
    {
      "epoch": 0.4228864893336845,
      "grad_norm": 2.1039559841156006,
      "learning_rate": 2.8868843824071634e-05,
      "loss": 1.8087,
      "step": 16057
    },
    {
      "epoch": 0.4229128259151962,
      "grad_norm": 2.312376022338867,
      "learning_rate": 2.8867526994996053e-05,
      "loss": 1.724,
      "step": 16058
    },
    {
      "epoch": 0.42293916249670793,
      "grad_norm": 3.3007993698120117,
      "learning_rate": 2.8866210165920465e-05,
      "loss": 1.0639,
      "step": 16059
    },
    {
      "epoch": 0.4229654990782196,
      "grad_norm": 2.103898763656616,
      "learning_rate": 2.8864893336844877e-05,
      "loss": 2.3941,
      "step": 16060
    },
    {
      "epoch": 0.42299183565973136,
      "grad_norm": 2.1146938800811768,
      "learning_rate": 2.886357650776929e-05,
      "loss": 2.715,
      "step": 16061
    },
    {
      "epoch": 0.4230181722412431,
      "grad_norm": 2.1182732582092285,
      "learning_rate": 2.886225967869371e-05,
      "loss": 2.145,
      "step": 16062
    },
    {
      "epoch": 0.4230445088227548,
      "grad_norm": 3.7761735916137695,
      "learning_rate": 2.886094284961812e-05,
      "loss": 2.6465,
      "step": 16063
    },
    {
      "epoch": 0.42307084540426654,
      "grad_norm": 8.249588012695312,
      "learning_rate": 2.8859626020542536e-05,
      "loss": 1.8051,
      "step": 16064
    },
    {
      "epoch": 0.4230971819857782,
      "grad_norm": 1.4332759380340576,
      "learning_rate": 2.885830919146695e-05,
      "loss": 1.9378,
      "step": 16065
    },
    {
      "epoch": 0.42312351856728997,
      "grad_norm": 2.812303304672241,
      "learning_rate": 2.885699236239136e-05,
      "loss": 1.3506,
      "step": 16066
    },
    {
      "epoch": 0.4231498551488017,
      "grad_norm": 2.758058547973633,
      "learning_rate": 2.885567553331578e-05,
      "loss": 1.8187,
      "step": 16067
    },
    {
      "epoch": 0.4231761917303134,
      "grad_norm": 1.809330701828003,
      "learning_rate": 2.8854358704240192e-05,
      "loss": 1.8237,
      "step": 16068
    },
    {
      "epoch": 0.42320252831182514,
      "grad_norm": 3.5669260025024414,
      "learning_rate": 2.8853041875164604e-05,
      "loss": 1.7585,
      "step": 16069
    },
    {
      "epoch": 0.42322886489333683,
      "grad_norm": 1.5892715454101562,
      "learning_rate": 2.8851725046089016e-05,
      "loss": 1.8642,
      "step": 16070
    },
    {
      "epoch": 0.4232552014748486,
      "grad_norm": 2.5576107501983643,
      "learning_rate": 2.8850408217013432e-05,
      "loss": 1.5692,
      "step": 16071
    },
    {
      "epoch": 0.42328153805636026,
      "grad_norm": 2.669217824935913,
      "learning_rate": 2.8849091387937847e-05,
      "loss": 1.5435,
      "step": 16072
    },
    {
      "epoch": 0.423307874637872,
      "grad_norm": 1.5828750133514404,
      "learning_rate": 2.8847774558862263e-05,
      "loss": 2.1899,
      "step": 16073
    },
    {
      "epoch": 0.42333421121938375,
      "grad_norm": 2.548888921737671,
      "learning_rate": 2.8846457729786675e-05,
      "loss": 1.6104,
      "step": 16074
    },
    {
      "epoch": 0.42336054780089544,
      "grad_norm": 4.546311855316162,
      "learning_rate": 2.8845140900711087e-05,
      "loss": 1.282,
      "step": 16075
    },
    {
      "epoch": 0.4233868843824072,
      "grad_norm": 1.860329270362854,
      "learning_rate": 2.8843824071635506e-05,
      "loss": 1.7186,
      "step": 16076
    },
    {
      "epoch": 0.42341322096391887,
      "grad_norm": 1.7091431617736816,
      "learning_rate": 2.884250724255992e-05,
      "loss": 1.9196,
      "step": 16077
    },
    {
      "epoch": 0.4234395575454306,
      "grad_norm": 2.13338565826416,
      "learning_rate": 2.884119041348433e-05,
      "loss": 1.9921,
      "step": 16078
    },
    {
      "epoch": 0.4234658941269423,
      "grad_norm": 1.7698370218276978,
      "learning_rate": 2.8839873584408743e-05,
      "loss": 1.771,
      "step": 16079
    },
    {
      "epoch": 0.42349223070845404,
      "grad_norm": 2.5107030868530273,
      "learning_rate": 2.883855675533316e-05,
      "loss": 0.5269,
      "step": 16080
    },
    {
      "epoch": 0.4235185672899658,
      "grad_norm": 1.9949605464935303,
      "learning_rate": 2.8837239926257574e-05,
      "loss": 1.9259,
      "step": 16081
    },
    {
      "epoch": 0.4235449038714775,
      "grad_norm": 4.140947341918945,
      "learning_rate": 2.883592309718199e-05,
      "loss": 0.6839,
      "step": 16082
    },
    {
      "epoch": 0.4235712404529892,
      "grad_norm": 3.2024660110473633,
      "learning_rate": 2.8834606268106402e-05,
      "loss": 2.2879,
      "step": 16083
    },
    {
      "epoch": 0.4235975770345009,
      "grad_norm": 1.5304476022720337,
      "learning_rate": 2.8833289439030814e-05,
      "loss": 1.6538,
      "step": 16084
    },
    {
      "epoch": 0.42362391361601265,
      "grad_norm": 1.7870818376541138,
      "learning_rate": 2.8831972609955226e-05,
      "loss": 1.9474,
      "step": 16085
    },
    {
      "epoch": 0.42365025019752434,
      "grad_norm": 2.1189584732055664,
      "learning_rate": 2.8830655780879645e-05,
      "loss": 1.1247,
      "step": 16086
    },
    {
      "epoch": 0.4236765867790361,
      "grad_norm": 1.9312340021133423,
      "learning_rate": 2.8829338951804058e-05,
      "loss": 1.953,
      "step": 16087
    },
    {
      "epoch": 0.4237029233605478,
      "grad_norm": 2.249635934829712,
      "learning_rate": 2.882802212272847e-05,
      "loss": 1.5643,
      "step": 16088
    },
    {
      "epoch": 0.4237292599420595,
      "grad_norm": 1.6765087842941284,
      "learning_rate": 2.8826705293652882e-05,
      "loss": 2.3425,
      "step": 16089
    },
    {
      "epoch": 0.42375559652357125,
      "grad_norm": 2.5360443592071533,
      "learning_rate": 2.88253884645773e-05,
      "loss": 1.6056,
      "step": 16090
    },
    {
      "epoch": 0.42378193310508294,
      "grad_norm": 3.798560619354248,
      "learning_rate": 2.8824071635501713e-05,
      "loss": 0.9149,
      "step": 16091
    },
    {
      "epoch": 0.4238082696865947,
      "grad_norm": 6.564652442932129,
      "learning_rate": 2.882275480642613e-05,
      "loss": 1.5192,
      "step": 16092
    },
    {
      "epoch": 0.4238346062681064,
      "grad_norm": 2.571826696395874,
      "learning_rate": 2.882143797735054e-05,
      "loss": 1.8611,
      "step": 16093
    },
    {
      "epoch": 0.4238609428496181,
      "grad_norm": 3.632727861404419,
      "learning_rate": 2.8820121148274953e-05,
      "loss": 0.7822,
      "step": 16094
    },
    {
      "epoch": 0.42388727943112986,
      "grad_norm": 2.8242435455322266,
      "learning_rate": 2.8818804319199372e-05,
      "loss": 1.3055,
      "step": 16095
    },
    {
      "epoch": 0.42391361601264155,
      "grad_norm": 2.136639356613159,
      "learning_rate": 2.8817487490123784e-05,
      "loss": 1.6273,
      "step": 16096
    },
    {
      "epoch": 0.4239399525941533,
      "grad_norm": 2.631714105606079,
      "learning_rate": 2.8816170661048197e-05,
      "loss": 2.1361,
      "step": 16097
    },
    {
      "epoch": 0.423966289175665,
      "grad_norm": 2.5165181159973145,
      "learning_rate": 2.881485383197261e-05,
      "loss": 1.7164,
      "step": 16098
    },
    {
      "epoch": 0.4239926257571767,
      "grad_norm": 3.3852922916412354,
      "learning_rate": 2.8813537002897025e-05,
      "loss": 0.5745,
      "step": 16099
    },
    {
      "epoch": 0.42401896233868847,
      "grad_norm": 2.246051073074341,
      "learning_rate": 2.881222017382144e-05,
      "loss": 1.9521,
      "step": 16100
    },
    {
      "epoch": 0.42404529892020015,
      "grad_norm": 2.8720548152923584,
      "learning_rate": 2.8810903344745856e-05,
      "loss": 1.7878,
      "step": 16101
    },
    {
      "epoch": 0.4240716355017119,
      "grad_norm": 2.173828363418579,
      "learning_rate": 2.8809586515670268e-05,
      "loss": 1.4198,
      "step": 16102
    },
    {
      "epoch": 0.4240979720832236,
      "grad_norm": 3.1550397872924805,
      "learning_rate": 2.880826968659468e-05,
      "loss": 2.065,
      "step": 16103
    },
    {
      "epoch": 0.42412430866473533,
      "grad_norm": 3.2687010765075684,
      "learning_rate": 2.8806952857519092e-05,
      "loss": 1.5574,
      "step": 16104
    },
    {
      "epoch": 0.424150645246247,
      "grad_norm": 2.112921953201294,
      "learning_rate": 2.880563602844351e-05,
      "loss": 1.754,
      "step": 16105
    },
    {
      "epoch": 0.42417698182775876,
      "grad_norm": 1.4761923551559448,
      "learning_rate": 2.8804319199367924e-05,
      "loss": 1.6966,
      "step": 16106
    },
    {
      "epoch": 0.4242033184092705,
      "grad_norm": 2.3852739334106445,
      "learning_rate": 2.8803002370292336e-05,
      "loss": 1.8944,
      "step": 16107
    },
    {
      "epoch": 0.4242296549907822,
      "grad_norm": 3.7875638008117676,
      "learning_rate": 2.880168554121675e-05,
      "loss": 1.1141,
      "step": 16108
    },
    {
      "epoch": 0.42425599157229393,
      "grad_norm": 3.1541240215301514,
      "learning_rate": 2.8800368712141167e-05,
      "loss": 0.2556,
      "step": 16109
    },
    {
      "epoch": 0.4242823281538056,
      "grad_norm": 2.062598705291748,
      "learning_rate": 2.8799051883065583e-05,
      "loss": 2.0313,
      "step": 16110
    },
    {
      "epoch": 0.42430866473531736,
      "grad_norm": 1.8286455869674683,
      "learning_rate": 2.8797735053989995e-05,
      "loss": 2.028,
      "step": 16111
    },
    {
      "epoch": 0.42433500131682905,
      "grad_norm": 2.5597963333129883,
      "learning_rate": 2.8796418224914407e-05,
      "loss": 1.7595,
      "step": 16112
    },
    {
      "epoch": 0.4243613378983408,
      "grad_norm": 6.226208209991455,
      "learning_rate": 2.879510139583882e-05,
      "loss": 1.8319,
      "step": 16113
    },
    {
      "epoch": 0.42438767447985254,
      "grad_norm": 1.7025924921035767,
      "learning_rate": 2.8793784566763238e-05,
      "loss": 1.6761,
      "step": 16114
    },
    {
      "epoch": 0.4244140110613642,
      "grad_norm": 2.502087354660034,
      "learning_rate": 2.879246773768765e-05,
      "loss": 1.4152,
      "step": 16115
    },
    {
      "epoch": 0.42444034764287597,
      "grad_norm": 2.1164026260375977,
      "learning_rate": 2.8791150908612063e-05,
      "loss": 2.0156,
      "step": 16116
    },
    {
      "epoch": 0.42446668422438766,
      "grad_norm": 4.253668308258057,
      "learning_rate": 2.8789834079536475e-05,
      "loss": 1.4962,
      "step": 16117
    },
    {
      "epoch": 0.4244930208058994,
      "grad_norm": 1.780917763710022,
      "learning_rate": 2.878851725046089e-05,
      "loss": 1.6653,
      "step": 16118
    },
    {
      "epoch": 0.4245193573874111,
      "grad_norm": 1.9676209688186646,
      "learning_rate": 2.8787200421385306e-05,
      "loss": 1.3094,
      "step": 16119
    },
    {
      "epoch": 0.42454569396892283,
      "grad_norm": 1.6165271997451782,
      "learning_rate": 2.878588359230972e-05,
      "loss": 1.7085,
      "step": 16120
    },
    {
      "epoch": 0.4245720305504346,
      "grad_norm": 1.3759636878967285,
      "learning_rate": 2.8784566763234134e-05,
      "loss": 1.5248,
      "step": 16121
    },
    {
      "epoch": 0.42459836713194626,
      "grad_norm": 2.0079309940338135,
      "learning_rate": 2.8783249934158546e-05,
      "loss": 1.3013,
      "step": 16122
    },
    {
      "epoch": 0.424624703713458,
      "grad_norm": 2.1813623905181885,
      "learning_rate": 2.8781933105082965e-05,
      "loss": 1.5161,
      "step": 16123
    },
    {
      "epoch": 0.4246510402949697,
      "grad_norm": 2.741903066635132,
      "learning_rate": 2.8780616276007377e-05,
      "loss": 1.6275,
      "step": 16124
    },
    {
      "epoch": 0.42467737687648144,
      "grad_norm": 2.2224152088165283,
      "learning_rate": 2.877929944693179e-05,
      "loss": 0.9563,
      "step": 16125
    },
    {
      "epoch": 0.4247037134579931,
      "grad_norm": 1.5816574096679688,
      "learning_rate": 2.87779826178562e-05,
      "loss": 1.986,
      "step": 16126
    },
    {
      "epoch": 0.42473005003950487,
      "grad_norm": 2.1497292518615723,
      "learning_rate": 2.8776665788780617e-05,
      "loss": 2.0907,
      "step": 16127
    },
    {
      "epoch": 0.4247563866210166,
      "grad_norm": 3.007984161376953,
      "learning_rate": 2.8775348959705033e-05,
      "loss": 0.763,
      "step": 16128
    },
    {
      "epoch": 0.4247827232025283,
      "grad_norm": 1.6721152067184448,
      "learning_rate": 2.877403213062945e-05,
      "loss": 1.8409,
      "step": 16129
    },
    {
      "epoch": 0.42480905978404004,
      "grad_norm": 2.0486481189727783,
      "learning_rate": 2.877271530155386e-05,
      "loss": 1.5911,
      "step": 16130
    },
    {
      "epoch": 0.42483539636555173,
      "grad_norm": 2.5473532676696777,
      "learning_rate": 2.8771398472478273e-05,
      "loss": 0.8705,
      "step": 16131
    },
    {
      "epoch": 0.4248617329470635,
      "grad_norm": 1.962107539176941,
      "learning_rate": 2.8770081643402685e-05,
      "loss": 1.304,
      "step": 16132
    },
    {
      "epoch": 0.4248880695285752,
      "grad_norm": 5.7192230224609375,
      "learning_rate": 2.8768764814327104e-05,
      "loss": 1.2819,
      "step": 16133
    },
    {
      "epoch": 0.4249144061100869,
      "grad_norm": 1.6997631788253784,
      "learning_rate": 2.8767447985251516e-05,
      "loss": 2.1681,
      "step": 16134
    },
    {
      "epoch": 0.42494074269159865,
      "grad_norm": 2.3568761348724365,
      "learning_rate": 2.876613115617593e-05,
      "loss": 1.8898,
      "step": 16135
    },
    {
      "epoch": 0.42496707927311034,
      "grad_norm": 2.2935540676116943,
      "learning_rate": 2.876481432710034e-05,
      "loss": 1.224,
      "step": 16136
    },
    {
      "epoch": 0.4249934158546221,
      "grad_norm": 2.655606746673584,
      "learning_rate": 2.8763497498024756e-05,
      "loss": 1.0649,
      "step": 16137
    },
    {
      "epoch": 0.42501975243613377,
      "grad_norm": 2.6828486919403076,
      "learning_rate": 2.8762180668949172e-05,
      "loss": 1.9586,
      "step": 16138
    },
    {
      "epoch": 0.4250460890176455,
      "grad_norm": 3.726508617401123,
      "learning_rate": 2.8760863839873587e-05,
      "loss": 1.5692,
      "step": 16139
    },
    {
      "epoch": 0.42507242559915726,
      "grad_norm": 7.708461284637451,
      "learning_rate": 2.8759547010798e-05,
      "loss": 1.3258,
      "step": 16140
    },
    {
      "epoch": 0.42509876218066894,
      "grad_norm": 4.94729471206665,
      "learning_rate": 2.8758230181722412e-05,
      "loss": 2.9704,
      "step": 16141
    },
    {
      "epoch": 0.4251250987621807,
      "grad_norm": 2.347440004348755,
      "learning_rate": 2.875691335264683e-05,
      "loss": 2.0847,
      "step": 16142
    },
    {
      "epoch": 0.4251514353436924,
      "grad_norm": 1.7017192840576172,
      "learning_rate": 2.8755596523571243e-05,
      "loss": 1.9728,
      "step": 16143
    },
    {
      "epoch": 0.4251777719252041,
      "grad_norm": 2.722093105316162,
      "learning_rate": 2.8754279694495655e-05,
      "loss": 0.7793,
      "step": 16144
    },
    {
      "epoch": 0.4252041085067158,
      "grad_norm": 1.8978090286254883,
      "learning_rate": 2.8752962865420067e-05,
      "loss": 1.8681,
      "step": 16145
    },
    {
      "epoch": 0.42523044508822755,
      "grad_norm": 2.12430477142334,
      "learning_rate": 2.8751646036344483e-05,
      "loss": 2.6236,
      "step": 16146
    },
    {
      "epoch": 0.4252567816697393,
      "grad_norm": 4.18602180480957,
      "learning_rate": 2.87503292072689e-05,
      "loss": 1.6183,
      "step": 16147
    },
    {
      "epoch": 0.425283118251251,
      "grad_norm": 1.5522241592407227,
      "learning_rate": 2.8749012378193314e-05,
      "loss": 1.9192,
      "step": 16148
    },
    {
      "epoch": 0.4253094548327627,
      "grad_norm": 2.20501971244812,
      "learning_rate": 2.8747695549117726e-05,
      "loss": 1.7048,
      "step": 16149
    },
    {
      "epoch": 0.4253357914142744,
      "grad_norm": 1.509581208229065,
      "learning_rate": 2.874637872004214e-05,
      "loss": 1.4333,
      "step": 16150
    },
    {
      "epoch": 0.42536212799578615,
      "grad_norm": 1.7848845720291138,
      "learning_rate": 2.874506189096655e-05,
      "loss": 1.7186,
      "step": 16151
    },
    {
      "epoch": 0.42538846457729784,
      "grad_norm": 2.1628212928771973,
      "learning_rate": 2.874374506189097e-05,
      "loss": 1.5007,
      "step": 16152
    },
    {
      "epoch": 0.4254148011588096,
      "grad_norm": 2.2218024730682373,
      "learning_rate": 2.8742428232815382e-05,
      "loss": 1.3011,
      "step": 16153
    },
    {
      "epoch": 0.42544113774032133,
      "grad_norm": 2.8692786693573,
      "learning_rate": 2.8741111403739794e-05,
      "loss": 1.2333,
      "step": 16154
    },
    {
      "epoch": 0.425467474321833,
      "grad_norm": 2.673438310623169,
      "learning_rate": 2.873979457466421e-05,
      "loss": 2.1072,
      "step": 16155
    },
    {
      "epoch": 0.42549381090334476,
      "grad_norm": 2.060781240463257,
      "learning_rate": 2.8738477745588625e-05,
      "loss": 1.981,
      "step": 16156
    },
    {
      "epoch": 0.42552014748485645,
      "grad_norm": 1.7278990745544434,
      "learning_rate": 2.873716091651304e-05,
      "loss": 1.6209,
      "step": 16157
    },
    {
      "epoch": 0.4255464840663682,
      "grad_norm": 2.8339948654174805,
      "learning_rate": 2.8735844087437453e-05,
      "loss": 1.2938,
      "step": 16158
    },
    {
      "epoch": 0.4255728206478799,
      "grad_norm": 2.2948098182678223,
      "learning_rate": 2.8734527258361866e-05,
      "loss": 1.3713,
      "step": 16159
    },
    {
      "epoch": 0.4255991572293916,
      "grad_norm": 1.9484632015228271,
      "learning_rate": 2.8733210429286278e-05,
      "loss": 1.2785,
      "step": 16160
    },
    {
      "epoch": 0.42562549381090337,
      "grad_norm": 3.2406296730041504,
      "learning_rate": 2.8731893600210697e-05,
      "loss": 0.6005,
      "step": 16161
    },
    {
      "epoch": 0.42565183039241505,
      "grad_norm": 2.459770917892456,
      "learning_rate": 2.873057677113511e-05,
      "loss": 0.7126,
      "step": 16162
    },
    {
      "epoch": 0.4256781669739268,
      "grad_norm": 2.5703790187835693,
      "learning_rate": 2.872925994205952e-05,
      "loss": 1.6406,
      "step": 16163
    },
    {
      "epoch": 0.4257045035554385,
      "grad_norm": 2.0524210929870605,
      "learning_rate": 2.8727943112983933e-05,
      "loss": 1.4072,
      "step": 16164
    },
    {
      "epoch": 0.42573084013695023,
      "grad_norm": 2.111962080001831,
      "learning_rate": 2.872662628390835e-05,
      "loss": 1.9734,
      "step": 16165
    },
    {
      "epoch": 0.4257571767184619,
      "grad_norm": 1.6880728006362915,
      "learning_rate": 2.8725309454832765e-05,
      "loss": 1.4628,
      "step": 16166
    },
    {
      "epoch": 0.42578351329997366,
      "grad_norm": 2.5431389808654785,
      "learning_rate": 2.872399262575718e-05,
      "loss": 1.1938,
      "step": 16167
    },
    {
      "epoch": 0.4258098498814854,
      "grad_norm": 2.4238691329956055,
      "learning_rate": 2.8722675796681592e-05,
      "loss": 2.3023,
      "step": 16168
    },
    {
      "epoch": 0.4258361864629971,
      "grad_norm": 6.673384666442871,
      "learning_rate": 2.8721358967606005e-05,
      "loss": 1.9657,
      "step": 16169
    },
    {
      "epoch": 0.42586252304450883,
      "grad_norm": 2.0809438228607178,
      "learning_rate": 2.8720042138530417e-05,
      "loss": 1.079,
      "step": 16170
    },
    {
      "epoch": 0.4258888596260205,
      "grad_norm": 1.7094929218292236,
      "learning_rate": 2.8718725309454836e-05,
      "loss": 1.9744,
      "step": 16171
    },
    {
      "epoch": 0.42591519620753227,
      "grad_norm": 2.3218657970428467,
      "learning_rate": 2.8717408480379248e-05,
      "loss": 1.4451,
      "step": 16172
    },
    {
      "epoch": 0.425941532789044,
      "grad_norm": 1.6189780235290527,
      "learning_rate": 2.871609165130366e-05,
      "loss": 1.8154,
      "step": 16173
    },
    {
      "epoch": 0.4259678693705557,
      "grad_norm": 3.1208300590515137,
      "learning_rate": 2.8714774822228076e-05,
      "loss": 1.5869,
      "step": 16174
    },
    {
      "epoch": 0.42599420595206744,
      "grad_norm": 1.9011083841323853,
      "learning_rate": 2.871345799315249e-05,
      "loss": 1.6565,
      "step": 16175
    },
    {
      "epoch": 0.4260205425335791,
      "grad_norm": 2.1710498332977295,
      "learning_rate": 2.8712141164076907e-05,
      "loss": 1.8116,
      "step": 16176
    },
    {
      "epoch": 0.42604687911509087,
      "grad_norm": 1.9681960344314575,
      "learning_rate": 2.871082433500132e-05,
      "loss": 0.6007,
      "step": 16177
    },
    {
      "epoch": 0.42607321569660256,
      "grad_norm": 2.6918060779571533,
      "learning_rate": 2.870950750592573e-05,
      "loss": 1.838,
      "step": 16178
    },
    {
      "epoch": 0.4260995522781143,
      "grad_norm": 2.6637935638427734,
      "learning_rate": 2.8708190676850144e-05,
      "loss": 2.1202,
      "step": 16179
    },
    {
      "epoch": 0.42612588885962605,
      "grad_norm": 2.059605836868286,
      "learning_rate": 2.8706873847774563e-05,
      "loss": 2.0905,
      "step": 16180
    },
    {
      "epoch": 0.42615222544113773,
      "grad_norm": 5.3449859619140625,
      "learning_rate": 2.8705557018698975e-05,
      "loss": 0.9404,
      "step": 16181
    },
    {
      "epoch": 0.4261785620226495,
      "grad_norm": 1.579784631729126,
      "learning_rate": 2.8704240189623387e-05,
      "loss": 1.3057,
      "step": 16182
    },
    {
      "epoch": 0.42620489860416116,
      "grad_norm": 1.8917019367218018,
      "learning_rate": 2.8702923360547803e-05,
      "loss": 1.8748,
      "step": 16183
    },
    {
      "epoch": 0.4262312351856729,
      "grad_norm": 2.118757486343384,
      "learning_rate": 2.8701606531472215e-05,
      "loss": 1.6654,
      "step": 16184
    },
    {
      "epoch": 0.4262575717671846,
      "grad_norm": 4.138052940368652,
      "learning_rate": 2.8700289702396634e-05,
      "loss": 1.891,
      "step": 16185
    },
    {
      "epoch": 0.42628390834869634,
      "grad_norm": 1.875789999961853,
      "learning_rate": 2.8698972873321046e-05,
      "loss": 1.3612,
      "step": 16186
    },
    {
      "epoch": 0.4263102449302081,
      "grad_norm": 2.8380467891693115,
      "learning_rate": 2.8697656044245458e-05,
      "loss": 1.7744,
      "step": 16187
    },
    {
      "epoch": 0.42633658151171977,
      "grad_norm": 2.5055181980133057,
      "learning_rate": 2.869633921516987e-05,
      "loss": 1.8268,
      "step": 16188
    },
    {
      "epoch": 0.4263629180932315,
      "grad_norm": 3.3114044666290283,
      "learning_rate": 2.869502238609429e-05,
      "loss": 2.2882,
      "step": 16189
    },
    {
      "epoch": 0.4263892546747432,
      "grad_norm": 4.40805196762085,
      "learning_rate": 2.86937055570187e-05,
      "loss": 1.8025,
      "step": 16190
    },
    {
      "epoch": 0.42641559125625494,
      "grad_norm": 1.9764550924301147,
      "learning_rate": 2.8692388727943114e-05,
      "loss": 1.0319,
      "step": 16191
    },
    {
      "epoch": 0.42644192783776663,
      "grad_norm": 1.6714836359024048,
      "learning_rate": 2.8691071898867526e-05,
      "loss": 2.482,
      "step": 16192
    },
    {
      "epoch": 0.4264682644192784,
      "grad_norm": 1.6736153364181519,
      "learning_rate": 2.868975506979194e-05,
      "loss": 1.6923,
      "step": 16193
    },
    {
      "epoch": 0.4264946010007901,
      "grad_norm": 1.7568933963775635,
      "learning_rate": 2.8688438240716357e-05,
      "loss": 2.3367,
      "step": 16194
    },
    {
      "epoch": 0.4265209375823018,
      "grad_norm": 2.0135488510131836,
      "learning_rate": 2.8687121411640773e-05,
      "loss": 0.7544,
      "step": 16195
    },
    {
      "epoch": 0.42654727416381355,
      "grad_norm": 1.9826570749282837,
      "learning_rate": 2.8685804582565185e-05,
      "loss": 1.8779,
      "step": 16196
    },
    {
      "epoch": 0.42657361074532524,
      "grad_norm": 2.129650354385376,
      "learning_rate": 2.8684487753489597e-05,
      "loss": 2.0082,
      "step": 16197
    },
    {
      "epoch": 0.426599947326837,
      "grad_norm": 2.671468496322632,
      "learning_rate": 2.868317092441401e-05,
      "loss": 0.7993,
      "step": 16198
    },
    {
      "epoch": 0.42662628390834867,
      "grad_norm": 3.3096554279327393,
      "learning_rate": 2.868185409533843e-05,
      "loss": 1.3271,
      "step": 16199
    },
    {
      "epoch": 0.4266526204898604,
      "grad_norm": 2.8333301544189453,
      "learning_rate": 2.868053726626284e-05,
      "loss": 1.8865,
      "step": 16200
    },
    {
      "epoch": 0.42667895707137216,
      "grad_norm": 4.446846008300781,
      "learning_rate": 2.8679220437187253e-05,
      "loss": 1.9822,
      "step": 16201
    },
    {
      "epoch": 0.42670529365288384,
      "grad_norm": 3.2590596675872803,
      "learning_rate": 2.867790360811167e-05,
      "loss": 1.6059,
      "step": 16202
    },
    {
      "epoch": 0.4267316302343956,
      "grad_norm": 2.5601813793182373,
      "learning_rate": 2.867658677903608e-05,
      "loss": 1.9725,
      "step": 16203
    },
    {
      "epoch": 0.4267579668159073,
      "grad_norm": 3.2188899517059326,
      "learning_rate": 2.86752699499605e-05,
      "loss": 2.2611,
      "step": 16204
    },
    {
      "epoch": 0.426784303397419,
      "grad_norm": 2.8070755004882812,
      "learning_rate": 2.8673953120884912e-05,
      "loss": 1.3266,
      "step": 16205
    },
    {
      "epoch": 0.42681063997893076,
      "grad_norm": 1.637706995010376,
      "learning_rate": 2.8672636291809324e-05,
      "loss": 2.1,
      "step": 16206
    },
    {
      "epoch": 0.42683697656044245,
      "grad_norm": 1.4981625080108643,
      "learning_rate": 2.8671319462733736e-05,
      "loss": 1.7029,
      "step": 16207
    },
    {
      "epoch": 0.4268633131419542,
      "grad_norm": 2.7051784992218018,
      "learning_rate": 2.8670002633658155e-05,
      "loss": 0.7684,
      "step": 16208
    },
    {
      "epoch": 0.4268896497234659,
      "grad_norm": 1.3664602041244507,
      "learning_rate": 2.8668685804582567e-05,
      "loss": 1.6947,
      "step": 16209
    },
    {
      "epoch": 0.4269159863049776,
      "grad_norm": 1.5344047546386719,
      "learning_rate": 2.866736897550698e-05,
      "loss": 1.6018,
      "step": 16210
    },
    {
      "epoch": 0.4269423228864893,
      "grad_norm": 2.9111382961273193,
      "learning_rate": 2.8666052146431395e-05,
      "loss": 0.7497,
      "step": 16211
    },
    {
      "epoch": 0.42696865946800105,
      "grad_norm": 2.486198663711548,
      "learning_rate": 2.8664735317355807e-05,
      "loss": 1.1718,
      "step": 16212
    },
    {
      "epoch": 0.4269949960495128,
      "grad_norm": 1.9173650741577148,
      "learning_rate": 2.8663418488280226e-05,
      "loss": 3.0041,
      "step": 16213
    },
    {
      "epoch": 0.4270213326310245,
      "grad_norm": 1.6175276041030884,
      "learning_rate": 2.866210165920464e-05,
      "loss": 1.1924,
      "step": 16214
    },
    {
      "epoch": 0.42704766921253623,
      "grad_norm": 1.9312342405319214,
      "learning_rate": 2.866078483012905e-05,
      "loss": 2.1435,
      "step": 16215
    },
    {
      "epoch": 0.4270740057940479,
      "grad_norm": 3.4961044788360596,
      "learning_rate": 2.8659468001053463e-05,
      "loss": 1.2657,
      "step": 16216
    },
    {
      "epoch": 0.42710034237555966,
      "grad_norm": 1.938255786895752,
      "learning_rate": 2.8658151171977875e-05,
      "loss": 2.1474,
      "step": 16217
    },
    {
      "epoch": 0.42712667895707135,
      "grad_norm": 2.5104610919952393,
      "learning_rate": 2.8656834342902294e-05,
      "loss": 0.7439,
      "step": 16218
    },
    {
      "epoch": 0.4271530155385831,
      "grad_norm": 2.8039329051971436,
      "learning_rate": 2.8655517513826707e-05,
      "loss": 1.8992,
      "step": 16219
    },
    {
      "epoch": 0.42717935212009484,
      "grad_norm": 1.9874725341796875,
      "learning_rate": 2.865420068475112e-05,
      "loss": 1.7834,
      "step": 16220
    },
    {
      "epoch": 0.4272056887016065,
      "grad_norm": 2.7436165809631348,
      "learning_rate": 2.8652883855675534e-05,
      "loss": 1.8501,
      "step": 16221
    },
    {
      "epoch": 0.42723202528311827,
      "grad_norm": 1.694075584411621,
      "learning_rate": 2.865156702659995e-05,
      "loss": 1.9433,
      "step": 16222
    },
    {
      "epoch": 0.42725836186462995,
      "grad_norm": 2.357699155807495,
      "learning_rate": 2.8650250197524366e-05,
      "loss": 0.4571,
      "step": 16223
    },
    {
      "epoch": 0.4272846984461417,
      "grad_norm": 1.7154619693756104,
      "learning_rate": 2.8648933368448778e-05,
      "loss": 2.0802,
      "step": 16224
    },
    {
      "epoch": 0.4273110350276534,
      "grad_norm": 1.6425429582595825,
      "learning_rate": 2.864761653937319e-05,
      "loss": 2.1626,
      "step": 16225
    },
    {
      "epoch": 0.42733737160916513,
      "grad_norm": 1.6343300342559814,
      "learning_rate": 2.8646299710297602e-05,
      "loss": 1.759,
      "step": 16226
    },
    {
      "epoch": 0.42736370819067687,
      "grad_norm": 1.6940662860870361,
      "learning_rate": 2.864498288122202e-05,
      "loss": 1.5748,
      "step": 16227
    },
    {
      "epoch": 0.42739004477218856,
      "grad_norm": 5.884566783905029,
      "learning_rate": 2.8643666052146433e-05,
      "loss": 1.0705,
      "step": 16228
    },
    {
      "epoch": 0.4274163813537003,
      "grad_norm": 1.9199550151824951,
      "learning_rate": 2.8642349223070846e-05,
      "loss": 2.1897,
      "step": 16229
    },
    {
      "epoch": 0.427442717935212,
      "grad_norm": 1.7519567012786865,
      "learning_rate": 2.864103239399526e-05,
      "loss": 2.28,
      "step": 16230
    },
    {
      "epoch": 0.42746905451672373,
      "grad_norm": 1.876975178718567,
      "learning_rate": 2.8639715564919673e-05,
      "loss": 1.6683,
      "step": 16231
    },
    {
      "epoch": 0.4274953910982354,
      "grad_norm": 3.0923686027526855,
      "learning_rate": 2.8638398735844092e-05,
      "loss": 0.9194,
      "step": 16232
    },
    {
      "epoch": 0.42752172767974717,
      "grad_norm": 2.1883909702301025,
      "learning_rate": 2.8637081906768505e-05,
      "loss": 0.9229,
      "step": 16233
    },
    {
      "epoch": 0.4275480642612589,
      "grad_norm": 1.7090818881988525,
      "learning_rate": 2.8635765077692917e-05,
      "loss": 1.8723,
      "step": 16234
    },
    {
      "epoch": 0.4275744008427706,
      "grad_norm": 3.919332981109619,
      "learning_rate": 2.863444824861733e-05,
      "loss": 1.8673,
      "step": 16235
    },
    {
      "epoch": 0.42760073742428234,
      "grad_norm": 3.2115514278411865,
      "learning_rate": 2.863313141954174e-05,
      "loss": 1.9871,
      "step": 16236
    },
    {
      "epoch": 0.42762707400579403,
      "grad_norm": 2.33235764503479,
      "learning_rate": 2.863181459046616e-05,
      "loss": 1.8323,
      "step": 16237
    },
    {
      "epoch": 0.42765341058730577,
      "grad_norm": 2.1285839080810547,
      "learning_rate": 2.8630497761390572e-05,
      "loss": 2.2213,
      "step": 16238
    },
    {
      "epoch": 0.4276797471688175,
      "grad_norm": 2.269050359725952,
      "learning_rate": 2.8629180932314985e-05,
      "loss": 1.4518,
      "step": 16239
    },
    {
      "epoch": 0.4277060837503292,
      "grad_norm": 2.024420738220215,
      "learning_rate": 2.86278641032394e-05,
      "loss": 2.097,
      "step": 16240
    },
    {
      "epoch": 0.42773242033184095,
      "grad_norm": 1.906901478767395,
      "learning_rate": 2.8626547274163816e-05,
      "loss": 2.1857,
      "step": 16241
    },
    {
      "epoch": 0.42775875691335263,
      "grad_norm": 2.3087074756622314,
      "learning_rate": 2.862523044508823e-05,
      "loss": 2.4593,
      "step": 16242
    },
    {
      "epoch": 0.4277850934948644,
      "grad_norm": 1.434691309928894,
      "learning_rate": 2.8623913616012644e-05,
      "loss": 1.7301,
      "step": 16243
    },
    {
      "epoch": 0.42781143007637606,
      "grad_norm": 1.7201263904571533,
      "learning_rate": 2.8622596786937056e-05,
      "loss": 1.1348,
      "step": 16244
    },
    {
      "epoch": 0.4278377666578878,
      "grad_norm": 1.6074718236923218,
      "learning_rate": 2.8621279957861468e-05,
      "loss": 1.6182,
      "step": 16245
    },
    {
      "epoch": 0.42786410323939955,
      "grad_norm": 2.414060115814209,
      "learning_rate": 2.8619963128785887e-05,
      "loss": 1.4731,
      "step": 16246
    },
    {
      "epoch": 0.42789043982091124,
      "grad_norm": 3.246783971786499,
      "learning_rate": 2.86186462997103e-05,
      "loss": 2.5249,
      "step": 16247
    },
    {
      "epoch": 0.427916776402423,
      "grad_norm": 1.9376178979873657,
      "learning_rate": 2.861732947063471e-05,
      "loss": 1.985,
      "step": 16248
    },
    {
      "epoch": 0.42794311298393467,
      "grad_norm": 1.9497365951538086,
      "learning_rate": 2.8616012641559127e-05,
      "loss": 1.4396,
      "step": 16249
    },
    {
      "epoch": 0.4279694495654464,
      "grad_norm": 3.112078905105591,
      "learning_rate": 2.861469581248354e-05,
      "loss": 1.7332,
      "step": 16250
    },
    {
      "epoch": 0.4279957861469581,
      "grad_norm": 2.140838861465454,
      "learning_rate": 2.8613378983407958e-05,
      "loss": 0.6323,
      "step": 16251
    },
    {
      "epoch": 0.42802212272846984,
      "grad_norm": 2.209380626678467,
      "learning_rate": 2.861206215433237e-05,
      "loss": 1.6335,
      "step": 16252
    },
    {
      "epoch": 0.4280484593099816,
      "grad_norm": 1.6760517358779907,
      "learning_rate": 2.8610745325256783e-05,
      "loss": 1.8567,
      "step": 16253
    },
    {
      "epoch": 0.4280747958914933,
      "grad_norm": 1.9600154161453247,
      "learning_rate": 2.8609428496181195e-05,
      "loss": 2.5125,
      "step": 16254
    },
    {
      "epoch": 0.428101132473005,
      "grad_norm": 1.9491287469863892,
      "learning_rate": 2.8608111667105614e-05,
      "loss": 2.3887,
      "step": 16255
    },
    {
      "epoch": 0.4281274690545167,
      "grad_norm": 1.8916538953781128,
      "learning_rate": 2.8606794838030026e-05,
      "loss": 0.2084,
      "step": 16256
    },
    {
      "epoch": 0.42815380563602845,
      "grad_norm": 2.9186549186706543,
      "learning_rate": 2.8605478008954438e-05,
      "loss": 1.6804,
      "step": 16257
    },
    {
      "epoch": 0.42818014221754014,
      "grad_norm": 1.9597272872924805,
      "learning_rate": 2.8604161179878854e-05,
      "loss": 0.7496,
      "step": 16258
    },
    {
      "epoch": 0.4282064787990519,
      "grad_norm": 3.079464912414551,
      "learning_rate": 2.8602844350803266e-05,
      "loss": 2.2019,
      "step": 16259
    },
    {
      "epoch": 0.4282328153805636,
      "grad_norm": 1.9543516635894775,
      "learning_rate": 2.8601527521727685e-05,
      "loss": 0.6934,
      "step": 16260
    },
    {
      "epoch": 0.4282591519620753,
      "grad_norm": 4.954905986785889,
      "learning_rate": 2.8600210692652097e-05,
      "loss": 1.1708,
      "step": 16261
    },
    {
      "epoch": 0.42828548854358706,
      "grad_norm": 3.6004514694213867,
      "learning_rate": 2.859889386357651e-05,
      "loss": 1.4056,
      "step": 16262
    },
    {
      "epoch": 0.42831182512509874,
      "grad_norm": 1.6310805082321167,
      "learning_rate": 2.859757703450092e-05,
      "loss": 1.6486,
      "step": 16263
    },
    {
      "epoch": 0.4283381617066105,
      "grad_norm": 2.038309097290039,
      "learning_rate": 2.8596260205425334e-05,
      "loss": 1.5376,
      "step": 16264
    },
    {
      "epoch": 0.4283644982881222,
      "grad_norm": 2.912075996398926,
      "learning_rate": 2.8594943376349753e-05,
      "loss": 2.2676,
      "step": 16265
    },
    {
      "epoch": 0.4283908348696339,
      "grad_norm": 2.8357462882995605,
      "learning_rate": 2.8593626547274165e-05,
      "loss": 1.5262,
      "step": 16266
    },
    {
      "epoch": 0.42841717145114566,
      "grad_norm": 2.675537586212158,
      "learning_rate": 2.8592309718198577e-05,
      "loss": 1.7726,
      "step": 16267
    },
    {
      "epoch": 0.42844350803265735,
      "grad_norm": 3.3788669109344482,
      "learning_rate": 2.8590992889122993e-05,
      "loss": 1.441,
      "step": 16268
    },
    {
      "epoch": 0.4284698446141691,
      "grad_norm": 2.9190473556518555,
      "learning_rate": 2.8589676060047405e-05,
      "loss": 0.4958,
      "step": 16269
    },
    {
      "epoch": 0.4284961811956808,
      "grad_norm": 5.56238317489624,
      "learning_rate": 2.8588359230971824e-05,
      "loss": 1.7465,
      "step": 16270
    },
    {
      "epoch": 0.4285225177771925,
      "grad_norm": 1.975429654121399,
      "learning_rate": 2.8587042401896236e-05,
      "loss": 1.1192,
      "step": 16271
    },
    {
      "epoch": 0.42854885435870427,
      "grad_norm": 1.9163085222244263,
      "learning_rate": 2.858572557282065e-05,
      "loss": 1.9938,
      "step": 16272
    },
    {
      "epoch": 0.42857519094021596,
      "grad_norm": 2.503218173980713,
      "learning_rate": 2.858440874374506e-05,
      "loss": 1.2884,
      "step": 16273
    },
    {
      "epoch": 0.4286015275217277,
      "grad_norm": 3.577244520187378,
      "learning_rate": 2.858309191466948e-05,
      "loss": 1.0316,
      "step": 16274
    },
    {
      "epoch": 0.4286278641032394,
      "grad_norm": 3.658782720565796,
      "learning_rate": 2.8581775085593892e-05,
      "loss": 1.1198,
      "step": 16275
    },
    {
      "epoch": 0.42865420068475113,
      "grad_norm": 2.451082468032837,
      "learning_rate": 2.8580458256518304e-05,
      "loss": 1.4398,
      "step": 16276
    },
    {
      "epoch": 0.4286805372662628,
      "grad_norm": 1.8286141157150269,
      "learning_rate": 2.857914142744272e-05,
      "loss": 2.6038,
      "step": 16277
    },
    {
      "epoch": 0.42870687384777456,
      "grad_norm": 2.7148830890655518,
      "learning_rate": 2.8577824598367132e-05,
      "loss": 1.314,
      "step": 16278
    },
    {
      "epoch": 0.4287332104292863,
      "grad_norm": 1.568585753440857,
      "learning_rate": 2.857650776929155e-05,
      "loss": 2.0191,
      "step": 16279
    },
    {
      "epoch": 0.428759547010798,
      "grad_norm": 2.098501443862915,
      "learning_rate": 2.8575190940215963e-05,
      "loss": 1.6439,
      "step": 16280
    },
    {
      "epoch": 0.42878588359230974,
      "grad_norm": 1.7789452075958252,
      "learning_rate": 2.8573874111140375e-05,
      "loss": 2.2273,
      "step": 16281
    },
    {
      "epoch": 0.4288122201738214,
      "grad_norm": 2.408709764480591,
      "learning_rate": 2.8572557282064788e-05,
      "loss": 1.5854,
      "step": 16282
    },
    {
      "epoch": 0.42883855675533317,
      "grad_norm": 1.5255910158157349,
      "learning_rate": 2.85712404529892e-05,
      "loss": 1.9672,
      "step": 16283
    },
    {
      "epoch": 0.42886489333684485,
      "grad_norm": 3.492717742919922,
      "learning_rate": 2.856992362391362e-05,
      "loss": 1.2016,
      "step": 16284
    },
    {
      "epoch": 0.4288912299183566,
      "grad_norm": 3.4919309616088867,
      "learning_rate": 2.856860679483803e-05,
      "loss": 1.942,
      "step": 16285
    },
    {
      "epoch": 0.42891756649986834,
      "grad_norm": 2.008453369140625,
      "learning_rate": 2.8567289965762447e-05,
      "loss": 1.498,
      "step": 16286
    },
    {
      "epoch": 0.42894390308138003,
      "grad_norm": 1.629468321800232,
      "learning_rate": 2.856597313668686e-05,
      "loss": 2.1533,
      "step": 16287
    },
    {
      "epoch": 0.4289702396628918,
      "grad_norm": 2.5563361644744873,
      "learning_rate": 2.8564656307611278e-05,
      "loss": 1.2936,
      "step": 16288
    },
    {
      "epoch": 0.42899657624440346,
      "grad_norm": 1.9235506057739258,
      "learning_rate": 2.856333947853569e-05,
      "loss": 1.0663,
      "step": 16289
    },
    {
      "epoch": 0.4290229128259152,
      "grad_norm": 1.5925219058990479,
      "learning_rate": 2.8562022649460102e-05,
      "loss": 1.638,
      "step": 16290
    },
    {
      "epoch": 0.4290492494074269,
      "grad_norm": 3.5170552730560303,
      "learning_rate": 2.8560705820384514e-05,
      "loss": 1.3771,
      "step": 16291
    },
    {
      "epoch": 0.42907558598893863,
      "grad_norm": 1.7583720684051514,
      "learning_rate": 2.8559388991308927e-05,
      "loss": 1.9183,
      "step": 16292
    },
    {
      "epoch": 0.4291019225704504,
      "grad_norm": 1.329180359840393,
      "learning_rate": 2.8558072162233346e-05,
      "loss": 0.4357,
      "step": 16293
    },
    {
      "epoch": 0.42912825915196207,
      "grad_norm": 4.115122318267822,
      "learning_rate": 2.8556755333157758e-05,
      "loss": 1.6922,
      "step": 16294
    },
    {
      "epoch": 0.4291545957334738,
      "grad_norm": 1.391040325164795,
      "learning_rate": 2.855543850408217e-05,
      "loss": 1.5295,
      "step": 16295
    },
    {
      "epoch": 0.4291809323149855,
      "grad_norm": 3.9990217685699463,
      "learning_rate": 2.8554121675006586e-05,
      "loss": 1.4757,
      "step": 16296
    },
    {
      "epoch": 0.42920726889649724,
      "grad_norm": 1.9993547201156616,
      "learning_rate": 2.8552804845930998e-05,
      "loss": 2.0338,
      "step": 16297
    },
    {
      "epoch": 0.42923360547800893,
      "grad_norm": 2.128110885620117,
      "learning_rate": 2.8551488016855417e-05,
      "loss": 2.2539,
      "step": 16298
    },
    {
      "epoch": 0.42925994205952067,
      "grad_norm": 3.921907424926758,
      "learning_rate": 2.855017118777983e-05,
      "loss": 1.1863,
      "step": 16299
    },
    {
      "epoch": 0.4292862786410324,
      "grad_norm": 1.785980224609375,
      "learning_rate": 2.854885435870424e-05,
      "loss": 1.802,
      "step": 16300
    },
    {
      "epoch": 0.4293126152225441,
      "grad_norm": 3.599180221557617,
      "learning_rate": 2.8547537529628653e-05,
      "loss": 2.0091,
      "step": 16301
    },
    {
      "epoch": 0.42933895180405585,
      "grad_norm": 2.1746342182159424,
      "learning_rate": 2.8546220700553066e-05,
      "loss": 1.7942,
      "step": 16302
    },
    {
      "epoch": 0.42936528838556753,
      "grad_norm": 2.8276827335357666,
      "learning_rate": 2.8544903871477485e-05,
      "loss": 0.3247,
      "step": 16303
    },
    {
      "epoch": 0.4293916249670793,
      "grad_norm": 4.218794345855713,
      "learning_rate": 2.8543587042401897e-05,
      "loss": 1.0507,
      "step": 16304
    },
    {
      "epoch": 0.429417961548591,
      "grad_norm": 1.9216302633285522,
      "learning_rate": 2.8542270213326312e-05,
      "loss": 2.2101,
      "step": 16305
    },
    {
      "epoch": 0.4294442981301027,
      "grad_norm": 2.8261210918426514,
      "learning_rate": 2.8540953384250725e-05,
      "loss": 1.2542,
      "step": 16306
    },
    {
      "epoch": 0.42947063471161445,
      "grad_norm": 2.011619806289673,
      "learning_rate": 2.8539636555175144e-05,
      "loss": 1.7692,
      "step": 16307
    },
    {
      "epoch": 0.42949697129312614,
      "grad_norm": 3.0594727993011475,
      "learning_rate": 2.8538319726099556e-05,
      "loss": 0.8944,
      "step": 16308
    },
    {
      "epoch": 0.4295233078746379,
      "grad_norm": 1.8649332523345947,
      "learning_rate": 2.8537002897023968e-05,
      "loss": 1.2833,
      "step": 16309
    },
    {
      "epoch": 0.42954964445614957,
      "grad_norm": 1.578713297843933,
      "learning_rate": 2.853568606794838e-05,
      "loss": 1.7837,
      "step": 16310
    },
    {
      "epoch": 0.4295759810376613,
      "grad_norm": 3.6511008739471436,
      "learning_rate": 2.8534369238872792e-05,
      "loss": 2.4896,
      "step": 16311
    },
    {
      "epoch": 0.42960231761917306,
      "grad_norm": 1.9747178554534912,
      "learning_rate": 2.853305240979721e-05,
      "loss": 1.8094,
      "step": 16312
    },
    {
      "epoch": 0.42962865420068475,
      "grad_norm": 2.046444892883301,
      "learning_rate": 2.8531735580721624e-05,
      "loss": 1.6579,
      "step": 16313
    },
    {
      "epoch": 0.4296549907821965,
      "grad_norm": 4.64616060256958,
      "learning_rate": 2.8530418751646036e-05,
      "loss": 1.9714,
      "step": 16314
    },
    {
      "epoch": 0.4296813273637082,
      "grad_norm": 4.983663558959961,
      "learning_rate": 2.852910192257045e-05,
      "loss": 1.3401,
      "step": 16315
    },
    {
      "epoch": 0.4297076639452199,
      "grad_norm": 2.8350348472595215,
      "learning_rate": 2.8527785093494864e-05,
      "loss": 2.3301,
      "step": 16316
    },
    {
      "epoch": 0.4297340005267316,
      "grad_norm": 1.832379937171936,
      "learning_rate": 2.8526468264419283e-05,
      "loss": 1.6732,
      "step": 16317
    },
    {
      "epoch": 0.42976033710824335,
      "grad_norm": 1.5315078496932983,
      "learning_rate": 2.8525151435343695e-05,
      "loss": 2.3453,
      "step": 16318
    },
    {
      "epoch": 0.4297866736897551,
      "grad_norm": 2.1844592094421387,
      "learning_rate": 2.8523834606268107e-05,
      "loss": 1.7826,
      "step": 16319
    },
    {
      "epoch": 0.4298130102712668,
      "grad_norm": 2.5775365829467773,
      "learning_rate": 2.852251777719252e-05,
      "loss": 1.9782,
      "step": 16320
    },
    {
      "epoch": 0.4298393468527785,
      "grad_norm": 1.9965500831604004,
      "learning_rate": 2.8521200948116938e-05,
      "loss": 1.702,
      "step": 16321
    },
    {
      "epoch": 0.4298656834342902,
      "grad_norm": 1.993498682975769,
      "learning_rate": 2.851988411904135e-05,
      "loss": 1.7707,
      "step": 16322
    },
    {
      "epoch": 0.42989202001580196,
      "grad_norm": 2.2596898078918457,
      "learning_rate": 2.8518567289965763e-05,
      "loss": 0.5715,
      "step": 16323
    },
    {
      "epoch": 0.42991835659731364,
      "grad_norm": 3.460712432861328,
      "learning_rate": 2.8517250460890178e-05,
      "loss": 1.5325,
      "step": 16324
    },
    {
      "epoch": 0.4299446931788254,
      "grad_norm": 3.9280924797058105,
      "learning_rate": 2.851593363181459e-05,
      "loss": 1.4971,
      "step": 16325
    },
    {
      "epoch": 0.42997102976033713,
      "grad_norm": 1.9667539596557617,
      "learning_rate": 2.851461680273901e-05,
      "loss": 1.1646,
      "step": 16326
    },
    {
      "epoch": 0.4299973663418488,
      "grad_norm": 3.4098823070526123,
      "learning_rate": 2.851329997366342e-05,
      "loss": 2.3861,
      "step": 16327
    },
    {
      "epoch": 0.43002370292336056,
      "grad_norm": 2.2132582664489746,
      "learning_rate": 2.8511983144587834e-05,
      "loss": 1.357,
      "step": 16328
    },
    {
      "epoch": 0.43005003950487225,
      "grad_norm": 2.117368459701538,
      "learning_rate": 2.8510666315512246e-05,
      "loss": 0.4702,
      "step": 16329
    },
    {
      "epoch": 0.430076376086384,
      "grad_norm": 1.898853063583374,
      "learning_rate": 2.8509349486436658e-05,
      "loss": 1.3479,
      "step": 16330
    },
    {
      "epoch": 0.4301027126678957,
      "grad_norm": 2.343618869781494,
      "learning_rate": 2.8508032657361077e-05,
      "loss": 1.9506,
      "step": 16331
    },
    {
      "epoch": 0.4301290492494074,
      "grad_norm": 1.6748192310333252,
      "learning_rate": 2.850671582828549e-05,
      "loss": 1.7221,
      "step": 16332
    },
    {
      "epoch": 0.43015538583091917,
      "grad_norm": 2.4770283699035645,
      "learning_rate": 2.8505398999209905e-05,
      "loss": 0.3854,
      "step": 16333
    },
    {
      "epoch": 0.43018172241243086,
      "grad_norm": 2.526390790939331,
      "learning_rate": 2.8504082170134317e-05,
      "loss": 2.1985,
      "step": 16334
    },
    {
      "epoch": 0.4302080589939426,
      "grad_norm": 1.5955396890640259,
      "learning_rate": 2.850276534105873e-05,
      "loss": 2.0176,
      "step": 16335
    },
    {
      "epoch": 0.4302343955754543,
      "grad_norm": 2.1602351665496826,
      "learning_rate": 2.850144851198315e-05,
      "loss": 2.0718,
      "step": 16336
    },
    {
      "epoch": 0.43026073215696603,
      "grad_norm": 2.4446284770965576,
      "learning_rate": 2.850013168290756e-05,
      "loss": 2.1879,
      "step": 16337
    },
    {
      "epoch": 0.4302870687384777,
      "grad_norm": 1.8843352794647217,
      "learning_rate": 2.8498814853831973e-05,
      "loss": 1.5665,
      "step": 16338
    },
    {
      "epoch": 0.43031340531998946,
      "grad_norm": 2.402031660079956,
      "learning_rate": 2.8497498024756385e-05,
      "loss": 1.3319,
      "step": 16339
    },
    {
      "epoch": 0.4303397419015012,
      "grad_norm": 1.7032818794250488,
      "learning_rate": 2.8496181195680804e-05,
      "loss": 1.4115,
      "step": 16340
    },
    {
      "epoch": 0.4303660784830129,
      "grad_norm": 3.1331326961517334,
      "learning_rate": 2.8494864366605216e-05,
      "loss": 1.5468,
      "step": 16341
    },
    {
      "epoch": 0.43039241506452464,
      "grad_norm": 1.6129095554351807,
      "learning_rate": 2.849354753752963e-05,
      "loss": 1.495,
      "step": 16342
    },
    {
      "epoch": 0.4304187516460363,
      "grad_norm": 2.6606736183166504,
      "learning_rate": 2.8492230708454044e-05,
      "loss": 1.5746,
      "step": 16343
    },
    {
      "epoch": 0.43044508822754807,
      "grad_norm": 4.29287576675415,
      "learning_rate": 2.8490913879378456e-05,
      "loss": 0.8516,
      "step": 16344
    },
    {
      "epoch": 0.4304714248090598,
      "grad_norm": 1.9414594173431396,
      "learning_rate": 2.8489597050302875e-05,
      "loss": 2.2639,
      "step": 16345
    },
    {
      "epoch": 0.4304977613905715,
      "grad_norm": 2.6296045780181885,
      "learning_rate": 2.8488280221227288e-05,
      "loss": 1.6935,
      "step": 16346
    },
    {
      "epoch": 0.43052409797208324,
      "grad_norm": 4.665785312652588,
      "learning_rate": 2.84869633921517e-05,
      "loss": 1.1521,
      "step": 16347
    },
    {
      "epoch": 0.43055043455359493,
      "grad_norm": 5.824530124664307,
      "learning_rate": 2.8485646563076112e-05,
      "loss": 1.3369,
      "step": 16348
    },
    {
      "epoch": 0.4305767711351067,
      "grad_norm": 3.34356689453125,
      "learning_rate": 2.8484329734000524e-05,
      "loss": 1.7588,
      "step": 16349
    },
    {
      "epoch": 0.43060310771661836,
      "grad_norm": 1.9068779945373535,
      "learning_rate": 2.8483012904924943e-05,
      "loss": 1.8227,
      "step": 16350
    },
    {
      "epoch": 0.4306294442981301,
      "grad_norm": 3.3791561126708984,
      "learning_rate": 2.8481696075849355e-05,
      "loss": 1.8706,
      "step": 16351
    },
    {
      "epoch": 0.43065578087964185,
      "grad_norm": 2.809380531311035,
      "learning_rate": 2.848037924677377e-05,
      "loss": 1.5818,
      "step": 16352
    },
    {
      "epoch": 0.43068211746115354,
      "grad_norm": 2.157747983932495,
      "learning_rate": 2.8479062417698183e-05,
      "loss": 1.8067,
      "step": 16353
    },
    {
      "epoch": 0.4307084540426653,
      "grad_norm": 1.7366297245025635,
      "learning_rate": 2.8477745588622602e-05,
      "loss": 1.6633,
      "step": 16354
    },
    {
      "epoch": 0.43073479062417697,
      "grad_norm": 2.163754463195801,
      "learning_rate": 2.8476428759547014e-05,
      "loss": 1.707,
      "step": 16355
    },
    {
      "epoch": 0.4307611272056887,
      "grad_norm": 2.410123109817505,
      "learning_rate": 2.8475111930471427e-05,
      "loss": 1.8286,
      "step": 16356
    },
    {
      "epoch": 0.4307874637872004,
      "grad_norm": 1.6513198614120483,
      "learning_rate": 2.847379510139584e-05,
      "loss": 1.8406,
      "step": 16357
    },
    {
      "epoch": 0.43081380036871214,
      "grad_norm": 3.8365464210510254,
      "learning_rate": 2.847247827232025e-05,
      "loss": 1.2741,
      "step": 16358
    },
    {
      "epoch": 0.4308401369502239,
      "grad_norm": 4.509221076965332,
      "learning_rate": 2.847116144324467e-05,
      "loss": 1.6658,
      "step": 16359
    },
    {
      "epoch": 0.43086647353173557,
      "grad_norm": 1.5654850006103516,
      "learning_rate": 2.8469844614169082e-05,
      "loss": 2.6995,
      "step": 16360
    },
    {
      "epoch": 0.4308928101132473,
      "grad_norm": 2.0247111320495605,
      "learning_rate": 2.8468527785093498e-05,
      "loss": 2.5882,
      "step": 16361
    },
    {
      "epoch": 0.430919146694759,
      "grad_norm": 3.857208251953125,
      "learning_rate": 2.846721095601791e-05,
      "loss": 1.3429,
      "step": 16362
    },
    {
      "epoch": 0.43094548327627075,
      "grad_norm": 2.45232892036438,
      "learning_rate": 2.8465894126942322e-05,
      "loss": 1.9087,
      "step": 16363
    },
    {
      "epoch": 0.43097181985778243,
      "grad_norm": 1.613037347793579,
      "learning_rate": 2.846457729786674e-05,
      "loss": 1.7584,
      "step": 16364
    },
    {
      "epoch": 0.4309981564392942,
      "grad_norm": 2.3972525596618652,
      "learning_rate": 2.8463260468791153e-05,
      "loss": 1.4356,
      "step": 16365
    },
    {
      "epoch": 0.4310244930208059,
      "grad_norm": 1.428754448890686,
      "learning_rate": 2.8461943639715566e-05,
      "loss": 0.2875,
      "step": 16366
    },
    {
      "epoch": 0.4310508296023176,
      "grad_norm": 2.2029333114624023,
      "learning_rate": 2.8460626810639978e-05,
      "loss": 1.7077,
      "step": 16367
    },
    {
      "epoch": 0.43107716618382935,
      "grad_norm": 2.456862449645996,
      "learning_rate": 2.845930998156439e-05,
      "loss": 1.5504,
      "step": 16368
    },
    {
      "epoch": 0.43110350276534104,
      "grad_norm": 2.1207661628723145,
      "learning_rate": 2.845799315248881e-05,
      "loss": 2.138,
      "step": 16369
    },
    {
      "epoch": 0.4311298393468528,
      "grad_norm": 2.8177073001861572,
      "learning_rate": 2.845667632341322e-05,
      "loss": 2.0563,
      "step": 16370
    },
    {
      "epoch": 0.43115617592836447,
      "grad_norm": 1.7018980979919434,
      "learning_rate": 2.8455359494337637e-05,
      "loss": 1.4509,
      "step": 16371
    },
    {
      "epoch": 0.4311825125098762,
      "grad_norm": 5.539059162139893,
      "learning_rate": 2.845404266526205e-05,
      "loss": 1.1372,
      "step": 16372
    },
    {
      "epoch": 0.43120884909138796,
      "grad_norm": 3.2257654666900635,
      "learning_rate": 2.8452725836186468e-05,
      "loss": 2.1277,
      "step": 16373
    },
    {
      "epoch": 0.43123518567289965,
      "grad_norm": 3.392864942550659,
      "learning_rate": 2.845140900711088e-05,
      "loss": 1.1504,
      "step": 16374
    },
    {
      "epoch": 0.4312615222544114,
      "grad_norm": 1.6590944528579712,
      "learning_rate": 2.8450092178035292e-05,
      "loss": 1.9739,
      "step": 16375
    },
    {
      "epoch": 0.4312878588359231,
      "grad_norm": 2.2991340160369873,
      "learning_rate": 2.8448775348959705e-05,
      "loss": 1.8108,
      "step": 16376
    },
    {
      "epoch": 0.4313141954174348,
      "grad_norm": 1.9851282835006714,
      "learning_rate": 2.8447458519884117e-05,
      "loss": 1.8127,
      "step": 16377
    },
    {
      "epoch": 0.43134053199894656,
      "grad_norm": 1.7383449077606201,
      "learning_rate": 2.8446141690808536e-05,
      "loss": 2.2359,
      "step": 16378
    },
    {
      "epoch": 0.43136686858045825,
      "grad_norm": 2.4224584102630615,
      "learning_rate": 2.8444824861732948e-05,
      "loss": 0.7735,
      "step": 16379
    },
    {
      "epoch": 0.43139320516197,
      "grad_norm": 3.781935930252075,
      "learning_rate": 2.8443508032657364e-05,
      "loss": 1.8866,
      "step": 16380
    },
    {
      "epoch": 0.4314195417434817,
      "grad_norm": 1.9859447479248047,
      "learning_rate": 2.8442191203581776e-05,
      "loss": 1.9327,
      "step": 16381
    },
    {
      "epoch": 0.4314458783249934,
      "grad_norm": 2.18391752243042,
      "learning_rate": 2.8440874374506188e-05,
      "loss": 1.5739,
      "step": 16382
    },
    {
      "epoch": 0.4314722149065051,
      "grad_norm": 1.9801342487335205,
      "learning_rate": 2.8439557545430607e-05,
      "loss": 1.3973,
      "step": 16383
    },
    {
      "epoch": 0.43149855148801686,
      "grad_norm": 2.1786768436431885,
      "learning_rate": 2.843824071635502e-05,
      "loss": 1.7656,
      "step": 16384
    },
    {
      "epoch": 0.4315248880695286,
      "grad_norm": 3.137179374694824,
      "learning_rate": 2.843692388727943e-05,
      "loss": 2.1335,
      "step": 16385
    },
    {
      "epoch": 0.4315512246510403,
      "grad_norm": 1.8687926530838013,
      "learning_rate": 2.8435607058203844e-05,
      "loss": 2.5486,
      "step": 16386
    },
    {
      "epoch": 0.43157756123255203,
      "grad_norm": 3.301006317138672,
      "learning_rate": 2.8434290229128263e-05,
      "loss": 0.3844,
      "step": 16387
    },
    {
      "epoch": 0.4316038978140637,
      "grad_norm": 4.079667568206787,
      "learning_rate": 2.8432973400052675e-05,
      "loss": 0.9807,
      "step": 16388
    },
    {
      "epoch": 0.43163023439557546,
      "grad_norm": 3.8437492847442627,
      "learning_rate": 2.843165657097709e-05,
      "loss": 1.8664,
      "step": 16389
    },
    {
      "epoch": 0.43165657097708715,
      "grad_norm": 1.8211383819580078,
      "learning_rate": 2.8430339741901503e-05,
      "loss": 2.09,
      "step": 16390
    },
    {
      "epoch": 0.4316829075585989,
      "grad_norm": 2.4085426330566406,
      "learning_rate": 2.8429022912825915e-05,
      "loss": 1.765,
      "step": 16391
    },
    {
      "epoch": 0.43170924414011064,
      "grad_norm": 1.6934255361557007,
      "learning_rate": 2.8427706083750334e-05,
      "loss": 2.0942,
      "step": 16392
    },
    {
      "epoch": 0.4317355807216223,
      "grad_norm": 1.5776584148406982,
      "learning_rate": 2.8426389254674746e-05,
      "loss": 1.9435,
      "step": 16393
    },
    {
      "epoch": 0.43176191730313407,
      "grad_norm": 2.0898146629333496,
      "learning_rate": 2.8425072425599158e-05,
      "loss": 1.8544,
      "step": 16394
    },
    {
      "epoch": 0.43178825388464576,
      "grad_norm": 2.8114964962005615,
      "learning_rate": 2.842375559652357e-05,
      "loss": 1.2003,
      "step": 16395
    },
    {
      "epoch": 0.4318145904661575,
      "grad_norm": 2.9184961318969727,
      "learning_rate": 2.8422438767447983e-05,
      "loss": 2.6132,
      "step": 16396
    },
    {
      "epoch": 0.4318409270476692,
      "grad_norm": 2.2989752292633057,
      "learning_rate": 2.84211219383724e-05,
      "loss": 2.1767,
      "step": 16397
    },
    {
      "epoch": 0.43186726362918093,
      "grad_norm": 2.08762788772583,
      "learning_rate": 2.8419805109296814e-05,
      "loss": 1.0648,
      "step": 16398
    },
    {
      "epoch": 0.4318936002106927,
      "grad_norm": 2.394007682800293,
      "learning_rate": 2.841848828022123e-05,
      "loss": 1.9494,
      "step": 16399
    },
    {
      "epoch": 0.43191993679220436,
      "grad_norm": 2.7329792976379395,
      "learning_rate": 2.8417171451145642e-05,
      "loss": 1.4813,
      "step": 16400
    },
    {
      "epoch": 0.4319462733737161,
      "grad_norm": 1.8522226810455322,
      "learning_rate": 2.841585462207006e-05,
      "loss": 1.7778,
      "step": 16401
    },
    {
      "epoch": 0.4319726099552278,
      "grad_norm": 2.5576720237731934,
      "learning_rate": 2.8414537792994473e-05,
      "loss": 1.631,
      "step": 16402
    },
    {
      "epoch": 0.43199894653673954,
      "grad_norm": 3.952995777130127,
      "learning_rate": 2.8413220963918885e-05,
      "loss": 0.7376,
      "step": 16403
    },
    {
      "epoch": 0.4320252831182512,
      "grad_norm": 1.7281336784362793,
      "learning_rate": 2.8411904134843297e-05,
      "loss": 1.6091,
      "step": 16404
    },
    {
      "epoch": 0.43205161969976297,
      "grad_norm": 2.3904290199279785,
      "learning_rate": 2.841058730576771e-05,
      "loss": 1.9631,
      "step": 16405
    },
    {
      "epoch": 0.4320779562812747,
      "grad_norm": 3.5860753059387207,
      "learning_rate": 2.840927047669213e-05,
      "loss": 2.4206,
      "step": 16406
    },
    {
      "epoch": 0.4321042928627864,
      "grad_norm": 4.733084201812744,
      "learning_rate": 2.840795364761654e-05,
      "loss": 1.1107,
      "step": 16407
    },
    {
      "epoch": 0.43213062944429814,
      "grad_norm": 1.4516595602035522,
      "learning_rate": 2.8406636818540956e-05,
      "loss": 1.8459,
      "step": 16408
    },
    {
      "epoch": 0.43215696602580983,
      "grad_norm": 4.789068698883057,
      "learning_rate": 2.840531998946537e-05,
      "loss": 1.3535,
      "step": 16409
    },
    {
      "epoch": 0.4321833026073216,
      "grad_norm": 1.5293134450912476,
      "learning_rate": 2.840400316038978e-05,
      "loss": 1.8594,
      "step": 16410
    },
    {
      "epoch": 0.4322096391888333,
      "grad_norm": 2.584707736968994,
      "learning_rate": 2.84026863313142e-05,
      "loss": 1.5222,
      "step": 16411
    },
    {
      "epoch": 0.432235975770345,
      "grad_norm": 1.5266469717025757,
      "learning_rate": 2.8401369502238612e-05,
      "loss": 2.3191,
      "step": 16412
    },
    {
      "epoch": 0.43226231235185675,
      "grad_norm": 1.7189316749572754,
      "learning_rate": 2.8400052673163024e-05,
      "loss": 1.7969,
      "step": 16413
    },
    {
      "epoch": 0.43228864893336844,
      "grad_norm": 1.8869389295578003,
      "learning_rate": 2.8398735844087436e-05,
      "loss": 2.1453,
      "step": 16414
    },
    {
      "epoch": 0.4323149855148802,
      "grad_norm": 1.6695712804794312,
      "learning_rate": 2.839741901501185e-05,
      "loss": 1.4628,
      "step": 16415
    },
    {
      "epoch": 0.43234132209639187,
      "grad_norm": 4.673247814178467,
      "learning_rate": 2.8396102185936268e-05,
      "loss": 0.482,
      "step": 16416
    },
    {
      "epoch": 0.4323676586779036,
      "grad_norm": 1.7116349935531616,
      "learning_rate": 2.839478535686068e-05,
      "loss": 2.653,
      "step": 16417
    },
    {
      "epoch": 0.43239399525941535,
      "grad_norm": 1.8586455583572388,
      "learning_rate": 2.8393468527785095e-05,
      "loss": 1.7581,
      "step": 16418
    },
    {
      "epoch": 0.43242033184092704,
      "grad_norm": 6.648214817047119,
      "learning_rate": 2.8392151698709508e-05,
      "loss": 1.9338,
      "step": 16419
    },
    {
      "epoch": 0.4324466684224388,
      "grad_norm": 2.884911298751831,
      "learning_rate": 2.8390834869633927e-05,
      "loss": 1.7442,
      "step": 16420
    },
    {
      "epoch": 0.4324730050039505,
      "grad_norm": 3.1319708824157715,
      "learning_rate": 2.838951804055834e-05,
      "loss": 1.619,
      "step": 16421
    },
    {
      "epoch": 0.4324993415854622,
      "grad_norm": 1.5436334609985352,
      "learning_rate": 2.838820121148275e-05,
      "loss": 1.5905,
      "step": 16422
    },
    {
      "epoch": 0.4325256781669739,
      "grad_norm": 1.6361109018325806,
      "learning_rate": 2.8386884382407163e-05,
      "loss": 1.7514,
      "step": 16423
    },
    {
      "epoch": 0.43255201474848565,
      "grad_norm": 2.9640231132507324,
      "learning_rate": 2.8385567553331575e-05,
      "loss": 1.2714,
      "step": 16424
    },
    {
      "epoch": 0.4325783513299974,
      "grad_norm": 3.1678459644317627,
      "learning_rate": 2.8384250724255994e-05,
      "loss": 0.7944,
      "step": 16425
    },
    {
      "epoch": 0.4326046879115091,
      "grad_norm": 1.590768814086914,
      "learning_rate": 2.8382933895180407e-05,
      "loss": 1.4295,
      "step": 16426
    },
    {
      "epoch": 0.4326310244930208,
      "grad_norm": 1.7419683933258057,
      "learning_rate": 2.8381617066104822e-05,
      "loss": 1.554,
      "step": 16427
    },
    {
      "epoch": 0.4326573610745325,
      "grad_norm": 2.3588669300079346,
      "learning_rate": 2.8380300237029234e-05,
      "loss": 1.8619,
      "step": 16428
    },
    {
      "epoch": 0.43268369765604425,
      "grad_norm": 1.926154613494873,
      "learning_rate": 2.8378983407953647e-05,
      "loss": 1.2702,
      "step": 16429
    },
    {
      "epoch": 0.43271003423755594,
      "grad_norm": 4.687825679779053,
      "learning_rate": 2.8377666578878066e-05,
      "loss": 1.198,
      "step": 16430
    },
    {
      "epoch": 0.4327363708190677,
      "grad_norm": 1.9334768056869507,
      "learning_rate": 2.8376349749802478e-05,
      "loss": 2.1527,
      "step": 16431
    },
    {
      "epoch": 0.4327627074005794,
      "grad_norm": 1.7973761558532715,
      "learning_rate": 2.837503292072689e-05,
      "loss": 2.0785,
      "step": 16432
    },
    {
      "epoch": 0.4327890439820911,
      "grad_norm": 1.5084431171417236,
      "learning_rate": 2.8373716091651302e-05,
      "loss": 1.7387,
      "step": 16433
    },
    {
      "epoch": 0.43281538056360286,
      "grad_norm": 2.261404037475586,
      "learning_rate": 2.837239926257572e-05,
      "loss": 0.9094,
      "step": 16434
    },
    {
      "epoch": 0.43284171714511455,
      "grad_norm": 3.752955436706543,
      "learning_rate": 2.8371082433500133e-05,
      "loss": 2.2964,
      "step": 16435
    },
    {
      "epoch": 0.4328680537266263,
      "grad_norm": 1.6899412870407104,
      "learning_rate": 2.836976560442455e-05,
      "loss": 1.3676,
      "step": 16436
    },
    {
      "epoch": 0.432894390308138,
      "grad_norm": 3.6785402297973633,
      "learning_rate": 2.836844877534896e-05,
      "loss": 1.9773,
      "step": 16437
    },
    {
      "epoch": 0.4329207268896497,
      "grad_norm": 1.5583354234695435,
      "learning_rate": 2.8367131946273373e-05,
      "loss": 1.2187,
      "step": 16438
    },
    {
      "epoch": 0.43294706347116146,
      "grad_norm": 1.9530384540557861,
      "learning_rate": 2.8365815117197792e-05,
      "loss": 1.8882,
      "step": 16439
    },
    {
      "epoch": 0.43297340005267315,
      "grad_norm": 4.859649181365967,
      "learning_rate": 2.8364498288122205e-05,
      "loss": 0.9564,
      "step": 16440
    },
    {
      "epoch": 0.4329997366341849,
      "grad_norm": 1.9613500833511353,
      "learning_rate": 2.8363181459046617e-05,
      "loss": 2.6533,
      "step": 16441
    },
    {
      "epoch": 0.4330260732156966,
      "grad_norm": 1.6729531288146973,
      "learning_rate": 2.836186462997103e-05,
      "loss": 1.797,
      "step": 16442
    },
    {
      "epoch": 0.4330524097972083,
      "grad_norm": 1.5913046598434448,
      "learning_rate": 2.836054780089544e-05,
      "loss": 1.7418,
      "step": 16443
    },
    {
      "epoch": 0.43307874637872007,
      "grad_norm": 3.0342214107513428,
      "learning_rate": 2.835923097181986e-05,
      "loss": 1.3378,
      "step": 16444
    },
    {
      "epoch": 0.43310508296023176,
      "grad_norm": 4.726546764373779,
      "learning_rate": 2.8357914142744272e-05,
      "loss": 1.392,
      "step": 16445
    },
    {
      "epoch": 0.4331314195417435,
      "grad_norm": 2.0671494007110596,
      "learning_rate": 2.8356597313668688e-05,
      "loss": 1.8711,
      "step": 16446
    },
    {
      "epoch": 0.4331577561232552,
      "grad_norm": 2.8993146419525146,
      "learning_rate": 2.83552804845931e-05,
      "loss": 2.1863,
      "step": 16447
    },
    {
      "epoch": 0.43318409270476693,
      "grad_norm": 1.5807468891143799,
      "learning_rate": 2.8353963655517512e-05,
      "loss": 2.6426,
      "step": 16448
    },
    {
      "epoch": 0.4332104292862786,
      "grad_norm": 1.556812047958374,
      "learning_rate": 2.835264682644193e-05,
      "loss": 2.0733,
      "step": 16449
    },
    {
      "epoch": 0.43323676586779036,
      "grad_norm": 2.182522773742676,
      "learning_rate": 2.8351329997366344e-05,
      "loss": 2.089,
      "step": 16450
    },
    {
      "epoch": 0.4332631024493021,
      "grad_norm": 2.5087220668792725,
      "learning_rate": 2.8350013168290756e-05,
      "loss": 1.3287,
      "step": 16451
    },
    {
      "epoch": 0.4332894390308138,
      "grad_norm": 1.84579336643219,
      "learning_rate": 2.8348696339215168e-05,
      "loss": 1.7719,
      "step": 16452
    },
    {
      "epoch": 0.43331577561232554,
      "grad_norm": 1.531093716621399,
      "learning_rate": 2.8347379510139587e-05,
      "loss": 2.1779,
      "step": 16453
    },
    {
      "epoch": 0.4333421121938372,
      "grad_norm": 2.491909980773926,
      "learning_rate": 2.8346062681064e-05,
      "loss": 1.1051,
      "step": 16454
    },
    {
      "epoch": 0.43336844877534897,
      "grad_norm": 2.3045425415039062,
      "learning_rate": 2.8344745851988415e-05,
      "loss": 1.3491,
      "step": 16455
    },
    {
      "epoch": 0.43339478535686066,
      "grad_norm": 1.6769473552703857,
      "learning_rate": 2.8343429022912827e-05,
      "loss": 1.0622,
      "step": 16456
    },
    {
      "epoch": 0.4334211219383724,
      "grad_norm": 5.449374675750732,
      "learning_rate": 2.834211219383724e-05,
      "loss": 0.9839,
      "step": 16457
    },
    {
      "epoch": 0.43344745851988414,
      "grad_norm": 2.3553080558776855,
      "learning_rate": 2.8340795364761658e-05,
      "loss": 1.5497,
      "step": 16458
    },
    {
      "epoch": 0.43347379510139583,
      "grad_norm": 2.1973652839660645,
      "learning_rate": 2.833947853568607e-05,
      "loss": 1.4848,
      "step": 16459
    },
    {
      "epoch": 0.4335001316829076,
      "grad_norm": 1.6975960731506348,
      "learning_rate": 2.8338161706610483e-05,
      "loss": 0.3975,
      "step": 16460
    },
    {
      "epoch": 0.43352646826441926,
      "grad_norm": 2.5517799854278564,
      "learning_rate": 2.8336844877534895e-05,
      "loss": 1.1243,
      "step": 16461
    },
    {
      "epoch": 0.433552804845931,
      "grad_norm": 2.489300012588501,
      "learning_rate": 2.833552804845931e-05,
      "loss": 1.3312,
      "step": 16462
    },
    {
      "epoch": 0.4335791414274427,
      "grad_norm": 3.0198323726654053,
      "learning_rate": 2.8334211219383726e-05,
      "loss": 1.0234,
      "step": 16463
    },
    {
      "epoch": 0.43360547800895444,
      "grad_norm": 3.0601694583892822,
      "learning_rate": 2.8332894390308142e-05,
      "loss": 1.3195,
      "step": 16464
    },
    {
      "epoch": 0.4336318145904662,
      "grad_norm": 1.7922035455703735,
      "learning_rate": 2.8331577561232554e-05,
      "loss": 1.2245,
      "step": 16465
    },
    {
      "epoch": 0.43365815117197787,
      "grad_norm": 3.8335766792297363,
      "learning_rate": 2.8330260732156966e-05,
      "loss": 1.001,
      "step": 16466
    },
    {
      "epoch": 0.4336844877534896,
      "grad_norm": 2.2135560512542725,
      "learning_rate": 2.8328943903081385e-05,
      "loss": 0.287,
      "step": 16467
    },
    {
      "epoch": 0.4337108243350013,
      "grad_norm": 1.8694093227386475,
      "learning_rate": 2.8327627074005797e-05,
      "loss": 1.7988,
      "step": 16468
    },
    {
      "epoch": 0.43373716091651304,
      "grad_norm": 1.8954025506973267,
      "learning_rate": 2.832631024493021e-05,
      "loss": 1.6016,
      "step": 16469
    },
    {
      "epoch": 0.43376349749802473,
      "grad_norm": 3.6818227767944336,
      "learning_rate": 2.8324993415854622e-05,
      "loss": 1.608,
      "step": 16470
    },
    {
      "epoch": 0.4337898340795365,
      "grad_norm": 3.071455240249634,
      "learning_rate": 2.8323676586779034e-05,
      "loss": 1.9356,
      "step": 16471
    },
    {
      "epoch": 0.4338161706610482,
      "grad_norm": 2.14438533782959,
      "learning_rate": 2.8322359757703453e-05,
      "loss": 2.3003,
      "step": 16472
    },
    {
      "epoch": 0.4338425072425599,
      "grad_norm": 6.42063045501709,
      "learning_rate": 2.8321042928627865e-05,
      "loss": 1.4109,
      "step": 16473
    },
    {
      "epoch": 0.43386884382407165,
      "grad_norm": 2.8539631366729736,
      "learning_rate": 2.831972609955228e-05,
      "loss": 1.4741,
      "step": 16474
    },
    {
      "epoch": 0.43389518040558334,
      "grad_norm": 2.316908359527588,
      "learning_rate": 2.8318409270476693e-05,
      "loss": 1.9613,
      "step": 16475
    },
    {
      "epoch": 0.4339215169870951,
      "grad_norm": 3.846804618835449,
      "learning_rate": 2.8317092441401105e-05,
      "loss": 0.8859,
      "step": 16476
    },
    {
      "epoch": 0.43394785356860677,
      "grad_norm": 3.548475503921509,
      "learning_rate": 2.8315775612325524e-05,
      "loss": 1.6674,
      "step": 16477
    },
    {
      "epoch": 0.4339741901501185,
      "grad_norm": 3.4965856075286865,
      "learning_rate": 2.8314458783249936e-05,
      "loss": 0.9479,
      "step": 16478
    },
    {
      "epoch": 0.43400052673163025,
      "grad_norm": 3.1194825172424316,
      "learning_rate": 2.831314195417435e-05,
      "loss": 1.5284,
      "step": 16479
    },
    {
      "epoch": 0.43402686331314194,
      "grad_norm": 1.6658433675765991,
      "learning_rate": 2.831182512509876e-05,
      "loss": 1.9135,
      "step": 16480
    },
    {
      "epoch": 0.4340531998946537,
      "grad_norm": 2.425234317779541,
      "learning_rate": 2.8310508296023176e-05,
      "loss": 1.8135,
      "step": 16481
    },
    {
      "epoch": 0.4340795364761654,
      "grad_norm": 1.751568078994751,
      "learning_rate": 2.8309191466947592e-05,
      "loss": 1.6672,
      "step": 16482
    },
    {
      "epoch": 0.4341058730576771,
      "grad_norm": 1.8724992275238037,
      "learning_rate": 2.8307874637872008e-05,
      "loss": 1.8876,
      "step": 16483
    },
    {
      "epoch": 0.43413220963918886,
      "grad_norm": 2.105757236480713,
      "learning_rate": 2.830655780879642e-05,
      "loss": 1.536,
      "step": 16484
    },
    {
      "epoch": 0.43415854622070055,
      "grad_norm": 2.004755735397339,
      "learning_rate": 2.8305240979720832e-05,
      "loss": 1.9267,
      "step": 16485
    },
    {
      "epoch": 0.4341848828022123,
      "grad_norm": 1.4966754913330078,
      "learning_rate": 2.830392415064525e-05,
      "loss": 1.4707,
      "step": 16486
    },
    {
      "epoch": 0.434211219383724,
      "grad_norm": 2.4996042251586914,
      "learning_rate": 2.8302607321569663e-05,
      "loss": 1.3735,
      "step": 16487
    },
    {
      "epoch": 0.4342375559652357,
      "grad_norm": 2.1456947326660156,
      "learning_rate": 2.8301290492494075e-05,
      "loss": 1.7783,
      "step": 16488
    },
    {
      "epoch": 0.4342638925467474,
      "grad_norm": 1.7817798852920532,
      "learning_rate": 2.8299973663418488e-05,
      "loss": 1.7208,
      "step": 16489
    },
    {
      "epoch": 0.43429022912825915,
      "grad_norm": 1.697743535041809,
      "learning_rate": 2.82986568343429e-05,
      "loss": 1.6933,
      "step": 16490
    },
    {
      "epoch": 0.4343165657097709,
      "grad_norm": 4.356972694396973,
      "learning_rate": 2.829734000526732e-05,
      "loss": 1.722,
      "step": 16491
    },
    {
      "epoch": 0.4343429022912826,
      "grad_norm": 2.211265802383423,
      "learning_rate": 2.8296023176191734e-05,
      "loss": 1.6289,
      "step": 16492
    },
    {
      "epoch": 0.4343692388727943,
      "grad_norm": 2.077662706375122,
      "learning_rate": 2.8294706347116147e-05,
      "loss": 1.8589,
      "step": 16493
    },
    {
      "epoch": 0.434395575454306,
      "grad_norm": 1.6653528213500977,
      "learning_rate": 2.829338951804056e-05,
      "loss": 1.6447,
      "step": 16494
    },
    {
      "epoch": 0.43442191203581776,
      "grad_norm": 2.913416862487793,
      "learning_rate": 2.829207268896497e-05,
      "loss": 1.32,
      "step": 16495
    },
    {
      "epoch": 0.43444824861732945,
      "grad_norm": 2.471480369567871,
      "learning_rate": 2.829075585988939e-05,
      "loss": 2.0304,
      "step": 16496
    },
    {
      "epoch": 0.4344745851988412,
      "grad_norm": 1.7987360954284668,
      "learning_rate": 2.8289439030813802e-05,
      "loss": 2.0854,
      "step": 16497
    },
    {
      "epoch": 0.43450092178035293,
      "grad_norm": 1.8381974697113037,
      "learning_rate": 2.8288122201738214e-05,
      "loss": 1.6529,
      "step": 16498
    },
    {
      "epoch": 0.4345272583618646,
      "grad_norm": 1.5506705045700073,
      "learning_rate": 2.8286805372662627e-05,
      "loss": 1.3963,
      "step": 16499
    },
    {
      "epoch": 0.43455359494337636,
      "grad_norm": 2.0843544006347656,
      "learning_rate": 2.8285488543587046e-05,
      "loss": 1.1241,
      "step": 16500
    },
    {
      "epoch": 0.43457993152488805,
      "grad_norm": 2.3446781635284424,
      "learning_rate": 2.8284171714511458e-05,
      "loss": 1.8986,
      "step": 16501
    },
    {
      "epoch": 0.4346062681063998,
      "grad_norm": 4.117410659790039,
      "learning_rate": 2.8282854885435873e-05,
      "loss": 1.0084,
      "step": 16502
    },
    {
      "epoch": 0.4346326046879115,
      "grad_norm": 4.413883686065674,
      "learning_rate": 2.8281538056360286e-05,
      "loss": 0.7021,
      "step": 16503
    },
    {
      "epoch": 0.4346589412694232,
      "grad_norm": 1.922536015510559,
      "learning_rate": 2.8280221227284698e-05,
      "loss": 2.0783,
      "step": 16504
    },
    {
      "epoch": 0.43468527785093497,
      "grad_norm": 1.968959927558899,
      "learning_rate": 2.8278904398209117e-05,
      "loss": 1.6969,
      "step": 16505
    },
    {
      "epoch": 0.43471161443244666,
      "grad_norm": 2.5623135566711426,
      "learning_rate": 2.827758756913353e-05,
      "loss": 1.3383,
      "step": 16506
    },
    {
      "epoch": 0.4347379510139584,
      "grad_norm": 2.795820951461792,
      "learning_rate": 2.827627074005794e-05,
      "loss": 1.6475,
      "step": 16507
    },
    {
      "epoch": 0.4347642875954701,
      "grad_norm": 2.042825698852539,
      "learning_rate": 2.8274953910982353e-05,
      "loss": 2.4634,
      "step": 16508
    },
    {
      "epoch": 0.43479062417698183,
      "grad_norm": 2.9703381061553955,
      "learning_rate": 2.827363708190677e-05,
      "loss": 1.4563,
      "step": 16509
    },
    {
      "epoch": 0.4348169607584935,
      "grad_norm": 1.5627120733261108,
      "learning_rate": 2.8272320252831185e-05,
      "loss": 0.5622,
      "step": 16510
    },
    {
      "epoch": 0.43484329734000526,
      "grad_norm": 2.4037153720855713,
      "learning_rate": 2.82710034237556e-05,
      "loss": 1.7923,
      "step": 16511
    },
    {
      "epoch": 0.434869633921517,
      "grad_norm": 3.218648672103882,
      "learning_rate": 2.8269686594680012e-05,
      "loss": 0.837,
      "step": 16512
    },
    {
      "epoch": 0.4348959705030287,
      "grad_norm": 2.02287220954895,
      "learning_rate": 2.8268369765604425e-05,
      "loss": 1.9907,
      "step": 16513
    },
    {
      "epoch": 0.43492230708454044,
      "grad_norm": 1.8718514442443848,
      "learning_rate": 2.8267052936528837e-05,
      "loss": 1.9464,
      "step": 16514
    },
    {
      "epoch": 0.4349486436660521,
      "grad_norm": 1.9429402351379395,
      "learning_rate": 2.8265736107453256e-05,
      "loss": 1.8846,
      "step": 16515
    },
    {
      "epoch": 0.43497498024756387,
      "grad_norm": 1.9777008295059204,
      "learning_rate": 2.8264419278377668e-05,
      "loss": 1.9679,
      "step": 16516
    },
    {
      "epoch": 0.4350013168290756,
      "grad_norm": 2.007272958755493,
      "learning_rate": 2.826310244930208e-05,
      "loss": 2.0907,
      "step": 16517
    },
    {
      "epoch": 0.4350276534105873,
      "grad_norm": 2.2078537940979004,
      "learning_rate": 2.8261785620226492e-05,
      "loss": 2.3834,
      "step": 16518
    },
    {
      "epoch": 0.43505398999209904,
      "grad_norm": 4.378069877624512,
      "learning_rate": 2.826046879115091e-05,
      "loss": 1.8814,
      "step": 16519
    },
    {
      "epoch": 0.43508032657361073,
      "grad_norm": 2.3533949851989746,
      "learning_rate": 2.8259151962075324e-05,
      "loss": 1.5736,
      "step": 16520
    },
    {
      "epoch": 0.4351066631551225,
      "grad_norm": 2.167644739151001,
      "learning_rate": 2.825783513299974e-05,
      "loss": 0.3618,
      "step": 16521
    },
    {
      "epoch": 0.43513299973663416,
      "grad_norm": 2.1816017627716064,
      "learning_rate": 2.825651830392415e-05,
      "loss": 1.7769,
      "step": 16522
    },
    {
      "epoch": 0.4351593363181459,
      "grad_norm": 1.5967211723327637,
      "learning_rate": 2.8255201474848564e-05,
      "loss": 0.2387,
      "step": 16523
    },
    {
      "epoch": 0.43518567289965765,
      "grad_norm": 1.6066944599151611,
      "learning_rate": 2.8253884645772983e-05,
      "loss": 2.1092,
      "step": 16524
    },
    {
      "epoch": 0.43521200948116934,
      "grad_norm": 2.3991541862487793,
      "learning_rate": 2.8252567816697395e-05,
      "loss": 0.4484,
      "step": 16525
    },
    {
      "epoch": 0.4352383460626811,
      "grad_norm": 1.8292269706726074,
      "learning_rate": 2.8251250987621807e-05,
      "loss": 1.7973,
      "step": 16526
    },
    {
      "epoch": 0.43526468264419277,
      "grad_norm": 1.5651960372924805,
      "learning_rate": 2.824993415854622e-05,
      "loss": 1.4939,
      "step": 16527
    },
    {
      "epoch": 0.4352910192257045,
      "grad_norm": 4.2878098487854,
      "learning_rate": 2.8248617329470635e-05,
      "loss": 0.4767,
      "step": 16528
    },
    {
      "epoch": 0.4353173558072162,
      "grad_norm": 2.1620655059814453,
      "learning_rate": 2.824730050039505e-05,
      "loss": 1.6215,
      "step": 16529
    },
    {
      "epoch": 0.43534369238872794,
      "grad_norm": 3.5705878734588623,
      "learning_rate": 2.8245983671319466e-05,
      "loss": 1.4668,
      "step": 16530
    },
    {
      "epoch": 0.4353700289702397,
      "grad_norm": 4.136373519897461,
      "learning_rate": 2.824466684224388e-05,
      "loss": 1.3312,
      "step": 16531
    },
    {
      "epoch": 0.4353963655517514,
      "grad_norm": 4.730039119720459,
      "learning_rate": 2.824335001316829e-05,
      "loss": 0.9081,
      "step": 16532
    },
    {
      "epoch": 0.4354227021332631,
      "grad_norm": 2.257317304611206,
      "learning_rate": 2.824203318409271e-05,
      "loss": 1.7244,
      "step": 16533
    },
    {
      "epoch": 0.4354490387147748,
      "grad_norm": 3.972649097442627,
      "learning_rate": 2.8240716355017122e-05,
      "loss": 1.5251,
      "step": 16534
    },
    {
      "epoch": 0.43547537529628655,
      "grad_norm": 2.4832606315612793,
      "learning_rate": 2.8239399525941534e-05,
      "loss": 2.2719,
      "step": 16535
    },
    {
      "epoch": 0.43550171187779824,
      "grad_norm": 1.695595622062683,
      "learning_rate": 2.8238082696865946e-05,
      "loss": 1.4979,
      "step": 16536
    },
    {
      "epoch": 0.43552804845931,
      "grad_norm": 2.467546224594116,
      "learning_rate": 2.8236765867790362e-05,
      "loss": 1.3863,
      "step": 16537
    },
    {
      "epoch": 0.4355543850408217,
      "grad_norm": 1.8082926273345947,
      "learning_rate": 2.8235449038714777e-05,
      "loss": 0.4872,
      "step": 16538
    },
    {
      "epoch": 0.4355807216223334,
      "grad_norm": 2.463932991027832,
      "learning_rate": 2.8234132209639193e-05,
      "loss": 2.1371,
      "step": 16539
    },
    {
      "epoch": 0.43560705820384515,
      "grad_norm": 2.318068027496338,
      "learning_rate": 2.8232815380563605e-05,
      "loss": 0.4452,
      "step": 16540
    },
    {
      "epoch": 0.43563339478535684,
      "grad_norm": 1.7529547214508057,
      "learning_rate": 2.8231498551488017e-05,
      "loss": 1.4515,
      "step": 16541
    },
    {
      "epoch": 0.4356597313668686,
      "grad_norm": 3.1177496910095215,
      "learning_rate": 2.823018172241243e-05,
      "loss": 2.0095,
      "step": 16542
    },
    {
      "epoch": 0.4356860679483803,
      "grad_norm": 2.89117169380188,
      "learning_rate": 2.822886489333685e-05,
      "loss": 1.6061,
      "step": 16543
    },
    {
      "epoch": 0.435712404529892,
      "grad_norm": 3.2339375019073486,
      "learning_rate": 2.822754806426126e-05,
      "loss": 1.7907,
      "step": 16544
    },
    {
      "epoch": 0.43573874111140376,
      "grad_norm": 2.899411916732788,
      "learning_rate": 2.8226231235185673e-05,
      "loss": 1.3294,
      "step": 16545
    },
    {
      "epoch": 0.43576507769291545,
      "grad_norm": 1.6597256660461426,
      "learning_rate": 2.8224914406110085e-05,
      "loss": 1.0791,
      "step": 16546
    },
    {
      "epoch": 0.4357914142744272,
      "grad_norm": 5.410963535308838,
      "learning_rate": 2.82235975770345e-05,
      "loss": 0.7657,
      "step": 16547
    },
    {
      "epoch": 0.4358177508559389,
      "grad_norm": 1.9413758516311646,
      "learning_rate": 2.8222280747958916e-05,
      "loss": 2.031,
      "step": 16548
    },
    {
      "epoch": 0.4358440874374506,
      "grad_norm": 1.704094648361206,
      "learning_rate": 2.8220963918883332e-05,
      "loss": 1.775,
      "step": 16549
    },
    {
      "epoch": 0.43587042401896237,
      "grad_norm": 1.5150890350341797,
      "learning_rate": 2.8219647089807744e-05,
      "loss": 1.7933,
      "step": 16550
    },
    {
      "epoch": 0.43589676060047405,
      "grad_norm": 2.559528350830078,
      "learning_rate": 2.8218330260732156e-05,
      "loss": 1.7279,
      "step": 16551
    },
    {
      "epoch": 0.4359230971819858,
      "grad_norm": 2.9103341102600098,
      "learning_rate": 2.8217013431656575e-05,
      "loss": 2.1717,
      "step": 16552
    },
    {
      "epoch": 0.4359494337634975,
      "grad_norm": 3.6804163455963135,
      "learning_rate": 2.8215696602580988e-05,
      "loss": 1.9862,
      "step": 16553
    },
    {
      "epoch": 0.43597577034500923,
      "grad_norm": 1.056111216545105,
      "learning_rate": 2.82143797735054e-05,
      "loss": 0.2084,
      "step": 16554
    },
    {
      "epoch": 0.4360021069265209,
      "grad_norm": 5.031483173370361,
      "learning_rate": 2.8213062944429812e-05,
      "loss": 1.4148,
      "step": 16555
    },
    {
      "epoch": 0.43602844350803266,
      "grad_norm": 1.9605884552001953,
      "learning_rate": 2.8211746115354228e-05,
      "loss": 1.6704,
      "step": 16556
    },
    {
      "epoch": 0.4360547800895444,
      "grad_norm": 1.915305733680725,
      "learning_rate": 2.8210429286278643e-05,
      "loss": 1.7103,
      "step": 16557
    },
    {
      "epoch": 0.4360811166710561,
      "grad_norm": 1.579426646232605,
      "learning_rate": 2.820911245720306e-05,
      "loss": 1.5001,
      "step": 16558
    },
    {
      "epoch": 0.43610745325256783,
      "grad_norm": 1.9474209547042847,
      "learning_rate": 2.820779562812747e-05,
      "loss": 2.1016,
      "step": 16559
    },
    {
      "epoch": 0.4361337898340795,
      "grad_norm": 1.8780663013458252,
      "learning_rate": 2.8206478799051883e-05,
      "loss": 1.8185,
      "step": 16560
    },
    {
      "epoch": 0.43616012641559126,
      "grad_norm": 2.063614845275879,
      "learning_rate": 2.8205161969976295e-05,
      "loss": 1.955,
      "step": 16561
    },
    {
      "epoch": 0.43618646299710295,
      "grad_norm": 2.269077777862549,
      "learning_rate": 2.8203845140900714e-05,
      "loss": 2.2423,
      "step": 16562
    },
    {
      "epoch": 0.4362127995786147,
      "grad_norm": 1.7708632946014404,
      "learning_rate": 2.8202528311825127e-05,
      "loss": 1.8232,
      "step": 16563
    },
    {
      "epoch": 0.43623913616012644,
      "grad_norm": 1.8033145666122437,
      "learning_rate": 2.820121148274954e-05,
      "loss": 1.59,
      "step": 16564
    },
    {
      "epoch": 0.4362654727416381,
      "grad_norm": 4.104933738708496,
      "learning_rate": 2.8199894653673954e-05,
      "loss": 0.6668,
      "step": 16565
    },
    {
      "epoch": 0.43629180932314987,
      "grad_norm": 3.9449446201324463,
      "learning_rate": 2.819857782459837e-05,
      "loss": 1.6086,
      "step": 16566
    },
    {
      "epoch": 0.43631814590466156,
      "grad_norm": 2.564462184906006,
      "learning_rate": 2.8197260995522786e-05,
      "loss": 1.0605,
      "step": 16567
    },
    {
      "epoch": 0.4363444824861733,
      "grad_norm": 1.6224631071090698,
      "learning_rate": 2.8195944166447198e-05,
      "loss": 1.6911,
      "step": 16568
    },
    {
      "epoch": 0.436370819067685,
      "grad_norm": 3.5223100185394287,
      "learning_rate": 2.819462733737161e-05,
      "loss": 1.7488,
      "step": 16569
    },
    {
      "epoch": 0.43639715564919673,
      "grad_norm": 3.0731112957000732,
      "learning_rate": 2.8193310508296022e-05,
      "loss": 1.5548,
      "step": 16570
    },
    {
      "epoch": 0.4364234922307085,
      "grad_norm": 3.0118627548217773,
      "learning_rate": 2.819199367922044e-05,
      "loss": 1.6599,
      "step": 16571
    },
    {
      "epoch": 0.43644982881222016,
      "grad_norm": 2.1013436317443848,
      "learning_rate": 2.8190676850144853e-05,
      "loss": 2.0399,
      "step": 16572
    },
    {
      "epoch": 0.4364761653937319,
      "grad_norm": 1.8144354820251465,
      "learning_rate": 2.8189360021069266e-05,
      "loss": 1.8986,
      "step": 16573
    },
    {
      "epoch": 0.4365025019752436,
      "grad_norm": 2.1434435844421387,
      "learning_rate": 2.8188043191993678e-05,
      "loss": 1.9027,
      "step": 16574
    },
    {
      "epoch": 0.43652883855675534,
      "grad_norm": 3.2731213569641113,
      "learning_rate": 2.8186726362918093e-05,
      "loss": 0.3589,
      "step": 16575
    },
    {
      "epoch": 0.436555175138267,
      "grad_norm": 1.8018608093261719,
      "learning_rate": 2.818540953384251e-05,
      "loss": 2.1207,
      "step": 16576
    },
    {
      "epoch": 0.43658151171977877,
      "grad_norm": 2.136406183242798,
      "learning_rate": 2.8184092704766925e-05,
      "loss": 1.8915,
      "step": 16577
    },
    {
      "epoch": 0.4366078483012905,
      "grad_norm": 1.9887404441833496,
      "learning_rate": 2.8182775875691337e-05,
      "loss": 2.1259,
      "step": 16578
    },
    {
      "epoch": 0.4366341848828022,
      "grad_norm": 3.9231603145599365,
      "learning_rate": 2.818145904661575e-05,
      "loss": 1.4021,
      "step": 16579
    },
    {
      "epoch": 0.43666052146431394,
      "grad_norm": 1.7362685203552246,
      "learning_rate": 2.818014221754016e-05,
      "loss": 0.2168,
      "step": 16580
    },
    {
      "epoch": 0.43668685804582563,
      "grad_norm": 1.8164308071136475,
      "learning_rate": 2.817882538846458e-05,
      "loss": 1.7385,
      "step": 16581
    },
    {
      "epoch": 0.4367131946273374,
      "grad_norm": 2.889979362487793,
      "learning_rate": 2.8177508559388992e-05,
      "loss": 2.1276,
      "step": 16582
    },
    {
      "epoch": 0.4367395312088491,
      "grad_norm": 1.8994073867797852,
      "learning_rate": 2.8176191730313405e-05,
      "loss": 1.5463,
      "step": 16583
    },
    {
      "epoch": 0.4367658677903608,
      "grad_norm": 2.7641348838806152,
      "learning_rate": 2.817487490123782e-05,
      "loss": 0.8215,
      "step": 16584
    },
    {
      "epoch": 0.43679220437187255,
      "grad_norm": 3.9808731079101562,
      "learning_rate": 2.8173558072162236e-05,
      "loss": 1.3807,
      "step": 16585
    },
    {
      "epoch": 0.43681854095338424,
      "grad_norm": 3.0710442066192627,
      "learning_rate": 2.817224124308665e-05,
      "loss": 0.8945,
      "step": 16586
    },
    {
      "epoch": 0.436844877534896,
      "grad_norm": 2.3296892642974854,
      "learning_rate": 2.8170924414011064e-05,
      "loss": 0.7532,
      "step": 16587
    },
    {
      "epoch": 0.43687121411640767,
      "grad_norm": 2.7271907329559326,
      "learning_rate": 2.8169607584935476e-05,
      "loss": 1.5595,
      "step": 16588
    },
    {
      "epoch": 0.4368975506979194,
      "grad_norm": 3.0335512161254883,
      "learning_rate": 2.8168290755859888e-05,
      "loss": 0.4242,
      "step": 16589
    },
    {
      "epoch": 0.43692388727943116,
      "grad_norm": 2.5800588130950928,
      "learning_rate": 2.8166973926784307e-05,
      "loss": 1.4019,
      "step": 16590
    },
    {
      "epoch": 0.43695022386094284,
      "grad_norm": 2.7184219360351562,
      "learning_rate": 2.816565709770872e-05,
      "loss": 1.6803,
      "step": 16591
    },
    {
      "epoch": 0.4369765604424546,
      "grad_norm": 2.4640495777130127,
      "learning_rate": 2.816434026863313e-05,
      "loss": 1.7125,
      "step": 16592
    },
    {
      "epoch": 0.4370028970239663,
      "grad_norm": 2.405949592590332,
      "learning_rate": 2.8163023439557544e-05,
      "loss": 1.9497,
      "step": 16593
    },
    {
      "epoch": 0.437029233605478,
      "grad_norm": 2.864454984664917,
      "learning_rate": 2.816170661048196e-05,
      "loss": 2.098,
      "step": 16594
    },
    {
      "epoch": 0.4370555701869897,
      "grad_norm": 3.776113271713257,
      "learning_rate": 2.8160389781406375e-05,
      "loss": 0.7326,
      "step": 16595
    },
    {
      "epoch": 0.43708190676850145,
      "grad_norm": 2.1270062923431396,
      "learning_rate": 2.815907295233079e-05,
      "loss": 1.8915,
      "step": 16596
    },
    {
      "epoch": 0.4371082433500132,
      "grad_norm": 2.74699330329895,
      "learning_rate": 2.8157756123255203e-05,
      "loss": 1.8578,
      "step": 16597
    },
    {
      "epoch": 0.4371345799315249,
      "grad_norm": 1.6508920192718506,
      "learning_rate": 2.8156439294179615e-05,
      "loss": 1.7382,
      "step": 16598
    },
    {
      "epoch": 0.4371609165130366,
      "grad_norm": 1.529170274734497,
      "learning_rate": 2.8155122465104034e-05,
      "loss": 1.9671,
      "step": 16599
    },
    {
      "epoch": 0.4371872530945483,
      "grad_norm": 1.6650422811508179,
      "learning_rate": 2.8153805636028446e-05,
      "loss": 1.6926,
      "step": 16600
    },
    {
      "epoch": 0.43721358967606005,
      "grad_norm": 2.509626626968384,
      "learning_rate": 2.815248880695286e-05,
      "loss": 1.7362,
      "step": 16601
    },
    {
      "epoch": 0.43723992625757174,
      "grad_norm": 1.6635472774505615,
      "learning_rate": 2.815117197787727e-05,
      "loss": 1.7993,
      "step": 16602
    },
    {
      "epoch": 0.4372662628390835,
      "grad_norm": 2.379213571548462,
      "learning_rate": 2.8149855148801686e-05,
      "loss": 1.6131,
      "step": 16603
    },
    {
      "epoch": 0.43729259942059523,
      "grad_norm": 3.236212968826294,
      "learning_rate": 2.8148538319726102e-05,
      "loss": 1.7876,
      "step": 16604
    },
    {
      "epoch": 0.4373189360021069,
      "grad_norm": 1.8496614694595337,
      "learning_rate": 2.8147221490650517e-05,
      "loss": 1.8803,
      "step": 16605
    },
    {
      "epoch": 0.43734527258361866,
      "grad_norm": 2.900906562805176,
      "learning_rate": 2.814590466157493e-05,
      "loss": 1.5396,
      "step": 16606
    },
    {
      "epoch": 0.43737160916513035,
      "grad_norm": 3.22273325920105,
      "learning_rate": 2.8144587832499342e-05,
      "loss": 0.3887,
      "step": 16607
    },
    {
      "epoch": 0.4373979457466421,
      "grad_norm": 1.628406286239624,
      "learning_rate": 2.8143271003423754e-05,
      "loss": 1.5892,
      "step": 16608
    },
    {
      "epoch": 0.4374242823281538,
      "grad_norm": 3.8181023597717285,
      "learning_rate": 2.8141954174348173e-05,
      "loss": 1.5064,
      "step": 16609
    },
    {
      "epoch": 0.4374506189096655,
      "grad_norm": 2.1755239963531494,
      "learning_rate": 2.8140637345272585e-05,
      "loss": 1.5169,
      "step": 16610
    },
    {
      "epoch": 0.43747695549117727,
      "grad_norm": 2.7795848846435547,
      "learning_rate": 2.8139320516196997e-05,
      "loss": 1.6898,
      "step": 16611
    },
    {
      "epoch": 0.43750329207268895,
      "grad_norm": 1.8404885530471802,
      "learning_rate": 2.8138003687121413e-05,
      "loss": 1.7695,
      "step": 16612
    },
    {
      "epoch": 0.4375296286542007,
      "grad_norm": 2.442927360534668,
      "learning_rate": 2.8136686858045825e-05,
      "loss": 1.8158,
      "step": 16613
    },
    {
      "epoch": 0.4375559652357124,
      "grad_norm": 2.6817586421966553,
      "learning_rate": 2.8135370028970244e-05,
      "loss": 1.2455,
      "step": 16614
    },
    {
      "epoch": 0.43758230181722413,
      "grad_norm": 2.126110553741455,
      "learning_rate": 2.8134053199894656e-05,
      "loss": 1.7892,
      "step": 16615
    },
    {
      "epoch": 0.4376086383987358,
      "grad_norm": 1.8690754175186157,
      "learning_rate": 2.813273637081907e-05,
      "loss": 1.7856,
      "step": 16616
    },
    {
      "epoch": 0.43763497498024756,
      "grad_norm": 1.87594735622406,
      "learning_rate": 2.813141954174348e-05,
      "loss": 1.724,
      "step": 16617
    },
    {
      "epoch": 0.4376613115617593,
      "grad_norm": 3.224095106124878,
      "learning_rate": 2.81301027126679e-05,
      "loss": 1.1844,
      "step": 16618
    },
    {
      "epoch": 0.437687648143271,
      "grad_norm": 2.0419390201568604,
      "learning_rate": 2.8128785883592312e-05,
      "loss": 1.6161,
      "step": 16619
    },
    {
      "epoch": 0.43771398472478273,
      "grad_norm": 1.8997228145599365,
      "learning_rate": 2.8127469054516724e-05,
      "loss": 1.7896,
      "step": 16620
    },
    {
      "epoch": 0.4377403213062944,
      "grad_norm": 2.3550984859466553,
      "learning_rate": 2.8126152225441136e-05,
      "loss": 2.0752,
      "step": 16621
    },
    {
      "epoch": 0.43776665788780617,
      "grad_norm": 1.678864598274231,
      "learning_rate": 2.8124835396365552e-05,
      "loss": 2.3086,
      "step": 16622
    },
    {
      "epoch": 0.4377929944693179,
      "grad_norm": 1.7242493629455566,
      "learning_rate": 2.8123518567289968e-05,
      "loss": 1.5141,
      "step": 16623
    },
    {
      "epoch": 0.4378193310508296,
      "grad_norm": 1.6488767862319946,
      "learning_rate": 2.8122201738214383e-05,
      "loss": 2.1585,
      "step": 16624
    },
    {
      "epoch": 0.43784566763234134,
      "grad_norm": 1.7621309757232666,
      "learning_rate": 2.8120884909138795e-05,
      "loss": 1.5271,
      "step": 16625
    },
    {
      "epoch": 0.437872004213853,
      "grad_norm": 1.4883825778961182,
      "learning_rate": 2.8119568080063208e-05,
      "loss": 1.9442,
      "step": 16626
    },
    {
      "epoch": 0.43789834079536477,
      "grad_norm": 3.7244820594787598,
      "learning_rate": 2.811825125098762e-05,
      "loss": 1.4252,
      "step": 16627
    },
    {
      "epoch": 0.43792467737687646,
      "grad_norm": 3.383676767349243,
      "learning_rate": 2.811693442191204e-05,
      "loss": 1.572,
      "step": 16628
    },
    {
      "epoch": 0.4379510139583882,
      "grad_norm": 2.4104089736938477,
      "learning_rate": 2.811561759283645e-05,
      "loss": 1.9732,
      "step": 16629
    },
    {
      "epoch": 0.43797735053989995,
      "grad_norm": 1.6271495819091797,
      "learning_rate": 2.8114300763760863e-05,
      "loss": 1.367,
      "step": 16630
    },
    {
      "epoch": 0.43800368712141163,
      "grad_norm": 1.797842025756836,
      "learning_rate": 2.811298393468528e-05,
      "loss": 2.0346,
      "step": 16631
    },
    {
      "epoch": 0.4380300237029234,
      "grad_norm": 3.5027220249176025,
      "learning_rate": 2.8111667105609694e-05,
      "loss": 1.8181,
      "step": 16632
    },
    {
      "epoch": 0.43805636028443506,
      "grad_norm": 2.5124311447143555,
      "learning_rate": 2.811035027653411e-05,
      "loss": 2.122,
      "step": 16633
    },
    {
      "epoch": 0.4380826968659468,
      "grad_norm": 4.779937744140625,
      "learning_rate": 2.8109033447458522e-05,
      "loss": 0.5993,
      "step": 16634
    },
    {
      "epoch": 0.4381090334474585,
      "grad_norm": 1.7733968496322632,
      "learning_rate": 2.8107716618382934e-05,
      "loss": 0.4342,
      "step": 16635
    },
    {
      "epoch": 0.43813537002897024,
      "grad_norm": 3.3720638751983643,
      "learning_rate": 2.8106399789307347e-05,
      "loss": 2.2293,
      "step": 16636
    },
    {
      "epoch": 0.438161706610482,
      "grad_norm": 1.5262728929519653,
      "learning_rate": 2.8105082960231766e-05,
      "loss": 1.5734,
      "step": 16637
    },
    {
      "epoch": 0.43818804319199367,
      "grad_norm": 1.7679376602172852,
      "learning_rate": 2.8103766131156178e-05,
      "loss": 1.1915,
      "step": 16638
    },
    {
      "epoch": 0.4382143797735054,
      "grad_norm": 3.353997230529785,
      "learning_rate": 2.810244930208059e-05,
      "loss": 1.4492,
      "step": 16639
    },
    {
      "epoch": 0.4382407163550171,
      "grad_norm": 1.382959008216858,
      "learning_rate": 2.8101132473005006e-05,
      "loss": 0.346,
      "step": 16640
    },
    {
      "epoch": 0.43826705293652884,
      "grad_norm": 3.6660995483398438,
      "learning_rate": 2.8099815643929418e-05,
      "loss": 1.2867,
      "step": 16641
    },
    {
      "epoch": 0.43829338951804053,
      "grad_norm": 1.4647947549819946,
      "learning_rate": 2.8098498814853837e-05,
      "loss": 1.975,
      "step": 16642
    },
    {
      "epoch": 0.4383197260995523,
      "grad_norm": 1.6251085996627808,
      "learning_rate": 2.809718198577825e-05,
      "loss": 2.3856,
      "step": 16643
    },
    {
      "epoch": 0.438346062681064,
      "grad_norm": 1.5047577619552612,
      "learning_rate": 2.809586515670266e-05,
      "loss": 1.6583,
      "step": 16644
    },
    {
      "epoch": 0.4383723992625757,
      "grad_norm": 4.208713054656982,
      "learning_rate": 2.8094548327627074e-05,
      "loss": 2.1076,
      "step": 16645
    },
    {
      "epoch": 0.43839873584408745,
      "grad_norm": 2.881943702697754,
      "learning_rate": 2.8093231498551486e-05,
      "loss": 1.4483,
      "step": 16646
    },
    {
      "epoch": 0.43842507242559914,
      "grad_norm": 2.7489466667175293,
      "learning_rate": 2.8091914669475905e-05,
      "loss": 1.7601,
      "step": 16647
    },
    {
      "epoch": 0.4384514090071109,
      "grad_norm": 3.5693933963775635,
      "learning_rate": 2.8090597840400317e-05,
      "loss": 0.9019,
      "step": 16648
    },
    {
      "epoch": 0.43847774558862257,
      "grad_norm": 1.6769102811813354,
      "learning_rate": 2.808928101132473e-05,
      "loss": 1.7101,
      "step": 16649
    },
    {
      "epoch": 0.4385040821701343,
      "grad_norm": 5.1734161376953125,
      "learning_rate": 2.8087964182249145e-05,
      "loss": 1.3493,
      "step": 16650
    },
    {
      "epoch": 0.43853041875164606,
      "grad_norm": 4.826345920562744,
      "learning_rate": 2.808664735317356e-05,
      "loss": 1.1355,
      "step": 16651
    },
    {
      "epoch": 0.43855675533315774,
      "grad_norm": 2.712402582168579,
      "learning_rate": 2.8085330524097976e-05,
      "loss": 2.102,
      "step": 16652
    },
    {
      "epoch": 0.4385830919146695,
      "grad_norm": 6.370478630065918,
      "learning_rate": 2.8084013695022388e-05,
      "loss": 2.3326,
      "step": 16653
    },
    {
      "epoch": 0.4386094284961812,
      "grad_norm": 2.029810905456543,
      "learning_rate": 2.80826968659468e-05,
      "loss": 1.8469,
      "step": 16654
    },
    {
      "epoch": 0.4386357650776929,
      "grad_norm": 2.1381540298461914,
      "learning_rate": 2.8081380036871213e-05,
      "loss": 1.8223,
      "step": 16655
    },
    {
      "epoch": 0.43866210165920466,
      "grad_norm": 2.176945924758911,
      "learning_rate": 2.808006320779563e-05,
      "loss": 1.6776,
      "step": 16656
    },
    {
      "epoch": 0.43868843824071635,
      "grad_norm": 3.5617611408233643,
      "learning_rate": 2.8078746378720044e-05,
      "loss": 0.9677,
      "step": 16657
    },
    {
      "epoch": 0.4387147748222281,
      "grad_norm": 3.248983383178711,
      "learning_rate": 2.8077429549644456e-05,
      "loss": 1.4099,
      "step": 16658
    },
    {
      "epoch": 0.4387411114037398,
      "grad_norm": 1.565604329109192,
      "learning_rate": 2.807611272056887e-05,
      "loss": 1.5412,
      "step": 16659
    },
    {
      "epoch": 0.4387674479852515,
      "grad_norm": 1.7660753726959229,
      "learning_rate": 2.8074795891493284e-05,
      "loss": 1.741,
      "step": 16660
    },
    {
      "epoch": 0.4387937845667632,
      "grad_norm": 2.713979721069336,
      "learning_rate": 2.8073479062417703e-05,
      "loss": 0.8884,
      "step": 16661
    },
    {
      "epoch": 0.43882012114827496,
      "grad_norm": 3.372013807296753,
      "learning_rate": 2.8072162233342115e-05,
      "loss": 1.1915,
      "step": 16662
    },
    {
      "epoch": 0.4388464577297867,
      "grad_norm": 4.073408126831055,
      "learning_rate": 2.8070845404266527e-05,
      "loss": 1.0933,
      "step": 16663
    },
    {
      "epoch": 0.4388727943112984,
      "grad_norm": 2.7456178665161133,
      "learning_rate": 2.806952857519094e-05,
      "loss": 1.3222,
      "step": 16664
    },
    {
      "epoch": 0.43889913089281013,
      "grad_norm": 2.710808277130127,
      "learning_rate": 2.806821174611536e-05,
      "loss": 1.5726,
      "step": 16665
    },
    {
      "epoch": 0.4389254674743218,
      "grad_norm": 4.102010250091553,
      "learning_rate": 2.806689491703977e-05,
      "loss": 0.1941,
      "step": 16666
    },
    {
      "epoch": 0.43895180405583356,
      "grad_norm": 2.143749713897705,
      "learning_rate": 2.8065578087964183e-05,
      "loss": 1.9777,
      "step": 16667
    },
    {
      "epoch": 0.43897814063734525,
      "grad_norm": 3.6210286617279053,
      "learning_rate": 2.80642612588886e-05,
      "loss": 1.636,
      "step": 16668
    },
    {
      "epoch": 0.439004477218857,
      "grad_norm": 1.8862544298171997,
      "learning_rate": 2.806294442981301e-05,
      "loss": 1.5848,
      "step": 16669
    },
    {
      "epoch": 0.43903081380036874,
      "grad_norm": 2.8148388862609863,
      "learning_rate": 2.806162760073743e-05,
      "loss": 1.8438,
      "step": 16670
    },
    {
      "epoch": 0.4390571503818804,
      "grad_norm": 1.6165825128555298,
      "learning_rate": 2.8060310771661842e-05,
      "loss": 1.2156,
      "step": 16671
    },
    {
      "epoch": 0.43908348696339217,
      "grad_norm": 6.481845855712891,
      "learning_rate": 2.8058993942586254e-05,
      "loss": 2.7948,
      "step": 16672
    },
    {
      "epoch": 0.43910982354490385,
      "grad_norm": 2.490011215209961,
      "learning_rate": 2.8057677113510666e-05,
      "loss": 2.065,
      "step": 16673
    },
    {
      "epoch": 0.4391361601264156,
      "grad_norm": 3.7045841217041016,
      "learning_rate": 2.805636028443508e-05,
      "loss": 2.1497,
      "step": 16674
    },
    {
      "epoch": 0.4391624967079273,
      "grad_norm": 1.612959384918213,
      "learning_rate": 2.8055043455359497e-05,
      "loss": 1.2847,
      "step": 16675
    },
    {
      "epoch": 0.43918883328943903,
      "grad_norm": 2.541501045227051,
      "learning_rate": 2.805372662628391e-05,
      "loss": 2.3188,
      "step": 16676
    },
    {
      "epoch": 0.4392151698709508,
      "grad_norm": 2.371471643447876,
      "learning_rate": 2.8052409797208322e-05,
      "loss": 1.6169,
      "step": 16677
    },
    {
      "epoch": 0.43924150645246246,
      "grad_norm": 1.4706392288208008,
      "learning_rate": 2.8051092968132737e-05,
      "loss": 0.4849,
      "step": 16678
    },
    {
      "epoch": 0.4392678430339742,
      "grad_norm": 1.6664483547210693,
      "learning_rate": 2.804977613905715e-05,
      "loss": 2.3979,
      "step": 16679
    },
    {
      "epoch": 0.4392941796154859,
      "grad_norm": 2.094409942626953,
      "learning_rate": 2.804845930998157e-05,
      "loss": 1.6646,
      "step": 16680
    },
    {
      "epoch": 0.43932051619699763,
      "grad_norm": 1.8919168710708618,
      "learning_rate": 2.804714248090598e-05,
      "loss": 1.6576,
      "step": 16681
    },
    {
      "epoch": 0.4393468527785093,
      "grad_norm": 1.5826287269592285,
      "learning_rate": 2.8045825651830393e-05,
      "loss": 1.6483,
      "step": 16682
    },
    {
      "epoch": 0.43937318936002107,
      "grad_norm": 4.1608710289001465,
      "learning_rate": 2.8044508822754805e-05,
      "loss": 1.9458,
      "step": 16683
    },
    {
      "epoch": 0.4393995259415328,
      "grad_norm": 2.2314000129699707,
      "learning_rate": 2.8043191993679224e-05,
      "loss": 1.8213,
      "step": 16684
    },
    {
      "epoch": 0.4394258625230445,
      "grad_norm": 1.8401408195495605,
      "learning_rate": 2.8041875164603636e-05,
      "loss": 1.965,
      "step": 16685
    },
    {
      "epoch": 0.43945219910455624,
      "grad_norm": 2.4395196437835693,
      "learning_rate": 2.804055833552805e-05,
      "loss": 1.1922,
      "step": 16686
    },
    {
      "epoch": 0.43947853568606793,
      "grad_norm": 2.1906843185424805,
      "learning_rate": 2.8039241506452464e-05,
      "loss": 1.9512,
      "step": 16687
    },
    {
      "epoch": 0.43950487226757967,
      "grad_norm": 2.4255566596984863,
      "learning_rate": 2.8037924677376876e-05,
      "loss": 2.1641,
      "step": 16688
    },
    {
      "epoch": 0.4395312088490914,
      "grad_norm": 2.333284854888916,
      "learning_rate": 2.8036607848301295e-05,
      "loss": 1.564,
      "step": 16689
    },
    {
      "epoch": 0.4395575454306031,
      "grad_norm": 1.7572884559631348,
      "learning_rate": 2.8035291019225708e-05,
      "loss": 1.8175,
      "step": 16690
    },
    {
      "epoch": 0.43958388201211485,
      "grad_norm": 3.4218058586120605,
      "learning_rate": 2.803397419015012e-05,
      "loss": 1.6107,
      "step": 16691
    },
    {
      "epoch": 0.43961021859362653,
      "grad_norm": 3.5521857738494873,
      "learning_rate": 2.8032657361074532e-05,
      "loss": 1.0413,
      "step": 16692
    },
    {
      "epoch": 0.4396365551751383,
      "grad_norm": 2.0270421504974365,
      "learning_rate": 2.8031340531998944e-05,
      "loss": 1.9875,
      "step": 16693
    },
    {
      "epoch": 0.43966289175664997,
      "grad_norm": 2.2825734615325928,
      "learning_rate": 2.8030023702923363e-05,
      "loss": 2.2645,
      "step": 16694
    },
    {
      "epoch": 0.4396892283381617,
      "grad_norm": 1.893298625946045,
      "learning_rate": 2.8028706873847775e-05,
      "loss": 1.9476,
      "step": 16695
    },
    {
      "epoch": 0.43971556491967345,
      "grad_norm": 2.5775043964385986,
      "learning_rate": 2.8027390044772188e-05,
      "loss": 1.6936,
      "step": 16696
    },
    {
      "epoch": 0.43974190150118514,
      "grad_norm": 1.5541459321975708,
      "learning_rate": 2.8026073215696603e-05,
      "loss": 1.87,
      "step": 16697
    },
    {
      "epoch": 0.4397682380826969,
      "grad_norm": 2.2654428482055664,
      "learning_rate": 2.802475638662102e-05,
      "loss": 1.2083,
      "step": 16698
    },
    {
      "epoch": 0.43979457466420857,
      "grad_norm": 1.7209984064102173,
      "learning_rate": 2.8023439557545434e-05,
      "loss": 1.4076,
      "step": 16699
    },
    {
      "epoch": 0.4398209112457203,
      "grad_norm": 1.937251329421997,
      "learning_rate": 2.8022122728469847e-05,
      "loss": 2.0598,
      "step": 16700
    },
    {
      "epoch": 0.439847247827232,
      "grad_norm": 2.355435371398926,
      "learning_rate": 2.802080589939426e-05,
      "loss": 1.7337,
      "step": 16701
    },
    {
      "epoch": 0.43987358440874375,
      "grad_norm": 1.8486247062683105,
      "learning_rate": 2.801948907031867e-05,
      "loss": 1.635,
      "step": 16702
    },
    {
      "epoch": 0.4398999209902555,
      "grad_norm": 2.5294671058654785,
      "learning_rate": 2.801817224124309e-05,
      "loss": 1.8208,
      "step": 16703
    },
    {
      "epoch": 0.4399262575717672,
      "grad_norm": 1.5869202613830566,
      "learning_rate": 2.8016855412167502e-05,
      "loss": 1.8462,
      "step": 16704
    },
    {
      "epoch": 0.4399525941532789,
      "grad_norm": 2.247387647628784,
      "learning_rate": 2.8015538583091915e-05,
      "loss": 2.0876,
      "step": 16705
    },
    {
      "epoch": 0.4399789307347906,
      "grad_norm": 1.6417685747146606,
      "learning_rate": 2.801422175401633e-05,
      "loss": 2.2676,
      "step": 16706
    },
    {
      "epoch": 0.44000526731630235,
      "grad_norm": 2.436558961868286,
      "learning_rate": 2.8012904924940742e-05,
      "loss": 2.796,
      "step": 16707
    },
    {
      "epoch": 0.44003160389781404,
      "grad_norm": 2.547675132751465,
      "learning_rate": 2.801158809586516e-05,
      "loss": 2.246,
      "step": 16708
    },
    {
      "epoch": 0.4400579404793258,
      "grad_norm": 3.9826619625091553,
      "learning_rate": 2.8010271266789574e-05,
      "loss": 1.3622,
      "step": 16709
    },
    {
      "epoch": 0.4400842770608375,
      "grad_norm": 2.6392602920532227,
      "learning_rate": 2.8008954437713986e-05,
      "loss": 1.5256,
      "step": 16710
    },
    {
      "epoch": 0.4401106136423492,
      "grad_norm": 3.0870163440704346,
      "learning_rate": 2.8007637608638398e-05,
      "loss": 1.3004,
      "step": 16711
    },
    {
      "epoch": 0.44013695022386096,
      "grad_norm": 3.5953996181488037,
      "learning_rate": 2.8006320779562817e-05,
      "loss": 2.0061,
      "step": 16712
    },
    {
      "epoch": 0.44016328680537264,
      "grad_norm": 1.9704854488372803,
      "learning_rate": 2.800500395048723e-05,
      "loss": 1.189,
      "step": 16713
    },
    {
      "epoch": 0.4401896233868844,
      "grad_norm": 1.8509185314178467,
      "learning_rate": 2.800368712141164e-05,
      "loss": 1.483,
      "step": 16714
    },
    {
      "epoch": 0.4402159599683961,
      "grad_norm": 1.6682113409042358,
      "learning_rate": 2.8002370292336057e-05,
      "loss": 1.6067,
      "step": 16715
    },
    {
      "epoch": 0.4402422965499078,
      "grad_norm": 5.056559085845947,
      "learning_rate": 2.800105346326047e-05,
      "loss": 1.0704,
      "step": 16716
    },
    {
      "epoch": 0.44026863313141956,
      "grad_norm": 1.7665220499038696,
      "learning_rate": 2.7999736634184888e-05,
      "loss": 1.7946,
      "step": 16717
    },
    {
      "epoch": 0.44029496971293125,
      "grad_norm": 2.2904481887817383,
      "learning_rate": 2.79984198051093e-05,
      "loss": 2.0682,
      "step": 16718
    },
    {
      "epoch": 0.440321306294443,
      "grad_norm": 2.2061214447021484,
      "learning_rate": 2.7997102976033713e-05,
      "loss": 2.6347,
      "step": 16719
    },
    {
      "epoch": 0.4403476428759547,
      "grad_norm": 1.9872568845748901,
      "learning_rate": 2.7995786146958125e-05,
      "loss": 1.5115,
      "step": 16720
    },
    {
      "epoch": 0.4403739794574664,
      "grad_norm": 3.753920078277588,
      "learning_rate": 2.7994469317882537e-05,
      "loss": 2.1913,
      "step": 16721
    },
    {
      "epoch": 0.44040031603897817,
      "grad_norm": 1.6522375345230103,
      "learning_rate": 2.7993152488806956e-05,
      "loss": 1.9022,
      "step": 16722
    },
    {
      "epoch": 0.44042665262048986,
      "grad_norm": 2.209785223007202,
      "learning_rate": 2.7991835659731368e-05,
      "loss": 2.0994,
      "step": 16723
    },
    {
      "epoch": 0.4404529892020016,
      "grad_norm": 1.7904874086380005,
      "learning_rate": 2.799051883065578e-05,
      "loss": 1.589,
      "step": 16724
    },
    {
      "epoch": 0.4404793257835133,
      "grad_norm": 2.5237224102020264,
      "learning_rate": 2.7989202001580196e-05,
      "loss": 1.3336,
      "step": 16725
    },
    {
      "epoch": 0.44050566236502503,
      "grad_norm": 2.4693028926849365,
      "learning_rate": 2.7987885172504608e-05,
      "loss": 2.1436,
      "step": 16726
    },
    {
      "epoch": 0.4405319989465367,
      "grad_norm": 1.6800041198730469,
      "learning_rate": 2.7986568343429027e-05,
      "loss": 2.0454,
      "step": 16727
    },
    {
      "epoch": 0.44055833552804846,
      "grad_norm": 2.56982421875,
      "learning_rate": 2.798525151435344e-05,
      "loss": 0.9311,
      "step": 16728
    },
    {
      "epoch": 0.4405846721095602,
      "grad_norm": 2.5149993896484375,
      "learning_rate": 2.798393468527785e-05,
      "loss": 1.6821,
      "step": 16729
    },
    {
      "epoch": 0.4406110086910719,
      "grad_norm": 1.8810791969299316,
      "learning_rate": 2.7982617856202264e-05,
      "loss": 2.4493,
      "step": 16730
    },
    {
      "epoch": 0.44063734527258364,
      "grad_norm": 1.9849098920822144,
      "learning_rate": 2.7981301027126683e-05,
      "loss": 1.7455,
      "step": 16731
    },
    {
      "epoch": 0.4406636818540953,
      "grad_norm": 5.133772850036621,
      "learning_rate": 2.7979984198051095e-05,
      "loss": 1.224,
      "step": 16732
    },
    {
      "epoch": 0.44069001843560707,
      "grad_norm": 1.9071556329727173,
      "learning_rate": 2.7978667368975507e-05,
      "loss": 1.7529,
      "step": 16733
    },
    {
      "epoch": 0.44071635501711875,
      "grad_norm": 1.4554475545883179,
      "learning_rate": 2.7977350539899923e-05,
      "loss": 1.5876,
      "step": 16734
    },
    {
      "epoch": 0.4407426915986305,
      "grad_norm": 3.959616184234619,
      "learning_rate": 2.7976033710824335e-05,
      "loss": 0.7249,
      "step": 16735
    },
    {
      "epoch": 0.44076902818014224,
      "grad_norm": 4.518592357635498,
      "learning_rate": 2.7974716881748754e-05,
      "loss": 2.0631,
      "step": 16736
    },
    {
      "epoch": 0.44079536476165393,
      "grad_norm": 1.9767451286315918,
      "learning_rate": 2.7973400052673166e-05,
      "loss": 1.2939,
      "step": 16737
    },
    {
      "epoch": 0.4408217013431657,
      "grad_norm": 1.6754765510559082,
      "learning_rate": 2.797208322359758e-05,
      "loss": 0.5993,
      "step": 16738
    },
    {
      "epoch": 0.44084803792467736,
      "grad_norm": 5.184513092041016,
      "learning_rate": 2.797076639452199e-05,
      "loss": 1.2088,
      "step": 16739
    },
    {
      "epoch": 0.4408743745061891,
      "grad_norm": 3.1669740676879883,
      "learning_rate": 2.7969449565446403e-05,
      "loss": 1.1096,
      "step": 16740
    },
    {
      "epoch": 0.4409007110877008,
      "grad_norm": 1.533186674118042,
      "learning_rate": 2.7968132736370822e-05,
      "loss": 0.6209,
      "step": 16741
    },
    {
      "epoch": 0.44092704766921254,
      "grad_norm": 1.5440374612808228,
      "learning_rate": 2.7966815907295234e-05,
      "loss": 1.3923,
      "step": 16742
    },
    {
      "epoch": 0.4409533842507243,
      "grad_norm": 3.21589994430542,
      "learning_rate": 2.796549907821965e-05,
      "loss": 1.1905,
      "step": 16743
    },
    {
      "epoch": 0.44097972083223597,
      "grad_norm": 1.814430594444275,
      "learning_rate": 2.7964182249144062e-05,
      "loss": 1.9907,
      "step": 16744
    },
    {
      "epoch": 0.4410060574137477,
      "grad_norm": 1.5405874252319336,
      "learning_rate": 2.796286542006848e-05,
      "loss": 1.8106,
      "step": 16745
    },
    {
      "epoch": 0.4410323939952594,
      "grad_norm": 1.4020309448242188,
      "learning_rate": 2.7961548590992893e-05,
      "loss": 1.9008,
      "step": 16746
    },
    {
      "epoch": 0.44105873057677114,
      "grad_norm": 3.371359348297119,
      "learning_rate": 2.7960231761917305e-05,
      "loss": 1.745,
      "step": 16747
    },
    {
      "epoch": 0.44108506715828283,
      "grad_norm": 4.1858930587768555,
      "learning_rate": 2.7958914932841717e-05,
      "loss": 1.5352,
      "step": 16748
    },
    {
      "epoch": 0.44111140373979457,
      "grad_norm": 2.824721336364746,
      "learning_rate": 2.795759810376613e-05,
      "loss": 1.6957,
      "step": 16749
    },
    {
      "epoch": 0.4411377403213063,
      "grad_norm": 1.7364839315414429,
      "learning_rate": 2.795628127469055e-05,
      "loss": 1.9256,
      "step": 16750
    },
    {
      "epoch": 0.441164076902818,
      "grad_norm": 2.467883825302124,
      "learning_rate": 2.795496444561496e-05,
      "loss": 1.9208,
      "step": 16751
    },
    {
      "epoch": 0.44119041348432975,
      "grad_norm": 1.568540096282959,
      "learning_rate": 2.7953647616539373e-05,
      "loss": 1.3809,
      "step": 16752
    },
    {
      "epoch": 0.44121675006584143,
      "grad_norm": 2.7559430599212646,
      "learning_rate": 2.795233078746379e-05,
      "loss": 1.5351,
      "step": 16753
    },
    {
      "epoch": 0.4412430866473532,
      "grad_norm": 3.288858413696289,
      "learning_rate": 2.79510139583882e-05,
      "loss": 1.8216,
      "step": 16754
    },
    {
      "epoch": 0.4412694232288649,
      "grad_norm": 1.822719931602478,
      "learning_rate": 2.794969712931262e-05,
      "loss": 1.8733,
      "step": 16755
    },
    {
      "epoch": 0.4412957598103766,
      "grad_norm": 2.376509666442871,
      "learning_rate": 2.7948380300237032e-05,
      "loss": 1.6281,
      "step": 16756
    },
    {
      "epoch": 0.44132209639188835,
      "grad_norm": 1.8745875358581543,
      "learning_rate": 2.7947063471161444e-05,
      "loss": 1.7885,
      "step": 16757
    },
    {
      "epoch": 0.44134843297340004,
      "grad_norm": 3.068058729171753,
      "learning_rate": 2.7945746642085856e-05,
      "loss": 1.3098,
      "step": 16758
    },
    {
      "epoch": 0.4413747695549118,
      "grad_norm": 2.4959139823913574,
      "learning_rate": 2.794442981301027e-05,
      "loss": 1.2738,
      "step": 16759
    },
    {
      "epoch": 0.44140110613642347,
      "grad_norm": 1.7837127447128296,
      "learning_rate": 2.7943112983934688e-05,
      "loss": 1.5153,
      "step": 16760
    },
    {
      "epoch": 0.4414274427179352,
      "grad_norm": 2.578883171081543,
      "learning_rate": 2.79417961548591e-05,
      "loss": 1.7842,
      "step": 16761
    },
    {
      "epoch": 0.44145377929944696,
      "grad_norm": 2.0582690238952637,
      "learning_rate": 2.7940479325783515e-05,
      "loss": 0.691,
      "step": 16762
    },
    {
      "epoch": 0.44148011588095865,
      "grad_norm": 1.7992372512817383,
      "learning_rate": 2.7939162496707928e-05,
      "loss": 2.1008,
      "step": 16763
    },
    {
      "epoch": 0.4415064524624704,
      "grad_norm": 3.998180866241455,
      "learning_rate": 2.7937845667632347e-05,
      "loss": 1.9433,
      "step": 16764
    },
    {
      "epoch": 0.4415327890439821,
      "grad_norm": 2.38696551322937,
      "learning_rate": 2.793652883855676e-05,
      "loss": 1.677,
      "step": 16765
    },
    {
      "epoch": 0.4415591256254938,
      "grad_norm": 1.9280604124069214,
      "learning_rate": 2.793521200948117e-05,
      "loss": 1.7596,
      "step": 16766
    },
    {
      "epoch": 0.4415854622070055,
      "grad_norm": 3.6320953369140625,
      "learning_rate": 2.7933895180405583e-05,
      "loss": 1.9152,
      "step": 16767
    },
    {
      "epoch": 0.44161179878851725,
      "grad_norm": 2.4863052368164062,
      "learning_rate": 2.7932578351329996e-05,
      "loss": 1.2012,
      "step": 16768
    },
    {
      "epoch": 0.441638135370029,
      "grad_norm": 1.4355405569076538,
      "learning_rate": 2.7931261522254414e-05,
      "loss": 1.1698,
      "step": 16769
    },
    {
      "epoch": 0.4416644719515407,
      "grad_norm": 1.891467809677124,
      "learning_rate": 2.7929944693178827e-05,
      "loss": 2.1571,
      "step": 16770
    },
    {
      "epoch": 0.4416908085330524,
      "grad_norm": 3.448801279067993,
      "learning_rate": 2.7928627864103242e-05,
      "loss": 1.5949,
      "step": 16771
    },
    {
      "epoch": 0.4417171451145641,
      "grad_norm": 4.097621917724609,
      "learning_rate": 2.7927311035027655e-05,
      "loss": 2.553,
      "step": 16772
    },
    {
      "epoch": 0.44174348169607586,
      "grad_norm": 1.3943984508514404,
      "learning_rate": 2.7925994205952067e-05,
      "loss": 0.5025,
      "step": 16773
    },
    {
      "epoch": 0.44176981827758754,
      "grad_norm": 1.7040599584579468,
      "learning_rate": 2.7924677376876486e-05,
      "loss": 1.3488,
      "step": 16774
    },
    {
      "epoch": 0.4417961548590993,
      "grad_norm": 4.253444194793701,
      "learning_rate": 2.7923360547800898e-05,
      "loss": 1.8751,
      "step": 16775
    },
    {
      "epoch": 0.44182249144061103,
      "grad_norm": 4.625461101531982,
      "learning_rate": 2.792204371872531e-05,
      "loss": 0.7516,
      "step": 16776
    },
    {
      "epoch": 0.4418488280221227,
      "grad_norm": 1.7392487525939941,
      "learning_rate": 2.7920726889649722e-05,
      "loss": 1.8613,
      "step": 16777
    },
    {
      "epoch": 0.44187516460363446,
      "grad_norm": 1.668346881866455,
      "learning_rate": 2.791941006057414e-05,
      "loss": 2.122,
      "step": 16778
    },
    {
      "epoch": 0.44190150118514615,
      "grad_norm": 4.862741947174072,
      "learning_rate": 2.7918093231498554e-05,
      "loss": 1.3039,
      "step": 16779
    },
    {
      "epoch": 0.4419278377666579,
      "grad_norm": 2.018497943878174,
      "learning_rate": 2.7916776402422966e-05,
      "loss": 1.6052,
      "step": 16780
    },
    {
      "epoch": 0.4419541743481696,
      "grad_norm": 1.9620503187179565,
      "learning_rate": 2.791545957334738e-05,
      "loss": 1.1701,
      "step": 16781
    },
    {
      "epoch": 0.4419805109296813,
      "grad_norm": 2.497812509536743,
      "learning_rate": 2.7914142744271794e-05,
      "loss": 1.8402,
      "step": 16782
    },
    {
      "epoch": 0.44200684751119307,
      "grad_norm": 1.4044374227523804,
      "learning_rate": 2.7912825915196213e-05,
      "loss": 1.7316,
      "step": 16783
    },
    {
      "epoch": 0.44203318409270476,
      "grad_norm": 2.0211944580078125,
      "learning_rate": 2.7911509086120625e-05,
      "loss": 1.4478,
      "step": 16784
    },
    {
      "epoch": 0.4420595206742165,
      "grad_norm": 1.9349212646484375,
      "learning_rate": 2.7910192257045037e-05,
      "loss": 2.0301,
      "step": 16785
    },
    {
      "epoch": 0.4420858572557282,
      "grad_norm": 2.043325185775757,
      "learning_rate": 2.790887542796945e-05,
      "loss": 1.3926,
      "step": 16786
    },
    {
      "epoch": 0.44211219383723993,
      "grad_norm": 2.8606808185577393,
      "learning_rate": 2.790755859889386e-05,
      "loss": 1.8603,
      "step": 16787
    },
    {
      "epoch": 0.4421385304187516,
      "grad_norm": 5.086670398712158,
      "learning_rate": 2.790624176981828e-05,
      "loss": 0.8439,
      "step": 16788
    },
    {
      "epoch": 0.44216486700026336,
      "grad_norm": 3.1945712566375732,
      "learning_rate": 2.7904924940742693e-05,
      "loss": 1.6982,
      "step": 16789
    },
    {
      "epoch": 0.4421912035817751,
      "grad_norm": 3.6901233196258545,
      "learning_rate": 2.7903608111667108e-05,
      "loss": 1.8002,
      "step": 16790
    },
    {
      "epoch": 0.4422175401632868,
      "grad_norm": 3.3323347568511963,
      "learning_rate": 2.790229128259152e-05,
      "loss": 2.2336,
      "step": 16791
    },
    {
      "epoch": 0.44224387674479854,
      "grad_norm": 5.013552188873291,
      "learning_rate": 2.7900974453515933e-05,
      "loss": 2.144,
      "step": 16792
    },
    {
      "epoch": 0.4422702133263102,
      "grad_norm": 2.5579257011413574,
      "learning_rate": 2.789965762444035e-05,
      "loss": 1.5682,
      "step": 16793
    },
    {
      "epoch": 0.44229654990782197,
      "grad_norm": 1.526326060295105,
      "learning_rate": 2.7898340795364764e-05,
      "loss": 1.4033,
      "step": 16794
    },
    {
      "epoch": 0.4423228864893337,
      "grad_norm": 1.6585986614227295,
      "learning_rate": 2.7897023966289176e-05,
      "loss": 2.1911,
      "step": 16795
    },
    {
      "epoch": 0.4423492230708454,
      "grad_norm": 2.5212934017181396,
      "learning_rate": 2.7895707137213588e-05,
      "loss": 0.2163,
      "step": 16796
    },
    {
      "epoch": 0.44237555965235714,
      "grad_norm": 1.8194775581359863,
      "learning_rate": 2.7894390308138007e-05,
      "loss": 2.0501,
      "step": 16797
    },
    {
      "epoch": 0.44240189623386883,
      "grad_norm": 1.8742762804031372,
      "learning_rate": 2.789307347906242e-05,
      "loss": 2.1743,
      "step": 16798
    },
    {
      "epoch": 0.4424282328153806,
      "grad_norm": 1.4154751300811768,
      "learning_rate": 2.789175664998683e-05,
      "loss": 2.158,
      "step": 16799
    },
    {
      "epoch": 0.44245456939689226,
      "grad_norm": 2.6107194423675537,
      "learning_rate": 2.7890439820911247e-05,
      "loss": 2.042,
      "step": 16800
    },
    {
      "epoch": 0.442480905978404,
      "grad_norm": 2.6754653453826904,
      "learning_rate": 2.788912299183566e-05,
      "loss": 2.268,
      "step": 16801
    },
    {
      "epoch": 0.44250724255991575,
      "grad_norm": 1.4304776191711426,
      "learning_rate": 2.788780616276008e-05,
      "loss": 1.39,
      "step": 16802
    },
    {
      "epoch": 0.44253357914142744,
      "grad_norm": 3.638984203338623,
      "learning_rate": 2.788648933368449e-05,
      "loss": 1.4697,
      "step": 16803
    },
    {
      "epoch": 0.4425599157229392,
      "grad_norm": 1.9505577087402344,
      "learning_rate": 2.7885172504608903e-05,
      "loss": 1.3988,
      "step": 16804
    },
    {
      "epoch": 0.44258625230445087,
      "grad_norm": 6.658775329589844,
      "learning_rate": 2.7883855675533315e-05,
      "loss": 1.7174,
      "step": 16805
    },
    {
      "epoch": 0.4426125888859626,
      "grad_norm": 1.8260091543197632,
      "learning_rate": 2.7882538846457727e-05,
      "loss": 1.6299,
      "step": 16806
    },
    {
      "epoch": 0.4426389254674743,
      "grad_norm": 2.0658438205718994,
      "learning_rate": 2.7881222017382146e-05,
      "loss": 1.519,
      "step": 16807
    },
    {
      "epoch": 0.44266526204898604,
      "grad_norm": 1.3855957984924316,
      "learning_rate": 2.787990518830656e-05,
      "loss": 1.2043,
      "step": 16808
    },
    {
      "epoch": 0.4426915986304978,
      "grad_norm": 3.339667797088623,
      "learning_rate": 2.7878588359230974e-05,
      "loss": 1.3137,
      "step": 16809
    },
    {
      "epoch": 0.4427179352120095,
      "grad_norm": 2.0247812271118164,
      "learning_rate": 2.7877271530155386e-05,
      "loss": 1.8398,
      "step": 16810
    },
    {
      "epoch": 0.4427442717935212,
      "grad_norm": 1.5882487297058105,
      "learning_rate": 2.7875954701079805e-05,
      "loss": 1.8964,
      "step": 16811
    },
    {
      "epoch": 0.4427706083750329,
      "grad_norm": 2.187187910079956,
      "learning_rate": 2.7874637872004217e-05,
      "loss": 1.5122,
      "step": 16812
    },
    {
      "epoch": 0.44279694495654465,
      "grad_norm": 2.3871068954467773,
      "learning_rate": 2.787332104292863e-05,
      "loss": 1.4994,
      "step": 16813
    },
    {
      "epoch": 0.44282328153805633,
      "grad_norm": 2.455298662185669,
      "learning_rate": 2.7872004213853042e-05,
      "loss": 1.6469,
      "step": 16814
    },
    {
      "epoch": 0.4428496181195681,
      "grad_norm": 2.073849678039551,
      "learning_rate": 2.7870687384777454e-05,
      "loss": 2.3492,
      "step": 16815
    },
    {
      "epoch": 0.4428759547010798,
      "grad_norm": 2.083263635635376,
      "learning_rate": 2.7869370555701873e-05,
      "loss": 0.7781,
      "step": 16816
    },
    {
      "epoch": 0.4429022912825915,
      "grad_norm": 3.7390551567077637,
      "learning_rate": 2.7868053726626285e-05,
      "loss": 0.8679,
      "step": 16817
    },
    {
      "epoch": 0.44292862786410325,
      "grad_norm": 2.083997964859009,
      "learning_rate": 2.78667368975507e-05,
      "loss": 1.4691,
      "step": 16818
    },
    {
      "epoch": 0.44295496444561494,
      "grad_norm": 1.7258881330490112,
      "learning_rate": 2.7865420068475113e-05,
      "loss": 2.1395,
      "step": 16819
    },
    {
      "epoch": 0.4429813010271267,
      "grad_norm": 2.7336554527282715,
      "learning_rate": 2.7864103239399525e-05,
      "loss": 2.7409,
      "step": 16820
    },
    {
      "epoch": 0.44300763760863837,
      "grad_norm": 2.1014702320098877,
      "learning_rate": 2.7862786410323944e-05,
      "loss": 1.6637,
      "step": 16821
    },
    {
      "epoch": 0.4430339741901501,
      "grad_norm": 3.676614761352539,
      "learning_rate": 2.7861469581248356e-05,
      "loss": 2.2768,
      "step": 16822
    },
    {
      "epoch": 0.44306031077166186,
      "grad_norm": 2.7421998977661133,
      "learning_rate": 2.786015275217277e-05,
      "loss": 1.7537,
      "step": 16823
    },
    {
      "epoch": 0.44308664735317355,
      "grad_norm": 1.6489410400390625,
      "learning_rate": 2.785883592309718e-05,
      "loss": 2.6386,
      "step": 16824
    },
    {
      "epoch": 0.4431129839346853,
      "grad_norm": 1.9031383991241455,
      "learning_rate": 2.7857519094021593e-05,
      "loss": 1.5666,
      "step": 16825
    },
    {
      "epoch": 0.443139320516197,
      "grad_norm": 2.2424068450927734,
      "learning_rate": 2.7856202264946012e-05,
      "loss": 1.6432,
      "step": 16826
    },
    {
      "epoch": 0.4431656570977087,
      "grad_norm": 3.7087645530700684,
      "learning_rate": 2.7854885435870424e-05,
      "loss": 1.3907,
      "step": 16827
    },
    {
      "epoch": 0.44319199367922046,
      "grad_norm": 2.816002130508423,
      "learning_rate": 2.785356860679484e-05,
      "loss": 2.2441,
      "step": 16828
    },
    {
      "epoch": 0.44321833026073215,
      "grad_norm": 2.118114709854126,
      "learning_rate": 2.7852251777719252e-05,
      "loss": 2.7322,
      "step": 16829
    },
    {
      "epoch": 0.4432446668422439,
      "grad_norm": 2.9789841175079346,
      "learning_rate": 2.785093494864367e-05,
      "loss": 1.5145,
      "step": 16830
    },
    {
      "epoch": 0.4432710034237556,
      "grad_norm": 2.5113940238952637,
      "learning_rate": 2.7849618119568083e-05,
      "loss": 0.5272,
      "step": 16831
    },
    {
      "epoch": 0.4432973400052673,
      "grad_norm": 2.500525712966919,
      "learning_rate": 2.7848301290492496e-05,
      "loss": 1.838,
      "step": 16832
    },
    {
      "epoch": 0.443323676586779,
      "grad_norm": 2.9666333198547363,
      "learning_rate": 2.7846984461416908e-05,
      "loss": 0.8301,
      "step": 16833
    },
    {
      "epoch": 0.44335001316829076,
      "grad_norm": 1.8912339210510254,
      "learning_rate": 2.784566763234132e-05,
      "loss": 1.0398,
      "step": 16834
    },
    {
      "epoch": 0.4433763497498025,
      "grad_norm": 2.5475335121154785,
      "learning_rate": 2.784435080326574e-05,
      "loss": 1.86,
      "step": 16835
    },
    {
      "epoch": 0.4434026863313142,
      "grad_norm": 2.1660566329956055,
      "learning_rate": 2.784303397419015e-05,
      "loss": 1.6221,
      "step": 16836
    },
    {
      "epoch": 0.44342902291282593,
      "grad_norm": 2.804410457611084,
      "learning_rate": 2.7841717145114567e-05,
      "loss": 2.5628,
      "step": 16837
    },
    {
      "epoch": 0.4434553594943376,
      "grad_norm": 3.7074782848358154,
      "learning_rate": 2.784040031603898e-05,
      "loss": 1.7598,
      "step": 16838
    },
    {
      "epoch": 0.44348169607584936,
      "grad_norm": 2.791229724884033,
      "learning_rate": 2.783908348696339e-05,
      "loss": 1.3794,
      "step": 16839
    },
    {
      "epoch": 0.44350803265736105,
      "grad_norm": 1.6307436227798462,
      "learning_rate": 2.783776665788781e-05,
      "loss": 1.5339,
      "step": 16840
    },
    {
      "epoch": 0.4435343692388728,
      "grad_norm": 4.305070877075195,
      "learning_rate": 2.7836449828812222e-05,
      "loss": 1.7818,
      "step": 16841
    },
    {
      "epoch": 0.44356070582038454,
      "grad_norm": 2.9363365173339844,
      "learning_rate": 2.7835132999736635e-05,
      "loss": 1.6574,
      "step": 16842
    },
    {
      "epoch": 0.4435870424018962,
      "grad_norm": 2.335444450378418,
      "learning_rate": 2.7833816170661047e-05,
      "loss": 1.6074,
      "step": 16843
    },
    {
      "epoch": 0.44361337898340797,
      "grad_norm": 2.6473681926727295,
      "learning_rate": 2.7832499341585466e-05,
      "loss": 1.4399,
      "step": 16844
    },
    {
      "epoch": 0.44363971556491966,
      "grad_norm": 1.644146203994751,
      "learning_rate": 2.7831182512509878e-05,
      "loss": 0.8183,
      "step": 16845
    },
    {
      "epoch": 0.4436660521464314,
      "grad_norm": 1.9546242952346802,
      "learning_rate": 2.7829865683434294e-05,
      "loss": 1.7601,
      "step": 16846
    },
    {
      "epoch": 0.4436923887279431,
      "grad_norm": 4.175062656402588,
      "learning_rate": 2.7828548854358706e-05,
      "loss": 2.1668,
      "step": 16847
    },
    {
      "epoch": 0.44371872530945483,
      "grad_norm": 1.99675714969635,
      "learning_rate": 2.7827232025283118e-05,
      "loss": 2.4218,
      "step": 16848
    },
    {
      "epoch": 0.4437450618909666,
      "grad_norm": 1.891690969467163,
      "learning_rate": 2.7825915196207537e-05,
      "loss": 1.7463,
      "step": 16849
    },
    {
      "epoch": 0.44377139847247826,
      "grad_norm": 5.1198320388793945,
      "learning_rate": 2.782459836713195e-05,
      "loss": 1.5876,
      "step": 16850
    },
    {
      "epoch": 0.44379773505399,
      "grad_norm": 2.1891446113586426,
      "learning_rate": 2.782328153805636e-05,
      "loss": 1.8012,
      "step": 16851
    },
    {
      "epoch": 0.4438240716355017,
      "grad_norm": 2.445836067199707,
      "learning_rate": 2.7821964708980774e-05,
      "loss": 1.7342,
      "step": 16852
    },
    {
      "epoch": 0.44385040821701344,
      "grad_norm": 1.9329490661621094,
      "learning_rate": 2.7820647879905186e-05,
      "loss": 0.5168,
      "step": 16853
    },
    {
      "epoch": 0.4438767447985251,
      "grad_norm": 1.816304087638855,
      "learning_rate": 2.7819331050829605e-05,
      "loss": 1.6229,
      "step": 16854
    },
    {
      "epoch": 0.44390308138003687,
      "grad_norm": 1.3820242881774902,
      "learning_rate": 2.7818014221754017e-05,
      "loss": 1.4886,
      "step": 16855
    },
    {
      "epoch": 0.4439294179615486,
      "grad_norm": 1.9624178409576416,
      "learning_rate": 2.7816697392678433e-05,
      "loss": 1.6843,
      "step": 16856
    },
    {
      "epoch": 0.4439557545430603,
      "grad_norm": 2.3891029357910156,
      "learning_rate": 2.7815380563602845e-05,
      "loss": 1.1776,
      "step": 16857
    },
    {
      "epoch": 0.44398209112457204,
      "grad_norm": 3.7418570518493652,
      "learning_rate": 2.7814063734527257e-05,
      "loss": 1.0569,
      "step": 16858
    },
    {
      "epoch": 0.44400842770608373,
      "grad_norm": 2.0556480884552,
      "learning_rate": 2.7812746905451676e-05,
      "loss": 0.5321,
      "step": 16859
    },
    {
      "epoch": 0.4440347642875955,
      "grad_norm": 7.790976047515869,
      "learning_rate": 2.7811430076376088e-05,
      "loss": 1.1386,
      "step": 16860
    },
    {
      "epoch": 0.4440611008691072,
      "grad_norm": 2.29221510887146,
      "learning_rate": 2.78101132473005e-05,
      "loss": 1.7673,
      "step": 16861
    },
    {
      "epoch": 0.4440874374506189,
      "grad_norm": 1.8455910682678223,
      "learning_rate": 2.7808796418224913e-05,
      "loss": 1.4893,
      "step": 16862
    },
    {
      "epoch": 0.44411377403213065,
      "grad_norm": 1.9344048500061035,
      "learning_rate": 2.780747958914933e-05,
      "loss": 1.937,
      "step": 16863
    },
    {
      "epoch": 0.44414011061364234,
      "grad_norm": 1.6465835571289062,
      "learning_rate": 2.7806162760073744e-05,
      "loss": 1.6376,
      "step": 16864
    },
    {
      "epoch": 0.4441664471951541,
      "grad_norm": 1.7242937088012695,
      "learning_rate": 2.780484593099816e-05,
      "loss": 1.8259,
      "step": 16865
    },
    {
      "epoch": 0.44419278377666577,
      "grad_norm": 2.9298622608184814,
      "learning_rate": 2.780352910192257e-05,
      "loss": 2.0969,
      "step": 16866
    },
    {
      "epoch": 0.4442191203581775,
      "grad_norm": 2.841571569442749,
      "learning_rate": 2.7802212272846984e-05,
      "loss": 1.343,
      "step": 16867
    },
    {
      "epoch": 0.44424545693968925,
      "grad_norm": 3.596860885620117,
      "learning_rate": 2.7800895443771403e-05,
      "loss": 1.51,
      "step": 16868
    },
    {
      "epoch": 0.44427179352120094,
      "grad_norm": 1.8692944049835205,
      "learning_rate": 2.7799578614695815e-05,
      "loss": 1.8425,
      "step": 16869
    },
    {
      "epoch": 0.4442981301027127,
      "grad_norm": 1.5739389657974243,
      "learning_rate": 2.7798261785620227e-05,
      "loss": 1.5801,
      "step": 16870
    },
    {
      "epoch": 0.4443244666842244,
      "grad_norm": 2.111631393432617,
      "learning_rate": 2.779694495654464e-05,
      "loss": 1.9658,
      "step": 16871
    },
    {
      "epoch": 0.4443508032657361,
      "grad_norm": 1.912125587463379,
      "learning_rate": 2.779562812746905e-05,
      "loss": 2.4808,
      "step": 16872
    },
    {
      "epoch": 0.4443771398472478,
      "grad_norm": 3.359158992767334,
      "learning_rate": 2.779431129839347e-05,
      "loss": 0.8423,
      "step": 16873
    },
    {
      "epoch": 0.44440347642875955,
      "grad_norm": 2.570143461227417,
      "learning_rate": 2.7792994469317883e-05,
      "loss": 1.1699,
      "step": 16874
    },
    {
      "epoch": 0.4444298130102713,
      "grad_norm": 2.8838579654693604,
      "learning_rate": 2.77916776402423e-05,
      "loss": 0.8558,
      "step": 16875
    },
    {
      "epoch": 0.444456149591783,
      "grad_norm": 3.3629441261291504,
      "learning_rate": 2.779036081116671e-05,
      "loss": 1.8851,
      "step": 16876
    },
    {
      "epoch": 0.4444824861732947,
      "grad_norm": 1.9049854278564453,
      "learning_rate": 2.778904398209113e-05,
      "loss": 2.054,
      "step": 16877
    },
    {
      "epoch": 0.4445088227548064,
      "grad_norm": 3.3422224521636963,
      "learning_rate": 2.7787727153015542e-05,
      "loss": 1.9309,
      "step": 16878
    },
    {
      "epoch": 0.44453515933631815,
      "grad_norm": 1.9425464868545532,
      "learning_rate": 2.7786410323939954e-05,
      "loss": 1.5455,
      "step": 16879
    },
    {
      "epoch": 0.44456149591782984,
      "grad_norm": 3.086204767227173,
      "learning_rate": 2.7785093494864366e-05,
      "loss": 1.569,
      "step": 16880
    },
    {
      "epoch": 0.4445878324993416,
      "grad_norm": 1.6296151876449585,
      "learning_rate": 2.778377666578878e-05,
      "loss": 1.8838,
      "step": 16881
    },
    {
      "epoch": 0.4446141690808533,
      "grad_norm": 1.8207261562347412,
      "learning_rate": 2.7782459836713197e-05,
      "loss": 2.1768,
      "step": 16882
    },
    {
      "epoch": 0.444640505662365,
      "grad_norm": 3.021439790725708,
      "learning_rate": 2.778114300763761e-05,
      "loss": 1.2047,
      "step": 16883
    },
    {
      "epoch": 0.44466684224387676,
      "grad_norm": 2.3001673221588135,
      "learning_rate": 2.7779826178562025e-05,
      "loss": 1.7757,
      "step": 16884
    },
    {
      "epoch": 0.44469317882538845,
      "grad_norm": 1.5953797101974487,
      "learning_rate": 2.7778509349486437e-05,
      "loss": 2.1521,
      "step": 16885
    },
    {
      "epoch": 0.4447195154069002,
      "grad_norm": 2.8892202377319336,
      "learning_rate": 2.777719252041085e-05,
      "loss": 1.5184,
      "step": 16886
    },
    {
      "epoch": 0.4447458519884119,
      "grad_norm": 2.0880119800567627,
      "learning_rate": 2.777587569133527e-05,
      "loss": 1.5588,
      "step": 16887
    },
    {
      "epoch": 0.4447721885699236,
      "grad_norm": 1.7109434604644775,
      "learning_rate": 2.777455886225968e-05,
      "loss": 1.5284,
      "step": 16888
    },
    {
      "epoch": 0.44479852515143536,
      "grad_norm": 1.627532720565796,
      "learning_rate": 2.7773242033184093e-05,
      "loss": 2.217,
      "step": 16889
    },
    {
      "epoch": 0.44482486173294705,
      "grad_norm": 2.318840265274048,
      "learning_rate": 2.7771925204108505e-05,
      "loss": 1.7881,
      "step": 16890
    },
    {
      "epoch": 0.4448511983144588,
      "grad_norm": 2.502875566482544,
      "learning_rate": 2.777060837503292e-05,
      "loss": 1.7811,
      "step": 16891
    },
    {
      "epoch": 0.4448775348959705,
      "grad_norm": 1.4858766794204712,
      "learning_rate": 2.7769291545957337e-05,
      "loss": 1.5904,
      "step": 16892
    },
    {
      "epoch": 0.4449038714774822,
      "grad_norm": 1.8423118591308594,
      "learning_rate": 2.7767974716881752e-05,
      "loss": 2.4283,
      "step": 16893
    },
    {
      "epoch": 0.44493020805899397,
      "grad_norm": 3.259521722793579,
      "learning_rate": 2.7766657887806164e-05,
      "loss": 1.1515,
      "step": 16894
    },
    {
      "epoch": 0.44495654464050566,
      "grad_norm": 3.624154806137085,
      "learning_rate": 2.7765341058730577e-05,
      "loss": 1.757,
      "step": 16895
    },
    {
      "epoch": 0.4449828812220174,
      "grad_norm": 2.348252058029175,
      "learning_rate": 2.7764024229654996e-05,
      "loss": 1.184,
      "step": 16896
    },
    {
      "epoch": 0.4450092178035291,
      "grad_norm": 2.777730941772461,
      "learning_rate": 2.7762707400579408e-05,
      "loss": 1.1515,
      "step": 16897
    },
    {
      "epoch": 0.44503555438504083,
      "grad_norm": 2.3152735233306885,
      "learning_rate": 2.776139057150382e-05,
      "loss": 2.2853,
      "step": 16898
    },
    {
      "epoch": 0.4450618909665525,
      "grad_norm": 1.5822404623031616,
      "learning_rate": 2.7760073742428232e-05,
      "loss": 1.7404,
      "step": 16899
    },
    {
      "epoch": 0.44508822754806426,
      "grad_norm": 1.573306918144226,
      "learning_rate": 2.7758756913352644e-05,
      "loss": 2.1833,
      "step": 16900
    },
    {
      "epoch": 0.445114564129576,
      "grad_norm": 1.7900347709655762,
      "learning_rate": 2.7757440084277063e-05,
      "loss": 1.7033,
      "step": 16901
    },
    {
      "epoch": 0.4451409007110877,
      "grad_norm": 2.898188591003418,
      "learning_rate": 2.7756123255201476e-05,
      "loss": 0.6554,
      "step": 16902
    },
    {
      "epoch": 0.44516723729259944,
      "grad_norm": 3.6206510066986084,
      "learning_rate": 2.775480642612589e-05,
      "loss": 2.0676,
      "step": 16903
    },
    {
      "epoch": 0.4451935738741111,
      "grad_norm": 1.631095290184021,
      "learning_rate": 2.7753489597050303e-05,
      "loss": 1.7796,
      "step": 16904
    },
    {
      "epoch": 0.44521991045562287,
      "grad_norm": 1.7430720329284668,
      "learning_rate": 2.7752172767974716e-05,
      "loss": 0.9468,
      "step": 16905
    },
    {
      "epoch": 0.44524624703713456,
      "grad_norm": 2.8095223903656006,
      "learning_rate": 2.7750855938899135e-05,
      "loss": 1.6888,
      "step": 16906
    },
    {
      "epoch": 0.4452725836186463,
      "grad_norm": 1.7991313934326172,
      "learning_rate": 2.7749539109823547e-05,
      "loss": 1.6587,
      "step": 16907
    },
    {
      "epoch": 0.44529892020015804,
      "grad_norm": 1.771796703338623,
      "learning_rate": 2.774822228074796e-05,
      "loss": 1.3306,
      "step": 16908
    },
    {
      "epoch": 0.44532525678166973,
      "grad_norm": 1.8517221212387085,
      "learning_rate": 2.774690545167237e-05,
      "loss": 2.137,
      "step": 16909
    },
    {
      "epoch": 0.4453515933631815,
      "grad_norm": 3.1675400733947754,
      "learning_rate": 2.774558862259679e-05,
      "loss": 1.8773,
      "step": 16910
    },
    {
      "epoch": 0.44537792994469316,
      "grad_norm": 2.7926506996154785,
      "learning_rate": 2.7744271793521202e-05,
      "loss": 1.8622,
      "step": 16911
    },
    {
      "epoch": 0.4454042665262049,
      "grad_norm": 1.9208461046218872,
      "learning_rate": 2.7742954964445618e-05,
      "loss": 2.5781,
      "step": 16912
    },
    {
      "epoch": 0.4454306031077166,
      "grad_norm": 2.115795373916626,
      "learning_rate": 2.774163813537003e-05,
      "loss": 1.7638,
      "step": 16913
    },
    {
      "epoch": 0.44545693968922834,
      "grad_norm": 1.5698108673095703,
      "learning_rate": 2.7740321306294442e-05,
      "loss": 1.7687,
      "step": 16914
    },
    {
      "epoch": 0.4454832762707401,
      "grad_norm": 3.2958595752716064,
      "learning_rate": 2.773900447721886e-05,
      "loss": 1.2381,
      "step": 16915
    },
    {
      "epoch": 0.44550961285225177,
      "grad_norm": 1.9297618865966797,
      "learning_rate": 2.7737687648143274e-05,
      "loss": 1.583,
      "step": 16916
    },
    {
      "epoch": 0.4455359494337635,
      "grad_norm": 1.9617490768432617,
      "learning_rate": 2.7736370819067686e-05,
      "loss": 1.5403,
      "step": 16917
    },
    {
      "epoch": 0.4455622860152752,
      "grad_norm": 2.4254512786865234,
      "learning_rate": 2.7735053989992098e-05,
      "loss": 2.1346,
      "step": 16918
    },
    {
      "epoch": 0.44558862259678694,
      "grad_norm": 1.8264394998550415,
      "learning_rate": 2.7733737160916514e-05,
      "loss": 1.9176,
      "step": 16919
    },
    {
      "epoch": 0.44561495917829863,
      "grad_norm": 3.0311176776885986,
      "learning_rate": 2.773242033184093e-05,
      "loss": 1.1973,
      "step": 16920
    },
    {
      "epoch": 0.4456412957598104,
      "grad_norm": 5.307843208312988,
      "learning_rate": 2.7731103502765345e-05,
      "loss": 1.7294,
      "step": 16921
    },
    {
      "epoch": 0.4456676323413221,
      "grad_norm": 2.270634889602661,
      "learning_rate": 2.7729786673689757e-05,
      "loss": 1.4195,
      "step": 16922
    },
    {
      "epoch": 0.4456939689228338,
      "grad_norm": 1.8978369235992432,
      "learning_rate": 2.772846984461417e-05,
      "loss": 1.6679,
      "step": 16923
    },
    {
      "epoch": 0.44572030550434555,
      "grad_norm": 3.815642833709717,
      "learning_rate": 2.772715301553858e-05,
      "loss": 1.0659,
      "step": 16924
    },
    {
      "epoch": 0.44574664208585724,
      "grad_norm": 1.8492826223373413,
      "learning_rate": 2.7725836186463e-05,
      "loss": 1.4072,
      "step": 16925
    },
    {
      "epoch": 0.445772978667369,
      "grad_norm": 1.870880126953125,
      "learning_rate": 2.7724519357387413e-05,
      "loss": 1.9449,
      "step": 16926
    },
    {
      "epoch": 0.44579931524888067,
      "grad_norm": 2.6569020748138428,
      "learning_rate": 2.7723202528311825e-05,
      "loss": 2.0963,
      "step": 16927
    },
    {
      "epoch": 0.4458256518303924,
      "grad_norm": 2.126800060272217,
      "learning_rate": 2.7721885699236237e-05,
      "loss": 1.9566,
      "step": 16928
    },
    {
      "epoch": 0.44585198841190415,
      "grad_norm": 2.0076680183410645,
      "learning_rate": 2.7720568870160656e-05,
      "loss": 2.3031,
      "step": 16929
    },
    {
      "epoch": 0.44587832499341584,
      "grad_norm": 2.5603325366973877,
      "learning_rate": 2.7719252041085068e-05,
      "loss": 0.8739,
      "step": 16930
    },
    {
      "epoch": 0.4459046615749276,
      "grad_norm": 3.1730544567108154,
      "learning_rate": 2.7717935212009484e-05,
      "loss": 1.7421,
      "step": 16931
    },
    {
      "epoch": 0.4459309981564393,
      "grad_norm": 1.8664087057113647,
      "learning_rate": 2.7716618382933896e-05,
      "loss": 1.6497,
      "step": 16932
    },
    {
      "epoch": 0.445957334737951,
      "grad_norm": 1.5322927236557007,
      "learning_rate": 2.7715301553858308e-05,
      "loss": 1.4631,
      "step": 16933
    },
    {
      "epoch": 0.44598367131946276,
      "grad_norm": 1.8551011085510254,
      "learning_rate": 2.7713984724782727e-05,
      "loss": 0.9272,
      "step": 16934
    },
    {
      "epoch": 0.44601000790097445,
      "grad_norm": 2.6447277069091797,
      "learning_rate": 2.771266789570714e-05,
      "loss": 0.9992,
      "step": 16935
    },
    {
      "epoch": 0.4460363444824862,
      "grad_norm": 3.992147922515869,
      "learning_rate": 2.771135106663155e-05,
      "loss": 1.4683,
      "step": 16936
    },
    {
      "epoch": 0.4460626810639979,
      "grad_norm": 1.8378742933273315,
      "learning_rate": 2.7710034237555964e-05,
      "loss": 1.9912,
      "step": 16937
    },
    {
      "epoch": 0.4460890176455096,
      "grad_norm": 1.8693233728408813,
      "learning_rate": 2.770871740848038e-05,
      "loss": 1.2862,
      "step": 16938
    },
    {
      "epoch": 0.4461153542270213,
      "grad_norm": 1.7912756204605103,
      "learning_rate": 2.7707400579404795e-05,
      "loss": 2.0776,
      "step": 16939
    },
    {
      "epoch": 0.44614169080853305,
      "grad_norm": 1.797825813293457,
      "learning_rate": 2.770608375032921e-05,
      "loss": 2.1249,
      "step": 16940
    },
    {
      "epoch": 0.4461680273900448,
      "grad_norm": 2.193272352218628,
      "learning_rate": 2.7704766921253623e-05,
      "loss": 0.9412,
      "step": 16941
    },
    {
      "epoch": 0.4461943639715565,
      "grad_norm": 2.6819419860839844,
      "learning_rate": 2.7703450092178035e-05,
      "loss": 1.4018,
      "step": 16942
    },
    {
      "epoch": 0.44622070055306823,
      "grad_norm": 1.5943903923034668,
      "learning_rate": 2.7702133263102454e-05,
      "loss": 0.2727,
      "step": 16943
    },
    {
      "epoch": 0.4462470371345799,
      "grad_norm": 3.126850128173828,
      "learning_rate": 2.7700816434026866e-05,
      "loss": 1.4446,
      "step": 16944
    },
    {
      "epoch": 0.44627337371609166,
      "grad_norm": 2.14018177986145,
      "learning_rate": 2.769949960495128e-05,
      "loss": 1.9665,
      "step": 16945
    },
    {
      "epoch": 0.44629971029760335,
      "grad_norm": 2.214160442352295,
      "learning_rate": 2.769818277587569e-05,
      "loss": 2.0631,
      "step": 16946
    },
    {
      "epoch": 0.4463260468791151,
      "grad_norm": 2.8749234676361084,
      "learning_rate": 2.7696865946800106e-05,
      "loss": 2.1724,
      "step": 16947
    },
    {
      "epoch": 0.44635238346062683,
      "grad_norm": 2.9379866123199463,
      "learning_rate": 2.7695549117724522e-05,
      "loss": 1.7791,
      "step": 16948
    },
    {
      "epoch": 0.4463787200421385,
      "grad_norm": 2.057192087173462,
      "learning_rate": 2.7694232288648937e-05,
      "loss": 1.4764,
      "step": 16949
    },
    {
      "epoch": 0.44640505662365026,
      "grad_norm": 2.3252058029174805,
      "learning_rate": 2.769291545957335e-05,
      "loss": 2.189,
      "step": 16950
    },
    {
      "epoch": 0.44643139320516195,
      "grad_norm": 1.5831563472747803,
      "learning_rate": 2.7691598630497762e-05,
      "loss": 1.5904,
      "step": 16951
    },
    {
      "epoch": 0.4464577297866737,
      "grad_norm": 1.7704025506973267,
      "learning_rate": 2.7690281801422174e-05,
      "loss": 1.7409,
      "step": 16952
    },
    {
      "epoch": 0.4464840663681854,
      "grad_norm": 1.6311105489730835,
      "learning_rate": 2.7688964972346593e-05,
      "loss": 1.8551,
      "step": 16953
    },
    {
      "epoch": 0.4465104029496971,
      "grad_norm": 1.8063361644744873,
      "learning_rate": 2.7687648143271005e-05,
      "loss": 1.7091,
      "step": 16954
    },
    {
      "epoch": 0.44653673953120887,
      "grad_norm": 1.9200565814971924,
      "learning_rate": 2.7686331314195418e-05,
      "loss": 1.5417,
      "step": 16955
    },
    {
      "epoch": 0.44656307611272056,
      "grad_norm": 1.5721930265426636,
      "learning_rate": 2.768501448511983e-05,
      "loss": 1.8376,
      "step": 16956
    },
    {
      "epoch": 0.4465894126942323,
      "grad_norm": 1.7521731853485107,
      "learning_rate": 2.7683697656044245e-05,
      "loss": 1.8468,
      "step": 16957
    },
    {
      "epoch": 0.446615749275744,
      "grad_norm": 2.1449780464172363,
      "learning_rate": 2.768238082696866e-05,
      "loss": 2.3527,
      "step": 16958
    },
    {
      "epoch": 0.44664208585725573,
      "grad_norm": 1.9583295583724976,
      "learning_rate": 2.7681063997893077e-05,
      "loss": 2.2316,
      "step": 16959
    },
    {
      "epoch": 0.4466684224387674,
      "grad_norm": 2.2941904067993164,
      "learning_rate": 2.767974716881749e-05,
      "loss": 1.235,
      "step": 16960
    },
    {
      "epoch": 0.44669475902027916,
      "grad_norm": 2.3666372299194336,
      "learning_rate": 2.76784303397419e-05,
      "loss": 1.3692,
      "step": 16961
    },
    {
      "epoch": 0.4467210956017909,
      "grad_norm": 1.5040719509124756,
      "learning_rate": 2.767711351066632e-05,
      "loss": 1.6571,
      "step": 16962
    },
    {
      "epoch": 0.4467474321833026,
      "grad_norm": 1.6939072608947754,
      "learning_rate": 2.7675796681590732e-05,
      "loss": 1.983,
      "step": 16963
    },
    {
      "epoch": 0.44677376876481434,
      "grad_norm": 1.7759827375411987,
      "learning_rate": 2.7674479852515144e-05,
      "loss": 1.5155,
      "step": 16964
    },
    {
      "epoch": 0.446800105346326,
      "grad_norm": 1.6097828149795532,
      "learning_rate": 2.7673163023439557e-05,
      "loss": 1.9206,
      "step": 16965
    },
    {
      "epoch": 0.44682644192783777,
      "grad_norm": 3.1648993492126465,
      "learning_rate": 2.7671846194363972e-05,
      "loss": 1.3184,
      "step": 16966
    },
    {
      "epoch": 0.4468527785093495,
      "grad_norm": 3.994750499725342,
      "learning_rate": 2.7670529365288388e-05,
      "loss": 1.7561,
      "step": 16967
    },
    {
      "epoch": 0.4468791150908612,
      "grad_norm": 3.123556137084961,
      "learning_rate": 2.7669212536212803e-05,
      "loss": 0.9361,
      "step": 16968
    },
    {
      "epoch": 0.44690545167237294,
      "grad_norm": 3.8806233406066895,
      "learning_rate": 2.7667895707137216e-05,
      "loss": 1.4404,
      "step": 16969
    },
    {
      "epoch": 0.44693178825388463,
      "grad_norm": 3.531498432159424,
      "learning_rate": 2.7666578878061628e-05,
      "loss": 1.6582,
      "step": 16970
    },
    {
      "epoch": 0.4469581248353964,
      "grad_norm": 1.766258955001831,
      "learning_rate": 2.766526204898604e-05,
      "loss": 1.7279,
      "step": 16971
    },
    {
      "epoch": 0.44698446141690806,
      "grad_norm": 2.598158121109009,
      "learning_rate": 2.766394521991046e-05,
      "loss": 1.9407,
      "step": 16972
    },
    {
      "epoch": 0.4470107979984198,
      "grad_norm": 1.6280368566513062,
      "learning_rate": 2.766262839083487e-05,
      "loss": 1.9219,
      "step": 16973
    },
    {
      "epoch": 0.44703713457993155,
      "grad_norm": 2.0324461460113525,
      "learning_rate": 2.7661311561759283e-05,
      "loss": 1.1257,
      "step": 16974
    },
    {
      "epoch": 0.44706347116144324,
      "grad_norm": 4.4627790451049805,
      "learning_rate": 2.7659994732683696e-05,
      "loss": 1.02,
      "step": 16975
    },
    {
      "epoch": 0.447089807742955,
      "grad_norm": 1.9029957056045532,
      "learning_rate": 2.7658677903608115e-05,
      "loss": 1.5703,
      "step": 16976
    },
    {
      "epoch": 0.44711614432446667,
      "grad_norm": 3.1215009689331055,
      "learning_rate": 2.7657361074532527e-05,
      "loss": 1.8366,
      "step": 16977
    },
    {
      "epoch": 0.4471424809059784,
      "grad_norm": 4.092123508453369,
      "learning_rate": 2.7656044245456942e-05,
      "loss": 2.1572,
      "step": 16978
    },
    {
      "epoch": 0.4471688174874901,
      "grad_norm": 2.174711227416992,
      "learning_rate": 2.7654727416381355e-05,
      "loss": 2.4771,
      "step": 16979
    },
    {
      "epoch": 0.44719515406900184,
      "grad_norm": 2.23246693611145,
      "learning_rate": 2.7653410587305767e-05,
      "loss": 1.8769,
      "step": 16980
    },
    {
      "epoch": 0.4472214906505136,
      "grad_norm": 2.5172669887542725,
      "learning_rate": 2.7652093758230186e-05,
      "loss": 1.5685,
      "step": 16981
    },
    {
      "epoch": 0.4472478272320253,
      "grad_norm": 2.2094478607177734,
      "learning_rate": 2.7650776929154598e-05,
      "loss": 2.0639,
      "step": 16982
    },
    {
      "epoch": 0.447274163813537,
      "grad_norm": 2.4007728099823,
      "learning_rate": 2.764946010007901e-05,
      "loss": 1.3648,
      "step": 16983
    },
    {
      "epoch": 0.4473005003950487,
      "grad_norm": 2.9891598224639893,
      "learning_rate": 2.7648143271003422e-05,
      "loss": 1.4025,
      "step": 16984
    },
    {
      "epoch": 0.44732683697656045,
      "grad_norm": 2.407750368118286,
      "learning_rate": 2.7646826441927838e-05,
      "loss": 0.4806,
      "step": 16985
    },
    {
      "epoch": 0.44735317355807214,
      "grad_norm": 2.238537549972534,
      "learning_rate": 2.7645509612852254e-05,
      "loss": 2.125,
      "step": 16986
    },
    {
      "epoch": 0.4473795101395839,
      "grad_norm": 1.7120082378387451,
      "learning_rate": 2.764419278377667e-05,
      "loss": 2.0576,
      "step": 16987
    },
    {
      "epoch": 0.4474058467210956,
      "grad_norm": 3.8782999515533447,
      "learning_rate": 2.764287595470108e-05,
      "loss": 0.9971,
      "step": 16988
    },
    {
      "epoch": 0.4474321833026073,
      "grad_norm": 2.8225274085998535,
      "learning_rate": 2.7641559125625494e-05,
      "loss": 0.3103,
      "step": 16989
    },
    {
      "epoch": 0.44745851988411905,
      "grad_norm": 1.5613248348236084,
      "learning_rate": 2.7640242296549906e-05,
      "loss": 0.4772,
      "step": 16990
    },
    {
      "epoch": 0.44748485646563074,
      "grad_norm": 1.8757102489471436,
      "learning_rate": 2.7638925467474325e-05,
      "loss": 1.378,
      "step": 16991
    },
    {
      "epoch": 0.4475111930471425,
      "grad_norm": 1.6405189037322998,
      "learning_rate": 2.7637608638398737e-05,
      "loss": 1.7303,
      "step": 16992
    },
    {
      "epoch": 0.4475375296286542,
      "grad_norm": 5.172881126403809,
      "learning_rate": 2.763629180932315e-05,
      "loss": 1.3605,
      "step": 16993
    },
    {
      "epoch": 0.4475638662101659,
      "grad_norm": 1.9535326957702637,
      "learning_rate": 2.7634974980247565e-05,
      "loss": 1.4646,
      "step": 16994
    },
    {
      "epoch": 0.44759020279167766,
      "grad_norm": 1.6215181350708008,
      "learning_rate": 2.763365815117198e-05,
      "loss": 1.4769,
      "step": 16995
    },
    {
      "epoch": 0.44761653937318935,
      "grad_norm": 2.584655284881592,
      "learning_rate": 2.7632341322096396e-05,
      "loss": 2.0825,
      "step": 16996
    },
    {
      "epoch": 0.4476428759547011,
      "grad_norm": 3.165177822113037,
      "learning_rate": 2.7631024493020808e-05,
      "loss": 1.8981,
      "step": 16997
    },
    {
      "epoch": 0.4476692125362128,
      "grad_norm": 2.9137821197509766,
      "learning_rate": 2.762970766394522e-05,
      "loss": 2.0165,
      "step": 16998
    },
    {
      "epoch": 0.4476955491177245,
      "grad_norm": 2.2985646724700928,
      "learning_rate": 2.7628390834869633e-05,
      "loss": 1.7323,
      "step": 16999
    },
    {
      "epoch": 0.44772188569923627,
      "grad_norm": 2.334723472595215,
      "learning_rate": 2.762707400579405e-05,
      "loss": 2.57,
      "step": 17000
    },
    {
      "epoch": 0.44774822228074795,
      "grad_norm": 2.095701217651367,
      "learning_rate": 2.7625757176718464e-05,
      "loss": 1.963,
      "step": 17001
    },
    {
      "epoch": 0.4477745588622597,
      "grad_norm": 4.5567731857299805,
      "learning_rate": 2.7624440347642876e-05,
      "loss": 2.1609,
      "step": 17002
    },
    {
      "epoch": 0.4478008954437714,
      "grad_norm": 3.805265426635742,
      "learning_rate": 2.7623123518567288e-05,
      "loss": 1.3903,
      "step": 17003
    },
    {
      "epoch": 0.44782723202528313,
      "grad_norm": 1.9035910367965698,
      "learning_rate": 2.7621806689491704e-05,
      "loss": 2.1309,
      "step": 17004
    },
    {
      "epoch": 0.4478535686067948,
      "grad_norm": 3.8158488273620605,
      "learning_rate": 2.762048986041612e-05,
      "loss": 1.3963,
      "step": 17005
    },
    {
      "epoch": 0.44787990518830656,
      "grad_norm": 1.8790903091430664,
      "learning_rate": 2.7619173031340535e-05,
      "loss": 2.5806,
      "step": 17006
    },
    {
      "epoch": 0.4479062417698183,
      "grad_norm": 2.065577745437622,
      "learning_rate": 2.7617856202264947e-05,
      "loss": 1.6602,
      "step": 17007
    },
    {
      "epoch": 0.44793257835133,
      "grad_norm": 2.0469491481781006,
      "learning_rate": 2.761653937318936e-05,
      "loss": 1.9029,
      "step": 17008
    },
    {
      "epoch": 0.44795891493284173,
      "grad_norm": 1.4174691438674927,
      "learning_rate": 2.761522254411378e-05,
      "loss": 1.5197,
      "step": 17009
    },
    {
      "epoch": 0.4479852515143534,
      "grad_norm": 1.9062668085098267,
      "learning_rate": 2.761390571503819e-05,
      "loss": 1.5491,
      "step": 17010
    },
    {
      "epoch": 0.44801158809586517,
      "grad_norm": 1.7393251657485962,
      "learning_rate": 2.7612588885962603e-05,
      "loss": 1.7952,
      "step": 17011
    },
    {
      "epoch": 0.44803792467737685,
      "grad_norm": 1.9266389608383179,
      "learning_rate": 2.7611272056887015e-05,
      "loss": 1.7722,
      "step": 17012
    },
    {
      "epoch": 0.4480642612588886,
      "grad_norm": 2.281890630722046,
      "learning_rate": 2.760995522781143e-05,
      "loss": 1.9477,
      "step": 17013
    },
    {
      "epoch": 0.44809059784040034,
      "grad_norm": 3.4346511363983154,
      "learning_rate": 2.7608638398735846e-05,
      "loss": 1.1425,
      "step": 17014
    },
    {
      "epoch": 0.448116934421912,
      "grad_norm": 1.7877534627914429,
      "learning_rate": 2.7607321569660262e-05,
      "loss": 2.0013,
      "step": 17015
    },
    {
      "epoch": 0.44814327100342377,
      "grad_norm": 2.627500534057617,
      "learning_rate": 2.7606004740584674e-05,
      "loss": 1.4983,
      "step": 17016
    },
    {
      "epoch": 0.44816960758493546,
      "grad_norm": 2.3048906326293945,
      "learning_rate": 2.7604687911509086e-05,
      "loss": 1.6608,
      "step": 17017
    },
    {
      "epoch": 0.4481959441664472,
      "grad_norm": 1.485011339187622,
      "learning_rate": 2.76033710824335e-05,
      "loss": 1.573,
      "step": 17018
    },
    {
      "epoch": 0.4482222807479589,
      "grad_norm": 1.5875369310379028,
      "learning_rate": 2.7602054253357918e-05,
      "loss": 1.5857,
      "step": 17019
    },
    {
      "epoch": 0.44824861732947063,
      "grad_norm": 4.307639122009277,
      "learning_rate": 2.760073742428233e-05,
      "loss": 1.2746,
      "step": 17020
    },
    {
      "epoch": 0.4482749539109824,
      "grad_norm": 2.0398101806640625,
      "learning_rate": 2.7599420595206742e-05,
      "loss": 1.9623,
      "step": 17021
    },
    {
      "epoch": 0.44830129049249406,
      "grad_norm": 3.6416165828704834,
      "learning_rate": 2.7598103766131158e-05,
      "loss": 1.0117,
      "step": 17022
    },
    {
      "epoch": 0.4483276270740058,
      "grad_norm": 1.6380103826522827,
      "learning_rate": 2.7596786937055573e-05,
      "loss": 1.8844,
      "step": 17023
    },
    {
      "epoch": 0.4483539636555175,
      "grad_norm": 3.5386505126953125,
      "learning_rate": 2.759547010797999e-05,
      "loss": 1.7687,
      "step": 17024
    },
    {
      "epoch": 0.44838030023702924,
      "grad_norm": 2.216176748275757,
      "learning_rate": 2.75941532789044e-05,
      "loss": 0.6895,
      "step": 17025
    },
    {
      "epoch": 0.4484066368185409,
      "grad_norm": 1.5757381916046143,
      "learning_rate": 2.7592836449828813e-05,
      "loss": 1.4709,
      "step": 17026
    },
    {
      "epoch": 0.44843297340005267,
      "grad_norm": 2.5789008140563965,
      "learning_rate": 2.7591519620753225e-05,
      "loss": 1.5665,
      "step": 17027
    },
    {
      "epoch": 0.4484593099815644,
      "grad_norm": 3.134542465209961,
      "learning_rate": 2.7590202791677644e-05,
      "loss": 1.4515,
      "step": 17028
    },
    {
      "epoch": 0.4484856465630761,
      "grad_norm": 2.4615089893341064,
      "learning_rate": 2.7588885962602057e-05,
      "loss": 1.675,
      "step": 17029
    },
    {
      "epoch": 0.44851198314458784,
      "grad_norm": 4.316629886627197,
      "learning_rate": 2.758756913352647e-05,
      "loss": 1.9403,
      "step": 17030
    },
    {
      "epoch": 0.44853831972609953,
      "grad_norm": 1.775210976600647,
      "learning_rate": 2.758625230445088e-05,
      "loss": 2.1424,
      "step": 17031
    },
    {
      "epoch": 0.4485646563076113,
      "grad_norm": 3.5381767749786377,
      "learning_rate": 2.7584935475375297e-05,
      "loss": 0.805,
      "step": 17032
    },
    {
      "epoch": 0.448590992889123,
      "grad_norm": 1.6372946500778198,
      "learning_rate": 2.7583618646299712e-05,
      "loss": 1.1744,
      "step": 17033
    },
    {
      "epoch": 0.4486173294706347,
      "grad_norm": 2.1658003330230713,
      "learning_rate": 2.7582301817224128e-05,
      "loss": 1.9981,
      "step": 17034
    },
    {
      "epoch": 0.44864366605214645,
      "grad_norm": 2.8261516094207764,
      "learning_rate": 2.758098498814854e-05,
      "loss": 0.9075,
      "step": 17035
    },
    {
      "epoch": 0.44867000263365814,
      "grad_norm": 2.115002393722534,
      "learning_rate": 2.7579668159072952e-05,
      "loss": 2.0283,
      "step": 17036
    },
    {
      "epoch": 0.4486963392151699,
      "grad_norm": 5.122161865234375,
      "learning_rate": 2.7578351329997364e-05,
      "loss": 2.2937,
      "step": 17037
    },
    {
      "epoch": 0.44872267579668157,
      "grad_norm": 1.5687413215637207,
      "learning_rate": 2.7577034500921783e-05,
      "loss": 1.5673,
      "step": 17038
    },
    {
      "epoch": 0.4487490123781933,
      "grad_norm": 2.282477378845215,
      "learning_rate": 2.7575717671846196e-05,
      "loss": 1.4989,
      "step": 17039
    },
    {
      "epoch": 0.44877534895970506,
      "grad_norm": 1.9963371753692627,
      "learning_rate": 2.7574400842770608e-05,
      "loss": 2.1559,
      "step": 17040
    },
    {
      "epoch": 0.44880168554121674,
      "grad_norm": 3.23002028465271,
      "learning_rate": 2.7573084013695023e-05,
      "loss": 1.5559,
      "step": 17041
    },
    {
      "epoch": 0.4488280221227285,
      "grad_norm": 1.7360063791275024,
      "learning_rate": 2.757176718461944e-05,
      "loss": 2.0031,
      "step": 17042
    },
    {
      "epoch": 0.4488543587042402,
      "grad_norm": 3.6184260845184326,
      "learning_rate": 2.7570450355543855e-05,
      "loss": 1.2703,
      "step": 17043
    },
    {
      "epoch": 0.4488806952857519,
      "grad_norm": 1.872036337852478,
      "learning_rate": 2.7569133526468267e-05,
      "loss": 2.6207,
      "step": 17044
    },
    {
      "epoch": 0.4489070318672636,
      "grad_norm": 1.8810017108917236,
      "learning_rate": 2.756781669739268e-05,
      "loss": 1.5997,
      "step": 17045
    },
    {
      "epoch": 0.44893336844877535,
      "grad_norm": 2.8621761798858643,
      "learning_rate": 2.756649986831709e-05,
      "loss": 2.0346,
      "step": 17046
    },
    {
      "epoch": 0.4489597050302871,
      "grad_norm": 1.5929418802261353,
      "learning_rate": 2.756518303924151e-05,
      "loss": 2.2243,
      "step": 17047
    },
    {
      "epoch": 0.4489860416117988,
      "grad_norm": 2.0779407024383545,
      "learning_rate": 2.7563866210165922e-05,
      "loss": 1.8046,
      "step": 17048
    },
    {
      "epoch": 0.4490123781933105,
      "grad_norm": 1.6420633792877197,
      "learning_rate": 2.7562549381090335e-05,
      "loss": 1.7348,
      "step": 17049
    },
    {
      "epoch": 0.4490387147748222,
      "grad_norm": 2.041776418685913,
      "learning_rate": 2.7561232552014747e-05,
      "loss": 1.4654,
      "step": 17050
    },
    {
      "epoch": 0.44906505135633396,
      "grad_norm": 1.5560107231140137,
      "learning_rate": 2.7559915722939162e-05,
      "loss": 1.5134,
      "step": 17051
    },
    {
      "epoch": 0.44909138793784564,
      "grad_norm": 1.781523585319519,
      "learning_rate": 2.755859889386358e-05,
      "loss": 1.7816,
      "step": 17052
    },
    {
      "epoch": 0.4491177245193574,
      "grad_norm": 1.7064391374588013,
      "learning_rate": 2.7557282064787994e-05,
      "loss": 1.6485,
      "step": 17053
    },
    {
      "epoch": 0.44914406110086913,
      "grad_norm": 2.2161433696746826,
      "learning_rate": 2.7555965235712406e-05,
      "loss": 1.9475,
      "step": 17054
    },
    {
      "epoch": 0.4491703976823808,
      "grad_norm": 4.413087844848633,
      "learning_rate": 2.7554648406636818e-05,
      "loss": 2.0317,
      "step": 17055
    },
    {
      "epoch": 0.44919673426389256,
      "grad_norm": 5.025756359100342,
      "learning_rate": 2.7553331577561237e-05,
      "loss": 1.2113,
      "step": 17056
    },
    {
      "epoch": 0.44922307084540425,
      "grad_norm": 1.4062490463256836,
      "learning_rate": 2.755201474848565e-05,
      "loss": 1.5441,
      "step": 17057
    },
    {
      "epoch": 0.449249407426916,
      "grad_norm": 2.8166022300720215,
      "learning_rate": 2.755069791941006e-05,
      "loss": 0.5119,
      "step": 17058
    },
    {
      "epoch": 0.4492757440084277,
      "grad_norm": 4.330456256866455,
      "learning_rate": 2.7549381090334474e-05,
      "loss": 0.634,
      "step": 17059
    },
    {
      "epoch": 0.4493020805899394,
      "grad_norm": 1.5672374963760376,
      "learning_rate": 2.754806426125889e-05,
      "loss": 0.8187,
      "step": 17060
    },
    {
      "epoch": 0.44932841717145117,
      "grad_norm": 2.9968671798706055,
      "learning_rate": 2.7546747432183305e-05,
      "loss": 1.5142,
      "step": 17061
    },
    {
      "epoch": 0.44935475375296285,
      "grad_norm": 3.419109344482422,
      "learning_rate": 2.754543060310772e-05,
      "loss": 2.0302,
      "step": 17062
    },
    {
      "epoch": 0.4493810903344746,
      "grad_norm": 1.6215554475784302,
      "learning_rate": 2.7544113774032133e-05,
      "loss": 1.7236,
      "step": 17063
    },
    {
      "epoch": 0.4494074269159863,
      "grad_norm": 2.2841813564300537,
      "learning_rate": 2.7542796944956545e-05,
      "loss": 1.6661,
      "step": 17064
    },
    {
      "epoch": 0.44943376349749803,
      "grad_norm": 2.1528613567352295,
      "learning_rate": 2.7541480115880957e-05,
      "loss": 1.6542,
      "step": 17065
    },
    {
      "epoch": 0.44946010007900977,
      "grad_norm": 3.6613404750823975,
      "learning_rate": 2.7540163286805376e-05,
      "loss": 1.2189,
      "step": 17066
    },
    {
      "epoch": 0.44948643666052146,
      "grad_norm": 2.3041701316833496,
      "learning_rate": 2.7538846457729788e-05,
      "loss": 2.2027,
      "step": 17067
    },
    {
      "epoch": 0.4495127732420332,
      "grad_norm": 1.9654216766357422,
      "learning_rate": 2.75375296286542e-05,
      "loss": 1.6361,
      "step": 17068
    },
    {
      "epoch": 0.4495391098235449,
      "grad_norm": 2.672168016433716,
      "learning_rate": 2.7536212799578616e-05,
      "loss": 2.4648,
      "step": 17069
    },
    {
      "epoch": 0.44956544640505663,
      "grad_norm": 1.6782050132751465,
      "learning_rate": 2.7534895970503028e-05,
      "loss": 1.8492,
      "step": 17070
    },
    {
      "epoch": 0.4495917829865683,
      "grad_norm": 2.02051043510437,
      "learning_rate": 2.7533579141427447e-05,
      "loss": 0.6029,
      "step": 17071
    },
    {
      "epoch": 0.44961811956808007,
      "grad_norm": 4.1116862297058105,
      "learning_rate": 2.753226231235186e-05,
      "loss": 2.2187,
      "step": 17072
    },
    {
      "epoch": 0.4496444561495918,
      "grad_norm": 8.349372863769531,
      "learning_rate": 2.7530945483276272e-05,
      "loss": 1.546,
      "step": 17073
    },
    {
      "epoch": 0.4496707927311035,
      "grad_norm": 1.8985421657562256,
      "learning_rate": 2.7529628654200684e-05,
      "loss": 1.5,
      "step": 17074
    },
    {
      "epoch": 0.44969712931261524,
      "grad_norm": 4.587907791137695,
      "learning_rate": 2.7528311825125103e-05,
      "loss": 0.8677,
      "step": 17075
    },
    {
      "epoch": 0.44972346589412693,
      "grad_norm": 2.4465866088867188,
      "learning_rate": 2.7526994996049515e-05,
      "loss": 2.7048,
      "step": 17076
    },
    {
      "epoch": 0.44974980247563867,
      "grad_norm": 2.727853298187256,
      "learning_rate": 2.7525678166973927e-05,
      "loss": 0.6312,
      "step": 17077
    },
    {
      "epoch": 0.44977613905715036,
      "grad_norm": 3.176668405532837,
      "learning_rate": 2.752436133789834e-05,
      "loss": 1.3678,
      "step": 17078
    },
    {
      "epoch": 0.4498024756386621,
      "grad_norm": 2.50823974609375,
      "learning_rate": 2.7523044508822755e-05,
      "loss": 0.9341,
      "step": 17079
    },
    {
      "epoch": 0.44982881222017385,
      "grad_norm": 1.958359956741333,
      "learning_rate": 2.752172767974717e-05,
      "loss": 0.9493,
      "step": 17080
    },
    {
      "epoch": 0.44985514880168553,
      "grad_norm": 1.9944642782211304,
      "learning_rate": 2.7520410850671586e-05,
      "loss": 0.6526,
      "step": 17081
    },
    {
      "epoch": 0.4498814853831973,
      "grad_norm": 2.1743321418762207,
      "learning_rate": 2.7519094021596e-05,
      "loss": 2.2238,
      "step": 17082
    },
    {
      "epoch": 0.44990782196470896,
      "grad_norm": 3.004852294921875,
      "learning_rate": 2.751777719252041e-05,
      "loss": 1.8927,
      "step": 17083
    },
    {
      "epoch": 0.4499341585462207,
      "grad_norm": 1.7980796098709106,
      "learning_rate": 2.7516460363444823e-05,
      "loss": 0.5407,
      "step": 17084
    },
    {
      "epoch": 0.4499604951277324,
      "grad_norm": 1.8922024965286255,
      "learning_rate": 2.7515143534369242e-05,
      "loss": 1.9667,
      "step": 17085
    },
    {
      "epoch": 0.44998683170924414,
      "grad_norm": 2.068293809890747,
      "learning_rate": 2.7513826705293654e-05,
      "loss": 0.4991,
      "step": 17086
    },
    {
      "epoch": 0.4500131682907559,
      "grad_norm": 2.132648468017578,
      "learning_rate": 2.7512509876218066e-05,
      "loss": 1.8514,
      "step": 17087
    },
    {
      "epoch": 0.45003950487226757,
      "grad_norm": 1.6560941934585571,
      "learning_rate": 2.7511193047142482e-05,
      "loss": 1.515,
      "step": 17088
    },
    {
      "epoch": 0.4500658414537793,
      "grad_norm": 1.729984164237976,
      "learning_rate": 2.7509876218066898e-05,
      "loss": 1.789,
      "step": 17089
    },
    {
      "epoch": 0.450092178035291,
      "grad_norm": 3.2434918880462646,
      "learning_rate": 2.7508559388991313e-05,
      "loss": 1.7042,
      "step": 17090
    },
    {
      "epoch": 0.45011851461680275,
      "grad_norm": 1.3891850709915161,
      "learning_rate": 2.7507242559915725e-05,
      "loss": 1.9058,
      "step": 17091
    },
    {
      "epoch": 0.45014485119831443,
      "grad_norm": 1.4968249797821045,
      "learning_rate": 2.7505925730840138e-05,
      "loss": 2.0176,
      "step": 17092
    },
    {
      "epoch": 0.4501711877798262,
      "grad_norm": 2.123619794845581,
      "learning_rate": 2.750460890176455e-05,
      "loss": 1.4141,
      "step": 17093
    },
    {
      "epoch": 0.4501975243613379,
      "grad_norm": 1.6175100803375244,
      "learning_rate": 2.750329207268897e-05,
      "loss": 1.7863,
      "step": 17094
    },
    {
      "epoch": 0.4502238609428496,
      "grad_norm": 2.0571160316467285,
      "learning_rate": 2.750197524361338e-05,
      "loss": 1.7893,
      "step": 17095
    },
    {
      "epoch": 0.45025019752436135,
      "grad_norm": 3.049384593963623,
      "learning_rate": 2.7500658414537793e-05,
      "loss": 1.7014,
      "step": 17096
    },
    {
      "epoch": 0.45027653410587304,
      "grad_norm": 4.013835430145264,
      "learning_rate": 2.749934158546221e-05,
      "loss": 0.8901,
      "step": 17097
    },
    {
      "epoch": 0.4503028706873848,
      "grad_norm": 2.2871170043945312,
      "learning_rate": 2.749802475638662e-05,
      "loss": 1.2354,
      "step": 17098
    },
    {
      "epoch": 0.45032920726889647,
      "grad_norm": 1.9408328533172607,
      "learning_rate": 2.749670792731104e-05,
      "loss": 1.8038,
      "step": 17099
    },
    {
      "epoch": 0.4503555438504082,
      "grad_norm": 1.685389518737793,
      "learning_rate": 2.7495391098235452e-05,
      "loss": 2.1319,
      "step": 17100
    },
    {
      "epoch": 0.45038188043191996,
      "grad_norm": 2.0820672512054443,
      "learning_rate": 2.7494074269159864e-05,
      "loss": 1.7891,
      "step": 17101
    },
    {
      "epoch": 0.45040821701343164,
      "grad_norm": 1.918860673904419,
      "learning_rate": 2.7492757440084277e-05,
      "loss": 2.04,
      "step": 17102
    },
    {
      "epoch": 0.4504345535949434,
      "grad_norm": 3.4740636348724365,
      "learning_rate": 2.749144061100869e-05,
      "loss": 0.958,
      "step": 17103
    },
    {
      "epoch": 0.4504608901764551,
      "grad_norm": 3.9740121364593506,
      "learning_rate": 2.7490123781933108e-05,
      "loss": 1.8255,
      "step": 17104
    },
    {
      "epoch": 0.4504872267579668,
      "grad_norm": 2.113982915878296,
      "learning_rate": 2.748880695285752e-05,
      "loss": 1.3238,
      "step": 17105
    },
    {
      "epoch": 0.45051356333947856,
      "grad_norm": 4.4182047843933105,
      "learning_rate": 2.7487490123781932e-05,
      "loss": 0.8614,
      "step": 17106
    },
    {
      "epoch": 0.45053989992099025,
      "grad_norm": 1.969622254371643,
      "learning_rate": 2.7486173294706348e-05,
      "loss": 1.3336,
      "step": 17107
    },
    {
      "epoch": 0.450566236502502,
      "grad_norm": 2.5526673793792725,
      "learning_rate": 2.7484856465630763e-05,
      "loss": 1.6763,
      "step": 17108
    },
    {
      "epoch": 0.4505925730840137,
      "grad_norm": 5.080016136169434,
      "learning_rate": 2.748353963655518e-05,
      "loss": 1.8847,
      "step": 17109
    },
    {
      "epoch": 0.4506189096655254,
      "grad_norm": 3.4436593055725098,
      "learning_rate": 2.748222280747959e-05,
      "loss": 1.2964,
      "step": 17110
    },
    {
      "epoch": 0.4506452462470371,
      "grad_norm": 2.404585123062134,
      "learning_rate": 2.7480905978404003e-05,
      "loss": 2.4136,
      "step": 17111
    },
    {
      "epoch": 0.45067158282854886,
      "grad_norm": 1.9660687446594238,
      "learning_rate": 2.7479589149328416e-05,
      "loss": 2.5152,
      "step": 17112
    },
    {
      "epoch": 0.4506979194100606,
      "grad_norm": 1.7157256603240967,
      "learning_rate": 2.7478272320252835e-05,
      "loss": 1.9111,
      "step": 17113
    },
    {
      "epoch": 0.4507242559915723,
      "grad_norm": 1.777350902557373,
      "learning_rate": 2.7476955491177247e-05,
      "loss": 1.5846,
      "step": 17114
    },
    {
      "epoch": 0.45075059257308403,
      "grad_norm": 1.5402629375457764,
      "learning_rate": 2.747563866210166e-05,
      "loss": 1.2306,
      "step": 17115
    },
    {
      "epoch": 0.4507769291545957,
      "grad_norm": 1.6992673873901367,
      "learning_rate": 2.7474321833026075e-05,
      "loss": 1.742,
      "step": 17116
    },
    {
      "epoch": 0.45080326573610746,
      "grad_norm": 2.3362605571746826,
      "learning_rate": 2.7473005003950487e-05,
      "loss": 2.2791,
      "step": 17117
    },
    {
      "epoch": 0.45082960231761915,
      "grad_norm": 1.3017711639404297,
      "learning_rate": 2.7471688174874906e-05,
      "loss": 0.3301,
      "step": 17118
    },
    {
      "epoch": 0.4508559388991309,
      "grad_norm": 2.409926414489746,
      "learning_rate": 2.7470371345799318e-05,
      "loss": 2.1464,
      "step": 17119
    },
    {
      "epoch": 0.45088227548064264,
      "grad_norm": 2.541557788848877,
      "learning_rate": 2.746905451672373e-05,
      "loss": 0.843,
      "step": 17120
    },
    {
      "epoch": 0.4509086120621543,
      "grad_norm": 1.480195164680481,
      "learning_rate": 2.7467737687648142e-05,
      "loss": 2.0305,
      "step": 17121
    },
    {
      "epoch": 0.45093494864366607,
      "grad_norm": 4.073674201965332,
      "learning_rate": 2.746642085857256e-05,
      "loss": 1.6269,
      "step": 17122
    },
    {
      "epoch": 0.45096128522517775,
      "grad_norm": 1.9352281093597412,
      "learning_rate": 2.7465104029496974e-05,
      "loss": 1.1902,
      "step": 17123
    },
    {
      "epoch": 0.4509876218066895,
      "grad_norm": 1.8534413576126099,
      "learning_rate": 2.7463787200421386e-05,
      "loss": 1.9605,
      "step": 17124
    },
    {
      "epoch": 0.4510139583882012,
      "grad_norm": 3.0895206928253174,
      "learning_rate": 2.74624703713458e-05,
      "loss": 0.9396,
      "step": 17125
    },
    {
      "epoch": 0.45104029496971293,
      "grad_norm": 3.2080650329589844,
      "learning_rate": 2.7461153542270214e-05,
      "loss": 1.5976,
      "step": 17126
    },
    {
      "epoch": 0.4510666315512247,
      "grad_norm": 1.801740050315857,
      "learning_rate": 2.7459836713194633e-05,
      "loss": 2.035,
      "step": 17127
    },
    {
      "epoch": 0.45109296813273636,
      "grad_norm": 2.3313302993774414,
      "learning_rate": 2.7458519884119045e-05,
      "loss": 1.7296,
      "step": 17128
    },
    {
      "epoch": 0.4511193047142481,
      "grad_norm": 2.4068751335144043,
      "learning_rate": 2.7457203055043457e-05,
      "loss": 2.4214,
      "step": 17129
    },
    {
      "epoch": 0.4511456412957598,
      "grad_norm": 1.6548713445663452,
      "learning_rate": 2.745588622596787e-05,
      "loss": 1.3809,
      "step": 17130
    },
    {
      "epoch": 0.45117197787727154,
      "grad_norm": 1.804884433746338,
      "learning_rate": 2.745456939689228e-05,
      "loss": 1.3989,
      "step": 17131
    },
    {
      "epoch": 0.4511983144587832,
      "grad_norm": 2.324575424194336,
      "learning_rate": 2.74532525678167e-05,
      "loss": 1.8062,
      "step": 17132
    },
    {
      "epoch": 0.45122465104029497,
      "grad_norm": 2.1455283164978027,
      "learning_rate": 2.7451935738741113e-05,
      "loss": 0.7217,
      "step": 17133
    },
    {
      "epoch": 0.4512509876218067,
      "grad_norm": 2.6046602725982666,
      "learning_rate": 2.7450618909665525e-05,
      "loss": 1.4877,
      "step": 17134
    },
    {
      "epoch": 0.4512773242033184,
      "grad_norm": 2.62692928314209,
      "learning_rate": 2.744930208058994e-05,
      "loss": 1.9191,
      "step": 17135
    },
    {
      "epoch": 0.45130366078483014,
      "grad_norm": 1.8383461236953735,
      "learning_rate": 2.7447985251514353e-05,
      "loss": 1.6445,
      "step": 17136
    },
    {
      "epoch": 0.45132999736634183,
      "grad_norm": 2.3609025478363037,
      "learning_rate": 2.7446668422438772e-05,
      "loss": 2.0814,
      "step": 17137
    },
    {
      "epoch": 0.45135633394785357,
      "grad_norm": 4.005642890930176,
      "learning_rate": 2.7445351593363184e-05,
      "loss": 1.7183,
      "step": 17138
    },
    {
      "epoch": 0.4513826705293653,
      "grad_norm": 1.4061229228973389,
      "learning_rate": 2.7444034764287596e-05,
      "loss": 1.2178,
      "step": 17139
    },
    {
      "epoch": 0.451409007110877,
      "grad_norm": 2.076510190963745,
      "learning_rate": 2.744271793521201e-05,
      "loss": 2.2251,
      "step": 17140
    },
    {
      "epoch": 0.45143534369238875,
      "grad_norm": 2.146402597427368,
      "learning_rate": 2.7441401106136427e-05,
      "loss": 2.0432,
      "step": 17141
    },
    {
      "epoch": 0.45146168027390043,
      "grad_norm": 2.8456602096557617,
      "learning_rate": 2.744008427706084e-05,
      "loss": 0.889,
      "step": 17142
    },
    {
      "epoch": 0.4514880168554122,
      "grad_norm": 2.419118881225586,
      "learning_rate": 2.7438767447985252e-05,
      "loss": 1.2221,
      "step": 17143
    },
    {
      "epoch": 0.45151435343692387,
      "grad_norm": 1.3562119007110596,
      "learning_rate": 2.7437450618909667e-05,
      "loss": 0.3291,
      "step": 17144
    },
    {
      "epoch": 0.4515406900184356,
      "grad_norm": 2.6415436267852783,
      "learning_rate": 2.743613378983408e-05,
      "loss": 1.3642,
      "step": 17145
    },
    {
      "epoch": 0.45156702659994735,
      "grad_norm": 1.6824222803115845,
      "learning_rate": 2.74348169607585e-05,
      "loss": 1.5869,
      "step": 17146
    },
    {
      "epoch": 0.45159336318145904,
      "grad_norm": 1.6827759742736816,
      "learning_rate": 2.743350013168291e-05,
      "loss": 1.4758,
      "step": 17147
    },
    {
      "epoch": 0.4516196997629708,
      "grad_norm": 3.5246641635894775,
      "learning_rate": 2.7432183302607323e-05,
      "loss": 2.3005,
      "step": 17148
    },
    {
      "epoch": 0.45164603634448247,
      "grad_norm": 2.109797954559326,
      "learning_rate": 2.7430866473531735e-05,
      "loss": 1.4442,
      "step": 17149
    },
    {
      "epoch": 0.4516723729259942,
      "grad_norm": 2.881183385848999,
      "learning_rate": 2.7429549644456147e-05,
      "loss": 1.8403,
      "step": 17150
    },
    {
      "epoch": 0.4516987095075059,
      "grad_norm": 3.3041598796844482,
      "learning_rate": 2.7428232815380566e-05,
      "loss": 1.2424,
      "step": 17151
    },
    {
      "epoch": 0.45172504608901765,
      "grad_norm": 3.8640763759613037,
      "learning_rate": 2.742691598630498e-05,
      "loss": 0.5971,
      "step": 17152
    },
    {
      "epoch": 0.4517513826705294,
      "grad_norm": 3.526885986328125,
      "learning_rate": 2.742559915722939e-05,
      "loss": 1.8192,
      "step": 17153
    },
    {
      "epoch": 0.4517777192520411,
      "grad_norm": 9.659627914428711,
      "learning_rate": 2.7424282328153806e-05,
      "loss": 1.1899,
      "step": 17154
    },
    {
      "epoch": 0.4518040558335528,
      "grad_norm": 2.479926347732544,
      "learning_rate": 2.7422965499078222e-05,
      "loss": 1.5591,
      "step": 17155
    },
    {
      "epoch": 0.4518303924150645,
      "grad_norm": 3.6006155014038086,
      "learning_rate": 2.7421648670002638e-05,
      "loss": 1.4015,
      "step": 17156
    },
    {
      "epoch": 0.45185672899657625,
      "grad_norm": 1.6525331735610962,
      "learning_rate": 2.742033184092705e-05,
      "loss": 1.9961,
      "step": 17157
    },
    {
      "epoch": 0.45188306557808794,
      "grad_norm": 1.9206260442733765,
      "learning_rate": 2.7419015011851462e-05,
      "loss": 1.7597,
      "step": 17158
    },
    {
      "epoch": 0.4519094021595997,
      "grad_norm": 3.4888410568237305,
      "learning_rate": 2.7417698182775874e-05,
      "loss": 1.6017,
      "step": 17159
    },
    {
      "epoch": 0.4519357387411114,
      "grad_norm": 3.253868341445923,
      "learning_rate": 2.7416381353700293e-05,
      "loss": 1.4429,
      "step": 17160
    },
    {
      "epoch": 0.4519620753226231,
      "grad_norm": 1.461851716041565,
      "learning_rate": 2.7415064524624705e-05,
      "loss": 2.1199,
      "step": 17161
    },
    {
      "epoch": 0.45198841190413486,
      "grad_norm": 3.6245734691619873,
      "learning_rate": 2.7413747695549118e-05,
      "loss": 1.3559,
      "step": 17162
    },
    {
      "epoch": 0.45201474848564654,
      "grad_norm": 2.16041898727417,
      "learning_rate": 2.7412430866473533e-05,
      "loss": 1.7796,
      "step": 17163
    },
    {
      "epoch": 0.4520410850671583,
      "grad_norm": 2.043455123901367,
      "learning_rate": 2.7411114037397945e-05,
      "loss": 1.9891,
      "step": 17164
    },
    {
      "epoch": 0.45206742164867,
      "grad_norm": 1.6806464195251465,
      "learning_rate": 2.7409797208322364e-05,
      "loss": 1.8386,
      "step": 17165
    },
    {
      "epoch": 0.4520937582301817,
      "grad_norm": 1.6694210767745972,
      "learning_rate": 2.7408480379246777e-05,
      "loss": 1.8001,
      "step": 17166
    },
    {
      "epoch": 0.45212009481169346,
      "grad_norm": 1.7431142330169678,
      "learning_rate": 2.740716355017119e-05,
      "loss": 1.2596,
      "step": 17167
    },
    {
      "epoch": 0.45214643139320515,
      "grad_norm": 1.8233349323272705,
      "learning_rate": 2.74058467210956e-05,
      "loss": 1.8809,
      "step": 17168
    },
    {
      "epoch": 0.4521727679747169,
      "grad_norm": 1.9612683057785034,
      "learning_rate": 2.7404529892020013e-05,
      "loss": 2.1106,
      "step": 17169
    },
    {
      "epoch": 0.4521991045562286,
      "grad_norm": 5.293704509735107,
      "learning_rate": 2.7403213062944432e-05,
      "loss": 1.6231,
      "step": 17170
    },
    {
      "epoch": 0.4522254411377403,
      "grad_norm": 2.054372549057007,
      "learning_rate": 2.7401896233868844e-05,
      "loss": 1.7198,
      "step": 17171
    },
    {
      "epoch": 0.45225177771925207,
      "grad_norm": 2.097214937210083,
      "learning_rate": 2.740057940479326e-05,
      "loss": 1.6886,
      "step": 17172
    },
    {
      "epoch": 0.45227811430076376,
      "grad_norm": 1.6854910850524902,
      "learning_rate": 2.7399262575717672e-05,
      "loss": 1.6825,
      "step": 17173
    },
    {
      "epoch": 0.4523044508822755,
      "grad_norm": 1.6675678491592407,
      "learning_rate": 2.739794574664209e-05,
      "loss": 1.4672,
      "step": 17174
    },
    {
      "epoch": 0.4523307874637872,
      "grad_norm": 2.632181167602539,
      "learning_rate": 2.7396628917566503e-05,
      "loss": 1.7371,
      "step": 17175
    },
    {
      "epoch": 0.45235712404529893,
      "grad_norm": 2.109461784362793,
      "learning_rate": 2.7395312088490916e-05,
      "loss": 1.8966,
      "step": 17176
    },
    {
      "epoch": 0.4523834606268106,
      "grad_norm": 2.290513277053833,
      "learning_rate": 2.7393995259415328e-05,
      "loss": 0.8599,
      "step": 17177
    },
    {
      "epoch": 0.45240979720832236,
      "grad_norm": 1.9919025897979736,
      "learning_rate": 2.739267843033974e-05,
      "loss": 0.7075,
      "step": 17178
    },
    {
      "epoch": 0.4524361337898341,
      "grad_norm": 2.0513458251953125,
      "learning_rate": 2.739136160126416e-05,
      "loss": 2.245,
      "step": 17179
    },
    {
      "epoch": 0.4524624703713458,
      "grad_norm": 2.7437376976013184,
      "learning_rate": 2.739004477218857e-05,
      "loss": 1.2284,
      "step": 17180
    },
    {
      "epoch": 0.45248880695285754,
      "grad_norm": 2.015800952911377,
      "learning_rate": 2.7388727943112983e-05,
      "loss": 1.4751,
      "step": 17181
    },
    {
      "epoch": 0.4525151435343692,
      "grad_norm": 1.7488065958023071,
      "learning_rate": 2.73874111140374e-05,
      "loss": 2.5183,
      "step": 17182
    },
    {
      "epoch": 0.45254148011588097,
      "grad_norm": 3.3676867485046387,
      "learning_rate": 2.738609428496181e-05,
      "loss": 1.5646,
      "step": 17183
    },
    {
      "epoch": 0.45256781669739266,
      "grad_norm": 1.915235996246338,
      "learning_rate": 2.738477745588623e-05,
      "loss": 1.5624,
      "step": 17184
    },
    {
      "epoch": 0.4525941532789044,
      "grad_norm": 1.3848034143447876,
      "learning_rate": 2.7383460626810642e-05,
      "loss": 1.2423,
      "step": 17185
    },
    {
      "epoch": 0.45262048986041614,
      "grad_norm": 3.018489122390747,
      "learning_rate": 2.7382143797735055e-05,
      "loss": 2.4648,
      "step": 17186
    },
    {
      "epoch": 0.45264682644192783,
      "grad_norm": 1.512101650238037,
      "learning_rate": 2.7380826968659467e-05,
      "loss": 1.5661,
      "step": 17187
    },
    {
      "epoch": 0.4526731630234396,
      "grad_norm": 3.8546786308288574,
      "learning_rate": 2.7379510139583886e-05,
      "loss": 0.9695,
      "step": 17188
    },
    {
      "epoch": 0.45269949960495126,
      "grad_norm": 2.109097719192505,
      "learning_rate": 2.7378193310508298e-05,
      "loss": 1.5202,
      "step": 17189
    },
    {
      "epoch": 0.452725836186463,
      "grad_norm": 3.920694589614868,
      "learning_rate": 2.737687648143271e-05,
      "loss": 2.5446,
      "step": 17190
    },
    {
      "epoch": 0.4527521727679747,
      "grad_norm": 1.7120757102966309,
      "learning_rate": 2.7375559652357126e-05,
      "loss": 1.5632,
      "step": 17191
    },
    {
      "epoch": 0.45277850934948644,
      "grad_norm": 1.8904166221618652,
      "learning_rate": 2.7374242823281538e-05,
      "loss": 1.6217,
      "step": 17192
    },
    {
      "epoch": 0.4528048459309982,
      "grad_norm": 5.083609104156494,
      "learning_rate": 2.7372925994205957e-05,
      "loss": 2.3104,
      "step": 17193
    },
    {
      "epoch": 0.45283118251250987,
      "grad_norm": 1.9159715175628662,
      "learning_rate": 2.737160916513037e-05,
      "loss": 1.9583,
      "step": 17194
    },
    {
      "epoch": 0.4528575190940216,
      "grad_norm": 1.686370849609375,
      "learning_rate": 2.737029233605478e-05,
      "loss": 1.5409,
      "step": 17195
    },
    {
      "epoch": 0.4528838556755333,
      "grad_norm": 2.9615120887756348,
      "learning_rate": 2.7368975506979194e-05,
      "loss": 0.836,
      "step": 17196
    },
    {
      "epoch": 0.45291019225704504,
      "grad_norm": 2.1956305503845215,
      "learning_rate": 2.7367658677903606e-05,
      "loss": 1.726,
      "step": 17197
    },
    {
      "epoch": 0.45293652883855673,
      "grad_norm": 2.52850079536438,
      "learning_rate": 2.7366341848828025e-05,
      "loss": 0.328,
      "step": 17198
    },
    {
      "epoch": 0.4529628654200685,
      "grad_norm": 1.5329182147979736,
      "learning_rate": 2.7365025019752437e-05,
      "loss": 2.0223,
      "step": 17199
    },
    {
      "epoch": 0.4529892020015802,
      "grad_norm": 2.016688346862793,
      "learning_rate": 2.7363708190676853e-05,
      "loss": 1.279,
      "step": 17200
    },
    {
      "epoch": 0.4530155385830919,
      "grad_norm": 1.8355118036270142,
      "learning_rate": 2.7362391361601265e-05,
      "loss": 1.5419,
      "step": 17201
    },
    {
      "epoch": 0.45304187516460365,
      "grad_norm": 1.8816572427749634,
      "learning_rate": 2.7361074532525677e-05,
      "loss": 1.8123,
      "step": 17202
    },
    {
      "epoch": 0.45306821174611533,
      "grad_norm": 1.9784739017486572,
      "learning_rate": 2.7359757703450096e-05,
      "loss": 1.827,
      "step": 17203
    },
    {
      "epoch": 0.4530945483276271,
      "grad_norm": 3.907982349395752,
      "learning_rate": 2.735844087437451e-05,
      "loss": 1.9803,
      "step": 17204
    },
    {
      "epoch": 0.4531208849091388,
      "grad_norm": 2.0782470703125,
      "learning_rate": 2.735712404529892e-05,
      "loss": 1.6839,
      "step": 17205
    },
    {
      "epoch": 0.4531472214906505,
      "grad_norm": 1.8123726844787598,
      "learning_rate": 2.7355807216223333e-05,
      "loss": 1.8258,
      "step": 17206
    },
    {
      "epoch": 0.45317355807216225,
      "grad_norm": 1.8910233974456787,
      "learning_rate": 2.7354490387147752e-05,
      "loss": 2.1613,
      "step": 17207
    },
    {
      "epoch": 0.45319989465367394,
      "grad_norm": 1.8334705829620361,
      "learning_rate": 2.7353173558072164e-05,
      "loss": 1.9071,
      "step": 17208
    },
    {
      "epoch": 0.4532262312351857,
      "grad_norm": 2.4042649269104004,
      "learning_rate": 2.7351856728996576e-05,
      "loss": 1.6075,
      "step": 17209
    },
    {
      "epoch": 0.45325256781669737,
      "grad_norm": 4.0542473793029785,
      "learning_rate": 2.7350539899920992e-05,
      "loss": 2.3992,
      "step": 17210
    },
    {
      "epoch": 0.4532789043982091,
      "grad_norm": 2.590764045715332,
      "learning_rate": 2.7349223070845404e-05,
      "loss": 1.7578,
      "step": 17211
    },
    {
      "epoch": 0.45330524097972086,
      "grad_norm": 2.123406171798706,
      "learning_rate": 2.7347906241769823e-05,
      "loss": 1.0294,
      "step": 17212
    },
    {
      "epoch": 0.45333157756123255,
      "grad_norm": 1.5296351909637451,
      "learning_rate": 2.7346589412694235e-05,
      "loss": 1.3012,
      "step": 17213
    },
    {
      "epoch": 0.4533579141427443,
      "grad_norm": 2.7377827167510986,
      "learning_rate": 2.7345272583618647e-05,
      "loss": 1.2522,
      "step": 17214
    },
    {
      "epoch": 0.453384250724256,
      "grad_norm": 2.5996670722961426,
      "learning_rate": 2.734395575454306e-05,
      "loss": 2.3996,
      "step": 17215
    },
    {
      "epoch": 0.4534105873057677,
      "grad_norm": 2.616281509399414,
      "learning_rate": 2.7342638925467472e-05,
      "loss": 1.3469,
      "step": 17216
    },
    {
      "epoch": 0.4534369238872794,
      "grad_norm": 2.360745429992676,
      "learning_rate": 2.734132209639189e-05,
      "loss": 0.8809,
      "step": 17217
    },
    {
      "epoch": 0.45346326046879115,
      "grad_norm": 1.5613644123077393,
      "learning_rate": 2.7340005267316303e-05,
      "loss": 1.4582,
      "step": 17218
    },
    {
      "epoch": 0.4534895970503029,
      "grad_norm": 2.0324182510375977,
      "learning_rate": 2.733868843824072e-05,
      "loss": 1.5494,
      "step": 17219
    },
    {
      "epoch": 0.4535159336318146,
      "grad_norm": 3.080385208129883,
      "learning_rate": 2.733737160916513e-05,
      "loss": 1.1272,
      "step": 17220
    },
    {
      "epoch": 0.4535422702133263,
      "grad_norm": 4.435571670532227,
      "learning_rate": 2.733605478008955e-05,
      "loss": 1.8973,
      "step": 17221
    },
    {
      "epoch": 0.453568606794838,
      "grad_norm": 1.5928090810775757,
      "learning_rate": 2.7334737951013962e-05,
      "loss": 1.6691,
      "step": 17222
    },
    {
      "epoch": 0.45359494337634976,
      "grad_norm": 1.9368505477905273,
      "learning_rate": 2.7333421121938374e-05,
      "loss": 1.7477,
      "step": 17223
    },
    {
      "epoch": 0.45362127995786145,
      "grad_norm": 2.5404584407806396,
      "learning_rate": 2.7332104292862786e-05,
      "loss": 2.0429,
      "step": 17224
    },
    {
      "epoch": 0.4536476165393732,
      "grad_norm": 1.8371343612670898,
      "learning_rate": 2.73307874637872e-05,
      "loss": 1.9979,
      "step": 17225
    },
    {
      "epoch": 0.45367395312088493,
      "grad_norm": 2.2765278816223145,
      "learning_rate": 2.7329470634711618e-05,
      "loss": 1.1807,
      "step": 17226
    },
    {
      "epoch": 0.4537002897023966,
      "grad_norm": 2.139545202255249,
      "learning_rate": 2.732815380563603e-05,
      "loss": 1.8717,
      "step": 17227
    },
    {
      "epoch": 0.45372662628390836,
      "grad_norm": 1.8141471147537231,
      "learning_rate": 2.7326836976560445e-05,
      "loss": 2.3454,
      "step": 17228
    },
    {
      "epoch": 0.45375296286542005,
      "grad_norm": 6.5997161865234375,
      "learning_rate": 2.7325520147484858e-05,
      "loss": 1.2423,
      "step": 17229
    },
    {
      "epoch": 0.4537792994469318,
      "grad_norm": 2.7536022663116455,
      "learning_rate": 2.732420331840927e-05,
      "loss": 1.9389,
      "step": 17230
    },
    {
      "epoch": 0.4538056360284435,
      "grad_norm": 1.8371065855026245,
      "learning_rate": 2.732288648933369e-05,
      "loss": 1.7975,
      "step": 17231
    },
    {
      "epoch": 0.4538319726099552,
      "grad_norm": 2.3389861583709717,
      "learning_rate": 2.73215696602581e-05,
      "loss": 1.3338,
      "step": 17232
    },
    {
      "epoch": 0.45385830919146697,
      "grad_norm": 2.5396571159362793,
      "learning_rate": 2.7320252831182513e-05,
      "loss": 2.6465,
      "step": 17233
    },
    {
      "epoch": 0.45388464577297866,
      "grad_norm": 2.511237382888794,
      "learning_rate": 2.7318936002106925e-05,
      "loss": 0.6535,
      "step": 17234
    },
    {
      "epoch": 0.4539109823544904,
      "grad_norm": 1.5506068468093872,
      "learning_rate": 2.7317619173031338e-05,
      "loss": 2.3247,
      "step": 17235
    },
    {
      "epoch": 0.4539373189360021,
      "grad_norm": 2.2335588932037354,
      "learning_rate": 2.7316302343955757e-05,
      "loss": 2.3492,
      "step": 17236
    },
    {
      "epoch": 0.45396365551751383,
      "grad_norm": 4.997518539428711,
      "learning_rate": 2.731498551488017e-05,
      "loss": 1.336,
      "step": 17237
    },
    {
      "epoch": 0.4539899920990255,
      "grad_norm": 3.5385055541992188,
      "learning_rate": 2.7313668685804584e-05,
      "loss": 1.5828,
      "step": 17238
    },
    {
      "epoch": 0.45401632868053726,
      "grad_norm": 2.7677905559539795,
      "learning_rate": 2.7312351856728997e-05,
      "loss": 1.4542,
      "step": 17239
    },
    {
      "epoch": 0.454042665262049,
      "grad_norm": 1.9444832801818848,
      "learning_rate": 2.7311035027653416e-05,
      "loss": 1.2108,
      "step": 17240
    },
    {
      "epoch": 0.4540690018435607,
      "grad_norm": 5.952491760253906,
      "learning_rate": 2.7309718198577828e-05,
      "loss": 1.1722,
      "step": 17241
    },
    {
      "epoch": 0.45409533842507244,
      "grad_norm": 1.6658071279525757,
      "learning_rate": 2.730840136950224e-05,
      "loss": 0.3918,
      "step": 17242
    },
    {
      "epoch": 0.4541216750065841,
      "grad_norm": 2.8430464267730713,
      "learning_rate": 2.7307084540426652e-05,
      "loss": 1.4477,
      "step": 17243
    },
    {
      "epoch": 0.45414801158809587,
      "grad_norm": 4.561838626861572,
      "learning_rate": 2.7305767711351064e-05,
      "loss": 0.9918,
      "step": 17244
    },
    {
      "epoch": 0.4541743481696076,
      "grad_norm": 4.652225494384766,
      "learning_rate": 2.7304450882275483e-05,
      "loss": 1.3644,
      "step": 17245
    },
    {
      "epoch": 0.4542006847511193,
      "grad_norm": 2.326185941696167,
      "learning_rate": 2.7303134053199896e-05,
      "loss": 1.5605,
      "step": 17246
    },
    {
      "epoch": 0.45422702133263104,
      "grad_norm": 2.047363758087158,
      "learning_rate": 2.730181722412431e-05,
      "loss": 1.9757,
      "step": 17247
    },
    {
      "epoch": 0.45425335791414273,
      "grad_norm": 1.630068302154541,
      "learning_rate": 2.7300500395048723e-05,
      "loss": 1.8384,
      "step": 17248
    },
    {
      "epoch": 0.4542796944956545,
      "grad_norm": 1.7588789463043213,
      "learning_rate": 2.7299183565973136e-05,
      "loss": 0.6989,
      "step": 17249
    },
    {
      "epoch": 0.45430603107716616,
      "grad_norm": 4.718081951141357,
      "learning_rate": 2.7297866736897555e-05,
      "loss": 1.2469,
      "step": 17250
    },
    {
      "epoch": 0.4543323676586779,
      "grad_norm": 3.3542864322662354,
      "learning_rate": 2.7296549907821967e-05,
      "loss": 1.7339,
      "step": 17251
    },
    {
      "epoch": 0.45435870424018965,
      "grad_norm": 2.1271185874938965,
      "learning_rate": 2.729523307874638e-05,
      "loss": 1.877,
      "step": 17252
    },
    {
      "epoch": 0.45438504082170134,
      "grad_norm": 2.3422200679779053,
      "learning_rate": 2.729391624967079e-05,
      "loss": 1.3132,
      "step": 17253
    },
    {
      "epoch": 0.4544113774032131,
      "grad_norm": 1.6371201276779175,
      "learning_rate": 2.729259942059521e-05,
      "loss": 1.5208,
      "step": 17254
    },
    {
      "epoch": 0.45443771398472477,
      "grad_norm": 2.037470817565918,
      "learning_rate": 2.7291282591519622e-05,
      "loss": 1.7234,
      "step": 17255
    },
    {
      "epoch": 0.4544640505662365,
      "grad_norm": 4.676828861236572,
      "learning_rate": 2.7289965762444035e-05,
      "loss": 2.13,
      "step": 17256
    },
    {
      "epoch": 0.4544903871477482,
      "grad_norm": 2.1511476039886475,
      "learning_rate": 2.728864893336845e-05,
      "loss": 1.208,
      "step": 17257
    },
    {
      "epoch": 0.45451672372925994,
      "grad_norm": 1.7112656831741333,
      "learning_rate": 2.7287332104292863e-05,
      "loss": 1.1046,
      "step": 17258
    },
    {
      "epoch": 0.4545430603107717,
      "grad_norm": 1.607563853263855,
      "learning_rate": 2.728601527521728e-05,
      "loss": 1.2543,
      "step": 17259
    },
    {
      "epoch": 0.4545693968922834,
      "grad_norm": 2.384552478790283,
      "learning_rate": 2.7284698446141694e-05,
      "loss": 1.6309,
      "step": 17260
    },
    {
      "epoch": 0.4545957334737951,
      "grad_norm": 4.566499710083008,
      "learning_rate": 2.7283381617066106e-05,
      "loss": 1.6098,
      "step": 17261
    },
    {
      "epoch": 0.4546220700553068,
      "grad_norm": 2.1666014194488525,
      "learning_rate": 2.7282064787990518e-05,
      "loss": 1.3774,
      "step": 17262
    },
    {
      "epoch": 0.45464840663681855,
      "grad_norm": 2.6022233963012695,
      "learning_rate": 2.728074795891493e-05,
      "loss": 1.7932,
      "step": 17263
    },
    {
      "epoch": 0.45467474321833024,
      "grad_norm": 2.702962875366211,
      "learning_rate": 2.727943112983935e-05,
      "loss": 1.159,
      "step": 17264
    },
    {
      "epoch": 0.454701079799842,
      "grad_norm": 3.4606845378875732,
      "learning_rate": 2.727811430076376e-05,
      "loss": 1.7029,
      "step": 17265
    },
    {
      "epoch": 0.4547274163813537,
      "grad_norm": 3.5899014472961426,
      "learning_rate": 2.7276797471688177e-05,
      "loss": 1.2461,
      "step": 17266
    },
    {
      "epoch": 0.4547537529628654,
      "grad_norm": 2.5236380100250244,
      "learning_rate": 2.727548064261259e-05,
      "loss": 2.1335,
      "step": 17267
    },
    {
      "epoch": 0.45478008954437715,
      "grad_norm": 2.2835521697998047,
      "learning_rate": 2.7274163813537e-05,
      "loss": 1.9793,
      "step": 17268
    },
    {
      "epoch": 0.45480642612588884,
      "grad_norm": 6.296321392059326,
      "learning_rate": 2.727284698446142e-05,
      "loss": 2.0144,
      "step": 17269
    },
    {
      "epoch": 0.4548327627074006,
      "grad_norm": 2.8443493843078613,
      "learning_rate": 2.7271530155385833e-05,
      "loss": 2.0607,
      "step": 17270
    },
    {
      "epoch": 0.45485909928891227,
      "grad_norm": 2.301936388015747,
      "learning_rate": 2.7270213326310245e-05,
      "loss": 2.331,
      "step": 17271
    },
    {
      "epoch": 0.454885435870424,
      "grad_norm": 2.5129923820495605,
      "learning_rate": 2.7268896497234657e-05,
      "loss": 0.8484,
      "step": 17272
    },
    {
      "epoch": 0.45491177245193576,
      "grad_norm": 1.8395601511001587,
      "learning_rate": 2.7267579668159076e-05,
      "loss": 0.8047,
      "step": 17273
    },
    {
      "epoch": 0.45493810903344745,
      "grad_norm": 1.784755825996399,
      "learning_rate": 2.726626283908349e-05,
      "loss": 1.9117,
      "step": 17274
    },
    {
      "epoch": 0.4549644456149592,
      "grad_norm": 2.107417583465576,
      "learning_rate": 2.7264946010007904e-05,
      "loss": 0.4097,
      "step": 17275
    },
    {
      "epoch": 0.4549907821964709,
      "grad_norm": 1.615294098854065,
      "learning_rate": 2.7263629180932316e-05,
      "loss": 1.5856,
      "step": 17276
    },
    {
      "epoch": 0.4550171187779826,
      "grad_norm": 3.5520570278167725,
      "learning_rate": 2.726231235185673e-05,
      "loss": 1.2245,
      "step": 17277
    },
    {
      "epoch": 0.45504345535949436,
      "grad_norm": 1.9660377502441406,
      "learning_rate": 2.7260995522781147e-05,
      "loss": 2.0264,
      "step": 17278
    },
    {
      "epoch": 0.45506979194100605,
      "grad_norm": 2.0342516899108887,
      "learning_rate": 2.725967869370556e-05,
      "loss": 2.1092,
      "step": 17279
    },
    {
      "epoch": 0.4550961285225178,
      "grad_norm": 2.382925033569336,
      "learning_rate": 2.7258361864629972e-05,
      "loss": 1.9994,
      "step": 17280
    },
    {
      "epoch": 0.4551224651040295,
      "grad_norm": 2.6056907176971436,
      "learning_rate": 2.7257045035554384e-05,
      "loss": 1.5397,
      "step": 17281
    },
    {
      "epoch": 0.4551488016855412,
      "grad_norm": 2.2213563919067383,
      "learning_rate": 2.7255728206478796e-05,
      "loss": 1.6928,
      "step": 17282
    },
    {
      "epoch": 0.4551751382670529,
      "grad_norm": 6.961357116699219,
      "learning_rate": 2.7254411377403215e-05,
      "loss": 2.4899,
      "step": 17283
    },
    {
      "epoch": 0.45520147484856466,
      "grad_norm": 4.042150974273682,
      "learning_rate": 2.7253094548327627e-05,
      "loss": 1.0525,
      "step": 17284
    },
    {
      "epoch": 0.4552278114300764,
      "grad_norm": 2.232088088989258,
      "learning_rate": 2.7251777719252043e-05,
      "loss": 1.8718,
      "step": 17285
    },
    {
      "epoch": 0.4552541480115881,
      "grad_norm": 1.5000238418579102,
      "learning_rate": 2.7250460890176455e-05,
      "loss": 1.59,
      "step": 17286
    },
    {
      "epoch": 0.45528048459309983,
      "grad_norm": 3.8233962059020996,
      "learning_rate": 2.7249144061100874e-05,
      "loss": 1.535,
      "step": 17287
    },
    {
      "epoch": 0.4553068211746115,
      "grad_norm": 1.9231579303741455,
      "learning_rate": 2.7247827232025286e-05,
      "loss": 1.8897,
      "step": 17288
    },
    {
      "epoch": 0.45533315775612326,
      "grad_norm": 1.9409904479980469,
      "learning_rate": 2.72465104029497e-05,
      "loss": 0.4959,
      "step": 17289
    },
    {
      "epoch": 0.45535949433763495,
      "grad_norm": 1.9427393674850464,
      "learning_rate": 2.724519357387411e-05,
      "loss": 1.8005,
      "step": 17290
    },
    {
      "epoch": 0.4553858309191467,
      "grad_norm": 3.4019124507904053,
      "learning_rate": 2.7243876744798523e-05,
      "loss": 1.533,
      "step": 17291
    },
    {
      "epoch": 0.45541216750065844,
      "grad_norm": 1.7695775032043457,
      "learning_rate": 2.7242559915722942e-05,
      "loss": 1.3177,
      "step": 17292
    },
    {
      "epoch": 0.4554385040821701,
      "grad_norm": 2.4496066570281982,
      "learning_rate": 2.7241243086647354e-05,
      "loss": 2.011,
      "step": 17293
    },
    {
      "epoch": 0.45546484066368187,
      "grad_norm": 4.076964855194092,
      "learning_rate": 2.723992625757177e-05,
      "loss": 1.7417,
      "step": 17294
    },
    {
      "epoch": 0.45549117724519356,
      "grad_norm": 1.7601439952850342,
      "learning_rate": 2.7238609428496182e-05,
      "loss": 1.1369,
      "step": 17295
    },
    {
      "epoch": 0.4555175138267053,
      "grad_norm": 4.183624744415283,
      "learning_rate": 2.7237292599420594e-05,
      "loss": 1.1913,
      "step": 17296
    },
    {
      "epoch": 0.455543850408217,
      "grad_norm": 2.683528423309326,
      "learning_rate": 2.7235975770345013e-05,
      "loss": 1.7364,
      "step": 17297
    },
    {
      "epoch": 0.45557018698972873,
      "grad_norm": 2.3656113147735596,
      "learning_rate": 2.7234658941269425e-05,
      "loss": 2.0034,
      "step": 17298
    },
    {
      "epoch": 0.4555965235712405,
      "grad_norm": 2.0474541187286377,
      "learning_rate": 2.7233342112193838e-05,
      "loss": 0.9772,
      "step": 17299
    },
    {
      "epoch": 0.45562286015275216,
      "grad_norm": 2.562882900238037,
      "learning_rate": 2.723202528311825e-05,
      "loss": 2.0706,
      "step": 17300
    },
    {
      "epoch": 0.4556491967342639,
      "grad_norm": 2.677858829498291,
      "learning_rate": 2.7230708454042665e-05,
      "loss": 0.5187,
      "step": 17301
    },
    {
      "epoch": 0.4556755333157756,
      "grad_norm": 4.931310653686523,
      "learning_rate": 2.722939162496708e-05,
      "loss": 1.8102,
      "step": 17302
    },
    {
      "epoch": 0.45570186989728734,
      "grad_norm": 1.7191283702850342,
      "learning_rate": 2.7228074795891497e-05,
      "loss": 1.812,
      "step": 17303
    },
    {
      "epoch": 0.455728206478799,
      "grad_norm": 6.109428405761719,
      "learning_rate": 2.722675796681591e-05,
      "loss": 1.0289,
      "step": 17304
    },
    {
      "epoch": 0.45575454306031077,
      "grad_norm": 1.7250807285308838,
      "learning_rate": 2.722544113774032e-05,
      "loss": 1.7173,
      "step": 17305
    },
    {
      "epoch": 0.4557808796418225,
      "grad_norm": 1.4788899421691895,
      "learning_rate": 2.722412430866474e-05,
      "loss": 1.8731,
      "step": 17306
    },
    {
      "epoch": 0.4558072162233342,
      "grad_norm": 2.513375997543335,
      "learning_rate": 2.7222807479589152e-05,
      "loss": 1.8095,
      "step": 17307
    },
    {
      "epoch": 0.45583355280484594,
      "grad_norm": 2.1410164833068848,
      "learning_rate": 2.7221490650513564e-05,
      "loss": 1.6,
      "step": 17308
    },
    {
      "epoch": 0.45585988938635763,
      "grad_norm": 1.7630873918533325,
      "learning_rate": 2.7220173821437977e-05,
      "loss": 0.6304,
      "step": 17309
    },
    {
      "epoch": 0.4558862259678694,
      "grad_norm": 1.8701564073562622,
      "learning_rate": 2.721885699236239e-05,
      "loss": 2.7404,
      "step": 17310
    },
    {
      "epoch": 0.4559125625493811,
      "grad_norm": 1.9941834211349487,
      "learning_rate": 2.7217540163286808e-05,
      "loss": 1.8618,
      "step": 17311
    },
    {
      "epoch": 0.4559388991308928,
      "grad_norm": 1.7726746797561646,
      "learning_rate": 2.721622333421122e-05,
      "loss": 0.3673,
      "step": 17312
    },
    {
      "epoch": 0.45596523571240455,
      "grad_norm": 1.9800612926483154,
      "learning_rate": 2.7214906505135636e-05,
      "loss": 2.0318,
      "step": 17313
    },
    {
      "epoch": 0.45599157229391624,
      "grad_norm": 1.7566994428634644,
      "learning_rate": 2.7213589676060048e-05,
      "loss": 0.436,
      "step": 17314
    },
    {
      "epoch": 0.456017908875428,
      "grad_norm": 2.314666748046875,
      "learning_rate": 2.721227284698446e-05,
      "loss": 1.8503,
      "step": 17315
    },
    {
      "epoch": 0.45604424545693967,
      "grad_norm": 3.5411014556884766,
      "learning_rate": 2.721095601790888e-05,
      "loss": 1.3966,
      "step": 17316
    },
    {
      "epoch": 0.4560705820384514,
      "grad_norm": 2.390590190887451,
      "learning_rate": 2.720963918883329e-05,
      "loss": 1.0384,
      "step": 17317
    },
    {
      "epoch": 0.45609691861996315,
      "grad_norm": 2.4419076442718506,
      "learning_rate": 2.7208322359757704e-05,
      "loss": 1.5649,
      "step": 17318
    },
    {
      "epoch": 0.45612325520147484,
      "grad_norm": 1.7367279529571533,
      "learning_rate": 2.7207005530682116e-05,
      "loss": 1.7707,
      "step": 17319
    },
    {
      "epoch": 0.4561495917829866,
      "grad_norm": 1.8437200784683228,
      "learning_rate": 2.7205688701606535e-05,
      "loss": 1.5748,
      "step": 17320
    },
    {
      "epoch": 0.4561759283644983,
      "grad_norm": 1.2369630336761475,
      "learning_rate": 2.7204371872530947e-05,
      "loss": 1.1754,
      "step": 17321
    },
    {
      "epoch": 0.45620226494601,
      "grad_norm": 1.8583474159240723,
      "learning_rate": 2.7203055043455363e-05,
      "loss": 1.6233,
      "step": 17322
    },
    {
      "epoch": 0.4562286015275217,
      "grad_norm": 2.628200054168701,
      "learning_rate": 2.7201738214379775e-05,
      "loss": 1.5336,
      "step": 17323
    },
    {
      "epoch": 0.45625493810903345,
      "grad_norm": 2.8460469245910645,
      "learning_rate": 2.7200421385304187e-05,
      "loss": 2.0944,
      "step": 17324
    },
    {
      "epoch": 0.4562812746905452,
      "grad_norm": 2.3785526752471924,
      "learning_rate": 2.7199104556228606e-05,
      "loss": 1.9822,
      "step": 17325
    },
    {
      "epoch": 0.4563076112720569,
      "grad_norm": 2.2259302139282227,
      "learning_rate": 2.7197787727153018e-05,
      "loss": 1.3858,
      "step": 17326
    },
    {
      "epoch": 0.4563339478535686,
      "grad_norm": 1.6828503608703613,
      "learning_rate": 2.719647089807743e-05,
      "loss": 1.0838,
      "step": 17327
    },
    {
      "epoch": 0.4563602844350803,
      "grad_norm": 1.918329119682312,
      "learning_rate": 2.7195154069001843e-05,
      "loss": 2.2113,
      "step": 17328
    },
    {
      "epoch": 0.45638662101659205,
      "grad_norm": 1.6512056589126587,
      "learning_rate": 2.7193837239926255e-05,
      "loss": 0.4725,
      "step": 17329
    },
    {
      "epoch": 0.45641295759810374,
      "grad_norm": 1.6636810302734375,
      "learning_rate": 2.7192520410850674e-05,
      "loss": 2.1088,
      "step": 17330
    },
    {
      "epoch": 0.4564392941796155,
      "grad_norm": 4.567899703979492,
      "learning_rate": 2.7191203581775086e-05,
      "loss": 1.3406,
      "step": 17331
    },
    {
      "epoch": 0.45646563076112723,
      "grad_norm": 4.268378734588623,
      "learning_rate": 2.71898867526995e-05,
      "loss": 1.2136,
      "step": 17332
    },
    {
      "epoch": 0.4564919673426389,
      "grad_norm": 4.233364582061768,
      "learning_rate": 2.7188569923623914e-05,
      "loss": 1.8267,
      "step": 17333
    },
    {
      "epoch": 0.45651830392415066,
      "grad_norm": 1.507789134979248,
      "learning_rate": 2.7187253094548333e-05,
      "loss": 2.28,
      "step": 17334
    },
    {
      "epoch": 0.45654464050566235,
      "grad_norm": 4.422889232635498,
      "learning_rate": 2.7185936265472745e-05,
      "loss": 1.8436,
      "step": 17335
    },
    {
      "epoch": 0.4565709770871741,
      "grad_norm": 1.651748538017273,
      "learning_rate": 2.7184619436397157e-05,
      "loss": 1.9707,
      "step": 17336
    },
    {
      "epoch": 0.4565973136686858,
      "grad_norm": 5.289831161499023,
      "learning_rate": 2.718330260732157e-05,
      "loss": 2.3291,
      "step": 17337
    },
    {
      "epoch": 0.4566236502501975,
      "grad_norm": 4.125948905944824,
      "learning_rate": 2.718198577824598e-05,
      "loss": 0.7519,
      "step": 17338
    },
    {
      "epoch": 0.45664998683170926,
      "grad_norm": 1.4943476915359497,
      "learning_rate": 2.71806689491704e-05,
      "loss": 1.7798,
      "step": 17339
    },
    {
      "epoch": 0.45667632341322095,
      "grad_norm": 2.3546130657196045,
      "learning_rate": 2.7179352120094813e-05,
      "loss": 1.9694,
      "step": 17340
    },
    {
      "epoch": 0.4567026599947327,
      "grad_norm": 4.659780025482178,
      "learning_rate": 2.717803529101923e-05,
      "loss": 0.3928,
      "step": 17341
    },
    {
      "epoch": 0.4567289965762444,
      "grad_norm": 2.033904552459717,
      "learning_rate": 2.717671846194364e-05,
      "loss": 1.0258,
      "step": 17342
    },
    {
      "epoch": 0.4567553331577561,
      "grad_norm": 1.9254486560821533,
      "learning_rate": 2.7175401632868053e-05,
      "loss": 0.7842,
      "step": 17343
    },
    {
      "epoch": 0.45678166973926787,
      "grad_norm": 3.2039225101470947,
      "learning_rate": 2.7174084803792472e-05,
      "loss": 1.5024,
      "step": 17344
    },
    {
      "epoch": 0.45680800632077956,
      "grad_norm": 4.272639751434326,
      "learning_rate": 2.7172767974716884e-05,
      "loss": 1.0394,
      "step": 17345
    },
    {
      "epoch": 0.4568343429022913,
      "grad_norm": 3.3715627193450928,
      "learning_rate": 2.7171451145641296e-05,
      "loss": 1.906,
      "step": 17346
    },
    {
      "epoch": 0.456860679483803,
      "grad_norm": 3.528592586517334,
      "learning_rate": 2.717013431656571e-05,
      "loss": 1.0343,
      "step": 17347
    },
    {
      "epoch": 0.45688701606531473,
      "grad_norm": 6.667778491973877,
      "learning_rate": 2.7168817487490124e-05,
      "loss": 2.2159,
      "step": 17348
    },
    {
      "epoch": 0.4569133526468264,
      "grad_norm": 2.012467622756958,
      "learning_rate": 2.716750065841454e-05,
      "loss": 2.0511,
      "step": 17349
    },
    {
      "epoch": 0.45693968922833816,
      "grad_norm": 5.494542598724365,
      "learning_rate": 2.7166183829338955e-05,
      "loss": 1.0639,
      "step": 17350
    },
    {
      "epoch": 0.4569660258098499,
      "grad_norm": 2.0097978115081787,
      "learning_rate": 2.7164867000263367e-05,
      "loss": 0.7831,
      "step": 17351
    },
    {
      "epoch": 0.4569923623913616,
      "grad_norm": 1.4346563816070557,
      "learning_rate": 2.716355017118778e-05,
      "loss": 0.416,
      "step": 17352
    },
    {
      "epoch": 0.45701869897287334,
      "grad_norm": 3.5429351329803467,
      "learning_rate": 2.71622333421122e-05,
      "loss": 1.9427,
      "step": 17353
    },
    {
      "epoch": 0.457045035554385,
      "grad_norm": 1.6483004093170166,
      "learning_rate": 2.716091651303661e-05,
      "loss": 1.8775,
      "step": 17354
    },
    {
      "epoch": 0.45707137213589677,
      "grad_norm": 3.4772427082061768,
      "learning_rate": 2.7159599683961023e-05,
      "loss": 1.2938,
      "step": 17355
    },
    {
      "epoch": 0.45709770871740846,
      "grad_norm": 2.2466113567352295,
      "learning_rate": 2.7158282854885435e-05,
      "loss": 1.8039,
      "step": 17356
    },
    {
      "epoch": 0.4571240452989202,
      "grad_norm": 3.016786575317383,
      "learning_rate": 2.7156966025809847e-05,
      "loss": 1.81,
      "step": 17357
    },
    {
      "epoch": 0.45715038188043194,
      "grad_norm": 1.6325918436050415,
      "learning_rate": 2.7155649196734266e-05,
      "loss": 1.2971,
      "step": 17358
    },
    {
      "epoch": 0.45717671846194363,
      "grad_norm": 4.765496253967285,
      "learning_rate": 2.715433236765868e-05,
      "loss": 1.2005,
      "step": 17359
    },
    {
      "epoch": 0.4572030550434554,
      "grad_norm": 3.836104393005371,
      "learning_rate": 2.7153015538583094e-05,
      "loss": 1.8405,
      "step": 17360
    },
    {
      "epoch": 0.45722939162496706,
      "grad_norm": 1.4749449491500854,
      "learning_rate": 2.7151698709507506e-05,
      "loss": 2.0393,
      "step": 17361
    },
    {
      "epoch": 0.4572557282064788,
      "grad_norm": 3.033538818359375,
      "learning_rate": 2.715038188043192e-05,
      "loss": 0.9233,
      "step": 17362
    },
    {
      "epoch": 0.4572820647879905,
      "grad_norm": 1.9844825267791748,
      "learning_rate": 2.7149065051356338e-05,
      "loss": 0.5922,
      "step": 17363
    },
    {
      "epoch": 0.45730840136950224,
      "grad_norm": 3.0275564193725586,
      "learning_rate": 2.714774822228075e-05,
      "loss": 1.7589,
      "step": 17364
    },
    {
      "epoch": 0.457334737951014,
      "grad_norm": 2.288918972015381,
      "learning_rate": 2.7146431393205162e-05,
      "loss": 1.1355,
      "step": 17365
    },
    {
      "epoch": 0.45736107453252567,
      "grad_norm": 1.8926944732666016,
      "learning_rate": 2.7145114564129574e-05,
      "loss": 1.7833,
      "step": 17366
    },
    {
      "epoch": 0.4573874111140374,
      "grad_norm": 1.709686517715454,
      "learning_rate": 2.7143797735053993e-05,
      "loss": 2.2145,
      "step": 17367
    },
    {
      "epoch": 0.4574137476955491,
      "grad_norm": 1.7107502222061157,
      "learning_rate": 2.7142480905978405e-05,
      "loss": 2.0242,
      "step": 17368
    },
    {
      "epoch": 0.45744008427706084,
      "grad_norm": 1.6546893119812012,
      "learning_rate": 2.714116407690282e-05,
      "loss": 1.8925,
      "step": 17369
    },
    {
      "epoch": 0.45746642085857253,
      "grad_norm": 3.4109745025634766,
      "learning_rate": 2.7139847247827233e-05,
      "loss": 1.1984,
      "step": 17370
    },
    {
      "epoch": 0.4574927574400843,
      "grad_norm": 1.6220732927322388,
      "learning_rate": 2.7138530418751645e-05,
      "loss": 2.0254,
      "step": 17371
    },
    {
      "epoch": 0.457519094021596,
      "grad_norm": 3.6341445446014404,
      "learning_rate": 2.7137213589676064e-05,
      "loss": 1.8416,
      "step": 17372
    },
    {
      "epoch": 0.4575454306031077,
      "grad_norm": 2.507774591445923,
      "learning_rate": 2.7135896760600477e-05,
      "loss": 2.9992,
      "step": 17373
    },
    {
      "epoch": 0.45757176718461945,
      "grad_norm": 2.6956207752227783,
      "learning_rate": 2.713457993152489e-05,
      "loss": 2.029,
      "step": 17374
    },
    {
      "epoch": 0.45759810376613114,
      "grad_norm": 2.8928754329681396,
      "learning_rate": 2.71332631024493e-05,
      "loss": 1.6337,
      "step": 17375
    },
    {
      "epoch": 0.4576244403476429,
      "grad_norm": 2.1158711910247803,
      "learning_rate": 2.7131946273373717e-05,
      "loss": 2.5666,
      "step": 17376
    },
    {
      "epoch": 0.4576507769291546,
      "grad_norm": 2.768433094024658,
      "learning_rate": 2.7130629444298132e-05,
      "loss": 1.2013,
      "step": 17377
    },
    {
      "epoch": 0.4576771135106663,
      "grad_norm": 1.7517542839050293,
      "learning_rate": 2.7129312615222548e-05,
      "loss": 1.5358,
      "step": 17378
    },
    {
      "epoch": 0.45770345009217805,
      "grad_norm": 2.7721211910247803,
      "learning_rate": 2.712799578614696e-05,
      "loss": 1.983,
      "step": 17379
    },
    {
      "epoch": 0.45772978667368974,
      "grad_norm": 2.01784348487854,
      "learning_rate": 2.7126678957071372e-05,
      "loss": 0.702,
      "step": 17380
    },
    {
      "epoch": 0.4577561232552015,
      "grad_norm": 3.2808613777160645,
      "learning_rate": 2.7125362127995785e-05,
      "loss": 1.3785,
      "step": 17381
    },
    {
      "epoch": 0.4577824598367132,
      "grad_norm": 4.0102057456970215,
      "learning_rate": 2.7124045298920204e-05,
      "loss": 0.9703,
      "step": 17382
    },
    {
      "epoch": 0.4578087964182249,
      "grad_norm": 1.7842707633972168,
      "learning_rate": 2.7122728469844616e-05,
      "loss": 1.9043,
      "step": 17383
    },
    {
      "epoch": 0.45783513299973666,
      "grad_norm": 3.1047117710113525,
      "learning_rate": 2.7121411640769028e-05,
      "loss": 1.136,
      "step": 17384
    },
    {
      "epoch": 0.45786146958124835,
      "grad_norm": 1.8260467052459717,
      "learning_rate": 2.712009481169344e-05,
      "loss": 1.5915,
      "step": 17385
    },
    {
      "epoch": 0.4578878061627601,
      "grad_norm": 1.6174358129501343,
      "learning_rate": 2.711877798261786e-05,
      "loss": 2.2055,
      "step": 17386
    },
    {
      "epoch": 0.4579141427442718,
      "grad_norm": 1.6015636920928955,
      "learning_rate": 2.711746115354227e-05,
      "loss": 2.1223,
      "step": 17387
    },
    {
      "epoch": 0.4579404793257835,
      "grad_norm": 2.2054030895233154,
      "learning_rate": 2.7116144324466687e-05,
      "loss": 2.1959,
      "step": 17388
    },
    {
      "epoch": 0.4579668159072952,
      "grad_norm": 3.391774892807007,
      "learning_rate": 2.71148274953911e-05,
      "loss": 1.6875,
      "step": 17389
    },
    {
      "epoch": 0.45799315248880695,
      "grad_norm": 1.5481411218643188,
      "learning_rate": 2.711351066631551e-05,
      "loss": 2.0179,
      "step": 17390
    },
    {
      "epoch": 0.4580194890703187,
      "grad_norm": 2.3364076614379883,
      "learning_rate": 2.711219383723993e-05,
      "loss": 2.3304,
      "step": 17391
    },
    {
      "epoch": 0.4580458256518304,
      "grad_norm": 1.7283635139465332,
      "learning_rate": 2.7110877008164343e-05,
      "loss": 1.9451,
      "step": 17392
    },
    {
      "epoch": 0.45807216223334213,
      "grad_norm": 2.9483442306518555,
      "learning_rate": 2.7109560179088755e-05,
      "loss": 1.1101,
      "step": 17393
    },
    {
      "epoch": 0.4580984988148538,
      "grad_norm": 1.640470027923584,
      "learning_rate": 2.7108243350013167e-05,
      "loss": 1.5166,
      "step": 17394
    },
    {
      "epoch": 0.45812483539636556,
      "grad_norm": 1.80405592918396,
      "learning_rate": 2.7106926520937583e-05,
      "loss": 1.8124,
      "step": 17395
    },
    {
      "epoch": 0.45815117197787725,
      "grad_norm": 3.0634729862213135,
      "learning_rate": 2.7105609691861998e-05,
      "loss": 1.051,
      "step": 17396
    },
    {
      "epoch": 0.458177508559389,
      "grad_norm": 1.7485488653182983,
      "learning_rate": 2.7104292862786414e-05,
      "loss": 1.4824,
      "step": 17397
    },
    {
      "epoch": 0.45820384514090073,
      "grad_norm": 2.075005292892456,
      "learning_rate": 2.7102976033710826e-05,
      "loss": 1.8233,
      "step": 17398
    },
    {
      "epoch": 0.4582301817224124,
      "grad_norm": 3.7069613933563232,
      "learning_rate": 2.7101659204635238e-05,
      "loss": 2.0435,
      "step": 17399
    },
    {
      "epoch": 0.45825651830392417,
      "grad_norm": 2.6145107746124268,
      "learning_rate": 2.7100342375559657e-05,
      "loss": 1.1847,
      "step": 17400
    },
    {
      "epoch": 0.45828285488543585,
      "grad_norm": 2.9730470180511475,
      "learning_rate": 2.709902554648407e-05,
      "loss": 1.0361,
      "step": 17401
    },
    {
      "epoch": 0.4583091914669476,
      "grad_norm": 1.979385495185852,
      "learning_rate": 2.709770871740848e-05,
      "loss": 1.4995,
      "step": 17402
    },
    {
      "epoch": 0.4583355280484593,
      "grad_norm": 4.62583065032959,
      "learning_rate": 2.7096391888332894e-05,
      "loss": 1.3785,
      "step": 17403
    },
    {
      "epoch": 0.458361864629971,
      "grad_norm": 1.8206110000610352,
      "learning_rate": 2.709507505925731e-05,
      "loss": 1.7719,
      "step": 17404
    },
    {
      "epoch": 0.45838820121148277,
      "grad_norm": 3.4109480381011963,
      "learning_rate": 2.7093758230181725e-05,
      "loss": 1.2555,
      "step": 17405
    },
    {
      "epoch": 0.45841453779299446,
      "grad_norm": 5.451360702514648,
      "learning_rate": 2.709244140110614e-05,
      "loss": 2.0528,
      "step": 17406
    },
    {
      "epoch": 0.4584408743745062,
      "grad_norm": 2.0876517295837402,
      "learning_rate": 2.7091124572030553e-05,
      "loss": 1.3576,
      "step": 17407
    },
    {
      "epoch": 0.4584672109560179,
      "grad_norm": 2.019777536392212,
      "learning_rate": 2.7089807742954965e-05,
      "loss": 1.7,
      "step": 17408
    },
    {
      "epoch": 0.45849354753752963,
      "grad_norm": 1.627122402191162,
      "learning_rate": 2.7088490913879377e-05,
      "loss": 2.1738,
      "step": 17409
    },
    {
      "epoch": 0.4585198841190413,
      "grad_norm": 2.301504611968994,
      "learning_rate": 2.7087174084803796e-05,
      "loss": 1.9525,
      "step": 17410
    },
    {
      "epoch": 0.45854622070055306,
      "grad_norm": 2.6174991130828857,
      "learning_rate": 2.708585725572821e-05,
      "loss": 1.3099,
      "step": 17411
    },
    {
      "epoch": 0.4585725572820648,
      "grad_norm": 2.4257068634033203,
      "learning_rate": 2.708454042665262e-05,
      "loss": 1.1037,
      "step": 17412
    },
    {
      "epoch": 0.4585988938635765,
      "grad_norm": 2.256848096847534,
      "learning_rate": 2.7083223597577033e-05,
      "loss": 2.1147,
      "step": 17413
    },
    {
      "epoch": 0.45862523044508824,
      "grad_norm": 1.6314328908920288,
      "learning_rate": 2.708190676850145e-05,
      "loss": 1.7321,
      "step": 17414
    },
    {
      "epoch": 0.4586515670265999,
      "grad_norm": 1.7508454322814941,
      "learning_rate": 2.7080589939425864e-05,
      "loss": 0.6495,
      "step": 17415
    },
    {
      "epoch": 0.45867790360811167,
      "grad_norm": 2.508268356323242,
      "learning_rate": 2.707927311035028e-05,
      "loss": 1.9909,
      "step": 17416
    },
    {
      "epoch": 0.4587042401896234,
      "grad_norm": 2.0246963500976562,
      "learning_rate": 2.7077956281274692e-05,
      "loss": 1.5806,
      "step": 17417
    },
    {
      "epoch": 0.4587305767711351,
      "grad_norm": 1.9890328645706177,
      "learning_rate": 2.7076639452199104e-05,
      "loss": 1.4487,
      "step": 17418
    },
    {
      "epoch": 0.45875691335264684,
      "grad_norm": 2.3599395751953125,
      "learning_rate": 2.7075322623123523e-05,
      "loss": 0.6264,
      "step": 17419
    },
    {
      "epoch": 0.45878324993415853,
      "grad_norm": 2.685049533843994,
      "learning_rate": 2.7074005794047935e-05,
      "loss": 0.3283,
      "step": 17420
    },
    {
      "epoch": 0.4588095865156703,
      "grad_norm": 2.487574815750122,
      "learning_rate": 2.7072688964972347e-05,
      "loss": 2.1188,
      "step": 17421
    },
    {
      "epoch": 0.45883592309718196,
      "grad_norm": 2.0520687103271484,
      "learning_rate": 2.707137213589676e-05,
      "loss": 1.9028,
      "step": 17422
    },
    {
      "epoch": 0.4588622596786937,
      "grad_norm": 1.7135155200958252,
      "learning_rate": 2.7070055306821175e-05,
      "loss": 2.0268,
      "step": 17423
    },
    {
      "epoch": 0.45888859626020545,
      "grad_norm": 2.03914213180542,
      "learning_rate": 2.706873847774559e-05,
      "loss": 1.9026,
      "step": 17424
    },
    {
      "epoch": 0.45891493284171714,
      "grad_norm": 3.514350175857544,
      "learning_rate": 2.7067421648670006e-05,
      "loss": 1.1091,
      "step": 17425
    },
    {
      "epoch": 0.4589412694232289,
      "grad_norm": 3.3137736320495605,
      "learning_rate": 2.706610481959442e-05,
      "loss": 1.9172,
      "step": 17426
    },
    {
      "epoch": 0.45896760600474057,
      "grad_norm": 2.468156099319458,
      "learning_rate": 2.706478799051883e-05,
      "loss": 2.3359,
      "step": 17427
    },
    {
      "epoch": 0.4589939425862523,
      "grad_norm": 2.0728931427001953,
      "learning_rate": 2.7063471161443243e-05,
      "loss": 2.3017,
      "step": 17428
    },
    {
      "epoch": 0.459020279167764,
      "grad_norm": 2.697718620300293,
      "learning_rate": 2.7062154332367662e-05,
      "loss": 1.3245,
      "step": 17429
    },
    {
      "epoch": 0.45904661574927574,
      "grad_norm": 3.767880439758301,
      "learning_rate": 2.7060837503292074e-05,
      "loss": 1.7804,
      "step": 17430
    },
    {
      "epoch": 0.4590729523307875,
      "grad_norm": 1.5344229936599731,
      "learning_rate": 2.7059520674216486e-05,
      "loss": 1.4511,
      "step": 17431
    },
    {
      "epoch": 0.4590992889122992,
      "grad_norm": 2.57387113571167,
      "learning_rate": 2.70582038451409e-05,
      "loss": 1.6768,
      "step": 17432
    },
    {
      "epoch": 0.4591256254938109,
      "grad_norm": 2.7254576683044434,
      "learning_rate": 2.7056887016065318e-05,
      "loss": 1.6601,
      "step": 17433
    },
    {
      "epoch": 0.4591519620753226,
      "grad_norm": 1.8453673124313354,
      "learning_rate": 2.705557018698973e-05,
      "loss": 1.3177,
      "step": 17434
    },
    {
      "epoch": 0.45917829865683435,
      "grad_norm": 2.5454792976379395,
      "learning_rate": 2.7054253357914145e-05,
      "loss": 2.777,
      "step": 17435
    },
    {
      "epoch": 0.45920463523834604,
      "grad_norm": 1.5997931957244873,
      "learning_rate": 2.7052936528838558e-05,
      "loss": 0.6221,
      "step": 17436
    },
    {
      "epoch": 0.4592309718198578,
      "grad_norm": 2.054805040359497,
      "learning_rate": 2.705161969976297e-05,
      "loss": 1.8622,
      "step": 17437
    },
    {
      "epoch": 0.4592573084013695,
      "grad_norm": 1.8314502239227295,
      "learning_rate": 2.705030287068739e-05,
      "loss": 1.6375,
      "step": 17438
    },
    {
      "epoch": 0.4592836449828812,
      "grad_norm": 1.7336448431015015,
      "learning_rate": 2.70489860416118e-05,
      "loss": 1.9737,
      "step": 17439
    },
    {
      "epoch": 0.45930998156439296,
      "grad_norm": 2.5273306369781494,
      "learning_rate": 2.7047669212536213e-05,
      "loss": 0.7548,
      "step": 17440
    },
    {
      "epoch": 0.45933631814590464,
      "grad_norm": 3.0589494705200195,
      "learning_rate": 2.7046352383460626e-05,
      "loss": 1.5871,
      "step": 17441
    },
    {
      "epoch": 0.4593626547274164,
      "grad_norm": 1.9810404777526855,
      "learning_rate": 2.704503555438504e-05,
      "loss": 1.9559,
      "step": 17442
    },
    {
      "epoch": 0.4593889913089281,
      "grad_norm": 2.150716543197632,
      "learning_rate": 2.7043718725309457e-05,
      "loss": 1.8882,
      "step": 17443
    },
    {
      "epoch": 0.4594153278904398,
      "grad_norm": 2.692133903503418,
      "learning_rate": 2.7042401896233872e-05,
      "loss": 1.4891,
      "step": 17444
    },
    {
      "epoch": 0.45944166447195156,
      "grad_norm": 1.7243131399154663,
      "learning_rate": 2.7041085067158285e-05,
      "loss": 1.8789,
      "step": 17445
    },
    {
      "epoch": 0.45946800105346325,
      "grad_norm": 1.5809563398361206,
      "learning_rate": 2.7039768238082697e-05,
      "loss": 1.55,
      "step": 17446
    },
    {
      "epoch": 0.459494337634975,
      "grad_norm": 1.7917300462722778,
      "learning_rate": 2.703845140900711e-05,
      "loss": 2.1017,
      "step": 17447
    },
    {
      "epoch": 0.4595206742164867,
      "grad_norm": 3.135422468185425,
      "learning_rate": 2.7037134579931528e-05,
      "loss": 1.8175,
      "step": 17448
    },
    {
      "epoch": 0.4595470107979984,
      "grad_norm": 2.8379335403442383,
      "learning_rate": 2.703581775085594e-05,
      "loss": 0.8343,
      "step": 17449
    },
    {
      "epoch": 0.45957334737951017,
      "grad_norm": 3.434492349624634,
      "learning_rate": 2.7034500921780352e-05,
      "loss": 2.0049,
      "step": 17450
    },
    {
      "epoch": 0.45959968396102185,
      "grad_norm": 3.260800361633301,
      "learning_rate": 2.7033184092704768e-05,
      "loss": 1.8594,
      "step": 17451
    },
    {
      "epoch": 0.4596260205425336,
      "grad_norm": 2.1037139892578125,
      "learning_rate": 2.7031867263629184e-05,
      "loss": 2.1083,
      "step": 17452
    },
    {
      "epoch": 0.4596523571240453,
      "grad_norm": 1.7813032865524292,
      "learning_rate": 2.70305504345536e-05,
      "loss": 2.3705,
      "step": 17453
    },
    {
      "epoch": 0.45967869370555703,
      "grad_norm": 3.1875293254852295,
      "learning_rate": 2.702923360547801e-05,
      "loss": 0.6947,
      "step": 17454
    },
    {
      "epoch": 0.4597050302870687,
      "grad_norm": 3.766479730606079,
      "learning_rate": 2.7027916776402424e-05,
      "loss": 1.2198,
      "step": 17455
    },
    {
      "epoch": 0.45973136686858046,
      "grad_norm": 3.2631001472473145,
      "learning_rate": 2.7026599947326836e-05,
      "loss": 2.3352,
      "step": 17456
    },
    {
      "epoch": 0.4597577034500922,
      "grad_norm": 4.20930814743042,
      "learning_rate": 2.7025283118251255e-05,
      "loss": 1.7662,
      "step": 17457
    },
    {
      "epoch": 0.4597840400316039,
      "grad_norm": 1.9887974262237549,
      "learning_rate": 2.7023966289175667e-05,
      "loss": 1.4584,
      "step": 17458
    },
    {
      "epoch": 0.45981037661311563,
      "grad_norm": 2.7577054500579834,
      "learning_rate": 2.702264946010008e-05,
      "loss": 2.2135,
      "step": 17459
    },
    {
      "epoch": 0.4598367131946273,
      "grad_norm": 1.8073235750198364,
      "learning_rate": 2.702133263102449e-05,
      "loss": 1.9439,
      "step": 17460
    },
    {
      "epoch": 0.45986304977613907,
      "grad_norm": 1.8886864185333252,
      "learning_rate": 2.7020015801948907e-05,
      "loss": 2.2911,
      "step": 17461
    },
    {
      "epoch": 0.45988938635765075,
      "grad_norm": 1.6016392707824707,
      "learning_rate": 2.7018698972873323e-05,
      "loss": 1.6465,
      "step": 17462
    },
    {
      "epoch": 0.4599157229391625,
      "grad_norm": 1.7485017776489258,
      "learning_rate": 2.7017382143797738e-05,
      "loss": 1.6884,
      "step": 17463
    },
    {
      "epoch": 0.45994205952067424,
      "grad_norm": 1.9669878482818604,
      "learning_rate": 2.701606531472215e-05,
      "loss": 1.4297,
      "step": 17464
    },
    {
      "epoch": 0.45996839610218593,
      "grad_norm": 1.8073192834854126,
      "learning_rate": 2.7014748485646563e-05,
      "loss": 2.2036,
      "step": 17465
    },
    {
      "epoch": 0.45999473268369767,
      "grad_norm": 2.008207082748413,
      "learning_rate": 2.701343165657098e-05,
      "loss": 1.7573,
      "step": 17466
    },
    {
      "epoch": 0.46002106926520936,
      "grad_norm": 2.4656214714050293,
      "learning_rate": 2.7012114827495394e-05,
      "loss": 1.8519,
      "step": 17467
    },
    {
      "epoch": 0.4600474058467211,
      "grad_norm": 2.4187605381011963,
      "learning_rate": 2.7010797998419806e-05,
      "loss": 2.0717,
      "step": 17468
    },
    {
      "epoch": 0.4600737424282328,
      "grad_norm": 1.9050865173339844,
      "learning_rate": 2.7009481169344218e-05,
      "loss": 1.7168,
      "step": 17469
    },
    {
      "epoch": 0.46010007900974453,
      "grad_norm": 2.115663528442383,
      "learning_rate": 2.7008164340268634e-05,
      "loss": 2.1858,
      "step": 17470
    },
    {
      "epoch": 0.4601264155912563,
      "grad_norm": 3.364642381668091,
      "learning_rate": 2.700684751119305e-05,
      "loss": 1.3902,
      "step": 17471
    },
    {
      "epoch": 0.46015275217276796,
      "grad_norm": 1.4723210334777832,
      "learning_rate": 2.7005530682117465e-05,
      "loss": 1.9311,
      "step": 17472
    },
    {
      "epoch": 0.4601790887542797,
      "grad_norm": 4.462168216705322,
      "learning_rate": 2.7004213853041877e-05,
      "loss": 0.6964,
      "step": 17473
    },
    {
      "epoch": 0.4602054253357914,
      "grad_norm": 2.6608057022094727,
      "learning_rate": 2.700289702396629e-05,
      "loss": 2.1537,
      "step": 17474
    },
    {
      "epoch": 0.46023176191730314,
      "grad_norm": 2.631505012512207,
      "learning_rate": 2.70015801948907e-05,
      "loss": 0.7878,
      "step": 17475
    },
    {
      "epoch": 0.4602580984988148,
      "grad_norm": 1.8283488750457764,
      "learning_rate": 2.700026336581512e-05,
      "loss": 2.6337,
      "step": 17476
    },
    {
      "epoch": 0.46028443508032657,
      "grad_norm": 6.193769454956055,
      "learning_rate": 2.6998946536739533e-05,
      "loss": 1.2673,
      "step": 17477
    },
    {
      "epoch": 0.4603107716618383,
      "grad_norm": 1.632066011428833,
      "learning_rate": 2.6997629707663945e-05,
      "loss": 1.7221,
      "step": 17478
    },
    {
      "epoch": 0.46033710824335,
      "grad_norm": 2.579348087310791,
      "learning_rate": 2.699631287858836e-05,
      "loss": 1.722,
      "step": 17479
    },
    {
      "epoch": 0.46036344482486175,
      "grad_norm": 3.4655957221984863,
      "learning_rate": 2.6994996049512773e-05,
      "loss": 1.2046,
      "step": 17480
    },
    {
      "epoch": 0.46038978140637343,
      "grad_norm": 5.123465061187744,
      "learning_rate": 2.6993679220437192e-05,
      "loss": 1.6401,
      "step": 17481
    },
    {
      "epoch": 0.4604161179878852,
      "grad_norm": 3.545768976211548,
      "learning_rate": 2.6992362391361604e-05,
      "loss": 1.5401,
      "step": 17482
    },
    {
      "epoch": 0.4604424545693969,
      "grad_norm": 1.8814715147018433,
      "learning_rate": 2.6991045562286016e-05,
      "loss": 1.4878,
      "step": 17483
    },
    {
      "epoch": 0.4604687911509086,
      "grad_norm": 1.5582077503204346,
      "learning_rate": 2.698972873321043e-05,
      "loss": 1.7208,
      "step": 17484
    },
    {
      "epoch": 0.46049512773242035,
      "grad_norm": 2.813190221786499,
      "learning_rate": 2.6988411904134847e-05,
      "loss": 1.38,
      "step": 17485
    },
    {
      "epoch": 0.46052146431393204,
      "grad_norm": 1.8609251976013184,
      "learning_rate": 2.698709507505926e-05,
      "loss": 2.4392,
      "step": 17486
    },
    {
      "epoch": 0.4605478008954438,
      "grad_norm": 3.437260627746582,
      "learning_rate": 2.6985778245983672e-05,
      "loss": 2.458,
      "step": 17487
    },
    {
      "epoch": 0.46057413747695547,
      "grad_norm": 2.206209421157837,
      "learning_rate": 2.6984461416908084e-05,
      "loss": 1.6908,
      "step": 17488
    },
    {
      "epoch": 0.4606004740584672,
      "grad_norm": 2.317737579345703,
      "learning_rate": 2.69831445878325e-05,
      "loss": 1.9039,
      "step": 17489
    },
    {
      "epoch": 0.46062681063997896,
      "grad_norm": 2.2961573600769043,
      "learning_rate": 2.6981827758756915e-05,
      "loss": 2.0017,
      "step": 17490
    },
    {
      "epoch": 0.46065314722149064,
      "grad_norm": 1.447906255722046,
      "learning_rate": 2.698051092968133e-05,
      "loss": 0.3744,
      "step": 17491
    },
    {
      "epoch": 0.4606794838030024,
      "grad_norm": 2.1364970207214355,
      "learning_rate": 2.6979194100605743e-05,
      "loss": 1.3722,
      "step": 17492
    },
    {
      "epoch": 0.4607058203845141,
      "grad_norm": 1.826667308807373,
      "learning_rate": 2.6977877271530155e-05,
      "loss": 1.3515,
      "step": 17493
    },
    {
      "epoch": 0.4607321569660258,
      "grad_norm": 1.814029335975647,
      "learning_rate": 2.6976560442454567e-05,
      "loss": 1.809,
      "step": 17494
    },
    {
      "epoch": 0.4607584935475375,
      "grad_norm": 3.5878870487213135,
      "learning_rate": 2.6975243613378986e-05,
      "loss": 1.332,
      "step": 17495
    },
    {
      "epoch": 0.46078483012904925,
      "grad_norm": 2.3971469402313232,
      "learning_rate": 2.69739267843034e-05,
      "loss": 1.5314,
      "step": 17496
    },
    {
      "epoch": 0.460811166710561,
      "grad_norm": 3.9298312664031982,
      "learning_rate": 2.697260995522781e-05,
      "loss": 1.8334,
      "step": 17497
    },
    {
      "epoch": 0.4608375032920727,
      "grad_norm": 2.1526389122009277,
      "learning_rate": 2.6971293126152226e-05,
      "loss": 1.3654,
      "step": 17498
    },
    {
      "epoch": 0.4608638398735844,
      "grad_norm": 1.6501330137252808,
      "learning_rate": 2.6969976297076642e-05,
      "loss": 1.9574,
      "step": 17499
    },
    {
      "epoch": 0.4608901764550961,
      "grad_norm": 1.8999844789505005,
      "learning_rate": 2.6968659468001058e-05,
      "loss": 1.9529,
      "step": 17500
    },
    {
      "epoch": 0.46091651303660786,
      "grad_norm": 2.108538866043091,
      "learning_rate": 2.696734263892547e-05,
      "loss": 1.6076,
      "step": 17501
    },
    {
      "epoch": 0.46094284961811954,
      "grad_norm": 3.4208884239196777,
      "learning_rate": 2.6966025809849882e-05,
      "loss": 1.666,
      "step": 17502
    },
    {
      "epoch": 0.4609691861996313,
      "grad_norm": 2.0282604694366455,
      "learning_rate": 2.6964708980774294e-05,
      "loss": 2.0153,
      "step": 17503
    },
    {
      "epoch": 0.46099552278114303,
      "grad_norm": 1.6982851028442383,
      "learning_rate": 2.6963392151698713e-05,
      "loss": 1.7184,
      "step": 17504
    },
    {
      "epoch": 0.4610218593626547,
      "grad_norm": 1.7129539251327515,
      "learning_rate": 2.6962075322623126e-05,
      "loss": 1.1235,
      "step": 17505
    },
    {
      "epoch": 0.46104819594416646,
      "grad_norm": 1.9388326406478882,
      "learning_rate": 2.6960758493547538e-05,
      "loss": 1.7993,
      "step": 17506
    },
    {
      "epoch": 0.46107453252567815,
      "grad_norm": 2.903170585632324,
      "learning_rate": 2.6959441664471953e-05,
      "loss": 1.7934,
      "step": 17507
    },
    {
      "epoch": 0.4611008691071899,
      "grad_norm": 2.008204460144043,
      "learning_rate": 2.6958124835396366e-05,
      "loss": 1.7812,
      "step": 17508
    },
    {
      "epoch": 0.4611272056887016,
      "grad_norm": 1.6647323369979858,
      "learning_rate": 2.6956808006320785e-05,
      "loss": 1.8555,
      "step": 17509
    },
    {
      "epoch": 0.4611535422702133,
      "grad_norm": 2.406745195388794,
      "learning_rate": 2.6955491177245197e-05,
      "loss": 0.7043,
      "step": 17510
    },
    {
      "epoch": 0.46117987885172507,
      "grad_norm": 1.7166666984558105,
      "learning_rate": 2.695417434816961e-05,
      "loss": 2.0748,
      "step": 17511
    },
    {
      "epoch": 0.46120621543323675,
      "grad_norm": 2.892624616622925,
      "learning_rate": 2.695285751909402e-05,
      "loss": 1.3902,
      "step": 17512
    },
    {
      "epoch": 0.4612325520147485,
      "grad_norm": 2.0963261127471924,
      "learning_rate": 2.6951540690018433e-05,
      "loss": 1.8431,
      "step": 17513
    },
    {
      "epoch": 0.4612588885962602,
      "grad_norm": 2.036587953567505,
      "learning_rate": 2.6950223860942852e-05,
      "loss": 1.5784,
      "step": 17514
    },
    {
      "epoch": 0.46128522517777193,
      "grad_norm": 1.982434630393982,
      "learning_rate": 2.6948907031867265e-05,
      "loss": 1.3778,
      "step": 17515
    },
    {
      "epoch": 0.4613115617592837,
      "grad_norm": 2.5579171180725098,
      "learning_rate": 2.6947590202791677e-05,
      "loss": 1.7708,
      "step": 17516
    },
    {
      "epoch": 0.46133789834079536,
      "grad_norm": 1.8268247842788696,
      "learning_rate": 2.6946273373716092e-05,
      "loss": 1.9356,
      "step": 17517
    },
    {
      "epoch": 0.4613642349223071,
      "grad_norm": 1.5559669733047485,
      "learning_rate": 2.6944956544640508e-05,
      "loss": 1.7244,
      "step": 17518
    },
    {
      "epoch": 0.4613905715038188,
      "grad_norm": 1.5587821006774902,
      "learning_rate": 2.6943639715564924e-05,
      "loss": 2.2493,
      "step": 17519
    },
    {
      "epoch": 0.46141690808533053,
      "grad_norm": 3.959388494491577,
      "learning_rate": 2.6942322886489336e-05,
      "loss": 2.49,
      "step": 17520
    },
    {
      "epoch": 0.4614432446668422,
      "grad_norm": 4.955875873565674,
      "learning_rate": 2.6941006057413748e-05,
      "loss": 0.7397,
      "step": 17521
    },
    {
      "epoch": 0.46146958124835397,
      "grad_norm": 1.5485541820526123,
      "learning_rate": 2.693968922833816e-05,
      "loss": 1.82,
      "step": 17522
    },
    {
      "epoch": 0.4614959178298657,
      "grad_norm": 2.207554578781128,
      "learning_rate": 2.693837239926258e-05,
      "loss": 1.9246,
      "step": 17523
    },
    {
      "epoch": 0.4615222544113774,
      "grad_norm": 2.386310338973999,
      "learning_rate": 2.693705557018699e-05,
      "loss": 1.6201,
      "step": 17524
    },
    {
      "epoch": 0.46154859099288914,
      "grad_norm": 1.8456358909606934,
      "learning_rate": 2.6935738741111404e-05,
      "loss": 1.4431,
      "step": 17525
    },
    {
      "epoch": 0.46157492757440083,
      "grad_norm": 2.117990016937256,
      "learning_rate": 2.693442191203582e-05,
      "loss": 2.1145,
      "step": 17526
    },
    {
      "epoch": 0.46160126415591257,
      "grad_norm": 1.796316385269165,
      "learning_rate": 2.693310508296023e-05,
      "loss": 1.8411,
      "step": 17527
    },
    {
      "epoch": 0.46162760073742426,
      "grad_norm": 2.610105037689209,
      "learning_rate": 2.693178825388465e-05,
      "loss": 1.1028,
      "step": 17528
    },
    {
      "epoch": 0.461653937318936,
      "grad_norm": 1.6163783073425293,
      "learning_rate": 2.6930471424809063e-05,
      "loss": 1.8427,
      "step": 17529
    },
    {
      "epoch": 0.46168027390044775,
      "grad_norm": 2.9518046379089355,
      "learning_rate": 2.6929154595733475e-05,
      "loss": 1.1236,
      "step": 17530
    },
    {
      "epoch": 0.46170661048195943,
      "grad_norm": 3.709421396255493,
      "learning_rate": 2.6927837766657887e-05,
      "loss": 1.2948,
      "step": 17531
    },
    {
      "epoch": 0.4617329470634712,
      "grad_norm": 2.0739338397979736,
      "learning_rate": 2.6926520937582306e-05,
      "loss": 2.27,
      "step": 17532
    },
    {
      "epoch": 0.46175928364498287,
      "grad_norm": 1.6710008382797241,
      "learning_rate": 2.6925204108506718e-05,
      "loss": 1.7208,
      "step": 17533
    },
    {
      "epoch": 0.4617856202264946,
      "grad_norm": 2.491178512573242,
      "learning_rate": 2.692388727943113e-05,
      "loss": 0.2786,
      "step": 17534
    },
    {
      "epoch": 0.4618119568080063,
      "grad_norm": 3.1380090713500977,
      "learning_rate": 2.6922570450355543e-05,
      "loss": 1.739,
      "step": 17535
    },
    {
      "epoch": 0.46183829338951804,
      "grad_norm": 1.7793279886245728,
      "learning_rate": 2.6921253621279958e-05,
      "loss": 1.8548,
      "step": 17536
    },
    {
      "epoch": 0.4618646299710298,
      "grad_norm": 2.807365655899048,
      "learning_rate": 2.6919936792204374e-05,
      "loss": 1.5687,
      "step": 17537
    },
    {
      "epoch": 0.46189096655254147,
      "grad_norm": 2.846344470977783,
      "learning_rate": 2.691861996312879e-05,
      "loss": 2.0694,
      "step": 17538
    },
    {
      "epoch": 0.4619173031340532,
      "grad_norm": 1.955567717552185,
      "learning_rate": 2.69173031340532e-05,
      "loss": 2.7026,
      "step": 17539
    },
    {
      "epoch": 0.4619436397155649,
      "grad_norm": 1.646132230758667,
      "learning_rate": 2.6915986304977614e-05,
      "loss": 1.7977,
      "step": 17540
    },
    {
      "epoch": 0.46196997629707665,
      "grad_norm": 3.220524549484253,
      "learning_rate": 2.6914669475902026e-05,
      "loss": 1.4106,
      "step": 17541
    },
    {
      "epoch": 0.46199631287858833,
      "grad_norm": 1.5704196691513062,
      "learning_rate": 2.6913352646826445e-05,
      "loss": 1.7109,
      "step": 17542
    },
    {
      "epoch": 0.4620226494601001,
      "grad_norm": 2.640713691711426,
      "learning_rate": 2.6912035817750857e-05,
      "loss": 1.1525,
      "step": 17543
    },
    {
      "epoch": 0.4620489860416118,
      "grad_norm": 3.208357810974121,
      "learning_rate": 2.691071898867527e-05,
      "loss": 1.7592,
      "step": 17544
    },
    {
      "epoch": 0.4620753226231235,
      "grad_norm": 1.7211967706680298,
      "learning_rate": 2.6909402159599685e-05,
      "loss": 0.9374,
      "step": 17545
    },
    {
      "epoch": 0.46210165920463525,
      "grad_norm": 1.4923902750015259,
      "learning_rate": 2.6908085330524097e-05,
      "loss": 1.2124,
      "step": 17546
    },
    {
      "epoch": 0.46212799578614694,
      "grad_norm": 3.00479793548584,
      "learning_rate": 2.6906768501448516e-05,
      "loss": 2.6912,
      "step": 17547
    },
    {
      "epoch": 0.4621543323676587,
      "grad_norm": 1.9975330829620361,
      "learning_rate": 2.690545167237293e-05,
      "loss": 1.3521,
      "step": 17548
    },
    {
      "epoch": 0.46218066894917037,
      "grad_norm": 3.134265184402466,
      "learning_rate": 2.690413484329734e-05,
      "loss": 1.334,
      "step": 17549
    },
    {
      "epoch": 0.4622070055306821,
      "grad_norm": 2.316401720046997,
      "learning_rate": 2.6902818014221753e-05,
      "loss": 1.8305,
      "step": 17550
    },
    {
      "epoch": 0.46223334211219386,
      "grad_norm": 1.648706316947937,
      "learning_rate": 2.6901501185146172e-05,
      "loss": 0.2379,
      "step": 17551
    },
    {
      "epoch": 0.46225967869370554,
      "grad_norm": 4.71798038482666,
      "learning_rate": 2.6900184356070584e-05,
      "loss": 1.7815,
      "step": 17552
    },
    {
      "epoch": 0.4622860152752173,
      "grad_norm": 3.6050045490264893,
      "learning_rate": 2.6898867526994996e-05,
      "loss": 1.2606,
      "step": 17553
    },
    {
      "epoch": 0.462312351856729,
      "grad_norm": 2.0452992916107178,
      "learning_rate": 2.6897550697919412e-05,
      "loss": 1.833,
      "step": 17554
    },
    {
      "epoch": 0.4623386884382407,
      "grad_norm": 4.002376556396484,
      "learning_rate": 2.6896233868843824e-05,
      "loss": 1.8259,
      "step": 17555
    },
    {
      "epoch": 0.46236502501975246,
      "grad_norm": 3.458533525466919,
      "learning_rate": 2.6894917039768243e-05,
      "loss": 1.0992,
      "step": 17556
    },
    {
      "epoch": 0.46239136160126415,
      "grad_norm": 1.8810521364212036,
      "learning_rate": 2.6893600210692655e-05,
      "loss": 1.8099,
      "step": 17557
    },
    {
      "epoch": 0.4624176981827759,
      "grad_norm": 2.9881768226623535,
      "learning_rate": 2.6892283381617067e-05,
      "loss": 1.0115,
      "step": 17558
    },
    {
      "epoch": 0.4624440347642876,
      "grad_norm": 4.67957067489624,
      "learning_rate": 2.689096655254148e-05,
      "loss": 1.2525,
      "step": 17559
    },
    {
      "epoch": 0.4624703713457993,
      "grad_norm": 2.159923791885376,
      "learning_rate": 2.6889649723465892e-05,
      "loss": 1.3093,
      "step": 17560
    },
    {
      "epoch": 0.462496707927311,
      "grad_norm": 1.7475998401641846,
      "learning_rate": 2.688833289439031e-05,
      "loss": 0.5041,
      "step": 17561
    },
    {
      "epoch": 0.46252304450882276,
      "grad_norm": 1.7486635446548462,
      "learning_rate": 2.6887016065314723e-05,
      "loss": 2.6904,
      "step": 17562
    },
    {
      "epoch": 0.4625493810903345,
      "grad_norm": 3.956477642059326,
      "learning_rate": 2.6885699236239135e-05,
      "loss": 1.6588,
      "step": 17563
    },
    {
      "epoch": 0.4625757176718462,
      "grad_norm": 1.904270887374878,
      "learning_rate": 2.688438240716355e-05,
      "loss": 1.8687,
      "step": 17564
    },
    {
      "epoch": 0.46260205425335793,
      "grad_norm": 1.4893171787261963,
      "learning_rate": 2.6883065578087967e-05,
      "loss": 2.3775,
      "step": 17565
    },
    {
      "epoch": 0.4626283908348696,
      "grad_norm": 1.8342007398605347,
      "learning_rate": 2.6881748749012382e-05,
      "loss": 1.9912,
      "step": 17566
    },
    {
      "epoch": 0.46265472741638136,
      "grad_norm": 2.115168809890747,
      "learning_rate": 2.6880431919936794e-05,
      "loss": 1.0927,
      "step": 17567
    },
    {
      "epoch": 0.46268106399789305,
      "grad_norm": 2.3055005073547363,
      "learning_rate": 2.6879115090861207e-05,
      "loss": 1.7059,
      "step": 17568
    },
    {
      "epoch": 0.4627074005794048,
      "grad_norm": 1.8641294240951538,
      "learning_rate": 2.687779826178562e-05,
      "loss": 1.3708,
      "step": 17569
    },
    {
      "epoch": 0.46273373716091654,
      "grad_norm": 2.036135196685791,
      "learning_rate": 2.6876481432710038e-05,
      "loss": 1.9027,
      "step": 17570
    },
    {
      "epoch": 0.4627600737424282,
      "grad_norm": 1.6262738704681396,
      "learning_rate": 2.687516460363445e-05,
      "loss": 1.4936,
      "step": 17571
    },
    {
      "epoch": 0.46278641032393997,
      "grad_norm": 1.7485545873641968,
      "learning_rate": 2.6873847774558862e-05,
      "loss": 2.2133,
      "step": 17572
    },
    {
      "epoch": 0.46281274690545166,
      "grad_norm": 1.6541633605957031,
      "learning_rate": 2.6872530945483278e-05,
      "loss": 1.5818,
      "step": 17573
    },
    {
      "epoch": 0.4628390834869634,
      "grad_norm": 2.2026326656341553,
      "learning_rate": 2.687121411640769e-05,
      "loss": 1.7335,
      "step": 17574
    },
    {
      "epoch": 0.4628654200684751,
      "grad_norm": 4.983541488647461,
      "learning_rate": 2.686989728733211e-05,
      "loss": 1.23,
      "step": 17575
    },
    {
      "epoch": 0.46289175664998683,
      "grad_norm": 2.145479679107666,
      "learning_rate": 2.686858045825652e-05,
      "loss": 2.0333,
      "step": 17576
    },
    {
      "epoch": 0.4629180932314986,
      "grad_norm": 3.012795925140381,
      "learning_rate": 2.6867263629180933e-05,
      "loss": 1.6883,
      "step": 17577
    },
    {
      "epoch": 0.46294442981301026,
      "grad_norm": 2.079068660736084,
      "learning_rate": 2.6865946800105346e-05,
      "loss": 1.524,
      "step": 17578
    },
    {
      "epoch": 0.462970766394522,
      "grad_norm": 2.492637872695923,
      "learning_rate": 2.6864629971029758e-05,
      "loss": 1.9061,
      "step": 17579
    },
    {
      "epoch": 0.4629971029760337,
      "grad_norm": 5.192150115966797,
      "learning_rate": 2.6863313141954177e-05,
      "loss": 1.5762,
      "step": 17580
    },
    {
      "epoch": 0.46302343955754544,
      "grad_norm": 3.4451215267181396,
      "learning_rate": 2.686199631287859e-05,
      "loss": 1.5547,
      "step": 17581
    },
    {
      "epoch": 0.4630497761390571,
      "grad_norm": 1.6375210285186768,
      "learning_rate": 2.6860679483803005e-05,
      "loss": 1.8389,
      "step": 17582
    },
    {
      "epoch": 0.46307611272056887,
      "grad_norm": 6.425418853759766,
      "learning_rate": 2.6859362654727417e-05,
      "loss": 1.8512,
      "step": 17583
    },
    {
      "epoch": 0.4631024493020806,
      "grad_norm": 2.2132797241210938,
      "learning_rate": 2.6858045825651836e-05,
      "loss": 1.9915,
      "step": 17584
    },
    {
      "epoch": 0.4631287858835923,
      "grad_norm": 3.884831666946411,
      "learning_rate": 2.6856728996576248e-05,
      "loss": 1.3174,
      "step": 17585
    },
    {
      "epoch": 0.46315512246510404,
      "grad_norm": 2.4993481636047363,
      "learning_rate": 2.685541216750066e-05,
      "loss": 1.9246,
      "step": 17586
    },
    {
      "epoch": 0.46318145904661573,
      "grad_norm": 2.1750473976135254,
      "learning_rate": 2.6854095338425072e-05,
      "loss": 1.3254,
      "step": 17587
    },
    {
      "epoch": 0.46320779562812747,
      "grad_norm": 1.5546739101409912,
      "learning_rate": 2.6852778509349485e-05,
      "loss": 1.5789,
      "step": 17588
    },
    {
      "epoch": 0.4632341322096392,
      "grad_norm": 2.34140944480896,
      "learning_rate": 2.6851461680273904e-05,
      "loss": 0.9459,
      "step": 17589
    },
    {
      "epoch": 0.4632604687911509,
      "grad_norm": 2.98567533493042,
      "learning_rate": 2.6850144851198316e-05,
      "loss": 1.0636,
      "step": 17590
    },
    {
      "epoch": 0.46328680537266265,
      "grad_norm": 2.0883536338806152,
      "learning_rate": 2.6848828022122728e-05,
      "loss": 1.5716,
      "step": 17591
    },
    {
      "epoch": 0.46331314195417433,
      "grad_norm": 1.701937198638916,
      "learning_rate": 2.6847511193047144e-05,
      "loss": 1.8402,
      "step": 17592
    },
    {
      "epoch": 0.4633394785356861,
      "grad_norm": 1.5772747993469238,
      "learning_rate": 2.6846194363971556e-05,
      "loss": 2.0436,
      "step": 17593
    },
    {
      "epoch": 0.46336581511719777,
      "grad_norm": 2.340604782104492,
      "learning_rate": 2.6844877534895975e-05,
      "loss": 1.521,
      "step": 17594
    },
    {
      "epoch": 0.4633921516987095,
      "grad_norm": 3.2818398475646973,
      "learning_rate": 2.6843560705820387e-05,
      "loss": 1.6473,
      "step": 17595
    },
    {
      "epoch": 0.46341848828022125,
      "grad_norm": 1.6537885665893555,
      "learning_rate": 2.68422438767448e-05,
      "loss": 1.1602,
      "step": 17596
    },
    {
      "epoch": 0.46344482486173294,
      "grad_norm": 2.0647153854370117,
      "learning_rate": 2.684092704766921e-05,
      "loss": 2.0969,
      "step": 17597
    },
    {
      "epoch": 0.4634711614432447,
      "grad_norm": 2.1821203231811523,
      "learning_rate": 2.683961021859363e-05,
      "loss": 1.9732,
      "step": 17598
    },
    {
      "epoch": 0.46349749802475637,
      "grad_norm": 2.088315486907959,
      "learning_rate": 2.6838293389518043e-05,
      "loss": 2.0616,
      "step": 17599
    },
    {
      "epoch": 0.4635238346062681,
      "grad_norm": 2.128448009490967,
      "learning_rate": 2.6836976560442455e-05,
      "loss": 2.3946,
      "step": 17600
    },
    {
      "epoch": 0.4635501711877798,
      "grad_norm": 2.9358444213867188,
      "learning_rate": 2.683565973136687e-05,
      "loss": 1.9484,
      "step": 17601
    },
    {
      "epoch": 0.46357650776929155,
      "grad_norm": 1.4702099561691284,
      "learning_rate": 2.6834342902291283e-05,
      "loss": 1.7624,
      "step": 17602
    },
    {
      "epoch": 0.4636028443508033,
      "grad_norm": 2.299086093902588,
      "learning_rate": 2.68330260732157e-05,
      "loss": 0.5126,
      "step": 17603
    },
    {
      "epoch": 0.463629180932315,
      "grad_norm": 5.270533084869385,
      "learning_rate": 2.6831709244140114e-05,
      "loss": 1.5283,
      "step": 17604
    },
    {
      "epoch": 0.4636555175138267,
      "grad_norm": 1.7086987495422363,
      "learning_rate": 2.6830392415064526e-05,
      "loss": 1.2519,
      "step": 17605
    },
    {
      "epoch": 0.4636818540953384,
      "grad_norm": 2.6047730445861816,
      "learning_rate": 2.6829075585988938e-05,
      "loss": 1.3696,
      "step": 17606
    },
    {
      "epoch": 0.46370819067685015,
      "grad_norm": 2.140594720840454,
      "learning_rate": 2.682775875691335e-05,
      "loss": 1.711,
      "step": 17607
    },
    {
      "epoch": 0.46373452725836184,
      "grad_norm": 4.106740951538086,
      "learning_rate": 2.682644192783777e-05,
      "loss": 0.3154,
      "step": 17608
    },
    {
      "epoch": 0.4637608638398736,
      "grad_norm": 2.4060792922973633,
      "learning_rate": 2.682512509876218e-05,
      "loss": 1.1164,
      "step": 17609
    },
    {
      "epoch": 0.4637872004213853,
      "grad_norm": 5.173916339874268,
      "learning_rate": 2.6823808269686594e-05,
      "loss": 1.0893,
      "step": 17610
    },
    {
      "epoch": 0.463813537002897,
      "grad_norm": 2.4310548305511475,
      "learning_rate": 2.682249144061101e-05,
      "loss": 1.3136,
      "step": 17611
    },
    {
      "epoch": 0.46383987358440876,
      "grad_norm": 1.8239504098892212,
      "learning_rate": 2.682117461153542e-05,
      "loss": 1.1987,
      "step": 17612
    },
    {
      "epoch": 0.46386621016592045,
      "grad_norm": 3.905519485473633,
      "learning_rate": 2.681985778245984e-05,
      "loss": 1.2127,
      "step": 17613
    },
    {
      "epoch": 0.4638925467474322,
      "grad_norm": 2.3367154598236084,
      "learning_rate": 2.6818540953384253e-05,
      "loss": 1.2624,
      "step": 17614
    },
    {
      "epoch": 0.4639188833289439,
      "grad_norm": 4.140835762023926,
      "learning_rate": 2.6817224124308665e-05,
      "loss": 1.4969,
      "step": 17615
    },
    {
      "epoch": 0.4639452199104556,
      "grad_norm": 3.3306164741516113,
      "learning_rate": 2.6815907295233077e-05,
      "loss": 1.5663,
      "step": 17616
    },
    {
      "epoch": 0.46397155649196736,
      "grad_norm": 2.9801230430603027,
      "learning_rate": 2.6814590466157496e-05,
      "loss": 1.3286,
      "step": 17617
    },
    {
      "epoch": 0.46399789307347905,
      "grad_norm": 2.134817600250244,
      "learning_rate": 2.681327363708191e-05,
      "loss": 1.7088,
      "step": 17618
    },
    {
      "epoch": 0.4640242296549908,
      "grad_norm": 2.7753093242645264,
      "learning_rate": 2.681195680800632e-05,
      "loss": 1.7023,
      "step": 17619
    },
    {
      "epoch": 0.4640505662365025,
      "grad_norm": 2.452871561050415,
      "learning_rate": 2.6810639978930736e-05,
      "loss": 1.6421,
      "step": 17620
    },
    {
      "epoch": 0.4640769028180142,
      "grad_norm": 1.9739270210266113,
      "learning_rate": 2.680932314985515e-05,
      "loss": 1.2964,
      "step": 17621
    },
    {
      "epoch": 0.46410323939952597,
      "grad_norm": 2.002150535583496,
      "learning_rate": 2.6808006320779567e-05,
      "loss": 2.2155,
      "step": 17622
    },
    {
      "epoch": 0.46412957598103766,
      "grad_norm": 4.563379764556885,
      "learning_rate": 2.680668949170398e-05,
      "loss": 1.5765,
      "step": 17623
    },
    {
      "epoch": 0.4641559125625494,
      "grad_norm": 1.8958802223205566,
      "learning_rate": 2.6805372662628392e-05,
      "loss": 1.0206,
      "step": 17624
    },
    {
      "epoch": 0.4641822491440611,
      "grad_norm": 3.142764091491699,
      "learning_rate": 2.6804055833552804e-05,
      "loss": 1.3983,
      "step": 17625
    },
    {
      "epoch": 0.46420858572557283,
      "grad_norm": 1.6672254800796509,
      "learning_rate": 2.6802739004477216e-05,
      "loss": 1.6891,
      "step": 17626
    },
    {
      "epoch": 0.4642349223070845,
      "grad_norm": 1.7124546766281128,
      "learning_rate": 2.6801422175401635e-05,
      "loss": 1.7931,
      "step": 17627
    },
    {
      "epoch": 0.46426125888859626,
      "grad_norm": 2.0575461387634277,
      "learning_rate": 2.6800105346326048e-05,
      "loss": 1.6328,
      "step": 17628
    },
    {
      "epoch": 0.464287595470108,
      "grad_norm": 2.658080577850342,
      "learning_rate": 2.6798788517250463e-05,
      "loss": 1.9511,
      "step": 17629
    },
    {
      "epoch": 0.4643139320516197,
      "grad_norm": 2.051379919052124,
      "learning_rate": 2.6797471688174875e-05,
      "loss": 0.7385,
      "step": 17630
    },
    {
      "epoch": 0.46434026863313144,
      "grad_norm": 1.6465489864349365,
      "learning_rate": 2.6796154859099294e-05,
      "loss": 1.8276,
      "step": 17631
    },
    {
      "epoch": 0.4643666052146431,
      "grad_norm": 1.5366212129592896,
      "learning_rate": 2.6794838030023707e-05,
      "loss": 1.8191,
      "step": 17632
    },
    {
      "epoch": 0.46439294179615487,
      "grad_norm": 1.9839245080947876,
      "learning_rate": 2.679352120094812e-05,
      "loss": 2.2081,
      "step": 17633
    },
    {
      "epoch": 0.46441927837766656,
      "grad_norm": 2.529362440109253,
      "learning_rate": 2.679220437187253e-05,
      "loss": 2.2052,
      "step": 17634
    },
    {
      "epoch": 0.4644456149591783,
      "grad_norm": 2.7027270793914795,
      "learning_rate": 2.6790887542796943e-05,
      "loss": 0.8896,
      "step": 17635
    },
    {
      "epoch": 0.46447195154069004,
      "grad_norm": 2.4789376258850098,
      "learning_rate": 2.6789570713721362e-05,
      "loss": 2.0316,
      "step": 17636
    },
    {
      "epoch": 0.46449828812220173,
      "grad_norm": 1.7484478950500488,
      "learning_rate": 2.6788253884645774e-05,
      "loss": 1.8745,
      "step": 17637
    },
    {
      "epoch": 0.4645246247037135,
      "grad_norm": 2.5566487312316895,
      "learning_rate": 2.6786937055570187e-05,
      "loss": 1.5511,
      "step": 17638
    },
    {
      "epoch": 0.46455096128522516,
      "grad_norm": 2.914454936981201,
      "learning_rate": 2.6785620226494602e-05,
      "loss": 1.4368,
      "step": 17639
    },
    {
      "epoch": 0.4645772978667369,
      "grad_norm": 1.9978471994400024,
      "learning_rate": 2.6784303397419014e-05,
      "loss": 1.3768,
      "step": 17640
    },
    {
      "epoch": 0.4646036344482486,
      "grad_norm": 2.215916156768799,
      "learning_rate": 2.6782986568343433e-05,
      "loss": 1.1244,
      "step": 17641
    },
    {
      "epoch": 0.46462997102976034,
      "grad_norm": 1.8007752895355225,
      "learning_rate": 2.6781669739267846e-05,
      "loss": 1.6294,
      "step": 17642
    },
    {
      "epoch": 0.4646563076112721,
      "grad_norm": 3.295771598815918,
      "learning_rate": 2.6780352910192258e-05,
      "loss": 0.6972,
      "step": 17643
    },
    {
      "epoch": 0.46468264419278377,
      "grad_norm": 2.87211275100708,
      "learning_rate": 2.677903608111667e-05,
      "loss": 1.4349,
      "step": 17644
    },
    {
      "epoch": 0.4647089807742955,
      "grad_norm": 1.689457893371582,
      "learning_rate": 2.677771925204109e-05,
      "loss": 2.3205,
      "step": 17645
    },
    {
      "epoch": 0.4647353173558072,
      "grad_norm": 2.2953498363494873,
      "learning_rate": 2.67764024229655e-05,
      "loss": 1.4227,
      "step": 17646
    },
    {
      "epoch": 0.46476165393731894,
      "grad_norm": 2.3698196411132812,
      "learning_rate": 2.6775085593889913e-05,
      "loss": 1.5988,
      "step": 17647
    },
    {
      "epoch": 0.46478799051883063,
      "grad_norm": 1.7133572101593018,
      "learning_rate": 2.677376876481433e-05,
      "loss": 2.2402,
      "step": 17648
    },
    {
      "epoch": 0.4648143271003424,
      "grad_norm": 2.118105411529541,
      "learning_rate": 2.677245193573874e-05,
      "loss": 1.3295,
      "step": 17649
    },
    {
      "epoch": 0.4648406636818541,
      "grad_norm": 4.4868998527526855,
      "learning_rate": 2.677113510666316e-05,
      "loss": 1.3086,
      "step": 17650
    },
    {
      "epoch": 0.4648670002633658,
      "grad_norm": 2.33290433883667,
      "learning_rate": 2.6769818277587572e-05,
      "loss": 0.7342,
      "step": 17651
    },
    {
      "epoch": 0.46489333684487755,
      "grad_norm": 2.0869383811950684,
      "learning_rate": 2.6768501448511985e-05,
      "loss": 1.6012,
      "step": 17652
    },
    {
      "epoch": 0.46491967342638923,
      "grad_norm": 3.0033810138702393,
      "learning_rate": 2.6767184619436397e-05,
      "loss": 1.715,
      "step": 17653
    },
    {
      "epoch": 0.464946010007901,
      "grad_norm": 2.684756278991699,
      "learning_rate": 2.676586779036081e-05,
      "loss": 2.2651,
      "step": 17654
    },
    {
      "epoch": 0.4649723465894127,
      "grad_norm": 1.6370489597320557,
      "learning_rate": 2.6764550961285228e-05,
      "loss": 1.7333,
      "step": 17655
    },
    {
      "epoch": 0.4649986831709244,
      "grad_norm": 1.6218193769454956,
      "learning_rate": 2.676323413220964e-05,
      "loss": 1.6966,
      "step": 17656
    },
    {
      "epoch": 0.46502501975243615,
      "grad_norm": 2.917945146560669,
      "learning_rate": 2.6761917303134056e-05,
      "loss": 1.7306,
      "step": 17657
    },
    {
      "epoch": 0.46505135633394784,
      "grad_norm": 1.9902558326721191,
      "learning_rate": 2.6760600474058468e-05,
      "loss": 1.3865,
      "step": 17658
    },
    {
      "epoch": 0.4650776929154596,
      "grad_norm": 1.5651133060455322,
      "learning_rate": 2.675928364498288e-05,
      "loss": 1.7864,
      "step": 17659
    },
    {
      "epoch": 0.46510402949697127,
      "grad_norm": 2.4701952934265137,
      "learning_rate": 2.67579668159073e-05,
      "loss": 2.3128,
      "step": 17660
    },
    {
      "epoch": 0.465130366078483,
      "grad_norm": 2.6418309211730957,
      "learning_rate": 2.675664998683171e-05,
      "loss": 1.1298,
      "step": 17661
    },
    {
      "epoch": 0.46515670265999476,
      "grad_norm": 1.8996413946151733,
      "learning_rate": 2.6755333157756124e-05,
      "loss": 1.6912,
      "step": 17662
    },
    {
      "epoch": 0.46518303924150645,
      "grad_norm": 1.6829640865325928,
      "learning_rate": 2.6754016328680536e-05,
      "loss": 0.8139,
      "step": 17663
    },
    {
      "epoch": 0.4652093758230182,
      "grad_norm": 1.9768600463867188,
      "learning_rate": 2.6752699499604955e-05,
      "loss": 1.5581,
      "step": 17664
    },
    {
      "epoch": 0.4652357124045299,
      "grad_norm": 2.5917341709136963,
      "learning_rate": 2.6751382670529367e-05,
      "loss": 1.7122,
      "step": 17665
    },
    {
      "epoch": 0.4652620489860416,
      "grad_norm": 4.4162726402282715,
      "learning_rate": 2.675006584145378e-05,
      "loss": 1.7994,
      "step": 17666
    },
    {
      "epoch": 0.4652883855675533,
      "grad_norm": 3.4718453884124756,
      "learning_rate": 2.6748749012378195e-05,
      "loss": 1.2841,
      "step": 17667
    },
    {
      "epoch": 0.46531472214906505,
      "grad_norm": 2.15373158454895,
      "learning_rate": 2.6747432183302607e-05,
      "loss": 2.0418,
      "step": 17668
    },
    {
      "epoch": 0.4653410587305768,
      "grad_norm": 1.4824116230010986,
      "learning_rate": 2.6746115354227026e-05,
      "loss": 1.5394,
      "step": 17669
    },
    {
      "epoch": 0.4653673953120885,
      "grad_norm": 5.731241703033447,
      "learning_rate": 2.6744798525151438e-05,
      "loss": 1.1126,
      "step": 17670
    },
    {
      "epoch": 0.4653937318936002,
      "grad_norm": 2.0641798973083496,
      "learning_rate": 2.674348169607585e-05,
      "loss": 1.853,
      "step": 17671
    },
    {
      "epoch": 0.4654200684751119,
      "grad_norm": 6.711053371429443,
      "learning_rate": 2.6742164867000263e-05,
      "loss": 1.9473,
      "step": 17672
    },
    {
      "epoch": 0.46544640505662366,
      "grad_norm": 2.259662389755249,
      "learning_rate": 2.6740848037924675e-05,
      "loss": 2.1195,
      "step": 17673
    },
    {
      "epoch": 0.46547274163813535,
      "grad_norm": 1.7201156616210938,
      "learning_rate": 2.6739531208849094e-05,
      "loss": 1.3219,
      "step": 17674
    },
    {
      "epoch": 0.4654990782196471,
      "grad_norm": 2.102354049682617,
      "learning_rate": 2.6738214379773506e-05,
      "loss": 0.5926,
      "step": 17675
    },
    {
      "epoch": 0.46552541480115883,
      "grad_norm": 2.0202291011810303,
      "learning_rate": 2.673689755069792e-05,
      "loss": 0.6954,
      "step": 17676
    },
    {
      "epoch": 0.4655517513826705,
      "grad_norm": 3.3457882404327393,
      "learning_rate": 2.6735580721622334e-05,
      "loss": 1.0044,
      "step": 17677
    },
    {
      "epoch": 0.46557808796418226,
      "grad_norm": 3.845686197280884,
      "learning_rate": 2.6734263892546753e-05,
      "loss": 1.5838,
      "step": 17678
    },
    {
      "epoch": 0.46560442454569395,
      "grad_norm": 1.4386787414550781,
      "learning_rate": 2.6732947063471165e-05,
      "loss": 1.6776,
      "step": 17679
    },
    {
      "epoch": 0.4656307611272057,
      "grad_norm": 2.5082128047943115,
      "learning_rate": 2.6731630234395577e-05,
      "loss": 1.9394,
      "step": 17680
    },
    {
      "epoch": 0.4656570977087174,
      "grad_norm": 2.361893653869629,
      "learning_rate": 2.673031340531999e-05,
      "loss": 2.3201,
      "step": 17681
    },
    {
      "epoch": 0.4656834342902291,
      "grad_norm": 2.3069963455200195,
      "learning_rate": 2.6728996576244402e-05,
      "loss": 2.1801,
      "step": 17682
    },
    {
      "epoch": 0.46570977087174087,
      "grad_norm": 1.9952476024627686,
      "learning_rate": 2.672767974716882e-05,
      "loss": 0.6037,
      "step": 17683
    },
    {
      "epoch": 0.46573610745325256,
      "grad_norm": 1.860139012336731,
      "learning_rate": 2.6726362918093233e-05,
      "loss": 1.4967,
      "step": 17684
    },
    {
      "epoch": 0.4657624440347643,
      "grad_norm": 3.3513987064361572,
      "learning_rate": 2.672504608901765e-05,
      "loss": 1.5793,
      "step": 17685
    },
    {
      "epoch": 0.465788780616276,
      "grad_norm": 2.2791037559509277,
      "learning_rate": 2.672372925994206e-05,
      "loss": 1.7081,
      "step": 17686
    },
    {
      "epoch": 0.46581511719778773,
      "grad_norm": 1.890899419784546,
      "learning_rate": 2.6722412430866473e-05,
      "loss": 1.5473,
      "step": 17687
    },
    {
      "epoch": 0.4658414537792994,
      "grad_norm": 3.320352792739868,
      "learning_rate": 2.6721095601790892e-05,
      "loss": 0.8191,
      "step": 17688
    },
    {
      "epoch": 0.46586779036081116,
      "grad_norm": 1.7549089193344116,
      "learning_rate": 2.6719778772715304e-05,
      "loss": 1.816,
      "step": 17689
    },
    {
      "epoch": 0.4658941269423229,
      "grad_norm": 1.6220976114273071,
      "learning_rate": 2.6718461943639716e-05,
      "loss": 1.558,
      "step": 17690
    },
    {
      "epoch": 0.4659204635238346,
      "grad_norm": 2.1625964641571045,
      "learning_rate": 2.671714511456413e-05,
      "loss": 1.6821,
      "step": 17691
    },
    {
      "epoch": 0.46594680010534634,
      "grad_norm": 1.5719045400619507,
      "learning_rate": 2.671582828548854e-05,
      "loss": 1.9854,
      "step": 17692
    },
    {
      "epoch": 0.465973136686858,
      "grad_norm": 2.02211856842041,
      "learning_rate": 2.671451145641296e-05,
      "loss": 1.8916,
      "step": 17693
    },
    {
      "epoch": 0.46599947326836977,
      "grad_norm": 2.3192813396453857,
      "learning_rate": 2.6713194627337372e-05,
      "loss": 1.6054,
      "step": 17694
    },
    {
      "epoch": 0.4660258098498815,
      "grad_norm": 1.6684606075286865,
      "learning_rate": 2.6711877798261788e-05,
      "loss": 1.1422,
      "step": 17695
    },
    {
      "epoch": 0.4660521464313932,
      "grad_norm": 2.1002683639526367,
      "learning_rate": 2.67105609691862e-05,
      "loss": 2.1899,
      "step": 17696
    },
    {
      "epoch": 0.46607848301290494,
      "grad_norm": 1.7114143371582031,
      "learning_rate": 2.670924414011062e-05,
      "loss": 0.2672,
      "step": 17697
    },
    {
      "epoch": 0.46610481959441663,
      "grad_norm": 4.462927341461182,
      "learning_rate": 2.670792731103503e-05,
      "loss": 1.9146,
      "step": 17698
    },
    {
      "epoch": 0.4661311561759284,
      "grad_norm": 1.6798112392425537,
      "learning_rate": 2.6706610481959443e-05,
      "loss": 1.3381,
      "step": 17699
    },
    {
      "epoch": 0.46615749275744006,
      "grad_norm": 1.7059693336486816,
      "learning_rate": 2.6705293652883855e-05,
      "loss": 1.9451,
      "step": 17700
    },
    {
      "epoch": 0.4661838293389518,
      "grad_norm": 1.4935001134872437,
      "learning_rate": 2.6703976823808268e-05,
      "loss": 1.7626,
      "step": 17701
    },
    {
      "epoch": 0.46621016592046355,
      "grad_norm": 1.8951994180679321,
      "learning_rate": 2.6702659994732687e-05,
      "loss": 2.3542,
      "step": 17702
    },
    {
      "epoch": 0.46623650250197524,
      "grad_norm": 2.825207233428955,
      "learning_rate": 2.67013431656571e-05,
      "loss": 0.9514,
      "step": 17703
    },
    {
      "epoch": 0.466262839083487,
      "grad_norm": 4.246274948120117,
      "learning_rate": 2.6700026336581514e-05,
      "loss": 1.3566,
      "step": 17704
    },
    {
      "epoch": 0.46628917566499867,
      "grad_norm": 2.1827847957611084,
      "learning_rate": 2.6698709507505927e-05,
      "loss": 2.1029,
      "step": 17705
    },
    {
      "epoch": 0.4663155122465104,
      "grad_norm": 2.0452218055725098,
      "learning_rate": 2.669739267843034e-05,
      "loss": 1.7554,
      "step": 17706
    },
    {
      "epoch": 0.4663418488280221,
      "grad_norm": 1.929308295249939,
      "learning_rate": 2.6696075849354758e-05,
      "loss": 1.3093,
      "step": 17707
    },
    {
      "epoch": 0.46636818540953384,
      "grad_norm": 1.682302713394165,
      "learning_rate": 2.669475902027917e-05,
      "loss": 1.5466,
      "step": 17708
    },
    {
      "epoch": 0.4663945219910456,
      "grad_norm": 1.8157784938812256,
      "learning_rate": 2.6693442191203582e-05,
      "loss": 0.9867,
      "step": 17709
    },
    {
      "epoch": 0.4664208585725573,
      "grad_norm": 6.393119812011719,
      "learning_rate": 2.6692125362127994e-05,
      "loss": 1.3851,
      "step": 17710
    },
    {
      "epoch": 0.466447195154069,
      "grad_norm": 1.5145739316940308,
      "learning_rate": 2.6690808533052413e-05,
      "loss": 1.6247,
      "step": 17711
    },
    {
      "epoch": 0.4664735317355807,
      "grad_norm": 1.7255464792251587,
      "learning_rate": 2.6689491703976826e-05,
      "loss": 1.9207,
      "step": 17712
    },
    {
      "epoch": 0.46649986831709245,
      "grad_norm": 13.894375801086426,
      "learning_rate": 2.6688174874901238e-05,
      "loss": 1.4912,
      "step": 17713
    },
    {
      "epoch": 0.46652620489860414,
      "grad_norm": 2.2871508598327637,
      "learning_rate": 2.6686858045825653e-05,
      "loss": 1.4475,
      "step": 17714
    },
    {
      "epoch": 0.4665525414801159,
      "grad_norm": 1.7496458292007446,
      "learning_rate": 2.6685541216750066e-05,
      "loss": 2.0623,
      "step": 17715
    },
    {
      "epoch": 0.4665788780616276,
      "grad_norm": 2.191917896270752,
      "learning_rate": 2.6684224387674485e-05,
      "loss": 1.9308,
      "step": 17716
    },
    {
      "epoch": 0.4666052146431393,
      "grad_norm": 1.4847074747085571,
      "learning_rate": 2.6682907558598897e-05,
      "loss": 1.2611,
      "step": 17717
    },
    {
      "epoch": 0.46663155122465105,
      "grad_norm": 1.7192156314849854,
      "learning_rate": 2.668159072952331e-05,
      "loss": 1.5391,
      "step": 17718
    },
    {
      "epoch": 0.46665788780616274,
      "grad_norm": 2.1765732765197754,
      "learning_rate": 2.668027390044772e-05,
      "loss": 0.4912,
      "step": 17719
    },
    {
      "epoch": 0.4666842243876745,
      "grad_norm": 2.5378360748291016,
      "learning_rate": 2.6678957071372133e-05,
      "loss": 1.4397,
      "step": 17720
    },
    {
      "epoch": 0.4667105609691862,
      "grad_norm": 1.8803786039352417,
      "learning_rate": 2.6677640242296552e-05,
      "loss": 1.7115,
      "step": 17721
    },
    {
      "epoch": 0.4667368975506979,
      "grad_norm": 1.6991279125213623,
      "learning_rate": 2.6676323413220965e-05,
      "loss": 1.83,
      "step": 17722
    },
    {
      "epoch": 0.46676323413220966,
      "grad_norm": 4.650506496429443,
      "learning_rate": 2.667500658414538e-05,
      "loss": 1.0469,
      "step": 17723
    },
    {
      "epoch": 0.46678957071372135,
      "grad_norm": 1.472511887550354,
      "learning_rate": 2.6673689755069792e-05,
      "loss": 1.763,
      "step": 17724
    },
    {
      "epoch": 0.4668159072952331,
      "grad_norm": 5.536637306213379,
      "learning_rate": 2.6672372925994205e-05,
      "loss": 0.6791,
      "step": 17725
    },
    {
      "epoch": 0.4668422438767448,
      "grad_norm": 1.6572816371917725,
      "learning_rate": 2.6671056096918624e-05,
      "loss": 1.9978,
      "step": 17726
    },
    {
      "epoch": 0.4668685804582565,
      "grad_norm": 1.8187482357025146,
      "learning_rate": 2.6669739267843036e-05,
      "loss": 1.5962,
      "step": 17727
    },
    {
      "epoch": 0.46689491703976826,
      "grad_norm": 2.826907157897949,
      "learning_rate": 2.6668422438767448e-05,
      "loss": 1.0366,
      "step": 17728
    },
    {
      "epoch": 0.46692125362127995,
      "grad_norm": 2.902482509613037,
      "learning_rate": 2.666710560969186e-05,
      "loss": 2.1595,
      "step": 17729
    },
    {
      "epoch": 0.4669475902027917,
      "grad_norm": 1.9059252738952637,
      "learning_rate": 2.666578878061628e-05,
      "loss": 2.1371,
      "step": 17730
    },
    {
      "epoch": 0.4669739267843034,
      "grad_norm": 3.6826422214508057,
      "learning_rate": 2.666447195154069e-05,
      "loss": 1.5158,
      "step": 17731
    },
    {
      "epoch": 0.4670002633658151,
      "grad_norm": 1.3771990537643433,
      "learning_rate": 2.6663155122465107e-05,
      "loss": 0.6562,
      "step": 17732
    },
    {
      "epoch": 0.4670265999473268,
      "grad_norm": 2.0774521827697754,
      "learning_rate": 2.666183829338952e-05,
      "loss": 0.9267,
      "step": 17733
    },
    {
      "epoch": 0.46705293652883856,
      "grad_norm": 1.9947038888931274,
      "learning_rate": 2.666052146431393e-05,
      "loss": 0.8888,
      "step": 17734
    },
    {
      "epoch": 0.4670792731103503,
      "grad_norm": 1.6749708652496338,
      "learning_rate": 2.665920463523835e-05,
      "loss": 1.7476,
      "step": 17735
    },
    {
      "epoch": 0.467105609691862,
      "grad_norm": 3.1274302005767822,
      "learning_rate": 2.6657887806162763e-05,
      "loss": 1.3334,
      "step": 17736
    },
    {
      "epoch": 0.46713194627337373,
      "grad_norm": 1.623986840248108,
      "learning_rate": 2.6656570977087175e-05,
      "loss": 1.9714,
      "step": 17737
    },
    {
      "epoch": 0.4671582828548854,
      "grad_norm": 1.9798483848571777,
      "learning_rate": 2.6655254148011587e-05,
      "loss": 2.3071,
      "step": 17738
    },
    {
      "epoch": 0.46718461943639716,
      "grad_norm": 2.107879877090454,
      "learning_rate": 2.6653937318936e-05,
      "loss": 2.0739,
      "step": 17739
    },
    {
      "epoch": 0.46721095601790885,
      "grad_norm": 1.8291912078857422,
      "learning_rate": 2.6652620489860418e-05,
      "loss": 0.6253,
      "step": 17740
    },
    {
      "epoch": 0.4672372925994206,
      "grad_norm": 1.672823429107666,
      "learning_rate": 2.665130366078483e-05,
      "loss": 1.4551,
      "step": 17741
    },
    {
      "epoch": 0.46726362918093234,
      "grad_norm": 3.9399280548095703,
      "learning_rate": 2.6649986831709246e-05,
      "loss": 0.9126,
      "step": 17742
    },
    {
      "epoch": 0.467289965762444,
      "grad_norm": 1.855733036994934,
      "learning_rate": 2.6648670002633658e-05,
      "loss": 0.7685,
      "step": 17743
    },
    {
      "epoch": 0.46731630234395577,
      "grad_norm": 1.9839106798171997,
      "learning_rate": 2.6647353173558077e-05,
      "loss": 0.8939,
      "step": 17744
    },
    {
      "epoch": 0.46734263892546746,
      "grad_norm": 1.5897464752197266,
      "learning_rate": 2.664603634448249e-05,
      "loss": 2.4577,
      "step": 17745
    },
    {
      "epoch": 0.4673689755069792,
      "grad_norm": 1.4755569696426392,
      "learning_rate": 2.6644719515406902e-05,
      "loss": 1.9235,
      "step": 17746
    },
    {
      "epoch": 0.4673953120884909,
      "grad_norm": 4.208126068115234,
      "learning_rate": 2.6643402686331314e-05,
      "loss": 0.7671,
      "step": 17747
    },
    {
      "epoch": 0.46742164867000263,
      "grad_norm": 2.061633348464966,
      "learning_rate": 2.6642085857255726e-05,
      "loss": 1.8405,
      "step": 17748
    },
    {
      "epoch": 0.4674479852515144,
      "grad_norm": 4.189487457275391,
      "learning_rate": 2.6640769028180145e-05,
      "loss": 0.8534,
      "step": 17749
    },
    {
      "epoch": 0.46747432183302606,
      "grad_norm": 3.2876577377319336,
      "learning_rate": 2.6639452199104557e-05,
      "loss": 1.3976,
      "step": 17750
    },
    {
      "epoch": 0.4675006584145378,
      "grad_norm": 2.196082830429077,
      "learning_rate": 2.6638135370028973e-05,
      "loss": 2.6455,
      "step": 17751
    },
    {
      "epoch": 0.4675269949960495,
      "grad_norm": 2.737238883972168,
      "learning_rate": 2.6636818540953385e-05,
      "loss": 1.1856,
      "step": 17752
    },
    {
      "epoch": 0.46755333157756124,
      "grad_norm": 4.558173656463623,
      "learning_rate": 2.6635501711877797e-05,
      "loss": 1.3998,
      "step": 17753
    },
    {
      "epoch": 0.4675796681590729,
      "grad_norm": 2.6295478343963623,
      "learning_rate": 2.6634184882802216e-05,
      "loss": 0.7673,
      "step": 17754
    },
    {
      "epoch": 0.46760600474058467,
      "grad_norm": 2.6075387001037598,
      "learning_rate": 2.663286805372663e-05,
      "loss": 1.9034,
      "step": 17755
    },
    {
      "epoch": 0.4676323413220964,
      "grad_norm": 2.605494260787964,
      "learning_rate": 2.663155122465104e-05,
      "loss": 2.1248,
      "step": 17756
    },
    {
      "epoch": 0.4676586779036081,
      "grad_norm": 2.8191702365875244,
      "learning_rate": 2.6630234395575453e-05,
      "loss": 1.8037,
      "step": 17757
    },
    {
      "epoch": 0.46768501448511984,
      "grad_norm": 1.878612756729126,
      "learning_rate": 2.662891756649987e-05,
      "loss": 1.8943,
      "step": 17758
    },
    {
      "epoch": 0.46771135106663153,
      "grad_norm": 1.842537522315979,
      "learning_rate": 2.6627600737424284e-05,
      "loss": 1.8705,
      "step": 17759
    },
    {
      "epoch": 0.4677376876481433,
      "grad_norm": 1.7836134433746338,
      "learning_rate": 2.66262839083487e-05,
      "loss": 0.5769,
      "step": 17760
    },
    {
      "epoch": 0.467764024229655,
      "grad_norm": 1.5259770154953003,
      "learning_rate": 2.6624967079273112e-05,
      "loss": 0.2331,
      "step": 17761
    },
    {
      "epoch": 0.4677903608111667,
      "grad_norm": 2.6390223503112793,
      "learning_rate": 2.6623650250197524e-05,
      "loss": 2.0226,
      "step": 17762
    },
    {
      "epoch": 0.46781669739267845,
      "grad_norm": 2.742978572845459,
      "learning_rate": 2.6622333421121943e-05,
      "loss": 2.0972,
      "step": 17763
    },
    {
      "epoch": 0.46784303397419014,
      "grad_norm": 2.2002484798431396,
      "learning_rate": 2.6621016592046355e-05,
      "loss": 1.7524,
      "step": 17764
    },
    {
      "epoch": 0.4678693705557019,
      "grad_norm": 1.9573888778686523,
      "learning_rate": 2.6619699762970768e-05,
      "loss": 1.6812,
      "step": 17765
    },
    {
      "epoch": 0.46789570713721357,
      "grad_norm": 1.5877116918563843,
      "learning_rate": 2.661838293389518e-05,
      "loss": 1.8839,
      "step": 17766
    },
    {
      "epoch": 0.4679220437187253,
      "grad_norm": 2.439161777496338,
      "learning_rate": 2.6617066104819592e-05,
      "loss": 1.6529,
      "step": 17767
    },
    {
      "epoch": 0.46794838030023705,
      "grad_norm": 1.6073988676071167,
      "learning_rate": 2.661574927574401e-05,
      "loss": 1.499,
      "step": 17768
    },
    {
      "epoch": 0.46797471688174874,
      "grad_norm": 1.8914847373962402,
      "learning_rate": 2.6614432446668423e-05,
      "loss": 1.3774,
      "step": 17769
    },
    {
      "epoch": 0.4680010534632605,
      "grad_norm": 4.3826189041137695,
      "learning_rate": 2.661311561759284e-05,
      "loss": 0.4888,
      "step": 17770
    },
    {
      "epoch": 0.4680273900447722,
      "grad_norm": 1.7288110256195068,
      "learning_rate": 2.661179878851725e-05,
      "loss": 1.8116,
      "step": 17771
    },
    {
      "epoch": 0.4680537266262839,
      "grad_norm": 3.684248924255371,
      "learning_rate": 2.6610481959441663e-05,
      "loss": 1.4851,
      "step": 17772
    },
    {
      "epoch": 0.4680800632077956,
      "grad_norm": 1.9046207666397095,
      "learning_rate": 2.6609165130366082e-05,
      "loss": 1.6437,
      "step": 17773
    },
    {
      "epoch": 0.46810639978930735,
      "grad_norm": 3.1491525173187256,
      "learning_rate": 2.6607848301290494e-05,
      "loss": 1.2911,
      "step": 17774
    },
    {
      "epoch": 0.4681327363708191,
      "grad_norm": 3.6731884479522705,
      "learning_rate": 2.6606531472214907e-05,
      "loss": 2.0269,
      "step": 17775
    },
    {
      "epoch": 0.4681590729523308,
      "grad_norm": 2.171692132949829,
      "learning_rate": 2.660521464313932e-05,
      "loss": 2.4977,
      "step": 17776
    },
    {
      "epoch": 0.4681854095338425,
      "grad_norm": 2.751981019973755,
      "learning_rate": 2.6603897814063738e-05,
      "loss": 1.9733,
      "step": 17777
    },
    {
      "epoch": 0.4682117461153542,
      "grad_norm": 1.8141319751739502,
      "learning_rate": 2.660258098498815e-05,
      "loss": 1.6459,
      "step": 17778
    },
    {
      "epoch": 0.46823808269686595,
      "grad_norm": 2.73344087600708,
      "learning_rate": 2.6601264155912566e-05,
      "loss": 1.9443,
      "step": 17779
    },
    {
      "epoch": 0.46826441927837764,
      "grad_norm": 2.477538585662842,
      "learning_rate": 2.6599947326836978e-05,
      "loss": 1.2884,
      "step": 17780
    },
    {
      "epoch": 0.4682907558598894,
      "grad_norm": 1.8230829238891602,
      "learning_rate": 2.659863049776139e-05,
      "loss": 2.0744,
      "step": 17781
    },
    {
      "epoch": 0.46831709244140113,
      "grad_norm": 2.1775026321411133,
      "learning_rate": 2.659731366868581e-05,
      "loss": 1.5417,
      "step": 17782
    },
    {
      "epoch": 0.4683434290229128,
      "grad_norm": 1.6220475435256958,
      "learning_rate": 2.659599683961022e-05,
      "loss": 1.8476,
      "step": 17783
    },
    {
      "epoch": 0.46836976560442456,
      "grad_norm": 1.5239176750183105,
      "learning_rate": 2.6594680010534633e-05,
      "loss": 1.5773,
      "step": 17784
    },
    {
      "epoch": 0.46839610218593625,
      "grad_norm": 1.610787034034729,
      "learning_rate": 2.6593363181459046e-05,
      "loss": 1.6851,
      "step": 17785
    },
    {
      "epoch": 0.468422438767448,
      "grad_norm": 4.847218990325928,
      "learning_rate": 2.6592046352383458e-05,
      "loss": 1.3044,
      "step": 17786
    },
    {
      "epoch": 0.4684487753489597,
      "grad_norm": 2.2366230487823486,
      "learning_rate": 2.6590729523307877e-05,
      "loss": 1.9313,
      "step": 17787
    },
    {
      "epoch": 0.4684751119304714,
      "grad_norm": 4.066692352294922,
      "learning_rate": 2.6589412694232292e-05,
      "loss": 1.9231,
      "step": 17788
    },
    {
      "epoch": 0.46850144851198317,
      "grad_norm": 2.2789418697357178,
      "learning_rate": 2.6588095865156705e-05,
      "loss": 0.8924,
      "step": 17789
    },
    {
      "epoch": 0.46852778509349485,
      "grad_norm": 1.9849071502685547,
      "learning_rate": 2.6586779036081117e-05,
      "loss": 1.7841,
      "step": 17790
    },
    {
      "epoch": 0.4685541216750066,
      "grad_norm": 1.7239867448806763,
      "learning_rate": 2.658546220700553e-05,
      "loss": 0.6548,
      "step": 17791
    },
    {
      "epoch": 0.4685804582565183,
      "grad_norm": 5.8985700607299805,
      "learning_rate": 2.6584145377929948e-05,
      "loss": 1.1502,
      "step": 17792
    },
    {
      "epoch": 0.46860679483803,
      "grad_norm": 1.953762173652649,
      "learning_rate": 2.658282854885436e-05,
      "loss": 0.5572,
      "step": 17793
    },
    {
      "epoch": 0.46863313141954177,
      "grad_norm": 1.599268913269043,
      "learning_rate": 2.6581511719778772e-05,
      "loss": 1.6468,
      "step": 17794
    },
    {
      "epoch": 0.46865946800105346,
      "grad_norm": 1.4985997676849365,
      "learning_rate": 2.6580194890703185e-05,
      "loss": 1.8841,
      "step": 17795
    },
    {
      "epoch": 0.4686858045825652,
      "grad_norm": 1.9278223514556885,
      "learning_rate": 2.6578878061627604e-05,
      "loss": 1.6121,
      "step": 17796
    },
    {
      "epoch": 0.4687121411640769,
      "grad_norm": 2.5737249851226807,
      "learning_rate": 2.6577561232552016e-05,
      "loss": 2.1832,
      "step": 17797
    },
    {
      "epoch": 0.46873847774558863,
      "grad_norm": 2.342465400695801,
      "learning_rate": 2.657624440347643e-05,
      "loss": 2.401,
      "step": 17798
    },
    {
      "epoch": 0.4687648143271003,
      "grad_norm": 3.326110601425171,
      "learning_rate": 2.6574927574400844e-05,
      "loss": 0.7558,
      "step": 17799
    },
    {
      "epoch": 0.46879115090861206,
      "grad_norm": 1.8555642366409302,
      "learning_rate": 2.6573610745325256e-05,
      "loss": 1.6713,
      "step": 17800
    },
    {
      "epoch": 0.4688174874901238,
      "grad_norm": 1.4029220342636108,
      "learning_rate": 2.6572293916249675e-05,
      "loss": 1.1246,
      "step": 17801
    },
    {
      "epoch": 0.4688438240716355,
      "grad_norm": 7.696796417236328,
      "learning_rate": 2.6570977087174087e-05,
      "loss": 2.863,
      "step": 17802
    },
    {
      "epoch": 0.46887016065314724,
      "grad_norm": 1.6174547672271729,
      "learning_rate": 2.65696602580985e-05,
      "loss": 2.256,
      "step": 17803
    },
    {
      "epoch": 0.4688964972346589,
      "grad_norm": 4.064257621765137,
      "learning_rate": 2.656834342902291e-05,
      "loss": 1.4144,
      "step": 17804
    },
    {
      "epoch": 0.46892283381617067,
      "grad_norm": 1.9717520475387573,
      "learning_rate": 2.6567026599947327e-05,
      "loss": 1.9244,
      "step": 17805
    },
    {
      "epoch": 0.46894917039768236,
      "grad_norm": 2.2301363945007324,
      "learning_rate": 2.6565709770871743e-05,
      "loss": 1.712,
      "step": 17806
    },
    {
      "epoch": 0.4689755069791941,
      "grad_norm": 1.8241791725158691,
      "learning_rate": 2.6564392941796158e-05,
      "loss": 1.6233,
      "step": 17807
    },
    {
      "epoch": 0.46900184356070584,
      "grad_norm": 2.6408090591430664,
      "learning_rate": 2.656307611272057e-05,
      "loss": 2.1652,
      "step": 17808
    },
    {
      "epoch": 0.46902818014221753,
      "grad_norm": 2.1685352325439453,
      "learning_rate": 2.6561759283644983e-05,
      "loss": 1.9775,
      "step": 17809
    },
    {
      "epoch": 0.4690545167237293,
      "grad_norm": 3.425508499145508,
      "learning_rate": 2.6560442454569402e-05,
      "loss": 2.2296,
      "step": 17810
    },
    {
      "epoch": 0.46908085330524096,
      "grad_norm": 1.4673404693603516,
      "learning_rate": 2.6559125625493814e-05,
      "loss": 1.8413,
      "step": 17811
    },
    {
      "epoch": 0.4691071898867527,
      "grad_norm": 1.9336313009262085,
      "learning_rate": 2.6557808796418226e-05,
      "loss": 2.6583,
      "step": 17812
    },
    {
      "epoch": 0.4691335264682644,
      "grad_norm": 1.85164213180542,
      "learning_rate": 2.655649196734264e-05,
      "loss": 2.2075,
      "step": 17813
    },
    {
      "epoch": 0.46915986304977614,
      "grad_norm": 1.7610957622528076,
      "learning_rate": 2.655517513826705e-05,
      "loss": 1.7066,
      "step": 17814
    },
    {
      "epoch": 0.4691861996312879,
      "grad_norm": 1.6094934940338135,
      "learning_rate": 2.655385830919147e-05,
      "loss": 1.6337,
      "step": 17815
    },
    {
      "epoch": 0.46921253621279957,
      "grad_norm": 1.860531210899353,
      "learning_rate": 2.6552541480115882e-05,
      "loss": 1.6072,
      "step": 17816
    },
    {
      "epoch": 0.4692388727943113,
      "grad_norm": 2.4694485664367676,
      "learning_rate": 2.6551224651040297e-05,
      "loss": 1.3012,
      "step": 17817
    },
    {
      "epoch": 0.469265209375823,
      "grad_norm": 1.6511313915252686,
      "learning_rate": 2.654990782196471e-05,
      "loss": 1.4186,
      "step": 17818
    },
    {
      "epoch": 0.46929154595733474,
      "grad_norm": 1.8920927047729492,
      "learning_rate": 2.6548590992889122e-05,
      "loss": 2.0614,
      "step": 17819
    },
    {
      "epoch": 0.46931788253884643,
      "grad_norm": 4.017232894897461,
      "learning_rate": 2.654727416381354e-05,
      "loss": 1.0195,
      "step": 17820
    },
    {
      "epoch": 0.4693442191203582,
      "grad_norm": 1.7876805067062378,
      "learning_rate": 2.6545957334737953e-05,
      "loss": 2.0411,
      "step": 17821
    },
    {
      "epoch": 0.4693705557018699,
      "grad_norm": 3.1118571758270264,
      "learning_rate": 2.6544640505662365e-05,
      "loss": 1.5229,
      "step": 17822
    },
    {
      "epoch": 0.4693968922833816,
      "grad_norm": 1.5109143257141113,
      "learning_rate": 2.6543323676586777e-05,
      "loss": 1.8546,
      "step": 17823
    },
    {
      "epoch": 0.46942322886489335,
      "grad_norm": 1.5621002912521362,
      "learning_rate": 2.6542006847511193e-05,
      "loss": 1.9555,
      "step": 17824
    },
    {
      "epoch": 0.46944956544640504,
      "grad_norm": 1.9511710405349731,
      "learning_rate": 2.654069001843561e-05,
      "loss": 1.85,
      "step": 17825
    },
    {
      "epoch": 0.4694759020279168,
      "grad_norm": 2.2528293132781982,
      "learning_rate": 2.6539373189360024e-05,
      "loss": 1.5248,
      "step": 17826
    },
    {
      "epoch": 0.4695022386094285,
      "grad_norm": 1.5066628456115723,
      "learning_rate": 2.6538056360284436e-05,
      "loss": 1.6815,
      "step": 17827
    },
    {
      "epoch": 0.4695285751909402,
      "grad_norm": 1.8154276609420776,
      "learning_rate": 2.653673953120885e-05,
      "loss": 0.7288,
      "step": 17828
    },
    {
      "epoch": 0.46955491177245195,
      "grad_norm": 2.893597364425659,
      "learning_rate": 2.6535422702133268e-05,
      "loss": 1.793,
      "step": 17829
    },
    {
      "epoch": 0.46958124835396364,
      "grad_norm": 1.9779239892959595,
      "learning_rate": 2.653410587305768e-05,
      "loss": 1.9619,
      "step": 17830
    },
    {
      "epoch": 0.4696075849354754,
      "grad_norm": 4.148197174072266,
      "learning_rate": 2.6532789043982092e-05,
      "loss": 1.1131,
      "step": 17831
    },
    {
      "epoch": 0.4696339215169871,
      "grad_norm": 1.5699442625045776,
      "learning_rate": 2.6531472214906504e-05,
      "loss": 1.6655,
      "step": 17832
    },
    {
      "epoch": 0.4696602580984988,
      "grad_norm": 1.997094988822937,
      "learning_rate": 2.653015538583092e-05,
      "loss": 0.9221,
      "step": 17833
    },
    {
      "epoch": 0.46968659468001056,
      "grad_norm": 1.4420344829559326,
      "learning_rate": 2.6528838556755335e-05,
      "loss": 1.3992,
      "step": 17834
    },
    {
      "epoch": 0.46971293126152225,
      "grad_norm": 2.0375351905822754,
      "learning_rate": 2.652752172767975e-05,
      "loss": 1.6928,
      "step": 17835
    },
    {
      "epoch": 0.469739267843034,
      "grad_norm": 1.5903856754302979,
      "learning_rate": 2.6526204898604163e-05,
      "loss": 0.8805,
      "step": 17836
    },
    {
      "epoch": 0.4697656044245457,
      "grad_norm": 2.220182418823242,
      "learning_rate": 2.6524888069528575e-05,
      "loss": 1.5697,
      "step": 17837
    },
    {
      "epoch": 0.4697919410060574,
      "grad_norm": 4.313830375671387,
      "learning_rate": 2.6523571240452988e-05,
      "loss": 1.0573,
      "step": 17838
    },
    {
      "epoch": 0.4698182775875691,
      "grad_norm": 4.469368934631348,
      "learning_rate": 2.6522254411377407e-05,
      "loss": 1.5517,
      "step": 17839
    },
    {
      "epoch": 0.46984461416908085,
      "grad_norm": 2.0591492652893066,
      "learning_rate": 2.652093758230182e-05,
      "loss": 2.056,
      "step": 17840
    },
    {
      "epoch": 0.4698709507505926,
      "grad_norm": 2.1079163551330566,
      "learning_rate": 2.651962075322623e-05,
      "loss": 1.9779,
      "step": 17841
    },
    {
      "epoch": 0.4698972873321043,
      "grad_norm": 2.601976156234741,
      "learning_rate": 2.6518303924150643e-05,
      "loss": 1.542,
      "step": 17842
    },
    {
      "epoch": 0.46992362391361603,
      "grad_norm": 3.2868494987487793,
      "learning_rate": 2.6516987095075062e-05,
      "loss": 0.8687,
      "step": 17843
    },
    {
      "epoch": 0.4699499604951277,
      "grad_norm": 2.5924289226531982,
      "learning_rate": 2.6515670265999474e-05,
      "loss": 0.5282,
      "step": 17844
    },
    {
      "epoch": 0.46997629707663946,
      "grad_norm": 1.372226595878601,
      "learning_rate": 2.651435343692389e-05,
      "loss": 1.534,
      "step": 17845
    },
    {
      "epoch": 0.47000263365815115,
      "grad_norm": 2.584397077560425,
      "learning_rate": 2.6513036607848302e-05,
      "loss": 1.6455,
      "step": 17846
    },
    {
      "epoch": 0.4700289702396629,
      "grad_norm": 1.9899561405181885,
      "learning_rate": 2.6511719778772714e-05,
      "loss": 2.0372,
      "step": 17847
    },
    {
      "epoch": 0.47005530682117463,
      "grad_norm": 1.6505098342895508,
      "learning_rate": 2.6510402949697133e-05,
      "loss": 0.5741,
      "step": 17848
    },
    {
      "epoch": 0.4700816434026863,
      "grad_norm": 2.215165376663208,
      "learning_rate": 2.6509086120621546e-05,
      "loss": 1.6832,
      "step": 17849
    },
    {
      "epoch": 0.47010797998419807,
      "grad_norm": 1.9001377820968628,
      "learning_rate": 2.6507769291545958e-05,
      "loss": 1.8595,
      "step": 17850
    },
    {
      "epoch": 0.47013431656570975,
      "grad_norm": 1.6046241521835327,
      "learning_rate": 2.650645246247037e-05,
      "loss": 0.3189,
      "step": 17851
    },
    {
      "epoch": 0.4701606531472215,
      "grad_norm": 1.8992936611175537,
      "learning_rate": 2.6505135633394786e-05,
      "loss": 1.89,
      "step": 17852
    },
    {
      "epoch": 0.4701869897287332,
      "grad_norm": 2.1740028858184814,
      "learning_rate": 2.65038188043192e-05,
      "loss": 0.4599,
      "step": 17853
    },
    {
      "epoch": 0.47021332631024493,
      "grad_norm": 1.810161828994751,
      "learning_rate": 2.6502501975243617e-05,
      "loss": 1.7993,
      "step": 17854
    },
    {
      "epoch": 0.47023966289175667,
      "grad_norm": 3.468501091003418,
      "learning_rate": 2.650118514616803e-05,
      "loss": 0.9058,
      "step": 17855
    },
    {
      "epoch": 0.47026599947326836,
      "grad_norm": 1.8690470457077026,
      "learning_rate": 2.649986831709244e-05,
      "loss": 1.6319,
      "step": 17856
    },
    {
      "epoch": 0.4702923360547801,
      "grad_norm": 1.8879598379135132,
      "learning_rate": 2.6498551488016853e-05,
      "loss": 1.692,
      "step": 17857
    },
    {
      "epoch": 0.4703186726362918,
      "grad_norm": 2.298372983932495,
      "learning_rate": 2.6497234658941272e-05,
      "loss": 1.6616,
      "step": 17858
    },
    {
      "epoch": 0.47034500921780353,
      "grad_norm": 2.492218017578125,
      "learning_rate": 2.6495917829865685e-05,
      "loss": 1.9995,
      "step": 17859
    },
    {
      "epoch": 0.4703713457993152,
      "grad_norm": 1.637429118156433,
      "learning_rate": 2.6494601000790097e-05,
      "loss": 2.4575,
      "step": 17860
    },
    {
      "epoch": 0.47039768238082696,
      "grad_norm": 3.52032208442688,
      "learning_rate": 2.6493284171714512e-05,
      "loss": 1.7719,
      "step": 17861
    },
    {
      "epoch": 0.4704240189623387,
      "grad_norm": 2.648847818374634,
      "learning_rate": 2.6491967342638928e-05,
      "loss": 1.8128,
      "step": 17862
    },
    {
      "epoch": 0.4704503555438504,
      "grad_norm": 1.9688808917999268,
      "learning_rate": 2.6490650513563344e-05,
      "loss": 1.785,
      "step": 17863
    },
    {
      "epoch": 0.47047669212536214,
      "grad_norm": 2.3816568851470947,
      "learning_rate": 2.6489333684487756e-05,
      "loss": 1.5061,
      "step": 17864
    },
    {
      "epoch": 0.4705030287068738,
      "grad_norm": 1.7334983348846436,
      "learning_rate": 2.6488016855412168e-05,
      "loss": 1.4209,
      "step": 17865
    },
    {
      "epoch": 0.47052936528838557,
      "grad_norm": 1.994781732559204,
      "learning_rate": 2.648670002633658e-05,
      "loss": 1.7904,
      "step": 17866
    },
    {
      "epoch": 0.4705557018698973,
      "grad_norm": 1.6072520017623901,
      "learning_rate": 2.6485383197261e-05,
      "loss": 1.8495,
      "step": 17867
    },
    {
      "epoch": 0.470582038451409,
      "grad_norm": 1.9076699018478394,
      "learning_rate": 2.648406636818541e-05,
      "loss": 1.8467,
      "step": 17868
    },
    {
      "epoch": 0.47060837503292074,
      "grad_norm": 2.63183331489563,
      "learning_rate": 2.6482749539109824e-05,
      "loss": 0.3947,
      "step": 17869
    },
    {
      "epoch": 0.47063471161443243,
      "grad_norm": 2.840358257293701,
      "learning_rate": 2.6481432710034236e-05,
      "loss": 1.7305,
      "step": 17870
    },
    {
      "epoch": 0.4706610481959442,
      "grad_norm": 1.6165119409561157,
      "learning_rate": 2.648011588095865e-05,
      "loss": 1.5707,
      "step": 17871
    },
    {
      "epoch": 0.47068738477745586,
      "grad_norm": 1.985432744026184,
      "learning_rate": 2.6478799051883067e-05,
      "loss": 2.2516,
      "step": 17872
    },
    {
      "epoch": 0.4707137213589676,
      "grad_norm": 2.791285991668701,
      "learning_rate": 2.6477482222807483e-05,
      "loss": 2.2673,
      "step": 17873
    },
    {
      "epoch": 0.47074005794047935,
      "grad_norm": 2.594158172607422,
      "learning_rate": 2.6476165393731895e-05,
      "loss": 1.5482,
      "step": 17874
    },
    {
      "epoch": 0.47076639452199104,
      "grad_norm": 1.4221909046173096,
      "learning_rate": 2.6474848564656307e-05,
      "loss": 1.4854,
      "step": 17875
    },
    {
      "epoch": 0.4707927311035028,
      "grad_norm": 1.8340985774993896,
      "learning_rate": 2.6473531735580726e-05,
      "loss": 2.3747,
      "step": 17876
    },
    {
      "epoch": 0.47081906768501447,
      "grad_norm": 1.8144675493240356,
      "learning_rate": 2.647221490650514e-05,
      "loss": 2.0887,
      "step": 17877
    },
    {
      "epoch": 0.4708454042665262,
      "grad_norm": 1.6621029376983643,
      "learning_rate": 2.647089807742955e-05,
      "loss": 1.6556,
      "step": 17878
    },
    {
      "epoch": 0.4708717408480379,
      "grad_norm": 4.519250392913818,
      "learning_rate": 2.6469581248353963e-05,
      "loss": 1.6906,
      "step": 17879
    },
    {
      "epoch": 0.47089807742954964,
      "grad_norm": 2.018160820007324,
      "learning_rate": 2.646826441927838e-05,
      "loss": 1.36,
      "step": 17880
    },
    {
      "epoch": 0.4709244140110614,
      "grad_norm": 5.2710185050964355,
      "learning_rate": 2.6466947590202794e-05,
      "loss": 1.416,
      "step": 17881
    },
    {
      "epoch": 0.4709507505925731,
      "grad_norm": 3.0456268787384033,
      "learning_rate": 2.646563076112721e-05,
      "loss": 1.1083,
      "step": 17882
    },
    {
      "epoch": 0.4709770871740848,
      "grad_norm": 1.931447982788086,
      "learning_rate": 2.6464313932051622e-05,
      "loss": 2.1125,
      "step": 17883
    },
    {
      "epoch": 0.4710034237555965,
      "grad_norm": 1.7303372621536255,
      "learning_rate": 2.6462997102976034e-05,
      "loss": 0.5131,
      "step": 17884
    },
    {
      "epoch": 0.47102976033710825,
      "grad_norm": 1.66446852684021,
      "learning_rate": 2.6461680273900446e-05,
      "loss": 1.5977,
      "step": 17885
    },
    {
      "epoch": 0.47105609691861994,
      "grad_norm": 3.5797624588012695,
      "learning_rate": 2.6460363444824865e-05,
      "loss": 1.2369,
      "step": 17886
    },
    {
      "epoch": 0.4710824335001317,
      "grad_norm": 1.716167688369751,
      "learning_rate": 2.6459046615749277e-05,
      "loss": 0.5763,
      "step": 17887
    },
    {
      "epoch": 0.4711087700816434,
      "grad_norm": 1.7607096433639526,
      "learning_rate": 2.645772978667369e-05,
      "loss": 1.8462,
      "step": 17888
    },
    {
      "epoch": 0.4711351066631551,
      "grad_norm": 3.1781349182128906,
      "learning_rate": 2.6456412957598102e-05,
      "loss": 1.4719,
      "step": 17889
    },
    {
      "epoch": 0.47116144324466686,
      "grad_norm": 4.016254425048828,
      "learning_rate": 2.6455096128522517e-05,
      "loss": 0.9295,
      "step": 17890
    },
    {
      "epoch": 0.47118777982617854,
      "grad_norm": 1.9094973802566528,
      "learning_rate": 2.6453779299446933e-05,
      "loss": 1.9185,
      "step": 17891
    },
    {
      "epoch": 0.4712141164076903,
      "grad_norm": 2.1751456260681152,
      "learning_rate": 2.645246247037135e-05,
      "loss": 1.6324,
      "step": 17892
    },
    {
      "epoch": 0.471240452989202,
      "grad_norm": 1.5610445737838745,
      "learning_rate": 2.645114564129576e-05,
      "loss": 1.6727,
      "step": 17893
    },
    {
      "epoch": 0.4712667895707137,
      "grad_norm": 2.7875514030456543,
      "learning_rate": 2.6449828812220173e-05,
      "loss": 1.2816,
      "step": 17894
    },
    {
      "epoch": 0.47129312615222546,
      "grad_norm": 1.7084176540374756,
      "learning_rate": 2.6448511983144592e-05,
      "loss": 1.4615,
      "step": 17895
    },
    {
      "epoch": 0.47131946273373715,
      "grad_norm": 2.6210553646087646,
      "learning_rate": 2.6447195154069004e-05,
      "loss": 0.7052,
      "step": 17896
    },
    {
      "epoch": 0.4713457993152489,
      "grad_norm": 4.016355514526367,
      "learning_rate": 2.6445878324993416e-05,
      "loss": 1.0527,
      "step": 17897
    },
    {
      "epoch": 0.4713721358967606,
      "grad_norm": 1.9485079050064087,
      "learning_rate": 2.644456149591783e-05,
      "loss": 1.3472,
      "step": 17898
    },
    {
      "epoch": 0.4713984724782723,
      "grad_norm": 1.6094517707824707,
      "learning_rate": 2.6443244666842244e-05,
      "loss": 1.1668,
      "step": 17899
    },
    {
      "epoch": 0.47142480905978407,
      "grad_norm": 1.8517144918441772,
      "learning_rate": 2.644192783776666e-05,
      "loss": 1.6535,
      "step": 17900
    },
    {
      "epoch": 0.47145114564129575,
      "grad_norm": 2.763026714324951,
      "learning_rate": 2.6440611008691075e-05,
      "loss": 1.4338,
      "step": 17901
    },
    {
      "epoch": 0.4714774822228075,
      "grad_norm": 1.8545913696289062,
      "learning_rate": 2.6439294179615488e-05,
      "loss": 2.0696,
      "step": 17902
    },
    {
      "epoch": 0.4715038188043192,
      "grad_norm": 4.549075603485107,
      "learning_rate": 2.64379773505399e-05,
      "loss": 2.1094,
      "step": 17903
    },
    {
      "epoch": 0.47153015538583093,
      "grad_norm": 1.748066782951355,
      "learning_rate": 2.6436660521464312e-05,
      "loss": 1.3943,
      "step": 17904
    },
    {
      "epoch": 0.4715564919673426,
      "grad_norm": 5.617864608764648,
      "learning_rate": 2.643534369238873e-05,
      "loss": 2.3394,
      "step": 17905
    },
    {
      "epoch": 0.47158282854885436,
      "grad_norm": 2.0360803604125977,
      "learning_rate": 2.6434026863313143e-05,
      "loss": 1.8966,
      "step": 17906
    },
    {
      "epoch": 0.4716091651303661,
      "grad_norm": 2.6220288276672363,
      "learning_rate": 2.6432710034237555e-05,
      "loss": 2.1283,
      "step": 17907
    },
    {
      "epoch": 0.4716355017118778,
      "grad_norm": 2.480895519256592,
      "learning_rate": 2.643139320516197e-05,
      "loss": 2.0587,
      "step": 17908
    },
    {
      "epoch": 0.47166183829338953,
      "grad_norm": 2.9045658111572266,
      "learning_rate": 2.6430076376086387e-05,
      "loss": 1.8739,
      "step": 17909
    },
    {
      "epoch": 0.4716881748749012,
      "grad_norm": 1.576075553894043,
      "learning_rate": 2.6428759547010802e-05,
      "loss": 1.7105,
      "step": 17910
    },
    {
      "epoch": 0.47171451145641297,
      "grad_norm": 1.9509483575820923,
      "learning_rate": 2.6427442717935214e-05,
      "loss": 1.4462,
      "step": 17911
    },
    {
      "epoch": 0.47174084803792465,
      "grad_norm": 2.1344716548919678,
      "learning_rate": 2.6426125888859627e-05,
      "loss": 2.1516,
      "step": 17912
    },
    {
      "epoch": 0.4717671846194364,
      "grad_norm": 2.135622978210449,
      "learning_rate": 2.642480905978404e-05,
      "loss": 1.8478,
      "step": 17913
    },
    {
      "epoch": 0.47179352120094814,
      "grad_norm": 3.6220579147338867,
      "learning_rate": 2.6423492230708458e-05,
      "loss": 2.0593,
      "step": 17914
    },
    {
      "epoch": 0.47181985778245983,
      "grad_norm": 2.3813655376434326,
      "learning_rate": 2.642217540163287e-05,
      "loss": 1.8815,
      "step": 17915
    },
    {
      "epoch": 0.47184619436397157,
      "grad_norm": 1.762906551361084,
      "learning_rate": 2.6420858572557282e-05,
      "loss": 1.7766,
      "step": 17916
    },
    {
      "epoch": 0.47187253094548326,
      "grad_norm": 2.101269483566284,
      "learning_rate": 2.6419541743481694e-05,
      "loss": 1.9748,
      "step": 17917
    },
    {
      "epoch": 0.471898867526995,
      "grad_norm": 4.175267219543457,
      "learning_rate": 2.641822491440611e-05,
      "loss": 1.6209,
      "step": 17918
    },
    {
      "epoch": 0.4719252041085067,
      "grad_norm": 1.826284408569336,
      "learning_rate": 2.6416908085330526e-05,
      "loss": 1.7092,
      "step": 17919
    },
    {
      "epoch": 0.47195154069001843,
      "grad_norm": 2.0861289501190186,
      "learning_rate": 2.641559125625494e-05,
      "loss": 1.2414,
      "step": 17920
    },
    {
      "epoch": 0.4719778772715302,
      "grad_norm": 2.045217275619507,
      "learning_rate": 2.6414274427179353e-05,
      "loss": 1.5074,
      "step": 17921
    },
    {
      "epoch": 0.47200421385304187,
      "grad_norm": 4.767573356628418,
      "learning_rate": 2.6412957598103766e-05,
      "loss": 1.2399,
      "step": 17922
    },
    {
      "epoch": 0.4720305504345536,
      "grad_norm": 1.6747349500656128,
      "learning_rate": 2.6411640769028178e-05,
      "loss": 1.9176,
      "step": 17923
    },
    {
      "epoch": 0.4720568870160653,
      "grad_norm": 2.0052506923675537,
      "learning_rate": 2.6410323939952597e-05,
      "loss": 2.047,
      "step": 17924
    },
    {
      "epoch": 0.47208322359757704,
      "grad_norm": 1.6829941272735596,
      "learning_rate": 2.640900711087701e-05,
      "loss": 0.3368,
      "step": 17925
    },
    {
      "epoch": 0.4721095601790887,
      "grad_norm": 2.3700201511383057,
      "learning_rate": 2.640769028180142e-05,
      "loss": 0.8107,
      "step": 17926
    },
    {
      "epoch": 0.47213589676060047,
      "grad_norm": 1.6656286716461182,
      "learning_rate": 2.6406373452725837e-05,
      "loss": 0.4055,
      "step": 17927
    },
    {
      "epoch": 0.4721622333421122,
      "grad_norm": 1.7245970964431763,
      "learning_rate": 2.6405056623650252e-05,
      "loss": 1.5221,
      "step": 17928
    },
    {
      "epoch": 0.4721885699236239,
      "grad_norm": 2.6271071434020996,
      "learning_rate": 2.6403739794574668e-05,
      "loss": 1.541,
      "step": 17929
    },
    {
      "epoch": 0.47221490650513565,
      "grad_norm": 3.7257206439971924,
      "learning_rate": 2.640242296549908e-05,
      "loss": 0.2719,
      "step": 17930
    },
    {
      "epoch": 0.47224124308664733,
      "grad_norm": 1.7708683013916016,
      "learning_rate": 2.6401106136423493e-05,
      "loss": 1.7959,
      "step": 17931
    },
    {
      "epoch": 0.4722675796681591,
      "grad_norm": 1.5768908262252808,
      "learning_rate": 2.6399789307347905e-05,
      "loss": 0.3727,
      "step": 17932
    },
    {
      "epoch": 0.4722939162496708,
      "grad_norm": 1.5334714651107788,
      "learning_rate": 2.6398472478272324e-05,
      "loss": 2.0055,
      "step": 17933
    },
    {
      "epoch": 0.4723202528311825,
      "grad_norm": 1.5198733806610107,
      "learning_rate": 2.6397155649196736e-05,
      "loss": 1.9119,
      "step": 17934
    },
    {
      "epoch": 0.47234658941269425,
      "grad_norm": 2.3848814964294434,
      "learning_rate": 2.6395838820121148e-05,
      "loss": 1.7405,
      "step": 17935
    },
    {
      "epoch": 0.47237292599420594,
      "grad_norm": 2.7062489986419678,
      "learning_rate": 2.6394521991045564e-05,
      "loss": 0.9261,
      "step": 17936
    },
    {
      "epoch": 0.4723992625757177,
      "grad_norm": 1.8381402492523193,
      "learning_rate": 2.6393205161969976e-05,
      "loss": 2.0577,
      "step": 17937
    },
    {
      "epoch": 0.47242559915722937,
      "grad_norm": 3.229065418243408,
      "learning_rate": 2.6391888332894395e-05,
      "loss": 1.9767,
      "step": 17938
    },
    {
      "epoch": 0.4724519357387411,
      "grad_norm": 1.5846070051193237,
      "learning_rate": 2.6390571503818807e-05,
      "loss": 0.4382,
      "step": 17939
    },
    {
      "epoch": 0.47247827232025286,
      "grad_norm": 2.2891552448272705,
      "learning_rate": 2.638925467474322e-05,
      "loss": 1.0503,
      "step": 17940
    },
    {
      "epoch": 0.47250460890176454,
      "grad_norm": 1.9028269052505493,
      "learning_rate": 2.638793784566763e-05,
      "loss": 1.7442,
      "step": 17941
    },
    {
      "epoch": 0.4725309454832763,
      "grad_norm": 3.3676249980926514,
      "learning_rate": 2.638662101659205e-05,
      "loss": 1.8799,
      "step": 17942
    },
    {
      "epoch": 0.472557282064788,
      "grad_norm": 1.905550241470337,
      "learning_rate": 2.6385304187516463e-05,
      "loss": 1.6442,
      "step": 17943
    },
    {
      "epoch": 0.4725836186462997,
      "grad_norm": 1.5947978496551514,
      "learning_rate": 2.6383987358440875e-05,
      "loss": 1.7902,
      "step": 17944
    },
    {
      "epoch": 0.4726099552278114,
      "grad_norm": 2.3338849544525146,
      "learning_rate": 2.6382670529365287e-05,
      "loss": 1.7916,
      "step": 17945
    },
    {
      "epoch": 0.47263629180932315,
      "grad_norm": 3.3827695846557617,
      "learning_rate": 2.6381353700289703e-05,
      "loss": 0.7726,
      "step": 17946
    },
    {
      "epoch": 0.4726626283908349,
      "grad_norm": 2.1535520553588867,
      "learning_rate": 2.638003687121412e-05,
      "loss": 2.0259,
      "step": 17947
    },
    {
      "epoch": 0.4726889649723466,
      "grad_norm": 2.0482892990112305,
      "learning_rate": 2.6378720042138534e-05,
      "loss": 2.9368,
      "step": 17948
    },
    {
      "epoch": 0.4727153015538583,
      "grad_norm": 2.736971855163574,
      "learning_rate": 2.6377403213062946e-05,
      "loss": 0.9796,
      "step": 17949
    },
    {
      "epoch": 0.47274163813537,
      "grad_norm": 3.0548458099365234,
      "learning_rate": 2.637608638398736e-05,
      "loss": 1.293,
      "step": 17950
    },
    {
      "epoch": 0.47276797471688176,
      "grad_norm": 3.1834046840667725,
      "learning_rate": 2.637476955491177e-05,
      "loss": 1.6554,
      "step": 17951
    },
    {
      "epoch": 0.47279431129839344,
      "grad_norm": 1.7934788465499878,
      "learning_rate": 2.637345272583619e-05,
      "loss": 1.4249,
      "step": 17952
    },
    {
      "epoch": 0.4728206478799052,
      "grad_norm": 3.496013641357422,
      "learning_rate": 2.6372135896760602e-05,
      "loss": 0.8848,
      "step": 17953
    },
    {
      "epoch": 0.47284698446141693,
      "grad_norm": 1.5670204162597656,
      "learning_rate": 2.6370819067685014e-05,
      "loss": 1.9357,
      "step": 17954
    },
    {
      "epoch": 0.4728733210429286,
      "grad_norm": 1.8909703493118286,
      "learning_rate": 2.636950223860943e-05,
      "loss": 1.8578,
      "step": 17955
    },
    {
      "epoch": 0.47289965762444036,
      "grad_norm": 2.0545642375946045,
      "learning_rate": 2.6368185409533845e-05,
      "loss": 2.5095,
      "step": 17956
    },
    {
      "epoch": 0.47292599420595205,
      "grad_norm": 2.18129825592041,
      "learning_rate": 2.636686858045826e-05,
      "loss": 1.8496,
      "step": 17957
    },
    {
      "epoch": 0.4729523307874638,
      "grad_norm": 2.0868897438049316,
      "learning_rate": 2.6365551751382673e-05,
      "loss": 1.4307,
      "step": 17958
    },
    {
      "epoch": 0.4729786673689755,
      "grad_norm": 3.3818743228912354,
      "learning_rate": 2.6364234922307085e-05,
      "loss": 1.6019,
      "step": 17959
    },
    {
      "epoch": 0.4730050039504872,
      "grad_norm": 2.294980525970459,
      "learning_rate": 2.6362918093231497e-05,
      "loss": 1.5046,
      "step": 17960
    },
    {
      "epoch": 0.47303134053199897,
      "grad_norm": 3.9720730781555176,
      "learning_rate": 2.6361601264155916e-05,
      "loss": 1.7485,
      "step": 17961
    },
    {
      "epoch": 0.47305767711351066,
      "grad_norm": 1.6498215198516846,
      "learning_rate": 2.636028443508033e-05,
      "loss": 1.7652,
      "step": 17962
    },
    {
      "epoch": 0.4730840136950224,
      "grad_norm": 4.694159030914307,
      "learning_rate": 2.635896760600474e-05,
      "loss": 0.8819,
      "step": 17963
    },
    {
      "epoch": 0.4731103502765341,
      "grad_norm": 3.2613956928253174,
      "learning_rate": 2.6357650776929156e-05,
      "loss": 1.0847,
      "step": 17964
    },
    {
      "epoch": 0.47313668685804583,
      "grad_norm": 1.9620249271392822,
      "learning_rate": 2.635633394785357e-05,
      "loss": 1.7778,
      "step": 17965
    },
    {
      "epoch": 0.4731630234395576,
      "grad_norm": 1.7562206983566284,
      "learning_rate": 2.6355017118777988e-05,
      "loss": 1.9489,
      "step": 17966
    },
    {
      "epoch": 0.47318936002106926,
      "grad_norm": 2.8412766456604004,
      "learning_rate": 2.63537002897024e-05,
      "loss": 1.81,
      "step": 17967
    },
    {
      "epoch": 0.473215696602581,
      "grad_norm": 1.8545047044754028,
      "learning_rate": 2.6352383460626812e-05,
      "loss": 1.7426,
      "step": 17968
    },
    {
      "epoch": 0.4732420331840927,
      "grad_norm": 2.510364294052124,
      "learning_rate": 2.6351066631551224e-05,
      "loss": 2.2014,
      "step": 17969
    },
    {
      "epoch": 0.47326836976560444,
      "grad_norm": 1.8358944654464722,
      "learning_rate": 2.6349749802475636e-05,
      "loss": 1.4303,
      "step": 17970
    },
    {
      "epoch": 0.4732947063471161,
      "grad_norm": 1.7235865592956543,
      "learning_rate": 2.6348432973400055e-05,
      "loss": 2.365,
      "step": 17971
    },
    {
      "epoch": 0.47332104292862787,
      "grad_norm": 1.8355944156646729,
      "learning_rate": 2.6347116144324468e-05,
      "loss": 1.6013,
      "step": 17972
    },
    {
      "epoch": 0.4733473795101396,
      "grad_norm": 2.8692994117736816,
      "learning_rate": 2.634579931524888e-05,
      "loss": 1.0572,
      "step": 17973
    },
    {
      "epoch": 0.4733737160916513,
      "grad_norm": 1.9615451097488403,
      "learning_rate": 2.6344482486173295e-05,
      "loss": 1.7517,
      "step": 17974
    },
    {
      "epoch": 0.47340005267316304,
      "grad_norm": 1.7628802061080933,
      "learning_rate": 2.634316565709771e-05,
      "loss": 1.8108,
      "step": 17975
    },
    {
      "epoch": 0.47342638925467473,
      "grad_norm": 1.5804575681686401,
      "learning_rate": 2.6341848828022127e-05,
      "loss": 1.7977,
      "step": 17976
    },
    {
      "epoch": 0.47345272583618647,
      "grad_norm": 4.294353008270264,
      "learning_rate": 2.634053199894654e-05,
      "loss": 1.0865,
      "step": 17977
    },
    {
      "epoch": 0.47347906241769816,
      "grad_norm": 3.211378335952759,
      "learning_rate": 2.633921516987095e-05,
      "loss": 0.8451,
      "step": 17978
    },
    {
      "epoch": 0.4735053989992099,
      "grad_norm": 1.733656644821167,
      "learning_rate": 2.6337898340795363e-05,
      "loss": 1.308,
      "step": 17979
    },
    {
      "epoch": 0.47353173558072165,
      "grad_norm": 1.7178661823272705,
      "learning_rate": 2.6336581511719782e-05,
      "loss": 1.8142,
      "step": 17980
    },
    {
      "epoch": 0.47355807216223333,
      "grad_norm": 2.3543434143066406,
      "learning_rate": 2.6335264682644194e-05,
      "loss": 1.7021,
      "step": 17981
    },
    {
      "epoch": 0.4735844087437451,
      "grad_norm": 3.654123306274414,
      "learning_rate": 2.6333947853568607e-05,
      "loss": 2.5898,
      "step": 17982
    },
    {
      "epoch": 0.47361074532525677,
      "grad_norm": 4.701235771179199,
      "learning_rate": 2.6332631024493022e-05,
      "loss": 1.4104,
      "step": 17983
    },
    {
      "epoch": 0.4736370819067685,
      "grad_norm": 3.404085874557495,
      "learning_rate": 2.6331314195417434e-05,
      "loss": 0.8055,
      "step": 17984
    },
    {
      "epoch": 0.4736634184882802,
      "grad_norm": 1.8302545547485352,
      "learning_rate": 2.6329997366341853e-05,
      "loss": 1.7302,
      "step": 17985
    },
    {
      "epoch": 0.47368975506979194,
      "grad_norm": 1.708420753479004,
      "learning_rate": 2.6328680537266266e-05,
      "loss": 2.1536,
      "step": 17986
    },
    {
      "epoch": 0.4737160916513037,
      "grad_norm": 2.7158589363098145,
      "learning_rate": 2.6327363708190678e-05,
      "loss": 2.1039,
      "step": 17987
    },
    {
      "epoch": 0.47374242823281537,
      "grad_norm": 1.9999464750289917,
      "learning_rate": 2.632604687911509e-05,
      "loss": 0.6847,
      "step": 17988
    },
    {
      "epoch": 0.4737687648143271,
      "grad_norm": 2.6629791259765625,
      "learning_rate": 2.632473005003951e-05,
      "loss": 2.8301,
      "step": 17989
    },
    {
      "epoch": 0.4737951013958388,
      "grad_norm": 2.133427143096924,
      "learning_rate": 2.632341322096392e-05,
      "loss": 1.8183,
      "step": 17990
    },
    {
      "epoch": 0.47382143797735055,
      "grad_norm": 2.132113218307495,
      "learning_rate": 2.6322096391888334e-05,
      "loss": 0.4059,
      "step": 17991
    },
    {
      "epoch": 0.47384777455886223,
      "grad_norm": 2.1298694610595703,
      "learning_rate": 2.6320779562812746e-05,
      "loss": 1.7905,
      "step": 17992
    },
    {
      "epoch": 0.473874111140374,
      "grad_norm": 2.297606945037842,
      "learning_rate": 2.631946273373716e-05,
      "loss": 1.2471,
      "step": 17993
    },
    {
      "epoch": 0.4739004477218857,
      "grad_norm": 2.932337760925293,
      "learning_rate": 2.6318145904661577e-05,
      "loss": 0.6356,
      "step": 17994
    },
    {
      "epoch": 0.4739267843033974,
      "grad_norm": 2.2184042930603027,
      "learning_rate": 2.6316829075585993e-05,
      "loss": 1.4082,
      "step": 17995
    },
    {
      "epoch": 0.47395312088490915,
      "grad_norm": 1.7243010997772217,
      "learning_rate": 2.6315512246510405e-05,
      "loss": 1.5012,
      "step": 17996
    },
    {
      "epoch": 0.47397945746642084,
      "grad_norm": 1.9842314720153809,
      "learning_rate": 2.6314195417434817e-05,
      "loss": 1.4877,
      "step": 17997
    },
    {
      "epoch": 0.4740057940479326,
      "grad_norm": 2.626722574234009,
      "learning_rate": 2.631287858835923e-05,
      "loss": 2.0232,
      "step": 17998
    },
    {
      "epoch": 0.47403213062944427,
      "grad_norm": 2.446789264678955,
      "learning_rate": 2.6311561759283648e-05,
      "loss": 1.7035,
      "step": 17999
    },
    {
      "epoch": 0.474058467210956,
      "grad_norm": 3.0998497009277344,
      "learning_rate": 2.631024493020806e-05,
      "loss": 1.8519,
      "step": 18000
    },
    {
      "epoch": 0.47408480379246776,
      "grad_norm": 3.261470317840576,
      "learning_rate": 2.6308928101132473e-05,
      "loss": 1.7502,
      "step": 18001
    },
    {
      "epoch": 0.47411114037397944,
      "grad_norm": 3.209693431854248,
      "learning_rate": 2.6307611272056888e-05,
      "loss": 0.7119,
      "step": 18002
    },
    {
      "epoch": 0.4741374769554912,
      "grad_norm": 1.4962162971496582,
      "learning_rate": 2.63062944429813e-05,
      "loss": 1.766,
      "step": 18003
    },
    {
      "epoch": 0.4741638135370029,
      "grad_norm": 1.7053147554397583,
      "learning_rate": 2.630497761390572e-05,
      "loss": 1.629,
      "step": 18004
    },
    {
      "epoch": 0.4741901501185146,
      "grad_norm": 3.461153268814087,
      "learning_rate": 2.630366078483013e-05,
      "loss": 1.2987,
      "step": 18005
    },
    {
      "epoch": 0.47421648670002636,
      "grad_norm": 5.332024574279785,
      "learning_rate": 2.6302343955754544e-05,
      "loss": 1.4677,
      "step": 18006
    },
    {
      "epoch": 0.47424282328153805,
      "grad_norm": 3.26020884513855,
      "learning_rate": 2.6301027126678956e-05,
      "loss": 2.0604,
      "step": 18007
    },
    {
      "epoch": 0.4742691598630498,
      "grad_norm": 1.7316187620162964,
      "learning_rate": 2.6299710297603375e-05,
      "loss": 1.8331,
      "step": 18008
    },
    {
      "epoch": 0.4742954964445615,
      "grad_norm": 3.970562219619751,
      "learning_rate": 2.6298393468527787e-05,
      "loss": 1.4967,
      "step": 18009
    },
    {
      "epoch": 0.4743218330260732,
      "grad_norm": 1.3757078647613525,
      "learning_rate": 2.62970766394522e-05,
      "loss": 2.077,
      "step": 18010
    },
    {
      "epoch": 0.4743481696075849,
      "grad_norm": 2.3905718326568604,
      "learning_rate": 2.6295759810376615e-05,
      "loss": 1.4518,
      "step": 18011
    },
    {
      "epoch": 0.47437450618909666,
      "grad_norm": 4.833202362060547,
      "learning_rate": 2.6294442981301027e-05,
      "loss": 2.1774,
      "step": 18012
    },
    {
      "epoch": 0.4744008427706084,
      "grad_norm": 2.995392322540283,
      "learning_rate": 2.6293126152225446e-05,
      "loss": 1.8886,
      "step": 18013
    },
    {
      "epoch": 0.4744271793521201,
      "grad_norm": 1.6298213005065918,
      "learning_rate": 2.629180932314986e-05,
      "loss": 1.6117,
      "step": 18014
    },
    {
      "epoch": 0.47445351593363183,
      "grad_norm": 2.83526611328125,
      "learning_rate": 2.629049249407427e-05,
      "loss": 0.5354,
      "step": 18015
    },
    {
      "epoch": 0.4744798525151435,
      "grad_norm": 2.0443403720855713,
      "learning_rate": 2.6289175664998683e-05,
      "loss": 1.8038,
      "step": 18016
    },
    {
      "epoch": 0.47450618909665526,
      "grad_norm": 2.0520477294921875,
      "learning_rate": 2.6287858835923095e-05,
      "loss": 0.98,
      "step": 18017
    },
    {
      "epoch": 0.47453252567816695,
      "grad_norm": 1.6275315284729004,
      "learning_rate": 2.6286542006847514e-05,
      "loss": 1.947,
      "step": 18018
    },
    {
      "epoch": 0.4745588622596787,
      "grad_norm": 1.6136394739151,
      "learning_rate": 2.6285225177771926e-05,
      "loss": 1.8218,
      "step": 18019
    },
    {
      "epoch": 0.47458519884119044,
      "grad_norm": 3.588017225265503,
      "learning_rate": 2.628390834869634e-05,
      "loss": 1.897,
      "step": 18020
    },
    {
      "epoch": 0.4746115354227021,
      "grad_norm": 1.9499776363372803,
      "learning_rate": 2.6282591519620754e-05,
      "loss": 2.0482,
      "step": 18021
    },
    {
      "epoch": 0.47463787200421387,
      "grad_norm": 1.7800002098083496,
      "learning_rate": 2.628127469054517e-05,
      "loss": 1.8747,
      "step": 18022
    },
    {
      "epoch": 0.47466420858572556,
      "grad_norm": 4.833249568939209,
      "learning_rate": 2.6279957861469585e-05,
      "loss": 1.8213,
      "step": 18023
    },
    {
      "epoch": 0.4746905451672373,
      "grad_norm": 2.453986883163452,
      "learning_rate": 2.6278641032393997e-05,
      "loss": 1.0012,
      "step": 18024
    },
    {
      "epoch": 0.474716881748749,
      "grad_norm": 2.8022005558013916,
      "learning_rate": 2.627732420331841e-05,
      "loss": 0.8829,
      "step": 18025
    },
    {
      "epoch": 0.47474321833026073,
      "grad_norm": 1.5881845951080322,
      "learning_rate": 2.6276007374242822e-05,
      "loss": 1.8127,
      "step": 18026
    },
    {
      "epoch": 0.4747695549117725,
      "grad_norm": 2.413527011871338,
      "learning_rate": 2.627469054516724e-05,
      "loss": 2.3022,
      "step": 18027
    },
    {
      "epoch": 0.47479589149328416,
      "grad_norm": 1.7777903079986572,
      "learning_rate": 2.6273373716091653e-05,
      "loss": 0.6391,
      "step": 18028
    },
    {
      "epoch": 0.4748222280747959,
      "grad_norm": 2.0796475410461426,
      "learning_rate": 2.6272056887016065e-05,
      "loss": 1.2444,
      "step": 18029
    },
    {
      "epoch": 0.4748485646563076,
      "grad_norm": 1.6656315326690674,
      "learning_rate": 2.627074005794048e-05,
      "loss": 1.5637,
      "step": 18030
    },
    {
      "epoch": 0.47487490123781934,
      "grad_norm": 1.4762217998504639,
      "learning_rate": 2.6269423228864893e-05,
      "loss": 1.8589,
      "step": 18031
    },
    {
      "epoch": 0.474901237819331,
      "grad_norm": 1.888584852218628,
      "learning_rate": 2.6268106399789312e-05,
      "loss": 2.2992,
      "step": 18032
    },
    {
      "epoch": 0.47492757440084277,
      "grad_norm": 3.5453085899353027,
      "learning_rate": 2.6266789570713724e-05,
      "loss": 1.7032,
      "step": 18033
    },
    {
      "epoch": 0.4749539109823545,
      "grad_norm": 2.0390994548797607,
      "learning_rate": 2.6265472741638136e-05,
      "loss": 1.3441,
      "step": 18034
    },
    {
      "epoch": 0.4749802475638662,
      "grad_norm": 1.6482908725738525,
      "learning_rate": 2.626415591256255e-05,
      "loss": 0.3402,
      "step": 18035
    },
    {
      "epoch": 0.47500658414537794,
      "grad_norm": 3.6160387992858887,
      "learning_rate": 2.626283908348696e-05,
      "loss": 0.8216,
      "step": 18036
    },
    {
      "epoch": 0.47503292072688963,
      "grad_norm": 1.7986429929733276,
      "learning_rate": 2.626152225441138e-05,
      "loss": 1.7053,
      "step": 18037
    },
    {
      "epoch": 0.4750592573084014,
      "grad_norm": 1.9232220649719238,
      "learning_rate": 2.6260205425335792e-05,
      "loss": 2.0929,
      "step": 18038
    },
    {
      "epoch": 0.4750855938899131,
      "grad_norm": 2.988654851913452,
      "learning_rate": 2.6258888596260208e-05,
      "loss": 1.2714,
      "step": 18039
    },
    {
      "epoch": 0.4751119304714248,
      "grad_norm": 1.9279531240463257,
      "learning_rate": 2.625757176718462e-05,
      "loss": 1.3988,
      "step": 18040
    },
    {
      "epoch": 0.47513826705293655,
      "grad_norm": 3.3459930419921875,
      "learning_rate": 2.625625493810904e-05,
      "loss": 1.2646,
      "step": 18041
    },
    {
      "epoch": 0.47516460363444823,
      "grad_norm": 4.728421688079834,
      "learning_rate": 2.625493810903345e-05,
      "loss": 0.8438,
      "step": 18042
    },
    {
      "epoch": 0.47519094021596,
      "grad_norm": 2.0103824138641357,
      "learning_rate": 2.6253621279957863e-05,
      "loss": 1.8723,
      "step": 18043
    },
    {
      "epoch": 0.47521727679747167,
      "grad_norm": 2.583697557449341,
      "learning_rate": 2.6252304450882275e-05,
      "loss": 2.0797,
      "step": 18044
    },
    {
      "epoch": 0.4752436133789834,
      "grad_norm": 6.7954559326171875,
      "learning_rate": 2.6250987621806688e-05,
      "loss": 2.6605,
      "step": 18045
    },
    {
      "epoch": 0.47526994996049515,
      "grad_norm": 3.5881361961364746,
      "learning_rate": 2.6249670792731107e-05,
      "loss": 2.1259,
      "step": 18046
    },
    {
      "epoch": 0.47529628654200684,
      "grad_norm": 2.464256525039673,
      "learning_rate": 2.624835396365552e-05,
      "loss": 0.9051,
      "step": 18047
    },
    {
      "epoch": 0.4753226231235186,
      "grad_norm": 7.696624755859375,
      "learning_rate": 2.624703713457993e-05,
      "loss": 2.0595,
      "step": 18048
    },
    {
      "epoch": 0.47534895970503027,
      "grad_norm": 5.125574588775635,
      "learning_rate": 2.6245720305504347e-05,
      "loss": 0.432,
      "step": 18049
    },
    {
      "epoch": 0.475375296286542,
      "grad_norm": 2.2241811752319336,
      "learning_rate": 2.624440347642876e-05,
      "loss": 1.2333,
      "step": 18050
    },
    {
      "epoch": 0.4754016328680537,
      "grad_norm": 2.0132031440734863,
      "learning_rate": 2.6243086647353178e-05,
      "loss": 1.7055,
      "step": 18051
    },
    {
      "epoch": 0.47542796944956545,
      "grad_norm": 1.9278985261917114,
      "learning_rate": 2.624176981827759e-05,
      "loss": 1.8635,
      "step": 18052
    },
    {
      "epoch": 0.4754543060310772,
      "grad_norm": 2.030756711959839,
      "learning_rate": 2.6240452989202002e-05,
      "loss": 0.4427,
      "step": 18053
    },
    {
      "epoch": 0.4754806426125889,
      "grad_norm": 1.6091283559799194,
      "learning_rate": 2.6239136160126415e-05,
      "loss": 1.5705,
      "step": 18054
    },
    {
      "epoch": 0.4755069791941006,
      "grad_norm": 2.5038156509399414,
      "learning_rate": 2.6237819331050834e-05,
      "loss": 0.2297,
      "step": 18055
    },
    {
      "epoch": 0.4755333157756123,
      "grad_norm": 2.3296658992767334,
      "learning_rate": 2.6236502501975246e-05,
      "loss": 1.9659,
      "step": 18056
    },
    {
      "epoch": 0.47555965235712405,
      "grad_norm": 2.38259220123291,
      "learning_rate": 2.6235185672899658e-05,
      "loss": 1.9652,
      "step": 18057
    },
    {
      "epoch": 0.47558598893863574,
      "grad_norm": 1.7944610118865967,
      "learning_rate": 2.6233868843824074e-05,
      "loss": 0.7851,
      "step": 18058
    },
    {
      "epoch": 0.4756123255201475,
      "grad_norm": 1.4154953956604004,
      "learning_rate": 2.6232552014748486e-05,
      "loss": 1.5766,
      "step": 18059
    },
    {
      "epoch": 0.4756386621016592,
      "grad_norm": 1.9560688734054565,
      "learning_rate": 2.6231235185672905e-05,
      "loss": 1.9532,
      "step": 18060
    },
    {
      "epoch": 0.4756649986831709,
      "grad_norm": 1.9043681621551514,
      "learning_rate": 2.6229918356597317e-05,
      "loss": 1.5029,
      "step": 18061
    },
    {
      "epoch": 0.47569133526468266,
      "grad_norm": 2.039125680923462,
      "learning_rate": 2.622860152752173e-05,
      "loss": 1.3208,
      "step": 18062
    },
    {
      "epoch": 0.47571767184619435,
      "grad_norm": 1.4919660091400146,
      "learning_rate": 2.622728469844614e-05,
      "loss": 1.4258,
      "step": 18063
    },
    {
      "epoch": 0.4757440084277061,
      "grad_norm": 2.184338331222534,
      "learning_rate": 2.6225967869370554e-05,
      "loss": 1.5314,
      "step": 18064
    },
    {
      "epoch": 0.4757703450092178,
      "grad_norm": 3.4698526859283447,
      "learning_rate": 2.6224651040294973e-05,
      "loss": 1.8335,
      "step": 18065
    },
    {
      "epoch": 0.4757966815907295,
      "grad_norm": 2.0690839290618896,
      "learning_rate": 2.6223334211219385e-05,
      "loss": 1.3201,
      "step": 18066
    },
    {
      "epoch": 0.47582301817224126,
      "grad_norm": 1.9289405345916748,
      "learning_rate": 2.62220173821438e-05,
      "loss": 1.8343,
      "step": 18067
    },
    {
      "epoch": 0.47584935475375295,
      "grad_norm": 1.8096168041229248,
      "learning_rate": 2.6220700553068213e-05,
      "loss": 0.6406,
      "step": 18068
    },
    {
      "epoch": 0.4758756913352647,
      "grad_norm": 1.51333749294281,
      "learning_rate": 2.6219383723992625e-05,
      "loss": 2.2432,
      "step": 18069
    },
    {
      "epoch": 0.4759020279167764,
      "grad_norm": 1.675201416015625,
      "learning_rate": 2.6218066894917044e-05,
      "loss": 1.9982,
      "step": 18070
    },
    {
      "epoch": 0.4759283644982881,
      "grad_norm": 3.723294734954834,
      "learning_rate": 2.6216750065841456e-05,
      "loss": 1.6848,
      "step": 18071
    },
    {
      "epoch": 0.47595470107979987,
      "grad_norm": 1.902472734451294,
      "learning_rate": 2.6215433236765868e-05,
      "loss": 1.6854,
      "step": 18072
    },
    {
      "epoch": 0.47598103766131156,
      "grad_norm": 2.0708296298980713,
      "learning_rate": 2.621411640769028e-05,
      "loss": 1.6196,
      "step": 18073
    },
    {
      "epoch": 0.4760073742428233,
      "grad_norm": 2.0596871376037598,
      "learning_rate": 2.62127995786147e-05,
      "loss": 0.9181,
      "step": 18074
    },
    {
      "epoch": 0.476033710824335,
      "grad_norm": 1.582932949066162,
      "learning_rate": 2.621148274953911e-05,
      "loss": 1.9876,
      "step": 18075
    },
    {
      "epoch": 0.47606004740584673,
      "grad_norm": 2.512171983718872,
      "learning_rate": 2.6210165920463524e-05,
      "loss": 2.3788,
      "step": 18076
    },
    {
      "epoch": 0.4760863839873584,
      "grad_norm": 2.1394431591033936,
      "learning_rate": 2.620884909138794e-05,
      "loss": 0.6411,
      "step": 18077
    },
    {
      "epoch": 0.47611272056887016,
      "grad_norm": 1.6871570348739624,
      "learning_rate": 2.620753226231235e-05,
      "loss": 1.7547,
      "step": 18078
    },
    {
      "epoch": 0.4761390571503819,
      "grad_norm": 3.2810633182525635,
      "learning_rate": 2.620621543323677e-05,
      "loss": 1.3224,
      "step": 18079
    },
    {
      "epoch": 0.4761653937318936,
      "grad_norm": 2.3216912746429443,
      "learning_rate": 2.6204898604161183e-05,
      "loss": 0.2762,
      "step": 18080
    },
    {
      "epoch": 0.47619173031340534,
      "grad_norm": 2.0798046588897705,
      "learning_rate": 2.6203581775085595e-05,
      "loss": 1.8598,
      "step": 18081
    },
    {
      "epoch": 0.476218066894917,
      "grad_norm": 2.1319127082824707,
      "learning_rate": 2.6202264946010007e-05,
      "loss": 1.3377,
      "step": 18082
    },
    {
      "epoch": 0.47624440347642877,
      "grad_norm": 2.078587055206299,
      "learning_rate": 2.620094811693442e-05,
      "loss": 1.2751,
      "step": 18083
    },
    {
      "epoch": 0.47627074005794046,
      "grad_norm": 2.120863914489746,
      "learning_rate": 2.619963128785884e-05,
      "loss": 0.5739,
      "step": 18084
    },
    {
      "epoch": 0.4762970766394522,
      "grad_norm": 1.5160518884658813,
      "learning_rate": 2.619831445878325e-05,
      "loss": 1.5942,
      "step": 18085
    },
    {
      "epoch": 0.47632341322096394,
      "grad_norm": 2.545600175857544,
      "learning_rate": 2.6196997629707666e-05,
      "loss": 1.5817,
      "step": 18086
    },
    {
      "epoch": 0.47634974980247563,
      "grad_norm": 2.5309865474700928,
      "learning_rate": 2.619568080063208e-05,
      "loss": 1.8868,
      "step": 18087
    },
    {
      "epoch": 0.4763760863839874,
      "grad_norm": 3.169647216796875,
      "learning_rate": 2.6194363971556497e-05,
      "loss": 1.0168,
      "step": 18088
    },
    {
      "epoch": 0.47640242296549906,
      "grad_norm": 1.8885867595672607,
      "learning_rate": 2.619304714248091e-05,
      "loss": 1.8344,
      "step": 18089
    },
    {
      "epoch": 0.4764287595470108,
      "grad_norm": 1.8588078022003174,
      "learning_rate": 2.6191730313405322e-05,
      "loss": 1.9987,
      "step": 18090
    },
    {
      "epoch": 0.4764550961285225,
      "grad_norm": 2.69921875,
      "learning_rate": 2.6190413484329734e-05,
      "loss": 1.5042,
      "step": 18091
    },
    {
      "epoch": 0.47648143271003424,
      "grad_norm": 1.9179502725601196,
      "learning_rate": 2.6189096655254146e-05,
      "loss": 2.1202,
      "step": 18092
    },
    {
      "epoch": 0.476507769291546,
      "grad_norm": 1.971161127090454,
      "learning_rate": 2.6187779826178565e-05,
      "loss": 0.6615,
      "step": 18093
    },
    {
      "epoch": 0.47653410587305767,
      "grad_norm": 4.208078384399414,
      "learning_rate": 2.6186462997102977e-05,
      "loss": 1.2328,
      "step": 18094
    },
    {
      "epoch": 0.4765604424545694,
      "grad_norm": 2.099757432937622,
      "learning_rate": 2.618514616802739e-05,
      "loss": 2.2761,
      "step": 18095
    },
    {
      "epoch": 0.4765867790360811,
      "grad_norm": 1.5804299116134644,
      "learning_rate": 2.6183829338951805e-05,
      "loss": 1.5915,
      "step": 18096
    },
    {
      "epoch": 0.47661311561759284,
      "grad_norm": 2.426229953765869,
      "learning_rate": 2.6182512509876217e-05,
      "loss": 0.1391,
      "step": 18097
    },
    {
      "epoch": 0.47663945219910453,
      "grad_norm": 3.338651418685913,
      "learning_rate": 2.6181195680800636e-05,
      "loss": 0.873,
      "step": 18098
    },
    {
      "epoch": 0.4766657887806163,
      "grad_norm": 3.4374072551727295,
      "learning_rate": 2.617987885172505e-05,
      "loss": 1.6158,
      "step": 18099
    },
    {
      "epoch": 0.476692125362128,
      "grad_norm": 1.865473985671997,
      "learning_rate": 2.617856202264946e-05,
      "loss": 2.1476,
      "step": 18100
    },
    {
      "epoch": 0.4767184619436397,
      "grad_norm": 5.584752559661865,
      "learning_rate": 2.6177245193573873e-05,
      "loss": 0.7848,
      "step": 18101
    },
    {
      "epoch": 0.47674479852515145,
      "grad_norm": 2.1740102767944336,
      "learning_rate": 2.6175928364498285e-05,
      "loss": 1.6618,
      "step": 18102
    },
    {
      "epoch": 0.47677113510666314,
      "grad_norm": 1.813247561454773,
      "learning_rate": 2.6174611535422704e-05,
      "loss": 2.3947,
      "step": 18103
    },
    {
      "epoch": 0.4767974716881749,
      "grad_norm": 2.389674663543701,
      "learning_rate": 2.6173294706347116e-05,
      "loss": 0.7996,
      "step": 18104
    },
    {
      "epoch": 0.4768238082696866,
      "grad_norm": 2.842863082885742,
      "learning_rate": 2.6171977877271532e-05,
      "loss": 1.9356,
      "step": 18105
    },
    {
      "epoch": 0.4768501448511983,
      "grad_norm": 2.1585025787353516,
      "learning_rate": 2.6170661048195944e-05,
      "loss": 2.4486,
      "step": 18106
    },
    {
      "epoch": 0.47687648143271005,
      "grad_norm": 2.1451637744903564,
      "learning_rate": 2.6169344219120363e-05,
      "loss": 2.1151,
      "step": 18107
    },
    {
      "epoch": 0.47690281801422174,
      "grad_norm": 5.163798809051514,
      "learning_rate": 2.6168027390044775e-05,
      "loss": 0.9405,
      "step": 18108
    },
    {
      "epoch": 0.4769291545957335,
      "grad_norm": 1.7317043542861938,
      "learning_rate": 2.6166710560969188e-05,
      "loss": 2.1485,
      "step": 18109
    },
    {
      "epoch": 0.47695549117724517,
      "grad_norm": 1.8907835483551025,
      "learning_rate": 2.61653937318936e-05,
      "loss": 1.6107,
      "step": 18110
    },
    {
      "epoch": 0.4769818277587569,
      "grad_norm": 3.3649308681488037,
      "learning_rate": 2.6164076902818012e-05,
      "loss": 1.0804,
      "step": 18111
    },
    {
      "epoch": 0.47700816434026866,
      "grad_norm": 4.079046726226807,
      "learning_rate": 2.616276007374243e-05,
      "loss": 1.4524,
      "step": 18112
    },
    {
      "epoch": 0.47703450092178035,
      "grad_norm": 1.972220540046692,
      "learning_rate": 2.6161443244666843e-05,
      "loss": 2.1815,
      "step": 18113
    },
    {
      "epoch": 0.4770608375032921,
      "grad_norm": 1.8180415630340576,
      "learning_rate": 2.616012641559126e-05,
      "loss": 2.4495,
      "step": 18114
    },
    {
      "epoch": 0.4770871740848038,
      "grad_norm": 1.9662320613861084,
      "learning_rate": 2.615880958651567e-05,
      "loss": 1.647,
      "step": 18115
    },
    {
      "epoch": 0.4771135106663155,
      "grad_norm": 2.095525026321411,
      "learning_rate": 2.6157492757440083e-05,
      "loss": 1.5167,
      "step": 18116
    },
    {
      "epoch": 0.4771398472478272,
      "grad_norm": 2.066540479660034,
      "learning_rate": 2.6156175928364502e-05,
      "loss": 1.2558,
      "step": 18117
    },
    {
      "epoch": 0.47716618382933895,
      "grad_norm": 3.7202439308166504,
      "learning_rate": 2.6154859099288915e-05,
      "loss": 1.4575,
      "step": 18118
    },
    {
      "epoch": 0.4771925204108507,
      "grad_norm": 1.8115872144699097,
      "learning_rate": 2.6153542270213327e-05,
      "loss": 2.5709,
      "step": 18119
    },
    {
      "epoch": 0.4772188569923624,
      "grad_norm": 2.043302536010742,
      "learning_rate": 2.615222544113774e-05,
      "loss": 1.1423,
      "step": 18120
    },
    {
      "epoch": 0.4772451935738741,
      "grad_norm": 2.0807416439056396,
      "learning_rate": 2.6150908612062158e-05,
      "loss": 1.6837,
      "step": 18121
    },
    {
      "epoch": 0.4772715301553858,
      "grad_norm": 3.1593334674835205,
      "learning_rate": 2.614959178298657e-05,
      "loss": 1.8308,
      "step": 18122
    },
    {
      "epoch": 0.47729786673689756,
      "grad_norm": 3.4992871284484863,
      "learning_rate": 2.6148274953910982e-05,
      "loss": 2.1369,
      "step": 18123
    },
    {
      "epoch": 0.47732420331840925,
      "grad_norm": 1.5538591146469116,
      "learning_rate": 2.6146958124835398e-05,
      "loss": 2.5318,
      "step": 18124
    },
    {
      "epoch": 0.477350539899921,
      "grad_norm": 1.703216552734375,
      "learning_rate": 2.614564129575981e-05,
      "loss": 2.0921,
      "step": 18125
    },
    {
      "epoch": 0.47737687648143273,
      "grad_norm": 2.0972647666931152,
      "learning_rate": 2.614432446668423e-05,
      "loss": 2.5599,
      "step": 18126
    },
    {
      "epoch": 0.4774032130629444,
      "grad_norm": 4.844968795776367,
      "learning_rate": 2.614300763760864e-05,
      "loss": 1.8993,
      "step": 18127
    },
    {
      "epoch": 0.47742954964445616,
      "grad_norm": 1.9064234495162964,
      "learning_rate": 2.6141690808533054e-05,
      "loss": 1.6282,
      "step": 18128
    },
    {
      "epoch": 0.47745588622596785,
      "grad_norm": 3.336031198501587,
      "learning_rate": 2.6140373979457466e-05,
      "loss": 1.6488,
      "step": 18129
    },
    {
      "epoch": 0.4774822228074796,
      "grad_norm": 1.554504632949829,
      "learning_rate": 2.6139057150381878e-05,
      "loss": 0.3676,
      "step": 18130
    },
    {
      "epoch": 0.4775085593889913,
      "grad_norm": 1.4876564741134644,
      "learning_rate": 2.6137740321306297e-05,
      "loss": 1.8622,
      "step": 18131
    },
    {
      "epoch": 0.477534895970503,
      "grad_norm": 3.5113332271575928,
      "learning_rate": 2.613642349223071e-05,
      "loss": 1.1884,
      "step": 18132
    },
    {
      "epoch": 0.47756123255201477,
      "grad_norm": 2.460200786590576,
      "learning_rate": 2.6135106663155125e-05,
      "loss": 0.6754,
      "step": 18133
    },
    {
      "epoch": 0.47758756913352646,
      "grad_norm": 2.1989099979400635,
      "learning_rate": 2.6133789834079537e-05,
      "loss": 2.3714,
      "step": 18134
    },
    {
      "epoch": 0.4776139057150382,
      "grad_norm": 2.667492389678955,
      "learning_rate": 2.613247300500395e-05,
      "loss": 0.6263,
      "step": 18135
    },
    {
      "epoch": 0.4776402422965499,
      "grad_norm": 2.299617052078247,
      "learning_rate": 2.6131156175928368e-05,
      "loss": 2.8219,
      "step": 18136
    },
    {
      "epoch": 0.47766657887806163,
      "grad_norm": 1.612807035446167,
      "learning_rate": 2.612983934685278e-05,
      "loss": 0.2344,
      "step": 18137
    },
    {
      "epoch": 0.4776929154595734,
      "grad_norm": 1.7575221061706543,
      "learning_rate": 2.6128522517777193e-05,
      "loss": 1.6634,
      "step": 18138
    },
    {
      "epoch": 0.47771925204108506,
      "grad_norm": 4.1696457862854,
      "learning_rate": 2.6127205688701605e-05,
      "loss": 0.5157,
      "step": 18139
    },
    {
      "epoch": 0.4777455886225968,
      "grad_norm": 1.7531259059906006,
      "learning_rate": 2.6125888859626024e-05,
      "loss": 2.1103,
      "step": 18140
    },
    {
      "epoch": 0.4777719252041085,
      "grad_norm": 1.9951260089874268,
      "learning_rate": 2.6124572030550436e-05,
      "loss": 1.7099,
      "step": 18141
    },
    {
      "epoch": 0.47779826178562024,
      "grad_norm": 2.5088744163513184,
      "learning_rate": 2.612325520147485e-05,
      "loss": 1.277,
      "step": 18142
    },
    {
      "epoch": 0.4778245983671319,
      "grad_norm": 1.8694902658462524,
      "learning_rate": 2.6121938372399264e-05,
      "loss": 1.7016,
      "step": 18143
    },
    {
      "epoch": 0.47785093494864367,
      "grad_norm": 2.1027448177337646,
      "learning_rate": 2.6120621543323676e-05,
      "loss": 2.3466,
      "step": 18144
    },
    {
      "epoch": 0.4778772715301554,
      "grad_norm": 5.901076793670654,
      "learning_rate": 2.6119304714248095e-05,
      "loss": 1.5969,
      "step": 18145
    },
    {
      "epoch": 0.4779036081116671,
      "grad_norm": 1.9883469343185425,
      "learning_rate": 2.6117987885172507e-05,
      "loss": 1.7854,
      "step": 18146
    },
    {
      "epoch": 0.47792994469317884,
      "grad_norm": 2.4258086681365967,
      "learning_rate": 2.611667105609692e-05,
      "loss": 2.0893,
      "step": 18147
    },
    {
      "epoch": 0.47795628127469053,
      "grad_norm": 1.5321277379989624,
      "learning_rate": 2.611535422702133e-05,
      "loss": 2.0222,
      "step": 18148
    },
    {
      "epoch": 0.4779826178562023,
      "grad_norm": 2.2107131481170654,
      "learning_rate": 2.6114037397945744e-05,
      "loss": 0.276,
      "step": 18149
    },
    {
      "epoch": 0.47800895443771396,
      "grad_norm": 2.2959635257720947,
      "learning_rate": 2.6112720568870163e-05,
      "loss": 2.1733,
      "step": 18150
    },
    {
      "epoch": 0.4780352910192257,
      "grad_norm": 2.7273948192596436,
      "learning_rate": 2.6111403739794575e-05,
      "loss": 2.3528,
      "step": 18151
    },
    {
      "epoch": 0.47806162760073745,
      "grad_norm": 1.675957441329956,
      "learning_rate": 2.611008691071899e-05,
      "loss": 1.7183,
      "step": 18152
    },
    {
      "epoch": 0.47808796418224914,
      "grad_norm": 2.2190258502960205,
      "learning_rate": 2.6108770081643403e-05,
      "loss": 1.826,
      "step": 18153
    },
    {
      "epoch": 0.4781143007637609,
      "grad_norm": 3.512843370437622,
      "learning_rate": 2.6107453252567822e-05,
      "loss": 2.0663,
      "step": 18154
    },
    {
      "epoch": 0.47814063734527257,
      "grad_norm": 1.5189660787582397,
      "learning_rate": 2.6106136423492234e-05,
      "loss": 2.2426,
      "step": 18155
    },
    {
      "epoch": 0.4781669739267843,
      "grad_norm": 5.404087543487549,
      "learning_rate": 2.6104819594416646e-05,
      "loss": 2.1597,
      "step": 18156
    },
    {
      "epoch": 0.478193310508296,
      "grad_norm": 2.591707229614258,
      "learning_rate": 2.610350276534106e-05,
      "loss": 2.4166,
      "step": 18157
    },
    {
      "epoch": 0.47821964708980774,
      "grad_norm": 4.782519817352295,
      "learning_rate": 2.610218593626547e-05,
      "loss": 1.4625,
      "step": 18158
    },
    {
      "epoch": 0.4782459836713195,
      "grad_norm": 2.67582106590271,
      "learning_rate": 2.610086910718989e-05,
      "loss": 1.7045,
      "step": 18159
    },
    {
      "epoch": 0.4782723202528312,
      "grad_norm": 1.6655147075653076,
      "learning_rate": 2.6099552278114302e-05,
      "loss": 1.421,
      "step": 18160
    },
    {
      "epoch": 0.4782986568343429,
      "grad_norm": 3.701991081237793,
      "learning_rate": 2.6098235449038717e-05,
      "loss": 1.9044,
      "step": 18161
    },
    {
      "epoch": 0.4783249934158546,
      "grad_norm": 1.59872567653656,
      "learning_rate": 2.609691861996313e-05,
      "loss": 2.3976,
      "step": 18162
    },
    {
      "epoch": 0.47835132999736635,
      "grad_norm": 1.5172570943832397,
      "learning_rate": 2.6095601790887542e-05,
      "loss": 1.5677,
      "step": 18163
    },
    {
      "epoch": 0.47837766657887804,
      "grad_norm": 2.137535333633423,
      "learning_rate": 2.609428496181196e-05,
      "loss": 1.8379,
      "step": 18164
    },
    {
      "epoch": 0.4784040031603898,
      "grad_norm": 5.482143402099609,
      "learning_rate": 2.6092968132736373e-05,
      "loss": 1.0263,
      "step": 18165
    },
    {
      "epoch": 0.4784303397419015,
      "grad_norm": 1.9915279150009155,
      "learning_rate": 2.6091651303660785e-05,
      "loss": 1.9052,
      "step": 18166
    },
    {
      "epoch": 0.4784566763234132,
      "grad_norm": 2.3375186920166016,
      "learning_rate": 2.6090334474585197e-05,
      "loss": 0.7127,
      "step": 18167
    },
    {
      "epoch": 0.47848301290492495,
      "grad_norm": 1.8854601383209229,
      "learning_rate": 2.608901764550961e-05,
      "loss": 2.0186,
      "step": 18168
    },
    {
      "epoch": 0.47850934948643664,
      "grad_norm": 1.6205345392227173,
      "learning_rate": 2.608770081643403e-05,
      "loss": 2.5091,
      "step": 18169
    },
    {
      "epoch": 0.4785356860679484,
      "grad_norm": 1.597391128540039,
      "learning_rate": 2.608638398735844e-05,
      "loss": 1.7359,
      "step": 18170
    },
    {
      "epoch": 0.4785620226494601,
      "grad_norm": 3.3700568675994873,
      "learning_rate": 2.6085067158282856e-05,
      "loss": 1.5659,
      "step": 18171
    },
    {
      "epoch": 0.4785883592309718,
      "grad_norm": 1.645609974861145,
      "learning_rate": 2.608375032920727e-05,
      "loss": 1.4701,
      "step": 18172
    },
    {
      "epoch": 0.47861469581248356,
      "grad_norm": 2.141824245452881,
      "learning_rate": 2.6082433500131688e-05,
      "loss": 2.4564,
      "step": 18173
    },
    {
      "epoch": 0.47864103239399525,
      "grad_norm": 1.818982481956482,
      "learning_rate": 2.60811166710561e-05,
      "loss": 1.02,
      "step": 18174
    },
    {
      "epoch": 0.478667368975507,
      "grad_norm": 1.7090140581130981,
      "learning_rate": 2.6079799841980512e-05,
      "loss": 1.7205,
      "step": 18175
    },
    {
      "epoch": 0.4786937055570187,
      "grad_norm": 2.0209977626800537,
      "learning_rate": 2.6078483012904924e-05,
      "loss": 1.3472,
      "step": 18176
    },
    {
      "epoch": 0.4787200421385304,
      "grad_norm": 1.909293532371521,
      "learning_rate": 2.6077166183829337e-05,
      "loss": 1.2932,
      "step": 18177
    },
    {
      "epoch": 0.47874637872004216,
      "grad_norm": 2.426377773284912,
      "learning_rate": 2.6075849354753756e-05,
      "loss": 1.3127,
      "step": 18178
    },
    {
      "epoch": 0.47877271530155385,
      "grad_norm": 5.718318939208984,
      "learning_rate": 2.6074532525678168e-05,
      "loss": 2.0245,
      "step": 18179
    },
    {
      "epoch": 0.4787990518830656,
      "grad_norm": 3.2824246883392334,
      "learning_rate": 2.6073215696602583e-05,
      "loss": 1.6523,
      "step": 18180
    },
    {
      "epoch": 0.4788253884645773,
      "grad_norm": 2.88808274269104,
      "learning_rate": 2.6071898867526996e-05,
      "loss": 1.4564,
      "step": 18181
    },
    {
      "epoch": 0.478851725046089,
      "grad_norm": 1.67848539352417,
      "learning_rate": 2.6070582038451408e-05,
      "loss": 1.81,
      "step": 18182
    },
    {
      "epoch": 0.4788780616276007,
      "grad_norm": 6.433162689208984,
      "learning_rate": 2.6069265209375827e-05,
      "loss": 1.011,
      "step": 18183
    },
    {
      "epoch": 0.47890439820911246,
      "grad_norm": 2.4767961502075195,
      "learning_rate": 2.606794838030024e-05,
      "loss": 1.3045,
      "step": 18184
    },
    {
      "epoch": 0.4789307347906242,
      "grad_norm": 5.748504638671875,
      "learning_rate": 2.606663155122465e-05,
      "loss": 1.2878,
      "step": 18185
    },
    {
      "epoch": 0.4789570713721359,
      "grad_norm": 1.5263091325759888,
      "learning_rate": 2.6065314722149063e-05,
      "loss": 1.87,
      "step": 18186
    },
    {
      "epoch": 0.47898340795364763,
      "grad_norm": 1.9353070259094238,
      "learning_rate": 2.6063997893073482e-05,
      "loss": 2.8552,
      "step": 18187
    },
    {
      "epoch": 0.4790097445351593,
      "grad_norm": 1.4300810098648071,
      "learning_rate": 2.6062681063997895e-05,
      "loss": 2.3514,
      "step": 18188
    },
    {
      "epoch": 0.47903608111667106,
      "grad_norm": 1.5277667045593262,
      "learning_rate": 2.606136423492231e-05,
      "loss": 1.7833,
      "step": 18189
    },
    {
      "epoch": 0.47906241769818275,
      "grad_norm": 1.8952159881591797,
      "learning_rate": 2.6060047405846722e-05,
      "loss": 1.7228,
      "step": 18190
    },
    {
      "epoch": 0.4790887542796945,
      "grad_norm": 1.6774052381515503,
      "learning_rate": 2.6058730576771135e-05,
      "loss": 1.7671,
      "step": 18191
    },
    {
      "epoch": 0.47911509086120624,
      "grad_norm": 2.0082924365997314,
      "learning_rate": 2.6057413747695554e-05,
      "loss": 1.6773,
      "step": 18192
    },
    {
      "epoch": 0.4791414274427179,
      "grad_norm": 4.2807297706604,
      "learning_rate": 2.6056096918619966e-05,
      "loss": 0.799,
      "step": 18193
    },
    {
      "epoch": 0.47916776402422967,
      "grad_norm": 2.550044059753418,
      "learning_rate": 2.6054780089544378e-05,
      "loss": 2.0308,
      "step": 18194
    },
    {
      "epoch": 0.47919410060574136,
      "grad_norm": 1.816925287246704,
      "learning_rate": 2.605346326046879e-05,
      "loss": 1.781,
      "step": 18195
    },
    {
      "epoch": 0.4792204371872531,
      "grad_norm": 1.9360864162445068,
      "learning_rate": 2.6052146431393202e-05,
      "loss": 2.1195,
      "step": 18196
    },
    {
      "epoch": 0.4792467737687648,
      "grad_norm": 2.2189080715179443,
      "learning_rate": 2.605082960231762e-05,
      "loss": 1.9188,
      "step": 18197
    },
    {
      "epoch": 0.47927311035027653,
      "grad_norm": 1.4427847862243652,
      "learning_rate": 2.6049512773242034e-05,
      "loss": 1.5472,
      "step": 18198
    },
    {
      "epoch": 0.4792994469317883,
      "grad_norm": 2.8791027069091797,
      "learning_rate": 2.604819594416645e-05,
      "loss": 1.3314,
      "step": 18199
    },
    {
      "epoch": 0.47932578351329996,
      "grad_norm": 5.540414333343506,
      "learning_rate": 2.604687911509086e-05,
      "loss": 2.2421,
      "step": 18200
    },
    {
      "epoch": 0.4793521200948117,
      "grad_norm": 5.057022571563721,
      "learning_rate": 2.6045562286015274e-05,
      "loss": 0.7574,
      "step": 18201
    },
    {
      "epoch": 0.4793784566763234,
      "grad_norm": 2.3301544189453125,
      "learning_rate": 2.6044245456939693e-05,
      "loss": 1.0735,
      "step": 18202
    },
    {
      "epoch": 0.47940479325783514,
      "grad_norm": 1.5644989013671875,
      "learning_rate": 2.6042928627864105e-05,
      "loss": 1.4331,
      "step": 18203
    },
    {
      "epoch": 0.4794311298393468,
      "grad_norm": 3.384436845779419,
      "learning_rate": 2.6041611798788517e-05,
      "loss": 0.5878,
      "step": 18204
    },
    {
      "epoch": 0.47945746642085857,
      "grad_norm": 4.111256122589111,
      "learning_rate": 2.604029496971293e-05,
      "loss": 2.0897,
      "step": 18205
    },
    {
      "epoch": 0.4794838030023703,
      "grad_norm": 2.0982885360717773,
      "learning_rate": 2.6038978140637348e-05,
      "loss": 1.9615,
      "step": 18206
    },
    {
      "epoch": 0.479510139583882,
      "grad_norm": 3.7931666374206543,
      "learning_rate": 2.603766131156176e-05,
      "loss": 0.831,
      "step": 18207
    },
    {
      "epoch": 0.47953647616539374,
      "grad_norm": 4.855052947998047,
      "learning_rate": 2.6036344482486176e-05,
      "loss": 0.9936,
      "step": 18208
    },
    {
      "epoch": 0.47956281274690543,
      "grad_norm": 1.6595323085784912,
      "learning_rate": 2.6035027653410588e-05,
      "loss": 1.7072,
      "step": 18209
    },
    {
      "epoch": 0.4795891493284172,
      "grad_norm": 1.57021963596344,
      "learning_rate": 2.6033710824335e-05,
      "loss": 1.811,
      "step": 18210
    },
    {
      "epoch": 0.4796154859099289,
      "grad_norm": 1.3585107326507568,
      "learning_rate": 2.603239399525942e-05,
      "loss": 1.5773,
      "step": 18211
    },
    {
      "epoch": 0.4796418224914406,
      "grad_norm": 1.8488690853118896,
      "learning_rate": 2.603107716618383e-05,
      "loss": 1.7188,
      "step": 18212
    },
    {
      "epoch": 0.47966815907295235,
      "grad_norm": 3.2584662437438965,
      "learning_rate": 2.6029760337108244e-05,
      "loss": 1.1955,
      "step": 18213
    },
    {
      "epoch": 0.47969449565446404,
      "grad_norm": 1.6268631219863892,
      "learning_rate": 2.6028443508032656e-05,
      "loss": 1.4492,
      "step": 18214
    },
    {
      "epoch": 0.4797208322359758,
      "grad_norm": 1.6208415031433105,
      "learning_rate": 2.602712667895707e-05,
      "loss": 1.8537,
      "step": 18215
    },
    {
      "epoch": 0.47974716881748747,
      "grad_norm": 1.8901538848876953,
      "learning_rate": 2.6025809849881487e-05,
      "loss": 2.321,
      "step": 18216
    },
    {
      "epoch": 0.4797735053989992,
      "grad_norm": 1.631158471107483,
      "learning_rate": 2.6024493020805903e-05,
      "loss": 0.445,
      "step": 18217
    },
    {
      "epoch": 0.47979984198051095,
      "grad_norm": 2.484971523284912,
      "learning_rate": 2.6023176191730315e-05,
      "loss": 2.1254,
      "step": 18218
    },
    {
      "epoch": 0.47982617856202264,
      "grad_norm": 1.7783170938491821,
      "learning_rate": 2.6021859362654727e-05,
      "loss": 1.9419,
      "step": 18219
    },
    {
      "epoch": 0.4798525151435344,
      "grad_norm": 2.196690082550049,
      "learning_rate": 2.6020542533579146e-05,
      "loss": 2.1618,
      "step": 18220
    },
    {
      "epoch": 0.4798788517250461,
      "grad_norm": 1.8459336757659912,
      "learning_rate": 2.601922570450356e-05,
      "loss": 1.5345,
      "step": 18221
    },
    {
      "epoch": 0.4799051883065578,
      "grad_norm": 2.015507936477661,
      "learning_rate": 2.601790887542797e-05,
      "loss": 1.4419,
      "step": 18222
    },
    {
      "epoch": 0.4799315248880695,
      "grad_norm": 1.6010900735855103,
      "learning_rate": 2.6016592046352383e-05,
      "loss": 1.935,
      "step": 18223
    },
    {
      "epoch": 0.47995786146958125,
      "grad_norm": 1.7097499370574951,
      "learning_rate": 2.6015275217276795e-05,
      "loss": 1.0837,
      "step": 18224
    },
    {
      "epoch": 0.479984198051093,
      "grad_norm": 3.9856553077697754,
      "learning_rate": 2.6013958388201214e-05,
      "loss": 2.4888,
      "step": 18225
    },
    {
      "epoch": 0.4800105346326047,
      "grad_norm": 3.3391363620758057,
      "learning_rate": 2.6012641559125626e-05,
      "loss": 1.6933,
      "step": 18226
    },
    {
      "epoch": 0.4800368712141164,
      "grad_norm": 2.165415048599243,
      "learning_rate": 2.6011324730050042e-05,
      "loss": 2.046,
      "step": 18227
    },
    {
      "epoch": 0.4800632077956281,
      "grad_norm": 3.911407232284546,
      "learning_rate": 2.6010007900974454e-05,
      "loss": 0.6511,
      "step": 18228
    },
    {
      "epoch": 0.48008954437713985,
      "grad_norm": 1.8012775182724,
      "learning_rate": 2.6008691071898866e-05,
      "loss": 1.3854,
      "step": 18229
    },
    {
      "epoch": 0.48011588095865154,
      "grad_norm": 2.567828416824341,
      "learning_rate": 2.6007374242823285e-05,
      "loss": 0.7743,
      "step": 18230
    },
    {
      "epoch": 0.4801422175401633,
      "grad_norm": 2.0181589126586914,
      "learning_rate": 2.6006057413747697e-05,
      "loss": 2.3429,
      "step": 18231
    },
    {
      "epoch": 0.48016855412167503,
      "grad_norm": 1.8341394662857056,
      "learning_rate": 2.600474058467211e-05,
      "loss": 1.6094,
      "step": 18232
    },
    {
      "epoch": 0.4801948907031867,
      "grad_norm": 3.8895981311798096,
      "learning_rate": 2.6003423755596522e-05,
      "loss": 1.0459,
      "step": 18233
    },
    {
      "epoch": 0.48022122728469846,
      "grad_norm": 2.5247411727905273,
      "learning_rate": 2.600210692652094e-05,
      "loss": 1.2766,
      "step": 18234
    },
    {
      "epoch": 0.48024756386621015,
      "grad_norm": 1.8207776546478271,
      "learning_rate": 2.6000790097445353e-05,
      "loss": 2.0494,
      "step": 18235
    },
    {
      "epoch": 0.4802739004477219,
      "grad_norm": 2.1421167850494385,
      "learning_rate": 2.599947326836977e-05,
      "loss": 2.0554,
      "step": 18236
    },
    {
      "epoch": 0.4803002370292336,
      "grad_norm": 1.9610754251480103,
      "learning_rate": 2.599815643929418e-05,
      "loss": 1.6792,
      "step": 18237
    },
    {
      "epoch": 0.4803265736107453,
      "grad_norm": 2.152301073074341,
      "learning_rate": 2.5996839610218593e-05,
      "loss": 1.2206,
      "step": 18238
    },
    {
      "epoch": 0.48035291019225707,
      "grad_norm": 1.690805435180664,
      "learning_rate": 2.5995522781143012e-05,
      "loss": 2.1232,
      "step": 18239
    },
    {
      "epoch": 0.48037924677376875,
      "grad_norm": 1.7222952842712402,
      "learning_rate": 2.5994205952067424e-05,
      "loss": 0.591,
      "step": 18240
    },
    {
      "epoch": 0.4804055833552805,
      "grad_norm": 1.9552398920059204,
      "learning_rate": 2.5992889122991837e-05,
      "loss": 1.0758,
      "step": 18241
    },
    {
      "epoch": 0.4804319199367922,
      "grad_norm": 2.508448839187622,
      "learning_rate": 2.599157229391625e-05,
      "loss": 1.3292,
      "step": 18242
    },
    {
      "epoch": 0.48045825651830393,
      "grad_norm": 2.046084403991699,
      "learning_rate": 2.5990255464840664e-05,
      "loss": 1.4425,
      "step": 18243
    },
    {
      "epoch": 0.48048459309981567,
      "grad_norm": 1.7621982097625732,
      "learning_rate": 2.598893863576508e-05,
      "loss": 1.9327,
      "step": 18244
    },
    {
      "epoch": 0.48051092968132736,
      "grad_norm": 3.080350399017334,
      "learning_rate": 2.5987621806689496e-05,
      "loss": 0.7348,
      "step": 18245
    },
    {
      "epoch": 0.4805372662628391,
      "grad_norm": 2.543154001235962,
      "learning_rate": 2.5986304977613908e-05,
      "loss": 1.0225,
      "step": 18246
    },
    {
      "epoch": 0.4805636028443508,
      "grad_norm": 2.1923797130584717,
      "learning_rate": 2.598498814853832e-05,
      "loss": 2.7072,
      "step": 18247
    },
    {
      "epoch": 0.48058993942586253,
      "grad_norm": 1.423647403717041,
      "learning_rate": 2.5983671319462732e-05,
      "loss": 1.8857,
      "step": 18248
    },
    {
      "epoch": 0.4806162760073742,
      "grad_norm": 2.073836326599121,
      "learning_rate": 2.598235449038715e-05,
      "loss": 1.231,
      "step": 18249
    },
    {
      "epoch": 0.48064261258888596,
      "grad_norm": 2.9497969150543213,
      "learning_rate": 2.5981037661311563e-05,
      "loss": 1.6494,
      "step": 18250
    },
    {
      "epoch": 0.4806689491703977,
      "grad_norm": 1.5961450338363647,
      "learning_rate": 2.5979720832235976e-05,
      "loss": 2.1089,
      "step": 18251
    },
    {
      "epoch": 0.4806952857519094,
      "grad_norm": 1.5951591730117798,
      "learning_rate": 2.5978404003160388e-05,
      "loss": 1.7174,
      "step": 18252
    },
    {
      "epoch": 0.48072162233342114,
      "grad_norm": 2.378539800643921,
      "learning_rate": 2.5977087174084807e-05,
      "loss": 1.7267,
      "step": 18253
    },
    {
      "epoch": 0.4807479589149328,
      "grad_norm": 2.2286441326141357,
      "learning_rate": 2.597577034500922e-05,
      "loss": 1.2339,
      "step": 18254
    },
    {
      "epoch": 0.48077429549644457,
      "grad_norm": 2.03003191947937,
      "learning_rate": 2.5974453515933635e-05,
      "loss": 2.0078,
      "step": 18255
    },
    {
      "epoch": 0.48080063207795626,
      "grad_norm": 1.9827549457550049,
      "learning_rate": 2.5973136686858047e-05,
      "loss": 1.9373,
      "step": 18256
    },
    {
      "epoch": 0.480826968659468,
      "grad_norm": 3.3949034214019775,
      "learning_rate": 2.597181985778246e-05,
      "loss": 2.0879,
      "step": 18257
    },
    {
      "epoch": 0.48085330524097974,
      "grad_norm": 1.7559983730316162,
      "learning_rate": 2.5970503028706878e-05,
      "loss": 1.0735,
      "step": 18258
    },
    {
      "epoch": 0.48087964182249143,
      "grad_norm": 2.3310892581939697,
      "learning_rate": 2.596918619963129e-05,
      "loss": 0.3768,
      "step": 18259
    },
    {
      "epoch": 0.4809059784040032,
      "grad_norm": 1.615135908126831,
      "learning_rate": 2.5967869370555702e-05,
      "loss": 2.1364,
      "step": 18260
    },
    {
      "epoch": 0.48093231498551486,
      "grad_norm": 3.7796401977539062,
      "learning_rate": 2.5966552541480115e-05,
      "loss": 1.1534,
      "step": 18261
    },
    {
      "epoch": 0.4809586515670266,
      "grad_norm": 2.0392417907714844,
      "learning_rate": 2.596523571240453e-05,
      "loss": 1.9045,
      "step": 18262
    },
    {
      "epoch": 0.4809849881485383,
      "grad_norm": 6.02974271774292,
      "learning_rate": 2.5963918883328946e-05,
      "loss": 0.8431,
      "step": 18263
    },
    {
      "epoch": 0.48101132473005004,
      "grad_norm": 1.9574013948440552,
      "learning_rate": 2.596260205425336e-05,
      "loss": 2.3259,
      "step": 18264
    },
    {
      "epoch": 0.4810376613115618,
      "grad_norm": 4.27133321762085,
      "learning_rate": 2.5961285225177774e-05,
      "loss": 0.8533,
      "step": 18265
    },
    {
      "epoch": 0.48106399789307347,
      "grad_norm": 2.9611685276031494,
      "learning_rate": 2.5959968396102186e-05,
      "loss": 1.4665,
      "step": 18266
    },
    {
      "epoch": 0.4810903344745852,
      "grad_norm": 2.1282074451446533,
      "learning_rate": 2.5958651567026605e-05,
      "loss": 0.51,
      "step": 18267
    },
    {
      "epoch": 0.4811166710560969,
      "grad_norm": 3.044217109680176,
      "learning_rate": 2.5957334737951017e-05,
      "loss": 2.7244,
      "step": 18268
    },
    {
      "epoch": 0.48114300763760864,
      "grad_norm": 1.874723196029663,
      "learning_rate": 2.595601790887543e-05,
      "loss": 1.4811,
      "step": 18269
    },
    {
      "epoch": 0.48116934421912033,
      "grad_norm": 1.7602438926696777,
      "learning_rate": 2.595470107979984e-05,
      "loss": 1.9281,
      "step": 18270
    },
    {
      "epoch": 0.4811956808006321,
      "grad_norm": 2.15421462059021,
      "learning_rate": 2.5953384250724254e-05,
      "loss": 2.1792,
      "step": 18271
    },
    {
      "epoch": 0.4812220173821438,
      "grad_norm": 3.2429089546203613,
      "learning_rate": 2.5952067421648673e-05,
      "loss": 2.1389,
      "step": 18272
    },
    {
      "epoch": 0.4812483539636555,
      "grad_norm": 1.878605604171753,
      "learning_rate": 2.5950750592573085e-05,
      "loss": 1.7448,
      "step": 18273
    },
    {
      "epoch": 0.48127469054516725,
      "grad_norm": 2.79583477973938,
      "learning_rate": 2.59494337634975e-05,
      "loss": 1.6102,
      "step": 18274
    },
    {
      "epoch": 0.48130102712667894,
      "grad_norm": 2.2260453701019287,
      "learning_rate": 2.5948116934421913e-05,
      "loss": 1.8317,
      "step": 18275
    },
    {
      "epoch": 0.4813273637081907,
      "grad_norm": 2.351231813430786,
      "learning_rate": 2.5946800105346325e-05,
      "loss": 1.7722,
      "step": 18276
    },
    {
      "epoch": 0.4813537002897024,
      "grad_norm": 2.5532896518707275,
      "learning_rate": 2.5945483276270744e-05,
      "loss": 2.4244,
      "step": 18277
    },
    {
      "epoch": 0.4813800368712141,
      "grad_norm": 4.415773391723633,
      "learning_rate": 2.5944166447195156e-05,
      "loss": 1.5416,
      "step": 18278
    },
    {
      "epoch": 0.48140637345272586,
      "grad_norm": 2.400348663330078,
      "learning_rate": 2.5942849618119568e-05,
      "loss": 1.9609,
      "step": 18279
    },
    {
      "epoch": 0.48143271003423754,
      "grad_norm": 2.8336586952209473,
      "learning_rate": 2.594153278904398e-05,
      "loss": 0.6991,
      "step": 18280
    },
    {
      "epoch": 0.4814590466157493,
      "grad_norm": 1.6775230169296265,
      "learning_rate": 2.5940215959968396e-05,
      "loss": 1.4768,
      "step": 18281
    },
    {
      "epoch": 0.481485383197261,
      "grad_norm": 1.9266937971115112,
      "learning_rate": 2.593889913089281e-05,
      "loss": 1.7842,
      "step": 18282
    },
    {
      "epoch": 0.4815117197787727,
      "grad_norm": 2.667290210723877,
      "learning_rate": 2.5937582301817227e-05,
      "loss": 1.1571,
      "step": 18283
    },
    {
      "epoch": 0.48153805636028446,
      "grad_norm": 1.645289421081543,
      "learning_rate": 2.593626547274164e-05,
      "loss": 1.5977,
      "step": 18284
    },
    {
      "epoch": 0.48156439294179615,
      "grad_norm": 3.0950942039489746,
      "learning_rate": 2.593494864366605e-05,
      "loss": 2.1776,
      "step": 18285
    },
    {
      "epoch": 0.4815907295233079,
      "grad_norm": 2.822568893432617,
      "learning_rate": 2.593363181459047e-05,
      "loss": 1.7076,
      "step": 18286
    },
    {
      "epoch": 0.4816170661048196,
      "grad_norm": 1.8029379844665527,
      "learning_rate": 2.5932314985514883e-05,
      "loss": 1.6493,
      "step": 18287
    },
    {
      "epoch": 0.4816434026863313,
      "grad_norm": 1.8903818130493164,
      "learning_rate": 2.5930998156439295e-05,
      "loss": 1.4257,
      "step": 18288
    },
    {
      "epoch": 0.481669739267843,
      "grad_norm": 1.7043659687042236,
      "learning_rate": 2.5929681327363707e-05,
      "loss": 1.8823,
      "step": 18289
    },
    {
      "epoch": 0.48169607584935475,
      "grad_norm": 2.1602940559387207,
      "learning_rate": 2.5928364498288123e-05,
      "loss": 1.569,
      "step": 18290
    },
    {
      "epoch": 0.4817224124308665,
      "grad_norm": 1.5170824527740479,
      "learning_rate": 2.592704766921254e-05,
      "loss": 2.5523,
      "step": 18291
    },
    {
      "epoch": 0.4817487490123782,
      "grad_norm": 1.7479232549667358,
      "learning_rate": 2.5925730840136954e-05,
      "loss": 2.0913,
      "step": 18292
    },
    {
      "epoch": 0.48177508559388993,
      "grad_norm": 2.471151351928711,
      "learning_rate": 2.5924414011061366e-05,
      "loss": 1.2299,
      "step": 18293
    },
    {
      "epoch": 0.4818014221754016,
      "grad_norm": 2.0613322257995605,
      "learning_rate": 2.592309718198578e-05,
      "loss": 0.9024,
      "step": 18294
    },
    {
      "epoch": 0.48182775875691336,
      "grad_norm": 2.1863367557525635,
      "learning_rate": 2.592178035291019e-05,
      "loss": 1.3717,
      "step": 18295
    },
    {
      "epoch": 0.48185409533842505,
      "grad_norm": 1.8875648975372314,
      "learning_rate": 2.592046352383461e-05,
      "loss": 1.5818,
      "step": 18296
    },
    {
      "epoch": 0.4818804319199368,
      "grad_norm": 3.3052406311035156,
      "learning_rate": 2.5919146694759022e-05,
      "loss": 1.8997,
      "step": 18297
    },
    {
      "epoch": 0.48190676850144853,
      "grad_norm": 2.7625679969787598,
      "learning_rate": 2.5917829865683434e-05,
      "loss": 1.493,
      "step": 18298
    },
    {
      "epoch": 0.4819331050829602,
      "grad_norm": 2.534208297729492,
      "learning_rate": 2.5916513036607846e-05,
      "loss": 1.599,
      "step": 18299
    },
    {
      "epoch": 0.48195944166447197,
      "grad_norm": 4.247109413146973,
      "learning_rate": 2.5915196207532265e-05,
      "loss": 1.4251,
      "step": 18300
    },
    {
      "epoch": 0.48198577824598365,
      "grad_norm": 2.1316025257110596,
      "learning_rate": 2.5913879378456678e-05,
      "loss": 2.5669,
      "step": 18301
    },
    {
      "epoch": 0.4820121148274954,
      "grad_norm": 3.536114454269409,
      "learning_rate": 2.5912562549381093e-05,
      "loss": 0.5308,
      "step": 18302
    },
    {
      "epoch": 0.4820384514090071,
      "grad_norm": 2.2415454387664795,
      "learning_rate": 2.5911245720305505e-05,
      "loss": 1.7294,
      "step": 18303
    },
    {
      "epoch": 0.48206478799051883,
      "grad_norm": 1.5645477771759033,
      "learning_rate": 2.5909928891229918e-05,
      "loss": 1.7197,
      "step": 18304
    },
    {
      "epoch": 0.48209112457203057,
      "grad_norm": 1.7059056758880615,
      "learning_rate": 2.5908612062154337e-05,
      "loss": 0.9997,
      "step": 18305
    },
    {
      "epoch": 0.48211746115354226,
      "grad_norm": 3.156730890274048,
      "learning_rate": 2.590729523307875e-05,
      "loss": 0.336,
      "step": 18306
    },
    {
      "epoch": 0.482143797735054,
      "grad_norm": 1.3372992277145386,
      "learning_rate": 2.590597840400316e-05,
      "loss": 1.411,
      "step": 18307
    },
    {
      "epoch": 0.4821701343165657,
      "grad_norm": 2.8801941871643066,
      "learning_rate": 2.5904661574927573e-05,
      "loss": 1.6704,
      "step": 18308
    },
    {
      "epoch": 0.48219647089807743,
      "grad_norm": 1.8397315740585327,
      "learning_rate": 2.590334474585199e-05,
      "loss": 2.1636,
      "step": 18309
    },
    {
      "epoch": 0.4822228074795891,
      "grad_norm": 2.7086033821105957,
      "learning_rate": 2.5902027916776404e-05,
      "loss": 0.6801,
      "step": 18310
    },
    {
      "epoch": 0.48224914406110087,
      "grad_norm": 4.560272693634033,
      "learning_rate": 2.590071108770082e-05,
      "loss": 0.6993,
      "step": 18311
    },
    {
      "epoch": 0.4822754806426126,
      "grad_norm": 1.794438362121582,
      "learning_rate": 2.5899394258625232e-05,
      "loss": 2.3285,
      "step": 18312
    },
    {
      "epoch": 0.4823018172241243,
      "grad_norm": 1.6634902954101562,
      "learning_rate": 2.5898077429549644e-05,
      "loss": 1.7613,
      "step": 18313
    },
    {
      "epoch": 0.48232815380563604,
      "grad_norm": 2.5338780879974365,
      "learning_rate": 2.5896760600474057e-05,
      "loss": 0.7794,
      "step": 18314
    },
    {
      "epoch": 0.4823544903871477,
      "grad_norm": 1.5597007274627686,
      "learning_rate": 2.5895443771398476e-05,
      "loss": 1.9812,
      "step": 18315
    },
    {
      "epoch": 0.48238082696865947,
      "grad_norm": 1.5586351156234741,
      "learning_rate": 2.5894126942322888e-05,
      "loss": 2.5375,
      "step": 18316
    },
    {
      "epoch": 0.4824071635501712,
      "grad_norm": 2.1696550846099854,
      "learning_rate": 2.58928101132473e-05,
      "loss": 1.6872,
      "step": 18317
    },
    {
      "epoch": 0.4824335001316829,
      "grad_norm": 2.1278188228607178,
      "learning_rate": 2.5891493284171716e-05,
      "loss": 1.9642,
      "step": 18318
    },
    {
      "epoch": 0.48245983671319465,
      "grad_norm": 1.742224097251892,
      "learning_rate": 2.589017645509613e-05,
      "loss": 0.7188,
      "step": 18319
    },
    {
      "epoch": 0.48248617329470633,
      "grad_norm": 4.575339317321777,
      "learning_rate": 2.5888859626020547e-05,
      "loss": 1.0558,
      "step": 18320
    },
    {
      "epoch": 0.4825125098762181,
      "grad_norm": 5.727512359619141,
      "learning_rate": 2.588754279694496e-05,
      "loss": 1.3297,
      "step": 18321
    },
    {
      "epoch": 0.48253884645772976,
      "grad_norm": 1.7224305868148804,
      "learning_rate": 2.588622596786937e-05,
      "loss": 1.606,
      "step": 18322
    },
    {
      "epoch": 0.4825651830392415,
      "grad_norm": 1.7859320640563965,
      "learning_rate": 2.5884909138793783e-05,
      "loss": 1.5833,
      "step": 18323
    },
    {
      "epoch": 0.48259151962075325,
      "grad_norm": 1.3457127809524536,
      "learning_rate": 2.5883592309718202e-05,
      "loss": 1.9783,
      "step": 18324
    },
    {
      "epoch": 0.48261785620226494,
      "grad_norm": 1.807705044746399,
      "learning_rate": 2.5882275480642615e-05,
      "loss": 1.6852,
      "step": 18325
    },
    {
      "epoch": 0.4826441927837767,
      "grad_norm": 3.553330421447754,
      "learning_rate": 2.5880958651567027e-05,
      "loss": 1.1183,
      "step": 18326
    },
    {
      "epoch": 0.48267052936528837,
      "grad_norm": 3.3810200691223145,
      "learning_rate": 2.587964182249144e-05,
      "loss": 2.8145,
      "step": 18327
    },
    {
      "epoch": 0.4826968659468001,
      "grad_norm": 2.4965574741363525,
      "learning_rate": 2.5878324993415855e-05,
      "loss": 1.7972,
      "step": 18328
    },
    {
      "epoch": 0.4827232025283118,
      "grad_norm": 2.483815908432007,
      "learning_rate": 2.587700816434027e-05,
      "loss": 0.8307,
      "step": 18329
    },
    {
      "epoch": 0.48274953910982354,
      "grad_norm": 2.7984135150909424,
      "learning_rate": 2.5875691335264686e-05,
      "loss": 2.2607,
      "step": 18330
    },
    {
      "epoch": 0.4827758756913353,
      "grad_norm": 1.6728575229644775,
      "learning_rate": 2.5874374506189098e-05,
      "loss": 1.8287,
      "step": 18331
    },
    {
      "epoch": 0.482802212272847,
      "grad_norm": 2.076826810836792,
      "learning_rate": 2.587305767711351e-05,
      "loss": 1.2793,
      "step": 18332
    },
    {
      "epoch": 0.4828285488543587,
      "grad_norm": 2.7556533813476562,
      "learning_rate": 2.587174084803793e-05,
      "loss": 2.1633,
      "step": 18333
    },
    {
      "epoch": 0.4828548854358704,
      "grad_norm": 2.11775803565979,
      "learning_rate": 2.587042401896234e-05,
      "loss": 1.8455,
      "step": 18334
    },
    {
      "epoch": 0.48288122201738215,
      "grad_norm": 3.899864435195923,
      "learning_rate": 2.5869107189886754e-05,
      "loss": 1.2825,
      "step": 18335
    },
    {
      "epoch": 0.48290755859889384,
      "grad_norm": 2.0632147789001465,
      "learning_rate": 2.5867790360811166e-05,
      "loss": 2.3838,
      "step": 18336
    },
    {
      "epoch": 0.4829338951804056,
      "grad_norm": 1.4783732891082764,
      "learning_rate": 2.586647353173558e-05,
      "loss": 0.6448,
      "step": 18337
    },
    {
      "epoch": 0.4829602317619173,
      "grad_norm": 3.414386510848999,
      "learning_rate": 2.5865156702659997e-05,
      "loss": 1.6303,
      "step": 18338
    },
    {
      "epoch": 0.482986568343429,
      "grad_norm": 2.6229095458984375,
      "learning_rate": 2.5863839873584413e-05,
      "loss": 1.1264,
      "step": 18339
    },
    {
      "epoch": 0.48301290492494076,
      "grad_norm": 1.6566879749298096,
      "learning_rate": 2.5862523044508825e-05,
      "loss": 1.7613,
      "step": 18340
    },
    {
      "epoch": 0.48303924150645244,
      "grad_norm": 2.410510778427124,
      "learning_rate": 2.5861206215433237e-05,
      "loss": 0.9519,
      "step": 18341
    },
    {
      "epoch": 0.4830655780879642,
      "grad_norm": 1.8161232471466064,
      "learning_rate": 2.585988938635765e-05,
      "loss": 1.752,
      "step": 18342
    },
    {
      "epoch": 0.4830919146694759,
      "grad_norm": 3.2136988639831543,
      "learning_rate": 2.5858572557282068e-05,
      "loss": 1.9701,
      "step": 18343
    },
    {
      "epoch": 0.4831182512509876,
      "grad_norm": 1.500807523727417,
      "learning_rate": 2.585725572820648e-05,
      "loss": 1.8475,
      "step": 18344
    },
    {
      "epoch": 0.48314458783249936,
      "grad_norm": 2.430117607116699,
      "learning_rate": 2.5855938899130893e-05,
      "loss": 2.1502,
      "step": 18345
    },
    {
      "epoch": 0.48317092441401105,
      "grad_norm": 1.8485294580459595,
      "learning_rate": 2.5854622070055305e-05,
      "loss": 2.0114,
      "step": 18346
    },
    {
      "epoch": 0.4831972609955228,
      "grad_norm": 3.348314046859741,
      "learning_rate": 2.585330524097972e-05,
      "loss": 1.9875,
      "step": 18347
    },
    {
      "epoch": 0.4832235975770345,
      "grad_norm": 1.8929311037063599,
      "learning_rate": 2.585198841190414e-05,
      "loss": 1.7272,
      "step": 18348
    },
    {
      "epoch": 0.4832499341585462,
      "grad_norm": 1.764998197555542,
      "learning_rate": 2.585067158282855e-05,
      "loss": 1.7428,
      "step": 18349
    },
    {
      "epoch": 0.48327627074005797,
      "grad_norm": 4.091507911682129,
      "learning_rate": 2.5849354753752964e-05,
      "loss": 2.518,
      "step": 18350
    },
    {
      "epoch": 0.48330260732156965,
      "grad_norm": 3.2935714721679688,
      "learning_rate": 2.5848037924677376e-05,
      "loss": 1.1943,
      "step": 18351
    },
    {
      "epoch": 0.4833289439030814,
      "grad_norm": 2.3311710357666016,
      "learning_rate": 2.5846721095601795e-05,
      "loss": 0.9638,
      "step": 18352
    },
    {
      "epoch": 0.4833552804845931,
      "grad_norm": 1.7031148672103882,
      "learning_rate": 2.5845404266526207e-05,
      "loss": 2.1818,
      "step": 18353
    },
    {
      "epoch": 0.48338161706610483,
      "grad_norm": 3.7465553283691406,
      "learning_rate": 2.584408743745062e-05,
      "loss": 1.2991,
      "step": 18354
    },
    {
      "epoch": 0.4834079536476165,
      "grad_norm": 3.8707756996154785,
      "learning_rate": 2.5842770608375032e-05,
      "loss": 2.0623,
      "step": 18355
    },
    {
      "epoch": 0.48343429022912826,
      "grad_norm": 1.9105716943740845,
      "learning_rate": 2.5841453779299447e-05,
      "loss": 1.3937,
      "step": 18356
    },
    {
      "epoch": 0.48346062681064,
      "grad_norm": 2.868964672088623,
      "learning_rate": 2.5840136950223863e-05,
      "loss": 1.844,
      "step": 18357
    },
    {
      "epoch": 0.4834869633921517,
      "grad_norm": 2.0943338871002197,
      "learning_rate": 2.583882012114828e-05,
      "loss": 1.582,
      "step": 18358
    },
    {
      "epoch": 0.48351329997366344,
      "grad_norm": 3.172908067703247,
      "learning_rate": 2.583750329207269e-05,
      "loss": 1.6974,
      "step": 18359
    },
    {
      "epoch": 0.4835396365551751,
      "grad_norm": 2.3387062549591064,
      "learning_rate": 2.5836186462997103e-05,
      "loss": 1.1011,
      "step": 18360
    },
    {
      "epoch": 0.48356597313668687,
      "grad_norm": 3.5115859508514404,
      "learning_rate": 2.5834869633921515e-05,
      "loss": 1.3581,
      "step": 18361
    },
    {
      "epoch": 0.48359230971819855,
      "grad_norm": 2.8621480464935303,
      "learning_rate": 2.5833552804845934e-05,
      "loss": 1.1026,
      "step": 18362
    },
    {
      "epoch": 0.4836186462997103,
      "grad_norm": 2.5172183513641357,
      "learning_rate": 2.5832235975770346e-05,
      "loss": 1.8828,
      "step": 18363
    },
    {
      "epoch": 0.48364498288122204,
      "grad_norm": 2.953972816467285,
      "learning_rate": 2.583091914669476e-05,
      "loss": 0.8589,
      "step": 18364
    },
    {
      "epoch": 0.48367131946273373,
      "grad_norm": 1.8448625802993774,
      "learning_rate": 2.5829602317619174e-05,
      "loss": 1.9318,
      "step": 18365
    },
    {
      "epoch": 0.48369765604424547,
      "grad_norm": 1.8998172283172607,
      "learning_rate": 2.582828548854359e-05,
      "loss": 1.8291,
      "step": 18366
    },
    {
      "epoch": 0.48372399262575716,
      "grad_norm": 1.906227469444275,
      "learning_rate": 2.5826968659468005e-05,
      "loss": 1.3863,
      "step": 18367
    },
    {
      "epoch": 0.4837503292072689,
      "grad_norm": 1.814586877822876,
      "learning_rate": 2.5825651830392418e-05,
      "loss": 1.9128,
      "step": 18368
    },
    {
      "epoch": 0.4837766657887806,
      "grad_norm": 2.9641284942626953,
      "learning_rate": 2.582433500131683e-05,
      "loss": 1.5397,
      "step": 18369
    },
    {
      "epoch": 0.48380300237029233,
      "grad_norm": 4.049939155578613,
      "learning_rate": 2.5823018172241242e-05,
      "loss": 1.5262,
      "step": 18370
    },
    {
      "epoch": 0.4838293389518041,
      "grad_norm": 1.870164155960083,
      "learning_rate": 2.582170134316566e-05,
      "loss": 0.9732,
      "step": 18371
    },
    {
      "epoch": 0.48385567553331577,
      "grad_norm": 3.756333351135254,
      "learning_rate": 2.5820384514090073e-05,
      "loss": 1.1043,
      "step": 18372
    },
    {
      "epoch": 0.4838820121148275,
      "grad_norm": 2.0509378910064697,
      "learning_rate": 2.5819067685014485e-05,
      "loss": 1.9517,
      "step": 18373
    },
    {
      "epoch": 0.4839083486963392,
      "grad_norm": 5.011322975158691,
      "learning_rate": 2.5817750855938898e-05,
      "loss": 1.9828,
      "step": 18374
    },
    {
      "epoch": 0.48393468527785094,
      "grad_norm": 2.274775743484497,
      "learning_rate": 2.5816434026863313e-05,
      "loss": 2.181,
      "step": 18375
    },
    {
      "epoch": 0.48396102185936263,
      "grad_norm": 3.4376394748687744,
      "learning_rate": 2.581511719778773e-05,
      "loss": 2.1538,
      "step": 18376
    },
    {
      "epoch": 0.48398735844087437,
      "grad_norm": 1.577967643737793,
      "learning_rate": 2.5813800368712144e-05,
      "loss": 2.4193,
      "step": 18377
    },
    {
      "epoch": 0.4840136950223861,
      "grad_norm": 2.75225830078125,
      "learning_rate": 2.5812483539636557e-05,
      "loss": 1.4493,
      "step": 18378
    },
    {
      "epoch": 0.4840400316038978,
      "grad_norm": 1.5984172821044922,
      "learning_rate": 2.581116671056097e-05,
      "loss": 1.8152,
      "step": 18379
    },
    {
      "epoch": 0.48406636818540955,
      "grad_norm": 2.0202548503875732,
      "learning_rate": 2.580984988148538e-05,
      "loss": 1.726,
      "step": 18380
    },
    {
      "epoch": 0.48409270476692123,
      "grad_norm": 1.554030418395996,
      "learning_rate": 2.58085330524098e-05,
      "loss": 1.9233,
      "step": 18381
    },
    {
      "epoch": 0.484119041348433,
      "grad_norm": 1.8082032203674316,
      "learning_rate": 2.5807216223334212e-05,
      "loss": 1.015,
      "step": 18382
    },
    {
      "epoch": 0.4841453779299447,
      "grad_norm": 1.779860258102417,
      "learning_rate": 2.5805899394258624e-05,
      "loss": 0.5784,
      "step": 18383
    },
    {
      "epoch": 0.4841717145114564,
      "grad_norm": 2.2123851776123047,
      "learning_rate": 2.580458256518304e-05,
      "loss": 1.8172,
      "step": 18384
    },
    {
      "epoch": 0.48419805109296815,
      "grad_norm": 1.8945305347442627,
      "learning_rate": 2.5803265736107456e-05,
      "loss": 1.8525,
      "step": 18385
    },
    {
      "epoch": 0.48422438767447984,
      "grad_norm": 2.622844934463501,
      "learning_rate": 2.580194890703187e-05,
      "loss": 0.7779,
      "step": 18386
    },
    {
      "epoch": 0.4842507242559916,
      "grad_norm": 2.0889954566955566,
      "learning_rate": 2.5800632077956283e-05,
      "loss": 1.5266,
      "step": 18387
    },
    {
      "epoch": 0.48427706083750327,
      "grad_norm": 2.3298048973083496,
      "learning_rate": 2.5799315248880696e-05,
      "loss": 2.2032,
      "step": 18388
    },
    {
      "epoch": 0.484303397419015,
      "grad_norm": 1.5841577053070068,
      "learning_rate": 2.5797998419805108e-05,
      "loss": 1.8769,
      "step": 18389
    },
    {
      "epoch": 0.48432973400052676,
      "grad_norm": 3.13581919670105,
      "learning_rate": 2.5796681590729527e-05,
      "loss": 1.656,
      "step": 18390
    },
    {
      "epoch": 0.48435607058203844,
      "grad_norm": 2.65350341796875,
      "learning_rate": 2.579536476165394e-05,
      "loss": 0.3908,
      "step": 18391
    },
    {
      "epoch": 0.4843824071635502,
      "grad_norm": 3.9427220821380615,
      "learning_rate": 2.579404793257835e-05,
      "loss": 0.5817,
      "step": 18392
    },
    {
      "epoch": 0.4844087437450619,
      "grad_norm": 2.438901424407959,
      "learning_rate": 2.5792731103502767e-05,
      "loss": 2.3633,
      "step": 18393
    },
    {
      "epoch": 0.4844350803265736,
      "grad_norm": 2.4363667964935303,
      "learning_rate": 2.579141427442718e-05,
      "loss": 2.085,
      "step": 18394
    },
    {
      "epoch": 0.4844614169080853,
      "grad_norm": 1.9361634254455566,
      "learning_rate": 2.5790097445351598e-05,
      "loss": 2.1063,
      "step": 18395
    },
    {
      "epoch": 0.48448775348959705,
      "grad_norm": 2.282162666320801,
      "learning_rate": 2.578878061627601e-05,
      "loss": 0.898,
      "step": 18396
    },
    {
      "epoch": 0.4845140900711088,
      "grad_norm": 1.8887823820114136,
      "learning_rate": 2.5787463787200422e-05,
      "loss": 1.9128,
      "step": 18397
    },
    {
      "epoch": 0.4845404266526205,
      "grad_norm": 2.7833309173583984,
      "learning_rate": 2.5786146958124835e-05,
      "loss": 1.4563,
      "step": 18398
    },
    {
      "epoch": 0.4845667632341322,
      "grad_norm": 1.7120144367218018,
      "learning_rate": 2.5784830129049254e-05,
      "loss": 1.7172,
      "step": 18399
    },
    {
      "epoch": 0.4845930998156439,
      "grad_norm": 2.7394728660583496,
      "learning_rate": 2.5783513299973666e-05,
      "loss": 1.1691,
      "step": 18400
    },
    {
      "epoch": 0.48461943639715566,
      "grad_norm": 2.6390151977539062,
      "learning_rate": 2.5782196470898078e-05,
      "loss": 0.4177,
      "step": 18401
    },
    {
      "epoch": 0.48464577297866734,
      "grad_norm": 1.8348826169967651,
      "learning_rate": 2.578087964182249e-05,
      "loss": 3.2067,
      "step": 18402
    },
    {
      "epoch": 0.4846721095601791,
      "grad_norm": 2.5606539249420166,
      "learning_rate": 2.5779562812746906e-05,
      "loss": 2.3002,
      "step": 18403
    },
    {
      "epoch": 0.48469844614169083,
      "grad_norm": 2.3585174083709717,
      "learning_rate": 2.577824598367132e-05,
      "loss": 1.7013,
      "step": 18404
    },
    {
      "epoch": 0.4847247827232025,
      "grad_norm": 1.8755167722702026,
      "learning_rate": 2.5776929154595737e-05,
      "loss": 1.9228,
      "step": 18405
    },
    {
      "epoch": 0.48475111930471426,
      "grad_norm": 1.3895341157913208,
      "learning_rate": 2.577561232552015e-05,
      "loss": 1.8269,
      "step": 18406
    },
    {
      "epoch": 0.48477745588622595,
      "grad_norm": 1.6985033750534058,
      "learning_rate": 2.577429549644456e-05,
      "loss": 1.8735,
      "step": 18407
    },
    {
      "epoch": 0.4848037924677377,
      "grad_norm": 3.8967151641845703,
      "learning_rate": 2.5772978667368974e-05,
      "loss": 0.7406,
      "step": 18408
    },
    {
      "epoch": 0.4848301290492494,
      "grad_norm": 4.177510738372803,
      "learning_rate": 2.5771661838293393e-05,
      "loss": 1.4695,
      "step": 18409
    },
    {
      "epoch": 0.4848564656307611,
      "grad_norm": 1.816548228263855,
      "learning_rate": 2.5770345009217805e-05,
      "loss": 1.8405,
      "step": 18410
    },
    {
      "epoch": 0.48488280221227287,
      "grad_norm": 2.2118592262268066,
      "learning_rate": 2.5769028180142217e-05,
      "loss": 1.6375,
      "step": 18411
    },
    {
      "epoch": 0.48490913879378456,
      "grad_norm": 1.968703269958496,
      "learning_rate": 2.5767711351066633e-05,
      "loss": 1.5512,
      "step": 18412
    },
    {
      "epoch": 0.4849354753752963,
      "grad_norm": 2.9055683612823486,
      "learning_rate": 2.5766394521991045e-05,
      "loss": 1.5991,
      "step": 18413
    },
    {
      "epoch": 0.484961811956808,
      "grad_norm": 2.03057861328125,
      "learning_rate": 2.5765077692915464e-05,
      "loss": 0.9919,
      "step": 18414
    },
    {
      "epoch": 0.48498814853831973,
      "grad_norm": 1.6863213777542114,
      "learning_rate": 2.5763760863839876e-05,
      "loss": 1.9363,
      "step": 18415
    },
    {
      "epoch": 0.4850144851198315,
      "grad_norm": 2.704174041748047,
      "learning_rate": 2.5762444034764288e-05,
      "loss": 1.5539,
      "step": 18416
    },
    {
      "epoch": 0.48504082170134316,
      "grad_norm": 6.385892391204834,
      "learning_rate": 2.57611272056887e-05,
      "loss": 1.2474,
      "step": 18417
    },
    {
      "epoch": 0.4850671582828549,
      "grad_norm": 1.6319364309310913,
      "learning_rate": 2.575981037661312e-05,
      "loss": 2.1538,
      "step": 18418
    },
    {
      "epoch": 0.4850934948643666,
      "grad_norm": 2.219048261642456,
      "learning_rate": 2.5758493547537532e-05,
      "loss": 1.7055,
      "step": 18419
    },
    {
      "epoch": 0.48511983144587834,
      "grad_norm": 1.4454690217971802,
      "learning_rate": 2.5757176718461944e-05,
      "loss": 2.0454,
      "step": 18420
    },
    {
      "epoch": 0.48514616802739,
      "grad_norm": 1.5924925804138184,
      "learning_rate": 2.575585988938636e-05,
      "loss": 1.7619,
      "step": 18421
    },
    {
      "epoch": 0.48517250460890177,
      "grad_norm": 1.780044436454773,
      "learning_rate": 2.5754543060310772e-05,
      "loss": 1.9857,
      "step": 18422
    },
    {
      "epoch": 0.4851988411904135,
      "grad_norm": 2.2101314067840576,
      "learning_rate": 2.575322623123519e-05,
      "loss": 1.9862,
      "step": 18423
    },
    {
      "epoch": 0.4852251777719252,
      "grad_norm": 2.3404316902160645,
      "learning_rate": 2.5751909402159603e-05,
      "loss": 1.8558,
      "step": 18424
    },
    {
      "epoch": 0.48525151435343694,
      "grad_norm": 2.050753355026245,
      "learning_rate": 2.5750592573084015e-05,
      "loss": 1.8381,
      "step": 18425
    },
    {
      "epoch": 0.48527785093494863,
      "grad_norm": 3.0578525066375732,
      "learning_rate": 2.5749275744008427e-05,
      "loss": 0.5958,
      "step": 18426
    },
    {
      "epoch": 0.4853041875164604,
      "grad_norm": 1.8695067167282104,
      "learning_rate": 2.574795891493284e-05,
      "loss": 1.622,
      "step": 18427
    },
    {
      "epoch": 0.48533052409797206,
      "grad_norm": 2.1240391731262207,
      "learning_rate": 2.574664208585726e-05,
      "loss": 1.2508,
      "step": 18428
    },
    {
      "epoch": 0.4853568606794838,
      "grad_norm": 2.0256903171539307,
      "learning_rate": 2.574532525678167e-05,
      "loss": 1.4946,
      "step": 18429
    },
    {
      "epoch": 0.48538319726099555,
      "grad_norm": 2.7355430126190186,
      "learning_rate": 2.5744008427706083e-05,
      "loss": 2.0422,
      "step": 18430
    },
    {
      "epoch": 0.48540953384250723,
      "grad_norm": 1.7283477783203125,
      "learning_rate": 2.57426915986305e-05,
      "loss": 1.4768,
      "step": 18431
    },
    {
      "epoch": 0.485435870424019,
      "grad_norm": 3.126765727996826,
      "learning_rate": 2.5741374769554914e-05,
      "loss": 1.2515,
      "step": 18432
    },
    {
      "epoch": 0.48546220700553067,
      "grad_norm": 3.3315203189849854,
      "learning_rate": 2.574005794047933e-05,
      "loss": 2.1508,
      "step": 18433
    },
    {
      "epoch": 0.4854885435870424,
      "grad_norm": 1.8240033388137817,
      "learning_rate": 2.5738741111403742e-05,
      "loss": 1.3215,
      "step": 18434
    },
    {
      "epoch": 0.4855148801685541,
      "grad_norm": 1.2926433086395264,
      "learning_rate": 2.5737424282328154e-05,
      "loss": 1.5452,
      "step": 18435
    },
    {
      "epoch": 0.48554121675006584,
      "grad_norm": 3.809260845184326,
      "learning_rate": 2.5736107453252566e-05,
      "loss": 1.4537,
      "step": 18436
    },
    {
      "epoch": 0.4855675533315776,
      "grad_norm": 2.7715842723846436,
      "learning_rate": 2.5734790624176985e-05,
      "loss": 0.8447,
      "step": 18437
    },
    {
      "epoch": 0.48559388991308927,
      "grad_norm": 5.419256687164307,
      "learning_rate": 2.5733473795101398e-05,
      "loss": 2.0753,
      "step": 18438
    },
    {
      "epoch": 0.485620226494601,
      "grad_norm": 1.581094741821289,
      "learning_rate": 2.573215696602581e-05,
      "loss": 1.4794,
      "step": 18439
    },
    {
      "epoch": 0.4856465630761127,
      "grad_norm": 2.4569525718688965,
      "learning_rate": 2.5730840136950225e-05,
      "loss": 1.0419,
      "step": 18440
    },
    {
      "epoch": 0.48567289965762445,
      "grad_norm": 2.8132851123809814,
      "learning_rate": 2.5729523307874638e-05,
      "loss": 1.2611,
      "step": 18441
    },
    {
      "epoch": 0.48569923623913613,
      "grad_norm": 3.084397315979004,
      "learning_rate": 2.5728206478799057e-05,
      "loss": 2.0227,
      "step": 18442
    },
    {
      "epoch": 0.4857255728206479,
      "grad_norm": 3.458346128463745,
      "learning_rate": 2.572688964972347e-05,
      "loss": 2.8425,
      "step": 18443
    },
    {
      "epoch": 0.4857519094021596,
      "grad_norm": 2.111849069595337,
      "learning_rate": 2.572557282064788e-05,
      "loss": 1.9156,
      "step": 18444
    },
    {
      "epoch": 0.4857782459836713,
      "grad_norm": 1.8623846769332886,
      "learning_rate": 2.5724255991572293e-05,
      "loss": 2.0972,
      "step": 18445
    },
    {
      "epoch": 0.48580458256518305,
      "grad_norm": 2.8974735736846924,
      "learning_rate": 2.5722939162496705e-05,
      "loss": 1.4019,
      "step": 18446
    },
    {
      "epoch": 0.48583091914669474,
      "grad_norm": 1.8519647121429443,
      "learning_rate": 2.5721622333421124e-05,
      "loss": 2.0956,
      "step": 18447
    },
    {
      "epoch": 0.4858572557282065,
      "grad_norm": 3.604668617248535,
      "learning_rate": 2.5720305504345537e-05,
      "loss": 0.9024,
      "step": 18448
    },
    {
      "epoch": 0.4858835923097182,
      "grad_norm": 2.3852856159210205,
      "learning_rate": 2.571898867526995e-05,
      "loss": 1.6787,
      "step": 18449
    },
    {
      "epoch": 0.4859099288912299,
      "grad_norm": 1.5685434341430664,
      "learning_rate": 2.5717671846194364e-05,
      "loss": 1.896,
      "step": 18450
    },
    {
      "epoch": 0.48593626547274166,
      "grad_norm": 3.2516274452209473,
      "learning_rate": 2.571635501711878e-05,
      "loss": 1.9251,
      "step": 18451
    },
    {
      "epoch": 0.48596260205425335,
      "grad_norm": 1.8900718688964844,
      "learning_rate": 2.5715038188043196e-05,
      "loss": 1.4476,
      "step": 18452
    },
    {
      "epoch": 0.4859889386357651,
      "grad_norm": 3.0719430446624756,
      "learning_rate": 2.5713721358967608e-05,
      "loss": 1.9875,
      "step": 18453
    },
    {
      "epoch": 0.4860152752172768,
      "grad_norm": 2.0585649013519287,
      "learning_rate": 2.571240452989202e-05,
      "loss": 0.7618,
      "step": 18454
    },
    {
      "epoch": 0.4860416117987885,
      "grad_norm": 1.8559067249298096,
      "learning_rate": 2.5711087700816432e-05,
      "loss": 1.5609,
      "step": 18455
    },
    {
      "epoch": 0.48606794838030026,
      "grad_norm": 2.169935941696167,
      "learning_rate": 2.570977087174085e-05,
      "loss": 2.1225,
      "step": 18456
    },
    {
      "epoch": 0.48609428496181195,
      "grad_norm": 2.295045852661133,
      "learning_rate": 2.5708454042665263e-05,
      "loss": 1.6724,
      "step": 18457
    },
    {
      "epoch": 0.4861206215433237,
      "grad_norm": 1.7074811458587646,
      "learning_rate": 2.5707137213589676e-05,
      "loss": 1.6424,
      "step": 18458
    },
    {
      "epoch": 0.4861469581248354,
      "grad_norm": 4.542028903961182,
      "learning_rate": 2.570582038451409e-05,
      "loss": 1.4893,
      "step": 18459
    },
    {
      "epoch": 0.4861732947063471,
      "grad_norm": 1.9196478128433228,
      "learning_rate": 2.5704503555438503e-05,
      "loss": 1.4574,
      "step": 18460
    },
    {
      "epoch": 0.4861996312878588,
      "grad_norm": 2.9205195903778076,
      "learning_rate": 2.5703186726362922e-05,
      "loss": 1.4391,
      "step": 18461
    },
    {
      "epoch": 0.48622596786937056,
      "grad_norm": 1.8574875593185425,
      "learning_rate": 2.5701869897287335e-05,
      "loss": 2.1417,
      "step": 18462
    },
    {
      "epoch": 0.4862523044508823,
      "grad_norm": 2.81107234954834,
      "learning_rate": 2.5700553068211747e-05,
      "loss": 2.5147,
      "step": 18463
    },
    {
      "epoch": 0.486278641032394,
      "grad_norm": 2.6533210277557373,
      "learning_rate": 2.569923623913616e-05,
      "loss": 1.7436,
      "step": 18464
    },
    {
      "epoch": 0.48630497761390573,
      "grad_norm": 4.165168762207031,
      "learning_rate": 2.5697919410060578e-05,
      "loss": 2.0054,
      "step": 18465
    },
    {
      "epoch": 0.4863313141954174,
      "grad_norm": 1.9256856441497803,
      "learning_rate": 2.569660258098499e-05,
      "loss": 1.9637,
      "step": 18466
    },
    {
      "epoch": 0.48635765077692916,
      "grad_norm": 1.7320513725280762,
      "learning_rate": 2.5695285751909402e-05,
      "loss": 1.8098,
      "step": 18467
    },
    {
      "epoch": 0.48638398735844085,
      "grad_norm": 2.1129705905914307,
      "learning_rate": 2.5693968922833818e-05,
      "loss": 1.1088,
      "step": 18468
    },
    {
      "epoch": 0.4864103239399526,
      "grad_norm": 5.155686378479004,
      "learning_rate": 2.569265209375823e-05,
      "loss": 1.3488,
      "step": 18469
    },
    {
      "epoch": 0.48643666052146434,
      "grad_norm": 3.6876680850982666,
      "learning_rate": 2.569133526468265e-05,
      "loss": 1.6031,
      "step": 18470
    },
    {
      "epoch": 0.486462997102976,
      "grad_norm": 1.8359041213989258,
      "learning_rate": 2.569001843560706e-05,
      "loss": 1.4716,
      "step": 18471
    },
    {
      "epoch": 0.48648933368448777,
      "grad_norm": 2.9589924812316895,
      "learning_rate": 2.5688701606531474e-05,
      "loss": 0.6421,
      "step": 18472
    },
    {
      "epoch": 0.48651567026599946,
      "grad_norm": 1.6885493993759155,
      "learning_rate": 2.5687384777455886e-05,
      "loss": 2.1624,
      "step": 18473
    },
    {
      "epoch": 0.4865420068475112,
      "grad_norm": 3.8228204250335693,
      "learning_rate": 2.5686067948380298e-05,
      "loss": 1.9418,
      "step": 18474
    },
    {
      "epoch": 0.4865683434290229,
      "grad_norm": 3.8486132621765137,
      "learning_rate": 2.5684751119304717e-05,
      "loss": 1.9678,
      "step": 18475
    },
    {
      "epoch": 0.48659468001053463,
      "grad_norm": 2.923513412475586,
      "learning_rate": 2.568343429022913e-05,
      "loss": 1.6008,
      "step": 18476
    },
    {
      "epoch": 0.4866210165920464,
      "grad_norm": 2.0346717834472656,
      "learning_rate": 2.568211746115354e-05,
      "loss": 1.6052,
      "step": 18477
    },
    {
      "epoch": 0.48664735317355806,
      "grad_norm": 4.474893569946289,
      "learning_rate": 2.5680800632077957e-05,
      "loss": 1.6256,
      "step": 18478
    },
    {
      "epoch": 0.4866736897550698,
      "grad_norm": 4.485257625579834,
      "learning_rate": 2.567948380300237e-05,
      "loss": 1.468,
      "step": 18479
    },
    {
      "epoch": 0.4867000263365815,
      "grad_norm": 2.5774567127227783,
      "learning_rate": 2.5678166973926788e-05,
      "loss": 1.6469,
      "step": 18480
    },
    {
      "epoch": 0.48672636291809324,
      "grad_norm": 2.047030448913574,
      "learning_rate": 2.56768501448512e-05,
      "loss": 2.1457,
      "step": 18481
    },
    {
      "epoch": 0.4867526994996049,
      "grad_norm": 1.8974688053131104,
      "learning_rate": 2.5675533315775613e-05,
      "loss": 2.2784,
      "step": 18482
    },
    {
      "epoch": 0.48677903608111667,
      "grad_norm": 4.346259593963623,
      "learning_rate": 2.5674216486700025e-05,
      "loss": 1.9088,
      "step": 18483
    },
    {
      "epoch": 0.4868053726626284,
      "grad_norm": 1.8092743158340454,
      "learning_rate": 2.5672899657624444e-05,
      "loss": 1.8927,
      "step": 18484
    },
    {
      "epoch": 0.4868317092441401,
      "grad_norm": 3.205528736114502,
      "learning_rate": 2.5671582828548856e-05,
      "loss": 1.3568,
      "step": 18485
    },
    {
      "epoch": 0.48685804582565184,
      "grad_norm": 1.6814615726470947,
      "learning_rate": 2.567026599947327e-05,
      "loss": 2.2728,
      "step": 18486
    },
    {
      "epoch": 0.48688438240716353,
      "grad_norm": 5.064326763153076,
      "learning_rate": 2.5668949170397684e-05,
      "loss": 1.3509,
      "step": 18487
    },
    {
      "epoch": 0.4869107189886753,
      "grad_norm": 3.4415476322174072,
      "learning_rate": 2.5667632341322096e-05,
      "loss": 2.8161,
      "step": 18488
    },
    {
      "epoch": 0.486937055570187,
      "grad_norm": 2.522833824157715,
      "learning_rate": 2.5666315512246515e-05,
      "loss": 1.0294,
      "step": 18489
    },
    {
      "epoch": 0.4869633921516987,
      "grad_norm": 2.1585633754730225,
      "learning_rate": 2.5664998683170927e-05,
      "loss": 1.3564,
      "step": 18490
    },
    {
      "epoch": 0.48698972873321045,
      "grad_norm": 2.394378900527954,
      "learning_rate": 2.566368185409534e-05,
      "loss": 2.4195,
      "step": 18491
    },
    {
      "epoch": 0.48701606531472214,
      "grad_norm": 2.5911078453063965,
      "learning_rate": 2.5662365025019752e-05,
      "loss": 1.2326,
      "step": 18492
    },
    {
      "epoch": 0.4870424018962339,
      "grad_norm": 1.7802774906158447,
      "learning_rate": 2.5661048195944164e-05,
      "loss": 0.6364,
      "step": 18493
    },
    {
      "epoch": 0.48706873847774557,
      "grad_norm": 2.638225555419922,
      "learning_rate": 2.5659731366868583e-05,
      "loss": 1.918,
      "step": 18494
    },
    {
      "epoch": 0.4870950750592573,
      "grad_norm": 1.846629023551941,
      "learning_rate": 2.5658414537792995e-05,
      "loss": 1.646,
      "step": 18495
    },
    {
      "epoch": 0.48712141164076905,
      "grad_norm": 2.197716236114502,
      "learning_rate": 2.565709770871741e-05,
      "loss": 1.7315,
      "step": 18496
    },
    {
      "epoch": 0.48714774822228074,
      "grad_norm": 1.91517174243927,
      "learning_rate": 2.5655780879641823e-05,
      "loss": 1.6118,
      "step": 18497
    },
    {
      "epoch": 0.4871740848037925,
      "grad_norm": 4.605914115905762,
      "learning_rate": 2.5654464050566242e-05,
      "loss": 1.6682,
      "step": 18498
    },
    {
      "epoch": 0.48720042138530417,
      "grad_norm": 2.0153045654296875,
      "learning_rate": 2.5653147221490654e-05,
      "loss": 1.3794,
      "step": 18499
    },
    {
      "epoch": 0.4872267579668159,
      "grad_norm": 1.849891185760498,
      "learning_rate": 2.5651830392415066e-05,
      "loss": 1.4523,
      "step": 18500
    }
  ],
  "logging_steps": 1,
  "max_steps": 37970,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.6567889688508826e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}