GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning
Agrawal, Lakshya A, Tan, Shangyin, Dilara Soylu, Noah Ziems, Khare, Rishi, Krista Opsahl-Ong, Arnav Singhvi, Herumb Shandilya e mais 9
De quem é
- University of California, Berkeleyafiliação declarada pelos autores, segundo o OpenAlex
Outros papers das mesmas organizações
- Lost in the Middle: How Language Models Use Long ContextsStanford University · 6.873 citações
- Structured information extraction from scientific text with large language modelsLawrence Berkeley National Laboratory · 1.409 citações
- FAST: Efficient Action Tokenization for Vision-Language-Action ModelsUniversity of California, Berkeley · 1.025 citações
- A Review of Safe Reinforcement Learning: Methods, Theories, and ApplicationsTongji University · 903 citações
- Union through UNITY: Cosmology with 2000 SNe Using a Unified Bayesian FrameworkLawrence Berkeley National Laboratory · 665 citações
- Evaluating Real-World Robot Manipulation Policies in SimulationUniversity of California, Berkeley · 624 citações