protologue

EvoPrompt

EvoPrompt optimizes prompts with evolutionary algorithms, such as genetic algorithms and differential evolution, in which a language model performs the crossover and mutation steps on a population of prompts.

Description

Using the model as the evolutionary operator keeps every candidate a readable, coherent prompt while borrowing the exploration properties of classical evolutionary search. It needs only API access to the model, not its gradients.

Sources

  1. Guo et al. (2023). EvoPrompt: Connecting LLMs with Evolutionary Algorithms Yields Powerful Prompt Optimizers.

Cite this entry

Protologue. (2026). EvoPrompt. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0154). https://protologue.com/t/evoprompt/

BibTeX
@misc{protologue_evoprompt,
  title = {EvoPrompt},
  author = {{Protologue}},
  year = {2026},
  howpublished = {Protologue: A Taxonomy of Prompting and LLM Techniques, v1.4.0},
  note = {Entry PTL-0154},
  url = {https://protologue.com/t/evoprompt/}
}

Markdown JSON