Files
llama-3-8b-base-beta-dpo-hh…/all_results.json
ModelHub XC 2349f340b8 初始化项目,由ModelHub XC社区提供模型
Model: W-61/llama-3-8b-base-beta-dpo-hh-harmless-8xh200
Source: Original Platform
2026-05-25 19:35:17 +08:00

21 lines
730 B
JSON

{
"epoch": 1.0,
"eval_beta_dpo/beta_used": 0.09322389215230942,
"eval_beta_dpo/beta_used_raw": 0.09322389215230942,
"eval_beta_dpo/gap_mean": 9.061779022216797,
"eval_beta_dpo/gap_std": 15.212827682495117,
"eval_beta_dpo/mask_keep_frac": 1.0,
"eval_logits/chosen": -0.42951661348342896,
"eval_logits/rejected": -0.41630053520202637,
"eval_loss": 0.5633630752563477,
"eval_runtime": 18.8064,
"eval_samples": 2303,
"eval_samples_per_second": 122.458,
"eval_steps_per_second": 0.957,
"total_flos": 0.0,
"train_loss": 0.5772968926213005,
"train_runtime": 1407.4268,
"train_samples": 42336,
"train_samples_per_second": 30.08,
"train_steps_per_second": 0.234
}