[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
fix some para
yichuan520030910320851f0f04c3c4
diff --git a/examples/wechat_history_reader_leann.py b/examples/wechat_history_reader_leann.pyindex 971ea74..d002174 100644--- a/examples/wechat_history_reader_leann.py+++ b/examples/wechat_history_reader_leann.py@@ -78,7 +78,7 @@ def create_leann_index_from_multiple_wechat_exports()# Create text splitter with 256 chunk size- text_splitter = SentenceSplitter(chunk_size=256, chunk_overlap=128)+ text_splitter = SentenceSplitter(chunk_size=192, chunk_overlap=64)# Convert Documents to text strings and chunk themall_texts = []diff --git a/packages/leann-core/src/leann/embedding_compute.py b/packages/leann-core/src/leann/embedding_compute.pyindex ce85055..14a21f8 100644--- a/packages/leann-core/src/leann/embedding_compute.py+++ b/packages/leann-core/src/leann/embedding_compute.py@@ -101,7 +101,7 @@ def compute_embeddings_sentence_transformers(if device == "mps":batch_size = 128 # MPS optimal batch size from benchmarkif model_name == "Qwen/Qwen3-Embedding-0.6B":- batch_size = 64+ batch_size = 32elif device == "cuda":batch_size = 256 # CUDA optimal batch size# Keep original batch_size for CPU