Mathematical Reasoning on Mathematical Reasoning Out-of-distribution
76.5MMLU-ProTRACE
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| TRACEModel=Qwen3-14B2026.06 | 76.5 | 96.5 | 60.4 | 77.8 | |
| PCLModel=Qwen3-14B2026.06 | 76.1 | 96.2 | 59.7 | 77.3 | |
| TreePOModel=Qwen3-14B2026.06 | 76 | 96.4 | 59.8 | 77.4 | |
| GRPOModel=Qwen3-14B2026.06 | 75.9 | 96.1 | 59.4 | 77.1 | |
| ReActModel=Qwen3-14B2026.06 | 75.1 | 96.2 | 59.2 | 76.8 | |
| TRACEModel=Qwen3-8B2026.06 | 73.4 | 95.7 | 56.9 | 75.3 | |
| TreePOModel=Qwen3-8B2026.06 | 73 | 95.2 | 56.4 | 74.9 | |
| PCLModel=Qwen3-8B2026.06 | 72.9 | 95.4 | 56.2 | 74.8 | |
| GRPOModel=Qwen3-8B2026.06 | 72.7 | 95.3 | 55.9 | 74.6 | |
| ReActModel=Qwen3-8B2026.06 | 68.5 | 95.3 | 52.4 | 72.1 |