{ "best_metric": 0.15095455944538116, "best_model_checkpoint": "AlexWang99/byt5_add_5k/checkpoint-329", "epoch": 47.0, "eval_steps": 500, "global_step": 329, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "eval_loss": 1.0333173274993896, "eval_runtime": 10.7395, "eval_samples_per_second": 931.141, "eval_steps_per_second": 1.21, "step": 7 }, { "epoch": 2.0, "eval_loss": 0.9484947323799133, "eval_runtime": 10.7979, "eval_samples_per_second": 926.109, "eval_steps_per_second": 1.204, "step": 14 }, { "epoch": 3.0, "eval_loss": 0.8662577271461487, "eval_runtime": 10.9139, "eval_samples_per_second": 916.266, "eval_steps_per_second": 1.191, "step": 21 }, { "epoch": 4.0, "eval_loss": 0.816236138343811, "eval_runtime": 10.8703, "eval_samples_per_second": 919.938, "eval_steps_per_second": 1.196, "step": 28 }, { "epoch": 5.0, "eval_loss": 0.7725976705551147, "eval_runtime": 11.0507, "eval_samples_per_second": 904.923, "eval_steps_per_second": 1.176, "step": 35 }, { "epoch": 6.0, "eval_loss": 0.7079625725746155, "eval_runtime": 10.9553, "eval_samples_per_second": 912.803, "eval_steps_per_second": 1.187, "step": 42 }, { "epoch": 7.0, "eval_loss": 0.6394063234329224, "eval_runtime": 10.8958, "eval_samples_per_second": 917.786, "eval_steps_per_second": 1.193, "step": 49 }, { "epoch": 8.0, "eval_loss": 0.5353133678436279, "eval_runtime": 11.1585, "eval_samples_per_second": 896.181, "eval_steps_per_second": 1.165, "step": 56 }, { "epoch": 9.0, "eval_loss": 0.501313328742981, "eval_runtime": 10.9739, "eval_samples_per_second": 911.249, "eval_steps_per_second": 1.185, "step": 63 }, { "epoch": 10.0, "eval_loss": 0.45894965529441833, "eval_runtime": 10.8763, "eval_samples_per_second": 919.43, "eval_steps_per_second": 1.195, "step": 70 }, { "epoch": 11.0, "eval_loss": 0.4472431540489197, "eval_runtime": 10.9138, "eval_samples_per_second": 916.274, "eval_steps_per_second": 1.191, "step": 77 }, { "epoch": 12.0, "eval_loss": 0.4227907359600067, "eval_runtime": 10.9807, "eval_samples_per_second": 910.685, "eval_steps_per_second": 1.184, "step": 84 }, { "epoch": 13.0, "eval_loss": 0.39944639801979065, "eval_runtime": 11.1342, "eval_samples_per_second": 898.131, "eval_steps_per_second": 1.168, "step": 91 }, { "epoch": 14.0, "eval_loss": 0.38267987966537476, "eval_runtime": 10.9641, "eval_samples_per_second": 912.068, "eval_steps_per_second": 1.186, "step": 98 }, { "epoch": 15.0, "eval_loss": 0.319256991147995, "eval_runtime": 10.7912, "eval_samples_per_second": 926.681, "eval_steps_per_second": 1.205, "step": 105 }, { "epoch": 16.0, "eval_loss": 0.3410109579563141, "eval_runtime": 10.7915, "eval_samples_per_second": 926.652, "eval_steps_per_second": 1.205, "step": 112 }, { "epoch": 17.0, "eval_loss": 0.32387295365333557, "eval_runtime": 10.7997, "eval_samples_per_second": 925.948, "eval_steps_per_second": 1.204, "step": 119 }, { "epoch": 18.0, "eval_loss": 0.2893741726875305, "eval_runtime": 10.8261, "eval_samples_per_second": 923.697, "eval_steps_per_second": 1.201, "step": 126 }, { "epoch": 19.0, "eval_loss": 0.26235708594322205, "eval_runtime": 10.9401, "eval_samples_per_second": 914.071, "eval_steps_per_second": 1.188, "step": 133 }, { "epoch": 20.0, "eval_loss": 0.26693689823150635, "eval_runtime": 10.9641, "eval_samples_per_second": 912.07, "eval_steps_per_second": 1.186, "step": 140 }, { "epoch": 21.0, "eval_loss": 0.2766812741756439, "eval_runtime": 11.0072, "eval_samples_per_second": 908.497, "eval_steps_per_second": 1.181, "step": 147 }, { "epoch": 22.0, "eval_loss": 0.24048474431037903, "eval_runtime": 10.8863, "eval_samples_per_second": 918.586, "eval_steps_per_second": 1.194, "step": 154 }, { "epoch": 23.0, "eval_loss": 0.23866714537143707, "eval_runtime": 10.7952, "eval_samples_per_second": 926.334, "eval_steps_per_second": 1.204, "step": 161 }, { "epoch": 24.0, "eval_loss": 0.24236781895160675, "eval_runtime": 11.0325, "eval_samples_per_second": 906.412, "eval_steps_per_second": 1.178, "step": 168 }, { "epoch": 25.0, "eval_loss": 0.2102736085653305, "eval_runtime": 10.9008, "eval_samples_per_second": 917.366, "eval_steps_per_second": 1.193, "step": 175 }, { "epoch": 26.0, "eval_loss": 0.22354699671268463, "eval_runtime": 10.8227, "eval_samples_per_second": 923.988, "eval_steps_per_second": 1.201, "step": 182 }, { "epoch": 27.0, "eval_loss": 0.21701256930828094, "eval_runtime": 10.8903, "eval_samples_per_second": 918.249, "eval_steps_per_second": 1.194, "step": 189 }, { "epoch": 28.0, "eval_loss": 0.20713335275650024, "eval_runtime": 11.0977, "eval_samples_per_second": 901.088, "eval_steps_per_second": 1.171, "step": 196 }, { "epoch": 29.0, "eval_loss": 0.19325299561023712, "eval_runtime": 10.8557, "eval_samples_per_second": 921.179, "eval_steps_per_second": 1.198, "step": 203 }, { "epoch": 30.0, "eval_loss": 0.19234418869018555, "eval_runtime": 10.8944, "eval_samples_per_second": 917.9, "eval_steps_per_second": 1.193, "step": 210 }, { "epoch": 31.0, "eval_loss": 0.19294323027133942, "eval_runtime": 11.1467, "eval_samples_per_second": 897.123, "eval_steps_per_second": 1.166, "step": 217 }, { "epoch": 32.0, "eval_loss": 0.18856723606586456, "eval_runtime": 10.9076, "eval_samples_per_second": 916.795, "eval_steps_per_second": 1.192, "step": 224 }, { "epoch": 33.0, "eval_loss": 0.1928592473268509, "eval_runtime": 10.9684, "eval_samples_per_second": 911.712, "eval_steps_per_second": 1.185, "step": 231 }, { "epoch": 34.0, "eval_loss": 0.17712894082069397, "eval_runtime": 10.9664, "eval_samples_per_second": 911.873, "eval_steps_per_second": 1.185, "step": 238 }, { "epoch": 35.0, "eval_loss": 0.1747332066297531, "eval_runtime": 10.9577, "eval_samples_per_second": 912.602, "eval_steps_per_second": 1.186, "step": 245 }, { "epoch": 36.0, "eval_loss": 0.1693752259016037, "eval_runtime": 10.8762, "eval_samples_per_second": 919.44, "eval_steps_per_second": 1.195, "step": 252 }, { "epoch": 37.0, "eval_loss": 0.17445072531700134, "eval_runtime": 10.872, "eval_samples_per_second": 919.793, "eval_steps_per_second": 1.196, "step": 259 }, { "epoch": 38.0, "eval_loss": 0.1689160168170929, "eval_runtime": 11.1193, "eval_samples_per_second": 899.339, "eval_steps_per_second": 1.169, "step": 266 }, { "epoch": 39.0, "eval_loss": 0.16354702413082123, "eval_runtime": 11.1231, "eval_samples_per_second": 899.033, "eval_steps_per_second": 1.169, "step": 273 }, { "epoch": 40.0, "eval_loss": 0.1702088564634323, "eval_runtime": 11.117, "eval_samples_per_second": 899.521, "eval_steps_per_second": 1.169, "step": 280 }, { "epoch": 41.0, "eval_loss": 0.16428792476654053, "eval_runtime": 10.9632, "eval_samples_per_second": 912.145, "eval_steps_per_second": 1.186, "step": 287 }, { "epoch": 42.0, "eval_loss": 0.16399501264095306, "eval_runtime": 10.8713, "eval_samples_per_second": 919.853, "eval_steps_per_second": 1.196, "step": 294 }, { "epoch": 43.0, "eval_loss": 0.15999095141887665, "eval_runtime": 10.8706, "eval_samples_per_second": 919.913, "eval_steps_per_second": 1.196, "step": 301 }, { "epoch": 44.0, "eval_loss": 0.16396267712116241, "eval_runtime": 11.1122, "eval_samples_per_second": 899.908, "eval_steps_per_second": 1.17, "step": 308 }, { "epoch": 45.0, "eval_loss": 0.16081862151622772, "eval_runtime": 10.8849, "eval_samples_per_second": 918.705, "eval_steps_per_second": 1.194, "step": 315 }, { "epoch": 46.0, "eval_loss": 0.1533430963754654, "eval_runtime": 10.8716, "eval_samples_per_second": 919.827, "eval_steps_per_second": 1.196, "step": 322 }, { "epoch": 47.0, "eval_loss": 0.15095455944538116, "eval_runtime": 10.9494, "eval_samples_per_second": 913.291, "eval_steps_per_second": 1.187, "step": 329 } ], "logging_steps": 500, "max_steps": 350, "num_train_epochs": 50, "save_steps": 500, "total_flos": 6747067207680000.0, "trial_name": null, "trial_params": null }