GEPA with Lakshya A. Agrawal - Weaviate Podcast #127!

Lakshya A. Agrawal is a Ph.D. student at U.C. Berkeley! Lakshya has lead the research behind GEPA, one of the newest innovations in DSPy and the use of Large Language Models as Optimizers! GEPA makes three key innovations on how exactly we use LLMs to propose prompts for LLMs, (1) Pareto-Optimal Candidate Selection, (2) Reflective Prompt Mutation, and (3) System-Aware Merging. The podcast discusses all of these details further, as well as topics such as Test-Time Training and the LangProBe benchmarks used in the paper! I hope you find the podcast useful! Links: GEPA: Reflective Prompt Evolution can Outperform Reinforcement Learning: https://arxiv.org/pdf/2507.19457 Lakshya A. Agrawal: https://lakshyaaagrawal.github.io/ DSPy 3.0: https://github.com/stanfordnlp/dspy/releases/tag/3.0.0 Chapters: 0:00 Welcome Lakshya! 4:50 Natural Language Rewards 11:38 Evolution of DSPy 14:55 Pareto-Optimal Candidate Selection 19:58 Domain Specific Knowledge 23:05 Exploration and Diversity 28:35 Test-Time Training 40:10 LangProBe Benchmark 48:05 Prompt Optimization and RL 52:20 Database Tuning with GEPA 1:00:00 Exciting Directions for AI