{
  "best_metric": 0.06656570732593536,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.922722029988466,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00461361014994233,
      "grad_norm": 15.18424129486084,
      "learning_rate": 1e-05,
      "loss": 4.4125,
      "step": 1
    },
    {
      "epoch": 0.00461361014994233,
      "eval_loss": 1.3732987642288208,
      "eval_runtime": 17.0765,
      "eval_samples_per_second": 21.374,
      "eval_steps_per_second": 5.388,
      "step": 1
    },
    {
      "epoch": 0.00922722029988466,
      "grad_norm": 20.024776458740234,
      "learning_rate": 2e-05,
      "loss": 4.6565,
      "step": 2
    },
    {
      "epoch": 0.01384083044982699,
      "grad_norm": 18.63164520263672,
      "learning_rate": 3e-05,
      "loss": 4.6303,
      "step": 3
    },
    {
      "epoch": 0.01845444059976932,
      "grad_norm": 13.807875633239746,
      "learning_rate": 4e-05,
      "loss": 4.5021,
      "step": 4
    },
    {
      "epoch": 0.02306805074971165,
      "grad_norm": 7.613506317138672,
      "learning_rate": 5e-05,
      "loss": 3.985,
      "step": 5
    },
    {
      "epoch": 0.02768166089965398,
      "grad_norm": 6.131557941436768,
      "learning_rate": 6e-05,
      "loss": 3.7741,
      "step": 6
    },
    {
      "epoch": 0.03229527104959631,
      "grad_norm": 6.234939098358154,
      "learning_rate": 7e-05,
      "loss": 3.5177,
      "step": 7
    },
    {
      "epoch": 0.03690888119953864,
      "grad_norm": 4.510298728942871,
      "learning_rate": 8e-05,
      "loss": 3.0088,
      "step": 8
    },
    {
      "epoch": 0.04152249134948097,
      "grad_norm": 5.073261260986328,
      "learning_rate": 9e-05,
      "loss": 2.8429,
      "step": 9
    },
    {
      "epoch": 0.0461361014994233,
      "grad_norm": 5.410125255584717,
      "learning_rate": 0.0001,
      "loss": 2.5398,
      "step": 10
    },
    {
      "epoch": 0.05074971164936563,
      "grad_norm": 4.456904888153076,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.4516,
      "step": 11
    },
    {
      "epoch": 0.05536332179930796,
      "grad_norm": 3.9411356449127197,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.2086,
      "step": 12
    },
    {
      "epoch": 0.05997693194925029,
      "grad_norm": 3.0666022300720215,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.9033,
      "step": 13
    },
    {
      "epoch": 0.06459054209919261,
      "grad_norm": 3.4097535610198975,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.0089,
      "step": 14
    },
    {
      "epoch": 0.06920415224913495,
      "grad_norm": 2.8611998558044434,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.6352,
      "step": 15
    },
    {
      "epoch": 0.07381776239907728,
      "grad_norm": 2.490839958190918,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5876,
      "step": 16
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 2.0047426223754883,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.367,
      "step": 17
    },
    {
      "epoch": 0.08304498269896193,
      "grad_norm": 2.3783082962036133,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.4764,
      "step": 18
    },
    {
      "epoch": 0.08765859284890427,
      "grad_norm": 2.6521780490875244,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.3002,
      "step": 19
    },
    {
      "epoch": 0.0922722029988466,
      "grad_norm": 2.160538673400879,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3112,
      "step": 20
    },
    {
      "epoch": 0.09688581314878893,
      "grad_norm": 2.36342453956604,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1729,
      "step": 21
    },
    {
      "epoch": 0.10149942329873125,
      "grad_norm": 2.79667067527771,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.3722,
      "step": 22
    },
    {
      "epoch": 0.1061130334486736,
      "grad_norm": 2.74226975440979,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.2075,
      "step": 23
    },
    {
      "epoch": 0.11072664359861592,
      "grad_norm": 2.2346718311309814,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.0829,
      "step": 24
    },
    {
      "epoch": 0.11534025374855825,
      "grad_norm": 2.790926456451416,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.1215,
      "step": 25
    },
    {
      "epoch": 0.11995386389850057,
      "grad_norm": 2.3403475284576416,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.9618,
      "step": 26
    },
    {
      "epoch": 0.1245674740484429,
      "grad_norm": 2.535047769546509,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.9964,
      "step": 27
    },
    {
      "epoch": 0.12918108419838523,
      "grad_norm": 2.7655372619628906,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.9765,
      "step": 28
    },
    {
      "epoch": 0.13379469434832755,
      "grad_norm": 2.4994313716888428,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0133,
      "step": 29
    },
    {
      "epoch": 0.1384083044982699,
      "grad_norm": 2.193906545639038,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9688,
      "step": 30
    },
    {
      "epoch": 0.14302191464821223,
      "grad_norm": 2.7496867179870605,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.9134,
      "step": 31
    },
    {
      "epoch": 0.14763552479815456,
      "grad_norm": 2.234182834625244,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.8893,
      "step": 32
    },
    {
      "epoch": 0.1522491349480969,
      "grad_norm": 2.0011980533599854,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.7804,
      "step": 33
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 2.2012534141540527,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.8649,
      "step": 34
    },
    {
      "epoch": 0.16147635524798154,
      "grad_norm": 2.191406726837158,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.7678,
      "step": 35
    },
    {
      "epoch": 0.16608996539792387,
      "grad_norm": 2.2175698280334473,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.7798,
      "step": 36
    },
    {
      "epoch": 0.1707035755478662,
      "grad_norm": 2.2709202766418457,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.8145,
      "step": 37
    },
    {
      "epoch": 0.17531718569780855,
      "grad_norm": 2.383368730545044,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.8203,
      "step": 38
    },
    {
      "epoch": 0.17993079584775087,
      "grad_norm": 2.196934938430786,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.7534,
      "step": 39
    },
    {
      "epoch": 0.1845444059976932,
      "grad_norm": 1.544556975364685,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.6179,
      "step": 40
    },
    {
      "epoch": 0.18915801614763553,
      "grad_norm": 1.7371619939804077,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.6035,
      "step": 41
    },
    {
      "epoch": 0.19377162629757785,
      "grad_norm": 2.1010055541992188,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.7008,
      "step": 42
    },
    {
      "epoch": 0.19838523644752018,
      "grad_norm": 2.0607075691223145,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.5768,
      "step": 43
    },
    {
      "epoch": 0.2029988465974625,
      "grad_norm": 2.099403142929077,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.5462,
      "step": 44
    },
    {
      "epoch": 0.20761245674740483,
      "grad_norm": 1.884232759475708,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.5677,
      "step": 45
    },
    {
      "epoch": 0.2122260668973472,
      "grad_norm": 2.212244749069214,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.5578,
      "step": 46
    },
    {
      "epoch": 0.21683967704728951,
      "grad_norm": 2.2206742763519287,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.6375,
      "step": 47
    },
    {
      "epoch": 0.22145328719723184,
      "grad_norm": 2.314852237701416,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.5721,
      "step": 48
    },
    {
      "epoch": 0.22606689734717417,
      "grad_norm": 2.848503828048706,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6016,
      "step": 49
    },
    {
      "epoch": 0.2306805074971165,
      "grad_norm": 3.2887659072875977,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.8696,
      "step": 50
    },
    {
      "epoch": 0.2306805074971165,
      "eval_loss": 0.17636431753635406,
      "eval_runtime": 17.4137,
      "eval_samples_per_second": 20.961,
      "eval_steps_per_second": 5.283,
      "step": 50
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 1.778917670249939,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.5403,
      "step": 51
    },
    {
      "epoch": 0.23990772779700115,
      "grad_norm": 1.6891322135925293,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.5421,
      "step": 52
    },
    {
      "epoch": 0.24452133794694347,
      "grad_norm": 2.0808722972869873,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6795,
      "step": 53
    },
    {
      "epoch": 0.2491349480968858,
      "grad_norm": 2.1992642879486084,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.6254,
      "step": 54
    },
    {
      "epoch": 0.2537485582468281,
      "grad_norm": 1.891418695449829,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.7448,
      "step": 55
    },
    {
      "epoch": 0.25836216839677045,
      "grad_norm": 1.8931927680969238,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.5738,
      "step": 56
    },
    {
      "epoch": 0.2629757785467128,
      "grad_norm": 1.633174180984497,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.6468,
      "step": 57
    },
    {
      "epoch": 0.2675893886966551,
      "grad_norm": 1.6375372409820557,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.668,
      "step": 58
    },
    {
      "epoch": 0.2722029988465975,
      "grad_norm": 1.6769412755966187,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.6342,
      "step": 59
    },
    {
      "epoch": 0.2768166089965398,
      "grad_norm": 1.4747549295425415,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.6371,
      "step": 60
    },
    {
      "epoch": 0.28143021914648214,
      "grad_norm": 1.8303107023239136,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.5018,
      "step": 61
    },
    {
      "epoch": 0.28604382929642447,
      "grad_norm": 1.4639233350753784,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.4807,
      "step": 62
    },
    {
      "epoch": 0.2906574394463668,
      "grad_norm": 1.581974983215332,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.5338,
      "step": 63
    },
    {
      "epoch": 0.2952710495963091,
      "grad_norm": 1.5630853176116943,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.5133,
      "step": 64
    },
    {
      "epoch": 0.29988465974625145,
      "grad_norm": 1.7212315797805786,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.5826,
      "step": 65
    },
    {
      "epoch": 0.3044982698961938,
      "grad_norm": 1.5302855968475342,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.5601,
      "step": 66
    },
    {
      "epoch": 0.3091118800461361,
      "grad_norm": 1.5532318353652954,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.5429,
      "step": 67
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 1.5127520561218262,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.5589,
      "step": 68
    },
    {
      "epoch": 0.31833910034602075,
      "grad_norm": 1.3508917093276978,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.5395,
      "step": 69
    },
    {
      "epoch": 0.3229527104959631,
      "grad_norm": 1.8375931978225708,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.5359,
      "step": 70
    },
    {
      "epoch": 0.3275663206459054,
      "grad_norm": 1.5403379201889038,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.4771,
      "step": 71
    },
    {
      "epoch": 0.33217993079584773,
      "grad_norm": 1.9747036695480347,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5319,
      "step": 72
    },
    {
      "epoch": 0.33679354094579006,
      "grad_norm": 1.5574091672897339,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.4915,
      "step": 73
    },
    {
      "epoch": 0.3414071510957324,
      "grad_norm": 1.5811783075332642,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.6,
      "step": 74
    },
    {
      "epoch": 0.3460207612456747,
      "grad_norm": 1.632620930671692,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5039,
      "step": 75
    },
    {
      "epoch": 0.3506343713956171,
      "grad_norm": 1.5223382711410522,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.5048,
      "step": 76
    },
    {
      "epoch": 0.3552479815455594,
      "grad_norm": 1.8742772340774536,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.4492,
      "step": 77
    },
    {
      "epoch": 0.35986159169550175,
      "grad_norm": 1.5278382301330566,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4546,
      "step": 78
    },
    {
      "epoch": 0.3644752018454441,
      "grad_norm": 1.8101459741592407,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.5013,
      "step": 79
    },
    {
      "epoch": 0.3690888119953864,
      "grad_norm": 1.4167555570602417,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.4598,
      "step": 80
    },
    {
      "epoch": 0.3737024221453287,
      "grad_norm": 2.231569766998291,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.527,
      "step": 81
    },
    {
      "epoch": 0.37831603229527105,
      "grad_norm": 1.5968513488769531,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.5133,
      "step": 82
    },
    {
      "epoch": 0.3829296424452134,
      "grad_norm": 1.4470057487487793,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.443,
      "step": 83
    },
    {
      "epoch": 0.3875432525951557,
      "grad_norm": 1.7189449071884155,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.4486,
      "step": 84
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 1.599415898323059,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.4868,
      "step": 85
    },
    {
      "epoch": 0.39677047289504036,
      "grad_norm": 1.8464813232421875,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.4435,
      "step": 86
    },
    {
      "epoch": 0.4013840830449827,
      "grad_norm": 1.677215337753296,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.4457,
      "step": 87
    },
    {
      "epoch": 0.405997693194925,
      "grad_norm": 1.5768344402313232,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.4538,
      "step": 88
    },
    {
      "epoch": 0.41061130334486734,
      "grad_norm": 1.4954752922058105,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.4343,
      "step": 89
    },
    {
      "epoch": 0.41522491349480967,
      "grad_norm": 1.5641493797302246,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4731,
      "step": 90
    },
    {
      "epoch": 0.419838523644752,
      "grad_norm": 1.2303073406219482,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.2784,
      "step": 91
    },
    {
      "epoch": 0.4244521337946944,
      "grad_norm": 1.716969609260559,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.423,
      "step": 92
    },
    {
      "epoch": 0.4290657439446367,
      "grad_norm": 1.7639166116714478,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.43,
      "step": 93
    },
    {
      "epoch": 0.43367935409457903,
      "grad_norm": 1.487332820892334,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3776,
      "step": 94
    },
    {
      "epoch": 0.43829296424452135,
      "grad_norm": 1.837235927581787,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.4869,
      "step": 95
    },
    {
      "epoch": 0.4429065743944637,
      "grad_norm": 1.4869582653045654,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3266,
      "step": 96
    },
    {
      "epoch": 0.447520184544406,
      "grad_norm": 1.269347071647644,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.2237,
      "step": 97
    },
    {
      "epoch": 0.45213379469434833,
      "grad_norm": 2.248960256576538,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.4066,
      "step": 98
    },
    {
      "epoch": 0.45674740484429066,
      "grad_norm": 2.063490152359009,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.4325,
      "step": 99
    },
    {
      "epoch": 0.461361014994233,
      "grad_norm": 2.7826852798461914,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.4356,
      "step": 100
    },
    {
      "epoch": 0.461361014994233,
      "eval_loss": 0.09896985441446304,
      "eval_runtime": 17.3559,
      "eval_samples_per_second": 21.03,
      "eval_steps_per_second": 5.301,
      "step": 100
    },
    {
      "epoch": 0.4659746251441753,
      "grad_norm": 1.3039848804473877,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3601,
      "step": 101
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 1.0854651927947998,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.2982,
      "step": 102
    },
    {
      "epoch": 0.47520184544405997,
      "grad_norm": 1.2118704319000244,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3719,
      "step": 103
    },
    {
      "epoch": 0.4798154555940023,
      "grad_norm": 1.4792733192443848,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.4309,
      "step": 104
    },
    {
      "epoch": 0.4844290657439446,
      "grad_norm": 1.5741987228393555,
      "learning_rate": 5e-05,
      "loss": 0.4081,
      "step": 105
    },
    {
      "epoch": 0.48904267589388695,
      "grad_norm": 1.5493136644363403,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3973,
      "step": 106
    },
    {
      "epoch": 0.4936562860438293,
      "grad_norm": 1.5050002336502075,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.4248,
      "step": 107
    },
    {
      "epoch": 0.4982698961937716,
      "grad_norm": 1.2711061239242554,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.3203,
      "step": 108
    },
    {
      "epoch": 0.5028835063437139,
      "grad_norm": 1.2149827480316162,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.2997,
      "step": 109
    },
    {
      "epoch": 0.5074971164936563,
      "grad_norm": 1.2261457443237305,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3121,
      "step": 110
    },
    {
      "epoch": 0.5121107266435986,
      "grad_norm": 1.1415098905563354,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.3037,
      "step": 111
    },
    {
      "epoch": 0.5167243367935409,
      "grad_norm": 1.6010178327560425,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.4267,
      "step": 112
    },
    {
      "epoch": 0.5213379469434832,
      "grad_norm": 1.3134310245513916,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3301,
      "step": 113
    },
    {
      "epoch": 0.5259515570934256,
      "grad_norm": 1.5067267417907715,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3875,
      "step": 114
    },
    {
      "epoch": 0.5305651672433679,
      "grad_norm": 1.1298049688339233,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3,
      "step": 115
    },
    {
      "epoch": 0.5351787773933102,
      "grad_norm": 1.1532821655273438,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3633,
      "step": 116
    },
    {
      "epoch": 0.5397923875432526,
      "grad_norm": 1.3989962339401245,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3472,
      "step": 117
    },
    {
      "epoch": 0.544405997693195,
      "grad_norm": 1.8116097450256348,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3939,
      "step": 118
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 1.5105483531951904,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3922,
      "step": 119
    },
    {
      "epoch": 0.5536332179930796,
      "grad_norm": 1.1212730407714844,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2863,
      "step": 120
    },
    {
      "epoch": 0.558246828143022,
      "grad_norm": 1.6151677370071411,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.3524,
      "step": 121
    },
    {
      "epoch": 0.5628604382929643,
      "grad_norm": 1.3998459577560425,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3398,
      "step": 122
    },
    {
      "epoch": 0.5674740484429066,
      "grad_norm": 1.2283453941345215,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3482,
      "step": 123
    },
    {
      "epoch": 0.5720876585928489,
      "grad_norm": 1.6418330669403076,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3156,
      "step": 124
    },
    {
      "epoch": 0.5767012687427913,
      "grad_norm": 1.3278775215148926,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.3147,
      "step": 125
    },
    {
      "epoch": 0.5813148788927336,
      "grad_norm": 1.148337483406067,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.258,
      "step": 126
    },
    {
      "epoch": 0.5859284890426759,
      "grad_norm": 1.2191426753997803,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3288,
      "step": 127
    },
    {
      "epoch": 0.5905420991926182,
      "grad_norm": 1.4731138944625854,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3776,
      "step": 128
    },
    {
      "epoch": 0.5951557093425606,
      "grad_norm": 1.4280023574829102,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.4394,
      "step": 129
    },
    {
      "epoch": 0.5997693194925029,
      "grad_norm": 1.4137822389602661,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3002,
      "step": 130
    },
    {
      "epoch": 0.6043829296424452,
      "grad_norm": 1.7817277908325195,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.4398,
      "step": 131
    },
    {
      "epoch": 0.6089965397923875,
      "grad_norm": 1.3498079776763916,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.4034,
      "step": 132
    },
    {
      "epoch": 0.6136101499423299,
      "grad_norm": 1.5424808263778687,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.4193,
      "step": 133
    },
    {
      "epoch": 0.6182237600922722,
      "grad_norm": 1.3509687185287476,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3755,
      "step": 134
    },
    {
      "epoch": 0.6228373702422145,
      "grad_norm": 1.941940426826477,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.303,
      "step": 135
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 1.5953118801116943,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.4121,
      "step": 136
    },
    {
      "epoch": 0.6320645905420992,
      "grad_norm": 1.118159532546997,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.2373,
      "step": 137
    },
    {
      "epoch": 0.6366782006920415,
      "grad_norm": 0.9691784977912903,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.2244,
      "step": 138
    },
    {
      "epoch": 0.6412918108419838,
      "grad_norm": 1.2845728397369385,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3074,
      "step": 139
    },
    {
      "epoch": 0.6459054209919262,
      "grad_norm": 1.2723478078842163,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2689,
      "step": 140
    },
    {
      "epoch": 0.6505190311418685,
      "grad_norm": 1.2604954242706299,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.2914,
      "step": 141
    },
    {
      "epoch": 0.6551326412918108,
      "grad_norm": 1.2579132318496704,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2537,
      "step": 142
    },
    {
      "epoch": 0.6597462514417531,
      "grad_norm": 1.6300617456436157,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.301,
      "step": 143
    },
    {
      "epoch": 0.6643598615916955,
      "grad_norm": 1.1282131671905518,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.1795,
      "step": 144
    },
    {
      "epoch": 0.6689734717416378,
      "grad_norm": 1.4947746992111206,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3037,
      "step": 145
    },
    {
      "epoch": 0.6735870818915801,
      "grad_norm": 1.6214041709899902,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.3567,
      "step": 146
    },
    {
      "epoch": 0.6782006920415224,
      "grad_norm": 1.575844168663025,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.3014,
      "step": 147
    },
    {
      "epoch": 0.6828143021914648,
      "grad_norm": 1.8102182149887085,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2842,
      "step": 148
    },
    {
      "epoch": 0.6874279123414071,
      "grad_norm": 1.95388925075531,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3852,
      "step": 149
    },
    {
      "epoch": 0.6920415224913494,
      "grad_norm": 2.8386013507843018,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.4423,
      "step": 150
    },
    {
      "epoch": 0.6920415224913494,
      "eval_loss": 0.07570867240428925,
      "eval_runtime": 17.4113,
      "eval_samples_per_second": 20.963,
      "eval_steps_per_second": 5.284,
      "step": 150
    },
    {
      "epoch": 0.6966551326412919,
      "grad_norm": 1.2515736818313599,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2714,
      "step": 151
    },
    {
      "epoch": 0.7012687427912342,
      "grad_norm": 1.2252408266067505,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.2586,
      "step": 152
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 1.4171631336212158,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3136,
      "step": 153
    },
    {
      "epoch": 0.7104959630911188,
      "grad_norm": 1.2745884656906128,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.3139,
      "step": 154
    },
    {
      "epoch": 0.7151095732410612,
      "grad_norm": 1.3520458936691284,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.3028,
      "step": 155
    },
    {
      "epoch": 0.7197231833910035,
      "grad_norm": 1.210499882698059,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.2581,
      "step": 156
    },
    {
      "epoch": 0.7243367935409458,
      "grad_norm": 1.2807996273040771,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3584,
      "step": 157
    },
    {
      "epoch": 0.7289504036908881,
      "grad_norm": 1.23085355758667,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2617,
      "step": 158
    },
    {
      "epoch": 0.7335640138408305,
      "grad_norm": 1.2482898235321045,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.2737,
      "step": 159
    },
    {
      "epoch": 0.7381776239907728,
      "grad_norm": 1.022756814956665,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.2355,
      "step": 160
    },
    {
      "epoch": 0.7427912341407151,
      "grad_norm": 1.1853199005126953,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.2744,
      "step": 161
    },
    {
      "epoch": 0.7474048442906575,
      "grad_norm": 1.272267460823059,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2566,
      "step": 162
    },
    {
      "epoch": 0.7520184544405998,
      "grad_norm": 0.9737191796302795,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.2373,
      "step": 163
    },
    {
      "epoch": 0.7566320645905421,
      "grad_norm": 1.0141868591308594,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2161,
      "step": 164
    },
    {
      "epoch": 0.7612456747404844,
      "grad_norm": 1.3028624057769775,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2798,
      "step": 165
    },
    {
      "epoch": 0.7658592848904268,
      "grad_norm": 1.302142858505249,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.2916,
      "step": 166
    },
    {
      "epoch": 0.7704728950403691,
      "grad_norm": 1.2708913087844849,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3564,
      "step": 167
    },
    {
      "epoch": 0.7750865051903114,
      "grad_norm": 1.2147297859191895,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2636,
      "step": 168
    },
    {
      "epoch": 0.7797001153402537,
      "grad_norm": 1.0937905311584473,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2505,
      "step": 169
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 1.1087138652801514,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.2666,
      "step": 170
    },
    {
      "epoch": 0.7889273356401384,
      "grad_norm": 1.5358645915985107,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3249,
      "step": 171
    },
    {
      "epoch": 0.7935409457900807,
      "grad_norm": 1.3537219762802124,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.3088,
      "step": 172
    },
    {
      "epoch": 0.798154555940023,
      "grad_norm": 1.2292712926864624,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.3016,
      "step": 173
    },
    {
      "epoch": 0.8027681660899654,
      "grad_norm": 1.060456395149231,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2449,
      "step": 174
    },
    {
      "epoch": 0.8073817762399077,
      "grad_norm": 1.2052295207977295,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.306,
      "step": 175
    },
    {
      "epoch": 0.81199538638985,
      "grad_norm": 1.1273924112319946,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.2947,
      "step": 176
    },
    {
      "epoch": 0.8166089965397924,
      "grad_norm": 1.0607858896255493,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.2293,
      "step": 177
    },
    {
      "epoch": 0.8212226066897347,
      "grad_norm": 1.218924641609192,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.2562,
      "step": 178
    },
    {
      "epoch": 0.825836216839677,
      "grad_norm": 1.3182764053344727,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3036,
      "step": 179
    },
    {
      "epoch": 0.8304498269896193,
      "grad_norm": 1.4145666360855103,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.3533,
      "step": 180
    },
    {
      "epoch": 0.8350634371395617,
      "grad_norm": 1.3908517360687256,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.3539,
      "step": 181
    },
    {
      "epoch": 0.839677047289504,
      "grad_norm": 1.1159809827804565,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.2625,
      "step": 182
    },
    {
      "epoch": 0.8442906574394463,
      "grad_norm": 1.467221975326538,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2809,
      "step": 183
    },
    {
      "epoch": 0.8489042675893888,
      "grad_norm": 1.589326024055481,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3439,
      "step": 184
    },
    {
      "epoch": 0.8535178777393311,
      "grad_norm": 1.3742806911468506,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.26,
      "step": 185
    },
    {
      "epoch": 0.8581314878892734,
      "grad_norm": 1.5459033250808716,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.2925,
      "step": 186
    },
    {
      "epoch": 0.8627450980392157,
      "grad_norm": 1.3778057098388672,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2951,
      "step": 187
    },
    {
      "epoch": 0.8673587081891581,
      "grad_norm": 1.0586920976638794,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2079,
      "step": 188
    },
    {
      "epoch": 0.8719723183391004,
      "grad_norm": 1.2481350898742676,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.2253,
      "step": 189
    },
    {
      "epoch": 0.8765859284890427,
      "grad_norm": 1.4716179370880127,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.2934,
      "step": 190
    },
    {
      "epoch": 0.881199538638985,
      "grad_norm": 1.3267749547958374,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2175,
      "step": 191
    },
    {
      "epoch": 0.8858131487889274,
      "grad_norm": 1.1983189582824707,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.2571,
      "step": 192
    },
    {
      "epoch": 0.8904267589388697,
      "grad_norm": 1.2302207946777344,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2555,
      "step": 193
    },
    {
      "epoch": 0.895040369088812,
      "grad_norm": 1.3070582151412964,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.2569,
      "step": 194
    },
    {
      "epoch": 0.8996539792387543,
      "grad_norm": 1.2101037502288818,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.2154,
      "step": 195
    },
    {
      "epoch": 0.9042675893886967,
      "grad_norm": 1.6718045473098755,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.3347,
      "step": 196
    },
    {
      "epoch": 0.908881199538639,
      "grad_norm": 1.6115975379943848,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.256,
      "step": 197
    },
    {
      "epoch": 0.9134948096885813,
      "grad_norm": 1.6082204580307007,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.2166,
      "step": 198
    },
    {
      "epoch": 0.9181084198385236,
      "grad_norm": 1.9207119941711426,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2874,
      "step": 199
    },
    {
      "epoch": 0.922722029988466,
      "grad_norm": 1.6422442197799683,
      "learning_rate": 0.0,
      "loss": 0.2269,
      "step": 200
    },
    {
      "epoch": 0.922722029988466,
      "eval_loss": 0.06656570732593536,
      "eval_runtime": 17.3802,
      "eval_samples_per_second": 21.001,
      "eval_steps_per_second": 5.293,
      "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": 1.50336002064384e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}