[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
ci: make arch smoke test hermetic
Abi737395cf21ba
diff --git a/.github/workflows/build-reusable.yml b/.github/workflows/build-reusable.ymlindex fe1c655..e3c333b 100644--- a/.github/workflows/build-reusable.yml+++ b/.github/workflows/build-reusable.yml@@ -659,14 +659,36 @@ jobs:run: |source .venv/bin/activate || source .venv/Scripts/activatepython - <<'PY'+ import numpy as npimport leannimport leann_backend_hnsw as himport leann_backend_diskann as dimport leann_backend_ivf as ivffrom leann import LeannBuilder, LeannSearcher- b = LeannBuilder(backend_name="hnsw")++ b = LeannBuilder(+ backend_name="hnsw",+ dimensions=2,+ is_compact=False,+ is_recompute=False,+ )b.add_text("hello arch")- b.build_index("arch_demo.leann")- s = LeannSearcher("arch_demo.leann")- print("search:", s.search("hello", top_k=1))+ b.build_index_from_arrays(+ "arch_demo.leann",+ ["0"],+ np.asarray([[1.0, 0.0]], dtype=np.float32),+ )++ with LeannSearcher(+ "arch_demo.leann",+ recompute_embeddings=False,+ enable_warmup=False,+ ) as s:+ s.backend_impl.compute_query_embedding = lambda *args, **kwargs: np.asarray(+ [[1.0, 0.0]], dtype=np.float32+ )+ result = s.search("hello", top_k=1)++ assert result and result[0].text == "hello arch"+ print("arch smoke ok")PY