{
  "best_metric": 0.28693631291389465,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 0.023144576454250887,
  "eval_steps": 50,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00015429717636167257,
      "grad_norm": 12.967491149902344,
      "learning_rate": 1e-05,
      "loss": 4.608,
      "step": 1
    },
    {
      "epoch": 0.00015429717636167257,
      "eval_loss": 3.070103406906128,
      "eval_runtime": 1161.2284,
      "eval_samples_per_second": 9.4,
      "eval_steps_per_second": 2.35,
      "step": 1
    },
    {
      "epoch": 0.00030859435272334514,
      "grad_norm": 16.873348236083984,
      "learning_rate": 2e-05,
      "loss": 5.5859,
      "step": 2
    },
    {
      "epoch": 0.00046289152908501776,
      "grad_norm": 18.696292877197266,
      "learning_rate": 3e-05,
      "loss": 5.4013,
      "step": 3
    },
    {
      "epoch": 0.0006171887054466903,
      "grad_norm": 32.20201873779297,
      "learning_rate": 4e-05,
      "loss": 4.9054,
      "step": 4
    },
    {
      "epoch": 0.0007714858818083629,
      "grad_norm": 30.779666900634766,
      "learning_rate": 5e-05,
      "loss": 3.3895,
      "step": 5
    },
    {
      "epoch": 0.0009257830581700355,
      "grad_norm": 12.106324195861816,
      "learning_rate": 6e-05,
      "loss": 2.0783,
      "step": 6
    },
    {
      "epoch": 0.001080080234531708,
      "grad_norm": 16.368005752563477,
      "learning_rate": 7e-05,
      "loss": 2.0905,
      "step": 7
    },
    {
      "epoch": 0.0012343774108933806,
      "grad_norm": 12.284082412719727,
      "learning_rate": 8e-05,
      "loss": 2.0212,
      "step": 8
    },
    {
      "epoch": 0.0013886745872550533,
      "grad_norm": 4.247934341430664,
      "learning_rate": 9e-05,
      "loss": 1.8962,
      "step": 9
    },
    {
      "epoch": 0.0015429717636167258,
      "grad_norm": 4.680515289306641,
      "learning_rate": 0.0001,
      "loss": 1.9638,
      "step": 10
    },
    {
      "epoch": 0.0016972689399783983,
      "grad_norm": 6.497641086578369,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.6292,
      "step": 11
    },
    {
      "epoch": 0.001851566116340071,
      "grad_norm": 9.55294132232666,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8568,
      "step": 12
    },
    {
      "epoch": 0.0020058632927017436,
      "grad_norm": 4.845419883728027,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.5581,
      "step": 13
    },
    {
      "epoch": 0.002160160469063416,
      "grad_norm": 5.300878524780273,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.8431,
      "step": 14
    },
    {
      "epoch": 0.0023144576454250886,
      "grad_norm": 5.047299861907959,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.5672,
      "step": 15
    },
    {
      "epoch": 0.002468754821786761,
      "grad_norm": 4.657147407531738,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.573,
      "step": 16
    },
    {
      "epoch": 0.002623051998148434,
      "grad_norm": 4.983191013336182,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.6145,
      "step": 17
    },
    {
      "epoch": 0.0027773491745101066,
      "grad_norm": 5.245067119598389,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.981,
      "step": 18
    },
    {
      "epoch": 0.002931646350871779,
      "grad_norm": 4.325369834899902,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.2232,
      "step": 19
    },
    {
      "epoch": 0.0030859435272334516,
      "grad_norm": 4.289376735687256,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3971,
      "step": 20
    },
    {
      "epoch": 0.003240240703595124,
      "grad_norm": 3.2687885761260986,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.0999,
      "step": 21
    },
    {
      "epoch": 0.0033945378799567967,
      "grad_norm": 6.781163692474365,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.7097,
      "step": 22
    },
    {
      "epoch": 0.003548835056318469,
      "grad_norm": 4.246693134307861,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.2993,
      "step": 23
    },
    {
      "epoch": 0.003703132232680142,
      "grad_norm": 3.9915480613708496,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.3736,
      "step": 24
    },
    {
      "epoch": 0.0038574294090418146,
      "grad_norm": 5.187669277191162,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.8441,
      "step": 25
    },
    {
      "epoch": 0.004011726585403487,
      "grad_norm": 4.719375133514404,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.3637,
      "step": 26
    },
    {
      "epoch": 0.00416602376176516,
      "grad_norm": 5.235927104949951,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5306,
      "step": 27
    },
    {
      "epoch": 0.004320320938126832,
      "grad_norm": 8.455881118774414,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.6825,
      "step": 28
    },
    {
      "epoch": 0.004474618114488505,
      "grad_norm": 5.1970672607421875,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.9581,
      "step": 29
    },
    {
      "epoch": 0.004628915290850177,
      "grad_norm": 4.755765914916992,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.9184,
      "step": 30
    },
    {
      "epoch": 0.00478321246721185,
      "grad_norm": 5.133935928344727,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.107,
      "step": 31
    },
    {
      "epoch": 0.004937509643573522,
      "grad_norm": 5.140307903289795,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.8789,
      "step": 32
    },
    {
      "epoch": 0.005091806819935195,
      "grad_norm": 5.53304386138916,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.8521,
      "step": 33
    },
    {
      "epoch": 0.005246103996296868,
      "grad_norm": 4.627854347229004,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.1042,
      "step": 34
    },
    {
      "epoch": 0.00540040117265854,
      "grad_norm": 5.7089056968688965,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.2423,
      "step": 35
    },
    {
      "epoch": 0.005554698349020213,
      "grad_norm": 4.635379314422607,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.958,
      "step": 36
    },
    {
      "epoch": 0.005708995525381885,
      "grad_norm": 4.911176681518555,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.8614,
      "step": 37
    },
    {
      "epoch": 0.005863292701743558,
      "grad_norm": 4.997920513153076,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.6707,
      "step": 38
    },
    {
      "epoch": 0.00601758987810523,
      "grad_norm": 3.992204427719116,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.5909,
      "step": 39
    },
    {
      "epoch": 0.006171887054466903,
      "grad_norm": 5.433296203613281,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.2718,
      "step": 40
    },
    {
      "epoch": 0.006326184230828576,
      "grad_norm": 4.329758644104004,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.7311,
      "step": 41
    },
    {
      "epoch": 0.006480481407190248,
      "grad_norm": 5.09123420715332,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.9057,
      "step": 42
    },
    {
      "epoch": 0.006634778583551921,
      "grad_norm": 5.60385799407959,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.305,
      "step": 43
    },
    {
      "epoch": 0.006789075759913593,
      "grad_norm": 5.289751052856445,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5586,
      "step": 44
    },
    {
      "epoch": 0.006943372936275266,
      "grad_norm": 5.216174602508545,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.7909,
      "step": 45
    },
    {
      "epoch": 0.007097670112636938,
      "grad_norm": 10.532437324523926,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.2109,
      "step": 46
    },
    {
      "epoch": 0.007251967288998611,
      "grad_norm": 6.646415710449219,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.0328,
      "step": 47
    },
    {
      "epoch": 0.007406264465360284,
      "grad_norm": 6.463160514831543,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.8959,
      "step": 48
    },
    {
      "epoch": 0.007560561641721956,
      "grad_norm": 8.218077659606934,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.99,
      "step": 49
    },
    {
      "epoch": 0.007714858818083629,
      "grad_norm": 6.032686710357666,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.5979,
      "step": 50
    },
    {
      "epoch": 0.007714858818083629,
      "eval_loss": 0.39734092354774475,
      "eval_runtime": 1166.4255,
      "eval_samples_per_second": 9.359,
      "eval_steps_per_second": 2.34,
      "step": 50
    },
    {
      "epoch": 0.007869155994445301,
      "grad_norm": 3.748234748840332,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.9452,
      "step": 51
    },
    {
      "epoch": 0.008023453170806974,
      "grad_norm": 8.013319969177246,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.8907,
      "step": 52
    },
    {
      "epoch": 0.008177750347168647,
      "grad_norm": 2.42112135887146,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.9899,
      "step": 53
    },
    {
      "epoch": 0.00833204752353032,
      "grad_norm": 2.3565921783447266,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.1361,
      "step": 54
    },
    {
      "epoch": 0.008486344699891991,
      "grad_norm": 4.064480781555176,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.1217,
      "step": 55
    },
    {
      "epoch": 0.008640641876253664,
      "grad_norm": 2.9545719623565674,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.0962,
      "step": 56
    },
    {
      "epoch": 0.008794939052615337,
      "grad_norm": 2.3463966846466064,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.9859,
      "step": 57
    },
    {
      "epoch": 0.00894923622897701,
      "grad_norm": 3.115272045135498,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.0806,
      "step": 58
    },
    {
      "epoch": 0.009103533405338681,
      "grad_norm": 2.863109827041626,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.1752,
      "step": 59
    },
    {
      "epoch": 0.009257830581700354,
      "grad_norm": 2.607872724533081,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.1884,
      "step": 60
    },
    {
      "epoch": 0.009412127758062027,
      "grad_norm": 2.4852263927459717,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.9903,
      "step": 61
    },
    {
      "epoch": 0.0095664249344237,
      "grad_norm": 2.4692223072052,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.1448,
      "step": 62
    },
    {
      "epoch": 0.009720722110785373,
      "grad_norm": 2.596566915512085,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.0819,
      "step": 63
    },
    {
      "epoch": 0.009875019287147044,
      "grad_norm": 2.9963595867156982,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.9952,
      "step": 64
    },
    {
      "epoch": 0.010029316463508717,
      "grad_norm": 15.808581352233887,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.0957,
      "step": 65
    },
    {
      "epoch": 0.01018361363987039,
      "grad_norm": 3.4665415287017822,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3545,
      "step": 66
    },
    {
      "epoch": 0.010337910816232063,
      "grad_norm": 4.240460395812988,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.4656,
      "step": 67
    },
    {
      "epoch": 0.010492207992593736,
      "grad_norm": 3.0784738063812256,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.227,
      "step": 68
    },
    {
      "epoch": 0.010646505168955408,
      "grad_norm": 3.0800323486328125,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.1006,
      "step": 69
    },
    {
      "epoch": 0.01080080234531708,
      "grad_norm": 3.473702907562256,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.2243,
      "step": 70
    },
    {
      "epoch": 0.010955099521678753,
      "grad_norm": 3.2348060607910156,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.05,
      "step": 71
    },
    {
      "epoch": 0.011109396698040426,
      "grad_norm": 4.6401047706604,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6483,
      "step": 72
    },
    {
      "epoch": 0.0112636938744021,
      "grad_norm": 2.943392515182495,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.1169,
      "step": 73
    },
    {
      "epoch": 0.01141799105076377,
      "grad_norm": 3.2355434894561768,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.2571,
      "step": 74
    },
    {
      "epoch": 0.011572288227125443,
      "grad_norm": 3.3314385414123535,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4786,
      "step": 75
    },
    {
      "epoch": 0.011726585403487116,
      "grad_norm": 3.3201792240142822,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.2187,
      "step": 76
    },
    {
      "epoch": 0.01188088257984879,
      "grad_norm": 3.2873687744140625,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2086,
      "step": 77
    },
    {
      "epoch": 0.01203517975621046,
      "grad_norm": 4.427672386169434,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.4086,
      "step": 78
    },
    {
      "epoch": 0.012189476932572134,
      "grad_norm": 4.180029392242432,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.4025,
      "step": 79
    },
    {
      "epoch": 0.012343774108933806,
      "grad_norm": 3.9144394397735596,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.5826,
      "step": 80
    },
    {
      "epoch": 0.01249807128529548,
      "grad_norm": 9.054929733276367,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.1001,
      "step": 81
    },
    {
      "epoch": 0.012652368461657152,
      "grad_norm": 4.338362216949463,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.403,
      "step": 82
    },
    {
      "epoch": 0.012806665638018824,
      "grad_norm": 3.729764699935913,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.1555,
      "step": 83
    },
    {
      "epoch": 0.012960962814380497,
      "grad_norm": 4.436242580413818,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.5528,
      "step": 84
    },
    {
      "epoch": 0.01311525999074217,
      "grad_norm": 3.7678236961364746,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.4008,
      "step": 85
    },
    {
      "epoch": 0.013269557167103842,
      "grad_norm": 3.5236315727233887,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4119,
      "step": 86
    },
    {
      "epoch": 0.013423854343465515,
      "grad_norm": 5.569118976593018,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.6854,
      "step": 87
    },
    {
      "epoch": 0.013578151519827187,
      "grad_norm": 3.5289385318756104,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.2744,
      "step": 88
    },
    {
      "epoch": 0.01373244869618886,
      "grad_norm": 5.369927406311035,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.9102,
      "step": 89
    },
    {
      "epoch": 0.013886745872550533,
      "grad_norm": 5.753623962402344,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.9137,
      "step": 90
    },
    {
      "epoch": 0.014041043048912205,
      "grad_norm": 4.6428914070129395,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.6249,
      "step": 91
    },
    {
      "epoch": 0.014195340225273877,
      "grad_norm": 4.946249485015869,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.8101,
      "step": 92
    },
    {
      "epoch": 0.01434963740163555,
      "grad_norm": 4.09445858001709,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.5066,
      "step": 93
    },
    {
      "epoch": 0.014503934577997223,
      "grad_norm": 4.676825523376465,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.6826,
      "step": 94
    },
    {
      "epoch": 0.014658231754358896,
      "grad_norm": 4.851978302001953,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.564,
      "step": 95
    },
    {
      "epoch": 0.014812528930720568,
      "grad_norm": 4.22647762298584,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.654,
      "step": 96
    },
    {
      "epoch": 0.01496682610708224,
      "grad_norm": 6.007287502288818,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.8697,
      "step": 97
    },
    {
      "epoch": 0.015121123283443913,
      "grad_norm": 5.582359313964844,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.7139,
      "step": 98
    },
    {
      "epoch": 0.015275420459805586,
      "grad_norm": 5.950445175170898,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.8765,
      "step": 99
    },
    {
      "epoch": 0.015429717636167259,
      "grad_norm": 8.614642143249512,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.2272,
      "step": 100
    },
    {
      "epoch": 0.015429717636167259,
      "eval_loss": 0.3122243583202362,
      "eval_runtime": 1165.3462,
      "eval_samples_per_second": 9.367,
      "eval_steps_per_second": 2.342,
      "step": 100
    },
    {
      "epoch": 0.015584014812528931,
      "grad_norm": 1.8591800928115845,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.9351,
      "step": 101
    },
    {
      "epoch": 0.015738311988890603,
      "grad_norm": 2.8303134441375732,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.8034,
      "step": 102
    },
    {
      "epoch": 0.015892609165252276,
      "grad_norm": 2.5067758560180664,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.9066,
      "step": 103
    },
    {
      "epoch": 0.01604690634161395,
      "grad_norm": 2.0450499057769775,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.8944,
      "step": 104
    },
    {
      "epoch": 0.01620120351797562,
      "grad_norm": 1.7207564115524292,
      "learning_rate": 5e-05,
      "loss": 0.7016,
      "step": 105
    },
    {
      "epoch": 0.016355500694337295,
      "grad_norm": 2.03654408454895,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.8353,
      "step": 106
    },
    {
      "epoch": 0.016509797870698967,
      "grad_norm": 2.0523147583007812,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.7695,
      "step": 107
    },
    {
      "epoch": 0.01666409504706064,
      "grad_norm": 2.134726047515869,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.6979,
      "step": 108
    },
    {
      "epoch": 0.01681839222342231,
      "grad_norm": 2.85213565826416,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.2285,
      "step": 109
    },
    {
      "epoch": 0.016972689399783983,
      "grad_norm": 2.42657470703125,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.951,
      "step": 110
    },
    {
      "epoch": 0.017126986576145656,
      "grad_norm": 2.6871676445007324,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.0429,
      "step": 111
    },
    {
      "epoch": 0.01728128375250733,
      "grad_norm": 3.100673198699951,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.0275,
      "step": 112
    },
    {
      "epoch": 0.017435580928869,
      "grad_norm": 2.3215911388397217,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.9375,
      "step": 113
    },
    {
      "epoch": 0.017589878105230675,
      "grad_norm": 2.6662847995758057,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.9412,
      "step": 114
    },
    {
      "epoch": 0.017744175281592348,
      "grad_norm": 2.5803070068359375,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.0703,
      "step": 115
    },
    {
      "epoch": 0.01789847245795402,
      "grad_norm": 3.0866687297821045,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.8465,
      "step": 116
    },
    {
      "epoch": 0.018052769634315693,
      "grad_norm": 1.8994590044021606,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.7841,
      "step": 117
    },
    {
      "epoch": 0.018207066810677363,
      "grad_norm": 2.8474042415618896,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.0694,
      "step": 118
    },
    {
      "epoch": 0.018361363987039036,
      "grad_norm": 3.5810625553131104,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2461,
      "step": 119
    },
    {
      "epoch": 0.01851566116340071,
      "grad_norm": 2.151620626449585,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.7699,
      "step": 120
    },
    {
      "epoch": 0.018669958339762382,
      "grad_norm": 2.897122383117676,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.9868,
      "step": 121
    },
    {
      "epoch": 0.018824255516124055,
      "grad_norm": 2.601444959640503,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.9727,
      "step": 122
    },
    {
      "epoch": 0.018978552692485728,
      "grad_norm": 3.0749759674072266,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.0328,
      "step": 123
    },
    {
      "epoch": 0.0191328498688474,
      "grad_norm": 2.693816900253296,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.9484,
      "step": 124
    },
    {
      "epoch": 0.019287147045209074,
      "grad_norm": 3.4224655628204346,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.9939,
      "step": 125
    },
    {
      "epoch": 0.019441444221570747,
      "grad_norm": 3.3732192516326904,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.0497,
      "step": 126
    },
    {
      "epoch": 0.01959574139793242,
      "grad_norm": 3.816964626312256,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.2592,
      "step": 127
    },
    {
      "epoch": 0.01975003857429409,
      "grad_norm": 3.1708836555480957,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.0902,
      "step": 128
    },
    {
      "epoch": 0.019904335750655762,
      "grad_norm": 3.912808418273926,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.3219,
      "step": 129
    },
    {
      "epoch": 0.020058632927017435,
      "grad_norm": 3.3545548915863037,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.1524,
      "step": 130
    },
    {
      "epoch": 0.020212930103379108,
      "grad_norm": 4.286497592926025,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.4666,
      "step": 131
    },
    {
      "epoch": 0.02036722727974078,
      "grad_norm": 4.28722620010376,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.4111,
      "step": 132
    },
    {
      "epoch": 0.020521524456102454,
      "grad_norm": 3.121302604675293,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.2172,
      "step": 133
    },
    {
      "epoch": 0.020675821632464127,
      "grad_norm": 3.8379640579223633,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4888,
      "step": 134
    },
    {
      "epoch": 0.0208301188088258,
      "grad_norm": 3.8402748107910156,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4676,
      "step": 135
    },
    {
      "epoch": 0.020984415985187473,
      "grad_norm": 3.382737159729004,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.2978,
      "step": 136
    },
    {
      "epoch": 0.021138713161549142,
      "grad_norm": 3.805022716522217,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.6522,
      "step": 137
    },
    {
      "epoch": 0.021293010337910815,
      "grad_norm": 3.469876766204834,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.0157,
      "step": 138
    },
    {
      "epoch": 0.021447307514272488,
      "grad_norm": 4.098129749298096,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.5612,
      "step": 139
    },
    {
      "epoch": 0.02160160469063416,
      "grad_norm": 5.493364334106445,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.425,
      "step": 140
    },
    {
      "epoch": 0.021755901866995834,
      "grad_norm": 3.308091878890991,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.1539,
      "step": 141
    },
    {
      "epoch": 0.021910199043357507,
      "grad_norm": 5.106802463531494,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.0486,
      "step": 142
    },
    {
      "epoch": 0.02206449621971918,
      "grad_norm": 3.9729108810424805,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.6753,
      "step": 143
    },
    {
      "epoch": 0.022218793396080853,
      "grad_norm": 4.53525447845459,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.6125,
      "step": 144
    },
    {
      "epoch": 0.022373090572442526,
      "grad_norm": 4.052093982696533,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.5345,
      "step": 145
    },
    {
      "epoch": 0.0225273877488042,
      "grad_norm": 4.750132083892822,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.7757,
      "step": 146
    },
    {
      "epoch": 0.022681684925165868,
      "grad_norm": 4.727255344390869,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.2851,
      "step": 147
    },
    {
      "epoch": 0.02283598210152754,
      "grad_norm": 4.936407566070557,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.653,
      "step": 148
    },
    {
      "epoch": 0.022990279277889214,
      "grad_norm": 5.240731716156006,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.3967,
      "step": 149
    },
    {
      "epoch": 0.023144576454250887,
      "grad_norm": 6.619349002838135,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.9061,
      "step": 150
    },
    {
      "epoch": 0.023144576454250887,
      "eval_loss": 0.28693631291389465,
      "eval_runtime": 1164.6298,
      "eval_samples_per_second": 9.373,
      "eval_steps_per_second": 2.343,
      "step": 150
    }
  ],
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.48136509800448e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}