Linkup Open-Sources SPARSEUP: <150M Sparse Retrieval Model Scoring 56+ on BEIR-13
antoine_chaffin · x · 2026-09-17
Linkup has released SPARSEUP, its first model and the missing sparse companion to DenseOn and LateOn. Same backbone and data, under 150M parameters, it scores 56+ on BEIR-13. The model is open-sourced under Apache 2.0, with blog post and weights available. Commenters are optimistic about combining sparse representations with late interaction.
Related event: Linkup Open-Sources SparseUp, Top Sparse Retriever Under 150M on BEIR-13(10 posts)→
More from Models
- OpenAI reportedly close to solving Hodge Conjecture, another $1M Millennium Prize problem — Distinct-Question-16 · 2026-09-18
- Pharma founder: Gemini's native Google Workspace access beats handing data to Claude or ChatGPT — pdamodaran · 2026-09-18
- Pichai confirms Gemini 4 is in training, with coding and agentic coding as top priorities — pdamodaran · 2026-09-18
- Sentdex Benchmarks LLMs on Halite: DSV4.1 Flash vs GLM 5.3 Flash in NVFP4 — Sentdex · 2026-09-18
- Six real-world uses for TypeSafe's Jev judgment model, 4x faster than Gemini in evals — HamelHusain · 2026-09-18
- Google DeepMind to host llama.cpp x Gemma side event at Nerdearla — mervenoyann · 2026-09-18