{ "best_metric": 17.571297148114077, "best_model_checkpoint": "./checkpoint-1000", "epoch": 43.0024, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.898977360288234e-06, "loss": 0.817, "step": 25 }, { "epoch": 0.01, "learning_rate": 6.160712527409633e-06, "loss": 0.3452, "step": 50 }, { "epoch": 0.01, "learning_rate": 6.85912902234906e-06, "loss": 0.3043, "step": 75 }, { "epoch": 0.02, "learning_rate": 7.344547104469332e-06, "loss": 0.2484, "step": 100 }, { "epoch": 1.0, "learning_rate": 7.716963756434345e-06, "loss": 0.1897, "step": 125 }, { "epoch": 1.01, "learning_rate": 8.019180844200955e-06, "loss": 0.1751, "step": 150 }, { "epoch": 1.01, "learning_rate": 8.27351214279797e-06, "loss": 0.1499, "step": 175 }, { "epoch": 1.02, "learning_rate": 8.49307723936858e-06, "loss": 0.145, "step": 200 }, { "epoch": 1.02, "learning_rate": 8.686247975778677e-06, "loss": 0.1039, "step": 225 }, { "epoch": 2.0, "learning_rate": 8.858694625217149e-06, "loss": 0.0958, "step": 250 }, { "epoch": 2.01, "learning_rate": 9.014436199608479e-06, "loss": 0.086, "step": 275 }, { "epoch": 2.01, "learning_rate": 9.156425255148058e-06, "loss": 0.0686, "step": 300 }, { "epoch": 2.02, "learning_rate": 9.28689473531776e-06, "loss": 0.0684, "step": 325 }, { "epoch": 3.0, "learning_rate": 9.407574351377137e-06, "loss": 0.0482, "step": 350 }, { "epoch": 3.01, "learning_rate": 9.519831289296397e-06, "loss": 0.0504, "step": 375 }, { "epoch": 3.01, "learning_rate": 9.624764935335318e-06, "loss": 0.0446, "step": 400 }, { "epoch": 3.02, "learning_rate": 9.723272550712454e-06, "loss": 0.0396, "step": 425 }, { "epoch": 3.02, "learning_rate": 9.816095971633122e-06, "loss": 0.0309, "step": 450 }, { "epoch": 4.0, "learning_rate": 9.90385555539545e-06, "loss": 0.0269, "step": 475 }, { "epoch": 4.01, "learning_rate": 9.987075336738768e-06, "loss": 0.03, "step": 500 }, { "epoch": 4.01, "learning_rate": 9.955555555555556e-06, "loss": 0.025, "step": 525 }, { "epoch": 4.02, "learning_rate": 9.9e-06, "loss": 0.0214, "step": 550 }, { "epoch": 4.02, "learning_rate": 9.844444444444446e-06, "loss": 0.0176, "step": 575 }, { "epoch": 5.0, "learning_rate": 9.78888888888889e-06, "loss": 0.0186, "step": 600 }, { "epoch": 5.01, "learning_rate": 9.733333333333334e-06, "loss": 0.0179, "step": 625 }, { "epoch": 5.01, "learning_rate": 9.677777777777778e-06, "loss": 0.0141, "step": 650 }, { "epoch": 5.02, "learning_rate": 9.622222222222222e-06, "loss": 0.0131, "step": 675 }, { "epoch": 6.0, "learning_rate": 9.566666666666668e-06, "loss": 0.0111, "step": 700 }, { "epoch": 6.01, "learning_rate": 9.511111111111112e-06, "loss": 0.0115, "step": 725 }, { "epoch": 6.01, "learning_rate": 9.455555555555557e-06, "loss": 0.0097, "step": 750 }, { "epoch": 6.02, "learning_rate": 9.4e-06, "loss": 0.0091, "step": 775 }, { "epoch": 6.02, "learning_rate": 9.344444444444446e-06, "loss": 0.0099, "step": 800 }, { "epoch": 7.0, "learning_rate": 9.28888888888889e-06, "loss": 0.0073, "step": 825 }, { "epoch": 7.01, "learning_rate": 9.233333333333334e-06, "loss": 0.0062, "step": 850 }, { "epoch": 7.01, "learning_rate": 9.17777777777778e-06, "loss": 0.006, "step": 875 }, { "epoch": 7.02, "learning_rate": 9.122222222222223e-06, "loss": 0.0046, "step": 900 }, { "epoch": 7.02, "learning_rate": 9.066666666666667e-06, "loss": 0.0045, "step": 925 }, { "epoch": 8.0, "learning_rate": 9.011111111111111e-06, "loss": 0.0053, "step": 950 }, { "epoch": 8.01, "learning_rate": 8.955555555555555e-06, "loss": 0.0039, "step": 975 }, { "epoch": 8.01, "learning_rate": 8.900000000000001e-06, "loss": 0.0046, "step": 1000 }, { "epoch": 8.01, "eval_loss": 0.28076171875, "eval_runtime": 1237.4696, "eval_samples_per_second": 3.118, "eval_steps_per_second": 0.098, "eval_wer": 17.571297148114077, "step": 1000 }, { "epoch": 8.02, "learning_rate": 8.844444444444445e-06, "loss": 0.0037, "step": 1025 }, { "epoch": 9.0, "learning_rate": 8.788888888888891e-06, "loss": 0.0035, "step": 1050 }, { "epoch": 9.01, "learning_rate": 8.733333333333333e-06, "loss": 0.0038, "step": 1075 }, { "epoch": 9.01, "learning_rate": 8.677777777777779e-06, "loss": 0.003, "step": 1100 }, { "epoch": 9.02, "learning_rate": 8.622222222222223e-06, "loss": 0.0037, "step": 1125 }, { "epoch": 9.02, "learning_rate": 8.566666666666667e-06, "loss": 0.0036, "step": 1150 }, { "epoch": 10.0, "learning_rate": 8.511111111111113e-06, "loss": 0.0034, "step": 1175 }, { "epoch": 10.01, "learning_rate": 8.455555555555555e-06, "loss": 0.0032, "step": 1200 }, { "epoch": 10.01, "learning_rate": 8.400000000000001e-06, "loss": 0.0028, "step": 1225 }, { "epoch": 10.02, "learning_rate": 8.344444444444445e-06, "loss": 0.0026, "step": 1250 }, { "epoch": 10.02, "learning_rate": 8.288888888888889e-06, "loss": 0.0022, "step": 1275 }, { "epoch": 11.0, "learning_rate": 8.233333333333335e-06, "loss": 0.002, "step": 1300 }, { "epoch": 11.01, "learning_rate": 8.177777777777779e-06, "loss": 0.0019, "step": 1325 }, { "epoch": 11.01, "learning_rate": 8.122222222222223e-06, "loss": 0.0015, "step": 1350 }, { "epoch": 11.02, "learning_rate": 8.066666666666667e-06, "loss": 0.0013, "step": 1375 }, { "epoch": 12.0, "learning_rate": 8.011111111111113e-06, "loss": 0.0011, "step": 1400 }, { "epoch": 12.01, "learning_rate": 7.955555555555557e-06, "loss": 0.0012, "step": 1425 }, { "epoch": 12.01, "learning_rate": 7.9e-06, "loss": 0.0011, "step": 1450 }, { "epoch": 12.02, "learning_rate": 7.844444444444446e-06, "loss": 0.0012, "step": 1475 }, { "epoch": 12.02, "learning_rate": 7.788888888888889e-06, "loss": 0.001, "step": 1500 }, { "epoch": 13.0, "learning_rate": 7.733333333333334e-06, "loss": 0.0009, "step": 1525 }, { "epoch": 13.01, "learning_rate": 7.677777777777778e-06, "loss": 0.0009, "step": 1550 }, { "epoch": 13.01, "learning_rate": 7.622222222222223e-06, "loss": 0.0012, "step": 1575 }, { "epoch": 13.02, "learning_rate": 7.566666666666667e-06, "loss": 0.0009, "step": 1600 }, { "epoch": 14.0, "learning_rate": 7.511111111111111e-06, "loss": 0.0008, "step": 1625 }, { "epoch": 14.01, "learning_rate": 7.455555555555556e-06, "loss": 0.0008, "step": 1650 }, { "epoch": 14.01, "learning_rate": 7.4e-06, "loss": 0.0008, "step": 1675 }, { "epoch": 14.02, "learning_rate": 7.344444444444445e-06, "loss": 0.0008, "step": 1700 }, { "epoch": 14.02, "learning_rate": 7.28888888888889e-06, "loss": 0.0007, "step": 1725 }, { "epoch": 15.0, "learning_rate": 7.233333333333334e-06, "loss": 0.0007, "step": 1750 }, { "epoch": 15.01, "learning_rate": 7.177777777777778e-06, "loss": 0.0007, "step": 1775 }, { "epoch": 15.01, "learning_rate": 7.122222222222222e-06, "loss": 0.0007, "step": 1800 }, { "epoch": 15.02, "learning_rate": 7.066666666666667e-06, "loss": 0.0007, "step": 1825 }, { "epoch": 15.02, "learning_rate": 7.011111111111112e-06, "loss": 0.0006, "step": 1850 }, { "epoch": 16.0, "learning_rate": 6.955555555555557e-06, "loss": 0.0006, "step": 1875 }, { "epoch": 16.01, "learning_rate": 6.9e-06, "loss": 0.0006, "step": 1900 }, { "epoch": 16.01, "learning_rate": 6.844444444444445e-06, "loss": 0.0006, "step": 1925 }, { "epoch": 16.02, "learning_rate": 6.788888888888889e-06, "loss": 0.0006, "step": 1950 }, { "epoch": 17.0, "learning_rate": 6.733333333333334e-06, "loss": 0.0005, "step": 1975 }, { "epoch": 17.01, "learning_rate": 6.677777777777779e-06, "loss": 0.0006, "step": 2000 }, { "epoch": 17.01, "eval_loss": 0.30908203125, "eval_runtime": 1241.2535, "eval_samples_per_second": 3.109, "eval_steps_per_second": 0.097, "eval_wer": 17.69160002830656, "step": 2000 }, { "epoch": 17.01, "learning_rate": 6.6222222222222236e-06, "loss": 0.0006, "step": 2025 }, { "epoch": 17.02, "learning_rate": 6.566666666666667e-06, "loss": 0.0005, "step": 2050 }, { "epoch": 17.02, "learning_rate": 6.511111111111112e-06, "loss": 0.0005, "step": 2075 }, { "epoch": 18.0, "learning_rate": 6.455555555555556e-06, "loss": 0.0005, "step": 2100 }, { "epoch": 18.01, "learning_rate": 6.4000000000000006e-06, "loss": 0.0005, "step": 2125 }, { "epoch": 18.01, "learning_rate": 6.3444444444444454e-06, "loss": 0.0005, "step": 2150 }, { "epoch": 18.02, "learning_rate": 6.28888888888889e-06, "loss": 0.0005, "step": 2175 }, { "epoch": 18.02, "learning_rate": 6.2333333333333335e-06, "loss": 0.0005, "step": 2200 }, { "epoch": 19.0, "learning_rate": 6.177777777777778e-06, "loss": 0.0005, "step": 2225 }, { "epoch": 19.01, "learning_rate": 6.1222222222222224e-06, "loss": 0.0005, "step": 2250 }, { "epoch": 19.01, "learning_rate": 6.066666666666667e-06, "loss": 0.0005, "step": 2275 }, { "epoch": 19.02, "learning_rate": 6.011111111111112e-06, "loss": 0.0004, "step": 2300 }, { "epoch": 20.0, "learning_rate": 5.955555555555555e-06, "loss": 0.0004, "step": 2325 }, { "epoch": 20.01, "learning_rate": 5.9e-06, "loss": 0.0004, "step": 2350 }, { "epoch": 20.01, "learning_rate": 5.844444444444445e-06, "loss": 0.0004, "step": 2375 }, { "epoch": 20.02, "learning_rate": 5.788888888888889e-06, "loss": 0.0004, "step": 2400 }, { "epoch": 20.02, "learning_rate": 5.733333333333334e-06, "loss": 0.0004, "step": 2425 }, { "epoch": 21.0, "learning_rate": 5.677777777777779e-06, "loss": 0.0004, "step": 2450 }, { "epoch": 21.01, "learning_rate": 5.622222222222222e-06, "loss": 0.0004, "step": 2475 }, { "epoch": 21.01, "learning_rate": 5.566666666666667e-06, "loss": 0.0004, "step": 2500 }, { "epoch": 21.02, "learning_rate": 5.511111111111112e-06, "loss": 0.0004, "step": 2525 }, { "epoch": 21.02, "learning_rate": 5.455555555555556e-06, "loss": 0.0004, "step": 2550 }, { "epoch": 22.0, "learning_rate": 5.400000000000001e-06, "loss": 0.0004, "step": 2575 }, { "epoch": 22.01, "learning_rate": 5.344444444444446e-06, "loss": 0.0004, "step": 2600 }, { "epoch": 22.01, "learning_rate": 5.288888888888889e-06, "loss": 0.0004, "step": 2625 }, { "epoch": 22.02, "learning_rate": 5.233333333333334e-06, "loss": 0.0004, "step": 2650 }, { "epoch": 23.0, "learning_rate": 5.177777777777779e-06, "loss": 0.0003, "step": 2675 }, { "epoch": 23.01, "learning_rate": 5.122222222222223e-06, "loss": 0.0004, "step": 2700 }, { "epoch": 23.01, "learning_rate": 5.0666666666666676e-06, "loss": 0.0003, "step": 2725 }, { "epoch": 23.02, "learning_rate": 5.011111111111111e-06, "loss": 0.0003, "step": 2750 }, { "epoch": 23.02, "learning_rate": 4.9555555555555565e-06, "loss": 0.0003, "step": 2775 }, { "epoch": 24.0, "learning_rate": 4.9000000000000005e-06, "loss": 0.0003, "step": 2800 }, { "epoch": 24.01, "learning_rate": 4.8444444444444446e-06, "loss": 0.0003, "step": 2825 }, { "epoch": 24.01, "learning_rate": 4.7888888888888894e-06, "loss": 0.0003, "step": 2850 }, { "epoch": 24.02, "learning_rate": 4.7333333333333335e-06, "loss": 0.0003, "step": 2875 }, { "epoch": 24.02, "learning_rate": 4.677777777777778e-06, "loss": 0.0003, "step": 2900 }, { "epoch": 25.0, "learning_rate": 4.622222222222222e-06, "loss": 0.0003, "step": 2925 }, { "epoch": 25.01, "learning_rate": 4.566666666666667e-06, "loss": 0.0003, "step": 2950 }, { "epoch": 25.02, "learning_rate": 4.511111111111111e-06, "loss": 0.0003, "step": 2975 }, { "epoch": 25.02, "learning_rate": 4.455555555555555e-06, "loss": 0.0003, "step": 3000 }, { "epoch": 25.02, "eval_loss": 0.32666015625, "eval_runtime": 1253.1843, "eval_samples_per_second": 3.079, "eval_steps_per_second": 0.097, "eval_wer": 17.688061708300896, "step": 3000 }, { "epoch": 26.0, "learning_rate": 4.404444444444445e-06, "loss": 0.0003, "step": 3025 }, { "epoch": 26.01, "learning_rate": 4.348888888888889e-06, "loss": 0.0003, "step": 3050 }, { "epoch": 26.01, "learning_rate": 4.2933333333333334e-06, "loss": 0.0003, "step": 3075 }, { "epoch": 26.02, "learning_rate": 4.2377777777777775e-06, "loss": 0.0003, "step": 3100 }, { "epoch": 26.02, "learning_rate": 4.182222222222222e-06, "loss": 0.0003, "step": 3125 }, { "epoch": 27.0, "learning_rate": 4.126666666666667e-06, "loss": 0.0003, "step": 3150 }, { "epoch": 27.01, "learning_rate": 4.071111111111111e-06, "loss": 0.0003, "step": 3175 }, { "epoch": 27.01, "learning_rate": 4.015555555555556e-06, "loss": 0.0003, "step": 3200 }, { "epoch": 27.02, "learning_rate": 3.96e-06, "loss": 0.0003, "step": 3225 }, { "epoch": 28.0, "learning_rate": 3.904444444444444e-06, "loss": 0.0003, "step": 3250 }, { "epoch": 28.01, "learning_rate": 3.848888888888889e-06, "loss": 0.0003, "step": 3275 }, { "epoch": 28.01, "learning_rate": 3.793333333333334e-06, "loss": 0.0003, "step": 3300 }, { "epoch": 28.02, "learning_rate": 3.737777777777778e-06, "loss": 0.0003, "step": 3325 }, { "epoch": 28.02, "learning_rate": 3.6822222222222225e-06, "loss": 0.0003, "step": 3350 }, { "epoch": 29.0, "learning_rate": 3.6266666666666674e-06, "loss": 0.0003, "step": 3375 }, { "epoch": 29.01, "learning_rate": 3.5711111111111114e-06, "loss": 0.0003, "step": 3400 }, { "epoch": 29.01, "learning_rate": 3.515555555555556e-06, "loss": 0.0003, "step": 3425 }, { "epoch": 29.02, "learning_rate": 3.46e-06, "loss": 0.0002, "step": 3450 }, { "epoch": 29.02, "learning_rate": 3.404444444444445e-06, "loss": 0.0002, "step": 3475 }, { "epoch": 30.0, "learning_rate": 3.3488888888888892e-06, "loss": 0.0003, "step": 3500 }, { "epoch": 30.01, "learning_rate": 3.2933333333333333e-06, "loss": 0.0002, "step": 3525 }, { "epoch": 30.01, "learning_rate": 3.237777777777778e-06, "loss": 0.0002, "step": 3550 }, { "epoch": 30.02, "learning_rate": 3.1822222222222226e-06, "loss": 0.0002, "step": 3575 }, { "epoch": 31.0, "learning_rate": 3.1266666666666667e-06, "loss": 0.0002, "step": 3600 }, { "epoch": 31.01, "learning_rate": 3.0711111111111115e-06, "loss": 0.0002, "step": 3625 }, { "epoch": 31.01, "learning_rate": 3.015555555555556e-06, "loss": 0.0002, "step": 3650 }, { "epoch": 31.02, "learning_rate": 2.96e-06, "loss": 0.0002, "step": 3675 }, { "epoch": 31.02, "learning_rate": 2.904444444444445e-06, "loss": 0.0002, "step": 3700 }, { "epoch": 32.0, "learning_rate": 2.8488888888888894e-06, "loss": 0.0002, "step": 3725 }, { "epoch": 32.01, "learning_rate": 2.7933333333333334e-06, "loss": 0.0002, "step": 3750 }, { "epoch": 32.01, "learning_rate": 2.7377777777777783e-06, "loss": 0.0002, "step": 3775 }, { "epoch": 32.02, "learning_rate": 2.6822222222222223e-06, "loss": 0.0002, "step": 3800 }, { "epoch": 32.02, "learning_rate": 2.6266666666666668e-06, "loss": 0.0002, "step": 3825 }, { "epoch": 33.0, "learning_rate": 2.5711111111111112e-06, "loss": 0.0002, "step": 3850 }, { "epoch": 33.01, "learning_rate": 2.5155555555555557e-06, "loss": 0.0002, "step": 3875 }, { "epoch": 33.01, "learning_rate": 2.46e-06, "loss": 0.0002, "step": 3900 }, { "epoch": 33.02, "learning_rate": 2.4044444444444446e-06, "loss": 0.0002, "step": 3925 }, { "epoch": 34.0, "learning_rate": 2.348888888888889e-06, "loss": 0.0002, "step": 3950 }, { "epoch": 34.01, "learning_rate": 2.2933333333333335e-06, "loss": 0.0002, "step": 3975 }, { "epoch": 34.01, "learning_rate": 2.237777777777778e-06, "loss": 0.0002, "step": 4000 }, { "epoch": 34.01, "eval_loss": 0.338134765625, "eval_runtime": 1277.681, "eval_samples_per_second": 3.02, "eval_steps_per_second": 0.095, "eval_wer": 17.7977496284764, "step": 4000 }, { "epoch": 34.02, "learning_rate": 2.1866666666666668e-06, "loss": 0.0002, "step": 4025 }, { "epoch": 34.02, "learning_rate": 2.1311111111111112e-06, "loss": 0.0002, "step": 4050 }, { "epoch": 35.0, "learning_rate": 2.0755555555555557e-06, "loss": 0.0002, "step": 4075 }, { "epoch": 35.01, "learning_rate": 2.02e-06, "loss": 0.0002, "step": 4100 }, { "epoch": 35.01, "learning_rate": 1.9644444444444446e-06, "loss": 0.0002, "step": 4125 }, { "epoch": 35.02, "learning_rate": 1.908888888888889e-06, "loss": 0.0002, "step": 4150 }, { "epoch": 35.02, "learning_rate": 1.8533333333333333e-06, "loss": 0.0002, "step": 4175 }, { "epoch": 36.0, "learning_rate": 1.797777777777778e-06, "loss": 0.0002, "step": 4200 }, { "epoch": 36.01, "learning_rate": 1.7422222222222224e-06, "loss": 0.0002, "step": 4225 }, { "epoch": 36.01, "learning_rate": 1.6866666666666667e-06, "loss": 0.0002, "step": 4250 }, { "epoch": 36.02, "learning_rate": 1.6311111111111114e-06, "loss": 0.0002, "step": 4275 }, { "epoch": 37.0, "learning_rate": 1.5755555555555558e-06, "loss": 0.0002, "step": 4300 }, { "epoch": 37.01, "learning_rate": 1.52e-06, "loss": 0.0002, "step": 4325 }, { "epoch": 37.01, "learning_rate": 1.4644444444444445e-06, "loss": 0.0002, "step": 4350 }, { "epoch": 37.02, "learning_rate": 1.4088888888888892e-06, "loss": 0.0002, "step": 4375 }, { "epoch": 37.02, "learning_rate": 1.3533333333333334e-06, "loss": 0.0002, "step": 4400 }, { "epoch": 38.0, "learning_rate": 1.2977777777777779e-06, "loss": 0.0002, "step": 4425 }, { "epoch": 38.01, "learning_rate": 1.2422222222222224e-06, "loss": 0.0002, "step": 4450 }, { "epoch": 38.01, "learning_rate": 1.1866666666666668e-06, "loss": 0.0002, "step": 4475 }, { "epoch": 38.02, "learning_rate": 1.131111111111111e-06, "loss": 0.0002, "step": 4500 }, { "epoch": 39.0, "learning_rate": 1.0755555555555557e-06, "loss": 0.0002, "step": 4525 }, { "epoch": 39.01, "learning_rate": 1.02e-06, "loss": 0.0002, "step": 4550 }, { "epoch": 39.01, "learning_rate": 9.644444444444444e-07, "loss": 0.0002, "step": 4575 }, { "epoch": 39.02, "learning_rate": 9.08888888888889e-07, "loss": 0.0002, "step": 4600 }, { "epoch": 39.02, "learning_rate": 8.533333333333334e-07, "loss": 0.0002, "step": 4625 }, { "epoch": 40.0, "learning_rate": 7.977777777777779e-07, "loss": 0.0002, "step": 4650 }, { "epoch": 40.01, "learning_rate": 7.422222222222223e-07, "loss": 0.0002, "step": 4675 }, { "epoch": 40.01, "learning_rate": 6.866666666666667e-07, "loss": 0.0002, "step": 4700 }, { "epoch": 40.02, "learning_rate": 6.311111111111112e-07, "loss": 0.0002, "step": 4725 }, { "epoch": 40.02, "learning_rate": 5.755555555555555e-07, "loss": 0.0002, "step": 4750 }, { "epoch": 41.0, "learning_rate": 5.2e-07, "loss": 0.0002, "step": 4775 }, { "epoch": 41.01, "learning_rate": 4.6444444444444446e-07, "loss": 0.0002, "step": 4800 }, { "epoch": 41.01, "learning_rate": 4.0888888888888897e-07, "loss": 0.0002, "step": 4825 }, { "epoch": 41.02, "learning_rate": 3.533333333333334e-07, "loss": 0.0002, "step": 4850 }, { "epoch": 42.0, "learning_rate": 2.977777777777778e-07, "loss": 0.0002, "step": 4875 }, { "epoch": 42.01, "learning_rate": 2.4222222222222224e-07, "loss": 0.0002, "step": 4900 }, { "epoch": 42.01, "learning_rate": 1.866666666666667e-07, "loss": 0.0002, "step": 4925 }, { "epoch": 42.02, "learning_rate": 1.3111111111111113e-07, "loss": 0.0002, "step": 4950 }, { "epoch": 42.02, "learning_rate": 7.555555555555556e-08, "loss": 0.0002, "step": 4975 }, { "epoch": 43.0, "learning_rate": 2e-08, "loss": 0.0002, "step": 5000 }, { "epoch": 43.0, "eval_loss": 0.3427734375, "eval_runtime": 1211.7822, "eval_samples_per_second": 3.185, "eval_steps_per_second": 0.1, "eval_wer": 17.804826268487723, "step": 5000 }, { "epoch": 43.0, "step": 5000, "total_flos": 9.21239635788101e+19, "train_loss": 0.01692909588813782, "train_runtime": 93041.5511, "train_samples_per_second": 3.439, "train_steps_per_second": 0.054 } ], "max_steps": 5000, "num_train_epochs": 9223372036854775807, "total_flos": 9.21239635788101e+19, "trial_name": null, "trial_params": null }