| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.1308703819661865, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.8308, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.7539, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.5e-06, | |
| "loss": 1.6532, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.5445, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.4521, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3e-06, | |
| "loss": 1.3696, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.5e-06, | |
| "loss": 1.3173, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.2789, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.5e-06, | |
| "loss": 1.2491, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-06, | |
| "loss": 1.2166, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 1.1947, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6e-06, | |
| "loss": 1.1663, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 1.1512, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 7e-06, | |
| "loss": 1.1223, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.1129, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.0833, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.5e-06, | |
| "loss": 1.0725, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9e-06, | |
| "loss": 1.0538, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.5e-06, | |
| "loss": 1.046, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0349, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.973684210526316e-06, | |
| "loss": 1.0174, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.947368421052632e-06, | |
| "loss": 1.003, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.921052631578947e-06, | |
| "loss": 0.9998, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.894736842105264e-06, | |
| "loss": 0.9833, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.868421052631579e-06, | |
| "loss": 0.9839, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.842105263157896e-06, | |
| "loss": 0.9652, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.815789473684212e-06, | |
| "loss": 0.9608, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.789473684210527e-06, | |
| "loss": 0.9534, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.763157894736844e-06, | |
| "loss": 0.9464, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.736842105263159e-06, | |
| "loss": 0.9402, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.710526315789474e-06, | |
| "loss": 0.9294, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.68421052631579e-06, | |
| "loss": 0.9249, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.657894736842106e-06, | |
| "loss": 0.9214, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.631578947368422e-06, | |
| "loss": 0.9149, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.605263157894737e-06, | |
| "loss": 0.9118, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.578947368421054e-06, | |
| "loss": 0.902, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.552631578947369e-06, | |
| "loss": 0.9092, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.526315789473684e-06, | |
| "loss": 0.8913, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.8939, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.473684210526315e-06, | |
| "loss": 0.8879, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.447368421052632e-06, | |
| "loss": 0.8821, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.421052631578949e-06, | |
| "loss": 0.8779, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.394736842105264e-06, | |
| "loss": 0.8715, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.36842105263158e-06, | |
| "loss": 0.8779, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.342105263157895e-06, | |
| "loss": 0.8674, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.315789473684212e-06, | |
| "loss": 0.8569, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.289473684210527e-06, | |
| "loss": 0.862, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.263157894736842e-06, | |
| "loss": 0.8519, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.236842105263159e-06, | |
| "loss": 0.8532, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.210526315789474e-06, | |
| "loss": 0.8577, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.18421052631579e-06, | |
| "loss": 0.8511, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.157894736842105e-06, | |
| "loss": 0.8403, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.131578947368422e-06, | |
| "loss": 0.8424, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.105263157894739e-06, | |
| "loss": 0.8373, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.078947368421054e-06, | |
| "loss": 0.8389, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.05263157894737e-06, | |
| "loss": 0.8304, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.026315789473685e-06, | |
| "loss": 0.8236, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9e-06, | |
| "loss": 0.829, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.973684210526317e-06, | |
| "loss": 0.8244, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.947368421052632e-06, | |
| "loss": 0.8182, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.921052631578949e-06, | |
| "loss": 0.8168, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.894736842105264e-06, | |
| "loss": 0.8227, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.86842105263158e-06, | |
| "loss": 0.8134, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.842105263157895e-06, | |
| "loss": 0.8058, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.81578947368421e-06, | |
| "loss": 0.8106, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.789473684210527e-06, | |
| "loss": 0.8099, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.763157894736842e-06, | |
| "loss": 0.8019, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.736842105263158e-06, | |
| "loss": 0.8049, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.710526315789475e-06, | |
| "loss": 0.8047, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.68421052631579e-06, | |
| "loss": 0.8032, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.657894736842107e-06, | |
| "loss": 0.8027, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.631578947368422e-06, | |
| "loss": 0.794, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.605263157894738e-06, | |
| "loss": 0.806, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.578947368421053e-06, | |
| "loss": 0.792, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.552631578947368e-06, | |
| "loss": 0.7941, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.526315789473685e-06, | |
| "loss": 0.7842, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.7795, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.473684210526317e-06, | |
| "loss": 0.7875, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.447368421052632e-06, | |
| "loss": 0.7862, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.421052631578948e-06, | |
| "loss": 0.7837, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.394736842105263e-06, | |
| "loss": 0.7832, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.36842105263158e-06, | |
| "loss": 0.7742, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.342105263157897e-06, | |
| "loss": 0.7664, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 8.315789473684212e-06, | |
| "loss": 0.774, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.289473684210526e-06, | |
| "loss": 0.776, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.263157894736843e-06, | |
| "loss": 0.764, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.236842105263158e-06, | |
| "loss": 0.7741, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.210526315789475e-06, | |
| "loss": 0.7676, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.18421052631579e-06, | |
| "loss": 0.7649, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.157894736842106e-06, | |
| "loss": 0.7735, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.131578947368421e-06, | |
| "loss": 0.7689, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.105263157894736e-06, | |
| "loss": 0.7638, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.078947368421053e-06, | |
| "loss": 0.7588, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.052631578947368e-06, | |
| "loss": 0.7591, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.026315789473685e-06, | |
| "loss": 0.7607, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.7575, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.973684210526316e-06, | |
| "loss": 0.7655, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.947368421052633e-06, | |
| "loss": 0.7632, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.921052631578948e-06, | |
| "loss": 0.7508, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.894736842105265e-06, | |
| "loss": 0.7563, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.86842105263158e-06, | |
| "loss": 0.7511, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.842105263157895e-06, | |
| "loss": 0.7537, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.815789473684211e-06, | |
| "loss": 0.7516, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.789473684210526e-06, | |
| "loss": 0.7519, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.763157894736843e-06, | |
| "loss": 0.7517, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.736842105263158e-06, | |
| "loss": 0.7407, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.710526315789474e-06, | |
| "loss": 0.7466, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.68421052631579e-06, | |
| "loss": 0.754, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.657894736842106e-06, | |
| "loss": 0.7435, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.631578947368423e-06, | |
| "loss": 0.7448, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.605263157894738e-06, | |
| "loss": 0.7438, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.578947368421054e-06, | |
| "loss": 0.748, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.552631578947369e-06, | |
| "loss": 0.739, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.526315789473685e-06, | |
| "loss": 0.7413, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.7363, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 7.473684210526316e-06, | |
| "loss": 0.7342, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.447368421052632e-06, | |
| "loss": 0.7282, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.421052631578948e-06, | |
| "loss": 0.7379, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.3947368421052635e-06, | |
| "loss": 0.7396, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.368421052631579e-06, | |
| "loss": 0.7378, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.342105263157895e-06, | |
| "loss": 0.7454, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.315789473684212e-06, | |
| "loss": 0.7392, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.289473684210528e-06, | |
| "loss": 0.7247, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.263157894736843e-06, | |
| "loss": 0.7322, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.236842105263158e-06, | |
| "loss": 0.7289, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.210526315789474e-06, | |
| "loss": 0.7256, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.18421052631579e-06, | |
| "loss": 0.7199, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.157894736842106e-06, | |
| "loss": 0.7191, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.131578947368422e-06, | |
| "loss": 0.7105, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 7.1052631578947375e-06, | |
| "loss": 0.7178, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.078947368421053e-06, | |
| "loss": 0.7302, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.052631578947369e-06, | |
| "loss": 0.7168, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.026315789473684e-06, | |
| "loss": 0.7161, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7e-06, | |
| "loss": 0.7157, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.973684210526316e-06, | |
| "loss": 0.7117, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.947368421052632e-06, | |
| "loss": 0.7104, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.921052631578948e-06, | |
| "loss": 0.7138, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.894736842105264e-06, | |
| "loss": 0.7123, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.86842105263158e-06, | |
| "loss": 0.7142, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.842105263157896e-06, | |
| "loss": 0.7057, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.8157894736842115e-06, | |
| "loss": 0.7171, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.789473684210527e-06, | |
| "loss": 0.7157, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.763157894736842e-06, | |
| "loss": 0.7068, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.736842105263158e-06, | |
| "loss": 0.7081, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.710526315789474e-06, | |
| "loss": 0.7013, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.68421052631579e-06, | |
| "loss": 0.7128, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.6578947368421055e-06, | |
| "loss": 0.7083, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.631578947368421e-06, | |
| "loss": 0.7051, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.605263157894738e-06, | |
| "loss": 0.7112, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.578947368421054e-06, | |
| "loss": 0.7078, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.55263157894737e-06, | |
| "loss": 0.7072, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.526315789473685e-06, | |
| "loss": 0.7063, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.7047, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.473684210526316e-06, | |
| "loss": 0.7071, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.447368421052632e-06, | |
| "loss": 0.7024, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.421052631578948e-06, | |
| "loss": 0.7003, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.394736842105264e-06, | |
| "loss": 0.7076, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.3684210526315795e-06, | |
| "loss": 0.7036, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.342105263157895e-06, | |
| "loss": 0.7055, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.31578947368421e-06, | |
| "loss": 0.7014, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.289473684210526e-06, | |
| "loss": 0.7069, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.263157894736842e-06, | |
| "loss": 0.6965, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.236842105263159e-06, | |
| "loss": 0.7042, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.2105263157894745e-06, | |
| "loss": 0.7048, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.18421052631579e-06, | |
| "loss": 0.7009, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.157894736842106e-06, | |
| "loss": 0.701, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.131578947368422e-06, | |
| "loss": 0.696, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.105263157894738e-06, | |
| "loss": 0.6878, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.0789473684210535e-06, | |
| "loss": 0.6932, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.0526315789473685e-06, | |
| "loss": 0.6904, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.026315789473684e-06, | |
| "loss": 0.694, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6e-06, | |
| "loss": 0.6965, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.973684210526316e-06, | |
| "loss": 0.6904, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.947368421052632e-06, | |
| "loss": 0.6884, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.921052631578948e-06, | |
| "loss": 0.6889, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.8947368421052634e-06, | |
| "loss": 0.6886, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.86842105263158e-06, | |
| "loss": 0.6879, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.842105263157896e-06, | |
| "loss": 0.6855, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.815789473684212e-06, | |
| "loss": 0.6907, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.789473684210527e-06, | |
| "loss": 0.6902, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.7631578947368425e-06, | |
| "loss": 0.695, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.736842105263158e-06, | |
| "loss": 0.6845, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.710526315789474e-06, | |
| "loss": 0.6847, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.68421052631579e-06, | |
| "loss": 0.6888, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.657894736842106e-06, | |
| "loss": 0.6906, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.631578947368422e-06, | |
| "loss": 0.6826, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.605263157894737e-06, | |
| "loss": 0.6833, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.578947368421052e-06, | |
| "loss": 0.6914, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.552631578947368e-06, | |
| "loss": 0.6806, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.526315789473685e-06, | |
| "loss": 0.6816, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.6816, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.4736842105263165e-06, | |
| "loss": 0.6885, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.447368421052632e-06, | |
| "loss": 0.688, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5.421052631578948e-06, | |
| "loss": 0.6812, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.394736842105264e-06, | |
| "loss": 0.6866, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.36842105263158e-06, | |
| "loss": 0.6789, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 5.342105263157895e-06, | |
| "loss": 0.6831, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 5.315789473684211e-06, | |
| "loss": 0.6907, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 5.289473684210526e-06, | |
| "loss": 0.6811, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 5.263157894736842e-06, | |
| "loss": 0.6834, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 5.236842105263158e-06, | |
| "loss": 0.6796, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 5.210526315789474e-06, | |
| "loss": 0.6807, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 5.18421052631579e-06, | |
| "loss": 0.6723, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 5.157894736842106e-06, | |
| "loss": 0.6734, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 5.131578947368422e-06, | |
| "loss": 0.6799, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 5.105263157894738e-06, | |
| "loss": 0.6774, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 5.078947368421053e-06, | |
| "loss": 0.6729, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 5.052631578947369e-06, | |
| "loss": 0.6756, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 5.026315789473685e-06, | |
| "loss": 0.6781, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 5e-06, | |
| "loss": 0.6755, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.973684210526316e-06, | |
| "loss": 0.6776, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.947368421052632e-06, | |
| "loss": 0.6737, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.921052631578948e-06, | |
| "loss": 0.6757, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.894736842105264e-06, | |
| "loss": 0.681, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.8684210526315795e-06, | |
| "loss": 0.6748, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.842105263157895e-06, | |
| "loss": 0.6734, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.815789473684211e-06, | |
| "loss": 0.6688, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.789473684210527e-06, | |
| "loss": 0.6709, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.763157894736842e-06, | |
| "loss": 0.6642, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.736842105263158e-06, | |
| "loss": 0.6696, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.710526315789474e-06, | |
| "loss": 0.6744, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.68421052631579e-06, | |
| "loss": 0.6744, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.657894736842106e-06, | |
| "loss": 0.6684, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.631578947368421e-06, | |
| "loss": 0.6654, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.605263157894737e-06, | |
| "loss": 0.673, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.578947368421053e-06, | |
| "loss": 0.6696, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.552631578947369e-06, | |
| "loss": 0.6712, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.526315789473685e-06, | |
| "loss": 0.6674, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.6735, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.473684210526316e-06, | |
| "loss": 0.6693, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.447368421052632e-06, | |
| "loss": 0.6688, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.4210526315789476e-06, | |
| "loss": 0.6668, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.394736842105263e-06, | |
| "loss": 0.6773, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.368421052631579e-06, | |
| "loss": 0.6686, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.342105263157895e-06, | |
| "loss": 0.6702, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.315789473684211e-06, | |
| "loss": 0.6684, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.289473684210527e-06, | |
| "loss": 0.664, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.2631578947368425e-06, | |
| "loss": 0.6695, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.236842105263158e-06, | |
| "loss": 0.6747, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.210526315789474e-06, | |
| "loss": 0.6634, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.18421052631579e-06, | |
| "loss": 0.671, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.157894736842106e-06, | |
| "loss": 0.6701, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.1315789473684216e-06, | |
| "loss": 0.6719, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.105263157894737e-06, | |
| "loss": 0.665, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.078947368421053e-06, | |
| "loss": 0.6642, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.052631578947368e-06, | |
| "loss": 0.6584, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.026315789473684e-06, | |
| "loss": 0.6655, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.6672, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.9736842105263165e-06, | |
| "loss": 0.6661, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.947368421052632e-06, | |
| "loss": 0.6669, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.921052631578947e-06, | |
| "loss": 0.6669, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.894736842105263e-06, | |
| "loss": 0.6626, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.868421052631579e-06, | |
| "loss": 0.6711, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 3.842105263157895e-06, | |
| "loss": 0.665, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.815789473684211e-06, | |
| "loss": 0.6651, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.790526315789474e-06, | |
| "loss": 0.6642, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.7642105263157897e-06, | |
| "loss": 0.6591, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.7378947368421055e-06, | |
| "loss": 0.6679, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.7115789473684218e-06, | |
| "loss": 0.6618, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.685263157894737e-06, | |
| "loss": 0.6583, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.658947368421053e-06, | |
| "loss": 0.6624, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.632631578947369e-06, | |
| "loss": 0.6579, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.6063157894736846e-06, | |
| "loss": 0.6578, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.58e-06, | |
| "loss": 0.6605, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.553684210526316e-06, | |
| "loss": 0.6623, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.527368421052632e-06, | |
| "loss": 0.658, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.501052631578948e-06, | |
| "loss": 0.6591, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.4747368421052637e-06, | |
| "loss": 0.6469, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.448421052631579e-06, | |
| "loss": 0.6549, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.422105263157895e-06, | |
| "loss": 0.6605, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.3957894736842108e-06, | |
| "loss": 0.6581, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 3.369473684210526e-06, | |
| "loss": 0.654, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.3431578947368424e-06, | |
| "loss": 0.6543, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.3168421052631582e-06, | |
| "loss": 0.6558, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.290526315789474e-06, | |
| "loss": 0.6522, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.26421052631579e-06, | |
| "loss": 0.6614, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.2378947368421052e-06, | |
| "loss": 0.6561, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.211578947368421e-06, | |
| "loss": 0.6568, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.185263157894737e-06, | |
| "loss": 0.6539, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.158947368421053e-06, | |
| "loss": 0.6524, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.132631578947369e-06, | |
| "loss": 0.6537, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.1063157894736843e-06, | |
| "loss": 0.6509, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.08e-06, | |
| "loss": 0.6532, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.053684210526316e-06, | |
| "loss": 0.6487, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.027368421052632e-06, | |
| "loss": 0.6531, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.001052631578948e-06, | |
| "loss": 0.6432, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.9747368421052634e-06, | |
| "loss": 0.655, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.9484210526315792e-06, | |
| "loss": 0.6508, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.922105263157895e-06, | |
| "loss": 0.6515, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.895789473684211e-06, | |
| "loss": 0.6507, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.8694736842105263e-06, | |
| "loss": 0.65, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.843157894736842e-06, | |
| "loss": 0.6516, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.8168421052631583e-06, | |
| "loss": 0.6514, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.790526315789474e-06, | |
| "loss": 0.6684, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.76421052631579e-06, | |
| "loss": 0.6481, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.7378947368421054e-06, | |
| "loss": 0.6592, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.711578947368421e-06, | |
| "loss": 0.6539, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.685263157894737e-06, | |
| "loss": 0.6544, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.6589473684210524e-06, | |
| "loss": 0.6437, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.632631578947369e-06, | |
| "loss": 0.6558, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.6063157894736845e-06, | |
| "loss": 0.6502, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.5800000000000003e-06, | |
| "loss": 0.6436, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.553684210526316e-06, | |
| "loss": 0.6483, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.5273684210526315e-06, | |
| "loss": 0.6416, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.5010526315789473e-06, | |
| "loss": 0.6529, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.4747368421052636e-06, | |
| "loss": 0.6535, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.448421052631579e-06, | |
| "loss": 0.6563, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 2.422105263157895e-06, | |
| "loss": 0.6484, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.3957894736842106e-06, | |
| "loss": 0.6437, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.3694736842105264e-06, | |
| "loss": 0.6515, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.3431578947368422e-06, | |
| "loss": 0.6575, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 2.316842105263158e-06, | |
| "loss": 0.6465, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.290526315789474e-06, | |
| "loss": 0.6481, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 2.2642105263157897e-06, | |
| "loss": 0.6474, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.2378947368421055e-06, | |
| "loss": 0.6421, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.2115789473684213e-06, | |
| "loss": 0.6488, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.185263157894737e-06, | |
| "loss": 0.64, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 2.1589473684210525e-06, | |
| "loss": 0.645, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 2.1326315789473688e-06, | |
| "loss": 0.6466, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 2.106315789473684e-06, | |
| "loss": 0.6465, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 2.08e-06, | |
| "loss": 0.652, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 2.0536842105263162e-06, | |
| "loss": 0.6479, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 2.0273684210526316e-06, | |
| "loss": 0.6423, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 2.0010526315789474e-06, | |
| "loss": 0.6473, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 1.9747368421052633e-06, | |
| "loss": 0.6527, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.948421052631579e-06, | |
| "loss": 0.6479, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.922105263157895e-06, | |
| "loss": 0.6536, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 1.8957894736842105e-06, | |
| "loss": 0.6385, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.8694736842105265e-06, | |
| "loss": 0.6461, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.8431578947368424e-06, | |
| "loss": 0.645, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 1.816842105263158e-06, | |
| "loss": 0.6505, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 1.790526315789474e-06, | |
| "loss": 0.6525, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1.7642105263157896e-06, | |
| "loss": 0.644, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1.7378947368421054e-06, | |
| "loss": 0.6494, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1.711578947368421e-06, | |
| "loss": 0.6442, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1.685263157894737e-06, | |
| "loss": 0.6437, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.6589473684210527e-06, | |
| "loss": 0.6466, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 1.6326315789473685e-06, | |
| "loss": 0.6459, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 1.6063157894736845e-06, | |
| "loss": 0.6473, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 1.5800000000000001e-06, | |
| "loss": 0.6406, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 1.553684210526316e-06, | |
| "loss": 0.6403, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 1.5273684210526318e-06, | |
| "loss": 0.6455, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 1.5010526315789476e-06, | |
| "loss": 0.6458, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.4747368421052632e-06, | |
| "loss": 0.6448, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 1.448421052631579e-06, | |
| "loss": 0.6479, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.422105263157895e-06, | |
| "loss": 0.6485, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.3957894736842106e-06, | |
| "loss": 0.6503, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.3694736842105262e-06, | |
| "loss": 0.6443, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.3431578947368423e-06, | |
| "loss": 0.6481, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.316842105263158e-06, | |
| "loss": 0.646, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.2905263157894737e-06, | |
| "loss": 0.6418, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.2642105263157895e-06, | |
| "loss": 0.6428, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.2378947368421053e-06, | |
| "loss": 0.645, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.2115789473684212e-06, | |
| "loss": 0.6446, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.185263157894737e-06, | |
| "loss": 0.6466, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.1589473684210526e-06, | |
| "loss": 0.6481, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.1326315789473686e-06, | |
| "loss": 0.6421, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.1063157894736842e-06, | |
| "loss": 0.642, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.08e-06, | |
| "loss": 0.6442, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.0536842105263159e-06, | |
| "loss": 0.6464, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.0273684210526317e-06, | |
| "loss": 0.6424, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.0010526315789475e-06, | |
| "loss": 0.6463, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 9.74736842105263e-07, | |
| "loss": 0.6437, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 9.48421052631579e-07, | |
| "loss": 0.6417, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.221052631578948e-07, | |
| "loss": 0.6487, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 8.957894736842106e-07, | |
| "loss": 0.6467, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 8.694736842105264e-07, | |
| "loss": 0.6435, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 8.431578947368421e-07, | |
| "loss": 0.6439, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 8.16842105263158e-07, | |
| "loss": 0.6421, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 7.905263157894738e-07, | |
| "loss": 0.6397, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 7.642105263157895e-07, | |
| "loss": 0.64, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 7.378947368421054e-07, | |
| "loss": 0.6434, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 7.115789473684211e-07, | |
| "loss": 0.6419, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 6.852631578947369e-07, | |
| "loss": 0.6495, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 6.589473684210527e-07, | |
| "loss": 0.6455, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 6.326315789473684e-07, | |
| "loss": 0.6437, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 6.063157894736842e-07, | |
| "loss": 0.6426, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 5.800000000000001e-07, | |
| "loss": 0.646, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.536842105263159e-07, | |
| "loss": 0.6445, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.273684210526316e-07, | |
| "loss": 0.6347, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 5.010526315789474e-07, | |
| "loss": 0.643, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.747368421052632e-07, | |
| "loss": 0.6403, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.4842105263157894e-07, | |
| "loss": 0.6472, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.2315789473684214e-07, | |
| "loss": 0.6415, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.968421052631579e-07, | |
| "loss": 0.6406, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.7052631578947377e-07, | |
| "loss": 0.6355, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.4421052631578954e-07, | |
| "loss": 0.637, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.178947368421053e-07, | |
| "loss": 0.6448, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.9157894736842107e-07, | |
| "loss": 0.6421, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.6526315789473684e-07, | |
| "loss": 0.6391, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.3894736842105265e-07, | |
| "loss": 0.6313, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.1263157894736842e-07, | |
| "loss": 0.6504, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 1.8631578947368424e-07, | |
| "loss": 0.6498, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 1.6e-07, | |
| "loss": 0.6423, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 1.3368421052631582e-07, | |
| "loss": 0.6395, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 1.0736842105263159e-07, | |
| "loss": 0.6414, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 8.105263157894738e-08, | |
| "loss": 0.6416, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.473684210526316e-08, | |
| "loss": 0.6369, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.842105263157895e-08, | |
| "loss": 0.6404, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.105263157894737e-09, | |
| "loss": 0.645, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "step": 10000, | |
| "total_flos": 6.300743273312256e+19, | |
| "train_loss": 0.736913056564331, | |
| "train_runtime": 18479.0778, | |
| "train_samples_per_second": 138.535, | |
| "train_steps_per_second": 0.541 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 4, | |
| "total_flos": 6.300743273312256e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |