Mathematical Reasoning on MATH500 (Acc, Len, L-Acc)
88.8AccuracyBingo-A
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Bingo-ABackbone=DeepSeek-R1-Distill-Qwen-7B2025.06 | 88.8 | 1,400 | 80.9 | |
| PPOBackbone=DeepSeek-R1-Distill-Qwen-7B2025.06 | 88.4 | 1,536 | 79.7 | |
| Bingo-EBackbone=DeepSeek-R1-Distill-Qwen-7B2025.06 | 87.2 | 1,252 | 80.3 | |
| BaseBackbone=DeepSeek-R1-Distill-Qwen-7B2025.06 | 82.8 | 3,033 | 65.7 | |
| Bingo-ABackbone=Qwen2.5-Math-7B-Instruct2025.06 | 82.6 | 656 | 73 | |
| Bingo-ABackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 82.2 | 894 | 77.6 | |
| PPOBackbone=Qwen2.5-Math-7B-Instruct2025.06 | 82 | 670 | 72.3 | |
| Bingo-EBackbone=Qwen2.5-Math-7B-Instruct2025.06 | 81.6 | 559 | 73.6 | |
| PPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 81.4 | 2,771 | 66.2 | |
| BaseBackbone=Qwen2.5-Math-7B-Instruct2025.06 | 80.8 | 727 | 70.3 | |
| Bingo-EBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 80.6 | 779 | 76.7 | |
| BaseBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.06 | 63.2 | 3,913 | 45.7 |