Research map / Architectures and efficiency / RNNs and normalization
Model-enhanced Vector Index
Hailin Zhang, Yujing Wang, Qi Chen, Ruiheng Chang, Ting Zhang, Ziming Miao, Yingyan Hou, Yang Ding, Xupeng Miao, Haonan Wang, Bochen Pang, Yuefeng Zhan and 7 more
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 |
|---|---|---|---|
| ensemble MEVI Top-1000 with AR2 (HNSW) | AR2 (HNSW) · MSMARCO Passage Dev | MRR@10: 35.54 -> 39.16 | Improved |
| ensemble MEVI Top-1000 with T5-ANCE (HNSW) | T5-ANCE (HNSW) · MSMARCO Passage Dev | MRR@10: 33.21 -> 35.83 | Improved |
| ensemble MEVI Top-1000 with AR2 (HNSW) | AR2 (HNSW) · MSMARCO Passage Dev | Recall@1000: 87.11 -> 97.65 | Improved |
| ensemble MEVI Top-1000 with AR2 (HNSW) | AR2 (HNSW) · Natural Questions Test | Recall@5: 70.89 -> 75.93 | Improved |
| ensemble MEVI Top-1000 with AR2 (HNSW) | AR2 (HNSW) · Natural Questions Test | Recall@100: 83.02 -> 88.98 | Improved |
| retrieve more clusters (Top-10 to Top-1000) | MEVI · MSMARCO Passage Dev | Recall@1000: 66.82 -> 95.17 | Improved |
| residual quantization instead of hierarchical KMeans (ablation) | MEVI · MSMARCO Passage Dev | Recall@1000 (Top-1000): 94.17 -> 95.17 | Improved |
| residual quantization instead of hierarchical KMeans, Top-10 clusters (ablation) | MEVI · MSMARCO Passage Dev | Recall@50: 65.09 -> 63.25 | Hurt |
| RQ(4x5) vs RQ(4x4) codebook (ablation) | MEVI · MSMARCO Passage Dev | MRR@10: 29.15 -> 33.36 | Improved |
| RQ(5x5) vs RQ(4x5) codebook (ablation) | MEVI · MSMARCO Passage Dev | MRR@10: 33.36 -> 34.12 | Improved |
| drop 10% of documents in training (dynamic update), MEVI Top-1000 (ablation) | MEVI · MSMARCO Passage Dev | MRR@10: 35.76 -> 35.38 | Hurt |
| drop 10% of documents in training, T5-ANCE (ablation) | T5-ANCE (HNSW) · MSMARCO Passage Dev | MRR@10: 33.21 -> 29.34 | Hurt |
Closest papers by meaning
- MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encoding (NeurIPS 2024 · similarity 0.88)
- NUDGE: Lightweight Non-Parametric Fine-Tuning of Embeddings for Retrieval (ICLR 2025 · similarity 0.87)
- NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models (ICLR 2025 · similarity 0.87)
- Hierarchical Retrieval: The Geometry and a Pretrain-Finetune Recipe (NeurIPS 2025 · similarity 0.87)
- Your Embedding Model Is SMARTer Than You Think (NeurIPS 2026 · similarity 0.86)
- Semi-Parametric Retrieval via Binary Bag-of-Tokens Index (ICLR 2025 · similarity 0.86)
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.
