Step-level verification on MATH
67.6PR-AUCReProbe, Hidden States, GPT-OSS-anno
Evaluation results
| Method | Links | |
|---|---|---|
| ReProbe, Hidden States, GPT-OSS-anno# Sample=10.8K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.11 | 67.6 | |
| Qwen2.5-Math-PRM-7B# Sample=860K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.11 | 66.8 | |
| Universal-PRM-Qwen2.5-Math-7B# Sample=690K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.11 | 57.8 | |
| Qwen2.5-Math-7B-PRM800K# Sample=263K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.11 | 40.9 | |
| Skywork-PRM-1.5B# Sample=Unk, Verifying Model=Qwen3-32B, Thinking Mode=native2025.11 | 30.3 | |
| H4-Qwen2.5-PRM-1.5B-0.2# Sample=369K, Verifying Model=Qwen3-32B, Thinking Mode=native2025.11 | 13.3 |