Mathematical Reasoning on BIGMATH (Competition Subsets and Pass@8)
49.95MATH500 ScoreRLVR
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| RLVRBackbone=LLAMA-3.1-8B-INSTRUCT2025.10 | 49.95 | 5 | 3.33 | 26.25 | 18.13 | 1.25 | 17.32 | 37.31 | |
| RLERBackbone=LLAMA-3.1-8B-INSTRUCT, k=2, Gk=82025.10 | 48.35 | 4.17 | 2.92 | 23.75 | 16.67 | 1.25 | 16.19 | 35.07 | |
| INTUITORBackbone=LLAMA-3.1-8B-INSTRUCT, Category=RLIR2025.10 | 46.25 | 3.33 | 2.5 | 21.56 | 12.78 | 0.42 | 14.47 | 31.72 | |
| pre-RLBackbone=LLAMA-3.1-8B-INSTRUCT2025.10 | 45.75 | 3.75 | 0 | 19.38 | 12.22 | 0 | 13.52 | 30.2 | |
| SCBackbone=LLAMA-3.1-8B-INSTRUCT, Category=RLIR2025.10 | 42.65 | 2.5 | 2.08 | 24.37 | 12.78 | 0 | 14.06 | 30.85 | |
| JudgeBackbone=LLAMA-3.1-8B-INSTRUCT, Category=RLIR2025.10 | 37.22 | 0.42 | 1.25 | 15.62 | 10.83 | 0 | 10.89 | 25.39 |