Skip to content

Conversation

@codelion
Copy link
Member

Updated the controller to use actual top programs for prompt context, ensuring the LLM receives only high-performing examples, and passed inspirations separately. Incremented the package version to 0.0.8 in both pyproject.toml and setup.py.

fixes #92

codelion added 2 commits June 23, 2025 11:58
Updated the controller to use actual top programs for prompt context, ensuring the LLM receives only high-performing examples, and passed inspirations separately. Incremented the package version to 0.0.8 in both pyproject.toml and setup.py.
Introduces an 'inspirations' parameter to PromptSampler and updates the evolution history formatting to include a new inspirations section. Adds logic for formatting, categorizing, and describing inspiration programs, and extends the templates to support this new section. This enhances prompt diversity and creativity by surfacing diverse and creative program examples.
@codelion codelion merged commit 7e6e58e into main Jun 23, 2025
3 checks passed
@codelion codelion deleted the fix-separate-inspirations-from-top-programs branch June 23, 2025 04:07
0x0f0f0f pushed a commit to 0x0f0f0f/openevolve that referenced this pull request Jul 7, 2025
…separate-inspirations-from-top-programs

Refine prompt context and bump version to 0.0.8
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Why are inspirations passed to top_programs during prompt construction?

2 participants