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

Step-level reasoning verification on Meetings

73.4PR-AUC

ReProbe, Hidden States, GPT-OSS-anno

Updated 1mo ago

Evaluation Results

Method	Links
ReProbe, Hidden States, GPT-OSS-anno 2025.11		73.4
Qwen2.5-Math-PRM-7B 2025.11		66
ReProbe, Attn+Logit, Qwen3-8B-anno 2025.11		64.6
Universal-PRM-Qwen2.5-Math-7B 2025.11		63.4
Qwen2.5-Math-7B-PRM800k 2025.11		59.9
Qwen2.5-Math-7B 2025.11		59.4
Qwen2.5-Math-7B-PRM800K 2025.11		59.1
Math-Shepherd-PRM-7B 2025.11		58.4
Skywork-PRM-1.5B 2025.11		56.5
Universal-PRM-Qwen2.5-Math-7B 2025.11		55.9
MaxProb 2025.11		54.8
MaxEntropy 2025.11		53.3
Perplexity 2025.11		50.8
H4-Qwen2.5-PRM-1.5B-0.2 2025.11		47.6
Skywork-PRM-1.5B 2025.11		46.7
Random 2025.11		46.3
RLHFlow-PRM-Deepseek-8B 2025.11		45.5
RLHFlow-PRM-Mistral-8B 2025.11		42.4
H4-Qwen2.5-PRM-1.5B-0.2 2025.11		32.5