https://community.databricks.com/t5/databricks-tv/how-to-improve-your-llm-quality/ba-p/81079 <P><IFRAME src="https://www.youtube.com/embed/UYXQeOgwDV8?si=S41B1Cn8m6yjVbry" width="560" height="315" frameborder="0" allowfullscreen="" title="YouTube video player" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin"></IFRAME></P> Tue, 30 Jul 2024 15:45:53 GMT lara_rachidi 2024-07-30T15:45:53Z https://community.databricks.com/t5/databricks-tv/how-to-improve-your-llm-quality/ba-p/81079 <P><SPAN>We deep dive into a blog post explaining how Databricks engineers leveraged an open source LLM to produce a large training dataset for AI-generated comments in Unity Catalog, cleaned up the dataset using heuristic filters, fine-tuned another smaller LLM to supersede the current model, and improved the automated evaluation framework. Link to blog: <A href="https://www.databricks.com/blog/how-we-improved-databricksiq-llm-quality-ai-generated-table-comments" target="_blank">https://www.databricks.com/blog/how-we-improved-databricksiq-llm-quality-ai-generated-table-comments</A></SPAN></P> Tue, 30 Jul 2024 15:45:53 GMT https://community.databricks.com/t5/databricks-tv/how-to-improve-your-llm-quality/ba-p/81079 lara_rachidi 2024-07-30T15:45:53Z