Abstract: Prompt engineering/optimising manually is like writing assembly code; if you're a pro, you get the most amount of control and max performance with manually adapting Prompt, but DSPy provides the higher-level abstraction DSL for the functions the LLM-based system is supposed to carry out before getting onto the bare metal (the LLM). Get smaller, cleaner code with DSPy. (And auto-optimization as a bonus)
talk-data.com
E
Speaker
Eito Miyamura
1
talks
Co-Founder
EdisonWatch
Co-Founder - EdisonWatch
Bio from: Agentic AI + Evals at Databricks
Filter by Event / Source
Talks & appearances
1 activities · Newest first