All Tools
S
Vector DBFreeOpen Source
SCANN
Scalable nearest neighbor search library
Apache-2.0
ABOUT
Finding nearest neighbors in large high-dimensional vector datasets is computationally prohibitive with exhaustive search. ScaNN uses approximate nearest neighbor search with anisotropic vector quantization and search space pruning to significantly speed up maximum inner-product search and other distance computations while maintaining high accuracy, enabling real-time semantic search, recommendation, and retrieval at billion-scale.
INSTALL
pip install scannINTEGRATION GUIDE
1. Semantic text search and document retrieval
2. Image similarity search and visual recommendation
3. Embedding-based recommendation systems
4. Large-scale machine learning inference and retrieval
TAGS
vector-searchapproximate-nearest-neighborsimilarity-searchquantizationtensorflowmachine-learningembeddingsMIPSANN