| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 125000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9e-05, | |
| "loss": 2.4039, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.8e-05, | |
| "loss": 2.241, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.7e-05, | |
| "loss": 2.1848, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 2.1211, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.5e-05, | |
| "loss": 2.078, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 2.017, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.3e-05, | |
| "loss": 1.9922, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.2e-05, | |
| "loss": 1.9616, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.1e-05, | |
| "loss": 1.9331, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4e-05, | |
| "loss": 1.9139, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 1.8636, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.8e-05, | |
| "loss": 1.8417, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.7e-05, | |
| "loss": 1.8372, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.6e-05, | |
| "loss": 1.8238, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.7966, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 1.7481, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.3e-05, | |
| "loss": 1.7472, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 1.7346, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.1e-05, | |
| "loss": 1.7196, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3e-05, | |
| "loss": 1.7059, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 2.9e-05, | |
| "loss": 1.6791, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 1.6594, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.6478, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 1.6371, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.6264, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.5921, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 1.5893, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 1.5792, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.5578, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2e-05, | |
| "loss": 1.5571, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 1.9e-05, | |
| "loss": 1.5221, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.5163, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 1.5183, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.4948, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.4956, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 1.4621, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 1.4563, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.435, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 1.4374, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 1e-05, | |
| "loss": 1.4418, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "learning_rate": 9e-06, | |
| "loss": 1.4116, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.4004, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 1.3932, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 6e-06, | |
| "loss": 1.3945, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 5e-06, | |
| "loss": 1.3841, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.3666, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 3e-06, | |
| "loss": 1.3577, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.3621, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.3601, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.3639, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 125000, | |
| "train_runtime": 35714.6581, | |
| "train_samples_per_second": 3.5 | |
| } | |
| ], | |
| "max_steps": 125000, | |
| "num_train_epochs": 10, | |
| "total_flos": 144180534768813144, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |