{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.5530429503718315,
  "global_step": 30000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 9.4799,
      "step": 200
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.533333333333333e-06,
      "loss": 7.4409,
      "step": 400
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.866666666666667e-06,
      "loss": 6.6304,
      "step": 600
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.3183333333333333e-05,
      "loss": 5.8163,
      "step": 800
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.6516666666666667e-05,
      "loss": 5.2109,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.985e-05,
      "loss": 4.7702,
      "step": 1200
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.3183333333333336e-05,
      "loss": 4.4959,
      "step": 1400
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.6516666666666666e-05,
      "loss": 4.3014,
      "step": 1600
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.985e-05,
      "loss": 4.1589,
      "step": 1800
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.318333333333333e-05,
      "loss": 4.0593,
      "step": 2000
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.651666666666667e-05,
      "loss": 3.9601,
      "step": 2200
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.9850000000000006e-05,
      "loss": 3.8843,
      "step": 2400
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.318333333333334e-05,
      "loss": 3.8441,
      "step": 2600
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.651666666666667e-05,
      "loss": 3.8172,
      "step": 2800
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.9850000000000006e-05,
      "loss": 3.7464,
      "step": 3000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.968108198363667e-05,
      "loss": 3.7161,
      "step": 3200
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.934713641676407e-05,
      "loss": 3.6457,
      "step": 3400
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.901319084989147e-05,
      "loss": 3.61,
      "step": 3600
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.867924528301887e-05,
      "loss": 3.579,
      "step": 3800
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.834529971614627e-05,
      "loss": 3.5302,
      "step": 4000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.8011354149273667e-05,
      "loss": 3.5092,
      "step": 4200
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.767740858240107e-05,
      "loss": 3.485,
      "step": 4400
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.734346301552847e-05,
      "loss": 3.4616,
      "step": 4600
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.7009517448655874e-05,
      "loss": 3.4466,
      "step": 4800
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.667557188178327e-05,
      "loss": 3.4108,
      "step": 5000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.6341626314910676e-05,
      "loss": 3.3687,
      "step": 5200
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.6007680748038074e-05,
      "loss": 3.3308,
      "step": 5400
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.567373518116547e-05,
      "loss": 3.3314,
      "step": 5600
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.533978961429287e-05,
      "loss": 3.3101,
      "step": 5800
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.500584404742027e-05,
      "loss": 3.3053,
      "step": 6000
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.467189848054767e-05,
      "loss": 3.2995,
      "step": 6200
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.433795291367508e-05,
      "loss": 3.2495,
      "step": 6400
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.4004007346802475e-05,
      "loss": 3.2186,
      "step": 6600
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.367006177992987e-05,
      "loss": 2.9888,
      "step": 6800
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.333611621305728e-05,
      "loss": 2.98,
      "step": 7000
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.3002170646184675e-05,
      "loss": 2.9622,
      "step": 7200
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.266822507931207e-05,
      "loss": 2.9751,
      "step": 7400
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.233427951243947e-05,
      "loss": 2.9455,
      "step": 7600
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.200033394556687e-05,
      "loss": 2.9657,
      "step": 7800
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.1666388378694274e-05,
      "loss": 2.9428,
      "step": 8000
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.133244281182168e-05,
      "loss": 2.96,
      "step": 8200
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.0998497244949076e-05,
      "loss": 2.935,
      "step": 8400
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.0664551678076474e-05,
      "loss": 2.9543,
      "step": 8600
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.033060611120388e-05,
      "loss": 2.9213,
      "step": 8800
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.9996660544331277e-05,
      "loss": 2.9299,
      "step": 9000
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.9662714977458674e-05,
      "loss": 2.9429,
      "step": 9200
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.932876941058607e-05,
      "loss": 2.936,
      "step": 9400
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.899649357154784e-05,
      "loss": 2.8932,
      "step": 9600
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.8662548004675245e-05,
      "loss": 2.9069,
      "step": 9800
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.832860243780264e-05,
      "loss": 2.8815,
      "step": 10000
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.799465687093004e-05,
      "loss": 2.8844,
      "step": 10200
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.766071130405744e-05,
      "loss": 2.8921,
      "step": 10400
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.73284354650192e-05,
      "loss": 2.8557,
      "step": 10600
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.6994489898146604e-05,
      "loss": 2.8768,
      "step": 10800
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.6660544331274e-05,
      "loss": 2.8527,
      "step": 11000
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.632659876440141e-05,
      "loss": 2.8839,
      "step": 11200
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.5992653197528805e-05,
      "loss": 2.8512,
      "step": 11400
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.565870763065621e-05,
      "loss": 2.8428,
      "step": 11600
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.532476206378361e-05,
      "loss": 2.8329,
      "step": 11800
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.4990816496911005e-05,
      "loss": 2.8266,
      "step": 12000
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.46568709300384e-05,
      "loss": 2.8282,
      "step": 12200
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.4324595091000164e-05,
      "loss": 2.8202,
      "step": 12400
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.399064952412757e-05,
      "loss": 2.8035,
      "step": 12600
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.365670395725497e-05,
      "loss": 2.8211,
      "step": 12800
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.332275839038237e-05,
      "loss": 2.796,
      "step": 13000
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.298881282350977e-05,
      "loss": 2.766,
      "step": 13200
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.265486725663717e-05,
      "loss": 2.5018,
      "step": 13400
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.232092168976457e-05,
      "loss": 2.5239,
      "step": 13600
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.198697612289197e-05,
      "loss": 2.4989,
      "step": 13800
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.165303055601937e-05,
      "loss": 2.5141,
      "step": 14000
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.1319084989146766e-05,
      "loss": 2.4981,
      "step": 14200
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.098513942227417e-05,
      "loss": 2.5149,
      "step": 14400
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.0651193855401575e-05,
      "loss": 2.4817,
      "step": 14600
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.031724828852897e-05,
      "loss": 2.5131,
      "step": 14800
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.998330272165637e-05,
      "loss": 2.5058,
      "step": 15000
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.964935715478377e-05,
      "loss": 2.5254,
      "step": 15200
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.9315411587911173e-05,
      "loss": 2.5074,
      "step": 15400
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.898146602103857e-05,
      "loss": 2.5147,
      "step": 15600
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.8647520454165972e-05,
      "loss": 2.4908,
      "step": 15800
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.8315244615127734e-05,
      "loss": 2.478,
      "step": 16000
    },
    {
      "epoch": 2.46,
      "learning_rate": 2.7981299048255138e-05,
      "loss": 2.5051,
      "step": 16200
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.7647353481382536e-05,
      "loss": 2.5101,
      "step": 16400
    },
    {
      "epoch": 2.52,
      "learning_rate": 2.7313407914509937e-05,
      "loss": 2.502,
      "step": 16600
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.6979462347637335e-05,
      "loss": 2.5053,
      "step": 16800
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.6645516780764733e-05,
      "loss": 2.4969,
      "step": 17000
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.6311571213892138e-05,
      "loss": 2.4882,
      "step": 17200
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.597762564701954e-05,
      "loss": 2.4755,
      "step": 17400
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.5643680080146937e-05,
      "loss": 2.4798,
      "step": 17600
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.5309734513274335e-05,
      "loss": 2.4795,
      "step": 17800
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.4975788946401736e-05,
      "loss": 2.4913,
      "step": 18000
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.464184337952914e-05,
      "loss": 2.4894,
      "step": 18200
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.430789781265654e-05,
      "loss": 2.4678,
      "step": 18400
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.3973952245783936e-05,
      "loss": 2.4418,
      "step": 18600
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.3640006678911338e-05,
      "loss": 2.4705,
      "step": 18800
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.330606111203874e-05,
      "loss": 2.4659,
      "step": 19000
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.297211554516614e-05,
      "loss": 2.4752,
      "step": 19200
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.2638169978293538e-05,
      "loss": 2.4694,
      "step": 19400
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.230422441142094e-05,
      "loss": 2.4442,
      "step": 19600
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.197027884454834e-05,
      "loss": 2.4312,
      "step": 19800
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.1636333277675742e-05,
      "loss": 2.2147,
      "step": 20000
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.130238771080314e-05,
      "loss": 2.2146,
      "step": 20200
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.0968442143930538e-05,
      "loss": 2.213,
      "step": 20400
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.0634496577057942e-05,
      "loss": 2.2176,
      "step": 20600
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.030055101018534e-05,
      "loss": 2.2169,
      "step": 20800
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.9968275171147105e-05,
      "loss": 2.223,
      "step": 21000
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.9634329604274503e-05,
      "loss": 2.2279,
      "step": 21200
    },
    {
      "epoch": 3.25,
      "learning_rate": 1.9300384037401907e-05,
      "loss": 2.2322,
      "step": 21400
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.8966438470529305e-05,
      "loss": 2.2428,
      "step": 21600
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.8632492903656703e-05,
      "loss": 2.235,
      "step": 21800
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.8298547336784104e-05,
      "loss": 2.2323,
      "step": 22000
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.7964601769911506e-05,
      "loss": 2.1997,
      "step": 22200
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.7630656203038907e-05,
      "loss": 2.2353,
      "step": 22400
    },
    {
      "epoch": 3.43,
      "learning_rate": 1.7296710636166305e-05,
      "loss": 2.2258,
      "step": 22600
    },
    {
      "epoch": 3.46,
      "learning_rate": 1.6962765069293706e-05,
      "loss": 2.2269,
      "step": 22800
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.6628819502421107e-05,
      "loss": 2.2351,
      "step": 23000
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.629654366338287e-05,
      "loss": 2.2387,
      "step": 23200
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.596259809651027e-05,
      "loss": 2.2285,
      "step": 23400
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.5628652529637668e-05,
      "loss": 2.2378,
      "step": 23600
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.5294706962765072e-05,
      "loss": 2.2274,
      "step": 23800
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.496076139589247e-05,
      "loss": 2.2081,
      "step": 24000
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.4626815829019871e-05,
      "loss": 2.2183,
      "step": 24200
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.4292870262147271e-05,
      "loss": 2.2316,
      "step": 24400
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.3958924695274672e-05,
      "loss": 2.2066,
      "step": 24600
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.3624979128402072e-05,
      "loss": 2.2214,
      "step": 24800
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.329103356152947e-05,
      "loss": 2.2308,
      "step": 25000
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.2957087994656873e-05,
      "loss": 2.2389,
      "step": 25200
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.262314242778427e-05,
      "loss": 2.226,
      "step": 25400
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.2289196860911672e-05,
      "loss": 2.2305,
      "step": 25600
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.1955251294039071e-05,
      "loss": 2.2206,
      "step": 25800
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.1621305727166473e-05,
      "loss": 2.2303,
      "step": 26000
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.1287360160293872e-05,
      "loss": 2.218,
      "step": 26200
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.0953414593421272e-05,
      "loss": 2.1781,
      "step": 26400
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.0619469026548673e-05,
      "loss": 2.0351,
      "step": 26600
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.0285523459676073e-05,
      "loss": 2.0511,
      "step": 26800
    },
    {
      "epoch": 4.1,
      "learning_rate": 9.951577892803474e-06,
      "loss": 2.0529,
      "step": 27000
    },
    {
      "epoch": 4.13,
      "learning_rate": 9.617632325930873e-06,
      "loss": 2.0306,
      "step": 27200
    },
    {
      "epoch": 4.16,
      "learning_rate": 9.283686759058275e-06,
      "loss": 2.0486,
      "step": 27400
    },
    {
      "epoch": 4.19,
      "learning_rate": 8.949741192185674e-06,
      "loss": 2.0505,
      "step": 27600
    },
    {
      "epoch": 4.22,
      "learning_rate": 8.615795625313076e-06,
      "loss": 2.0395,
      "step": 27800
    },
    {
      "epoch": 4.25,
      "learning_rate": 8.281850058440473e-06,
      "loss": 2.0387,
      "step": 28000
    },
    {
      "epoch": 4.28,
      "learning_rate": 7.947904491567875e-06,
      "loss": 2.0537,
      "step": 28200
    },
    {
      "epoch": 4.31,
      "learning_rate": 7.615628652529638e-06,
      "loss": 2.0339,
      "step": 28400
    },
    {
      "epoch": 4.34,
      "learning_rate": 7.281683085657039e-06,
      "loss": 2.0676,
      "step": 28600
    },
    {
      "epoch": 4.37,
      "learning_rate": 6.947737518784439e-06,
      "loss": 2.0498,
      "step": 28800
    },
    {
      "epoch": 4.4,
      "learning_rate": 6.61379195191184e-06,
      "loss": 2.0341,
      "step": 29000
    },
    {
      "epoch": 4.43,
      "learning_rate": 6.279846385039238e-06,
      "loss": 2.0446,
      "step": 29200
    },
    {
      "epoch": 4.46,
      "learning_rate": 5.94590081816664e-06,
      "loss": 2.0435,
      "step": 29400
    },
    {
      "epoch": 4.49,
      "learning_rate": 5.611955251294039e-06,
      "loss": 2.0513,
      "step": 29600
    },
    {
      "epoch": 4.52,
      "learning_rate": 5.27800968442144e-06,
      "loss": 2.0378,
      "step": 29800
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.94406411754884e-06,
      "loss": 2.0463,
      "step": 30000
    }
  ],
  "max_steps": 32945,
  "num_train_epochs": 5,
  "total_flos": 1.0567419081666048e+17,
  "trial_name": null,
  "trial_params": null
}