[MLsys2026 Best Paper]: https://arxiv.org/abs/2506.08276. RAG on Everything with LEANN. Enjoy 97% storage savings while running a fast, accurate, and 100% private RAG application on your personal device.
Python aifaissgpt-osslangchainllama-indexllmlocalstorageoffline-firstollamaprivacypythonragretrieval-augmented-generationvector-databasevector-searchvectors
Commits on main
fix(core): skip empty/invalid chunks before embedding; guard OpenAI embeddings (#55)
14e84d9 Andy Lee committed 1 year ago
Docs/Core: Low-Resource Setups, SkyPilot Option, and No-Recompute (#45)
db3c63c Andy Lee committed 1 year ago
add chunk size in leann build& fix batch size in oai& docs
42c8370 yichuan520030910320 committed 1 year ago
feat(core,diskann): robust embedding server (no-hang) + DiskANN fast mode (graph partition) (#29)
fafdf8f Andy Lee committed 1 year ago
feat: support multiple input formats for --docs argument (#39)
eab1343 Yichuan Wang committed 1 year ago
fix: detect and report Ollama embedding dimension inconsistency (#37)
e8fca2c Andy Lee committed 1 year ago
Merge branch 'main' of https://github.com/yichuan-w/LEANN
ac36307 yichuan520030910320 committed 1 year ago
[Readme]update embedding model config according to reddit feedback
9801aa5 yichuan520030910320 committed 1 year ago
Add Ollama embedding support to enable local embedding models (#22)
3ff5aac Andy Lee committed 1 year ago