| { | |
| "best_metric": 1.2186108827590942, | |
| "best_model_checkpoint": "wikitext_roberta-base/checkpoint-666", | |
| "epoch": 19.986666666666668, | |
| "global_step": 740, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.8988, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.8877, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3e-06, | |
| "loss": 1.9457, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.8727, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5e-06, | |
| "loss": 1.9502, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6e-06, | |
| "loss": 1.911, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 1.925, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.8293, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9e-06, | |
| "loss": 1.7548, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8128, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 1.7672, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.789, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 1.7381, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 1.6687, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.7386, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.6627, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 1.5889, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.5649, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.9e-05, | |
| "loss": 1.5465, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2e-05, | |
| "loss": 1.5523, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2e-05, | |
| "loss": 1.5625, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.639, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 1.5218, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 1.5603, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.5555, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.5e-05, | |
| "loss": 1.5245, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 1.5195, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.5227, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 1.4743, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.9e-05, | |
| "loss": 1.5171, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3e-05, | |
| "loss": 1.4961, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.1e-05, | |
| "loss": 1.5427, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 1.4519, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.3e-05, | |
| "loss": 1.4714, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 1.4477, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.4796, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.6e-05, | |
| "loss": 1.4175, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_accuracy": 0.7193657677192031, | |
| "eval_loss": 1.3355050086975098, | |
| "eval_runtime": 13.2653, | |
| "eval_samples_per_second": 37.391, | |
| "eval_steps_per_second": 4.674, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.7e-05, | |
| "loss": 2.1673, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.8e-05, | |
| "loss": 1.4677, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 1.4678, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4e-05, | |
| "loss": 1.4979, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.1e-05, | |
| "loss": 1.4639, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.2e-05, | |
| "loss": 1.4553, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.3e-05, | |
| "loss": 1.3852, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.4783, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.416, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 1.4261, | |
| "step": 47 | |
| }, | |