List your company
ResearchFormerly WizWand

Reward Modeling on PPE-IFEval

0.75AccuracyGemini-2.5-Flash
0.44840.52670.6050.6833Feb 2, 2026
Updated 8mo ago

Evaluation results

MethodLinks
Gemini-2.5-Flash2026.020.75
Rubric-ARM-voting@5voting=52026.020.72
Rubric-RM-voting@5voting=52026.020.708
Rubric-ARM2026.020.708
Rubric-RM2026.020.67
RM-R1-32B (DeepSeek-Dist)Backbone=DeepSeek-Dist, Size=32B2026.020.632
RM-R1-14B (DeepSeek-Dist)Backbone=DeepSeek-Dist, Size=14B2026.020.612
API (Rubric+Judge)Rubric Model=GPT-4.1-Mini, Judge Model=Gemini-2.5-Flash Lite2026.020.61
RM-R1-32B (Qwen-2.5-Inst)Backbone=Qwen-2.5-Inst, Size=32B2026.020.604
RRM-32BSize=32B2026.020.602
API (direct Judge)Judge Model=Gemini-2.5-Flash Lite2026.020.592
RM-R1-14B (Qwen-2.5-Inst)Backbone=Qwen-2.5-Inst, Size=14B2026.020.59
Claude-3.5-Sonnet2026.020.58
RM-R1-7B (Qwen-2.5-Inst)Backbone=Qwen-2.5-Inst, Size=7B2026.020.552
Qwen-3-8B (Rubric+Judge)Backbone=Qwen-3-8B2026.020.538
RM-R1-7B (DeepSeek-Dist)Backbone=DeepSeek-Dist, Size=7B2026.020.51
RRM-7BSize=7B2026.020.51
JudgeLRM-7BSize=7B2026.020.46