{
  "best_metric": 0.8862866759300232,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8429926238145417,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004214963119072708,
      "grad_norm": 40.88788604736328,
      "learning_rate": 1e-05,
      "loss": 7.2132,
      "step": 1
    },
    {
      "epoch": 0.004214963119072708,
      "eval_loss": 8.23657512664795,
      "eval_runtime": 6.5433,
      "eval_samples_per_second": 61.131,
      "eval_steps_per_second": 15.283,
      "step": 1
    },
    {
      "epoch": 0.008429926238145416,
      "grad_norm": 50.758140563964844,
      "learning_rate": 2e-05,
      "loss": 7.8276,
      "step": 2
    },
    {
      "epoch": 0.012644889357218124,
      "grad_norm": 29.915327072143555,
      "learning_rate": 3e-05,
      "loss": 7.1081,
      "step": 3
    },
    {
      "epoch": 0.01685985247629083,
      "grad_norm": 24.2876033782959,
      "learning_rate": 4e-05,
      "loss": 6.7063,
      "step": 4
    },
    {
      "epoch": 0.02107481559536354,
      "grad_norm": 12.967751502990723,
      "learning_rate": 5e-05,
      "loss": 6.1091,
      "step": 5
    },
    {
      "epoch": 0.02528977871443625,
      "grad_norm": 10.861435890197754,
      "learning_rate": 6e-05,
      "loss": 5.8751,
      "step": 6
    },
    {
      "epoch": 0.029504741833508957,
      "grad_norm": 8.685080528259277,
      "learning_rate": 7e-05,
      "loss": 5.4433,
      "step": 7
    },
    {
      "epoch": 0.03371970495258166,
      "grad_norm": 9.255840301513672,
      "learning_rate": 8e-05,
      "loss": 4.8627,
      "step": 8
    },
    {
      "epoch": 0.037934668071654375,
      "grad_norm": 6.5735697746276855,
      "learning_rate": 9e-05,
      "loss": 4.5257,
      "step": 9
    },
    {
      "epoch": 0.04214963119072708,
      "grad_norm": 4.860849380493164,
      "learning_rate": 0.0001,
      "loss": 4.0828,
      "step": 10
    },
    {
      "epoch": 0.04636459430979979,
      "grad_norm": 4.477214336395264,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.1133,
      "step": 11
    },
    {
      "epoch": 0.0505795574288725,
      "grad_norm": 4.521075248718262,
      "learning_rate": 9.997266286704631e-05,
      "loss": 3.6587,
      "step": 12
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 5.390371799468994,
      "learning_rate": 9.993849845741524e-05,
      "loss": 3.6041,
      "step": 13
    },
    {
      "epoch": 0.059009483667017915,
      "grad_norm": 4.206800937652588,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.3321,
      "step": 14
    },
    {
      "epoch": 0.06322444678609063,
      "grad_norm": 3.838667154312134,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.8629,
      "step": 15
    },
    {
      "epoch": 0.06743940990516333,
      "grad_norm": 3.8978095054626465,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.1538,
      "step": 16
    },
    {
      "epoch": 0.07165437302423604,
      "grad_norm": 3.9196834564208984,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.112,
      "step": 17
    },
    {
      "epoch": 0.07586933614330875,
      "grad_norm": 3.479748487472534,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.7872,
      "step": 18
    },
    {
      "epoch": 0.08008429926238145,
      "grad_norm": 4.987894058227539,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.8133,
      "step": 19
    },
    {
      "epoch": 0.08429926238145416,
      "grad_norm": 3.634122133255005,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.619,
      "step": 20
    },
    {
      "epoch": 0.08851422550052687,
      "grad_norm": 7.234653472900391,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.6316,
      "step": 21
    },
    {
      "epoch": 0.09272918861959958,
      "grad_norm": 3.858067750930786,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.747,
      "step": 22
    },
    {
      "epoch": 0.09694415173867228,
      "grad_norm": 3.198962450027466,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2918,
      "step": 23
    },
    {
      "epoch": 0.101159114857745,
      "grad_norm": 3.7858288288116455,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.5512,
      "step": 24
    },
    {
      "epoch": 0.1053740779768177,
      "grad_norm": 3.425922393798828,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.413,
      "step": 25
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 3.4858334064483643,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.4267,
      "step": 26
    },
    {
      "epoch": 0.11380400421496312,
      "grad_norm": 3.4881060123443604,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.4678,
      "step": 27
    },
    {
      "epoch": 0.11801896733403583,
      "grad_norm": 3.4506871700286865,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.1047,
      "step": 28
    },
    {
      "epoch": 0.12223393045310854,
      "grad_norm": 4.144366264343262,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.2491,
      "step": 29
    },
    {
      "epoch": 0.12644889357218125,
      "grad_norm": 3.483288526535034,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.0538,
      "step": 30
    },
    {
      "epoch": 0.13066385669125394,
      "grad_norm": 3.5999341011047363,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.0198,
      "step": 31
    },
    {
      "epoch": 0.13487881981032665,
      "grad_norm": 3.165118455886841,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.767,
      "step": 32
    },
    {
      "epoch": 0.13909378292939936,
      "grad_norm": 4.484847545623779,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.4543,
      "step": 33
    },
    {
      "epoch": 0.14330874604847207,
      "grad_norm": 3.8553950786590576,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.1239,
      "step": 34
    },
    {
      "epoch": 0.1475237091675448,
      "grad_norm": 3.8386988639831543,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.2084,
      "step": 35
    },
    {
      "epoch": 0.1517386722866175,
      "grad_norm": 3.6721622943878174,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.8781,
      "step": 36
    },
    {
      "epoch": 0.1559536354056902,
      "grad_norm": 3.515183687210083,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.679,
      "step": 37
    },
    {
      "epoch": 0.1601685985247629,
      "grad_norm": 3.4947457313537598,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.8612,
      "step": 38
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 3.608725070953369,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.8341,
      "step": 39
    },
    {
      "epoch": 0.16859852476290832,
      "grad_norm": 3.7523324489593506,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.7345,
      "step": 40
    },
    {
      "epoch": 0.17281348788198103,
      "grad_norm": 3.6245577335357666,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.5875,
      "step": 41
    },
    {
      "epoch": 0.17702845100105374,
      "grad_norm": 4.004650592803955,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.7884,
      "step": 42
    },
    {
      "epoch": 0.18124341412012646,
      "grad_norm": 3.693814277648926,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.7175,
      "step": 43
    },
    {
      "epoch": 0.18545837723919917,
      "grad_norm": 3.998577117919922,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5736,
      "step": 44
    },
    {
      "epoch": 0.18967334035827185,
      "grad_norm": 3.8152501583099365,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.444,
      "step": 45
    },
    {
      "epoch": 0.19388830347734456,
      "grad_norm": 4.811203956604004,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.739,
      "step": 46
    },
    {
      "epoch": 0.19810326659641728,
      "grad_norm": 3.8776473999023438,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.5367,
      "step": 47
    },
    {
      "epoch": 0.20231822971549,
      "grad_norm": 4.629638195037842,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.2577,
      "step": 48
    },
    {
      "epoch": 0.2065331928345627,
      "grad_norm": 4.332769393920898,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3416,
      "step": 49
    },
    {
      "epoch": 0.2107481559536354,
      "grad_norm": 5.999655723571777,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.0822,
      "step": 50
    },
    {
      "epoch": 0.2107481559536354,
      "eval_loss": 1.55375337600708,
      "eval_runtime": 6.5162,
      "eval_samples_per_second": 61.386,
      "eval_steps_per_second": 15.346,
      "step": 50
    },
    {
      "epoch": 0.21496311907270813,
      "grad_norm": 4.117035865783691,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.1453,
      "step": 51
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 4.088257789611816,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.9938,
      "step": 52
    },
    {
      "epoch": 0.22339304531085352,
      "grad_norm": 3.515298843383789,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.0991,
      "step": 53
    },
    {
      "epoch": 0.22760800842992623,
      "grad_norm": 3.0887224674224854,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.6539,
      "step": 54
    },
    {
      "epoch": 0.23182297154899895,
      "grad_norm": 2.7529525756835938,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5724,
      "step": 55
    },
    {
      "epoch": 0.23603793466807166,
      "grad_norm": 3.4280331134796143,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.8182,
      "step": 56
    },
    {
      "epoch": 0.24025289778714437,
      "grad_norm": 3.0896048545837402,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.6041,
      "step": 57
    },
    {
      "epoch": 0.24446786090621708,
      "grad_norm": 3.1821529865264893,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.6608,
      "step": 58
    },
    {
      "epoch": 0.24868282402528977,
      "grad_norm": 2.834794759750366,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.6976,
      "step": 59
    },
    {
      "epoch": 0.2528977871443625,
      "grad_norm": 2.920761823654175,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.5668,
      "step": 60
    },
    {
      "epoch": 0.2571127502634352,
      "grad_norm": 2.5749521255493164,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.364,
      "step": 61
    },
    {
      "epoch": 0.2613277133825079,
      "grad_norm": 3.0735316276550293,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.477,
      "step": 62
    },
    {
      "epoch": 0.2655426765015806,
      "grad_norm": 2.9889707565307617,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.494,
      "step": 63
    },
    {
      "epoch": 0.2697576396206533,
      "grad_norm": 2.7958595752716064,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4037,
      "step": 64
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 2.937525749206543,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.5477,
      "step": 65
    },
    {
      "epoch": 0.2781875658587987,
      "grad_norm": 2.924337863922119,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3599,
      "step": 66
    },
    {
      "epoch": 0.28240252897787144,
      "grad_norm": 3.3123669624328613,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.4913,
      "step": 67
    },
    {
      "epoch": 0.28661749209694415,
      "grad_norm": 2.940045118331909,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.2275,
      "step": 68
    },
    {
      "epoch": 0.29083245521601686,
      "grad_norm": 3.051992654800415,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4199,
      "step": 69
    },
    {
      "epoch": 0.2950474183350896,
      "grad_norm": 3.1050009727478027,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4687,
      "step": 70
    },
    {
      "epoch": 0.2992623814541623,
      "grad_norm": 3.2320985794067383,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.2468,
      "step": 71
    },
    {
      "epoch": 0.303477344573235,
      "grad_norm": 3.6176812648773193,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.5032,
      "step": 72
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 3.5661792755126953,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.5375,
      "step": 73
    },
    {
      "epoch": 0.3119072708113804,
      "grad_norm": 2.896869421005249,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.1279,
      "step": 74
    },
    {
      "epoch": 0.31612223393045313,
      "grad_norm": 3.137770414352417,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.2923,
      "step": 75
    },
    {
      "epoch": 0.3203371970495258,
      "grad_norm": 3.3340225219726562,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.3779,
      "step": 76
    },
    {
      "epoch": 0.3245521601685985,
      "grad_norm": 3.598785638809204,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.4817,
      "step": 77
    },
    {
      "epoch": 0.3287671232876712,
      "grad_norm": 3.4752955436706543,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.1593,
      "step": 78
    },
    {
      "epoch": 0.3329820864067439,
      "grad_norm": 3.72232723236084,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3676,
      "step": 79
    },
    {
      "epoch": 0.33719704952581664,
      "grad_norm": 3.4522814750671387,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.1664,
      "step": 80
    },
    {
      "epoch": 0.34141201264488935,
      "grad_norm": 3.5752956867218018,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.1711,
      "step": 81
    },
    {
      "epoch": 0.34562697576396206,
      "grad_norm": 3.416863203048706,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3242,
      "step": 82
    },
    {
      "epoch": 0.3498419388830348,
      "grad_norm": 4.146493434906006,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.279,
      "step": 83
    },
    {
      "epoch": 0.3540569020021075,
      "grad_norm": 3.5011348724365234,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.2263,
      "step": 84
    },
    {
      "epoch": 0.3582718651211802,
      "grad_norm": 3.209510087966919,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.9726,
      "step": 85
    },
    {
      "epoch": 0.3624868282402529,
      "grad_norm": 3.6212852001190186,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.1824,
      "step": 86
    },
    {
      "epoch": 0.3667017913593256,
      "grad_norm": 3.9444656372070312,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.4081,
      "step": 87
    },
    {
      "epoch": 0.37091675447839834,
      "grad_norm": 3.8729257583618164,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.0846,
      "step": 88
    },
    {
      "epoch": 0.37513171759747105,
      "grad_norm": 3.9269540309906006,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2498,
      "step": 89
    },
    {
      "epoch": 0.3793466807165437,
      "grad_norm": 3.8559165000915527,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.1026,
      "step": 90
    },
    {
      "epoch": 0.3835616438356164,
      "grad_norm": 3.275670051574707,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.0651,
      "step": 91
    },
    {
      "epoch": 0.38777660695468913,
      "grad_norm": 3.741572856903076,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.0514,
      "step": 92
    },
    {
      "epoch": 0.39199157007376184,
      "grad_norm": 3.4798471927642822,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.8789,
      "step": 93
    },
    {
      "epoch": 0.39620653319283455,
      "grad_norm": 3.079193353652954,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.6738,
      "step": 94
    },
    {
      "epoch": 0.40042149631190727,
      "grad_norm": 4.083731174468994,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1807,
      "step": 95
    },
    {
      "epoch": 0.40463645943098,
      "grad_norm": 3.9977304935455322,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.9,
      "step": 96
    },
    {
      "epoch": 0.4088514225500527,
      "grad_norm": 3.484790086746216,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.6684,
      "step": 97
    },
    {
      "epoch": 0.4130663856691254,
      "grad_norm": 4.130280494689941,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.8525,
      "step": 98
    },
    {
      "epoch": 0.4172813487881981,
      "grad_norm": 4.026912689208984,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9243,
      "step": 99
    },
    {
      "epoch": 0.4214963119072708,
      "grad_norm": 4.30971622467041,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.9518,
      "step": 100
    },
    {
      "epoch": 0.4214963119072708,
      "eval_loss": 1.0880745649337769,
      "eval_runtime": 6.533,
      "eval_samples_per_second": 61.227,
      "eval_steps_per_second": 15.307,
      "step": 100
    },
    {
      "epoch": 0.42571127502634354,
      "grad_norm": 3.386011838912964,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2544,
      "step": 101
    },
    {
      "epoch": 0.42992623814541625,
      "grad_norm": 3.37831974029541,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.3462,
      "step": 102
    },
    {
      "epoch": 0.43414120126448896,
      "grad_norm": 3.2945053577423096,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3816,
      "step": 103
    },
    {
      "epoch": 0.4383561643835616,
      "grad_norm": 3.1703274250030518,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5241,
      "step": 104
    },
    {
      "epoch": 0.44257112750263433,
      "grad_norm": 2.873652696609497,
      "learning_rate": 5e-05,
      "loss": 1.2792,
      "step": 105
    },
    {
      "epoch": 0.44678609062170704,
      "grad_norm": 2.8146026134490967,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2281,
      "step": 106
    },
    {
      "epoch": 0.45100105374077976,
      "grad_norm": 3.0346248149871826,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.3371,
      "step": 107
    },
    {
      "epoch": 0.45521601685985247,
      "grad_norm": 3.360595941543579,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.2377,
      "step": 108
    },
    {
      "epoch": 0.4594309799789252,
      "grad_norm": 3.307637929916382,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.2842,
      "step": 109
    },
    {
      "epoch": 0.4636459430979979,
      "grad_norm": 3.021942377090454,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1708,
      "step": 110
    },
    {
      "epoch": 0.4678609062170706,
      "grad_norm": 3.195174217224121,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.402,
      "step": 111
    },
    {
      "epoch": 0.4720758693361433,
      "grad_norm": 2.8297829627990723,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.141,
      "step": 112
    },
    {
      "epoch": 0.47629083245521603,
      "grad_norm": 3.01725172996521,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.3173,
      "step": 113
    },
    {
      "epoch": 0.48050579557428874,
      "grad_norm": 2.8402228355407715,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.0624,
      "step": 114
    },
    {
      "epoch": 0.48472075869336145,
      "grad_norm": 3.3632452487945557,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1446,
      "step": 115
    },
    {
      "epoch": 0.48893572181243417,
      "grad_norm": 3.1667258739471436,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.2906,
      "step": 116
    },
    {
      "epoch": 0.4931506849315068,
      "grad_norm": 2.796734094619751,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.9742,
      "step": 117
    },
    {
      "epoch": 0.49736564805057953,
      "grad_norm": 3.2416276931762695,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2833,
      "step": 118
    },
    {
      "epoch": 0.5015806111696522,
      "grad_norm": 3.4378788471221924,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2747,
      "step": 119
    },
    {
      "epoch": 0.505795574288725,
      "grad_norm": 2.8093643188476562,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.9897,
      "step": 120
    },
    {
      "epoch": 0.5100105374077977,
      "grad_norm": 3.5080909729003906,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.0083,
      "step": 121
    },
    {
      "epoch": 0.5142255005268704,
      "grad_norm": 2.8501391410827637,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2174,
      "step": 122
    },
    {
      "epoch": 0.5184404636459431,
      "grad_norm": 3.107459783554077,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.168,
      "step": 123
    },
    {
      "epoch": 0.5226554267650158,
      "grad_norm": 3.094845771789551,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.234,
      "step": 124
    },
    {
      "epoch": 0.5268703898840885,
      "grad_norm": 2.82763934135437,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.0466,
      "step": 125
    },
    {
      "epoch": 0.5310853530031612,
      "grad_norm": 3.1835439205169678,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.9544,
      "step": 126
    },
    {
      "epoch": 0.5353003161222339,
      "grad_norm": 2.8917644023895264,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.9952,
      "step": 127
    },
    {
      "epoch": 0.5395152792413066,
      "grad_norm": 3.3423712253570557,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.2208,
      "step": 128
    },
    {
      "epoch": 0.5437302423603794,
      "grad_norm": 3.1876165866851807,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0151,
      "step": 129
    },
    {
      "epoch": 0.547945205479452,
      "grad_norm": 3.201247215270996,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2059,
      "step": 130
    },
    {
      "epoch": 0.5521601685985248,
      "grad_norm": 2.7853102684020996,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.9169,
      "step": 131
    },
    {
      "epoch": 0.5563751317175974,
      "grad_norm": 3.1830129623413086,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.8751,
      "step": 132
    },
    {
      "epoch": 0.5605900948366702,
      "grad_norm": 3.2747745513916016,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.0821,
      "step": 133
    },
    {
      "epoch": 0.5648050579557429,
      "grad_norm": 4.344180107116699,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.15,
      "step": 134
    },
    {
      "epoch": 0.5690200210748156,
      "grad_norm": 3.5275580883026123,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.031,
      "step": 135
    },
    {
      "epoch": 0.5732349841938883,
      "grad_norm": 3.548412799835205,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0051,
      "step": 136
    },
    {
      "epoch": 0.5774499473129611,
      "grad_norm": 3.4377214908599854,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.1083,
      "step": 137
    },
    {
      "epoch": 0.5816649104320337,
      "grad_norm": 3.2830352783203125,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.8664,
      "step": 138
    },
    {
      "epoch": 0.5858798735511064,
      "grad_norm": 2.5875048637390137,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.6063,
      "step": 139
    },
    {
      "epoch": 0.5900948366701791,
      "grad_norm": 3.1425721645355225,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.7587,
      "step": 140
    },
    {
      "epoch": 0.5943097997892518,
      "grad_norm": 3.090977668762207,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.7278,
      "step": 141
    },
    {
      "epoch": 0.5985247629083246,
      "grad_norm": 3.6454904079437256,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.038,
      "step": 142
    },
    {
      "epoch": 0.6027397260273972,
      "grad_norm": 3.254973888397217,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.8658,
      "step": 143
    },
    {
      "epoch": 0.60695468914647,
      "grad_norm": 3.190573215484619,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.8059,
      "step": 144
    },
    {
      "epoch": 0.6111696522655427,
      "grad_norm": 3.104487180709839,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.637,
      "step": 145
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 3.4784772396087646,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.7202,
      "step": 146
    },
    {
      "epoch": 0.6195995785036881,
      "grad_norm": 3.98059344291687,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.9319,
      "step": 147
    },
    {
      "epoch": 0.6238145416227608,
      "grad_norm": 3.632870674133301,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.8554,
      "step": 148
    },
    {
      "epoch": 0.6280295047418335,
      "grad_norm": 4.012639045715332,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.7925,
      "step": 149
    },
    {
      "epoch": 0.6322444678609063,
      "grad_norm": 5.00639533996582,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.8196,
      "step": 150
    },
    {
      "epoch": 0.6322444678609063,
      "eval_loss": 0.9291651844978333,
      "eval_runtime": 6.5338,
      "eval_samples_per_second": 61.22,
      "eval_steps_per_second": 15.305,
      "step": 150
    },
    {
      "epoch": 0.6364594309799789,
      "grad_norm": 2.848217487335205,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.1926,
      "step": 151
    },
    {
      "epoch": 0.6406743940990516,
      "grad_norm": 3.1672067642211914,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.1999,
      "step": 152
    },
    {
      "epoch": 0.6448893572181243,
      "grad_norm": 3.413862943649292,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.2104,
      "step": 153
    },
    {
      "epoch": 0.649104320337197,
      "grad_norm": 3.187894105911255,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.2231,
      "step": 154
    },
    {
      "epoch": 0.6533192834562698,
      "grad_norm": 3.935333251953125,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1542,
      "step": 155
    },
    {
      "epoch": 0.6575342465753424,
      "grad_norm": 3.1985912322998047,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.0731,
      "step": 156
    },
    {
      "epoch": 0.6617492096944152,
      "grad_norm": 3.2511754035949707,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.0841,
      "step": 157
    },
    {
      "epoch": 0.6659641728134879,
      "grad_norm": 2.8101887702941895,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1023,
      "step": 158
    },
    {
      "epoch": 0.6701791359325606,
      "grad_norm": 3.1054890155792236,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.0365,
      "step": 159
    },
    {
      "epoch": 0.6743940990516333,
      "grad_norm": 3.1497960090637207,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.2406,
      "step": 160
    },
    {
      "epoch": 0.678609062170706,
      "grad_norm": 3.1722278594970703,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2494,
      "step": 161
    },
    {
      "epoch": 0.6828240252897787,
      "grad_norm": 2.920506238937378,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.099,
      "step": 162
    },
    {
      "epoch": 0.6870389884088515,
      "grad_norm": 2.694607973098755,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.9515,
      "step": 163
    },
    {
      "epoch": 0.6912539515279241,
      "grad_norm": 3.49603533744812,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0193,
      "step": 164
    },
    {
      "epoch": 0.6954689146469969,
      "grad_norm": 3.4482979774475098,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.2229,
      "step": 165
    },
    {
      "epoch": 0.6996838777660696,
      "grad_norm": 3.1300268173217773,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.0126,
      "step": 166
    },
    {
      "epoch": 0.7038988408851422,
      "grad_norm": 3.126187324523926,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.1116,
      "step": 167
    },
    {
      "epoch": 0.708113804004215,
      "grad_norm": 3.690009355545044,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0759,
      "step": 168
    },
    {
      "epoch": 0.7123287671232876,
      "grad_norm": 3.056969404220581,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.9331,
      "step": 169
    },
    {
      "epoch": 0.7165437302423604,
      "grad_norm": 3.0779786109924316,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.9349,
      "step": 170
    },
    {
      "epoch": 0.7207586933614331,
      "grad_norm": 3.3711771965026855,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.8955,
      "step": 171
    },
    {
      "epoch": 0.7249736564805058,
      "grad_norm": 3.033543825149536,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.8737,
      "step": 172
    },
    {
      "epoch": 0.7291886195995785,
      "grad_norm": 3.4564316272735596,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.9108,
      "step": 173
    },
    {
      "epoch": 0.7334035827186512,
      "grad_norm": 2.8985610008239746,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0421,
      "step": 174
    },
    {
      "epoch": 0.7376185458377239,
      "grad_norm": 3.2784736156463623,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0429,
      "step": 175
    },
    {
      "epoch": 0.7418335089567967,
      "grad_norm": 2.9215872287750244,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9404,
      "step": 176
    },
    {
      "epoch": 0.7460484720758693,
      "grad_norm": 3.2456657886505127,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9858,
      "step": 177
    },
    {
      "epoch": 0.7502634351949421,
      "grad_norm": 3.1223134994506836,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.9826,
      "step": 178
    },
    {
      "epoch": 0.7544783983140148,
      "grad_norm": 3.4771385192871094,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.14,
      "step": 179
    },
    {
      "epoch": 0.7586933614330874,
      "grad_norm": 2.9304895401000977,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.7536,
      "step": 180
    },
    {
      "epoch": 0.7629083245521602,
      "grad_norm": 3.4102566242218018,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0803,
      "step": 181
    },
    {
      "epoch": 0.7671232876712328,
      "grad_norm": 3.1297965049743652,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.9483,
      "step": 182
    },
    {
      "epoch": 0.7713382507903056,
      "grad_norm": 3.374161720275879,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.9244,
      "step": 183
    },
    {
      "epoch": 0.7755532139093783,
      "grad_norm": 3.2307045459747314,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.9322,
      "step": 184
    },
    {
      "epoch": 0.779768177028451,
      "grad_norm": 3.7904913425445557,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.1284,
      "step": 185
    },
    {
      "epoch": 0.7839831401475237,
      "grad_norm": 3.5597307682037354,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.9433,
      "step": 186
    },
    {
      "epoch": 0.7881981032665965,
      "grad_norm": 3.9056553840637207,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0799,
      "step": 187
    },
    {
      "epoch": 0.7924130663856691,
      "grad_norm": 3.586214780807495,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.8109,
      "step": 188
    },
    {
      "epoch": 0.7966280295047419,
      "grad_norm": 3.483254909515381,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8241,
      "step": 189
    },
    {
      "epoch": 0.8008429926238145,
      "grad_norm": 3.586808919906616,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.835,
      "step": 190
    },
    {
      "epoch": 0.8050579557428873,
      "grad_norm": 3.5435309410095215,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.7889,
      "step": 191
    },
    {
      "epoch": 0.80927291886196,
      "grad_norm": 3.3583855628967285,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.8096,
      "step": 192
    },
    {
      "epoch": 0.8134878819810326,
      "grad_norm": 3.4656455516815186,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.9019,
      "step": 193
    },
    {
      "epoch": 0.8177028451001054,
      "grad_norm": 3.1683249473571777,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.8147,
      "step": 194
    },
    {
      "epoch": 0.821917808219178,
      "grad_norm": 4.181645393371582,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.8447,
      "step": 195
    },
    {
      "epoch": 0.8261327713382508,
      "grad_norm": 3.5625126361846924,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.5977,
      "step": 196
    },
    {
      "epoch": 0.8303477344573235,
      "grad_norm": 3.9427285194396973,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.702,
      "step": 197
    },
    {
      "epoch": 0.8345626975763962,
      "grad_norm": 4.489522933959961,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.7069,
      "step": 198
    },
    {
      "epoch": 0.8387776606954689,
      "grad_norm": 5.069953918457031,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.1038,
      "step": 199
    },
    {
      "epoch": 0.8429926238145417,
      "grad_norm": 5.032002925872803,
      "learning_rate": 0.0,
      "loss": 0.8499,
      "step": 200
    },
    {
      "epoch": 0.8429926238145417,
      "eval_loss": 0.8862866759300232,
      "eval_runtime": 6.5371,
      "eval_samples_per_second": 61.19,
      "eval_steps_per_second": 15.297,
      "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
}