Cross-lingual Question Answering on MKQA French
59.67fEMCroSearch-R1
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CroSearch-R1Backbone=Qwen2.5-7B-Instruct, Paradigm=Post-training2026.04 | 59.67 | 74.04 | |
| Search-R1Backbone=Qwen2.5-7B-Instruct, Paradigm=Post-training2026.04 | 57.91 | 70.36 | |
| CroSearch-R1Backbone=Qwen2.5-3B-Instruct, Paradigm=Post-training2026.04 | 54 | 72.39 | |
| SFTBackbone=Qwen2.5-7B-Instruct, Paradigm=Post-training2026.04 | 51.92 | 59.53 | |
| SFTBackbone=Qwen2.5-3B-Instruct, Paradigm=Post-training2026.04 | 48 | 56.56 | |
| Search-R1Backbone=Qwen2.5-3B-Instruct, Paradigm=Post-training2026.04 | 44.73 | 61.35 | |
| RAGBackbone=Qwen2.5-7B-Instruct, Paradigm=Training-free2026.04 | 44.64 | 53.53 | |
| IRCoTBackbone=Qwen2.5-7B-Instruct, Paradigm=Training-free2026.04 | 41.52 | 51.44 | |
| Search-o1Backbone=Qwen2.5-7B-Instruct, Paradigm=Training-free2026.04 | 41.48 | 53.57 | |
| RAGBackbone=Qwen2.5-3B-Instruct, Paradigm=Training-free2026.04 | 40.72 | 50.2 | |
| IRCoTBackbone=Qwen2.5-3B-Instruct, Paradigm=Training-free2026.04 | 39.88 | 49.29 | |
| Search-o1Backbone=Qwen2.5-3B-Instruct, Paradigm=Training-free2026.04 | 38.62 | 51.05 | |
| Direct InferenceBackbone=Qwen2.5-7B-Instruct, Paradigm=Training-free2026.04 | 35.79 | 46.86 | |
| Direct InferenceBackbone=Qwen2.5-3B-Instruct, Paradigm=Training-free2026.04 | 29.21 | 40.04 |