| { | |
| "best_metric": 18.684759916492695, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 476.0004, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.5e-08, | |
| "loss": 0.521, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5e-08, | |
| "loss": 0.51, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 7.5e-08, | |
| "loss": 0.4781, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 1e-07, | |
| "loss": 0.4499, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 1.25e-07, | |
| "loss": 0.4025, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 1.5e-07, | |
| "loss": 0.351, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 1.75e-07, | |
| "loss": 0.3063, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 2e-07, | |
| "loss": 0.2693, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 2.25e-07, | |
| "loss": 0.2478, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 2.5e-07, | |
| "loss": 0.2309, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 2.75e-07, | |
| "loss": 0.2101, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 3e-07, | |
| "loss": 0.2018, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 3.25e-07, | |
| "loss": 0.1795, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "learning_rate": 3.5e-07, | |
| "loss": 0.1726, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "learning_rate": 3.75e-07, | |
| "loss": 0.1548, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "learning_rate": 4e-07, | |
| "loss": 0.1459, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 4.2499999999999995e-07, | |
| "loss": 0.1307, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "learning_rate": 4.5e-07, | |
| "loss": 0.12, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "learning_rate": 4.7499999999999995e-07, | |
| "loss": 0.1076, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "learning_rate": 5e-07, | |
| "loss": 0.0974, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "learning_rate": 4.972222222222222e-07, | |
| "loss": 0.0871, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "learning_rate": 4.944444444444445e-07, | |
| "loss": 0.0773, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "learning_rate": 4.916666666666666e-07, | |
| "loss": 0.0695, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "learning_rate": 4.888888888888889e-07, | |
| "loss": 0.0625, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "learning_rate": 4.861111111111111e-07, | |
| "loss": 0.056, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "learning_rate": 4.833333333333333e-07, | |
| "loss": 0.0522, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "learning_rate": 4.805555555555555e-07, | |
| "loss": 0.0463, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "learning_rate": 4.777777777777778e-07, | |
| "loss": 0.0424, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "learning_rate": 4.7499999999999995e-07, | |
| "loss": 0.0395, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "learning_rate": 4.722222222222222e-07, | |
| "loss": 0.0363, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "learning_rate": 4.694444444444444e-07, | |
| "loss": 0.0328, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "learning_rate": 4.6666666666666666e-07, | |
| "loss": 0.0314, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "learning_rate": 4.6388888888888886e-07, | |
| "loss": 0.029, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 4.611111111111111e-07, | |
| "loss": 0.0271, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 83.0, | |
| "learning_rate": 4.5833333333333327e-07, | |
| "loss": 0.0252, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 85.0, | |
| "learning_rate": 4.555555555555555e-07, | |
| "loss": 0.0242, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 88.0, | |
| "learning_rate": 4.527777777777778e-07, | |
| "loss": 0.0221, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 90.0, | |
| "learning_rate": 4.5e-07, | |
| "loss": 0.0212, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 92.0, | |
| "learning_rate": 4.4722222222222223e-07, | |
| "loss": 0.0204, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "learning_rate": 4.444444444444444e-07, | |
| "loss": 0.0191, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "eval_loss": 0.2880958318710327, | |
| "eval_runtime": 289.7816, | |
| "eval_samples_per_second": 2.588, | |
| "eval_steps_per_second": 0.083, | |
| "eval_wer": 18.684759916492695, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 97.0, | |
| "learning_rate": 4.4166666666666664e-07, | |
| "loss": 0.0182, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 99.0, | |
| "learning_rate": 4.3888888888888884e-07, | |
| "loss": 0.0174, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 102.0, | |
| "learning_rate": 4.361111111111111e-07, | |
| "loss": 0.0165, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 104.0, | |
| "learning_rate": 4.3333333333333335e-07, | |
| "loss": 0.0159, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 107.0, | |
| "learning_rate": 4.3055555555555555e-07, | |
| "loss": 0.015, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 109.0, | |
| "learning_rate": 4.2777777777777775e-07, | |
| "loss": 0.0148, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 111.0, | |
| "learning_rate": 4.2499999999999995e-07, | |
| "loss": 0.0143, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 114.0, | |
| "learning_rate": 4.222222222222222e-07, | |
| "loss": 0.0136, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 116.0, | |
| "learning_rate": 4.194444444444444e-07, | |
| "loss": 0.0132, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 119.0, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "loss": 0.0127, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 121.0, | |
| "learning_rate": 4.1388888888888887e-07, | |
| "loss": 0.0124, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 123.0, | |
| "learning_rate": 4.1111111111111107e-07, | |
| "loss": 0.0118, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 126.0, | |
| "learning_rate": 4.083333333333333e-07, | |
| "loss": 0.0116, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 128.0, | |
| "learning_rate": 4.055555555555555e-07, | |
| "loss": 0.0111, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 130.0, | |
| "learning_rate": 4.027777777777778e-07, | |
| "loss": 0.0108, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 133.0, | |
| "learning_rate": 4e-07, | |
| "loss": 0.0104, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 135.0, | |
| "learning_rate": 3.972222222222222e-07, | |
| "loss": 0.0102, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 138.0, | |
| "learning_rate": 3.9444444444444444e-07, | |
| "loss": 0.0099, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 140.0, | |
| "learning_rate": 3.9166666666666664e-07, | |
| "loss": 0.0097, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 142.0, | |
| "learning_rate": 3.888888888888889e-07, | |
| "loss": 0.0094, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 145.0, | |
| "learning_rate": 3.861111111111111e-07, | |
| "loss": 0.0092, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 147.0, | |
| "learning_rate": 3.8333333333333335e-07, | |
| "loss": 0.0089, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 149.0, | |
| "learning_rate": 3.805555555555555e-07, | |
| "loss": 0.0087, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 152.0, | |
| "learning_rate": 3.7777777777777775e-07, | |
| "loss": 0.0085, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 154.0, | |
| "learning_rate": 3.75e-07, | |
| "loss": 0.0083, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 157.0, | |
| "learning_rate": 3.722222222222222e-07, | |
| "loss": 0.008, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 159.0, | |
| "learning_rate": 3.6944444444444447e-07, | |
| "loss": 0.0079, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 161.0, | |
| "learning_rate": 3.666666666666666e-07, | |
| "loss": 0.0077, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 164.0, | |
| "learning_rate": 3.6388888888888887e-07, | |
| "loss": 0.0076, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 166.0, | |
| "learning_rate": 3.6111111111111107e-07, | |
| "loss": 0.0074, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 169.0, | |
| "learning_rate": 3.583333333333333e-07, | |
| "loss": 0.0073, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 171.0, | |
| "learning_rate": 3.5555555555555553e-07, | |
| "loss": 0.0071, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 173.0, | |
| "learning_rate": 3.527777777777778e-07, | |
| "loss": 0.0069, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 176.0, | |
| "learning_rate": 3.5e-07, | |
| "loss": 0.0068, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 178.0, | |
| "learning_rate": 3.472222222222222e-07, | |
| "loss": 0.0067, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 180.0, | |
| "learning_rate": 3.4444444444444444e-07, | |
| "loss": 0.0065, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 183.0, | |
| "learning_rate": 3.4166666666666664e-07, | |
| "loss": 0.0065, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 185.0, | |
| "learning_rate": 3.388888888888889e-07, | |
| "loss": 0.0063, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 188.0, | |
| "learning_rate": 3.361111111111111e-07, | |
| "loss": 0.0062, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 190.0, | |
| "learning_rate": 3.333333333333333e-07, | |
| "loss": 0.0061, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 190.0, | |
| "eval_loss": 0.305788516998291, | |
| "eval_runtime": 275.9778, | |
| "eval_samples_per_second": 2.718, | |
| "eval_steps_per_second": 0.087, | |
| "eval_wer": 18.893528183716075, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 192.0, | |
| "learning_rate": 3.3055555555555556e-07, | |
| "loss": 0.006, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 195.0, | |
| "learning_rate": 3.2777777777777776e-07, | |
| "loss": 0.0059, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 197.0, | |
| "learning_rate": 3.25e-07, | |
| "loss": 0.0057, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 199.0, | |
| "learning_rate": 3.222222222222222e-07, | |
| "loss": 0.0057, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 202.0, | |
| "learning_rate": 3.194444444444444e-07, | |
| "loss": 0.0056, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 204.0, | |
| "learning_rate": 3.166666666666666e-07, | |
| "loss": 0.0055, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 207.0, | |
| "learning_rate": 3.1388888888888887e-07, | |
| "loss": 0.0054, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 209.0, | |
| "learning_rate": 3.111111111111111e-07, | |
| "loss": 0.0053, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 211.0, | |
| "learning_rate": 3.0833333333333333e-07, | |
| "loss": 0.0052, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 214.0, | |
| "learning_rate": 3.055555555555556e-07, | |
| "loss": 0.0052, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 216.0, | |
| "learning_rate": 3.0277777777777773e-07, | |
| "loss": 0.0051, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 219.0, | |
| "learning_rate": 3e-07, | |
| "loss": 0.005, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 221.0, | |
| "learning_rate": 2.972222222222222e-07, | |
| "loss": 0.0049, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 223.0, | |
| "learning_rate": 2.9444444444444444e-07, | |
| "loss": 0.0049, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 226.0, | |
| "learning_rate": 2.916666666666667e-07, | |
| "loss": 0.0048, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 228.0, | |
| "learning_rate": 2.8888888888888885e-07, | |
| "loss": 0.0047, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 230.0, | |
| "learning_rate": 2.861111111111111e-07, | |
| "loss": 0.0047, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 233.0, | |
| "learning_rate": 2.833333333333333e-07, | |
| "loss": 0.0046, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 235.0, | |
| "learning_rate": 2.8055555555555556e-07, | |
| "loss": 0.0045, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 238.0, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0045, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 240.0, | |
| "learning_rate": 2.75e-07, | |
| "loss": 0.0044, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 242.0, | |
| "learning_rate": 2.7222222222222216e-07, | |
| "loss": 0.0043, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 245.0, | |
| "learning_rate": 2.694444444444444e-07, | |
| "loss": 0.0043, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 247.0, | |
| "learning_rate": 2.6666666666666667e-07, | |
| "loss": 0.0042, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 249.0, | |
| "learning_rate": 2.638888888888889e-07, | |
| "loss": 0.0042, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 252.0, | |
| "learning_rate": 2.6111111111111113e-07, | |
| "loss": 0.0041, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 254.0, | |
| "learning_rate": 2.5833333333333333e-07, | |
| "loss": 0.0041, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 257.0, | |
| "learning_rate": 2.5555555555555553e-07, | |
| "loss": 0.004, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 259.0, | |
| "learning_rate": 2.5277777777777773e-07, | |
| "loss": 0.004, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 261.0, | |
| "learning_rate": 2.5e-07, | |
| "loss": 0.0039, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 264.0, | |
| "learning_rate": 2.4722222222222224e-07, | |
| "loss": 0.0039, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 266.0, | |
| "learning_rate": 2.4444444444444445e-07, | |
| "loss": 0.0038, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 269.0, | |
| "learning_rate": 2.4166666666666665e-07, | |
| "loss": 0.0038, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 271.0, | |
| "learning_rate": 2.388888888888889e-07, | |
| "loss": 0.0037, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 273.0, | |
| "learning_rate": 2.361111111111111e-07, | |
| "loss": 0.0037, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 276.0, | |
| "learning_rate": 2.3333333333333333e-07, | |
| "loss": 0.0037, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 278.0, | |
| "learning_rate": 2.3055555555555556e-07, | |
| "loss": 0.0036, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 280.0, | |
| "learning_rate": 2.2777777777777776e-07, | |
| "loss": 0.0036, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 283.0, | |
| "learning_rate": 2.25e-07, | |
| "loss": 0.0035, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 285.0, | |
| "learning_rate": 2.222222222222222e-07, | |
| "loss": 0.0035, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 285.0, | |
| "eval_loss": 0.317928284406662, | |
| "eval_runtime": 272.2274, | |
| "eval_samples_per_second": 2.755, | |
| "eval_steps_per_second": 0.088, | |
| "eval_wer": 19.227557411273484, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 288.0, | |
| "learning_rate": 2.1944444444444442e-07, | |
| "loss": 0.0035, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 290.0, | |
| "learning_rate": 2.1666666666666667e-07, | |
| "loss": 0.0034, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 292.0, | |
| "learning_rate": 2.1388888888888888e-07, | |
| "loss": 0.0034, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 295.0, | |
| "learning_rate": 2.111111111111111e-07, | |
| "loss": 0.0034, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 297.0, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "loss": 0.0033, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 299.0, | |
| "learning_rate": 2.0555555555555553e-07, | |
| "loss": 0.0033, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 302.0, | |
| "learning_rate": 2.0277777777777776e-07, | |
| "loss": 0.0033, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 304.0, | |
| "learning_rate": 2e-07, | |
| "loss": 0.0032, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 307.0, | |
| "learning_rate": 1.9722222222222222e-07, | |
| "loss": 0.0032, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 309.0, | |
| "learning_rate": 1.9444444444444445e-07, | |
| "loss": 0.0032, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 311.0, | |
| "learning_rate": 1.9166666666666668e-07, | |
| "loss": 0.0031, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 314.0, | |
| "learning_rate": 1.8888888888888888e-07, | |
| "loss": 0.0031, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 316.0, | |
| "learning_rate": 1.861111111111111e-07, | |
| "loss": 0.0031, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 319.0, | |
| "learning_rate": 1.833333333333333e-07, | |
| "loss": 0.003, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 321.0, | |
| "learning_rate": 1.8055555555555554e-07, | |
| "loss": 0.003, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 323.0, | |
| "learning_rate": 1.7777777777777776e-07, | |
| "loss": 0.003, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 326.0, | |
| "learning_rate": 1.75e-07, | |
| "loss": 0.003, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 328.0, | |
| "learning_rate": 1.7222222222222222e-07, | |
| "loss": 0.0029, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 330.0, | |
| "learning_rate": 1.6944444444444445e-07, | |
| "loss": 0.0029, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 333.0, | |
| "learning_rate": 1.6666666666666665e-07, | |
| "loss": 0.0029, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 335.0, | |
| "learning_rate": 1.6388888888888888e-07, | |
| "loss": 0.0029, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 338.0, | |
| "learning_rate": 1.611111111111111e-07, | |
| "loss": 0.0028, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 340.0, | |
| "learning_rate": 1.583333333333333e-07, | |
| "loss": 0.0028, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 342.0, | |
| "learning_rate": 1.5555555555555556e-07, | |
| "loss": 0.0028, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 345.0, | |
| "learning_rate": 1.527777777777778e-07, | |
| "loss": 0.0028, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 347.0, | |
| "learning_rate": 1.5e-07, | |
| "loss": 0.0028, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 349.0, | |
| "learning_rate": 1.4722222222222222e-07, | |
| "loss": 0.0027, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 352.0, | |
| "learning_rate": 1.4444444444444442e-07, | |
| "loss": 0.0027, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 354.0, | |
| "learning_rate": 1.4166666666666665e-07, | |
| "loss": 0.0027, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 357.0, | |
| "learning_rate": 1.3888888888888888e-07, | |
| "loss": 0.0027, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 359.0, | |
| "learning_rate": 1.3611111111111108e-07, | |
| "loss": 0.0027, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 361.0, | |
| "learning_rate": 1.3333333333333334e-07, | |
| "loss": 0.0026, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 364.0, | |
| "learning_rate": 1.3055555555555556e-07, | |
| "loss": 0.0026, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 366.0, | |
| "learning_rate": 1.2777777777777777e-07, | |
| "loss": 0.0026, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 369.0, | |
| "learning_rate": 1.25e-07, | |
| "loss": 0.0026, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 371.0, | |
| "learning_rate": 1.2222222222222222e-07, | |
| "loss": 0.0026, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 373.0, | |
| "learning_rate": 1.1944444444444445e-07, | |
| "loss": 0.0026, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 376.0, | |
| "learning_rate": 1.1666666666666667e-07, | |
| "loss": 0.0025, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 378.0, | |
| "learning_rate": 1.1388888888888888e-07, | |
| "loss": 0.0025, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 380.0, | |
| "learning_rate": 1.111111111111111e-07, | |
| "loss": 0.0025, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 380.0, | |
| "eval_loss": 0.3262755572795868, | |
| "eval_runtime": 274.332, | |
| "eval_samples_per_second": 2.734, | |
| "eval_steps_per_second": 0.087, | |
| "eval_wer": 19.45024356297843, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 383.0, | |
| "learning_rate": 1.0833333333333334e-07, | |
| "loss": 0.0025, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 385.0, | |
| "learning_rate": 1.0555555555555555e-07, | |
| "loss": 0.0025, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 388.0, | |
| "learning_rate": 1.0277777777777777e-07, | |
| "loss": 0.0025, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 390.0, | |
| "learning_rate": 1e-07, | |
| "loss": 0.0025, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 392.0, | |
| "learning_rate": 9.722222222222222e-08, | |
| "loss": 0.0024, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 395.0, | |
| "learning_rate": 9.444444444444444e-08, | |
| "loss": 0.0024, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 397.0, | |
| "learning_rate": 9.166666666666665e-08, | |
| "loss": 0.0024, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 399.0, | |
| "learning_rate": 8.888888888888888e-08, | |
| "loss": 0.0024, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 402.0, | |
| "learning_rate": 8.611111111111111e-08, | |
| "loss": 0.0024, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 404.0, | |
| "learning_rate": 8.333333333333333e-08, | |
| "loss": 0.0024, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 407.0, | |
| "learning_rate": 8.055555555555555e-08, | |
| "loss": 0.0023, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 409.0, | |
| "learning_rate": 7.777777777777778e-08, | |
| "loss": 0.0023, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 411.0, | |
| "learning_rate": 7.5e-08, | |
| "loss": 0.0023, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 414.0, | |
| "learning_rate": 7.222222222222221e-08, | |
| "loss": 0.0023, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 416.0, | |
| "learning_rate": 6.944444444444444e-08, | |
| "loss": 0.0023, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 419.0, | |
| "learning_rate": 6.666666666666667e-08, | |
| "loss": 0.0023, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 421.0, | |
| "learning_rate": 6.388888888888888e-08, | |
| "loss": 0.0023, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 423.0, | |
| "learning_rate": 6.111111111111111e-08, | |
| "loss": 0.0023, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 426.0, | |
| "learning_rate": 5.833333333333333e-08, | |
| "loss": 0.0023, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 428.0, | |
| "learning_rate": 5.555555555555555e-08, | |
| "loss": 0.0023, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 430.0, | |
| "learning_rate": 5.2777777777777776e-08, | |
| "loss": 0.0023, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 433.0, | |
| "learning_rate": 5e-08, | |
| "loss": 0.0022, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 435.0, | |
| "learning_rate": 4.722222222222222e-08, | |
| "loss": 0.0023, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 438.0, | |
| "learning_rate": 4.444444444444444e-08, | |
| "loss": 0.0022, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 440.0, | |
| "learning_rate": 4.166666666666666e-08, | |
| "loss": 0.0022, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 442.0, | |
| "learning_rate": 3.888888888888889e-08, | |
| "loss": 0.0022, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 445.0, | |
| "learning_rate": 3.6111111111111106e-08, | |
| "loss": 0.0022, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 447.0, | |
| "learning_rate": 3.3333333333333334e-08, | |
| "loss": 0.0022, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 449.0, | |
| "learning_rate": 3.0555555555555556e-08, | |
| "loss": 0.0022, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 452.0, | |
| "learning_rate": 2.7777777777777774e-08, | |
| "loss": 0.0022, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 454.0, | |
| "learning_rate": 2.5e-08, | |
| "loss": 0.0022, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 457.0, | |
| "learning_rate": 2.222222222222222e-08, | |
| "loss": 0.0022, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 459.0, | |
| "learning_rate": 1.9444444444444445e-08, | |
| "loss": 0.0022, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 461.0, | |
| "learning_rate": 1.6666666666666667e-08, | |
| "loss": 0.0022, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 464.0, | |
| "learning_rate": 1.3888888888888887e-08, | |
| "loss": 0.0022, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 466.0, | |
| "learning_rate": 1.111111111111111e-08, | |
| "loss": 0.0022, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 469.0, | |
| "learning_rate": 8.333333333333334e-09, | |
| "loss": 0.0022, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 471.0, | |
| "learning_rate": 5.555555555555555e-09, | |
| "loss": 0.0022, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 473.0, | |
| "learning_rate": 2.7777777777777776e-09, | |
| "loss": 0.0022, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 476.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0022, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 476.0, | |
| "eval_loss": 0.3296935260295868, | |
| "eval_runtime": 285.7041, | |
| "eval_samples_per_second": 2.625, | |
| "eval_steps_per_second": 0.084, | |
| "eval_wer": 19.491997216423105, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 476.0, | |
| "step": 5000, | |
| "total_flos": 8.918789513969664e+19, | |
| "train_loss": 0.03452818280979991, | |
| "train_runtime": 46791.6642, | |
| "train_samples_per_second": 6.839, | |
| "train_steps_per_second": 0.107 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "total_flos": 8.918789513969664e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |