List your company
ResearchFormerly WizWand

Step-level reasoning verification on Trips

82.5PR-AUCQwen2.5-Math-7B-PRM800k
39.13250.39161.6572.909Nov 9, 2025
Updated 6mo ago

Evaluation results

MethodLinks
Qwen2.5-Math-7B-PRM800k# Sample=265K2025.1182.5
Qwen2.5-Math-7B# Sample=860K2025.1179.1
ReProbe, Attn+Logit, Qwen3-8B-anno# Sample=32K2025.1175.6
ReProbe, Hidden States, GPT-OSS-anno# Sample=10.8K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.1175.2
Math-Shepherd-PRM-7B# Sample=440K2025.1174.7
Universal-PRM-Qwen2.5-Math-7B# Sample=690K2025.1174.1
Universal-PRM-Qwen2.5-Math-7B# Sample=690K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.1173.7
Qwen2.5-Math-PRM-7B# Sample=860K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.1169.6
MaxProb# Sample=-2025.1161.8
MaxEntropy# Sample=-2025.1158.5
Qwen2.5-Math-7B-PRM800K# Sample=263K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.1157.3
RLHFlow-PRM-Deepseek-8B# Sample=253K2025.1155.8
Perplexity# Sample=-2025.1155.7
Random# Sample=-2025.1155.2
H4-Qwen2.5-PRM-1.5B-0.2# Sample=369K2025.1153.4
H4-Qwen2.5-PRM-1.5B-0.2# Sample=369K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.1147
RLHFlow-PRM-Mistral-8B# Sample=273K2025.1146.2
Skywork-PRM-1.5B# Sample=Unk, Verifying Model=Qwen3-32B, Thinking Mode=native2025.1142
Skywork-PRM-1.5B# Sample=Unk2025.1140.8