Mathematical Reasoning on MATH 500 (EM, #Tool)
84.6EM (Accuracy)ParetoPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ParetoPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 84.6 | 1.2 | |
| ToRL-GRPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 82.2 | 2.2 | |
| OTC-GRPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 81 | 1.5 | |
| MO-GRPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 80.8 | 1.9 | |
| ParetoPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 80 | 0.9 | |
| Qwen2.5-Math-7B-Ins-TIRBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 78.8 | 1.6 | |
| ToRL-GRPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 77.8 | 2.1 | |
| Qwen2.5-Math-7B-InsBase Model=Qwen2.5-Math-7B, Tool Use=No2026.06 | 74.8 | — | |
| OTC-GRPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 74 | 1.3 | |
| Qwen2.5-Math-1.5B-Ins-TIRBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 73.8 | 1.3 | |
| MO-GRPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 71.2 | 2 | |
| Qwen2.5-Math-1.5B-InsBase Model=Qwen2.5-Math-1.5B, Tool Use=No2026.06 | 66 | — |