Question Answering on 2WikiQA (test)
57.9EMSearch-R1
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Search-R1LLM Generator=Qwen2.5-7B-Instruct2026.02 | 57.9 | 64.9 | |
| MultiCube-RAGLLM Generator=Qwen2.5-7B-Instruct2026.02 | 56.3 | 63.2 | |
| s3LLM Generator=Qwen2.5-7B-Instruct2026.02 | 47.7 | 55.2 | |
| TrustMarginBackbone=LLaMA3.1-8B2026.06 | 32.7 | 39.48 | |
| BM25-RAGBackbone=LLaMA3.1-8B2026.06 | 27.3 | 34.3 | |
| DTR-RAGBackbone=LLaMA3.1-8B2026.06 | 27.3 | 33.67 | |
| IRCoTBackbone=LLaMA3.1-8B2026.06 | 26.2 | 33.12 | |
| DirectBackbone=LLaMA3.1-8B2026.06 | 24.9 | 30.98 | |
| FLAREBackbone=LLaMA3.1-8B2026.06 | 24.9 | 31.61 | |
| TrustMarginBackbone=LLaMA3.2-3B2026.06 | 22.9 | 30.29 | |
| CLeHe-RAGBackbone=LLaMA3.1-8B2026.06 | 22.9 | 27.13 | |
| DTR-RAGBackbone=LLaMA3.2-3B2026.06 | 21.5 | 27.65 | |
| IRCoTBackbone=LLaMA3.2-3B2026.06 | 20.7 | 27.45 | |
| CLeHe-RAGBackbone=LLaMA3.2-1B2026.06 | 20.4 | 26.14 | |
| DTR-RAGBackbone=LLaMA3.2-1B2026.06 | 20.4 | 26.26 | |
| TrustMarginBackbone=LLaMA3.2-1B2026.06 | 20.4 | 27.29 | |
| IRCoTBackbone=LLaMA3.2-1B2026.06 | 20.1 | 26.32 | |
| BM25-RAGBackbone=LLaMA3.2-1B2026.06 | 19.4 | 26.4 | |
| BM25-RAGBackbone=LLaMA3.2-3B2026.06 | 18.8 | 25.33 | |
| DirectBackbone=LLaMA3.2-3B2026.06 | 17.9 | 24.27 | |
| FLAREBackbone=LLaMA3.2-3B2026.06 | 17.9 | 24.04 | |
| FLAREBackbone=LLaMA3.2-1B2026.06 | 16.6 | 22.77 | |
| CLeHe-RAGBackbone=LLaMA3.2-3B2026.06 | 16.3 | 23.13 | |
| DirectBackbone=LLaMA3.2-1B2026.06 | 16.1 | 22.17 |