Mathematical Reasoning on Olympiad (EM, #Tool)
52.2EM AccuracyParetoPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ParetoPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 52.2 | 0.9 | |
| ToRL-GRPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 49.9 | 2.8 | |
| ParetoPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 48.1 | 0.8 | |
| MO-GRPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 46.3 | 1.7 | |
| OTC-GRPOBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 46.1 | 1.2 | |
| Qwen2.5-Math-7B-Ins-TIRBase Model=Qwen2.5-Math-7B, Tool Use=Yes2026.06 | 45 | 1.8 | |
| ToRL-GRPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 44 | 2.7 | |
| OTC-GRPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 42.1 | 1.2 | |
| Qwen2.5-Math-1.5B-Ins-TIRBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 41.3 | 1.5 | |
| MO-GRPOBase Model=Qwen2.5-Math-1.5B, Tool Use=Yes2026.06 | 41.2 | 2 | |
| Qwen2.5-Math-7B-InsBase Model=Qwen2.5-Math-7B, Tool Use=No2026.06 | 32.4 | — | |
| Qwen2.5-Math-1.5B-InsBase Model=Qwen2.5-Math-1.5B, Tool Use=No2026.06 | 31 | — |