| { | |
| "best_metric": 16.318103103769815, | |
| "best_model_checkpoint": "./checkpoint-20000", | |
| "epoch": 8.0364, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 1.2203, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.6019, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.2567, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.2, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.2052, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.2403, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.2248, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.1989, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.1801, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.1493, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.976410256410257e-06, | |
| "loss": 0.1371, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.950769230769232e-06, | |
| "loss": 0.1455, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.925128205128206e-06, | |
| "loss": 0.1899, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.899487179487181e-06, | |
| "loss": 0.1778, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.873846153846155e-06, | |
| "loss": 0.1741, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.84820512820513e-06, | |
| "loss": 0.1721, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.822564102564104e-06, | |
| "loss": 0.165, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.796923076923077e-06, | |
| "loss": 0.1645, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.771282051282051e-06, | |
| "loss": 0.1572, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.745641025641026e-06, | |
| "loss": 0.1631, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.72e-06, | |
| "loss": 0.1295, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.694358974358975e-06, | |
| "loss": 0.1294, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.668717948717949e-06, | |
| "loss": 0.1376, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.643076923076924e-06, | |
| "loss": 0.1263, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.617435897435898e-06, | |
| "loss": 0.1242, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.591794871794873e-06, | |
| "loss": 0.15, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.566153846153847e-06, | |
| "loss": 0.1288, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.540512820512822e-06, | |
| "loss": 0.1266, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.514871794871796e-06, | |
| "loss": 0.1351, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.48923076923077e-06, | |
| "loss": 0.1834, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.463589743589743e-06, | |
| "loss": 0.1825, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.437948717948719e-06, | |
| "loss": 0.199, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.412307692307692e-06, | |
| "loss": 0.1937, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.386666666666668e-06, | |
| "loss": 0.1646, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.361025641025641e-06, | |
| "loss": 0.1683, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.335384615384617e-06, | |
| "loss": 0.1739, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.30974358974359e-06, | |
| "loss": 0.1759, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.284102564102566e-06, | |
| "loss": 0.1591, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.25846153846154e-06, | |
| "loss": 0.1646, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.232820512820515e-06, | |
| "loss": 0.1658, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.207179487179488e-06, | |
| "loss": 0.2146, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.181538461538464e-06, | |
| "loss": 0.182, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.155897435897436e-06, | |
| "loss": 0.148, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.130256410256411e-06, | |
| "loss": 0.1327, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.104615384615385e-06, | |
| "loss": 0.1309, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.07897435897436e-06, | |
| "loss": 0.1244, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.053333333333334e-06, | |
| "loss": 0.1468, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.027692307692309e-06, | |
| "loss": 0.139, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.002051282051283e-06, | |
| "loss": 0.1555, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.976410256410258e-06, | |
| "loss": 0.1563, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_cer": 5.173836959751708, | |
| "eval_loss": 0.25240397453308105, | |
| "eval_runtime": 2113.4886, | |
| "eval_samples_per_second": 4.237, | |
| "eval_steps_per_second": 0.53, | |
| "eval_wer": 19.857022334709825, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.950769230769232e-06, | |
| "loss": 0.1312, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.925128205128207e-06, | |
| "loss": 0.1127, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.89948717948718e-06, | |
| "loss": 0.0998, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.873846153846156e-06, | |
| "loss": 0.1194, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.848205128205128e-06, | |
| "loss": 0.0745, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.822564102564103e-06, | |
| "loss": 0.0757, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.796923076923077e-06, | |
| "loss": 0.066, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.771282051282052e-06, | |
| "loss": 0.0517, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.745641025641026e-06, | |
| "loss": 0.0501, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.720000000000001e-06, | |
| "loss": 0.039, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.694358974358975e-06, | |
| "loss": 0.0525, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.66871794871795e-06, | |
| "loss": 0.0562, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.643076923076924e-06, | |
| "loss": 0.0507, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.6174358974359e-06, | |
| "loss": 0.0443, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.591794871794873e-06, | |
| "loss": 0.0549, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.566153846153848e-06, | |
| "loss": 0.0544, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.54051282051282e-06, | |
| "loss": 0.0419, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.514871794871795e-06, | |
| "loss": 0.0386, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.489230769230769e-06, | |
| "loss": 0.0303, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.463589743589744e-06, | |
| "loss": 0.0385, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.437948717948718e-06, | |
| "loss": 0.0415, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.412307692307693e-06, | |
| "loss": 0.0405, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.386666666666667e-06, | |
| "loss": 0.0409, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.361025641025642e-06, | |
| "loss": 0.0419, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.335384615384616e-06, | |
| "loss": 0.0341, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.309743589743591e-06, | |
| "loss": 0.0321, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.284102564102565e-06, | |
| "loss": 0.0423, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.25846153846154e-06, | |
| "loss": 0.0517, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.232820512820512e-06, | |
| "loss": 0.0665, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.207179487179488e-06, | |
| "loss": 0.0779, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.181538461538461e-06, | |
| "loss": 0.0554, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.155897435897437e-06, | |
| "loss": 0.0486, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.13025641025641e-06, | |
| "loss": 0.0523, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.104615384615386e-06, | |
| "loss": 0.0579, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.07897435897436e-06, | |
| "loss": 0.0447, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.053333333333335e-06, | |
| "loss": 0.0479, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.027692307692308e-06, | |
| "loss": 0.049, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.002051282051284e-06, | |
| "loss": 0.04, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.976410256410257e-06, | |
| "loss": 0.0541, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.950769230769233e-06, | |
| "loss": 0.0549, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.925128205128205e-06, | |
| "loss": 0.0444, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.89948717948718e-06, | |
| "loss": 0.0335, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.873846153846154e-06, | |
| "loss": 0.0327, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.848205128205129e-06, | |
| "loss": 0.0449, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.822564102564103e-06, | |
| "loss": 0.0433, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.796923076923078e-06, | |
| "loss": 0.0509, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.771282051282052e-06, | |
| "loss": 0.0521, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.745641025641027e-06, | |
| "loss": 0.0503, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.72e-06, | |
| "loss": 0.0469, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.694358974358976e-06, | |
| "loss": 0.032, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "eval_cer": 4.7792782381880015, | |
| "eval_loss": 0.25666213035583496, | |
| "eval_runtime": 2116.6961, | |
| "eval_samples_per_second": 4.23, | |
| "eval_steps_per_second": 0.529, | |
| "eval_wer": 18.562674836818967, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.66871794871795e-06, | |
| "loss": 0.0277, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.643076923076925e-06, | |
| "loss": 0.0396, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.617435897435898e-06, | |
| "loss": 0.0299, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.591794871794872e-06, | |
| "loss": 0.0239, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.566153846153847e-06, | |
| "loss": 0.0228, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.540512820512821e-06, | |
| "loss": 0.0169, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.514871794871795e-06, | |
| "loss": 0.0157, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.489230769230769e-06, | |
| "loss": 0.0111, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.463589743589744e-06, | |
| "loss": 0.0141, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.437948717948718e-06, | |
| "loss": 0.0186, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.412307692307693e-06, | |
| "loss": 0.0151, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.386666666666667e-06, | |
| "loss": 0.0153, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.361025641025642e-06, | |
| "loss": 0.019, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.335384615384616e-06, | |
| "loss": 0.019, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.309743589743591e-06, | |
| "loss": 0.0168, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.2841025641025645e-06, | |
| "loss": 0.0155, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.258461538461539e-06, | |
| "loss": 0.015, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.2328205128205135e-06, | |
| "loss": 0.0153, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.207179487179487e-06, | |
| "loss": 0.0133, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.181538461538462e-06, | |
| "loss": 0.0109, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.155897435897436e-06, | |
| "loss": 0.0094, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.130256410256411e-06, | |
| "loss": 0.0103, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.104615384615385e-06, | |
| "loss": 0.0111, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.07897435897436e-06, | |
| "loss": 0.0119, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.053846153846155e-06, | |
| "loss": 0.0123, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.028205128205129e-06, | |
| "loss": 0.0158, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.002564102564104e-06, | |
| "loss": 0.0168, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.976923076923078e-06, | |
| "loss": 0.0271, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.951282051282052e-06, | |
| "loss": 0.02, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.925641025641026e-06, | |
| "loss": 0.0206, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.0169, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.8743589743589745e-06, | |
| "loss": 0.0139, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.848717948717949e-06, | |
| "loss": 0.0218, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.8230769230769235e-06, | |
| "loss": 0.0159, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.797435897435898e-06, | |
| "loss": 0.0175, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.7717948717948725e-06, | |
| "loss": 0.0161, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.746153846153847e-06, | |
| "loss": 0.0171, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.7205128205128215e-06, | |
| "loss": 0.0119, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.694871794871796e-06, | |
| "loss": 0.0178, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.6692307692307705e-06, | |
| "loss": 0.0138, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.643589743589744e-06, | |
| "loss": 0.0152, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.617948717948719e-06, | |
| "loss": 0.0149, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.592307692307692e-06, | |
| "loss": 0.0105, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.0234, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.541025641025641e-06, | |
| "loss": 0.0155, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.515384615384616e-06, | |
| "loss": 0.0181, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.48974358974359e-06, | |
| "loss": 0.0149, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.464102564102565e-06, | |
| "loss": 0.0127, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.438461538461539e-06, | |
| "loss": 0.0131, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.412820512820514e-06, | |
| "loss": 0.013, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "eval_cer": 4.66635850364843, | |
| "eval_loss": 0.26371335983276367, | |
| "eval_runtime": 2122.3121, | |
| "eval_samples_per_second": 4.219, | |
| "eval_steps_per_second": 0.528, | |
| "eval_wer": 17.772301407575153, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.387179487179488e-06, | |
| "loss": 0.0121, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.361538461538463e-06, | |
| "loss": 0.0136, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.335897435897436e-06, | |
| "loss": 0.0053, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.310256410256411e-06, | |
| "loss": 0.0104, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.284615384615385e-06, | |
| "loss": 0.0042, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.258974358974359e-06, | |
| "loss": 0.0073, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.0068, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.207692307692308e-06, | |
| "loss": 0.0047, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.1820512820512825e-06, | |
| "loss": 0.0069, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.156410256410257e-06, | |
| "loss": 0.0071, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.1307692307692315e-06, | |
| "loss": 0.0084, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.105128205128206e-06, | |
| "loss": 0.0102, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.0794871794871805e-06, | |
| "loss": 0.0066, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.053846153846155e-06, | |
| "loss": 0.0072, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.028205128205129e-06, | |
| "loss": 0.0042, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.002564102564103e-06, | |
| "loss": 0.0024, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.976923076923078e-06, | |
| "loss": 0.0064, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.951282051282051e-06, | |
| "loss": 0.0068, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.925641025641026e-06, | |
| "loss": 0.0044, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.0045, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.8748717948717945e-06, | |
| "loss": 0.009, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.849230769230769e-06, | |
| "loss": 0.0037, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.8235897435897435e-06, | |
| "loss": 0.0042, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.797948717948718e-06, | |
| "loss": 0.0108, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.7723076923076925e-06, | |
| "loss": 0.0053, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.746666666666667e-06, | |
| "loss": 0.0061, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.7210256410256415e-06, | |
| "loss": 0.01, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.695384615384616e-06, | |
| "loss": 0.0075, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.6697435897435905e-06, | |
| "loss": 0.0065, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.644102564102565e-06, | |
| "loss": 0.0087, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.6184615384615395e-06, | |
| "loss": 0.0096, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.592820512820514e-06, | |
| "loss": 0.0054, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.567179487179487e-06, | |
| "loss": 0.0045, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.541538461538461e-06, | |
| "loss": 0.0069, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.515897435897436e-06, | |
| "loss": 0.0055, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.49025641025641e-06, | |
| "loss": 0.0041, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.464615384615385e-06, | |
| "loss": 0.0051, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.438974358974359e-06, | |
| "loss": 0.0047, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.413333333333334e-06, | |
| "loss": 0.0034, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.387692307692308e-06, | |
| "loss": 0.005, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.362051282051283e-06, | |
| "loss": 0.003, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.336410256410257e-06, | |
| "loss": 0.0102, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.310769230769232e-06, | |
| "loss": 0.0043, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.285128205128206e-06, | |
| "loss": 0.009, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.259487179487181e-06, | |
| "loss": 0.0049, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.2338461538461535e-06, | |
| "loss": 0.006, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.208205128205128e-06, | |
| "loss": 0.0048, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.1825641025641025e-06, | |
| "loss": 0.0054, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.156923076923077e-06, | |
| "loss": 0.0043, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.1312820512820515e-06, | |
| "loss": 0.0057, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "eval_cer": 4.366229735530095, | |
| "eval_loss": 0.27028778195381165, | |
| "eval_runtime": 2186.2317, | |
| "eval_samples_per_second": 4.096, | |
| "eval_steps_per_second": 0.512, | |
| "eval_wer": 17.05963323120643, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.105641025641026e-06, | |
| "loss": 0.0044, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.0800000000000005e-06, | |
| "loss": 0.0035, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.054358974358975e-06, | |
| "loss": 0.0025, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.0287179487179495e-06, | |
| "loss": 0.0039, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.003076923076924e-06, | |
| "loss": 0.0018, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.977435897435898e-06, | |
| "loss": 0.0024, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.951794871794872e-06, | |
| "loss": 0.0013, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.926153846153847e-06, | |
| "loss": 0.0029, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.900512820512821e-06, | |
| "loss": 0.0036, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.874871794871796e-06, | |
| "loss": 0.003, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.849230769230769e-06, | |
| "loss": 0.0032, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.823589743589744e-06, | |
| "loss": 0.0031, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.797948717948718e-06, | |
| "loss": 0.0054, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.772307692307693e-06, | |
| "loss": 0.0044, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.746666666666667e-06, | |
| "loss": 0.0016, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.721025641025642e-06, | |
| "loss": 0.0019, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.695384615384615e-06, | |
| "loss": 0.0012, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.66974358974359e-06, | |
| "loss": 0.0013, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.644615384615385e-06, | |
| "loss": 0.0043, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.6189743589743595e-06, | |
| "loss": 0.0023, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.593333333333333e-06, | |
| "loss": 0.0011, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.567692307692308e-06, | |
| "loss": 0.0028, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.542051282051282e-06, | |
| "loss": 0.0046, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.516410256410257e-06, | |
| "loss": 0.004, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.490769230769231e-06, | |
| "loss": 0.004, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.465128205128206e-06, | |
| "loss": 0.0035, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.439487179487179e-06, | |
| "loss": 0.0021, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.413846153846154e-06, | |
| "loss": 0.003, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.388205128205128e-06, | |
| "loss": 0.0057, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.362564102564103e-06, | |
| "loss": 0.0019, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.336923076923077e-06, | |
| "loss": 0.0032, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.311282051282052e-06, | |
| "loss": 0.0016, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.285641025641025e-06, | |
| "loss": 0.0024, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.26e-06, | |
| "loss": 0.0031, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.234358974358974e-06, | |
| "loss": 0.0034, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.208717948717949e-06, | |
| "loss": 0.0011, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.183076923076923e-06, | |
| "loss": 0.0014, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.157435897435898e-06, | |
| "loss": 0.0031, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 4.131794871794872e-06, | |
| "loss": 0.0022, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 4.106153846153846e-06, | |
| "loss": 0.0025, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 4.0805128205128205e-06, | |
| "loss": 0.0029, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 4.054871794871795e-06, | |
| "loss": 0.0033, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 4.0292307692307695e-06, | |
| "loss": 0.0044, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 4.003589743589744e-06, | |
| "loss": 0.0018, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.9779487179487185e-06, | |
| "loss": 0.0027, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.952307692307692e-06, | |
| "loss": 0.0022, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.926666666666667e-06, | |
| "loss": 0.0028, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.901025641025641e-06, | |
| "loss": 0.003, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.875384615384616e-06, | |
| "loss": 0.0019, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.84974358974359e-06, | |
| "loss": 0.0012, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "eval_cer": 5.2286459537095125, | |
| "eval_loss": 0.2696499526500702, | |
| "eval_runtime": 2194.6715, | |
| "eval_samples_per_second": 4.08, | |
| "eval_steps_per_second": 0.51, | |
| "eval_wer": 17.832245459793082, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.824102564102565e-06, | |
| "loss": 0.0012, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.7984615384615387e-06, | |
| "loss": 0.0023, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.772820512820513e-06, | |
| "loss": 0.0007, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.7471794871794877e-06, | |
| "loss": 0.0011, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.721538461538462e-06, | |
| "loss": 0.0042, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.6958974358974362e-06, | |
| "loss": 0.0021, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.6702564102564103e-06, | |
| "loss": 0.0011, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.644615384615385e-06, | |
| "loss": 0.0005, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.6189743589743593e-06, | |
| "loss": 0.0009, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.593333333333334e-06, | |
| "loss": 0.0017, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.5676923076923083e-06, | |
| "loss": 0.0009, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.5420512820512824e-06, | |
| "loss": 0.0013, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.516923076923077e-06, | |
| "loss": 0.0044, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.4912820512820515e-06, | |
| "loss": 0.0006, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.465641025641026e-06, | |
| "loss": 0.001, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.0017, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.4143589743589746e-06, | |
| "loss": 0.0009, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.388717948717949e-06, | |
| "loss": 0.0021, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.363076923076923e-06, | |
| "loss": 0.0021, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.3374358974358977e-06, | |
| "loss": 0.001, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.311794871794872e-06, | |
| "loss": 0.0014, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.2861538461538467e-06, | |
| "loss": 0.0009, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.2605128205128207e-06, | |
| "loss": 0.0017, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.2348717948717952e-06, | |
| "loss": 0.0038, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.2092307692307693e-06, | |
| "loss": 0.0009, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.1835897435897438e-06, | |
| "loss": 0.0022, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.1579487179487183e-06, | |
| "loss": 0.0015, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.1323076923076928e-06, | |
| "loss": 0.0015, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.106666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.0810256410256413e-06, | |
| "loss": 0.0007, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.0553846153846154e-06, | |
| "loss": 0.0006, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.02974358974359e-06, | |
| "loss": 0.001, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.0041025641025644e-06, | |
| "loss": 0.0006, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.978461538461539e-06, | |
| "loss": 0.0005, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.952820512820513e-06, | |
| "loss": 0.0005, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.9271794871794875e-06, | |
| "loss": 0.0027, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.9015384615384615e-06, | |
| "loss": 0.0004, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.875897435897436e-06, | |
| "loss": 0.0011, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.8502564102564105e-06, | |
| "loss": 0.0024, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.824615384615385e-06, | |
| "loss": 0.0012, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.798974358974359e-06, | |
| "loss": 0.0016, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.7733333333333336e-06, | |
| "loss": 0.0008, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.7476923076923077e-06, | |
| "loss": 0.0044, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.722051282051282e-06, | |
| "loss": 0.001, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.6964102564102567e-06, | |
| "loss": 0.0007, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.670769230769231e-06, | |
| "loss": 0.0008, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.6451282051282052e-06, | |
| "loss": 0.0007, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.6194871794871797e-06, | |
| "loss": 0.0014, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.593846153846154e-06, | |
| "loss": 0.0006, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.5682051282051283e-06, | |
| "loss": 0.003, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "eval_cer": 4.297223231089246, | |
| "eval_loss": 0.28001004457473755, | |
| "eval_runtime": 2165.3419, | |
| "eval_samples_per_second": 4.135, | |
| "eval_steps_per_second": 0.517, | |
| "eval_wer": 16.719950268638158, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.542564102564103e-06, | |
| "loss": 0.0003, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.5169230769230773e-06, | |
| "loss": 0.0004, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.4912820512820514e-06, | |
| "loss": 0.0008, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.465641025641026e-06, | |
| "loss": 0.0004, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.0003, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.4143589743589744e-06, | |
| "loss": 0.0004, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.388717948717949e-06, | |
| "loss": 0.0003, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 2.3630769230769234e-06, | |
| "loss": 0.0006, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 2.3374358974358975e-06, | |
| "loss": 0.0003, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 2.311794871794872e-06, | |
| "loss": 0.0006, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 2.2861538461538465e-06, | |
| "loss": 0.0003, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.2605128205128206e-06, | |
| "loss": 0.0003, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.234871794871795e-06, | |
| "loss": 0.0003, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.2092307692307695e-06, | |
| "loss": 0.0003, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.1835897435897436e-06, | |
| "loss": 0.0007, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 2.157948717948718e-06, | |
| "loss": 0.0004, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 2.1323076923076926e-06, | |
| "loss": 0.0003, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 2.1066666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 2.081025641025641e-06, | |
| "loss": 0.0006, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 2.0553846153846157e-06, | |
| "loss": 0.0005, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 2.0297435897435897e-06, | |
| "loss": 0.0034, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 2.0041025641025642e-06, | |
| "loss": 0.0008, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.9784615384615387e-06, | |
| "loss": 0.0004, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.952820512820513e-06, | |
| "loss": 0.001, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.9271794871794873e-06, | |
| "loss": 0.0003, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.9015384615384616e-06, | |
| "loss": 0.0003, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.875897435897436e-06, | |
| "loss": 0.0003, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.8502564102564106e-06, | |
| "loss": 0.0002, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.8246153846153847e-06, | |
| "loss": 0.0003, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.7989743589743592e-06, | |
| "loss": 0.0006, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.7733333333333336e-06, | |
| "loss": 0.0004, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.7476923076923077e-06, | |
| "loss": 0.0003, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.7220512820512822e-06, | |
| "loss": 0.0002, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.6964102564102567e-06, | |
| "loss": 0.0002, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.6707692307692308e-06, | |
| "loss": 0.0002, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.6451282051282053e-06, | |
| "loss": 0.0002, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.6194871794871798e-06, | |
| "loss": 0.002, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.5938461538461538e-06, | |
| "loss": 0.0003, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.5682051282051283e-06, | |
| "loss": 0.0005, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.5425641025641028e-06, | |
| "loss": 0.0003, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.516923076923077e-06, | |
| "loss": 0.0003, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.4912820512820514e-06, | |
| "loss": 0.0014, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.465641025641026e-06, | |
| "loss": 0.0003, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.0004, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.4143589743589745e-06, | |
| "loss": 0.0002, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.388717948717949e-06, | |
| "loss": 0.0002, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.363076923076923e-06, | |
| "loss": 0.0003, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.3374358974358975e-06, | |
| "loss": 0.0002, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.311794871794872e-06, | |
| "loss": 0.0002, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.286153846153846e-06, | |
| "loss": 0.0003, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "eval_cer": 4.201802753656684, | |
| "eval_loss": 0.2833803594112396, | |
| "eval_runtime": 2120.5198, | |
| "eval_samples_per_second": 4.223, | |
| "eval_steps_per_second": 0.528, | |
| "eval_wer": 16.409129257137785, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.2605128205128206e-06, | |
| "loss": 0.0002, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.234871794871795e-06, | |
| "loss": 0.0002, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.2092307692307694e-06, | |
| "loss": 0.001, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.1835897435897437e-06, | |
| "loss": 0.0013, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.1579487179487182e-06, | |
| "loss": 0.0003, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.1323076923076924e-06, | |
| "loss": 0.0002, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.1066666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.0810256410256412e-06, | |
| "loss": 0.0001, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.0553846153846155e-06, | |
| "loss": 0.0002, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.0297435897435898e-06, | |
| "loss": 0.0002, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.0041025641025643e-06, | |
| "loss": 0.0002, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 9.784615384615386e-07, | |
| "loss": 0.0002, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 9.528205128205129e-07, | |
| "loss": 0.0002, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 9.271794871794874e-07, | |
| "loss": 0.0002, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 9.015384615384616e-07, | |
| "loss": 0.0002, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 8.758974358974359e-07, | |
| "loss": 0.0002, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 8.502564102564104e-07, | |
| "loss": 0.0002, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 8.246153846153847e-07, | |
| "loss": 0.0002, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 7.98974358974359e-07, | |
| "loss": 0.0011, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 7.733333333333335e-07, | |
| "loss": 0.0002, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 7.476923076923078e-07, | |
| "loss": 0.0008, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 7.22051282051282e-07, | |
| "loss": 0.0002, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 6.964102564102565e-07, | |
| "loss": 0.0021, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 6.707692307692308e-07, | |
| "loss": 0.0002, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 6.451282051282051e-07, | |
| "loss": 0.0002, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 6.194871794871796e-07, | |
| "loss": 0.0002, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 5.938461538461539e-07, | |
| "loss": 0.0002, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 5.682051282051283e-07, | |
| "loss": 0.0002, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 5.425641025641027e-07, | |
| "loss": 0.0002, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 5.16923076923077e-07, | |
| "loss": 0.0002, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 4.912820512820513e-07, | |
| "loss": 0.0002, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 4.656410256410257e-07, | |
| "loss": 0.0001, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 4.4e-07, | |
| "loss": 0.0001, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 4.143589743589744e-07, | |
| "loss": 0.0001, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 3.8923076923076924e-07, | |
| "loss": 0.0008, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 3.6358974358974363e-07, | |
| "loss": 0.0002, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 3.37948717948718e-07, | |
| "loss": 0.0002, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 3.1230769230769236e-07, | |
| "loss": 0.0002, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.6102564102564103e-07, | |
| "loss": 0.0002, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.3538461538461542e-07, | |
| "loss": 0.002, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.0974358974358976e-07, | |
| "loss": 0.0002, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 1.841025641025641e-07, | |
| "loss": 0.0002, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 1.5846153846153848e-07, | |
| "loss": 0.0002, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 1.3282051282051282e-07, | |
| "loss": 0.0001, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 1.0717948717948718e-07, | |
| "loss": 0.0001, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 8.153846153846155e-08, | |
| "loss": 0.0001, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 5.58974358974359e-08, | |
| "loss": 0.0002, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 3.025641025641026e-08, | |
| "loss": 0.0002, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 4.615384615384616e-09, | |
| "loss": 0.0002, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "eval_cer": 4.145012711724502, | |
| "eval_loss": 0.2859713137149811, | |
| "eval_runtime": 2113.5111, | |
| "eval_samples_per_second": 4.237, | |
| "eval_steps_per_second": 0.53, | |
| "eval_wer": 16.318103103769815, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "step": 20000, | |
| "total_flos": 4.61621206351872e+19, | |
| "train_loss": 0.0347061085105408, | |
| "train_runtime": 57083.3768, | |
| "train_samples_per_second": 2.803, | |
| "train_steps_per_second": 0.35 | |
| } | |
| ], | |
| "max_steps": 20000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 4.61621206351872e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |