| { | |
| "best_metric": 16.3839925843315, | |
| "best_model_checkpoint": "./checkpoint-9000", | |
| "epoch": 1.0, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 0.6709, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 0.596, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.4857, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.3658, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.4504, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.4356, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.4217, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.4439, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.3948, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.3626, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.3838, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.4015, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.3427, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.3582, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.3408, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.3252, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.3157, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.3813, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.3489, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.3676, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.975789473684211e-06, | |
| "loss": 0.3151, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.949473684210526e-06, | |
| "loss": 0.3342, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.923157894736844e-06, | |
| "loss": 0.3241, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.89684210526316e-06, | |
| "loss": 0.2749, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.870526315789474e-06, | |
| "loss": 0.3745, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.84421052631579e-06, | |
| "loss": 0.3269, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.817894736842106e-06, | |
| "loss": 0.3545, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.791578947368422e-06, | |
| "loss": 0.3464, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.765263157894737e-06, | |
| "loss": 0.3367, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.738947368421054e-06, | |
| "loss": 0.3749, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.712631578947369e-06, | |
| "loss": 0.3322, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.686315789473684e-06, | |
| "loss": 0.3304, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.3169, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.633684210526316e-06, | |
| "loss": 0.3407, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.607368421052632e-06, | |
| "loss": 0.363, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.581052631578947e-06, | |
| "loss": 0.329, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.554736842105264e-06, | |
| "loss": 0.3262, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.52842105263158e-06, | |
| "loss": 0.3153, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.502105263157896e-06, | |
| "loss": 0.3095, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.475789473684212e-06, | |
| "loss": 0.2963, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.9114714860916138, | |
| "eval_runtime": 6612.5212, | |
| "eval_samples_per_second": 0.855, | |
| "eval_steps_per_second": 0.054, | |
| "eval_wer": 27.364106426429906, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.449473684210527e-06, | |
| "loss": 0.3157, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.423157894736842e-06, | |
| "loss": 0.3128, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.396842105263159e-06, | |
| "loss": 0.3081, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.370526315789474e-06, | |
| "loss": 0.3011, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.34421052631579e-06, | |
| "loss": 0.3038, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.317894736842105e-06, | |
| "loss": 0.3061, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.291578947368422e-06, | |
| "loss": 0.3351, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.265263157894737e-06, | |
| "loss": 0.3328, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.238947368421052e-06, | |
| "loss": 0.3272, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.21263157894737e-06, | |
| "loss": 0.369, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.186315789473685e-06, | |
| "loss": 0.3143, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.16e-06, | |
| "loss": 0.2937, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.133684210526317e-06, | |
| "loss": 0.3536, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.107368421052632e-06, | |
| "loss": 0.3534, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.081052631578949e-06, | |
| "loss": 0.3063, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.054736842105264e-06, | |
| "loss": 0.3121, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.02842105263158e-06, | |
| "loss": 0.3115, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.002105263157895e-06, | |
| "loss": 0.3373, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.97578947368421e-06, | |
| "loss": 0.3083, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.949473684210527e-06, | |
| "loss": 0.2566, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.923157894736842e-06, | |
| "loss": 0.3019, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.896842105263159e-06, | |
| "loss": 0.2855, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.870526315789474e-06, | |
| "loss": 0.2961, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.84421052631579e-06, | |
| "loss": 0.2786, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.817894736842107e-06, | |
| "loss": 0.3422, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.791578947368422e-06, | |
| "loss": 0.2714, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.765263157894739e-06, | |
| "loss": 0.2986, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.738947368421053e-06, | |
| "loss": 0.264, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.712631578947368e-06, | |
| "loss": 0.3402, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.686315789473685e-06, | |
| "loss": 0.2863, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.66e-06, | |
| "loss": 0.258, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.633684210526317e-06, | |
| "loss": 0.3005, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.607368421052632e-06, | |
| "loss": 0.2873, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.581052631578948e-06, | |
| "loss": 0.3047, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.554736842105263e-06, | |
| "loss": 0.3019, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.528421052631578e-06, | |
| "loss": 0.3317, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.502105263157897e-06, | |
| "loss": 0.293, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.475789473684212e-06, | |
| "loss": 0.2969, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.449473684210527e-06, | |
| "loss": 0.2888, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.423157894736843e-06, | |
| "loss": 0.2676, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.8795959949493408, | |
| "eval_runtime": 6297.5185, | |
| "eval_samples_per_second": 0.898, | |
| "eval_steps_per_second": 0.056, | |
| "eval_wer": 24.10237098619701, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.396842105263158e-06, | |
| "loss": 0.248, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.370526315789475e-06, | |
| "loss": 0.2437, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.34421052631579e-06, | |
| "loss": 0.2451, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.317894736842107e-06, | |
| "loss": 0.2911, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.291578947368422e-06, | |
| "loss": 0.3174, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.265263157894737e-06, | |
| "loss": 0.3139, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.238947368421053e-06, | |
| "loss": 0.2514, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.212631578947368e-06, | |
| "loss": 0.2619, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.186315789473685e-06, | |
| "loss": 0.3192, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.16e-06, | |
| "loss": 0.3024, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.133684210526316e-06, | |
| "loss": 0.2978, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.107368421052633e-06, | |
| "loss": 0.3104, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.081052631578948e-06, | |
| "loss": 0.2983, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.054736842105265e-06, | |
| "loss": 0.2777, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.02842105263158e-06, | |
| "loss": 0.2971, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.002105263157895e-06, | |
| "loss": 0.2638, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.975789473684211e-06, | |
| "loss": 0.24, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.949473684210526e-06, | |
| "loss": 0.2579, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.923157894736843e-06, | |
| "loss": 0.2687, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.896842105263158e-06, | |
| "loss": 0.2591, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.870526315789475e-06, | |
| "loss": 0.3053, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.84421052631579e-06, | |
| "loss": 0.2626, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.817894736842105e-06, | |
| "loss": 0.2799, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.791578947368423e-06, | |
| "loss": 0.2293, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.765263157894738e-06, | |
| "loss": 0.251, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.738947368421053e-06, | |
| "loss": 0.3127, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.71263157894737e-06, | |
| "loss": 0.2738, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.686315789473685e-06, | |
| "loss": 0.2327, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 0.2591, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.633684210526316e-06, | |
| "loss": 0.2121, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.607368421052632e-06, | |
| "loss": 0.3053, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.581052631578948e-06, | |
| "loss": 0.2791, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.554736842105264e-06, | |
| "loss": 0.2519, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.5284210526315794e-06, | |
| "loss": 0.2564, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.502105263157895e-06, | |
| "loss": 0.2464, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.475789473684211e-06, | |
| "loss": 0.3187, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.449473684210526e-06, | |
| "loss": 0.3153, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.4231578947368436e-06, | |
| "loss": 0.2961, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.3968421052631585e-06, | |
| "loss": 0.2807, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.370526315789474e-06, | |
| "loss": 0.3166, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 0.8467275500297546, | |
| "eval_runtime": 5805.1114, | |
| "eval_samples_per_second": 0.974, | |
| "eval_steps_per_second": 0.061, | |
| "eval_wer": 20.17003896122706, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.34421052631579e-06, | |
| "loss": 0.2834, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.317894736842106e-06, | |
| "loss": 0.2622, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.291578947368422e-06, | |
| "loss": 0.2616, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.265263157894738e-06, | |
| "loss": 0.2829, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.2389473684210534e-06, | |
| "loss": 0.2634, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.212631578947369e-06, | |
| "loss": 0.2936, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.186315789473684e-06, | |
| "loss": 0.2626, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.16e-06, | |
| "loss": 0.3031, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.133684210526316e-06, | |
| "loss": 0.2826, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.107368421052632e-06, | |
| "loss": 0.2545, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.0810526315789475e-06, | |
| "loss": 0.248, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.054736842105264e-06, | |
| "loss": 0.2606, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.02842105263158e-06, | |
| "loss": 0.2495, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.002105263157896e-06, | |
| "loss": 0.2657, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.975789473684212e-06, | |
| "loss": 0.3054, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.9494736842105275e-06, | |
| "loss": 0.2985, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.9231578947368424e-06, | |
| "loss": 0.2544, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.896842105263158e-06, | |
| "loss": 0.2464, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.870526315789474e-06, | |
| "loss": 0.256, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.84421052631579e-06, | |
| "loss": 0.2706, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.817894736842106e-06, | |
| "loss": 0.2725, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.7915789473684215e-06, | |
| "loss": 0.2693, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.765263157894737e-06, | |
| "loss": 0.2447, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.738947368421052e-06, | |
| "loss": 0.2581, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.71263157894737e-06, | |
| "loss": 0.2797, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.686315789473685e-06, | |
| "loss": 0.252, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 0.2335, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.6336842105263164e-06, | |
| "loss": 0.2673, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.607368421052632e-06, | |
| "loss": 0.2552, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.581052631578948e-06, | |
| "loss": 0.2655, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.554736842105264e-06, | |
| "loss": 0.2437, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.52842105263158e-06, | |
| "loss": 0.225, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.5021052631578955e-06, | |
| "loss": 0.2483, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.4757894736842105e-06, | |
| "loss": 0.2731, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.449473684210526e-06, | |
| "loss": 0.2181, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.423157894736842e-06, | |
| "loss": 0.231, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.396842105263158e-06, | |
| "loss": 0.2596, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.370526315789474e-06, | |
| "loss": 0.2612, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.3442105263157904e-06, | |
| "loss": 0.22, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.317894736842106e-06, | |
| "loss": 0.2797, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.8756310343742371, | |
| "eval_runtime": 6384.9348, | |
| "eval_samples_per_second": 0.886, | |
| "eval_steps_per_second": 0.055, | |
| "eval_wer": 29.48886925539157, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.292631578947369e-06, | |
| "loss": 0.298, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.266315789473685e-06, | |
| "loss": 0.2248, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.24e-06, | |
| "loss": 0.2592, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.213684210526316e-06, | |
| "loss": 0.2575, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.187368421052632e-06, | |
| "loss": 0.2678, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.161052631578948e-06, | |
| "loss": 0.2751, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.134736842105264e-06, | |
| "loss": 0.2259, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.108421052631579e-06, | |
| "loss": 0.2301, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.0821052631578945e-06, | |
| "loss": 0.2767, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.055789473684212e-06, | |
| "loss": 0.2747, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.029473684210527e-06, | |
| "loss": 0.2657, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.003157894736843e-06, | |
| "loss": 0.2437, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.976842105263159e-06, | |
| "loss": 0.2511, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.950526315789474e-06, | |
| "loss": 0.2957, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.92421052631579e-06, | |
| "loss": 0.2479, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.897894736842106e-06, | |
| "loss": 0.2582, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.871578947368422e-06, | |
| "loss": 0.2621, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.845263157894737e-06, | |
| "loss": 0.2489, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.818947368421053e-06, | |
| "loss": 0.2428, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.7926315789473685e-06, | |
| "loss": 0.2415, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.766315789473684e-06, | |
| "loss": 0.2493, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.74e-06, | |
| "loss": 0.2741, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.713684210526316e-06, | |
| "loss": 0.2756, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.687368421052633e-06, | |
| "loss": 0.2737, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.661052631578948e-06, | |
| "loss": 0.2446, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.634736842105264e-06, | |
| "loss": 0.2724, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.60842105263158e-06, | |
| "loss": 0.2267, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.582105263157895e-06, | |
| "loss": 0.2626, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.555789473684211e-06, | |
| "loss": 0.2432, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.529473684210527e-06, | |
| "loss": 0.2654, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.5031578947368425e-06, | |
| "loss": 0.2832, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.476842105263158e-06, | |
| "loss": 0.263, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.450526315789474e-06, | |
| "loss": 0.2561, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.42421052631579e-06, | |
| "loss": 0.2831, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.397894736842105e-06, | |
| "loss": 0.2593, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.371578947368421e-06, | |
| "loss": 0.235, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.345263157894738e-06, | |
| "loss": 0.2342, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.318947368421053e-06, | |
| "loss": 0.2304, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.292631578947369e-06, | |
| "loss": 0.2422, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.266315789473685e-06, | |
| "loss": 0.2302, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_loss": 0.8522536754608154, | |
| "eval_runtime": 5802.9054, | |
| "eval_samples_per_second": 0.975, | |
| "eval_steps_per_second": 0.061, | |
| "eval_wer": 19.641382906304766, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.240000000000001e-06, | |
| "loss": 0.272, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.2136842105263165e-06, | |
| "loss": 0.2574, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.187368421052632e-06, | |
| "loss": 0.3026, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.161052631578948e-06, | |
| "loss": 0.2287, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.134736842105263e-06, | |
| "loss": 0.2492, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.108421052631579e-06, | |
| "loss": 0.2267, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.082105263157895e-06, | |
| "loss": 0.2427, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.0557894736842105e-06, | |
| "loss": 0.2528, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.029473684210526e-06, | |
| "loss": 0.2463, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.003157894736842e-06, | |
| "loss": 0.2235, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.976842105263158e-06, | |
| "loss": 0.2199, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.950526315789474e-06, | |
| "loss": 0.2966, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.92421052631579e-06, | |
| "loss": 0.2853, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.897894736842106e-06, | |
| "loss": 0.2454, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.871578947368421e-06, | |
| "loss": 0.2306, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.845263157894737e-06, | |
| "loss": 0.2181, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.818947368421053e-06, | |
| "loss": 0.2565, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.792631578947369e-06, | |
| "loss": 0.2506, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.7663157894736845e-06, | |
| "loss": 0.2515, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.74e-06, | |
| "loss": 0.2233, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.713684210526316e-06, | |
| "loss": 0.2675, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.687368421052632e-06, | |
| "loss": 0.2312, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.661052631578948e-06, | |
| "loss": 0.2872, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.634736842105264e-06, | |
| "loss": 0.298, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.6084210526315794e-06, | |
| "loss": 0.2632, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.582105263157895e-06, | |
| "loss": 0.2398, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.55578947368421e-06, | |
| "loss": 0.2713, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.529473684210527e-06, | |
| "loss": 0.2764, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.503157894736843e-06, | |
| "loss": 0.2476, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.4768421052631585e-06, | |
| "loss": 0.2391, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.450526315789474e-06, | |
| "loss": 0.2045, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.424210526315789e-06, | |
| "loss": 0.2232, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.397894736842105e-06, | |
| "loss": 0.2835, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.371578947368421e-06, | |
| "loss": 0.2671, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.345263157894738e-06, | |
| "loss": 0.2324, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.3189473684210535e-06, | |
| "loss": 0.2082, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.2926315789473684e-06, | |
| "loss": 0.2369, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.266315789473684e-06, | |
| "loss": 0.2593, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.24e-06, | |
| "loss": 0.3048, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.213684210526316e-06, | |
| "loss": 0.2803, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.8714997172355652, | |
| "eval_runtime": 5666.5233, | |
| "eval_samples_per_second": 0.998, | |
| "eval_steps_per_second": 0.062, | |
| "eval_wer": 19.74132062627638, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.1873684210526325e-06, | |
| "loss": 0.236, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.1610526315789475e-06, | |
| "loss": 0.2652, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.134736842105263e-06, | |
| "loss": 0.2346, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.108421052631579e-06, | |
| "loss": 0.2464, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.082105263157895e-06, | |
| "loss": 0.2253, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.055789473684211e-06, | |
| "loss": 0.2244, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.029473684210527e-06, | |
| "loss": 0.2151, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.0031578947368424e-06, | |
| "loss": 0.2573, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.977894736842106e-06, | |
| "loss": 0.241, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.951578947368422e-06, | |
| "loss": 0.2797, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9252631578947366e-06, | |
| "loss": 0.2502, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.898947368421052e-06, | |
| "loss": 0.2128, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.872631578947369e-06, | |
| "loss": 0.2127, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.846315789473685e-06, | |
| "loss": 0.2245, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.820000000000001e-06, | |
| "loss": 0.2224, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.793684210526316e-06, | |
| "loss": 0.1967, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.767368421052632e-06, | |
| "loss": 0.2379, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.7410526315789473e-06, | |
| "loss": 0.2202, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.714736842105263e-06, | |
| "loss": 0.2221, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.6884210526315794e-06, | |
| "loss": 0.2227, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.662105263157895e-06, | |
| "loss": 0.2374, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.635789473684211e-06, | |
| "loss": 0.25, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.6094736842105264e-06, | |
| "loss": 0.244, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.5831578947368422e-06, | |
| "loss": 0.2068, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.556842105263158e-06, | |
| "loss": 0.2312, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.5305263157894743e-06, | |
| "loss": 0.2171, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.50421052631579e-06, | |
| "loss": 0.2417, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.4778947368421055e-06, | |
| "loss": 0.2452, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.4515789473684213e-06, | |
| "loss": 0.2459, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.425263157894737e-06, | |
| "loss": 0.2265, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.398947368421053e-06, | |
| "loss": 0.2276, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.3726315789473683e-06, | |
| "loss": 0.258, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.3463157894736846e-06, | |
| "loss": 0.2292, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 0.3045, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2936842105263162e-06, | |
| "loss": 0.2752, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.267368421052632e-06, | |
| "loss": 0.2702, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2410526315789474e-06, | |
| "loss": 0.2365, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2147368421052633e-06, | |
| "loss": 0.2123, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.188421052631579e-06, | |
| "loss": 0.2672, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1621052631578953e-06, | |
| "loss": 0.2794, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 0.8547870516777039, | |
| "eval_runtime": 5789.6547, | |
| "eval_samples_per_second": 0.977, | |
| "eval_steps_per_second": 0.061, | |
| "eval_wer": 18.68400851643179, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.135789473684211e-06, | |
| "loss": 0.2769, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1094736842105265e-06, | |
| "loss": 0.2691, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0831578947368423e-06, | |
| "loss": 0.2285, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.056842105263158e-06, | |
| "loss": 0.218, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0305263157894736e-06, | |
| "loss": 0.2302, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0042105263157894e-06, | |
| "loss": 0.2808, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9778947368421056e-06, | |
| "loss": 0.2532, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9515789473684214e-06, | |
| "loss": 0.2556, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9252631578947373e-06, | |
| "loss": 0.2608, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.8989473684210526e-06, | |
| "loss": 0.2286, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8726315789473685e-06, | |
| "loss": 0.2501, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8463157894736843e-06, | |
| "loss": 0.2358, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.82e-06, | |
| "loss": 0.2021, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.7936842105263163e-06, | |
| "loss": 0.2187, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7673684210526317e-06, | |
| "loss": 0.2468, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7410526315789476e-06, | |
| "loss": 0.2527, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7147368421052634e-06, | |
| "loss": 0.2422, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.688421052631579e-06, | |
| "loss": 0.2499, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6621052631578946e-06, | |
| "loss": 0.2997, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.635789473684211e-06, | |
| "loss": 0.2786, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6094736842105267e-06, | |
| "loss": 0.2804, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5831578947368425e-06, | |
| "loss": 0.2601, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5568421052631583e-06, | |
| "loss": 0.2857, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5305263157894737e-06, | |
| "loss": 0.2386, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5042105263157895e-06, | |
| "loss": 0.2035, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.4778947368421053e-06, | |
| "loss": 0.2247, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.451578947368421e-06, | |
| "loss": 0.2178, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.425263157894737e-06, | |
| "loss": 0.2212, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.3989473684210528e-06, | |
| "loss": 0.2123, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3726315789473686e-06, | |
| "loss": 0.2143, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3463157894736844e-06, | |
| "loss": 0.2002, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3200000000000002e-06, | |
| "loss": 0.2219, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.293684210526316e-06, | |
| "loss": 0.2597, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.267368421052632e-06, | |
| "loss": 0.2168, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2410526315789473e-06, | |
| "loss": 0.2489, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2147368421052635e-06, | |
| "loss": 0.1965, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.188421052631579e-06, | |
| "loss": 0.2449, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1621052631578947e-06, | |
| "loss": 0.2284, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.135789473684211e-06, | |
| "loss": 0.2127, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1094736842105264e-06, | |
| "loss": 0.2173, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.8542773723602295, | |
| "eval_runtime": 5521.1614, | |
| "eval_samples_per_second": 1.024, | |
| "eval_steps_per_second": 0.064, | |
| "eval_wer": 17.901887229697437, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.083157894736842e-06, | |
| "loss": 0.2428, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.056842105263158e-06, | |
| "loss": 0.2218, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.030526315789474e-06, | |
| "loss": 0.231, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0042105263157896e-06, | |
| "loss": 0.2091, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9778947368421055e-06, | |
| "loss": 0.2813, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9515789473684213e-06, | |
| "loss": 0.2378, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.925263157894737e-06, | |
| "loss": 0.221, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8989473684210527e-06, | |
| "loss": 0.2133, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8726315789473687e-06, | |
| "loss": 0.2172, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8463157894736843e-06, | |
| "loss": 0.2301, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8200000000000002e-06, | |
| "loss": 0.2039, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7936842105263158e-06, | |
| "loss": 0.2103, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7673684210526318e-06, | |
| "loss": 0.2143, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.742105263157895e-06, | |
| "loss": 0.2607, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.7157894736842107e-06, | |
| "loss": 0.2785, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6894736842105264e-06, | |
| "loss": 0.2195, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6631578947368424e-06, | |
| "loss": 0.1813, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.636842105263158e-06, | |
| "loss": 0.2221, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6105263157894738e-06, | |
| "loss": 0.1919, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5842105263157894e-06, | |
| "loss": 0.2215, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5578947368421054e-06, | |
| "loss": 0.2439, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5315789473684213e-06, | |
| "loss": 0.2338, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.5052631578947369e-06, | |
| "loss": 0.2049, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.478947368421053e-06, | |
| "loss": 0.1977, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4526315789473685e-06, | |
| "loss": 0.2129, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4263157894736843e-06, | |
| "loss": 0.2319, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.2209, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.373684210526316e-06, | |
| "loss": 0.23, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3473684210526316e-06, | |
| "loss": 0.2418, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.3210526315789474e-06, | |
| "loss": 0.2623, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2947368421052634e-06, | |
| "loss": 0.2338, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.268421052631579e-06, | |
| "loss": 0.2409, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2421052631578948e-06, | |
| "loss": 0.2761, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2157894736842107e-06, | |
| "loss": 0.2146, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1894736842105265e-06, | |
| "loss": 0.2272, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.163157894736842e-06, | |
| "loss": 0.2265, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.136842105263158e-06, | |
| "loss": 0.2196, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.110526315789474e-06, | |
| "loss": 0.2143, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0842105263157895e-06, | |
| "loss": 0.2129, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0578947368421054e-06, | |
| "loss": 0.217, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_loss": 0.8518440127372742, | |
| "eval_runtime": 5194.633, | |
| "eval_samples_per_second": 1.089, | |
| "eval_steps_per_second": 0.068, | |
| "eval_wer": 16.3839925843315, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0315789473684212e-06, | |
| "loss": 0.2194, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.005263157894737e-06, | |
| "loss": 0.1901, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.789473684210526e-07, | |
| "loss": 0.1898, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.526315789473685e-07, | |
| "loss": 0.224, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.263157894736844e-07, | |
| "loss": 0.2209, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.29, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.736842105263159e-07, | |
| "loss": 0.1973, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.473684210526316e-07, | |
| "loss": 0.182, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.210526315789474e-07, | |
| "loss": 0.2654, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.947368421052632e-07, | |
| "loss": 0.2342, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.684210526315789e-07, | |
| "loss": 0.2254, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.421052631578949e-07, | |
| "loss": 0.2037, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.157894736842106e-07, | |
| "loss": 0.2146, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.894736842105264e-07, | |
| "loss": 0.218, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.631578947368422e-07, | |
| "loss": 0.2405, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.368421052631579e-07, | |
| "loss": 0.2306, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.105263157894738e-07, | |
| "loss": 0.2291, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.842105263157896e-07, | |
| "loss": 0.2411, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.578947368421053e-07, | |
| "loss": 0.2503, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.315789473684211e-07, | |
| "loss": 0.1941, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.052631578947369e-07, | |
| "loss": 0.1873, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.789473684210526e-07, | |
| "loss": 0.2644, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.526315789473685e-07, | |
| "loss": 0.2611, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.2631578947368427e-07, | |
| "loss": 0.2303, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.2403, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.736842105263158e-07, | |
| "loss": 0.2652, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.4736842105263157e-07, | |
| "loss": 0.2225, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.210526315789474e-07, | |
| "loss": 0.2045, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.9473684210526315e-07, | |
| "loss": 0.1908, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.6842105263157897e-07, | |
| "loss": 0.2981, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.4210526315789473e-07, | |
| "loss": 0.2498, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.1578947368421053e-07, | |
| "loss": 0.2385, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.8947368421052634e-07, | |
| "loss": 0.2189, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.631578947368421e-07, | |
| "loss": 0.2586, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.368421052631579e-07, | |
| "loss": 0.2696, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1052631578947368e-07, | |
| "loss": 0.2125, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.421052631578947e-08, | |
| "loss": 0.2214, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.789473684210527e-08, | |
| "loss": 0.2398, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.157894736842106e-08, | |
| "loss": 0.2005, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.263157894736842e-09, | |
| "loss": 0.1718, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.8487818241119385, | |
| "eval_runtime": 5428.8291, | |
| "eval_samples_per_second": 1.042, | |
| "eval_steps_per_second": 0.065, | |
| "eval_wer": 16.588213142534364, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 10000, | |
| "total_flos": 3.265935704064e+20, | |
| "train_loss": 0.2656231692790985, | |
| "train_runtime": 120187.1267, | |
| "train_samples_per_second": 2.663, | |
| "train_steps_per_second": 0.083 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 3.265935704064e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |