| { | |
| "best_metric": 41.233830845771145, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 434.0018, | |
| "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": 4.8e-08, | |
| "loss": 0.8504, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.8e-08, | |
| "loss": 0.8357, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 1.4799999999999998e-07, | |
| "loss": 0.822, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 1.98e-07, | |
| "loss": 0.7952, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 2.48e-07, | |
| "loss": 0.7591, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 2.98e-07, | |
| "loss": 0.7329, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 3.4799999999999994e-07, | |
| "loss": 0.698, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 3.98e-07, | |
| "loss": 0.6683, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 4.48e-07, | |
| "loss": 0.63, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 4.979999999999999e-07, | |
| "loss": 0.5954, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 5.48e-07, | |
| "loss": 0.5689, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 5.979999999999999e-07, | |
| "loss": 0.5462, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 6.48e-07, | |
| "loss": 0.5142, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 6.979999999999999e-07, | |
| "loss": 0.4942, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "learning_rate": 7.48e-07, | |
| "loss": 0.4715, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "learning_rate": 7.98e-07, | |
| "loss": 0.4464, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "learning_rate": 8.48e-07, | |
| "loss": 0.4342, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "learning_rate": 8.98e-07, | |
| "loss": 0.4111, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "learning_rate": 9.479999999999999e-07, | |
| "loss": 0.3904, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "learning_rate": 9.98e-07, | |
| "loss": 0.366, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "learning_rate": 9.946666666666666e-07, | |
| "loss": 0.3524, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "learning_rate": 9.891111111111112e-07, | |
| "loss": 0.3366, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "learning_rate": 9.835555555555554e-07, | |
| "loss": 0.3176, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "learning_rate": 9.78e-07, | |
| "loss": 0.3023, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "learning_rate": 9.724444444444444e-07, | |
| "loss": 0.2867, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "learning_rate": 9.668888888888888e-07, | |
| "loss": 0.2696, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "learning_rate": 9.613333333333334e-07, | |
| "loss": 0.2593, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 9.557777777777776e-07, | |
| "loss": 0.2462, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "learning_rate": 9.502222222222221e-07, | |
| "loss": 0.2344, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "learning_rate": 9.446666666666666e-07, | |
| "loss": 0.222, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "learning_rate": 9.39111111111111e-07, | |
| "loss": 0.2085, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "learning_rate": 9.335555555555555e-07, | |
| "loss": 0.2002, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "learning_rate": 9.28e-07, | |
| "loss": 0.1892, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "learning_rate": 9.224444444444443e-07, | |
| "loss": 0.1802, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "learning_rate": 9.168888888888889e-07, | |
| "loss": 0.1698, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "learning_rate": 9.113333333333333e-07, | |
| "loss": 0.1601, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 9.057777777777778e-07, | |
| "loss": 0.1531, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 82.0, | |
| "learning_rate": 9.002222222222222e-07, | |
| "loss": 0.1436, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 84.0, | |
| "learning_rate": 8.946666666666667e-07, | |
| "loss": 0.1375, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 86.0, | |
| "learning_rate": 8.89111111111111e-07, | |
| "loss": 0.1284, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 86.0, | |
| "eval_loss": 0.5030277371406555, | |
| "eval_runtime": 204.9632, | |
| "eval_samples_per_second": 4.25, | |
| "eval_steps_per_second": 0.137, | |
| "eval_wer": 41.233830845771145, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 89.0, | |
| "learning_rate": 8.835555555555555e-07, | |
| "loss": 0.1222, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 91.0, | |
| "learning_rate": 8.78e-07, | |
| "loss": 0.1155, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 93.0, | |
| "learning_rate": 8.724444444444444e-07, | |
| "loss": 0.1082, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "learning_rate": 8.668888888888889e-07, | |
| "loss": 0.1026, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 97.0, | |
| "learning_rate": 8.613333333333332e-07, | |
| "loss": 0.0954, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 99.0, | |
| "learning_rate": 8.557777777777777e-07, | |
| "loss": 0.0914, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 102.0, | |
| "learning_rate": 8.502222222222221e-07, | |
| "loss": 0.0857, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 104.0, | |
| "learning_rate": 8.446666666666666e-07, | |
| "loss": 0.0802, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 106.0, | |
| "learning_rate": 8.391111111111111e-07, | |
| "loss": 0.0756, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 108.0, | |
| "learning_rate": 8.335555555555555e-07, | |
| "loss": 0.071, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 110.0, | |
| "learning_rate": 8.28e-07, | |
| "loss": 0.0663, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 113.0, | |
| "learning_rate": 8.224444444444444e-07, | |
| "loss": 0.0629, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 115.0, | |
| "learning_rate": 8.168888888888889e-07, | |
| "loss": 0.0583, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 117.0, | |
| "learning_rate": 8.113333333333333e-07, | |
| "loss": 0.0559, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 119.0, | |
| "learning_rate": 8.057777777777778e-07, | |
| "loss": 0.0529, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 121.0, | |
| "learning_rate": 8.002222222222222e-07, | |
| "loss": 0.0483, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 123.0, | |
| "learning_rate": 7.946666666666666e-07, | |
| "loss": 0.0467, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 126.0, | |
| "learning_rate": 7.891111111111111e-07, | |
| "loss": 0.0439, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 128.0, | |
| "learning_rate": 7.835555555555555e-07, | |
| "loss": 0.0421, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 130.0, | |
| "learning_rate": 7.78e-07, | |
| "loss": 0.0391, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 132.0, | |
| "learning_rate": 7.724444444444444e-07, | |
| "loss": 0.0377, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 134.0, | |
| "learning_rate": 7.668888888888888e-07, | |
| "loss": 0.036, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 136.0, | |
| "learning_rate": 7.613333333333333e-07, | |
| "loss": 0.0344, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 139.0, | |
| "learning_rate": 7.557777777777777e-07, | |
| "loss": 0.0329, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 141.0, | |
| "learning_rate": 7.502222222222222e-07, | |
| "loss": 0.031, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 143.0, | |
| "learning_rate": 7.446666666666666e-07, | |
| "loss": 0.03, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 145.0, | |
| "learning_rate": 7.391111111111111e-07, | |
| "loss": 0.0285, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 147.0, | |
| "learning_rate": 7.335555555555554e-07, | |
| "loss": 0.0277, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 149.0, | |
| "learning_rate": 7.28e-07, | |
| "loss": 0.0264, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 152.0, | |
| "learning_rate": 7.224444444444445e-07, | |
| "loss": 0.0255, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 154.0, | |
| "learning_rate": 7.168888888888889e-07, | |
| "loss": 0.0243, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 156.0, | |
| "learning_rate": 7.113333333333334e-07, | |
| "loss": 0.0233, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 158.0, | |
| "learning_rate": 7.057777777777777e-07, | |
| "loss": 0.0229, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 160.0, | |
| "learning_rate": 7.002222222222222e-07, | |
| "loss": 0.0218, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 163.0, | |
| "learning_rate": 6.946666666666666e-07, | |
| "loss": 0.0211, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 165.0, | |
| "learning_rate": 6.891111111111111e-07, | |
| "loss": 0.0202, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 167.0, | |
| "learning_rate": 6.835555555555555e-07, | |
| "loss": 0.0196, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 169.0, | |
| "learning_rate": 6.78e-07, | |
| "loss": 0.0189, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 171.0, | |
| "learning_rate": 6.724444444444444e-07, | |
| "loss": 0.0186, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 173.0, | |
| "learning_rate": 6.668888888888888e-07, | |
| "loss": 0.0179, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 173.0, | |
| "eval_loss": 0.5767329931259155, | |
| "eval_runtime": 156.6136, | |
| "eval_samples_per_second": 5.561, | |
| "eval_steps_per_second": 0.179, | |
| "eval_wer": 41.61194029850746, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 176.0, | |
| "learning_rate": 6.613333333333333e-07, | |
| "loss": 0.0173, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 178.0, | |
| "learning_rate": 6.557777777777777e-07, | |
| "loss": 0.0169, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 180.0, | |
| "learning_rate": 6.502222222222222e-07, | |
| "loss": 0.0162, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 182.0, | |
| "learning_rate": 6.446666666666666e-07, | |
| "loss": 0.0158, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 184.0, | |
| "learning_rate": 6.39111111111111e-07, | |
| "loss": 0.0155, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 186.0, | |
| "learning_rate": 6.335555555555556e-07, | |
| "loss": 0.0149, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 189.0, | |
| "learning_rate": 6.28e-07, | |
| "loss": 0.0146, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 191.0, | |
| "learning_rate": 6.224444444444445e-07, | |
| "loss": 0.0143, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 193.0, | |
| "learning_rate": 6.168888888888889e-07, | |
| "loss": 0.0137, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 195.0, | |
| "learning_rate": 6.113333333333333e-07, | |
| "loss": 0.0136, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 197.0, | |
| "learning_rate": 6.057777777777777e-07, | |
| "loss": 0.013, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 199.0, | |
| "learning_rate": 6.002222222222222e-07, | |
| "loss": 0.0128, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 202.0, | |
| "learning_rate": 5.946666666666667e-07, | |
| "loss": 0.0125, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 204.0, | |
| "learning_rate": 5.891111111111111e-07, | |
| "loss": 0.0122, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 206.0, | |
| "learning_rate": 5.835555555555556e-07, | |
| "loss": 0.012, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 208.0, | |
| "learning_rate": 5.779999999999999e-07, | |
| "loss": 0.0115, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 210.0, | |
| "learning_rate": 5.724444444444444e-07, | |
| "loss": 0.0114, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 213.0, | |
| "learning_rate": 5.668888888888888e-07, | |
| "loss": 0.0111, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 215.0, | |
| "learning_rate": 5.613333333333333e-07, | |
| "loss": 0.011, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 217.0, | |
| "learning_rate": 5.557777777777778e-07, | |
| "loss": 0.0106, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 219.0, | |
| "learning_rate": 5.502222222222221e-07, | |
| "loss": 0.0104, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 221.0, | |
| "learning_rate": 5.446666666666666e-07, | |
| "loss": 0.0102, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 223.0, | |
| "learning_rate": 5.39111111111111e-07, | |
| "loss": 0.01, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 226.0, | |
| "learning_rate": 5.335555555555556e-07, | |
| "loss": 0.0098, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 228.0, | |
| "learning_rate": 5.28e-07, | |
| "loss": 0.0096, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 230.0, | |
| "learning_rate": 5.224444444444445e-07, | |
| "loss": 0.0094, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 232.0, | |
| "learning_rate": 5.168888888888888e-07, | |
| "loss": 0.0092, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 234.0, | |
| "learning_rate": 5.113333333333333e-07, | |
| "loss": 0.0091, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 236.0, | |
| "learning_rate": 5.057777777777778e-07, | |
| "loss": 0.009, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 239.0, | |
| "learning_rate": 5.002222222222222e-07, | |
| "loss": 0.0088, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 241.0, | |
| "learning_rate": 4.946666666666666e-07, | |
| "loss": 0.0086, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 243.0, | |
| "learning_rate": 4.891111111111111e-07, | |
| "loss": 0.0084, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 245.0, | |
| "learning_rate": 4.835555555555555e-07, | |
| "loss": 0.0084, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 247.0, | |
| "learning_rate": 4.779999999999999e-07, | |
| "loss": 0.0082, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 249.0, | |
| "learning_rate": 4.724444444444444e-07, | |
| "loss": 0.008, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 252.0, | |
| "learning_rate": 4.668888888888889e-07, | |
| "loss": 0.0079, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 254.0, | |
| "learning_rate": 4.613333333333333e-07, | |
| "loss": 0.0078, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 256.0, | |
| "learning_rate": 4.5577777777777774e-07, | |
| "loss": 0.0076, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 258.0, | |
| "learning_rate": 4.5022222222222225e-07, | |
| "loss": 0.0076, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 260.0, | |
| "learning_rate": 4.4466666666666665e-07, | |
| "loss": 0.0074, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 260.0, | |
| "eval_loss": 0.617143988609314, | |
| "eval_runtime": 152.8225, | |
| "eval_samples_per_second": 5.699, | |
| "eval_steps_per_second": 0.183, | |
| "eval_wer": 41.44776119402985, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 263.0, | |
| "learning_rate": 4.391111111111111e-07, | |
| "loss": 0.0073, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 265.0, | |
| "learning_rate": 4.3355555555555556e-07, | |
| "loss": 0.0072, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 267.0, | |
| "learning_rate": 4.2799999999999997e-07, | |
| "loss": 0.0071, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 269.0, | |
| "learning_rate": 4.224444444444444e-07, | |
| "loss": 0.007, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 271.0, | |
| "learning_rate": 4.1688888888888883e-07, | |
| "loss": 0.0069, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 273.0, | |
| "learning_rate": 4.113333333333333e-07, | |
| "loss": 0.0068, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 276.0, | |
| "learning_rate": 4.057777777777778e-07, | |
| "loss": 0.0067, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 278.0, | |
| "learning_rate": 4.002222222222222e-07, | |
| "loss": 0.0066, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 280.0, | |
| "learning_rate": 3.9466666666666665e-07, | |
| "loss": 0.0065, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 282.0, | |
| "learning_rate": 3.891111111111111e-07, | |
| "loss": 0.0064, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 284.0, | |
| "learning_rate": 3.835555555555555e-07, | |
| "loss": 0.0063, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 286.0, | |
| "learning_rate": 3.7799999999999997e-07, | |
| "loss": 0.0063, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 289.0, | |
| "learning_rate": 3.7244444444444443e-07, | |
| "loss": 0.0062, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 291.0, | |
| "learning_rate": 3.668888888888889e-07, | |
| "loss": 0.0061, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 293.0, | |
| "learning_rate": 3.6133333333333334e-07, | |
| "loss": 0.006, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 295.0, | |
| "learning_rate": 3.557777777777778e-07, | |
| "loss": 0.006, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 297.0, | |
| "learning_rate": 3.502222222222222e-07, | |
| "loss": 0.0058, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 299.0, | |
| "learning_rate": 3.4466666666666666e-07, | |
| "loss": 0.0058, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 302.0, | |
| "learning_rate": 3.3911111111111106e-07, | |
| "loss": 0.0057, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 304.0, | |
| "learning_rate": 3.335555555555555e-07, | |
| "loss": 0.0057, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 306.0, | |
| "learning_rate": 3.28e-07, | |
| "loss": 0.0056, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 308.0, | |
| "learning_rate": 3.2244444444444443e-07, | |
| "loss": 0.0056, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 310.0, | |
| "learning_rate": 3.168888888888889e-07, | |
| "loss": 0.0055, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 313.0, | |
| "learning_rate": 3.1133333333333334e-07, | |
| "loss": 0.0054, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 315.0, | |
| "learning_rate": 3.0577777777777775e-07, | |
| "loss": 0.0054, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 317.0, | |
| "learning_rate": 3.002222222222222e-07, | |
| "loss": 0.0053, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 319.0, | |
| "learning_rate": 2.9466666666666666e-07, | |
| "loss": 0.0053, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 321.0, | |
| "learning_rate": 2.8911111111111106e-07, | |
| "loss": 0.0052, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 323.0, | |
| "learning_rate": 2.8355555555555557e-07, | |
| "loss": 0.0051, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 326.0, | |
| "learning_rate": 2.7800000000000003e-07, | |
| "loss": 0.0051, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 328.0, | |
| "learning_rate": 2.7244444444444443e-07, | |
| "loss": 0.005, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 330.0, | |
| "learning_rate": 2.668888888888889e-07, | |
| "loss": 0.005, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 332.0, | |
| "learning_rate": 2.613333333333333e-07, | |
| "loss": 0.005, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 334.0, | |
| "learning_rate": 2.5577777777777775e-07, | |
| "loss": 0.0049, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 336.0, | |
| "learning_rate": 2.502222222222222e-07, | |
| "loss": 0.0049, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 339.0, | |
| "learning_rate": 2.4466666666666666e-07, | |
| "loss": 0.0048, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 341.0, | |
| "learning_rate": 2.391111111111111e-07, | |
| "loss": 0.0048, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 343.0, | |
| "learning_rate": 2.3355555555555552e-07, | |
| "loss": 0.0047, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 345.0, | |
| "learning_rate": 2.28e-07, | |
| "loss": 0.0047, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 347.0, | |
| "learning_rate": 2.2244444444444443e-07, | |
| "loss": 0.0047, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 347.0, | |
| "eval_loss": 0.6419367790222168, | |
| "eval_runtime": 210.1791, | |
| "eval_samples_per_second": 4.144, | |
| "eval_steps_per_second": 0.133, | |
| "eval_wer": 42.179104477611936, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 349.0, | |
| "learning_rate": 2.1688888888888886e-07, | |
| "loss": 0.0046, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 352.0, | |
| "learning_rate": 2.1133333333333335e-07, | |
| "loss": 0.0046, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 354.0, | |
| "learning_rate": 2.0577777777777778e-07, | |
| "loss": 0.0045, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 356.0, | |
| "learning_rate": 2.002222222222222e-07, | |
| "loss": 0.0045, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 358.0, | |
| "learning_rate": 1.9466666666666664e-07, | |
| "loss": 0.0045, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 360.0, | |
| "learning_rate": 1.8911111111111112e-07, | |
| "loss": 0.0044, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 363.0, | |
| "learning_rate": 1.8355555555555555e-07, | |
| "loss": 0.0044, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 365.0, | |
| "learning_rate": 1.7799999999999998e-07, | |
| "loss": 0.0044, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 367.0, | |
| "learning_rate": 1.7244444444444444e-07, | |
| "loss": 0.0044, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 369.0, | |
| "learning_rate": 1.668888888888889e-07, | |
| "loss": 0.0043, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 371.0, | |
| "learning_rate": 1.6133333333333332e-07, | |
| "loss": 0.0043, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 373.0, | |
| "learning_rate": 1.5577777777777775e-07, | |
| "loss": 0.0043, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 376.0, | |
| "learning_rate": 1.5022222222222224e-07, | |
| "loss": 0.0043, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 378.0, | |
| "learning_rate": 1.4466666666666667e-07, | |
| "loss": 0.0042, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 380.0, | |
| "learning_rate": 1.391111111111111e-07, | |
| "loss": 0.0042, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 382.0, | |
| "learning_rate": 1.3355555555555555e-07, | |
| "loss": 0.0042, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 384.0, | |
| "learning_rate": 1.28e-07, | |
| "loss": 0.0042, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 386.0, | |
| "learning_rate": 1.2244444444444444e-07, | |
| "loss": 0.0041, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 389.0, | |
| "learning_rate": 1.1688888888888888e-07, | |
| "loss": 0.0041, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 391.0, | |
| "learning_rate": 1.1133333333333334e-07, | |
| "loss": 0.0041, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 393.0, | |
| "learning_rate": 1.0577777777777777e-07, | |
| "loss": 0.0041, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 395.0, | |
| "learning_rate": 1.0022222222222222e-07, | |
| "loss": 0.0041, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 397.0, | |
| "learning_rate": 9.466666666666665e-08, | |
| "loss": 0.004, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 399.0, | |
| "learning_rate": 8.911111111111111e-08, | |
| "loss": 0.004, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 402.0, | |
| "learning_rate": 8.355555555555554e-08, | |
| "loss": 0.004, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 404.0, | |
| "learning_rate": 7.8e-08, | |
| "loss": 0.004, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 406.0, | |
| "learning_rate": 7.244444444444445e-08, | |
| "loss": 0.004, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 408.0, | |
| "learning_rate": 6.688888888888888e-08, | |
| "loss": 0.004, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 410.0, | |
| "learning_rate": 6.133333333333333e-08, | |
| "loss": 0.004, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 413.0, | |
| "learning_rate": 5.5777777777777777e-08, | |
| "loss": 0.0039, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 415.0, | |
| "learning_rate": 5.022222222222222e-08, | |
| "loss": 0.004, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 417.0, | |
| "learning_rate": 4.466666666666666e-08, | |
| "loss": 0.0039, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 419.0, | |
| "learning_rate": 3.9111111111111106e-08, | |
| "loss": 0.0039, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 421.0, | |
| "learning_rate": 3.355555555555555e-08, | |
| "loss": 0.0039, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 423.0, | |
| "learning_rate": 2.8e-08, | |
| "loss": 0.0039, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 426.0, | |
| "learning_rate": 2.2444444444444442e-08, | |
| "loss": 0.0039, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 428.0, | |
| "learning_rate": 1.6888888888888886e-08, | |
| "loss": 0.0039, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 430.0, | |
| "learning_rate": 1.1333333333333334e-08, | |
| "loss": 0.0039, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 432.0, | |
| "learning_rate": 5.777777777777777e-09, | |
| "loss": 0.0039, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 434.0, | |
| "learning_rate": 2.2222222222222221e-10, | |
| "loss": 0.0039, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 434.0, | |
| "eval_loss": 0.6517490148544312, | |
| "eval_runtime": 158.9123, | |
| "eval_samples_per_second": 5.481, | |
| "eval_steps_per_second": 0.176, | |
| "eval_wer": 41.97512437810945, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 434.0, | |
| "step": 5000, | |
| "total_flos": 2.044555714953216e+19, | |
| "train_loss": 0.096768384963274, | |
| "train_runtime": 21588.374, | |
| "train_samples_per_second": 14.823, | |
| "train_steps_per_second": 0.232 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "total_flos": 2.044555714953216e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |