GEPA Reflective Prompt Evolution Can Outperform Reinforcement Learning
---
created:
- 2025-08-02T12:02
modified: 2025-08-02 12:04
tags:
- nlp
- llm
- large-language-model
- optimisation
- optim
- RL
- reinforcement-learning
- fine-tuning
- gpt
type:
- note
status:
- in-progress
---
## References
* [original paper (arxiv)](https://arxiv.org/abs/2507.19457)
## Related
* [LLM Prompting Strategies](LLM%20Prompting%20Strategies.md)when to use it
Community prompt sourced from the open-source GitHub repo J-sephB-lt-n/knowledge-base (GPL-3.0). A "GEPA Reflective Prompt Evolution Can Outperform Reinforcement Learning" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.
tags
roleplaycommunitygeneral
source
J-sephB-lt-n/knowledge-base · GPL-3.0