Files
TinyLlama-1.1B-IPO-PKU-Safe…/eval_results.json
ModelHub XC d540a44812 初始化项目,由ModelHub XC社区提供模型
Model: AIPlans/TinyLlama-1.1B-IPO-PKU-SafeRLHF
Source: Original Platform
2026-06-06 11:06:21 +08:00

15 lines
563 B
JSON

{
"epoch": 0.9997600191984641,
"eval_logits/chosen": -2.4037210941314697,
"eval_logits/rejected": -2.271228313446045,
"eval_logps/chosen": -5.422507286071777,
"eval_logps/rejected": -6.740255355834961,
"eval_loss": 18.408336639404297,
"eval_rewards/accuracies": 0.738882839679718,
"eval_rewards/chosen": -0.34205058217048645,
"eval_rewards/margins": 0.13162928819656372,
"eval_rewards/rejected": -0.47367987036705017,
"eval_runtime": 42.4366,
"eval_samples_per_second": 86.906,
"eval_steps_per_second": 21.727
}