| { | |
| "best_metric": 0.41696879267692566, | |
| "best_model_checkpoint": "tiny_bert_rand_50_v1_qqp/checkpoint-4266", | |
| "epoch": 8.0, | |
| "eval_steps": 500, | |
| "global_step": 11376, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.562728762626648, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.4963, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7838980954736582, | |
| "eval_combined_score": 0.7335150724711706, | |
| "eval_f1": 0.6831320494686831, | |
| "eval_loss": 0.45065996050834656, | |
| "eval_runtime": 14.2982, | |
| "eval_samples_per_second": 2827.632, | |
| "eval_steps_per_second": 11.05, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 2.0054335594177246, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.4096, | |
| "step": 2844 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8030175612169181, | |
| "eval_combined_score": 0.7601426914292486, | |
| "eval_f1": 0.7172678216415791, | |
| "eval_loss": 0.4217735528945923, | |
| "eval_runtime": 14.2195, | |
| "eval_samples_per_second": 2843.269, | |
| "eval_steps_per_second": 11.111, | |
| "step": 2844 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 3.368314027786255, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.3533, | |
| "step": 4266 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8144694533762058, | |
| "eval_combined_score": 0.7785063930687448, | |
| "eval_f1": 0.7425433327612837, | |
| "eval_loss": 0.41696879267692566, | |
| "eval_runtime": 14.0807, | |
| "eval_samples_per_second": 2871.299, | |
| "eval_steps_per_second": 11.221, | |
| "step": 4266 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 3.216777801513672, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.3086, | |
| "step": 5688 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8168439277764037, | |
| "eval_combined_score": 0.7798048797877398, | |
| "eval_f1": 0.742765831799076, | |
| "eval_loss": 0.4282667338848114, | |
| "eval_runtime": 14.1794, | |
| "eval_samples_per_second": 2851.323, | |
| "eval_steps_per_second": 11.143, | |
| "step": 5688 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 2.892664909362793, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.2704, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8229532525352461, | |
| "eval_combined_score": 0.7891009167688882, | |
| "eval_f1": 0.7552485810025302, | |
| "eval_loss": 0.4386835992336273, | |
| "eval_runtime": 14.1116, | |
| "eval_samples_per_second": 2865.017, | |
| "eval_steps_per_second": 11.196, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 4.031663417816162, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.2413, | |
| "step": 8532 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.818748454118229, | |
| "eval_combined_score": 0.7912949489411641, | |
| "eval_f1": 0.7638414437640992, | |
| "eval_loss": 0.45857173204421997, | |
| "eval_runtime": 14.0672, | |
| "eval_samples_per_second": 2874.053, | |
| "eval_steps_per_second": 11.232, | |
| "step": 8532 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 4.435065269470215, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.215, | |
| "step": 9954 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8246846401187237, | |
| "eval_combined_score": 0.7946718240104874, | |
| "eval_f1": 0.7646590079022512, | |
| "eval_loss": 0.48123985528945923, | |
| "eval_runtime": 14.088, | |
| "eval_samples_per_second": 2869.812, | |
| "eval_steps_per_second": 11.215, | |
| "step": 9954 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 4.721343517303467, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.193, | |
| "step": 11376 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.8243630967103636, | |
| "eval_combined_score": 0.793399975306438, | |
| "eval_f1": 0.7624368539025125, | |
| "eval_loss": 0.5110756754875183, | |
| "eval_runtime": 14.3423, | |
| "eval_samples_per_second": 2818.93, | |
| "eval_steps_per_second": 11.016, | |
| "step": 11376 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "step": 11376, | |
| "total_flos": 7.633075201391002e+16, | |
| "train_loss": 0.31095045494798534, | |
| "train_runtime": 2030.4673, | |
| "train_samples_per_second": 8959.662, | |
| "train_steps_per_second": 35.017 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 71100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "save_steps": 500, | |
| "total_flos": 7.633075201391002e+16, | |
| "train_batch_size": 256, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |