| { | |
| "best_metric": 3.520382881164551, | |
| "best_model_checkpoint": "/users/zyong2/data/zyong2/bigscience/data/processed/024/bloom-350m_az_bitfit_10000samples_-1vocab_original-frozen/checkpoint-25000", | |
| "epoch": 19.856519150625125, | |
| "global_step": 25000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9e-05, | |
| "loss": 4.249, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 8e-05, | |
| "loss": 3.9172, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "eval_loss": 3.8250739574432373, | |
| "eval_runtime": 519.4242, | |
| "eval_samples_per_second": 9.445, | |
| "eval_steps_per_second": 4.723, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 7e-05, | |
| "loss": 3.7702, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 6e-05, | |
| "loss": 3.6827, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "eval_loss": 3.64797043800354, | |
| "eval_runtime": 519.0865, | |
| "eval_samples_per_second": 9.451, | |
| "eval_steps_per_second": 4.726, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 5e-05, | |
| "loss": 3.624, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "learning_rate": 4e-05, | |
| "loss": 3.585, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "eval_loss": 3.5690791606903076, | |
| "eval_runtime": 519.0226, | |
| "eval_samples_per_second": 9.452, | |
| "eval_steps_per_second": 4.726, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 13.9, | |
| "learning_rate": 3e-05, | |
| "loss": 3.5588, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 2e-05, | |
| "loss": 3.54, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "eval_loss": 3.531822443008423, | |
| "eval_runtime": 518.4812, | |
| "eval_samples_per_second": 9.462, | |
| "eval_steps_per_second": 4.731, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "learning_rate": 1e-05, | |
| "loss": 3.5249, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 19.86, | |
| "learning_rate": 0.0, | |
| "loss": 3.5218, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 19.86, | |
| "eval_loss": 3.520382881164551, | |
| "eval_runtime": 518.5767, | |
| "eval_samples_per_second": 9.461, | |
| "eval_steps_per_second": 4.73, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 19.86, | |
| "step": 25000, | |
| "total_flos": 3.7165924872093696e+17, | |
| "train_loss": 3.6973518359375, | |
| "train_runtime": 53135.1163, | |
| "train_samples_per_second": 3.764, | |
| "train_steps_per_second": 0.47 | |
| } | |
| ], | |
| "max_steps": 25000, | |
| "num_train_epochs": 20, | |
| "total_flos": 3.7165924872093696e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |