#artificial intelligence
May 2026
No More K-means: Single-Stage Sparse Coding for Efficient Multi-Vector Retrieval
This paper proposes Single-stage Sparse Retrieval (SSR), a paradigm shift that replaces expensive clustering with efficient sparse coding, and utilizes Sparse Autoencoder (SAE) to project token embeddings into a high-dimensional but highly sparse representation.
Li-Xuan Guo, Yi-Fei Wang, Tiansheng Wen et al.
· arXiv.org · 0 citations