Translator-call mode selection on PolyMath High
73.74Macro F1LUAR
Evaluation Results
| Method | Links | |
|---|---|---|
| LUARBackbone=Qwen3-4B2026.06 | 73.74 | |
| LUARBackbone=Qwen3-8B2026.06 | 69.74 | |
| BOUNDARY-SFTBackbone=Qwen3-4B2026.06 | 66.85 | |
| ST(qr)Backbone=Qwen3-8B2026.06 | 60.73 | |
| BOUNDARY-SFTBackbone=Qwen3-8B2026.06 | 60.54 | |
| ST(q)Backbone=Qwen3-8B2026.06 | 60.41 | |
| ST(qr)Backbone=Qwen3-4B2026.06 | 58.24 | |
| ST(q)Backbone=Qwen3-4B2026.06 | 53.58 | |
| SELF-ASSESSMENTBackbone=Qwen3-4B2026.06 | 52.93 | |
| SELF-ASSESSMENTBackbone=Qwen3-8B2026.06 | 48.6 | |
| NATIVE-TOOL-USEBackbone=Qwen3-4B2026.06 | 48.54 | |
| NATIVE-TOOL-USEBackbone=Qwen3-8B2026.06 | 45.6 |