Multilingual Mathematical Reasoning on MMATH In-Domain Languages (test)
26.3Accuracy (Ar)Qwen2.5-7B-Instruct + LANG
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Qwen2.5-7B-Instruct + LANGBackbone=Qwen2.5-7B-Instruct, Method=LANG2026.05 | 26.3 | 28.1 | — | 28.5 | 31.1 | 28 | 22.5 | 32.1 | 28.1 | |
| Qwen2.5-7B-Instruct + mGRPOBackbone=Qwen2.5-7B-Instruct, Method=mGRPO2026.05 | 24.7 | 25.1 | — | 20.5 | 27.8 | 26.5 | 20.1 | 31 | 25.1 | |
| Qwen2.5-7B-Instruct + LC-GRPOBackbone=Qwen2.5-7B-Instruct, Method=LC-GRPO2026.05 | 24.3 | 26 | — | 23.6 | 28.2 | 25 | 24.7 | 30.2 | 26 | |
| Qwen2.5-7B-Instruct + M-ThinkerBackbone=Qwen2.5-7B-Instruct, Method=M-Thinker2026.05 | 21.4 | 23.9 | — | 20.7 | 22.7 | 24.5 | 25.7 | 28.6 | 23.9 | |
| Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct, Metric Type=Accuracy2026.05 | 20.3 | — | 23.9 | 21.5 | 24.1 | 25.4 | 23.8 | 28.2 | 23.9 | |
| Qwen2.5-7B-Instruct + DITBackbone=Qwen2.5-7B-Instruct, Method=DIT2026.05 | 18.3 | 20.4 | — | 16.8 | 19.2 | 19.8 | 19.4 | 29 | 20.4 | |
| Qwen2.5-7B-Instruct + LCPBackbone=Qwen2.5-7B-Instruct, Method=LCP2026.05 | 18 | 20.8 | — | 16.1 | 23.2 | 18.6 | 19.5 | 29.6 | 20.8 | |
| Qwen2.5-7B-Instruct + QRTBackbone=Qwen2.5-7B-Instruct, Method=QRT2026.05 | 17.7 | 20.8 | — | 16.1 | 20.3 | 19.8 | 23.1 | 27.7 | 20.8 | |
| Qwen2.5-3B-Instruct + mGRPOBackbone=Qwen2.5-3B-Instruct, Method=mGRPO2026.05 | 15 | 15.8 | — | 15.3 | 18.2 | 12.2 | 14.1 | 19.9 | 15.8 | |
| Qwen2.5-3B-Instruct + LANGBackbone=Qwen2.5-3B-Instruct, Method=LANG2026.05 | 14.5 | 18.3 | — | 19.5 | 20.7 | 15.4 | 16.9 | 23.1 | 18.3 | |
| Qwen2.5-3B-Instruct + LC-GRPOBackbone=Qwen2.5-3B-Instruct, Method=LC-GRPO2026.05 | 14.3 | 17.3 | — | 17.7 | 19.4 | 14.6 | 14.9 | 22.9 | 17.3 | |
| Qwen2.5-3B-Instruct + LCPBackbone=Qwen2.5-3B-Instruct, Method=LCP2026.05 | 13.7 | 15.8 | — | 16.6 | 16.4 | 12.4 | 14.3 | 21.3 | 15.8 | |
| Qwen2.5-3B-InstructBackbone=Qwen2.5-3B-Instruct, Metric Type=Accuracy2026.05 | 13.1 | — | 15.4 | 11 | 16.7 | 13.7 | 15.1 | 23 | 15.4 | |
| Qwen2.5-7B-Instruct + M-SFTBackbone=Qwen2.5-7B-Instruct, Method=M-SFT2026.05 | 12.9 | 18.7 | — | 15.2 | 12.6 | 19.8 | 19.4 | 32.4 | 18.7 | |
| Qwen2.5-3B-Instruct + M-SFTBackbone=Qwen2.5-3B-Instruct, Method=M-SFT2026.05 | 12.8 | 16.8 | — | 18.9 | 18.5 | 13.1 | 15.3 | 22.4 | 16.8 | |
| Qwen2.5-3B-Instruct + M-ThinkerBackbone=Qwen2.5-3B-Instruct, Method=M-Thinker2026.05 | 12.8 | 16.3 | — | 17.5 | 19.6 | 11.8 | 13.8 | 22.4 | 16.3 | |
| Qwen2.5-3B-Instruct + DITBackbone=Qwen2.5-3B-Instruct, Method=DIT2026.05 | 12.5 | 16.4 | — | 17.6 | 17 | 12.3 | 15.7 | 23.2 | 16.4 | |
| Qwen2.5-3B-Instruct + QRTBackbone=Qwen2.5-3B-Instruct, Method=QRT2026.05 | 12.4 | 16.7 | — | 18.5 | 17.2 | 13.2 | 16.6 | 22.3 | 16.7 | |
| Qwen2.5-3B-InstructBackbone=Qwen2.5-3B-Instruct, Metric Type=LC&Acc2026.05 | 3 | 7.2 | — | 0.9 | 0.7 | 2.1 | 13.7 | 23 | 7.2 | |
| Qwen2.5-3B-Instruct + Vanilla GRPOBackbone=Qwen2.5-3B-Instruct, Method=Vanilla GRPO2026.05 | 2.5 | 10.4 | — | 4.2 | 9 | 6.9 | 17.2 | 22.4 | 10.4 | |
| Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct, Metric Type=LC&Acc2026.05 | 0.3 | 9 | — | 0.5 | 0.2 | 3.6 | 21 | 28.2 | 9 | |
| Qwen2.5-7B-Instruct + Vanilla GRPOBackbone=Qwen2.5-7B-Instruct, Method=Vanilla GRPO2026.05 | 0 | 5.4 | — | 0.7 | 0 | 0.1 | 0.7 | 30.9 | 5.4 |