| { | |
| "best_metric": 2.68359375, | |
| "best_model_checkpoint": "/home/zhengxinyong/outputs/bloom-1b1_de_continual-pretrain_100000samples_-1vocab_original/checkpoint-25000", | |
| "epoch": 2.7475546763380594, | |
| "global_step": 25000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0001, | |
| "loss": 3.1701, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0001, | |
| "loss": 2.8341, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_loss": 2.873046875, | |
| "eval_runtime": 3.4814, | |
| "eval_samples_per_second": 230.94, | |
| "eval_steps_per_second": 29.011, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0001, | |
| "loss": 2.7255, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0001, | |
| "loss": 2.5869, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "eval_loss": 2.759765625, | |
| "eval_runtime": 3.473, | |
| "eval_samples_per_second": 231.499, | |
| "eval_steps_per_second": 29.081, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0001, | |
| "loss": 2.4538, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0001, | |
| "loss": 2.4623, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "eval_loss": 2.705078125, | |
| "eval_runtime": 3.4584, | |
| "eval_samples_per_second": 232.479, | |
| "eval_steps_per_second": 29.204, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.0001, | |
| "loss": 2.4534, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.0001, | |
| "loss": 2.2463, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_loss": 2.712890625, | |
| "eval_runtime": 3.4672, | |
| "eval_samples_per_second": 231.886, | |
| "eval_steps_per_second": 29.13, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 0.0001, | |
| "loss": 2.2084, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.0001, | |
| "loss": 2.2451, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "eval_loss": 2.68359375, | |
| "eval_runtime": 3.4826, | |
| "eval_samples_per_second": 230.861, | |
| "eval_steps_per_second": 29.001, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "step": 25000, | |
| "total_flos": 8.355380415011226e+17, | |
| "train_loss": 2.53857546875, | |
| "train_runtime": 4808.8208, | |
| "train_samples_per_second": 41.59, | |
| "train_steps_per_second": 5.199 | |
| } | |
| ], | |
| "max_steps": 25000, | |
| "num_train_epochs": 3, | |
| "total_flos": 8.355380415011226e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |