| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9999962488962377, | |
| "global_step": 166617, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5e-05, | |
| "loss": 3.0218, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-05, | |
| "loss": 2.376, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-05, | |
| "loss": 2.2374, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-05, | |
| "loss": 2.1414, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0934, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0597, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5e-05, | |
| "loss": 2.0249, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9947, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-05, | |
| "loss": 1.969, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9525, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9319, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9208, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-05, | |
| "loss": 1.9116, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-05, | |
| "loss": 1.899, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8899, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8775, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8701, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8654, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8557, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8487, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8401, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8307, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8303, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8217, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8192, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8156, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8077, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5e-05, | |
| "loss": 1.802, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7997, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7962, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7918, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7898, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7861, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7817, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7743, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7747, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7718, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7672, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7676, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7618, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7554, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7548, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7542, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7487, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7504, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7464, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7497, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7451, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5e-05, | |
| "loss": 1.745, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7409, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7416, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7341, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7313, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7351, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7269, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7292, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7279, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7259, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7254, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7262, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7213, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7207, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5e-05, | |
| "loss": 1.72, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7178, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.71, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7128, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7101, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7108, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7105, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7076, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7055, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7068, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5e-05, | |
| "loss": 1.707, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7009, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6997, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6973, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6963, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6965, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6944, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6976, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6853, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6877, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6891, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6898, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6844, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6861, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6819, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6905, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6818, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6855, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6987, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6898, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6831, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6782, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6789, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6766, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6708, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6719, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6703, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5e-05, | |
| "loss": 1.667, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6622, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6677, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6671, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.665, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6644, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6605, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6554, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.659, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6573, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6538, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6549, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6529, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6565, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6556, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6513, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6533, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6517, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6497, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6454, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6492, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6468, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.643, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6443, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6452, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6479, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6467, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6442, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6393, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5e-05, | |
| "loss": 1.639, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5e-05, | |
| "loss": 1.639, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6413, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6402, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6377, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6341, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6347, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6432, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6381, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6327, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6392, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6367, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6341, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6363, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6303, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6307, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.633, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6341, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6289, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6357, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6292, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.629, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6286, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6318, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6278, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6252, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6302, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6286, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5e-05, | |
| "loss": 1.629, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6246, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6275, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6251, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6248, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6224, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6265, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6228, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6258, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6215, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6231, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6136, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6216, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6201, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6197, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6164, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6236, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6169, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6159, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5e-05, | |
| "loss": 1.616, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6125, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6174, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6174, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6125, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6177, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6111, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6166, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6126, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6142, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6111, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6117, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6123, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6146, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6097, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6109, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5e-05, | |
| "loss": 1.612, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6137, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6088, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6127, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5e-05, | |
| "loss": 1.608, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6069, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6115, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5e-05, | |
| "loss": 1.608, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6049, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6104, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5e-05, | |
| "loss": 1.607, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6036, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6053, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6072, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6054, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6039, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5e-05, | |
| "loss": 1.606, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6046, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6053, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6047, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6021, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6033, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5e-05, | |
| "loss": 1.602, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6026, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6046, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6088, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6035, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6004, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6015, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6001, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6006, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6007, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6023, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5983, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5976, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6027, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6017, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5954, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6038, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5949, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5934, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5981, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5918, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5938, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5979, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5936, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5953, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5914, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5962, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5931, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5891, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5919, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5947, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5943, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5897, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5e-05, | |
| "loss": 1.587, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5e-05, | |
| "loss": 1.595, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5921, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5899, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5918, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5938, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5911, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5926, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5932, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5895, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5891, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5884, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5875, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5901, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5e-05, | |
| "loss": 1.59, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5892, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5896, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 5e-05, | |
| "loss": 1.589, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5864, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5866, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5919, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5839, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5828, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5888, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5881, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5854, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5887, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5847, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5833, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5e-05, | |
| "loss": 1.586, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5838, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5843, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5837, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5864, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5867, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5821, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5867, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5846, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5863, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5865, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5867, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5827, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5816, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5798, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5e-05, | |
| "loss": 1.579, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5825, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5811, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5786, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5813, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5829, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5e-05, | |
| "loss": 1.58, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5797, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5e-05, | |
| "loss": 1.579, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5804, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5787, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5814, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5796, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5742, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5762, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5822, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5e-05, | |
| "loss": 1.578, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5816, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5772, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5786, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5784, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5784, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5786, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5769, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5e-05, | |
| "loss": 1.577, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5767, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5769, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5e-05, | |
| "loss": 1.576, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5736, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5746, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5692, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5742, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5767, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5728, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5704, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5698, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5765, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5712, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5747, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5784, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5e-05, | |
| "loss": 1.575, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5729, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5715, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 166617, | |
| "total_flos": 9.903255447295492e+18, | |
| "train_loss": 0.8373472213664984, | |
| "train_runtime": 584567.9437, | |
| "train_samples_per_second": 18.242, | |
| "train_steps_per_second": 0.285 | |
| } | |
| ], | |
| "max_steps": 166617, | |
| "num_train_epochs": 1, | |
| "total_flos": 9.903255447295492e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |