Mathematical Reasoning on MATH-P
48.4Pass@1FPA
Evaluation Results
| Method | Links | |
|---|---|---|
| FPABackbone=Qwen3-4B-Base2025.09 | 48.4 | |
| FPABackbone=Qwen2.5-Math-7B2025.09 | 44.6 | |
| Off-RLBackbone=Qwen2.5-Math-7B2025.09 | 43.9 | |
| DPOBackbone=Qwen3-4B-Base2025.09 | 43.7 | |
| A*-POBackbone=Qwen3-4B-Base2025.09 | 42.6 | |
| A*-POBackbone=Qwen2.5-Math-7B2025.09 | 42.3 | |
| Off-RLBackbone=Qwen3-4B-Base2025.09 | 41.4 | |
| KTOBackbone=Qwen3-4B-Base2025.09 | 41.1 | |
| SFTBackbone=Qwen2.5-Math-7B2025.09 | 40.1 | |
| BaseBackbone=Qwen2.5-Math-7B2025.09 | 39.8 | |
| RPOBackbone=Qwen3-4B-Base2025.09 | 39 | |
| RPOBackbone=Qwen2.5-Math-7B2025.09 | 38.4 | |
| KTOBackbone=Qwen2.5-Math-7B2025.09 | 38.4 | |
| DPOBackbone=Qwen2.5-Math-7B2025.09 | 38.1 | |
| SFTBackbone=Qwen3-4B-Base2025.09 | 34.6 | |
| BaseBackbone=Qwen3-4B-Base2025.09 | 33.2 | |
| A*-POBackbone=Llama-3.2-3B-Instruct2025.09 | 21.6 | |
| RPOBackbone=Llama-3.2-3B-Instruct2025.09 | 21.4 | |
| DPOBackbone=Llama-3.2-3B-Instruct2025.09 | 21.3 | |
| KTOBackbone=Llama-3.2-3B-Instruct2025.09 | 21.2 | |
| Off-RLBackbone=Llama-3.2-3B-Instruct2025.09 | 21.1 | |
| BaseBackbone=Llama-3.2-3B-Instruct2025.09 | 20.9 | |
| FPABackbone=Llama-3.2-3B-Instruct2025.09 | 20.7 | |
| SFTBackbone=Llama-3.2-3B-Instruct2025.09 | 20.6 |