Share your thoughts, 1 month free Claude Pro on usSee more

Reward Modeling on PPE-IFEval

0.75Accuracy

Gemini-2.5-Flash

Updated 1mo ago

Evaluation Results

Method	Links
Gemini-2.5-Flash 2026.02		0.75
Rubric-ARM-voting@5 2026.02		0.72
Rubric-RM-voting@5 2026.02		0.708
Rubric-ARM 2026.02		0.708
Rubric-RM 2026.02		0.67
RM-R1-32B (DeepSeek-Dist) 2026.02		0.632
RM-R1-14B (DeepSeek-Dist) 2026.02		0.612
API (Rubric+Judge) 2026.02		0.61
RM-R1-32B (Qwen-2.5-Inst) 2026.02		0.604
RRM-32B 2026.02		0.602
API (direct Judge) 2026.02		0.592
RM-R1-14B (Qwen-2.5-Inst) 2026.02		0.59
Claude-3.5-Sonnet 2026.02		0.58
RM-R1-7B (Qwen-2.5-Inst) 2026.02		0.552
Qwen-3-8B (Rubric+Judge) 2026.02		0.538
RM-R1-7B (DeepSeek-Dist) 2026.02		0.51
RRM-7B 2026.02		0.51
JudgeLRM-7B 2026.02		0.46