{ "best_metric": 4.976505766766339, "best_model_checkpoint": "./checkpoint-1000", "epoch": 70.54673721340389, "eval_steps": 1000, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.09, "learning_rate": 4.8e-08, "loss": 0.9708, "step": 25 }, { "epoch": 0.18, "learning_rate": 9.8e-08, "loss": 0.9737, "step": 50 }, { "epoch": 0.26, "learning_rate": 1.44e-07, "loss": 0.9134, "step": 75 }, { "epoch": 0.35, "learning_rate": 1.94e-07, "loss": 0.7953, "step": 100 }, { "epoch": 0.44, "learning_rate": 2.4399999999999996e-07, "loss": 0.4903, "step": 125 }, { "epoch": 0.53, "learning_rate": 2.9399999999999996e-07, "loss": 0.2757, "step": 150 }, { "epoch": 0.62, "learning_rate": 3.4399999999999996e-07, "loss": 0.2419, "step": 175 }, { "epoch": 0.71, "learning_rate": 3.94e-07, "loss": 0.2164, "step": 200 }, { "epoch": 0.79, "learning_rate": 4.44e-07, "loss": 0.2014, "step": 225 }, { "epoch": 0.88, "learning_rate": 4.94e-07, "loss": 0.1934, "step": 250 }, { "epoch": 0.97, "learning_rate": 5.44e-07, "loss": 0.1794, "step": 275 }, { "epoch": 1.06, "learning_rate": 5.939999999999999e-07, "loss": 0.1707, "step": 300 }, { "epoch": 1.15, "learning_rate": 6.44e-07, "loss": 0.1708, "step": 325 }, { "epoch": 1.23, "learning_rate": 6.939999999999999e-07, "loss": 0.1602, "step": 350 }, { "epoch": 1.32, "learning_rate": 7.44e-07, "loss": 0.1499, "step": 375 }, { "epoch": 1.41, "learning_rate": 7.94e-07, "loss": 0.1664, "step": 400 }, { "epoch": 1.5, "learning_rate": 8.439999999999999e-07, "loss": 0.1514, "step": 425 }, { "epoch": 1.59, "learning_rate": 8.939999999999999e-07, "loss": 0.1504, "step": 450 }, { "epoch": 1.68, "learning_rate": 9.439999999999999e-07, "loss": 0.1566, "step": 475 }, { "epoch": 1.76, "learning_rate": 9.94e-07, "loss": 0.1467, "step": 500 }, { "epoch": 1.85, "learning_rate": 9.98871794871795e-07, "loss": 0.1492, "step": 525 }, { "epoch": 1.94, "learning_rate": 9.975897435897437e-07, "loss": 0.1402, "step": 550 }, { "epoch": 2.03, "learning_rate": 9.963076923076924e-07, "loss": 0.139, "step": 575 }, { "epoch": 2.12, "learning_rate": 9.950256410256411e-07, "loss": 0.1155, "step": 600 }, { "epoch": 2.2, "learning_rate": 9.937435897435898e-07, "loss": 0.1199, "step": 625 }, { "epoch": 2.29, "learning_rate": 9.924615384615386e-07, "loss": 0.1228, "step": 650 }, { "epoch": 2.38, "learning_rate": 9.91179487179487e-07, "loss": 0.1166, "step": 675 }, { "epoch": 2.47, "learning_rate": 9.898974358974358e-07, "loss": 0.1153, "step": 700 }, { "epoch": 2.56, "learning_rate": 9.886153846153845e-07, "loss": 0.1201, "step": 725 }, { "epoch": 2.65, "learning_rate": 9.873333333333333e-07, "loss": 0.1184, "step": 750 }, { "epoch": 2.73, "learning_rate": 9.86051282051282e-07, "loss": 0.1068, "step": 775 }, { "epoch": 2.82, "learning_rate": 9.847692307692307e-07, "loss": 0.1115, "step": 800 }, { "epoch": 2.91, "learning_rate": 9.834871794871794e-07, "loss": 0.119, "step": 825 }, { "epoch": 3.0, "learning_rate": 9.822051282051282e-07, "loss": 0.1153, "step": 850 }, { "epoch": 3.09, "learning_rate": 9.809230769230769e-07, "loss": 0.0948, "step": 875 }, { "epoch": 3.17, "learning_rate": 9.796410256410256e-07, "loss": 0.091, "step": 900 }, { "epoch": 3.26, "learning_rate": 9.783589743589743e-07, "loss": 0.0956, "step": 925 }, { "epoch": 3.35, "learning_rate": 9.77128205128205e-07, "loss": 0.088, "step": 950 }, { "epoch": 3.44, "learning_rate": 9.758461538461538e-07, "loss": 0.0857, "step": 975 }, { "epoch": 3.53, "learning_rate": 9.745641025641025e-07, "loss": 0.0874, "step": 1000 }, { "epoch": 3.53, "eval_loss": 0.15932698547840118, "eval_runtime": 2459.4199, "eval_samples_per_second": 3.689, "eval_steps_per_second": 0.231, "eval_wer": 4.976505766766339, "step": 1000 }, { "epoch": 3.62, "learning_rate": 9.732820512820512e-07, "loss": 0.0867, "step": 1025 }, { "epoch": 3.7, "learning_rate": 9.72e-07, "loss": 0.0901, "step": 1050 }, { "epoch": 3.79, "learning_rate": 9.707179487179487e-07, "loss": 0.0886, "step": 1075 }, { "epoch": 3.88, "learning_rate": 9.694358974358974e-07, "loss": 0.0952, "step": 1100 }, { "epoch": 3.97, "learning_rate": 9.681538461538461e-07, "loss": 0.0878, "step": 1125 }, { "epoch": 4.06, "learning_rate": 9.668717948717949e-07, "loss": 0.0773, "step": 1150 }, { "epoch": 4.14, "learning_rate": 9.655897435897436e-07, "loss": 0.0664, "step": 1175 }, { "epoch": 4.23, "learning_rate": 9.643076923076923e-07, "loss": 0.067, "step": 1200 }, { "epoch": 4.32, "learning_rate": 9.63025641025641e-07, "loss": 0.0696, "step": 1225 }, { "epoch": 4.41, "learning_rate": 9.617435897435898e-07, "loss": 0.0711, "step": 1250 }, { "epoch": 4.5, "learning_rate": 9.604615384615385e-07, "loss": 0.0709, "step": 1275 }, { "epoch": 4.59, "learning_rate": 9.591794871794872e-07, "loss": 0.0697, "step": 1300 }, { "epoch": 4.67, "learning_rate": 9.57897435897436e-07, "loss": 0.072, "step": 1325 }, { "epoch": 4.76, "learning_rate": 9.566153846153847e-07, "loss": 0.0707, "step": 1350 }, { "epoch": 4.85, "learning_rate": 9.553333333333334e-07, "loss": 0.0691, "step": 1375 }, { "epoch": 4.94, "learning_rate": 9.541025641025641e-07, "loss": 0.0717, "step": 1400 }, { "epoch": 5.03, "learning_rate": 9.528205128205128e-07, "loss": 0.0672, "step": 1425 }, { "epoch": 5.11, "learning_rate": 9.515384615384615e-07, "loss": 0.0553, "step": 1450 }, { "epoch": 5.2, "learning_rate": 9.502564102564102e-07, "loss": 0.0534, "step": 1475 }, { "epoch": 5.29, "learning_rate": 9.490256410256409e-07, "loss": 0.053, "step": 1500 }, { "epoch": 5.38, "learning_rate": 9.477435897435897e-07, "loss": 0.0501, "step": 1525 }, { "epoch": 5.47, "learning_rate": 9.464615384615384e-07, "loss": 0.0494, "step": 1550 }, { "epoch": 5.56, "learning_rate": 9.451794871794871e-07, "loss": 0.0577, "step": 1575 }, { "epoch": 5.64, "learning_rate": 9.438974358974358e-07, "loss": 0.0523, "step": 1600 }, { "epoch": 5.73, "learning_rate": 9.426153846153846e-07, "loss": 0.0527, "step": 1625 }, { "epoch": 5.82, "learning_rate": 9.413333333333333e-07, "loss": 0.0542, "step": 1650 }, { "epoch": 5.91, "learning_rate": 9.40051282051282e-07, "loss": 0.0543, "step": 1675 }, { "epoch": 6.0, "learning_rate": 9.387692307692307e-07, "loss": 0.0557, "step": 1700 }, { "epoch": 6.08, "learning_rate": 9.374871794871795e-07, "loss": 0.0386, "step": 1725 }, { "epoch": 6.17, "learning_rate": 9.362051282051282e-07, "loss": 0.0361, "step": 1750 }, { "epoch": 6.26, "learning_rate": 9.349230769230769e-07, "loss": 0.036, "step": 1775 }, { "epoch": 6.35, "learning_rate": 9.336410256410256e-07, "loss": 0.0385, "step": 1800 }, { "epoch": 6.44, "learning_rate": 9.323589743589744e-07, "loss": 0.04, "step": 1825 }, { "epoch": 6.53, "learning_rate": 9.31076923076923e-07, "loss": 0.0388, "step": 1850 }, { "epoch": 6.61, "learning_rate": 9.297948717948717e-07, "loss": 0.0409, "step": 1875 }, { "epoch": 6.7, "learning_rate": 9.285128205128204e-07, "loss": 0.0421, "step": 1900 }, { "epoch": 6.79, "learning_rate": 9.272307692307692e-07, "loss": 0.0401, "step": 1925 }, { "epoch": 6.88, "learning_rate": 9.259487179487179e-07, "loss": 0.0354, "step": 1950 }, { "epoch": 6.97, "learning_rate": 9.246666666666666e-07, "loss": 0.0458, "step": 1975 }, { "epoch": 7.05, "learning_rate": 9.233846153846153e-07, "loss": 0.0318, "step": 2000 }, { "epoch": 7.05, "eval_loss": 0.22633017599582672, "eval_runtime": 2408.3448, "eval_samples_per_second": 3.767, "eval_steps_per_second": 0.235, "eval_wer": 5.436532711201656, "step": 2000 }, { "epoch": 7.14, "learning_rate": 9.221025641025641e-07, "loss": 0.0281, "step": 2025 }, { "epoch": 7.23, "learning_rate": 9.208205128205128e-07, "loss": 0.0273, "step": 2050 }, { "epoch": 7.32, "learning_rate": 9.195384615384615e-07, "loss": 0.0271, "step": 2075 }, { "epoch": 7.41, "learning_rate": 9.182564102564102e-07, "loss": 0.0283, "step": 2100 }, { "epoch": 7.5, "learning_rate": 9.16974358974359e-07, "loss": 0.0275, "step": 2125 }, { "epoch": 7.58, "learning_rate": 9.156923076923077e-07, "loss": 0.0265, "step": 2150 }, { "epoch": 7.67, "learning_rate": 9.144102564102564e-07, "loss": 0.0285, "step": 2175 }, { "epoch": 7.76, "learning_rate": 9.131282051282051e-07, "loss": 0.0271, "step": 2200 }, { "epoch": 7.85, "learning_rate": 9.118461538461539e-07, "loss": 0.0324, "step": 2225 }, { "epoch": 7.94, "learning_rate": 9.105641025641026e-07, "loss": 0.03, "step": 2250 }, { "epoch": 8.02, "learning_rate": 9.092820512820513e-07, "loss": 0.0254, "step": 2275 }, { "epoch": 8.11, "learning_rate": 9.08e-07, "loss": 0.0198, "step": 2300 }, { "epoch": 8.2, "learning_rate": 9.067179487179488e-07, "loss": 0.0212, "step": 2325 }, { "epoch": 8.29, "learning_rate": 9.054358974358975e-07, "loss": 0.0198, "step": 2350 }, { "epoch": 8.38, "learning_rate": 9.041538461538462e-07, "loss": 0.0177, "step": 2375 }, { "epoch": 8.47, "learning_rate": 9.028717948717949e-07, "loss": 0.02, "step": 2400 }, { "epoch": 8.55, "learning_rate": 9.015897435897435e-07, "loss": 0.0215, "step": 2425 }, { "epoch": 8.64, "learning_rate": 9.003076923076922e-07, "loss": 0.0198, "step": 2450 }, { "epoch": 8.73, "learning_rate": 8.990256410256409e-07, "loss": 0.0218, "step": 2475 }, { "epoch": 8.82, "learning_rate": 8.977435897435896e-07, "loss": 0.0207, "step": 2500 }, { "epoch": 8.91, "learning_rate": 8.964615384615383e-07, "loss": 0.0199, "step": 2525 }, { "epoch": 8.99, "learning_rate": 8.951794871794871e-07, "loss": 0.0219, "step": 2550 }, { "epoch": 9.08, "learning_rate": 8.938974358974358e-07, "loss": 0.0134, "step": 2575 }, { "epoch": 9.17, "learning_rate": 8.926153846153845e-07, "loss": 0.0153, "step": 2600 }, { "epoch": 9.26, "learning_rate": 8.913333333333332e-07, "loss": 0.0155, "step": 2625 }, { "epoch": 9.35, "learning_rate": 8.90051282051282e-07, "loss": 0.0142, "step": 2650 }, { "epoch": 9.44, "learning_rate": 8.887692307692307e-07, "loss": 0.0162, "step": 2675 }, { "epoch": 9.52, "learning_rate": 8.874871794871794e-07, "loss": 0.0156, "step": 2700 }, { "epoch": 9.61, "learning_rate": 8.862564102564102e-07, "loss": 0.0158, "step": 2725 }, { "epoch": 9.7, "learning_rate": 8.849743589743589e-07, "loss": 0.0169, "step": 2750 }, { "epoch": 9.79, "learning_rate": 8.836923076923076e-07, "loss": 0.0156, "step": 2775 }, { "epoch": 9.88, "learning_rate": 8.824102564102563e-07, "loss": 0.0158, "step": 2800 }, { "epoch": 9.96, "learning_rate": 8.811282051282051e-07, "loss": 0.0167, "step": 2825 }, { "epoch": 10.05, "learning_rate": 8.798461538461538e-07, "loss": 0.0144, "step": 2850 }, { "epoch": 10.14, "learning_rate": 8.785641025641025e-07, "loss": 0.01, "step": 2875 }, { "epoch": 10.23, "learning_rate": 8.772820512820512e-07, "loss": 0.0119, "step": 2900 }, { "epoch": 10.32, "learning_rate": 8.76e-07, "loss": 0.0138, "step": 2925 }, { "epoch": 10.41, "learning_rate": 8.747179487179487e-07, "loss": 0.0114, "step": 2950 }, { "epoch": 10.49, "learning_rate": 8.734358974358974e-07, "loss": 0.0114, "step": 2975 }, { "epoch": 10.58, "learning_rate": 8.721538461538461e-07, "loss": 0.0121, "step": 3000 }, { "epoch": 10.58, "eval_loss": 0.2965580224990845, "eval_runtime": 2304.8744, "eval_samples_per_second": 3.936, "eval_steps_per_second": 0.246, "eval_wer": 5.563040120921368, "step": 3000 }, { "epoch": 10.67, "learning_rate": 8.708717948717949e-07, "loss": 0.0124, "step": 3025 }, { "epoch": 10.76, "learning_rate": 8.695897435897436e-07, "loss": 0.0124, "step": 3050 }, { "epoch": 10.85, "learning_rate": 8.683076923076923e-07, "loss": 0.0123, "step": 3075 }, { "epoch": 10.93, "learning_rate": 8.67025641025641e-07, "loss": 0.0107, "step": 3100 }, { "epoch": 11.02, "learning_rate": 8.657435897435898e-07, "loss": 0.0117, "step": 3125 }, { "epoch": 11.11, "learning_rate": 8.644615384615385e-07, "loss": 0.0101, "step": 3150 }, { "epoch": 11.2, "learning_rate": 8.631794871794872e-07, "loss": 0.0082, "step": 3175 }, { "epoch": 11.29, "learning_rate": 8.618974358974359e-07, "loss": 0.009, "step": 3200 }, { "epoch": 11.38, "learning_rate": 8.606153846153845e-07, "loss": 0.0091, "step": 3225 }, { "epoch": 11.46, "learning_rate": 8.593333333333333e-07, "loss": 0.0111, "step": 3250 }, { "epoch": 11.55, "learning_rate": 8.58051282051282e-07, "loss": 0.0097, "step": 3275 }, { "epoch": 11.64, "learning_rate": 8.567692307692307e-07, "loss": 0.0091, "step": 3300 }, { "epoch": 11.73, "learning_rate": 8.554871794871794e-07, "loss": 0.0094, "step": 3325 }, { "epoch": 11.82, "learning_rate": 8.542051282051282e-07, "loss": 0.01, "step": 3350 }, { "epoch": 11.9, "learning_rate": 8.529230769230769e-07, "loss": 0.0096, "step": 3375 }, { "epoch": 11.99, "learning_rate": 8.516410256410256e-07, "loss": 0.0097, "step": 3400 }, { "epoch": 12.08, "learning_rate": 8.503589743589743e-07, "loss": 0.0078, "step": 3425 }, { "epoch": 12.17, "learning_rate": 8.490769230769231e-07, "loss": 0.0067, "step": 3450 }, { "epoch": 12.26, "learning_rate": 8.477948717948718e-07, "loss": 0.0074, "step": 3475 }, { "epoch": 12.35, "learning_rate": 8.465128205128205e-07, "loss": 0.0075, "step": 3500 }, { "epoch": 12.43, "learning_rate": 8.452307692307691e-07, "loss": 0.0074, "step": 3525 }, { "epoch": 12.52, "learning_rate": 8.439487179487179e-07, "loss": 0.0068, "step": 3550 }, { "epoch": 12.61, "learning_rate": 8.426666666666666e-07, "loss": 0.0092, "step": 3575 }, { "epoch": 12.7, "learning_rate": 8.413846153846153e-07, "loss": 0.0091, "step": 3600 }, { "epoch": 12.79, "learning_rate": 8.40102564102564e-07, "loss": 0.0084, "step": 3625 }, { "epoch": 12.87, "learning_rate": 8.388205128205128e-07, "loss": 0.0078, "step": 3650 }, { "epoch": 12.96, "learning_rate": 8.375384615384615e-07, "loss": 0.0079, "step": 3675 }, { "epoch": 13.05, "learning_rate": 8.362564102564102e-07, "loss": 0.0069, "step": 3700 }, { "epoch": 13.14, "learning_rate": 8.349743589743589e-07, "loss": 0.0057, "step": 3725 }, { "epoch": 13.23, "learning_rate": 8.336923076923077e-07, "loss": 0.0066, "step": 3750 }, { "epoch": 13.32, "learning_rate": 8.324102564102564e-07, "loss": 0.0063, "step": 3775 }, { "epoch": 13.4, "learning_rate": 8.311282051282051e-07, "loss": 0.0071, "step": 3800 }, { "epoch": 13.49, "learning_rate": 8.298461538461538e-07, "loss": 0.0064, "step": 3825 }, { "epoch": 13.58, "learning_rate": 8.285641025641025e-07, "loss": 0.0066, "step": 3850 }, { "epoch": 13.67, "learning_rate": 8.272820512820512e-07, "loss": 0.0073, "step": 3875 }, { "epoch": 13.76, "learning_rate": 8.259999999999999e-07, "loss": 0.0065, "step": 3900 }, { "epoch": 13.84, "learning_rate": 8.247179487179486e-07, "loss": 0.0075, "step": 3925 }, { "epoch": 13.93, "learning_rate": 8.234358974358974e-07, "loss": 0.0064, "step": 3950 }, { "epoch": 14.02, "learning_rate": 8.221538461538461e-07, "loss": 0.0062, "step": 3975 }, { "epoch": 14.11, "learning_rate": 8.208717948717948e-07, "loss": 0.005, "step": 4000 }, { "epoch": 14.11, "eval_loss": 0.3400271236896515, "eval_runtime": 2264.7792, "eval_samples_per_second": 4.006, "eval_steps_per_second": 0.25, "eval_wer": 5.612328722110867, "step": 4000 }, { "epoch": 14.2, "learning_rate": 8.195897435897435e-07, "loss": 0.0055, "step": 4025 }, { "epoch": 14.29, "learning_rate": 8.183076923076923e-07, "loss": 0.0057, "step": 4050 }, { "epoch": 14.37, "learning_rate": 8.17025641025641e-07, "loss": 0.0059, "step": 4075 }, { "epoch": 14.46, "learning_rate": 8.157435897435897e-07, "loss": 0.0065, "step": 4100 }, { "epoch": 14.55, "learning_rate": 8.144615384615384e-07, "loss": 0.0058, "step": 4125 }, { "epoch": 14.64, "learning_rate": 8.131794871794872e-07, "loss": 0.0056, "step": 4150 }, { "epoch": 14.73, "learning_rate": 8.118974358974359e-07, "loss": 0.0056, "step": 4175 }, { "epoch": 14.81, "learning_rate": 8.106153846153846e-07, "loss": 0.0063, "step": 4200 }, { "epoch": 14.9, "learning_rate": 8.093333333333333e-07, "loss": 0.0063, "step": 4225 }, { "epoch": 14.99, "learning_rate": 8.08051282051282e-07, "loss": 0.0059, "step": 4250 }, { "epoch": 15.08, "learning_rate": 8.067692307692308e-07, "loss": 0.0053, "step": 4275 }, { "epoch": 15.17, "learning_rate": 8.054871794871795e-07, "loss": 0.0054, "step": 4300 }, { "epoch": 15.26, "learning_rate": 8.042051282051282e-07, "loss": 0.0043, "step": 4325 }, { "epoch": 15.34, "learning_rate": 8.02923076923077e-07, "loss": 0.0045, "step": 4350 }, { "epoch": 15.43, "learning_rate": 8.016410256410257e-07, "loss": 0.0047, "step": 4375 }, { "epoch": 15.52, "learning_rate": 8.003589743589744e-07, "loss": 0.0045, "step": 4400 }, { "epoch": 15.61, "learning_rate": 7.990769230769231e-07, "loss": 0.0052, "step": 4425 }, { "epoch": 15.7, "learning_rate": 7.977948717948717e-07, "loss": 0.0053, "step": 4450 }, { "epoch": 15.78, "learning_rate": 7.965128205128205e-07, "loss": 0.0051, "step": 4475 }, { "epoch": 15.87, "learning_rate": 7.952307692307692e-07, "loss": 0.0056, "step": 4500 }, { "epoch": 15.96, "learning_rate": 7.939487179487179e-07, "loss": 0.0051, "step": 4525 }, { "epoch": 16.05, "learning_rate": 7.926666666666666e-07, "loss": 0.0053, "step": 4550 }, { "epoch": 16.14, "learning_rate": 7.913846153846154e-07, "loss": 0.0039, "step": 4575 }, { "epoch": 16.23, "learning_rate": 7.90102564102564e-07, "loss": 0.0046, "step": 4600 }, { "epoch": 16.31, "learning_rate": 7.888205128205127e-07, "loss": 0.0037, "step": 4625 }, { "epoch": 16.4, "learning_rate": 7.875384615384614e-07, "loss": 0.0049, "step": 4650 }, { "epoch": 16.49, "learning_rate": 7.862564102564102e-07, "loss": 0.0043, "step": 4675 }, { "epoch": 16.58, "learning_rate": 7.849743589743589e-07, "loss": 0.0041, "step": 4700 }, { "epoch": 16.67, "learning_rate": 7.836923076923076e-07, "loss": 0.0041, "step": 4725 }, { "epoch": 16.75, "learning_rate": 7.824102564102563e-07, "loss": 0.0053, "step": 4750 }, { "epoch": 16.84, "learning_rate": 7.811282051282051e-07, "loss": 0.0063, "step": 4775 }, { "epoch": 16.93, "learning_rate": 7.798461538461538e-07, "loss": 0.0058, "step": 4800 }, { "epoch": 17.02, "learning_rate": 7.785641025641025e-07, "loss": 0.0049, "step": 4825 }, { "epoch": 17.11, "learning_rate": 7.772820512820512e-07, "loss": 0.004, "step": 4850 }, { "epoch": 17.2, "learning_rate": 7.76e-07, "loss": 0.0039, "step": 4875 }, { "epoch": 17.28, "learning_rate": 7.747179487179487e-07, "loss": 0.0042, "step": 4900 }, { "epoch": 17.37, "learning_rate": 7.734358974358974e-07, "loss": 0.004, "step": 4925 }, { "epoch": 17.46, "learning_rate": 7.721538461538461e-07, "loss": 0.0038, "step": 4950 }, { "epoch": 17.55, "learning_rate": 7.708717948717949e-07, "loss": 0.004, "step": 4975 }, { "epoch": 17.64, "learning_rate": 7.695897435897436e-07, "loss": 0.0036, "step": 5000 }, { "epoch": 17.64, "eval_loss": 0.3554035723209381, "eval_runtime": 2261.6132, "eval_samples_per_second": 4.011, "eval_steps_per_second": 0.251, "eval_wer": 5.659974369927381, "step": 5000 }, { "epoch": 17.72, "learning_rate": 7.683076923076923e-07, "loss": 0.0034, "step": 5025 }, { "epoch": 17.81, "learning_rate": 7.67025641025641e-07, "loss": 0.0045, "step": 5050 }, { "epoch": 17.9, "learning_rate": 7.657435897435897e-07, "loss": 0.004, "step": 5075 }, { "epoch": 17.99, "learning_rate": 7.644615384615384e-07, "loss": 0.0052, "step": 5100 }, { "epoch": 18.08, "learning_rate": 7.631794871794871e-07, "loss": 0.0049, "step": 5125 }, { "epoch": 18.17, "learning_rate": 7.618974358974358e-07, "loss": 0.0037, "step": 5150 }, { "epoch": 18.25, "learning_rate": 7.606153846153846e-07, "loss": 0.0034, "step": 5175 }, { "epoch": 18.34, "learning_rate": 7.593333333333333e-07, "loss": 0.0034, "step": 5200 }, { "epoch": 18.43, "learning_rate": 7.58051282051282e-07, "loss": 0.0039, "step": 5225 }, { "epoch": 18.52, "learning_rate": 7.567692307692307e-07, "loss": 0.0034, "step": 5250 }, { "epoch": 18.61, "learning_rate": 7.554871794871795e-07, "loss": 0.0037, "step": 5275 }, { "epoch": 18.69, "learning_rate": 7.542051282051282e-07, "loss": 0.0038, "step": 5300 }, { "epoch": 18.78, "learning_rate": 7.529230769230769e-07, "loss": 0.0036, "step": 5325 }, { "epoch": 18.87, "learning_rate": 7.516410256410256e-07, "loss": 0.0036, "step": 5350 }, { "epoch": 18.96, "learning_rate": 7.503589743589744e-07, "loss": 0.0037, "step": 5375 }, { "epoch": 19.05, "learning_rate": 7.490769230769231e-07, "loss": 0.0039, "step": 5400 }, { "epoch": 19.14, "learning_rate": 7.477948717948718e-07, "loss": 0.0039, "step": 5425 }, { "epoch": 19.22, "learning_rate": 7.465128205128205e-07, "loss": 0.004, "step": 5450 }, { "epoch": 19.31, "learning_rate": 7.452307692307692e-07, "loss": 0.0039, "step": 5475 }, { "epoch": 19.4, "learning_rate": 7.43948717948718e-07, "loss": 0.0028, "step": 5500 }, { "epoch": 19.49, "learning_rate": 7.426666666666667e-07, "loss": 0.0034, "step": 5525 }, { "epoch": 19.58, "learning_rate": 7.413846153846154e-07, "loss": 0.0034, "step": 5550 }, { "epoch": 19.66, "learning_rate": 7.401025641025641e-07, "loss": 0.004, "step": 5575 }, { "epoch": 19.75, "learning_rate": 7.388205128205129e-07, "loss": 0.003, "step": 5600 }, { "epoch": 19.84, "learning_rate": 7.375384615384616e-07, "loss": 0.0028, "step": 5625 }, { "epoch": 19.93, "learning_rate": 7.362564102564103e-07, "loss": 0.0031, "step": 5650 }, { "epoch": 20.02, "learning_rate": 7.349743589743588e-07, "loss": 0.0041, "step": 5675 }, { "epoch": 20.11, "learning_rate": 7.336923076923076e-07, "loss": 0.0029, "step": 5700 }, { "epoch": 20.19, "learning_rate": 7.324102564102563e-07, "loss": 0.004, "step": 5725 }, { "epoch": 20.28, "learning_rate": 7.31128205128205e-07, "loss": 0.0034, "step": 5750 }, { "epoch": 20.37, "learning_rate": 7.298461538461537e-07, "loss": 0.0025, "step": 5775 }, { "epoch": 20.46, "learning_rate": 7.285641025641025e-07, "loss": 0.0036, "step": 5800 }, { "epoch": 20.55, "learning_rate": 7.272820512820512e-07, "loss": 0.0029, "step": 5825 }, { "epoch": 20.63, "learning_rate": 7.259999999999999e-07, "loss": 0.0025, "step": 5850 }, { "epoch": 20.72, "learning_rate": 7.247179487179486e-07, "loss": 0.0029, "step": 5875 }, { "epoch": 20.81, "learning_rate": 7.234358974358974e-07, "loss": 0.0033, "step": 5900 }, { "epoch": 20.9, "learning_rate": 7.221538461538461e-07, "loss": 0.003, "step": 5925 }, { "epoch": 20.99, "learning_rate": 7.208717948717948e-07, "loss": 0.0032, "step": 5950 }, { "epoch": 21.08, "learning_rate": 7.195897435897435e-07, "loss": 0.0032, "step": 5975 }, { "epoch": 21.16, "learning_rate": 7.183076923076923e-07, "loss": 0.0034, "step": 6000 }, { "epoch": 21.16, "eval_loss": 0.3640270531177521, "eval_runtime": 2254.0986, "eval_samples_per_second": 4.025, "eval_steps_per_second": 0.252, "eval_wer": 5.636973022705616, "step": 6000 }, { "epoch": 21.25, "learning_rate": 7.17025641025641e-07, "loss": 0.0025, "step": 6025 }, { "epoch": 21.34, "learning_rate": 7.157435897435897e-07, "loss": 0.0027, "step": 6050 }, { "epoch": 21.43, "learning_rate": 7.144615384615384e-07, "loss": 0.0027, "step": 6075 }, { "epoch": 21.52, "learning_rate": 7.131794871794872e-07, "loss": 0.0029, "step": 6100 }, { "epoch": 21.6, "learning_rate": 7.118974358974359e-07, "loss": 0.003, "step": 6125 }, { "epoch": 21.69, "learning_rate": 7.106153846153846e-07, "loss": 0.0029, "step": 6150 }, { "epoch": 21.78, "learning_rate": 7.093333333333333e-07, "loss": 0.0031, "step": 6175 }, { "epoch": 21.87, "learning_rate": 7.080512820512821e-07, "loss": 0.0032, "step": 6200 }, { "epoch": 21.96, "learning_rate": 7.067692307692308e-07, "loss": 0.0032, "step": 6225 }, { "epoch": 22.05, "learning_rate": 7.054871794871795e-07, "loss": 0.0034, "step": 6250 }, { "epoch": 22.13, "learning_rate": 7.042051282051281e-07, "loss": 0.0036, "step": 6275 }, { "epoch": 22.22, "learning_rate": 7.029230769230769e-07, "loss": 0.0025, "step": 6300 }, { "epoch": 22.31, "learning_rate": 7.016410256410256e-07, "loss": 0.0024, "step": 6325 }, { "epoch": 22.4, "learning_rate": 7.003589743589743e-07, "loss": 0.004, "step": 6350 }, { "epoch": 22.49, "learning_rate": 6.99076923076923e-07, "loss": 0.0035, "step": 6375 }, { "epoch": 22.57, "learning_rate": 6.977948717948717e-07, "loss": 0.0025, "step": 6400 }, { "epoch": 22.66, "learning_rate": 6.965128205128205e-07, "loss": 0.0028, "step": 6425 }, { "epoch": 22.75, "learning_rate": 6.952307692307692e-07, "loss": 0.0032, "step": 6450 }, { "epoch": 22.84, "learning_rate": 6.939487179487179e-07, "loss": 0.0026, "step": 6475 }, { "epoch": 22.93, "learning_rate": 6.926666666666666e-07, "loss": 0.0032, "step": 6500 }, { "epoch": 23.02, "learning_rate": 6.913846153846154e-07, "loss": 0.0032, "step": 6525 }, { "epoch": 23.1, "learning_rate": 6.901025641025641e-07, "loss": 0.0025, "step": 6550 }, { "epoch": 23.19, "learning_rate": 6.888205128205128e-07, "loss": 0.0032, "step": 6575 }, { "epoch": 23.28, "learning_rate": 6.875384615384615e-07, "loss": 0.0022, "step": 6600 }, { "epoch": 23.37, "learning_rate": 6.862564102564103e-07, "loss": 0.0022, "step": 6625 }, { "epoch": 23.46, "learning_rate": 6.84974358974359e-07, "loss": 0.0026, "step": 6650 }, { "epoch": 23.54, "learning_rate": 6.836923076923077e-07, "loss": 0.0022, "step": 6675 }, { "epoch": 23.63, "learning_rate": 6.824102564102564e-07, "loss": 0.0021, "step": 6700 }, { "epoch": 23.72, "learning_rate": 6.811282051282052e-07, "loss": 0.0021, "step": 6725 }, { "epoch": 23.81, "learning_rate": 6.798461538461539e-07, "loss": 0.0026, "step": 6750 }, { "epoch": 23.9, "learning_rate": 6.785641025641026e-07, "loss": 0.0025, "step": 6775 }, { "epoch": 23.99, "learning_rate": 6.772820512820512e-07, "loss": 0.0023, "step": 6800 }, { "epoch": 24.07, "learning_rate": 6.76e-07, "loss": 0.0024, "step": 6825 }, { "epoch": 24.16, "learning_rate": 6.747179487179487e-07, "loss": 0.0022, "step": 6850 }, { "epoch": 24.25, "learning_rate": 6.734358974358974e-07, "loss": 0.0019, "step": 6875 }, { "epoch": 24.34, "learning_rate": 6.72153846153846e-07, "loss": 0.0019, "step": 6900 }, { "epoch": 24.43, "learning_rate": 6.708717948717948e-07, "loss": 0.0026, "step": 6925 }, { "epoch": 24.51, "learning_rate": 6.695897435897435e-07, "loss": 0.0022, "step": 6950 }, { "epoch": 24.6, "learning_rate": 6.683076923076922e-07, "loss": 0.0021, "step": 6975 }, { "epoch": 24.69, "learning_rate": 6.670256410256409e-07, "loss": 0.0021, "step": 7000 }, { "epoch": 24.69, "eval_loss": 0.37135618925094604, "eval_runtime": 2254.6079, "eval_samples_per_second": 4.024, "eval_steps_per_second": 0.251, "eval_wer": 5.648473696316499, "step": 7000 }, { "epoch": 24.78, "learning_rate": 6.657435897435897e-07, "loss": 0.0031, "step": 7025 }, { "epoch": 24.87, "learning_rate": 6.644615384615384e-07, "loss": 0.0022, "step": 7050 }, { "epoch": 24.96, "learning_rate": 6.631794871794871e-07, "loss": 0.0019, "step": 7075 }, { "epoch": 25.04, "learning_rate": 6.618974358974358e-07, "loss": 0.0018, "step": 7100 }, { "epoch": 25.13, "learning_rate": 6.606153846153846e-07, "loss": 0.0019, "step": 7125 }, { "epoch": 25.22, "learning_rate": 6.593333333333333e-07, "loss": 0.0017, "step": 7150 }, { "epoch": 25.31, "learning_rate": 6.58051282051282e-07, "loss": 0.0017, "step": 7175 }, { "epoch": 25.4, "learning_rate": 6.567692307692307e-07, "loss": 0.0021, "step": 7200 }, { "epoch": 25.49, "learning_rate": 6.554871794871795e-07, "loss": 0.0019, "step": 7225 }, { "epoch": 25.57, "learning_rate": 6.542051282051282e-07, "loss": 0.0021, "step": 7250 }, { "epoch": 25.66, "learning_rate": 6.529230769230769e-07, "loss": 0.0028, "step": 7275 }, { "epoch": 25.75, "learning_rate": 6.516410256410256e-07, "loss": 0.0019, "step": 7300 }, { "epoch": 25.84, "learning_rate": 6.503589743589744e-07, "loss": 0.002, "step": 7325 }, { "epoch": 25.93, "learning_rate": 6.490769230769231e-07, "loss": 0.0021, "step": 7350 }, { "epoch": 26.01, "learning_rate": 6.477948717948718e-07, "loss": 0.0023, "step": 7375 }, { "epoch": 26.1, "learning_rate": 6.465128205128205e-07, "loss": 0.0024, "step": 7400 }, { "epoch": 26.19, "learning_rate": 6.452307692307693e-07, "loss": 0.0021, "step": 7425 }, { "epoch": 26.28, "learning_rate": 6.43948717948718e-07, "loss": 0.0018, "step": 7450 }, { "epoch": 26.37, "learning_rate": 6.426666666666667e-07, "loss": 0.0016, "step": 7475 }, { "epoch": 26.46, "learning_rate": 6.413846153846153e-07, "loss": 0.002, "step": 7500 }, { "epoch": 26.54, "learning_rate": 6.40102564102564e-07, "loss": 0.0016, "step": 7525 }, { "epoch": 26.63, "learning_rate": 6.388205128205128e-07, "loss": 0.0019, "step": 7550 }, { "epoch": 26.72, "learning_rate": 6.375384615384615e-07, "loss": 0.0015, "step": 7575 }, { "epoch": 26.81, "learning_rate": 6.362564102564102e-07, "loss": 0.0016, "step": 7600 }, { "epoch": 26.9, "learning_rate": 6.349743589743589e-07, "loss": 0.0017, "step": 7625 }, { "epoch": 26.98, "learning_rate": 6.336923076923077e-07, "loss": 0.002, "step": 7650 }, { "epoch": 27.07, "learning_rate": 6.324102564102564e-07, "loss": 0.0014, "step": 7675 }, { "epoch": 27.16, "learning_rate": 6.311282051282051e-07, "loss": 0.0016, "step": 7700 }, { "epoch": 27.25, "learning_rate": 6.298461538461538e-07, "loss": 0.0018, "step": 7725 }, { "epoch": 27.34, "learning_rate": 6.285641025641026e-07, "loss": 0.0024, "step": 7750 }, { "epoch": 27.43, "learning_rate": 6.272820512820513e-07, "loss": 0.0017, "step": 7775 }, { "epoch": 27.51, "learning_rate": 6.26e-07, "loss": 0.0021, "step": 7800 }, { "epoch": 27.6, "learning_rate": 6.247179487179487e-07, "loss": 0.0019, "step": 7825 }, { "epoch": 27.69, "learning_rate": 6.234358974358975e-07, "loss": 0.0016, "step": 7850 }, { "epoch": 27.78, "learning_rate": 6.221538461538461e-07, "loss": 0.0018, "step": 7875 }, { "epoch": 27.87, "learning_rate": 6.208717948717948e-07, "loss": 0.0018, "step": 7900 }, { "epoch": 27.95, "learning_rate": 6.195897435897435e-07, "loss": 0.0019, "step": 7925 }, { "epoch": 28.04, "learning_rate": 6.183076923076923e-07, "loss": 0.0014, "step": 7950 }, { "epoch": 28.13, "learning_rate": 6.17025641025641e-07, "loss": 0.0014, "step": 7975 }, { "epoch": 28.22, "learning_rate": 6.157435897435897e-07, "loss": 0.0016, "step": 8000 }, { "epoch": 28.22, "eval_loss": 0.396211177110672, "eval_runtime": 2252.8753, "eval_samples_per_second": 4.027, "eval_steps_per_second": 0.252, "eval_wer": 5.625472349094732, "step": 8000 }, { "epoch": 28.31, "learning_rate": 6.144615384615384e-07, "loss": 0.0017, "step": 8025 }, { "epoch": 28.4, "learning_rate": 6.131794871794872e-07, "loss": 0.0017, "step": 8050 }, { "epoch": 28.48, "learning_rate": 6.118974358974359e-07, "loss": 0.0028, "step": 8075 }, { "epoch": 28.57, "learning_rate": 6.106153846153845e-07, "loss": 0.003, "step": 8100 }, { "epoch": 28.66, "learning_rate": 6.093333333333332e-07, "loss": 0.0021, "step": 8125 }, { "epoch": 28.75, "learning_rate": 6.08051282051282e-07, "loss": 0.0019, "step": 8150 }, { "epoch": 28.84, "learning_rate": 6.067692307692307e-07, "loss": 0.0025, "step": 8175 }, { "epoch": 28.92, "learning_rate": 6.054871794871794e-07, "loss": 0.0019, "step": 8200 }, { "epoch": 29.01, "learning_rate": 6.042051282051281e-07, "loss": 0.0016, "step": 8225 }, { "epoch": 29.1, "learning_rate": 6.029230769230769e-07, "loss": 0.0015, "step": 8250 }, { "epoch": 29.19, "learning_rate": 6.016410256410256e-07, "loss": 0.0012, "step": 8275 }, { "epoch": 29.28, "learning_rate": 6.003589743589743e-07, "loss": 0.0012, "step": 8300 }, { "epoch": 29.37, "learning_rate": 5.99076923076923e-07, "loss": 0.0015, "step": 8325 }, { "epoch": 29.45, "learning_rate": 5.977948717948718e-07, "loss": 0.0038, "step": 8350 }, { "epoch": 29.54, "learning_rate": 5.965128205128205e-07, "loss": 0.0014, "step": 8375 }, { "epoch": 29.63, "learning_rate": 5.952307692307692e-07, "loss": 0.0013, "step": 8400 }, { "epoch": 29.72, "learning_rate": 5.939487179487179e-07, "loss": 0.0023, "step": 8425 }, { "epoch": 29.81, "learning_rate": 5.926666666666667e-07, "loss": 0.002, "step": 8450 }, { "epoch": 29.89, "learning_rate": 5.913846153846154e-07, "loss": 0.0022, "step": 8475 }, { "epoch": 29.98, "learning_rate": 5.901025641025641e-07, "loss": 0.0018, "step": 8500 }, { "epoch": 30.07, "learning_rate": 5.888205128205128e-07, "loss": 0.0023, "step": 8525 }, { "epoch": 30.16, "learning_rate": 5.875384615384616e-07, "loss": 0.0019, "step": 8550 }, { "epoch": 30.25, "learning_rate": 5.862564102564103e-07, "loss": 0.0015, "step": 8575 }, { "epoch": 30.34, "learning_rate": 5.84974358974359e-07, "loss": 0.0014, "step": 8600 }, { "epoch": 30.42, "learning_rate": 5.836923076923077e-07, "loss": 0.0013, "step": 8625 }, { "epoch": 30.51, "learning_rate": 5.824102564102565e-07, "loss": 0.0023, "step": 8650 }, { "epoch": 30.6, "learning_rate": 5.811282051282052e-07, "loss": 0.0018, "step": 8675 }, { "epoch": 30.69, "learning_rate": 5.798461538461539e-07, "loss": 0.0015, "step": 8700 }, { "epoch": 30.78, "learning_rate": 5.785641025641025e-07, "loss": 0.0026, "step": 8725 }, { "epoch": 30.86, "learning_rate": 5.773333333333334e-07, "loss": 0.0019, "step": 8750 }, { "epoch": 30.95, "learning_rate": 5.760512820512821e-07, "loss": 0.0016, "step": 8775 }, { "epoch": 31.04, "learning_rate": 5.747692307692308e-07, "loss": 0.0016, "step": 8800 }, { "epoch": 31.13, "learning_rate": 5.734871794871795e-07, "loss": 0.0021, "step": 8825 }, { "epoch": 31.22, "learning_rate": 5.722051282051283e-07, "loss": 0.0017, "step": 8850 }, { "epoch": 31.31, "learning_rate": 5.70923076923077e-07, "loss": 0.0017, "step": 8875 }, { "epoch": 31.39, "learning_rate": 5.696410256410256e-07, "loss": 0.0018, "step": 8900 }, { "epoch": 31.48, "learning_rate": 5.683589743589743e-07, "loss": 0.0016, "step": 8925 }, { "epoch": 31.57, "learning_rate": 5.670769230769231e-07, "loss": 0.0013, "step": 8950 }, { "epoch": 31.66, "learning_rate": 5.657948717948717e-07, "loss": 0.0018, "step": 8975 }, { "epoch": 31.75, "learning_rate": 5.645128205128204e-07, "loss": 0.0013, "step": 9000 }, { "epoch": 31.75, "eval_loss": 0.3959801197052002, "eval_runtime": 2256.2096, "eval_samples_per_second": 4.021, "eval_steps_per_second": 0.251, "eval_wer": 5.673117996911248, "step": 9000 }, { "epoch": 31.83, "learning_rate": 5.632307692307691e-07, "loss": 0.0017, "step": 9025 }, { "epoch": 31.92, "learning_rate": 5.619487179487179e-07, "loss": 0.0015, "step": 9050 }, { "epoch": 32.01, "learning_rate": 5.606666666666666e-07, "loss": 0.0018, "step": 9075 }, { "epoch": 32.1, "learning_rate": 5.593846153846153e-07, "loss": 0.0017, "step": 9100 }, { "epoch": 32.19, "learning_rate": 5.58102564102564e-07, "loss": 0.001, "step": 9125 }, { "epoch": 32.28, "learning_rate": 5.568205128205128e-07, "loss": 0.0012, "step": 9150 }, { "epoch": 32.36, "learning_rate": 5.555384615384615e-07, "loss": 0.0011, "step": 9175 }, { "epoch": 32.45, "learning_rate": 5.542564102564102e-07, "loss": 0.0016, "step": 9200 }, { "epoch": 32.54, "learning_rate": 5.529743589743589e-07, "loss": 0.0012, "step": 9225 }, { "epoch": 32.63, "learning_rate": 5.516923076923077e-07, "loss": 0.0014, "step": 9250 }, { "epoch": 32.72, "learning_rate": 5.504102564102564e-07, "loss": 0.0014, "step": 9275 }, { "epoch": 32.8, "learning_rate": 5.491282051282051e-07, "loss": 0.0014, "step": 9300 }, { "epoch": 32.89, "learning_rate": 5.478461538461538e-07, "loss": 0.0013, "step": 9325 }, { "epoch": 32.98, "learning_rate": 5.465641025641026e-07, "loss": 0.0021, "step": 9350 }, { "epoch": 33.07, "learning_rate": 5.452820512820513e-07, "loss": 0.0011, "step": 9375 }, { "epoch": 33.16, "learning_rate": 5.44e-07, "loss": 0.0015, "step": 9400 }, { "epoch": 33.25, "learning_rate": 5.427179487179487e-07, "loss": 0.0012, "step": 9425 }, { "epoch": 33.33, "learning_rate": 5.414358974358975e-07, "loss": 0.0012, "step": 9450 }, { "epoch": 33.42, "learning_rate": 5.401538461538462e-07, "loss": 0.0011, "step": 9475 }, { "epoch": 33.51, "learning_rate": 5.388717948717948e-07, "loss": 0.0015, "step": 9500 }, { "epoch": 33.6, "learning_rate": 5.375897435897435e-07, "loss": 0.0014, "step": 9525 }, { "epoch": 33.69, "learning_rate": 5.363076923076922e-07, "loss": 0.001, "step": 9550 }, { "epoch": 33.77, "learning_rate": 5.35025641025641e-07, "loss": 0.001, "step": 9575 }, { "epoch": 33.86, "learning_rate": 5.337435897435897e-07, "loss": 0.0013, "step": 9600 }, { "epoch": 33.95, "learning_rate": 5.324615384615384e-07, "loss": 0.0013, "step": 9625 }, { "epoch": 34.04, "learning_rate": 5.311794871794871e-07, "loss": 0.0012, "step": 9650 }, { "epoch": 34.13, "learning_rate": 5.298974358974359e-07, "loss": 0.001, "step": 9675 }, { "epoch": 34.22, "learning_rate": 5.286153846153846e-07, "loss": 0.0011, "step": 9700 }, { "epoch": 34.3, "learning_rate": 5.273333333333333e-07, "loss": 0.0009, "step": 9725 }, { "epoch": 34.39, "learning_rate": 5.26051282051282e-07, "loss": 0.0009, "step": 9750 }, { "epoch": 34.48, "learning_rate": 5.247692307692308e-07, "loss": 0.001, "step": 9775 }, { "epoch": 34.57, "learning_rate": 5.234871794871795e-07, "loss": 0.0011, "step": 9800 }, { "epoch": 34.66, "learning_rate": 5.222051282051282e-07, "loss": 0.0009, "step": 9825 }, { "epoch": 34.74, "learning_rate": 5.209230769230769e-07, "loss": 0.001, "step": 9850 }, { "epoch": 34.83, "learning_rate": 5.196410256410257e-07, "loss": 0.0009, "step": 9875 }, { "epoch": 34.92, "learning_rate": 5.183589743589744e-07, "loss": 0.0012, "step": 9900 }, { "epoch": 35.01, "learning_rate": 5.170769230769231e-07, "loss": 0.0009, "step": 9925 }, { "epoch": 35.1, "learning_rate": 5.157948717948718e-07, "loss": 0.0008, "step": 9950 }, { "epoch": 35.19, "learning_rate": 5.145128205128206e-07, "loss": 0.0009, "step": 9975 }, { "epoch": 35.27, "learning_rate": 5.132307692307693e-07, "loss": 0.0009, "step": 10000 }, { "epoch": 35.27, "eval_loss": 0.410661518573761, "eval_runtime": 2238.8448, "eval_samples_per_second": 4.052, "eval_steps_per_second": 0.253, "eval_wer": 5.702691157624947, "step": 10000 }, { "epoch": 35.36, "learning_rate": 5.11948717948718e-07, "loss": 0.001, "step": 10025 }, { "epoch": 35.45, "learning_rate": 5.106666666666667e-07, "loss": 0.0013, "step": 10050 }, { "epoch": 35.54, "learning_rate": 5.093846153846155e-07, "loss": 0.0011, "step": 10075 }, { "epoch": 35.63, "learning_rate": 5.08102564102564e-07, "loss": 0.0009, "step": 10100 }, { "epoch": 35.71, "learning_rate": 5.068205128205127e-07, "loss": 0.001, "step": 10125 }, { "epoch": 35.8, "learning_rate": 5.055384615384614e-07, "loss": 0.001, "step": 10150 }, { "epoch": 35.89, "learning_rate": 5.042564102564102e-07, "loss": 0.001, "step": 10175 }, { "epoch": 35.98, "learning_rate": 5.029743589743589e-07, "loss": 0.0009, "step": 10200 }, { "epoch": 36.07, "learning_rate": 5.016923076923076e-07, "loss": 0.0017, "step": 10225 }, { "epoch": 36.16, "learning_rate": 5.004102564102563e-07, "loss": 0.0009, "step": 10250 }, { "epoch": 36.24, "learning_rate": 4.99128205128205e-07, "loss": 0.0013, "step": 10275 }, { "epoch": 36.33, "learning_rate": 4.978461538461538e-07, "loss": 0.0016, "step": 10300 }, { "epoch": 36.42, "learning_rate": 4.965641025641025e-07, "loss": 0.0022, "step": 10325 }, { "epoch": 36.51, "learning_rate": 4.952820512820512e-07, "loss": 0.0021, "step": 10350 }, { "epoch": 36.6, "learning_rate": 4.94e-07, "loss": 0.0009, "step": 10375 }, { "epoch": 36.68, "learning_rate": 4.927179487179487e-07, "loss": 0.0017, "step": 10400 }, { "epoch": 36.77, "learning_rate": 4.914358974358974e-07, "loss": 0.0013, "step": 10425 }, { "epoch": 36.86, "learning_rate": 4.901538461538461e-07, "loss": 0.0014, "step": 10450 }, { "epoch": 36.95, "learning_rate": 4.888717948717948e-07, "loss": 0.002, "step": 10475 }, { "epoch": 37.04, "learning_rate": 4.875897435897436e-07, "loss": 0.0009, "step": 10500 }, { "epoch": 37.13, "learning_rate": 4.863076923076923e-07, "loss": 0.0012, "step": 10525 }, { "epoch": 37.21, "learning_rate": 4.85025641025641e-07, "loss": 0.001, "step": 10550 }, { "epoch": 37.3, "learning_rate": 4.837435897435897e-07, "loss": 0.0008, "step": 10575 }, { "epoch": 37.39, "learning_rate": 4.824615384615384e-07, "loss": 0.0009, "step": 10600 }, { "epoch": 37.48, "learning_rate": 4.811794871794871e-07, "loss": 0.0013, "step": 10625 }, { "epoch": 37.57, "learning_rate": 4.798974358974358e-07, "loss": 0.0011, "step": 10650 }, { "epoch": 37.65, "learning_rate": 4.786153846153845e-07, "loss": 0.0019, "step": 10675 }, { "epoch": 37.74, "learning_rate": 4.773333333333333e-07, "loss": 0.0014, "step": 10700 }, { "epoch": 37.83, "learning_rate": 4.7605128205128204e-07, "loss": 0.0009, "step": 10725 }, { "epoch": 37.92, "learning_rate": 4.7476923076923077e-07, "loss": 0.0015, "step": 10750 }, { "epoch": 38.01, "learning_rate": 4.734871794871795e-07, "loss": 0.001, "step": 10775 }, { "epoch": 38.1, "learning_rate": 4.722051282051282e-07, "loss": 0.001, "step": 10800 }, { "epoch": 38.18, "learning_rate": 4.7092307692307694e-07, "loss": 0.0009, "step": 10825 }, { "epoch": 38.27, "learning_rate": 4.6964102564102567e-07, "loss": 0.0009, "step": 10850 }, { "epoch": 38.36, "learning_rate": 4.683589743589743e-07, "loss": 0.0008, "step": 10875 }, { "epoch": 38.45, "learning_rate": 4.67076923076923e-07, "loss": 0.0012, "step": 10900 }, { "epoch": 38.54, "learning_rate": 4.6579487179487174e-07, "loss": 0.0008, "step": 10925 }, { "epoch": 38.62, "learning_rate": 4.6451282051282046e-07, "loss": 0.001, "step": 10950 }, { "epoch": 38.71, "learning_rate": 4.632307692307692e-07, "loss": 0.0027, "step": 10975 }, { "epoch": 38.8, "learning_rate": 4.619487179487179e-07, "loss": 0.0008, "step": 11000 }, { "epoch": 38.8, "eval_loss": 0.39808371663093567, "eval_runtime": 2241.5703, "eval_samples_per_second": 4.047, "eval_steps_per_second": 0.253, "eval_wer": 5.986922091151054, "step": 11000 }, { "epoch": 38.89, "learning_rate": 4.6066666666666664e-07, "loss": 0.0009, "step": 11025 }, { "epoch": 38.98, "learning_rate": 4.5938461538461536e-07, "loss": 0.0008, "step": 11050 }, { "epoch": 39.07, "learning_rate": 4.581025641025641e-07, "loss": 0.001, "step": 11075 }, { "epoch": 39.15, "learning_rate": 4.568205128205128e-07, "loss": 0.0008, "step": 11100 }, { "epoch": 39.24, "learning_rate": 4.5553846153846154e-07, "loss": 0.0007, "step": 11125 }, { "epoch": 39.33, "learning_rate": 4.5425641025641026e-07, "loss": 0.0011, "step": 11150 }, { "epoch": 39.42, "learning_rate": 4.5297435897435893e-07, "loss": 0.0009, "step": 11175 }, { "epoch": 39.51, "learning_rate": 4.5169230769230766e-07, "loss": 0.0008, "step": 11200 }, { "epoch": 39.59, "learning_rate": 4.504102564102564e-07, "loss": 0.0008, "step": 11225 }, { "epoch": 39.68, "learning_rate": 4.491282051282051e-07, "loss": 0.0008, "step": 11250 }, { "epoch": 39.77, "learning_rate": 4.4784615384615383e-07, "loss": 0.0008, "step": 11275 }, { "epoch": 39.86, "learning_rate": 4.4656410256410256e-07, "loss": 0.0008, "step": 11300 }, { "epoch": 39.95, "learning_rate": 4.452820512820513e-07, "loss": 0.0009, "step": 11325 }, { "epoch": 40.04, "learning_rate": 4.44e-07, "loss": 0.0008, "step": 11350 }, { "epoch": 40.12, "learning_rate": 4.4271794871794873e-07, "loss": 0.0012, "step": 11375 }, { "epoch": 40.21, "learning_rate": 4.4143589743589745e-07, "loss": 0.0013, "step": 11400 }, { "epoch": 40.3, "learning_rate": 4.401538461538462e-07, "loss": 0.0009, "step": 11425 }, { "epoch": 40.39, "learning_rate": 4.3887179487179485e-07, "loss": 0.0009, "step": 11450 }, { "epoch": 40.48, "learning_rate": 4.375897435897435e-07, "loss": 0.0016, "step": 11475 }, { "epoch": 40.56, "learning_rate": 4.3630769230769225e-07, "loss": 0.0008, "step": 11500 }, { "epoch": 40.65, "learning_rate": 4.3502564102564097e-07, "loss": 0.0007, "step": 11525 }, { "epoch": 40.74, "learning_rate": 4.337435897435897e-07, "loss": 0.0007, "step": 11550 }, { "epoch": 40.83, "learning_rate": 4.324615384615384e-07, "loss": 0.0008, "step": 11575 }, { "epoch": 40.92, "learning_rate": 4.3117948717948715e-07, "loss": 0.0009, "step": 11600 }, { "epoch": 41.01, "learning_rate": 4.2989743589743587e-07, "loss": 0.0009, "step": 11625 }, { "epoch": 41.09, "learning_rate": 4.286153846153846e-07, "loss": 0.0006, "step": 11650 }, { "epoch": 41.18, "learning_rate": 4.273333333333333e-07, "loss": 0.0007, "step": 11675 }, { "epoch": 41.27, "learning_rate": 4.2605128205128205e-07, "loss": 0.0013, "step": 11700 }, { "epoch": 41.36, "learning_rate": 4.2476923076923077e-07, "loss": 0.0008, "step": 11725 }, { "epoch": 41.45, "learning_rate": 4.234871794871795e-07, "loss": 0.001, "step": 11750 }, { "epoch": 41.53, "learning_rate": 4.222051282051282e-07, "loss": 0.0016, "step": 11775 }, { "epoch": 41.62, "learning_rate": 4.209230769230769e-07, "loss": 0.0008, "step": 11800 }, { "epoch": 41.71, "learning_rate": 4.196410256410256e-07, "loss": 0.0008, "step": 11825 }, { "epoch": 41.8, "learning_rate": 4.1835897435897434e-07, "loss": 0.0008, "step": 11850 }, { "epoch": 41.89, "learning_rate": 4.1707692307692307e-07, "loss": 0.001, "step": 11875 }, { "epoch": 41.98, "learning_rate": 4.157948717948718e-07, "loss": 0.0009, "step": 11900 }, { "epoch": 42.06, "learning_rate": 4.145128205128205e-07, "loss": 0.0007, "step": 11925 }, { "epoch": 42.15, "learning_rate": 4.1323076923076924e-07, "loss": 0.0016, "step": 11950 }, { "epoch": 42.24, "learning_rate": 4.119487179487179e-07, "loss": 0.0007, "step": 11975 }, { "epoch": 42.33, "learning_rate": 4.1066666666666664e-07, "loss": 0.0006, "step": 12000 }, { "epoch": 42.33, "eval_loss": 0.40968000888824463, "eval_runtime": 2244.3687, "eval_samples_per_second": 4.042, "eval_steps_per_second": 0.253, "eval_wer": 5.701048204251964, "step": 12000 }, { "epoch": 42.42, "learning_rate": 4.0938461538461536e-07, "loss": 0.0008, "step": 12025 }, { "epoch": 42.5, "learning_rate": 4.081025641025641e-07, "loss": 0.0008, "step": 12050 }, { "epoch": 42.59, "learning_rate": 4.068205128205128e-07, "loss": 0.0007, "step": 12075 }, { "epoch": 42.68, "learning_rate": 4.055384615384615e-07, "loss": 0.0008, "step": 12100 }, { "epoch": 42.77, "learning_rate": 4.042564102564102e-07, "loss": 0.0006, "step": 12125 }, { "epoch": 42.86, "learning_rate": 4.0297435897435893e-07, "loss": 0.0011, "step": 12150 }, { "epoch": 42.95, "learning_rate": 4.0169230769230766e-07, "loss": 0.0008, "step": 12175 }, { "epoch": 43.03, "learning_rate": 4.004102564102564e-07, "loss": 0.0006, "step": 12200 }, { "epoch": 43.12, "learning_rate": 3.991282051282051e-07, "loss": 0.0007, "step": 12225 }, { "epoch": 43.21, "learning_rate": 3.9784615384615383e-07, "loss": 0.0006, "step": 12250 }, { "epoch": 43.3, "learning_rate": 3.9656410256410256e-07, "loss": 0.0006, "step": 12275 }, { "epoch": 43.39, "learning_rate": 3.952820512820513e-07, "loss": 0.0008, "step": 12300 }, { "epoch": 43.47, "learning_rate": 3.94e-07, "loss": 0.0006, "step": 12325 }, { "epoch": 43.56, "learning_rate": 3.9271794871794873e-07, "loss": 0.0006, "step": 12350 }, { "epoch": 43.65, "learning_rate": 3.9143589743589746e-07, "loss": 0.0007, "step": 12375 }, { "epoch": 43.74, "learning_rate": 3.9015384615384613e-07, "loss": 0.0008, "step": 12400 }, { "epoch": 43.83, "learning_rate": 3.8887179487179485e-07, "loss": 0.0009, "step": 12425 }, { "epoch": 43.92, "learning_rate": 3.875897435897436e-07, "loss": 0.0006, "step": 12450 }, { "epoch": 44.0, "learning_rate": 3.863076923076923e-07, "loss": 0.0008, "step": 12475 }, { "epoch": 44.09, "learning_rate": 3.85025641025641e-07, "loss": 0.0006, "step": 12500 }, { "epoch": 44.18, "learning_rate": 3.837435897435897e-07, "loss": 0.0006, "step": 12525 }, { "epoch": 44.27, "learning_rate": 3.824615384615384e-07, "loss": 0.0006, "step": 12550 }, { "epoch": 44.36, "learning_rate": 3.8117948717948715e-07, "loss": 0.0006, "step": 12575 }, { "epoch": 44.44, "learning_rate": 3.798974358974359e-07, "loss": 0.0006, "step": 12600 }, { "epoch": 44.53, "learning_rate": 3.786153846153846e-07, "loss": 0.0006, "step": 12625 }, { "epoch": 44.62, "learning_rate": 3.773333333333333e-07, "loss": 0.0006, "step": 12650 }, { "epoch": 44.71, "learning_rate": 3.7605128205128205e-07, "loss": 0.0009, "step": 12675 }, { "epoch": 44.8, "learning_rate": 3.747692307692307e-07, "loss": 0.0006, "step": 12700 }, { "epoch": 44.89, "learning_rate": 3.7348717948717945e-07, "loss": 0.0006, "step": 12725 }, { "epoch": 44.97, "learning_rate": 3.7220512820512817e-07, "loss": 0.0006, "step": 12750 }, { "epoch": 45.06, "learning_rate": 3.709230769230769e-07, "loss": 0.0006, "step": 12775 }, { "epoch": 45.15, "learning_rate": 3.696410256410256e-07, "loss": 0.0006, "step": 12800 }, { "epoch": 45.24, "learning_rate": 3.6835897435897434e-07, "loss": 0.0005, "step": 12825 }, { "epoch": 45.33, "learning_rate": 3.6707692307692307e-07, "loss": 0.0005, "step": 12850 }, { "epoch": 45.41, "learning_rate": 3.657948717948718e-07, "loss": 0.0005, "step": 12875 }, { "epoch": 45.5, "learning_rate": 3.645128205128205e-07, "loss": 0.0005, "step": 12900 }, { "epoch": 45.59, "learning_rate": 3.6323076923076924e-07, "loss": 0.0005, "step": 12925 }, { "epoch": 45.68, "learning_rate": 3.6194871794871797e-07, "loss": 0.0005, "step": 12950 }, { "epoch": 45.77, "learning_rate": 3.606666666666667e-07, "loss": 0.0005, "step": 12975 }, { "epoch": 45.86, "learning_rate": 3.5938461538461537e-07, "loss": 0.0005, "step": 13000 }, { "epoch": 45.86, "eval_loss": 0.4226188063621521, "eval_runtime": 2237.7538, "eval_samples_per_second": 4.054, "eval_steps_per_second": 0.253, "eval_wer": 5.8144119869878095, "step": 13000 }, { "epoch": 45.94, "learning_rate": 3.581025641025641e-07, "loss": 0.0005, "step": 13025 }, { "epoch": 46.03, "learning_rate": 3.5682051282051276e-07, "loss": 0.0005, "step": 13050 }, { "epoch": 46.12, "learning_rate": 3.555384615384615e-07, "loss": 0.0005, "step": 13075 }, { "epoch": 46.21, "learning_rate": 3.542564102564102e-07, "loss": 0.0005, "step": 13100 }, { "epoch": 46.3, "learning_rate": 3.5297435897435894e-07, "loss": 0.0005, "step": 13125 }, { "epoch": 46.38, "learning_rate": 3.5169230769230766e-07, "loss": 0.0005, "step": 13150 }, { "epoch": 46.47, "learning_rate": 3.504102564102564e-07, "loss": 0.0005, "step": 13175 }, { "epoch": 46.56, "learning_rate": 3.491282051282051e-07, "loss": 0.0005, "step": 13200 }, { "epoch": 46.65, "learning_rate": 3.4784615384615384e-07, "loss": 0.0005, "step": 13225 }, { "epoch": 46.74, "learning_rate": 3.4656410256410256e-07, "loss": 0.0005, "step": 13250 }, { "epoch": 46.83, "learning_rate": 3.452820512820513e-07, "loss": 0.0005, "step": 13275 }, { "epoch": 46.91, "learning_rate": 3.4399999999999996e-07, "loss": 0.0005, "step": 13300 }, { "epoch": 47.0, "learning_rate": 3.427179487179487e-07, "loss": 0.0005, "step": 13325 }, { "epoch": 47.09, "learning_rate": 3.414358974358974e-07, "loss": 0.0005, "step": 13350 }, { "epoch": 47.18, "learning_rate": 3.4015384615384613e-07, "loss": 0.0005, "step": 13375 }, { "epoch": 47.27, "learning_rate": 3.3887179487179486e-07, "loss": 0.0005, "step": 13400 }, { "epoch": 47.35, "learning_rate": 3.375897435897436e-07, "loss": 0.0005, "step": 13425 }, { "epoch": 47.44, "learning_rate": 3.363076923076923e-07, "loss": 0.0005, "step": 13450 }, { "epoch": 47.53, "learning_rate": 3.3502564102564103e-07, "loss": 0.0005, "step": 13475 }, { "epoch": 47.62, "learning_rate": 3.3374358974358976e-07, "loss": 0.0005, "step": 13500 }, { "epoch": 47.71, "learning_rate": 3.324615384615385e-07, "loss": 0.0005, "step": 13525 }, { "epoch": 47.8, "learning_rate": 3.311794871794872e-07, "loss": 0.0005, "step": 13550 }, { "epoch": 47.88, "learning_rate": 3.2989743589743593e-07, "loss": 0.0005, "step": 13575 }, { "epoch": 47.97, "learning_rate": 3.286153846153846e-07, "loss": 0.0005, "step": 13600 }, { "epoch": 48.06, "learning_rate": 3.2733333333333327e-07, "loss": 0.0005, "step": 13625 }, { "epoch": 48.15, "learning_rate": 3.26051282051282e-07, "loss": 0.0004, "step": 13650 }, { "epoch": 48.24, "learning_rate": 3.247692307692307e-07, "loss": 0.0005, "step": 13675 }, { "epoch": 48.32, "learning_rate": 3.2348717948717945e-07, "loss": 0.0005, "step": 13700 }, { "epoch": 48.41, "learning_rate": 3.2220512820512817e-07, "loss": 0.0005, "step": 13725 }, { "epoch": 48.5, "learning_rate": 3.209230769230769e-07, "loss": 0.0005, "step": 13750 }, { "epoch": 48.59, "learning_rate": 3.196410256410256e-07, "loss": 0.0005, "step": 13775 }, { "epoch": 48.68, "learning_rate": 3.1835897435897435e-07, "loss": 0.0005, "step": 13800 }, { "epoch": 48.77, "learning_rate": 3.1707692307692307e-07, "loss": 0.0004, "step": 13825 }, { "epoch": 48.85, "learning_rate": 3.157948717948718e-07, "loss": 0.0005, "step": 13850 }, { "epoch": 48.94, "learning_rate": 3.145128205128205e-07, "loss": 0.0005, "step": 13875 }, { "epoch": 49.03, "learning_rate": 3.1323076923076925e-07, "loss": 0.0004, "step": 13900 }, { "epoch": 49.12, "learning_rate": 3.119487179487179e-07, "loss": 0.0004, "step": 13925 }, { "epoch": 49.21, "learning_rate": 3.1066666666666664e-07, "loss": 0.0005, "step": 13950 }, { "epoch": 49.29, "learning_rate": 3.0938461538461537e-07, "loss": 0.0004, "step": 13975 }, { "epoch": 49.38, "learning_rate": 3.081025641025641e-07, "loss": 0.0004, "step": 14000 }, { "epoch": 49.38, "eval_loss": 0.4330260455608368, "eval_runtime": 2236.0285, "eval_samples_per_second": 4.057, "eval_steps_per_second": 0.254, "eval_wer": 5.825912660598692, "step": 14000 }, { "epoch": 49.47, "learning_rate": 3.068205128205128e-07, "loss": 0.0004, "step": 14025 }, { "epoch": 49.56, "learning_rate": 3.0553846153846154e-07, "loss": 0.0004, "step": 14050 }, { "epoch": 49.65, "learning_rate": 3.0425641025641027e-07, "loss": 0.0004, "step": 14075 }, { "epoch": 49.74, "learning_rate": 3.02974358974359e-07, "loss": 0.0004, "step": 14100 }, { "epoch": 49.82, "learning_rate": 3.016923076923077e-07, "loss": 0.0004, "step": 14125 }, { "epoch": 49.91, "learning_rate": 3.004102564102564e-07, "loss": 0.0004, "step": 14150 }, { "epoch": 50.0, "learning_rate": 2.991282051282051e-07, "loss": 0.0004, "step": 14175 }, { "epoch": 50.09, "learning_rate": 2.9784615384615384e-07, "loss": 0.0004, "step": 14200 }, { "epoch": 50.18, "learning_rate": 2.965641025641025e-07, "loss": 0.0004, "step": 14225 }, { "epoch": 50.26, "learning_rate": 2.9528205128205123e-07, "loss": 0.0004, "step": 14250 }, { "epoch": 50.35, "learning_rate": 2.9399999999999996e-07, "loss": 0.0004, "step": 14275 }, { "epoch": 50.44, "learning_rate": 2.927179487179487e-07, "loss": 0.0004, "step": 14300 }, { "epoch": 50.53, "learning_rate": 2.914358974358974e-07, "loss": 0.0004, "step": 14325 }, { "epoch": 50.62, "learning_rate": 2.9015384615384613e-07, "loss": 0.0004, "step": 14350 }, { "epoch": 50.71, "learning_rate": 2.8887179487179486e-07, "loss": 0.0004, "step": 14375 }, { "epoch": 50.79, "learning_rate": 2.875897435897436e-07, "loss": 0.0004, "step": 14400 }, { "epoch": 50.88, "learning_rate": 2.863076923076923e-07, "loss": 0.0004, "step": 14425 }, { "epoch": 50.97, "learning_rate": 2.8502564102564103e-07, "loss": 0.0004, "step": 14450 }, { "epoch": 51.06, "learning_rate": 2.8374358974358976e-07, "loss": 0.0004, "step": 14475 }, { "epoch": 51.15, "learning_rate": 2.824615384615385e-07, "loss": 0.0004, "step": 14500 }, { "epoch": 51.23, "learning_rate": 2.8117948717948715e-07, "loss": 0.0004, "step": 14525 }, { "epoch": 51.32, "learning_rate": 2.798974358974359e-07, "loss": 0.0004, "step": 14550 }, { "epoch": 51.41, "learning_rate": 2.786153846153846e-07, "loss": 0.0004, "step": 14575 }, { "epoch": 51.5, "learning_rate": 2.7733333333333333e-07, "loss": 0.0004, "step": 14600 }, { "epoch": 51.59, "learning_rate": 2.7605128205128205e-07, "loss": 0.0004, "step": 14625 }, { "epoch": 51.68, "learning_rate": 2.747692307692308e-07, "loss": 0.0004, "step": 14650 }, { "epoch": 51.76, "learning_rate": 2.7348717948717945e-07, "loss": 0.0004, "step": 14675 }, { "epoch": 51.85, "learning_rate": 2.722051282051282e-07, "loss": 0.0004, "step": 14700 }, { "epoch": 51.94, "learning_rate": 2.709230769230769e-07, "loss": 0.0004, "step": 14725 }, { "epoch": 52.03, "learning_rate": 2.696410256410256e-07, "loss": 0.0004, "step": 14750 }, { "epoch": 52.12, "learning_rate": 2.6835897435897435e-07, "loss": 0.0004, "step": 14775 }, { "epoch": 52.2, "learning_rate": 2.670769230769231e-07, "loss": 0.0004, "step": 14800 }, { "epoch": 52.29, "learning_rate": 2.6579487179487175e-07, "loss": 0.0004, "step": 14825 }, { "epoch": 52.38, "learning_rate": 2.6451282051282047e-07, "loss": 0.0004, "step": 14850 }, { "epoch": 52.47, "learning_rate": 2.632307692307692e-07, "loss": 0.0004, "step": 14875 }, { "epoch": 52.56, "learning_rate": 2.619487179487179e-07, "loss": 0.0004, "step": 14900 }, { "epoch": 52.65, "learning_rate": 2.6066666666666664e-07, "loss": 0.0004, "step": 14925 }, { "epoch": 52.73, "learning_rate": 2.5938461538461537e-07, "loss": 0.0004, "step": 14950 }, { "epoch": 52.82, "learning_rate": 2.581025641025641e-07, "loss": 0.0004, "step": 14975 }, { "epoch": 52.91, "learning_rate": 2.568205128205128e-07, "loss": 0.0004, "step": 15000 }, { "epoch": 52.91, "eval_loss": 0.4414743185043335, "eval_runtime": 2236.5929, "eval_samples_per_second": 4.056, "eval_steps_per_second": 0.254, "eval_wer": 5.791410639766044, "step": 15000 }, { "epoch": 53.0, "learning_rate": 2.5553846153846154e-07, "loss": 0.0004, "step": 15025 }, { "epoch": 53.09, "learning_rate": 2.5425641025641027e-07, "loss": 0.0004, "step": 15050 }, { "epoch": 53.17, "learning_rate": 2.52974358974359e-07, "loss": 0.0004, "step": 15075 }, { "epoch": 53.26, "learning_rate": 2.516923076923077e-07, "loss": 0.0004, "step": 15100 }, { "epoch": 53.35, "learning_rate": 2.5041025641025644e-07, "loss": 0.0004, "step": 15125 }, { "epoch": 53.44, "learning_rate": 2.491282051282051e-07, "loss": 0.0004, "step": 15150 }, { "epoch": 53.53, "learning_rate": 2.4784615384615384e-07, "loss": 0.0004, "step": 15175 }, { "epoch": 53.62, "learning_rate": 2.4656410256410256e-07, "loss": 0.0004, "step": 15200 }, { "epoch": 53.7, "learning_rate": 2.4528205128205124e-07, "loss": 0.0004, "step": 15225 }, { "epoch": 53.79, "learning_rate": 2.4399999999999996e-07, "loss": 0.0004, "step": 15250 }, { "epoch": 53.88, "learning_rate": 2.427179487179487e-07, "loss": 0.0004, "step": 15275 }, { "epoch": 53.97, "learning_rate": 2.414358974358974e-07, "loss": 0.0004, "step": 15300 }, { "epoch": 54.06, "learning_rate": 2.4015384615384614e-07, "loss": 0.0004, "step": 15325 }, { "epoch": 54.14, "learning_rate": 2.3887179487179486e-07, "loss": 0.0004, "step": 15350 }, { "epoch": 54.23, "learning_rate": 2.3758974358974356e-07, "loss": 0.0004, "step": 15375 }, { "epoch": 54.32, "learning_rate": 2.3630769230769228e-07, "loss": 0.0003, "step": 15400 }, { "epoch": 54.41, "learning_rate": 2.35025641025641e-07, "loss": 0.0003, "step": 15425 }, { "epoch": 54.5, "learning_rate": 2.3374358974358973e-07, "loss": 0.0003, "step": 15450 }, { "epoch": 54.59, "learning_rate": 2.3246153846153846e-07, "loss": 0.0003, "step": 15475 }, { "epoch": 54.67, "learning_rate": 2.3117948717948718e-07, "loss": 0.0003, "step": 15500 }, { "epoch": 54.76, "learning_rate": 2.2989743589743588e-07, "loss": 0.0003, "step": 15525 }, { "epoch": 54.85, "learning_rate": 2.286153846153846e-07, "loss": 0.0003, "step": 15550 }, { "epoch": 54.94, "learning_rate": 2.2733333333333333e-07, "loss": 0.0004, "step": 15575 }, { "epoch": 55.03, "learning_rate": 2.2605128205128206e-07, "loss": 0.0003, "step": 15600 }, { "epoch": 55.11, "learning_rate": 2.2476923076923075e-07, "loss": 0.0003, "step": 15625 }, { "epoch": 55.2, "learning_rate": 2.2348717948717948e-07, "loss": 0.0003, "step": 15650 }, { "epoch": 55.29, "learning_rate": 2.2220512820512818e-07, "loss": 0.0003, "step": 15675 }, { "epoch": 55.38, "learning_rate": 2.209230769230769e-07, "loss": 0.0003, "step": 15700 }, { "epoch": 55.47, "learning_rate": 2.1964102564102563e-07, "loss": 0.0003, "step": 15725 }, { "epoch": 55.56, "learning_rate": 2.1835897435897435e-07, "loss": 0.0003, "step": 15750 }, { "epoch": 55.64, "learning_rate": 2.1707692307692308e-07, "loss": 0.0003, "step": 15775 }, { "epoch": 55.73, "learning_rate": 2.157948717948718e-07, "loss": 0.0003, "step": 15800 }, { "epoch": 55.82, "learning_rate": 2.145128205128205e-07, "loss": 0.0003, "step": 15825 }, { "epoch": 55.91, "learning_rate": 2.1323076923076922e-07, "loss": 0.0003, "step": 15850 }, { "epoch": 56.0, "learning_rate": 2.1194871794871795e-07, "loss": 0.0003, "step": 15875 }, { "epoch": 56.08, "learning_rate": 2.1066666666666665e-07, "loss": 0.0003, "step": 15900 }, { "epoch": 56.17, "learning_rate": 2.0938461538461537e-07, "loss": 0.0003, "step": 15925 }, { "epoch": 56.26, "learning_rate": 2.081025641025641e-07, "loss": 0.0003, "step": 15950 }, { "epoch": 56.35, "learning_rate": 2.068205128205128e-07, "loss": 0.0003, "step": 15975 }, { "epoch": 56.44, "learning_rate": 2.0553846153846152e-07, "loss": 0.0003, "step": 16000 }, { "epoch": 56.44, "eval_loss": 0.44896265864372253, "eval_runtime": 2257.6284, "eval_samples_per_second": 4.018, "eval_steps_per_second": 0.251, "eval_wer": 5.78483882627411, "step": 16000 }, { "epoch": 56.53, "learning_rate": 2.0425641025641024e-07, "loss": 0.0003, "step": 16025 }, { "epoch": 56.61, "learning_rate": 2.0297435897435897e-07, "loss": 0.0003, "step": 16050 }, { "epoch": 56.7, "learning_rate": 2.016923076923077e-07, "loss": 0.0003, "step": 16075 }, { "epoch": 56.79, "learning_rate": 2.0041025641025642e-07, "loss": 0.0003, "step": 16100 }, { "epoch": 56.88, "learning_rate": 1.9912820512820512e-07, "loss": 0.0003, "step": 16125 }, { "epoch": 56.97, "learning_rate": 1.9784615384615384e-07, "loss": 0.0003, "step": 16150 }, { "epoch": 57.05, "learning_rate": 1.9656410256410254e-07, "loss": 0.0003, "step": 16175 }, { "epoch": 57.14, "learning_rate": 1.9528205128205127e-07, "loss": 0.0003, "step": 16200 }, { "epoch": 57.23, "learning_rate": 1.94e-07, "loss": 0.0003, "step": 16225 }, { "epoch": 57.32, "learning_rate": 1.9271794871794871e-07, "loss": 0.0003, "step": 16250 }, { "epoch": 57.41, "learning_rate": 1.9143589743589741e-07, "loss": 0.0003, "step": 16275 }, { "epoch": 57.5, "learning_rate": 1.9015384615384614e-07, "loss": 0.0003, "step": 16300 }, { "epoch": 57.58, "learning_rate": 1.8887179487179486e-07, "loss": 0.0003, "step": 16325 }, { "epoch": 57.67, "learning_rate": 1.875897435897436e-07, "loss": 0.0003, "step": 16350 }, { "epoch": 57.76, "learning_rate": 1.863076923076923e-07, "loss": 0.0003, "step": 16375 }, { "epoch": 57.85, "learning_rate": 1.8502564102564104e-07, "loss": 0.0003, "step": 16400 }, { "epoch": 57.94, "learning_rate": 1.837435897435897e-07, "loss": 0.0003, "step": 16425 }, { "epoch": 58.02, "learning_rate": 1.8246153846153843e-07, "loss": 0.0003, "step": 16450 }, { "epoch": 58.11, "learning_rate": 1.8117948717948716e-07, "loss": 0.0003, "step": 16475 }, { "epoch": 58.2, "learning_rate": 1.7989743589743588e-07, "loss": 0.0003, "step": 16500 }, { "epoch": 58.29, "learning_rate": 1.786153846153846e-07, "loss": 0.0003, "step": 16525 }, { "epoch": 58.38, "learning_rate": 1.7733333333333333e-07, "loss": 0.0003, "step": 16550 }, { "epoch": 58.47, "learning_rate": 1.7605128205128203e-07, "loss": 0.0003, "step": 16575 }, { "epoch": 58.55, "learning_rate": 1.7476923076923076e-07, "loss": 0.0003, "step": 16600 }, { "epoch": 58.64, "learning_rate": 1.7348717948717948e-07, "loss": 0.0003, "step": 16625 }, { "epoch": 58.73, "learning_rate": 1.722051282051282e-07, "loss": 0.0003, "step": 16650 }, { "epoch": 58.82, "learning_rate": 1.7092307692307693e-07, "loss": 0.0003, "step": 16675 }, { "epoch": 58.91, "learning_rate": 1.6964102564102566e-07, "loss": 0.0003, "step": 16700 }, { "epoch": 58.99, "learning_rate": 1.6835897435897435e-07, "loss": 0.0003, "step": 16725 }, { "epoch": 59.08, "learning_rate": 1.6707692307692305e-07, "loss": 0.0003, "step": 16750 }, { "epoch": 59.17, "learning_rate": 1.6579487179487178e-07, "loss": 0.0003, "step": 16775 }, { "epoch": 59.26, "learning_rate": 1.645128205128205e-07, "loss": 0.0003, "step": 16800 }, { "epoch": 59.35, "learning_rate": 1.6323076923076923e-07, "loss": 0.0003, "step": 16825 }, { "epoch": 59.44, "learning_rate": 1.6194871794871795e-07, "loss": 0.0003, "step": 16850 }, { "epoch": 59.52, "learning_rate": 1.6066666666666668e-07, "loss": 0.0003, "step": 16875 }, { "epoch": 59.61, "learning_rate": 1.5938461538461537e-07, "loss": 0.0003, "step": 16900 }, { "epoch": 59.7, "learning_rate": 1.581025641025641e-07, "loss": 0.0003, "step": 16925 }, { "epoch": 59.79, "learning_rate": 1.5682051282051282e-07, "loss": 0.0003, "step": 16950 }, { "epoch": 59.88, "learning_rate": 1.5553846153846152e-07, "loss": 0.0003, "step": 16975 }, { "epoch": 59.96, "learning_rate": 1.5425641025641025e-07, "loss": 0.0003, "step": 17000 }, { "epoch": 59.96, "eval_loss": 0.45534029603004456, "eval_runtime": 2259.4705, "eval_samples_per_second": 4.015, "eval_steps_per_second": 0.251, "eval_wer": 5.801268360003943, "step": 17000 }, { "epoch": 60.05, "learning_rate": 1.5297435897435897e-07, "loss": 0.0003, "step": 17025 }, { "epoch": 60.14, "learning_rate": 1.5169230769230767e-07, "loss": 0.0003, "step": 17050 }, { "epoch": 60.23, "learning_rate": 1.504102564102564e-07, "loss": 0.0003, "step": 17075 }, { "epoch": 60.32, "learning_rate": 1.4912820512820512e-07, "loss": 0.0003, "step": 17100 }, { "epoch": 60.41, "learning_rate": 1.4784615384615384e-07, "loss": 0.0003, "step": 17125 }, { "epoch": 60.49, "learning_rate": 1.4656410256410257e-07, "loss": 0.0003, "step": 17150 }, { "epoch": 60.58, "learning_rate": 1.452820512820513e-07, "loss": 0.0003, "step": 17175 }, { "epoch": 60.67, "learning_rate": 1.44e-07, "loss": 0.0003, "step": 17200 }, { "epoch": 60.76, "learning_rate": 1.4271794871794872e-07, "loss": 0.0003, "step": 17225 }, { "epoch": 60.85, "learning_rate": 1.4143589743589742e-07, "loss": 0.0003, "step": 17250 }, { "epoch": 60.93, "learning_rate": 1.4015384615384614e-07, "loss": 0.0003, "step": 17275 }, { "epoch": 61.02, "learning_rate": 1.3887179487179487e-07, "loss": 0.0003, "step": 17300 }, { "epoch": 61.11, "learning_rate": 1.375897435897436e-07, "loss": 0.0003, "step": 17325 }, { "epoch": 61.2, "learning_rate": 1.363076923076923e-07, "loss": 0.0003, "step": 17350 }, { "epoch": 61.29, "learning_rate": 1.35025641025641e-07, "loss": 0.0003, "step": 17375 }, { "epoch": 61.38, "learning_rate": 1.3374358974358974e-07, "loss": 0.0003, "step": 17400 }, { "epoch": 61.46, "learning_rate": 1.3246153846153846e-07, "loss": 0.0003, "step": 17425 }, { "epoch": 61.55, "learning_rate": 1.311794871794872e-07, "loss": 0.0003, "step": 17450 }, { "epoch": 61.64, "learning_rate": 1.298974358974359e-07, "loss": 0.0003, "step": 17475 }, { "epoch": 61.73, "learning_rate": 1.286153846153846e-07, "loss": 0.0003, "step": 17500 }, { "epoch": 61.82, "learning_rate": 1.273333333333333e-07, "loss": 0.0003, "step": 17525 }, { "epoch": 61.9, "learning_rate": 1.2605128205128203e-07, "loss": 0.0003, "step": 17550 }, { "epoch": 61.99, "learning_rate": 1.2476923076923076e-07, "loss": 0.0003, "step": 17575 }, { "epoch": 62.08, "learning_rate": 1.2348717948717948e-07, "loss": 0.0002, "step": 17600 }, { "epoch": 62.17, "learning_rate": 1.222051282051282e-07, "loss": 0.0003, "step": 17625 }, { "epoch": 62.26, "learning_rate": 1.209230769230769e-07, "loss": 0.0002, "step": 17650 }, { "epoch": 62.35, "learning_rate": 1.1964102564102563e-07, "loss": 0.0002, "step": 17675 }, { "epoch": 62.43, "learning_rate": 1.1835897435897436e-07, "loss": 0.0003, "step": 17700 }, { "epoch": 62.52, "learning_rate": 1.1707692307692307e-07, "loss": 0.0003, "step": 17725 }, { "epoch": 62.61, "learning_rate": 1.1579487179487179e-07, "loss": 0.0003, "step": 17750 }, { "epoch": 62.7, "learning_rate": 1.145128205128205e-07, "loss": 0.0003, "step": 17775 }, { "epoch": 62.79, "learning_rate": 1.1323076923076923e-07, "loss": 0.0003, "step": 17800 }, { "epoch": 62.87, "learning_rate": 1.1194871794871795e-07, "loss": 0.0003, "step": 17825 }, { "epoch": 62.96, "learning_rate": 1.1066666666666665e-07, "loss": 0.0003, "step": 17850 }, { "epoch": 63.05, "learning_rate": 1.0938461538461538e-07, "loss": 0.0002, "step": 17875 }, { "epoch": 63.14, "learning_rate": 1.081025641025641e-07, "loss": 0.0002, "step": 17900 }, { "epoch": 63.23, "learning_rate": 1.0682051282051281e-07, "loss": 0.0002, "step": 17925 }, { "epoch": 63.32, "learning_rate": 1.0553846153846154e-07, "loss": 0.0002, "step": 17950 }, { "epoch": 63.4, "learning_rate": 1.0425641025641026e-07, "loss": 0.0003, "step": 17975 }, { "epoch": 63.49, "learning_rate": 1.0297435897435896e-07, "loss": 0.0002, "step": 18000 }, { "epoch": 63.49, "eval_loss": 0.46248582005500793, "eval_runtime": 2251.3847, "eval_samples_per_second": 4.03, "eval_steps_per_second": 0.252, "eval_wer": 5.7963394998849935, "step": 18000 }, { "epoch": 63.58, "learning_rate": 1.0169230769230769e-07, "loss": 0.0002, "step": 18025 }, { "epoch": 63.67, "learning_rate": 1.0041025641025641e-07, "loss": 0.0002, "step": 18050 }, { "epoch": 63.76, "learning_rate": 9.912820512820512e-08, "loss": 0.0002, "step": 18075 }, { "epoch": 63.84, "learning_rate": 9.784615384615385e-08, "loss": 0.0002, "step": 18100 }, { "epoch": 63.93, "learning_rate": 9.656410256410256e-08, "loss": 0.0002, "step": 18125 }, { "epoch": 64.02, "learning_rate": 9.528205128205127e-08, "loss": 0.0002, "step": 18150 }, { "epoch": 64.11, "learning_rate": 9.4e-08, "loss": 0.0002, "step": 18175 }, { "epoch": 64.2, "learning_rate": 9.271794871794872e-08, "loss": 0.0002, "step": 18200 }, { "epoch": 64.29, "learning_rate": 9.143589743589743e-08, "loss": 0.0002, "step": 18225 }, { "epoch": 64.37, "learning_rate": 9.015384615384616e-08, "loss": 0.0002, "step": 18250 }, { "epoch": 64.46, "learning_rate": 8.887179487179487e-08, "loss": 0.0002, "step": 18275 }, { "epoch": 64.55, "learning_rate": 8.758974358974358e-08, "loss": 0.0002, "step": 18300 }, { "epoch": 64.64, "learning_rate": 8.63076923076923e-08, "loss": 0.0002, "step": 18325 }, { "epoch": 64.73, "learning_rate": 8.502564102564103e-08, "loss": 0.0002, "step": 18350 }, { "epoch": 64.81, "learning_rate": 8.374358974358974e-08, "loss": 0.0002, "step": 18375 }, { "epoch": 64.9, "learning_rate": 8.246153846153845e-08, "loss": 0.0002, "step": 18400 }, { "epoch": 64.99, "learning_rate": 8.117948717948718e-08, "loss": 0.0002, "step": 18425 }, { "epoch": 65.08, "learning_rate": 7.989743589743589e-08, "loss": 0.0002, "step": 18450 }, { "epoch": 65.17, "learning_rate": 7.861538461538461e-08, "loss": 0.0002, "step": 18475 }, { "epoch": 65.26, "learning_rate": 7.733333333333334e-08, "loss": 0.0002, "step": 18500 }, { "epoch": 65.34, "learning_rate": 7.605128205128204e-08, "loss": 0.0002, "step": 18525 }, { "epoch": 65.43, "learning_rate": 7.476923076923076e-08, "loss": 0.0002, "step": 18550 }, { "epoch": 65.52, "learning_rate": 7.348717948717949e-08, "loss": 0.0002, "step": 18575 }, { "epoch": 65.61, "learning_rate": 7.22051282051282e-08, "loss": 0.0002, "step": 18600 }, { "epoch": 65.7, "learning_rate": 7.092307692307692e-08, "loss": 0.0002, "step": 18625 }, { "epoch": 65.78, "learning_rate": 6.964102564102565e-08, "loss": 0.0002, "step": 18650 }, { "epoch": 65.87, "learning_rate": 6.835897435897434e-08, "loss": 0.0002, "step": 18675 }, { "epoch": 65.96, "learning_rate": 6.707692307692307e-08, "loss": 0.0002, "step": 18700 }, { "epoch": 66.05, "learning_rate": 6.57948717948718e-08, "loss": 0.0002, "step": 18725 }, { "epoch": 66.14, "learning_rate": 6.45128205128205e-08, "loss": 0.0002, "step": 18750 }, { "epoch": 66.23, "learning_rate": 6.323076923076923e-08, "loss": 0.0002, "step": 18775 }, { "epoch": 66.31, "learning_rate": 6.194871794871794e-08, "loss": 0.0002, "step": 18800 }, { "epoch": 66.4, "learning_rate": 6.066666666666667e-08, "loss": 0.0002, "step": 18825 }, { "epoch": 66.49, "learning_rate": 5.938461538461538e-08, "loss": 0.0002, "step": 18850 }, { "epoch": 66.58, "learning_rate": 5.8102564102564104e-08, "loss": 0.0002, "step": 18875 }, { "epoch": 66.67, "learning_rate": 5.6820512820512815e-08, "loss": 0.0002, "step": 18900 }, { "epoch": 66.75, "learning_rate": 5.5538461538461533e-08, "loss": 0.0002, "step": 18925 }, { "epoch": 66.84, "learning_rate": 5.425641025641026e-08, "loss": 0.0002, "step": 18950 }, { "epoch": 66.93, "learning_rate": 5.297435897435897e-08, "loss": 0.0002, "step": 18975 }, { "epoch": 67.02, "learning_rate": 5.169230769230769e-08, "loss": 0.0002, "step": 19000 }, { "epoch": 67.02, "eval_loss": 0.466307133436203, "eval_runtime": 2249.6805, "eval_samples_per_second": 4.033, "eval_steps_per_second": 0.252, "eval_wer": 5.852199914566424, "step": 19000 }, { "epoch": 67.11, "learning_rate": 5.041025641025641e-08, "loss": 0.0002, "step": 19025 }, { "epoch": 67.2, "learning_rate": 4.9128205128205124e-08, "loss": 0.0002, "step": 19050 }, { "epoch": 67.28, "learning_rate": 4.784615384615384e-08, "loss": 0.0002, "step": 19075 }, { "epoch": 67.37, "learning_rate": 4.656410256410257e-08, "loss": 0.0002, "step": 19100 }, { "epoch": 67.46, "learning_rate": 4.528205128205128e-08, "loss": 0.0002, "step": 19125 }, { "epoch": 67.55, "learning_rate": 4.4e-08, "loss": 0.0002, "step": 19150 }, { "epoch": 67.64, "learning_rate": 4.2717948717948715e-08, "loss": 0.0002, "step": 19175 }, { "epoch": 67.72, "learning_rate": 4.143589743589743e-08, "loss": 0.0002, "step": 19200 }, { "epoch": 67.81, "learning_rate": 4.015384615384615e-08, "loss": 0.0002, "step": 19225 }, { "epoch": 67.9, "learning_rate": 3.887179487179487e-08, "loss": 0.0002, "step": 19250 }, { "epoch": 67.99, "learning_rate": 3.758974358974359e-08, "loss": 0.0002, "step": 19275 }, { "epoch": 68.08, "learning_rate": 3.6307692307692306e-08, "loss": 0.0002, "step": 19300 }, { "epoch": 68.17, "learning_rate": 3.5025641025641024e-08, "loss": 0.0002, "step": 19325 }, { "epoch": 68.25, "learning_rate": 3.374358974358974e-08, "loss": 0.0002, "step": 19350 }, { "epoch": 68.34, "learning_rate": 3.2461538461538454e-08, "loss": 0.0002, "step": 19375 }, { "epoch": 68.43, "learning_rate": 3.117948717948718e-08, "loss": 0.0002, "step": 19400 }, { "epoch": 68.52, "learning_rate": 2.98974358974359e-08, "loss": 0.0002, "step": 19425 }, { "epoch": 68.61, "learning_rate": 2.8615384615384615e-08, "loss": 0.0002, "step": 19450 }, { "epoch": 68.69, "learning_rate": 2.7333333333333333e-08, "loss": 0.0002, "step": 19475 }, { "epoch": 68.78, "learning_rate": 2.6051282051282048e-08, "loss": 0.0002, "step": 19500 }, { "epoch": 68.87, "learning_rate": 2.476923076923077e-08, "loss": 0.0002, "step": 19525 }, { "epoch": 68.96, "learning_rate": 2.3487179487179488e-08, "loss": 0.0002, "step": 19550 }, { "epoch": 69.05, "learning_rate": 2.2205128205128203e-08, "loss": 0.0002, "step": 19575 }, { "epoch": 69.14, "learning_rate": 2.092307692307692e-08, "loss": 0.0002, "step": 19600 }, { "epoch": 69.22, "learning_rate": 1.9641025641025642e-08, "loss": 0.0002, "step": 19625 }, { "epoch": 69.31, "learning_rate": 1.8358974358974357e-08, "loss": 0.0002, "step": 19650 }, { "epoch": 69.4, "learning_rate": 1.7076923076923075e-08, "loss": 0.0002, "step": 19675 }, { "epoch": 69.49, "learning_rate": 1.5794871794871794e-08, "loss": 0.0002, "step": 19700 }, { "epoch": 69.58, "learning_rate": 1.4512820512820512e-08, "loss": 0.0002, "step": 19725 }, { "epoch": 69.66, "learning_rate": 1.323076923076923e-08, "loss": 0.0002, "step": 19750 }, { "epoch": 69.75, "learning_rate": 1.1948717948717948e-08, "loss": 0.0002, "step": 19775 }, { "epoch": 69.84, "learning_rate": 1.0666666666666666e-08, "loss": 0.0002, "step": 19800 }, { "epoch": 69.93, "learning_rate": 9.384615384615384e-09, "loss": 0.0002, "step": 19825 }, { "epoch": 70.02, "learning_rate": 8.102564102564103e-09, "loss": 0.0002, "step": 19850 }, { "epoch": 70.11, "learning_rate": 6.820512820512821e-09, "loss": 0.0002, "step": 19875 }, { "epoch": 70.19, "learning_rate": 5.538461538461538e-09, "loss": 0.0002, "step": 19900 }, { "epoch": 70.28, "learning_rate": 4.256410256410256e-09, "loss": 0.0002, "step": 19925 }, { "epoch": 70.37, "learning_rate": 2.9743589743589745e-09, "loss": 0.0002, "step": 19950 }, { "epoch": 70.46, "learning_rate": 1.6923076923076922e-09, "loss": 0.0002, "step": 19975 }, { "epoch": 70.55, "learning_rate": 4.102564102564102e-10, "loss": 0.0002, "step": 20000 }, { "epoch": 70.55, "eval_loss": 0.46797338128089905, "eval_runtime": 2252.2761, "eval_samples_per_second": 4.028, "eval_steps_per_second": 0.252, "eval_wer": 5.875201261788191, "step": 20000 }, { "epoch": 70.55, "step": 20000, "total_flos": 2.717658906624e+21, "train_loss": 0.01680180715448223, "train_runtime": 273693.4473, "train_samples_per_second": 4.677, "train_steps_per_second": 0.073 } ], "logging_steps": 25, "max_steps": 20000, "num_input_tokens_seen": 0, "num_train_epochs": 71, "save_steps": 1000, "total_flos": 2.717658906624e+21, "train_batch_size": 32, "trial_name": null, "trial_params": null }