| { | |
| "best_metric": 23.14374107567825, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 71.42857142857143, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.898977360288234e-06, | |
| "loss": 0.8611, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.160712527409633e-06, | |
| "loss": 0.3224, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.85912902234906e-06, | |
| "loss": 0.2375, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.344547104469332e-06, | |
| "loss": 0.1304, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.716963756434345e-06, | |
| "loss": 0.1404, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 8.019180844200955e-06, | |
| "loss": 0.1005, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.27351214279797e-06, | |
| "loss": 0.0496, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 8.49307723936858e-06, | |
| "loss": 0.0468, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 8.686247975778677e-06, | |
| "loss": 0.0305, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.858694625217149e-06, | |
| "loss": 0.0245, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 9.014436199608479e-06, | |
| "loss": 0.0204, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 9.156425255148058e-06, | |
| "loss": 0.0161, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 9.28689473531776e-06, | |
| "loss": 0.0129, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 9.407574351377137e-06, | |
| "loss": 0.013, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 9.519831289296397e-06, | |
| "loss": 0.0084, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 9.624764935335318e-06, | |
| "loss": 0.0063, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 9.723272550712454e-06, | |
| "loss": 0.0091, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 9.816095971633122e-06, | |
| "loss": 0.008, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 9.90385555539545e-06, | |
| "loss": 0.0082, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 9.987075336738768e-06, | |
| "loss": 0.0085, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 9.955555555555556e-06, | |
| "loss": 0.0063, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.0084, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 9.844444444444446e-06, | |
| "loss": 0.0049, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 9.78888888888889e-06, | |
| "loss": 0.0059, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 0.0066, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 0.0053, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 0.004, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 0.0054, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 0.0041, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 0.0042, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 11.07, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.0031, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 11.43, | |
| "learning_rate": 9.344444444444446e-06, | |
| "loss": 0.0042, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 11.79, | |
| "learning_rate": 9.28888888888889e-06, | |
| "loss": 0.0029, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 12.14, | |
| "learning_rate": 9.233333333333334e-06, | |
| "loss": 0.0029, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 9.17777777777778e-06, | |
| "loss": 0.0021, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 12.86, | |
| "learning_rate": 9.122222222222223e-06, | |
| "loss": 0.0018, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 0.0015, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 9.011111111111111e-06, | |
| "loss": 0.0018, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 8.955555555555555e-06, | |
| "loss": 0.0012, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.001, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "eval_loss": 0.398193359375, | |
| "eval_runtime": 830.3901, | |
| "eval_samples_per_second": 2.73, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 23.14374107567825, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 14.64, | |
| "learning_rate": 8.844444444444445e-06, | |
| "loss": 0.0008, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 8.788888888888891e-06, | |
| "loss": 0.0007, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 15.36, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 0.0006, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 15.71, | |
| "learning_rate": 8.677777777777779e-06, | |
| "loss": 0.0005, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 16.07, | |
| "learning_rate": 8.622222222222223e-06, | |
| "loss": 0.0008, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 16.43, | |
| "learning_rate": 8.566666666666667e-06, | |
| "loss": 0.0007, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 16.79, | |
| "learning_rate": 8.511111111111113e-06, | |
| "loss": 0.0006, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 17.14, | |
| "learning_rate": 8.455555555555555e-06, | |
| "loss": 0.0005, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 17.5, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.0005, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 17.86, | |
| "learning_rate": 8.344444444444445e-06, | |
| "loss": 0.0007, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 18.21, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 0.0009, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 18.57, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 0.0004, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 18.93, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 0.0005, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 19.29, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 0.0004, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 19.64, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 0.0009, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 20.36, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 0.0014, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 20.71, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.0015, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 21.07, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 0.0018, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 21.43, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 0.0016, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 21.79, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 0.0039, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 22.14, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 0.0059, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 22.5, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 0.0044, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 22.86, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 0.0036, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 23.21, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 0.0044, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 23.57, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 0.0052, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 23.93, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.0042, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 24.29, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 0.0025, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 24.64, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.0025, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 0.0028, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 25.36, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 0.0028, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 25.71, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 0.0018, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 26.07, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 0.0018, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 26.43, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 0.001, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 26.79, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 0.001, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 27.14, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.0008, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 27.5, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 0.0009, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 27.86, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 0.0007, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 28.21, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 0.0008, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 28.57, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.0013, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 28.57, | |
| "eval_loss": 0.434326171875, | |
| "eval_runtime": 808.1708, | |
| "eval_samples_per_second": 2.805, | |
| "eval_steps_per_second": 0.088, | |
| "eval_wer": 24.036173250832938, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 28.93, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.0009, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 29.29, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 29.64, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.0003, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 0.0012, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 30.36, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.0008, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 30.71, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 0.0003, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 31.07, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 31.43, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.0002, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 31.79, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 32.14, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 0.0002, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 32.5, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 32.86, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 33.21, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 0.0002, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 33.57, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.0002, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 33.93, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 34.29, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 34.64, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 0.0002, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 35.36, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 0.0002, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 35.71, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 36.07, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 0.0002, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 36.43, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 0.0002, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 36.79, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.0002, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 37.14, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 0.0002, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 37.5, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 37.86, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 38.21, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 0.0002, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 38.57, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 38.93, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 0.0002, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 39.29, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 39.64, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 0.0001, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.0001, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 40.36, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 0.0001, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 40.71, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 0.0001, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 41.07, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 41.43, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 41.79, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 0.0001, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 42.14, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 42.5, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 42.86, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 0.0001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 42.86, | |
| "eval_loss": 0.45654296875, | |
| "eval_runtime": 844.8019, | |
| "eval_samples_per_second": 2.683, | |
| "eval_steps_per_second": 0.084, | |
| "eval_wer": 23.322227510709187, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 43.21, | |
| "learning_rate": 4.404444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 43.57, | |
| "learning_rate": 4.348888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 43.93, | |
| "learning_rate": 4.2933333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 44.29, | |
| "learning_rate": 4.2377777777777775e-06, | |
| "loss": 0.0001, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 44.64, | |
| "learning_rate": 4.182222222222222e-06, | |
| "loss": 0.0001, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "learning_rate": 4.126666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 45.36, | |
| "learning_rate": 4.071111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 45.71, | |
| "learning_rate": 4.015555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 46.07, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.0001, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 46.43, | |
| "learning_rate": 3.904444444444444e-06, | |
| "loss": 0.0001, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 46.79, | |
| "learning_rate": 3.848888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 47.14, | |
| "learning_rate": 3.793333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 47.5, | |
| "learning_rate": 3.737777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 47.86, | |
| "learning_rate": 3.6822222222222225e-06, | |
| "loss": 0.0001, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 48.21, | |
| "learning_rate": 3.6266666666666674e-06, | |
| "loss": 0.0001, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 48.57, | |
| "learning_rate": 3.5711111111111114e-06, | |
| "loss": 0.0001, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 48.93, | |
| "learning_rate": 3.515555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 49.29, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.0001, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 49.64, | |
| "learning_rate": 3.404444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 3.3488888888888892e-06, | |
| "loss": 0.0001, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 50.36, | |
| "learning_rate": 3.2933333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 50.71, | |
| "learning_rate": 3.237777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 51.07, | |
| "learning_rate": 3.1822222222222226e-06, | |
| "loss": 0.0001, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 51.43, | |
| "learning_rate": 3.1266666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 51.79, | |
| "learning_rate": 3.0711111111111115e-06, | |
| "loss": 0.0001, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 52.14, | |
| "learning_rate": 3.015555555555556e-06, | |
| "loss": 0.0001, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 52.5, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.0001, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 52.86, | |
| "learning_rate": 2.904444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 53.21, | |
| "learning_rate": 2.8488888888888894e-06, | |
| "loss": 0.0001, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 53.57, | |
| "learning_rate": 2.7933333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 53.93, | |
| "learning_rate": 2.7377777777777783e-06, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 54.29, | |
| "learning_rate": 2.6822222222222223e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 54.64, | |
| "learning_rate": 2.6266666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "learning_rate": 2.5711111111111112e-06, | |
| "loss": 0.0001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 55.36, | |
| "learning_rate": 2.5155555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 55.71, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.0001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 56.07, | |
| "learning_rate": 2.4044444444444446e-06, | |
| "loss": 0.0001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 56.43, | |
| "learning_rate": 2.348888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 56.79, | |
| "learning_rate": 2.2933333333333335e-06, | |
| "loss": 0.0001, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 57.14, | |
| "learning_rate": 2.237777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 57.14, | |
| "eval_loss": 0.469970703125, | |
| "eval_runtime": 827.6996, | |
| "eval_samples_per_second": 2.739, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 23.39362208472156, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 57.5, | |
| "learning_rate": 2.1866666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 57.86, | |
| "learning_rate": 2.1311111111111112e-06, | |
| "loss": 0.0001, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 58.21, | |
| "learning_rate": 2.0755555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 58.57, | |
| "learning_rate": 2.02e-06, | |
| "loss": 0.0001, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 58.93, | |
| "learning_rate": 1.9644444444444446e-06, | |
| "loss": 0.0001, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 59.29, | |
| "learning_rate": 1.908888888888889e-06, | |
| "loss": 0.0001, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 59.64, | |
| "learning_rate": 1.8533333333333333e-06, | |
| "loss": 0.0001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 1.797777777777778e-06, | |
| "loss": 0.0001, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 60.36, | |
| "learning_rate": 1.7422222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 60.71, | |
| "learning_rate": 1.6866666666666667e-06, | |
| "loss": 0.0001, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 61.07, | |
| "learning_rate": 1.6311111111111114e-06, | |
| "loss": 0.0001, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 61.43, | |
| "learning_rate": 1.5755555555555558e-06, | |
| "loss": 0.0001, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 61.79, | |
| "learning_rate": 1.52e-06, | |
| "loss": 0.0001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 62.14, | |
| "learning_rate": 1.4644444444444445e-06, | |
| "loss": 0.0001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 62.5, | |
| "learning_rate": 1.4088888888888892e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 62.86, | |
| "learning_rate": 1.3533333333333334e-06, | |
| "loss": 0.0001, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 63.21, | |
| "learning_rate": 1.2977777777777779e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 63.57, | |
| "learning_rate": 1.2422222222222224e-06, | |
| "loss": 0.0001, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 63.93, | |
| "learning_rate": 1.1866666666666668e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 64.29, | |
| "learning_rate": 1.131111111111111e-06, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 64.64, | |
| "learning_rate": 1.0755555555555557e-06, | |
| "loss": 0.0001, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "learning_rate": 1.02e-06, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 65.36, | |
| "learning_rate": 9.644444444444444e-07, | |
| "loss": 0.0001, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 65.71, | |
| "learning_rate": 9.08888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 66.07, | |
| "learning_rate": 8.533333333333334e-07, | |
| "loss": 0.0001, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 66.43, | |
| "learning_rate": 7.977777777777779e-07, | |
| "loss": 0.0001, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 66.79, | |
| "learning_rate": 7.422222222222223e-07, | |
| "loss": 0.0001, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 67.14, | |
| "learning_rate": 6.866666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 67.5, | |
| "learning_rate": 6.311111111111112e-07, | |
| "loss": 0.0001, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 67.86, | |
| "learning_rate": 5.755555555555555e-07, | |
| "loss": 0.0001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 68.21, | |
| "learning_rate": 5.2e-07, | |
| "loss": 0.0001, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 68.57, | |
| "learning_rate": 4.6444444444444446e-07, | |
| "loss": 0.0001, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 68.93, | |
| "learning_rate": 4.0888888888888897e-07, | |
| "loss": 0.0001, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 69.29, | |
| "learning_rate": 3.533333333333334e-07, | |
| "loss": 0.0001, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 69.64, | |
| "learning_rate": 2.977777777777778e-07, | |
| "loss": 0.0001, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "learning_rate": 2.4222222222222224e-07, | |
| "loss": 0.0001, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 70.36, | |
| "learning_rate": 1.866666666666667e-07, | |
| "loss": 0.0001, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 70.71, | |
| "learning_rate": 1.3111111111111113e-07, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 71.07, | |
| "learning_rate": 7.555555555555556e-08, | |
| "loss": 0.0001, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 71.43, | |
| "learning_rate": 2e-08, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 71.43, | |
| "eval_loss": 0.475341796875, | |
| "eval_runtime": 788.365, | |
| "eval_samples_per_second": 2.876, | |
| "eval_steps_per_second": 0.09, | |
| "eval_wer": 23.453117563065206, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 71.43, | |
| "step": 5000, | |
| "total_flos": 3.2644864551847304e+20, | |
| "train_loss": 0.011088945543766021, | |
| "train_runtime": 79780.349, | |
| "train_samples_per_second": 4.011, | |
| "train_steps_per_second": 0.063 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 72, | |
| "total_flos": 3.2644864551847304e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |