Math Reasoning on MATH (test) (EM)
52.7Exact Match (EM)VEROIC
Evaluation results
| Method | Links | |
|---|---|---|
| VEROICInference Paths=LLaMA-3.1-8B -> LLaMA-3.1-70B2026.04 | 52.7 | |
| VEROICInference Paths=Qwen-2.5-7B -> Qwen-2.5-32B2026.04 | 52.3 | |
| RISK PREDICTORInference Paths=LLaMA-3.1-8B -> LLaMA-3.1-70B2026.04 | 52.1 | |
| RISK PREDICTORInference Paths=Qwen-2.5-7B -> Qwen-2.5-32B2026.04 | 51.8 | |
| CONFIDENCE-THRESHOLDInference Paths=LLaMA-3.1-8B -> LLaMA-3.1-70B2026.04 | 51.5 | |
| OUTPUT-TRIGGERInference Paths=LLaMA-3.1-8B -> LLaMA-3.1-70B2026.04 | 51.2 | |
| CONFIDENCE-THRESHOLDInference Paths=Qwen-2.5-7B -> Qwen-2.5-32B2026.04 | 50.9 | |
| OUTPUT-TRIGGERInference Paths=Qwen-2.5-7B -> Qwen-2.5-32B2026.04 | 50.6 | |
| PERIODICInference Paths=LLaMA-3.1-8B -> LLaMA-3.1-70B2026.04 | 49.8 | |
| RANDOMInference Paths=LLaMA-3.1-8B -> LLaMA-3.1-70B2026.04 | 49.4 | |
| PERIODICInference Paths=Qwen-2.5-7B -> Qwen-2.5-32B2026.04 | 48.9 | |
| RANDOMInference Paths=Qwen-2.5-7B -> Qwen-2.5-32B2026.04 | 48.5 | |
| FIXEDInference Paths=LLaMA-3.1-8B -> LLaMA-3.1-70B2026.04 | 45.8 | |
| FIXEDInference Paths=Qwen-2.5-7B -> Qwen-2.5-32B2026.04 | 44.8 |