{ "best_metric": 173.47916334025228, "best_model_checkpoint": "./whisper-training-blog/checkpoint-396", "epoch": 2.330357142857143, "global_step": 448, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 1.6666666666666668e-07, "loss": 1.7185, "step": 4 }, { "epoch": 0.02, "learning_rate": 3.3333333333333335e-07, "loss": 1.6808, "step": 8 }, { "epoch": 0.03, "learning_rate": 5.555555555555555e-07, "loss": 1.712, "step": 12 }, { "epoch": 0.04, "learning_rate": 7.777777777777778e-07, "loss": 1.543, "step": 16 }, { "epoch": 0.04, "learning_rate": 1e-06, "loss": 1.609, "step": 20 }, { "epoch": 0.05, "learning_rate": 1.2222222222222223e-06, "loss": 1.5657, "step": 24 }, { "epoch": 0.06, "learning_rate": 1.4444444444444445e-06, "loss": 1.608, "step": 28 }, { "epoch": 0.07, "learning_rate": 1.6666666666666667e-06, "loss": 1.5075, "step": 32 }, { "epoch": 0.08, "learning_rate": 1.8888888888888888e-06, "loss": 1.5183, "step": 36 }, { "epoch": 0.09, "learning_rate": 2.1111111111111114e-06, "loss": 1.4948, "step": 40 }, { "epoch": 0.1, "learning_rate": 2.3333333333333336e-06, "loss": 1.4112, "step": 44 }, { "epoch": 0.1, "eval_loss": 1.4919109344482422, "eval_runtime": 118.4455, "eval_samples_per_second": 2.786, "eval_steps_per_second": 0.355, "eval_wer": 245.29778061631805, "step": 44 }, { "epoch": 0.11, "learning_rate": 2.5555555555555557e-06, "loss": 1.3192, "step": 48 }, { "epoch": 0.12, "learning_rate": 2.7777777777777775e-06, "loss": 1.3334, "step": 52 }, { "epoch": 0.12, "learning_rate": 3e-06, "loss": 1.2895, "step": 56 }, { "epoch": 0.13, "learning_rate": 3.2222222222222222e-06, "loss": 1.2749, "step": 60 }, { "epoch": 0.14, "learning_rate": 3.4444444444444444e-06, "loss": 1.3856, "step": 64 }, { "epoch": 0.15, "learning_rate": 3.6666666666666666e-06, "loss": 1.2078, "step": 68 }, { "epoch": 0.16, "learning_rate": 3.888888888888889e-06, "loss": 1.2185, "step": 72 }, { "epoch": 0.17, "learning_rate": 4.111111111111111e-06, "loss": 1.1661, "step": 76 }, { "epoch": 0.18, "learning_rate": 4.333333333333333e-06, "loss": 1.1351, "step": 80 }, { "epoch": 0.19, "learning_rate": 4.555555555555555e-06, "loss": 1.0729, "step": 84 }, { "epoch": 0.2, "learning_rate": 4.777777777777778e-06, "loss": 1.0501, "step": 88 }, { "epoch": 0.2, "eval_loss": 1.225467324256897, "eval_runtime": 119.277, "eval_samples_per_second": 2.767, "eval_steps_per_second": 0.352, "eval_wer": 219.94251955931662, "step": 88 }, { "epoch": 0.21, "learning_rate": 4.9999999999999996e-06, "loss": 1.1278, "step": 92 }, { "epoch": 0.21, "learning_rate": 5.2222222222222226e-06, "loss": 1.0563, "step": 96 }, { "epoch": 0.22, "learning_rate": 5.444444444444445e-06, "loss": 0.9929, "step": 100 }, { "epoch": 0.23, "learning_rate": 5.666666666666667e-06, "loss": 1.0879, "step": 104 }, { "epoch": 0.24, "learning_rate": 5.888888888888889e-06, "loss": 1.0417, "step": 108 }, { "epoch": 0.25, "learning_rate": 6.111111111111111e-06, "loss": 0.9337, "step": 112 }, { "epoch": 0.26, "learning_rate": 6.333333333333333e-06, "loss": 0.9688, "step": 116 }, { "epoch": 0.27, "learning_rate": 6.555555555555556e-06, "loss": 0.9559, "step": 120 }, { "epoch": 0.28, "learning_rate": 6.777777777777778e-06, "loss": 1.0192, "step": 124 }, { "epoch": 0.29, "learning_rate": 7e-06, "loss": 0.8509, "step": 128 }, { "epoch": 0.29, "learning_rate": 7.222222222222222e-06, "loss": 0.9033, "step": 132 }, { "epoch": 0.29, "eval_loss": 1.120314359664917, "eval_runtime": 111.0003, "eval_samples_per_second": 2.973, "eval_steps_per_second": 0.378, "eval_wer": 205.7799776464953, "step": 132 }, { "epoch": 0.3, "learning_rate": 7.444444444444444e-06, "loss": 0.9171, "step": 136 }, { "epoch": 0.31, "learning_rate": 7.428115015974441e-06, "loss": 0.9445, "step": 140 }, { "epoch": 0.32, "learning_rate": 7.332268370607029e-06, "loss": 0.8926, "step": 144 }, { "epoch": 0.33, "learning_rate": 7.236421725239617e-06, "loss": 0.9975, "step": 148 }, { "epoch": 1.0, "learning_rate": 7.140575079872205e-06, "loss": 1.239, "step": 152 }, { "epoch": 1.01, "learning_rate": 7.044728434504793e-06, "loss": 0.9148, "step": 156 }, { "epoch": 1.02, "learning_rate": 6.948881789137381e-06, "loss": 1.0117, "step": 160 }, { "epoch": 1.03, "learning_rate": 6.853035143769968e-06, "loss": 0.9081, "step": 164 }, { "epoch": 1.04, "learning_rate": 6.7571884984025565e-06, "loss": 0.8434, "step": 168 }, { "epoch": 1.05, "learning_rate": 6.6613418530351436e-06, "loss": 0.8672, "step": 172 }, { "epoch": 1.06, "learning_rate": 6.565495207667732e-06, "loss": 0.8142, "step": 176 }, { "epoch": 1.06, "eval_loss": 1.0674585103988647, "eval_runtime": 102.4925, "eval_samples_per_second": 3.22, "eval_steps_per_second": 0.41, "eval_wer": 192.87881207089254, "step": 176 }, { "epoch": 1.07, "learning_rate": 6.469648562300319e-06, "loss": 0.8329, "step": 180 }, { "epoch": 1.08, "learning_rate": 6.373801916932908e-06, "loss": 0.8191, "step": 184 }, { "epoch": 1.08, "learning_rate": 6.277955271565495e-06, "loss": 0.8659, "step": 188 }, { "epoch": 1.09, "learning_rate": 6.182108626198084e-06, "loss": 0.821, "step": 192 }, { "epoch": 1.1, "learning_rate": 6.086261980830671e-06, "loss": 0.8518, "step": 196 }, { "epoch": 1.11, "learning_rate": 5.990415335463259e-06, "loss": 0.7813, "step": 200 }, { "epoch": 1.12, "learning_rate": 5.894568690095847e-06, "loss": 0.7541, "step": 204 }, { "epoch": 1.13, "learning_rate": 5.798722044728435e-06, "loss": 0.7938, "step": 208 }, { "epoch": 1.14, "learning_rate": 5.702875399361023e-06, "loss": 0.8182, "step": 212 }, { "epoch": 1.15, "learning_rate": 5.607028753993611e-06, "loss": 0.7461, "step": 216 }, { "epoch": 1.16, "learning_rate": 5.5111821086261985e-06, "loss": 0.8029, "step": 220 }, { "epoch": 1.16, "eval_loss": 1.0393497943878174, "eval_runtime": 101.6759, "eval_samples_per_second": 3.246, "eval_steps_per_second": 0.413, "eval_wer": 178.42886795465432, "step": 220 }, { "epoch": 1.17, "learning_rate": 5.415335463258786e-06, "loss": 0.754, "step": 224 }, { "epoch": 1.17, "learning_rate": 5.319488817891374e-06, "loss": 0.7662, "step": 228 }, { "epoch": 1.18, "learning_rate": 5.223642172523962e-06, "loss": 0.7253, "step": 232 }, { "epoch": 1.19, "learning_rate": 5.127795527156549e-06, "loss": 0.6944, "step": 236 }, { "epoch": 1.2, "learning_rate": 5.031948881789138e-06, "loss": 0.7281, "step": 240 }, { "epoch": 1.21, "learning_rate": 4.936102236421725e-06, "loss": 0.7841, "step": 244 }, { "epoch": 1.22, "learning_rate": 4.840255591054313e-06, "loss": 0.7154, "step": 248 }, { "epoch": 1.23, "learning_rate": 4.744408945686901e-06, "loss": 0.7077, "step": 252 }, { "epoch": 1.24, "learning_rate": 4.648562300319489e-06, "loss": 0.7812, "step": 256 }, { "epoch": 1.25, "learning_rate": 4.552715654952077e-06, "loss": 0.6817, "step": 260 }, { "epoch": 1.25, "learning_rate": 4.456869009584665e-06, "loss": 0.6324, "step": 264 }, { "epoch": 1.25, "eval_loss": 1.0302220582962036, "eval_runtime": 114.7389, "eval_samples_per_second": 2.876, "eval_steps_per_second": 0.366, "eval_wer": 216.6054606418649, "step": 264 }, { "epoch": 1.26, "learning_rate": 4.361022364217253e-06, "loss": 0.6826, "step": 268 }, { "epoch": 1.27, "learning_rate": 4.26517571884984e-06, "loss": 0.7187, "step": 272 }, { "epoch": 1.28, "learning_rate": 4.169329073482428e-06, "loss": 0.6699, "step": 276 }, { "epoch": 1.29, "learning_rate": 4.0734824281150155e-06, "loss": 0.5677, "step": 280 }, { "epoch": 1.3, "learning_rate": 3.977635782747604e-06, "loss": 0.6833, "step": 284 }, { "epoch": 1.31, "learning_rate": 3.881789137380191e-06, "loss": 0.6293, "step": 288 }, { "epoch": 1.32, "learning_rate": 3.7859424920127796e-06, "loss": 0.672, "step": 292 }, { "epoch": 1.33, "learning_rate": 3.6900958466453675e-06, "loss": 0.7121, "step": 296 }, { "epoch": 1.33, "learning_rate": 3.5942492012779555e-06, "loss": 0.8766, "step": 300 }, { "epoch": 2.01, "learning_rate": 3.4984025559105434e-06, "loss": 0.6726, "step": 304 }, { "epoch": 2.02, "learning_rate": 3.4025559105431313e-06, "loss": 0.6971, "step": 308 }, { "epoch": 2.02, "eval_loss": 1.0134836435317993, "eval_runtime": 101.0358, "eval_samples_per_second": 3.266, "eval_steps_per_second": 0.416, "eval_wer": 179.3709085102986, "step": 308 }, { "epoch": 2.03, "learning_rate": 3.306709265175719e-06, "loss": 0.7432, "step": 312 }, { "epoch": 2.04, "learning_rate": 3.2108626198083067e-06, "loss": 0.6264, "step": 316 }, { "epoch": 2.04, "learning_rate": 3.1150159744408946e-06, "loss": 0.6604, "step": 320 }, { "epoch": 2.05, "learning_rate": 3.0191693290734825e-06, "loss": 0.6409, "step": 324 }, { "epoch": 2.06, "learning_rate": 2.9233226837060704e-06, "loss": 0.6539, "step": 328 }, { "epoch": 2.07, "learning_rate": 2.8274760383386583e-06, "loss": 0.611, "step": 332 }, { "epoch": 2.08, "learning_rate": 2.7316293929712462e-06, "loss": 0.6828, "step": 336 }, { "epoch": 2.09, "learning_rate": 2.635782747603834e-06, "loss": 0.6499, "step": 340 }, { "epoch": 2.1, "learning_rate": 2.539936102236422e-06, "loss": 0.6642, "step": 344 }, { "epoch": 2.11, "learning_rate": 2.44408945686901e-06, "loss": 0.6147, "step": 348 }, { "epoch": 2.12, "learning_rate": 2.3482428115015974e-06, "loss": 0.6051, "step": 352 }, { "epoch": 2.12, "eval_loss": 1.0064767599105835, "eval_runtime": 107.523, "eval_samples_per_second": 3.069, "eval_steps_per_second": 0.391, "eval_wer": 194.63515886955133, "step": 352 }, { "epoch": 2.12, "learning_rate": 2.2523961661341854e-06, "loss": 0.6237, "step": 356 }, { "epoch": 2.13, "learning_rate": 2.1565495207667733e-06, "loss": 0.5992, "step": 360 }, { "epoch": 2.14, "learning_rate": 2.060702875399361e-06, "loss": 0.6777, "step": 364 }, { "epoch": 2.15, "learning_rate": 1.964856230031949e-06, "loss": 0.6012, "step": 368 }, { "epoch": 2.16, "learning_rate": 1.8690095846645368e-06, "loss": 0.622, "step": 372 }, { "epoch": 2.17, "learning_rate": 1.7731629392971245e-06, "loss": 0.6288, "step": 376 }, { "epoch": 2.18, "learning_rate": 1.6773162939297124e-06, "loss": 0.6079, "step": 380 }, { "epoch": 2.19, "learning_rate": 1.5814696485623003e-06, "loss": 0.5887, "step": 384 }, { "epoch": 2.2, "learning_rate": 1.4856230031948882e-06, "loss": 0.5932, "step": 388 }, { "epoch": 2.21, "learning_rate": 1.389776357827476e-06, "loss": 0.6373, "step": 392 }, { "epoch": 2.21, "learning_rate": 1.2939297124600638e-06, "loss": 0.6048, "step": 396 }, { "epoch": 2.21, "eval_loss": 1.002966284751892, "eval_runtime": 97.8102, "eval_samples_per_second": 3.374, "eval_steps_per_second": 0.429, "eval_wer": 173.47916334025228, "step": 396 }, { "epoch": 2.22, "learning_rate": 1.1980830670926517e-06, "loss": 0.5774, "step": 400 }, { "epoch": 2.23, "learning_rate": 1.1022364217252397e-06, "loss": 0.6325, "step": 404 }, { "epoch": 2.24, "learning_rate": 1.0063897763578274e-06, "loss": 0.601, "step": 408 }, { "epoch": 2.25, "learning_rate": 9.105431309904153e-07, "loss": 0.5618, "step": 412 }, { "epoch": 2.26, "learning_rate": 8.146964856230032e-07, "loss": 0.5729, "step": 416 }, { "epoch": 2.27, "learning_rate": 7.188498402555911e-07, "loss": 0.5755, "step": 420 }, { "epoch": 2.28, "learning_rate": 6.230031948881789e-07, "loss": 0.6207, "step": 424 }, { "epoch": 2.29, "learning_rate": 5.271565495207668e-07, "loss": 0.5041, "step": 428 }, { "epoch": 2.29, "learning_rate": 4.313099041533546e-07, "loss": 0.5544, "step": 432 }, { "epoch": 2.3, "learning_rate": 3.3546325878594247e-07, "loss": 0.5608, "step": 436 }, { "epoch": 2.31, "learning_rate": 2.3961661341853033e-07, "loss": 0.585, "step": 440 }, { "epoch": 2.31, "eval_loss": 1.0049320459365845, "eval_runtime": 105.5439, "eval_samples_per_second": 3.127, "eval_steps_per_second": 0.398, "eval_wer": 186.6677311192719, "step": 440 }, { "epoch": 2.32, "learning_rate": 1.437699680511182e-07, "loss": 0.5486, "step": 444 }, { "epoch": 2.33, "learning_rate": 4.792332268370607e-08, "loss": 0.6387, "step": 448 }, { "epoch": 2.33, "step": 448, "total_flos": 1.7572960198656e+17, "train_loss": 0.8751795228038516, "train_runtime": 2133.32, "train_samples_per_second": 3.36, "train_steps_per_second": 0.21 } ], "max_steps": 448, "num_train_epochs": 9223372036854775807, "total_flos": 1.7572960198656e+17, "trial_name": null, "trial_params": null }