{ "best_metric": 44.080171349061175, "best_model_checkpoint": "./whisper_tiny_cs/checkpoint-1000", "epoch": 65.57377049180327, "global_step": 3000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.22, "learning_rate": 2.5938375012788124e-06, "loss": 1.9088, "step": 10 }, { "epoch": 0.44, "learning_rate": 4.0137332755197495e-06, "loss": 1.6421, "step": 20 }, { "epoch": 0.66, "learning_rate": 4.716577826569394e-06, "loss": 1.3807, "step": 30 }, { "epoch": 0.87, "learning_rate": 5.187675002557625e-06, "loss": 1.1869, "step": 40 }, { "epoch": 1.09, "learning_rate": 5.542526105605248e-06, "loss": 1.0727, "step": 50 }, { "epoch": 1.31, "learning_rate": 5.827293423354003e-06, "loss": 0.9867, "step": 60 }, { "epoch": 1.53, "learning_rate": 6.065146451806229e-06, "loss": 0.9234, "step": 70 }, { "epoch": 1.75, "learning_rate": 6.2693786409359715e-06, "loss": 0.874, "step": 80 }, { "epoch": 1.97, "learning_rate": 6.448328170811893e-06, "loss": 0.849, "step": 90 }, { "epoch": 2.19, "learning_rate": 6.607570776798562e-06, "loss": 0.7855, "step": 100 }, { "epoch": 2.4, "learning_rate": 6.751019550882568e-06, "loss": 0.7623, "step": 110 }, { "epoch": 2.62, "learning_rate": 6.881526630756396e-06, "loss": 0.7337, "step": 120 }, { "epoch": 2.84, "learning_rate": 7.001235150391877e-06, "loss": 0.7062, "step": 130 }, { "epoch": 3.06, "learning_rate": 7.11179636123406e-06, "loss": 0.6861, "step": 140 }, { "epoch": 3.28, "learning_rate": 7.214509519281458e-06, "loss": 0.6584, "step": 150 }, { "epoch": 3.5, "learning_rate": 7.310415327848206e-06, "loss": 0.6371, "step": 160 }, { "epoch": 3.72, "learning_rate": 7.400360293466852e-06, "loss": 0.6267, "step": 170 }, { "epoch": 3.93, "learning_rate": 7.485042226047167e-06, "loss": 0.6112, "step": 180 }, { "epoch": 4.15, "learning_rate": 7.565043147393057e-06, "loss": 0.5846, "step": 190 }, { "epoch": 4.37, "learning_rate": 7.640853571188253e-06, "loss": 0.5748, "step": 200 }, { "epoch": 4.59, "learning_rate": 7.712890734563846e-06, "loss": 0.5518, "step": 210 }, { "epoch": 4.81, "learning_rate": 7.781512503836438e-06, "loss": 0.5368, "step": 220 }, { "epoch": 5.03, "learning_rate": 7.847028130491337e-06, "loss": 0.5437, "step": 230 }, { "epoch": 5.25, "learning_rate": 7.909706676567024e-06, "loss": 0.499, "step": 240 }, { "epoch": 5.46, "learning_rate": 7.969783690344598e-06, "loss": 0.5025, "step": 250 }, { "epoch": 5.68, "learning_rate": 8.027466551039499e-06, "loss": 0.4953, "step": 260 }, { "epoch": 5.9, "learning_rate": 8.082938788770226e-06, "loss": 0.4931, "step": 270 }, { "epoch": 6.12, "learning_rate": 8.13636360688406e-06, "loss": 0.4646, "step": 280 }, { "epoch": 6.34, "learning_rate": 8.187886777096811e-06, "loss": 0.4616, "step": 290 }, { "epoch": 6.56, "learning_rate": 8.237639036863129e-06, "loss": 0.4489, "step": 300 }, { "epoch": 6.78, "learning_rate": 8.285738088271934e-06, "loss": 0.4407, "step": 310 }, { "epoch": 6.99, "learning_rate": 8.332290275394681e-06, "loss": 0.4325, "step": 320 }, { "epoch": 7.21, "learning_rate": 8.377392000226465e-06, "loss": 0.4117, "step": 330 }, { "epoch": 7.43, "learning_rate": 8.421130924632814e-06, "loss": 0.4053, "step": 340 }, { "epoch": 7.65, "learning_rate": 8.463586995975924e-06, "loss": 0.3979, "step": 350 }, { "epoch": 7.87, "learning_rate": 8.504833326576252e-06, "loss": 0.4018, "step": 360 }, { "epoch": 8.09, "learning_rate": 8.544936951314702e-06, "loss": 0.3902, "step": 370 }, { "epoch": 8.31, "learning_rate": 8.583959483092205e-06, "loss": 0.3659, "step": 380 }, { "epoch": 8.52, "learning_rate": 8.621957682239185e-06, "loss": 0.3719, "step": 390 }, { "epoch": 8.74, "learning_rate": 8.658983953085041e-06, "loss": 0.3638, "step": 400 }, { "epoch": 8.96, "learning_rate": 8.695086778590648e-06, "loss": 0.3671, "step": 410 }, { "epoch": 9.18, "learning_rate": 8.730311102089144e-06, "loss": 0.3467, "step": 420 }, { "epoch": 9.4, "learning_rate": 8.76469866367573e-06, "loss": 0.3334, "step": 430 }, { "epoch": 9.62, "learning_rate": 8.798288297561955e-06, "loss": 0.3344, "step": 440 }, { "epoch": 9.84, "learning_rate": 8.831116195707142e-06, "loss": 0.3262, "step": 450 }, { "epoch": 10.05, "learning_rate": 8.863216142214784e-06, "loss": 0.3256, "step": 460 }, { "epoch": 10.27, "learning_rate": 8.8946197223e-06, "loss": 0.3059, "step": 470 }, { "epoch": 10.49, "learning_rate": 8.925356509068313e-06, "loss": 0.2991, "step": 480 }, { "epoch": 10.71, "learning_rate": 8.955454230874312e-06, "loss": 0.3073, "step": 490 }, { "epoch": 10.93, "learning_rate": 8.984938921633993e-06, "loss": 0.311, "step": 500 }, { "epoch": 11.15, "learning_rate": 9.013835056132664e-06, "loss": 0.281, "step": 510 }, { "epoch": 11.37, "learning_rate": 9.042165672090707e-06, "loss": 0.2794, "step": 520 }, { "epoch": 11.58, "learning_rate": 9.069952480512465e-06, "loss": 0.2811, "step": 530 }, { "epoch": 11.8, "learning_rate": 9.097215965642567e-06, "loss": 0.2774, "step": 540 }, { "epoch": 12.02, "learning_rate": 9.123975475682459e-06, "loss": 0.2753, "step": 550 }, { "epoch": 12.24, "learning_rate": 9.150249305273527e-06, "loss": 0.2582, "step": 560 }, { "epoch": 12.46, "learning_rate": 9.176054770627573e-06, "loss": 0.252, "step": 570 }, { "epoch": 12.68, "learning_rate": 9.201408278077374e-06, "loss": 0.25, "step": 580 }, { "epoch": 12.9, "learning_rate": 9.22632538672697e-06, "loss": 0.2559, "step": 590 }, { "epoch": 13.11, "learning_rate": 9.25082086580079e-06, "loss": 0.2397, "step": 600 }, { "epoch": 13.33, "learning_rate": 9.274908747220955e-06, "loss": 0.229, "step": 610 }, { "epoch": 13.55, "learning_rate": 9.298602373881418e-06, "loss": 0.2302, "step": 620 }, { "epoch": 13.77, "learning_rate": 9.321914444034767e-06, "loss": 0.2321, "step": 630 }, { "epoch": 13.99, "learning_rate": 9.344857052161382e-06, "loss": 0.2328, "step": 640 }, { "epoch": 14.21, "learning_rate": 9.367441726650282e-06, "loss": 0.2087, "step": 650 }, { "epoch": 14.43, "learning_rate": 9.389679464585535e-06, "loss": 0.2066, "step": 660 }, { "epoch": 14.64, "learning_rate": 9.411580763901005e-06, "loss": 0.2128, "step": 670 }, { "epoch": 14.86, "learning_rate": 9.433155653138788e-06, "loss": 0.2113, "step": 680 }, { "epoch": 15.08, "learning_rate": 9.454413719022506e-06, "loss": 0.2027, "step": 690 }, { "epoch": 15.3, "learning_rate": 9.475364132035208e-06, "loss": 0.1912, "step": 700 }, { "epoch": 15.52, "learning_rate": 9.496015670172681e-06, "loss": 0.1897, "step": 710 }, { "epoch": 15.74, "learning_rate": 9.516376741026186e-06, "loss": 0.1915, "step": 720 }, { "epoch": 15.96, "learning_rate": 9.536455402333646e-06, "loss": 0.1893, "step": 730 }, { "epoch": 16.17, "learning_rate": 9.556259381124999e-06, "loss": 0.1775, "step": 740 }, { "epoch": 16.39, "learning_rate": 9.575796091575562e-06, "loss": 0.1696, "step": 750 }, { "epoch": 16.61, "learning_rate": 9.59507265167069e-06, "loss": 0.168, "step": 760 }, { "epoch": 16.83, "learning_rate": 9.614095898775347e-06, "loss": 0.1755, "step": 770 }, { "epoch": 17.05, "learning_rate": 9.632872404193962e-06, "loss": 0.17, "step": 780 }, { "epoch": 17.27, "learning_rate": 9.651408486798028e-06, "loss": 0.152, "step": 790 }, { "epoch": 17.49, "learning_rate": 9.669710225792232e-06, "loss": 0.1567, "step": 800 }, { "epoch": 17.7, "learning_rate": 9.687783472683637e-06, "loss": 0.1571, "step": 810 }, { "epoch": 17.92, "learning_rate": 9.705633862512872e-06, "loss": 0.1555, "step": 820 }, { "epoch": 18.14, "learning_rate": 9.723266824401275e-06, "loss": 0.1457, "step": 830 }, { "epoch": 18.36, "learning_rate": 9.740687591463389e-06, "loss": 0.1401, "step": 840 }, { "epoch": 18.58, "learning_rate": 9.75790121013008e-06, "loss": 0.1371, "step": 850 }, { "epoch": 18.8, "learning_rate": 9.774912548923845e-06, "loss": 0.1392, "step": 860 }, { "epoch": 19.02, "learning_rate": 9.791726306724489e-06, "loss": 0.1429, "step": 870 }, { "epoch": 19.23, "learning_rate": 9.80834702056027e-06, "loss": 0.1246, "step": 880 }, { "epoch": 19.45, "learning_rate": 9.824779072956837e-06, "loss": 0.1228, "step": 890 }, { "epoch": 19.67, "learning_rate": 9.841026698873751e-06, "loss": 0.1253, "step": 900 }, { "epoch": 19.89, "learning_rate": 9.857093992256045e-06, "loss": 0.1258, "step": 910 }, { "epoch": 20.11, "learning_rate": 9.87298491222617e-06, "loss": 0.1209, "step": 920 }, { "epoch": 20.33, "learning_rate": 9.888703288939783e-06, "loss": 0.109, "step": 930 }, { "epoch": 20.55, "learning_rate": 9.904252829127018e-06, "loss": 0.1131, "step": 940 }, { "epoch": 20.77, "learning_rate": 9.919637121339311e-06, "loss": 0.1119, "step": 950 }, { "epoch": 20.98, "learning_rate": 9.934859640920334e-06, "loss": 0.1105, "step": 960 }, { "epoch": 21.2, "learning_rate": 9.949923754718312e-06, "loss": 0.0991, "step": 970 }, { "epoch": 21.42, "learning_rate": 9.96483272555564e-06, "loss": 0.0974, "step": 980 }, { "epoch": 21.64, "learning_rate": 9.979589716470704e-06, "loss": 0.0975, "step": 990 }, { "epoch": 21.86, "learning_rate": 9.994197794745664e-06, "loss": 0.1007, "step": 1000 }, { "epoch": 21.86, "eval_loss": 0.54296875, "eval_runtime": 99.3921, "eval_samples_per_second": 88.83, "eval_steps_per_second": 1.851, "eval_wer": 44.080171349061175, "step": 1000 }, { "epoch": 22.08, "learning_rate": 9.975000000000002e-06, "loss": 0.0972, "step": 1010 }, { "epoch": 22.3, "learning_rate": 9.925e-06, "loss": 0.0862, "step": 1020 }, { "epoch": 22.51, "learning_rate": 9.875000000000001e-06, "loss": 0.0875, "step": 1030 }, { "epoch": 22.73, "learning_rate": 9.825000000000002e-06, "loss": 0.0873, "step": 1040 }, { "epoch": 22.95, "learning_rate": 9.775e-06, "loss": 0.0884, "step": 1050 }, { "epoch": 23.17, "learning_rate": 9.725000000000001e-06, "loss": 0.075, "step": 1060 }, { "epoch": 23.39, "learning_rate": 9.675000000000001e-06, "loss": 0.0751, "step": 1070 }, { "epoch": 23.61, "learning_rate": 9.625e-06, "loss": 0.0772, "step": 1080 }, { "epoch": 23.83, "learning_rate": 9.575e-06, "loss": 0.0782, "step": 1090 }, { "epoch": 24.04, "learning_rate": 9.525000000000001e-06, "loss": 0.0762, "step": 1100 }, { "epoch": 24.26, "learning_rate": 9.475000000000002e-06, "loss": 0.0653, "step": 1110 }, { "epoch": 24.48, "learning_rate": 9.425e-06, "loss": 0.0668, "step": 1120 }, { "epoch": 24.7, "learning_rate": 9.375000000000001e-06, "loss": 0.0678, "step": 1130 }, { "epoch": 24.92, "learning_rate": 9.325000000000001e-06, "loss": 0.0671, "step": 1140 }, { "epoch": 25.14, "learning_rate": 9.275e-06, "loss": 0.0636, "step": 1150 }, { "epoch": 25.36, "learning_rate": 9.225e-06, "loss": 0.0565, "step": 1160 }, { "epoch": 25.57, "learning_rate": 9.175000000000001e-06, "loss": 0.0576, "step": 1170 }, { "epoch": 25.79, "learning_rate": 9.125e-06, "loss": 0.0593, "step": 1180 }, { "epoch": 26.01, "learning_rate": 9.075e-06, "loss": 0.0591, "step": 1190 }, { "epoch": 26.23, "learning_rate": 9.025e-06, "loss": 0.049, "step": 1200 }, { "epoch": 26.45, "learning_rate": 8.975e-06, "loss": 0.0518, "step": 1210 }, { "epoch": 26.67, "learning_rate": 8.925e-06, "loss": 0.0502, "step": 1220 }, { "epoch": 26.89, "learning_rate": 8.875e-06, "loss": 0.0523, "step": 1230 }, { "epoch": 27.1, "learning_rate": 8.825000000000001e-06, "loss": 0.0486, "step": 1240 }, { "epoch": 27.32, "learning_rate": 8.775e-06, "loss": 0.0437, "step": 1250 }, { "epoch": 27.54, "learning_rate": 8.725000000000002e-06, "loss": 0.0439, "step": 1260 }, { "epoch": 27.76, "learning_rate": 8.675e-06, "loss": 0.0451, "step": 1270 }, { "epoch": 27.98, "learning_rate": 8.625000000000001e-06, "loss": 0.0458, "step": 1280 }, { "epoch": 28.2, "learning_rate": 8.575000000000002e-06, "loss": 0.0404, "step": 1290 }, { "epoch": 28.42, "learning_rate": 8.525e-06, "loss": 0.0393, "step": 1300 }, { "epoch": 28.63, "learning_rate": 8.475000000000001e-06, "loss": 0.0392, "step": 1310 }, { "epoch": 28.85, "learning_rate": 8.425000000000001e-06, "loss": 0.0395, "step": 1320 }, { "epoch": 29.07, "learning_rate": 8.375e-06, "loss": 0.0386, "step": 1330 }, { "epoch": 29.29, "learning_rate": 8.325e-06, "loss": 0.0351, "step": 1340 }, { "epoch": 29.51, "learning_rate": 8.275000000000001e-06, "loss": 0.0353, "step": 1350 }, { "epoch": 29.73, "learning_rate": 8.225e-06, "loss": 0.0357, "step": 1360 }, { "epoch": 29.95, "learning_rate": 8.175e-06, "loss": 0.0353, "step": 1370 }, { "epoch": 30.16, "learning_rate": 8.125000000000001e-06, "loss": 0.033, "step": 1380 }, { "epoch": 30.38, "learning_rate": 8.075000000000001e-06, "loss": 0.0311, "step": 1390 }, { "epoch": 30.6, "learning_rate": 8.025e-06, "loss": 0.0318, "step": 1400 }, { "epoch": 30.82, "learning_rate": 7.975e-06, "loss": 0.0319, "step": 1410 }, { "epoch": 31.04, "learning_rate": 7.925000000000001e-06, "loss": 0.0314, "step": 1420 }, { "epoch": 31.26, "learning_rate": 7.875e-06, "loss": 0.0285, "step": 1430 }, { "epoch": 31.48, "learning_rate": 7.825e-06, "loss": 0.0286, "step": 1440 }, { "epoch": 31.69, "learning_rate": 7.775000000000001e-06, "loss": 0.0294, "step": 1450 }, { "epoch": 31.91, "learning_rate": 7.725e-06, "loss": 0.0292, "step": 1460 }, { "epoch": 32.13, "learning_rate": 7.675e-06, "loss": 0.0269, "step": 1470 }, { "epoch": 32.35, "learning_rate": 7.625e-06, "loss": 0.0261, "step": 1480 }, { "epoch": 32.57, "learning_rate": 7.575e-06, "loss": 0.0262, "step": 1490 }, { "epoch": 32.79, "learning_rate": 7.525e-06, "loss": 0.0264, "step": 1500 }, { "epoch": 33.01, "learning_rate": 7.475000000000001e-06, "loss": 0.0264, "step": 1510 }, { "epoch": 33.22, "learning_rate": 7.425000000000001e-06, "loss": 0.0239, "step": 1520 }, { "epoch": 33.44, "learning_rate": 7.375000000000001e-06, "loss": 0.024, "step": 1530 }, { "epoch": 33.66, "learning_rate": 7.325000000000001e-06, "loss": 0.0244, "step": 1540 }, { "epoch": 33.88, "learning_rate": 7.275000000000001e-06, "loss": 0.0242, "step": 1550 }, { "epoch": 34.1, "learning_rate": 7.225000000000001e-06, "loss": 0.0235, "step": 1560 }, { "epoch": 34.32, "learning_rate": 7.175000000000001e-06, "loss": 0.0221, "step": 1570 }, { "epoch": 34.54, "learning_rate": 7.125e-06, "loss": 0.0221, "step": 1580 }, { "epoch": 34.75, "learning_rate": 7.075000000000001e-06, "loss": 0.0222, "step": 1590 }, { "epoch": 34.97, "learning_rate": 7.0250000000000005e-06, "loss": 0.0223, "step": 1600 }, { "epoch": 35.19, "learning_rate": 6.975000000000001e-06, "loss": 0.0205, "step": 1610 }, { "epoch": 35.41, "learning_rate": 6.925000000000001e-06, "loss": 0.0204, "step": 1620 }, { "epoch": 35.63, "learning_rate": 6.875e-06, "loss": 0.0207, "step": 1630 }, { "epoch": 35.85, "learning_rate": 6.825000000000001e-06, "loss": 0.0208, "step": 1640 }, { "epoch": 36.07, "learning_rate": 6.775e-06, "loss": 0.0202, "step": 1650 }, { "epoch": 36.28, "learning_rate": 6.725000000000001e-06, "loss": 0.0194, "step": 1660 }, { "epoch": 36.5, "learning_rate": 6.6750000000000005e-06, "loss": 0.0194, "step": 1670 }, { "epoch": 36.72, "learning_rate": 6.625e-06, "loss": 0.019, "step": 1680 }, { "epoch": 36.94, "learning_rate": 6.5750000000000006e-06, "loss": 0.0192, "step": 1690 }, { "epoch": 37.16, "learning_rate": 6.525e-06, "loss": 0.0183, "step": 1700 }, { "epoch": 37.38, "learning_rate": 6.475e-06, "loss": 0.0176, "step": 1710 }, { "epoch": 37.6, "learning_rate": 6.425e-06, "loss": 0.0179, "step": 1720 }, { "epoch": 37.81, "learning_rate": 6.375e-06, "loss": 0.0182, "step": 1730 }, { "epoch": 38.03, "learning_rate": 6.3250000000000004e-06, "loss": 0.018, "step": 1740 }, { "epoch": 38.25, "learning_rate": 6.275e-06, "loss": 0.0167, "step": 1750 }, { "epoch": 38.47, "learning_rate": 6.225000000000001e-06, "loss": 0.0167, "step": 1760 }, { "epoch": 38.69, "learning_rate": 6.175000000000001e-06, "loss": 0.0169, "step": 1770 }, { "epoch": 38.91, "learning_rate": 6.125000000000001e-06, "loss": 0.0168, "step": 1780 }, { "epoch": 39.13, "learning_rate": 6.075000000000001e-06, "loss": 0.0167, "step": 1790 }, { "epoch": 39.34, "learning_rate": 6.025000000000001e-06, "loss": 0.0157, "step": 1800 }, { "epoch": 39.56, "learning_rate": 5.975e-06, "loss": 0.016, "step": 1810 }, { "epoch": 39.78, "learning_rate": 5.925000000000001e-06, "loss": 0.0158, "step": 1820 }, { "epoch": 40.0, "learning_rate": 5.8750000000000005e-06, "loss": 0.016, "step": 1830 }, { "epoch": 40.22, "learning_rate": 5.825000000000001e-06, "loss": 0.0151, "step": 1840 }, { "epoch": 40.44, "learning_rate": 5.775000000000001e-06, "loss": 0.0151, "step": 1850 }, { "epoch": 40.66, "learning_rate": 5.725e-06, "loss": 0.0152, "step": 1860 }, { "epoch": 40.87, "learning_rate": 5.675000000000001e-06, "loss": 0.0148, "step": 1870 }, { "epoch": 41.09, "learning_rate": 5.625e-06, "loss": 0.0146, "step": 1880 }, { "epoch": 41.31, "learning_rate": 5.575000000000001e-06, "loss": 0.0144, "step": 1890 }, { "epoch": 41.53, "learning_rate": 5.5250000000000005e-06, "loss": 0.0142, "step": 1900 }, { "epoch": 41.75, "learning_rate": 5.475e-06, "loss": 0.0144, "step": 1910 }, { "epoch": 41.97, "learning_rate": 5.4250000000000006e-06, "loss": 0.0143, "step": 1920 }, { "epoch": 42.19, "learning_rate": 5.375e-06, "loss": 0.0136, "step": 1930 }, { "epoch": 42.4, "learning_rate": 5.325e-06, "loss": 0.0135, "step": 1940 }, { "epoch": 42.62, "learning_rate": 5.275e-06, "loss": 0.0136, "step": 1950 }, { "epoch": 42.84, "learning_rate": 5.225e-06, "loss": 0.0136, "step": 1960 }, { "epoch": 43.06, "learning_rate": 5.1750000000000004e-06, "loss": 0.0135, "step": 1970 }, { "epoch": 43.28, "learning_rate": 5.125e-06, "loss": 0.0129, "step": 1980 }, { "epoch": 43.5, "learning_rate": 5.075e-06, "loss": 0.0128, "step": 1990 }, { "epoch": 43.72, "learning_rate": 5.025e-06, "loss": 0.013, "step": 2000 }, { "epoch": 43.72, "eval_loss": 0.64892578125, "eval_runtime": 97.0258, "eval_samples_per_second": 90.996, "eval_steps_per_second": 1.896, "eval_wer": 44.918155808826974, "step": 2000 }, { "epoch": 43.93, "learning_rate": 4.975000000000001e-06, "loss": 0.0131, "step": 2010 }, { "epoch": 44.15, "learning_rate": 4.925e-06, "loss": 0.0127, "step": 2020 }, { "epoch": 44.37, "learning_rate": 4.875e-06, "loss": 0.0125, "step": 2030 }, { "epoch": 44.59, "learning_rate": 4.825e-06, "loss": 0.0125, "step": 2040 }, { "epoch": 44.81, "learning_rate": 4.775e-06, "loss": 0.0123, "step": 2050 }, { "epoch": 45.03, "learning_rate": 4.7250000000000005e-06, "loss": 0.0124, "step": 2060 }, { "epoch": 45.25, "learning_rate": 4.675000000000001e-06, "loss": 0.0118, "step": 2070 }, { "epoch": 45.46, "learning_rate": 4.625000000000001e-06, "loss": 0.0118, "step": 2080 }, { "epoch": 45.68, "learning_rate": 4.575e-06, "loss": 0.012, "step": 2090 }, { "epoch": 45.9, "learning_rate": 4.525000000000001e-06, "loss": 0.012, "step": 2100 }, { "epoch": 46.12, "learning_rate": 4.475e-06, "loss": 0.0117, "step": 2110 }, { "epoch": 46.34, "learning_rate": 4.425e-06, "loss": 0.0114, "step": 2120 }, { "epoch": 46.56, "learning_rate": 4.3750000000000005e-06, "loss": 0.0116, "step": 2130 }, { "epoch": 46.78, "learning_rate": 4.325e-06, "loss": 0.0115, "step": 2140 }, { "epoch": 46.99, "learning_rate": 4.2750000000000006e-06, "loss": 0.0115, "step": 2150 }, { "epoch": 47.21, "learning_rate": 4.225e-06, "loss": 0.0111, "step": 2160 }, { "epoch": 47.43, "learning_rate": 4.175e-06, "loss": 0.011, "step": 2170 }, { "epoch": 47.65, "learning_rate": 4.125e-06, "loss": 0.0111, "step": 2180 }, { "epoch": 47.87, "learning_rate": 4.075e-06, "loss": 0.0111, "step": 2190 }, { "epoch": 48.09, "learning_rate": 4.0250000000000004e-06, "loss": 0.0108, "step": 2200 }, { "epoch": 48.31, "learning_rate": 3.975000000000001e-06, "loss": 0.0106, "step": 2210 }, { "epoch": 48.52, "learning_rate": 3.9250000000000005e-06, "loss": 0.0107, "step": 2220 }, { "epoch": 48.74, "learning_rate": 3.875e-06, "loss": 0.0108, "step": 2230 }, { "epoch": 48.96, "learning_rate": 3.825000000000001e-06, "loss": 0.0107, "step": 2240 }, { "epoch": 49.18, "learning_rate": 3.7750000000000003e-06, "loss": 0.0104, "step": 2250 }, { "epoch": 49.4, "learning_rate": 3.7250000000000003e-06, "loss": 0.0103, "step": 2260 }, { "epoch": 49.62, "learning_rate": 3.6750000000000004e-06, "loss": 0.0102, "step": 2270 }, { "epoch": 49.84, "learning_rate": 3.625e-06, "loss": 0.0104, "step": 2280 }, { "epoch": 50.05, "learning_rate": 3.575e-06, "loss": 0.0102, "step": 2290 }, { "epoch": 50.27, "learning_rate": 3.525e-06, "loss": 0.0099, "step": 2300 }, { "epoch": 50.49, "learning_rate": 3.475e-06, "loss": 0.0101, "step": 2310 }, { "epoch": 50.71, "learning_rate": 3.4250000000000007e-06, "loss": 0.0101, "step": 2320 }, { "epoch": 50.93, "learning_rate": 3.3750000000000003e-06, "loss": 0.0102, "step": 2330 }, { "epoch": 51.15, "learning_rate": 3.3250000000000004e-06, "loss": 0.0099, "step": 2340 }, { "epoch": 51.37, "learning_rate": 3.2750000000000004e-06, "loss": 0.0096, "step": 2350 }, { "epoch": 51.58, "learning_rate": 3.2250000000000005e-06, "loss": 0.0098, "step": 2360 }, { "epoch": 51.8, "learning_rate": 3.175e-06, "loss": 0.0098, "step": 2370 }, { "epoch": 52.02, "learning_rate": 3.125e-06, "loss": 0.0097, "step": 2380 }, { "epoch": 52.24, "learning_rate": 3.075e-06, "loss": 0.0095, "step": 2390 }, { "epoch": 52.46, "learning_rate": 3.0250000000000003e-06, "loss": 0.0093, "step": 2400 }, { "epoch": 52.68, "learning_rate": 2.9750000000000003e-06, "loss": 0.0096, "step": 2410 }, { "epoch": 52.9, "learning_rate": 2.925e-06, "loss": 0.0097, "step": 2420 }, { "epoch": 53.11, "learning_rate": 2.875e-06, "loss": 0.0094, "step": 2430 }, { "epoch": 53.33, "learning_rate": 2.825e-06, "loss": 0.0092, "step": 2440 }, { "epoch": 53.55, "learning_rate": 2.7750000000000005e-06, "loss": 0.0094, "step": 2450 }, { "epoch": 53.77, "learning_rate": 2.7250000000000006e-06, "loss": 0.0091, "step": 2460 }, { "epoch": 53.99, "learning_rate": 2.6750000000000002e-06, "loss": 0.0094, "step": 2470 }, { "epoch": 54.21, "learning_rate": 2.6250000000000003e-06, "loss": 0.009, "step": 2480 }, { "epoch": 54.43, "learning_rate": 2.5750000000000003e-06, "loss": 0.0092, "step": 2490 }, { "epoch": 54.64, "learning_rate": 2.5250000000000004e-06, "loss": 0.0089, "step": 2500 }, { "epoch": 54.86, "learning_rate": 2.475e-06, "loss": 0.0091, "step": 2510 }, { "epoch": 55.08, "learning_rate": 2.425e-06, "loss": 0.009, "step": 2520 }, { "epoch": 55.3, "learning_rate": 2.375e-06, "loss": 0.0089, "step": 2530 }, { "epoch": 55.52, "learning_rate": 2.325e-06, "loss": 0.0088, "step": 2540 }, { "epoch": 55.74, "learning_rate": 2.2750000000000002e-06, "loss": 0.0089, "step": 2550 }, { "epoch": 55.96, "learning_rate": 2.2250000000000003e-06, "loss": 0.0089, "step": 2560 }, { "epoch": 56.17, "learning_rate": 2.1750000000000004e-06, "loss": 0.0087, "step": 2570 }, { "epoch": 56.39, "learning_rate": 2.125e-06, "loss": 0.0086, "step": 2580 }, { "epoch": 56.61, "learning_rate": 2.075e-06, "loss": 0.0087, "step": 2590 }, { "epoch": 56.83, "learning_rate": 2.025e-06, "loss": 0.0088, "step": 2600 }, { "epoch": 57.05, "learning_rate": 1.975e-06, "loss": 0.0086, "step": 2610 }, { "epoch": 57.27, "learning_rate": 1.925e-06, "loss": 0.0085, "step": 2620 }, { "epoch": 57.49, "learning_rate": 1.8750000000000003e-06, "loss": 0.0085, "step": 2630 }, { "epoch": 57.7, "learning_rate": 1.825e-06, "loss": 0.0086, "step": 2640 }, { "epoch": 57.92, "learning_rate": 1.7750000000000002e-06, "loss": 0.0086, "step": 2650 }, { "epoch": 58.14, "learning_rate": 1.725e-06, "loss": 0.0085, "step": 2660 }, { "epoch": 58.36, "learning_rate": 1.6750000000000003e-06, "loss": 0.0085, "step": 2670 }, { "epoch": 58.58, "learning_rate": 1.6250000000000001e-06, "loss": 0.0083, "step": 2680 }, { "epoch": 58.8, "learning_rate": 1.5750000000000002e-06, "loss": 0.0084, "step": 2690 }, { "epoch": 59.02, "learning_rate": 1.525e-06, "loss": 0.0084, "step": 2700 }, { "epoch": 59.23, "learning_rate": 1.475e-06, "loss": 0.0083, "step": 2710 }, { "epoch": 59.45, "learning_rate": 1.425e-06, "loss": 0.0083, "step": 2720 }, { "epoch": 59.67, "learning_rate": 1.3750000000000002e-06, "loss": 0.0083, "step": 2730 }, { "epoch": 59.89, "learning_rate": 1.3250000000000002e-06, "loss": 0.0084, "step": 2740 }, { "epoch": 60.11, "learning_rate": 1.275e-06, "loss": 0.0082, "step": 2750 }, { "epoch": 60.33, "learning_rate": 1.2250000000000001e-06, "loss": 0.0081, "step": 2760 }, { "epoch": 60.55, "learning_rate": 1.175e-06, "loss": 0.0083, "step": 2770 }, { "epoch": 60.77, "learning_rate": 1.125e-06, "loss": 0.0082, "step": 2780 }, { "epoch": 60.98, "learning_rate": 1.075e-06, "loss": 0.0083, "step": 2790 }, { "epoch": 61.2, "learning_rate": 1.025e-06, "loss": 0.0081, "step": 2800 }, { "epoch": 61.42, "learning_rate": 9.750000000000002e-07, "loss": 0.0081, "step": 2810 }, { "epoch": 61.64, "learning_rate": 9.25e-07, "loss": 0.0081, "step": 2820 }, { "epoch": 61.86, "learning_rate": 8.75e-07, "loss": 0.0081, "step": 2830 }, { "epoch": 62.08, "learning_rate": 8.250000000000001e-07, "loss": 0.008, "step": 2840 }, { "epoch": 62.3, "learning_rate": 7.750000000000001e-07, "loss": 0.008, "step": 2850 }, { "epoch": 62.51, "learning_rate": 7.25e-07, "loss": 0.0081, "step": 2860 }, { "epoch": 62.73, "learning_rate": 6.750000000000001e-07, "loss": 0.0081, "step": 2870 }, { "epoch": 62.95, "learning_rate": 6.25e-07, "loss": 0.008, "step": 2880 }, { "epoch": 63.17, "learning_rate": 5.750000000000001e-07, "loss": 0.008, "step": 2890 }, { "epoch": 63.39, "learning_rate": 5.250000000000001e-07, "loss": 0.008, "step": 2900 }, { "epoch": 63.61, "learning_rate": 4.7500000000000006e-07, "loss": 0.008, "step": 2910 }, { "epoch": 63.83, "learning_rate": 4.2500000000000006e-07, "loss": 0.008, "step": 2920 }, { "epoch": 64.04, "learning_rate": 3.75e-07, "loss": 0.008, "step": 2930 }, { "epoch": 64.26, "learning_rate": 3.25e-07, "loss": 0.008, "step": 2940 }, { "epoch": 64.48, "learning_rate": 2.75e-07, "loss": 0.0078, "step": 2950 }, { "epoch": 64.7, "learning_rate": 2.2500000000000002e-07, "loss": 0.008, "step": 2960 }, { "epoch": 64.92, "learning_rate": 1.7500000000000002e-07, "loss": 0.0081, "step": 2970 }, { "epoch": 65.14, "learning_rate": 1.2500000000000002e-07, "loss": 0.0079, "step": 2980 }, { "epoch": 65.36, "learning_rate": 7.500000000000001e-08, "loss": 0.0079, "step": 2990 }, { "epoch": 65.57, "learning_rate": 2.5000000000000002e-08, "loss": 0.0079, "step": 3000 }, { "epoch": 65.57, "eval_loss": 0.67822265625, "eval_runtime": 99.2826, "eval_samples_per_second": 88.928, "eval_steps_per_second": 1.853, "eval_wer": 45.51693948063724, "step": 3000 }, { "epoch": 65.57, "step": 3000, "total_flos": 2.8360954968696422e+19, "train_loss": 0.14609307404607536, "train_runtime": 15667.3381, "train_samples_per_second": 73.529, "train_steps_per_second": 0.191 } ], "max_steps": 3000, "num_train_epochs": 67, "total_flos": 2.8360954968696422e+19, "trial_name": null, "trial_params": null }