NAVER's SPLADE v3 is a sparse encoder based on BERT that produces sparse high-dimensional embeddings optimized for information retrieval. It can be used with the sentence-transformers library to encode text and compute similarities. The model excels at both lexical and semantic matching and is provided with full weights on Hugging Face.
In the Foundation models & chat space, Splade takes a focused approach. It focuses on creating effective sparse vector representations for accurate lexical and semantic search in documents. It is built as an open-source project for search engineers and NLP developers. Splade is open source under the Open Source license. It runs on the web, the command line, and API.
It is developed by NAVER, and the product first shipped in 2021. Development happens publicly on GitHub with 999 stars. Key capabilities include Sparse Embeddings, Semantic Search, and Text Retrieval.
Latest indexed changes and source events
naver/splade-v3 verified by the PulseGate indexer
