初始化项目,由ModelHub XC社区提供模型

Model: CriteriaPO/qwen2.5-3b-dpo-mini
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-08-15 05:21:18 +08:00
commit b7f7e4e78f
58 changed files with 340467 additions and 0 deletions

16
eval_results.json Normal file
View File

@@ -0,0 +1,16 @@
{
"epoch": 1.9999360981532366,
"eval_logits/chosen": -2.984375,
"eval_logits/rejected": -2.765625,
"eval_logps/chosen": -720.0,
"eval_logps/rejected": -1080.0,
"eval_loss": 0.23926551640033722,
"eval_rewards/accuracies": 0.8974654674530029,
"eval_rewards/chosen": -5.625,
"eval_rewards/margins": 3.78125,
"eval_rewards/rejected": -9.375,
"eval_runtime": 1835.1519,
"eval_samples": 55582,
"eval_samples_per_second": 30.246,
"eval_steps_per_second": 0.473
}