{
  "best_metric": 1.07815420627594,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.11114198388441233,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005557099194220616,
      "grad_norm": 0.5824400186538696,
      "learning_rate": 1e-05,
      "loss": 0.9854,
      "step": 1
    },
    {
      "epoch": 0.0005557099194220616,
      "eval_loss": 1.5876773595809937,
      "eval_runtime": 72.4591,
      "eval_samples_per_second": 41.83,
      "eval_steps_per_second": 10.461,
      "step": 1
    },
    {
      "epoch": 0.0011114198388441233,
      "grad_norm": 0.7623146772384644,
      "learning_rate": 2e-05,
      "loss": 1.075,
      "step": 2
    },
    {
      "epoch": 0.0016671297582661851,
      "grad_norm": 0.5391097068786621,
      "learning_rate": 3e-05,
      "loss": 1.0145,
      "step": 3
    },
    {
      "epoch": 0.0022228396776882466,
      "grad_norm": 0.5953089594841003,
      "learning_rate": 4e-05,
      "loss": 1.1138,
      "step": 4
    },
    {
      "epoch": 0.0027785495971103086,
      "grad_norm": 0.5542176365852356,
      "learning_rate": 5e-05,
      "loss": 1.0475,
      "step": 5
    },
    {
      "epoch": 0.0033342595165323703,
      "grad_norm": 0.4748373329639435,
      "learning_rate": 6e-05,
      "loss": 1.0721,
      "step": 6
    },
    {
      "epoch": 0.003889969435954432,
      "grad_norm": 0.4489298462867737,
      "learning_rate": 7e-05,
      "loss": 1.0362,
      "step": 7
    },
    {
      "epoch": 0.004445679355376493,
      "grad_norm": 0.5009029507637024,
      "learning_rate": 8e-05,
      "loss": 1.085,
      "step": 8
    },
    {
      "epoch": 0.005001389274798555,
      "grad_norm": 0.4855598211288452,
      "learning_rate": 9e-05,
      "loss": 1.0349,
      "step": 9
    },
    {
      "epoch": 0.005557099194220617,
      "grad_norm": 0.45389288663864136,
      "learning_rate": 0.0001,
      "loss": 1.104,
      "step": 10
    },
    {
      "epoch": 0.006112809113642679,
      "grad_norm": 0.42139554023742676,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.0432,
      "step": 11
    },
    {
      "epoch": 0.0066685190330647405,
      "grad_norm": 0.4515708386898041,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.116,
      "step": 12
    },
    {
      "epoch": 0.007224228952486802,
      "grad_norm": 0.48064443469047546,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.0675,
      "step": 13
    },
    {
      "epoch": 0.007779938871908864,
      "grad_norm": 0.4572497010231018,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.1427,
      "step": 14
    },
    {
      "epoch": 0.008335648791330925,
      "grad_norm": 0.438105970621109,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.1194,
      "step": 15
    },
    {
      "epoch": 0.008891358710752986,
      "grad_norm": 0.39443475008010864,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.1491,
      "step": 16
    },
    {
      "epoch": 0.009447068630175049,
      "grad_norm": 0.4367530643939972,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1117,
      "step": 17
    },
    {
      "epoch": 0.01000277854959711,
      "grad_norm": 0.48059606552124023,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.0913,
      "step": 18
    },
    {
      "epoch": 0.010558488469019172,
      "grad_norm": 0.3961222767829895,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.1326,
      "step": 19
    },
    {
      "epoch": 0.011114198388441235,
      "grad_norm": 0.3907172977924347,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.1345,
      "step": 20
    },
    {
      "epoch": 0.011669908307863295,
      "grad_norm": 0.5000215172767639,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.0895,
      "step": 21
    },
    {
      "epoch": 0.012225618227285358,
      "grad_norm": 0.3938838839530945,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.2404,
      "step": 22
    },
    {
      "epoch": 0.012781328146707419,
      "grad_norm": 0.40758854150772095,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1137,
      "step": 23
    },
    {
      "epoch": 0.013337038066129481,
      "grad_norm": 0.422472208738327,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.033,
      "step": 24
    },
    {
      "epoch": 0.013892747985551542,
      "grad_norm": 0.4575078785419464,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.1317,
      "step": 25
    },
    {
      "epoch": 0.014448457904973604,
      "grad_norm": 0.4240778386592865,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.1015,
      "step": 26
    },
    {
      "epoch": 0.015004167824395665,
      "grad_norm": 0.44602343440055847,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.2184,
      "step": 27
    },
    {
      "epoch": 0.015559877743817728,
      "grad_norm": 0.4262112081050873,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.1052,
      "step": 28
    },
    {
      "epoch": 0.01611558766323979,
      "grad_norm": 0.4655727744102478,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0972,
      "step": 29
    },
    {
      "epoch": 0.01667129758266185,
      "grad_norm": 0.5123158693313599,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.123,
      "step": 30
    },
    {
      "epoch": 0.01722700750208391,
      "grad_norm": 0.488567054271698,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.1982,
      "step": 31
    },
    {
      "epoch": 0.017782717421505972,
      "grad_norm": 0.5412582755088806,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.291,
      "step": 32
    },
    {
      "epoch": 0.018338427340928037,
      "grad_norm": 0.5219689607620239,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2081,
      "step": 33
    },
    {
      "epoch": 0.018894137260350097,
      "grad_norm": 0.5576116442680359,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.1646,
      "step": 34
    },
    {
      "epoch": 0.019449847179772158,
      "grad_norm": 0.5750767588615417,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.1442,
      "step": 35
    },
    {
      "epoch": 0.02000555709919422,
      "grad_norm": 0.6544929146766663,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.2047,
      "step": 36
    },
    {
      "epoch": 0.020561267018616283,
      "grad_norm": 0.7206011414527893,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3443,
      "step": 37
    },
    {
      "epoch": 0.021116976938038344,
      "grad_norm": 0.6689258217811584,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.13,
      "step": 38
    },
    {
      "epoch": 0.021672686857460405,
      "grad_norm": 0.7947595119476318,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.3251,
      "step": 39
    },
    {
      "epoch": 0.02222839677688247,
      "grad_norm": 0.7253210544586182,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.2069,
      "step": 40
    },
    {
      "epoch": 0.02278410669630453,
      "grad_norm": 0.7657760381698608,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1924,
      "step": 41
    },
    {
      "epoch": 0.02333981661572659,
      "grad_norm": 0.788730800151825,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.264,
      "step": 42
    },
    {
      "epoch": 0.02389552653514865,
      "grad_norm": 0.9794409871101379,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.406,
      "step": 43
    },
    {
      "epoch": 0.024451236454570716,
      "grad_norm": 0.9536454081535339,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.3045,
      "step": 44
    },
    {
      "epoch": 0.025006946373992776,
      "grad_norm": 1.371140956878662,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.3805,
      "step": 45
    },
    {
      "epoch": 0.025562656293414837,
      "grad_norm": 1.137738585472107,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.2789,
      "step": 46
    },
    {
      "epoch": 0.026118366212836898,
      "grad_norm": 1.1242486238479614,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.2741,
      "step": 47
    },
    {
      "epoch": 0.026674076132258962,
      "grad_norm": 1.3122904300689697,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.2571,
      "step": 48
    },
    {
      "epoch": 0.027229786051681023,
      "grad_norm": 1.3932980298995972,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.354,
      "step": 49
    },
    {
      "epoch": 0.027785495971103084,
      "grad_norm": 2.2601118087768555,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4594,
      "step": 50
    },
    {
      "epoch": 0.027785495971103084,
      "eval_loss": 1.1749505996704102,
      "eval_runtime": 72.5207,
      "eval_samples_per_second": 41.795,
      "eval_steps_per_second": 10.452,
      "step": 50
    },
    {
      "epoch": 0.028341205890525144,
      "grad_norm": 0.8685677647590637,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.9233,
      "step": 51
    },
    {
      "epoch": 0.02889691580994721,
      "grad_norm": 0.652723491191864,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.8858,
      "step": 52
    },
    {
      "epoch": 0.02945262572936927,
      "grad_norm": 0.4941800832748413,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.9968,
      "step": 53
    },
    {
      "epoch": 0.03000833564879133,
      "grad_norm": 0.3875123858451843,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.9955,
      "step": 54
    },
    {
      "epoch": 0.03056404556821339,
      "grad_norm": 0.34053778648376465,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.9355,
      "step": 55
    },
    {
      "epoch": 0.031119755487635455,
      "grad_norm": 0.32027530670166016,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.9261,
      "step": 56
    },
    {
      "epoch": 0.03167546540705751,
      "grad_norm": 0.30107083916664124,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.9832,
      "step": 57
    },
    {
      "epoch": 0.03223117532647958,
      "grad_norm": 0.334947407245636,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.0217,
      "step": 58
    },
    {
      "epoch": 0.03278688524590164,
      "grad_norm": 0.309049516916275,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.9614,
      "step": 59
    },
    {
      "epoch": 0.0333425951653237,
      "grad_norm": 0.3308800756931305,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.9614,
      "step": 60
    },
    {
      "epoch": 0.03389830508474576,
      "grad_norm": 0.3481154143810272,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.0493,
      "step": 61
    },
    {
      "epoch": 0.03445401500416782,
      "grad_norm": 0.34031444787979126,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.0398,
      "step": 62
    },
    {
      "epoch": 0.035009724923589884,
      "grad_norm": 0.32300055027008057,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.0842,
      "step": 63
    },
    {
      "epoch": 0.035565434843011945,
      "grad_norm": 0.33104395866394043,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.0351,
      "step": 64
    },
    {
      "epoch": 0.03612114476243401,
      "grad_norm": 0.32372066378593445,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.0475,
      "step": 65
    },
    {
      "epoch": 0.03667685468185607,
      "grad_norm": 0.3336024582386017,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.0588,
      "step": 66
    },
    {
      "epoch": 0.037232564601278134,
      "grad_norm": 0.30212101340293884,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0105,
      "step": 67
    },
    {
      "epoch": 0.037788274520700195,
      "grad_norm": 0.33015307784080505,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.0991,
      "step": 68
    },
    {
      "epoch": 0.038343984440122256,
      "grad_norm": 0.3651010990142822,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.032,
      "step": 69
    },
    {
      "epoch": 0.038899694359544316,
      "grad_norm": 0.3300033509731293,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.0194,
      "step": 70
    },
    {
      "epoch": 0.03945540427896638,
      "grad_norm": 0.34712526202201843,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.1186,
      "step": 71
    },
    {
      "epoch": 0.04001111419838844,
      "grad_norm": 0.3352060616016388,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1779,
      "step": 72
    },
    {
      "epoch": 0.040566824117810506,
      "grad_norm": 0.30711618065834045,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.048,
      "step": 73
    },
    {
      "epoch": 0.041122534037232567,
      "grad_norm": 0.34901267290115356,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.0142,
      "step": 74
    },
    {
      "epoch": 0.04167824395665463,
      "grad_norm": 0.363225519657135,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1125,
      "step": 75
    },
    {
      "epoch": 0.04223395387607669,
      "grad_norm": 0.3958677053451538,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.0855,
      "step": 76
    },
    {
      "epoch": 0.04278966379549875,
      "grad_norm": 0.3785785436630249,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.1018,
      "step": 77
    },
    {
      "epoch": 0.04334537371492081,
      "grad_norm": 0.4093249440193176,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.1562,
      "step": 78
    },
    {
      "epoch": 0.04390108363434287,
      "grad_norm": 0.3871191143989563,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1055,
      "step": 79
    },
    {
      "epoch": 0.04445679355376494,
      "grad_norm": 0.45775848627090454,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.1413,
      "step": 80
    },
    {
      "epoch": 0.045012503473187,
      "grad_norm": 0.49375495314598083,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.1434,
      "step": 81
    },
    {
      "epoch": 0.04556821339260906,
      "grad_norm": 0.4653330445289612,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.1428,
      "step": 82
    },
    {
      "epoch": 0.04612392331203112,
      "grad_norm": 0.5244958400726318,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2182,
      "step": 83
    },
    {
      "epoch": 0.04667963323145318,
      "grad_norm": 0.515475332736969,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.0964,
      "step": 84
    },
    {
      "epoch": 0.04723534315087524,
      "grad_norm": 0.5070421099662781,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.1998,
      "step": 85
    },
    {
      "epoch": 0.0477910530702973,
      "grad_norm": 0.5222674012184143,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.1623,
      "step": 86
    },
    {
      "epoch": 0.048346762989719363,
      "grad_norm": 0.5940301418304443,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.135,
      "step": 87
    },
    {
      "epoch": 0.04890247290914143,
      "grad_norm": 0.6131709218025208,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1874,
      "step": 88
    },
    {
      "epoch": 0.04945818282856349,
      "grad_norm": 0.685854434967041,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.1919,
      "step": 89
    },
    {
      "epoch": 0.05001389274798555,
      "grad_norm": 0.6810467839241028,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2603,
      "step": 90
    },
    {
      "epoch": 0.050569602667407614,
      "grad_norm": 0.8020808100700378,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.2785,
      "step": 91
    },
    {
      "epoch": 0.051125312586829674,
      "grad_norm": 0.7462509274482727,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2539,
      "step": 92
    },
    {
      "epoch": 0.051681022506251735,
      "grad_norm": 0.8378671407699585,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.2623,
      "step": 93
    },
    {
      "epoch": 0.052236732425673796,
      "grad_norm": 0.7647984027862549,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.2319,
      "step": 94
    },
    {
      "epoch": 0.05279244234509586,
      "grad_norm": 0.8100523352622986,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1638,
      "step": 95
    },
    {
      "epoch": 0.053348152264517924,
      "grad_norm": 0.9509829878807068,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.2408,
      "step": 96
    },
    {
      "epoch": 0.053903862183939985,
      "grad_norm": 0.9600633978843689,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.3337,
      "step": 97
    },
    {
      "epoch": 0.054459572103362046,
      "grad_norm": 1.0256901979446411,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.3092,
      "step": 98
    },
    {
      "epoch": 0.05501528202278411,
      "grad_norm": 1.230337142944336,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.3623,
      "step": 99
    },
    {
      "epoch": 0.05557099194220617,
      "grad_norm": 1.3636703491210938,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.4335,
      "step": 100
    },
    {
      "epoch": 0.05557099194220617,
      "eval_loss": 1.1121288537979126,
      "eval_runtime": 72.6261,
      "eval_samples_per_second": 41.734,
      "eval_steps_per_second": 10.437,
      "step": 100
    },
    {
      "epoch": 0.05612670186162823,
      "grad_norm": 0.35126662254333496,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.7206,
      "step": 101
    },
    {
      "epoch": 0.05668241178105029,
      "grad_norm": 0.3719412386417389,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.8151,
      "step": 102
    },
    {
      "epoch": 0.05723812170047236,
      "grad_norm": 0.42008474469184875,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.9282,
      "step": 103
    },
    {
      "epoch": 0.05779383161989442,
      "grad_norm": 0.3835548460483551,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.8687,
      "step": 104
    },
    {
      "epoch": 0.05834954153931648,
      "grad_norm": 0.36141228675842285,
      "learning_rate": 5e-05,
      "loss": 0.9697,
      "step": 105
    },
    {
      "epoch": 0.05890525145873854,
      "grad_norm": 0.33506307005882263,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.935,
      "step": 106
    },
    {
      "epoch": 0.0594609613781606,
      "grad_norm": 0.32544660568237305,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.986,
      "step": 107
    },
    {
      "epoch": 0.06001667129758266,
      "grad_norm": 0.27874335646629333,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.8796,
      "step": 108
    },
    {
      "epoch": 0.06057238121700472,
      "grad_norm": 0.26315340399742126,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.9322,
      "step": 109
    },
    {
      "epoch": 0.06112809113642678,
      "grad_norm": 0.2800102233886719,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.9671,
      "step": 110
    },
    {
      "epoch": 0.06168380105584885,
      "grad_norm": 0.27754732966423035,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.9139,
      "step": 111
    },
    {
      "epoch": 0.06223951097527091,
      "grad_norm": 0.2846638262271881,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.0149,
      "step": 112
    },
    {
      "epoch": 0.06279522089469297,
      "grad_norm": 0.2971688508987427,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.0451,
      "step": 113
    },
    {
      "epoch": 0.06335093081411503,
      "grad_norm": 0.2871069610118866,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.984,
      "step": 114
    },
    {
      "epoch": 0.06390664073353709,
      "grad_norm": 0.29142001271247864,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.9882,
      "step": 115
    },
    {
      "epoch": 0.06446235065295916,
      "grad_norm": 0.2938911020755768,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.9948,
      "step": 116
    },
    {
      "epoch": 0.06501806057238121,
      "grad_norm": 0.307596892118454,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.0425,
      "step": 117
    },
    {
      "epoch": 0.06557377049180328,
      "grad_norm": 0.31925082206726074,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.9976,
      "step": 118
    },
    {
      "epoch": 0.06612948041122534,
      "grad_norm": 0.31774935126304626,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.1184,
      "step": 119
    },
    {
      "epoch": 0.0666851903306474,
      "grad_norm": 0.32030215859413147,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.0316,
      "step": 120
    },
    {
      "epoch": 0.06724090025006946,
      "grad_norm": 0.35295090079307556,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.1201,
      "step": 121
    },
    {
      "epoch": 0.06779661016949153,
      "grad_norm": 0.3115982115268707,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.037,
      "step": 122
    },
    {
      "epoch": 0.06835232008891359,
      "grad_norm": 0.32781195640563965,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.9565,
      "step": 123
    },
    {
      "epoch": 0.06890803000833565,
      "grad_norm": 0.36501210927963257,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.095,
      "step": 124
    },
    {
      "epoch": 0.06946373992775771,
      "grad_norm": 0.34200319647789,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.0184,
      "step": 125
    },
    {
      "epoch": 0.07001944984717977,
      "grad_norm": 0.3334757387638092,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.9712,
      "step": 126
    },
    {
      "epoch": 0.07057515976660184,
      "grad_norm": 0.40706580877304077,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.0333,
      "step": 127
    },
    {
      "epoch": 0.07113086968602389,
      "grad_norm": 0.37194889783859253,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.0282,
      "step": 128
    },
    {
      "epoch": 0.07168657960544596,
      "grad_norm": 0.388510525226593,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.1057,
      "step": 129
    },
    {
      "epoch": 0.07224228952486803,
      "grad_norm": 0.39619317650794983,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.0826,
      "step": 130
    },
    {
      "epoch": 0.07279799944429008,
      "grad_norm": 0.43052905797958374,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1265,
      "step": 131
    },
    {
      "epoch": 0.07335370936371215,
      "grad_norm": 0.4574252963066101,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.1076,
      "step": 132
    },
    {
      "epoch": 0.0739094192831342,
      "grad_norm": 0.42951375246047974,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.0856,
      "step": 133
    },
    {
      "epoch": 0.07446512920255627,
      "grad_norm": 0.4537774920463562,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.1315,
      "step": 134
    },
    {
      "epoch": 0.07502083912197832,
      "grad_norm": 0.46146097779273987,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.1864,
      "step": 135
    },
    {
      "epoch": 0.07557654904140039,
      "grad_norm": 0.5010975003242493,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0196,
      "step": 136
    },
    {
      "epoch": 0.07613225896082246,
      "grad_norm": 0.5995842814445496,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.1385,
      "step": 137
    },
    {
      "epoch": 0.07668796888024451,
      "grad_norm": 0.5713678002357483,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1103,
      "step": 138
    },
    {
      "epoch": 0.07724367879966658,
      "grad_norm": 0.6867542266845703,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.3628,
      "step": 139
    },
    {
      "epoch": 0.07779938871908863,
      "grad_norm": 0.6472092270851135,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.0568,
      "step": 140
    },
    {
      "epoch": 0.0783550986385107,
      "grad_norm": 0.6878289580345154,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2264,
      "step": 141
    },
    {
      "epoch": 0.07891080855793275,
      "grad_norm": 0.6812445521354675,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.254,
      "step": 142
    },
    {
      "epoch": 0.07946651847735482,
      "grad_norm": 0.679113507270813,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1547,
      "step": 143
    },
    {
      "epoch": 0.08002222839677688,
      "grad_norm": 0.938140332698822,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2103,
      "step": 144
    },
    {
      "epoch": 0.08057793831619894,
      "grad_norm": 0.8095746636390686,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1483,
      "step": 145
    },
    {
      "epoch": 0.08113364823562101,
      "grad_norm": 0.9538931846618652,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.2137,
      "step": 146
    },
    {
      "epoch": 0.08168935815504307,
      "grad_norm": 0.993846595287323,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.0651,
      "step": 147
    },
    {
      "epoch": 0.08224506807446513,
      "grad_norm": 0.966073751449585,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.2755,
      "step": 148
    },
    {
      "epoch": 0.08280077799388719,
      "grad_norm": 1.1758711338043213,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.3301,
      "step": 149
    },
    {
      "epoch": 0.08335648791330925,
      "grad_norm": 1.6939853429794312,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.2407,
      "step": 150
    },
    {
      "epoch": 0.08335648791330925,
      "eval_loss": 1.0815174579620361,
      "eval_runtime": 72.6188,
      "eval_samples_per_second": 41.739,
      "eval_steps_per_second": 10.438,
      "step": 150
    },
    {
      "epoch": 0.08391219783273131,
      "grad_norm": 0.30047351121902466,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.8292,
      "step": 151
    },
    {
      "epoch": 0.08446790775215338,
      "grad_norm": 0.3198910057544708,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.8585,
      "step": 152
    },
    {
      "epoch": 0.08502361767157544,
      "grad_norm": 0.2920304536819458,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.8605,
      "step": 153
    },
    {
      "epoch": 0.0855793275909975,
      "grad_norm": 0.31984180212020874,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.9555,
      "step": 154
    },
    {
      "epoch": 0.08613503751041957,
      "grad_norm": 0.29660314321517944,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.9483,
      "step": 155
    },
    {
      "epoch": 0.08669074742984162,
      "grad_norm": 0.28220507502555847,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.8908,
      "step": 156
    },
    {
      "epoch": 0.08724645734926369,
      "grad_norm": 0.28584742546081543,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.9208,
      "step": 157
    },
    {
      "epoch": 0.08780216726868574,
      "grad_norm": 0.30322790145874023,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.8816,
      "step": 158
    },
    {
      "epoch": 0.08835787718810781,
      "grad_norm": 0.29575324058532715,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.9232,
      "step": 159
    },
    {
      "epoch": 0.08891358710752988,
      "grad_norm": 0.3057366609573364,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.0517,
      "step": 160
    },
    {
      "epoch": 0.08946929702695193,
      "grad_norm": 0.2996023893356323,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.9902,
      "step": 161
    },
    {
      "epoch": 0.090025006946374,
      "grad_norm": 0.2995615303516388,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.9786,
      "step": 162
    },
    {
      "epoch": 0.09058071686579605,
      "grad_norm": 0.2993999421596527,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0684,
      "step": 163
    },
    {
      "epoch": 0.09113642678521812,
      "grad_norm": 0.29088151454925537,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.9621,
      "step": 164
    },
    {
      "epoch": 0.09169213670464017,
      "grad_norm": 0.3247353136539459,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.9998,
      "step": 165
    },
    {
      "epoch": 0.09224784662406224,
      "grad_norm": 0.2907552123069763,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.9709,
      "step": 166
    },
    {
      "epoch": 0.0928035565434843,
      "grad_norm": 0.31431370973587036,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0878,
      "step": 167
    },
    {
      "epoch": 0.09335926646290636,
      "grad_norm": 0.30374595522880554,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0518,
      "step": 168
    },
    {
      "epoch": 0.09391497638232843,
      "grad_norm": 0.3017720580101013,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.067,
      "step": 169
    },
    {
      "epoch": 0.09447068630175048,
      "grad_norm": 0.31449365615844727,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.9681,
      "step": 170
    },
    {
      "epoch": 0.09502639622117255,
      "grad_norm": 0.340142160654068,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.0287,
      "step": 171
    },
    {
      "epoch": 0.0955821061405946,
      "grad_norm": 0.33678263425827026,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.0596,
      "step": 172
    },
    {
      "epoch": 0.09613781606001667,
      "grad_norm": 0.3631838858127594,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0915,
      "step": 173
    },
    {
      "epoch": 0.09669352597943873,
      "grad_norm": 0.3573366403579712,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.1511,
      "step": 174
    },
    {
      "epoch": 0.0972492358988608,
      "grad_norm": 0.35506105422973633,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0163,
      "step": 175
    },
    {
      "epoch": 0.09780494581828286,
      "grad_norm": 0.3673512935638428,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9928,
      "step": 176
    },
    {
      "epoch": 0.09836065573770492,
      "grad_norm": 0.37375813722610474,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.0247,
      "step": 177
    },
    {
      "epoch": 0.09891636565712698,
      "grad_norm": 0.3927251696586609,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.0163,
      "step": 178
    },
    {
      "epoch": 0.09947207557654904,
      "grad_norm": 0.3984759449958801,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.1519,
      "step": 179
    },
    {
      "epoch": 0.1000277854959711,
      "grad_norm": 0.4091857969760895,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.9839,
      "step": 180
    },
    {
      "epoch": 0.10058349541539316,
      "grad_norm": 0.4268726706504822,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0537,
      "step": 181
    },
    {
      "epoch": 0.10113920533481523,
      "grad_norm": 0.4150637686252594,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.1172,
      "step": 182
    },
    {
      "epoch": 0.1016949152542373,
      "grad_norm": 0.49816185235977173,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.1863,
      "step": 183
    },
    {
      "epoch": 0.10225062517365935,
      "grad_norm": 0.4841300845146179,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.1023,
      "step": 184
    },
    {
      "epoch": 0.10280633509308142,
      "grad_norm": 0.5328807830810547,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.1539,
      "step": 185
    },
    {
      "epoch": 0.10336204501250347,
      "grad_norm": 0.6443712115287781,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.0665,
      "step": 186
    },
    {
      "epoch": 0.10391775493192554,
      "grad_norm": 0.5393214225769043,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.1732,
      "step": 187
    },
    {
      "epoch": 0.10447346485134759,
      "grad_norm": 0.6496522426605225,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.2112,
      "step": 188
    },
    {
      "epoch": 0.10502917477076966,
      "grad_norm": 0.626882016658783,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.0787,
      "step": 189
    },
    {
      "epoch": 0.10558488469019171,
      "grad_norm": 0.7077686190605164,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.241,
      "step": 190
    },
    {
      "epoch": 0.10614059460961378,
      "grad_norm": 0.696556568145752,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2266,
      "step": 191
    },
    {
      "epoch": 0.10669630452903585,
      "grad_norm": 0.8313038349151611,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.1929,
      "step": 192
    },
    {
      "epoch": 0.1072520144484579,
      "grad_norm": 0.7896590828895569,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.3328,
      "step": 193
    },
    {
      "epoch": 0.10780772436787997,
      "grad_norm": 0.8421167135238647,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.2122,
      "step": 194
    },
    {
      "epoch": 0.10836343428730202,
      "grad_norm": 0.8936784267425537,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.1984,
      "step": 195
    },
    {
      "epoch": 0.10891914420672409,
      "grad_norm": 0.9015226364135742,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.1352,
      "step": 196
    },
    {
      "epoch": 0.10947485412614615,
      "grad_norm": 0.9489641189575195,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2486,
      "step": 197
    },
    {
      "epoch": 0.11003056404556821,
      "grad_norm": 1.0115143060684204,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.3163,
      "step": 198
    },
    {
      "epoch": 0.11058627396499028,
      "grad_norm": 1.2494778633117676,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.2106,
      "step": 199
    },
    {
      "epoch": 0.11114198388441233,
      "grad_norm": 1.4850845336914062,
      "learning_rate": 0.0,
      "loss": 1.365,
      "step": 200
    },
    {
      "epoch": 0.11114198388441233,
      "eval_loss": 1.07815420627594,
      "eval_runtime": 72.5916,
      "eval_samples_per_second": 41.754,
      "eval_steps_per_second": 10.442,
      "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": 5.039561801465856e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}