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

Step-level correctness assessment on ProofNet (test)

32.9PR-AUC

Universal-PRM-Qwen2.5-Math-7B

10.95616.65322.3528.047Nov 9, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.11
32.9
2025.11
31
2025.11
30.1
2025.11
28.1
2025.11
26.4
2025.11
26
2025.11
24.3
2025.11
19.4
2025.11
18.5
2025.11
17.6
2025.11
17.3
2025.11
17.2
2025.11
17
2025.11
16.8
2025.11
15.9
2025.11
15.5
2025.11
15.3
2025.11
14.7
2025.11
14.7
2025.11
14.7
2025.11
13.6
2025.11
11.8