{
  "best_metric": 1.0638600587844849,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.16266775111834078,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008133387555917039,
      "grad_norm": 0.37575653195381165,
      "learning_rate": 1e-05,
      "loss": 1.0661,
      "step": 1
    },
    {
      "epoch": 0.0008133387555917039,
      "eval_loss": 1.7443177700042725,
      "eval_runtime": 87.8063,
      "eval_samples_per_second": 23.586,
      "eval_steps_per_second": 5.899,
      "step": 1
    },
    {
      "epoch": 0.0016266775111834079,
      "grad_norm": 0.4422896206378937,
      "learning_rate": 2e-05,
      "loss": 1.1632,
      "step": 2
    },
    {
      "epoch": 0.0024400162667751117,
      "grad_norm": 0.4560563862323761,
      "learning_rate": 3e-05,
      "loss": 1.1966,
      "step": 3
    },
    {
      "epoch": 0.0032533550223668157,
      "grad_norm": 0.4433326721191406,
      "learning_rate": 4e-05,
      "loss": 1.119,
      "step": 4
    },
    {
      "epoch": 0.004066693777958519,
      "grad_norm": 0.48522961139678955,
      "learning_rate": 5e-05,
      "loss": 1.2333,
      "step": 5
    },
    {
      "epoch": 0.004880032533550223,
      "grad_norm": 0.4333650767803192,
      "learning_rate": 6e-05,
      "loss": 1.0603,
      "step": 6
    },
    {
      "epoch": 0.0056933712891419274,
      "grad_norm": 0.4161796569824219,
      "learning_rate": 7e-05,
      "loss": 1.2085,
      "step": 7
    },
    {
      "epoch": 0.0065067100447336315,
      "grad_norm": 0.4600462317466736,
      "learning_rate": 8e-05,
      "loss": 1.2437,
      "step": 8
    },
    {
      "epoch": 0.0073200488003253355,
      "grad_norm": 0.4632268249988556,
      "learning_rate": 9e-05,
      "loss": 1.1157,
      "step": 9
    },
    {
      "epoch": 0.008133387555917039,
      "grad_norm": 0.4042969048023224,
      "learning_rate": 0.0001,
      "loss": 1.1244,
      "step": 10
    },
    {
      "epoch": 0.008946726311508743,
      "grad_norm": 0.47048935294151306,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.2068,
      "step": 11
    },
    {
      "epoch": 0.009760065067100447,
      "grad_norm": 0.5726286172866821,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.1264,
      "step": 12
    },
    {
      "epoch": 0.01057340382269215,
      "grad_norm": 0.5205187797546387,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.1207,
      "step": 13
    },
    {
      "epoch": 0.011386742578283855,
      "grad_norm": 0.4981187582015991,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.1173,
      "step": 14
    },
    {
      "epoch": 0.012200081333875559,
      "grad_norm": 0.44847434759140015,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.1536,
      "step": 15
    },
    {
      "epoch": 0.013013420089467263,
      "grad_norm": 0.5005471706390381,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.167,
      "step": 16
    },
    {
      "epoch": 0.013826758845058967,
      "grad_norm": 0.4815964102745056,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1728,
      "step": 17
    },
    {
      "epoch": 0.014640097600650671,
      "grad_norm": 0.49710598587989807,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.2169,
      "step": 18
    },
    {
      "epoch": 0.015453436356242375,
      "grad_norm": 0.5447195768356323,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.1496,
      "step": 19
    },
    {
      "epoch": 0.016266775111834077,
      "grad_norm": 0.5534111857414246,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.0239,
      "step": 20
    },
    {
      "epoch": 0.01708011386742578,
      "grad_norm": 0.5421622395515442,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1137,
      "step": 21
    },
    {
      "epoch": 0.017893452623017485,
      "grad_norm": 0.5853100419044495,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.239,
      "step": 22
    },
    {
      "epoch": 0.01870679137860919,
      "grad_norm": 0.5909392833709717,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.115,
      "step": 23
    },
    {
      "epoch": 0.019520130134200894,
      "grad_norm": 0.6172589659690857,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1728,
      "step": 24
    },
    {
      "epoch": 0.020333468889792598,
      "grad_norm": 0.5866591930389404,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.1261,
      "step": 25
    },
    {
      "epoch": 0.0211468076453843,
      "grad_norm": 0.619943380355835,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.0471,
      "step": 26
    },
    {
      "epoch": 0.021960146400976006,
      "grad_norm": 0.7176154255867004,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.2538,
      "step": 27
    },
    {
      "epoch": 0.02277348515656771,
      "grad_norm": 0.7767391204833984,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.1681,
      "step": 28
    },
    {
      "epoch": 0.023586823912159414,
      "grad_norm": 0.7608658075332642,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0935,
      "step": 29
    },
    {
      "epoch": 0.024400162667751118,
      "grad_norm": 0.9368367791175842,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3343,
      "step": 30
    },
    {
      "epoch": 0.025213501423342822,
      "grad_norm": 0.8900392651557922,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.2107,
      "step": 31
    },
    {
      "epoch": 0.026026840178934526,
      "grad_norm": 1.1127827167510986,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.2866,
      "step": 32
    },
    {
      "epoch": 0.02684017893452623,
      "grad_norm": 0.9749177098274231,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.0908,
      "step": 33
    },
    {
      "epoch": 0.027653517690117934,
      "grad_norm": 1.0239723920822144,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0952,
      "step": 34
    },
    {
      "epoch": 0.028466856445709638,
      "grad_norm": 1.1572667360305786,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.0909,
      "step": 35
    },
    {
      "epoch": 0.029280195201301342,
      "grad_norm": 1.2718197107315063,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.084,
      "step": 36
    },
    {
      "epoch": 0.030093533956893046,
      "grad_norm": 1.524453043937683,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3012,
      "step": 37
    },
    {
      "epoch": 0.03090687271248475,
      "grad_norm": 1.3463082313537598,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.0689,
      "step": 38
    },
    {
      "epoch": 0.03172021146807645,
      "grad_norm": 1.6751750707626343,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.0738,
      "step": 39
    },
    {
      "epoch": 0.032533550223668155,
      "grad_norm": 1.779761552810669,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.1019,
      "step": 40
    },
    {
      "epoch": 0.03334688897925986,
      "grad_norm": 1.7344919443130493,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1088,
      "step": 41
    },
    {
      "epoch": 0.03416022773485156,
      "grad_norm": 2.0268380641937256,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.1033,
      "step": 42
    },
    {
      "epoch": 0.03497356649044327,
      "grad_norm": 1.6721314191818237,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.815,
      "step": 43
    },
    {
      "epoch": 0.03578690524603497,
      "grad_norm": 1.7108746767044067,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.0455,
      "step": 44
    },
    {
      "epoch": 0.036600244001626675,
      "grad_norm": 1.9779719114303589,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.1687,
      "step": 45
    },
    {
      "epoch": 0.03741358275721838,
      "grad_norm": 2.065678834915161,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1324,
      "step": 46
    },
    {
      "epoch": 0.03822692151281008,
      "grad_norm": 2.068726062774658,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.0247,
      "step": 47
    },
    {
      "epoch": 0.03904026026840179,
      "grad_norm": 3.3183324337005615,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.9841,
      "step": 48
    },
    {
      "epoch": 0.03985359902399349,
      "grad_norm": 2.552274703979492,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.9452,
      "step": 49
    },
    {
      "epoch": 0.040666937779585195,
      "grad_norm": 4.016664981842041,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.497,
      "step": 50
    },
    {
      "epoch": 0.040666937779585195,
      "eval_loss": 1.2598134279251099,
      "eval_runtime": 89.0386,
      "eval_samples_per_second": 23.26,
      "eval_steps_per_second": 5.818,
      "step": 50
    },
    {
      "epoch": 0.0414802765351769,
      "grad_norm": 1.2362276315689087,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.1788,
      "step": 51
    },
    {
      "epoch": 0.0422936152907686,
      "grad_norm": 1.0195097923278809,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2343,
      "step": 52
    },
    {
      "epoch": 0.04310695404636031,
      "grad_norm": 0.7455459237098694,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.1409,
      "step": 53
    },
    {
      "epoch": 0.04392029280195201,
      "grad_norm": 0.5190573930740356,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.1568,
      "step": 54
    },
    {
      "epoch": 0.044733631557543715,
      "grad_norm": 0.41744327545166016,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.1231,
      "step": 55
    },
    {
      "epoch": 0.04554697031313542,
      "grad_norm": 0.40008601546287537,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.217,
      "step": 56
    },
    {
      "epoch": 0.046360309068727124,
      "grad_norm": 0.3854612410068512,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.0603,
      "step": 57
    },
    {
      "epoch": 0.04717364782431883,
      "grad_norm": 0.41007179021835327,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.0678,
      "step": 58
    },
    {
      "epoch": 0.04798698657991053,
      "grad_norm": 0.37711963057518005,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.0819,
      "step": 59
    },
    {
      "epoch": 0.048800325335502236,
      "grad_norm": 0.41456639766693115,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.0696,
      "step": 60
    },
    {
      "epoch": 0.04961366409109394,
      "grad_norm": 0.42507752776145935,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.0832,
      "step": 61
    },
    {
      "epoch": 0.050427002846685644,
      "grad_norm": 0.40908849239349365,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.0709,
      "step": 62
    },
    {
      "epoch": 0.05124034160227735,
      "grad_norm": 0.3968433439731598,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.2347,
      "step": 63
    },
    {
      "epoch": 0.05205368035786905,
      "grad_norm": 0.39582815766334534,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.1774,
      "step": 64
    },
    {
      "epoch": 0.052867019113460756,
      "grad_norm": 0.4031344950199127,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.1364,
      "step": 65
    },
    {
      "epoch": 0.05368035786905246,
      "grad_norm": 0.40360987186431885,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.1707,
      "step": 66
    },
    {
      "epoch": 0.054493696624644164,
      "grad_norm": 0.42251041531562805,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.1812,
      "step": 67
    },
    {
      "epoch": 0.05530703538023587,
      "grad_norm": 0.477252334356308,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.1249,
      "step": 68
    },
    {
      "epoch": 0.05612037413582757,
      "grad_norm": 0.45225009322166443,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.0997,
      "step": 69
    },
    {
      "epoch": 0.056933712891419276,
      "grad_norm": 0.4852558970451355,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.1717,
      "step": 70
    },
    {
      "epoch": 0.05774705164701098,
      "grad_norm": 0.48301947116851807,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.1074,
      "step": 71
    },
    {
      "epoch": 0.058560390402602684,
      "grad_norm": 0.5047841668128967,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1409,
      "step": 72
    },
    {
      "epoch": 0.05937372915819439,
      "grad_norm": 0.5363867878913879,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.1792,
      "step": 73
    },
    {
      "epoch": 0.06018706791378609,
      "grad_norm": 0.5447888970375061,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.232,
      "step": 74
    },
    {
      "epoch": 0.061000406669377796,
      "grad_norm": 0.5834814310073853,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1886,
      "step": 75
    },
    {
      "epoch": 0.0618137454249695,
      "grad_norm": 0.6308369040489197,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.1922,
      "step": 76
    },
    {
      "epoch": 0.0626270841805612,
      "grad_norm": 0.6759172081947327,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.1149,
      "step": 77
    },
    {
      "epoch": 0.0634404229361529,
      "grad_norm": 0.6794350743293762,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.2673,
      "step": 78
    },
    {
      "epoch": 0.06425376169174461,
      "grad_norm": 0.7519348859786987,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1591,
      "step": 79
    },
    {
      "epoch": 0.06506710044733631,
      "grad_norm": 0.8293099999427795,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.2551,
      "step": 80
    },
    {
      "epoch": 0.06588043920292802,
      "grad_norm": 0.8560991287231445,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.266,
      "step": 81
    },
    {
      "epoch": 0.06669377795851972,
      "grad_norm": 0.9848523736000061,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.219,
      "step": 82
    },
    {
      "epoch": 0.06750711671411143,
      "grad_norm": 0.988909900188446,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2014,
      "step": 83
    },
    {
      "epoch": 0.06832045546970313,
      "grad_norm": 0.9696521162986755,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.0757,
      "step": 84
    },
    {
      "epoch": 0.06913379422529484,
      "grad_norm": 1.0835824012756348,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.0505,
      "step": 85
    },
    {
      "epoch": 0.06994713298088653,
      "grad_norm": 1.0682529211044312,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.1699,
      "step": 86
    },
    {
      "epoch": 0.07076047173647824,
      "grad_norm": 1.308504581451416,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.1774,
      "step": 87
    },
    {
      "epoch": 0.07157381049206994,
      "grad_norm": 1.1644800901412964,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.9633,
      "step": 88
    },
    {
      "epoch": 0.07238714924766165,
      "grad_norm": 1.180165410041809,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.9451,
      "step": 89
    },
    {
      "epoch": 0.07320048800325335,
      "grad_norm": 1.3333361148834229,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.0087,
      "step": 90
    },
    {
      "epoch": 0.07401382675884506,
      "grad_norm": 1.3818581104278564,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.9659,
      "step": 91
    },
    {
      "epoch": 0.07482716551443676,
      "grad_norm": 1.3823522329330444,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.8774,
      "step": 92
    },
    {
      "epoch": 0.07564050427002847,
      "grad_norm": 1.890332818031311,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.2164,
      "step": 93
    },
    {
      "epoch": 0.07645384302562017,
      "grad_norm": 2.019684076309204,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.1214,
      "step": 94
    },
    {
      "epoch": 0.07726718178121188,
      "grad_norm": 1.5795767307281494,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.8816,
      "step": 95
    },
    {
      "epoch": 0.07808052053680357,
      "grad_norm": 1.788567066192627,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.0409,
      "step": 96
    },
    {
      "epoch": 0.07889385929239529,
      "grad_norm": 1.9877651929855347,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.9022,
      "step": 97
    },
    {
      "epoch": 0.07970719804798698,
      "grad_norm": 1.6864415407180786,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.7545,
      "step": 98
    },
    {
      "epoch": 0.0805205368035787,
      "grad_norm": 2.085402250289917,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9831,
      "step": 99
    },
    {
      "epoch": 0.08133387555917039,
      "grad_norm": 3.7699906826019287,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.1156,
      "step": 100
    },
    {
      "epoch": 0.08133387555917039,
      "eval_loss": 1.187030553817749,
      "eval_runtime": 88.9107,
      "eval_samples_per_second": 23.293,
      "eval_steps_per_second": 5.826,
      "step": 100
    },
    {
      "epoch": 0.0821472143147621,
      "grad_norm": 0.7749425172805786,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.0667,
      "step": 101
    },
    {
      "epoch": 0.0829605530703538,
      "grad_norm": 0.9096212387084961,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2133,
      "step": 102
    },
    {
      "epoch": 0.08377389182594551,
      "grad_norm": 0.6741483807563782,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.1435,
      "step": 103
    },
    {
      "epoch": 0.0845872305815372,
      "grad_norm": 0.5287403464317322,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.1874,
      "step": 104
    },
    {
      "epoch": 0.08540056933712892,
      "grad_norm": 0.44233056902885437,
      "learning_rate": 5e-05,
      "loss": 1.1832,
      "step": 105
    },
    {
      "epoch": 0.08621390809272061,
      "grad_norm": 0.3863862454891205,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.0603,
      "step": 106
    },
    {
      "epoch": 0.08702724684831233,
      "grad_norm": 0.3736167550086975,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.113,
      "step": 107
    },
    {
      "epoch": 0.08784058560390402,
      "grad_norm": 0.34162628650665283,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.0863,
      "step": 108
    },
    {
      "epoch": 0.08865392435949573,
      "grad_norm": 0.3608766198158264,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.1283,
      "step": 109
    },
    {
      "epoch": 0.08946726311508743,
      "grad_norm": 0.3732871115207672,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.0999,
      "step": 110
    },
    {
      "epoch": 0.09028060187067914,
      "grad_norm": 0.3927890956401825,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.1831,
      "step": 111
    },
    {
      "epoch": 0.09109394062627084,
      "grad_norm": 0.38079723715782166,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1255,
      "step": 112
    },
    {
      "epoch": 0.09190727938186255,
      "grad_norm": 0.41005364060401917,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.1393,
      "step": 113
    },
    {
      "epoch": 0.09272061813745425,
      "grad_norm": 0.3990439474582672,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.0819,
      "step": 114
    },
    {
      "epoch": 0.09353395689304596,
      "grad_norm": 0.4078267812728882,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1341,
      "step": 115
    },
    {
      "epoch": 0.09434729564863766,
      "grad_norm": 0.4114376902580261,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.092,
      "step": 116
    },
    {
      "epoch": 0.09516063440422937,
      "grad_norm": 0.42537808418273926,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.1877,
      "step": 117
    },
    {
      "epoch": 0.09597397315982106,
      "grad_norm": 0.44751498103141785,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2194,
      "step": 118
    },
    {
      "epoch": 0.09678731191541277,
      "grad_norm": 0.4612998962402344,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.1079,
      "step": 119
    },
    {
      "epoch": 0.09760065067100447,
      "grad_norm": 0.49598780274391174,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.2836,
      "step": 120
    },
    {
      "epoch": 0.09841398942659618,
      "grad_norm": 0.48062801361083984,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.1549,
      "step": 121
    },
    {
      "epoch": 0.09922732818218788,
      "grad_norm": 0.4495367407798767,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.0882,
      "step": 122
    },
    {
      "epoch": 0.10004066693777959,
      "grad_norm": 0.4912242591381073,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2911,
      "step": 123
    },
    {
      "epoch": 0.10085400569337129,
      "grad_norm": 0.5084016919136047,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1159,
      "step": 124
    },
    {
      "epoch": 0.101667344448963,
      "grad_norm": 0.532623827457428,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.2222,
      "step": 125
    },
    {
      "epoch": 0.1024806832045547,
      "grad_norm": 0.5241797566413879,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.2952,
      "step": 126
    },
    {
      "epoch": 0.1032940219601464,
      "grad_norm": 0.5172199010848999,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.1479,
      "step": 127
    },
    {
      "epoch": 0.1041073607157381,
      "grad_norm": 0.6069467663764954,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.1961,
      "step": 128
    },
    {
      "epoch": 0.10492069947132981,
      "grad_norm": 0.6096017956733704,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0515,
      "step": 129
    },
    {
      "epoch": 0.10573403822692151,
      "grad_norm": 0.7095544934272766,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.1301,
      "step": 130
    },
    {
      "epoch": 0.10654737698251322,
      "grad_norm": 0.7477601766586304,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1677,
      "step": 131
    },
    {
      "epoch": 0.10736071573810492,
      "grad_norm": 0.7499041557312012,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.1445,
      "step": 132
    },
    {
      "epoch": 0.10817405449369663,
      "grad_norm": 0.7380723357200623,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1145,
      "step": 133
    },
    {
      "epoch": 0.10898739324928833,
      "grad_norm": 0.8323284983634949,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.1791,
      "step": 134
    },
    {
      "epoch": 0.10980073200488004,
      "grad_norm": 0.8600062131881714,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.1232,
      "step": 135
    },
    {
      "epoch": 0.11061407076047174,
      "grad_norm": 1.059281826019287,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0529,
      "step": 136
    },
    {
      "epoch": 0.11142740951606345,
      "grad_norm": 1.0456724166870117,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2178,
      "step": 137
    },
    {
      "epoch": 0.11224074827165514,
      "grad_norm": 1.1753181219100952,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1653,
      "step": 138
    },
    {
      "epoch": 0.11305408702724686,
      "grad_norm": 1.0999313592910767,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.0019,
      "step": 139
    },
    {
      "epoch": 0.11386742578283855,
      "grad_norm": 1.330169439315796,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.9902,
      "step": 140
    },
    {
      "epoch": 0.11468076453843025,
      "grad_norm": 1.2811272144317627,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.9978,
      "step": 141
    },
    {
      "epoch": 0.11549410329402196,
      "grad_norm": 1.394974946975708,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.0387,
      "step": 142
    },
    {
      "epoch": 0.11630744204961366,
      "grad_norm": 1.524612545967102,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.9911,
      "step": 143
    },
    {
      "epoch": 0.11712078080520537,
      "grad_norm": 1.4745184183120728,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.7549,
      "step": 144
    },
    {
      "epoch": 0.11793411956079707,
      "grad_norm": 1.5490124225616455,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.0546,
      "step": 145
    },
    {
      "epoch": 0.11874745831638878,
      "grad_norm": 1.9890395402908325,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.0849,
      "step": 146
    },
    {
      "epoch": 0.11956079707198047,
      "grad_norm": 1.5341365337371826,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.8449,
      "step": 147
    },
    {
      "epoch": 0.12037413582757218,
      "grad_norm": 1.9113014936447144,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.9846,
      "step": 148
    },
    {
      "epoch": 0.12118747458316388,
      "grad_norm": 1.9153579473495483,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.8372,
      "step": 149
    },
    {
      "epoch": 0.12200081333875559,
      "grad_norm": 3.9733545780181885,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.5363,
      "step": 150
    },
    {
      "epoch": 0.12200081333875559,
      "eval_loss": 1.0768191814422607,
      "eval_runtime": 88.8274,
      "eval_samples_per_second": 23.315,
      "eval_steps_per_second": 5.832,
      "step": 150
    },
    {
      "epoch": 0.12281415209434729,
      "grad_norm": 0.30643174052238464,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.9751,
      "step": 151
    },
    {
      "epoch": 0.123627490849939,
      "grad_norm": 0.3268009126186371,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.0955,
      "step": 152
    },
    {
      "epoch": 0.1244408296055307,
      "grad_norm": 0.33609554171562195,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.1659,
      "step": 153
    },
    {
      "epoch": 0.1252541683611224,
      "grad_norm": 0.3275425136089325,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.1242,
      "step": 154
    },
    {
      "epoch": 0.1260675071167141,
      "grad_norm": 0.33968231081962585,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.0668,
      "step": 155
    },
    {
      "epoch": 0.1268808458723058,
      "grad_norm": 0.37512120604515076,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.1883,
      "step": 156
    },
    {
      "epoch": 0.12769418462789753,
      "grad_norm": 0.3543924391269684,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.0778,
      "step": 157
    },
    {
      "epoch": 0.12850752338348922,
      "grad_norm": 0.3728344440460205,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1004,
      "step": 158
    },
    {
      "epoch": 0.12932086213908092,
      "grad_norm": 0.34430885314941406,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.0505,
      "step": 159
    },
    {
      "epoch": 0.13013420089467262,
      "grad_norm": 0.408173143863678,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.1292,
      "step": 160
    },
    {
      "epoch": 0.13094753965026434,
      "grad_norm": 0.37590157985687256,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.1467,
      "step": 161
    },
    {
      "epoch": 0.13176087840585604,
      "grad_norm": 0.38756778836250305,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.1703,
      "step": 162
    },
    {
      "epoch": 0.13257421716144774,
      "grad_norm": 0.37330153584480286,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0813,
      "step": 163
    },
    {
      "epoch": 0.13338755591703944,
      "grad_norm": 0.3720373213291168,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0597,
      "step": 164
    },
    {
      "epoch": 0.13420089467263116,
      "grad_norm": 0.3890669047832489,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.0766,
      "step": 165
    },
    {
      "epoch": 0.13501423342822286,
      "grad_norm": 0.39143073558807373,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.1514,
      "step": 166
    },
    {
      "epoch": 0.13582757218381455,
      "grad_norm": 0.41639575362205505,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.1938,
      "step": 167
    },
    {
      "epoch": 0.13664091093940625,
      "grad_norm": 0.41437551379203796,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0823,
      "step": 168
    },
    {
      "epoch": 0.13745424969499798,
      "grad_norm": 0.4499380588531494,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.2052,
      "step": 169
    },
    {
      "epoch": 0.13826758845058967,
      "grad_norm": 0.4648504853248596,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.1096,
      "step": 170
    },
    {
      "epoch": 0.13908092720618137,
      "grad_norm": 0.49872690439224243,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.2844,
      "step": 171
    },
    {
      "epoch": 0.13989426596177307,
      "grad_norm": 0.5160312652587891,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.2801,
      "step": 172
    },
    {
      "epoch": 0.1407076047173648,
      "grad_norm": 0.49740076065063477,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.1684,
      "step": 173
    },
    {
      "epoch": 0.1415209434729565,
      "grad_norm": 0.5344133973121643,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.3055,
      "step": 174
    },
    {
      "epoch": 0.1423342822285482,
      "grad_norm": 0.5410744547843933,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.1928,
      "step": 175
    },
    {
      "epoch": 0.14314762098413988,
      "grad_norm": 0.5793537497520447,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.1174,
      "step": 176
    },
    {
      "epoch": 0.1439609597397316,
      "grad_norm": 0.6002209186553955,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.2152,
      "step": 177
    },
    {
      "epoch": 0.1447742984953233,
      "grad_norm": 0.5854225158691406,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.1934,
      "step": 178
    },
    {
      "epoch": 0.145587637250915,
      "grad_norm": 0.6765329241752625,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.238,
      "step": 179
    },
    {
      "epoch": 0.1464009760065067,
      "grad_norm": 0.6818188428878784,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.159,
      "step": 180
    },
    {
      "epoch": 0.14721431476209842,
      "grad_norm": 0.7453911304473877,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2572,
      "step": 181
    },
    {
      "epoch": 0.14802765351769012,
      "grad_norm": 1.0769611597061157,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.0332,
      "step": 182
    },
    {
      "epoch": 0.14884099227328182,
      "grad_norm": 0.8125038146972656,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.0758,
      "step": 183
    },
    {
      "epoch": 0.14965433102887352,
      "grad_norm": 0.9278245568275452,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.0299,
      "step": 184
    },
    {
      "epoch": 0.15046766978446524,
      "grad_norm": 1.0640056133270264,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.9823,
      "step": 185
    },
    {
      "epoch": 0.15128100854005694,
      "grad_norm": 1.0738080739974976,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1573,
      "step": 186
    },
    {
      "epoch": 0.15209434729564864,
      "grad_norm": 1.1078773736953735,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0005,
      "step": 187
    },
    {
      "epoch": 0.15290768605124033,
      "grad_norm": 1.1760276556015015,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.1245,
      "step": 188
    },
    {
      "epoch": 0.15372102480683206,
      "grad_norm": 1.3052631616592407,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.9939,
      "step": 189
    },
    {
      "epoch": 0.15453436356242375,
      "grad_norm": 1.1983450651168823,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.9377,
      "step": 190
    },
    {
      "epoch": 0.15534770231801545,
      "grad_norm": 1.4003032445907593,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.1285,
      "step": 191
    },
    {
      "epoch": 0.15616104107360715,
      "grad_norm": 1.4053841829299927,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.0471,
      "step": 192
    },
    {
      "epoch": 0.15697437982919887,
      "grad_norm": 1.5955594778060913,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1501,
      "step": 193
    },
    {
      "epoch": 0.15778771858479057,
      "grad_norm": 1.717362403869629,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.9466,
      "step": 194
    },
    {
      "epoch": 0.15860105734038227,
      "grad_norm": 1.7957724332809448,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.0075,
      "step": 195
    },
    {
      "epoch": 0.15941439609597396,
      "grad_norm": 1.660839319229126,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2567,
      "step": 196
    },
    {
      "epoch": 0.1602277348515657,
      "grad_norm": 1.6973848342895508,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.7597,
      "step": 197
    },
    {
      "epoch": 0.1610410736071574,
      "grad_norm": 1.566506266593933,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.7693,
      "step": 198
    },
    {
      "epoch": 0.16185441236274908,
      "grad_norm": 2.0964441299438477,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.8482,
      "step": 199
    },
    {
      "epoch": 0.16266775111834078,
      "grad_norm": 3.0530765056610107,
      "learning_rate": 0.0,
      "loss": 0.9245,
      "step": 200
    },
    {
      "epoch": 0.16266775111834078,
      "eval_loss": 1.0638600587844849,
      "eval_runtime": 89.1861,
      "eval_samples_per_second": 23.221,
      "eval_steps_per_second": 5.808,
      "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": 1.146618963296256e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}