Fast PEFT serving at scale is hard! At Databricks, I've been working on building an inference runtime that outperforms open source, and we talk about how we did it in this blog: databricks.com/blog/fast-peft…
(1/2)
Check out my latest article: The importance of programming in modern society, and the wide assortment of jobs available in this field. linkedin.com/pulse/importan… via @LinkedIn