arXiv paper cuts learned sparse retrieval index memory with medoids and 4-bit quantization
_reachsumit · x · 2026-10-09
A new arXiv paper by Nardini et al. substantially shrinks learned sparse retrieval index memory without sacrificing efficiency:
- Inverted index: building on SEISMIC, replaces costly per-block summaries with medoids (real documents elected as block representatives), collapsing per-block metadata to a single doc ID.
- Forward index: vocabulary reordering plus SIMD-friendly DOTPACKING8 bit-packing for Δ-gaps and per-component 4-bit quantization; the compression plugs into any forward-index-based system like KANNOLO.
- Introduces JUMPDOT, a blocked dot-product kernel for queries with few non-zero entries.
Evaluated on MS MARCO with three state-of-the-art sparse encoders, showing markedly improved speed-space trade-offs.
More from Research
- RFD4-Proteina's 'underrated' docs detail partial diffusion for protein redesign — AllThingsApx · 2026-10-23
- MPI-Sec PhD Thesis on Vulnerability Detection Benchmarks Includes Pre-LLM Classic "Top Score on the Wrong Exam" — mboehme_ · 2026-10-23
- LiveNerf's 29-day data finds no nerfing of Claude Opus 5.5 — TheOnlyVibemaster · 2026-10-23
- New paper makes 9 falsifiable predictions toward a major AI-enabled terror attack by 2027-2028 — StephenLCasper · 2026-10-23
- HumanoidPose recovers humanoid robots' full-body joint states from RGB video alone — cholz · 2026-10-23
- Sophont's Holon-S1 predicts genes and biomarkers from routine pathology slides — iScienceLuvr · 2026-10-23