| { | |
| "best_metric": 0.6759016194822504, | |
| "best_model_checkpoint": "omarmomen/tf_babylm_1/finetune/main_verb_relative_token_position/checkpoint-1200", | |
| "epoch": 10.0, | |
| "global_step": 1280, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.56, | |
| "eval_accuracy": 0.6680949330329895, | |
| "eval_f1": 0.6634077045207655, | |
| "eval_loss": 2.8229305744171143, | |
| "eval_mcc": 0.3361689150945025, | |
| "eval_runtime": 32.1081, | |
| "eval_samples_per_second": 755.947, | |
| "eval_steps_per_second": 94.493, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "eval_accuracy": 0.6663645505905151, | |
| "eval_f1": 0.6715340309888862, | |
| "eval_loss": 2.9007394313812256, | |
| "eval_mcc": 0.3331342188992681, | |
| "eval_runtime": 32.2031, | |
| "eval_samples_per_second": 753.716, | |
| "eval_steps_per_second": 94.214, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.0468750000000002e-05, | |
| "loss": 0.0336, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "eval_accuracy": 0.6681361198425293, | |
| "eval_f1": 0.674637476269338, | |
| "eval_loss": 3.2524144649505615, | |
| "eval_mcc": 0.33683820534555325, | |
| "eval_runtime": 32.2451, | |
| "eval_samples_per_second": 752.735, | |
| "eval_steps_per_second": 94.092, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "eval_accuracy": 0.668506920337677, | |
| "eval_f1": 0.6750666343590986, | |
| "eval_loss": 3.3832876682281494, | |
| "eval_mcc": 0.3375888849891447, | |
| "eval_runtime": 32.0441, | |
| "eval_samples_per_second": 757.456, | |
| "eval_steps_per_second": 94.682, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1.09375e-05, | |
| "loss": 0.0001, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "eval_accuracy": 0.6669001579284668, | |
| "eval_f1": 0.6638672930611567, | |
| "eval_loss": 3.474618434906006, | |
| "eval_mcc": 0.3337682158312302, | |
| "eval_runtime": 31.9554, | |
| "eval_samples_per_second": 759.559, | |
| "eval_steps_per_second": 94.945, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "eval_accuracy": 0.6693721413612366, | |
| "eval_f1": 0.6759016194822504, | |
| "eval_loss": 3.4603095054626465, | |
| "eval_mcc": 0.33931939844597186, | |
| "eval_runtime": 32.022, | |
| "eval_samples_per_second": 757.979, | |
| "eval_steps_per_second": 94.747, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 1280, | |
| "total_flos": 2243671795921920.0, | |
| "train_loss": 0.013182738378236535, | |
| "train_runtime": 408.6986, | |
| "train_samples_per_second": 200.074, | |
| "train_steps_per_second": 3.132 | |
| } | |
| ], | |
| "max_steps": 1280, | |
| "num_train_epochs": 10, | |
| "total_flos": 2243671795921920.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |