| { | |
| "best_metric": 9.045873924973758, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 21.1864406779661, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.973833272194737e-06, | |
| "loss": 0.3246, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 6.195318418690893e-06, | |
| "loss": 0.1691, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.881634451095711e-06, | |
| "loss": 0.1644, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 7.361221988663844e-06, | |
| "loss": 0.1458, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 7.730207550743121e-06, | |
| "loss": 0.1389, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.03016458599496e-06, | |
| "loss": 0.1376, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.282894746203441e-06, | |
| "loss": 0.1374, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.501266121799902e-06, | |
| "loss": 0.1287, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.686247975778677e-06, | |
| "loss": 0.1225, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.852140188761744e-06, | |
| "loss": 0.0911, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.00848753507038e-06, | |
| "loss": 0.0701, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.150979862726452e-06, | |
| "loss": 0.068, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.281874101213678e-06, | |
| "loss": 0.0673, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.402917005361869e-06, | |
| "loss": 0.0679, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.51548820454122e-06, | |
| "loss": 0.0671, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.620696382156558e-06, | |
| "loss": 0.0739, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.719445885591654e-06, | |
| "loss": 0.0675, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.812484046603779e-06, | |
| "loss": 0.0662, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.900435550016748e-06, | |
| "loss": 0.0615, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.98382788472848e-06, | |
| "loss": 0.0323, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 9.957777777777779e-06, | |
| "loss": 0.0333, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 9.902222222222223e-06, | |
| "loss": 0.0356, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.846666666666668e-06, | |
| "loss": 0.0348, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 9.791111111111112e-06, | |
| "loss": 0.034, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 9.735555555555556e-06, | |
| "loss": 0.0339, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 9.68e-06, | |
| "loss": 0.0354, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.624444444444445e-06, | |
| "loss": 0.0374, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 9.56888888888889e-06, | |
| "loss": 0.0371, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.513333333333334e-06, | |
| "loss": 0.0227, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 9.457777777777778e-06, | |
| "loss": 0.0173, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 9.402222222222222e-06, | |
| "loss": 0.0172, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 9.346666666666666e-06, | |
| "loss": 0.0166, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 9.291111111111112e-06, | |
| "loss": 0.0196, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 9.235555555555556e-06, | |
| "loss": 0.0166, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 9.180000000000002e-06, | |
| "loss": 0.0188, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 9.124444444444444e-06, | |
| "loss": 0.0201, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 9.06888888888889e-06, | |
| "loss": 0.0204, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 9.013333333333334e-06, | |
| "loss": 0.0177, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 8.957777777777778e-06, | |
| "loss": 0.0101, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 8.902222222222224e-06, | |
| "loss": 0.0106, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "eval_loss": 0.1624755859375, | |
| "eval_runtime": 2123.3545, | |
| "eval_samples_per_second": 3.633, | |
| "eval_steps_per_second": 0.454, | |
| "eval_wer": 9.988766321062228, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 8.846666666666668e-06, | |
| "loss": 0.0098, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 8.791111111111112e-06, | |
| "loss": 0.0109, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 8.735555555555556e-06, | |
| "loss": 0.0098, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 8.68e-06, | |
| "loss": 0.0095, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 8.624444444444446e-06, | |
| "loss": 0.0109, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 8.56888888888889e-06, | |
| "loss": 0.0102, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 8.513333333333335e-06, | |
| "loss": 0.0114, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 8.457777777777778e-06, | |
| "loss": 0.0097, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 8.402222222222223e-06, | |
| "loss": 0.0075, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 8.346666666666668e-06, | |
| "loss": 0.0063, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 8.291111111111112e-06, | |
| "loss": 0.0063, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 8.235555555555557e-06, | |
| "loss": 0.0057, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 8.18e-06, | |
| "loss": 0.0058, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 8.124444444444445e-06, | |
| "loss": 0.0065, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 8.06888888888889e-06, | |
| "loss": 0.0059, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 8.013333333333333e-06, | |
| "loss": 0.0061, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.957777777777779e-06, | |
| "loss": 0.0069, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 7.902222222222223e-06, | |
| "loss": 0.0042, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 7.846666666666667e-06, | |
| "loss": 0.0036, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 7.791111111111111e-06, | |
| "loss": 0.004, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 7.735555555555557e-06, | |
| "loss": 0.0043, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 7.680000000000001e-06, | |
| "loss": 0.0049, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 7.624444444444445e-06, | |
| "loss": 0.0044, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 7.56888888888889e-06, | |
| "loss": 0.0042, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 7.513333333333334e-06, | |
| "loss": 0.0047, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 7.457777777777778e-06, | |
| "loss": 0.006, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 7.402222222222223e-06, | |
| "loss": 0.0034, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 7.346666666666668e-06, | |
| "loss": 0.0029, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 7.291111111111112e-06, | |
| "loss": 0.0031, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 7.235555555555556e-06, | |
| "loss": 0.003, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 7.180000000000001e-06, | |
| "loss": 0.0028, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 7.124444444444445e-06, | |
| "loss": 0.0039, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 7.06888888888889e-06, | |
| "loss": 0.0031, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 7.0133333333333345e-06, | |
| "loss": 0.0027, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 6.9577777777777785e-06, | |
| "loss": 0.0034, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 6.902222222222223e-06, | |
| "loss": 0.0033, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.16, | |
| "learning_rate": 6.846666666666667e-06, | |
| "loss": 0.0032, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 6.7911111111111115e-06, | |
| "loss": 0.0033, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 6.735555555555556e-06, | |
| "loss": 0.0035, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 6.680000000000001e-06, | |
| "loss": 0.0034, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "eval_loss": 0.18408203125, | |
| "eval_runtime": 2143.7282, | |
| "eval_samples_per_second": 3.598, | |
| "eval_steps_per_second": 0.45, | |
| "eval_wer": 9.830389863906742, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 6.6244444444444445e-06, | |
| "loss": 0.0044, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 6.568888888888889e-06, | |
| "loss": 0.0038, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 6.513333333333333e-06, | |
| "loss": 0.0043, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 6.457777777777778e-06, | |
| "loss": 0.0033, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 6.402222222222223e-06, | |
| "loss": 0.0034, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 6.346666666666668e-06, | |
| "loss": 0.0027, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 6.291111111111111e-06, | |
| "loss": 0.0032, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 6.235555555555556e-06, | |
| "loss": 0.0028, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 6.18e-06, | |
| "loss": 0.0027, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 6.124444444444445e-06, | |
| "loss": 0.0029, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 6.06888888888889e-06, | |
| "loss": 0.0036, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 6.013333333333335e-06, | |
| "loss": 0.0032, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 5.957777777777778e-06, | |
| "loss": 0.0039, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 5.902222222222223e-06, | |
| "loss": 0.0035, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 5.846666666666667e-06, | |
| "loss": 0.003, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 10.17, | |
| "learning_rate": 5.791111111111112e-06, | |
| "loss": 0.0021, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 5.735555555555557e-06, | |
| "loss": 0.0021, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 10.38, | |
| "learning_rate": 5.68e-06, | |
| "loss": 0.002, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 10.49, | |
| "learning_rate": 5.624444444444445e-06, | |
| "loss": 0.0024, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 5.56888888888889e-06, | |
| "loss": 0.0025, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.7, | |
| "learning_rate": 5.513333333333334e-06, | |
| "loss": 0.0021, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 5.4577777777777785e-06, | |
| "loss": 0.0018, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "learning_rate": 5.402222222222223e-06, | |
| "loss": 0.0022, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.346666666666667e-06, | |
| "loss": 0.0016, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 5.2911111111111115e-06, | |
| "loss": 0.0011, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 5.235555555555556e-06, | |
| "loss": 0.0012, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 11.33, | |
| "learning_rate": 5.18e-06, | |
| "loss": 0.0012, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 5.124444444444445e-06, | |
| "loss": 0.0011, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 5.06888888888889e-06, | |
| "loss": 0.001, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 11.65, | |
| "learning_rate": 5.013333333333333e-06, | |
| "loss": 0.0012, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 11.76, | |
| "learning_rate": 4.957777777777778e-06, | |
| "loss": 0.0012, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 11.86, | |
| "learning_rate": 4.902222222222222e-06, | |
| "loss": 0.0017, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 11.97, | |
| "learning_rate": 4.846666666666667e-06, | |
| "loss": 0.0011, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 12.08, | |
| "learning_rate": 4.791111111111111e-06, | |
| "loss": 0.0008, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 12.18, | |
| "learning_rate": 4.735555555555556e-06, | |
| "loss": 0.0007, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 12.29, | |
| "learning_rate": 4.680000000000001e-06, | |
| "loss": 0.0008, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 12.39, | |
| "learning_rate": 4.624444444444445e-06, | |
| "loss": 0.001, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 4.568888888888889e-06, | |
| "loss": 0.0012, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 12.61, | |
| "learning_rate": 4.513333333333333e-06, | |
| "loss": 0.0012, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 4.457777777777778e-06, | |
| "loss": 0.0011, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "eval_loss": 0.191650390625, | |
| "eval_runtime": 2117.549, | |
| "eval_samples_per_second": 3.643, | |
| "eval_steps_per_second": 0.456, | |
| "eval_wer": 9.403141746929155, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.82, | |
| "learning_rate": 4.402222222222223e-06, | |
| "loss": 0.001, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 12.92, | |
| "learning_rate": 4.346666666666667e-06, | |
| "loss": 0.0011, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 4.291111111111112e-06, | |
| "loss": 0.0011, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 13.14, | |
| "learning_rate": 4.235555555555556e-06, | |
| "loss": 0.0012, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 13.24, | |
| "learning_rate": 4.18e-06, | |
| "loss": 0.0007, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "learning_rate": 4.124444444444445e-06, | |
| "loss": 0.0007, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "learning_rate": 4.0688888888888896e-06, | |
| "loss": 0.0007, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 4.013333333333334e-06, | |
| "loss": 0.0007, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 13.67, | |
| "learning_rate": 3.9577777777777785e-06, | |
| "loss": 0.001, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 13.77, | |
| "learning_rate": 3.9022222222222225e-06, | |
| "loss": 0.0006, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "learning_rate": 3.8466666666666665e-06, | |
| "loss": 0.0007, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 13.98, | |
| "learning_rate": 3.7911111111111114e-06, | |
| "loss": 0.0006, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "learning_rate": 3.7355555555555555e-06, | |
| "loss": 0.0005, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 3.6800000000000003e-06, | |
| "loss": 0.0005, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 14.3, | |
| "learning_rate": 3.624444444444445e-06, | |
| "loss": 0.0009, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 14.41, | |
| "learning_rate": 3.568888888888889e-06, | |
| "loss": 0.0008, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 14.51, | |
| "learning_rate": 3.5133333333333337e-06, | |
| "loss": 0.0006, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 3.457777777777778e-06, | |
| "loss": 0.0006, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 14.72, | |
| "learning_rate": 3.4022222222222222e-06, | |
| "loss": 0.0006, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "learning_rate": 3.346666666666667e-06, | |
| "loss": 0.0007, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 3.2911111111111116e-06, | |
| "loss": 0.0007, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 3.2355555555555556e-06, | |
| "loss": 0.0006, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 15.15, | |
| "learning_rate": 3.1800000000000005e-06, | |
| "loss": 0.0006, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 15.25, | |
| "learning_rate": 3.124444444444445e-06, | |
| "loss": 0.0007, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 15.36, | |
| "learning_rate": 3.068888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 15.47, | |
| "learning_rate": 3.013333333333334e-06, | |
| "loss": 0.0005, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 15.57, | |
| "learning_rate": 2.957777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 15.68, | |
| "learning_rate": 2.9022222222222223e-06, | |
| "loss": 0.0004, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 15.78, | |
| "learning_rate": 2.8466666666666672e-06, | |
| "loss": 0.0004, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 2.7911111111111113e-06, | |
| "loss": 0.0004, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 2.7355555555555557e-06, | |
| "loss": 0.0004, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 16.1, | |
| "learning_rate": 2.68e-06, | |
| "loss": 0.0004, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 16.21, | |
| "learning_rate": 2.6244444444444446e-06, | |
| "loss": 0.0004, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 16.31, | |
| "learning_rate": 2.568888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 2.5133333333333336e-06, | |
| "loss": 0.0004, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 16.53, | |
| "learning_rate": 2.457777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 16.63, | |
| "learning_rate": 2.4022222222222225e-06, | |
| "loss": 0.0004, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 16.74, | |
| "learning_rate": 2.346666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 16.84, | |
| "learning_rate": 2.2911111111111114e-06, | |
| "loss": 0.0004, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "learning_rate": 2.235555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "eval_loss": 0.20751953125, | |
| "eval_runtime": 2104.1262, | |
| "eval_samples_per_second": 3.666, | |
| "eval_steps_per_second": 0.459, | |
| "eval_wer": 9.117695806707058, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 17.06, | |
| "learning_rate": 2.1800000000000003e-06, | |
| "loss": 0.0004, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 17.16, | |
| "learning_rate": 2.1244444444444443e-06, | |
| "loss": 0.0003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 17.27, | |
| "learning_rate": 2.0688888888888892e-06, | |
| "loss": 0.0003, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 17.37, | |
| "learning_rate": 2.0133333333333337e-06, | |
| "loss": 0.0003, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 1.9577777777777777e-06, | |
| "loss": 0.0003, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 17.58, | |
| "learning_rate": 1.9022222222222222e-06, | |
| "loss": 0.0003, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "learning_rate": 1.8466666666666668e-06, | |
| "loss": 0.0003, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 17.8, | |
| "learning_rate": 1.7911111111111113e-06, | |
| "loss": 0.0003, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 17.9, | |
| "learning_rate": 1.7355555555555555e-06, | |
| "loss": 0.0003, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.6800000000000002e-06, | |
| "loss": 0.0003, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 18.11, | |
| "learning_rate": 1.6244444444444447e-06, | |
| "loss": 0.0003, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 1.568888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 18.33, | |
| "learning_rate": 1.5133333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 18.43, | |
| "learning_rate": 1.457777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 18.54, | |
| "learning_rate": 1.4022222222222223e-06, | |
| "loss": 0.0003, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 18.64, | |
| "learning_rate": 1.3466666666666668e-06, | |
| "loss": 0.0003, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 1.2911111111111112e-06, | |
| "loss": 0.0003, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 18.86, | |
| "learning_rate": 1.2355555555555557e-06, | |
| "loss": 0.0003, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 18.96, | |
| "learning_rate": 1.1800000000000001e-06, | |
| "loss": 0.0003, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 19.07, | |
| "learning_rate": 1.1244444444444446e-06, | |
| "loss": 0.0003, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 19.17, | |
| "learning_rate": 1.068888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 19.28, | |
| "learning_rate": 1.0133333333333333e-06, | |
| "loss": 0.0003, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 19.39, | |
| "learning_rate": 9.57777777777778e-07, | |
| "loss": 0.0003, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 19.49, | |
| "learning_rate": 9.022222222222222e-07, | |
| "loss": 0.0003, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 19.6, | |
| "learning_rate": 8.466666666666668e-07, | |
| "loss": 0.0003, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 19.7, | |
| "learning_rate": 7.911111111111111e-07, | |
| "loss": 0.0003, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 19.81, | |
| "learning_rate": 7.355555555555556e-07, | |
| "loss": 0.0003, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 19.92, | |
| "learning_rate": 6.800000000000001e-07, | |
| "loss": 0.0003, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 6.244444444444445e-07, | |
| "loss": 0.0003, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 20.13, | |
| "learning_rate": 5.68888888888889e-07, | |
| "loss": 0.0003, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 20.23, | |
| "learning_rate": 5.133333333333334e-07, | |
| "loss": 0.0003, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 20.34, | |
| "learning_rate": 4.5777777777777784e-07, | |
| "loss": 0.0003, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 20.44, | |
| "learning_rate": 4.0222222222222224e-07, | |
| "loss": 0.0003, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 20.55, | |
| "learning_rate": 3.466666666666667e-07, | |
| "loss": 0.0003, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 20.66, | |
| "learning_rate": 2.9111111111111116e-07, | |
| "loss": 0.0003, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 20.76, | |
| "learning_rate": 2.3555555555555556e-07, | |
| "loss": 0.0003, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 20.87, | |
| "learning_rate": 1.8e-07, | |
| "loss": 0.0003, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 20.97, | |
| "learning_rate": 1.2444444444444446e-07, | |
| "loss": 0.0003, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 21.08, | |
| "learning_rate": 6.888888888888889e-08, | |
| "loss": 0.0003, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "learning_rate": 1.3333333333333334e-08, | |
| "loss": 0.0003, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "eval_loss": 0.2120361328125, | |
| "eval_runtime": 2098.0347, | |
| "eval_samples_per_second": 3.677, | |
| "eval_steps_per_second": 0.46, | |
| "eval_wer": 9.045873924973758, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "step": 5000, | |
| "total_flos": 6.786121739322608e+20, | |
| "train_loss": 0.015194951736927033, | |
| "train_runtime": 67979.462, | |
| "train_samples_per_second": 4.707, | |
| "train_steps_per_second": 0.074 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 22, | |
| "total_flos": 6.786121739322608e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |