{ "best_metric": 6.598745817992301, "best_model_checkpoint": "./checkpoint-3000", "epoch": 11.0045, "global_step": 6000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 3.96e-07, "loss": 1.249, "step": 25 }, { "epoch": 0.01, "learning_rate": 8.46e-07, "loss": 1.0081, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.296e-06, "loss": 0.4711, "step": 75 }, { "epoch": 0.02, "learning_rate": 1.746e-06, "loss": 0.281, "step": 100 }, { "epoch": 0.02, "learning_rate": 2.196e-06, "loss": 0.219, "step": 125 }, { "epoch": 0.03, "learning_rate": 2.6459999999999997e-06, "loss": 0.2109, "step": 150 }, { "epoch": 0.03, "learning_rate": 3.0959999999999997e-06, "loss": 0.2049, "step": 175 }, { "epoch": 0.03, "learning_rate": 3.5460000000000004e-06, "loss": 0.1752, "step": 200 }, { "epoch": 0.04, "learning_rate": 3.996e-06, "loss": 0.1854, "step": 225 }, { "epoch": 0.04, "learning_rate": 4.446e-06, "loss": 0.172, "step": 250 }, { "epoch": 0.05, "learning_rate": 4.896e-06, "loss": 0.1769, "step": 275 }, { "epoch": 0.05, "learning_rate": 5.346e-06, "loss": 0.18, "step": 300 }, { "epoch": 0.05, "learning_rate": 5.796e-06, "loss": 0.1905, "step": 325 }, { "epoch": 0.06, "learning_rate": 6.246e-06, "loss": 0.1491, "step": 350 }, { "epoch": 0.06, "learning_rate": 6.696e-06, "loss": 0.1631, "step": 375 }, { "epoch": 0.07, "learning_rate": 7.146000000000001e-06, "loss": 0.1645, "step": 400 }, { "epoch": 0.07, "learning_rate": 7.596e-06, "loss": 0.1531, "step": 425 }, { "epoch": 0.07, "learning_rate": 8.046e-06, "loss": 0.1653, "step": 450 }, { "epoch": 0.08, "learning_rate": 8.496e-06, "loss": 0.1695, "step": 475 }, { "epoch": 0.08, "learning_rate": 8.946e-06, "loss": 0.1693, "step": 500 }, { "epoch": 0.09, "learning_rate": 8.964000000000001e-06, "loss": 0.1649, "step": 525 }, { "epoch": 1.0, "learning_rate": 8.92309090909091e-06, "loss": 0.1554, "step": 550 }, { "epoch": 1.01, "learning_rate": 8.882181818181818e-06, "loss": 0.1702, "step": 575 }, { "epoch": 1.01, "learning_rate": 8.841272727272727e-06, "loss": 0.1502, "step": 600 }, { "epoch": 1.01, "learning_rate": 8.800363636363636e-06, "loss": 0.1362, "step": 625 }, { "epoch": 1.02, "learning_rate": 8.759454545454546e-06, "loss": 0.1332, "step": 650 }, { "epoch": 1.02, "learning_rate": 8.718545454545455e-06, "loss": 0.1116, "step": 675 }, { "epoch": 1.03, "learning_rate": 8.677636363636364e-06, "loss": 0.1231, "step": 700 }, { "epoch": 1.03, "learning_rate": 8.636727272727272e-06, "loss": 0.1102, "step": 725 }, { "epoch": 1.03, "learning_rate": 8.595818181818183e-06, "loss": 0.0975, "step": 750 }, { "epoch": 1.04, "learning_rate": 8.554909090909091e-06, "loss": 0.1098, "step": 775 }, { "epoch": 1.04, "learning_rate": 8.514e-06, "loss": 0.0917, "step": 800 }, { "epoch": 1.05, "learning_rate": 8.473090909090909e-06, "loss": 0.0914, "step": 825 }, { "epoch": 1.05, "learning_rate": 8.43218181818182e-06, "loss": 0.0903, "step": 850 }, { "epoch": 1.06, "learning_rate": 8.391272727272728e-06, "loss": 0.088, "step": 875 }, { "epoch": 1.06, "learning_rate": 8.350363636363637e-06, "loss": 0.0816, "step": 900 }, { "epoch": 1.06, "learning_rate": 8.309454545454545e-06, "loss": 0.071, "step": 925 }, { "epoch": 1.07, "learning_rate": 8.268545454545454e-06, "loss": 0.0717, "step": 950 }, { "epoch": 1.07, "learning_rate": 8.227636363636365e-06, "loss": 0.0709, "step": 975 }, { "epoch": 1.08, "learning_rate": 8.186727272727273e-06, "loss": 0.0704, "step": 1000 }, { "epoch": 1.08, "learning_rate": 8.145818181818182e-06, "loss": 0.0695, "step": 1025 }, { "epoch": 1.08, "learning_rate": 8.10490909090909e-06, "loss": 0.0739, "step": 1050 }, { "epoch": 1.09, "learning_rate": 8.064000000000001e-06, "loss": 0.0689, "step": 1075 }, { "epoch": 2.0, "learning_rate": 8.02309090909091e-06, "loss": 0.0679, "step": 1100 }, { "epoch": 2.01, "learning_rate": 7.982181818181818e-06, "loss": 0.0706, "step": 1125 }, { "epoch": 2.01, "learning_rate": 7.941272727272727e-06, "loss": 0.0617, "step": 1150 }, { "epoch": 2.01, "learning_rate": 7.900363636363636e-06, "loss": 0.0568, "step": 1175 }, { "epoch": 2.02, "learning_rate": 7.859454545454546e-06, "loss": 0.0661, "step": 1200 }, { "epoch": 2.02, "learning_rate": 7.818545454545455e-06, "loss": 0.0474, "step": 1225 }, { "epoch": 2.03, "learning_rate": 7.777636363636364e-06, "loss": 0.0568, "step": 1250 }, { "epoch": 2.03, "learning_rate": 7.736727272727272e-06, "loss": 0.0495, "step": 1275 }, { "epoch": 2.04, "learning_rate": 7.695818181818183e-06, "loss": 0.0458, "step": 1300 }, { "epoch": 2.04, "learning_rate": 7.65490909090909e-06, "loss": 0.0484, "step": 1325 }, { "epoch": 2.04, "learning_rate": 7.614e-06, "loss": 0.0433, "step": 1350 }, { "epoch": 2.05, "learning_rate": 7.573090909090909e-06, "loss": 0.0494, "step": 1375 }, { "epoch": 2.05, "learning_rate": 7.5321818181818186e-06, "loss": 0.0397, "step": 1400 }, { "epoch": 2.06, "learning_rate": 7.491272727272728e-06, "loss": 0.0372, "step": 1425 }, { "epoch": 2.06, "learning_rate": 7.450363636363636e-06, "loss": 0.0425, "step": 1450 }, { "epoch": 2.06, "learning_rate": 7.4094545454545455e-06, "loss": 0.0348, "step": 1475 }, { "epoch": 2.07, "learning_rate": 7.368545454545455e-06, "loss": 0.0333, "step": 1500 }, { "epoch": 2.07, "eval_loss": 0.20725364983081818, "eval_runtime": 4481.6137, "eval_samples_per_second": 1.94, "eval_steps_per_second": 0.121, "eval_wer": 6.977017871253005, "step": 1500 }, { "epoch": 2.07, "learning_rate": 7.327636363636365e-06, "loss": 0.038, "step": 1525 }, { "epoch": 2.08, "learning_rate": 7.2867272727272725e-06, "loss": 0.0362, "step": 1550 }, { "epoch": 2.08, "learning_rate": 7.245818181818182e-06, "loss": 0.0398, "step": 1575 }, { "epoch": 2.09, "learning_rate": 7.204909090909092e-06, "loss": 0.038, "step": 1600 }, { "epoch": 2.09, "learning_rate": 7.164e-06, "loss": 0.0362, "step": 1625 }, { "epoch": 3.0, "learning_rate": 7.123090909090909e-06, "loss": 0.037, "step": 1650 }, { "epoch": 3.01, "learning_rate": 7.082181818181818e-06, "loss": 0.0303, "step": 1675 }, { "epoch": 3.01, "learning_rate": 7.041272727272727e-06, "loss": 0.0281, "step": 1700 }, { "epoch": 3.02, "learning_rate": 7.000363636363637e-06, "loss": 0.0278, "step": 1725 }, { "epoch": 3.02, "learning_rate": 6.959454545454545e-06, "loss": 0.0296, "step": 1750 }, { "epoch": 3.02, "learning_rate": 6.918545454545454e-06, "loss": 0.0279, "step": 1775 }, { "epoch": 3.03, "learning_rate": 6.877636363636364e-06, "loss": 0.025, "step": 1800 }, { "epoch": 3.03, "learning_rate": 6.8367272727272734e-06, "loss": 0.0265, "step": 1825 }, { "epoch": 3.04, "learning_rate": 6.795818181818182e-06, "loss": 0.0223, "step": 1850 }, { "epoch": 3.04, "learning_rate": 6.754909090909091e-06, "loss": 0.0247, "step": 1875 }, { "epoch": 3.05, "learning_rate": 6.7140000000000004e-06, "loss": 0.0189, "step": 1900 }, { "epoch": 3.05, "learning_rate": 6.673090909090909e-06, "loss": 0.0186, "step": 1925 }, { "epoch": 3.05, "learning_rate": 6.632181818181819e-06, "loss": 0.0228, "step": 1950 }, { "epoch": 3.06, "learning_rate": 6.591272727272727e-06, "loss": 0.0181, "step": 1975 }, { "epoch": 3.06, "learning_rate": 6.550363636363636e-06, "loss": 0.0181, "step": 2000 }, { "epoch": 3.07, "learning_rate": 6.509454545454546e-06, "loss": 0.0164, "step": 2025 }, { "epoch": 3.07, "learning_rate": 6.468545454545455e-06, "loss": 0.0189, "step": 2050 }, { "epoch": 3.07, "learning_rate": 6.427636363636363e-06, "loss": 0.0168, "step": 2075 }, { "epoch": 3.08, "learning_rate": 6.386727272727273e-06, "loss": 0.0209, "step": 2100 }, { "epoch": 3.08, "learning_rate": 6.345818181818182e-06, "loss": 0.0157, "step": 2125 }, { "epoch": 3.09, "learning_rate": 6.304909090909092e-06, "loss": 0.0175, "step": 2150 }, { "epoch": 4.0, "learning_rate": 6.264e-06, "loss": 0.0164, "step": 2175 }, { "epoch": 4.0, "learning_rate": 6.223090909090909e-06, "loss": 0.0149, "step": 2200 }, { "epoch": 4.01, "learning_rate": 6.182181818181819e-06, "loss": 0.0199, "step": 2225 }, { "epoch": 4.01, "learning_rate": 6.1412727272727275e-06, "loss": 0.0127, "step": 2250 }, { "epoch": 4.02, "learning_rate": 6.100363636363636e-06, "loss": 0.0114, "step": 2275 }, { "epoch": 4.02, "learning_rate": 6.059454545454546e-06, "loss": 0.0129, "step": 2300 }, { "epoch": 4.03, "learning_rate": 6.0185454545454545e-06, "loss": 0.012, "step": 2325 }, { "epoch": 4.03, "learning_rate": 5.977636363636364e-06, "loss": 0.0125, "step": 2350 }, { "epoch": 4.03, "learning_rate": 5.936727272727274e-06, "loss": 0.0136, "step": 2375 }, { "epoch": 4.04, "learning_rate": 5.8958181818181815e-06, "loss": 0.0148, "step": 2400 }, { "epoch": 4.04, "learning_rate": 5.854909090909091e-06, "loss": 0.0076, "step": 2425 }, { "epoch": 4.05, "learning_rate": 5.814000000000001e-06, "loss": 0.0111, "step": 2450 }, { "epoch": 4.05, "learning_rate": 5.773090909090909e-06, "loss": 0.0139, "step": 2475 }, { "epoch": 4.05, "learning_rate": 5.732181818181818e-06, "loss": 0.0117, "step": 2500 }, { "epoch": 4.06, "learning_rate": 5.6912727272727276e-06, "loss": 0.0144, "step": 2525 }, { "epoch": 4.06, "learning_rate": 5.650363636363636e-06, "loss": 0.0086, "step": 2550 }, { "epoch": 4.07, "learning_rate": 5.609454545454546e-06, "loss": 0.0119, "step": 2575 }, { "epoch": 4.07, "learning_rate": 5.5685454545454545e-06, "loss": 0.0075, "step": 2600 }, { "epoch": 4.08, "learning_rate": 5.527636363636363e-06, "loss": 0.0091, "step": 2625 }, { "epoch": 4.08, "learning_rate": 5.486727272727273e-06, "loss": 0.0091, "step": 2650 }, { "epoch": 4.08, "learning_rate": 5.445818181818182e-06, "loss": 0.0076, "step": 2675 }, { "epoch": 4.09, "learning_rate": 5.40490909090909e-06, "loss": 0.0085, "step": 2700 }, { "epoch": 5.0, "learning_rate": 5.364e-06, "loss": 0.0083, "step": 2725 }, { "epoch": 5.01, "learning_rate": 5.323090909090909e-06, "loss": 0.0077, "step": 2750 }, { "epoch": 5.01, "learning_rate": 5.282181818181819e-06, "loss": 0.0071, "step": 2775 }, { "epoch": 5.01, "learning_rate": 5.241272727272728e-06, "loss": 0.0068, "step": 2800 }, { "epoch": 5.02, "learning_rate": 5.200363636363636e-06, "loss": 0.0076, "step": 2825 }, { "epoch": 5.02, "learning_rate": 5.159454545454546e-06, "loss": 0.0062, "step": 2850 }, { "epoch": 5.03, "learning_rate": 5.118545454545455e-06, "loss": 0.0078, "step": 2875 }, { "epoch": 5.03, "learning_rate": 5.077636363636364e-06, "loss": 0.0057, "step": 2900 }, { "epoch": 5.04, "learning_rate": 5.036727272727273e-06, "loss": 0.0085, "step": 2925 }, { "epoch": 5.04, "learning_rate": 4.995818181818182e-06, "loss": 0.0065, "step": 2950 }, { "epoch": 5.04, "learning_rate": 4.954909090909091e-06, "loss": 0.0059, "step": 2975 }, { "epoch": 5.05, "learning_rate": 4.914000000000001e-06, "loss": 0.0061, "step": 3000 }, { "epoch": 5.05, "eval_loss": 0.2628188133239746, "eval_runtime": 4477.0926, "eval_samples_per_second": 1.942, "eval_steps_per_second": 0.122, "eval_wer": 6.598745817992301, "step": 3000 }, { "epoch": 5.05, "learning_rate": 4.873090909090909e-06, "loss": 0.0034, "step": 3025 }, { "epoch": 5.06, "learning_rate": 4.832181818181818e-06, "loss": 0.0071, "step": 3050 }, { "epoch": 5.06, "learning_rate": 4.791272727272728e-06, "loss": 0.0051, "step": 3075 }, { "epoch": 5.06, "learning_rate": 4.750363636363636e-06, "loss": 0.0042, "step": 3100 }, { "epoch": 5.07, "learning_rate": 4.709454545454545e-06, "loss": 0.0057, "step": 3125 }, { "epoch": 5.07, "learning_rate": 4.668545454545455e-06, "loss": 0.0039, "step": 3150 }, { "epoch": 5.08, "learning_rate": 4.627636363636363e-06, "loss": 0.0043, "step": 3175 }, { "epoch": 5.08, "learning_rate": 4.586727272727273e-06, "loss": 0.0053, "step": 3200 }, { "epoch": 5.08, "learning_rate": 4.5458181818181825e-06, "loss": 0.0044, "step": 3225 }, { "epoch": 5.09, "learning_rate": 4.50490909090909e-06, "loss": 0.0039, "step": 3250 }, { "epoch": 6.0, "learning_rate": 4.464e-06, "loss": 0.0038, "step": 3275 }, { "epoch": 6.01, "learning_rate": 4.4230909090909095e-06, "loss": 0.004, "step": 3300 }, { "epoch": 6.01, "learning_rate": 4.382181818181818e-06, "loss": 0.0031, "step": 3325 }, { "epoch": 6.02, "learning_rate": 4.341272727272728e-06, "loss": 0.0036, "step": 3350 }, { "epoch": 6.02, "learning_rate": 4.3003636363636365e-06, "loss": 0.0049, "step": 3375 }, { "epoch": 6.02, "learning_rate": 4.259454545454546e-06, "loss": 0.0024, "step": 3400 }, { "epoch": 6.03, "learning_rate": 4.218545454545455e-06, "loss": 0.0038, "step": 3425 }, { "epoch": 6.03, "learning_rate": 4.1776363636363635e-06, "loss": 0.0038, "step": 3450 }, { "epoch": 6.04, "learning_rate": 4.136727272727273e-06, "loss": 0.0023, "step": 3475 }, { "epoch": 6.04, "learning_rate": 4.095818181818182e-06, "loss": 0.0027, "step": 3500 }, { "epoch": 6.04, "learning_rate": 4.0549090909090905e-06, "loss": 0.0036, "step": 3525 }, { "epoch": 6.05, "learning_rate": 4.014e-06, "loss": 0.0041, "step": 3550 }, { "epoch": 6.05, "learning_rate": 3.973090909090909e-06, "loss": 0.0028, "step": 3575 }, { "epoch": 6.06, "learning_rate": 3.932181818181818e-06, "loss": 0.002, "step": 3600 }, { "epoch": 6.06, "learning_rate": 3.891272727272727e-06, "loss": 0.0027, "step": 3625 }, { "epoch": 6.07, "learning_rate": 3.8503636363636366e-06, "loss": 0.0022, "step": 3650 }, { "epoch": 6.07, "learning_rate": 3.8094545454545457e-06, "loss": 0.0028, "step": 3675 }, { "epoch": 6.07, "learning_rate": 3.7685454545454544e-06, "loss": 0.0039, "step": 3700 }, { "epoch": 6.08, "learning_rate": 3.727636363636364e-06, "loss": 0.0021, "step": 3725 }, { "epoch": 6.08, "learning_rate": 3.6867272727272727e-06, "loss": 0.0024, "step": 3750 }, { "epoch": 6.09, "learning_rate": 3.6458181818181822e-06, "loss": 0.0019, "step": 3775 }, { "epoch": 6.09, "learning_rate": 3.604909090909091e-06, "loss": 0.0017, "step": 3800 }, { "epoch": 7.0, "learning_rate": 3.564e-06, "loss": 0.0024, "step": 3825 }, { "epoch": 7.01, "learning_rate": 3.5230909090909092e-06, "loss": 0.0024, "step": 3850 }, { "epoch": 7.01, "learning_rate": 3.4821818181818184e-06, "loss": 0.002, "step": 3875 }, { "epoch": 7.02, "learning_rate": 3.441272727272727e-06, "loss": 0.0013, "step": 3900 }, { "epoch": 7.02, "learning_rate": 3.4003636363636366e-06, "loss": 0.0021, "step": 3925 }, { "epoch": 7.02, "learning_rate": 3.3594545454545454e-06, "loss": 0.0015, "step": 3950 }, { "epoch": 7.03, "learning_rate": 3.318545454545455e-06, "loss": 0.0017, "step": 3975 }, { "epoch": 7.03, "learning_rate": 3.2776363636363636e-06, "loss": 0.0023, "step": 4000 }, { "epoch": 7.04, "learning_rate": 3.2367272727272728e-06, "loss": 0.0022, "step": 4025 }, { "epoch": 7.04, "learning_rate": 3.195818181818182e-06, "loss": 0.0026, "step": 4050 }, { "epoch": 7.05, "learning_rate": 3.154909090909091e-06, "loss": 0.0014, "step": 4075 }, { "epoch": 7.05, "learning_rate": 3.1139999999999997e-06, "loss": 0.0021, "step": 4100 }, { "epoch": 7.05, "learning_rate": 3.0730909090909093e-06, "loss": 0.0018, "step": 4125 }, { "epoch": 7.06, "learning_rate": 3.0321818181818184e-06, "loss": 0.0009, "step": 4150 }, { "epoch": 7.06, "learning_rate": 2.991272727272727e-06, "loss": 0.0018, "step": 4175 }, { "epoch": 7.07, "learning_rate": 2.9503636363636367e-06, "loss": 0.0014, "step": 4200 }, { "epoch": 7.07, "learning_rate": 2.9094545454545454e-06, "loss": 0.0013, "step": 4225 }, { "epoch": 7.07, "learning_rate": 2.868545454545455e-06, "loss": 0.0011, "step": 4250 }, { "epoch": 7.08, "learning_rate": 2.8276363636363637e-06, "loss": 0.0023, "step": 4275 }, { "epoch": 7.08, "learning_rate": 2.786727272727273e-06, "loss": 0.002, "step": 4300 }, { "epoch": 7.09, "learning_rate": 2.745818181818182e-06, "loss": 0.0012, "step": 4325 }, { "epoch": 8.0, "learning_rate": 2.704909090909091e-06, "loss": 0.0009, "step": 4350 }, { "epoch": 8.01, "learning_rate": 2.664e-06, "loss": 0.0007, "step": 4375 }, { "epoch": 8.01, "learning_rate": 2.6230909090909094e-06, "loss": 0.001, "step": 4400 }, { "epoch": 8.01, "learning_rate": 2.582181818181818e-06, "loss": 0.0006, "step": 4425 }, { "epoch": 8.02, "learning_rate": 2.5412727272727272e-06, "loss": 0.0008, "step": 4450 }, { "epoch": 8.02, "learning_rate": 2.5003636363636364e-06, "loss": 0.0016, "step": 4475 }, { "epoch": 8.03, "learning_rate": 2.4594545454545455e-06, "loss": 0.0007, "step": 4500 }, { "epoch": 8.03, "eval_loss": 0.29601243138313293, "eval_runtime": 4468.4473, "eval_samples_per_second": 1.945, "eval_steps_per_second": 0.122, "eval_wer": 6.697937156402885, "step": 4500 }, { "epoch": 8.03, "learning_rate": 2.4185454545454542e-06, "loss": 0.0008, "step": 4525 }, { "epoch": 8.03, "learning_rate": 2.3776363636363638e-06, "loss": 0.0013, "step": 4550 }, { "epoch": 8.04, "learning_rate": 2.3367272727272725e-06, "loss": 0.0017, "step": 4575 }, { "epoch": 8.04, "learning_rate": 2.295818181818182e-06, "loss": 0.0006, "step": 4600 }, { "epoch": 8.05, "learning_rate": 2.254909090909091e-06, "loss": 0.0006, "step": 4625 }, { "epoch": 8.05, "learning_rate": 2.214e-06, "loss": 0.0006, "step": 4650 }, { "epoch": 8.06, "learning_rate": 2.173090909090909e-06, "loss": 0.0008, "step": 4675 }, { "epoch": 8.06, "learning_rate": 2.132181818181818e-06, "loss": 0.0007, "step": 4700 }, { "epoch": 8.06, "learning_rate": 2.0912727272727273e-06, "loss": 0.001, "step": 4725 }, { "epoch": 8.07, "learning_rate": 2.0503636363636364e-06, "loss": 0.0005, "step": 4750 }, { "epoch": 8.07, "learning_rate": 2.0094545454545456e-06, "loss": 0.0005, "step": 4775 }, { "epoch": 8.08, "learning_rate": 1.9685454545454543e-06, "loss": 0.0005, "step": 4800 }, { "epoch": 8.08, "learning_rate": 1.927636363636364e-06, "loss": 0.0006, "step": 4825 }, { "epoch": 8.08, "learning_rate": 1.886727272727273e-06, "loss": 0.0005, "step": 4850 }, { "epoch": 8.09, "learning_rate": 1.845818181818182e-06, "loss": 0.0005, "step": 4875 }, { "epoch": 9.0, "learning_rate": 1.804909090909091e-06, "loss": 0.0006, "step": 4900 }, { "epoch": 9.01, "learning_rate": 1.7640000000000002e-06, "loss": 0.0007, "step": 4925 }, { "epoch": 9.01, "learning_rate": 1.723090909090909e-06, "loss": 0.0005, "step": 4950 }, { "epoch": 9.01, "learning_rate": 1.6821818181818182e-06, "loss": 0.0014, "step": 4975 }, { "epoch": 9.02, "learning_rate": 1.6412727272727274e-06, "loss": 0.0005, "step": 5000 }, { "epoch": 9.02, "learning_rate": 1.6003636363636365e-06, "loss": 0.0004, "step": 5025 }, { "epoch": 9.03, "learning_rate": 1.5594545454545454e-06, "loss": 0.0005, "step": 5050 }, { "epoch": 9.03, "learning_rate": 1.5185454545454546e-06, "loss": 0.0004, "step": 5075 }, { "epoch": 9.04, "learning_rate": 1.4776363636363637e-06, "loss": 0.0009, "step": 5100 }, { "epoch": 9.04, "learning_rate": 1.4367272727272726e-06, "loss": 0.0004, "step": 5125 }, { "epoch": 9.04, "learning_rate": 1.3958181818181818e-06, "loss": 0.0004, "step": 5150 }, { "epoch": 9.05, "learning_rate": 1.354909090909091e-06, "loss": 0.0004, "step": 5175 }, { "epoch": 9.05, "learning_rate": 1.314e-06, "loss": 0.0006, "step": 5200 }, { "epoch": 9.06, "learning_rate": 1.273090909090909e-06, "loss": 0.0004, "step": 5225 }, { "epoch": 9.06, "learning_rate": 1.232181818181818e-06, "loss": 0.0009, "step": 5250 }, { "epoch": 9.06, "learning_rate": 1.1912727272727272e-06, "loss": 0.0004, "step": 5275 }, { "epoch": 9.07, "learning_rate": 1.1503636363636366e-06, "loss": 0.0004, "step": 5300 }, { "epoch": 9.07, "learning_rate": 1.1094545454545455e-06, "loss": 0.0004, "step": 5325 }, { "epoch": 9.08, "learning_rate": 1.0685454545454544e-06, "loss": 0.0004, "step": 5350 }, { "epoch": 9.08, "learning_rate": 1.0276363636363636e-06, "loss": 0.0004, "step": 5375 }, { "epoch": 9.09, "learning_rate": 9.86727272727273e-07, "loss": 0.0004, "step": 5400 }, { "epoch": 9.09, "learning_rate": 9.458181818181818e-07, "loss": 0.0004, "step": 5425 }, { "epoch": 10.0, "learning_rate": 9.04909090909091e-07, "loss": 0.0004, "step": 5450 }, { "epoch": 10.01, "learning_rate": 8.64e-07, "loss": 0.0004, "step": 5475 }, { "epoch": 10.01, "learning_rate": 8.230909090909091e-07, "loss": 0.0004, "step": 5500 }, { "epoch": 10.02, "learning_rate": 7.821818181818182e-07, "loss": 0.0004, "step": 5525 }, { "epoch": 10.02, "learning_rate": 7.412727272727273e-07, "loss": 0.0004, "step": 5550 }, { "epoch": 10.02, "learning_rate": 7.003636363636363e-07, "loss": 0.0004, "step": 5575 }, { "epoch": 10.03, "learning_rate": 6.594545454545454e-07, "loss": 0.0004, "step": 5600 }, { "epoch": 10.03, "learning_rate": 6.185454545454546e-07, "loss": 0.0005, "step": 5625 }, { "epoch": 10.04, "learning_rate": 5.776363636363637e-07, "loss": 0.0004, "step": 5650 }, { "epoch": 10.04, "learning_rate": 5.367272727272727e-07, "loss": 0.0004, "step": 5675 }, { "epoch": 10.04, "learning_rate": 4.958181818181818e-07, "loss": 0.0004, "step": 5700 }, { "epoch": 10.05, "learning_rate": 4.5490909090909094e-07, "loss": 0.0004, "step": 5725 }, { "epoch": 10.05, "learning_rate": 4.1400000000000003e-07, "loss": 0.0004, "step": 5750 }, { "epoch": 10.06, "learning_rate": 3.7309090909090906e-07, "loss": 0.0004, "step": 5775 }, { "epoch": 10.06, "learning_rate": 3.321818181818182e-07, "loss": 0.0004, "step": 5800 }, { "epoch": 10.07, "learning_rate": 2.912727272727273e-07, "loss": 0.0003, "step": 5825 }, { "epoch": 10.07, "learning_rate": 2.5036363636363636e-07, "loss": 0.0004, "step": 5850 }, { "epoch": 10.07, "learning_rate": 2.0945454545454544e-07, "loss": 0.0004, "step": 5875 }, { "epoch": 10.08, "learning_rate": 1.6854545454545455e-07, "loss": 0.0004, "step": 5900 }, { "epoch": 10.08, "learning_rate": 1.2763636363636364e-07, "loss": 0.0004, "step": 5925 }, { "epoch": 10.09, "learning_rate": 8.672727272727272e-08, "loss": 0.0004, "step": 5950 }, { "epoch": 11.0, "learning_rate": 4.581818181818182e-08, "loss": 0.0004, "step": 5975 }, { "epoch": 11.0, "learning_rate": 4.90909090909091e-09, "loss": 0.0004, "step": 6000 }, { "epoch": 11.0, "eval_loss": 0.32118794322013855, "eval_runtime": 4478.9685, "eval_samples_per_second": 1.941, "eval_steps_per_second": 0.121, "eval_wer": 6.6794438560212495, "step": 6000 }, { "epoch": 11.0, "step": 6000, "total_flos": 1.9595614224384e+20, "train_loss": 0.04298678698204458, "train_runtime": 52040.1074, "train_samples_per_second": 3.689, "train_steps_per_second": 0.115 } ], "max_steps": 6000, "num_train_epochs": 9223372036854775807, "total_flos": 1.9595614224384e+20, "trial_name": null, "trial_params": null }