| { | |
| "best_metric": 15.59975139838409, | |
| "best_model_checkpoint": "./wpsmth/checkpoint-7000", | |
| "epoch": 4.147428571428572, | |
| "global_step": 7000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 0.6372, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 0.5862, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.5218, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.4872, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.4607, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.4075, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.3671, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.3425, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.3203, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.283, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.2712, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.2436, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.2213, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.2121, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.2346, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.1849, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.1719, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.1652, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.1479, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.1316, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.964615384615387e-06, | |
| "loss": 0.1134, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.926153846153847e-06, | |
| "loss": 0.1258, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.887692307692308e-06, | |
| "loss": 0.1184, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.84923076923077e-06, | |
| "loss": 0.1047, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.810769230769232e-06, | |
| "loss": 0.0924, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.772307692307694e-06, | |
| "loss": 0.0915, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.733846153846154e-06, | |
| "loss": 0.1113, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.695384615384617e-06, | |
| "loss": 0.11, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.656923076923079e-06, | |
| "loss": 0.0981, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.61846153846154e-06, | |
| "loss": 0.0604, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.58e-06, | |
| "loss": 0.0641, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.541538461538462e-06, | |
| "loss": 0.0575, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.503076923076924e-06, | |
| "loss": 0.0578, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.464615384615386e-06, | |
| "loss": 0.0516, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.426153846153847e-06, | |
| "loss": 0.0607, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.387692307692309e-06, | |
| "loss": 0.0809, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.349230769230771e-06, | |
| "loss": 0.0768, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.310769230769232e-06, | |
| "loss": 0.0724, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.272307692307692e-06, | |
| "loss": 0.0619, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.233846153846154e-06, | |
| "loss": 0.0473, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "eval_loss": 0.24591997265815735, | |
| "eval_runtime": 83.7474, | |
| "eval_samples_per_second": 2.388, | |
| "eval_steps_per_second": 0.084, | |
| "eval_wer": 21.6283405842138, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.195384615384616e-06, | |
| "loss": 0.0425, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.156923076923079e-06, | |
| "loss": 0.0399, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.118461538461539e-06, | |
| "loss": 0.0321, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.080000000000001e-06, | |
| "loss": 0.0513, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.041538461538463e-06, | |
| "loss": 0.0425, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.003076923076924e-06, | |
| "loss": 0.0374, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.964615384615384e-06, | |
| "loss": 0.0392, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.926153846153846e-06, | |
| "loss": 0.0378, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.887692307692309e-06, | |
| "loss": 0.0374, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.84923076923077e-06, | |
| "loss": 0.037, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.810769230769231e-06, | |
| "loss": 0.0392, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.772307692307693e-06, | |
| "loss": 0.0366, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.733846153846156e-06, | |
| "loss": 0.0326, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.695384615384616e-06, | |
| "loss": 0.0273, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.656923076923077e-06, | |
| "loss": 0.0337, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.618461538461539e-06, | |
| "loss": 0.0326, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.580000000000001e-06, | |
| "loss": 0.0301, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.541538461538463e-06, | |
| "loss": 0.031, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.503076923076924e-06, | |
| "loss": 0.0304, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.464615384615386e-06, | |
| "loss": 0.0349, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.426153846153848e-06, | |
| "loss": 0.0397, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.387692307692308e-06, | |
| "loss": 0.0598, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.349230769230769e-06, | |
| "loss": 0.045, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.310769230769231e-06, | |
| "loss": 0.048, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.272307692307693e-06, | |
| "loss": 0.0474, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.233846153846155e-06, | |
| "loss": 0.0359, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.195384615384616e-06, | |
| "loss": 0.0328, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.156923076923078e-06, | |
| "loss": 0.0405, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.11846153846154e-06, | |
| "loss": 0.0293, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.08e-06, | |
| "loss": 0.0332, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.041538461538461e-06, | |
| "loss": 0.0372, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.003076923076923e-06, | |
| "loss": 0.0279, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.964615384615385e-06, | |
| "loss": 0.0348, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.926153846153848e-06, | |
| "loss": 0.0347, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.887692307692308e-06, | |
| "loss": 0.0604, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.84923076923077e-06, | |
| "loss": 0.0334, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.810769230769232e-06, | |
| "loss": 0.03, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.772307692307693e-06, | |
| "loss": 0.0334, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.733846153846153e-06, | |
| "loss": 0.0306, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.695384615384615e-06, | |
| "loss": 0.0253, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "eval_loss": 0.1970207393169403, | |
| "eval_runtime": 85.4101, | |
| "eval_samples_per_second": 2.342, | |
| "eval_steps_per_second": 0.082, | |
| "eval_wer": 17.215661901802363, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.656923076923078e-06, | |
| "loss": 0.0206, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.618461538461539e-06, | |
| "loss": 0.0328, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.58e-06, | |
| "loss": 0.0291, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.5415384615384624e-06, | |
| "loss": 0.0236, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.503076923076924e-06, | |
| "loss": 0.0278, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.464615384615386e-06, | |
| "loss": 0.0298, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.426153846153846e-06, | |
| "loss": 0.0381, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.387692307692308e-06, | |
| "loss": 0.0415, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.34923076923077e-06, | |
| "loss": 0.0295, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.310769230769231e-06, | |
| "loss": 0.0209, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.272307692307693e-06, | |
| "loss": 0.0167, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.233846153846155e-06, | |
| "loss": 0.0206, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.195384615384616e-06, | |
| "loss": 0.0177, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.156923076923078e-06, | |
| "loss": 0.0194, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.118461538461539e-06, | |
| "loss": 0.0261, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.08e-06, | |
| "loss": 0.0363, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.041538461538462e-06, | |
| "loss": 0.0322, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.0030769230769235e-06, | |
| "loss": 0.034, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.964615384615386e-06, | |
| "loss": 0.023, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.926153846153847e-06, | |
| "loss": 0.0171, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.887692307692308e-06, | |
| "loss": 0.0167, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.8492307692307704e-06, | |
| "loss": 0.0132, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.810769230769231e-06, | |
| "loss": 0.0133, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.772307692307692e-06, | |
| "loss": 0.0209, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.733846153846154e-06, | |
| "loss": 0.0145, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.695384615384616e-06, | |
| "loss": 0.0131, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.656923076923078e-06, | |
| "loss": 0.0153, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.618461538461539e-06, | |
| "loss": 0.0158, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.5800000000000005e-06, | |
| "loss": 0.0177, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.541538461538463e-06, | |
| "loss": 0.0158, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.503076923076923e-06, | |
| "loss": 0.0157, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.4646153846153845e-06, | |
| "loss": 0.0173, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.426153846153847e-06, | |
| "loss": 0.0127, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.387692307692308e-06, | |
| "loss": 0.0128, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.34923076923077e-06, | |
| "loss": 0.016, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.3107692307692315e-06, | |
| "loss": 0.0148, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.272307692307693e-06, | |
| "loss": 0.0131, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.233846153846155e-06, | |
| "loss": 0.0141, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.195384615384616e-06, | |
| "loss": 0.0138, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.156923076923077e-06, | |
| "loss": 0.0181, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.20171760022640228, | |
| "eval_runtime": 87.4636, | |
| "eval_samples_per_second": 2.287, | |
| "eval_steps_per_second": 0.08, | |
| "eval_wer": 17.899316345556247, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.118461538461539e-06, | |
| "loss": 0.0268, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.08e-06, | |
| "loss": 0.0236, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.041538461538462e-06, | |
| "loss": 0.0209, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.003076923076924e-06, | |
| "loss": 0.027, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 5.964615384615385e-06, | |
| "loss": 0.0235, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 5.926153846153847e-06, | |
| "loss": 0.0167, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.8876923076923085e-06, | |
| "loss": 0.0157, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.849230769230769e-06, | |
| "loss": 0.0186, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.810769230769231e-06, | |
| "loss": 0.0115, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 5.7723076923076925e-06, | |
| "loss": 0.0192, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 5.733846153846155e-06, | |
| "loss": 0.0173, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.695384615384616e-06, | |
| "loss": 0.014, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.656923076923077e-06, | |
| "loss": 0.0166, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.6184615384615395e-06, | |
| "loss": 0.026, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.580000000000001e-06, | |
| "loss": 0.0302, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.541538461538461e-06, | |
| "loss": 0.0166, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.5030769230769234e-06, | |
| "loss": 0.012, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.464615384615385e-06, | |
| "loss": 0.0164, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.426153846153847e-06, | |
| "loss": 0.0143, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.387692307692308e-06, | |
| "loss": 0.0123, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.3492307692307696e-06, | |
| "loss": 0.0113, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.310769230769232e-06, | |
| "loss": 0.0148, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.272307692307693e-06, | |
| "loss": 0.0136, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.2338461538461535e-06, | |
| "loss": 0.0087, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.195384615384616e-06, | |
| "loss": 0.0139, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.156923076923077e-06, | |
| "loss": 0.0156, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.118461538461539e-06, | |
| "loss": 0.0193, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.0800000000000005e-06, | |
| "loss": 0.0209, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.041538461538462e-06, | |
| "loss": 0.0108, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.003076923076924e-06, | |
| "loss": 0.01, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.964615384615385e-06, | |
| "loss": 0.0063, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.926153846153847e-06, | |
| "loss": 0.01, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.887692307692308e-06, | |
| "loss": 0.0081, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.849230769230769e-06, | |
| "loss": 0.0097, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.8107692307692314e-06, | |
| "loss": 0.0158, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.772307692307693e-06, | |
| "loss": 0.0162, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.733846153846154e-06, | |
| "loss": 0.0165, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.695384615384615e-06, | |
| "loss": 0.0154, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.6569230769230776e-06, | |
| "loss": 0.0095, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.618461538461539e-06, | |
| "loss": 0.0088, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "eval_loss": 0.21484118700027466, | |
| "eval_runtime": 81.2702, | |
| "eval_samples_per_second": 2.461, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 16.842759477936607, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.58e-06, | |
| "loss": 0.0077, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.5415384615384615e-06, | |
| "loss": 0.0051, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.503076923076924e-06, | |
| "loss": 0.0079, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.464615384615385e-06, | |
| "loss": 0.0085, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.426153846153846e-06, | |
| "loss": 0.0057, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.387692307692308e-06, | |
| "loss": 0.0069, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.34923076923077e-06, | |
| "loss": 0.0068, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.310769230769231e-06, | |
| "loss": 0.008, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.2723076923076925e-06, | |
| "loss": 0.0081, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.233846153846154e-06, | |
| "loss": 0.0077, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.195384615384616e-06, | |
| "loss": 0.008, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.156923076923077e-06, | |
| "loss": 0.0079, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.118461538461539e-06, | |
| "loss": 0.0058, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.08e-06, | |
| "loss": 0.007, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.041538461538462e-06, | |
| "loss": 0.008, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.003076923076923e-06, | |
| "loss": 0.0075, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.964615384615385e-06, | |
| "loss": 0.007, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.926153846153846e-06, | |
| "loss": 0.0069, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.887692307692308e-06, | |
| "loss": 0.0072, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.8492307692307695e-06, | |
| "loss": 0.0093, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.8107692307692313e-06, | |
| "loss": 0.0154, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.772307692307693e-06, | |
| "loss": 0.0096, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.733846153846154e-06, | |
| "loss": 0.0116, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.6953846153846156e-06, | |
| "loss": 0.0117, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.6569230769230774e-06, | |
| "loss": 0.0088, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.618461538461539e-06, | |
| "loss": 0.0076, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.58e-06, | |
| "loss": 0.0101, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.5415384615384618e-06, | |
| "loss": 0.0071, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.5030769230769235e-06, | |
| "loss": 0.0053, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.4646153846153853e-06, | |
| "loss": 0.0095, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.426153846153846e-06, | |
| "loss": 0.0091, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.387692307692308e-06, | |
| "loss": 0.0071, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.3492307692307696e-06, | |
| "loss": 0.0087, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.3107692307692314e-06, | |
| "loss": 0.0167, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.2723076923076923e-06, | |
| "loss": 0.0109, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.233846153846154e-06, | |
| "loss": 0.008, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.1953846153846158e-06, | |
| "loss": 0.0061, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.1569230769230775e-06, | |
| "loss": 0.0083, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.1184615384615384e-06, | |
| "loss": 0.0067, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.08e-06, | |
| "loss": 0.0055, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "eval_loss": 0.2165648639202118, | |
| "eval_runtime": 84.4921, | |
| "eval_samples_per_second": 2.367, | |
| "eval_steps_per_second": 0.083, | |
| "eval_wer": 15.848353014294592, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.041538461538462e-06, | |
| "loss": 0.0064, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.0030769230769236e-06, | |
| "loss": 0.0052, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.9646153846153845e-06, | |
| "loss": 0.0052, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.9261538461538463e-06, | |
| "loss": 0.0041, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.887692307692308e-06, | |
| "loss": 0.0061, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.8492307692307698e-06, | |
| "loss": 0.0082, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.8107692307692307e-06, | |
| "loss": 0.0094, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.7723076923076924e-06, | |
| "loss": 0.0077, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.733846153846154e-06, | |
| "loss": 0.0043, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.695384615384616e-06, | |
| "loss": 0.0034, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.656923076923077e-06, | |
| "loss": 0.0034, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.6184615384615385e-06, | |
| "loss": 0.0044, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.5800000000000003e-06, | |
| "loss": 0.0034, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.541538461538462e-06, | |
| "loss": 0.0051, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.503076923076923e-06, | |
| "loss": 0.0085, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.4646153846153847e-06, | |
| "loss": 0.0071, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.4261538461538464e-06, | |
| "loss": 0.0082, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3876923076923077e-06, | |
| "loss": 0.007, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3492307692307695e-06, | |
| "loss": 0.0044, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.310769230769231e-06, | |
| "loss": 0.004, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.2723076923076925e-06, | |
| "loss": 0.0031, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.233846153846154e-06, | |
| "loss": 0.0022, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.1953846153846156e-06, | |
| "loss": 0.0041, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.156923076923077e-06, | |
| "loss": 0.0035, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.1184615384615387e-06, | |
| "loss": 0.0027, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.08e-06, | |
| "loss": 0.0032, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.0415384615384617e-06, | |
| "loss": 0.0043, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.003076923076923e-06, | |
| "loss": 0.004, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.964615384615385e-06, | |
| "loss": 0.0041, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.926153846153846e-06, | |
| "loss": 0.0038, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.8876923076923079e-06, | |
| "loss": 0.0043, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.8492307692307692e-06, | |
| "loss": 0.0035, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.810769230769231e-06, | |
| "loss": 0.0026, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.7723076923076922e-06, | |
| "loss": 0.0039, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.733846153846154e-06, | |
| "loss": 0.0043, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.6953846153846153e-06, | |
| "loss": 0.0033, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.656923076923077e-06, | |
| "loss": 0.0036, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.6184615384615384e-06, | |
| "loss": 0.0035, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.5800000000000001e-06, | |
| "loss": 0.0038, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.5415384615384614e-06, | |
| "loss": 0.0048, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.22611263394355774, | |
| "eval_runtime": 89.7451, | |
| "eval_samples_per_second": 2.229, | |
| "eval_steps_per_second": 0.078, | |
| "eval_wer": 16.034804226227468, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 1.5030769230769232e-06, | |
| "loss": 0.0065, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 1.4646153846153845e-06, | |
| "loss": 0.0046, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 1.4261538461538462e-06, | |
| "loss": 0.0058, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 1.3876923076923076e-06, | |
| "loss": 0.0063, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 1.3492307692307693e-06, | |
| "loss": 0.0037, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.3107692307692308e-06, | |
| "loss": 0.0035, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.2723076923076924e-06, | |
| "loss": 0.0053, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.233846153846154e-06, | |
| "loss": 0.0029, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.1953846153846154e-06, | |
| "loss": 0.0036, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.156923076923077e-06, | |
| "loss": 0.005, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.1184615384615385e-06, | |
| "loss": 0.0038, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.08e-06, | |
| "loss": 0.0037, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.0415384615384616e-06, | |
| "loss": 0.0044, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.003076923076923e-06, | |
| "loss": 0.0098, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 9.646153846153846e-07, | |
| "loss": 0.0046, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 9.261538461538462e-07, | |
| "loss": 0.0037, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 8.876923076923077e-07, | |
| "loss": 0.0035, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 8.492307692307692e-07, | |
| "loss": 0.0048, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 8.107692307692308e-07, | |
| "loss": 0.0032, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 7.723076923076923e-07, | |
| "loss": 0.0025, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 7.338461538461538e-07, | |
| "loss": 0.0032, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 6.953846153846154e-07, | |
| "loss": 0.0032, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 6.569230769230769e-07, | |
| "loss": 0.0016, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 6.184615384615385e-07, | |
| "loss": 0.0032, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 5.800000000000001e-07, | |
| "loss": 0.0032, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 5.415384615384616e-07, | |
| "loss": 0.0042, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 5.030769230769231e-07, | |
| "loss": 0.0048, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.6461538461538465e-07, | |
| "loss": 0.0035, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.261538461538462e-07, | |
| "loss": 0.0018, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.876923076923077e-07, | |
| "loss": 0.0017, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.4923076923076925e-07, | |
| "loss": 0.0024, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.107692307692308e-07, | |
| "loss": 0.0019, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 2.7230769230769237e-07, | |
| "loss": 0.0022, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 2.3384615384615387e-07, | |
| "loss": 0.0033, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 1.953846153846154e-07, | |
| "loss": 0.0045, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 1.5692307692307694e-07, | |
| "loss": 0.0044, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 1.1846153846153847e-07, | |
| "loss": 0.0053, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 8e-08, | |
| "loss": 0.0032, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 4.153846153846154e-08, | |
| "loss": 0.0031, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.0769230769230774e-09, | |
| "loss": 0.0026, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "eval_loss": 0.2262488156557083, | |
| "eval_runtime": 80.9315, | |
| "eval_samples_per_second": 2.471, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 15.59975139838409, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "step": 7000, | |
| "total_flos": 1.292804876353536e+20, | |
| "train_loss": 0.04174862339294382, | |
| "train_runtime": 55685.6569, | |
| "train_samples_per_second": 8.045, | |
| "train_steps_per_second": 0.126 | |
| } | |
| ], | |
| "max_steps": 7000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.292804876353536e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |