| { | |
| "best_metric": 0.42420005798339844, | |
| "best_model_checkpoint": "tiny_bert_km_100_v1_qqp/checkpoint-2844", | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 9954, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.8514636754989624, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.4988, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.782067771456839, | |
| "eval_combined_score": 0.7278249056224662, | |
| "eval_f1": 0.6735820397880932, | |
| "eval_loss": 0.45090559124946594, | |
| "eval_runtime": 14.3079, | |
| "eval_samples_per_second": 2825.71, | |
| "eval_steps_per_second": 11.043, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 1.9057090282440186, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.4121, | |
| "step": 2844 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8003957457333664, | |
| "eval_combined_score": 0.7589954652519015, | |
| "eval_f1": 0.7175951847704367, | |
| "eval_loss": 0.42420005798339844, | |
| "eval_runtime": 14.2122, | |
| "eval_samples_per_second": 2844.741, | |
| "eval_steps_per_second": 11.117, | |
| "step": 2844 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 2.9178097248077393, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.3559, | |
| "step": 4266 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8094484293841206, | |
| "eval_combined_score": 0.7703114852622327, | |
| "eval_f1": 0.7311745411403447, | |
| "eval_loss": 0.42494192719459534, | |
| "eval_runtime": 14.1152, | |
| "eval_samples_per_second": 2864.287, | |
| "eval_steps_per_second": 11.194, | |
| "step": 4266 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 3.0772738456726074, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.3085, | |
| "step": 5688 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8128122681177343, | |
| "eval_combined_score": 0.7678691200733652, | |
| "eval_f1": 0.7229259720289961, | |
| "eval_loss": 0.4429996609687805, | |
| "eval_runtime": 14.07, | |
| "eval_samples_per_second": 2873.494, | |
| "eval_steps_per_second": 11.23, | |
| "step": 5688 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 3.853461980819702, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.2704, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8162750432846896, | |
| "eval_combined_score": 0.7877563470021269, | |
| "eval_f1": 0.7592376507195644, | |
| "eval_loss": 0.43376293778419495, | |
| "eval_runtime": 14.0745, | |
| "eval_samples_per_second": 2872.565, | |
| "eval_steps_per_second": 11.226, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 2.85068941116333, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.2367, | |
| "step": 8532 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8189710610932476, | |
| "eval_combined_score": 0.7891982478986272, | |
| "eval_f1": 0.7594254347040068, | |
| "eval_loss": 0.44623711705207825, | |
| "eval_runtime": 14.7957, | |
| "eval_samples_per_second": 2732.543, | |
| "eval_steps_per_second": 10.679, | |
| "step": 8532 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 5.932079315185547, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.2093, | |
| "step": 9954 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8209003215434083, | |
| "eval_combined_score": 0.789698401441417, | |
| "eval_f1": 0.7584964813394257, | |
| "eval_loss": 0.4795359671115875, | |
| "eval_runtime": 14.7926, | |
| "eval_samples_per_second": 2733.125, | |
| "eval_steps_per_second": 10.681, | |
| "step": 9954 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "step": 9954, | |
| "total_flos": 6.678940801217126e+16, | |
| "train_loss": 0.3273723741025122, | |
| "train_runtime": 1785.6259, | |
| "train_samples_per_second": 10188.192, | |
| "train_steps_per_second": 39.818 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 71100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "save_steps": 500, | |
| "total_flos": 6.678940801217126e+16, | |
| "train_batch_size": 256, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |