{
  "best_metric": 0.642096996307373,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8592910848549946,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004296455424274973,
      "grad_norm": 3.7213289737701416,
      "learning_rate": 1e-05,
      "loss": 1.3846,
      "step": 1
    },
    {
      "epoch": 0.004296455424274973,
      "eval_loss": 3.0925545692443848,
      "eval_runtime": 6.6461,
      "eval_samples_per_second": 58.982,
      "eval_steps_per_second": 14.745,
      "step": 1
    },
    {
      "epoch": 0.008592910848549946,
      "grad_norm": 4.857822895050049,
      "learning_rate": 2e-05,
      "loss": 1.8958,
      "step": 2
    },
    {
      "epoch": 0.01288936627282492,
      "grad_norm": 5.981305122375488,
      "learning_rate": 3e-05,
      "loss": 2.0238,
      "step": 3
    },
    {
      "epoch": 0.017185821697099892,
      "grad_norm": 5.285318374633789,
      "learning_rate": 4e-05,
      "loss": 2.1652,
      "step": 4
    },
    {
      "epoch": 0.021482277121374866,
      "grad_norm": 3.245223045349121,
      "learning_rate": 5e-05,
      "loss": 1.8591,
      "step": 5
    },
    {
      "epoch": 0.02577873254564984,
      "grad_norm": 3.0641939640045166,
      "learning_rate": 6e-05,
      "loss": 1.7127,
      "step": 6
    },
    {
      "epoch": 0.03007518796992481,
      "grad_norm": 3.007746458053589,
      "learning_rate": 7e-05,
      "loss": 1.5769,
      "step": 7
    },
    {
      "epoch": 0.034371643394199784,
      "grad_norm": 3.19362735748291,
      "learning_rate": 8e-05,
      "loss": 1.3185,
      "step": 8
    },
    {
      "epoch": 0.03866809881847476,
      "grad_norm": 2.9073708057403564,
      "learning_rate": 9e-05,
      "loss": 1.2654,
      "step": 9
    },
    {
      "epoch": 0.04296455424274973,
      "grad_norm": 2.2574188709259033,
      "learning_rate": 0.0001,
      "loss": 1.2634,
      "step": 10
    },
    {
      "epoch": 0.047261009667024706,
      "grad_norm": 2.9579427242279053,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.4193,
      "step": 11
    },
    {
      "epoch": 0.05155746509129968,
      "grad_norm": 1.50870680809021,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.9728,
      "step": 12
    },
    {
      "epoch": 0.055853920515574654,
      "grad_norm": 1.8414061069488525,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.1333,
      "step": 13
    },
    {
      "epoch": 0.06015037593984962,
      "grad_norm": 1.6953611373901367,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.0486,
      "step": 14
    },
    {
      "epoch": 0.0644468313641246,
      "grad_norm": 1.4926620721817017,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.0834,
      "step": 15
    },
    {
      "epoch": 0.06874328678839957,
      "grad_norm": 1.9215829372406006,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.0414,
      "step": 16
    },
    {
      "epoch": 0.07303974221267455,
      "grad_norm": 1.6022037267684937,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.9147,
      "step": 17
    },
    {
      "epoch": 0.07733619763694952,
      "grad_norm": 1.7988508939743042,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.9868,
      "step": 18
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 1.4760843515396118,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.9118,
      "step": 19
    },
    {
      "epoch": 0.08592910848549946,
      "grad_norm": 2.768462657928467,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.9134,
      "step": 20
    },
    {
      "epoch": 0.09022556390977443,
      "grad_norm": 1.8300420045852661,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1868,
      "step": 21
    },
    {
      "epoch": 0.09452201933404941,
      "grad_norm": 2.1668691635131836,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1904,
      "step": 22
    },
    {
      "epoch": 0.09881847475832438,
      "grad_norm": 2.031832218170166,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.0189,
      "step": 23
    },
    {
      "epoch": 0.10311493018259936,
      "grad_norm": 2.1706461906433105,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1083,
      "step": 24
    },
    {
      "epoch": 0.10741138560687433,
      "grad_norm": 2.0704774856567383,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.0931,
      "step": 25
    },
    {
      "epoch": 0.11170784103114931,
      "grad_norm": 2.4387216567993164,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.1432,
      "step": 26
    },
    {
      "epoch": 0.11600429645542427,
      "grad_norm": 2.168456792831421,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.1215,
      "step": 27
    },
    {
      "epoch": 0.12030075187969924,
      "grad_norm": 1.818217158317566,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.9617,
      "step": 28
    },
    {
      "epoch": 0.12459720730397422,
      "grad_norm": 1.9304856061935425,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.038,
      "step": 29
    },
    {
      "epoch": 0.1288936627282492,
      "grad_norm": 1.8753042221069336,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.0944,
      "step": 30
    },
    {
      "epoch": 0.13319011815252416,
      "grad_norm": 2.0664823055267334,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.9842,
      "step": 31
    },
    {
      "epoch": 0.13748657357679914,
      "grad_norm": 2.7358603477478027,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.0543,
      "step": 32
    },
    {
      "epoch": 0.14178302900107412,
      "grad_norm": 2.3741488456726074,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.9306,
      "step": 33
    },
    {
      "epoch": 0.1460794844253491,
      "grad_norm": 2.5049633979797363,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0233,
      "step": 34
    },
    {
      "epoch": 0.15037593984962405,
      "grad_norm": 2.42488694190979,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.247,
      "step": 35
    },
    {
      "epoch": 0.15467239527389903,
      "grad_norm": 2.060638189315796,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.9915,
      "step": 36
    },
    {
      "epoch": 0.158968850698174,
      "grad_norm": 2.7562460899353027,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.953,
      "step": 37
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 1.9984441995620728,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.9137,
      "step": 38
    },
    {
      "epoch": 0.16756176154672395,
      "grad_norm": 1.9816590547561646,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.933,
      "step": 39
    },
    {
      "epoch": 0.17185821697099893,
      "grad_norm": 2.506340742111206,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.0371,
      "step": 40
    },
    {
      "epoch": 0.1761546723952739,
      "grad_norm": 2.7125437259674072,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1284,
      "step": 41
    },
    {
      "epoch": 0.18045112781954886,
      "grad_norm": 2.310711145401001,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.0132,
      "step": 42
    },
    {
      "epoch": 0.18474758324382384,
      "grad_norm": 2.6945035457611084,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.927,
      "step": 43
    },
    {
      "epoch": 0.18904403866809882,
      "grad_norm": 2.5864429473876953,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.7936,
      "step": 44
    },
    {
      "epoch": 0.1933404940923738,
      "grad_norm": 2.4397449493408203,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.9487,
      "step": 45
    },
    {
      "epoch": 0.19763694951664876,
      "grad_norm": 2.9100451469421387,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1686,
      "step": 46
    },
    {
      "epoch": 0.20193340494092374,
      "grad_norm": 3.3991737365722656,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.1151,
      "step": 47
    },
    {
      "epoch": 0.20622986036519872,
      "grad_norm": 8.242486953735352,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5061,
      "step": 48
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 9.160292625427246,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6933,
      "step": 49
    },
    {
      "epoch": 0.21482277121374865,
      "grad_norm": 2.3462066650390625,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.5101,
      "step": 50
    },
    {
      "epoch": 0.21482277121374865,
      "eval_loss": 1.189436912536621,
      "eval_runtime": 6.4418,
      "eval_samples_per_second": 60.852,
      "eval_steps_per_second": 15.213,
      "step": 50
    },
    {
      "epoch": 0.21911922663802363,
      "grad_norm": 5.1871843338012695,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.2066,
      "step": 51
    },
    {
      "epoch": 0.22341568206229862,
      "grad_norm": 4.914016246795654,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2274,
      "step": 52
    },
    {
      "epoch": 0.22771213748657357,
      "grad_norm": 4.083756923675537,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.2134,
      "step": 53
    },
    {
      "epoch": 0.23200859291084855,
      "grad_norm": 3.1531567573547363,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.0547,
      "step": 54
    },
    {
      "epoch": 0.23630504833512353,
      "grad_norm": 2.098278045654297,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.975,
      "step": 55
    },
    {
      "epoch": 0.24060150375939848,
      "grad_norm": 1.6753902435302734,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.914,
      "step": 56
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 1.6551477909088135,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.8647,
      "step": 57
    },
    {
      "epoch": 0.24919441460794844,
      "grad_norm": 1.695286750793457,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.861,
      "step": 58
    },
    {
      "epoch": 0.2534908700322234,
      "grad_norm": 1.93912935256958,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.8712,
      "step": 59
    },
    {
      "epoch": 0.2577873254564984,
      "grad_norm": 1.7504533529281616,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.0136,
      "step": 60
    },
    {
      "epoch": 0.2620837808807734,
      "grad_norm": 1.331573724746704,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.7822,
      "step": 61
    },
    {
      "epoch": 0.2663802363050483,
      "grad_norm": 1.2616267204284668,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.7795,
      "step": 62
    },
    {
      "epoch": 0.2706766917293233,
      "grad_norm": 1.3351017236709595,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.738,
      "step": 63
    },
    {
      "epoch": 0.2749731471535983,
      "grad_norm": 1.184287667274475,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.7535,
      "step": 64
    },
    {
      "epoch": 0.27926960257787325,
      "grad_norm": 1.1478452682495117,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.7001,
      "step": 65
    },
    {
      "epoch": 0.28356605800214824,
      "grad_norm": 1.2412298917770386,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.7398,
      "step": 66
    },
    {
      "epoch": 0.2878625134264232,
      "grad_norm": 1.1977274417877197,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.6985,
      "step": 67
    },
    {
      "epoch": 0.2921589688506982,
      "grad_norm": 1.1713128089904785,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.7861,
      "step": 68
    },
    {
      "epoch": 0.2964554242749731,
      "grad_norm": 1.2367181777954102,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.6473,
      "step": 69
    },
    {
      "epoch": 0.3007518796992481,
      "grad_norm": 1.2370542287826538,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.7719,
      "step": 70
    },
    {
      "epoch": 0.3050483351235231,
      "grad_norm": 1.3135274648666382,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.8051,
      "step": 71
    },
    {
      "epoch": 0.30934479054779807,
      "grad_norm": 1.3031071424484253,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.6723,
      "step": 72
    },
    {
      "epoch": 0.31364124597207305,
      "grad_norm": 1.2682934999465942,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.7005,
      "step": 73
    },
    {
      "epoch": 0.317937701396348,
      "grad_norm": 1.3928266763687134,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.7691,
      "step": 74
    },
    {
      "epoch": 0.322234156820623,
      "grad_norm": 1.6352171897888184,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.8317,
      "step": 75
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 1.7115323543548584,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.8111,
      "step": 76
    },
    {
      "epoch": 0.3308270676691729,
      "grad_norm": 1.5842256546020508,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.7893,
      "step": 77
    },
    {
      "epoch": 0.3351235230934479,
      "grad_norm": 1.8490431308746338,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.8194,
      "step": 78
    },
    {
      "epoch": 0.3394199785177229,
      "grad_norm": 2.136834144592285,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.0555,
      "step": 79
    },
    {
      "epoch": 0.34371643394199786,
      "grad_norm": 1.652445673942566,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.7407,
      "step": 80
    },
    {
      "epoch": 0.34801288936627284,
      "grad_norm": 1.7499744892120361,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.8236,
      "step": 81
    },
    {
      "epoch": 0.3523093447905478,
      "grad_norm": 1.6274521350860596,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.8268,
      "step": 82
    },
    {
      "epoch": 0.35660580021482274,
      "grad_norm": 1.6025863885879517,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.7074,
      "step": 83
    },
    {
      "epoch": 0.3609022556390977,
      "grad_norm": 2.065844774246216,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.9598,
      "step": 84
    },
    {
      "epoch": 0.3651987110633727,
      "grad_norm": 1.4625885486602783,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.7081,
      "step": 85
    },
    {
      "epoch": 0.3694951664876477,
      "grad_norm": 1.8904412984848022,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.8011,
      "step": 86
    },
    {
      "epoch": 0.37379162191192267,
      "grad_norm": 1.763213038444519,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.85,
      "step": 87
    },
    {
      "epoch": 0.37808807733619765,
      "grad_norm": 1.809980869293213,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.8742,
      "step": 88
    },
    {
      "epoch": 0.38238453276047263,
      "grad_norm": 1.7497470378875732,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.7314,
      "step": 89
    },
    {
      "epoch": 0.3866809881847476,
      "grad_norm": 1.658308982849121,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.6642,
      "step": 90
    },
    {
      "epoch": 0.39097744360902253,
      "grad_norm": 2.3599700927734375,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.6236,
      "step": 91
    },
    {
      "epoch": 0.3952738990332975,
      "grad_norm": 2.1833715438842773,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.8379,
      "step": 92
    },
    {
      "epoch": 0.3995703544575725,
      "grad_norm": 1.9399131536483765,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.6815,
      "step": 93
    },
    {
      "epoch": 0.4038668098818475,
      "grad_norm": 1.9606842994689941,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.7068,
      "step": 94
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 2.4859061241149902,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.9544,
      "step": 95
    },
    {
      "epoch": 0.41245972073039744,
      "grad_norm": 2.272859573364258,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.7826,
      "step": 96
    },
    {
      "epoch": 0.4167561761546724,
      "grad_norm": 2.4689385890960693,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.8473,
      "step": 97
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 2.3891687393188477,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.7084,
      "step": 98
    },
    {
      "epoch": 0.4253490870032223,
      "grad_norm": 3.3651132583618164,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.1093,
      "step": 99
    },
    {
      "epoch": 0.4296455424274973,
      "grad_norm": 1.983601450920105,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3948,
      "step": 100
    },
    {
      "epoch": 0.4296455424274973,
      "eval_loss": 0.8723666667938232,
      "eval_runtime": 6.4139,
      "eval_samples_per_second": 61.117,
      "eval_steps_per_second": 15.279,
      "step": 100
    },
    {
      "epoch": 0.4339419978517723,
      "grad_norm": 2.4359476566314697,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.8804,
      "step": 101
    },
    {
      "epoch": 0.43823845327604727,
      "grad_norm": 2.6817538738250732,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.0145,
      "step": 102
    },
    {
      "epoch": 0.44253490870032225,
      "grad_norm": 2.778102397918701,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.1181,
      "step": 103
    },
    {
      "epoch": 0.44683136412459723,
      "grad_norm": 2.5488288402557373,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.9167,
      "step": 104
    },
    {
      "epoch": 0.45112781954887216,
      "grad_norm": 1.968772292137146,
      "learning_rate": 5e-05,
      "loss": 0.8341,
      "step": 105
    },
    {
      "epoch": 0.45542427497314714,
      "grad_norm": 1.894675612449646,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.8267,
      "step": 106
    },
    {
      "epoch": 0.4597207303974221,
      "grad_norm": 1.5120309591293335,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.6824,
      "step": 107
    },
    {
      "epoch": 0.4640171858216971,
      "grad_norm": 1.3843655586242676,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.7507,
      "step": 108
    },
    {
      "epoch": 0.4683136412459721,
      "grad_norm": 1.2969028949737549,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.7291,
      "step": 109
    },
    {
      "epoch": 0.47261009667024706,
      "grad_norm": 1.2888069152832031,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.7187,
      "step": 110
    },
    {
      "epoch": 0.47690655209452204,
      "grad_norm": 1.3067426681518555,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.7991,
      "step": 111
    },
    {
      "epoch": 0.48120300751879697,
      "grad_norm": 1.396358609199524,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.7577,
      "step": 112
    },
    {
      "epoch": 0.48549946294307195,
      "grad_norm": 1.2107123136520386,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.6896,
      "step": 113
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 1.2302546501159668,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.6197,
      "step": 114
    },
    {
      "epoch": 0.4940923737916219,
      "grad_norm": 1.155816674232483,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.5821,
      "step": 115
    },
    {
      "epoch": 0.4983888292158969,
      "grad_norm": 1.385813593864441,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.7065,
      "step": 116
    },
    {
      "epoch": 0.5026852846401718,
      "grad_norm": 1.3972958326339722,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.6254,
      "step": 117
    },
    {
      "epoch": 0.5069817400644469,
      "grad_norm": 1.1042088270187378,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.5305,
      "step": 118
    },
    {
      "epoch": 0.5112781954887218,
      "grad_norm": 1.5464346408843994,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.6977,
      "step": 119
    },
    {
      "epoch": 0.5155746509129968,
      "grad_norm": 1.3161275386810303,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.5552,
      "step": 120
    },
    {
      "epoch": 0.5198711063372717,
      "grad_norm": 1.4442644119262695,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.5969,
      "step": 121
    },
    {
      "epoch": 0.5241675617615468,
      "grad_norm": 1.8110568523406982,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.7694,
      "step": 122
    },
    {
      "epoch": 0.5284640171858217,
      "grad_norm": 1.5310821533203125,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.6087,
      "step": 123
    },
    {
      "epoch": 0.5327604726100966,
      "grad_norm": 1.6861546039581299,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.7382,
      "step": 124
    },
    {
      "epoch": 0.5370569280343717,
      "grad_norm": 1.5911574363708496,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.7375,
      "step": 125
    },
    {
      "epoch": 0.5413533834586466,
      "grad_norm": 1.7347263097763062,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.7894,
      "step": 126
    },
    {
      "epoch": 0.5456498388829216,
      "grad_norm": 1.7925807237625122,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.7615,
      "step": 127
    },
    {
      "epoch": 0.5499462943071965,
      "grad_norm": 1.813718557357788,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.7268,
      "step": 128
    },
    {
      "epoch": 0.5542427497314716,
      "grad_norm": 1.7056523561477661,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7985,
      "step": 129
    },
    {
      "epoch": 0.5585392051557465,
      "grad_norm": 1.7043051719665527,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.6707,
      "step": 130
    },
    {
      "epoch": 0.5628356605800214,
      "grad_norm": 1.7093660831451416,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.6382,
      "step": 131
    },
    {
      "epoch": 0.5671321160042965,
      "grad_norm": 1.7664512395858765,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.7141,
      "step": 132
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.6907439231872559,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.5828,
      "step": 133
    },
    {
      "epoch": 0.5757250268528464,
      "grad_norm": 1.8687456846237183,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.9244,
      "step": 134
    },
    {
      "epoch": 0.5800214822771214,
      "grad_norm": 1.8165651559829712,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.7633,
      "step": 135
    },
    {
      "epoch": 0.5843179377013964,
      "grad_norm": 1.4862018823623657,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.5513,
      "step": 136
    },
    {
      "epoch": 0.5886143931256713,
      "grad_norm": 1.6100958585739136,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.6509,
      "step": 137
    },
    {
      "epoch": 0.5929108485499462,
      "grad_norm": 1.5446810722351074,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.6008,
      "step": 138
    },
    {
      "epoch": 0.5972073039742213,
      "grad_norm": 3.325167655944824,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.6454,
      "step": 139
    },
    {
      "epoch": 0.6015037593984962,
      "grad_norm": 2.6420528888702393,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.6219,
      "step": 140
    },
    {
      "epoch": 0.6058002148227712,
      "grad_norm": 1.7877399921417236,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.7111,
      "step": 141
    },
    {
      "epoch": 0.6100966702470462,
      "grad_norm": 1.682938814163208,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.6371,
      "step": 142
    },
    {
      "epoch": 0.6143931256713212,
      "grad_norm": 1.8125334978103638,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.6172,
      "step": 143
    },
    {
      "epoch": 0.6186895810955961,
      "grad_norm": 1.7639728784561157,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.5436,
      "step": 144
    },
    {
      "epoch": 0.6229860365198711,
      "grad_norm": 1.9774454832077026,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.6925,
      "step": 145
    },
    {
      "epoch": 0.6272824919441461,
      "grad_norm": 2.3576865196228027,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.7764,
      "step": 146
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 2.6232426166534424,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.8236,
      "step": 147
    },
    {
      "epoch": 0.635875402792696,
      "grad_norm": 3.0793042182922363,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1341,
      "step": 148
    },
    {
      "epoch": 0.640171858216971,
      "grad_norm": 2.714630603790283,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.7163,
      "step": 149
    },
    {
      "epoch": 0.644468313641246,
      "grad_norm": 2.3378677368164062,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.4872,
      "step": 150
    },
    {
      "epoch": 0.644468313641246,
      "eval_loss": 0.6852211356163025,
      "eval_runtime": 6.4193,
      "eval_samples_per_second": 61.066,
      "eval_steps_per_second": 15.267,
      "step": 150
    },
    {
      "epoch": 0.6487647690655209,
      "grad_norm": 1.0330160856246948,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.4858,
      "step": 151
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 1.379602074623108,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.687,
      "step": 152
    },
    {
      "epoch": 0.6573576799140709,
      "grad_norm": 1.5182565450668335,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.7997,
      "step": 153
    },
    {
      "epoch": 0.6616541353383458,
      "grad_norm": 1.4632099866867065,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.6679,
      "step": 154
    },
    {
      "epoch": 0.6659505907626209,
      "grad_norm": 1.5550950765609741,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.7364,
      "step": 155
    },
    {
      "epoch": 0.6702470461868958,
      "grad_norm": 1.2299954891204834,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.6155,
      "step": 156
    },
    {
      "epoch": 0.6745435016111708,
      "grad_norm": 1.5277807712554932,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.736,
      "step": 157
    },
    {
      "epoch": 0.6788399570354458,
      "grad_norm": 1.2997380495071411,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.6228,
      "step": 158
    },
    {
      "epoch": 0.6831364124597207,
      "grad_norm": 1.328229308128357,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.6507,
      "step": 159
    },
    {
      "epoch": 0.6874328678839957,
      "grad_norm": 1.426734447479248,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.7477,
      "step": 160
    },
    {
      "epoch": 0.6917293233082706,
      "grad_norm": 1.3138563632965088,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.6586,
      "step": 161
    },
    {
      "epoch": 0.6960257787325457,
      "grad_norm": 1.3582028150558472,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.6254,
      "step": 162
    },
    {
      "epoch": 0.7003222341568206,
      "grad_norm": 1.2654234170913696,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.5467,
      "step": 163
    },
    {
      "epoch": 0.7046186895810956,
      "grad_norm": 1.3099442720413208,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.5932,
      "step": 164
    },
    {
      "epoch": 0.7089151450053706,
      "grad_norm": 1.2985515594482422,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.6976,
      "step": 165
    },
    {
      "epoch": 0.7132116004296455,
      "grad_norm": 1.2208131551742554,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.5715,
      "step": 166
    },
    {
      "epoch": 0.7175080558539205,
      "grad_norm": 1.192247986793518,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.589,
      "step": 167
    },
    {
      "epoch": 0.7218045112781954,
      "grad_norm": 1.2376370429992676,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.5438,
      "step": 168
    },
    {
      "epoch": 0.7261009667024705,
      "grad_norm": 1.1177722215652466,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.5245,
      "step": 169
    },
    {
      "epoch": 0.7303974221267454,
      "grad_norm": 1.2670952081680298,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.5502,
      "step": 170
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 1.1297271251678467,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.5725,
      "step": 171
    },
    {
      "epoch": 0.7389903329752954,
      "grad_norm": 1.1443334817886353,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.5625,
      "step": 172
    },
    {
      "epoch": 0.7432867883995704,
      "grad_norm": 1.2655280828475952,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.5852,
      "step": 173
    },
    {
      "epoch": 0.7475832438238453,
      "grad_norm": 1.4558875560760498,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.6581,
      "step": 174
    },
    {
      "epoch": 0.7518796992481203,
      "grad_norm": 1.5853036642074585,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.704,
      "step": 175
    },
    {
      "epoch": 0.7561761546723953,
      "grad_norm": 1.4298464059829712,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.6426,
      "step": 176
    },
    {
      "epoch": 0.7604726100966702,
      "grad_norm": 1.5150086879730225,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.6552,
      "step": 177
    },
    {
      "epoch": 0.7647690655209453,
      "grad_norm": 1.4844558238983154,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.5472,
      "step": 178
    },
    {
      "epoch": 0.7690655209452202,
      "grad_norm": 1.4962722063064575,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.574,
      "step": 179
    },
    {
      "epoch": 0.7733619763694952,
      "grad_norm": 1.7102340459823608,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.7301,
      "step": 180
    },
    {
      "epoch": 0.7776584317937701,
      "grad_norm": 1.661906123161316,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.6172,
      "step": 181
    },
    {
      "epoch": 0.7819548872180451,
      "grad_norm": 1.5829746723175049,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.5403,
      "step": 182
    },
    {
      "epoch": 0.7862513426423201,
      "grad_norm": 1.674821376800537,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.6475,
      "step": 183
    },
    {
      "epoch": 0.790547798066595,
      "grad_norm": 1.914728045463562,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.7731,
      "step": 184
    },
    {
      "epoch": 0.7948442534908701,
      "grad_norm": 1.59696364402771,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.587,
      "step": 185
    },
    {
      "epoch": 0.799140708915145,
      "grad_norm": 1.774781346321106,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.7186,
      "step": 186
    },
    {
      "epoch": 0.80343716433942,
      "grad_norm": 1.749155879020691,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.6595,
      "step": 187
    },
    {
      "epoch": 0.807733619763695,
      "grad_norm": 1.7926386594772339,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.6383,
      "step": 188
    },
    {
      "epoch": 0.8120300751879699,
      "grad_norm": 2.03888201713562,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.7565,
      "step": 189
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 1.7739629745483398,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.5497,
      "step": 190
    },
    {
      "epoch": 0.8206229860365198,
      "grad_norm": 1.7532708644866943,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.5183,
      "step": 191
    },
    {
      "epoch": 0.8249194414607949,
      "grad_norm": 2.1690104007720947,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.5761,
      "step": 192
    },
    {
      "epoch": 0.8292158968850698,
      "grad_norm": 1.905128002166748,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.5924,
      "step": 193
    },
    {
      "epoch": 0.8335123523093448,
      "grad_norm": 2.5123884677886963,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.6385,
      "step": 194
    },
    {
      "epoch": 0.8378088077336198,
      "grad_norm": 2.2678191661834717,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.6812,
      "step": 195
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 2.4593465328216553,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.5903,
      "step": 196
    },
    {
      "epoch": 0.8464017185821697,
      "grad_norm": 2.827348470687866,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.9272,
      "step": 197
    },
    {
      "epoch": 0.8506981740064447,
      "grad_norm": 3.3419406414031982,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.7683,
      "step": 198
    },
    {
      "epoch": 0.8549946294307197,
      "grad_norm": 4.755130767822266,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.7716,
      "step": 199
    },
    {
      "epoch": 0.8592910848549946,
      "grad_norm": 2.8084490299224854,
      "learning_rate": 0.0,
      "loss": 0.3677,
      "step": 200
    },
    {
      "epoch": 0.8592910848549946,
      "eval_loss": 0.642096996307373,
      "eval_runtime": 6.4189,
      "eval_samples_per_second": 61.07,
      "eval_steps_per_second": 15.267,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.3317552734208e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}