analiza aiqbench
This research introduces Single-stage Sparse Retrieval (SSR), which replaces computationally expensive clustering methods like K-means with sparse coding via Sparse Autoencoders to address the scalability bottlenecks of multi-vector retrieval models. By projecting token embeddings into high-dimensional sparse representations, the approach eliminates the need for vector clustering, enabling the use of inverted indexing for precise, high-throughput retrieval. Experimental results on the BEIR benchmark demonstrate a significant trade-off resolution, achieving a 15x reduction in indexing time, halved retrieval latency, and improved performance over existing baselines.
No More K-means: Single-Stage Sparse Coding for Efficient Multi-Vector Retrieval
arXiv:2605.30120v3 Announce Type: replace-cross Abstract: Multi-vector retrieval (MVR) models, exemplified by ColBERT, have established new benchmarks in retrieval accuracy by preserving fine-grained token-level interactions. However, this granularity imposes prohibitive storage and retrieval efficiency bottlenecks: to manage the immense memory footprint and computational overhead of billion-scale token vectors, state-of-the-art systems are forced to rely on aggressive dimension reduction and complex clustering (e.g., K-means). This compromise introduces two critical limitations: excessive indexing latency of clustering large-scale corpora and semantic information loss inherent to compression. In this paper, we propose Single-stage Sparse Retrieval (SSR}, a paradigm shift that replaces expensive clustering with efficient sparse coding. Instead of compressing features into low-dimensional dense vectors, we utilize Sparse Autoencoder (SAE) to project token embeddings into a high-dimensional but highly sparse representation. This transformation enables us to bypass vector clustering entirely and leverage inverted indexing for precise, high-throughput retrieval. Extensive experiments on the BEIR benchmark demonstrate that SSR achieves a "trifecta" of improvements: it reduces indexing time by 15x compared to ColBERTv2, halves retrieval latency, and simultaneously improves retrieval performance over leading baselines.
Abstract:Multi-vector retrieval (MVR) models, exemplified by ColBERT, have established new benchmarks in retrieval accuracy by preserving fine-grained token-level interactions. However, this granularity imposes prohibitive storage and retrieval efficiency bottlenecks: to manage the immense memory footprint and computational overhead of billion-scale token vectors, state-of-the-art systems are forced to rely on aggressive dimension reduction and complex clustering (e.g., K-means). This compromise introduces two critical limitations: excessive indexing latency of clustering large-scale corpora and semantic information loss inherent to compression. In this paper, we propose Single-stage Sparse Retrieval (SSR}, a paradigm shift that replaces expensive clustering with efficient sparse coding. Instead of compressing features into low-dimensional dense vectors, we utilize Sparse Autoencoder (SAE) to project token embeddings into a high-dimensional but highly sparse representation. This transformation enables us to bypass vector clustering entirely and leverage inverted indexing for precise, high-throughput retrieval. Extensive experiments on the BEIR benchmark demonstrate that SSR achieves a "trifecta" of improvements: it reduces indexing time by 15x compared to ColBERTv2, halves retrieval latency, and simultaneously improves retrieval performance over leading baselines.
Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2605.30120 [cs.IR]
(or arXiv:2605.30120v3 [cs.IR] for this version)
Text integral sub licența deschisă a publicației · 189 cuvinte
Subiecte conexe
Frontier post-training recipe review with Finbarr Timbers
Interconnects · 16 iun. 2026
[AINews] not much happened today
Latent Space · 3 oct. 2026
[AINews] OpenAI reports Navier-Stokes singularity find in 88 hours using Astra-next, roughly 10,000 agents and 130B tokens (>$40M), a contender for second ever Millennium Prize awarded
Latent Space · 9 sept. 2026
Granite 4.2 LLMs: How They're Built
Hugging Face · 25 aug. 2026