Rethinking the Role of Token Retrieval in Multi-Vector Retrieval

arXiv:2304.01982v3 Announce Type: replace
Abstract: Multi-vector retrieval models such as ColBERT [Khattab and Zaharia, 2020] allow token-level interactions between queries and documents, and hence achieve state of the art on many information retrieval benchmarks. However, their non-linear scoring function cannot be scaled to millions of documents, necessitating a three-stage process for inference: retrieving initial candidates via token retrieval, accessing all token vectors, and scoring the initial candidate documents. The non-linear scoring function is applied over all token vectors of each …

abstract art arxiv benchmarks cs.cl cs.ir documents function however information interactions linear non-linear process queries retrieval role scoring stage state state of the art token type vector