| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "global_step": 25000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.9604e-05, |
| "loss": 2.46, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.9204e-05, |
| "loss": 2.2671, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.8804e-05, |
| "loss": 2.2099, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.8404000000000002e-05, |
| "loss": 2.1318, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.80048e-05, |
| "loss": 2.1846, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.76048e-05, |
| "loss": 2.1693, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.7204800000000002e-05, |
| "loss": 2.1344, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.6804800000000002e-05, |
| "loss": 2.0914, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.64048e-05, |
| "loss": 2.0775, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.60056e-05, |
| "loss": 2.0589, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.5605600000000002e-05, |
| "loss": 2.1024, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.52056e-05, |
| "loss": 2.0772, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.4806400000000001e-05, |
| "loss": 2.0638, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.44064e-05, |
| "loss": 2.0966, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.4006400000000002e-05, |
| "loss": 1.9985, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.3606400000000002e-05, |
| "loss": 2.0813, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.3206400000000001e-05, |
| "loss": 2.0429, |
| "step": 8500 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.2806400000000002e-05, |
| "loss": 2.0296, |
| "step": 9000 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.24064e-05, |
| "loss": 2.0589, |
| "step": 9500 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.20064e-05, |
| "loss": 1.9988, |
| "step": 10000 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.16064e-05, |
| "loss": 2.0147, |
| "step": 10500 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.1206400000000002e-05, |
| "loss": 2.0093, |
| "step": 11000 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.0806400000000002e-05, |
| "loss": 2.0379, |
| "step": 11500 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.0407200000000001e-05, |
| "loss": 2.0037, |
| "step": 12000 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.0007200000000001e-05, |
| "loss": 1.9776, |
| "step": 12500 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.607200000000002e-06, |
| "loss": 1.9606, |
| "step": 13000 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.2072e-06, |
| "loss": 2.028, |
| "step": 13500 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.808000000000001e-06, |
| "loss": 1.9674, |
| "step": 14000 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 8.408e-06, |
| "loss": 1.9895, |
| "step": 14500 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 8.008e-06, |
| "loss": 1.9596, |
| "step": 15000 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 7.608000000000001e-06, |
| "loss": 2.0047, |
| "step": 15500 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 7.208e-06, |
| "loss": 1.9908, |
| "step": 16000 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 6.808800000000001e-06, |
| "loss": 1.9246, |
| "step": 16500 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 6.4088e-06, |
| "loss": 1.9711, |
| "step": 17000 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 6.0088e-06, |
| "loss": 1.9285, |
| "step": 17500 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 5.608800000000001e-06, |
| "loss": 1.9388, |
| "step": 18000 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 5.2096e-06, |
| "loss": 1.9361, |
| "step": 18500 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 4.8096000000000005e-06, |
| "loss": 1.9566, |
| "step": 19000 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 4.4104e-06, |
| "loss": 1.9158, |
| "step": 19500 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 4.010400000000001e-06, |
| "loss": 1.927, |
| "step": 20000 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.6104000000000006e-06, |
| "loss": 1.9152, |
| "step": 20500 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.2104000000000002e-06, |
| "loss": 1.906, |
| "step": 21000 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 2.8104000000000002e-06, |
| "loss": 1.9252, |
| "step": 21500 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 2.4104000000000002e-06, |
| "loss": 1.9316, |
| "step": 22000 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 2.0112e-06, |
| "loss": 1.8907, |
| "step": 22500 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 1.6112000000000003e-06, |
| "loss": 1.8786, |
| "step": 23000 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.2112000000000002e-06, |
| "loss": 1.9196, |
| "step": 23500 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 8.112000000000001e-07, |
| "loss": 1.9025, |
| "step": 24000 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 4.112e-07, |
| "loss": 1.9201, |
| "step": 24500 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 1.12e-08, |
| "loss": 1.8927, |
| "step": 25000 |
| } |
| ], |
| "max_steps": 25000, |
| "num_train_epochs": 1, |
| "total_flos": 4.818482588575334e+16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
