| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 100.0, | |
| "global_step": 25800, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.903294573643411e-05, | |
| "loss": 0.981, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 4.8063953488372096e-05, | |
| "loss": 0.7307, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 4.709496124031008e-05, | |
| "loss": 0.6521, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 4.612596899224806e-05, | |
| "loss": 0.5851, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.69, | |
| "learning_rate": 4.515697674418605e-05, | |
| "loss": 0.5504, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 4.418798449612404e-05, | |
| "loss": 0.5169, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 4.321899224806202e-05, | |
| "loss": 0.4854, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 15.5, | |
| "learning_rate": 4.2250000000000004e-05, | |
| "loss": 0.4565, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 17.44, | |
| "learning_rate": 4.128100775193799e-05, | |
| "loss": 0.4407, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 19.38, | |
| "learning_rate": 4.031201550387597e-05, | |
| "loss": 0.4256, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.32, | |
| "learning_rate": 3.9343023255813954e-05, | |
| "loss": 0.4087, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 23.26, | |
| "learning_rate": 3.837403100775194e-05, | |
| "loss": 0.3882, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 25.19, | |
| "learning_rate": 3.740697674418605e-05, | |
| "loss": 0.3757, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 27.13, | |
| "learning_rate": 3.643798449612403e-05, | |
| "loss": 0.3613, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 29.07, | |
| "learning_rate": 3.5468992248062015e-05, | |
| "loss": 0.3477, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 3.45e-05, | |
| "loss": 0.3302, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 32.95, | |
| "learning_rate": 3.353100775193799e-05, | |
| "loss": 0.3222, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 34.88, | |
| "learning_rate": 3.256201550387597e-05, | |
| "loss": 0.3118, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 36.82, | |
| "learning_rate": 3.1594961240310076e-05, | |
| "loss": 0.3015, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 38.76, | |
| "learning_rate": 3.062596899224806e-05, | |
| "loss": 0.2976, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 40.7, | |
| "learning_rate": 2.965697674418605e-05, | |
| "loss": 0.2893, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 42.64, | |
| "learning_rate": 2.8687984496124033e-05, | |
| "loss": 0.2786, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 44.57, | |
| "learning_rate": 2.7718992248062013e-05, | |
| "loss": 0.2696, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 46.51, | |
| "learning_rate": 2.6750000000000003e-05, | |
| "loss": 0.2646, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 48.45, | |
| "learning_rate": 2.5781007751937987e-05, | |
| "loss": 0.2588, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 50.39, | |
| "learning_rate": 2.481201550387597e-05, | |
| "loss": 0.2484, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 52.33, | |
| "learning_rate": 2.384496124031008e-05, | |
| "loss": 0.251, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 54.26, | |
| "learning_rate": 2.287596899224806e-05, | |
| "loss": 0.2415, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 56.2, | |
| "learning_rate": 2.1906976744186047e-05, | |
| "loss": 0.2297, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 58.14, | |
| "learning_rate": 2.0937984496124034e-05, | |
| "loss": 0.2255, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 60.08, | |
| "learning_rate": 1.997093023255814e-05, | |
| "loss": 0.2217, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 62.02, | |
| "learning_rate": 1.9001937984496125e-05, | |
| "loss": 0.2172, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 63.95, | |
| "learning_rate": 1.8032945736434108e-05, | |
| "loss": 0.2149, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 65.89, | |
| "learning_rate": 1.7063953488372095e-05, | |
| "loss": 0.2102, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 67.83, | |
| "learning_rate": 1.6096899224806202e-05, | |
| "loss": 0.2059, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 69.77, | |
| "learning_rate": 1.5127906976744186e-05, | |
| "loss": 0.2011, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 71.71, | |
| "learning_rate": 1.415891472868217e-05, | |
| "loss": 0.1939, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 73.64, | |
| "learning_rate": 1.3189922480620156e-05, | |
| "loss": 0.1919, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 75.58, | |
| "learning_rate": 1.2224806201550389e-05, | |
| "loss": 0.1903, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 77.52, | |
| "learning_rate": 1.1257751937984498e-05, | |
| "loss": 0.1862, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 79.46, | |
| "learning_rate": 1.0288759689922481e-05, | |
| "loss": 0.1833, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 81.4, | |
| "learning_rate": 9.319767441860464e-06, | |
| "loss": 0.1811, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 83.33, | |
| "learning_rate": 8.350775193798451e-06, | |
| "loss": 0.1778, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 85.27, | |
| "learning_rate": 7.381782945736435e-06, | |
| "loss": 0.1749, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 87.21, | |
| "learning_rate": 6.412790697674419e-06, | |
| "loss": 0.1727, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 89.15, | |
| "learning_rate": 5.443798449612403e-06, | |
| "loss": 0.1666, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 91.09, | |
| "learning_rate": 4.4748062015503875e-06, | |
| "loss": 0.1707, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 93.02, | |
| "learning_rate": 3.5077519379844964e-06, | |
| "loss": 0.1687, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 94.96, | |
| "learning_rate": 2.538759689922481e-06, | |
| "loss": 0.1668, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 96.9, | |
| "learning_rate": 1.5697674418604653e-06, | |
| "loss": 0.1644, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 98.84, | |
| "learning_rate": 6.007751937984496e-07, | |
| "loss": 0.1636, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "step": 25800, | |
| "total_flos": 1.18478524985856e+17, | |
| "train_loss": 0.3071320072440214, | |
| "train_runtime": 5938.5496, | |
| "train_samples_per_second": 277.239, | |
| "train_steps_per_second": 4.344 | |
| } | |
| ], | |
| "max_steps": 25800, | |
| "num_train_epochs": 100, | |
| "total_flos": 1.18478524985856e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |