analiza aiqbench
ColBERTSaR addresses the scalability bottleneck of ColBERT-based retrieval by converting heavy embedding indexes into compact inverted indexes via product quantization. This approach reduces index storage by 50-70% compared to one-bit PLAID while maintaining retrieval effectiveness and eliminating costly decompression stages. The technique offers a practical path to deploying dense neural retrieval architectures in resource-constrained environments without sacrificing performance.
ColBERTSaR: Sparsified ColBERT Index via Product Quantization
arXiv:2606.05568v1 Announce Type: cross Abstract: While ColBERT is an effective neural retrieval architecture, it requires a heavy index structure to support candidate set retrieval based on approximated token embeddings, gathering and decompressing document token embeddings, and applying the MaxSim operation. Indexes in PLAID and similar ColBERT implementations require five to ten times the disk storage of the original raw text, which limits their scalability. Furthermore, prior work has identified that the gathering and decompression stages are the primary inefficiencies at query time. Limiting the number of document tokens that must be gathered by thresholding and score approximation does not eliminate the need for the entire index to support ad hoc queries. In this work, we propose an embedding quantization approach that turns a ColBERT index into a true inverted index. We show that, theoretically, ColBERT with embedding quantization is equivalent to learned-sparse retrieval except for the scoring mechanism. Empirically, we demonstrate that our index is 50-70% smaller than a one-bit PLAID index while retaining retrieval effectiveness.
Abstract:While ColBERT is an effective neural retrieval architecture, it requires a heavy index structure to support candidate set retrieval based on approximated token embeddings, gathering and decompressing document token embeddings, and applying the MaxSim operation. Indexes in PLAID and similar ColBERT implementations require five to ten times the disk storage of the original raw text, which limits their scalability. Furthermore, prior work has identified that the gathering and decompression stages are the primary inefficiencies at query time. Limiting the number of document tokens that must be gathered by thresholding and score approximation does not eliminate the need for the entire index to support ad hoc queries. In this work, we propose an embedding quantization approach that turns a ColBERT index into a true inverted index. We show that, theoretically, ColBERT with embedding quantization is equivalent to learned-sparse retrieval except for the scoring mechanism. Empirically, we demonstrate that our index is 50-70% smaller than a one-bit PLAID index while retaining retrieval effectiveness.
Information Retrieval (cs.IR); Computation and Language (cs.CL)
arXiv:2606.05568 [cs.IR]
(or arXiv:2606.05568v1 [cs.IR] for this version)
https://doi.org/10.48550/arXiv.2606.05568 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
https://doi.org/10.1145/3805712.3809920 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 269 cuvinte