Information Retrieval on TREC DL Hard
37.5nDCG@10MUGI + ReFormeR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MUGI + ReFormeREnhancement=ReFormeR2026.04 | 37.5 | 25 | 81.4 | |
| Query2Doc (FS) + ReFormeRPrompting Strategy=Few-shot, Enhancement=ReFormeR2026.04 | 36.5 | 23.3 | 79.9 | |
| Query2Doc (CoT) + ReFormeRPrompting Strategy=Chain-of-Thought, Enhancement=ReFormeR2026.04 | 35.3 | 21.8 | 79.8 | |
| Query2Doc (ZS)Prompting Strategy=Zero-shot2026.04 | 34.8 | 21.9 | 78.3 | |
| MUGI2026.04 | 34.7 | 22.4 | 79.4 | |
| Query2Doc (ZS) + ReFormeRPrompting Strategy=Zero-shot, Enhancement=ReFormeR2026.04 | 34.2 | 22.7 | 75.6 | |
| ReFormeR2026.04 | 33.7 | 20.7 | 71.8 | |
| Query2Doc (FS)Prompting Strategy=Few-shot2026.04 | 33.4 | 21.5 | 78.2 | |
| Query2Doc (CoT)Prompting Strategy=Chain-of-Thought2026.04 | 33.2 | 21 | 75.1 | |
| GenQR2026.04 | 31.3 | 18.6 | 71.9 | |
| QA-Expand + ReFormeREnhancement=ReFormeR2026.04 | 30.2 | 20.9 | 76.2 | |
| BM252026.04 | 29 | 16.4 | 67.8 | |
| QA-Expand2026.04 | 28.4 | 18.9 | 72.5 | |
| BM25 + RocchioEnhancement=Rocchio2026.04 | 27.7 | 16.4 | 70.4 | |
| BM25 + RM3Enhancement=RM32026.04 | 26.4 | 15.4 | 69.9 | |
| GenQREnsemble2026.04 | 19.7 | 12.2 | 65.1 | |
| CSQEBackbone=GPT-4.12026.06 | 0.366 | — | — | |
| MUGIBackbone=GPT-4.12026.06 | 0.365 | — | — | |
| LameRBackbone=GPT-4.12026.06 | 0.355 | — | — | |
| Q2DBackbone=GPT-4.1, Evaluation Protocol=Zero-shot2026.06 | 0.35 | — | — | |
| Q2KBackbone=GPT-4.12026.06 | 0.345 | — | — | |
| STORMBackbone=Qwen3-8B, Max tokens=642026.06 | 0.342 | — | — | |
| QA-ExpaBackbone=GPT-4.12026.06 | 0.302 | — | — | |
| GenQRBackbone=GPT-4.12026.06 | 0.292 | — | — | |
| GenQREBackbone=GPT-4.12026.06 | 0.27 | — | — | |
| RM3Backbone=GPT-4.12026.06 | 0.251 | — | — |