| { | |
| "best_metric": 0.68276149, | |
| "best_model_checkpoint": "/mnt/cachenew/yangzekang/wsdm_lym/swift_wsdm/output/solar-10-7b-instruct-v1/v13-20240206-111010/checkpoint-1700", | |
| "epoch": 1.9144144144144144, | |
| "eval_steps": 1700, | |
| "global_step": 1700, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "acc": 0.67516661, | |
| "epoch": 0.0, | |
| "learning_rate": 9.345794392523364e-07, | |
| "loss": 1.46797299, | |
| "step": 1 | |
| }, | |
| { | |
| "acc": 0.75082575, | |
| "epoch": 0.11, | |
| "learning_rate": 9.252336448598131e-05, | |
| "loss": 0.95137394, | |
| "step": 100 | |
| }, | |
| { | |
| "acc": 0.78081596, | |
| "epoch": 0.23, | |
| "learning_rate": 9.732946298984034e-05, | |
| "loss": 0.77715546, | |
| "step": 200 | |
| }, | |
| { | |
| "acc": 0.7879702, | |
| "epoch": 0.34, | |
| "learning_rate": 9.44267053701016e-05, | |
| "loss": 0.7487645, | |
| "step": 300 | |
| }, | |
| { | |
| "acc": 0.79198853, | |
| "epoch": 0.45, | |
| "learning_rate": 9.152394775036285e-05, | |
| "loss": 0.73725586, | |
| "step": 400 | |
| }, | |
| { | |
| "acc": 0.79181892, | |
| "epoch": 0.56, | |
| "learning_rate": 8.86211901306241e-05, | |
| "loss": 0.73659386, | |
| "step": 500 | |
| }, | |
| { | |
| "acc": 0.791054, | |
| "epoch": 0.68, | |
| "learning_rate": 8.571843251088535e-05, | |
| "loss": 0.73336884, | |
| "step": 600 | |
| }, | |
| { | |
| "acc": 0.79418472, | |
| "epoch": 0.79, | |
| "learning_rate": 8.284470246734399e-05, | |
| "loss": 0.72362808, | |
| "step": 700 | |
| }, | |
| { | |
| "acc": 0.79279587, | |
| "epoch": 0.9, | |
| "learning_rate": 7.994194484760524e-05, | |
| "loss": 0.72401505, | |
| "step": 800 | |
| }, | |
| { | |
| "acc": 0.79312164, | |
| "epoch": 1.01, | |
| "learning_rate": 7.703918722786648e-05, | |
| "loss": 0.7227565, | |
| "step": 900 | |
| }, | |
| { | |
| "acc": 0.8135437, | |
| "epoch": 1.13, | |
| "learning_rate": 7.413642960812773e-05, | |
| "loss": 0.63391575, | |
| "step": 1000 | |
| }, | |
| { | |
| "acc": 0.81166725, | |
| "epoch": 1.24, | |
| "learning_rate": 7.123367198838897e-05, | |
| "loss": 0.63856529, | |
| "step": 1100 | |
| }, | |
| { | |
| "acc": 0.81463379, | |
| "epoch": 1.35, | |
| "learning_rate": 6.833091436865022e-05, | |
| "loss": 0.63027618, | |
| "step": 1200 | |
| }, | |
| { | |
| "acc": 0.81466286, | |
| "epoch": 1.46, | |
| "learning_rate": 6.542815674891147e-05, | |
| "loss": 0.62910736, | |
| "step": 1300 | |
| }, | |
| { | |
| "acc": 0.81553825, | |
| "epoch": 1.58, | |
| "learning_rate": 6.252539912917271e-05, | |
| "loss": 0.62564072, | |
| "step": 1400 | |
| }, | |
| { | |
| "acc": 0.81496376, | |
| "epoch": 1.69, | |
| "learning_rate": 5.9622641509433966e-05, | |
| "loss": 0.62815582, | |
| "step": 1500 | |
| }, | |
| { | |
| "acc": 0.81624107, | |
| "epoch": 1.8, | |
| "learning_rate": 5.671988388969521e-05, | |
| "loss": 0.6277507, | |
| "step": 1600 | |
| }, | |
| { | |
| "acc": 0.81836472, | |
| "epoch": 1.91, | |
| "learning_rate": 5.381712626995646e-05, | |
| "loss": 0.61489536, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_acc": 0.8124290204157093, | |
| "eval_loss": 0.6827614903450012, | |
| "eval_runtime": 19.6247, | |
| "eval_samples_per_second": 7.338, | |
| "eval_steps_per_second": 3.669, | |
| "step": 1700 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 3552, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 1700, | |
| "total_flos": 2.357705064388231e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |