| { | |
| "best_metric": 23.795498749652683, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 526.3157894736842, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.4e-09, | |
| "loss": 0.8459, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 9.4e-09, | |
| "loss": 0.8681, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 1.4399999999999998e-08, | |
| "loss": 0.8349, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 1.94e-08, | |
| "loss": 0.8471, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "learning_rate": 2.4399999999999997e-08, | |
| "loss": 0.8299, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 2.94e-08, | |
| "loss": 0.8228, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 18.42, | |
| "learning_rate": 3.44e-08, | |
| "loss": 0.8015, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 21.05, | |
| "learning_rate": 3.94e-08, | |
| "loss": 0.8041, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 23.68, | |
| "learning_rate": 4.44e-08, | |
| "loss": 0.797, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 26.32, | |
| "learning_rate": 4.94e-08, | |
| "loss": 0.7771, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 28.95, | |
| "learning_rate": 5.4400000000000004e-08, | |
| "loss": 0.7623, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 31.58, | |
| "learning_rate": 5.9399999999999996e-08, | |
| "loss": 0.7524, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 34.21, | |
| "learning_rate": 6.44e-08, | |
| "loss": 0.7384, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 36.84, | |
| "learning_rate": 6.939999999999999e-08, | |
| "loss": 0.715, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 39.47, | |
| "learning_rate": 7.439999999999999e-08, | |
| "loss": 0.6994, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 42.11, | |
| "learning_rate": 7.94e-08, | |
| "loss": 0.6721, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 44.74, | |
| "learning_rate": 8.439999999999999e-08, | |
| "loss": 0.676, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 47.37, | |
| "learning_rate": 8.939999999999999e-08, | |
| "loss": 0.6614, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 9.439999999999999e-08, | |
| "loss": 0.625, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 52.63, | |
| "learning_rate": 9.939999999999999e-08, | |
| "loss": 0.5845, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 55.26, | |
| "learning_rate": 9.95111111111111e-08, | |
| "loss": 0.5787, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 57.89, | |
| "learning_rate": 9.895555555555555e-08, | |
| "loss": 0.5624, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 60.53, | |
| "learning_rate": 9.84e-08, | |
| "loss": 0.5237, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 63.16, | |
| "learning_rate": 9.784444444444444e-08, | |
| "loss": 0.5334, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 65.79, | |
| "learning_rate": 9.728888888888889e-08, | |
| "loss": 0.5064, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 68.42, | |
| "learning_rate": 9.673333333333334e-08, | |
| "loss": 0.5078, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 71.05, | |
| "learning_rate": 9.617777777777777e-08, | |
| "loss": 0.485, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 73.68, | |
| "learning_rate": 9.562222222222221e-08, | |
| "loss": 0.4678, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 76.32, | |
| "learning_rate": 9.506666666666666e-08, | |
| "loss": 0.4636, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 78.95, | |
| "learning_rate": 9.45111111111111e-08, | |
| "loss": 0.4707, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 81.58, | |
| "learning_rate": 9.395555555555555e-08, | |
| "loss": 0.4725, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 84.21, | |
| "learning_rate": 9.34e-08, | |
| "loss": 0.4595, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 86.84, | |
| "learning_rate": 9.284444444444444e-08, | |
| "loss": 0.4449, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 89.47, | |
| "learning_rate": 9.228888888888889e-08, | |
| "loss": 0.4254, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 92.11, | |
| "learning_rate": 9.173333333333333e-08, | |
| "loss": 0.4222, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 94.74, | |
| "learning_rate": 9.117777777777777e-08, | |
| "loss": 0.4247, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 97.37, | |
| "learning_rate": 9.062222222222222e-08, | |
| "loss": 0.4306, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "learning_rate": 9.006666666666665e-08, | |
| "loss": 0.4017, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 102.63, | |
| "learning_rate": 8.95111111111111e-08, | |
| "loss": 0.4176, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 105.26, | |
| "learning_rate": 8.895555555555555e-08, | |
| "loss": 0.3835, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 105.26, | |
| "eval_loss": 0.5891541838645935, | |
| "eval_runtime": 160.0944, | |
| "eval_samples_per_second": 4.223, | |
| "eval_steps_per_second": 0.137, | |
| "eval_wer": 25.423728813559322, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 107.89, | |
| "learning_rate": 8.84e-08, | |
| "loss": 0.4049, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 110.53, | |
| "learning_rate": 8.784444444444445e-08, | |
| "loss": 0.396, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 113.16, | |
| "learning_rate": 8.72888888888889e-08, | |
| "loss": 0.3819, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 115.79, | |
| "learning_rate": 8.673333333333333e-08, | |
| "loss": 0.3813, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 118.42, | |
| "learning_rate": 8.617777777777778e-08, | |
| "loss": 0.3786, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 121.05, | |
| "learning_rate": 8.562222222222221e-08, | |
| "loss": 0.3623, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 123.68, | |
| "learning_rate": 8.506666666666666e-08, | |
| "loss": 0.3702, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 126.32, | |
| "learning_rate": 8.451111111111111e-08, | |
| "loss": 0.3578, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 128.95, | |
| "learning_rate": 8.395555555555556e-08, | |
| "loss": 0.3559, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 131.58, | |
| "learning_rate": 8.339999999999999e-08, | |
| "loss": 0.3548, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 134.21, | |
| "learning_rate": 8.284444444444444e-08, | |
| "loss": 0.345, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 136.84, | |
| "learning_rate": 8.228888888888889e-08, | |
| "loss": 0.3574, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 139.47, | |
| "learning_rate": 8.173333333333334e-08, | |
| "loss": 0.3467, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 142.11, | |
| "learning_rate": 8.117777777777777e-08, | |
| "loss": 0.3511, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 144.74, | |
| "learning_rate": 8.062222222222221e-08, | |
| "loss": 0.3376, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 147.37, | |
| "learning_rate": 8.006666666666666e-08, | |
| "loss": 0.3369, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 150.0, | |
| "learning_rate": 7.95111111111111e-08, | |
| "loss": 0.3302, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 152.63, | |
| "learning_rate": 7.895555555555555e-08, | |
| "loss": 0.3339, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 155.26, | |
| "learning_rate": 7.84e-08, | |
| "loss": 0.33, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 157.89, | |
| "learning_rate": 7.784444444444445e-08, | |
| "loss": 0.3285, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 160.53, | |
| "learning_rate": 7.728888888888888e-08, | |
| "loss": 0.3236, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 163.16, | |
| "learning_rate": 7.673333333333333e-08, | |
| "loss": 0.3167, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 165.79, | |
| "learning_rate": 7.617777777777778e-08, | |
| "loss": 0.3117, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 168.42, | |
| "learning_rate": 7.562222222222222e-08, | |
| "loss": 0.3191, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 171.05, | |
| "learning_rate": 7.506666666666666e-08, | |
| "loss": 0.3058, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 173.68, | |
| "learning_rate": 7.45111111111111e-08, | |
| "loss": 0.3177, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 176.32, | |
| "learning_rate": 7.395555555555555e-08, | |
| "loss": 0.3017, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 178.95, | |
| "learning_rate": 7.34e-08, | |
| "loss": 0.3059, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 181.58, | |
| "learning_rate": 7.284444444444444e-08, | |
| "loss": 0.3124, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 184.21, | |
| "learning_rate": 7.228888888888889e-08, | |
| "loss": 0.2987, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 186.84, | |
| "learning_rate": 7.173333333333334e-08, | |
| "loss": 0.2981, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 189.47, | |
| "learning_rate": 7.117777777777778e-08, | |
| "loss": 0.3084, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 192.11, | |
| "learning_rate": 7.062222222222221e-08, | |
| "loss": 0.2858, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 194.74, | |
| "learning_rate": 7.006666666666666e-08, | |
| "loss": 0.2978, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 197.37, | |
| "learning_rate": 6.951111111111111e-08, | |
| "loss": 0.2818, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 200.0, | |
| "learning_rate": 6.895555555555556e-08, | |
| "loss": 0.2902, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 202.63, | |
| "learning_rate": 6.84e-08, | |
| "loss": 0.2838, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 205.26, | |
| "learning_rate": 6.784444444444444e-08, | |
| "loss": 0.2992, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 207.89, | |
| "learning_rate": 6.728888888888889e-08, | |
| "loss": 0.2774, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 210.53, | |
| "learning_rate": 6.673333333333334e-08, | |
| "loss": 0.2837, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 210.53, | |
| "eval_loss": 0.5525617003440857, | |
| "eval_runtime": 144.9276, | |
| "eval_samples_per_second": 4.664, | |
| "eval_steps_per_second": 0.152, | |
| "eval_wer": 23.895526535148655, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 213.16, | |
| "learning_rate": 6.617777777777777e-08, | |
| "loss": 0.2805, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 215.79, | |
| "learning_rate": 6.562222222222222e-08, | |
| "loss": 0.2759, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 218.42, | |
| "learning_rate": 6.506666666666665e-08, | |
| "loss": 0.2667, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 221.05, | |
| "learning_rate": 6.45111111111111e-08, | |
| "loss": 0.2777, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 223.68, | |
| "learning_rate": 6.395555555555555e-08, | |
| "loss": 0.2672, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 226.32, | |
| "learning_rate": 6.34e-08, | |
| "loss": 0.2739, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 228.95, | |
| "learning_rate": 6.284444444444445e-08, | |
| "loss": 0.2819, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 231.58, | |
| "learning_rate": 6.22888888888889e-08, | |
| "loss": 0.2646, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 234.21, | |
| "learning_rate": 6.173333333333333e-08, | |
| "loss": 0.2618, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 236.84, | |
| "learning_rate": 6.117777777777778e-08, | |
| "loss": 0.2711, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 239.47, | |
| "learning_rate": 6.062222222222221e-08, | |
| "loss": 0.2644, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 242.11, | |
| "learning_rate": 6.006666666666666e-08, | |
| "loss": 0.2494, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 244.74, | |
| "learning_rate": 5.951111111111111e-08, | |
| "loss": 0.2601, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 247.37, | |
| "learning_rate": 5.895555555555556e-08, | |
| "loss": 0.2538, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 250.0, | |
| "learning_rate": 5.8399999999999994e-08, | |
| "loss": 0.2551, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 252.63, | |
| "learning_rate": 5.784444444444444e-08, | |
| "loss": 0.2494, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 255.26, | |
| "learning_rate": 5.7288888888888884e-08, | |
| "loss": 0.2518, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 257.89, | |
| "learning_rate": 5.673333333333333e-08, | |
| "loss": 0.2455, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 260.53, | |
| "learning_rate": 5.617777777777778e-08, | |
| "loss": 0.2606, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 263.16, | |
| "learning_rate": 5.5622222222222215e-08, | |
| "loss": 0.2533, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 265.79, | |
| "learning_rate": 5.5066666666666663e-08, | |
| "loss": 0.2436, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 268.42, | |
| "learning_rate": 5.4511111111111105e-08, | |
| "loss": 0.2346, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 271.05, | |
| "learning_rate": 5.3955555555555553e-08, | |
| "loss": 0.2456, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 273.68, | |
| "learning_rate": 5.34e-08, | |
| "loss": 0.2462, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 276.32, | |
| "learning_rate": 5.284444444444445e-08, | |
| "loss": 0.24, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 278.95, | |
| "learning_rate": 5.2288888888888885e-08, | |
| "loss": 0.2466, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 281.58, | |
| "learning_rate": 5.173333333333333e-08, | |
| "loss": 0.24, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 284.21, | |
| "learning_rate": 5.1177777777777775e-08, | |
| "loss": 0.2419, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 286.84, | |
| "learning_rate": 5.0622222222222223e-08, | |
| "loss": 0.2366, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 289.47, | |
| "learning_rate": 5.0066666666666665e-08, | |
| "loss": 0.232, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 292.11, | |
| "learning_rate": 4.9511111111111107e-08, | |
| "loss": 0.235, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 294.74, | |
| "learning_rate": 4.8955555555555555e-08, | |
| "loss": 0.234, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 297.37, | |
| "learning_rate": 4.8399999999999997e-08, | |
| "loss": 0.2368, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 300.0, | |
| "learning_rate": 4.7844444444444445e-08, | |
| "loss": 0.2231, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 302.63, | |
| "learning_rate": 4.7288888888888887e-08, | |
| "loss": 0.224, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 305.26, | |
| "learning_rate": 4.673333333333333e-08, | |
| "loss": 0.2285, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 307.89, | |
| "learning_rate": 4.6177777777777777e-08, | |
| "loss": 0.2265, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 310.53, | |
| "learning_rate": 4.562222222222222e-08, | |
| "loss": 0.2238, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 313.16, | |
| "learning_rate": 4.5066666666666667e-08, | |
| "loss": 0.2224, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 315.79, | |
| "learning_rate": 4.451111111111111e-08, | |
| "loss": 0.2323, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 315.79, | |
| "eval_loss": 0.5431821942329407, | |
| "eval_runtime": 144.0249, | |
| "eval_samples_per_second": 4.694, | |
| "eval_steps_per_second": 0.153, | |
| "eval_wer": 24.012225618227284, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 318.42, | |
| "learning_rate": 4.395555555555555e-08, | |
| "loss": 0.2231, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 321.05, | |
| "learning_rate": 4.34e-08, | |
| "loss": 0.2132, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 323.68, | |
| "learning_rate": 4.2844444444444446e-08, | |
| "loss": 0.2275, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 326.32, | |
| "learning_rate": 4.228888888888889e-08, | |
| "loss": 0.2149, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 328.95, | |
| "learning_rate": 4.173333333333333e-08, | |
| "loss": 0.214, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 331.58, | |
| "learning_rate": 4.117777777777778e-08, | |
| "loss": 0.2181, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 334.21, | |
| "learning_rate": 4.062222222222222e-08, | |
| "loss": 0.2153, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 336.84, | |
| "learning_rate": 4.006666666666667e-08, | |
| "loss": 0.2266, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 339.47, | |
| "learning_rate": 3.951111111111111e-08, | |
| "loss": 0.2154, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 342.11, | |
| "learning_rate": 3.895555555555555e-08, | |
| "loss": 0.2088, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 344.74, | |
| "learning_rate": 3.84e-08, | |
| "loss": 0.2104, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 347.37, | |
| "learning_rate": 3.784444444444445e-08, | |
| "loss": 0.2115, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 350.0, | |
| "learning_rate": 3.728888888888888e-08, | |
| "loss": 0.2103, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 352.63, | |
| "learning_rate": 3.673333333333333e-08, | |
| "loss": 0.209, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 355.26, | |
| "learning_rate": 3.617777777777777e-08, | |
| "loss": 0.216, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 357.89, | |
| "learning_rate": 3.562222222222222e-08, | |
| "loss": 0.2125, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 360.53, | |
| "learning_rate": 3.506666666666667e-08, | |
| "loss": 0.2161, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 363.16, | |
| "learning_rate": 3.4511111111111105e-08, | |
| "loss": 0.2148, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 365.79, | |
| "learning_rate": 3.395555555555555e-08, | |
| "loss": 0.2093, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 368.42, | |
| "learning_rate": 3.34e-08, | |
| "loss": 0.215, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 371.05, | |
| "learning_rate": 3.284444444444444e-08, | |
| "loss": 0.2142, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 373.68, | |
| "learning_rate": 3.2288888888888885e-08, | |
| "loss": 0.2069, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 376.32, | |
| "learning_rate": 3.173333333333333e-08, | |
| "loss": 0.2067, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 378.95, | |
| "learning_rate": 3.1177777777777775e-08, | |
| "loss": 0.2084, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 381.58, | |
| "learning_rate": 3.062222222222222e-08, | |
| "loss": 0.209, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 384.21, | |
| "learning_rate": 3.006666666666667e-08, | |
| "loss": 0.2139, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 386.84, | |
| "learning_rate": 2.951111111111111e-08, | |
| "loss": 0.2019, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 389.47, | |
| "learning_rate": 2.8955555555555555e-08, | |
| "loss": 0.1983, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 392.11, | |
| "learning_rate": 2.8399999999999996e-08, | |
| "loss": 0.2043, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 394.74, | |
| "learning_rate": 2.784444444444444e-08, | |
| "loss": 0.2044, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 397.37, | |
| "learning_rate": 2.728888888888889e-08, | |
| "loss": 0.2025, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 400.0, | |
| "learning_rate": 2.673333333333333e-08, | |
| "loss": 0.2036, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 402.63, | |
| "learning_rate": 2.6177777777777776e-08, | |
| "loss": 0.1942, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 405.26, | |
| "learning_rate": 2.5622222222222224e-08, | |
| "loss": 0.1977, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 407.89, | |
| "learning_rate": 2.5066666666666663e-08, | |
| "loss": 0.2042, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 410.53, | |
| "learning_rate": 2.451111111111111e-08, | |
| "loss": 0.1933, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 413.16, | |
| "learning_rate": 2.3955555555555553e-08, | |
| "loss": 0.1988, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 415.79, | |
| "learning_rate": 2.34e-08, | |
| "loss": 0.1943, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 418.42, | |
| "learning_rate": 2.2844444444444443e-08, | |
| "loss": 0.2033, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 421.05, | |
| "learning_rate": 2.2288888888888888e-08, | |
| "loss": 0.1961, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 421.05, | |
| "eval_loss": 0.5402421951293945, | |
| "eval_runtime": 143.5894, | |
| "eval_samples_per_second": 4.708, | |
| "eval_steps_per_second": 0.153, | |
| "eval_wer": 23.795498749652683, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 423.68, | |
| "learning_rate": 2.1733333333333333e-08, | |
| "loss": 0.2009, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 426.32, | |
| "learning_rate": 2.1177777777777778e-08, | |
| "loss": 0.1961, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 428.95, | |
| "learning_rate": 2.0622222222222223e-08, | |
| "loss": 0.1987, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 431.58, | |
| "learning_rate": 2.0066666666666664e-08, | |
| "loss": 0.1951, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 434.21, | |
| "learning_rate": 1.9533333333333333e-08, | |
| "loss": 0.1914, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 436.84, | |
| "learning_rate": 1.8977777777777775e-08, | |
| "loss": 0.1955, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 439.47, | |
| "learning_rate": 1.8422222222222223e-08, | |
| "loss": 0.1967, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 442.11, | |
| "learning_rate": 1.7866666666666665e-08, | |
| "loss": 0.2011, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 444.74, | |
| "learning_rate": 1.731111111111111e-08, | |
| "loss": 0.1953, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 447.37, | |
| "learning_rate": 1.6755555555555554e-08, | |
| "loss": 0.1927, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 450.0, | |
| "learning_rate": 1.62e-08, | |
| "loss": 0.1918, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 452.63, | |
| "learning_rate": 1.5644444444444444e-08, | |
| "loss": 0.195, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 455.26, | |
| "learning_rate": 1.5088888888888886e-08, | |
| "loss": 0.1939, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 457.89, | |
| "learning_rate": 1.4533333333333333e-08, | |
| "loss": 0.1916, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 460.53, | |
| "learning_rate": 1.3977777777777778e-08, | |
| "loss": 0.1968, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 463.16, | |
| "learning_rate": 1.3422222222222221e-08, | |
| "loss": 0.193, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 465.79, | |
| "learning_rate": 1.2866666666666668e-08, | |
| "loss": 0.1888, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 468.42, | |
| "learning_rate": 1.2311111111111111e-08, | |
| "loss": 0.1858, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 471.05, | |
| "learning_rate": 1.1755555555555554e-08, | |
| "loss": 0.1872, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 473.68, | |
| "learning_rate": 1.12e-08, | |
| "loss": 0.1919, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 476.32, | |
| "learning_rate": 1.0644444444444443e-08, | |
| "loss": 0.1919, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 478.95, | |
| "learning_rate": 1.0088888888888888e-08, | |
| "loss": 0.1901, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 481.58, | |
| "learning_rate": 9.533333333333334e-09, | |
| "loss": 0.1947, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 484.21, | |
| "learning_rate": 8.977777777777778e-09, | |
| "loss": 0.189, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 486.84, | |
| "learning_rate": 8.422222222222223e-09, | |
| "loss": 0.1872, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 489.47, | |
| "learning_rate": 7.866666666666666e-09, | |
| "loss": 0.1848, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 492.11, | |
| "learning_rate": 7.311111111111111e-09, | |
| "loss": 0.1876, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 494.74, | |
| "learning_rate": 6.755555555555554e-09, | |
| "loss": 0.1911, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 497.37, | |
| "learning_rate": 6.199999999999999e-09, | |
| "loss": 0.1882, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 500.0, | |
| "learning_rate": 5.644444444444444e-09, | |
| "loss": 0.1872, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 502.63, | |
| "learning_rate": 5.088888888888888e-09, | |
| "loss": 0.1895, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 505.26, | |
| "learning_rate": 4.533333333333333e-09, | |
| "loss": 0.1895, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 507.89, | |
| "learning_rate": 3.9777777777777776e-09, | |
| "loss": 0.1834, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 510.53, | |
| "learning_rate": 3.422222222222222e-09, | |
| "loss": 0.1923, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 513.16, | |
| "learning_rate": 2.8666666666666667e-09, | |
| "loss": 0.1884, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 515.79, | |
| "learning_rate": 2.311111111111111e-09, | |
| "loss": 0.1834, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 518.42, | |
| "learning_rate": 1.7555555555555556e-09, | |
| "loss": 0.1964, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 521.05, | |
| "learning_rate": 1.2e-09, | |
| "loss": 0.1923, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 523.68, | |
| "learning_rate": 6.444444444444445e-10, | |
| "loss": 0.1866, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 526.32, | |
| "learning_rate": 8.888888888888889e-11, | |
| "loss": 0.1863, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 526.32, | |
| "eval_loss": 0.5395069122314453, | |
| "eval_runtime": 143.5193, | |
| "eval_samples_per_second": 4.71, | |
| "eval_steps_per_second": 0.153, | |
| "eval_wer": 23.795498749652683, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 526.32, | |
| "step": 5000, | |
| "total_flos": 1.97317080317952e+19, | |
| "train_loss": 0.31816294193267824, | |
| "train_runtime": 12909.7746, | |
| "train_samples_per_second": 24.787, | |
| "train_steps_per_second": 0.387 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 556, | |
| "save_steps": 1000, | |
| "total_flos": 1.97317080317952e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |