STEM Reasoning on TheoremQA
36.8AccuracyBingo-A
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Bingo-ABase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 36.8 | 1,648 | 32.9 | |
| Bingo-EBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 36.7 | 1,584 | 33 | |
| DASTBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 35.2 | 2,325 | 29.8 | |
| DemystifyingBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 35.1 | 1,976 | 30.6 | |
| Effi. ReasoningBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 34.8 | 3,560 | 26.2 | |
| kimi-k1.5Base Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 34.4 | 2,136 | 29.6 | |
| Vanilla PPOBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 32.3 | 4,146 | 22.7 | |
| O1-PrunerBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 28.6 | 524 | 27.6 |