| { | |
| "best_metric": 0.6843704, | |
| "best_model_checkpoint": "/mnt/cachenew/yangzekang/wsdm_lym/swift_wsdm/output/solar-10-7b-instruct-v1/v27-20240209-133614/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": 0.75091121, | |
| "epoch": 0.11, | |
| "learning_rate": 9.158878504672898e-05, | |
| "loss": 0.95139783, | |
| "step": 100 | |
| }, | |
| { | |
| "acc": 0.78103409, | |
| "epoch": 0.23, | |
| "learning_rate": 9.735849056603774e-05, | |
| "loss": 0.77608315, | |
| "step": 200 | |
| }, | |
| { | |
| "acc": 0.78829514, | |
| "epoch": 0.34, | |
| "learning_rate": 9.4455732946299e-05, | |
| "loss": 0.74766434, | |
| "step": 300 | |
| }, | |
| { | |
| "acc": 0.79177124, | |
| "epoch": 0.45, | |
| "learning_rate": 9.155297532656024e-05, | |
| "loss": 0.73638321, | |
| "step": 400 | |
| }, | |
| { | |
| "acc": 0.79178207, | |
| "epoch": 0.56, | |
| "learning_rate": 8.865021770682148e-05, | |
| "loss": 0.73636147, | |
| "step": 500 | |
| }, | |
| { | |
| "acc": 0.79106674, | |
| "epoch": 0.68, | |
| "learning_rate": 8.574746008708273e-05, | |
| "loss": 0.73325569, | |
| "step": 600 | |
| }, | |
| { | |
| "acc": 0.79414444, | |
| "epoch": 0.79, | |
| "learning_rate": 8.284470246734399e-05, | |
| "loss": 0.72352669, | |
| "step": 700 | |
| }, | |
| { | |
| "acc": 0.79252029, | |
| "epoch": 0.9, | |
| "learning_rate": 7.994194484760524e-05, | |
| "loss": 0.72396622, | |
| "step": 800 | |
| }, | |
| { | |
| "acc": 0.793386, | |
| "epoch": 1.01, | |
| "learning_rate": 7.703918722786648e-05, | |
| "loss": 0.72276489, | |
| "step": 900 | |
| }, | |
| { | |
| "acc": 0.81338135, | |
| "epoch": 1.13, | |
| "learning_rate": 7.413642960812773e-05, | |
| "loss": 0.63367615, | |
| "step": 1000 | |
| }, | |
| { | |
| "acc": 0.81188278, | |
| "epoch": 1.24, | |
| "learning_rate": 7.123367198838897e-05, | |
| "loss": 0.63777443, | |
| "step": 1100 | |
| }, | |
| { | |
| "acc": 0.81501755, | |
| "epoch": 1.35, | |
| "learning_rate": 6.833091436865022e-05, | |
| "loss": 0.62992439, | |
| "step": 1200 | |
| }, | |
| { | |
| "acc": 0.81464638, | |
| "epoch": 1.46, | |
| "learning_rate": 6.542815674891147e-05, | |
| "loss": 0.62890068, | |
| "step": 1300 | |
| }, | |
| { | |
| "acc": 0.81589249, | |
| "epoch": 1.58, | |
| "learning_rate": 6.252539912917271e-05, | |
| "loss": 0.62468796, | |
| "step": 1400 | |
| }, | |
| { | |
| "acc": 0.8147139, | |
| "epoch": 1.69, | |
| "learning_rate": 5.9622641509433966e-05, | |
| "loss": 0.62785313, | |
| "step": 1500 | |
| }, | |
| { | |
| "acc": 0.81618683, | |
| "epoch": 1.8, | |
| "learning_rate": 5.671988388969521e-05, | |
| "loss": 0.62735779, | |
| "step": 1600 | |
| }, | |
| { | |
| "acc": 0.81866211, | |
| "epoch": 1.91, | |
| "learning_rate": 5.381712626995646e-05, | |
| "loss": 0.61408382, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_acc": 0.8115838681562475, | |
| "eval_loss": 0.6843703985214233, | |
| "eval_runtime": 19.3211, | |
| "eval_samples_per_second": 7.453, | |
| "eval_steps_per_second": 3.726, | |
| "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 | |
| } | |