| { | |
| "best_metric": 0.8883299798792756, | |
| "best_model_checkpoint": "PhoWhisper-small-vispeech-classifier-v4/checkpoint-490", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 490, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01020408163265306, | |
| "grad_norm": 51640.66015625, | |
| "learning_rate": 5.1020408163265303e-08, | |
| "loss": 2.0778, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02040816326530612, | |
| "grad_norm": 46773.35546875, | |
| "learning_rate": 1.0204081632653061e-07, | |
| "loss": 2.0809, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.030612244897959183, | |
| "grad_norm": 45343.12109375, | |
| "learning_rate": 1.5306122448979592e-07, | |
| "loss": 2.0771, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04081632653061224, | |
| "grad_norm": 49285.91796875, | |
| "learning_rate": 2.0408163265306121e-07, | |
| "loss": 2.077, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05102040816326531, | |
| "grad_norm": 63271.82421875, | |
| "learning_rate": 2.5510204081632656e-07, | |
| "loss": 2.0755, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.061224489795918366, | |
| "grad_norm": 51649.79296875, | |
| "learning_rate": 3.0612244897959183e-07, | |
| "loss": 2.0764, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07142857142857142, | |
| "grad_norm": 51633.98828125, | |
| "learning_rate": 3.5714285714285716e-07, | |
| "loss": 2.0763, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08163265306122448, | |
| "grad_norm": 61777.23828125, | |
| "learning_rate": 4.0816326530612243e-07, | |
| "loss": 2.0772, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09183673469387756, | |
| "grad_norm": 58467.15625, | |
| "learning_rate": 4.591836734693878e-07, | |
| "loss": 2.0743, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 48974.37890625, | |
| "learning_rate": 5.102040816326531e-07, | |
| "loss": 2.072, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11224489795918367, | |
| "grad_norm": 71007.515625, | |
| "learning_rate": 5.612244897959184e-07, | |
| "loss": 2.0751, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.12244897959183673, | |
| "grad_norm": 58237.77734375, | |
| "learning_rate": 6.122448979591837e-07, | |
| "loss": 2.0702, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1326530612244898, | |
| "grad_norm": 58785.83203125, | |
| "learning_rate": 6.632653061224491e-07, | |
| "loss": 2.0713, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14285714285714285, | |
| "grad_norm": 52771.265625, | |
| "learning_rate": 7.142857142857143e-07, | |
| "loss": 2.069, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15306122448979592, | |
| "grad_norm": 69647.078125, | |
| "learning_rate": 7.653061224489796e-07, | |
| "loss": 2.0684, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.16326530612244897, | |
| "grad_norm": 59148.47265625, | |
| "learning_rate": 8.163265306122449e-07, | |
| "loss": 2.0634, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.17346938775510204, | |
| "grad_norm": 66723.921875, | |
| "learning_rate": 8.673469387755103e-07, | |
| "loss": 2.064, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1836734693877551, | |
| "grad_norm": 72545.203125, | |
| "learning_rate": 9.183673469387756e-07, | |
| "loss": 2.0593, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.19387755102040816, | |
| "grad_norm": 59112.55859375, | |
| "learning_rate": 9.69387755102041e-07, | |
| "loss": 2.0553, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 68369.265625, | |
| "learning_rate": 1.0204081632653063e-06, | |
| "loss": 2.0566, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21428571428571427, | |
| "grad_norm": 51080.62890625, | |
| "learning_rate": 1.0714285714285714e-06, | |
| "loss": 2.0505, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.22448979591836735, | |
| "grad_norm": 78760.0859375, | |
| "learning_rate": 1.122448979591837e-06, | |
| "loss": 2.0496, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.23469387755102042, | |
| "grad_norm": 57198.7890625, | |
| "learning_rate": 1.1734693877551022e-06, | |
| "loss": 2.0458, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.24489795918367346, | |
| "grad_norm": 98369.4609375, | |
| "learning_rate": 1.2244897959183673e-06, | |
| "loss": 2.0336, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.25510204081632654, | |
| "grad_norm": 80801.5703125, | |
| "learning_rate": 1.2755102040816329e-06, | |
| "loss": 2.0354, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2653061224489796, | |
| "grad_norm": 82179.9765625, | |
| "learning_rate": 1.3265306122448982e-06, | |
| "loss": 2.0263, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2755102040816326, | |
| "grad_norm": 83225.078125, | |
| "learning_rate": 1.3775510204081633e-06, | |
| "loss": 2.0168, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2857142857142857, | |
| "grad_norm": 62451.64453125, | |
| "learning_rate": 1.4285714285714286e-06, | |
| "loss": 2.023, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.29591836734693877, | |
| "grad_norm": 75567.9453125, | |
| "learning_rate": 1.479591836734694e-06, | |
| "loss": 2.0206, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 98461.484375, | |
| "learning_rate": 1.5306122448979593e-06, | |
| "loss": 1.9984, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3163265306122449, | |
| "grad_norm": 75078.1484375, | |
| "learning_rate": 1.5816326530612248e-06, | |
| "loss": 1.9899, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.32653061224489793, | |
| "grad_norm": 97895.671875, | |
| "learning_rate": 1.6326530612244897e-06, | |
| "loss": 1.9758, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.336734693877551, | |
| "grad_norm": 65880.2421875, | |
| "learning_rate": 1.6836734693877552e-06, | |
| "loss": 1.9918, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3469387755102041, | |
| "grad_norm": 70675.578125, | |
| "learning_rate": 1.7346938775510206e-06, | |
| "loss": 1.9715, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.35714285714285715, | |
| "grad_norm": 64327.19921875, | |
| "learning_rate": 1.7857142857142859e-06, | |
| "loss": 1.9597, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3673469387755102, | |
| "grad_norm": 80610.6171875, | |
| "learning_rate": 1.8367346938775512e-06, | |
| "loss": 1.9632, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.37755102040816324, | |
| "grad_norm": 110899.375, | |
| "learning_rate": 1.8877551020408163e-06, | |
| "loss": 1.9476, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.3877551020408163, | |
| "grad_norm": 111132.5703125, | |
| "learning_rate": 1.938775510204082e-06, | |
| "loss": 1.9369, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3979591836734694, | |
| "grad_norm": 88722.625, | |
| "learning_rate": 1.989795918367347e-06, | |
| "loss": 1.9452, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 105874.296875, | |
| "learning_rate": 2.0408163265306125e-06, | |
| "loss": 1.9459, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.41836734693877553, | |
| "grad_norm": 97286.0078125, | |
| "learning_rate": 2.0918367346938776e-06, | |
| "loss": 1.918, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.42857142857142855, | |
| "grad_norm": 82848.0625, | |
| "learning_rate": 2.1428571428571427e-06, | |
| "loss": 1.9135, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4387755102040816, | |
| "grad_norm": 114854.4609375, | |
| "learning_rate": 2.1938775510204083e-06, | |
| "loss": 1.8869, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.4489795918367347, | |
| "grad_norm": 86661.25, | |
| "learning_rate": 2.244897959183674e-06, | |
| "loss": 1.8861, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.45918367346938777, | |
| "grad_norm": 127591.3984375, | |
| "learning_rate": 2.295918367346939e-06, | |
| "loss": 1.8396, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.46938775510204084, | |
| "grad_norm": 71038.8671875, | |
| "learning_rate": 2.3469387755102044e-06, | |
| "loss": 1.8995, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.47959183673469385, | |
| "grad_norm": 101402.421875, | |
| "learning_rate": 2.3979591836734696e-06, | |
| "loss": 1.8699, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.4897959183673469, | |
| "grad_norm": 113157.3828125, | |
| "learning_rate": 2.4489795918367347e-06, | |
| "loss": 1.8599, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 115485.0546875, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.8214, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 139369.8125, | |
| "learning_rate": 2.5510204081632657e-06, | |
| "loss": 1.8539, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5204081632653061, | |
| "grad_norm": 141401.46875, | |
| "learning_rate": 2.602040816326531e-06, | |
| "loss": 1.7507, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5306122448979592, | |
| "grad_norm": 125767.3984375, | |
| "learning_rate": 2.6530612244897964e-06, | |
| "loss": 1.7948, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5408163265306123, | |
| "grad_norm": 148848.5625, | |
| "learning_rate": 2.7040816326530615e-06, | |
| "loss": 1.7871, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5510204081632653, | |
| "grad_norm": 152509.4375, | |
| "learning_rate": 2.7551020408163266e-06, | |
| "loss": 1.8054, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5612244897959183, | |
| "grad_norm": 114681.078125, | |
| "learning_rate": 2.8061224489795917e-06, | |
| "loss": 1.7711, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 119104.984375, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 1.7139, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5816326530612245, | |
| "grad_norm": 123160.5, | |
| "learning_rate": 2.908163265306123e-06, | |
| "loss": 1.703, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.5918367346938775, | |
| "grad_norm": 157218.203125, | |
| "learning_rate": 2.959183673469388e-06, | |
| "loss": 1.7685, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6020408163265306, | |
| "grad_norm": 134804.828125, | |
| "learning_rate": 3.0102040816326534e-06, | |
| "loss": 1.7053, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 153977.625, | |
| "learning_rate": 3.0612244897959185e-06, | |
| "loss": 1.6711, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6224489795918368, | |
| "grad_norm": 204690.125, | |
| "learning_rate": 3.112244897959184e-06, | |
| "loss": 1.5873, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.6326530612244898, | |
| "grad_norm": 158604.1875, | |
| "learning_rate": 3.1632653061224496e-06, | |
| "loss": 1.6642, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6428571428571429, | |
| "grad_norm": 181877.484375, | |
| "learning_rate": 3.2142857142857147e-06, | |
| "loss": 1.6204, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6530612244897959, | |
| "grad_norm": 157735.734375, | |
| "learning_rate": 3.2653061224489794e-06, | |
| "loss": 1.6152, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6632653061224489, | |
| "grad_norm": 121935.3828125, | |
| "learning_rate": 3.316326530612245e-06, | |
| "loss": 1.6357, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.673469387755102, | |
| "grad_norm": 133579.859375, | |
| "learning_rate": 3.3673469387755105e-06, | |
| "loss": 1.6085, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.6836734693877551, | |
| "grad_norm": 154886.953125, | |
| "learning_rate": 3.4183673469387756e-06, | |
| "loss": 1.5609, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.6938775510204082, | |
| "grad_norm": 151655.0625, | |
| "learning_rate": 3.469387755102041e-06, | |
| "loss": 1.5905, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7040816326530612, | |
| "grad_norm": 198471.609375, | |
| "learning_rate": 3.5204081632653062e-06, | |
| "loss": 1.5052, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 191516.453125, | |
| "learning_rate": 3.5714285714285718e-06, | |
| "loss": 1.5798, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7244897959183674, | |
| "grad_norm": 207511.296875, | |
| "learning_rate": 3.6224489795918373e-06, | |
| "loss": 1.5382, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.7346938775510204, | |
| "grad_norm": 170820.609375, | |
| "learning_rate": 3.6734693877551024e-06, | |
| "loss": 1.485, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7448979591836735, | |
| "grad_norm": 183925.796875, | |
| "learning_rate": 3.724489795918368e-06, | |
| "loss": 1.427, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.7551020408163265, | |
| "grad_norm": 302829.75, | |
| "learning_rate": 3.7755102040816327e-06, | |
| "loss": 1.4242, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7653061224489796, | |
| "grad_norm": 292990.5625, | |
| "learning_rate": 3.826530612244898e-06, | |
| "loss": 1.484, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.7755102040816326, | |
| "grad_norm": 189517.421875, | |
| "learning_rate": 3.877551020408164e-06, | |
| "loss": 1.5039, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.7857142857142857, | |
| "grad_norm": 239271.890625, | |
| "learning_rate": 3.928571428571429e-06, | |
| "loss": 1.3332, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.7959183673469388, | |
| "grad_norm": 167386.28125, | |
| "learning_rate": 3.979591836734694e-06, | |
| "loss": 1.429, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8061224489795918, | |
| "grad_norm": 165701.9375, | |
| "learning_rate": 4.03061224489796e-06, | |
| "loss": 1.3167, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 284793.46875, | |
| "learning_rate": 4.081632653061225e-06, | |
| "loss": 1.3867, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.826530612244898, | |
| "grad_norm": 159807.671875, | |
| "learning_rate": 4.13265306122449e-06, | |
| "loss": 1.404, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.8367346938775511, | |
| "grad_norm": 225698.859375, | |
| "learning_rate": 4.183673469387755e-06, | |
| "loss": 1.3842, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8469387755102041, | |
| "grad_norm": 148945.90625, | |
| "learning_rate": 4.234693877551021e-06, | |
| "loss": 1.4605, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.8571428571428571, | |
| "grad_norm": 301350.15625, | |
| "learning_rate": 4.2857142857142855e-06, | |
| "loss": 1.4279, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8673469387755102, | |
| "grad_norm": 216519.640625, | |
| "learning_rate": 4.336734693877551e-06, | |
| "loss": 1.3122, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.8775510204081632, | |
| "grad_norm": 243937.1875, | |
| "learning_rate": 4.3877551020408165e-06, | |
| "loss": 1.2084, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.8877551020408163, | |
| "grad_norm": 195260.390625, | |
| "learning_rate": 4.438775510204082e-06, | |
| "loss": 1.3024, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.8979591836734694, | |
| "grad_norm": 218123.65625, | |
| "learning_rate": 4.489795918367348e-06, | |
| "loss": 1.2303, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9081632653061225, | |
| "grad_norm": 343990.03125, | |
| "learning_rate": 4.540816326530613e-06, | |
| "loss": 1.2012, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 157196.390625, | |
| "learning_rate": 4.591836734693878e-06, | |
| "loss": 1.2632, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9285714285714286, | |
| "grad_norm": 238103.46875, | |
| "learning_rate": 4.642857142857144e-06, | |
| "loss": 1.3131, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.9387755102040817, | |
| "grad_norm": 158442.625, | |
| "learning_rate": 4.693877551020409e-06, | |
| "loss": 1.2849, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.9489795918367347, | |
| "grad_norm": 178316.640625, | |
| "learning_rate": 4.744897959183674e-06, | |
| "loss": 1.0987, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.9591836734693877, | |
| "grad_norm": 240848.03125, | |
| "learning_rate": 4.795918367346939e-06, | |
| "loss": 1.154, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.9693877551020408, | |
| "grad_norm": 677730.8125, | |
| "learning_rate": 4.846938775510204e-06, | |
| "loss": 1.1682, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.9795918367346939, | |
| "grad_norm": 300622.875, | |
| "learning_rate": 4.897959183673469e-06, | |
| "loss": 1.1212, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.9897959183673469, | |
| "grad_norm": 234745.03125, | |
| "learning_rate": 4.948979591836735e-06, | |
| "loss": 1.1859, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 254966.984375, | |
| "learning_rate": 5e-06, | |
| "loss": 1.1763, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8883299798792756, | |
| "eval_loss": 0.968666136264801, | |
| "eval_runtime": 208.5533, | |
| "eval_samples_per_second": 14.299, | |
| "eval_steps_per_second": 0.599, | |
| "step": 490 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 4900, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.49679069336576e+18, | |
| "train_batch_size": 24, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |