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

Reward Modeling on SCAN HPD

82.88Accuracy

gemini-2.5-pro

Updated 1mo ago

Evaluation Results

Method	Links
gemini-2.5-pro 2026.01		82.88
deepseek-r1-0528 2026.01		82.27
gpt-5-2025-08-07 2026.01		81.47
o3-2025-04-16 2026.01		81.47
qwen-plus-latest 2026.01		80.48
RM-NLHF-Qwen-32B 2026.01		78.88
qwen3-max 2026.01		78.65
RM-R1-Qwen-32B 2026.01		77.95
INF-ORM-Llama3.1-70B 2026.01		76.84
deepseek-v3 2026.01		76.36
RRM-Qwen-32B 2026.01		76.04
gpt-4o-latest 2026.01		75.72
claude-3-7-sonnet-20250219 2026.01		75.24
R1-Distill-Qwen-32B 2026.01		74.92
Skywork-Reward-Llama-3.1-8B-v0.2 2026.01		72.05
URM-LLaMa-3.1-8B 2026.01		70.13
RM-NLHF-Qwen-7B 2026.01		68.22
RRM-Qwen-7B 2026.01		66.45
ArmoRM-Llama3-8B-v0.1 2026.01		66.28
R1-Distill-Llama-8B 2026.01		65.92
RM-R1-Qwen-7B 2026.01		64.38
DeepSeek-R1-Distill-Qwen-7B 2026.01		60.15