| { | |
| "best_metric": 3.7314870357513428, | |
| "best_model_checkpoint": "output/enigma/checkpoint-66", | |
| "epoch": 2.0, | |
| "global_step": 66, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00012957411177772776, | |
| "loss": 4.4237, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001083919039965842, | |
| "loss": 4.1256, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.836279790554737e-05, | |
| "loss": 3.9414, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.616313771642488e-05, | |
| "loss": 3.7963, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.8951844985992186e-05, | |
| "loss": 3.7966, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.778782010045488e-06, | |
| "loss": 3.3801, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 3.9204065799713135, | |
| "eval_runtime": 2.3835, | |
| "eval_samples_per_second": 20.139, | |
| "eval_steps_per_second": 2.517, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.2396913677783124e-06, | |
| "loss": 3.4099, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.4676757700644785e-05, | |
| "loss": 3.9212, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.0102530108070535e-05, | |
| "loss": 3.831, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.186411942550872e-05, | |
| "loss": 3.4868, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00010290000000000001, | |
| "loss": 3.5598, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.000126309992352219, | |
| "loss": 3.6865, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0001368893738885136, | |
| "loss": 3.4504, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 3.7314870357513428, | |
| "eval_runtime": 1.9498, | |
| "eval_samples_per_second": 23.079, | |
| "eval_steps_per_second": 3.077, | |
| "step": 66 | |
| } | |
| ], | |
| "max_steps": 66, | |
| "num_train_epochs": 2, | |
| "total_flos": 68066574336000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |