KURE-Sparse-base/nano land at 79.84/77.16 nDCG@10 on Korean retrieval, searchable via inverted index

tomaarsen · x · 2026-10-23

tomaarsen covers the sparse members of the KURE retrieval family: KURE-Sparse-base/nano learn weighted terms that can be searched with an inverted index.

• Two sizes: 593M / 149M parameters
• The team reports 79.84 / 77.16 nDCG@10 across nine Korean retrieval tasks (the chart shows Sparse-base ahead of mitra-650m's 78.24 among sparse models)
• Both load with SparseEncoder and trustremotecode=True

The sparse route complements KURE's dense, late-interaction, and reranker variants, fitting setups built on classic search-engine stacks.

Related event: Korea University open-sources KURE, a full retrieval family spanning all four Sentence Transformers model types(8 posts)→

Original post →

More from Models

• Why frontier AI labs' bug bounties are cheap: their models flood the supply — lxrjl · 2026-10-23
• 7 OpenAI updates this week that cut your API bill, from 12-hour caching to Decisions API — MrCashMahon · 2026-10-23
• 100 classic human-written texts stress-test Pangram AI detector: formatting alone triggers false positives — GolinoHudson · 2026-10-23
• IQ2_XXS quant + 64k context makes a 27B model spin forever instead of coding — mailto_devnull · 2026-10-23
• UK AISI: Mythos 5 ignored DO NOT ATTACK warnings, delivered malware in 60% of tests — Hesamation · 2026-10-23
• Decision models are shockingly sensitive to question order and random noise — antoine_chaffin · 2026-10-23

Models channel →