Computer Science > Software Engineering
[Submitted on 18 Jul 2024 (this version), latest version 19 Jul 2024 (v2)]
Title:AI-Assisted SQL Authoring at Industry Scale
View PDF HTML (experimental)Abstract:SqlCompose is a tool that uses generative AI to assist with data analytics tasks, specifically SQL queries. It addresses the challenges of SQL being declarative, having formal table schemas, and often being written in a non-linear manner. The authors develop an internal SQL benchmark to test the performance of the Public Llama model and find that it performs well, with a BLEU score of 53% for single-line predictions and 24% for multi-line predictions. They then fine-tune the Llama model on their internal data and database schemas, resulting in a substantial improvement in performance. They also develop a fill-in-the-middle model, SqlComposeFIM, which is aware of the context before and after the line(s) that need to be completed, and this model outperforms the other two models by 35 percentage points. Additionally, they measure how often the models get the correct table names and find that SqlComposeFIM is able to do this 75% of the time, a major improvement over the other two models. The authors also roll out SqlComposeFIM at Meta and receive positive feedback from users, including completing tedious or repetitive SQL clauses, suggesting boilerplate coding, and help in eliminating the need to remember difficult SQL syntax. However, some users report table and column name hallucinations, which has been reduced with the release of SqlComposeFIM. Overall, the SqlCompose models consistently outperform public and internal LLMs despite their smaller size, providing early indications that smaller specialist models can outperform larger general purpose models.
Submission history
From: Rui Abreu [view email][v1] Thu, 18 Jul 2024 08:33:39 UTC (3,042 KB)
[v2] Fri, 19 Jul 2024 22:52:39 UTC (3,042 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
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.