| { | |
| "best_metric": 0.9517181383210092, | |
| "best_model_checkpoint": "logs/unfair_tos/bert-base-uncased/seed_1/checkpoint-2076", | |
| "epoch": 6.0, | |
| "eval_steps": 500, | |
| "global_step": 4152, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.7225433526011561, | |
| "grad_norm": 0.4137115776538849, | |
| "learning_rate": 2.8916184971098265e-05, | |
| "loss": 0.0965, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.03946217894554138, | |
| "eval_macro-f1": 0.3257253069723768, | |
| "eval_micro-f1": 0.9201138104618078, | |
| "eval_runtime": 3.109, | |
| "eval_samples_per_second": 731.749, | |
| "eval_steps_per_second": 91.67, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.4450867052023122, | |
| "grad_norm": 0.5951205492019653, | |
| "learning_rate": 2.7832369942196533e-05, | |
| "loss": 0.0428, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.030047794803977013, | |
| "eval_macro-f1": 0.5948063101905734, | |
| "eval_micro-f1": 0.926039387308534, | |
| "eval_runtime": 3.1251, | |
| "eval_samples_per_second": 727.972, | |
| "eval_steps_per_second": 91.196, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 2.167630057803468, | |
| "grad_norm": 0.040406469255685806, | |
| "learning_rate": 2.6748554913294797e-05, | |
| "loss": 0.0294, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.8901734104046244, | |
| "grad_norm": 0.47717490792274475, | |
| "learning_rate": 2.5664739884393065e-05, | |
| "loss": 0.0202, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.025113314390182495, | |
| "eval_macro-f1": 0.7620096966057794, | |
| "eval_micro-f1": 0.9517181383210092, | |
| "eval_runtime": 3.0811, | |
| "eval_samples_per_second": 738.38, | |
| "eval_steps_per_second": 92.5, | |
| "step": 2076 | |
| }, | |
| { | |
| "epoch": 3.61271676300578, | |
| "grad_norm": 0.00729083176702261, | |
| "learning_rate": 2.458092485549133e-05, | |
| "loss": 0.0136, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.028495851904153824, | |
| "eval_macro-f1": 0.7233574363382008, | |
| "eval_micro-f1": 0.937227550130776, | |
| "eval_runtime": 3.1146, | |
| "eval_samples_per_second": 730.425, | |
| "eval_steps_per_second": 91.504, | |
| "step": 2768 | |
| }, | |
| { | |
| "epoch": 4.335260115606936, | |
| "grad_norm": 0.01607728749513626, | |
| "learning_rate": 2.3497109826589597e-05, | |
| "loss": 0.01, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.030019745230674744, | |
| "eval_macro-f1": 0.7251725230841095, | |
| "eval_micro-f1": 0.9452084697664266, | |
| "eval_runtime": 3.0696, | |
| "eval_samples_per_second": 741.132, | |
| "eval_steps_per_second": 92.845, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 5.057803468208093, | |
| "grad_norm": 0.006081081461161375, | |
| "learning_rate": 2.2413294797687862e-05, | |
| "loss": 0.0087, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.780346820809249, | |
| "grad_norm": 4.081294536590576, | |
| "learning_rate": 2.132947976878613e-05, | |
| "loss": 0.0068, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 0.02861410565674305, | |
| "eval_macro-f1": 0.7558502112595478, | |
| "eval_micro-f1": 0.9500763025942882, | |
| "eval_runtime": 3.0792, | |
| "eval_samples_per_second": 738.833, | |
| "eval_steps_per_second": 92.557, | |
| "step": 4152 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "step": 4152, | |
| "total_flos": 2183413154955264.0, | |
| "train_loss": 0.027697743765421227, | |
| "train_runtime": 253.8343, | |
| "train_samples_per_second": 435.875, | |
| "train_steps_per_second": 54.524 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 13840, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 2183413154955264.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |