Research map / Architectures and efficiency / Attention and transformers
Rethinking the Role of Token Retrieval in Multi-Vector Retrieval
Jinhyuk Lee, Zhuyun Dai, Sai Meher Karthik Duddu, Tao Lei, Iftekhar Naim, Ming-Wei Chang, Vincent Y Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train token retrieval with in-batch top-k token alignment rather than document-level sum-of-max alignment | T5-based multi-vector retrieval · BEIR (13 datasets) | nDCG@10 average: 50.8 -> 52.7 | Improved |
| score candidates from retrieved tokens and impute missing similarities rather than gathering and scoring all document tokens | XTR · MS MARCO-derived scoring estimate | FLOPs/query: 0.36 × 10^9 -> 0.09 × 10^6 | Improved |
| reduce token retrieval depth from k′=40,000 to k′=1,000 (ablation) | XTR-base · BEIR (13 datasets) | nDCG@10 average: 49.1 -> 47.9 | Hurt |
| reduce token retrieval depth from k′=40,000 to k′=1,000 (ablation) | XTR-base · MS MARCO | nDCG@10: 45.0 -> 43.2 | Hurt |
| train multilingual retrieval on English MS MARCO plus MIRACL rather than English MS MARCO alone | mXTR-xxl · MIRACL (18 languages) | nDCG@10 average: 63.5 -> 65.9 | Improved |
Closest papers by meaning
- MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encoding (NeurIPS 2024 · similarity 0.86)
- BTR: Binary Token Representations for Efficient Retrieval Augmented Language Models (ICLR 2024 · similarity 0.85)
- Benchmarking and Building Long-Context Retrieval Models with LoCo and M2-BERT (ICML 2024 · similarity 0.85)
- xRAG: Extreme Context Compression for Retrieval-augmented Generation with One Token (NeurIPS 2024 · similarity 0.85)
- Can Language Models Actually Retrieve In-Context? Drowning in Documents at Million Token Scale (NeurIPS 2026 · similarity 0.85)
- ColPali: Efficient Document Retrieval with Vision Language Models (ICLR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
