analiza aiqbench
This research addresses the efficiency bottleneck in Text-to-SQL systems by introducing dynamic LLM routing strategies that match query complexity with appropriately sized models. The approach successfully maintains high accuracy on complex queries while significantly reducing latency and computational costs for simpler tasks. By optimizing the trade-off between performance and resource expenditure, this method offers a practical pathway for scaling database interaction tools without incurring prohibitive expenses.
Towards Optimizing SQL Generation via LLM Routing
arXiv:2411.04319v1 Announce Type: cross Abstract: Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capable large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.
Abstract:Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capable large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.
Databases (cs.DB); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2411.04319 [cs.DB]
(or arXiv:2411.04319v1 [cs.DB] for this version)
https://doi.org/10.48550/arXiv.2411.04319 Focus to learn more arXiv-issued DOI via DataCite
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 · 208 cuvinte
Subiecte conexe
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · 17 sept. 2026
[AINews] DeepSeek v4.1-Flash: 763B-P8B-D16B novel causal Encoder–Decoder architecture with vision marks the Return of the Whale
Latent Space · 12 sept. 2026
[AINews] not much happened today
Latent Space · 18 sept. 2026
[AINews] not much happened today
Latent Space · 10 sept. 2026