| { | |
| "best_metric": 2.75215744972229, | |
| "best_model_checkpoint": "PEFT/adapters-lib/output/parallel/dataset-5400/checkpoint-5136", | |
| "epoch": 13.0, | |
| "eval_steps": 500, | |
| "global_step": 8346, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_gen_len": 16.2889, | |
| "eval_loss": 3.750108242034912, | |
| "eval_rouge-1": 18.3573, | |
| "eval_rouge-2": 7.2689, | |
| "eval_rouge-l": 18.0787, | |
| "eval_runtime": 67.5801, | |
| "eval_samples_per_second": 6.659, | |
| "eval_steps_per_second": 0.843, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_gen_len": 18.9889, | |
| "eval_loss": 3.052328109741211, | |
| "eval_rouge-1": 35.6298, | |
| "eval_rouge-2": 18.8236, | |
| "eval_rouge-l": 35.061, | |
| "eval_runtime": 75.9845, | |
| "eval_samples_per_second": 5.922, | |
| "eval_steps_per_second": 0.75, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_gen_len": 17.7733, | |
| "eval_loss": 2.960456371307373, | |
| "eval_rouge-1": 38.5671, | |
| "eval_rouge-2": 21.0729, | |
| "eval_rouge-l": 37.9242, | |
| "eval_runtime": 70.7318, | |
| "eval_samples_per_second": 6.362, | |
| "eval_steps_per_second": 0.806, | |
| "step": 1926 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_gen_len": 17.3978, | |
| "eval_loss": 2.899169683456421, | |
| "eval_rouge-1": 40.4934, | |
| "eval_rouge-2": 22.7632, | |
| "eval_rouge-l": 39.8547, | |
| "eval_runtime": 69.4305, | |
| "eval_samples_per_second": 6.481, | |
| "eval_steps_per_second": 0.821, | |
| "step": 2568 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_gen_len": 17.2022, | |
| "eval_loss": 2.8467295169830322, | |
| "eval_rouge-1": 41.7604, | |
| "eval_rouge-2": 24.0399, | |
| "eval_rouge-l": 41.092, | |
| "eval_runtime": 70.3771, | |
| "eval_samples_per_second": 6.394, | |
| "eval_steps_per_second": 0.81, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_gen_len": 16.8956, | |
| "eval_loss": 2.812547206878662, | |
| "eval_rouge-1": 41.5256, | |
| "eval_rouge-2": 23.8131, | |
| "eval_rouge-l": 40.8416, | |
| "eval_runtime": 68.6799, | |
| "eval_samples_per_second": 6.552, | |
| "eval_steps_per_second": 0.83, | |
| "step": 3852 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_gen_len": 16.1022, | |
| "eval_loss": 2.783888816833496, | |
| "eval_rouge-1": 42.7127, | |
| "eval_rouge-2": 24.6471, | |
| "eval_rouge-l": 42.1972, | |
| "eval_runtime": 63.7095, | |
| "eval_samples_per_second": 7.063, | |
| "eval_steps_per_second": 0.895, | |
| "step": 4494 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 8.715112540192926e-05, | |
| "loss": 3.4332, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_gen_len": 15.7444, | |
| "eval_loss": 2.75215744972229, | |
| "eval_rouge-1": 42.9172, | |
| "eval_rouge-2": 25.6169, | |
| "eval_rouge-l": 42.4864, | |
| "eval_runtime": 62.162, | |
| "eval_samples_per_second": 7.239, | |
| "eval_steps_per_second": 0.917, | |
| "step": 5136 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_gen_len": 19.3733, | |
| "eval_loss": 3.0449774265289307, | |
| "eval_rouge-1": 33.8509, | |
| "eval_rouge-2": 18.7234, | |
| "eval_rouge-l": 33.1807, | |
| "eval_runtime": 80.7729, | |
| "eval_samples_per_second": 5.571, | |
| "eval_steps_per_second": 0.706, | |
| "step": 5778 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_gen_len": 18.5022, | |
| "eval_loss": 2.987593650817871, | |
| "eval_rouge-1": 34.5358, | |
| "eval_rouge-2": 18.8882, | |
| "eval_rouge-l": 33.5787, | |
| "eval_runtime": 70.1569, | |
| "eval_samples_per_second": 6.414, | |
| "eval_steps_per_second": 0.812, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_gen_len": 18.0133, | |
| "eval_loss": 2.951963424682617, | |
| "eval_rouge-1": 36.0373, | |
| "eval_rouge-2": 20.1364, | |
| "eval_rouge-l": 35.0084, | |
| "eval_runtime": 67.9106, | |
| "eval_samples_per_second": 6.626, | |
| "eval_steps_per_second": 0.839, | |
| "step": 7062 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_gen_len": 17.6444, | |
| "eval_loss": 2.919402599334717, | |
| "eval_rouge-1": 36.9999, | |
| "eval_rouge-2": 22.0519, | |
| "eval_rouge-l": 36.0277, | |
| "eval_runtime": 67.2477, | |
| "eval_samples_per_second": 6.692, | |
| "eval_steps_per_second": 0.848, | |
| "step": 7704 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_gen_len": 18.0467, | |
| "eval_loss": 2.8770110607147217, | |
| "eval_rouge-1": 39.3542, | |
| "eval_rouge-2": 23.7167, | |
| "eval_rouge-l": 38.2318, | |
| "eval_runtime": 68.3111, | |
| "eval_samples_per_second": 6.588, | |
| "eval_steps_per_second": 0.834, | |
| "step": 8346 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "step": 8346, | |
| "total_flos": 1.840140038622413e+16, | |
| "train_loss": 3.136590390419063, | |
| "train_runtime": 4227.0028, | |
| "train_samples_per_second": 60.717, | |
| "train_steps_per_second": 7.594 | |
| } | |
| ], | |
| "logging_steps": 5000, | |
| "max_steps": 32100, | |
| "num_train_epochs": 50, | |
| "save_steps": 500, | |
| "total_flos": 1.840140038622413e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |