{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.18625442354255914, "global_step": 1000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1e-08, "loss": 24.0683, "step": 5 }, { "epoch": 0.0, "learning_rate": 3.5e-08, "loss": 17.9067, "step": 10 }, { "epoch": 0.0, "learning_rate": 6.000000000000001e-08, "loss": 16.9004, "step": 15 }, { "epoch": 0.0, "learning_rate": 8.500000000000001e-08, "loss": 16.8411, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.1e-07, "loss": 15.0697, "step": 25 }, { "epoch": 0.01, "learning_rate": 1.35e-07, "loss": 15.2061, "step": 30 }, { "epoch": 0.01, "learning_rate": 1.6e-07, "loss": 14.8504, "step": 35 }, { "epoch": 0.01, "learning_rate": 1.85e-07, "loss": 14.6816, "step": 40 }, { "epoch": 0.01, "learning_rate": 2.1000000000000003e-07, "loss": 14.9665, "step": 45 }, { "epoch": 0.01, "learning_rate": 2.3500000000000003e-07, "loss": 14.3427, "step": 50 }, { "epoch": 0.01, "learning_rate": 2.6e-07, "loss": 18.4574, "step": 55 }, { "epoch": 0.01, "learning_rate": 2.8e-07, "loss": 18.379, "step": 60 }, { "epoch": 0.01, "learning_rate": 3.0500000000000004e-07, "loss": 17.7823, "step": 65 }, { "epoch": 0.01, "learning_rate": 3.3e-07, "loss": 15.2998, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.55e-07, "loss": 15.2838, "step": 75 }, { "epoch": 0.01, "learning_rate": 3.8e-07, "loss": 14.8496, "step": 80 }, { "epoch": 0.02, "learning_rate": 4.0500000000000004e-07, "loss": 14.3004, "step": 85 }, { "epoch": 0.02, "learning_rate": 4.3e-07, "loss": 14.2005, "step": 90 }, { "epoch": 0.02, "learning_rate": 4.5500000000000004e-07, "loss": 14.5054, "step": 95 }, { "epoch": 0.02, "learning_rate": 4.800000000000001e-07, "loss": 14.3888, "step": 100 }, { "epoch": 0.02, "learning_rate": 5.05e-07, "loss": 17.3825, "step": 105 }, { "epoch": 0.02, "learning_rate": 5.3e-07, "loss": 17.0812, "step": 110 }, { "epoch": 0.02, "learning_rate": 5.550000000000001e-07, "loss": 17.0701, "step": 115 }, { "epoch": 0.02, "learning_rate": 5.800000000000001e-07, "loss": 14.9025, "step": 120 }, { "epoch": 0.02, "learning_rate": 6.05e-07, "loss": 14.4225, "step": 125 }, { "epoch": 0.02, "learning_rate": 6.3e-07, "loss": 13.9425, "step": 130 }, { "epoch": 0.03, "learning_rate": 6.550000000000001e-07, "loss": 13.8037, "step": 135 }, { "epoch": 0.03, "learning_rate": 6.800000000000001e-07, "loss": 13.4372, "step": 140 }, { "epoch": 0.03, "learning_rate": 7.05e-07, "loss": 13.6807, "step": 145 }, { "epoch": 0.03, "learning_rate": 7.3e-07, "loss": 13.742, "step": 150 }, { "epoch": 0.03, "learning_rate": 7.550000000000001e-07, "loss": 15.1545, "step": 155 }, { "epoch": 0.03, "learning_rate": 7.8e-07, "loss": 16.1792, "step": 160 }, { "epoch": 0.03, "learning_rate": 8.000000000000001e-07, "loss": 14.6018, "step": 165 }, { "epoch": 0.03, "learning_rate": 8.250000000000001e-07, "loss": 12.6032, "step": 170 }, { "epoch": 0.03, "learning_rate": 8.500000000000001e-07, "loss": 13.0217, "step": 175 }, { "epoch": 0.03, "learning_rate": 8.75e-07, "loss": 12.2336, "step": 180 }, { "epoch": 0.03, "learning_rate": 9.000000000000001e-07, "loss": 11.8643, "step": 185 }, { "epoch": 0.04, "learning_rate": 9.25e-07, "loss": 11.8572, "step": 190 }, { "epoch": 0.04, "learning_rate": 9.500000000000001e-07, "loss": 10.8976, "step": 195 }, { "epoch": 0.04, "learning_rate": 9.750000000000002e-07, "loss": 12.2617, "step": 200 }, { "epoch": 0.04, "learning_rate": 1.0000000000000002e-06, "loss": 13.7753, "step": 205 }, { "epoch": 0.04, "learning_rate": 1.025e-06, "loss": 11.6137, "step": 210 }, { "epoch": 0.04, "learning_rate": 1.0500000000000001e-06, "loss": 10.6402, "step": 215 }, { "epoch": 0.04, "learning_rate": 1.075e-06, "loss": 12.9591, "step": 220 }, { "epoch": 0.04, "learning_rate": 1.1e-06, "loss": 9.8613, "step": 225 }, { "epoch": 0.04, "learning_rate": 1.125e-06, "loss": 10.487, "step": 230 }, { "epoch": 0.04, "learning_rate": 1.1500000000000002e-06, "loss": 9.3773, "step": 235 }, { "epoch": 0.04, "learning_rate": 1.175e-06, "loss": 9.5665, "step": 240 }, { "epoch": 0.05, "learning_rate": 1.2000000000000002e-06, "loss": 8.2941, "step": 245 }, { "epoch": 0.05, "learning_rate": 1.2250000000000001e-06, "loss": 8.5563, "step": 250 }, { "epoch": 0.05, "learning_rate": 1.25e-06, "loss": 10.4941, "step": 255 }, { "epoch": 0.05, "learning_rate": 1.275e-06, "loss": 8.4986, "step": 260 }, { "epoch": 0.05, "learning_rate": 1.3e-06, "loss": 10.2295, "step": 265 }, { "epoch": 0.05, "learning_rate": 1.3250000000000002e-06, "loss": 8.7026, "step": 270 }, { "epoch": 0.05, "learning_rate": 1.3500000000000002e-06, "loss": 8.6009, "step": 275 }, { "epoch": 0.05, "learning_rate": 1.3750000000000002e-06, "loss": 7.7613, "step": 280 }, { "epoch": 0.05, "learning_rate": 1.4000000000000001e-06, "loss": 7.8609, "step": 285 }, { "epoch": 0.05, "learning_rate": 1.425e-06, "loss": 7.0097, "step": 290 }, { "epoch": 0.05, "learning_rate": 1.45e-06, "loss": 5.5692, "step": 295 }, { "epoch": 0.06, "learning_rate": 1.475e-06, "loss": 5.6402, "step": 300 }, { "epoch": 0.06, "learning_rate": 1.5e-06, "loss": 9.0815, "step": 305 }, { "epoch": 0.06, "learning_rate": 1.525e-06, "loss": 8.0803, "step": 310 }, { "epoch": 0.06, "learning_rate": 1.5500000000000002e-06, "loss": 7.6229, "step": 315 }, { "epoch": 0.06, "learning_rate": 1.5750000000000002e-06, "loss": 5.5354, "step": 320 }, { "epoch": 0.06, "learning_rate": 1.6000000000000001e-06, "loss": 8.1564, "step": 325 }, { "epoch": 0.06, "learning_rate": 1.6250000000000001e-06, "loss": 7.0378, "step": 330 }, { "epoch": 0.06, "learning_rate": 1.6500000000000003e-06, "loss": 5.3662, "step": 335 }, { "epoch": 0.06, "learning_rate": 1.6750000000000003e-06, "loss": 7.6436, "step": 340 }, { "epoch": 0.06, "learning_rate": 1.7000000000000002e-06, "loss": 5.3403, "step": 345 }, { "epoch": 0.07, "learning_rate": 1.725e-06, "loss": 8.1018, "step": 350 }, { "epoch": 0.07, "learning_rate": 1.745e-06, "loss": 7.5364, "step": 355 }, { "epoch": 0.07, "learning_rate": 1.77e-06, "loss": 5.2764, "step": 360 }, { "epoch": 0.07, "learning_rate": 1.7950000000000002e-06, "loss": 6.4889, "step": 365 }, { "epoch": 0.07, "learning_rate": 1.8200000000000002e-06, "loss": 5.1625, "step": 370 }, { "epoch": 0.07, "learning_rate": 1.8450000000000001e-06, "loss": 7.4417, "step": 375 }, { "epoch": 0.07, "learning_rate": 1.87e-06, "loss": 4.8924, "step": 380 }, { "epoch": 0.07, "learning_rate": 1.895e-06, "loss": 9.9698, "step": 385 }, { "epoch": 0.07, "learning_rate": 1.9200000000000003e-06, "loss": 7.4896, "step": 390 }, { "epoch": 0.07, "learning_rate": 1.945e-06, "loss": 6.7582, "step": 395 }, { "epoch": 0.07, "learning_rate": 1.97e-06, "loss": 6.0409, "step": 400 }, { "epoch": 0.08, "learning_rate": 1.9950000000000004e-06, "loss": 7.4142, "step": 405 }, { "epoch": 0.08, "learning_rate": 2.02e-06, "loss": 7.9007, "step": 410 }, { "epoch": 0.08, "learning_rate": 2.045e-06, "loss": 5.15, "step": 415 }, { "epoch": 0.08, "learning_rate": 2.07e-06, "loss": 5.6287, "step": 420 }, { "epoch": 0.08, "learning_rate": 2.0950000000000003e-06, "loss": 4.7277, "step": 425 }, { "epoch": 0.08, "learning_rate": 2.12e-06, "loss": 4.9955, "step": 430 }, { "epoch": 0.08, "learning_rate": 2.1450000000000002e-06, "loss": 6.4111, "step": 435 }, { "epoch": 0.08, "learning_rate": 2.17e-06, "loss": 5.7243, "step": 440 }, { "epoch": 0.08, "learning_rate": 2.195e-06, "loss": 5.3146, "step": 445 }, { "epoch": 0.08, "learning_rate": 2.2200000000000003e-06, "loss": 4.8272, "step": 450 }, { "epoch": 0.08, "learning_rate": 2.245e-06, "loss": 6.3851, "step": 455 }, { "epoch": 0.09, "learning_rate": 2.2700000000000003e-06, "loss": 5.0858, "step": 460 }, { "epoch": 0.09, "learning_rate": 2.2950000000000005e-06, "loss": 13.9773, "step": 465 }, { "epoch": 0.09, "learning_rate": 2.3200000000000002e-06, "loss": 5.1323, "step": 470 }, { "epoch": 0.09, "learning_rate": 2.345e-06, "loss": 4.5519, "step": 475 }, { "epoch": 0.09, "learning_rate": 2.37e-06, "loss": 4.9828, "step": 480 }, { "epoch": 0.09, "learning_rate": 2.395e-06, "loss": 10.9601, "step": 485 }, { "epoch": 0.09, "learning_rate": 2.42e-06, "loss": 4.715, "step": 490 }, { "epoch": 0.09, "learning_rate": 2.4450000000000003e-06, "loss": 6.1112, "step": 495 }, { "epoch": 0.09, "learning_rate": 2.47e-06, "loss": 4.8677, "step": 500 }, { "epoch": 0.09, "eval_loss": 6.384158611297607, "eval_runtime": 97.6612, "eval_samples_per_second": 12.226, "eval_steps_per_second": 1.536, "eval_wer": 1.0, "step": 500 }, { "epoch": 0.09, "learning_rate": 2.4950000000000003e-06, "loss": 5.6367, "step": 505 }, { "epoch": 0.09, "learning_rate": 2.52e-06, "loss": 4.798, "step": 510 }, { "epoch": 0.1, "learning_rate": 2.545e-06, "loss": 4.7321, "step": 515 }, { "epoch": 0.1, "learning_rate": 2.5700000000000004e-06, "loss": 9.0061, "step": 520 }, { "epoch": 0.1, "learning_rate": 2.595e-06, "loss": 4.5565, "step": 525 }, { "epoch": 0.1, "learning_rate": 2.6200000000000003e-06, "loss": 4.9119, "step": 530 }, { "epoch": 0.1, "learning_rate": 2.6450000000000005e-06, "loss": 4.6625, "step": 535 }, { "epoch": 0.1, "learning_rate": 2.6700000000000003e-06, "loss": 4.4122, "step": 540 }, { "epoch": 0.1, "learning_rate": 2.6950000000000005e-06, "loss": 6.515, "step": 545 }, { "epoch": 0.1, "learning_rate": 2.7200000000000002e-06, "loss": 6.2983, "step": 550 }, { "epoch": 0.1, "learning_rate": 2.7450000000000004e-06, "loss": 7.8164, "step": 555 }, { "epoch": 0.1, "learning_rate": 2.7700000000000006e-06, "loss": 5.5597, "step": 560 }, { "epoch": 0.11, "learning_rate": 2.7950000000000003e-06, "loss": 4.1535, "step": 565 }, { "epoch": 0.11, "learning_rate": 2.82e-06, "loss": 5.9433, "step": 570 }, { "epoch": 0.11, "learning_rate": 2.845e-06, "loss": 4.8775, "step": 575 }, { "epoch": 0.11, "learning_rate": 2.87e-06, "loss": 6.3474, "step": 580 }, { "epoch": 0.11, "learning_rate": 2.8950000000000002e-06, "loss": 8.6641, "step": 585 }, { "epoch": 0.11, "learning_rate": 2.92e-06, "loss": 6.0309, "step": 590 }, { "epoch": 0.11, "learning_rate": 2.945e-06, "loss": 4.1472, "step": 595 }, { "epoch": 0.11, "learning_rate": 2.97e-06, "loss": 4.5871, "step": 600 }, { "epoch": 0.11, "learning_rate": 2.995e-06, "loss": 5.0069, "step": 605 }, { "epoch": 0.11, "learning_rate": 3.0200000000000003e-06, "loss": 4.694, "step": 610 }, { "epoch": 0.11, "learning_rate": 3.045e-06, "loss": 4.502, "step": 615 }, { "epoch": 0.12, "learning_rate": 3.0700000000000003e-06, "loss": 7.2696, "step": 620 }, { "epoch": 0.12, "learning_rate": 3.0950000000000004e-06, "loss": 4.5793, "step": 625 }, { "epoch": 0.12, "learning_rate": 3.12e-06, "loss": 4.1937, "step": 630 }, { "epoch": 0.12, "learning_rate": 3.1450000000000004e-06, "loss": 6.3539, "step": 635 }, { "epoch": 0.12, "learning_rate": 3.17e-06, "loss": 4.1621, "step": 640 }, { "epoch": 0.12, "learning_rate": 3.1950000000000003e-06, "loss": 4.4824, "step": 645 }, { "epoch": 0.12, "learning_rate": 3.2200000000000005e-06, "loss": 4.4693, "step": 650 }, { "epoch": 0.12, "learning_rate": 3.2450000000000003e-06, "loss": 5.6584, "step": 655 }, { "epoch": 0.12, "learning_rate": 3.2700000000000005e-06, "loss": 4.813, "step": 660 }, { "epoch": 0.12, "learning_rate": 3.2950000000000002e-06, "loss": 6.1425, "step": 665 }, { "epoch": 0.12, "learning_rate": 3.3200000000000004e-06, "loss": 7.5331, "step": 670 }, { "epoch": 0.13, "learning_rate": 3.3450000000000006e-06, "loss": 9.772, "step": 675 }, { "epoch": 0.13, "learning_rate": 3.3700000000000003e-06, "loss": 4.1922, "step": 680 }, { "epoch": 0.13, "learning_rate": 3.3950000000000005e-06, "loss": 3.8468, "step": 685 }, { "epoch": 0.13, "learning_rate": 3.4200000000000007e-06, "loss": 3.5808, "step": 690 }, { "epoch": 0.13, "learning_rate": 3.445e-06, "loss": 5.5091, "step": 695 }, { "epoch": 0.13, "learning_rate": 3.4700000000000002e-06, "loss": 3.768, "step": 700 }, { "epoch": 0.13, "learning_rate": 3.495e-06, "loss": 4.9197, "step": 705 }, { "epoch": 0.13, "learning_rate": 3.52e-06, "loss": 4.1167, "step": 710 }, { "epoch": 0.13, "learning_rate": 3.545e-06, "loss": 7.1805, "step": 715 }, { "epoch": 0.13, "learning_rate": 3.57e-06, "loss": 7.7838, "step": 720 }, { "epoch": 0.14, "learning_rate": 3.5950000000000003e-06, "loss": 6.4851, "step": 725 }, { "epoch": 0.14, "learning_rate": 3.62e-06, "loss": 4.2357, "step": 730 }, { "epoch": 0.14, "learning_rate": 3.6450000000000003e-06, "loss": 5.7078, "step": 735 }, { "epoch": 0.14, "learning_rate": 3.6700000000000004e-06, "loss": 5.5984, "step": 740 }, { "epoch": 0.14, "learning_rate": 3.695e-06, "loss": 5.8425, "step": 745 }, { "epoch": 0.14, "learning_rate": 3.7200000000000004e-06, "loss": 3.9582, "step": 750 }, { "epoch": 0.14, "learning_rate": 3.745e-06, "loss": 6.1284, "step": 755 }, { "epoch": 0.14, "learning_rate": 3.7700000000000003e-06, "loss": 6.4174, "step": 760 }, { "epoch": 0.14, "learning_rate": 3.7950000000000005e-06, "loss": 4.0111, "step": 765 }, { "epoch": 0.14, "learning_rate": 3.820000000000001e-06, "loss": 6.0067, "step": 770 }, { "epoch": 0.14, "learning_rate": 3.8450000000000005e-06, "loss": 5.5284, "step": 775 }, { "epoch": 0.15, "learning_rate": 3.87e-06, "loss": 5.7256, "step": 780 }, { "epoch": 0.15, "learning_rate": 3.895000000000001e-06, "loss": 5.4687, "step": 785 }, { "epoch": 0.15, "learning_rate": 3.920000000000001e-06, "loss": 5.5711, "step": 790 }, { "epoch": 0.15, "learning_rate": 3.945e-06, "loss": 3.9495, "step": 795 }, { "epoch": 0.15, "learning_rate": 3.97e-06, "loss": 7.5433, "step": 800 }, { "epoch": 0.15, "learning_rate": 3.995000000000001e-06, "loss": 6.6596, "step": 805 }, { "epoch": 0.15, "learning_rate": 4.0200000000000005e-06, "loss": 3.8816, "step": 810 }, { "epoch": 0.15, "learning_rate": 4.045e-06, "loss": 6.0984, "step": 815 }, { "epoch": 0.15, "learning_rate": 4.07e-06, "loss": 3.6398, "step": 820 }, { "epoch": 0.15, "learning_rate": 4.095e-06, "loss": 6.2519, "step": 825 }, { "epoch": 0.15, "learning_rate": 4.12e-06, "loss": 5.4348, "step": 830 }, { "epoch": 0.16, "learning_rate": 4.145e-06, "loss": 5.7349, "step": 835 }, { "epoch": 0.16, "learning_rate": 4.17e-06, "loss": 5.6213, "step": 840 }, { "epoch": 0.16, "learning_rate": 4.1950000000000005e-06, "loss": 3.7084, "step": 845 }, { "epoch": 0.16, "learning_rate": 4.22e-06, "loss": 7.2329, "step": 850 }, { "epoch": 0.16, "learning_rate": 4.245e-06, "loss": 5.7584, "step": 855 }, { "epoch": 0.16, "learning_rate": 4.270000000000001e-06, "loss": 6.9726, "step": 860 }, { "epoch": 0.16, "learning_rate": 4.295e-06, "loss": 3.7044, "step": 865 }, { "epoch": 0.16, "learning_rate": 4.32e-06, "loss": 4.0139, "step": 870 }, { "epoch": 0.16, "learning_rate": 4.345000000000001e-06, "loss": 6.19, "step": 875 }, { "epoch": 0.16, "learning_rate": 4.3700000000000005e-06, "loss": 6.5308, "step": 880 }, { "epoch": 0.16, "learning_rate": 4.395e-06, "loss": 5.7261, "step": 885 }, { "epoch": 0.17, "learning_rate": 4.42e-06, "loss": 3.8616, "step": 890 }, { "epoch": 0.17, "learning_rate": 4.445000000000001e-06, "loss": 3.4939, "step": 895 }, { "epoch": 0.17, "learning_rate": 4.47e-06, "loss": 3.5178, "step": 900 }, { "epoch": 0.17, "learning_rate": 4.495e-06, "loss": 3.6419, "step": 905 }, { "epoch": 0.17, "learning_rate": 4.520000000000001e-06, "loss": 8.8566, "step": 910 }, { "epoch": 0.17, "learning_rate": 4.5450000000000005e-06, "loss": 3.7769, "step": 915 }, { "epoch": 0.17, "learning_rate": 4.57e-06, "loss": 3.6874, "step": 920 }, { "epoch": 0.17, "learning_rate": 4.595000000000001e-06, "loss": 5.7525, "step": 925 }, { "epoch": 0.17, "learning_rate": 4.620000000000001e-06, "loss": 5.7784, "step": 930 }, { "epoch": 0.17, "learning_rate": 4.645e-06, "loss": 5.0083, "step": 935 }, { "epoch": 0.18, "learning_rate": 4.670000000000001e-06, "loss": 3.4095, "step": 940 }, { "epoch": 0.18, "learning_rate": 4.695e-06, "loss": 3.212, "step": 945 }, { "epoch": 0.18, "learning_rate": 4.7200000000000005e-06, "loss": 5.4597, "step": 950 }, { "epoch": 0.18, "learning_rate": 4.745e-06, "loss": 3.7717, "step": 955 }, { "epoch": 0.18, "learning_rate": 4.77e-06, "loss": 3.3286, "step": 960 }, { "epoch": 0.18, "learning_rate": 4.795e-06, "loss": 3.3198, "step": 965 }, { "epoch": 0.18, "learning_rate": 4.8200000000000004e-06, "loss": 3.6952, "step": 970 }, { "epoch": 0.18, "learning_rate": 4.845e-06, "loss": 3.3483, "step": 975 }, { "epoch": 0.18, "learning_rate": 4.87e-06, "loss": 4.8449, "step": 980 }, { "epoch": 0.18, "learning_rate": 4.8950000000000006e-06, "loss": 3.5239, "step": 985 }, { "epoch": 0.18, "learning_rate": 4.92e-06, "loss": 5.7172, "step": 990 }, { "epoch": 0.19, "learning_rate": 4.945e-06, "loss": 3.9009, "step": 995 }, { "epoch": 0.19, "learning_rate": 4.970000000000001e-06, "loss": 3.293, "step": 1000 }, { "epoch": 0.19, "eval_loss": 5.126453876495361, "eval_runtime": 138.3538, "eval_samples_per_second": 8.63, "eval_steps_per_second": 1.084, "eval_wer": 1.0, "step": 1000 } ], "max_steps": 161070, "num_train_epochs": 30, "total_flos": 9.309229789594176e+16, "trial_name": null, "trial_params": null }