{
  "best_metric": 0.2527673840522766,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.12876227265411236,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006438113632705617,
      "grad_norm": 2.7586772441864014,
      "learning_rate": 1e-05,
      "loss": 0.484,
      "step": 1
    },
    {
      "epoch": 0.0006438113632705617,
      "eval_loss": 0.5094354152679443,
      "eval_runtime": 213.6979,
      "eval_samples_per_second": 12.242,
      "eval_steps_per_second": 3.06,
      "step": 1
    },
    {
      "epoch": 0.0012876227265411235,
      "grad_norm": 3.4913101196289062,
      "learning_rate": 2e-05,
      "loss": 0.8068,
      "step": 2
    },
    {
      "epoch": 0.001931434089811685,
      "grad_norm": 5.203159809112549,
      "learning_rate": 3e-05,
      "loss": 1.0453,
      "step": 3
    },
    {
      "epoch": 0.002575245453082247,
      "grad_norm": 3.369285821914673,
      "learning_rate": 4e-05,
      "loss": 1.3956,
      "step": 4
    },
    {
      "epoch": 0.003219056816352809,
      "grad_norm": 2.530057191848755,
      "learning_rate": 5e-05,
      "loss": 0.7828,
      "step": 5
    },
    {
      "epoch": 0.00386286817962337,
      "grad_norm": 2.487403392791748,
      "learning_rate": 6e-05,
      "loss": 0.8683,
      "step": 6
    },
    {
      "epoch": 0.0045066795428939325,
      "grad_norm": 2.891543388366699,
      "learning_rate": 7e-05,
      "loss": 1.054,
      "step": 7
    },
    {
      "epoch": 0.005150490906164494,
      "grad_norm": 2.854285478591919,
      "learning_rate": 8e-05,
      "loss": 1.1143,
      "step": 8
    },
    {
      "epoch": 0.005794302269435055,
      "grad_norm": 2.239809274673462,
      "learning_rate": 9e-05,
      "loss": 0.7577,
      "step": 9
    },
    {
      "epoch": 0.006438113632705618,
      "grad_norm": 2.365220069885254,
      "learning_rate": 0.0001,
      "loss": 0.8837,
      "step": 10
    },
    {
      "epoch": 0.007081924995976179,
      "grad_norm": 2.4386515617370605,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.8666,
      "step": 11
    },
    {
      "epoch": 0.00772573635924674,
      "grad_norm": 3.054266929626465,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.9865,
      "step": 12
    },
    {
      "epoch": 0.008369547722517303,
      "grad_norm": 2.756296157836914,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.8786,
      "step": 13
    },
    {
      "epoch": 0.009013359085787865,
      "grad_norm": 2.560093641281128,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.782,
      "step": 14
    },
    {
      "epoch": 0.009657170449058426,
      "grad_norm": 2.6688358783721924,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.2252,
      "step": 15
    },
    {
      "epoch": 0.010300981812328988,
      "grad_norm": 2.5514702796936035,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.9558,
      "step": 16
    },
    {
      "epoch": 0.01094479317559955,
      "grad_norm": 2.711026668548584,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.0278,
      "step": 17
    },
    {
      "epoch": 0.01158860453887011,
      "grad_norm": 2.954331398010254,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.2308,
      "step": 18
    },
    {
      "epoch": 0.012232415902140673,
      "grad_norm": 2.981369733810425,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.0474,
      "step": 19
    },
    {
      "epoch": 0.012876227265411235,
      "grad_norm": 2.8534157276153564,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.1713,
      "step": 20
    },
    {
      "epoch": 0.013520038628681796,
      "grad_norm": 3.0949838161468506,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.2187,
      "step": 21
    },
    {
      "epoch": 0.014163849991952358,
      "grad_norm": 2.8564131259918213,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.0249,
      "step": 22
    },
    {
      "epoch": 0.01480766135522292,
      "grad_norm": 3.3575549125671387,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.3171,
      "step": 23
    },
    {
      "epoch": 0.01545147271849348,
      "grad_norm": 2.9042158126831055,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.0252,
      "step": 24
    },
    {
      "epoch": 0.016095284081764045,
      "grad_norm": 3.1608076095581055,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.0897,
      "step": 25
    },
    {
      "epoch": 0.016739095445034605,
      "grad_norm": 3.195927858352661,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.3316,
      "step": 26
    },
    {
      "epoch": 0.017382906808305166,
      "grad_norm": 3.554035186767578,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.2097,
      "step": 27
    },
    {
      "epoch": 0.01802671817157573,
      "grad_norm": 3.1218101978302,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.0435,
      "step": 28
    },
    {
      "epoch": 0.01867052953484629,
      "grad_norm": 3.078212261199951,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.2053,
      "step": 29
    },
    {
      "epoch": 0.01931434089811685,
      "grad_norm": 3.353097677230835,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3273,
      "step": 30
    },
    {
      "epoch": 0.019958152261387415,
      "grad_norm": 3.1882622241973877,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.1413,
      "step": 31
    },
    {
      "epoch": 0.020601963624657976,
      "grad_norm": 3.6276164054870605,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.3342,
      "step": 32
    },
    {
      "epoch": 0.021245774987928536,
      "grad_norm": 3.6932384967803955,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2814,
      "step": 33
    },
    {
      "epoch": 0.0218895863511991,
      "grad_norm": 3.4194512367248535,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.107,
      "step": 34
    },
    {
      "epoch": 0.02253339771446966,
      "grad_norm": 3.30964732170105,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.1435,
      "step": 35
    },
    {
      "epoch": 0.02317720907774022,
      "grad_norm": 3.6015894412994385,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3716,
      "step": 36
    },
    {
      "epoch": 0.023821020441010785,
      "grad_norm": 3.4707467555999756,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.0779,
      "step": 37
    },
    {
      "epoch": 0.024464831804281346,
      "grad_norm": 4.103673934936523,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.2257,
      "step": 38
    },
    {
      "epoch": 0.025108643167551906,
      "grad_norm": 7.245471000671387,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.245,
      "step": 39
    },
    {
      "epoch": 0.02575245453082247,
      "grad_norm": 4.206423282623291,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.2589,
      "step": 40
    },
    {
      "epoch": 0.02639626589409303,
      "grad_norm": 4.436548709869385,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.2072,
      "step": 41
    },
    {
      "epoch": 0.02704007725736359,
      "grad_norm": 3.9481594562530518,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.3028,
      "step": 42
    },
    {
      "epoch": 0.027683888620634155,
      "grad_norm": 4.500491142272949,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.2124,
      "step": 43
    },
    {
      "epoch": 0.028327699983904716,
      "grad_norm": 4.823503017425537,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.343,
      "step": 44
    },
    {
      "epoch": 0.028971511347175277,
      "grad_norm": 3.5415332317352295,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.0797,
      "step": 45
    },
    {
      "epoch": 0.02961532271044584,
      "grad_norm": 4.574649810791016,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.3344,
      "step": 46
    },
    {
      "epoch": 0.0302591340737164,
      "grad_norm": 4.903590202331543,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.3402,
      "step": 47
    },
    {
      "epoch": 0.03090294543698696,
      "grad_norm": 5.884121894836426,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3943,
      "step": 48
    },
    {
      "epoch": 0.03154675680025752,
      "grad_norm": 8.597677230834961,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.9095,
      "step": 49
    },
    {
      "epoch": 0.03219056816352809,
      "grad_norm": 12.897464752197266,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.9138,
      "step": 50
    },
    {
      "epoch": 0.03219056816352809,
      "eval_loss": 0.3263968825340271,
      "eval_runtime": 215.2328,
      "eval_samples_per_second": 12.154,
      "eval_steps_per_second": 3.039,
      "step": 50
    },
    {
      "epoch": 0.03283437952679865,
      "grad_norm": 3.8128509521484375,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.4371,
      "step": 51
    },
    {
      "epoch": 0.03347819089006921,
      "grad_norm": 2.112067461013794,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.673,
      "step": 52
    },
    {
      "epoch": 0.03412200225333977,
      "grad_norm": 1.6706172227859497,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.7727,
      "step": 53
    },
    {
      "epoch": 0.03476581361661033,
      "grad_norm": 2.110781669616699,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.8855,
      "step": 54
    },
    {
      "epoch": 0.03540962497988089,
      "grad_norm": 1.8794212341308594,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.9929,
      "step": 55
    },
    {
      "epoch": 0.03605343634315146,
      "grad_norm": 1.877468228340149,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.7695,
      "step": 56
    },
    {
      "epoch": 0.03669724770642202,
      "grad_norm": 1.7253862619400024,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.9385,
      "step": 57
    },
    {
      "epoch": 0.03734105906969258,
      "grad_norm": 1.7417266368865967,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.924,
      "step": 58
    },
    {
      "epoch": 0.03798487043296314,
      "grad_norm": 1.572616457939148,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.7326,
      "step": 59
    },
    {
      "epoch": 0.0386286817962337,
      "grad_norm": 1.7766157388687134,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.8458,
      "step": 60
    },
    {
      "epoch": 0.03927249315950426,
      "grad_norm": 2.061432123184204,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.9238,
      "step": 61
    },
    {
      "epoch": 0.03991630452277483,
      "grad_norm": 1.9468456506729126,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.9941,
      "step": 62
    },
    {
      "epoch": 0.04056011588604539,
      "grad_norm": 1.9039522409439087,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.9502,
      "step": 63
    },
    {
      "epoch": 0.04120392724931595,
      "grad_norm": 1.956261396408081,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.9128,
      "step": 64
    },
    {
      "epoch": 0.04184773861258651,
      "grad_norm": 2.431889772415161,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.1457,
      "step": 65
    },
    {
      "epoch": 0.04249154997585707,
      "grad_norm": 2.198906660079956,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.1591,
      "step": 66
    },
    {
      "epoch": 0.04313536133912763,
      "grad_norm": 2.406588554382324,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.2245,
      "step": 67
    },
    {
      "epoch": 0.0437791727023982,
      "grad_norm": 2.2110981941223145,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.0465,
      "step": 68
    },
    {
      "epoch": 0.04442298406566876,
      "grad_norm": 2.0902910232543945,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.9651,
      "step": 69
    },
    {
      "epoch": 0.04506679542893932,
      "grad_norm": 2.509665012359619,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.2153,
      "step": 70
    },
    {
      "epoch": 0.04571060679220988,
      "grad_norm": 2.3155574798583984,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.8746,
      "step": 71
    },
    {
      "epoch": 0.04635441815548044,
      "grad_norm": 2.4643378257751465,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1928,
      "step": 72
    },
    {
      "epoch": 0.046998229518751,
      "grad_norm": 2.5876643657684326,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.0966,
      "step": 73
    },
    {
      "epoch": 0.04764204088202157,
      "grad_norm": 2.661036252975464,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.0991,
      "step": 74
    },
    {
      "epoch": 0.04828585224529213,
      "grad_norm": 2.38126802444458,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.8888,
      "step": 75
    },
    {
      "epoch": 0.04892966360856269,
      "grad_norm": 2.3816940784454346,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.1485,
      "step": 76
    },
    {
      "epoch": 0.04957347497183325,
      "grad_norm": 2.7507896423339844,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2601,
      "step": 77
    },
    {
      "epoch": 0.05021728633510381,
      "grad_norm": 2.6402173042297363,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0965,
      "step": 78
    },
    {
      "epoch": 0.05086109769837437,
      "grad_norm": 2.464062213897705,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.0722,
      "step": 79
    },
    {
      "epoch": 0.05150490906164494,
      "grad_norm": 2.7562756538391113,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.222,
      "step": 80
    },
    {
      "epoch": 0.0521487204249155,
      "grad_norm": 2.982177495956421,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.1727,
      "step": 81
    },
    {
      "epoch": 0.05279253178818606,
      "grad_norm": 3.4546287059783936,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.5217,
      "step": 82
    },
    {
      "epoch": 0.05343634315145662,
      "grad_norm": 2.525982141494751,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.0534,
      "step": 83
    },
    {
      "epoch": 0.05408015451472718,
      "grad_norm": 2.6467645168304443,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1894,
      "step": 84
    },
    {
      "epoch": 0.05472396587799774,
      "grad_norm": 2.534043073654175,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.1111,
      "step": 85
    },
    {
      "epoch": 0.05536777724126831,
      "grad_norm": 4.064790725708008,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3789,
      "step": 86
    },
    {
      "epoch": 0.05601158860453887,
      "grad_norm": 3.106489419937134,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.1505,
      "step": 87
    },
    {
      "epoch": 0.05665539996780943,
      "grad_norm": 3.576272487640381,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1347,
      "step": 88
    },
    {
      "epoch": 0.05729921133107999,
      "grad_norm": 2.7399938106536865,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.0732,
      "step": 89
    },
    {
      "epoch": 0.05794302269435055,
      "grad_norm": 3.0096888542175293,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.9872,
      "step": 90
    },
    {
      "epoch": 0.058586834057621114,
      "grad_norm": 3.079171895980835,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.0885,
      "step": 91
    },
    {
      "epoch": 0.05923064542089168,
      "grad_norm": 2.9458725452423096,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2014,
      "step": 92
    },
    {
      "epoch": 0.05987445678416224,
      "grad_norm": 3.212573289871216,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.0985,
      "step": 93
    },
    {
      "epoch": 0.0605182681474328,
      "grad_norm": 3.55605411529541,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.2574,
      "step": 94
    },
    {
      "epoch": 0.06116207951070336,
      "grad_norm": 3.6658647060394287,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.18,
      "step": 95
    },
    {
      "epoch": 0.06180589087397392,
      "grad_norm": 3.6560888290405273,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.2603,
      "step": 96
    },
    {
      "epoch": 0.06244970223724449,
      "grad_norm": 4.703493595123291,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.3327,
      "step": 97
    },
    {
      "epoch": 0.06309351360051504,
      "grad_norm": 4.822589874267578,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.1194,
      "step": 98
    },
    {
      "epoch": 0.0637373249637856,
      "grad_norm": 6.346423625946045,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.4549,
      "step": 99
    },
    {
      "epoch": 0.06438113632705618,
      "grad_norm": 7.962794780731201,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.9194,
      "step": 100
    },
    {
      "epoch": 0.06438113632705618,
      "eval_loss": 0.28992101550102234,
      "eval_runtime": 215.3493,
      "eval_samples_per_second": 12.148,
      "eval_steps_per_second": 3.037,
      "step": 100
    },
    {
      "epoch": 0.06502494769032674,
      "grad_norm": 1.655432105064392,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.5385,
      "step": 101
    },
    {
      "epoch": 0.0656687590535973,
      "grad_norm": 1.6436342000961304,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.6655,
      "step": 102
    },
    {
      "epoch": 0.06631257041686786,
      "grad_norm": 1.8506100177764893,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.832,
      "step": 103
    },
    {
      "epoch": 0.06695638178013842,
      "grad_norm": 1.301578402519226,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.7436,
      "step": 104
    },
    {
      "epoch": 0.06760019314340898,
      "grad_norm": 1.4872246980667114,
      "learning_rate": 5e-05,
      "loss": 0.9278,
      "step": 105
    },
    {
      "epoch": 0.06824400450667954,
      "grad_norm": 1.491284966468811,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.8245,
      "step": 106
    },
    {
      "epoch": 0.0688878158699501,
      "grad_norm": 1.6801183223724365,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.972,
      "step": 107
    },
    {
      "epoch": 0.06953162723322066,
      "grad_norm": 1.4707728624343872,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.7708,
      "step": 108
    },
    {
      "epoch": 0.07017543859649122,
      "grad_norm": 1.550600290298462,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.8869,
      "step": 109
    },
    {
      "epoch": 0.07081924995976178,
      "grad_norm": 1.574897289276123,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.8787,
      "step": 110
    },
    {
      "epoch": 0.07146306132303235,
      "grad_norm": 1.89789879322052,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.0283,
      "step": 111
    },
    {
      "epoch": 0.07210687268630292,
      "grad_norm": 1.9088706970214844,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1371,
      "step": 112
    },
    {
      "epoch": 0.07275068404957348,
      "grad_norm": 2.1733453273773193,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.306,
      "step": 113
    },
    {
      "epoch": 0.07339449541284404,
      "grad_norm": 2.116964340209961,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.9345,
      "step": 114
    },
    {
      "epoch": 0.0740383067761146,
      "grad_norm": 1.9357224702835083,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.084,
      "step": 115
    },
    {
      "epoch": 0.07468211813938516,
      "grad_norm": 1.759748101234436,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.8709,
      "step": 116
    },
    {
      "epoch": 0.07532592950265572,
      "grad_norm": 2.208758592605591,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.1747,
      "step": 117
    },
    {
      "epoch": 0.07596974086592628,
      "grad_norm": 1.853180170059204,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.8755,
      "step": 118
    },
    {
      "epoch": 0.07661355222919684,
      "grad_norm": 2.052518606185913,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0298,
      "step": 119
    },
    {
      "epoch": 0.0772573635924674,
      "grad_norm": 2.3367223739624023,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3256,
      "step": 120
    },
    {
      "epoch": 0.07790117495573796,
      "grad_norm": 2.0620906352996826,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.9399,
      "step": 121
    },
    {
      "epoch": 0.07854498631900853,
      "grad_norm": 2.0867626667022705,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.9868,
      "step": 122
    },
    {
      "epoch": 0.07918879768227909,
      "grad_norm": 2.5150322914123535,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.1192,
      "step": 123
    },
    {
      "epoch": 0.07983260904554966,
      "grad_norm": 2.711125135421753,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.2906,
      "step": 124
    },
    {
      "epoch": 0.08047642040882022,
      "grad_norm": 2.1506714820861816,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.2054,
      "step": 125
    },
    {
      "epoch": 0.08112023177209078,
      "grad_norm": 2.3625669479370117,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.102,
      "step": 126
    },
    {
      "epoch": 0.08176404313536134,
      "grad_norm": 2.0529468059539795,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.0799,
      "step": 127
    },
    {
      "epoch": 0.0824078544986319,
      "grad_norm": 2.3510608673095703,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.1709,
      "step": 128
    },
    {
      "epoch": 0.08305166586190246,
      "grad_norm": 2.228466510772705,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.9181,
      "step": 129
    },
    {
      "epoch": 0.08369547722517302,
      "grad_norm": 2.654266595840454,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.3584,
      "step": 130
    },
    {
      "epoch": 0.08433928858844358,
      "grad_norm": 2.6218576431274414,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1335,
      "step": 131
    },
    {
      "epoch": 0.08498309995171414,
      "grad_norm": 2.3853259086608887,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.0661,
      "step": 132
    },
    {
      "epoch": 0.0856269113149847,
      "grad_norm": 2.351012945175171,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.878,
      "step": 133
    },
    {
      "epoch": 0.08627072267825527,
      "grad_norm": 2.1597161293029785,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.8321,
      "step": 134
    },
    {
      "epoch": 0.08691453404152583,
      "grad_norm": 3.058601140975952,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.3161,
      "step": 135
    },
    {
      "epoch": 0.0875583454047964,
      "grad_norm": 2.6464040279388428,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.1124,
      "step": 136
    },
    {
      "epoch": 0.08820215676806696,
      "grad_norm": 2.425506591796875,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.9552,
      "step": 137
    },
    {
      "epoch": 0.08884596813133752,
      "grad_norm": 2.6761343479156494,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.0614,
      "step": 138
    },
    {
      "epoch": 0.08948977949460808,
      "grad_norm": 2.7491350173950195,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.0011,
      "step": 139
    },
    {
      "epoch": 0.09013359085787864,
      "grad_norm": 3.208111047744751,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.036,
      "step": 140
    },
    {
      "epoch": 0.0907774022211492,
      "grad_norm": 3.00718092918396,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.1736,
      "step": 141
    },
    {
      "epoch": 0.09142121358441976,
      "grad_norm": 2.9277632236480713,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.049,
      "step": 142
    },
    {
      "epoch": 0.09206502494769032,
      "grad_norm": 3.495861053466797,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4006,
      "step": 143
    },
    {
      "epoch": 0.09270883631096088,
      "grad_norm": 2.722851037979126,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.984,
      "step": 144
    },
    {
      "epoch": 0.09335264767423145,
      "grad_norm": 3.886777400970459,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1988,
      "step": 145
    },
    {
      "epoch": 0.093996459037502,
      "grad_norm": 3.3488070964813232,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.392,
      "step": 146
    },
    {
      "epoch": 0.09464027040077258,
      "grad_norm": 4.061428070068359,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.3468,
      "step": 147
    },
    {
      "epoch": 0.09528408176404314,
      "grad_norm": 3.773306131362915,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1455,
      "step": 148
    },
    {
      "epoch": 0.0959278931273137,
      "grad_norm": 4.972281455993652,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.4787,
      "step": 149
    },
    {
      "epoch": 0.09657170449058426,
      "grad_norm": 12.073335647583008,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.0815,
      "step": 150
    },
    {
      "epoch": 0.09657170449058426,
      "eval_loss": 0.2579449415206909,
      "eval_runtime": 215.3156,
      "eval_samples_per_second": 12.15,
      "eval_steps_per_second": 3.037,
      "step": 150
    },
    {
      "epoch": 0.09721551585385482,
      "grad_norm": 1.0818828344345093,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.3422,
      "step": 151
    },
    {
      "epoch": 0.09785932721712538,
      "grad_norm": 1.2756608724594116,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.8004,
      "step": 152
    },
    {
      "epoch": 0.09850313858039594,
      "grad_norm": 1.5277647972106934,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.8189,
      "step": 153
    },
    {
      "epoch": 0.0991469499436665,
      "grad_norm": 1.5677112340927124,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.7028,
      "step": 154
    },
    {
      "epoch": 0.09979076130693706,
      "grad_norm": 1.6006295680999756,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.8025,
      "step": 155
    },
    {
      "epoch": 0.10043457267020763,
      "grad_norm": 1.6912732124328613,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.9051,
      "step": 156
    },
    {
      "epoch": 0.10107838403347819,
      "grad_norm": 1.667656660079956,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.9367,
      "step": 157
    },
    {
      "epoch": 0.10172219539674875,
      "grad_norm": 1.5066521167755127,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.8188,
      "step": 158
    },
    {
      "epoch": 0.10236600676001932,
      "grad_norm": 1.7178314924240112,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.8506,
      "step": 159
    },
    {
      "epoch": 0.10300981812328988,
      "grad_norm": 1.6823207139968872,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.8289,
      "step": 160
    },
    {
      "epoch": 0.10365362948656044,
      "grad_norm": 1.7377427816390991,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.0078,
      "step": 161
    },
    {
      "epoch": 0.104297440849831,
      "grad_norm": 1.9322346448898315,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0257,
      "step": 162
    },
    {
      "epoch": 0.10494125221310156,
      "grad_norm": 1.743464469909668,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.975,
      "step": 163
    },
    {
      "epoch": 0.10558506357637212,
      "grad_norm": 1.5859054327011108,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.7812,
      "step": 164
    },
    {
      "epoch": 0.10622887493964268,
      "grad_norm": 1.9096014499664307,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.9688,
      "step": 165
    },
    {
      "epoch": 0.10687268630291324,
      "grad_norm": 2.0253822803497314,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.0641,
      "step": 166
    },
    {
      "epoch": 0.1075164976661838,
      "grad_norm": 2.075273036956787,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.9957,
      "step": 167
    },
    {
      "epoch": 0.10816030902945437,
      "grad_norm": 2.046759843826294,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0708,
      "step": 168
    },
    {
      "epoch": 0.10880412039272493,
      "grad_norm": 1.9082063436508179,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.0296,
      "step": 169
    },
    {
      "epoch": 0.10944793175599549,
      "grad_norm": 1.8657265901565552,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.8831,
      "step": 170
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 2.117521047592163,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.9232,
      "step": 171
    },
    {
      "epoch": 0.11073555448253662,
      "grad_norm": 1.9781477451324463,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.9381,
      "step": 172
    },
    {
      "epoch": 0.11137936584580718,
      "grad_norm": 2.2230567932128906,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0224,
      "step": 173
    },
    {
      "epoch": 0.11202317720907774,
      "grad_norm": 2.08274507522583,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0277,
      "step": 174
    },
    {
      "epoch": 0.1126669885723483,
      "grad_norm": 2.2180800437927246,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.97,
      "step": 175
    },
    {
      "epoch": 0.11331079993561886,
      "grad_norm": 2.1694130897521973,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9793,
      "step": 176
    },
    {
      "epoch": 0.11395461129888942,
      "grad_norm": 2.3947696685791016,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.0581,
      "step": 177
    },
    {
      "epoch": 0.11459842266215998,
      "grad_norm": 2.3367292881011963,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.221,
      "step": 178
    },
    {
      "epoch": 0.11524223402543055,
      "grad_norm": 2.221740961074829,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.1238,
      "step": 179
    },
    {
      "epoch": 0.1158860453887011,
      "grad_norm": 2.7792890071868896,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.9372,
      "step": 180
    },
    {
      "epoch": 0.11652985675197167,
      "grad_norm": 2.6455326080322266,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.1726,
      "step": 181
    },
    {
      "epoch": 0.11717366811524223,
      "grad_norm": 2.2467410564422607,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.934,
      "step": 182
    },
    {
      "epoch": 0.1178174794785128,
      "grad_norm": 2.599761962890625,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.2378,
      "step": 183
    },
    {
      "epoch": 0.11846129084178336,
      "grad_norm": 2.3840746879577637,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.9829,
      "step": 184
    },
    {
      "epoch": 0.11910510220505392,
      "grad_norm": 2.811389923095703,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.4107,
      "step": 185
    },
    {
      "epoch": 0.11974891356832448,
      "grad_norm": 2.4712958335876465,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1729,
      "step": 186
    },
    {
      "epoch": 0.12039272493159504,
      "grad_norm": 2.361564874649048,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.9774,
      "step": 187
    },
    {
      "epoch": 0.1210365362948656,
      "grad_norm": 2.514312744140625,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0611,
      "step": 188
    },
    {
      "epoch": 0.12168034765813616,
      "grad_norm": 2.4720776081085205,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.1268,
      "step": 189
    },
    {
      "epoch": 0.12232415902140673,
      "grad_norm": 2.8340442180633545,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.1006,
      "step": 190
    },
    {
      "epoch": 0.12296797038467729,
      "grad_norm": 2.550647258758545,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.0606,
      "step": 191
    },
    {
      "epoch": 0.12361178174794785,
      "grad_norm": 2.700902223587036,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.9209,
      "step": 192
    },
    {
      "epoch": 0.12425559311121841,
      "grad_norm": 2.4640536308288574,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.8696,
      "step": 193
    },
    {
      "epoch": 0.12489940447448898,
      "grad_norm": 2.9509224891662598,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.1862,
      "step": 194
    },
    {
      "epoch": 0.12554321583775954,
      "grad_norm": 3.1793181896209717,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.1368,
      "step": 195
    },
    {
      "epoch": 0.1261870272010301,
      "grad_norm": 3.549727439880371,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.3533,
      "step": 196
    },
    {
      "epoch": 0.12683083856430066,
      "grad_norm": 3.851001024246216,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.4117,
      "step": 197
    },
    {
      "epoch": 0.1274746499275712,
      "grad_norm": 3.3996798992156982,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.1925,
      "step": 198
    },
    {
      "epoch": 0.12811846129084178,
      "grad_norm": 5.141419887542725,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.4439,
      "step": 199
    },
    {
      "epoch": 0.12876227265411236,
      "grad_norm": 7.753669738769531,
      "learning_rate": 0.0,
      "loss": 1.867,
      "step": 200
    },
    {
      "epoch": 0.12876227265411236,
      "eval_loss": 0.2527673840522766,
      "eval_runtime": 215.5969,
      "eval_samples_per_second": 12.134,
      "eval_steps_per_second": 3.033,
      "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": 3.123156669777838e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}