| { | |
| "best_metric": 48.22644445885481, | |
| "best_model_checkpoint": "./output/small/yt-special-batch4-small/checkpoint-5000", | |
| "epoch": 0.4294425835265825, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 2.482, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 2.2377, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 1.9238, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 1.7416, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.5879, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.9e-06, | |
| "loss": 1.4014, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.105, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 1.0308, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.0176, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.9895, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 1.0954, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.8852, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.9041, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.8532, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.8955, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.8801, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.073, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.9524, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.0207, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.8515, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.955555555555556e-06, | |
| "loss": 0.8429, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.1065, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.844444444444446e-06, | |
| "loss": 1.0153, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.78888888888889e-06, | |
| "loss": 0.9275, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 0.8207, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 0.8609, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 1.0398, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 0.8857, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 1.0582, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 1.1059, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.9196, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.344444444444446e-06, | |
| "loss": 0.9251, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.28888888888889e-06, | |
| "loss": 0.7997, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.233333333333334e-06, | |
| "loss": 0.8166, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.17777777777778e-06, | |
| "loss": 1.0085, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.122222222222223e-06, | |
| "loss": 0.9693, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 0.8625, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.011111111111111e-06, | |
| "loss": 0.9412, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.955555555555555e-06, | |
| "loss": 0.8437, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 1.0296, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 0.9364269971847534, | |
| "eval_runtime": 602.1345, | |
| "eval_samples_per_second": 4.071, | |
| "eval_steps_per_second": 2.036, | |
| "eval_wer": 69.13300045392646, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.844444444444445e-06, | |
| "loss": 0.8542, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.788888888888891e-06, | |
| "loss": 0.9021, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 1.0371, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.677777777777779e-06, | |
| "loss": 0.8405, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.622222222222223e-06, | |
| "loss": 0.9133, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.566666666666667e-06, | |
| "loss": 0.8717, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.511111111111113e-06, | |
| "loss": 0.99, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.455555555555555e-06, | |
| "loss": 1.0554, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.9705, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.344444444444445e-06, | |
| "loss": 0.8173, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 0.9625, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 0.8858, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 0.9709, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 0.9056, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.891, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 0.7428, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 0.9, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.8744, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 0.9524, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 0.8547, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 0.9014, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 0.7328, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 0.8926, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 0.8251, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 0.8198, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 0.767, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.8833, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 0.7859, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.8662, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 0.8423, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 0.9215, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 0.8694, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 0.9849, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 0.7727, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 0.8345, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.7282, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 0.8742, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 0.7725, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 0.8694, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.8092, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 0.8502616286277771, | |
| "eval_runtime": 584.526, | |
| "eval_samples_per_second": 4.193, | |
| "eval_steps_per_second": 2.097, | |
| "eval_wer": 59.14013358407366, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.861, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 1.0144, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.8564, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 0.7732, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.922, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 0.955, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 0.7927, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.7831, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 0.8221, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 0.7984, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 0.963, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 1.0482, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 0.9075, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9e-06, | |
| "loss": 1.0663, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 0.8319, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 0.7351, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 0.8353, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 0.8339, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 0.8664, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 0.9061, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 0.7423, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 0.694, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.7212, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 0.9159, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 0.8811, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.7263, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 0.9562, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 0.6958, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 0.8412, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 0.7806, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 0.7606, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.7649, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 0.7376, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 1.0152, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.896, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.6582, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 0.8983, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.8131, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.6541, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 0.9109, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_loss": 0.8034260272979736, | |
| "eval_runtime": 567.7225, | |
| "eval_samples_per_second": 4.317, | |
| "eval_steps_per_second": 2.16, | |
| "eval_wer": 50.42474547694702, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.9081, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 0.959, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 0.7276, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 0.7866, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 0.8353, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 0.8532, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 0.8191, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 0.7893, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 0.6504, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.8162, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 0.8175, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 0.7899, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 0.786, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 0.8321, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 0.739, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 0.7737, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 0.8008, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 0.6954, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.8135, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 0.6532, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 0.8961, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 0.8073, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 0.7586, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 0.6529, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 0.7675, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 0.8188, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 0.8119, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.8591, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 0.7659, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 0.6989, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 0.778, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 0.7373, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 0.839, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 0.8074, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 0.8113, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 0.8354, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.8255, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 0.5901, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 0.8427, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 0.7291, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_loss": 0.7616028189659119, | |
| "eval_runtime": 559.7733, | |
| "eval_samples_per_second": 4.379, | |
| "eval_steps_per_second": 2.19, | |
| "eval_wer": 48.38207638933921, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 0.7638, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 0.751, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 0.7375, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 0.9296, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 0.7254, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.8582, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.6712, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.7964, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.714, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.689, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.6816, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.568888888888889e-06, | |
| "loss": 0.9035, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5133333333333334e-06, | |
| "loss": 0.7253, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.457777777777778e-06, | |
| "loss": 0.7956, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4022222222222223e-06, | |
| "loss": 0.7964, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3466666666666668e-06, | |
| "loss": 0.7326, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.2911111111111112e-06, | |
| "loss": 0.8501, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.2355555555555557e-06, | |
| "loss": 0.6447, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.1800000000000001e-06, | |
| "loss": 0.6771, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.1244444444444446e-06, | |
| "loss": 0.7192, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.068888888888889e-06, | |
| "loss": 0.7319, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0133333333333333e-06, | |
| "loss": 0.5776, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.57777777777778e-07, | |
| "loss": 0.8475, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.022222222222222e-07, | |
| "loss": 0.7619, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.466666666666668e-07, | |
| "loss": 0.7445, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.911111111111111e-07, | |
| "loss": 0.7901, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.355555555555556e-07, | |
| "loss": 0.7235, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.800000000000001e-07, | |
| "loss": 0.9231, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.244444444444445e-07, | |
| "loss": 0.8385, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.68888888888889e-07, | |
| "loss": 0.7913, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.133333333333334e-07, | |
| "loss": 0.6996, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.5777777777777784e-07, | |
| "loss": 0.7122, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.0222222222222224e-07, | |
| "loss": 0.7256, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.466666666666667e-07, | |
| "loss": 0.7686, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9111111111111116e-07, | |
| "loss": 0.8249, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.3555555555555556e-07, | |
| "loss": 0.6176, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8e-07, | |
| "loss": 0.7734, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2444444444444446e-07, | |
| "loss": 0.6663, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.888888888888889e-08, | |
| "loss": 0.7035, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.3333333333333334e-08, | |
| "loss": 0.7631, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 0.7389790415763855, | |
| "eval_runtime": 549.7598, | |
| "eval_samples_per_second": 4.458, | |
| "eval_steps_per_second": 2.23, | |
| "eval_wer": 48.22644445885481, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "step": 5000, | |
| "total_flos": 5.7717080064e+18, | |
| "train_loss": 0.8695706056594849, | |
| "train_runtime": 5861.515, | |
| "train_samples_per_second": 3.412, | |
| "train_steps_per_second": 0.853 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 1, | |
| "total_flos": 5.7717080064e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |