{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.996706915477497,
  "eval_steps": 500,
  "global_step": 1365,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0021953896816684962,
      "grad_norm": 37.74675866275334,
      "learning_rate": 3.64963503649635e-09,
      "loss": 1.4854,
      "step": 1
    },
    {
      "epoch": 0.010976948408342482,
      "grad_norm": 34.58078409329761,
      "learning_rate": 1.824817518248175e-08,
      "loss": 1.4947,
      "step": 5
    },
    {
      "epoch": 0.021953896816684963,
      "grad_norm": 33.64622164796075,
      "learning_rate": 3.64963503649635e-08,
      "loss": 1.4655,
      "step": 10
    },
    {
      "epoch": 0.03293084522502744,
      "grad_norm": 33.96621368420792,
      "learning_rate": 5.474452554744526e-08,
      "loss": 1.4454,
      "step": 15
    },
    {
      "epoch": 0.043907793633369926,
      "grad_norm": 32.64608823456708,
      "learning_rate": 7.2992700729927e-08,
      "loss": 1.441,
      "step": 20
    },
    {
      "epoch": 0.054884742041712405,
      "grad_norm": 28.7407961705511,
      "learning_rate": 9.124087591240875e-08,
      "loss": 1.4399,
      "step": 25
    },
    {
      "epoch": 0.06586169045005488,
      "grad_norm": 26.49133383604883,
      "learning_rate": 1.0948905109489052e-07,
      "loss": 1.3907,
      "step": 30
    },
    {
      "epoch": 0.07683863885839737,
      "grad_norm": 28.558137483756873,
      "learning_rate": 1.2773722627737227e-07,
      "loss": 1.3469,
      "step": 35
    },
    {
      "epoch": 0.08781558726673985,
      "grad_norm": 16.332214000825115,
      "learning_rate": 1.45985401459854e-07,
      "loss": 1.2666,
      "step": 40
    },
    {
      "epoch": 0.09879253567508232,
      "grad_norm": 10.940594279582514,
      "learning_rate": 1.6423357664233575e-07,
      "loss": 1.2482,
      "step": 45
    },
    {
      "epoch": 0.10976948408342481,
      "grad_norm": 5.741537178926412,
      "learning_rate": 1.824817518248175e-07,
      "loss": 1.2249,
      "step": 50
    },
    {
      "epoch": 0.1207464324917673,
      "grad_norm": 5.018901761568715,
      "learning_rate": 2.0072992700729928e-07,
      "loss": 1.1839,
      "step": 55
    },
    {
      "epoch": 0.13172338090010977,
      "grad_norm": 4.321418760394092,
      "learning_rate": 2.1897810218978103e-07,
      "loss": 1.1806,
      "step": 60
    },
    {
      "epoch": 0.14270032930845225,
      "grad_norm": 3.6868934108802405,
      "learning_rate": 2.3722627737226276e-07,
      "loss": 1.1367,
      "step": 65
    },
    {
      "epoch": 0.15367727771679474,
      "grad_norm": 3.6056758938325633,
      "learning_rate": 2.5547445255474454e-07,
      "loss": 1.1194,
      "step": 70
    },
    {
      "epoch": 0.16465422612513722,
      "grad_norm": 2.9888643007945923,
      "learning_rate": 2.737226277372263e-07,
      "loss": 1.1277,
      "step": 75
    },
    {
      "epoch": 0.1756311745334797,
      "grad_norm": 3.1811524089107737,
      "learning_rate": 2.91970802919708e-07,
      "loss": 1.1041,
      "step": 80
    },
    {
      "epoch": 0.18660812294182216,
      "grad_norm": 2.7324489147282356,
      "learning_rate": 3.102189781021898e-07,
      "loss": 1.1085,
      "step": 85
    },
    {
      "epoch": 0.19758507135016465,
      "grad_norm": 2.612737685100592,
      "learning_rate": 3.284671532846715e-07,
      "loss": 1.0872,
      "step": 90
    },
    {
      "epoch": 0.20856201975850713,
      "grad_norm": 2.7784872016274806,
      "learning_rate": 3.467153284671533e-07,
      "loss": 1.1009,
      "step": 95
    },
    {
      "epoch": 0.21953896816684962,
      "grad_norm": 2.6806669714481695,
      "learning_rate": 3.64963503649635e-07,
      "loss": 1.0742,
      "step": 100
    },
    {
      "epoch": 0.2305159165751921,
      "grad_norm": 2.8589613220847774,
      "learning_rate": 3.8321167883211675e-07,
      "loss": 1.0928,
      "step": 105
    },
    {
      "epoch": 0.2414928649835346,
      "grad_norm": 2.6851891388693887,
      "learning_rate": 4.0145985401459856e-07,
      "loss": 1.079,
      "step": 110
    },
    {
      "epoch": 0.2524698133918771,
      "grad_norm": 2.3561189558185136,
      "learning_rate": 4.1970802919708026e-07,
      "loss": 1.0441,
      "step": 115
    },
    {
      "epoch": 0.26344676180021953,
      "grad_norm": 2.3680695140089796,
      "learning_rate": 4.3795620437956206e-07,
      "loss": 1.077,
      "step": 120
    },
    {
      "epoch": 0.27442371020856204,
      "grad_norm": 2.9917242722942095,
      "learning_rate": 4.5620437956204376e-07,
      "loss": 1.0486,
      "step": 125
    },
    {
      "epoch": 0.2854006586169045,
      "grad_norm": 2.331882690510773,
      "learning_rate": 4.744525547445255e-07,
      "loss": 1.0478,
      "step": 130
    },
    {
      "epoch": 0.29637760702524696,
      "grad_norm": 2.54506147430991,
      "learning_rate": 4.927007299270073e-07,
      "loss": 1.0501,
      "step": 135
    },
    {
      "epoch": 0.30735455543358947,
      "grad_norm": 3.343713651464699,
      "learning_rate": 4.999926370237027e-07,
      "loss": 1.0464,
      "step": 140
    },
    {
      "epoch": 0.31833150384193193,
      "grad_norm": 2.345761304992979,
      "learning_rate": 4.999476426280587e-07,
      "loss": 1.039,
      "step": 145
    },
    {
      "epoch": 0.32930845225027444,
      "grad_norm": 2.5391392458703286,
      "learning_rate": 4.998617517322294e-07,
      "loss": 1.0316,
      "step": 150
    },
    {
      "epoch": 0.3402854006586169,
      "grad_norm": 2.472489561391851,
      "learning_rate": 4.997349783897061e-07,
      "loss": 1.0417,
      "step": 155
    },
    {
      "epoch": 0.3512623490669594,
      "grad_norm": 2.3737303149457016,
      "learning_rate": 4.99567343343177e-07,
      "loss": 1.0195,
      "step": 160
    },
    {
      "epoch": 0.36223929747530187,
      "grad_norm": 2.3800187550942047,
      "learning_rate": 4.993588740211331e-07,
      "loss": 1.0474,
      "step": 165
    },
    {
      "epoch": 0.3732162458836443,
      "grad_norm": 2.330985495803907,
      "learning_rate": 4.991096045333809e-07,
      "loss": 1.0262,
      "step": 170
    },
    {
      "epoch": 0.38419319429198684,
      "grad_norm": 2.4303555873767935,
      "learning_rate": 4.988195756654605e-07,
      "loss": 1.0142,
      "step": 175
    },
    {
      "epoch": 0.3951701427003293,
      "grad_norm": 2.4094455102172585,
      "learning_rate": 4.984888348719731e-07,
      "loss": 1.0188,
      "step": 180
    },
    {
      "epoch": 0.4061470911086718,
      "grad_norm": 2.1476465324301057,
      "learning_rate": 4.981174362688158e-07,
      "loss": 1.0142,
      "step": 185
    },
    {
      "epoch": 0.41712403951701427,
      "grad_norm": 2.2526911826436615,
      "learning_rate": 4.977054406243274e-07,
      "loss": 1.0328,
      "step": 190
    },
    {
      "epoch": 0.4281009879253567,
      "grad_norm": 2.3224686636469443,
      "learning_rate": 4.972529153493455e-07,
      "loss": 0.9933,
      "step": 195
    },
    {
      "epoch": 0.43907793633369924,
      "grad_norm": 2.4302653074531437,
      "learning_rate": 4.967599344861768e-07,
      "loss": 0.9954,
      "step": 200
    },
    {
      "epoch": 0.4500548847420417,
      "grad_norm": 2.4835737176654904,
      "learning_rate": 4.96226578696482e-07,
      "loss": 0.9985,
      "step": 205
    },
    {
      "epoch": 0.4610318331503842,
      "grad_norm": 2.4230184538015354,
      "learning_rate": 4.956529352480782e-07,
      "loss": 0.9987,
      "step": 210
    },
    {
      "epoch": 0.47200878155872666,
      "grad_norm": 2.329153987703765,
      "learning_rate": 4.950390980006599e-07,
      "loss": 1.0098,
      "step": 215
    },
    {
      "epoch": 0.4829857299670692,
      "grad_norm": 2.1643982961769574,
      "learning_rate": 4.943851673904419e-07,
      "loss": 1.0133,
      "step": 220
    },
    {
      "epoch": 0.49396267837541163,
      "grad_norm": 2.3197403816872977,
      "learning_rate": 4.936912504137257e-07,
      "loss": 1.0142,
      "step": 225
    },
    {
      "epoch": 0.5049396267837541,
      "grad_norm": 2.34747990249977,
      "learning_rate": 4.929574606093926e-07,
      "loss": 0.9976,
      "step": 230
    },
    {
      "epoch": 0.5159165751920965,
      "grad_norm": 2.2485105183919107,
      "learning_rate": 4.92183918040327e-07,
      "loss": 0.9891,
      "step": 235
    },
    {
      "epoch": 0.5268935236004391,
      "grad_norm": 2.2790091103112977,
      "learning_rate": 4.913707492737708e-07,
      "loss": 1.002,
      "step": 240
    },
    {
      "epoch": 0.5378704720087816,
      "grad_norm": 2.239563511547285,
      "learning_rate": 4.905180873606156e-07,
      "loss": 0.9884,
      "step": 245
    },
    {
      "epoch": 0.5488474204171241,
      "grad_norm": 2.201197024971912,
      "learning_rate": 4.896260718136315e-07,
      "loss": 1.0053,
      "step": 250
    },
    {
      "epoch": 0.5598243688254665,
      "grad_norm": 2.4477006618500976,
      "learning_rate": 4.886948485846413e-07,
      "loss": 0.9905,
      "step": 255
    },
    {
      "epoch": 0.570801317233809,
      "grad_norm": 2.1778270375289357,
      "learning_rate": 4.877245700406389e-07,
      "loss": 1.006,
      "step": 260
    },
    {
      "epoch": 0.5817782656421515,
      "grad_norm": 2.470318023588166,
      "learning_rate": 4.867153949388592e-07,
      "loss": 1.0001,
      "step": 265
    },
    {
      "epoch": 0.5927552140504939,
      "grad_norm": 2.246482390471278,
      "learning_rate": 4.856674884008027e-07,
      "loss": 1.0016,
      "step": 270
    },
    {
      "epoch": 0.6037321624588364,
      "grad_norm": 2.555761228606553,
      "learning_rate": 4.845810218852175e-07,
      "loss": 0.9909,
      "step": 275
    },
    {
      "epoch": 0.6147091108671789,
      "grad_norm": 2.6187160779641405,
      "learning_rate": 4.834561731600457e-07,
      "loss": 0.9879,
      "step": 280
    },
    {
      "epoch": 0.6256860592755215,
      "grad_norm": 2.2540909179670288,
      "learning_rate": 4.822931262733367e-07,
      "loss": 0.9723,
      "step": 285
    },
    {
      "epoch": 0.6366630076838639,
      "grad_norm": 2.3392602017439468,
      "learning_rate": 4.810920715231334e-07,
      "loss": 0.9813,
      "step": 290
    },
    {
      "epoch": 0.6476399560922064,
      "grad_norm": 2.5885537238335607,
      "learning_rate": 4.798532054263356e-07,
      "loss": 0.9672,
      "step": 295
    },
    {
      "epoch": 0.6586169045005489,
      "grad_norm": 2.3642433721473712,
      "learning_rate": 4.785767306865457e-07,
      "loss": 0.9774,
      "step": 300
    },
    {
      "epoch": 0.6695938529088913,
      "grad_norm": 2.408185221870625,
      "learning_rate": 4.772628561609021e-07,
      "loss": 0.985,
      "step": 305
    },
    {
      "epoch": 0.6805708013172338,
      "grad_norm": 2.2324134890318246,
      "learning_rate": 4.7591179682590644e-07,
      "loss": 0.9725,
      "step": 310
    },
    {
      "epoch": 0.6915477497255763,
      "grad_norm": 2.322340624031069,
      "learning_rate": 4.7452377374224874e-07,
      "loss": 0.9806,
      "step": 315
    },
    {
      "epoch": 0.7025246981339188,
      "grad_norm": 2.211381205162829,
      "learning_rate": 4.730990140186373e-07,
      "loss": 0.999,
      "step": 320
    },
    {
      "epoch": 0.7135016465422612,
      "grad_norm": 2.2193259266838155,
      "learning_rate": 4.7163775077463963e-07,
      "loss": 0.9728,
      "step": 325
    },
    {
      "epoch": 0.7244785949506037,
      "grad_norm": 2.307395076998798,
      "learning_rate": 4.7014022310253875e-07,
      "loss": 0.9924,
      "step": 330
    },
    {
      "epoch": 0.7354555433589463,
      "grad_norm": 2.242602418548775,
      "learning_rate": 4.6860667602821316e-07,
      "loss": 0.9861,
      "step": 335
    },
    {
      "epoch": 0.7464324917672887,
      "grad_norm": 2.2686164540827547,
      "learning_rate": 4.670373604710456e-07,
      "loss": 0.9803,
      "step": 340
    },
    {
      "epoch": 0.7574094401756312,
      "grad_norm": 2.1375492767276048,
      "learning_rate": 4.6543253320286755e-07,
      "loss": 0.9617,
      "step": 345
    },
    {
      "epoch": 0.7683863885839737,
      "grad_norm": 2.2558323943911684,
      "learning_rate": 4.6379245680594625e-07,
      "loss": 0.9531,
      "step": 350
    },
    {
      "epoch": 0.7793633369923162,
      "grad_norm": 2.39500346963894,
      "learning_rate": 4.621173996300207e-07,
      "loss": 0.9897,
      "step": 355
    },
    {
      "epoch": 0.7903402854006586,
      "grad_norm": 2.636774268293073,
      "learning_rate": 4.6040763574839435e-07,
      "loss": 0.9899,
      "step": 360
    },
    {
      "epoch": 0.8013172338090011,
      "grad_norm": 2.3125464385328236,
      "learning_rate": 4.5866344491309104e-07,
      "loss": 0.9844,
      "step": 365
    },
    {
      "epoch": 0.8122941822173436,
      "grad_norm": 2.158764874076257,
      "learning_rate": 4.568851125090822e-07,
      "loss": 0.9789,
      "step": 370
    },
    {
      "epoch": 0.823271130625686,
      "grad_norm": 2.146303203312366,
      "learning_rate": 4.550729295075918e-07,
      "loss": 0.9894,
      "step": 375
    },
    {
      "epoch": 0.8342480790340285,
      "grad_norm": 2.1234461105972686,
      "learning_rate": 4.532271924184874e-07,
      "loss": 0.9573,
      "step": 380
    },
    {
      "epoch": 0.845225027442371,
      "grad_norm": 2.5991666840318524,
      "learning_rate": 4.513482032417655e-07,
      "loss": 0.978,
      "step": 385
    },
    {
      "epoch": 0.8562019758507134,
      "grad_norm": 2.459654668779688,
      "learning_rate": 4.494362694181384e-07,
      "loss": 0.9774,
      "step": 390
    },
    {
      "epoch": 0.867178924259056,
      "grad_norm": 2.247292046481084,
      "learning_rate": 4.474917037787297e-07,
      "loss": 0.9753,
      "step": 395
    },
    {
      "epoch": 0.8781558726673985,
      "grad_norm": 2.2211212349255574,
      "learning_rate": 4.455148244938901e-07,
      "loss": 0.9789,
      "step": 400
    },
    {
      "epoch": 0.889132821075741,
      "grad_norm": 2.3018670472595644,
      "learning_rate": 4.435059550211371e-07,
      "loss": 0.9748,
      "step": 405
    },
    {
      "epoch": 0.9001097694840834,
      "grad_norm": 2.1721337174418784,
      "learning_rate": 4.414654240522315e-07,
      "loss": 0.9741,
      "step": 410
    },
    {
      "epoch": 0.9110867178924259,
      "grad_norm": 2.28819746756527,
      "learning_rate": 4.3939356545939677e-07,
      "loss": 0.9756,
      "step": 415
    },
    {
      "epoch": 0.9220636663007684,
      "grad_norm": 2.4856430019366806,
      "learning_rate": 4.372907182406902e-07,
      "loss": 0.9599,
      "step": 420
    },
    {
      "epoch": 0.9330406147091108,
      "grad_norm": 2.279118517164672,
      "learning_rate": 4.3515722646453657e-07,
      "loss": 0.9766,
      "step": 425
    },
    {
      "epoch": 0.9440175631174533,
      "grad_norm": 2.283587407407711,
      "learning_rate": 4.329934392134314e-07,
      "loss": 0.9708,
      "step": 430
    },
    {
      "epoch": 0.9549945115257958,
      "grad_norm": 2.3436121330856676,
      "learning_rate": 4.3079971052682387e-07,
      "loss": 0.9752,
      "step": 435
    },
    {
      "epoch": 0.9659714599341384,
      "grad_norm": 2.33617606208673,
      "learning_rate": 4.2857639934318874e-07,
      "loss": 0.9719,
      "step": 440
    },
    {
      "epoch": 0.9769484083424808,
      "grad_norm": 2.2203473882036966,
      "learning_rate": 4.26323869441297e-07,
      "loss": 0.9613,
      "step": 445
    },
    {
      "epoch": 0.9879253567508233,
      "grad_norm": 2.5027842654915005,
      "learning_rate": 4.240424893806941e-07,
      "loss": 0.9864,
      "step": 450
    },
    {
      "epoch": 0.9989023051591658,
      "grad_norm": 2.333812012249256,
      "learning_rate": 4.217326324413962e-07,
      "loss": 0.9643,
      "step": 455
    },
    {
      "epoch": 0.9989023051591658,
      "eval_loss": 1.019766926765442,
      "eval_runtime": 4.8917,
      "eval_samples_per_second": 54.786,
      "eval_steps_per_second": 1.022,
      "step": 455
    },
    {
      "epoch": 1.0098792535675083,
      "grad_norm": 2.6384758953681495,
      "learning_rate": 4.1939467656281387e-07,
      "loss": 0.9411,
      "step": 460
    },
    {
      "epoch": 1.0208562019758507,
      "grad_norm": 2.4368918808468263,
      "learning_rate": 4.170290042819137e-07,
      "loss": 0.9304,
      "step": 465
    },
    {
      "epoch": 1.031833150384193,
      "grad_norm": 2.606275797390668,
      "learning_rate": 4.146360026706276e-07,
      "loss": 0.9401,
      "step": 470
    },
    {
      "epoch": 1.0428100987925357,
      "grad_norm": 2.3703029555878175,
      "learning_rate": 4.122160632725195e-07,
      "loss": 0.9411,
      "step": 475
    },
    {
      "epoch": 1.0537870472008781,
      "grad_norm": 2.3433873192058785,
      "learning_rate": 4.097695820387216e-07,
      "loss": 0.9314,
      "step": 480
    },
    {
      "epoch": 1.0647639956092205,
      "grad_norm": 2.3059951796342757,
      "learning_rate": 4.0729695926314813e-07,
      "loss": 0.9212,
      "step": 485
    },
    {
      "epoch": 1.0757409440175631,
      "grad_norm": 2.246905867491328,
      "learning_rate": 4.0479859951699934e-07,
      "loss": 0.9299,
      "step": 490
    },
    {
      "epoch": 1.0867178924259056,
      "grad_norm": 2.332204582181345,
      "learning_rate": 4.0227491158256564e-07,
      "loss": 0.9283,
      "step": 495
    },
    {
      "epoch": 1.0976948408342482,
      "grad_norm": 2.3052192318083096,
      "learning_rate": 3.9972630838634236e-07,
      "loss": 0.9081,
      "step": 500
    },
    {
      "epoch": 1.1086717892425906,
      "grad_norm": 2.2938816727121174,
      "learning_rate": 3.9715320693146653e-07,
      "loss": 0.9244,
      "step": 505
    },
    {
      "epoch": 1.119648737650933,
      "grad_norm": 2.3376274658971794,
      "learning_rate": 3.9455602822948695e-07,
      "loss": 0.9356,
      "step": 510
    },
    {
      "epoch": 1.1306256860592756,
      "grad_norm": 2.2234966705346437,
      "learning_rate": 3.9193519723147795e-07,
      "loss": 0.937,
      "step": 515
    },
    {
      "epoch": 1.141602634467618,
      "grad_norm": 2.3167888423276897,
      "learning_rate": 3.892911427585089e-07,
      "loss": 0.9253,
      "step": 520
    },
    {
      "epoch": 1.1525795828759604,
      "grad_norm": 2.2822984548495,
      "learning_rate": 3.8662429743148046e-07,
      "loss": 0.93,
      "step": 525
    },
    {
      "epoch": 1.163556531284303,
      "grad_norm": 2.3315853272899885,
      "learning_rate": 3.839350976003386e-07,
      "loss": 0.9307,
      "step": 530
    },
    {
      "epoch": 1.1745334796926454,
      "grad_norm": 2.225438330230274,
      "learning_rate": 3.8122398327267956e-07,
      "loss": 0.9191,
      "step": 535
    },
    {
      "epoch": 1.1855104281009878,
      "grad_norm": 2.282678706637911,
      "learning_rate": 3.7849139804175527e-07,
      "loss": 0.9242,
      "step": 540
    },
    {
      "epoch": 1.1964873765093305,
      "grad_norm": 2.4151431361330262,
      "learning_rate": 3.7573778901389264e-07,
      "loss": 0.9129,
      "step": 545
    },
    {
      "epoch": 1.2074643249176729,
      "grad_norm": 2.1432543425535986,
      "learning_rate": 3.729636067353377e-07,
      "loss": 0.9352,
      "step": 550
    },
    {
      "epoch": 1.2184412733260155,
      "grad_norm": 2.2103634060126045,
      "learning_rate": 3.701693051185375e-07,
      "loss": 0.9217,
      "step": 555
    },
    {
      "epoch": 1.2294182217343579,
      "grad_norm": 2.29979398895129,
      "learning_rate": 3.6735534136787034e-07,
      "loss": 0.9241,
      "step": 560
    },
    {
      "epoch": 1.2403951701427003,
      "grad_norm": 2.211918385111886,
      "learning_rate": 3.6452217590483843e-07,
      "loss": 0.9316,
      "step": 565
    },
    {
      "epoch": 1.2513721185510427,
      "grad_norm": 2.188755024370649,
      "learning_rate": 3.6167027229273296e-07,
      "loss": 0.9089,
      "step": 570
    },
    {
      "epoch": 1.2623490669593853,
      "grad_norm": 2.4206253706110576,
      "learning_rate": 3.5880009716078643e-07,
      "loss": 0.9285,
      "step": 575
    },
    {
      "epoch": 1.2733260153677277,
      "grad_norm": 2.2945442413712684,
      "learning_rate": 3.559121201278219e-07,
      "loss": 0.918,
      "step": 580
    },
    {
      "epoch": 1.2843029637760703,
      "grad_norm": 2.345758700673061,
      "learning_rate": 3.5300681372541473e-07,
      "loss": 0.9305,
      "step": 585
    },
    {
      "epoch": 1.2952799121844127,
      "grad_norm": 2.2789322285102314,
      "learning_rate": 3.5008465332057594e-07,
      "loss": 0.9281,
      "step": 590
    },
    {
      "epoch": 1.3062568605927551,
      "grad_norm": 2.243522596217181,
      "learning_rate": 3.471461170379732e-07,
      "loss": 0.914,
      "step": 595
    },
    {
      "epoch": 1.3172338090010978,
      "grad_norm": 2.323467820978964,
      "learning_rate": 3.4419168568169935e-07,
      "loss": 0.9361,
      "step": 600
    },
    {
      "epoch": 1.3282107574094402,
      "grad_norm": 2.194684758387474,
      "learning_rate": 3.4122184265660395e-07,
      "loss": 0.9311,
      "step": 605
    },
    {
      "epoch": 1.3391877058177828,
      "grad_norm": 2.2140499236992603,
      "learning_rate": 3.3823707388919736e-07,
      "loss": 0.9125,
      "step": 610
    },
    {
      "epoch": 1.3501646542261252,
      "grad_norm": 2.258349553066923,
      "learning_rate": 3.352378677481444e-07,
      "loss": 0.9426,
      "step": 615
    },
    {
      "epoch": 1.3611416026344676,
      "grad_norm": 2.358073970259772,
      "learning_rate": 3.3222471496435643e-07,
      "loss": 0.9188,
      "step": 620
    },
    {
      "epoch": 1.37211855104281,
      "grad_norm": 2.293914944788209,
      "learning_rate": 3.291981085506986e-07,
      "loss": 0.9046,
      "step": 625
    },
    {
      "epoch": 1.3830954994511526,
      "grad_norm": 2.239065960199277,
      "learning_rate": 3.2615854372132226e-07,
      "loss": 0.924,
      "step": 630
    },
    {
      "epoch": 1.394072447859495,
      "grad_norm": 2.236539103492653,
      "learning_rate": 3.231065178106384e-07,
      "loss": 0.9229,
      "step": 635
    },
    {
      "epoch": 1.4050493962678376,
      "grad_norm": 2.1945349736134707,
      "learning_rate": 3.2004253019194324e-07,
      "loss": 0.9272,
      "step": 640
    },
    {
      "epoch": 1.41602634467618,
      "grad_norm": 2.2398945956325513,
      "learning_rate": 3.169670821957112e-07,
      "loss": 0.9102,
      "step": 645
    },
    {
      "epoch": 1.4270032930845225,
      "grad_norm": 2.289801618738276,
      "learning_rate": 3.1388067702756655e-07,
      "loss": 0.9429,
      "step": 650
    },
    {
      "epoch": 1.4379802414928649,
      "grad_norm": 2.245541374176715,
      "learning_rate": 3.1078381968594895e-07,
      "loss": 0.9115,
      "step": 655
    },
    {
      "epoch": 1.4489571899012075,
      "grad_norm": 2.278894612214528,
      "learning_rate": 3.076770168794854e-07,
      "loss": 0.9398,
      "step": 660
    },
    {
      "epoch": 1.4599341383095499,
      "grad_norm": 2.3065149845422392,
      "learning_rate": 3.045607769440829e-07,
      "loss": 0.8946,
      "step": 665
    },
    {
      "epoch": 1.4709110867178925,
      "grad_norm": 2.3594569504860696,
      "learning_rate": 3.014356097597535e-07,
      "loss": 0.9137,
      "step": 670
    },
    {
      "epoch": 1.481888035126235,
      "grad_norm": 2.4349120794769354,
      "learning_rate": 2.983020266671886e-07,
      "loss": 0.9113,
      "step": 675
    },
    {
      "epoch": 1.4928649835345773,
      "grad_norm": 2.2209808300678318,
      "learning_rate": 2.951605403840921e-07,
      "loss": 0.8983,
      "step": 680
    },
    {
      "epoch": 1.5038419319429197,
      "grad_norm": 2.3286466229872635,
      "learning_rate": 2.920116649212909e-07,
      "loss": 0.9312,
      "step": 685
    },
    {
      "epoch": 1.5148188803512623,
      "grad_norm": 2.238157735155838,
      "learning_rate": 2.888559154986307e-07,
      "loss": 0.921,
      "step": 690
    },
    {
      "epoch": 1.525795828759605,
      "grad_norm": 2.2522331987696944,
      "learning_rate": 2.856938084606769e-07,
      "loss": 0.9027,
      "step": 695
    },
    {
      "epoch": 1.5367727771679474,
      "grad_norm": 2.225248865709702,
      "learning_rate": 2.825258611922292e-07,
      "loss": 0.9186,
      "step": 700
    },
    {
      "epoch": 1.5477497255762898,
      "grad_norm": 2.2372307174396036,
      "learning_rate": 2.793525920336677e-07,
      "loss": 0.9179,
      "step": 705
    },
    {
      "epoch": 1.5587266739846322,
      "grad_norm": 2.227278818178374,
      "learning_rate": 2.7617452019614144e-07,
      "loss": 0.9068,
      "step": 710
    },
    {
      "epoch": 1.5697036223929748,
      "grad_norm": 2.1887205060374133,
      "learning_rate": 2.7299216567661503e-07,
      "loss": 0.9128,
      "step": 715
    },
    {
      "epoch": 1.5806805708013172,
      "grad_norm": 2.1730778041819487,
      "learning_rate": 2.6980604917278675e-07,
      "loss": 0.9072,
      "step": 720
    },
    {
      "epoch": 1.5916575192096598,
      "grad_norm": 2.306897540604203,
      "learning_rate": 2.6661669199789174e-07,
      "loss": 0.9241,
      "step": 725
    },
    {
      "epoch": 1.6026344676180022,
      "grad_norm": 2.2513680878615956,
      "learning_rate": 2.6342461599540456e-07,
      "loss": 0.9183,
      "step": 730
    },
    {
      "epoch": 1.6136114160263446,
      "grad_norm": 2.2683353014176504,
      "learning_rate": 2.6023034345365515e-07,
      "loss": 0.9101,
      "step": 735
    },
    {
      "epoch": 1.624588364434687,
      "grad_norm": 2.3773000711715793,
      "learning_rate": 2.5703439702037155e-07,
      "loss": 0.9226,
      "step": 740
    },
    {
      "epoch": 1.6355653128430296,
      "grad_norm": 2.2707149058977047,
      "learning_rate": 2.538372996171648e-07,
      "loss": 0.9099,
      "step": 745
    },
    {
      "epoch": 1.6465422612513723,
      "grad_norm": 2.3321984673401808,
      "learning_rate": 2.506395743539677e-07,
      "loss": 0.9053,
      "step": 750
    },
    {
      "epoch": 1.6575192096597147,
      "grad_norm": 2.289265422242154,
      "learning_rate": 2.474417444434436e-07,
      "loss": 0.9167,
      "step": 755
    },
    {
      "epoch": 1.668496158068057,
      "grad_norm": 2.29419980236518,
      "learning_rate": 2.442443331153783e-07,
      "loss": 0.9255,
      "step": 760
    },
    {
      "epoch": 1.6794731064763995,
      "grad_norm": 2.2919378273617785,
      "learning_rate": 2.4104786353106926e-07,
      "loss": 0.9039,
      "step": 765
    },
    {
      "epoch": 1.690450054884742,
      "grad_norm": 2.2503081710593347,
      "learning_rate": 2.3785285869772525e-07,
      "loss": 0.908,
      "step": 770
    },
    {
      "epoch": 1.7014270032930845,
      "grad_norm": 2.273983108472163,
      "learning_rate": 2.3465984138289237e-07,
      "loss": 0.9178,
      "step": 775
    },
    {
      "epoch": 1.7124039517014271,
      "grad_norm": 2.2199735303427333,
      "learning_rate": 2.3146933402891812e-07,
      "loss": 0.9174,
      "step": 780
    },
    {
      "epoch": 1.7233809001097695,
      "grad_norm": 2.246832371307188,
      "learning_rate": 2.2828185866746987e-07,
      "loss": 0.9063,
      "step": 785
    },
    {
      "epoch": 1.734357848518112,
      "grad_norm": 2.262285299215957,
      "learning_rate": 2.2509793683411932e-07,
      "loss": 0.9282,
      "step": 790
    },
    {
      "epoch": 1.7453347969264543,
      "grad_norm": 2.179628826722484,
      "learning_rate": 2.2191808948300904e-07,
      "loss": 0.8962,
      "step": 795
    },
    {
      "epoch": 1.756311745334797,
      "grad_norm": 2.363447036072627,
      "learning_rate": 2.1874283690161364e-07,
      "loss": 0.9288,
      "step": 800
    },
    {
      "epoch": 1.7672886937431396,
      "grad_norm": 2.2413506946996025,
      "learning_rate": 2.1557269862561042e-07,
      "loss": 0.9105,
      "step": 805
    },
    {
      "epoch": 1.778265642151482,
      "grad_norm": 2.213099523042175,
      "learning_rate": 2.124081933538726e-07,
      "loss": 0.9152,
      "step": 810
    },
    {
      "epoch": 1.7892425905598244,
      "grad_norm": 2.3911262169971232,
      "learning_rate": 2.0924983886359966e-07,
      "loss": 0.9114,
      "step": 815
    },
    {
      "epoch": 1.8002195389681668,
      "grad_norm": 2.3443041782008,
      "learning_rate": 2.0609815192559863e-07,
      "loss": 0.906,
      "step": 820
    },
    {
      "epoch": 1.8111964873765092,
      "grad_norm": 2.2448633071392736,
      "learning_rate": 2.0295364821972993e-07,
      "loss": 0.9229,
      "step": 825
    },
    {
      "epoch": 1.8221734357848518,
      "grad_norm": 2.335369643205432,
      "learning_rate": 1.998168422505315e-07,
      "loss": 0.9055,
      "step": 830
    },
    {
      "epoch": 1.8331503841931944,
      "grad_norm": 2.260199044110363,
      "learning_rate": 1.966882472630356e-07,
      "loss": 0.916,
      "step": 835
    },
    {
      "epoch": 1.8441273326015368,
      "grad_norm": 2.1940831038305624,
      "learning_rate": 1.9356837515879137e-07,
      "loss": 0.9128,
      "step": 840
    },
    {
      "epoch": 1.8551042810098792,
      "grad_norm": 2.2395348963129176,
      "learning_rate": 1.904577364121077e-07,
      "loss": 0.9137,
      "step": 845
    },
    {
      "epoch": 1.8660812294182216,
      "grad_norm": 2.2788272374999288,
      "learning_rate": 1.873568399865288e-07,
      "loss": 0.8989,
      "step": 850
    },
    {
      "epoch": 1.8770581778265643,
      "grad_norm": 2.364940762438768,
      "learning_rate": 1.8426619325155772e-07,
      "loss": 0.9136,
      "step": 855
    },
    {
      "epoch": 1.8880351262349067,
      "grad_norm": 2.248184524351583,
      "learning_rate": 1.8118630189964055e-07,
      "loss": 0.9234,
      "step": 860
    },
    {
      "epoch": 1.8990120746432493,
      "grad_norm": 2.265297755363968,
      "learning_rate": 1.7811766986342457e-07,
      "loss": 0.9003,
      "step": 865
    },
    {
      "epoch": 1.9099890230515917,
      "grad_norm": 2.3424580199936447,
      "learning_rate": 1.7506079923330492e-07,
      "loss": 0.918,
      "step": 870
    },
    {
      "epoch": 1.920965971459934,
      "grad_norm": 2.262902517301985,
      "learning_rate": 1.7201619017527224e-07,
      "loss": 0.9305,
      "step": 875
    },
    {
      "epoch": 1.9319429198682765,
      "grad_norm": 2.2767358868319385,
      "learning_rate": 1.6898434084907548e-07,
      "loss": 0.9004,
      "step": 880
    },
    {
      "epoch": 1.942919868276619,
      "grad_norm": 2.233676995391588,
      "learning_rate": 1.6596574732671286e-07,
      "loss": 0.9107,
      "step": 885
    },
    {
      "epoch": 1.9538968166849617,
      "grad_norm": 2.2349418347267984,
      "learning_rate": 1.6296090351126445e-07,
      "loss": 0.9012,
      "step": 890
    },
    {
      "epoch": 1.9648737650933041,
      "grad_norm": 2.2644252517319043,
      "learning_rate": 1.5997030105607967e-07,
      "loss": 0.9055,
      "step": 895
    },
    {
      "epoch": 1.9758507135016465,
      "grad_norm": 2.3639914159405824,
      "learning_rate": 1.5699442928433255e-07,
      "loss": 0.8966,
      "step": 900
    },
    {
      "epoch": 1.986827661909989,
      "grad_norm": 2.2554267100734346,
      "learning_rate": 1.5403377510895898e-07,
      "loss": 0.9192,
      "step": 905
    },
    {
      "epoch": 1.9978046103183313,
      "grad_norm": 2.202331515306409,
      "learning_rate": 1.5108882295298748e-07,
      "loss": 0.9077,
      "step": 910
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.9618459939956665,
      "eval_runtime": 3.545,
      "eval_samples_per_second": 75.599,
      "eval_steps_per_second": 1.41,
      "step": 911
    },
    {
      "epoch": 2.008781558726674,
      "grad_norm": 2.3776213420255745,
      "learning_rate": 1.4816005467027793e-07,
      "loss": 0.8907,
      "step": 915
    },
    {
      "epoch": 2.0197585071350166,
      "grad_norm": 2.2333706796571384,
      "learning_rate": 1.452479494666809e-07,
      "loss": 0.8927,
      "step": 920
    },
    {
      "epoch": 2.030735455543359,
      "grad_norm": 2.2112426814454964,
      "learning_rate": 1.4235298382162897e-07,
      "loss": 0.8852,
      "step": 925
    },
    {
      "epoch": 2.0417124039517014,
      "grad_norm": 2.236063023873273,
      "learning_rate": 1.3947563141017593e-07,
      "loss": 0.8955,
      "step": 930
    },
    {
      "epoch": 2.052689352360044,
      "grad_norm": 2.329499119971109,
      "learning_rate": 1.3661636302549355e-07,
      "loss": 0.8726,
      "step": 935
    },
    {
      "epoch": 2.063666300768386,
      "grad_norm": 2.226335466772911,
      "learning_rate": 1.3377564650184008e-07,
      "loss": 0.8844,
      "step": 940
    },
    {
      "epoch": 2.074643249176729,
      "grad_norm": 2.359491075414422,
      "learning_rate": 1.3095394663801344e-07,
      "loss": 0.8888,
      "step": 945
    },
    {
      "epoch": 2.0856201975850714,
      "grad_norm": 2.2422498605446894,
      "learning_rate": 1.2815172512130079e-07,
      "loss": 0.8946,
      "step": 950
    },
    {
      "epoch": 2.096597145993414,
      "grad_norm": 2.34035910547768,
      "learning_rate": 1.2536944045193644e-07,
      "loss": 0.8925,
      "step": 955
    },
    {
      "epoch": 2.1075740944017562,
      "grad_norm": 2.3092350865810105,
      "learning_rate": 1.2260754786808286e-07,
      "loss": 0.8769,
      "step": 960
    },
    {
      "epoch": 2.1185510428100987,
      "grad_norm": 2.285345952327998,
      "learning_rate": 1.198664992713437e-07,
      "loss": 0.8637,
      "step": 965
    },
    {
      "epoch": 2.129527991218441,
      "grad_norm": 2.31950118158217,
      "learning_rate": 1.1714674315282406e-07,
      "loss": 0.8851,
      "step": 970
    },
    {
      "epoch": 2.140504939626784,
      "grad_norm": 2.260100588550131,
      "learning_rate": 1.144487245197481e-07,
      "loss": 0.877,
      "step": 975
    },
    {
      "epoch": 2.1514818880351263,
      "grad_norm": 2.28406902577705,
      "learning_rate": 1.1177288482264652e-07,
      "loss": 0.8731,
      "step": 980
    },
    {
      "epoch": 2.1624588364434687,
      "grad_norm": 2.349769500619024,
      "learning_rate": 1.091196618831268e-07,
      "loss": 0.8741,
      "step": 985
    },
    {
      "epoch": 2.173435784851811,
      "grad_norm": 2.319775552144385,
      "learning_rate": 1.0648948982223657e-07,
      "loss": 0.9009,
      "step": 990
    },
    {
      "epoch": 2.1844127332601535,
      "grad_norm": 2.854032904166289,
      "learning_rate": 1.0388279898943222e-07,
      "loss": 0.8867,
      "step": 995
    },
    {
      "epoch": 2.1953896816684964,
      "grad_norm": 2.3639396187236508,
      "learning_rate": 1.0130001589216566e-07,
      "loss": 0.8848,
      "step": 1000
    },
    {
      "epoch": 2.2063666300768388,
      "grad_norm": 2.3793062354194143,
      "learning_rate": 9.874156312609835e-08,
      "loss": 0.8811,
      "step": 1005
    },
    {
      "epoch": 2.217343578485181,
      "grad_norm": 2.287306409448599,
      "learning_rate": 9.620785930595679e-08,
      "loss": 0.8742,
      "step": 1010
    },
    {
      "epoch": 2.2283205268935236,
      "grad_norm": 2.308427796984264,
      "learning_rate": 9.369931899703821e-08,
      "loss": 0.8879,
      "step": 1015
    },
    {
      "epoch": 2.239297475301866,
      "grad_norm": 2.266095319232965,
      "learning_rate": 9.121635264737923e-08,
      "loss": 0.8976,
      "step": 1020
    },
    {
      "epoch": 2.2502744237102084,
      "grad_norm": 2.290278874944272,
      "learning_rate": 8.87593665205987e-08,
      "loss": 0.8957,
      "step": 1025
    },
    {
      "epoch": 2.261251372118551,
      "grad_norm": 2.351021776661905,
      "learning_rate": 8.632876262942415e-08,
      "loss": 0.8833,
      "step": 1030
    },
    {
      "epoch": 2.2722283205268936,
      "grad_norm": 2.290071235509179,
      "learning_rate": 8.392493866991487e-08,
      "loss": 0.8728,
      "step": 1035
    },
    {
      "epoch": 2.283205268935236,
      "grad_norm": 2.2366625219176037,
      "learning_rate": 8.154828795639057e-08,
      "loss": 0.8925,
      "step": 1040
    },
    {
      "epoch": 2.2941822173435784,
      "grad_norm": 2.2935122329416253,
      "learning_rate": 7.9199199357077e-08,
      "loss": 0.8954,
      "step": 1045
    },
    {
      "epoch": 2.305159165751921,
      "grad_norm": 2.2174582268909346,
      "learning_rate": 7.687805723047952e-08,
      "loss": 0.8755,
      "step": 1050
    },
    {
      "epoch": 2.3161361141602637,
      "grad_norm": 2.320147905008012,
      "learning_rate": 7.45852413624943e-08,
      "loss": 0.8672,
      "step": 1055
    },
    {
      "epoch": 2.327113062568606,
      "grad_norm": 2.311040151424324,
      "learning_rate": 7.232112690426712e-08,
      "loss": 0.8732,
      "step": 1060
    },
    {
      "epoch": 2.3380900109769485,
      "grad_norm": 2.286645866168397,
      "learning_rate": 7.008608431081179e-08,
      "loss": 0.8862,
      "step": 1065
    },
    {
      "epoch": 2.349066959385291,
      "grad_norm": 2.243167376226312,
      "learning_rate": 6.78804792803955e-08,
      "loss": 0.8708,
      "step": 1070
    },
    {
      "epoch": 2.3600439077936333,
      "grad_norm": 2.315245531135358,
      "learning_rate": 6.570467269470375e-08,
      "loss": 0.8822,
      "step": 1075
    },
    {
      "epoch": 2.3710208562019757,
      "grad_norm": 2.268683761726241,
      "learning_rate": 6.355902055979253e-08,
      "loss": 0.879,
      "step": 1080
    },
    {
      "epoch": 2.3819978046103185,
      "grad_norm": 2.312610053338687,
      "learning_rate": 6.144387394783829e-08,
      "loss": 0.8647,
      "step": 1085
    },
    {
      "epoch": 2.392974753018661,
      "grad_norm": 2.2428629980309616,
      "learning_rate": 5.935957893969587e-08,
      "loss": 0.8762,
      "step": 1090
    },
    {
      "epoch": 2.4039517014270033,
      "grad_norm": 2.35765208556055,
      "learning_rate": 5.730647656827242e-08,
      "loss": 0.8841,
      "step": 1095
    },
    {
      "epoch": 2.4149286498353457,
      "grad_norm": 2.2494376360363164,
      "learning_rate": 5.528490276272732e-08,
      "loss": 0.8711,
      "step": 1100
    },
    {
      "epoch": 2.425905598243688,
      "grad_norm": 2.2596581762379793,
      "learning_rate": 5.329518829350788e-08,
      "loss": 0.8724,
      "step": 1105
    },
    {
      "epoch": 2.436882546652031,
      "grad_norm": 2.3221019724494982,
      "learning_rate": 5.1337658718227926e-08,
      "loss": 0.8716,
      "step": 1110
    },
    {
      "epoch": 2.4478594950603734,
      "grad_norm": 2.2548297711291467,
      "learning_rate": 4.941263432840062e-08,
      "loss": 0.8703,
      "step": 1115
    },
    {
      "epoch": 2.4588364434687158,
      "grad_norm": 2.2559502603970776,
      "learning_rate": 4.7520430097031855e-08,
      "loss": 0.891,
      "step": 1120
    },
    {
      "epoch": 2.469813391877058,
      "grad_norm": 2.348427252684778,
      "learning_rate": 4.566135562708437e-08,
      "loss": 0.8697,
      "step": 1125
    },
    {
      "epoch": 2.4807903402854006,
      "grad_norm": 2.216073741504324,
      "learning_rate": 4.383571510082051e-08,
      "loss": 0.8682,
      "step": 1130
    },
    {
      "epoch": 2.491767288693743,
      "grad_norm": 2.295163968424724,
      "learning_rate": 4.20438072300319e-08,
      "loss": 0.8837,
      "step": 1135
    },
    {
      "epoch": 2.5027442371020854,
      "grad_norm": 2.2516483968280343,
      "learning_rate": 4.028592520716387e-08,
      "loss": 0.8638,
      "step": 1140
    },
    {
      "epoch": 2.5137211855104282,
      "grad_norm": 2.269953195416105,
      "learning_rate": 3.8562356657343584e-08,
      "loss": 0.8788,
      "step": 1145
    },
    {
      "epoch": 2.5246981339187706,
      "grad_norm": 2.3024875486631133,
      "learning_rate": 3.6873383591318394e-08,
      "loss": 0.8714,
      "step": 1150
    },
    {
      "epoch": 2.535675082327113,
      "grad_norm": 2.2241713120206037,
      "learning_rate": 3.521928235931346e-08,
      "loss": 0.887,
      "step": 1155
    },
    {
      "epoch": 2.5466520307354554,
      "grad_norm": 2.2936145333162994,
      "learning_rate": 3.3600323605815107e-08,
      "loss": 0.87,
      "step": 1160
    },
    {
      "epoch": 2.5576289791437983,
      "grad_norm": 2.32545502497619,
      "learning_rate": 3.201677222528784e-08,
      "loss": 0.8686,
      "step": 1165
    },
    {
      "epoch": 2.5686059275521407,
      "grad_norm": 2.2489169681563195,
      "learning_rate": 3.0468887318832406e-08,
      "loss": 0.8746,
      "step": 1170
    },
    {
      "epoch": 2.579582875960483,
      "grad_norm": 2.2858684159165734,
      "learning_rate": 2.8956922151791547e-08,
      "loss": 0.8894,
      "step": 1175
    },
    {
      "epoch": 2.5905598243688255,
      "grad_norm": 2.2940920483922382,
      "learning_rate": 2.748112411231046e-08,
      "loss": 0.8682,
      "step": 1180
    },
    {
      "epoch": 2.601536772777168,
      "grad_norm": 2.2496855474263366,
      "learning_rate": 2.6041734670859488e-08,
      "loss": 0.8694,
      "step": 1185
    },
    {
      "epoch": 2.6125137211855103,
      "grad_norm": 2.3184342673634055,
      "learning_rate": 2.463898934072417e-08,
      "loss": 0.8742,
      "step": 1190
    },
    {
      "epoch": 2.6234906695938527,
      "grad_norm": 2.2891655132723576,
      "learning_rate": 2.3273117639470958e-08,
      "loss": 0.8692,
      "step": 1195
    },
    {
      "epoch": 2.6344676180021955,
      "grad_norm": 2.186621510859105,
      "learning_rate": 2.1944343051393173e-08,
      "loss": 0.8871,
      "step": 1200
    },
    {
      "epoch": 2.645444566410538,
      "grad_norm": 2.1909337313948467,
      "learning_rate": 2.0652882990944532e-08,
      "loss": 0.882,
      "step": 1205
    },
    {
      "epoch": 2.6564215148188803,
      "grad_norm": 2.3021129840080827,
      "learning_rate": 1.9398948767165774e-08,
      "loss": 0.8798,
      "step": 1210
    },
    {
      "epoch": 2.6673984632272227,
      "grad_norm": 2.233026500612387,
      "learning_rate": 1.818274554911034e-08,
      "loss": 0.87,
      "step": 1215
    },
    {
      "epoch": 2.6783754116355656,
      "grad_norm": 2.3034061687210925,
      "learning_rate": 1.7004472332274117e-08,
      "loss": 0.8878,
      "step": 1220
    },
    {
      "epoch": 2.689352360043908,
      "grad_norm": 2.2448202497203957,
      "learning_rate": 1.586432190603626e-08,
      "loss": 0.8924,
      "step": 1225
    },
    {
      "epoch": 2.7003293084522504,
      "grad_norm": 2.297719442799476,
      "learning_rate": 1.4762480822114731e-08,
      "loss": 0.8702,
      "step": 1230
    },
    {
      "epoch": 2.711306256860593,
      "grad_norm": 2.2378202672720273,
      "learning_rate": 1.3699129364042522e-08,
      "loss": 0.8907,
      "step": 1235
    },
    {
      "epoch": 2.722283205268935,
      "grad_norm": 2.2880262820978703,
      "learning_rate": 1.267444151766986e-08,
      "loss": 0.9005,
      "step": 1240
    },
    {
      "epoch": 2.7332601536772776,
      "grad_norm": 2.338012677352183,
      "learning_rate": 1.1688584942696366e-08,
      "loss": 0.8904,
      "step": 1245
    },
    {
      "epoch": 2.74423710208562,
      "grad_norm": 2.289922529284634,
      "learning_rate": 1.0741720945238731e-08,
      "loss": 0.8986,
      "step": 1250
    },
    {
      "epoch": 2.755214050493963,
      "grad_norm": 2.2470416957804273,
      "learning_rate": 9.834004451437699e-09,
      "loss": 0.8728,
      "step": 1255
    },
    {
      "epoch": 2.7661909989023052,
      "grad_norm": 2.2478830819509303,
      "learning_rate": 8.965583982108865e-09,
      "loss": 0.8754,
      "step": 1260
    },
    {
      "epoch": 2.7771679473106476,
      "grad_norm": 2.244470092578961,
      "learning_rate": 8.136601628441875e-09,
      "loss": 0.8828,
      "step": 1265
    },
    {
      "epoch": 2.78814489571899,
      "grad_norm": 2.2299781074531326,
      "learning_rate": 7.347193028751364e-09,
      "loss": 0.8755,
      "step": 1270
    },
    {
      "epoch": 2.7991218441273324,
      "grad_norm": 2.379183267890298,
      "learning_rate": 6.597487346283626e-09,
      "loss": 0.8697,
      "step": 1275
    },
    {
      "epoch": 2.8100987925356753,
      "grad_norm": 2.3021772414641837,
      "learning_rate": 5.8876072480831264e-09,
      "loss": 0.8849,
      "step": 1280
    },
    {
      "epoch": 2.8210757409440177,
      "grad_norm": 2.2739127597741144,
      "learning_rate": 5.217668884921505e-09,
      "loss": 0.8889,
      "step": 1285
    },
    {
      "epoch": 2.83205268935236,
      "grad_norm": 2.239037460712834,
      "learning_rate": 4.587781872293056e-09,
      "loss": 0.8636,
      "step": 1290
    },
    {
      "epoch": 2.8430296377607025,
      "grad_norm": 2.3065882078621436,
      "learning_rate": 3.998049272479431e-09,
      "loss": 0.8643,
      "step": 1295
    },
    {
      "epoch": 2.854006586169045,
      "grad_norm": 2.2987652140661265,
      "learning_rate": 3.4485675776863843e-09,
      "loss": 0.882,
      "step": 1300
    },
    {
      "epoch": 2.8649835345773873,
      "grad_norm": 2.2776905750235956,
      "learning_rate": 2.9394266942558976e-09,
      "loss": 0.8883,
      "step": 1305
    },
    {
      "epoch": 2.8759604829857297,
      "grad_norm": 2.2719810844272406,
      "learning_rate": 2.4707099279556164e-09,
      "loss": 0.9002,
      "step": 1310
    },
    {
      "epoch": 2.8869374313940726,
      "grad_norm": 2.393629212210359,
      "learning_rate": 2.04249397034828e-09,
      "loss": 0.8737,
      "step": 1315
    },
    {
      "epoch": 2.897914379802415,
      "grad_norm": 2.318705736365544,
      "learning_rate": 1.6548488862435095e-09,
      "loss": 0.8641,
      "step": 1320
    },
    {
      "epoch": 2.9088913282107574,
      "grad_norm": 2.1750311426072604,
      "learning_rate": 1.3078381022336715e-09,
      "loss": 0.8636,
      "step": 1325
    },
    {
      "epoch": 2.9198682766190998,
      "grad_norm": 2.2319410876071166,
      "learning_rate": 1.0015183963161811e-09,
      "loss": 0.8631,
      "step": 1330
    },
    {
      "epoch": 2.9308452250274426,
      "grad_norm": 2.2205987624431143,
      "learning_rate": 7.359398886032653e-10,
      "loss": 0.8888,
      "step": 1335
    },
    {
      "epoch": 2.941822173435785,
      "grad_norm": 2.255451203517086,
      "learning_rate": 5.111460331214124e-10,
      "loss": 0.8845,
      "step": 1340
    },
    {
      "epoch": 2.9527991218441274,
      "grad_norm": 2.360197105909993,
      "learning_rate": 3.271736107015033e-10,
      "loss": 0.8999,
      "step": 1345
    },
    {
      "epoch": 2.96377607025247,
      "grad_norm": 2.2839001061008553,
      "learning_rate": 1.8405272296045937e-10,
      "loss": 0.8876,
      "step": 1350
    },
    {
      "epoch": 2.974753018660812,
      "grad_norm": 2.240462357412488,
      "learning_rate": 8.180678737629287e-11,
      "loss": 0.8671,
      "step": 1355
    },
    {
      "epoch": 2.9857299670691546,
      "grad_norm": 2.282162777030651,
      "learning_rate": 2.0452533456311037e-11,
      "loss": 0.875,
      "step": 1360
    },
    {
      "epoch": 2.996706915477497,
      "grad_norm": 2.320056946149888,
      "learning_rate": 0.0,
      "loss": 0.8919,
      "step": 1365
    },
    {
      "epoch": 2.996706915477497,
      "eval_loss": 0.9529117345809937,
      "eval_runtime": 3.471,
      "eval_samples_per_second": 77.211,
      "eval_steps_per_second": 1.44,
      "step": 1365
    },
    {
      "epoch": 2.996706915477497,
      "step": 1365,
      "total_flos": 571501770178560.0,
      "train_loss": 0.9509169136648213,
      "train_runtime": 15780.8295,
      "train_samples_per_second": 22.162,
      "train_steps_per_second": 0.086
    }
  ],
  "logging_steps": 5,
  "max_steps": 1365,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 571501770178560.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}